sokol/sokol_gfx.h

#if defined(SOKOL_IMPL) && !defined(SOKOL_GFX_IMPL)
#define SOKOL_GFX_IMPL
#endif
#ifndef SOKOL_GFX_INCLUDED
/*
    sokol_gfx.h -- simple 3D API wrapper

    Project URL: https://github.com/floooh/sokol

    Example code: https://github.com/floooh/sokol-samples

    Do this:
        #define SOKOL_IMPL or
        #define SOKOL_GFX_IMPL
    before you include this file in *one* C or C++ file to create the
    implementation.

    In the same place define one of the following to select the rendering
    backend:
        #define SOKOL_GLCORE
        #define SOKOL_GLES3
        #define SOKOL_D3D11
        #define SOKOL_METAL
        #define SOKOL_WGPU
        #define SOKOL_DUMMY_BACKEND

    I.e. for the desktop GL it should look like this:

    #include ...
    #include ...
    #define SOKOL_IMPL
    #define SOKOL_GLCORE
    #include "sokol_gfx.h"

    The dummy backend replaces the platform-specific backend code with empty
    stub functions. This is useful for writing tests that need to run on the
    command line.

    Optionally provide the following defines with your own implementations:

    SOKOL_ASSERT(c)             - your own assert macro (default: assert(c))
    SOKOL_UNREACHABLE()         - a guard macro for unreachable code (default: assert(false))
    SOKOL_GFX_API_DECL          - public function declaration prefix (default: extern)
    SOKOL_API_DECL              - same as SOKOL_GFX_API_DECL
    SOKOL_API_IMPL              - public function implementation prefix (default: -)
    SOKOL_TRACE_HOOKS           - enable trace hook callbacks (search below for TRACE HOOKS)
    SOKOL_EXTERNAL_GL_LOADER    - indicates that you're using your own GL loader, in this case
                                  sokol_gfx.h will not include any platform GL headers and disable
                                  the integrated Win32 GL loader

    If sokol_gfx.h is compiled as a DLL, define the following before
    including the declaration or implementation:

    SOKOL_DLL

    On Windows, SOKOL_DLL will define SOKOL_GFX_API_DECL as __declspec(dllexport)
    or __declspec(dllimport) as needed.

    If you want to compile without deprecated structs and functions,
    define:

    SOKOL_NO_DEPRECATED

    Optionally define the following to force debug checks and validations
    even in release mode:

    SOKOL_DEBUG - by default this is defined if _DEBUG is defined

    sokol_gfx DOES NOT:
    ===================
    - create a window, swapchain or the 3D-API context/device, you must do this
      before sokol_gfx is initialized, and pass any required information
      (like 3D device pointers) to the sokol_gfx initialization call

    - present the rendered frame, how this is done exactly usually depends
      on how the window and 3D-API context/device was created

    - provide a unified shader language, instead 3D-API-specific shader
      source-code or shader-bytecode must be provided (for the "official"
      offline shader cross-compiler / code-generator, see here:
      https://github.com/floooh/sokol-tools/blob/master/docs/sokol-shdc.md)


    STEP BY STEP
    ============
    --- to initialize sokol_gfx, after creating a window and a 3D-API
        context/device, call:

            sg_setup(const sg_desc*)

        Depending on the selected 3D backend, sokol-gfx requires some
        information about its runtime environment, like a GPU device pointer,
        default swapchain pixel formats and so on. If you are using sokol_app.h
        for the window system glue, you can use a helper function provided in
        the sokol_glue.h header:

            #include "sokol_gfx.h"
            #include "sokol_app.h"
            #include "sokol_glue.h"
            //...
            sg_setup(&(sg_desc){
                .environment = sglue_environment(),
            });

        To get any logging output for errors and from the validation layer, you
        need to provide a logging callback. Easiest way is through sokol_log.h:

            #include "sokol_log.h"
            //...
            sg_setup(&(sg_desc){
                //...
                .logger.func = slog_func,
            });

    --- create resource objects (at least buffers, shaders and pipelines,
        and optionally images, samplers and render/compute-pass-attachments):

            sg_buffer sg_make_buffer(const sg_buffer_desc*)
            sg_image sg_make_image(const sg_image_desc*)
            sg_sampler sg_make_sampler(const sg_sampler_desc*)
            sg_shader sg_make_shader(const sg_shader_desc*)
            sg_pipeline sg_make_pipeline(const sg_pipeline_desc*)
            sg_attachments sg_make_attachments(const sg_attachments_desc*)

    --- start a render- or compute-pass:

            sg_begin_pass(const sg_pass* pass);

        Typically, render passes render into an externally provided swapchain which
        presents the rendering result on the display. Such a 'swapchain pass'
        is started like this:

            sg_begin_pass(&(sg_pass){ .action = { ... }, .swapchain = sglue_swapchain() })

        ...where .action is an sg_pass_action struct containing actions to be performed
        at the start and end of a render pass (such as clearing the render surfaces to
        a specific color), and .swapchain is an sg_swapchain struct with all the required
        information to render into the swapchain's surfaces.

        To start an 'offscreen render pass' into sokol-gfx image objects, an sg_attachment
        object handle is required instead of an sg_swapchain struct. An offscreen
        pass is started like this (assuming attachments is an sg_attachments handle):

            sg_begin_pass(&(sg_pass){ .action = { ... }, .attachments = attachments });

        To start a compute-pass, just set the .compute item to true:

            sg_begin_pass(&(sg_pass){ .compute = true });

        If the compute pass writes into storage images, provide those as
        'storage attachments' via an sg_attachments object:

            sg_begin_pass(&(sg_pass){ .compute = true, .attachments = attattachments });

    --- set the pipeline state for the next draw call with:

            sg_apply_pipeline(sg_pipeline pip)

    --- fill an sg_bindings struct with the resource bindings for the next
        draw- or dispatch-call (0..N vertex buffers, 0 or 1 index buffer, 0..N images,
        samplers and storage-buffers), and call

            sg_apply_bindings(const sg_bindings* bindings)

        ...to update the resource bindings. Note that in a compute pass, no vertex-
        or index-buffer bindings are allowed and will be rejected by the validation
        layer.

    --- optionally update shader uniform data with:

            sg_apply_uniforms(int ub_slot, const sg_range* data)

        Read the section 'UNIFORM DATA LAYOUT' to learn about the expected memory layout
        of the uniform data passed into sg_apply_uniforms().

    --- kick off a draw call with:

            sg_draw(int base_element, int num_elements, int num_instances)

        The sg_draw() function unifies all the different ways to render primitives
        in a single call (indexed vs non-indexed rendering, and instanced vs non-instanced
        rendering). In case of indexed rendering, base_element and num_element specify
        indices in the currently bound index buffer. In case of non-indexed rendering
        base_element and num_elements specify vertices in the currently bound
        vertex-buffer(s). To perform instanced rendering, the rendering pipeline
        must be setup for instancing (see sg_pipeline_desc below), a separate vertex buffer
        containing per-instance data must be bound, and the num_instances parameter
        must be > 1.

    --- ...or kick of a dispatch call to invoke a compute shader workload:

            sg_dispatch(int num_groups_x, int num_groups_y, int num_groups_z)

        The dispatch args define the number of 'compute workgroups' processed
        by the currently applied compute shader.

    --- finish the current pass with:

            sg_end_pass()

    --- when done with the current frame, call

            sg_commit()

    --- at the end of your program, shutdown sokol_gfx with:

            sg_shutdown()

    --- if you need to destroy resources before sg_shutdown(), call:

            sg_destroy_buffer(sg_buffer buf)
            sg_destroy_image(sg_image img)
            sg_destroy_sampler(sg_sampler smp)
            sg_destroy_shader(sg_shader shd)
            sg_destroy_pipeline(sg_pipeline pip)
            sg_destroy_attachments(sg_attachments atts)

    --- to set a new viewport rectangle, call:

            sg_apply_viewport(int x, int y, int width, int height, bool origin_top_left)

        ...or if you want to specify the viewport rectangle with float values:

            sg_apply_viewportf(float x, float y, float width, float height, bool origin_top_left)

    --- to set a new scissor rect, call:

            sg_apply_scissor_rect(int x, int y, int width, int height, bool origin_top_left)

        ...or with float values:

            sg_apply_scissor_rectf(float x, float y, float width, float height, bool origin_top_left)

        Both sg_apply_viewport() and sg_apply_scissor_rect() must be called
        inside a rendering pass (e.g. not in a compute pass, or outside a pass)

        Note that sg_begin_default_pass() and sg_begin_pass() will reset both the
        viewport and scissor rectangles to cover the entire framebuffer.

    --- to update (overwrite) the content of buffer and image resources, call:

            sg_update_buffer(sg_buffer buf, const sg_range* data)
            sg_update_image(sg_image img, const sg_image_data* data)

        Buffers and images to be updated must have been created with
        sg_buffer_desc.usage.dynamic_update or .stream_update.

        Only one update per frame is allowed for buffer and image resources when
        using the sg_update_*() functions. The rationale is to have a simple
        protection from the CPU scribbling over data the GPU is currently
        using, or the CPU having to wait for the GPU

        Buffer and image updates can be partial, as long as a rendering
        operation only references the valid (updated) data in the
        buffer or image.

    --- to append a chunk of data to a buffer resource, call:

            int sg_append_buffer(sg_buffer buf, const sg_range* data)

        The difference to sg_update_buffer() is that sg_append_buffer()
        can be called multiple times per frame to append new data to the
        buffer piece by piece, optionally interleaved with draw calls referencing
        the previously written data.

        sg_append_buffer() returns a byte offset to the start of the
        written data, this offset can be assigned to
        sg_bindings.vertex_buffer_offsets[n] or
        sg_bindings.index_buffer_offset

        Code example:

        for (...) {
            const void* data = ...;
            const int num_bytes = ...;
            int offset = sg_append_buffer(buf, &(sg_range) { .ptr=data, .size=num_bytes });
            bindings.vertex_buffer_offsets[0] = offset;
            sg_apply_pipeline(pip);
            sg_apply_bindings(&bindings);
            sg_apply_uniforms(...);
            sg_draw(...);
        }

        A buffer to be used with sg_append_buffer() must have been created
        with sg_buffer_desc.usage.dynamic_update or .stream_update.

        If the application appends more data to the buffer then fits into
        the buffer, the buffer will go into the "overflow" state for the
        rest of the frame.

        Any draw calls attempting to render an overflown buffer will be
        silently dropped (in debug mode this will also result in a
        validation error).

        You can also check manually if a buffer is in overflow-state by calling

            bool sg_query_buffer_overflow(sg_buffer buf)

        You can manually check to see if an overflow would occur before adding
        any data to a buffer by calling

            bool sg_query_buffer_will_overflow(sg_buffer buf, size_t size)

        NOTE: Due to restrictions in underlying 3D-APIs, appended chunks of
        data will be 4-byte aligned in the destination buffer. This means
        that there will be gaps in index buffers containing 16-bit indices
        when the number of indices in a call to sg_append_buffer() is
        odd. This isn't a problem when each call to sg_append_buffer()
        is associated with one draw call, but will be problematic when
        a single indexed draw call spans several appended chunks of indices.

    --- to check at runtime for optional features, limits and pixelformat support,
        call:

            sg_features sg_query_features()
            sg_limits sg_query_limits()
            sg_pixelformat_info sg_query_pixelformat(sg_pixel_format fmt)

    --- if you need to call into the underlying 3D-API directly, you must call:

            sg_reset_state_cache()

        ...before calling sokol_gfx functions again

    --- you can inspect the original sg_desc structure handed to sg_setup()
        by calling sg_query_desc(). This will return an sg_desc struct with
        the default values patched in instead of any zero-initialized values

    --- you can get a desc struct matching the creation attributes of a
        specific resource object via:

            sg_buffer_desc sg_query_buffer_desc(sg_buffer buf)
            sg_image_desc sg_query_image_desc(sg_image img)
            sg_sampler_desc sg_query_sampler_desc(sg_sampler smp)
            sg_shader_desc sq_query_shader_desc(sg_shader shd)
            sg_pipeline_desc sg_query_pipeline_desc(sg_pipeline pip)
            sg_attachments_desc sg_query_attachments_desc(sg_attachments atts)

        ...but NOTE that the returned desc structs may be incomplete, only
        creation attributes that are kept around internally after resource
        creation will be filled in, and in some cases (like shaders) that's
        very little. Any missing attributes will be set to zero. The returned
        desc structs might still be useful as partial blueprint for creating
        similar resources if filled up with the missing attributes.

        Calling the query-desc functions on an invalid resource will return
        completely zeroed structs (it makes sense to check  the resource state
        with sg_query_*_state() first)

    --- you can query the default resource creation parameters through the functions

            sg_buffer_desc sg_query_buffer_defaults(const sg_buffer_desc* desc)
            sg_image_desc sg_query_image_defaults(const sg_image_desc* desc)
            sg_sampler_desc sg_query_sampler_defaults(const sg_sampler_desc* desc)
            sg_shader_desc sg_query_shader_defaults(const sg_shader_desc* desc)
            sg_pipeline_desc sg_query_pipeline_defaults(const sg_pipeline_desc* desc)
            sg_attachments_desc sg_query_attachments_defaults(const sg_attachments_desc* desc)

        These functions take a pointer to a desc structure which may contain
        zero-initialized items for default values. These zero-init values
        will be replaced with their concrete values in the returned desc
        struct.

    --- you can inspect various internal resource runtime values via:

            sg_buffer_info sg_query_buffer_info(sg_buffer buf)
            sg_image_info sg_query_image_info(sg_image img)
            sg_sampler_info sg_query_sampler_info(sg_sampler smp)
            sg_shader_info sg_query_shader_info(sg_shader shd)
            sg_pipeline_info sg_query_pipeline_info(sg_pipeline pip)
            sg_attachments_info sg_query_attachments_info(sg_attachments atts)

        ...please note that the returned info-structs are tied quite closely
        to sokol_gfx.h internals, and may change more often than other
        public API functions and structs.

    --- you can query frame stats and control stats collection via:

            sg_query_frame_stats()
            sg_enable_frame_stats()
            sg_disable_frame_stats()
            sg_frame_stats_enabled()

    --- you can ask at runtime what backend sokol_gfx.h has been compiled for:

            sg_backend sg_query_backend(void)

    --- call the following helper functions to compute the number of
        bytes in a texture row or surface for a specific pixel format.
        These functions might be helpful when preparing image data for consumption
        by sg_make_image() or sg_update_image():

            int sg_query_row_pitch(sg_pixel_format fmt, int width, int int row_align_bytes);
            int sg_query_surface_pitch(sg_pixel_format fmt, int width, int height, int row_align_bytes);

        Width and height are generally in number pixels, but note that 'row' has different meaning
        for uncompressed vs compressed pixel formats: for uncompressed formats, a row is identical
        with a single line if pixels, while in compressed formats, one row is a line of *compression blocks*.

        This is why calling sg_query_surface_pitch() for a compressed pixel format and height
        N, N+1, N+2, ... may return the same result.

        The row_align_bytes parameter is for added flexibility. For image data that goes into
        the sg_make_image() or sg_update_image() this should generally be 1, because these
        functions take tightly packed image data as input no matter what alignment restrictions
        exist in the backend 3D APIs.

    ON INITIALIZATION:
    ==================
    When calling sg_setup(), a pointer to an sg_desc struct must be provided
    which contains initialization options. These options provide two types
    of information to sokol-gfx:

        (1) upper bounds and limits needed to allocate various internal
            data structures:
                - the max number of resources of each type that can
                  be alive at the same time, this is used for allocating
                  internal pools
                - the max overall size of uniform data that can be
                  updated per frame, including a worst-case alignment
                  per uniform update (this worst-case alignment is 256 bytes)
                - the max size of all dynamic resource updates (sg_update_buffer,
                  sg_append_buffer and sg_update_image) per frame
                - the max number of compute-dispatch calls in a compute pass
            Not all of those limit values are used by all backends, but it is
            good practice to provide them none-the-less.

        (2) 3D backend "environment information" in a nested sg_environment struct:
            - pointers to backend-specific context- or device-objects (for instance
              the D3D11, WebGPU or Metal device objects)
            - defaults for external swapchain pixel formats and sample counts,
              these will be used as default values in image and pipeline objects,
              and the sg_swapchain struct passed into sg_begin_pass()
            Usually you provide a complete sg_environment struct through
            a helper function, as an example look at the sglue_environment()
            function in the sokol_glue.h header.

    See the documentation block of the sg_desc struct below for more information.


    ON RENDER PASSES
    ================
    Relevant samples:
        - https://floooh.github.io/sokol-html5/offscreen-sapp.html
        - https://floooh.github.io/sokol-html5/offscreen-msaa-sapp.html
        - https://floooh.github.io/sokol-html5/mrt-sapp.html
        - https://floooh.github.io/sokol-html5/mrt-pixelformats-sapp.html

    A render pass groups rendering commands into a set of render target images
    (called 'pass attachments'). Render target images can be used in subsequent
    passes as textures (it is invalid to use the same image both as render target
    and as texture in the same pass).

    The following sokol-gfx functions must only be called inside a render-pass:

        sg_apply_viewport[f]
        sg_apply_scissor_rect[f]
        sg_draw

    The following function may be called inside a render- or compute-pass, but
    not outside a pass:

        sg_apply_pipeline
        sg_apply_bindings
        sg_apply_uniforms

    A frame must have at least one 'swapchain render pass' which renders into an
    externally provided swapchain provided as an sg_swapchain struct to the
    sg_begin_pass() function. If you use sokol_gfx.h together with sokol_app.h,
    just call the sglue_swapchain() helper function in sokol_glue.h to
    provide the swapchain information. Otherwise the following information
    must be provided:

        - the color pixel-format of the swapchain's render surface
        - an optional depth/stencil pixel format if the swapchain
          has a depth/stencil buffer
        - an optional sample-count for MSAA rendering
        - NOTE: the above three values can be zero-initialized, in that
          case the defaults from the sg_environment struct will be used that
          had been passed to the sg_setup() function.
        - a number of backend specific objects:
            - GL/GLES3: just a GL framebuffer handle
            - D3D11:
                - an ID3D11RenderTargetView for the rendering surface
                - if MSAA is used, an ID3D11RenderTargetView as
                  MSAA resolve-target
                - an optional ID3D11DepthStencilView for the
                  depth/stencil buffer
            - WebGPU
                - a WGPUTextureView object for the rendering surface
                - if MSAA is used, a WGPUTextureView object as MSAA resolve target
                - an optional WGPUTextureView for the
            - Metal (NOTE that the roles of provided surfaces is slightly
              different in Metal than in D3D11 or WebGPU, notably, the
              CAMetalDrawable is either rendered to directly, or serves
              as MSAA resolve target):
                - a CAMetalDrawable object which is either rendered
                  into directly, or in case of MSAA rendering, serves
                  as MSAA-resolve-target
                - if MSAA is used, an multisampled MTLTexture where
                  rendering goes into
                - an optional MTLTexture for the depth/stencil buffer

    It's recommended that you create a helper function which returns an
    initialized sg_swapchain struct by value. This can then be directly plugged
    into the sg_begin_pass function like this:

        sg_begin_pass(&(sg_pass){ .swapchain = sglue_swapchain() });

    As an example for such a helper function check out the function sglue_swapchain()
    in the sokol_glue.h header.

    For offscreen render passes, the render target images used in a render pass
    are baked into an immutable sg_attachments object.

    For a simple offscreen scenario with one color-, one depth-stencil-render
    target and without multisampling, creating an attachment object looks like this:

    First create two render target images, one with a color pixel format,
    and one with the depth- or depth-stencil pixel format. Both images
    must have the same dimensions:

        const sg_image color_img = sg_make_image(&(sg_image_desc){
            .render_target = true,
            .width = 256,
            .height = 256,
            .pixel_format = SG_PIXELFORMAT_RGBA8,
            .sample_count = 1,
        });
        const sg_image depth_img = sg_make_image(&(sg_image_desc){
            .render_target = true,
            .width = 256,
            .height = 256,
            .pixel_format = SG_PIXELFORMAT_DEPTH,
            .sample_count = 1,
        });

    NOTE: when creating render target images, have in mind that some default values
    are aligned with the default environment attributes in the sg_environment struct
    that was passed into the sg_setup() call:

        - the default value for sg_image_desc.pixel_format is taken from
          sg_environment.defaults.color_format
        - the default value for sg_image_desc.sample_count is taken from
          sg_environment.defaults.sample_count
        - the default value for sg_image_desc.num_mipmaps is always 1

    Next create an attachments object:

        const sg_attachments atts = sg_make_attachments(&(sg_attachments_desc){
            .colors[0].image = color_img,
            .depth_stencil.image = depth_img,
        });

    This attachments object is then passed into the sg_begin_pass() function
    in place of the swapchain struct:

        sg_begin_pass(&(sg_pass){ .attachments = atts });

    Swapchain and offscreen passes form dependency trees each with a swapchain
    pass at the root, offscreen passes as nodes, and render target images as
    dependencies between passes.

    sg_pass_action structs are used to define actions that should happen at the
    start and end of rendering passes (such as clearing pass attachments to a
    specific color or depth-value, or performing an MSAA resolve operation at
    the end of a pass).

    A typical sg_pass_action object which clears the color attachment to black
    might look like this:

        const sg_pass_action = {
            .colors[0] = {
                .load_action = SG_LOADACTION_CLEAR,
                .clear_value = { 0.0f, 0.0f, 0.0f, 1.0f }
            }
        };

    This omits the defaults for the color attachment store action, and
    the depth-stencil-attachments actions. The same pass action with the
    defaults explicitly filled in would look like this:

        const sg_pass_action pass_action = {
            .colors[0] = {
                .load_action = SG_LOADACTION_CLEAR,
                .store_action = SG_STOREACTION_STORE,
                .clear_value = { 0.0f, 0.0f, 0.0f, 1.0f }
            },
            .depth = = {
                .load_action = SG_LOADACTION_CLEAR,
                .store_action = SG_STOREACTION_DONTCARE,
                .clear_value = 1.0f,
            },
            .stencil = {
                .load_action = SG_LOADACTION_CLEAR,
                .store_action = SG_STOREACTION_DONTCARE,
                .clear_value = 0
            }
        };

    With the sg_pass object and sg_pass_action struct in place everything
    is ready now for the actual render pass:

    Using such this prepared sg_pass_action in a swapchain pass looks like
    this:

        sg_begin_pass(&(sg_pass){
            .action = pass_action,
            .swapchain = sglue_swapchain()
        });
        ...
        sg_end_pass();

    ...of alternatively in one offscreen pass:

        sg_begin_pass(&(sg_pass){
            .action = pass_action,
            .attachments = attachments,
        });
        ...
        sg_end_pass();

    Offscreen rendering can also go into a mipmap, or a slice/face of
    a cube-, array- or 3d-image (which some restrictions, for instance
    it's not possible to create a 3D image with a depth/stencil pixel format,
    these exceptions are generally caught by the sokol-gfx validation layer).

    The mipmap/slice selection happens at attachments creation time, for instance
    to render into mipmap 2 of slice 3 of an array texture:

        const sg_attachments atts = sg_make_attachments(&(sg_attachments_desc){
            .colors[0] = {
                .image = color_img,
                .mip_level = 2,
                .slice = 3,
            },
            .depth_stencil.image = depth_img,
        });

    If MSAA offscreen rendering is desired, the multi-sample rendering result
    must be 'resolved' into a separate 'resolve image', before that image can
    be used as texture.

    Creating a simple attachments object for multisampled rendering requires
    3 attachment images: the color attachment image which has a sample
    count > 1, a resolve attachment image of the same size and pixel format
    but a sample count == 1, and a depth/stencil attachment image with
    the same size and sample count as the color attachment image:

        const sg_image color_img = sg_make_image(&(sg_image_desc){
            .render_target = true,
            .width = 256,
            .height = 256,
            .pixel_format = SG_PIXELFORMAT_RGBA8,
            .sample_count = 4,
        });
        const sg_image resolve_img = sg_make_image(&(sg_image_desc){
            .render_target = true,
            .width = 256,
            .height = 256,
            .pixel_format = SG_PIXELFORMAT_RGBA8,
            .sample_count = 1,
        });
        const sg_image depth_img = sg_make_image(&(sg_image_desc){
            .render_target = true,
            .width = 256,
            .height = 256,
            .pixel_format = SG_PIXELFORMAT_DEPTH,
            .sample_count = 4,
        });

    ...create the attachments object:

        const sg_attachments atts = sg_make_attachments(&(sg_attachments_desc){
            .colors[0].image = color_img,
            .resolves[0].image = resolve_img,
            .depth_stencil.image = depth_img,
        });

    If an attachments object defines a resolve image in a specific resolve attachment slot,
    an 'msaa resolve operation' will happen in sg_end_pass().

    In this scenario, the content of the MSAA color attachment doesn't need to be
    preserved (since it's only needed inside sg_end_pass for the msaa-resolve), so
    the .store_action should be set to "don't care":

        const sg_pass_action = {
            .colors[0] = {
                .load_action = SG_LOADACTION_CLEAR,
                .store_action = SG_STOREACTION_DONTCARE,
                .clear_value = { 0.0f, 0.0f, 0.0f, 1.0f }
            }
        };

    The actual render pass looks as usual:

        sg_begin_pass(&(sg_pass){ .action = pass_action, .attachments = atts });
        ...
        sg_end_pass();

    ...after sg_end_pass() the only difference to the non-msaa scenario is that the
    rendering result which is going to be used as texture in a followup pass is
    in 'resolve_img', not in 'color_img' (in fact, trying to bind color_img as a
    texture would result in a validation error).


    ON COMPUTE PASSES
    =================
    Compute passes are used to update the content of storage buffers and
    storage images by running compute shader code on
    the GPU. Updating storage resources with a compute shader will almost always
    be more efficient than computing the same data on the CPU and then uploading
    it via `sg_update_buffer()` or `sg_update_image()`.

    NOTE: compute passes are only supported on the following platforms and
    backends:

        - macOS and iOS with Metal
        - Windows with D3D11 and OpenGL
        - Linux with OpenGL or GLES3.1+
        - Web with WebGPU
        - Android with GLES3.1+

    ...this means compute shaders can't be used on the following platform/backend
    combos (the same restrictions apply to using storage buffers without compute
    shaders):

        - macOS with GL
        - iOS with GLES3
        - Web with WebGL2

    A compute pass which only updates storage buffers is started with:

        sg_begin_pass(&(sg_pass){ .compute = true });

    ...if the compute pass updates storage images, the images must be 'bound'
    via an sg_attachments object:

        sg_begin_pass(&(sg_pass){ .compute = true, .attachments = attachments });

    Image objects in such a compute pass attachments object must be created with
    `storage_attachment` usage:

        sg_image storage_image = sg_make_image(&(sg_image_desc){
            .usage = {
                .storage_attachment = true,
            },
            // ...
        });

    ...a compute pass is finished with a regular:

        sg_end_pass();

    Typically the following functions will be called inside a compute pass:

        sg_apply_pipeline()
        sg_apply_bindings()
        sg_apply_uniforms()
        sg_dispatch()

    The following functions are disallowed inside a compute pass
    and will cause validation layer errors:

        sg_apply_viewport[f]()
        sg_apply_scissor_rect[f]()
        sg_draw()

    Only special 'compute shaders' and 'compute pipelines' can be used in
    compute passes. A compute shader only has a compute-function instead
    of a vertex- and fragment-function pair, and it doesn't accept vertex-
    and index-buffers as input, only storage-buffers, textures, non-filtering
    samplers and images via storage attachments (more details on compute shaders in
    the following section).

    A compute pipeline is created by providing a compute shader object,
    setting the .compute creation parameter to true and not defining any
    'render state':

        sg_pipeline pip = sg_make_pipeline(&(sg_pipeline_desc){
            .compute = true,
            .shader = compute_shader,
        });

    The sg_apply_bindings and sg_apply_uniforms calls are the same as in
    render passes, with the exception that no vertex- and index-buffers
    can be bound in the sg_apply_bindings call.

    Finally to kick off a compute workload, call sg_dispatch with the
    number of workgroups in the x, y and z-dimension:

        sg_dispatch(int num_groups_x, int num_groups_y, int num_groups_z)

    Also see the following compute-shader samples:

        - https://floooh.github.io/sokol-webgpu/instancing-compute-sapp.html
        - https://floooh.github.io/sokol-webgpu/computeboids-sapp.html
        - https://floooh.github.io/sokol-webgpu/imageblur-sapp.html


    ON SHADER CREATION
    ==================
    sokol-gfx doesn't come with an integrated shader cross-compiler, instead
    backend-specific shader sources or binary blobs need to be provided when
    creating a shader object, along with reflection information about the
    shader resource binding interface needed to bind sokol-gfx resources to the
    proper shader inputs.

    The easiest way to provide all this shader creation data is to use the
    sokol-shdc shader compiler tool to compile shaders from a common
    GLSL syntax into backend-specific sources or binary blobs, along with
    shader interface information and uniform blocks and storage buffer array items
    mapped to C structs.

    To create a shader using a C header which has been code-generated by sokol-shdc:

        // include the C header code-generated by sokol-shdc:
        #include "myshader.glsl.h"
        ...

        // create shader using a code-generated helper function from the C header:
        sg_shader shd = sg_make_shader(myshader_shader_desc(sg_query_backend()));

    The samples in the 'sapp' subdirectory of the sokol-samples project
    also use the sokol-shdc approach:

        https://github.com/floooh/sokol-samples/tree/master/sapp

    If you're planning to use sokol-shdc, you can stop reading here, instead
    continue with the sokol-shdc documentation:

        https://github.com/floooh/sokol-tools/blob/master/docs/sokol-shdc.md

    To create shaders with backend-specific shader code or binary blobs,
    the sg_make_shader() function requires the following information:

    - Shader code or shader binary blobs for the vertex- and fragment-, or the
      compute-shader-stage:
        - for the desktop GL backend, source code can be provided in '#version 410' or
          '#version 430', version 430 is required when using storage buffers and
          compute shaders support, but note that this is not available on macOS
        - for the GLES3 backend, source code must be provided in '#version 300 es' or
          '#version 310 es' syntax (version 310 is required for storage buffer and
          compute shader support, but note that this is not supported on WebGL2)
        - for the D3D11 backend, shaders can be provided as source or binary
          blobs, the source code should be in HLSL4.0 (for compatibility with old
          low-end GPUs) or preferably in HLSL5.0 syntax, note that when
          shader source code is provided for the D3D11 backend, sokol-gfx will
          dynamically load 'd3dcompiler_47.dll'
        - for the Metal backends, shaders can be provided as source or binary blobs, the
          MSL version should be in 'metal-1.1' (other versions may work but are not tested)
        - for the WebGPU backend, shaders must be provided as WGSL source code
        - optionally the following shader-code related attributes can be provided:
            - an entry function name (only on D3D11 or Metal, but not OpenGL)
            - on D3D11 only, a compilation target (default is "vs_4_0" and "ps_4_0")

    - Information about the input vertex attributes used by the vertex shader,
      most of that backend-specific:
        - An optional 'base type' (float, signed-/unsigned-int) for each vertex
          attribute. When provided, this used by the validation layer to check
          that the CPU-side input vertex format is compatible with the input
          vertex declaration of the vertex shader.
        - Metal: no location information needed since vertex attributes are always bound
          by their attribute location defined in the shader via '[[attribute(N)]]'
        - WebGPU: no location information needed since vertex attributes are always
          bound by their attribute location defined in the shader via `@location(N)`
        - GLSL: vertex attribute names can be optionally provided, in that case their
          location will be looked up by name, otherwise, the vertex attribute location
          can be defined with 'layout(location = N)'
        - D3D11: a 'semantic name' and 'semantic index' must be provided for each vertex
          attribute, e.g. if the vertex attribute is defined as 'TEXCOORD1' in the shader,
          the semantic name would be 'TEXCOORD', and the semantic index would be '1'

      NOTE that vertex attributes currently must not have gaps. This requirement
      may be relaxed in the future.

    - Specifically for Metal compute shaders, the 'number of threads per threadgroup'
      must be provided. Normally this is extracted by sokol-shdc from the GLSL
      shader source code. For instance the following statement in the input
      GLSL:

        layout(local_size_x=64, local_size_y=1, local_size_z=1) in;

      ...will be communicated to the sokol-gfx Metal backend in the
      code-generated sg_shader_desc struct:

        (sg_shader_desc){
            .mtl_threads_per_threadgroup = { .x = 64, .y = 1, .z = 1 },
        }

    - Information about each uniform block binding used in the shader:
        - the shader stage of the uniform block (vertex, fragment or compute)
        - the size of the uniform block in number of bytes
        - a memory layout hint (currently 'native' or 'std140') where 'native' defines a
          backend-specific memory layout which shouldn't be used for cross-platform code.
          Only std140 guarantees a backend-agnostic memory layout.
        - a backend-specific bind slot:
            - D3D11/HLSL: the buffer register N (`register(bN)`) where N is 0..7
            - Metal/MSL: the buffer bind slot N (`[[buffer(N)]]`) where N is 0..7
            - WebGPU: the binding N in `@group(0) @binding(N)` where N is 0..15
        - For GLSL only: a description of the internal uniform block layout, which maps
          member types and their offsets on the CPU side to uniform variable names
          in the GLSL shader
        - please also NOTE the documentation sections about UNIFORM DATA LAYOUT
          and CROSS-BACKEND COMMON UNIFORM DATA LAYOUT below!

    - A description of each storage buffer binding used in the shader:
        - the shader stage of the storage buffer
        - a boolean 'readonly' flag, this is used for validation and hazard
          tracking in some 3D backends. Note that in render passes, only
          readonly storage buffer bindings are allowed. In compute passes, any
          read/write storage buffer binding is assumed to be written to by the
          compute shader.
        - a backend-specific bind slot:
            - D3D11/HLSL:
                - for readonly storage buffer bindings: the texture register N
                  (`register(tN)`) where N is 0..23 (in HLSL, readonly storage
                  buffers and textures share the same bind space for
                  'shader resource views')
                - for read/write storage buffer buffer bindings: the UAV register N
                  (`register(uN)`) where N is 0..11 (in HLSL, readwrite storage
                  buffers use their own bind space for 'unordered access views')
            - Metal/MSL: the buffer bind slot N (`[[buffer(N)]]`) where N is 8..15
            - WebGPU/WGSL: the binding N in `@group(0) @binding(N)` where N is 0..127
            - GL/GLSL: the buffer binding N in `layout(binding=N)` where N is 0..7
        - note that storage buffer bindings are not supported on all backends
          and platforms

    - A description of each storage image binding used in the shader (only supported
      in compute shaders):
        - the shader stage (*must* be compute)
        - the expected image type:
            - SG_IMAGETYPE_2D
            - SG_IMAGETYPE_CUBE
            - SG_IMAGETYPE_3D
            - SG_IMAGETYPE_ARRAY
        - the 'access pixel format', this is currently limited to:
            - SG_PIXELFORMAT_RGBA8
            - SG_PIXELFORMAT_RGBA8SN/UI/SI
            - SG_PIXELFORMAT_RGBA16UI/SI/F
            - SG_PIXELFORMAT_R32UIUI/SI/F
            - SG_PIXELFORMAT_RG32UI/SI/F
            - SG_PIXELFORMAT_RGBA32UI/SI/F
        - the access type (readwrite or writeonly)
        - a backend-specific bind slot:
            - D3D11/HLSL: the UAV register N (`register(uN)` where N is 0..11, the
              bind slot must not collide with UAV storage buffer bindings
            - Metal/MSL: the texture bind slot N (`[[texture(N)]])` where N is 0..19,
              the bind slot must not collide with other texture bindings on the same
              stage
            - WebGPU/WGSL: the binding N in `@group(2) @binding(N)` where N is 0..3
            - GL/GLSL: the buffer binding N in `layout(binding=N)` where N is 0..3
        - note that storage image bindings are not supported on all backends and platforms

    - A description of each texture binding used in the shader:
        - the shader stage of the texture (vertex, fragment or compute)
        - the expected image type:
            - SG_IMAGETYPE_2D
            - SG_IMAGETYPE_CUBE
            - SG_IMAGETYPE_3D
            - SG_IMAGETYPE_ARRAY
        - the expected 'image sample type':
            - SG_IMAGESAMPLETYPE_FLOAT
            - SG_IMAGESAMPLETYPE_DEPTH
            - SG_IMAGESAMPLETYPE_SINT
            - SG_IMAGESAMPLETYPE_UINT
            - SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT
        - a flag whether the texture is expected to be multisampled
        - a backend-specific bind slot:
            - D3D11/HLSL: the texture register N (`register(tN)`) where N is 0..23
              (in HLSL, readonly storage buffers and texture share the same bind space)
            - Metal/MSL: the texture bind slot N (`[[texture(N)]]`) where N is 0..19
              (the bind slot must not collide with storage image bindings on the same stage)
            - WebGPU/WGSL: the binding N in `@group(0) @binding(N)` where N is 0..127

    - A description of each sampler used in the shader:
        - the shader stage of the sampler (vertex, fragment or compute)
        - the expected sampler type:
            - SG_SAMPLERTYPE_FILTERING,
            - SG_SAMPLERTYPE_NONFILTERING,
            - SG_SAMPLERTYPE_COMPARISON,
        - a backend-specific bind slot:
            - D3D11/HLSL: the sampler register N (`register(sN)`) where N is 0..15
            - Metal/MSL: the sampler bind slot N (`[[sampler(N)]]`) where N is 0..15
            - WebGPU/WGSL: the binding N in `@group(0) @binding(N)` where N is 0..127

    - An array of 'image-sampler-pairs' used by the shader to sample textures,
      for D3D11, Metal and WebGPU this is used for validation purposes to check
      whether the texture and sampler are compatible with each other (especially
      WebGPU is very picky about combining the correct
      texture-sample-type with the correct sampler-type). For GLSL an
      additional 'combined-image-sampler name' must be provided because 'OpenGL
      style GLSL' cannot handle separate texture and sampler objects, but still
      groups them into a traditional GLSL 'sampler object'.

    Compatibility rules for image-sample-type vs sampler-type are as follows:

        - SG_IMAGESAMPLETYPE_FLOAT => (SG_SAMPLERTYPE_FILTERING or SG_SAMPLERTYPE_NONFILTERING)
        - SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT => SG_SAMPLERTYPE_NONFILTERING
        - SG_IMAGESAMPLETYPE_SINT => SG_SAMPLERTYPE_NONFILTERING
        - SG_IMAGESAMPLETYPE_UINT => SG_SAMPLERTYPE_NONFILTERING
        - SG_IMAGESAMPLETYPE_DEPTH => SG_SAMPLERTYPE_COMPARISON

    Backend-specific bindslot ranges (not relevant when using sokol-shdc):

        - D3D11/HLSL:
            - separate bindslot space per shader stage
            - uniform block bindings (as cbuffer): `register(b0..b7)`
            - texture- and readonly storage buffer bindings: `register(t0..t23)`
            - read/write storage buffer and storage image bindings: `register(u0..u11)`
            - samplers: `register(s0..s15)`
        - Metal/MSL:
            - separate bindslot space per shader stage
            - uniform blocks: `[[buffer(0..7)]]`
            - storage buffers: `[[buffer(8..15)]]`
            - textures and storage image bindings: `[[texture(0..19)]]`
            - samplers: `[[sampler(0..15)]]`
        - WebGPU/WGSL:
            - common bindslot space across shader stages
            - uniform blocks: `@group(0) @binding(0..15)`
            - textures, samplers and storage buffers: `@group(1) @binding(0..127)`
            - storage image bindings: `@group(2) @binding(0..3)`
        - GL/GLSL:
            - uniforms and image-samplers are bound by name
            - storage buffer bindings: `layout(std430, binding=0..7)` (common
              bindslot space across shader stages)
            - storage image bindings: `layout(binding=0..3, [access_format])`

    For example code of how to create backend-specific shader objects,
    please refer to the following samples:

        - for D3D11:    https://github.com/floooh/sokol-samples/tree/master/d3d11
        - for Metal:    https://github.com/floooh/sokol-samples/tree/master/metal
        - for OpenGL:   https://github.com/floooh/sokol-samples/tree/master/glfw
        - for GLES3:    https://github.com/floooh/sokol-samples/tree/master/html5
        - for WebGPU:   https://github.com/floooh/sokol-samples/tree/master/wgpu


    ON SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT AND SG_SAMPLERTYPE_NONFILTERING
    ========================================================================
    The WebGPU backend introduces the concept of 'unfilterable-float' textures,
    which can only be combined with 'nonfiltering' samplers (this is a restriction
    specific to WebGPU, but since the same sokol-gfx code should work across
    all backend, the sokol-gfx validation layer also enforces this restriction
    - the alternative would be undefined behaviour in some backend APIs on
    some devices).

    The background is that some mobile devices (most notably iOS devices) can
    not perform linear filtering when sampling textures with certain pixel
    formats, most notable the 32F formats:

        - SG_PIXELFORMAT_R32F
        - SG_PIXELFORMAT_RG32F
        - SG_PIXELFORMAT_RGBA32F

    The information of whether a shader is going to be used with such an
    unfilterable-float texture must already be provided in the sg_shader_desc
    struct when creating the shader (see the above section "ON SHADER CREATION").

    If you are using the sokol-shdc shader compiler, the information whether a
    texture/sampler binding expects an 'unfilterable-float/nonfiltering'
    texture/sampler combination cannot be inferred from the shader source
    alone, you'll need to provide this hint via annotation-tags. For instance
    here is an example from the ozz-skin-sapp.c sample shader which samples an
    RGBA32F texture with skinning matrices in the vertex shader:

    ```glsl
    @image_sample_type joint_tex unfilterable_float
    uniform texture2D joint_tex;
    @sampler_type smp nonfiltering
    uniform sampler smp;
    ```

    This will result in SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT and
    SG_SAMPLERTYPE_NONFILTERING being written to the code-generated
    sg_shader_desc struct.


    ON VERTEX FORMATS
    =================
    Sokol-gfx implements the same strict mapping rules from CPU-side
    vertex component formats to GPU-side vertex input data types:

    - float and packed normalized CPU-side formats must be used as
      floating point base type in the vertex shader
    - packed signed-integer CPU-side formats must be used as signed
      integer base type in the vertex shader
    - packed unsigned-integer CPU-side formats must be used as unsigned
      integer base type in the vertex shader

    These mapping rules are enforced by the sokol-gfx validation layer,
    but only when sufficient reflection information is provided in
    `sg_shader_desc.attrs[].base_type`. This is the case when sokol-shdc
    is used, otherwise the default base_type will be SG_SHADERATTRBASETYPE_UNDEFINED
    which causes the sokol-gfx validation check to be skipped (of course you
    can also provide the per-attribute base type information manually when
    not using sokol-shdc).

    The detailed mapping rules from SG_VERTEXFORMAT_* to GLSL data types
    are as follows:

    - FLOAT[*] => float, vec*
    - BYTE4N => vec* (scaled to -1.0 .. +1.0)
    - UBYTE4N => vec* (scaled to 0.0 .. +1.0)
    - SHORT[*]N => vec* (scaled to -1.0 .. +1.0)
    - USHORT[*]N => vec* (scaled to 0.0 .. +1.0)
    - INT[*] => int, ivec*
    - UINT[*] => uint, uvec*
    - BYTE4 => int*
    - UBYTE4 => uint*
    - SHORT[*] => int*
    - USHORT[*] => uint*

    NOTE that sokol-gfx only provides vertex formats with sizes of a multiple
    of 4 (e.g. BYTE4N but not BYTE2N). This is because vertex components must
    be 4-byte aligned anyway.


    UNIFORM DATA LAYOUT:
    ====================
    NOTE: if you use the sokol-shdc shader compiler tool, you don't need to worry
    about the following details.

    The data that's passed into the sg_apply_uniforms() function must adhere to
    specific layout rules so that the GPU shader finds the uniform block
    items at the right offset.

    For the D3D11 and Metal backends, sokol-gfx only cares about the size of uniform
    blocks, but not about the internal layout. The data will just be copied into
    a uniform/constant buffer in a single operation and it's up you to arrange the
    CPU-side layout so that it matches the GPU side layout. This also means that with
    the D3D11 and Metal backends you are not limited to a 'cross-platform' subset
    of uniform variable types.

    If you ever only use one of the D3D11, Metal *or* WebGPU backend, you can stop reading here.

    For the GL backends, the internal layout of uniform blocks matters though,
    and you are limited to a small number of uniform variable types. This is
    because sokol-gfx must be able to locate the uniform block members in order
    to upload them to the GPU with glUniformXXX() calls.

    To describe the uniform block layout to sokol-gfx, the following information
    must be passed to the sg_make_shader() call in the sg_shader_desc struct:

        - a hint about the used packing rule (either SG_UNIFORMLAYOUT_NATIVE or
          SG_UNIFORMLAYOUT_STD140)
        - a list of the uniform block members types in the correct order they
          appear on the CPU side

    For example if the GLSL shader has the following uniform declarations:

        uniform mat4 mvp;
        uniform vec2 offset0;
        uniform vec2 offset1;
        uniform vec2 offset2;

    ...and on the CPU side, there's a similar C struct:

        typedef struct {
            float mvp[16];
            float offset0[2];
            float offset1[2];
            float offset2[2];
        } params_t;

    ...the uniform block description in the sg_shader_desc must look like this:

        sg_shader_desc desc = {
            .vs.uniform_blocks[0] = {
                .size = sizeof(params_t),
                .layout = SG_UNIFORMLAYOUT_NATIVE,  // this is the default and can be omitted
                .uniforms = {
                    // order must be the same as in 'params_t':
                    [0] = { .name = "mvp", .type = SG_UNIFORMTYPE_MAT4 },
                    [1] = { .name = "offset0", .type = SG_UNIFORMTYPE_VEC2 },
                    [2] = { .name = "offset1", .type = SG_UNIFORMTYPE_VEC2 },
                    [3] = { .name = "offset2", .type = SG_UNIFORMTYPE_VEC2 },
                }
            }
        };

    With this information sokol-gfx can now compute the correct offsets of the data items
    within the uniform block struct.

    The SG_UNIFORMLAYOUT_NATIVE packing rule works fine if only the GL backends are used,
    but for proper D3D11/Metal/GL a subset of the std140 layout must be used which is
    described in the next section:


    CROSS-BACKEND COMMON UNIFORM DATA LAYOUT
    ========================================
    For cross-platform / cross-3D-backend code it is important that the same uniform block
    layout on the CPU side can be used for all sokol-gfx backends. To achieve this,
    a common subset of the std140 layout must be used:

    - The uniform block layout hint in sg_shader_desc must be explicitly set to
      SG_UNIFORMLAYOUT_STD140.
    - Only the following GLSL uniform types can be used (with their associated sokol-gfx enums):
        - float => SG_UNIFORMTYPE_FLOAT
        - vec2  => SG_UNIFORMTYPE_FLOAT2
        - vec3  => SG_UNIFORMTYPE_FLOAT3
        - vec4  => SG_UNIFORMTYPE_FLOAT4
        - int   => SG_UNIFORMTYPE_INT
        - ivec2 => SG_UNIFORMTYPE_INT2
        - ivec3 => SG_UNIFORMTYPE_INT3
        - ivec4 => SG_UNIFORMTYPE_INT4
        - mat4  => SG_UNIFORMTYPE_MAT4
    - Alignment for those types must be as follows (in bytes):
        - float => 4
        - vec2  => 8
        - vec3  => 16
        - vec4  => 16
        - int   => 4
        - ivec2 => 8
        - ivec3 => 16
        - ivec4 => 16
        - mat4  => 16
    - Arrays are only allowed for the following types: vec4, int4, mat4.

    Note that the HLSL cbuffer layout rules are slightly different from the
    std140 layout rules, this means that the cbuffer declarations in HLSL code
    must be tweaked so that the layout is compatible with std140.

    The by far easiest way to tackle the common uniform block layout problem is
    to use the sokol-shdc shader cross-compiler tool!


    ON STORAGE BUFFERS
    ==================
    The two main purpose of storage buffers are:

        - to be populated by compute shaders with dynamically generated data
        - for providing random-access data to all shader stages

    Storage buffers can be used to pass large amounts of random access structured
    data from the CPU side to the shaders. They are similar to data textures, but are
    more convenient to use both on the CPU and shader side since they can be accessed
    in shaders as as a 1-dimensional array of struct items.

    Storage buffers are *NOT* supported on the following platform/backend combos:

    - macOS+GL (because storage buffers require GL 4.3, while macOS only goes up to GL 4.1)
    - platforms which only support a GLES3.0 context (WebGL2 and iOS)

    To use storage buffers, the following steps are required:

        - write a shader which uses storage buffers (vertex- and fragment-shaders
          can only read from storage buffers, while compute-shaders can both read
          and write storage buffers)
        - create one or more storage buffers via sg_make_buffer() with the
          `.usage.storage_buffer = true`
        - when creating a shader via sg_make_shader(), populate the sg_shader_desc
          struct with binding info (when using sokol-shdc, this step will be taken care
          of automatically)
            - which storage buffer bind slots on the vertex-, fragment- or compute-stage
              are occupied
            - whether the storage buffer on that bind slot is readonly (readonly
              bindings are required for vertex- and fragment-shaders, and in compute
              shaders the readonly flag is used to control hazard tracking in some
              3D backends)

        - when calling sg_apply_bindings(), apply the matching bind slots with the previously
          created storage buffers
        - ...and that's it.

    For more details, see the following backend-agnostic sokol samples:

    - simple vertex pulling from a storage buffer:
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/vertexpull-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/vertexpull-sapp.glsl
    - instanced rendering via storage buffers (vertex- and instance-pulling):
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/instancing-pull-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/instancing-pull-sapp.glsl
    - storage buffers both on the vertex- and fragment-stage:
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/sbuftex-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/sbuftex-sapp.glsl
    - the Ozz animation sample rewritten to pull all rendering data from storage buffers:
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/ozz-storagebuffer-sapp.cc
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/ozz-storagebuffer-sapp.glsl
    - the instancing sample modified to use compute shaders:
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/instancing-compute-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/instancing-compute-sapp.glsl
    - the Compute Boids sample ported to sokol-gfx:
        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/computeboids-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/computeboids-sapp.glsl

    ...also see the following backend-specific vertex pulling samples (those also don't use sokol-shdc):

    - D3D11: https://github.com/floooh/sokol-samples/blob/master/d3d11/vertexpulling-d3d11.c
    - desktop GL: https://github.com/floooh/sokol-samples/blob/master/glfw/vertexpulling-glfw.c
    - Metal: https://github.com/floooh/sokol-samples/blob/master/metal/vertexpulling-metal.c
    - WebGPU: https://github.com/floooh/sokol-samples/blob/master/wgpu/vertexpulling-wgpu.c

    ...and the backend specific compute shader samples:

    - D3D11: https://github.com/floooh/sokol-samples/blob/master/d3d11/instancing-compute-d3d11.c
    - desktop GL: https://github.com/floooh/sokol-samples/blob/master/glfw/instancing-compute-glfw.c
    - Metal: https://github.com/floooh/sokol-samples/blob/master/metal/instancing-compute-metal.c
    - WebGPU: https://github.com/floooh/sokol-samples/blob/master/wgpu/instancing-compute-wgpu.c

    Storage buffer shader authoring caveats when using sokol-shdc:

        - declare a read-only storage buffer interface block with `layout(binding=N) readonly buffer [name] { ... }`
          (where 'N' is the index in `sg_bindings.storage_buffers[N]`)
        - ...or a read/write storage buffer interface block with `layout(binding=N) buffer [name] { ... }`
        - declare a struct which describes a single array item in the storage buffer interface block
        - only put a single flexible array member into the storage buffer interface block

    E.g. a complete example in 'sokol-shdc GLSL':

        ```glsl
        @vs
        // declare a struct:
        struct sb_vertex {
            vec3 pos;
            vec4 color;
        }
        // declare a buffer interface block with a single flexible struct array:
        layout(binding=0) readonly buffer vertices {
            sb_vertex vtx[];
        }
        // in the shader function, access the storage buffer like this:
        void main() {
            vec3 pos = vtx[gl_VertexIndex].pos;
            ...
        }
        @end
        ```

    In a compute shader you can read and write the same item in the same
    storage buffer (but you'll have to be careful for random access since
    many threads of the same compute function run in parallel):

        @cs
        struct sb_item {
            vec3 pos;
            vec3 vel;
        }
        layout(binding=0) buffer items_ssbo {
            sb_item items[];
        }
        layout(local_size_x=64, local_size_y=1, local_size_z=1) in;
        void main() {
            uint idx = gl_GlobalInvocationID.x;
            vec3 pos = items[idx].pos;
            ...
            items[idx].pos = pos;
        }
        @end

    Backend-specific storage-buffer caveats (not relevant when using sokol-shdc):

        D3D11:
            - storage buffers are created as 'raw' Byte Address Buffers
              (https://learn.microsoft.com/en-us/windows/win32/direct3d11/overviews-direct3d-11-resources-intro#raw-views-of-buffers)
            - in HLSL, use a ByteAddressBuffer for readonly access of the buffer content:
              (https://learn.microsoft.com/en-us/windows/win32/direct3dhlsl/sm5-object-byteaddressbuffer)
            - ...or RWByteAddressBuffer for read/write access:
              (https://learn.microsoft.com/en-us/windows/win32/direct3dhlsl/sm5-object-rwbyteaddressbuffer)
            - readonly-storage buffers and textures are both bound as 'shader-resource-view' and
              share the same bind slots (declared as `register(tN)` in HLSL), where N must be in the range 0..23)
            - read/write storage buffers and storage images are bound as 'unordered-access-view'
              (declared as `register(uN)` in HLSL where N is in the range 0..11)

        Metal:
            - in Metal there is no internal difference between vertex-, uniform- and
              storage-buffers, all are bound to the same 'buffer bind slots' with the
              following reserved ranges:
                - vertex shader stage:
                    - uniform buffers: slots 0..7
                    - storage buffers: slots 8..15
                    - vertex buffers: slots 15..23
                - fragment shader stage:
                    - uniform buffers: slots 0..7
                    - storage buffers: slots 8..15
            - this means in MSL, storage buffer bindings start at [[buffer(8)]] both in
              the vertex and fragment stage

        GL:
            - the GL backend doesn't use name-lookup to find storage buffer bindings, this
              means you must annotate buffers with `layout(std430, binding=N)` in GLSL
            - ...where N is 0..7 in the vertex shader, and 8..15 in the fragment shader

        WebGPU:
            - in WGSL, textures, samplers and storage buffers all use a shared
              bindspace across all shader stages on bindgroup 1:

              `@group(1) @binding(0..127)

    ON STORAGE IMAGES:
    ==================
    To write pixel data to texture objects in compute shaders, first an image
    object must be created with `storage_attachment usage`:

        sg_image storage_image = sg_make_image(&(sg_image_desc){
            .usage = {
                .storage_attachment = true,
            },
            .width = ...,
            .height = ...,
            .pixel_format = ...,
        });

    ...next the image object must be wrapped in an attachment object, this allows
    to pick a specific mipmap level or slice to be accessed by the compute shader:

        sg_attachments storage_attachment = sg_make_attachment(&(sg_attachments_desc){
            .storages[0] = {
                .image = storage_image,
                .mip_level = ...,
                .slice = ...,
            },
        });

    Finally 'bind' the storage image as pass attachment in the `sg_begin_pass`
    call of a compute pass:

        sg_begin_pass(&(sg_pass){ .compute = true, .attachments = storage_attachments });
        ...
        sg_end_pass();

    Storage attachments should only be accessed as `readwrite` or `writeonly` mode
    in compute shaders because if the limited bind space of up to 4 slots. For
    readonly access, just bind the storage image as regular texture via
    `sg_apply_bindings()`.

    For an example of using storage images in compute shaders see imageblur-sapp:

        - C code: https://github.com/floooh/sokol-samples/blob/master/sapp/imageblur-sapp.c
        - shader: https://github.com/floooh/sokol-samples/blob/master/sapp/imageblur-sapp.glsl

    NOTE: in the (hopefully not-too-distant) future, working with storage
    images will change by moving the resource binding from pass attachments to
    regular bindings via `sg_apply_bindings()`, but this requires the
    introduction of resource view objects into sokol-gfx (see planning
    ticket: https://github.com/floooh/sokol/issues/1252)

    TRACE HOOKS:
    ============
    sokol_gfx.h optionally allows to install "trace hook" callbacks for
    each public API functions. When a public API function is called, and
    a trace hook callback has been installed for this function, the
    callback will be invoked with the parameters and result of the function.
    This is useful for things like debugging- and profiling-tools, or
    keeping track of resource creation and destruction.

    To use the trace hook feature:

    --- Define SOKOL_TRACE_HOOKS before including the implementation.

    --- Setup an sg_trace_hooks structure with your callback function
        pointers (keep all function pointers you're not interested
        in zero-initialized), optionally set the user_data member
        in the sg_trace_hooks struct.

    --- Install the trace hooks by calling sg_install_trace_hooks(),
        the return value of this function is another sg_trace_hooks
        struct which contains the previously set of trace hooks.
        You should keep this struct around, and call those previous
        functions pointers from your own trace callbacks for proper
        chaining.

    As an example of how trace hooks are used, have a look at the
    imgui/sokol_gfx_imgui.h header which implements a realtime
    debugging UI for sokol_gfx.h on top of Dear ImGui.


    MEMORY ALLOCATION OVERRIDE
    ==========================
    You can override the memory allocation functions at initialization time
    like this:

        void* my_alloc(size_t size, void* user_data) {
            return malloc(size);
        }

        void my_free(void* ptr, void* user_data) {
            free(ptr);
        }

        ...
            sg_setup(&(sg_desc){
                // ...
                .allocator = {
                    .alloc_fn = my_alloc,
                    .free_fn = my_free,
                    .user_data = ...,
                }
            });
        ...

    If no overrides are provided, malloc and free will be used.

    This only affects memory allocation calls done by sokol_gfx.h
    itself though, not any allocations in OS libraries.


    ERROR REPORTING AND LOGGING
    ===========================
    To get any logging information at all you need to provide a logging callback in the setup call
    the easiest way is to use sokol_log.h:

        #include "sokol_log.h"

        sg_setup(&(sg_desc){ .logger.func = slog_func });

    To override logging with your own callback, first write a logging function like this:

        void my_log(const char* tag,                // e.g. 'sg'
                    uint32_t log_level,             // 0=panic, 1=error, 2=warn, 3=info
                    uint32_t log_item_id,           // SG_LOGITEM_*
                    const char* message_or_null,    // a message string, may be nullptr in release mode
                    uint32_t line_nr,               // line number in sokol_gfx.h
                    const char* filename_or_null,   // source filename, may be nullptr in release mode
                    void* user_data)
        {
            ...
        }

    ...and then setup sokol-gfx like this:

        sg_setup(&(sg_desc){
            .logger = {
                .func = my_log,
                .user_data = my_user_data,
            }
        });

    The provided logging function must be reentrant (e.g. be callable from
    different threads).

    If you don't want to provide your own custom logger it is highly recommended to use
    the standard logger in sokol_log.h instead, otherwise you won't see any warnings or
    errors.


    COMMIT LISTENERS
    ================
    It's possible to hook callback functions into sokol-gfx which are called from
    inside sg_commit() in unspecified order. This is mainly useful for libraries
    that build on top of sokol_gfx.h to be notified about the end/start of a frame.

    To add a commit listener, call:

        static void my_commit_listener(void* user_data) {
            ...
        }

        bool success = sg_add_commit_listener((sg_commit_listener){
            .func = my_commit_listener,
            .user_data = ...,
        });

    The function returns false if the internal array of commit listeners is full,
    or the same commit listener had already been added.

    If the function returns true, my_commit_listener() will be called each frame
    from inside sg_commit().

    By default, 1024 distinct commit listeners can be added, but this number
    can be tweaked in the sg_setup() call:

        sg_setup(&(sg_desc){
            .max_commit_listeners = 2048,
        });

    An sg_commit_listener item is equal to another if both the function
    pointer and user_data field are equal.

    To remove a commit listener:

        bool success = sg_remove_commit_listener((sg_commit_listener){
            .func = my_commit_listener,
            .user_data = ...,
        });

    ...where the .func and .user_data field are equal to a previous
    sg_add_commit_listener() call. The function returns true if the commit
    listener item was found and removed, and false otherwise.


    RESOURCE CREATION AND DESTRUCTION IN DETAIL
    ===========================================
    The 'vanilla' way to create resource objects is with the 'make functions':

        sg_buffer sg_make_buffer(const sg_buffer_desc* desc)
        sg_image sg_make_image(const sg_image_desc* desc)
        sg_sampler sg_make_sampler(const sg_sampler_desc* desc)
        sg_shader sg_make_shader(const sg_shader_desc* desc)
        sg_pipeline sg_make_pipeline(const sg_pipeline_desc* desc)
        sg_attachments sg_make_attachments(const sg_attachments_desc* desc)

    This will result in one of three cases:

        1. The returned handle is invalid. This happens when there are no more
           free slots in the resource pool for this resource type. An invalid
           handle is associated with the INVALID resource state, for instance:

                sg_buffer buf = sg_make_buffer(...)
                if (sg_query_buffer_state(buf) == SG_RESOURCESTATE_INVALID) {
                    // buffer pool is exhausted
                }

        2. The returned handle is valid, but creating the underlying resource
           has failed for some reason. This results in a resource object in the
           FAILED state. The reason *why* resource creation has failed differ
           by resource type. Look for log messages with more details. A failed
           resource state can be checked with:

                sg_buffer buf = sg_make_buffer(...)
                if (sg_query_buffer_state(buf) == SG_RESOURCESTATE_FAILED) {
                    // creating the resource has failed
                }

        3. And finally, if everything goes right, the returned resource is
           in resource state VALID and ready to use. This can be checked
           with:

                sg_buffer buf = sg_make_buffer(...)
                if (sg_query_buffer_state(buf) == SG_RESOURCESTATE_VALID) {
                    // creating the resource has failed
                }

    When calling the 'make functions', the created resource goes through a number
    of states:

        - INITIAL: the resource slot associated with the new resource is currently
          free (technically, there is no resource yet, just an empty pool slot)
        - ALLOC: a handle for the new resource has been allocated, this just means
          a pool slot has been reserved.
        - VALID or FAILED: in VALID state any 3D API backend resource objects have
          been successfully created, otherwise if anything went wrong, the resource
          will be in FAILED state.

    Sometimes it makes sense to first grab a handle, but initialize the
    underlying resource at a later time. For instance when loading data
    asynchronously from a slow data source, you may know what buffers and
    textures are needed at an early stage of the loading process, but actually
    loading the buffer or texture content can only be completed at a later time.

    For such situations, sokol-gfx resource objects can be created in two steps.
    You can allocate a handle upfront with one of the 'alloc functions':

        sg_buffer sg_alloc_buffer(void)
        sg_image sg_alloc_image(void)
        sg_sampler sg_alloc_sampler(void)
        sg_shader sg_alloc_shader(void)
        sg_pipeline sg_alloc_pipeline(void)
        sg_attachments sg_alloc_attachments(void)

    This will return a handle with the underlying resource object in the
    ALLOC state:

        sg_image img = sg_alloc_image();
        if (sg_query_image_state(img) == SG_RESOURCESTATE_ALLOC) {
            // allocating an image handle has succeeded, otherwise
            // the image pool is full
        }

    Such an 'incomplete' handle can be used in most sokol-gfx rendering functions
    without doing any harm, sokol-gfx will simply skip any rendering operation
    that involve resources which are not in VALID state.

    At a later time (for instance once the texture has completed loading
    asynchronously), the resource creation can be completed by calling one of
    the 'init functions', those functions take an existing resource handle and
    'desc struct':

        void sg_init_buffer(sg_buffer buf, const sg_buffer_desc* desc)
        void sg_init_image(sg_image img, const sg_image_desc* desc)
        void sg_init_sampler(sg_sampler smp, const sg_sampler_desc* desc)
        void sg_init_shader(sg_shader shd, const sg_shader_desc* desc)
        void sg_init_pipeline(sg_pipeline pip, const sg_pipeline_desc* desc)
        void sg_init_attachments(sg_attachments atts, const sg_attachments_desc* desc)

    The init functions expect a resource in ALLOC state, and after the function
    returns, the resource will be either in VALID or FAILED state. Calling
    an 'alloc function' followed by the matching 'init function' is fully
    equivalent with calling the 'make function' alone.

    Destruction can also happen as a two-step process. The 'uninit functions'
    will put a resource object from the VALID or FAILED state back into the
    ALLOC state:

        void sg_uninit_buffer(sg_buffer buf)
        void sg_uninit_image(sg_image img)
        void sg_uninit_sampler(sg_sampler smp)
        void sg_uninit_shader(sg_shader shd)
        void sg_uninit_pipeline(sg_pipeline pip)
        void sg_uninit_attachments(sg_attachments pass)

    Calling the 'uninit functions' with a resource that is not in the VALID or
    FAILED state is a no-op.

    To finally free the pool slot for recycling call the 'dealloc functions':

        void sg_dealloc_buffer(sg_buffer buf)
        void sg_dealloc_image(sg_image img)
        void sg_dealloc_sampler(sg_sampler smp)
        void sg_dealloc_shader(sg_shader shd)
        void sg_dealloc_pipeline(sg_pipeline pip)
        void sg_dealloc_attachments(sg_attachments atts)

    Calling the 'dealloc functions' on a resource that's not in ALLOC state is
    a no-op, but will generate a warning log message.

    Calling an 'uninit function' and 'dealloc function' in sequence is equivalent
    with calling the associated 'destroy function':

        void sg_destroy_buffer(sg_buffer buf)
        void sg_destroy_image(sg_image img)
        void sg_destroy_sampler(sg_sampler smp)
        void sg_destroy_shader(sg_shader shd)
        void sg_destroy_pipeline(sg_pipeline pip)
        void sg_destroy_attachments(sg_attachments atts)

    The 'destroy functions' can be called on resources in any state and generally
    do the right thing (for instance if the resource is in ALLOC state, the destroy
    function will be equivalent to the 'dealloc function' and skip the 'uninit part').

    And finally to close the circle, the 'fail functions' can be called to manually
    put a resource in ALLOC state into the FAILED state:

        sg_fail_buffer(sg_buffer buf)
        sg_fail_image(sg_image img)
        sg_fail_sampler(sg_sampler smp)
        sg_fail_shader(sg_shader shd)
        sg_fail_pipeline(sg_pipeline pip)
        sg_fail_attachments(sg_attachments atts)

    This is recommended if anything went wrong outside of sokol-gfx during asynchronous
    resource setup (for instance a file loading operation failed). In this case,
    the 'fail function' should be called instead of the 'init function'.

    Calling a 'fail function' on a resource that's not in ALLOC state is a no-op,
    but will generate a warning log message.

    NOTE: that two-step resource creation usually only makes sense for buffers
    and images, but not for samplers, shaders, pipelines or attachments. Most notably, trying
    to create a pipeline object with a shader that's not in VALID state will
    trigger a validation layer error, or if the validation layer is disabled,
    result in a pipeline object in FAILED state. Same when trying to create
    an attachments object with invalid image objects.


    WEBGPU CAVEATS
    ==============
    For a general overview and design notes of the WebGPU backend see:

        https://floooh.github.io/2023/10/16/sokol-webgpu.html

    In general, don't expect an automatic speedup when switching from the WebGL2
    backend to the WebGPU backend. Some WebGPU functions currently actually
    have a higher CPU overhead than similar WebGL2 functions, leading to the
    paradoxical situation that some WebGPU code may be slower than similar WebGL2
    code.

    - when writing WGSL shader code by hand, a specific bind-slot convention
      must be used:

      All uniform block structs must use `@group(0)` and bindings in the
      range 0..15

        @group(0) @binding(0..15)

      All textures, samplers and storage buffers must use `@group(1)` and
      bindings must be in the range 0..127:

        @group(1) @binding(0..127)

      All storage image attachments must use `@group(2)` and bindings
      must be in the range 0..3:

        @group(2) @binding(0..3)

      Note that the number of texture, sampler and storage buffer bindings
      is still limited despite the large bind range:

        - up to 16 textures and sampler across all shader stages
        - up to 8 storage buffers across all shader stages

      If you use sokol-shdc to generate WGSL shader code, you don't need to worry
      about the above binding conventions since sokol-shdc.

    - The sokol-gfx WebGPU backend uses the sg_desc.uniform_buffer_size item
      to allocate a single per-frame uniform buffer which must be big enough
      to hold all data written by sg_apply_uniforms() during a single frame,
      including a worst-case 256-byte alignment (e.g. each sg_apply_uniform
      call will cost at least 256 bytes of uniform buffer size). The default size
      is 4 MB, which is enough for 16384 sg_apply_uniform() calls per
      frame (assuming the uniform data 'payload' is less than 256 bytes
      per call). These rules are the same as for the Metal backend, so if
      you are already using the Metal backend you'll be fine.

    - sg_apply_bindings(): the sokol-gfx WebGPU backend implements a bindgroup
      cache to prevent excessive creation and destruction of BindGroup objects
      when calling sg_apply_bindings(). The number of slots in the bindgroups
      cache is defined in sg_desc.wgpu_bindgroups_cache_size when calling
      sg_setup. The cache size must be a power-of-2 number, with the default being
      1024. The bindgroups cache behaviour can be observed by calling the new
      function sg_query_frame_stats(), where the following struct items are
      of interest:

        .wgpu.num_bindgroup_cache_hits
        .wgpu.num_bindgroup_cache_misses
        .wgpu.num_bindgroup_cache_collisions
        .wgpu_num_bindgroup_cache_invalidates
        .wgpu.num_bindgroup_cache_vs_hash_key_mismatch

      The value to pay attention to is `.wgpu.num_bindgroup_cache_collisions`,
      if this number is consistently higher than a few percent of the
      .wgpu.num_set_bindgroup value, it might be a good idea to bump the
      bindgroups cache size to the next power-of-2.

    - sg_apply_viewport(): WebGPU currently has a unique restriction that viewport
      rectangles must be contained entirely within the framebuffer. As a shitty
      workaround sokol_gfx.h will clip incoming viewport rectangles against
      the framebuffer, but this will distort the clipspace-to-screenspace mapping.
      There's no proper way to handle this inside sokol_gfx.h, this must be fixed
      in a future WebGPU update (see: https://github.com/gpuweb/gpuweb/issues/373
      and https://github.com/gpuweb/gpuweb/pull/5025)

    - The sokol shader compiler generally adds `diagnostic(off, derivative_uniformity);`
      into the WGSL output. Currently only the Chrome WebGPU implementation seems
      to recognize this.

    - Likewise, the following sokol-gfx pixel formats are not supported in WebGPU:
      R16, R16SN, RG16, RG16SN, RGBA16, RGBA16SN.
      Unlike unsupported vertex formats, unsupported pixel formats can be queried
      in cross-backend code via sg_query_pixelformat() though.

    - The Emscripten WebGPU shim currently doesn't support the Closure minification
      post-link-step (e.g. currently the emcc argument '--closure 1' or '--closure 2'
      will generate broken Javascript code.

    - sokol-gfx requires the WebGPU device feature `depth32float-stencil8` to be enabled
      (this should be widely supported)

    - sokol-gfx expects that the WebGPU device feature `float32-filterable` to *not* be
      enabled (since this would exclude all iOS devices)


    LICENSE
    =======
    zlib/libpng license

    Copyright (c) 2018 Andre Weissflog

    This software is provided 'as-is', without any express or implied warranty.
    In no event will the authors be held liable for any damages arising from the
    use of this software.

    Permission is granted to anyone to use this software for any purpose,
    including commercial applications, and to alter it and redistribute it
    freely, subject to the following restrictions:

        1. The origin of this software must not be misrepresented; you must not
        claim that you wrote the original software. If you use this software in a
        product, an acknowledgment in the product documentation would be
        appreciated but is not required.

        2. Altered source versions must be plainly marked as such, and must not
        be misrepresented as being the original software.

        3. This notice may not be removed or altered from any source
        distribution.
*/
#define SOKOL_GFX_INCLUDED (1)
#include <stddef.h>     // size_t
#include <stdint.h>
#include <stdbool.h>

#if defined(SOKOL_API_DECL) && !defined(SOKOL_GFX_API_DECL)
#define SOKOL_GFX_API_DECL SOKOL_API_DECL
#endif
#ifndef SOKOL_GFX_API_DECL
#if defined(_WIN32) && defined(SOKOL_DLL) && defined(SOKOL_GFX_IMPL)
#define SOKOL_GFX_API_DECL __declspec(dllexport)
#elif defined(_WIN32) && defined(SOKOL_DLL)
#define SOKOL_GFX_API_DECL __declspec(dllimport)
#else
#define SOKOL_GFX_API_DECL extern
#endif
#endif

#ifdef __cplusplus
extern "C" {
#endif

/*
    Resource id typedefs:

    sg_buffer:      vertex- and index-buffers
    sg_image:       images used as textures and render-pass attachments
    sg_sampler      sampler objects describing how a texture is sampled in a shader
    sg_shader:      vertex- and fragment-shaders and shader interface information
    sg_pipeline:    associated shader and vertex-layouts, and render states
    sg_attachments: a baked collection of render pass attachment images

    Instead of pointers, resource creation functions return a 32-bit
    handle which uniquely identifies the resource object.

    The 32-bit resource id is split into a 16-bit pool index in the lower bits,
    and a 16-bit 'generation counter' in the upper bits. The index allows fast
    pool lookups, and combined with the generation-counter it allows to detect
    'dangling accesses' (trying to use an object which no longer exists, and
    its pool slot has been reused for a new object)

    The resource ids are wrapped into a strongly-typed struct so that
    trying to pass an incompatible resource id is a compile error.
*/
typedef struct sg_buffer        { uint32_t id; } sg_buffer;
typedef struct sg_image         { uint32_t id; } sg_image;
typedef struct sg_sampler       { uint32_t id; } sg_sampler;
typedef struct sg_shader        { uint32_t id; } sg_shader;
typedef struct sg_pipeline      { uint32_t id; } sg_pipeline;
typedef struct sg_attachments   { uint32_t id; } sg_attachments;

/*
    sg_range is a pointer-size-pair struct used to pass memory blobs into
    sokol-gfx. When initialized from a value type (array or struct), you can
    use the SG_RANGE() macro to build an sg_range struct. For functions which
    take either a sg_range pointer, or a (C++) sg_range reference, use the
    SG_RANGE_REF macro as a solution which compiles both in C and C++.
*/
typedef struct sg_range {
    const void* ptr;
    size_t size;
} sg_range;

// disabling this for every includer isn't great, but the warnings are also quite pointless
#if defined(_MSC_VER)
#pragma warning(disable:4221)   // /W4 only: nonstandard extension used: 'x': cannot be initialized using address of automatic variable 'y'
#pragma warning(disable:4204)   // VS2015: nonstandard extension used: non-constant aggregate initializer
#endif
#if defined(__cplusplus)
#define SG_RANGE(x) sg_range{ &x, sizeof(x) }
#define SG_RANGE_REF(x) sg_range{ &x, sizeof(x) }
#else
#define SG_RANGE(x) (sg_range){ &x, sizeof(x) }
#define SG_RANGE_REF(x) &(sg_range){ &x, sizeof(x) }
#endif

// various compile-time constants in the public API
enum {
    SG_INVALID_ID = 0,
    SG_NUM_INFLIGHT_FRAMES = 2,
    SG_MAX_COLOR_ATTACHMENTS = 4,
    SG_MAX_STORAGE_ATTACHMENTS = 4,
    SG_MAX_UNIFORMBLOCK_MEMBERS = 16,
    SG_MAX_VERTEX_ATTRIBUTES = 16,
    SG_MAX_MIPMAPS = 16,
    SG_MAX_TEXTUREARRAY_LAYERS = 128,
    SG_MAX_UNIFORMBLOCK_BINDSLOTS = 8,
    SG_MAX_VERTEXBUFFER_BINDSLOTS = 8,
    SG_MAX_IMAGE_BINDSLOTS = 16,
    SG_MAX_SAMPLER_BINDSLOTS = 16,
    SG_MAX_STORAGEBUFFER_BINDSLOTS = 8,
    SG_MAX_IMAGE_SAMPLER_PAIRS = 16,
};

/*
    sg_color

    An RGBA color value.
*/
typedef struct sg_color { float r, g, b, a; } sg_color;

/*
    sg_backend

    The active 3D-API backend, use the function sg_query_backend()
    to get the currently active backend.
*/
typedef enum sg_backend {
    SG_BACKEND_GLCORE,
    SG_BACKEND_GLES3,
    SG_BACKEND_D3D11,
    SG_BACKEND_METAL_IOS,
    SG_BACKEND_METAL_MACOS,
    SG_BACKEND_METAL_SIMULATOR,
    SG_BACKEND_WGPU,
    SG_BACKEND_DUMMY,
} sg_backend;

/*
    sg_pixel_format

    sokol_gfx.h basically uses the same pixel formats as WebGPU, since these
    are supported on most newer GPUs.

    A pixelformat name consist of three parts:

        - components (R, RG, RGB or RGBA)
        - bit width per component (8, 16 or 32)
        - component data type:
            - unsigned normalized (no postfix)
            - signed normalized (SN postfix)
            - unsigned integer (UI postfix)
            - signed integer (SI postfix)
            - float (F postfix)

    Not all pixel formats can be used for everything, call sg_query_pixelformat()
    to inspect the capabilities of a given pixelformat. The function returns
    an sg_pixelformat_info struct with the following members:

        - sample: the pixelformat can be sampled as texture at least with
                  nearest filtering
        - filter: the pixelformat can be sampled as texture with linear
                  filtering
        - render: the pixelformat can be used as render-pass attachment
        - blend:  blending is supported when used as render-pass attachment
        - msaa:   multisample-antialiasing is supported when used
                  as render-pass attachment
        - depth:  the pixelformat can be used for depth-stencil attachments
        - compressed: this is a block-compressed format
        - bytes_per_pixel: the numbers of bytes in a pixel (0 for compressed formats)

    The default pixel format for texture images is SG_PIXELFORMAT_RGBA8.

    The default pixel format for render target images is platform-dependent
    and taken from the sg_environment struct passed into sg_setup(). Typically
    the default formats are:

        - for the Metal, D3D11 and WebGPU backends: SG_PIXELFORMAT_BGRA8
        - for GL backends: SG_PIXELFORMAT_RGBA8
*/
typedef enum sg_pixel_format {
    _SG_PIXELFORMAT_DEFAULT,    // value 0 reserved for default-init
    SG_PIXELFORMAT_NONE,

    SG_PIXELFORMAT_R8,
    SG_PIXELFORMAT_R8SN,
    SG_PIXELFORMAT_R8UI,
    SG_PIXELFORMAT_R8SI,

    SG_PIXELFORMAT_R16,
    SG_PIXELFORMAT_R16SN,
    SG_PIXELFORMAT_R16UI,
    SG_PIXELFORMAT_R16SI,
    SG_PIXELFORMAT_R16F,
    SG_PIXELFORMAT_RG8,
    SG_PIXELFORMAT_RG8SN,
    SG_PIXELFORMAT_RG8UI,
    SG_PIXELFORMAT_RG8SI,

    SG_PIXELFORMAT_R32UI,
    SG_PIXELFORMAT_R32SI,
    SG_PIXELFORMAT_R32F,
    SG_PIXELFORMAT_RG16,
    SG_PIXELFORMAT_RG16SN,
    SG_PIXELFORMAT_RG16UI,
    SG_PIXELFORMAT_RG16SI,
    SG_PIXELFORMAT_RG16F,
    SG_PIXELFORMAT_RGBA8,
    SG_PIXELFORMAT_SRGB8A8,
    SG_PIXELFORMAT_RGBA8SN,
    SG_PIXELFORMAT_RGBA8UI,
    SG_PIXELFORMAT_RGBA8SI,
    SG_PIXELFORMAT_BGRA8,
    SG_PIXELFORMAT_RGB10A2,
    SG_PIXELFORMAT_RG11B10F,
    SG_PIXELFORMAT_RGB9E5,

    SG_PIXELFORMAT_RG32UI,
    SG_PIXELFORMAT_RG32SI,
    SG_PIXELFORMAT_RG32F,
    SG_PIXELFORMAT_RGBA16,
    SG_PIXELFORMAT_RGBA16SN,
    SG_PIXELFORMAT_RGBA16UI,
    SG_PIXELFORMAT_RGBA16SI,
    SG_PIXELFORMAT_RGBA16F,

    SG_PIXELFORMAT_RGBA32UI,
    SG_PIXELFORMAT_RGBA32SI,
    SG_PIXELFORMAT_RGBA32F,

    // NOTE: when adding/removing pixel formats before DEPTH, also update sokol_app.h/_SAPP_PIXELFORMAT_*
    SG_PIXELFORMAT_DEPTH,
    SG_PIXELFORMAT_DEPTH_STENCIL,

    // NOTE: don't put any new compressed format in front of here
    SG_PIXELFORMAT_BC1_RGBA,
    SG_PIXELFORMAT_BC2_RGBA,
    SG_PIXELFORMAT_BC3_RGBA,
    SG_PIXELFORMAT_BC3_SRGBA,
    SG_PIXELFORMAT_BC4_R,
    SG_PIXELFORMAT_BC4_RSN,
    SG_PIXELFORMAT_BC5_RG,
    SG_PIXELFORMAT_BC5_RGSN,
    SG_PIXELFORMAT_BC6H_RGBF,
    SG_PIXELFORMAT_BC6H_RGBUF,
    SG_PIXELFORMAT_BC7_RGBA,
    SG_PIXELFORMAT_BC7_SRGBA,
    SG_PIXELFORMAT_ETC2_RGB8,
    SG_PIXELFORMAT_ETC2_SRGB8,
    SG_PIXELFORMAT_ETC2_RGB8A1,
    SG_PIXELFORMAT_ETC2_RGBA8,
    SG_PIXELFORMAT_ETC2_SRGB8A8,
    SG_PIXELFORMAT_EAC_R11,
    SG_PIXELFORMAT_EAC_R11SN,
    SG_PIXELFORMAT_EAC_RG11,
    SG_PIXELFORMAT_EAC_RG11SN,

    SG_PIXELFORMAT_ASTC_4x4_RGBA,
    SG_PIXELFORMAT_ASTC_4x4_SRGBA,

    _SG_PIXELFORMAT_NUM,
    _SG_PIXELFORMAT_FORCE_U32 = 0x7FFFFFFF
} sg_pixel_format;

/*
    Runtime information about a pixel format, returned by sg_query_pixelformat().
*/
typedef struct sg_pixelformat_info {
    bool sample;            // pixel format can be sampled in shaders at least with nearest filtering
    bool filter;            // pixel format can be sampled with linear filtering
    bool render;            // pixel format can be used as render-pass attachment
    bool blend;             // pixel format supports alpha-blending when used as render-pass attachment
    bool msaa;              // pixel format supports MSAA when used as render-pass attachment
    bool depth;             // pixel format is a depth format
    bool compressed;        // true if this is a hardware-compressed format
    bool read;              // true if format supports compute shader read access
    bool write;             // true if format supports compute shader write access
    int bytes_per_pixel;    // NOTE: this is 0 for compressed formats, use sg_query_row_pitch() / sg_query_surface_pitch() as alternative
} sg_pixelformat_info;

/*
    Runtime information about available optional features, returned by sg_query_features()
*/
typedef struct sg_features {
    bool origin_top_left;               // framebuffer- and texture-origin is in top left corner
    bool image_clamp_to_border;         // border color and clamp-to-border uv-wrap mode is supported
    bool mrt_independent_blend_state;   // multiple-render-target rendering can use per-render-target blend state
    bool mrt_independent_write_mask;    // multiple-render-target rendering can use per-render-target color write masks
    bool compute;                       // storage buffers and compute shaders are supported
    bool msaa_image_bindings;           // if true, multisampled images can be bound as texture resources
    bool separate_buffer_types;         // cannot use the same buffer for vertex and indices (onlu WebGL2)
} sg_features;

/*
    Runtime information about resource limits, returned by sg_query_limit()
*/
typedef struct sg_limits {
    int max_image_size_2d;          // max width/height of SG_IMAGETYPE_2D images
    int max_image_size_cube;        // max width/height of SG_IMAGETYPE_CUBE images
    int max_image_size_3d;          // max width/height/depth of SG_IMAGETYPE_3D images
    int max_image_size_array;       // max width/height of SG_IMAGETYPE_ARRAY images
    int max_image_array_layers;     // max number of layers in SG_IMAGETYPE_ARRAY images
    int max_vertex_attrs;           // max number of vertex attributes, clamped to SG_MAX_VERTEX_ATTRIBUTES
    int gl_max_vertex_uniform_components;    // <= GL_MAX_VERTEX_UNIFORM_COMPONENTS (only on GL backends)
    int gl_max_combined_texture_image_units; // <= GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS (only on GL backends)
} sg_limits;

/*
    sg_resource_state

    The current state of a resource in its resource pool.
    Resources start in the INITIAL state, which means the
    pool slot is unoccupied and can be allocated. When a resource is
    created, first an id is allocated, and the resource pool slot
    is set to state ALLOC. After allocation, the resource is
    initialized, which may result in the VALID or FAILED state. The
    reason why allocation and initialization are separate is because
    some resource types (e.g. buffers and images) might be asynchronously
    initialized by the user application. If a resource which is not
    in the VALID state is attempted to be used for rendering, rendering
    operations will silently be dropped.

    The special INVALID state is returned in sg_query_xxx_state() if no
    resource object exists for the provided resource id.
*/
typedef enum sg_resource_state {
    SG_RESOURCESTATE_INITIAL,
    SG_RESOURCESTATE_ALLOC,
    SG_RESOURCESTATE_VALID,
    SG_RESOURCESTATE_FAILED,
    SG_RESOURCESTATE_INVALID,
    _SG_RESOURCESTATE_FORCE_U32 = 0x7FFFFFFF
} sg_resource_state;

/*
    sg_index_type

    Indicates whether indexed rendering (fetching vertex-indices from an
    index buffer) is used, and if yes, the index data type (16- or 32-bits).

    This is used in the sg_pipeline_desc.index_type member when creating a
    pipeline object.

    The default index type is SG_INDEXTYPE_NONE.
*/
typedef enum sg_index_type {
    _SG_INDEXTYPE_DEFAULT,   // value 0 reserved for default-init
    SG_INDEXTYPE_NONE,
    SG_INDEXTYPE_UINT16,
    SG_INDEXTYPE_UINT32,
    _SG_INDEXTYPE_NUM,
    _SG_INDEXTYPE_FORCE_U32 = 0x7FFFFFFF
} sg_index_type;

/*
    sg_image_type

    Indicates the basic type of an image object (2D-texture, cubemap,
    3D-texture or 2D-array-texture). Used in the sg_image_desc.type member when
    creating an image, and in sg_shader_image_desc to describe a sampled texture
    in the shader (both must match and will be checked in the validation layer
    when calling sg_apply_bindings).

    The default image type when creating an image is SG_IMAGETYPE_2D.
*/
typedef enum sg_image_type {
    _SG_IMAGETYPE_DEFAULT,  // value 0 reserved for default-init
    SG_IMAGETYPE_2D,
    SG_IMAGETYPE_CUBE,
    SG_IMAGETYPE_3D,
    SG_IMAGETYPE_ARRAY,
    _SG_IMAGETYPE_NUM,
    _SG_IMAGETYPE_FORCE_U32 = 0x7FFFFFFF
} sg_image_type;

/*
    sg_image_sample_type

    The basic data type of a texture sample as expected by a shader.
    Must be provided in sg_shader_image and used by the validation
    layer in sg_apply_bindings() to check if the provided image object
    is compatible with what the shader expects. Apart from the sokol-gfx
    validation layer, WebGPU is the only backend API which actually requires
    matching texture and sampler type to be provided upfront for validation
    (other 3D APIs treat texture/sampler type mismatches as undefined behaviour).

    NOTE that the following texture pixel formats require the use
    of SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT, combined with a sampler
    of type SG_SAMPLERTYPE_NONFILTERING:

    - SG_PIXELFORMAT_R32F
    - SG_PIXELFORMAT_RG32F
    - SG_PIXELFORMAT_RGBA32F

    (when using sokol-shdc, also check out the meta tags `@image_sample_type`
    and `@sampler_type`)
*/
typedef enum sg_image_sample_type {
    _SG_IMAGESAMPLETYPE_DEFAULT,  // value 0 reserved for default-init
    SG_IMAGESAMPLETYPE_FLOAT,
    SG_IMAGESAMPLETYPE_DEPTH,
    SG_IMAGESAMPLETYPE_SINT,
    SG_IMAGESAMPLETYPE_UINT,
    SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT,
    _SG_IMAGESAMPLETYPE_NUM,
    _SG_IMAGESAMPLETYPE_FORCE_U32 = 0x7FFFFFFF
} sg_image_sample_type;

/*
    sg_sampler_type

    The basic type of a texture sampler (sampling vs comparison) as
    defined in a shader. Must be provided in sg_shader_sampler_desc.

    sg_image_sample_type and sg_sampler_type for a texture/sampler
    pair must be compatible with each other, specifically only
    the following pairs are allowed:

    - SG_IMAGESAMPLETYPE_FLOAT => (SG_SAMPLERTYPE_FILTERING or SG_SAMPLERTYPE_NONFILTERING)
    - SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT => SG_SAMPLERTYPE_NONFILTERING
    - SG_IMAGESAMPLETYPE_SINT => SG_SAMPLERTYPE_NONFILTERING
    - SG_IMAGESAMPLETYPE_UINT => SG_SAMPLERTYPE_NONFILTERING
    - SG_IMAGESAMPLETYPE_DEPTH => SG_SAMPLERTYPE_COMPARISON
*/
typedef enum sg_sampler_type {
    _SG_SAMPLERTYPE_DEFAULT,
    SG_SAMPLERTYPE_FILTERING,
    SG_SAMPLERTYPE_NONFILTERING,
    SG_SAMPLERTYPE_COMPARISON,
    _SG_SAMPLERTYPE_NUM,
    _SG_SAMPLERTYPE_FORCE_U32,
} sg_sampler_type;

/*
    sg_cube_face

    The cubemap faces. Use these as indices in the sg_image_desc.content
    array.
*/
typedef enum sg_cube_face {
    SG_CUBEFACE_POS_X,
    SG_CUBEFACE_NEG_X,
    SG_CUBEFACE_POS_Y,
    SG_CUBEFACE_NEG_Y,
    SG_CUBEFACE_POS_Z,
    SG_CUBEFACE_NEG_Z,
    SG_CUBEFACE_NUM,
    _SG_CUBEFACE_FORCE_U32 = 0x7FFFFFFF
} sg_cube_face;

/*
    sg_primitive_type

    This is the common subset of 3D primitive types supported across all 3D
    APIs. This is used in the sg_pipeline_desc.primitive_type member when
    creating a pipeline object.

    The default primitive type is SG_PRIMITIVETYPE_TRIANGLES.
*/
typedef enum sg_primitive_type {
    _SG_PRIMITIVETYPE_DEFAULT,  // value 0 reserved for default-init
    SG_PRIMITIVETYPE_POINTS,
    SG_PRIMITIVETYPE_LINES,
    SG_PRIMITIVETYPE_LINE_STRIP,
    SG_PRIMITIVETYPE_TRIANGLES,
    SG_PRIMITIVETYPE_TRIANGLE_STRIP,
    _SG_PRIMITIVETYPE_NUM,
    _SG_PRIMITIVETYPE_FORCE_U32 = 0x7FFFFFFF
} sg_primitive_type;

/*
    sg_filter

    The filtering mode when sampling a texture image. This is
    used in the sg_sampler_desc.min_filter, sg_sampler_desc.mag_filter
    and sg_sampler_desc.mipmap_filter members when creating a sampler object.

    For the default is SG_FILTER_NEAREST.
*/
typedef enum sg_filter {
    _SG_FILTER_DEFAULT, // value 0 reserved for default-init
    SG_FILTER_NEAREST,
    SG_FILTER_LINEAR,
    _SG_FILTER_NUM,
    _SG_FILTER_FORCE_U32 = 0x7FFFFFFF
} sg_filter;

/*
    sg_wrap

    The texture coordinates wrapping mode when sampling a texture
    image. This is used in the sg_image_desc.wrap_u, .wrap_v
    and .wrap_w members when creating an image.

    The default wrap mode is SG_WRAP_REPEAT.

    NOTE: SG_WRAP_CLAMP_TO_BORDER is not supported on all backends
    and platforms. To check for support, call sg_query_features()
    and check the "clamp_to_border" boolean in the returned
    sg_features struct.

    Platforms which don't support SG_WRAP_CLAMP_TO_BORDER will silently fall back
    to SG_WRAP_CLAMP_TO_EDGE without a validation error.
*/
typedef enum sg_wrap {
    _SG_WRAP_DEFAULT,   // value 0 reserved for default-init
    SG_WRAP_REPEAT,
    SG_WRAP_CLAMP_TO_EDGE,
    SG_WRAP_CLAMP_TO_BORDER,
    SG_WRAP_MIRRORED_REPEAT,
    _SG_WRAP_NUM,
    _SG_WRAP_FORCE_U32 = 0x7FFFFFFF
} sg_wrap;

/*
    sg_border_color

    The border color to use when sampling a texture, and the UV wrap
    mode is SG_WRAP_CLAMP_TO_BORDER.

    The default border color is SG_BORDERCOLOR_OPAQUE_BLACK
*/
typedef enum sg_border_color {
    _SG_BORDERCOLOR_DEFAULT,    // value 0 reserved for default-init
    SG_BORDERCOLOR_TRANSPARENT_BLACK,
    SG_BORDERCOLOR_OPAQUE_BLACK,
    SG_BORDERCOLOR_OPAQUE_WHITE,
    _SG_BORDERCOLOR_NUM,
    _SG_BORDERCOLOR_FORCE_U32 = 0x7FFFFFFF
} sg_border_color;

/*
    sg_vertex_format

    The data type of a vertex component. This is used to describe
    the layout of input vertex data when creating a pipeline object.

    NOTE that specific mapping rules exist from the CPU-side vertex
    formats to the vertex attribute base type in the vertex shader code
    (see doc header section 'ON VERTEX FORMATS').
*/
typedef enum sg_vertex_format {
    SG_VERTEXFORMAT_INVALID,
    SG_VERTEXFORMAT_FLOAT,
    SG_VERTEXFORMAT_FLOAT2,
    SG_VERTEXFORMAT_FLOAT3,
    SG_VERTEXFORMAT_FLOAT4,
    SG_VERTEXFORMAT_INT,
    SG_VERTEXFORMAT_INT2,
    SG_VERTEXFORMAT_INT3,
    SG_VERTEXFORMAT_INT4,
    SG_VERTEXFORMAT_UINT,
    SG_VERTEXFORMAT_UINT2,
    SG_VERTEXFORMAT_UINT3,
    SG_VERTEXFORMAT_UINT4,
    SG_VERTEXFORMAT_BYTE4,
    SG_VERTEXFORMAT_BYTE4N,
    SG_VERTEXFORMAT_UBYTE4,
    SG_VERTEXFORMAT_UBYTE4N,
    SG_VERTEXFORMAT_SHORT2,
    SG_VERTEXFORMAT_SHORT2N,
    SG_VERTEXFORMAT_USHORT2,
    SG_VERTEXFORMAT_USHORT2N,
    SG_VERTEXFORMAT_SHORT4,
    SG_VERTEXFORMAT_SHORT4N,
    SG_VERTEXFORMAT_USHORT4,
    SG_VERTEXFORMAT_USHORT4N,
    SG_VERTEXFORMAT_UINT10_N2,
    SG_VERTEXFORMAT_HALF2,
    SG_VERTEXFORMAT_HALF4,
    _SG_VERTEXFORMAT_NUM,
    _SG_VERTEXFORMAT_FORCE_U32 = 0x7FFFFFFF
} sg_vertex_format;

/*
    sg_vertex_step

    Defines whether the input pointer of a vertex input stream is advanced
    'per vertex' or 'per instance'. The default step-func is
    SG_VERTEXSTEP_PER_VERTEX. SG_VERTEXSTEP_PER_INSTANCE is used with
    instanced-rendering.

    The vertex-step is part of the vertex-layout definition
    when creating pipeline objects.
*/
typedef enum sg_vertex_step {
    _SG_VERTEXSTEP_DEFAULT,     // value 0 reserved for default-init
    SG_VERTEXSTEP_PER_VERTEX,
    SG_VERTEXSTEP_PER_INSTANCE,
    _SG_VERTEXSTEP_NUM,
    _SG_VERTEXSTEP_FORCE_U32 = 0x7FFFFFFF
} sg_vertex_step;

/*
    sg_uniform_type

    The data type of a uniform block member. This is used to
    describe the internal layout of uniform blocks when creating
    a shader object. This is only required for the GL backend, all
    other backends will ignore the interior layout of uniform blocks.
*/
typedef enum sg_uniform_type {
    SG_UNIFORMTYPE_INVALID,
    SG_UNIFORMTYPE_FLOAT,
    SG_UNIFORMTYPE_FLOAT2,
    SG_UNIFORMTYPE_FLOAT3,
    SG_UNIFORMTYPE_FLOAT4,
    SG_UNIFORMTYPE_INT,
    SG_UNIFORMTYPE_INT2,
    SG_UNIFORMTYPE_INT3,
    SG_UNIFORMTYPE_INT4,
    SG_UNIFORMTYPE_MAT4,
    _SG_UNIFORMTYPE_NUM,
    _SG_UNIFORMTYPE_FORCE_U32 = 0x7FFFFFFF
} sg_uniform_type;

/*
    sg_uniform_layout

    A hint for the interior memory layout of uniform blocks. This is
    only relevant for the GL backend where the internal layout
    of uniform blocks must be known to sokol-gfx. For all other backends the
    internal memory layout of uniform blocks doesn't matter, sokol-gfx
    will just pass uniform data as an opaque memory blob to the
    3D backend.

    SG_UNIFORMLAYOUT_NATIVE (default)
        Native layout means that a 'backend-native' memory layout
        is used. For the GL backend this means that uniforms
        are packed tightly in memory (e.g. there are no padding
        bytes).

    SG_UNIFORMLAYOUT_STD140
        The memory layout is a subset of std140. Arrays are only
        allowed for the FLOAT4, INT4 and MAT4. Alignment is as
        is as follows:

            FLOAT, INT:         4 byte alignment
            FLOAT2, INT2:       8 byte alignment
            FLOAT3, INT3:       16 byte alignment(!)
            FLOAT4, INT4:       16 byte alignment
            MAT4:               16 byte alignment
            FLOAT4[], INT4[]:   16 byte alignment

        The overall size of the uniform block must be a multiple
        of 16.

    For more information search for 'UNIFORM DATA LAYOUT' in the documentation block
    at the start of the header.
*/
typedef enum sg_uniform_layout {
    _SG_UNIFORMLAYOUT_DEFAULT,     // value 0 reserved for default-init
    SG_UNIFORMLAYOUT_NATIVE,       // default: layout depends on currently active backend
    SG_UNIFORMLAYOUT_STD140,       // std140: memory layout according to std140
    _SG_UNIFORMLAYOUT_NUM,
    _SG_UNIFORMLAYOUT_FORCE_U32 = 0x7FFFFFFF
} sg_uniform_layout;

/*
    sg_cull_mode

    The face-culling mode, this is used in the
    sg_pipeline_desc.cull_mode member when creating a
    pipeline object.

    The default cull mode is SG_CULLMODE_NONE
*/
typedef enum sg_cull_mode {
    _SG_CULLMODE_DEFAULT,   // value 0 reserved for default-init
    SG_CULLMODE_NONE,
    SG_CULLMODE_FRONT,
    SG_CULLMODE_BACK,
    _SG_CULLMODE_NUM,
    _SG_CULLMODE_FORCE_U32 = 0x7FFFFFFF
} sg_cull_mode;

/*
    sg_face_winding

    The vertex-winding rule that determines a front-facing primitive. This
    is used in the member sg_pipeline_desc.face_winding
    when creating a pipeline object.

    The default winding is SG_FACEWINDING_CW (clockwise)
*/
typedef enum sg_face_winding {
    _SG_FACEWINDING_DEFAULT,    // value 0 reserved for default-init
    SG_FACEWINDING_CCW,
    SG_FACEWINDING_CW,
    _SG_FACEWINDING_NUM,
    _SG_FACEWINDING_FORCE_U32 = 0x7FFFFFFF
} sg_face_winding;

/*
    sg_compare_func

    The compare-function for configuring depth- and stencil-ref tests
    in pipeline objects, and for texture samplers which perform a comparison
    instead of regular sampling operation.

    Used in the following structs:

    sg_pipeline_desc
        .depth
            .compare
        .stencil
            .front.compare
            .back.compare

    sg_sampler_desc
        .compare

    The default compare func for depth- and stencil-tests is
    SG_COMPAREFUNC_ALWAYS.

    The default compare func for samplers is SG_COMPAREFUNC_NEVER.
*/
typedef enum sg_compare_func {
    _SG_COMPAREFUNC_DEFAULT,    // value 0 reserved for default-init
    SG_COMPAREFUNC_NEVER,
    SG_COMPAREFUNC_LESS,
    SG_COMPAREFUNC_EQUAL,
    SG_COMPAREFUNC_LESS_EQUAL,
    SG_COMPAREFUNC_GREATER,
    SG_COMPAREFUNC_NOT_EQUAL,
    SG_COMPAREFUNC_GREATER_EQUAL,
    SG_COMPAREFUNC_ALWAYS,
    _SG_COMPAREFUNC_NUM,
    _SG_COMPAREFUNC_FORCE_U32 = 0x7FFFFFFF
} sg_compare_func;

/*
    sg_stencil_op

    The operation performed on a currently stored stencil-value when a
    comparison test passes or fails. This is used when creating a pipeline
    object in the following sg_pipeline_desc struct items:

    sg_pipeline_desc
        .stencil
            .front
                .fail_op
                .depth_fail_op
                .pass_op
            .back
                .fail_op
                .depth_fail_op
                .pass_op

    The default value is SG_STENCILOP_KEEP.
*/
typedef enum sg_stencil_op {
    _SG_STENCILOP_DEFAULT,      // value 0 reserved for default-init
    SG_STENCILOP_KEEP,
    SG_STENCILOP_ZERO,
    SG_STENCILOP_REPLACE,
    SG_STENCILOP_INCR_CLAMP,
    SG_STENCILOP_DECR_CLAMP,
    SG_STENCILOP_INVERT,
    SG_STENCILOP_INCR_WRAP,
    SG_STENCILOP_DECR_WRAP,
    _SG_STENCILOP_NUM,
    _SG_STENCILOP_FORCE_U32 = 0x7FFFFFFF
} sg_stencil_op;

/*
    sg_blend_factor

    The source and destination factors in blending operations.
    This is used in the following members when creating a pipeline object:

    sg_pipeline_desc
        .colors[i]
            .blend
                .src_factor_rgb
                .dst_factor_rgb
                .src_factor_alpha
                .dst_factor_alpha

    The default value is SG_BLENDFACTOR_ONE for source
    factors, and for the destination SG_BLENDFACTOR_ZERO if the associated
    blend-op is ADD, SUBTRACT or REVERSE_SUBTRACT or SG_BLENDFACTOR_ONE
    if the associated blend-op is MIN or MAX.
*/
typedef enum sg_blend_factor {
    _SG_BLENDFACTOR_DEFAULT,    // value 0 reserved for default-init
    SG_BLENDFACTOR_ZERO,
    SG_BLENDFACTOR_ONE,
    SG_BLENDFACTOR_SRC_COLOR,
    SG_BLENDFACTOR_ONE_MINUS_SRC_COLOR,
    SG_BLENDFACTOR_SRC_ALPHA,
    SG_BLENDFACTOR_ONE_MINUS_SRC_ALPHA,
    SG_BLENDFACTOR_DST_COLOR,
    SG_BLENDFACTOR_ONE_MINUS_DST_COLOR,
    SG_BLENDFACTOR_DST_ALPHA,
    SG_BLENDFACTOR_ONE_MINUS_DST_ALPHA,
    SG_BLENDFACTOR_SRC_ALPHA_SATURATED,
    SG_BLENDFACTOR_BLEND_COLOR,
    SG_BLENDFACTOR_ONE_MINUS_BLEND_COLOR,
    SG_BLENDFACTOR_BLEND_ALPHA,
    SG_BLENDFACTOR_ONE_MINUS_BLEND_ALPHA,
    _SG_BLENDFACTOR_NUM,
    _SG_BLENDFACTOR_FORCE_U32 = 0x7FFFFFFF
} sg_blend_factor;

/*
    sg_blend_op

    Describes how the source and destination values are combined in the
    fragment blending operation. It is used in the following struct items
    when creating a pipeline object:

    sg_pipeline_desc
        .colors[i]
            .blend
                .op_rgb
                .op_alpha

    The default value is SG_BLENDOP_ADD.
*/
typedef enum sg_blend_op {
    _SG_BLENDOP_DEFAULT,    // value 0 reserved for default-init
    SG_BLENDOP_ADD,
    SG_BLENDOP_SUBTRACT,
    SG_BLENDOP_REVERSE_SUBTRACT,
    SG_BLENDOP_MIN,
    SG_BLENDOP_MAX,
    _SG_BLENDOP_NUM,
    _SG_BLENDOP_FORCE_U32 = 0x7FFFFFFF
} sg_blend_op;

/*
    sg_color_mask

    Selects the active color channels when writing a fragment color to the
    framebuffer. This is used in the members
    sg_pipeline_desc.colors[i].write_mask when creating a pipeline object.

    The default colormask is SG_COLORMASK_RGBA (write all colors channels)

    NOTE: since the color mask value 0 is reserved for the default value
    (SG_COLORMASK_RGBA), use SG_COLORMASK_NONE if all color channels
    should be disabled.
*/
typedef enum sg_color_mask {
    _SG_COLORMASK_DEFAULT = 0,    // value 0 reserved for default-init
    SG_COLORMASK_NONE   = 0x10,   // special value for 'all channels disabled
    SG_COLORMASK_R      = 0x1,
    SG_COLORMASK_G      = 0x2,
    SG_COLORMASK_RG     = 0x3,
    SG_COLORMASK_B      = 0x4,
    SG_COLORMASK_RB     = 0x5,
    SG_COLORMASK_GB     = 0x6,
    SG_COLORMASK_RGB    = 0x7,
    SG_COLORMASK_A      = 0x8,
    SG_COLORMASK_RA     = 0x9,
    SG_COLORMASK_GA     = 0xA,
    SG_COLORMASK_RGA    = 0xB,
    SG_COLORMASK_BA     = 0xC,
    SG_COLORMASK_RBA    = 0xD,
    SG_COLORMASK_GBA    = 0xE,
    SG_COLORMASK_RGBA   = 0xF,
    _SG_COLORMASK_FORCE_U32 = 0x7FFFFFFF
} sg_color_mask;

/*
    sg_load_action

    Defines the load action that should be performed at the start of a render pass:

    SG_LOADACTION_CLEAR:        clear the render target
    SG_LOADACTION_LOAD:         load the previous content of the render target
    SG_LOADACTION_DONTCARE:     leave the render target in an undefined state

    This is used in the sg_pass_action structure.

    The default load action for all pass attachments is SG_LOADACTION_CLEAR,
    with the values rgba = { 0.5f, 0.5f, 0.5f, 1.0f }, depth=1.0f and stencil=0.

    If you want to override the default behaviour, it is important to not
    only set the clear color, but the 'action' field as well (as long as this
    is _SG_LOADACTION_DEFAULT, the value fields will be ignored).
*/
typedef enum sg_load_action {
    _SG_LOADACTION_DEFAULT,
    SG_LOADACTION_CLEAR,
    SG_LOADACTION_LOAD,
    SG_LOADACTION_DONTCARE,
    _SG_LOADACTION_FORCE_U32 = 0x7FFFFFFF
} sg_load_action;

/*
    sg_store_action

    Defines the store action that should be performed at the end of a render pass:

    SG_STOREACTION_STORE:       store the rendered content to the color attachment image
    SG_STOREACTION_DONTCARE:    allows the GPU to discard the rendered content
*/
typedef enum sg_store_action {
    _SG_STOREACTION_DEFAULT,
    SG_STOREACTION_STORE,
    SG_STOREACTION_DONTCARE,
    _SG_STOREACTION_FORCE_U32 = 0x7FFFFFFF
} sg_store_action;


/*
    sg_pass_action

    The sg_pass_action struct defines the actions to be performed
    at the start and end of a render pass.

    - at the start of the pass: whether the render attachments should be cleared,
      loaded with their previous content, or start in an undefined state
    - for clear operations: the clear value (color, depth, or stencil values)
    - at the end of the pass: whether the rendering result should be
      stored back into the render attachment or discarded
*/
typedef struct sg_color_attachment_action {
    sg_load_action load_action;         // default: SG_LOADACTION_CLEAR
    sg_store_action store_action;       // default: SG_STOREACTION_STORE
    sg_color clear_value;               // default: { 0.5f, 0.5f, 0.5f, 1.0f }
} sg_color_attachment_action;

typedef struct sg_depth_attachment_action {
    sg_load_action load_action;         // default: SG_LOADACTION_CLEAR
    sg_store_action store_action;       // default: SG_STOREACTION_DONTCARE
    float clear_value;                  // default: 1.0
} sg_depth_attachment_action;

typedef struct sg_stencil_attachment_action {
    sg_load_action load_action;         // default: SG_LOADACTION_CLEAR
    sg_store_action store_action;       // default: SG_STOREACTION_DONTCARE
    uint8_t clear_value;                // default: 0
} sg_stencil_attachment_action;

typedef struct sg_pass_action {
    sg_color_attachment_action colors[SG_MAX_COLOR_ATTACHMENTS];
    sg_depth_attachment_action depth;
    sg_stencil_attachment_action stencil;
} sg_pass_action;

/*
    sg_swapchain

    Used in sg_begin_pass() to provide details about an external swapchain
    (pixel formats, sample count and backend-API specific render surface objects).

    The following information must be provided:

    - the width and height of the swapchain surfaces in number of pixels,
    - the pixel format of the render- and optional msaa-resolve-surface
    - the pixel format of the optional depth- or depth-stencil-surface
    - the MSAA sample count for the render and depth-stencil surface

    If the pixel formats and MSAA sample counts are left zero-initialized,
    their defaults are taken from the sg_environment struct provided in the
    sg_setup() call.

    The width and height *must* be > 0.

    Additionally the following backend API specific objects must be passed in
    as 'type erased' void pointers:

    GL:
        - on all GL backends, a GL framebuffer object must be provided. This
          can be zero for the default framebuffer.

    D3D11:
        - an ID3D11RenderTargetView for the rendering surface, without
          MSAA rendering this surface will also be displayed
        - an optional ID3D11DepthStencilView for the depth- or depth/stencil
          buffer surface
        - when MSAA rendering is used, another ID3D11RenderTargetView
          which serves as MSAA resolve target and will be displayed

    WebGPU (same as D3D11, except different types)
        - a WGPUTextureView for the rendering surface, without
          MSAA rendering this surface will also be displayed
        - an optional WGPUTextureView for the depth- or depth/stencil
          buffer surface
        - when MSAA rendering is used, another WGPUTextureView
          which serves as MSAA resolve target and will be displayed

    Metal (NOTE that the roles of provided surfaces is slightly different
    than on D3D11 or WebGPU in case of MSAA vs non-MSAA rendering):

        - A current CAMetalDrawable (NOT an MTLDrawable!) which will be presented.
          This will either be rendered to directly (if no MSAA is used), or serve
          as MSAA-resolve target.
        - an optional MTLTexture for the depth- or depth-stencil buffer
        - an optional multisampled MTLTexture which serves as intermediate
          rendering surface which will then be resolved into the
          CAMetalDrawable.

    NOTE that for Metal you must use an ObjC __bridge cast to
    properly tunnel the ObjC object id through a C void*, e.g.:

        swapchain.metal.current_drawable = (__bridge const void*) [mtkView currentDrawable];

    On all other backends you shouldn't need to mess with the reference count.

    It's a good practice to write a helper function which returns an initialized
    sg_swapchain structs, which can then be plugged directly into
    sg_pass.swapchain. Look at the function sglue_swapchain() in the sokol_glue.h
    as an example.
*/
typedef struct sg_metal_swapchain {
    const void* current_drawable;       // CAMetalDrawable (NOT MTLDrawable!!!)
    const void* depth_stencil_texture;  // MTLTexture
    const void* msaa_color_texture;     // MTLTexture
} sg_metal_swapchain;

typedef struct sg_d3d11_swapchain {
    const void* render_view;            // ID3D11RenderTargetView
    const void* resolve_view;           // ID3D11RenderTargetView
    const void* depth_stencil_view;     // ID3D11DepthStencilView
} sg_d3d11_swapchain;

typedef struct sg_wgpu_swapchain {
    const void* render_view;            // WGPUTextureView
    const void* resolve_view;           // WGPUTextureView
    const void* depth_stencil_view;     // WGPUTextureView
} sg_wgpu_swapchain;

typedef struct sg_gl_swapchain {
    uint32_t framebuffer;               // GL framebuffer object
} sg_gl_swapchain;

typedef struct sg_swapchain {
    int width;
    int height;
    int sample_count;
    sg_pixel_format color_format;
    sg_pixel_format depth_format;
    sg_metal_swapchain metal;
    sg_d3d11_swapchain d3d11;
    sg_wgpu_swapchain wgpu;
    sg_gl_swapchain gl;
} sg_swapchain;

/*
    sg_pass

    The sg_pass structure is passed as argument into the sg_begin_pass()
    function.

    For a swapchain render pass, provide an sg_pass_action and sg_swapchain
    struct (for instance via the sglue_swapchain() helper function from
    sokol_glue.h):

        sg_begin_pass(&(sg_pass){
            .action = { ... },
            .swapchain = sglue_swapchain(),
        });

    For an offscreen render pass, provide an sg_pass_action struct and
    an sg_attachments handle:

        sg_begin_pass(&(sg_pass){
            .action = { ... },
            .attachments = attachments,
        });

    You can also omit the .action object to get default pass action behaviour
    (clear to color=grey, depth=1 and stencil=0).

    For a compute pass, just set the sg_pass.compute boolean to true:

        sg_begin_pass(&(sg_pass){ .compute = true });
*/
typedef struct sg_pass {
    uint32_t _start_canary;
    bool compute;
    sg_pass_action action;
    sg_attachments attachments;
    sg_swapchain swapchain;
    const char* label;
    uint32_t _end_canary;
} sg_pass;

/*
    sg_bindings

    The sg_bindings structure defines the buffers, images and
    samplers resource bindings for the next draw call.

    To update the resource bindings, call sg_apply_bindings() with
    a pointer to a populated sg_bindings struct. Note that
    sg_apply_bindings() must be called after sg_apply_pipeline()
    and that bindings are not preserved across sg_apply_pipeline()
    calls, even when the new pipeline uses the same 'bindings layout'.

    A resource binding struct contains:

    - 1..N vertex buffers
    - 0..N vertex buffer offsets
    - 0..1 index buffers
    - 0..1 index buffer offsets
    - 0..N images
    - 0..N samplers
    - 0..N storage buffers

    Where 'N' is defined in the following constants:

    - SG_MAX_VERTEXBUFFER_BINDSLOTS
    - SG_MAX_IMAGE_BINDLOTS
    - SG_MAX_SAMPLER_BINDSLOTS
    - SG_MAX_STORAGEBUFFER_BINDGLOTS

    Note that inside compute passes vertex- and index-buffer-bindings are
    disallowed.

    When using sokol-shdc for shader authoring, the `layout(binding=N)`
    annotation in the shader code directly maps to the slot index for that
    resource type in the bindings struct, for instance the following vertex-
    and fragment-shader interface for sokol-shdc:

        @vs vs
        layout(binding=0) uniform vs_params { ... };
        layout(binding=0) readonly buffer ssbo { ... };
        layout(binding=0) uniform texture2D vs_tex;
        layout(binding=0) uniform sampler vs_smp;
        ...
        @end

        @fs fs
        layout(binding=1) uniform fs_params { ... };
        layout(binding=1) uniform texture2D fs_tex;
        layout(binding=1) uniform sampler fs_smp;
        ...
        @end

    ...would map to the following sg_bindings struct:

        const sg_bindings bnd = {
            .vertex_buffers[0] = ...,
            .images[0] = vs_tex,
            .images[1] = fs_tex,
            .samplers[0] = vs_smp,
            .samplers[1] = fs_smp,
            .storage_buffers[0] = ssbo,
        };

    ...alternatively you can use code-generated slot indices:

        const sg_bindings bnd = {
            .vertex_buffers[0] = ...,
            .images[IMG_vs_tex] = vs_tex,
            .images[IMG_fs_tex] = fs_tex,
            .samplers[SMP_vs_smp] = vs_smp,
            .samplers[SMP_fs_smp] = fs_smp,
            .storage_buffers[SBUF_ssbo] = ssbo,
        };

    Resource bindslots for a specific shader/pipeline may have gaps, and an
    sg_bindings struct may have populated bind slots which are not used by a
    specific shader. This allows to use the same sg_bindings struct across
    different shader variants.

    When not using sokol-shdc, the bindslot indices in the sg_bindings
    struct need to match the per-resource reflection info slot indices
    in the sg_shader_desc struct (for details about that see the
    sg_shader_desc struct documentation).

    The optional buffer offsets can be used to put different unrelated
    chunks of vertex- and/or index-data into the same buffer objects.
*/
typedef struct sg_bindings {
    uint32_t _start_canary;
    sg_buffer vertex_buffers[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    int vertex_buffer_offsets[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    sg_buffer index_buffer;
    int index_buffer_offset;
    sg_image images[SG_MAX_IMAGE_BINDSLOTS];
    sg_sampler samplers[SG_MAX_SAMPLER_BINDSLOTS];
    sg_buffer storage_buffers[SG_MAX_STORAGEBUFFER_BINDSLOTS];
    uint32_t _end_canary;
} sg_bindings;

/*
    sg_buffer_usage

    Describes how a buffer object is going to be used:

    .vertex_buffer (default: true)
        the buffer will bound as vertex buffer via sg_bindings.vertex_buffers[]
    .index_buffer (default: false)
        the buffer will bound as index buffer via sg_bindings.index_buffer
    .storage_buffer (default: false)
        the buffer will bound as storage buffer via sg_bindings.storage_buffers[]
    .immutable (default: true)
        the buffer content will never be updated from the CPU side (but
        may be written to by a compute shader)
    .dynamic_update (default: false)
        the buffer content will be infrequently updated from the CPU side
    .stream_upate (default: false)
        the buffer content will be updated each frame from the CPU side
*/
typedef struct sg_buffer_usage {
    bool vertex_buffer;
    bool index_buffer;
    bool storage_buffer;
    bool immutable;
    bool dynamic_update;
    bool stream_update;
} sg_buffer_usage;

/*
    sg_buffer_desc

    Creation parameters for sg_buffer objects, used in the sg_make_buffer() call.

    The default configuration is:

    .size:      0       (*must* be >0 for buffers without data)
    .usage              .vertex_buffer = true, .immutable = true
    .data.ptr   0       (*must* be valid for immutable buffers without storage buffer usage)
    .data.size  0       (*must* be > 0 for immutable buffers without storage buffer usage)
    .label      0       (optional string label)

    For immutable buffers which are initialized with initial data,
    keep the .size item zero-initialized, and set the size together with the
    pointer to the initial data in the .data item.

    For immutable or mutable buffers without initial data, keep the .data item
    zero-initialized, and set the buffer size in the .size item instead.

    You can also set both size values, but currently both size values must
    be identical (this may change in the future when the dynamic resource
    management may become more flexible).

    NOTE: Immutable buffers without storage-buffer-usage *must* be created
    with initial content, this restriction doesn't apply to storage buffer usage,
    because storage buffers may also get their initial content by running
    a compute shader on them.

    NOTE: Buffers without initial data will have undefined content, e.g.
    do *not* expect the buffer to be zero-initialized!

    ADVANCED TOPIC: Injecting native 3D-API buffers:

    The following struct members allow to inject your own GL, Metal
    or D3D11 buffers into sokol_gfx:

    .gl_buffers[SG_NUM_INFLIGHT_FRAMES]
    .mtl_buffers[SG_NUM_INFLIGHT_FRAMES]
    .d3d11_buffer

    You must still provide all other struct items except the .data item, and
    these must match the creation parameters of the native buffers you provide.
    For sg_buffer_desc.usage.immutable buffers, only provide a single native
    3D-API buffer, otherwise you need to provide SG_NUM_INFLIGHT_FRAMES buffers
    (only for GL and Metal, not D3D11). Providing multiple buffers for GL and
    Metal is necessary because sokol_gfx will rotate through them when calling
    sg_update_buffer() to prevent lock-stalls.

    Note that it is expected that immutable injected buffer have already been
    initialized with content, and the .content member must be 0!

    Also you need to call sg_reset_state_cache() after calling native 3D-API
    functions, and before calling any sokol_gfx function.
*/
typedef struct sg_buffer_desc {
    uint32_t _start_canary;
    size_t size;
    sg_buffer_usage usage;
    sg_range data;
    const char* label;
    // optionally inject backend-specific resources
    uint32_t gl_buffers[SG_NUM_INFLIGHT_FRAMES];
    const void* mtl_buffers[SG_NUM_INFLIGHT_FRAMES];
    const void* d3d11_buffer;
    const void* wgpu_buffer;
    uint32_t _end_canary;
} sg_buffer_desc;

/*
    sg_image_usage

    Describes how the image object is going to be used:

    .render_attachment (default: false)
        the image object is used as color-, resolve- or depth-stencil-
        attachment in a render pass
    .storage_attachment (default: false)
        the image object is used as storage-attachment in a
        compute pass (to be written to by compute shaders)
    .immutable (default: true)
        the image content cannot be updated from the CPU side
        (but may be updated by the GPU in a render- or compute-pass)
    .dynamic_update (default: false)
        the image content is updated infrequently by the CPU
    .stream_update (default: false)
        the image content is updated each frame by the CPU via

    Note that the usage as texture binding is implicit and always allowed.
*/
typedef struct sg_image_usage {
    bool render_attachment;
    bool storage_attachment;
    bool immutable;
    bool dynamic_update;
    bool stream_update;
} sg_image_usage;

/*
    sg_image_data

    Defines the content of an image through a 2D array of sg_range structs.
    The first array dimension is the cubemap face, and the second array
    dimension the mipmap level.
*/
typedef struct sg_image_data {
    sg_range subimage[SG_CUBEFACE_NUM][SG_MAX_MIPMAPS];
} sg_image_data;

/*
    sg_image_desc

    Creation parameters for sg_image objects, used in the sg_make_image() call.

    The default configuration is:

    .type               SG_IMAGETYPE_2D
    .usage              .immutable = true
    .width              0 (must be set to >0)
    .height             0 (must be set to >0)
    .num_slices         1 (3D textures: depth; array textures: number of layers)
    .num_mipmaps        1
    .pixel_format       SG_PIXELFORMAT_RGBA8 for textures, or sg_desc.environment.defaults.color_format for render targets
    .sample_count       1 for textures, or sg_desc.environment.defaults.sample_count for render targets
    .data               an sg_image_data struct to define the initial content
    .label              0 (optional string label for trace hooks)

    Q: Why is the default sample_count for render targets identical with the
    "default sample count" from sg_desc.environment.defaults.sample_count?

    A: So that it matches the default sample count in pipeline objects. Even
    though it is a bit strange/confusing that offscreen render targets by default
    get the same sample count as 'default swapchains', but it's better that
    an offscreen render target created with default parameters matches
    a pipeline object created with default parameters.

    NOTE:

    Regular (non-attachment) images with usage.immutable must be fully initialized by
    providing a valid .data member which points to initialization data.

    Images with usage.render_attachment or usage.storage_attachment must
    *not* be created with initial content. Be aware that the initial
    content of render- and storage-attachment images is undefined.

    ADVANCED TOPIC: Injecting native 3D-API textures:

    The following struct members allow to inject your own GL, Metal or D3D11
    textures into sokol_gfx:

    .gl_textures[SG_NUM_INFLIGHT_FRAMES]
    .mtl_textures[SG_NUM_INFLIGHT_FRAMES]
    .d3d11_texture
    .d3d11_shader_resource_view
    .wgpu_texture
    .wgpu_texture_view

    For GL, you can also specify the texture target or leave it empty to use
    the default texture target for the image type (GL_TEXTURE_2D for
    SG_IMAGETYPE_2D etc)

    For D3D11 and WebGPU, either only provide a texture, or both a texture and
    shader-resource-view / texture-view object. If you want to use access the
    injected texture in a shader you *must* provide a shader-resource-view.

    The same rules apply as for injecting native buffers (see sg_buffer_desc
    documentation for more details).
*/
typedef struct sg_image_desc {
    uint32_t _start_canary;
    sg_image_type type;
    sg_image_usage usage;
    int width;
    int height;
    int num_slices;
    int num_mipmaps;
    sg_pixel_format pixel_format;
    int sample_count;
    sg_image_data data;
    const char* label;
    // optionally inject backend-specific resources
    uint32_t gl_textures[SG_NUM_INFLIGHT_FRAMES];
    uint32_t gl_texture_target;
    const void* mtl_textures[SG_NUM_INFLIGHT_FRAMES];
    const void* d3d11_texture;
    const void* d3d11_shader_resource_view;
    const void* wgpu_texture;
    const void* wgpu_texture_view;
    uint32_t _end_canary;
} sg_image_desc;

/*
    sg_sampler_desc

    Creation parameters for sg_sampler objects, used in the sg_make_sampler() call

    .min_filter:        SG_FILTER_NEAREST
    .mag_filter:        SG_FILTER_NEAREST
    .mipmap_filter      SG_FILTER_NEAREST
    .wrap_u:            SG_WRAP_REPEAT
    .wrap_v:            SG_WRAP_REPEAT
    .wrap_w:            SG_WRAP_REPEAT (only SG_IMAGETYPE_3D)
    .min_lod            0.0f
    .max_lod            FLT_MAX
    .border_color       SG_BORDERCOLOR_OPAQUE_BLACK
    .compare            SG_COMPAREFUNC_NEVER
    .max_anisotropy     1 (must be 1..16)

*/
typedef struct sg_sampler_desc {
    uint32_t _start_canary;
    sg_filter min_filter;
    sg_filter mag_filter;
    sg_filter mipmap_filter;
    sg_wrap wrap_u;
    sg_wrap wrap_v;
    sg_wrap wrap_w;
    float min_lod;
    float max_lod;
    sg_border_color border_color;
    sg_compare_func compare;
    uint32_t max_anisotropy;
    const char* label;
    // optionally inject backend-specific resources
    uint32_t gl_sampler;
    const void* mtl_sampler;
    const void* d3d11_sampler;
    const void* wgpu_sampler;
    uint32_t _end_canary;
} sg_sampler_desc;

/*
    sg_shader_desc

    Used as parameter of sg_make_shader() to create a shader object which
    communicates shader source or bytecode and shader interface
    reflection information to sokol-gfx.

    If you use sokol-shdc you can ignore the following information since
    the sg_shader_desc struct will be code-generated.

    Otherwise you need to provide the following information to the
    sg_make_shader() call:

    - a vertex- and fragment-shader function:
        - the shader source or bytecode
        - an optional entry point name
        - for D3D11: an optional compile target when source code is provided
          (the defaults are "vs_4_0" and "ps_4_0")

    - ...or alternatively, a compute function:
        - the shader source or bytecode
        - an optional entry point name
        - for D3D11: an optional compile target when source code is provided
          (the default is "cs_5_0")

    - vertex attributes required by some backends (not for compute shaders):
        - the vertex attribute base type (undefined, float, signed int, unsigned int),
          this information is only used in the validation layer to check that the
          pipeline object vertex formats are compatible with the input vertex attribute
          type used in the vertex shader. NOTE that the default base type
          'undefined' skips the validation layer check.
        - for the GL backend: optional vertex attribute names used for name lookup
        - for the D3D11 backend: semantic names and indices

    - only for compute shaders on the Metal backend:
        - the workgroup size aka 'threads per thread-group'

          In other 3D APIs this is declared in the shader code:
            - GLSL: `layout(local_size_x=x, local_size_y=y, local_size_y=z) in;`
            - HLSL: `[numthreads(x, y, z)]`
            - WGSL: `@workgroup_size(x, y, z)`
          ...but in Metal the workgroup size is declared on the CPU side

    - reflection information for each uniform block binding used by the shader:
        - the shader stage the uniform block appears in (SG_SHADERSTAGE_*)
        - the size in bytes of the uniform block
        - backend-specific bindslots:
            - HLSL: the constant buffer register `register(b0..7)`
            - MSL: the buffer attribute `[[buffer(0..7)]]`
            - WGSL: the binding in `@group(0) @binding(0..15)`
        - GLSL only: a description of the uniform block interior
            - the memory layout standard (SG_UNIFORMLAYOUT_*)
            - for each member in the uniform block:
                - the member type (SG_UNIFORM_*)
                - if the member is an array, the array count
                - the member name

    - reflection information for each texture binding used by the shader:
        - the shader stage the texture appears in (SG_SHADERSTAGE_*)
        - the image type (SG_IMAGETYPE_*)
        - the image-sample type (SG_IMAGESAMPLETYPE_*)
        - whether the texture is multisampled
        - backend specific bindslots:
            - HLSL: the texture register `register(t0..23)`
            - MSL: the texture attribute `[[texture(0..19)]]`
            - WGSL: the binding in `@group(1) @binding(0..127)`

    - reflection information for each sampler used by the shader:
        - the shader stage the sampler appears in (SG_SHADERSTAGE_*)
        - the sampler type (SG_SAMPLERTYPE_*)
        - backend specific bindslots:
            - HLSL: the sampler register `register(s0..15)`
            - MSL: the sampler attribute `[[sampler(0..15)]]`
            - WGSL: the binding in `@group(0) @binding(0..127)`

    - reflection information for each storage buffer binding used by the shader:
        - the shader stage the storage buffer appears in (SG_SHADERSTAGE_*)
        - whether the storage buffer is readonly
        - backend specific bindslots:
            - HLSL:
                - for readonly storage buffer bindings: `register(t0..23)`
                - for read/write storage buffer bindings: `register(u0..11)`
            - MSL: the buffer attribute `[[buffer(8..15)]]`
            - WGSL: the binding in `@group(1) @binding(0..127)`
            - GL: the binding in `layout(binding=0..7)`

    - reflection information for each storage image binding used by the shader:
        - the shader stage (*must* be SG_SHADERSTAGE_COMPUTE)
        - whether the storage image is writeonly or readwrite (for readonly
          access use a regular texture binding instead)
        - the image type expected by the shader (SG_IMAGETYPE_*)
        - the access pixel format expected by the shader (SG_PIXELFORMAT_*),
          note that only a subset of pixel formats is allowed for storage image
          bindings
        - backend specific bindslots:
            - HLSL: the UAV register `register(u0..u11)`
            - MSL: the texture attribute `[[texture(0..19)]]`
            - WGSL: the binding in `@group(2) @binding(0..3)`
            - GLSL: the binding in `layout(binding=0..3, [access_format])`

    - reflection information for each combined image-sampler object
      used by the shader:
        - the shader stage (SG_SHADERSTAGE_*)
        - the texture's array index in the sg_shader_desc.images[] array
        - the sampler's array index in the sg_shader_desc.samplers[] array
        - GLSL only: the name of the combined image-sampler object

    The number and order of items in the sg_shader_desc.attrs[]
    array corresponds to the items in sg_pipeline_desc.layout.attrs.

        - sg_shader_desc.attrs[N] => sg_pipeline_desc.layout.attrs[N]

    NOTE that vertex attribute indices currently cannot have gaps.

    The items index in the sg_shader_desc.uniform_blocks[] array corresponds
    to the ub_slot arg in sg_apply_uniforms():

        - sg_shader_desc.uniform_blocks[N] => sg_apply_uniforms(N, ...)

    The items in the shader_desc images, samplers and storage_buffers
    arrays correspond to the same array items in the sg_bindings struct:

        - sg_shader_desc.images[N] => sg_bindings.images[N]
        - sg_shader_desc.samplers[N] => sg_bindings.samplers[N]
        - sg_shader_desc.storage_buffers[N] => sg_bindings.storage_buffers[N]

    For all GL backends, shader source-code must be provided. For D3D11 and Metal,
    either shader source-code or byte-code can be provided.

    NOTE that the uniform block, image, sampler, storage_buffer and
    storage_image arrays may have gaps. This allows to use the same sg_bindings
    struct for different related shader variants.

