On 28/10/11 08:13, Ronald S. Bultje wrote:
> Hi,
> 
> On Fri, Oct 28, 2011 at 8:11 AM, Luca Barbato <[email protected]> wrote:
>> On 10/28/11 7:19 AM, Ronald S. Bultje wrote:
>>>
>>> Index: a/libavcodec/vp8.c
>>> ===================================================================
>>> --- a/libavcodec/vp8.c  (revision 106016)
>>> +++ b/libavcodec/vp8.c  (working copy)
>>> @@ -50,7 +50,8 @@
>>>      int ret;
>>>      if ((ret = ff_thread_get_buffer(s->avctx, f))<  0)
>>>          return ret;
>>> -    if (!s->maps_are_invalid&&  s->num_maps_to_be_freed) {
>>> +    if (s->num_maps_to_be_freed) {
>>> +        assert(!s->maps_are_invalid);
>>>          f->ref_index[0] =
>>> s->segmentation_maps[--s->num_maps_to_be_freed];
>>>      } else if (!(f->ref_index[0] = av_mallocz(s->mb_width *
>>> s->mb_height))) {
>>>          ff_thread_release_buffer(s->avctx, f);
>>> @@ -59,39 +60,50 @@
>>>      return 0;
>>>  }
>>
>> The rest looks more or less ok, but that puzzles me.
> 
> It should never trigger. If it does trigger, something is very bad.
> 
> There's another check (the memleak check further down) that I may
> change into an assert later on also, because again, it should never
> trigger and I can probably proove that.

Ok then, please commit anytime you like =)

lu

-- 

Luca Barbato
Gentoo/linux
http://dev.gentoo.org/~lu_zero

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to