avformat: Avoid allocation for AVFormatInternal

Do this by allocating AVFormatContext together with the data that is
currently in AVFormatInternal; or rather: Put AVFormatContext at the
beginning of a new structure called FFFormatContext (which encompasses
more than just the internal fields and is a proper context in its own
right, hence the name) and remove AVFormatInternal altogether.

The biggest simplifications occured in avformat_alloc_context(), where
one can now simply call avformat_free_context() in case of errors.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This commit is contained in:
Andreas Rheinhardt
2021-08-24 14:58:07 +02:00
parent dfbf41775c
commit fed0282508
32 changed files with 135 additions and 138 deletions

View File

@@ -441,9 +441,10 @@ static int64_t dv_frame_offset(AVFormatContext *s, DVDemuxContext *c,
int64_t timestamp, int flags)
{
// FIXME: sys may be wrong if last dv_read_packet() failed (buffer is junk)
FFFormatContext *const si = ffformatcontext(s);
const int frame_size = c->sys->frame_size;
int64_t offset;
int64_t size = avio_size(s->pb) - s->internal->data_offset;
int64_t size = avio_size(s->pb) - si->data_offset;
int64_t max_offset = ((size - 1) / frame_size) * frame_size;
offset = frame_size * timestamp;
@@ -453,7 +454,7 @@ static int64_t dv_frame_offset(AVFormatContext *s, DVDemuxContext *c,
else if (offset < 0)
offset = 0;
return offset + s->internal->data_offset;
return offset + si->data_offset;
}
void ff_dv_offset_reset(DVDemuxContext *c, int64_t frame_offset)