xfs/fstests: largeish pile of bug fixes

index 82f2e0dd224997..3290dd8524a69a 100644
@@ -163,7 +163,8 @@ struct xfs_getfsmap_info {
 	xfs_daddr_t		next_daddr;	/* next daddr we expect */
 	/* daddr of low fsmap key when we're using the rtbitmap */
 	xfs_daddr_t		low_daddr;
-	xfs_daddr_t		end_daddr;	/* daddr of high fsmap key */
+	/* daddr of high fsmap key, or the last daddr on the device */
+	xfs_daddr_t		end_daddr;
 	u64			missing_owner;	/* owner of holes */
 	u32			dev;		/* device id */
 	/*

@@ -387,8 +388,8 @@ xfs_getfsmap_group_helper(
 	 * we calculated from userspace's high key to synthesize the record.
 	 * Note that if the btree query found a mapping, there won't be a gap.
 	 */
-	if (info->last && info->end_daddr != XFS_BUF_DADDR_NULL)
-		frec->start_daddr = info->end_daddr;
+	if (info->last)
+		frec->start_daddr = info->end_daddr + 1;
 	else
 		frec->start_daddr = xfs_gbno_to_daddr(xg, startblock);
 

@@ -736,11 +737,10 @@ xfs_getfsmap_rtdev_rtbitmap_helper(
 	 * we calculated from userspace's high key to synthesize the record.
 	 * Note that if the btree query found a mapping, there won't be a gap.
 	 */
-	if (info->last && info->end_daddr != XFS_BUF_DADDR_NULL) {
-		frec.start_daddr = info->end_daddr;
-	} else {
+	if (info->last)
+		frec.start_daddr = info->end_daddr + 1;
+	else
 		frec.start_daddr = xfs_rtb_to_daddr(mp, start_rtb);
-	}
 
 	frec.len_daddr = XFS_FSB_TO_BB(mp, rtbcount);
 	return xfs_getfsmap_helper(tp, info, &frec);

@@ -933,7 +933,10 @@ xfs_getfsmap(
 	struct xfs_trans		*tp = NULL;
 	struct xfs_fsmap		dkeys[2];	/* per-dev keys */
 	struct xfs_getfsmap_dev		handlers[XFS_GETFSMAP_DEVS];
-	struct xfs_getfsmap_info	info = { NULL };
+	struct xfs_getfsmap_info	info = {
+		.fsmap_recs		= fsmap_recs,
+		.head			= head,
+	};
 	bool				use_rmap;
 	int				i;
 	int				error = 0;

@@ -998,9 +1001,6 @@ xfs_getfsmap(
 
 	info.next_daddr = head->fmh_keys[0].fmr_physical +
 			  head->fmh_keys[0].fmr_length;
-	info.end_daddr = XFS_BUF_DADDR_NULL;
-	info.fsmap_recs = fsmap_recs;
-	info.head = head;
 
 	/* For each device we support... */
 	for (i = 0; i < XFS_GETFSMAP_DEVS; i++) {

@@ -1013,17 +1013,23 @@ xfs_getfsmap(
 			break;
 
 		/*
-		 * If this device number matches the high key, we have
-		 * to pass the high key to the handler to limit the
-		 * query results.  If the device number exceeds the
-		 * low key, zero out the low key so that we get
-		 * everything from the beginning.
+		 * If this device number matches the high key, we have to pass
+		 * the high key to the handler to limit the query results, and
+		 * set the end_daddr so that we can synthesize records at the
+		 * end of the query range or device.
 		 */
 		if (handlers[i].dev == head->fmh_keys[1].fmr_device) {
 			dkeys[1] = head->fmh_keys[1];
 			info.end_daddr = min(handlers[i].nr_sectors - 1,
 					     dkeys[1].fmr_physical);
+		} else {
+			info.end_daddr = handlers[i].nr_sectors - 1;
 		}
+
+		/*
+		 * If the device number exceeds the low key, zero out the low
+		 * key so that we get everything from the beginning.
+		 */
 		if (handlers[i].dev > head->fmh_keys[0].fmr_device)
 			memset(&dkeys[0], 0, sizeof(struct xfs_fsmap));
 

index b78db651346518..80467d6bc76389 100644
@@ -196,36 +196,45 @@ xchk_setup_metapath_dqinode(
 	struct xfs_scrub	*sc,
 	xfs_dqtype_t		type)
 {
+	struct xfs_quotainfo	*qi = sc->mp->m_quotainfo;
 	struct xfs_trans	*tp = NULL;
 	struct xfs_inode	*dp = NULL;
 	struct xfs_inode	*ip = NULL;
-	const char		*path;
 	int			error;
 
+	if (!qi)
+		return -ENOENT;
+
+	switch (type) {
+	case XFS_DQTYPE_USER:
+		ip = qi->qi_uquotaip;
+		break;
+	case XFS_DQTYPE_GROUP:
+		ip = qi->qi_gquotaip;
+		break;
+	case XFS_DQTYPE_PROJ:
+		ip = qi->qi_pquotaip;
+		break;
+	default:
+		ASSERT(0);
+		return -EINVAL;
+	}
+	if (!ip)
+		return -ENOENT;
+
 	error = xfs_trans_alloc_empty(sc->mp, &tp);
 	if (error)
 		return error;
 
 	error = xfs_dqinode_load_parent(tp, &dp);
-	if (error)
-		goto out_cancel;
-
-	error = xfs_dqinode_load(tp, dp, type, &ip);
-	if (error)
-		goto out_dp;
-
 	xfs_trans_cancel(tp);
-	tp = NULL;
+	if (error)
+		return error;
 
-	path = kasprintf(GFP_KERNEL, "%s", xfs_dqinode_path(type));
-	error = xchk_setup_metapath_scan(sc, dp, path, ip);
+	error = xchk_setup_metapath_scan(sc, dp,
+			kstrdup(xfs_dqinode_path(type), GFP_KERNEL), ip);
 
-	xfs_irele(ip);
-out_dp:
 	xfs_irele(dp);
-out_cancel:
-	if (tp)
-		xfs_trans_cancel(tp);
 	return error;
 }
 #else

index 80467d6bc76389..c678cba1ffc3f7 100644
@@ -171,23 +171,13 @@ static int
 xchk_setup_metapath_quotadir(
 	struct xfs_scrub	*sc)
 {
-	struct xfs_trans	*tp;
-	struct xfs_inode	*dp = NULL;
-	int			error;
+	struct xfs_quotainfo	*qi = sc->mp->m_quotainfo;
 
-	error = xfs_trans_alloc_empty(sc->mp, &tp);
-	if (error)
-		return error;
+	if (!qi || !qi->qi_dirip)
+		return -ENOENT;
 
-	error = xfs_dqinode_load_parent(tp, &dp);
-	xfs_trans_cancel(tp);
-	if (error)
-		return error;
-
-	error = xchk_setup_metapath_scan(sc, sc->mp->m_metadirip,
-			kasprintf(GFP_KERNEL, "quota"), dp);
-	xfs_irele(dp);
-	return error;
+	return xchk_setup_metapath_scan(sc, sc->mp->m_metadirip,
+			kstrdup("quota", GFP_KERNEL), qi->qi_dirip);
 }
 
 /* Scan a quota inode under the /quota directory. */

@@ -197,10 +187,7 @@ xchk_setup_metapath_dqinode(
 	xfs_dqtype_t		type)
 {
 	struct xfs_quotainfo	*qi = sc->mp->m_quotainfo;
-	struct xfs_trans	*tp = NULL;
-	struct xfs_inode	*dp = NULL;
 	struct xfs_inode	*ip = NULL;
-	int			error;
 
 	if (!qi)
 		return -ENOENT;

@@ -222,20 +209,8 @@ xchk_setup_metapath_dqinode(
 	if (!ip)
 		return -ENOENT;
 
-	error = xfs_trans_alloc_empty(sc->mp, &tp);
-	if (error)
-		return error;
-
-	error = xfs_dqinode_load_parent(tp, &dp);
-	xfs_trans_cancel(tp);
-	if (error)
-		return error;
-
-	error = xchk_setup_metapath_scan(sc, dp,
+	return xchk_setup_metapath_scan(sc, qi->qi_dirip,
 			kstrdup(xfs_dqinode_path(type), GFP_KERNEL), ip);
-
-	xfs_irele(dp);
-	return error;
 }
 #else
 # define xchk_setup_metapath_quotadir(...)	(-ENOENT)

