Hi, On Fri, Oct 14, 2011 at 7:06 PM, Jason Garrett-Glaser <[email protected]> wrote: > On Fri, Oct 14, 2011 at 3:21 PM, Janne Grunau <[email protected]> wrote: >> From: "Ronald S. Bultje" <[email protected]> >> >> Signed-off-by: Janne Grunau <[email protected]> >> --- >> libavcodec/vp8.c | 1 + >> 1 files changed, 1 insertions(+), 0 deletions(-) >> >> diff --git a/libavcodec/vp8.c b/libavcodec/vp8.c >> index 691324e..7442b99 100644 >> --- a/libavcodec/vp8.c >> +++ b/libavcodec/vp8.c >> @@ -492,6 +492,7 @@ void decode_mvs(VP8Context *s, VP8Macroblock *mb, int >> mb_x, int mb_y) >> >> AV_ZERO32(&near_mv[0]); >> AV_ZERO32(&near_mv[1]); >> + AV_ZERO32(&near_mv[2]); > > Is this actually necessary? Reading uninitialized-but-valid memory is > okay if the value isn't actually used.
So I saw this value being used in motion compensation for the actual file that triggered this. If it hadn't been used there, I indeed wouldn't have cared, but if it's used for MC, then it may lead to bad stuff... I can ask if we can make the particular file that triggered it available for further analysis. Ronald _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
