On Thu, Sep 22, 2022 at 06:47:27PM +0300, Amir Goldstein wrote:
From: Dave Chinner dchinner@redhat.com
commit 9a5280b312e2e7898b6397b2ca3cfd03f67d7be1 upstream.
[backport for 5.10.y]
The O_TMPFILE creation implementation creates a specific order of operations for inode allocation/freeing and unlinked list modification. Currently both are serialised by the AGI, so the order doesn't strictly matter as long as the are both in the same transaction.
However, if we want to move the unlinked list insertions largely out from under the AGI lock, then we have to be concerned about the order in which we do unlinked list modification operations. O_TMPFILE creation tells us this order is inode allocation/free, then unlinked list modification.
Change xfs_ifree() to use this same ordering on unlinked list removal. This way we always guarantee that when we enter the iunlinked list removal code from this path, we already have the AGI locked and we don't have to worry about lock nesting AGI reads inside unlink list locks because it's already locked and attached to the transaction.
We can do this safely as the inode freeing and unlinked list removal are done in the same transaction and hence are atomic operations with respect to log recovery.
Reported-by: Frank Hofmann fhofmann@cloudflare.com Fixes: 298f7bec503f ("xfs: pin inode backing buffer to the inode log item") Signed-off-by: Dave Chinner dchinner@redhat.com Reviewed-by: Darrick J. Wong darrick.wong@oracle.com Signed-off-by: Dave Chinner david@fromorbit.com Signed-off-by: Amir Goldstein amir73il@gmail.com Acked-by: Darrick J. Wong djwong@kernel.org
fs/xfs/xfs_inode.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-)
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index 1f61e085676b..929ed3bc5619 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -2669,14 +2669,13 @@ xfs_ifree_cluster( } /*
- This is called to return an inode to the inode free list.
- The inode should already be truncated to 0 length and have
- no pages associated with it. This routine also assumes that
- the inode is already a part of the transaction.
- This is called to return an inode to the inode free list. The inode should
- already be truncated to 0 length and have no pages associated with it. This
- routine also assumes that the inode is already a part of the transaction.
- The on-disk copy of the inode will have been added to the list
- of unlinked inodes in the AGI. We need to remove the inode from
- that list atomically with respect to freeing it here.
- The on-disk copy of the inode will have been added to the list of unlinked
- inodes in the AGI. We need to remove the inode from that list atomically with
*/
- respect to freeing it here.
int xfs_ifree( @@ -2694,13 +2693,16 @@ xfs_ifree( ASSERT(ip->i_d.di_nblocks == 0); /*
* Pull the on-disk inode from the AGI unlinked list.
* Free the inode first so that we guarantee that the AGI lock is going* to be taken before we remove the inode from the unlinked list. This* makes the AGI lock -> unlinked list modification order the same as* used in O_TMPFILE creation.
- error = xfs_iunlink_remove(tp, ip);
- error = xfs_difree(tp, ip->i_ino, &xic); if (error) return error;
- error = xfs_difree(tp, ip->i_ino, &xic);
- error = xfs_iunlink_remove(tp, ip); if (error) return error;
2.25.1
Any specific reason you do not want 6f5097e3367a ("xfs: fix xfs_ifree() error handling to not leak perag ref") also applied? That commit fixes this one (or so it says.)
That is part of the 5.15 queue right now, but not 5.10, is that a problem?
thanks,
greg k-h