128 lines not on this page · raw message

index e919c7f62f5780..35b64bc3a7a867 100644
@@ -55,6 +55,7 @@ struct xfs_quotainfo {
 	struct xfs_inode	*qi_uquotaip;	/* user quota inode */
 	struct xfs_inode	*qi_gquotaip;	/* group quota inode */
 	struct xfs_inode	*qi_pquotaip;	/* project quota inode */
+	struct xfs_inode	*qi_dirip;	/* quota metadir */
 	struct list_lru		qi_lru;
 	int			qi_dquots;
 	struct mutex		qi_quotaofflock;/* to serialize quotaoff */

index 2b5fc5fd16435d..c748866ef92368 100644
@@ -5144,7 +5144,7 @@ xfs_btree_count_blocks_helper(
 	int			level,
 	void			*data)
 {
-	xfs_extlen_t		*blocks = data;
+	xfs_filblks_t		*blocks = data;
 	(*blocks)++;
 
 	return 0;

@@ -5154,7 +5154,7 @@ xfs_btree_count_blocks_helper(
 int
 xfs_btree_count_blocks(
 	struct xfs_btree_cur	*cur,
-	xfs_extlen_t		*blocks)
+	xfs_filblks_t		*blocks)
 {
 	*blocks = 0;
 	return xfs_btree_visit_blocks(cur, xfs_btree_count_blocks_helper,

index 3b739459ebb0f4..c5bff273cae255 100644
@@ -484,7 +484,7 @@ typedef int (*xfs_btree_visit_blocks_fn)(struct xfs_btree_cur *cur, int level,
 int xfs_btree_visit_blocks(struct xfs_btree_cur *cur,
 		xfs_btree_visit_blocks_fn fn, unsigned int flags, void *data);
 
-int xfs_btree_count_blocks(struct xfs_btree_cur *cur, xfs_extlen_t *blocks);
+int xfs_btree_count_blocks(struct xfs_btree_cur *cur, xfs_filblks_t *blocks);
 
 union xfs_btree_rec *xfs_btree_rec_addr(struct xfs_btree_cur *cur, int n,
 		struct xfs_btree_block *block);

index 9b34896dd1a32f..6f270d8f4270cb 100644
@@ -744,6 +744,7 @@ xfs_finobt_count_blocks(
 {
 	struct xfs_buf		*agbp = NULL;
 	struct xfs_btree_cur	*cur;
+	xfs_filblks_t		blocks;
 	int			error;
 
 	error = xfs_ialloc_read_agi(pag, tp, 0, &agbp);

@@ -751,9 +752,10 @@ xfs_finobt_count_blocks(
 		return error;
 
 	cur = xfs_finobt_init_cursor(pag, tp, agbp);
-	error = xfs_btree_count_blocks(cur, tree_blocks);
+	error = xfs_btree_count_blocks(cur, &blocks);
 	xfs_btree_del_cursor(cur, error);
 	xfs_trans_brelse(tp, agbp);
+	*tree_blocks = blocks;
 
 	return error;
 }

index 61f80a6410c738..1d41b85478da9d 100644
@@ -458,7 +458,7 @@ xchk_agf_xref_btreeblks(
 {
 	struct xfs_agf		*agf = sc->sa.agf_bp->b_addr;
 	struct xfs_mount	*mp = sc->mp;
-	xfs_agblock_t		blocks;
+	xfs_filblks_t		blocks;
 	xfs_agblock_t		btreeblks;
 	int			error;
 

@@ -507,7 +507,7 @@ xchk_agf_xref_refcblks(
 	struct xfs_scrub	*sc)
 {
 	struct xfs_agf		*agf = sc->sa.agf_bp->b_addr;
-	xfs_agblock_t		blocks;
+	xfs_filblks_t		blocks;
 	int			error;
 
 	if (!sc->sa.refc_cur)

@@ -840,7 +840,7 @@ xchk_agi_xref_fiblocks(
 	struct xfs_scrub	*sc)
 {
 	struct xfs_agi		*agi = sc->sa.agi_bp->b_addr;
-	xfs_agblock_t		blocks;
+	xfs_filblks_t		blocks;
 	int			error = 0;
 
 	if (!xfs_has_inobtcounts(sc->mp))

index 0fad0baaba2f69..b45d2b32051a63 100644
@@ -256,7 +256,7 @@ xrep_agf_calc_from_btrees(
 	struct xfs_agf		*agf = agf_bp->b_addr;
 	struct xfs_mount	*mp = sc->mp;
 	xfs_agblock_t		btreeblks;
-	xfs_agblock_t		blocks;
+	xfs_filblks_t		blocks;
 	int			error;
 
 	/* Update the AGF counters from the bnobt. */

@@ -946,7 +946,7 @@ xrep_agi_calc_from_btrees(
 	if (error)
 		goto err;
 	if (xfs_has_inobtcounts(mp)) {
-		xfs_agblock_t	blocks;
+		xfs_filblks_t	blocks;
 
 		error = xfs_btree_count_blocks(cur, &blocks);
 		if (error)

@@ -959,7 +959,7 @@ xrep_agi_calc_from_btrees(
 	agi->agi_freecount = cpu_to_be32(freecount);
 
 	if (xfs_has_finobt(mp) && xfs_has_inobtcounts(mp)) {
-		xfs_agblock_t	blocks;
+		xfs_filblks_t	blocks;
 
 		cur = xfs_finobt_init_cursor(sc->sa.pag, sc->tp, agi_bp);
 		error = xfs_btree_count_blocks(cur, &blocks);

index 4a50f8e0004092..ca23cf4db6c5ef 100644
@@ -261,7 +261,7 @@ xchk_fscount_btreeblks(
 	struct xchk_fscounters	*fsc,
 	xfs_agnumber_t		agno)
 {
-	xfs_extlen_t		blocks;
+	xfs_filblks_t		blocks;
 	int			error;
 
 	error = xchk_ag_init_existing(sc, agno, &sc->sa);

index abad54c3621d44..4dc7c83dc08a40 100644
@@ -650,8 +650,8 @@ xchk_iallocbt_xref_rmap_btreeblks(
 	struct xfs_scrub	*sc)
 {
 	xfs_filblks_t		blocks;
-	xfs_extlen_t		inobt_blocks = 0;
-	xfs_extlen_t		finobt_blocks = 0;
+	xfs_filblks_t		inobt_blocks = 0;
+	xfs_filblks_t		finobt_blocks = 0;
 	int			error;
 
 	if (!sc->sa.ino_cur || !sc->sa.rmap_cur ||

index 2b6be75e942415..1c5e45cc64190c 100644
@@ -491,7 +491,7 @@ xchk_refcount_xref_rmap(
 	struct xfs_scrub	*sc,
 	xfs_filblks_t		cow_blocks)
 {
-	xfs_extlen_t		refcbt_blocks = 0;
+	xfs_filblks_t		refcbt_blocks = 0;
 	xfs_filblks_t		blocks;
 	int			error;
 

index a59bbe767a7dc4..0836fea2d6d814 100644
@@ -103,7 +103,7 @@ xfs_bmap_count_blocks(
 	struct xfs_mount	*mp = ip->i_mount;
 	struct xfs_ifork	*ifp = xfs_ifork_ptr(ip, whichfork);
 	struct xfs_btree_cur	*cur;
-	xfs_extlen_t		btblocks = 0;
+	xfs_filblks_t		btblocks = 0;
 	int			error;
 
 	*nextents = 0;

index c6de6b865ef11c..73562ff1c956f0 100644
@@ -1228,6 +1228,14 @@ xfs_file_remap_range(
 	xfs_iunlock2_remapping(src, dest);
 	if (ret)
 		trace_xfs_reflink_remap_range_error(dest, ret, _RET_IP_);
+	/*
+	 * If the caller did not set CAN_SHORTEN, then it is not prepared to
+	 * handle partial results -- either the whole remap succeeds, or we
+	 * must say why it did not.  In this case, any error should be returned
+	 * to the caller.
+	 */
+	if (ret && remapped < len && !(remap_flags & REMAP_FILE_CAN_SHORTEN))
+		return ret;
 	return remapped > 0 ? remapped : ret;
 }
 

index ce86bdad37fa42..ccc6ca5934ca6a 100644
@@ -71,7 +71,8 @@
 /* Map our scrub type to a sick mask and a set of health update functions. */
 
 enum xchk_health_group {
-	XHG_FS = 1,
+	XHG_NONE = 1,
+	XHG_FS,
 	XHG_AG,
 	XHG_INO,
 	XHG_RTGROUP,

