Re: Memory corruption during hibernation since 2.6.31

From: Hugh Dickins
Date: Fri Jul 30 2010 - 00:03:20 EST

On Thu, Jul 29, 2010 at 4:40 PM, Rafael J. Wysocki <rjw@xxxxxxx> wrote:
> On Thursday, July 29, 2010, Andrea Arcangeli wrote:
> I'm afraid that would be difficult to achieve in general. ÂBesides, there's
> no reason why kernel threads that need not be frozen should care about the
> freezing thing at all. ÂIt's much simpler to require the ones that need to be
> frozen to cooperate.

But it's looking as if any thread which might have to allocate some memory
(and ksmd is on balance a freer of memory, but nonetheless has to do
slab allocations to get there) would need to be frozen.

Except that does not go far enough, because freezing the hibernating
thread (which itself makes memory allocations) won't work out well!

>> But for now I'm afraid the below is needed (only ksm.c part applies to
>> upstream).
> Looks good to me.
> Can you please prepare a patch against mainline for Ondrej to try?

Andrea wasn't proposing his patch for Ondrej's corruption, and I don't
suppose Ondrej even has ksmd running. Any confusion there is my
fault, for linking Andrea's concerns about suspend and hibernation
with this particular thread on Ondrej's bug.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at