    For D3D11, if source code is provided, the d3dcompiler_47.dll will be loaded
    on demand. If this fails, shader creation will fail. When compiling HLSL
    source code, you can provide an optional target string via
    sg_shader_stage_desc.d3d11_target, the default target is "vs_4_0" for the
    vertex shader stage and "ps_4_0" for the pixel shader stage.
*/
typedef enum sg_shader_stage {
    SG_SHADERSTAGE_NONE,
    SG_SHADERSTAGE_VERTEX,
    SG_SHADERSTAGE_FRAGMENT,
    SG_SHADERSTAGE_COMPUTE,
    _SG_SHADERSTAGE_FORCE_U32 = 0x7FFFFFFF,
} sg_shader_stage;

typedef struct sg_shader_function {
    const char* source;
    sg_range bytecode;
    const char* entry;
    const char* d3d11_target;   // default: "vs_4_0" or "ps_4_0"
} sg_shader_function;

typedef enum sg_shader_attr_base_type {
    SG_SHADERATTRBASETYPE_UNDEFINED,
    SG_SHADERATTRBASETYPE_FLOAT,
    SG_SHADERATTRBASETYPE_SINT,
    SG_SHADERATTRBASETYPE_UINT,
    _SG_SHADERATTRBASETYPE_FORCE_U32 = 0x7FFFFFFF,
} sg_shader_attr_base_type;

typedef struct sg_shader_vertex_attr {
    sg_shader_attr_base_type base_type;  // default: UNDEFINED (disables validation)
    const char* glsl_name;      // [optional] GLSL attribute name
    const char* hlsl_sem_name;  // HLSL semantic name
    uint8_t hlsl_sem_index;     // HLSL semantic index
} sg_shader_vertex_attr;

typedef struct sg_glsl_shader_uniform {
    sg_uniform_type type;
    uint16_t array_count;       // 0 or 1 for scalars, >1 for arrays
    const char* glsl_name;      // glsl name binding is required on GL 4.1 and WebGL2
} sg_glsl_shader_uniform;

typedef struct sg_shader_uniform_block {
    sg_shader_stage stage;
    uint32_t size;
    uint8_t hlsl_register_b_n;  // HLSL register(bn)
    uint8_t msl_buffer_n;       // MSL [[buffer(n)]]
    uint8_t wgsl_group0_binding_n; // WGSL @group(0) @binding(n)
    sg_uniform_layout layout;
    sg_glsl_shader_uniform glsl_uniforms[SG_MAX_UNIFORMBLOCK_MEMBERS];
} sg_shader_uniform_block;

typedef struct sg_shader_image {
    sg_shader_stage stage;
    sg_image_type image_type;
    sg_image_sample_type sample_type;
    bool multisampled;
    uint8_t hlsl_register_t_n;      // HLSL register(tn) bind slot
    uint8_t msl_texture_n;          // MSL [[texture(n)]] bind slot
    uint8_t wgsl_group1_binding_n;  // WGSL @group(1) @binding(n) bind slot
} sg_shader_image;

typedef struct sg_shader_sampler {
    sg_shader_stage stage;
    sg_sampler_type sampler_type;
    uint8_t hlsl_register_s_n;      // HLSL register(sn) bind slot
    uint8_t msl_sampler_n;          // MSL [[sampler(n)]] bind slot
    uint8_t wgsl_group1_binding_n;  // WGSL @group(1) @binding(n) bind slot
} sg_shader_sampler;

typedef struct sg_shader_storage_buffer {
    sg_shader_stage stage;
    bool readonly;
    uint8_t hlsl_register_t_n;      // HLSL register(tn) bind slot (for readonly access)
    uint8_t hlsl_register_u_n;      // HLSL register(un) bind slot (for read/write access)
    uint8_t msl_buffer_n;           // MSL [[buffer(n)]] bind slot
    uint8_t wgsl_group1_binding_n;  // WGSL @group(1) @binding(n) bind slot
    uint8_t glsl_binding_n;         // GLSL layout(binding=n)
} sg_shader_storage_buffer;

typedef struct sg_shader_storage_image {
    sg_shader_stage stage;          // must be NONE or COMPUTE
    sg_image_type image_type;
    sg_pixel_format access_format;  // shader-access pixel format
    bool writeonly;                 // false means read/write access
    uint8_t hlsl_register_u_n;      // HLSL register(un) bind slot
    uint8_t msl_texture_n;          // MSL [[texture(n)]] bind slot
    uint8_t wgsl_group2_binding_n;  // WGSL @group(2) @binding(n) bind slot
    uint8_t glsl_binding_n;         // GLSL layout(binding=n)
} sg_shader_storage_image;

typedef struct sg_shader_image_sampler_pair {
    sg_shader_stage stage;
    uint8_t image_slot;
    uint8_t sampler_slot;
    const char* glsl_name;          // glsl name binding required because of GL 4.1 and WebGL2
} sg_shader_image_sampler_pair;

typedef struct sg_mtl_shader_threads_per_threadgroup {
    int x, y, z;
} sg_mtl_shader_threads_per_threadgroup;

typedef struct sg_shader_desc {
    uint32_t _start_canary;
    sg_shader_function vertex_func;
    sg_shader_function fragment_func;
    sg_shader_function compute_func;
    sg_shader_vertex_attr attrs[SG_MAX_VERTEX_ATTRIBUTES];
    sg_shader_uniform_block uniform_blocks[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
    sg_shader_storage_buffer storage_buffers[SG_MAX_STORAGEBUFFER_BINDSLOTS];
    sg_shader_image images[SG_MAX_IMAGE_BINDSLOTS];
    sg_shader_sampler samplers[SG_MAX_SAMPLER_BINDSLOTS];
    sg_shader_image_sampler_pair image_sampler_pairs[SG_MAX_IMAGE_SAMPLER_PAIRS];
    sg_shader_storage_image storage_images[SG_MAX_STORAGE_ATTACHMENTS];
    sg_mtl_shader_threads_per_threadgroup mtl_threads_per_threadgroup;
    const char* label;
    uint32_t _end_canary;
} sg_shader_desc;

/*
    sg_pipeline_desc

    The sg_pipeline_desc struct defines all creation parameters for an
    sg_pipeline object, used as argument to the sg_make_pipeline() function:

    Pipeline objects come in two flavours:

    - render pipelines for use in render passes
    - compute pipelines for use in compute passes

    A compute pipeline only requires a compute shader object but no
    'render state', while a render pipeline requires a vertex/fragment shader
    object and additional render state declarations:

    - the vertex layout for all input vertex buffers
    - a shader object
    - the 3D primitive type (points, lines, triangles, ...)
    - the index type (none, 16- or 32-bit)
    - all the fixed-function-pipeline state (depth-, stencil-, blend-state, etc...)

    If the vertex data has no gaps between vertex components, you can omit
    the .layout.buffers[].stride and layout.attrs[].offset items (leave them
    default-initialized to 0), sokol-gfx will then compute the offsets and
    strides from the vertex component formats (.layout.attrs[].format).
    Please note that ALL vertex attribute offsets must be 0 in order for the
    automatic offset computation to kick in.

    Note that if you use vertex-pulling from storage buffers instead of
    fixed-function vertex input you can simply omit the entire nested .layout
    struct.

    The default configuration is as follows:

    .compute:               false (must be set to true for a compute pipeline)
    .shader:                0 (must be initialized with a valid sg_shader id!)
    .layout:
        .buffers[]:         vertex buffer layouts
            .stride:        0 (if no stride is given it will be computed)
            .step_func      SG_VERTEXSTEP_PER_VERTEX
            .step_rate      1
        .attrs[]:           vertex attribute declarations
            .buffer_index   0 the vertex buffer bind slot
            .offset         0 (offsets can be omitted if the vertex layout has no gaps)
            .format         SG_VERTEXFORMAT_INVALID (must be initialized!)
    .depth:
        .pixel_format:      sg_desc.context.depth_format
        .compare:           SG_COMPAREFUNC_ALWAYS
        .write_enabled:     false
        .bias:              0.0f
        .bias_slope_scale:  0.0f
        .bias_clamp:        0.0f
    .stencil:
        .enabled:           false
        .front/back:
            .compare:       SG_COMPAREFUNC_ALWAYS
            .fail_op:       SG_STENCILOP_KEEP
            .depth_fail_op: SG_STENCILOP_KEEP
            .pass_op:       SG_STENCILOP_KEEP
        .read_mask:         0
        .write_mask:        0
        .ref:               0
    .color_count            1
    .colors[0..color_count]
        .pixel_format       sg_desc.context.color_format
        .write_mask:        SG_COLORMASK_RGBA
        .blend:
            .enabled:           false
            .src_factor_rgb:    SG_BLENDFACTOR_ONE
            .dst_factor_rgb:    SG_BLENDFACTOR_ZERO
            .op_rgb:            SG_BLENDOP_ADD
            .src_factor_alpha:  SG_BLENDFACTOR_ONE
            .dst_factor_alpha:  SG_BLENDFACTOR_ZERO
            .op_alpha:          SG_BLENDOP_ADD
    .primitive_type:            SG_PRIMITIVETYPE_TRIANGLES
    .index_type:                SG_INDEXTYPE_NONE
    .cull_mode:                 SG_CULLMODE_NONE
    .face_winding:              SG_FACEWINDING_CW
    .sample_count:              sg_desc.context.sample_count
    .blend_color:               (sg_color) { 0.0f, 0.0f, 0.0f, 0.0f }
    .alpha_to_coverage_enabled: false
    .label  0       (optional string label for trace hooks)
*/
typedef struct sg_vertex_buffer_layout_state {
    int stride;
    sg_vertex_step step_func;
    int step_rate;
} sg_vertex_buffer_layout_state;

typedef struct sg_vertex_attr_state {
    int buffer_index;
    int offset;
    sg_vertex_format format;
} sg_vertex_attr_state;

typedef struct sg_vertex_layout_state {
    sg_vertex_buffer_layout_state buffers[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    sg_vertex_attr_state attrs[SG_MAX_VERTEX_ATTRIBUTES];
} sg_vertex_layout_state;

typedef struct sg_stencil_face_state {
    sg_compare_func compare;
    sg_stencil_op fail_op;
    sg_stencil_op depth_fail_op;
    sg_stencil_op pass_op;
} sg_stencil_face_state;

typedef struct sg_stencil_state {
    bool enabled;
    sg_stencil_face_state front;
    sg_stencil_face_state back;
    uint8_t read_mask;
    uint8_t write_mask;
    uint8_t ref;
} sg_stencil_state;

typedef struct sg_depth_state {
    sg_pixel_format pixel_format;
    sg_compare_func compare;
    bool write_enabled;
    float bias;
    float bias_slope_scale;
    float bias_clamp;
} sg_depth_state;

typedef struct sg_blend_state {
    bool enabled;
    sg_blend_factor src_factor_rgb;
    sg_blend_factor dst_factor_rgb;
    sg_blend_op op_rgb;
    sg_blend_factor src_factor_alpha;
    sg_blend_factor dst_factor_alpha;
    sg_blend_op op_alpha;
} sg_blend_state;

typedef struct sg_color_target_state {
    sg_pixel_format pixel_format;
    sg_color_mask write_mask;
    sg_blend_state blend;
} sg_color_target_state;

typedef struct sg_pipeline_desc {
    uint32_t _start_canary;
    bool compute;
    sg_shader shader;
    sg_vertex_layout_state layout;
    sg_depth_state depth;
    sg_stencil_state stencil;
    int color_count;
    sg_color_target_state colors[SG_MAX_COLOR_ATTACHMENTS];
    sg_primitive_type primitive_type;
    sg_index_type index_type;
    sg_cull_mode cull_mode;
    sg_face_winding face_winding;
    int sample_count;
    sg_color blend_color;
    bool alpha_to_coverage_enabled;
    const char* label;
    uint32_t _end_canary;
} sg_pipeline_desc;

/*
    sg_attachments_desc

    Creation parameters for an sg_attachments object, used as argument to the
    sg_make_attachments() function.

    An attachments object bundles either bundles 'render attachments' for
    a render pass, or 'storage attachments' for a compute pass which writes
    to storage images.

    Render attachments are:

        - 0..4 color attachment images
        - 0..4 msaa-resolve attachment images
        - 0 or one depth-stencil attachment image

    Note that all types of render attachment images must be created with
    `sg_image_desc.usage.render_attachment = true`. At least one color-attachment
    or depth-stencil-attachment image must be provided in a render pass
    (only providing a depth-stencil-attachment is useful for depth-only passes).

    Alternatively provide 1..4 storage attachment images which must be created
    with `sg_image_desc.usage.storage_attachment = true`.

    An sg_attachments object cannot have both render- and storage-attachments.

    Each attachment definition consists of an image object, and two additional indices
    describing which subimage the pass will render into: one mipmap index, and if the image
    is a cubemap, array-texture or 3D-texture, the face-index, array-layer or
    depth-slice.

    All attachments must have the same width and height.

    All color attachments and the depth-stencil attachment must have the
    same sample count.

    If a resolve attachment is set, an MSAA-resolve operation from the
    associated color attachment image into the resolve attachment image will take
    place in the sg_end_pass() function. In this case, the color attachment
    must have a (sample_count>1), and the resolve attachment a
    (sample_count==1). The resolve attachment also must have the same pixel
    format as the color attachment.

    NOTE that MSAA depth-stencil attachments cannot be msaa-resolved!
*/
typedef struct sg_attachment_desc {
    sg_image image;
    int mip_level;
    int slice;      // cube texture: face; array texture: layer; 3D texture: slice
} sg_attachment_desc;

typedef struct sg_attachments_desc {
    uint32_t _start_canary;
    sg_attachment_desc colors[SG_MAX_COLOR_ATTACHMENTS];
    sg_attachment_desc resolves[SG_MAX_COLOR_ATTACHMENTS];
    sg_attachment_desc depth_stencil;
    sg_attachment_desc storages[SG_MAX_STORAGE_ATTACHMENTS];
    const char* label;
    uint32_t _end_canary;
} sg_attachments_desc;

/*
    sg_trace_hooks

    Installable callback functions to keep track of the sokol-gfx calls,
    this is useful for debugging, or keeping track of resource creation
    and destruction.

    Trace hooks are installed with sg_install_trace_hooks(), this returns
    another sg_trace_hooks struct with the previous set of
    trace hook function pointers. These should be invoked by the
    new trace hooks to form a proper call chain.
*/
typedef struct sg_trace_hooks {
    void* user_data;
    void (*reset_state_cache)(void* user_data);
    void (*make_buffer)(const sg_buffer_desc* desc, sg_buffer result, void* user_data);
    void (*make_image)(const sg_image_desc* desc, sg_image result, void* user_data);
    void (*make_sampler)(const sg_sampler_desc* desc, sg_sampler result, void* user_data);
    void (*make_shader)(const sg_shader_desc* desc, sg_shader result, void* user_data);
    void (*make_pipeline)(const sg_pipeline_desc* desc, sg_pipeline result, void* user_data);
    void (*make_attachments)(const sg_attachments_desc* desc, sg_attachments result, void* user_data);
    void (*destroy_buffer)(sg_buffer buf, void* user_data);
    void (*destroy_image)(sg_image img, void* user_data);
    void (*destroy_sampler)(sg_sampler smp, void* user_data);
    void (*destroy_shader)(sg_shader shd, void* user_data);
    void (*destroy_pipeline)(sg_pipeline pip, void* user_data);
    void (*destroy_attachments)(sg_attachments atts, void* user_data);
    void (*update_buffer)(sg_buffer buf, const sg_range* data, void* user_data);
    void (*update_image)(sg_image img, const sg_image_data* data, void* user_data);
    void (*append_buffer)(sg_buffer buf, const sg_range* data, int result, void* user_data);
    void (*begin_pass)(const sg_pass* pass, void* user_data);
    void (*apply_viewport)(int x, int y, int width, int height, bool origin_top_left, void* user_data);
    void (*apply_scissor_rect)(int x, int y, int width, int height, bool origin_top_left, void* user_data);
    void (*apply_pipeline)(sg_pipeline pip, void* user_data);
    void (*apply_bindings)(const sg_bindings* bindings, void* user_data);
    void (*apply_uniforms)(int ub_index, const sg_range* data, void* user_data);
    void (*draw)(int base_element, int num_elements, int num_instances, void* user_data);
    void (*dispatch)(int num_groups_x, int num_groups_y, int num_groups_z, void* user_data);
    void (*end_pass)(void* user_data);
    void (*commit)(void* user_data);
    void (*alloc_buffer)(sg_buffer result, void* user_data);
    void (*alloc_image)(sg_image result, void* user_data);
    void (*alloc_sampler)(sg_sampler result, void* user_data);
    void (*alloc_shader)(sg_shader result, void* user_data);
    void (*alloc_pipeline)(sg_pipeline result, void* user_data);
    void (*alloc_attachments)(sg_attachments result, void* user_data);
    void (*dealloc_buffer)(sg_buffer buf_id, void* user_data);
    void (*dealloc_image)(sg_image img_id, void* user_data);
    void (*dealloc_sampler)(sg_sampler smp_id, void* user_data);
    void (*dealloc_shader)(sg_shader shd_id, void* user_data);
    void (*dealloc_pipeline)(sg_pipeline pip_id, void* user_data);
    void (*dealloc_attachments)(sg_attachments atts_id, void* user_data);
    void (*init_buffer)(sg_buffer buf_id, const sg_buffer_desc* desc, void* user_data);
    void (*init_image)(sg_image img_id, const sg_image_desc* desc, void* user_data);
    void (*init_sampler)(sg_sampler smp_id, const sg_sampler_desc* desc, void* user_data);
    void (*init_shader)(sg_shader shd_id, const sg_shader_desc* desc, void* user_data);
    void (*init_pipeline)(sg_pipeline pip_id, const sg_pipeline_desc* desc, void* user_data);
    void (*init_attachments)(sg_attachments atts_id, const sg_attachments_desc* desc, void* user_data);
    void (*uninit_buffer)(sg_buffer buf_id, void* user_data);
    void (*uninit_image)(sg_image img_id, void* user_data);
    void (*uninit_sampler)(sg_sampler smp_id, void* user_data);
    void (*uninit_shader)(sg_shader shd_id, void* user_data);
    void (*uninit_pipeline)(sg_pipeline pip_id, void* user_data);
    void (*uninit_attachments)(sg_attachments atts_id, void* user_data);
    void (*fail_buffer)(sg_buffer buf_id, void* user_data);
    void (*fail_image)(sg_image img_id, void* user_data);
    void (*fail_sampler)(sg_sampler smp_id, void* user_data);
    void (*fail_shader)(sg_shader shd_id, void* user_data);
    void (*fail_pipeline)(sg_pipeline pip_id, void* user_data);
    void (*fail_attachments)(sg_attachments atts_id, void* user_data);
    void (*push_debug_group)(const char* name, void* user_data);
    void (*pop_debug_group)(void* user_data);
} sg_trace_hooks;

/*
    sg_buffer_info
    sg_image_info
    sg_sampler_info
    sg_shader_info
    sg_pipeline_info
    sg_attachments_info

    These structs contain various internal resource attributes which
    might be useful for debug-inspection. Please don't rely on the
    actual content of those structs too much, as they are quite closely
    tied to sokol_gfx.h internals and may change more frequently than
    the other public API elements.

    The *_info structs are used as the return values of the following functions:

    sg_query_buffer_info()
    sg_query_image_info()
    sg_query_sampler_info()
    sg_query_shader_info()
    sg_query_pipeline_info()
    sg_query_attachments_info()
*/
typedef struct sg_slot_info {
    sg_resource_state state;    // the current state of this resource slot
    uint32_t res_id;            // type-neutral resource if (e.g. sg_buffer.id)
    uint32_t uninit_count;
} sg_slot_info;

typedef struct sg_buffer_info {
    sg_slot_info slot;              // resource pool slot info
    uint32_t update_frame_index;    // frame index of last sg_update_buffer()
    uint32_t append_frame_index;    // frame index of last sg_append_buffer()
    int append_pos;                 // current position in buffer for sg_append_buffer()
    bool append_overflow;           // is buffer in overflow state (due to sg_append_buffer)
    int num_slots;                  // number of renaming-slots for dynamically updated buffers
    int active_slot;                // currently active write-slot for dynamically updated buffers
} sg_buffer_info;

typedef struct sg_image_info {
    sg_slot_info slot;              // resource pool slot info
    uint32_t upd_frame_index;       // frame index of last sg_update_image()
    int num_slots;                  // number of renaming-slots for dynamically updated images
    int active_slot;                // currently active write-slot for dynamically updated images
} sg_image_info;

typedef struct sg_sampler_info {
    sg_slot_info slot;              // resource pool slot info
} sg_sampler_info;

typedef struct sg_shader_info {
    sg_slot_info slot;              // resource pool slot info
} sg_shader_info;

typedef struct sg_pipeline_info {
    sg_slot_info slot;              // resource pool slot info
} sg_pipeline_info;

typedef struct sg_attachments_info {
    sg_slot_info slot;              // resource pool slot info
} sg_attachments_info;

/*
    sg_frame_stats

    Allows to track generic and backend-specific stats about a
    render frame. Obtained by calling sg_query_frame_stats(). The returned
    struct contains information about the *previous* frame.
*/
typedef struct sg_frame_stats_gl {
    uint32_t num_bind_buffer;
    uint32_t num_active_texture;
    uint32_t num_bind_texture;
    uint32_t num_bind_sampler;
    uint32_t num_use_program;
    uint32_t num_render_state;
    uint32_t num_vertex_attrib_pointer;
    uint32_t num_vertex_attrib_divisor;
    uint32_t num_enable_vertex_attrib_array;
    uint32_t num_disable_vertex_attrib_array;
    uint32_t num_uniform;
    uint32_t num_memory_barriers;
} sg_frame_stats_gl;

typedef struct sg_frame_stats_d3d11_pass {
    uint32_t num_om_set_render_targets;
    uint32_t num_clear_render_target_view;
    uint32_t num_clear_depth_stencil_view;
    uint32_t num_resolve_subresource;
} sg_frame_stats_d3d11_pass;

typedef struct sg_frame_stats_d3d11_pipeline {
    uint32_t num_rs_set_state;
    uint32_t num_om_set_depth_stencil_state;
    uint32_t num_om_set_blend_state;
    uint32_t num_ia_set_primitive_topology;
    uint32_t num_ia_set_input_layout;
    uint32_t num_vs_set_shader;
    uint32_t num_vs_set_constant_buffers;
    uint32_t num_ps_set_shader;
    uint32_t num_ps_set_constant_buffers;
    uint32_t num_cs_set_shader;
    uint32_t num_cs_set_constant_buffers;
} sg_frame_stats_d3d11_pipeline;

typedef struct sg_frame_stats_d3d11_bindings {
    uint32_t num_ia_set_vertex_buffers;
    uint32_t num_ia_set_index_buffer;
    uint32_t num_vs_set_shader_resources;
    uint32_t num_vs_set_samplers;
    uint32_t num_ps_set_shader_resources;
    uint32_t num_ps_set_samplers;
    uint32_t num_cs_set_shader_resources;
    uint32_t num_cs_set_samplers;
    uint32_t num_cs_set_unordered_access_views;
} sg_frame_stats_d3d11_bindings;

typedef struct sg_frame_stats_d3d11_uniforms {
    uint32_t num_update_subresource;
} sg_frame_stats_d3d11_uniforms;

typedef struct sg_frame_stats_d3d11_draw {
    uint32_t num_draw_indexed_instanced;
    uint32_t num_draw_indexed;
    uint32_t num_draw_instanced;
    uint32_t num_draw;
} sg_frame_stats_d3d11_draw;

typedef struct sg_frame_stats_d3d11 {
    sg_frame_stats_d3d11_pass pass;
    sg_frame_stats_d3d11_pipeline pipeline;
    sg_frame_stats_d3d11_bindings bindings;
    sg_frame_stats_d3d11_uniforms uniforms;
    sg_frame_stats_d3d11_draw draw;
    uint32_t num_map;
    uint32_t num_unmap;
} sg_frame_stats_d3d11;

typedef struct sg_frame_stats_metal_idpool {
    uint32_t num_added;
    uint32_t num_released;
    uint32_t num_garbage_collected;
} sg_frame_stats_metal_idpool;

typedef struct sg_frame_stats_metal_pipeline {
    uint32_t num_set_blend_color;
    uint32_t num_set_cull_mode;
    uint32_t num_set_front_facing_winding;
    uint32_t num_set_stencil_reference_value;
    uint32_t num_set_depth_bias;
    uint32_t num_set_render_pipeline_state;
    uint32_t num_set_depth_stencil_state;
} sg_frame_stats_metal_pipeline;

typedef struct sg_frame_stats_metal_bindings {
    uint32_t num_set_vertex_buffer;
    uint32_t num_set_vertex_texture;
    uint32_t num_set_vertex_sampler_state;
    uint32_t num_set_fragment_buffer;
    uint32_t num_set_fragment_texture;
    uint32_t num_set_fragment_sampler_state;
    uint32_t num_set_compute_buffer;
    uint32_t num_set_compute_texture;
    uint32_t num_set_compute_sampler_state;
} sg_frame_stats_metal_bindings;

typedef struct sg_frame_stats_metal_uniforms {
    uint32_t num_set_vertex_buffer_offset;
    uint32_t num_set_fragment_buffer_offset;
    uint32_t num_set_compute_buffer_offset;
} sg_frame_stats_metal_uniforms;

typedef struct sg_frame_stats_metal {
    sg_frame_stats_metal_idpool idpool;
    sg_frame_stats_metal_pipeline pipeline;
    sg_frame_stats_metal_bindings bindings;
    sg_frame_stats_metal_uniforms uniforms;
} sg_frame_stats_metal;

typedef struct sg_frame_stats_wgpu_uniforms {
    uint32_t num_set_bindgroup;
    uint32_t size_write_buffer;
} sg_frame_stats_wgpu_uniforms;

typedef struct sg_frame_stats_wgpu_bindings {
    uint32_t num_set_vertex_buffer;
    uint32_t num_skip_redundant_vertex_buffer;
    uint32_t num_set_index_buffer;
    uint32_t num_skip_redundant_index_buffer;
    uint32_t num_create_bindgroup;
    uint32_t num_discard_bindgroup;
    uint32_t num_set_bindgroup;
    uint32_t num_skip_redundant_bindgroup;
    uint32_t num_bindgroup_cache_hits;
    uint32_t num_bindgroup_cache_misses;
    uint32_t num_bindgroup_cache_collisions;
    uint32_t num_bindgroup_cache_invalidates;
    uint32_t num_bindgroup_cache_hash_vs_key_mismatch;
} sg_frame_stats_wgpu_bindings;

typedef struct sg_frame_stats_wgpu {
    sg_frame_stats_wgpu_uniforms uniforms;
    sg_frame_stats_wgpu_bindings bindings;
} sg_frame_stats_wgpu;

typedef struct sg_frame_stats {
    uint32_t frame_index;   // current frame counter, starts at 0

    uint32_t num_passes;
    uint32_t num_apply_viewport;
    uint32_t num_apply_scissor_rect;
    uint32_t num_apply_pipeline;
    uint32_t num_apply_bindings;
    uint32_t num_apply_uniforms;
    uint32_t num_draw;
    uint32_t num_dispatch;
    uint32_t num_update_buffer;
    uint32_t num_append_buffer;
    uint32_t num_update_image;

    uint32_t size_apply_uniforms;
    uint32_t size_update_buffer;
    uint32_t size_append_buffer;
    uint32_t size_update_image;

    sg_frame_stats_gl gl;
    sg_frame_stats_d3d11 d3d11;
    sg_frame_stats_metal metal;
    sg_frame_stats_wgpu wgpu;
} sg_frame_stats;

/*
    sg_log_item

    An enum with a unique item for each log message, warning, error
    and validation layer message. Note that these messages are only
    visible when a logger function is installed in the sg_setup() call.
*/
#define _SG_LOG_ITEMS \
    _SG_LOGITEM_XMACRO(OK, "Ok") \
    _SG_LOGITEM_XMACRO(MALLOC_FAILED, "memory allocation failed") \
    _SG_LOGITEM_XMACRO(GL_TEXTURE_FORMAT_NOT_SUPPORTED, "pixel format not supported for texture (gl)") \
    _SG_LOGITEM_XMACRO(GL_3D_TEXTURES_NOT_SUPPORTED, "3d textures not supported (gl)") \
    _SG_LOGITEM_XMACRO(GL_ARRAY_TEXTURES_NOT_SUPPORTED, "array textures not supported (gl)") \
    _SG_LOGITEM_XMACRO(GL_STORAGEBUFFER_GLSL_BINDING_OUT_OF_RANGE, "GLSL storage buffer bindslot is out of range (must be 0..7) (gl)") \
    _SG_LOGITEM_XMACRO(GL_STORAGEIMAGE_GLSL_BINDING_OUT_OF_RANGE, "GLSL storage image bindslot is out of range (must be 0..3) (gl)") \
    _SG_LOGITEM_XMACRO(GL_SHADER_COMPILATION_FAILED, "shader compilation failed (gl)") \
    _SG_LOGITEM_XMACRO(GL_SHADER_LINKING_FAILED, "shader linking failed (gl)") \
    _SG_LOGITEM_XMACRO(GL_VERTEX_ATTRIBUTE_NOT_FOUND_IN_SHADER, "vertex attribute not found in shader; NOTE: may be caused by GL driver's GLSL compiler removing unused globals") \
    _SG_LOGITEM_XMACRO(GL_UNIFORMBLOCK_NAME_NOT_FOUND_IN_SHADER, "uniform block name not found in shader; NOTE: may be caused by GL driver's GLSL compiler removing unused globals") \
    _SG_LOGITEM_XMACRO(GL_IMAGE_SAMPLER_NAME_NOT_FOUND_IN_SHADER, "image-sampler name not found in shader; NOTE: may be caused by GL driver's GLSL compiler removing unused globals") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_UNDEFINED, "framebuffer completeness check failed with GL_FRAMEBUFFER_UNDEFINED (gl)") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_INCOMPLETE_ATTACHMENT, "framebuffer completeness check failed with GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT (gl)") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MISSING_ATTACHMENT, "framebuffer completeness check failed with GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT (gl)") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_UNSUPPORTED, "framebuffer completeness check failed with GL_FRAMEBUFFER_UNSUPPORTED (gl)") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MULTISAMPLE, "framebuffer completeness check failed with GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE (gl)") \
    _SG_LOGITEM_XMACRO(GL_FRAMEBUFFER_STATUS_UNKNOWN, "framebuffer completeness check failed (unknown reason) (gl)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_BUFFER_FAILED, "CreateBuffer() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_BUFFER_SRV_FAILED, "CreateShaderResourceView() failed for storage buffer (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_BUFFER_UAV_FAILED, "CreateUnorderedAccessView() failed for storage buffer (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_DEPTH_TEXTURE_UNSUPPORTED_PIXEL_FORMAT, "pixel format not supported for depth-stencil texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_DEPTH_TEXTURE_FAILED, "CreateTexture2D() failed for depth-stencil texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_2D_TEXTURE_UNSUPPORTED_PIXEL_FORMAT, "pixel format not supported for 2d-, cube- or array-texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_2D_TEXTURE_FAILED, "CreateTexture2D() failed for 2d-, cube- or array-texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_2D_SRV_FAILED, "CreateShaderResourceView() failed for 2d-, cube- or array-texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_3D_TEXTURE_UNSUPPORTED_PIXEL_FORMAT, "pixel format not supported for 3D texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_3D_TEXTURE_FAILED, "CreateTexture3D() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_3D_SRV_FAILED, "CreateShaderResourceView() failed for 3d texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_MSAA_TEXTURE_FAILED, "CreateTexture2D() failed for MSAA render target texture (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_SAMPLER_STATE_FAILED, "CreateSamplerState() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_UNIFORMBLOCK_HLSL_REGISTER_B_OUT_OF_RANGE, "uniform block 'hlsl_register_b_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(D3D11_STORAGEBUFFER_HLSL_REGISTER_T_OUT_OF_RANGE, "storage buffer 'hlsl_register_t_n' is out of range (must be 0..23)") \
    _SG_LOGITEM_XMACRO(D3D11_STORAGEBUFFER_HLSL_REGISTER_U_OUT_OF_RANGE, "storage buffer 'hlsl_register_u_n' is out of range (must be 0..11)") \
    _SG_LOGITEM_XMACRO(D3D11_IMAGE_HLSL_REGISTER_T_OUT_OF_RANGE, "image 'hlsl_register_t_n' is out of range (must be 0..23)") \
    _SG_LOGITEM_XMACRO(D3D11_SAMPLER_HLSL_REGISTER_S_OUT_OF_RANGE, "sampler 'hlsl_register_s_n' is out of rang (must be 0..15)") \
    _SG_LOGITEM_XMACRO(D3D11_STORAGEIMAGE_HLSL_REGISTER_U_OUT_OF_RANGE, "storage image 'hlsl_register_u_n' is out of range (must be 0..11)") \
    _SG_LOGITEM_XMACRO(D3D11_LOAD_D3DCOMPILER_47_DLL_FAILED, "loading d3dcompiler_47.dll failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_SHADER_COMPILATION_FAILED, "shader compilation failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_SHADER_COMPILATION_OUTPUT, "") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_CONSTANT_BUFFER_FAILED, "CreateBuffer() failed for uniform constant buffer (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_INPUT_LAYOUT_FAILED, "CreateInputLayout() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_RASTERIZER_STATE_FAILED, "CreateRasterizerState() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_DEPTH_STENCIL_STATE_FAILED, "CreateDepthStencilState() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_BLEND_STATE_FAILED, "CreateBlendState() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_RTV_FAILED, "CreateRenderTargetView() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_DSV_FAILED, "CreateDepthStencilView() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_CREATE_UAV_FAILED, "CreateUnorderedAccessView() failed (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_MAP_FOR_UPDATE_BUFFER_FAILED, "Map() failed when updating buffer (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_MAP_FOR_APPEND_BUFFER_FAILED, "Map() failed when appending to buffer (d3d11)") \
    _SG_LOGITEM_XMACRO(D3D11_MAP_FOR_UPDATE_IMAGE_FAILED, "Map() failed when updating image (d3d11)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_BUFFER_FAILED, "failed to create buffer object (metal)") \
    _SG_LOGITEM_XMACRO(METAL_TEXTURE_FORMAT_NOT_SUPPORTED, "pixel format not supported for texture (metal)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_TEXTURE_FAILED, "failed to create texture object (metal)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_SAMPLER_FAILED, "failed to create sampler object (metal)") \
    _SG_LOGITEM_XMACRO(METAL_SHADER_COMPILATION_FAILED, "shader compilation failed (metal)") \
    _SG_LOGITEM_XMACRO(METAL_SHADER_CREATION_FAILED, "shader creation failed (metal)") \
    _SG_LOGITEM_XMACRO(METAL_SHADER_COMPILATION_OUTPUT, "") \
    _SG_LOGITEM_XMACRO(METAL_SHADER_ENTRY_NOT_FOUND, "shader entry function not found (metal)") \
    _SG_LOGITEM_XMACRO(METAL_UNIFORMBLOCK_MSL_BUFFER_SLOT_OUT_OF_RANGE, "uniform block 'msl_buffer_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(METAL_STORAGEBUFFER_MSL_BUFFER_SLOT_OUT_OF_RANGE, "storage buffer 'msl_buffer_n' is out of range (must be 8..15)") \
    _SG_LOGITEM_XMACRO(METAL_STORAGEIMAGE_MSL_TEXTURE_SLOT_OUT_OF_RANGE, "storage image 'msl_texture_n' is out of range (must be 0..19)") \
    _SG_LOGITEM_XMACRO(METAL_IMAGE_MSL_TEXTURE_SLOT_OUT_OF_RANGE, "image 'msl_texture_n' is out of range (must be 0..19)") \
    _SG_LOGITEM_XMACRO(METAL_SAMPLER_MSL_SAMPLER_SLOT_OUT_OF_RANGE, "sampler 'msl_sampler_n' is out of range (must be 0..15)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_CPS_FAILED, "failed to create compute pipeline state (metal)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_CPS_OUTPUT, "") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_RPS_FAILED, "failed to create render pipeline state (metal)") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_RPS_OUTPUT, "") \
    _SG_LOGITEM_XMACRO(METAL_CREATE_DSS_FAILED, "failed to create depth stencil state (metal)") \
    _SG_LOGITEM_XMACRO(WGPU_BINDGROUPS_POOL_EXHAUSTED, "bindgroups pool exhausted (increase sg_desc.bindgroups_cache_size) (wgpu)") \
    _SG_LOGITEM_XMACRO(WGPU_BINDGROUPSCACHE_SIZE_GREATER_ONE, "sg_desc.wgpu_bindgroups_cache_size must be > 1 (wgpu)") \
    _SG_LOGITEM_XMACRO(WGPU_BINDGROUPSCACHE_SIZE_POW2, "sg_desc.wgpu_bindgroups_cache_size must be a power of 2 (wgpu)") \
    _SG_LOGITEM_XMACRO(WGPU_CREATEBINDGROUP_FAILED, "wgpuDeviceCreateBindGroup failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_BUFFER_FAILED, "wgpuDeviceCreateBuffer() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_TEXTURE_FAILED, "wgpuDeviceCreateTexture() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_TEXTURE_VIEW_FAILED, "wgpuTextureCreateView() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_SAMPLER_FAILED, "wgpuDeviceCreateSampler() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_SHADER_MODULE_FAILED, "wgpuDeviceCreateShaderModule() failed") \
    _SG_LOGITEM_XMACRO(WGPU_SHADER_CREATE_BINDGROUP_LAYOUT_FAILED, "wgpuDeviceCreateBindGroupLayout() for shader stage failed") \
    _SG_LOGITEM_XMACRO(WGPU_UNIFORMBLOCK_WGSL_GROUP0_BINDING_OUT_OF_RANGE, "uniform block 'wgsl_group0_binding_n' is out of range (must be 0..15)") \
    _SG_LOGITEM_XMACRO(WGPU_STORAGEBUFFER_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "storage buffer 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(WGPU_IMAGE_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "image 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(WGPU_SAMPLER_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "sampler 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(WGPU_STORAGEIMAGE_WGSL_GROUP2_BINDING_OUT_OF_RANGE, "storage image 'wgsl_group2_binding_n' is out of range (must be 0..3)") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_PIPELINE_LAYOUT_FAILED, "wgpuDeviceCreatePipelineLayout() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_RENDER_PIPELINE_FAILED, "wgpuDeviceCreateRenderPipeline() failed") \
    _SG_LOGITEM_XMACRO(WGPU_CREATE_COMPUTE_PIPELINE_FAILED, "wgpuDeviceCreateComputePipeline() failed") \
    _SG_LOGITEM_XMACRO(WGPU_ATTACHMENTS_CREATE_TEXTURE_VIEW_FAILED, "wgpuTextureCreateView() failed in create attachments") \
    _SG_LOGITEM_XMACRO(IDENTICAL_COMMIT_LISTENER, "attempting to add identical commit listener") \
    _SG_LOGITEM_XMACRO(COMMIT_LISTENER_ARRAY_FULL, "commit listener array full") \
    _SG_LOGITEM_XMACRO(TRACE_HOOKS_NOT_ENABLED, "sg_install_trace_hooks() called, but SOKOL_TRACE_HOOKS is not defined") \
    _SG_LOGITEM_XMACRO(DEALLOC_BUFFER_INVALID_STATE, "sg_dealloc_buffer(): buffer must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(DEALLOC_IMAGE_INVALID_STATE, "sg_dealloc_image(): image must be in alloc state") \
    _SG_LOGITEM_XMACRO(DEALLOC_SAMPLER_INVALID_STATE, "sg_dealloc_sampler(): sampler must be in alloc state") \
    _SG_LOGITEM_XMACRO(DEALLOC_SHADER_INVALID_STATE, "sg_dealloc_shader(): shader must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(DEALLOC_PIPELINE_INVALID_STATE, "sg_dealloc_pipeline(): pipeline must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(DEALLOC_ATTACHMENTS_INVALID_STATE, "sg_dealloc_attachments(): attachments must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_BUFFER_INVALID_STATE, "sg_init_buffer(): buffer must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_IMAGE_INVALID_STATE, "sg_init_image(): image must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_SAMPLER_INVALID_STATE, "sg_init_sampler(): sampler must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_SHADER_INVALID_STATE, "sg_init_shader(): shader must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_PIPELINE_INVALID_STATE, "sg_init_pipeline(): pipeline must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(INIT_ATTACHMENTS_INVALID_STATE, "sg_init_attachments(): pass must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(UNINIT_BUFFER_INVALID_STATE, "sg_uninit_buffer(): buffer must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(UNINIT_IMAGE_INVALID_STATE, "sg_uninit_image(): image must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(UNINIT_SAMPLER_INVALID_STATE, "sg_uninit_sampler(): sampler must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(UNINIT_SHADER_INVALID_STATE, "sg_uninit_shader(): shader must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(UNINIT_PIPELINE_INVALID_STATE, "sg_uninit_pipeline(): pipeline must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(UNINIT_ATTACHMENTS_INVALID_STATE, "sg_uninit_attachments(): attachments must be in VALID or FAILED state") \
    _SG_LOGITEM_XMACRO(FAIL_BUFFER_INVALID_STATE, "sg_fail_buffer(): buffer must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(FAIL_IMAGE_INVALID_STATE, "sg_fail_image(): image must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(FAIL_SAMPLER_INVALID_STATE, "sg_fail_sampler(): sampler must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(FAIL_SHADER_INVALID_STATE, "sg_fail_shader(): shader must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(FAIL_PIPELINE_INVALID_STATE, "sg_fail_pipeline(): pipeline must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(FAIL_ATTACHMENTS_INVALID_STATE, "sg_fail_attachments(): attachments must be in ALLOC state") \
    _SG_LOGITEM_XMACRO(BUFFER_POOL_EXHAUSTED, "buffer pool exhausted") \
    _SG_LOGITEM_XMACRO(IMAGE_POOL_EXHAUSTED, "image pool exhausted") \
    _SG_LOGITEM_XMACRO(SAMPLER_POOL_EXHAUSTED, "sampler pool exhausted") \
    _SG_LOGITEM_XMACRO(SHADER_POOL_EXHAUSTED, "shader pool exhausted") \
    _SG_LOGITEM_XMACRO(PIPELINE_POOL_EXHAUSTED, "pipeline pool exhausted") \
    _SG_LOGITEM_XMACRO(PASS_POOL_EXHAUSTED, "pass pool exhausted") \
    _SG_LOGITEM_XMACRO(BEGINPASS_ATTACHMENT_INVALID, "sg_begin_pass: an attachment was provided that no longer exists") \
    _SG_LOGITEM_XMACRO(APPLY_BINDINGS_STORAGE_BUFFER_TRACKER_EXHAUSTED, "sg_apply_bindings: too many read/write storage buffers in pass (bump sg_desc.max_dispatch_calls_per_pass") \
    _SG_LOGITEM_XMACRO(DRAW_WITHOUT_BINDINGS, "attempting to draw without resource bindings") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_CANARY, "sg_buffer_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_IMMUTABLE_DYNAMIC_STREAM, "sg_buffer_desc.usage: only one of .immutable, .dynamic_update, .stream_update can be true") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_SEPARATE_BUFFER_TYPES, "sg_buffer_desc.usage: on WebGL2, only one of .vertex_buffer or .index_buffer can be true (check sg_features.separate_buffer_types)") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_EXPECT_NONZERO_SIZE, "sg_buffer_desc.size must be greater zero") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_EXPECT_MATCHING_DATA_SIZE, "sg_buffer_desc.size and .data.size must be equal") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_EXPECT_ZERO_DATA_SIZE, "sg_buffer_desc.data.size expected to be zero") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_EXPECT_NO_DATA, "sg_buffer_desc.data.ptr must be null for dynamic/stream buffers") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_EXPECT_DATA, "sg_buffer_desc: initial content data must be provided for immutable buffers without storage buffer usage") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_STORAGEBUFFER_SUPPORTED, "storage buffers not supported by the backend 3D API (requires OpenGL >= 4.3)") \
    _SG_LOGITEM_XMACRO(VALIDATE_BUFFERDESC_STORAGEBUFFER_SIZE_MULTIPLE_4, "size of storage buffers must be a multiple of 4") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDATA_NODATA, "sg_image_data: no data (.ptr and/or .size is zero)") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDATA_DATA_SIZE, "sg_image_data: data size doesn't match expected surface size") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_CANARY, "sg_image_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_IMMUTABLE_DYNAMIC_STREAM, "sg_image_desc.usage: only one of .immutable, .dynamic_update, .stream_update can be true") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDER_VS_STORAGE_ATTACHMENT, "sg_image_desc.usage: only one of .render_attachment or .storage_attachment can be true") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_WIDTH, "sg_image_desc.width must be > 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_HEIGHT, "sg_image_desc.height must be > 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_NONRT_PIXELFORMAT, "invalid pixel format for non-render-target image") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_MSAA_BUT_NO_ATTACHMENT, "non-attachment images cannot be multisampled") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_DEPTH_3D_IMAGE, "3D images cannot have a depth/stencil image format") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_ATTACHMENT_EXPECT_IMMUTABLE, "render/storage attachment images must be sg_image_usage.immutable") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_ATTACHMENT_EXPECT_NO_DATA, "render/storage attachment images cannot be initialized with data") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_NO_MSAA_SUPPORT, "multisampling not supported for this pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_NUM_MIPMAPS, "multisample images must have num_mipmaps == 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_3D_IMAGE, "3D images cannot have a sample_count > 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_CUBE_IMAGE, "cube images cannot have sample_count > 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_ARRAY_IMAGE, "array images cannot have sample_count > 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_RENDERATTACHMENT_PIXELFORMAT, "invalid pixel format for render attachment image") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_STORAGEATTACHMENT_PIXELFORMAT, "invalid pixel format for storage attachment image") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_STORAGEATTACHMENT_EXPECT_NO_MSAA, "storage attachment images cannot be multisampled") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_INJECTED_NO_DATA, "images with injected textures cannot be initialized with data") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_DYNAMIC_NO_DATA, "dynamic/stream-update images cannot be initialized with data") \
    _SG_LOGITEM_XMACRO(VALIDATE_IMAGEDESC_COMPRESSED_IMMUTABLE, "compressed images must be immutable") \
    _SG_LOGITEM_XMACRO(VALIDATE_SAMPLERDESC_CANARY, "sg_sampler_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_SAMPLERDESC_ANISTROPIC_REQUIRES_LINEAR_FILTERING, "sg_sampler_desc.max_anisotropy > 1 requires min/mag/mipmap_filter to be SG_FILTER_LINEAR") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_CANARY, "sg_shader_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_VERTEX_SOURCE, "vertex shader source code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_FRAGMENT_SOURCE, "fragment shader source code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_COMPUTE_SOURCE, "compute shader source code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_VERTEX_SOURCE_OR_BYTECODE, "vertex shader source or byte code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_FRAGMENT_SOURCE_OR_BYTECODE, "fragment shader source or byte code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_COMPUTE_SOURCE_OR_BYTECODE, "compute shader source or byte code expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_INVALID_SHADER_COMBO, "cannot combine compute shaders with vertex or fragment shaders") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_NO_BYTECODE_SIZE, "shader byte code length (in bytes) required") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_METAL_THREADS_PER_THREADGROUP, "sg_shader_desc.mtl_threads_per_threadgroup must be initialized for compute shaders (metal)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_NO_CONT_MEMBERS, "uniform block members must occupy continuous slots") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_SIZE_IS_ZERO, "bound uniform block size cannot be zero") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_METAL_BUFFER_SLOT_OUT_OF_RANGE, "uniform block 'msl_buffer_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_METAL_BUFFER_SLOT_COLLISION, "uniform block 'msl_buffer_n' must be unique across uniform blocks and storage buffers in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_HLSL_REGISTER_B_OUT_OF_RANGE, "uniform block 'hlsl_register_b_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_HLSL_REGISTER_B_COLLISION, "uniform block 'hlsl_register_b_n' must be unique across uniform blocks in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_WGSL_GROUP0_BINDING_OUT_OF_RANGE, "uniform block 'wgsl_group0_binding_n' is out of range (must be 0..15)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_WGSL_GROUP0_BINDING_COLLISION, "uniform block 'wgsl_group0_binding_n' must be unique across all uniform blocks") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_NO_MEMBERS, "GL backend requires uniform block member declarations") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_UNIFORM_GLSL_NAME, "uniform block member 'glsl_name' missing") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_SIZE_MISMATCH, "size of uniform block members doesn't match uniform block size") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_ARRAY_COUNT, "uniform array count must be >= 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_UNIFORMBLOCK_STD140_ARRAY_TYPE, "uniform arrays only allowed for FLOAT4, INT4, MAT4 in std140 layout") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_METAL_BUFFER_SLOT_OUT_OF_RANGE, "storage buffer 'msl_buffer_n' is out of range (must be 8..15)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_METAL_BUFFER_SLOT_COLLISION, "storage buffer 'msl_buffer_n' must be unique across uniform blocks and storage buffer in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_T_OUT_OF_RANGE, "storage buffer 'hlsl_register_t_n' is out of range (must be 0..23)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_T_COLLISION, "storage_buffer 'hlsl_register_t_n' must be unique across read-only storage buffers and images in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_U_OUT_OF_RANGE, "storage buffer 'hlsl_register_u_n' is out of range (must be 0..11)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_U_COLLISION, "storage_buffer 'hlsl_register_u_n' must be unique across read/write storage buffers and storage images in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_GLSL_BINDING_OUT_OF_RANGE, "storage buffer 'glsl_binding_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_GLSL_BINDING_COLLISION, "storage buffer 'glsl_binding_n' must be unique across shader stages") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "storage buffer 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEBUFFER_WGSL_GROUP1_BINDING_COLLISION, "storage buffer 'wgsl_group1_binding_n' must be unique across all images, samplers and storage buffers") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_EXPECT_COMPUTE_STAGE, "storage images are only allowed on the compute stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_METAL_TEXTURE_SLOT_OUT_OF_RANGE, "storage image 'msl_texture_n' is out of range (must be 0..19") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_METAL_TEXTURE_SLOT_COLLISION, "storage image 'msl_texture_n' must be unique across images and storage images in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_HLSL_REGISTER_U_OUT_OF_RANGE, "storage image 'hlsl_register_u_n' is out of range (must be 0..11)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_HLSL_REGISTER_U_COLLISION, "storage image 'hlsl_register_u_n' must be unique across storage images and read/write storage buffers in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_GLSL_BINDING_OUT_OF_RANGE, "storage image 'glsl_binding_n' is out of range (must be 0..4)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_GLSL_BINDING_COLLISION, "storage image 'glsl_binding_n' must be unique across shader stages") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_WGSL_GROUP2_BINDING_OUT_OF_RANGE, "storage image 'wgsl_group2_binding_n' is out of range (must be 0..7)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_STORAGEIMAGE_WGSL_GROUP2_BINDING_COLLISION, "storage image 'wgsl_group2_binding_n' must be unique in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_METAL_TEXTURE_SLOT_OUT_OF_RANGE, "image 'msl_texture_n' is out of range (must be 0..19)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_METAL_TEXTURE_SLOT_COLLISION, "image 'msl_texture_n' must be unique across images and storage images in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_HLSL_REGISTER_T_OUT_OF_RANGE, "image 'hlsl_register_t_n' is out of range (must be 0..23)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_HLSL_REGISTER_T_COLLISION, "image 'hlsl_register_t_n' must be unique across images and storage buffers in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "image 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_WGSL_GROUP1_BINDING_COLLISION, "image 'wgsl_group1_binding_n' must be unique across all images, samplers and storage buffers") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_METAL_SAMPLER_SLOT_OUT_OF_RANGE, "sampler 'msl_sampler_n' is out of range (must be 0..15)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_METAL_SAMPLER_SLOT_COLLISION, "sampler 'msl_sampler_n' must be unique in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_HLSL_REGISTER_S_OUT_OF_RANGE, "sampler 'hlsl_register_s_n' is out of rang (must be 0..15)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_HLSL_REGISTER_S_COLLISION, "sampler 'hlsl_register_s_n' must be unique in same shader stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_WGSL_GROUP1_BINDING_OUT_OF_RANGE, "sampler 'wgsl_group1_binding_n' is out of range (must be 0..127)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_WGSL_GROUP1_BINDING_COLLISION, "sampler 'wgsl_group1_binding_n' must be unique across all images, samplers and storage buffers") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_IMAGE_SLOT_OUT_OF_RANGE, "image-sampler-pair image slot index is out of range (sg_shader_desc.image_sampler_pairs[].image_slot)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_SAMPLER_SLOT_OUT_OF_RANGE, "image-sampler-pair sampler slot index is out of range (sg_shader_desc.image_sampler_pairs[].sampler_slot)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_IMAGE_STAGE_MISMATCH, "image-sampler-pair stage doesn't match referenced image stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_SAMPLER_STAGE_MISMATCH, "image-sampler-pair stage doesn't match referenced sampler stage") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_GLSL_NAME, "image-sampler-pair 'glsl_name' missing") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_NONFILTERING_SAMPLER_REQUIRED, "image sample type UNFILTERABLE_FLOAT, UINT, SINT can only be used with NONFILTERING sampler") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_COMPARISON_SAMPLER_REQUIRED, "image sample type DEPTH can only be used with COMPARISON sampler") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_IMAGE_NOT_REFERENCED_BY_IMAGE_SAMPLER_PAIRS, "one or more images are not referenced by by image-sampler-pairs (sg_shader_desc.image_sampler_pairs[].image_slot)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_SAMPLER_NOT_REFERENCED_BY_IMAGE_SAMPLER_PAIRS, "one or more samplers are not referenced by image-sampler-pairs (sg_shader_desc.image_sampler_pairs[].sampler_slot)") \
    _SG_LOGITEM_XMACRO(VALIDATE_SHADERDESC_ATTR_STRING_TOO_LONG, "vertex attribute name/semantic string too long (max len 16)") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_CANARY, "sg_pipeline_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_SHADER, "sg_pipeline_desc.shader missing or invalid") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_COMPUTE_SHADER_EXPECTED, "sg_pipeline_desc.shader must be a compute shader") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_NO_COMPUTE_SHADER_EXPECTED, "sg_pipeline_desc.compute is false, but shader is a compute shader") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_NO_CONT_ATTRS, "sg_pipeline_desc.layout.attrs is not continuous") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH, "sg_pipeline_desc.layout.attrs[].format is incompatible with sg_shader_desc.attrs[].base_type") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_LAYOUT_STRIDE4, "sg_pipeline_desc.layout.buffers[].stride must be multiple of 4") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_ATTR_SEMANTICS, "D3D11 missing vertex attribute semantics in shader") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_SHADER_READONLY_STORAGEBUFFERS, "sg_pipeline_desc.shader: only readonly storage buffer bindings allowed in render pipelines") \
    _SG_LOGITEM_XMACRO(VALIDATE_PIPELINEDESC_BLENDOP_MINMAX_REQUIRES_BLENDFACTOR_ONE, "SG_BLENDOP_MIN/MAX requires all blend factors to be SG_BLENDFACTOR_ONE") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_CANARY, "sg_attachments_desc not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_NO_ATTACHMENTS, "sg_attachments_desc no color, depth-stencil or storage attachments") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_NO_CONT_COLOR_ATTS, "color attachments must occupy continuous slots") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_IMAGE, "color attachment image is not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_MIPLEVEL, "color attachment mip level is higher than number of mipmaps in image") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_FACE, "color attachment image is cubemap, but face index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_LAYER, "color attachment image is array texture, but layer index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_SLICE, "color attachment image is 3d texture, but slice value is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_IMAGE_NO_RENDERATTACHMENT, "color attachment images must be sg_image_desc.usage.render_attachment=true") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_COLOR_INV_PIXELFORMAT, "color attachment images must be renderable color pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_IMAGE_SIZES, "all color and depth attachment images must have the same size") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_IMAGE_SAMPLE_COUNTS, "all color and depth attachment images must have the same sample count") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_COLOR_IMAGE_MSAA, "resolve attachments must have a color attachment image with sample count > 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE, "resolve attachment image not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_SAMPLE_COUNT, "pass resolve attachment image sample count must be 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_MIPLEVEL, "resolve attachment mip level is higher than number of mipmaps in image") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_FACE, "resolve attachment is cubemap, but face index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_LAYER, "resolve attachment is array texture, but layer index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_SLICE, "resolve attachment is 3d texture, but slice value is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_NO_RT, "resolve attachment image must have render_target=true") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_SIZES, "resolve attachment size must match color attachment image size") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_FORMAT, "resolve attachment pixel format must match color attachment pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_INV_PIXELFORMAT, "depth attachment image must be depth or depth-stencil pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE, "depth attachment image is not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_MIPLEVEL, "depth attachment mip level is higher than number of mipmaps in image") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_FACE, "depth attachment image is cubemap, but face index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_LAYER, "depth attachment image is array texture, but layer index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_SLICE, "depth attachment image is 3d texture, but slice value is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_NO_RENDERATTACHMENT, "depth attachment image must be sg_image_desc.usage.render_attachment=true") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_SIZES, "depth attachment image size must match color attachment image size") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_SAMPLE_COUNT, "depth attachment sample count must match color attachment sample count") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_IMAGE, "storage attachment image is not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_MIPLEVEL, "storage attachment mip level is higher than number of mipmaps in image") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_FACE, "storage attachment image is cubemap, but face index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_LAYER, "storage attachment image is array texture, but layer index is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_SLICE, "storage attachment image is 3d texture, but slice value is too big") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_IMAGE_NO_STORAGEATTACHMENT, "storage attachment images must be sg_image_desc.usage.storage_attachment=true") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_STORAGE_INV_PIXELFORMAT, "storage attachment pixel format must have .compute_readwrite or .compute_writeonly capabilities") \
    _SG_LOGITEM_XMACRO(VALIDATE_ATTACHMENTSDESC_RENDER_VS_STORAGE_ATTACHMENTS, "cannot use color/depth and storage attachment images on the same sg_attachments object") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_CANARY, "sg_begin_pass: pass struct not initialized") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_ATTACHMENTS_EXISTS, "sg_begin_pass: attachments object no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_ATTACHMENTS_VALID, "sg_begin_pass: attachments object not in resource state VALID") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_COMPUTEPASS_STORAGE_ATTACHMENTS_ONLY, "sg_begin_pass: only storage attachments allowed on compute pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_RENDERPASS_RENDER_ATTACHMENTS_ONLY, "sg_begin_pass: a render pass cannot have storage attachments") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_COLOR_ATTACHMENT_IMAGE, "sg_begin_pass: one or more color attachment images are not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_RESOLVE_ATTACHMENT_IMAGE, "sg_begin_pass: one or more resolve attachment images are not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_DEPTHSTENCIL_ATTACHMENT_IMAGE, "sg_begin_pass: one or more depth-stencil attachment images are not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_STORAGE_ATTACHMENT_IMAGE, "sg_begin_pass: one or more storage attachment images are not valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_WIDTH, "sg_begin_pass: expected pass.swapchain.width > 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_WIDTH_NOTSET, "sg_begin_pass: expected pass.swapchain.width == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_HEIGHT, "sg_begin_pass: expected pass.swapchain.height > 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_HEIGHT_NOTSET, "sg_begin_pass: expected pass.swapchain.height == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_SAMPLECOUNT, "sg_begin_pass: expected pass.swapchain.sample_count > 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_SAMPLECOUNT_NOTSET, "sg_begin_pass: expected pass.swapchain.sample_count == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_COLORFORMAT, "sg_begin_pass: expected pass.swapchain.color_format to be valid") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_COLORFORMAT_NOTSET, "sg_begin_pass: expected pass.swapchain.color_format to be unset") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_DEPTHFORMAT_NOTSET, "sg_begin_pass: expected pass.swapchain.depth_format to be unset") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_CURRENTDRAWABLE, "sg_begin_pass: expected pass.swapchain.metal.current_drawable != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_CURRENTDRAWABLE_NOTSET, "sg_begin_pass: expected pass.swapchain.metal.current_drawable == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_DEPTHSTENCILTEXTURE, "sg_begin_pass: expected pass.swapchain.metal.depth_stencil_texture != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_DEPTHSTENCILTEXTURE_NOTSET, "sg_begin_pass: expected pass.swapchain.metal.depth_stencil_texture == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_MSAACOLORTEXTURE, "sg_begin_pass: expected pass.swapchain.metal.msaa_color_texture != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_MSAACOLORTEXTURE_NOTSET, "sg_begin_pass: expected pass.swapchain.metal.msaa_color_texture == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RENDERVIEW, "sg_begin_pass: expected pass.swapchain.d3d11.render_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RENDERVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.d3d11.render_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RESOLVEVIEW, "sg_begin_pass: expected pass.swapchain.d3d11.resolve_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RESOLVEVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.d3d11.resolve_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_DEPTHSTENCILVIEW, "sg_begin_pass: expected pass.swapchain.d3d11.depth_stencil_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_DEPTHSTENCILVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.d3d11.depth_stencil_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RENDERVIEW, "sg_begin_pass: expected pass.swapchain.wgpu.render_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RENDERVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.wgpu.render_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RESOLVEVIEW, "sg_begin_pass: expected pass.swapchain.wgpu.resolve_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RESOLVEVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.wgpu.resolve_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_DEPTHSTENCILVIEW, "sg_begin_pass: expected pass.swapchain.wgpu.depth_stencil_view != 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_DEPTHSTENCILVIEW_NOTSET, "sg_begin_pass: expected pass.swapchain.wgpu.depth_stencil_view == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_BEGINPASS_SWAPCHAIN_GL_EXPECT_FRAMEBUFFER_NOTSET, "sg_begin_pass: expected pass.swapchain.gl.framebuffer == 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_AVP_RENDERPASS_EXPECTED, "sg_apply_viewport: must be called in a render pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_ASR_RENDERPASS_EXPECTED, "sg_apply_scissor_rect: must be called in a render pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PIPELINE_VALID_ID, "sg_apply_pipeline: invalid pipeline id provided") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PIPELINE_EXISTS, "sg_apply_pipeline: pipeline object no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PIPELINE_VALID, "sg_apply_pipeline: pipeline object not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PASS_EXPECTED, "sg_apply_pipeline: must be called in a pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PIPELINE_SHADER_ALIVE, "sg_apply_pipeline: shader object associated with pipeline no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_PIPELINE_SHADER_VALID, "sg_apply_pipeline: shader object associated with pipeline not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_COMPUTEPASS_EXPECTED, "sg_apply_pipeline: trying to apply compute pipeline in render pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_RENDERPASS_EXPECTED, "sg_apply_pipeline: trying to apply render pipeline in compute pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_CURPASS_ATTACHMENTS_ALIVE, "sg_apply_pipeline: current pass attachments no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_CURPASS_ATTACHMENTS_VALID, "sg_apply_pipeline: current pass attachments not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_ATT_COUNT, "sg_apply_pipeline: number of pipeline color attachments doesn't match number of pass color attachments") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_COLOR_FORMAT, "sg_apply_pipeline: pipeline color attachment pixel format doesn't match pass color attachment pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_DEPTH_FORMAT, "sg_apply_pipeline: pipeline depth pixel_format doesn't match pass depth attachment pixel format") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_SAMPLE_COUNT, "sg_apply_pipeline: pipeline MSAA sample count doesn't match render pass attachment sample count") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_EXPECTED_STORAGE_ATTACHMENT_IMAGE, "sg_apply_pipeline: shader expects storage image binding but compute pass doesn't have storage attachment image at expected bind slot") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_ALIVE, "sg_apply_pipeline: compute pass storage image attachment no longer exists") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_VALID, "sg_apply_pipeline: compute pass storage image attachment is not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_STORAGE_ATTACHMENT_PIXELFORMAT, "sg_apply_pipeline: compute pass storage image attachment pixel format doesn't match sg_shader_desc.storage_images[].access_format") \
    _SG_LOGITEM_XMACRO(VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_TYPE, "sg_apply_pipeline: compute pass storage image attachment image type doesn't match sg_shader_desc.storage_images[].image_type") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_PASS_EXPECTED, "sg_apply_bindings: must be called in a pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EMPTY_BINDINGS, "sg_apply_bindings: the provided sg_bindings struct is empty") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_NO_PIPELINE, "sg_apply_bindings: must be called after sg_apply_pipeline") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_PIPELINE_ALIVE, "sg_apply_bindings: currently applied pipeline object no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_PIPELINE_VALID, "sg_apply_bindings: currently applied pipeline object not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_PIPELINE_SHADER_ALIVE, "sg_apply_bindings: shader associated with currently applied pipeline is no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_PIPELINE_SHADER_VALID, "sg_apply_bindings: shader associated with currently applied pipeline is not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_COMPUTE_EXPECTED_NO_VBS, "sg_apply_bindings: vertex buffer bindings not allowed in a compute pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_COMPUTE_EXPECTED_NO_IB, "sg_apply_bindings: index buffer binding not allowed in compute pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_VB, "sg_apply_bindings: vertex buffer binding is missing or buffer handle is invalid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_VB_ALIVE, "sg_apply_bindings: vertex buffer no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_VB_TYPE, "sg_apply_bindings: buffer in vertex buffer slot doesn't have vertex buffer usage (sg_buffer_desc.usage.storage_buffer)") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_VB_OVERFLOW, "sg_apply_bindings: buffer in vertex buffer slot is overflown") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_NO_IB, "sg_apply_bindings: pipeline object defines indexed rendering, but no index buffer provided") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IB, "sg_apply_bindings: pipeline object defines non-indexed rendering, but index buffer provided") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IB_ALIVE, "sg_apply_bindings: index buffer no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IB_TYPE, "sg_apply_bindings: buffer in index buffer slot doesn't have index buffer usage (sg_buffer_desc.usage.index_buffer)") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IB_OVERFLOW, "sg_apply_bindings: buffer in index buffer slot is overflown") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_IMAGE_BINDING, "sg_apply_bindings: image binding is missing or the image handle is invalid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMG_ALIVE, "sg_apply_bindings: bound image no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_TYPE_MISMATCH, "sg_apply_bindings: type of bound image doesn't match shader desc") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_MULTISAMPLED_IMAGE, "sg_apply_bindings: expected image with sample_count > 1") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_MSAA, "sg_apply_bindings: cannot bind image with sample_count>1") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_FILTERABLE_IMAGE, "sg_apply_bindings: filterable image expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_DEPTH_IMAGE, "sg_apply_bindings: depth image expected") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_SAMPLER_BINDING, "sg_apply_bindings: sampler binding is missing or the sampler handle is invalid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_UNEXPECTED_SAMPLER_COMPARE_NEVER, "sg_apply_bindings: shader expects SG_SAMPLERTYPE_COMPARISON but sampler has SG_COMPAREFUNC_NEVER") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_SAMPLER_COMPARE_NEVER, "sg_apply_bindings: shader expects SG_SAMPLERTYPE_FILTERING or SG_SAMPLERTYPE_NONFILTERING but sampler doesn't have SG_COMPAREFUNC_NEVER") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_NONFILTERING_SAMPLER, "sg_apply_bindings: shader expected SG_SAMPLERTYPE_NONFILTERING, but sampler has SG_FILTER_LINEAR filters") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_SMP_ALIVE, "sg_apply_bindings: bound sampler no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_EXPECTED_STORAGEBUFFER_BINDING, "sg_apply_bindings: storage buffer binding is missing or the buffer handle is invalid") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_STORAGEBUFFER_ALIVE, "sg_apply_bindings: bound storage buffer no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_STORAGEBUFFER_BINDING_BUFFERTYPE, "sg_apply_bindings: buffer bound to storage buffer slot doesn't have storage buffer usage (sg_buffer_desc.usage.storage_buffer)") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_STORAGEBUFFER_READWRITE_IMMUTABLE, "sg_apply_bindings: storage buffers bound as read/write must have usage immutable") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_BINDING_VS_DEPTHSTENCIL_ATTACHMENT, "sg_apply_bindings: cannot bind image in the same pass it is used as depth-stencil attachment") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_BINDING_VS_COLOR_ATTACHMENT, "sg_apply_bindings: cannot bind image in the same pass it is used as color attachment") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_BINDING_VS_RESOLVE_ATTACHMENT, "sg_apply_bindings: cannot bind image in the same pass it is used as resolve attachment") \
    _SG_LOGITEM_XMACRO(VALIDATE_ABND_IMAGE_BINDING_VS_STORAGE_ATTACHMENT, "sg_apply_bindings: cannot bind image in the same pass it is used as storage attachment") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_PASS_EXPECTED, "sg_apply_uniforms: must be called in a pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_NO_PIPELINE, "sg_apply_uniforms: must be called after sg_apply_pipeline()") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_PIPELINE_ALIVE, "sg_apply_uniforms: currently applied pipeline object no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_PIPELINE_VALID, "sg_apply_uniforms: currently applied pipeline object not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_PIPELINE_SHADER_ALIVE, "sg_apply_uniforms: shader associated with currently applied pipeline is no longer alive") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_PIPELINE_SHADER_VALID, "sg_apply_uniforms: shader associated with currently applied pipeline is not in valid state") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_NO_UNIFORMBLOCK_AT_SLOT, "sg_apply_uniforms: no uniform block declaration at this shader stage UB slot") \
    _SG_LOGITEM_XMACRO(VALIDATE_AU_SIZE, "sg_apply_uniforms: data size doesn't match declared uniform block size") \
    _SG_LOGITEM_XMACRO(VALIDATE_DRAW_RENDERPASS_EXPECTED, "sg_draw: must be called in a render pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_DRAW_BASEELEMENT, "sg_draw: base_element cannot be < 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_DRAW_NUMELEMENTS, "sg_draw: num_elements cannot be < 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_DRAW_NUMINSTANCES, "sg_draw: num_instances cannot be < 0") \
    _SG_LOGITEM_XMACRO(VALIDATE_DRAW_REQUIRED_BINDINGS_OR_UNIFORMS_MISSING, "sg_draw: call to sg_apply_bindings() and/or sg_apply_uniforms() missing after sg_apply_pipeline()") \
    _SG_LOGITEM_XMACRO(VALIDATE_DISPATCH_COMPUTEPASS_EXPECTED, "sg_dispatch: must be called in a compute pass") \
    _SG_LOGITEM_XMACRO(VALIDATE_DISPATCH_NUMGROUPSX, "sg_dispatch: num_groups_x must be >=0 and <65536") \
    _SG_LOGITEM_XMACRO(VALIDATE_DISPATCH_NUMGROUPSY, "sg_dispatch: num_groups_y must be >=0 and <65536") \
    _SG_LOGITEM_XMACRO(VALIDATE_DISPATCH_NUMGROUPSZ, "sg_dispatch: num_groups_z must be >=0 and <65536") \
    _SG_LOGITEM_XMACRO(VALIDATE_DISPATCH_REQUIRED_BINDINGS_OR_UNIFORMS_MISSING, "sg_dispatch: call to sg_apply_bindings() and/or sg_apply_uniforms() missing after sg_apply_pipeline()") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDATEBUF_USAGE, "sg_update_buffer: cannot update immutable buffer") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDATEBUF_SIZE, "sg_update_buffer: update size is bigger than buffer size") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDATEBUF_ONCE, "sg_update_buffer: only one update allowed per buffer and frame") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDATEBUF_APPEND, "sg_update_buffer: cannot call sg_update_buffer and sg_append_buffer in same frame") \
    _SG_LOGITEM_XMACRO(VALIDATE_APPENDBUF_USAGE, "sg_append_buffer: cannot append to immutable buffer") \
    _SG_LOGITEM_XMACRO(VALIDATE_APPENDBUF_SIZE, "sg_append_buffer: overall appended size is bigger than buffer size") \
    _SG_LOGITEM_XMACRO(VALIDATE_APPENDBUF_UPDATE, "sg_append_buffer: cannot call sg_append_buffer and sg_update_buffer in same frame") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDIMG_USAGE, "sg_update_image: cannot update immutable image") \
    _SG_LOGITEM_XMACRO(VALIDATE_UPDIMG_ONCE, "sg_update_image: only one update allowed per image and frame") \
    _SG_LOGITEM_XMACRO(VALIDATION_FAILED, "validation layer checks failed") \

#define _SG_LOGITEM_XMACRO(item,msg) SG_LOGITEM_##item,
typedef enum sg_log_item {
    _SG_LOG_ITEMS
} sg_log_item;
#undef _SG_LOGITEM_XMACRO

/*
    sg_desc

    The sg_desc struct contains configuration values for sokol_gfx,
    it is used as parameter to the sg_setup() call.

    The default configuration is:

    .buffer_pool_size               128
    .image_pool_size                128
    .sampler_pool_size              64
    .shader_pool_size               32
    .pipeline_pool_size             64
    .attachments_pool_size          16
    .uniform_buffer_size            4 MB (4*1024*1024)
    .max_dispatch_calls_per_pass    1024
    .max_commit_listeners           1024
    .disable_validation             false
    .mtl_force_managed_storage_mode false
    .wgpu_disable_bindgroups_cache  false
    .wgpu_bindgroups_cache_size     1024

    .allocator.alloc_fn     0 (in this case, malloc() will be called)
    .allocator.free_fn      0 (in this case, free() will be called)
    .allocator.user_data    0

    .environment.defaults.color_format: default value depends on selected backend:
        all GL backends:    SG_PIXELFORMAT_RGBA8
        Metal and D3D11:    SG_PIXELFORMAT_BGRA8
        WebGPU:             *no default* (must be queried from WebGPU swapchain object)
    .environment.defaults.depth_format: SG_PIXELFORMAT_DEPTH_STENCIL
    .environment.defaults.sample_count: 1

    Metal specific:
        (NOTE: All Objective-C object references are transferred through
        a bridged cast (__bridge const void*) to sokol_gfx, which will use an
        unretained bridged cast (__bridge id<xxx>) to retrieve the Objective-C
        references back. Since the bridge cast is unretained, the caller
        must hold a strong reference to the Objective-C object until sg_setup()
        returns.

        .mtl_force_managed_storage_mode
            when enabled, Metal buffers and texture resources are created in managed storage
            mode, otherwise sokol-gfx will decide whether to create buffers and
            textures in managed or shared storage mode (this is mainly a debugging option)
        .mtl_use_command_buffer_with_retained_references
            when true, the sokol-gfx Metal backend will use Metal command buffers which
            bump the reference count of resource objects as long as they are inflight,
            this is slower than the default command-buffer-with-unretained-references
            method, this may be a workaround when confronted with lifetime validation
            errors from the Metal validation layer until a proper fix has been implemented
        .environment.metal.device
            a pointer to the MTLDevice object

    D3D11 specific:
        .environment.d3d11.device
            a pointer to the ID3D11Device object, this must have been created
            before sg_setup() is called
        .environment.d3d11.device_context
            a pointer to the ID3D11DeviceContext object
        .d3d11_shader_debugging
            set this to true to compile shaders which are provided as HLSL source
            code with debug information and without optimization, this allows
            shader debugging in tools like RenderDoc, to output source code
            instead of byte code from sokol-shdc, omit the `--binary` cmdline
            option

    WebGPU specific:
        .wgpu_disable_bindgroups_cache
            When this is true, the WebGPU backend will create and immediately
            release a BindGroup object in the sg_apply_bindings() call, only
            use this for debugging purposes.
        .wgpu_bindgroups_cache_size
            The size of the bindgroups cache for re-using BindGroup objects
            between sg_apply_bindings() calls. The smaller the cache size,
            the more likely are cache slot collisions which will cause
            a BindGroups object to be destroyed and a new one created.
            Use the information returned by sg_query_stats() to check
            if this is a frequent occurrence, and increase the cache size as
            needed (the default is 1024).
            NOTE: wgpu_bindgroups_cache_size must be a power-of-2 number!
        .environment.wgpu.device
            a WGPUDevice handle

    When using sokol_gfx.h and sokol_app.h together, consider using the
    helper function sglue_environment() in the sokol_glue.h header to
    initialize the sg_desc.environment nested struct. sglue_environment() returns
    a completely initialized sg_environment struct with information
    provided by sokol_app.h.
*/
typedef struct sg_environment_defaults {
    sg_pixel_format color_format;
    sg_pixel_format depth_format;
    int sample_count;
} sg_environment_defaults;

typedef struct sg_metal_environment {
    const void* device;
} sg_metal_environment;

typedef struct sg_d3d11_environment {
    const void* device;
    const void* device_context;
} sg_d3d11_environment;

typedef struct sg_wgpu_environment {
    const void* device;
} sg_wgpu_environment;

typedef struct sg_environment {
    sg_environment_defaults defaults;
    sg_metal_environment metal;
    sg_d3d11_environment d3d11;
    sg_wgpu_environment wgpu;
} sg_environment;

/*
    sg_commit_listener

    Used with function sg_add_commit_listener() to add a callback
    which will be called in sg_commit(). This is useful for libraries
    building on top of sokol-gfx to be notified about when a frame
    ends (instead of having to guess, or add a manual 'new-frame'
    function.
*/
typedef struct sg_commit_listener {
    void (*func)(void* user_data);
    void* user_data;
} sg_commit_listener;

/*
    sg_allocator

    Used in sg_desc to provide custom memory-alloc and -free functions
    to sokol_gfx.h. If memory management should be overridden, both the
    alloc_fn and free_fn function must be provided (e.g. it's not valid to
    override one function but not the other).
*/
typedef struct sg_allocator {
    void* (*alloc_fn)(size_t size, void* user_data);
    void (*free_fn)(void* ptr, void* user_data);
    void* user_data;
} sg_allocator;

/*
    sg_logger

    Used in sg_desc to provide a logging function. Please be aware
    that without logging function, sokol-gfx will be completely
    silent, e.g. it will not report errors, warnings and
    validation layer messages. For maximum error verbosity,
    compile in debug mode (e.g. NDEBUG *not* defined) and provide a
    compatible logger function in the sg_setup() call
    (for instance the standard logging function from sokol_log.h).
*/
typedef struct sg_logger {
    void (*func)(
        const char* tag,                // always "sg"
        uint32_t log_level,             // 0=panic, 1=error, 2=warning, 3=info
        uint32_t log_item_id,           // SG_LOGITEM_*
        const char* message_or_null,    // a message string, may be nullptr in release mode
        uint32_t line_nr,               // line number in sokol_gfx.h
        const char* filename_or_null,   // source filename, may be nullptr in release mode
        void* user_data);
    void* user_data;
} sg_logger;

typedef struct sg_desc {
    uint32_t _start_canary;
    int buffer_pool_size;
    int image_pool_size;
    int sampler_pool_size;
    int shader_pool_size;
    int pipeline_pool_size;
    int attachments_pool_size;
    int uniform_buffer_size;
    int max_dispatch_calls_per_pass;    // max expected number of dispatch calls per pass (default: 1024)
    int max_commit_listeners;
    bool disable_validation;    // disable validation layer even in debug mode, useful for tests
    bool d3d11_shader_debugging;    // if true, HLSL shaders are compiled with D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION
    bool mtl_force_managed_storage_mode; // for debugging: use Metal managed storage mode for resources even with UMA
    bool mtl_use_command_buffer_with_retained_references;    // Metal: use a managed MTLCommandBuffer which ref-counts used resources
    bool wgpu_disable_bindgroups_cache;  // set to true to disable the WebGPU backend BindGroup cache
    int wgpu_bindgroups_cache_size;      // number of slots in the WebGPU bindgroup cache (must be 2^N)
    sg_allocator allocator;
    sg_logger logger; // optional log function override
    sg_environment environment;
    uint32_t _end_canary;
} sg_desc;

// setup and misc functions
SOKOL_GFX_API_DECL void sg_setup(const sg_desc* desc);
SOKOL_GFX_API_DECL void sg_shutdown(void);
SOKOL_GFX_API_DECL bool sg_isvalid(void);
SOKOL_GFX_API_DECL void sg_reset_state_cache(void);
SOKOL_GFX_API_DECL sg_trace_hooks sg_install_trace_hooks(const sg_trace_hooks* trace_hooks);
SOKOL_GFX_API_DECL void sg_push_debug_group(const char* name);
SOKOL_GFX_API_DECL void sg_pop_debug_group(void);
SOKOL_GFX_API_DECL bool sg_add_commit_listener(sg_commit_listener listener);
SOKOL_GFX_API_DECL bool sg_remove_commit_listener(sg_commit_listener listener);

// resource creation, destruction and updating
SOKOL_GFX_API_DECL sg_buffer sg_make_buffer(const sg_buffer_desc* desc);
SOKOL_GFX_API_DECL sg_image sg_make_image(const sg_image_desc* desc);
SOKOL_GFX_API_DECL sg_sampler sg_make_sampler(const sg_sampler_desc* desc);
SOKOL_GFX_API_DECL sg_shader sg_make_shader(const sg_shader_desc* desc);
SOKOL_GFX_API_DECL sg_pipeline sg_make_pipeline(const sg_pipeline_desc* desc);
SOKOL_GFX_API_DECL sg_attachments sg_make_attachments(const sg_attachments_desc* desc);
SOKOL_GFX_API_DECL void sg_destroy_buffer(sg_buffer buf);
SOKOL_GFX_API_DECL void sg_destroy_image(sg_image img);
SOKOL_GFX_API_DECL void sg_destroy_sampler(sg_sampler smp);
SOKOL_GFX_API_DECL void sg_destroy_shader(sg_shader shd);
SOKOL_GFX_API_DECL void sg_destroy_pipeline(sg_pipeline pip);
SOKOL_GFX_API_DECL void sg_destroy_attachments(sg_attachments atts);
SOKOL_GFX_API_DECL void sg_update_buffer(sg_buffer buf, const sg_range* data);
SOKOL_GFX_API_DECL void sg_update_image(sg_image img, const sg_image_data* data);
SOKOL_GFX_API_DECL int sg_append_buffer(sg_buffer buf, const sg_range* data);
SOKOL_GFX_API_DECL bool sg_query_buffer_overflow(sg_buffer buf);
SOKOL_GFX_API_DECL bool sg_query_buffer_will_overflow(sg_buffer buf, size_t size);

// render and compute functions
SOKOL_GFX_API_DECL void sg_begin_pass(const sg_pass* pass);
SOKOL_GFX_API_DECL void sg_apply_viewport(int x, int y, int width, int height, bool origin_top_left);
SOKOL_GFX_API_DECL void sg_apply_viewportf(float x, float y, float width, float height, bool origin_top_left);
SOKOL_GFX_API_DECL void sg_apply_scissor_rect(int x, int y, int width, int height, bool origin_top_left);
SOKOL_GFX_API_DECL void sg_apply_scissor_rectf(float x, float y, float width, float height, bool origin_top_left);
SOKOL_GFX_API_DECL void sg_apply_pipeline(sg_pipeline pip);
SOKOL_GFX_API_DECL void sg_apply_bindings(const sg_bindings* bindings);
SOKOL_GFX_API_DECL void sg_apply_uniforms(int ub_slot, const sg_range* data);
SOKOL_GFX_API_DECL void sg_draw(int base_element, int num_elements, int num_instances);
SOKOL_GFX_API_DECL void sg_dispatch(int num_groups_x, int num_groups_y, int num_groups_z);
SOKOL_GFX_API_DECL void sg_end_pass(void);
SOKOL_GFX_API_DECL void sg_commit(void);

// getting information
SOKOL_GFX_API_DECL sg_desc sg_query_desc(void);
SOKOL_GFX_API_DECL sg_backend sg_query_backend(void);
SOKOL_GFX_API_DECL sg_features sg_query_features(void);
SOKOL_GFX_API_DECL sg_limits sg_query_limits(void);
SOKOL_GFX_API_DECL sg_pixelformat_info sg_query_pixelformat(sg_pixel_format fmt);
SOKOL_GFX_API_DECL int sg_query_row_pitch(sg_pixel_format fmt, int width, int row_align_bytes);
SOKOL_GFX_API_DECL int sg_query_surface_pitch(sg_pixel_format fmt, int width, int height, int row_align_bytes);
// get current state of a resource (INITIAL, ALLOC, VALID, FAILED, INVALID)
SOKOL_GFX_API_DECL sg_resource_state sg_query_buffer_state(sg_buffer buf);
SOKOL_GFX_API_DECL sg_resource_state sg_query_image_state(sg_image img);
SOKOL_GFX_API_DECL sg_resource_state sg_query_sampler_state(sg_sampler smp);
SOKOL_GFX_API_DECL sg_resource_state sg_query_shader_state(sg_shader shd);
SOKOL_GFX_API_DECL sg_resource_state sg_query_pipeline_state(sg_pipeline pip);
SOKOL_GFX_API_DECL sg_resource_state sg_query_attachments_state(sg_attachments atts);
// get runtime information about a resource
SOKOL_GFX_API_DECL sg_buffer_info sg_query_buffer_info(sg_buffer buf);
SOKOL_GFX_API_DECL sg_image_info sg_query_image_info(sg_image img);
SOKOL_GFX_API_DECL sg_sampler_info sg_query_sampler_info(sg_sampler smp);
SOKOL_GFX_API_DECL sg_shader_info sg_query_shader_info(sg_shader shd);
SOKOL_GFX_API_DECL sg_pipeline_info sg_query_pipeline_info(sg_pipeline pip);
SOKOL_GFX_API_DECL sg_attachments_info sg_query_attachments_info(sg_attachments atts);
// get desc structs matching a specific resource (NOTE that not all creation attributes may be provided)
SOKOL_GFX_API_DECL sg_buffer_desc sg_query_buffer_desc(sg_buffer buf);
SOKOL_GFX_API_DECL sg_image_desc sg_query_image_desc(sg_image img);
SOKOL_GFX_API_DECL sg_sampler_desc sg_query_sampler_desc(sg_sampler smp);
SOKOL_GFX_API_DECL sg_shader_desc sg_query_shader_desc(sg_shader shd);
SOKOL_GFX_API_DECL sg_pipeline_desc sg_query_pipeline_desc(sg_pipeline pip);
SOKOL_GFX_API_DECL sg_attachments_desc sg_query_attachments_desc(sg_attachments atts);
// get resource creation desc struct with their default values replaced
SOKOL_GFX_API_DECL sg_buffer_desc sg_query_buffer_defaults(const sg_buffer_desc* desc);
SOKOL_GFX_API_DECL sg_image_desc sg_query_image_defaults(const sg_image_desc* desc);
SOKOL_GFX_API_DECL sg_sampler_desc sg_query_sampler_defaults(const sg_sampler_desc* desc);
SOKOL_GFX_API_DECL sg_shader_desc sg_query_shader_defaults(const sg_shader_desc* desc);
SOKOL_GFX_API_DECL sg_pipeline_desc sg_query_pipeline_defaults(const sg_pipeline_desc* desc);
SOKOL_GFX_API_DECL sg_attachments_desc sg_query_attachments_defaults(const sg_attachments_desc* desc);
// assorted query functions
SOKOL_GFX_API_DECL size_t sg_query_buffer_size(sg_buffer buf);
SOKOL_GFX_API_DECL sg_buffer_usage sg_query_buffer_usage(sg_buffer buf);
SOKOL_GFX_API_DECL sg_image_type sg_query_image_type(sg_image img);
SOKOL_GFX_API_DECL int sg_query_image_width(sg_image img);
SOKOL_GFX_API_DECL int sg_query_image_height(sg_image img);
SOKOL_GFX_API_DECL int sg_query_image_num_slices(sg_image img);
SOKOL_GFX_API_DECL int sg_query_image_num_mipmaps(sg_image img);
SOKOL_GFX_API_DECL sg_pixel_format sg_query_image_pixelformat(sg_image img);
SOKOL_GFX_API_DECL sg_image_usage sg_query_image_usage(sg_image img);
SOKOL_GFX_API_DECL int sg_query_image_sample_count(sg_image img);

// separate resource allocation and initialization (for async setup)
SOKOL_GFX_API_DECL sg_buffer sg_alloc_buffer(void);
SOKOL_GFX_API_DECL sg_image sg_alloc_image(void);
SOKOL_GFX_API_DECL sg_sampler sg_alloc_sampler(void);
SOKOL_GFX_API_DECL sg_shader sg_alloc_shader(void);
SOKOL_GFX_API_DECL sg_pipeline sg_alloc_pipeline(void);
SOKOL_GFX_API_DECL sg_attachments sg_alloc_attachments(void);
SOKOL_GFX_API_DECL void sg_dealloc_buffer(sg_buffer buf);
SOKOL_GFX_API_DECL void sg_dealloc_image(sg_image img);
SOKOL_GFX_API_DECL void sg_dealloc_sampler(sg_sampler smp);
SOKOL_GFX_API_DECL void sg_dealloc_shader(sg_shader shd);
SOKOL_GFX_API_DECL void sg_dealloc_pipeline(sg_pipeline pip);
SOKOL_GFX_API_DECL void sg_dealloc_attachments(sg_attachments attachments);
SOKOL_GFX_API_DECL void sg_init_buffer(sg_buffer buf, const sg_buffer_desc* desc);
SOKOL_GFX_API_DECL void sg_init_image(sg_image img, const sg_image_desc* desc);
SOKOL_GFX_API_DECL void sg_init_sampler(sg_sampler smg, const sg_sampler_desc* desc);
SOKOL_GFX_API_DECL void sg_init_shader(sg_shader shd, const sg_shader_desc* desc);
SOKOL_GFX_API_DECL void sg_init_pipeline(sg_pipeline pip, const sg_pipeline_desc* desc);
SOKOL_GFX_API_DECL void sg_init_attachments(sg_attachments attachments, const sg_attachments_desc* desc);
SOKOL_GFX_API_DECL void sg_uninit_buffer(sg_buffer buf);
SOKOL_GFX_API_DECL void sg_uninit_image(sg_image img);
SOKOL_GFX_API_DECL void sg_uninit_sampler(sg_sampler smp);
SOKOL_GFX_API_DECL void sg_uninit_shader(sg_shader shd);
SOKOL_GFX_API_DECL void sg_uninit_pipeline(sg_pipeline pip);
SOKOL_GFX_API_DECL void sg_uninit_attachments(sg_attachments atts);
SOKOL_GFX_API_DECL void sg_fail_buffer(sg_buffer buf);
SOKOL_GFX_API_DECL void sg_fail_image(sg_image img);
SOKOL_GFX_API_DECL void sg_fail_sampler(sg_sampler smp);
SOKOL_GFX_API_DECL void sg_fail_shader(sg_shader shd);
SOKOL_GFX_API_DECL void sg_fail_pipeline(sg_pipeline pip);
SOKOL_GFX_API_DECL void sg_fail_attachments(sg_attachments atts);

// frame stats
SOKOL_GFX_API_DECL void sg_enable_frame_stats(void);
SOKOL_GFX_API_DECL void sg_disable_frame_stats(void);
SOKOL_GFX_API_DECL bool sg_frame_stats_enabled(void);
SOKOL_GFX_API_DECL sg_frame_stats sg_query_frame_stats(void);

/* Backend-specific structs and functions, these may come in handy for mixing
   sokol-gfx rendering with 'native backend' rendering functions.

   This group of functions will be expanded as needed.
*/

typedef struct sg_d3d11_buffer_info {
    const void* buf;      // ID3D11Buffer*
} sg_d3d11_buffer_info;

typedef struct sg_d3d11_image_info {
    const void* tex2d;    // ID3D11Texture2D*
    const void* tex3d;    // ID3D11Texture3D*
    const void* res;      // ID3D11Resource* (either tex2d or tex3d)
    const void* srv;      // ID3D11ShaderResourceView*
} sg_d3d11_image_info;

typedef struct sg_d3d11_sampler_info {
    const void* smp;      // ID3D11SamplerState*
} sg_d3d11_sampler_info;

typedef struct sg_d3d11_shader_info {
    const void* cbufs[SG_MAX_UNIFORMBLOCK_BINDSLOTS]; // ID3D11Buffer* (constant buffers by bind slot)
    const void* vs;   // ID3D11VertexShader*
    const void* fs;   // ID3D11PixelShader*
} sg_d3d11_shader_info;

typedef struct sg_d3d11_pipeline_info {
    const void* il;   // ID3D11InputLayout*
    const void* rs;   // ID3D11RasterizerState*
    const void* dss;  // ID3D11DepthStencilState*
    const void* bs;   // ID3D11BlendState*
} sg_d3d11_pipeline_info;

typedef struct sg_d3d11_attachments_info {
    const void* color_rtv[SG_MAX_COLOR_ATTACHMENTS];      // ID3D11RenderTargetView
    const void* resolve_rtv[SG_MAX_COLOR_ATTACHMENTS];    // ID3D11RenderTargetView
    const void* dsv;  // ID3D11DepthStencilView
} sg_d3d11_attachments_info;

typedef struct sg_mtl_buffer_info {
    const void* buf[SG_NUM_INFLIGHT_FRAMES];  // id<MTLBuffer>
    int active_slot;
} sg_mtl_buffer_info;

typedef struct sg_mtl_image_info {
    const void* tex[SG_NUM_INFLIGHT_FRAMES]; // id<MTLTexture>
    int active_slot;
} sg_mtl_image_info;

typedef struct sg_mtl_sampler_info {
    const void* smp;  // id<MTLSamplerState>
} sg_mtl_sampler_info;

typedef struct sg_mtl_shader_info {
    const void* vertex_lib;     // id<MTLLibrary>
    const void* fragment_lib;   // id<MTLLibrary>
    const void* vertex_func;    // id<MTLFunction>
    const void* fragment_func;  // id<MTLFunction>
} sg_mtl_shader_info;

typedef struct sg_mtl_pipeline_info {
    const void* rps;      // id<MTLRenderPipelineState>
    const void* dss;      // id<MTLDepthStencilState>
} sg_mtl_pipeline_info;

typedef struct sg_wgpu_buffer_info {
    const void* buf;  // WGPUBuffer
} sg_wgpu_buffer_info;

typedef struct sg_wgpu_image_info {
    const void* tex;  // WGPUTexture
    const void* view; // WGPUTextureView
} sg_wgpu_image_info;

typedef struct sg_wgpu_sampler_info {
    const void* smp;  // WGPUSampler
} sg_wgpu_sampler_info;

typedef struct sg_wgpu_shader_info {
    const void* vs_mod;   // WGPUShaderModule
    const void* fs_mod;   // WGPUShaderModule
    const void* bgl;      // WGPUBindGroupLayout;
} sg_wgpu_shader_info;

typedef struct sg_wgpu_pipeline_info {
    const void* render_pipeline;   // WGPURenderPipeline
    const void* compute_pipeline;  // WGPUComputePipeline
} sg_wgpu_pipeline_info;

typedef struct sg_wgpu_attachments_info {
    const void* color_view[SG_MAX_COLOR_ATTACHMENTS];     // WGPUTextureView
    const void* resolve_view[SG_MAX_COLOR_ATTACHMENTS];    // WGPUTextureView
    const void* ds_view;  // WGPUTextureView
} sg_wgpu_attachments_info;

typedef struct sg_gl_buffer_info {
    uint32_t buf[SG_NUM_INFLIGHT_FRAMES];
    int active_slot;
} sg_gl_buffer_info;

typedef struct sg_gl_image_info {
    uint32_t tex[SG_NUM_INFLIGHT_FRAMES];
    uint32_t tex_target;
    uint32_t msaa_render_buffer;
    int active_slot;
} sg_gl_image_info;

typedef struct sg_gl_sampler_info {
    uint32_t smp;
} sg_gl_sampler_info;

typedef struct sg_gl_shader_info {
    uint32_t prog;
} sg_gl_shader_info;

typedef struct sg_gl_attachments_info {
    uint32_t framebuffer;
    uint32_t msaa_resolve_framebuffer[SG_MAX_COLOR_ATTACHMENTS];
} sg_gl_attachments_info;

// D3D11: return ID3D11Device
SOKOL_GFX_API_DECL const void* sg_d3d11_device(void);
// D3D11: return ID3D11DeviceContext
SOKOL_GFX_API_DECL const void* sg_d3d11_device_context(void);
// D3D11: get internal buffer resource objects
SOKOL_GFX_API_DECL sg_d3d11_buffer_info sg_d3d11_query_buffer_info(sg_buffer buf);
// D3D11: get internal image resource objects
SOKOL_GFX_API_DECL sg_d3d11_image_info sg_d3d11_query_image_info(sg_image img);
// D3D11: get internal sampler resource objects
SOKOL_GFX_API_DECL sg_d3d11_sampler_info sg_d3d11_query_sampler_info(sg_sampler smp);
// D3D11: get internal shader resource objects
SOKOL_GFX_API_DECL sg_d3d11_shader_info sg_d3d11_query_shader_info(sg_shader shd);
// D3D11: get internal pipeline resource objects
SOKOL_GFX_API_DECL sg_d3d11_pipeline_info sg_d3d11_query_pipeline_info(sg_pipeline pip);
// D3D11: get internal pass resource objects
SOKOL_GFX_API_DECL sg_d3d11_attachments_info sg_d3d11_query_attachments_info(sg_attachments atts);

// Metal: return __bridge-casted MTLDevice
SOKOL_GFX_API_DECL const void* sg_mtl_device(void);
// Metal: return __bridge-casted MTLRenderCommandEncoder when inside render pass (otherwise zero)
SOKOL_GFX_API_DECL const void* sg_mtl_render_command_encoder(void);
// Metal: return __bridge-casted MTLComputeCommandEncoder when inside compute pass (otherwise zero)
SOKOL_GFX_API_DECL const void* sg_mtl_compute_command_encoder(void);
// Metal: get internal __bridge-casted buffer resource objects
SOKOL_GFX_API_DECL sg_mtl_buffer_info sg_mtl_query_buffer_info(sg_buffer buf);
// Metal: get internal __bridge-casted image resource objects
SOKOL_GFX_API_DECL sg_mtl_image_info sg_mtl_query_image_info(sg_image img);
// Metal: get internal __bridge-casted sampler resource objects
SOKOL_GFX_API_DECL sg_mtl_sampler_info sg_mtl_query_sampler_info(sg_sampler smp);
// Metal: get internal __bridge-casted shader resource objects
SOKOL_GFX_API_DECL sg_mtl_shader_info sg_mtl_query_shader_info(sg_shader shd);
// Metal: get internal __bridge-casted pipeline resource objects
SOKOL_GFX_API_DECL sg_mtl_pipeline_info sg_mtl_query_pipeline_info(sg_pipeline pip);

// WebGPU: return WGPUDevice object
SOKOL_GFX_API_DECL const void* sg_wgpu_device(void);
// WebGPU: return WGPUQueue object
SOKOL_GFX_API_DECL const void* sg_wgpu_queue(void);
// WebGPU: return this frame's WGPUCommandEncoder
SOKOL_GFX_API_DECL const void* sg_wgpu_command_encoder(void);
// WebGPU: return WGPURenderPassEncoder of current pass (returns 0 when outside pass or in a compute pass)
SOKOL_GFX_API_DECL const void* sg_wgpu_render_pass_encoder(void);
// WebGPU: return WGPUComputePassEncoder of current pass (returns 0 when outside pass or in a render pass)
SOKOL_GFX_API_DECL const void* sg_wgpu_compute_pass_encoder(void);
// WebGPU: get internal buffer resource objects
SOKOL_GFX_API_DECL sg_wgpu_buffer_info sg_wgpu_query_buffer_info(sg_buffer buf);
// WebGPU: get internal image resource objects
SOKOL_GFX_API_DECL sg_wgpu_image_info sg_wgpu_query_image_info(sg_image img);
// WebGPU: get internal sampler resource objects
SOKOL_GFX_API_DECL sg_wgpu_sampler_info sg_wgpu_query_sampler_info(sg_sampler smp);
// WebGPU: get internal shader resource objects
SOKOL_GFX_API_DECL sg_wgpu_shader_info sg_wgpu_query_shader_info(sg_shader shd);
// WebGPU: get internal pipeline resource objects
SOKOL_GFX_API_DECL sg_wgpu_pipeline_info sg_wgpu_query_pipeline_info(sg_pipeline pip);
// WebGPU: get internal pass resource objects
SOKOL_GFX_API_DECL sg_wgpu_attachments_info sg_wgpu_query_attachments_info(sg_attachments atts);

// GL: get internal buffer resource objects
SOKOL_GFX_API_DECL sg_gl_buffer_info sg_gl_query_buffer_info(sg_buffer buf);
// GL: get internal image resource objects
SOKOL_GFX_API_DECL sg_gl_image_info sg_gl_query_image_info(sg_image img);
// GL: get internal sampler resource objects
SOKOL_GFX_API_DECL sg_gl_sampler_info sg_gl_query_sampler_info(sg_sampler smp);
// GL: get internal shader resource objects
SOKOL_GFX_API_DECL sg_gl_shader_info sg_gl_query_shader_info(sg_shader shd);
// GL: get internal pass resource objects
SOKOL_GFX_API_DECL sg_gl_attachments_info sg_gl_query_attachments_info(sg_attachments atts);

#ifdef __cplusplus
} // extern "C"

// reference-based equivalents for c++
inline void sg_setup(const sg_desc& desc) { return sg_setup(&desc); }

inline sg_buffer sg_make_buffer(const sg_buffer_desc& desc) { return sg_make_buffer(&desc); }
inline sg_image sg_make_image(const sg_image_desc& desc) { return sg_make_image(&desc); }
inline sg_sampler sg_make_sampler(const sg_sampler_desc& desc) { return sg_make_sampler(&desc); }
inline sg_shader sg_make_shader(const sg_shader_desc& desc) { return sg_make_shader(&desc); }
inline sg_pipeline sg_make_pipeline(const sg_pipeline_desc& desc) { return sg_make_pipeline(&desc); }
inline sg_attachments sg_make_attachments(const sg_attachments_desc& desc) { return sg_make_attachments(&desc); }
inline void sg_update_image(sg_image img, const sg_image_data& data) { return sg_update_image(img, &data); }

inline void sg_begin_pass(const sg_pass& pass) { return sg_begin_pass(&pass); }
inline void sg_apply_bindings(const sg_bindings& bindings) { return sg_apply_bindings(&bindings); }
inline void sg_apply_uniforms(int ub_slot, const sg_range& data) { return sg_apply_uniforms(ub_slot, &data); }

inline sg_buffer_desc sg_query_buffer_defaults(const sg_buffer_desc& desc) { return sg_query_buffer_defaults(&desc); }
inline sg_image_desc sg_query_image_defaults(const sg_image_desc& desc) { return sg_query_image_defaults(&desc); }
inline sg_sampler_desc sg_query_sampler_defaults(const sg_sampler_desc& desc) { return sg_query_sampler_defaults(&desc); }
inline sg_shader_desc sg_query_shader_defaults(const sg_shader_desc& desc) { return sg_query_shader_defaults(&desc); }
inline sg_pipeline_desc sg_query_pipeline_defaults(const sg_pipeline_desc& desc) { return sg_query_pipeline_defaults(&desc); }
inline sg_attachments_desc sg_query_attachments_defaults(const sg_attachments_desc& desc) { return sg_query_attachments_defaults(&desc); }

inline void sg_init_buffer(sg_buffer buf, const sg_buffer_desc& desc) { return sg_init_buffer(buf, &desc); }
inline void sg_init_image(sg_image img, const sg_image_desc& desc) { return sg_init_image(img, &desc); }
inline void sg_init_sampler(sg_sampler smp, const sg_sampler_desc& desc) { return sg_init_sampler(smp, &desc); }
inline void sg_init_shader(sg_shader shd, const sg_shader_desc& desc) { return sg_init_shader(shd, &desc); }
inline void sg_init_pipeline(sg_pipeline pip, const sg_pipeline_desc& desc) { return sg_init_pipeline(pip, &desc); }
inline void sg_init_attachments(sg_attachments atts, const sg_attachments_desc& desc) { return sg_init_attachments(atts, &desc); }

inline void sg_update_buffer(sg_buffer buf_id, const sg_range& data) { return sg_update_buffer(buf_id, &data); }
inline int sg_append_buffer(sg_buffer buf_id, const sg_range& data) { return sg_append_buffer(buf_id, &data); }
#endif
#endif // SOKOL_GFX_INCLUDED

// ██ ███    ███ ██████  ██      ███████ ███    ███ ███████ ███    ██ ████████  █████  ████████ ██  ██████  ███    ██
// ██ ████  ████ ██   ██ ██      ██      ████  ████ ██      ████   ██    ██    ██   ██    ██    ██ ██    ██ ████   ██
// ██ ██ ████ ██ ██████  ██      █████   ██ ████ ██ █████   ██ ██  ██    ██    ███████    ██    ██ ██    ██ ██ ██  ██
// ██ ██  ██  ██ ██      ██      ██      ██  ██  ██ ██      ██  ██ ██    ██    ██   ██    ██    ██ ██    ██ ██  ██ ██
// ██ ██      ██ ██      ███████ ███████ ██      ██ ███████ ██   ████    ██    ██   ██    ██    ██  ██████  ██   ████
//
// >>implementation
#ifdef SOKOL_GFX_IMPL
#define SOKOL_GFX_IMPL_INCLUDED (1)

#if !(defined(SOKOL_GLCORE)||defined(SOKOL_GLES3)||defined(SOKOL_D3D11)||defined(SOKOL_METAL)||defined(SOKOL_WGPU)||defined(SOKOL_DUMMY_BACKEND))
#error "Please select a backend with SOKOL_GLCORE, SOKOL_GLES3, SOKOL_D3D11, SOKOL_METAL, SOKOL_WGPU or SOKOL_DUMMY_BACKEND"
#endif
#if defined(SOKOL_MALLOC) || defined(SOKOL_CALLOC) || defined(SOKOL_FREE)
#error "SOKOL_MALLOC/CALLOC/FREE macros are no longer supported, please use sg_desc.allocator to override memory allocation functions"
#endif

#include <stdlib.h> // malloc, free, qsort
#include <string.h> // memset
#include <float.h> // FLT_MAX

#ifndef SOKOL_API_IMPL
    #define SOKOL_API_IMPL
#endif
#ifndef SOKOL_DEBUG
    #ifndef NDEBUG
        #define SOKOL_DEBUG
    #endif
#endif
#ifndef SOKOL_ASSERT
    #include <assert.h>
    #define SOKOL_ASSERT(c) assert(c)
#endif
#ifndef SOKOL_UNREACHABLE
    #define SOKOL_UNREACHABLE SOKOL_ASSERT(false)
#endif

#ifndef _SOKOL_PRIVATE
    #if defined(__GNUC__) || defined(__clang__)
        #define _SOKOL_PRIVATE __attribute__((unused)) static
    #else
        #define _SOKOL_PRIVATE static
    #endif
#endif

#ifndef _SOKOL_UNUSED
    #define _SOKOL_UNUSED(x) (void)(x)
#endif

#if defined(SOKOL_TRACE_HOOKS)
#define _SG_TRACE_ARGS(fn, ...) if (_sg.hooks.fn) { _sg.hooks.fn(__VA_ARGS__, _sg.hooks.user_data); }
#define _SG_TRACE_NOARGS(fn) if (_sg.hooks.fn) { _sg.hooks.fn(_sg.hooks.user_data); }
#else
#define _SG_TRACE_ARGS(fn, ...)
#define _SG_TRACE_NOARGS(fn)
#endif

// default clear values
#ifndef SG_DEFAULT_CLEAR_RED
#define SG_DEFAULT_CLEAR_RED (0.5f)
#endif
#ifndef SG_DEFAULT_CLEAR_GREEN
#define SG_DEFAULT_CLEAR_GREEN (0.5f)
#endif
#ifndef SG_DEFAULT_CLEAR_BLUE
#define SG_DEFAULT_CLEAR_BLUE (0.5f)
#endif
#ifndef SG_DEFAULT_CLEAR_ALPHA
#define SG_DEFAULT_CLEAR_ALPHA (1.0f)
#endif
#ifndef SG_DEFAULT_CLEAR_DEPTH
#define SG_DEFAULT_CLEAR_DEPTH (1.0f)
#endif
#ifndef SG_DEFAULT_CLEAR_STENCIL
#define SG_DEFAULT_CLEAR_STENCIL (0)
#endif

#ifdef _MSC_VER
#pragma warning(push)
#pragma warning(disable:4115)   // named type definition in parentheses
#pragma warning(disable:4505)   // unreferenced local function has been removed
#pragma warning(disable:4201)   // nonstandard extension used: nameless struct/union (needed by d3d11.h)
#pragma warning(disable:4054)   // 'type cast': from function pointer
#pragma warning(disable:4055)   // 'type cast': from data pointer
#endif

#if defined(SOKOL_D3D11)
    #ifndef D3D11_NO_HELPERS
    #define D3D11_NO_HELPERS
    #endif
    #ifndef WIN32_LEAN_AND_MEAN
    #define WIN32_LEAN_AND_MEAN
    #endif
    #ifndef NOMINMAX
    #define NOMINMAX
    #endif
    #include <d3d11.h>
    #include <d3dcompiler.h>
    #ifdef _MSC_VER
    #pragma comment (lib, "kernel32")
    #pragma comment (lib, "user32")
    #pragma comment (lib, "dxgi")
    #pragma comment (lib, "d3d11")
    #endif
#elif defined(SOKOL_METAL)
    // see https://clang.llvm.org/docs/LanguageExtensions.html#automatic-reference-counting
    #if !defined(__cplusplus)
        #if __has_feature(objc_arc) && !__has_feature(objc_arc_fields)
            #error "sokol_gfx.h requires __has_feature(objc_arc_field) if ARC is enabled (use a more recent compiler version)"
        #endif
    #endif
    #include <TargetConditionals.h>
    #include <AvailabilityMacros.h>
    #if defined(TARGET_OS_IPHONE) && !TARGET_OS_IPHONE
        #define _SG_TARGET_MACOS (1)
    #else
        #define _SG_TARGET_IOS (1)
        #if defined(TARGET_IPHONE_SIMULATOR) && TARGET_IPHONE_SIMULATOR
            #define _SG_TARGET_IOS_SIMULATOR (1)
        #endif
    #endif
    #import <Metal/Metal.h>
    #import <QuartzCore/CoreAnimation.h> // needed for CAMetalDrawable
#elif defined(SOKOL_WGPU)
    #include <webgpu/webgpu.h>
    #if defined(__EMSCRIPTEN__)
        #include <emscripten/emscripten.h>
    #endif
#elif defined(SOKOL_GLCORE) || defined(SOKOL_GLES3)
    #define _SOKOL_ANY_GL (1)

    // include platform specific GL headers (or on Win32: use an embedded GL loader)
    #if !defined(SOKOL_EXTERNAL_GL_LOADER)
        #if defined(_WIN32)
            #if defined(SOKOL_GLCORE)
                #define _SOKOL_USE_WIN32_GL_LOADER (1)
                #ifndef WIN32_LEAN_AND_MEAN
                #define WIN32_LEAN_AND_MEAN
                #endif
                #ifndef NOMINMAX
                #define NOMINMAX
                #endif
                #include <windows.h>
                #pragma comment (lib, "kernel32")   // GetProcAddress()
                #define _SOKOL_GL_HAS_COMPUTE (1)
                #define _SOKOL_GL_HAS_TEXSTORAGE (1)
            #endif
        #elif defined(__APPLE__)
            #include <TargetConditionals.h>
            #ifndef GL_SILENCE_DEPRECATION
                #define GL_SILENCE_DEPRECATION
            #endif
            #if defined(TARGET_OS_IPHONE) && !TARGET_OS_IPHONE
                #include <OpenGL/gl3.h>
            #else
                #include <OpenGLES/ES3/gl.h>
                #include <OpenGLES/ES3/glext.h>
                #define _SOKOL_GL_HAS_TEXSTORAGE (1)
            #endif
        #elif defined(__EMSCRIPTEN__)
            #if defined(SOKOL_GLES3)
                #include <GLES3/gl3.h>
                #define _SOKOL_GL_HAS_TEXSTORAGE (1)
            #endif
        #elif defined(__ANDROID__)
            #include <GLES3/gl31.h>
            #define _SOKOL_GL_HAS_COMPUTE (1)
            #define _SOKOL_GL_HAS_TEXSTORAGE (1)
        #elif defined(__linux__) || defined(__unix__)
            #if defined(SOKOL_GLCORE)
                #define GL_GLEXT_PROTOTYPES
                #include <GL/gl.h>
            #else
                #include <GLES3/gl31.h>
                #include <GLES3/gl3ext.h>
            #endif
            #define _SOKOL_GL_HAS_COMPUTE (1)
            #define _SOKOL_GL_HAS_TEXSTORAGE (1)
        #endif
    #endif

    // optional GL loader definitions (only on Win32)
    #if defined(_SOKOL_USE_WIN32_GL_LOADER)
        #define __gl_h_ 1
        #define __gl32_h_ 1
        #define __gl31_h_ 1
        #define __GL_H__ 1
        #define __glext_h_ 1
        #define __GLEXT_H_ 1
        #define __gltypes_h_ 1
        #define __glcorearb_h_ 1
        #define __gl_glcorearb_h_ 1
        #define GL_APIENTRY APIENTRY

        typedef unsigned int  GLenum;
        typedef unsigned int  GLuint;
        typedef int  GLsizei;
        typedef char  GLchar;
        typedef ptrdiff_t  GLintptr;
        typedef ptrdiff_t  GLsizeiptr;
        typedef double  GLclampd;
        typedef unsigned short  GLushort;
        typedef unsigned char  GLubyte;
        typedef unsigned char  GLboolean;
        typedef uint64_t  GLuint64;
        typedef double  GLdouble;
        typedef unsigned short  GLhalf;
        typedef float  GLclampf;
        typedef unsigned int  GLbitfield;
        typedef signed char  GLbyte;
        typedef short  GLshort;
        typedef void  GLvoid;
        typedef int64_t  GLint64;
        typedef float  GLfloat;
        typedef int  GLint;
        #define GL_INT_2_10_10_10_REV 0x8D9F
        #define GL_R32F 0x822E
        #define GL_PROGRAM_POINT_SIZE 0x8642
        #define GL_DEPTH_ATTACHMENT 0x8D00
        #define GL_DEPTH_STENCIL_ATTACHMENT 0x821A
        #define GL_COLOR_ATTACHMENT2 0x8CE2
        #define GL_COLOR_ATTACHMENT0 0x8CE0
        #define GL_R16F 0x822D
        #define GL_COLOR_ATTACHMENT22 0x8CF6
        #define GL_DRAW_FRAMEBUFFER 0x8CA9
        #define GL_FRAMEBUFFER_COMPLETE 0x8CD5
        #define GL_NUM_EXTENSIONS 0x821D
        #define GL_INFO_LOG_LENGTH 0x8B84
        #define GL_VERTEX_SHADER 0x8B31
        #define GL_INCR 0x1E02
        #define GL_DYNAMIC_DRAW 0x88E8
        #define GL_STATIC_DRAW 0x88E4
        #define GL_TEXTURE_CUBE_MAP_POSITIVE_Z 0x8519
        #define GL_TEXTURE_CUBE_MAP 0x8513
        #define GL_FUNC_SUBTRACT 0x800A
        #define GL_FUNC_REVERSE_SUBTRACT 0x800B
        #define GL_CONSTANT_COLOR 0x8001
        #define GL_DECR_WRAP 0x8508
        #define GL_R8 0x8229
        #define GL_LINEAR_MIPMAP_LINEAR 0x2703
        #define GL_ELEMENT_ARRAY_BUFFER 0x8893
        #define GL_SHORT 0x1402
        #define GL_DEPTH_TEST 0x0B71
        #define GL_TEXTURE_CUBE_MAP_NEGATIVE_Y 0x8518
        #define GL_LINK_STATUS 0x8B82
        #define GL_TEXTURE_CUBE_MAP_POSITIVE_Y 0x8517
        #define GL_SAMPLE_ALPHA_TO_COVERAGE 0x809E
        #define GL_RGBA16F 0x881A
        #define GL_CONSTANT_ALPHA 0x8003
        #define GL_READ_FRAMEBUFFER 0x8CA8
        #define GL_TEXTURE0 0x84C0
        #define GL_TEXTURE_MIN_LOD 0x813A
        #define GL_CLAMP_TO_EDGE 0x812F
        #define GL_UNSIGNED_SHORT_5_6_5 0x8363
        #define GL_TEXTURE_WRAP_R 0x8072
        #define GL_UNSIGNED_SHORT_5_5_5_1 0x8034
        #define GL_NEAREST_MIPMAP_NEAREST 0x2700
        #define GL_UNSIGNED_SHORT_4_4_4_4 0x8033
        #define GL_SRC_ALPHA_SATURATE 0x0308
        #define GL_STREAM_DRAW 0x88E0
        #define GL_ONE 1
        #define GL_NEAREST_MIPMAP_LINEAR 0x2702
        #define GL_RGB10_A2 0x8059
        #define GL_RGBA8 0x8058
        #define GL_SRGB8_ALPHA8 0x8C43
        #define GL_COLOR_ATTACHMENT1 0x8CE1
        #define GL_RGBA4 0x8056
        #define GL_RGB8 0x8051
        #define GL_ARRAY_BUFFER 0x8892
        #define GL_STENCIL 0x1802
        #define GL_TEXTURE_2D 0x0DE1
        #define GL_DEPTH 0x1801
        #define GL_FRONT 0x0404
        #define GL_STENCIL_BUFFER_BIT 0x00000400
        #define GL_REPEAT 0x2901
        #define GL_RGBA 0x1908
        #define GL_TEXTURE_CUBE_MAP_POSITIVE_X 0x8515
        #define GL_DECR 0x1E03
        #define GL_FRAGMENT_SHADER 0x8B30
        #define GL_COMPUTE_SHADER 0x91B9
        #define GL_FLOAT 0x1406
        #define GL_TEXTURE_MAX_LOD 0x813B
        #define GL_DEPTH_COMPONENT 0x1902
        #define GL_ONE_MINUS_DST_ALPHA 0x0305
        #define GL_COLOR 0x1800
        #define GL_TEXTURE_2D_ARRAY 0x8C1A
        #define GL_TRIANGLES 0x0004
        #define GL_UNSIGNED_BYTE 0x1401
        #define GL_TEXTURE_MAG_FILTER 0x2800
        #define GL_ONE_MINUS_CONSTANT_ALPHA 0x8004
        #define GL_NONE 0
        #define GL_SRC_COLOR 0x0300
        #define GL_BYTE 0x1400
        #define GL_TEXTURE_CUBE_MAP_NEGATIVE_Z 0x851A
        #define GL_LINE_STRIP 0x0003
        #define GL_TEXTURE_3D 0x806F
        #define GL_CW 0x0900
        #define GL_LINEAR 0x2601
        #define GL_RENDERBUFFER 0x8D41
        #define GL_GEQUAL 0x0206
        #define GL_COLOR_BUFFER_BIT 0x00004000
        #define GL_RGBA32F 0x8814
        #define GL_BLEND 0x0BE2
        #define GL_ONE_MINUS_SRC_ALPHA 0x0303
        #define GL_ONE_MINUS_CONSTANT_COLOR 0x8002
        #define GL_TEXTURE_WRAP_T 0x2803
        #define GL_TEXTURE_WRAP_S 0x2802
        #define GL_TEXTURE_MIN_FILTER 0x2801
        #define GL_LINEAR_MIPMAP_NEAREST 0x2701
        #define GL_EXTENSIONS 0x1F03
        #define GL_NO_ERROR 0
        #define GL_REPLACE 0x1E01
        #define GL_KEEP 0x1E00
        #define GL_CCW 0x0901
        #define GL_TEXTURE_CUBE_MAP_NEGATIVE_X 0x8516
        #define GL_RGB 0x1907
        #define GL_TRIANGLE_STRIP 0x0005
        #define GL_FALSE 0
        #define GL_ZERO 0
        #define GL_CULL_FACE 0x0B44
        #define GL_INVERT 0x150A
        #define GL_INT 0x1404
        #define GL_UNSIGNED_INT 0x1405
        #define GL_UNSIGNED_SHORT 0x1403
        #define GL_NEAREST 0x2600
        #define GL_SCISSOR_TEST 0x0C11
        #define GL_LEQUAL 0x0203
        #define GL_STENCIL_TEST 0x0B90
        #define GL_DITHER 0x0BD0
        #define GL_DEPTH_COMPONENT32F 0x8CAC
        #define GL_EQUAL 0x0202
        #define GL_FRAMEBUFFER 0x8D40
        #define GL_RGB5 0x8050
        #define GL_LINES 0x0001
        #define GL_DEPTH_BUFFER_BIT 0x00000100
        #define GL_SRC_ALPHA 0x0302
        #define GL_INCR_WRAP 0x8507
        #define GL_LESS 0x0201
        #define GL_MULTISAMPLE 0x809D
        #define GL_FRAMEBUFFER_BINDING 0x8CA6
        #define GL_BACK 0x0405
        #define GL_ALWAYS 0x0207
        #define GL_FUNC_ADD 0x8006
        #define GL_ONE_MINUS_DST_COLOR 0x0307
        #define GL_NOTEQUAL 0x0205
        #define GL_DST_COLOR 0x0306
        #define GL_COMPILE_STATUS 0x8B81
        #define GL_RED 0x1903
        #define GL_COLOR_ATTACHMENT3 0x8CE3
        #define GL_DST_ALPHA 0x0304
        #define GL_RGB5_A1 0x8057
        #define GL_GREATER 0x0204
        #define GL_POLYGON_OFFSET_FILL 0x8037
        #define GL_TRUE 1
        #define GL_NEVER 0x0200
        #define GL_POINTS 0x0000
        #define GL_ONE_MINUS_SRC_COLOR 0x0301
        #define GL_MIRRORED_REPEAT 0x8370
        #define GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS 0x8B4D
        #define GL_R11F_G11F_B10F 0x8C3A
        #define GL_UNSIGNED_INT_10F_11F_11F_REV 0x8C3B
        #define GL_RGB9_E5 0x8C3D
        #define GL_UNSIGNED_INT_5_9_9_9_REV 0x8C3E
        #define GL_RGBA32UI 0x8D70
        #define GL_RGB32UI 0x8D71
        #define GL_RGBA16UI 0x8D76
        #define GL_RGB16UI 0x8D77
        #define GL_RGBA8UI 0x8D7C
        #define GL_RGB8UI 0x8D7D
        #define GL_RGBA32I 0x8D82
        #define GL_RGB32I 0x8D83
        #define GL_RGBA16I 0x8D88
        #define GL_RGB16I 0x8D89
        #define GL_RGBA8I 0x8D8E
        #define GL_RGB8I 0x8D8F
        #define GL_RED_INTEGER 0x8D94
        #define GL_RG 0x8227
        #define GL_RG_INTEGER 0x8228
        #define GL_R8 0x8229
        #define GL_R16 0x822A
        #define GL_RG8 0x822B
        #define GL_RG16 0x822C
        #define GL_R16F 0x822D
        #define GL_R32F 0x822E
        #define GL_RG16F 0x822F
        #define GL_RG32F 0x8230
        #define GL_R8I 0x8231
        #define GL_R8UI 0x8232
        #define GL_R16I 0x8233
        #define GL_R16UI 0x8234
        #define GL_R32I 0x8235
        #define GL_R32UI 0x8236
        #define GL_RG8I 0x8237
        #define GL_RG8UI 0x8238
        #define GL_RG16I 0x8239
        #define GL_RG16UI 0x823A
        #define GL_RG32I 0x823B
        #define GL_RG32UI 0x823C
        #define GL_RGBA_INTEGER 0x8D99
        #define GL_R8_SNORM 0x8F94
        #define GL_RG8_SNORM 0x8F95
        #define GL_RGB8_SNORM 0x8F96
        #define GL_RGBA8_SNORM 0x8F97
        #define GL_R16_SNORM 0x8F98
        #define GL_RG16_SNORM 0x8F99
        #define GL_RGB16_SNORM 0x8F9A
        #define GL_RGBA16_SNORM 0x8F9B
        #define GL_RGBA16 0x805B
        #define GL_MAX_TEXTURE_SIZE 0x0D33
        #define GL_MAX_CUBE_MAP_TEXTURE_SIZE 0x851C
        #define GL_MAX_3D_TEXTURE_SIZE 0x8073
        #define GL_MAX_ARRAY_TEXTURE_LAYERS 0x88FF
        #define GL_MAX_VERTEX_ATTRIBS 0x8869
        #define GL_CLAMP_TO_BORDER 0x812D
        #define GL_TEXTURE_BORDER_COLOR 0x1004
        #define GL_CURRENT_PROGRAM 0x8B8D
        #define GL_MAX_VERTEX_UNIFORM_COMPONENTS  0x8B4A
        #define GL_UNPACK_ALIGNMENT 0x0CF5
        #define GL_FRAMEBUFFER_SRGB 0x8DB9
        #define GL_TEXTURE_COMPARE_MODE 0x884C
        #define GL_TEXTURE_COMPARE_FUNC 0x884D
        #define GL_COMPARE_REF_TO_TEXTURE 0x884E
        #define GL_TEXTURE_CUBE_MAP_SEAMLESS 0x884F
        #define GL_TEXTURE_MAX_LEVEL 0x813D
        #define GL_FRAMEBUFFER_UNDEFINED 0x8219
        #define GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT 0x8CD6
        #define GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT 0x8CD7
        #define GL_FRAMEBUFFER_UNSUPPORTED 0x8CDD
        #define GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE 0x8D56
        #define GL_MAJOR_VERSION 0x821B
        #define GL_MINOR_VERSION 0x821C
        #define GL_TEXTURE_2D_MULTISAMPLE 0x9100
        #define GL_TEXTURE_2D_MULTISAMPLE_ARRAY 0x9102
        #define GL_SHADER_STORAGE_BARRIER_BIT 0x2000
        #define GL_VERTEX_ATTRIB_ARRAY_BARRIER_BIT 0x00000001
        #define GL_ELEMENT_ARRAY_BARRIER_BIT 0x00000002
        #define GL_TEXTURE_FETCH_BARRIER_BIT 0x00000008
        #define GL_SHADER_IMAGE_ACCESS_BARRIER_BIT 0x00000020
        #define GL_MIN 0x8007
        #define GL_MAX 0x8008
        #define GL_WRITE_ONLY 0x88B9
        #define GL_READ_WRITE 0x88BA
    #endif

    #ifndef GL_UNSIGNED_INT_2_10_10_10_REV
    #define GL_UNSIGNED_INT_2_10_10_10_REV 0x8368
    #endif
    #ifndef GL_UNSIGNED_INT_24_8
    #define GL_UNSIGNED_INT_24_8 0x84FA
    #endif
    #ifndef GL_TEXTURE_MAX_ANISOTROPY_EXT
    #define GL_TEXTURE_MAX_ANISOTROPY_EXT 0x84FE
    #endif
    #ifndef GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT
    #define GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT 0x84FF
    #endif
    #ifndef GL_COMPRESSED_RGBA_S3TC_DXT1_EXT
    #define GL_COMPRESSED_RGBA_S3TC_DXT1_EXT 0x83F1
    #endif
    #ifndef GL_COMPRESSED_RGBA_S3TC_DXT3_EXT
    #define GL_COMPRESSED_RGBA_S3TC_DXT3_EXT 0x83F2
    #endif
    #ifndef GL_COMPRESSED_RGBA_S3TC_DXT5_EXT
    #define GL_COMPRESSED_RGBA_S3TC_DXT5_EXT 0x83F3
    #endif
    #ifndef GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT
    #define GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT 0x8C4F
    #endif
    #ifndef GL_COMPRESSED_RED_RGTC1
    #define GL_COMPRESSED_RED_RGTC1 0x8DBB
    #endif
    #ifndef GL_COMPRESSED_SIGNED_RED_RGTC1
    #define GL_COMPRESSED_SIGNED_RED_RGTC1 0x8DBC
    #endif
    #ifndef GL_COMPRESSED_RED_GREEN_RGTC2
    #define GL_COMPRESSED_RED_GREEN_RGTC2 0x8DBD
    #endif
    #ifndef GL_COMPRESSED_SIGNED_RED_GREEN_RGTC2
    #define GL_COMPRESSED_SIGNED_RED_GREEN_RGTC2 0x8DBE
    #endif
    #ifndef GL_COMPRESSED_RGBA_BPTC_UNORM_ARB
    #define GL_COMPRESSED_RGBA_BPTC_UNORM_ARB 0x8E8C
    #endif
    #ifndef GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB
    #define GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB 0x8E8D
    #endif
    #ifndef GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB
    #define GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB 0x8E8E
    #endif
    #ifndef GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB
    #define GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB 0x8E8F
    #endif
    #ifndef GL_COMPRESSED_RGB8_ETC2
    #define GL_COMPRESSED_RGB8_ETC2 0x9274
    #endif
    #ifndef GL_COMPRESSED_SRGB8_ETC2
    #define GL_COMPRESSED_SRGB8_ETC2 0x9275
    #endif
    #ifndef GL_COMPRESSED_RGBA8_ETC2_EAC
    #define GL_COMPRESSED_RGBA8_ETC2_EAC 0x9278
    #endif
    #ifndef GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC
    #define GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC 0x9279
    #endif
    #ifndef GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2
    #define GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2 0x9276
    #endif
    #ifndef GL_COMPRESSED_R11_EAC
    #define GL_COMPRESSED_R11_EAC 0x9270
    #endif
    #ifndef GL_COMPRESSED_SIGNED_R11_EAC
    #define GL_COMPRESSED_SIGNED_R11_EAC 0x9271
    #endif
    #ifndef GL_COMPRESSED_RG11_EAC
    #define GL_COMPRESSED_RG11_EAC 0x9272
    #endif
    #ifndef GL_COMPRESSED_SIGNED_RG11_EAC
    #define GL_COMPRESSED_SIGNED_RG11_EAC 0x9273
    #endif
    #ifndef GL_COMPRESSED_RGBA_ASTC_4x4_KHR
    #define GL_COMPRESSED_RGBA_ASTC_4x4_KHR 0x93B0
    #endif
    #ifndef GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR
    #define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR 0x93D0
    #endif
    #ifndef GL_DEPTH24_STENCIL8
    #define GL_DEPTH24_STENCIL8 0x88F0
    #endif
    #ifndef GL_HALF_FLOAT
    #define GL_HALF_FLOAT 0x140B
    #endif
    #ifndef GL_DEPTH_STENCIL
    #define GL_DEPTH_STENCIL 0x84F9
    #endif
    #ifndef GL_LUMINANCE
    #define GL_LUMINANCE 0x1909
    #endif
    #ifndef GL_COMPUTE_SHADER
    #define GL_COMPUTE_SHADER 0x91B9
    #endif
    #ifndef _SG_GL_CHECK_ERROR
    #define _SG_GL_CHECK_ERROR() { SOKOL_ASSERT(glGetError() == GL_NO_ERROR); }
    #endif
    // make some GL constants generally available to simplify compilation,
    // use of those constants will be filtered by runtime flags
    #ifndef GL_SHADER_STORAGE_BUFFER
    #define GL_SHADER_STORAGE_BUFFER 0x90D2
    #endif
#endif

#if defined(SOKOL_GLES3)
    // on WebGL2, GL_FRAMEBUFFER_UNDEFINED technically doesn't exist (it is defined
    // in the Emscripten headers, but may not exist in other WebGL2 shims)
    // see: https://github.com/floooh/sokol/pull/933
    #ifndef GL_FRAMEBUFFER_UNDEFINED
    #define GL_FRAMEBUFFER_UNDEFINED 0x8219
    #endif
#endif

// ███████ ████████ ██████  ██    ██  ██████ ████████ ███████
// ██         ██    ██   ██ ██    ██ ██         ██    ██
// ███████    ██    ██████  ██    ██ ██         ██    ███████
//      ██    ██    ██   ██ ██    ██ ██         ██         ██
// ███████    ██    ██   ██  ██████   ██████    ██    ███████
//
// >>structs
// resource pool slots
typedef struct {
    uint32_t id;
    uint32_t uninit_count;
    sg_resource_state state;
} _sg_slot_t;

// resource pool housekeeping struct
typedef struct {
    int size;
    int queue_top;
    uint32_t* gen_ctrs;
    int* free_queue;
} _sg_pool_t;

// resource func forward decls
struct _sg_buffer_s;
struct _sg_image_s;
struct _sg_sampler_s;
struct _sg_shader_s;
struct _sg_pipeline_s;
struct _sg_attachments_s;

// a general resource slot reference useful for caches
typedef struct _sg_sref_s {
    uint32_t id;
    uint32_t uninit_count;
} _sg_sref_t;

// safe (in debug mode) internal resource references
typedef struct _sg_buffer_ref_s {
    struct _sg_buffer_s* ptr;
    _sg_sref_t sref;
} _sg_buffer_ref_t;

typedef struct _sg_image_ref_s {
    struct _sg_image_s* ptr;
    _sg_sref_t sref;
} _sg_image_ref_t;

typedef struct _sg_sampler_ref_t {
    struct _sg_sampler_s* ptr;
    _sg_sref_t sref;
} _sg_sampler_ref_t;

typedef struct _sg_shader_ref_s {
    struct _sg_shader_s* ptr;
    _sg_sref_t sref;
} _sg_shader_ref_t;

typedef struct _sg_pipeline_ref_s {
    struct _sg_pipeline_s* ptr;
    _sg_sref_t sref;
} _sg_pipeline_ref_t;

typedef struct _sg_attachments_ref_s {
    struct _sg_attachments_s* ptr;
    _sg_sref_t sref;

} _sg_attachments_ref_t;

// resource tracking (for keeping track of gpu-written storage resources
typedef struct {
    uint32_t size;
    uint32_t cur;
    uint32_t* items;
} _sg_tracker_t;

// constants
enum {
    _SG_STRING_SIZE = 32,
    _SG_SLOT_SHIFT = 16,
    _SG_SLOT_MASK = (1<<_SG_SLOT_SHIFT)-1,
    _SG_MAX_POOL_SIZE = (1<<_SG_SLOT_SHIFT),
    _SG_DEFAULT_BUFFER_POOL_SIZE = 128,
    _SG_DEFAULT_IMAGE_POOL_SIZE = 128,
    _SG_DEFAULT_SAMPLER_POOL_SIZE = 64,
    _SG_DEFAULT_SHADER_POOL_SIZE = 32,
    _SG_DEFAULT_PIPELINE_POOL_SIZE = 64,
    _SG_DEFAULT_ATTACHMENTS_POOL_SIZE = 16,
    _SG_DEFAULT_UB_SIZE = 4 * 1024 * 1024,
    _SG_DEFAULT_MAX_DISPATCH_CALLS_PER_PASS = 1024,
    _SG_DEFAULT_MAX_COMMIT_LISTENERS = 1024,
    _SG_DEFAULT_WGPU_BINDGROUP_CACHE_SIZE = 1024,
};

