Merge tag 'xfs-5.14-merge-6' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux

Pull xfs updates from Darrick Wong:
 "Most of the work this cycle has been on refactoring various parts of
  the codebase. The biggest non-cleanup changes are (1) reducing the
  number of cache flushes sent when writing the log; (2) a substantial
  number of log recovery fixes; and (3) I started accepting pull
  requests from contributors if the commits in their branches match
  what's been sent to the list.

  For a week or so I /had/ staged a major cleanup of the logging code
  from Dave Chinner, but it exposed so many lurking bugs in other parts
  of the logging and log recovery code that I decided to defer that
  patchset until we can address those latent bugs.

  Larger cleanups this time include walking the incore inode cache (me)
  and rework of the extended attribute code (Allison) to prepare it for
  adding logged xattr updates (and directory tree parent pointers) in
  future releases.

  Summary:

   - Refactor the buffer cache to use bulk page allocation

   - Convert agnumber-based AG iteration to walk per-AG structures

   - Clean up some unit conversions and other code warts

   - Reduce spinlock contention in the directio fastpath

   - Collapse all the inode cache walks into a single function

   - Remove indirect function calls from the inode cache walk code

   - Dramatically reduce the number of cache flushes sent when writing
     log buffers

   - Preserve inode sickness reports for longer

   - Rename xfs_eofblocks since it controls inode cache walks

   - Refactor the extended attribute code to prepare it for the addition
     of log intent items to make xattrs fully transactional

   - A few fixes to earlier large patchsets

   - Log recovery fixes so that we don't accidentally mark the log clean
     when log intent recovery fails

   - Fix some latent SOB errors

   - Clean up shutdown messages that get logged to dmesg

   - Fix a regression in the online shrink code

   - Fix a UAF in the buffer logging code if the fs goes offline

   - Fix uninitialized error variables

   - Fix a UAF in the CIL when commited log item callbacks race with a
     shutdown

   - Fix a bug where the CIL could hang trying to push part of the log
     ring buffer that hasn't been filled yet"

* tag 'xfs-5.14-merge-6' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux: (102 commits)
  xfs: don't wait on future iclogs when pushing the CIL
  xfs: Fix a CIL UAF by getting get rid of the iclog callback lock
  xfs: remove callback dequeue loop from xlog_state_do_iclog_callbacks
  xfs: don't nest icloglock inside ic_callback_lock
  xfs: Initialize error in xfs_attr_remove_iter
  xfs: fix endianness issue in xfs_ag_shrink_space
  xfs: remove dead stale buf unpin handling code
  xfs: hold buffer across unpin and potential shutdown processing
  xfs: force the log offline when log intent item recovery fails
  xfs: fix log intent recovery ENOSPC shutdowns when inactivating inodes
  xfs: shorten the shutdown messages to a single line
  xfs: print name of function causing fs shutdown instead of hex pointer
  xfs: fix type mismatches in the inode reclaim functions
  xfs: separate primary inode selection criteria in xfs_iget_cache_hit
  xfs: refactor the inode recycling code
  xfs: add iclog state trace events
  xfs: xfs_log_force_lsn isn't passed a LSN
  xfs: Fix CIL throttle hang when CIL space used going backwards
  xfs: journal IO cache flush reductions
  xfs: remove need_start_rec parameter from xlog_write()
  ...

92 files changed, 3827 insertions, 3036 deletions

diff --git a/fs/xfs/libxfs/xfs_ag.c b/fs/xfs/libxfs/xfs_ag.c
index c68a36688474..778ec52cce70 100644
--- a/fs/xfs/libxfs/xfs_ag.c
+++ b/fs/xfs/libxfs/xfs_ag.c
@@ -27,6 +27,276 @@
 #include "xfs_defer.h"
 #include "xfs_log_format.h"
 #include "xfs_trans.h"
+#include "xfs_trace.h"
+#include "xfs_inode.h"
+#include "xfs_icache.h"
+
+
+/*
+ * Passive reference counting access wrappers to the perag structures.  If the
+ * per-ag structure is to be freed, the freeing code is responsible for cleaning
+ * up objects with passive references before freeing the structure. This is
+ * things like cached buffers.
+ */
+struct xfs_perag *
+xfs_perag_get(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno)
+{
+	struct xfs_perag	*pag;
+	int			ref = 0;
+
+	rcu_read_lock();
+	pag = radix_tree_lookup(&mp->m_perag_tree, agno);
+	if (pag) {
+		ASSERT(atomic_read(&pag->pag_ref) >= 0);
+		ref = atomic_inc_return(&pag->pag_ref);
+	}
+	rcu_read_unlock();
+	trace_xfs_perag_get(mp, agno, ref, _RET_IP_);
+	return pag;
+}
+
+/*
+ * search from @first to find the next perag with the given tag set.
+ */
+struct xfs_perag *
+xfs_perag_get_tag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		first,
+	unsigned int		tag)
+{
+	struct xfs_perag	*pag;
+	int			found;
+	int			ref;
+
+	rcu_read_lock();
+	found = radix_tree_gang_lookup_tag(&mp->m_perag_tree,
+					(void **)&pag, first, 1, tag);
+	if (found <= 0) {
+		rcu_read_unlock();
+		return NULL;
+	}
+	ref = atomic_inc_return(&pag->pag_ref);
+	rcu_read_unlock();
+	trace_xfs_perag_get_tag(mp, pag->pag_agno, ref, _RET_IP_);
+	return pag;
+}
+
+void
+xfs_perag_put(
+	struct xfs_perag	*pag)
+{
+	int	ref;
+
+	ASSERT(atomic_read(&pag->pag_ref) > 0);
+	ref = atomic_dec_return(&pag->pag_ref);
+	trace_xfs_perag_put(pag->pag_mount, pag->pag_agno, ref, _RET_IP_);
+}
+
+/*
+ * xfs_initialize_perag_data
+ *
+ * Read in each per-ag structure so we can count up the number of
+ * allocated inodes, free inodes and used filesystem blocks as this
+ * information is no longer persistent in the superblock. Once we have
+ * this information, write it into the in-core superblock structure.
+ */
+int
+xfs_initialize_perag_data(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agcount)
+{
+	xfs_agnumber_t		index;
+	struct xfs_perag	*pag;
+	struct xfs_sb		*sbp = &mp->m_sb;
+	uint64_t		ifree = 0;
+	uint64_t		ialloc = 0;
+	uint64_t		bfree = 0;
+	uint64_t		bfreelst = 0;
+	uint64_t		btree = 0;
+	uint64_t		fdblocks;
+	int			error = 0;
+
+	for (index = 0; index < agcount; index++) {
+		/*
+		 * read the agf, then the agi. This gets us
+		 * all the information we need and populates the
+		 * per-ag structures for us.
+		 */
+		error = xfs_alloc_pagf_init(mp, NULL, index, 0);
+		if (error)
+			return error;
+
+		error = xfs_ialloc_pagi_init(mp, NULL, index);
+		if (error)
+			return error;
+		pag = xfs_perag_get(mp, index);
+		ifree += pag->pagi_freecount;
+		ialloc += pag->pagi_count;
+		bfree += pag->pagf_freeblks;
+		bfreelst += pag->pagf_flcount;
+		btree += pag->pagf_btreeblks;
+		xfs_perag_put(pag);
+	}
+	fdblocks = bfree + bfreelst + btree;
+
+	/*
+	 * If the new summary counts are obviously incorrect, fail the
+	 * mount operation because that implies the AGFs are also corrupt.
+	 * Clear FS_COUNTERS so that we don't unmount with a dirty log, which
+	 * will prevent xfs_repair from fixing anything.
+	 */
+	if (fdblocks > sbp->sb_dblocks || ifree > ialloc) {
+		xfs_alert(mp, "AGF corruption. Please run xfs_repair.");
+		error = -EFSCORRUPTED;
+		goto out;
+	}
+
+	/* Overwrite incore superblock counters with just-read data */
+	spin_lock(&mp->m_sb_lock);
+	sbp->sb_ifree = ifree;
+	sbp->sb_icount = ialloc;
+	sbp->sb_fdblocks = fdblocks;
+	spin_unlock(&mp->m_sb_lock);
+
+	xfs_reinit_percpu_counters(mp);
+out:
+	xfs_fs_mark_healthy(mp, XFS_SICK_FS_COUNTERS);
+	return error;
+}
+
+STATIC void
+__xfs_free_perag(
+	struct rcu_head	*head)
+{
+	struct xfs_perag *pag = container_of(head, struct xfs_perag, rcu_head);
+
+	ASSERT(!delayed_work_pending(&pag->pag_blockgc_work));
+	ASSERT(atomic_read(&pag->pag_ref) == 0);
+	kmem_free(pag);
+}
+
+/*
+ * Free up the per-ag resources associated with the mount structure.
+ */
+void
+xfs_free_perag(
+	struct xfs_mount	*mp)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		spin_lock(&mp->m_perag_lock);
+		pag = radix_tree_delete(&mp->m_perag_tree, agno);
+		spin_unlock(&mp->m_perag_lock);
+		ASSERT(pag);
+		ASSERT(atomic_read(&pag->pag_ref) == 0);
+
+		cancel_delayed_work_sync(&pag->pag_blockgc_work);
+		xfs_iunlink_destroy(pag);
+		xfs_buf_hash_destroy(pag);
+
+		call_rcu(&pag->rcu_head, __xfs_free_perag);
+	}
+}
+
+int
+xfs_initialize_perag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agcount,
+	xfs_agnumber_t		*maxagi)
+{
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		index;
+	xfs_agnumber_t		first_initialised = NULLAGNUMBER;
+	int			error;
+
+	/*
+	 * Walk the current per-ag tree so we don't try to initialise AGs
+	 * that already exist (growfs case). Allocate and insert all the
+	 * AGs we don't find ready for initialisation.
+	 */
+	for (index = 0; index < agcount; index++) {
+		pag = xfs_perag_get(mp, index);
+		if (pag) {
+			xfs_perag_put(pag);
+			continue;
+		}
+
+		pag = kmem_zalloc(sizeof(*pag), KM_MAYFAIL);
+		if (!pag) {
+			error = -ENOMEM;
+			goto out_unwind_new_pags;
+		}
+		pag->pag_agno = index;
+		pag->pag_mount = mp;
+
+		error = radix_tree_preload(GFP_NOFS);
+		if (error)
+			goto out_free_pag;
+
+		spin_lock(&mp->m_perag_lock);
+		if (radix_tree_insert(&mp->m_perag_tree, index, pag)) {
+			WARN_ON_ONCE(1);
+			spin_unlock(&mp->m_perag_lock);
+			radix_tree_preload_end();
+			error = -EEXIST;
+			goto out_free_pag;
+		}
+		spin_unlock(&mp->m_perag_lock);
+		radix_tree_preload_end();
+
+		/* Place kernel structure only init below this point. */
+		spin_lock_init(&pag->pag_ici_lock);
+		spin_lock_init(&pag->pagb_lock);
+		spin_lock_init(&pag->pag_state_lock);
+		INIT_DELAYED_WORK(&pag->pag_blockgc_work, xfs_blockgc_worker);
+		INIT_RADIX_TREE(&pag->pag_ici_root, GFP_ATOMIC);
+		init_waitqueue_head(&pag->pagb_wait);
+		pag->pagb_count = 0;
+		pag->pagb_tree = RB_ROOT;
+
+		error = xfs_buf_hash_init(pag);
+		if (error)
+			goto out_remove_pag;
+
+		error = xfs_iunlink_init(pag);
+		if (error)
+			goto out_hash_destroy;
+
+		/* first new pag is fully initialized */
+		if (first_initialised == NULLAGNUMBER)
+			first_initialised = index;
+	}
+
+	index = xfs_set_inode_alloc(mp, agcount);
+
+	if (maxagi)
+		*maxagi = index;
+
+	mp->m_ag_prealloc_blocks = xfs_prealloc_blocks(mp);
+	return 0;
+
+out_hash_destroy:
+	xfs_buf_hash_destroy(pag);
+out_remove_pag:
+	radix_tree_delete(&mp->m_perag_tree, index);
+out_free_pag:
+	kmem_free(pag);
+out_unwind_new_pags:
+	/* unwind any prior newly initialized pags */
+	for (index = first_initialised; index < agcount; index++) {
+		pag = radix_tree_delete(&mp->m_perag_tree, index);
+		if (!pag)
+			break;
+		xfs_buf_hash_destroy(pag);
+		xfs_iunlink_destroy(pag);
+		kmem_free(pag);
+	}
+	return error;
+}
 
 static int
 xfs_get_aghdr_buf(
@@ -43,7 +313,6 @@ xfs_get_aghdr_buf(
 	if (error)
 		return error;
 
-	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));
 	bp->b_bn = blkno;
 	bp->b_maps[0].bm_bn = blkno;
 	bp->b_ops = ops;
@@ -510,6 +779,7 @@ xfs_ag_shrink_space(
 	struct xfs_buf		*agibp, *agfbp;
 	struct xfs_agi		*agi;
 	struct xfs_agf		*agf;
+	xfs_agblock_t		aglen;
 	int			error, err2;
 
 	ASSERT(agno == mp->m_sb.sb_agcount - 1);
@@ -524,14 +794,14 @@ xfs_ag_shrink_space(
 		return error;
 
 	agf = agfbp->b_addr;
+	aglen = be32_to_cpu(agi->agi_length);
 	/* some extra paranoid checks before we shrink the ag */
 	if (XFS_IS_CORRUPT(mp, agf->agf_length != agi->agi_length))
 		return -EFSCORRUPTED;
-	if (delta >= agi->agi_length)
+	if (delta >= aglen)
 		return -EINVAL;
 
-	args.fsbno = XFS_AGB_TO_FSB(mp, agno,
-				    be32_to_cpu(agi->agi_length) - delta);
+	args.fsbno = XFS_AGB_TO_FSB(mp, agno, aglen - delta);
 
 	/*
 	 * Disable perag reservations so it doesn't cause the allocation request
@@ -646,7 +916,7 @@ xfs_ag_extend_space(
 	 * XFS_RMAP_OINFO_SKIP_UPDATE is used here to tell the rmap btree that
 	 * this doesn't actually exist in the rmap btree.
 	 */
-	error = xfs_rmap_free(tp, bp, id->agno,
+	error = xfs_rmap_free(tp, bp, bp->b_pag,
 				be32_to_cpu(agf->agf_length) - len,
 				len, &XFS_RMAP_OINFO_SKIP_UPDATE);
 	if (error)
diff --git a/fs/xfs/libxfs/xfs_ag.h b/fs/xfs/libxfs/xfs_ag.h
index 4535de1d88ea..4c6f9045baca 100644
--- a/fs/xfs/libxfs/xfs_ag.h
+++ b/fs/xfs/libxfs/xfs_ag.h
@@ -9,6 +9,142 @@
 
 struct xfs_mount;
 struct xfs_trans;
+struct xfs_perag;
+
+/*
+ * Per-ag infrastructure
+ */
+
+/* per-AG block reservation data structures*/
+struct xfs_ag_resv {
+	/* number of blocks originally reserved here */
+	xfs_extlen_t			ar_orig_reserved;
+	/* number of blocks reserved here */
+	xfs_extlen_t			ar_reserved;
+	/* number of blocks originally asked for */
+	xfs_extlen_t			ar_asked;
+};
+
+/*
+ * Per-ag incore structure, copies of information in agf and agi, to improve the
+ * performance of allocation group selection.
+ */
+struct xfs_perag {
+	struct xfs_mount *pag_mount;	/* owner filesystem */
+	xfs_agnumber_t	pag_agno;	/* AG this structure belongs to */
+	atomic_t	pag_ref;	/* perag reference count */
+	char		pagf_init;	/* this agf's entry is initialized */
+	char		pagi_init;	/* this agi's entry is initialized */
+	char		pagf_metadata;	/* the agf is preferred to be metadata */
+	char		pagi_inodeok;	/* The agi is ok for inodes */
+	uint8_t		pagf_levels[XFS_BTNUM_AGF];
+					/* # of levels in bno & cnt btree */
+	bool		pagf_agflreset; /* agfl requires reset before use */
+	uint32_t	pagf_flcount;	/* count of blocks in freelist */
+	xfs_extlen_t	pagf_freeblks;	/* total free blocks */
+	xfs_extlen_t	pagf_longest;	/* longest free space */
+	uint32_t	pagf_btreeblks;	/* # of blocks held in AGF btrees */
+	xfs_agino_t	pagi_freecount;	/* number of free inodes */
+	xfs_agino_t	pagi_count;	/* number of allocated inodes */
+
+	/*
+	 * Inode allocation search lookup optimisation.
+	 * If the pagino matches, the search for new inodes
+	 * doesn't need to search the near ones again straight away
+	 */
+	xfs_agino_t	pagl_pagino;
+	xfs_agino_t	pagl_leftrec;
+	xfs_agino_t	pagl_rightrec;
+
+	int		pagb_count;	/* pagb slots in use */
+	uint8_t		pagf_refcount_level; /* recount btree height */
+
+	/* Blocks reserved for all kinds of metadata. */
+	struct xfs_ag_resv	pag_meta_resv;
+	/* Blocks reserved for the reverse mapping btree. */
+	struct xfs_ag_resv	pag_rmapbt_resv;
+
+	/* -- kernel only structures below this line -- */
+
+	/*
+	 * Bitsets of per-ag metadata that have been checked and/or are sick.
+	 * Callers should hold pag_state_lock before accessing this field.
+	 */
+	uint16_t	pag_checked;
+	uint16_t	pag_sick;
+	spinlock_t	pag_state_lock;
+
+	spinlock_t	pagb_lock;	/* lock for pagb_tree */
+	struct rb_root	pagb_tree;	/* ordered tree of busy extents */
+	unsigned int	pagb_gen;	/* generation count for pagb_tree */
+	wait_queue_head_t pagb_wait;	/* woken when pagb_gen changes */
+
+	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
+
+	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
+	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
+	int		pag_ici_reclaimable;	/* reclaimable inodes */
+	unsigned long	pag_ici_reclaim_cursor;	/* reclaim restart point */
+
+	/* buffer cache index */
+	spinlock_t	pag_buf_lock;	/* lock for pag_buf_hash */
+	struct rhashtable pag_buf_hash;
+
+	/* for rcu-safe freeing */
+	struct rcu_head	rcu_head;
+
+	/* background prealloc block trimming */
+	struct delayed_work	pag_blockgc_work;
+
+	/*
+	 * Unlinked inode information.  This incore information reflects
+	 * data stored in the AGI, so callers must hold the AGI buffer lock
+	 * or have some other means to control concurrency.
+	 */
+	struct rhashtable	pagi_unlinked_hash;
+};
+
+int xfs_initialize_perag(struct xfs_mount *mp, xfs_agnumber_t agcount,
+			xfs_agnumber_t *maxagi);
+int xfs_initialize_perag_data(struct xfs_mount *mp, xfs_agnumber_t agno);
+void xfs_free_perag(struct xfs_mount *mp);
+
+struct xfs_perag *xfs_perag_get(struct xfs_mount *mp, xfs_agnumber_t agno);
+struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *mp, xfs_agnumber_t agno,
+		unsigned int tag);
+void xfs_perag_put(struct xfs_perag *pag);
+
+/*
+ * Perag iteration APIs
+ *
+ * XXX: for_each_perag_range() usage really needs an iterator to clean up when
+ * we terminate at end_agno because we may have taken a reference to the perag
+ * beyond end_agno. Right now callers have to be careful to catch and clean that
+ * up themselves. This is not necessary for the callers of for_each_perag() and
+ * for_each_perag_from() because they terminate at sb_agcount where there are
+ * no perag structures in tree beyond end_agno.
+ */
+#define for_each_perag_range(mp, next_agno, end_agno, pag) \
+	for ((pag) = xfs_perag_get((mp), (next_agno)); \
+		(pag) != NULL && (next_agno) <= (end_agno); \
+		(next_agno) = (pag)->pag_agno + 1, \
+		xfs_perag_put(pag), \
+		(pag) = xfs_perag_get((mp), (next_agno)))
+
+#define for_each_perag_from(mp, next_agno, pag) \
+	for_each_perag_range((mp), (next_agno), (mp)->m_sb.sb_agcount, (pag))
+
+
+#define for_each_perag(mp, agno, pag) \
+	(agno) = 0; \
+	for_each_perag_from((mp), (agno), (pag))
+
+#define for_each_perag_tag(mp, agno, pag, tag) \
+	for ((agno) = 0, (pag) = xfs_perag_get_tag((mp), 0, (tag)); \
+		(pag) != NULL; \
+		(agno) = (pag)->pag_agno + 1, \
+		xfs_perag_put(pag), \
+		(pag) = xfs_perag_get_tag((mp), (agno), (tag)))
 
 struct aghdr_init_data {
 	/* per ag data */
diff --git a/fs/xfs/libxfs/xfs_ag_resv.c b/fs/xfs/libxfs/xfs_ag_resv.c
index e0d5cdc57cc2..2aa2b3484c28 100644
--- a/fs/xfs/libxfs/xfs_ag_resv.c
+++ b/fs/xfs/libxfs/xfs_ag_resv.c
@@ -19,7 +19,7 @@
 #include "xfs_btree.h"
 #include "xfs_refcount_btree.h"
 #include "xfs_ialloc_btree.h"
-#include "xfs_sb.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 
 /*
@@ -250,7 +250,6 @@ xfs_ag_resv_init(
 	struct xfs_trans		*tp)
 {
 	struct xfs_mount		*mp = pag->pag_mount;
-	xfs_agnumber_t			agno = pag->pag_agno;
 	xfs_extlen_t			ask;
 	xfs_extlen_t			used;
 	int				error = 0, error2;
@@ -260,11 +259,11 @@ xfs_ag_resv_init(
 	if (pag->pag_meta_resv.ar_asked == 0) {
 		ask = used = 0;
 
-		error = xfs_refcountbt_calc_reserves(mp, tp, agno, &ask, &used);
+		error = xfs_refcountbt_calc_reserves(mp, tp, pag, &ask, &used);
 		if (error)
 			goto out;
 
-		error = xfs_finobt_calc_reserves(mp, tp, agno, &ask, &used);
+		error = xfs_finobt_calc_reserves(mp, tp, pag, &ask, &used);
 		if (error)
 			goto out;
 
@@ -282,7 +281,7 @@ xfs_ag_resv_init(
 
 			mp->m_finobt_nores = true;
 
-			error = xfs_refcountbt_calc_reserves(mp, tp, agno, &ask,
+			error = xfs_refcountbt_calc_reserves(mp, tp, pag, &ask,
 					&used);
 			if (error)
 				goto out;
@@ -300,7 +299,7 @@ xfs_ag_resv_init(
 	if (pag->pag_rmapbt_resv.ar_asked == 0) {
 		ask = used = 0;
 
-		error = xfs_rmapbt_calc_reserves(mp, tp, agno, &ask, &used);
+		error = xfs_rmapbt_calc_reserves(mp, tp, pag, &ask, &used);
 		if (error)
 			goto out;
 
diff --git a/fs/xfs/libxfs/xfs_ag_resv.h b/fs/xfs/libxfs/xfs_ag_resv.h
index 8a8eb4bc48bb..b74b210008ea 100644
--- a/fs/xfs/libxfs/xfs_ag_resv.h
+++ b/fs/xfs/libxfs/xfs_ag_resv.h
@@ -18,6 +18,21 @@ void xfs_ag_resv_alloc_extent(struct xfs_perag *pag, enum xfs_ag_resv_type type,
 void xfs_ag_resv_free_extent(struct xfs_perag *pag, enum xfs_ag_resv_type type,
 		struct xfs_trans *tp, xfs_extlen_t len);
 
+static inline struct xfs_ag_resv *
+xfs_perag_resv(
+	struct xfs_perag	*pag,
+	enum xfs_ag_resv_type	type)
+{
+	switch (type) {
+	case XFS_AG_RESV_METADATA:
+		return &pag->pag_meta_resv;
+	case XFS_AG_RESV_RMAPBT:
+		return &pag->pag_rmapbt_resv;
+	default:
+		return NULL;
+	}
+}
+
 /*
  * RMAPBT reservation accounting wrappers. Since rmapbt blocks are sourced from
  * the AGFL, they are allocated one at a time and the reservation updates don't
diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
index af3d5f9271f6..6929157d8d6e 100644
--- a/fs/xfs/libxfs/xfs_alloc.c
+++ b/fs/xfs/libxfs/xfs_alloc.c
@@ -10,7 +10,6 @@
 #include "xfs_shared.h"
 #include "xfs_trans_resv.h"
 #include "xfs_bit.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_defer.h"
 #include "xfs_btree.h"
@@ -24,6 +23,7 @@
 #include "xfs_trans.h"
 #include "xfs_buf_item.h"
 #include "xfs_log.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 #include "xfs_bmap.h"
 
@@ -230,7 +230,7 @@ xfs_alloc_get_rec(
 	int			*stat)	/* output: success/failure */
 {
 	struct xfs_mount	*mp = cur->bc_mp;
-	xfs_agnumber_t		agno = cur->bc_ag.agno;
+	xfs_agnumber_t		agno = cur->bc_ag.pag->pag_agno;
 	union xfs_btree_rec	*rec;
 	int			error;
 
@@ -776,7 +776,7 @@ xfs_alloc_cur_setup(
 	 */
 	if (!acur->cnt)
 		acur->cnt = xfs_allocbt_init_cursor(args->mp, args->tp,
-					args->agbp, args->agno, XFS_BTNUM_CNT);
+					args->agbp, args->pag, XFS_BTNUM_CNT);
 	error = xfs_alloc_lookup_ge(acur->cnt, 0, args->maxlen, &i);
 	if (error)
 		return error;
@@ -786,10 +786,10 @@ xfs_alloc_cur_setup(
 	 */
 	if (!acur->bnolt)
 		acur->bnolt = xfs_allocbt_init_cursor(args->mp, args->tp,
-					args->agbp, args->agno, XFS_BTNUM_BNO);
+					args->agbp, args->pag, XFS_BTNUM_BNO);
 	if (!acur->bnogt)
 		acur->bnogt = xfs_allocbt_init_cursor(args->mp, args->tp,
-					args->agbp, args->agno, XFS_BTNUM_BNO);
+					args->agbp, args->pag, XFS_BTNUM_BNO);
 	return i == 1 ? 0 : -ENOSPC;
 }
 
@@ -1063,7 +1063,7 @@ xfs_alloc_ag_vextent_small(
 	if (fbno == NULLAGBLOCK)
 		goto out;
 
-	xfs_extent_busy_reuse(args->mp, args->agno, fbno, 1,
+	xfs_extent_busy_reuse(args->mp, args->pag, fbno, 1,
 			      (args->datatype & XFS_ALLOC_NOBUSY));
 
 	if (args->datatype & XFS_ALLOC_USERDATA) {
@@ -1089,7 +1089,7 @@ xfs_alloc_ag_vextent_small(
 	 * If we're feeding an AGFL block to something that doesn't live in the
 	 * free space, we need to clear out the OWN_AG rmap.
 	 */
-	error = xfs_rmap_free(args->tp, args->agbp, args->agno, fbno, 1,
+	error = xfs_rmap_free(args->tp, args->agbp, args->pag, fbno, 1,
 			      &XFS_RMAP_OINFO_AG);
 	if (error)
 		goto error;
@@ -1166,7 +1166,7 @@ xfs_alloc_ag_vextent(
 
 	/* if not file data, insert new block into the reverse map btree */
 	if (!xfs_rmap_should_skip_owner_update(&args->oinfo)) {
-		error = xfs_rmap_alloc(args->tp, args->agbp, args->agno,
+		error = xfs_rmap_alloc(args->tp, args->agbp, args->pag,
 				       args->agbno, args->len, &args->oinfo);
 		if (error)
 			return error;
@@ -1178,7 +1178,7 @@ xfs_alloc_ag_vextent(
 		if (error)
 			return error;
 
-		ASSERT(!xfs_extent_busy_search(args->mp, args->agno,
+		ASSERT(!xfs_extent_busy_search(args->mp, args->pag,
 					      args->agbno, args->len));
 	}
 
@@ -1217,7 +1217,7 @@ xfs_alloc_ag_vextent_exact(
 	 * Allocate/initialize a cursor for the by-number freespace btree.
 	 */
 	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
-					  args->agno, XFS_BTNUM_BNO);
+					  args->pag, XFS_BTNUM_BNO);
 
 	/*
 	 * Lookup bno and minlen in the btree (minlen is irrelevant, really).
@@ -1277,7 +1277,7 @@ xfs_alloc_ag_vextent_exact(
 	 * Allocate/initialize a cursor for the by-size btree.
 	 */
 	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_CNT);
+					args->pag, XFS_BTNUM_CNT);
 	ASSERT(args->agbno + args->len <= be32_to_cpu(agf->agf_length));
 	error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen, args->agbno,
 				      args->len, XFSA_FIXUP_BNO_OK);
@@ -1674,9 +1674,8 @@ restart:
 	 * Allocate and initialize a cursor for the by-size btree.
 	 */
 	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_CNT);
+					args->pag, XFS_BTNUM_CNT);
 	bno_cur = NULL;
-	busy = false;
 
 	/*
 	 * Look for an entry >= maxlen+alignment-1 blocks.
@@ -1837,7 +1836,7 @@ restart:
 	 * Allocate and initialize a cursor for the by-block tree.
 	 */
 	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_BNO);
+					args->pag, XFS_BTNUM_BNO);
 	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
 			rbno, rlen, XFSA_FIXUP_CNT_OK)))
 		goto error0;
@@ -1896,12 +1895,13 @@ xfs_free_ag_extent(
 	int				haveright; /* have a right neighbor */
 	int				i;
 	int				error;
+	struct xfs_perag		*pag = agbp->b_pag;
 
 	bno_cur = cnt_cur = NULL;
 	mp = tp->t_mountp;
 
 	if (!xfs_rmap_should_skip_owner_update(oinfo)) {
-		error = xfs_rmap_free(tp, agbp, agno, bno, len, oinfo);
+		error = xfs_rmap_free(tp, agbp, pag, bno, len, oinfo);
 		if (error)
 			goto error0;
 	}
@@ -1909,7 +1909,7 @@ xfs_free_ag_extent(
 	/*
 	 * Allocate and initialize a cursor for the by-block btree.
 	 */
-	bno_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_BNO);
+	bno_cur = xfs_allocbt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_BNO);
 	/*
 	 * Look for a neighboring block on the left (lower block numbers)
 	 * that is contiguous with this space.
@@ -1979,7 +1979,7 @@ xfs_free_ag_extent(
 	/*
 	 * Now allocate and initialize a cursor for the by-size tree.
 	 */
-	cnt_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_CNT);
+	cnt_cur = xfs_allocbt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_CNT);
 	/*
 	 * Have both left and right contiguous neighbors.
 	 * Merge all three into a single free block.
@@ -2490,7 +2490,7 @@ xfs_exact_minlen_extent_available(
 	int			error = 0;
 