@@ -83,6 +84,7 @@ struct xchk_health_map {
 };
 
 static const struct xchk_health_map type_to_health_flag[XFS_SCRUB_TYPE_NR] = {
+	[XFS_SCRUB_TYPE_PROBE]		= { XHG_NONE,  0 },
 	[XFS_SCRUB_TYPE_SB]		= { XHG_AG,  XFS_SICK_AG_SB },
 	[XFS_SCRUB_TYPE_AGF]		= { XHG_AG,  XFS_SICK_AG_AGF },
 	[XFS_SCRUB_TYPE_AGFL]		= { XHG_AG,  XFS_SICK_AG_AGFL },

@@ -133,7 +135,7 @@ xchk_mark_healthy_if_clean(
 {
 	if (!(sc->sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
 				  XFS_SCRUB_OFLAG_XCORRUPT)))
-		sc->sick_mask |= mask;
+		sc->healthy_mask |= mask;
 }
 
 /*

@@ -189,6 +191,7 @@ xchk_update_health(
 {
 	struct xfs_perag	*pag;
 	struct xfs_rtgroup	*rtg;
+	unsigned int		mask = sc->sick_mask;
 	bool			bad;
 
 	/*

@@ -203,50 +206,56 @@ xchk_update_health(
 		return;
 	}
 
-	if (!sc->sick_mask)
-		return;
-
 	bad = (sc->sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
 				   XFS_SCRUB_OFLAG_XCORRUPT));
+	if (!bad)
+		mask |= sc->healthy_mask;
 	switch (type_to_health_flag[sc->sm->sm_type].group) {
+	case XHG_NONE:
+		break;
 	case XHG_AG:
+		if (!mask)
+			return;
 		pag = xfs_perag_get(sc->mp, sc->sm->sm_agno);
 		if (bad)
-			xfs_group_mark_corrupt(pag_group(pag), sc->sick_mask);
+			xfs_group_mark_corrupt(pag_group(pag), mask);
 		else
-			xfs_group_mark_healthy(pag_group(pag), sc->sick_mask);
+			xfs_group_mark_healthy(pag_group(pag), mask);
 		xfs_perag_put(pag);
 		break;
 	case XHG_INO:
 		if (!sc->ip)
 			return;
-		if (bad) {
-			unsigned int	mask = sc->sick_mask;
-
-			/*
-			 * If we're coming in for repairs then we don't want
-			 * sickness flags to propagate to the incore health
-			 * status if the inode gets inactivated before we can
-			 * fix it.
-			 */
-			if (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)
-				mask |= XFS_SICK_INO_FORGET;
+		/*
+		 * If we're coming in for repairs then we don't want sickness
+		 * flags to propagate to the incore health status if the inode
+		 * gets inactivated before we can fix it.
+		 */
+		if (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)
+			mask |= XFS_SICK_INO_FORGET;
+		if (!mask)
+			return;
+		if (bad)
 			xfs_inode_mark_corrupt(sc->ip, mask);
-		} else
-			xfs_inode_mark_healthy(sc->ip, sc->sick_mask);
+		else
+			xfs_inode_mark_healthy(sc->ip, mask);
 		break;
 	case XHG_FS:
+		if (!mask)
+			return;
 		if (bad)
-			xfs_fs_mark_corrupt(sc->mp, sc->sick_mask);
+			xfs_fs_mark_corrupt(sc->mp, mask);
 		else
-			xfs_fs_mark_healthy(sc->mp, sc->sick_mask);
+			xfs_fs_mark_healthy(sc->mp, mask);
 		break;
 	case XHG_RTGROUP:
+		if (!mask)
+			return;
 		rtg = xfs_rtgroup_get(sc->mp, sc->sm->sm_agno);
 		if (bad)
-			xfs_group_mark_corrupt(rtg_group(rtg), sc->sick_mask);
+			xfs_group_mark_corrupt(rtg_group(rtg), mask);
 		else
-			xfs_group_mark_healthy(rtg_group(rtg), sc->sick_mask);
+			xfs_group_mark_healthy(rtg_group(rtg), mask);
 		xfs_rtgroup_put(rtg);
 		break;
 	default:

index a7fda3e2b01377..5dbbe93cb49bfa 100644
@@ -184,6 +184,12 @@ struct xfs_scrub {
 	 */
 	unsigned int			sick_mask;
 
+	/*
+	 * Clear these XFS_SICK_* flags but only if the scan is ok.  Useful for
+	 * removing ZAPPED flags after a repair.
+	 */
+	unsigned int			healthy_mask;
+
 	/* next time we want to cond_resched() */
 	struct xchk_relax		relax;
 

index d015a86ef460fb..953ce7be78dc2f 100644
@@ -36,6 +36,7 @@
 #include "scrub/tempfile.h"
 #include "scrub/tempexch.h"
 #include "scrub/reap.h"
+#include "scrub/health.h"
 
 /*
  * Symbolic Link Repair

@@ -233,7 +234,7 @@ xrep_symlink_salvage(
 	 * target zapped flag.
 	 */
 	if (buflen == 0) {
-		sc->sick_mask |= XFS_SICK_INO_SYMLINK_ZAPPED;
+		xchk_mark_healthy_if_clean(sc, XFS_SICK_INO_SYMLINK_ZAPPED);
 		sprintf(target_buf, DUMMY_TARGET);
 	}
 

index 4b7f7860e37ece..dc3802c7f678ce 100644
@@ -223,6 +223,7 @@ xrep_tempfile_adjust_directory_tree(
 	if (error)
 		goto out_ilock;
 
+	xfs_iflags_set(sc->tempip, XFS_IRECOVERY);
 	xfs_qm_dqdetach(sc->tempip);
 out_ilock:
 	xrep_tempfile_iunlock(sc);

@@ -246,6 +247,8 @@ xrep_tempfile_remove_metadir(
 
 	ASSERT(sc->tp == NULL);
 
+	xfs_iflags_clear(sc->tempip, XFS_IRECOVERY);
+
 	xfs_ilock(sc->tempip, XFS_IOLOCK_EXCL);
 	sc->temp_ilock_flags |= XFS_IOLOCK_EXCL;
 

@@ -945,10 +948,13 @@ xrep_is_tempfile(
 
 	/*
 	 * Files in the metadata directory tree also have S_PRIVATE set and
-	 * IOP_XATTR unset, so we must distinguish them separately.
+	 * IOP_XATTR unset, so we must distinguish them separately.  We (ab)use
+	 * the IRECOVERY flag to mark temporary metadir inodes knowing that the
+	 * end of log recovery clears IRECOVERY, so the only ones that can
+	 * exist during online repair are the ones we create.
 	 */
 	if (xfs_has_metadir(mp) && (ip->i_diflags2 & XFS_DIFLAG2_METADATA))
-		return false;
+		return __xfs_iflags_test(ip, XFS_IRECOVERY);
 
 	if (IS_PRIVATE(inode) && !(inode->i_opflags & IOP_XATTR))
 		return true;

index 2a4485fb990846..bd6b37beabacdd 100644
@@ -231,7 +231,7 @@ xfs_iflags_clear(xfs_inode_t *ip, unsigned long flags)
 }
 
 static inline int
-__xfs_iflags_test(xfs_inode_t *ip, unsigned long flags)
+__xfs_iflags_test(const struct xfs_inode *ip, unsigned long flags)
 {
 	return (ip->i_flags & flags);
 }

index 0cb534d71119a5..fcfa6e0eb3ad2a 100644
@@ -1827,7 +1827,7 @@ xfs_rtallocate_rtg(
 	 * For an allocation to an empty file at offset 0, pick an extent that
 	 * will space things out in the rt area.
 	 */
-	if (bno_hint)
+	if (bno_hint != NULLFSBLOCK)
 		start = xfs_rtb_to_rtx(args.mp, bno_hint);
 	else if (!xfs_has_rtgroups(args.mp) && initial_user_data)
 		start = xfs_rtpick_extent(args.rtg, tp, maxlen);

index e74bb059f24fa1..4f3bfc884aff29 100644
@@ -496,7 +496,7 @@ xfs_rtginode_create(
 
 	error = xfs_metadir_create(&upd, S_IFREG);
 	if (error)
-		return error;
+		goto out_cancel;
 
 	xfs_rtginode_lockdep_setup(upd.ip, rtg_rgno(rtg), type);
 

index c748866ef92368..68ee1c299c25fd 100644
@@ -3557,14 +3557,31 @@ xfs_btree_insrec(
 	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
 
