dnrops
/
gimp
mirror of https://github.com/GNOME/gimp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Reorged, to address bug #142907 and bug #143069. Accel implementations 

2004-06-13  Manish Singh  <yosh@gimp.org>

        * app/base/cpu-accel.c: Reorged, to address bug #142907 and
        bug #143069. Accel implementations #define HAVE_ACCEL, and cpu_accel()
        keys on that. Both PPC and X86 implementations check for __GNUC__.
        X86 stuff is only used with USE_MMX is defined. The SSE OS check
        is now checked in arch_accel(), not cpu_accel(). Finally, the
        arch x86_64 checks now are EM64T aware (which didn't matter in
        practice).
This commit is contained in:
Manish Singh 2004-06-14 00:52:38 +00:00 committed by Manish Singh
parent 7b3cc62c52
commit 0829a3b65e
3 changed files with 102 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
ChangeLog
View File

@ -1,3 +1,13 @@
2004-06-13 Manish Singh <yosh@gimp.org>
* app/base/cpu-accel.c: Reorged, to address bug #142907 and
bug #143069. Accel implementations #define HAVE_ACCEL, and cpu_accel()
keys on that. Both PPC and X86 implementations check for __GNUC__.
X86 stuff is only used with USE_MMX is defined. The SSE OS check
is now checked in arch_accel(), not cpu_accel(). Finally, the
arch x86_64 checks now are EM64T aware (which didn't matter in
practice).
2004-06-13 Philip Lafleur <plafleur@cvs.gnome.org>
* app/display/gimpdisplayshell-preview.c: use drawable_mask_bounds()

80
app/base/cpu-accel.c
View File

@ -37,8 +37,9 @@
#include "cpu-accel.h"
#if defined(ARCH_X86) && defined(USE_MMX) && defined(__GNUC__)
#ifdef ARCH_X86
#define HAVE_ACCEL 1
typedef enum
{
@ -135,6 +136,8 @@ arch_get_vendor (void)
#ifdef ARCH_X86_64
if (strcmp (id, "AuthenticAMD") == 0)
return ARCH_X86_VENDOR_AMD;
else if (strcmp (id, "GenuineIntel") == 0)
return ARCH_X86_VENDOR_INTEL;
#else
if (strcmp (id, "GenuineIntel") == 0)
return ARCH_X86_VENDOR_INTEL;
@ -287,6 +290,33 @@ arch_accel_cyrix (void)
return caps;
}
#ifdef USE_SSE
static jmp_buf sigill_return;
static void
sigill_handler (gint n)
{
longjmp (sigill_return, 1);
}
static gboolean
arch_accel_sse_os_support (void)
{
if (setjmp (sigill_return))
{
return FALSE;
}
else
{
signal (SIGILL, sigill_handler);
__asm__ __volatile__ ("xorps %xmm0, %xmm0");
signal (SIGILL, SIG_DFL);
}
return TRUE;
}
#endif /* USE_SSE */
static guint32
arch_accel (void)
{
@ -320,21 +350,20 @@ arch_accel (void)
break;
}
#ifdef USE_SSE
if ((caps & CPU_ACCEL_X86_SSE) && !arch_accel_sse_os_support ())
caps &= ~(CPU_ACCEL_X86_SSE | CPU_ACCEL_X86_SSE2);
#endif
return caps;
}
static jmp_buf sigill_return;
static void
sigill_handler (gint n)
{
longjmp (sigill_return, 1);
}
#endif /* ARCH_X86 */
#endif /* ARCH_X86 && USE_MMX && __GNUC__ */
#if defined (ARCH_PPC) && defined (USE_ALTIVEC)
#if defined (ARCH_PPC) && defined (USE_ALTIVEC) && defined(__GNUC__)
#define HAVE_ACCEL 1
static sigjmp_buf jmpbuf;
static volatile sig_atomic_t canjump = 0;
@ -366,22 +395,22 @@ arch_accel (void)
canjump = 1;
asm volatile ("mtspr 256, %0\n\t"
"vand %%v0, %%v0, %%v0"
:
: "r" (-1));
"vand %%v0, %%v0, %%v0"
:
: "r" (-1));
signal (SIGILL, SIG_DFL);
return CPU_ACCEL_PPC_ALTIVEC;
}
#endif /* ARCH_PPC */
#endif /* ARCH_PPC && USE_ALTIVEC && __GNUC__ */
guint32
cpu_accel (void)
{
#if defined (ARCH_X86) || (defined (ARCH_PPC) && defined (USE_ALTIVEC))
#ifdef HAVE_ACCEL
static guint32 accel = ~0U;
if (accel != ~0U)
@ -389,26 +418,9 @@ cpu_accel (void)
accel = arch_accel ();
#ifdef USE_SSE
/* test OS support for SSE */
if (accel & CPU_ACCEL_X86_SSE)
{
if (setjmp (sigill_return))
{
accel &= ~(CPU_ACCEL_X86_SSE | CPU_ACCEL_X86_SSE2);
}
else
{
signal (SIGILL, sigill_handler);
__asm__ __volatile__ ("xorps %xmm0, %xmm0");
signal (SIGILL, SIG_DFL);
}
}
#endif /* USE_SSE */
return accel;
#else /* !ARCH_X86 && !ARCH_PPC/USE_ALTIVEC */
#else /* !HAVE_ACCEL */
return 0;
#endif
}