// fixed-size string
typedef struct {
    char buf[_SG_STRING_SIZE];
} _sg_str_t;

typedef struct {
    int size;
    int append_pos;
    bool append_overflow;
    uint32_t update_frame_index;
    uint32_t append_frame_index;
    int num_slots;
    int active_slot;
    sg_buffer_usage usage;
} _sg_buffer_common_t;

typedef struct {
    uint32_t upd_frame_index;
    int num_slots;
    int active_slot;
    sg_image_type type;
    int width;
    int height;
    int num_slices;
    int num_mipmaps;
    sg_image_usage usage;
    sg_pixel_format pixel_format;
    int sample_count;
} _sg_image_common_t;

typedef struct {
    sg_filter min_filter;
    sg_filter mag_filter;
    sg_filter mipmap_filter;
    sg_wrap wrap_u;
    sg_wrap wrap_v;
    sg_wrap wrap_w;
    float min_lod;
    float max_lod;
    sg_border_color border_color;
    sg_compare_func compare;
    uint32_t max_anisotropy;
} _sg_sampler_common_t;

typedef struct {
    sg_shader_attr_base_type base_type;
} _sg_shader_attr_t;

typedef struct {
    sg_shader_stage stage;
    uint32_t size;
} _sg_shader_uniform_block_t;

typedef struct {
    sg_shader_stage stage;
    bool readonly;
} _sg_shader_storage_buffer_t;

typedef struct {
    sg_shader_stage stage;
    sg_image_type image_type;
    sg_pixel_format access_format;
    bool writeonly;
} _sg_shader_storage_image_t;

typedef struct {
    sg_shader_stage stage;
    sg_image_type image_type;
    sg_image_sample_type sample_type;
    bool multisampled;
} _sg_shader_image_t;

typedef struct {
    sg_shader_stage stage;
    sg_sampler_type sampler_type;
} _sg_shader_sampler_t;

typedef struct {
    sg_shader_stage stage;
    uint8_t image_slot;
    uint8_t sampler_slot;
} _sg_shader_image_sampler_t;

typedef struct {
    uint32_t required_bindings_and_uniforms;
    bool is_compute;
    _sg_shader_attr_t attrs[SG_MAX_VERTEX_ATTRIBUTES];
    _sg_shader_uniform_block_t uniform_blocks[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
    _sg_shader_storage_buffer_t storage_buffers[SG_MAX_STORAGEBUFFER_BINDSLOTS];
    _sg_shader_image_t images[SG_MAX_IMAGE_BINDSLOTS];
    _sg_shader_sampler_t samplers[SG_MAX_SAMPLER_BINDSLOTS];
    _sg_shader_image_sampler_t image_samplers[SG_MAX_IMAGE_SAMPLER_PAIRS];
    _sg_shader_storage_image_t storage_images[SG_MAX_STORAGE_ATTACHMENTS];
} _sg_shader_common_t;

typedef struct {
    bool vertex_buffer_layout_active[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    bool use_instanced_draw;
    bool is_compute;
    uint32_t required_bindings_and_uniforms;
    _sg_shader_ref_t shader;
    sg_vertex_layout_state layout;
    sg_depth_state depth;
    sg_stencil_state stencil;
    int color_count;
    sg_color_target_state colors[SG_MAX_COLOR_ATTACHMENTS];
    sg_primitive_type primitive_type;
    sg_index_type index_type;
    sg_cull_mode cull_mode;
    sg_face_winding face_winding;
    int sample_count;
    sg_color blend_color;
    bool alpha_to_coverage_enabled;
} _sg_pipeline_common_t;

typedef struct {
    _sg_image_ref_t image;
    int mip_level;
    int slice;
} _sg_attachment_common_t;

typedef struct {
    int width;
    int height;
    int num_colors;
    bool has_render_attachments;
    bool has_storage_attachments;
    _sg_attachment_common_t colors[SG_MAX_COLOR_ATTACHMENTS];
    _sg_attachment_common_t resolves[SG_MAX_COLOR_ATTACHMENTS];
    _sg_attachment_common_t depth_stencil;
    _sg_attachment_common_t storages[SG_MAX_STORAGE_ATTACHMENTS];
} _sg_attachments_common_t;

#if defined(SOKOL_DUMMY_BACKEND)
typedef struct _sg_buffer_s {
    _sg_slot_t slot;
    _sg_buffer_common_t cmn;
} _sg_dummy_buffer_t;
typedef _sg_dummy_buffer_t _sg_buffer_t;

typedef struct _sg_image_s {
    _sg_slot_t slot;
    _sg_image_common_t cmn;
} _sg_dummy_image_t;
typedef _sg_dummy_image_t _sg_image_t;

typedef struct _sg_sampler_s {
    _sg_slot_t slot;
    _sg_sampler_common_t cmn;
} _sg_dummy_sampler_t;
typedef _sg_dummy_sampler_t _sg_sampler_t;

typedef struct _sg_shader_s {
    _sg_slot_t slot;
    _sg_shader_common_t cmn;
} _sg_dummy_shader_t;
typedef _sg_dummy_shader_t _sg_shader_t;

typedef struct _sg_pipeline_s {
    _sg_slot_t slot;
    _sg_pipeline_common_t cmn;
} _sg_dummy_pipeline_t;
typedef _sg_dummy_pipeline_t _sg_pipeline_t;

typedef struct _sg_attachments_s {
    _sg_slot_t slot;
    _sg_attachments_common_t cmn;
} _sg_dummy_attachments_t;
typedef _sg_dummy_attachments_t _sg_attachments_t;

#elif defined(_SOKOL_ANY_GL)

typedef enum {
    _SG_GL_GPUDIRTY_VERTEXBUFFER = (1<<0),
    _SG_GL_GPUDIRTY_INDEXBUFFER = (1<<1),
    _SG_GL_GPUDIRTY_STORAGEBUFFER = (1<<2),
    _SG_GL_GPUDIRTY_BUFFER_ALL = _SG_GL_GPUDIRTY_VERTEXBUFFER | _SG_GL_GPUDIRTY_INDEXBUFFER | _SG_GL_GPUDIRTY_STORAGEBUFFER,
} _sg_gl_gpudirty_t;

typedef struct _sg_buffer_s {
    _sg_slot_t slot;
    _sg_buffer_common_t cmn;
    struct {
        GLuint buf[SG_NUM_INFLIGHT_FRAMES];
        bool injected;  // if true, external buffers were injected with sg_buffer_desc.gl_buffers
        uint8_t gpu_dirty_flags; // combination of _sg_gl_gpudirty_t flags
    } gl;
} _sg_gl_buffer_t;
typedef _sg_gl_buffer_t _sg_buffer_t;

typedef struct _sg_image_s {
    _sg_slot_t slot;
    _sg_image_common_t cmn;
    struct {
        GLenum target;
        GLuint msaa_render_buffer;
        GLuint tex[SG_NUM_INFLIGHT_FRAMES];
        bool injected;  // if true, external textures were injected with sg_image_desc.gl_textures
    } gl;
} _sg_gl_image_t;
typedef _sg_gl_image_t _sg_image_t;

typedef struct _sg_sampler_s {
    _sg_slot_t slot;
    _sg_sampler_common_t cmn;
    struct {
        GLuint smp;
        bool injected;  // true if external sampler was injects in sg_sampler_desc.gl_sampler
    } gl;
} _sg_gl_sampler_t;
typedef _sg_gl_sampler_t _sg_sampler_t;

typedef struct {
    GLint gl_loc;
    sg_uniform_type type;
    uint16_t count;
    uint16_t offset;
} _sg_gl_uniform_t;

typedef struct {
    int num_uniforms;
    _sg_gl_uniform_t uniforms[SG_MAX_UNIFORMBLOCK_MEMBERS];
} _sg_gl_uniform_block_t;

typedef struct {
    _sg_str_t name;
} _sg_gl_shader_attr_t;

typedef struct _sg_shader_s {
    _sg_slot_t slot;
    _sg_shader_common_t cmn;
    struct {
        GLuint prog;
        _sg_gl_shader_attr_t attrs[SG_MAX_VERTEX_ATTRIBUTES];
        _sg_gl_uniform_block_t uniform_blocks[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        uint8_t sbuf_binding[SG_MAX_STORAGEBUFFER_BINDSLOTS];
        uint8_t simg_binding[SG_MAX_STORAGE_ATTACHMENTS];
        int8_t tex_slot[SG_MAX_IMAGE_SAMPLER_PAIRS]; // GL texture unit index
    } gl;
} _sg_gl_shader_t;
typedef _sg_gl_shader_t _sg_shader_t;

typedef struct {
    int8_t vb_index;        // -1 if attr is not enabled
    int8_t divisor;         // -1 if not initialized
    uint8_t stride;
    uint8_t size;
    uint8_t normalized;
    int offset;
    GLenum type;
    sg_shader_attr_base_type base_type;
} _sg_gl_attr_t;

typedef struct _sg_pipeline_s {
    _sg_slot_t slot;
    _sg_pipeline_common_t cmn;
    struct {
        _sg_gl_attr_t attrs[SG_MAX_VERTEX_ATTRIBUTES];
        sg_depth_state depth;
        sg_stencil_state stencil;
        sg_primitive_type primitive_type;
        sg_blend_state blend;
        sg_color_mask color_write_mask[SG_MAX_COLOR_ATTACHMENTS];
        sg_cull_mode cull_mode;
        sg_face_winding face_winding;
        int sample_count;
        bool alpha_to_coverage_enabled;
    } gl;
} _sg_gl_pipeline_t;
typedef _sg_gl_pipeline_t _sg_pipeline_t;

typedef struct _sg_attachments_s {
    _sg_slot_t slot;
    _sg_attachments_common_t cmn;
    struct {
        GLuint fb;
        GLuint msaa_resolve_framebuffer[SG_MAX_COLOR_ATTACHMENTS];
    } gl;
} _sg_gl_attachments_t;
typedef _sg_gl_attachments_t _sg_attachments_t;

typedef struct {
    _sg_gl_attr_t gl_attr;
    GLuint gl_vbuf;
} _sg_gl_cache_attr_t;

typedef struct {
    GLenum target;
    GLuint texture;
    GLuint sampler;
} _sg_gl_cache_texture_sampler_bind_slot;

#define _SG_GL_MAX_SBUF_BINDINGS (SG_MAX_STORAGEBUFFER_BINDSLOTS)
#define _SG_GL_MAX_SIMG_BINDINGS (SG_MAX_STORAGE_ATTACHMENTS)
#define _SG_GL_MAX_IMG_SMP_BINDINGS (SG_MAX_IMAGE_SAMPLER_PAIRS)
typedef struct {
    sg_depth_state depth;
    sg_stencil_state stencil;
    sg_blend_state blend;
    sg_color_mask color_write_mask[SG_MAX_COLOR_ATTACHMENTS];
    sg_cull_mode cull_mode;
    sg_face_winding face_winding;
    bool polygon_offset_enabled;
    int sample_count;
    sg_color blend_color;
    bool alpha_to_coverage_enabled;
    _sg_gl_cache_attr_t attrs[SG_MAX_VERTEX_ATTRIBUTES];
    GLuint vertex_buffer;
    GLuint index_buffer;
    GLuint storage_buffer;  // general bind point
    GLuint storage_buffers[_SG_GL_MAX_SBUF_BINDINGS];
    GLuint stored_vertex_buffer;
    GLuint stored_index_buffer;
    GLuint stored_storage_buffer;
    GLuint prog;
    _sg_gl_cache_texture_sampler_bind_slot texture_samplers[_SG_GL_MAX_IMG_SMP_BINDINGS];
    _sg_gl_cache_texture_sampler_bind_slot stored_texture_sampler;
    int cur_ib_offset;
    GLenum cur_primitive_type;
    GLenum cur_index_type;
    GLenum cur_active_texture;
    _sg_sref_t cur_pip;
} _sg_gl_state_cache_t;

typedef struct {
    bool valid;
    GLuint vao;
    _sg_gl_state_cache_t cache;
    bool ext_anisotropic;
    GLint max_anisotropy;
    sg_store_action color_store_actions[SG_MAX_COLOR_ATTACHMENTS];
    sg_store_action depth_store_action;
    sg_store_action stencil_store_action;
    #if _SOKOL_USE_WIN32_GL_LOADER
    HINSTANCE opengl32_dll;
    #endif
} _sg_gl_backend_t;

#elif defined(SOKOL_D3D11)

typedef struct _sg_buffer_s {
    _sg_slot_t slot;
    _sg_buffer_common_t cmn;
    struct {
        ID3D11Buffer* buf;
        ID3D11ShaderResourceView* srv;
        ID3D11UnorderedAccessView* uav;
    } d3d11;
} _sg_d3d11_buffer_t;
typedef _sg_d3d11_buffer_t _sg_buffer_t;

typedef struct _sg_image_s {
    _sg_slot_t slot;
    _sg_image_common_t cmn;
    struct {
        DXGI_FORMAT format;
        ID3D11Texture2D* tex2d;
        ID3D11Texture3D* tex3d;
        ID3D11Resource* res;    // either tex2d or tex3d
        ID3D11ShaderResourceView* srv;
    } d3d11;
} _sg_d3d11_image_t;
typedef _sg_d3d11_image_t _sg_image_t;

typedef struct _sg_sampler_s {
    _sg_slot_t slot;
    _sg_sampler_common_t cmn;
    struct {
        ID3D11SamplerState* smp;
    } d3d11;
} _sg_d3d11_sampler_t;
typedef _sg_d3d11_sampler_t _sg_sampler_t;

typedef struct {
    _sg_str_t sem_name;
    int sem_index;
} _sg_d3d11_shader_attr_t;

#define _SG_D3D11_MAX_STAGE_UB_BINDINGS (SG_MAX_UNIFORMBLOCK_BINDSLOTS)
#define _SG_D3D11_MAX_STAGE_SRV_BINDINGS (SG_MAX_IMAGE_BINDSLOTS + SG_MAX_STORAGEBUFFER_BINDSLOTS)
#define _SG_D3D11_MAX_STAGE_UAV_BINDINGS (SG_MAX_STORAGEBUFFER_BINDSLOTS + SG_MAX_STORAGE_ATTACHMENTS)
#define _SG_D3D11_MAX_STAGE_SMP_BINDINGS (SG_MAX_SAMPLER_BINDSLOTS)

typedef struct _sg_shader_s {
    _sg_slot_t slot;
    _sg_shader_common_t cmn;
    struct {
        _sg_d3d11_shader_attr_t attrs[SG_MAX_VERTEX_ATTRIBUTES];
        ID3D11VertexShader* vs;
        ID3D11PixelShader* fs;
        ID3D11ComputeShader* cs;
        void* vs_blob;
        size_t vs_blob_length;
        uint8_t ub_register_b_n[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        uint8_t img_register_t_n[SG_MAX_IMAGE_BINDSLOTS];
        uint8_t smp_register_s_n[SG_MAX_SAMPLER_BINDSLOTS];
        uint8_t sbuf_register_t_n[SG_MAX_STORAGEBUFFER_BINDSLOTS];
        uint8_t sbuf_register_u_n[SG_MAX_STORAGEBUFFER_BINDSLOTS];
        uint8_t simg_register_u_n[SG_MAX_STORAGE_ATTACHMENTS];
        ID3D11Buffer* all_cbufs[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        ID3D11Buffer* vs_cbufs[_SG_D3D11_MAX_STAGE_UB_BINDINGS];
        ID3D11Buffer* fs_cbufs[_SG_D3D11_MAX_STAGE_UB_BINDINGS];
        ID3D11Buffer* cs_cbufs[_SG_D3D11_MAX_STAGE_UB_BINDINGS];
    } d3d11;
} _sg_d3d11_shader_t;
typedef _sg_d3d11_shader_t _sg_shader_t;

typedef struct _sg_pipeline_s {
    _sg_slot_t slot;
    _sg_pipeline_common_t cmn;
    struct {
        UINT stencil_ref;
        UINT vb_strides[SG_MAX_VERTEXBUFFER_BINDSLOTS];
        D3D_PRIMITIVE_TOPOLOGY topology;
        DXGI_FORMAT index_format;
        ID3D11InputLayout* il;
        ID3D11RasterizerState* rs;
        ID3D11DepthStencilState* dss;
        ID3D11BlendState* bs;
    } d3d11;
} _sg_d3d11_pipeline_t;
typedef _sg_d3d11_pipeline_t _sg_pipeline_t;

typedef struct {
    union {
        ID3D11RenderTargetView* rtv;
        ID3D11DepthStencilView* dsv;
        ID3D11UnorderedAccessView* uav;
    } view;
} _sg_d3d11_attachment_t;

typedef struct _sg_attachments_s {
    _sg_slot_t slot;
    _sg_attachments_common_t cmn;
    struct {
        _sg_d3d11_attachment_t colors[SG_MAX_COLOR_ATTACHMENTS];
        _sg_d3d11_attachment_t resolves[SG_MAX_COLOR_ATTACHMENTS];
        _sg_d3d11_attachment_t depth_stencil;
        _sg_d3d11_attachment_t storages[SG_MAX_STORAGE_ATTACHMENTS];
    } d3d11;
} _sg_d3d11_attachments_t;
typedef _sg_d3d11_attachments_t _sg_attachments_t;

typedef struct {
    bool valid;
    ID3D11Device* dev;
    ID3D11DeviceContext* ctx;
    bool use_indexed_draw;
    bool use_instanced_draw;
    _sg_pipeline_t* cur_pipeline;
    sg_pipeline cur_pipeline_id;
    struct {
        ID3D11RenderTargetView* render_view;
        ID3D11RenderTargetView* resolve_view;
    } cur_pass;
    // on-demand loaded d3dcompiler_47.dll handles
    HINSTANCE d3dcompiler_dll;
    bool d3dcompiler_dll_load_failed;
    pD3DCompile D3DCompile_func;
    // global subresourcedata array for texture updates
    D3D11_SUBRESOURCE_DATA subres_data[SG_MAX_MIPMAPS * SG_MAX_TEXTUREARRAY_LAYERS];
} _sg_d3d11_backend_t;

#elif defined(SOKOL_METAL)

#if defined(_SG_TARGET_MACOS) || defined(_SG_TARGET_IOS_SIMULATOR)
#define _SG_MTL_UB_ALIGN (256)
#else
#define _SG_MTL_UB_ALIGN (16)
#endif
#define _SG_MTL_INVALID_SLOT_INDEX (0)

typedef struct {
    uint32_t frame_index;   // frame index at which it is safe to release this resource
    int slot_index;
} _sg_mtl_release_item_t;

typedef struct {
    NSMutableArray* pool;
    int num_slots;
    int free_queue_top;
    int* free_queue;
    int release_queue_front;
    int release_queue_back;
    _sg_mtl_release_item_t* release_queue;
} _sg_mtl_idpool_t;

typedef struct _sg_buffer_s {
    _sg_slot_t slot;
    _sg_buffer_common_t cmn;
    struct {
        int buf[SG_NUM_INFLIGHT_FRAMES];  // index into _sg_mtl_pool
    } mtl;
} _sg_mtl_buffer_t;
typedef _sg_mtl_buffer_t _sg_buffer_t;

typedef struct _sg_image_s {
    _sg_slot_t slot;
    _sg_image_common_t cmn;
    struct {
        int tex[SG_NUM_INFLIGHT_FRAMES];
    } mtl;
} _sg_mtl_image_t;
typedef _sg_mtl_image_t _sg_image_t;

typedef struct _sg_sampler_s {
    _sg_slot_t slot;
    _sg_sampler_common_t cmn;
    struct {
        int sampler_state;
    } mtl;
} _sg_mtl_sampler_t;
typedef _sg_mtl_sampler_t _sg_sampler_t;

typedef struct {
    int mtl_lib;
    int mtl_func;
} _sg_mtl_shader_func_t;

typedef struct _sg_shader_s {
    _sg_slot_t slot;
    _sg_shader_common_t cmn;
    struct {
        _sg_mtl_shader_func_t vertex_func;
        _sg_mtl_shader_func_t fragment_func;
        _sg_mtl_shader_func_t compute_func;
        MTLSize threads_per_threadgroup;
        uint8_t ub_buffer_n[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        uint8_t img_texture_n[SG_MAX_IMAGE_BINDSLOTS];
        uint8_t smp_sampler_n[SG_MAX_SAMPLER_BINDSLOTS];
        uint8_t sbuf_buffer_n[SG_MAX_STORAGEBUFFER_BINDSLOTS];
        uint8_t simg_texture_n[SG_MAX_STORAGE_ATTACHMENTS];
    } mtl;
} _sg_mtl_shader_t;
typedef _sg_mtl_shader_t _sg_shader_t;

typedef struct _sg_pipeline_s {
    _sg_slot_t slot;
    _sg_pipeline_common_t cmn;
    struct {
        MTLPrimitiveType prim_type;
        int index_size;
        MTLIndexType index_type;
        MTLCullMode cull_mode;
        MTLWinding winding;
        uint32_t stencil_ref;
        MTLSize threads_per_threadgroup;
        int cps;    // MTLComputePipelineState
        int rps;    // MTLRenderPipelineState
        int dss;    // MTLDepthStencilState
    } mtl;
} _sg_mtl_pipeline_t;
typedef _sg_mtl_pipeline_t _sg_pipeline_t;

typedef struct _sg_attachments_s {
    _sg_slot_t slot;
    _sg_attachments_common_t cmn;
    struct {
        int storage_views[SG_MAX_STORAGE_ATTACHMENTS];
    } mtl;
} _sg_mtl_attachments_t;
typedef _sg_mtl_attachments_t _sg_attachments_t;

// resource binding state cache
#define _SG_MTL_MAX_STAGE_UB_BINDINGS (SG_MAX_UNIFORMBLOCK_BINDSLOTS)
#define _SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS (_SG_MTL_MAX_STAGE_UB_BINDINGS + SG_MAX_STORAGEBUFFER_BINDSLOTS)
#define _SG_MTL_MAX_STAGE_BUFFER_BINDINGS (_SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS + SG_MAX_VERTEXBUFFER_BINDSLOTS)
#define _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS (SG_MAX_IMAGE_BINDSLOTS + SG_MAX_STORAGE_ATTACHMENTS)
#define _SG_MTL_MAX_STAGE_SAMPLER_BINDINGS (SG_MAX_SAMPLER_BINDSLOTS)
typedef struct {
    _sg_sref_t cur_pip;
    _sg_buffer_ref_t cur_ibuf;
    int cur_ibuf_offset;
    int cur_vs_buffer_offsets[_SG_MTL_MAX_STAGE_BUFFER_BINDINGS];
    _sg_sref_t cur_vsbufs[_SG_MTL_MAX_STAGE_BUFFER_BINDINGS];
    _sg_sref_t cur_fsbufs[_SG_MTL_MAX_STAGE_BUFFER_BINDINGS];
    _sg_sref_t cur_csbufs[_SG_MTL_MAX_STAGE_BUFFER_BINDINGS];
    _sg_sref_t cur_vsimgs[_SG_MTL_MAX_STAGE_TEXTURE_BINDINGS];
    _sg_sref_t cur_fsimgs[_SG_MTL_MAX_STAGE_TEXTURE_BINDINGS];
    _sg_sref_t cur_vssmps[_SG_MTL_MAX_STAGE_SAMPLER_BINDINGS];
    _sg_sref_t cur_fssmps[_SG_MTL_MAX_STAGE_SAMPLER_BINDINGS];
    _sg_sref_t cur_cssmps[_SG_MTL_MAX_STAGE_SAMPLER_BINDINGS];
    // NOTE: special case: uint64_t for storage images, because we need
    // to differentiate between storage pass attachments and regular
    // textures bound to compute stages (but both binding types live
    // in the texture bind space in Metal)
    // This special case will be removed in the view update!
    uint64_t cur_cs_image_ids[_SG_MTL_MAX_STAGE_TEXTURE_BINDINGS];
} _sg_mtl_state_cache_t;

typedef struct {
    bool valid;
    bool use_shared_storage_mode;
    uint32_t cur_frame_rotate_index;
    int ub_size;
    int cur_ub_offset;
    uint8_t* cur_ub_base_ptr;
    _sg_mtl_state_cache_t state_cache;
    _sg_mtl_idpool_t idpool;
    dispatch_semaphore_t sem;
    id<MTLDevice> device;
    id<MTLCommandQueue> cmd_queue;
    id<MTLCommandBuffer> cmd_buffer;
    id<MTLRenderCommandEncoder> render_cmd_encoder;
    id<MTLComputeCommandEncoder> compute_cmd_encoder;
    id<CAMetalDrawable> cur_drawable;
    id<MTLBuffer> uniform_buffers[SG_NUM_INFLIGHT_FRAMES];
} _sg_mtl_backend_t;

#elif defined(SOKOL_WGPU)

#define _SG_WGPU_ROWPITCH_ALIGN (256)
#define _SG_WGPU_MAX_UNIFORM_UPDATE_SIZE (1<<16) // also see WGPULimits.maxUniformBufferBindingSize
#define _SG_WGPU_MAX_BINDGROUPS (3) // 0: uniforms, 1: images, samplers, storage buffers, 2: storage images (only in compute passes)
#define _SG_WGPU_UB_BINDGROUP_INDEX (0)
#define _SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX (1)
#define _SG_WGPU_SIMG_BINDGROUP_INDEX (2)
#define _SG_WGPU_MAX_UB_BINDGROUP_ENTRIES (SG_MAX_UNIFORMBLOCK_BINDSLOTS)
#define _SG_WGPU_MAX_UB_BINDGROUP_BIND_SLOTS (2 * SG_MAX_UNIFORMBLOCK_BINDSLOTS)
#define _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES (SG_MAX_IMAGE_BINDSLOTS + SG_MAX_SAMPLER_BINDSLOTS + SG_MAX_STORAGEBUFFER_BINDSLOTS)
#define _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS (128)
#define _SG_WGPU_MAX_SIMG_BIND_SLOTS (SG_MAX_STORAGE_ATTACHMENTS)
#define _SG_WGPU_MAX_SIMG_BINDGROUP_ENTRIES (SG_MAX_STORAGE_ATTACHMENTS)

typedef struct _sg_buffer_s {
    _sg_slot_t slot;
    _sg_buffer_common_t cmn;
    struct {
        WGPUBuffer buf;
    } wgpu;
} _sg_wgpu_buffer_t;
typedef _sg_wgpu_buffer_t _sg_buffer_t;

typedef struct _sg_image_s {
    _sg_slot_t slot;
    _sg_image_common_t cmn;
    struct {
        WGPUTexture tex;
        WGPUTextureView view;
    } wgpu;
} _sg_wgpu_image_t;
typedef _sg_wgpu_image_t _sg_image_t;

typedef struct _sg_sampler_s {
    _sg_slot_t slot;
    _sg_sampler_common_t cmn;
    struct {
        WGPUSampler smp;
    } wgpu;
} _sg_wgpu_sampler_t;
typedef _sg_wgpu_sampler_t _sg_sampler_t;

typedef struct {
    WGPUShaderModule module;
    _sg_str_t entry;
} _sg_wgpu_shader_func_t;

typedef struct _sg_shader_s {
    _sg_slot_t slot;
    _sg_shader_common_t cmn;
    struct {
        _sg_wgpu_shader_func_t vertex_func;
        _sg_wgpu_shader_func_t fragment_func;
        _sg_wgpu_shader_func_t compute_func;
        WGPUBindGroupLayout bgl_ub;
        WGPUBindGroup bg_ub;
        WGPUBindGroupLayout bgl_img_smp_sbuf;
        WGPUBindGroupLayout bgl_simg;
        // a mapping of sokol-gfx bind slots to setBindGroup dynamic-offset-array indices
        uint8_t ub_num_dynoffsets;
        uint8_t ub_dynoffsets[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        // indexed by sokol-gfx bind slot:
        uint8_t ub_grp0_bnd_n[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
        uint8_t img_grp1_bnd_n[SG_MAX_IMAGE_BINDSLOTS];
        uint8_t smp_grp1_bnd_n[SG_MAX_SAMPLER_BINDSLOTS];
        uint8_t sbuf_grp1_bnd_n[SG_MAX_STORAGEBUFFER_BINDSLOTS];
        uint8_t simg_grp2_bnd_n[SG_MAX_STORAGE_ATTACHMENTS];
    } wgpu;
} _sg_wgpu_shader_t;
typedef _sg_wgpu_shader_t _sg_shader_t;

typedef struct _sg_pipeline_s {
    _sg_slot_t slot;
    _sg_pipeline_common_t cmn;
    struct {
        WGPURenderPipeline rpip;
        WGPUComputePipeline cpip;
        WGPUColor blend_color;
    } wgpu;
} _sg_wgpu_pipeline_t;
typedef _sg_wgpu_pipeline_t _sg_pipeline_t;

typedef struct {
    WGPUTextureView view;
} _sg_wgpu_attachment_t;

typedef struct _sg_attachments_s {
    _sg_slot_t slot;
    _sg_attachments_common_t cmn;
    struct {
        _sg_wgpu_attachment_t colors[SG_MAX_COLOR_ATTACHMENTS];
        _sg_wgpu_attachment_t resolves[SG_MAX_COLOR_ATTACHMENTS];
        _sg_wgpu_attachment_t depth_stencil;
        _sg_wgpu_attachment_t storages[SG_MAX_STORAGE_ATTACHMENTS];
    } wgpu;
} _sg_wgpu_attachments_t;
typedef _sg_wgpu_attachments_t _sg_attachments_t;

// a pool of per-frame uniform buffers
typedef struct {
    uint32_t num_bytes;
    uint32_t offset;    // current offset into buf
    uint8_t* staging;   // intermediate buffer for uniform data updates
    WGPUBuffer buf;     // the GPU-side uniform buffer
    uint32_t bind_offsets[SG_MAX_UNIFORMBLOCK_BINDSLOTS];   // NOTE: index is sokol-gfx ub slot index!
} _sg_wgpu_uniform_buffer_t;

typedef struct {
    uint32_t id;
} _sg_wgpu_bindgroup_handle_t;

typedef enum {
    _SG_WGPU_BINDGROUPSCACHEITEMTYPE_NONE           = 0,
    _SG_WGPU_BINDGROUPSCACHEITEMTYPE_IMAGE          = 0x00001111,
    _SG_WGPU_BINDGROUPSCACHEITEMTYPE_SAMPLER        = 0x00002222,
    _SG_WGPU_BINDGROUPSCACHEITEMTYPE_STORAGEBUFFER  = 0x00003333,
    _SG_WGPU_BINDGROUPSCACHEITEMTYPE_PIPELINE       = 0x00004444,
} _sg_wgpu_bindgroups_cache_item_type_t;

#define _SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS (1 + _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES)
typedef struct {
    uint64_t hash;
    // the format of cache key items is BBBBTTTTIIIIIIII
    // where
    //  - BBBB is 2x the WGPU binding
    //  - TTTT is the _sg_wgpu_bindgroups_cache_item_type_t
    //  - IIIIIIII is the resource id
    //
    // where the item type is a per-resource-type bit pattern
    uint64_t items[_SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS];
} _sg_wgpu_bindgroups_cache_key_t;

typedef struct {
    uint32_t num;           // must be 2^n
    uint32_t index_mask;    // mask to turn hash into valid index
    _sg_wgpu_bindgroup_handle_t* items;
} _sg_wgpu_bindgroups_cache_t;

typedef struct {
    _sg_slot_t slot;
    WGPUBindGroup bindgroup;
    _sg_wgpu_bindgroups_cache_key_t key;
} _sg_wgpu_bindgroup_t;

typedef struct {
    _sg_pool_t pool;
    _sg_wgpu_bindgroup_t* bindgroups;
} _sg_wgpu_bindgroups_pool_t;

typedef struct {
    struct {
        sg_buffer buffer;
        uint64_t offset;
    } vbs[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    struct {
        sg_buffer buffer;
        uint64_t offset;
    } ib;
    _sg_wgpu_bindgroup_handle_t bg;
} _sg_wgpu_bindings_cache_t;

// the WGPU backend state
typedef struct {
    bool valid;
    bool use_indexed_draw;
    WGPUDevice dev;
    WGPUSupportedLimits limits;
    WGPUQueue queue;
    WGPUCommandEncoder cmd_enc;
    WGPURenderPassEncoder rpass_enc;
    WGPUComputePassEncoder cpass_enc;
    WGPUBindGroup empty_bind_group;
    const _sg_pipeline_t* cur_pipeline;
    sg_pipeline cur_pipeline_id;
    _sg_wgpu_uniform_buffer_t uniform;
    _sg_wgpu_bindings_cache_t bindings_cache;
    _sg_wgpu_bindgroups_cache_t bindgroups_cache;
    _sg_wgpu_bindgroups_pool_t bindgroups_pool;
} _sg_wgpu_backend_t;
#endif

// this *MUST* remain 0
#define _SG_INVALID_SLOT_INDEX (0)

typedef struct _sg_pools_s {
    _sg_pool_t buffer_pool;
    _sg_pool_t image_pool;
    _sg_pool_t sampler_pool;
    _sg_pool_t shader_pool;
    _sg_pool_t pipeline_pool;
    _sg_pool_t attachments_pool;
    _sg_buffer_t* buffers;
    _sg_image_t* images;
    _sg_sampler_t* samplers;
    _sg_shader_t* shaders;
    _sg_pipeline_t* pipelines;
    _sg_attachments_t* attachments;
} _sg_pools_t;

typedef struct {
    int num;        // number of allocated commit listener items
    int upper;      // the current upper index (no valid items past this point)
    sg_commit_listener* items;
} _sg_commit_listeners_t;

// resolved pass attachments struct
typedef struct {
    _sg_image_t* color_images[SG_MAX_COLOR_ATTACHMENTS];
    _sg_image_t* resolve_images[SG_MAX_COLOR_ATTACHMENTS];
    _sg_image_t* ds_image;
    _sg_image_t* storage_images[SG_MAX_STORAGE_ATTACHMENTS];
} _sg_attachments_ptrs_t;

// resolved resource bindings struct
typedef struct {
    _sg_pipeline_t* pip;
    int vb_offsets[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    int ib_offset;
    _sg_buffer_t* vbs[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    _sg_buffer_t* ib;
    _sg_image_t* imgs[SG_MAX_IMAGE_BINDSLOTS];
    _sg_sampler_t* smps[SG_MAX_SAMPLER_BINDSLOTS];
    _sg_buffer_t* sbufs[SG_MAX_STORAGEBUFFER_BINDSLOTS];
    _sg_image_t* simgs[SG_MAX_STORAGE_ATTACHMENTS];
} _sg_bindings_ptrs_t;

typedef struct {
    bool sample;
    bool filter;
    bool render;
    bool blend;
    bool msaa;
    bool depth;
    bool read;
    bool write;
} _sg_pixelformat_info_t;

typedef struct {
    bool valid;
    sg_desc desc;       // original desc with default values patched in
    uint32_t frame_index;
    struct {
        bool valid;
        bool in_pass;
        bool is_compute;
        _sg_attachments_ref_t atts; // null in a swapchain pass
        int width;
        int height;
        struct {
            sg_pixel_format color_fmt;
            sg_pixel_format depth_fmt;
            int sample_count;
        } swapchain;
    } cur_pass;
    _sg_pipeline_ref_t cur_pip;
    bool next_draw_valid;
    uint32_t required_bindings_and_uniforms;    // used to check that bindings and uniforms are applied after applying pipeline
    uint32_t applied_bindings_and_uniforms;     // bits 0..7: uniform blocks, bit 8: bindings
    #if defined(SOKOL_DEBUG)
    sg_log_item validate_error;
    #endif
    struct {
        _sg_tracker_t readwrite_sbufs;  // tracks read/write storage buffers used in compute pass
    } compute;
    _sg_pools_t pools;
    sg_backend backend;
    sg_features features;
    sg_limits limits;
    _sg_pixelformat_info_t formats[_SG_PIXELFORMAT_NUM];
    bool stats_enabled;
    sg_frame_stats stats;
    sg_frame_stats prev_stats;
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_backend_t gl;
    #elif defined(SOKOL_METAL)
    _sg_mtl_backend_t mtl;
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_backend_t d3d11;
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_backend_t wgpu;
    #endif
    #if defined(SOKOL_TRACE_HOOKS)
    sg_trace_hooks hooks;
    #endif
    _sg_commit_listeners_t commit_listeners;
} _sg_state_t;
static _sg_state_t _sg;

// ██       ██████   ██████   ██████  ██ ███    ██  ██████
// ██      ██    ██ ██       ██       ██ ████   ██ ██
// ██      ██    ██ ██   ███ ██   ███ ██ ██ ██  ██ ██   ███
// ██      ██    ██ ██    ██ ██    ██ ██ ██  ██ ██ ██    ██
// ███████  ██████   ██████   ██████  ██ ██   ████  ██████
//
// >>logging
#if defined(SOKOL_DEBUG)
#define _SG_LOGITEM_XMACRO(item,msg) #item ": " msg,
static const char* _sg_log_messages[] = {
    _SG_LOG_ITEMS
};
#undef _SG_LOGITEM_XMACRO
#endif // SOKOL_DEBUG

#define _SG_PANIC(code) _sg_log(SG_LOGITEM_ ##code, 0, 0, __LINE__)
#define _SG_ERROR(code) _sg_log(SG_LOGITEM_ ##code, 1, 0, __LINE__)
#define _SG_WARN(code) _sg_log(SG_LOGITEM_ ##code, 2, 0, __LINE__)
#define _SG_INFO(code) _sg_log(SG_LOGITEM_ ##code, 3, 0, __LINE__)
#define _SG_LOGMSG(code,msg) _sg_log(SG_LOGITEM_ ##code, 3, msg, __LINE__)
#define _SG_VALIDATE(cond,code) if (!(cond)){ _sg.validate_error = SG_LOGITEM_ ##code; _sg_log(SG_LOGITEM_ ##code, 1, 0, __LINE__); }

static void _sg_log(sg_log_item log_item, uint32_t log_level, const char* msg, uint32_t line_nr) {
    if (_sg.desc.logger.func) {
        const char* filename = 0;
        #if defined(SOKOL_DEBUG)
            filename = __FILE__;
            if (0 == msg) {
                msg = _sg_log_messages[log_item];
            }
        #endif
        _sg.desc.logger.func("sg", log_level, (uint32_t)log_item, msg, line_nr, filename, _sg.desc.logger.user_data);
    } else {
        // for log level PANIC it would be 'undefined behaviour' to continue
        if (log_level == 0) {
            abort();
        }
    }
}

// ███    ███ ███████ ███    ███  ██████  ██████  ██    ██
// ████  ████ ██      ████  ████ ██    ██ ██   ██  ██  ██
// ██ ████ ██ █████   ██ ████ ██ ██    ██ ██████    ████
// ██  ██  ██ ██      ██  ██  ██ ██    ██ ██   ██    ██
// ██      ██ ███████ ██      ██  ██████  ██   ██    ██
//
// >>memory

// a helper macro to clear a struct with potentially ARC'ed ObjC references
#if defined(SOKOL_METAL)
    #if defined(__cplusplus)
        #define _SG_CLEAR_ARC_STRUCT(type, item) { item = type(); }
    #else
        #define _SG_CLEAR_ARC_STRUCT(type, item) { item = (type) { 0 }; }
    #endif
#else
    #define _SG_CLEAR_ARC_STRUCT(type, item) { _sg_clear(&item, sizeof(item)); }
#endif

_SOKOL_PRIVATE void _sg_clear(void* ptr, size_t size) {
    SOKOL_ASSERT(ptr && (size > 0));
    memset(ptr, 0, size);
}

_SOKOL_PRIVATE void* _sg_malloc(size_t size) {
    SOKOL_ASSERT(size > 0);
    void* ptr;
    if (_sg.desc.allocator.alloc_fn) {
        ptr = _sg.desc.allocator.alloc_fn(size, _sg.desc.allocator.user_data);
    } else {
        ptr = malloc(size);
    }
    if (0 == ptr) {
        _SG_PANIC(MALLOC_FAILED);
    }
    return ptr;
}

_SOKOL_PRIVATE void* _sg_malloc_clear(size_t size) {
    void* ptr = _sg_malloc(size);
    _sg_clear(ptr, size);
    return ptr;
}

_SOKOL_PRIVATE void _sg_free(void* ptr) {
    if (_sg.desc.allocator.free_fn) {
        _sg.desc.allocator.free_fn(ptr, _sg.desc.allocator.user_data);
    } else {
        free(ptr);
    }
}

_SOKOL_PRIVATE bool _sg_strempty(const _sg_str_t* str) {
    return 0 == str->buf[0];
}

_SOKOL_PRIVATE const char* _sg_strptr(const _sg_str_t* str) {
    return &str->buf[0];
}

_SOKOL_PRIVATE void _sg_strcpy(_sg_str_t* dst, const char* src) {
    SOKOL_ASSERT(dst);
    if (src) {
        #if defined(_MSC_VER)
        strncpy_s(dst->buf, _SG_STRING_SIZE, src, (_SG_STRING_SIZE-1));
        #else
        strncpy(dst->buf, src, _SG_STRING_SIZE);
        #endif
        dst->buf[_SG_STRING_SIZE-1] = 0;
    } else {
        _sg_clear(dst->buf, _SG_STRING_SIZE);
    }
}

// ██████   ██████   ██████  ██
// ██   ██ ██    ██ ██    ██ ██
// ██████  ██    ██ ██    ██ ██
// ██      ██    ██ ██    ██ ██
// ██       ██████   ██████  ███████
//
// >>pool
_SOKOL_PRIVATE void _sg_pool_init(_sg_pool_t* pool, int num) {
    SOKOL_ASSERT(pool && (num >= 1));
    // slot 0 is reserved for the 'invalid id', so bump the pool size by 1
    pool->size = num + 1;
    pool->queue_top = 0;
    // generation counters indexable by pool slot index, slot 0 is reserved
    size_t gen_ctrs_size = sizeof(uint32_t) * (size_t)pool->size;
    pool->gen_ctrs = (uint32_t*)_sg_malloc_clear(gen_ctrs_size);
    // it's not a bug to only reserve 'num' here
    pool->free_queue = (int*) _sg_malloc_clear(sizeof(int) * (size_t)num);
    // never allocate the zero-th pool item since the invalid id is 0
    for (int i = pool->size-1; i >= 1; i--) {
        pool->free_queue[pool->queue_top++] = i;
    }
}

_SOKOL_PRIVATE void _sg_pool_discard(_sg_pool_t* pool) {
    SOKOL_ASSERT(pool);
    SOKOL_ASSERT(pool->free_queue);
    _sg_free(pool->free_queue);
    pool->free_queue = 0;
    SOKOL_ASSERT(pool->gen_ctrs);
    _sg_free(pool->gen_ctrs);
    pool->gen_ctrs = 0;
    pool->size = 0;
    pool->queue_top = 0;
}

_SOKOL_PRIVATE int _sg_pool_alloc_index(_sg_pool_t* pool) {
    SOKOL_ASSERT(pool);
    SOKOL_ASSERT(pool->free_queue);
    if (pool->queue_top > 0) {
        int slot_index = pool->free_queue[--pool->queue_top];
        SOKOL_ASSERT((slot_index > 0) && (slot_index < pool->size));
        return slot_index;
    } else {
        // pool exhausted
        return _SG_INVALID_SLOT_INDEX;
    }
}

_SOKOL_PRIVATE void _sg_pool_free_index(_sg_pool_t* pool, int slot_index) {
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < pool->size));
    SOKOL_ASSERT(pool);
    SOKOL_ASSERT(pool->free_queue);
    SOKOL_ASSERT(pool->queue_top < pool->size);
    #ifdef SOKOL_DEBUG
    // debug check against double-free
    for (int i = 0; i < pool->queue_top; i++) {
        SOKOL_ASSERT(pool->free_queue[i] != slot_index);
    }
    #endif
    pool->free_queue[pool->queue_top++] = slot_index;
    SOKOL_ASSERT(pool->queue_top <= (pool->size-1));
}

_SOKOL_PRIVATE void _sg_slot_reset(_sg_slot_t* slot) {
    SOKOL_ASSERT(slot);
    _sg_clear(slot, sizeof(_sg_slot_t));
}

_SOKOL_PRIVATE void _sg_reset_buffer_to_alloc_state(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    _sg_slot_t slot = buf->slot;
    _sg_clear(buf, sizeof(*buf));
    buf->slot = slot;
    buf->slot.uninit_count += 1;
    buf->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_reset_image_to_alloc_state(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    _sg_slot_t slot = img->slot;
    _sg_clear(img, sizeof(*img));
    img->slot = slot;
    img->slot.uninit_count += 1;
    img->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_reset_sampler_to_alloc_state(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    _sg_slot_t slot = smp->slot;
    _sg_clear(smp, sizeof(*smp));
    smp->slot = slot;
    smp->slot.uninit_count += 1;
    smp->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_reset_shader_to_alloc_state(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    _sg_slot_t slot = shd->slot;
    _sg_clear(shd, sizeof(*shd));
    shd->slot = slot;
    shd->slot.uninit_count += 1;
    shd->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_reset_pipeline_to_alloc_state(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _sg_slot_t slot = pip->slot;
    _sg_clear(pip, sizeof(*pip));
    pip->slot = slot;
    pip->slot.uninit_count += 1;
    pip->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_reset_attachments_to_alloc_state(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    _sg_slot_t slot = atts->slot;
    _sg_clear(atts, sizeof(*atts));
    atts->slot = slot;
    atts->slot.uninit_count += 1;
    atts->slot.state = SG_RESOURCESTATE_ALLOC;
}

_SOKOL_PRIVATE void _sg_setup_pools(_sg_pools_t* p, const sg_desc* desc) {
    SOKOL_ASSERT(p);
    SOKOL_ASSERT(desc);
    // note: the pools here will have an additional item, since slot 0 is reserved
    SOKOL_ASSERT((desc->buffer_pool_size > 0) && (desc->buffer_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->buffer_pool, desc->buffer_pool_size);
    size_t buffer_pool_byte_size = sizeof(_sg_buffer_t) * (size_t)p->buffer_pool.size;
    p->buffers = (_sg_buffer_t*) _sg_malloc_clear(buffer_pool_byte_size);

    SOKOL_ASSERT((desc->image_pool_size > 0) && (desc->image_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->image_pool, desc->image_pool_size);
    size_t image_pool_byte_size = sizeof(_sg_image_t) * (size_t)p->image_pool.size;
    p->images = (_sg_image_t*) _sg_malloc_clear(image_pool_byte_size);

    SOKOL_ASSERT((desc->sampler_pool_size > 0) && (desc->sampler_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->sampler_pool, desc->sampler_pool_size);
    size_t sampler_pool_byte_size = sizeof(_sg_sampler_t) * (size_t)p->sampler_pool.size;
    p->samplers = (_sg_sampler_t*) _sg_malloc_clear(sampler_pool_byte_size);

    SOKOL_ASSERT((desc->shader_pool_size > 0) && (desc->shader_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->shader_pool, desc->shader_pool_size);
    size_t shader_pool_byte_size = sizeof(_sg_shader_t) * (size_t)p->shader_pool.size;
    p->shaders = (_sg_shader_t*) _sg_malloc_clear(shader_pool_byte_size);

    SOKOL_ASSERT((desc->pipeline_pool_size > 0) && (desc->pipeline_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->pipeline_pool, desc->pipeline_pool_size);
    size_t pipeline_pool_byte_size = sizeof(_sg_pipeline_t) * (size_t)p->pipeline_pool.size;
    p->pipelines = (_sg_pipeline_t*) _sg_malloc_clear(pipeline_pool_byte_size);

    SOKOL_ASSERT((desc->attachments_pool_size > 0) && (desc->attachments_pool_size < _SG_MAX_POOL_SIZE));
    _sg_pool_init(&p->attachments_pool, desc->attachments_pool_size);
    size_t attachments_pool_byte_size = sizeof(_sg_attachments_t) * (size_t)p->attachments_pool.size;
    p->attachments = (_sg_attachments_t*) _sg_malloc_clear(attachments_pool_byte_size);
}

_SOKOL_PRIVATE void _sg_discard_pools(_sg_pools_t* p) {
    SOKOL_ASSERT(p);
    _sg_free(p->attachments); p->attachments = 0;
    _sg_free(p->pipelines);   p->pipelines = 0;
    _sg_free(p->shaders);     p->shaders = 0;
    _sg_free(p->samplers);    p->samplers = 0;
    _sg_free(p->images);      p->images = 0;
    _sg_free(p->buffers);     p->buffers = 0;
    _sg_pool_discard(&p->attachments_pool);
    _sg_pool_discard(&p->pipeline_pool);
    _sg_pool_discard(&p->shader_pool);
    _sg_pool_discard(&p->sampler_pool);
    _sg_pool_discard(&p->image_pool);
    _sg_pool_discard(&p->buffer_pool);
}

/* allocate the slot at slot_index:
    - bump the slot's generation counter
    - create a resource id from the generation counter and slot index
    - set the slot's id to this id
    - set the slot's state to ALLOC
    - return the resource id
*/
_SOKOL_PRIVATE uint32_t _sg_slot_alloc(_sg_pool_t* pool, _sg_slot_t* slot, int slot_index) {
    /* FIXME: add handling for an overflowing generation counter,
       for now, just overflow (another option is to disable
       the slot)
    */
    SOKOL_ASSERT(pool && pool->gen_ctrs);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < pool->size));
    SOKOL_ASSERT(slot->id == SG_INVALID_ID);
    SOKOL_ASSERT(slot->state == SG_RESOURCESTATE_INITIAL);
    uint32_t ctr = ++pool->gen_ctrs[slot_index];
    slot->id = (ctr<<_SG_SLOT_SHIFT)|(slot_index & _SG_SLOT_MASK);
    slot->state = SG_RESOURCESTATE_ALLOC;
    return slot->id;
}

// extract slot index from id
_SOKOL_PRIVATE int _sg_slot_index(uint32_t id) {
    int slot_index = (int) (id & _SG_SLOT_MASK);
    SOKOL_ASSERT(_SG_INVALID_SLOT_INDEX != slot_index);
    return slot_index;
}

// returns pointer to resource by id without matching id check
_SOKOL_PRIVATE _sg_buffer_t* _sg_buffer_at(uint32_t buf_id) {
    SOKOL_ASSERT(SG_INVALID_ID != buf_id);
    int slot_index = _sg_slot_index(buf_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.buffer_pool.size));
    return &_sg.pools.buffers[slot_index];
}

_SOKOL_PRIVATE _sg_image_t* _sg_image_at(uint32_t img_id) {
    SOKOL_ASSERT(SG_INVALID_ID != img_id);
    int slot_index = _sg_slot_index(img_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.image_pool.size));
    return &_sg.pools.images[slot_index];
}

_SOKOL_PRIVATE _sg_sampler_t* _sg_sampler_at(uint32_t smp_id) {
    SOKOL_ASSERT(SG_INVALID_ID != smp_id);
    int slot_index = _sg_slot_index(smp_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.sampler_pool.size));
    return &_sg.pools.samplers[slot_index];
}

_SOKOL_PRIVATE _sg_shader_t* _sg_shader_at(uint32_t shd_id) {
    SOKOL_ASSERT(SG_INVALID_ID != shd_id);
    int slot_index = _sg_slot_index(shd_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.shader_pool.size));
    return &_sg.pools.shaders[slot_index];
}

_SOKOL_PRIVATE _sg_pipeline_t* _sg_pipeline_at(uint32_t pip_id) {
    SOKOL_ASSERT(SG_INVALID_ID != pip_id);
    int slot_index = _sg_slot_index(pip_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.pipeline_pool.size));
    return &_sg.pools.pipelines[slot_index];
}

_SOKOL_PRIVATE _sg_attachments_t* _sg_attachments_at(uint32_t atts_id) {
    SOKOL_ASSERT(SG_INVALID_ID != atts_id);
    int slot_index = _sg_slot_index(atts_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < _sg.pools.attachments_pool.size));
    return &_sg.pools.attachments[slot_index];
}

// returns pointer to resource with matching id check, may return 0
_SOKOL_PRIVATE _sg_buffer_t* _sg_lookup_buffer(uint32_t buf_id) {
    if (SG_INVALID_ID != buf_id) {
        _sg_buffer_t* buf = _sg_buffer_at(buf_id);
        if (buf->slot.id == buf_id) {
            return buf;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_image_t* _sg_lookup_image(uint32_t img_id) {
    if (SG_INVALID_ID != img_id) {
        _sg_image_t* img = _sg_image_at(img_id);
        if (img->slot.id == img_id) {
            return img;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_sampler_t* _sg_lookup_sampler(uint32_t smp_id) {
    if (SG_INVALID_ID != smp_id) {
        _sg_sampler_t* smp = _sg_sampler_at(smp_id);
        if (smp->slot.id == smp_id) {
            return smp;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_shader_t* _sg_lookup_shader(uint32_t shd_id) {
    if (SG_INVALID_ID != shd_id) {
        _sg_shader_t* shd = _sg_shader_at(shd_id);
        if (shd->slot.id == shd_id) {
            return shd;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_pipeline_t* _sg_lookup_pipeline(uint32_t pip_id) {
    if (SG_INVALID_ID != pip_id) {
        _sg_pipeline_t* pip = _sg_pipeline_at(pip_id);
        if (pip->slot.id == pip_id) {
            return pip;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_attachments_t* _sg_lookup_attachments(uint32_t atts_id) {
    if (SG_INVALID_ID != atts_id) {
        _sg_attachments_t* atts = _sg_attachments_at(atts_id);
        if (atts->slot.id == atts_id) {
            return atts;
        }
    }
    return 0;
}

// ██████  ███████ ███████ ███████
// ██   ██ ██      ██      ██
// ██████  █████   █████   ███████
// ██   ██ ██      ██           ██
// ██   ██ ███████ ██      ███████
//
// >>refs
_SOKOL_PRIVATE _sg_sref_t _sg_sref(const _sg_slot_t* slot) {
    _sg_sref_t sref; _sg_clear(&sref, sizeof(sref));
    if (slot) {
        sref.id = slot->id;
        sref.uninit_count = slot->uninit_count;
    }
    return sref;
}

_SOKOL_PRIVATE bool _sg_sref_eql(const _sg_sref_t* sref, const _sg_slot_t* slot) {
    SOKOL_ASSERT(sref && slot);
    return (sref->id == slot->id) && (sref->uninit_count == slot->uninit_count);
}

_SOKOL_PRIVATE _sg_buffer_ref_t _sg_buffer_ref(_sg_buffer_t* buf_or_null) {
    _sg_buffer_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (buf_or_null) {
        _sg_buffer_t* buf = buf_or_null;
        SOKOL_ASSERT(buf->slot.id != SG_INVALID_ID);
        ref.ptr = buf;
        ref.sref = _sg_sref(&buf->slot);
    }
    return ref;
}

_SOKOL_PRIVATE _sg_image_ref_t _sg_image_ref(_sg_image_t* img_or_null) {
    _sg_image_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (img_or_null) {
        _sg_image_t* img = img_or_null;
        SOKOL_ASSERT(img->slot.id != SG_INVALID_ID);
        ref.ptr = img;
        ref.sref = _sg_sref(&img->slot);
    }
    return ref;
}

_SOKOL_PRIVATE _sg_sampler_ref_t _sg_sampler_ref(_sg_sampler_t* smp_or_null) {
    _sg_sampler_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (smp_or_null) {
        _sg_sampler_t* smp = smp_or_null;
        SOKOL_ASSERT(smp->slot.id != SG_INVALID_ID);
        ref.ptr = smp;
        ref.sref = _sg_sref(&smp->slot);
    }
    return ref;
}

_SOKOL_PRIVATE _sg_shader_ref_t _sg_shader_ref(_sg_shader_t* shd_or_null) {
    _sg_shader_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (shd_or_null) {
        _sg_shader_t* shd = shd_or_null;
        SOKOL_ASSERT(shd->slot.id != SG_INVALID_ID);
        ref.ptr = shd;
        ref.sref = _sg_sref(&shd->slot);
    }
    return ref;
}

_SOKOL_PRIVATE _sg_pipeline_ref_t _sg_pipeline_ref(_sg_pipeline_t* pip_or_null) {
    _sg_pipeline_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (pip_or_null) {
        _sg_pipeline_t* pip = pip_or_null;
        SOKOL_ASSERT(pip->slot.id != SG_INVALID_ID);
        ref.ptr = pip;
        ref.sref = _sg_sref(&pip->slot);
    }
    return ref;
}

_SOKOL_PRIVATE _sg_attachments_ref_t _sg_attachments_ref(_sg_attachments_t* atts_or_null) {
    _sg_attachments_ref_t ref; _sg_clear(&ref, sizeof(ref));
    if (atts_or_null) {
        _sg_attachments_t* atts = atts_or_null;
        SOKOL_ASSERT(atts->slot.id != SG_INVALID_ID);
        ref.ptr = atts;
        ref.sref = _sg_sref(&atts->slot);
    }
    return ref;
}

#define _SG_IMPL_RES_EQL(NAME,REF,RES) _SOKOL_PRIVATE bool NAME(const REF* ref, const RES* res) { SOKOL_ASSERT(ref && res); return _sg_sref_eql(&ref->sref, &res->slot); }
_SG_IMPL_RES_EQL(_sg_buffer_ref_eql, _sg_buffer_ref_t, _sg_buffer_t)
_SG_IMPL_RES_EQL(_sg_image_ref_eql, _sg_image_ref_t, _sg_image_t)
_SG_IMPL_RES_EQL(_sg_sampler_ref_eql, _sg_sampler_ref_t, _sg_sampler_t)
_SG_IMPL_RES_EQL(_sg_shader_ref_eql, _sg_shader_ref_t, _sg_shader_t)
_SG_IMPL_RES_EQL(_sg_pipeline_ref_eql, _sg_pipeline_ref_t, _sg_pipeline_t)
_SG_IMPL_RES_EQL(_sg_attachments_ref_eql, _sg_attachments_ref_t, _sg_attachments_t)

#define _SG_IMPL_RES_NULL(NAME,REF) _SOKOL_PRIVATE bool NAME(const REF* ref) { SOKOL_ASSERT(ref); return SG_INVALID_ID == ref->sref.id; }
_SG_IMPL_RES_NULL(_sg_buffer_ref_null, _sg_buffer_ref_t)
_SG_IMPL_RES_NULL(_sg_image_ref_null, _sg_image_ref_t)
_SG_IMPL_RES_NULL(_sg_sampler_ref_null, _sg_sampler_ref_t)
_SG_IMPL_RES_NULL(_sg_shader_ref_null, _sg_shader_ref_t)
_SG_IMPL_RES_NULL(_sg_pipeline_ref_null, _sg_pipeline_ref_t)
_SG_IMPL_RES_NULL(_sg_attachments_ref_null, _sg_attachments_ref_t)

#define _SG_IMPL_RES_ALIVE(NAME,REF) _SOKOL_PRIVATE bool NAME(const REF* ref) { SOKOL_ASSERT(ref); return ref->ptr && _sg_sref_eql(&ref->sref, &ref->ptr->slot); }
_SG_IMPL_RES_ALIVE(_sg_buffer_ref_alive, _sg_buffer_ref_t)
_SG_IMPL_RES_ALIVE(_sg_image_ref_alive, _sg_image_ref_t)
_SG_IMPL_RES_ALIVE(_sg_sampler_ref_alive, _sg_sampler_ref_t)
_SG_IMPL_RES_ALIVE(_sg_shader_ref_alive, _sg_shader_ref_t)
_SG_IMPL_RES_ALIVE(_sg_pipeline_ref_alive, _sg_pipeline_ref_t)
_SG_IMPL_RES_ALIVE(_sg_attachments_ref_alive, _sg_attachments_ref_t)

#define _SG_IMPL_RES_PTR(NAME,REF,RES) _SOKOL_PRIVATE RES* NAME(const REF* ref) { SOKOL_ASSERT(ref && ref->ptr && _sg_sref_eql(&ref->sref, &ref->ptr->slot)); return ref->ptr; }
_SG_IMPL_RES_PTR(_sg_buffer_ref_ptr, _sg_buffer_ref_t, _sg_buffer_t)
_SG_IMPL_RES_PTR(_sg_image_ref_ptr, _sg_image_ref_t, _sg_image_t)
_SG_IMPL_RES_PTR(_sg_sampler_ref_ptr, _sg_sampler_ref_t, _sg_sampler_t)
_SG_IMPL_RES_PTR(_sg_shader_ref_ptr, _sg_shader_ref_t, _sg_shader_t)
_SG_IMPL_RES_PTR(_sg_pipeline_ref_ptr, _sg_pipeline_ref_t, _sg_pipeline_t)
_SG_IMPL_RES_PTR(_sg_attachments_ref_ptr, _sg_attachments_ref_t, _sg_attachments_t)

#define _SG_IMPL_RES_PTR_OR_NULL(NAME,REF,RES) _SOKOL_PRIVATE RES* NAME(const REF* ref) { SOKOL_ASSERT(ref); if (SG_INVALID_ID != ref->sref.id) { SOKOL_ASSERT(_sg_sref_eql(&ref->sref, &ref->ptr->slot)); return ref->ptr; } else { return 0; } }
_SG_IMPL_RES_PTR_OR_NULL(_sg_buffer_ref_ptr_or_null, _sg_buffer_ref_t, _sg_buffer_t)
_SG_IMPL_RES_PTR_OR_NULL(_sg_image_ref_ptr_or_null, _sg_image_ref_t, _sg_image_t)
_SG_IMPL_RES_PTR_OR_NULL(_sg_sampler_ref_ptr_or_null, _sg_sampler_ref_t, _sg_sampler_t)
_SG_IMPL_RES_PTR_OR_NULL(_sg_shader_ref_ptr_or_null, _sg_shader_ref_t, _sg_shader_t)
_SG_IMPL_RES_PTR_OR_NULL(_sg_pipeline_ref_ptr_or_null, _sg_pipeline_ref_t, _sg_pipeline_t)
_SG_IMPL_RES_PTR_OR_NULL(_sg_attachments_ref_ptr_or_null, _sg_attachments_ref_t, _sg_attachments_t)

// ████████ ██████   █████   ██████ ██   ██ ███████ ██████
//    ██    ██   ██ ██   ██ ██      ██  ██  ██      ██   ██
//    ██    ██████  ███████ ██      █████   █████   ██████
//    ██    ██   ██ ██   ██ ██      ██  ██  ██      ██   ██
//    ██    ██   ██ ██   ██  ██████ ██   ██ ███████ ██   ██
//
// >>tracker
_SOKOL_PRIVATE void _sg_tracker_init(_sg_tracker_t* tracker, uint32_t num) {
    SOKOL_ASSERT(tracker);
    SOKOL_ASSERT(num > 0);
    SOKOL_ASSERT(0 == tracker->size);
    SOKOL_ASSERT(0 == tracker->cur);
    SOKOL_ASSERT(0 == tracker->items);
    tracker->size = (uint32_t)num;
    tracker->items = (uint32_t*)_sg_malloc_clear(num * sizeof(uint32_t));
}

_SOKOL_PRIVATE void _sg_tracker_discard(_sg_tracker_t* tracker) {
    SOKOL_ASSERT(tracker);
    if (tracker->items) {
        _sg_free(tracker->items);
    }
    tracker->size = 0;
    tracker->cur = 0;
    tracker->items = 0;
}

_SOKOL_PRIVATE void _sg_tracker_reset(_sg_tracker_t* tracker) {
    SOKOL_ASSERT(tracker && tracker->items);
    tracker->cur = 0;
}

_SOKOL_PRIVATE bool _sg_tracker_add(_sg_tracker_t* tracker, uint32_t res_id) {
    SOKOL_ASSERT(tracker && tracker->items);
    if (tracker->cur < tracker->size) {
        tracker->items[tracker->cur++] = res_id;
        return true;
    } else {
        return false;
    }
}

// ██   ██ ███████ ██      ██████  ███████ ██████  ███████
// ██   ██ ██      ██      ██   ██ ██      ██   ██ ██
// ███████ █████   ██      ██████  █████   ██████  ███████
// ██   ██ ██      ██      ██      ██      ██   ██      ██
// ██   ██ ███████ ███████ ██      ███████ ██   ██ ███████
//
// >>helpers

// helper macros
#define _sg_def(val, def) (((val) == 0) ? (def) : (val))
#define _sg_def_flt(val, def) (((val) == 0.0f) ? (def) : (val))
#define _sg_min(a,b) (((a)<(b))?(a):(b))
#define _sg_max(a,b) (((a)>(b))?(a):(b))
#define _sg_clamp(v,v0,v1) (((v)<(v0))?(v0):(((v)>(v1))?(v1):(v)))
#define _sg_fequal(val,cmp,delta) ((((val)-(cmp))> -(delta))&&(((val)-(cmp))<(delta)))
#define _sg_ispow2(val) ((val&(val-1))==0)
#define _sg_stats_add(key,val) {if(_sg.stats_enabled){ _sg.stats.key+=val;}}

_SOKOL_PRIVATE uint32_t _sg_align_u32(uint32_t val, uint32_t align) {
    SOKOL_ASSERT((align > 0) && ((align & (align - 1)) == 0));
    return (val + (align - 1)) & ~(align - 1);
}

typedef struct { int x, y, w, h; } _sg_recti_t;

_SOKOL_PRIVATE _sg_recti_t _sg_clipi(int x, int y, int w, int h, int clip_width, int clip_height) {
    x = _sg_min(_sg_max(0, x), clip_width-1);
    y = _sg_min(_sg_max(0, y), clip_height-1);
    if ((x + w) > clip_width) {
        w = clip_width - x;
    }
    if ((y + h) > clip_height) {
        h = clip_height - y;
    }
    w = _sg_max(w, 1);
    h = _sg_max(h, 1);
    const _sg_recti_t res = { x, y, w, h };
    return res;
}

_SOKOL_PRIVATE void _sg_buffer_common_init(_sg_buffer_common_t* cmn, const sg_buffer_desc* desc) {
    cmn->size = (int)desc->size;
    cmn->append_pos = 0;
    cmn->append_overflow = false;
    cmn->update_frame_index = 0;
    cmn->append_frame_index = 0;
    cmn->num_slots = desc->usage.immutable ? 1 : SG_NUM_INFLIGHT_FRAMES;
    cmn->active_slot = 0;
    cmn->usage = desc->usage;
}

_SOKOL_PRIVATE void _sg_image_common_init(_sg_image_common_t* cmn, const sg_image_desc* desc) {
    cmn->upd_frame_index = 0;
    cmn->num_slots = desc->usage.immutable ? 1 : SG_NUM_INFLIGHT_FRAMES;
    cmn->active_slot = 0;
    cmn->type = desc->type;
    cmn->width = desc->width;
    cmn->height = desc->height;
    cmn->num_slices = desc->num_slices;
    cmn->num_mipmaps = desc->num_mipmaps;
    cmn->usage = desc->usage;
    cmn->pixel_format = desc->pixel_format;
    cmn->sample_count = desc->sample_count;
}

_SOKOL_PRIVATE void _sg_sampler_common_init(_sg_sampler_common_t* cmn, const sg_sampler_desc* desc) {
    cmn->min_filter = desc->min_filter;
    cmn->mag_filter = desc->mag_filter;
    cmn->mipmap_filter = desc->mipmap_filter;
    cmn->wrap_u = desc->wrap_u;
    cmn->wrap_v = desc->wrap_v;
    cmn->wrap_w = desc->wrap_w;
    cmn->min_lod = desc->min_lod;
    cmn->max_lod = desc->max_lod;
    cmn->border_color = desc->border_color;
    cmn->compare = desc->compare;
    cmn->max_anisotropy = desc->max_anisotropy;
}

_SOKOL_PRIVATE void _sg_shader_common_init(_sg_shader_common_t* cmn, const sg_shader_desc* desc) {
    cmn->is_compute = desc->compute_func.source || desc->compute_func.bytecode.ptr;
    for (size_t i = 0; i < SG_MAX_VERTEX_ATTRIBUTES; i++) {
        cmn->attrs[i].base_type = desc->attrs[i].base_type;
    }
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        const sg_shader_uniform_block* src = &desc->uniform_blocks[i];
        _sg_shader_uniform_block_t* dst = &cmn->uniform_blocks[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            cmn->required_bindings_and_uniforms |= (1 << i);
            dst->stage = src->stage;
            dst->size = src->size;
        }
    }
    const uint32_t required_bindings_flag = (1 << SG_MAX_UNIFORMBLOCK_BINDSLOTS);
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        const sg_shader_storage_buffer* src = &desc->storage_buffers[i];
        _sg_shader_storage_buffer_t* dst = &cmn->storage_buffers[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            cmn->required_bindings_and_uniforms |= required_bindings_flag;
            dst->stage = src->stage;
            dst->readonly = src->readonly;
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        const sg_shader_image* src = &desc->images[i];
        _sg_shader_image_t* dst = &cmn->images[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            cmn->required_bindings_and_uniforms |= required_bindings_flag;
            dst->stage = src->stage;
            dst->image_type = src->image_type;
            dst->sample_type = src->sample_type;
            dst->multisampled = src->multisampled;
        }
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        const sg_shader_sampler* src = &desc->samplers[i];
        _sg_shader_sampler_t* dst = &cmn->samplers[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            cmn->required_bindings_and_uniforms |= required_bindings_flag;
            dst->stage = src->stage;
            dst->sampler_type = src->sampler_type;
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_SAMPLER_PAIRS; i++) {
        const sg_shader_image_sampler_pair* src = &desc->image_sampler_pairs[i];
        _sg_shader_image_sampler_t* dst = &cmn->image_samplers[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            dst->stage = src->stage;
            SOKOL_ASSERT((src->image_slot >= 0) && (src->image_slot < SG_MAX_IMAGE_BINDSLOTS));
            SOKOL_ASSERT(desc->images[src->image_slot].stage == src->stage);
            dst->image_slot = src->image_slot;
            SOKOL_ASSERT((src->sampler_slot >= 0) && (src->sampler_slot < SG_MAX_SAMPLER_BINDSLOTS));
            SOKOL_ASSERT(desc->samplers[src->sampler_slot].stage == src->stage);
            dst->sampler_slot = src->sampler_slot;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        const sg_shader_storage_image* src = &desc->storage_images[i];
        _sg_shader_storage_image_t* dst = &cmn->storage_images[i];
        if (src->stage != SG_SHADERSTAGE_NONE) {
            dst->stage = src->stage;
            dst->image_type = src->image_type;
            dst->access_format = src->access_format;
            dst->writeonly = src->writeonly;
        }
    }
}

_SOKOL_PRIVATE void _sg_pipeline_common_init(_sg_pipeline_common_t* cmn, const sg_pipeline_desc* desc, _sg_shader_t* shd) {
    SOKOL_ASSERT((desc->color_count >= 0) && (desc->color_count <= SG_MAX_COLOR_ATTACHMENTS));

    // FIXME: most of this isn't needed for compute pipelines

    const uint32_t required_bindings_flag = (1 << SG_MAX_UNIFORMBLOCK_BINDSLOTS);
    for (int i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
        const sg_vertex_attr_state* a_state = &desc->layout.attrs[i];
        if (a_state->format != SG_VERTEXFORMAT_INVALID) {
            SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
            cmn->vertex_buffer_layout_active[a_state->buffer_index] = true;
            cmn->required_bindings_and_uniforms |= required_bindings_flag;
        }
    }
    cmn->is_compute = desc->compute;
    cmn->use_instanced_draw = false;
    cmn->shader = _sg_shader_ref(shd);
    cmn->layout = desc->layout;
    cmn->depth = desc->depth;
    cmn->stencil = desc->stencil;
    cmn->color_count = desc->color_count;
    for (int i = 0; i < desc->color_count; i++) {
        cmn->colors[i] = desc->colors[i];
    }
    cmn->primitive_type = desc->primitive_type;
    cmn->index_type = desc->index_type;
    if (cmn->index_type != SG_INDEXTYPE_NONE) {
        cmn->required_bindings_and_uniforms |= required_bindings_flag;
    }
    cmn->cull_mode = desc->cull_mode;
    cmn->face_winding = desc->face_winding;
    cmn->sample_count = desc->sample_count;
    cmn->blend_color = desc->blend_color;
    cmn->alpha_to_coverage_enabled = desc->alpha_to_coverage_enabled;
}

_SOKOL_PRIVATE void _sg_attachment_common_init(_sg_attachment_common_t* cmn, const sg_attachment_desc* desc, _sg_image_t* img) {
    cmn->image = _sg_image_ref(img);
    cmn->mip_level = desc->mip_level;
    cmn->slice = desc->slice;
}

_SOKOL_PRIVATE void _sg_attachments_common_init(_sg_attachments_common_t* cmn, const sg_attachments_desc* desc, const _sg_attachments_ptrs_t* ptrs, int width, int height) {
    // NOTE: width/height will be 0 for storage image attachments
    cmn->width = width;
    cmn->height = height;
    for (size_t i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        if (desc->colors[i].image.id != SG_INVALID_ID) {
            cmn->num_colors++;
            _sg_attachment_common_init(&cmn->colors[i], &desc->colors[i], ptrs->color_images[i]);
            _sg_attachment_common_init(&cmn->resolves[i], &desc->resolves[i], ptrs->resolve_images[i]);
            cmn->has_render_attachments = true;
        }
    }
    if (desc->depth_stencil.image.id != SG_INVALID_ID) {
        _sg_attachment_common_init(&cmn->depth_stencil, &desc->depth_stencil, ptrs->ds_image);
        cmn->has_render_attachments = true;
    }
    // NOTE: storage attachment slots may be non-continuous,
    // so a 'num_storages' doesn't make sense
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (desc->storages[i].image.id != SG_INVALID_ID) {
            cmn->has_storage_attachments = true;
            _sg_attachment_common_init(&cmn->storages[i], &desc->storages[i], ptrs->storage_images[i]);
        }
    }
}

_SOKOL_PRIVATE int _sg_vertexformat_bytesize(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:     return 4;
        case SG_VERTEXFORMAT_FLOAT2:    return 8;
        case SG_VERTEXFORMAT_FLOAT3:    return 12;
        case SG_VERTEXFORMAT_FLOAT4:    return 16;
        case SG_VERTEXFORMAT_INT:       return 4;
        case SG_VERTEXFORMAT_INT2:      return 8;
        case SG_VERTEXFORMAT_INT3:      return 12;
        case SG_VERTEXFORMAT_INT4:      return 16;
        case SG_VERTEXFORMAT_UINT:      return 4;
        case SG_VERTEXFORMAT_UINT2:     return 8;
        case SG_VERTEXFORMAT_UINT3:     return 12;
        case SG_VERTEXFORMAT_UINT4:     return 16;
        case SG_VERTEXFORMAT_BYTE4:     return 4;
        case SG_VERTEXFORMAT_BYTE4N:    return 4;
        case SG_VERTEXFORMAT_UBYTE4:    return 4;
        case SG_VERTEXFORMAT_UBYTE4N:   return 4;
        case SG_VERTEXFORMAT_SHORT2:    return 4;
        case SG_VERTEXFORMAT_SHORT2N:   return 4;
        case SG_VERTEXFORMAT_USHORT2:   return 4;
        case SG_VERTEXFORMAT_USHORT2N:  return 4;
        case SG_VERTEXFORMAT_SHORT4:    return 8;
        case SG_VERTEXFORMAT_SHORT4N:   return 8;
        case SG_VERTEXFORMAT_USHORT4:   return 8;
        case SG_VERTEXFORMAT_USHORT4N:  return 8;
        case SG_VERTEXFORMAT_UINT10_N2: return 4;
        case SG_VERTEXFORMAT_HALF2:     return 4;
        case SG_VERTEXFORMAT_HALF4:     return 8;
        case SG_VERTEXFORMAT_INVALID:   return 0;
        default:
            SOKOL_UNREACHABLE;
            return -1;
    }
}

_SOKOL_PRIVATE const char* _sg_vertexformat_to_string(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:     return "FLOAT";
        case SG_VERTEXFORMAT_FLOAT2:    return "FLOAT2";
        case SG_VERTEXFORMAT_FLOAT3:    return "FLOAT3";
        case SG_VERTEXFORMAT_FLOAT4:    return "FLOAT4";
        case SG_VERTEXFORMAT_INT:       return "INT";
        case SG_VERTEXFORMAT_INT2:      return "INT2";
        case SG_VERTEXFORMAT_INT3:      return "INT3";
        case SG_VERTEXFORMAT_INT4:      return "INT4";
        case SG_VERTEXFORMAT_UINT:      return "UINT";
        case SG_VERTEXFORMAT_UINT2:     return "UINT2";
        case SG_VERTEXFORMAT_UINT3:     return "UINT3";
        case SG_VERTEXFORMAT_UINT4:     return "UINT4";
        case SG_VERTEXFORMAT_BYTE4:     return "BYTE4";
        case SG_VERTEXFORMAT_BYTE4N:    return "BYTE4N";
        case SG_VERTEXFORMAT_UBYTE4:    return "UBYTE4";
        case SG_VERTEXFORMAT_UBYTE4N:   return "UBYTE4N";
        case SG_VERTEXFORMAT_SHORT2:    return "SHORT2";
        case SG_VERTEXFORMAT_SHORT2N:   return "SHORT2N";
        case SG_VERTEXFORMAT_USHORT2:   return "USHORT2";
        case SG_VERTEXFORMAT_USHORT2N:  return "USHORT2N";
        case SG_VERTEXFORMAT_SHORT4:    return "SHORT4";
        case SG_VERTEXFORMAT_SHORT4N:   return "SHORT4N";
        case SG_VERTEXFORMAT_USHORT4:   return "USHORT4";
        case SG_VERTEXFORMAT_USHORT4N:  return "USHORT4N";
        case SG_VERTEXFORMAT_UINT10_N2: return "UINT10_N2";
        case SG_VERTEXFORMAT_HALF2:     return "HALF2";
        case SG_VERTEXFORMAT_HALF4:     return "HALF4";
        default:
            SOKOL_UNREACHABLE;
            return "INVALID";
    }
}

_SOKOL_PRIVATE const char* _sg_shaderattrbasetype_to_string(sg_shader_attr_base_type b) {
    switch (b) {
        case SG_SHADERATTRBASETYPE_UNDEFINED:   return "UNDEFINED";
        case SG_SHADERATTRBASETYPE_FLOAT:       return "FLOAT";
        case SG_SHADERATTRBASETYPE_SINT:        return "SINT";
        case SG_SHADERATTRBASETYPE_UINT:        return "UINT";
        default:
            SOKOL_UNREACHABLE;
            return "INVALID";
    }
}

_SOKOL_PRIVATE sg_shader_attr_base_type _sg_vertexformat_basetype(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:
        case SG_VERTEXFORMAT_FLOAT2:
        case SG_VERTEXFORMAT_FLOAT3:
        case SG_VERTEXFORMAT_FLOAT4:
        case SG_VERTEXFORMAT_HALF2:
        case SG_VERTEXFORMAT_HALF4:
        case SG_VERTEXFORMAT_BYTE4N:
        case SG_VERTEXFORMAT_UBYTE4N:
        case SG_VERTEXFORMAT_SHORT2N:
        case SG_VERTEXFORMAT_USHORT2N:
        case SG_VERTEXFORMAT_SHORT4N:
        case SG_VERTEXFORMAT_USHORT4N:
        case SG_VERTEXFORMAT_UINT10_N2:
            return SG_SHADERATTRBASETYPE_FLOAT;
        case SG_VERTEXFORMAT_INT:
        case SG_VERTEXFORMAT_INT2:
        case SG_VERTEXFORMAT_INT3:
        case SG_VERTEXFORMAT_INT4:
        case SG_VERTEXFORMAT_BYTE4:
        case SG_VERTEXFORMAT_SHORT2:
        case SG_VERTEXFORMAT_SHORT4:
            return SG_SHADERATTRBASETYPE_SINT;
        case SG_VERTEXFORMAT_UINT:
        case SG_VERTEXFORMAT_UINT2:
        case SG_VERTEXFORMAT_UINT3:
        case SG_VERTEXFORMAT_UINT4:
        case SG_VERTEXFORMAT_UBYTE4:
        case SG_VERTEXFORMAT_USHORT2:
        case SG_VERTEXFORMAT_USHORT4:
            return SG_SHADERATTRBASETYPE_UINT;
        default:
            SOKOL_UNREACHABLE;
            return SG_SHADERATTRBASETYPE_UNDEFINED;
    }
}

_SOKOL_PRIVATE uint32_t _sg_uniform_alignment(sg_uniform_type type, int array_count, sg_uniform_layout ub_layout) {
    if (ub_layout == SG_UNIFORMLAYOUT_NATIVE) {
        return 1;
    } else {
        SOKOL_ASSERT(array_count > 0);
        if (array_count == 1) {
            switch (type) {
                case SG_UNIFORMTYPE_FLOAT:
                case SG_UNIFORMTYPE_INT:
                    return 4;
                case SG_UNIFORMTYPE_FLOAT2:
                case SG_UNIFORMTYPE_INT2:
                    return 8;
                case SG_UNIFORMTYPE_FLOAT3:
                case SG_UNIFORMTYPE_FLOAT4:
                case SG_UNIFORMTYPE_INT3:
                case SG_UNIFORMTYPE_INT4:
                    return 16;
                case SG_UNIFORMTYPE_MAT4:
                    return 16;
                default:
                    SOKOL_UNREACHABLE;
                    return 1;
            }
        } else {
            return 16;
        }
    }
}

_SOKOL_PRIVATE uint32_t _sg_uniform_size(sg_uniform_type type, int array_count, sg_uniform_layout ub_layout) {
    SOKOL_ASSERT(array_count > 0);
    if (array_count == 1) {
        switch (type) {
            case SG_UNIFORMTYPE_FLOAT:
            case SG_UNIFORMTYPE_INT:
                return 4;
            case SG_UNIFORMTYPE_FLOAT2:
            case SG_UNIFORMTYPE_INT2:
                return 8;
            case SG_UNIFORMTYPE_FLOAT3:
            case SG_UNIFORMTYPE_INT3:
                return 12;
            case SG_UNIFORMTYPE_FLOAT4:
            case SG_UNIFORMTYPE_INT4:
                return 16;
            case SG_UNIFORMTYPE_MAT4:
                return 64;
            default:
                SOKOL_UNREACHABLE;
                return 0;
        }
    } else {
        if (ub_layout == SG_UNIFORMLAYOUT_NATIVE) {
            switch (type) {
                case SG_UNIFORMTYPE_FLOAT:
                case SG_UNIFORMTYPE_INT:
                    return 4 * (uint32_t)array_count;
                case SG_UNIFORMTYPE_FLOAT2:
                case SG_UNIFORMTYPE_INT2:
                    return 8 * (uint32_t)array_count;
                case SG_UNIFORMTYPE_FLOAT3:
                case SG_UNIFORMTYPE_INT3:
                    return 12 * (uint32_t)array_count;
                case SG_UNIFORMTYPE_FLOAT4:
                case SG_UNIFORMTYPE_INT4:
                    return 16 * (uint32_t)array_count;
                case SG_UNIFORMTYPE_MAT4:
                    return 64 * (uint32_t)array_count;
                default:
                    SOKOL_UNREACHABLE;
                    return 0;
            }
        } else {
            switch (type) {
                case SG_UNIFORMTYPE_FLOAT:
                case SG_UNIFORMTYPE_FLOAT2:
                case SG_UNIFORMTYPE_FLOAT3:
                case SG_UNIFORMTYPE_FLOAT4:
                case SG_UNIFORMTYPE_INT:
                case SG_UNIFORMTYPE_INT2:
                case SG_UNIFORMTYPE_INT3:
                case SG_UNIFORMTYPE_INT4:
                    return 16 * (uint32_t)array_count;
                case SG_UNIFORMTYPE_MAT4:
                    return 64 * (uint32_t)array_count;
                default:
                    SOKOL_UNREACHABLE;
                    return 0;
            }
        }
    }
}

_SOKOL_PRIVATE bool _sg_is_compressed_pixel_format(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_BC1_RGBA:
        case SG_PIXELFORMAT_BC2_RGBA:
        case SG_PIXELFORMAT_BC3_RGBA:
        case SG_PIXELFORMAT_BC3_SRGBA:
        case SG_PIXELFORMAT_BC4_R:
        case SG_PIXELFORMAT_BC4_RSN:
        case SG_PIXELFORMAT_BC5_RG:
        case SG_PIXELFORMAT_BC5_RGSN:
        case SG_PIXELFORMAT_BC6H_RGBF:
        case SG_PIXELFORMAT_BC6H_RGBUF:
        case SG_PIXELFORMAT_BC7_RGBA:
        case SG_PIXELFORMAT_BC7_SRGBA:
        case SG_PIXELFORMAT_ETC2_RGB8:
        case SG_PIXELFORMAT_ETC2_SRGB8:
        case SG_PIXELFORMAT_ETC2_RGB8A1:
        case SG_PIXELFORMAT_ETC2_RGBA8:
        case SG_PIXELFORMAT_ETC2_SRGB8A8:
        case SG_PIXELFORMAT_EAC_R11:
        case SG_PIXELFORMAT_EAC_R11SN:
        case SG_PIXELFORMAT_EAC_RG11:
        case SG_PIXELFORMAT_EAC_RG11SN:
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:
            return true;
        default:
            return false;
    }
}

_SOKOL_PRIVATE bool _sg_is_valid_attachment_color_format(sg_pixel_format fmt) {
    const int fmt_index = (int) fmt;
    SOKOL_ASSERT((fmt_index >= 0) && (fmt_index < _SG_PIXELFORMAT_NUM));
    return _sg.formats[fmt_index].render && !_sg.formats[fmt_index].depth;
}

_SOKOL_PRIVATE bool _sg_is_valid_attachment_depth_format(sg_pixel_format fmt) {
    const int fmt_index = (int) fmt;
    SOKOL_ASSERT((fmt_index >= 0) && (fmt_index < _SG_PIXELFORMAT_NUM));
    return _sg.formats[fmt_index].render && _sg.formats[fmt_index].depth;
}

_SOKOL_PRIVATE bool _sg_is_valid_attachment_storage_format(sg_pixel_format fmt) {
    const int fmt_index = (int) fmt;
    SOKOL_ASSERT((fmt_index >= 0) && (fmt_index < _SG_PIXELFORMAT_NUM));
    return _sg.formats[fmt_index].read || _sg.formats[fmt_index].write;
}

_SOKOL_PRIVATE bool _sg_is_depth_or_depth_stencil_format(sg_pixel_format fmt) {
    return (SG_PIXELFORMAT_DEPTH == fmt) || (SG_PIXELFORMAT_DEPTH_STENCIL == fmt);
}

_SOKOL_PRIVATE bool _sg_is_depth_stencil_format(sg_pixel_format fmt) {
    return (SG_PIXELFORMAT_DEPTH_STENCIL == fmt);
}

_SOKOL_PRIVATE int _sg_pixelformat_bytesize(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:
        case SG_PIXELFORMAT_R8SN:
        case SG_PIXELFORMAT_R8UI:
        case SG_PIXELFORMAT_R8SI:
            return 1;
        case SG_PIXELFORMAT_R16:
        case SG_PIXELFORMAT_R16SN:
        case SG_PIXELFORMAT_R16UI:
        case SG_PIXELFORMAT_R16SI:
        case SG_PIXELFORMAT_R16F:
        case SG_PIXELFORMAT_RG8:
        case SG_PIXELFORMAT_RG8SN:
        case SG_PIXELFORMAT_RG8UI:
        case SG_PIXELFORMAT_RG8SI:
            return 2;
        case SG_PIXELFORMAT_R32UI:
        case SG_PIXELFORMAT_R32SI:
        case SG_PIXELFORMAT_R32F:
        case SG_PIXELFORMAT_RG16:
        case SG_PIXELFORMAT_RG16SN:
        case SG_PIXELFORMAT_RG16UI:
        case SG_PIXELFORMAT_RG16SI:
        case SG_PIXELFORMAT_RG16F:
        case SG_PIXELFORMAT_RGBA8:
        case SG_PIXELFORMAT_SRGB8A8:
        case SG_PIXELFORMAT_RGBA8SN:
        case SG_PIXELFORMAT_RGBA8UI:
        case SG_PIXELFORMAT_RGBA8SI:
        case SG_PIXELFORMAT_BGRA8:
        case SG_PIXELFORMAT_RGB10A2:
        case SG_PIXELFORMAT_RG11B10F:
        case SG_PIXELFORMAT_RGB9E5:
            return 4;
        case SG_PIXELFORMAT_RG32UI:
        case SG_PIXELFORMAT_RG32SI:
        case SG_PIXELFORMAT_RG32F:
        case SG_PIXELFORMAT_RGBA16:
        case SG_PIXELFORMAT_RGBA16SN:
        case SG_PIXELFORMAT_RGBA16UI:
        case SG_PIXELFORMAT_RGBA16SI:
        case SG_PIXELFORMAT_RGBA16F:
            return 8;
        case SG_PIXELFORMAT_RGBA32UI:
        case SG_PIXELFORMAT_RGBA32SI:
        case SG_PIXELFORMAT_RGBA32F:
            return 16;
        case SG_PIXELFORMAT_DEPTH:
        case SG_PIXELFORMAT_DEPTH_STENCIL:
            return 4;
        default:
            SOKOL_UNREACHABLE;
            return 0;
    }
}

_SOKOL_PRIVATE int _sg_roundup(int val, int round_to) {
    return (val+(round_to-1)) & ~(round_to-1);
}

_SOKOL_PRIVATE uint32_t _sg_roundup_u32(uint32_t val, uint32_t round_to) {
    return (val+(round_to-1)) & ~(round_to-1);
}

_SOKOL_PRIVATE uint64_t _sg_roundup_u64(uint64_t val, uint64_t round_to) {
    return (val+(round_to-1)) & ~(round_to-1);
}

_SOKOL_PRIVATE bool _sg_multiple_u64(uint64_t val, uint64_t of) {
    return (val & (of-1)) == 0;
}

/* return row pitch for an image

    see ComputePitch in https://github.com/microsoft/DirectXTex/blob/master/DirectXTex/DirectXTexUtil.cpp
*/
_SOKOL_PRIVATE int _sg_row_pitch(sg_pixel_format fmt, int width, int row_align) {
    int pitch;
    switch (fmt) {
        case SG_PIXELFORMAT_BC1_RGBA:
        case SG_PIXELFORMAT_BC4_R:
        case SG_PIXELFORMAT_BC4_RSN:
        case SG_PIXELFORMAT_ETC2_RGB8:
        case SG_PIXELFORMAT_ETC2_SRGB8:
        case SG_PIXELFORMAT_ETC2_RGB8A1:
        case SG_PIXELFORMAT_EAC_R11:
        case SG_PIXELFORMAT_EAC_R11SN:
            pitch = ((width + 3) / 4) * 8;
            pitch = pitch < 8 ? 8 : pitch;
            break;
        case SG_PIXELFORMAT_BC2_RGBA:
        case SG_PIXELFORMAT_BC3_RGBA:
        case SG_PIXELFORMAT_BC3_SRGBA:
        case SG_PIXELFORMAT_BC5_RG:
        case SG_PIXELFORMAT_BC5_RGSN:
        case SG_PIXELFORMAT_BC6H_RGBF:
        case SG_PIXELFORMAT_BC6H_RGBUF:
        case SG_PIXELFORMAT_BC7_RGBA:
        case SG_PIXELFORMAT_BC7_SRGBA:
        case SG_PIXELFORMAT_ETC2_RGBA8:
        case SG_PIXELFORMAT_ETC2_SRGB8A8:
        case SG_PIXELFORMAT_EAC_RG11:
        case SG_PIXELFORMAT_EAC_RG11SN:
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:
            pitch = ((width + 3) / 4) * 16;
            pitch = pitch < 16 ? 16 : pitch;
            break;
        default:
            pitch = width * _sg_pixelformat_bytesize(fmt);
            break;
    }
    pitch = _sg_roundup(pitch, row_align);
    return pitch;
}

// compute the number of rows in a surface depending on pixel format
_SOKOL_PRIVATE int _sg_num_rows(sg_pixel_format fmt, int height) {
    int num_rows;
    switch (fmt) {
        case SG_PIXELFORMAT_BC1_RGBA:
        case SG_PIXELFORMAT_BC4_R:
        case SG_PIXELFORMAT_BC4_RSN:
        case SG_PIXELFORMAT_ETC2_RGB8:
        case SG_PIXELFORMAT_ETC2_SRGB8:
        case SG_PIXELFORMAT_ETC2_RGB8A1:
        case SG_PIXELFORMAT_ETC2_RGBA8:
        case SG_PIXELFORMAT_ETC2_SRGB8A8:
        case SG_PIXELFORMAT_EAC_R11:
        case SG_PIXELFORMAT_EAC_R11SN:
        case SG_PIXELFORMAT_EAC_RG11:
        case SG_PIXELFORMAT_EAC_RG11SN:
        case SG_PIXELFORMAT_BC2_RGBA:
        case SG_PIXELFORMAT_BC3_RGBA:
        case SG_PIXELFORMAT_BC3_SRGBA:
        case SG_PIXELFORMAT_BC5_RG:
        case SG_PIXELFORMAT_BC5_RGSN:
        case SG_PIXELFORMAT_BC6H_RGBF:
        case SG_PIXELFORMAT_BC6H_RGBUF:
        case SG_PIXELFORMAT_BC7_RGBA:
        case SG_PIXELFORMAT_BC7_SRGBA:
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:
            num_rows = ((height + 3) / 4);
            break;
        default:
            num_rows = height;
            break;
    }
    if (num_rows < 1) {
        num_rows = 1;
    }
    return num_rows;
}

// return size of a mipmap level
_SOKOL_PRIVATE int _sg_miplevel_dim(int base_dim, int mip_level) {
    return _sg_max(base_dim >> mip_level, 1);
}

/* return pitch of a 2D subimage / texture slice
    see ComputePitch in https://github.com/microsoft/DirectXTex/blob/master/DirectXTex/DirectXTexUtil.cpp
*/
_SOKOL_PRIVATE int _sg_surface_pitch(sg_pixel_format fmt, int width, int height, int row_align) {
    int num_rows = _sg_num_rows(fmt, height);
    return num_rows * _sg_row_pitch(fmt, width, row_align);
}

// capability table pixel format helper functions
_SOKOL_PRIVATE void _sg_pixelformat_all(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->filter = true;
    pfi->blend = true;
    pfi->render = true;
    pfi->msaa = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_s(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sf(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->filter = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sr(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->render = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sfr(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->filter = true;
    pfi->render = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_srmd(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->render = true;
    pfi->msaa = true;
    pfi->depth = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_srm(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->render = true;
    pfi->msaa = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sfrm(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->filter = true;
    pfi->render = true;
    pfi->msaa = true;
}
_SOKOL_PRIVATE void _sg_pixelformat_sbrm(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->blend = true;
    pfi->render = true;
    pfi->msaa = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sbr(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->blend = true;
    pfi->render = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_sfbr(_sg_pixelformat_info_t* pfi) {
    pfi->sample = true;
    pfi->filter = true;
    pfi->blend = true;
    pfi->render = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_compute_all(_sg_pixelformat_info_t* pfi) {
    pfi->read = true;
    pfi->write = true;
}

_SOKOL_PRIVATE void _sg_pixelformat_compute_writeonly(_sg_pixelformat_info_t* pfi) {
    pfi->read = false;
    pfi->write = true;
}

_SOKOL_PRIVATE sg_pass_action _sg_pass_action_defaults(const sg_pass_action* action) {
    SOKOL_ASSERT(action);
    sg_pass_action res = *action;
    for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        if (res.colors[i].load_action == _SG_LOADACTION_DEFAULT) {
            res.colors[i].load_action = SG_LOADACTION_CLEAR;
            res.colors[i].clear_value.r = SG_DEFAULT_CLEAR_RED;
            res.colors[i].clear_value.g = SG_DEFAULT_CLEAR_GREEN;
            res.colors[i].clear_value.b = SG_DEFAULT_CLEAR_BLUE;
            res.colors[i].clear_value.a = SG_DEFAULT_CLEAR_ALPHA;
        }
        if (res.colors[i].store_action == _SG_STOREACTION_DEFAULT) {
            res.colors[i].store_action = SG_STOREACTION_STORE;
        }
    }
    if (res.depth.load_action == _SG_LOADACTION_DEFAULT) {
        res.depth.load_action = SG_LOADACTION_CLEAR;
        res.depth.clear_value = SG_DEFAULT_CLEAR_DEPTH;
    }
    if (res.depth.store_action == _SG_STOREACTION_DEFAULT) {
        res.depth.store_action = SG_STOREACTION_DONTCARE;
    }
    if (res.stencil.load_action == _SG_LOADACTION_DEFAULT) {
        res.stencil.load_action = SG_LOADACTION_CLEAR;
        res.stencil.clear_value = SG_DEFAULT_CLEAR_STENCIL;
    }
    if (res.stencil.store_action == _SG_STOREACTION_DEFAULT) {
        res.stencil.store_action = SG_STOREACTION_DONTCARE;
    }
    return res;
}

// ██████  ██    ██ ███    ███ ███    ███ ██    ██     ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ██   ██ ██    ██ ████  ████ ████  ████  ██  ██      ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██   ██ ██    ██ ██ ████ ██ ██ ████ ██   ████       ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██   ██ ██    ██ ██  ██  ██ ██  ██  ██    ██        ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
// ██████   ██████  ██      ██ ██      ██    ██        ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>dummy backend
#if defined(SOKOL_DUMMY_BACKEND)

_SOKOL_PRIVATE void _sg_dummy_setup_backend(const sg_desc* desc) {
    SOKOL_ASSERT(desc);
    _SOKOL_UNUSED(desc);
    _sg.backend = SG_BACKEND_DUMMY;
    for (int i = SG_PIXELFORMAT_R8; i < SG_PIXELFORMAT_BC1_RGBA; i++) {
        _sg.formats[i].sample = true;
        _sg.formats[i].filter = true;
        _sg.formats[i].render = true;
        _sg.formats[i].blend = true;
        _sg.formats[i].msaa = true;
    }
    _sg.formats[SG_PIXELFORMAT_DEPTH].depth = true;
    _sg.formats[SG_PIXELFORMAT_DEPTH_STENCIL].depth = true;
}

_SOKOL_PRIVATE void _sg_dummy_discard_backend(void) {
    // empty
}

_SOKOL_PRIVATE void _sg_dummy_reset_state_cache(void) {
    // empty
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && desc);
    _SOKOL_UNUSED(buf);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    _SOKOL_UNUSED(buf);
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && desc);
    _SOKOL_UNUSED(img);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_image(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    _SOKOL_UNUSED(img);
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && desc);
    _SOKOL_UNUSED(smp);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    _SOKOL_UNUSED(smp);
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && desc);
    _SOKOL_UNUSED(shd);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    _SOKOL_UNUSED(shd);
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_pipeline(_sg_pipeline_t* pip, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && desc);
    _SOKOL_UNUSED(pip);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _SOKOL_UNUSED(pip);
}

_SOKOL_PRIVATE sg_resource_state _sg_dummy_create_attachments(_sg_attachments_t* atts, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && desc);
    _SOKOL_UNUSED(atts);
    _SOKOL_UNUSED(desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_dummy_discard_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    _SOKOL_UNUSED(atts);
}

_SOKOL_PRIVATE void _sg_dummy_begin_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass);
    _SOKOL_UNUSED(pass);
}

_SOKOL_PRIVATE void _sg_dummy_end_pass(void) {
    // empty
}

_SOKOL_PRIVATE void _sg_dummy_commit(void) {
    // empty
}

_SOKOL_PRIVATE void _sg_dummy_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    _SOKOL_UNUSED(x);
    _SOKOL_UNUSED(y);
    _SOKOL_UNUSED(w);
    _SOKOL_UNUSED(h);
    _SOKOL_UNUSED(origin_top_left);
}

_SOKOL_PRIVATE void _sg_dummy_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    _SOKOL_UNUSED(x);
    _SOKOL_UNUSED(y);
    _SOKOL_UNUSED(w);
    _SOKOL_UNUSED(h);
    _SOKOL_UNUSED(origin_top_left);
}

_SOKOL_PRIVATE void _sg_dummy_apply_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _SOKOL_UNUSED(pip);
}

_SOKOL_PRIVATE bool _sg_dummy_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip);
    _SOKOL_UNUSED(bnd);
    return true;
}

_SOKOL_PRIVATE void _sg_dummy_apply_uniforms(int ub_slot, const sg_range* data) {
    _SOKOL_UNUSED(ub_slot);
    _SOKOL_UNUSED(data);
}

_SOKOL_PRIVATE void _sg_dummy_draw(int base_element, int num_elements, int num_instances) {
    _SOKOL_UNUSED(base_element);
    _SOKOL_UNUSED(num_elements);
    _SOKOL_UNUSED(num_instances);
}

_SOKOL_PRIVATE void _sg_dummy_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    _SOKOL_UNUSED(num_groups_x);
    _SOKOL_UNUSED(num_groups_y);
    _SOKOL_UNUSED(num_groups_z);
}

_SOKOL_PRIVATE void _sg_dummy_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    _SOKOL_UNUSED(data);
    if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
        buf->cmn.active_slot = 0;
    }
}

_SOKOL_PRIVATE bool _sg_dummy_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    _SOKOL_UNUSED(data);
    if (new_frame) {
        if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
            buf->cmn.active_slot = 0;
        }
    }
    return true;
}

_SOKOL_PRIVATE void _sg_dummy_update_image(_sg_image_t* img, const sg_image_data* data) {
    SOKOL_ASSERT(img && data);
    _SOKOL_UNUSED(data);
    if (++img->cmn.active_slot >= img->cmn.num_slots) {
        img->cmn.active_slot = 0;
    }
}

//  ██████  ██████  ███████ ███    ██  ██████  ██          ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ██    ██ ██   ██ ██      ████   ██ ██       ██          ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██    ██ ██████  █████   ██ ██  ██ ██   ███ ██          ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██    ██ ██      ██      ██  ██ ██ ██    ██ ██          ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
//  ██████  ██      ███████ ██   ████  ██████  ███████     ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>opengl backend
#elif defined(_SOKOL_ANY_GL)

// optional GL loader for win32
#if defined(_SOKOL_USE_WIN32_GL_LOADER)

#ifndef SG_GL_FUNCS_EXT
#define SG_GL_FUNCS_EXT
#endif

// X Macro list of GL function names and signatures
#define _SG_GL_FUNCS \
    SG_GL_FUNCS_EXT \
    _SG_XMACRO(glBindVertexArray,                 void, (GLuint array)) \
    _SG_XMACRO(glFramebufferTextureLayer,         void, (GLenum target, GLenum attachment, GLuint texture, GLint level, GLint layer)) \
    _SG_XMACRO(glGenFramebuffers,                 void, (GLsizei n, GLuint * framebuffers)) \
    _SG_XMACRO(glBindFramebuffer,                 void, (GLenum target, GLuint framebuffer)) \
    _SG_XMACRO(glBindRenderbuffer,                void, (GLenum target, GLuint renderbuffer)) \
    _SG_XMACRO(glGetStringi,                      const GLubyte *, (GLenum name, GLuint index)) \
    _SG_XMACRO(glClearBufferfi,                   void, (GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil)) \
    _SG_XMACRO(glClearBufferfv,                   void, (GLenum buffer, GLint drawbuffer, const GLfloat * value)) \
    _SG_XMACRO(glClearBufferuiv,                  void, (GLenum buffer, GLint drawbuffer, const GLuint * value)) \
    _SG_XMACRO(glClearBufferiv,                   void, (GLenum buffer, GLint drawbuffer, const GLint * value)) \
    _SG_XMACRO(glDeleteRenderbuffers,             void, (GLsizei n, const GLuint * renderbuffers)) \
    _SG_XMACRO(glUniform1fv,                      void, (GLint location, GLsizei count, const GLfloat * value)) \
    _SG_XMACRO(glUniform2fv,                      void, (GLint location, GLsizei count, const GLfloat * value)) \
    _SG_XMACRO(glUniform3fv,                      void, (GLint location, GLsizei count, const GLfloat * value)) \
    _SG_XMACRO(glUniform4fv,                      void, (GLint location, GLsizei count, const GLfloat * value)) \
    _SG_XMACRO(glUniform1iv,                      void, (GLint location, GLsizei count, const GLint * value)) \
    _SG_XMACRO(glUniform2iv,                      void, (GLint location, GLsizei count, const GLint * value)) \
    _SG_XMACRO(glUniform3iv,                      void, (GLint location, GLsizei count, const GLint * value)) \
    _SG_XMACRO(glUniform4iv,                      void, (GLint location, GLsizei count, const GLint * value)) \
    _SG_XMACRO(glUniformMatrix4fv,                void, (GLint location, GLsizei count, GLboolean transpose, const GLfloat * value)) \
    _SG_XMACRO(glUseProgram,                      void, (GLuint program)) \
    _SG_XMACRO(glShaderSource,                    void, (GLuint shader, GLsizei count, const GLchar *const* string, const GLint * length)) \
    _SG_XMACRO(glLinkProgram,                     void, (GLuint program)) \
    _SG_XMACRO(glGetUniformLocation,              GLint, (GLuint program, const GLchar * name)) \
    _SG_XMACRO(glGetShaderiv,                     void, (GLuint shader, GLenum pname, GLint * params)) \
    _SG_XMACRO(glGetProgramInfoLog,               void, (GLuint program, GLsizei bufSize, GLsizei * length, GLchar * infoLog)) \
    _SG_XMACRO(glGetAttribLocation,               GLint, (GLuint program, const GLchar * name)) \
    _SG_XMACRO(glDisableVertexAttribArray,        void, (GLuint index)) \
    _SG_XMACRO(glDeleteShader,                    void, (GLuint shader)) \
    _SG_XMACRO(glDeleteProgram,                   void, (GLuint program)) \
    _SG_XMACRO(glCompileShader,                   void, (GLuint shader)) \
    _SG_XMACRO(glStencilFuncSeparate,             void, (GLenum face, GLenum func, GLint ref, GLuint mask)) \
    _SG_XMACRO(glStencilOpSeparate,               void, (GLenum face, GLenum sfail, GLenum dpfail, GLenum dppass)) \
    _SG_XMACRO(glRenderbufferStorageMultisample,  void, (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height)) \
    _SG_XMACRO(glDrawBuffers,                     void, (GLsizei n, const GLenum * bufs)) \
    _SG_XMACRO(glVertexAttribDivisor,             void, (GLuint index, GLuint divisor)) \
    _SG_XMACRO(glBufferSubData,                   void, (GLenum target, GLintptr offset, GLsizeiptr size, const void * data)) \
    _SG_XMACRO(glGenBuffers,                      void, (GLsizei n, GLuint * buffers)) \
    _SG_XMACRO(glCheckFramebufferStatus,          GLenum, (GLenum target)) \
    _SG_XMACRO(glFramebufferRenderbuffer,         void, (GLenum target, GLenum attachment, GLenum renderbuffertarget, GLuint renderbuffer)) \
    _SG_XMACRO(glCompressedTexImage2D,            void, (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const void * data)) \
    _SG_XMACRO(glCompressedTexImage3D,            void, (GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const void * data)) \
    _SG_XMACRO(glActiveTexture,                   void, (GLenum texture)) \
    _SG_XMACRO(glTexSubImage3D,                   void, (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const void * pixels)) \
    _SG_XMACRO(glRenderbufferStorage,             void, (GLenum target, GLenum internalformat, GLsizei width, GLsizei height)) \
    _SG_XMACRO(glGenTextures,                     void, (GLsizei n, GLuint * textures)) \
    _SG_XMACRO(glPolygonOffset,                   void, (GLfloat factor, GLfloat units)) \
    _SG_XMACRO(glDrawElements,                    void, (GLenum mode, GLsizei count, GLenum type, const void * indices)) \
    _SG_XMACRO(glDeleteFramebuffers,              void, (GLsizei n, const GLuint * framebuffers)) \
    _SG_XMACRO(glBlendEquationSeparate,           void, (GLenum modeRGB, GLenum modeAlpha)) \
    _SG_XMACRO(glDeleteTextures,                  void, (GLsizei n, const GLuint * textures)) \
    _SG_XMACRO(glGetProgramiv,                    void, (GLuint program, GLenum pname, GLint * params)) \
    _SG_XMACRO(glBindTexture,                     void, (GLenum target, GLuint texture)) \
    _SG_XMACRO(glTexImage3D,                      void, (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLenum format, GLenum type, const void * pixels)) \
    _SG_XMACRO(glCreateShader,                    GLuint, (GLenum type)) \
    _SG_XMACRO(glTexSubImage2D,                   void, (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const void * pixels)) \
    _SG_XMACRO(glFramebufferTexture2D,            void, (GLenum target, GLenum attachment, GLenum textarget, GLuint texture, GLint level)) \
    _SG_XMACRO(glCreateProgram,                   GLuint, (void)) \
    _SG_XMACRO(glViewport,                        void, (GLint x, GLint y, GLsizei width, GLsizei height)) \
    _SG_XMACRO(glDeleteBuffers,                   void, (GLsizei n, const GLuint * buffers)) \
    _SG_XMACRO(glDrawArrays,                      void, (GLenum mode, GLint first, GLsizei count)) \
    _SG_XMACRO(glDrawElementsInstanced,           void, (GLenum mode, GLsizei count, GLenum type, const void * indices, GLsizei instancecount)) \
    _SG_XMACRO(glVertexAttribPointer,             void, (GLuint index, GLint size, GLenum type, GLboolean normalized, GLsizei stride, const void * pointer)) \
    _SG_XMACRO(glVertexAttribIPointer,            void, (GLuint index, GLint size, GLenum type, GLsizei stride, const void * pointer)) \
    _SG_XMACRO(glUniform1i,                       void, (GLint location, GLint v0)) \
    _SG_XMACRO(glDisable,                         void, (GLenum cap)) \
    _SG_XMACRO(glColorMask,                       void, (GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha)) \
    _SG_XMACRO(glColorMaski,                      void, (GLuint buf, GLboolean red, GLboolean green, GLboolean blue, GLboolean alpha)) \
    _SG_XMACRO(glBindBuffer,                      void, (GLenum target, GLuint buffer)) \
    _SG_XMACRO(glDeleteVertexArrays,              void, (GLsizei n, const GLuint * arrays)) \
    _SG_XMACRO(glDepthMask,                       void, (GLboolean flag)) \
    _SG_XMACRO(glDrawArraysInstanced,             void, (GLenum mode, GLint first, GLsizei count, GLsizei instancecount)) \
    _SG_XMACRO(glScissor,                         void, (GLint x, GLint y, GLsizei width, GLsizei height)) \
    _SG_XMACRO(glGenRenderbuffers,                void, (GLsizei n, GLuint * renderbuffers)) \
    _SG_XMACRO(glBufferData,                      void, (GLenum target, GLsizeiptr size, const void * data, GLenum usage)) \
    _SG_XMACRO(glBlendFuncSeparate,               void, (GLenum sfactorRGB, GLenum dfactorRGB, GLenum sfactorAlpha, GLenum dfactorAlpha)) \
    _SG_XMACRO(glTexParameteri,                   void, (GLenum target, GLenum pname, GLint param)) \
    _SG_XMACRO(glGetIntegerv,                     void, (GLenum pname, GLint * data)) \
    _SG_XMACRO(glEnable,                          void, (GLenum cap)) \
    _SG_XMACRO(glBlitFramebuffer,                 void, (GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter)) \
    _SG_XMACRO(glStencilMask,                     void, (GLuint mask)) \
    _SG_XMACRO(glAttachShader,                    void, (GLuint program, GLuint shader)) \
    _SG_XMACRO(glGetError,                        GLenum, (void)) \
    _SG_XMACRO(glBlendColor,                      void, (GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha)) \
    _SG_XMACRO(glTexParameterf,                   void, (GLenum target, GLenum pname, GLfloat param)) \
    _SG_XMACRO(glTexParameterfv,                  void, (GLenum target, GLenum pname, const GLfloat* params)) \
    _SG_XMACRO(glGetShaderInfoLog,                void, (GLuint shader, GLsizei bufSize, GLsizei * length, GLchar * infoLog)) \
    _SG_XMACRO(glDepthFunc,                       void, (GLenum func)) \
    _SG_XMACRO(glStencilOp ,                      void, (GLenum fail, GLenum zfail, GLenum zpass)) \
    _SG_XMACRO(glStencilFunc,                     void, (GLenum func, GLint ref, GLuint mask)) \
    _SG_XMACRO(glEnableVertexAttribArray,         void, (GLuint index)) \
    _SG_XMACRO(glBlendFunc,                       void, (GLenum sfactor, GLenum dfactor)) \
    _SG_XMACRO(glReadBuffer,                      void, (GLenum src)) \
    _SG_XMACRO(glTexImage2D,                      void, (GLenum target, GLint level, GLint internalformat, GLsizei width, GLsizei height, GLint border, GLenum format, GLenum type, const void * pixels)) \
    _SG_XMACRO(glGenVertexArrays,                 void, (GLsizei n, GLuint * arrays)) \
    _SG_XMACRO(glFrontFace,                       void, (GLenum mode)) \
    _SG_XMACRO(glCullFace,                        void, (GLenum mode)) \
    _SG_XMACRO(glPixelStorei,                     void, (GLenum pname, GLint param)) \
    _SG_XMACRO(glBindSampler,                     void, (GLuint unit, GLuint sampler)) \
    _SG_XMACRO(glGenSamplers,                     void, (GLsizei n, GLuint* samplers)) \
    _SG_XMACRO(glSamplerParameteri,               void, (GLuint sampler, GLenum pname, GLint param)) \
    _SG_XMACRO(glSamplerParameterf,               void, (GLuint sampler, GLenum pname, GLfloat param)) \
    _SG_XMACRO(glSamplerParameterfv,              void, (GLuint sampler, GLenum pname, const GLfloat* params)) \
    _SG_XMACRO(glDeleteSamplers,                  void, (GLsizei n, const GLuint* samplers)) \
    _SG_XMACRO(glBindBufferBase,                  void, (GLenum target, GLuint index, GLuint buffer)) \
    _SG_XMACRO(glTexImage2DMultisample,           void, (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height, GLboolean fixedsamplelocations)) \
    _SG_XMACRO(glTexImage3DMultisample,           void, (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations)) \
    _SG_XMACRO(glDispatchCompute,                 void, (GLuint num_groups_x, GLuint num_groups_y, GLuint num_groups_z)) \
    _SG_XMACRO(glMemoryBarrier,                   void, (GLbitfield barriers)) \
    _SG_XMACRO(glBindImageTexture,                void, (GLuint unit, GLuint texture, GLint level, GLboolean layered, GLint layer, GLenum access, GLenum format)) \
    _SG_XMACRO(glTexStorage2DMultisample,         void, (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height, GLboolean fixedsamplelocations)) \
    _SG_XMACRO(glTexStorage2D,                    void, (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height)) \
    _SG_XMACRO(glTexStorage3DMultisample,         void, (GLenum target, GLsizei samples, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations)) \
    _SG_XMACRO(glTexStorage3D,                    void, (GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth)) \
    _SG_XMACRO(glCompressedTexSubImage2D,         void, (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const void *data)) \
    _SG_XMACRO(glCompressedTexSubImage3D,         void, (GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void *data))

// generate GL function pointer typedefs
#define _SG_XMACRO(name, ret, args) typedef ret (GL_APIENTRY* PFN_ ## name) args;
_SG_GL_FUNCS
#undef _SG_XMACRO

// generate GL function pointers
#define _SG_XMACRO(name, ret, args) static PFN_ ## name name;
_SG_GL_FUNCS
#undef _SG_XMACRO

// helper function to lookup GL functions in GL DLL
typedef PROC (WINAPI * _sg_wglGetProcAddress)(LPCSTR);
_SOKOL_PRIVATE void* _sg_gl_getprocaddr(const char* name, _sg_wglGetProcAddress wgl_getprocaddress) {
    void* proc_addr = (void*) wgl_getprocaddress(name);
    if (0 == proc_addr) {
        proc_addr = (void*) GetProcAddress(_sg.gl.opengl32_dll, name);
    }
    SOKOL_ASSERT(proc_addr);
    return proc_addr;
}

// populate GL function pointers
_SOKOL_PRIVATE  void _sg_gl_load_opengl(void) {
    SOKOL_ASSERT(0 == _sg.gl.opengl32_dll);
    _sg.gl.opengl32_dll = LoadLibraryA("opengl32.dll");
    SOKOL_ASSERT(_sg.gl.opengl32_dll);
    _sg_wglGetProcAddress wgl_getprocaddress = (_sg_wglGetProcAddress) GetProcAddress(_sg.gl.opengl32_dll, "wglGetProcAddress");
    SOKOL_ASSERT(wgl_getprocaddress);
    #define _SG_XMACRO(name, ret, args) name = (PFN_ ## name) _sg_gl_getprocaddr(#name, wgl_getprocaddress);
    _SG_GL_FUNCS
    #undef _SG_XMACRO
}

_SOKOL_PRIVATE void _sg_gl_unload_opengl(void) {
    SOKOL_ASSERT(_sg.gl.opengl32_dll);
    FreeLibrary(_sg.gl.opengl32_dll);
    _sg.gl.opengl32_dll = 0;
}
#endif // _SOKOL_USE_WIN32_GL_LOADER

//-- type translation ----------------------------------------------------------
_SOKOL_PRIVATE GLenum _sg_gl_buffer_target(const sg_buffer_usage* usg) {
    // NOTE: the buffer target returned here is only used for the bind point
    // to copy data into the buffer, expect for WebGL2, the bind point doesn't
    // need to match the later usage of the buffer (but because of the WebGL2
    // restriction we cannot simply select a random bind point, because in WebGL2
    // a buffer cannot 'switch' bind points later.
    if (usg->vertex_buffer) {
        return GL_ARRAY_BUFFER;
    } else if (usg->index_buffer) {
        return GL_ELEMENT_ARRAY_BUFFER;
    } else if (usg->storage_buffer) {
        return GL_SHADER_STORAGE_BUFFER;
    } else {
        SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_texture_target(sg_image_type t, int sample_count) {
    #if defined(SOKOL_GLCORE)
        const bool msaa = sample_count > 1;
        if (msaa) {
            switch (t) {
                case SG_IMAGETYPE_2D: return GL_TEXTURE_2D_MULTISAMPLE;
                case SG_IMAGETYPE_ARRAY: return GL_TEXTURE_2D_MULTISAMPLE_ARRAY;
                default: SOKOL_UNREACHABLE; return 0;
            }
        } else {
            switch (t) {
                case SG_IMAGETYPE_2D:   return GL_TEXTURE_2D;
                case SG_IMAGETYPE_CUBE: return GL_TEXTURE_CUBE_MAP;
                case SG_IMAGETYPE_3D:       return GL_TEXTURE_3D;
                case SG_IMAGETYPE_ARRAY:    return GL_TEXTURE_2D_ARRAY;
                default: SOKOL_UNREACHABLE; return 0;
            }
        }
    #else
        SOKOL_ASSERT(sample_count == 1); _SOKOL_UNUSED(sample_count);
        switch (t) {
            case SG_IMAGETYPE_2D:   return GL_TEXTURE_2D;
            case SG_IMAGETYPE_CUBE: return GL_TEXTURE_CUBE_MAP;
            case SG_IMAGETYPE_3D:       return GL_TEXTURE_3D;
            case SG_IMAGETYPE_ARRAY:    return GL_TEXTURE_2D_ARRAY;
            default: SOKOL_UNREACHABLE; return 0;
        }
    #endif
}

_SOKOL_PRIVATE GLenum _sg_gl_buffer_usage(const sg_buffer_usage* usg) {
    if (usg->immutable) {
        return GL_STATIC_DRAW;
    } else if (usg->dynamic_update) {
        return GL_DYNAMIC_DRAW;
    } else if (usg->stream_update) {
        return GL_STREAM_DRAW;
    } else {
        SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_shader_stage(sg_shader_stage stage) {
    switch (stage) {
        case SG_SHADERSTAGE_VERTEX:   return GL_VERTEX_SHADER;
        case SG_SHADERSTAGE_FRAGMENT: return GL_FRAGMENT_SHADER;
        case SG_SHADERSTAGE_COMPUTE:  return GL_COMPUTE_SHADER;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLint _sg_gl_vertexformat_size(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:     return 1;
        case SG_VERTEXFORMAT_FLOAT2:    return 2;
        case SG_VERTEXFORMAT_FLOAT3:    return 3;
        case SG_VERTEXFORMAT_FLOAT4:    return 4;
        case SG_VERTEXFORMAT_INT:       return 1;
        case SG_VERTEXFORMAT_INT2:      return 2;
        case SG_VERTEXFORMAT_INT3:      return 3;
        case SG_VERTEXFORMAT_INT4:      return 4;
        case SG_VERTEXFORMAT_UINT:      return 1;
        case SG_VERTEXFORMAT_UINT2:     return 2;
        case SG_VERTEXFORMAT_UINT3:     return 3;
        case SG_VERTEXFORMAT_UINT4:     return 4;
        case SG_VERTEXFORMAT_BYTE4:     return 4;
        case SG_VERTEXFORMAT_BYTE4N:    return 4;
        case SG_VERTEXFORMAT_UBYTE4:    return 4;
        case SG_VERTEXFORMAT_UBYTE4N:   return 4;
        case SG_VERTEXFORMAT_SHORT2:    return 2;
        case SG_VERTEXFORMAT_SHORT2N:   return 2;
        case SG_VERTEXFORMAT_USHORT2:   return 2;
        case SG_VERTEXFORMAT_USHORT2N:  return 2;
        case SG_VERTEXFORMAT_SHORT4:    return 4;
        case SG_VERTEXFORMAT_SHORT4N:   return 4;
        case SG_VERTEXFORMAT_USHORT4:   return 4;
        case SG_VERTEXFORMAT_USHORT4N:  return 4;
        case SG_VERTEXFORMAT_UINT10_N2: return 4;
        case SG_VERTEXFORMAT_HALF2:     return 2;
        case SG_VERTEXFORMAT_HALF4:     return 4;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_vertexformat_type(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:
        case SG_VERTEXFORMAT_FLOAT2:
        case SG_VERTEXFORMAT_FLOAT3:
        case SG_VERTEXFORMAT_FLOAT4:
            return GL_FLOAT;
        case SG_VERTEXFORMAT_INT:
        case SG_VERTEXFORMAT_INT2:
        case SG_VERTEXFORMAT_INT3:
        case SG_VERTEXFORMAT_INT4:
            return GL_INT;
        case SG_VERTEXFORMAT_UINT:
        case SG_VERTEXFORMAT_UINT2:
        case SG_VERTEXFORMAT_UINT3:
        case SG_VERTEXFORMAT_UINT4:
            return GL_UNSIGNED_INT;
        case SG_VERTEXFORMAT_BYTE4:
        case SG_VERTEXFORMAT_BYTE4N:
            return GL_BYTE;
        case SG_VERTEXFORMAT_UBYTE4:
        case SG_VERTEXFORMAT_UBYTE4N:
            return GL_UNSIGNED_BYTE;
        case SG_VERTEXFORMAT_SHORT2:
        case SG_VERTEXFORMAT_SHORT2N:
        case SG_VERTEXFORMAT_SHORT4:
        case SG_VERTEXFORMAT_SHORT4N:
            return GL_SHORT;
        case SG_VERTEXFORMAT_USHORT2:
        case SG_VERTEXFORMAT_USHORT2N:
        case SG_VERTEXFORMAT_USHORT4:
        case SG_VERTEXFORMAT_USHORT4N:
            return GL_UNSIGNED_SHORT;
        case SG_VERTEXFORMAT_UINT10_N2:
            return GL_UNSIGNED_INT_2_10_10_10_REV;
        case SG_VERTEXFORMAT_HALF2:
        case SG_VERTEXFORMAT_HALF4:
            return GL_HALF_FLOAT;
        default:
            SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLboolean _sg_gl_vertexformat_normalized(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_BYTE4N:
        case SG_VERTEXFORMAT_UBYTE4N:
        case SG_VERTEXFORMAT_SHORT2N:
        case SG_VERTEXFORMAT_USHORT2N:
        case SG_VERTEXFORMAT_SHORT4N:
        case SG_VERTEXFORMAT_USHORT4N:
        case SG_VERTEXFORMAT_UINT10_N2:
            return GL_TRUE;
        default:
            return GL_FALSE;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_primitive_type(sg_primitive_type t) {
    switch (t) {
        case SG_PRIMITIVETYPE_POINTS:           return GL_POINTS;
        case SG_PRIMITIVETYPE_LINES:            return GL_LINES;
        case SG_PRIMITIVETYPE_LINE_STRIP:       return GL_LINE_STRIP;
        case SG_PRIMITIVETYPE_TRIANGLES:        return GL_TRIANGLES;
        case SG_PRIMITIVETYPE_TRIANGLE_STRIP:   return GL_TRIANGLE_STRIP;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_index_type(sg_index_type t) {
    switch (t) {
        case SG_INDEXTYPE_NONE:     return 0;
        case SG_INDEXTYPE_UINT16:   return GL_UNSIGNED_SHORT;
        case SG_INDEXTYPE_UINT32:   return GL_UNSIGNED_INT;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_compare_func(sg_compare_func cmp) {
    switch (cmp) {
        case SG_COMPAREFUNC_NEVER:          return GL_NEVER;
        case SG_COMPAREFUNC_LESS:           return GL_LESS;
        case SG_COMPAREFUNC_EQUAL:          return GL_EQUAL;
        case SG_COMPAREFUNC_LESS_EQUAL:     return GL_LEQUAL;
        case SG_COMPAREFUNC_GREATER:        return GL_GREATER;
        case SG_COMPAREFUNC_NOT_EQUAL:      return GL_NOTEQUAL;
        case SG_COMPAREFUNC_GREATER_EQUAL:  return GL_GEQUAL;
        case SG_COMPAREFUNC_ALWAYS:         return GL_ALWAYS;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_stencil_op(sg_stencil_op op) {
    switch (op) {
        case SG_STENCILOP_KEEP:         return GL_KEEP;
        case SG_STENCILOP_ZERO:         return GL_ZERO;
        case SG_STENCILOP_REPLACE:      return GL_REPLACE;
        case SG_STENCILOP_INCR_CLAMP:   return GL_INCR;
        case SG_STENCILOP_DECR_CLAMP:   return GL_DECR;
        case SG_STENCILOP_INVERT:       return GL_INVERT;
        case SG_STENCILOP_INCR_WRAP:    return GL_INCR_WRAP;
        case SG_STENCILOP_DECR_WRAP:    return GL_DECR_WRAP;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_blend_factor(sg_blend_factor f) {
    switch (f) {
        case SG_BLENDFACTOR_ZERO:                   return GL_ZERO;
        case SG_BLENDFACTOR_ONE:                    return GL_ONE;
        case SG_BLENDFACTOR_SRC_COLOR:              return GL_SRC_COLOR;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_COLOR:    return GL_ONE_MINUS_SRC_COLOR;
        case SG_BLENDFACTOR_SRC_ALPHA:              return GL_SRC_ALPHA;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_ALPHA:    return GL_ONE_MINUS_SRC_ALPHA;
        case SG_BLENDFACTOR_DST_COLOR:              return GL_DST_COLOR;
        case SG_BLENDFACTOR_ONE_MINUS_DST_COLOR:    return GL_ONE_MINUS_DST_COLOR;
        case SG_BLENDFACTOR_DST_ALPHA:              return GL_DST_ALPHA;
        case SG_BLENDFACTOR_ONE_MINUS_DST_ALPHA:    return GL_ONE_MINUS_DST_ALPHA;
        case SG_BLENDFACTOR_SRC_ALPHA_SATURATED:    return GL_SRC_ALPHA_SATURATE;
        case SG_BLENDFACTOR_BLEND_COLOR:            return GL_CONSTANT_COLOR;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_COLOR:  return GL_ONE_MINUS_CONSTANT_COLOR;
        case SG_BLENDFACTOR_BLEND_ALPHA:            return GL_CONSTANT_ALPHA;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_ALPHA:  return GL_ONE_MINUS_CONSTANT_ALPHA;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_blend_op(sg_blend_op op) {
    switch (op) {
        case SG_BLENDOP_ADD:                return GL_FUNC_ADD;
        case SG_BLENDOP_SUBTRACT:           return GL_FUNC_SUBTRACT;
        case SG_BLENDOP_REVERSE_SUBTRACT:   return GL_FUNC_REVERSE_SUBTRACT;
        case SG_BLENDOP_MIN:                return GL_MIN;
        case SG_BLENDOP_MAX:                return GL_MAX;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_min_filter(sg_filter min_f, sg_filter mipmap_f) {
    if (min_f == SG_FILTER_NEAREST) {
        switch (mipmap_f) {
            case SG_FILTER_NEAREST: return GL_NEAREST_MIPMAP_NEAREST;
            case SG_FILTER_LINEAR:  return GL_NEAREST_MIPMAP_LINEAR;
            default: SOKOL_UNREACHABLE; return (GLenum)0;
        }
    } else if (min_f == SG_FILTER_LINEAR) {
        switch (mipmap_f) {
            case SG_FILTER_NEAREST: return GL_LINEAR_MIPMAP_NEAREST;
            case SG_FILTER_LINEAR:  return GL_LINEAR_MIPMAP_LINEAR;
            default: SOKOL_UNREACHABLE; return (GLenum)0;
        }
    } else {
        SOKOL_UNREACHABLE; return (GLenum)0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_mag_filter(sg_filter mag_f) {
    if (mag_f == SG_FILTER_NEAREST) {
        return GL_NEAREST;
    } else {
        return GL_LINEAR;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_wrap(sg_wrap w) {
    switch (w) {
        case SG_WRAP_CLAMP_TO_EDGE:     return GL_CLAMP_TO_EDGE;
        #if defined(SOKOL_GLCORE)
        case SG_WRAP_CLAMP_TO_BORDER:   return GL_CLAMP_TO_BORDER;
        #else
        case SG_WRAP_CLAMP_TO_BORDER:   return GL_CLAMP_TO_EDGE;
        #endif
        case SG_WRAP_REPEAT:            return GL_REPEAT;
        case SG_WRAP_MIRRORED_REPEAT:   return GL_MIRRORED_REPEAT;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_teximage_type(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:
        case SG_PIXELFORMAT_R8UI:
        case SG_PIXELFORMAT_RG8:
        case SG_PIXELFORMAT_RG8UI:
        case SG_PIXELFORMAT_RGBA8:
        case SG_PIXELFORMAT_SRGB8A8:
        case SG_PIXELFORMAT_RGBA8UI:
        case SG_PIXELFORMAT_BGRA8:
            return GL_UNSIGNED_BYTE;
        case SG_PIXELFORMAT_R8SN:
        case SG_PIXELFORMAT_R8SI:
        case SG_PIXELFORMAT_RG8SN:
        case SG_PIXELFORMAT_RG8SI:
        case SG_PIXELFORMAT_RGBA8SN:
        case SG_PIXELFORMAT_RGBA8SI:
            return GL_BYTE;
        case SG_PIXELFORMAT_R16:
        case SG_PIXELFORMAT_R16UI:
        case SG_PIXELFORMAT_RG16:
        case SG_PIXELFORMAT_RG16UI:
        case SG_PIXELFORMAT_RGBA16:
        case SG_PIXELFORMAT_RGBA16UI:
            return GL_UNSIGNED_SHORT;
        case SG_PIXELFORMAT_R16SN:
        case SG_PIXELFORMAT_R16SI:
        case SG_PIXELFORMAT_RG16SN:
        case SG_PIXELFORMAT_RG16SI:
        case SG_PIXELFORMAT_RGBA16SN:
        case SG_PIXELFORMAT_RGBA16SI:
            return GL_SHORT;
        case SG_PIXELFORMAT_R16F:
        case SG_PIXELFORMAT_RG16F:
        case SG_PIXELFORMAT_RGBA16F:
            return GL_HALF_FLOAT;
        case SG_PIXELFORMAT_R32UI:
        case SG_PIXELFORMAT_RG32UI:
        case SG_PIXELFORMAT_RGBA32UI:
            return GL_UNSIGNED_INT;
        case SG_PIXELFORMAT_R32SI:
        case SG_PIXELFORMAT_RG32SI:
        case SG_PIXELFORMAT_RGBA32SI:
            return GL_INT;
        case SG_PIXELFORMAT_R32F:
        case SG_PIXELFORMAT_RG32F:
        case SG_PIXELFORMAT_RGBA32F:
            return GL_FLOAT;
        case SG_PIXELFORMAT_RGB10A2:
            return GL_UNSIGNED_INT_2_10_10_10_REV;
        case SG_PIXELFORMAT_RG11B10F:
            return GL_UNSIGNED_INT_10F_11F_11F_REV;
        case SG_PIXELFORMAT_RGB9E5:
            return GL_UNSIGNED_INT_5_9_9_9_REV;
        case SG_PIXELFORMAT_DEPTH:
            return GL_FLOAT;
        case SG_PIXELFORMAT_DEPTH_STENCIL:
            return GL_UNSIGNED_INT_24_8;
        default:
            SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_teximage_format(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:
        case SG_PIXELFORMAT_R8SN:
        case SG_PIXELFORMAT_R16:
        case SG_PIXELFORMAT_R16SN:
        case SG_PIXELFORMAT_R16F:
        case SG_PIXELFORMAT_R32F:
            return GL_RED;
        case SG_PIXELFORMAT_R8UI:
        case SG_PIXELFORMAT_R8SI:
        case SG_PIXELFORMAT_R16UI:
        case SG_PIXELFORMAT_R16SI:
        case SG_PIXELFORMAT_R32UI:
        case SG_PIXELFORMAT_R32SI:
            return GL_RED_INTEGER;
        case SG_PIXELFORMAT_RG8:
        case SG_PIXELFORMAT_RG8SN:
        case SG_PIXELFORMAT_RG16:
        case SG_PIXELFORMAT_RG16SN:
        case SG_PIXELFORMAT_RG16F:
        case SG_PIXELFORMAT_RG32F:
            return GL_RG;
        case SG_PIXELFORMAT_RG8UI:
        case SG_PIXELFORMAT_RG8SI:
        case SG_PIXELFORMAT_RG16UI:
        case SG_PIXELFORMAT_RG16SI:
        case SG_PIXELFORMAT_RG32UI:
        case SG_PIXELFORMAT_RG32SI:
            return GL_RG_INTEGER;
        case SG_PIXELFORMAT_RGBA8:
        case SG_PIXELFORMAT_SRGB8A8:
        case SG_PIXELFORMAT_RGBA8SN:
        case SG_PIXELFORMAT_RGBA16:
        case SG_PIXELFORMAT_RGBA16SN:
        case SG_PIXELFORMAT_RGBA16F:
        case SG_PIXELFORMAT_RGBA32F:
        case SG_PIXELFORMAT_RGB10A2:
            return GL_RGBA;
        case SG_PIXELFORMAT_RGBA8UI:
        case SG_PIXELFORMAT_RGBA8SI:
        case SG_PIXELFORMAT_RGBA16UI:
        case SG_PIXELFORMAT_RGBA16SI:
        case SG_PIXELFORMAT_RGBA32UI:
        case SG_PIXELFORMAT_RGBA32SI:
            return GL_RGBA_INTEGER;
        case SG_PIXELFORMAT_RG11B10F:
        case SG_PIXELFORMAT_RGB9E5:
            return GL_RGB;
        case SG_PIXELFORMAT_DEPTH:
            return GL_DEPTH_COMPONENT;
        case SG_PIXELFORMAT_DEPTH_STENCIL:
            return GL_DEPTH_STENCIL;
        case SG_PIXELFORMAT_BC1_RGBA:
            return GL_COMPRESSED_RGBA_S3TC_DXT1_EXT;
        case SG_PIXELFORMAT_BC2_RGBA:
            return GL_COMPRESSED_RGBA_S3TC_DXT3_EXT;
        case SG_PIXELFORMAT_BC3_RGBA:
            return GL_COMPRESSED_RGBA_S3TC_DXT5_EXT;
        case SG_PIXELFORMAT_BC3_SRGBA:
            return GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT;
        case SG_PIXELFORMAT_BC4_R:
            return GL_COMPRESSED_RED_RGTC1;
        case SG_PIXELFORMAT_BC4_RSN:
            return GL_COMPRESSED_SIGNED_RED_RGTC1;
        case SG_PIXELFORMAT_BC5_RG:
            return GL_COMPRESSED_RED_GREEN_RGTC2;
        case SG_PIXELFORMAT_BC5_RGSN:
            return GL_COMPRESSED_SIGNED_RED_GREEN_RGTC2;
        case SG_PIXELFORMAT_BC6H_RGBF:
            return GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB;
        case SG_PIXELFORMAT_BC6H_RGBUF:
            return GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB;
        case SG_PIXELFORMAT_BC7_RGBA:
            return GL_COMPRESSED_RGBA_BPTC_UNORM_ARB;
        case SG_PIXELFORMAT_BC7_SRGBA:
            return GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB;
        case SG_PIXELFORMAT_ETC2_RGB8:
            return GL_COMPRESSED_RGB8_ETC2;
        case SG_PIXELFORMAT_ETC2_SRGB8:
            return GL_COMPRESSED_SRGB8_ETC2;
        case SG_PIXELFORMAT_ETC2_RGB8A1:
            return GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2;
        case SG_PIXELFORMAT_ETC2_RGBA8:
            return GL_COMPRESSED_RGBA8_ETC2_EAC;
        case SG_PIXELFORMAT_ETC2_SRGB8A8:
            return GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC;
        case SG_PIXELFORMAT_EAC_R11:
            return GL_COMPRESSED_R11_EAC;
        case SG_PIXELFORMAT_EAC_R11SN:
            return GL_COMPRESSED_SIGNED_R11_EAC;
        case SG_PIXELFORMAT_EAC_RG11:
            return GL_COMPRESSED_RG11_EAC;
        case SG_PIXELFORMAT_EAC_RG11SN:
            return GL_COMPRESSED_SIGNED_RG11_EAC;
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:
            return GL_COMPRESSED_RGBA_ASTC_4x4_KHR;
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:
            return GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR;
        default:
            SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_teximage_internal_format(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:         return GL_R8;
        case SG_PIXELFORMAT_R8SN:       return GL_R8_SNORM;
        case SG_PIXELFORMAT_R8UI:       return GL_R8UI;
        case SG_PIXELFORMAT_R8SI:       return GL_R8I;
        #if !defined(SOKOL_GLES3)
            case SG_PIXELFORMAT_R16:        return GL_R16;
            case SG_PIXELFORMAT_R16SN:      return GL_R16_SNORM;
        #endif
        case SG_PIXELFORMAT_R16UI:      return GL_R16UI;
        case SG_PIXELFORMAT_R16SI:      return GL_R16I;
        case SG_PIXELFORMAT_R16F:       return GL_R16F;
        case SG_PIXELFORMAT_RG8:        return GL_RG8;
        case SG_PIXELFORMAT_RG8SN:      return GL_RG8_SNORM;
        case SG_PIXELFORMAT_RG8UI:      return GL_RG8UI;
        case SG_PIXELFORMAT_RG8SI:      return GL_RG8I;
        case SG_PIXELFORMAT_R32UI:      return GL_R32UI;
        case SG_PIXELFORMAT_R32SI:      return GL_R32I;
        case SG_PIXELFORMAT_R32F:       return GL_R32F;
        #if !defined(SOKOL_GLES3)
            case SG_PIXELFORMAT_RG16:       return GL_RG16;
            case SG_PIXELFORMAT_RG16SN:     return GL_RG16_SNORM;
        #endif
        case SG_PIXELFORMAT_RG16UI:     return GL_RG16UI;
        case SG_PIXELFORMAT_RG16SI:     return GL_RG16I;
        case SG_PIXELFORMAT_RG16F:      return GL_RG16F;
        case SG_PIXELFORMAT_RGBA8:      return GL_RGBA8;
        case SG_PIXELFORMAT_SRGB8A8:    return GL_SRGB8_ALPHA8;
        case SG_PIXELFORMAT_RGBA8SN:    return GL_RGBA8_SNORM;
        case SG_PIXELFORMAT_RGBA8UI:    return GL_RGBA8UI;
        case SG_PIXELFORMAT_RGBA8SI:    return GL_RGBA8I;
        case SG_PIXELFORMAT_RGB10A2:    return GL_RGB10_A2;
        case SG_PIXELFORMAT_RG11B10F:   return GL_R11F_G11F_B10F;
        case SG_PIXELFORMAT_RGB9E5:     return GL_RGB9_E5;
        case SG_PIXELFORMAT_RG32UI:     return GL_RG32UI;
        case SG_PIXELFORMAT_RG32SI:     return GL_RG32I;
        case SG_PIXELFORMAT_RG32F:      return GL_RG32F;
        #if !defined(SOKOL_GLES3)
            case SG_PIXELFORMAT_RGBA16:     return GL_RGBA16;
            case SG_PIXELFORMAT_RGBA16SN:   return GL_RGBA16_SNORM;
        #endif
        case SG_PIXELFORMAT_RGBA16UI:   return GL_RGBA16UI;
        case SG_PIXELFORMAT_RGBA16SI:   return GL_RGBA16I;
        case SG_PIXELFORMAT_RGBA16F:    return GL_RGBA16F;
        case SG_PIXELFORMAT_RGBA32UI:   return GL_RGBA32UI;
        case SG_PIXELFORMAT_RGBA32SI:   return GL_RGBA32I;
        case SG_PIXELFORMAT_RGBA32F:    return GL_RGBA32F;
        case SG_PIXELFORMAT_DEPTH:      return GL_DEPTH_COMPONENT32F;
        case SG_PIXELFORMAT_DEPTH_STENCIL:      return GL_DEPTH24_STENCIL8;
        case SG_PIXELFORMAT_BC1_RGBA:           return GL_COMPRESSED_RGBA_S3TC_DXT1_EXT;
        case SG_PIXELFORMAT_BC2_RGBA:           return GL_COMPRESSED_RGBA_S3TC_DXT3_EXT;
        case SG_PIXELFORMAT_BC3_RGBA:           return GL_COMPRESSED_RGBA_S3TC_DXT5_EXT;
        case SG_PIXELFORMAT_BC3_SRGBA:          return GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT;
        case SG_PIXELFORMAT_BC4_R:              return GL_COMPRESSED_RED_RGTC1;
        case SG_PIXELFORMAT_BC4_RSN:            return GL_COMPRESSED_SIGNED_RED_RGTC1;
        case SG_PIXELFORMAT_BC5_RG:             return GL_COMPRESSED_RED_GREEN_RGTC2;
        case SG_PIXELFORMAT_BC5_RGSN:           return GL_COMPRESSED_SIGNED_RED_GREEN_RGTC2;
        case SG_PIXELFORMAT_BC6H_RGBF:          return GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB;
        case SG_PIXELFORMAT_BC6H_RGBUF:         return GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB;
        case SG_PIXELFORMAT_BC7_RGBA:           return GL_COMPRESSED_RGBA_BPTC_UNORM_ARB;
        case SG_PIXELFORMAT_BC7_SRGBA:          return GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB;
        case SG_PIXELFORMAT_ETC2_RGB8:          return GL_COMPRESSED_RGB8_ETC2;
        case SG_PIXELFORMAT_ETC2_SRGB8:         return GL_COMPRESSED_SRGB8_ETC2;
        case SG_PIXELFORMAT_ETC2_RGB8A1:        return GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2;
        case SG_PIXELFORMAT_ETC2_RGBA8:         return GL_COMPRESSED_RGBA8_ETC2_EAC;
        case SG_PIXELFORMAT_ETC2_SRGB8A8:       return GL_COMPRESSED_SRGB8_ALPHA8_ETC2_EAC;
        case SG_PIXELFORMAT_EAC_R11:            return GL_COMPRESSED_R11_EAC;
        case SG_PIXELFORMAT_EAC_R11SN:          return GL_COMPRESSED_SIGNED_R11_EAC;
        case SG_PIXELFORMAT_EAC_RG11:           return GL_COMPRESSED_RG11_EAC;
        case SG_PIXELFORMAT_EAC_RG11SN:         return GL_COMPRESSED_SIGNED_RG11_EAC;
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:      return GL_COMPRESSED_RGBA_ASTC_4x4_KHR;
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:     return GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_cubeface_target(int face_index) {
    switch (face_index) {
        case 0: return GL_TEXTURE_CUBE_MAP_POSITIVE_X;
        case 1: return GL_TEXTURE_CUBE_MAP_NEGATIVE_X;
        case 2: return GL_TEXTURE_CUBE_MAP_POSITIVE_Y;
        case 3: return GL_TEXTURE_CUBE_MAP_NEGATIVE_Y;
        case 4: return GL_TEXTURE_CUBE_MAP_POSITIVE_Z;
        case 5: return GL_TEXTURE_CUBE_MAP_NEGATIVE_Z;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