 	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, agbp,
-			args->agno, XFS_BTNUM_CNT);
+					args->pag, XFS_BTNUM_CNT);
 	error = xfs_alloc_lookup_ge(cnt_cur, 0, args->minlen, stat);
 	if (error)
 		goto out;
@@ -2693,21 +2693,21 @@ out_no_agbp:
  * Get a block from the freelist.
  * Returns with the buffer for the block gotten.
  */
-int				/* error */
+int
 xfs_alloc_get_freelist(
-	xfs_trans_t	*tp,	/* transaction pointer */
-	struct xfs_buf	*agbp,	/* buffer containing the agf structure */
-	xfs_agblock_t	*bnop,	/* block address retrieved from freelist */
-	int		btreeblk) /* destination is a AGF btree */
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	xfs_agblock_t		*bnop,
+	int			btreeblk)
 {
-	struct xfs_agf	*agf = agbp->b_addr;
-	struct xfs_buf	*agflbp;/* buffer for a.g. freelist structure */
-	xfs_agblock_t	bno;	/* block number returned */
-	__be32		*agfl_bno;
-	int		error;
-	int		logflags;
-	xfs_mount_t	*mp = tp->t_mountp;
-	xfs_perag_t	*pag;	/* per allocation group data */
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_buf		*agflbp;
+	xfs_agblock_t		bno;
+	__be32			*agfl_bno;
+	int			error;
+	int			logflags;
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
 
 	/*
 	 * Freelist is empty, give up.
@@ -2817,20 +2817,20 @@ xfs_alloc_pagf_init(
 /*
  * Put the block on the freelist for the allocation group.
  */
-int					/* error */
+int
 xfs_alloc_put_freelist(
-	xfs_trans_t		*tp,	/* transaction pointer */
-	struct xfs_buf		*agbp,	/* buffer for a.g. freelist header */
-	struct xfs_buf		*agflbp,/* buffer for a.g. free block array */
-	xfs_agblock_t		bno,	/* block being freed */
-	int			btreeblk) /* block came from a AGF btree */
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_buf		*agflbp,
+	xfs_agblock_t		bno,
+	int			btreeblk)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
 	struct xfs_agf		*agf = agbp->b_addr;
-	__be32			*blockp;/* pointer to array entry */
+	struct xfs_perag	*pag;
+	__be32			*blockp;
 	int			error;
 	int			logflags;
-	xfs_perag_t		*pag;	/* per allocation group data */
 	__be32			*agfl_bno;
 	int			startoff;
 
@@ -3292,7 +3292,7 @@ error0:
 int
 xfs_free_extent_fix_freelist(
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	struct xfs_buf		**agbp)
 {
 	struct xfs_alloc_arg	args;
@@ -3301,7 +3301,8 @@ xfs_free_extent_fix_freelist(
 	memset(&args, 0, sizeof(struct xfs_alloc_arg));
 	args.tp = tp;
 	args.mp = tp->t_mountp;
-	args.agno = agno;
+	args.agno = pag->pag_agno;
+	args.pag = pag;
 
 	/*
 	 * validate that the block number is legal - the enables us to detect
@@ -3310,17 +3311,12 @@ xfs_free_extent_fix_freelist(
 	if (args.agno >= args.mp->m_sb.sb_agcount)
 		return -EFSCORRUPTED;
 
-	args.pag = xfs_perag_get(args.mp, args.agno);
-	ASSERT(args.pag);
-
 	error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING);
 	if (error)
-		goto out;
+		return error;
 
 	*agbp = args.agbp;
-out:
-	xfs_perag_put(args.pag);
-	return error;
+	return 0;
 }
 
 /*
@@ -3344,6 +3340,7 @@ __xfs_free_extent(
 	struct xfs_agf			*agf;
 	int				error;
 	unsigned int			busy_flags = 0;
+	struct xfs_perag		*pag;
 
 	ASSERT(len != 0);
 	ASSERT(type != XFS_AG_RESV_AGFL);
@@ -3352,33 +3349,37 @@ __xfs_free_extent(
 			XFS_ERRTAG_FREE_EXTENT))
 		return -EIO;
 
-	error = xfs_free_extent_fix_freelist(tp, agno, &agbp);
+	pag = xfs_perag_get(mp, agno);
+	error = xfs_free_extent_fix_freelist(tp, pag, &agbp);
 	if (error)
-		return error;
+		goto err;
 	agf = agbp->b_addr;
 
 	if (XFS_IS_CORRUPT(mp, agbno >= mp->m_sb.sb_agblocks)) {
 		error = -EFSCORRUPTED;
-		goto err;
+		goto err_release;
 	}
 
 	/* validate the extent size is legal now we have the agf locked */
 	if (XFS_IS_CORRUPT(mp, agbno + len > be32_to_cpu(agf->agf_length))) {
 		error = -EFSCORRUPTED;
-		goto err;
+		goto err_release;
 	}
 
 	error = xfs_free_ag_extent(tp, agbp, agno, agbno, len, oinfo, type);
 	if (error)
-		goto err;
+		goto err_release;
 
 	if (skip_discard)
 		busy_flags |= XFS_EXTENT_BUSY_SKIP_DISCARD;
-	xfs_extent_busy_insert(tp, agno, agbno, len, busy_flags);
+	xfs_extent_busy_insert(tp, pag, agbno, len, busy_flags);
+	xfs_perag_put(pag);
 	return 0;
 
-err:
+err_release:
 	xfs_trans_brelse(tp, agbp);
+err:
+	xfs_perag_put(pag);
 	return error;
 }
 
diff --git a/fs/xfs/libxfs/xfs_alloc.h b/fs/xfs/libxfs/xfs_alloc.h
index a4427c5775c2..e30900b6f8ba 100644
--- a/fs/xfs/libxfs/xfs_alloc.h
+++ b/fs/xfs/libxfs/xfs_alloc.h
@@ -214,7 +214,7 @@ int xfs_alloc_read_agfl(struct xfs_mount *mp, struct xfs_trans *tp,
 int xfs_free_agfl_block(struct xfs_trans *, xfs_agnumber_t, xfs_agblock_t,
 			struct xfs_buf *, struct xfs_owner_info *);
 int xfs_alloc_fix_freelist(struct xfs_alloc_arg *args, int flags);
-int xfs_free_extent_fix_freelist(struct xfs_trans *tp, xfs_agnumber_t agno,
+int xfs_free_extent_fix_freelist(struct xfs_trans *tp, struct xfs_perag *pag,
 		struct xfs_buf **agbp);
 
 xfs_extlen_t xfs_prealloc_blocks(struct xfs_mount *mp);
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c b/fs/xfs/libxfs/xfs_alloc_btree.c
index a43e4c50e69b..6b363f78cfa2 100644
--- a/fs/xfs/libxfs/xfs_alloc_btree.c
+++ b/fs/xfs/libxfs/xfs_alloc_btree.c
@@ -9,7 +9,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_btree.h"
 #include "xfs_btree_staging.h"
@@ -19,6 +18,7 @@
 #include "xfs_error.h"
 #include "xfs_trace.h"
 #include "xfs_trans.h"
+#include "xfs_ag.h"
 
 
 STATIC struct xfs_btree_cur *
@@ -26,8 +26,7 @@ xfs_allocbt_dup_cursor(
 	struct xfs_btree_cur	*cur)
 {
 	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
-			cur->bc_ag.agbp, cur->bc_ag.agno,
-			cur->bc_btnum);
+			cur->bc_ag.agbp, cur->bc_ag.pag, cur->bc_btnum);
 }
 
 STATIC void
@@ -39,13 +38,12 @@ xfs_allocbt_set_root(
 	struct xfs_buf		*agbp = cur->bc_ag.agbp;
 	struct xfs_agf		*agf = agbp->b_addr;
 	int			btnum = cur->bc_btnum;
-	struct xfs_perag	*pag = agbp->b_pag;
 
 	ASSERT(ptr->s != 0);
 
 	agf->agf_roots[btnum] = ptr->s;
 	be32_add_cpu(&agf->agf_levels[btnum], inc);
-	pag->pagf_levels[btnum] += inc;
+	cur->bc_ag.pag->pagf_levels[btnum] += inc;
 
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
 }
@@ -72,7 +70,7 @@ xfs_allocbt_alloc_block(
 	}
 
 	atomic64_inc(&cur->bc_mp->m_allocbt_blks);
-	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_ag.agno, bno, 1, false);
+	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_ag.agbp->b_pag, bno, 1, false);
 
 	new->s = cpu_to_be32(bno);
 
@@ -86,7 +84,6 @@ xfs_allocbt_free_block(
 	struct xfs_buf		*bp)
 {
 	struct xfs_buf		*agbp = cur->bc_ag.agbp;
-	struct xfs_agf		*agf = agbp->b_addr;
 	xfs_agblock_t		bno;
 	int			error;
 
@@ -96,7 +93,7 @@ xfs_allocbt_free_block(
 		return error;
 
 	atomic64_dec(&cur->bc_mp->m_allocbt_blks);
-	xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
+	xfs_extent_busy_insert(cur->bc_tp, agbp->b_pag, bno, 1,
 			      XFS_EXTENT_BUSY_SKIP_DISCARD);
 	return 0;
 }
@@ -225,7 +222,7 @@ xfs_allocbt_init_ptr_from_cur(
 {
 	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_ag.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
 
 	ptr->s = agf->agf_roots[cur->bc_btnum];
 }
@@ -473,7 +470,7 @@ STATIC struct xfs_btree_cur *
 xfs_allocbt_init_common(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)
 {
 	struct xfs_btree_cur	*cur;
@@ -486,6 +483,7 @@ xfs_allocbt_init_common(
 	cur->bc_mp = mp;
 	cur->bc_btnum = btnum;
 	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+	cur->bc_ag.abt.active = false;
 
 	if (btnum == XFS_BTNUM_CNT) {
 		cur->bc_ops = &xfs_cntbt_ops;
@@ -496,8 +494,9 @@ xfs_allocbt_init_common(
 		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtb_2);
 	}
 
-	cur->bc_ag.agno = agno;
-	cur->bc_ag.abt.active = false;
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
 
 	if (xfs_sb_version_hascrc(&mp->m_sb))
 		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
@@ -513,13 +512,13 @@ xfs_allocbt_init_cursor(
 	struct xfs_mount	*mp,		/* file system mount point */
 	struct xfs_trans	*tp,		/* transaction pointer */
 	struct xfs_buf		*agbp,		/* buffer for agf structure */
-	xfs_agnumber_t		agno,		/* allocation group number */
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)		/* btree identifier */
 {
 	struct xfs_agf		*agf = agbp->b_addr;
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_allocbt_init_common(mp, tp, agno, btnum);
+	cur = xfs_allocbt_init_common(mp, tp, pag, btnum);
 	if (btnum == XFS_BTNUM_CNT)
 		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
 	else
@@ -535,12 +534,12 @@ struct xfs_btree_cur *
 xfs_allocbt_stage_cursor(
 	struct xfs_mount	*mp,
 	struct xbtree_afakeroot	*afake,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)
 {
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_allocbt_init_common(mp, NULL, agno, btnum);
+	cur = xfs_allocbt_init_common(mp, NULL, pag, btnum);
 	xfs_btree_stage_afakeroot(cur, afake);
 	return cur;
 }
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.h b/fs/xfs/libxfs/xfs_alloc_btree.h
index a5b998e950fe..9eb4c667a6b8 100644
--- a/fs/xfs/libxfs/xfs_alloc_btree.h
+++ b/fs/xfs/libxfs/xfs_alloc_btree.h
@@ -13,6 +13,7 @@
 struct xfs_buf;
 struct xfs_btree_cur;
 struct xfs_mount;
+struct xfs_perag;
 struct xbtree_afakeroot;
 
 /*
@@ -46,11 +47,11 @@ struct xbtree_afakeroot;
 		 (maxrecs) * sizeof(xfs_alloc_key_t) + \
 		 ((index) - 1) * sizeof(xfs_alloc_ptr_t)))
 
-extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *,
-		struct xfs_trans *, struct xfs_buf *,
-		xfs_agnumber_t, xfs_btnum_t);
+extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_buf *bp,
+		struct xfs_perag *pag, xfs_btnum_t btnum);
 struct xfs_btree_cur *xfs_allocbt_stage_cursor(struct xfs_mount *mp,
-		struct xbtree_afakeroot *afake, xfs_agnumber_t agno,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag,
 		xfs_btnum_t btnum);
 extern int xfs_allocbt_maxrecs(struct xfs_mount *, int, int);
 extern xfs_extlen_t xfs_allocbt_calc_size(struct xfs_mount *mp,
diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
index 96146f425e50..d9d7d5137b73 100644
--- a/fs/xfs/libxfs/xfs_attr.c
+++ b/fs/xfs/libxfs/xfs_attr.c
@@ -44,20 +44,27 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args);
  * Internal routines when attribute list is one block.
  */
 STATIC int xfs_attr_leaf_get(xfs_da_args_t *args);
-STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args);
 STATIC int xfs_attr_leaf_removename(xfs_da_args_t *args);
 STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
+STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp);
 
 /*
  * Internal routines when attribute list is more than one block.
  */
 STATIC int xfs_attr_node_get(xfs_da_args_t *args);
-STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
-STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
+STATIC void xfs_attr_restore_rmt_blk(struct xfs_da_args *args);
+STATIC int xfs_attr_node_addname(struct xfs_delattr_context *dac);
+STATIC int xfs_attr_node_addname_find_attr(struct xfs_delattr_context *dac);
+STATIC int xfs_attr_node_addname_clear_incomplete(
+				struct xfs_delattr_context *dac);
 STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
 				 struct xfs_da_state **state);
 STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
 STATIC int xfs_attr_refillstate(xfs_da_state_t *state);
+STATIC int xfs_attr_set_iter(struct xfs_delattr_context *dac,
+			     struct xfs_buf **leaf_bp);
+STATIC int xfs_attr_node_removename(struct xfs_da_args *args,
+				    struct xfs_da_state *state);
 
 int
 xfs_inode_hasattr(
@@ -237,27 +244,77 @@ xfs_attr_is_shortform(
 }
 
 /*
- * Attempts to set an attr in shortform, or converts short form to leaf form if
- * there is not enough room.  If the attr is set, the transaction is committed
- * and set to NULL.
+ * Checks to see if a delayed attribute transaction should be rolled.  If so,
+ * transaction is finished or rolled as needed.
  */
 STATIC int
-xfs_attr_set_shortform(
-	struct xfs_da_args	*args,
-	struct xfs_buf		**leaf_bp)
+xfs_attr_trans_roll(
+	struct xfs_delattr_context	*dac)
 {
-	struct xfs_inode	*dp = args->dp;
-	int			error, error2 = 0;
+	struct xfs_da_args		*args = dac->da_args;
+	int				error;
+
+	if (dac->flags & XFS_DAC_DEFER_FINISH) {
+		/*
+		 * The caller wants us to finish all the deferred ops so that we
+		 * avoid pinning the log tail with a large number of deferred
+		 * ops.
+		 */
+		dac->flags &= ~XFS_DAC_DEFER_FINISH;
+		error = xfs_defer_finish(&args->trans);
+	} else
+		error = xfs_trans_roll_inode(&args->trans, args->dp);
+
+	return error;
+}
+
+/*
+ * Set the attribute specified in @args.
+ */
+int
+xfs_attr_set_args(
+	struct xfs_da_args		*args)
+{
+	struct xfs_buf			*leaf_bp = NULL;
+	int				error = 0;
+	struct xfs_delattr_context	dac = {
+		.da_args	= args,
+	};
+
+	do {
+		error = xfs_attr_set_iter(&dac, &leaf_bp);
+		if (error != -EAGAIN)
+			break;
+
+		error = xfs_attr_trans_roll(&dac);
+		if (error) {
+			if (leaf_bp)
+				xfs_trans_brelse(args->trans, leaf_bp);
+			return error;
+		}
+	} while (true);
+
+	return error;
+}
+
+STATIC int
+xfs_attr_sf_addname(
+	struct xfs_delattr_context	*dac,
+	struct xfs_buf			**leaf_bp)
+{
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_inode		*dp = args->dp;
+	int				error = 0;
 
 	/*
 	 * Try to add the attr to the attribute list in the inode.
 	 */
 	error = xfs_attr_try_sf_addname(dp, args);
-	if (error != -ENOSPC) {
-		error2 = xfs_trans_commit(args->trans);
-		args->trans = NULL;
-		return error ? error : error2;
-	}
+
+	/* Should only be 0, -EEXIST or -ENOSPC */
+	if (error != -ENOSPC)
+		return error;
+
 	/*
 	 * It won't fit in the shortform, transform to a leaf block.  GROT:
 	 * another possible req'mt for a double-split btree op.
@@ -269,85 +326,300 @@ xfs_attr_set_shortform(
 	/*
 	 * Prevent the leaf buffer from being unlocked so that a concurrent AIL
 	 * push cannot grab the half-baked leaf buffer and run into problems
-	 * with the write verifier. Once we're done rolling the transaction we
-	 * can release the hold and add the attr to the leaf.
+	 * with the write verifier.
 	 */
 	xfs_trans_bhold(args->trans, *leaf_bp);
-	error = xfs_defer_finish(&args->trans);
-	xfs_trans_bhold_release(args->trans, *leaf_bp);
-	if (error) {
-		xfs_trans_brelse(args->trans, *leaf_bp);
-		return error;
-	}
 
-	return 0;
+	/*
+	 * We're still in XFS_DAS_UNINIT state here.  We've converted
+	 * the attr fork to leaf format and will restart with the leaf
+	 * add.
+	 */
+	dac->flags |= XFS_DAC_DEFER_FINISH;
+	return -EAGAIN;
 }
 
 /*
  * Set the attribute specified in @args.
+ * This routine is meant to function as a delayed operation, and may return
+ * -EAGAIN when the transaction needs to be rolled.  Calling functions will need
+ * to handle this, and recall the function until a successful error code is
+ * returned.
  */
 int
-xfs_attr_set_args(
-	struct xfs_da_args	*args)
+xfs_attr_set_iter(
+	struct xfs_delattr_context	*dac,
+	struct xfs_buf			**leaf_bp)
 {
-	struct xfs_inode	*dp = args->dp;
-	struct xfs_buf          *leaf_bp = NULL;
-	int			error = 0;
+	struct xfs_da_args              *args = dac->da_args;
+	struct xfs_inode		*dp = args->dp;
+	struct xfs_buf			*bp = NULL;
+	int				forkoff, error = 0;
+
+	/* State machine switch */
+	switch (dac->dela_state) {
+	case XFS_DAS_UNINIT:
+		/*
+		 * If the fork is shortform, attempt to add the attr. If there
+		 * is no space, this converts to leaf format and returns
+		 * -EAGAIN with the leaf buffer held across the roll. The caller
+		 * will deal with a transaction roll error, but otherwise
+		 * release the hold once we return with a clean transaction.
+		 */
+		if (xfs_attr_is_shortform(dp))
+			return xfs_attr_sf_addname(dac, leaf_bp);
+		if (*leaf_bp != NULL) {
+			xfs_trans_bhold_release(args->trans, *leaf_bp);
+			*leaf_bp = NULL;
+		}
 
-	/*
-	 * If the attribute list is already in leaf format, jump straight to
-	 * leaf handling.  Otherwise, try to add the attribute to the shortform
-	 * list; if there's no room then convert the list to leaf format and try
-	 * again.
-	 */
-	if (xfs_attr_is_shortform(dp)) {
+		if (xfs_attr_is_leaf(dp)) {
+			error = xfs_attr_leaf_try_add(args, *leaf_bp);
+			if (error == -ENOSPC) {
+				error = xfs_attr3_leaf_to_node(args);
+				if (error)
+					return error;
+
+				/*
+				 * Finish any deferred work items and roll the
+				 * transaction once more.  The goal here is to
+				 * call node_addname with the inode and
+				 * transaction in the same state (inode locked
+				 * and joined, transaction clean) no matter how
+				 * we got to this step.
+				 *
+				 * At this point, we are still in
+				 * XFS_DAS_UNINIT, but when we come back, we'll
+				 * be a node, so we'll fall down into the node
+				 * handling code below
+				 */
+				dac->flags |= XFS_DAC_DEFER_FINISH;
+				return -EAGAIN;
+			} else if (error) {
+				return error;
+			}
+
+			dac->dela_state = XFS_DAS_FOUND_LBLK;
+		} else {
+			error = xfs_attr_node_addname_find_attr(dac);
+			if (error)
+				return error;
+
+			error = xfs_attr_node_addname(dac);
+			if (error)
+				return error;
+
+			dac->dela_state = XFS_DAS_FOUND_NBLK;
+		}
+		return -EAGAIN;
+	case XFS_DAS_FOUND_LBLK:
+		/*
+		 * If there was an out-of-line value, allocate the blocks we
+		 * identified for its storage and copy the value.  This is done
+		 * after we create the attribute so that we don't overflow the
+		 * maximum size of a transaction and/or hit a deadlock.
+		 */
+
+		/* Open coded xfs_attr_rmtval_set without trans handling */
+		if ((dac->flags & XFS_DAC_LEAF_ADDNAME_INIT) == 0) {
+			dac->flags |= XFS_DAC_LEAF_ADDNAME_INIT;
+			if (args->rmtblkno > 0) {
+				error = xfs_attr_rmtval_find_space(dac);
+				if (error)
+					return error;
+			}
+		}
 
 		/*
-		 * If the attr was successfully set in shortform, the
-		 * transaction is committed and set to NULL.  Otherwise, is it
-		 * converted from shortform to leaf, and the transaction is
-		 * retained.
+		 * Repeat allocating remote blocks for the attr value until
+		 * blkcnt drops to zero.
 		 */
-		error = xfs_attr_set_shortform(args, &leaf_bp);
-		if (error || !args->trans)
+		if (dac->blkcnt > 0) {
+			error = xfs_attr_rmtval_set_blk(dac);
+			if (error)
+				return error;
+			return -EAGAIN;
+		}
+
+		error = xfs_attr_rmtval_set_value(args);
+		if (error)
 			return error;
-	}
 
-	if (xfs_attr_is_leaf(dp)) {
-		error = xfs_attr_leaf_addname(args);
-		if (error != -ENOSPC)
+		/*
+		 * If this is not a rename, clear the incomplete flag and we're
+		 * done.
+		 */
+		if (!(args->op_flags & XFS_DA_OP_RENAME)) {
+			if (args->rmtblkno > 0)
+				error = xfs_attr3_leaf_clearflag(args);
 			return error;
+		}
 
 		/*
-		 * Promote the attribute list to the Btree format.
+		 * If this is an atomic rename operation, we must "flip" the
+		 * incomplete flags on the "new" and "old" attribute/value pairs
+		 * so that one disappears and one appears atomically.  Then we
+		 * must remove the "old" attribute/value pair.
+		 *
+		 * In a separate transaction, set the incomplete flag on the
+		 * "old" attr and clear the incomplete flag on the "new" attr.
 		 */
-		error = xfs_attr3_leaf_to_node(args);
+		error = xfs_attr3_leaf_flipflags(args);
 		if (error)
 			return error;
+		/*
+		 * Commit the flag value change and start the next trans in
+		 * series.
+		 */
+		dac->dela_state = XFS_DAS_FLIP_LFLAG;
+		return -EAGAIN;
+	case XFS_DAS_FLIP_LFLAG:
+		/*
+		 * Dismantle the "old" attribute/value pair by removing a
+		 * "remote" value (if it exists).
+		 */
+		xfs_attr_restore_rmt_blk(args);
+		error = xfs_attr_rmtval_invalidate(args);
+		if (error)
+			return error;
+
+		/* fallthrough */
+	case XFS_DAS_RM_LBLK:
+		/* Set state in case xfs_attr_rmtval_remove returns -EAGAIN */
+		dac->dela_state = XFS_DAS_RM_LBLK;
+		if (args->rmtblkno) {
+			error = __xfs_attr_rmtval_remove(dac);
+			if (error)
+				return error;
+
+			dac->dela_state = XFS_DAS_RD_LEAF;
+			return -EAGAIN;
+		}
 
+		/* fallthrough */
+	case XFS_DAS_RD_LEAF:
 		/*
-		 * Finish any deferred work items and roll the transaction once
-		 * more.  The goal here is to call node_addname with the inode
-		 * and transaction in the same state (inode locked and joined,
-		 * transaction clean) no matter how we got to this step.
+		 * This is the last step for leaf format. Read the block with
+		 * the old attr, remove the old attr, check for shortform
+		 * conversion and return.
 		 */
-		error = xfs_defer_finish(&args->trans);
+		error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno,
+					   &bp);
 		if (error)
 			return error;
 
+		xfs_attr3_leaf_remove(bp, args);
+
+		forkoff = xfs_attr_shortform_allfit(bp, dp);
+		if (forkoff)
+			error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
+			/* bp is gone due to xfs_da_shrink_inode */
+
+		return error;
+
+	case XFS_DAS_FOUND_NBLK:
+		/*
+		 * Find space for remote blocks and fall into the allocation
+		 * state.
+		 */
+		if (args->rmtblkno > 0) {
+			error = xfs_attr_rmtval_find_space(dac);
+			if (error)
+				return error;
+		}
+
+		/* fallthrough */
+	case XFS_DAS_ALLOC_NODE:
+		/*
+		 * If there was an out-of-line value, allocate the blocks we
+		 * identified for its storage and copy the value.  This is done
+		 * after we create the attribute so that we don't overflow the
+		 * maximum size of a transaction and/or hit a deadlock.
+		 */
+		dac->dela_state = XFS_DAS_ALLOC_NODE;
+		if (args->rmtblkno > 0) {
+			if (dac->blkcnt > 0) {
+				error = xfs_attr_rmtval_set_blk(dac);
+				if (error)
+					return error;
+				return -EAGAIN;
+			}
+
+			error = xfs_attr_rmtval_set_value(args);
+			if (error)
+				return error;
+		}
+
 		/*
-		 * Commit the current trans (including the inode) and
-		 * start a new one.
+		 * If this was not a rename, clear the incomplete flag and we're
+		 * done.
 		 */
-		error = xfs_trans_roll_inode(&args->trans, dp);
+		if (!(args->op_flags & XFS_DA_OP_RENAME)) {
+			if (args->rmtblkno > 0)
+				error = xfs_attr3_leaf_clearflag(args);
+			goto out;
+		}
+
+		/*
+		 * If this is an atomic rename operation, we must "flip" the
+		 * incomplete flags on the "new" and "old" attribute/value pairs
+		 * so that one disappears and one appears atomically.  Then we
+		 * must remove the "old" attribute/value pair.
+		 *
+		 * In a separate transaction, set the incomplete flag on the
+		 * "old" attr and clear the incomplete flag on the "new" attr.
+		 */
+		error = xfs_attr3_leaf_flipflags(args);
+		if (error)
+			goto out;
+		/*
+		 * Commit the flag value change and start the next trans in
+		 * series
+		 */
+		dac->dela_state = XFS_DAS_FLIP_NFLAG;
+		return -EAGAIN;
+
+	case XFS_DAS_FLIP_NFLAG:
+		/*
+		 * Dismantle the "old" attribute/value pair by removing a
+		 * "remote" value (if it exists).
+		 */
+		xfs_attr_restore_rmt_blk(args);
+
+		error = xfs_attr_rmtval_invalidate(args);
 		if (error)
 			return error;
-	}
 
-	error = xfs_attr_node_addname(args);
+		/* fallthrough */
+	case XFS_DAS_RM_NBLK:
+		/* Set state in case xfs_attr_rmtval_remove returns -EAGAIN */
+		dac->dela_state = XFS_DAS_RM_NBLK;
+		if (args->rmtblkno) {
+			error = __xfs_attr_rmtval_remove(dac);
+			if (error)
+				return error;
+
+			dac->dela_state = XFS_DAS_CLR_FLAG;
+			return -EAGAIN;
+		}
+
+		/* fallthrough */
+	case XFS_DAS_CLR_FLAG:
+		/*
+		 * The last state for node format. Look up the old attr and
+		 * remove it.
+		 */
+		error = xfs_attr_node_addname_clear_incomplete(dac);
+		break;
+	default:
+		ASSERT(0);
+		break;
+	}
+out:
 	return error;
 }
 
+
 /*
  * Return EEXIST if attr is found, or ENOATTR if not
  */
@@ -382,16 +654,25 @@ xfs_has_attr(
  */
 int
 xfs_attr_remove_args(
-	struct xfs_da_args      *args)
+	struct xfs_da_args	*args)
 {
-	if (!xfs_inode_hasattr(args->dp))
-		return -ENOATTR;
+	int				error;
+	struct xfs_delattr_context	dac = {
+		.da_args	= args,
+	};
 
-	if (args->dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL)
-		return xfs_attr_shortform_remove(args);
-	if (xfs_attr_is_leaf(args->dp))
-		return xfs_attr_leaf_removename(args);
-	return xfs_attr_node_removename(args);
+	do {
+		error = xfs_attr_remove_iter(&dac);
+		if (error != -EAGAIN)
+			break;
+
+		error = xfs_attr_trans_roll(&dac);
+		if (error)
+			return error;
+
+	} while (true);
+
+	return error;
 }
 
 /*
@@ -559,7 +840,7 @@ xfs_attr_shortform_addname(xfs_da_args_t *args)
 	if (retval == -EEXIST) {
 		if (args->attr_flags & XATTR_CREATE)
 			return retval;
-		retval = xfs_attr_shortform_remove(args);
+		retval = xfs_attr_sf_removename(args);
 		if (retval)
 			return retval;
 		/*
@@ -670,115 +951,6 @@ out_brelse:
 	return retval;
 }
 
