diff mbox series

[v4,19/30] btrfs: don't clear page extent mapped if we're not invalidating the full page

Message ID 20210531085106.259490-20-wqu@suse.com (mailing list archive)
State New, archived
Headers show
Series btrfs: add data write support for subpage | expand

Commit Message

Qu Wenruo May 31, 2021, 8:50 a.m. UTC
[BUG]
With current btrfs subpage rw support, the following script can lead to
fs hang:

  mkfs.btrfs -f -s 4k $dev
  mount $dev -o nospace_cache $mnt

  fsstress -w -n 100 -p 1 -s 1608140256 -v -d $mnt

The fs will hang at btrfs_start_ordered_extent().

[CAUSE]
In above test case, btrfs_invalidate() will be called with the following
parameters:
  offset = 0 length = 53248 page dirty = 1 subpage dirty bitmap = 0x2000

Since @offset is 0, btrfs_invalidate() will try to invalidate the full
page, and finally call clear_page_extent_mapped() which will detach
btrfs subpage structure from the page.

And since the page no longer has btrfs subpage structure, the subpage
dirty bitmap will be cleared, preventing the dirty range from
written back, thus no way to wake up the ordered extent.

[FIX]
Just follow other fses, only to invalidate the page if the range covers
the full page.

There are cases like truncate_setsize() which can call
btrfs_invalidatepage() with offset == 0 and length != 0 for the last
page of an inode.

Although the old code will still try to invalidate the full page, we are
still safe to just wait for ordered extent to finish.
So it shouldn't cause extra problems.

Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 fs/btrfs/inode.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
diff mbox series

Patch

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index e756423a7a70..c3d5abf3df5f 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -8363,7 +8363,19 @@  static void btrfs_invalidatepage(struct page *page, unsigned int offset,
 	 */
 	wait_on_page_writeback(page);
 
-	if (offset) {
+	/*
+	 * For subpage case, we have call sites like
+	 * btrfs_punch_hole_lock_range() which passes range not aligned to
+	 * sectorsize.
+	 * If the range doesn't cover the full page, we don't need to and
+	 * shouldn't clear page extent mapped, as page->private can still
+	 * record subpage dirty bits for other part of the range.
+	 *
+	 * For cases where can invalidate the full even the range doesn't
+	 * cover the full page, like invalidating the last page, we're
+	 * still safe to wait for ordered extent to finish.
+	 */
+	if (!(offset == 0 && length == PAGE_SIZE)) {
 		btrfs_releasepage(page, GFP_NOFS);
 		return;
 	}