| Message ID | 20221125213714.4115729-3-jannh@google.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [v3,1/3] mm/khugepaged: Take the right locks for page table retraction | expand |
On 25.11.22 22:37, Jann Horn wrote: > Any codepath that zaps page table entries must invoke MMU notifiers to > ensure that secondary MMUs (like KVM) don't keep accessing pages which > aren't mapped anymore. Secondary MMUs don't hold their own references to > pages that are mirrored over, so failing to notify them can lead to page > use-after-free. > > I'm marking this as addressing an issue introduced in commit f3f0e1d2150b > ("khugepaged: add support of collapse for tmpfs/shmem pages"), but most of > the security impact of this only came in commit 27e1f8273113 ("khugepaged: > enable collapse pmd for pte-mapped THP"), which actually omitted flushes > for the removal of present PTEs, not just for the removal of empty page > tables. > > Cc: stable@kernel.org > Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages") I'm curious, do you have a working reproducer for this? Change looks sane on quick glimpse.
On Mon, Nov 28, 2022 at 6:37 PM David Hildenbrand <david@redhat.com> wrote: > > On 25.11.22 22:37, Jann Horn wrote: > > Any codepath that zaps page table entries must invoke MMU notifiers to > > ensure that secondary MMUs (like KVM) don't keep accessing pages which > > aren't mapped anymore. Secondary MMUs don't hold their own references to > > pages that are mirrored over, so failing to notify them can lead to page > > use-after-free. > > > > I'm marking this as addressing an issue introduced in commit f3f0e1d2150b > > ("khugepaged: add support of collapse for tmpfs/shmem pages"), but most of > > the security impact of this only came in commit 27e1f8273113 ("khugepaged: > > enable collapse pmd for pte-mapped THP"), which actually omitted flushes > > for the removal of present PTEs, not just for the removal of empty page > > tables. > > > > Cc: stable@kernel.org > > Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages") > > I'm curious, do you have a working reproducer for this? You're on the CC list of my bug report to security@kernel.org with title "khugepaged races with rmap-based zap, races with GUP-fast, and fails to call MMU notifiers". That has an attached reproducer thp_ro_no_notify_kvm.c that is able to read PAGE_POISON out of freed file THP pages through KVM.
On 28.11.22 18:57, Jann Horn wrote: > On Mon, Nov 28, 2022 at 6:37 PM David Hildenbrand <david@redhat.com> wrote: >> >> On 25.11.22 22:37, Jann Horn wrote: >>> Any codepath that zaps page table entries must invoke MMU notifiers to >>> ensure that secondary MMUs (like KVM) don't keep accessing pages which >>> aren't mapped anymore. Secondary MMUs don't hold their own references to >>> pages that are mirrored over, so failing to notify them can lead to page >>> use-after-free. >>> >>> I'm marking this as addressing an issue introduced in commit f3f0e1d2150b >>> ("khugepaged: add support of collapse for tmpfs/shmem pages"), but most of >>> the security impact of this only came in commit 27e1f8273113 ("khugepaged: >>> enable collapse pmd for pte-mapped THP"), which actually omitted flushes >>> for the removal of present PTEs, not just for the removal of empty page >>> tables. >>> >>> Cc: stable@kernel.org >>> Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages") >> >> I'm curious, do you have a working reproducer for this? > > You're on the CC list of my bug report to security@kernel.org > with title "khugepaged races with rmap-based zap, races with GUP-fast, > and fails to call MMU notifiers". That has an attached reproducer > thp_ro_no_notify_kvm.c that is able to read PAGE_POISON out of freed > file THP pages through KVM. > Ah, the mail from early October, thanks (drowning in mail). You're amazingly skilled at writing reproducers.
diff --git a/mm/khugepaged.c b/mm/khugepaged.c index c3d3ce596bff7..49eb4b4981d88 100644 --- a/mm/khugepaged.c +++ b/mm/khugepaged.c @@ -1404,6 +1404,7 @@ static void collapse_and_free_pmd(struct mm_struct *mm, struct vm_area_struct *v unsigned long addr, pmd_t *pmdp) { pmd_t pmd; + struct mmu_notifier_range range; mmap_assert_write_locked(mm); if (vma->vm_file) @@ -1415,8 +1416,12 @@ static void collapse_and_free_pmd(struct mm_struct *mm, struct vm_area_struct *v if (vma->anon_vma) lockdep_assert_held_write(&vma->anon_vma->root->rwsem); + mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, NULL, mm, addr, + addr + HPAGE_PMD_SIZE); + mmu_notifier_invalidate_range_start(&range); pmd = pmdp_collapse_flush(vma, addr, pmdp); tlb_remove_table_sync_one(); + mmu_notifier_invalidate_range_end(&range); mm_dec_nr_ptes(mm); page_table_check_pte_clear_range(mm, addr, pmd); pte_free(mm, pmd_pgtable(pmd));
Any codepath that zaps page table entries must invoke MMU notifiers to ensure that secondary MMUs (like KVM) don't keep accessing pages which aren't mapped anymore. Secondary MMUs don't hold their own references to pages that are mirrored over, so failing to notify them can lead to page use-after-free. I'm marking this as addressing an issue introduced in commit f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages"), but most of the security impact of this only came in commit 27e1f8273113 ("khugepaged: enable collapse pmd for pte-mapped THP"), which actually omitted flushes for the removal of present PTEs, not just for the removal of empty page tables. Cc: stable@kernel.org Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages") Signed-off-by: Jann Horn <jannh@google.com> --- mm/khugepaged.c | 5 +++++ 1 file changed, 5 insertions(+)