Merge branch 'master' into issue1244/compute-ms2

.github/workflows/gen_bindings.yml

@@ -295,8 +295,8 @@ jobs:
         with:
           repository: kassane/sokol-d
       - uses: mlugg/setup-zig@v1
-        with:
-          version: master
+        #with:
+        #  version: master
       - uses: dlang-community/setup-dlang@v1
         with:
           compiler: ldc-master

CHANGELOG.md

@@ -1,5 +1,24 @@
 ## Updates
 
+### 07-May-2025
+
+A memory alignment fix for raw mouse input on Windows via `GetRawInputData()`.
+Raw mouse input is used for 'mouse-lock mode', not for regular mouse input.
+`GetRawInputData()` requires the output buffer to be 8-byte aligned, but
+sokol_app.h didn't guarantee this (and in reality it doesn't really seem to
+matter, the problem only showed up when running under ASAN - props to Zig). The
+new solution works the same as GLFW: first a dummy `GetRawInputData()` is
+performed to query the required buffer size, then a buffer is allocated on
+demand (which typically only happens once), then the actual `GetRawInputData()`
+call is performed. This depends on `malloc()` returning memory that's at least
+8-byte aligned (malloc on 64-bit Windows guarantees 16-byte alignment).
+
+Many thanks to @roig for noticing the issue and providing an initial fix
+via PR https://github.com/floooh/sokol/pull/1262 (which I merged but then replaced
+with the above heap-allocation solution - this is safe for any data size returned
+by `GetRawInputData()` - I'm not actually sure though if anything greater than
+`sizeof(RAWINPUTDATA)` is ever returned.
+
 ### 05-Apr-2025
 
 - Compute shaders are now supported on platforms that support GLES3.1

bindgen/README.md

@@ -24,7 +24,7 @@ To update the Zig bindings:
 > git clone https://github.com/floooh/sokol-nim
 > git clone https://github.com/floooh/sokol-odin
 > git clone https://github.com/floooh/sokol-rust
-> git clone https://github.com/floooh/sokol-d
+> git clone https://github.com/kassane/sokol-d
 > git clone https://github.com/colinbellino/sokol-jai
 > git clone https://github.com/floooh/sokol-c3
 > python3 gen_all.py

sokol_app.h

@@ -2675,7 +2675,10 @@ typedef struct {
         bool tracked;
         uint8_t capture_mask;
     } mouse;
-    uint8_t raw_input_data[256];
+    struct {
+        size_t size;
+        void* ptr;
+    } raw_input_data;
 } _sapp_win32_t;
 
 #if defined(SOKOL_GLCORE)
@@ -7254,6 +7257,32 @@ _SOKOL_PRIVATE void _sapp_win32_lock_mouse(bool lock) {
     _sapp.win32.mouse.requested_lock = lock;
 }
 
+_SOKOL_PRIVATE void _sapp_win32_free_raw_input_data(void) {
+    if (_sapp.win32.raw_input_data.ptr) {
+        _sapp_free(_sapp.win32.raw_input_data.ptr);
+        _sapp.win32.raw_input_data.ptr = 0;
+        _sapp.win32.raw_input_data.size = 0;
+    }
+}
+
+_SOKOL_PRIVATE void _sapp_win32_alloc_raw_input_data(size_t size) {
+    SOKOL_ASSERT(!_sapp.win32.raw_input_data.ptr);
+    SOKOL_ASSERT(size > 0);
+    _sapp.win32.raw_input_data.ptr = _sapp_malloc(size);
+    _sapp.win32.raw_input_data.size = size;
+    SOKOL_ASSERT(_sapp.win32.raw_input_data.ptr);
+}
+
+_SOKOL_PRIVATE void* _sapp_win32_ensure_raw_input_data(size_t required_size) {
+    if (required_size > _sapp.win32.raw_input_data.size) {
+        _sapp_win32_free_raw_input_data();
+        _sapp_win32_alloc_raw_input_data(required_size);
+    }
+    // we expect that malloc() returns at least 8-byte aligned memory
+    SOKOL_ASSERT((((uintptr_t)_sapp.win32.raw_input_data.ptr) & 7) == 0);
+    return _sapp.win32.raw_input_data.ptr;
+}
+
 _SOKOL_PRIVATE void _sapp_win32_do_lock_mouse(void) {
     _sapp.mouse.locked = true;
 
