Re: [PATCH v8 1/3] mm/hwpoison: fix race between hugetlb free/demotion and memory_failure_hugetlb()

["HORIGUCHI NAOYA(堀口 直也)" <naoya.horiguchi@nec.com>]

On Fri, Apr 15, 2022 at 09:55:14AM +0800, Miaohe Lin wrote:
> On 2022/4/15 1:56, Mike Kravetz wrote:
> > On 4/8/22 06:53, Naoya Horiguchi wrote:
> >> From: Naoya Horiguchi <naoya.horiguchi@nec.com>
> >>
> >> There is a race condition between memory_failure_hugetlb() and hugetlb
> >> free/demotion, which causes setting PageHWPoison flag on the wrong page.
> >> The one simple result is that wrong processes can be killed, but another
> >> (more serious) one is that the actual error is left unhandled, so no one
> >> prevents later access to it, and that might lead to more serious results
> >> like consuming corrupted data.
> >>
> >> Think about the below race window:
> >>
> >>   CPU 1                                   CPU 2
> >>   memory_failure_hugetlb
> >>   struct page *head = compound_head(p);
> >>                                           hugetlb page might be freed to
> >>                                           buddy, or even changed to another
> >>                                           compound page.
> >>
> >>   get_hwpoison_page -- page is not what we want now...
> >>
> >> The current code first does prechecks roughly and then reconfirms
> >> after taking refcount, but it's found that it makes code overly
> >> complicated, so move the prechecks in a single hugetlb_lock range.
> >>
> >> A newly introduced function, try_memory_failure_hugetlb(), always
> >> takes hugetlb_lock (even for non-hugetlb pages).  That can be
> >> improved, but memory_failure() is rare in principle, so should
> >> not be a big problem.
> ...
> > 
> > The above code works as designed, but may be a bit confusing.  If HPageFreed()
> > we KNOW ref count is zero, so no need to even call get_page_unless_zero() as
> > it will always return false in this case.  It might be more clear if written
> > as separate else if statements such as:
> > 
> > 	} else if (HPageFreed(head)) {
> > 		ret = 0;
> > 	} else if (HPageMigratable(head)) {
> > 		ret = get_page_unless_zero(head);
> > 		if (ret)
> > 			count_increased = true;
> > 
> 
> This code here is consistent with the logic in get_hwpoison_huge_page. If change is required,
> they might need to be changed together.
> 
> BTW: They look a bit confusing for me at first but I get used to it later. ;)

Thank you for comments, Mike, Miaohe.
Patch 1/3 is to stable, so I'll submit a separate cleanup patch for these changes.

> 
> Thanks!
> 
> > Not insisting this be changed.  Just easier to understand IMO.
> > 
> > Again, thanks for your work on this!
> > 
> > Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>

Thank you!

- Naoya Horiguchi