-
-/*
- * Add a name to the leaf attribute list structure
- *
- * This leaf block cannot have a "remote" value, we only call this routine
- * if bmap_one_block() says there is only one block (ie: no remote blks).
- */
-STATIC int
-xfs_attr_leaf_addname(
-	struct xfs_da_args	*args)
-{
-	int			error, forkoff;
-	struct xfs_buf		*bp = NULL;
-	struct xfs_inode	*dp = args->dp;
-
-	trace_xfs_attr_leaf_addname(args);
-
-	error = xfs_attr_leaf_try_add(args, bp);
-	if (error)
-		return error;
-
-	/*
-	 * Commit the transaction that added the attr name so that
-	 * later routines can manage their own transactions.
-	 */
-	error = xfs_trans_roll_inode(&args->trans, dp);
-	if (error)
-		return error;
-
-	/*
-	 * If there was an out-of-line value, allocate the blocks we
-	 * identified for its storage and copy the value.  This is done
-	 * after we create the attribute so that we don't overflow the
-	 * maximum size of a transaction and/or hit a deadlock.
-	 */
-	if (args->rmtblkno > 0) {
-		error = xfs_attr_rmtval_set(args);
-		if (error)
-			return error;
-	}
-
-	if (!(args->op_flags & XFS_DA_OP_RENAME)) {
-		/*
-		 * Added a "remote" value, just clear the incomplete flag.
-		 */
-		if (args->rmtblkno > 0)
-			error = xfs_attr3_leaf_clearflag(args);
-
-		return error;
-	}
-
-	/*
-	 * If this is an atomic rename operation, we must "flip" the incomplete
-	 * flags on the "new" and "old" attribute/value pairs so that one
-	 * disappears and one appears atomically.  Then we must remove the "old"
-	 * attribute/value pair.
-	 *
-	 * In a separate transaction, set the incomplete flag on the "old" attr
-	 * and clear the incomplete flag on the "new" attr.
-	 */
-
-	error = xfs_attr3_leaf_flipflags(args);
-	if (error)
-		return error;
-	/*
-	 * Commit the flag value change and start the next trans in series.
-	 */
-	error = xfs_trans_roll_inode(&args->trans, args->dp);
-	if (error)
-		return error;
-
-	/*
-	 * Dismantle the "old" attribute/value pair by removing a "remote" value
-	 * (if it exists).
-	 */
-	xfs_attr_restore_rmt_blk(args);
-
-	if (args->rmtblkno) {
-		error = xfs_attr_rmtval_invalidate(args);
-		if (error)
-			return error;
-
-		error = xfs_attr_rmtval_remove(args);
-		if (error)
-			return error;
-	}
-
-	/*
-	 * Read in the block containing the "old" attr, then remove the "old"
-	 * attr from that block (neat, huh!)
-	 */
-	error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno,
-				   &bp);
-	if (error)
-		return error;
-
-	xfs_attr3_leaf_remove(bp, args);
-
-	/*
-	 * If the result is small enough, shrink it all into the inode.
-	 */
-	forkoff = xfs_attr_shortform_allfit(bp, dp);
-	if (forkoff)
-		error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
-		/* bp is gone due to xfs_da_shrink_inode */
-
-	return error;
-}
-
 /*
  * Return EEXIST if attr is found, or ENOATTR if not
  */
@@ -909,48 +1081,26 @@ xfs_attr_node_hasname(
  * External routines when attribute list size > geo->blksize
  *========================================================================*/
 
-/*
- * Add a name to a Btree-format attribute list.
- *
- * This will involve walking down the Btree, and may involve splitting
- * leaf nodes and even splitting intermediate nodes up to and including
- * the root node (a special case of an intermediate node).
- *
- * "Remote" attribute values confuse the issue and atomic rename operations
- * add a whole extra layer of confusion on top of that.
- */
 STATIC int
-xfs_attr_node_addname(
-	struct xfs_da_args	*args)
+xfs_attr_node_addname_find_attr(
+	struct xfs_delattr_context	*dac)
 {
-	struct xfs_da_state	*state;
-	struct xfs_da_state_blk	*blk;
-	struct xfs_inode	*dp;
-	int			retval, error;
-
-	trace_xfs_attr_node_addname(args);
+	struct xfs_da_args		*args = dac->da_args;
+	int				retval;
 
 	/*
-	 * Fill in bucket of arguments/results/context to carry around.
-	 */
-	dp = args->dp;
-restart:
-	/*
 	 * Search to see if name already exists, and get back a pointer
 	 * to where it should go.
 	 */
-	error = 0;
-	retval = xfs_attr_node_hasname(args, &state);
+	retval = xfs_attr_node_hasname(args, &dac->da_state);
 	if (retval != -ENOATTR && retval != -EEXIST)
-		goto out;
+		return retval;
 
-	blk = &state->path.blk[ state->path.active-1 ];
-	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
 	if (retval == -ENOATTR && (args->attr_flags & XATTR_REPLACE))
-		goto out;
+		goto error;
 	if (retval == -EEXIST) {
 		if (args->attr_flags & XATTR_CREATE)
-			goto out;
+			goto error;
 
 		trace_xfs_attr_node_replace(args);
 
@@ -968,8 +1118,44 @@ restart:
 		args->rmtvaluelen = 0;
 	}
 
-	retval = xfs_attr3_leaf_add(blk->bp, state->args);
-	if (retval == -ENOSPC) {
+	return 0;
+error:
+	if (dac->da_state)
+		xfs_da_state_free(dac->da_state);
+	return retval;
+}
+
+/*
+ * Add a name to a Btree-format attribute list.
+ *
+ * This will involve walking down the Btree, and may involve splitting
+ * leaf nodes and even splitting intermediate nodes up to and including
+ * the root node (a special case of an intermediate node).
+ *
+ * "Remote" attribute values confuse the issue and atomic rename operations
+ * add a whole extra layer of confusion on top of that.
+ *
+ * This routine is meant to function as a delayed operation, and may return
+ * -EAGAIN when the transaction needs to be rolled.  Calling functions will need
+ * to handle this, and recall the function until a successful error code is
+ *returned.
+ */
+STATIC int
+xfs_attr_node_addname(
+	struct xfs_delattr_context	*dac)
+{
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_da_state		*state = dac->da_state;
+	struct xfs_da_state_blk		*blk;
+	int				error;
+
+	trace_xfs_attr_node_addname(args);
+
+	blk = &state->path.blk[state->path.active-1];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+	error = xfs_attr3_leaf_add(blk->bp, state->args);
+	if (error == -ENOSPC) {
 		if (state->path.active == 1) {
 			/*
 			 * Its really a single leaf node, but it had
@@ -981,19 +1167,16 @@ restart:
 			error = xfs_attr3_leaf_to_node(args);
 			if (error)
 				goto out;
-			error = xfs_defer_finish(&args->trans);
-			if (error)
-				goto out;
 
 			/*
-			 * Commit the node conversion and start the next
-			 * trans in the chain.
+			 * Now that we have converted the leaf to a node, we can
+			 * roll the transaction, and try xfs_attr3_leaf_add
+			 * again on re-entry.  No need to set dela_state to do
+			 * this. dela_state is still unset by this function at
+			 * this point.
 			 */
-			error = xfs_trans_roll_inode(&args->trans, dp);
-			if (error)
-				goto out;
-
-			goto restart;
+			dac->flags |= XFS_DAC_DEFER_FINISH;
+			return -EAGAIN;
 		}
 
 		/*
@@ -1005,9 +1188,7 @@ restart:
 		error = xfs_da3_split(state);
 		if (error)
 			goto out;
-		error = xfs_defer_finish(&args->trans);
-		if (error)
-			goto out;
+		dac->flags |= XFS_DAC_DEFER_FINISH;
 	} else {
 		/*
 		 * Addition succeeded, update Btree hashvals.
@@ -1015,77 +1196,21 @@ restart:
 		xfs_da3_fixhashpath(state, &state->path);
 	}
 
-	/*
-	 * Kill the state structure, we're done with it and need to
-	 * allow the buffers to come back later.
-	 */
-	xfs_da_state_free(state);
-	state = NULL;
-
-	/*
-	 * Commit the leaf addition or btree split and start the next
-	 * trans in the chain.
-	 */
-	error = xfs_trans_roll_inode(&args->trans, dp);
-	if (error)
-		goto out;
-
-	/*
-	 * If there was an out-of-line value, allocate the blocks we
-	 * identified for its storage and copy the value.  This is done
-	 * after we create the attribute so that we don't overflow the
-	 * maximum size of a transaction and/or hit a deadlock.
-	 */
-	if (args->rmtblkno > 0) {
-		error = xfs_attr_rmtval_set(args);
-		if (error)
-			return error;
-	}
-
-	if (!(args->op_flags & XFS_DA_OP_RENAME)) {
-		/*
-		 * Added a "remote" value, just clear the incomplete flag.
-		 */
-		if (args->rmtblkno > 0)
-			error = xfs_attr3_leaf_clearflag(args);
-		retval = error;
-		goto out;
-	}
-
-	/*
-	 * If this is an atomic rename operation, we must "flip" the incomplete
-	 * flags on the "new" and "old" attribute/value pairs so that one
-	 * disappears and one appears atomically.  Then we must remove the "old"
-	 * attribute/value pair.
-	 *
-	 * In a separate transaction, set the incomplete flag on the "old" attr
-	 * and clear the incomplete flag on the "new" attr.
-	 */
-	error = xfs_attr3_leaf_flipflags(args);
-	if (error)
-		goto out;
-	/*
-	 * Commit the flag value change and start the next trans in series
-	 */
-	error = xfs_trans_roll_inode(&args->trans, args->dp);
-	if (error)
-		goto out;
-
-	/*
-	 * Dismantle the "old" attribute/value pair by removing a "remote" value
-	 * (if it exists).
-	 */
-	xfs_attr_restore_rmt_blk(args);
+out:
+	if (state)
+		xfs_da_state_free(state);
+	return error;
+}
 
-	if (args->rmtblkno) {
-		error = xfs_attr_rmtval_invalidate(args);
-		if (error)
-			return error;
 
-		error = xfs_attr_rmtval_remove(args);
-		if (error)
-			return error;
-	}
+STATIC int
+xfs_attr_node_addname_clear_incomplete(
+	struct xfs_delattr_context	*dac)
+{
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_da_state		*state = NULL;
+	int				retval = 0;
+	int				error = 0;
 
 	/*
 	 * Re-find the "old" attribute entry after any split ops. The INCOMPLETE
@@ -1098,13 +1223,7 @@ restart:
 	if (error)
 		goto out;
 
-	/*
-	 * Remove the name and update the hashvals in the tree.
-	 */
-	blk = &state->path.blk[state->path.active-1];
-	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
-	error = xfs_attr3_leaf_remove(blk->bp, args);
-	xfs_da3_fixhashpath(state, &state->path);
+	error = xfs_attr_node_removename(args, state);
 
 	/*
 	 * Check to see if the tree needs to be collapsed.
@@ -1190,14 +1309,16 @@ xfs_attr_leaf_mark_incomplete(
  */
 STATIC
 int xfs_attr_node_removename_setup(
-	struct xfs_da_args	*args,
-	struct xfs_da_state	**state)
+	struct xfs_delattr_context	*dac)
 {
-	int			error;
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_da_state		**state = &dac->da_state;
+	int				error;
 
 	error = xfs_attr_node_hasname(args, state);
 	if (error != -EEXIST)
 		return error;
+	error = 0;
 
 	ASSERT((*state)->path.blk[(*state)->path.active - 1].bp != NULL);
 	ASSERT((*state)->path.blk[(*state)->path.active - 1].magic ==
@@ -1206,97 +1327,164 @@ int xfs_attr_node_removename_setup(
 	if (args->rmtblkno > 0) {
 		error = xfs_attr_leaf_mark_incomplete(args, *state);
 		if (error)
-			return error;
+			goto out;
 
-		return xfs_attr_rmtval_invalidate(args);
+		error = xfs_attr_rmtval_invalidate(args);
 	}
+out:
+	if (error)
+		xfs_da_state_free(*state);
 
-	return 0;
+	return error;
 }
 
 STATIC int
-xfs_attr_node_remove_rmt(
+xfs_attr_node_removename(
 	struct xfs_da_args	*args,
 	struct xfs_da_state	*state)
 {
-	int			error = 0;
-
-	error = xfs_attr_rmtval_remove(args);
-	if (error)
-		return error;
+	struct xfs_da_state_blk	*blk;
+	int			retval;
 
 	/*
-	 * Refill the state structure with buffers, the prior calls released our
-	 * buffers.
+	 * Remove the name and update the hashvals in the tree.
 	 */
-	return xfs_attr_refillstate(state);
+	blk = &state->path.blk[state->path.active-1];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+	retval = xfs_attr3_leaf_remove(blk->bp, args);
+	xfs_da3_fixhashpath(state, &state->path);
+
+	return retval;
 }
 
 /*
- * Remove a name from a B-tree attribute list.
+ * Remove the attribute specified in @args.
  *
  * This will involve walking down the Btree, and may involve joining
  * leaf nodes and even joining intermediate nodes up to and including
  * the root node (a special case of an intermediate node).
+ *
+ * This routine is meant to function as either an in-line or delayed operation,
+ * and may return -EAGAIN when the transaction needs to be rolled.  Calling
+ * functions will need to handle this, and call the function until a
+ * successful error code is returned.
  */
-STATIC int
-xfs_attr_node_removename(
-	struct xfs_da_args	*args)
+int
+xfs_attr_remove_iter(
+	struct xfs_delattr_context	*dac)
 {
-	struct xfs_da_state	*state;
-	struct xfs_da_state_blk	*blk;
-	int			retval, error;
-	struct xfs_inode	*dp = args->dp;
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_da_state		*state = dac->da_state;
+	int				retval, error = 0;
+	struct xfs_inode		*dp = args->dp;
 
 	trace_xfs_attr_node_removename(args);
 
-	error = xfs_attr_node_removename_setup(args, &state);
-	if (error)
-		goto out;
+	switch (dac->dela_state) {
+	case XFS_DAS_UNINIT:
+		if (!xfs_inode_hasattr(dp))
+			return -ENOATTR;
 
-	/*
-	 * If there is an out-of-line value, de-allocate the blocks.
-	 * This is done before we remove the attribute so that we don't
-	 * overflow the maximum size of a transaction and/or hit a deadlock.
-	 */
-	if (args->rmtblkno > 0) {
-		error = xfs_attr_node_remove_rmt(args, state);
-		if (error)
-			goto out;
-	}
+		/*
+		 * Shortform or leaf formats don't require transaction rolls and
+		 * thus state transitions. Call the right helper and return.
+		 */
+		if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL)
+			return xfs_attr_sf_removename(args);
 
-	/*
-	 * Remove the name and update the hashvals in the tree.
-	 */
-	blk = &state->path.blk[ state->path.active-1 ];
-	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
-	retval = xfs_attr3_leaf_remove(blk->bp, args);
-	xfs_da3_fixhashpath(state, &state->path);
+		if (xfs_attr_is_leaf(dp))
+			return xfs_attr_leaf_removename(args);
 
-	/*
-	 * Check to see if the tree needs to be collapsed.
-	 */
-	if (retval && (state->path.active > 1)) {
-		error = xfs_da3_join(state);
-		if (error)
-			goto out;
-		error = xfs_defer_finish(&args->trans);
-		if (error)
-			goto out;
 		/*
-		 * Commit the Btree join operation and start a new trans.
+		 * Node format may require transaction rolls. Set up the
+		 * state context and fall into the state machine.
 		 */
-		error = xfs_trans_roll_inode(&args->trans, dp);
-		if (error)
-			goto out;
-	}
+		if (!dac->da_state) {
+			error = xfs_attr_node_removename_setup(dac);
+			if (error)
+				return error;
+			state = dac->da_state;
+		}
 
-	/*
-	 * If the result is small enough, push it all into the inode.
-	 */
-	if (xfs_attr_is_leaf(dp))
-		error = xfs_attr_node_shrink(args, state);
+		/* fallthrough */
+	case XFS_DAS_RMTBLK:
+		dac->dela_state = XFS_DAS_RMTBLK;
 
+		/*
+		 * If there is an out-of-line value, de-allocate the blocks.
+		 * This is done before we remove the attribute so that we don't
+		 * overflow the maximum size of a transaction and/or hit a
+		 * deadlock.
+		 */
+		if (args->rmtblkno > 0) {
+			/*
+			 * May return -EAGAIN. Roll and repeat until all remote
+			 * blocks are removed.
+			 */
+			error = __xfs_attr_rmtval_remove(dac);
+			if (error == -EAGAIN)
+				return error;
+			else if (error)
+				goto out;
+
+			/*
+			 * Refill the state structure with buffers (the prior
+			 * calls released our buffers) and close out this
+			 * transaction before proceeding.
+			 */
+			ASSERT(args->rmtblkno == 0);
+			error = xfs_attr_refillstate(state);
+			if (error)
+				goto out;
+			dac->dela_state = XFS_DAS_RM_NAME;
+			dac->flags |= XFS_DAC_DEFER_FINISH;
+			return -EAGAIN;
+		}
+
+		/* fallthrough */
+	case XFS_DAS_RM_NAME:
+		/*
+		 * If we came here fresh from a transaction roll, reattach all
+		 * the buffers to the current transaction.
+		 */
+		if (dac->dela_state == XFS_DAS_RM_NAME) {
+			error = xfs_attr_refillstate(state);
+			if (error)
+				goto out;
+		}
+
+		retval = xfs_attr_node_removename(args, state);
+
+		/*
+		 * Check to see if the tree needs to be collapsed. If so, roll
+		 * the transacton and fall into the shrink state.
+		 */
+		if (retval && (state->path.active > 1)) {
+			error = xfs_da3_join(state);
+			if (error)
+				goto out;
+
+			dac->flags |= XFS_DAC_DEFER_FINISH;
+			dac->dela_state = XFS_DAS_RM_SHRINK;
+			return -EAGAIN;
+		}
+
+		/* fallthrough */
+	case XFS_DAS_RM_SHRINK:
+		/*
+		 * If the result is small enough, push it all into the inode.
+		 * This is our final state so it's safe to return a dirty
+		 * transaction.
+		 */
+		if (xfs_attr_is_leaf(dp))
+			error = xfs_attr_node_shrink(args, state);
+		ASSERT(error != -EAGAIN);
+		break;
+	default:
+		ASSERT(0);
+		error = -EINVAL;
+		goto out;
+	}
 out:
 	if (state)
 		xfs_da_state_free(state);
diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
index 2b1f61987a9d..8de5d1d2733e 100644
--- a/fs/xfs/libxfs/xfs_attr.h
+++ b/fs/xfs/libxfs/xfs_attr.h
@@ -74,6 +74,406 @@ struct xfs_attr_list_context {
 };
 
 
+/*
+ * ========================================================================
+ * Structure used to pass context around among the delayed routines.
+ * ========================================================================
+ */
+
+/*
+ * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
+ * states indicate places where the function would return -EAGAIN, and then
+ * immediately resume from after being called by the calling function. States
+ * marked as a "subroutine state" indicate that they belong to a subroutine, and
+ * so the calling function needs to pass them back to that subroutine to allow
+ * it to finish where it left off. But they otherwise do not have a role in the
+ * calling function other than just passing through.
+ *
+ * xfs_attr_remove_iter()
+ *              │
+ *              v
+ *        have attr to remove? ──n──> done
+ *              │
+ *              y
+ *              │
+ *              v
+ *        are we short form? ──y──> xfs_attr_shortform_remove ──> done
+ *              │
+ *              n
+ *              │
+ *              V
+ *        are we leaf form? ──y──> xfs_attr_leaf_removename ──> done
+ *              │
+ *              n
+ *              │
+ *              V
+ *   ┌── need to setup state?
+ *   │          │
+ *   n          y
+ *   │          │
+ *   │          v
+ *   │ find attr and get state
+ *   │ attr has remote blks? ──n─┐
+ *   │          │                v
+ *   │          │         find and invalidate
+ *   │          y         the remote blocks.
+ *   │          │         mark attr incomplete
+ *   │          ├────────────────┘
+ *   └──────────┤
+ *              │
+ *              v
+ *   Have remote blks to remove? ───y─────┐
+ *              │        ^          remove the blks
+ *              │        │                │
+ *              │        │                v
+ *              │  XFS_DAS_RMTBLK <─n── done?
+ *              │  re-enter with          │
+ *              │  one less blk to        y
+ *              │      remove             │
+ *              │                         V
+ *              │                  refill the state
+ *              n                         │
+ *              │                         v
+ *              │                   XFS_DAS_RM_NAME
+ *              │                         │
+ *              ├─────────────────────────┘
+ *              │
+ *              v
+ *       remove leaf and
+ *       update hash with
+ *   xfs_attr_node_remove_cleanup
+ *              │
+ *              v
+ *           need to
+ *        shrink tree? ─n─┐
+ *              │         │
+ *              y         │
+ *              │         │
+ *              v         │
+ *          join leaf     │
+ *              │         │
+ *              v         │
+ *      XFS_DAS_RM_SHRINK │
+ *              │         │
+ *              v         │
+ *       do the shrink    │
+ *              │         │
+ *              v         │
+ *          free state <──┘
+ *              │
+ *              v
+ *            done
+ *
+ *
+ * Below is a state machine diagram for attr set operations.
+ *
+ * It seems the challenge with understanding this system comes from trying to
+ * absorb the state machine all at once, when really one should only be looking
+ * at it with in the context of a single function. Once a state sensitive
+ * function is called, the idea is that it "takes ownership" of the
+ * state machine. It isn't concerned with the states that may have belonged to
+ * it's calling parent. Only the states relevant to itself or any other
+ * subroutines there in. Once a calling function hands off the state machine to
+ * a subroutine, it needs to respect the simple rule that it doesn't "own" the
+ * state machine anymore, and it's the responsibility of that calling function
+ * to propagate the -EAGAIN back up the call stack. Upon reentry, it is
+ * committed to re-calling that subroutine until it returns something other than
+ * -EAGAIN. Once that subroutine signals completion (by returning anything other
+ * than -EAGAIN), the calling function can resume using the state machine.
+ *
+ *  xfs_attr_set_iter()
+ *              │
+ *              v
+ *   ┌─y─ has an attr fork?
+ *   │          |
+ *   │          n
+ *   │          |
+ *   │          V
+ *   │       add a fork
+ *   │          │
+ *   └──────────┤
+ *              │
+ *              V
+ *   ┌─── is shortform?
+ *   │          │
+ *   │          y
+ *   │          │
+ *   │          V
+ *   │   xfs_attr_set_fmt
+ *   │          |
+ *   │          V
+ *   │ xfs_attr_try_sf_addname
+ *   │          │
+ *   │          V
+ *   │      had enough ──y──> done
+ *   │        space?
+ *   n          │
+ *   │          n
+ *   │          │
+ *   │          V
+ *   │   transform to leaf
+ *   │          │
+ *   │          V
+ *   │   hold the leaf buffer
+ *   │          │
+ *   │          V
+ *   │     return -EAGAIN
+ *   │      Re-enter in
+ *   │       leaf form
+ *   │
+ *   └─> release leaf buffer
+ *          if needed
+ *              │
+ *              V
+ *   ┌───n── fork has
+ *   │      only 1 blk?
+ *   │          │
+ *   │          y
+ *   │          │
+ *   │          v
+ *   │ xfs_attr_leaf_try_add()
+ *   │          │
+ *   │          v
+ *   │      had enough ──────────────y─────────────┐
+ *   │        space?                               │
+ *   │          │                                  │
+ *   │          n                                  │
+ *   │          │                                  │
+ *   │          v                                  │
+ *   │    return -EAGAIN                           │
+ *   │      re-enter in                            │
+ *   │        node form                            │
+ *   │          │                                  │
+ *   └──────────┤                                  │
+ *              │                                  │
+ *              V                                  │
+ * xfs_attr_node_addname_find_attr                 │
+ *        determines if this                       │
+ *       is create or rename                       │
+ *     find space to store attr                    │
+ *              │                                  │
+ *              v                                  │
+ *     xfs_attr_node_addname                       │
+ *              │                                  │
+ *              v                                  │
+ *   fits in a node leaf? ────n─────┐              │
+ *              │     ^             v              │
+ *              │     │       single leaf node?    │
+ *              │     │         │            │     │
+ *              y     │         y            n     │
+ *              │     │         │            │     │
+ *              v     │         v            v     │
+ *            update  │    grow the leaf  split if │
+ *           hashvals └── return -EAGAIN   needed  │
+ *              │         retry leaf add     │     │
+ *              │           on reentry       │     │
+ *              ├────────────────────────────┘     │
+ *              │                                  │
+ *              v                                  │
+ *         need to alloc                           │
+ *   ┌─y── or flip flag?                           │
+ *   │          │                                  │
+ *   │          n                                  │
+ *   │          │                                  │
+ *   │          v                                  │
+ *   │         done                                │
+ *   │                                             │
+ *   │                                             │
+ *   │         XFS_DAS_FOUND_LBLK <────────────────┘
+ *   │                  │
+ *   │                  V
+ *   │        xfs_attr_leaf_addname()
+ *   │                  │
+ *   │                  v
+ *   │      ┌──first time through?
+ *   │      │          │
+ *   │      │          y
+ *   │      │          │
+ *   │      n          v
+ *   │      │    if we have rmt blks
+ *   │      │    find space for them
+ *   │      │          │
+ *   │      └──────────┤
+ *   │                 │
+ *   │                 v
+ *   │            still have
+ *   │      ┌─n─ blks to alloc? <──┐
+ *   │      │          │           │
+ *   │      │          y           │
+ *   │      │          │           │
+ *   │      │          v           │
+ *   │      │     alloc one blk    │
+ *   │      │     return -EAGAIN ──┘
+ *   │      │    re-enter with one
+ *   │      │    less blk to alloc
+ *   │      │
+ *   │      │
+ *   │      └───> set the rmt
+ *   │               value
+ *   │                 │
+ *   │                 v
+ *   │               was this
+ *   │              a rename? ──n─┐
+ *   │                 │          │
+ *   │                 y          │
+ *   │                 │          │
+ *   │                 v          │
+ *   │           flip incomplete  │
+ *   │               flag         │
+ *   │                 │          │
+ *   │                 v          │
+ *   │         XFS_DAS_FLIP_LFLAG │
+ *   │                 │          │
+ *   │                 v          │
+ *   │          need to remove    │
+ *   │              old bks? ──n──┤
+ *   │                 │          │
+ *   │                 y          │
+ *   │                 │          │
+ *   │                 V          │
+ *   │               remove       │
+ *   │        ┌───> old blks      │
+ *   │        │        │          │
+ *   │ XFS_DAS_RM_LBLK │          │
+ *   │        ^        │          │
+ *   │        │        v          │
+ *   │        └──y── more to      │
+ *   │              remove?       │
+ *   │                 │          │
+ *   │                 n          │
+ *   │                 │          │
+ *   │                 v          │
+ *   │          XFS_DAS_RD_LEAF   │
+ *   │                 │          │
+ *   │                 v          │
+ *   │            remove leaf     │
+ *   │                 │          │
+ *   │                 v          │
+ *   │            shrink to sf    │
+ *   │             if needed      │
+ *   │                 │          │
+ *   │                 v          │
+ *   │                done <──────┘
+ *   │
+ *   └──────> XFS_DAS_FOUND_NBLK
+ *                     │
+ *                     v
+ *       ┌─────n──  need to
+ *       │        alloc blks?
+ *       │             │
+ *       │             y
+ *       │             │
+ *       │             v
+ *       │        find space
+ *       │             │
+ *       │             v
+ *       │  ┌─>XFS_DAS_ALLOC_NODE
+ *       │  │          │
+ *       │  │          v
+ *       │  │      alloc blk
+ *       │  │          │
+ *       │  │          v
+ *       │  └──y── need to alloc
+ *       │         more blocks?
+ *       │             │
+ *       │             n
+ *       │             │
+ *       │             v
+ *       │      set the rmt value
+ *       │             │
+ *       │             v
+ *       │          was this
+ *       └────────> a rename? ──n─┐
+ *                     │          │
+ *                     y          │
+ *                     │          │
+ *                     v          │
+ *               flip incomplete  │
+ *                   flag         │
+ *                     │          │
+ *                     v          │
+ *             XFS_DAS_FLIP_NFLAG │
+ *                     │          │
+ *                     v          │
+ *                 need to        │
+ *               remove blks? ─n──┤
+ *                     │          │
+ *                     y          │
+ *                     │          │
+ *                     v          │
+ *                   remove       │
+ *        ┌────────> old blks     │
+ *        │            │          │
+ *  XFS_DAS_RM_NBLK    │          │
+ *        ^            │          │
+ *        │            v          │
+ *        └──────y── more to      │
+ *                   remove       │
+ *                     │          │
+ *                     n          │
+ *                     │          │
+ *                     v          │
+ *              XFS_DAS_CLR_FLAG  │
+ *                     │          │
+ *                     v          │
+ *                clear flags     │
+ *                     │          │
+ *                     ├──────────┘
+ *                     │
+ *                     v
+ *                   done
+ */
+
+/*
+ * Enum values for xfs_delattr_context.da_state
+ *
+ * These values are used by delayed attribute operations to keep track  of where
+ * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
+ * calling function to roll the transaction, and then call the subroutine to
+ * finish the operation.  The enum is then used by the subroutine to jump back
+ * to where it was and resume executing where it left off.
+ */
+enum xfs_delattr_state {
+	XFS_DAS_UNINIT		= 0,  /* No state has been set yet */
+	XFS_DAS_RMTBLK,		      /* Removing remote blks */
+	XFS_DAS_RM_NAME,	      /* Remove attr name */
+	XFS_DAS_RM_SHRINK,	      /* We are shrinking the tree */
+	XFS_DAS_FOUND_LBLK,	      /* We found leaf blk for attr */
+	XFS_DAS_FOUND_NBLK,	      /* We found node blk for attr */
+	XFS_DAS_FLIP_LFLAG,	      /* Flipped leaf INCOMPLETE attr flag */
+	XFS_DAS_RM_LBLK,	      /* A rename is removing leaf blocks */
+	XFS_DAS_RD_LEAF,	      /* Read in the new leaf */
+	XFS_DAS_ALLOC_NODE,	      /* We are allocating node blocks */
+	XFS_DAS_FLIP_NFLAG,	      /* Flipped node INCOMPLETE attr flag */
+	XFS_DAS_RM_NBLK,	      /* A rename is removing node blocks */
+	XFS_DAS_CLR_FLAG,	      /* Clear incomplete flag */
+};
+
+/*
+ * Defines for xfs_delattr_context.flags
+ */
+#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
+#define XFS_DAC_LEAF_ADDNAME_INIT	0x02 /* xfs_attr_leaf_addname init*/
+
+/*
+ * Context used for keeping track of delayed attribute operations
+ */
+struct xfs_delattr_context {
+	struct xfs_da_args      *da_args;
+
+	/* Used in xfs_attr_rmtval_set_blk to roll through allocating blocks */
+	struct xfs_bmbt_irec	map;
+	xfs_dablk_t		lblkno;
+	int			blkcnt;
+
+	/* Used in xfs_attr_node_removename to roll through removing blocks */
+	struct xfs_da_state     *da_state;
+
+	/* Used to keep track of current state of delayed operation */
+	unsigned int            flags;
+	enum xfs_delattr_state  dela_state;
+};
+
 /*========================================================================
  * Function prototypes for the kernel.
  *========================================================================*/
@@ -92,6 +492,9 @@ int xfs_attr_set(struct xfs_da_args *args);
 int xfs_attr_set_args(struct xfs_da_args *args);
 int xfs_has_attr(struct xfs_da_args *args);
 int xfs_attr_remove_args(struct xfs_da_args *args);
+int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
 bool xfs_attr_namecheck(const void *name, size_t length);
+void xfs_delattr_context_init(struct xfs_delattr_context *dac,
+			      struct xfs_da_args *args);
 
