4.14-stable review patch. If anyone has any objections, please let me know.
------------------
From: Carlos Maiolino cmaiolino@redhat.com
[ Upstream commit 373b0589dc8d58bc09c9a28d03611ae4fb216057 ]
Once the inode item writeback errors is already fixed, it's time to fix the same problem in dquot code.
Although there were no reports of users hitting this bug in dquot code (at least none I've seen), the bug is there and I was already planning to fix it when the correct approach to fix the inodes part was decided.
This patch aims to fix the same problem in dquot code, regarding failed buffers being unable to be resubmitted once they are flush locked.
Tested with the recently test-case sent to fstests list by Hou Tao.
Reviewed-by: Brian Foster bfoster@redhat.com Signed-off-by: Carlos Maiolino cmaiolino@redhat.com Reviewed-by: Darrick J. Wong darrick.wong@oracle.com Signed-off-by: Darrick J. Wong darrick.wong@oracle.com Signed-off-by: Sasha Levin alexander.levin@verizon.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- fs/xfs/xfs_dquot.c | 14 +++++++++++--- fs/xfs/xfs_dquot_item.c | 40 ++++++++++++++++++++++++++++++++++++++-- 2 files changed, 49 insertions(+), 5 deletions(-)
--- a/fs/xfs/xfs_dquot.c +++ b/fs/xfs/xfs_dquot.c @@ -987,14 +987,22 @@ xfs_qm_dqflush_done( * holding the lock before removing the dquot from the AIL. */ if ((lip->li_flags & XFS_LI_IN_AIL) && - lip->li_lsn == qip->qli_flush_lsn) { + ((lip->li_lsn == qip->qli_flush_lsn) || + (lip->li_flags & XFS_LI_FAILED))) {
/* xfs_trans_ail_delete() drops the AIL lock. */ spin_lock(&ailp->xa_lock); - if (lip->li_lsn == qip->qli_flush_lsn) + if (lip->li_lsn == qip->qli_flush_lsn) { xfs_trans_ail_delete(ailp, lip, SHUTDOWN_CORRUPT_INCORE); - else + } else { + /* + * Clear the failed state since we are about to drop the + * flush lock + */ + if (lip->li_flags & XFS_LI_FAILED) + xfs_clear_li_failed(lip); spin_unlock(&ailp->xa_lock); + } }
/* --- a/fs/xfs/xfs_dquot_item.c +++ b/fs/xfs/xfs_dquot_item.c @@ -137,6 +137,26 @@ xfs_qm_dqunpin_wait( wait_event(dqp->q_pinwait, (atomic_read(&dqp->q_pincount) == 0)); }
+/* + * Callback used to mark a buffer with XFS_LI_FAILED when items in the buffer + * have been failed during writeback + * + * this informs the AIL that the dquot is already flush locked on the next push, + * and acquires a hold on the buffer to ensure that it isn't reclaimed before + * dirty data makes it to disk. + */ +STATIC void +xfs_dquot_item_error( + struct xfs_log_item *lip, + struct xfs_buf *bp) +{ + struct xfs_dquot *dqp; + + dqp = DQUOT_ITEM(lip)->qli_dquot; + ASSERT(!completion_done(&dqp->q_flush)); + xfs_set_li_failed(lip, bp); +} + STATIC uint xfs_qm_dquot_logitem_push( struct xfs_log_item *lip, @@ -144,13 +164,28 @@ xfs_qm_dquot_logitem_push( __acquires(&lip->li_ailp->xa_lock) { struct xfs_dquot *dqp = DQUOT_ITEM(lip)->qli_dquot; - struct xfs_buf *bp = NULL; + struct xfs_buf *bp = lip->li_buf; uint rval = XFS_ITEM_SUCCESS; int error;
if (atomic_read(&dqp->q_pincount) > 0) return XFS_ITEM_PINNED;
+ /* + * The buffer containing this item failed to be written back + * previously. Resubmit the buffer for IO + */ + if (lip->li_flags & XFS_LI_FAILED) { + if (!xfs_buf_trylock(bp)) + return XFS_ITEM_LOCKED; + + if (!xfs_buf_resubmit_failed_buffers(bp, lip, buffer_list)) + rval = XFS_ITEM_FLUSHING; + + xfs_buf_unlock(bp); + return rval; + } + if (!xfs_dqlock_nowait(dqp)) return XFS_ITEM_LOCKED;
@@ -242,7 +277,8 @@ static const struct xfs_item_ops xfs_dqu .iop_unlock = xfs_qm_dquot_logitem_unlock, .iop_committed = xfs_qm_dquot_logitem_committed, .iop_push = xfs_qm_dquot_logitem_push, - .iop_committing = xfs_qm_dquot_logitem_committing + .iop_committing = xfs_qm_dquot_logitem_committing, + .iop_error = xfs_dquot_item_error };
/*