// see: https://www.khronos.org/registry/OpenGL-Refpages/es3.0/html/glTexImage2D.xhtml
_SOKOL_PRIVATE void _sg_gl_init_pixelformats(bool has_bgra) {
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_R8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R8SI]);
    #if !defined(SOKOL_GLES3)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R16SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG8SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    #if !defined(SOKOL_GLES3)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG16SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_SRGB8A8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    if (has_bgra) {
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_BGRA8]);
    }
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGB10A2]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGB9E5]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    #if !defined(SOKOL_GLES3)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH]);
    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH_STENCIL]);
}

// FIXME: OES_half_float_blend
_SOKOL_PRIVATE void _sg_gl_init_pixelformats_half_float(bool has_colorbuffer_half_float) {
    if (has_colorbuffer_half_float) {
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16F]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16F]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    } else {
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_R16F]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG16F]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    }
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_float(bool has_colorbuffer_float, bool has_texture_float_linear, bool has_float_blend) {
    if (has_texture_float_linear) {
        if (has_colorbuffer_float) {
            if (has_float_blend) {
                _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R32F]);
                _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG32F]);
                _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
            } else {
                _sg_pixelformat_sfrm(&_sg.formats[SG_PIXELFORMAT_R32F]);
                _sg_pixelformat_sfrm(&_sg.formats[SG_PIXELFORMAT_RG32F]);
                _sg_pixelformat_sfrm(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
            }
            _sg_pixelformat_sfrm(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);
        } else {
            _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_R32F]);
            _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG32F]);
            _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
            _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);
        }
    } else {
        if (has_colorbuffer_float) {
            _sg_pixelformat_sbrm(&_sg.formats[SG_PIXELFORMAT_R32F]);
            _sg_pixelformat_sbrm(&_sg.formats[SG_PIXELFORMAT_RG32F]);
            _sg_pixelformat_sbrm(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
            _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);
        } else {
            _sg_pixelformat_s(&_sg.formats[SG_PIXELFORMAT_R32F]);
            _sg_pixelformat_s(&_sg.formats[SG_PIXELFORMAT_RG32F]);
            _sg_pixelformat_s(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
            _sg_pixelformat_s(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);
        }
    }
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_s3tc(void) {
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC1_RGBA]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC2_RGBA]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_RGBA]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_SRGBA]);
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_rgtc(void) {
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_R]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_RSN]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RG]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RGSN]);
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_bptc(void) {
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBF]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBUF]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_RGBA]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_SRGBA]);
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_etc2(void) {
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8A1]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGBA8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8A8]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11SN]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11SN]);
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_astc(void) {
     _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_RGBA]);
     _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_SRGBA]);
}

_SOKOL_PRIVATE void _sg_gl_init_pixelformats_compute(void) {
    // using Vulkan's conservative default caps (see: https://github.com/gpuweb/gpuweb/issues/513)
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
}

_SOKOL_PRIVATE void _sg_gl_init_limits(void) {
    _SG_GL_CHECK_ERROR();
    GLint gl_int;
    glGetIntegerv(GL_MAX_TEXTURE_SIZE, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.max_image_size_2d = gl_int;
    _sg.limits.max_image_size_array = gl_int;
    glGetIntegerv(GL_MAX_CUBE_MAP_TEXTURE_SIZE, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.max_image_size_cube = gl_int;
    glGetIntegerv(GL_MAX_VERTEX_ATTRIBS, &gl_int);
    _SG_GL_CHECK_ERROR();
    if (gl_int > SG_MAX_VERTEX_ATTRIBUTES) {
        gl_int = SG_MAX_VERTEX_ATTRIBUTES;
    }
    _sg.limits.max_vertex_attrs = gl_int;
    glGetIntegerv(GL_MAX_VERTEX_UNIFORM_COMPONENTS, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.gl_max_vertex_uniform_components = gl_int;
    glGetIntegerv(GL_MAX_3D_TEXTURE_SIZE, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.max_image_size_3d = gl_int;
    glGetIntegerv(GL_MAX_ARRAY_TEXTURE_LAYERS, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.max_image_array_layers = gl_int;
    if (_sg.gl.ext_anisotropic) {
        glGetIntegerv(GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT, &gl_int);
        _SG_GL_CHECK_ERROR();
        _sg.gl.max_anisotropy = gl_int;
    } else {
        _sg.gl.max_anisotropy = 1;
    }
    glGetIntegerv(GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS, &gl_int);
    _SG_GL_CHECK_ERROR();
    _sg.limits.gl_max_combined_texture_image_units = gl_int;
}

#if defined(SOKOL_GLCORE)
_SOKOL_PRIVATE void _sg_gl_init_caps_glcore(void) {
    _sg.backend = SG_BACKEND_GLCORE;

    GLint major_version = 0;
    GLint minor_version = 0;
    glGetIntegerv(GL_MAJOR_VERSION, &major_version);
    glGetIntegerv(GL_MINOR_VERSION, &minor_version);
    const int version = major_version * 100 + minor_version * 10;
    _sg.features.origin_top_left = false;
    _sg.features.image_clamp_to_border = true;
    _sg.features.mrt_independent_blend_state = false;
    _sg.features.mrt_independent_write_mask = true;
    _sg.features.compute = version >= 430;
    #if defined(__APPLE__)
    _sg.features.msaa_image_bindings = false;
    #else
    _sg.features.msaa_image_bindings = true;
    #endif

    // scan extensions
    bool has_s3tc = false;  // BC1..BC3
    bool has_rgtc = false;  // BC4 and BC5
    bool has_bptc = false;  // BC6H and BC7
    bool has_etc2 = false;
    bool has_astc = false;
    GLint num_ext = 0;
    glGetIntegerv(GL_NUM_EXTENSIONS, &num_ext);
    for (int i = 0; i < num_ext; i++) {
        const char* ext = (const char*) glGetStringi(GL_EXTENSIONS, (GLuint)i);
        if (ext) {
            if (strstr(ext, "_texture_compression_s3tc")) {
                has_s3tc = true;
            } else if (strstr(ext, "_texture_compression_rgtc")) {
                has_rgtc = true;
            } else if (strstr(ext, "_texture_compression_bptc")) {
                has_bptc = true;
            } else if (strstr(ext, "_ES3_compatibility")) {
                has_etc2 = true;
            } else if (strstr(ext, "_texture_filter_anisotropic")) {
                _sg.gl.ext_anisotropic = true;
            } else if (strstr(ext, "_texture_compression_astc_ldr")) {
                has_astc = true;
            }
        }
    }

    // limits
    _sg_gl_init_limits();

    // pixel formats
    const bool has_bgra = false;    // not a bug
    const bool has_colorbuffer_float = true;
    const bool has_colorbuffer_half_float = true;
    const bool has_texture_float_linear = true; // FIXME???
    const bool has_float_blend = true;
    _sg_gl_init_pixelformats(has_bgra);
    _sg_gl_init_pixelformats_float(has_colorbuffer_float, has_texture_float_linear, has_float_blend);
    _sg_gl_init_pixelformats_half_float(has_colorbuffer_half_float);
    if (has_s3tc) {
        _sg_gl_init_pixelformats_s3tc();
    }
    if (has_rgtc) {
        _sg_gl_init_pixelformats_rgtc();
    }
    if (has_bptc) {
        _sg_gl_init_pixelformats_bptc();
    }
    if (has_etc2) {
        _sg_gl_init_pixelformats_etc2();
    }
    if (has_astc) {
        _sg_gl_init_pixelformats_astc();
    }
    if (_sg.features.compute) {
        _sg_gl_init_pixelformats_compute();
    }
}
#endif

#if defined(SOKOL_GLES3)
_SOKOL_PRIVATE void _sg_gl_init_caps_gles3(void) {
    _sg.backend = SG_BACKEND_GLES3;

    GLint major_version = 0;
    GLint minor_version = 0;
    glGetIntegerv(GL_MAJOR_VERSION, &major_version);
    glGetIntegerv(GL_MINOR_VERSION, &minor_version);
    const int version = major_version * 100 + minor_version * 10;
    _sg.features.origin_top_left = false;
    _sg.features.image_clamp_to_border = false;
    _sg.features.mrt_independent_blend_state = false;
    _sg.features.mrt_independent_write_mask = false;
    _sg.features.compute = version >= 310;
    _sg.features.msaa_image_bindings = false;
    #if defined(__EMSCRIPTEN__)
    _sg.features.separate_buffer_types = true;
    #else
    _sg.features.separate_buffer_types = false;
    #endif

    bool has_s3tc = false;  // BC1..BC3
    bool has_rgtc = false;  // BC4 and BC5
    bool has_bptc = false;  // BC6H and BC7
    #if defined(__EMSCRIPTEN__)
        bool has_etc2 = false;
    #else
        bool has_etc2 = true;
    #endif
    bool has_astc = false;
    bool has_colorbuffer_float = false;
    bool has_colorbuffer_half_float = false;
    bool has_texture_float_linear = false;
    bool has_float_blend = false;
    GLint num_ext = 0;
    glGetIntegerv(GL_NUM_EXTENSIONS, &num_ext);
    for (int i = 0; i < num_ext; i++) {
        const char* ext = (const char*) glGetStringi(GL_EXTENSIONS, (GLuint)i);
        if (ext) {
            if (strstr(ext, "_texture_compression_s3tc")) {
                has_s3tc = true;
            } else if (strstr(ext, "_compressed_texture_s3tc")) {
                has_s3tc = true;
            } else if (strstr(ext, "_texture_compression_rgtc")) {
                has_rgtc = true;
            } else if (strstr(ext, "_texture_compression_bptc")) {
                has_bptc = true;
            } else if (strstr(ext, "_compressed_texture_etc")) {
                has_etc2 = true;
            } else if (strstr(ext, "_compressed_texture_astc")) {
                has_astc = true;
            } else if (strstr(ext, "_color_buffer_float")) {
                has_colorbuffer_float = true;
            } else if (strstr(ext, "_color_buffer_half_float")) {
                has_colorbuffer_half_float = true;
            } else if (strstr(ext, "_texture_float_linear")) {
                has_texture_float_linear = true;
            } else if (strstr(ext, "_float_blend")) {
                has_float_blend = true;
            } else if (strstr(ext, "_texture_filter_anisotropic")) {
                _sg.gl.ext_anisotropic = true;
            }
        }
    }

    /* on WebGL2, color_buffer_float also includes 16-bit formats
       see: https://developer.mozilla.org/en-US/docs/Web/API/EXT_color_buffer_float
    */
    #if defined(__EMSCRIPTEN__)
    if (!has_colorbuffer_half_float && has_colorbuffer_float) {
        has_colorbuffer_half_float = has_colorbuffer_float;
    }
    #endif

    // limits
    _sg_gl_init_limits();

    // pixel formats
    const bool has_bgra = false;    // not a bug
    _sg_gl_init_pixelformats(has_bgra);
    _sg_gl_init_pixelformats_float(has_colorbuffer_float, has_texture_float_linear, has_float_blend);
    _sg_gl_init_pixelformats_half_float(has_colorbuffer_half_float);
    if (has_s3tc) {
        _sg_gl_init_pixelformats_s3tc();
    }
    if (has_rgtc) {
        _sg_gl_init_pixelformats_rgtc();
    }
    if (has_bptc) {
        _sg_gl_init_pixelformats_bptc();
    }
    if (has_etc2) {
        _sg_gl_init_pixelformats_etc2();
    }
    if (has_astc) {
        _sg_gl_init_pixelformats_astc();
    }
    if (_sg.features.compute) {
        _sg_gl_init_pixelformats_compute();
    }
}
#endif

//-- state cache implementation ------------------------------------------------
_SOKOL_PRIVATE void _sg_gl_cache_clear_buffer_bindings(bool force) {
    if (force || (_sg.gl.cache.vertex_buffer != 0)) {
        glBindBuffer(GL_ARRAY_BUFFER, 0);
        _sg.gl.cache.vertex_buffer = 0;
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    if (force || (_sg.gl.cache.index_buffer != 0)) {
        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
        _sg.gl.cache.index_buffer = 0;
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    if (force || (_sg.gl.cache.storage_buffer != 0)) {
        if (_sg.features.compute) {
            glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
        }
        _sg.gl.cache.storage_buffer = 0;
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    for (size_t i = 0; i < _SG_GL_MAX_SBUF_BINDINGS; i++) {
        if (force || (_sg.gl.cache.storage_buffers[i] != 0)) {
            if (_sg.features.compute) {
                glBindBufferBase(GL_SHADER_STORAGE_BUFFER, (GLuint)i, 0);
            }
            _sg.gl.cache.storage_buffers[i] = 0;
            _sg_stats_add(gl.num_bind_buffer, 1);
        }
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_bind_buffer(GLenum target, GLuint buffer) {
    SOKOL_ASSERT((GL_ARRAY_BUFFER == target) || (GL_ELEMENT_ARRAY_BUFFER == target) || (GL_SHADER_STORAGE_BUFFER == target));
    if (target == GL_ARRAY_BUFFER) {
        if (_sg.gl.cache.vertex_buffer != buffer) {
            _sg.gl.cache.vertex_buffer = buffer;
            glBindBuffer(target, buffer);
            _sg_stats_add(gl.num_bind_buffer, 1);
        }
    } else if (target == GL_ELEMENT_ARRAY_BUFFER) {
        if (_sg.gl.cache.index_buffer != buffer) {
            _sg.gl.cache.index_buffer = buffer;
            glBindBuffer(target, buffer);
            _sg_stats_add(gl.num_bind_buffer, 1);
        }
    } else if (target == GL_SHADER_STORAGE_BUFFER) {
        if (_sg.gl.cache.storage_buffer != buffer) {
            _sg.gl.cache.storage_buffer = buffer;
            if (_sg.features.compute) {
                glBindBuffer(target, buffer);
            }
            _sg_stats_add(gl.num_bind_buffer, 1);
        }
    } else {
        SOKOL_UNREACHABLE;
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_bind_storage_buffer(uint8_t glsl_binding_n, GLuint buffer) {
    SOKOL_ASSERT(glsl_binding_n < _SG_GL_MAX_SBUF_BINDINGS);
    if (_sg.gl.cache.storage_buffers[glsl_binding_n] != buffer) {
        _sg.gl.cache.storage_buffers[glsl_binding_n] = buffer;
        _sg.gl.cache.storage_buffer = buffer; // not a bug
        if (_sg.features.compute) {
            glBindBufferBase(GL_SHADER_STORAGE_BUFFER, glsl_binding_n, buffer);
        }
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_store_buffer_binding(GLenum target) {
    if (target == GL_ARRAY_BUFFER) {
        _sg.gl.cache.stored_vertex_buffer = _sg.gl.cache.vertex_buffer;
    } else if (target == GL_ELEMENT_ARRAY_BUFFER) {
        _sg.gl.cache.stored_index_buffer = _sg.gl.cache.index_buffer;
    } else if (target == GL_SHADER_STORAGE_BUFFER) {
        _sg.gl.cache.stored_storage_buffer = _sg.gl.cache.storage_buffer;
    } else {
        SOKOL_UNREACHABLE;
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_restore_buffer_binding(GLenum target) {
    if (target == GL_ARRAY_BUFFER) {
        if (_sg.gl.cache.stored_vertex_buffer != 0) {
            // we only care about restoring valid ids
            _sg_gl_cache_bind_buffer(target, _sg.gl.cache.stored_vertex_buffer);
            _sg.gl.cache.stored_vertex_buffer = 0;
        }
    } else if (target == GL_ELEMENT_ARRAY_BUFFER) {
        if (_sg.gl.cache.stored_index_buffer != 0) {
            // we only care about restoring valid ids
            _sg_gl_cache_bind_buffer(target, _sg.gl.cache.stored_index_buffer);
            _sg.gl.cache.stored_index_buffer = 0;
        }
    } else if (target == GL_SHADER_STORAGE_BUFFER) {
        if (_sg.gl.cache.stored_storage_buffer != 0) {
            // we only care about restoring valid ids
            _sg_gl_cache_bind_buffer(target, _sg.gl.cache.stored_storage_buffer);
            _sg.gl.cache.stored_storage_buffer = 0;
        }
    } else {
        SOKOL_UNREACHABLE;
    }
}

// called from _sg_gl_discard_buffer()
_SOKOL_PRIVATE void _sg_gl_cache_invalidate_buffer(GLuint buf) {
    if (buf == _sg.gl.cache.vertex_buffer) {
        _sg.gl.cache.vertex_buffer = 0;
        glBindBuffer(GL_ARRAY_BUFFER, 0);
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    if (buf == _sg.gl.cache.index_buffer) {
        _sg.gl.cache.index_buffer = 0;
        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    if (buf == _sg.gl.cache.storage_buffer) {
        _sg.gl.cache.storage_buffer = 0;
        glBindBuffer(GL_SHADER_STORAGE_BUFFER, 0);
        _sg_stats_add(gl.num_bind_buffer, 1);
    }
    for (size_t i = 0; i < _SG_GL_MAX_SBUF_BINDINGS; i++) {
        if (buf == _sg.gl.cache.storage_buffers[i]) {
            _sg.gl.cache.storage_buffers[i] = 0;
            _sg.gl.cache.storage_buffer = 0; // not a bug!
            glBindBufferBase(GL_SHADER_STORAGE_BUFFER, (GLuint)i, 0);
            _sg_stats_add(gl.num_bind_buffer, 1);
        }
    }
    if (buf == _sg.gl.cache.stored_vertex_buffer) {
        _sg.gl.cache.stored_vertex_buffer = 0;
    }
    if (buf == _sg.gl.cache.stored_index_buffer) {
        _sg.gl.cache.stored_index_buffer = 0;
    }
    if (buf == _sg.gl.cache.stored_storage_buffer) {
        _sg.gl.cache.stored_storage_buffer = 0;
    }
    for (int i = 0; i < SG_MAX_VERTEX_ATTRIBUTES; i++) {
        if (buf == _sg.gl.cache.attrs[i].gl_vbuf) {
            _sg.gl.cache.attrs[i].gl_vbuf = 0;
        }
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_active_texture(GLenum texture) {
    _SG_GL_CHECK_ERROR();
    if (_sg.gl.cache.cur_active_texture != texture) {
        _sg.gl.cache.cur_active_texture = texture;
        glActiveTexture(texture);
        _sg_stats_add(gl.num_active_texture, 1);
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_cache_clear_texture_sampler_bindings(bool force) {
    _SG_GL_CHECK_ERROR();
    for (int i = 0; (i < _SG_GL_MAX_IMG_SMP_BINDINGS) && (i < _sg.limits.gl_max_combined_texture_image_units); i++) {
        if (force || (_sg.gl.cache.texture_samplers[i].texture != 0)) {
            GLenum gl_texture_unit = (GLenum) (GL_TEXTURE0 + i);
            glActiveTexture(gl_texture_unit);
            _sg_stats_add(gl.num_active_texture, 1);
            glBindTexture(GL_TEXTURE_2D, 0);
            glBindTexture(GL_TEXTURE_CUBE_MAP, 0);
            glBindTexture(GL_TEXTURE_3D, 0);
            glBindTexture(GL_TEXTURE_2D_ARRAY, 0);
            _sg_stats_add(gl.num_bind_texture, 4);
            glBindSampler((GLuint)i, 0);
            _sg_stats_add(gl.num_bind_sampler, 1);
            _sg.gl.cache.texture_samplers[i].target = 0;
            _sg.gl.cache.texture_samplers[i].texture = 0;
            _sg.gl.cache.texture_samplers[i].sampler = 0;
            _sg.gl.cache.cur_active_texture = gl_texture_unit;
        }
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_cache_bind_texture_sampler(int8_t gl_tex_slot, GLenum target, GLuint texture, GLuint sampler) {
    /* it's valid to call this function with target=0 and/or texture=0
       target=0 will unbind the previous binding, texture=0 will clear
       the new binding
    */
    SOKOL_ASSERT((gl_tex_slot >= 0) && (gl_tex_slot < _SG_GL_MAX_IMG_SMP_BINDINGS));
    if (gl_tex_slot >= _sg.limits.gl_max_combined_texture_image_units) {
        return;
    }
    _SG_GL_CHECK_ERROR();
    _sg_gl_cache_texture_sampler_bind_slot* slot = &_sg.gl.cache.texture_samplers[gl_tex_slot];
    if ((slot->target != target) || (slot->texture != texture) || (slot->sampler != sampler)) {
        _sg_gl_cache_active_texture((GLenum)(GL_TEXTURE0 + gl_tex_slot));
        // if the target has changed, clear the previous binding on that target
        if ((target != slot->target) && (slot->target != 0)) {
            glBindTexture(slot->target, 0);
            _SG_GL_CHECK_ERROR();
            _sg_stats_add(gl.num_bind_texture, 1);
        }
        // apply new binding (can be 0 to unbind)
        if (target != 0) {
            glBindTexture(target, texture);
            _SG_GL_CHECK_ERROR();
            _sg_stats_add(gl.num_bind_texture, 1);
        }
        // apply new sampler (can be 0 to unbind)
        glBindSampler((GLuint)gl_tex_slot, sampler);
        _SG_GL_CHECK_ERROR();
        _sg_stats_add(gl.num_bind_sampler, 1);

        slot->target = target;
        slot->texture = texture;
        slot->sampler = sampler;
    }
}

_SOKOL_PRIVATE void _sg_gl_cache_store_texture_sampler_binding(int8_t gl_tex_slot) {
    SOKOL_ASSERT((gl_tex_slot >= 0) && (gl_tex_slot < _SG_GL_MAX_IMG_SMP_BINDINGS));
    _sg.gl.cache.stored_texture_sampler = _sg.gl.cache.texture_samplers[gl_tex_slot];
}

_SOKOL_PRIVATE void _sg_gl_cache_restore_texture_sampler_binding(int8_t gl_tex_slot) {
    SOKOL_ASSERT((gl_tex_slot >= 0) && (gl_tex_slot < _SG_GL_MAX_IMG_SMP_BINDINGS));
    _sg_gl_cache_texture_sampler_bind_slot* slot = &_sg.gl.cache.stored_texture_sampler;
    if (slot->texture != 0) {
        // we only care about restoring valid ids
        SOKOL_ASSERT(slot->target != 0);
        _sg_gl_cache_bind_texture_sampler(gl_tex_slot, slot->target, slot->texture, slot->sampler);
        slot->target = 0;
        slot->texture = 0;
        slot->sampler = 0;
    }
}

// called from _sg_gl_discard_texture() and _sg_gl_discard_sampler()
_SOKOL_PRIVATE void _sg_gl_cache_invalidate_texture_sampler(GLuint tex, GLuint smp) {
    _SG_GL_CHECK_ERROR();
    for (size_t i = 0; i < _SG_GL_MAX_IMG_SMP_BINDINGS; i++) {
        _sg_gl_cache_texture_sampler_bind_slot* slot = &_sg.gl.cache.texture_samplers[i];
        if ((0 != slot->target) && ((tex == slot->texture) || (smp == slot->sampler))) {
            _sg_gl_cache_active_texture((GLenum)(GL_TEXTURE0 + i));
            glBindTexture(slot->target, 0);
            _SG_GL_CHECK_ERROR();
            _sg_stats_add(gl.num_bind_texture, 1);
            glBindSampler((GLuint)i, 0);
            _SG_GL_CHECK_ERROR();
            _sg_stats_add(gl.num_bind_sampler, 1);
            slot->target = 0;
            slot->texture = 0;
            slot->sampler = 0;
        }
    }
    if ((tex == _sg.gl.cache.stored_texture_sampler.texture) || (smp == _sg.gl.cache.stored_texture_sampler.sampler)) {
        _sg.gl.cache.stored_texture_sampler.target = 0;
        _sg.gl.cache.stored_texture_sampler.texture = 0;
        _sg.gl.cache.stored_texture_sampler.sampler = 0;
    }
}

// called from _sg_gl_discard_shader()
_SOKOL_PRIVATE void _sg_gl_cache_invalidate_program(GLuint prog) {
    if (prog == _sg.gl.cache.prog) {
        _sg.gl.cache.prog = 0;
        glUseProgram(0);
        _sg_stats_add(gl.num_use_program, 1);
    }
}

// called from _sg_gl_discard_pipeline()
_SOKOL_PRIVATE void _sg_gl_cache_invalidate_pipeline(_sg_pipeline_t* pip) {
    if (_sg_sref_eql(&_sg.gl.cache.cur_pip, &pip->slot)) {
        _sg.gl.cache.cur_pip = _sg_sref(0);
    }
}

_SOKOL_PRIVATE void _sg_gl_reset_state_cache(void) {
    _SG_GL_CHECK_ERROR();
    glBindVertexArray(_sg.gl.vao);
    _SG_GL_CHECK_ERROR();
    _sg_clear(&_sg.gl.cache, sizeof(_sg.gl.cache));
    _sg_gl_cache_clear_buffer_bindings(true);
    _SG_GL_CHECK_ERROR();
    _sg_gl_cache_clear_texture_sampler_bindings(true);
    _SG_GL_CHECK_ERROR();
    for (int i = 0; i < _sg.limits.max_vertex_attrs; i++) {
        _sg_gl_attr_t* attr = &_sg.gl.cache.attrs[i].gl_attr;
        attr->vb_index = -1;
        attr->divisor = -1;
        glDisableVertexAttribArray((GLuint)i);
        _SG_GL_CHECK_ERROR();
        _sg_stats_add(gl.num_disable_vertex_attrib_array, 1);
    }
    _sg.gl.cache.cur_primitive_type = GL_TRIANGLES;

    // shader program
    glGetIntegerv(GL_CURRENT_PROGRAM, (GLint*)&_sg.gl.cache.prog);
    _SG_GL_CHECK_ERROR();

    // depth and stencil state
    _sg.gl.cache.depth.compare = SG_COMPAREFUNC_ALWAYS;
    _sg.gl.cache.stencil.front.compare = SG_COMPAREFUNC_ALWAYS;
    _sg.gl.cache.stencil.front.fail_op = SG_STENCILOP_KEEP;
    _sg.gl.cache.stencil.front.depth_fail_op = SG_STENCILOP_KEEP;
    _sg.gl.cache.stencil.front.pass_op = SG_STENCILOP_KEEP;
    _sg.gl.cache.stencil.back.compare = SG_COMPAREFUNC_ALWAYS;
    _sg.gl.cache.stencil.back.fail_op = SG_STENCILOP_KEEP;
    _sg.gl.cache.stencil.back.depth_fail_op = SG_STENCILOP_KEEP;
    _sg.gl.cache.stencil.back.pass_op = SG_STENCILOP_KEEP;
    glEnable(GL_DEPTH_TEST);
    glDepthFunc(GL_ALWAYS);
    glDepthMask(GL_FALSE);
    glDisable(GL_STENCIL_TEST);
    glStencilFunc(GL_ALWAYS, 0, 0);
    glStencilOp(GL_KEEP, GL_KEEP, GL_KEEP);
    glStencilMask(0);
    _sg_stats_add(gl.num_render_state, 7);

    // blend state
    _sg.gl.cache.blend.src_factor_rgb = SG_BLENDFACTOR_ONE;
    _sg.gl.cache.blend.dst_factor_rgb = SG_BLENDFACTOR_ZERO;
    _sg.gl.cache.blend.op_rgb = SG_BLENDOP_ADD;
    _sg.gl.cache.blend.src_factor_alpha = SG_BLENDFACTOR_ONE;
    _sg.gl.cache.blend.dst_factor_alpha = SG_BLENDFACTOR_ZERO;
    _sg.gl.cache.blend.op_alpha = SG_BLENDOP_ADD;
    glDisable(GL_BLEND);
    glBlendFuncSeparate(GL_ONE, GL_ZERO, GL_ONE, GL_ZERO);
    glBlendEquationSeparate(GL_FUNC_ADD, GL_FUNC_ADD);
    glBlendColor(0.0f, 0.0f, 0.0f, 0.0f);
    _sg_stats_add(gl.num_render_state, 4);

    // standalone state
    for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        _sg.gl.cache.color_write_mask[i] = SG_COLORMASK_RGBA;
    }
    _sg.gl.cache.cull_mode = SG_CULLMODE_NONE;
    _sg.gl.cache.face_winding = SG_FACEWINDING_CW;
    _sg.gl.cache.sample_count = 1;
    glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
    glPolygonOffset(0.0f, 0.0f);
    glDisable(GL_POLYGON_OFFSET_FILL);
    glDisable(GL_CULL_FACE);
    glFrontFace(GL_CW);
    glCullFace(GL_BACK);
    glEnable(GL_SCISSOR_TEST);
    glDisable(GL_SAMPLE_ALPHA_TO_COVERAGE);
    glEnable(GL_DITHER);
    glDisable(GL_POLYGON_OFFSET_FILL);
    _sg_stats_add(gl.num_render_state, 10);
    #if defined(SOKOL_GLCORE)
        glEnable(GL_MULTISAMPLE);
        glEnable(GL_PROGRAM_POINT_SIZE);
        _sg_stats_add(gl.num_render_state, 2);
    #endif
}

_SOKOL_PRIVATE void _sg_gl_setup_backend(const sg_desc* desc) {
    _SOKOL_UNUSED(desc);

    // assumes that _sg.gl is already zero-initialized
    _sg.gl.valid = true;

    #if defined(_SOKOL_USE_WIN32_GL_LOADER)
    _sg_gl_load_opengl();
    #endif

    // clear initial GL error state
    #if defined(SOKOL_DEBUG)
        while (glGetError() != GL_NO_ERROR);
    #endif
    #if defined(SOKOL_GLCORE)
        _sg_gl_init_caps_glcore();
    #elif defined(SOKOL_GLES3)
        _sg_gl_init_caps_gles3();
    #endif

    glGenVertexArrays(1, &_sg.gl.vao);
    glBindVertexArray(_sg.gl.vao);
    _SG_GL_CHECK_ERROR();
    // incoming texture data is generally expected to be packed tightly
    glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
    #if defined(SOKOL_GLCORE)
        // enable seamless cubemap sampling (only desktop GL)
        glEnable(GL_TEXTURE_CUBE_MAP_SEAMLESS);
    #endif
    _sg_gl_reset_state_cache();
}

_SOKOL_PRIVATE void _sg_gl_discard_backend(void) {
    SOKOL_ASSERT(_sg.gl.valid);
    if (_sg.gl.vao) {
        glDeleteVertexArrays(1, &_sg.gl.vao);
    }
    #if defined(_SOKOL_USE_WIN32_GL_LOADER)
    _sg_gl_unload_opengl();
    #endif
    _sg.gl.valid = false;
}

//-- GL backend resource creation and destruction ------------------------------
_SOKOL_PRIVATE sg_resource_state _sg_gl_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && desc);
    _SG_GL_CHECK_ERROR();
    buf->gl.injected = (0 != desc->gl_buffers[0]);
    const GLenum gl_target = _sg_gl_buffer_target(&buf->cmn.usage);
    const GLenum gl_usage  = _sg_gl_buffer_usage(&buf->cmn.usage);
    for (int slot = 0; slot < buf->cmn.num_slots; slot++) {
        GLuint gl_buf = 0;
        if (buf->gl.injected) {
            SOKOL_ASSERT(desc->gl_buffers[slot]);
            gl_buf = desc->gl_buffers[slot];
        } else {
            glGenBuffers(1, &gl_buf);
            SOKOL_ASSERT(gl_buf);
            _sg_gl_cache_store_buffer_binding(gl_target);
            _sg_gl_cache_bind_buffer(gl_target, gl_buf);
            glBufferData(gl_target, buf->cmn.size, 0, gl_usage);
            if (desc->data.ptr) {
                glBufferSubData(gl_target, 0, buf->cmn.size, desc->data.ptr);
            }
            _sg_gl_cache_restore_buffer_binding(gl_target);
        }
        buf->gl.buf[slot] = gl_buf;
    }
    _SG_GL_CHECK_ERROR();
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    _SG_GL_CHECK_ERROR();
    for (int slot = 0; slot < buf->cmn.num_slots; slot++) {
        if (buf->gl.buf[slot]) {
            _sg_gl_cache_invalidate_buffer(buf->gl.buf[slot]);
            if (!buf->gl.injected) {
                glDeleteBuffers(1, &buf->gl.buf[slot]);
            }
        }
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE bool _sg_gl_supported_texture_format(sg_pixel_format fmt) {
    const int fmt_index = (int) fmt;
    SOKOL_ASSERT((fmt_index > SG_PIXELFORMAT_NONE) && (fmt_index < _SG_PIXELFORMAT_NUM));
    return _sg.formats[fmt_index].sample;
}

_SOKOL_PRIVATE void _sg_gl_texstorage(const _sg_image_t* img) {
    const GLenum tgt = img->gl.target;
    const int num_mips = img->cmn.num_mipmaps;
    #if defined(_SOKOL_GL_HAS_TEXSTORAGE)
        const GLenum ifmt = _sg_gl_teximage_internal_format(img->cmn.pixel_format);
        const bool msaa = img->cmn.sample_count > 1;
        const int w = img->cmn.width;
        const int h = img->cmn.height;
        if ((SG_IMAGETYPE_2D == img->cmn.type) || (SG_IMAGETYPE_CUBE == img->cmn.type)) {
            #if defined(SOKOL_GLCORE)
                if (msaa) {
                    glTexStorage2DMultisample(tgt, img->cmn.sample_count, ifmt, w, h, GL_TRUE);
                } else {
                    glTexStorage2D(tgt, num_mips, ifmt, w, h);
                }
            #else
                SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                glTexStorage2D(tgt, num_mips, ifmt, w, h);
            #endif
        } else if ((SG_IMAGETYPE_3D == img->cmn.type) || (SG_IMAGETYPE_ARRAY == img->cmn.type)) {
            const int depth = img->cmn.num_slices;
            #if defined(SOKOL_GLCORE)
                if (msaa) {
                    // NOTE: MSAA works only for array textures, not 3D textures
                    glTexStorage3DMultisample(tgt, img->cmn.sample_count, ifmt, w, h, depth, GL_TRUE);
                } else {
                    glTexStorage3D(tgt, num_mips, ifmt, w, h, depth);
                }
            #else
                SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                glTexStorage3D(tgt, num_mips, ifmt, w, h, depth);
            #endif
        }
    #else
        glTexParameteri(tgt, GL_TEXTURE_MAX_LEVEL, num_mips - 1);
    #endif
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_teximage(const _sg_image_t* img, GLenum tgt, int mip_index, int w, int h, int depth, const GLvoid* data_ptr, GLsizei data_size) {
    const bool compressed = _sg_is_compressed_pixel_format(img->cmn.pixel_format);
    #if defined(_SOKOL_GL_HAS_TEXSTORAGE)
        if (data_ptr == 0) {
            return;
        }
        SOKOL_ASSERT(img->cmn.sample_count == 1);
        if ((SG_IMAGETYPE_2D == img->cmn.type) || (SG_IMAGETYPE_CUBE == img->cmn.type)) {
            if (compressed) {
                const GLenum ifmt = _sg_gl_teximage_internal_format(img->cmn.pixel_format);
                glCompressedTexSubImage2D(tgt, mip_index, 0, 0, w, h, ifmt, data_size, data_ptr);
            } else {
                const GLenum type = _sg_gl_teximage_type(img->cmn.pixel_format);
                const GLenum fmt = _sg_gl_teximage_format(img->cmn.pixel_format);
                glTexSubImage2D(tgt, mip_index, 0, 0, w, h, fmt, type, data_ptr);
            }
        } else if ((SG_IMAGETYPE_3D == img->cmn.type) || (SG_IMAGETYPE_ARRAY == img->cmn.type)) {
            if (compressed) {
                const GLenum ifmt = _sg_gl_teximage_internal_format(img->cmn.pixel_format);
                glCompressedTexSubImage3D(tgt, mip_index, 0, 0, 0, w, h, depth, ifmt, data_size, data_ptr);
            } else {
                const GLenum type = _sg_gl_teximage_type(img->cmn.pixel_format);
                const GLenum fmt = _sg_gl_teximage_format(img->cmn.pixel_format);
                glTexSubImage3D(tgt, mip_index, 0, 0, 0, w, h, depth, fmt, type, data_ptr);
            }
        }
    #else
        const GLenum ifmt = _sg_gl_teximage_internal_format(img->cmn.pixel_format);
        const bool msaa = img->cmn.sample_count > 1;
        if ((SG_IMAGETYPE_2D == img->cmn.type) || (SG_IMAGETYPE_CUBE == img->cmn.type)) {
            if (compressed) {
                SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                glCompressedTexImage2D(tgt, mip_index, ifmt, w, h, 0, data_size, data_ptr);
            } else {
                const GLenum type = _sg_gl_teximage_type(img->cmn.pixel_format);
                const GLenum fmt = _sg_gl_teximage_format(img->cmn.pixel_format);
                #if defined(SOKOL_GLCORE)
                    if (msaa) {
                        glTexImage2DMultisample(tgt, img->cmn.sample_count, (GLint)ifmt, w, h, GL_TRUE);
                    } else {
                        glTexImage2D(tgt, mip_index, (GLint)ifmt, w, h, 0, fmt, type, data_ptr);
                    }
                #else
                    SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                    glTexImage2D(tgt, mip_index, (GLint)ifmt, w, h, 0, fmt, type, data_ptr);
                #endif
            }
        } else if ((SG_IMAGETYPE_3D == img->cmn.type) || (SG_IMAGETYPE_ARRAY == img->cmn.type)) {
            if (compressed) {
                SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                glCompressedTexImage3D(tgt, mip_index, ifmt, w, h, depth, 0, data_size, data_ptr);
            } else {
                const GLenum type = _sg_gl_teximage_type(img->cmn.pixel_format);
                const GLenum fmt = _sg_gl_teximage_format(img->cmn.pixel_format);
                #if defined(SOKOL_GLCORE)
                    if (msaa) {
                        // NOTE: MSAA works only for array textures, not 3D textures
                        glTexImage3DMultisample(tgt, img->cmn.sample_count, (GLint)ifmt, w, h, depth, GL_TRUE);
                    } else {
                        glTexImage3D(tgt, mip_index, (GLint)ifmt, w, h, depth, 0, fmt, type, data_ptr);
                    }
                #else
                    SOKOL_ASSERT(!msaa); _SOKOL_UNUSED(msaa);
                    glTexImage3D(tgt, mip_index, (GLint)ifmt, w, h, depth, 0, fmt, type, data_ptr);
                #endif
            }
        }
    #endif
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE sg_resource_state _sg_gl_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && desc);
    _SG_GL_CHECK_ERROR();
    img->gl.injected = (0 != desc->gl_textures[0]);

    // check if texture format is support
    if (!_sg_gl_supported_texture_format(img->cmn.pixel_format)) {
        _SG_ERROR(GL_TEXTURE_FORMAT_NOT_SUPPORTED);
        return SG_RESOURCESTATE_FAILED;
    }

    // GLES3/WebGL2/macOS doesn't have support for multisampled textures, so create a render buffer object instead
    const bool msaa = img->cmn.sample_count > 1;
    if (!_sg.features.msaa_image_bindings && img->cmn.usage.render_attachment && msaa) {
        const GLenum gl_internal_format = _sg_gl_teximage_internal_format(img->cmn.pixel_format);
        glGenRenderbuffers(1, &img->gl.msaa_render_buffer);
        glBindRenderbuffer(GL_RENDERBUFFER, img->gl.msaa_render_buffer);
        glRenderbufferStorageMultisample(GL_RENDERBUFFER, img->cmn.sample_count, gl_internal_format, img->cmn.width, img->cmn.height);
    } else if (img->gl.injected) {
        img->gl.target = _sg_gl_texture_target(img->cmn.type, img->cmn.sample_count);
        // inject externally GL textures
        for (int slot = 0; slot < img->cmn.num_slots; slot++) {
            SOKOL_ASSERT(desc->gl_textures[slot]);
            img->gl.tex[slot] = desc->gl_textures[slot];
        }
        if (desc->gl_texture_target) {
            img->gl.target = (GLenum)desc->gl_texture_target;
        }
    } else {
        // create our own GL texture(s)
        img->gl.target = _sg_gl_texture_target(img->cmn.type, img->cmn.sample_count);
        for (int slot = 0; slot < img->cmn.num_slots; slot++) {
            glGenTextures(1, &img->gl.tex[slot]);
            SOKOL_ASSERT(img->gl.tex[slot]);
            _sg_gl_cache_store_texture_sampler_binding(0);
            _sg_gl_cache_bind_texture_sampler(0, img->gl.target, img->gl.tex[slot], 0);
            _sg_gl_texstorage(img);
            const int num_faces = img->cmn.type == SG_IMAGETYPE_CUBE ? 6 : 1;
            for (int face_index = 0; face_index < num_faces; face_index++) {
                for (int mip_index = 0; mip_index < img->cmn.num_mipmaps; mip_index++) {
                    GLenum gl_img_target = img->gl.target;
                    if (SG_IMAGETYPE_CUBE == img->cmn.type) {
                        gl_img_target = _sg_gl_cubeface_target(face_index);
                    }
                    const GLvoid* data_ptr = desc->data.subimage[face_index][mip_index].ptr;
                    const GLsizei data_size = (GLsizei)desc->data.subimage[face_index][mip_index].size;
                    const int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
                    const int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
                    const int mip_depth = (SG_IMAGETYPE_3D == img->cmn.type) ? _sg_miplevel_dim(img->cmn.num_slices, mip_index) : img->cmn.num_slices;
                    _sg_gl_teximage(img, gl_img_target, mip_index, mip_width, mip_height, mip_depth, data_ptr, data_size);
                }
            }
            _sg_gl_cache_restore_texture_sampler_binding(0);
        }
    }
    _SG_GL_CHECK_ERROR();
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_image(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    _SG_GL_CHECK_ERROR();
    for (int slot = 0; slot < img->cmn.num_slots; slot++) {
        if (img->gl.tex[slot]) {
            _sg_gl_cache_invalidate_texture_sampler(img->gl.tex[slot], 0);
            if (!img->gl.injected) {
                glDeleteTextures(1, &img->gl.tex[slot]);
            }
        }
    }
    if (img->gl.msaa_render_buffer) {
        glDeleteRenderbuffers(1, &img->gl.msaa_render_buffer);
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE sg_resource_state _sg_gl_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && desc);
    _SG_GL_CHECK_ERROR();
    smp->gl.injected = (0 != desc->gl_sampler);
    if (smp->gl.injected) {
        smp->gl.smp = (GLuint) desc->gl_sampler;
    } else {
        glGenSamplers(1, &smp->gl.smp);
        SOKOL_ASSERT(smp->gl.smp);

        const GLenum gl_min_filter = _sg_gl_min_filter(smp->cmn.min_filter, smp->cmn.mipmap_filter);
        const GLenum gl_mag_filter = _sg_gl_mag_filter(smp->cmn.mag_filter);
        glSamplerParameteri(smp->gl.smp, GL_TEXTURE_MIN_FILTER, (GLint)gl_min_filter);
        glSamplerParameteri(smp->gl.smp, GL_TEXTURE_MAG_FILTER, (GLint)gl_mag_filter);
        // GL spec has strange defaults for mipmap min/max lod: -1000 to +1000
        const float min_lod = _sg_clamp(desc->min_lod, 0.0f, 1000.0f);
        const float max_lod = _sg_clamp(desc->max_lod, 0.0f, 1000.0f);
        glSamplerParameterf(smp->gl.smp, GL_TEXTURE_MIN_LOD, min_lod);
        glSamplerParameterf(smp->gl.smp, GL_TEXTURE_MAX_LOD, max_lod);
        glSamplerParameteri(smp->gl.smp, GL_TEXTURE_WRAP_S, (GLint)_sg_gl_wrap(smp->cmn.wrap_u));
        glSamplerParameteri(smp->gl.smp, GL_TEXTURE_WRAP_T, (GLint)_sg_gl_wrap(smp->cmn.wrap_v));
        glSamplerParameteri(smp->gl.smp, GL_TEXTURE_WRAP_R, (GLint)_sg_gl_wrap(smp->cmn.wrap_w));
        #if defined(SOKOL_GLCORE)
        float border[4];
        switch (smp->cmn.border_color) {
            case SG_BORDERCOLOR_TRANSPARENT_BLACK:
                border[0] = 0.0f; border[1] = 0.0f; border[2] = 0.0f; border[3] = 0.0f;
                break;
            case SG_BORDERCOLOR_OPAQUE_WHITE:
                border[0] = 1.0f; border[1] = 1.0f; border[2] = 1.0f; border[3] = 1.0f;
                break;
            default:
                border[0] = 0.0f; border[1] = 0.0f; border[2] = 0.0f; border[3] = 1.0f;
                break;
        }
        glSamplerParameterfv(smp->gl.smp, GL_TEXTURE_BORDER_COLOR, border);
        #endif
        if (smp->cmn.compare != SG_COMPAREFUNC_NEVER) {
            glSamplerParameteri(smp->gl.smp, GL_TEXTURE_COMPARE_MODE, GL_COMPARE_REF_TO_TEXTURE);
            glSamplerParameteri(smp->gl.smp, GL_TEXTURE_COMPARE_FUNC, (GLint)_sg_gl_compare_func(smp->cmn.compare));
        } else {
            glSamplerParameteri(smp->gl.smp, GL_TEXTURE_COMPARE_MODE, GL_NONE);
        }
        if (_sg.gl.ext_anisotropic && (smp->cmn.max_anisotropy > 1)) {
            GLint max_aniso = (GLint) smp->cmn.max_anisotropy;
            if (max_aniso > _sg.gl.max_anisotropy) {
                max_aniso = _sg.gl.max_anisotropy;
            }
            glSamplerParameteri(smp->gl.smp, GL_TEXTURE_MAX_ANISOTROPY_EXT, max_aniso);
        }
    }
    _SG_GL_CHECK_ERROR();
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    _SG_GL_CHECK_ERROR();
    _sg_gl_cache_invalidate_texture_sampler(0, smp->gl.smp);
    if (!smp->gl.injected) {
        glDeleteSamplers(1, &smp->gl.smp);
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE GLuint _sg_gl_compile_shader(sg_shader_stage stage, const char* src) {
    SOKOL_ASSERT(src);
    _SG_GL_CHECK_ERROR();
    GLuint gl_shd = glCreateShader(_sg_gl_shader_stage(stage));
    glShaderSource(gl_shd, 1, &src, 0);
    glCompileShader(gl_shd);
    GLint compile_status = 0;
    glGetShaderiv(gl_shd, GL_COMPILE_STATUS, &compile_status);
    if (!compile_status) {
        // compilation failed, log error and delete shader
        GLint log_len = 0;
        glGetShaderiv(gl_shd, GL_INFO_LOG_LENGTH, &log_len);
        if (log_len > 0) {
            GLchar* log_buf = (GLchar*) _sg_malloc((size_t)log_len);
            glGetShaderInfoLog(gl_shd, log_len, &log_len, log_buf);
            _SG_ERROR(GL_SHADER_COMPILATION_FAILED);
            _SG_LOGMSG(GL_SHADER_COMPILATION_FAILED, log_buf);
            _sg_free(log_buf);
        }
        glDeleteShader(gl_shd);
        gl_shd = 0;
    }
    _SG_GL_CHECK_ERROR();
    return gl_shd;
}

// NOTE: this is an out-of-range check for GLSL bindslots that's also active in release mode
_SOKOL_PRIVATE bool _sg_gl_ensure_glsl_bindslot_ranges(const sg_shader_desc* desc) {
    SOKOL_ASSERT(desc);
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (desc->storage_buffers[i].glsl_binding_n >= _SG_GL_MAX_SBUF_BINDINGS) {
            _SG_ERROR(GL_STORAGEBUFFER_GLSL_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (desc->storage_images[i].glsl_binding_n >= _SG_GL_MAX_SIMG_BINDINGS) {
            _SG_ERROR(GL_STORAGEIMAGE_GLSL_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    return true;
}

_SOKOL_PRIVATE sg_resource_state _sg_gl_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && desc);
    SOKOL_ASSERT(!shd->gl.prog);
    _SG_GL_CHECK_ERROR();

    // perform a fatal range-check on GLSL bindslots that's also active
    // in release mode to avoid potential out-of-bounds array accesses
    if (!_sg_gl_ensure_glsl_bindslot_ranges(desc)) {
        return SG_RESOURCESTATE_FAILED;
    }

    // copy the optional vertex attribute names over
    for (int i = 0; i < SG_MAX_VERTEX_ATTRIBUTES; i++) {
        _sg_strcpy(&shd->gl.attrs[i].name, desc->attrs[i].glsl_name);
    }

    const bool has_vs = desc->vertex_func.source;
    const bool has_fs = desc->fragment_func.source;
    const bool has_cs = desc->compute_func.source;
    SOKOL_ASSERT((has_vs && has_fs) || has_cs);
    GLuint gl_prog = glCreateProgram();
    if (has_vs && has_fs) {
        GLuint gl_vs = _sg_gl_compile_shader(SG_SHADERSTAGE_VERTEX, desc->vertex_func.source);
        GLuint gl_fs = _sg_gl_compile_shader(SG_SHADERSTAGE_FRAGMENT, desc->fragment_func.source);
        if (!(gl_vs && gl_fs)) {
            glDeleteProgram(gl_prog);
            if (gl_vs) { glDeleteShader(gl_vs); }
            if (gl_fs) { glDeleteShader(gl_fs); }
            return SG_RESOURCESTATE_FAILED;
        }
        glAttachShader(gl_prog, gl_vs);
        glAttachShader(gl_prog, gl_fs);
        glLinkProgram(gl_prog);
        glDeleteShader(gl_vs);
        glDeleteShader(gl_fs);
        _SG_GL_CHECK_ERROR();
    } else if (has_cs) {
        GLuint gl_cs = _sg_gl_compile_shader(SG_SHADERSTAGE_COMPUTE, desc->compute_func.source);
        if (!gl_cs) {
            glDeleteProgram(gl_prog);
            return SG_RESOURCESTATE_FAILED;
        }
        glAttachShader(gl_prog, gl_cs);
        glLinkProgram(gl_prog);
        glDeleteShader(gl_cs);
        _SG_GL_CHECK_ERROR();
    } else {
        SOKOL_UNREACHABLE;
    }
    GLint link_status;
    glGetProgramiv(gl_prog, GL_LINK_STATUS, &link_status);
    if (!link_status) {
        GLint log_len = 0;
        glGetProgramiv(gl_prog, GL_INFO_LOG_LENGTH, &log_len);
        if (log_len > 0) {
            GLchar* log_buf = (GLchar*) _sg_malloc((size_t)log_len);
            glGetProgramInfoLog(gl_prog, log_len, &log_len, log_buf);
            _SG_ERROR(GL_SHADER_LINKING_FAILED);
            _SG_LOGMSG(GL_SHADER_LINKING_FAILED, log_buf);
            _sg_free(log_buf);
        }
        glDeleteProgram(gl_prog);
        return SG_RESOURCESTATE_FAILED;
    }
    shd->gl.prog = gl_prog;

    // resolve uniforms
    _SG_GL_CHECK_ERROR();
    for (size_t ub_index = 0; ub_index < SG_MAX_UNIFORMBLOCK_BINDSLOTS; ub_index++) {
        const sg_shader_uniform_block* ub_desc = &desc->uniform_blocks[ub_index];
        if (ub_desc->stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(ub_desc->size > 0);
        _sg_gl_uniform_block_t* ub = &shd->gl.uniform_blocks[ub_index];
        SOKOL_ASSERT(ub->num_uniforms == 0);
        uint32_t cur_uniform_offset = 0;
        for (int u_index = 0; u_index < SG_MAX_UNIFORMBLOCK_MEMBERS; u_index++) {
            const sg_glsl_shader_uniform* u_desc = &ub_desc->glsl_uniforms[u_index];
            if (u_desc->type == SG_UNIFORMTYPE_INVALID) {
                break;
            }
            const uint32_t u_align = _sg_uniform_alignment(u_desc->type, u_desc->array_count, ub_desc->layout);
            const uint32_t u_size = _sg_uniform_size(u_desc->type, u_desc->array_count, ub_desc->layout);
            cur_uniform_offset = _sg_align_u32(cur_uniform_offset, u_align);
            _sg_gl_uniform_t* u = &ub->uniforms[u_index];
            u->type = u_desc->type;
            u->count = (uint16_t) u_desc->array_count;
            u->offset = (uint16_t) cur_uniform_offset;
            SOKOL_ASSERT(u_desc->glsl_name);
            u->gl_loc = glGetUniformLocation(gl_prog, u_desc->glsl_name);
            if (u->gl_loc == -1) {
                _SG_WARN(GL_UNIFORMBLOCK_NAME_NOT_FOUND_IN_SHADER);
                _SG_LOGMSG(GL_UNIFORMBLOCK_NAME_NOT_FOUND_IN_SHADER, u_desc->glsl_name);
            }
            cur_uniform_offset += u_size;
            ub->num_uniforms++;
        }
        if (ub_desc->layout == SG_UNIFORMLAYOUT_STD140) {
            cur_uniform_offset = _sg_align_u32(cur_uniform_offset, 16);
        }
        SOKOL_ASSERT(ub_desc->size == (size_t)cur_uniform_offset);
        _SOKOL_UNUSED(cur_uniform_offset);
    }

    // copy storage buffer bind slots
    for (size_t sbuf_index = 0; sbuf_index < SG_MAX_STORAGEBUFFER_BINDSLOTS; sbuf_index++) {
        const sg_shader_storage_buffer* sbuf_desc = &desc->storage_buffers[sbuf_index];
        if (sbuf_desc->stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(sbuf_desc->glsl_binding_n < _SG_GL_MAX_SBUF_BINDINGS);
        shd->gl.sbuf_binding[sbuf_index] = sbuf_desc->glsl_binding_n;
    }

    // copy storage image bind slots
    for (size_t simg_index = 0; simg_index < SG_MAX_STORAGE_ATTACHMENTS; simg_index++) {
        const sg_shader_storage_image* simg_desc = &desc->storage_images[simg_index];
        if (simg_desc->stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(simg_desc->glsl_binding_n < _SG_GL_MAX_SIMG_BINDINGS);
        shd->gl.simg_binding[simg_index] = simg_desc->glsl_binding_n;
    }

    // record image sampler location in shader program
    _SG_GL_CHECK_ERROR();
    GLuint cur_prog = 0;
    glGetIntegerv(GL_CURRENT_PROGRAM, (GLint*)&cur_prog);
    glUseProgram(gl_prog);
    GLint gl_tex_slot = 0;
    for (size_t img_smp_index = 0; img_smp_index < SG_MAX_IMAGE_SAMPLER_PAIRS; img_smp_index++) {
        const sg_shader_image_sampler_pair* img_smp_desc = &desc->image_sampler_pairs[img_smp_index];
        if (img_smp_desc->stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(img_smp_desc->glsl_name);
        GLint gl_loc = glGetUniformLocation(gl_prog, img_smp_desc->glsl_name);
        if (gl_loc != -1) {
            glUniform1i(gl_loc, gl_tex_slot);
            shd->gl.tex_slot[img_smp_index] = (int8_t)gl_tex_slot++;
        } else {
            shd->gl.tex_slot[img_smp_index] = -1;
            _SG_WARN(GL_IMAGE_SAMPLER_NAME_NOT_FOUND_IN_SHADER);
            _SG_LOGMSG(GL_IMAGE_SAMPLER_NAME_NOT_FOUND_IN_SHADER, img_smp_desc->glsl_name);
        }
    }

    // it's legal to call glUseProgram with 0
    glUseProgram(cur_prog);
    _SG_GL_CHECK_ERROR();
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    _SG_GL_CHECK_ERROR();
    if (shd->gl.prog) {
        _sg_gl_cache_invalidate_program(shd->gl.prog);
        glDeleteProgram(shd->gl.prog);
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE sg_resource_state _sg_gl_create_pipeline(_sg_pipeline_t* pip, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && desc);
    SOKOL_ASSERT(_sg.limits.max_vertex_attrs <= SG_MAX_VERTEX_ATTRIBUTES);
    if (pip->cmn.is_compute) {
        // shortcut for compute pipelines
        return SG_RESOURCESTATE_VALID;
    }
    pip->gl.primitive_type = desc->primitive_type;
    pip->gl.depth = desc->depth;
    pip->gl.stencil = desc->stencil;
    // FIXME: blend color and write mask per draw-buffer-attachment (requires GL4)
    pip->gl.blend = desc->colors[0].blend;
    for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        pip->gl.color_write_mask[i] = desc->colors[i].write_mask;
    }
    pip->gl.cull_mode = desc->cull_mode;
    pip->gl.face_winding = desc->face_winding;
    pip->gl.sample_count = desc->sample_count;
    pip->gl.alpha_to_coverage_enabled = desc->alpha_to_coverage_enabled;

    // NOTE: GLSL compilers may remove unused vertex attributes so we can't rely
    // on the 'prepopulated' vertex_buffer_layout_active[] state and need to
    // fill this array from scratch with the actual info after GLSL compilation
    for (int i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
        pip->cmn.vertex_buffer_layout_active[i] = false;
    }

    // resolve vertex attributes
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
    SOKOL_ASSERT(shd->gl.prog);
    for (int attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
        pip->gl.attrs[attr_index].vb_index = -1;
    }
    for (int attr_index = 0; attr_index < _sg.limits.max_vertex_attrs; attr_index++) {
        const sg_vertex_attr_state* a_state = &desc->layout.attrs[attr_index];
        if (a_state->format == SG_VERTEXFORMAT_INVALID) {
            break;
        }
        SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
        const sg_vertex_buffer_layout_state* l_state = &desc->layout.buffers[a_state->buffer_index];
        const sg_vertex_step step_func = l_state->step_func;
        const int step_rate = l_state->step_rate;
        GLint attr_loc = attr_index;
        if (!_sg_strempty(&shd->gl.attrs[attr_index].name)) {
            attr_loc = glGetAttribLocation(shd->gl.prog, _sg_strptr(&shd->gl.attrs[attr_index].name));
        }
        if (attr_loc != -1) {
            SOKOL_ASSERT(attr_loc < (GLint)_sg.limits.max_vertex_attrs);
            _sg_gl_attr_t* gl_attr = &pip->gl.attrs[attr_loc];
            SOKOL_ASSERT(gl_attr->vb_index == -1);
            gl_attr->vb_index = (int8_t) a_state->buffer_index;
            if (step_func == SG_VERTEXSTEP_PER_VERTEX) {
                gl_attr->divisor = 0;
            } else {
                gl_attr->divisor = (int8_t) step_rate;
                pip->cmn.use_instanced_draw = true;
            }
            SOKOL_ASSERT(l_state->stride > 0);
            gl_attr->stride = (uint8_t) l_state->stride;
            gl_attr->offset = a_state->offset;
            gl_attr->size = (uint8_t) _sg_gl_vertexformat_size(a_state->format);
            gl_attr->type = _sg_gl_vertexformat_type(a_state->format);
            gl_attr->normalized = _sg_gl_vertexformat_normalized(a_state->format);
            gl_attr->base_type = _sg_vertexformat_basetype(a_state->format);
            pip->cmn.vertex_buffer_layout_active[a_state->buffer_index] = true;
        } else {
            _SG_WARN(GL_VERTEX_ATTRIBUTE_NOT_FOUND_IN_SHADER);
            _SG_LOGMSG(GL_VERTEX_ATTRIBUTE_NOT_FOUND_IN_SHADER, _sg_strptr(&shd->gl.attrs[attr_index].name));
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _sg_gl_cache_invalidate_pipeline(pip);
}

_SOKOL_PRIVATE void _sg_gl_fb_attach_texture(const _sg_attachment_common_t* att, GLenum gl_att_type) {
    const _sg_image_t* img = _sg_image_ref_ptr(&att->image);
    const GLuint gl_tex = img->gl.tex[0];
    SOKOL_ASSERT(gl_tex);
    const GLuint gl_target = img->gl.target;
    SOKOL_ASSERT(gl_target);
    const int mip_level = att->mip_level;
    const int slice = att->slice;
    switch (img->cmn.type) {
        case SG_IMAGETYPE_2D:
            glFramebufferTexture2D(GL_FRAMEBUFFER, gl_att_type, gl_target, gl_tex, mip_level);
            break;
        case SG_IMAGETYPE_CUBE:
            glFramebufferTexture2D(GL_FRAMEBUFFER, gl_att_type, _sg_gl_cubeface_target(slice), gl_tex, mip_level);
            break;
        default:
            glFramebufferTextureLayer(GL_FRAMEBUFFER, gl_att_type, gl_tex, mip_level, slice);
            break;
    }
}

_SOKOL_PRIVATE GLenum _sg_gl_depth_stencil_attachment_type(const _sg_image_t* ds_img) {
    if (_sg_is_depth_stencil_format(ds_img->cmn.pixel_format)) {
        return GL_DEPTH_STENCIL_ATTACHMENT;
    } else {
        return GL_DEPTH_ATTACHMENT;
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_gl_create_attachments(_sg_attachments_t* atts, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && desc);
    _SOKOL_UNUSED(desc);
    _SG_GL_CHECK_ERROR();

    // if this is a compute pass attachment we're done here
    if (atts->cmn.has_storage_attachments) {
        SOKOL_ASSERT(!atts->cmn.has_render_attachments);
        return SG_RESOURCESTATE_VALID;
    }

    // store current framebuffer binding (restored at end of function)
    GLuint gl_orig_fb;
    glGetIntegerv(GL_FRAMEBUFFER_BINDING, (GLint*)&gl_orig_fb);

    // create a framebuffer object
    glGenFramebuffers(1, &atts->gl.fb);
    glBindFramebuffer(GL_FRAMEBUFFER, atts->gl.fb);

    // attach color attachments to framebuffer
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        const _sg_image_t* color_img = _sg_image_ref_ptr(&atts->cmn.colors[i].image);
        const GLuint gl_msaa_render_buffer = color_img->gl.msaa_render_buffer;
        if (gl_msaa_render_buffer) {
            glFramebufferRenderbuffer(GL_FRAMEBUFFER, (GLenum)(GL_COLOR_ATTACHMENT0+i), GL_RENDERBUFFER, gl_msaa_render_buffer);
        } else {
            const GLenum gl_att_type = (GLenum)(GL_COLOR_ATTACHMENT0 + i);
            _sg_gl_fb_attach_texture(&atts->cmn.colors[i], gl_att_type);
        }
    }
    // attach depth-stencil attachment
    if (!_sg_image_ref_null(&atts->cmn.depth_stencil.image)) {
        const _sg_image_t* ds_img = _sg_image_ref_ptr(&atts->cmn.depth_stencil.image);
        const GLenum gl_att_type = _sg_gl_depth_stencil_attachment_type(ds_img);
        const GLuint gl_msaa_render_buffer = ds_img->gl.msaa_render_buffer;
        if (gl_msaa_render_buffer) {
            glFramebufferRenderbuffer(GL_FRAMEBUFFER, gl_att_type, GL_RENDERBUFFER, gl_msaa_render_buffer);
        } else {
            _sg_gl_fb_attach_texture(&atts->cmn.depth_stencil, gl_att_type);
        }
    }

    // check if framebuffer is complete
    {
        const GLenum fb_status = glCheckFramebufferStatus(GL_FRAMEBUFFER);
        if (fb_status != GL_FRAMEBUFFER_COMPLETE) {
            switch (fb_status) {
                case GL_FRAMEBUFFER_UNDEFINED:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNDEFINED);
                    break;
                case GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_ATTACHMENT);
                    break;
                case GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MISSING_ATTACHMENT);
                    break;
                case GL_FRAMEBUFFER_UNSUPPORTED:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNSUPPORTED);
                    break;
                case GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MULTISAMPLE);
                    break;
                default:
                    _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNKNOWN);
                    break;
            }
            return SG_RESOURCESTATE_FAILED;
        }
    }

    // setup color attachments for the framebuffer
    static const GLenum gl_draw_bufs[SG_MAX_COLOR_ATTACHMENTS] = {
        GL_COLOR_ATTACHMENT0,
        GL_COLOR_ATTACHMENT1,
        GL_COLOR_ATTACHMENT2,
        GL_COLOR_ATTACHMENT3
    };
    glDrawBuffers(atts->cmn.num_colors, gl_draw_bufs);

    // create MSAA resolve framebuffers if necessary
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        _sg_attachment_common_t* cmn_resolve_att = &atts->cmn.resolves[i];
        if (!_sg_image_ref_null(&cmn_resolve_att->image)) {
            SOKOL_ASSERT(0 == atts->gl.msaa_resolve_framebuffer[i]);
            glGenFramebuffers(1, &atts->gl.msaa_resolve_framebuffer[i]);
            glBindFramebuffer(GL_FRAMEBUFFER, atts->gl.msaa_resolve_framebuffer[i]);
            _sg_gl_fb_attach_texture(cmn_resolve_att, GL_COLOR_ATTACHMENT0);
            // check if framebuffer is complete
            const GLenum fb_status = glCheckFramebufferStatus(GL_FRAMEBUFFER);
            if (fb_status != GL_FRAMEBUFFER_COMPLETE) {
                switch (fb_status) {
                    case GL_FRAMEBUFFER_UNDEFINED:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNDEFINED);
                        break;
                    case GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_ATTACHMENT);
                        break;
                    case GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MISSING_ATTACHMENT);
                        break;
                    case GL_FRAMEBUFFER_UNSUPPORTED:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNSUPPORTED);
                        break;
                    case GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_INCOMPLETE_MULTISAMPLE);
                        break;
                    default:
                        _SG_ERROR(GL_FRAMEBUFFER_STATUS_UNKNOWN);
                        break;
                }
                return SG_RESOURCESTATE_FAILED;
            }
            // setup color attachments for the framebuffer
            glDrawBuffers(1, &gl_draw_bufs[0]);
        }
    }

    // restore original framebuffer binding
    glBindFramebuffer(GL_FRAMEBUFFER, gl_orig_fb);
    _SG_GL_CHECK_ERROR();
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_gl_discard_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    _SG_GL_CHECK_ERROR();
    if (0 != atts->gl.fb) {
        glDeleteFramebuffers(1, &atts->gl.fb);
    }
    for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        if (atts->gl.msaa_resolve_framebuffer[i]) {
            glDeleteFramebuffers(1, &atts->gl.msaa_resolve_framebuffer[i]);
        }
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_begin_pass(const sg_pass* pass) {
    // FIXME: what if a texture used as render target is still bound, should we
    // unbind all currently bound textures in begin pass?
    SOKOL_ASSERT(pass);
    _SG_GL_CHECK_ERROR();
    const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);

    // early out if this a compute pass
    if (pass->compute) {
        // first pipeline in pass needs to re-apply storage attachments
        if (atts && atts->cmn.has_storage_attachments) {
            _sg.gl.cache.cur_pip = _sg_sref(0);
        }
        return;
    }

    const sg_swapchain* swapchain = &pass->swapchain;
    const sg_pass_action* action = &pass->action;

    // bind the render pass framebuffer
    //
    // FIXME: Disabling SRGB conversion for the default framebuffer is
    // a crude hack to make behaviour for sRGB render target textures
    // identical with the Metal and D3D11 swapchains created by sokol-app.
    //
    // This will need a cleaner solution (e.g. allowing to configure
    // sokol_app.h with an sRGB or RGB framebuffer.
    if (atts) {
        // offscreen pass
        SOKOL_ASSERT(atts->gl.fb);
        #if defined(SOKOL_GLCORE)
        glEnable(GL_FRAMEBUFFER_SRGB);
        #endif
        glBindFramebuffer(GL_FRAMEBUFFER, atts->gl.fb);
    } else {
        // default pass
        #if defined(SOKOL_GLCORE)
        glDisable(GL_FRAMEBUFFER_SRGB);
        #endif
        // NOTE: on some platforms, the default framebuffer of a context
        // is null, so we can't actually assert here that the
        // framebuffer has been provided
        glBindFramebuffer(GL_FRAMEBUFFER, swapchain->gl.framebuffer);
    }
    glViewport(0, 0, _sg.cur_pass.width, _sg.cur_pass.height);
    glScissor(0, 0, _sg.cur_pass.width, _sg.cur_pass.height);

    // number of color attachments
    const int num_color_atts = atts ? atts->cmn.num_colors : 1;

    // clear color and depth-stencil attachments if needed
    bool clear_any_color = false;
    for (int i = 0; i < num_color_atts; i++) {
        if (SG_LOADACTION_CLEAR == action->colors[i].load_action) {
            clear_any_color = true;
            break;
        }
    }
    const bool clear_depth = (action->depth.load_action == SG_LOADACTION_CLEAR);
    const bool clear_stencil = (action->stencil.load_action == SG_LOADACTION_CLEAR);

    bool need_pip_cache_flush = false;
    if (clear_any_color) {
        bool need_color_mask_flush = false;
        // NOTE: not a bug to iterate over all possible color attachments
        for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
            if (SG_COLORMASK_RGBA != _sg.gl.cache.color_write_mask[i]) {
                need_pip_cache_flush = true;
                need_color_mask_flush = true;
                _sg.gl.cache.color_write_mask[i] = SG_COLORMASK_RGBA;
            }
        }
        if (need_color_mask_flush) {
            glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
        }
    }
    if (clear_depth) {
        if (!_sg.gl.cache.depth.write_enabled) {
            need_pip_cache_flush = true;
            _sg.gl.cache.depth.write_enabled = true;
            glDepthMask(GL_TRUE);
        }
        if (_sg.gl.cache.depth.compare != SG_COMPAREFUNC_ALWAYS) {
            need_pip_cache_flush = true;
            _sg.gl.cache.depth.compare = SG_COMPAREFUNC_ALWAYS;
            glDepthFunc(GL_ALWAYS);
        }
    }
    if (clear_stencil) {
        if (_sg.gl.cache.stencil.write_mask != 0xFF) {
            need_pip_cache_flush = true;
            _sg.gl.cache.stencil.write_mask = 0xFF;
            glStencilMask(0xFF);
        }
    }
    if (need_pip_cache_flush) {
        // we messed with the state cache directly, need to clear cached
        // pipeline to force re-evaluation in next sg_apply_pipeline()
        _sg.gl.cache.cur_pip = _sg_sref(0);
    }
    for (int i = 0; i < num_color_atts; i++) {
        if (action->colors[i].load_action == SG_LOADACTION_CLEAR) {
            glClearBufferfv(GL_COLOR, i, &action->colors[i].clear_value.r);
        }
    }
    if ((atts == 0) || !_sg_image_ref_null(&atts->cmn.depth_stencil.image)) {
        if (clear_depth && clear_stencil) {
            glClearBufferfi(GL_DEPTH_STENCIL, 0, action->depth.clear_value, action->stencil.clear_value);
        } else if (clear_depth) {
            glClearBufferfv(GL_DEPTH, 0, &action->depth.clear_value);
        } else if (clear_stencil) {
            GLint val = (GLint) action->stencil.clear_value;
            glClearBufferiv(GL_STENCIL, 0, &val);
        }
    }
    // keep store actions for end-pass
    for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        _sg.gl.color_store_actions[i] = action->colors[i].store_action;
    }
    _sg.gl.depth_store_action = action->depth.store_action;
    _sg.gl.stencil_store_action = action->stencil.store_action;

    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_end_render_pass(void) {
    const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);
    if (atts) {
        bool fb_read_bound = false;
        bool fb_draw_bound = false;
        const int num_color_atts = atts->cmn.num_colors;
        for (int i = 0; i < num_color_atts; i++) {
            // perform MSAA resolve if needed
            if (atts->gl.msaa_resolve_framebuffer[i] != 0) {
                if (!fb_read_bound) {
                    SOKOL_ASSERT(atts->gl.fb);
                    glBindFramebuffer(GL_READ_FRAMEBUFFER, atts->gl.fb);
                    fb_read_bound = true;
                }
                const _sg_image_t* img = _sg_image_ref_ptr(&atts->cmn.colors[i].image);
                const int w = img->cmn.width;
                const int h = img->cmn.height;
                glBindFramebuffer(GL_DRAW_FRAMEBUFFER, atts->gl.msaa_resolve_framebuffer[i]);
                glReadBuffer((GLenum)(GL_COLOR_ATTACHMENT0 + i));
                glBlitFramebuffer(0, 0, w, h, 0, 0, w, h, GL_COLOR_BUFFER_BIT, GL_NEAREST);
                fb_draw_bound = true;
            }
        }

        // invalidate framebuffers
        _SOKOL_UNUSED(fb_draw_bound);
        #if defined(SOKOL_GLES3)
        // need to restore framebuffer binding before invalidate if the MSAA resolve had changed the binding
        if (fb_draw_bound) {
            glBindFramebuffer(GL_FRAMEBUFFER, atts->gl.fb);
        }
        GLenum invalidate_atts[SG_MAX_COLOR_ATTACHMENTS + 2] = { 0 };
        int att_index = 0;
        for (int i = 0; i < num_color_atts; i++) {
            if (_sg.gl.color_store_actions[i] == SG_STOREACTION_DONTCARE) {
                invalidate_atts[att_index++] = (GLenum)(GL_COLOR_ATTACHMENT0 + i);
            }
        }
        if ((_sg.gl.depth_store_action == SG_STOREACTION_DONTCARE) && (_sg.cur_pass.atts->cmn.depth_stencil.image_id.id != SG_INVALID_ID)) {
            invalidate_atts[att_index++] = GL_DEPTH_ATTACHMENT;
        }
        if ((_sg.gl.stencil_store_action == SG_STOREACTION_DONTCARE) && (_sg.cur_pass.atts->cmn.depth_stencil.image_id.id != SG_INVALID_ID)) {
            invalidate_atts[att_index++] = GL_STENCIL_ATTACHMENT;
        }
        if (att_index > 0) {
            glInvalidateFramebuffer(GL_DRAW_FRAMEBUFFER, att_index, invalidate_atts);
        }
        #endif
    }
}

_SOKOL_PRIVATE void _sg_gl_end_compute_pass(void) {
    #if defined(_SOKOL_GL_HAS_COMPUTE)
    const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);
    if (atts && atts->cmn.has_storage_attachments) {
        glMemoryBarrier(GL_TEXTURE_FETCH_BARRIER_BIT|GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
    }
    #endif
}

_SOKOL_PRIVATE void _sg_gl_end_pass(void) {
    _SG_GL_CHECK_ERROR();
    if (_sg.cur_pass.is_compute) {
        _sg_gl_end_compute_pass();
    } else {
        _sg_gl_end_render_pass();
    }
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    y = origin_top_left ? (_sg.cur_pass.height - (y+h)) : y;
    glViewport(x, y, w, h);
}

_SOKOL_PRIVATE void _sg_gl_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    y = origin_top_left ? (_sg.cur_pass.height - (y+h)) : y;
    glScissor(x, y, w, h);
}

_SOKOL_PRIVATE void _sg_gl_apply_render_pipeline_state(_sg_pipeline_t* pip) {
    // update render pipeline state
    _sg.gl.cache.cur_primitive_type = _sg_gl_primitive_type(pip->gl.primitive_type);
    _sg.gl.cache.cur_index_type = _sg_gl_index_type(pip->cmn.index_type);

    // update depth state
    {
        const sg_depth_state* state_ds = &pip->gl.depth;
        sg_depth_state* cache_ds = &_sg.gl.cache.depth;
        if (state_ds->compare != cache_ds->compare) {
            cache_ds->compare = state_ds->compare;
            glDepthFunc(_sg_gl_compare_func(state_ds->compare));
            _sg_stats_add(gl.num_render_state, 1);
        }
        if (state_ds->write_enabled != cache_ds->write_enabled) {
            cache_ds->write_enabled = state_ds->write_enabled;
            glDepthMask(state_ds->write_enabled);
            _sg_stats_add(gl.num_render_state, 1);
        }
        if (!_sg_fequal(state_ds->bias, cache_ds->bias, 0.000001f) ||
            !_sg_fequal(state_ds->bias_slope_scale, cache_ds->bias_slope_scale, 0.000001f))
        {
            /* according to ANGLE's D3D11 backend:
                D3D11 SlopeScaledDepthBias ==> GL polygonOffsetFactor
                D3D11 DepthBias ==> GL polygonOffsetUnits
                DepthBiasClamp has no meaning on GL
            */
            cache_ds->bias = state_ds->bias;
            cache_ds->bias_slope_scale = state_ds->bias_slope_scale;
            glPolygonOffset(state_ds->bias_slope_scale, state_ds->bias);
            _sg_stats_add(gl.num_render_state, 1);
            bool po_enabled = true;
            if (_sg_fequal(state_ds->bias, 0.0f, 0.000001f) &&
                _sg_fequal(state_ds->bias_slope_scale, 0.0f, 0.000001f))
            {
                po_enabled = false;
            }
            if (po_enabled != _sg.gl.cache.polygon_offset_enabled) {
                _sg.gl.cache.polygon_offset_enabled = po_enabled;
                if (po_enabled) {
                    glEnable(GL_POLYGON_OFFSET_FILL);
                } else {
                    glDisable(GL_POLYGON_OFFSET_FILL);
                }
                _sg_stats_add(gl.num_render_state, 1);
            }
        }
    }

    // update stencil state
    {
        const sg_stencil_state* state_ss = &pip->gl.stencil;
        sg_stencil_state* cache_ss = &_sg.gl.cache.stencil;
        if (state_ss->enabled != cache_ss->enabled) {
            cache_ss->enabled = state_ss->enabled;
            if (state_ss->enabled) {
                glEnable(GL_STENCIL_TEST);
            } else {
                glDisable(GL_STENCIL_TEST);
            }
            _sg_stats_add(gl.num_render_state, 1);
        }
        if (state_ss->write_mask != cache_ss->write_mask) {
            cache_ss->write_mask = state_ss->write_mask;
            glStencilMask(state_ss->write_mask);
            _sg_stats_add(gl.num_render_state, 1);
        }
        for (int i = 0; i < 2; i++) {
            const sg_stencil_face_state* state_sfs = (i==0)? &state_ss->front : &state_ss->back;
            sg_stencil_face_state* cache_sfs = (i==0)? &cache_ss->front : &cache_ss->back;
            GLenum gl_face = (i==0)? GL_FRONT : GL_BACK;
            if ((state_sfs->compare != cache_sfs->compare) ||
                (state_ss->read_mask != cache_ss->read_mask) ||
                (state_ss->ref != cache_ss->ref))
            {
                cache_sfs->compare = state_sfs->compare;
                glStencilFuncSeparate(gl_face,
                    _sg_gl_compare_func(state_sfs->compare),
                    state_ss->ref,
                    state_ss->read_mask);
                _sg_stats_add(gl.num_render_state, 1);
            }
            if ((state_sfs->fail_op != cache_sfs->fail_op) ||
                (state_sfs->depth_fail_op != cache_sfs->depth_fail_op) ||
                (state_sfs->pass_op != cache_sfs->pass_op))
            {
                cache_sfs->fail_op = state_sfs->fail_op;
                cache_sfs->depth_fail_op = state_sfs->depth_fail_op;
                cache_sfs->pass_op = state_sfs->pass_op;
                glStencilOpSeparate(gl_face,
                    _sg_gl_stencil_op(state_sfs->fail_op),
                    _sg_gl_stencil_op(state_sfs->depth_fail_op),
                    _sg_gl_stencil_op(state_sfs->pass_op));
                _sg_stats_add(gl.num_render_state, 1);
            }
        }
        cache_ss->read_mask = state_ss->read_mask;
        cache_ss->ref = state_ss->ref;
    }

    if (pip->cmn.color_count > 0) {
        // update blend state
        // FIXME: separate blend state per color attachment
        const sg_blend_state* state_bs = &pip->gl.blend;
        sg_blend_state* cache_bs = &_sg.gl.cache.blend;
        if (state_bs->enabled != cache_bs->enabled) {
            cache_bs->enabled = state_bs->enabled;
            if (state_bs->enabled) {
                glEnable(GL_BLEND);
            } else {
                glDisable(GL_BLEND);
            }
            _sg_stats_add(gl.num_render_state, 1);
        }
        if ((state_bs->src_factor_rgb != cache_bs->src_factor_rgb) ||
            (state_bs->dst_factor_rgb != cache_bs->dst_factor_rgb) ||
            (state_bs->src_factor_alpha != cache_bs->src_factor_alpha) ||
            (state_bs->dst_factor_alpha != cache_bs->dst_factor_alpha))
        {
            cache_bs->src_factor_rgb = state_bs->src_factor_rgb;
            cache_bs->dst_factor_rgb = state_bs->dst_factor_rgb;
            cache_bs->src_factor_alpha = state_bs->src_factor_alpha;
            cache_bs->dst_factor_alpha = state_bs->dst_factor_alpha;
            glBlendFuncSeparate(_sg_gl_blend_factor(state_bs->src_factor_rgb),
                _sg_gl_blend_factor(state_bs->dst_factor_rgb),
                _sg_gl_blend_factor(state_bs->src_factor_alpha),
                _sg_gl_blend_factor(state_bs->dst_factor_alpha));
            _sg_stats_add(gl.num_render_state, 1);
        }
        if ((state_bs->op_rgb != cache_bs->op_rgb) || (state_bs->op_alpha != cache_bs->op_alpha)) {
            cache_bs->op_rgb = state_bs->op_rgb;
            cache_bs->op_alpha = state_bs->op_alpha;
            glBlendEquationSeparate(_sg_gl_blend_op(state_bs->op_rgb), _sg_gl_blend_op(state_bs->op_alpha));
            _sg_stats_add(gl.num_render_state, 1);
        }

        // standalone color target state
        for (GLuint i = 0; i < (GLuint)pip->cmn.color_count; i++) {
            if (pip->gl.color_write_mask[i] != _sg.gl.cache.color_write_mask[i]) {
                const sg_color_mask cm = pip->gl.color_write_mask[i];
                _sg.gl.cache.color_write_mask[i] = cm;
                #ifdef SOKOL_GLCORE
                    glColorMaski(i,
                                (cm & SG_COLORMASK_R) != 0,
                                (cm & SG_COLORMASK_G) != 0,
                                (cm & SG_COLORMASK_B) != 0,
                                (cm & SG_COLORMASK_A) != 0);
                #else
                    if (0 == i) {
                        glColorMask((cm & SG_COLORMASK_R) != 0,
                                    (cm & SG_COLORMASK_G) != 0,
                                    (cm & SG_COLORMASK_B) != 0,
                                    (cm & SG_COLORMASK_A) != 0);
                    }
                #endif
                _sg_stats_add(gl.num_render_state, 1);
            }
        }

        if (!_sg_fequal(pip->cmn.blend_color.r, _sg.gl.cache.blend_color.r, 0.0001f) ||
            !_sg_fequal(pip->cmn.blend_color.g, _sg.gl.cache.blend_color.g, 0.0001f) ||
            !_sg_fequal(pip->cmn.blend_color.b, _sg.gl.cache.blend_color.b, 0.0001f) ||
            !_sg_fequal(pip->cmn.blend_color.a, _sg.gl.cache.blend_color.a, 0.0001f))
        {
            sg_color c = pip->cmn.blend_color;
            _sg.gl.cache.blend_color = c;
            glBlendColor(c.r, c.g, c.b, c.a);
            _sg_stats_add(gl.num_render_state, 1);
        }
    } // pip->cmn.color_count > 0

    if (pip->gl.cull_mode != _sg.gl.cache.cull_mode) {
        _sg.gl.cache.cull_mode = pip->gl.cull_mode;
        if (SG_CULLMODE_NONE == pip->gl.cull_mode) {
            glDisable(GL_CULL_FACE);
            _sg_stats_add(gl.num_render_state, 1);
        } else {
            glEnable(GL_CULL_FACE);
            GLenum gl_mode = (SG_CULLMODE_FRONT == pip->gl.cull_mode) ? GL_FRONT : GL_BACK;
            glCullFace(gl_mode);
            _sg_stats_add(gl.num_render_state, 2);
        }
    }
    if (pip->gl.face_winding != _sg.gl.cache.face_winding) {
        _sg.gl.cache.face_winding = pip->gl.face_winding;
        GLenum gl_winding = (SG_FACEWINDING_CW == pip->gl.face_winding) ? GL_CW : GL_CCW;
        glFrontFace(gl_winding);
        _sg_stats_add(gl.num_render_state, 1);
    }
    if (pip->gl.alpha_to_coverage_enabled != _sg.gl.cache.alpha_to_coverage_enabled) {
        _sg.gl.cache.alpha_to_coverage_enabled = pip->gl.alpha_to_coverage_enabled;
        if (pip->gl.alpha_to_coverage_enabled) {
            glEnable(GL_SAMPLE_ALPHA_TO_COVERAGE);
        } else {
            glDisable(GL_SAMPLE_ALPHA_TO_COVERAGE);
        }
        _sg_stats_add(gl.num_render_state, 1);
    }
    #ifdef SOKOL_GLCORE
    if (pip->gl.sample_count != _sg.gl.cache.sample_count) {
        _sg.gl.cache.sample_count = pip->gl.sample_count;
        if (pip->gl.sample_count > 1) {
            glEnable(GL_MULTISAMPLE);
        } else {
            glDisable(GL_MULTISAMPLE);
        }
        _sg_stats_add(gl.num_render_state, 1);
    }
    #endif
}

_SOKOL_PRIVATE void _sg_gl_apply_compute_pipeline_state(_sg_pipeline_t* pip) {
    #if defined(_SOKOL_GL_HAS_COMPUTE)
    // apply storage attachment images (if any)
    const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);
    if (atts) {
        const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
        for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
            if (shd->cmn.storage_images[i].stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            SOKOL_ASSERT(shd->cmn.storage_images[i].stage == SG_SHADERSTAGE_COMPUTE);
            SOKOL_ASSERT(shd->gl.simg_binding[i] < _SG_GL_MAX_SIMG_BINDINGS);
            _sg_image_t* img = _sg_image_ref_ptr(&atts->cmn.storages[i].image);
            GLuint gl_unit = shd->gl.simg_binding[i];
            GLuint gl_tex = img->gl.tex[img->cmn.active_slot];
            GLint level = atts->cmn.storages[i].mip_level;
            GLint layer = atts->cmn.storages[i].slice;
            GLboolean layered = shd->cmn.storage_images[i].image_type != SG_IMAGETYPE_2D;
            GLenum access = shd->cmn.storage_images[i].writeonly ? GL_WRITE_ONLY : GL_READ_WRITE;
            GLenum format = _sg_gl_teximage_internal_format(shd->cmn.storage_images[i].access_format);
            // FIXME: go through state cache, use attachment id as key
            glBindImageTexture(gl_unit, gl_tex, level, layered, layer, access, format);
            _SG_GL_CHECK_ERROR();
        }
    }
    #else
    _SOKOL_UNUSED(pip);
    #endif
}

_SOKOL_PRIVATE void _sg_gl_apply_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _SG_GL_CHECK_ERROR();
    if (!_sg_sref_eql(&_sg.gl.cache.cur_pip, &pip->slot)) {
        _sg.gl.cache.cur_pip = _sg_sref(&pip->slot);

        // bind shader program
        const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
        if (shd->gl.prog != _sg.gl.cache.prog) {
            _sg.gl.cache.prog = shd->gl.prog;
            glUseProgram(shd->gl.prog);
            _sg_stats_add(gl.num_use_program, 1);
        }

        if (pip->cmn.is_compute) {
            _sg_gl_apply_compute_pipeline_state(pip);
        } else {
            _sg_gl_apply_render_pipeline_state(pip);
        }
    }
    _SG_GL_CHECK_ERROR();
}

#if defined(_SOKOL_GL_HAS_COMPUTE)
_SOKOL_PRIVATE void _sg_gl_handle_memory_barriers(const _sg_shader_t* shd, const _sg_bindings_ptrs_t* bnd) {
    if (!_sg.features.compute) {
        return;
    }
    GLbitfield gl_barrier_bits = 0;

    // if vertex-, index- or storage-buffer bindings have been written
    // by a compute shader before, a barrier must be issued
    for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
        _sg_buffer_t* buf = bnd->vbs[i];
        if (!buf) {
            continue;
        }
        if (buf->gl.gpu_dirty_flags & _SG_GL_GPUDIRTY_VERTEXBUFFER) {
            gl_barrier_bits |= GL_VERTEX_ATTRIB_ARRAY_BARRIER_BIT;
            buf->gl.gpu_dirty_flags &= (uint8_t)~_SG_GL_GPUDIRTY_VERTEXBUFFER;
        }
    }
    if (bnd->ib) {
        _sg_buffer_t* buf = bnd->ib;
        if (buf->gl.gpu_dirty_flags & _SG_GL_GPUDIRTY_INDEXBUFFER) {
            gl_barrier_bits |= GL_ELEMENT_ARRAY_BARRIER_BIT;
            buf->gl.gpu_dirty_flags &= (uint8_t)~_SG_GL_GPUDIRTY_INDEXBUFFER;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        _sg_buffer_t* buf = bnd->sbufs[i];
        if (!buf) {
            continue;
        }
        SOKOL_ASSERT(shd->cmn.storage_buffers[i].stage != SG_SHADERSTAGE_NONE);
        if (buf->gl.gpu_dirty_flags & _SG_GL_GPUDIRTY_STORAGEBUFFER) {
            gl_barrier_bits |= GL_SHADER_STORAGE_BARRIER_BIT;
            buf->gl.gpu_dirty_flags &= (uint8_t)~_SG_GL_GPUDIRTY_STORAGEBUFFER;
        }
    }

    // mark storage buffers as dirty which will be written by compute shaders
    // (don't merge this into the above loop, this would mess up the dirty
    // dirty flags if the same buffer is bound multiple times)
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        _sg_buffer_t* buf = bnd->sbufs[i];
        if (!buf) {
            continue;
        }
        if (!shd->cmn.storage_buffers[i].readonly) {
            buf->gl.gpu_dirty_flags = _SG_GL_GPUDIRTY_BUFFER_ALL;
        }
    }
    if (0 != gl_barrier_bits) {
        glMemoryBarrier(gl_barrier_bits);
        _sg_stats_add(gl.num_memory_barriers, 1);
    }
}
#endif

_SOKOL_PRIVATE bool _sg_gl_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip);
    _SG_GL_CHECK_ERROR();
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&bnd->pip->cmn.shader);

    // bind combined image-samplers
    _SG_GL_CHECK_ERROR();
    for (size_t img_smp_index = 0; img_smp_index < SG_MAX_IMAGE_SAMPLER_PAIRS; img_smp_index++) {
        const _sg_shader_image_sampler_t* img_smp = &shd->cmn.image_samplers[img_smp_index];
        if (img_smp->stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        const int8_t gl_tex_slot = (GLint)shd->gl.tex_slot[img_smp_index];
        if (gl_tex_slot != -1) {
            SOKOL_ASSERT(img_smp->image_slot < SG_MAX_IMAGE_BINDSLOTS);
            SOKOL_ASSERT(img_smp->sampler_slot < SG_MAX_SAMPLER_BINDSLOTS);
            const _sg_image_t* img = bnd->imgs[img_smp->image_slot];
            const _sg_sampler_t* smp = bnd->smps[img_smp->sampler_slot];
            SOKOL_ASSERT(img);
            SOKOL_ASSERT(smp);
            const GLenum gl_tgt = img->gl.target;
            const GLuint gl_tex = img->gl.tex[img->cmn.active_slot];
            const GLuint gl_smp = smp->gl.smp;
            _sg_gl_cache_bind_texture_sampler(gl_tex_slot, gl_tgt, gl_tex, gl_smp);
        }
    }
    _SG_GL_CHECK_ERROR();

    // bind storage buffers
    for (size_t sbuf_index = 0; sbuf_index < SG_MAX_STORAGEBUFFER_BINDSLOTS; sbuf_index++) {
        if (shd->cmn.storage_buffers[sbuf_index].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        const _sg_buffer_t* sbuf = bnd->sbufs[sbuf_index];
        const uint8_t binding = shd->gl.sbuf_binding[sbuf_index];
        GLuint gl_sbuf = sbuf->gl.buf[sbuf->cmn.active_slot];
        _sg_gl_cache_bind_storage_buffer(binding, gl_sbuf);
    }
    _SG_GL_CHECK_ERROR();

    if (!bnd->pip->cmn.is_compute) {
        // index buffer (can be 0)
        const GLuint gl_ib = bnd->ib ? bnd->ib->gl.buf[bnd->ib->cmn.active_slot] : 0;
        _sg_gl_cache_bind_buffer(GL_ELEMENT_ARRAY_BUFFER, gl_ib);
        _sg.gl.cache.cur_ib_offset = bnd->ib_offset;

        // vertex attributes
        for (GLuint attr_index = 0; attr_index < (GLuint)_sg.limits.max_vertex_attrs; attr_index++) {
            _sg_gl_attr_t* attr = &bnd->pip->gl.attrs[attr_index];
            _sg_gl_cache_attr_t* cache_attr = &_sg.gl.cache.attrs[attr_index];
            bool cache_attr_dirty = false;
            int vb_offset = 0;
            GLuint gl_vb = 0;
            if (attr->vb_index >= 0) {
                // attribute is enabled
                SOKOL_ASSERT(attr->vb_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
                _sg_buffer_t* vb = bnd->vbs[attr->vb_index];
                SOKOL_ASSERT(vb);
                gl_vb = vb->gl.buf[vb->cmn.active_slot];
                vb_offset = bnd->vb_offsets[attr->vb_index] + attr->offset;
                if ((gl_vb != cache_attr->gl_vbuf) ||
                    (attr->size != cache_attr->gl_attr.size) ||
                    (attr->type != cache_attr->gl_attr.type) ||
                    (attr->normalized != cache_attr->gl_attr.normalized) ||
                    (attr->base_type != cache_attr->gl_attr.base_type) ||
                    (attr->stride != cache_attr->gl_attr.stride) ||
                    (vb_offset != cache_attr->gl_attr.offset) ||
                    (cache_attr->gl_attr.divisor != attr->divisor))
                {
                    _sg_gl_cache_bind_buffer(GL_ARRAY_BUFFER, gl_vb);
                    if (attr->base_type == SG_SHADERATTRBASETYPE_FLOAT) {
                        glVertexAttribPointer(attr_index, attr->size, attr->type, attr->normalized, attr->stride, (const GLvoid*)(GLintptr)vb_offset);
                    } else {
                        glVertexAttribIPointer(attr_index, attr->size, attr->type, attr->stride, (const GLvoid*)(GLintptr)vb_offset);
                    }
                    _sg_stats_add(gl.num_vertex_attrib_pointer, 1);
                    glVertexAttribDivisor(attr_index, (GLuint)attr->divisor);
                    _sg_stats_add(gl.num_vertex_attrib_divisor, 1);
                    cache_attr_dirty = true;
                }
                if (cache_attr->gl_attr.vb_index == -1) {
                    glEnableVertexAttribArray(attr_index);
                    _sg_stats_add(gl.num_enable_vertex_attrib_array, 1);
                    cache_attr_dirty = true;
                }
            } else {
                // attribute is disabled
                if (cache_attr->gl_attr.vb_index != -1) {
                    glDisableVertexAttribArray(attr_index);
                    _sg_stats_add(gl.num_disable_vertex_attrib_array, 1);
                    cache_attr_dirty = true;
                }
            }
            if (cache_attr_dirty) {
                cache_attr->gl_attr = *attr;
                cache_attr->gl_attr.offset = vb_offset;
                cache_attr->gl_vbuf = gl_vb;
            }
        }
        _SG_GL_CHECK_ERROR();
    }

    // take care of storage buffer memory barriers (this needs to happen after the bindings are set)
    #if defined(_SOKOL_GL_HAS_COMPUTE)
    _sg_gl_handle_memory_barriers(shd, bnd);
    _SG_GL_CHECK_ERROR();
    #endif

    return true;
}

_SOKOL_PRIVATE void _sg_gl_apply_uniforms(int ub_slot, const sg_range* data) {
    SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
    const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
    SOKOL_ASSERT(SG_SHADERSTAGE_NONE != shd->cmn.uniform_blocks[ub_slot].stage);
    SOKOL_ASSERT(data->size == shd->cmn.uniform_blocks[ub_slot].size);
    const _sg_gl_uniform_block_t* gl_ub = &shd->gl.uniform_blocks[ub_slot];
    for (int u_index = 0; u_index < gl_ub->num_uniforms; u_index++) {
        const _sg_gl_uniform_t* u = &gl_ub->uniforms[u_index];
        SOKOL_ASSERT(u->type != SG_UNIFORMTYPE_INVALID);
        if (u->gl_loc == -1) {
            continue;
        }
        _sg_stats_add(gl.num_uniform, 1);
        GLfloat* fptr = (GLfloat*) (((uint8_t*)data->ptr) + u->offset);
        GLint* iptr = (GLint*) (((uint8_t*)data->ptr) + u->offset);
        switch (u->type) {
            case SG_UNIFORMTYPE_INVALID:
                break;
            case SG_UNIFORMTYPE_FLOAT:
                glUniform1fv(u->gl_loc, u->count, fptr);
                break;
            case SG_UNIFORMTYPE_FLOAT2:
                glUniform2fv(u->gl_loc, u->count, fptr);
                break;
            case SG_UNIFORMTYPE_FLOAT3:
                glUniform3fv(u->gl_loc, u->count, fptr);
                break;
            case SG_UNIFORMTYPE_FLOAT4:
                glUniform4fv(u->gl_loc, u->count, fptr);
                break;
            case SG_UNIFORMTYPE_INT:
                glUniform1iv(u->gl_loc, u->count, iptr);
                break;
            case SG_UNIFORMTYPE_INT2:
                glUniform2iv(u->gl_loc, u->count, iptr);
                break;
            case SG_UNIFORMTYPE_INT3:
                glUniform3iv(u->gl_loc, u->count, iptr);
                break;
            case SG_UNIFORMTYPE_INT4:
                glUniform4iv(u->gl_loc, u->count, iptr);
                break;
            case SG_UNIFORMTYPE_MAT4:
                glUniformMatrix4fv(u->gl_loc, u->count, GL_FALSE, fptr);
                break;
            default:
                SOKOL_UNREACHABLE;
                break;
        }
    }
}

_SOKOL_PRIVATE void _sg_gl_draw(int base_element, int num_elements, int num_instances) {
    const GLenum i_type = _sg.gl.cache.cur_index_type;
    const GLenum p_type = _sg.gl.cache.cur_primitive_type;
    const bool use_instanced_draw = (num_instances > 1) || _sg_pipeline_ref_ptr(&_sg.cur_pip)->cmn.use_instanced_draw;
    if (0 != i_type) {
        // indexed rendering
        const int i_size = (i_type == GL_UNSIGNED_SHORT) ? 2 : 4;
        const int ib_offset = _sg.gl.cache.cur_ib_offset;
        const GLvoid* indices = (const GLvoid*)(GLintptr)(base_element*i_size+ib_offset);
        if (use_instanced_draw) {
            glDrawElementsInstanced(p_type, num_elements, i_type, indices, num_instances);
        } else {
            glDrawElements(p_type, num_elements, i_type, indices);
        }
    } else {
        // non-indexed rendering
        if (use_instanced_draw) {
            glDrawArraysInstanced(p_type, base_element, num_elements, num_instances);
        } else {
            glDrawArrays(p_type, base_element, num_elements);
        }
    }
}

_SOKOL_PRIVATE void _sg_gl_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    #if defined(_SOKOL_GL_HAS_COMPUTE)
    if (!_sg.features.compute) {
        return;
    }
    glDispatchCompute((GLuint)num_groups_x, (GLuint)num_groups_y, (GLuint)num_groups_z);
    #else
    (void)num_groups_x; (void)num_groups_y; (void)num_groups_z;
    #endif
}

_SOKOL_PRIVATE void _sg_gl_commit(void) {
    // "soft" clear bindings (only those that are actually bound)
    _sg_gl_cache_clear_buffer_bindings(false);
    _sg_gl_cache_clear_texture_sampler_bindings(false);
}

_SOKOL_PRIVATE void _sg_gl_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    // only one update per buffer per frame allowed
    if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
        buf->cmn.active_slot = 0;
    }
    GLenum gl_tgt = _sg_gl_buffer_target(&buf->cmn.usage);
    SOKOL_ASSERT(buf->cmn.active_slot < SG_NUM_INFLIGHT_FRAMES);
    GLuint gl_buf = buf->gl.buf[buf->cmn.active_slot];
    SOKOL_ASSERT(gl_buf);
    _SG_GL_CHECK_ERROR();
    _sg_gl_cache_store_buffer_binding(gl_tgt);
    _sg_gl_cache_bind_buffer(gl_tgt, gl_buf);
    glBufferSubData(gl_tgt, 0, (GLsizeiptr)data->size, data->ptr);
    _sg_gl_cache_restore_buffer_binding(gl_tgt);
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    if (new_frame) {
        if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
            buf->cmn.active_slot = 0;
        }
    }
    GLenum gl_tgt = _sg_gl_buffer_target(&buf->cmn.usage);
    SOKOL_ASSERT(buf->cmn.active_slot < SG_NUM_INFLIGHT_FRAMES);
    GLuint gl_buf = buf->gl.buf[buf->cmn.active_slot];
    SOKOL_ASSERT(gl_buf);
    _SG_GL_CHECK_ERROR();
    _sg_gl_cache_store_buffer_binding(gl_tgt);
    _sg_gl_cache_bind_buffer(gl_tgt, gl_buf);
    glBufferSubData(gl_tgt, buf->cmn.append_pos, (GLsizeiptr)data->size, data->ptr);
    _sg_gl_cache_restore_buffer_binding(gl_tgt);
    _SG_GL_CHECK_ERROR();
}

_SOKOL_PRIVATE void _sg_gl_update_image(_sg_image_t* img, const sg_image_data* data) {
    SOKOL_ASSERT(img && data);
    // only one update per image per frame allowed
    if (++img->cmn.active_slot >= img->cmn.num_slots) {
        img->cmn.active_slot = 0;
    }
    SOKOL_ASSERT(img->cmn.active_slot < SG_NUM_INFLIGHT_FRAMES);
    SOKOL_ASSERT(0 != img->gl.tex[img->cmn.active_slot]);
    _sg_gl_cache_store_texture_sampler_binding(0);
    _sg_gl_cache_bind_texture_sampler(0, img->gl.target, img->gl.tex[img->cmn.active_slot], 0);
    const GLenum gl_img_format = _sg_gl_teximage_format(img->cmn.pixel_format);
    const GLenum gl_img_type = _sg_gl_teximage_type(img->cmn.pixel_format);
    const int num_faces = img->cmn.type == SG_IMAGETYPE_CUBE ? 6 : 1;
    const int num_mips = img->cmn.num_mipmaps;
    for (int face_index = 0; face_index < num_faces; face_index++) {
        for (int mip_index = 0; mip_index < num_mips; mip_index++) {
            GLenum gl_img_target = img->gl.target;
            if (SG_IMAGETYPE_CUBE == img->cmn.type) {
                gl_img_target = _sg_gl_cubeface_target(face_index);
            }
            const GLvoid* data_ptr = data->subimage[face_index][mip_index].ptr;
            int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
            int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
            if ((SG_IMAGETYPE_2D == img->cmn.type) || (SG_IMAGETYPE_CUBE == img->cmn.type)) {
                glTexSubImage2D(gl_img_target, mip_index,
                    0, 0,
                    mip_width, mip_height,
                    gl_img_format, gl_img_type,
                    data_ptr);
            } else if ((SG_IMAGETYPE_3D == img->cmn.type) || (SG_IMAGETYPE_ARRAY == img->cmn.type)) {
                int mip_depth = img->cmn.num_slices;
                if (SG_IMAGETYPE_3D == img->cmn.type) {
                    mip_depth = _sg_miplevel_dim(img->cmn.num_slices, mip_index);
                }
                glTexSubImage3D(gl_img_target, mip_index,
                    0, 0, 0,
                    mip_width, mip_height, mip_depth,
                    gl_img_format, gl_img_type,
                    data_ptr);

            }
        }
    }
    _sg_gl_cache_restore_texture_sampler_binding(0);
}

// ██████  ██████  ██████   ██  ██     ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ██   ██      ██ ██   ██ ███ ███     ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██   ██  █████  ██   ██  ██  ██     ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██   ██      ██ ██   ██  ██  ██     ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
// ██████  ██████  ██████   ██  ██     ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>d3d11 backend
#elif defined(SOKOL_D3D11)

#if defined(__cplusplus)
#define _sg_d3d11_AddRef(self) (self)->AddRef()
#else
#define _sg_d3d11_AddRef(self) (self)->lpVtbl->AddRef(self)
#endif

#if defined(__cplusplus)
#define _sg_d3d11_Release(self) (self)->Release()
#else
#define _sg_d3d11_Release(self) (self)->lpVtbl->Release(self)
#endif

// NOTE: This needs to be a macro since we can't use the polymorphism in C. It's called on many kinds of resources.
// NOTE: Based on microsoft docs, it's fine to call this with pData=NULL if DataSize is also zero.
#if defined(__cplusplus)
#define _sg_d3d11_SetPrivateData(self, guid, DataSize, pData) (self)->SetPrivateData(guid, DataSize, pData)
#else
#define _sg_d3d11_SetPrivateData(self, guid, DataSize, pData) (self)->lpVtbl->SetPrivateData(self, guid, DataSize, pData)
#endif

#if defined(__cplusplus)
#define _sg_win32_refguid(guid) guid
#else
#define _sg_win32_refguid(guid) &guid
#endif

static const GUID _sg_d3d11_WKPDID_D3DDebugObjectName = { 0x429b8c22,0x9188,0x4b0c, {0x87,0x42,0xac,0xb0,0xbf,0x85,0xc2,0x00} };

#if defined(SOKOL_DEBUG)
#define _sg_d3d11_setlabel(self, label) _sg_d3d11_SetPrivateData(self, _sg_win32_refguid(_sg_d3d11_WKPDID_D3DDebugObjectName), label ? (UINT)strlen(label) : 0, label)
#else
#define _sg_d3d11_setlabel(self, label)
#endif


//-- D3D11 C/C++ wrappers ------------------------------------------------------
static inline HRESULT _sg_d3d11_CheckFormatSupport(ID3D11Device* self, DXGI_FORMAT Format, UINT* pFormatSupport) {
    #if defined(__cplusplus)
        return self->CheckFormatSupport(Format, pFormatSupport);
    #else
        return self->lpVtbl->CheckFormatSupport(self, Format, pFormatSupport);
    #endif
}

static inline void _sg_d3d11_OMSetRenderTargets(ID3D11DeviceContext* self, UINT NumViews, ID3D11RenderTargetView* const* ppRenderTargetViews, ID3D11DepthStencilView *pDepthStencilView) {
    #if defined(__cplusplus)
        self->OMSetRenderTargets(NumViews, ppRenderTargetViews, pDepthStencilView);
    #else
        self->lpVtbl->OMSetRenderTargets(self, NumViews, ppRenderTargetViews, pDepthStencilView);
    #endif
}

static inline void _sg_d3d11_RSSetState(ID3D11DeviceContext* self, ID3D11RasterizerState* pRasterizerState) {
    #if defined(__cplusplus)
        self->RSSetState(pRasterizerState);
    #else
        self->lpVtbl->RSSetState(self, pRasterizerState);
    #endif
}

static inline void _sg_d3d11_OMSetDepthStencilState(ID3D11DeviceContext* self, ID3D11DepthStencilState* pDepthStencilState, UINT StencilRef) {
    #if defined(__cplusplus)
        self->OMSetDepthStencilState(pDepthStencilState, StencilRef);
    #else
        self->lpVtbl->OMSetDepthStencilState(self, pDepthStencilState, StencilRef);
    #endif
}

static inline void _sg_d3d11_OMSetBlendState(ID3D11DeviceContext* self, ID3D11BlendState* pBlendState, const FLOAT BlendFactor[4], UINT SampleMask) {
    #if defined(__cplusplus)
        self->OMSetBlendState(pBlendState, BlendFactor, SampleMask);
    #else
        self->lpVtbl->OMSetBlendState(self, pBlendState, BlendFactor, SampleMask);
    #endif
}

static inline void _sg_d3d11_IASetVertexBuffers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumBuffers, ID3D11Buffer* const* ppVertexBuffers, const UINT* pStrides, const UINT* pOffsets) {
    #if defined(__cplusplus)
        self->IASetVertexBuffers(StartSlot, NumBuffers, ppVertexBuffers, pStrides, pOffsets);
    #else
        self->lpVtbl->IASetVertexBuffers(self, StartSlot, NumBuffers, ppVertexBuffers, pStrides, pOffsets);
    #endif
}

static inline void _sg_d3d11_IASetIndexBuffer(ID3D11DeviceContext* self, ID3D11Buffer* pIndexBuffer, DXGI_FORMAT Format, UINT Offset) {
    #if defined(__cplusplus)
        self->IASetIndexBuffer(pIndexBuffer, Format, Offset);
    #else
        self->lpVtbl->IASetIndexBuffer(self, pIndexBuffer, Format, Offset);
    #endif
}

static inline void _sg_d3d11_IASetInputLayout(ID3D11DeviceContext* self, ID3D11InputLayout* pInputLayout) {
    #if defined(__cplusplus)
        self->IASetInputLayout(pInputLayout);
    #else
        self->lpVtbl->IASetInputLayout(self, pInputLayout);
    #endif
}

static inline void _sg_d3d11_VSSetShader(ID3D11DeviceContext* self, ID3D11VertexShader* pVertexShader, ID3D11ClassInstance* const* ppClassInstances, UINT NumClassInstances) {
    #if defined(__cplusplus)
        self->VSSetShader(pVertexShader, ppClassInstances, NumClassInstances);
    #else
        self->lpVtbl->VSSetShader(self, pVertexShader, ppClassInstances, NumClassInstances);
    #endif
}

static inline void _sg_d3d11_PSSetShader(ID3D11DeviceContext* self, ID3D11PixelShader* pPixelShader, ID3D11ClassInstance* const* ppClassInstances, UINT NumClassInstances) {
    #if defined(__cplusplus)
        self->PSSetShader(pPixelShader, ppClassInstances, NumClassInstances);
    #else
        self->lpVtbl->PSSetShader(self, pPixelShader, ppClassInstances, NumClassInstances);
    #endif
}

static inline void _sg_d3d11_CSSetShader(ID3D11DeviceContext* self, ID3D11ComputeShader* pComputeShader, ID3D11ClassInstance* const* ppClassInstances, UINT NumClassInstances) {
    #if defined(__cplusplus)
        self->CSSetShader(pComputeShader, ppClassInstances, NumClassInstances);
    #else
        self->lpVtbl->CSSetShader(self, pComputeShader, ppClassInstances, NumClassInstances);
    #endif
}

static inline void _sg_d3d11_VSSetConstantBuffers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumBuffers, ID3D11Buffer* const* ppConstantBuffers) {
    #if defined(__cplusplus)
        self->VSSetConstantBuffers(StartSlot, NumBuffers, ppConstantBuffers);
    #else
        self->lpVtbl->VSSetConstantBuffers(self, StartSlot, NumBuffers, ppConstantBuffers);
    #endif
}

static inline void _sg_d3d11_PSSetConstantBuffers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumBuffers, ID3D11Buffer* const* ppConstantBuffers) {
    #if defined(__cplusplus)
        self->PSSetConstantBuffers(StartSlot, NumBuffers, ppConstantBuffers);
    #else
        self->lpVtbl->PSSetConstantBuffers(self, StartSlot, NumBuffers, ppConstantBuffers);
    #endif
}

static inline void _sg_d3d11_CSSetConstantBuffers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumBuffers, ID3D11Buffer* const* ppConstantBuffers) {
    #if defined(__cplusplus)
        self->CSSetConstantBuffers(StartSlot, NumBuffers, ppConstantBuffers);
    #else
        self->lpVtbl->CSSetConstantBuffers(self, StartSlot, NumBuffers, ppConstantBuffers);
    #endif
}

static inline void _sg_d3d11_VSSetShaderResources(ID3D11DeviceContext* self, UINT StartSlot, UINT NumViews, ID3D11ShaderResourceView* const* ppShaderResourceViews) {
    #if defined(__cplusplus)
        self->VSSetShaderResources(StartSlot, NumViews, ppShaderResourceViews);
    #else
        self->lpVtbl->VSSetShaderResources(self, StartSlot, NumViews, ppShaderResourceViews);
    #endif
}

static inline void _sg_d3d11_PSSetShaderResources(ID3D11DeviceContext* self, UINT StartSlot, UINT NumViews, ID3D11ShaderResourceView* const* ppShaderResourceViews) {
    #if defined(__cplusplus)
        self->PSSetShaderResources(StartSlot, NumViews, ppShaderResourceViews);
    #else
        self->lpVtbl->PSSetShaderResources(self, StartSlot, NumViews, ppShaderResourceViews);
    #endif
}

static inline void _sg_d3d11_CSSetShaderResources(ID3D11DeviceContext* self, UINT StartSlot, UINT NumViews, ID3D11ShaderResourceView* const* ppShaderResourceViews) {
    #if defined(__cplusplus)
        self->CSSetShaderResources(StartSlot, NumViews, ppShaderResourceViews);
    #else
        self->lpVtbl->CSSetShaderResources(self, StartSlot, NumViews, ppShaderResourceViews);
    #endif
}

static inline void _sg_d3d11_VSSetSamplers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumSamplers, ID3D11SamplerState* const* ppSamplers) {
    #if defined(__cplusplus)
        self->VSSetSamplers(StartSlot, NumSamplers, ppSamplers);
    #else
        self->lpVtbl->VSSetSamplers(self, StartSlot, NumSamplers, ppSamplers);
    #endif
}

static inline void _sg_d3d11_PSSetSamplers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumSamplers, ID3D11SamplerState* const* ppSamplers) {
    #if defined(__cplusplus)
        self->PSSetSamplers(StartSlot, NumSamplers, ppSamplers);
    #else
        self->lpVtbl->PSSetSamplers(self, StartSlot, NumSamplers, ppSamplers);
    #endif
}

static inline void _sg_d3d11_CSSetSamplers(ID3D11DeviceContext* self, UINT StartSlot, UINT NumSamplers, ID3D11SamplerState* const* ppSamplers) {
    #if defined(__cplusplus)
        self->CSSetSamplers(StartSlot, NumSamplers, ppSamplers);
    #else
        self->lpVtbl->CSSetSamplers(self, StartSlot, NumSamplers, ppSamplers);
    #endif
}

static inline void _sg_d3d11_CSSetUnorderedAccessViews(ID3D11DeviceContext* self, UINT StartSlot, UINT NumUAVs, ID3D11UnorderedAccessView* const* ppUnorderedAccessViews, const UINT* pUAVInitialCounts) {
    #if defined(__cplusplus)
        self->CSSetUnorderedAccessViews(StartSlot, NumUAVs, ppUnorderedAccessViews, pUAVInitialCounts);
    #else
        self->lpVtbl->CSSetUnorderedAccessViews(self, StartSlot, NumUAVs, ppUnorderedAccessViews, pUAVInitialCounts);
    #endif
}

static inline HRESULT _sg_d3d11_CreateBuffer(ID3D11Device* self, const D3D11_BUFFER_DESC* pDesc, const D3D11_SUBRESOURCE_DATA* pInitialData, ID3D11Buffer** ppBuffer) {
    #if defined(__cplusplus)
        return self->CreateBuffer(pDesc, pInitialData, ppBuffer);
    #else
        return self->lpVtbl->CreateBuffer(self, pDesc, pInitialData, ppBuffer);
    #endif
}

static inline HRESULT _sg_d3d11_CreateTexture2D(ID3D11Device* self, const D3D11_TEXTURE2D_DESC* pDesc, const D3D11_SUBRESOURCE_DATA* pInitialData, ID3D11Texture2D** ppTexture2D) {
    #if defined(__cplusplus)
        return self->CreateTexture2D(pDesc, pInitialData, ppTexture2D);
    #else
        return self->lpVtbl->CreateTexture2D(self, pDesc, pInitialData, ppTexture2D);
    #endif
}

static inline HRESULT _sg_d3d11_CreateShaderResourceView(ID3D11Device* self, ID3D11Resource* pResource, const D3D11_SHADER_RESOURCE_VIEW_DESC* pDesc, ID3D11ShaderResourceView** ppSRView) {
    #if defined(__cplusplus)
        return self->CreateShaderResourceView(pResource, pDesc, ppSRView);
    #else
        return self->lpVtbl->CreateShaderResourceView(self, pResource, pDesc, ppSRView);
    #endif
}

static inline HRESULT _sg_d3d11_CreateUnorderedAccessView(ID3D11Device* self, ID3D11Resource* pResource, const D3D11_UNORDERED_ACCESS_VIEW_DESC* pDesc, ID3D11UnorderedAccessView** ppUAVView) {
    #if defined(__cplusplus)
        return self->CreateUnorderedAccessView(pResource, pDesc, ppUAVView);
    #else
        return self->lpVtbl->CreateUnorderedAccessView(self, pResource, pDesc, ppUAVView);
    #endif
}

static inline void _sg_d3d11_GetResource(ID3D11View* self, ID3D11Resource** ppResource) {
    #if defined(__cplusplus)
        self->GetResource(ppResource);
    #else
        self->lpVtbl->GetResource(self, ppResource);
    #endif
}

static inline HRESULT _sg_d3d11_CreateTexture3D(ID3D11Device* self, const D3D11_TEXTURE3D_DESC* pDesc, const D3D11_SUBRESOURCE_DATA* pInitialData, ID3D11Texture3D** ppTexture3D) {
    #if defined(__cplusplus)
        return self->CreateTexture3D(pDesc, pInitialData, ppTexture3D);
    #else
        return self->lpVtbl->CreateTexture3D(self, pDesc, pInitialData, ppTexture3D);
    #endif
}

static inline HRESULT _sg_d3d11_CreateSamplerState(ID3D11Device* self, const D3D11_SAMPLER_DESC* pSamplerDesc, ID3D11SamplerState** ppSamplerState) {
    #if defined(__cplusplus)
        return self->CreateSamplerState(pSamplerDesc, ppSamplerState);
    #else
        return self->lpVtbl->CreateSamplerState(self, pSamplerDesc, ppSamplerState);
    #endif
}

static inline LPVOID _sg_d3d11_GetBufferPointer(ID3D10Blob* self) {
    #if defined(__cplusplus)
        return self->GetBufferPointer();
    #else
        return self->lpVtbl->GetBufferPointer(self);
    #endif
}

static inline SIZE_T _sg_d3d11_GetBufferSize(ID3D10Blob* self) {
    #if defined(__cplusplus)
        return self->GetBufferSize();
    #else
        return self->lpVtbl->GetBufferSize(self);
    #endif
}

static inline HRESULT _sg_d3d11_CreateVertexShader(ID3D11Device* self, const void* pShaderBytecode, SIZE_T BytecodeLength, ID3D11ClassLinkage* pClassLinkage, ID3D11VertexShader** ppVertexShader) {
    #if defined(__cplusplus)
        return self->CreateVertexShader(pShaderBytecode, BytecodeLength, pClassLinkage, ppVertexShader);
    #else
        return self->lpVtbl->CreateVertexShader(self, pShaderBytecode, BytecodeLength, pClassLinkage, ppVertexShader);
    #endif
}

static inline HRESULT _sg_d3d11_CreatePixelShader(ID3D11Device* self, const void* pShaderBytecode, SIZE_T BytecodeLength, ID3D11ClassLinkage* pClassLinkage, ID3D11PixelShader** ppPixelShader) {
    #if defined(__cplusplus)
        return self->CreatePixelShader(pShaderBytecode, BytecodeLength, pClassLinkage, ppPixelShader);
    #else
        return self->lpVtbl->CreatePixelShader(self, pShaderBytecode, BytecodeLength, pClassLinkage, ppPixelShader);
    #endif
}

static inline HRESULT _sg_d3d11_CreateComputeShader(ID3D11Device* self, const void* pShaderBytecode, SIZE_T BytecodeLength, ID3D11ClassLinkage* pClassLinkage, ID3D11ComputeShader** ppComputeShader) {
    #if defined(__cplusplus)
        return self->CreateComputeShader(pShaderBytecode, BytecodeLength, pClassLinkage, ppComputeShader);
    #else
        return self->lpVtbl->CreateComputeShader(self, pShaderBytecode, BytecodeLength, pClassLinkage, ppComputeShader);
    #endif
}

static inline HRESULT _sg_d3d11_CreateInputLayout(ID3D11Device* self, const D3D11_INPUT_ELEMENT_DESC* pInputElementDescs, UINT NumElements, const void* pShaderBytecodeWithInputSignature, SIZE_T BytecodeLength, ID3D11InputLayout **ppInputLayout) {
    #if defined(__cplusplus)
        return self->CreateInputLayout(pInputElementDescs, NumElements, pShaderBytecodeWithInputSignature, BytecodeLength, ppInputLayout);
    #else
        return self->lpVtbl->CreateInputLayout(self, pInputElementDescs, NumElements, pShaderBytecodeWithInputSignature, BytecodeLength, ppInputLayout);
    #endif
}

static inline HRESULT _sg_d3d11_CreateRasterizerState(ID3D11Device* self, const D3D11_RASTERIZER_DESC* pRasterizerDesc, ID3D11RasterizerState** ppRasterizerState) {
    #if defined(__cplusplus)
        return self->CreateRasterizerState(pRasterizerDesc, ppRasterizerState);
    #else
        return self->lpVtbl->CreateRasterizerState(self, pRasterizerDesc, ppRasterizerState);
    #endif
}

static inline HRESULT _sg_d3d11_CreateDepthStencilState(ID3D11Device* self, const D3D11_DEPTH_STENCIL_DESC* pDepthStencilDesc, ID3D11DepthStencilState** ppDepthStencilState) {
    #if defined(__cplusplus)
        return self->CreateDepthStencilState(pDepthStencilDesc, ppDepthStencilState);
    #else
        return self->lpVtbl->CreateDepthStencilState(self, pDepthStencilDesc, ppDepthStencilState);
    #endif
}

static inline HRESULT _sg_d3d11_CreateBlendState(ID3D11Device* self, const D3D11_BLEND_DESC* pBlendStateDesc, ID3D11BlendState** ppBlendState) {
    #if defined(__cplusplus)
        return self->CreateBlendState(pBlendStateDesc, ppBlendState);
    #else
        return self->lpVtbl->CreateBlendState(self, pBlendStateDesc, ppBlendState);
    #endif
}

static inline HRESULT _sg_d3d11_CreateRenderTargetView(ID3D11Device* self, ID3D11Resource *pResource, const D3D11_RENDER_TARGET_VIEW_DESC* pDesc, ID3D11RenderTargetView** ppRTView) {
    #if defined(__cplusplus)
        return self->CreateRenderTargetView(pResource, pDesc, ppRTView);
    #else
        return self->lpVtbl->CreateRenderTargetView(self, pResource, pDesc, ppRTView);
    #endif
}

static inline HRESULT _sg_d3d11_CreateDepthStencilView(ID3D11Device* self, ID3D11Resource* pResource, const D3D11_DEPTH_STENCIL_VIEW_DESC* pDesc, ID3D11DepthStencilView** ppDepthStencilView) {
    #if defined(__cplusplus)
        return self->CreateDepthStencilView(pResource, pDesc, ppDepthStencilView);
    #else
        return self->lpVtbl->CreateDepthStencilView(self, pResource, pDesc, ppDepthStencilView);
    #endif
}

static inline void _sg_d3d11_RSSetViewports(ID3D11DeviceContext* self, UINT NumViewports, const D3D11_VIEWPORT* pViewports) {
    #if defined(__cplusplus)
        self->RSSetViewports(NumViewports, pViewports);
    #else
        self->lpVtbl->RSSetViewports(self, NumViewports, pViewports);
    #endif
}

static inline void _sg_d3d11_RSSetScissorRects(ID3D11DeviceContext* self, UINT NumRects, const D3D11_RECT* pRects) {
    #if defined(__cplusplus)
        self->RSSetScissorRects(NumRects, pRects);
    #else
        self->lpVtbl->RSSetScissorRects(self, NumRects, pRects);
    #endif
}

static inline void _sg_d3d11_ClearRenderTargetView(ID3D11DeviceContext* self, ID3D11RenderTargetView* pRenderTargetView, const FLOAT ColorRGBA[4]) {
    #if defined(__cplusplus)
        self->ClearRenderTargetView(pRenderTargetView, ColorRGBA);
    #else
        self->lpVtbl->ClearRenderTargetView(self, pRenderTargetView, ColorRGBA);
    #endif
}

static inline void _sg_d3d11_ClearDepthStencilView(ID3D11DeviceContext* self, ID3D11DepthStencilView* pDepthStencilView, UINT ClearFlags, FLOAT Depth, UINT8 Stencil) {
    #if defined(__cplusplus)
        self->ClearDepthStencilView(pDepthStencilView, ClearFlags, Depth, Stencil);
    #else
        self->lpVtbl->ClearDepthStencilView(self, pDepthStencilView, ClearFlags, Depth, Stencil);
    #endif
}

static inline void _sg_d3d11_ResolveSubresource(ID3D11DeviceContext* self, ID3D11Resource* pDstResource, UINT DstSubresource, ID3D11Resource* pSrcResource, UINT SrcSubresource, DXGI_FORMAT Format) {
    #if defined(__cplusplus)
        self->ResolveSubresource(pDstResource, DstSubresource, pSrcResource, SrcSubresource, Format);
    #else
        self->lpVtbl->ResolveSubresource(self, pDstResource, DstSubresource, pSrcResource, SrcSubresource, Format);
    #endif
}

static inline void _sg_d3d11_IASetPrimitiveTopology(ID3D11DeviceContext* self, D3D11_PRIMITIVE_TOPOLOGY Topology) {
    #if defined(__cplusplus)
        self->IASetPrimitiveTopology(Topology);
    #else
        self->lpVtbl->IASetPrimitiveTopology(self, Topology);
    #endif
}

static inline void _sg_d3d11_UpdateSubresource(ID3D11DeviceContext* self, ID3D11Resource* pDstResource, UINT DstSubresource, const D3D11_BOX* pDstBox, const void* pSrcData, UINT SrcRowPitch, UINT SrcDepthPitch) {
    #if defined(__cplusplus)
        self->UpdateSubresource(pDstResource, DstSubresource, pDstBox, pSrcData, SrcRowPitch, SrcDepthPitch);
    #else
        self->lpVtbl->UpdateSubresource(self, pDstResource, DstSubresource, pDstBox, pSrcData, SrcRowPitch, SrcDepthPitch);
    #endif
}

static inline void _sg_d3d11_DrawIndexed(ID3D11DeviceContext* self, UINT IndexCount, UINT StartIndexLocation, INT  BaseVertexLocation) {
    #if defined(__cplusplus)
        self->DrawIndexed(IndexCount, StartIndexLocation, BaseVertexLocation);
    #else
        self->lpVtbl->DrawIndexed(self, IndexCount, StartIndexLocation, BaseVertexLocation);
    #endif
}

static inline void _sg_d3d11_DrawIndexedInstanced(ID3D11DeviceContext* self, UINT IndexCountPerInstance, UINT InstanceCount, UINT StartIndexLocation, INT BaseVertexLocation, UINT StartInstanceLocation) {
    #if defined(__cplusplus)
        self->DrawIndexedInstanced(IndexCountPerInstance, InstanceCount, StartIndexLocation, BaseVertexLocation, StartInstanceLocation);
    #else
        self->lpVtbl->DrawIndexedInstanced(self, IndexCountPerInstance, InstanceCount, StartIndexLocation, BaseVertexLocation, StartInstanceLocation);
    #endif
}

static inline void _sg_d3d11_Draw(ID3D11DeviceContext* self, UINT VertexCount, UINT StartVertexLocation) {
    #if defined(__cplusplus)
        self->Draw(VertexCount, StartVertexLocation);
    #else
        self->lpVtbl->Draw(self, VertexCount, StartVertexLocation);
    #endif
}

static inline void _sg_d3d11_DrawInstanced(ID3D11DeviceContext* self, UINT VertexCountPerInstance, UINT InstanceCount, UINT StartVertexLocation, UINT StartInstanceLocation) {
    #if defined(__cplusplus)
        self->DrawInstanced(VertexCountPerInstance, InstanceCount, StartVertexLocation, StartInstanceLocation);
    #else
        self->lpVtbl->DrawInstanced(self, VertexCountPerInstance, InstanceCount, StartVertexLocation, StartInstanceLocation);
    #endif
}

static inline void _sg_d3d11_Dispatch(ID3D11DeviceContext* self, UINT ThreadGroupCountX, UINT ThreadGroupCountY, UINT ThreadGroupCountZ) {
    #if defined(__cplusplus)
        self->Dispatch(ThreadGroupCountX, ThreadGroupCountY, ThreadGroupCountZ);
    #else
        self->lpVtbl->Dispatch(self, ThreadGroupCountX, ThreadGroupCountY, ThreadGroupCountZ);
    #endif
}

static inline HRESULT _sg_d3d11_Map(ID3D11DeviceContext* self, ID3D11Resource* pResource, UINT Subresource, D3D11_MAP MapType, UINT MapFlags, D3D11_MAPPED_SUBRESOURCE* pMappedResource) {
    #if defined(__cplusplus)
        return self->Map(pResource, Subresource, MapType, MapFlags, pMappedResource);
    #else
        return self->lpVtbl->Map(self, pResource, Subresource, MapType, MapFlags, pMappedResource);
    #endif
}

static inline void _sg_d3d11_Unmap(ID3D11DeviceContext* self, ID3D11Resource* pResource, UINT Subresource) {
    #if defined(__cplusplus)
        self->Unmap(pResource, Subresource);
    #else
        self->lpVtbl->Unmap(self, pResource, Subresource);
    #endif
}

static inline void _sg_d3d11_ClearState(ID3D11DeviceContext* self) {
    #if defined(__cplusplus)
        self->ClearState();
    #else
        self->lpVtbl->ClearState(self);
    #endif
}