 #endif	/* __XFS_ATTR_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
index 556184b63061..b910bd209949 100644
--- a/fs/xfs/libxfs/xfs_attr_leaf.c
+++ b/fs/xfs/libxfs/xfs_attr_leaf.c
@@ -19,14 +19,15 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_bmap.h"
 #include "xfs_attr_sf.h"
-#include "xfs_attr_remote.h"
 #include "xfs_attr.h"
+#include "xfs_attr_remote.h"
 #include "xfs_attr_leaf.h"
 #include "xfs_error.h"
 #include "xfs_trace.h"
 #include "xfs_buf_item.h"
 #include "xfs_dir2.h"
 #include "xfs_log.h"
+#include "xfs_ag.h"
 
 
 /*
@@ -773,7 +774,7 @@ xfs_attr_fork_remove(
  * Remove an attribute from the shortform attribute list structure.
  */
 int
-xfs_attr_shortform_remove(
+xfs_attr_sf_removename(
 	struct xfs_da_args		*args)
 {
 	struct xfs_attr_shortform	*sf;
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.h b/fs/xfs/libxfs/xfs_attr_leaf.h
index 9b1c59f40a26..efa757f1e912 100644
--- a/fs/xfs/libxfs/xfs_attr_leaf.h
+++ b/fs/xfs/libxfs/xfs_attr_leaf.h
@@ -51,7 +51,7 @@ int	xfs_attr_shortform_lookup(struct xfs_da_args *args);
 int	xfs_attr_shortform_getvalue(struct xfs_da_args *args);
 int	xfs_attr_shortform_to_leaf(struct xfs_da_args *args,
 			struct xfs_buf **leaf_bp);
-int	xfs_attr_shortform_remove(struct xfs_da_args *args);
+int	xfs_attr_sf_removename(struct xfs_da_args *args);
 int	xfs_attr_sf_findname(struct xfs_da_args *args,
 			     struct xfs_attr_sf_entry **sfep,
 			     unsigned int *basep);
diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
index 48d8e9caf86f..0c8bee3abc3b 100644
--- a/fs/xfs/libxfs/xfs_attr_remote.c
+++ b/fs/xfs/libxfs/xfs_attr_remote.c
@@ -439,9 +439,9 @@ xfs_attr_rmtval_get(
 
 /*
  * Find a "hole" in the attribute address space large enough for us to drop the
- * new attribute's value into
+ * new attributes value into
  */
-STATIC int
+int
 xfs_attr_rmt_find_hole(
 	struct xfs_da_args	*args)
 {
@@ -468,7 +468,7 @@ xfs_attr_rmt_find_hole(
 	return 0;
 }
 
-STATIC int
+int
 xfs_attr_rmtval_set_value(
 	struct xfs_da_args	*args)
 {
@@ -562,69 +562,66 @@ xfs_attr_rmtval_stale(
 }
 
 /*
- * Write the value associated with an attribute into the out-of-line buffer
- * that we have defined for it.
+ * Find a hole for the attr and store it in the delayed attr context.  This
+ * initializes the context to roll through allocating an attr extent for a
+ * delayed attr operation
  */
 int
-xfs_attr_rmtval_set(
-	struct xfs_da_args	*args)
+xfs_attr_rmtval_find_space(
+	struct xfs_delattr_context	*dac)
 {
-	struct xfs_inode	*dp = args->dp;
-	struct xfs_bmbt_irec	map;
-	xfs_dablk_t		lblkno;
-	int			blkcnt;
-	int			nmap;
-	int			error;
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_bmbt_irec		*map = &dac->map;
+	int				error;
 
-	trace_xfs_attr_rmtval_set(args);
+	dac->lblkno = 0;
+	dac->blkcnt = 0;
+	args->rmtblkcnt = 0;
+	args->rmtblkno = 0;
+	memset(map, 0, sizeof(struct xfs_bmbt_irec));
 
 	error = xfs_attr_rmt_find_hole(args);
 	if (error)
 		return error;
 
-	blkcnt = args->rmtblkcnt;
-	lblkno = (xfs_dablk_t)args->rmtblkno;
-	/*
-	 * Roll through the "value", allocating blocks on disk as required.
-	 */
-	while (blkcnt > 0) {
-		/*
-		 * Allocate a single extent, up to the size of the value.
-		 *
-		 * Note that we have to consider this a data allocation as we
-		 * write the remote attribute without logging the contents.
-		 * Hence we must ensure that we aren't using blocks that are on
-		 * the busy list so that we don't overwrite blocks which have
-		 * recently been freed but their transactions are not yet
-		 * committed to disk. If we overwrite the contents of a busy
-		 * extent and then crash then the block may not contain the
-		 * correct metadata after log recovery occurs.
-		 */
-		nmap = 1;
-		error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
-				  blkcnt, XFS_BMAPI_ATTRFORK, args->total, &map,
-				  &nmap);
-		if (error)
-			return error;
-		error = xfs_defer_finish(&args->trans);
-		if (error)
-			return error;
+	dac->blkcnt = args->rmtblkcnt;
+	dac->lblkno = args->rmtblkno;
 
-		ASSERT(nmap == 1);
-		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
-		       (map.br_startblock != HOLESTARTBLOCK));
-		lblkno += map.br_blockcount;
-		blkcnt -= map.br_blockcount;
+	return 0;
+}
 
-		/*
-		 * Start the next trans in the chain.
-		 */
-		error = xfs_trans_roll_inode(&args->trans, dp);
-		if (error)
-			return error;
-	}
+/*
+ * Write one block of the value associated with an attribute into the
+ * out-of-line buffer that we have defined for it. This is similar to a subset
+ * of xfs_attr_rmtval_set, but records the current block to the delayed attr
+ * context, and leaves transaction handling to the caller.
+ */
+int
+xfs_attr_rmtval_set_blk(
+	struct xfs_delattr_context	*dac)
+{
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_inode		*dp = args->dp;
+	struct xfs_bmbt_irec		*map = &dac->map;
+	int nmap;
+	int error;
+
+	nmap = 1;
+	error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)dac->lblkno,
+			dac->blkcnt, XFS_BMAPI_ATTRFORK, args->total,
+			map, &nmap);
+	if (error)
+		return error;
+
+	ASSERT(nmap == 1);
+	ASSERT((map->br_startblock != DELAYSTARTBLOCK) &&
+	       (map->br_startblock != HOLESTARTBLOCK));
 
-	return xfs_attr_rmtval_set_value(args);
+	/* roll attribute extent map forwards */
+	dac->lblkno += map->br_blockcount;
+	dac->blkcnt -= map->br_blockcount;
+
+	return 0;
 }
 
 /*
@@ -669,47 +666,17 @@ xfs_attr_rmtval_invalidate(
 }
 
 /*
- * Remove the value associated with an attribute by deleting the
- * out-of-line buffer that it is stored on.
- */
-int
-xfs_attr_rmtval_remove(
-	struct xfs_da_args      *args)
-{
-	int			error;
-	int			retval;
-
-	trace_xfs_attr_rmtval_remove(args);
-
-	/*
-	 * Keep de-allocating extents until the remote-value region is gone.
-	 */
-	do {
-		retval = __xfs_attr_rmtval_remove(args);
-		if (retval && retval != -EAGAIN)
-			return retval;
-
-		/*
-		 * Close out trans and start the next one in the chain.
-		 */
-		error = xfs_trans_roll_inode(&args->trans, args->dp);
-		if (error)
-			return error;
-	} while (retval == -EAGAIN);
-
-	return 0;
-}
-
-/*
  * Remove the value associated with an attribute by deleting the out-of-line
- * buffer that it is stored on. Returns EAGAIN for the caller to refresh the
- * transaction and re-call the function
+ * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the
+ * transaction and re-call the function.  Callers should keep calling this
+ * routine until it returns something other than -EAGAIN.
  */
 int
 __xfs_attr_rmtval_remove(
-	struct xfs_da_args	*args)
+	struct xfs_delattr_context	*dac)
 {
-	int			error, done;
+	struct xfs_da_args		*args = dac->da_args;
+	int				error, done;
 
 	/*
 	 * Unmap value blocks for this attr.
@@ -719,12 +686,20 @@ __xfs_attr_rmtval_remove(
 	if (error)
 		return error;
 
-	error = xfs_defer_finish(&args->trans);
-	if (error)
-		return error;
-
-	if (!done)
+	/*
+	 * We don't need an explicit state here to pick up where we left off. We
+	 * can figure it out using the !done return code. The actual value of
+	 * attr->xattri_dela_state may be some value reminiscent of the calling
+	 * function, but it's value is irrelevant with in the context of this
+	 * function. Once we are done here, the next state is set as needed by
+	 * the parent
+	 */
+	if (!done) {
+		dac->flags |= XFS_DAC_DEFER_FINISH;
 		return -EAGAIN;
+	}
 
-	return error;
+	args->rmtblkno = 0;
+	args->rmtblkcnt = 0;
+	return 0;
 }
diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
index 9eee615da156..61b85b918db8 100644
--- a/fs/xfs/libxfs/xfs_attr_remote.h
+++ b/fs/xfs/libxfs/xfs_attr_remote.h
@@ -9,10 +9,12 @@
 int xfs_attr3_rmt_blocks(struct xfs_mount *mp, int attrlen);
 
 int xfs_attr_rmtval_get(struct xfs_da_args *args);
-int xfs_attr_rmtval_set(struct xfs_da_args *args);
-int xfs_attr_rmtval_remove(struct xfs_da_args *args);
 int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
 		xfs_buf_flags_t incore_flags);
 int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
-int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
+int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
+int xfs_attr_rmt_find_hole(struct xfs_da_args *args);
+int xfs_attr_rmtval_set_value(struct xfs_da_args *args);
+int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac);
+int xfs_attr_rmtval_find_space(struct xfs_delattr_context *dac);
 #endif /* __XFS_ATTR_REMOTE_H__ */
diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
index a3e0e6f672d6..948092babb6a 100644
--- a/fs/xfs/libxfs/xfs_bmap.c
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -31,6 +31,7 @@
 #include "xfs_attr_leaf.h"
 #include "xfs_filestream.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 #include "xfs_refcount.h"
 #include "xfs_icache.h"
@@ -1028,7 +1029,7 @@ xfs_bmap_add_attrfork_local(
 /*
  * Set an inode attr fork offset based on the format of the data fork.
  */
-int
+static int
 xfs_bmap_set_attrforkoff(
 	struct xfs_inode	*ip,
 	int			size,
diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
index f9a390ecfb1d..67641f669918 100644
--- a/fs/xfs/libxfs/xfs_bmap.h
+++ b/fs/xfs/libxfs/xfs_bmap.h
@@ -187,7 +187,6 @@ void	xfs_trim_extent(struct xfs_bmbt_irec *irec, xfs_fileoff_t bno,
 		xfs_filblks_t len);
 unsigned int xfs_bmap_compute_attr_offset(struct xfs_mount *mp);
 int	xfs_bmap_add_attrfork(struct xfs_inode *ip, int size, int rsvd);
-int	xfs_bmap_set_attrforkoff(struct xfs_inode *ip, int size, int *version);
 void	xfs_bmap_local_to_extents_empty(struct xfs_trans *tp,
 		struct xfs_inode *ip, int whichfork);
 void	__xfs_bmap_add_free(struct xfs_trans *tp, xfs_fsblock_t bno,
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index 5b6fcb9b44e2..be74a6b53689 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -21,6 +21,7 @@
 #include "xfs_alloc.h"
 #include "xfs_log.h"
 #include "xfs_btree_staging.h"
+#include "xfs_ag.h"
 
 /*
  * Cursor allocation zone.
@@ -215,7 +216,7 @@ xfs_btree_check_sptr(
 {
 	if (level <= 0)
 		return false;
-	return xfs_verify_agbno(cur->bc_mp, cur->bc_ag.agno, agbno);
+	return xfs_verify_agbno(cur->bc_mp, cur->bc_ag.pag->pag_agno, agbno);
 }
 
 /*
@@ -244,7 +245,7 @@ xfs_btree_check_ptr(
 			return 0;
 		xfs_err(cur->bc_mp,
 "AG %u: Corrupt btree %d pointer at level %d index %d.",
-				cur->bc_ag.agno, cur->bc_btnum,
+				cur->bc_ag.pag->pag_agno, cur->bc_btnum,
 				level, index);
 	}
 
@@ -376,6 +377,8 @@ xfs_btree_del_cursor(
 	       XFS_FORCED_SHUTDOWN(cur->bc_mp));
 	if (unlikely(cur->bc_flags & XFS_BTREE_STAGING))
 		kmem_free(cur->bc_ops);
+	if (!(cur->bc_flags & XFS_BTREE_LONG_PTRS) && cur->bc_ag.pag)
+		xfs_perag_put(cur->bc_ag.pag);
 	kmem_cache_free(xfs_btree_cur_zone, cur);
 }
 
@@ -885,13 +888,13 @@ xfs_btree_readahead_sblock(
 
 
 	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
-		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.agno,
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				     left, 1, cur->bc_ops->buf_ops);
 		rval++;
 	}
 
 	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
-		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.agno,
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				     right, 1, cur->bc_ops->buf_ops);
 		rval++;
 	}
@@ -949,7 +952,7 @@ xfs_btree_ptr_to_daddr(
 		*daddr = XFS_FSB_TO_DADDR(cur->bc_mp, fsbno);
 	} else {
 		agbno = be32_to_cpu(ptr->s);
-		*daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_ag.agno,
+		*daddr = XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				agbno);
 	}
 
@@ -1150,7 +1153,7 @@ xfs_btree_init_block_cur(
 	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 		owner = cur->bc_ino.ip->i_ino;
 	else
-		owner = cur->bc_ag.agno;
+		owner = cur->bc_ag.pag->pag_agno;
 
 	xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
 				 cur->bc_btnum, level, numrecs,
diff --git a/fs/xfs/libxfs/xfs_btree.h b/fs/xfs/libxfs/xfs_btree.h
index 10e50cbacacf..4dbdc659c396 100644
--- a/fs/xfs/libxfs/xfs_btree.h
+++ b/fs/xfs/libxfs/xfs_btree.h
@@ -11,6 +11,7 @@ struct xfs_inode;
 struct xfs_mount;
 struct xfs_trans;
 struct xfs_ifork;
+struct xfs_perag;
 
 extern kmem_zone_t	*xfs_btree_cur_zone;
 
@@ -180,11 +181,11 @@ union xfs_btree_irec {
 
 /* Per-AG btree information. */
 struct xfs_btree_cur_ag {
+	struct xfs_perag	*pag;
 	union {
 		struct xfs_buf		*agbp;
 		struct xbtree_afakeroot	*afake;	/* for staging cursor */
 	};
-	xfs_agnumber_t		agno;
 	union {
 		struct {
 			unsigned long nr_ops;	/* # record updates */
@@ -231,6 +232,13 @@ typedef struct xfs_btree_cur
 	uint8_t		bc_blocklog;	/* log2(blocksize) of btree blocks */
 	xfs_btnum_t	bc_btnum;	/* identifies which btree type */
 	int		bc_statoff;	/* offset of btre stats array */
+
+	/*
+	 * Short btree pointers need an agno to be able to turn the pointers
+	 * into physical addresses for IO, so the btree cursor switches between
+	 * bc_ino and bc_ag based on whether XFS_BTREE_LONG_PTRS is set for the
+	 * cursor.
+	 */
 	union {
 		struct xfs_btree_cur_ag	bc_ag;
 		struct xfs_btree_cur_ino bc_ino;
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index eefdb518fe64..57d9cb632983 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -10,7 +10,6 @@
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
 #include "xfs_bit.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_inode.h"
 #include "xfs_btree.h"
@@ -27,6 +26,7 @@
 #include "xfs_trace.h"
 #include "xfs_log.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 
 /*
  * Lookup a record by ino in the btree given by cur.
@@ -105,7 +105,7 @@ xfs_inobt_get_rec(
 	int				*stat)
 {
 	struct xfs_mount		*mp = cur->bc_mp;
-	xfs_agnumber_t			agno = cur->bc_ag.agno;
+	xfs_agnumber_t			agno = cur->bc_ag.pag->pag_agno;
 	union xfs_btree_rec		*rec;
 	int				error;
 	uint64_t			realfree;
@@ -172,18 +172,17 @@ xfs_inobt_insert(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
 	xfs_agino_t		newino,
 	xfs_agino_t		newlen,
 	xfs_btnum_t		btnum)
 {
 	struct xfs_btree_cur	*cur;
-	struct xfs_agi		*agi = agbp->b_addr;
-	xfs_agnumber_t		agno = be32_to_cpu(agi->agi_seqno);
 	xfs_agino_t		thisino;
 	int			i;
 	int			error;
 
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, btnum);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, btnum);
 
 	for (thisino = newino;
 	     thisino < newino + newlen;
@@ -215,10 +214,9 @@ xfs_inobt_insert(
  * Verify that the number of free inodes in the AGI is correct.
  */
 #ifdef DEBUG
-STATIC int
+static int
 xfs_check_agi_freecount(
-	struct xfs_btree_cur	*cur,
-	struct xfs_agi		*agi)
+	struct xfs_btree_cur	*cur)
 {
 	if (cur->bc_nlevels == 1) {
 		xfs_inobt_rec_incore_t rec;
@@ -244,12 +242,12 @@ xfs_check_agi_freecount(
 		} while (i == 1);
 
 		if (!XFS_FORCED_SHUTDOWN(cur->bc_mp))
-			ASSERT(freecount == be32_to_cpu(agi->agi_freecount));
+			ASSERT(freecount == cur->bc_ag.pag->pagi_freecount);
 	}
 	return 0;
 }
 #else
-#define xfs_check_agi_freecount(cur, agi)	0
+#define xfs_check_agi_freecount(cur)	0
 #endif
 
 /*
@@ -520,18 +518,17 @@ xfs_inobt_insert_sprec(
 	struct xfs_mount		*mp,
 	struct xfs_trans		*tp,
 	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
 	int				btnum,
 	struct xfs_inobt_rec_incore	*nrec,	/* in/out: new/merged rec. */
 	bool				merge)	/* merge or replace */
 {
 	struct xfs_btree_cur		*cur;
-	struct xfs_agi			*agi = agbp->b_addr;
-	xfs_agnumber_t			agno = be32_to_cpu(agi->agi_seqno);
 	int				error;
 	int				i;
 	struct xfs_inobt_rec_incore	rec;
 
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, btnum);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, btnum);
 
 	/* the new record is pre-aligned so we know where to look */
 	error = xfs_inobt_lookup(cur, nrec->ir_startino, XFS_LOOKUP_EQ, &i);
@@ -578,14 +575,14 @@ xfs_inobt_insert_sprec(
 			goto error;
 		}
 
-		trace_xfs_irec_merge_pre(mp, agno, rec.ir_startino,
+		trace_xfs_irec_merge_pre(mp, pag->pag_agno, rec.ir_startino,
 					 rec.ir_holemask, nrec->ir_startino,
 					 nrec->ir_holemask);
 
 		/* merge to nrec to output the updated record */
 		__xfs_inobt_rec_merge(nrec, &rec);
 
-		trace_xfs_irec_merge_post(mp, agno, nrec->ir_startino,
+		trace_xfs_irec_merge_post(mp, pag->pag_agno, nrec->ir_startino,
 					  nrec->ir_holemask);
 
 		error = xfs_inobt_rec_check_count(mp, nrec);
@@ -606,28 +603,28 @@ error:
 }
 
 /*
- * Allocate new inodes in the allocation group specified by agbp.
- * Returns 0 if inodes were allocated in this AG; 1 if there was no space
- * in this AG; or the usual negative error code.
+ * Allocate new inodes in the allocation group specified by agbp.  Returns 0 if
+ * inodes were allocated in this AG; -EAGAIN if there was no space in this AG so
+ * the caller knows it can try another AG, a hard -ENOSPC when over the maximum
+ * inode count threshold, or the usual negative error code for other errors.
  */
 STATIC int
 xfs_ialloc_ag_alloc(
 	struct xfs_trans	*tp,
-	struct xfs_buf		*agbp)
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
 {
 	struct xfs_agi		*agi;
 	struct xfs_alloc_arg	args;
-	xfs_agnumber_t		agno;
 	int			error;
 	xfs_agino_t		newino;		/* new first inode's number */
 	xfs_agino_t		newlen;		/* new number of inodes */
 	int			isaligned = 0;	/* inode allocation at stripe */
 						/* unit boundary */
 	/* init. to full chunk */
-	uint16_t		allocmask = (uint16_t) -1;
 	struct xfs_inobt_rec_incore rec;
-	struct xfs_perag	*pag;
 	struct xfs_ino_geometry	*igeo = M_IGEO(tp->t_mountp);
+	uint16_t		allocmask = (uint16_t) -1;
 	int			do_sparse = 0;
 
 	memset(&args, 0, sizeof(args));
@@ -660,14 +657,13 @@ xfs_ialloc_ag_alloc(
 	 */
 	agi = agbp->b_addr;
 	newino = be32_to_cpu(agi->agi_newino);
-	agno = be32_to_cpu(agi->agi_seqno);
 	args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) +
 		     igeo->ialloc_blks;
 	if (do_sparse)
 		goto sparse_alloc;
 	if (likely(newino != NULLAGINO &&
 		  (args.agbno < be32_to_cpu(agi->agi_length)))) {
-		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
 		args.type = XFS_ALLOCTYPE_THIS_BNO;
 		args.prod = 1;
 
@@ -727,7 +723,7 @@ xfs_ialloc_ag_alloc(
 		 * For now, just allocate blocks up front.
 		 */
 		args.agbno = be32_to_cpu(agi->agi_root);
-		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
 		/*
 		 * Allocate a fixed-size extent of inodes.
 		 */
@@ -748,7 +744,7 @@ xfs_ialloc_ag_alloc(
 	if (isaligned && args.fsbno == NULLFSBLOCK) {
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 		args.agbno = be32_to_cpu(agi->agi_root);
-		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
 		args.alignment = igeo->cluster_align;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
@@ -764,7 +760,7 @@ xfs_ialloc_ag_alloc(
 sparse_alloc:
 		args.type = XFS_ALLOCTYPE_NEAR_BNO;
 		args.agbno = be32_to_cpu(agi->agi_root);
-		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, pag->pag_agno, args.agbno);
 		args.alignment = args.mp->m_sb.sb_spino_align;
 		args.prod = 1;
 
@@ -796,7 +792,7 @@ sparse_alloc:
 	}
 
 	if (args.fsbno == NULLFSBLOCK)
-		return 1;
+		return -EAGAIN;
 
 	ASSERT(args.len == args.minlen);
 
@@ -809,7 +805,7 @@ sparse_alloc:
 	 * rather than a linear progression to prevent the next generation
 	 * number from being easily guessable.
 	 */
-	error = xfs_ialloc_inode_init(args.mp, tp, NULL, newlen, agno,
+	error = xfs_ialloc_inode_init(args.mp, tp, NULL, newlen, pag->pag_agno,
 			args.agbno, args.len, prandom_u32());
 
 	if (error)
@@ -836,12 +832,12 @@ sparse_alloc:
 		 * if necessary. If a merge does occur, rec is updated to the
 		 * merged record.
 		 */
-		error = xfs_inobt_insert_sprec(args.mp, tp, agbp, XFS_BTNUM_INO,
-					       &rec, true);
+		error = xfs_inobt_insert_sprec(args.mp, tp, agbp, pag,
+				XFS_BTNUM_INO, &rec, true);
 		if (error == -EFSCORRUPTED) {
 			xfs_alert(args.mp,
 	"invalid sparse inode record: ino 0x%llx holemask 0x%x count %u",
-				  XFS_AGINO_TO_INO(args.mp, agno,
+				  XFS_AGINO_TO_INO(args.mp, pag->pag_agno,
 						   rec.ir_startino),
 				  rec.ir_holemask, rec.ir_count);
 			xfs_force_shutdown(args.mp, SHUTDOWN_CORRUPT_INCORE);
@@ -861,21 +857,20 @@ sparse_alloc:
 		 * existing record with this one.
 		 */
 		if (xfs_sb_version_hasfinobt(&args.mp->m_sb)) {
-			error = xfs_inobt_insert_sprec(args.mp, tp, agbp,
-						       XFS_BTNUM_FINO, &rec,
-						       false);
+			error = xfs_inobt_insert_sprec(args.mp, tp, agbp, pag,
+				       XFS_BTNUM_FINO, &rec, false);
 			if (error)
 				return error;
 		}
 	} else {
 		/* full chunk - insert new records to both btrees */
-		error = xfs_inobt_insert(args.mp, tp, agbp, newino, newlen,
+		error = xfs_inobt_insert(args.mp, tp, agbp, pag, newino, newlen,
 					 XFS_BTNUM_INO);
 		if (error)
 			return error;
 
 		if (xfs_sb_version_hasfinobt(&args.mp->m_sb)) {
-			error = xfs_inobt_insert(args.mp, tp, agbp, newino,
+			error = xfs_inobt_insert(args.mp, tp, agbp, pag, newino,
 						 newlen, XFS_BTNUM_FINO);
 			if (error)
 				return error;
@@ -887,7 +882,6 @@ sparse_alloc:
 	 */
 	be32_add_cpu(&agi->agi_count, newlen);
 	be32_add_cpu(&agi->agi_freecount, newlen);
-	pag = agbp->b_pag;
 	pag->pagi_freecount += newlen;
 	pag->pagi_count += newlen;
 	agi->agi_newino = cpu_to_be32(newino);
@@ -905,139 +899,6 @@ sparse_alloc:
 	return 0;
 }
 
-STATIC xfs_agnumber_t
-xfs_ialloc_next_ag(
-	xfs_mount_t	*mp)
-{
-	xfs_agnumber_t	agno;
-
-	spin_lock(&mp->m_agirotor_lock);
-	agno = mp->m_agirotor;
-	if (++mp->m_agirotor >= mp->m_maxagi)
-		mp->m_agirotor = 0;
-	spin_unlock(&mp->m_agirotor_lock);
-
-	return agno;
-}
-
-/*
- * Select an allocation group to look for a free inode in, based on the parent
- * inode and the mode.  Return the allocation group buffer.
- */
-STATIC xfs_agnumber_t
-xfs_ialloc_ag_select(
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_ino_t	parent,		/* parent directory inode number */
-	umode_t		mode)		/* bits set to indicate file type */
-{
-	xfs_agnumber_t	agcount;	/* number of ag's in the filesystem */
-	xfs_agnumber_t	agno;		/* current ag number */
-	int		flags;		/* alloc buffer locking flags */
-	xfs_extlen_t	ineed;		/* blocks needed for inode allocation */
-	xfs_extlen_t	longest = 0;	/* longest extent available */
-	xfs_mount_t	*mp;		/* mount point structure */
-	int		needspace;	/* file mode implies space allocated */
-	xfs_perag_t	*pag;		/* per allocation group data */
-	xfs_agnumber_t	pagno;		/* parent (starting) ag number */
-	int		error;
-
-	/*
-	 * Files of these types need at least one block if length > 0
-	 * (and they won't fit in the inode, but that's hard to figure out).
-	 */
-	needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
-	mp = tp->t_mountp;
-	agcount = mp->m_maxagi;
-	if (S_ISDIR(mode))
-		pagno = xfs_ialloc_next_ag(mp);
-	else {
-		pagno = XFS_INO_TO_AGNO(mp, parent);
-		if (pagno >= agcount)
-			pagno = 0;
-	}
-
-	ASSERT(pagno < agcount);
-
-	/*
-	 * Loop through allocation groups, looking for one with a little
-	 * free space in it.  Note we don't look for free inodes, exactly.
-	 * Instead, we include whether there is a need to allocate inodes
-	 * to mean that blocks must be allocated for them,
-	 * if none are currently free.
-	 */
-	agno = pagno;
-	flags = XFS_ALLOC_FLAG_TRYLOCK;
-	for (;;) {
-		pag = xfs_perag_get(mp, agno);
-		if (!pag->pagi_inodeok) {
-			xfs_ialloc_next_ag(mp);
-			goto nextag;
-		}
-
-		if (!pag->pagi_init) {
-			error = xfs_ialloc_pagi_init(mp, tp, agno);
-			if (error)
-				goto nextag;
-		}
-
-		if (pag->pagi_freecount) {
-			xfs_perag_put(pag);
-			return agno;
-		}
-
-		if (!pag->pagf_init) {
-			error = xfs_alloc_pagf_init(mp, tp, agno, flags);
-			if (error)
-				goto nextag;
-		}
-
-		/*
-		 * Check that there is enough free space for the file plus a
-		 * chunk of inodes if we need to allocate some. If this is the
-		 * first pass across the AGs, take into account the potential
-		 * space needed for alignment of inode chunks when checking the
-		 * longest contiguous free space in the AG - this prevents us
-		 * from getting ENOSPC because we have free space larger than
-		 * ialloc_blks but alignment constraints prevent us from using
-		 * it.
-		 *
-		 * If we can't find an AG with space for full alignment slack to
-		 * be taken into account, we must be near ENOSPC in all AGs.
-		 * Hence we don't include alignment for the second pass and so
-		 * if we fail allocation due to alignment issues then it is most
-		 * likely a real ENOSPC condition.
-		 */
-		ineed = M_IGEO(mp)->ialloc_min_blks;
-		if (flags && ineed > 1)
-			ineed += M_IGEO(mp)->cluster_align;
-		longest = pag->pagf_longest;
-		if (!longest)
-			longest = pag->pagf_flcount > 0;
-
-		if (pag->pagf_freeblks >= needspace + ineed &&
-		    longest >= ineed) {
-			xfs_perag_put(pag);
-			return agno;
-		}
-nextag:
-		xfs_perag_put(pag);
-		/*
-		 * No point in iterating over the rest, if we're shutting
-		 * down.
-		 */
-		if (XFS_FORCED_SHUTDOWN(mp))
-			return NULLAGNUMBER;
-		agno++;
-		if (agno >= agcount)
-			agno = 0;
-		if (agno == pagno) {
-			if (flags == 0)
-				return NULLAGNUMBER;
-			flags = 0;
-		}
-	}
-}
-
 /*
  * Try to retrieve the next record to the left/right from the current one.
  */
@@ -1123,15 +984,14 @@ STATIC int
 xfs_dialloc_ag_inobt(
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
 	xfs_ino_t		parent,
 	xfs_ino_t		*inop)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
 	struct xfs_agi		*agi = agbp->b_addr;
-	xfs_agnumber_t		agno = be32_to_cpu(agi->agi_seqno);
 	xfs_agnumber_t		pagno = XFS_INO_TO_AGNO(mp, parent);
 	xfs_agino_t		pagino = XFS_INO_TO_AGINO(mp, parent);
-	struct xfs_perag	*pag = agbp->b_pag;
 	struct xfs_btree_cur	*cur, *tcur;
 	struct xfs_inobt_rec_incore rec, trec;
 	xfs_ino_t		ino;
@@ -1145,7 +1005,7 @@ xfs_dialloc_ag_inobt(
 	ASSERT(pag->pagi_freecount > 0);
 
  restart_pagno:
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
 	/*
 	 * If pagino is 0 (this is the root inode allocation) use newino.
 	 * This must work because we've just allocated some.
@@ -1153,14 +1013,14 @@ xfs_dialloc_ag_inobt(
 	if (!pagino)
 		pagino = be32_to_cpu(agi->agi_newino);
 
