On Fri, Dec 12, 2014 at 11:32 AM, Luca Barbato <[email protected]> wrote:
> On 12/12/14 04:35, Vittorio Giovara wrote:
>> Signed-off-by: Vittorio Giovara <[email protected]>
>> ---
>> Blame j-b.
>> Vittorio
>>
>> libavcodec/vp3.c | 52 +++++++++++++++++++++++++++++++++++++---------------
>> 1 file changed, 37 insertions(+), 15 deletions(-)
>>
>> diff --git a/libavcodec/vp3.c b/libavcodec/vp3.c
>> - vp3_draw_horiz_band(s, s->avctx->height);
>> + vp3_draw_horiz_band(s, s->height);
>>
>> + /* output frame, offset as needed */
>> if ((ret = av_frame_ref(data, s->current_frame.f)) < 0)
>> return ret;
>> + for (i = 0; i < 3; i++) {
>> + AVFrame *dst = data;
>> + int off = (s->offset_x >> (i && s->chroma_y_shift)) +
>> + (s->offset_y >> (i && s->chroma_y_shift)) *
>> dst->linesize[i];
>
> are we sure it won't get over the buffer? (are the offsets validated
> somewhere?).
Actually no, I'll add a check for it.
>> + dst->data[i] += off;
>> + }
>> *got_frame = 1;
>>
>> if (!HAVE_THREADS || !(s->avctx->active_thread_type & FF_THREAD_FRAME))
>> {
>> @@ -2259,16 +2268,29 @@ static int theora_decode_header(AVCodecContext
>> *avctx, GetBitContext *gb)
>> skip_bits(gb, 3); /* reserved */
>> }
>>
>> -// align_get_bits(gb);
>> -
>> - if (visible_width <= s->width && visible_width > s->width - 16 &&
>> - visible_height <= s->height && visible_height > s->height - 16 &&
>> - !offset_x && (offset_y == s->height - visible_height))
>> - ret = ff_set_dimensions(avctx, visible_width, visible_height);
>> - else
>> - ret = ff_set_dimensions(avctx, s->width, s->height);
>
>
> Is that enough?
What do you mean? Right now s->width and s->height contain the full
frame size, so context width/height and coded_w/h gets initialized
correctly.
--
Vittorio
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
