[PATCH] iWMMXt support was Re: [Ffmpeg-devel] iWMMXt support

[Guillaume Poirier]

Hi,

Reimar D?ffinger wrote:
> Hello,
> On Thu, Mar 15, 2007 at 12:10:56AM +0100, Guillaume POIRIER wrote:
> 
>>Couldn't be done in the same way as we do Altivec support detection on 
>>Linux?
>>
>>The idea is simple: install a signal handler for SIGILL, try to
>>exectute an iWMMXt instruction, and if it triggers a SIGILL, then that
>>means that no support is there, otherwise, iWMMXt is available.
> 
> 
> And that one is still broken. It sent a patch to make it restore at
> least the signal handler again to its original value, but it was
> suggested to remove the whole hack instead.

Maybe it's a hack, but I'd like to see someone coming up with a proper
way to test for Altivec support. I mean, there's gotta be a proper
way, it's just that Google isn't that informative :-(

Most people agree that x86 sucks in almost every way possible, but at
least it has a standard procedure to test for CPU capabilities.


> Many months later, this
> completely broken code remains unchanged.
> Changing a global attribute that can't be done in a thread-safe way from
> a library is really evil anyway.

I agree with you. Here's what man signal has to say:

NOTES
 The effects of this call in a multi-threaded process are unspecified.

:-(

Guillaume