Move math helpers to lj_vmmath.c. Add missing log2/exp2 for Symbian.

src/Makefile

@@ -359,7 +359,7 @@ LJLIB_C= $(LJLIB_O:.o=.c)
 
 LJCORE_O= lj_gc.o lj_err.o lj_char.o lj_bc.o lj_obj.o \
 	  lj_str.o lj_tab.o lj_func.o lj_udata.o lj_meta.o \
-	  lj_state.o lj_dispatch.o lj_vmevent.o lj_api.o \
+	  lj_state.o lj_dispatch.o lj_vmevent.o lj_vmmath.o lj_api.o \
 	  lj_lex.o lj_parse.o \
 	  lj_ir.o lj_opt_mem.o lj_opt_fold.o lj_opt_narrow.o \
 	  lj_opt_dce.o lj_opt_loop.o lj_opt_split.o \

src/Makefile.dep

@@ -166,22 +166,24 @@ lj_udata.o: lj_udata.c lj_obj.h lua.h luaconf.h lj_def.h lj_arch.h \
 lj_vmevent.o: lj_vmevent.c lj_obj.h lua.h luaconf.h lj_def.h lj_arch.h \
  lj_str.h lj_tab.h lj_state.h lj_dispatch.h lj_bc.h lj_jit.h lj_ir.h \
  lj_vm.h lj_vmevent.h
+lj_vmmath.o: lj_vmmath.c lj_obj.h lua.h luaconf.h lj_def.h lj_arch.h \
+ lj_ir.h lj_vm.h
 ljamalg.o: ljamalg.c lua.h luaconf.h lauxlib.h lj_gc.c lj_obj.h lj_def.h \
  lj_arch.h lj_gc.h lj_err.h lj_errmsg.h lj_str.h lj_tab.h lj_func.h \
  lj_udata.h lj_meta.h lj_state.h lj_frame.h lj_bc.h lj_ctype.h lj_cdata.h \
  lj_trace.h lj_jit.h lj_ir.h lj_dispatch.h lj_traceerr.h lj_vm.h lj_err.c \
  lj_ff.h lj_ffdef.h lj_char.c lj_char.h lj_bc.c lj_bcdef.h lj_obj.c \
  lj_str.c lj_tab.c lj_func.c lj_udata.c lj_meta.c lj_state.c lj_lex.h \
- lj_alloc.h lj_dispatch.c luajit.h lj_vmevent.c lj_vmevent.h lj_api.c \
- lj_parse.h lj_lex.c lualib.h lj_parse.c lj_ctype.c lj_cdata.c lj_cconv.h \
- lj_cconv.c lj_ccall.c lj_ccall.h lj_carith.c lj_carith.h lj_clib.c \
- lj_clib.h lj_cparse.c lj_cparse.h lj_lib.c lj_lib.h lj_ir.c lj_ircall.h \
- lj_iropt.h lj_opt_mem.c lj_opt_fold.c lj_folddef.h lj_opt_narrow.c \
- lj_opt_dce.c lj_opt_loop.c lj_snap.h lj_opt_split.c lj_mcode.c \
- lj_mcode.h lj_snap.c lj_target.h lj_target_*.h lj_record.c lj_record.h \
- lj_ffrecord.h lj_crecord.c lj_crecord.h lj_ffrecord.c lj_recdef.h \
- lj_asm.c lj_asm.h lj_emit_*.h lj_asm_*.h lj_trace.c lj_gdbjit.h \
- lj_gdbjit.c lj_alloc.c lib_aux.c lib_base.c lj_libdef.h lib_math.c \
- lib_string.c lib_table.c lib_io.c lib_os.c lib_package.c lib_debug.c \
- lib_bit.c lib_jit.c lib_ffi.c lib_init.c
+ lj_alloc.h lj_dispatch.c luajit.h lj_vmevent.c lj_vmevent.h lj_vmmath.c \
+ lj_api.c lj_parse.h lj_lex.c lualib.h lj_parse.c lj_ctype.c lj_cdata.c \
+ lj_cconv.h lj_cconv.c lj_ccall.c lj_ccall.h lj_carith.c lj_carith.h \
+ lj_clib.c lj_clib.h lj_cparse.c lj_cparse.h lj_lib.c lj_lib.h lj_ir.c \
+ lj_ircall.h lj_iropt.h lj_opt_mem.c lj_opt_fold.c lj_folddef.h \
+ lj_opt_narrow.c lj_opt_dce.c lj_opt_loop.c lj_snap.h lj_opt_split.c \
+ lj_mcode.c lj_mcode.h lj_snap.c lj_target.h lj_target_*.h lj_record.c \
+ lj_record.h lj_ffrecord.h lj_crecord.c lj_crecord.h lj_ffrecord.c \
+ lj_recdef.h lj_asm.c lj_asm.h lj_emit_*.h lj_asm_*.h lj_trace.c \
+ lj_gdbjit.h lj_gdbjit.c lj_alloc.c lib_aux.c lib_base.c lj_libdef.h \
+ lib_math.c lib_string.c lib_table.c lib_io.c lib_os.c lib_package.c \
+ lib_debug.c lib_bit.c lib_jit.c lib_ffi.c lib_init.c
 luajit.o: luajit.c lua.h luaconf.h lauxlib.h lualib.h luajit.h lj_arch.h

