From: Josef Bacik josef@toxicpanda.com
[ Upstream commit a7d16d9a07bbcb7dcd5214a1bea75c808830bc0d ]
This is a long time leftover from when I originally added the free space inode, the point was to catch cases where we weren't honoring the NOCOW flag. However there exists a race with relocation, if we allocate our free space inode in a block group that is about to be relocated, we could trigger the COW path before the relocation has the opportunity to find the extents and delete the free space cache. In production where we have auto-relocation enabled we're seeing this WARN_ON_ONCE() around 5k times in a 2 week period, so not super common but enough that it's at the top of our metrics.
We're properly handling the error here, and with us phasing out v1 space cache anyway just drop the WARN_ON_ONCE.
Signed-off-by: Josef Bacik josef@toxicpanda.com Reviewed-by: David Sterba dsterba@suse.com Signed-off-by: David Sterba dsterba@suse.com Signed-off-by: Sasha Levin sashal@kernel.org --- fs/btrfs/inode.c | 1 - 1 file changed, 1 deletion(-)
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 0f4408f9dadd..ffa2fde25f9c 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -1130,7 +1130,6 @@ static noinline int cow_file_range(struct btrfs_inode *inode, int ret = 0;
if (btrfs_is_free_space_inode(inode)) { - WARN_ON_ONCE(1); ret = -EINVAL; goto out_unlock; }