xfs: on-stack delayed write buffer lists

Queue delwri buffers on a local on-stack list instead of a per-buftarg one,
and write back the buffers per-process instead of by waking up xfsbufd.

This is now easily doable given that we have very few places left that write
delwri buffers:

 - log recovery:
	Only done at mount time, and already forcing out the buffers
	synchronously using xfs_flush_buftarg

 - quotacheck:
	Same story.

 - dquot reclaim:
	Writes out dirty dquots on the LRU under memory pressure.  We might
	want to look into doing more of this via xfsaild, but it's already
	more optimal than the synchronous inode reclaim that writes each
	buffer synchronously.

 - xfsaild:
	This is the main beneficiary of the change.  By keeping a local list
	of buffers to write we reduce latency of writing out buffers, and
	more importably we can remove all the delwri list promotions which
	were hitting the buffer cache hard under sustained metadata loads.

The implementation is very straight forward - xfs_buf_delwri_queue now gets
a new list_head pointer that it adds the delwri buffers to, and all callers
need to eventually submit the list using xfs_buf_delwi_submit or
xfs_buf_delwi_submit_nowait.  Buffers that already are on a delwri list are
skipped in xfs_buf_delwri_queue, assuming they already are on another delwri
list.  The biggest change to pass down the buffer list was done to the AIL
pushing. Now that we operate on buffers the trylock, push and pushbuf log
item methods are merged into a single push routine, which tries to lock the
item, and if possible add the buffer that needs writeback to the buffer list.
This leads to much simpler code than the previous split but requires the
individual IOP_PUSH instances to unlock and reacquire the AIL around calls
to blocking routines.

Given that xfsailds now also handle writing out buffers, the conditions for
log forcing and the sleep times needed some small changes.  The most
important one is that we consider an AIL busy as long we still have buffers
to push, and the other one is that we do increment the pushed LSN for
buffers that are under flushing at this moment, but still count them towards
the stuck items for restart purposes.  Without this we could hammer on stuck
items without ever forcing the log and not make progress under heavy random
delete workloads on fast flash storage devices.

[ Dave Chinner:
	- rebase on previous patches.
	- improved comments for XBF_DELWRI_Q handling
	- fix XBF_ASYNC handling in queue submission (test 106 failure)
	- rename delwri submit function buffer list parameters for clarity
	- xfs_efd_item_push() should return XFS_ITEM_PINNED ]

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>

1 files changed, 71 insertions, 77 deletions

diff --git a/fs/xfs/xfs_qm.c b/fs/xfs/xfs_qm.c
index 95aecf52475..755a9bd749d 100644
--- a/fs/xfs/xfs_qm.c
+++ b/fs/xfs/xfs_qm.c
@@ -65,7 +65,8 @@ STATIC int
 xfs_qm_dquot_walk(
 	struct xfs_mount	*mp,
 	int			type,
-	int			(*execute)(struct xfs_dquot *dqp))
+	int			(*execute)(struct xfs_dquot *dqp, void *data),
+	void			*data)
 {
 	struct xfs_quotainfo	*qi = mp->m_quotainfo;
 	struct radix_tree_root	*tree = XFS_DQUOT_TREE(qi, type);
@@ -97,7 +98,7 @@ restart:
 
 			next_index = be32_to_cpu(dqp->q_core.d_id) + 1;
 
-			error = execute(batch[i]);
+			error = execute(batch[i], data);
 			if (error == EAGAIN) {
 				skipped++;
 				continue;
@@ -129,7 +130,8 @@ restart:
  */
 STATIC int
 xfs_qm_dqpurge(
-	struct xfs_dquot	*dqp)
+	struct xfs_dquot	*dqp,
+	void			*data)
 {
 	struct xfs_mount	*mp = dqp->q_mount;
 	struct xfs_quotainfo	*qi = mp->m_quotainfo;
@@ -153,21 +155,7 @@ xfs_qm_dqpurge(
 
 	dqp->dq_flags |= XFS_DQ_FREEING;
 
