mm/userfaultfd: fix uffd-wp handling for THP migration entries
authorDavid Hildenbrand <david@redhat.com>
Wed, 5 Apr 2023 16:02:35 +0000 (18:02 +0200)
committerAndrew Morton <akpm@linux-foundation.org>
Sun, 16 Apr 2023 17:41:24 +0000 (10:41 -0700)
Looks like what we fixed for hugetlb in commit 44f86392bdd1 ("mm/hugetlb:
fix uffd-wp handling for migration entries in
hugetlb_change_protection()") similarly applies to THP.

Setting/clearing uffd-wp on THP migration entries is not implemented
properly.  Further, while removing migration PMDs considers the uffd-wp
bit, inserting migration PMDs does not consider the uffd-wp bit.

We have to set/clear independently of the migration entry type in
change_huge_pmd() and properly copy the uffd-wp bit in
set_pmd_migration_entry().

Verified using a simple reproducer that triggers migration of a THP, that
the set_pmd_migration_entry() no longer loses the uffd-wp bit.

Link: https://lkml.kernel.org/r/20230405160236.587705-2-david@redhat.com
Fixes: f45ec5ff16a7 ("userfaultfd: wp: support swap and page migration")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org>
Cc: Muhammad Usama Anjum <usama.anjum@collabora.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
mm/huge_memory.c

index 032fb0e..e3706a2 100644 (file)
@@ -1838,10 +1838,10 @@ int change_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
        if (is_swap_pmd(*pmd)) {
                swp_entry_t entry = pmd_to_swp_entry(*pmd);
                struct page *page = pfn_swap_entry_to_page(entry);
+               pmd_t newpmd;
 
                VM_BUG_ON(!is_pmd_migration_entry(*pmd));
                if (is_writable_migration_entry(entry)) {
-                       pmd_t newpmd;
                        /*
                         * A protection check is difficult so
                         * just be safe and disable write
@@ -1855,8 +1855,16 @@ int change_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
                                newpmd = pmd_swp_mksoft_dirty(newpmd);
                        if (pmd_swp_uffd_wp(*pmd))
                                newpmd = pmd_swp_mkuffd_wp(newpmd);
-                       set_pmd_at(mm, addr, pmd, newpmd);
+               } else {
+                       newpmd = *pmd;
                }
+
+               if (uffd_wp)
+                       newpmd = pmd_swp_mkuffd_wp(newpmd);
+               else if (uffd_wp_resolve)
+                       newpmd = pmd_swp_clear_uffd_wp(newpmd);
+               if (!pmd_same(*pmd, newpmd))
+                       set_pmd_at(mm, addr, pmd, newpmd);
                goto unlock;
        }
 #endif
@@ -3251,6 +3259,8 @@ int set_pmd_migration_entry(struct page_vma_mapped_walk *pvmw,
        pmdswp = swp_entry_to_pmd(entry);
        if (pmd_soft_dirty(pmdval))
                pmdswp = pmd_swp_mksoft_dirty(pmdswp);
+       if (pmd_uffd_wp(pmdval))
+               pmdswp = pmd_swp_mkuffd_wp(pmdswp);
        set_pmd_at(mm, address, pvmw->pmd, pmdswp);
        page_remove_rmap(page, vma, true);
        put_page(page);