src/buildvm_x64.h

@@ -879,7 +879,7 @@ enum {
   GLOB_ff_math_log,
   GLOB_ff_math_log10,
   GLOB_ff_math_exp,
-  GLOB_vm_exp,
+  GLOB_vm_exp_x87,
   GLOB_ff_math_sin,
   GLOB_ff_math_cos,
   GLOB_ff_math_tan,
@@ -939,7 +939,7 @@ enum {
   GLOB_vm_ceil_sse,
   GLOB_vm_trunc_sse,
   GLOB_vm_mod,
-  GLOB_vm_exp2,
+  GLOB_vm_exp2_x87,
   GLOB_vm_exp2raw,
   GLOB_vm_pow_sse,
   GLOB_vm_powi_sse,
@@ -1040,7 +1040,7 @@ static const char *const globnames[] = {
   "ff_math_log",
   "ff_math_log10",
   "ff_math_exp",
-  "vm_exp",
+  "vm_exp_x87",
   "ff_math_sin",
   "ff_math_cos",
   "ff_math_tan",
@@ -1100,7 +1100,7 @@ static const char *const globnames[] = {
   "vm_ceil_sse",
   "vm_trunc_sse",
   "vm_mod",
-  "vm_exp2",
+  "vm_exp2_x87",
   "vm_exp2raw",
   "vm_pow_sse",
   "vm_powi_sse",
@@ -1132,9 +1132,9 @@ static const char *const extnames[] = {
   "lj_tab_next",
   "lj_tab_getinth@8",
   "lj_ffh_coroutine_wrap_err@8",
-  "lj_wrapper_sinh",
-  "lj_wrapper_cosh",
-  "lj_wrapper_tanh",
+  "lj_vm_sinh",
+  "lj_vm_cosh",
+  "lj_vm_tanh",
   "lj_str_new",
   "lj_tab_len@4",
   "lj_gc_step@4",

src/buildvm_x64win.h

@@ -876,7 +876,7 @@ enum {
   GLOB_ff_math_log,
   GLOB_ff_math_log10,
   GLOB_ff_math_exp,
-  GLOB_vm_exp,
+  GLOB_vm_exp_x87,
   GLOB_ff_math_sin,
   GLOB_ff_math_cos,
   GLOB_ff_math_tan,
@@ -936,7 +936,7 @@ enum {
   GLOB_vm_ceil_sse,
   GLOB_vm_trunc_sse,
   GLOB_vm_mod,
-  GLOB_vm_exp2,
+  GLOB_vm_exp2_x87,
   GLOB_vm_exp2raw,
   GLOB_vm_pow_sse,
   GLOB_vm_powi_sse,
@@ -1037,7 +1037,7 @@ static const char *const globnames[] = {
   "ff_math_log",
   "ff_math_log10",
   "ff_math_exp",
-  "vm_exp",
+  "vm_exp_x87",
   "ff_math_sin",
   "ff_math_cos",
   "ff_math_tan",
@@ -1097,7 +1097,7 @@ static const char *const globnames[] = {
   "vm_ceil_sse",
   "vm_trunc_sse",
   "vm_mod",
-  "vm_exp2",
+  "vm_exp2_x87",
   "vm_exp2raw",
   "vm_pow_sse",
   "vm_powi_sse",
@@ -1128,9 +1128,9 @@ static const char *const extnames[] = {
   "lj_tab_next",
   "lj_tab_getinth@8",
   "lj_ffh_coroutine_wrap_err@8",
-  "lj_wrapper_sinh",
-  "lj_wrapper_cosh",
-  "lj_wrapper_tanh",
+  "lj_vm_sinh",
+  "lj_vm_cosh",
+  "lj_vm_tanh",
   "lj_str_new",
   "lj_tab_len@4",
   "lj_gc_step@4",

