[Ffmpeg-devel] h264 decoder on ARM cpu

Gábor Kovács picard
Wed May 11 10:49:05 CEST 2005


I checked and it's always odd:

h->b8_stride= s->mb_width*2 + 1;

bye, Picard

----- Original Message ----- 
From: "Stefan Gehrer" <stefan.gehrer at freescale.com>
To: "FFMpeg development discussions and patches" <ffmpeg-devel at mplayerhq.hu>
Sent: Wednesday, May 11, 2005 10:24 AM
Subject: Re: [Ffmpeg-devel] h264 decoder on ARM cpu


> G?bor Kov?cs wrote:
>
>> I think I found a memory access alignment problem in h264.c line 1379 (when b8_stride is odd):
>>
>> *(uint16_t*)&s->current_picture.ref_index[list][b8_xy + y*h->b8_stride]= (LIST_NOT_USED&0xFF)*0x0101;
>>
>> (I split it into two uint8_t access)
>>
> I am not that much into that code, but guessing from the name, b8_stride is the width
> of the frame in 8x8 blocks. As the frames are always coded in Macroblocks (16x16),
> b8_stride can never become odd.
>
> Regards
> Stefan Gehrer
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/ffmpeg-devel
> 





More information about the ffmpeg-devel mailing list