 	/*
-	 * If we just inserted into a new tree block, we have to
-	 * recalculate nkey here because nkey is out of date.
+	 * Update btree keys to reflect the newly added record or keyptr.
+	 * There are three cases here to be aware of.  Normally, all we have to
+	 * do is walk towards the root, updating keys as necessary.
 	 *
-	 * Otherwise we're just updating an existing block (having shoved
-	 * some records into the new tree block), so use the regular key
-	 * update mechanism.
+	 * If the caller had us target a full block for the insertion, we dealt
+	 * with that by calling the _make_block_unfull function.  If the
+	 * "make unfull" function splits the block, it'll hand us back the key
+	 * and pointer of the new block.  We haven't yet added the new block to
+	 * the next level up, so if we decide to add the new record to the new
+	 * block (bp->b_bn != old_bn), we have to update the caller's pointer
+	 * so that the caller adds the new block with the correct key.
+	 *
+	 * However, there is a third possibility-- if the selected block is the
+	 * root block of an inode-rooted btree and cannot be expanded further,
+	 * the "make unfull" function moves the root block contents to a new
+	 * block and updates the root block to point to the new block.  In this
+	 * case, no block pointer is passed back because the block has already
+	 * been added to the btree.  In this case, we need to use the regular
+	 * key update function, just like the first case.  This is critical for
+	 * overlapping btrees, because the high key must be updated to reflect
+	 * the entire tree, not just the subtree accessible through the first
+	 * child of the root (which is now two levels down from the root).
 	 */
-	if (bp && xfs_buf_daddr(bp) != old_bn) {
+	if (!xfs_btree_ptr_is_null(cur, &nptr) &&
+	    bp && xfs_buf_daddr(bp) != old_bn) {
 		xfs_btree_get_keys(cur, block, lkey);
 	} else if (xfs_btree_needs_key_update(cur, optr)) {
 		error = xfs_btree_update_keys(cur, level);

index 9b38f5ad1eaf07..d2ae7e93acb08e 100644
@@ -605,7 +605,7 @@ TRACE_EVENT(xchk_ifork_btree_op_error,
 	TP_fast_assign(
 		xfs_fsblock_t fsbno = xchk_btree_cur_fsbno(cur, level);
 		__entry->dev = sc->mp->m_super->s_dev;
-		__entry->ino = sc->ip->i_ino;
+		__entry->ino = cur->bc_ino.ip->i_ino;
 		__entry->whichfork = cur->bc_ino.whichfork;
 		__entry->type = sc->sm->sm_type;
 		__assign_str(name);

index 30fbed27cf05cc..05b18e30368e4b 100644
@@ -1435,5 +1435,8 @@ xfs_trans_alloc_dir(
 
 out_cancel:
 	xfs_trans_cancel(tp);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	if (dp != ip)
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	return error;
 }

index 05b18e30368e4b..4a517250efc911 100644
@@ -860,13 +860,6 @@ __xfs_trans_commit(
 
 	trace_xfs_trans_commit(tp, _RET_IP_);
 
-	error = xfs_trans_run_precommits(tp);
-	if (error) {
-		if (tp->t_flags & XFS_TRANS_PERM_LOG_RES)
-			xfs_defer_cancel(tp);
-		goto out_unreserve;
-	}
-
 	/*
 	 * Finish deferred items on final commit. Only permanent transactions
 	 * should ever have deferred ops.

@@ -877,13 +870,12 @@ __xfs_trans_commit(
 		error = xfs_defer_finish_noroll(&tp);
 		if (error)
 			goto out_unreserve;
-
-		/* Run precommits from final tx in defer chain. */
-		error = xfs_trans_run_precommits(tp);
-		if (error)
-			goto out_unreserve;
 	}
 
+	error = xfs_trans_run_precommits(tp);
+	if (error)
+		goto out_unreserve;
+
 	/*
 	 * If there is nothing to be logged by the transaction,
 	 * then unlock all of the items associated with the

index 4a517250efc911..26bb2343082af4 100644
@@ -860,18 +860,6 @@ __xfs_trans_commit(
 
 	trace_xfs_trans_commit(tp, _RET_IP_);
 
-	/*
-	 * Finish deferred items on final commit. Only permanent transactions
-	 * should ever have deferred ops.
-	 */
-	WARN_ON_ONCE(!list_empty(&tp->t_dfops) &&
-		     !(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
-	if (!regrant && (tp->t_flags & XFS_TRANS_PERM_LOG_RES)) {
-		error = xfs_defer_finish_noroll(&tp);
-		if (error)
-			goto out_unreserve;
-	}
-
 	error = xfs_trans_run_precommits(tp);
 	if (error)
 		goto out_unreserve;

@@ -950,6 +938,20 @@ int
 xfs_trans_commit(
 	struct xfs_trans	*tp)
 {
+	/*
+	 * Finish deferred items on final commit. Only permanent transactions
+	 * should ever have deferred ops.
+	 */
+	WARN_ON_ONCE(!list_empty(&tp->t_dfops) &&
+		     !(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
+	if (tp->t_flags & XFS_TRANS_PERM_LOG_RES) {
+		int error = xfs_defer_finish_noroll(&tp);
+		if (error) {
+			xfs_trans_cancel(tp);
+			return error;
+		}
+	}
+
 	return __xfs_trans_commit(tp, false);
 }
 

index 26bb2343082af4..427a8ba0ab99e2 100644
@@ -860,6 +860,13 @@ __xfs_trans_commit(
 
 	trace_xfs_trans_commit(tp, _RET_IP_);
 
+	/*
+	 * Commit per-transaction changes that are not already tracked through
+	 * log items.  This can add dirty log items to the transaction.
+	 */
+	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
+		xfs_trans_apply_sb_deltas(tp);
+
 	error = xfs_trans_run_precommits(tp);
 	if (error)
 		goto out_unreserve;

@@ -890,8 +897,6 @@ __xfs_trans_commit(
 	/*
 	 * If we need to update the superblock, then do it now.
 	 */
-	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
-		xfs_trans_apply_sb_deltas(tp);
 	xfs_trans_apply_dquot_deltas(tp);
 
 	xlog_cil_commit(log, tp, &commit_seq, regrant);

index fa1317cc396c96..b864ed59787780 100644
@@ -101,7 +101,8 @@ extern void xfs_trans_free_dqinfo(struct xfs_trans *);
 extern void xfs_trans_mod_dquot_byino(struct xfs_trans *, struct xfs_inode *,
 		uint, int64_t);
 extern void xfs_trans_apply_dquot_deltas(struct xfs_trans *);
-extern void xfs_trans_unreserve_and_mod_dquots(struct xfs_trans *);
+void xfs_trans_unreserve_and_mod_dquots(struct xfs_trans *tp,
+		bool already_locked);
 int xfs_trans_reserve_quota_nblks(struct xfs_trans *tp, struct xfs_inode *ip,
 		int64_t dblocks, int64_t rblocks, bool force);
 extern int xfs_trans_reserve_quota_bydquots(struct xfs_trans *,

@@ -172,8 +173,8 @@ static inline void xfs_trans_mod_dquot_byino(struct xfs_trans *tp,
 		struct xfs_inode *ip, uint field, int64_t delta)
 {
 }
-#define xfs_trans_apply_dquot_deltas(tp)
-#define xfs_trans_unreserve_and_mod_dquots(tp)
+#define xfs_trans_apply_dquot_deltas(tp, a)
+#define xfs_trans_unreserve_and_mod_dquots(tp, a)
 static inline int xfs_trans_reserve_quota_nblks(struct xfs_trans *tp,
 		struct xfs_inode *ip, int64_t dblocks, int64_t rblocks,
 		bool force)

index 427a8ba0ab99e2..4cd25717c9d130 100644
@@ -866,6 +866,7 @@ __xfs_trans_commit(
 	 */
 	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
 		xfs_trans_apply_sb_deltas(tp);
+	xfs_trans_apply_dquot_deltas(tp);
 
 	error = xfs_trans_run_precommits(tp);
 	if (error)

@@ -894,11 +895,6 @@ __xfs_trans_commit(
 
 	ASSERT(tp->t_ticket != NULL);
 
-	/*
-	 * If we need to update the superblock, then do it now.
-	 */
-	xfs_trans_apply_dquot_deltas(tp);
-
 	xlog_cil_commit(log, tp, &commit_seq, regrant);
 
 	xfs_trans_free(tp);