-	/*
-	 * If we're turning off quotas, we have to make sure that, for
-	 * example, we don't delete quota disk blocks while dquots are
-	 * in the process of getting written to those disk blocks.
-	 * This dquot might well be on AIL, and we can't leave it there
-	 * if we're turning off quotas. Basically, we need this flush
-	 * lock, and are willing to block on it.
-	 */
-	if (!xfs_dqflock_nowait(dqp)) {
-		/*
-		 * Block on the flush lock after nudging dquot buffer,
-		 * if it is incore.
-		 */
-		xfs_dqflock_pushbuf_wait(dqp);
-	}
+	xfs_dqflock(dqp);
 
 	/*
 	 * If we are turning this type of quotas off, we don't care
@@ -231,11 +219,11 @@ xfs_qm_dqpurge_all(
 	uint			flags)
 {
 	if (flags & XFS_QMOPT_UQUOTA)
-		xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_dqpurge);
+		xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_dqpurge, NULL);
 	if (flags & XFS_QMOPT_GQUOTA)
-		xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_dqpurge);
+		xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_dqpurge, NULL);
 	if (flags & XFS_QMOPT_PQUOTA)
-		xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_dqpurge);
+		xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_dqpurge, NULL);
 }
 
 /*
@@ -876,15 +864,16 @@ xfs_qm_reset_dqcounts(
 
 STATIC int
 xfs_qm_dqiter_bufs(
-	xfs_mount_t	*mp,
-	xfs_dqid_t	firstid,
-	xfs_fsblock_t	bno,
-	xfs_filblks_t	blkcnt,
-	uint		flags)
+	struct xfs_mount	*mp,
+	xfs_dqid_t		firstid,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		blkcnt,
+	uint			flags,
+	struct list_head	*buffer_list)
 {
-	xfs_buf_t	*bp;
-	int		error;
-	int		type;
+	struct xfs_buf		*bp;
+	int			error;
+	int			type;
 
 	ASSERT(blkcnt > 0);
 	type = flags & XFS_QMOPT_UQUOTA ? XFS_DQ_USER :
@@ -908,7 +897,7 @@ xfs_qm_dqiter_bufs(
 			break;
 
 		xfs_qm_reset_dqcounts(mp, bp, firstid, type);
-		xfs_buf_delwri_queue(bp);
+		xfs_buf_delwri_queue(bp, buffer_list);
 		xfs_buf_relse(bp);
 		/*
 		 * goto the next block.
@@ -916,6 +905,7 @@ xfs_qm_dqiter_bufs(
 		bno++;
 		firstid += mp->m_quotainfo->qi_dqperchunk;
 	}
+
 	return error;
 }
 
@@ -925,11 +915,12 @@ xfs_qm_dqiter_bufs(
  */
 STATIC int
 xfs_qm_dqiterate(
-	xfs_mount_t	*mp,
-	xfs_inode_t	*qip,
-	uint		flags)
+	struct xfs_mount	*mp,
+	struct xfs_inode	*qip,
+	uint			flags,
+	struct list_head	*buffer_list)
 {
-	xfs_bmbt_irec_t		*map;
+	struct xfs_bmbt_irec	*map;
 	int			i, nmaps;	/* number of map entries */
 	int			error;		/* return value */
 	xfs_fileoff_t		lblkno;
@@ -996,21 +987,17 @@ xfs_qm_dqiterate(
 			 * Iterate thru all the blks in the extent and
 			 * reset the counters of all the dquots inside them.
 			 */
-			if ((error = xfs_qm_dqiter_bufs(mp,
-						       firstid,
-						       map[i].br_startblock,
-						       map[i].br_blockcount,
-						       flags))) {
-				break;
-			}
+			error = xfs_qm_dqiter_bufs(mp, firstid,
+						   map[i].br_startblock,
+						   map[i].br_blockcount,
+						   flags, buffer_list);
+			if (error)
+				goto out;
 		}
-
-		if (error)
-			break;
 	} while (nmaps > 0);
 
