This is the mail archive of the mailing list for the glibc project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] BZ #14059 - HAS_FMA4 check needs to also check for AVX

On 05/11/2012 02:40 PM, Carlos O'Donell wrote:
On Fri, May 11, 2012 at 4:36 PM, Jeff Law<> wrote:
On 05/11/2012 09:22 AM, Carlos O'Donell wrote:

On Fri, May 11, 2012 at 10:31 AM, Jeff Law<> wrote:

On 05/11/2012 06:05 AM, Carlos O'Donell wrote:

Here's a combined patch that compiled fine for me on Linux/x86-64. Jeff, could you test it under Xen, please?

I've finally got functional access to a box which exhibits the problem;
yes, I'll take this patch for a spin and report back.

Would you test this final unified patch?

This patch doesn't work. It causes infinite mutual recursion between init_cpu_features and get_cpu_features.

AFAICT, you can't use the CPUID_XXX macros (which call get_cpu_features)
until init_cpu_features is complete.

Shucks. OK, I'll poke at this on Monday. I horribly confused though. The testsuite didn't blow up for me on a system *with* AVX running RHEL6.1. I would have expected init_cpu_features to be called and the blowup to result in the testsuite blowing up.

I'll rewrite the macros.
It happens ;-) Watch the && HAS_AVX test on the FMA check as well, I think that's going to set up the mutual recursion as well.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]