btrfs: adjust dirty_metadata_bytes after writeback failure of extent buffer

Before, if a eb failed to write out, we would end up triggering a
BUG_ON(). As of f4340622e0226 ("btrfs: extent_io: Move the BUG_ON() in
flush_write_bio() one level up"), we no longer BUG_ON(), so we should
make life consistent and add back the unwritten bytes to
dirty_metadata_bytes.

Fixes: f4340622e022 ("btrfs: extent_io: Move the BUG_ON() in flush_write_bio() one level up")
CC: stable@vger.kernel.org # 5.2+
Reviewed-by: Filipe Manana <fdmanana@kernel.org>
Signed-off-by: Dennis Zhou <dennis@kernel.org>
Signed-off-by: David Sterba <dsterba@suse.com>

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 4dc5e69..67066ec 100644 (file)
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -3728,12 +3728,21 @@ static void end_extent_buffer_writeback(struct extent_buffer *eb)
 static void set_btree_ioerr(struct page *page)
 {
        struct extent_buffer *eb = (struct extent_buffer *)page->private;
+       struct btrfs_fs_info *fs_info;
 
        SetPageError(page);
        if (test_and_set_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags))
                return;
 
        /*
+        * If we error out, we should add back the dirty_metadata_bytes
+        * to make it consistent.
+        */
+       fs_info = eb->fs_info;
+       percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
+                                eb->len, fs_info->dirty_metadata_batch);
+
+       /*
         * If writeback for a btree extent that doesn't belong to a log tree
         * failed, increment the counter transaction->eb_write_errors.
         * We do this because while the transaction is running and before it's