Re: [PATCH 1/7] hugetlb, rmap: add reverse mapping for hugepage

[Andi Kleen]

Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx> writes:

Adding a few more recent hugetlb hackers in cc. Folks, please consider
reviewing the hugetlb.c parts of the original patch kit in linux-mm.

> While hugepage is not currently swappable, rmapping can be useful
> for memory error handler.
> Using rmap, memory error handler can collect processes affected
> by hugepage errors and unmap them to contain error's effect.

Thanks.

I reviewed all the patches and they look good to me. I can merge
them through the hwpoison git tree.

But before merging it there I would like to have some review
and acks from mm hackers on the mm/hugetlb.c parts, which
do (relatively minor) changes outside memory-failure.c

I think you also had a patch for mce-test, can you send me that
one too?

BTW I wonder: did you verify that the 1GB page support works?
I would expect it does, but it would be good to double check.
One would need a Westmere server or AMD Family10h+ system to test that.

> Current status of hugepage rmap differs depending on mapping mode:
> - for shared hugepage:
>   we can collect processes using a hugepage through pagecache,
>   but can not unmap the hugepage because of the lack of mapcount.
> - for privately mapped hugepage:
>   we can neither collect processes nor unmap the hugepage.

I hope these points can be eventually addressed too, but this
is a good first step and closes an important hole in hwpoison
coverage.

-Andi


-- 
ak@xxxxxxxxxxxxxxx -- Speaking for myself only.
--
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/