Re: [PATCH] mm: numa: Recheck for transhuge pages under lock during protection changes

[Andrew Morton]

On Wed, 19 Mar 2014 14:38:32 +0000 Mel Gorman <mgorman@xxxxxxx> wrote:

> On Fri, Mar 14, 2014 at 11:15:37PM -0400, Sasha Levin wrote:
> > On 03/12/2014 06:36 AM, Mel Gorman wrote:
> > >Andrew, this should go with the patches
> > >mmnuma-reorganize-change_pmd_range.patch
> > >mmnuma-reorganize-change_pmd_range-fix.patch
> > >move-mmu-notifier-call-from-change_protection-to-change_pmd_range.patch
> > >in mmotm please.
> > >
> > >Thanks.
> > >
> > >---8<---
> > >From: Mel Gorman<mgorman@xxxxxxx>
> > >Subject: [PATCH] mm: numa: Recheck for transhuge pages under lock during protection changes
> > >
> > >Sasha Levin reported the following bug using trinity
> > 
> > I'm seeing a different issue with this patch. A NULL ptr deref occurs in the
> > pte_offset_map_lock() macro right before the new recheck code:
> > 
> 
> This on top?
> 
> I tried testing it but got all sorts of carnage that trinity throw up
> in the mix and ordinary testing does not trigger the race. I've no idea
> which of the current mess of trinity-exposed bugs you've encountered and
> got fixed already.
> 

Where are we at with this one?


With this patch I'm now sitting on

mmnuma-reorganize-change_pmd_range.patch
mmnuma-reorganize-change_pmd_range-fix.patch
mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes.patch
mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes-fix.patch
move-mmu-notifier-call-from-change_protection-to-change_pmd_range.patch

Where the first four are prep work for the fifth.

move-mmu-notifier-call-from-change_protection-to-change_pmd_range.patch
is a performance optimisation for KVM guests.  Should I drop all five,
try again next time?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/