@@ -924,7 +920,7 @@ __xfs_trans_commit(
 	 * the dqinfo portion to be.  All that means is that we have some
 	 * (non-persistent) quota reservations that need to be unreserved.
 	 */
-	xfs_trans_unreserve_and_mod_dquots(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp, true);
 	if (tp->t_ticket) {
 		if (regrant && !xlog_is_shutdown(log))
 			xfs_log_ticket_regrant(log, tp->t_ticket);

@@ -1018,7 +1014,7 @@ xfs_trans_cancel(
 	}
 #endif
 	xfs_trans_unreserve_and_mod_sb(tp);
-	xfs_trans_unreserve_and_mod_dquots(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp, false);
 
 	if (tp->t_ticket) {
 		xfs_log_ticket_ungrant(log, tp->t_ticket);

index 481ba3dc9f190d..713b6d243e5631 100644
@@ -606,6 +606,24 @@ xfs_trans_apply_dquot_deltas(
 			ASSERT(dqp->q_blk.reserved >= dqp->q_blk.count);
 			ASSERT(dqp->q_ino.reserved >= dqp->q_ino.count);
 			ASSERT(dqp->q_rtb.reserved >= dqp->q_rtb.count);
+
+			/*
+			 * We've applied the count changes and given back
+			 * whatever reservation we didn't use.  Zero out the
+			 * dqtrx fields.
+			 */
+			qtrx->qt_blk_res = 0;
+			qtrx->qt_bcount_delta = 0;
+			qtrx->qt_delbcnt_delta = 0;
+
+			qtrx->qt_rtblk_res = 0;
+			qtrx->qt_rtblk_res_used = 0;
+			qtrx->qt_rtbcount_delta = 0;
+			qtrx->qt_delrtb_delta = 0;
+
+			qtrx->qt_ino_res = 0;
+			qtrx->qt_ino_res_used = 0;
+			qtrx->qt_icount_delta = 0;
 		}
 	}
 }

@@ -642,7 +660,8 @@ xfs_trans_unreserve_and_mod_dquots_hook(
  */
 void
 xfs_trans_unreserve_and_mod_dquots(
-	struct xfs_trans	*tp)
+	struct xfs_trans	*tp,
+	bool			already_locked)
 {
 	int			i, j;
 	struct xfs_dquot	*dqp;

@@ -671,10 +690,12 @@ xfs_trans_unreserve_and_mod_dquots(
 			 * about the number of blocks used field, or deltas.
 			 * Also we don't bother to zero the fields.
 			 */
-			locked = false;
+			locked = already_locked;
 			if (qtrx->qt_blk_res) {
-				xfs_dqlock(dqp);
-				locked = true;
+				if (!locked) {
+					xfs_dqlock(dqp);
+					locked = true;
+				}
 				dqp->q_blk.reserved -=
 					(xfs_qcnt_t)qtrx->qt_blk_res;
 			}

@@ -695,7 +716,7 @@ xfs_trans_unreserve_and_mod_dquots(
 				dqp->q_rtb.reserved -=
 					(xfs_qcnt_t)qtrx->qt_rtblk_res;
 			}
-			if (locked)
+			if (locked && !already_locked)
 				xfs_dqunlock(dqp);
 
 		}

index ff982d983989b0..6ec4087e38dfc8 100644
@@ -1238,6 +1238,42 @@ xfs_qm_dqflush_check(
 	return NULL;
 }
 
+/*
+ * Get the buffer containing the on-disk dquot.
+ *
+ * Requires dquot flush lock, will clear the dirty flag, delete the quota log
+ * item from the AIL, and shut down the system if something goes wrong.
+ */
+int
+xfs_dquot_read_buf(
+	struct xfs_trans	*tp,
+	struct xfs_dquot	*dqp,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = dqp->q_mount;
+	struct xfs_buf		*bp = NULL;
+	int			error;
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, dqp->q_blkno,
+				   mp->m_quotainfo->qi_dqchunklen, XBF_TRYLOCK,
+				   &bp, &xfs_dquot_buf_ops);
+	if (error == -EAGAIN)
+		return error;
+	if (xfs_metadata_is_sick(error))
+		xfs_dquot_mark_sick(dqp);
+	if (error)
+		goto out_abort;
+
+	*bpp = bp;
+	return 0;
+
+out_abort:
+	dqp->q_flags &= ~XFS_DQFLAG_DIRTY;
+	xfs_trans_ail_delete(&dqp->q_logitem.qli_item, 0);
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+	return error;
+}
+
 /*
  * Write a modified dquot to disk.
  * The dquot must be locked and the flush lock too taken by caller.

@@ -1249,11 +1285,10 @@ xfs_qm_dqflush_check(
 int
 xfs_qm_dqflush(
 	struct xfs_dquot	*dqp,
-	struct xfs_buf		**bpp)
+	struct xfs_buf		*bp)
 {
 	struct xfs_mount	*mp = dqp->q_mount;
 	struct xfs_log_item	*lip = &dqp->q_logitem.qli_item;
-	struct xfs_buf		*bp;
 	struct xfs_dqblk	*dqblk;
 	xfs_failaddr_t		fa;
 	int			error;

@@ -1263,28 +1298,12 @@ xfs_qm_dqflush(
 
 	trace_xfs_dqflush(dqp);
 
-	*bpp = NULL;
-
 	xfs_qm_dqunpin_wait(dqp);
 
-	/*
-	 * Get the buffer containing the on-disk dquot
-	 */
-	error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp, dqp->q_blkno,
-				   mp->m_quotainfo->qi_dqchunklen, XBF_TRYLOCK,
-				   &bp, &xfs_dquot_buf_ops);
-	if (error == -EAGAIN)
-		goto out_unlock;
-	if (xfs_metadata_is_sick(error))
-		xfs_dquot_mark_sick(dqp);
-	if (error)
-		goto out_abort;
-
 	fa = xfs_qm_dqflush_check(dqp);
 	if (fa) {
 		xfs_alert(mp, "corrupt dquot ID 0x%x in memory at %pS",
 				dqp->q_id, fa);
-		xfs_buf_relse(bp);
 		xfs_dquot_mark_sick(dqp);
 		error = -EFSCORRUPTED;
 		goto out_abort;

@@ -1334,14 +1353,12 @@ xfs_qm_dqflush(
 	}
 
 	trace_xfs_dqflush_done(dqp);
-	*bpp = bp;
 	return 0;
 
 out_abort:
 	dqp->q_flags &= ~XFS_DQFLAG_DIRTY;
 	xfs_trans_ail_delete(lip, 0);
 	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
-out_unlock:
 	xfs_dqfunlock(dqp);
 	return error;
 }

index d73d179df00958..50f8404c41176c 100644
@@ -214,7 +214,9 @@ void xfs_dquot_to_disk(struct xfs_disk_dquot *ddqp, struct xfs_dquot *dqp);
 #define XFS_DQ_IS_DIRTY(dqp)	((dqp)->q_flags & XFS_DQFLAG_DIRTY)
 
 void		xfs_qm_dqdestroy(struct xfs_dquot *dqp);
-int		xfs_qm_dqflush(struct xfs_dquot *dqp, struct xfs_buf **bpp);
+int		xfs_dquot_read_buf(struct xfs_trans *tp, struct xfs_dquot *dqp,
+				struct xfs_buf **bpp);
+int		xfs_qm_dqflush(struct xfs_dquot *dqp, struct xfs_buf *bp);
 void		xfs_qm_dqunpin_wait(struct xfs_dquot *dqp);
 void		xfs_qm_adjust_dqtimers(struct xfs_dquot *d);
 void		xfs_qm_adjust_dqlimits(struct xfs_dquot *d);

index 7d19091215b080..56ecc5ed01934d 100644
@@ -155,14 +155,26 @@ xfs_qm_dquot_logitem_push(
 
 	spin_unlock(&lip->li_ailp->ail_lock);
 
-	error = xfs_qm_dqflush(dqp, &bp);
+	error = xfs_dquot_read_buf(NULL, dqp, &bp);
+	if (error) {
+		if (error == -EAGAIN)
+			rval = XFS_ITEM_LOCKED;
+		xfs_dqfunlock(dqp);
+		goto out_relock_ail;
+	}
+
+	/*
+	 * dqflush completes dqflock on error, and the delwri ioend does it on
+	 * success.
+	 */
+	error = xfs_qm_dqflush(dqp, bp);
 	if (!error) {
 		if (!xfs_buf_delwri_queue(bp, buffer_list))
 			rval = XFS_ITEM_FLUSHING;
-		xfs_buf_relse(bp);
-	} else if (error == -EAGAIN)
-		rval = XFS_ITEM_LOCKED;
+	}
+	xfs_buf_relse(bp);
 
