Re: [PATCH 2/2] hugetlbfs: add swap entry check infollow_hugetlb_page()

[Michal Hocko]

On Fri 29-03-13 13:23:38, Naoya Horiguchi wrote:
> Hi,
> 
> On Fri, Mar 29, 2013 at 02:57:30PM +0100, Michal Hocko wrote:
> > On Thu 28-03-13 11:42:38, Naoya Horiguchi wrote:
> > [...]
> > > @@ -2968,7 +2968,8 @@ long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
> > >  		 * first, for the page indexing below to work.
> > >  		 */
> > >  		pte = huge_pte_offset(mm, vaddr & huge_page_mask(h));
> > > -		absent = !pte || huge_pte_none(huge_ptep_get(pte));
> > > +		absent = !pte || huge_pte_none(huge_ptep_get(pte)) ||
> > > +			is_swap_pte(huge_ptep_get(pte));
> > 
> > is_swap_pte doesn't seem right. Shouldn't you use is_hugetlb_entry_hwpoisoned
> > instead?
> 
> I tested only hwpoisoned hugepage, but the same can happen for hugepages
> under migration. So I intended to filter out all types of swap entries.
> The local variable 'absent' seems to mean whether data on the address
> is immediately available, so swap type entry isn't included in it.

OK, I didn't consider huge pages under migration and I was merely worried
that is_hugetlb_entry_hwpoisoned sounds more appropriate than
is_swap_pte.

Could you add a comment which would clarify that is_swap_pte covers both
migration and hwpoison pages, please? Something like:

		/*
		 * is_swap_pte test covers both is_hugetlb_entry_hwpoisoned
		 * and hugepages under migration in which case
		 * hugetlb_fault waits for the migration and bails out
		 * properly for HWPosined pages.
		 */
		 absent = !pte || huge_pte_none(huge_ptep_get(pte)) ||
		 	 is_swap_pte(huge_ptep_get(pte));

Other than that feel free to add
Reviewed-by: Michal Hocko <mhocko@xxxxxxx>

Thanks.
-- 
Michal Hocko
SUSE Labs
--
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/