[FFmpeg-cvslog] swscale: fix crashes in yuv2yuvX on x86-32.

Ronald S. Bultje git at videolan.org
Tue Feb 14 03:43:17 CET 2012


ffmpeg | branch: master | Ronald S. Bultje <rsbultje at gmail.com> | Mon Feb 13 13:28:50 2012 -0800| [771bab7f573fdb4a5c9456ecaabf15c5a14a22b0] | committer: Ronald S. Bultje

swscale: fix crashes in yuv2yuvX on x86-32.

They were introduced in an earlier commit that introduced use of named
arguments. One cause was a typo, a second cause appears to be a bug in
x264asm that I work around by not using named arguments.

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=771bab7f573fdb4a5c9456ecaabf15c5a14a22b0
---

 libswscale/x86/output.asm |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/libswscale/x86/output.asm b/libswscale/x86/output.asm
index 6017bf1..c4daa82 100644
--- a/libswscale/x86/output.asm
+++ b/libswscale/x86/output.asm
@@ -59,7 +59,7 @@ SECTION .text
 %macro yuv2planeX_fn 3
 
 %if ARCH_X86_32
-%define cntr_reg filterq
+%define cntr_reg fltsizeq
 %define movsx mov
 %else
 %define cntr_reg r11
@@ -146,7 +146,7 @@ cglobal yuv2planeX_%1, %3, 7, %2, filter, fltsize, src, dst, w, dither, offset
     mova            m1, [yuv2yuvX_%1_start]
     mova            m2,  m1
 %endif ; %1 == 8/9/10/16
-    movsx     cntr_reg,  fltsizem
+    movsx     cntr_reg,  r1m ; FIXME should be fltsizem, but the assembler does the wrong thing b/c of SUB above
 .filterloop_ %+ %%i:
     ; input pixels
     mov             r6, [srcq+gprsize*cntr_reg-2*gprsize]



More information about the ffmpeg-cvslog mailing list