Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1197081 > unrolled thread

Re: [tip:x86/mm] x86/mm/mtrr: Clean up mtrr_type_lookup()

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [tip:x86/mm] x86/mm/mtrr: Clean up mtrr_type_lookup() Peter Zijlstra <peterz@infradead.org> - 2015-07-31 17:10 +0200
    Re: [tip:x86/mm] x86/mm/mtrr: Clean up mtrr_type_lookup() Borislav Petkov <bp@alien8.de> - 2015-07-31 17:30 +0200

#1197081 — Re: [tip:x86/mm] x86/mm/mtrr: Clean up mtrr_type_lookup()

On Fri, Jul 31, 2015 at 04:44:52PM +0200, Borislav Petkov wrote:
> On Fri, Jul 31, 2015 at 03:18:02PM +0200, Peter Zijlstra wrote:
> > Using these functions with preemption enabled is racy against MTRR
> > updates. And if that race is ok, at the very least explain that it is
> > indeed racy and why this is not a problem.
> 
> Right, so Luis has been working on burying direct MTRR access so
> after that work is done, we'll be using only PAT for changing memory
> attributes. Look at arch_phys_wc_add() and all those fbdev users of
> mtrr_add() which get converted to that thing...

Drivers don't do those lookups afaict.

But its things like set_memory_XX(), and afaict that's all buggy against
MTRR modifications.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1197119

On Fri, Jul 31, 2015 at 05:08:06PM +0200, Peter Zijlstra wrote:
> But its things like set_memory_XX(), and afaict that's all buggy against
> MTRR modifications.

I think the idea is to not do any MTRR modifications at some point:

From Documentation/x86/pat.txt:

"... Ideally mtrr_add() usage will be phased out in favor of
arch_phys_wc_add() which will be a no-op on PAT enabled systems. The
region over which a arch_phys_wc_add() is made, should already have been
ioremapped with WC attributes or PAT entries, this can be done by using
ioremap_wc() / set_memory_wc()."

-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web