src/buildvm_x86.dasc

@@ -2028,7 +2028,7 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
   }
   |.ffunc_n math_log, fldln2;	fyl2x;		jmp ->fff_resn
   |.ffunc_n math_log10, fldlg2;	fyl2x;		jmp ->fff_resn
-  |.ffunc_n math_exp;	call ->vm_exp;		jmp ->fff_resn
+  |.ffunc_n math_exp;	call ->vm_exp_x87;	jmp ->fff_resn
   |
   |.ffunc_n math_sin;	fsin;			jmp ->fff_resn
   |.ffunc_n math_cos;	fcos;			jmp ->fff_resn
@@ -2055,7 +2055,7 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
   |  .endif
   ||}
   |  mov RB, BASE
-  |  call extern lj_wrapper_ .. func
+  |  call extern lj_vm_ .. func
   |  mov BASE, RB
   |  .if X64
   |    jmp ->fff_resxmm0
@@ -3169,9 +3169,9 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
   |// FP exponentiation e^x and 2^x. Called by math.exp fast function and
   |// from JIT code. Arg/ret on x87 stack. No int/xmm regs modified.
   |// Caveat: needs 3 slots on x87 stack!
-  |->vm_exp:
+  |->vm_exp_x87:
   |  fldl2e; fmulp st1				// e^x ==> 2^(x*log2(e))
-  |->vm_exp2:
+  |->vm_exp2_x87:
   |  .if X64WIN
   |    .define expscratch, dword [rsp+8]	// Use scratch area.
   |  .elif X64
@@ -3444,11 +3444,11 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
     |  cmp fpmop, 5; ja >2
     |  .if X64WIN; pop rax; .endif
     |  je >1
-    |  call ->vm_exp
+    |  call ->vm_exp_x87
     |  .if X64WIN; push rax; .endif
     |  jmp >7
     |1:
-    |  call ->vm_exp2
+    |  call ->vm_exp2_x87
     |  .if X64WIN; push rax; .endif
     |  jmp >7
     |2: ; cmp fpmop, 7; je >1; ja >2
@@ -3487,7 +3487,7 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
     |  fld qword [esp+4]
     |  ret
     |2: ; fld qword [esp+4]
-    |  cmp fpmop, 5; jb ->vm_exp; je ->vm_exp2
+    |  cmp fpmop, 5; jb ->vm_exp_x87; je ->vm_exp2_x87
     |2: ; cmp fpmop, 7; je >1; ja >2
     |  fldln2; fxch; fyl2x; ret
     |1: ; fld1; fxch; fyl2x; ret
@@ -3505,7 +3505,7 @@ static void build_subroutines(BuildCtx *ctx, int cmov, int sse)
     |  cmp fpmop, 1; jb ->vm_floor; je ->vm_ceil
     |  cmp fpmop, 3; jb ->vm_trunc; ja >2
     |  fsqrt; ret
-    |2: ; cmp fpmop, 5; jb ->vm_exp; je ->vm_exp2
+    |2: ; cmp fpmop, 5; jb ->vm_exp_x87; je ->vm_exp2_x87
     |  cmp fpmop, 7; je >1; ja >2
     |  fldln2; fxch; fyl2x; ret
     |1: ; fld1; fxch; fyl2x; ret

