On Wed, May 12, 2021 at 02:19:32PM +0200, Michal Hocko wrote:
> On Wed 12-05-21 00:10:15, Naoya Horiguchi wrote:
> > From: Naoya Horiguchi <naoya.horiguchi@xxxxxxx>
> >
> > When hugetlb page fault (under overcommiting situation) and
> > memory_failure() race, VM_BUG_ON_PAGE() is triggered by the following race:
> >
> > CPU0: CPU1:
> >
> > gather_surplus_pages()
> > page = alloc_surplus_huge_page()
> > memory_failure_hugetlb()
> > get_hwpoison_page(page)
> > __get_hwpoison_page(page)
> > get_page_unless_zero(page)
> > zero = put_page_testzero(page)
> > VM_BUG_ON_PAGE(!zero, page)
> > enqueue_huge_page(h, page)
> > put_page(page)
> >
> > __get_hwpoison_page() only checks page refcount before taking additional
> > one for memory error handling, which is wrong because there's time
> > windows where compound pages have non-zero refcount during initialization.
> >
> > So makes __get_hwpoison_page() check page status a bit more for a few
> > types of compound pages. PageSlab() check is added because otherwise
> > "non anonymous thp" path is wrongly chosen.
>
> This should really describe the fix in more details. E.g.
> [...]
> > @@ -1095,30 +1095,43 @@ static int __get_hwpoison_page(struct page *page)
> > {
> > struct page *head = compound_head(page);
> >
> > - if (!PageHuge(head) && PageTransHuge(head)) {
> > - /*
> > - * Non anonymous thp exists only in allocation/free time. We
> > - * can't handle such a case correctly, so let's give it up.
> > - * This should be better than triggering BUG_ON when kernel
> > - * tries to touch the "partially handled" page.
> > - */
> > - if (!PageAnon(head)) {
> > - pr_err("Memory failure: %#lx: non anonymous thp\n",
> > - page_to_pfn(page));
> > - return 0;
> > + if (PageCompound(page)) {
>
> So you do rely on PageCompound to be true. Which is prone to races as
> well. All you need is to check before prep_compound_page and run into
> get_page_unless_zero (down in this function) before hugetlb reaches
> put_page_testzero. Or do I miss something that would prevent from that?
No, you're right, the race can still happen (I only considered the case when
a compound page is already allocated, that was insufficient...). Any check
outside locking seems harmful, so does the checking like below make more
sense?
static int __get_hwpoison_page(struct page *page)
{
struct page *head = compound_head(page);
int ret = 0;
#ifdef CONFIG_HUGETLB_PAGE // this #ifdef is needed for hugetlb_lock
spin_lock(&hugetlb_lock);
if (PageHuge(head) && (HPageFreed(head) || HPageMigratable(head)))
ret = get_page_unless_zero(head);
spin_unlock(&hugetlb_lock);
if (ret > 0)
return ret;
#endif
// other types of compound page.
...
return get_page_unless_zero(page);
}
- Naoya
