[FFmpeg-devel] [PATCH v4] lavf/matroskadec: Normalize noncompliant A_QUICKTIME/V_QUICKTIME private data

[Mats Peterson]

On 01/10/2016 02:20 AM, Michael Niedermayer wrote:

>> +        if (!(track->codec_priv.data = av_realloc(track->codec_priv.data,
>> +                track->codec_priv.size + 4)))
>> +            return AVERROR(ENOMEM);
>
> in case of this failing the memory is not freed and leaks
>
> see av_reallocp() if you want it to be freed, alternatively the code
> could continue without the move/update
>

OK, but I thought an AVERROR(ENOMEM) eventually causes FFmpeg to exit?

And another thing: I notice the following regarding both av_realloc() 
and av_reallocp():

"Pointers originating from the av_malloc() family of functions must not 
be passed to av_realloc(). The former can be implemented using 
memalign() (or other functions), and there is no guarantee that pointers 
from such functions can be passed to realloc() at all. The situation is 
undefined according to POSIX and may crash with some libc implementations."

I suppose that ebml_read_binary() in matroskadec.c is the function used 
to read the private data, and it allocates memory using 
av_fast_padded_malloc() for this. Is it safe to reallocate a pointer 
from this function then? So far I've had no problems, but of course I 
want to be absolutely sure.

Mats