Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
Diego Biurrun di...@biurrun.de writes:
---
The #define indirection will be reused by the yasm version of cpuid_test,
so I'm adding it here already to avoid code churn.
libavutil/x86/cpu.c | 27 ++-
1 files changed, 18 insertions(+), 9 deletions(-)
diff --git a/libavutil/x86/cpu.c b/libavutil/x86/cpu.c
index 9cf68e4..e3a9fb6 100644
--- a/libavutil/x86/cpu.c
+++ b/libavutil/x86/cpu.c
@@ -86,6 +86,23 @@
#endif /* HAVE_INLINE_ASM */
+#if HAVE_INLINE_ASM || HAVE_RWEFLAGS
+#define have_cpuid cpuid_test
+
+static int cpuid_test(void)
+{
+x86_reg a, c;
+
+/* Check if CPUID is supported by attempting to toggle the ID bit in
+ * the EFLAGS register. */
+get_eflags(a);
+set_eflags(a ^ 0x20);
+get_eflags(c);
+
+return a != c;
+}
+#endif
+
/* Function to test if multimedia instructions are supported... */
int ff_get_cpu_flags_x86(void)
{
@@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void)
union { int i[3]; char c[12]; } vendor;
#if ARCH_X86_32
-x86_reg a, c;
-
-/* Check if CPUID is supported by attempting to toggle the ID bit in
- * the EFLAGS register. */
-get_eflags(a);
-set_eflags(a ^ 0x20);
-get_eflags(c);
-
-if (a == c)
+if (!have_cpuid())
return 0; /* CPUID not supported */
#endif
You could #define have_cpuid() 1 for x86_64 and get rid of this #if.
--
Måns Rullgård
m...@mansr.com
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. So if I set it unconditionally for x86_64, I have to go back to HAVE_YASM || HAVE_INLINE_ASM in 5/5 or define have_cpuid conditionally on x86_64 somehow. Not worth the trouble - let's keep it the way it is right now. Diego ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. I thought you tested for cpuid there. -- Måns Rullgård m...@mansr.com ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
On Thu, Oct 04, 2012 at 03:30:17PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. I thought you tested for cpuid there. No. Diego ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 03:30:17PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. I thought you tested for cpuid there. No. Why not? -- Måns Rullgård m...@mansr.com ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
On Thu, Oct 04, 2012 at 03:36:40PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 03:30:17PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. I thought you tested for cpuid there. No. Why not? Because there is no cpuid symbol that I could check for. The HAVE_CPUID that I remove in 3/5 denotes the compiler intrinsic. Diego ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 2/5] x86: cpu: Break out test for cpuid capabilities into separate function
Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 03:36:40PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 03:30:17PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: On Thu, Oct 04, 2012 at 02:04:27PM +0100, Måns Rullgård wrote: Diego Biurrun di...@biurrun.de writes: --- a/libavutil/x86/cpu.c +++ b/libavutil/x86/cpu.c @@ -96,15 +113,7 @@ int ff_get_cpu_flags_x86(void) #if ARCH_X86_32 -x86_reg a, c; - -/* Check if CPUID is supported by attempting to toggle the ID bit in - * the EFLAGS register. */ -get_eflags(a); -set_eflags(a ^ 0x20); -get_eflags(c); - -if (a == c) +if (!have_cpuid()) return 0; /* CPUID not supported */ #endif You could #define have_cpuid() 1 for x86_64 and get rid of this #if. But I rely on have_cpuid to mean that I have either yasm or inline asm available in 5/5. I thought you tested for cpuid there. No. Why not? Because there is no cpuid symbol that I could check for. The HAVE_CPUID that I remove in 3/5 denotes the compiler intrinsic. #define cpuid(index, eax, ebx, ecx, edx) ^^ what's that then? -- Måns Rullgård m...@mansr.com ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