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error0;
 
 	/*
 	 * If in the same AG as the parent, try to get near the parent.
 	 */
-	if (pagno == agno) {
+	if (pagno == pag->pag_agno) {
 		int		doneleft;	/* done, to the left */
 		int		doneright;	/* done, to the right */
 
@@ -1363,7 +1223,7 @@ alloc_inode:
 	ASSERT(offset < XFS_INODES_PER_CHUNK);
 	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
 				   XFS_INODES_PER_CHUNK) == 0);
-	ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset);
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, rec.ir_startino + offset);
 	rec.ir_free &= ~XFS_INOBT_MASK(offset);
 	rec.ir_freecount--;
 	error = xfs_inobt_update(cur, &rec);
@@ -1373,7 +1233,7 @@ alloc_inode:
 	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
 	pag->pagi_freecount--;
 
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error0;
 
@@ -1568,16 +1428,16 @@ xfs_dialloc_ag_update_inobt(
  * The caller selected an AG for us, and made sure that free inodes are
  * available.
  */
-int
+static int
 xfs_dialloc_ag(
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag,
 	xfs_ino_t		parent,
 	xfs_ino_t		*inop)
 {
 	struct xfs_mount		*mp = tp->t_mountp;
 	struct xfs_agi			*agi = agbp->b_addr;
-	xfs_agnumber_t			agno = be32_to_cpu(agi->agi_seqno);
 	xfs_agnumber_t			pagno = XFS_INO_TO_AGNO(mp, parent);
 	xfs_agino_t			pagino = XFS_INO_TO_AGINO(mp, parent);
 	struct xfs_btree_cur		*cur;	/* finobt cursor */
@@ -1589,7 +1449,7 @@ xfs_dialloc_ag(
 	int				i;
 
 	if (!xfs_sb_version_hasfinobt(&mp->m_sb))
-		return xfs_dialloc_ag_inobt(tp, agbp, parent, inop);
+		return xfs_dialloc_ag_inobt(tp, agbp, pag, parent, inop);
 
 	/*
 	 * If pagino is 0 (this is the root inode allocation) use newino.
@@ -1598,9 +1458,9 @@ xfs_dialloc_ag(
 	if (!pagino)
 		pagino = be32_to_cpu(agi->agi_newino);
 
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_FINO);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_FINO);
 
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error_cur;
 
@@ -1609,7 +1469,7 @@ xfs_dialloc_ag(
 	 * parent. If so, find the closest available inode to the parent. If
 	 * not, consider the agi hint or find the first free inode in the AG.
 	 */
-	if (agno == pagno)
+	if (pag->pag_agno == pagno)
 		error = xfs_dialloc_ag_finobt_near(pagino, &cur, &rec);
 	else
 		error = xfs_dialloc_ag_finobt_newino(agi, cur, &rec);
@@ -1621,7 +1481,7 @@ xfs_dialloc_ag(
 	ASSERT(offset < XFS_INODES_PER_CHUNK);
 	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
 				   XFS_INODES_PER_CHUNK) == 0);
-	ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset);
+	ino = XFS_AGINO_TO_INO(mp, pag->pag_agno, rec.ir_startino + offset);
 
 	/*
 	 * Modify or remove the finobt record.
@@ -1641,9 +1501,9 @@ xfs_dialloc_ag(
 	 * the original freecount. If all is well, make the equivalent update to
 	 * the inobt using the finobt record and offset information.
 	 */
-	icur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO);
+	icur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
 
-	error = xfs_check_agi_freecount(icur, agi);
+	error = xfs_check_agi_freecount(icur);
 	if (error)
 		goto error_icur;
 
@@ -1657,14 +1517,14 @@ xfs_dialloc_ag(
 	 */
 	be32_add_cpu(&agi->agi_freecount, -1);
 	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
-	agbp->b_pag->pagi_freecount--;
+	pag->pagi_freecount--;
 
 	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
 
-	error = xfs_check_agi_freecount(icur, agi);
+	error = xfs_check_agi_freecount(icur);
 	if (error)
 		goto error_icur;
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error_icur;
 
@@ -1708,54 +1568,195 @@ xfs_dialloc_roll(
 	/* Re-attach the quota info that we detached from prev trx. */
 	tp->t_dqinfo = dqinfo;
 
+	/*
+	 * Join the buffer even on commit error so that the buffer is released
+	 * when the caller cancels the transaction and doesn't have to handle
+	 * this error case specially.
+	 */
+	xfs_trans_bjoin(tp, agibp);
 	*tpp = tp;
+	return error;
+}
+
+static xfs_agnumber_t
+xfs_ialloc_next_ag(
+	xfs_mount_t	*mp)
+{
+	xfs_agnumber_t	agno;
+
+	spin_lock(&mp->m_agirotor_lock);
+	agno = mp->m_agirotor;
+	if (++mp->m_agirotor >= mp->m_maxagi)
+		mp->m_agirotor = 0;
+	spin_unlock(&mp->m_agirotor_lock);
+
+	return agno;
+}
+
+static bool
+xfs_dialloc_good_ag(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	umode_t			mode,
+	int			flags,
+	bool			ok_alloc)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	xfs_extlen_t		ineed;
+	xfs_extlen_t		longest = 0;
+	int			needspace;
+	int			error;
+
+	if (!pag->pagi_inodeok)
+		return false;
+
+	if (!pag->pagi_init) {
+		error = xfs_ialloc_pagi_init(mp, tp, pag->pag_agno);
+		if (error)
+			return false;
+	}
+
+	if (pag->pagi_freecount)
+		return true;
+	if (!ok_alloc)
+		return false;
+
+	if (!pag->pagf_init) {
+		error = xfs_alloc_pagf_init(mp, tp, pag->pag_agno, flags);
+		if (error)
+			return false;
+	}
+
+	/*
+	 * Check that there is enough free space for the file plus a chunk of
+	 * inodes if we need to allocate some. If this is the first pass across
+	 * the AGs, take into account the potential space needed for alignment
+	 * of inode chunks when checking the longest contiguous free space in
+	 * the AG - this prevents us from getting ENOSPC because we have free
+	 * space larger than ialloc_blks but alignment constraints prevent us
+	 * from using it.
+	 *
+	 * If we can't find an AG with space for full alignment slack to be
+	 * taken into account, we must be near ENOSPC in all AGs.  Hence we
+	 * don't include alignment for the second pass and so if we fail
+	 * allocation due to alignment issues then it is most likely a real
+	 * ENOSPC condition.
+	 *
+	 * XXX(dgc): this calculation is now bogus thanks to the per-ag
+	 * reservations that xfs_alloc_fix_freelist() now does via
+	 * xfs_alloc_space_available(). When the AG fills up, pagf_freeblks will
+	 * be more than large enough for the check below to succeed, but
+	 * xfs_alloc_space_available() will fail because of the non-zero
+	 * metadata reservation and hence we won't actually be able to allocate
+	 * more inodes in this AG. We do soooo much unnecessary work near ENOSPC
+	 * because of this.
+	 */
+	ineed = M_IGEO(mp)->ialloc_min_blks;
+	if (flags && ineed > 1)
+		ineed += M_IGEO(mp)->cluster_align;
+	longest = pag->pagf_longest;
+	if (!longest)
+		longest = pag->pagf_flcount > 0;
+	needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
+
+	if (pag->pagf_freeblks < needspace + ineed || longest < ineed)
+		return false;
+	return true;
+}
+
+static int
+xfs_dialloc_try_ag(
+	struct xfs_trans	**tpp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		parent,
+	xfs_ino_t		*new_ino,
+	bool			ok_alloc)
+{
+	struct xfs_buf		*agbp;
+	xfs_ino_t		ino;
+	int			error;
+
+	/*
+	 * Then read in the AGI buffer and recheck with the AGI buffer
+	 * lock held.
+	 */
+	error = xfs_ialloc_read_agi(pag->pag_mount, *tpp, pag->pag_agno, &agbp);
 	if (error)
 		return error;
-	xfs_trans_bjoin(tp, agibp);
-	return 0;
+
+	if (!pag->pagi_freecount) {
+		if (!ok_alloc) {
+			error = -EAGAIN;
+			goto out_release;
+		}
+
+		error = xfs_ialloc_ag_alloc(*tpp, agbp, pag);
+		if (error < 0)
+			goto out_release;
+
+		/*
+		 * We successfully allocated space for an inode cluster in this
+		 * AG.  Roll the transaction so that we can allocate one of the
+		 * new inodes.
+		 */
+		ASSERT(pag->pagi_freecount > 0);
+		error = xfs_dialloc_roll(tpp, agbp);
+		if (error)
+			goto out_release;
+	}
+
+	/* Allocate an inode in the found AG */
+	error = xfs_dialloc_ag(*tpp, agbp, pag, parent, &ino);
+	if (!error)
+		*new_ino = ino;
+	return error;
+
+out_release:
+	xfs_trans_brelse(*tpp, agbp);
+	return error;
 }
 
 /*
- * Select and prepare an AG for inode allocation.
+ * Allocate an on-disk inode.
  *
  * Mode is used to tell whether the new inode is a directory and hence where to
- * locate it.
- *
- * This function will ensure that the selected AG has free inodes available to
- * allocate from. The selected AGI will be returned locked to the caller, and it
- * will allocate more free inodes if required. If no free inodes are found or
- * can be allocated, no AGI will be returned.
+ * locate it. The on-disk inode that is allocated will be returned in @new_ino
+ * on success, otherwise an error will be set to indicate the failure (e.g.
+ * -ENOSPC).
  */
 int
-xfs_dialloc_select_ag(
+xfs_dialloc(
 	struct xfs_trans	**tpp,
 	xfs_ino_t		parent,
 	umode_t			mode,
-	struct xfs_buf		**IO_agbp)
+	xfs_ino_t		*new_ino)
 {
 	struct xfs_mount	*mp = (*tpp)->t_mountp;
-	struct xfs_buf		*agbp;
 	xfs_agnumber_t		agno;
-	int			error;
-	bool			noroom = false;
+	int			error = 0;
 	xfs_agnumber_t		start_agno;
 	struct xfs_perag	*pag;
 	struct xfs_ino_geometry	*igeo = M_IGEO(mp);
-	bool			okalloc = true;
-
-	*IO_agbp = NULL;
+	bool			ok_alloc = true;
+	int			flags;
+	xfs_ino_t		ino;
 
 	/*
-	 * We do not have an agbp, so select an initial allocation
-	 * group for inode allocation.
+	 * Directories, symlinks, and regular files frequently allocate at least
+	 * one block, so factor that potential expansion when we examine whether
+	 * an AG has enough space for file creation.
 	 */
-	start_agno = xfs_ialloc_ag_select(*tpp, parent, mode);
-	if (start_agno == NULLAGNUMBER)
-		return 0;
+	if (S_ISDIR(mode))
+		start_agno = xfs_ialloc_next_ag(mp);
+	else {
+		start_agno = XFS_INO_TO_AGNO(mp, parent);
+		if (start_agno >= mp->m_maxagi)
+			start_agno = 0;
+	}
 
 	/*
 	 * If we have already hit the ceiling of inode blocks then clear
-	 * okalloc so we scan all available agi structures for a free
+	 * ok_alloc so we scan all available agi structures for a free
 	 * inode.
 	 *
 	 * Read rough value of mp->m_icount by percpu_counter_read_positive,
@@ -1764,8 +1765,7 @@ xfs_dialloc_select_ag(
 	if (igeo->maxicount &&
 	    percpu_counter_read_positive(&mp->m_icount) + igeo->ialloc_inos
 							> igeo->maxicount) {
-		noroom = true;
-		okalloc = false;
+		ok_alloc = false;
 	}
 
 	/*
@@ -1774,82 +1774,36 @@ xfs_dialloc_select_ag(
 	 * allocation groups upward, wrapping at the end.
 	 */
 	agno = start_agno;
+	flags = XFS_ALLOC_FLAG_TRYLOCK;
 	for (;;) {
 		pag = xfs_perag_get(mp, agno);
-		if (!pag->pagi_inodeok) {
-			xfs_ialloc_next_ag(mp);
-			goto nextag;
-		}
-
-		if (!pag->pagi_init) {
-			error = xfs_ialloc_pagi_init(mp, *tpp, agno);
-			if (error)
+		if (xfs_dialloc_good_ag(*tpp, pag, mode, flags, ok_alloc)) {
+			error = xfs_dialloc_try_ag(tpp, pag, parent,
+					&ino, ok_alloc);
+			if (error != -EAGAIN)
 				break;
 		}
 
-		/*
-		 * Do a first racy fast path check if this AG is usable.
-		 */
-		if (!pag->pagi_freecount && !okalloc)
-			goto nextag;
-
-		/*
-		 * Then read in the AGI buffer and recheck with the AGI buffer
-		 * lock held.
-		 */
-		error = xfs_ialloc_read_agi(mp, *tpp, agno, &agbp);
-		if (error)
-			break;
-
-		if (pag->pagi_freecount) {
-			xfs_perag_put(pag);
-			goto found_ag;
-		}
-
-		if (!okalloc)
-			goto nextag_relse_buffer;
-
-		error = xfs_ialloc_ag_alloc(*tpp, agbp);
-		if (error < 0) {
-			xfs_trans_brelse(*tpp, agbp);
-
-			if (error == -ENOSPC)
-				error = 0;
+		if (XFS_FORCED_SHUTDOWN(mp)) {
+			error = -EFSCORRUPTED;
 			break;
 		}
-
-		if (error == 0) {
-			/*
-			 * We successfully allocated space for an inode cluster
-			 * in this AG.  Roll the transaction so that we can
-			 * allocate one of the new inodes.
-			 */
-			ASSERT(pag->pagi_freecount > 0);
-			xfs_perag_put(pag);
-
-			error = xfs_dialloc_roll(tpp, agbp);
-			if (error) {
-				xfs_buf_relse(agbp);
-				return error;
+		if (++agno == mp->m_maxagi)
+			agno = 0;
+		if (agno == start_agno) {
+			if (!flags) {
+				error = -ENOSPC;
+				break;
 			}
-			goto found_ag;
+			flags = 0;
 		}
-
-nextag_relse_buffer:
-		xfs_trans_brelse(*tpp, agbp);
-nextag:
 		xfs_perag_put(pag);
-		if (++agno == mp->m_sb.sb_agcount)
-			agno = 0;
-		if (agno == start_agno)
-			return noroom ? -ENOSPC : 0;
 	}
 
+	if (!error)
+		*new_ino = ino;
 	xfs_perag_put(pag);
 	return error;
-found_ag:
-	*IO_agbp = agbp;
-	return 0;
 }
 
 /*
@@ -1935,12 +1889,12 @@ xfs_difree_inobt(
 	struct xfs_mount		*mp,
 	struct xfs_trans		*tp,
 	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
 	xfs_agino_t			agino,
 	struct xfs_icluster		*xic,
 	struct xfs_inobt_rec_incore	*orec)
 {
 	struct xfs_agi			*agi = agbp->b_addr;
-	xfs_agnumber_t			agno = be32_to_cpu(agi->agi_seqno);
 	struct xfs_btree_cur		*cur;
 	struct xfs_inobt_rec_incore	rec;
 	int				ilen;
@@ -1954,9 +1908,9 @@ xfs_difree_inobt(
 	/*
 	 * Initialize the cursor.
 	 */
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
 
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error0;
 
@@ -2005,7 +1959,8 @@ xfs_difree_inobt(
 		struct xfs_perag	*pag = agbp->b_pag;
 
 		xic->deleted = true;
-		xic->first_ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino);
+		xic->first_ino = XFS_AGINO_TO_INO(mp, pag->pag_agno,
+				rec.ir_startino);
 		xic->alloc = xfs_inobt_irec_to_allocmask(&rec);
 
 		/*
@@ -2028,7 +1983,7 @@ xfs_difree_inobt(
 			goto error0;
 		}
 
-		xfs_difree_inode_chunk(tp, agno, &rec);
+		xfs_difree_inode_chunk(tp, pag->pag_agno, &rec);
 	} else {
 		xic->deleted = false;
 
@@ -2044,11 +1999,11 @@ xfs_difree_inobt(
 		 */
 		be32_add_cpu(&agi->agi_freecount, 1);
 		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
-		agbp->b_pag->pagi_freecount++;
+		pag->pagi_freecount++;
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, 1);
 	}
 
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error0;
 
@@ -2069,18 +2024,17 @@ xfs_difree_finobt(
 	struct xfs_mount		*mp,
 	struct xfs_trans		*tp,
 	struct xfs_buf			*agbp,
+	struct xfs_perag		*pag,
 	xfs_agino_t			agino,
 	struct xfs_inobt_rec_incore	*ibtrec) /* inobt record */
 {
-	struct xfs_agi			*agi = agbp->b_addr;
-	xfs_agnumber_t			agno = be32_to_cpu(agi->agi_seqno);
 	struct xfs_btree_cur		*cur;
 	struct xfs_inobt_rec_incore	rec;
 	int				offset = agino - ibtrec->ir_startino;
 	int				error;
 	int				i;
 
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_FINO);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_FINO);
 
 	error = xfs_inobt_lookup(cur, ibtrec->ir_startino, XFS_LOOKUP_EQ, &i);
 	if (error)
@@ -2158,7 +2112,7 @@ xfs_difree_finobt(
 	}
 
 out:
-	error = xfs_check_agi_freecount(cur, agi);
+	error = xfs_check_agi_freecount(cur);
 	if (error)
 		goto error;
 
@@ -2178,36 +2132,33 @@ error:
  */
 int
 xfs_difree(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	xfs_ino_t		inode,		/* inode to be freed */
-	struct xfs_icluster	*xic)	/* cluster info if deleted */
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	xfs_ino_t		inode,
+	struct xfs_icluster	*xic)
 {
 	/* REFERENCED */
 	xfs_agblock_t		agbno;	/* block number containing inode */
 	struct xfs_buf		*agbp;	/* buffer for allocation group header */
 	xfs_agino_t		agino;	/* allocation group inode number */
-	xfs_agnumber_t		agno;	/* allocation group number */
 	int			error;	/* error return value */
-	struct xfs_mount	*mp;	/* mount structure for filesystem */
+	struct xfs_mount	*mp = tp->t_mountp;
 	struct xfs_inobt_rec_incore rec;/* btree record */
 
-	mp = tp->t_mountp;
-
 	/*
 	 * Break up inode number into its components.
 	 */
-	agno = XFS_INO_TO_AGNO(mp, inode);
-	if (agno >= mp->m_sb.sb_agcount)  {
-		xfs_warn(mp, "%s: agno >= mp->m_sb.sb_agcount (%d >= %d).",
-			__func__, agno, mp->m_sb.sb_agcount);
+	if (pag->pag_agno != XFS_INO_TO_AGNO(mp, inode)) {
+		xfs_warn(mp, "%s: agno != pag->pag_agno (%d != %d).",
+			__func__, XFS_INO_TO_AGNO(mp, inode), pag->pag_agno);
 		ASSERT(0);
 		return -EINVAL;
 	}
 	agino = XFS_INO_TO_AGINO(mp, inode);
-	if (inode != XFS_AGINO_TO_INO(mp, agno, agino))  {
+	if (inode != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino))  {
 		xfs_warn(mp, "%s: inode != XFS_AGINO_TO_INO() (%llu != %llu).",
 			__func__, (unsigned long long)inode,
-			(unsigned long long)XFS_AGINO_TO_INO(mp, agno, agino));
+			(unsigned long long)XFS_AGINO_TO_INO(mp, pag->pag_agno, agino));
 		ASSERT(0);
 		return -EINVAL;
 	}
@@ -2221,7 +2172,7 @@ xfs_difree(
 	/*
 	 * Get the allocation group header.
 	 */
-	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
+	error = xfs_ialloc_read_agi(mp, tp, pag->pag_agno, &agbp);
 	if (error) {
 		xfs_warn(mp, "%s: xfs_ialloc_read_agi() returned error %d.",
 			__func__, error);
@@ -2231,7 +2182,7 @@ xfs_difree(
 	/*
 	 * Fix up the inode allocation btree.
 	 */
-	error = xfs_difree_inobt(mp, tp, agbp, agino, xic, &rec);
+	error = xfs_difree_inobt(mp, tp, agbp, pag, agino, xic, &rec);
 	if (error)
 		goto error0;
 
@@ -2239,7 +2190,7 @@ xfs_difree(
 	 * Fix up the free inode btree.
 	 */
 	if (xfs_sb_version_hasfinobt(&mp->m_sb)) {
-		error = xfs_difree_finobt(mp, tp, agbp, agino, &rec);
+		error = xfs_difree_finobt(mp, tp, agbp, pag, agino, &rec);
 		if (error)
 			goto error0;
 	}
@@ -2254,7 +2205,7 @@ STATIC int
 xfs_imap_lookup(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agino_t		agino,
 	xfs_agblock_t		agbno,
 	xfs_agblock_t		*chunk_agbno,
@@ -2267,11 +2218,11 @@ xfs_imap_lookup(
 	int			error;
 	int			i;
 
-	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
+	error = xfs_ialloc_read_agi(mp, tp, pag->pag_agno, &agbp);
 	if (error) {
 		xfs_alert(mp,
 			"%s: xfs_ialloc_read_agi() returned error %d, agno %d",
-			__func__, error, agno);
+			__func__, error, pag->pag_agno);
 		return error;
 	}
 
@@ -2281,7 +2232,7 @@ xfs_imap_lookup(
 	 * we have a record, we need to ensure it contains the inode number
 	 * we are looking up.
 	 */
-	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, pag, XFS_BTNUM_INO);
 	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i);
 	if (!error) {
 		if (i)
@@ -2315,42 +2266,44 @@ xfs_imap_lookup(
  */
 int
 xfs_imap(
-	xfs_mount_t	 *mp,	/* file system mount structure */
-	xfs_trans_t	 *tp,	/* transaction pointer */
-	xfs_ino_t	ino,	/* inode to locate */
-	struct xfs_imap	*imap,	/* location map structure */
-	uint		flags)	/* flags for inode btree lookup */
+	struct xfs_mount	 *mp,	/* file system mount structure */
+	struct xfs_trans	 *tp,	/* transaction pointer */
+	xfs_ino_t		ino,	/* inode to locate */
+	struct xfs_imap		*imap,	/* location map structure */
+	uint			flags)	/* flags for inode btree lookup */
 {
-	xfs_agblock_t	agbno;	/* block number of inode in the alloc group */
-	xfs_agino_t	agino;	/* inode number within alloc group */
-	xfs_agnumber_t	agno;	/* allocation group number */
-	xfs_agblock_t	chunk_agbno;	/* first block in inode chunk */
-	xfs_agblock_t	cluster_agbno;	/* first block in inode cluster */
-	int		error;	/* error code */
-	int		offset;	/* index of inode in its buffer */
-	xfs_agblock_t	offset_agbno;	/* blks from chunk start to inode */
+	xfs_agblock_t		agbno;	/* block number of inode in the alloc group */
+	xfs_agino_t		agino;	/* inode number within alloc group */
+	xfs_agblock_t		chunk_agbno;	/* first block in inode chunk */
+	xfs_agblock_t		cluster_agbno;	/* first block in inode cluster */
+	int			error;	/* error code */
+	int			offset;	/* index of inode in its buffer */
+	xfs_agblock_t		offset_agbno;	/* blks from chunk start to inode */
+	struct xfs_perag	*pag;
 
 	ASSERT(ino != NULLFSINO);
 
 	/*
 	 * Split up the inode number into its parts.
 	 */
-	agno = XFS_INO_TO_AGNO(mp, ino);
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ino));
 	agino = XFS_INO_TO_AGINO(mp, ino);
 	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
-	if (agno >= mp->m_sb.sb_agcount || agbno >= mp->m_sb.sb_agblocks ||
-	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
+	if (!pag || agbno >= mp->m_sb.sb_agblocks ||
+	    ino != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino)) {
+		error = -EINVAL;
 #ifdef DEBUG
 		/*
 		 * Don't output diagnostic information for untrusted inodes
 		 * as they can be invalid without implying corruption.
 		 */
 		if (flags & XFS_IGET_UNTRUSTED)
-			return -EINVAL;
-		if (agno >= mp->m_sb.sb_agcount) {
+			goto out_drop;
+		if (!pag) {
 			xfs_alert(mp,
 				"%s: agno (%d) >= mp->m_sb.sb_agcount (%d)",
-				__func__, agno, mp->m_sb.sb_agcount);
+				__func__, XFS_INO_TO_AGNO(mp, ino),
+				mp->m_sb.sb_agcount);
 		}
 		if (agbno >= mp->m_sb.sb_agblocks) {
 			xfs_alert(mp,
@@ -2358,15 +2311,15 @@ xfs_imap(
 				__func__, (unsigned long long)agbno,
 				(unsigned long)mp->m_sb.sb_agblocks);
 		}
-		if (ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
+		if (pag && ino != XFS_AGINO_TO_INO(mp, pag->pag_agno, agino)) {
 			xfs_alert(mp,
 		"%s: ino (0x%llx) != XFS_AGINO_TO_INO() (0x%llx)",
 				__func__, ino,
-				XFS_AGINO_TO_INO(mp, agno, agino));
+				XFS_AGINO_TO_INO(mp, pag->pag_agno, agino));
 		}
 		xfs_stack_trace();
 #endif /* DEBUG */
-		return -EINVAL;
+		goto out_drop;
 	}
 
 	/*
@@ -2377,10 +2330,10 @@ xfs_imap(
 	 * in all cases where an untrusted inode number is passed.
 	 */
 	if (flags & XFS_IGET_UNTRUSTED) {
-		error = xfs_imap_lookup(mp, tp, agno, agino, agbno,
+		error = xfs_imap_lookup(mp, tp, pag, agino, agbno,
 					&chunk_agbno, &offset_agbno, flags);
 		if (error)
-			return error;
+			goto out_drop;
 		goto out_map;
 	}
 
@@ -2392,11 +2345,12 @@ xfs_imap(
 		offset = XFS_INO_TO_OFFSET(mp, ino);
 		ASSERT(offset < mp->m_sb.sb_inopblock);
 
-		imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, agbno);
+		imap->im_blkno = XFS_AGB_TO_DADDR(mp, pag->pag_agno, agbno);
 		imap->im_len = XFS_FSB_TO_BB(mp, 1);
 		imap->im_boffset = (unsigned short)(offset <<
 							mp->m_sb.sb_inodelog);
-		return 0;
+		error = 0;
+		goto out_drop;
 	}
 
 	/*
@@ -2408,10 +2362,10 @@ xfs_imap(
 		offset_agbno = agbno & M_IGEO(mp)->inoalign_mask;
 		chunk_agbno = agbno - offset_agbno;
 	} else {
-		error = xfs_imap_lookup(mp, tp, agno, agino, agbno,
+		error = xfs_imap_lookup(mp, tp, pag, agino, agbno,
 					&chunk_agbno, &offset_agbno, flags);
 		if (error)
-			return error;
+			goto out_drop;
 	}
 
 out_map:
@@ -2422,7 +2376,7 @@ out_map:
 	offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
 		XFS_INO_TO_OFFSET(mp, ino);
 
-	imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, cluster_agbno);
+	imap->im_blkno = XFS_AGB_TO_DADDR(mp, pag->pag_agno, cluster_agbno);
 	imap->im_len = XFS_FSB_TO_BB(mp, M_IGEO(mp)->blocks_per_cluster);
 	imap->im_boffset = (unsigned short)(offset << mp->m_sb.sb_inodelog);
 
@@ -2439,9 +2393,14 @@ out_map:
 			__func__, (unsigned long long) imap->im_blkno,
 			(unsigned long long) imap->im_len,
 			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
-		return -EINVAL;
+		error = -EINVAL;
+		goto out_drop;
 	}
-	return 0;
+	error = 0;
+out_drop:
+	if (pag)
+		xfs_perag_put(pag);
+	return error;
 }
 
 /*
diff --git a/fs/xfs/libxfs/xfs_ialloc.h b/fs/xfs/libxfs/xfs_ialloc.h
index 3511086a7ae1..9df7c80408ff 100644
--- a/fs/xfs/libxfs/xfs_ialloc.h
+++ b/fs/xfs/libxfs/xfs_ialloc.h
@@ -33,42 +33,14 @@ xfs_make_iptr(struct xfs_mount *mp, struct xfs_buf *b, int o)
 }
 