+out_relock_ail:
 	spin_lock(&lip->li_ailp->ail_lock);
 out_unlock:
 	xfs_dqunlock(dqp);

index a4fa21dfd6b4ad..341fe4821c2d77 100644
@@ -148,17 +148,28 @@ xfs_qm_dqpurge(
 		 * We don't care about getting disk errors here. We need
 		 * to purge this dquot anyway, so we go ahead regardless.
 		 */
-		error = xfs_qm_dqflush(dqp, &bp);
+		error = xfs_dquot_read_buf(NULL, dqp, &bp);
+		if (error == -EAGAIN) {
+			xfs_dqfunlock(dqp);
+			dqp->q_flags &= ~XFS_DQFLAG_FREEING;
+			goto out_unlock;
+		}
+		if (error)
+			goto out_funlock;
+
+		/*
+		 * dqflush completes dqflock on error, and the bwrite ioend
+		 * does it on success.
+		 */
+		error = xfs_qm_dqflush(dqp, bp);
 		if (!error) {
 			error = xfs_bwrite(bp);
 			xfs_buf_relse(bp);
-		} else if (error == -EAGAIN) {
-			dqp->q_flags &= ~XFS_DQFLAG_FREEING;
-			goto out_unlock;
 		}
 		xfs_dqflock(dqp);
 	}
 
+out_funlock:
 	ASSERT(atomic_read(&dqp->q_pincount) == 0);
 	ASSERT(xlog_is_shutdown(dqp->q_logitem.qli_item.li_log) ||
 		!test_bit(XFS_LI_IN_AIL, &dqp->q_logitem.qli_item.li_flags));

@@ -495,7 +506,17 @@ xfs_qm_dquot_isolate(
 		/* we have to drop the LRU lock to flush the dquot */
 		spin_unlock(lru_lock);
 
-		error = xfs_qm_dqflush(dqp, &bp);
+		error = xfs_dquot_read_buf(NULL, dqp, &bp);
+		if (error) {
+			xfs_dqfunlock(dqp);
+			goto out_unlock_dirty;
+		}
+
+		/*
+		 * dqflush completes dqflock on error, and the delwri ioend
+		 * does it on success.
+		 */
+		error = xfs_qm_dqflush(dqp, bp);
 		if (error)
 			goto out_unlock_dirty;
 

@@ -1491,11 +1512,13 @@ xfs_qm_flush_one(
 		goto out_unlock;
 	}
 
-	error = xfs_qm_dqflush(dqp, &bp);
+	error = xfs_dquot_read_buf(NULL, dqp, &bp);
 	if (error)
 		goto out_unlock;
 
-	xfs_buf_delwri_queue(bp, buffer_list);
+	error = xfs_qm_dqflush(dqp, bp);
+	if (!error)
+		xfs_buf_delwri_queue(bp, buffer_list);
 	xfs_buf_relse(bp);
 out_unlock:
 	xfs_dqunlock(dqp);

index 6ec4087e38dfc8..4ba042786cfb7b 100644
@@ -1142,8 +1142,9 @@ static void
 xfs_qm_dqflush_done(
 	struct xfs_log_item	*lip)
 {
-	struct xfs_dq_logitem	*qip = (struct xfs_dq_logitem *)lip;
-	struct xfs_dquot	*dqp = qip->qli_dquot;
+	struct xfs_dq_logitem	*qlip =
+			container_of(lip, struct xfs_dq_logitem, qli_item);
+	struct xfs_dquot	*dqp = qlip->qli_dquot;
 	struct xfs_ail		*ailp = lip->li_ailp;
 	xfs_lsn_t		tail_lsn;
 

@@ -1156,12 +1157,12 @@ xfs_qm_dqflush_done(
 	 * holding the lock before removing the dquot from the AIL.
 	 */
 	if (test_bit(XFS_LI_IN_AIL, &lip->li_flags) &&
-	    ((lip->li_lsn == qip->qli_flush_lsn) ||
+	    ((lip->li_lsn == qlip->qli_flush_lsn) ||
 	     test_bit(XFS_LI_FAILED, &lip->li_flags))) {
 
 		spin_lock(&ailp->ail_lock);
 		xfs_clear_li_failed(lip);
-		if (lip->li_lsn == qip->qli_flush_lsn) {
+		if (lip->li_lsn == qlip->qli_flush_lsn) {
 			/* xfs_ail_update_finish() drops the AIL lock */
 			tail_lsn = xfs_ail_delete_one(ailp, lip);
 			xfs_ail_update_finish(ailp, tail_lsn);

@@ -1319,7 +1320,7 @@ xfs_qm_dqflush(
 	dqp->q_flags &= ~XFS_DQFLAG_DIRTY;
 
 	xfs_trans_ail_copy_lsn(mp->m_ail, &dqp->q_logitem.qli_flush_lsn,
-					&dqp->q_logitem.qli_item.li_lsn);
+			&lip->li_lsn);
 
 	/*
 	 * copy the lsn into the on-disk dquot now while we have the in memory

@@ -1331,7 +1332,7 @@ xfs_qm_dqflush(
 	 * of a dquot without an up-to-date CRC getting to disk.
 	 */
 	if (xfs_has_crc(mp)) {
-		dqblk->dd_lsn = cpu_to_be64(dqp->q_logitem.qli_item.li_lsn);
+		dqblk->dd_lsn = cpu_to_be64(lip->li_lsn);
 		xfs_update_cksum((char *)dqblk, sizeof(struct xfs_dqblk),
 				 XFS_DQUOT_CRC_OFF);
 	}

@@ -1341,7 +1342,7 @@ xfs_qm_dqflush(
 	 * the AIL and release the flush lock once the dquot is synced to disk.
 	 */
 	bp->b_flags |= _XBF_DQUOTS;
-	list_add_tail(&dqp->q_logitem.qli_item.li_bio_list, &bp->b_li_list);
+	list_add_tail(&lip->li_bio_list, &bp->b_li_list);
 
 	/*
 	 * If the buffer is pinned then push on the log so we won't

index 4ba042786cfb7b..c495f7ad80018f 100644
@@ -75,8 +75,24 @@ void
 xfs_qm_dqdestroy(
 	struct xfs_dquot	*dqp)
 {
+	struct xfs_dq_logitem	*qlip = &dqp->q_logitem;
+	struct xfs_buf		*bp = NULL;
+
 	ASSERT(list_empty(&dqp->q_lru));
 
+	/*
+	 * Detach the dquot buffer if it's still attached, because we can get
+	 * called through dqpurge after a log shutdown.
+	 */
+	spin_lock(&qlip->qli_lock);
+	if (qlip->qli_item.li_buf) {
+		bp = qlip->qli_item.li_buf;
+		qlip->qli_item.li_buf = NULL;
+	}
+	spin_unlock(&qlip->qli_lock);
+	if (bp)
+		xfs_buf_rele(bp);
+
 	kvfree(dqp->q_logitem.qli_item.li_lv_shadow);
 	mutex_destroy(&dqp->q_qlock);
 

@@ -1146,6 +1162,7 @@ xfs_qm_dqflush_done(
 			container_of(lip, struct xfs_dq_logitem, qli_item);
 	struct xfs_dquot	*dqp = qlip->qli_dquot;
 	struct xfs_ail		*ailp = lip->li_ailp;
+	struct xfs_buf		*bp = NULL;
 	xfs_lsn_t		tail_lsn;
 
 	/*

@@ -1175,6 +1192,19 @@ xfs_qm_dqflush_done(
 	 * Release the dq's flush lock since we're done with it.
 	 */
 	xfs_dqfunlock(dqp);
+
+	/*
+	 * If this dquot hasn't been dirtied since initiating the last dqflush,
+	 * release the buffer reference.
+	 */
+	spin_lock(&qlip->qli_lock);
+	if (!qlip->qli_dirty) {
+		bp = lip->li_buf;
+		lip->li_buf = NULL;
+	}
+	spin_unlock(&qlip->qli_lock);
+	if (bp)
+		xfs_buf_rele(bp);
 }
 
 void

@@ -1197,7 +1227,7 @@ xfs_buf_dquot_io_fail(
 
 	spin_lock(&bp->b_mount->m_ail->ail_lock);
 	list_for_each_entry(lip, &bp->b_li_list, li_bio_list)
-		xfs_set_li_failed(lip, bp);
+		set_bit(XFS_LI_FAILED, &lip->li_flags);
 	spin_unlock(&bp->b_mount->m_ail->ail_lock);
 }
 

