[FFmpeg-devel] [PATCH 2/2] mpegaudio: call ff_mpegaudiodec_init_mmx() only from float decoder

[Måns Rullgård]

Michael Niedermayer <michaelni at gmx.at> writes:

> On Thu, Jul 01, 2010 at 06:14:16PM +0100, M?ns Rullg?rd wrote:
>> Vitor Sessak <vitor1001 at gmail.com> writes:
>> 
>> > On 07/01/2010 06:26 PM, Mans Rullgard wrote:
>> >> The mmx code is floating-point only, and this function does not know
>> >> from which decoder it is called.  Without this change, the integer
>> >> decoder only "works" because the size of the context struct is smaller
>> >> in this case, and the mmx init function writes the function pointer
>> >> outside the allocated context.
>> >> ---
>> >>   libavcodec/mpegaudiodec.c |    2 +-
>> >>   1 files changed, 1 insertions(+), 1 deletions(-)
>> >>
>> >> diff --git a/libavcodec/mpegaudiodec.c b/libavcodec/mpegaudiodec.c
>> >> index 3edfc65..af5b3b8 100644
>> >> --- a/libavcodec/mpegaudiodec.c
>> >> +++ b/libavcodec/mpegaudiodec.c
>> >> @@ -320,7 +320,7 @@ static av_cold int decode_init(AVCodecContext * avctx)
>> >>
>> >>       s->avctx = avctx;
>> >>       s->apply_window_mp3 = apply_window_mp3_c;
>> >> -#if HAVE_MMX
>> >> +#if HAVE_MMX&&  CONFIG_FLOAT
>> >
>> > I was thinking about committing something similar. Your patch should
>> > fix issue2054. I would also move the line
>> >
>> >       s->apply_window_mp3 = apply_window_mp3_c;
>> >
>> > inside a #if CONFIG_FLOAT.
>> 
>> Don't we want to allow asm optimisation of the fixed-point one as well?
>
> on systems with mmx optimized float decoding will liekly be faster
> anywaym fix this in whichever way you prefer

I was thinking of FPU-less ARM systems and the like.

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