src/buildvm_x86.h

@@ -925,7 +925,7 @@ enum {
   GLOB_ff_math_log,
   GLOB_ff_math_log10,
   GLOB_ff_math_exp,
-  GLOB_vm_exp,
+  GLOB_vm_exp_x87,
   GLOB_ff_math_sin,
   GLOB_ff_math_cos,
   GLOB_ff_math_tan,
@@ -985,7 +985,7 @@ enum {
   GLOB_vm_ceil_sse,
   GLOB_vm_trunc_sse,
   GLOB_vm_mod,
-  GLOB_vm_exp2,
+  GLOB_vm_exp2_x87,
   GLOB_vm_exp2raw,
   GLOB_vm_pow_sse,
   GLOB_vm_powi_sse,
@@ -1086,7 +1086,7 @@ static const char *const globnames[] = {
   "ff_math_log",
   "ff_math_log10",
   "ff_math_exp",
-  "vm_exp",
+  "vm_exp_x87",
   "ff_math_sin",
   "ff_math_cos",
   "ff_math_tan",
@@ -1146,7 +1146,7 @@ static const char *const globnames[] = {
   "vm_ceil_sse",
   "vm_trunc_sse",
   "vm_mod",
-  "vm_exp2",
+  "vm_exp2_x87",
   "vm_exp2raw",
   "vm_pow_sse",
   "vm_powi_sse",
@@ -1177,9 +1177,9 @@ static const char *const extnames[] = {
   "lj_tab_next",
   "lj_tab_getinth@8",
   "lj_ffh_coroutine_wrap_err@8",
-  "lj_wrapper_sinh",
-  "lj_wrapper_cosh",
-  "lj_wrapper_tanh",
+  "lj_vm_sinh",
+  "lj_vm_cosh",
+  "lj_vm_tanh",
   "lj_str_new",
   "lj_tab_len@4",
   "lj_gc_step@4",

src/lib_math.c

@@ -84,10 +84,6 @@ LJLIB_ASM_(math_max)		LJLIB_REC(math_minmax IR_MAX)
 LJLIB_PUSH(3.14159265358979323846) LJLIB_SET(pi)
 LJLIB_PUSH(1e310) LJLIB_SET(huge)
 
-LJ_FUNCA double lj_wrapper_sinh(double x) { return sinh(x); }
-LJ_FUNCA double lj_wrapper_cosh(double x) { return cosh(x); }
-LJ_FUNCA double lj_wrapper_tanh(double x) { return tanh(x); }
-
 /* ------------------------------------------------------------------------ */
 
 /* This implements a Tausworthe PRNG with period 2^223. Based on:

src/lj_asm_x86.h

@@ -1591,8 +1591,8 @@ static void asm_fpmath(ASMState *as, IRIns *ir)
     }
     emit_rmro(as, XO_FSTPq, XOg_FSTPq, RID_ESP, ofs);
     switch (fpm) {  /* st0 = lj_vm_*(st0) */
-    case IRFPM_EXP: emit_call(as, lj_vm_exp); break;
-    case IRFPM_EXP2: emit_call(as, lj_vm_exp2); break;
+    case IRFPM_EXP: emit_call(as, lj_vm_exp_x87); break;
+    case IRFPM_EXP2: emit_call(as, lj_vm_exp2_x87); break;
     case IRFPM_SIN: emit_x87op(as, XI_FSIN); break;
     case IRFPM_COS: emit_x87op(as, XI_FCOS); break;
     case IRFPM_TAN: emit_x87op(as, XI_FPOP); emit_x87op(as, XI_FPTAN); break;

src/lj_ir.c

@@ -37,72 +37,6 @@
 /* Pass IR on to next optimization in chain (FOLD). */
 #define emitir(ot, a, b)        (lj_ir_set(J, (ot), (a), (b)), lj_opt_fold(J))
 
