On 2012-12-12 14:39:34 -0800, Ronald S. Bultje wrote: > Hi, > > On Wed, Dec 12, 2012 at 12:30 PM, Janne Grunau <[email protected]> wrote: > > Fixes hang in HPCAMAPALQ_BRCM_B.264_s14038 while waiting on invalid > > field 2. > > --- > > libavcodec/h264.c | 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/libavcodec/h264.c b/libavcodec/h264.c > > index 546b046..d5a54e2 100644 > > --- a/libavcodec/h264.c > > +++ b/libavcodec/h264.c > > @@ -433,7 +433,8 @@ static void await_references(H264Context *h) > > ff_thread_await_progress(&ref_pic->f, > > FFMIN((row >> 1), pic_height > > - 1), > > 0); > > - } else if (FIELD_PICTURE && !ref_field_picture) { // field > > referencing one field of a frame > > + } else if (FIELD_PICTURE && > > + (!ref_field_picture || ref_field > 1)) { // > > field referencing one field of a frame or complementary field pair > > I don't understand this one. If we're referencing two fields, > shouldn't ref_field_picture automatically be true?
it isn't and the code that marks both fields of a complementory field pair as available doesn't touch field_picture. Janne _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