80
libgimpbase/gimpcpuaccel.c
View File

@ -37,8 +37,9 @@
#include "cpu-accel.h"
#if defined(ARCH_X86) && defined(USE_MMX) && defined(__GNUC__)
#ifdef ARCH_X86
#define HAVE_ACCEL 1
typedef enum
{
@ -135,6 +136,8 @@ arch_get_vendor (void)
#ifdef ARCH_X86_64
if (strcmp (id, "AuthenticAMD") == 0)
return ARCH_X86_VENDOR_AMD;
else if (strcmp (id, "GenuineIntel") == 0)
return ARCH_X86_VENDOR_INTEL;
#else
if (strcmp (id, "GenuineIntel") == 0)
return ARCH_X86_VENDOR_INTEL;
@ -287,6 +290,33 @@ arch_accel_cyrix (void)
return caps;
}
#ifdef USE_SSE
static jmp_buf sigill_return;
static void
sigill_handler (gint n)
{
longjmp (sigill_return, 1);
}
static gboolean
arch_accel_sse_os_support (void)
{
if (setjmp (sigill_return))
{
return FALSE;
}
else
{
signal (SIGILL, sigill_handler);
__asm__ __volatile__ ("xorps %xmm0, %xmm0");
signal (SIGILL, SIG_DFL);
}
return TRUE;
}
#endif /* USE_SSE */
static guint32
arch_accel (void)
{
@ -320,21 +350,20 @@ arch_accel (void)
break;
}
#ifdef USE_SSE
if ((caps & CPU_ACCEL_X86_SSE) && !arch_accel_sse_os_support ())
caps &= ~(CPU_ACCEL_X86_SSE | CPU_ACCEL_X86_SSE2);
#endif
return caps;
}
static jmp_buf sigill_return;
static void
sigill_handler (gint n)
{
longjmp (sigill_return, 1);
}
#endif /* ARCH_X86 */
#endif /* ARCH_X86 && USE_MMX && __GNUC__ */
#if defined (ARCH_PPC) && defined (USE_ALTIVEC)
#if defined (ARCH_PPC) && defined (USE_ALTIVEC) && defined(__GNUC__)
#define HAVE_ACCEL 1
static sigjmp_buf jmpbuf;
static volatile sig_atomic_t canjump = 0;
@ -366,22 +395,22 @@ arch_accel (void)
canjump = 1;
asm volatile ("mtspr 256, %0\n\t"
"vand %%v0, %%v0, %%v0"
:
: "r" (-1));
"vand %%v0, %%v0, %%v0"
:
: "r" (-1));
signal (SIGILL, SIG_DFL);
return CPU_ACCEL_PPC_ALTIVEC;
}
#endif /* ARCH_PPC */
#endif /* ARCH_PPC && USE_ALTIVEC && __GNUC__ */
guint32
cpu_accel (void)
{
#if defined (ARCH_X86) || (defined (ARCH_PPC) && defined (USE_ALTIVEC))
#ifdef HAVE_ACCEL
static guint32 accel = ~0U;
if (accel != ~0U)
@ -389,26 +418,9 @@ cpu_accel (void)
accel = arch_accel ();
#ifdef USE_SSE
/* test OS support for SSE */
if (accel & CPU_ACCEL_X86_SSE)
{
if (setjmp (sigill_return))
{
accel &= ~(CPU_ACCEL_X86_SSE | CPU_ACCEL_X86_SSE2);
}
else
{
signal (SIGILL, sigill_handler);
__asm__ __volatile__ ("xorps %xmm0, %xmm0");
signal (SIGILL, SIG_DFL);
}
}
#endif /* USE_SSE */
return accel;
#else /* !ARCH_X86 && !ARCH_PPC/USE_ALTIVEC */
#else /* !HAVE_ACCEL */
return 0;
#endif
}