1 files changed, 44 insertions, 66 deletions

diff --git a/src/lj_vmmath.c b/src/lj_vmmath.c
index 2da5f6b7..3351e72b 100644
--- a/src/lj_vmmath.c
+++ b/src/lj_vmmath.c
@@ -13,16 +13,40 @@
 #include "lj_ir.h"
 #include "lj_vm.h"
 
-/* -- Helper functions for generated machine code ------------------------- */
+/* -- Wrapper functions --------------------------------------------------- */
 
-#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); }
+#if LJ_TARGET_X86 && __ELF__ && __PIC__
+/* Wrapper functions to deal with the ELF/x86 PIC disaster. */
+LJ_FUNCA double lj_wrap_log(double x) { return log(x); }
+LJ_FUNCA double lj_wrap_log10(double x) { return log10(x); }
+LJ_FUNCA double lj_wrap_exp(double x) { return exp(x); }
+LJ_FUNCA double lj_wrap_sin(double x) { return sin(x); }
+LJ_FUNCA double lj_wrap_cos(double x) { return cos(x); }
+LJ_FUNCA double lj_wrap_tan(double x) { return tan(x); }
+LJ_FUNCA double lj_wrap_asin(double x) { return asin(x); }
+LJ_FUNCA double lj_wrap_acos(double x) { return acos(x); }
+LJ_FUNCA double lj_wrap_atan(double x) { return atan(x); }
+LJ_FUNCA double lj_wrap_sinh(double x) { return sinh(x); }
+LJ_FUNCA double lj_wrap_cosh(double x) { return cosh(x); }
+LJ_FUNCA double lj_wrap_tanh(double x) { return tanh(x); }
+LJ_FUNCA double lj_wrap_atan2(double x, double y) { return atan2(x, y); }
+LJ_FUNCA double lj_wrap_pow(double x, double y) { return pow(x, y); }
+LJ_FUNCA double lj_wrap_fmod(double x, double y) { return fmod(x, y); }
 #endif
 
-#if !LJ_TARGET_X86ORX64
+/* -- Helper functions ---------------------------------------------------- */
+
+/* Required to prevent the C compiler from applying FMA optimizations.
+**
+** Yes, there's -ffp-contract and the FP_CONTRACT pragma ... in theory.
+** But the current state of C compilers is a mess in this regard.
+** Also, this function is not performance sensitive at all.
+*/
+LJ_NOINLINE static double lj_vm_floormul(double x, double y)
+{
+  return lj_vm_floor(x / y) * y;
+}
+
 double lj_vm_foldarith(double x, double y, int op)
 {
   switch (op) {
@@ -30,42 +54,27 @@ double lj_vm_foldarith(double x, double y, int op)
   case IR_SUB - IR_ADD: return x-y; break;
   case IR_MUL - IR_ADD: return x*y; break;
   case IR_DIV - IR_ADD: return x/y; break;
-  case IR_MOD - IR_ADD: return x-lj_vm_floor(x/y)*y; break;
+  case IR_MOD - IR_ADD: return x-lj_vm_floormul(x, y); break;
   case IR_POW - IR_ADD: return pow(x, y); break;
   case IR_NEG - IR_ADD: return -x; break;
   case IR_ABS - IR_ADD: return fabs(x); break;
 #if LJ_HASJIT
-  case IR_ATAN2 - IR_ADD: return atan2(x, y); break;
   case IR_LDEXP - IR_ADD: return ldexp(x, (int)y); break;
-  case IR_MIN - IR_ADD: return x > y ? y : x; break;
-  case IR_MAX - IR_ADD: return x < y ? y : x; break;
+  case IR_MIN - IR_ADD: return x < y ? x : y; break;
+  case IR_MAX - IR_ADD: return x > y ? x : y; break;
 #endif
   default: return x;
   }
 }
-#endif
 
-#if LJ_HASJIT
-
-#ifdef LUAJIT_NO_LOG2
-double lj_vm_log2(double a)
-{
-  return log(a) * 1.4426950408889634074;
-}
-#endif
-
-#ifdef LUAJIT_NO_EXP2
-double lj_vm_exp2(double a)
-{
-  return exp(a * 0.6931471805599453);
-}
-#endif
+/* -- Helper functions for generated machine code ------------------------- */
 
-#if !(LJ_TARGET_ARM || LJ_TARGET_PPC)
+#if (LJ_HASJIT && !(LJ_TARGET_ARM || LJ_TARGET_ARM64 || LJ_TARGET_PPC)) || LJ_TARGET_MIPS
 int32_t LJ_FASTCALL lj_vm_modi(int32_t a, int32_t b)
 {
   uint32_t y, ua, ub;
-  lua_assert(b != 0);  /* This must be checked before using this function. */
+  /* This must be checked before using this function. */
+  lj_assertX(b != 0, "modulo with zero divisor");
   ua = a < 0 ? ~(uint32_t)a+1u : (uint32_t)a;
   ub = b < 0 ? ~(uint32_t)b+1u : (uint32_t)b;
   y = ua % ub;
@@ -75,38 +84,14 @@ int32_t LJ_FASTCALL lj_vm_modi(int32_t a, int32_t b)
 }
 #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;
-}
+#if LJ_HASJIT
 
-/* Signed x^k. */
-double lj_vm_powi(double x, int32_t k)
+#ifdef LUAJIT_NO_LOG2
+double lj_vm_log2(double a)
 {
-  if (k > 1)
-    return lj_vm_powui(x, (uint32_t)k);
-  else if (k == 1)
-    return x;
-  else if (k == 0)
-    return 1.0;
-  else
-    return 1.0 / lj_vm_powui(x, ~(uint32_t)k+1u);
+  return log(a) * 1.4426950408889634074;
 }
+#endif
 
 /* Computes fpm(x) for extended math functions. */
 double lj_vm_foldfpm(double x, int fpm)
@@ -116,19 +101,12 @@ double lj_vm_foldfpm(double x, int fpm)
   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);
+  default: lj_assertX(0, "bad fpm %d", fpm);
   }
   return 0;
 }
-#endif
 
 #if LJ_HASFFI
 int lj_vm_errno(void)