diff mbox series

[5/6] btrfs: remove pointless barrier from btrfs_sync_file()

Message ID ba7f7646404185a0713f55c50be90e659d29a566.1696415673.git.fdmanana@suse.com (mailing list archive)
State New, archived
Headers show
Series btrfs: fix some data races during fsync and cleanups | expand

Commit Message

Filipe Manana Oct. 4, 2023, 10:38 a.m. UTC 
From: Filipe Manana <fdmanana@suse.com>

The memory barrier (smp_mb()) at btrfs_sync_file() is completely redundant
now that fs_info->last_trans_committed is read using READ_ONCE(), with the
helper btrfs_get_last_trans_committed(), and written using WRITE_ONCE()
with the helper btrfs_set_last_trans_committed().

This barrier was introduced in 2011, by commit a4abeea41adf ("Btrfs: kill
trans_mutex"), but even back then it was not correct since the writer side
(in btrfs_commit_transaction()), did not issue a pairing memory barrier
after it updated fs_info->last_trans_committed.

So remove this barrier.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
---
 fs/btrfs/file.c | 1 -
 1 file changed, 1 deletion(-)
diff mbox series

Patch

diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 92e6f224bff9..92419cb8508a 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1889,7 +1889,6 @@  int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 
 	atomic_inc(&root->log_batch);
 
-	smp_mb();
 	if (skip_inode_logging(&ctx)) {
 		/*
 		 * We've had everything committed since the last time we were