On Tue, Feb 07, 2017 at 01:44:54PM -0800, Andrew Morton wrote:
> On Tue, 7 Feb 2017 17:33:47 +0300 "Kirill A. Shutemov"
> <[email protected]> wrote:
>
> > lock_pte_protection() uses pmd_lock() to make sure that we have stable
> > PTE page table before walking pte range.
> >
> > That's not necessary. We only need to make sure that PTE page table is
> > established. It cannot vanish under us as long as we hold mmap_sem at
> > least for read.
> >
> > And we already have helper for that -- pmd_trans_unstable().
>
> http://ozlabs.org/~akpm/mmots/broken-out/mm-mprotect-use-pmd_trans_unstable-instead-of-taking-the-pmd_lock.patch
> already did this?
Right. Except, it doesn't drop unneeded pmd_trans_unstable(pmd) check after
__split_huge_pmd().
Could you fold this part of my patch into Andrea's?
diff --git a/mm/mprotect.c b/mm/mprotect.c
index f9c07f54dd62..e919e4613eab 100644
--- a/mm/mprotect.c
+++ b/mm/mprotect.c
@@ -177,8 +149,6 @@ static inline unsigned long change_pmd_range(struct
vm_area_struct *vma,
if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd)) {
if (next - addr != HPAGE_PMD_SIZE) {
__split_huge_pmd(vma, pmd, addr, false, NULL);
- if (pmd_trans_unstable(pmd))
- continue;
} else {
int nr_ptes = change_huge_pmd(vma, pmd, addr,
newprot, prot_numa);
--
Kirill A. Shutemov
