Skip to contentSkip to conversation

Bug in hugetlb_cow()?

PATCH Doug Doan 1 participant 1 message May 26, 2010

Subsystems (2)
Memory management HugeTLB
lore
Share & tools
lore

Doug Doan

author
#0 lore raw

Hi,

I'm working on a driver that relies on mmu notifications, and found that it is lacking for huge pages.

From handle_mm_fault(),

If we take the normal page path, this is what happens:

handle_pte_fault -> do_wp_page -> ptep_clear_flush_notify -> mmu_notifier_invalidate_page

If we're dealing with huge pages, this is what happens:

hugetlb_fault -> hugetlb_cow -> (no mmu notifiers)

Below is my patch, can anyone comment?

mm/hugetlb.c +6 -0
index 4c9e6bb..96d9937 100644
@@ -2345,11 +2345,17 @@ retry_avoidcopy:
        ptep = huge_pte_offset(mm, address & huge_page_mask(h));
        if (likely(pte_same(huge_ptep_get(ptep), pte))) {
                /* Break COW */
+               mmu_notifier_invalidate_range_start(mm,
+                       address & huge_page_mask(h),
+                       (address & huge_page_mask(h)) + huge_page_size(h));
                huge_ptep_clear_flush(vma, address, ptep);
                set_huge_pte_at(mm, address, ptep,
                                make_huge_pte(vma, new_page, 1));
                /* Make the old page be freed below */
                new_page = old_page;
+               mmu_notifier_invalidate_range_end(mm,
+                       address & huge_page_mask(h),
+                       (address & huge_page_mask(h)) + huge_page_size(h));
        }
        page_cache_release(new_page);
        page_cache_release(old_page);

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>