diff mbox series

[13/15] btrfs: Call iomap_dio_complete() without inode_lock

Message ID 20200921144353.31319-14-rgoldwyn@suse.de (mailing list archive)
State New, archived
Headers show
Series BTRFS DIO inode locking/D_SYNC fix | expand

Commit Message

Goldwyn Rodrigues Sept. 21, 2020, 2:43 p.m. UTC 
From: Goldwyn Rodrigues <rgoldwyn@suse.com>

If direct writes are called with O_DIRECT | O_DSYNC, it will result in a
deadlock because iomap_dio_rw() is called under i_rwsem which calls
iomap_dio_complete()
  generic_write_sync()
    btrfs_sync_file().

btrfs_sync_file() requires i_rwsem, so call __iomap_dio_rw() with the
i_rwsem locked, and call iomap_dio_complete() after unlocking i_rwsem.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
---
 fs/btrfs/file.c | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

Comments

Josef Bacik Sept. 22, 2020, 3:11 p.m. UTC | #1 
On 9/21/20 10:43 AM, Goldwyn Rodrigues wrote:
> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
> 
> If direct writes are called with O_DIRECT | O_DSYNC, it will result in a
> deadlock because iomap_dio_rw() is called under i_rwsem which calls
> iomap_dio_complete()
>    generic_write_sync()
>      btrfs_sync_file().
> 
> btrfs_sync_file() requires i_rwsem, so call __iomap_dio_rw() with the
> i_rwsem locked, and call iomap_dio_complete() after unlocking i_rwsem.
> 
> Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>

Reviewed-by: Josef Bacik <josef@toxicpanda.com>

Thanks,

Josef
diff mbox series

Patch

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 193af84f5405..9c7a2d4b4148 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1981,6 +1981,7 @@  static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
 	loff_t endbyte;
 	int err;
 	int ilock_flags = 0;
+	struct iomap_dio *dio = NULL;
 
 	if (iocb->ki_flags & IOCB_NOWAIT)
 		ilock_flags |= BTRFS_ILOCK_TRY;
@@ -2022,22 +2023,19 @@  static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
 		goto buffered;
 	}
 
-	/*
-	 * We have are actually a sync iocb, so we need our fancy endio to know
-	 * if we need to sync.
-	 */
-	if (current->journal_info)
-		written = iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops,
-				&btrfs_sync_dops, is_sync_kiocb(iocb));
-	else
-		written = iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops,
-				&btrfs_dio_ops, is_sync_kiocb(iocb));
-
-	if (written == -ENOTBLK)
-		written = 0;
+	dio = __iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops,
+			&btrfs_dio_ops, is_sync_kiocb(iocb));
 
 	btrfs_inode_unlock(inode, ilock_flags);
 
+	if (IS_ERR_OR_NULL(dio)) {
+		err = PTR_ERR_OR_ZERO(dio);
+		if (err < 0 && err != -ENOTBLK)
+			goto out;
+	} else {
+		written = iomap_dio_complete(dio);
+	}
+
 	if (written < 0 || !iov_iter_count(from)) {
 		err = written;
 		goto out;