[FFmpeg-cvslog] cabac: x86 version of get_cabac_bypass

[Jason Garrett-Glaser]

ffmpeg | branch: master | Jason Garrett-Glaser <jason at x264.com> | Mon Jul  1 12:06:22 2013 -0700| [d222f6e39e263c3455b9e1108cccda035889622d] | committer: Luca Barbato

cabac: x86 version of get_cabac_bypass

Signed-off-by: Luca Barbato <lu_zero at gentoo.org>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d222f6e39e263c3455b9e1108cccda035889622d
---

 libavcodec/cabac_functions.h |    3 ++-
 libavcodec/x86/cabac.h       |   40 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 42 insertions(+), 1 deletion(-)

diff --git a/libavcodec/cabac_functions.h b/libavcodec/cabac_functions.h
index 484ba85..afc7c98 100644
--- a/libavcodec/cabac_functions.h
+++ b/libavcodec/cabac_functions.h
@@ -113,6 +113,7 @@ static int av_unused get_cabac(CABACContext *c, uint8_t * const state){
     return get_cabac_inline(c,state);
 }
 
+#ifndef get_cabac_bypass
 static int av_unused get_cabac_bypass(CABACContext *c){
     int range;
     c->low += c->low;
@@ -128,7 +129,7 @@ static int av_unused get_cabac_bypass(CABACContext *c){
         return 1;
     }
 }
-
+#endif
 
 #ifndef get_cabac_bypass_sign
 static av_always_inline int get_cabac_bypass_sign(CABACContext *c, int val){
diff --git a/libavcodec/x86/cabac.h b/libavcodec/x86/cabac.h
index a74cf0b..fdb0a29 100644
--- a/libavcodec/x86/cabac.h
+++ b/libavcodec/x86/cabac.h
@@ -229,5 +229,45 @@ static av_always_inline int get_cabac_bypass_sign_x86(CABACContext *c, int val)
     return val;
 }
 
+#define get_cabac_bypass get_cabac_bypass_x86
+static av_always_inline int get_cabac_bypass_x86(CABACContext *c)
+{
+    x86_reg tmp;
+    int res;
+    __asm__ volatile(
+        "movl        %c6(%2), %k1       \n\t"
+        "movl        %c3(%2), %%eax     \n\t"
+        "shl             $17, %k1       \n\t"
+        "add           %%eax, %%eax     \n\t"
+        "sub             %k1, %%eax     \n\t"
+        "cltd                           \n\t"
+        "and           %%edx, %k1       \n\t"
+        "add             %k1, %%eax     \n\t"
+        "inc           %%edx            \n\t"
+        "test           %%ax, %%ax      \n\t"
+        "jnz              1f            \n\t"
+        "mov         %c4(%2), %1        \n\t"
+        "subl        $0xFFFF, %%eax     \n\t"
+        "movzwl         (%1), %%ecx     \n\t"
+        "bswap         %%ecx            \n\t"
+        "shrl            $15, %%ecx     \n\t"
+        "addl          %%ecx, %%eax     \n\t"
+        "cmp         %c5(%2), %1        \n\t"
+        "jge              1f            \n\t"
+        "add"OPSIZE"      $2, %c4(%2)   \n\t"
+        "1:                             \n\t"
+        "movl          %%eax, %c3(%2)   \n\t"
+
+        : "=&d"(res), "=&r"(tmp)
+        : "r"(c),
+          "i"(offsetof(CABACContext, low)),
+          "i"(offsetof(CABACContext, bytestream)),
+          "i"(offsetof(CABACContext, bytestream_end)),
+          "i"(offsetof(CABACContext, range))
+        : "%eax", "%ecx", "memory"
+    );
+    return res;
+}
+
 #endif /* HAVE_INLINE_ASM */
 #endif /* AVCODEC_X86_CABAC_H */