Re: [PATCH 3/4] Check whether pages are poisoned before copying

[Andi Kleen]

> On an extended note, I don't understand why hwpoison code should not
> handle Ksm pages the same way as other user-space pages. While it is

Noone has done the work for it so far.

However we did some evaluation some time ago with some simple 
KVM setups and it turned out KSM was only very little physical memory 
overall (< few percent) Normally it's better to focus on areas with more 
payoff (= more memory), unless you can find a workload where KSM is actually 
significant.

> It appears that the restriction for PageKsm() can be removed from
> memory-failure.c code without making the races any better or worse. Any
> opinions on that?

I don't know. The KSM code paths would need to be audited first.
Hugh added the checks originally and may have an opinion.

-Andi
--
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/