[FFmpeg-cvslog] h264: change deblock_h_chroma_8_mmxext() to prevent valgrind confusion.

Michael Niedermayer git at videolan.org
Fri Feb 17 21:41:19 CET 2012


ffmpeg | branch: master | Michael Niedermayer <michaelni at gmx.at> | Fri Feb 17 20:41:33 2012 +0100| [f9caec0cf9a2bd06efa3bc800649713e8531cad0] | committer: Michael Niedermayer

h264: change deblock_h_chroma_8_mmxext() to prevent valgrind confusion.

Signed-off-by: Michael Niedermayer <michaelni at gmx.at>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f9caec0cf9a2bd06efa3bc800649713e8531cad0
---

 libavcodec/x86/h264_deblock.asm |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/libavcodec/x86/h264_deblock.asm b/libavcodec/x86/h264_deblock.asm
index 9f517bf..08625e4 100644
--- a/libavcodec/x86/h264_deblock.asm
+++ b/libavcodec/x86/h264_deblock.asm
@@ -841,7 +841,11 @@ cglobal deblock_h_chroma_8_mmxext, 5,7
     TRANSPOSE4x8_LOAD  bw, wd, dq, PASS8ROWS(t5, r0, r1, t6)
     movq  buf0, m0
     movq  buf1, m3
-    call ff_chroma_inter_body_mmxext
+    LOAD_MASK  r2d, r3d
+    movd       m6, [r4] ; tc0
+    punpcklbw  m6, m6
+    pand       m7, m6
+    DEBLOCK_P0_Q0
     movq  m0, buf0
     movq  m3, buf1
     TRANSPOSE8x4B_STORE PASS8ROWS(t5, r0, r1, t6)



More information about the ffmpeg-cvslog mailing list