diff options
| -rw-r--r-- | HISTORY | 3 | ||||
| -rw-r--r-- | dxr3memcpy.c | 16 | ||||
| -rw-r--r-- | dxr3memcpy.h | 5 |
3 files changed, 20 insertions, 4 deletions
@@ -235,4 +235,5 @@ NOTE: I havent found time to include all of the languages, will be done in pre2 - fixed output of anamorphic video, if tv aspect is configured to 16:9 in DVB setup menu thanks Seppo Ingalsuo <seppo.ingalsuo@iki.fi> - rewrote i18n support -- fixed void cDxr3Interface::SetAudioDigitalPCM() - thanks to Stephan Skrodzki <skrodzki@contcast.de> for patch
\ No newline at end of file +- fixed void cDxr3Interface::SetAudioDigitalPCM() - thanks to Stephan Skrodzki <skrodzki@contcast.de> for patch +- dxr3memcpy.c/h: fix to support older compilers like gcc-2.95
\ No newline at end of file diff --git a/dxr3memcpy.c b/dxr3memcpy.c index a8814b6..3a545f3 100644 --- a/dxr3memcpy.c +++ b/dxr3memcpy.c @@ -93,6 +93,10 @@ int d0, d1, d2; #define MMX1_MIN_LEN 0x800 /* 2K blocks */ #define MIN_LEN 0x40 /* 64-byte blocks */ + +// Test for GCC > 3.2.0 +#if GCC_VERSION > 30200 + // ================================== /* SSE note: i tried to move 128 bytes a time instead of 64 but it didn't make any measureable difference. i'm using 64 for the sake of @@ -256,7 +260,7 @@ static void * mmx2_memcpy(void * to, const void * from, size_t len) " prefetchnta 224(%0)\n" " prefetchnta 256(%0)\n" " prefetchnta 288(%0)\n" - : : "r" (from) ); + :: "r" (from) ); if(len >= MIN_LEN) { @@ -308,6 +312,8 @@ static void * mmx2_memcpy(void * to, const void * from, size_t len) return retval; } +#endif /*GCC_VERSION > 30200*/ + // ================================== static void *linux_kernel_memcpy(void *to, const void *from, size_t len) { return __memcpy(to,from,len); @@ -344,6 +350,9 @@ cDxr3MemcpyBench::cDxr3MemcpyBench(uint32_t config_flags) routine.cpu_require = 0; m_methods.push_back(routine); + // Test for GCC > 3.2.0 + #if GCC_VERSION > 30200 + // MMX optimized memcpy() routine.name = "MMX optimized memcpy()"; routine.function = mmx_memcpy; @@ -356,7 +365,7 @@ cDxr3MemcpyBench::cDxr3MemcpyBench(uint32_t config_flags) routine.cpu_require = CC_MMXEXT; m_methods.push_back(routine); - #ifndef __FreeBSD__ + # ifndef __FreeBSD__ // SSE optimized memcpy() routine.name = "SSE optimized memcpy()"; @@ -364,7 +373,8 @@ cDxr3MemcpyBench::cDxr3MemcpyBench(uint32_t config_flags) routine.cpu_require = CC_MMXEXT|CC_SSE; m_methods.push_back(routine); - #endif /*__FreeBSD__*/ + # endif /*__FreeBSD__*/ + #endif /*GCC_VERSION > 30200*/ // #endif /*ARCH_X86*/ diff --git a/dxr3memcpy.h b/dxr3memcpy.h index 71bb673..e360b32 100644 --- a/dxr3memcpy.h +++ b/dxr3memcpy.h @@ -57,6 +57,11 @@ // size of buffer for benchmark #define BUFSIZE 1024*1024 +// get gcc version +#define GCC_VERSION (__GNUC__ * 10000 \ + + __GNUC_MINOR__ * 100 \ + + __GNUC_PATCHLEVEL__) + enum { MEMCPY_PROBE = 0, MEMCPY_GLIBC, |