On Wed, Jul 13, 2011 at 01:33:58PM +0100, Måns Rullgård wrote: > Diego Biurrun <[email protected]> writes: > > > On Wed, Jul 13, 2011 at 12:54:46PM +0100, Mans Rullgard wrote: > >> These buffers have int16_t elements so the padding at the end needs > >> to be two bytes, not one. > >> > >> --- a/libswscale/utils.c > >> +++ b/libswscale/utils.c > >> @@ -1045,19 +1045,19 @@ int sws_init_context(SwsContext *c, SwsFilter > >> *srcFilter, SwsFilter *dstFilter) > >> /* align at 16 bytes for AltiVec */ > >> for (i=0; i<c->vLumBufSize; i++) { > >> - FF_ALLOCZ_OR_GOTO(c, c->lumPixBuf[i+c->vLumBufSize], > >> dst_stride+1, fail); > >> + FF_ALLOCZ_OR_GOTO(c, c->lumPixBuf[i+c->vLumBufSize], > >> dst_stride+2, fail); > > > > Could this be done with sizeof or similar instead of magic numbers? > > I guess sizeof(c->lumPixBuf[0]) would work. It's a bit of a mouthful though.
Yes, it's longish, but I think the magic number gave rise to the bug in the first place, so I thought I'd raise the point. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