 /*
- * Allocate an inode on disk.
- * Mode is used to tell whether the new inode will need space, and whether
- * it is a directory.
- *
- * There are two phases to inode allocation: selecting an AG and ensuring
- * that it contains free inodes, followed by allocating one of the free
- * inodes. xfs_dialloc_select_ag() does the former and returns a locked AGI
- * to the caller, ensuring that followup call to xfs_dialloc_ag() will
- * have free inodes to allocate from. xfs_dialloc_ag() will return the inode
- * number of the free inode we allocated.
+ * Allocate an inode on disk.  Mode is used to tell whether the new inode will
+ * need space, and whether it is a directory.
  */
-int					/* error */
-xfs_dialloc_select_ag(
-	struct xfs_trans **tpp,		/* double pointer of transaction */
-	xfs_ino_t	parent,		/* parent inode (directory) */
-	umode_t		mode,		/* mode bits for new inode */
-	struct xfs_buf	**IO_agbp);
-
-int
-xfs_dialloc_ag(
-	struct xfs_trans	*tp,
-	struct xfs_buf		*agbp,
-	xfs_ino_t		parent,
-	xfs_ino_t		*inop);
+int xfs_dialloc(struct xfs_trans **tpp, xfs_ino_t parent, umode_t mode,
+		xfs_ino_t *new_ino);
 
-/*
- * Free disk inode.  Carefully avoids touching the incore inode, all
- * manipulations incore are the caller's responsibility.
- * The on-disk inode is not changed by this operation, only the
- * btree (free inode mask) is changed.
- */
-int					/* error */
-xfs_difree(
-	struct xfs_trans *tp,		/* transaction pointer */
-	xfs_ino_t	inode,		/* inode to be freed */
-	struct xfs_icluster *ifree);	/* cluster info if deleted */
+int xfs_difree(struct xfs_trans *tp, struct xfs_perag *pag,
+		xfs_ino_t ino, struct xfs_icluster *ifree);
 
 /*
  * Return the location of the inode in imap, for mapping it into a buffer.
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.c b/fs/xfs/libxfs/xfs_ialloc_btree.c
index 4c5831646bd9..823a038939f8 100644
--- a/fs/xfs/libxfs/xfs_ialloc_btree.c
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.c
@@ -20,6 +20,7 @@
 #include "xfs_trace.h"
 #include "xfs_trans.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 
 STATIC int
 xfs_inobt_get_minrecs(
@@ -34,8 +35,7 @@ xfs_inobt_dup_cursor(
 	struct xfs_btree_cur	*cur)
 {
 	return xfs_inobt_init_cursor(cur->bc_mp, cur->bc_tp,
-			cur->bc_ag.agbp, cur->bc_ag.agno,
-			cur->bc_btnum);
+			cur->bc_ag.agbp, cur->bc_ag.pag, cur->bc_btnum);
 }
 
 STATIC void
@@ -102,7 +102,7 @@ __xfs_inobt_alloc_block(
 	args.tp = cur->bc_tp;
 	args.mp = cur->bc_mp;
 	args.oinfo = XFS_RMAP_OINFO_INOBT;
-	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_ag.agno, sbno);
+	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_ag.pag->pag_agno, sbno);
 	args.minlen = 1;
 	args.maxlen = 1;
 	args.prod = 1;
@@ -235,7 +235,7 @@ xfs_inobt_init_ptr_from_cur(
 {
 	struct xfs_agi		*agi = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_ag.agno == be32_to_cpu(agi->agi_seqno));
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agi->agi_seqno));
 
 	ptr->s = agi->agi_root;
 }
@@ -247,7 +247,7 @@ xfs_finobt_init_ptr_from_cur(
 {
 	struct xfs_agi		*agi = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_ag.agno == be32_to_cpu(agi->agi_seqno));
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agi->agi_seqno));
 	ptr->s = agi->agi_free_root;
 }
 
@@ -427,7 +427,7 @@ static struct xfs_btree_cur *
 xfs_inobt_init_common(
 	struct xfs_mount	*mp,		/* file system mount point */
 	struct xfs_trans	*tp,		/* transaction pointer */
-	xfs_agnumber_t		agno,		/* allocation group number */
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)		/* ialloc or free ino btree */
 {
 	struct xfs_btree_cur	*cur;
@@ -449,7 +449,9 @@ xfs_inobt_init_common(
 	if (xfs_sb_version_hascrc(&mp->m_sb))
 		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
 
-	cur->bc_ag.agno = agno;
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
 	return cur;
 }
 
@@ -459,13 +461,13 @@ xfs_inobt_init_cursor(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)
 {
 	struct xfs_btree_cur	*cur;
 	struct xfs_agi		*agi = agbp->b_addr;
 
-	cur = xfs_inobt_init_common(mp, tp, agno, btnum);
+	cur = xfs_inobt_init_common(mp, tp, pag, btnum);
 	if (btnum == XFS_BTNUM_INO)
 		cur->bc_nlevels = be32_to_cpu(agi->agi_level);
 	else
@@ -479,12 +481,12 @@ struct xfs_btree_cur *
 xfs_inobt_stage_cursor(
 	struct xfs_mount	*mp,
 	struct xbtree_afakeroot	*afake,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum)
 {
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_inobt_init_common(mp, NULL, agno, btnum);
+	cur = xfs_inobt_init_common(mp, NULL, pag, btnum);
 	xfs_btree_stage_afakeroot(cur, afake);
 	return cur;
 }
@@ -656,7 +658,7 @@ int
 xfs_inobt_cur(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		which,
 	struct xfs_btree_cur	**curpp,
 	struct xfs_buf		**agi_bpp)
@@ -667,11 +669,11 @@ xfs_inobt_cur(
 	ASSERT(*agi_bpp == NULL);
 	ASSERT(*curpp == NULL);
 
-	error = xfs_ialloc_read_agi(mp, tp, agno, agi_bpp);
+	error = xfs_ialloc_read_agi(mp, tp, pag->pag_agno, agi_bpp);
 	if (error)
 		return error;
 
-	cur = xfs_inobt_init_cursor(mp, tp, *agi_bpp, agno, which);
+	cur = xfs_inobt_init_cursor(mp, tp, *agi_bpp, pag, which);
 	*curpp = cur;
 	return 0;
 }
@@ -680,7 +682,7 @@ static int
 xfs_inobt_count_blocks(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_btnum_t		btnum,
 	xfs_extlen_t		*tree_blocks)
 {
@@ -688,7 +690,7 @@ xfs_inobt_count_blocks(
 	struct xfs_btree_cur	*cur = NULL;
 	int			error;
 
-	error = xfs_inobt_cur(mp, tp, agno, btnum, &cur, &agbp);
+	error = xfs_inobt_cur(mp, tp, pag, btnum, &cur, &agbp);
 	if (error)
 		return error;
 
@@ -704,14 +706,14 @@ static int
 xfs_finobt_read_blocks(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_extlen_t		*tree_blocks)
 {
 	struct xfs_buf		*agbp;
 	struct xfs_agi		*agi;
 	int			error;
 
-	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
+	error = xfs_ialloc_read_agi(mp, tp, pag->pag_agno, &agbp);
 	if (error)
 		return error;
 
@@ -728,7 +730,7 @@ int
 xfs_finobt_calc_reserves(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_extlen_t		*ask,
 	xfs_extlen_t		*used)
 {
@@ -739,14 +741,14 @@ xfs_finobt_calc_reserves(
 		return 0;
 
 	if (xfs_sb_version_hasinobtcounts(&mp->m_sb))
-		error = xfs_finobt_read_blocks(mp, tp, agno, &tree_len);
+		error = xfs_finobt_read_blocks(mp, tp, pag, &tree_len);
 	else
-		error = xfs_inobt_count_blocks(mp, tp, agno, XFS_BTNUM_FINO,
+		error = xfs_inobt_count_blocks(mp, tp, pag, XFS_BTNUM_FINO,
 				&tree_len);
 	if (error)
 		return error;
 
-	*ask += xfs_inobt_max_size(mp, agno);
+	*ask += xfs_inobt_max_size(mp, pag->pag_agno);
 	*used += tree_len;
 	return 0;
 }
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.h b/fs/xfs/libxfs/xfs_ialloc_btree.h
index 35bbd978c272..e530c82b2217 100644
--- a/fs/xfs/libxfs/xfs_ialloc_btree.h
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.h
@@ -13,6 +13,7 @@
 struct xfs_buf;
 struct xfs_btree_cur;
 struct xfs_mount;
+struct xfs_perag;
 
 /*
  * Btree block header size depends on a superblock flag.
@@ -45,11 +46,11 @@ struct xfs_mount;
 		 (maxrecs) * sizeof(xfs_inobt_key_t) + \
 		 ((index) - 1) * sizeof(xfs_inobt_ptr_t)))
 
-extern struct xfs_btree_cur *xfs_inobt_init_cursor(struct xfs_mount *,
-		struct xfs_trans *, struct xfs_buf *, xfs_agnumber_t,
-		xfs_btnum_t);
+extern struct xfs_btree_cur *xfs_inobt_init_cursor(struct xfs_mount *mp,
+		struct xfs_trans *tp, struct xfs_buf *agbp,
+		struct xfs_perag *pag, xfs_btnum_t btnum);
 struct xfs_btree_cur *xfs_inobt_stage_cursor(struct xfs_mount *mp,
-		struct xbtree_afakeroot *afake, xfs_agnumber_t agno,
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag,
 		xfs_btnum_t btnum);
 extern int xfs_inobt_maxrecs(struct xfs_mount *, int, int);
 
@@ -64,11 +65,11 @@ int xfs_inobt_rec_check_count(struct xfs_mount *,
 #endif	/* DEBUG */
 
 int xfs_finobt_calc_reserves(struct xfs_mount *mp, struct xfs_trans *tp,
-		xfs_agnumber_t agno, xfs_extlen_t *ask, xfs_extlen_t *used);
+		struct xfs_perag *pag, xfs_extlen_t *ask, xfs_extlen_t *used);
 extern xfs_extlen_t xfs_iallocbt_calc_size(struct xfs_mount *mp,
 		unsigned long long len);
 int xfs_inobt_cur(struct xfs_mount *mp, struct xfs_trans *tp,
-		xfs_agnumber_t agno, xfs_btnum_t btnum,
+		struct xfs_perag *pag, xfs_btnum_t btnum,
 		struct xfs_btree_cur **curpp, struct xfs_buf **agi_bpp);
 
 void xfs_inobt_commit_staged_btree(struct xfs_btree_cur *cur,
diff --git a/fs/xfs/libxfs/xfs_inode_buf.c b/fs/xfs/libxfs/xfs_inode_buf.c
index f3254a4f4cb4..04ce361688f7 100644
--- a/fs/xfs/libxfs/xfs_inode_buf.c
+++ b/fs/xfs/libxfs/xfs_inode_buf.c
@@ -612,7 +612,7 @@ xfs_inode_validate_extsize(
 	 */
 
 	if (rt_flag)
-		blocksize_bytes = mp->m_sb.sb_rextsize << mp->m_sb.sb_blocklog;
+		blocksize_bytes = XFS_FSB_TO_B(mp, mp->m_sb.sb_rextsize);
 	else
 		blocksize_bytes = mp->m_sb.sb_blocksize;
 
diff --git a/fs/xfs/libxfs/xfs_log_format.h b/fs/xfs/libxfs/xfs_log_format.h
index 3e15ea29fb8d..d548ea4b6aab 100644
--- a/fs/xfs/libxfs/xfs_log_format.h
+++ b/fs/xfs/libxfs/xfs_log_format.h
@@ -34,9 +34,6 @@ typedef uint32_t xlog_tid_t;
 #define XLOG_MIN_RECORD_BSHIFT	14		/* 16384 == 1 << 14 */
 #define XLOG_BIG_RECORD_BSHIFT	15		/* 32k == 1 << 15 */
 #define XLOG_MAX_RECORD_BSHIFT	18		/* 256k == 1 << 18 */
-#define XLOG_BTOLSUNIT(log, b)  (((b)+(log)->l_mp->m_sb.sb_logsunit-1) / \
-                                 (log)->l_mp->m_sb.sb_logsunit)
-#define XLOG_LSUNITTOB(log, su) ((su) * (log)->l_mp->m_sb.sb_logsunit)
 
 #define XLOG_HEADER_SIZE	512
 
diff --git a/fs/xfs/libxfs/xfs_refcount.c b/fs/xfs/libxfs/xfs_refcount.c
index 2037b9f23069..860a0c9801ba 100644
--- a/fs/xfs/libxfs/xfs_refcount.c
+++ b/fs/xfs/libxfs/xfs_refcount.c
@@ -22,6 +22,7 @@
 #include "xfs_bit.h"
 #include "xfs_refcount.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 
 /* Allowable refcount adjustment amounts. */
 enum xfs_refc_adjust_op {
@@ -46,7 +47,7 @@ xfs_refcount_lookup_le(
 	xfs_agblock_t		bno,
 	int			*stat)
 {
-	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.agno, bno,
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno, bno,
 			XFS_LOOKUP_LE);
 	cur->bc_rec.rc.rc_startblock = bno;
 	cur->bc_rec.rc.rc_blockcount = 0;
@@ -63,7 +64,7 @@ xfs_refcount_lookup_ge(
 	xfs_agblock_t		bno,
 	int			*stat)
 {
-	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.agno, bno,
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno, bno,
 			XFS_LOOKUP_GE);
 	cur->bc_rec.rc.rc_startblock = bno;
 	cur->bc_rec.rc.rc_blockcount = 0;
@@ -80,7 +81,7 @@ xfs_refcount_lookup_eq(
 	xfs_agblock_t		bno,
 	int			*stat)
 {
-	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.agno, bno,
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_ag.pag->pag_agno, bno,
 			XFS_LOOKUP_LE);
 	cur->bc_rec.rc.rc_startblock = bno;
 	cur->bc_rec.rc.rc_blockcount = 0;
@@ -108,7 +109,7 @@ xfs_refcount_get_rec(
 	int				*stat)
 {
 	struct xfs_mount		*mp = cur->bc_mp;
-	xfs_agnumber_t			agno = cur->bc_ag.agno;
+	xfs_agnumber_t			agno = cur->bc_ag.pag->pag_agno;
 	union xfs_btree_rec		*rec;
 	int				error;
 	xfs_agblock_t			realstart;
@@ -119,7 +120,7 @@ xfs_refcount_get_rec(
 
 	xfs_refcount_btrec_to_irec(rec, irec);
 
-	agno = cur->bc_ag.agno;
+	agno = cur->bc_ag.pag->pag_agno;
 	if (irec->rc_blockcount == 0 || irec->rc_blockcount > MAXREFCEXTLEN)
 		goto out_bad_rec;
 
@@ -144,7 +145,7 @@ xfs_refcount_get_rec(
 	if (irec->rc_refcount == 0 || irec->rc_refcount > MAXREFCOUNT)
 		goto out_bad_rec;
 
-	trace_xfs_refcount_get(cur->bc_mp, cur->bc_ag.agno, irec);
+	trace_xfs_refcount_get(cur->bc_mp, cur->bc_ag.pag->pag_agno, irec);
 	return 0;
 
 out_bad_rec:
@@ -169,14 +170,14 @@ xfs_refcount_update(
 	union xfs_btree_rec	rec;
 	int			error;
 
-	trace_xfs_refcount_update(cur->bc_mp, cur->bc_ag.agno, irec);
+	trace_xfs_refcount_update(cur->bc_mp, cur->bc_ag.pag->pag_agno, irec);
 	rec.refc.rc_startblock = cpu_to_be32(irec->rc_startblock);
 	rec.refc.rc_blockcount = cpu_to_be32(irec->rc_blockcount);
 	rec.refc.rc_refcount = cpu_to_be32(irec->rc_refcount);
 	error = xfs_btree_update(cur, &rec);
 	if (error)
 		trace_xfs_refcount_update_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -193,7 +194,7 @@ xfs_refcount_insert(
 {
 	int				error;
 
-	trace_xfs_refcount_insert(cur->bc_mp, cur->bc_ag.agno, irec);
+	trace_xfs_refcount_insert(cur->bc_mp, cur->bc_ag.pag->pag_agno, irec);
 	cur->bc_rec.rc.rc_startblock = irec->rc_startblock;
 	cur->bc_rec.rc.rc_blockcount = irec->rc_blockcount;
 	cur->bc_rec.rc.rc_refcount = irec->rc_refcount;
@@ -208,7 +209,7 @@ xfs_refcount_insert(
 out_error:
 	if (error)
 		trace_xfs_refcount_insert_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -234,7 +235,7 @@ xfs_refcount_delete(
 		error = -EFSCORRUPTED;
 		goto out_error;
 	}
-	trace_xfs_refcount_delete(cur->bc_mp, cur->bc_ag.agno, &irec);
+	trace_xfs_refcount_delete(cur->bc_mp, cur->bc_ag.pag->pag_agno, &irec);
 	error = xfs_btree_delete(cur, i);
 	if (XFS_IS_CORRUPT(cur->bc_mp, *i != 1)) {
 		error = -EFSCORRUPTED;
@@ -246,7 +247,7 @@ xfs_refcount_delete(
 out_error:
 	if (error)
 		trace_xfs_refcount_delete_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -366,7 +367,7 @@ xfs_refcount_split_extent(
 		return 0;
 
 	*shape_changed = true;
-	trace_xfs_refcount_split_extent(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_split_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			&rcext, agbno);
 
 	/* Establish the right extent. */
@@ -391,7 +392,7 @@ xfs_refcount_split_extent(
 
 out_error:
 	trace_xfs_refcount_split_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -411,7 +412,7 @@ xfs_refcount_merge_center_extents(
 	int				found_rec;
 
 	trace_xfs_refcount_merge_center_extents(cur->bc_mp,
-			cur->bc_ag.agno, left, center, right);
+			cur->bc_ag.pag->pag_agno, left, center, right);
 
 	/*
 	 * Make sure the center and right extents are not in the btree.
@@ -468,7 +469,7 @@ xfs_refcount_merge_center_extents(
 
 out_error:
 	trace_xfs_refcount_merge_center_extents_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -487,7 +488,7 @@ xfs_refcount_merge_left_extent(
 	int				found_rec;
 
 	trace_xfs_refcount_merge_left_extent(cur->bc_mp,
-			cur->bc_ag.agno, left, cleft);
+			cur->bc_ag.pag->pag_agno, left, cleft);
 
 	/* If the extent at agbno (cleft) wasn't synthesized, remove it. */
 	if (cleft->rc_refcount > 1) {
@@ -530,7 +531,7 @@ xfs_refcount_merge_left_extent(
 
 out_error:
 	trace_xfs_refcount_merge_left_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -548,7 +549,7 @@ xfs_refcount_merge_right_extent(
 	int				found_rec;
 
 	trace_xfs_refcount_merge_right_extent(cur->bc_mp,
-			cur->bc_ag.agno, cright, right);
+			cur->bc_ag.pag->pag_agno, cright, right);
 
 	/*
 	 * If the extent ending at agbno+aglen (cright) wasn't synthesized,
@@ -594,7 +595,7 @@ xfs_refcount_merge_right_extent(
 
 out_error:
 	trace_xfs_refcount_merge_right_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -679,13 +680,13 @@ xfs_refcount_find_left_extents(
 		cleft->rc_blockcount = aglen;
 		cleft->rc_refcount = 1;
 	}
-	trace_xfs_refcount_find_left_extent(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_find_left_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			left, cleft, agbno);
 	return error;
 
 out_error:
 	trace_xfs_refcount_find_left_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -768,13 +769,13 @@ xfs_refcount_find_right_extents(
 		cright->rc_blockcount = aglen;
 		cright->rc_refcount = 1;
 	}
-	trace_xfs_refcount_find_right_extent(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_find_right_extent(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			cright, right, agbno + aglen);
 	return error;
 
 out_error:
 	trace_xfs_refcount_find_right_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -952,7 +953,7 @@ xfs_refcount_adjust_extents(
 					ext.rc_startblock - *agbno);
 			tmp.rc_refcount = 1 + adj;
 			trace_xfs_refcount_modify_extent(cur->bc_mp,
-					cur->bc_ag.agno, &tmp);
+					cur->bc_ag.pag->pag_agno, &tmp);
 
 			/*
 			 * Either cover the hole (increment) or
@@ -971,7 +972,7 @@ xfs_refcount_adjust_extents(
 				cur->bc_ag.refc.nr_ops++;
 			} else {
 				fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
-						cur->bc_ag.agno,
+						cur->bc_ag.pag->pag_agno,
 						tmp.rc_startblock);
 				xfs_bmap_add_free(cur->bc_tp, fsbno,
 						  tmp.rc_blockcount, oinfo);
@@ -998,7 +999,7 @@ xfs_refcount_adjust_extents(
 			goto skip;
 		ext.rc_refcount += adj;
 		trace_xfs_refcount_modify_extent(cur->bc_mp,
-				cur->bc_ag.agno, &ext);
+				cur->bc_ag.pag->pag_agno, &ext);
 		if (ext.rc_refcount > 1) {
 			error = xfs_refcount_update(cur, &ext);
 			if (error)
@@ -1016,7 +1017,7 @@ xfs_refcount_adjust_extents(
 			goto advloop;
 		} else {
 			fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
-					cur->bc_ag.agno,
+					cur->bc_ag.pag->pag_agno,
 					ext.rc_startblock);
 			xfs_bmap_add_free(cur->bc_tp, fsbno, ext.rc_blockcount,
 					  oinfo);
@@ -1035,7 +1036,7 @@ advloop:
 	return error;
 out_error:
 	trace_xfs_refcount_modify_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -1057,10 +1058,10 @@ xfs_refcount_adjust(
 	*new_agbno = agbno;
 	*new_aglen = aglen;
 	if (adj == XFS_REFCOUNT_ADJUST_INCREASE)
-		trace_xfs_refcount_increase(cur->bc_mp, cur->bc_ag.agno,
+		trace_xfs_refcount_increase(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				agbno, aglen);
 	else
-		trace_xfs_refcount_decrease(cur->bc_mp, cur->bc_ag.agno,
+		trace_xfs_refcount_decrease(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				agbno, aglen);
 
 	/*
@@ -1099,7 +1100,7 @@ xfs_refcount_adjust(
 	return 0;
 
 out_error:
-	trace_xfs_refcount_adjust_error(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_adjust_error(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			error, _RET_IP_);
 	return error;
 }
@@ -1142,30 +1143,30 @@ xfs_refcount_finish_one(
 	struct xfs_btree_cur		*rcur;
 	struct xfs_buf			*agbp = NULL;
 	int				error = 0;
-	xfs_agnumber_t			agno;
 	xfs_agblock_t			bno;
 	xfs_agblock_t			new_agbno;
 	unsigned long			nr_ops = 0;
 	int				shape_changes = 0;
+	struct xfs_perag		*pag;
 
-	agno = XFS_FSB_TO_AGNO(mp, startblock);
-	ASSERT(agno != NULLAGNUMBER);
+	pag = xfs_perag_get(mp, XFS_FSB_TO_AGNO(mp, startblock));
 	bno = XFS_FSB_TO_AGBNO(mp, startblock);
 
 	trace_xfs_refcount_deferred(mp, XFS_FSB_TO_AGNO(mp, startblock),
 			type, XFS_FSB_TO_AGBNO(mp, startblock),
 			blockcount);
 
-	if (XFS_TEST_ERROR(false, mp,
-			XFS_ERRTAG_REFCOUNT_FINISH_ONE))
-		return -EIO;
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_REFCOUNT_FINISH_ONE)) {
+		error = -EIO;
+		goto out_drop;
+	}
 
 	/*
 	 * If we haven't gotten a cursor or the cursor AG doesn't match
 	 * the startblock, get one now.
 	 */
 	rcur = *pcur;
-	if (rcur != NULL && rcur->bc_ag.agno != agno) {
+	if (rcur != NULL && rcur->bc_ag.pag != pag) {
 		nr_ops = rcur->bc_ag.refc.nr_ops;
 		shape_changes = rcur->bc_ag.refc.shape_changes;
 		xfs_refcount_finish_one_cleanup(tp, rcur, 0);
@@ -1173,12 +1174,12 @@ xfs_refcount_finish_one(
 		*pcur = NULL;
 	}
 	if (rcur == NULL) {
-		error = xfs_alloc_read_agf(tp->t_mountp, tp, agno,
+		error = xfs_alloc_read_agf(tp->t_mountp, tp, pag->pag_agno,
 				XFS_ALLOC_FLAG_FREEING, &agbp);
 		if (error)
-			return error;
+			goto out_drop;
 
-		rcur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno);
+		rcur = xfs_refcountbt_init_cursor(mp, tp, agbp, pag);
 		rcur->bc_ag.refc.nr_ops = nr_ops;
 		rcur->bc_ag.refc.shape_changes = shape_changes;
 	}
@@ -1188,12 +1189,12 @@ xfs_refcount_finish_one(
 	case XFS_REFCOUNT_INCREASE:
 		error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
 			new_len, XFS_REFCOUNT_ADJUST_INCREASE, NULL);
-		*new_fsb = XFS_AGB_TO_FSB(mp, agno, new_agbno);
+		*new_fsb = XFS_AGB_TO_FSB(mp, pag->pag_agno, new_agbno);
 		break;
 	case XFS_REFCOUNT_DECREASE:
 		error = xfs_refcount_adjust(rcur, bno, blockcount, &new_agbno,
 			new_len, XFS_REFCOUNT_ADJUST_DECREASE, NULL);
-		*new_fsb = XFS_AGB_TO_FSB(mp, agno, new_agbno);
+		*new_fsb = XFS_AGB_TO_FSB(mp, pag->pag_agno, new_agbno);
 		break;
 	case XFS_REFCOUNT_ALLOC_COW:
 		*new_fsb = startblock + blockcount;
@@ -1210,8 +1211,10 @@ xfs_refcount_finish_one(
 		error = -EFSCORRUPTED;
 	}
 	if (!error && *new_len > 0)
-		trace_xfs_refcount_finish_one_leftover(mp, agno, type,
+		trace_xfs_refcount_finish_one_leftover(mp, pag->pag_agno, type,
 				bno, blockcount, new_agbno, *new_len);
+out_drop:
+	xfs_perag_put(pag);
 	return error;
 }
 
@@ -1294,7 +1297,7 @@ xfs_refcount_find_shared(
 	int				have;
 	int				error;
 
-	trace_xfs_refcount_find_shared(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_find_shared(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			agbno, aglen);
 
 	/* By default, skip the whole range */
@@ -1374,12 +1377,12 @@ xfs_refcount_find_shared(
 
 done:
 	trace_xfs_refcount_find_shared_result(cur->bc_mp,
-			cur->bc_ag.agno, *fbno, *flen);
+			cur->bc_ag.pag->pag_agno, *fbno, *flen);
 
 out_error:
 	if (error)
 		trace_xfs_refcount_find_shared_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -1476,7 +1479,7 @@ xfs_refcount_adjust_cow_extents(
 		tmp.rc_blockcount = aglen;
 		tmp.rc_refcount = 1;
 		trace_xfs_refcount_modify_extent(cur->bc_mp,
-				cur->bc_ag.agno, &tmp);
+				cur->bc_ag.pag->pag_agno, &tmp);
 
 		error = xfs_refcount_insert(cur, &tmp,
 				&found_tmp);
@@ -1504,7 +1507,7 @@ xfs_refcount_adjust_cow_extents(
 
 		ext.rc_refcount = 0;
 		trace_xfs_refcount_modify_extent(cur->bc_mp,
-				cur->bc_ag.agno, &ext);
+				cur->bc_ag.pag->pag_agno, &ext);
 		error = xfs_refcount_delete(cur, &found_rec);
 		if (error)
 			goto out_error;
@@ -1520,7 +1523,7 @@ xfs_refcount_adjust_cow_extents(
 	return error;
 out_error:
 	trace_xfs_refcount_modify_extent_error(cur->bc_mp,
-			cur->bc_ag.agno, error, _RET_IP_);
+			cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -1566,7 +1569,7 @@ xfs_refcount_adjust_cow(
 	return 0;
 
 out_error:
-	trace_xfs_refcount_adjust_cow_error(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcount_adjust_cow_error(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			error, _RET_IP_);
 	return error;
 }
@@ -1580,7 +1583,7 @@ __xfs_refcount_cow_alloc(
 	xfs_agblock_t		agbno,
 	xfs_extlen_t		aglen)
 {
-	trace_xfs_refcount_cow_increase(rcur->bc_mp, rcur->bc_ag.agno,
+	trace_xfs_refcount_cow_increase(rcur->bc_mp, rcur->bc_ag.pag->pag_agno,
 			agbno, aglen);
 
 	/* Add refcount btree reservation */
@@ -1597,7 +1600,7 @@ __xfs_refcount_cow_free(
 	xfs_agblock_t		agbno,
 	xfs_extlen_t		aglen)
 {
-	trace_xfs_refcount_cow_decrease(rcur->bc_mp, rcur->bc_ag.agno,
+	trace_xfs_refcount_cow_decrease(rcur->bc_mp, rcur->bc_ag.pag->pag_agno,
 			agbno, aglen);
 
 	/* Remove refcount btree reservation */
@@ -1672,7 +1675,7 @@ xfs_refcount_recover_extent(
 int
 xfs_refcount_recover_cow_leftovers(
 	struct xfs_mount		*mp,
-	xfs_agnumber_t			agno)
+	struct xfs_perag		*pag)
 {
 	struct xfs_trans		*tp;
 	struct xfs_btree_cur		*cur;
@@ -1704,10 +1707,10 @@ xfs_refcount_recover_cow_leftovers(
 	if (error)
 		return error;
 
-	error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
+	error = xfs_alloc_read_agf(mp, tp, pag->pag_agno, 0, &agbp);
 	if (error)
 		goto out_trans;
-	cur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno);
+	cur = xfs_refcountbt_init_cursor(mp, tp, agbp, pag);
 
 	/* Find all the leftover CoW staging extents. */
 	memset(&low, 0, sizeof(low));
@@ -1729,11 +1732,12 @@ xfs_refcount_recover_cow_leftovers(
 		if (error)
 			goto out_free;
 
-		trace_xfs_refcount_recover_extent(mp, agno, &rr->rr_rrec);
+		trace_xfs_refcount_recover_extent(mp, pag->pag_agno,
+				&rr->rr_rrec);
 
 		/* Free the orphan record */
 		agbno = rr->rr_rrec.rc_startblock - XFS_REFC_COW_START;
-		fsb = XFS_AGB_TO_FSB(mp, agno, agbno);
+		fsb = XFS_AGB_TO_FSB(mp, pag->pag_agno, agbno);
 		xfs_refcount_free_cow_extent(tp, fsb,
 				rr->rr_rrec.rc_blockcount);
 
diff --git a/fs/xfs/libxfs/xfs_refcount.h b/fs/xfs/libxfs/xfs_refcount.h
index 209795539c8d..9f6e9aae4da0 100644
--- a/fs/xfs/libxfs/xfs_refcount.h
+++ b/fs/xfs/libxfs/xfs_refcount.h
@@ -6,6 +6,13 @@
 #ifndef __XFS_REFCOUNT_H__
 #define __XFS_REFCOUNT_H__
 