-/* -- Helper functions for generated machine code ------------------------- */
-
-#ifdef __ANDROID__
-/* Android doesn't have log2(). Oh well. */
-#define log2 lj_vm_log2
-static double lj_vm_log2(double a)
-{
-  return log(a) * 1.4426950408889634074;
-}
-#endif
-
-#if !LJ_TARGET_X86ORX64
-/* Unsigned x^k. */
-static double lj_vm_powui(double x, uint32_t k)
-{
-  double y;
-  lua_assert(k != 0);
-  for (; (k & 1) == 0; k >>= 1) x *= x;
-  y = x;
-  if ((k >>= 1) != 0) {
-    for (;;) {
-      x *= x;
-      if (k == 1) break;
-      if (k & 1) y *= x;
-      k >>= 1;
-    }
-    y *= x;
-  }
-  return y;
-}
-
-/* Signed x^k. */
-static double lj_vm_powi(double x, int32_t k)
-{
-  if (k > 1)
-    return lj_vm_powui(x, (uint32_t)k);
-  else if (k == 1)
-    return x;
-  else if (k == 0)
-    return 1;
-  else
-    return 1.0 / lj_vm_powui(x, (uint32_t)-k);
-}
-
-/* Computes fpm(x) for extended math functions. */
-double lj_vm_foldfpm(double x, int fpm)
-{
-  switch (fpm) {
-  case IRFPM_FLOOR: return lj_vm_floor(x);
-  case IRFPM_CEIL: return lj_vm_ceil(x);
-  case IRFPM_TRUNC: return lj_vm_trunc(x);
-  case IRFPM_SQRT: return sqrt(x);
-  case IRFPM_EXP: return exp(x);
-  case IRFPM_EXP2: return exp2(x);
-  case IRFPM_LOG: return log(x);
-  case IRFPM_LOG2: return log2(x);
-  case IRFPM_LOG10: return log10(x);
-  case IRFPM_SIN: return sin(x);
-  case IRFPM_COS: return cos(x);
-  case IRFPM_TAN: return tan(x);
-  default: lua_assert(0);
-  }
-  return 0;
-}
-#endif
-
 /* -- IR tables ----------------------------------------------------------- */
 
 /* IR instruction modes. */

src/lj_ircall.h

@@ -91,9 +91,9 @@ typedef struct CCallInfo {
   _(lj_vm_trunc,	ARG1_FP,   N, NUM, 0) \
   _(sqrt,		ARG1_FP,   N, NUM, 0) \
   _(exp,		ARG1_FP,   N, NUM, 0) \
-  _(exp2,		ARG1_FP,   N, NUM, 0) \
+  _(lj_vm_exp2,		ARG1_FP,   N, NUM, 0) \
   _(log,		ARG1_FP,   N, NUM, 0) \
-  _(log2,		ARG1_FP,   N, NUM, 0) \
+  _(lj_vm_log2,		ARG1_FP,   N, NUM, 0) \
   _(log10,		ARG1_FP,   N, NUM, 0) \
   _(sin,		ARG1_FP,   N, NUM, 0) \
   _(cos,		ARG1_FP,   N, NUM, 0) \

src/lj_opt_split.c

@@ -274,7 +274,7 @@ static void split_ir(jit_State *J)
 	    IRIns *irm12 = IR(irm3->op1);
 	    IRIns *irl1 = IR(irm12->op1);
 	    if (irm12->op1 > J->loopref && irl1->o == IR_CALLN &&
-		irl1->op2 == IRCALL_log2) {
+		irl1->op2 == IRCALL_lj_vm_log2) {
 	      IRRef tmp = irl1->op1;  /* Recycle first two args from LOG2. */
 	      IRRef arg3 = irm3->op2, arg4 = irm4->op2;
 	      J->cur.nins--;