@@ -1249,6 +1279,7 @@ int
 xfs_dquot_read_buf(
 	struct xfs_trans	*tp,
 	struct xfs_dquot	*dqp,
+	xfs_buf_flags_t		xbf_flags,
 	struct xfs_buf		**bpp)
 {
 	struct xfs_mount	*mp = dqp->q_mount;

@@ -1256,7 +1287,7 @@ xfs_dquot_read_buf(
 	int			error;
 
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, dqp->q_blkno,
-				   mp->m_quotainfo->qi_dqchunklen, XBF_TRYLOCK,
+				   mp->m_quotainfo->qi_dqchunklen, xbf_flags,
 				   &bp, &xfs_dquot_buf_ops);
 	if (error == -EAGAIN)
 		return error;

@@ -1275,6 +1306,77 @@ xfs_dquot_read_buf(
 	return error;
 }
 
+/*
+ * Attach a dquot buffer to this dquot to avoid allocating a buffer during a
+ * dqflush, since dqflush can be called from reclaim context.
+ */
+int
+xfs_dquot_attach_buf(
+	struct xfs_trans	*tp,
+	struct xfs_dquot	*dqp)
+{
+	struct xfs_dq_logitem	*qlip = &dqp->q_logitem;
+	struct xfs_log_item	*lip = &qlip->qli_item;
+	int			error;
+
+	spin_lock(&qlip->qli_lock);
+	if (!lip->li_buf) {
+		struct xfs_buf	*bp = NULL;
+
+		spin_unlock(&qlip->qli_lock);
+		error = xfs_dquot_read_buf(tp, dqp, 0, &bp);
+		if (error)
+			return error;
+
+		/*
+		 * Attach the dquot to the buffer so that the AIL does not have
+		 * to read the dquot buffer to push this item.
+		 */
+		xfs_buf_hold(bp);
+		spin_lock(&qlip->qli_lock);
+		lip->li_buf = bp;
+		xfs_trans_brelse(tp, bp);
+	}
+	qlip->qli_dirty = true;
+	spin_unlock(&qlip->qli_lock);
+
+	return 0;
+}
+
+/*
+ * Get a new reference the dquot buffer attached to this dquot for a dqflush
+ * operation.
+ *
+ * Returns 0 and a NULL bp if none was attached to the dquot; 0 and a locked
+ * bp; or -EAGAIN if the buffer could not be locked.
+ */
+int
+xfs_dquot_use_attached_buf(
+	struct xfs_dquot	*dqp,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_buf		*bp = dqp->q_logitem.qli_item.li_buf;
+
+	/*
+	 * A NULL buffer can happen if the dquot dirty flag was set but the
+	 * filesystem shut down before transaction commit happened.  In that
+	 * case we're not going to flush anyway.
+	 */
+	if (!bp) {
+		ASSERT(xfs_is_shutdown(dqp->q_mount));
+
+		*bpp = NULL;
+		return 0;
+	}
+
+	if (!xfs_buf_trylock(bp))
+		return -EAGAIN;
+
+	xfs_buf_hold(bp);
+	*bpp = bp;
+	return 0;
+}
+
 /*
  * Write a modified dquot to disk.
  * The dquot must be locked and the flush lock too taken by caller.

@@ -1289,7 +1391,8 @@ xfs_qm_dqflush(
 	struct xfs_buf		*bp)
 {
 	struct xfs_mount	*mp = dqp->q_mount;
-	struct xfs_log_item	*lip = &dqp->q_logitem.qli_item;
+	struct xfs_dq_logitem	*qlip = &dqp->q_logitem;
+	struct xfs_log_item	*lip = &qlip->qli_item;
 	struct xfs_dqblk	*dqblk;
 	xfs_failaddr_t		fa;
 	int			error;

@@ -1319,8 +1422,15 @@ xfs_qm_dqflush(
 	 */
 	dqp->q_flags &= ~XFS_DQFLAG_DIRTY;
 
-	xfs_trans_ail_copy_lsn(mp->m_ail, &dqp->q_logitem.qli_flush_lsn,
-			&lip->li_lsn);
+	/*
+	 * We hold the dquot lock, so nobody can dirty it while we're
+	 * scheduling the write out.  Clear the dirty-since-flush flag.
+	 */
+	spin_lock(&qlip->qli_lock);
+	qlip->qli_dirty = false;
+	spin_unlock(&qlip->qli_lock);
+
+	xfs_trans_ail_copy_lsn(mp->m_ail, &qlip->qli_flush_lsn, &lip->li_lsn);
 
 	/*
 	 * copy the lsn into the on-disk dquot now while we have the in memory

index 50f8404c41176c..362ca34f7c248b 100644
@@ -215,7 +215,7 @@ void xfs_dquot_to_disk(struct xfs_disk_dquot *ddqp, struct xfs_dquot *dqp);
 
 void		xfs_qm_dqdestroy(struct xfs_dquot *dqp);
 int		xfs_dquot_read_buf(struct xfs_trans *tp, struct xfs_dquot *dqp,
-				struct xfs_buf **bpp);
+				xfs_buf_flags_t flags, struct xfs_buf **bpp);
 int		xfs_qm_dqflush(struct xfs_dquot *dqp, struct xfs_buf *bp);
 void		xfs_qm_dqunpin_wait(struct xfs_dquot *dqp);
 void		xfs_qm_adjust_dqtimers(struct xfs_dquot *d);

@@ -239,6 +239,9 @@ void		xfs_dqlockn(struct xfs_dqtrx *q);
 
 void		xfs_dquot_set_prealloc_limits(struct xfs_dquot *);
 
+int xfs_dquot_attach_buf(struct xfs_trans *tp, struct xfs_dquot *dqp);
+int xfs_dquot_use_attached_buf(struct xfs_dquot *dqp, struct xfs_buf **bpp);
+
 static inline struct xfs_dquot *xfs_qm_dqhold(struct xfs_dquot *dqp)
 {
 	xfs_dqlock(dqp);

index 56ecc5ed01934d..271b195ebb9326 100644
@@ -123,8 +123,9 @@ xfs_qm_dquot_logitem_push(
 		__releases(&lip->li_ailp->ail_lock)
 		__acquires(&lip->li_ailp->ail_lock)
 {
-	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
-	struct xfs_buf		*bp = lip->li_buf;
+	struct xfs_dq_logitem	*qlip = DQUOT_ITEM(lip);
+	struct xfs_dquot	*dqp = qlip->qli_dquot;
+	struct xfs_buf		*bp;
 	uint			rval = XFS_ITEM_SUCCESS;
 	int			error;
 

@@ -155,11 +156,10 @@ xfs_qm_dquot_logitem_push(
 
 	spin_unlock(&lip->li_ailp->ail_lock);
 
-	error = xfs_dquot_read_buf(NULL, dqp, &bp);
-	if (error) {
-		if (error == -EAGAIN)
-			rval = XFS_ITEM_LOCKED;
+	error = xfs_dquot_use_attached_buf(dqp, &bp);
+	if (error == -EAGAIN) {
 		xfs_dqfunlock(dqp);
+		rval = XFS_ITEM_LOCKED;
 		goto out_relock_ail;
 	}
 

@@ -207,12 +207,10 @@ xfs_qm_dquot_logitem_committing(
 }
 
 #ifdef DEBUG_EXPENSIVE
-static int
-xfs_qm_dquot_logitem_precommit(
-	struct xfs_trans	*tp,
-	struct xfs_log_item	*lip)
+static void
+xfs_qm_dquot_logitem_precommit_check(
+	struct xfs_dquot	*dqp)
 {
-	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
 	struct xfs_mount	*mp = dqp->q_mount;
 	struct xfs_disk_dquot	ddq = { };
 	xfs_failaddr_t		fa;

@@ -228,13 +226,24 @@ xfs_qm_dquot_logitem_precommit(
 		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
 		ASSERT(fa == NULL);
 	}
-
-	return 0;
 }
 #else
-# define xfs_qm_dquot_logitem_precommit	NULL
+# define xfs_qm_dquot_logitem_precommit_check(...)	((void)0)
 #endif
 