//-- enum translation functions ------------------------------------------------
_SOKOL_PRIVATE D3D11_USAGE _sg_d3d11_image_usage(const sg_image_usage* usg) {
    if (usg->immutable) {
        if (usg->render_attachment || usg->storage_attachment) {
            return D3D11_USAGE_DEFAULT;
        } else {
            return D3D11_USAGE_IMMUTABLE;
        }
    } else {
        return D3D11_USAGE_DYNAMIC;
    }
}

_SOKOL_PRIVATE UINT _sg_d3d11_image_bind_flags(const sg_image_usage* usg, sg_pixel_format fmt) {
    UINT res = D3D11_BIND_SHADER_RESOURCE;
    if (usg->render_attachment) {
        if (_sg_is_depth_or_depth_stencil_format(fmt)) {
            res |= D3D11_BIND_DEPTH_STENCIL;
        } else {
            res |= D3D11_BIND_RENDER_TARGET;
        }
    } else if (usg->storage_attachment) {
        res |= D3D11_BIND_UNORDERED_ACCESS;
    }
    return res;
}

_SOKOL_PRIVATE UINT _sg_d3d11_image_cpu_access_flags(const sg_image_usage* usg) {
    if (usg->render_attachment || usg->storage_attachment || usg->immutable) {
        return 0;
    } else {
        return D3D11_CPU_ACCESS_WRITE;
    }
}

_SOKOL_PRIVATE D3D11_USAGE _sg_d3d11_buffer_usage(const sg_buffer_usage* usg) {
    if (usg->immutable) {
        return usg->storage_buffer ? D3D11_USAGE_DEFAULT : D3D11_USAGE_IMMUTABLE;
    } else {
        return D3D11_USAGE_DYNAMIC;
    }
}

_SOKOL_PRIVATE UINT _sg_d3d11_buffer_bind_flags(const sg_buffer_usage* usg) {
    UINT res = 0;
    if (usg->vertex_buffer) {
        res |= D3D11_BIND_VERTEX_BUFFER;
    }
    if (usg->index_buffer) {
        res |= D3D11_BIND_INDEX_BUFFER;
    }
    if (usg->storage_buffer) {
        if (usg->immutable) {
            res |= D3D11_BIND_SHADER_RESOURCE | D3D11_BIND_UNORDERED_ACCESS;
        } else {
            res |= D3D11_BIND_SHADER_RESOURCE;
        }
    }
    return res;
}

_SOKOL_PRIVATE UINT _sg_d3d11_buffer_misc_flags(const sg_buffer_usage* usg) {
    return usg->storage_buffer ? D3D11_RESOURCE_MISC_BUFFER_ALLOW_RAW_VIEWS : 0;
}

_SOKOL_PRIVATE UINT _sg_d3d11_buffer_cpu_access_flags(const sg_buffer_usage* usg) {
    return usg->immutable ? 0 : D3D11_CPU_ACCESS_WRITE;
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_texture_pixel_format(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:             return DXGI_FORMAT_R8_UNORM;
        case SG_PIXELFORMAT_R8SN:           return DXGI_FORMAT_R8_SNORM;
        case SG_PIXELFORMAT_R8UI:           return DXGI_FORMAT_R8_UINT;
        case SG_PIXELFORMAT_R8SI:           return DXGI_FORMAT_R8_SINT;
        case SG_PIXELFORMAT_R16:            return DXGI_FORMAT_R16_UNORM;
        case SG_PIXELFORMAT_R16SN:          return DXGI_FORMAT_R16_SNORM;
        case SG_PIXELFORMAT_R16UI:          return DXGI_FORMAT_R16_UINT;
        case SG_PIXELFORMAT_R16SI:          return DXGI_FORMAT_R16_SINT;
        case SG_PIXELFORMAT_R16F:           return DXGI_FORMAT_R16_FLOAT;
        case SG_PIXELFORMAT_RG8:            return DXGI_FORMAT_R8G8_UNORM;
        case SG_PIXELFORMAT_RG8SN:          return DXGI_FORMAT_R8G8_SNORM;
        case SG_PIXELFORMAT_RG8UI:          return DXGI_FORMAT_R8G8_UINT;
        case SG_PIXELFORMAT_RG8SI:          return DXGI_FORMAT_R8G8_SINT;
        case SG_PIXELFORMAT_R32UI:          return DXGI_FORMAT_R32_UINT;
        case SG_PIXELFORMAT_R32SI:          return DXGI_FORMAT_R32_SINT;
        case SG_PIXELFORMAT_R32F:           return DXGI_FORMAT_R32_FLOAT;
        case SG_PIXELFORMAT_RG16:           return DXGI_FORMAT_R16G16_UNORM;
        case SG_PIXELFORMAT_RG16SN:         return DXGI_FORMAT_R16G16_SNORM;
        case SG_PIXELFORMAT_RG16UI:         return DXGI_FORMAT_R16G16_UINT;
        case SG_PIXELFORMAT_RG16SI:         return DXGI_FORMAT_R16G16_SINT;
        case SG_PIXELFORMAT_RG16F:          return DXGI_FORMAT_R16G16_FLOAT;
        case SG_PIXELFORMAT_RGBA8:          return DXGI_FORMAT_R8G8B8A8_UNORM;
        case SG_PIXELFORMAT_SRGB8A8:        return DXGI_FORMAT_R8G8B8A8_UNORM_SRGB;
        case SG_PIXELFORMAT_RGBA8SN:        return DXGI_FORMAT_R8G8B8A8_SNORM;
        case SG_PIXELFORMAT_RGBA8UI:        return DXGI_FORMAT_R8G8B8A8_UINT;
        case SG_PIXELFORMAT_RGBA8SI:        return DXGI_FORMAT_R8G8B8A8_SINT;
        case SG_PIXELFORMAT_BGRA8:          return DXGI_FORMAT_B8G8R8A8_UNORM;
        case SG_PIXELFORMAT_RGB10A2:        return DXGI_FORMAT_R10G10B10A2_UNORM;
        case SG_PIXELFORMAT_RG11B10F:       return DXGI_FORMAT_R11G11B10_FLOAT;
        case SG_PIXELFORMAT_RGB9E5:         return DXGI_FORMAT_R9G9B9E5_SHAREDEXP;
        case SG_PIXELFORMAT_RG32UI:         return DXGI_FORMAT_R32G32_UINT;
        case SG_PIXELFORMAT_RG32SI:         return DXGI_FORMAT_R32G32_SINT;
        case SG_PIXELFORMAT_RG32F:          return DXGI_FORMAT_R32G32_FLOAT;
        case SG_PIXELFORMAT_RGBA16:         return DXGI_FORMAT_R16G16B16A16_UNORM;
        case SG_PIXELFORMAT_RGBA16SN:       return DXGI_FORMAT_R16G16B16A16_SNORM;
        case SG_PIXELFORMAT_RGBA16UI:       return DXGI_FORMAT_R16G16B16A16_UINT;
        case SG_PIXELFORMAT_RGBA16SI:       return DXGI_FORMAT_R16G16B16A16_SINT;
        case SG_PIXELFORMAT_RGBA16F:        return DXGI_FORMAT_R16G16B16A16_FLOAT;
        case SG_PIXELFORMAT_RGBA32UI:       return DXGI_FORMAT_R32G32B32A32_UINT;
        case SG_PIXELFORMAT_RGBA32SI:       return DXGI_FORMAT_R32G32B32A32_SINT;
        case SG_PIXELFORMAT_RGBA32F:        return DXGI_FORMAT_R32G32B32A32_FLOAT;
        case SG_PIXELFORMAT_DEPTH:          return DXGI_FORMAT_R32_TYPELESS;
        case SG_PIXELFORMAT_DEPTH_STENCIL:  return DXGI_FORMAT_R24G8_TYPELESS;
        case SG_PIXELFORMAT_BC1_RGBA:       return DXGI_FORMAT_BC1_UNORM;
        case SG_PIXELFORMAT_BC2_RGBA:       return DXGI_FORMAT_BC2_UNORM;
        case SG_PIXELFORMAT_BC3_RGBA:       return DXGI_FORMAT_BC3_UNORM;
        case SG_PIXELFORMAT_BC3_SRGBA:      return DXGI_FORMAT_BC3_UNORM_SRGB;
        case SG_PIXELFORMAT_BC4_R:          return DXGI_FORMAT_BC4_UNORM;
        case SG_PIXELFORMAT_BC4_RSN:        return DXGI_FORMAT_BC4_SNORM;
        case SG_PIXELFORMAT_BC5_RG:         return DXGI_FORMAT_BC5_UNORM;
        case SG_PIXELFORMAT_BC5_RGSN:       return DXGI_FORMAT_BC5_SNORM;
        case SG_PIXELFORMAT_BC6H_RGBF:      return DXGI_FORMAT_BC6H_SF16;
        case SG_PIXELFORMAT_BC6H_RGBUF:     return DXGI_FORMAT_BC6H_UF16;
        case SG_PIXELFORMAT_BC7_RGBA:       return DXGI_FORMAT_BC7_UNORM;
        case SG_PIXELFORMAT_BC7_SRGBA:      return DXGI_FORMAT_BC7_UNORM_SRGB;
        default:                            return DXGI_FORMAT_UNKNOWN;
    };
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_srv_pixel_format(sg_pixel_format fmt) {
    if (fmt == SG_PIXELFORMAT_DEPTH) {
        return DXGI_FORMAT_R32_FLOAT;
    } else if (fmt == SG_PIXELFORMAT_DEPTH_STENCIL) {
        return DXGI_FORMAT_R24_UNORM_X8_TYPELESS;
    } else {
        return _sg_d3d11_texture_pixel_format(fmt);
    }
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_dsv_pixel_format(sg_pixel_format fmt) {
    if (fmt == SG_PIXELFORMAT_DEPTH) {
        return DXGI_FORMAT_D32_FLOAT;
    } else if (fmt == SG_PIXELFORMAT_DEPTH_STENCIL) {
        return DXGI_FORMAT_D24_UNORM_S8_UINT;
    } else {
        return _sg_d3d11_texture_pixel_format(fmt);
    }
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_rtv_uav_pixel_format(sg_pixel_format fmt) {
    if (fmt == SG_PIXELFORMAT_DEPTH) {
        return DXGI_FORMAT_R32_FLOAT;
    } else if (fmt == SG_PIXELFORMAT_DEPTH_STENCIL) {
        return DXGI_FORMAT_R24_UNORM_X8_TYPELESS;
    } else {
        return _sg_d3d11_texture_pixel_format(fmt);
    }
}

_SOKOL_PRIVATE D3D11_PRIMITIVE_TOPOLOGY _sg_d3d11_primitive_topology(sg_primitive_type prim_type) {
    switch (prim_type) {
        case SG_PRIMITIVETYPE_POINTS:           return D3D11_PRIMITIVE_TOPOLOGY_POINTLIST;
        case SG_PRIMITIVETYPE_LINES:            return D3D11_PRIMITIVE_TOPOLOGY_LINELIST;
        case SG_PRIMITIVETYPE_LINE_STRIP:       return D3D11_PRIMITIVE_TOPOLOGY_LINESTRIP;
        case SG_PRIMITIVETYPE_TRIANGLES:        return D3D11_PRIMITIVE_TOPOLOGY_TRIANGLELIST;
        case SG_PRIMITIVETYPE_TRIANGLE_STRIP:   return D3D11_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP;
        default: SOKOL_UNREACHABLE; return (D3D11_PRIMITIVE_TOPOLOGY) 0;
    }
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_index_format(sg_index_type index_type) {
    switch (index_type) {
        case SG_INDEXTYPE_NONE:     return DXGI_FORMAT_UNKNOWN;
        case SG_INDEXTYPE_UINT16:   return DXGI_FORMAT_R16_UINT;
        case SG_INDEXTYPE_UINT32:   return DXGI_FORMAT_R32_UINT;
        default: SOKOL_UNREACHABLE; return (DXGI_FORMAT) 0;
    }
}

_SOKOL_PRIVATE D3D11_FILTER _sg_d3d11_filter(sg_filter min_f, sg_filter mag_f, sg_filter mipmap_f, bool comparison, uint32_t max_anisotropy) {
    uint32_t d3d11_filter = 0;
    if (max_anisotropy > 1) {
        // D3D11_FILTER_ANISOTROPIC = 0x55,
        d3d11_filter |= 0x55;
    } else {
        // D3D11_FILTER_MIN_MAG_MIP_POINT = 0,
        // D3D11_FILTER_MIN_MAG_POINT_MIP_LINEAR = 0x1,
        // D3D11_FILTER_MIN_POINT_MAG_LINEAR_MIP_POINT = 0x4,
        // D3D11_FILTER_MIN_POINT_MAG_MIP_LINEAR = 0x5,
        // D3D11_FILTER_MIN_LINEAR_MAG_MIP_POINT = 0x10,
        // D3D11_FILTER_MIN_LINEAR_MAG_POINT_MIP_LINEAR = 0x11,
        // D3D11_FILTER_MIN_MAG_LINEAR_MIP_POINT = 0x14,
        // D3D11_FILTER_MIN_MAG_MIP_LINEAR = 0x15,
        if (mipmap_f == SG_FILTER_LINEAR) {
            d3d11_filter |= 0x01;
        }
        if (mag_f == SG_FILTER_LINEAR) {
            d3d11_filter |= 0x04;
        }
        if (min_f == SG_FILTER_LINEAR) {
            d3d11_filter |= 0x10;
        }
    }
    // D3D11_FILTER_COMPARISON_MIN_MAG_MIP_POINT = 0x80,
    // D3D11_FILTER_COMPARISON_MIN_MAG_POINT_MIP_LINEAR = 0x81,
    // D3D11_FILTER_COMPARISON_MIN_POINT_MAG_LINEAR_MIP_POINT = 0x84,
    // D3D11_FILTER_COMPARISON_MIN_POINT_MAG_MIP_LINEAR = 0x85,
    // D3D11_FILTER_COMPARISON_MIN_LINEAR_MAG_MIP_POINT = 0x90,
    // D3D11_FILTER_COMPARISON_MIN_LINEAR_MAG_POINT_MIP_LINEAR = 0x91,
    // D3D11_FILTER_COMPARISON_MIN_MAG_LINEAR_MIP_POINT = 0x94,
    // D3D11_FILTER_COMPARISON_MIN_MAG_MIP_LINEAR = 0x95,
    // D3D11_FILTER_COMPARISON_ANISOTROPIC = 0xd5,
    if (comparison) {
        d3d11_filter |= 0x80;
    }
    return (D3D11_FILTER)d3d11_filter;
}

_SOKOL_PRIVATE D3D11_TEXTURE_ADDRESS_MODE _sg_d3d11_address_mode(sg_wrap m) {
    switch (m) {
        case SG_WRAP_REPEAT:            return D3D11_TEXTURE_ADDRESS_WRAP;
        case SG_WRAP_CLAMP_TO_EDGE:     return D3D11_TEXTURE_ADDRESS_CLAMP;
        case SG_WRAP_CLAMP_TO_BORDER:   return D3D11_TEXTURE_ADDRESS_BORDER;
        case SG_WRAP_MIRRORED_REPEAT:   return D3D11_TEXTURE_ADDRESS_MIRROR;
        default: SOKOL_UNREACHABLE; return (D3D11_TEXTURE_ADDRESS_MODE) 0;
    }
}

_SOKOL_PRIVATE DXGI_FORMAT _sg_d3d11_vertex_format(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:     return DXGI_FORMAT_R32_FLOAT;
        case SG_VERTEXFORMAT_FLOAT2:    return DXGI_FORMAT_R32G32_FLOAT;
        case SG_VERTEXFORMAT_FLOAT3:    return DXGI_FORMAT_R32G32B32_FLOAT;
        case SG_VERTEXFORMAT_FLOAT4:    return DXGI_FORMAT_R32G32B32A32_FLOAT;
        case SG_VERTEXFORMAT_INT:       return DXGI_FORMAT_R32_SINT;
        case SG_VERTEXFORMAT_INT2:      return DXGI_FORMAT_R32G32_SINT;
        case SG_VERTEXFORMAT_INT3:      return DXGI_FORMAT_R32G32B32_SINT;
        case SG_VERTEXFORMAT_INT4:      return DXGI_FORMAT_R32G32B32A32_SINT;
        case SG_VERTEXFORMAT_UINT:      return DXGI_FORMAT_R32_UINT;
        case SG_VERTEXFORMAT_UINT2:     return DXGI_FORMAT_R32G32_UINT;
        case SG_VERTEXFORMAT_UINT3:     return DXGI_FORMAT_R32G32B32_UINT;
        case SG_VERTEXFORMAT_UINT4:     return DXGI_FORMAT_R32G32B32A32_UINT;
        case SG_VERTEXFORMAT_BYTE4:     return DXGI_FORMAT_R8G8B8A8_SINT;
        case SG_VERTEXFORMAT_BYTE4N:    return DXGI_FORMAT_R8G8B8A8_SNORM;
        case SG_VERTEXFORMAT_UBYTE4:    return DXGI_FORMAT_R8G8B8A8_UINT;
        case SG_VERTEXFORMAT_UBYTE4N:   return DXGI_FORMAT_R8G8B8A8_UNORM;
        case SG_VERTEXFORMAT_SHORT2:    return DXGI_FORMAT_R16G16_SINT;
        case SG_VERTEXFORMAT_SHORT2N:   return DXGI_FORMAT_R16G16_SNORM;
        case SG_VERTEXFORMAT_USHORT2:   return DXGI_FORMAT_R16G16_UINT;
        case SG_VERTEXFORMAT_USHORT2N:  return DXGI_FORMAT_R16G16_UNORM;
        case SG_VERTEXFORMAT_SHORT4:    return DXGI_FORMAT_R16G16B16A16_SINT;
        case SG_VERTEXFORMAT_SHORT4N:   return DXGI_FORMAT_R16G16B16A16_SNORM;
        case SG_VERTEXFORMAT_USHORT4:   return DXGI_FORMAT_R16G16B16A16_UINT;
        case SG_VERTEXFORMAT_USHORT4N:  return DXGI_FORMAT_R16G16B16A16_UNORM;
        case SG_VERTEXFORMAT_UINT10_N2: return DXGI_FORMAT_R10G10B10A2_UNORM;
        case SG_VERTEXFORMAT_HALF2:     return DXGI_FORMAT_R16G16_FLOAT;
        case SG_VERTEXFORMAT_HALF4:     return DXGI_FORMAT_R16G16B16A16_FLOAT;
        default: SOKOL_UNREACHABLE; return (DXGI_FORMAT) 0;
    }
}

_SOKOL_PRIVATE D3D11_INPUT_CLASSIFICATION _sg_d3d11_input_classification(sg_vertex_step step) {
    switch (step) {
        case SG_VERTEXSTEP_PER_VERTEX:      return D3D11_INPUT_PER_VERTEX_DATA;
        case SG_VERTEXSTEP_PER_INSTANCE:    return D3D11_INPUT_PER_INSTANCE_DATA;
        default: SOKOL_UNREACHABLE; return (D3D11_INPUT_CLASSIFICATION) 0;
    }
}

_SOKOL_PRIVATE D3D11_CULL_MODE _sg_d3d11_cull_mode(sg_cull_mode m) {
    switch (m) {
        case SG_CULLMODE_NONE:      return D3D11_CULL_NONE;
        case SG_CULLMODE_FRONT:     return D3D11_CULL_FRONT;
        case SG_CULLMODE_BACK:      return D3D11_CULL_BACK;
        default: SOKOL_UNREACHABLE; return (D3D11_CULL_MODE) 0;
    }
}

_SOKOL_PRIVATE D3D11_COMPARISON_FUNC _sg_d3d11_compare_func(sg_compare_func f) {
    switch (f) {
        case SG_COMPAREFUNC_NEVER:          return D3D11_COMPARISON_NEVER;
        case SG_COMPAREFUNC_LESS:           return D3D11_COMPARISON_LESS;
        case SG_COMPAREFUNC_EQUAL:          return D3D11_COMPARISON_EQUAL;
        case SG_COMPAREFUNC_LESS_EQUAL:     return D3D11_COMPARISON_LESS_EQUAL;
        case SG_COMPAREFUNC_GREATER:        return D3D11_COMPARISON_GREATER;
        case SG_COMPAREFUNC_NOT_EQUAL:      return D3D11_COMPARISON_NOT_EQUAL;
        case SG_COMPAREFUNC_GREATER_EQUAL:  return D3D11_COMPARISON_GREATER_EQUAL;
        case SG_COMPAREFUNC_ALWAYS:         return D3D11_COMPARISON_ALWAYS;
        default: SOKOL_UNREACHABLE; return (D3D11_COMPARISON_FUNC) 0;
    }
}

_SOKOL_PRIVATE D3D11_STENCIL_OP _sg_d3d11_stencil_op(sg_stencil_op op) {
    switch (op) {
        case SG_STENCILOP_KEEP:         return D3D11_STENCIL_OP_KEEP;
        case SG_STENCILOP_ZERO:         return D3D11_STENCIL_OP_ZERO;
        case SG_STENCILOP_REPLACE:      return D3D11_STENCIL_OP_REPLACE;
        case SG_STENCILOP_INCR_CLAMP:   return D3D11_STENCIL_OP_INCR_SAT;
        case SG_STENCILOP_DECR_CLAMP:   return D3D11_STENCIL_OP_DECR_SAT;
        case SG_STENCILOP_INVERT:       return D3D11_STENCIL_OP_INVERT;
        case SG_STENCILOP_INCR_WRAP:    return D3D11_STENCIL_OP_INCR;
        case SG_STENCILOP_DECR_WRAP:    return D3D11_STENCIL_OP_DECR;
        default: SOKOL_UNREACHABLE; return (D3D11_STENCIL_OP) 0;
    }
}

_SOKOL_PRIVATE D3D11_BLEND _sg_d3d11_blend_factor(sg_blend_factor f) {
    switch (f) {
        case SG_BLENDFACTOR_ZERO:                   return D3D11_BLEND_ZERO;
        case SG_BLENDFACTOR_ONE:                    return D3D11_BLEND_ONE;
        case SG_BLENDFACTOR_SRC_COLOR:              return D3D11_BLEND_SRC_COLOR;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_COLOR:    return D3D11_BLEND_INV_SRC_COLOR;
        case SG_BLENDFACTOR_SRC_ALPHA:              return D3D11_BLEND_SRC_ALPHA;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_ALPHA:    return D3D11_BLEND_INV_SRC_ALPHA;
        case SG_BLENDFACTOR_DST_COLOR:              return D3D11_BLEND_DEST_COLOR;
        case SG_BLENDFACTOR_ONE_MINUS_DST_COLOR:    return D3D11_BLEND_INV_DEST_COLOR;
        case SG_BLENDFACTOR_DST_ALPHA:              return D3D11_BLEND_DEST_ALPHA;
        case SG_BLENDFACTOR_ONE_MINUS_DST_ALPHA:    return D3D11_BLEND_INV_DEST_ALPHA;
        case SG_BLENDFACTOR_SRC_ALPHA_SATURATED:    return D3D11_BLEND_SRC_ALPHA_SAT;
        case SG_BLENDFACTOR_BLEND_COLOR:            return D3D11_BLEND_BLEND_FACTOR;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_COLOR:  return D3D11_BLEND_INV_BLEND_FACTOR;
        case SG_BLENDFACTOR_BLEND_ALPHA:            return D3D11_BLEND_BLEND_FACTOR;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_ALPHA:  return D3D11_BLEND_INV_BLEND_FACTOR;
        default: SOKOL_UNREACHABLE; return (D3D11_BLEND) 0;
    }
}

_SOKOL_PRIVATE D3D11_BLEND_OP _sg_d3d11_blend_op(sg_blend_op op) {
    switch (op) {
        case SG_BLENDOP_ADD:                return D3D11_BLEND_OP_ADD;
        case SG_BLENDOP_SUBTRACT:           return D3D11_BLEND_OP_SUBTRACT;
        case SG_BLENDOP_REVERSE_SUBTRACT:   return D3D11_BLEND_OP_REV_SUBTRACT;
        case SG_BLENDOP_MIN:                return D3D11_BLEND_OP_MIN;
        case SG_BLENDOP_MAX:                return D3D11_BLEND_OP_MAX;
        default: SOKOL_UNREACHABLE; return (D3D11_BLEND_OP) 0;
    }
}

_SOKOL_PRIVATE UINT8 _sg_d3d11_color_write_mask(sg_color_mask m) {
    UINT8 res = 0;
    if (m & SG_COLORMASK_R) {
        res |= D3D11_COLOR_WRITE_ENABLE_RED;
    }
    if (m & SG_COLORMASK_G) {
        res |= D3D11_COLOR_WRITE_ENABLE_GREEN;
    }
    if (m & SG_COLORMASK_B) {
        res |= D3D11_COLOR_WRITE_ENABLE_BLUE;
    }
    if (m & SG_COLORMASK_A) {
        res |= D3D11_COLOR_WRITE_ENABLE_ALPHA;
    }
    return res;
}

_SOKOL_PRIVATE UINT _sg_d3d11_dxgi_fmt_caps(DXGI_FORMAT dxgi_fmt) {
    UINT dxgi_fmt_caps = 0;
    if (dxgi_fmt != DXGI_FORMAT_UNKNOWN) {
        HRESULT hr = _sg_d3d11_CheckFormatSupport(_sg.d3d11.dev, dxgi_fmt, &dxgi_fmt_caps);
        SOKOL_ASSERT(SUCCEEDED(hr) || (E_FAIL == hr));
        if (!SUCCEEDED(hr)) {
            dxgi_fmt_caps = 0;
        }
    }
    return dxgi_fmt_caps;
}

// see: https://docs.microsoft.com/en-us/windows/win32/direct3d11/overviews-direct3d-11-resources-limits#resource-limits-for-feature-level-11-hardware
_SOKOL_PRIVATE void _sg_d3d11_init_caps(void) {
    _sg.backend = SG_BACKEND_D3D11;

    _sg.features.origin_top_left = true;
    _sg.features.image_clamp_to_border = true;
    _sg.features.mrt_independent_blend_state = true;
    _sg.features.mrt_independent_write_mask = true;
    _sg.features.compute = true;
    _sg.features.msaa_image_bindings = true;

    _sg.limits.max_image_size_2d = 16 * 1024;
    _sg.limits.max_image_size_cube = 16 * 1024;
    _sg.limits.max_image_size_3d = 2 * 1024;
    _sg.limits.max_image_size_array = 16 * 1024;
    _sg.limits.max_image_array_layers = 2 * 1024;
    _sg.limits.max_vertex_attrs = SG_MAX_VERTEX_ATTRIBUTES;

    // see: https://docs.microsoft.com/en-us/windows/win32/api/d3d11/ne-d3d11-d3d11_format_support
    for (int fmt = (SG_PIXELFORMAT_NONE+1); fmt < _SG_PIXELFORMAT_NUM; fmt++) {
        const UINT srv_dxgi_fmt_caps = _sg_d3d11_dxgi_fmt_caps(_sg_d3d11_srv_pixel_format((sg_pixel_format)fmt));
        const UINT rtv_uav_dxgi_fmt_caps = _sg_d3d11_dxgi_fmt_caps(_sg_d3d11_rtv_uav_pixel_format((sg_pixel_format)fmt));
        const UINT dsv_dxgi_fmt_caps = _sg_d3d11_dxgi_fmt_caps(_sg_d3d11_dsv_pixel_format((sg_pixel_format)fmt));
        _sg_pixelformat_info_t* info = &_sg.formats[fmt];
        const bool render = 0 != (rtv_uav_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_RENDER_TARGET);
        const bool depth  = 0 != (dsv_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_DEPTH_STENCIL);
        info->sample = 0 != (srv_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_TEXTURE2D);
        info->filter = 0 != (srv_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_SHADER_SAMPLE);
        info->render = render || depth;
        if (depth) {
            info->blend = 0 != (dsv_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_BLENDABLE);
            info->msaa  = 0 != (dsv_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_MULTISAMPLE_RENDERTARGET);
        } else {
            info->blend = 0 != (rtv_uav_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_BLENDABLE);
            info->msaa  = 0 != (rtv_uav_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_MULTISAMPLE_RENDERTARGET);
        }
        info->depth = depth;
        info->read = info->write = 0 != (rtv_uav_dxgi_fmt_caps & D3D11_FORMAT_SUPPORT_TYPED_UNORDERED_ACCESS_VIEW);
    }
}

_SOKOL_PRIVATE void _sg_d3d11_setup_backend(const sg_desc* desc) {
    // assume _sg.d3d11 already is zero-initialized
    SOKOL_ASSERT(desc);
    SOKOL_ASSERT(desc->environment.d3d11.device);
    SOKOL_ASSERT(desc->environment.d3d11.device_context);
    _sg.d3d11.valid = true;
    _sg.d3d11.dev = (ID3D11Device*) desc->environment.d3d11.device;
    _sg.d3d11.ctx = (ID3D11DeviceContext*) desc->environment.d3d11.device_context;
    _sg_d3d11_init_caps();
}

_SOKOL_PRIVATE void _sg_d3d11_discard_backend(void) {
    SOKOL_ASSERT(_sg.d3d11.valid);
    _sg.d3d11.valid = false;
}

_SOKOL_PRIVATE void _sg_d3d11_clear_state(void) {
    // clear all the device context state, so that resource refs don't keep stuck in the d3d device context
    _sg_d3d11_ClearState(_sg.d3d11.ctx);
}

_SOKOL_PRIVATE void _sg_d3d11_reset_state_cache(void) {
    // there's currently no state cache in the D3D11 backend, so this is a no-op
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && desc);
    SOKOL_ASSERT(!buf->d3d11.buf);
    const bool injected = (0 != desc->d3d11_buffer);
    if (injected) {
        buf->d3d11.buf = (ID3D11Buffer*) desc->d3d11_buffer;
        _sg_d3d11_AddRef(buf->d3d11.buf);
    } else {
        D3D11_BUFFER_DESC d3d11_buf_desc;
        _sg_clear(&d3d11_buf_desc, sizeof(d3d11_buf_desc));
        d3d11_buf_desc.ByteWidth = (UINT)buf->cmn.size;
        d3d11_buf_desc.Usage = _sg_d3d11_buffer_usage(&buf->cmn.usage);
        d3d11_buf_desc.BindFlags = _sg_d3d11_buffer_bind_flags(&buf->cmn.usage);
        d3d11_buf_desc.CPUAccessFlags = _sg_d3d11_buffer_cpu_access_flags(&buf->cmn.usage);
        d3d11_buf_desc.MiscFlags = _sg_d3d11_buffer_misc_flags(&buf->cmn.usage);
        D3D11_SUBRESOURCE_DATA* init_data_ptr = 0;
        D3D11_SUBRESOURCE_DATA init_data;
        _sg_clear(&init_data, sizeof(init_data));
        if (desc->data.ptr) {
            init_data.pSysMem = desc->data.ptr;
            init_data_ptr = &init_data;
        }
        HRESULT hr = _sg_d3d11_CreateBuffer(_sg.d3d11.dev, &d3d11_buf_desc, init_data_ptr, &buf->d3d11.buf);
        if (!(SUCCEEDED(hr) && buf->d3d11.buf)) {
            _SG_ERROR(D3D11_CREATE_BUFFER_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }

        // for storage buffers need to create a shader-resource-view
        // for read-only access, and an unordered-access-view for
        // read-write access
        if (buf->cmn.usage.storage_buffer) {
            SOKOL_ASSERT(_sg_multiple_u64((uint64_t)buf->cmn.size, 4));
            D3D11_SHADER_RESOURCE_VIEW_DESC d3d11_srv_desc;
            _sg_clear(&d3d11_srv_desc, sizeof(d3d11_srv_desc));
            d3d11_srv_desc.Format = DXGI_FORMAT_R32_TYPELESS;
            d3d11_srv_desc.ViewDimension = D3D11_SRV_DIMENSION_BUFFEREX;
            d3d11_srv_desc.BufferEx.FirstElement = 0;
            d3d11_srv_desc.BufferEx.NumElements = ((UINT)buf->cmn.size) / 4;
            d3d11_srv_desc.BufferEx.Flags = D3D11_BUFFEREX_SRV_FLAG_RAW;
            hr = _sg_d3d11_CreateShaderResourceView(_sg.d3d11.dev, (ID3D11Resource*)buf->d3d11.buf, &d3d11_srv_desc, &buf->d3d11.srv);
            if (!(SUCCEEDED(hr) && buf->d3d11.srv)) {
                _SG_ERROR(D3D11_CREATE_BUFFER_SRV_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
            if (buf->cmn.usage.immutable) {
                D3D11_UNORDERED_ACCESS_VIEW_DESC d3d11_uav_desc;
                _sg_clear(&d3d11_uav_desc, sizeof(d3d11_uav_desc));
                d3d11_uav_desc.Format = DXGI_FORMAT_R32_TYPELESS;
                d3d11_uav_desc.ViewDimension = D3D11_UAV_DIMENSION_BUFFER;
                d3d11_uav_desc.Buffer.FirstElement = 0;
                d3d11_uav_desc.Buffer.NumElements = ((UINT)buf->cmn.size) / 4;
                d3d11_uav_desc.Buffer.Flags = D3D11_BUFFER_UAV_FLAG_RAW;
                hr = _sg_d3d11_CreateUnorderedAccessView(_sg.d3d11.dev, (ID3D11Resource*)buf->d3d11.buf, &d3d11_uav_desc, &buf->d3d11.uav);
                if (!(SUCCEEDED(hr) && buf->d3d11.uav)) {
                    _SG_ERROR(D3D11_CREATE_BUFFER_UAV_FAILED);
                    return SG_RESOURCESTATE_FAILED;
                }
            }
        }
        _sg_d3d11_setlabel(buf->d3d11.buf, desc->label);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_d3d11_discard_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    if (buf->d3d11.buf) {
        _sg_d3d11_Release(buf->d3d11.buf);
    }
    if (buf->d3d11.srv) {
        _sg_d3d11_Release(buf->d3d11.srv);
    }
    if (buf->d3d11.uav) {
        _sg_d3d11_Release(buf->d3d11.uav);
    }
}

_SOKOL_PRIVATE void _sg_d3d11_fill_subres_data(const _sg_image_t* img, const sg_image_data* data) {
    const int num_faces = (img->cmn.type == SG_IMAGETYPE_CUBE) ? 6:1;
    const int num_slices = (img->cmn.type == SG_IMAGETYPE_ARRAY) ? img->cmn.num_slices:1;
    int subres_index = 0;
    for (int face_index = 0; face_index < num_faces; face_index++) {
        for (int slice_index = 0; slice_index < num_slices; slice_index++) {
            for (int mip_index = 0; mip_index < img->cmn.num_mipmaps; mip_index++, subres_index++) {
                SOKOL_ASSERT(subres_index < (SG_MAX_MIPMAPS * SG_MAX_TEXTUREARRAY_LAYERS));
                D3D11_SUBRESOURCE_DATA* subres_data = &_sg.d3d11.subres_data[subres_index];
                const int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
                const int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
                const sg_range* subimg_data = &(data->subimage[face_index][mip_index]);
                const size_t slice_size = subimg_data->size / (size_t)num_slices;
                const size_t slice_offset = slice_size * (size_t)slice_index;
                const uint8_t* ptr = (const uint8_t*) subimg_data->ptr;
                subres_data->pSysMem = ptr + slice_offset;
                subres_data->SysMemPitch = (UINT)_sg_row_pitch(img->cmn.pixel_format, mip_width, 1);
                if (img->cmn.type == SG_IMAGETYPE_3D) {
                    // FIXME? const int mip_depth = _sg_miplevel_dim(img->depth, mip_index);
                    subres_data->SysMemSlicePitch = (UINT)_sg_surface_pitch(img->cmn.pixel_format, mip_width, mip_height, 1);
                } else {
                    subres_data->SysMemSlicePitch = 0;
                }
            }
        }
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && desc);
    SOKOL_ASSERT((0 == img->d3d11.tex2d) && (0 == img->d3d11.tex3d) && (0 == img->d3d11.res) && (0 == img->d3d11.srv));
    HRESULT hr;

    const bool injected = (0 != desc->d3d11_texture);
    const bool msaa = (img->cmn.sample_count > 1);
    SOKOL_ASSERT(!(msaa && (img->cmn.type == SG_IMAGETYPE_CUBE)));
    img->d3d11.format = _sg_d3d11_texture_pixel_format(img->cmn.pixel_format);
    if (img->d3d11.format == DXGI_FORMAT_UNKNOWN) {
        _SG_ERROR(D3D11_CREATE_2D_TEXTURE_UNSUPPORTED_PIXEL_FORMAT);
        return SG_RESOURCESTATE_FAILED;
    }

    // prepare initial content pointers
    D3D11_SUBRESOURCE_DATA* init_data = 0;
    if (!injected && desc->data.subimage[0][0].ptr) {
        _sg_d3d11_fill_subres_data(img, &desc->data);
        init_data = _sg.d3d11.subres_data;
    }
    if (img->cmn.type != SG_IMAGETYPE_3D) {
        // 2D-, cube- or array-texture
        // first check for injected texture and/or resource view
        if (injected) {
            img->d3d11.tex2d = (ID3D11Texture2D*) desc->d3d11_texture;
            _sg_d3d11_AddRef(img->d3d11.tex2d);
            img->d3d11.srv = (ID3D11ShaderResourceView*) desc->d3d11_shader_resource_view;
            if (img->d3d11.srv) {
                _sg_d3d11_AddRef(img->d3d11.srv);
            }
        } else {
            // if not injected, create 2D texture
            D3D11_TEXTURE2D_DESC d3d11_tex_desc;
            _sg_clear(&d3d11_tex_desc, sizeof(d3d11_tex_desc));
            d3d11_tex_desc.Width = (UINT)img->cmn.width;
            d3d11_tex_desc.Height = (UINT)img->cmn.height;
            d3d11_tex_desc.MipLevels = (UINT)img->cmn.num_mipmaps;
            switch (img->cmn.type) {
                case SG_IMAGETYPE_ARRAY:    d3d11_tex_desc.ArraySize = (UINT)img->cmn.num_slices; break;
                case SG_IMAGETYPE_CUBE:     d3d11_tex_desc.ArraySize = 6; break;
                default:                    d3d11_tex_desc.ArraySize = 1; break;
            }
            d3d11_tex_desc.Format = img->d3d11.format;
            d3d11_tex_desc.BindFlags = _sg_d3d11_image_bind_flags(&img->cmn.usage, img->cmn.pixel_format);
            d3d11_tex_desc.Usage = _sg_d3d11_image_usage(&img->cmn.usage);
            d3d11_tex_desc.CPUAccessFlags = _sg_d3d11_image_cpu_access_flags(&img->cmn.usage);
            d3d11_tex_desc.SampleDesc.Count = (UINT)img->cmn.sample_count;
            d3d11_tex_desc.SampleDesc.Quality = (UINT) (msaa ? D3D11_STANDARD_MULTISAMPLE_PATTERN : 0);
            d3d11_tex_desc.MiscFlags = (img->cmn.type == SG_IMAGETYPE_CUBE) ? D3D11_RESOURCE_MISC_TEXTURECUBE : 0;
            hr = _sg_d3d11_CreateTexture2D(_sg.d3d11.dev, &d3d11_tex_desc, init_data, &img->d3d11.tex2d);
            if (!(SUCCEEDED(hr) && img->d3d11.tex2d)) {
                _SG_ERROR(D3D11_CREATE_2D_TEXTURE_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
            _sg_d3d11_setlabel(img->d3d11.tex2d, desc->label);

            // create shader-resource-view for 2D texture
            D3D11_SHADER_RESOURCE_VIEW_DESC d3d11_srv_desc;
            _sg_clear(&d3d11_srv_desc, sizeof(d3d11_srv_desc));
            d3d11_srv_desc.Format = _sg_d3d11_srv_pixel_format(img->cmn.pixel_format);
            switch (img->cmn.type) {
                case SG_IMAGETYPE_2D:
                    d3d11_srv_desc.ViewDimension = msaa ? D3D11_SRV_DIMENSION_TEXTURE2DMS : D3D11_SRV_DIMENSION_TEXTURE2D;
                    d3d11_srv_desc.Texture2D.MipLevels = (UINT)img->cmn.num_mipmaps;
                    break;
                case SG_IMAGETYPE_CUBE:
                    d3d11_srv_desc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURECUBE;
                    d3d11_srv_desc.TextureCube.MipLevels = (UINT)img->cmn.num_mipmaps;
                    break;
                case SG_IMAGETYPE_ARRAY:
                    d3d11_srv_desc.ViewDimension = msaa ? D3D11_SRV_DIMENSION_TEXTURE2DMSARRAY : D3D11_SRV_DIMENSION_TEXTURE2DARRAY;
                    d3d11_srv_desc.Texture2DArray.MipLevels = (UINT)img->cmn.num_mipmaps;
                    d3d11_srv_desc.Texture2DArray.ArraySize = (UINT)img->cmn.num_slices;
                    break;
                default:
                    SOKOL_UNREACHABLE; break;
            }
            hr = _sg_d3d11_CreateShaderResourceView(_sg.d3d11.dev, (ID3D11Resource*)img->d3d11.tex2d, &d3d11_srv_desc, &img->d3d11.srv);
            if (!(SUCCEEDED(hr) && img->d3d11.srv)) {
                _SG_ERROR(D3D11_CREATE_2D_SRV_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
            _sg_d3d11_setlabel(img->d3d11.srv, desc->label);
        }
        SOKOL_ASSERT(img->d3d11.tex2d);
        img->d3d11.res = (ID3D11Resource*)img->d3d11.tex2d;
        _sg_d3d11_AddRef(img->d3d11.res);
    } else {
        // 3D texture - same procedure, first check if injected, than create non-injected
        if (injected) {
            img->d3d11.tex3d = (ID3D11Texture3D*) desc->d3d11_texture;
            _sg_d3d11_AddRef(img->d3d11.tex3d);
            img->d3d11.srv = (ID3D11ShaderResourceView*) desc->d3d11_shader_resource_view;
            if (img->d3d11.srv) {
                _sg_d3d11_AddRef(img->d3d11.srv);
            }
        } else {
            // not injected, create 3d texture
            D3D11_TEXTURE3D_DESC d3d11_tex_desc;
            _sg_clear(&d3d11_tex_desc, sizeof(d3d11_tex_desc));
            d3d11_tex_desc.Width = (UINT)img->cmn.width;
            d3d11_tex_desc.Height = (UINT)img->cmn.height;
            d3d11_tex_desc.Depth = (UINT)img->cmn.num_slices;
            d3d11_tex_desc.MipLevels = (UINT)img->cmn.num_mipmaps;
            d3d11_tex_desc.Format = img->d3d11.format;
            d3d11_tex_desc.BindFlags = _sg_d3d11_image_bind_flags(&img->cmn.usage, img->cmn.pixel_format);
            d3d11_tex_desc.Usage = _sg_d3d11_image_usage(&img->cmn.usage);
            d3d11_tex_desc.CPUAccessFlags = _sg_d3d11_image_cpu_access_flags(&img->cmn.usage);
            if (img->d3d11.format == DXGI_FORMAT_UNKNOWN) {
                _SG_ERROR(D3D11_CREATE_3D_TEXTURE_UNSUPPORTED_PIXEL_FORMAT);
                return SG_RESOURCESTATE_FAILED;
            }
            hr = _sg_d3d11_CreateTexture3D(_sg.d3d11.dev, &d3d11_tex_desc, init_data, &img->d3d11.tex3d);
            if (!(SUCCEEDED(hr) && img->d3d11.tex3d)) {
                _SG_ERROR(D3D11_CREATE_3D_TEXTURE_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
            _sg_d3d11_setlabel(img->d3d11.tex3d, desc->label);

            // create shader-resource-view for 3D texture
            if (!msaa) {
                D3D11_SHADER_RESOURCE_VIEW_DESC d3d11_srv_desc;
                _sg_clear(&d3d11_srv_desc, sizeof(d3d11_srv_desc));
                d3d11_srv_desc.Format = _sg_d3d11_srv_pixel_format(img->cmn.pixel_format);
                d3d11_srv_desc.ViewDimension = D3D11_SRV_DIMENSION_TEXTURE3D;
                d3d11_srv_desc.Texture3D.MipLevels = (UINT)img->cmn.num_mipmaps;
                hr = _sg_d3d11_CreateShaderResourceView(_sg.d3d11.dev, (ID3D11Resource*)img->d3d11.tex3d, &d3d11_srv_desc, &img->d3d11.srv);
                if (!(SUCCEEDED(hr) && img->d3d11.srv)) {
                    _SG_ERROR(D3D11_CREATE_3D_SRV_FAILED);
                    return SG_RESOURCESTATE_FAILED;
                }
                _sg_d3d11_setlabel(img->d3d11.srv, desc->label);
            }
        }
        SOKOL_ASSERT(img->d3d11.tex3d);
        img->d3d11.res = (ID3D11Resource*)img->d3d11.tex3d;
        _sg_d3d11_AddRef(img->d3d11.res);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_d3d11_discard_image(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    if (img->d3d11.tex2d) {
        _sg_d3d11_Release(img->d3d11.tex2d);
    }
    if (img->d3d11.tex3d) {
        _sg_d3d11_Release(img->d3d11.tex3d);
    }
    if (img->d3d11.res) {
        _sg_d3d11_Release(img->d3d11.res);
    }
    if (img->d3d11.srv) {
        _sg_d3d11_Release(img->d3d11.srv);
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && desc);
    SOKOL_ASSERT(0 == smp->d3d11.smp);
    const bool injected = (0 != desc->d3d11_sampler);
    if (injected) {
        smp->d3d11.smp = (ID3D11SamplerState*)desc->d3d11_sampler;
        _sg_d3d11_AddRef(smp->d3d11.smp);
    } else {
        D3D11_SAMPLER_DESC d3d11_smp_desc;
        _sg_clear(&d3d11_smp_desc, sizeof(d3d11_smp_desc));
        d3d11_smp_desc.Filter = _sg_d3d11_filter(desc->min_filter, desc->mag_filter, desc->mipmap_filter, desc->compare != SG_COMPAREFUNC_NEVER, desc->max_anisotropy);
        d3d11_smp_desc.AddressU = _sg_d3d11_address_mode(desc->wrap_u);
        d3d11_smp_desc.AddressV = _sg_d3d11_address_mode(desc->wrap_v);
        d3d11_smp_desc.AddressW = _sg_d3d11_address_mode(desc->wrap_w);
        d3d11_smp_desc.MipLODBias = 0.0f; // FIXME?
        switch (desc->border_color) {
            case SG_BORDERCOLOR_TRANSPARENT_BLACK:
                // all 0.0f
                break;
            case SG_BORDERCOLOR_OPAQUE_WHITE:
                for (int i = 0; i < 4; i++) {
                    d3d11_smp_desc.BorderColor[i] = 1.0f;
                }
                break;
            default:
                // opaque black
                d3d11_smp_desc.BorderColor[3] = 1.0f;
                break;
        }
        d3d11_smp_desc.MaxAnisotropy = desc->max_anisotropy;
        d3d11_smp_desc.ComparisonFunc = _sg_d3d11_compare_func(desc->compare);
        d3d11_smp_desc.MinLOD = desc->min_lod;
        d3d11_smp_desc.MaxLOD = desc->max_lod;
        HRESULT hr = _sg_d3d11_CreateSamplerState(_sg.d3d11.dev, &d3d11_smp_desc, &smp->d3d11.smp);
        if (!(SUCCEEDED(hr) && smp->d3d11.smp)) {
            _SG_ERROR(D3D11_CREATE_SAMPLER_STATE_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(smp->d3d11.smp, desc->label);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_d3d11_discard_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    if (smp->d3d11.smp) {
        _sg_d3d11_Release(smp->d3d11.smp);
    }
}

_SOKOL_PRIVATE bool _sg_d3d11_load_d3dcompiler_dll(void) {
    if ((0 == _sg.d3d11.d3dcompiler_dll) && !_sg.d3d11.d3dcompiler_dll_load_failed) {
        _sg.d3d11.d3dcompiler_dll = LoadLibraryA("d3dcompiler_47.dll");
        if (0 == _sg.d3d11.d3dcompiler_dll) {
            // don't attempt to load missing DLL in the future
            _SG_ERROR(D3D11_LOAD_D3DCOMPILER_47_DLL_FAILED);
            _sg.d3d11.d3dcompiler_dll_load_failed = true;
            return false;
        }
        // look up function pointers
        _sg.d3d11.D3DCompile_func = (pD3DCompile)(void*) GetProcAddress(_sg.d3d11.d3dcompiler_dll, "D3DCompile");
        SOKOL_ASSERT(_sg.d3d11.D3DCompile_func);
    }
    return 0 != _sg.d3d11.d3dcompiler_dll;
}

_SOKOL_PRIVATE ID3DBlob* _sg_d3d11_compile_shader(const sg_shader_function* shd_func) {
    if (!_sg_d3d11_load_d3dcompiler_dll()) {
        return NULL;
    }
    SOKOL_ASSERT(shd_func->d3d11_target);
    UINT flags1 = D3DCOMPILE_PACK_MATRIX_COLUMN_MAJOR;
    if (_sg.desc.d3d11_shader_debugging) {
        flags1 |= D3DCOMPILE_DEBUG | D3DCOMPILE_SKIP_OPTIMIZATION;
    } else {
        flags1 |= D3DCOMPILE_OPTIMIZATION_LEVEL3;
    }
    ID3DBlob* output = NULL;
    ID3DBlob* errors_or_warnings = NULL;
    HRESULT hr = _sg.d3d11.D3DCompile_func(
        shd_func->source,               // pSrcData
        strlen(shd_func->source),       // SrcDataSize
        NULL,                           // pSourceName
        NULL,                           // pDefines
        NULL,                           // pInclude
        shd_func->entry ? shd_func->entry : "main", // pEntryPoint
        shd_func->d3d11_target,         // pTarget
        flags1,     // Flags1
        0,          // Flags2
        &output,    // ppCode
        &errors_or_warnings);   // ppErrorMsgs
    if (FAILED(hr)) {
        _SG_ERROR(D3D11_SHADER_COMPILATION_FAILED);
    }
    if (errors_or_warnings) {
        _SG_WARN(D3D11_SHADER_COMPILATION_OUTPUT);
        _SG_LOGMSG(D3D11_SHADER_COMPILATION_OUTPUT, (LPCSTR)_sg_d3d11_GetBufferPointer(errors_or_warnings));
        _sg_d3d11_Release(errors_or_warnings); errors_or_warnings = NULL;
    }
    if (FAILED(hr)) {
        // just in case, usually output is NULL here
        if (output) {
            _sg_d3d11_Release(output);
            output = NULL;
        }
    }
    return output;
}

// NOTE: this is an out-of-range check for HLSL bindslots that's also active in release mode
_SOKOL_PRIVATE bool _sg_d3d11_ensure_hlsl_bindslot_ranges(const sg_shader_desc* desc) {
    SOKOL_ASSERT(desc);
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (desc->uniform_blocks[i].hlsl_register_b_n >= _SG_D3D11_MAX_STAGE_UB_BINDINGS) {
            _SG_ERROR(D3D11_UNIFORMBLOCK_HLSL_REGISTER_B_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (desc->storage_buffers[i].hlsl_register_t_n >= _SG_D3D11_MAX_STAGE_SRV_BINDINGS) {
            _SG_ERROR(D3D11_STORAGEBUFFER_HLSL_REGISTER_T_OUT_OF_RANGE);
            return false;
        }
        if (desc->storage_buffers[i].hlsl_register_u_n >= _SG_D3D11_MAX_STAGE_UAV_BINDINGS) {
            _SG_ERROR(D3D11_STORAGEBUFFER_HLSL_REGISTER_U_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (desc->images[i].hlsl_register_t_n >= _SG_D3D11_MAX_STAGE_SRV_BINDINGS) {
            _SG_ERROR(D3D11_IMAGE_HLSL_REGISTER_T_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (desc->samplers[i].hlsl_register_s_n >= _SG_D3D11_MAX_STAGE_SMP_BINDINGS) {
            _SG_ERROR(D3D11_SAMPLER_HLSL_REGISTER_S_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (desc->storage_images[i].hlsl_register_u_n >= _SG_D3D11_MAX_STAGE_UAV_BINDINGS) {
            _SG_ERROR(D3D11_STORAGEIMAGE_HLSL_REGISTER_U_OUT_OF_RANGE);
            return false;
        }
    }
    return true;
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && desc);
    SOKOL_ASSERT(!shd->d3d11.vs && !shd->d3d11.fs && !shd->d3d11.cs && !shd->d3d11.vs_blob);
    HRESULT hr;

    // perform a range-check on HLSL bindslots that's also active in release
    // mode to avoid potential out-of-bounds array accesses
    if (!_sg_d3d11_ensure_hlsl_bindslot_ranges(desc)) {
        return SG_RESOURCESTATE_FAILED;
    }

    // copy vertex attribute semantic names and indices
    for (size_t i = 0; i < SG_MAX_VERTEX_ATTRIBUTES; i++) {
        _sg_strcpy(&shd->d3d11.attrs[i].sem_name, desc->attrs[i].hlsl_sem_name);
        shd->d3d11.attrs[i].sem_index = desc->attrs[i].hlsl_sem_index;
    }

    // copy HLSL bind slots
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        shd->d3d11.ub_register_b_n[i] = desc->uniform_blocks[i].hlsl_register_b_n;
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        shd->d3d11.sbuf_register_t_n[i] = desc->storage_buffers[i].hlsl_register_t_n;
        shd->d3d11.sbuf_register_u_n[i] = desc->storage_buffers[i].hlsl_register_u_n;
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        shd->d3d11.img_register_t_n[i] = desc->images[i].hlsl_register_t_n;
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        shd->d3d11.smp_register_s_n[i] = desc->samplers[i].hlsl_register_s_n;
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        shd->d3d11.simg_register_u_n[i] = desc->storage_images[i].hlsl_register_u_n;
    }

    // create a D3D constant buffer for each uniform block
    for (size_t ub_index = 0; ub_index < SG_MAX_UNIFORMBLOCK_BINDSLOTS; ub_index++) {
        const sg_shader_stage stage = desc->uniform_blocks[ub_index].stage;
        if (stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        const _sg_shader_uniform_block_t* ub = &shd->cmn.uniform_blocks[ub_index];
        ID3D11Buffer* cbuf = 0;
        D3D11_BUFFER_DESC cb_desc;
        _sg_clear(&cb_desc, sizeof(cb_desc));
        cb_desc.ByteWidth = (UINT)_sg_roundup((int)ub->size, 16);
        cb_desc.Usage = D3D11_USAGE_DEFAULT;
        cb_desc.BindFlags = D3D11_BIND_CONSTANT_BUFFER;
        hr = _sg_d3d11_CreateBuffer(_sg.d3d11.dev, &cb_desc, NULL, &cbuf);
        if (!(SUCCEEDED(hr) && cbuf)) {
            _SG_ERROR(D3D11_CREATE_CONSTANT_BUFFER_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(cbuf, desc->label);
        shd->d3d11.all_cbufs[ub_index] = cbuf;

        const uint8_t d3d11_slot = shd->d3d11.ub_register_b_n[ub_index];
        SOKOL_ASSERT(d3d11_slot < _SG_D3D11_MAX_STAGE_UB_BINDINGS);
        if (stage == SG_SHADERSTAGE_VERTEX) {
            SOKOL_ASSERT(0 == shd->d3d11.vs_cbufs[d3d11_slot]);
            shd->d3d11.vs_cbufs[d3d11_slot] = cbuf;
        } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
            SOKOL_ASSERT(0 == shd->d3d11.fs_cbufs[d3d11_slot]);
            shd->d3d11.fs_cbufs[d3d11_slot] = cbuf;
        } else if (stage == SG_SHADERSTAGE_COMPUTE) {
            SOKOL_ASSERT(0 == shd->d3d11.cs_cbufs[d3d11_slot]);
            shd->d3d11.cs_cbufs[d3d11_slot] = cbuf;
        } else {
            SOKOL_UNREACHABLE;
        }
    }

    // create shader functions
    const bool has_vs = desc->vertex_func.bytecode.ptr || desc->vertex_func.source;
    const bool has_fs = desc->fragment_func.bytecode.ptr || desc->fragment_func.source;
    const bool has_cs = desc->compute_func.bytecode.ptr || desc->compute_func.source;
    bool vs_valid = false; bool fs_valid = false; bool cs_valid = false;
    if (has_vs) {
        const void* vs_ptr = 0; SIZE_T vs_length = 0;
        ID3DBlob* vs_blob = 0;
        if (desc->vertex_func.bytecode.ptr) {
            SOKOL_ASSERT(desc->vertex_func.bytecode.size > 0);
            vs_ptr = desc->vertex_func.bytecode.ptr;
            vs_length = desc->vertex_func.bytecode.size;
        } else {
            SOKOL_ASSERT(desc->vertex_func.source);
            vs_blob = _sg_d3d11_compile_shader(&desc->vertex_func);
            if (vs_blob) {
                vs_ptr = _sg_d3d11_GetBufferPointer(vs_blob);
                vs_length = _sg_d3d11_GetBufferSize(vs_blob);
            }
        }
        if (vs_ptr && (vs_length > 0)) {
            hr = _sg_d3d11_CreateVertexShader(_sg.d3d11.dev, vs_ptr, vs_length, NULL, &shd->d3d11.vs);
            vs_valid = SUCCEEDED(hr) && shd->d3d11.vs;
        }
        // set label, and need to store a copy of the vertex shader blob for the pipeline creation
        if (vs_valid) {
            _sg_d3d11_setlabel(shd->d3d11.vs, desc->label);
            shd->d3d11.vs_blob_length = vs_length;
            shd->d3d11.vs_blob = _sg_malloc((size_t)vs_length);
            SOKOL_ASSERT(shd->d3d11.vs_blob);
            memcpy(shd->d3d11.vs_blob, vs_ptr, vs_length);
        }
        if (vs_blob) {
            _sg_d3d11_Release(vs_blob);
        }
    }
    if (has_fs) {
        const void* fs_ptr = 0; SIZE_T fs_length = 0;
        ID3DBlob* fs_blob = 0;
        if (desc->fragment_func.bytecode.ptr) {
            SOKOL_ASSERT(desc->fragment_func.bytecode.size > 0);
            fs_ptr = desc->fragment_func.bytecode.ptr;
            fs_length = desc->fragment_func.bytecode.size;
        } else {
            SOKOL_ASSERT(desc->fragment_func.source);
            fs_blob = _sg_d3d11_compile_shader(&desc->fragment_func);
            if (fs_blob) {
                fs_ptr = _sg_d3d11_GetBufferPointer(fs_blob);
                fs_length = _sg_d3d11_GetBufferSize(fs_blob);
            }
        }
        if (fs_ptr && (fs_length > 0)) {
            hr = _sg_d3d11_CreatePixelShader(_sg.d3d11.dev, fs_ptr, fs_length, NULL, &shd->d3d11.fs);
            fs_valid = SUCCEEDED(hr) && shd->d3d11.fs;
        }
        if (fs_valid) {
            _sg_d3d11_setlabel(shd->d3d11.fs, desc->label);
        }
        if (fs_blob) {
            _sg_d3d11_Release(fs_blob);
        }
    }
    if (has_cs) {
        const void* cs_ptr = 0; SIZE_T cs_length = 0;
        ID3DBlob* cs_blob = 0;
        if (desc->compute_func.bytecode.ptr) {
            SOKOL_ASSERT(desc->compute_func.bytecode.size > 0);
            cs_ptr = desc->compute_func.bytecode.ptr;
            cs_length = desc->compute_func.bytecode.size;
        } else {
            SOKOL_ASSERT(desc->compute_func.source);
            cs_blob = _sg_d3d11_compile_shader(&desc->compute_func);
            if (cs_blob) {
                cs_ptr = _sg_d3d11_GetBufferPointer(cs_blob);
                cs_length = _sg_d3d11_GetBufferSize(cs_blob);
            }
        }
        if (cs_ptr && (cs_length > 0)) {
            hr = _sg_d3d11_CreateComputeShader(_sg.d3d11.dev, cs_ptr, cs_length, NULL, &shd->d3d11.cs);
            cs_valid = SUCCEEDED(hr) && shd->d3d11.cs;
        }
        if (cs_blob) {
            _sg_d3d11_Release(cs_blob);
        }
    }
    if ((vs_valid && fs_valid) || cs_valid) {
        return SG_RESOURCESTATE_VALID;
    } else {
        return SG_RESOURCESTATE_FAILED;
    }
}

_SOKOL_PRIVATE void _sg_d3d11_discard_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    if (shd->d3d11.vs) {
        _sg_d3d11_Release(shd->d3d11.vs);
    }
    if (shd->d3d11.fs) {
        _sg_d3d11_Release(shd->d3d11.fs);
    }
    if (shd->d3d11.cs) {
        _sg_d3d11_Release(shd->d3d11.cs);
    }
    if (shd->d3d11.vs_blob) {
        _sg_free(shd->d3d11.vs_blob);
    }
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (shd->d3d11.all_cbufs[i]) {
            _sg_d3d11_Release(shd->d3d11.all_cbufs[i]);
        }
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_pipeline(_sg_pipeline_t* pip, _sg_shader_t* shd, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && shd && desc);
    SOKOL_ASSERT(desc->shader.id == shd->slot.id);
    SOKOL_ASSERT(shd->slot.state == SG_RESOURCESTATE_VALID);

    pip->shader = shd;

    // if this is a compute pipeline, we're done here
    if (pip->cmn.is_compute) {
        return SG_RESOURCESTATE_VALID;
    }

    // a render pipeline...
    SOKOL_ASSERT(shd->d3d11.vs_blob && shd->d3d11.vs_blob_length > 0);
    SOKOL_ASSERT(!pip->d3d11.il && !pip->d3d11.rs && !pip->d3d11.dss && !pip->d3d11.bs);

    pip->d3d11.index_format = _sg_d3d11_index_format(pip->cmn.index_type);
    pip->d3d11.topology = _sg_d3d11_primitive_topology(desc->primitive_type);
    pip->d3d11.stencil_ref = desc->stencil.ref;

    // create input layout object
    HRESULT hr;
    D3D11_INPUT_ELEMENT_DESC d3d11_comps[SG_MAX_VERTEX_ATTRIBUTES];
    _sg_clear(d3d11_comps, sizeof(d3d11_comps));
    size_t attr_index = 0;
    for (; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
        const sg_vertex_attr_state* a_state = &desc->layout.attrs[attr_index];
        if (a_state->format == SG_VERTEXFORMAT_INVALID) {
            break;
        }
        SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
        SOKOL_ASSERT(pip->cmn.vertex_buffer_layout_active[a_state->buffer_index]);
        const sg_vertex_buffer_layout_state* l_state = &desc->layout.buffers[a_state->buffer_index];
        const sg_vertex_step step_func = l_state->step_func;
        const int step_rate = l_state->step_rate;
        D3D11_INPUT_ELEMENT_DESC* d3d11_comp = &d3d11_comps[attr_index];
        d3d11_comp->SemanticName = _sg_strptr(&shd->d3d11.attrs[attr_index].sem_name);
        d3d11_comp->SemanticIndex = (UINT)shd->d3d11.attrs[attr_index].sem_index;
        d3d11_comp->Format = _sg_d3d11_vertex_format(a_state->format);
        d3d11_comp->InputSlot = (UINT)a_state->buffer_index;
        d3d11_comp->AlignedByteOffset = (UINT)a_state->offset;
        d3d11_comp->InputSlotClass = _sg_d3d11_input_classification(step_func);
        if (SG_VERTEXSTEP_PER_INSTANCE == step_func) {
            d3d11_comp->InstanceDataStepRate = (UINT)step_rate;
            pip->cmn.use_instanced_draw = true;
        }
    }
    for (size_t layout_index = 0; layout_index < SG_MAX_VERTEXBUFFER_BINDSLOTS; layout_index++) {
        if (pip->cmn.vertex_buffer_layout_active[layout_index]) {
            const sg_vertex_buffer_layout_state* l_state = &desc->layout.buffers[layout_index];
            SOKOL_ASSERT(l_state->stride > 0);
            pip->d3d11.vb_strides[layout_index] = (UINT)l_state->stride;
        } else {
            pip->d3d11.vb_strides[layout_index] = 0;
        }
    }
    if (attr_index > 0) {
        hr = _sg_d3d11_CreateInputLayout(_sg.d3d11.dev,
            d3d11_comps,                // pInputElementDesc
            (UINT)attr_index,           // NumElements
            shd->d3d11.vs_blob,         // pShaderByteCodeWithInputSignature
            shd->d3d11.vs_blob_length,  // BytecodeLength
            &pip->d3d11.il);
        if (!(SUCCEEDED(hr) && pip->d3d11.il)) {
            _SG_ERROR(D3D11_CREATE_INPUT_LAYOUT_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(pip->d3d11.il, desc->label);
    }

    // create rasterizer state
    D3D11_RASTERIZER_DESC rs_desc;
    _sg_clear(&rs_desc, sizeof(rs_desc));
    rs_desc.FillMode = D3D11_FILL_SOLID;
    rs_desc.CullMode = _sg_d3d11_cull_mode(desc->cull_mode);
    rs_desc.FrontCounterClockwise = desc->face_winding == SG_FACEWINDING_CCW;
    rs_desc.DepthBias = (INT) pip->cmn.depth.bias;
    rs_desc.DepthBiasClamp = pip->cmn.depth.bias_clamp;
    rs_desc.SlopeScaledDepthBias = pip->cmn.depth.bias_slope_scale;
    rs_desc.DepthClipEnable = TRUE;
    rs_desc.ScissorEnable = TRUE;
    rs_desc.MultisampleEnable = desc->sample_count > 1;
    rs_desc.AntialiasedLineEnable = FALSE;
    hr = _sg_d3d11_CreateRasterizerState(_sg.d3d11.dev, &rs_desc, &pip->d3d11.rs);
    if (!(SUCCEEDED(hr) && pip->d3d11.rs)) {
        _SG_ERROR(D3D11_CREATE_RASTERIZER_STATE_FAILED);
        return SG_RESOURCESTATE_FAILED;
    }
    _sg_d3d11_setlabel(pip->d3d11.rs, desc->label);

    // create depth-stencil state
    D3D11_DEPTH_STENCIL_DESC dss_desc;
    _sg_clear(&dss_desc, sizeof(dss_desc));
    dss_desc.DepthEnable = TRUE;
    dss_desc.DepthWriteMask = desc->depth.write_enabled ? D3D11_DEPTH_WRITE_MASK_ALL : D3D11_DEPTH_WRITE_MASK_ZERO;
    dss_desc.DepthFunc = _sg_d3d11_compare_func(desc->depth.compare);
    dss_desc.StencilEnable = desc->stencil.enabled;
    dss_desc.StencilReadMask = desc->stencil.read_mask;
    dss_desc.StencilWriteMask = desc->stencil.write_mask;
    const sg_stencil_face_state* sf = &desc->stencil.front;
    dss_desc.FrontFace.StencilFailOp = _sg_d3d11_stencil_op(sf->fail_op);
    dss_desc.FrontFace.StencilDepthFailOp = _sg_d3d11_stencil_op(sf->depth_fail_op);
    dss_desc.FrontFace.StencilPassOp = _sg_d3d11_stencil_op(sf->pass_op);
    dss_desc.FrontFace.StencilFunc = _sg_d3d11_compare_func(sf->compare);
    const sg_stencil_face_state* sb = &desc->stencil.back;
    dss_desc.BackFace.StencilFailOp = _sg_d3d11_stencil_op(sb->fail_op);
    dss_desc.BackFace.StencilDepthFailOp = _sg_d3d11_stencil_op(sb->depth_fail_op);
    dss_desc.BackFace.StencilPassOp = _sg_d3d11_stencil_op(sb->pass_op);
    dss_desc.BackFace.StencilFunc = _sg_d3d11_compare_func(sb->compare);
    hr = _sg_d3d11_CreateDepthStencilState(_sg.d3d11.dev, &dss_desc, &pip->d3d11.dss);
    if (!(SUCCEEDED(hr) && pip->d3d11.dss)) {
        _SG_ERROR(D3D11_CREATE_DEPTH_STENCIL_STATE_FAILED);
        return SG_RESOURCESTATE_FAILED;
    }
    _sg_d3d11_setlabel(pip->d3d11.dss, desc->label);

    // create blend state
    D3D11_BLEND_DESC bs_desc;
    _sg_clear(&bs_desc, sizeof(bs_desc));
    bs_desc.AlphaToCoverageEnable = desc->alpha_to_coverage_enabled;
    bs_desc.IndependentBlendEnable = TRUE;
    {
        size_t i = 0;
        for (i = 0; i < (size_t)desc->color_count; i++) {
            const sg_blend_state* src = &desc->colors[i].blend;
            D3D11_RENDER_TARGET_BLEND_DESC* dst = &bs_desc.RenderTarget[i];
            dst->BlendEnable = src->enabled;
            dst->SrcBlend = _sg_d3d11_blend_factor(src->src_factor_rgb);
            dst->DestBlend = _sg_d3d11_blend_factor(src->dst_factor_rgb);
            dst->BlendOp = _sg_d3d11_blend_op(src->op_rgb);
            dst->SrcBlendAlpha = _sg_d3d11_blend_factor(src->src_factor_alpha);
            dst->DestBlendAlpha = _sg_d3d11_blend_factor(src->dst_factor_alpha);
            dst->BlendOpAlpha = _sg_d3d11_blend_op(src->op_alpha);
            dst->RenderTargetWriteMask = _sg_d3d11_color_write_mask(desc->colors[i].write_mask);
        }
        for (; i < 8; i++) {
            D3D11_RENDER_TARGET_BLEND_DESC* dst = &bs_desc.RenderTarget[i];
            dst->BlendEnable = FALSE;
            dst->SrcBlend = dst->SrcBlendAlpha = D3D11_BLEND_ONE;
            dst->DestBlend = dst->DestBlendAlpha = D3D11_BLEND_ZERO;
            dst->BlendOp = dst->BlendOpAlpha = D3D11_BLEND_OP_ADD;
            dst->RenderTargetWriteMask = D3D11_COLOR_WRITE_ENABLE_ALL;
        }
    }
    hr = _sg_d3d11_CreateBlendState(_sg.d3d11.dev, &bs_desc, &pip->d3d11.bs);
    if (!(SUCCEEDED(hr) && pip->d3d11.bs)) {
        _SG_ERROR(D3D11_CREATE_BLEND_STATE_FAILED);
        return SG_RESOURCESTATE_FAILED;
    }
    _sg_d3d11_setlabel(pip->d3d11.bs, desc->label);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_d3d11_discard_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    if (pip == _sg.d3d11.cur_pipeline) {
        _sg.d3d11.cur_pipeline = 0;
        _sg.d3d11.cur_pipeline_id.id = SG_INVALID_ID;
    }
    if (pip->d3d11.il) {
        _sg_d3d11_Release(pip->d3d11.il);
    }
    if (pip->d3d11.rs) {
        _sg_d3d11_Release(pip->d3d11.rs);
    }
    if (pip->d3d11.dss) {
        _sg_d3d11_Release(pip->d3d11.dss);
    }
    if (pip->d3d11.bs) {
        _sg_d3d11_Release(pip->d3d11.bs);
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_d3d11_create_attachments(_sg_attachments_t* atts, const _sg_attachments_ptrs_t* atts_ptrs, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && atts_ptrs && desc);
    SOKOL_ASSERT(_sg.d3d11.dev);

    // copy image pointers
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        const sg_attachment_desc* color_desc = &desc->colors[i];
        _SOKOL_UNUSED(color_desc);
        SOKOL_ASSERT(color_desc->image.id != SG_INVALID_ID);
        SOKOL_ASSERT(0 == atts->d3d11.colors[i].image);
        SOKOL_ASSERT(atts_ptrs->color_images[i]);
        _sg_image_t* clr_img = atts_ptrs->color_images[i];
        SOKOL_ASSERT(clr_img->slot.id == color_desc->image.id);
        SOKOL_ASSERT(_sg_is_valid_attachment_color_format(clr_img->cmn.pixel_format));
        atts->d3d11.colors[i].image = clr_img;

        const sg_attachment_desc* resolve_desc = &desc->resolves[i];
        if (resolve_desc->image.id != SG_INVALID_ID) {
            SOKOL_ASSERT(0 == atts->d3d11.resolves[i].image);
            SOKOL_ASSERT(atts_ptrs->resolve_images[i]);
            _sg_image_t* rsv_img = atts_ptrs->resolve_images[i];
            SOKOL_ASSERT(rsv_img->slot.id == resolve_desc->image.id);
            SOKOL_ASSERT(clr_img->cmn.pixel_format == rsv_img->cmn.pixel_format);
            atts->d3d11.resolves[i].image = rsv_img;
        }
    }
    SOKOL_ASSERT(0 == atts->d3d11.depth_stencil.image);
    const sg_attachment_desc* ds_desc = &desc->depth_stencil;
    if (ds_desc->image.id != SG_INVALID_ID) {
        SOKOL_ASSERT(atts_ptrs->ds_image);
        _sg_image_t* ds_img = atts_ptrs->ds_image;
        SOKOL_ASSERT(ds_img->slot.id == ds_desc->image.id);
        SOKOL_ASSERT(_sg_is_valid_attachment_depth_format(ds_img->cmn.pixel_format));
        atts->d3d11.depth_stencil.image = ds_img;
    }
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        const sg_attachment_desc* storage_desc = &desc->storages[i];
        if (storage_desc->image.id != SG_INVALID_ID) {
            SOKOL_ASSERT(0 == atts->d3d11.storages[i].image);
            SOKOL_ASSERT(atts_ptrs->storage_images[i]);
            _sg_image_t* stg_img = atts_ptrs->storage_images[i];
            SOKOL_ASSERT(stg_img->slot.id == storage_desc->image.id);
            atts->d3d11.storages[i].image = stg_img;
        }
    }

    // create render-target views
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        const _sg_attachment_common_t* cmn_color_att = &atts->cmn.colors[i];
        const _sg_image_t* clr_img = atts_ptrs->color_images[i];
        SOKOL_ASSERT(0 == atts->d3d11.colors[i].view.rtv);
        const bool msaa = clr_img->cmn.sample_count > 1;
        D3D11_RENDER_TARGET_VIEW_DESC d3d11_rtv_desc;
        _sg_clear(&d3d11_rtv_desc, sizeof(d3d11_rtv_desc));
        d3d11_rtv_desc.Format = _sg_d3d11_rtv_uav_pixel_format(clr_img->cmn.pixel_format);
        switch (clr_img->cmn.type) {
            case SG_IMAGETYPE_2D:
                if (msaa) {
                    d3d11_rtv_desc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DMS;
                } else {
                    d3d11_rtv_desc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2D;
                    d3d11_rtv_desc.Texture2D.MipSlice = (UINT)cmn_color_att->mip_level;
                }
                break;
            case SG_IMAGETYPE_CUBE:
            case SG_IMAGETYPE_ARRAY:
                if (msaa) {
                    d3d11_rtv_desc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DMSARRAY;
                    d3d11_rtv_desc.Texture2DMSArray.FirstArraySlice = (UINT)cmn_color_att->slice;
                    d3d11_rtv_desc.Texture2DMSArray.ArraySize = 1;
                } else {
                    d3d11_rtv_desc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE2DARRAY;
                    d3d11_rtv_desc.Texture2DArray.MipSlice = (UINT)cmn_color_att->mip_level;
                    d3d11_rtv_desc.Texture2DArray.FirstArraySlice = (UINT)cmn_color_att->slice;
                    d3d11_rtv_desc.Texture2DArray.ArraySize = 1;
                }
                break;
            case SG_IMAGETYPE_3D:
                SOKOL_ASSERT(!msaa);
                d3d11_rtv_desc.ViewDimension = D3D11_RTV_DIMENSION_TEXTURE3D;
                d3d11_rtv_desc.Texture3D.MipSlice = (UINT)cmn_color_att->mip_level;
                d3d11_rtv_desc.Texture3D.FirstWSlice = (UINT)cmn_color_att->slice;
                d3d11_rtv_desc.Texture3D.WSize = 1;
                break;
            default: SOKOL_UNREACHABLE; break;
        }
        SOKOL_ASSERT(clr_img->d3d11.res);
        HRESULT hr = _sg_d3d11_CreateRenderTargetView(_sg.d3d11.dev, clr_img->d3d11.res, &d3d11_rtv_desc, &atts->d3d11.colors[i].view.rtv);
        if (!(SUCCEEDED(hr) && atts->d3d11.colors[i].view.rtv)) {
            _SG_ERROR(D3D11_CREATE_RTV_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(atts->d3d11.colors[i].view.rtv, desc->label);
    }
    SOKOL_ASSERT(0 == atts->d3d11.depth_stencil.view.dsv);
    if (ds_desc->image.id != SG_INVALID_ID) {
        const _sg_attachment_common_t* cmn_ds_att = &atts->cmn.depth_stencil;
        _sg_image_t* ds_img = atts_ptrs->ds_image;
        const bool msaa = ds_img->cmn.sample_count > 1;
        D3D11_DEPTH_STENCIL_VIEW_DESC d3d11_dsv_desc;
        _sg_clear(&d3d11_dsv_desc, sizeof(d3d11_dsv_desc));
        d3d11_dsv_desc.Format = _sg_d3d11_dsv_pixel_format(ds_img->cmn.pixel_format);
        SOKOL_ASSERT(ds_img && ds_img->cmn.type != SG_IMAGETYPE_3D);
        switch(ds_img->cmn.type) {
            case SG_IMAGETYPE_2D:
                if (msaa) {
                    d3d11_dsv_desc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DMS;
                } else {
                    d3d11_dsv_desc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2D;
                    d3d11_dsv_desc.Texture2D.MipSlice = (UINT)cmn_ds_att->mip_level;
                }
                break;
            case SG_IMAGETYPE_CUBE:
            case SG_IMAGETYPE_ARRAY:
                if (msaa) {
                    d3d11_dsv_desc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DMSARRAY;
                    d3d11_dsv_desc.Texture2DMSArray.FirstArraySlice = (UINT)cmn_ds_att->slice;
                    d3d11_dsv_desc.Texture2DMSArray.ArraySize = 1;
                } else {
                    d3d11_dsv_desc.ViewDimension = D3D11_DSV_DIMENSION_TEXTURE2DARRAY;
                    d3d11_dsv_desc.Texture2DArray.MipSlice = (UINT)cmn_ds_att->mip_level;
                    d3d11_dsv_desc.Texture2DArray.FirstArraySlice = (UINT)cmn_ds_att->slice;
                    d3d11_dsv_desc.Texture2DArray.ArraySize = 1;
                }
                break;
            default: SOKOL_UNREACHABLE; break;
        }
        SOKOL_ASSERT(ds_img->d3d11.res);
        HRESULT hr = _sg_d3d11_CreateDepthStencilView(_sg.d3d11.dev, ds_img->d3d11.res, &d3d11_dsv_desc, &atts->d3d11.depth_stencil.view.dsv);
        if (!(SUCCEEDED(hr) && atts->d3d11.depth_stencil.view.dsv)) {
            _SG_ERROR(D3D11_CREATE_DSV_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(atts->d3d11.depth_stencil.view.dsv, desc->label);
    }

    // create storage attachments unordered access views
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        const _sg_attachment_common_t* cmn_stg_att = &atts->cmn.storages[i];
        const _sg_image_t* stg_img = atts_ptrs->storage_images[i];
        if (!stg_img) {
            continue;
        }
        SOKOL_ASSERT(stg_img->cmn.sample_count == 1);
        SOKOL_ASSERT(0 == atts->d3d11.storages[i].view.uav);
        D3D11_UNORDERED_ACCESS_VIEW_DESC d3d11_uav_desc;
        _sg_clear(&d3d11_uav_desc, sizeof(d3d11_uav_desc));
        d3d11_uav_desc.Format = _sg_d3d11_rtv_uav_pixel_format(stg_img->cmn.pixel_format);
        switch (stg_img->cmn.type) {
            case SG_IMAGETYPE_2D:
                d3d11_uav_desc.ViewDimension = D3D11_UAV_DIMENSION_TEXTURE2D;
                d3d11_uav_desc.Texture2D.MipSlice = (UINT)cmn_stg_att->mip_level;
                break;
            case SG_IMAGETYPE_CUBE:
            case SG_IMAGETYPE_ARRAY:
                d3d11_uav_desc.ViewDimension = D3D11_UAV_DIMENSION_TEXTURE2DARRAY;
                d3d11_uav_desc.Texture2DArray.MipSlice = (UINT)cmn_stg_att->mip_level;
                d3d11_uav_desc.Texture2DArray.FirstArraySlice = (UINT)cmn_stg_att->slice;
                d3d11_uav_desc.Texture2DArray.ArraySize = 1;
                break;
            case SG_IMAGETYPE_3D:
                d3d11_uav_desc.ViewDimension = D3D11_UAV_DIMENSION_TEXTURE3D;
                d3d11_uav_desc.Texture3D.MipSlice = (UINT)cmn_stg_att->mip_level;
                d3d11_uav_desc.Texture3D.FirstWSlice = (UINT)cmn_stg_att->slice;
                d3d11_uav_desc.Texture3D.WSize = 1;
                break;
            default: SOKOL_UNREACHABLE; break;
        }
        SOKOL_ASSERT(stg_img->d3d11.res);
        HRESULT hr = _sg_d3d11_CreateUnorderedAccessView(_sg.d3d11.dev, stg_img->d3d11.res, &d3d11_uav_desc, &atts->d3d11.storages[i].view.uav);
        if (!(SUCCEEDED(hr) && atts->d3d11.storages[i].view.uav)) {
            _SG_ERROR(D3D11_CREATE_UAV_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        _sg_d3d11_setlabel(atts->d3d11.storages[i].view.uav, desc->label);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_d3d11_discard_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    for (size_t i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
        if (atts->d3d11.colors[i].view.rtv) {
            _sg_d3d11_Release(atts->d3d11.colors[i].view.rtv);
        }
        if (atts->d3d11.resolves[i].view.rtv) {
            _sg_d3d11_Release(atts->d3d11.resolves[i].view.rtv);
        }
    }
    if (atts->d3d11.depth_stencil.view.dsv) {
        _sg_d3d11_Release(atts->d3d11.depth_stencil.view.dsv);
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (atts->d3d11.storages[i].view.uav) {
            _sg_d3d11_Release(atts->d3d11.storages[i].view.uav);
        }
    }
}

_SOKOL_PRIVATE void _sg_d3d11_begin_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass);
    if (_sg.cur_pass.is_compute) {
        // nothing to do in compute passes
        return;
    }
    const _sg_attachments_t* atts = _sg.cur_pass.atts;
    const sg_swapchain* swapchain = &pass->swapchain;
    const sg_pass_action* action = &pass->action;

    int num_rtvs = 0;
    ID3D11RenderTargetView* rtvs[SG_MAX_COLOR_ATTACHMENTS] = { 0 };
    ID3D11DepthStencilView* dsv = 0;
    _sg.d3d11.cur_pass.render_view = 0;
    _sg.d3d11.cur_pass.resolve_view = 0;
    if (atts) {
        num_rtvs = atts->cmn.num_colors;
        for (size_t i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
            rtvs[i] = atts->d3d11.colors[i].view.rtv;
        }
        dsv = atts->d3d11.depth_stencil.view.dsv;
    } else {
        // NOTE: depth-stencil-view is optional
        SOKOL_ASSERT(swapchain->d3d11.render_view);
        num_rtvs = 1;
        rtvs[0] = (ID3D11RenderTargetView*) swapchain->d3d11.render_view;
        dsv = (ID3D11DepthStencilView*) swapchain->d3d11.depth_stencil_view;
        _sg.d3d11.cur_pass.render_view = (ID3D11RenderTargetView*) swapchain->d3d11.render_view;
        _sg.d3d11.cur_pass.resolve_view = (ID3D11RenderTargetView*) swapchain->d3d11.resolve_view;
    }
    // apply the render-target- and depth-stencil-views
    _sg_d3d11_OMSetRenderTargets(_sg.d3d11.ctx, SG_MAX_COLOR_ATTACHMENTS, rtvs, dsv);
    _sg_stats_add(d3d11.pass.num_om_set_render_targets, 1);

    // set viewport and scissor rect to cover whole screen
    D3D11_VIEWPORT vp;
    _sg_clear(&vp, sizeof(vp));
    vp.Width = (FLOAT) _sg.cur_pass.width;
    vp.Height = (FLOAT) _sg.cur_pass.height;
    vp.MaxDepth = 1.0f;
    _sg_d3d11_RSSetViewports(_sg.d3d11.ctx, 1, &vp);
    D3D11_RECT rect;
    rect.left = 0;
    rect.top = 0;
    rect.right = _sg.cur_pass.width;
    rect.bottom = _sg.cur_pass.height;
    _sg_d3d11_RSSetScissorRects(_sg.d3d11.ctx, 1, &rect);

    // perform clear action
    for (size_t i = 0; i < (size_t)num_rtvs; i++) {
        if (action->colors[i].load_action == SG_LOADACTION_CLEAR) {
            _sg_d3d11_ClearRenderTargetView(_sg.d3d11.ctx, rtvs[i], (float*)&action->colors[i].clear_value);
            _sg_stats_add(d3d11.pass.num_clear_render_target_view, 1);
        }
    }
    UINT ds_flags = 0;
    if (action->depth.load_action == SG_LOADACTION_CLEAR) {
        ds_flags |= D3D11_CLEAR_DEPTH;
    }
    if (action->stencil.load_action == SG_LOADACTION_CLEAR) {
        ds_flags |= D3D11_CLEAR_STENCIL;
    }
    if ((0 != ds_flags) && dsv) {
        _sg_d3d11_ClearDepthStencilView(_sg.d3d11.ctx, dsv, ds_flags, action->depth.clear_value, action->stencil.clear_value);
        _sg_stats_add(d3d11.pass.num_clear_depth_stencil_view, 1);
    }
}

// D3D11CalcSubresource only exists for C++
_SOKOL_PRIVATE UINT _sg_d3d11_calcsubresource(UINT mip_slice, UINT array_slice, UINT mip_levels) {
    return mip_slice + array_slice * mip_levels;
}

_SOKOL_PRIVATE void _sg_d3d11_end_pass(void) {
    SOKOL_ASSERT(_sg.d3d11.ctx);

    if (!_sg.cur_pass.is_compute) {
        // need to resolve MSAA render attachments into texture?
        if (_sg.cur_pass.atts_id.id != SG_INVALID_ID) {
            // ...for offscreen pass...
            SOKOL_ASSERT(_sg.cur_pass.atts && _sg.cur_pass.atts->slot.id == _sg.cur_pass.atts_id.id);
            for (size_t i = 0; i < (size_t)_sg.cur_pass.atts->cmn.num_colors; i++) {
                const _sg_image_t* resolve_img = _sg.cur_pass.atts->d3d11.resolves[i].image;
                if (resolve_img) {
                    const _sg_image_t* color_img = _sg.cur_pass.atts->d3d11.colors[i].image;
                    const _sg_attachment_common_t* cmn_color_att = &_sg.cur_pass.atts->cmn.colors[i];
                    const _sg_attachment_common_t* cmn_resolve_att = &_sg.cur_pass.atts->cmn.resolves[i];
                    SOKOL_ASSERT(resolve_img->slot.id == cmn_resolve_att->image_id.id);
                    SOKOL_ASSERT(color_img && (color_img->slot.id == cmn_color_att->image_id.id));
                    SOKOL_ASSERT(color_img->cmn.sample_count > 1);
                    SOKOL_ASSERT(resolve_img->cmn.sample_count == 1);
                    const UINT src_subres = _sg_d3d11_calcsubresource(
                        (UINT)cmn_color_att->mip_level,
                        (UINT)cmn_color_att->slice,
                        (UINT)color_img->cmn.num_mipmaps);
                    const UINT dst_subres = _sg_d3d11_calcsubresource(
                        (UINT)cmn_resolve_att->mip_level,
                        (UINT)cmn_resolve_att->slice,
                        (UINT)resolve_img->cmn.num_mipmaps);
                    _sg_d3d11_ResolveSubresource(_sg.d3d11.ctx,
                        resolve_img->d3d11.res,
                        dst_subres,
                        color_img->d3d11.res,
                        src_subres,
                        color_img->d3d11.format);
                    _sg_stats_add(d3d11.pass.num_resolve_subresource, 1);
                }
            }
        } else {
            // ...for swapchain pass...
            if (_sg.d3d11.cur_pass.resolve_view) {
                SOKOL_ASSERT(_sg.d3d11.cur_pass.render_view);
                SOKOL_ASSERT(_sg.cur_pass.swapchain.sample_count > 1);
                SOKOL_ASSERT(_sg.cur_pass.swapchain.color_fmt > SG_PIXELFORMAT_NONE);
                ID3D11Resource* d3d11_render_res = 0;
                ID3D11Resource* d3d11_resolve_res = 0;
                _sg_d3d11_GetResource((ID3D11View*)_sg.d3d11.cur_pass.render_view, &d3d11_render_res);
                _sg_d3d11_GetResource((ID3D11View*)_sg.d3d11.cur_pass.resolve_view, &d3d11_resolve_res);
                SOKOL_ASSERT(d3d11_render_res);
                SOKOL_ASSERT(d3d11_resolve_res);
                const sg_pixel_format color_fmt = _sg.cur_pass.swapchain.color_fmt;
                _sg_d3d11_ResolveSubresource(_sg.d3d11.ctx, d3d11_resolve_res, 0, d3d11_render_res, 0, _sg_d3d11_rtv_uav_pixel_format(color_fmt));
                _sg_d3d11_Release(d3d11_render_res);
                _sg_d3d11_Release(d3d11_resolve_res);
                _sg_stats_add(d3d11.pass.num_resolve_subresource, 1);
            }
        }
    }
    _sg.d3d11.cur_pass.render_view = 0;
    _sg.d3d11.cur_pass.resolve_view = 0;
    _sg.d3d11.cur_pipeline = 0;
    _sg.d3d11.cur_pipeline_id.id = SG_INVALID_ID;
    _sg_d3d11_clear_state();
}

_SOKOL_PRIVATE void _sg_d3d11_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(_sg.d3d11.ctx);
    D3D11_VIEWPORT vp;
    vp.TopLeftX = (FLOAT) x;
    vp.TopLeftY = (FLOAT) (origin_top_left ? y : (_sg.cur_pass.height - (y + h)));
    vp.Width = (FLOAT) w;
    vp.Height = (FLOAT) h;
    vp.MinDepth = 0.0f;
    vp.MaxDepth = 1.0f;
    _sg_d3d11_RSSetViewports(_sg.d3d11.ctx, 1, &vp);
}

_SOKOL_PRIVATE void _sg_d3d11_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(_sg.d3d11.ctx);
    D3D11_RECT rect;
    rect.left = x;
    rect.top = (origin_top_left ? y : (_sg.cur_pass.height - (y + h)));
    rect.right = x + w;
    rect.bottom = origin_top_left ? (y + h) : (_sg.cur_pass.height - y);
    _sg_d3d11_RSSetScissorRects(_sg.d3d11.ctx, 1, &rect);
}

_SOKOL_PRIVATE void _sg_d3d11_populate_storage_attachment_uavs(_sg_pipeline_t* pip, ID3D11UnorderedAccessView** d3d11_cs_uavs) {
    const _sg_attachments_t* atts = _sg.cur_pass.atts;
    SOKOL_ASSERT(atts);
    const _sg_shader_t* shd = pip->shader;
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (shd->cmn.storage_images[i].stage != SG_SHADERSTAGE_COMPUTE) {
            continue;
        }
        SOKOL_ASSERT(shd->d3d11.simg_register_u_n[i] < _SG_D3D11_MAX_STAGE_UAV_BINDINGS);
        SOKOL_ASSERT(atts->d3d11.storages[i].view.uav);
        SOKOL_ASSERT(0 == d3d11_cs_uavs[i]);
        d3d11_cs_uavs[shd->d3d11.simg_register_u_n[i]] = atts->d3d11.storages[i].view.uav;
    }
}

_SOKOL_PRIVATE void _sg_d3d11_apply_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    SOKOL_ASSERT(pip->shader && (pip->cmn.shader_id.id == pip->shader->slot.id));
    SOKOL_ASSERT(_sg.d3d11.ctx);

    _sg.d3d11.cur_pipeline = pip;
    _sg.d3d11.cur_pipeline_id.id = pip->slot.id;

    if (pip->cmn.is_compute) {
        // a compute pipeline
        SOKOL_ASSERT(pip->shader->d3d11.cs);
        _sg_d3d11_CSSetShader(_sg.d3d11.ctx, pip->shader->d3d11.cs, NULL, 0);
        _sg_d3d11_CSSetConstantBuffers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_UB_BINDINGS, pip->shader->d3d11.cs_cbufs);
        _sg_stats_add(d3d11.pipeline.num_cs_set_shader, 1);
        _sg_stats_add(d3d11.pipeline.num_cs_set_constant_buffers, 1);

        // bind storage attachment UAVs
        if (_sg.cur_pass.atts) {
            ID3D11UnorderedAccessView* d3d11_cs_uavs[_SG_D3D11_MAX_STAGE_UAV_BINDINGS] = {0};
            _sg_d3d11_populate_storage_attachment_uavs(pip, d3d11_cs_uavs);
            _sg_d3d11_CSSetUnorderedAccessViews(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_UAV_BINDINGS, d3d11_cs_uavs, NULL);
            _sg_stats_add(d3d11.bindings.num_cs_set_unordered_access_views, 1);
        }
    } else {
        // a render pipeline
        SOKOL_ASSERT(pip->d3d11.rs && pip->d3d11.bs && pip->d3d11.dss);
        SOKOL_ASSERT(pip->shader->d3d11.vs);
        SOKOL_ASSERT(pip->shader->d3d11.fs);

        _sg.d3d11.use_indexed_draw = (pip->d3d11.index_format != DXGI_FORMAT_UNKNOWN);
        _sg.d3d11.use_instanced_draw = pip->cmn.use_instanced_draw;

        _sg_d3d11_RSSetState(_sg.d3d11.ctx, pip->d3d11.rs);
        _sg_d3d11_OMSetDepthStencilState(_sg.d3d11.ctx, pip->d3d11.dss, pip->d3d11.stencil_ref);
        _sg_d3d11_OMSetBlendState(_sg.d3d11.ctx, pip->d3d11.bs, (float*)&pip->cmn.blend_color, 0xFFFFFFFF);
        _sg_d3d11_IASetPrimitiveTopology(_sg.d3d11.ctx, pip->d3d11.topology);
        _sg_d3d11_IASetInputLayout(_sg.d3d11.ctx, pip->d3d11.il);
        _sg_d3d11_VSSetShader(_sg.d3d11.ctx, pip->shader->d3d11.vs, NULL, 0);
        _sg_d3d11_VSSetConstantBuffers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_UB_BINDINGS, pip->shader->d3d11.vs_cbufs);
        _sg_d3d11_PSSetShader(_sg.d3d11.ctx, pip->shader->d3d11.fs, NULL, 0);
        _sg_d3d11_PSSetConstantBuffers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_UB_BINDINGS, pip->shader->d3d11.fs_cbufs);
        _sg_stats_add(d3d11.pipeline.num_rs_set_state, 1);
        _sg_stats_add(d3d11.pipeline.num_om_set_depth_stencil_state, 1);
        _sg_stats_add(d3d11.pipeline.num_om_set_blend_state, 1);
        _sg_stats_add(d3d11.pipeline.num_ia_set_primitive_topology, 1);
        _sg_stats_add(d3d11.pipeline.num_ia_set_input_layout, 1);
        _sg_stats_add(d3d11.pipeline.num_vs_set_shader, 1);
        _sg_stats_add(d3d11.pipeline.num_vs_set_constant_buffers, 1);
        _sg_stats_add(d3d11.pipeline.num_ps_set_shader, 1);
        _sg_stats_add(d3d11.pipeline.num_ps_set_constant_buffers, 1);
    }
}

_SOKOL_PRIVATE bool _sg_d3d11_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip && bnd->pip->shader);
    SOKOL_ASSERT(bnd->pip->shader->slot.id == bnd->pip->cmn.shader_id.id);
    SOKOL_ASSERT(_sg.d3d11.ctx);
    const _sg_shader_t* shd = bnd->pip->shader;
    const bool is_compute = bnd->pip->cmn.is_compute;

    // gather all the D3D11 resources into arrays
    ID3D11Buffer* d3d11_ib = bnd->ib ? bnd->ib->d3d11.buf : 0;
    ID3D11Buffer* d3d11_vbs[SG_MAX_VERTEXBUFFER_BINDSLOTS] = {0};
    UINT d3d11_vb_offsets[SG_MAX_VERTEXBUFFER_BINDSLOTS] = {0};
    ID3D11ShaderResourceView* d3d11_vs_srvs[_SG_D3D11_MAX_STAGE_SRV_BINDINGS] = {0};
    ID3D11ShaderResourceView* d3d11_fs_srvs[_SG_D3D11_MAX_STAGE_SRV_BINDINGS] = {0};
    ID3D11ShaderResourceView* d3d11_cs_srvs[_SG_D3D11_MAX_STAGE_SRV_BINDINGS] = {0};
    ID3D11SamplerState* d3d11_vs_smps[_SG_D3D11_MAX_STAGE_SMP_BINDINGS] = {0};
    ID3D11SamplerState* d3d11_fs_smps[_SG_D3D11_MAX_STAGE_SMP_BINDINGS] = {0};
    ID3D11SamplerState* d3d11_cs_smps[_SG_D3D11_MAX_STAGE_SMP_BINDINGS] = {0};
    ID3D11UnorderedAccessView* d3d11_cs_uavs[_SG_D3D11_MAX_STAGE_UAV_BINDINGS] = {0};

    if (!is_compute) {
        for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
            const _sg_buffer_t* vb = bnd->vbs[i];
            if (vb == 0) {
                continue;
            }
            SOKOL_ASSERT(vb->d3d11.buf);
            d3d11_vbs[i] = vb->d3d11.buf;
            d3d11_vb_offsets[i] = (UINT)bnd->vb_offsets[i];
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        const _sg_image_t* img = bnd->imgs[i];
        if (img == 0) {
            continue;
        }
        const sg_shader_stage stage = shd->cmn.images[i].stage;
        SOKOL_ASSERT(stage != SG_SHADERSTAGE_NONE);
        const uint8_t d3d11_slot = shd->d3d11.img_register_t_n[i];
        SOKOL_ASSERT(d3d11_slot < _SG_D3D11_MAX_STAGE_SRV_BINDINGS);
        SOKOL_ASSERT(img->d3d11.srv);
        ID3D11ShaderResourceView* d3d11_srv = img->d3d11.srv;
        switch (stage) {
            case SG_SHADERSTAGE_VERTEX: d3d11_vs_srvs[d3d11_slot] = d3d11_srv; break;
            case SG_SHADERSTAGE_FRAGMENT: d3d11_fs_srvs[d3d11_slot] = d3d11_srv; break;
            case SG_SHADERSTAGE_COMPUTE: d3d11_cs_srvs[d3d11_slot] = d3d11_srv; break;
            default: SOKOL_UNREACHABLE;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        const _sg_buffer_t* sbuf = bnd->sbufs[i];
        if (sbuf == 0) {
            continue;
        }
        const sg_shader_stage stage = shd->cmn.storage_buffers[i].stage;
        SOKOL_ASSERT(stage != SG_SHADERSTAGE_NONE);
        if (shd->cmn.storage_buffers[i].readonly) {
            SOKOL_ASSERT(sbuf->d3d11.srv);
            const uint8_t d3d11_slot = shd->d3d11.sbuf_register_t_n[i];
            SOKOL_ASSERT(d3d11_slot < _SG_D3D11_MAX_STAGE_SRV_BINDINGS);
            ID3D11ShaderResourceView* d3d11_srv = sbuf->d3d11.srv;
            switch (stage) {
                case SG_SHADERSTAGE_VERTEX: d3d11_vs_srvs[d3d11_slot] = d3d11_srv; break;
                case SG_SHADERSTAGE_FRAGMENT: d3d11_fs_srvs[d3d11_slot] = d3d11_srv; break;
                case SG_SHADERSTAGE_COMPUTE: d3d11_cs_srvs[d3d11_slot] = d3d11_srv; break;
                default: SOKOL_UNREACHABLE;
            }
        } else {
            SOKOL_ASSERT(sbuf->d3d11.uav);
            SOKOL_ASSERT(stage == SG_SHADERSTAGE_COMPUTE);
            const uint8_t d3d11_slot = shd->d3d11.sbuf_register_u_n[i];
            SOKOL_ASSERT(d3d11_slot < _SG_D3D11_MAX_STAGE_UAV_BINDINGS);
            d3d11_cs_uavs[d3d11_slot] = sbuf->d3d11.uav;
        }
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        const _sg_sampler_t* smp = bnd->smps[i];
        if (smp == 0) {
            continue;
        }
        const sg_shader_stage stage = shd->cmn.samplers[i].stage;
        SOKOL_ASSERT(stage != SG_SHADERSTAGE_NONE);
        const uint8_t d3d11_slot = shd->d3d11.smp_register_s_n[i];
        SOKOL_ASSERT(d3d11_slot < _SG_D3D11_MAX_STAGE_SMP_BINDINGS);
        SOKOL_ASSERT(smp->d3d11.smp);
        ID3D11SamplerState* d3d11_smp = smp->d3d11.smp;
        switch (stage) {
            case SG_SHADERSTAGE_VERTEX: d3d11_vs_smps[d3d11_slot] = d3d11_smp; break;
            case SG_SHADERSTAGE_FRAGMENT: d3d11_fs_smps[d3d11_slot] = d3d11_smp; break;
            case SG_SHADERSTAGE_COMPUTE: d3d11_cs_smps[d3d11_slot] = d3d11_smp; break;
            default: SOKOL_UNREACHABLE;
        }
    }
    if (is_compute) {
        // in a compute pass with storage attachments, also need to rebind the storage attachments
        if (_sg.cur_pass.atts) {
            _sg_d3d11_populate_storage_attachment_uavs(bnd->pip, d3d11_cs_uavs);
        }
        _sg_d3d11_CSSetShaderResources(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SRV_BINDINGS, d3d11_cs_srvs);
        _sg_d3d11_CSSetSamplers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SMP_BINDINGS, d3d11_cs_smps);
        _sg_d3d11_CSSetUnorderedAccessViews(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_UAV_BINDINGS, d3d11_cs_uavs, NULL);
        _sg_stats_add(d3d11.bindings.num_cs_set_shader_resources, 1);
        _sg_stats_add(d3d11.bindings.num_cs_set_samplers, 1);
        _sg_stats_add(d3d11.bindings.num_cs_set_unordered_access_views, 1);
    } else {
        _sg_d3d11_IASetVertexBuffers(_sg.d3d11.ctx, 0, SG_MAX_VERTEXBUFFER_BINDSLOTS, d3d11_vbs, bnd->pip->d3d11.vb_strides, d3d11_vb_offsets);
        _sg_d3d11_IASetIndexBuffer(_sg.d3d11.ctx, d3d11_ib, bnd->pip->d3d11.index_format, (UINT)bnd->ib_offset);
        _sg_d3d11_VSSetShaderResources(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SRV_BINDINGS, d3d11_vs_srvs);
        _sg_d3d11_PSSetShaderResources(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SRV_BINDINGS, d3d11_fs_srvs);
        _sg_d3d11_VSSetSamplers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SMP_BINDINGS, d3d11_vs_smps);
        _sg_d3d11_PSSetSamplers(_sg.d3d11.ctx, 0, _SG_D3D11_MAX_STAGE_SMP_BINDINGS, d3d11_fs_smps);
        _sg_stats_add(d3d11.bindings.num_ia_set_vertex_buffers, 1);
        _sg_stats_add(d3d11.bindings.num_ia_set_index_buffer, 1);
        _sg_stats_add(d3d11.bindings.num_vs_set_shader_resources, 1);
        _sg_stats_add(d3d11.bindings.num_ps_set_shader_resources, 1);
        _sg_stats_add(d3d11.bindings.num_vs_set_samplers, 1);
        _sg_stats_add(d3d11.bindings.num_ps_set_samplers, 1);
    }
    return true;
}

_SOKOL_PRIVATE void _sg_d3d11_apply_uniforms(int ub_slot, const sg_range* data) {
    SOKOL_ASSERT(_sg.d3d11.ctx);
    SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
    SOKOL_ASSERT(_sg.d3d11.cur_pipeline && _sg.d3d11.cur_pipeline->slot.id == _sg.d3d11.cur_pipeline_id.id);
    const _sg_shader_t* shd = _sg.d3d11.cur_pipeline->shader;
    SOKOL_ASSERT(shd && (shd->slot.id == _sg.d3d11.cur_pipeline->cmn.shader_id.id));
    SOKOL_ASSERT(data->size == shd->cmn.uniform_blocks[ub_slot].size);

    ID3D11Buffer* cbuf = shd->d3d11.all_cbufs[ub_slot];
    SOKOL_ASSERT(cbuf);
    _sg_d3d11_UpdateSubresource(_sg.d3d11.ctx, (ID3D11Resource*)cbuf, 0, NULL, data->ptr, 0, 0);
    _sg_stats_add(d3d11.uniforms.num_update_subresource, 1);
}

_SOKOL_PRIVATE void _sg_d3d11_draw(int base_element, int num_elements, int num_instances) {
    const bool use_instanced_draw = (num_instances > 1) || (_sg.d3d11.use_instanced_draw);
    if (_sg.d3d11.use_indexed_draw) {
        if (use_instanced_draw) {
            _sg_d3d11_DrawIndexedInstanced(_sg.d3d11.ctx, (UINT)num_elements, (UINT)num_instances, (UINT)base_element, 0, 0);
            _sg_stats_add(d3d11.draw.num_draw_indexed_instanced, 1);
        } else {
            _sg_d3d11_DrawIndexed(_sg.d3d11.ctx, (UINT)num_elements, (UINT)base_element, 0);
            _sg_stats_add(d3d11.draw.num_draw_indexed, 1);
        }
    } else {
        if (use_instanced_draw) {
            _sg_d3d11_DrawInstanced(_sg.d3d11.ctx, (UINT)num_elements, (UINT)num_instances, (UINT)base_element, 0);
            _sg_stats_add(d3d11.draw.num_draw_instanced, 1);
        } else {
            _sg_d3d11_Draw(_sg.d3d11.ctx, (UINT)num_elements, (UINT)base_element);
            _sg_stats_add(d3d11.draw.num_draw, 1);
        }
    }
}

_SOKOL_PRIVATE void _sg_d3d11_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    _sg_d3d11_Dispatch(_sg.d3d11.ctx, (UINT)num_groups_x, (UINT)num_groups_y, (UINT)num_groups_z);
}

_SOKOL_PRIVATE void _sg_d3d11_commit(void) {
    // empty
}

_SOKOL_PRIVATE void _sg_d3d11_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    SOKOL_ASSERT(_sg.d3d11.ctx);
    SOKOL_ASSERT(buf->d3d11.buf);
    D3D11_MAPPED_SUBRESOURCE d3d11_msr;
    HRESULT hr = _sg_d3d11_Map(_sg.d3d11.ctx, (ID3D11Resource*)buf->d3d11.buf, 0, D3D11_MAP_WRITE_DISCARD, 0, &d3d11_msr);
    _sg_stats_add(d3d11.num_map, 1);
    if (SUCCEEDED(hr)) {
        memcpy(d3d11_msr.pData, data->ptr, data->size);
        _sg_d3d11_Unmap(_sg.d3d11.ctx, (ID3D11Resource*)buf->d3d11.buf, 0);
        _sg_stats_add(d3d11.num_unmap, 1);
    } else {
        _SG_ERROR(D3D11_MAP_FOR_UPDATE_BUFFER_FAILED);
    }
}

_SOKOL_PRIVATE void _sg_d3d11_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    SOKOL_ASSERT(_sg.d3d11.ctx);
    SOKOL_ASSERT(buf->d3d11.buf);
    D3D11_MAP map_type = new_frame ? D3D11_MAP_WRITE_DISCARD : D3D11_MAP_WRITE_NO_OVERWRITE;
    D3D11_MAPPED_SUBRESOURCE d3d11_msr;
    HRESULT hr = _sg_d3d11_Map(_sg.d3d11.ctx, (ID3D11Resource*)buf->d3d11.buf, 0, map_type, 0, &d3d11_msr);
    _sg_stats_add(d3d11.num_map, 1);
    if (SUCCEEDED(hr)) {
        uint8_t* dst_ptr = (uint8_t*)d3d11_msr.pData + buf->cmn.append_pos;
        memcpy(dst_ptr, data->ptr, data->size);
        _sg_d3d11_Unmap(_sg.d3d11.ctx, (ID3D11Resource*)buf->d3d11.buf, 0);
        _sg_stats_add(d3d11.num_unmap, 1);
    } else {
        _SG_ERROR(D3D11_MAP_FOR_APPEND_BUFFER_FAILED);
    }
}

// see: https://learn.microsoft.com/en-us/windows/win32/direct3d11/overviews-direct3d-11-resources-subresources
// also see: https://learn.microsoft.com/en-us/windows/win32/api/d3d11/nf-d3d11-d3d11calcsubresource
_SOKOL_PRIVATE void _sg_d3d11_update_image(_sg_image_t* img, const sg_image_data* data) {
    SOKOL_ASSERT(img && data);
    SOKOL_ASSERT(_sg.d3d11.ctx);
    SOKOL_ASSERT(img->d3d11.res);
    const int num_faces = (img->cmn.type == SG_IMAGETYPE_CUBE) ? 6:1;
    const int num_slices = (img->cmn.type == SG_IMAGETYPE_ARRAY) ? img->cmn.num_slices:1;
    const int num_depth_slices = (img->cmn.type == SG_IMAGETYPE_3D) ? img->cmn.num_slices:1;
    UINT subres_index = 0;
    HRESULT hr;
    D3D11_MAPPED_SUBRESOURCE d3d11_msr;
    for (int face_index = 0; face_index < num_faces; face_index++) {
        for (int slice_index = 0; slice_index < num_slices; slice_index++) {
            for (int mip_index = 0; mip_index < img->cmn.num_mipmaps; mip_index++, subres_index++) {
                SOKOL_ASSERT(subres_index < (SG_MAX_MIPMAPS * SG_MAX_TEXTUREARRAY_LAYERS));
                const int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
                const int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
                const int src_row_pitch = _sg_row_pitch(img->cmn.pixel_format, mip_width, 1);
                const int src_depth_pitch = _sg_surface_pitch(img->cmn.pixel_format, mip_width, mip_height, 1);
                const sg_range* subimg_data = &(data->subimage[face_index][mip_index]);
                const size_t slice_size = subimg_data->size / (size_t)num_slices;
                SOKOL_ASSERT(slice_size == (size_t)(src_depth_pitch * num_depth_slices));
                const size_t slice_offset = slice_size * (size_t)slice_index;
                const uint8_t* slice_ptr = ((const uint8_t*)subimg_data->ptr) + slice_offset;
                hr = _sg_d3d11_Map(_sg.d3d11.ctx, img->d3d11.res, subres_index, D3D11_MAP_WRITE_DISCARD, 0, &d3d11_msr);
                _sg_stats_add(d3d11.num_map, 1);
                if (SUCCEEDED(hr)) {
                    const uint8_t* src_ptr = slice_ptr;
                    uint8_t* dst_ptr = (uint8_t*)d3d11_msr.pData;
                    for (int depth_index = 0; depth_index < num_depth_slices; depth_index++) {
                        if (src_row_pitch == (int)d3d11_msr.RowPitch) {
                            const size_t copy_size = slice_size / (size_t)num_depth_slices;
                            SOKOL_ASSERT((copy_size * (size_t)num_depth_slices) == slice_size);
                            memcpy(dst_ptr, src_ptr, copy_size);
                        } else {
                            SOKOL_ASSERT(src_row_pitch < (int)d3d11_msr.RowPitch);
                            const uint8_t* src_row_ptr = src_ptr;
                            uint8_t* dst_row_ptr = dst_ptr;
                            for (int row_index = 0; row_index < mip_height; row_index++) {
                                memcpy(dst_row_ptr, src_row_ptr, (size_t)src_row_pitch);
                                src_row_ptr += src_row_pitch;
                                dst_row_ptr += d3d11_msr.RowPitch;
                            }
                        }
                        src_ptr += src_depth_pitch;
                        dst_ptr += d3d11_msr.DepthPitch;
                    }
                    _sg_d3d11_Unmap(_sg.d3d11.ctx, img->d3d11.res, subres_index);
                    _sg_stats_add(d3d11.num_unmap, 1);
                } else {
                    _SG_ERROR(D3D11_MAP_FOR_UPDATE_IMAGE_FAILED);
                }
            }
        }
    }
}

// ███    ███ ███████ ████████  █████  ██          ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ████  ████ ██         ██    ██   ██ ██          ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██ ████ ██ █████      ██    ███████ ██          ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██  ██  ██ ██         ██    ██   ██ ██          ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
// ██      ██ ███████    ██    ██   ██ ███████     ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>metal backend
#elif defined(SOKOL_METAL)

#if __has_feature(objc_arc)
#define _SG_OBJC_RETAIN(obj) { }
#define _SG_OBJC_RELEASE(obj) { obj = nil; }
#else
#define _SG_OBJC_RETAIN(obj) { [obj retain]; }
#define _SG_OBJC_RELEASE(obj) { [obj release]; obj = nil; }
#endif

//-- enum translation functions ------------------------------------------------
_SOKOL_PRIVATE MTLLoadAction _sg_mtl_load_action(sg_load_action a) {
    switch (a) {
        case SG_LOADACTION_CLEAR:       return MTLLoadActionClear;
        case SG_LOADACTION_LOAD:        return MTLLoadActionLoad;
        case SG_LOADACTION_DONTCARE:    return MTLLoadActionDontCare;
        default: SOKOL_UNREACHABLE;     return (MTLLoadAction)0;
    }
}

_SOKOL_PRIVATE MTLStoreAction _sg_mtl_store_action(sg_store_action a, bool resolve) {
    switch (a) {
        case SG_STOREACTION_STORE:
            if (resolve) {
                return MTLStoreActionStoreAndMultisampleResolve;
            } else {
                return MTLStoreActionStore;
            }
            break;
        case SG_STOREACTION_DONTCARE:
            if (resolve) {
                return MTLStoreActionMultisampleResolve;
            } else {
                return MTLStoreActionDontCare;
            }
            break;
        default: SOKOL_UNREACHABLE; return (MTLStoreAction)0;
    }
}

_SOKOL_PRIVATE MTLResourceOptions _sg_mtl_resource_options_storage_mode_managed_or_shared(void) {
    #if defined(_SG_TARGET_MACOS)
    if (_sg.mtl.use_shared_storage_mode) {
        return MTLResourceStorageModeShared;
    } else {
        return MTLResourceStorageModeManaged;
    }
    #else
        // MTLResourceStorageModeManaged is not even defined on iOS SDK
        return MTLResourceStorageModeShared;
    #endif
}

_SOKOL_PRIVATE MTLResourceOptions _sg_mtl_buffer_resource_options(const sg_buffer_usage* usage) {
    if (usage->immutable) {
        return _sg_mtl_resource_options_storage_mode_managed_or_shared();
    } else {
        return MTLResourceCPUCacheModeWriteCombined | _sg_mtl_resource_options_storage_mode_managed_or_shared();
    }
}

_SOKOL_PRIVATE MTLVertexStepFunction _sg_mtl_step_function(sg_vertex_step step) {
    switch (step) {
        case SG_VERTEXSTEP_PER_VERTEX:      return MTLVertexStepFunctionPerVertex;
        case SG_VERTEXSTEP_PER_INSTANCE:    return MTLVertexStepFunctionPerInstance;
        default: SOKOL_UNREACHABLE; return (MTLVertexStepFunction)0;
    }
}

_SOKOL_PRIVATE MTLVertexFormat _sg_mtl_vertex_format(sg_vertex_format fmt) {
    switch (fmt) {
        case SG_VERTEXFORMAT_FLOAT:     return MTLVertexFormatFloat;
        case SG_VERTEXFORMAT_FLOAT2:    return MTLVertexFormatFloat2;
        case SG_VERTEXFORMAT_FLOAT3:    return MTLVertexFormatFloat3;
        case SG_VERTEXFORMAT_FLOAT4:    return MTLVertexFormatFloat4;
        case SG_VERTEXFORMAT_INT:       return MTLVertexFormatInt;
        case SG_VERTEXFORMAT_INT2:      return MTLVertexFormatInt2;
        case SG_VERTEXFORMAT_INT3:      return MTLVertexFormatInt3;
        case SG_VERTEXFORMAT_INT4:      return MTLVertexFormatInt4;
        case SG_VERTEXFORMAT_UINT:      return MTLVertexFormatUInt;
        case SG_VERTEXFORMAT_UINT2:     return MTLVertexFormatUInt2;
        case SG_VERTEXFORMAT_UINT3:     return MTLVertexFormatUInt3;
        case SG_VERTEXFORMAT_UINT4:     return MTLVertexFormatUInt4;
        case SG_VERTEXFORMAT_BYTE4:     return MTLVertexFormatChar4;
        case SG_VERTEXFORMAT_BYTE4N:    return MTLVertexFormatChar4Normalized;
        case SG_VERTEXFORMAT_UBYTE4:    return MTLVertexFormatUChar4;
        case SG_VERTEXFORMAT_UBYTE4N:   return MTLVertexFormatUChar4Normalized;
        case SG_VERTEXFORMAT_SHORT2:    return MTLVertexFormatShort2;
        case SG_VERTEXFORMAT_SHORT2N:   return MTLVertexFormatShort2Normalized;
        case SG_VERTEXFORMAT_USHORT2:   return MTLVertexFormatUShort2;
        case SG_VERTEXFORMAT_USHORT2N:  return MTLVertexFormatUShort2Normalized;
        case SG_VERTEXFORMAT_SHORT4:    return MTLVertexFormatShort4;
        case SG_VERTEXFORMAT_SHORT4N:   return MTLVertexFormatShort4Normalized;
        case SG_VERTEXFORMAT_USHORT4:   return MTLVertexFormatUShort4;
        case SG_VERTEXFORMAT_USHORT4N:  return MTLVertexFormatUShort4Normalized;
        case SG_VERTEXFORMAT_UINT10_N2: return MTLVertexFormatUInt1010102Normalized;
        case SG_VERTEXFORMAT_HALF2:     return MTLVertexFormatHalf2;
        case SG_VERTEXFORMAT_HALF4:     return MTLVertexFormatHalf4;
        default: SOKOL_UNREACHABLE; return (MTLVertexFormat)0;
    }
}

_SOKOL_PRIVATE MTLPrimitiveType _sg_mtl_primitive_type(sg_primitive_type t) {
    switch (t) {
        case SG_PRIMITIVETYPE_POINTS:           return MTLPrimitiveTypePoint;
        case SG_PRIMITIVETYPE_LINES:            return MTLPrimitiveTypeLine;
        case SG_PRIMITIVETYPE_LINE_STRIP:       return MTLPrimitiveTypeLineStrip;
        case SG_PRIMITIVETYPE_TRIANGLES:        return MTLPrimitiveTypeTriangle;
        case SG_PRIMITIVETYPE_TRIANGLE_STRIP:   return MTLPrimitiveTypeTriangleStrip;
        default: SOKOL_UNREACHABLE; return (MTLPrimitiveType)0;
    }
}

_SOKOL_PRIVATE MTLPixelFormat _sg_mtl_pixel_format(sg_pixel_format fmt) {
    switch (fmt) {
        case SG_PIXELFORMAT_R8:                     return MTLPixelFormatR8Unorm;
        case SG_PIXELFORMAT_R8SN:                   return MTLPixelFormatR8Snorm;
        case SG_PIXELFORMAT_R8UI:                   return MTLPixelFormatR8Uint;
        case SG_PIXELFORMAT_R8SI:                   return MTLPixelFormatR8Sint;
        case SG_PIXELFORMAT_R16:                    return MTLPixelFormatR16Unorm;
        case SG_PIXELFORMAT_R16SN:                  return MTLPixelFormatR16Snorm;
        case SG_PIXELFORMAT_R16UI:                  return MTLPixelFormatR16Uint;
        case SG_PIXELFORMAT_R16SI:                  return MTLPixelFormatR16Sint;
        case SG_PIXELFORMAT_R16F:                   return MTLPixelFormatR16Float;
        case SG_PIXELFORMAT_RG8:                    return MTLPixelFormatRG8Unorm;
        case SG_PIXELFORMAT_RG8SN:                  return MTLPixelFormatRG8Snorm;
        case SG_PIXELFORMAT_RG8UI:                  return MTLPixelFormatRG8Uint;
        case SG_PIXELFORMAT_RG8SI:                  return MTLPixelFormatRG8Sint;
        case SG_PIXELFORMAT_R32UI:                  return MTLPixelFormatR32Uint;
        case SG_PIXELFORMAT_R32SI:                  return MTLPixelFormatR32Sint;
        case SG_PIXELFORMAT_R32F:                   return MTLPixelFormatR32Float;
        case SG_PIXELFORMAT_RG16:                   return MTLPixelFormatRG16Unorm;
        case SG_PIXELFORMAT_RG16SN:                 return MTLPixelFormatRG16Snorm;
        case SG_PIXELFORMAT_RG16UI:                 return MTLPixelFormatRG16Uint;
        case SG_PIXELFORMAT_RG16SI:                 return MTLPixelFormatRG16Sint;
        case SG_PIXELFORMAT_RG16F:                  return MTLPixelFormatRG16Float;
        case SG_PIXELFORMAT_RGBA8:                  return MTLPixelFormatRGBA8Unorm;
        case SG_PIXELFORMAT_SRGB8A8:                return MTLPixelFormatRGBA8Unorm_sRGB;
        case SG_PIXELFORMAT_RGBA8SN:                return MTLPixelFormatRGBA8Snorm;
        case SG_PIXELFORMAT_RGBA8UI:                return MTLPixelFormatRGBA8Uint;
        case SG_PIXELFORMAT_RGBA8SI:                return MTLPixelFormatRGBA8Sint;
        case SG_PIXELFORMAT_BGRA8:                  return MTLPixelFormatBGRA8Unorm;
        case SG_PIXELFORMAT_RGB10A2:                return MTLPixelFormatRGB10A2Unorm;
        case SG_PIXELFORMAT_RG11B10F:               return MTLPixelFormatRG11B10Float;
        case SG_PIXELFORMAT_RGB9E5:                 return MTLPixelFormatRGB9E5Float;
        case SG_PIXELFORMAT_RG32UI:                 return MTLPixelFormatRG32Uint;
        case SG_PIXELFORMAT_RG32SI:                 return MTLPixelFormatRG32Sint;
        case SG_PIXELFORMAT_RG32F:                  return MTLPixelFormatRG32Float;
        case SG_PIXELFORMAT_RGBA16:                 return MTLPixelFormatRGBA16Unorm;
        case SG_PIXELFORMAT_RGBA16SN:               return MTLPixelFormatRGBA16Snorm;
        case SG_PIXELFORMAT_RGBA16UI:               return MTLPixelFormatRGBA16Uint;
        case SG_PIXELFORMAT_RGBA16SI:               return MTLPixelFormatRGBA16Sint;
        case SG_PIXELFORMAT_RGBA16F:                return MTLPixelFormatRGBA16Float;
        case SG_PIXELFORMAT_RGBA32UI:               return MTLPixelFormatRGBA32Uint;
        case SG_PIXELFORMAT_RGBA32SI:               return MTLPixelFormatRGBA32Sint;
        case SG_PIXELFORMAT_RGBA32F:                return MTLPixelFormatRGBA32Float;
        case SG_PIXELFORMAT_DEPTH:                  return MTLPixelFormatDepth32Float;
        case SG_PIXELFORMAT_DEPTH_STENCIL:          return MTLPixelFormatDepth32Float_Stencil8;
        #if defined(_SG_TARGET_MACOS)
        case SG_PIXELFORMAT_BC1_RGBA:               return MTLPixelFormatBC1_RGBA;
        case SG_PIXELFORMAT_BC2_RGBA:               return MTLPixelFormatBC2_RGBA;
        case SG_PIXELFORMAT_BC3_RGBA:               return MTLPixelFormatBC3_RGBA;
        case SG_PIXELFORMAT_BC3_SRGBA:              return MTLPixelFormatBC3_RGBA_sRGB;
        case SG_PIXELFORMAT_BC4_R:                  return MTLPixelFormatBC4_RUnorm;
        case SG_PIXELFORMAT_BC4_RSN:                return MTLPixelFormatBC4_RSnorm;
        case SG_PIXELFORMAT_BC5_RG:                 return MTLPixelFormatBC5_RGUnorm;
        case SG_PIXELFORMAT_BC5_RGSN:               return MTLPixelFormatBC5_RGSnorm;
        case SG_PIXELFORMAT_BC6H_RGBF:              return MTLPixelFormatBC6H_RGBFloat;
        case SG_PIXELFORMAT_BC6H_RGBUF:             return MTLPixelFormatBC6H_RGBUfloat;
        case SG_PIXELFORMAT_BC7_RGBA:               return MTLPixelFormatBC7_RGBAUnorm;
        case SG_PIXELFORMAT_BC7_SRGBA:              return MTLPixelFormatBC7_RGBAUnorm_sRGB;
        #else
        case SG_PIXELFORMAT_ETC2_RGB8:              return MTLPixelFormatETC2_RGB8;
        case SG_PIXELFORMAT_ETC2_SRGB8:             return MTLPixelFormatETC2_RGB8_sRGB;
        case SG_PIXELFORMAT_ETC2_RGB8A1:            return MTLPixelFormatETC2_RGB8A1;
        case SG_PIXELFORMAT_ETC2_RGBA8:             return MTLPixelFormatEAC_RGBA8;
        case SG_PIXELFORMAT_ETC2_SRGB8A8:           return MTLPixelFormatEAC_RGBA8_sRGB;
        case SG_PIXELFORMAT_EAC_R11:                return MTLPixelFormatEAC_R11Unorm;
        case SG_PIXELFORMAT_EAC_R11SN:              return MTLPixelFormatEAC_R11Snorm;
        case SG_PIXELFORMAT_EAC_RG11:               return MTLPixelFormatEAC_RG11Unorm;
        case SG_PIXELFORMAT_EAC_RG11SN:             return MTLPixelFormatEAC_RG11Snorm;
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:          return MTLPixelFormatASTC_4x4_LDR;
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA:         return MTLPixelFormatASTC_4x4_sRGB;
        #endif
        default: return MTLPixelFormatInvalid;
    }
}

_SOKOL_PRIVATE MTLColorWriteMask _sg_mtl_color_write_mask(sg_color_mask m) {
    MTLColorWriteMask mtl_mask = MTLColorWriteMaskNone;
    if (m & SG_COLORMASK_R) {
        mtl_mask |= MTLColorWriteMaskRed;
    }
    if (m & SG_COLORMASK_G) {
        mtl_mask |= MTLColorWriteMaskGreen;
    }
    if (m & SG_COLORMASK_B) {
        mtl_mask |= MTLColorWriteMaskBlue;
    }
    if (m & SG_COLORMASK_A) {
        mtl_mask |= MTLColorWriteMaskAlpha;
    }
    return mtl_mask;
}

