[FFmpeg-devel] [PATCH] configure: add command flag to disable yasm use

Dave Yeo daveryeo
Thu Feb 5 20:43:51 CET 2009

On 02/04/09 09:45 pm, Jason Garrett-Glaser wrote:
> On Wed, Feb 4, 2009 at 9:36 PM, Dave Yeo<daveryeo at telus.net>  wrote:
>> On 02/04/09 07:58 pm, Art Clarke wrote:
>>> But in general given that libx264 already requires yasm (build fails
>>> if it can't find yasm or you don't pass --disable-asm), more and more
>>> people are using FFMPEG for h.264 encoding, and slowly but surely Yasm
>>> code is increasing in FFMPEG, I think some change re: YASM
>>> configuration is a good idea.
>>> I suggest giving a rude WARNING on configure if yasm can't be found on
>>> the current system to give folks a heads up, and then in three months
>>> or so change configure to require yasm.
>>> If folks agree, I'm happy to submit a patch.
>> This would be very irritating on platforms that yasm doesn't support.
>> (and nasm also doesn't work)
> Which x86/x86_64 platforms are those?  I don't know of any off the top
> of my head.

OS/2, yasm doesn't support either of the native object formats, aout and 
nasm does support them. aout needs a lot of work to compile FFmpeg but 
ultimately fails the same way as obj,

nasm.exe -f obj -DARCH_X86_32 -DPREFIX -I 
/usr/src/ffmpeg/libavcodec/x86/ -o libavcodec/x86/fft_mmx.o 
/usr/src/ffmpeg/libavcodec/x86/fft_mmx.asm:406: error: 
beroset-p-603-invalid effective address
/usr/src/ffmpeg/libavcodec/x86/fft_mmx.asm:406: error: 
beroset-p-603-invalid effective address
error repeated numerous times till line #416

yasm -f elf does work but of course doesn't link. Nasm with a bit of 
work compiles x264 fine.

> But if they exist, why not just have the rude warning/error not appear
> on such platforms?

Yes, that will have to be the solution.

> Dark Shikari


More information about the ffmpeg-devel mailing list