+struct xfs_trans;
+struct xfs_mount;
+struct xfs_perag;
+struct xfs_btree_cur;
+struct xfs_bmbt_irec;
+struct xfs_refcount_irec;
+
 extern int xfs_refcount_lookup_le(struct xfs_btree_cur *cur,
 		xfs_agblock_t bno, int *stat);
 extern int xfs_refcount_lookup_ge(struct xfs_btree_cur *cur,
@@ -50,7 +57,7 @@ void xfs_refcount_alloc_cow_extent(struct xfs_trans *tp, xfs_fsblock_t fsb,
 void xfs_refcount_free_cow_extent(struct xfs_trans *tp, xfs_fsblock_t fsb,
 		xfs_extlen_t len);
 extern int xfs_refcount_recover_cow_leftovers(struct xfs_mount *mp,
-		xfs_agnumber_t agno);
+		struct xfs_perag *pag);
 
 /*
  * While we're adjusting the refcounts records of an extent, we have
diff --git a/fs/xfs/libxfs/xfs_refcount_btree.c b/fs/xfs/libxfs/xfs_refcount_btree.c
index a6ac60ae9421..92d336c17e83 100644
--- a/fs/xfs/libxfs/xfs_refcount_btree.c
+++ b/fs/xfs/libxfs/xfs_refcount_btree.c
@@ -9,7 +9,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_btree.h"
 #include "xfs_btree_staging.h"
@@ -20,13 +19,14 @@
 #include "xfs_trans.h"
 #include "xfs_bit.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 
 static struct xfs_btree_cur *
 xfs_refcountbt_dup_cursor(
 	struct xfs_btree_cur	*cur)
 {
 	return xfs_refcountbt_init_cursor(cur->bc_mp, cur->bc_tp,
-			cur->bc_ag.agbp, cur->bc_ag.agno);
+			cur->bc_ag.agbp, cur->bc_ag.pag);
 }
 
 STATIC void
@@ -65,7 +65,7 @@ xfs_refcountbt_alloc_block(
 	args.tp = cur->bc_tp;
 	args.mp = cur->bc_mp;
 	args.type = XFS_ALLOCTYPE_NEAR_BNO;
-	args.fsbno = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.agno,
+	args.fsbno = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			xfs_refc_block(args.mp));
 	args.oinfo = XFS_RMAP_OINFO_REFC;
 	args.minlen = args.maxlen = args.prod = 1;
@@ -74,13 +74,13 @@ xfs_refcountbt_alloc_block(
 	error = xfs_alloc_vextent(&args);
 	if (error)
 		goto out_error;
-	trace_xfs_refcountbt_alloc_block(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcountbt_alloc_block(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			args.agbno, 1);
 	if (args.fsbno == NULLFSBLOCK) {
 		*stat = 0;
 		return 0;
 	}
-	ASSERT(args.agno == cur->bc_ag.agno);
+	ASSERT(args.agno == cur->bc_ag.pag->pag_agno);
 	ASSERT(args.len == 1);
 
 	new->s = cpu_to_be32(args.agbno);
@@ -105,7 +105,7 @@ xfs_refcountbt_free_block(
 	xfs_fsblock_t		fsbno = XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(bp));
 	int			error;
 
-	trace_xfs_refcountbt_free_block(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_refcountbt_free_block(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			XFS_FSB_TO_AGBNO(cur->bc_mp, fsbno), 1);
 	be32_add_cpu(&agf->agf_refcount_blocks, -1);
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_REFCOUNT_BLOCKS);
@@ -170,7 +170,7 @@ xfs_refcountbt_init_ptr_from_cur(
 {
 	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_ag.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
 
 	ptr->s = agf->agf_refcount_root;
 }
@@ -316,12 +316,11 @@ static struct xfs_btree_cur *
 xfs_refcountbt_init_common(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_btree_cur	*cur;
 
-	ASSERT(agno != NULLAGNUMBER);
-	ASSERT(agno < mp->m_sb.sb_agcount);
+	ASSERT(pag->pag_agno < mp->m_sb.sb_agcount);
 
 	cur = kmem_cache_zalloc(xfs_btree_cur_zone, GFP_NOFS | __GFP_NOFAIL);
 	cur->bc_tp = tp;
@@ -330,9 +329,12 @@ xfs_refcountbt_init_common(
 	cur->bc_blocklog = mp->m_sb.sb_blocklog;
 	cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_refcbt_2);
 
-	cur->bc_ag.agno = agno;
 	cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
 
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+
 	cur->bc_ag.refc.nr_ops = 0;
 	cur->bc_ag.refc.shape_changes = 0;
 	cur->bc_ops = &xfs_refcountbt_ops;
@@ -345,12 +347,12 @@ xfs_refcountbt_init_cursor(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_agf		*agf = agbp->b_addr;
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_refcountbt_init_common(mp, tp, agno);
+	cur = xfs_refcountbt_init_common(mp, tp, pag);
 	cur->bc_nlevels = be32_to_cpu(agf->agf_refcount_level);
 	cur->bc_ag.agbp = agbp;
 	return cur;
@@ -361,11 +363,11 @@ struct xfs_btree_cur *
 xfs_refcountbt_stage_cursor(
 	struct xfs_mount	*mp,
 	struct xbtree_afakeroot	*afake,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_refcountbt_init_common(mp, NULL, agno);
+	cur = xfs_refcountbt_init_common(mp, NULL, pag);
 	xfs_btree_stage_afakeroot(cur, afake);
 	return cur;
 }
@@ -450,7 +452,7 @@ int
 xfs_refcountbt_calc_reserves(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_extlen_t		*ask,
 	xfs_extlen_t		*used)
 {
@@ -463,8 +465,7 @@ xfs_refcountbt_calc_reserves(
 	if (!xfs_sb_version_hasreflink(&mp->m_sb))
 		return 0;
 
-
-	error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
+	error = xfs_alloc_read_agf(mp, tp, pag->pag_agno, 0, &agbp);
 	if (error)
 		return error;
 
@@ -479,7 +480,7 @@ xfs_refcountbt_calc_reserves(
 	 * expansion.  We therefore can pretend the space isn't there.
 	 */
 	if (mp->m_sb.sb_logstart &&
-	    XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart) == agno)
+	    XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart) == pag->pag_agno)
 		agblocks -= mp->m_sb.sb_logblocks;
 
 	*ask += xfs_refcountbt_max_size(mp, agblocks);
diff --git a/fs/xfs/libxfs/xfs_refcount_btree.h b/fs/xfs/libxfs/xfs_refcount_btree.h
index 69dc515db671..bd9ed9e1e41f 100644
--- a/fs/xfs/libxfs/xfs_refcount_btree.h
+++ b/fs/xfs/libxfs/xfs_refcount_btree.h
@@ -13,6 +13,7 @@
 struct xfs_buf;
 struct xfs_btree_cur;
 struct xfs_mount;
+struct xfs_perag;
 struct xbtree_afakeroot;
 
 /*
@@ -46,9 +47,9 @@ struct xbtree_afakeroot;
 
 extern struct xfs_btree_cur *xfs_refcountbt_init_cursor(struct xfs_mount *mp,
 		struct xfs_trans *tp, struct xfs_buf *agbp,
-		xfs_agnumber_t agno);
+		struct xfs_perag *pag);
 struct xfs_btree_cur *xfs_refcountbt_stage_cursor(struct xfs_mount *mp,
-		struct xbtree_afakeroot *afake, xfs_agnumber_t agno);
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag);
 extern int xfs_refcountbt_maxrecs(int blocklen, bool leaf);
 extern void xfs_refcountbt_compute_maxlevels(struct xfs_mount *mp);
 
@@ -58,7 +59,7 @@ extern xfs_extlen_t xfs_refcountbt_max_size(struct xfs_mount *mp,
 		xfs_agblock_t agblocks);
 
 extern int xfs_refcountbt_calc_reserves(struct xfs_mount *mp,
-		struct xfs_trans *tp, xfs_agnumber_t agno, xfs_extlen_t *ask,
+		struct xfs_trans *tp, struct xfs_perag *pag, xfs_extlen_t *ask,
 		xfs_extlen_t *used);
 
 void xfs_refcountbt_commit_staged_btree(struct xfs_btree_cur *cur,
diff --git a/fs/xfs/libxfs/xfs_rmap.c b/fs/xfs/libxfs/xfs_rmap.c
index 10e0cf9949a2..d1dfad0204e3 100644
--- a/fs/xfs/libxfs/xfs_rmap.c
+++ b/fs/xfs/libxfs/xfs_rmap.c
@@ -11,6 +11,7 @@
 #include "xfs_trans_resv.h"
 #include "xfs_bit.h"
 #include "xfs_mount.h"
+#include "xfs_sb.h"
 #include "xfs_defer.h"
 #include "xfs_btree.h"
 #include "xfs_trans.h"
@@ -21,6 +22,7 @@
 #include "xfs_errortag.h"
 #include "xfs_error.h"
 #include "xfs_inode.h"
+#include "xfs_ag.h"
 
 /*
  * Lookup the first record less than or equal to [bno, len, owner, offset]
@@ -79,7 +81,7 @@ xfs_rmap_update(
 	union xfs_btree_rec	rec;
 	int			error;
 
-	trace_xfs_rmap_update(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_rmap_update(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 			irec->rm_startblock, irec->rm_blockcount,
 			irec->rm_owner, irec->rm_offset, irec->rm_flags);
 
@@ -91,7 +93,7 @@ xfs_rmap_update(
 	error = xfs_btree_update(cur, &rec);
 	if (error)
 		trace_xfs_rmap_update_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -107,7 +109,7 @@ xfs_rmap_insert(
 	int			i;
 	int			error;
 
-	trace_xfs_rmap_insert(rcur->bc_mp, rcur->bc_ag.agno, agbno,
+	trace_xfs_rmap_insert(rcur->bc_mp, rcur->bc_ag.pag->pag_agno, agbno,
 			len, owner, offset, flags);
 
 	error = xfs_rmap_lookup_eq(rcur, agbno, len, owner, offset, flags, &i);
@@ -133,7 +135,7 @@ xfs_rmap_insert(
 done:
 	if (error)
 		trace_xfs_rmap_insert_error(rcur->bc_mp,
-				rcur->bc_ag.agno, error, _RET_IP_);
+				rcur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -149,7 +151,7 @@ xfs_rmap_delete(
 	int			i;
 	int			error;
 
-	trace_xfs_rmap_delete(rcur->bc_mp, rcur->bc_ag.agno, agbno,
+	trace_xfs_rmap_delete(rcur->bc_mp, rcur->bc_ag.pag->pag_agno, agbno,
 			len, owner, offset, flags);
 
 	error = xfs_rmap_lookup_eq(rcur, agbno, len, owner, offset, flags, &i);
@@ -170,7 +172,7 @@ xfs_rmap_delete(
 done:
 	if (error)
 		trace_xfs_rmap_delete_error(rcur->bc_mp,
-				rcur->bc_ag.agno, error, _RET_IP_);
+				rcur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -197,7 +199,7 @@ xfs_rmap_get_rec(
 	int			*stat)
 {
 	struct xfs_mount	*mp = cur->bc_mp;
-	xfs_agnumber_t		agno = cur->bc_ag.agno;
+	xfs_agnumber_t		agno = cur->bc_ag.pag->pag_agno;
 	union xfs_btree_rec	*rec;
 	int			error;
 
@@ -260,7 +262,7 @@ xfs_rmap_find_left_neighbor_helper(
 	struct xfs_find_left_neighbor_info	*info = priv;
 
 	trace_xfs_rmap_find_left_neighbor_candidate(cur->bc_mp,
-			cur->bc_ag.agno, rec->rm_startblock,
+			cur->bc_ag.pag->pag_agno, rec->rm_startblock,
 			rec->rm_blockcount, rec->rm_owner, rec->rm_offset,
 			rec->rm_flags);
 
@@ -312,7 +314,7 @@ xfs_rmap_find_left_neighbor(
 	info.stat = stat;
 
 	trace_xfs_rmap_find_left_neighbor_query(cur->bc_mp,
-			cur->bc_ag.agno, bno, 0, owner, offset, flags);
+			cur->bc_ag.pag->pag_agno, bno, 0, owner, offset, flags);
 
 	error = xfs_rmap_query_range(cur, &info.high, &info.high,
 			xfs_rmap_find_left_neighbor_helper, &info);
@@ -320,7 +322,7 @@ xfs_rmap_find_left_neighbor(
 		error = 0;
 	if (*stat)
 		trace_xfs_rmap_find_left_neighbor_result(cur->bc_mp,
-				cur->bc_ag.agno, irec->rm_startblock,
+				cur->bc_ag.pag->pag_agno, irec->rm_startblock,
 				irec->rm_blockcount, irec->rm_owner,
 				irec->rm_offset, irec->rm_flags);
 	return error;
@@ -336,7 +338,7 @@ xfs_rmap_lookup_le_range_helper(
 	struct xfs_find_left_neighbor_info	*info = priv;
 
 	trace_xfs_rmap_lookup_le_range_candidate(cur->bc_mp,
-			cur->bc_ag.agno, rec->rm_startblock,
+			cur->bc_ag.pag->pag_agno, rec->rm_startblock,
 			rec->rm_blockcount, rec->rm_owner, rec->rm_offset,
 			rec->rm_flags);
 
@@ -385,14 +387,14 @@ xfs_rmap_lookup_le_range(
 	info.stat = stat;
 
 	trace_xfs_rmap_lookup_le_range(cur->bc_mp,
-			cur->bc_ag.agno, bno, 0, owner, offset, flags);
+			cur->bc_ag.pag->pag_agno, bno, 0, owner, offset, flags);
 	error = xfs_rmap_query_range(cur, &info.high, &info.high,
 			xfs_rmap_lookup_le_range_helper, &info);
 	if (error == -ECANCELED)
 		error = 0;
 	if (*stat)
 		trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
-				cur->bc_ag.agno, irec->rm_startblock,
+				cur->bc_ag.pag->pag_agno, irec->rm_startblock,
 				irec->rm_blockcount, irec->rm_owner,
 				irec->rm_offset, irec->rm_flags);
 	return error;
@@ -498,7 +500,7 @@ xfs_rmap_unmap(
 			(flags & XFS_RMAP_BMBT_BLOCK);
 	if (unwritten)
 		flags |= XFS_RMAP_UNWRITTEN;
-	trace_xfs_rmap_unmap(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_unmap(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 
 	/*
@@ -522,7 +524,7 @@ xfs_rmap_unmap(
 		goto out_error;
 	}
 	trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
-			cur->bc_ag.agno, ltrec.rm_startblock,
+			cur->bc_ag.pag->pag_agno, ltrec.rm_startblock,
 			ltrec.rm_blockcount, ltrec.rm_owner,
 			ltrec.rm_offset, ltrec.rm_flags);
 	ltoff = ltrec.rm_offset;
@@ -588,7 +590,7 @@ xfs_rmap_unmap(
 
 	if (ltrec.rm_startblock == bno && ltrec.rm_blockcount == len) {
 		/* exact match, simply remove the record from rmap tree */
-		trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 				ltrec.rm_startblock, ltrec.rm_blockcount,
 				ltrec.rm_owner, ltrec.rm_offset,
 				ltrec.rm_flags);
@@ -666,7 +668,7 @@ xfs_rmap_unmap(
 		else
 			cur->bc_rec.r.rm_offset = offset + len;
 		cur->bc_rec.r.rm_flags = flags;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno,
 				cur->bc_rec.r.rm_startblock,
 				cur->bc_rec.r.rm_blockcount,
 				cur->bc_rec.r.rm_owner,
@@ -678,11 +680,11 @@ xfs_rmap_unmap(
 	}
 
 out_done:
-	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 out_error:
 	if (error)
-		trace_xfs_rmap_unmap_error(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_unmap_error(mp, cur->bc_ag.pag->pag_agno,
 				error, _RET_IP_);
 	return error;
 }
@@ -694,7 +696,7 @@ int
 xfs_rmap_free(
 	struct xfs_trans		*tp,
 	struct xfs_buf			*agbp,
-	xfs_agnumber_t			agno,
+	struct xfs_perag		*pag,
 	xfs_agblock_t			bno,
 	xfs_extlen_t			len,
 	const struct xfs_owner_info	*oinfo)
@@ -706,7 +708,7 @@ xfs_rmap_free(
 	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
 		return 0;
 
-	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, agno);
+	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
 
 	error = xfs_rmap_unmap(cur, bno, len, false, oinfo);
 
@@ -773,7 +775,7 @@ xfs_rmap_map(
 			(flags & XFS_RMAP_BMBT_BLOCK);
 	if (unwritten)
 		flags |= XFS_RMAP_UNWRITTEN;
-	trace_xfs_rmap_map(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_map(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 	ASSERT(!xfs_rmap_should_skip_owner_update(oinfo));
 
@@ -795,7 +797,7 @@ xfs_rmap_map(
 			goto out_error;
 		}
 		trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
-				cur->bc_ag.agno, ltrec.rm_startblock,
+				cur->bc_ag.pag->pag_agno, ltrec.rm_startblock,
 				ltrec.rm_blockcount, ltrec.rm_owner,
 				ltrec.rm_offset, ltrec.rm_flags);
 
@@ -831,7 +833,7 @@ xfs_rmap_map(
 			goto out_error;
 		}
 		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
-			cur->bc_ag.agno, gtrec.rm_startblock,
+			cur->bc_ag.pag->pag_agno, gtrec.rm_startblock,
 			gtrec.rm_blockcount, gtrec.rm_owner,
 			gtrec.rm_offset, gtrec.rm_flags);
 		if (!xfs_rmap_is_mergeable(&gtrec, owner, flags))
@@ -870,7 +872,7 @@ xfs_rmap_map(
 			 * result: |rrrrrrrrrrrrrrrrrrrrrrrrrrrrr|
 			 */
 			ltrec.rm_blockcount += gtrec.rm_blockcount;
-			trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+			trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 					gtrec.rm_startblock,
 					gtrec.rm_blockcount,
 					gtrec.rm_owner,
@@ -921,7 +923,7 @@ xfs_rmap_map(
 		cur->bc_rec.r.rm_owner = owner;
 		cur->bc_rec.r.rm_offset = offset;
 		cur->bc_rec.r.rm_flags = flags;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno, bno, len,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			owner, offset, flags);
 		error = xfs_btree_insert(cur, &i);
 		if (error)
@@ -932,11 +934,11 @@ xfs_rmap_map(
 		}
 	}
 
-	trace_xfs_rmap_map_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_map_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 out_error:
 	if (error)
-		trace_xfs_rmap_map_error(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_map_error(mp, cur->bc_ag.pag->pag_agno,
 				error, _RET_IP_);
 	return error;
 }
@@ -948,7 +950,7 @@ int
 xfs_rmap_alloc(
 	struct xfs_trans		*tp,
 	struct xfs_buf			*agbp,
-	xfs_agnumber_t			agno,
+	struct xfs_perag		*pag,
 	xfs_agblock_t			bno,
 	xfs_extlen_t			len,
 	const struct xfs_owner_info	*oinfo)
@@ -960,7 +962,7 @@ xfs_rmap_alloc(
 	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
 		return 0;
 
-	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, agno);
+	cur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
 	error = xfs_rmap_map(cur, bno, len, false, oinfo);
 
 	xfs_btree_del_cursor(cur, error);
@@ -1010,7 +1012,7 @@ xfs_rmap_convert(
 			(flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK))));
 	oldext = unwritten ? XFS_RMAP_UNWRITTEN : 0;
 	new_endoff = offset + len;
-	trace_xfs_rmap_convert(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_convert(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 
 	/*
@@ -1034,7 +1036,7 @@ xfs_rmap_convert(
 		goto done;
 	}
 	trace_xfs_rmap_lookup_le_range_result(cur->bc_mp,
-			cur->bc_ag.agno, PREV.rm_startblock,
+			cur->bc_ag.pag->pag_agno, PREV.rm_startblock,
 			PREV.rm_blockcount, PREV.rm_owner,
 			PREV.rm_offset, PREV.rm_flags);
 
@@ -1076,7 +1078,7 @@ xfs_rmap_convert(
 			goto done;
 		}
 		trace_xfs_rmap_find_left_neighbor_result(cur->bc_mp,
-				cur->bc_ag.agno, LEFT.rm_startblock,
+				cur->bc_ag.pag->pag_agno, LEFT.rm_startblock,
 				LEFT.rm_blockcount, LEFT.rm_owner,
 				LEFT.rm_offset, LEFT.rm_flags);
 		if (LEFT.rm_startblock + LEFT.rm_blockcount == bno &&
@@ -1114,7 +1116,7 @@ xfs_rmap_convert(
 			goto done;
 		}
 		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
-				cur->bc_ag.agno, RIGHT.rm_startblock,
+				cur->bc_ag.pag->pag_agno, RIGHT.rm_startblock,
 				RIGHT.rm_blockcount, RIGHT.rm_owner,
 				RIGHT.rm_offset, RIGHT.rm_flags);
 		if (bno + len == RIGHT.rm_startblock &&
@@ -1132,7 +1134,7 @@ xfs_rmap_convert(
 	     RIGHT.rm_blockcount > XFS_RMAP_LEN_MAX)
 		state &= ~RMAP_RIGHT_CONTIG;
 
-	trace_xfs_rmap_convert_state(mp, cur->bc_ag.agno, state,
+	trace_xfs_rmap_convert_state(mp, cur->bc_ag.pag->pag_agno, state,
 			_RET_IP_);
 
 	/* reset the cursor back to PREV */
@@ -1162,7 +1164,7 @@ xfs_rmap_convert(
 			error = -EFSCORRUPTED;
 			goto done;
 		}
-		trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 				RIGHT.rm_startblock, RIGHT.rm_blockcount,
 				RIGHT.rm_owner, RIGHT.rm_offset,
 				RIGHT.rm_flags);
@@ -1180,7 +1182,7 @@ xfs_rmap_convert(
 			error = -EFSCORRUPTED;
 			goto done;
 		}
-		trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 				PREV.rm_startblock, PREV.rm_blockcount,
 				PREV.rm_owner, PREV.rm_offset,
 				PREV.rm_flags);
@@ -1210,7 +1212,7 @@ xfs_rmap_convert(
 		 * Setting all of a previous oldext extent to newext.
 		 * The left neighbor is contiguous, the right is not.
 		 */
-		trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 				PREV.rm_startblock, PREV.rm_blockcount,
 				PREV.rm_owner, PREV.rm_offset,
 				PREV.rm_flags);
@@ -1247,7 +1249,7 @@ xfs_rmap_convert(
 			error = -EFSCORRUPTED;
 			goto done;
 		}
-		trace_xfs_rmap_delete(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_delete(mp, cur->bc_ag.pag->pag_agno,
 				RIGHT.rm_startblock, RIGHT.rm_blockcount,
 				RIGHT.rm_owner, RIGHT.rm_offset,
 				RIGHT.rm_flags);
@@ -1326,7 +1328,7 @@ xfs_rmap_convert(
 		NEW.rm_blockcount = len;
 		NEW.rm_flags = newext;
 		cur->bc_rec.r = NEW;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno, bno,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno,
 				len, owner, offset, newext);
 		error = xfs_btree_insert(cur, &i);
 		if (error)
@@ -1383,7 +1385,7 @@ xfs_rmap_convert(
 		NEW.rm_blockcount = len;
 		NEW.rm_flags = newext;
 		cur->bc_rec.r = NEW;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno, bno,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno,
 				len, owner, offset, newext);
 		error = xfs_btree_insert(cur, &i);
 		if (error)
@@ -1414,7 +1416,7 @@ xfs_rmap_convert(
 		NEW = PREV;
 		NEW.rm_blockcount = offset - PREV.rm_offset;
 		cur->bc_rec.r = NEW;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno,
 				NEW.rm_startblock, NEW.rm_blockcount,
 				NEW.rm_owner, NEW.rm_offset,
 				NEW.rm_flags);
@@ -1441,7 +1443,7 @@ xfs_rmap_convert(
 		/* new middle extent - newext */
 		cur->bc_rec.r.rm_flags &= ~XFS_RMAP_UNWRITTEN;
 		cur->bc_rec.r.rm_flags |= newext;
-		trace_xfs_rmap_insert(mp, cur->bc_ag.agno, bno, len,
+		trace_xfs_rmap_insert(mp, cur->bc_ag.pag->pag_agno, bno, len,
 				owner, offset, newext);
 		error = xfs_btree_insert(cur, &i);
 		if (error)
@@ -1465,12 +1467,12 @@ xfs_rmap_convert(
 		ASSERT(0);
 	}
 
-	trace_xfs_rmap_convert_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_convert_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 done:
 	if (error)
 		trace_xfs_rmap_convert_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -1506,7 +1508,7 @@ xfs_rmap_convert_shared(
 			(flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK))));
 	oldext = unwritten ? XFS_RMAP_UNWRITTEN : 0;
 	new_endoff = offset + len;
-	trace_xfs_rmap_convert(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_convert(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 
 	/*
@@ -1573,7 +1575,7 @@ xfs_rmap_convert_shared(
 			goto done;
 		}
 		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
-				cur->bc_ag.agno, RIGHT.rm_startblock,
+				cur->bc_ag.pag->pag_agno, RIGHT.rm_startblock,
 				RIGHT.rm_blockcount, RIGHT.rm_owner,
 				RIGHT.rm_offset, RIGHT.rm_flags);
 		if (xfs_rmap_is_mergeable(&RIGHT, owner, newext))
@@ -1589,7 +1591,7 @@ xfs_rmap_convert_shared(
 	     RIGHT.rm_blockcount > XFS_RMAP_LEN_MAX)
 		state &= ~RMAP_RIGHT_CONTIG;
 
-	trace_xfs_rmap_convert_state(mp, cur->bc_ag.agno, state,
+	trace_xfs_rmap_convert_state(mp, cur->bc_ag.pag->pag_agno, state,
 			_RET_IP_);
 	/*
 	 * Switch out based on the FILLING and CONTIG state bits.
@@ -1880,12 +1882,12 @@ xfs_rmap_convert_shared(
 		ASSERT(0);
 	}
 
-	trace_xfs_rmap_convert_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_convert_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 done:
 	if (error)
 		trace_xfs_rmap_convert_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -1923,7 +1925,7 @@ xfs_rmap_unmap_shared(
 	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
 	if (unwritten)
 		flags |= XFS_RMAP_UNWRITTEN;
-	trace_xfs_rmap_unmap(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_unmap(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 
 	/*
@@ -2072,12 +2074,12 @@ xfs_rmap_unmap_shared(
 			goto out_error;
 	}
 
-	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_unmap_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 out_error:
 	if (error)
 		trace_xfs_rmap_unmap_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -2112,7 +2114,7 @@ xfs_rmap_map_shared(
 	xfs_owner_info_unpack(oinfo, &owner, &offset, &flags);
 	if (unwritten)
 		flags |= XFS_RMAP_UNWRITTEN;
-	trace_xfs_rmap_map(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_map(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 
 	/* Is there a left record that abuts our range? */
@@ -2138,7 +2140,7 @@ xfs_rmap_map_shared(
 			goto out_error;
 		}
 		trace_xfs_rmap_find_right_neighbor_result(cur->bc_mp,
-			cur->bc_ag.agno, gtrec.rm_startblock,
+			cur->bc_ag.pag->pag_agno, gtrec.rm_startblock,
 			gtrec.rm_blockcount, gtrec.rm_owner,
 			gtrec.rm_offset, gtrec.rm_flags);
 
@@ -2231,12 +2233,12 @@ xfs_rmap_map_shared(
 			goto out_error;
 	}
 
-	trace_xfs_rmap_map_done(mp, cur->bc_ag.agno, bno, len,
+	trace_xfs_rmap_map_done(mp, cur->bc_ag.pag->pag_agno, bno, len,
 			unwritten, oinfo);
 out_error:
 	if (error)
 		trace_xfs_rmap_map_error(cur->bc_mp,
-				cur->bc_ag.agno, error, _RET_IP_);
+				cur->bc_ag.pag->pag_agno, error, _RET_IP_);
 	return error;
 }
 
@@ -2362,31 +2364,32 @@ xfs_rmap_finish_one(
 	struct xfs_btree_cur		**pcur)
 {
 	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_perag		*pag;
 	struct xfs_btree_cur		*rcur;
 	struct xfs_buf			*agbp = NULL;
 	int				error = 0;
-	xfs_agnumber_t			agno;
 	struct xfs_owner_info		oinfo;
 	xfs_agblock_t			bno;
 	bool				unwritten;
 
-	agno = XFS_FSB_TO_AGNO(mp, startblock);
-	ASSERT(agno != NULLAGNUMBER);
+	pag = xfs_perag_get(mp, XFS_FSB_TO_AGNO(mp, startblock));
 	bno = XFS_FSB_TO_AGBNO(mp, startblock);
 
-	trace_xfs_rmap_deferred(mp, agno, type, bno, owner, whichfork,
+	trace_xfs_rmap_deferred(mp, pag->pag_agno, type, bno, owner, whichfork,
 			startoff, blockcount, state);
 
-	if (XFS_TEST_ERROR(false, mp,
-			XFS_ERRTAG_RMAP_FINISH_ONE))
-		return -EIO;
+	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_RMAP_FINISH_ONE)) {
+		error = -EIO;
+		goto out_drop;
+	}
+
 
 	/*
 	 * If we haven't gotten a cursor or the cursor AG doesn't match
 	 * the startblock, get one now.
 	 */
 	rcur = *pcur;
-	if (rcur != NULL && rcur->bc_ag.agno != agno) {
+	if (rcur != NULL && rcur->bc_ag.pag != pag) {
 		xfs_rmap_finish_one_cleanup(tp, rcur, 0);
 		rcur = NULL;
 		*pcur = NULL;
@@ -2397,13 +2400,15 @@ xfs_rmap_finish_one(
 		 * rmapbt, because a shape change could cause us to
 		 * allocate blocks.
 		 */
-		error = xfs_free_extent_fix_freelist(tp, agno, &agbp);
+		error = xfs_free_extent_fix_freelist(tp, pag, &agbp);
 		if (error)
-			return error;
-		if (XFS_IS_CORRUPT(tp->t_mountp, !agbp))
-			return -EFSCORRUPTED;
+			goto out_drop;
+		if (XFS_IS_CORRUPT(tp->t_mountp, !agbp)) {
+			error = -EFSCORRUPTED;
+			goto out_drop;
+		}
 