_SOKOL_PRIVATE MTLBlendOperation _sg_mtl_blend_op(sg_blend_op op) {
    switch (op) {
        case SG_BLENDOP_ADD:                return MTLBlendOperationAdd;
        case SG_BLENDOP_SUBTRACT:           return MTLBlendOperationSubtract;
        case SG_BLENDOP_REVERSE_SUBTRACT:   return MTLBlendOperationReverseSubtract;
        case SG_BLENDOP_MIN:                return MTLBlendOperationMin;
        case SG_BLENDOP_MAX:                return MTLBlendOperationMax;
        default: SOKOL_UNREACHABLE; return (MTLBlendOperation)0;
    }
}

_SOKOL_PRIVATE MTLBlendFactor _sg_mtl_blend_factor(sg_blend_factor f) {
    switch (f) {
        case SG_BLENDFACTOR_ZERO:                   return MTLBlendFactorZero;
        case SG_BLENDFACTOR_ONE:                    return MTLBlendFactorOne;
        case SG_BLENDFACTOR_SRC_COLOR:              return MTLBlendFactorSourceColor;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_COLOR:    return MTLBlendFactorOneMinusSourceColor;
        case SG_BLENDFACTOR_SRC_ALPHA:              return MTLBlendFactorSourceAlpha;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_ALPHA:    return MTLBlendFactorOneMinusSourceAlpha;
        case SG_BLENDFACTOR_DST_COLOR:              return MTLBlendFactorDestinationColor;
        case SG_BLENDFACTOR_ONE_MINUS_DST_COLOR:    return MTLBlendFactorOneMinusDestinationColor;
        case SG_BLENDFACTOR_DST_ALPHA:              return MTLBlendFactorDestinationAlpha;
        case SG_BLENDFACTOR_ONE_MINUS_DST_ALPHA:    return MTLBlendFactorOneMinusDestinationAlpha;
        case SG_BLENDFACTOR_SRC_ALPHA_SATURATED:    return MTLBlendFactorSourceAlphaSaturated;
        case SG_BLENDFACTOR_BLEND_COLOR:            return MTLBlendFactorBlendColor;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_COLOR:  return MTLBlendFactorOneMinusBlendColor;
        case SG_BLENDFACTOR_BLEND_ALPHA:            return MTLBlendFactorBlendAlpha;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_ALPHA:  return MTLBlendFactorOneMinusBlendAlpha;
        default: SOKOL_UNREACHABLE; return (MTLBlendFactor)0;
    }
}

_SOKOL_PRIVATE MTLCompareFunction _sg_mtl_compare_func(sg_compare_func f) {
    switch (f) {
        case SG_COMPAREFUNC_NEVER:          return MTLCompareFunctionNever;
        case SG_COMPAREFUNC_LESS:           return MTLCompareFunctionLess;
        case SG_COMPAREFUNC_EQUAL:          return MTLCompareFunctionEqual;
        case SG_COMPAREFUNC_LESS_EQUAL:     return MTLCompareFunctionLessEqual;
        case SG_COMPAREFUNC_GREATER:        return MTLCompareFunctionGreater;
        case SG_COMPAREFUNC_NOT_EQUAL:      return MTLCompareFunctionNotEqual;
        case SG_COMPAREFUNC_GREATER_EQUAL:  return MTLCompareFunctionGreaterEqual;
        case SG_COMPAREFUNC_ALWAYS:         return MTLCompareFunctionAlways;
        default: SOKOL_UNREACHABLE; return (MTLCompareFunction)0;
    }
}

_SOKOL_PRIVATE MTLStencilOperation _sg_mtl_stencil_op(sg_stencil_op op) {
    switch (op) {
        case SG_STENCILOP_KEEP:         return MTLStencilOperationKeep;
        case SG_STENCILOP_ZERO:         return MTLStencilOperationZero;
        case SG_STENCILOP_REPLACE:      return MTLStencilOperationReplace;
        case SG_STENCILOP_INCR_CLAMP:   return MTLStencilOperationIncrementClamp;
        case SG_STENCILOP_DECR_CLAMP:   return MTLStencilOperationDecrementClamp;
        case SG_STENCILOP_INVERT:       return MTLStencilOperationInvert;
        case SG_STENCILOP_INCR_WRAP:    return MTLStencilOperationIncrementWrap;
        case SG_STENCILOP_DECR_WRAP:    return MTLStencilOperationDecrementWrap;
        default: SOKOL_UNREACHABLE; return (MTLStencilOperation)0;
    }
}

_SOKOL_PRIVATE MTLCullMode _sg_mtl_cull_mode(sg_cull_mode m) {
    switch (m) {
        case SG_CULLMODE_NONE:  return MTLCullModeNone;
        case SG_CULLMODE_FRONT: return MTLCullModeFront;
        case SG_CULLMODE_BACK:  return MTLCullModeBack;
        default: SOKOL_UNREACHABLE; return (MTLCullMode)0;
    }
}

_SOKOL_PRIVATE MTLWinding _sg_mtl_winding(sg_face_winding w) {
    switch (w) {
        case SG_FACEWINDING_CW:     return MTLWindingClockwise;
        case SG_FACEWINDING_CCW:    return MTLWindingCounterClockwise;
        default: SOKOL_UNREACHABLE; return (MTLWinding)0;
    }
}

_SOKOL_PRIVATE MTLIndexType _sg_mtl_index_type(sg_index_type t) {
    switch (t) {
        case SG_INDEXTYPE_UINT16:   return MTLIndexTypeUInt16;
        case SG_INDEXTYPE_UINT32:   return MTLIndexTypeUInt32;
        default: SOKOL_UNREACHABLE; return (MTLIndexType)0;
    }
}

_SOKOL_PRIVATE int _sg_mtl_index_size(sg_index_type t) {
    switch (t) {
        case SG_INDEXTYPE_NONE:     return 0;
        case SG_INDEXTYPE_UINT16:   return 2;
        case SG_INDEXTYPE_UINT32:   return 4;
        default: SOKOL_UNREACHABLE; return 0;
    }
}

_SOKOL_PRIVATE MTLTextureType _sg_mtl_texture_type(sg_image_type t, bool msaa) {
    switch (t) {
        case SG_IMAGETYPE_2D:       return msaa ? MTLTextureType2DMultisample : MTLTextureType2D;
        case SG_IMAGETYPE_CUBE:     return MTLTextureTypeCube;
        case SG_IMAGETYPE_3D:       return MTLTextureType3D;
        // NOTE: MTLTextureType2DMultisampleArray requires macOS 10.14+, iOS 14.0+
        case SG_IMAGETYPE_ARRAY:    return MTLTextureType2DArray;
        default: SOKOL_UNREACHABLE; return (MTLTextureType)0;
    }
}

_SOKOL_PRIVATE MTLSamplerAddressMode _sg_mtl_address_mode(sg_wrap w) {
    if (_sg.features.image_clamp_to_border) {
        if (@available(macOS 12.0, iOS 14.0, *)) {
            // border color feature available
            switch (w) {
                case SG_WRAP_REPEAT:            return MTLSamplerAddressModeRepeat;
                case SG_WRAP_CLAMP_TO_EDGE:     return MTLSamplerAddressModeClampToEdge;
                case SG_WRAP_CLAMP_TO_BORDER:   return MTLSamplerAddressModeClampToBorderColor;
                case SG_WRAP_MIRRORED_REPEAT:   return MTLSamplerAddressModeMirrorRepeat;
                default: SOKOL_UNREACHABLE; return (MTLSamplerAddressMode)0;
            }
        }
    }
    // fallthrough: clamp to border no supported
    switch (w) {
        case SG_WRAP_REPEAT:            return MTLSamplerAddressModeRepeat;
        case SG_WRAP_CLAMP_TO_EDGE:     return MTLSamplerAddressModeClampToEdge;
        case SG_WRAP_CLAMP_TO_BORDER:   return MTLSamplerAddressModeClampToEdge;
        case SG_WRAP_MIRRORED_REPEAT:   return MTLSamplerAddressModeMirrorRepeat;
        default: SOKOL_UNREACHABLE; return (MTLSamplerAddressMode)0;
    }
}

_SOKOL_PRIVATE API_AVAILABLE(ios(14.0), macos(12.0)) MTLSamplerBorderColor _sg_mtl_border_color(sg_border_color c) {
    switch (c) {
        case SG_BORDERCOLOR_TRANSPARENT_BLACK: return MTLSamplerBorderColorTransparentBlack;
        case SG_BORDERCOLOR_OPAQUE_BLACK: return MTLSamplerBorderColorOpaqueBlack;
        case SG_BORDERCOLOR_OPAQUE_WHITE: return MTLSamplerBorderColorOpaqueWhite;
        default: SOKOL_UNREACHABLE; return (MTLSamplerBorderColor)0;
    }
}

_SOKOL_PRIVATE MTLSamplerMinMagFilter _sg_mtl_minmag_filter(sg_filter f) {
    switch (f) {
        case SG_FILTER_NEAREST:
            return MTLSamplerMinMagFilterNearest;
        case SG_FILTER_LINEAR:
            return MTLSamplerMinMagFilterLinear;
        default:
            SOKOL_UNREACHABLE; return (MTLSamplerMinMagFilter)0;
    }
}

_SOKOL_PRIVATE MTLSamplerMipFilter _sg_mtl_mipmap_filter(sg_filter f) {
    switch (f) {
        case SG_FILTER_NEAREST:
            return MTLSamplerMipFilterNearest;
        case SG_FILTER_LINEAR:
            return MTLSamplerMipFilterLinear;
        default:
            SOKOL_UNREACHABLE; return (MTLSamplerMipFilter)0;
    }
}

_SOKOL_PRIVATE size_t _sg_mtl_vertexbuffer_bindslot(size_t sokol_bindslot) {
    return sokol_bindslot + _SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS;
}

//-- a pool for all Metal resource objects, with deferred release queue ---------
_SOKOL_PRIVATE void _sg_mtl_init_pool(const sg_desc* desc) {
    _sg.mtl.idpool.num_slots = 2 *
        (
            2 * desc->buffer_pool_size +
            4 * desc->image_pool_size +
            1 * desc->sampler_pool_size +
            4 * desc->shader_pool_size +
            2 * desc->pipeline_pool_size +
            desc->attachments_pool_size +
            128
        );
    _sg.mtl.idpool.pool = [NSMutableArray arrayWithCapacity:(NSUInteger)_sg.mtl.idpool.num_slots];
    _SG_OBJC_RETAIN(_sg.mtl.idpool.pool);
    NSNull* null = [NSNull null];
    for (int i = 0; i < _sg.mtl.idpool.num_slots; i++) {
        [_sg.mtl.idpool.pool addObject:null];
    }
    SOKOL_ASSERT([_sg.mtl.idpool.pool count] == (NSUInteger)_sg.mtl.idpool.num_slots);
    // a queue of currently free slot indices
    _sg.mtl.idpool.free_queue_top = 0;
    _sg.mtl.idpool.free_queue = (int*)_sg_malloc_clear((size_t)_sg.mtl.idpool.num_slots * sizeof(int));
    // pool slot 0 is reserved!
    for (int i = _sg.mtl.idpool.num_slots-1; i >= 1; i--) {
        _sg.mtl.idpool.free_queue[_sg.mtl.idpool.free_queue_top++] = i;
    }
    // a circular queue which holds release items (frame index when a resource is to be released, and the resource's pool index
    _sg.mtl.idpool.release_queue_front = 0;
    _sg.mtl.idpool.release_queue_back = 0;
    _sg.mtl.idpool.release_queue = (_sg_mtl_release_item_t*)_sg_malloc_clear((size_t)_sg.mtl.idpool.num_slots * sizeof(_sg_mtl_release_item_t));
    for (int i = 0; i < _sg.mtl.idpool.num_slots; i++) {
        _sg.mtl.idpool.release_queue[i].frame_index = 0;
        _sg.mtl.idpool.release_queue[i].slot_index = _SG_MTL_INVALID_SLOT_INDEX;
    }
}

_SOKOL_PRIVATE void _sg_mtl_destroy_pool(void) {
    _sg_free(_sg.mtl.idpool.release_queue);  _sg.mtl.idpool.release_queue = 0;
    _sg_free(_sg.mtl.idpool.free_queue);     _sg.mtl.idpool.free_queue = 0;
    _SG_OBJC_RELEASE(_sg.mtl.idpool.pool);
}

// get a new free resource pool slot
_SOKOL_PRIVATE int _sg_mtl_alloc_pool_slot(void) {
    SOKOL_ASSERT(_sg.mtl.idpool.free_queue_top > 0);
    const int slot_index = _sg.mtl.idpool.free_queue[--_sg.mtl.idpool.free_queue_top];
    SOKOL_ASSERT((slot_index > 0) && (slot_index < _sg.mtl.idpool.num_slots));
    return slot_index;
}

// put a free resource pool slot back into the free-queue
_SOKOL_PRIVATE void _sg_mtl_free_pool_slot(int slot_index) {
    SOKOL_ASSERT(_sg.mtl.idpool.free_queue_top < _sg.mtl.idpool.num_slots);
    SOKOL_ASSERT((slot_index > 0) && (slot_index < _sg.mtl.idpool.num_slots));
    _sg.mtl.idpool.free_queue[_sg.mtl.idpool.free_queue_top++] = slot_index;
}

// add an MTLResource to the pool, return pool index or 0 if input was 'nil'
_SOKOL_PRIVATE int _sg_mtl_add_resource(id res) {
    if (nil == res) {
        return _SG_MTL_INVALID_SLOT_INDEX;
    }
    _sg_stats_add(metal.idpool.num_added, 1);
    const int slot_index = _sg_mtl_alloc_pool_slot();
    // NOTE: the NSMutableArray will take ownership of its items
    SOKOL_ASSERT([NSNull null] == _sg.mtl.idpool.pool[(NSUInteger)slot_index]);
    _sg.mtl.idpool.pool[(NSUInteger)slot_index] = res;
    return slot_index;
}

/*  mark an MTLResource for release, this will put the resource into the
    deferred-release queue, and the resource will then be released N frames later,
    the special pool index 0 will be ignored (this means that a nil
    value was provided to _sg_mtl_add_resource()
*/
_SOKOL_PRIVATE void _sg_mtl_release_resource(uint32_t frame_index, int slot_index) {
    if (slot_index == _SG_MTL_INVALID_SLOT_INDEX) {
        return;
    }
    _sg_stats_add(metal.idpool.num_released, 1);
    SOKOL_ASSERT((slot_index > 0) && (slot_index < _sg.mtl.idpool.num_slots));
    SOKOL_ASSERT([NSNull null] != _sg.mtl.idpool.pool[(NSUInteger)slot_index]);
    int release_index = _sg.mtl.idpool.release_queue_front++;
    if (_sg.mtl.idpool.release_queue_front >= _sg.mtl.idpool.num_slots) {
        // wrap-around
        _sg.mtl.idpool.release_queue_front = 0;
    }
    // release queue full?
    SOKOL_ASSERT(_sg.mtl.idpool.release_queue_front != _sg.mtl.idpool.release_queue_back);
    SOKOL_ASSERT(0 == _sg.mtl.idpool.release_queue[release_index].frame_index);
    const uint32_t safe_to_release_frame_index = frame_index + SG_NUM_INFLIGHT_FRAMES + 1;
    _sg.mtl.idpool.release_queue[release_index].frame_index = safe_to_release_frame_index;
    _sg.mtl.idpool.release_queue[release_index].slot_index = slot_index;
}

// run garbage-collection pass on all resources in the release-queue
_SOKOL_PRIVATE void _sg_mtl_garbage_collect(uint32_t frame_index) {
    while (_sg.mtl.idpool.release_queue_back != _sg.mtl.idpool.release_queue_front) {
        if (frame_index < _sg.mtl.idpool.release_queue[_sg.mtl.idpool.release_queue_back].frame_index) {
            // don't need to check further, release-items past this are too young
            break;
        }
        _sg_stats_add(metal.idpool.num_garbage_collected, 1);
        // safe to release this resource
        const int slot_index = _sg.mtl.idpool.release_queue[_sg.mtl.idpool.release_queue_back].slot_index;
        SOKOL_ASSERT((slot_index > 0) && (slot_index < _sg.mtl.idpool.num_slots));
        // note: the NSMutableArray takes ownership of its items, assigning an NSNull object will
        // release the object, no matter if using ARC or not
        SOKOL_ASSERT(_sg.mtl.idpool.pool[(NSUInteger)slot_index] != [NSNull null]);
        _sg.mtl.idpool.pool[(NSUInteger)slot_index] = [NSNull null];
        // put the now free pool index back on the free queue
        _sg_mtl_free_pool_slot(slot_index);
        // reset the release queue slot and advance the back index
        _sg.mtl.idpool.release_queue[_sg.mtl.idpool.release_queue_back].frame_index = 0;
        _sg.mtl.idpool.release_queue[_sg.mtl.idpool.release_queue_back].slot_index = _SG_MTL_INVALID_SLOT_INDEX;
        _sg.mtl.idpool.release_queue_back++;
        if (_sg.mtl.idpool.release_queue_back >= _sg.mtl.idpool.num_slots) {
            // wrap-around
            _sg.mtl.idpool.release_queue_back = 0;
        }
    }
}

_SOKOL_PRIVATE id _sg_mtl_id(int slot_index) {
    return _sg.mtl.idpool.pool[(NSUInteger)slot_index];
}

_SOKOL_PRIVATE void _sg_mtl_clear_state_cache(void) {
    _sg_clear(&_sg.mtl.state_cache, sizeof(_sg.mtl.state_cache));
}

// https://developer.apple.com/metal/Metal-Feature-Set-Tables.pdf
_SOKOL_PRIVATE void _sg_mtl_init_caps(void) {
    #if defined(_SG_TARGET_MACOS)
        _sg.backend = SG_BACKEND_METAL_MACOS;
    #elif defined(_SG_TARGET_IOS)
        #if defined(_SG_TARGET_IOS_SIMULATOR)
            _sg.backend = SG_BACKEND_METAL_SIMULATOR;
        #else
            _sg.backend = SG_BACKEND_METAL_IOS;
        #endif
    #endif
    _sg.features.origin_top_left = true;
    _sg.features.mrt_independent_blend_state = true;
    _sg.features.mrt_independent_write_mask = true;
    _sg.features.compute = true;
    _sg.features.msaa_image_bindings = true;

    _sg.features.image_clamp_to_border = false;
    #if (MAC_OS_X_VERSION_MAX_ALLOWED >= 120000) || (__IPHONE_OS_VERSION_MAX_ALLOWED >= 140000)
    if (@available(macOS 12.0, iOS 14.0, *)) {
        _sg.features.image_clamp_to_border = [_sg.mtl.device supportsFamily:MTLGPUFamilyApple7]
                                             || [_sg.mtl.device supportsFamily:MTLGPUFamilyMac2];
        #if (MAC_OS_X_VERSION_MAX_ALLOWED >= 130000) || (__IPHONE_OS_VERSION_MAX_ALLOWED >= 160000)
        if (!_sg.features.image_clamp_to_border) {
            if (@available(macOS 13.0, iOS 16.0, *)) {
                _sg.features.image_clamp_to_border = [_sg.mtl.device supportsFamily:MTLGPUFamilyMetal3];
            }
        }
        #endif
    }
    #endif

    #if defined(_SG_TARGET_MACOS)
        _sg.limits.max_image_size_2d = 16 * 1024;
        _sg.limits.max_image_size_cube = 16 * 1024;
        _sg.limits.max_image_size_3d = 2 * 1024;
        _sg.limits.max_image_size_array = 16 * 1024;
        _sg.limits.max_image_array_layers = 2 * 1024;
    #else
        // FIXME: newer iOS devices support 16k textures
        _sg.limits.max_image_size_2d = 8 * 1024;
        _sg.limits.max_image_size_cube = 8 * 1024;
        _sg.limits.max_image_size_3d = 2 * 1024;
        _sg.limits.max_image_size_array = 8 * 1024;
        _sg.limits.max_image_array_layers = 2 * 1024;
    #endif
    _sg.limits.max_vertex_attrs = SG_MAX_VERTEX_ATTRIBUTES;

    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R8SI]);
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16SN]);
    #else
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_R16]);
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_R16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_R16SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16F]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG8SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R32F]);
    #else
        _sg_pixelformat_sbr(&_sg.formats[SG_PIXELFORMAT_R32F]);
    #endif
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16SN]);
    #else
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_RG16]);
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_RG16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG16SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16F]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_SRGB8A8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_BGRA8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGB10A2]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGB9E5]);
        _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
        _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    #else
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGB9E5]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    #endif
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG32F]);
    #else
        _sg_pixelformat_sbr(&_sg.formats[SG_PIXELFORMAT_RG32F]);
    #endif
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16SN]);
    #else
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_RGBA16]);
        _sg_pixelformat_sfbr(&_sg.formats[SG_PIXELFORMAT_RGBA16SN]);
    #endif
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
        _sg_pixelformat_srm(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
        _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
    #else
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
    #endif
    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH]);
    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH_STENCIL]);
    #if defined(_SG_TARGET_MACOS)
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC1_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC2_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_SRGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_R]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_RSN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RG]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RGSN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBF]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBUF]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_SRGBA]);
    #else
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8A1]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGBA8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8A8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11SN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11SN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_SRGBA]);
    #endif

    // compute shader access (see: https://github.com/gpuweb/gpuweb/issues/513)
    // for now let's use the same conservative set on all backends even though
    // some backends are less restrictive
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
}

//-- main Metal backend state and functions ------------------------------------
_SOKOL_PRIVATE void _sg_mtl_setup_backend(const sg_desc* desc) {
    // assume already zero-initialized
    SOKOL_ASSERT(desc);
    SOKOL_ASSERT(desc->environment.metal.device);
    SOKOL_ASSERT(desc->uniform_buffer_size > 0);
    _sg_mtl_init_pool(desc);
    _sg_mtl_clear_state_cache();
    _sg.mtl.valid = true;
    _sg.mtl.ub_size = desc->uniform_buffer_size;
    _sg.mtl.sem = dispatch_semaphore_create(SG_NUM_INFLIGHT_FRAMES);
    _sg.mtl.device = (__bridge id<MTLDevice>) desc->environment.metal.device;
    _sg.mtl.cmd_queue = [_sg.mtl.device newCommandQueue];

    for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
        _sg.mtl.uniform_buffers[i] = [_sg.mtl.device
            newBufferWithLength:(NSUInteger)_sg.mtl.ub_size
            options:MTLResourceCPUCacheModeWriteCombined|MTLResourceStorageModeShared
        ];
        #if defined(SOKOL_DEBUG)
            _sg.mtl.uniform_buffers[i].label = [NSString stringWithFormat:@"sg-uniform-buffer.%d", i];
        #endif
    }

    if (desc->mtl_force_managed_storage_mode) {
        _sg.mtl.use_shared_storage_mode = false;
    } else if (@available(macOS 10.15, iOS 13.0, *)) {
        // on Intel Macs, always use managed resources even though the
        // device says it supports unified memory (because of texture restrictions)
        const bool is_apple_gpu = [_sg.mtl.device supportsFamily:MTLGPUFamilyApple1];
        if (!is_apple_gpu) {
            _sg.mtl.use_shared_storage_mode = false;
        } else {
            _sg.mtl.use_shared_storage_mode = true;
        }
    } else {
        #if defined(_SG_TARGET_MACOS)
            _sg.mtl.use_shared_storage_mode = false;
        #else
            _sg.mtl.use_shared_storage_mode = true;
        #endif
    }
    _sg_mtl_init_caps();
}

_SOKOL_PRIVATE void _sg_mtl_discard_backend(void) {
    SOKOL_ASSERT(_sg.mtl.valid);
    // wait for the last frame to finish
    for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
        dispatch_semaphore_wait(_sg.mtl.sem, DISPATCH_TIME_FOREVER);
    }
    // semaphore must be "relinquished" before destruction
    for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
        dispatch_semaphore_signal(_sg.mtl.sem);
    }
    _sg_mtl_garbage_collect(_sg.frame_index + SG_NUM_INFLIGHT_FRAMES + 2);
    _sg_mtl_destroy_pool();
    _sg.mtl.valid = false;

    _SG_OBJC_RELEASE(_sg.mtl.sem);
    _SG_OBJC_RELEASE(_sg.mtl.device);
    _SG_OBJC_RELEASE(_sg.mtl.cmd_queue);
    for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
        _SG_OBJC_RELEASE(_sg.mtl.uniform_buffers[i]);
    }
    // NOTE: MTLCommandBuffer, MTLRenderCommandEncoder and MTLComputeCommandEncoder are auto-released
    _sg.mtl.cmd_buffer = nil;
    _sg.mtl.render_cmd_encoder = nil;
    _sg.mtl.compute_cmd_encoder = nil;
}

_SOKOL_PRIVATE void _sg_mtl_reset_state_cache(void) {
    _sg_mtl_clear_state_cache();
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && desc);
    SOKOL_ASSERT(buf->cmn.size > 0);
    const bool injected = (0 != desc->mtl_buffers[0]);
    MTLResourceOptions mtl_options = _sg_mtl_buffer_resource_options(&buf->cmn.usage);
    for (int slot = 0; slot < buf->cmn.num_slots; slot++) {
        id<MTLBuffer> mtl_buf;
        if (injected) {
            SOKOL_ASSERT(desc->mtl_buffers[slot]);
            mtl_buf = (__bridge id<MTLBuffer>) desc->mtl_buffers[slot];
        } else {
            if (desc->data.ptr) {
                SOKOL_ASSERT(desc->data.size > 0);
                mtl_buf = [_sg.mtl.device newBufferWithBytes:desc->data.ptr length:(NSUInteger)buf->cmn.size options:mtl_options];
            } else {
                mtl_buf = [_sg.mtl.device newBufferWithLength:(NSUInteger)buf->cmn.size options:mtl_options];
            }
            if (nil == mtl_buf) {
                _SG_ERROR(METAL_CREATE_BUFFER_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
        }
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                mtl_buf.label = [NSString stringWithFormat:@"%s.%d", desc->label, slot];
            }
        #endif
        buf->mtl.buf[slot] = _sg_mtl_add_resource(mtl_buf);
        _SG_OBJC_RELEASE(mtl_buf);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_mtl_discard_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    for (int slot = 0; slot < buf->cmn.num_slots; slot++) {
        // it's valid to call release resource with '0'
        _sg_mtl_release_resource(_sg.frame_index, buf->mtl.buf[slot]);
    }
}

_SOKOL_PRIVATE void _sg_mtl_copy_image_data(const _sg_image_t* img, __unsafe_unretained id<MTLTexture> mtl_tex, const sg_image_data* data) {
    const int num_faces = (img->cmn.type == SG_IMAGETYPE_CUBE) ? 6:1;
    const int num_slices = (img->cmn.type == SG_IMAGETYPE_ARRAY) ? img->cmn.num_slices : 1;
    for (int face_index = 0; face_index < num_faces; face_index++) {
        for (int mip_index = 0; mip_index < img->cmn.num_mipmaps; mip_index++) {
            SOKOL_ASSERT(data->subimage[face_index][mip_index].ptr);
            SOKOL_ASSERT(data->subimage[face_index][mip_index].size > 0);
            const uint8_t* data_ptr = (const uint8_t*)data->subimage[face_index][mip_index].ptr;
            const int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
            const int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
            int bytes_per_row = _sg_row_pitch(img->cmn.pixel_format, mip_width, 1);
            int bytes_per_slice = _sg_surface_pitch(img->cmn.pixel_format, mip_width, mip_height, 1);
            /* bytesPerImage special case: https://developer.apple.com/documentation/metal/mtltexture/1515679-replaceregion

                "Supply a nonzero value only when you copy data to a MTLTextureType3D type texture"
            */
            MTLRegion region;
            int bytes_per_image;
            if (img->cmn.type == SG_IMAGETYPE_3D) {
                const int mip_depth = _sg_miplevel_dim(img->cmn.num_slices, mip_index);
                region = MTLRegionMake3D(0, 0, 0, (NSUInteger)mip_width, (NSUInteger)mip_height, (NSUInteger)mip_depth);
                bytes_per_image = bytes_per_slice;
                // FIXME: apparently the minimal bytes_per_image size for 3D texture is 4 KByte... somehow need to handle this
            } else {
                region = MTLRegionMake2D(0, 0, (NSUInteger)mip_width, (NSUInteger)mip_height);
                bytes_per_image = 0;
            }

            for (int slice_index = 0; slice_index < num_slices; slice_index++) {
                const int mtl_slice_index = (img->cmn.type == SG_IMAGETYPE_CUBE) ? face_index : slice_index;
                const int slice_offset = slice_index * bytes_per_slice;
                SOKOL_ASSERT((slice_offset + bytes_per_slice) <= (int)data->subimage[face_index][mip_index].size);
                [mtl_tex replaceRegion:region
                    mipmapLevel:(NSUInteger)mip_index
                    slice:(NSUInteger)mtl_slice_index
                    withBytes:data_ptr + slice_offset
                    bytesPerRow:(NSUInteger)bytes_per_row
                    bytesPerImage:(NSUInteger)bytes_per_image];
            }
        }
    }
}

_SOKOL_PRIVATE bool _sg_mtl_init_texdesc(MTLTextureDescriptor* mtl_desc, _sg_image_t* img) {
    mtl_desc.textureType = _sg_mtl_texture_type(img->cmn.type, img->cmn.sample_count > 1);
    mtl_desc.pixelFormat = _sg_mtl_pixel_format(img->cmn.pixel_format);
    if (MTLPixelFormatInvalid == mtl_desc.pixelFormat) {
        _SG_ERROR(METAL_TEXTURE_FORMAT_NOT_SUPPORTED);
        return false;
    }
    mtl_desc.width = (NSUInteger)img->cmn.width;
    mtl_desc.height = (NSUInteger)img->cmn.height;
    if (SG_IMAGETYPE_3D == img->cmn.type) {
        mtl_desc.depth = (NSUInteger)img->cmn.num_slices;
    } else {
        mtl_desc.depth = 1;
    }
    mtl_desc.mipmapLevelCount = (NSUInteger)img->cmn.num_mipmaps;
    if (SG_IMAGETYPE_ARRAY == img->cmn.type) {
        mtl_desc.arrayLength = (NSUInteger)img->cmn.num_slices;
    } else {
        mtl_desc.arrayLength = 1;
    }
    mtl_desc.sampleCount = (NSUInteger)img->cmn.sample_count;

    MTLTextureUsage mtl_tex_usage = MTLTextureUsageShaderRead;
    if (img->cmn.usage.render_attachment) {
        mtl_tex_usage |= MTLTextureUsageRenderTarget;
    } else if (img->cmn.usage.storage_attachment) {
        mtl_tex_usage |= MTLTextureUsageShaderWrite;
    }
    mtl_desc.usage = mtl_tex_usage;

    MTLResourceOptions mtl_res_options = 0;
    if (img->cmn.usage.render_attachment || img->cmn.usage.storage_attachment) {
        mtl_res_options |= MTLResourceStorageModePrivate;
    } else {
        mtl_res_options |= _sg_mtl_resource_options_storage_mode_managed_or_shared();
        if (!img->cmn.usage.immutable) {
            mtl_res_options |= MTLResourceCPUCacheModeWriteCombined;
        }
    }
    mtl_desc.resourceOptions = mtl_res_options;

    return true;
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && desc);
    const bool injected = (0 != desc->mtl_textures[0]);

    // first initialize all Metal resource pool slots to 'empty'
    for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
        img->mtl.tex[i] = _sg_mtl_add_resource(nil);
    }

    // initialize a Metal texture descriptor
    MTLTextureDescriptor* mtl_desc = [[MTLTextureDescriptor alloc] init];
    if (!_sg_mtl_init_texdesc(mtl_desc, img)) {
        _SG_OBJC_RELEASE(mtl_desc);
        return SG_RESOURCESTATE_FAILED;
    }
    for (int slot = 0; slot < img->cmn.num_slots; slot++) {
        id<MTLTexture> mtl_tex;
        if (injected) {
            SOKOL_ASSERT(desc->mtl_textures[slot]);
            mtl_tex = (__bridge id<MTLTexture>) desc->mtl_textures[slot];
        } else {
            mtl_tex = [_sg.mtl.device newTextureWithDescriptor:mtl_desc];
            if (nil == mtl_tex) {
                _SG_OBJC_RELEASE(mtl_desc);
                _SG_ERROR(METAL_CREATE_TEXTURE_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
            if (desc->data.subimage[0][0].ptr) {
                _sg_mtl_copy_image_data(img, mtl_tex, &desc->data);
            }
        }
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                mtl_tex.label = [NSString stringWithFormat:@"%s.%d", desc->label, slot];
            }
        #endif
        img->mtl.tex[slot] = _sg_mtl_add_resource(mtl_tex);
        _SG_OBJC_RELEASE(mtl_tex);
    }
    _SG_OBJC_RELEASE(mtl_desc);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_mtl_discard_image(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    // it's valid to call release resource with a 'null resource'
    for (int slot = 0; slot < img->cmn.num_slots; slot++) {
        _sg_mtl_release_resource(_sg.frame_index, img->mtl.tex[slot]);
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && desc);
    id<MTLSamplerState> mtl_smp;
    const bool injected = (0 != desc->mtl_sampler);
    if (injected) {
        SOKOL_ASSERT(desc->mtl_sampler);
        mtl_smp = (__bridge id<MTLSamplerState>) desc->mtl_sampler;
    } else {
        MTLSamplerDescriptor* mtl_desc = [[MTLSamplerDescriptor alloc] init];
        mtl_desc.sAddressMode = _sg_mtl_address_mode(desc->wrap_u);
        mtl_desc.tAddressMode = _sg_mtl_address_mode(desc->wrap_v);
        mtl_desc.rAddressMode = _sg_mtl_address_mode(desc->wrap_w);
        if (_sg.features.image_clamp_to_border) {
            if (@available(macOS 12.0, iOS 14.0, *)) {
                mtl_desc.borderColor  = _sg_mtl_border_color(desc->border_color);
            }
        }
        mtl_desc.minFilter = _sg_mtl_minmag_filter(desc->min_filter);
        mtl_desc.magFilter = _sg_mtl_minmag_filter(desc->mag_filter);
        mtl_desc.mipFilter = _sg_mtl_mipmap_filter(desc->mipmap_filter);
        mtl_desc.lodMinClamp = desc->min_lod;
        mtl_desc.lodMaxClamp = desc->max_lod;
        // FIXME: lodAverage?
        mtl_desc.maxAnisotropy = desc->max_anisotropy;
        mtl_desc.normalizedCoordinates = YES;
        mtl_desc.compareFunction = _sg_mtl_compare_func(desc->compare);
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                mtl_desc.label = [NSString stringWithUTF8String:desc->label];
            }
        #endif
        mtl_smp = [_sg.mtl.device newSamplerStateWithDescriptor:mtl_desc];
        _SG_OBJC_RELEASE(mtl_desc);
        if (nil == mtl_smp) {
            _SG_ERROR(METAL_CREATE_SAMPLER_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    smp->mtl.sampler_state = _sg_mtl_add_resource(mtl_smp);
    _SG_OBJC_RELEASE(mtl_smp);
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_mtl_discard_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    // it's valid to call release resource with a 'null resource'
    _sg_mtl_release_resource(_sg.frame_index, smp->mtl.sampler_state);
}

_SOKOL_PRIVATE id<MTLLibrary> _sg_mtl_compile_library(const char* src) {
    NSError* err = NULL;
    id<MTLLibrary> lib = [_sg.mtl.device
        newLibraryWithSource:[NSString stringWithUTF8String:src]
        options:nil
        error:&err
    ];
    if (err) {
        _SG_ERROR(METAL_SHADER_COMPILATION_FAILED);
        _SG_LOGMSG(METAL_SHADER_COMPILATION_OUTPUT, [err.localizedDescription UTF8String]);
    }
    return lib;
}

_SOKOL_PRIVATE id<MTLLibrary> _sg_mtl_library_from_bytecode(const void* ptr, size_t num_bytes) {
    NSError* err = NULL;
    dispatch_data_t lib_data = dispatch_data_create(ptr, num_bytes, NULL, DISPATCH_DATA_DESTRUCTOR_DEFAULT);
    id<MTLLibrary> lib = [_sg.mtl.device newLibraryWithData:lib_data error:&err];
    if (err) {
        _SG_ERROR(METAL_SHADER_CREATION_FAILED);
        _SG_LOGMSG(METAL_SHADER_COMPILATION_OUTPUT, [err.localizedDescription UTF8String]);
    }
    _SG_OBJC_RELEASE(lib_data);
    return lib;
}

_SOKOL_PRIVATE bool _sg_mtl_create_shader_func(const sg_shader_function* func, const char* label, const char* label_ext, _sg_mtl_shader_func_t* res) {
    SOKOL_ASSERT(res->mtl_lib == _SG_MTL_INVALID_SLOT_INDEX);
    SOKOL_ASSERT(res->mtl_func == _SG_MTL_INVALID_SLOT_INDEX);
    id<MTLLibrary> mtl_lib = nil;
    if (func->bytecode.ptr) {
        SOKOL_ASSERT(func->bytecode.size > 0);
        mtl_lib = _sg_mtl_library_from_bytecode(func->bytecode.ptr, func->bytecode.size);
    } else if (func->source) {
        mtl_lib = _sg_mtl_compile_library(func->source);
    }
    if (mtl_lib == nil) {
        return false;
    }
    #if defined(SOKOL_DEBUG)
    if (label) {
        SOKOL_ASSERT(label_ext);
        mtl_lib.label = [NSString stringWithFormat:@"%s.%s", label, label_ext];
    }
    #else
    _SOKOL_UNUSED(label);
    _SOKOL_UNUSED(label_ext);
    #endif
    SOKOL_ASSERT(func->entry);
    id<MTLFunction> mtl_func = [mtl_lib newFunctionWithName:[NSString stringWithUTF8String:func->entry]];
    if (mtl_func == nil) {
        _SG_ERROR(METAL_SHADER_ENTRY_NOT_FOUND);
        _SG_OBJC_RELEASE(mtl_lib);
        return false;
    }
    res->mtl_lib = _sg_mtl_add_resource(mtl_lib);
    res->mtl_func = _sg_mtl_add_resource(mtl_func);
    _SG_OBJC_RELEASE(mtl_lib);
    _SG_OBJC_RELEASE(mtl_func);
    return true;
}

_SOKOL_PRIVATE void _sg_mtl_discard_shader_func(const _sg_mtl_shader_func_t* func) {
    // it is valid to call _sg_mtl_release_resource with a 'null resource'
    _sg_mtl_release_resource(_sg.frame_index, func->mtl_func);
    _sg_mtl_release_resource(_sg.frame_index, func->mtl_lib);
}

// NOTE: this is an out-of-range check for MSL bindslots that's also active in release mode
_SOKOL_PRIVATE bool _sg_mtl_ensure_msl_bindslot_ranges(const sg_shader_desc* desc) {
    SOKOL_ASSERT(desc);
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (desc->uniform_blocks[i].msl_buffer_n >= _SG_MTL_MAX_STAGE_UB_BINDINGS) {
            _SG_ERROR(METAL_UNIFORMBLOCK_MSL_BUFFER_SLOT_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (desc->images[i].msl_texture_n >= _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS) {
            _SG_ERROR(METAL_IMAGE_MSL_TEXTURE_SLOT_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (desc->samplers[i].msl_sampler_n >= _SG_MTL_MAX_STAGE_SAMPLER_BINDINGS) {
            _SG_ERROR(METAL_SAMPLER_MSL_SAMPLER_SLOT_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (desc->storage_buffers[i].msl_buffer_n >= _SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS) {
            _SG_ERROR(METAL_STORAGEBUFFER_MSL_BUFFER_SLOT_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (desc->storage_images[i].msl_texture_n >= _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS) {
            _SG_ERROR(METAL_STORAGEIMAGE_MSL_TEXTURE_SLOT_OUT_OF_RANGE);
            return false;
        }
    }
    return true;
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && desc);

    // do a MSL bindslot range check also in release mode, and if that fails,
    // also fail shader creation
    if (!_sg_mtl_ensure_msl_bindslot_ranges(desc)) {
        return SG_RESOURCESTATE_FAILED;
    }

    shd->mtl.threads_per_threadgroup = MTLSizeMake(
        (NSUInteger)desc->mtl_threads_per_threadgroup.x,
        (NSUInteger)desc->mtl_threads_per_threadgroup.y,
        (NSUInteger)desc->mtl_threads_per_threadgroup.z);

    // copy resource bindslot mappings
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        shd->mtl.ub_buffer_n[i] = desc->uniform_blocks[i].msl_buffer_n;
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        shd->mtl.img_texture_n[i] = desc->images[i].msl_texture_n;
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        shd->mtl.smp_sampler_n[i] = desc->samplers[i].msl_sampler_n;
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        shd->mtl.sbuf_buffer_n[i] = desc->storage_buffers[i].msl_buffer_n;
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        shd->mtl.simg_texture_n[i] = desc->storage_images[i].msl_texture_n;
    }

    // create metal library and function objects
    bool shd_valid = true;
    if (desc->vertex_func.source || desc->vertex_func.bytecode.ptr) {
        shd_valid &= _sg_mtl_create_shader_func(&desc->vertex_func, desc->label, "vs", &shd->mtl.vertex_func);
    }
    if (desc->fragment_func.source || desc->fragment_func.bytecode.ptr) {
        shd_valid &= _sg_mtl_create_shader_func(&desc->fragment_func, desc->label, "fs", &shd->mtl.fragment_func);
    }
    if (desc->compute_func.source || desc->compute_func.bytecode.ptr) {
        shd_valid &= _sg_mtl_create_shader_func(&desc->compute_func, desc->label, "cs", &shd->mtl.compute_func);
    }
    if (!shd_valid) {
        _sg_mtl_discard_shader_func(&shd->mtl.vertex_func);
        _sg_mtl_discard_shader_func(&shd->mtl.fragment_func);
        _sg_mtl_discard_shader_func(&shd->mtl.compute_func);
    }
    return shd_valid ? SG_RESOURCESTATE_VALID : SG_RESOURCESTATE_FAILED;
}

_SOKOL_PRIVATE void _sg_mtl_discard_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    _sg_mtl_discard_shader_func(&shd->mtl.vertex_func);
    _sg_mtl_discard_shader_func(&shd->mtl.fragment_func);
    _sg_mtl_discard_shader_func(&shd->mtl.compute_func);
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_pipeline(_sg_pipeline_t* pip, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && desc);
    _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
    if (pip->cmn.is_compute) {
        NSError* err = NULL;
        MTLComputePipelineDescriptor* cp_desc = [[MTLComputePipelineDescriptor alloc] init];
        cp_desc.computeFunction = _sg_mtl_id(shd->mtl.compute_func.mtl_func);
        cp_desc.threadGroupSizeIsMultipleOfThreadExecutionWidth = true;
        for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
            const sg_shader_stage stage = shd->cmn.storage_buffers[i].stage;
            SOKOL_ASSERT((stage != SG_SHADERSTAGE_VERTEX) && (stage != SG_SHADERSTAGE_FRAGMENT));
            if ((stage == SG_SHADERSTAGE_COMPUTE) && shd->cmn.storage_buffers[i].readonly) {
                const NSUInteger mtl_slot = shd->mtl.sbuf_buffer_n[i];
                cp_desc.buffers[mtl_slot].mutability = MTLMutabilityImmutable;
            }
        }
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                cp_desc.label = [NSString stringWithFormat:@"%s", desc->label];
            }
        #endif
        id<MTLComputePipelineState> mtl_cps = [_sg.mtl.device
            newComputePipelineStateWithDescriptor:cp_desc
            options:MTLPipelineOptionNone
            reflection:nil
            error:&err];
        _SG_OBJC_RELEASE(cp_desc);
        if (nil == mtl_cps) {
            SOKOL_ASSERT(err);
            _SG_ERROR(METAL_CREATE_CPS_FAILED);
            _SG_LOGMSG(METAL_CREATE_CPS_OUTPUT, [err.localizedDescription UTF8String]);
            return SG_RESOURCESTATE_FAILED;
        }
        pip->mtl.cps = _sg_mtl_add_resource(mtl_cps);
        _SG_OBJC_RELEASE(mtl_cps);
        pip->mtl.threads_per_threadgroup = shd->mtl.threads_per_threadgroup;
    } else {
        sg_primitive_type prim_type = desc->primitive_type;
        pip->mtl.prim_type = _sg_mtl_primitive_type(prim_type);
        pip->mtl.index_size = _sg_mtl_index_size(pip->cmn.index_type);
        if (SG_INDEXTYPE_NONE != pip->cmn.index_type) {
            pip->mtl.index_type = _sg_mtl_index_type(pip->cmn.index_type);
        }
        pip->mtl.cull_mode = _sg_mtl_cull_mode(desc->cull_mode);
        pip->mtl.winding = _sg_mtl_winding(desc->face_winding);
        pip->mtl.stencil_ref = desc->stencil.ref;

        // create vertex-descriptor
        MTLVertexDescriptor* vtx_desc = [MTLVertexDescriptor vertexDescriptor];
        for (NSUInteger attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
            const sg_vertex_attr_state* a_state = &desc->layout.attrs[attr_index];
            if (a_state->format == SG_VERTEXFORMAT_INVALID) {
                break;
            }
            SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
            SOKOL_ASSERT(pip->cmn.vertex_buffer_layout_active[a_state->buffer_index]);
            vtx_desc.attributes[attr_index].format = _sg_mtl_vertex_format(a_state->format);
            vtx_desc.attributes[attr_index].offset = (NSUInteger)a_state->offset;
            vtx_desc.attributes[attr_index].bufferIndex = _sg_mtl_vertexbuffer_bindslot((size_t)a_state->buffer_index);
        }
        for (NSUInteger layout_index = 0; layout_index < SG_MAX_VERTEXBUFFER_BINDSLOTS; layout_index++) {
            if (pip->cmn.vertex_buffer_layout_active[layout_index]) {
                const sg_vertex_buffer_layout_state* l_state = &desc->layout.buffers[layout_index];
                const NSUInteger mtl_vb_slot = _sg_mtl_vertexbuffer_bindslot(layout_index);
                SOKOL_ASSERT(l_state->stride > 0);
                vtx_desc.layouts[mtl_vb_slot].stride = (NSUInteger)l_state->stride;
                vtx_desc.layouts[mtl_vb_slot].stepFunction = _sg_mtl_step_function(l_state->step_func);
                vtx_desc.layouts[mtl_vb_slot].stepRate = (NSUInteger)l_state->step_rate;
                if (SG_VERTEXSTEP_PER_INSTANCE == l_state->step_func) {
                    // NOTE: not actually used in _sg_mtl_draw()
                    pip->cmn.use_instanced_draw = true;
                }
            }
        }

        // render-pipeline descriptor
        MTLRenderPipelineDescriptor* rp_desc = [[MTLRenderPipelineDescriptor alloc] init];
        rp_desc.vertexDescriptor = vtx_desc;
        SOKOL_ASSERT(shd->mtl.vertex_func.mtl_func != _SG_MTL_INVALID_SLOT_INDEX);
        rp_desc.vertexFunction = _sg_mtl_id(shd->mtl.vertex_func.mtl_func);
        SOKOL_ASSERT(shd->mtl.fragment_func.mtl_func != _SG_MTL_INVALID_SLOT_INDEX);
        rp_desc.fragmentFunction = _sg_mtl_id(shd->mtl.fragment_func.mtl_func);
        rp_desc.rasterSampleCount = (NSUInteger)desc->sample_count;
        rp_desc.alphaToCoverageEnabled = desc->alpha_to_coverage_enabled;
        rp_desc.alphaToOneEnabled = NO;
        rp_desc.rasterizationEnabled = YES;
        rp_desc.depthAttachmentPixelFormat = _sg_mtl_pixel_format(desc->depth.pixel_format);
        if (desc->depth.pixel_format == SG_PIXELFORMAT_DEPTH_STENCIL) {
            rp_desc.stencilAttachmentPixelFormat = _sg_mtl_pixel_format(desc->depth.pixel_format);
        }
        for (NSUInteger i = 0; i < (NSUInteger)desc->color_count; i++) {
            SOKOL_ASSERT(i < SG_MAX_COLOR_ATTACHMENTS);
            const sg_color_target_state* cs = &desc->colors[i];
            rp_desc.colorAttachments[i].pixelFormat = _sg_mtl_pixel_format(cs->pixel_format);
            rp_desc.colorAttachments[i].writeMask = _sg_mtl_color_write_mask(cs->write_mask);
            rp_desc.colorAttachments[i].blendingEnabled = cs->blend.enabled;
            rp_desc.colorAttachments[i].alphaBlendOperation = _sg_mtl_blend_op(cs->blend.op_alpha);
            rp_desc.colorAttachments[i].rgbBlendOperation = _sg_mtl_blend_op(cs->blend.op_rgb);
            rp_desc.colorAttachments[i].destinationAlphaBlendFactor = _sg_mtl_blend_factor(cs->blend.dst_factor_alpha);
            rp_desc.colorAttachments[i].destinationRGBBlendFactor = _sg_mtl_blend_factor(cs->blend.dst_factor_rgb);
            rp_desc.colorAttachments[i].sourceAlphaBlendFactor = _sg_mtl_blend_factor(cs->blend.src_factor_alpha);
            rp_desc.colorAttachments[i].sourceRGBBlendFactor = _sg_mtl_blend_factor(cs->blend.src_factor_rgb);
        }
        // set buffer mutability for all read-only buffers (vertex buffers and read-only storage buffers)
        for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
            if (pip->cmn.vertex_buffer_layout_active[i]) {
                const NSUInteger mtl_slot = _sg_mtl_vertexbuffer_bindslot(i);
                rp_desc.vertexBuffers[mtl_slot].mutability = MTLMutabilityImmutable;
            }
        }
        for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
            const NSUInteger mtl_slot = shd->mtl.sbuf_buffer_n[i];
            const sg_shader_stage stage = shd->cmn.storage_buffers[i].stage;
            SOKOL_ASSERT(stage != SG_SHADERSTAGE_COMPUTE);
            if (stage == SG_SHADERSTAGE_VERTEX) {
                SOKOL_ASSERT(shd->cmn.storage_buffers[i].readonly);
                rp_desc.vertexBuffers[mtl_slot].mutability = MTLMutabilityImmutable;
            } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
                SOKOL_ASSERT(shd->cmn.storage_buffers[i].readonly);
                rp_desc.fragmentBuffers[mtl_slot].mutability = MTLMutabilityImmutable;
            }
        }
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                rp_desc.label = [NSString stringWithFormat:@"%s", desc->label];
            }
        #endif
        NSError* err = NULL;
        id<MTLRenderPipelineState> mtl_rps = [_sg.mtl.device newRenderPipelineStateWithDescriptor:rp_desc error:&err];
        _SG_OBJC_RELEASE(rp_desc);
        if (nil == mtl_rps) {
            SOKOL_ASSERT(err);
            _SG_ERROR(METAL_CREATE_RPS_FAILED);
            _SG_LOGMSG(METAL_CREATE_RPS_OUTPUT, [err.localizedDescription UTF8String]);
            return SG_RESOURCESTATE_FAILED;
        }
        pip->mtl.rps = _sg_mtl_add_resource(mtl_rps);
        _SG_OBJC_RELEASE(mtl_rps);

        // depth-stencil-state
        MTLDepthStencilDescriptor* ds_desc = [[MTLDepthStencilDescriptor alloc] init];
        ds_desc.depthCompareFunction = _sg_mtl_compare_func(desc->depth.compare);
        ds_desc.depthWriteEnabled = desc->depth.write_enabled;
        if (desc->stencil.enabled) {
            const sg_stencil_face_state* sb = &desc->stencil.back;
            ds_desc.backFaceStencil = [[MTLStencilDescriptor alloc] init];
            ds_desc.backFaceStencil.stencilFailureOperation = _sg_mtl_stencil_op(sb->fail_op);
            ds_desc.backFaceStencil.depthFailureOperation = _sg_mtl_stencil_op(sb->depth_fail_op);
            ds_desc.backFaceStencil.depthStencilPassOperation = _sg_mtl_stencil_op(sb->pass_op);
            ds_desc.backFaceStencil.stencilCompareFunction = _sg_mtl_compare_func(sb->compare);
            ds_desc.backFaceStencil.readMask = desc->stencil.read_mask;
            ds_desc.backFaceStencil.writeMask = desc->stencil.write_mask;
            const sg_stencil_face_state* sf = &desc->stencil.front;
            ds_desc.frontFaceStencil = [[MTLStencilDescriptor alloc] init];
            ds_desc.frontFaceStencil.stencilFailureOperation = _sg_mtl_stencil_op(sf->fail_op);
            ds_desc.frontFaceStencil.depthFailureOperation = _sg_mtl_stencil_op(sf->depth_fail_op);
            ds_desc.frontFaceStencil.depthStencilPassOperation = _sg_mtl_stencil_op(sf->pass_op);
            ds_desc.frontFaceStencil.stencilCompareFunction = _sg_mtl_compare_func(sf->compare);
            ds_desc.frontFaceStencil.readMask = desc->stencil.read_mask;
            ds_desc.frontFaceStencil.writeMask = desc->stencil.write_mask;
        }
        #if defined(SOKOL_DEBUG)
            if (desc->label) {
                ds_desc.label = [NSString stringWithFormat:@"%s.dss", desc->label];
            }
        #endif
        id<MTLDepthStencilState> mtl_dss = [_sg.mtl.device newDepthStencilStateWithDescriptor:ds_desc];
        _SG_OBJC_RELEASE(ds_desc);
        if (nil == mtl_dss) {
            _SG_ERROR(METAL_CREATE_DSS_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        pip->mtl.dss = _sg_mtl_add_resource(mtl_dss);
        _SG_OBJC_RELEASE(mtl_dss);
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_mtl_discard_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    // it's valid to call release resource with a 'null resource'
    _sg_mtl_release_resource(_sg.frame_index, pip->mtl.cps);
    _sg_mtl_release_resource(_sg.frame_index, pip->mtl.rps);
    _sg_mtl_release_resource(_sg.frame_index, pip->mtl.dss);
}

_SOKOL_PRIVATE sg_resource_state _sg_mtl_create_attachments(_sg_attachments_t* atts, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && desc);
    _SOKOL_UNUSED(desc);

    // create texture views for storage attachments
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        const _sg_image_t* stg_img = _sg_image_ref_ptr_or_null(&atts->cmn.storages[i].image);
        if (stg_img) {
            id<MTLTexture> mtl_tex_view = [_sg_mtl_id(stg_img->mtl.tex[0])
                newTextureViewWithPixelFormat: _sg_mtl_pixel_format(stg_img->cmn.pixel_format)
                textureType: _sg_mtl_texture_type(stg_img->cmn.type, false)
                levels: NSMakeRange((NSUInteger)atts->cmn.storages[i].mip_level, 1)
                slices: NSMakeRange((NSUInteger)atts->cmn.storages[i].slice, 1)];
            atts->mtl.storage_views[i] = _sg_mtl_add_resource(mtl_tex_view);
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_mtl_discard_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    _SOKOL_UNUSED(atts);
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        // it's valid to call _sg_mtl_release_resource with a null handle
        _sg_mtl_release_resource(_sg.frame_index, atts->mtl.storage_views[i]);
    }
}

_SOKOL_PRIVATE void _sg_mtl_bind_uniform_buffers(void) {
    // In the Metal backend, uniform buffer bindings happen once in sg_begin_pass() and
    // remain valid for the entire pass. Only binding offsets will be updated
    // in sg_apply_uniforms()
    if (_sg.cur_pass.is_compute) {
        SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
        for (size_t slot = 0; slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS; slot++) {
            [_sg.mtl.compute_cmd_encoder
                setBuffer:_sg.mtl.uniform_buffers[_sg.mtl.cur_frame_rotate_index]
                offset:0
                atIndex:slot];
        }
    } else {
        SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
        for (size_t slot = 0; slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS; slot++) {
            [_sg.mtl.render_cmd_encoder
                setVertexBuffer:_sg.mtl.uniform_buffers[_sg.mtl.cur_frame_rotate_index]
                offset:0
                atIndex:slot];
            [_sg.mtl.render_cmd_encoder
                setFragmentBuffer:_sg.mtl.uniform_buffers[_sg.mtl.cur_frame_rotate_index]
                offset:0
                atIndex:slot];
        }
    }
}

_SOKOL_PRIVATE void _sg_mtl_begin_compute_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass); (void)pass;
    SOKOL_ASSERT(nil != _sg.mtl.cmd_buffer);
    SOKOL_ASSERT(nil == _sg.mtl.compute_cmd_encoder);
    SOKOL_ASSERT(nil == _sg.mtl.render_cmd_encoder);

    _sg.mtl.compute_cmd_encoder = [_sg.mtl.cmd_buffer computeCommandEncoder];
    if (nil == _sg.mtl.compute_cmd_encoder) {
        _sg.cur_pass.valid = false;
        return;
    }

    #if defined(SOKOL_DEBUG)
    if (pass->label) {
        _sg.mtl.compute_cmd_encoder.label = [NSString stringWithUTF8String:pass->label];
    }
    #endif
}

_SOKOL_PRIVATE void _sg_mtl_begin_render_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass);
    SOKOL_ASSERT(nil != _sg.mtl.cmd_buffer);
    SOKOL_ASSERT(nil == _sg.mtl.render_cmd_encoder);
    SOKOL_ASSERT(nil == _sg.mtl.compute_cmd_encoder);

    const sg_swapchain* swapchain = &pass->swapchain;
    const sg_pass_action* action = &pass->action;

    MTLRenderPassDescriptor* pass_desc = [MTLRenderPassDescriptor renderPassDescriptor];
    SOKOL_ASSERT(pass_desc);
    const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);
    if (atts) {
        // setup pass descriptor for offscreen rendering
        for (NSUInteger i = 0; i < (NSUInteger)atts->cmn.num_colors; i++) {
            const _sg_attachment_common_t* cmn_color_att = &atts->cmn.colors[i];
            const _sg_image_t* color_att_img = _sg_image_ref_ptr(&cmn_color_att->image);
            const _sg_attachment_common_t* cmn_resolve_att = &atts->cmn.resolves[i];
            const _sg_image_t* resolve_att_img = _sg_image_ref_ptr_or_null(&cmn_resolve_att->image);
            SOKOL_ASSERT(color_att_img->slot.state == SG_RESOURCESTATE_VALID);
            SOKOL_ASSERT(color_att_img->mtl.tex[color_att_img->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
            pass_desc.colorAttachments[i].loadAction = _sg_mtl_load_action(action->colors[i].load_action);
            pass_desc.colorAttachments[i].storeAction = _sg_mtl_store_action(action->colors[i].store_action, resolve_att_img != 0);
            sg_color c = action->colors[i].clear_value;
            pass_desc.colorAttachments[i].clearColor = MTLClearColorMake(c.r, c.g, c.b, c.a);
            pass_desc.colorAttachments[i].texture = _sg_mtl_id(color_att_img->mtl.tex[color_att_img->cmn.active_slot]);
            pass_desc.colorAttachments[i].level = (NSUInteger)cmn_color_att->mip_level;
            switch (color_att_img->cmn.type) {
                case SG_IMAGETYPE_CUBE:
                case SG_IMAGETYPE_ARRAY:
                    pass_desc.colorAttachments[i].slice = (NSUInteger)cmn_color_att->slice;
                    break;
                case SG_IMAGETYPE_3D:
                    pass_desc.colorAttachments[i].depthPlane = (NSUInteger)cmn_color_att->slice;
                    break;
                default: break;
            }
            if (resolve_att_img) {
                SOKOL_ASSERT(resolve_att_img->slot.state == SG_RESOURCESTATE_VALID);
                SOKOL_ASSERT(resolve_att_img->mtl.tex[resolve_att_img->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
                pass_desc.colorAttachments[i].resolveTexture = _sg_mtl_id(resolve_att_img->mtl.tex[resolve_att_img->cmn.active_slot]);
                pass_desc.colorAttachments[i].resolveLevel = (NSUInteger)cmn_resolve_att->mip_level;
                switch (resolve_att_img->cmn.type) {
                    case SG_IMAGETYPE_CUBE:
                    case SG_IMAGETYPE_ARRAY:
                        pass_desc.colorAttachments[i].resolveSlice = (NSUInteger)cmn_resolve_att->slice;
                        break;
                    case SG_IMAGETYPE_3D:
                        pass_desc.colorAttachments[i].resolveDepthPlane = (NSUInteger)cmn_resolve_att->slice;
                        break;
                    default: break;
                }
            }
        }
        const _sg_image_t* ds_att_img = _sg_image_ref_ptr_or_null(&atts->cmn.depth_stencil.image);
        if (0 != ds_att_img) {
            SOKOL_ASSERT(ds_att_img->slot.state == SG_RESOURCESTATE_VALID);
            SOKOL_ASSERT(ds_att_img->mtl.tex[ds_att_img->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
            pass_desc.depthAttachment.texture = _sg_mtl_id(ds_att_img->mtl.tex[ds_att_img->cmn.active_slot]);
            pass_desc.depthAttachment.loadAction = _sg_mtl_load_action(action->depth.load_action);
            pass_desc.depthAttachment.storeAction = _sg_mtl_store_action(action->depth.store_action, false);
            pass_desc.depthAttachment.clearDepth = action->depth.clear_value;
            const _sg_attachment_common_t* cmn_ds_att = &atts->cmn.depth_stencil;
            switch (ds_att_img->cmn.type) {
                case SG_IMAGETYPE_CUBE:
                case SG_IMAGETYPE_ARRAY:
                    pass_desc.depthAttachment.slice = (NSUInteger)cmn_ds_att->slice;
                    break;
                case SG_IMAGETYPE_3D:
                    pass_desc.depthAttachment.resolveDepthPlane = (NSUInteger)cmn_ds_att->slice;
                    break;
                default: break;
            }
            if (_sg_is_depth_stencil_format(ds_att_img->cmn.pixel_format)) {
                pass_desc.stencilAttachment.texture = _sg_mtl_id(ds_att_img->mtl.tex[ds_att_img->cmn.active_slot]);
                pass_desc.stencilAttachment.loadAction = _sg_mtl_load_action(action->stencil.load_action);
                pass_desc.stencilAttachment.storeAction = _sg_mtl_store_action(action->depth.store_action, false);
                pass_desc.stencilAttachment.clearStencil = action->stencil.clear_value;
                switch (ds_att_img->cmn.type) {
                    case SG_IMAGETYPE_CUBE:
                    case SG_IMAGETYPE_ARRAY:
                        pass_desc.stencilAttachment.slice = (NSUInteger)cmn_ds_att->slice;
                        break;
                    case SG_IMAGETYPE_3D:
                        pass_desc.stencilAttachment.resolveDepthPlane = (NSUInteger)cmn_ds_att->slice;
                        break;
                    default: break;
                }
            }
        }
    } else {
        // setup pass descriptor for swapchain rendering
        //
        // NOTE: at least in macOS Sonoma this no longer seems to be the case, the
        // current drawable is also valid in a minimized window
        // ===
        // an MTKView current_drawable will not be valid if window is minimized, don't do any rendering in this case
        if (0 == swapchain->metal.current_drawable) {
            _sg.cur_pass.valid = false;
            return;
        }
        // pin the swapchain resources into memory so that they outlive their command buffer
        // (this is necessary because the command buffer doesn't retain references)
        int pass_desc_ref = _sg_mtl_add_resource(pass_desc);
        _sg_mtl_release_resource(_sg.frame_index, pass_desc_ref);

        _sg.mtl.cur_drawable = (__bridge id<CAMetalDrawable>) swapchain->metal.current_drawable;
        if (swapchain->sample_count > 1) {
            // multi-sampling: render into msaa texture, resolve into drawable texture
            id<MTLTexture> msaa_tex = (__bridge id<MTLTexture>) swapchain->metal.msaa_color_texture;
            SOKOL_ASSERT(msaa_tex != nil);
            pass_desc.colorAttachments[0].texture = msaa_tex;
            pass_desc.colorAttachments[0].resolveTexture = _sg.mtl.cur_drawable.texture;
            pass_desc.colorAttachments[0].storeAction = MTLStoreActionMultisampleResolve;
        } else {
            // non-msaa: render into current_drawable
            pass_desc.colorAttachments[0].texture = _sg.mtl.cur_drawable.texture;
            pass_desc.colorAttachments[0].storeAction = MTLStoreActionStore;
        }
        pass_desc.colorAttachments[0].loadAction = _sg_mtl_load_action(action->colors[0].load_action);
        const sg_color c = action->colors[0].clear_value;
        pass_desc.colorAttachments[0].clearColor = MTLClearColorMake(c.r, c.g, c.b, c.a);

        // optional depth-stencil texture
        if (swapchain->metal.depth_stencil_texture) {
            id<MTLTexture> ds_tex = (__bridge id<MTLTexture>) swapchain->metal.depth_stencil_texture;
            SOKOL_ASSERT(ds_tex != nil);
            pass_desc.depthAttachment.texture = ds_tex;
            pass_desc.depthAttachment.storeAction = MTLStoreActionDontCare;
            pass_desc.depthAttachment.loadAction = _sg_mtl_load_action(action->depth.load_action);
            pass_desc.depthAttachment.clearDepth = action->depth.clear_value;
            if (_sg_is_depth_stencil_format(swapchain->depth_format)) {
                pass_desc.stencilAttachment.texture = ds_tex;
                pass_desc.stencilAttachment.storeAction = MTLStoreActionDontCare;
                pass_desc.stencilAttachment.loadAction = _sg_mtl_load_action(action->stencil.load_action);
                pass_desc.stencilAttachment.clearStencil = action->stencil.clear_value;
            }
        }
    }

    // NOTE: at least in macOS Sonoma, the following is no longer the case, a valid
    // render command encoder is also returned in a minimized window
    // ===
    // create a render command encoder, this might return nil if window is minimized
    _sg.mtl.render_cmd_encoder = [_sg.mtl.cmd_buffer renderCommandEncoderWithDescriptor:pass_desc];
    if (nil == _sg.mtl.render_cmd_encoder) {
        _sg.cur_pass.valid = false;
        return;
    }

    #if defined(SOKOL_DEBUG)
    if (pass->label) {
        _sg.mtl.render_cmd_encoder.label = [NSString stringWithUTF8String:pass->label];
    }
    #endif
}

_SOKOL_PRIVATE void _sg_mtl_begin_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass);
    SOKOL_ASSERT(_sg.mtl.cmd_queue);
    SOKOL_ASSERT(nil == _sg.mtl.compute_cmd_encoder);
    SOKOL_ASSERT(nil == _sg.mtl.render_cmd_encoder);
    SOKOL_ASSERT(nil == _sg.mtl.cur_drawable);
    _sg_mtl_clear_state_cache();

    // if this is the first pass in the frame, create one command buffer and blit-cmd-encoder for the entire frame
    if (nil == _sg.mtl.cmd_buffer) {
        // block until the oldest frame in flight has finished
        dispatch_semaphore_wait(_sg.mtl.sem, DISPATCH_TIME_FOREVER);
        if (_sg.desc.mtl_use_command_buffer_with_retained_references) {
            _sg.mtl.cmd_buffer = [_sg.mtl.cmd_queue commandBuffer];
        } else {
            _sg.mtl.cmd_buffer = [_sg.mtl.cmd_queue commandBufferWithUnretainedReferences];
        }
        [_sg.mtl.cmd_buffer enqueue];
        [_sg.mtl.cmd_buffer addCompletedHandler:^(id<MTLCommandBuffer> cmd_buf) {
            // NOTE: this code is called on a different thread!
            _SOKOL_UNUSED(cmd_buf);
            dispatch_semaphore_signal(_sg.mtl.sem);
        }];
    }

    // if this is first pass in frame, get uniform buffer base pointer
    if (0 == _sg.mtl.cur_ub_base_ptr) {
        _sg.mtl.cur_ub_base_ptr = (uint8_t*)[_sg.mtl.uniform_buffers[_sg.mtl.cur_frame_rotate_index] contents];
    }

    if (pass->compute) {
        _sg_mtl_begin_compute_pass(pass);
    } else {
        _sg_mtl_begin_render_pass(pass);
    }

    // bind uniform buffers, those bindings remain valid for the entire pass
    if (_sg.cur_pass.valid) {
        _sg_mtl_bind_uniform_buffers();
    }
}

_SOKOL_PRIVATE void _sg_mtl_end_pass(void) {
    if (nil != _sg.mtl.render_cmd_encoder) {
        [_sg.mtl.render_cmd_encoder endEncoding];
        // NOTE: MTLRenderCommandEncoder is autoreleased
        _sg.mtl.render_cmd_encoder = nil;
    }
    if (nil != _sg.mtl.compute_cmd_encoder) {
        [_sg.mtl.compute_cmd_encoder endEncoding];
        // NOTE: MTLComputeCommandEncoder is autoreleased
        _sg.mtl.compute_cmd_encoder = nil;

        // synchronize any managed resources written by the GPU
        // NOTE: storage attachment images are currently not managed and are not eligible for syncing
        #if defined(_SG_TARGET_MACOS)
        if (_sg_mtl_resource_options_storage_mode_managed_or_shared() == MTLResourceStorageModeManaged) {
            const bool needs_sync = _sg.compute.readwrite_sbufs.cur > 0;
            if (needs_sync) {
                id<MTLBlitCommandEncoder> blit_cmd_encoder = [_sg.mtl.cmd_buffer blitCommandEncoder];
                for (uint32_t i = 0; i < _sg.compute.readwrite_sbufs.cur; i++) {
                    _sg_buffer_t* sbuf = _sg_lookup_buffer(_sg.compute.readwrite_sbufs.items[i]);
                    if (sbuf) {
                        [blit_cmd_encoder synchronizeResource:_sg_mtl_id(sbuf->mtl.buf[sbuf->cmn.active_slot])];
                    }
                }
                [blit_cmd_encoder endEncoding];
            }
        }
        #endif
    }
    // if this is a swapchain pass, present the drawable
    if (nil != _sg.mtl.cur_drawable) {
        [_sg.mtl.cmd_buffer presentDrawable:_sg.mtl.cur_drawable];
        _sg.mtl.cur_drawable = nil;
    }
}

_SOKOL_PRIVATE void _sg_mtl_commit(void) {
    SOKOL_ASSERT(nil == _sg.mtl.render_cmd_encoder);
    SOKOL_ASSERT(nil == _sg.mtl.compute_cmd_encoder);
    SOKOL_ASSERT(nil != _sg.mtl.cmd_buffer);

    // commit the frame's command buffer
    [_sg.mtl.cmd_buffer commit];

    // garbage-collect resources pending for release
    _sg_mtl_garbage_collect(_sg.frame_index);

    // rotate uniform buffer slot
    if (++_sg.mtl.cur_frame_rotate_index >= SG_NUM_INFLIGHT_FRAMES) {
        _sg.mtl.cur_frame_rotate_index = 0;
    }
    _sg.mtl.cur_ub_offset = 0;
    _sg.mtl.cur_ub_base_ptr = 0;
    // NOTE: MTLCommandBuffer is autoreleased
    _sg.mtl.cmd_buffer = nil;
}

_SOKOL_PRIVATE void _sg_mtl_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
    SOKOL_ASSERT(_sg.cur_pass.height > 0);
    MTLViewport vp;
    vp.originX = (double) x;
    vp.originY = (double) (origin_top_left ? y : (_sg.cur_pass.height - (y + h)));
    vp.width   = (double) w;
    vp.height  = (double) h;
    vp.znear   = 0.0;
    vp.zfar    = 1.0;
    [_sg.mtl.render_cmd_encoder setViewport:vp];
}

_SOKOL_PRIVATE void _sg_mtl_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
    SOKOL_ASSERT(_sg.cur_pass.width > 0);
    SOKOL_ASSERT(_sg.cur_pass.height > 0);
    // clip against framebuffer rect
    const _sg_recti_t clip = _sg_clipi(x, y, w, h, _sg.cur_pass.width, _sg.cur_pass.height);
    MTLScissorRect r;
    r.x = (NSUInteger)clip.x;
    r.y = (NSUInteger) (origin_top_left ? clip.y : (_sg.cur_pass.height - (clip.y + clip.h)));
    r.width = (NSUInteger)clip.w;
    r.height = (NSUInteger)clip.h;
    [_sg.mtl.render_cmd_encoder setScissorRect:r];
}

_SOKOL_PRIVATE void _sg_mtl_apply_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_pip, &pip->slot)) {
        _sg.mtl.state_cache.cur_pip = _sg_sref(&pip->slot);
        if (pip->cmn.is_compute) {
            SOKOL_ASSERT(_sg.cur_pass.is_compute);
            SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
            SOKOL_ASSERT(pip->mtl.cps != _SG_MTL_INVALID_SLOT_INDEX);
            [_sg.mtl.compute_cmd_encoder setComputePipelineState:_sg_mtl_id(pip->mtl.cps)];
            // apply storage image bindings for writing
            const _sg_attachments_t* atts = _sg_attachments_ref_ptr_or_null(&_sg.cur_pass.atts);
            if (atts) {
                const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
                for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
                    const sg_shader_stage stage = shd->cmn.storage_images[i].stage;
                    if (stage != SG_SHADERSTAGE_COMPUTE) {
                        continue;
                    }
                    const NSUInteger mtl_slot = shd->mtl.simg_texture_n[i];
                    SOKOL_ASSERT(mtl_slot < _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS);
                    const uint64_t cache_key = ((uint64_t)atts->cmn.storages[i].image.sref.id) << 32;
                    SOKOL_ASSERT(cache_key != 0);
                    if (_sg.mtl.state_cache.cur_cs_image_ids[mtl_slot] != cache_key) {
                        _sg.mtl.state_cache.cur_cs_image_ids[mtl_slot] = cache_key;
                        [_sg.mtl.compute_cmd_encoder setTexture:_sg_mtl_id(atts->mtl.storage_views[i]) atIndex:mtl_slot];
                        _sg_stats_add(metal.bindings.num_set_compute_texture, 1);
                    }
                }
            }
        } else {
            SOKOL_ASSERT(!_sg.cur_pass.is_compute);
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            sg_color c = pip->cmn.blend_color;
            [_sg.mtl.render_cmd_encoder setBlendColorRed:c.r green:c.g blue:c.b alpha:c.a];
            _sg_stats_add(metal.pipeline.num_set_blend_color, 1);
            [_sg.mtl.render_cmd_encoder setCullMode:pip->mtl.cull_mode];
            _sg_stats_add(metal.pipeline.num_set_cull_mode, 1);
            [_sg.mtl.render_cmd_encoder setFrontFacingWinding:pip->mtl.winding];
            _sg_stats_add(metal.pipeline.num_set_front_facing_winding, 1);
            [_sg.mtl.render_cmd_encoder setStencilReferenceValue:pip->mtl.stencil_ref];
            _sg_stats_add(metal.pipeline.num_set_stencil_reference_value, 1);
            [_sg.mtl.render_cmd_encoder setDepthBias:pip->cmn.depth.bias slopeScale:pip->cmn.depth.bias_slope_scale clamp:pip->cmn.depth.bias_clamp];
            _sg_stats_add(metal.pipeline.num_set_depth_bias, 1);
            SOKOL_ASSERT(pip->mtl.rps != _SG_MTL_INVALID_SLOT_INDEX);
            [_sg.mtl.render_cmd_encoder setRenderPipelineState:_sg_mtl_id(pip->mtl.rps)];
            _sg_stats_add(metal.pipeline.num_set_render_pipeline_state, 1);
            SOKOL_ASSERT(pip->mtl.dss != _SG_MTL_INVALID_SLOT_INDEX);
            [_sg.mtl.render_cmd_encoder setDepthStencilState:_sg_mtl_id(pip->mtl.dss)];
            _sg_stats_add(metal.pipeline.num_set_depth_stencil_state, 1);
        }
    }
}

_SOKOL_PRIVATE bool _sg_mtl_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip);
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&bnd->pip->cmn.shader);

    // don't set vertex- and index-buffers in compute passes
    if (!_sg.cur_pass.is_compute) {
        SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
        // store index buffer binding, this will be needed later in sg_draw()
        _sg.mtl.state_cache.cur_ibuf = _sg_buffer_ref(bnd->ib);
        if (bnd->ib) {
            SOKOL_ASSERT(bnd->pip->cmn.index_type != SG_INDEXTYPE_NONE);
        } else {
            SOKOL_ASSERT(bnd->pip->cmn.index_type == SG_INDEXTYPE_NONE);
        }
        _sg.mtl.state_cache.cur_ibuf_offset = bnd->ib_offset;
        // apply vertex buffers
        for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
            const _sg_buffer_t* vb = bnd->vbs[i];
            if (vb == 0) {
                continue;
            }
            const NSUInteger mtl_slot = _sg_mtl_vertexbuffer_bindslot(i);
            SOKOL_ASSERT(mtl_slot < _SG_MTL_MAX_STAGE_BUFFER_BINDINGS);
            const int vb_offset = bnd->vb_offsets[i];
            const bool ref_eql = _sg_sref_eql(&_sg.mtl.state_cache.cur_vsbufs[mtl_slot], &vb->slot);
            if (!ref_eql || (_sg.mtl.state_cache.cur_vs_buffer_offsets[mtl_slot] != vb_offset)) {
                _sg.mtl.state_cache.cur_vs_buffer_offsets[mtl_slot] = vb_offset;
                if (!ref_eql) {
                    // vertex buffer has changed
                    _sg.mtl.state_cache.cur_vsbufs[mtl_slot] = _sg_sref(&vb->slot);
                    SOKOL_ASSERT(vb->mtl.buf[vb->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
                    [_sg.mtl.render_cmd_encoder setVertexBuffer:_sg_mtl_id(vb->mtl.buf[vb->cmn.active_slot])
                        offset:(NSUInteger)vb_offset
                        atIndex:mtl_slot];
                } else {
                    // only vertex buffer offset has changed
                    [_sg.mtl.render_cmd_encoder setVertexBufferOffset:(NSUInteger)vb_offset atIndex:mtl_slot];
                }
                _sg_stats_add(metal.bindings.num_set_vertex_buffer, 1);
            }
        }
    }

    // apply image bindings
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        const _sg_image_t* img = bnd->imgs[i];
        if (img == 0) {
            continue;
        }
        SOKOL_ASSERT(img->mtl.tex[img->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
        const sg_shader_stage stage = shd->cmn.images[i].stage;
        SOKOL_ASSERT((stage == SG_SHADERSTAGE_VERTEX) || (stage == SG_SHADERSTAGE_FRAGMENT) || (stage == SG_SHADERSTAGE_COMPUTE));
        const NSUInteger mtl_slot = shd->mtl.img_texture_n[i];
        SOKOL_ASSERT(mtl_slot < _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS);
        if (stage == SG_SHADERSTAGE_VERTEX) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_vsimgs[mtl_slot], &img->slot)) {
                _sg.mtl.state_cache.cur_vsimgs[mtl_slot] = _sg_sref(&img->slot);
                [_sg.mtl.render_cmd_encoder setVertexTexture:_sg_mtl_id(img->mtl.tex[img->cmn.active_slot]) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_vertex_texture, 1);
            }
        } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_fsimgs[mtl_slot], &img->slot)) {
                _sg.mtl.state_cache.cur_fsimgs[mtl_slot] = _sg_sref(&img->slot);
                [_sg.mtl.render_cmd_encoder setFragmentTexture:_sg_mtl_id(img->mtl.tex[img->cmn.active_slot]) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_fragment_texture, 1);
            }
        } else if (stage == SG_SHADERSTAGE_COMPUTE) {
            SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
            if (_sg.mtl.state_cache.cur_cs_image_ids[mtl_slot] != img->slot.id) {
                _sg.mtl.state_cache.cur_cs_image_ids[mtl_slot] = img->slot.id;
                [_sg.mtl.compute_cmd_encoder setTexture:_sg_mtl_id(img->mtl.tex[img->cmn.active_slot]) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_compute_texture, 1);
            }
        }
    }

    // apply sampler bindings
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        const _sg_sampler_t* smp = bnd->smps[i];
        if (smp == 0) {
            continue;
        }
        SOKOL_ASSERT(smp->mtl.sampler_state != _SG_MTL_INVALID_SLOT_INDEX);
        const sg_shader_stage stage = shd->cmn.samplers[i].stage;
        SOKOL_ASSERT((stage == SG_SHADERSTAGE_VERTEX) || (stage == SG_SHADERSTAGE_FRAGMENT) || (stage == SG_SHADERSTAGE_COMPUTE));
        const NSUInteger mtl_slot = shd->mtl.smp_sampler_n[i];
        SOKOL_ASSERT(mtl_slot < _SG_MTL_MAX_STAGE_SAMPLER_BINDINGS);
        if (stage == SG_SHADERSTAGE_VERTEX) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_vssmps[mtl_slot], &smp->slot)) {
                _sg.mtl.state_cache.cur_vssmps[mtl_slot] = _sg_sref(&smp->slot);
                [_sg.mtl.render_cmd_encoder setVertexSamplerState:_sg_mtl_id(smp->mtl.sampler_state) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_vertex_sampler_state, 1);
            }
        } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_fssmps[mtl_slot], &smp->slot)) {
                _sg.mtl.state_cache.cur_fssmps[mtl_slot] = _sg_sref(&smp->slot);
                [_sg.mtl.render_cmd_encoder setFragmentSamplerState:_sg_mtl_id(smp->mtl.sampler_state) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_fragment_sampler_state, 1);
            }
        } else if (stage == SG_SHADERSTAGE_COMPUTE) {
            SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_cssmps[mtl_slot], &smp->slot)) {
                _sg.mtl.state_cache.cur_cssmps[mtl_slot] = _sg_sref(&smp->slot);
                [_sg.mtl.compute_cmd_encoder setSamplerState:_sg_mtl_id(smp->mtl.sampler_state) atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_compute_sampler_state, 1);
            }
        }
    }

    // apply storage buffer bindings
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        const _sg_buffer_t* sbuf = bnd->sbufs[i];
        if (sbuf == 0) {
            continue;
        }
        SOKOL_ASSERT(sbuf->mtl.buf[sbuf->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX);
        const sg_shader_stage stage = shd->cmn.storage_buffers[i].stage;
        SOKOL_ASSERT((stage == SG_SHADERSTAGE_VERTEX) || (stage == SG_SHADERSTAGE_FRAGMENT) || (stage == SG_SHADERSTAGE_COMPUTE));
        const NSUInteger mtl_slot = shd->mtl.sbuf_buffer_n[i];
        SOKOL_ASSERT(mtl_slot < _SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS);
        if (stage == SG_SHADERSTAGE_VERTEX) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_vsbufs[mtl_slot], &sbuf->slot)) {
                _sg.mtl.state_cache.cur_vsbufs[mtl_slot] = _sg_sref(&sbuf->slot);
                [_sg.mtl.render_cmd_encoder setVertexBuffer:_sg_mtl_id(sbuf->mtl.buf[sbuf->cmn.active_slot]) offset:0 atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_vertex_buffer, 1);
            }
        } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
            SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_fsbufs[mtl_slot], &sbuf->slot)) {
                _sg.mtl.state_cache.cur_fsbufs[mtl_slot] = _sg_sref(&sbuf->slot);
                [_sg.mtl.render_cmd_encoder setFragmentBuffer:_sg_mtl_id(sbuf->mtl.buf[sbuf->cmn.active_slot]) offset:0 atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_fragment_buffer, 1);
            }
        } else if (stage == SG_SHADERSTAGE_COMPUTE) {
            SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
            if (!_sg_sref_eql(&_sg.mtl.state_cache.cur_csbufs[mtl_slot], &sbuf->slot)) {
                _sg.mtl.state_cache.cur_csbufs[mtl_slot] = _sg_sref(&sbuf->slot);
                [_sg.mtl.compute_cmd_encoder setBuffer:_sg_mtl_id(sbuf->mtl.buf[sbuf->cmn.active_slot]) offset:0 atIndex:mtl_slot];
                _sg_stats_add(metal.bindings.num_set_compute_buffer, 1);
            }
        }
    }
    return true;
}

_SOKOL_PRIVATE void _sg_mtl_apply_uniforms(int ub_slot, const sg_range* data) {
    SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
    SOKOL_ASSERT(((size_t)_sg.mtl.cur_ub_offset + data->size) <= (size_t)_sg.mtl.ub_size);
    SOKOL_ASSERT((_sg.mtl.cur_ub_offset & (_SG_MTL_UB_ALIGN-1)) == 0);
    const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
    SOKOL_ASSERT(pip);
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
    SOKOL_ASSERT(data->size == shd->cmn.uniform_blocks[ub_slot].size);

    const sg_shader_stage stage = shd->cmn.uniform_blocks[ub_slot].stage;
    const NSUInteger mtl_slot = shd->mtl.ub_buffer_n[ub_slot];

    // copy to global uniform buffer, record offset into cmd encoder, and advance offset
    uint8_t* dst = &_sg.mtl.cur_ub_base_ptr[_sg.mtl.cur_ub_offset];
    memcpy(dst, data->ptr, data->size);
    if (stage == SG_SHADERSTAGE_VERTEX) {
        SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
        [_sg.mtl.render_cmd_encoder setVertexBufferOffset:(NSUInteger)_sg.mtl.cur_ub_offset atIndex:mtl_slot];
        _sg_stats_add(metal.uniforms.num_set_vertex_buffer_offset, 1);
    } else if (stage == SG_SHADERSTAGE_FRAGMENT) {
        SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
        [_sg.mtl.render_cmd_encoder setFragmentBufferOffset:(NSUInteger)_sg.mtl.cur_ub_offset atIndex:mtl_slot];
        _sg_stats_add(metal.uniforms.num_set_fragment_buffer_offset, 1);
    } else if (stage == SG_SHADERSTAGE_COMPUTE) {
        SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
        [_sg.mtl.compute_cmd_encoder setBufferOffset:(NSUInteger)_sg.mtl.cur_ub_offset atIndex:mtl_slot];
        _sg_stats_add(metal.uniforms.num_set_compute_buffer_offset, 1);
    } else {
        SOKOL_UNREACHABLE;
    }
    _sg.mtl.cur_ub_offset = _sg_roundup(_sg.mtl.cur_ub_offset + (int)data->size, _SG_MTL_UB_ALIGN);
}

_SOKOL_PRIVATE void _sg_mtl_draw(int base_element, int num_elements, int num_instances) {
    SOKOL_ASSERT(nil != _sg.mtl.render_cmd_encoder);
    const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
    SOKOL_ASSERT(pip);
    if (SG_INDEXTYPE_NONE != pip->cmn.index_type) {
        // indexed rendering
        const _sg_buffer_t* ib = _sg_buffer_ref_ptr(&_sg.mtl.state_cache.cur_ibuf);
        SOKOL_ASSERT(ib && (ib->mtl.buf[ib->cmn.active_slot] != _SG_MTL_INVALID_SLOT_INDEX));
        const NSUInteger index_buffer_offset = (NSUInteger) (_sg.mtl.state_cache.cur_ibuf_offset + base_element * pip->mtl.index_size);
        [_sg.mtl.render_cmd_encoder drawIndexedPrimitives:pip->mtl.prim_type
            indexCount:(NSUInteger)num_elements
            indexType:pip->mtl.index_type
            indexBuffer:_sg_mtl_id(ib->mtl.buf[ib->cmn.active_slot])
            indexBufferOffset:index_buffer_offset
            instanceCount:(NSUInteger)num_instances];
    } else {
        // non-indexed rendering
        [_sg.mtl.render_cmd_encoder drawPrimitives:pip->mtl.prim_type
            vertexStart:(NSUInteger)base_element
            vertexCount:(NSUInteger)num_elements
            instanceCount:(NSUInteger)num_instances];
    }
}

_SOKOL_PRIVATE void _sg_mtl_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    SOKOL_ASSERT(nil != _sg.mtl.compute_cmd_encoder);
    const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
    SOKOL_ASSERT(pip);
    const MTLSize thread_groups = MTLSizeMake(
        (NSUInteger)num_groups_x,
        (NSUInteger)num_groups_y,
        (NSUInteger)num_groups_z);
    const MTLSize threads_per_threadgroup = pip->mtl.threads_per_threadgroup;
    [_sg.mtl.compute_cmd_encoder dispatchThreadgroups:thread_groups threadsPerThreadgroup:threads_per_threadgroup];
}

_SOKOL_PRIVATE void _sg_mtl_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
        buf->cmn.active_slot = 0;
    }
    __unsafe_unretained id<MTLBuffer> mtl_buf = _sg_mtl_id(buf->mtl.buf[buf->cmn.active_slot]);
    void* dst_ptr = [mtl_buf contents];
    memcpy(dst_ptr, data->ptr, data->size);
    #if defined(_SG_TARGET_MACOS)
    if (_sg_mtl_resource_options_storage_mode_managed_or_shared() == MTLResourceStorageModeManaged) {
        [mtl_buf didModifyRange:NSMakeRange(0, data->size)];
    }
    #endif
}

_SOKOL_PRIVATE void _sg_mtl_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    SOKOL_ASSERT(buf && data && data->ptr && (data->size > 0));
    if (new_frame) {
        if (++buf->cmn.active_slot >= buf->cmn.num_slots) {
            buf->cmn.active_slot = 0;
        }
    }
    __unsafe_unretained id<MTLBuffer> mtl_buf = _sg_mtl_id(buf->mtl.buf[buf->cmn.active_slot]);
    uint8_t* dst_ptr = (uint8_t*) [mtl_buf contents];
    dst_ptr += buf->cmn.append_pos;
    memcpy(dst_ptr, data->ptr, data->size);
    #if defined(_SG_TARGET_MACOS)
    if (_sg_mtl_resource_options_storage_mode_managed_or_shared() == MTLResourceStorageModeManaged) {
        [mtl_buf didModifyRange:NSMakeRange((NSUInteger)buf->cmn.append_pos, (NSUInteger)data->size)];
    }
    #endif
}

_SOKOL_PRIVATE void _sg_mtl_update_image(_sg_image_t* img, const sg_image_data* data) {
    SOKOL_ASSERT(img && data);
    if (++img->cmn.active_slot >= img->cmn.num_slots) {
        img->cmn.active_slot = 0;
    }
    __unsafe_unretained id<MTLTexture> mtl_tex = _sg_mtl_id(img->mtl.tex[img->cmn.active_slot]);
    _sg_mtl_copy_image_data(img, mtl_tex, data);
}

_SOKOL_PRIVATE void _sg_mtl_push_debug_group(const char* name) {
    SOKOL_ASSERT(name);
    if (_sg.mtl.render_cmd_encoder) {
        [_sg.mtl.render_cmd_encoder pushDebugGroup:[NSString stringWithUTF8String:name]];
    } else if (_sg.mtl.compute_cmd_encoder) {
        [_sg.mtl.compute_cmd_encoder pushDebugGroup:[NSString stringWithUTF8String:name]];
    }
}

_SOKOL_PRIVATE void _sg_mtl_pop_debug_group(void) {
    if (_sg.mtl.render_cmd_encoder) {
        [_sg.mtl.render_cmd_encoder popDebugGroup];
    } else if (_sg.mtl.compute_cmd_encoder) {
        [_sg.mtl.compute_cmd_encoder popDebugGroup];
    }
}

// ██     ██ ███████ ██████   ██████  ██████  ██    ██     ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ██     ██ ██      ██   ██ ██       ██   ██ ██    ██     ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██  █  ██ █████   ██████  ██   ███ ██████  ██    ██     ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██ ███ ██ ██      ██   ██ ██    ██ ██      ██    ██     ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
//  ███ ███  ███████ ██████   ██████  ██       ██████      ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>webgpu
// >>wgpu
#elif defined(SOKOL_WGPU)

#if !defined(__EMSCRIPTEN__)
// FIXME: webgpu.h differences between Dawn and Emscripten webgpu.h
#define wgpuBufferReference wgpuBufferAddRef
#define wgpuTextureReference wgpuTextureAddRef
#define wgpuTextureViewReference wgpuTextureViewAddRef
#define wgpuSamplerReference wgpuSamplerAddRef
#define WGPUSType_ShaderModuleWGSLDescriptor WGPUSType_ShaderSourceWGSL
_SOKOL_PRIVATE WGPUStringView _sg_wgpu_stringview(const char* str) {
    WGPUStringView res;
    if (str) {
        res.data = str;
        res.length = strlen(str);
    } else {
        res.data = 0;
        res.length = 0;
    }
    return res;
}
_SOKOL_PRIVATE WGPUOptionalBool _sg_wgpu_optional_bool(bool b) {
    return b ? WGPUOptionalBool_True : WGPUOptionalBool_False;
}
#else
#define _sg_wgpu_stringview(str) str
#define _sg_wgpu_optional_bool(b) (b)
#endif

_SOKOL_PRIVATE WGPUBufferUsage _sg_wgpu_buffer_usage(const sg_buffer_usage* usg) {
    int res = 0;
    if (usg->vertex_buffer) {
        res |= WGPUBufferUsage_Vertex;
    }
    if (usg->index_buffer) {
        res |= WGPUBufferUsage_Index;
    }
    if (usg->storage_buffer) {
        res |= WGPUBufferUsage_Storage;
    }
    if (!usg->immutable) {
        res |= WGPUBufferUsage_CopyDst;
    }
    return (WGPUBufferUsage)res;
}

_SOKOL_PRIVATE WGPULoadOp _sg_wgpu_load_op(WGPUTextureView view, sg_load_action a) {
    if (0 == view) {
        return WGPULoadOp_Undefined;
    } else switch (a) {
        case SG_LOADACTION_CLEAR:
        case SG_LOADACTION_DONTCARE:
            return WGPULoadOp_Clear;
        case SG_LOADACTION_LOAD:
            return WGPULoadOp_Load;
        default:
            SOKOL_UNREACHABLE;
            return WGPULoadOp_Force32;
    }
}

_SOKOL_PRIVATE WGPUStoreOp _sg_wgpu_store_op(WGPUTextureView view, sg_store_action a) {
    if (0 == view) {
        return WGPUStoreOp_Undefined;
    } else switch (a) {
        case SG_STOREACTION_STORE:
            return WGPUStoreOp_Store;
        case SG_STOREACTION_DONTCARE:
            return WGPUStoreOp_Discard;
        default:
            SOKOL_UNREACHABLE;
            return WGPUStoreOp_Force32;
    }
}

_SOKOL_PRIVATE WGPUTextureViewDimension _sg_wgpu_texture_view_dimension(sg_image_type t) {
    switch (t) {
        case SG_IMAGETYPE_2D:       return WGPUTextureViewDimension_2D;
        case SG_IMAGETYPE_CUBE:     return WGPUTextureViewDimension_Cube;
        case SG_IMAGETYPE_3D:       return WGPUTextureViewDimension_3D;
        case SG_IMAGETYPE_ARRAY:    return WGPUTextureViewDimension_2DArray;
        default: SOKOL_UNREACHABLE; return WGPUTextureViewDimension_Force32;
    }
}

_SOKOL_PRIVATE WGPUTextureDimension _sg_wgpu_texture_dimension(sg_image_type t) {
    if (SG_IMAGETYPE_3D == t) {
        return WGPUTextureDimension_3D;
    } else {
        return WGPUTextureDimension_2D;
    }
}

_SOKOL_PRIVATE WGPUTextureSampleType _sg_wgpu_texture_sample_type(sg_image_sample_type t, bool msaa) {
    switch (t) {
        case SG_IMAGESAMPLETYPE_FLOAT:  return msaa ? WGPUTextureSampleType_UnfilterableFloat : WGPUTextureSampleType_Float;
        case SG_IMAGESAMPLETYPE_DEPTH:  return WGPUTextureSampleType_Depth;
        case SG_IMAGESAMPLETYPE_SINT:   return WGPUTextureSampleType_Sint;
        case SG_IMAGESAMPLETYPE_UINT:   return WGPUTextureSampleType_Uint;
        case SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT: return WGPUTextureSampleType_UnfilterableFloat;
        default: SOKOL_UNREACHABLE;     return WGPUTextureSampleType_Force32;
    }
}

_SOKOL_PRIVATE WGPUSamplerBindingType _sg_wgpu_sampler_binding_type(sg_sampler_type t) {
    switch (t) {
        case SG_SAMPLERTYPE_FILTERING: return WGPUSamplerBindingType_Filtering;
        case SG_SAMPLERTYPE_COMPARISON: return WGPUSamplerBindingType_Comparison;
        case SG_SAMPLERTYPE_NONFILTERING: return WGPUSamplerBindingType_NonFiltering;
        default: SOKOL_UNREACHABLE; return WGPUSamplerBindingType_Force32;
    }
}

_SOKOL_PRIVATE WGPUAddressMode _sg_wgpu_sampler_address_mode(sg_wrap m) {
    switch (m) {
        case SG_WRAP_REPEAT:
            return WGPUAddressMode_Repeat;
        case SG_WRAP_CLAMP_TO_EDGE:
        case SG_WRAP_CLAMP_TO_BORDER:
            return WGPUAddressMode_ClampToEdge;
        case SG_WRAP_MIRRORED_REPEAT:
            return WGPUAddressMode_MirrorRepeat;
        default:
            SOKOL_UNREACHABLE;
            return WGPUAddressMode_Force32;
    }
}

_SOKOL_PRIVATE WGPUFilterMode _sg_wgpu_sampler_minmag_filter(sg_filter f) {
    switch (f) {
        case SG_FILTER_NEAREST:
            return WGPUFilterMode_Nearest;
        case SG_FILTER_LINEAR:
            return WGPUFilterMode_Linear;
        default:
            SOKOL_UNREACHABLE;
            return WGPUFilterMode_Force32;
    }
}

_SOKOL_PRIVATE WGPUMipmapFilterMode _sg_wgpu_sampler_mipmap_filter(sg_filter f) {
    switch (f) {
        case SG_FILTER_NEAREST:
            return WGPUMipmapFilterMode_Nearest;
        case SG_FILTER_LINEAR:
            return WGPUMipmapFilterMode_Linear;
        default:
            SOKOL_UNREACHABLE;
            return WGPUMipmapFilterMode_Force32;
    }
}

_SOKOL_PRIVATE WGPUIndexFormat _sg_wgpu_indexformat(sg_index_type t) {
    // NOTE: there's no WGPUIndexFormat_None
    return (t == SG_INDEXTYPE_UINT16) ? WGPUIndexFormat_Uint16 : WGPUIndexFormat_Uint32;
}

_SOKOL_PRIVATE WGPUIndexFormat _sg_wgpu_stripindexformat(sg_primitive_type prim_type, sg_index_type idx_type) {
    if (idx_type == SG_INDEXTYPE_NONE) {
        return WGPUIndexFormat_Undefined;
    } else if ((prim_type == SG_PRIMITIVETYPE_LINE_STRIP) || (prim_type == SG_PRIMITIVETYPE_TRIANGLE_STRIP)) {
        return _sg_wgpu_indexformat(idx_type);
    } else {
        return WGPUIndexFormat_Undefined;
    }
}

_SOKOL_PRIVATE WGPUVertexStepMode _sg_wgpu_stepmode(sg_vertex_step s) {
    return (s == SG_VERTEXSTEP_PER_VERTEX) ? WGPUVertexStepMode_Vertex : WGPUVertexStepMode_Instance;
}

_SOKOL_PRIVATE WGPUVertexFormat _sg_wgpu_vertexformat(sg_vertex_format f) {
    switch (f) {
        case SG_VERTEXFORMAT_FLOAT:         return WGPUVertexFormat_Float32;
        case SG_VERTEXFORMAT_FLOAT2:        return WGPUVertexFormat_Float32x2;
        case SG_VERTEXFORMAT_FLOAT3:        return WGPUVertexFormat_Float32x3;
        case SG_VERTEXFORMAT_FLOAT4:        return WGPUVertexFormat_Float32x4;
        case SG_VERTEXFORMAT_INT:           return WGPUVertexFormat_Sint32;
        case SG_VERTEXFORMAT_INT2:          return WGPUVertexFormat_Sint32x2;
        case SG_VERTEXFORMAT_INT3:          return WGPUVertexFormat_Sint32x3;
        case SG_VERTEXFORMAT_INT4:          return WGPUVertexFormat_Sint32x4;
        case SG_VERTEXFORMAT_UINT:          return WGPUVertexFormat_Uint32;
        case SG_VERTEXFORMAT_UINT2:         return WGPUVertexFormat_Uint32x2;
        case SG_VERTEXFORMAT_UINT3:         return WGPUVertexFormat_Uint32x3;
        case SG_VERTEXFORMAT_UINT4:         return WGPUVertexFormat_Uint32x4;
        case SG_VERTEXFORMAT_BYTE4:         return WGPUVertexFormat_Sint8x4;
        case SG_VERTEXFORMAT_BYTE4N:        return WGPUVertexFormat_Snorm8x4;
        case SG_VERTEXFORMAT_UBYTE4:        return WGPUVertexFormat_Uint8x4;
        case SG_VERTEXFORMAT_UBYTE4N:       return WGPUVertexFormat_Unorm8x4;
        case SG_VERTEXFORMAT_SHORT2:        return WGPUVertexFormat_Sint16x2;
        case SG_VERTEXFORMAT_SHORT2N:       return WGPUVertexFormat_Snorm16x2;
        case SG_VERTEXFORMAT_USHORT2:       return WGPUVertexFormat_Uint16x2;
        case SG_VERTEXFORMAT_USHORT2N:      return WGPUVertexFormat_Unorm16x2;
        case SG_VERTEXFORMAT_SHORT4:        return WGPUVertexFormat_Sint16x4;
        case SG_VERTEXFORMAT_SHORT4N:       return WGPUVertexFormat_Snorm16x4;
        case SG_VERTEXFORMAT_USHORT4:       return WGPUVertexFormat_Uint16x4;
        case SG_VERTEXFORMAT_USHORT4N:      return WGPUVertexFormat_Unorm16x4;
        case SG_VERTEXFORMAT_UINT10_N2:     return WGPUVertexFormat_Unorm10_10_10_2;
        case SG_VERTEXFORMAT_HALF2:         return WGPUVertexFormat_Float16x2;
        case SG_VERTEXFORMAT_HALF4:         return WGPUVertexFormat_Float16x4;
        default:
            SOKOL_UNREACHABLE;
            return WGPUVertexFormat_Force32;
    }
}

_SOKOL_PRIVATE WGPUPrimitiveTopology _sg_wgpu_topology(sg_primitive_type t) {
    switch (t) {
        case SG_PRIMITIVETYPE_POINTS:           return WGPUPrimitiveTopology_PointList;
        case SG_PRIMITIVETYPE_LINES:            return WGPUPrimitiveTopology_LineList;
        case SG_PRIMITIVETYPE_LINE_STRIP:       return WGPUPrimitiveTopology_LineStrip;
        case SG_PRIMITIVETYPE_TRIANGLES:        return WGPUPrimitiveTopology_TriangleList;
        case SG_PRIMITIVETYPE_TRIANGLE_STRIP:   return WGPUPrimitiveTopology_TriangleStrip;
        default:
            SOKOL_UNREACHABLE;
            return WGPUPrimitiveTopology_Force32;
    }
}

_SOKOL_PRIVATE WGPUFrontFace _sg_wgpu_frontface(sg_face_winding fw) {
    return (fw == SG_FACEWINDING_CCW) ? WGPUFrontFace_CCW : WGPUFrontFace_CW;
}

_SOKOL_PRIVATE WGPUCullMode _sg_wgpu_cullmode(sg_cull_mode cm) {
    switch (cm) {
        case SG_CULLMODE_NONE:      return WGPUCullMode_None;
        case SG_CULLMODE_FRONT:     return WGPUCullMode_Front;
        case SG_CULLMODE_BACK:      return WGPUCullMode_Back;
        default:
            SOKOL_UNREACHABLE;
            return WGPUCullMode_Force32;
    }
}

_SOKOL_PRIVATE WGPUTextureFormat _sg_wgpu_textureformat(sg_pixel_format p) {
    switch (p) {
        case SG_PIXELFORMAT_NONE:           return WGPUTextureFormat_Undefined;
        case SG_PIXELFORMAT_R8:             return WGPUTextureFormat_R8Unorm;
        case SG_PIXELFORMAT_R8SN:           return WGPUTextureFormat_R8Snorm;
        case SG_PIXELFORMAT_R8UI:           return WGPUTextureFormat_R8Uint;
        case SG_PIXELFORMAT_R8SI:           return WGPUTextureFormat_R8Sint;
        case SG_PIXELFORMAT_R16UI:          return WGPUTextureFormat_R16Uint;
        case SG_PIXELFORMAT_R16SI:          return WGPUTextureFormat_R16Sint;
        case SG_PIXELFORMAT_R16F:           return WGPUTextureFormat_R16Float;
        case SG_PIXELFORMAT_RG8:            return WGPUTextureFormat_RG8Unorm;
        case SG_PIXELFORMAT_RG8SN:          return WGPUTextureFormat_RG8Snorm;
        case SG_PIXELFORMAT_RG8UI:          return WGPUTextureFormat_RG8Uint;
        case SG_PIXELFORMAT_RG8SI:          return WGPUTextureFormat_RG8Sint;
        case SG_PIXELFORMAT_R32UI:          return WGPUTextureFormat_R32Uint;
        case SG_PIXELFORMAT_R32SI:          return WGPUTextureFormat_R32Sint;
        case SG_PIXELFORMAT_R32F:           return WGPUTextureFormat_R32Float;
        case SG_PIXELFORMAT_RG16UI:         return WGPUTextureFormat_RG16Uint;
        case SG_PIXELFORMAT_RG16SI:         return WGPUTextureFormat_RG16Sint;
        case SG_PIXELFORMAT_RG16F:          return WGPUTextureFormat_RG16Float;
        case SG_PIXELFORMAT_RGBA8:          return WGPUTextureFormat_RGBA8Unorm;
        case SG_PIXELFORMAT_SRGB8A8:        return WGPUTextureFormat_RGBA8UnormSrgb;
        case SG_PIXELFORMAT_RGBA8SN:        return WGPUTextureFormat_RGBA8Snorm;
        case SG_PIXELFORMAT_RGBA8UI:        return WGPUTextureFormat_RGBA8Uint;
        case SG_PIXELFORMAT_RGBA8SI:        return WGPUTextureFormat_RGBA8Sint;
        case SG_PIXELFORMAT_BGRA8:          return WGPUTextureFormat_BGRA8Unorm;
        case SG_PIXELFORMAT_RGB10A2:        return WGPUTextureFormat_RGB10A2Unorm;
        case SG_PIXELFORMAT_RG11B10F:       return WGPUTextureFormat_RG11B10Ufloat;
        case SG_PIXELFORMAT_RG32UI:         return WGPUTextureFormat_RG32Uint;
        case SG_PIXELFORMAT_RG32SI:         return WGPUTextureFormat_RG32Sint;
        case SG_PIXELFORMAT_RG32F:          return WGPUTextureFormat_RG32Float;
        case SG_PIXELFORMAT_RGBA16UI:       return WGPUTextureFormat_RGBA16Uint;
        case SG_PIXELFORMAT_RGBA16SI:       return WGPUTextureFormat_RGBA16Sint;
        case SG_PIXELFORMAT_RGBA16F:        return WGPUTextureFormat_RGBA16Float;
        case SG_PIXELFORMAT_RGBA32UI:       return WGPUTextureFormat_RGBA32Uint;
        case SG_PIXELFORMAT_RGBA32SI:       return WGPUTextureFormat_RGBA32Sint;
        case SG_PIXELFORMAT_RGBA32F:        return WGPUTextureFormat_RGBA32Float;
        case SG_PIXELFORMAT_DEPTH:          return WGPUTextureFormat_Depth32Float;
        case SG_PIXELFORMAT_DEPTH_STENCIL:  return WGPUTextureFormat_Depth32FloatStencil8;
        case SG_PIXELFORMAT_BC1_RGBA:       return WGPUTextureFormat_BC1RGBAUnorm;
        case SG_PIXELFORMAT_BC2_RGBA:       return WGPUTextureFormat_BC2RGBAUnorm;
        case SG_PIXELFORMAT_BC3_RGBA:       return WGPUTextureFormat_BC3RGBAUnorm;
        case SG_PIXELFORMAT_BC3_SRGBA:      return WGPUTextureFormat_BC3RGBAUnormSrgb;
        case SG_PIXELFORMAT_BC4_R:          return WGPUTextureFormat_BC4RUnorm;
        case SG_PIXELFORMAT_BC4_RSN:        return WGPUTextureFormat_BC4RSnorm;
        case SG_PIXELFORMAT_BC5_RG:         return WGPUTextureFormat_BC5RGUnorm;
        case SG_PIXELFORMAT_BC5_RGSN:       return WGPUTextureFormat_BC5RGSnorm;
        case SG_PIXELFORMAT_BC6H_RGBF:      return WGPUTextureFormat_BC6HRGBFloat;
        case SG_PIXELFORMAT_BC6H_RGBUF:     return WGPUTextureFormat_BC6HRGBUfloat;
        case SG_PIXELFORMAT_BC7_RGBA:       return WGPUTextureFormat_BC7RGBAUnorm;
        case SG_PIXELFORMAT_BC7_SRGBA:      return WGPUTextureFormat_BC7RGBAUnormSrgb;
        case SG_PIXELFORMAT_ETC2_RGB8:      return WGPUTextureFormat_ETC2RGB8Unorm;
        case SG_PIXELFORMAT_ETC2_RGB8A1:    return WGPUTextureFormat_ETC2RGB8A1Unorm;
        case SG_PIXELFORMAT_ETC2_RGBA8:     return WGPUTextureFormat_ETC2RGBA8Unorm;
        case SG_PIXELFORMAT_ETC2_SRGB8:     return WGPUTextureFormat_ETC2RGB8UnormSrgb;
        case SG_PIXELFORMAT_ETC2_SRGB8A8:   return WGPUTextureFormat_ETC2RGBA8UnormSrgb;
        case SG_PIXELFORMAT_EAC_R11:        return WGPUTextureFormat_EACR11Unorm;
        case SG_PIXELFORMAT_EAC_R11SN:      return WGPUTextureFormat_EACR11Snorm;
        case SG_PIXELFORMAT_EAC_RG11:       return WGPUTextureFormat_EACRG11Unorm;
        case SG_PIXELFORMAT_EAC_RG11SN:     return WGPUTextureFormat_EACRG11Snorm;
        case SG_PIXELFORMAT_RGB9E5:         return WGPUTextureFormat_RGB9E5Ufloat;
        case SG_PIXELFORMAT_ASTC_4x4_RGBA:  return WGPUTextureFormat_ASTC4x4Unorm;
        case SG_PIXELFORMAT_ASTC_4x4_SRGBA: return WGPUTextureFormat_ASTC4x4UnormSrgb;
        // NOT SUPPORTED
        case SG_PIXELFORMAT_R16:
        case SG_PIXELFORMAT_R16SN:
        case SG_PIXELFORMAT_RG16:
        case SG_PIXELFORMAT_RG16SN:
        case SG_PIXELFORMAT_RGBA16:
        case SG_PIXELFORMAT_RGBA16SN:
            return WGPUTextureFormat_Undefined;

        default:
            SOKOL_UNREACHABLE;
            return WGPUTextureFormat_Force32;
    }
}

_SOKOL_PRIVATE WGPUCompareFunction _sg_wgpu_comparefunc(sg_compare_func f) {
    switch (f) {
        case SG_COMPAREFUNC_NEVER:          return WGPUCompareFunction_Never;
        case SG_COMPAREFUNC_LESS:           return WGPUCompareFunction_Less;
        case SG_COMPAREFUNC_EQUAL:          return WGPUCompareFunction_Equal;
        case SG_COMPAREFUNC_LESS_EQUAL:     return WGPUCompareFunction_LessEqual;
        case SG_COMPAREFUNC_GREATER:        return WGPUCompareFunction_Greater;
        case SG_COMPAREFUNC_NOT_EQUAL:      return WGPUCompareFunction_NotEqual;
        case SG_COMPAREFUNC_GREATER_EQUAL:  return WGPUCompareFunction_GreaterEqual;
        case SG_COMPAREFUNC_ALWAYS:         return WGPUCompareFunction_Always;
        default:
            SOKOL_UNREACHABLE;
            return WGPUCompareFunction_Force32;
    }
}

_SOKOL_PRIVATE WGPUStencilOperation _sg_wgpu_stencilop(sg_stencil_op op) {
    switch (op) {
        case SG_STENCILOP_KEEP:         return WGPUStencilOperation_Keep;
        case SG_STENCILOP_ZERO:         return WGPUStencilOperation_Zero;
        case SG_STENCILOP_REPLACE:      return WGPUStencilOperation_Replace;
        case SG_STENCILOP_INCR_CLAMP:   return WGPUStencilOperation_IncrementClamp;
        case SG_STENCILOP_DECR_CLAMP:   return WGPUStencilOperation_DecrementClamp;
        case SG_STENCILOP_INVERT:       return WGPUStencilOperation_Invert;
        case SG_STENCILOP_INCR_WRAP:    return WGPUStencilOperation_IncrementWrap;
        case SG_STENCILOP_DECR_WRAP:    return WGPUStencilOperation_DecrementWrap;
        default:
            SOKOL_UNREACHABLE;
            return WGPUStencilOperation_Force32;
    }
}

_SOKOL_PRIVATE WGPUBlendOperation _sg_wgpu_blendop(sg_blend_op op) {
    switch (op) {
        case SG_BLENDOP_ADD:                return WGPUBlendOperation_Add;
        case SG_BLENDOP_SUBTRACT:           return WGPUBlendOperation_Subtract;
        case SG_BLENDOP_REVERSE_SUBTRACT:   return WGPUBlendOperation_ReverseSubtract;
        case SG_BLENDOP_MIN:                return WGPUBlendOperation_Min;
        case SG_BLENDOP_MAX:                return WGPUBlendOperation_Max;
        default:
            SOKOL_UNREACHABLE;
            return WGPUBlendOperation_Force32;
    }
}

_SOKOL_PRIVATE WGPUBlendFactor _sg_wgpu_blendfactor(sg_blend_factor f) {
    switch (f) {
        case SG_BLENDFACTOR_ZERO:                   return WGPUBlendFactor_Zero;
        case SG_BLENDFACTOR_ONE:                    return WGPUBlendFactor_One;
        case SG_BLENDFACTOR_SRC_COLOR:              return WGPUBlendFactor_Src;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_COLOR:    return WGPUBlendFactor_OneMinusSrc;
        case SG_BLENDFACTOR_SRC_ALPHA:              return WGPUBlendFactor_SrcAlpha;
        case SG_BLENDFACTOR_ONE_MINUS_SRC_ALPHA:    return WGPUBlendFactor_OneMinusSrcAlpha;
        case SG_BLENDFACTOR_DST_COLOR:              return WGPUBlendFactor_Dst;
        case SG_BLENDFACTOR_ONE_MINUS_DST_COLOR:    return WGPUBlendFactor_OneMinusDst;
        case SG_BLENDFACTOR_DST_ALPHA:              return WGPUBlendFactor_DstAlpha;
        case SG_BLENDFACTOR_ONE_MINUS_DST_ALPHA:    return WGPUBlendFactor_OneMinusDstAlpha;
        case SG_BLENDFACTOR_SRC_ALPHA_SATURATED:    return WGPUBlendFactor_SrcAlphaSaturated;
        case SG_BLENDFACTOR_BLEND_COLOR:            return WGPUBlendFactor_Constant;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_COLOR:  return WGPUBlendFactor_OneMinusConstant;
        // FIXME: separate blend alpha value not supported?
        case SG_BLENDFACTOR_BLEND_ALPHA:            return WGPUBlendFactor_Constant;
        case SG_BLENDFACTOR_ONE_MINUS_BLEND_ALPHA:  return WGPUBlendFactor_OneMinusConstant;
        default:
            SOKOL_UNREACHABLE;
            return WGPUBlendFactor_Force32;
    }
}

_SOKOL_PRIVATE WGPUColorWriteMask _sg_wgpu_colorwritemask(uint8_t m) {
    // FIXME: change to WGPUColorWriteMask once Emscripten and Dawn webgpu.h agree
    int res = 0;
    if (0 != (m & SG_COLORMASK_R)) {
        res |= WGPUColorWriteMask_Red;
    }
    if (0 != (m & SG_COLORMASK_G)) {
        res |= WGPUColorWriteMask_Green;
    }
    if (0 != (m & SG_COLORMASK_B)) {
        res |= WGPUColorWriteMask_Blue;
    }
    if (0 != (m & SG_COLORMASK_A)) {
        res |= WGPUColorWriteMask_Alpha;
    }
    return (WGPUColorWriteMask)res;
}

_SOKOL_PRIVATE WGPUShaderStage _sg_wgpu_shader_stage(sg_shader_stage stage) {
    switch (stage) {
        case SG_SHADERSTAGE_VERTEX: return WGPUShaderStage_Vertex;
        case SG_SHADERSTAGE_FRAGMENT: return WGPUShaderStage_Fragment;
        case SG_SHADERSTAGE_COMPUTE: return WGPUShaderStage_Compute;
        default: SOKOL_UNREACHABLE; return WGPUShaderStage_None;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_init_caps(void) {
    _sg.backend = SG_BACKEND_WGPU;
    _sg.features.origin_top_left = true;
    _sg.features.image_clamp_to_border = false;
    _sg.features.mrt_independent_blend_state = true;
    _sg.features.mrt_independent_write_mask = true;
    _sg.features.compute = true;
    _sg.features.msaa_image_bindings = true;

    wgpuDeviceGetLimits(_sg.wgpu.dev, &_sg.wgpu.limits);

    const WGPULimits* l = &_sg.wgpu.limits.limits;
    _sg.limits.max_image_size_2d = (int) l->maxTextureDimension2D;
    _sg.limits.max_image_size_cube = (int) l->maxTextureDimension2D; // not a bug, see: https://github.com/gpuweb/gpuweb/issues/1327
    _sg.limits.max_image_size_3d = (int) l->maxTextureDimension3D;
    _sg.limits.max_image_size_array = (int) l->maxTextureDimension2D;
    _sg.limits.max_image_array_layers = (int) l->maxTextureArrayLayers;
    _sg.limits.max_vertex_attrs = SG_MAX_VERTEX_ATTRIBUTES;

    // NOTE: no WGPUTextureFormat_R16Unorm
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_SRGB8A8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_BGRA8]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_R16F]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RG16F]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    _sg_pixelformat_all(&_sg.formats[SG_PIXELFORMAT_RGB10A2]);

    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_R8SN]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG8SN]);
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);

    // FIXME: can be made renderable via extension
    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RG11B10F]);

    // NOTE: msaa rendering is possible in WebGPU, but no resolve
    // which is a combination that's not currently supported in sokol-gfx
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R8UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R8SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG8UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG8SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R16UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R16SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG16UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG16SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
    _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);

    if (wgpuDeviceHasFeature(_sg.wgpu.dev, WGPUFeatureName_Float32Filterable)) {
        _sg_pixelformat_sfr(&_sg.formats[SG_PIXELFORMAT_R32F]);
        _sg_pixelformat_sfr(&_sg.formats[SG_PIXELFORMAT_RG32F]);
        _sg_pixelformat_sfr(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
    } else {
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_R32F]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RG32F]);
        _sg_pixelformat_sr(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
    }

    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH]);
    _sg_pixelformat_srmd(&_sg.formats[SG_PIXELFORMAT_DEPTH_STENCIL]);

    _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_RGB9E5]);

    if (wgpuDeviceHasFeature(_sg.wgpu.dev, WGPUFeatureName_TextureCompressionBC)) {
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC1_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC2_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC3_SRGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_R]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC4_RSN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RG]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC5_RGSN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBF]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC6H_RGBUF]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_BC7_SRGBA]);
    }
    if (wgpuDeviceHasFeature(_sg.wgpu.dev, WGPUFeatureName_TextureCompressionETC2)) {
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGB8A1]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_RGBA8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ETC2_SRGB8A8]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_R11SN]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_EAC_RG11SN]);
    }

    if (wgpuDeviceHasFeature(_sg.wgpu.dev, WGPUFeatureName_TextureCompressionASTC)) {
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_RGBA]);
        _sg_pixelformat_sf(&_sg.formats[SG_PIXELFORMAT_ASTC_4x4_SRGBA]);
    }

    // see: https://github.com/gpuweb/gpuweb/issues/513
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SN]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA8SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA16F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_R32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RG32F]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32UI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32SI]);
    _sg_pixelformat_compute_all(&_sg.formats[SG_PIXELFORMAT_RGBA32F]);
}

_SOKOL_PRIVATE void _sg_wgpu_uniform_buffer_init(const sg_desc* desc) {
    SOKOL_ASSERT(0 == _sg.wgpu.uniform.staging);
    SOKOL_ASSERT(0 == _sg.wgpu.uniform.buf);

    // Add the max-uniform-update size (64 KB) to the requested buffer size,
    // this is to prevent validation errors in the WebGPU implementation
    // if the entire buffer size is used per frame. 64 KB is the allowed
    // max uniform update size on NVIDIA
    //
    // FIXME: is this still needed?
    _sg.wgpu.uniform.num_bytes = (uint32_t)(desc->uniform_buffer_size + _SG_WGPU_MAX_UNIFORM_UPDATE_SIZE);
    _sg.wgpu.uniform.staging = (uint8_t*)_sg_malloc(_sg.wgpu.uniform.num_bytes);

    WGPUBufferDescriptor ub_desc;
    _sg_clear(&ub_desc, sizeof(ub_desc));
    ub_desc.size = _sg.wgpu.uniform.num_bytes;
    ub_desc.usage = WGPUBufferUsage_Uniform|WGPUBufferUsage_CopyDst;
    _sg.wgpu.uniform.buf = wgpuDeviceCreateBuffer(_sg.wgpu.dev, &ub_desc);
    SOKOL_ASSERT(_sg.wgpu.uniform.buf);
}

_SOKOL_PRIVATE void _sg_wgpu_uniform_buffer_discard(void) {
    if (_sg.wgpu.uniform.buf) {
        wgpuBufferRelease(_sg.wgpu.uniform.buf);
        _sg.wgpu.uniform.buf = 0;
    }
    if (_sg.wgpu.uniform.staging) {
        _sg_free(_sg.wgpu.uniform.staging);
        _sg.wgpu.uniform.staging = 0;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_uniform_buffer_on_commit(void) {
    wgpuQueueWriteBuffer(_sg.wgpu.queue, _sg.wgpu.uniform.buf, 0, _sg.wgpu.uniform.staging, _sg.wgpu.uniform.offset);
    _sg_stats_add(wgpu.uniforms.size_write_buffer, _sg.wgpu.uniform.offset);
    _sg.wgpu.uniform.offset = 0;
    _sg_clear(_sg.wgpu.uniform.bind_offsets, sizeof(_sg.wgpu.uniform.bind_offsets));
}

_SOKOL_PRIVATE void _sg_wgpu_bindgroups_pool_init(const sg_desc* desc) {
    SOKOL_ASSERT((desc->wgpu_bindgroups_cache_size > 0) && (desc->wgpu_bindgroups_cache_size < _SG_MAX_POOL_SIZE));
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    SOKOL_ASSERT(0 == p->bindgroups);
    const int pool_size = desc->wgpu_bindgroups_cache_size;
    _sg_pool_init(&p->pool, pool_size);
    size_t pool_byte_size = sizeof(_sg_wgpu_bindgroup_t) * (size_t)p->pool.size;
    p->bindgroups = (_sg_wgpu_bindgroup_t*) _sg_malloc_clear(pool_byte_size);
}

_SOKOL_PRIVATE void _sg_wgpu_bindgroups_pool_discard(void) {
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    SOKOL_ASSERT(p->bindgroups);
    _sg_free(p->bindgroups); p->bindgroups = 0;
    _sg_pool_discard(&p->pool);
}

_SOKOL_PRIVATE _sg_wgpu_bindgroup_t* _sg_wgpu_bindgroup_at(uint32_t bg_id) {
    SOKOL_ASSERT(SG_INVALID_ID != bg_id);
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    int slot_index = _sg_slot_index(bg_id);
    SOKOL_ASSERT((slot_index > _SG_INVALID_SLOT_INDEX) && (slot_index < p->pool.size));
    return &p->bindgroups[slot_index];
}

_SOKOL_PRIVATE _sg_wgpu_bindgroup_t* _sg_wgpu_lookup_bindgroup(uint32_t bg_id) {
    if (SG_INVALID_ID != bg_id) {
        _sg_wgpu_bindgroup_t* bg = _sg_wgpu_bindgroup_at(bg_id);
        if (bg->slot.id == bg_id) {
            return bg;
        }
    }
    return 0;
}

_SOKOL_PRIVATE _sg_wgpu_bindgroup_handle_t _sg_wgpu_alloc_bindgroup(void) {
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    _sg_wgpu_bindgroup_handle_t res;
    int slot_index = _sg_pool_alloc_index(&p->pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&p->pool, &p->bindgroups[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(WGPU_BINDGROUPS_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE void _sg_wgpu_dealloc_bindgroup(_sg_wgpu_bindgroup_t* bg) {
    SOKOL_ASSERT(bg && (bg->slot.state == SG_RESOURCESTATE_ALLOC) && (bg->slot.id != SG_INVALID_ID));
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    _sg_pool_free_index(&p->pool, _sg_slot_index(bg->slot.id));
    _sg_slot_reset(&bg->slot);
}

_SOKOL_PRIVATE void _sg_wgpu_reset_bindgroup_to_alloc_state(_sg_wgpu_bindgroup_t* bg) {
    SOKOL_ASSERT(bg);
    _sg_slot_t slot = bg->slot;
    _sg_clear(bg, sizeof(_sg_wgpu_bindgroup_t));
    bg->slot = slot;
    bg->slot.state = SG_RESOURCESTATE_ALLOC;
}

// MurmurHash64B (see: https://github.com/aappleby/smhasher/blob/61a0530f28277f2e850bfc39600ce61d02b518de/src/MurmurHash2.cpp#L142)
_SOKOL_PRIVATE uint64_t _sg_wgpu_hash(const void* key, int len, uint64_t seed) {
    const uint32_t m = 0x5bd1e995;
    const int r = 24;
    uint32_t h1 = (uint32_t)seed ^ (uint32_t)len;
    uint32_t h2 = (uint32_t)(seed >> 32);
    const uint32_t * data = (const uint32_t *)key;
    while (len >= 8) {
        uint32_t k1 = *data++;
        k1 *= m; k1 ^= k1 >> r; k1 *= m;
        h1 *= m; h1 ^= k1;
        len -= 4;
        uint32_t k2 = *data++;
        k2 *= m; k2 ^= k2 >> r; k2 *= m;
        h2 *= m; h2 ^= k2;
        len -= 4;
    }
    if (len >= 4) {
        uint32_t k1 = *data++;
        k1 *= m; k1 ^= k1 >> r; k1 *= m;
        h1 *= m; h1 ^= k1;
        len -= 4;
    }
    switch(len) {
        case 3: h2 ^= (uint32_t)(((unsigned char*)data)[2] << 16);
        case 2: h2 ^= (uint32_t)(((unsigned char*)data)[1] << 8);
        case 1: h2 ^= ((unsigned char*)data)[0];
        h2 *= m;
    };
    h1 ^= h2 >> 18; h1 *= m;
    h2 ^= h1 >> 22; h2 *= m;
    h1 ^= h2 >> 17; h1 *= m;
    h2 ^= h1 >> 19; h2 *= m;
    uint64_t h = h1;
    h = (h << 32) | h2;
    return h;
}

_SOKOL_PRIVATE uint64_t _sg_wgpu_bindgroups_cache_item(_sg_wgpu_bindgroups_cache_item_type_t type, uint8_t wgpu_binding, uint32_t id) {
    // key pattern is bbbbttttiiiiiiii
    const uint64_t bb = (uint64_t)wgpu_binding;
    const uint64_t tttt = (uint64_t)type;
    const uint64_t iiiiiiii = (uint64_t)id;
    return (bb << 56) | (bb << 48) | (tttt << 32) | iiiiiiii;
}

_SOKOL_PRIVATE uint64_t _sg_wgpu_bindgroups_cache_pip_item(uint32_t id) {
    return _sg_wgpu_bindgroups_cache_item(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_PIPELINE, 0xFF, id);
}

_SOKOL_PRIVATE uint64_t _sg_wgpu_bindgroups_cache_image_item(uint8_t wgpu_binding, uint32_t id) {
    return _sg_wgpu_bindgroups_cache_item(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_IMAGE, wgpu_binding, id);
}

_SOKOL_PRIVATE uint64_t _sg_wgpu_bindgroups_cache_sampler_item(uint8_t wgpu_binding, uint32_t id) {
    return _sg_wgpu_bindgroups_cache_item(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_SAMPLER, wgpu_binding, id);
}

_SOKOL_PRIVATE uint64_t _sg_wgpu_bindgroups_cache_sbuf_item(uint8_t wgpu_binding, uint32_t id) {
    return _sg_wgpu_bindgroups_cache_item(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_STORAGEBUFFER, wgpu_binding, id);
}

_SOKOL_PRIVATE void _sg_wgpu_init_bindgroups_cache_key(_sg_wgpu_bindgroups_cache_key_t* key, const _sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip);
    const _sg_shader_t* shd = bnd->pip->shader;
    SOKOL_ASSERT(shd && shd->slot.id == bnd->pip->cmn.shader_id.id);

    _sg_clear(key->items, sizeof(key->items));
    key->items[0] = _sg_wgpu_bindgroups_cache_pip_item(bnd->pip->slot.id);
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (shd->cmn.images[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->imgs[i]);
        const size_t item_idx = i + 1;
        SOKOL_ASSERT(item_idx < _SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS);
        SOKOL_ASSERT(0 == key->items[item_idx]);
        const uint8_t wgpu_binding = shd->wgpu.img_grp1_bnd_n[i];
        const uint32_t id = bnd->imgs[i]->slot.id;
        key->items[item_idx] = _sg_wgpu_bindgroups_cache_image_item(wgpu_binding, id);
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (shd->cmn.samplers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->smps[i]);
        const size_t item_idx = i + 1 + SG_MAX_IMAGE_BINDSLOTS;
        SOKOL_ASSERT(item_idx < _SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS);
        SOKOL_ASSERT(0 == key->items[item_idx]);
        const uint8_t wgpu_binding = shd->wgpu.smp_grp1_bnd_n[i];
        const uint32_t id = bnd->smps[i]->slot.id;
        key->items[item_idx] = _sg_wgpu_bindgroups_cache_sampler_item(wgpu_binding, id);
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (shd->cmn.storage_buffers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->sbufs[i]);
        const size_t item_idx = i + 1 + SG_MAX_IMAGE_BINDSLOTS + SG_MAX_SAMPLER_BINDSLOTS;
        SOKOL_ASSERT(item_idx < _SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS);
        SOKOL_ASSERT(0 == key->items[item_idx]);
        const uint8_t wgpu_binding = shd->wgpu.sbuf_grp1_bnd_n[i];
        const uint32_t id = bnd->sbufs[i]->slot.id;
        key->items[item_idx] = _sg_wgpu_bindgroups_cache_sbuf_item(wgpu_binding, id);
    }
    key->hash = _sg_wgpu_hash(&key->items, (int)sizeof(key->items), 0x1234567887654321);
}

_SOKOL_PRIVATE bool _sg_wgpu_compare_bindgroups_cache_key(_sg_wgpu_bindgroups_cache_key_t* k0, _sg_wgpu_bindgroups_cache_key_t* k1) {
    SOKOL_ASSERT(k0 && k1);
    if (k0->hash != k1->hash) {
        return false;
    }
    if (memcmp(&k0->items, &k1->items, sizeof(k0->items)) != 0) {
        _sg_stats_add(wgpu.bindings.num_bindgroup_cache_hash_vs_key_mismatch, 1);
        return false;
    }
    return true;
}

_SOKOL_PRIVATE _sg_wgpu_bindgroup_t* _sg_wgpu_create_bindgroup(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(_sg.wgpu.dev);
    SOKOL_ASSERT(bnd->pip);
    const _sg_shader_t* shd = bnd->pip->shader;
    SOKOL_ASSERT(shd && (shd->slot.id == bnd->pip->cmn.shader_id.id));
    _sg_stats_add(wgpu.bindings.num_create_bindgroup, 1);
    _sg_wgpu_bindgroup_handle_t bg_id = _sg_wgpu_alloc_bindgroup();
    if (bg_id.id == SG_INVALID_ID) {
        return 0;
    }
    _sg_wgpu_bindgroup_t* bg = _sg_wgpu_bindgroup_at(bg_id.id);
    SOKOL_ASSERT(bg && (bg->slot.state == SG_RESOURCESTATE_ALLOC));