+out:
 	kmem_free(map);
-
 	return error;
 }
 
@@ -1203,8 +1190,10 @@ error0:
 
 STATIC int
 xfs_qm_flush_one(
-	struct xfs_dquot	*dqp)
+	struct xfs_dquot	*dqp,
+	void			*data)
 {
+	struct list_head	*buffer_list = data;
 	struct xfs_buf		*bp = NULL;
 	int			error = 0;
 
@@ -1214,14 +1203,12 @@ xfs_qm_flush_one(
 	if (!XFS_DQ_IS_DIRTY(dqp))
 		goto out_unlock;
 
-	if (!xfs_dqflock_nowait(dqp))
-		xfs_dqflock_pushbuf_wait(dqp);
-
+	xfs_dqflock(dqp);
 	error = xfs_qm_dqflush(dqp, &bp);
 	if (error)
 		goto out_unlock;
 
-	xfs_buf_delwri_queue(bp);
+	xfs_buf_delwri_queue(bp, buffer_list);
 	xfs_buf_relse(bp);
 out_unlock:
 	xfs_dqunlock(dqp);
@@ -1241,6 +1228,7 @@ xfs_qm_quotacheck(
 	size_t		structsz;
 	xfs_inode_t	*uip, *gip;
 	uint		flags;
+	LIST_HEAD	(buffer_list);
 
 	count = INT_MAX;
 	structsz = 1;
@@ -1259,7 +1247,8 @@ xfs_qm_quotacheck(
 	 */
 	uip = mp->m_quotainfo->qi_uquotaip;
 	if (uip) {
-		error = xfs_qm_dqiterate(mp, uip, XFS_QMOPT_UQUOTA);
+		error = xfs_qm_dqiterate(mp, uip, XFS_QMOPT_UQUOTA,
+					 &buffer_list);
 		if (error)
 			goto error_return;
 		flags |= XFS_UQUOTA_CHKD;
@@ -1268,7 +1257,8 @@ xfs_qm_quotacheck(
 	gip = mp->m_quotainfo->qi_gquotaip;
 	if (gip) {
 		error = xfs_qm_dqiterate(mp, gip, XFS_IS_GQUOTA_ON(mp) ?
-					XFS_QMOPT_GQUOTA : XFS_QMOPT_PQUOTA);
+					 XFS_QMOPT_GQUOTA : XFS_QMOPT_PQUOTA,
+					 &buffer_list);
 		if (error)
 			goto error_return;
 		flags |= XFS_OQUOTA_CHKD;
@@ -1291,19 +1281,27 @@ xfs_qm_quotacheck(
 	 * We've made all the changes that we need to make incore.  Flush them
 	 * down to disk buffers if everything was updated successfully.
 	 */
-	if (XFS_IS_UQUOTA_ON(mp))
-		error = xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_flush_one);
+	if (XFS_IS_UQUOTA_ON(mp)) {
+		error = xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_flush_one,
+					  &buffer_list);
+	}
 	if (XFS_IS_GQUOTA_ON(mp)) {
-		error2 = xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_flush_one);
+		error2 = xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_flush_one,
+					   &buffer_list);
 		if (!error)
 			error = error2;
 	}
 	if (XFS_IS_PQUOTA_ON(mp)) {
-		error2 = xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_flush_one);
+		error2 = xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_flush_one,
+					   &buffer_list);
 		if (!error)
 			error = error2;
 	}
 
+	error2 = xfs_buf_delwri_submit(&buffer_list);
+	if (!error)
+		error = error2;
+
 	/*
 	 * We can get this error if we couldn't do a dquot allocation inside
 	 * xfs_qm_dqusage_adjust (via bulkstat). We don't care about the
@@ -1317,15 +1315,6 @@ xfs_qm_quotacheck(
 	}
 
