Re: [PATCH 00/19] x86, mpx updates for 4.2 (take 7)

[Dave Hansen <dave@sr71.net>]

On 05/20/2015 03:05 AM, Ingo Molnar wrote:
>> > 
>> > This sees breakage unless either booted with 'noxsaves'
>> > or if it has Fenghua's set from here applied:
>> > 
>> > 	http://lkml.kernel.org/r/1429678319-61356-1-git-send-email-fenghua.yu@intel.com
>> > 
>> > This set is also available against 4.1-rc3 in git:
>> > 
>> >   git://git.kernel.org/pub/scm/linux/kernel/git/daveh/x86-mpx.git mpx-v22
> Yeah, so as a first step, could you please test that the patch below 
> solves the crashes as well, without having to specify 'noxsaves' on 
> the boot line?

Yes, that does seem to make it happy in lieu of Fenghua's patches.

> +	/*
> +	 * Quirk: we don't yet handle the XSAVES* instructions
> +	 * correctly, as we don't correctly convert between
> +	 * standard and compacted format when interfacing
> +	 * with user-space - so disable it for now.
> +	 *
> +	 * The difference is small: with recent CPUs the
> +	 * compacted format is only marginally smaller than
> +	 * the standard FPU state format.
> +	 *
> +	 * ( This is easy to backport while we are fixing
> +	 *   XSAVES* support. )
> +	 */
> +	setup_clear_cpu_cap(X86_FEATURE_XSAVES);
>  }

FWIW, I think it would be prudent to also clear X86_FEATURE_XSAVEC.

All of the issues I am aware of are related to the compact format, not
'xsaves' itself (although 'xsaves' does *use* the compact format of course).

The XSAVEC bit is the one that technically indicates the compact format
support, although I don't think there is any actual use of it in the
kernel at present.