    // create wgpu bindgroup object (also see _sg_wgpu_create_shader())
    WGPUBindGroupLayout bgl = bnd->pip->shader->wgpu.bgl_img_smp_sbuf;
    SOKOL_ASSERT(bgl);
    WGPUBindGroupEntry bg_entries[_SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES];
    _sg_clear(&bg_entries, sizeof(bg_entries));
    size_t bgl_index = 0;
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (shd->cmn.images[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->imgs[i]);
        SOKOL_ASSERT(bgl_index < _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES);
        WGPUBindGroupEntry* bg_entry = &bg_entries[bgl_index];
        bg_entry->binding = shd->wgpu.img_grp1_bnd_n[i];
        bg_entry->textureView = bnd->imgs[i]->wgpu.view;
        bgl_index += 1;
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (shd->cmn.samplers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->smps[i]);
        SOKOL_ASSERT(bgl_index < _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES);
        WGPUBindGroupEntry* bg_entry = &bg_entries[bgl_index];
        bg_entry->binding = shd->wgpu.smp_grp1_bnd_n[i];
        bg_entry->sampler = bnd->smps[i]->wgpu.smp;
        bgl_index += 1;
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (shd->cmn.storage_buffers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        SOKOL_ASSERT(bnd->sbufs[i]);
        SOKOL_ASSERT(bgl_index < _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES);
        WGPUBindGroupEntry* bg_entry = &bg_entries[bgl_index];
        bg_entry->binding = shd->wgpu.sbuf_grp1_bnd_n[i];
        bg_entry->buffer = bnd->sbufs[i]->wgpu.buf;
        bg_entry->size = (uint64_t) bnd->sbufs[i]->cmn.size;
        bgl_index += 1;
    }
    WGPUBindGroupDescriptor bg_desc;
    _sg_clear(&bg_desc, sizeof(bg_desc));
    bg_desc.layout = bgl;
    bg_desc.entryCount = bgl_index;
    bg_desc.entries = bg_entries;
    bg->bindgroup = wgpuDeviceCreateBindGroup(_sg.wgpu.dev, &bg_desc);
    if (bg->bindgroup == 0) {
        _SG_ERROR(WGPU_CREATEBINDGROUP_FAILED);
        bg->slot.state = SG_RESOURCESTATE_FAILED;
        return bg;
    }
    _sg_wgpu_init_bindgroups_cache_key(&bg->key, bnd);
    bg->slot.state = SG_RESOURCESTATE_VALID;
    return bg;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_bindgroup(_sg_wgpu_bindgroup_t* bg) {
    SOKOL_ASSERT(bg);
    _sg_stats_add(wgpu.bindings.num_discard_bindgroup, 1);
    if (bg->slot.state == SG_RESOURCESTATE_VALID) {
        if (bg->bindgroup) {
            wgpuBindGroupRelease(bg->bindgroup);
            bg->bindgroup = 0;
        }
        _sg_wgpu_reset_bindgroup_to_alloc_state(bg);
        SOKOL_ASSERT(bg->slot.state == SG_RESOURCESTATE_ALLOC);
    }
    if (bg->slot.state == SG_RESOURCESTATE_ALLOC) {
        _sg_wgpu_dealloc_bindgroup(bg);
        SOKOL_ASSERT(bg->slot.state == SG_RESOURCESTATE_INITIAL);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_discard_all_bindgroups(void) {
    _sg_wgpu_bindgroups_pool_t* p = &_sg.wgpu.bindgroups_pool;
    for (int i = 0; i < p->pool.size; i++) {
        sg_resource_state state = p->bindgroups[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_wgpu_discard_bindgroup(&p->bindgroups[i]);
        }
    }
}

_SOKOL_PRIVATE void _sg_wgpu_bindgroups_cache_init(const sg_desc* desc) {
    SOKOL_ASSERT(desc);
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.num == 0);
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.index_mask == 0);
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.items == 0);
    const int num = desc->wgpu_bindgroups_cache_size;
    if (num <= 1) {
        _SG_PANIC(WGPU_BINDGROUPSCACHE_SIZE_GREATER_ONE);
    }
    if (!_sg_ispow2(num)) {
        _SG_PANIC(WGPU_BINDGROUPSCACHE_SIZE_POW2);
    }
    _sg.wgpu.bindgroups_cache.num = (uint32_t)desc->wgpu_bindgroups_cache_size;
    _sg.wgpu.bindgroups_cache.index_mask = _sg.wgpu.bindgroups_cache.num - 1;
    size_t size_in_bytes = sizeof(_sg_wgpu_bindgroup_handle_t) * (size_t)num;
    _sg.wgpu.bindgroups_cache.items = (_sg_wgpu_bindgroup_handle_t*)_sg_malloc_clear(size_in_bytes);
}

_SOKOL_PRIVATE void _sg_wgpu_bindgroups_cache_discard(void) {
    if (_sg.wgpu.bindgroups_cache.items) {
        _sg_free(_sg.wgpu.bindgroups_cache.items);
        _sg.wgpu.bindgroups_cache.items = 0;
    }
    _sg.wgpu.bindgroups_cache.num = 0;
    _sg.wgpu.bindgroups_cache.index_mask = 0;
}

_SOKOL_PRIVATE void _sg_wgpu_bindgroups_cache_set(uint64_t hash, uint32_t bg_id) {
    uint32_t index = hash & _sg.wgpu.bindgroups_cache.index_mask;
    SOKOL_ASSERT(index < _sg.wgpu.bindgroups_cache.num);
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.items);
    _sg.wgpu.bindgroups_cache.items[index].id = bg_id;
}

_SOKOL_PRIVATE uint32_t _sg_wgpu_bindgroups_cache_get(uint64_t hash) {
    uint32_t index = hash & _sg.wgpu.bindgroups_cache.index_mask;
    SOKOL_ASSERT(index < _sg.wgpu.bindgroups_cache.num);
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.items);
    return _sg.wgpu.bindgroups_cache.items[index].id;
}

// called from wgpu resource destroy functions to also invalidate any
// bindgroups cache slot and bindgroup referencing that resource
_SOKOL_PRIVATE void _sg_wgpu_bindgroups_cache_invalidate(_sg_wgpu_bindgroups_cache_item_type_t type, uint32_t id) {
    const uint64_t key_mask = 0x0000FFFFFFFFFFFF;
    const uint64_t key_item = _sg_wgpu_bindgroups_cache_item(type, 0, id) & key_mask;
    SOKOL_ASSERT(_sg.wgpu.bindgroups_cache.items);
    for (uint32_t cache_item_idx = 0; cache_item_idx < _sg.wgpu.bindgroups_cache.num; cache_item_idx++) {
        const uint32_t bg_id = _sg.wgpu.bindgroups_cache.items[cache_item_idx].id;
        if (bg_id != SG_INVALID_ID) {
            _sg_wgpu_bindgroup_t* bg = _sg_wgpu_lookup_bindgroup(bg_id);
            SOKOL_ASSERT(bg && (bg->slot.state == SG_RESOURCESTATE_VALID));
            // check if resource is in bindgroup, if yes discard bindgroup and invalidate cache slot
            bool invalidate_cache_item = false;
            for (int key_item_idx = 0; key_item_idx < _SG_WGPU_BINDGROUPSCACHEKEY_NUM_ITEMS; key_item_idx++) {
                if ((bg->key.items[key_item_idx] & key_mask) == key_item) {
                    invalidate_cache_item = true;
                    break;
                }
            }
            if (invalidate_cache_item) {
                _sg_wgpu_discard_bindgroup(bg); bg = 0;
                _sg_wgpu_bindgroups_cache_set(cache_item_idx, SG_INVALID_ID);
                _sg_stats_add(wgpu.bindings.num_bindgroup_cache_invalidates, 1);
            }
        }
    }
}

_SOKOL_PRIVATE void _sg_wgpu_bindings_cache_clear(void) {
    memset(&_sg.wgpu.bindings_cache, 0, sizeof(_sg.wgpu.bindings_cache));
}

_SOKOL_PRIVATE bool _sg_wgpu_bindings_cache_vb_dirty(size_t index, const _sg_buffer_t* vb, uint64_t offset) {
    SOKOL_ASSERT((index >= 0) && (index < SG_MAX_VERTEXBUFFER_BINDSLOTS));
    if (vb) {
        return (_sg.wgpu.bindings_cache.vbs[index].buffer.id != vb->slot.id)
            || (_sg.wgpu.bindings_cache.vbs[index].offset != offset);
    } else {
        return _sg.wgpu.bindings_cache.vbs[index].buffer.id != SG_INVALID_ID;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_bindings_cache_vb_update(size_t index, const _sg_buffer_t* vb, uint64_t offset) {
    SOKOL_ASSERT((index >= 0) && (index < SG_MAX_VERTEXBUFFER_BINDSLOTS));
    if (vb) {
        _sg.wgpu.bindings_cache.vbs[index].buffer.id = vb->slot.id;
        _sg.wgpu.bindings_cache.vbs[index].offset = offset;
    } else {
        _sg.wgpu.bindings_cache.vbs[index].buffer.id = SG_INVALID_ID;
        _sg.wgpu.bindings_cache.vbs[index].offset = 0;
    }
}

_SOKOL_PRIVATE bool _sg_wgpu_bindings_cache_ib_dirty(const _sg_buffer_t* ib, uint64_t offset) {
    if (ib) {
        return (_sg.wgpu.bindings_cache.ib.buffer.id != ib->slot.id)
            || (_sg.wgpu.bindings_cache.ib.offset != offset);
    } else {
        return _sg.wgpu.bindings_cache.ib.buffer.id != SG_INVALID_ID;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_bindings_cache_ib_update(const _sg_buffer_t* ib, uint64_t offset) {
    if (ib) {
        _sg.wgpu.bindings_cache.ib.buffer.id = ib->slot.id;
        _sg.wgpu.bindings_cache.ib.offset = offset;
    } else {
        _sg.wgpu.bindings_cache.ib.buffer.id = SG_INVALID_ID;
        _sg.wgpu.bindings_cache.ib.offset = 0;
    }
}

_SOKOL_PRIVATE bool _sg_wgpu_bindings_cache_bg_dirty(const _sg_wgpu_bindgroup_t* bg) {
    if (bg) {
        return _sg.wgpu.bindings_cache.bg.id != bg->slot.id;
    } else {
        return _sg.wgpu.bindings_cache.bg.id != SG_INVALID_ID;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_bindings_cache_bg_update(const _sg_wgpu_bindgroup_t* bg) {
    if (bg) {
        _sg.wgpu.bindings_cache.bg.id = bg->slot.id;
    } else {
        _sg.wgpu.bindings_cache.bg.id = SG_INVALID_ID;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_set_bindgroup(size_t bg_idx, _sg_wgpu_bindgroup_t* bg) {
    if (_sg_wgpu_bindings_cache_bg_dirty(bg)) {
        _sg_wgpu_bindings_cache_bg_update(bg);
        _sg_stats_add(wgpu.bindings.num_set_bindgroup, 1);
        if (_sg.cur_pass.is_compute) {
            SOKOL_ASSERT(_sg.wgpu.cpass_enc);
            if (bg) {
                SOKOL_ASSERT(bg->slot.state == SG_RESOURCESTATE_VALID);
                SOKOL_ASSERT(bg->bindgroup);
                wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, bg_idx, bg->bindgroup, 0, 0);
            } else {
                wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, bg_idx, _sg.wgpu.empty_bind_group, 0, 0);
            }
        } else {
            SOKOL_ASSERT(_sg.wgpu.rpass_enc);
            if (bg) {
                SOKOL_ASSERT(bg->slot.state == SG_RESOURCESTATE_VALID);
                SOKOL_ASSERT(bg->bindgroup);
                wgpuRenderPassEncoderSetBindGroup(_sg.wgpu.rpass_enc, bg_idx, bg->bindgroup, 0, 0);
            } else {
                wgpuRenderPassEncoderSetBindGroup(_sg.wgpu.rpass_enc, bg_idx, _sg.wgpu.empty_bind_group, 0, 0);
            }
        }
    } else {
        _sg_stats_add(wgpu.bindings.num_skip_redundant_bindgroup, 1);
    }
}

_SOKOL_PRIVATE bool _sg_wgpu_apply_bindings_bindgroup(_sg_bindings_ptrs_t* bnd) {
    if (!_sg.desc.wgpu_disable_bindgroups_cache) {
        _sg_wgpu_bindgroup_t* bg = 0;
        _sg_wgpu_bindgroups_cache_key_t key;
        _sg_wgpu_init_bindgroups_cache_key(&key, bnd);
        uint32_t bg_id = _sg_wgpu_bindgroups_cache_get(key.hash);
        if (bg_id != SG_INVALID_ID) {
            // potential cache hit
            bg = _sg_wgpu_lookup_bindgroup(bg_id);
            SOKOL_ASSERT(bg && (bg->slot.state == SG_RESOURCESTATE_VALID));
            if (!_sg_wgpu_compare_bindgroups_cache_key(&key, &bg->key)) {
                // cache collision, need to delete cached bindgroup
                _sg_stats_add(wgpu.bindings.num_bindgroup_cache_collisions, 1);
                _sg_wgpu_discard_bindgroup(bg);
                _sg_wgpu_bindgroups_cache_set(key.hash, SG_INVALID_ID);
                bg = 0;
            } else {
                _sg_stats_add(wgpu.bindings.num_bindgroup_cache_hits, 1);
            }
        } else {
            _sg_stats_add(wgpu.bindings.num_bindgroup_cache_misses, 1);
        }
        if (bg == 0) {
            // either no cache entry yet, or cache collision, create new bindgroup and store in cache
            bg = _sg_wgpu_create_bindgroup(bnd);
            _sg_wgpu_bindgroups_cache_set(key.hash, bg->slot.id);
        }
        if (bg && bg->slot.state == SG_RESOURCESTATE_VALID) {
            _sg_wgpu_set_bindgroup(_SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX, bg);
        } else {
            return false;
        }
    } else {
        // bindgroups cache disabled, create and destroy bindgroup on the fly (expensive!)
        _sg_wgpu_bindgroup_t* bg = _sg_wgpu_create_bindgroup(bnd);
        if (bg) {
            if (bg->slot.state == SG_RESOURCESTATE_VALID) {
                _sg_wgpu_set_bindgroup(_SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX, bg);
            }
            _sg_wgpu_discard_bindgroup(bg);
        } else {
            return false;
        }
    }
    return true;
}

_SOKOL_PRIVATE bool _sg_wgpu_apply_index_buffer(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);
    const _sg_buffer_t* ib = bnd->ib;
    uint64_t offset = (uint64_t)bnd->ib_offset;
    if (_sg_wgpu_bindings_cache_ib_dirty(ib, offset)) {
        _sg_wgpu_bindings_cache_ib_update(ib, offset);
        if (ib) {
            const WGPUIndexFormat format = _sg_wgpu_indexformat(bnd->pip->cmn.index_type);
            const uint64_t buf_size = (uint64_t)ib->cmn.size;
            SOKOL_ASSERT(buf_size > offset);
            const uint64_t max_bytes = buf_size - offset;
            wgpuRenderPassEncoderSetIndexBuffer(_sg.wgpu.rpass_enc, ib->wgpu.buf, format, offset, max_bytes);
        /* FIXME: the else-pass should actually set a null index buffer, but that doesn't seem to work yet
        } else {
            wgpuRenderPassEncoderSetIndexBuffer(_sg.wgpu.rpass_enc, 0, WGPUIndexFormat_Undefined, 0, 0);
        */
        }
        _sg_stats_add(wgpu.bindings.num_set_index_buffer, 1);
    } else {
        _sg_stats_add(wgpu.bindings.num_skip_redundant_index_buffer, 1);
    }
    return true;
}

_SOKOL_PRIVATE bool _sg_wgpu_apply_vertex_buffers(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);
    for (uint32_t slot = 0; slot < SG_MAX_VERTEXBUFFER_BINDSLOTS; slot++) {
        const _sg_buffer_t* vb = bnd->vbs[slot];
        const uint64_t offset = (uint64_t)bnd->vb_offsets[slot];
        if (_sg_wgpu_bindings_cache_vb_dirty(slot, vb, offset)) {
            _sg_wgpu_bindings_cache_vb_update(slot, vb, offset);
            if (vb) {
                const uint64_t buf_size = (uint64_t)vb->cmn.size;
                SOKOL_ASSERT(buf_size > offset);
                const uint64_t max_bytes = buf_size - offset;
                wgpuRenderPassEncoderSetVertexBuffer(_sg.wgpu.rpass_enc, slot, vb->wgpu.buf, offset, max_bytes);
            /* FIXME: the else-pass should actually set a null vertex buffer, but that doesn't seem to work yet
            } else {
                wgpuRenderPassEncoderSetVertexBuffer(_sg.wgpu.rpass_enc, slot, 0, 0, 0);
            */
            }
            _sg_stats_add(wgpu.bindings.num_set_vertex_buffer, 1);
        } else {
            _sg_stats_add(wgpu.bindings.num_skip_redundant_vertex_buffer, 1);
        }
    }
    return true;
}

_SOKOL_PRIVATE void _sg_wgpu_setup_backend(const sg_desc* desc) {
    SOKOL_ASSERT(desc);
    SOKOL_ASSERT(desc->environment.wgpu.device);
    SOKOL_ASSERT(desc->uniform_buffer_size > 0);
    _sg.backend = SG_BACKEND_WGPU;
    _sg.wgpu.valid = true;
    _sg.wgpu.dev = (WGPUDevice) desc->environment.wgpu.device;
    _sg.wgpu.queue = wgpuDeviceGetQueue(_sg.wgpu.dev);
    SOKOL_ASSERT(_sg.wgpu.queue);

    _sg_wgpu_init_caps();
    _sg_wgpu_uniform_buffer_init(desc);
    _sg_wgpu_bindgroups_pool_init(desc);
    _sg_wgpu_bindgroups_cache_init(desc);
    _sg_wgpu_bindings_cache_clear();

    // create an empty bind group
    WGPUBindGroupLayoutDescriptor bgl_desc;
    _sg_clear(&bgl_desc, sizeof(bgl_desc));
    WGPUBindGroupLayout empty_bgl = wgpuDeviceCreateBindGroupLayout(_sg.wgpu.dev, &bgl_desc);
    SOKOL_ASSERT(empty_bgl);
    WGPUBindGroupDescriptor bg_desc;
    _sg_clear(&bg_desc, sizeof(bg_desc));
    bg_desc.layout = empty_bgl;
    _sg.wgpu.empty_bind_group = wgpuDeviceCreateBindGroup(_sg.wgpu.dev, &bg_desc);
    SOKOL_ASSERT(_sg.wgpu.empty_bind_group);
    wgpuBindGroupLayoutRelease(empty_bgl);

    // create initial per-frame command encoder
    WGPUCommandEncoderDescriptor cmd_enc_desc;
    _sg_clear(&cmd_enc_desc, sizeof(cmd_enc_desc));
    _sg.wgpu.cmd_enc = wgpuDeviceCreateCommandEncoder(_sg.wgpu.dev, &cmd_enc_desc);
    SOKOL_ASSERT(_sg.wgpu.cmd_enc);
}

_SOKOL_PRIVATE void _sg_wgpu_discard_backend(void) {
    SOKOL_ASSERT(_sg.wgpu.valid);
    SOKOL_ASSERT(_sg.wgpu.cmd_enc);
    _sg.wgpu.valid = false;
    _sg_wgpu_discard_all_bindgroups();
    _sg_wgpu_bindgroups_cache_discard();
    _sg_wgpu_bindgroups_pool_discard();
    _sg_wgpu_uniform_buffer_discard();
    wgpuBindGroupRelease(_sg.wgpu.empty_bind_group); _sg.wgpu.empty_bind_group = 0;
    wgpuCommandEncoderRelease(_sg.wgpu.cmd_enc); _sg.wgpu.cmd_enc = 0;
    wgpuQueueRelease(_sg.wgpu.queue); _sg.wgpu.queue = 0;
}

_SOKOL_PRIVATE void _sg_wgpu_reset_state_cache(void) {
    _sg_wgpu_bindings_cache_clear();
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && desc);
    SOKOL_ASSERT(buf->cmn.size > 0);
    const bool injected = (0 != desc->wgpu_buffer);
    if (injected) {
        buf->wgpu.buf = (WGPUBuffer) desc->wgpu_buffer;
        wgpuBufferReference(buf->wgpu.buf);
    } else {
        // buffer mapping size must be multiple of 4, so round up buffer size (only a problem
        // with index buffers containing odd number of indices)
        const uint64_t wgpu_buf_size = _sg_roundup_u64((uint64_t)buf->cmn.size, 4);
        const bool map_at_creation = buf->cmn.usage.immutable && (desc->data.ptr);

        WGPUBufferDescriptor wgpu_buf_desc;
        _sg_clear(&wgpu_buf_desc, sizeof(wgpu_buf_desc));
        wgpu_buf_desc.usage = _sg_wgpu_buffer_usage(&buf->cmn.usage);
        wgpu_buf_desc.size = wgpu_buf_size;
        wgpu_buf_desc.mappedAtCreation = map_at_creation;
        wgpu_buf_desc.label = _sg_wgpu_stringview(desc->label);
        buf->wgpu.buf = wgpuDeviceCreateBuffer(_sg.wgpu.dev, &wgpu_buf_desc);
        if (0 == buf->wgpu.buf) {
            _SG_ERROR(WGPU_CREATE_BUFFER_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        // NOTE: assume that WebGPU creates zero-initialized buffers
        if (map_at_creation) {
            SOKOL_ASSERT(desc->data.ptr && (desc->data.size > 0));
            SOKOL_ASSERT(desc->data.size <= (size_t)buf->cmn.size);
            // FIXME: inefficient on WASM
            void* ptr = wgpuBufferGetMappedRange(buf->wgpu.buf, 0, wgpu_buf_size);
            SOKOL_ASSERT(ptr);
            memcpy(ptr, desc->data.ptr, desc->data.size);
            wgpuBufferUnmap(buf->wgpu.buf);
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf);
    if (buf->cmn.usage.storage_buffer) {
        _sg_wgpu_bindgroups_cache_invalidate(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_STORAGEBUFFER, buf->slot.id);
    }
    if (buf->wgpu.buf) {
        wgpuBufferRelease(buf->wgpu.buf);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_copy_buffer_data(const _sg_buffer_t* buf, uint64_t offset, const sg_range* data) {
    SOKOL_ASSERT((offset + data->size) <= (size_t)buf->cmn.size);
    // WebGPU's write-buffer requires the size to be a multiple of four, so we may need to split the copy
    // operation into two writeBuffer calls
    uint64_t clamped_size = data->size & ~3UL;
    uint64_t extra_size = data->size & 3UL;
    SOKOL_ASSERT(extra_size < 4);
    wgpuQueueWriteBuffer(_sg.wgpu.queue, buf->wgpu.buf, offset, data->ptr, clamped_size);
    if (extra_size > 0) {
        const uint64_t extra_src_offset = clamped_size;
        const uint64_t extra_dst_offset = offset + clamped_size;
        uint8_t extra_data[4] = { 0 };
        uint8_t* extra_src_ptr = ((uint8_t*)data->ptr) + extra_src_offset;
        for (size_t i = 0; i < extra_size; i++) {
            extra_data[i] = extra_src_ptr[i];
        }
        wgpuQueueWriteBuffer(_sg.wgpu.queue, buf->wgpu.buf, extra_dst_offset, extra_src_ptr, 4);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_copy_image_data(const _sg_image_t* img, WGPUTexture wgpu_tex, const sg_image_data* data) {
    WGPUTextureDataLayout wgpu_layout;
    _sg_clear(&wgpu_layout, sizeof(wgpu_layout));
    WGPUImageCopyTexture wgpu_copy_tex;
    _sg_clear(&wgpu_copy_tex, sizeof(wgpu_copy_tex));
    wgpu_copy_tex.texture = wgpu_tex;
    wgpu_copy_tex.aspect = WGPUTextureAspect_All;
    WGPUExtent3D wgpu_extent;
    _sg_clear(&wgpu_extent, sizeof(wgpu_extent));
    const int num_faces = (img->cmn.type == SG_IMAGETYPE_CUBE) ? 6 : 1;
    for (int face_index = 0; face_index < num_faces; face_index++) {
        for (int mip_index = 0; mip_index < img->cmn.num_mipmaps; mip_index++) {
            wgpu_copy_tex.mipLevel = (uint32_t)mip_index;
            wgpu_copy_tex.origin.z = (uint32_t)face_index;
            int mip_width = _sg_miplevel_dim(img->cmn.width, mip_index);
            int mip_height = _sg_miplevel_dim(img->cmn.height, mip_index);
            int mip_slices;
            switch (img->cmn.type) {
                case SG_IMAGETYPE_CUBE:
                    mip_slices = 1;
                    break;
                case SG_IMAGETYPE_3D:
                    mip_slices = _sg_miplevel_dim(img->cmn.num_slices, mip_index);
                    break;
                default:
                    mip_slices = img->cmn.num_slices;
                    break;
            }
            const int row_pitch = _sg_row_pitch(img->cmn.pixel_format, mip_width, 1);
            const int num_rows = _sg_num_rows(img->cmn.pixel_format, mip_height);
            if (_sg_is_compressed_pixel_format(img->cmn.pixel_format)) {
                mip_width = _sg_roundup(mip_width, 4);
                mip_height = _sg_roundup(mip_height, 4);
            }
            wgpu_layout.offset = 0;
            wgpu_layout.bytesPerRow = (uint32_t)row_pitch;
            wgpu_layout.rowsPerImage = (uint32_t)num_rows;
            wgpu_extent.width = (uint32_t)mip_width;
            wgpu_extent.height = (uint32_t)mip_height;
            wgpu_extent.depthOrArrayLayers = (uint32_t)mip_slices;
            const sg_range* mip_data = &data->subimage[face_index][mip_index];
            wgpuQueueWriteTexture(_sg.wgpu.queue, &wgpu_copy_tex, mip_data->ptr, mip_data->size, &wgpu_layout, &wgpu_extent);
        }
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && desc);
    const bool injected = (0 != desc->wgpu_texture);
    if (injected) {
        img->wgpu.tex = (WGPUTexture)desc->wgpu_texture;
        wgpuTextureReference(img->wgpu.tex);
        img->wgpu.view = (WGPUTextureView)desc->wgpu_texture_view;
        if (img->wgpu.view) {
            wgpuTextureViewReference(img->wgpu.view);
        }
    } else {
        WGPUTextureDescriptor wgpu_tex_desc;
        _sg_clear(&wgpu_tex_desc, sizeof(wgpu_tex_desc));
        wgpu_tex_desc.label = _sg_wgpu_stringview(desc->label);
        wgpu_tex_desc.usage = WGPUTextureUsage_TextureBinding|WGPUTextureUsage_CopyDst;
        if (desc->usage.render_attachment) {
            wgpu_tex_desc.usage |= WGPUTextureUsage_RenderAttachment;
        }
        if (desc->usage.storage_attachment) {
            wgpu_tex_desc.usage |= WGPUTextureUsage_StorageBinding;
        }
        wgpu_tex_desc.dimension = _sg_wgpu_texture_dimension(img->cmn.type);
        wgpu_tex_desc.size.width = (uint32_t) img->cmn.width;
        wgpu_tex_desc.size.height = (uint32_t) img->cmn.height;
        if (desc->type == SG_IMAGETYPE_CUBE) {
            wgpu_tex_desc.size.depthOrArrayLayers = 6;
        } else {
            wgpu_tex_desc.size.depthOrArrayLayers = (uint32_t) img->cmn.num_slices;
        }
        wgpu_tex_desc.format = _sg_wgpu_textureformat(img->cmn.pixel_format);
        wgpu_tex_desc.mipLevelCount = (uint32_t) img->cmn.num_mipmaps;
        wgpu_tex_desc.sampleCount = (uint32_t) img->cmn.sample_count;
        img->wgpu.tex = wgpuDeviceCreateTexture(_sg.wgpu.dev, &wgpu_tex_desc);
        if (0 == img->wgpu.tex) {
            _SG_ERROR(WGPU_CREATE_TEXTURE_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
        if (desc->data.subimage[0][0].ptr) {
            _sg_wgpu_copy_image_data(img, img->wgpu.tex, &desc->data);
        }
        WGPUTextureViewDescriptor wgpu_texview_desc;
        _sg_clear(&wgpu_texview_desc, sizeof(wgpu_texview_desc));
        wgpu_texview_desc.label = _sg_wgpu_stringview(desc->label);
        wgpu_texview_desc.dimension = _sg_wgpu_texture_view_dimension(img->cmn.type);
        wgpu_texview_desc.mipLevelCount = (uint32_t)img->cmn.num_mipmaps;
        if (img->cmn.type == SG_IMAGETYPE_CUBE) {
            wgpu_texview_desc.arrayLayerCount = 6;
        } else if (img->cmn.type == SG_IMAGETYPE_ARRAY) {
            wgpu_texview_desc.arrayLayerCount = (uint32_t)img->cmn.num_slices;
        } else {
            wgpu_texview_desc.arrayLayerCount = 1;
        }
        if (_sg_is_depth_or_depth_stencil_format(img->cmn.pixel_format)) {
            wgpu_texview_desc.aspect = WGPUTextureAspect_DepthOnly;
        } else {
            wgpu_texview_desc.aspect = WGPUTextureAspect_All;
        }
        img->wgpu.view = wgpuTextureCreateView(img->wgpu.tex, &wgpu_texview_desc);
        if (0 == img->wgpu.view) {
            _SG_ERROR(WGPU_CREATE_TEXTURE_VIEW_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_image(_sg_image_t* img) {
    SOKOL_ASSERT(img);
    _sg_wgpu_bindgroups_cache_invalidate(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_IMAGE, img->slot.id);
    if (img->wgpu.view) {
        wgpuTextureViewRelease(img->wgpu.view);
        img->wgpu.view = 0;
    }
    if (img->wgpu.tex) {
        wgpuTextureRelease(img->wgpu.tex);
        img->wgpu.tex = 0;
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && desc);
    SOKOL_ASSERT(_sg.wgpu.dev);
    const bool injected = (0 != desc->wgpu_sampler);
    if (injected) {
        smp->wgpu.smp = (WGPUSampler) desc->wgpu_sampler;
        wgpuSamplerReference(smp->wgpu.smp);
    } else {
        WGPUSamplerDescriptor wgpu_desc;
        _sg_clear(&wgpu_desc, sizeof(wgpu_desc));
        wgpu_desc.label = _sg_wgpu_stringview(desc->label);
        wgpu_desc.addressModeU = _sg_wgpu_sampler_address_mode(desc->wrap_u);
        wgpu_desc.addressModeV = _sg_wgpu_sampler_address_mode(desc->wrap_v);
        wgpu_desc.addressModeW = _sg_wgpu_sampler_address_mode(desc->wrap_w);
        wgpu_desc.magFilter = _sg_wgpu_sampler_minmag_filter(desc->mag_filter);
        wgpu_desc.minFilter = _sg_wgpu_sampler_minmag_filter(desc->min_filter);
        wgpu_desc.mipmapFilter = _sg_wgpu_sampler_mipmap_filter(desc->mipmap_filter);
        wgpu_desc.lodMinClamp = desc->min_lod;
        wgpu_desc.lodMaxClamp = desc->max_lod;
        wgpu_desc.compare = _sg_wgpu_comparefunc(desc->compare);
        if (wgpu_desc.compare == WGPUCompareFunction_Never) {
            wgpu_desc.compare = WGPUCompareFunction_Undefined;
        }
        wgpu_desc.maxAnisotropy = (uint16_t)desc->max_anisotropy;
        smp->wgpu.smp = wgpuDeviceCreateSampler(_sg.wgpu.dev, &wgpu_desc);
        if (0 == smp->wgpu.smp) {
            _SG_ERROR(WGPU_CREATE_SAMPLER_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp);
    _sg_wgpu_bindgroups_cache_invalidate(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_SAMPLER, smp->slot.id);
    if (smp->wgpu.smp) {
        wgpuSamplerRelease(smp->wgpu.smp);
        smp->wgpu.smp = 0;
    }
}

_SOKOL_PRIVATE _sg_wgpu_shader_func_t _sg_wgpu_create_shader_func(const sg_shader_function* func, const char* label) {
    SOKOL_ASSERT(func);
    SOKOL_ASSERT(func->source);
    SOKOL_ASSERT(func->entry);

    _sg_wgpu_shader_func_t res;
    _sg_clear(&res, sizeof(res));
    _sg_strcpy(&res.entry, func->entry);

    WGPUShaderModuleWGSLDescriptor wgpu_shdmod_wgsl_desc;
    _sg_clear(&wgpu_shdmod_wgsl_desc, sizeof(wgpu_shdmod_wgsl_desc));
    wgpu_shdmod_wgsl_desc.chain.sType = WGPUSType_ShaderModuleWGSLDescriptor;
    wgpu_shdmod_wgsl_desc.code = _sg_wgpu_stringview(func->source);

    WGPUShaderModuleDescriptor wgpu_shdmod_desc;
    _sg_clear(&wgpu_shdmod_desc, sizeof(wgpu_shdmod_desc));
    wgpu_shdmod_desc.nextInChain = &wgpu_shdmod_wgsl_desc.chain;
    wgpu_shdmod_desc.label = _sg_wgpu_stringview(label);

    // NOTE: if compilation fails we won't actually find out in this call since
    // it always returns a valid module handle, and the GetCompilationInfo() call
    // is asynchronous
    res.module = wgpuDeviceCreateShaderModule(_sg.wgpu.dev, &wgpu_shdmod_desc);
    if (0 == res.module) {
        _SG_ERROR(WGPU_CREATE_SHADER_MODULE_FAILED);
    }
    return res;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_shader_func(_sg_wgpu_shader_func_t* func) {
    if (func->module) {
        wgpuShaderModuleRelease(func->module);
        func->module = 0;
    }
}

typedef struct { uint8_t sokol_slot, wgpu_slot; } _sg_wgpu_dynoffset_mapping_t;

_SOKOL_PRIVATE int _sg_wgpu_dynoffset_cmp(const void* a, const void* b) {
    const _sg_wgpu_dynoffset_mapping_t* aa = (const _sg_wgpu_dynoffset_mapping_t*)a;
    const _sg_wgpu_dynoffset_mapping_t* bb = (const _sg_wgpu_dynoffset_mapping_t*)b;
    if (aa->wgpu_slot < bb->wgpu_slot) return -1;
    else if (aa->wgpu_slot > bb->wgpu_slot) return 1;
    return 0;
}

// NOTE: this is an out-of-range check for WGSL bindslots that's also active in release mode
_SOKOL_PRIVATE bool _sg_wgpu_ensure_wgsl_bindslot_ranges(const sg_shader_desc* desc) {
    SOKOL_ASSERT(desc);
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (desc->uniform_blocks[i].wgsl_group0_binding_n >= _SG_WGPU_MAX_UB_BINDGROUP_BIND_SLOTS) {
            _SG_ERROR(WGPU_UNIFORMBLOCK_WGSL_GROUP0_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (desc->storage_buffers[i].wgsl_group1_binding_n >= _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS) {
            _SG_ERROR(WGPU_STORAGEBUFFER_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (desc->images[i].wgsl_group1_binding_n >= _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS) {
            _SG_ERROR(WGPU_IMAGE_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (desc->samplers[i].wgsl_group1_binding_n >= _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS) {
            _SG_ERROR(WGPU_SAMPLER_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            return false;
        }
    }
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (desc->storage_images[i].wgsl_group2_binding_n >= _SG_WGPU_MAX_SIMG_BIND_SLOTS) {
            _SG_ERROR(WGPU_STORAGEIMAGE_WGSL_GROUP2_BINDING_OUT_OF_RANGE);
        }
    }
    return true;
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && desc);
    SOKOL_ASSERT(shd->wgpu.vertex_func.module == 0);
    SOKOL_ASSERT(shd->wgpu.fragment_func.module == 0);
    SOKOL_ASSERT(shd->wgpu.compute_func.module == 0);
    SOKOL_ASSERT(shd->wgpu.bgl_ub == 0);
    SOKOL_ASSERT(shd->wgpu.bg_ub == 0);
    SOKOL_ASSERT(shd->wgpu.bgl_img_smp_sbuf == 0);

    // do a release-mode bounds-check on wgsl bindslots, even though out-of-range
    // bindslots can't cause out-of-bounds accesses in the wgpu backend, this
    // is done to be consistent with the other backends
    if (!_sg_wgpu_ensure_wgsl_bindslot_ranges(desc)) {
        return SG_RESOURCESTATE_FAILED;
    }

    // build shader modules
    bool shd_valid = true;
    if (desc->vertex_func.source) {
        shd->wgpu.vertex_func = _sg_wgpu_create_shader_func(&desc->vertex_func, desc->label);
        shd_valid &= shd->wgpu.vertex_func.module != 0;
    }
    if (desc->fragment_func.source) {
        shd->wgpu.fragment_func = _sg_wgpu_create_shader_func(&desc->fragment_func, desc->label);
        shd_valid &= shd->wgpu.fragment_func.module != 0;
    }
    if (desc->compute_func.source) {
        shd->wgpu.compute_func = _sg_wgpu_create_shader_func(&desc->compute_func, desc->label);
        shd_valid &= shd->wgpu.compute_func.module != 0;
    }
    if (!shd_valid) {
        _sg_wgpu_discard_shader_func(&shd->wgpu.vertex_func);
        _sg_wgpu_discard_shader_func(&shd->wgpu.fragment_func);
        _sg_wgpu_discard_shader_func(&shd->wgpu.compute_func);
        return SG_RESOURCESTATE_FAILED;
    }

    // create bind group layout and bind group for uniform blocks
    // NOTE also need to create a mapping of sokol ub bind slots to array indices
    // for the dynamic offsets array in the setBindGroup call
    SOKOL_ASSERT(_SG_WGPU_MAX_UB_BINDGROUP_ENTRIES <= _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES);
    SOKOL_ASSERT(_SG_WGPU_MAX_SIMG_BINDGROUP_ENTRIES <= _SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES);
    WGPUBindGroupLayoutEntry bgl_entries[_SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES];
    _sg_clear(bgl_entries, sizeof(bgl_entries));
    WGPUBindGroupLayoutDescriptor bgl_desc;
    _sg_clear(&bgl_desc, sizeof(bgl_desc));
    WGPUBindGroupEntry bg_entries[_SG_WGPU_MAX_IMG_SMP_SBUF_BINDGROUP_ENTRIES];
    _sg_clear(&bg_entries, sizeof(bg_entries));
    WGPUBindGroupDescriptor bg_desc;
    _sg_clear(&bg_desc, sizeof(bg_desc));
    _sg_wgpu_dynoffset_mapping_t dynoffset_map[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
    _sg_clear(dynoffset_map, sizeof(dynoffset_map));
    size_t bgl_index = 0;
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (shd->cmn.uniform_blocks[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        shd->wgpu.ub_grp0_bnd_n[i] = desc->uniform_blocks[i].wgsl_group0_binding_n;
        WGPUBindGroupEntry* bg_entry = &bg_entries[bgl_index];
        WGPUBindGroupLayoutEntry* bgl_entry = &bgl_entries[bgl_index];
        bgl_entry->binding = shd->wgpu.ub_grp0_bnd_n[i];
        bgl_entry->visibility = _sg_wgpu_shader_stage(shd->cmn.uniform_blocks[i].stage);
        bgl_entry->buffer.type = WGPUBufferBindingType_Uniform;
        bgl_entry->buffer.hasDynamicOffset = true;
        bg_entry->binding = bgl_entry->binding;
        bg_entry->buffer = _sg.wgpu.uniform.buf;
        bg_entry->size = _SG_WGPU_MAX_UNIFORM_UPDATE_SIZE;
        dynoffset_map[i].sokol_slot = i;
        dynoffset_map[i].wgpu_slot = bgl_entry->binding;
        bgl_index += 1;
    }
    bgl_desc.entryCount = bgl_index;
    bgl_desc.entries = bgl_entries;
    shd->wgpu.bgl_ub = wgpuDeviceCreateBindGroupLayout(_sg.wgpu.dev, &bgl_desc);
    SOKOL_ASSERT(shd->wgpu.bgl_ub);
    bg_desc.layout = shd->wgpu.bgl_ub;
    bg_desc.entryCount = bgl_index;
    bg_desc.entries = bg_entries;
    shd->wgpu.bg_ub = wgpuDeviceCreateBindGroup(_sg.wgpu.dev, &bg_desc);
    SOKOL_ASSERT(shd->wgpu.bg_ub);

    // sort the dynoffset_map by wgpu bindings, this is because the
    // dynamic offsets of the WebGPU setBindGroup call must be in
    // 'binding order', not 'bindgroup entry order'
    qsort(dynoffset_map, bgl_index, sizeof(_sg_wgpu_dynoffset_mapping_t), _sg_wgpu_dynoffset_cmp);
    shd->wgpu.ub_num_dynoffsets = bgl_index;
    for (uint8_t i = 0; i < bgl_index; i++) {
        const uint8_t sokol_slot = dynoffset_map[i].sokol_slot;
        shd->wgpu.ub_dynoffsets[sokol_slot] = i;
    }

    // create bind group layout for images, samplers and storage buffers
    _sg_clear(bgl_entries, sizeof(bgl_entries));
    _sg_clear(&bgl_desc, sizeof(bgl_desc));
    bgl_index = 0;
    for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (shd->cmn.images[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        const bool msaa = shd->cmn.images[i].multisampled;
        shd->wgpu.img_grp1_bnd_n[i] = desc->images[i].wgsl_group1_binding_n;
        WGPUBindGroupLayoutEntry* bgl_entry = &bgl_entries[bgl_index];
        bgl_entry->binding = shd->wgpu.img_grp1_bnd_n[i];
        bgl_entry->visibility = _sg_wgpu_shader_stage(shd->cmn.images[i].stage);
        bgl_entry->texture.viewDimension = _sg_wgpu_texture_view_dimension(shd->cmn.images[i].image_type);
        bgl_entry->texture.sampleType = _sg_wgpu_texture_sample_type(shd->cmn.images[i].sample_type, msaa);
        bgl_entry->texture.multisampled = msaa;
        bgl_index += 1;
    }
    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (shd->cmn.samplers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        shd->wgpu.smp_grp1_bnd_n[i] = desc->samplers[i].wgsl_group1_binding_n;
        WGPUBindGroupLayoutEntry* bgl_entry = &bgl_entries[bgl_index];
        bgl_entry->binding = shd->wgpu.smp_grp1_bnd_n[i];
        bgl_entry->visibility = _sg_wgpu_shader_stage(shd->cmn.samplers[i].stage);
        bgl_entry->sampler.type = _sg_wgpu_sampler_binding_type(shd->cmn.samplers[i].sampler_type);
        bgl_index += 1;
    }
    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (shd->cmn.storage_buffers[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        shd->wgpu.sbuf_grp1_bnd_n[i] = desc->storage_buffers[i].wgsl_group1_binding_n;
        WGPUBindGroupLayoutEntry* bgl_entry = &bgl_entries[bgl_index];
        bgl_entry->binding = shd->wgpu.sbuf_grp1_bnd_n[i];
        bgl_entry->visibility = _sg_wgpu_shader_stage(shd->cmn.storage_buffers[i].stage);
        if (shd->cmn.storage_buffers[i].readonly) {
            bgl_entry->buffer.type = WGPUBufferBindingType_ReadOnlyStorage;
        } else {
            bgl_entry->buffer.type = WGPUBufferBindingType_Storage;
        }
        bgl_index += 1;
    }
    bgl_desc.entryCount = bgl_index;
    bgl_desc.entries = bgl_entries;
    shd->wgpu.bgl_img_smp_sbuf = wgpuDeviceCreateBindGroupLayout(_sg.wgpu.dev, &bgl_desc);
    if (shd->wgpu.bgl_img_smp_sbuf == 0) {
        _SG_ERROR(WGPU_SHADER_CREATE_BINDGROUP_LAYOUT_FAILED);
        return SG_RESOURCESTATE_FAILED;
    }

    // create optional bindgroup layout for storage images (separate bindgroup because
    // those are not applied in sg_apply_bindings() but are defined as pass attachments)
    _sg_clear(bgl_entries, sizeof(bgl_entries));
    _sg_clear(&bgl_desc, sizeof(bgl_desc));
    bgl_index = 0;
    for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (shd->cmn.storage_images[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        shd->wgpu.simg_grp2_bnd_n[i] = desc->storage_images[i].wgsl_group2_binding_n;
        WGPUBindGroupLayoutEntry* bgl_entry = &bgl_entries[bgl_index];
        bgl_entry->binding = shd->wgpu.simg_grp2_bnd_n[i];
        bgl_entry->visibility = _sg_wgpu_shader_stage(shd->cmn.storage_images[i].stage);
        if (shd->cmn.storage_images[i].writeonly) {
            bgl_entry->storageTexture.access = WGPUStorageTextureAccess_WriteOnly;
        } else {
            bgl_entry->storageTexture.access = WGPUStorageTextureAccess_ReadWrite;
        }
        bgl_entry->storageTexture.format = _sg_wgpu_textureformat(desc->storage_images[i].access_format);
        bgl_entry->texture.viewDimension = _sg_wgpu_texture_view_dimension(shd->cmn.storage_images[i].image_type);
        bgl_index += 1;
    }
    if (bgl_index > 0) {
        bgl_desc.entryCount = bgl_index;
        bgl_desc.entries = bgl_entries;
        shd->wgpu.bgl_simg = wgpuDeviceCreateBindGroupLayout(_sg.wgpu.dev, &bgl_desc);
        if (shd->wgpu.bgl_simg == 0) {
            _SG_ERROR(WGPU_SHADER_CREATE_BINDGROUP_LAYOUT_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd);
    _sg_wgpu_discard_shader_func(&shd->wgpu.vertex_func);
    _sg_wgpu_discard_shader_func(&shd->wgpu.fragment_func);
    _sg_wgpu_discard_shader_func(&shd->wgpu.compute_func);
    if (shd->wgpu.bgl_ub) {
        wgpuBindGroupLayoutRelease(shd->wgpu.bgl_ub);
        shd->wgpu.bgl_ub = 0;
    }
    if (shd->wgpu.bg_ub) {
        wgpuBindGroupRelease(shd->wgpu.bg_ub);
        shd->wgpu.bg_ub = 0;
    }
    if (shd->wgpu.bgl_img_smp_sbuf) {
        wgpuBindGroupLayoutRelease(shd->wgpu.bgl_img_smp_sbuf);
        shd->wgpu.bgl_img_smp_sbuf = 0;
    }
    if (shd->wgpu.bgl_simg) {
        wgpuBindGroupLayoutRelease(shd->wgpu.bgl_simg);
        shd->wgpu.bgl_simg = 0;
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_pipeline(_sg_pipeline_t* pip, _sg_shader_t* shd, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && shd && desc);
    SOKOL_ASSERT(desc->shader.id == shd->slot.id);
    SOKOL_ASSERT(shd->wgpu.bgl_ub);
    SOKOL_ASSERT(shd->wgpu.bgl_img_smp_sbuf);
    pip->shader = shd;

    pip->wgpu.blend_color.r = (double) desc->blend_color.r;
    pip->wgpu.blend_color.g = (double) desc->blend_color.g;
    pip->wgpu.blend_color.b = (double) desc->blend_color.b;
    pip->wgpu.blend_color.a = (double) desc->blend_color.a;

    // - @group(0) for uniform blocks
    // - @group(1) for all image, sampler and storagebuffer resources
    // - @group(2) optional: storage image attachments in compute passes
    size_t num_bgls = 2;
    WGPUBindGroupLayout wgpu_bgl[_SG_WGPU_MAX_BINDGROUPS];
    _sg_clear(&wgpu_bgl, sizeof(wgpu_bgl));
    wgpu_bgl[_SG_WGPU_UB_BINDGROUP_INDEX ] = shd->wgpu.bgl_ub;
    wgpu_bgl[_SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX] = shd->wgpu.bgl_img_smp_sbuf;
    if (shd->wgpu.bgl_simg) {
        wgpu_bgl[_SG_WGPU_SIMG_BINDGROUP_INDEX] = shd->wgpu.bgl_simg;
        num_bgls += 1;
    }
    WGPUPipelineLayoutDescriptor wgpu_pl_desc;
    _sg_clear(&wgpu_pl_desc, sizeof(wgpu_pl_desc));
    wgpu_pl_desc.bindGroupLayoutCount = num_bgls;
    wgpu_pl_desc.bindGroupLayouts = &wgpu_bgl[0];
    const WGPUPipelineLayout wgpu_pip_layout = wgpuDeviceCreatePipelineLayout(_sg.wgpu.dev, &wgpu_pl_desc);
    if (0 == wgpu_pip_layout) {
        _SG_ERROR(WGPU_CREATE_PIPELINE_LAYOUT_FAILED);
        return SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT(wgpu_pip_layout);

    if (pip->cmn.is_compute) {
        WGPUComputePipelineDescriptor wgpu_pip_desc;
        _sg_clear(&wgpu_pip_desc, sizeof(wgpu_pip_desc));
        wgpu_pip_desc.label = _sg_wgpu_stringview(desc->label);
        wgpu_pip_desc.layout = wgpu_pip_layout;
        wgpu_pip_desc.compute.module = shd->wgpu.compute_func.module;
        wgpu_pip_desc.compute.entryPoint = shd->wgpu.compute_func.entry.buf;
        pip->wgpu.cpip = wgpuDeviceCreateComputePipeline(_sg.wgpu.dev, &wgpu_pip_desc);
        wgpuPipelineLayoutRelease(wgpu_pip_layout);
        if (0 == pip->wgpu.cpip) {
            _SG_ERROR(WGPU_CREATE_COMPUTE_PIPELINE_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    } else {
        WGPUVertexBufferLayout wgpu_vb_layouts[SG_MAX_VERTEXBUFFER_BINDSLOTS];
        _sg_clear(wgpu_vb_layouts, sizeof(wgpu_vb_layouts));
        WGPUVertexAttribute wgpu_vtx_attrs[SG_MAX_VERTEXBUFFER_BINDSLOTS][SG_MAX_VERTEX_ATTRIBUTES];
        _sg_clear(wgpu_vtx_attrs, sizeof(wgpu_vtx_attrs));
        int wgpu_vb_num = 0;
        for (int vb_idx = 0; vb_idx < SG_MAX_VERTEXBUFFER_BINDSLOTS; vb_idx++, wgpu_vb_num++) {
            const sg_vertex_buffer_layout_state* vbl_state = &desc->layout.buffers[vb_idx];
            if (0 == vbl_state->stride) {
                break;
            }
            wgpu_vb_layouts[vb_idx].arrayStride = (uint64_t)vbl_state->stride;
            wgpu_vb_layouts[vb_idx].stepMode = _sg_wgpu_stepmode(vbl_state->step_func);
            wgpu_vb_layouts[vb_idx].attributes = &wgpu_vtx_attrs[vb_idx][0];
        }
        for (int va_idx = 0; va_idx < SG_MAX_VERTEX_ATTRIBUTES; va_idx++) {
            const sg_vertex_attr_state* va_state = &desc->layout.attrs[va_idx];
            if (SG_VERTEXFORMAT_INVALID == va_state->format) {
                break;
            }
            const int vb_idx = va_state->buffer_index;
            SOKOL_ASSERT(vb_idx < SG_MAX_VERTEXBUFFER_BINDSLOTS);
            SOKOL_ASSERT(pip->cmn.vertex_buffer_layout_active[vb_idx]);
            const size_t wgpu_attr_idx = wgpu_vb_layouts[vb_idx].attributeCount;
            wgpu_vb_layouts[vb_idx].attributeCount += 1;
            wgpu_vtx_attrs[vb_idx][wgpu_attr_idx].format = _sg_wgpu_vertexformat(va_state->format);
            wgpu_vtx_attrs[vb_idx][wgpu_attr_idx].offset = (uint64_t)va_state->offset;
            wgpu_vtx_attrs[vb_idx][wgpu_attr_idx].shaderLocation = (uint32_t)va_idx;
        }

        WGPURenderPipelineDescriptor wgpu_pip_desc;
        _sg_clear(&wgpu_pip_desc, sizeof(wgpu_pip_desc));
        WGPUDepthStencilState wgpu_ds_state;
        _sg_clear(&wgpu_ds_state, sizeof(wgpu_ds_state));
        WGPUFragmentState wgpu_frag_state;
        _sg_clear(&wgpu_frag_state, sizeof(wgpu_frag_state));
        WGPUColorTargetState wgpu_ctgt_state[SG_MAX_COLOR_ATTACHMENTS];
        _sg_clear(&wgpu_ctgt_state, sizeof(wgpu_ctgt_state));
        WGPUBlendState wgpu_blend_state[SG_MAX_COLOR_ATTACHMENTS];
        _sg_clear(&wgpu_blend_state, sizeof(wgpu_blend_state));
        wgpu_pip_desc.label = _sg_wgpu_stringview(desc->label);
        wgpu_pip_desc.layout = wgpu_pip_layout;
        wgpu_pip_desc.vertex.module = shd->wgpu.vertex_func.module;
        wgpu_pip_desc.vertex.entryPoint = shd->wgpu.vertex_func.entry.buf;
        wgpu_pip_desc.vertex.bufferCount = (size_t)wgpu_vb_num;
        wgpu_pip_desc.vertex.buffers = &wgpu_vb_layouts[0];
        wgpu_pip_desc.primitive.topology = _sg_wgpu_topology(desc->primitive_type);
        wgpu_pip_desc.primitive.stripIndexFormat = _sg_wgpu_stripindexformat(desc->primitive_type, desc->index_type);
        wgpu_pip_desc.primitive.frontFace = _sg_wgpu_frontface(desc->face_winding);
        wgpu_pip_desc.primitive.cullMode = _sg_wgpu_cullmode(desc->cull_mode);
        if (SG_PIXELFORMAT_NONE != desc->depth.pixel_format) {
            wgpu_ds_state.format = _sg_wgpu_textureformat(desc->depth.pixel_format);
            wgpu_ds_state.depthWriteEnabled = _sg_wgpu_optional_bool(desc->depth.write_enabled);
            wgpu_ds_state.depthCompare = _sg_wgpu_comparefunc(desc->depth.compare);
            wgpu_ds_state.stencilFront.compare = _sg_wgpu_comparefunc(desc->stencil.front.compare);
            wgpu_ds_state.stencilFront.failOp = _sg_wgpu_stencilop(desc->stencil.front.fail_op);
            wgpu_ds_state.stencilFront.depthFailOp = _sg_wgpu_stencilop(desc->stencil.front.depth_fail_op);
            wgpu_ds_state.stencilFront.passOp = _sg_wgpu_stencilop(desc->stencil.front.pass_op);
            wgpu_ds_state.stencilBack.compare = _sg_wgpu_comparefunc(desc->stencil.back.compare);
            wgpu_ds_state.stencilBack.failOp = _sg_wgpu_stencilop(desc->stencil.back.fail_op);
            wgpu_ds_state.stencilBack.depthFailOp = _sg_wgpu_stencilop(desc->stencil.back.depth_fail_op);
            wgpu_ds_state.stencilBack.passOp = _sg_wgpu_stencilop(desc->stencil.back.pass_op);
            wgpu_ds_state.stencilReadMask = desc->stencil.read_mask;
            wgpu_ds_state.stencilWriteMask = desc->stencil.write_mask;
            wgpu_ds_state.depthBias = (int32_t)desc->depth.bias;
            wgpu_ds_state.depthBiasSlopeScale = desc->depth.bias_slope_scale;
            wgpu_ds_state.depthBiasClamp = desc->depth.bias_clamp;
            wgpu_pip_desc.depthStencil = &wgpu_ds_state;
        }
        wgpu_pip_desc.multisample.count = (uint32_t)desc->sample_count;
        wgpu_pip_desc.multisample.mask = 0xFFFFFFFF;
        wgpu_pip_desc.multisample.alphaToCoverageEnabled = desc->alpha_to_coverage_enabled;
        if (desc->color_count > 0) {
            wgpu_frag_state.module = shd->wgpu.fragment_func.module;
            wgpu_frag_state.entryPoint = shd->wgpu.fragment_func.entry.buf;
            wgpu_frag_state.targetCount = (size_t)desc->color_count;
            wgpu_frag_state.targets = &wgpu_ctgt_state[0];
            for (int i = 0; i < desc->color_count; i++) {
                SOKOL_ASSERT(i < SG_MAX_COLOR_ATTACHMENTS);
                wgpu_ctgt_state[i].format = _sg_wgpu_textureformat(desc->colors[i].pixel_format);
                wgpu_ctgt_state[i].writeMask = _sg_wgpu_colorwritemask(desc->colors[i].write_mask);
                if (desc->colors[i].blend.enabled) {
                    wgpu_ctgt_state[i].blend = &wgpu_blend_state[i];
                    wgpu_blend_state[i].color.operation = _sg_wgpu_blendop(desc->colors[i].blend.op_rgb);
                    wgpu_blend_state[i].color.srcFactor = _sg_wgpu_blendfactor(desc->colors[i].blend.src_factor_rgb);
                    wgpu_blend_state[i].color.dstFactor = _sg_wgpu_blendfactor(desc->colors[i].blend.dst_factor_rgb);
                    wgpu_blend_state[i].alpha.operation = _sg_wgpu_blendop(desc->colors[i].blend.op_alpha);
                    wgpu_blend_state[i].alpha.srcFactor = _sg_wgpu_blendfactor(desc->colors[i].blend.src_factor_alpha);
                    wgpu_blend_state[i].alpha.dstFactor = _sg_wgpu_blendfactor(desc->colors[i].blend.dst_factor_alpha);
                }
            }
            wgpu_pip_desc.fragment = &wgpu_frag_state;
        }
        pip->wgpu.rpip = wgpuDeviceCreateRenderPipeline(_sg.wgpu.dev, &wgpu_pip_desc);
        wgpuPipelineLayoutRelease(wgpu_pip_layout);
        if (0 == pip->wgpu.rpip) {
            _SG_ERROR(WGPU_CREATE_RENDER_PIPELINE_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    _sg_wgpu_bindgroups_cache_invalidate(_SG_WGPU_BINDGROUPSCACHEITEMTYPE_PIPELINE, pip->slot.id);
    if (pip == _sg.wgpu.cur_pipeline) {
        _sg.wgpu.cur_pipeline = 0;
        _sg.wgpu.cur_pipeline_id.id = SG_INVALID_ID;
    }
    if (pip->wgpu.rpip) {
        wgpuRenderPipelineRelease(pip->wgpu.rpip);
        pip->wgpu.rpip = 0;
    }
    if (pip->wgpu.cpip) {
        wgpuComputePipelineRelease(pip->wgpu.cpip);
        pip->wgpu.cpip = 0;
    }
}

_SOKOL_PRIVATE sg_resource_state _sg_wgpu_create_attachments(_sg_attachments_t* atts, const _sg_attachments_ptrs_t* atts_ptrs, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && atts_ptrs && desc);

    // copy image pointers and create renderable wgpu texture views
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        const sg_attachment_desc* color_desc = &desc->colors[i];
        _SOKOL_UNUSED(color_desc);
        SOKOL_ASSERT(color_desc->image.id != SG_INVALID_ID);
        SOKOL_ASSERT(0 == atts->wgpu.colors[i].image);
        SOKOL_ASSERT(atts_ptrs->color_images[i]);
        _sg_image_t* clr_img = atts_ptrs->color_images[i];
        SOKOL_ASSERT(clr_img->slot.id == color_desc->image.id);
        SOKOL_ASSERT(_sg_is_valid_attachment_color_format(clr_img->cmn.pixel_format));
        SOKOL_ASSERT(clr_img->wgpu.tex);
        atts->wgpu.colors[i].image = clr_img;

        WGPUTextureViewDescriptor wgpu_color_view_desc;
        _sg_clear(&wgpu_color_view_desc, sizeof(wgpu_color_view_desc));
        wgpu_color_view_desc.baseMipLevel = (uint32_t) color_desc->mip_level;
        wgpu_color_view_desc.mipLevelCount = 1;
        wgpu_color_view_desc.baseArrayLayer = (uint32_t) color_desc->slice;
        wgpu_color_view_desc.arrayLayerCount = 1;
        atts->wgpu.colors[i].view = wgpuTextureCreateView(clr_img->wgpu.tex, &wgpu_color_view_desc);
        if (0 == atts->wgpu.colors[i].view) {
            _SG_ERROR(WGPU_ATTACHMENTS_CREATE_TEXTURE_VIEW_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }

        const sg_attachment_desc* resolve_desc = &desc->resolves[i];
        if (resolve_desc->image.id != SG_INVALID_ID) {
            SOKOL_ASSERT(0 == atts->wgpu.resolves[i].image);
            SOKOL_ASSERT(atts_ptrs->resolve_images[i]);
            _sg_image_t* rsv_img = atts_ptrs->resolve_images[i];
            SOKOL_ASSERT(rsv_img->slot.id == resolve_desc->image.id);
            SOKOL_ASSERT(clr_img->cmn.pixel_format == rsv_img->cmn.pixel_format);
            SOKOL_ASSERT(rsv_img->wgpu.tex);
            atts->wgpu.resolves[i].image = rsv_img;

            WGPUTextureViewDescriptor wgpu_resolve_view_desc;
            _sg_clear(&wgpu_resolve_view_desc, sizeof(wgpu_resolve_view_desc));
            wgpu_resolve_view_desc.baseMipLevel = (uint32_t) resolve_desc->mip_level;
            wgpu_resolve_view_desc.mipLevelCount = 1;
            wgpu_resolve_view_desc.baseArrayLayer = (uint32_t) resolve_desc->slice;
            wgpu_resolve_view_desc.arrayLayerCount = 1;
            atts->wgpu.resolves[i].view = wgpuTextureCreateView(rsv_img->wgpu.tex, &wgpu_resolve_view_desc);
            if (0 == atts->wgpu.resolves[i].view) {
                _SG_ERROR(WGPU_ATTACHMENTS_CREATE_TEXTURE_VIEW_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
        }
    }
    SOKOL_ASSERT(0 == atts->wgpu.depth_stencil.image);
    const sg_attachment_desc* ds_desc = &desc->depth_stencil;
    if (ds_desc->image.id != SG_INVALID_ID) {
        SOKOL_ASSERT(atts_ptrs->ds_image);
        _sg_image_t* ds_img =atts_ptrs->ds_image;
        SOKOL_ASSERT(ds_img->slot.id == ds_desc->image.id);
        SOKOL_ASSERT(_sg_is_valid_attachment_depth_format(ds_img->cmn.pixel_format));
        SOKOL_ASSERT(ds_img->wgpu.tex);
        atts->wgpu.depth_stencil.image = ds_img;

        WGPUTextureViewDescriptor wgpu_ds_view_desc;
        _sg_clear(&wgpu_ds_view_desc, sizeof(wgpu_ds_view_desc));
        wgpu_ds_view_desc.baseMipLevel = (uint32_t) ds_desc->mip_level;
        wgpu_ds_view_desc.mipLevelCount = 1;
        wgpu_ds_view_desc.baseArrayLayer = (uint32_t) ds_desc->slice;
        wgpu_ds_view_desc.arrayLayerCount = 1;
        atts->wgpu.depth_stencil.view = wgpuTextureCreateView(ds_img->wgpu.tex, &wgpu_ds_view_desc);
        if (0 == atts->wgpu.depth_stencil.view) {
            _SG_ERROR(WGPU_ATTACHMENTS_CREATE_TEXTURE_VIEW_FAILED);
            return SG_RESOURCESTATE_FAILED;
        }
    }
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        const sg_attachment_desc* storage_desc = &desc->storages[i];
        if (storage_desc->image.id != SG_INVALID_ID) {
            SOKOL_ASSERT(0 == atts->wgpu.storages[i].image);
            SOKOL_ASSERT(atts_ptrs->storage_images[i]);
            _sg_image_t* stg_img = atts_ptrs->storage_images[i];
            SOKOL_ASSERT(stg_img->slot.id == storage_desc->image.id);
            SOKOL_ASSERT(_sg_is_valid_attachment_storage_format(stg_img->cmn.pixel_format));
            atts->wgpu.storages[i].image = stg_img;

            WGPUTextureViewDescriptor wgpu_storage_view_desc;
            _sg_clear(&wgpu_storage_view_desc, sizeof(wgpu_storage_view_desc));
            wgpu_storage_view_desc.baseMipLevel = (uint32_t) storage_desc->mip_level;
            wgpu_storage_view_desc.mipLevelCount = 1;
            wgpu_storage_view_desc.baseArrayLayer = (uint32_t) storage_desc->slice;
            wgpu_storage_view_desc.arrayLayerCount = 1;
            if (_sg_is_depth_or_depth_stencil_format(stg_img->cmn.pixel_format)) {
                wgpu_storage_view_desc.aspect = WGPUTextureAspect_DepthOnly;
            } else {
                wgpu_storage_view_desc.aspect = WGPUTextureAspect_All;
            }
            atts->wgpu.storages[i].view = wgpuTextureCreateView(stg_img->wgpu.tex, &wgpu_storage_view_desc);
            if (0 == atts->wgpu.storages[i].view) {
                _SG_ERROR(WGPU_ATTACHMENTS_CREATE_TEXTURE_VIEW_FAILED);
                return SG_RESOURCESTATE_FAILED;
            }
        }
    }
    return SG_RESOURCESTATE_VALID;
}

_SOKOL_PRIVATE void _sg_wgpu_discard_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts);
    for (int i = 0; i < atts->cmn.num_colors; i++) {
        if (atts->wgpu.colors[i].view) {
            wgpuTextureViewRelease(atts->wgpu.colors[i].view);
            atts->wgpu.colors[i].view = 0;
        }
        if (atts->wgpu.resolves[i].view) {
            wgpuTextureViewRelease(atts->wgpu.resolves[i].view);
            atts->wgpu.resolves[i].view = 0;
        }
    }
    if (atts->wgpu.depth_stencil.view) {
        wgpuTextureViewRelease(atts->wgpu.depth_stencil.view);
        atts->wgpu.depth_stencil.view = 0;
    }
    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
        if (atts->wgpu.storages[i].view) {
            wgpuTextureViewRelease(atts->wgpu.storages[i].view);
            atts->wgpu.storages[i].view = 0;
        }
    }
}

_SOKOL_PRIVATE void _sg_wgpu_init_color_att(WGPURenderPassColorAttachment* wgpu_att, const sg_color_attachment_action* action, WGPUTextureView color_view, WGPUTextureView resolve_view) {
    wgpu_att->depthSlice = WGPU_DEPTH_SLICE_UNDEFINED;
    wgpu_att->view = color_view;
    wgpu_att->resolveTarget = resolve_view;
    wgpu_att->loadOp = _sg_wgpu_load_op(color_view, action->load_action);
    wgpu_att->storeOp = _sg_wgpu_store_op(color_view, action->store_action);
    wgpu_att->clearValue.r = action->clear_value.r;
    wgpu_att->clearValue.g = action->clear_value.g;
    wgpu_att->clearValue.b = action->clear_value.b;
    wgpu_att->clearValue.a = action->clear_value.a;
}

_SOKOL_PRIVATE void _sg_wgpu_init_ds_att(WGPURenderPassDepthStencilAttachment* wgpu_att, const sg_pass_action* action, sg_pixel_format fmt, WGPUTextureView view) {
    wgpu_att->view = view;
    wgpu_att->depthLoadOp = _sg_wgpu_load_op(view, action->depth.load_action);
    wgpu_att->depthStoreOp = _sg_wgpu_store_op(view, action->depth.store_action);
    wgpu_att->depthClearValue = action->depth.clear_value;
    wgpu_att->depthReadOnly = false;
    if (_sg_is_depth_stencil_format(fmt)) {
        wgpu_att->stencilLoadOp = _sg_wgpu_load_op(view, action->stencil.load_action);
        wgpu_att->stencilStoreOp = _sg_wgpu_store_op(view, action->stencil.store_action);
    } else {
        wgpu_att->stencilLoadOp = WGPULoadOp_Undefined;
        wgpu_att->stencilStoreOp = WGPUStoreOp_Undefined;
    }
    wgpu_att->stencilClearValue = action->stencil.clear_value;
    wgpu_att->stencilReadOnly = false;
}

_SOKOL_PRIVATE void _sg_wgpu_begin_compute_pass(const sg_pass* pass) {
    WGPUComputePassDescriptor wgpu_pass_desc;
    _sg_clear(&wgpu_pass_desc, sizeof(wgpu_pass_desc));
    wgpu_pass_desc.label = _sg_wgpu_stringview(pass->label);
    _sg.wgpu.cpass_enc = wgpuCommandEncoderBeginComputePass(_sg.wgpu.cmd_enc, &wgpu_pass_desc);
    SOKOL_ASSERT(_sg.wgpu.cpass_enc);
    // clear initial bindings
    wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, _SG_WGPU_UB_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
    wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, _SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
    wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, _SG_WGPU_SIMG_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
    _sg_stats_add(wgpu.bindings.num_set_bindgroup, 1);
}

_SOKOL_PRIVATE void _sg_wgpu_begin_render_pass(const sg_pass* pass) {
    const _sg_attachments_t* atts = _sg.cur_pass.atts;
    const sg_swapchain* swapchain = &pass->swapchain;
    const sg_pass_action* action = &pass->action;

    WGPURenderPassDescriptor wgpu_pass_desc;
    WGPURenderPassColorAttachment wgpu_color_att[SG_MAX_COLOR_ATTACHMENTS];
    WGPURenderPassDepthStencilAttachment wgpu_ds_att;
    _sg_clear(&wgpu_pass_desc, sizeof(wgpu_pass_desc));
    _sg_clear(&wgpu_color_att, sizeof(wgpu_color_att));
    _sg_clear(&wgpu_ds_att, sizeof(wgpu_ds_att));
    wgpu_pass_desc.label = _sg_wgpu_stringview(pass->label);
    if (atts) {
        SOKOL_ASSERT(atts->slot.state == SG_RESOURCESTATE_VALID);
        for (int i = 0; i < atts->cmn.num_colors; i++) {
            _sg_wgpu_init_color_att(&wgpu_color_att[i], &action->colors[i], atts->wgpu.colors[i].view, atts->wgpu.resolves[i].view);
        }
        wgpu_pass_desc.colorAttachmentCount = (size_t)atts->cmn.num_colors;
        wgpu_pass_desc.colorAttachments = &wgpu_color_att[0];
        if (atts->wgpu.depth_stencil.image) {
            _sg_wgpu_init_ds_att(&wgpu_ds_att, action, atts->wgpu.depth_stencil.image->cmn.pixel_format, atts->wgpu.depth_stencil.view);
            wgpu_pass_desc.depthStencilAttachment = &wgpu_ds_att;
        }
    } else {
        WGPUTextureView wgpu_color_view = (WGPUTextureView) swapchain->wgpu.render_view;
        WGPUTextureView wgpu_resolve_view = (WGPUTextureView) swapchain->wgpu.resolve_view;
        WGPUTextureView wgpu_depth_stencil_view = (WGPUTextureView) swapchain->wgpu.depth_stencil_view;
        _sg_wgpu_init_color_att(&wgpu_color_att[0], &action->colors[0], wgpu_color_view, wgpu_resolve_view);
        wgpu_pass_desc.colorAttachmentCount = 1;
        wgpu_pass_desc.colorAttachments = &wgpu_color_att[0];
        if (wgpu_depth_stencil_view) {
            SOKOL_ASSERT(swapchain->depth_format > SG_PIXELFORMAT_NONE);
            _sg_wgpu_init_ds_att(&wgpu_ds_att, action, swapchain->depth_format, wgpu_depth_stencil_view);
            wgpu_pass_desc.depthStencilAttachment = &wgpu_ds_att;
        }
    }
    _sg.wgpu.rpass_enc = wgpuCommandEncoderBeginRenderPass(_sg.wgpu.cmd_enc, &wgpu_pass_desc);
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);

    wgpuRenderPassEncoderSetBindGroup(_sg.wgpu.rpass_enc, _SG_WGPU_UB_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
    wgpuRenderPassEncoderSetBindGroup(_sg.wgpu.rpass_enc, _SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
    _sg_stats_add(wgpu.bindings.num_set_bindgroup, 1);
}

_SOKOL_PRIVATE void _sg_wgpu_begin_pass(const sg_pass* pass) {
    SOKOL_ASSERT(pass);
    SOKOL_ASSERT(_sg.wgpu.dev);
    SOKOL_ASSERT(_sg.wgpu.cmd_enc);
    SOKOL_ASSERT(0 == _sg.wgpu.rpass_enc);
    SOKOL_ASSERT(0 == _sg.wgpu.cpass_enc);

    _sg.wgpu.cur_pipeline = 0;
    _sg.wgpu.cur_pipeline_id.id = SG_INVALID_ID;
    _sg_wgpu_bindings_cache_clear();

    if (pass->compute) {
        _sg_wgpu_begin_compute_pass(pass);
    } else {
        _sg_wgpu_begin_render_pass(pass);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_end_pass(void) {
    if (_sg.wgpu.rpass_enc) {
        wgpuRenderPassEncoderEnd(_sg.wgpu.rpass_enc);
        wgpuRenderPassEncoderRelease(_sg.wgpu.rpass_enc);
        _sg.wgpu.rpass_enc = 0;
    }
    if (_sg.wgpu.cpass_enc) {
        wgpuComputePassEncoderEnd(_sg.wgpu.cpass_enc);
        wgpuComputePassEncoderRelease(_sg.wgpu.cpass_enc);
        _sg.wgpu.cpass_enc = 0;
    }
}

_SOKOL_PRIVATE void _sg_wgpu_commit(void) {
    SOKOL_ASSERT(_sg.wgpu.cmd_enc);

    _sg_wgpu_uniform_buffer_on_commit();

    WGPUCommandBufferDescriptor cmd_buf_desc;
    _sg_clear(&cmd_buf_desc, sizeof(cmd_buf_desc));
    WGPUCommandBuffer wgpu_cmd_buf = wgpuCommandEncoderFinish(_sg.wgpu.cmd_enc, &cmd_buf_desc);
    SOKOL_ASSERT(wgpu_cmd_buf);
    wgpuCommandEncoderRelease(_sg.wgpu.cmd_enc);
    _sg.wgpu.cmd_enc = 0;

    wgpuQueueSubmit(_sg.wgpu.queue, 1, &wgpu_cmd_buf);
    wgpuCommandBufferRelease(wgpu_cmd_buf);

    // create a new render-command-encoder for next frame
    WGPUCommandEncoderDescriptor cmd_enc_desc;
    _sg_clear(&cmd_enc_desc, sizeof(cmd_enc_desc));
    _sg.wgpu.cmd_enc = wgpuDeviceCreateCommandEncoder(_sg.wgpu.dev, &cmd_enc_desc);
}

_SOKOL_PRIVATE void _sg_wgpu_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);
    // FIXME FIXME FIXME: CLIPPING THE VIEWPORT HERE IS WRONG!!!
    // (but currently required because WebGPU insists that the viewport rectangle must be
    // fully contained inside the framebuffer, but this doesn't make any sense, and also
    // isn't required by the backend APIs)
    const _sg_recti_t clip = _sg_clipi(x, y, w, h, _sg.cur_pass.width, _sg.cur_pass.height);
    float xf = (float) clip.x;
    float yf = (float) (origin_top_left ? clip.y : (_sg.cur_pass.height - (clip.y + clip.h)));
    float wf = (float) clip.w;
    float hf = (float) clip.h;
    wgpuRenderPassEncoderSetViewport(_sg.wgpu.rpass_enc, xf, yf, wf, hf, 0.0f, 1.0f);
}

_SOKOL_PRIVATE void _sg_wgpu_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);
    const _sg_recti_t clip = _sg_clipi(x, y, w, h, _sg.cur_pass.width, _sg.cur_pass.height);
    uint32_t sx = (uint32_t) clip.x;
    uint32_t sy = (uint32_t) (origin_top_left ? clip.y : (_sg.cur_pass.height - (clip.y + clip.h)));
    uint32_t sw = (uint32_t) clip.w;
    uint32_t sh = (uint32_t) clip.h;
    wgpuRenderPassEncoderSetScissorRect(_sg.wgpu.rpass_enc, sx, sy, sw, sh);
}

_SOKOL_PRIVATE void _sg_wgpu_set_ub_bindgroup(const _sg_shader_t* shd) {
    // NOTE: dynamic offsets must be in binding order, not in BindGroupEntry order
    SOKOL_ASSERT(shd->wgpu.ub_num_dynoffsets < SG_MAX_UNIFORMBLOCK_BINDSLOTS);
    uint32_t dyn_offsets[SG_MAX_UNIFORMBLOCK_BINDSLOTS];
    _sg_clear(dyn_offsets, sizeof(dyn_offsets));
    for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
        if (shd->cmn.uniform_blocks[i].stage == SG_SHADERSTAGE_NONE) {
            continue;
        }
        uint8_t dynoffset_index = shd->wgpu.ub_dynoffsets[i];
        SOKOL_ASSERT(dynoffset_index < shd->wgpu.ub_num_dynoffsets);
        dyn_offsets[dynoffset_index] = _sg.wgpu.uniform.bind_offsets[i];
    }
    if (_sg.cur_pass.is_compute) {
        SOKOL_ASSERT(_sg.wgpu.cpass_enc);
        wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc,
            _SG_WGPU_UB_BINDGROUP_INDEX,
            shd->wgpu.bg_ub,
            shd->wgpu.ub_num_dynoffsets,
            dyn_offsets);
    } else {
        SOKOL_ASSERT(_sg.wgpu.rpass_enc);
        wgpuRenderPassEncoderSetBindGroup(_sg.wgpu.rpass_enc,
            _SG_WGPU_UB_BINDGROUP_INDEX,
            shd->wgpu.bg_ub,
            shd->wgpu.ub_num_dynoffsets,
            dyn_offsets);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_apply_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip);
    SOKOL_ASSERT(pip->shader && (pip->shader->slot.id == pip->cmn.shader_id.id));
    _sg.wgpu.cur_pipeline = pip;
    _sg.wgpu.cur_pipeline_id.id = pip->slot.id;
    if (pip->cmn.is_compute) {
        SOKOL_ASSERT(_sg.cur_pass.is_compute);
        SOKOL_ASSERT(pip->wgpu.cpip);
        SOKOL_ASSERT(_sg.wgpu.cpass_enc);
        wgpuComputePassEncoderSetPipeline(_sg.wgpu.cpass_enc, pip->wgpu.cpip);

        // adhoc-create a storage attachment bindgroup without going through the bindgroups cache
        // FIXME: the 'resource view update' will get rid of this special case because then storage images
        // will be regular resource bindings
        if (pip->shader->wgpu.bgl_simg) {
            _sg_stats_add(wgpu.bindings.num_create_bindgroup, 1);
            _sg_shader_t* shd = pip->shader;
            SOKOL_ASSERT(shd);
            WGPUBindGroupLayout bgl = shd->wgpu.bgl_simg;
            WGPUBindGroupEntry bg_entries[_SG_WGPU_MAX_SIMG_BINDGROUP_ENTRIES];
            _sg_clear(&bg_entries, sizeof(bg_entries));
            size_t bgl_index = 0;
            for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
                if (shd->cmn.storage_images[i].stage == SG_SHADERSTAGE_NONE) {
                    continue;
                }
                SOKOL_ASSERT(_sg.cur_pass.atts);
                _sg_attachments_t* atts = _sg.cur_pass.atts;
                SOKOL_ASSERT(atts->wgpu.storages[i].view);
                WGPUBindGroupEntry* bg_entry = &bg_entries[bgl_index];
                bg_entry->binding = shd->wgpu.simg_grp2_bnd_n[i];
                bg_entry->textureView = atts->wgpu.storages[i].view;
                bgl_index++;
            }
            SOKOL_ASSERT(bgl_index > 0);
            WGPUBindGroupDescriptor bg_desc;
            _sg_clear(&bg_desc, sizeof(bg_desc));
            bg_desc.layout = bgl;
            bg_desc.entryCount = bgl_index;
            bg_desc.entries = bg_entries;
            WGPUBindGroup bg = wgpuDeviceCreateBindGroup(_sg.wgpu.dev, &bg_desc);
            if (bg) {
                wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, _SG_WGPU_SIMG_BINDGROUP_INDEX, bg, 0, 0);
                wgpuBindGroupRelease(bg);
            } else {
                _SG_ERROR(WGPU_CREATEBINDGROUP_FAILED);
            }
        } else {
            // no storage attachment bindings, clear bindgroup slot
            wgpuComputePassEncoderSetBindGroup(_sg.wgpu.cpass_enc, _SG_WGPU_SIMG_BINDGROUP_INDEX, _sg.wgpu.empty_bind_group, 0, 0);
        }
    } else {
        SOKOL_ASSERT(!_sg.cur_pass.is_compute);
        SOKOL_ASSERT(pip->wgpu.rpip);
        SOKOL_ASSERT(_sg.wgpu.rpass_enc);
        _sg.wgpu.use_indexed_draw = (pip->cmn.index_type != SG_INDEXTYPE_NONE);
        wgpuRenderPassEncoderSetPipeline(_sg.wgpu.rpass_enc, pip->wgpu.rpip);
        wgpuRenderPassEncoderSetBlendConstant(_sg.wgpu.rpass_enc, &pip->wgpu.blend_color);
        wgpuRenderPassEncoderSetStencilReference(_sg.wgpu.rpass_enc, pip->cmn.stencil.ref);
    }
    // bind groups must be set because pipelines without uniform blocks or resource bindings
    // will still create 'empty' BindGroupLayouts
    _sg_wgpu_set_ub_bindgroup(pip->shader);
    _sg_wgpu_set_bindgroup(_SG_WGPU_IMG_SMP_SBUF_BINDGROUP_INDEX, 0); // this will set the 'empty bind group'
}

_SOKOL_PRIVATE bool _sg_wgpu_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    SOKOL_ASSERT(bnd);
    SOKOL_ASSERT(bnd->pip->shader && (bnd->pip->cmn.shader_id.id == bnd->pip->shader->slot.id));
    bool retval = true;
    if (!_sg.cur_pass.is_compute) {
        retval &= _sg_wgpu_apply_index_buffer(bnd);
        retval &= _sg_wgpu_apply_vertex_buffers(bnd);
    }
    retval &= _sg_wgpu_apply_bindings_bindgroup(bnd);
    return retval;
}

_SOKOL_PRIVATE void _sg_wgpu_apply_uniforms(int ub_slot, const sg_range* data) {
    const uint32_t alignment = _sg.wgpu.limits.limits.minUniformBufferOffsetAlignment;
    SOKOL_ASSERT(_sg.wgpu.uniform.staging);
    SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
    SOKOL_ASSERT((_sg.wgpu.uniform.offset + data->size) <= _sg.wgpu.uniform.num_bytes);
    SOKOL_ASSERT((_sg.wgpu.uniform.offset & (alignment - 1)) == 0);
    const _sg_pipeline_t* pip = _sg.wgpu.cur_pipeline;
    SOKOL_ASSERT(pip && pip->shader);
    SOKOL_ASSERT(pip->slot.id == _sg.wgpu.cur_pipeline_id.id);
    const _sg_shader_t* shd = pip->shader;
    SOKOL_ASSERT(shd->slot.id == pip->cmn.shader_id.id);
    SOKOL_ASSERT(data->size == shd->cmn.uniform_blocks[ub_slot].size);
    SOKOL_ASSERT(data->size <= _SG_WGPU_MAX_UNIFORM_UPDATE_SIZE);

    _sg_stats_add(wgpu.uniforms.num_set_bindgroup, 1);
    memcpy(_sg.wgpu.uniform.staging + _sg.wgpu.uniform.offset, data->ptr, data->size);
    _sg.wgpu.uniform.bind_offsets[ub_slot] = _sg.wgpu.uniform.offset;
    _sg.wgpu.uniform.offset = _sg_roundup_u32(_sg.wgpu.uniform.offset + (uint32_t)data->size, alignment);

    _sg_wgpu_set_ub_bindgroup(shd);
}

_SOKOL_PRIVATE void _sg_wgpu_draw(int base_element, int num_elements, int num_instances) {
    SOKOL_ASSERT(_sg.wgpu.rpass_enc);
    SOKOL_ASSERT(_sg.wgpu.cur_pipeline && (_sg.wgpu.cur_pipeline->slot.id == _sg.wgpu.cur_pipeline_id.id));
    if (SG_INDEXTYPE_NONE != _sg.wgpu.cur_pipeline->cmn.index_type) {
        wgpuRenderPassEncoderDrawIndexed(_sg.wgpu.rpass_enc, (uint32_t)num_elements, (uint32_t)num_instances, (uint32_t)base_element, 0, 0);
    } else {
        wgpuRenderPassEncoderDraw(_sg.wgpu.rpass_enc, (uint32_t)num_elements, (uint32_t)num_instances, (uint32_t)base_element, 0);
    }
}

_SOKOL_PRIVATE void _sg_wgpu_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    SOKOL_ASSERT(_sg.wgpu.cpass_enc);
    wgpuComputePassEncoderDispatchWorkgroups(_sg.wgpu.cpass_enc,
        (uint32_t)num_groups_x,
        (uint32_t)num_groups_y,
        (uint32_t)num_groups_z);
}

_SOKOL_PRIVATE void _sg_wgpu_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    SOKOL_ASSERT(data && data->ptr && (data->size > 0));
    SOKOL_ASSERT(buf);
    _sg_wgpu_copy_buffer_data(buf, 0, data);
}

_SOKOL_PRIVATE void _sg_wgpu_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    SOKOL_ASSERT(data && data->ptr && (data->size > 0));
    _SOKOL_UNUSED(new_frame);
    _sg_wgpu_copy_buffer_data(buf, (uint64_t)buf->cmn.append_pos, data);
}

_SOKOL_PRIVATE void _sg_wgpu_update_image(_sg_image_t* img, const sg_image_data* data) {
    SOKOL_ASSERT(img && data);
    _sg_wgpu_copy_image_data(img, img->wgpu.tex, data);
}
#endif

//  ██████  ███████ ███    ██ ███████ ██████  ██  ██████     ██████   █████   ██████ ██   ██ ███████ ███    ██ ██████
// ██       ██      ████   ██ ██      ██   ██ ██ ██          ██   ██ ██   ██ ██      ██  ██  ██      ████   ██ ██   ██
// ██   ███ █████   ██ ██  ██ █████   ██████  ██ ██          ██████  ███████ ██      █████   █████   ██ ██  ██ ██   ██
// ██    ██ ██      ██  ██ ██ ██      ██   ██ ██ ██          ██   ██ ██   ██ ██      ██  ██  ██      ██  ██ ██ ██   ██
//  ██████  ███████ ██   ████ ███████ ██   ██ ██  ██████     ██████  ██   ██  ██████ ██   ██ ███████ ██   ████ ██████
//
// >>generic backend
static inline void _sg_setup_backend(const sg_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_setup_backend(desc);
    #elif defined(SOKOL_METAL)
    _sg_mtl_setup_backend(desc);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_setup_backend(desc);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_setup_backend(desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_setup_backend(desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_backend(void) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_backend();
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_backend();
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_backend();
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_backend();
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_backend();
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_reset_state_cache(void) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_reset_state_cache();
    #elif defined(SOKOL_METAL)
    _sg_mtl_reset_state_cache();
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_reset_state_cache();
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_reset_state_cache();
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_reset_state_cache();
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_buffer(buf, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_buffer(buf, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_buffer(buf, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_buffer(buf, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_buffer(buf, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_buffer(_sg_buffer_t* buf) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_buffer(buf);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_buffer(buf);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_buffer(buf);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_buffer(buf);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_buffer(buf);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_image(_sg_image_t* img, const sg_image_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_image(img, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_image(img, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_image(img, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_image(img, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_image(img, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_image(_sg_image_t* img) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_image(img);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_image(img);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_image(img);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_image(img);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_image(img);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_sampler(smp, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_sampler(smp, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_sampler(smp, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_sampler(smp, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_sampler(smp, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_sampler(_sg_sampler_t* smp) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_sampler(smp);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_sampler(smp);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_sampler(smp);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_sampler(smp);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_sampler(smp);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_shader(shd, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_shader(shd, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_shader(shd, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_shader(shd, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_shader(shd, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_shader(_sg_shader_t* shd) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_shader(shd);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_shader(shd);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_shader(shd);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_shader(shd);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_shader(shd);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_pipeline(_sg_pipeline_t* pip, const sg_pipeline_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_pipeline(pip, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_pipeline(pip, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_pipeline(pip, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_pipeline(pip, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_pipeline(pip, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_pipeline(_sg_pipeline_t* pip) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_pipeline(pip);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_pipeline(pip);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_pipeline(pip);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_discard_pipeline(pip);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_pipeline(pip);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline sg_resource_state _sg_create_attachments(_sg_attachments_t* atts, const sg_attachments_desc* desc) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_create_attachments(atts, desc);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_create_attachments(atts, desc);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_create_attachments(atts, desc);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_create_attachments(atts, desc);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_create_attachments(atts, desc);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_discard_attachments(_sg_attachments_t* atts) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_discard_attachments(atts);
    #elif defined(SOKOL_METAL)
    _sg_mtl_discard_attachments(atts);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_discard_attachments(atts);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_discard_attachments(atts);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_discard_attachments(atts);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_begin_pass(const sg_pass* pass) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_begin_pass(pass);
    #elif defined(SOKOL_METAL)
    _sg_mtl_begin_pass(pass);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_begin_pass(pass);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_begin_pass(pass);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_begin_pass(pass);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_end_pass(void) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_end_pass();
    #elif defined(SOKOL_METAL)
    _sg_mtl_end_pass();
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_end_pass();
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_end_pass();
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_end_pass();
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_apply_viewport(int x, int y, int w, int h, bool origin_top_left) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_apply_viewport(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_METAL)
    _sg_mtl_apply_viewport(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_apply_viewport(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_apply_viewport(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_apply_viewport(x, y, w, h, origin_top_left);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_apply_scissor_rect(int x, int y, int w, int h, bool origin_top_left) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_apply_scissor_rect(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_METAL)
    _sg_mtl_apply_scissor_rect(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_apply_scissor_rect(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_apply_scissor_rect(x, y, w, h, origin_top_left);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_apply_scissor_rect(x, y, w, h, origin_top_left);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_apply_pipeline(_sg_pipeline_t* pip) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_apply_pipeline(pip);
    #elif defined(SOKOL_METAL)
    _sg_mtl_apply_pipeline(pip);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_apply_pipeline(pip);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_apply_pipeline(pip);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_apply_pipeline(pip);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline bool _sg_apply_bindings(_sg_bindings_ptrs_t* bnd) {
    #if defined(_SOKOL_ANY_GL)
    return _sg_gl_apply_bindings(bnd);
    #elif defined(SOKOL_METAL)
    return _sg_mtl_apply_bindings(bnd);
    #elif defined(SOKOL_D3D11)
    return _sg_d3d11_apply_bindings(bnd);
    #elif defined(SOKOL_WGPU)
    return _sg_wgpu_apply_bindings(bnd);
    #elif defined(SOKOL_DUMMY_BACKEND)
    return _sg_dummy_apply_bindings(bnd);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_apply_uniforms(int ub_slot, const sg_range* data) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_apply_uniforms(ub_slot, data);
    #elif defined(SOKOL_METAL)
    _sg_mtl_apply_uniforms(ub_slot, data);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_apply_uniforms(ub_slot, data);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_apply_uniforms(ub_slot, data);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_apply_uniforms(ub_slot, data);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_draw(int base_element, int num_elements, int num_instances) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_draw(base_element, num_elements, num_instances);
    #elif defined(SOKOL_METAL)
    _sg_mtl_draw(base_element, num_elements, num_instances);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_draw(base_element, num_elements, num_instances);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_draw(base_element, num_elements, num_instances);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_draw(base_element, num_elements, num_instances);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_dispatch(num_groups_x, num_groups_y, num_groups_z);
    #elif defined(SOKOL_METAL)
    _sg_mtl_dispatch(num_groups_x, num_groups_y, num_groups_z);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_dispatch(num_groups_x, num_groups_y, num_groups_z);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_dispatch(num_groups_x, num_groups_y, num_groups_z);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_dispatch(num_groups_x, num_groups_y, num_groups_z);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_commit(void) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_commit();
    #elif defined(SOKOL_METAL)
    _sg_mtl_commit();
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_commit();
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_commit();
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_commit();
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_update_buffer(_sg_buffer_t* buf, const sg_range* data) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_update_buffer(buf, data);
    #elif defined(SOKOL_METAL)
    _sg_mtl_update_buffer(buf, data);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_update_buffer(buf, data);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_update_buffer(buf, data);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_update_buffer(buf, data);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_append_buffer(_sg_buffer_t* buf, const sg_range* data, bool new_frame) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_append_buffer(buf, data, new_frame);
    #elif defined(SOKOL_METAL)
    _sg_mtl_append_buffer(buf, data, new_frame);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_append_buffer(buf, data, new_frame);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_append_buffer(buf, data, new_frame);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_append_buffer(buf, data, new_frame);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_update_image(_sg_image_t* img, const sg_image_data* data) {
    #if defined(_SOKOL_ANY_GL)
    _sg_gl_update_image(img, data);
    #elif defined(SOKOL_METAL)
    _sg_mtl_update_image(img, data);
    #elif defined(SOKOL_D3D11)
    _sg_d3d11_update_image(img, data);
    #elif defined(SOKOL_WGPU)
    _sg_wgpu_update_image(img, data);
    #elif defined(SOKOL_DUMMY_BACKEND)
    _sg_dummy_update_image(img, data);
    #else
    #error("INVALID BACKEND");
    #endif
}

static inline void _sg_push_debug_group(const char* name) {
    #if defined(SOKOL_METAL)
    _sg_mtl_push_debug_group(name);
    #else
    _SOKOL_UNUSED(name);
    #endif
}

static inline void _sg_pop_debug_group(void) {
    #if defined(SOKOL_METAL)
    _sg_mtl_pop_debug_group();
    #endif
}

// ██    ██  █████  ██      ██ ██████   █████  ████████ ██  ██████  ███    ██
// ██    ██ ██   ██ ██      ██ ██   ██ ██   ██    ██    ██ ██    ██ ████   ██
// ██    ██ ███████ ██      ██ ██   ██ ███████    ██    ██ ██    ██ ██ ██  ██
//  ██  ██  ██   ██ ██      ██ ██   ██ ██   ██    ██    ██ ██    ██ ██  ██ ██
//   ████   ██   ██ ███████ ██ ██████  ██   ██    ██    ██  ██████  ██   ████
//
// >>validation
#if defined(SOKOL_DEBUG)
_SOKOL_PRIVATE void _sg_validate_begin(void) {
    _sg.validate_error = SG_LOGITEM_OK;
}

_SOKOL_PRIVATE bool _sg_validate_end(void) {
    if (_sg.validate_error != SG_LOGITEM_OK) {
        #if !defined(SOKOL_VALIDATE_NON_FATAL)
            _SG_PANIC(VALIDATION_FAILED);
            return false;
        #else
            return false;
        #endif
    } else {
        return true;
    }
}
#endif

_SOKOL_PRIVATE bool _sg_one(bool b0, bool b1, bool b2) {
    return (b0 && !b1 && !b2) || (!b0 && b1 && !b2) || (!b0 && !b1 && b2);
}

_SOKOL_PRIVATE bool _sg_validate_buffer_desc(const sg_buffer_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_BUFFERDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_BUFFERDESC_CANARY);
        _SG_VALIDATE(desc->size > 0, VALIDATE_BUFFERDESC_EXPECT_NONZERO_SIZE);
        _SG_VALIDATE(_sg_one(desc->usage.immutable, desc->usage.dynamic_update, desc->usage.stream_update), VALIDATE_BUFFERDESC_IMMUTABLE_DYNAMIC_STREAM);
        if (_sg.features.separate_buffer_types) {
            _SG_VALIDATE(_sg_one(desc->usage.vertex_buffer, desc->usage.index_buffer, desc->usage.storage_buffer), VALIDATE_BUFFERDESC_SEPARATE_BUFFER_TYPES);
        }
        bool injected = (0 != desc->gl_buffers[0]) ||
                        (0 != desc->mtl_buffers[0]) ||
                        (0 != desc->d3d11_buffer) ||
                        (0 != desc->wgpu_buffer);
        if (!injected && desc->usage.immutable) {
            if (desc->data.ptr) {
                _SG_VALIDATE(desc->size == desc->data.size, VALIDATE_BUFFERDESC_EXPECT_MATCHING_DATA_SIZE);
            } else {
                _SG_VALIDATE(desc->usage.storage_buffer, VALIDATE_BUFFERDESC_EXPECT_DATA);
                _SG_VALIDATE(desc->data.size == 0, VALIDATE_BUFFERDESC_EXPECT_ZERO_DATA_SIZE);
            }
        } else {
            _SG_VALIDATE(0 == desc->data.ptr, VALIDATE_BUFFERDESC_EXPECT_NO_DATA);
            _SG_VALIDATE(desc->data.size == 0, VALIDATE_BUFFERDESC_EXPECT_ZERO_DATA_SIZE);
        }
        if (desc->usage.storage_buffer) {
            _SG_VALIDATE(_sg.features.compute, VALIDATE_BUFFERDESC_STORAGEBUFFER_SUPPORTED);
            _SG_VALIDATE(_sg_multiple_u64(desc->size, 4), VALIDATE_BUFFERDESC_STORAGEBUFFER_SIZE_MULTIPLE_4);
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE void _sg_validate_image_data(const sg_image_data* data, sg_pixel_format fmt, int width, int height, int num_faces, int num_mips, int num_slices) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(data);
        _SOKOL_UNUSED(fmt);
        _SOKOL_UNUSED(width);
        _SOKOL_UNUSED(height);
        _SOKOL_UNUSED(num_faces);
        _SOKOL_UNUSED(num_mips);
        _SOKOL_UNUSED(num_slices);
    #else
        for (int face_index = 0; face_index < num_faces; face_index++) {
            for (int mip_index = 0; mip_index < num_mips; mip_index++) {
                const bool has_data = data->subimage[face_index][mip_index].ptr != 0;
                const bool has_size = data->subimage[face_index][mip_index].size > 0;
                _SG_VALIDATE(has_data && has_size, VALIDATE_IMAGEDATA_NODATA);
                const int mip_width = _sg_miplevel_dim(width, mip_index);
                const int mip_height = _sg_miplevel_dim(height, mip_index);
                const int bytes_per_slice = _sg_surface_pitch(fmt, mip_width, mip_height, 1);
                const int expected_size = bytes_per_slice * num_slices;
                _SG_VALIDATE(expected_size == (int)data->subimage[face_index][mip_index].size, VALIDATE_IMAGEDATA_DATA_SIZE);
            }
        }
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_image_desc(const sg_image_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_IMAGEDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_IMAGEDESC_CANARY);
        _SG_VALIDATE(_sg_one(desc->usage.immutable, desc->usage.dynamic_update, desc->usage.stream_update), VALIDATE_IMAGEDESC_IMMUTABLE_DYNAMIC_STREAM);
        if (desc->usage.render_attachment || desc->usage.storage_attachment) {
            _SG_VALIDATE(_sg_one(desc->usage.render_attachment, desc->usage.storage_attachment, false), VALIDATE_IMAGEDESC_RENDER_VS_STORAGE_ATTACHMENT);
        }
        _SG_VALIDATE(desc->width > 0, VALIDATE_IMAGEDESC_WIDTH);
        _SG_VALIDATE(desc->height > 0, VALIDATE_IMAGEDESC_HEIGHT);
        const sg_pixel_format fmt = desc->pixel_format;
        const sg_image_usage* usage = &desc->usage;
        const bool injected = (0 != desc->gl_textures[0]) ||
                              (0 != desc->mtl_textures[0]) ||
                              (0 != desc->d3d11_texture) ||
                              (0 != desc->wgpu_texture);
        if (_sg_is_depth_or_depth_stencil_format(fmt)) {
            _SG_VALIDATE(desc->type != SG_IMAGETYPE_3D, VALIDATE_IMAGEDESC_DEPTH_3D_IMAGE);
        }
        if (usage->render_attachment || usage->storage_attachment) {
            SOKOL_ASSERT(((int)fmt >= 0) && ((int)fmt < _SG_PIXELFORMAT_NUM));
            _SG_VALIDATE(usage->immutable, VALIDATE_IMAGEDESC_ATTACHMENT_EXPECT_IMMUTABLE);
            _SG_VALIDATE(desc->data.subimage[0][0].ptr==0, VALIDATE_IMAGEDESC_ATTACHMENT_EXPECT_NO_DATA);
            if (usage->render_attachment) {
                _SG_VALIDATE(_sg.formats[fmt].render, VALIDATE_IMAGEDESC_RENDERATTACHMENT_PIXELFORMAT);
                if (desc->sample_count > 1) {
                    _SG_VALIDATE(_sg.formats[fmt].msaa, VALIDATE_IMAGEDESC_RENDERATTACHMENT_NO_MSAA_SUPPORT);
                    _SG_VALIDATE(desc->num_mipmaps == 1, VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_NUM_MIPMAPS);
                    _SG_VALIDATE(desc->type != SG_IMAGETYPE_ARRAY, VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_ARRAY_IMAGE);
                    _SG_VALIDATE(desc->type != SG_IMAGETYPE_3D, VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_3D_IMAGE);
                    _SG_VALIDATE(desc->type != SG_IMAGETYPE_CUBE, VALIDATE_IMAGEDESC_RENDERATTACHMENT_MSAA_CUBE_IMAGE);
                }
            } else if (usage->storage_attachment) {
                _SG_VALIDATE(_sg_is_valid_attachment_storage_format(fmt), VALIDATE_IMAGEDESC_STORAGEATTACHMENT_PIXELFORMAT);
                // D3D11 doesn't allow multisampled UAVs (see: https://github.com/gpuweb/gpuweb/issues/513)
                _SG_VALIDATE(desc->sample_count == 1, VALIDATE_IMAGEDESC_STORAGEATTACHMENT_EXPECT_NO_MSAA);
            }
        } else {
            _SG_VALIDATE(desc->sample_count == 1, VALIDATE_IMAGEDESC_MSAA_BUT_NO_ATTACHMENT);
            const bool valid_nonrt_fmt = !_sg_is_valid_attachment_depth_format(fmt);
            _SG_VALIDATE(valid_nonrt_fmt, VALIDATE_IMAGEDESC_NONRT_PIXELFORMAT);
            const bool is_compressed = _sg_is_compressed_pixel_format(desc->pixel_format);
            if (is_compressed) {
                _SG_VALIDATE(usage->immutable, VALIDATE_IMAGEDESC_COMPRESSED_IMMUTABLE);
            }
            if (!injected && usage->immutable) {
                // image desc must have valid data
                _sg_validate_image_data(&desc->data,
                    desc->pixel_format,
                    desc->width,
                    desc->height,
                    (desc->type == SG_IMAGETYPE_CUBE) ? 6 : 1,
                    desc->num_mipmaps,
                    desc->num_slices);
            } else {
                // image desc must not have data
                for (int face_index = 0; face_index < SG_CUBEFACE_NUM; face_index++) {
                    for (int mip_index = 0; mip_index < SG_MAX_MIPMAPS; mip_index++) {
                        const bool no_data = 0 == desc->data.subimage[face_index][mip_index].ptr;
                        const bool no_size = 0 == desc->data.subimage[face_index][mip_index].size;
                        if (injected) {
                            _SG_VALIDATE(no_data && no_size, VALIDATE_IMAGEDESC_INJECTED_NO_DATA);
                        }
                        if (!usage->immutable) {
                            _SG_VALIDATE(no_data && no_size, VALIDATE_IMAGEDESC_DYNAMIC_NO_DATA);
                        }
                    }
                }
            }
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_sampler_desc(const sg_sampler_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_SAMPLERDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_SAMPLERDESC_CANARY);
        // restriction from WebGPU: when anisotropy > 1, all filters must be linear
        if (desc->max_anisotropy > 1) {
            _SG_VALIDATE((desc->min_filter == SG_FILTER_LINEAR)
                      && (desc->mag_filter == SG_FILTER_LINEAR)
                      && (desc->mipmap_filter == SG_FILTER_LINEAR),
                      VALIDATE_SAMPLERDESC_ANISTROPIC_REQUIRES_LINEAR_FILTERING);
        }
        return _sg_validate_end();
    #endif
}

typedef struct {
    uint64_t lo, hi;
} _sg_u128_t;

_SOKOL_PRIVATE _sg_u128_t _sg_u128(void) {
    _sg_u128_t res;
    _sg_clear(&res, sizeof(res));
    return res;
}

_SOKOL_PRIVATE _sg_u128_t _sg_validate_set_slot_bit(_sg_u128_t bits, sg_shader_stage stage, uint8_t slot) {
    switch (stage) {
        case SG_SHADERSTAGE_NONE:
            SOKOL_ASSERT(slot < 128);
            if (slot < 64) {
                bits.lo |= 1ULL << slot;
            } else {
                bits.hi |= 1ULL << (slot - 64);
            }
            break;
        case SG_SHADERSTAGE_VERTEX:
            SOKOL_ASSERT(slot < 64);
            bits.lo |= 1ULL << slot;
            break;
        case SG_SHADERSTAGE_FRAGMENT:
            SOKOL_ASSERT(slot < 64);
            bits.hi |= 1ULL << slot;
            break;
        case SG_SHADERSTAGE_COMPUTE:
            SOKOL_ASSERT(slot < 64);
            bits.lo |= 1ULL << slot;
            break;
        default:
            SOKOL_UNREACHABLE;
            break;
    }
    return bits;
}

_SOKOL_PRIVATE bool _sg_validate_slot_bits(_sg_u128_t bits, sg_shader_stage stage, uint8_t slot) {
    _sg_u128_t mask = _sg_u128();
    switch (stage) {
        case SG_SHADERSTAGE_NONE:
            SOKOL_ASSERT(slot < 128);
            if (slot < 64) {
                mask.lo = 1ULL << slot;
            } else {
                mask.hi = 1ULL << (slot - 64);
            }
            break;
        case SG_SHADERSTAGE_VERTEX:
            SOKOL_ASSERT(slot < 64);
            mask.lo = 1ULL << slot;
            break;
        case SG_SHADERSTAGE_FRAGMENT:
            SOKOL_ASSERT(slot < 64);
            mask.hi = 1ULL << slot;
            break;
        case SG_SHADERSTAGE_COMPUTE:
            SOKOL_ASSERT(slot < 64);
            mask.lo = 1ULL << slot;
            break;
        default:
            SOKOL_UNREACHABLE;
            break;
    }
    return ((bits.lo & mask.lo) == 0) && ((bits.hi & mask.hi) == 0);
}

_SOKOL_PRIVATE bool _sg_validate_shader_desc(const sg_shader_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        bool is_compute_shader = (desc->compute_func.source != 0) || (desc->compute_func.bytecode.ptr != 0);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_SHADERDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_SHADERDESC_CANARY);
        #if defined(SOKOL_GLCORE) || defined(SOKOL_GLES3) || defined(SOKOL_WGPU)
            // on GL or WebGPU, must provide shader source code
            if (is_compute_shader) {
                _SG_VALIDATE(0 != desc->compute_func.source, VALIDATE_SHADERDESC_COMPUTE_SOURCE);
            } else {
                _SG_VALIDATE(0 != desc->vertex_func.source, VALIDATE_SHADERDESC_VERTEX_SOURCE);
                _SG_VALIDATE(0 != desc->fragment_func.source, VALIDATE_SHADERDESC_FRAGMENT_SOURCE);
            }
        #elif defined(SOKOL_METAL) || defined(SOKOL_D3D11)
            // on Metal or D3D11, must provide shader source code or byte code
            if (is_compute_shader) {
                _SG_VALIDATE((0 != desc->compute_func.source) || (0 != desc->compute_func.bytecode.ptr), VALIDATE_SHADERDESC_COMPUTE_SOURCE_OR_BYTECODE);
            } else {
                _SG_VALIDATE((0 != desc->vertex_func.source)|| (0 != desc->vertex_func.bytecode.ptr), VALIDATE_SHADERDESC_VERTEX_SOURCE_OR_BYTECODE);
                _SG_VALIDATE((0 != desc->fragment_func.source) || (0 != desc->fragment_func.bytecode.ptr), VALIDATE_SHADERDESC_FRAGMENT_SOURCE_OR_BYTECODE);
            }
        #else
            // Dummy Backend, don't require source or bytecode
        #endif
        if (is_compute_shader) {
            _SG_VALIDATE((0 == desc->vertex_func.source) && (0 == desc->vertex_func.bytecode.ptr), VALIDATE_SHADERDESC_INVALID_SHADER_COMBO);
            _SG_VALIDATE((0 == desc->fragment_func.source) && (0 == desc->fragment_func.bytecode.ptr), VALIDATE_SHADERDESC_INVALID_SHADER_COMBO);
        } else {
            _SG_VALIDATE((0 == desc->compute_func.source) && (0 == desc->compute_func.bytecode.ptr), VALIDATE_SHADERDESC_INVALID_SHADER_COMBO);
        }
        #if defined(SOKOL_METAL)
        if (is_compute_shader) {
            _SG_VALIDATE(desc->mtl_threads_per_threadgroup.x > 0, VALIDATE_SHADERDESC_METAL_THREADS_PER_THREADGROUP);
            _SG_VALIDATE(desc->mtl_threads_per_threadgroup.y > 0, VALIDATE_SHADERDESC_METAL_THREADS_PER_THREADGROUP);
            _SG_VALIDATE(desc->mtl_threads_per_threadgroup.z > 0, VALIDATE_SHADERDESC_METAL_THREADS_PER_THREADGROUP);
        }
        #endif
        for (size_t i = 0; i < SG_MAX_VERTEX_ATTRIBUTES; i++) {
            if (desc->attrs[i].glsl_name) {
                _SG_VALIDATE(strlen(desc->attrs[i].glsl_name) < _SG_STRING_SIZE, VALIDATE_SHADERDESC_ATTR_STRING_TOO_LONG);
            }
            if (desc->attrs[i].hlsl_sem_name) {
                _SG_VALIDATE(strlen(desc->attrs[i].hlsl_sem_name) < _SG_STRING_SIZE, VALIDATE_SHADERDESC_ATTR_STRING_TOO_LONG);
            }
        }
        // if shader byte code, the size must also be provided
        if (0 != desc->vertex_func.bytecode.ptr) {
            _SG_VALIDATE(desc->vertex_func.bytecode.size > 0, VALIDATE_SHADERDESC_NO_BYTECODE_SIZE);
        }
        if (0 != desc->fragment_func.bytecode.ptr) {
            _SG_VALIDATE(desc->fragment_func.bytecode.size > 0, VALIDATE_SHADERDESC_NO_BYTECODE_SIZE);
        }
        if (0 != desc->compute_func.bytecode.ptr) {
            _SG_VALIDATE(desc->compute_func.bytecode.size > 0, VALIDATE_SHADERDESC_NO_BYTECODE_SIZE);
        }

        #if defined(SOKOL_METAL)
        _sg_u128_t msl_buf_bits = _sg_u128();
        _sg_u128_t msl_tex_bits = _sg_u128();
        _sg_u128_t msl_smp_bits = _sg_u128();
        #elif defined(SOKOL_D3D11)
        _sg_u128_t hlsl_buf_bits = _sg_u128();
        _sg_u128_t hlsl_srv_bits = _sg_u128();
        _sg_u128_t hlsl_uav_bits = _sg_u128();
        _sg_u128_t hlsl_smp_bits = _sg_u128();
        #elif defined(_SOKOL_ANY_GL)
        _sg_u128_t glsl_sbuf_bnd_bits = _sg_u128();
        _sg_u128_t glsl_simg_bnd_bits = _sg_u128();
        #elif defined(SOKOL_WGPU)
        _sg_u128_t wgsl_group0_bits = _sg_u128();
        _sg_u128_t wgsl_group1_bits = _sg_u128();
        _sg_u128_t wgsl_group2_bits = _sg_u128();
        #endif
        for (size_t ub_idx = 0; ub_idx < SG_MAX_UNIFORMBLOCK_BINDSLOTS; ub_idx++) {
            const sg_shader_uniform_block* ub_desc = &desc->uniform_blocks[ub_idx];
            if (ub_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            _SG_VALIDATE(ub_desc->size > 0, VALIDATE_SHADERDESC_UNIFORMBLOCK_SIZE_IS_ZERO);
            #if defined(SOKOL_METAL)
            _SG_VALIDATE(ub_desc->msl_buffer_n < _SG_MTL_MAX_STAGE_UB_BINDINGS, VALIDATE_SHADERDESC_UNIFORMBLOCK_METAL_BUFFER_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(msl_buf_bits, ub_desc->stage, ub_desc->msl_buffer_n), VALIDATE_SHADERDESC_UNIFORMBLOCK_METAL_BUFFER_SLOT_COLLISION);
            msl_buf_bits = _sg_validate_set_slot_bit(msl_buf_bits, ub_desc->stage, ub_desc->msl_buffer_n);
            #elif defined(SOKOL_D3D11)
            _SG_VALIDATE(ub_desc->hlsl_register_b_n < _SG_D3D11_MAX_STAGE_UB_BINDINGS, VALIDATE_SHADERDESC_UNIFORMBLOCK_HLSL_REGISTER_B_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(hlsl_buf_bits, ub_desc->stage, ub_desc->hlsl_register_b_n), VALIDATE_SHADERDESC_UNIFORMBLOCK_HLSL_REGISTER_B_COLLISION);
            hlsl_buf_bits = _sg_validate_set_slot_bit(hlsl_buf_bits, ub_desc->stage, ub_desc->hlsl_register_b_n);
            #elif defined(SOKOL_WGPU)
            _SG_VALIDATE(ub_desc->wgsl_group0_binding_n < _SG_WGPU_MAX_UB_BINDGROUP_BIND_SLOTS, VALIDATE_SHADERDESC_UNIFORMBLOCK_WGSL_GROUP0_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(wgsl_group0_bits, SG_SHADERSTAGE_NONE, ub_desc->wgsl_group0_binding_n), VALIDATE_SHADERDESC_UNIFORMBLOCK_WGSL_GROUP0_BINDING_COLLISION);
            wgsl_group0_bits = _sg_validate_set_slot_bit(wgsl_group0_bits, SG_SHADERSTAGE_NONE, ub_desc->wgsl_group0_binding_n);
            #endif
            #if defined(_SOKOL_ANY_GL)
            bool uniforms_continuous = true;
            uint32_t uniform_offset = 0;
            int num_uniforms = 0;
            for (size_t u_index = 0; u_index < SG_MAX_UNIFORMBLOCK_MEMBERS; u_index++) {
                const sg_glsl_shader_uniform* u_desc = &ub_desc->glsl_uniforms[u_index];
                if (u_desc->type != SG_UNIFORMTYPE_INVALID) {
                    _SG_VALIDATE(uniforms_continuous, VALIDATE_SHADERDESC_UNIFORMBLOCK_NO_CONT_MEMBERS);
                    _SG_VALIDATE(u_desc->glsl_name, VALIDATE_SHADERDESC_UNIFORMBLOCK_UNIFORM_GLSL_NAME);
                    const int array_count = u_desc->array_count;
                    _SG_VALIDATE(array_count > 0, VALIDATE_SHADERDESC_UNIFORMBLOCK_ARRAY_COUNT);
                    const uint32_t u_align = _sg_uniform_alignment(u_desc->type, array_count, ub_desc->layout);
                    const uint32_t u_size  = _sg_uniform_size(u_desc->type, array_count, ub_desc->layout);
                    uniform_offset = _sg_align_u32(uniform_offset, u_align);
                    uniform_offset += u_size;
                    num_uniforms++;
                    // with std140, arrays are only allowed for FLOAT4, INT4, MAT4
                    if (ub_desc->layout == SG_UNIFORMLAYOUT_STD140) {
                        if (array_count > 1) {
                            _SG_VALIDATE((u_desc->type == SG_UNIFORMTYPE_FLOAT4) || (u_desc->type == SG_UNIFORMTYPE_INT4) || (u_desc->type == SG_UNIFORMTYPE_MAT4), VALIDATE_SHADERDESC_UNIFORMBLOCK_STD140_ARRAY_TYPE);
                        }
                    }
                } else {
                    uniforms_continuous = false;
                }
            }
            if (ub_desc->layout == SG_UNIFORMLAYOUT_STD140) {
                uniform_offset = _sg_align_u32(uniform_offset, 16);
            }
            _SG_VALIDATE((size_t)uniform_offset == ub_desc->size, VALIDATE_SHADERDESC_UNIFORMBLOCK_SIZE_MISMATCH);
            _SG_VALIDATE(num_uniforms > 0, VALIDATE_SHADERDESC_UNIFORMBLOCK_NO_MEMBERS);
            #endif
        }

        for (size_t sbuf_idx = 0; sbuf_idx < SG_MAX_STORAGEBUFFER_BINDSLOTS; sbuf_idx++) {
            const sg_shader_storage_buffer* sbuf_desc = &desc->storage_buffers[sbuf_idx];
            if (sbuf_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            #if defined(SOKOL_METAL)
            _SG_VALIDATE((sbuf_desc->msl_buffer_n >= _SG_MTL_MAX_STAGE_UB_BINDINGS) && (sbuf_desc->msl_buffer_n < _SG_MTL_MAX_STAGE_UB_SBUF_BINDINGS), VALIDATE_SHADERDESC_STORAGEBUFFER_METAL_BUFFER_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(msl_buf_bits, sbuf_desc->stage, sbuf_desc->msl_buffer_n), VALIDATE_SHADERDESC_STORAGEBUFFER_METAL_BUFFER_SLOT_COLLISION);
            msl_buf_bits = _sg_validate_set_slot_bit(msl_buf_bits, sbuf_desc->stage, sbuf_desc->msl_buffer_n);
            #elif defined(SOKOL_D3D11)
            if (sbuf_desc->readonly) {
                _SG_VALIDATE(sbuf_desc->hlsl_register_t_n < _SG_D3D11_MAX_STAGE_SRV_BINDINGS, VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_T_OUT_OF_RANGE);
                _SG_VALIDATE(_sg_validate_slot_bits(hlsl_srv_bits, sbuf_desc->stage, sbuf_desc->hlsl_register_t_n), VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_T_COLLISION);
                hlsl_srv_bits = _sg_validate_set_slot_bit(hlsl_srv_bits, sbuf_desc->stage, sbuf_desc->hlsl_register_t_n);
            } else {
                _SG_VALIDATE(sbuf_desc->hlsl_register_u_n < _SG_D3D11_MAX_STAGE_UAV_BINDINGS, VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_U_OUT_OF_RANGE);
                _SG_VALIDATE(_sg_validate_slot_bits(hlsl_uav_bits, sbuf_desc->stage, sbuf_desc->hlsl_register_u_n), VALIDATE_SHADERDESC_STORAGEBUFFER_HLSL_REGISTER_U_COLLISION);
                hlsl_uav_bits = _sg_validate_set_slot_bit(hlsl_uav_bits, sbuf_desc->stage, sbuf_desc->hlsl_register_u_n);
            }
            #elif defined(_SOKOL_ANY_GL)
            _SG_VALIDATE(sbuf_desc->glsl_binding_n < _SG_GL_MAX_SBUF_BINDINGS, VALIDATE_SHADERDESC_STORAGEBUFFER_GLSL_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(glsl_sbuf_bnd_bits, SG_SHADERSTAGE_NONE, sbuf_desc->glsl_binding_n), VALIDATE_SHADERDESC_STORAGEBUFFER_GLSL_BINDING_COLLISION);
            glsl_sbuf_bnd_bits = _sg_validate_set_slot_bit(glsl_sbuf_bnd_bits, SG_SHADERSTAGE_NONE, sbuf_desc->glsl_binding_n);
            #elif defined(SOKOL_WGPU)
            _SG_VALIDATE(sbuf_desc->wgsl_group1_binding_n < _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS, VALIDATE_SHADERDESC_STORAGEBUFFER_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(wgsl_group1_bits, SG_SHADERSTAGE_NONE, sbuf_desc->wgsl_group1_binding_n), VALIDATE_SHADERDESC_STORAGEBUFFER_WGSL_GROUP1_BINDING_COLLISION);
            wgsl_group1_bits = _sg_validate_set_slot_bit(wgsl_group1_bits, SG_SHADERSTAGE_NONE, sbuf_desc->wgsl_group1_binding_n);
            #endif
        }

        for (size_t simg_idx = 0; simg_idx < SG_MAX_STORAGE_ATTACHMENTS; simg_idx++) {
            const sg_shader_storage_image* simg_desc = &desc->storage_images[simg_idx];
            if (simg_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            _SG_VALIDATE(simg_desc->stage == SG_SHADERSTAGE_COMPUTE, VALIDATE_SHADERDESC_STORAGEIMAGE_EXPECT_COMPUTE_STAGE);
            #if defined(SOKOL_METAL)
            _SG_VALIDATE(simg_desc->msl_texture_n < _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS, VALIDATE_SHADERDESC_STORAGEIMAGE_METAL_TEXTURE_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(msl_tex_bits, simg_desc->stage, simg_desc->msl_texture_n), VALIDATE_SHADERDESC_STORAGEIMAGE_METAL_TEXTURE_SLOT_COLLISION);
            msl_tex_bits = _sg_validate_set_slot_bit(msl_tex_bits, simg_desc->stage, simg_desc->msl_texture_n);
            #elif defined(SOKOL_D3D11)
            _SG_VALIDATE(simg_desc->hlsl_register_u_n < _SG_D3D11_MAX_STAGE_UAV_BINDINGS, VALIDATE_SHADERDESC_STORAGEIMAGE_HLSL_REGISTER_U_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(hlsl_uav_bits, simg_desc->stage, simg_desc->hlsl_register_u_n), VALIDATE_SHADERDESC_STORAGEIMAGE_HLSL_REGISTER_U_COLLISION);
            hlsl_uav_bits = _sg_validate_set_slot_bit(hlsl_uav_bits, simg_desc->stage, simg_desc->hlsl_register_u_n);
            #elif defined(_SOKOL_ANY_GL)
            _SG_VALIDATE(simg_desc->glsl_binding_n < _SG_GL_MAX_SIMG_BINDINGS, VALIDATE_SHADERDESC_STORAGEIMAGE_GLSL_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(glsl_simg_bnd_bits, SG_SHADERSTAGE_NONE, simg_desc->glsl_binding_n), VALIDATE_SHADERDESC_STORAGEIMAGE_GLSL_BINDING_COLLISION);
            glsl_simg_bnd_bits = _sg_validate_set_slot_bit(glsl_simg_bnd_bits, SG_SHADERSTAGE_NONE, simg_desc->glsl_binding_n);
            #elif defined(SOKOL_WGPU)
            _SG_VALIDATE(simg_desc->wgsl_group2_binding_n < _SG_WGPU_MAX_SIMG_BIND_SLOTS, VALIDATE_SHADERDESC_STORAGEIMAGE_WGSL_GROUP2_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(wgsl_group2_bits, SG_SHADERSTAGE_NONE, simg_desc->wgsl_group2_binding_n), VALIDATE_SHADERDESC_STORAGEIMAGE_WGSL_GROUP2_BINDING_COLLISION);
            wgsl_group2_bits = _sg_validate_set_slot_bit(wgsl_group2_bits, SG_SHADERSTAGE_NONE, simg_desc->wgsl_group2_binding_n);
            #endif
        }

        uint32_t img_slot_mask = 0;
        for (size_t img_idx = 0; img_idx < SG_MAX_IMAGE_BINDSLOTS; img_idx++) {
            const sg_shader_image* img_desc = &desc->images[img_idx];
            if (img_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            img_slot_mask |= (1 << img_idx);
            #if defined(SOKOL_METAL)
            _SG_VALIDATE(img_desc->msl_texture_n < _SG_MTL_MAX_STAGE_TEXTURE_BINDINGS, VALIDATE_SHADERDESC_IMAGE_METAL_TEXTURE_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(msl_tex_bits, img_desc->stage, img_desc->msl_texture_n), VALIDATE_SHADERDESC_IMAGE_METAL_TEXTURE_SLOT_COLLISION);
            msl_tex_bits = _sg_validate_set_slot_bit(msl_tex_bits, img_desc->stage, img_desc->msl_texture_n);
            #elif defined(SOKOL_D3D11)
            _SG_VALIDATE(img_desc->hlsl_register_t_n < _SG_D3D11_MAX_STAGE_SRV_BINDINGS, VALIDATE_SHADERDESC_IMAGE_HLSL_REGISTER_T_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(hlsl_srv_bits, img_desc->stage, img_desc->hlsl_register_t_n), VALIDATE_SHADERDESC_IMAGE_HLSL_REGISTER_T_COLLISION);
            hlsl_srv_bits = _sg_validate_set_slot_bit(hlsl_srv_bits, img_desc->stage, img_desc->hlsl_register_t_n);
            #elif defined(SOKOL_WGPU)
            _SG_VALIDATE(img_desc->wgsl_group1_binding_n < _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS, VALIDATE_SHADERDESC_IMAGE_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(wgsl_group1_bits, SG_SHADERSTAGE_NONE, img_desc->wgsl_group1_binding_n), VALIDATE_SHADERDESC_IMAGE_WGSL_GROUP1_BINDING_COLLISION);
            wgsl_group1_bits = _sg_validate_set_slot_bit(wgsl_group1_bits, SG_SHADERSTAGE_NONE, img_desc->wgsl_group1_binding_n);
            #endif
        }

        uint32_t smp_slot_mask = 0;
        for (size_t smp_idx = 0; smp_idx < SG_MAX_SAMPLER_BINDSLOTS; smp_idx++) {
            const sg_shader_sampler* smp_desc = &desc->samplers[smp_idx];
            if (smp_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            smp_slot_mask |= (1 << smp_idx);
            #if defined(SOKOL_METAL)
            _SG_VALIDATE(smp_desc->msl_sampler_n < _SG_MTL_MAX_STAGE_SAMPLER_BINDINGS, VALIDATE_SHADERDESC_SAMPLER_METAL_SAMPLER_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(msl_smp_bits, smp_desc->stage, smp_desc->msl_sampler_n), VALIDATE_SHADERDESC_SAMPLER_METAL_SAMPLER_SLOT_COLLISION);
            msl_smp_bits = _sg_validate_set_slot_bit(msl_smp_bits, smp_desc->stage, smp_desc->msl_sampler_n);
            #elif defined(SOKOL_D3D11)
            _SG_VALIDATE(smp_desc->hlsl_register_s_n < _SG_D3D11_MAX_STAGE_SMP_BINDINGS, VALIDATE_SHADERDESC_SAMPLER_HLSL_REGISTER_S_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(hlsl_smp_bits, smp_desc->stage, smp_desc->hlsl_register_s_n), VALIDATE_SHADERDESC_SAMPLER_HLSL_REGISTER_S_COLLISION);
            hlsl_smp_bits = _sg_validate_set_slot_bit(hlsl_smp_bits, smp_desc->stage, smp_desc->hlsl_register_s_n);
            #elif defined(SOKOL_WGPU)
            _SG_VALIDATE(smp_desc->wgsl_group1_binding_n < _SG_WGPU_MAX_IMG_SMP_SBUF_BIND_SLOTS, VALIDATE_SHADERDESC_SAMPLER_WGSL_GROUP1_BINDING_OUT_OF_RANGE);
            _SG_VALIDATE(_sg_validate_slot_bits(wgsl_group1_bits, SG_SHADERSTAGE_NONE, smp_desc->wgsl_group1_binding_n), VALIDATE_SHADERDESC_SAMPLER_WGSL_GROUP1_BINDING_COLLISION);
            wgsl_group1_bits = _sg_validate_set_slot_bit(wgsl_group1_bits, SG_SHADERSTAGE_NONE, smp_desc->wgsl_group1_binding_n);
            #endif
        }

        uint32_t ref_img_slot_mask = 0;
        uint32_t ref_smp_slot_mask = 0;
        for (size_t img_smp_idx = 0; img_smp_idx < SG_MAX_IMAGE_SAMPLER_PAIRS; img_smp_idx++) {
            const sg_shader_image_sampler_pair* img_smp_desc = &desc->image_sampler_pairs[img_smp_idx];
            if (img_smp_desc->stage == SG_SHADERSTAGE_NONE) {
                continue;
            }
            #if defined(_SOKOL_ANY_GL)
            _SG_VALIDATE(img_smp_desc->glsl_name != 0, VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_GLSL_NAME);
            #endif
            const bool img_slot_in_range = img_smp_desc->image_slot < SG_MAX_IMAGE_BINDSLOTS;
            const bool smp_slot_in_range = img_smp_desc->sampler_slot < SG_MAX_SAMPLER_BINDSLOTS;
            _SG_VALIDATE(img_slot_in_range, VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_IMAGE_SLOT_OUT_OF_RANGE);
            _SG_VALIDATE(smp_slot_in_range, VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_SAMPLER_SLOT_OUT_OF_RANGE);
            if (img_slot_in_range && smp_slot_in_range) {
                ref_img_slot_mask |= 1 << img_smp_desc->image_slot;
                ref_smp_slot_mask |= 1 << img_smp_desc->sampler_slot;
                const sg_shader_image* img_desc = &desc->images[img_smp_desc->image_slot];
                const sg_shader_sampler* smp_desc = &desc->samplers[img_smp_desc->sampler_slot];
                _SG_VALIDATE(img_desc->stage == img_smp_desc->stage, VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_IMAGE_STAGE_MISMATCH);
                _SG_VALIDATE(smp_desc->stage == img_smp_desc->stage, VALIDATE_SHADERDESC_IMAGE_SAMPLER_PAIR_SAMPLER_STAGE_MISMATCH);
                const bool needs_nonfiltering = (img_desc->sample_type == SG_IMAGESAMPLETYPE_UINT)
                                             || (img_desc->sample_type == SG_IMAGESAMPLETYPE_SINT)
                                             || (img_desc->sample_type == SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT);
                const bool needs_comparison = img_desc->sample_type == SG_IMAGESAMPLETYPE_DEPTH;
                if (needs_nonfiltering) {
                    _SG_VALIDATE(needs_nonfiltering && (smp_desc->sampler_type == SG_SAMPLERTYPE_NONFILTERING), VALIDATE_SHADERDESC_NONFILTERING_SAMPLER_REQUIRED);
                }
                if (needs_comparison) {
                    _SG_VALIDATE(needs_comparison && (smp_desc->sampler_type == SG_SAMPLERTYPE_COMPARISON), VALIDATE_SHADERDESC_COMPARISON_SAMPLER_REQUIRED);
                }
            }
        }
        // each image and sampler must be referenced by an image sampler
        _SG_VALIDATE(img_slot_mask == ref_img_slot_mask, VALIDATE_SHADERDESC_IMAGE_NOT_REFERENCED_BY_IMAGE_SAMPLER_PAIRS);
        _SG_VALIDATE(smp_slot_mask == ref_smp_slot_mask, VALIDATE_SHADERDESC_SAMPLER_NOT_REFERENCED_BY_IMAGE_SAMPLER_PAIRS);

        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_pipeline_desc(const sg_pipeline_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_PIPELINEDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_PIPELINEDESC_CANARY);
        _SG_VALIDATE(desc->shader.id != SG_INVALID_ID, VALIDATE_PIPELINEDESC_SHADER);
        const _sg_shader_t* shd = _sg_lookup_shader(desc->shader.id);
        _SG_VALIDATE(0 != shd, VALIDATE_PIPELINEDESC_SHADER);
        if (shd) {
            _SG_VALIDATE(shd->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_PIPELINEDESC_SHADER);
            if (desc->compute) {
                _SG_VALIDATE(shd->cmn.is_compute, VALIDATE_PIPELINEDESC_COMPUTE_SHADER_EXPECTED);
            } else {
                _SG_VALIDATE(!shd->cmn.is_compute, VALIDATE_PIPELINEDESC_NO_COMPUTE_SHADER_EXPECTED);
                bool attrs_cont = true;
                for (size_t attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
                    const sg_vertex_attr_state* a_state = &desc->layout.attrs[attr_index];
                    if (a_state->format == SG_VERTEXFORMAT_INVALID) {
                        attrs_cont = false;
                        continue;
                    }
                    _SG_VALIDATE(attrs_cont, VALIDATE_PIPELINEDESC_NO_CONT_ATTRS);
                    SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
                    // vertex format must match expected shader attribute base type (if provided)
                    if (shd->cmn.attrs[attr_index].base_type != SG_SHADERATTRBASETYPE_UNDEFINED) {
                        if (_sg_vertexformat_basetype(a_state->format) != shd->cmn.attrs[attr_index].base_type) {
                            _SG_VALIDATE(false, VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH);
                            _SG_LOGMSG(VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH, "attr format:");
                            _SG_LOGMSG(VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH, _sg_vertexformat_to_string(a_state->format));
                            _SG_LOGMSG(VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH, "shader attr base type:");
                            _SG_LOGMSG(VALIDATE_PIPELINEDESC_ATTR_BASETYPE_MISMATCH, _sg_shaderattrbasetype_to_string(shd->cmn.attrs[attr_index].base_type));
                        }
                    }
                    #if defined(SOKOL_D3D11)
                    // on D3D11, semantic names (and semantic indices) must be provided
                    _SG_VALIDATE(!_sg_strempty(&shd->d3d11.attrs[attr_index].sem_name), VALIDATE_PIPELINEDESC_ATTR_SEMANTICS);
                    #endif
                }
                // must only use readonly storage buffer bindings in render pipelines
                for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
                    if (shd->cmn.storage_buffers[i].stage != SG_SHADERSTAGE_NONE) {
                        _SG_VALIDATE(shd->cmn.storage_buffers[i].readonly, VALIDATE_PIPELINEDESC_SHADER_READONLY_STORAGEBUFFERS);
                    }
                }
                for (int buf_index = 0; buf_index < SG_MAX_VERTEXBUFFER_BINDSLOTS; buf_index++) {
                    const sg_vertex_buffer_layout_state* l_state = &desc->layout.buffers[buf_index];
                    if (l_state->stride == 0) {
                        continue;
                    }
                    _SG_VALIDATE(_sg_multiple_u64((uint64_t)l_state->stride, 4), VALIDATE_PIPELINEDESC_LAYOUT_STRIDE4);
                }
            }
        }
        for (size_t color_index = 0; color_index < (size_t)desc->color_count; color_index++) {
            const sg_blend_state* bs = &desc->colors[color_index].blend;
            if ((bs->op_rgb == SG_BLENDOP_MIN) || (bs->op_rgb == SG_BLENDOP_MAX)) {
                _SG_VALIDATE((bs->src_factor_rgb == SG_BLENDFACTOR_ONE) && (bs->dst_factor_rgb == SG_BLENDFACTOR_ONE), VALIDATE_PIPELINEDESC_BLENDOP_MINMAX_REQUIRES_BLENDFACTOR_ONE);
            }
            if ((bs->op_alpha == SG_BLENDOP_MIN) || (bs->op_alpha == SG_BLENDOP_MAX)) {
                _SG_VALIDATE((bs->src_factor_alpha == SG_BLENDFACTOR_ONE) && (bs->dst_factor_alpha == SG_BLENDFACTOR_ONE), VALIDATE_PIPELINEDESC_BLENDOP_MINMAX_REQUIRES_BLENDFACTOR_ONE);
            }
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_attachments_desc(const sg_attachments_desc* desc) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(desc);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(desc);
        _sg_validate_begin();
        _SG_VALIDATE(desc->_start_canary == 0, VALIDATE_ATTACHMENTSDESC_CANARY);
        _SG_VALIDATE(desc->_end_canary == 0, VALIDATE_ATTACHMENTSDESC_CANARY);