+static int
+xfs_qm_dquot_logitem_precommit(
+	struct xfs_trans	*tp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dq_logitem	*qlip = DQUOT_ITEM(lip);
+	struct xfs_dquot	*dqp = qlip->qli_dquot;
+
+	xfs_qm_dquot_logitem_precommit_check(dqp);
+
+	return xfs_dquot_attach_buf(tp, dqp);
+}
+
 static const struct xfs_item_ops xfs_dquot_item_ops = {
 	.iop_size	= xfs_qm_dquot_logitem_size,
 	.iop_precommit	= xfs_qm_dquot_logitem_precommit,

@@ -259,5 +268,7 @@ xfs_qm_dquot_logitem_init(
 
 	xfs_log_item_init(dqp->q_mount, &lp->qli_item, XFS_LI_DQUOT,
 					&xfs_dquot_item_ops);
+	spin_lock_init(&lp->qli_lock);
 	lp->qli_dquot = dqp;
+	lp->qli_dirty = false;
 }

index 794710c2447493..d66e52807d76d5 100644
@@ -14,6 +14,13 @@ struct xfs_dq_logitem {
 	struct xfs_log_item	qli_item;	/* common portion */
 	struct xfs_dquot	*qli_dquot;	/* dquot ptr */
 	xfs_lsn_t		qli_flush_lsn;	/* lsn at last flush */
+
+	/*
+	 * We use this spinlock to coordinate access to the li_buf pointer in
+	 * the log item and the qli_dirty flag.
+	 */
+	spinlock_t		qli_lock;
+	bool			qli_dirty;	/* dirtied since last flush? */
 };
 
 void xfs_qm_dquot_logitem_init(struct xfs_dquot *dqp);

index 341fe4821c2d77..a79c4a1bf27fab 100644
@@ -148,7 +148,7 @@ xfs_qm_dqpurge(
 		 * We don't care about getting disk errors here. We need
 		 * to purge this dquot anyway, so we go ahead regardless.
 		 */
-		error = xfs_dquot_read_buf(NULL, dqp, &bp);
+		error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
 		if (error == -EAGAIN) {
 			xfs_dqfunlock(dqp);
 			dqp->q_flags &= ~XFS_DQFLAG_FREEING;

@@ -506,7 +506,7 @@ xfs_qm_dquot_isolate(
 		/* we have to drop the LRU lock to flush the dquot */
 		spin_unlock(lru_lock);
 
-		error = xfs_dquot_read_buf(NULL, dqp, &bp);
+		error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
 		if (error) {
 			xfs_dqfunlock(dqp);
 			goto out_unlock_dirty;

@@ -1512,7 +1512,7 @@ xfs_qm_flush_one(
 		goto out_unlock;
 	}
 
-	error = xfs_dquot_read_buf(NULL, dqp, &bp);
+	error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
 	if (error)
 		goto out_unlock;
 

index 8ede9d099d1fea..f56d62dced97b1 100644
@@ -360,7 +360,7 @@ xfsaild_resubmit_item(
 
 	/* protected by ail_lock */
 	list_for_each_entry(lip, &bp->b_li_list, li_bio_list) {
-		if (bp->b_flags & _XBF_INODES)
+		if (bp->b_flags & (_XBF_INODES | _XBF_DQUOTS))
 			clear_bit(XFS_LI_FAILED, &lip->li_flags);
 		else
 			xfs_clear_li_failed(lip);

index c495f7ad80018f..c47f95c96fe0cf 100644
@@ -1275,11 +1275,10 @@ xfs_qm_dqflush_check(
  * Requires dquot flush lock, will clear the dirty flag, delete the quota log
  * item from the AIL, and shut down the system if something goes wrong.
  */
-int
+static int
 xfs_dquot_read_buf(
 	struct xfs_trans	*tp,
 	struct xfs_dquot	*dqp,
-	xfs_buf_flags_t		xbf_flags,
 	struct xfs_buf		**bpp)
 {
 	struct xfs_mount	*mp = dqp->q_mount;

@@ -1287,10 +1286,8 @@ xfs_dquot_read_buf(
 	int			error;
 
 	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, dqp->q_blkno,
-				   mp->m_quotainfo->qi_dqchunklen, xbf_flags,
+				   mp->m_quotainfo->qi_dqchunklen, 0,
 				   &bp, &xfs_dquot_buf_ops);
-	if (error == -EAGAIN)
-		return error;
 	if (xfs_metadata_is_sick(error))
 		xfs_dquot_mark_sick(dqp);
 	if (error)

@@ -1324,7 +1321,7 @@ xfs_dquot_attach_buf(
 		struct xfs_buf	*bp = NULL;
 
 		spin_unlock(&qlip->qli_lock);
-		error = xfs_dquot_read_buf(tp, dqp, 0, &bp);
+		error = xfs_dquot_read_buf(tp, dqp, &bp);
 		if (error)
 			return error;
 

index 362ca34f7c248b..1c5c911615bf7f 100644
@@ -214,8 +214,6 @@ void xfs_dquot_to_disk(struct xfs_disk_dquot *ddqp, struct xfs_dquot *dqp);
 #define XFS_DQ_IS_DIRTY(dqp)	((dqp)->q_flags & XFS_DQFLAG_DIRTY)
 
 void		xfs_qm_dqdestroy(struct xfs_dquot *dqp);
-int		xfs_dquot_read_buf(struct xfs_trans *tp, struct xfs_dquot *dqp,
-				xfs_buf_flags_t flags, struct xfs_buf **bpp);
 int		xfs_qm_dqflush(struct xfs_dquot *dqp, struct xfs_buf *bp);
 void		xfs_qm_dqunpin_wait(struct xfs_dquot *dqp);
 void		xfs_qm_adjust_dqtimers(struct xfs_dquot *d);

index a79c4a1bf27fab..e073ad51af1a3d 100644
@@ -148,13 +148,13 @@ xfs_qm_dqpurge(
 		 * We don't care about getting disk errors here. We need
 		 * to purge this dquot anyway, so we go ahead regardless.
 		 */
-		error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
+		error = xfs_dquot_use_attached_buf(dqp, &bp);
 		if (error == -EAGAIN) {
 			xfs_dqfunlock(dqp);
 			dqp->q_flags &= ~XFS_DQFLAG_FREEING;
 			goto out_unlock;
 		}
-		if (error)
+		if (!bp)
 			goto out_funlock;
 
 		/*

@@ -506,8 +506,8 @@ xfs_qm_dquot_isolate(
 		/* we have to drop the LRU lock to flush the dquot */
 		spin_unlock(lru_lock);
 
-		error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
-		if (error) {
+		error = xfs_dquot_use_attached_buf(dqp, &bp);
+		if (!bp || error == -EAGAIN) {
 			xfs_dqfunlock(dqp);
 			goto out_unlock_dirty;
 		}

@@ -1330,6 +1330,10 @@ xfs_qm_quotacheck_dqadjust(
 		return error;
 	}
 
+	error = xfs_dquot_attach_buf(NULL, dqp);
+	if (error)
+		return error;
+
 	trace_xfs_dqadjust(dqp);
 
 	/*

@@ -1512,9 +1516,13 @@ xfs_qm_flush_one(
 		goto out_unlock;
 	}
 
-	error = xfs_dquot_read_buf(NULL, dqp, XBF_TRYLOCK, &bp);
+	error = xfs_dquot_use_attached_buf(dqp, &bp);
 	if (error)
 		goto out_unlock;
+	if (!bp) {
+		error = -EFSCORRUPTED;
+		goto out_unlock;
+	}
 
 	error = xfs_qm_dqflush(dqp, bp);
 	if (!error)

index a809513a290cf4..3b5623611eba02 100644
@@ -494,12 +494,13 @@ xfs_validate_sb_common(
 				return -EINVAL;
 			}
 
-			if (!sbp->sb_spino_align ||
-			    sbp->sb_spino_align > sbp->sb_inoalignmt ||
-			    (sbp->sb_inoalignmt % sbp->sb_spino_align) != 0) {
+			if (sbp->sb_spino_align &&
+			    (sbp->sb_spino_align > sbp->sb_inoalignmt ||
+			     (sbp->sb_inoalignmt % sbp->sb_spino_align) != 0)) {
 				xfs_warn(mp,
-				"Sparse inode alignment (%u) is invalid.",
-					sbp->sb_spino_align);
+"Sparse inode alignment (%u) is invalid, must be integer factor of (%u).",
+					sbp->sb_spino_align,
+					sbp->sb_inoalignmt);
 				return -EINVAL;
 			}
 		} else if (sbp->sb_spino_align) {