[FFmpeg-devel] [PATCH] Fix non-rounding up to next 16-bit aligned bug in IFF decoder

[Måns Rullgård]

"Ronald S. Bultje" <rsbultje at gmail.com> writes:

> Hi,
>
> On Apr 29, 2010, at 8:15 AM, Sebastian Vater
> <cdgs.basty at googlemail.com> wrote:
>
>> Bad news here...
>>
>> I just got a report from the guy who did sent the IFF files which were
>> decoded wrong and he said that it doesn't work on big-endian.
>>
>> I assume that the tables have to be swapped around. Hence, what's the
>> ffmpeg prefered way of checking if it's big or little endian (I just
>> will then do sth. like):
>> #ifdef BIG_ENDIAN
>>  // declare table for BE here
>> #else
>>  // declare table for LE here
>
> No, as said on IRC, try replacing the AV_w/rn32() with w/r[bl]32().

That would make it slower.  Using different tables is better.

-- 
M?ns Rullg?rd
mans at mansr.com