mm/ksm: fix race with VMA iteration and mm_struct teardown

exit_mmap() will tear down the VMAs and maple tree with the mmap_lock held
in write mode.  Ensure that the maple tree is still valid by checking
ksm_test_exit() after taking the mmap_lock in read mode, but before the
for_each_vma() iterator dereferences a destroyed maple tree.

Since the maple tree is destroyed, the flags telling lockdep to check an
external lock has been cleared.  Skip the for_each_vma() iterator to avoid
dereferencing a maple tree without the external lock flag, which would
create a lockdep warning.

Link: https://lkml.kernel.org/r/20230308220310.3119196-1-Liam.Howlett@oracle.com
Fixes: a5f18ba ("mm/ksm: use vma iterators instead of vma linked list")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Reported-by: Pengfei Xu <pengfei.xu@intel.com>
  Link: https://lore.kernel.org/lkml/ZAdUUhSbaa6fHS36@xpf.sh.intel.com/
Reported-by: syzbot+2ee18845e89ae76342c5@syzkaller.appspotmail.com
  Link: https://syzkaller.appspot.com/bug?id=64a3e95957cd3deab99df7cd7b5a9475af92c93e
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: <heng.su@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Author: howlett

mm/ksm.c

@@ -988,9 +988,15 @@ static int unmerge_and_remove_all_rmap_items(void)
 
 		mm = mm_slot->slot.mm;
 		mmap_read_lock(mm);
+
+		/*
+		 * Exit right away if mm is exiting to avoid lockdep issue in
+		 * the maple tree
+		 */
+		if (ksm_test_exit(mm))
+			goto mm_exiting;
+
 		for_each_vma(vmi, vma) {
-			if (ksm_test_exit(mm))
-				break;
 			if (!(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma)
 				continue;
 			err = unmerge_ksm_pages(vma,
@@ -999,6 +1005,7 @@ static int unmerge_and_remove_all_rmap_items(void)
 				goto error;
 		}
 
+mm_exiting:
 		remove_trailing_rmap_items(&mm_slot->rmap_list);
 		mmap_read_unlock(mm);