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

[Andrea Arcangeli <aarcange@redhat.com>]

On Thu, Mar 17, 2011 at 05:27:10PM +0100, Andi Kleen wrote:
> > What I can tell is with the default khugepaged scan rate, the
> > collapse_huge_page will have an impact much smaller than KSM. It could
> > have more impact than KSM if you increase khugepaged load to 100% with
> > sysfs (because of the more memory that is covered by khugepaged
> > compared to only the shared portion of KSM). Then the window gets much
> > bigger, but still minor, if you can't trigger it with the testsuite
> > it's even less likely to ever happen in practice.
> 
> You mean randomly injecting errors?
> That tends to be hard and unreliable -- usually we try to have a 
> specific tester that is not random.

I meant the testsuite using MCE injection, called mce-test. I've run
it a couple of times for some hugetlbfs collision with THP (solved
some time ago).

> The measurement is simple: run the workloads and do some dumps
> with pagetypes and check if the memory with lots of pages 
> has a state that can be handled by memory_failure()
> 
> AFAIK  this hasn't been done so far with THP.

I'm unsure if there's already coverage for it in mce-test yet, the
biggest test I run was hugetlbfs related (MAP_ANONYMOUS|MAP_HUGETLB or
filebacked or still shm). Surely it'd be good idea to add THP
coverage.