 	/*
-	 * We didn't log anything, because if we crashed, we'll have to
-	 * start the quotacheck from scratch anyway. However, we must make
-	 * sure that our dquot changes are secure before we put the
-	 * quotacheck'd stamp on the superblock. So, here we do a synchronous
-	 * flush.
-	 */
-	xfs_flush_buftarg(mp->m_ddev_targp, 1);
-
-	/*
 	 * If one type of quotas is off, then it will lose its
 	 * quotachecked status, since we won't be doing accounting for
 	 * that type anymore.
@@ -1334,6 +1323,13 @@ xfs_qm_quotacheck(
 	mp->m_qflags |= flags;
 
  error_return:
+	while (!list_empty(&buffer_list)) {
+		struct xfs_buf *bp =
+			list_first_entry(&buffer_list, struct xfs_buf, b_list);
+		list_del_init(&bp->b_list);
+		xfs_buf_relse(bp);
+	}
+
 	if (error) {
 		xfs_warn(mp,
 	"Quotacheck: Unsuccessful (Error %d): Disabling quotas.",
@@ -1450,6 +1446,7 @@ xfs_qm_dqfree_one(
 STATIC void
 xfs_qm_dqreclaim_one(
 	struct xfs_dquot	*dqp,
+	struct list_head	*buffer_list,
 	struct list_head	*dispose_list)
 {
 	struct xfs_mount	*mp = dqp->q_mount;
@@ -1482,21 +1479,11 @@ xfs_qm_dqreclaim_one(
 	if (!xfs_dqflock_nowait(dqp))
 		goto out_busy;
 
-	/*
-	 * We have the flush lock so we know that this is not in the
-	 * process of being flushed. So, if this is dirty, flush it
-	 * DELWRI so that we don't get a freelist infested with
-	 * dirty dquots.
-	 */
 	if (XFS_DQ_IS_DIRTY(dqp)) {
 		struct xfs_buf	*bp = NULL;
 
 		trace_xfs_dqreclaim_dirty(dqp);
 
-		/*
-		 * We flush it delayed write, so don't bother releasing the
-		 * freelist lock.
-		 */
 		error = xfs_qm_dqflush(dqp, &bp);
 		if (error) {
 			xfs_warn(mp, "%s: dquot %p flush failed",
@@ -1504,7 +1491,7 @@ xfs_qm_dqreclaim_one(
 			goto out_busy;
 		}
 
-		xfs_buf_delwri_queue(bp);
+		xfs_buf_delwri_queue(bp, buffer_list);
 		xfs_buf_relse(bp);
 		/*
 		 * Give the dquot another try on the freelist, as the
@@ -1549,8 +1536,10 @@ xfs_qm_shake(
 	struct xfs_quotainfo	*qi =
 		container_of(shrink, struct xfs_quotainfo, qi_shrinker);
 	int			nr_to_scan = sc->nr_to_scan;
+	LIST_HEAD		(buffer_list);
 	LIST_HEAD		(dispose_list);
 	struct xfs_dquot	*dqp;
+	int			error;
 
 	if ((sc->gfp_mask & (__GFP_FS|__GFP_WAIT)) != (__GFP_FS|__GFP_WAIT))
 		return 0;
@@ -1563,15 +1552,20 @@ xfs_qm_shake(
 			break;
 		dqp = list_first_entry(&qi->qi_lru_list, struct xfs_dquot,
 				       q_lru);
-		xfs_qm_dqreclaim_one(dqp, &dispose_list);
+		xfs_qm_dqreclaim_one(dqp, &buffer_list, &dispose_list);
 	}
 	mutex_unlock(&qi->qi_lru_lock);
 
+	error = xfs_buf_delwri_submit(&buffer_list);
+	if (error)
+		xfs_warn(NULL, "%s: dquot reclaim failed", __func__);
+
 	while (!list_empty(&dispose_list)) {
 		dqp = list_first_entry(&dispose_list, struct xfs_dquot, q_lru);
 		list_del_init(&dqp->q_lru);
 		xfs_qm_dqfree_one(dqp);
 	}
+
 out:
 	return (qi->qi_lru_count / 100) * sysctl_vfs_cache_pressure;
 }