[FFmpeg-devel] [PATCH][RFC] QCP demuxer

Kenan Gillet kenan.gillet
Fri May 15 20:21:21 CEST 2009


Hi,
On May 15, 2009, at 8:53 AM, Michael Niedermayer wrote:

> On Thu, May 14, 2009 at 09:57:19PM -0700, Kenan Gillet wrote:
>> hi,
>>
>> attached is the round 2 of the patch.
> [...]
>> +    for (i=0; i<c->nb_rates; i++)
>> +        if (c->rates_mapping[i].mode == byte_mode)
>> +            return c->rates_mapping[i].size;
>> +
>
> id use a LUT here, besides nb_rates can be larger than the array

out of bond read fixed.

While in practice, the byte_mode should be 0-4 for QCELP and EVRC,
nothing says that it must be. Would it be acceptable to assume that it  
does
and have a 5  entries LUT ? A  255-entry LUT would make it more bullet  
proof
if the extra unused memory is not an issue?
Any other solution i missed?

Kenan



More information about the ffmpeg-devel mailing list