        // check color attachments
        bool has_color_atts = false;
        bool has_depth_stencil_att = false;
        {
            bool atts_cont = true;
            int color_width = -1, color_height = -1, color_sample_count = -1;
            for (int att_index = 0; att_index < SG_MAX_COLOR_ATTACHMENTS; att_index++) {
                const sg_attachment_desc* att = &desc->colors[att_index];
                if (att->image.id == SG_INVALID_ID) {
                    atts_cont = false;
                    continue;
                }
                has_color_atts = true;
                _SG_VALIDATE(atts_cont, VALIDATE_ATTACHMENTSDESC_NO_CONT_COLOR_ATTS);
                const _sg_image_t* img = _sg_lookup_image(att->image.id);
                _SG_VALIDATE(img, VALIDATE_ATTACHMENTSDESC_COLOR_IMAGE);
                if (0 != img) {
                    _SG_VALIDATE(img->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ATTACHMENTSDESC_COLOR_IMAGE);
                    _SG_VALIDATE(img->cmn.usage.render_attachment, VALIDATE_ATTACHMENTSDESC_COLOR_IMAGE_NO_RENDERATTACHMENT);
                    _SG_VALIDATE(att->mip_level < img->cmn.num_mipmaps, VALIDATE_ATTACHMENTSDESC_COLOR_MIPLEVEL);
                    if (img->cmn.type == SG_IMAGETYPE_CUBE) {
                        _SG_VALIDATE(att->slice < 6, VALIDATE_ATTACHMENTSDESC_COLOR_FACE);
                    } else if (img->cmn.type == SG_IMAGETYPE_ARRAY) {
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_COLOR_LAYER);
                    } else if (img->cmn.type == SG_IMAGETYPE_3D) {
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_COLOR_SLICE);
                    }
                    if (att_index == 0) {
                        color_width = _sg_miplevel_dim(img->cmn.width, att->mip_level);
                        color_height = _sg_miplevel_dim(img->cmn.height, att->mip_level);
                        color_sample_count = img->cmn.sample_count;
                    } else {
                        _SG_VALIDATE(color_width == _sg_miplevel_dim(img->cmn.width, att->mip_level), VALIDATE_ATTACHMENTSDESC_IMAGE_SIZES);
                        _SG_VALIDATE(color_height == _sg_miplevel_dim(img->cmn.height, att->mip_level), VALIDATE_ATTACHMENTSDESC_IMAGE_SIZES);
                        _SG_VALIDATE(color_sample_count == img->cmn.sample_count, VALIDATE_ATTACHMENTSDESC_IMAGE_SAMPLE_COUNTS);
                    }
                    _SG_VALIDATE(_sg_is_valid_attachment_color_format(img->cmn.pixel_format), VALIDATE_ATTACHMENTSDESC_COLOR_INV_PIXELFORMAT);

                    // check resolve attachment
                    const sg_attachment_desc* res_att = &desc->resolves[att_index];
                    if (res_att->image.id != SG_INVALID_ID) {
                        // associated color attachment must be MSAA
                        _SG_VALIDATE(img->cmn.sample_count > 1, VALIDATE_ATTACHMENTSDESC_RESOLVE_COLOR_IMAGE_MSAA);
                        const _sg_image_t* res_img = _sg_lookup_image(res_att->image.id);
                        _SG_VALIDATE(res_img, VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE);
                        if (res_img != 0) {
                            _SG_VALIDATE(res_img->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE);
                            _SG_VALIDATE(res_img->cmn.usage.render_attachment, VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_NO_RT);
                            _SG_VALIDATE(res_img->cmn.sample_count == 1, VALIDATE_ATTACHMENTSDESC_RESOLVE_SAMPLE_COUNT);
                            _SG_VALIDATE(res_att->mip_level < res_img->cmn.num_mipmaps, VALIDATE_ATTACHMENTSDESC_RESOLVE_MIPLEVEL);
                            if (res_img->cmn.type == SG_IMAGETYPE_CUBE) {
                                _SG_VALIDATE(res_att->slice < SG_CUBEFACE_NUM, VALIDATE_ATTACHMENTSDESC_RESOLVE_FACE);
                            } else if (res_img->cmn.type == SG_IMAGETYPE_ARRAY) {
                                _SG_VALIDATE(res_att->slice < res_img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_RESOLVE_LAYER);
                            } else if (res_img->cmn.type == SG_IMAGETYPE_3D) {
                                _SG_VALIDATE(res_att->slice < res_img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_RESOLVE_SLICE);
                            }
                            _SG_VALIDATE(img->cmn.pixel_format == res_img->cmn.pixel_format, VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_FORMAT);
                            _SG_VALIDATE(color_width == _sg_miplevel_dim(res_img->cmn.width, res_att->mip_level), VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_SIZES);
                            _SG_VALIDATE(color_height == _sg_miplevel_dim(res_img->cmn.height, res_att->mip_level), VALIDATE_ATTACHMENTSDESC_RESOLVE_IMAGE_SIZES);
                        }
                    }
                }
            }
            // check depth stencil attachments
            if (desc->depth_stencil.image.id != SG_INVALID_ID) {
                has_depth_stencil_att = true;
                const sg_attachment_desc* att = &desc->depth_stencil;
                const _sg_image_t* img = _sg_lookup_image(att->image.id);
                _SG_VALIDATE(img, VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE);
                if (img) {
                    _SG_VALIDATE(img->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE);
                    _SG_VALIDATE(att->mip_level < img->cmn.num_mipmaps, VALIDATE_ATTACHMENTSDESC_DEPTH_MIPLEVEL);
                    if (img->cmn.type == SG_IMAGETYPE_CUBE) {
                        _SG_VALIDATE(att->slice < 6, VALIDATE_ATTACHMENTSDESC_DEPTH_FACE);
                    } else if (img->cmn.type == SG_IMAGETYPE_ARRAY) {
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_DEPTH_LAYER);
                    } else if (img->cmn.type == SG_IMAGETYPE_3D) {
                        // NOTE: this can't actually happen because of VALIDATE_IMAGEDESC_DEPTH_3D_IMAGE
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_DEPTH_SLICE);
                    }
                    _SG_VALIDATE(img->cmn.usage.render_attachment, VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_NO_RENDERATTACHMENT);
                    _SG_VALIDATE((color_width == -1) || (color_width == _sg_miplevel_dim(img->cmn.width, att->mip_level)), VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_SIZES);
                    _SG_VALIDATE((color_height == -1) || (color_height == _sg_miplevel_dim(img->cmn.height, att->mip_level)), VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_SIZES);
                    _SG_VALIDATE((color_sample_count == -1) || (color_sample_count == img->cmn.sample_count), VALIDATE_ATTACHMENTSDESC_DEPTH_IMAGE_SAMPLE_COUNT);
                    _SG_VALIDATE(_sg_is_valid_attachment_depth_format(img->cmn.pixel_format), VALIDATE_ATTACHMENTSDESC_DEPTH_INV_PIXELFORMAT);
                }
            }
        }

        // check storage attachments (note: storage attachments don't need to continuous)
        bool has_storage_atts = false;
        {
            for (int att_index = 0; att_index < SG_MAX_STORAGE_ATTACHMENTS; att_index++) {
                const sg_attachment_desc* att = &desc->storages[att_index];
                if (att->image.id == SG_INVALID_ID) {
                    continue;
                }
                has_storage_atts = true;
                const _sg_image_t* img = _sg_lookup_image(att->image.id);
                _SG_VALIDATE(img, VALIDATE_ATTACHMENTSDESC_STORAGE_IMAGE);
                if (0 != img) {
                    _SG_VALIDATE(img->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ATTACHMENTSDESC_STORAGE_IMAGE);
                    _SG_VALIDATE(img->cmn.usage.storage_attachment, VALIDATE_ATTACHMENTSDESC_STORAGE_IMAGE_NO_STORAGEATTACHMENT);
                    _SG_VALIDATE(att->mip_level < img->cmn.num_mipmaps, VALIDATE_ATTACHMENTSDESC_STORAGE_MIPLEVEL);
                    if (img->cmn.type == SG_IMAGETYPE_CUBE) {
                        _SG_VALIDATE(att->slice < 6, VALIDATE_ATTACHMENTSDESC_STORAGE_FACE);
                    } else if (img->cmn.type == SG_IMAGETYPE_ARRAY) {
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_STORAGE_LAYER);
                    } else if (img->cmn.type == SG_IMAGETYPE_3D) {
                        _SG_VALIDATE(att->slice < img->cmn.num_slices, VALIDATE_ATTACHMENTSDESC_STORAGE_SLICE);
                    }
                    _SG_VALIDATE(_sg_is_valid_attachment_storage_format(img->cmn.pixel_format), VALIDATE_ATTACHMENTSDESC_STORAGE_INV_PIXELFORMAT);
                }
            }
        }
        _SG_VALIDATE(has_color_atts || has_depth_stencil_att || has_storage_atts, VALIDATE_ATTACHMENTSDESC_NO_ATTACHMENTS);
        if (has_color_atts || has_depth_stencil_att) {
            _SG_VALIDATE(!has_storage_atts, VALIDATE_ATTACHMENTSDESC_RENDER_VS_STORAGE_ATTACHMENTS);
        }
        if (has_storage_atts) {
            _SG_VALIDATE(!(has_color_atts || has_depth_stencil_att), VALIDATE_ATTACHMENTSDESC_RENDER_VS_STORAGE_ATTACHMENTS);
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_begin_pass(const sg_pass* pass) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(pass);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        const bool is_compute_pass = pass->compute;
        const bool is_swapchain_pass = !is_compute_pass && (pass->attachments.id == SG_INVALID_ID);
        const bool is_offscreen_pass = !(is_compute_pass || is_swapchain_pass);
        _sg_validate_begin();
        _SG_VALIDATE(pass->_start_canary == 0, VALIDATE_BEGINPASS_CANARY);
        _SG_VALIDATE(pass->_end_canary == 0, VALIDATE_BEGINPASS_CANARY);
        if (is_compute_pass) {
            // this is a compute pass with optional storage attachments
            if (pass->attachments.id) {
                const _sg_attachments_t* atts = _sg_lookup_attachments(pass->attachments.id);
                if (atts) {
                    _SG_VALIDATE(atts->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_ATTACHMENTS_VALID);
                    _SG_VALIDATE(!atts->cmn.has_render_attachments, VALIDATE_BEGINPASS_COMPUTEPASS_STORAGE_ATTACHMENTS_ONLY);
                    for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
                        const _sg_image_ref_t* storage_img_ref = &atts->cmn.storages[i].image;
                        if (!_sg_image_ref_null(storage_img_ref)) {
                            _SG_VALIDATE(_sg_image_ref_alive(storage_img_ref), VALIDATE_BEGINPASS_STORAGE_ATTACHMENT_IMAGE);
                            _SG_VALIDATE(_sg_image_ref_ptr(storage_img_ref)->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_STORAGE_ATTACHMENT_IMAGE);
                        }
                    }
                } else {
                    _SG_VALIDATE(atts != 0, VALIDATE_BEGINPASS_ATTACHMENTS_EXISTS);
                }
            }
        } else if (is_swapchain_pass) {
            // this is a swapchain pass
            _SG_VALIDATE(pass->swapchain.width > 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_WIDTH);
            _SG_VALIDATE(pass->swapchain.height > 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_HEIGHT);
            _SG_VALIDATE(pass->swapchain.sample_count > 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_SAMPLECOUNT);
            _SG_VALIDATE(pass->swapchain.color_format > SG_PIXELFORMAT_NONE, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_COLORFORMAT);
            // NOTE: depth buffer is optional, so depth_format is allowed to be invalid
            // NOTE: the GL framebuffer handle may actually be 0
            #if defined(SOKOL_METAL)
                _SG_VALIDATE(pass->swapchain.metal.current_drawable != 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_CURRENTDRAWABLE);
                if (pass->swapchain.depth_format == SG_PIXELFORMAT_NONE) {
                    _SG_VALIDATE(pass->swapchain.metal.depth_stencil_texture == 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_DEPTHSTENCILTEXTURE_NOTSET);
                } else {
                    _SG_VALIDATE(pass->swapchain.metal.depth_stencil_texture != 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_DEPTHSTENCILTEXTURE);
                }
                if (pass->swapchain.sample_count > 1) {
                    _SG_VALIDATE(pass->swapchain.metal.msaa_color_texture != 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_MSAACOLORTEXTURE);
                } else {
                    _SG_VALIDATE(pass->swapchain.metal.msaa_color_texture == 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_MSAACOLORTEXTURE_NOTSET);
                }
            #elif defined(SOKOL_D3D11)
                _SG_VALIDATE(pass->swapchain.d3d11.render_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RENDERVIEW);
                if (pass->swapchain.depth_format == SG_PIXELFORMAT_NONE) {
                    _SG_VALIDATE(pass->swapchain.d3d11.depth_stencil_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_DEPTHSTENCILVIEW_NOTSET);
                } else {
                    _SG_VALIDATE(pass->swapchain.d3d11.depth_stencil_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_DEPTHSTENCILVIEW);
                }
                if (pass->swapchain.sample_count > 1) {
                    _SG_VALIDATE(pass->swapchain.d3d11.resolve_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RESOLVEVIEW);
                } else {
                    _SG_VALIDATE(pass->swapchain.d3d11.resolve_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RESOLVEVIEW_NOTSET);
                }
            #elif defined(SOKOL_WGPU)
                _SG_VALIDATE(pass->swapchain.wgpu.render_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RENDERVIEW);
                if (pass->swapchain.depth_format == SG_PIXELFORMAT_NONE) {
                    _SG_VALIDATE(pass->swapchain.wgpu.depth_stencil_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_DEPTHSTENCILVIEW_NOTSET);
                } else {
                    _SG_VALIDATE(pass->swapchain.wgpu.depth_stencil_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_DEPTHSTENCILVIEW);
                }
                if (pass->swapchain.sample_count > 1) {
                    _SG_VALIDATE(pass->swapchain.wgpu.resolve_view != 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RESOLVEVIEW);
                } else {
                    _SG_VALIDATE(pass->swapchain.wgpu.resolve_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RESOLVEVIEW_NOTSET);
                }
            #endif
        } else {
            // this is an 'offscreen pass'
            const _sg_attachments_t* atts = _sg_lookup_attachments(pass->attachments.id);
            if (atts) {
                _SG_VALIDATE(atts->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_ATTACHMENTS_VALID);
                _SG_VALIDATE(!atts->cmn.has_storage_attachments, VALIDATE_BEGINPASS_RENDERPASS_RENDER_ATTACHMENTS_ONLY);
                for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
                    const _sg_image_ref_t* color_img_ref = &atts->cmn.colors[i].image;
                    if (!_sg_image_ref_null(color_img_ref)) {
                        _SG_VALIDATE(_sg_image_ref_alive(color_img_ref), VALIDATE_BEGINPASS_COLOR_ATTACHMENT_IMAGE);
                        _SG_VALIDATE(_sg_image_ref_ptr(color_img_ref)->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_COLOR_ATTACHMENT_IMAGE);
                    }
                    const _sg_image_ref_t* resolve_img_ref = &atts->cmn.resolves[i].image;
                    if (!_sg_image_ref_null(resolve_img_ref)) {
                        _SG_VALIDATE(_sg_image_ref_alive(resolve_img_ref), VALIDATE_BEGINPASS_RESOLVE_ATTACHMENT_IMAGE);
                        _SG_VALIDATE(_sg_image_ref_ptr(resolve_img_ref)->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_RESOLVE_ATTACHMENT_IMAGE);
                    }
                }
                const _sg_image_ref_t* ds_img_ref = &atts->cmn.depth_stencil.image;
                if (!_sg_image_ref_null(ds_img_ref)) {
                    _SG_VALIDATE(_sg_image_ref_alive(ds_img_ref), VALIDATE_BEGINPASS_DEPTHSTENCIL_ATTACHMENT_IMAGE);
                    _SG_VALIDATE(_sg_image_ref_ptr(ds_img_ref)->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_BEGINPASS_DEPTHSTENCIL_ATTACHMENT_IMAGE);
                }
            } else {
                _SG_VALIDATE(atts != 0, VALIDATE_BEGINPASS_ATTACHMENTS_EXISTS);
            }
        }
        if (is_compute_pass || is_offscreen_pass) {
            _SG_VALIDATE(pass->swapchain.width == 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_WIDTH_NOTSET);
            _SG_VALIDATE(pass->swapchain.height == 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_HEIGHT_NOTSET);
            _SG_VALIDATE(pass->swapchain.sample_count == 0, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_SAMPLECOUNT_NOTSET);
            _SG_VALIDATE(pass->swapchain.color_format == _SG_PIXELFORMAT_DEFAULT, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_COLORFORMAT_NOTSET);
            _SG_VALIDATE(pass->swapchain.depth_format == _SG_PIXELFORMAT_DEFAULT, VALIDATE_BEGINPASS_SWAPCHAIN_EXPECT_DEPTHFORMAT_NOTSET);
            #if defined(SOKOL_METAL)
                _SG_VALIDATE(pass->swapchain.metal.current_drawable == 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_CURRENTDRAWABLE_NOTSET);
                _SG_VALIDATE(pass->swapchain.metal.depth_stencil_texture == 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_DEPTHSTENCILTEXTURE_NOTSET);
                _SG_VALIDATE(pass->swapchain.metal.msaa_color_texture == 0, VALIDATE_BEGINPASS_SWAPCHAIN_METAL_EXPECT_MSAACOLORTEXTURE_NOTSET);
            #elif defined(SOKOL_D3D11)
                _SG_VALIDATE(pass->swapchain.d3d11.render_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RENDERVIEW_NOTSET);
                _SG_VALIDATE(pass->swapchain.d3d11.depth_stencil_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_DEPTHSTENCILVIEW_NOTSET);
                _SG_VALIDATE(pass->swapchain.d3d11.resolve_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_D3D11_EXPECT_RESOLVEVIEW_NOTSET);
            #elif defined(SOKOL_WGPU)
                _SG_VALIDATE(pass->swapchain.wgpu.render_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RENDERVIEW_NOTSET);
                _SG_VALIDATE(pass->swapchain.wgpu.depth_stencil_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_DEPTHSTENCILVIEW_NOTSET);
                _SG_VALIDATE(pass->swapchain.wgpu.resolve_view == 0, VALIDATE_BEGINPASS_SWAPCHAIN_WGPU_EXPECT_RESOLVEVIEW_NOTSET);
            #elif defined(_SOKOL_ANY_GL)
                _SG_VALIDATE(pass->swapchain.gl.framebuffer == 0, VALIDATE_BEGINPASS_SWAPCHAIN_GL_EXPECT_FRAMEBUFFER_NOTSET);
            #endif
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_apply_viewport(int x, int y, int width, int height, bool origin_top_left) {
    _SOKOL_UNUSED(x);
    _SOKOL_UNUSED(y);
    _SOKOL_UNUSED(width);
    _SOKOL_UNUSED(height);
    _SOKOL_UNUSED(origin_top_left);
    #if !defined(SOKOL_DEBUG)
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();
        _SG_VALIDATE(_sg.cur_pass.in_pass && !_sg.cur_pass.is_compute, VALIDATE_AVP_RENDERPASS_EXPECTED);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_apply_scissor_rect(int x, int y, int width, int height, bool origin_top_left) {
    _SOKOL_UNUSED(x);
    _SOKOL_UNUSED(y);
    _SOKOL_UNUSED(width);
    _SOKOL_UNUSED(height);
    _SOKOL_UNUSED(origin_top_left);
    #if !defined(SOKOL_DEBUG)
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();
        _SG_VALIDATE(_sg.cur_pass.in_pass && !_sg.cur_pass.is_compute, VALIDATE_ASR_RENDERPASS_EXPECTED);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_apply_pipeline(sg_pipeline pip_id) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(pip_id);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();
        // the pipeline object must be alive and valid
        _SG_VALIDATE(pip_id.id != SG_INVALID_ID, VALIDATE_APIP_PIPELINE_VALID_ID);
        const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
        _SG_VALIDATE(pip != 0, VALIDATE_APIP_PIPELINE_EXISTS);
        if (!pip) {
            return _sg_validate_end();
        }
        _SG_VALIDATE(pip->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_APIP_PIPELINE_VALID);
        // the pipeline's shader must be alive and valid
        _SG_VALIDATE(_sg.cur_pass.in_pass, VALIDATE_APIP_PASS_EXPECTED);
        _SG_VALIDATE(_sg_shader_ref_alive(&pip->cmn.shader), VALIDATE_APIP_PIPELINE_SHADER_ALIVE);
        const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
        _SG_VALIDATE(shd->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_APIP_PIPELINE_SHADER_VALID);

        // if pass attachments exist, check that the attachment object is still valid
        bool has_attachments = !_sg_attachments_ref_null(&_sg.cur_pass.atts);
        if (has_attachments) {
            _SG_VALIDATE(_sg_attachments_ref_alive(&_sg.cur_pass.atts), VALIDATE_APIP_CURPASS_ATTACHMENTS_ALIVE);
            const _sg_attachments_t* atts = _sg_attachments_ref_ptr(&_sg.cur_pass.atts);
            _SG_VALIDATE(atts->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_APIP_CURPASS_ATTACHMENTS_VALID);
        }
        if (pip->cmn.is_compute) {
            _SG_VALIDATE(_sg.cur_pass.is_compute, VALIDATE_APIP_COMPUTEPASS_EXPECTED);
            if (has_attachments) {
                const _sg_attachments_t* atts = _sg_attachments_ref_ptr(&_sg.cur_pass.atts);
                // a compute pass with storage attachments
                // check that the pass storage attachments match the shader expectations
                for (int i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
                    if (shd->cmn.storage_images[i].stage != SG_SHADERSTAGE_NONE) {
                        const _sg_image_ref_t* img_ref = &atts->cmn.storages[i].image;
                        bool img_null = _sg_image_ref_null(img_ref);
                        bool img_alive = _sg_image_ref_alive(img_ref);
                        _SG_VALIDATE(!img_null, VALIDATE_APIP_EXPECTED_STORAGE_ATTACHMENT_IMAGE);
                        if (!img_null) {
                            _SG_VALIDATE(img_alive, VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_ALIVE);
                            if (img_alive) {
                                const _sg_image_t* img = _sg_image_ref_ptr(img_ref);
                                _SG_VALIDATE(img->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_VALID);
                                _SG_VALIDATE(img->cmn.pixel_format == shd->cmn.storage_images[i].access_format, VALIDATE_APIP_STORAGE_ATTACHMENT_PIXELFORMAT);
                                _SG_VALIDATE(img->cmn.type == shd->cmn.storage_images[i].image_type, VALIDATE_APIP_STORAGE_ATTACHMENT_IMAGE_TYPE);
                            }
                        }
                    }
                }
            }
        } else {
            _SG_VALIDATE(!_sg.cur_pass.is_compute, VALIDATE_APIP_RENDERPASS_EXPECTED);
            // check that pipeline attributes match current pass attributes
            if (has_attachments) {
                const _sg_attachments_t* atts = _sg_attachments_ref_ptr(&_sg.cur_pass.atts);
                // an offscreen pass
                _SG_VALIDATE(pip->cmn.color_count == atts->cmn.num_colors, VALIDATE_APIP_ATT_COUNT);
                for (int i = 0; i < pip->cmn.color_count; i++) {
                    const _sg_image_t* att_img = _sg_image_ref_ptr(&atts->cmn.colors[i].image);
                    _SG_VALIDATE(pip->cmn.colors[i].pixel_format == att_img->cmn.pixel_format, VALIDATE_APIP_COLOR_FORMAT);
                    _SG_VALIDATE(pip->cmn.sample_count == att_img->cmn.sample_count, VALIDATE_APIP_SAMPLE_COUNT);
                }
                if (!_sg_image_ref_null(&atts->cmn.depth_stencil.image)) {
                    _SG_VALIDATE(pip->cmn.depth.pixel_format == _sg_image_ref_ptr(&atts->cmn.depth_stencil.image)->cmn.pixel_format, VALIDATE_APIP_DEPTH_FORMAT);
                } else {
                    _SG_VALIDATE(pip->cmn.depth.pixel_format == SG_PIXELFORMAT_NONE, VALIDATE_APIP_DEPTH_FORMAT);
                }
            } else {
                // default pass
                _SG_VALIDATE(pip->cmn.color_count == 1, VALIDATE_APIP_ATT_COUNT);
                _SG_VALIDATE(pip->cmn.colors[0].pixel_format == _sg.cur_pass.swapchain.color_fmt, VALIDATE_APIP_COLOR_FORMAT);
                _SG_VALIDATE(pip->cmn.depth.pixel_format == _sg.cur_pass.swapchain.depth_fmt, VALIDATE_APIP_DEPTH_FORMAT);
                _SG_VALIDATE(pip->cmn.sample_count == _sg.cur_pass.swapchain.sample_count, VALIDATE_APIP_SAMPLE_COUNT);
            }
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_apply_bindings(const sg_bindings* bindings) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(bindings);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();

        // must be called in a pass
        _SG_VALIDATE(_sg.cur_pass.in_pass, VALIDATE_ABND_PASS_EXPECTED);

        // bindings must not be empty
        bool has_any_bindings = bindings->index_buffer.id != SG_INVALID_ID;
        if (!has_any_bindings) for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
            has_any_bindings |= bindings->vertex_buffers[i].id != SG_INVALID_ID;
        }
        if (!has_any_bindings) for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
            has_any_bindings |= bindings->images[i].id != SG_INVALID_ID;
        }
        if (!has_any_bindings) for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
            has_any_bindings |= bindings->samplers[i].id != SG_INVALID_ID;
        }
        if (!has_any_bindings) for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
            has_any_bindings |= bindings->storage_buffers[i].id != SG_INVALID_ID;
        }
        _SG_VALIDATE(has_any_bindings, VALIDATE_ABND_EMPTY_BINDINGS);

        // a pipeline object must have been applied
        _SG_VALIDATE(!_sg_pipeline_ref_null(&_sg.cur_pip), VALIDATE_ABND_NO_PIPELINE);
        const bool pip_alive = _sg_pipeline_ref_alive(&_sg.cur_pip);
        _SG_VALIDATE(pip_alive, VALIDATE_ABND_PIPELINE_ALIVE);
        if (!pip_alive) {
            return _sg_validate_end();
        }
        const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
        _SG_VALIDATE(pip->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ABND_PIPELINE_VALID);
        _SG_VALIDATE(_sg_shader_ref_alive(&pip->cmn.shader), VALIDATE_ABND_PIPELINE_SHADER_ALIVE);
        const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
        _SG_VALIDATE(shd->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_ABND_PIPELINE_SHADER_VALID);

        if (_sg.cur_pass.is_compute) {
            for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
                _SG_VALIDATE(bindings->vertex_buffers[i].id == SG_INVALID_ID, VALIDATE_ABND_COMPUTE_EXPECTED_NO_VBS);
            }
        } else {
            // has expected vertex buffers, and vertex buffers still exist
            for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
                if (pip->cmn.vertex_buffer_layout_active[i]) {
                    _SG_VALIDATE(bindings->vertex_buffers[i].id != SG_INVALID_ID, VALIDATE_ABND_EXPECTED_VB);
                    // buffers in vertex-buffer-slots must have vertex buffer usage
                    if (bindings->vertex_buffers[i].id != SG_INVALID_ID) {
                        const _sg_buffer_t* buf = _sg_lookup_buffer(bindings->vertex_buffers[i].id);
                        _SG_VALIDATE(buf != 0, VALIDATE_ABND_VB_ALIVE);
                        if (buf && buf->slot.state == SG_RESOURCESTATE_VALID) {
                            _SG_VALIDATE(buf->cmn.usage.vertex_buffer, VALIDATE_ABND_VB_TYPE);
                            _SG_VALIDATE(!buf->cmn.append_overflow, VALIDATE_ABND_VB_OVERFLOW);
                        }
                    }
                }
            }
        }

        if (_sg.cur_pass.is_compute) {
            _SG_VALIDATE(bindings->index_buffer.id == SG_INVALID_ID, VALIDATE_ABND_COMPUTE_EXPECTED_NO_IB);
        } else {
            // index buffer expected or not, and index buffer still exists
            if (pip->cmn.index_type == SG_INDEXTYPE_NONE) {
                // pipeline defines non-indexed rendering, but index buffer provided
                _SG_VALIDATE(bindings->index_buffer.id == SG_INVALID_ID, VALIDATE_ABND_IB);
            } else {
                // pipeline defines indexed rendering, but no index buffer provided
                _SG_VALIDATE(bindings->index_buffer.id != SG_INVALID_ID, VALIDATE_ABND_NO_IB);
            }
            if (bindings->index_buffer.id != SG_INVALID_ID) {
                // buffer in index-buffer-slot must have index buffer usage
                const _sg_buffer_t* buf = _sg_lookup_buffer(bindings->index_buffer.id);
                _SG_VALIDATE(buf != 0, VALIDATE_ABND_IB_ALIVE);
                if (buf && buf->slot.state == SG_RESOURCESTATE_VALID) {
                    _SG_VALIDATE(buf->cmn.usage.index_buffer, VALIDATE_ABND_IB_TYPE);
                    _SG_VALIDATE(!buf->cmn.append_overflow, VALIDATE_ABND_IB_OVERFLOW);
                }
            }
        }

        // has expected images
        for (size_t i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
            if (shd->cmn.images[i].stage != SG_SHADERSTAGE_NONE) {
                _SG_VALIDATE(bindings->images[i].id != SG_INVALID_ID, VALIDATE_ABND_EXPECTED_IMAGE_BINDING);
                if (bindings->images[i].id != SG_INVALID_ID) {
                    const _sg_image_t* img = _sg_lookup_image(bindings->images[i].id);
                    _SG_VALIDATE(img != 0, VALIDATE_ABND_IMG_ALIVE);
                    if (img && img->slot.state == SG_RESOURCESTATE_VALID) {
                        _SG_VALIDATE(img->cmn.type == shd->cmn.images[i].image_type, VALIDATE_ABND_IMAGE_TYPE_MISMATCH);
                        if (!_sg.features.msaa_image_bindings) {
                            _SG_VALIDATE(img->cmn.sample_count == 1, VALIDATE_ABND_IMAGE_MSAA);
                        }
                        if (shd->cmn.images[i].multisampled) {
                            _SG_VALIDATE(img->cmn.sample_count > 1, VALIDATE_ABND_EXPECTED_MULTISAMPLED_IMAGE);
                        }
                        const _sg_pixelformat_info_t* info = &_sg.formats[img->cmn.pixel_format];
                        switch (shd->cmn.images[i].sample_type) {
                            case SG_IMAGESAMPLETYPE_FLOAT:
                                _SG_VALIDATE(info->filter, VALIDATE_ABND_EXPECTED_FILTERABLE_IMAGE);
                                break;
                            case SG_IMAGESAMPLETYPE_DEPTH:
                                _SG_VALIDATE(info->depth, VALIDATE_ABND_EXPECTED_DEPTH_IMAGE);
                                break;
                            default:
                                break;
                        }
                    }
                }
            }
        }

        // has expected samplers
        for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
            if (shd->cmn.samplers[i].stage != SG_SHADERSTAGE_NONE) {
                _SG_VALIDATE(bindings->samplers[i].id != SG_INVALID_ID, VALIDATE_ABND_EXPECTED_SAMPLER_BINDING);
                if (bindings->samplers[i].id != SG_INVALID_ID) {
                    const _sg_sampler_t* smp = _sg_lookup_sampler(bindings->samplers[i].id);
                    _SG_VALIDATE(smp != 0, VALIDATE_ABND_SMP_ALIVE);
                    if (smp) {
                        if (shd->cmn.samplers[i].sampler_type == SG_SAMPLERTYPE_COMPARISON) {
                            _SG_VALIDATE(smp->cmn.compare != SG_COMPAREFUNC_NEVER, VALIDATE_ABND_UNEXPECTED_SAMPLER_COMPARE_NEVER);
                        } else {
                            _SG_VALIDATE(smp->cmn.compare == SG_COMPAREFUNC_NEVER, VALIDATE_ABND_EXPECTED_SAMPLER_COMPARE_NEVER);
                        }
                        if (shd->cmn.samplers[i].sampler_type == SG_SAMPLERTYPE_NONFILTERING) {
                            const bool nonfiltering = (smp->cmn.min_filter != SG_FILTER_LINEAR)
                                                   && (smp->cmn.mag_filter != SG_FILTER_LINEAR)
                                                   && (smp->cmn.mipmap_filter != SG_FILTER_LINEAR);
                            _SG_VALIDATE(nonfiltering, VALIDATE_ABND_EXPECTED_NONFILTERING_SAMPLER);
                        }
                    }
                }
            }
        }

        // has expected storage buffers
        for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
            if (shd->cmn.storage_buffers[i].stage != SG_SHADERSTAGE_NONE) {
                _SG_VALIDATE(bindings->storage_buffers[i].id != SG_INVALID_ID, VALIDATE_ABND_EXPECTED_STORAGEBUFFER_BINDING);
                if (bindings->storage_buffers[i].id != SG_INVALID_ID) {
                    const _sg_buffer_t* sbuf = _sg_lookup_buffer(bindings->storage_buffers[i].id);
                    _SG_VALIDATE(sbuf != 0, VALIDATE_ABND_STORAGEBUFFER_ALIVE);
                    if (sbuf) {
                        _SG_VALIDATE(sbuf->cmn.usage.storage_buffer, VALIDATE_ABND_STORAGEBUFFER_BINDING_BUFFERTYPE);
                        // read/write bindings are only allowed for immutable buffers
                        if (!shd->cmn.storage_buffers[i].readonly) {
                            _SG_VALIDATE(sbuf->cmn.usage.immutable, VALIDATE_ABND_STORAGEBUFFER_READWRITE_IMMUTABLE);
                        }
                    }
                }
            }
        }

        // the same image cannot be bound as texture and pass attachment
        if (!_sg_attachments_ref_null(&_sg.cur_pass.atts)) {
            const _sg_attachments_t* atts = _sg_attachments_ref_ptr(&_sg.cur_pass.atts);
            for (size_t img_idx = 0; img_idx < SG_MAX_IMAGE_BINDSLOTS; img_idx++) {
                if (shd->cmn.images[img_idx].stage != SG_SHADERSTAGE_NONE) {
                    const uint32_t img_id = bindings->images[img_idx].id;
                    if (img_id == SG_INVALID_ID) {
                        continue;
                    }
                    _SG_VALIDATE(img_id != atts->cmn.depth_stencil.image.sref.id, VALIDATE_ABND_IMAGE_BINDING_VS_DEPTHSTENCIL_ATTACHMENT);
                    for (size_t att_idx = 0; att_idx < SG_MAX_COLOR_ATTACHMENTS; att_idx++) {
                        _SG_VALIDATE(img_id != atts->cmn.colors[att_idx].image.sref.id, VALIDATE_ABND_IMAGE_BINDING_VS_COLOR_ATTACHMENT);
                        _SG_VALIDATE(img_id != atts->cmn.resolves[att_idx].image.sref.id, VALIDATE_ABND_IMAGE_BINDING_VS_RESOLVE_ATTACHMENT);
                    }
                    for (size_t att_idx = 0; att_idx < SG_MAX_STORAGE_ATTACHMENTS; att_idx++) {
                        _SG_VALIDATE(img_id != atts->cmn.storages[att_idx].image.sref.id, VALIDATE_ABND_IMAGE_BINDING_VS_STORAGE_ATTACHMENT);
                    }
                }
            }
        }
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_apply_uniforms(int ub_slot, const sg_range* data) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(ub_slot);
        _SOKOL_UNUSED(data);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
        _sg_validate_begin();
        _SG_VALIDATE(_sg.cur_pass.in_pass, VALIDATE_AU_PASS_EXPECTED);
        _SG_VALIDATE(!_sg_pipeline_ref_null(&_sg.cur_pip), VALIDATE_AU_NO_PIPELINE);
        _SG_VALIDATE(_sg_pipeline_ref_alive(&_sg.cur_pip), VALIDATE_AU_PIPELINE_ALIVE);
        const _sg_pipeline_t* pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
        _SG_VALIDATE(pip->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_AU_PIPELINE_VALID);
        _SG_VALIDATE(_sg_shader_ref_alive(&pip->cmn.shader), VALIDATE_AU_PIPELINE_SHADER_ALIVE);
        const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
        _SG_VALIDATE(shd->slot.state == SG_RESOURCESTATE_VALID, VALIDATE_AU_PIPELINE_SHADER_VALID);
        _SG_VALIDATE(shd->cmn.uniform_blocks[ub_slot].stage != SG_SHADERSTAGE_NONE, VALIDATE_AU_NO_UNIFORMBLOCK_AT_SLOT);
        _SG_VALIDATE(data->size == shd->cmn.uniform_blocks[ub_slot].size, VALIDATE_AU_SIZE);

        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_draw(int base_element, int num_elements, int num_instances) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(base_element);
        _SOKOL_UNUSED(num_elements);
        _SOKOL_UNUSED(num_instances);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();
        _SG_VALIDATE(_sg.cur_pass.in_pass && !_sg.cur_pass.is_compute, VALIDATE_DRAW_RENDERPASS_EXPECTED);
        _SG_VALIDATE(base_element >= 0, VALIDATE_DRAW_BASEELEMENT);
        _SG_VALIDATE(num_elements >= 0, VALIDATE_DRAW_NUMELEMENTS);
        _SG_VALIDATE(num_instances >= 0, VALIDATE_DRAW_NUMINSTANCES);
        _SG_VALIDATE(_sg.required_bindings_and_uniforms == _sg.applied_bindings_and_uniforms, VALIDATE_DRAW_REQUIRED_BINDINGS_OR_UNIFORMS_MISSING);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(num_groups_x);
        _SOKOL_UNUSED(num_groups_y);
        _SOKOL_UNUSED(num_groups_z);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        _sg_validate_begin();
        _SG_VALIDATE(_sg.cur_pass.in_pass && _sg.cur_pass.is_compute, VALIDATE_DISPATCH_COMPUTEPASS_EXPECTED);
        _SG_VALIDATE((num_groups_x >= 0) && (num_groups_x < (1<<16)), VALIDATE_DISPATCH_NUMGROUPSX);
        _SG_VALIDATE((num_groups_y >= 0) && (num_groups_y < (1<<16)), VALIDATE_DISPATCH_NUMGROUPSY);
        _SG_VALIDATE((num_groups_z >= 0) && (num_groups_z < (1<<16)), VALIDATE_DISPATCH_NUMGROUPSZ);
        _SG_VALIDATE(_sg.required_bindings_and_uniforms == _sg.applied_bindings_and_uniforms, VALIDATE_DRAW_REQUIRED_BINDINGS_OR_UNIFORMS_MISSING);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_update_buffer(const _sg_buffer_t* buf, const sg_range* data) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(buf);
        _SOKOL_UNUSED(data);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(buf && data && data->ptr);
        _sg_validate_begin();
        _SG_VALIDATE(!buf->cmn.usage.immutable, VALIDATE_UPDATEBUF_USAGE);
        _SG_VALIDATE(buf->cmn.size >= (int)data->size, VALIDATE_UPDATEBUF_SIZE);
        _SG_VALIDATE(buf->cmn.update_frame_index != _sg.frame_index, VALIDATE_UPDATEBUF_ONCE);
        _SG_VALIDATE(buf->cmn.append_frame_index != _sg.frame_index, VALIDATE_UPDATEBUF_APPEND);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_append_buffer(const _sg_buffer_t* buf, const sg_range* data) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(buf);
        _SOKOL_UNUSED(data);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(buf && data && data->ptr);
        _sg_validate_begin();
        _SG_VALIDATE(!buf->cmn.usage.immutable, VALIDATE_APPENDBUF_USAGE);
        _SG_VALIDATE(buf->cmn.size >= (buf->cmn.append_pos + (int)data->size), VALIDATE_APPENDBUF_SIZE);
        _SG_VALIDATE(buf->cmn.update_frame_index != _sg.frame_index, VALIDATE_APPENDBUF_UPDATE);
        return _sg_validate_end();
    #endif
}

_SOKOL_PRIVATE bool _sg_validate_update_image(const _sg_image_t* img, const sg_image_data* data) {
    #if !defined(SOKOL_DEBUG)
        _SOKOL_UNUSED(img);
        _SOKOL_UNUSED(data);
        return true;
    #else
        if (_sg.desc.disable_validation) {
            return true;
        }
        SOKOL_ASSERT(img && data);
        _sg_validate_begin();
        _SG_VALIDATE(!img->cmn.usage.immutable, VALIDATE_UPDIMG_USAGE);
        _SG_VALIDATE(img->cmn.upd_frame_index != _sg.frame_index, VALIDATE_UPDIMG_ONCE);
        _sg_validate_image_data(data,
            img->cmn.pixel_format,
            img->cmn.width,
            img->cmn.height,
            (img->cmn.type == SG_IMAGETYPE_CUBE) ? 6 : 1,
            img->cmn.num_mipmaps,
            img->cmn.num_slices);
        return _sg_validate_end();
    #endif
}

// ██████  ███████ ███████  ██████  ██    ██ ██████   ██████ ███████ ███████
// ██   ██ ██      ██      ██    ██ ██    ██ ██   ██ ██      ██      ██
// ██████  █████   ███████ ██    ██ ██    ██ ██████  ██      █████   ███████
// ██   ██ ██           ██ ██    ██ ██    ██ ██   ██ ██      ██           ██
// ██   ██ ███████ ███████  ██████   ██████  ██   ██  ██████ ███████ ███████
//
// >>resources
_SOKOL_PRIVATE sg_buffer_usage _sg_buffer_usage_defaults(const sg_buffer_usage* usg) {
    sg_buffer_usage def = *usg;
    if (!(def.vertex_buffer || def.index_buffer || def.storage_buffer)) {
        def.vertex_buffer = true;
    }
    if (!(def.immutable || def.stream_update || def.dynamic_update)) {
        def.immutable = true;
    }
    return def;
}


_SOKOL_PRIVATE sg_buffer_desc _sg_buffer_desc_defaults(const sg_buffer_desc* desc) {
    sg_buffer_desc def = *desc;
    def.usage = _sg_buffer_usage_defaults(&def.usage);
    if (def.size == 0) {
        def.size = def.data.size;
    }
    return def;
}

_SOKOL_PRIVATE sg_image_usage _sg_image_usage_defaults(const sg_image_usage *usg) {
    sg_image_usage def = *usg;
    if (!(def.immutable || def.stream_update || def.dynamic_update)) {
        def.immutable = true;
    }
    return def;
}

_SOKOL_PRIVATE sg_image_desc _sg_image_desc_defaults(const sg_image_desc* desc) {
    sg_image_desc def = *desc;
    def.type = _sg_def(def.type, SG_IMAGETYPE_2D);
    def.usage = _sg_image_usage_defaults(&def.usage);
    def.num_slices = _sg_def(def.num_slices, 1);
    def.num_mipmaps = _sg_def(def.num_mipmaps, 1);
    if (def.usage.render_attachment) {
        def.pixel_format = _sg_def(def.pixel_format, _sg.desc.environment.defaults.color_format);
        def.sample_count = _sg_def(def.sample_count, _sg.desc.environment.defaults.sample_count);
    } else {
        def.pixel_format = _sg_def(def.pixel_format, SG_PIXELFORMAT_RGBA8);
        def.sample_count = _sg_def(def.sample_count, 1);
    }
    return def;
}

_SOKOL_PRIVATE sg_sampler_desc _sg_sampler_desc_defaults(const sg_sampler_desc* desc) {
    sg_sampler_desc def = *desc;
    def.min_filter = _sg_def(def.min_filter, SG_FILTER_NEAREST);
    def.mag_filter = _sg_def(def.mag_filter, SG_FILTER_NEAREST);
    def.mipmap_filter = _sg_def(def.mipmap_filter, SG_FILTER_NEAREST);
    def.wrap_u = _sg_def(def.wrap_u, SG_WRAP_REPEAT);
    def.wrap_v = _sg_def(def.wrap_v, SG_WRAP_REPEAT);
    def.wrap_w = _sg_def(def.wrap_w, SG_WRAP_REPEAT);
    def.max_lod = _sg_def_flt(def.max_lod, FLT_MAX);
    def.border_color = _sg_def(def.border_color, SG_BORDERCOLOR_OPAQUE_BLACK);
    def.compare = _sg_def(def.compare, SG_COMPAREFUNC_NEVER);
    def.max_anisotropy = _sg_def(def.max_anisotropy, 1);
    return def;
}

_SOKOL_PRIVATE sg_shader_desc _sg_shader_desc_defaults(const sg_shader_desc* desc) {
    sg_shader_desc def = *desc;
    #if defined(SOKOL_METAL)
        def.vertex_func.entry = _sg_def(def.vertex_func.entry, "_main");
        def.fragment_func.entry = _sg_def(def.fragment_func.entry, "_main");
        def.compute_func.entry = _sg_def(def.compute_func.entry, "_main");
    #else
        def.vertex_func.entry = _sg_def(def.vertex_func.entry, "main");
        def.fragment_func.entry = _sg_def(def.fragment_func.entry, "main");
        def.compute_func.entry = _sg_def(def.compute_func.entry, "main");
    #endif
    #if defined(SOKOL_D3D11)
        if (def.vertex_func.source) {
            def.vertex_func.d3d11_target = _sg_def(def.vertex_func.d3d11_target, "vs_4_0");
        }
        if (def.fragment_func.source) {
            def.fragment_func.d3d11_target = _sg_def(def.fragment_func.d3d11_target, "ps_4_0");
        }
        if (def.compute_func.source) {
            def.compute_func.d3d11_target = _sg_def(def.fragment_func.d3d11_target,"cs_5_0");
        }
    #endif
    def.mtl_threads_per_threadgroup.y = _sg_def(desc->mtl_threads_per_threadgroup.y, 1);
    def.mtl_threads_per_threadgroup.z = _sg_def(desc->mtl_threads_per_threadgroup.z, 1);
    for (size_t ub_index = 0; ub_index < SG_MAX_UNIFORMBLOCK_BINDSLOTS; ub_index++) {
        sg_shader_uniform_block* ub_desc = &def.uniform_blocks[ub_index];
        if (ub_desc->stage != SG_SHADERSTAGE_NONE) {
            ub_desc->layout = _sg_def(ub_desc->layout, SG_UNIFORMLAYOUT_NATIVE);
            for (size_t u_index = 0; u_index < SG_MAX_UNIFORMBLOCK_MEMBERS; u_index++) {
                sg_glsl_shader_uniform* u_desc = &ub_desc->glsl_uniforms[u_index];
                if (u_desc->type == SG_UNIFORMTYPE_INVALID) {
                    break;
                }
                u_desc->array_count = _sg_def(u_desc->array_count, 1);
            }
        }
    }
    for (size_t img_index = 0; img_index < SG_MAX_IMAGE_BINDSLOTS; img_index++) {
        sg_shader_image* img_desc = &def.images[img_index];
        if (img_desc->stage != SG_SHADERSTAGE_NONE) {
            img_desc->image_type = _sg_def(img_desc->image_type, SG_IMAGETYPE_2D);
            img_desc->sample_type = _sg_def(img_desc->sample_type, SG_IMAGESAMPLETYPE_FLOAT);
        }
    }
    for (size_t smp_index = 0; smp_index < SG_MAX_SAMPLER_BINDSLOTS; smp_index++) {
        sg_shader_sampler* smp_desc = &def.samplers[smp_index];
        if (smp_desc->stage != SG_SHADERSTAGE_NONE) {
            smp_desc->sampler_type = _sg_def(smp_desc->sampler_type, SG_SAMPLERTYPE_FILTERING);
        }
    }
    return def;
}

_SOKOL_PRIVATE sg_pipeline_desc _sg_pipeline_desc_defaults(const sg_pipeline_desc* desc) {
    sg_pipeline_desc def = *desc;

    // FIXME: should we actually do all this stuff for a compute pipeline?

    def.primitive_type = _sg_def(def.primitive_type, SG_PRIMITIVETYPE_TRIANGLES);
    def.index_type = _sg_def(def.index_type, SG_INDEXTYPE_NONE);
    def.cull_mode = _sg_def(def.cull_mode, SG_CULLMODE_NONE);
    def.face_winding = _sg_def(def.face_winding, SG_FACEWINDING_CW);
    def.sample_count = _sg_def(def.sample_count, _sg.desc.environment.defaults.sample_count);

    def.stencil.front.compare = _sg_def(def.stencil.front.compare, SG_COMPAREFUNC_ALWAYS);
    def.stencil.front.fail_op = _sg_def(def.stencil.front.fail_op, SG_STENCILOP_KEEP);
    def.stencil.front.depth_fail_op = _sg_def(def.stencil.front.depth_fail_op, SG_STENCILOP_KEEP);
    def.stencil.front.pass_op = _sg_def(def.stencil.front.pass_op, SG_STENCILOP_KEEP);
    def.stencil.back.compare = _sg_def(def.stencil.back.compare, SG_COMPAREFUNC_ALWAYS);
    def.stencil.back.fail_op = _sg_def(def.stencil.back.fail_op, SG_STENCILOP_KEEP);
    def.stencil.back.depth_fail_op = _sg_def(def.stencil.back.depth_fail_op, SG_STENCILOP_KEEP);
    def.stencil.back.pass_op = _sg_def(def.stencil.back.pass_op, SG_STENCILOP_KEEP);

    def.depth.compare = _sg_def(def.depth.compare, SG_COMPAREFUNC_ALWAYS);
    def.depth.pixel_format = _sg_def(def.depth.pixel_format, _sg.desc.environment.defaults.depth_format);
    if (def.colors[0].pixel_format == SG_PIXELFORMAT_NONE) {
        // special case depth-only rendering, enforce a color count of 0
        def.color_count = 0;
    } else {
        def.color_count = _sg_def(def.color_count, 1);
    }
    if (def.color_count > SG_MAX_COLOR_ATTACHMENTS) {
        def.color_count = SG_MAX_COLOR_ATTACHMENTS;
    }
    for (int i = 0; i < def.color_count; i++) {
        sg_color_target_state* cs = &def.colors[i];
        cs->pixel_format = _sg_def(cs->pixel_format, _sg.desc.environment.defaults.color_format);
        cs->write_mask = _sg_def(cs->write_mask, SG_COLORMASK_RGBA);
        sg_blend_state* bs = &def.colors[i].blend;
        bs->op_rgb = _sg_def(bs->op_rgb, SG_BLENDOP_ADD);
        bs->src_factor_rgb = _sg_def(bs->src_factor_rgb, SG_BLENDFACTOR_ONE);
        if ((bs->op_rgb == SG_BLENDOP_MIN) || (bs->op_rgb == SG_BLENDOP_MAX)) {
            bs->dst_factor_rgb = _sg_def(bs->dst_factor_rgb, SG_BLENDFACTOR_ONE);
        } else {
            bs->dst_factor_rgb = _sg_def(bs->dst_factor_rgb, SG_BLENDFACTOR_ZERO);
        }
        bs->op_alpha = _sg_def(bs->op_alpha, SG_BLENDOP_ADD);
        bs->src_factor_alpha = _sg_def(bs->src_factor_alpha, SG_BLENDFACTOR_ONE);
        if ((bs->op_alpha == SG_BLENDOP_MIN) || (bs->op_alpha == SG_BLENDOP_MAX)) {
            bs->dst_factor_alpha = _sg_def(bs->dst_factor_alpha, SG_BLENDFACTOR_ONE);
        } else {
            bs->dst_factor_alpha = _sg_def(bs->dst_factor_alpha, SG_BLENDFACTOR_ZERO);
        }
    }

    for (int attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
        sg_vertex_attr_state* a_state = &def.layout.attrs[attr_index];
        if (a_state->format == SG_VERTEXFORMAT_INVALID) {
            break;
        }
        SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
        sg_vertex_buffer_layout_state* l_state = &def.layout.buffers[a_state->buffer_index];
        l_state->step_func = _sg_def(l_state->step_func, SG_VERTEXSTEP_PER_VERTEX);
        l_state->step_rate = _sg_def(l_state->step_rate, 1);
    }

    // resolve vertex layout strides and offsets
    int auto_offset[SG_MAX_VERTEXBUFFER_BINDSLOTS];
    _sg_clear(auto_offset, sizeof(auto_offset));
    bool use_auto_offset = true;
    for (int attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
        // to use computed offsets, *all* attr offsets must be 0
        if (def.layout.attrs[attr_index].offset != 0) {
            use_auto_offset = false;
        }
    }
    for (int attr_index = 0; attr_index < SG_MAX_VERTEX_ATTRIBUTES; attr_index++) {
        sg_vertex_attr_state* a_state = &def.layout.attrs[attr_index];
        if (a_state->format == SG_VERTEXFORMAT_INVALID) {
            break;
        }
        SOKOL_ASSERT(a_state->buffer_index < SG_MAX_VERTEXBUFFER_BINDSLOTS);
        if (use_auto_offset) {
            a_state->offset = auto_offset[a_state->buffer_index];
        }
        auto_offset[a_state->buffer_index] += _sg_vertexformat_bytesize(a_state->format);
    }
    // compute vertex strides if needed
    for (int buf_index = 0; buf_index < SG_MAX_VERTEXBUFFER_BINDSLOTS; buf_index++) {
        sg_vertex_buffer_layout_state* l_state = &def.layout.buffers[buf_index];
        if (l_state->stride == 0) {
            l_state->stride = auto_offset[buf_index];
        }
    }

    return def;
}

_SOKOL_PRIVATE sg_attachments_desc _sg_attachments_desc_defaults(const sg_attachments_desc* desc) {
    sg_attachments_desc def = *desc;
    return def;
}

_SOKOL_PRIVATE sg_buffer _sg_alloc_buffer(void) {
    sg_buffer res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.buffer_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&_sg.pools.buffer_pool, &_sg.pools.buffers[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(BUFFER_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE sg_image _sg_alloc_image(void) {
    sg_image res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.image_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&_sg.pools.image_pool, &_sg.pools.images[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(IMAGE_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE sg_sampler _sg_alloc_sampler(void) {
    sg_sampler res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.sampler_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&_sg.pools.sampler_pool, &_sg.pools.samplers[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(SAMPLER_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE sg_shader _sg_alloc_shader(void) {
    sg_shader res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.shader_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&_sg.pools.shader_pool, &_sg.pools.shaders[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(SHADER_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE sg_pipeline _sg_alloc_pipeline(void) {
    sg_pipeline res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.pipeline_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id =_sg_slot_alloc(&_sg.pools.pipeline_pool, &_sg.pools.pipelines[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(PIPELINE_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE sg_attachments _sg_alloc_attachments(void) {
    sg_attachments res;
    int slot_index = _sg_pool_alloc_index(&_sg.pools.attachments_pool);
    if (_SG_INVALID_SLOT_INDEX != slot_index) {
        res.id = _sg_slot_alloc(&_sg.pools.attachments_pool, &_sg.pools.attachments[slot_index].slot, slot_index);
    } else {
        res.id = SG_INVALID_ID;
        _SG_ERROR(PASS_POOL_EXHAUSTED);
    }
    return res;
}

_SOKOL_PRIVATE void _sg_dealloc_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf && (buf->slot.state == SG_RESOURCESTATE_ALLOC) && (buf->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.buffer_pool, _sg_slot_index(buf->slot.id));
    _sg_slot_reset(&buf->slot);
}

_SOKOL_PRIVATE void _sg_dealloc_image(_sg_image_t* img) {
    SOKOL_ASSERT(img && (img->slot.state == SG_RESOURCESTATE_ALLOC) && (img->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.image_pool, _sg_slot_index(img->slot.id));
    _sg_slot_reset(&img->slot);
}

_SOKOL_PRIVATE void _sg_dealloc_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp && (smp->slot.state == SG_RESOURCESTATE_ALLOC) && (smp->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.sampler_pool, _sg_slot_index(smp->slot.id));
    _sg_slot_reset(&smp->slot);
}

_SOKOL_PRIVATE void _sg_dealloc_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd && (shd->slot.state == SG_RESOURCESTATE_ALLOC) && (shd->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.shader_pool, _sg_slot_index(shd->slot.id));
    _sg_slot_reset(&shd->slot);
}

_SOKOL_PRIVATE void _sg_dealloc_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip && (pip->slot.state == SG_RESOURCESTATE_ALLOC) && (pip->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.pipeline_pool, _sg_slot_index(pip->slot.id));
    _sg_slot_reset(&pip->slot);
}

_SOKOL_PRIVATE void _sg_dealloc_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts && (atts->slot.state == SG_RESOURCESTATE_ALLOC) && (atts->slot.id != SG_INVALID_ID));
    _sg_pool_free_index(&_sg.pools.attachments_pool, _sg_slot_index(atts->slot.id));
    _sg_slot_reset(&atts->slot);
}

_SOKOL_PRIVATE void _sg_init_buffer(_sg_buffer_t* buf, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(buf && (buf->slot.state == SG_RESOURCESTATE_ALLOC));
    SOKOL_ASSERT(desc);
    if (_sg_validate_buffer_desc(desc)) {
        _sg_buffer_common_init(&buf->cmn, desc);
        buf->slot.state = _sg_create_buffer(buf, desc);
    } else {
        buf->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((buf->slot.state == SG_RESOURCESTATE_VALID)||(buf->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_init_image(_sg_image_t* img, const sg_image_desc* desc) {
    SOKOL_ASSERT(img && (img->slot.state == SG_RESOURCESTATE_ALLOC));
    SOKOL_ASSERT(desc);
    if (_sg_validate_image_desc(desc)) {
        _sg_image_common_init(&img->cmn, desc);
        img->slot.state = _sg_create_image(img, desc);
    } else {
        img->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((img->slot.state == SG_RESOURCESTATE_VALID)||(img->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_init_sampler(_sg_sampler_t* smp, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(smp && (smp->slot.state == SG_RESOURCESTATE_ALLOC));
    SOKOL_ASSERT(desc);
    if (_sg_validate_sampler_desc(desc)) {
        _sg_sampler_common_init(&smp->cmn, desc);
        smp->slot.state = _sg_create_sampler(smp, desc);
    } else {
        smp->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((smp->slot.state == SG_RESOURCESTATE_VALID)||(smp->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_init_shader(_sg_shader_t* shd, const sg_shader_desc* desc) {
    SOKOL_ASSERT(shd && (shd->slot.state == SG_RESOURCESTATE_ALLOC));
    SOKOL_ASSERT(desc);
    if (_sg_validate_shader_desc(desc)) {
        _sg_shader_common_init(&shd->cmn, desc);
        shd->slot.state = _sg_create_shader(shd, desc);
    } else {
        shd->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((shd->slot.state == SG_RESOURCESTATE_VALID)||(shd->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_init_pipeline(_sg_pipeline_t* pip, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(pip && (pip->slot.state == SG_RESOURCESTATE_ALLOC));
    SOKOL_ASSERT(desc);
    if (_sg_validate_pipeline_desc(desc)) {
        _sg_shader_t* shd = _sg_lookup_shader(desc->shader.id);
        if (shd && (shd->slot.state == SG_RESOURCESTATE_VALID)) {
            _sg_pipeline_common_init(&pip->cmn, desc, shd);
            pip->slot.state = _sg_create_pipeline(pip, desc);
        } else {
            pip->slot.state = SG_RESOURCESTATE_FAILED;
        }
    } else {
        pip->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((pip->slot.state == SG_RESOURCESTATE_VALID)||(pip->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_init_attachments(_sg_attachments_t* atts, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(atts && atts->slot.state == SG_RESOURCESTATE_ALLOC);
    SOKOL_ASSERT(desc);
    if (_sg_validate_attachments_desc(desc)) {
        // resolve image pointers, track width and height of render attachments,
        // the validation layer already ensured that render attachments have the
        // same width and height (which doesn't matter for storage attachments)
        _sg_attachments_ptrs_t atts_ptrs;
        _sg_clear(&atts_ptrs, sizeof(atts_ptrs));
        int width = 0;
        int height = 0;
        for (size_t i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
            if (desc->colors[i].image.id) {
                _sg_image_t* img = _sg_lookup_image(desc->colors[i].image.id);
                if (!(img && img->slot.state == SG_RESOURCESTATE_VALID)) {
                    atts->slot.state = SG_RESOURCESTATE_FAILED;
                    return;
                }
                const int mip_level = desc->colors[i].mip_level;
                width = _sg_miplevel_dim(img->cmn.width, mip_level);
                height = _sg_miplevel_dim(img->cmn.height, mip_level);
                atts_ptrs.color_images[i] = img;
            }
            if (desc->resolves[i].image.id) {
                _sg_image_t* img = _sg_lookup_image(desc->resolves[i].image.id);
                if (!(img && img->slot.state == SG_RESOURCESTATE_VALID)) {
                    atts->slot.state = SG_RESOURCESTATE_FAILED;
                    return;
                }
                atts_ptrs.resolve_images[i] = img;
            }
        }
        if (desc->depth_stencil.image.id) {
            _sg_image_t* img = _sg_lookup_image(desc->depth_stencil.image.id);
            if (!(img && img->slot.state == SG_RESOURCESTATE_VALID)) {
                atts->slot.state = SG_RESOURCESTATE_FAILED;
                return;
            }
            const int mip_level = desc->depth_stencil.mip_level;
            width = _sg_miplevel_dim(img->cmn.width, mip_level);
            height = _sg_miplevel_dim(img->cmn.height, mip_level);
            atts_ptrs.ds_image = img;
        }
        for (size_t i = 0; i < SG_MAX_STORAGE_ATTACHMENTS; i++) {
            if (desc->storages[i].image.id) {
                _sg_image_t* img = _sg_lookup_image(desc->storages[i].image.id);
                if (!(img && img->slot.state == SG_RESOURCESTATE_VALID)) {
                    atts->slot.state = SG_RESOURCESTATE_FAILED;
                    return;
                }
                atts_ptrs.storage_images[i] = img;
            }
        }
        _sg_attachments_common_init(&atts->cmn, desc, &atts_ptrs, width, height);
        atts->slot.state = _sg_create_attachments(atts, desc);
    } else {
        atts->slot.state = SG_RESOURCESTATE_FAILED;
    }
    SOKOL_ASSERT((atts->slot.state == SG_RESOURCESTATE_VALID)||(atts->slot.state == SG_RESOURCESTATE_FAILED));
}

_SOKOL_PRIVATE void _sg_uninit_buffer(_sg_buffer_t* buf) {
    SOKOL_ASSERT(buf && ((buf->slot.state == SG_RESOURCESTATE_VALID) || (buf->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_buffer(buf);
    _sg_reset_buffer_to_alloc_state(buf);
}

_SOKOL_PRIVATE void _sg_uninit_image(_sg_image_t* img) {
    SOKOL_ASSERT(img && ((img->slot.state == SG_RESOURCESTATE_VALID) || (img->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_image(img);
    _sg_reset_image_to_alloc_state(img);
}

_SOKOL_PRIVATE void _sg_uninit_sampler(_sg_sampler_t* smp) {
    SOKOL_ASSERT(smp && ((smp->slot.state == SG_RESOURCESTATE_VALID) || (smp->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_sampler(smp);
    _sg_reset_sampler_to_alloc_state(smp);
}

_SOKOL_PRIVATE void _sg_uninit_shader(_sg_shader_t* shd) {
    SOKOL_ASSERT(shd && ((shd->slot.state == SG_RESOURCESTATE_VALID) || (shd->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_shader(shd);
    _sg_reset_shader_to_alloc_state(shd);
}

_SOKOL_PRIVATE void _sg_uninit_pipeline(_sg_pipeline_t* pip) {
    SOKOL_ASSERT(pip && ((pip->slot.state == SG_RESOURCESTATE_VALID) || (pip->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_pipeline(pip);
    _sg_reset_pipeline_to_alloc_state(pip);
}

_SOKOL_PRIVATE void _sg_uninit_attachments(_sg_attachments_t* atts) {
    SOKOL_ASSERT(atts && ((atts->slot.state == SG_RESOURCESTATE_VALID) || (atts->slot.state == SG_RESOURCESTATE_FAILED)));
    _sg_discard_attachments(atts);
    _sg_reset_attachments_to_alloc_state(atts);
}

_SOKOL_PRIVATE void _sg_setup_commit_listeners(const sg_desc* desc) {
    SOKOL_ASSERT(desc->max_commit_listeners > 0);
    SOKOL_ASSERT(0 == _sg.commit_listeners.items);
    SOKOL_ASSERT(0 == _sg.commit_listeners.num);
    SOKOL_ASSERT(0 == _sg.commit_listeners.upper);
    _sg.commit_listeners.num = desc->max_commit_listeners;
    const size_t size = (size_t)_sg.commit_listeners.num * sizeof(sg_commit_listener);
    _sg.commit_listeners.items = (sg_commit_listener*)_sg_malloc_clear(size);
}

_SOKOL_PRIVATE void _sg_discard_commit_listeners(void) {
    SOKOL_ASSERT(0 != _sg.commit_listeners.items);
    _sg_free(_sg.commit_listeners.items);
    _sg.commit_listeners.items = 0;
}

_SOKOL_PRIVATE void _sg_notify_commit_listeners(void) {
    SOKOL_ASSERT(_sg.commit_listeners.items);
    for (int i = 0; i < _sg.commit_listeners.upper; i++) {
        const sg_commit_listener* listener = &_sg.commit_listeners.items[i];
        if (listener->func) {
            listener->func(listener->user_data);
        }
    }
}

_SOKOL_PRIVATE bool _sg_add_commit_listener(const sg_commit_listener* new_listener) {
    SOKOL_ASSERT(new_listener && new_listener->func);
    SOKOL_ASSERT(_sg.commit_listeners.items);
    // first check if the listener hadn't been added already
    for (int i = 0; i < _sg.commit_listeners.upper; i++) {
        const sg_commit_listener* slot = &_sg.commit_listeners.items[i];
        if ((slot->func == new_listener->func) && (slot->user_data == new_listener->user_data)) {
            _SG_ERROR(IDENTICAL_COMMIT_LISTENER);
            return false;
        }
    }
    // first try to plug a hole
    sg_commit_listener* slot = 0;
    for (int i = 0; i < _sg.commit_listeners.upper; i++) {
        if (_sg.commit_listeners.items[i].func == 0) {
            slot = &_sg.commit_listeners.items[i];
            break;
        }
    }
    if (!slot) {
        // append to end
        if (_sg.commit_listeners.upper < _sg.commit_listeners.num) {
            slot = &_sg.commit_listeners.items[_sg.commit_listeners.upper++];
        }
    }
    if (!slot) {
        _SG_ERROR(COMMIT_LISTENER_ARRAY_FULL);
        return false;
    }
    *slot = *new_listener;
    return true;
}

_SOKOL_PRIVATE bool _sg_remove_commit_listener(const sg_commit_listener* listener) {
    SOKOL_ASSERT(listener && listener->func);
    SOKOL_ASSERT(_sg.commit_listeners.items);
    for (int i = 0; i < _sg.commit_listeners.upper; i++) {
        sg_commit_listener* slot = &_sg.commit_listeners.items[i];
        // both the function pointer and user data must match!
        if ((slot->func == listener->func) && (slot->user_data == listener->user_data)) {
            slot->func = 0;
            slot->user_data = 0;
            // NOTE: since _sg_add_commit_listener() already catches duplicates,
            // we don't need to worry about them here
            return true;
        }
    }
    return false;
}

_SOKOL_PRIVATE void _sg_setup_compute(const sg_desc* desc) {
    SOKOL_ASSERT(desc && (desc->max_dispatch_calls_per_pass > 0));
    const uint32_t max_tracked_sbufs = (uint32_t)desc->max_dispatch_calls_per_pass * SG_MAX_STORAGEBUFFER_BINDSLOTS;
    _sg_tracker_init(&_sg.compute.readwrite_sbufs, max_tracked_sbufs);
}

_SOKOL_PRIVATE void _sg_discard_compute(void) {
    _sg_tracker_discard(&_sg.compute.readwrite_sbufs);
}

_SOKOL_PRIVATE void _sg_compute_pass_track_storage_buffer(_sg_buffer_t* sbuf, bool readonly) {
    SOKOL_ASSERT(sbuf);
    if (!readonly) {
        _sg_tracker_add(&_sg.compute.readwrite_sbufs, sbuf->slot.id);
    }
}

_SOKOL_PRIVATE void _sg_compute_on_endpass(void) {
    SOKOL_ASSERT(_sg.cur_pass.in_pass);
    SOKOL_ASSERT(_sg.cur_pass.is_compute);
    _sg_tracker_reset(&_sg.compute.readwrite_sbufs);
}

_SOKOL_PRIVATE sg_desc _sg_desc_defaults(const sg_desc* desc) {
    /*
        NOTE: on WebGPU, the default color pixel format MUST be provided,
        cannot be a default compile-time constant.
    */
    sg_desc res = *desc;
    #if defined(SOKOL_WGPU)
        SOKOL_ASSERT(SG_PIXELFORMAT_NONE < res.environment.defaults.color_format);
    #elif defined(SOKOL_METAL) || defined(SOKOL_D3D11)
        res.environment.defaults.color_format = _sg_def(res.environment.defaults.color_format, SG_PIXELFORMAT_BGRA8);
    #else
        res.environment.defaults.color_format = _sg_def(res.environment.defaults.color_format, SG_PIXELFORMAT_RGBA8);
    #endif
    res.environment.defaults.depth_format = _sg_def(res.environment.defaults.depth_format, SG_PIXELFORMAT_DEPTH_STENCIL);
    res.environment.defaults.sample_count = _sg_def(res.environment.defaults.sample_count, 1);
    res.buffer_pool_size = _sg_def(res.buffer_pool_size, _SG_DEFAULT_BUFFER_POOL_SIZE);
    res.image_pool_size = _sg_def(res.image_pool_size, _SG_DEFAULT_IMAGE_POOL_SIZE);
    res.sampler_pool_size = _sg_def(res.sampler_pool_size, _SG_DEFAULT_SAMPLER_POOL_SIZE);
    res.shader_pool_size = _sg_def(res.shader_pool_size, _SG_DEFAULT_SHADER_POOL_SIZE);
    res.pipeline_pool_size = _sg_def(res.pipeline_pool_size, _SG_DEFAULT_PIPELINE_POOL_SIZE);
    res.attachments_pool_size = _sg_def(res.attachments_pool_size, _SG_DEFAULT_ATTACHMENTS_POOL_SIZE);
    res.uniform_buffer_size = _sg_def(res.uniform_buffer_size, _SG_DEFAULT_UB_SIZE);
    res.max_dispatch_calls_per_pass = _sg_def(res.max_dispatch_calls_per_pass, _SG_DEFAULT_MAX_DISPATCH_CALLS_PER_PASS);
    res.max_commit_listeners = _sg_def(res.max_commit_listeners, _SG_DEFAULT_MAX_COMMIT_LISTENERS);
    res.wgpu_bindgroups_cache_size = _sg_def(res.wgpu_bindgroups_cache_size, _SG_DEFAULT_WGPU_BINDGROUP_CACHE_SIZE);
    return res;
}

_SOKOL_PRIVATE sg_pass _sg_pass_defaults(const sg_pass* pass) {
    sg_pass res = *pass;
    if (!res.compute) {
        if (res.compute && res.attachments.id == SG_INVALID_ID) {
            // this is a swapchain-pass
            res.swapchain.sample_count = _sg_def(res.swapchain.sample_count, _sg.desc.environment.defaults.sample_count);
            res.swapchain.color_format = _sg_def(res.swapchain.color_format, _sg.desc.environment.defaults.color_format);
            res.swapchain.depth_format = _sg_def(res.swapchain.depth_format, _sg.desc.environment.defaults.depth_format);
        }
        res.action = _sg_pass_action_defaults(&res.action);
    }
    return res;
}

_SOKOL_PRIVATE void _sg_discard_all_resources(void) {
    /*  this is a bit dumb since it loops over all pool slots to
        find the occupied slots, on the other hand it is only ever
        executed at shutdown
        NOTE: ONLY EXECUTE THIS AT SHUTDOWN
              ...because the free queues will not be reset
              and the resource slots not be cleared!
    */
    for (int i = 1; i < _sg.pools.buffer_pool.size; i++) {
        sg_resource_state state = _sg.pools.buffers[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_buffer(&_sg.pools.buffers[i]);
        }
    }
    for (int i = 1; i < _sg.pools.image_pool.size; i++) {
        sg_resource_state state = _sg.pools.images[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_image(&_sg.pools.images[i]);
        }
    }
    for (int i = 1; i < _sg.pools.sampler_pool.size; i++) {
        sg_resource_state state = _sg.pools.samplers[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_sampler(&_sg.pools.samplers[i]);
        }
    }
    for (int i = 1; i < _sg.pools.shader_pool.size; i++) {
        sg_resource_state state = _sg.pools.shaders[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_shader(&_sg.pools.shaders[i]);
        }
    }
    for (int i = 1; i < _sg.pools.pipeline_pool.size; i++) {
        sg_resource_state state = _sg.pools.pipelines[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_pipeline(&_sg.pools.pipelines[i]);
        }
    }
    for (int i = 1; i < _sg.pools.attachments_pool.size; i++) {
        sg_resource_state state = _sg.pools.attachments[i].slot.state;
        if ((state == SG_RESOURCESTATE_VALID) || (state == SG_RESOURCESTATE_FAILED)) {
            _sg_discard_attachments(&_sg.pools.attachments[i]);
        }
    }
}

// ██████  ██    ██ ██████  ██      ██  ██████
// ██   ██ ██    ██ ██   ██ ██      ██ ██
// ██████  ██    ██ ██████  ██      ██ ██
// ██      ██    ██ ██   ██ ██      ██ ██
// ██       ██████  ██████  ███████ ██  ██████
//
// >>public
SOKOL_API_IMPL void sg_setup(const sg_desc* desc) {
    SOKOL_ASSERT(desc);
    SOKOL_ASSERT((desc->_start_canary == 0) && (desc->_end_canary == 0));
    SOKOL_ASSERT((desc->allocator.alloc_fn && desc->allocator.free_fn) || (!desc->allocator.alloc_fn && !desc->allocator.free_fn));
    _SG_CLEAR_ARC_STRUCT(_sg_state_t, _sg);
    _sg.desc = _sg_desc_defaults(desc);
    _sg_setup_pools(&_sg.pools, &_sg.desc);
    _sg_setup_compute(&_sg.desc);
    _sg_setup_commit_listeners(&_sg.desc);
    _sg.frame_index = 1;
    _sg.stats_enabled = true;
    _sg_setup_backend(&_sg.desc);
    _sg.valid = true;
}

SOKOL_API_IMPL void sg_shutdown(void) {
    _sg_discard_all_resources();
    _sg_discard_backend();
    _sg_discard_commit_listeners();
    _sg_discard_compute();
    _sg_discard_pools(&_sg.pools);
    _SG_CLEAR_ARC_STRUCT(_sg_state_t, _sg);
}

SOKOL_API_IMPL bool sg_isvalid(void) {
    return _sg.valid;
}

SOKOL_API_IMPL sg_desc sg_query_desc(void) {
    SOKOL_ASSERT(_sg.valid);
    return _sg.desc;
}

SOKOL_API_IMPL sg_backend sg_query_backend(void) {
    SOKOL_ASSERT(_sg.valid);
    return _sg.backend;
}

SOKOL_API_IMPL sg_features sg_query_features(void) {
    SOKOL_ASSERT(_sg.valid);
    return _sg.features;
}

SOKOL_API_IMPL sg_limits sg_query_limits(void) {
    SOKOL_ASSERT(_sg.valid);
    return _sg.limits;
}

SOKOL_API_IMPL sg_pixelformat_info sg_query_pixelformat(sg_pixel_format fmt) {
    SOKOL_ASSERT(_sg.valid);
    int fmt_index = (int) fmt;
    SOKOL_ASSERT((fmt_index > SG_PIXELFORMAT_NONE) && (fmt_index < _SG_PIXELFORMAT_NUM));
    const _sg_pixelformat_info_t* src = &_sg.formats[fmt_index];
    sg_pixelformat_info res;
    _sg_clear(&res, sizeof(res));
    res.sample = src->sample;
    res.filter = src->filter;
    res.render = src->render;
    res.blend = src->blend;
    res.msaa = src->msaa;
    res.depth = src->depth;
    res.compressed = _sg_is_compressed_pixel_format(fmt);
    res.read = src->read;
    res.write = src->write;
    if (!res.compressed) {
        res.bytes_per_pixel = _sg_pixelformat_bytesize(fmt);
    }
    return res;
}

SOKOL_API_IMPL int sg_query_row_pitch(sg_pixel_format fmt, int width, int row_align_bytes) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(width > 0);
    SOKOL_ASSERT((row_align_bytes > 0) && _sg_ispow2(row_align_bytes));
    SOKOL_ASSERT(((int)fmt > SG_PIXELFORMAT_NONE) && ((int)fmt < _SG_PIXELFORMAT_NUM));
    return _sg_row_pitch(fmt, width, row_align_bytes);
}

SOKOL_API_IMPL int sg_query_surface_pitch(sg_pixel_format fmt, int width, int height, int row_align_bytes) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT((width > 0) && (height > 0));
    SOKOL_ASSERT((row_align_bytes > 0) && _sg_ispow2(row_align_bytes));
    SOKOL_ASSERT(((int)fmt > SG_PIXELFORMAT_NONE) && ((int)fmt < _SG_PIXELFORMAT_NUM));
    return _sg_surface_pitch(fmt, width, height, row_align_bytes);
}

SOKOL_API_IMPL sg_frame_stats sg_query_frame_stats(void) {
    SOKOL_ASSERT(_sg.valid);
    return _sg.prev_stats;
}

SOKOL_API_IMPL sg_trace_hooks sg_install_trace_hooks(const sg_trace_hooks* trace_hooks) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(trace_hooks);
    _SOKOL_UNUSED(trace_hooks);
    #if defined(SOKOL_TRACE_HOOKS)
        sg_trace_hooks old_hooks = _sg.hooks;
        _sg.hooks = *trace_hooks;
    #else
        static sg_trace_hooks old_hooks;
        _SG_WARN(TRACE_HOOKS_NOT_ENABLED);
    #endif
    return old_hooks;
}

SOKOL_API_IMPL sg_buffer sg_alloc_buffer(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_buffer res = _sg_alloc_buffer();
    _SG_TRACE_ARGS(alloc_buffer, res);
    return res;
}

SOKOL_API_IMPL sg_image sg_alloc_image(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_image res = _sg_alloc_image();
    _SG_TRACE_ARGS(alloc_image, res);
    return res;
}

SOKOL_API_IMPL sg_sampler sg_alloc_sampler(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_sampler res = _sg_alloc_sampler();
    _SG_TRACE_ARGS(alloc_sampler, res);
    return res;
}

SOKOL_API_IMPL sg_shader sg_alloc_shader(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_shader res = _sg_alloc_shader();
    _SG_TRACE_ARGS(alloc_shader, res);
    return res;
}

SOKOL_API_IMPL sg_pipeline sg_alloc_pipeline(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_pipeline res = _sg_alloc_pipeline();
    _SG_TRACE_ARGS(alloc_pipeline, res);
    return res;
}

SOKOL_API_IMPL sg_attachments sg_alloc_attachments(void) {
    SOKOL_ASSERT(_sg.valid);
    sg_attachments res = _sg_alloc_attachments();
    _SG_TRACE_ARGS(alloc_attachments, res);
    return res;
}

SOKOL_API_IMPL void sg_dealloc_buffer(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        if (buf->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_buffer(buf);
        } else {
            _SG_ERROR(DEALLOC_BUFFER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_buffer, buf_id);
}

SOKOL_API_IMPL void sg_dealloc_image(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        if (img->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_image(img);
        } else {
            _SG_ERROR(DEALLOC_IMAGE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_image, img_id);
}

SOKOL_API_IMPL void sg_dealloc_sampler(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        if (smp->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_sampler(smp);
        } else {
            _SG_ERROR(DEALLOC_SAMPLER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_sampler, smp_id);
}

SOKOL_API_IMPL void sg_dealloc_shader(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        if (shd->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_shader(shd);
        } else {
            _SG_ERROR(DEALLOC_SHADER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_shader, shd_id);
}

SOKOL_API_IMPL void sg_dealloc_pipeline(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        if (pip->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_pipeline(pip);
        } else {
            _SG_ERROR(DEALLOC_PIPELINE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_pipeline, pip_id);
}

SOKOL_API_IMPL void sg_dealloc_attachments(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        if (atts->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_attachments(atts);
        } else {
            _SG_ERROR(DEALLOC_ATTACHMENTS_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(dealloc_attachments, atts_id);
}

SOKOL_API_IMPL void sg_init_buffer(sg_buffer buf_id, const sg_buffer_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_buffer_desc desc_def = _sg_buffer_desc_defaults(desc);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        if (buf->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_buffer(buf, &desc_def);
            SOKOL_ASSERT((buf->slot.state == SG_RESOURCESTATE_VALID) || (buf->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_BUFFER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_buffer, buf_id, &desc_def);
}

SOKOL_API_IMPL void sg_init_image(sg_image img_id, const sg_image_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_image_desc desc_def = _sg_image_desc_defaults(desc);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        if (img->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_image(img, &desc_def);
            SOKOL_ASSERT((img->slot.state == SG_RESOURCESTATE_VALID) || (img->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_IMAGE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_image, img_id, &desc_def);
}

SOKOL_API_IMPL void sg_init_sampler(sg_sampler smp_id, const sg_sampler_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_sampler_desc desc_def = _sg_sampler_desc_defaults(desc);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        if (smp->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_sampler(smp, &desc_def);
            SOKOL_ASSERT((smp->slot.state == SG_RESOURCESTATE_VALID) || (smp->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_SAMPLER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_sampler, smp_id, &desc_def);
}

SOKOL_API_IMPL void sg_init_shader(sg_shader shd_id, const sg_shader_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_shader_desc desc_def = _sg_shader_desc_defaults(desc);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        if (shd->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_shader(shd, &desc_def);
            SOKOL_ASSERT((shd->slot.state == SG_RESOURCESTATE_VALID) || (shd->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_SHADER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_shader, shd_id, &desc_def);
}

SOKOL_API_IMPL void sg_init_pipeline(sg_pipeline pip_id, const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_pipeline_desc desc_def = _sg_pipeline_desc_defaults(desc);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        if (pip->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_pipeline(pip, &desc_def);
            SOKOL_ASSERT((pip->slot.state == SG_RESOURCESTATE_VALID) || (pip->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_PIPELINE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_pipeline, pip_id, &desc_def);
}

SOKOL_API_IMPL void sg_init_attachments(sg_attachments atts_id, const sg_attachments_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    sg_attachments_desc desc_def = _sg_attachments_desc_defaults(desc);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        if (atts->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_init_attachments(atts, &desc_def);
            SOKOL_ASSERT((atts->slot.state == SG_RESOURCESTATE_VALID) || (atts->slot.state == SG_RESOURCESTATE_FAILED));
        } else {
            _SG_ERROR(INIT_ATTACHMENTS_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(init_attachments, atts_id, &desc_def);
}

SOKOL_API_IMPL void sg_uninit_buffer(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        if ((buf->slot.state == SG_RESOURCESTATE_VALID) || (buf->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_buffer(buf);
            SOKOL_ASSERT(buf->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_BUFFER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_buffer, buf_id);
}

SOKOL_API_IMPL void sg_uninit_image(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        if ((img->slot.state == SG_RESOURCESTATE_VALID) || (img->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_image(img);
            SOKOL_ASSERT(img->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_IMAGE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_image, img_id);
}

SOKOL_API_IMPL void sg_uninit_sampler(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        if ((smp->slot.state == SG_RESOURCESTATE_VALID) || (smp->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_sampler(smp);
            SOKOL_ASSERT(smp->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_SAMPLER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_sampler, smp_id);
}

SOKOL_API_IMPL void sg_uninit_shader(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        if ((shd->slot.state == SG_RESOURCESTATE_VALID) || (shd->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_shader(shd);
            SOKOL_ASSERT(shd->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_SHADER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_shader, shd_id);
}

SOKOL_API_IMPL void sg_uninit_pipeline(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        if ((pip->slot.state == SG_RESOURCESTATE_VALID) || (pip->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_pipeline(pip);
            SOKOL_ASSERT(pip->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_PIPELINE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_pipeline, pip_id);
}

SOKOL_API_IMPL void sg_uninit_attachments(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        if ((atts->slot.state == SG_RESOURCESTATE_VALID) || (atts->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_attachments(atts);
            SOKOL_ASSERT(atts->slot.state == SG_RESOURCESTATE_ALLOC);
        } else {
            _SG_ERROR(UNINIT_ATTACHMENTS_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(uninit_attachments, atts_id);
}

SOKOL_API_IMPL void sg_fail_buffer(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        if (buf->slot.state == SG_RESOURCESTATE_ALLOC) {
            buf->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_BUFFER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_buffer, buf_id);
}

SOKOL_API_IMPL void sg_fail_image(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        if (img->slot.state == SG_RESOURCESTATE_ALLOC) {
            img->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_IMAGE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_image, img_id);
}

SOKOL_API_IMPL void sg_fail_sampler(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        if (smp->slot.state == SG_RESOURCESTATE_ALLOC) {
            smp->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_SAMPLER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_sampler, smp_id);
}

SOKOL_API_IMPL void sg_fail_shader(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        if (shd->slot.state == SG_RESOURCESTATE_ALLOC) {
            shd->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_SHADER_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_shader, shd_id);
}

SOKOL_API_IMPL void sg_fail_pipeline(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        if (pip->slot.state == SG_RESOURCESTATE_ALLOC) {
            pip->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_PIPELINE_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_pipeline, pip_id);
}

SOKOL_API_IMPL void sg_fail_attachments(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        if (atts->slot.state == SG_RESOURCESTATE_ALLOC) {
            atts->slot.state = SG_RESOURCESTATE_FAILED;
        } else {
            _SG_ERROR(FAIL_ATTACHMENTS_INVALID_STATE);
        }
    }
    _SG_TRACE_ARGS(fail_attachments, atts_id);
}

SOKOL_API_IMPL sg_resource_state sg_query_buffer_state(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    sg_resource_state res = buf ? buf->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_resource_state sg_query_image_state(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    sg_resource_state res = img ? img->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_resource_state sg_query_sampler_state(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    sg_resource_state res = smp ? smp->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_resource_state sg_query_shader_state(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    sg_resource_state res = shd ? shd->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_resource_state sg_query_pipeline_state(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    sg_resource_state res = pip ? pip->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_resource_state sg_query_attachments_state(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    sg_resource_state res = atts ? atts->slot.state : SG_RESOURCESTATE_INVALID;
    return res;
}

SOKOL_API_IMPL sg_buffer sg_make_buffer(const sg_buffer_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_buffer_desc desc_def = _sg_buffer_desc_defaults(desc);
    sg_buffer buf_id = _sg_alloc_buffer();
    if (buf_id.id != SG_INVALID_ID) {
        _sg_buffer_t* buf = _sg_buffer_at(buf_id.id);
        SOKOL_ASSERT(buf && (buf->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_buffer(buf, &desc_def);
        SOKOL_ASSERT((buf->slot.state == SG_RESOURCESTATE_VALID) || (buf->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_buffer, &desc_def, buf_id);
    return buf_id;
}

SOKOL_API_IMPL sg_image sg_make_image(const sg_image_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_image_desc desc_def = _sg_image_desc_defaults(desc);
    sg_image img_id = _sg_alloc_image();
    if (img_id.id != SG_INVALID_ID) {
        _sg_image_t* img = _sg_image_at(img_id.id);
        SOKOL_ASSERT(img && (img->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_image(img, &desc_def);
        SOKOL_ASSERT((img->slot.state == SG_RESOURCESTATE_VALID) || (img->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_image, &desc_def, img_id);
    return img_id;
}

SOKOL_API_IMPL sg_sampler sg_make_sampler(const sg_sampler_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_sampler_desc desc_def = _sg_sampler_desc_defaults(desc);
    sg_sampler smp_id = _sg_alloc_sampler();
    if (smp_id.id != SG_INVALID_ID) {
        _sg_sampler_t* smp = _sg_sampler_at(smp_id.id);
        SOKOL_ASSERT(smp && (smp->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_sampler(smp, &desc_def);
        SOKOL_ASSERT((smp->slot.state == SG_RESOURCESTATE_VALID) || (smp->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_sampler, &desc_def, smp_id);
    return smp_id;
}

SOKOL_API_IMPL sg_shader sg_make_shader(const sg_shader_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_shader_desc desc_def = _sg_shader_desc_defaults(desc);
    sg_shader shd_id = _sg_alloc_shader();
    if (shd_id.id != SG_INVALID_ID) {
        _sg_shader_t* shd = _sg_shader_at(shd_id.id);
        SOKOL_ASSERT(shd && (shd->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_shader(shd, &desc_def);
        SOKOL_ASSERT((shd->slot.state == SG_RESOURCESTATE_VALID) || (shd->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_shader, &desc_def, shd_id);
    return shd_id;
}

SOKOL_API_IMPL sg_pipeline sg_make_pipeline(const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_pipeline_desc desc_def = _sg_pipeline_desc_defaults(desc);
    sg_pipeline pip_id = _sg_alloc_pipeline();
    if (pip_id.id != SG_INVALID_ID) {
        _sg_pipeline_t* pip = _sg_pipeline_at(pip_id.id);
        SOKOL_ASSERT(pip && (pip->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_pipeline(pip, &desc_def);
        SOKOL_ASSERT((pip->slot.state == SG_RESOURCESTATE_VALID) || (pip->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_pipeline, &desc_def, pip_id);
    return pip_id;
}

SOKOL_API_IMPL sg_attachments sg_make_attachments(const sg_attachments_desc* desc) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(desc);
    sg_attachments_desc desc_def = _sg_attachments_desc_defaults(desc);
    sg_attachments atts_id = _sg_alloc_attachments();
    if (atts_id.id != SG_INVALID_ID) {
        _sg_attachments_t* atts = _sg_attachments_at(atts_id.id);
        SOKOL_ASSERT(atts && (atts->slot.state == SG_RESOURCESTATE_ALLOC));
        _sg_init_attachments(atts, &desc_def);
        SOKOL_ASSERT((atts->slot.state == SG_RESOURCESTATE_VALID) || (atts->slot.state == SG_RESOURCESTATE_FAILED));
    }
    _SG_TRACE_ARGS(make_attachments, &desc_def, atts_id);
    return atts_id;
}

SOKOL_API_IMPL void sg_destroy_buffer(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_buffer, buf_id);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        if ((buf->slot.state == SG_RESOURCESTATE_VALID) || (buf->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_buffer(buf);
            SOKOL_ASSERT(buf->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (buf->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_buffer(buf);
            SOKOL_ASSERT(buf->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_destroy_image(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_image, img_id);
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        if ((img->slot.state == SG_RESOURCESTATE_VALID) || (img->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_image(img);
            SOKOL_ASSERT(img->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (img->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_image(img);
            SOKOL_ASSERT(img->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_destroy_sampler(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_sampler, smp_id);
    _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        if ((smp->slot.state == SG_RESOURCESTATE_VALID) || (smp->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_sampler(smp);
            SOKOL_ASSERT(smp->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (smp->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_sampler(smp);
            SOKOL_ASSERT(smp->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_destroy_shader(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_shader, shd_id);
    _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        if ((shd->slot.state == SG_RESOURCESTATE_VALID) || (shd->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_shader(shd);
            SOKOL_ASSERT(shd->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (shd->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_shader(shd);
            SOKOL_ASSERT(shd->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_destroy_pipeline(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_pipeline, pip_id);
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        if ((pip->slot.state == SG_RESOURCESTATE_VALID) || (pip->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_pipeline(pip);
            SOKOL_ASSERT(pip->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (pip->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_pipeline(pip);
            SOKOL_ASSERT(pip->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_destroy_attachments(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    _SG_TRACE_ARGS(destroy_attachments, atts_id);
    _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        if ((atts->slot.state == SG_RESOURCESTATE_VALID) || (atts->slot.state == SG_RESOURCESTATE_FAILED)) {
            _sg_uninit_attachments(atts);
            SOKOL_ASSERT(atts->slot.state == SG_RESOURCESTATE_ALLOC);
        }
        if (atts->slot.state == SG_RESOURCESTATE_ALLOC) {
            _sg_dealloc_attachments(atts);
            SOKOL_ASSERT(atts->slot.state == SG_RESOURCESTATE_INITIAL);
        }
    }
}

SOKOL_API_IMPL void sg_begin_pass(const sg_pass* pass) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(!_sg.cur_pass.valid);
    SOKOL_ASSERT(!_sg.cur_pass.in_pass);
    SOKOL_ASSERT(_sg_attachments_ref_null(&_sg.cur_pass.atts));
    SOKOL_ASSERT(pass);
    SOKOL_ASSERT((pass->_start_canary == 0) && (pass->_end_canary == 0));
    const sg_pass pass_def = _sg_pass_defaults(pass);
    if (!_sg_validate_begin_pass(&pass_def)) {
        return;
    }
    _sg.cur_pass.atts = _sg_attachments_ref(0);
    if (pass_def.attachments.id != SG_INVALID_ID) {
        _sg_attachments_t* atts = _sg_lookup_attachments(pass_def.attachments.id);
        if (0 == atts) {
            _SG_ERROR(BEGINPASS_ATTACHMENT_INVALID);
            return;
        }
        SOKOL_ASSERT(atts);
        _sg.cur_pass.atts = _sg_attachments_ref(atts);
        _sg.cur_pass.width = atts->cmn.width;
        _sg.cur_pass.height = atts->cmn.height;
    } else if (!pass_def.compute) {
        // a swapchain pass
        SOKOL_ASSERT(pass_def.swapchain.width > 0);
        SOKOL_ASSERT(pass_def.swapchain.height > 0);
        SOKOL_ASSERT(pass_def.swapchain.color_format > SG_PIXELFORMAT_NONE);
        SOKOL_ASSERT(pass_def.swapchain.sample_count > 0);
        _sg.cur_pass.width = pass_def.swapchain.width;
        _sg.cur_pass.height = pass_def.swapchain.height;
        _sg.cur_pass.swapchain.color_fmt = pass_def.swapchain.color_format;
        _sg.cur_pass.swapchain.depth_fmt = pass_def.swapchain.depth_format;
        _sg.cur_pass.swapchain.sample_count = pass_def.swapchain.sample_count;
    }
    _sg.cur_pass.valid = true;  // may be overruled by backend begin-pass functions
    _sg.cur_pass.in_pass = true;
    _sg.cur_pass.is_compute = pass_def.compute;
    _sg_begin_pass(&pass_def);
    _SG_TRACE_ARGS(begin_pass, &pass_def);
}

SOKOL_API_IMPL void sg_apply_viewport(int x, int y, int width, int height, bool origin_top_left) {
    SOKOL_ASSERT(_sg.valid);
    #if defined(SOKOL_DEBUG)
    if (!_sg_validate_apply_viewport(x, y, width, height, origin_top_left)) {
        return;
    }
    #endif
    _sg_stats_add(num_apply_viewport, 1);
    if (!_sg.cur_pass.valid) {
        return;
    }
    _sg_apply_viewport(x, y, width, height, origin_top_left);
    _SG_TRACE_ARGS(apply_viewport, x, y, width, height, origin_top_left);
}

SOKOL_API_IMPL void sg_apply_viewportf(float x, float y, float width, float height, bool origin_top_left) {
    sg_apply_viewport((int)x, (int)y, (int)width, (int)height, origin_top_left);
}

SOKOL_API_IMPL void sg_apply_scissor_rect(int x, int y, int width, int height, bool origin_top_left) {
    SOKOL_ASSERT(_sg.valid);
    #if defined(SOKOL_DEBUG)
    if (!_sg_validate_apply_scissor_rect(x, y, width, height, origin_top_left)) {
        return;
    }
    #endif
    _sg_stats_add(num_apply_scissor_rect, 1);
    if (!_sg.cur_pass.valid) {
        return;
    }
    _sg_apply_scissor_rect(x, y, width, height, origin_top_left);
    _SG_TRACE_ARGS(apply_scissor_rect, x, y, width, height, origin_top_left);
}

SOKOL_API_IMPL void sg_apply_scissor_rectf(float x, float y, float width, float height, bool origin_top_left) {
    sg_apply_scissor_rect((int)x, (int)y, (int)width, (int)height, origin_top_left);
}

SOKOL_API_IMPL void sg_apply_pipeline(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_stats_add(num_apply_pipeline, 1);
    if (!_sg_validate_apply_pipeline(pip_id)) {
        _sg.next_draw_valid = false;
        return;
    }
    if (!_sg.cur_pass.valid) {
        return;
    }
    _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    SOKOL_ASSERT(pip);
    _sg.cur_pip = _sg_pipeline_ref(pip);

    _sg.next_draw_valid = (SG_RESOURCESTATE_VALID == pip->slot.state);
    if (!_sg.next_draw_valid) {
        return;
    }

    _sg_apply_pipeline(pip);

    // set the expected bindings and uniform block flags
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&pip->cmn.shader);
    _sg.required_bindings_and_uniforms = pip->cmn.required_bindings_and_uniforms | shd->cmn.required_bindings_and_uniforms;
    _sg.applied_bindings_and_uniforms = 0;

    _SG_TRACE_ARGS(apply_pipeline, pip_id);
}

SOKOL_API_IMPL void sg_apply_bindings(const sg_bindings* bindings) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(bindings);
    SOKOL_ASSERT((bindings->_start_canary == 0) && (bindings->_end_canary==0));
    _sg_stats_add(num_apply_bindings, 1);
    _sg.applied_bindings_and_uniforms |= (1 << SG_MAX_UNIFORMBLOCK_BINDSLOTS);
    if (!_sg_validate_apply_bindings(bindings)) {
        _sg.next_draw_valid = false;
    }
    if (!_sg_pipeline_ref_alive(&_sg.cur_pip)) {
        _sg.next_draw_valid = false;
    }
    if (!_sg.cur_pass.valid) {
        return;
    }
    if (!_sg.next_draw_valid) {
        return;
    }

    _sg_bindings_ptrs_t bnd;
    _sg_clear(&bnd, sizeof(bnd));
    bnd.pip = _sg_pipeline_ref_ptr(&_sg.cur_pip);
    const _sg_shader_t* shd = _sg_shader_ref_ptr(&bnd.pip->cmn.shader);
    if (!_sg.cur_pass.is_compute) {
        for (size_t i = 0; i < SG_MAX_VERTEXBUFFER_BINDSLOTS; i++) {
            if (bnd.pip->cmn.vertex_buffer_layout_active[i]) {
                SOKOL_ASSERT(bindings->vertex_buffers[i].id != SG_INVALID_ID);
                bnd.vbs[i] = _sg_lookup_buffer(bindings->vertex_buffers[i].id);
                bnd.vb_offsets[i] = bindings->vertex_buffer_offsets[i];
                if (bnd.vbs[i]) {
                    _sg.next_draw_valid &= (SG_RESOURCESTATE_VALID == bnd.vbs[i]->slot.state);
                    _sg.next_draw_valid &= !bnd.vbs[i]->cmn.append_overflow;
                } else {
                    _sg.next_draw_valid = false;
                }
            }
        }
        if (bindings->index_buffer.id) {
            bnd.ib = _sg_lookup_buffer(bindings->index_buffer.id);
            bnd.ib_offset = bindings->index_buffer_offset;
            if (bnd.ib) {
                _sg.next_draw_valid &= (SG_RESOURCESTATE_VALID == bnd.ib->slot.state);
                _sg.next_draw_valid &= !bnd.ib->cmn.append_overflow;
            } else {
                _sg.next_draw_valid = false;
            }
        }
    }

    for (int i = 0; i < SG_MAX_IMAGE_BINDSLOTS; i++) {
        if (shd->cmn.images[i].stage != SG_SHADERSTAGE_NONE) {
            SOKOL_ASSERT(bindings->images[i].id != SG_INVALID_ID);
            bnd.imgs[i] = _sg_lookup_image(bindings->images[i].id);
            if (bnd.imgs[i]) {
                _sg.next_draw_valid &= (SG_RESOURCESTATE_VALID == bnd.imgs[i]->slot.state);
            } else {
                _sg.next_draw_valid = false;
            }
        }
    }

    for (size_t i = 0; i < SG_MAX_SAMPLER_BINDSLOTS; i++) {
        if (shd->cmn.samplers[i].stage != SG_SHADERSTAGE_NONE) {
            SOKOL_ASSERT(bindings->samplers[i].id != SG_INVALID_ID);
            bnd.smps[i] = _sg_lookup_sampler(bindings->samplers[i].id);
            if (bnd.smps[i]) {
                _sg.next_draw_valid &= (SG_RESOURCESTATE_VALID == bnd.smps[i]->slot.state);
            } else {
                _sg.next_draw_valid = false;
            }
        }
    }

    for (size_t i = 0; i < SG_MAX_STORAGEBUFFER_BINDSLOTS; i++) {
        if (shd->cmn.storage_buffers[i].stage != SG_SHADERSTAGE_NONE) {
            SOKOL_ASSERT(bindings->storage_buffers[i].id != SG_INVALID_ID);
            bnd.sbufs[i] = _sg_lookup_buffer(bindings->storage_buffers[i].id);
            if (bnd.sbufs[i]) {
                _sg.next_draw_valid &= (SG_RESOURCESTATE_VALID == bnd.sbufs[i]->slot.state);
                if (_sg.cur_pass.is_compute) {
                    _sg_compute_pass_track_storage_buffer(bnd.sbufs[i], shd->cmn.storage_buffers[i].readonly);
                }
            } else {
                _sg.next_draw_valid = false;
            }
        }
    }

    if (_sg.next_draw_valid) {
        _sg.next_draw_valid &= _sg_apply_bindings(&bnd);
        _SG_TRACE_ARGS(apply_bindings, bindings);
    }
}

SOKOL_API_IMPL void sg_apply_uniforms(int ub_slot, const sg_range* data) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT((ub_slot >= 0) && (ub_slot < SG_MAX_UNIFORMBLOCK_BINDSLOTS));
    SOKOL_ASSERT(data && data->ptr && (data->size > 0));
    _sg_stats_add(num_apply_uniforms, 1);
    _sg_stats_add(size_apply_uniforms, (uint32_t)data->size);
    _sg.applied_bindings_and_uniforms |= 1 << ub_slot;
    if (!_sg_validate_apply_uniforms(ub_slot, data)) {
        _sg.next_draw_valid = false;
        return;
    }
    if (!_sg.cur_pass.valid) {
        return;
    }
    if (!_sg.next_draw_valid) {
        return;
    }
    _sg_apply_uniforms(ub_slot, data);
    _SG_TRACE_ARGS(apply_uniforms, ub_slot, data);
}

SOKOL_API_IMPL void sg_draw(int base_element, int num_elements, int num_instances) {
    SOKOL_ASSERT(_sg.valid);
    #if defined(SOKOL_DEBUG)
    if (!_sg_validate_draw(base_element, num_elements, num_instances)) {
        return;
    }
    #endif
    _sg_stats_add(num_draw, 1);
    if (!_sg.cur_pass.valid) {
        return;
    }
    if (!_sg.next_draw_valid) {
        return;
    }
    // skip no-op draws
    if ((0 == num_elements) || (0 == num_instances)) {
        return;
    }
    _sg_draw(base_element, num_elements, num_instances);
    _SG_TRACE_ARGS(draw, base_element, num_elements, num_instances);
}

SOKOL_API_IMPL void sg_dispatch(int num_groups_x, int num_groups_y, int num_groups_z) {
    SOKOL_ASSERT(_sg.valid);
    #if defined(SOKOL_DEBUG)
    if (!_sg_validate_dispatch(num_groups_x, num_groups_y, num_groups_z)) {
        return;
    }
    #endif
    _sg_stats_add(num_dispatch, 1);
    if (!_sg.cur_pass.valid) {
        return;
    }
    if (!_sg.next_draw_valid) {
        return;
    }
    // skip no-op dispatches
    if ((0 == num_groups_x) || (0 == num_groups_y) || (0 == num_groups_z)) {
        return;
    }
    _sg_dispatch(num_groups_x, num_groups_y, num_groups_z);
    _SG_TRACE_ARGS(dispatch, num_groups_x, num_groups_y, num_groups_z);
}

SOKOL_API_IMPL void sg_end_pass(void) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(_sg.cur_pass.in_pass);
    _sg_stats_add(num_passes, 1);
    // NOTE: don't exit early if !_sg.cur_pass.valid
    _sg_end_pass();
    _sg.cur_pip = _sg_pipeline_ref(0);
    if (_sg.cur_pass.is_compute) {
        _sg_compute_on_endpass();
    }
    _sg_clear(&_sg.cur_pass, sizeof(_sg.cur_pass));
    _SG_TRACE_NOARGS(end_pass);
}

SOKOL_API_IMPL void sg_commit(void) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(!_sg.cur_pass.valid);
    SOKOL_ASSERT(!_sg.cur_pass.in_pass);
    _sg_commit();
    _sg.stats.frame_index = _sg.frame_index;
    _sg.prev_stats = _sg.stats;
    _sg_clear(&_sg.stats, sizeof(_sg.stats));
    _sg_notify_commit_listeners();
    _SG_TRACE_NOARGS(commit);
    _sg.frame_index++;
}

SOKOL_API_IMPL void sg_reset_state_cache(void) {
    SOKOL_ASSERT(_sg.valid);
    _sg_reset_state_cache();
    _SG_TRACE_NOARGS(reset_state_cache);
}

SOKOL_API_IMPL void sg_update_buffer(sg_buffer buf_id, const sg_range* data) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(data && data->ptr && (data->size > 0));
    _sg_stats_add(num_update_buffer, 1);
    _sg_stats_add(size_update_buffer, (uint32_t)data->size);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if ((data->size > 0) && buf && (buf->slot.state == SG_RESOURCESTATE_VALID)) {
        if (_sg_validate_update_buffer(buf, data)) {
            SOKOL_ASSERT(data->size <= (size_t)buf->cmn.size);
            // only one update allowed per buffer and frame
            SOKOL_ASSERT(buf->cmn.update_frame_index != _sg.frame_index);
            // update and append on same buffer in same frame not allowed
            SOKOL_ASSERT(buf->cmn.append_frame_index != _sg.frame_index);
            _sg_update_buffer(buf, data);
            buf->cmn.update_frame_index = _sg.frame_index;
        }
    }
    _SG_TRACE_ARGS(update_buffer, buf_id, data);
}

SOKOL_API_IMPL int sg_append_buffer(sg_buffer buf_id, const sg_range* data) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(data && data->ptr);
    _sg_stats_add(num_append_buffer, 1);
    _sg_stats_add(size_append_buffer, (uint32_t)data->size);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    int result;
    if (buf) {
        // rewind append cursor in a new frame
        if (buf->cmn.append_frame_index != _sg.frame_index) {
            buf->cmn.append_pos = 0;
            buf->cmn.append_overflow = false;
        }
        if (((size_t)buf->cmn.append_pos + data->size) > (size_t)buf->cmn.size) {
            buf->cmn.append_overflow = true;
        }
        const int start_pos = buf->cmn.append_pos;
        // NOTE: the multiple-of-4 requirement for the buffer offset is coming
        // from WebGPU, but we want identical behaviour between backends
        SOKOL_ASSERT(_sg_multiple_u64((uint64_t)start_pos, 4));
        if (buf->slot.state == SG_RESOURCESTATE_VALID) {
            if (_sg_validate_append_buffer(buf, data)) {
                if (!buf->cmn.append_overflow && (data->size > 0)) {
                    // update and append on same buffer in same frame not allowed
                    SOKOL_ASSERT(buf->cmn.update_frame_index != _sg.frame_index);
                    _sg_append_buffer(buf, data, buf->cmn.append_frame_index != _sg.frame_index);
                    buf->cmn.append_pos += (int) _sg_roundup_u64(data->size, 4);
                    buf->cmn.append_frame_index = _sg.frame_index;
                }
            }
        }
        result = start_pos;
    } else {
        // FIXME: should we return -1 here?
        result = 0;
    }
    _SG_TRACE_ARGS(append_buffer, buf_id, data, result);
    return result;
}

SOKOL_API_IMPL bool sg_query_buffer_overflow(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    bool result = buf ? buf->cmn.append_overflow : false;
    return result;
}

SOKOL_API_IMPL bool sg_query_buffer_will_overflow(sg_buffer buf_id, size_t size) {
    SOKOL_ASSERT(_sg.valid);
    _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    bool result = false;
    if (buf) {
        int append_pos = buf->cmn.append_pos;
        // rewind append cursor in a new frame
        if (buf->cmn.append_frame_index != _sg.frame_index) {
            append_pos = 0;
        }
        if ((append_pos + _sg_roundup((int)size, 4)) > buf->cmn.size) {
            result = true;
        }
    }
    return result;
}

SOKOL_API_IMPL void sg_update_image(sg_image img_id, const sg_image_data* data) {
    SOKOL_ASSERT(_sg.valid);
    _sg_stats_add(num_update_image, 1);
    for (int face_index = 0; face_index < SG_CUBEFACE_NUM; face_index++) {
        for (int mip_index = 0; mip_index < SG_MAX_MIPMAPS; mip_index++) {
            if (data->subimage[face_index][mip_index].size == 0) {
                break;
            }
            _sg_stats_add(size_update_image, (uint32_t)data->subimage[face_index][mip_index].size);
        }
    }
    _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img && img->slot.state == SG_RESOURCESTATE_VALID) {
        if (_sg_validate_update_image(img, data)) {
            SOKOL_ASSERT(img->cmn.upd_frame_index != _sg.frame_index);
            _sg_update_image(img, data);
            img->cmn.upd_frame_index = _sg.frame_index;
        }
    }
    _SG_TRACE_ARGS(update_image, img_id, data);
}

SOKOL_API_IMPL void sg_push_debug_group(const char* name) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(name);
    _sg_push_debug_group(name);
    _SG_TRACE_ARGS(push_debug_group, name);
}

SOKOL_API_IMPL void sg_pop_debug_group(void) {
    SOKOL_ASSERT(_sg.valid);
    _sg_pop_debug_group();
    _SG_TRACE_NOARGS(pop_debug_group);
}

SOKOL_API_IMPL bool sg_add_commit_listener(sg_commit_listener listener) {
    SOKOL_ASSERT(_sg.valid);
    return _sg_add_commit_listener(&listener);
}

SOKOL_API_IMPL bool sg_remove_commit_listener(sg_commit_listener listener) {
    SOKOL_ASSERT(_sg.valid);
    return _sg_remove_commit_listener(&listener);
}

SOKOL_API_IMPL void sg_enable_frame_stats(void) {
    SOKOL_ASSERT(_sg.valid);
    _sg.stats_enabled = true;
}

SOKOL_API_IMPL void sg_disable_frame_stats(void) {
    SOKOL_ASSERT(_sg.valid);
    _sg.stats_enabled = false;
}

SOKOL_API_IMPL bool sg_frame_stats_enabled(void) {
    return _sg.stats_enabled;
}

SOKOL_API_IMPL sg_buffer_info sg_query_buffer_info(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_buffer_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        info.slot.state = buf->slot.state;
        info.slot.res_id = buf->slot.id;
        info.slot.uninit_count = buf->slot.uninit_count;
        info.update_frame_index = buf->cmn.update_frame_index;
        info.append_frame_index = buf->cmn.append_frame_index;
        info.append_pos = buf->cmn.append_pos;
        info.append_overflow = buf->cmn.append_overflow;
        #if defined(SOKOL_D3D11)
        info.num_slots = 1;
        info.active_slot = 0;
        #else
        info.num_slots = buf->cmn.num_slots;
        info.active_slot = buf->cmn.active_slot;
        #endif
    }
    return info;
}

SOKOL_API_IMPL sg_image_info sg_query_image_info(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_image_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        info.slot.state = img->slot.state;
        info.slot.res_id = img->slot.id;
        info.slot.uninit_count = img->slot.uninit_count;
        info.upd_frame_index = img->cmn.upd_frame_index;
        #if defined(SOKOL_D3D11)
        info.num_slots = 1;
        info.active_slot = 0;
        #else
        info.num_slots = img->cmn.num_slots;
        info.active_slot = img->cmn.active_slot;
        #endif
    }
    return info;
}

SOKOL_API_IMPL sg_sampler_info sg_query_sampler_info(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_sampler_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        info.slot.state = smp->slot.state;
        info.slot.res_id = smp->slot.id;
        info.slot.uninit_count = smp->slot.uninit_count;
    }
    return info;
}

SOKOL_API_IMPL sg_shader_info sg_query_shader_info(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_shader_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        info.slot.state = shd->slot.state;
        info.slot.res_id = shd->slot.id;
        info.slot.uninit_count = shd->slot.uninit_count;
    }
    return info;
}

SOKOL_API_IMPL sg_pipeline_info sg_query_pipeline_info(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_pipeline_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        info.slot.state = pip->slot.state;
        info.slot.res_id = pip->slot.id;
        info.slot.uninit_count = pip->slot.uninit_count;
    }
    return info;
}

SOKOL_API_IMPL sg_attachments_info sg_query_attachments_info(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_attachments_info info;
    _sg_clear(&info, sizeof(info));
    const _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        info.slot.state = atts->slot.state;
        info.slot.res_id = atts->slot.id;
        info.slot.uninit_count = atts->slot.uninit_count;
    }
    return info;
}

SOKOL_API_IMPL sg_buffer_desc sg_query_buffer_desc(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_buffer_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        desc.size = (size_t)buf->cmn.size;
        desc.usage = buf->cmn.usage;
    }
    return desc;
}

SOKOL_API_IMPL size_t sg_query_buffer_size(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        return (size_t)buf->cmn.size;
    }
    return 0;
}

SOKOL_API_IMPL sg_buffer_usage sg_query_buffer_usage(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_buffer_usage usg;
    _sg_clear(&usg, sizeof(usg));
    const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
    if (buf) {
        usg = buf->cmn.usage;
    }
    return usg;
}

SOKOL_API_IMPL sg_image_desc sg_query_image_desc(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_image_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        desc.type = img->cmn.type;
        desc.width = img->cmn.width;
        desc.height = img->cmn.height;
        desc.num_slices = img->cmn.num_slices;
        desc.num_mipmaps = img->cmn.num_mipmaps;
        desc.usage = img->cmn.usage;
        desc.pixel_format = img->cmn.pixel_format;
        desc.sample_count = img->cmn.sample_count;
    }
    return desc;
}

SOKOL_API_IMPL sg_image_type sg_query_image_type(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.type;
    }
    return _SG_IMAGETYPE_DEFAULT;
}

SOKOL_API_IMPL int sg_query_image_width(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.width;
    }
    return 0;
}

SOKOL_API_IMPL int sg_query_image_height(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.height;
    }
    return 0;
}

SOKOL_API_IMPL int sg_query_image_num_slices(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.num_slices;
    }
    return 0;
}

SOKOL_API_IMPL int sg_query_image_num_mipmaps(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.num_mipmaps;
    }
    return 0;
}

SOKOL_API_IMPL sg_pixel_format sg_query_image_pixelformat(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.pixel_format;
    }
    return _SG_PIXELFORMAT_DEFAULT;
}

SOKOL_API_IMPL sg_image_usage sg_query_image_usage(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_image_usage usg;
    _sg_clear(&usg, sizeof(usg));
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        usg = img->cmn.usage;
    }
    return usg;
}

SOKOL_API_IMPL int sg_query_image_sample_count(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    SOKOL_ASSERT(_sg.valid);
    const _sg_image_t* img = _sg_lookup_image(img_id.id);
    if (img) {
        return img->cmn.sample_count;
    }
    return 0;
}

SOKOL_API_IMPL sg_sampler_desc sg_query_sampler_desc(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_sampler_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
    if (smp) {
        desc.min_filter = smp->cmn.min_filter;
        desc.mag_filter = smp->cmn.mag_filter;
        desc.mipmap_filter = smp->cmn.mipmap_filter;
        desc.wrap_u = smp->cmn.wrap_u;
        desc.wrap_v = smp->cmn.wrap_v;
        desc.wrap_w = smp->cmn.wrap_w;
        desc.min_lod = smp->cmn.min_lod;
        desc.max_lod = smp->cmn.max_lod;
        desc.border_color = smp->cmn.border_color;
        desc.compare = smp->cmn.compare;
        desc.max_anisotropy = smp->cmn.max_anisotropy;
    }
    return desc;
}

SOKOL_API_IMPL sg_shader_desc sg_query_shader_desc(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_shader_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
    if (shd) {
        for (size_t ub_idx = 0; ub_idx < SG_MAX_UNIFORMBLOCK_BINDSLOTS; ub_idx++) {
            sg_shader_uniform_block* ub_desc = &desc.uniform_blocks[ub_idx];
            const _sg_shader_uniform_block_t* ub = &shd->cmn.uniform_blocks[ub_idx];
            ub_desc->stage = ub->stage;
            ub_desc->size = ub->size;
        }
        for (size_t sbuf_idx = 0; sbuf_idx < SG_MAX_STORAGEBUFFER_BINDSLOTS; sbuf_idx++) {
            sg_shader_storage_buffer* sbuf_desc = &desc.storage_buffers[sbuf_idx];
            const _sg_shader_storage_buffer_t* sbuf = &shd->cmn.storage_buffers[sbuf_idx];
            sbuf_desc->stage = sbuf->stage;
            sbuf_desc->readonly = sbuf->readonly;
        }
        for (size_t simg_idx = 0; simg_idx < SG_MAX_STORAGE_ATTACHMENTS; simg_idx++) {
            sg_shader_storage_image* simg_desc = &desc.storage_images[simg_idx];
            const _sg_shader_storage_image_t* simg = &shd->cmn.storage_images[simg_idx];
            simg_desc->stage = simg->stage;
            simg_desc->access_format = simg->access_format;
            simg_desc->image_type = simg->image_type;
            simg_desc->writeonly = simg->writeonly;
        }
        for (size_t img_idx = 0; img_idx < SG_MAX_IMAGE_BINDSLOTS; img_idx++) {
            sg_shader_image* img_desc = &desc.images[img_idx];
            const _sg_shader_image_t* img = &shd->cmn.images[img_idx];
            img_desc->stage = img->stage;
            img_desc->image_type = img->image_type;
            img_desc->sample_type = img->sample_type;
            img_desc->multisampled = img->multisampled;
        }
        for (size_t smp_idx = 0; smp_idx < SG_MAX_SAMPLER_BINDSLOTS; smp_idx++) {
            sg_shader_sampler* smp_desc = &desc.samplers[smp_idx];
            const _sg_shader_sampler_t* smp = &shd->cmn.samplers[smp_idx];
            smp_desc->stage = smp->stage;
            smp_desc->sampler_type = smp->sampler_type;
        }
        for (size_t img_smp_idx = 0; img_smp_idx < SG_MAX_IMAGE_SAMPLER_PAIRS; img_smp_idx++) {
            sg_shader_image_sampler_pair* img_smp_desc = &desc.image_sampler_pairs[img_smp_idx];
            const _sg_shader_image_sampler_t* img_smp = &shd->cmn.image_samplers[img_smp_idx];
            img_smp_desc->stage = img_smp->stage;
            img_smp_desc->image_slot = img_smp->image_slot;
            img_smp_desc->sampler_slot = img_smp->sampler_slot;
        }
    }
    return desc;
}

SOKOL_API_IMPL sg_pipeline_desc sg_query_pipeline_desc(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_pipeline_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
    if (pip) {
        desc.compute = pip->cmn.is_compute;
        desc.shader.id = pip->cmn.shader.sref.id;
        desc.layout = pip->cmn.layout;
        desc.depth = pip->cmn.depth;
        desc.stencil = pip->cmn.stencil;
        desc.color_count = pip->cmn.color_count;
        for (int i = 0; i < pip->cmn.color_count; i++) {
            desc.colors[i] = pip->cmn.colors[i];
        }
        desc.primitive_type = pip->cmn.primitive_type;
        desc.index_type = pip->cmn.index_type;
        desc.cull_mode = pip->cmn.cull_mode;
        desc.face_winding = pip->cmn.face_winding;
        desc.sample_count = pip->cmn.sample_count;
        desc.blend_color = pip->cmn.blend_color;
        desc.alpha_to_coverage_enabled = pip->cmn.alpha_to_coverage_enabled;
    }
    return desc;
}

SOKOL_API_IMPL sg_attachments_desc sg_query_attachments_desc(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_attachments_desc desc;
    _sg_clear(&desc, sizeof(desc));
    const _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
    if (atts) {
        for (int i = 0; i < atts->cmn.num_colors; i++) {
            desc.colors[i].image.id = atts->cmn.colors[i].image.sref.id;
            desc.colors[i].mip_level = atts->cmn.colors[i].mip_level;
            desc.colors[i].slice = atts->cmn.colors[i].slice;
        }
        desc.depth_stencil.image.id = atts->cmn.depth_stencil.image.sref.id;
        desc.depth_stencil.mip_level = atts->cmn.depth_stencil.mip_level;
        desc.depth_stencil.slice = atts->cmn.depth_stencil.slice;
    }
    return desc;
}

SOKOL_API_IMPL sg_buffer_desc sg_query_buffer_defaults(const sg_buffer_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_buffer_desc_defaults(desc);
}

SOKOL_API_IMPL sg_image_desc sg_query_image_defaults(const sg_image_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_image_desc_defaults(desc);
}

SOKOL_API_IMPL sg_sampler_desc sg_query_sampler_defaults(const sg_sampler_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_sampler_desc_defaults(desc);
}

SOKOL_API_IMPL sg_shader_desc sg_query_shader_defaults(const sg_shader_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_shader_desc_defaults(desc);
}

SOKOL_API_IMPL sg_pipeline_desc sg_query_pipeline_defaults(const sg_pipeline_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_pipeline_desc_defaults(desc);
}

SOKOL_API_IMPL sg_attachments_desc sg_query_attachments_defaults(const sg_attachments_desc* desc) {
    SOKOL_ASSERT(_sg.valid && desc);
    return _sg_attachments_desc_defaults(desc);
}

SOKOL_API_IMPL const void* sg_d3d11_device(void) {
    #if defined(SOKOL_D3D11)
        return (const void*) _sg.d3d11.dev;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_d3d11_device_context(void) {
    #if defined(SOKOL_D3D11)
        return (const void*) _sg.d3d11.ctx;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL sg_d3d11_buffer_info sg_d3d11_query_buffer_info(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_buffer_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
        if (buf) {
            res.buf = (const void*) buf->d3d11.buf;
        }
    #else
        _SOKOL_UNUSED(buf_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_d3d11_image_info sg_d3d11_query_image_info(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_image_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_image_t* img = _sg_lookup_image(img_id.id);
        if (img) {
            res.tex2d = (const void*) img->d3d11.tex2d;
            res.tex3d = (const void*) img->d3d11.tex3d;
            res.res = (const void*) img->d3d11.res;
            res.srv = (const void*) img->d3d11.srv;
        }
    #else
        _SOKOL_UNUSED(img_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_d3d11_sampler_info sg_d3d11_query_sampler_info(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_sampler_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
        if (smp) {
            res.smp = (const void*) smp->d3d11.smp;
        }
    #else
        _SOKOL_UNUSED(smp_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_d3d11_shader_info sg_d3d11_query_shader_info(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_shader_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
        if (shd) {
            for (size_t i = 0; i < SG_MAX_UNIFORMBLOCK_BINDSLOTS; i++) {
                res.cbufs[i] = (const void*) shd->d3d11.all_cbufs[i];
            }
            res.vs = (const void*) shd->d3d11.vs;
            res.fs = (const void*) shd->d3d11.fs;
        }
    #else
        _SOKOL_UNUSED(shd_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_d3d11_pipeline_info sg_d3d11_query_pipeline_info(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_pipeline_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
        if (pip) {
            res.il = (const void*) pip->d3d11.il;
            res.rs = (const void*) pip->d3d11.rs;
            res.dss = (const void*) pip->d3d11.dss;
            res.bs = (const void*) pip->d3d11.bs;
        }
    #else
        _SOKOL_UNUSED(pip_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_d3d11_attachments_info sg_d3d11_query_attachments_info(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_d3d11_attachments_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_D3D11)
        const _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
        if (atts) {
            for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
                res.color_rtv[i] = (const void*) atts->d3d11.colors[i].view.rtv;
                res.resolve_rtv[i] = (const void*) atts->d3d11.resolves[i].view.rtv;
            }
            res.dsv = (const void*) atts->d3d11.depth_stencil.view.dsv;
        }
    #else
        _SOKOL_UNUSED(atts_id);
    #endif
    return res;
}

SOKOL_API_IMPL const void* sg_mtl_device(void) {
    #if defined(SOKOL_METAL)
        if (nil != _sg.mtl.device) {
            return (__bridge const void*) _sg.mtl.device;
        } else {
            return 0;
        }
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_mtl_render_command_encoder(void) {
    #if defined(SOKOL_METAL)
        if (nil != _sg.mtl.render_cmd_encoder) {
            return (__bridge const void*) _sg.mtl.render_cmd_encoder;
        } else {
            return 0;
        }
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_mtl_compute_command_encoder(void) {
    #if defined(SOKOL_METAL)
        if (nil != _sg.mtl.compute_cmd_encoder) {
            return (__bridge const void*) _sg.mtl.compute_cmd_encoder;
        } else {
            return 0;
        }
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL sg_mtl_buffer_info sg_mtl_query_buffer_info(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_mtl_buffer_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_METAL)
        const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
        if (buf) {
            for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
                if (buf->mtl.buf[i] != 0) {
                    res.buf[i] = (__bridge void*) _sg_mtl_id(buf->mtl.buf[i]);
                }
            }
            res.active_slot = buf->cmn.active_slot;
        }
    #else
        _SOKOL_UNUSED(buf_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_mtl_image_info sg_mtl_query_image_info(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_mtl_image_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_METAL)
        const _sg_image_t* img = _sg_lookup_image(img_id.id);
        if (img) {
            for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
                if (img->mtl.tex[i] != 0) {
                    res.tex[i] = (__bridge void*) _sg_mtl_id(img->mtl.tex[i]);
                }
            }
            res.active_slot = img->cmn.active_slot;
        }
    #else
        _SOKOL_UNUSED(img_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_mtl_sampler_info sg_mtl_query_sampler_info(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_mtl_sampler_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_METAL)
        const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
        if (smp) {
            if (smp->mtl.sampler_state != 0) {
                res.smp = (__bridge void*) _sg_mtl_id(smp->mtl.sampler_state);
            }
        }
    #else
        _SOKOL_UNUSED(smp_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_mtl_shader_info sg_mtl_query_shader_info(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_mtl_shader_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_METAL)
        const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
        if (shd) {
            const int vertex_lib  = shd->mtl.vertex_func.mtl_lib;
            const int vertex_func = shd->mtl.vertex_func.mtl_func;
            const int fragment_lib  = shd->mtl.fragment_func.mtl_lib;
            const int fragment_func = shd->mtl.fragment_func.mtl_func;
            if (vertex_lib != 0) {
                res.vertex_lib = (__bridge void*) _sg_mtl_id(vertex_lib);
            }
            if (fragment_lib != 0) {
                res.fragment_lib = (__bridge void*) _sg_mtl_id(fragment_lib);
            }
            if (vertex_func != 0) {
                res.vertex_func = (__bridge void*) _sg_mtl_id(vertex_func);
            }
            if (fragment_func != 0) {
                res.fragment_func = (__bridge void*) _sg_mtl_id(fragment_func);
            }
        }
    #else
        _SOKOL_UNUSED(shd_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_mtl_pipeline_info sg_mtl_query_pipeline_info(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_mtl_pipeline_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_METAL)
        const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
        if (pip) {
            if (pip->mtl.rps != 0) {
                res.rps = (__bridge void*) _sg_mtl_id(pip->mtl.rps);
            }
            if (pip->mtl.dss != 0) {
                res.dss = (__bridge void*) _sg_mtl_id(pip->mtl.dss);
            }
        }
    #else
        _SOKOL_UNUSED(pip_id);
    #endif
    return res;
}

SOKOL_API_IMPL const void* sg_wgpu_device(void) {
    #if defined(SOKOL_WGPU)
        return (const void*) _sg.wgpu.dev;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_wgpu_queue(void) {
    #if defined(SOKOL_WGPU)
        return (const void*) _sg.wgpu.queue;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_wgpu_command_encoder(void) {
    #if defined(SOKOL_WGPU)
        return (const void*) _sg.wgpu.cmd_enc;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_wgpu_render_pass_encoder(void) {
    #if defined(SOKOL_WGPU)
        return (const void*) _sg.wgpu.rpass_enc;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL const void* sg_wgpu_compute_pass_encoder(void) {
    #if defined(SOKOL_WGPU)
        return (const void*) _sg.wgpu.cpass_enc;
    #else
        return 0;
    #endif
}

SOKOL_API_IMPL sg_wgpu_buffer_info sg_wgpu_query_buffer_info(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_buffer_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
        if (buf) {
            res.buf = (const void*) buf->wgpu.buf;
        }
    #else
        _SOKOL_UNUSED(buf_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_wgpu_image_info sg_wgpu_query_image_info(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_image_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_image_t* img = _sg_lookup_image(img_id.id);
        if (img) {
            res.tex = (const void*) img->wgpu.tex;
            res.view = (const void*) img->wgpu.view;
        }
    #else
        _SOKOL_UNUSED(img_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_wgpu_sampler_info sg_wgpu_query_sampler_info(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_sampler_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
        if (smp) {
            res.smp = (const void*) smp->wgpu.smp;
        }
    #else
        _SOKOL_UNUSED(smp_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_wgpu_shader_info sg_wgpu_query_shader_info(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_shader_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
        if (shd) {
            res.vs_mod = (const void*) shd->wgpu.vertex_func.module;
            res.fs_mod = (const void*) shd->wgpu.fragment_func.module;
            res.bgl = (const void*) shd->wgpu.bgl_img_smp_sbuf;
        }
    #else
        _SOKOL_UNUSED(shd_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_wgpu_pipeline_info sg_wgpu_query_pipeline_info(sg_pipeline pip_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_pipeline_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_pipeline_t* pip = _sg_lookup_pipeline(pip_id.id);
        if (pip) {
            res.render_pipeline = (const void*) pip->wgpu.rpip;
            res.compute_pipeline = (const void*) pip->wgpu.cpip;
        }
    #else
        _SOKOL_UNUSED(pip_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_wgpu_attachments_info sg_wgpu_query_attachments_info(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_wgpu_attachments_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(SOKOL_WGPU)
        const _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
        if (atts) {
            for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
                res.color_view[i] = (const void*) atts->wgpu.colors[i].view;
                res.resolve_view[i] = (const void*) atts->wgpu.resolves[i].view;
            }
            res.ds_view = (const void*) atts->wgpu.depth_stencil.view;
        }
    #else
        _SOKOL_UNUSED(atts_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_gl_buffer_info sg_gl_query_buffer_info(sg_buffer buf_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_gl_buffer_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(_SOKOL_ANY_GL)
        const _sg_buffer_t* buf = _sg_lookup_buffer(buf_id.id);
        if (buf) {
            for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
                res.buf[i] = buf->gl.buf[i];
            }
            res.active_slot = buf->cmn.active_slot;
        }
    #else
        _SOKOL_UNUSED(buf_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_gl_image_info sg_gl_query_image_info(sg_image img_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_gl_image_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(_SOKOL_ANY_GL)
        const _sg_image_t* img = _sg_lookup_image(img_id.id);
        if (img) {
            for (int i = 0; i < SG_NUM_INFLIGHT_FRAMES; i++) {
                res.tex[i] = img->gl.tex[i];
            }
            res.tex_target = img->gl.target;
            res.msaa_render_buffer = img->gl.msaa_render_buffer;
            res.active_slot = img->cmn.active_slot;
        }
    #else
        _SOKOL_UNUSED(img_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_gl_sampler_info sg_gl_query_sampler_info(sg_sampler smp_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_gl_sampler_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(_SOKOL_ANY_GL)
        const _sg_sampler_t* smp = _sg_lookup_sampler(smp_id.id);
        if (smp) {
            res.smp = smp->gl.smp;
        }
    #else
        _SOKOL_UNUSED(smp_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_gl_shader_info sg_gl_query_shader_info(sg_shader shd_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_gl_shader_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(_SOKOL_ANY_GL)
        const _sg_shader_t* shd = _sg_lookup_shader(shd_id.id);
        if (shd) {
            res.prog = shd->gl.prog;
        }
    #else
        _SOKOL_UNUSED(shd_id);
    #endif
    return res;
}

SOKOL_API_IMPL sg_gl_attachments_info sg_gl_query_attachments_info(sg_attachments atts_id) {
    SOKOL_ASSERT(_sg.valid);
    sg_gl_attachments_info res;
    _sg_clear(&res, sizeof(res));
    #if defined(_SOKOL_ANY_GL)
        const _sg_attachments_t* atts = _sg_lookup_attachments(atts_id.id);
        if (atts) {
            res.framebuffer = atts->gl.fb;
            for (int i = 0; i < SG_MAX_COLOR_ATTACHMENTS; i++) {
                res.msaa_resolve_framebuffer[i] = atts->gl.msaa_resolve_framebuffer[i];
            }
        }
    #else
        _SOKOL_UNUSED(atts_id);
    #endif
    return res;
}

#ifdef _MSC_VER
#pragma warning(pop)
#endif

#endif // SOKOL_GFX_IMPL