src/lj_vm.h

@@ -51,14 +51,25 @@ LJ_ASMF void lj_vm_exit_interp(void);
 LJ_ASMF void lj_vm_floor_sse(void);
 LJ_ASMF void lj_vm_ceil_sse(void);
 LJ_ASMF void lj_vm_trunc_sse(void);
-LJ_ASMF void lj_vm_exp(void);
-LJ_ASMF void lj_vm_exp2(void);
+LJ_ASMF void lj_vm_exp_x87(void);
+LJ_ASMF void lj_vm_exp2_x87(void);
 LJ_ASMF void lj_vm_pow_sse(void);
 LJ_ASMF void lj_vm_powi_sse(void);
 #else
 LJ_ASMF double lj_vm_floor(double);
 LJ_ASMF double lj_vm_ceil(double);
 LJ_ASMF double lj_vm_trunc(double);
+LJ_ASMF double lj_vm_powi(double, int32_t);
+#if defined(__ANDROID__) || defined(__symbian__)
+LJ_ASMF double lj_vm_log2(double);
+#else
+#define lj_vm_log2	log2
+#endif
+#if defined(__symbian__)
+LJ_ASMF double lj_vm_exp2(double);
+#else
+#define lj_vm_exp2	exp2
+#endif
 #endif
 #endif
 

src/lj_vmmath.c

@@ -0,0 +1,97 @@
+/*
+** Math helper functions for assembler VM.
+** Copyright (C) 2005-2011 Mike Pall. See Copyright Notice in luajit.h
+*/
+
+#define lj_vmmath_c
+#define LUA_CORE
+
+#include <math.h>
+
+#include "lj_obj.h"
+#if LJ_HASJIT
+#include "lj_ir.h"
+#endif
+#include "lj_vm.h"
+
+/* -- Helper functions for generated machine code ------------------------- */
+
+#if LJ_TARGET_X86ORX64
+/* Wrapper functions to avoid linker issues on OSX. */
+LJ_FUNCA double lj_vm_sinh(double x) { return sinh(x); }
+LJ_FUNCA double lj_vm_cosh(double x) { return cosh(x); }
+LJ_FUNCA double lj_vm_tanh(double x) { return tanh(x); }
+#endif
+
+#if LJ_HASJIT
+
+#if defined(__ANDROID__) || defined(__symbian__)
+double lj_vm_log2(double a)
+{
+  return log(a) * 1.4426950408889634074;
+}
+#endif
+
+#if defined(__symbian__)
+double lj_vm_exp2(double a)
+{
+  return exp(a * 0.6931471805599453);
+}
+#endif
+
+#if !LJ_TARGET_X86ORX64
+/* Unsigned x^k. */
+static double lj_vm_powui(double x, uint32_t k)
+{
+  double y;
+  lua_assert(k != 0);
+  for (; (k & 1) == 0; k >>= 1) x *= x;
+  y = x;
+  if ((k >>= 1) != 0) {
+    for (;;) {
+      x *= x;
+      if (k == 1) break;
+      if (k & 1) y *= x;
+      k >>= 1;
+    }
+    y *= x;
+  }
+  return y;
+}
+
+/* Signed x^k. */
+double lj_vm_powi(double x, int32_t k)
+{
+  if (k > 1)
+    return lj_vm_powui(x, (uint32_t)k);
+  else if (k == 1)
+    return x;
+  else if (k == 0)
+    return 1;
+  else
+    return 1.0 / lj_vm_powui(x, (uint32_t)-k);
+}
+
+/* Computes fpm(x) for extended math functions. */
+double lj_vm_foldfpm(double x, int fpm)
+{
+  switch (fpm) {
+  case IRFPM_FLOOR: return lj_vm_floor(x);
+  case IRFPM_CEIL: return lj_vm_ceil(x);
+  case IRFPM_TRUNC: return lj_vm_trunc(x);
+  case IRFPM_SQRT: return sqrt(x);
+  case IRFPM_EXP: return exp(x);
+  case IRFPM_EXP2: return lj_vm_exp2(x);
+  case IRFPM_LOG: return log(x);
+  case IRFPM_LOG2: return lj_vm_log2(x);
+  case IRFPM_LOG10: return log10(x);
+  case IRFPM_SIN: return sin(x);
+  case IRFPM_COS: return cos(x);
+  case IRFPM_TAN: return tan(x);
+  default: lua_assert(0);
+  }
+  return 0;
+}
+#endif
+
+#endif

src/ljamalg.c

@@ -41,6 +41,7 @@
 #include "lj_state.c"
 #include "lj_dispatch.c"
 #include "lj_vmevent.c"
+#include "lj_vmmath.c"
 #include "lj_api.c"
 #include "lj_lex.c"
 #include "lj_parse.c"