@@ -7669,13 +7698,18 @@ _SOKOL_PRIVATE LRESULT CALLBACK _sapp_win32_wndproc(HWND hWnd, UINT uMsg, WPARAM
                 /* raw mouse input during mouse-lock */
                 if (_sapp.mouse.locked) {
                     HRAWINPUT ri = (HRAWINPUT) lParam;
-                    UINT size = sizeof(_sapp.win32.raw_input_data);
                     // see: https://docs.microsoft.com/en-us/windows/win32/api/winuser/nf-winuser-getrawinputdata
-                    if ((UINT)-1 == GetRawInputData(ri, RID_INPUT, &_sapp.win32.raw_input_data, &size, sizeof(RAWINPUTHEADER))) {
+                    // also see: https://github.com/glfw/glfw/blob/e7ea71be039836da3a98cea55ae5569cb5eb885c/src/win32_window.c#L912-L924
+
+                    // first poll for required size to alloc/grow input buffer, then get the actual data
+                    UINT size = 0;
+                    GetRawInputData(ri, RID_INPUT, NULL, &size, sizeof(RAWINPUTHEADER));
+                    void* raw_input_data_ptr = _sapp_win32_ensure_raw_input_data(size);
+                    if ((UINT)-1 == GetRawInputData(ri, RID_INPUT, raw_input_data_ptr, &size, sizeof(RAWINPUTHEADER))) {
                         _SAPP_ERROR(WIN32_GET_RAW_INPUT_DATA_FAILED);
                         break;
                     }
-                    const RAWINPUT* raw_mouse_data = (const RAWINPUT*) &_sapp.win32.raw_input_data;
+                    const RAWINPUT* raw_mouse_data = (const RAWINPUT*) raw_input_data_ptr;
                     if (raw_mouse_data->data.mouse.usFlags & MOUSE_MOVE_ABSOLUTE) {
                         /* mouse only reports absolute position
                            NOTE: This code is untested and will most likely behave wrong in Remote Desktop sessions.
@@ -8216,6 +8250,7 @@ _SOKOL_PRIVATE void _sapp_win32_run(const sapp_desc* desc) {
     _sapp_win32_destroy_window();
     _sapp_win32_destroy_icons();
     _sapp_win32_restore_console();
+    _sapp_win32_free_raw_input_data();
     _sapp_discard_state();
 }
 

sokol_gfx.h

@@ -452,7 +452,7 @@
         sg_apply_scissor_rect[f]
         sg_draw
 
-    The folling function may be called inside a render- or compute-pass, but
+    The following function may be called inside a render- or compute-pass, but
     not outside a pass:
 
         sg_apply_pipeline
@@ -818,7 +818,7 @@
         - for the GLES3 backend, source code must be provided in '#version 300 es' syntax
         - 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 preferrably in HLSL5.0 syntax, note that when
+          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
@@ -883,7 +883,7 @@
         - 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 assumbed to be written to by the
+          read/write storage buffer binding is assumed to be written to by the
           compute shader.
         - a backend-specific bind slot:
             - D3D11/HLSL:
@@ -978,7 +978,7 @@
         - 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 WebGPI:   https://github.com/floooh/sokol-samples/tree/master/wgpu
+        - for WebGPU:   https://github.com/floooh/sokol-samples/tree/master/wgpu
 
 
     ON SG_IMAGESAMPLETYPE_UNFILTERABLE_FLOAT AND SG_SAMPLERTYPE_NONFILTERING