[FFmpeg-cvslog] x86: XOP/FMA4 CPU detection support

[Jason Garrett-Glaser]

ffmpeg | branch: master | Jason Garrett-Glaser <jason at x264.com> | Mon Sep 26 14:44:47 2011 -0700| [96a59cf37b080080b7e45dd57828b40a7a2bbfe7] | committer: Jason Garrett-Glaser

x86: XOP/FMA4 CPU detection support

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=96a59cf37b080080b7e45dd57828b40a7a2bbfe7
---

 libavutil/avutil.h  |    4 ++--
 libavutil/cpu.c     |    2 ++
 libavutil/cpu.h     |    2 ++
 libavutil/x86/cpu.c |    9 +++++++++
 4 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/libavutil/avutil.h b/libavutil/avutil.h
index 5d378ce..d89b23c 100644
--- a/libavutil/avutil.h
+++ b/libavutil/avutil.h
@@ -40,8 +40,8 @@
 #define AV_VERSION(a, b, c) AV_VERSION_DOT(a, b, c)
 
 #define LIBAVUTIL_VERSION_MAJOR 51
-#define LIBAVUTIL_VERSION_MINOR 10
-#define LIBAVUTIL_VERSION_MICRO  2
+#define LIBAVUTIL_VERSION_MINOR 11
+#define LIBAVUTIL_VERSION_MICRO  0
 
 #define LIBAVUTIL_VERSION_INT   AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \
                                                LIBAVUTIL_VERSION_MINOR, \
diff --git a/libavutil/cpu.c b/libavutil/cpu.c
index baa7922..25895d6 100644
--- a/libavutil/cpu.c
+++ b/libavutil/cpu.c
@@ -60,6 +60,8 @@ static const struct {
     { AV_CPU_FLAG_SSE4,      "sse4.1"     },
     { AV_CPU_FLAG_SSE42,     "sse4.2"     },
     { AV_CPU_FLAG_AVX,       "avx"        },
+    { AV_CPU_FLAG_XOP,       "xop"        },
+    { AV_CPU_FLAG_FMA4,      "fma4"       },
     { AV_CPU_FLAG_3DNOW,     "3dnow"      },
     { AV_CPU_FLAG_3DNOWEXT,  "3dnowext"   },
 #endif
diff --git a/libavutil/cpu.h b/libavutil/cpu.h
index 777cdc0..df7bf44 100644
--- a/libavutil/cpu.h
+++ b/libavutil/cpu.h
@@ -38,6 +38,8 @@
 #define AV_CPU_FLAG_SSE4         0x0100 ///< Penryn SSE4.1 functions
 #define AV_CPU_FLAG_SSE42        0x0200 ///< Nehalem SSE4.2 functions
 #define AV_CPU_FLAG_AVX          0x4000 ///< AVX functions: requires OS support even if YMM registers aren't used
+#define AV_CPU_FLAG_XOP          0x0400 ///< Bulldozer XOP functions
+#define AV_CPU_FLAG_FMA4         0x0800 ///< Bulldozer FMA4 functions
 #define AV_CPU_FLAG_IWMMXT       0x0100 ///< XScale IWMMXT
 #define AV_CPU_FLAG_ALTIVEC      0x0001 ///< standard
 
diff --git a/libavutil/x86/cpu.c b/libavutil/x86/cpu.c
index f747e4d..3975e68 100644
--- a/libavutil/x86/cpu.c
+++ b/libavutil/x86/cpu.c
@@ -133,6 +133,15 @@ int ff_get_cpu_flags_x86(void)
             rval & AV_CPU_FLAG_SSE2 && !(ecx & 0x00000040)) {
             rval |= AV_CPU_FLAG_SSE2SLOW;
         }
+
+        /* XOP and FMA4 use the AVX instruction coding scheme, so they can't be
+         * used unless the OS has AVX support. */
+        if (rval & AV_CPU_FLAG_AVX) {
+            if (ecx & 0x00000800)
+                rval |= AV_CPU_FLAG_XOP;
+            if (ecx & 0x00010000)
+                rval |= AV_CPU_FLAG_FMA4;
+        }
     }
 
     if (!strncmp(vendor.c, "GenuineIntel", 12)) {