gcc/config/i386/crtfastmath.c

   1 /*
   2  * Copyright (C) 2005, 2007 Free Software Foundation, Inc.
   3  *
   4  * This file is free software; you can redistribute it and/or modify it
   5  * under the terms of the GNU General Public License as published by the
   6  * Free Software Foundation; either version 2, or (at your option) any
   7  * later version.
   8  *
   9  * In addition to the permissions in the GNU General Public License, the
  10  * Free Software Foundation gives you unlimited permission to link the
  11  * compiled version of this file with other programs, and to distribute
  12  * those programs without any restriction coming from the use of this
  13  * file.  (The General Public License restrictions do apply in other
  14  * respects; for example, they cover modification of the file, and
  15  * distribution when not linked into another program.)
  16  *
  17  * This file is distributed in the hope that it will be useful, but
  18  * WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20  * General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with this program; see the file COPYING.  If not, write to
  24  * the Free Software Foundation, 51 Franklin Street, Fifth Floor,
  25  * Boston, MA 02110-1301, USA.
  26  *
  27  *    As a special exception, if you link this library with files
  28  *    compiled with GCC to produce an executable, this does not cause
  29  *    the resulting executable to be covered by the GNU General Public License.
  30  *    This exception does not however invalidate any other reasons why
  31  *    the executable file might be covered by the GNU General Public License.
  32  */
  33
  34 #define MXCSR_DAZ (1 << 6)      /* Enable denormals are zero mode */
  35 #define MXCSR_FTZ (1 << 15)     /* Enable flush to zero mode */
  36
  37 #ifndef __x86_64__
  38 /* All 64-bit targets have SSE and DAZ;
  39    only check them explicitly for 32-bit ones. */
  40 #include "cpuid.h"
  41 #endif
  42
  43 static void __attribute__((constructor))
  44 #ifndef __x86_64__
  45 /* The i386 ABI only requires 4-byte stack alignment, so this is necessary
  46    to make sure the fxsave struct gets correct alignment.
  47    See PR27537 and PR28621.  */
  48 __attribute__ ((force_align_arg_pointer))
  49 #endif
  50 set_fast_math (void)
  51 {
  52 #ifndef __x86_64__
  53   unsigned int eax, ebx, ecx, edx;
  54
  55   if (!__get_cpuid (1, &eax, &ebx, &ecx, &edx))
  56     return;
  57
  58   if (edx & bit_SSE)
  59     {
  60       unsigned int mxcsr = __builtin_ia32_stmxcsr ();
  61
  62       mxcsr |= MXCSR_FTZ;
  63
  64       if (edx & bit_FXSAVE)
  65         {
  66           /* Check if DAZ is available.  */
  67           struct
  68             {
  69               unsigned short int cwd;
  70               unsigned short int swd;
  71               unsigned short int twd;
  72               unsigned short int fop;
  73               long int fip;
  74               long int fcs;
  75               long int foo;
  76               long int fos;
  77               long int mxcsr;
  78               long int mxcsr_mask;
  79               long int st_space[32];
  80               long int xmm_space[32];
  81               long int padding[56];
  82             } __attribute__ ((aligned (16))) fxsave;
  83
  84           __builtin_memset (&fxsave, 0, sizeof (fxsave));
  85
  86           asm volatile ("fxsave %0" : "=m" (fxsave) : "m" (fxsave));
  87
  88           if (fxsave.mxcsr_mask & MXCSR_DAZ)
  89             mxcsr |= MXCSR_DAZ;
  90         }
  91
  92       __builtin_ia32_ldmxcsr (mxcsr);
  93     }
  94 #else
  95   unsigned int mxcsr = __builtin_ia32_stmxcsr ();
  96   mxcsr |= MXCSR_DAZ | MXCSR_FTZ;
  97   __builtin_ia32_ldmxcsr (mxcsr);
  98 #endif
  99 }