[FFmpeg-devel] [PATCH] libavformat: Account for negative position differences in ff_configure_buffers_for_index

[Martin Storsjö]

On Fri, 24 Mar 2023, James Almer wrote:

> On 3/24/2023 6:09 PM, Martin Storsjö wrote:
>> When scanning through the index, account for the fact that the
>> compared samples may be located in an unexpected order in the file;
>> this function is mainly interested in the absolute difference between
>> file locations.
>> 
>> Signed-off-by: Martin Storsjö <martin at martin.st>
>> ---
>>   libavformat/seek.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>> 
>> diff --git a/libavformat/seek.c b/libavformat/seek.c
>> index 818549dfef..62a26d6b5f 100644
>> --- a/libavformat/seek.c
>> +++ b/libavformat/seek.c
>> @@ -210,7 +210,7 @@ void ff_configure_buffers_for_index(AVFormatContext *s, 
> int64_t time_tolerance)
>>                       int64_t e2_pts = av_rescale_q(e2->timestamp, 
> st2->time_base, AV_TIME_BASE_Q);
>>                       if (e2_pts < e1_pts || e2_pts - (uint64_t)e1_pts < 
> time_tolerance)
>>                           continue;
>> -                    pos_delta = FFMAX(pos_delta, e1->pos - e2->pos);
>> +                    pos_delta = FFMAX(pos_delta, llabs(e1->pos - 
> e2->pos));
>
> There's FFABS(), which gives a lot more hits with git grep than llabs.

Fair enough, I can change it to that too.

Since FFMAX is a macro, it ends up expanding the arguments twice, so with 
a call to llabs, we'd have two function calls. I guess it'd be better with 
an intermediate variable,

     int64_t delta = FFABS(e1->pos - e2->pos);
     pos_delta = FFMAX(pos_delta, delta);

WDYT?

// Martin