[FFmpeg-devel] [PATCH 2/2] avcodec/ac3enc: Avoid needlessly copying encoded packets around

[Andreas Rheinhardt]

James Almer:
> On 3/29/2021 2:12 PM, Andreas Rheinhardt wrote:
>> AC-3 and EAC-3 are codecs whose packet sizes are known in advance,
>> so one can use the min_size parameter of ff_alloc_packet2() to
>> allocate exactly this amount. This avoids a memcpy later in
>> av_packet_make_refcounted() in encode_simple_internal().
>>
>> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at outlook.com>
>> ---
>> Is there actually a reason not to use av_new_packet() (or
>> ff_get_encode_buffer()) directly?
> 
> I'm waiting for the bump when avcodec_encode_*() are removed before
> making most encoders DR1, and thus use ff_get_encode_buffer(). And that
> includes this one.
> 
> This patch LGTM, but it may not be worth applying if it's going to be
> changed again very soon.
> 

What are your plans for encoders that don't know the final size in
advance and overallocate?

>>
>>   libavcodec/ac3enc.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c
>> index 4cfd0afe12..fc4d45651d 100644
>> --- a/libavcodec/ac3enc.c
>> +++ b/libavcodec/ac3enc.c
>> @@ -1759,7 +1759,8 @@ int
>> ff_ac3_encode_frame_common_end(AVCodecContext *avctx, AVPacket *avpkt,
>>         ac3_quantize_mantissas(s);
>>   -    if ((ret = ff_alloc_packet2(avctx, avpkt, s->frame_size, 0)) < 0)
>> +    ret = ff_alloc_packet2(avctx, avpkt, s->frame_size, s->frame_size);
>> +    if (ret < 0)
>>           return ret;
>>       ac3_output_frame(s, avpkt->data);
>>  
> 
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> 
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".