-		rcur = xfs_rmapbt_init_cursor(mp, tp, agbp, agno);
+		rcur = xfs_rmapbt_init_cursor(mp, tp, agbp, pag);
 	}
 	*pcur = rcur;
 
@@ -2441,6 +2446,8 @@ xfs_rmap_finish_one(
 		ASSERT(0);
 		error = -EFSCORRUPTED;
 	}
+out_drop:
+	xfs_perag_put(pag);
 	return error;
 }
 
diff --git a/fs/xfs/libxfs/xfs_rmap.h b/fs/xfs/libxfs/xfs_rmap.h
index abe633403fd1..f2423cf7f1e2 100644
--- a/fs/xfs/libxfs/xfs_rmap.h
+++ b/fs/xfs/libxfs/xfs_rmap.h
@@ -6,6 +6,8 @@
 #ifndef __XFS_RMAP_H__
 #define __XFS_RMAP_H__
 
+struct xfs_perag;
+
 static inline void
 xfs_rmap_ino_bmbt_owner(
 	struct xfs_owner_info	*oi,
@@ -113,10 +115,10 @@ xfs_owner_info_pack(
 }
 
 int xfs_rmap_alloc(struct xfs_trans *tp, struct xfs_buf *agbp,
-		   xfs_agnumber_t agno, xfs_agblock_t bno, xfs_extlen_t len,
+		   struct xfs_perag *pag, xfs_agblock_t bno, xfs_extlen_t len,
 		   const struct xfs_owner_info *oinfo);
 int xfs_rmap_free(struct xfs_trans *tp, struct xfs_buf *agbp,
-		  xfs_agnumber_t agno, xfs_agblock_t bno, xfs_extlen_t len,
+		  struct xfs_perag *pag, xfs_agblock_t bno, xfs_extlen_t len,
 		  const struct xfs_owner_info *oinfo);
 
 int xfs_rmap_lookup_le(struct xfs_btree_cur *cur, xfs_agblock_t bno,
diff --git a/fs/xfs/libxfs/xfs_rmap_btree.c b/fs/xfs/libxfs/xfs_rmap_btree.c
index 9f5bcbd834c3..f29bc71b9950 100644
--- a/fs/xfs/libxfs/xfs_rmap_btree.c
+++ b/fs/xfs/libxfs/xfs_rmap_btree.c
@@ -9,7 +9,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_trans.h"
 #include "xfs_alloc.h"
@@ -20,6 +19,7 @@
 #include "xfs_trace.h"
 #include "xfs_error.h"
 #include "xfs_extent_busy.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 
 /*
@@ -52,7 +52,7 @@ xfs_rmapbt_dup_cursor(
 	struct xfs_btree_cur	*cur)
 {
 	return xfs_rmapbt_init_cursor(cur->bc_mp, cur->bc_tp,
-			cur->bc_ag.agbp, cur->bc_ag.agno);
+				cur->bc_ag.agbp, cur->bc_ag.pag);
 }
 
 STATIC void
@@ -64,13 +64,12 @@ xfs_rmapbt_set_root(
 	struct xfs_buf		*agbp = cur->bc_ag.agbp;
 	struct xfs_agf		*agf = agbp->b_addr;
 	int			btnum = cur->bc_btnum;
-	struct xfs_perag	*pag = agbp->b_pag;
 
 	ASSERT(ptr->s != 0);
 
 	agf->agf_roots[btnum] = ptr->s;
 	be32_add_cpu(&agf->agf_levels[btnum], inc);
-	pag->pagf_levels[btnum] += inc;
+	cur->bc_ag.pag->pagf_levels[btnum] += inc;
 
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
 }
@@ -84,6 +83,7 @@ xfs_rmapbt_alloc_block(
 {
 	struct xfs_buf		*agbp = cur->bc_ag.agbp;
 	struct xfs_agf		*agf = agbp->b_addr;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
 	int			error;
 	xfs_agblock_t		bno;
 
@@ -93,21 +93,19 @@ xfs_rmapbt_alloc_block(
 	if (error)
 		return error;
 
-	trace_xfs_rmapbt_alloc_block(cur->bc_mp, cur->bc_ag.agno,
-			bno, 1);
+	trace_xfs_rmapbt_alloc_block(cur->bc_mp, pag->pag_agno, bno, 1);
 	if (bno == NULLAGBLOCK) {
 		*stat = 0;
 		return 0;
 	}
 
-	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_ag.agno, bno, 1,
-			false);
+	xfs_extent_busy_reuse(cur->bc_mp, pag, bno, 1, false);
 
 	new->s = cpu_to_be32(bno);
 	be32_add_cpu(&agf->agf_rmap_blocks, 1);
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS);
 
-	xfs_ag_resv_rmapbt_alloc(cur->bc_mp, cur->bc_ag.agno);
+	xfs_ag_resv_rmapbt_alloc(cur->bc_mp, pag->pag_agno);
 
 	*stat = 1;
 	return 0;
@@ -120,12 +118,12 @@ xfs_rmapbt_free_block(
 {
 	struct xfs_buf		*agbp = cur->bc_ag.agbp;
 	struct xfs_agf		*agf = agbp->b_addr;
-	struct xfs_perag	*pag;
+	struct xfs_perag	*pag = cur->bc_ag.pag;
 	xfs_agblock_t		bno;
 	int			error;
 
 	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
-	trace_xfs_rmapbt_free_block(cur->bc_mp, cur->bc_ag.agno,
+	trace_xfs_rmapbt_free_block(cur->bc_mp, pag->pag_agno,
 			bno, 1);
 	be32_add_cpu(&agf->agf_rmap_blocks, -1);
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_RMAP_BLOCKS);
@@ -133,10 +131,9 @@ xfs_rmapbt_free_block(
 	if (error)
 		return error;
 
-	xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
+	xfs_extent_busy_insert(cur->bc_tp, pag, bno, 1,
 			      XFS_EXTENT_BUSY_SKIP_DISCARD);
 
-	pag = cur->bc_ag.agbp->b_pag;
 	xfs_ag_resv_free_extent(pag, XFS_AG_RESV_RMAPBT, NULL, 1);
 	return 0;
 }
@@ -215,7 +212,7 @@ xfs_rmapbt_init_ptr_from_cur(
 {
 	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_ag.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(cur->bc_ag.pag->pag_agno == be32_to_cpu(agf->agf_seqno));
 
 	ptr->s = agf->agf_roots[cur->bc_btnum];
 }
@@ -450,7 +447,7 @@ static struct xfs_btree_cur *
 xfs_rmapbt_init_common(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_btree_cur	*cur;
 
@@ -462,9 +459,12 @@ xfs_rmapbt_init_common(
 	cur->bc_flags = XFS_BTREE_CRC_BLOCKS | XFS_BTREE_OVERLAPPING;
 	cur->bc_blocklog = mp->m_sb.sb_blocklog;
 	cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_rmap_2);
-	cur->bc_ag.agno = agno;
 	cur->bc_ops = &xfs_rmapbt_ops;
 
+	/* take a reference for the cursor */
+	atomic_inc(&pag->pag_ref);
+	cur->bc_ag.pag = pag;
+
 	return cur;
 }
 
@@ -474,12 +474,12 @@ xfs_rmapbt_init_cursor(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
 	struct xfs_buf		*agbp,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_agf		*agf = agbp->b_addr;
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_rmapbt_init_common(mp, tp, agno);
+	cur = xfs_rmapbt_init_common(mp, tp, pag);
 	cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAP]);
 	cur->bc_ag.agbp = agbp;
 	return cur;
@@ -490,11 +490,11 @@ struct xfs_btree_cur *
 xfs_rmapbt_stage_cursor(
 	struct xfs_mount	*mp,
 	struct xbtree_afakeroot	*afake,
-	xfs_agnumber_t		agno)
+	struct xfs_perag	*pag)
 {
 	struct xfs_btree_cur	*cur;
 
-	cur = xfs_rmapbt_init_common(mp, NULL, agno);
+	cur = xfs_rmapbt_init_common(mp, NULL, pag);
 	xfs_btree_stage_afakeroot(cur, afake);
 	return cur;
 }
@@ -596,7 +596,7 @@ int
 xfs_rmapbt_calc_reserves(
 	struct xfs_mount	*mp,
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_extlen_t		*ask,
 	xfs_extlen_t		*used)
 {
@@ -609,7 +609,7 @@ xfs_rmapbt_calc_reserves(
 	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
 		return 0;
 
-	error = xfs_alloc_read_agf(mp, tp, agno, 0, &agbp);
+	error = xfs_alloc_read_agf(mp, tp, pag->pag_agno, 0, &agbp);
 	if (error)
 		return error;
 
@@ -624,7 +624,7 @@ xfs_rmapbt_calc_reserves(
 	 * expansion.  We therefore can pretend the space isn't there.
 	 */
 	if (mp->m_sb.sb_logstart &&
-	    XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart) == agno)
+	    XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart) == pag->pag_agno)
 		agblocks -= mp->m_sb.sb_logblocks;
 
 	/* Reserve 1% of the AG or enough for 1 block per record. */
diff --git a/fs/xfs/libxfs/xfs_rmap_btree.h b/fs/xfs/libxfs/xfs_rmap_btree.h
index 115c3455a734..88d8d18788a2 100644
--- a/fs/xfs/libxfs/xfs_rmap_btree.h
+++ b/fs/xfs/libxfs/xfs_rmap_btree.h
@@ -43,9 +43,9 @@ struct xbtree_afakeroot;
 
 struct xfs_btree_cur *xfs_rmapbt_init_cursor(struct xfs_mount *mp,
 				struct xfs_trans *tp, struct xfs_buf *bp,
-				xfs_agnumber_t agno);
+				struct xfs_perag *pag);
 struct xfs_btree_cur *xfs_rmapbt_stage_cursor(struct xfs_mount *mp,
-		struct xbtree_afakeroot *afake, xfs_agnumber_t agno);
+		struct xbtree_afakeroot *afake, struct xfs_perag *pag);
 void xfs_rmapbt_commit_staged_btree(struct xfs_btree_cur *cur,
 		struct xfs_trans *tp, struct xfs_buf *agbp);
 int xfs_rmapbt_maxrecs(int blocklen, int leaf);
@@ -57,6 +57,6 @@ extern xfs_extlen_t xfs_rmapbt_max_size(struct xfs_mount *mp,
 		xfs_agblock_t agblocks);
 
 extern int xfs_rmapbt_calc_reserves(struct xfs_mount *mp, struct xfs_trans *tp,
-		xfs_agnumber_t agno, xfs_extlen_t *ask, xfs_extlen_t *used);
+		struct xfs_perag *pag, xfs_extlen_t *ask, xfs_extlen_t *used);
 
 #endif	/* __XFS_RMAP_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c
index dfbbcbd448c1..04f5386446db 100644
--- a/fs/xfs/libxfs/xfs_sb.c
+++ b/fs/xfs/libxfs/xfs_sb.c
@@ -15,7 +15,6 @@
 #include "xfs_ialloc.h"
 #include "xfs_alloc.h"
 #include "xfs_error.h"
-#include "xfs_trace.h"
 #include "xfs_trans.h"
 #include "xfs_buf_item.h"
 #include "xfs_bmap_btree.h"
@@ -25,72 +24,12 @@
 #include "xfs_refcount_btree.h"
 #include "xfs_da_format.h"
 #include "xfs_health.h"
+#include "xfs_ag.h"
 
 /*
  * Physical superblock buffer manipulations. Shared with libxfs in userspace.
  */
 
-/*
- * Reference counting access wrappers to the perag structures.
- * Because we never free per-ag structures, the only thing we
- * have to protect against changes is the tree structure itself.
- */
-struct xfs_perag *
-xfs_perag_get(
-	struct xfs_mount	*mp,
-	xfs_agnumber_t		agno)
-{
-	struct xfs_perag	*pag;
-	int			ref = 0;
-
-	rcu_read_lock();
-	pag = radix_tree_lookup(&mp->m_perag_tree, agno);
-	if (pag) {
-		ASSERT(atomic_read(&pag->pag_ref) >= 0);
-		ref = atomic_inc_return(&pag->pag_ref);
-	}
-	rcu_read_unlock();
-	trace_xfs_perag_get(mp, agno, ref, _RET_IP_);
-	return pag;
-}
-
-/*
- * search from @first to find the next perag with the given tag set.
- */
-struct xfs_perag *
-xfs_perag_get_tag(
-	struct xfs_mount	*mp,
-	xfs_agnumber_t		first,
-	int			tag)
-{
-	struct xfs_perag	*pag;
-	int			found;
-	int			ref;
-
-	rcu_read_lock();
-	found = radix_tree_gang_lookup_tag(&mp->m_perag_tree,
-					(void **)&pag, first, 1, tag);
-	if (found <= 0) {
-		rcu_read_unlock();
-		return NULL;
-	}
-	ref = atomic_inc_return(&pag->pag_ref);
-	rcu_read_unlock();
-	trace_xfs_perag_get_tag(mp, pag->pag_agno, ref, _RET_IP_);
-	return pag;
-}
-
-void
-xfs_perag_put(
-	struct xfs_perag	*pag)
-{
-	int	ref;
-
-	ASSERT(atomic_read(&pag->pag_ref) > 0);
-	ref = atomic_dec_return(&pag->pag_ref);
-	trace_xfs_perag_put(pag->pag_mount, pag->pag_agno, ref, _RET_IP_);
-}
-
 /* Check all the superblock fields we care about when reading one in. */
 STATIC int
 xfs_validate_sb_read(
@@ -842,78 +781,6 @@ xfs_sb_mount_common(
 }
 
 /*
- * xfs_initialize_perag_data
- *
- * Read in each per-ag structure so we can count up the number of
- * allocated inodes, free inodes and used filesystem blocks as this
- * information is no longer persistent in the superblock. Once we have
- * this information, write it into the in-core superblock structure.
- */
-int
-xfs_initialize_perag_data(
-	struct xfs_mount *mp,
-	xfs_agnumber_t	agcount)
-{
-	xfs_agnumber_t	index;
-	xfs_perag_t	*pag;
-	xfs_sb_t	*sbp = &mp->m_sb;
-	uint64_t	ifree = 0;
-	uint64_t	ialloc = 0;
-	uint64_t	bfree = 0;
-	uint64_t	bfreelst = 0;
-	uint64_t	btree = 0;
-	uint64_t	fdblocks;
-	int		error = 0;
-
-	for (index = 0; index < agcount; index++) {
-		/*
-		 * read the agf, then the agi. This gets us
-		 * all the information we need and populates the
-		 * per-ag structures for us.
-		 */
-		error = xfs_alloc_pagf_init(mp, NULL, index, 0);
-		if (error)
-			return error;
-
-		error = xfs_ialloc_pagi_init(mp, NULL, index);
-		if (error)
-			return error;
-		pag = xfs_perag_get(mp, index);
-		ifree += pag->pagi_freecount;
-		ialloc += pag->pagi_count;
-		bfree += pag->pagf_freeblks;
-		bfreelst += pag->pagf_flcount;
-		btree += pag->pagf_btreeblks;
-		xfs_perag_put(pag);
-	}
-	fdblocks = bfree + bfreelst + btree;
-
-	/*
-	 * If the new summary counts are obviously incorrect, fail the
-	 * mount operation because that implies the AGFs are also corrupt.
-	 * Clear FS_COUNTERS so that we don't unmount with a dirty log, which
-	 * will prevent xfs_repair from fixing anything.
-	 */
-	if (fdblocks > sbp->sb_dblocks || ifree > ialloc) {
-		xfs_alert(mp, "AGF corruption. Please run xfs_repair.");
-		error = -EFSCORRUPTED;
-		goto out;
-	}
-
-	/* Overwrite incore superblock counters with just-read data */
-	spin_lock(&mp->m_sb_lock);
-	sbp->sb_ifree = ifree;
-	sbp->sb_icount = ialloc;
-	sbp->sb_fdblocks = fdblocks;
-	spin_unlock(&mp->m_sb_lock);
-
-	xfs_reinit_percpu_counters(mp);
-out:
-	xfs_fs_mark_healthy(mp, XFS_SICK_FS_COUNTERS);
-	return error;
-}
-
-/*
  * xfs_log_sb() can be used to copy arbitrary changes to the in-core superblock
  * into the superblock buffer to be logged.  It does not provide the higher
  * level of locking that is needed to protect the in-core superblock from
@@ -989,17 +856,18 @@ int
 xfs_update_secondary_sbs(
 	struct xfs_mount	*mp)
 {
-	xfs_agnumber_t		agno;
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno = 1;
 	int			saved_error = 0;
 	int			error = 0;
 	LIST_HEAD		(buffer_list);
 
 	/* update secondary superblocks. */
-	for (agno = 1; agno < mp->m_sb.sb_agcount; agno++) {
+	for_each_perag_from(mp, agno, pag) {
 		struct xfs_buf		*bp;
 
 		error = xfs_buf_get(mp->m_ddev_targp,
-				 XFS_AG_DADDR(mp, agno, XFS_SB_DADDR),
+				 XFS_AG_DADDR(mp, pag->pag_agno, XFS_SB_DADDR),
 				 XFS_FSS_TO_BB(mp, 1), &bp);
 		/*
 		 * If we get an error reading or writing alternate superblocks,
@@ -1011,7 +879,7 @@ xfs_update_secondary_sbs(
 		if (error) {
 			xfs_warn(mp,
 		"error allocating secondary superblock for ag %d",
-				agno);
+				pag->pag_agno);
 			if (!saved_error)
 				saved_error = error;
 			continue;
@@ -1032,7 +900,7 @@ xfs_update_secondary_sbs(
 		if (error) {
 			xfs_warn(mp,
 		"write error %d updating a secondary superblock near ag %d",
-				error, agno);
+				error, pag->pag_agno);
 			if (!saved_error)
 				saved_error = error;
 			continue;
diff --git a/fs/xfs/libxfs/xfs_sb.h b/fs/xfs/libxfs/xfs_sb.h
index f79f9dc632b6..0c1602d9b53d 100644
--- a/fs/xfs/libxfs/xfs_sb.h
+++ b/fs/xfs/libxfs/xfs_sb.h
@@ -13,15 +13,6 @@ struct xfs_trans;
 struct xfs_fsop_geom;
 struct xfs_perag;
 
-/*
- * perag get/put wrappers for ref counting
- */
-extern struct xfs_perag *xfs_perag_get(struct xfs_mount *, xfs_agnumber_t);
-extern struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *, xfs_agnumber_t,
-					   int tag);
-extern void	xfs_perag_put(struct xfs_perag *pag);
-extern int	xfs_initialize_perag_data(struct xfs_mount *, xfs_agnumber_t);
-
 extern void	xfs_log_sb(struct xfs_trans *tp);
 extern int	xfs_sync_sb(struct xfs_mount *mp, bool wait);
 extern int	xfs_sync_sb_buf(struct xfs_mount *mp);
diff --git a/fs/xfs/libxfs/xfs_shared.h b/fs/xfs/libxfs/xfs_shared.h
index 782fdd08f759..25c4cab58851 100644
--- a/fs/xfs/libxfs/xfs_shared.h
+++ b/fs/xfs/libxfs/xfs_shared.h
@@ -22,30 +22,26 @@ struct xfs_inode;
  * Buffer verifier operations are widely used, including userspace tools
  */
 extern const struct xfs_buf_ops xfs_agf_buf_ops;
-extern const struct xfs_buf_ops xfs_agi_buf_ops;
-extern const struct xfs_buf_ops xfs_agf_buf_ops;
 extern const struct xfs_buf_ops xfs_agfl_buf_ops;
-extern const struct xfs_buf_ops xfs_bnobt_buf_ops;
-extern const struct xfs_buf_ops xfs_cntbt_buf_ops;
-extern const struct xfs_buf_ops xfs_rmapbt_buf_ops;
-extern const struct xfs_buf_ops xfs_refcountbt_buf_ops;
+extern const struct xfs_buf_ops xfs_agi_buf_ops;
 extern const struct xfs_buf_ops xfs_attr3_leaf_buf_ops;
 extern const struct xfs_buf_ops xfs_attr3_rmt_buf_ops;
 extern const struct xfs_buf_ops xfs_bmbt_buf_ops;
+extern const struct xfs_buf_ops xfs_bnobt_buf_ops;
+extern const struct xfs_buf_ops xfs_cntbt_buf_ops;
 extern const struct xfs_buf_ops xfs_da3_node_buf_ops;
 extern const struct xfs_buf_ops xfs_dquot_buf_ops;
-extern const struct xfs_buf_ops xfs_symlink_buf_ops;
-extern const struct xfs_buf_ops xfs_agi_buf_ops;
-extern const struct xfs_buf_ops xfs_inobt_buf_ops;
+extern const struct xfs_buf_ops xfs_dquot_buf_ra_ops;
 extern const struct xfs_buf_ops xfs_finobt_buf_ops;
+extern const struct xfs_buf_ops xfs_inobt_buf_ops;
 extern const struct xfs_buf_ops xfs_inode_buf_ops;
 extern const struct xfs_buf_ops xfs_inode_buf_ra_ops;
-extern const struct xfs_buf_ops xfs_dquot_buf_ops;
-extern const struct xfs_buf_ops xfs_dquot_buf_ra_ops;
+extern const struct xfs_buf_ops xfs_refcountbt_buf_ops;
+extern const struct xfs_buf_ops xfs_rmapbt_buf_ops;
+extern const struct xfs_buf_ops xfs_rtbuf_ops;
 extern const struct xfs_buf_ops xfs_sb_buf_ops;
 extern const struct xfs_buf_ops xfs_sb_quiet_buf_ops;
 extern const struct xfs_buf_ops xfs_symlink_buf_ops;
-extern const struct xfs_buf_ops xfs_rtbuf_ops;
 
 /* log size calculation functions */
 int	xfs_log_calc_unit_res(struct xfs_mount *mp, int unit_bytes);
diff --git a/fs/xfs/libxfs/xfs_types.c b/fs/xfs/libxfs/xfs_types.c
index 04801362e1a7..e8f4abee7892 100644
--- a/fs/xfs/libxfs/xfs_types.c
+++ b/fs/xfs/libxfs/xfs_types.c
@@ -11,6 +11,7 @@
 #include "xfs_trans_resv.h"
 #include "xfs_bit.h"
 #include "xfs_mount.h"
+#include "xfs_ag.h"
 
 /* Find the size of the AG, in blocks. */
 inline xfs_agblock_t
@@ -222,12 +223,13 @@ xfs_icount_range(
 	unsigned long long	*max)
 {
 	unsigned long long	nr_inos = 0;
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno;
 
 	/* root, rtbitmap, rtsum all live in the first chunk */
 	*min = XFS_INODES_PER_CHUNK;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+	for_each_perag(mp, agno, pag) {
 		xfs_agino_t	first, last;
 
 		xfs_agino_range(mp, agno, &first, &last);
diff --git a/fs/xfs/libxfs/xfs_types.h b/fs/xfs/libxfs/xfs_types.h
index 064bd6e8c922..0870ef6f933d 100644
--- a/fs/xfs/libxfs/xfs_types.h
+++ b/fs/xfs/libxfs/xfs_types.h
@@ -21,6 +21,7 @@ typedef int32_t		xfs_suminfo_t;	/* type of bitmap summary info */
 typedef uint32_t	xfs_rtword_t;	/* word type for bitmap manipulations */
 
 typedef int64_t		xfs_lsn_t;	/* log sequence number */
+typedef int64_t		xfs_csn_t;	/* CIL sequence number */
 
 typedef uint32_t	xfs_dablk_t;	/* dir/attr block number (in file) */
 typedef uint32_t	xfs_dahash_t;	/* dir/attr hash value */
diff --git a/fs/xfs/scrub/agheader.c b/fs/xfs/scrub/agheader.c
index f96e84793cc9..be1a7e1e65f7 100644
--- a/fs/xfs/scrub/agheader.c
+++ b/fs/xfs/scrub/agheader.c
@@ -14,6 +14,7 @@
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_rmap.h"
+#include "xfs_ag.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 
diff --git a/fs/xfs/scrub/agheader_repair.c b/fs/xfs/scrub/agheader_repair.c
index 23690f824ffa..e95f8c98f0f7 100644
--- a/fs/xfs/scrub/agheader_repair.c
+++ b/fs/xfs/scrub/agheader_repair.c
@@ -20,6 +20,7 @@
 #include "xfs_rmap.h"
 #include "xfs_rmap_btree.h"
 #include "xfs_refcount_btree.h"
+#include "xfs_ag.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -245,8 +246,8 @@ xrep_agf_calc_from_btrees(
 	int			error;
 
 	/* Update the AGF counters from the bnobt. */
-	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
-			XFS_BTNUM_BNO);
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp,
+			sc->sa.pag, XFS_BTNUM_BNO);
 	error = xfs_alloc_query_all(cur, xrep_agf_walk_allocbt, &raa);
 	if (error)
 		goto err;
@@ -259,8 +260,8 @@ xrep_agf_calc_from_btrees(
 	agf->agf_longest = cpu_to_be32(raa.longest);
 
 	/* Update the AGF counters from the cntbt. */
-	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
-			XFS_BTNUM_CNT);
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp,
+			sc->sa.pag, XFS_BTNUM_CNT);
 	error = xfs_btree_count_blocks(cur, &blocks);
 	if (error)
 		goto err;
@@ -268,7 +269,7 @@ xrep_agf_calc_from_btrees(
 	btreeblks += blocks - 1;
 
 	/* Update the AGF counters from the rmapbt. */
-	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.pag);
 	error = xfs_btree_count_blocks(cur, &blocks);
 	if (error)
 		goto err;
@@ -281,7 +282,7 @@ xrep_agf_calc_from_btrees(
 	/* Update the AGF counters from the refcountbt. */
 	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
 		cur = xfs_refcountbt_init_cursor(mp, sc->tp, agf_bp,
-				sc->sa.agno);
+				sc->sa.pag);
 		error = xfs_btree_count_blocks(cur, &blocks);
 		if (error)
 			goto err;
@@ -453,7 +454,7 @@ xrep_agfl_walk_rmap(
 
 	/* Record all the OWN_AG blocks. */
 	if (rec->rm_owner == XFS_RMAP_OWN_AG) {
-		fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.agno,
+		fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.pag->pag_agno,
 				rec->rm_startblock);
 		error = xbitmap_set(ra->freesp, fsb, rec->rm_blockcount);
 		if (error)
@@ -489,23 +490,23 @@ xrep_agfl_collect_blocks(
 	xbitmap_init(&ra.agmetablocks);
 
 	/* Find all space used by the free space btrees & rmapbt. */
-	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.pag);
 	error = xfs_rmap_query_all(cur, xrep_agfl_walk_rmap, &ra);
 	if (error)
 		goto err;
 	xfs_btree_del_cursor(cur, error);
 
 	/* Find all blocks currently being used by the bnobt. */
-	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
-			XFS_BTNUM_BNO);
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp,
+			sc->sa.pag, XFS_BTNUM_BNO);
 	error = xbitmap_set_btblocks(&ra.agmetablocks, cur);
 	if (error)
 		goto err;
 	xfs_btree_del_cursor(cur, error);
 
 	/* Find all blocks currently being used by the cntbt. */
-	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
-			XFS_BTNUM_CNT);
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp,
+			sc->sa.pag, XFS_BTNUM_CNT);
 	error = xbitmap_set_btblocks(&ra.agmetablocks, cur);
 	if (error)
 		goto err;
@@ -805,8 +806,8 @@ xrep_agi_calc_from_btrees(
 	xfs_agino_t		freecount;
 	int			error;
 
-	cur = xfs_inobt_init_cursor(mp, sc->tp, agi_bp, sc->sa.agno,
-			XFS_BTNUM_INO);
+	cur = xfs_inobt_init_cursor(mp, sc->tp, agi_bp,
+			sc->sa.pag, XFS_BTNUM_INO);
 	error = xfs_ialloc_count_inodes(cur, &count, &freecount);
 	if (error)
 		goto err;
@@ -827,8 +828,8 @@ xrep_agi_calc_from_btrees(
 	    xfs_sb_version_hasinobtcounts(&mp->m_sb)) {
 		xfs_agblock_t	blocks;
 
-		cur = xfs_inobt_init_cursor(mp, sc->tp, agi_bp, sc->sa.agno,
-				XFS_BTNUM_FINO);
+		cur = xfs_inobt_init_cursor(mp, sc->tp, agi_bp,
+				sc->sa.pag, XFS_BTNUM_FINO);
 		error = xfs_btree_count_blocks(cur, &blocks);
 		if (error)
 			goto err;
diff --git a/fs/xfs/scrub/alloc.c b/fs/xfs/scrub/alloc.c
index 2720bd7fe53b..d5741980094a 100644
--- a/fs/xfs/scrub/alloc.c
+++ b/fs/xfs/scrub/alloc.c
@@ -15,6 +15,7 @@
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/btree.h"
+#include "xfs_ag.h"
 
 /*
  * Set us up to scrub free space btrees.
@@ -93,7 +94,7 @@ xchk_allocbt_rec(
 	union xfs_btree_rec	*rec)
 {
 	struct xfs_mount	*mp = bs->cur->bc_mp;
-	xfs_agnumber_t		agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t		agno = bs->cur->bc_ag.pag->pag_agno;
 	xfs_agblock_t		bno;
 	xfs_extlen_t		len;
 
diff --git a/fs/xfs/scrub/bmap.c b/fs/xfs/scrub/bmap.c
index 77d5c4a0f09f..1d146c9d9de1 100644
--- a/fs/xfs/scrub/bmap.c
+++ b/fs/xfs/scrub/bmap.c
@@ -22,6 +22,7 @@
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/btree.h"
+#include "xfs_ag.h"
 
 /* Set us up with an inode's bmap. */
 int
@@ -514,7 +515,7 @@ xchk_bmap_check_rmap(
 			xchk_fblock_set_corrupt(sc, sbcri->whichfork,
 					rec->rm_offset);
 		if (irec.br_startblock != XFS_AGB_TO_FSB(sc->mp,
-				cur->bc_ag.agno, rec->rm_startblock))
+				cur->bc_ag.pag->pag_agno, rec->rm_startblock))
 			xchk_fblock_set_corrupt(sc, sbcri->whichfork,
 					rec->rm_offset);
 		if (irec.br_blockcount > rec->rm_blockcount)
@@ -544,18 +545,18 @@ STATIC int
 xchk_bmap_check_ag_rmaps(
 	struct xfs_scrub		*sc,
 	int				whichfork,
-	xfs_agnumber_t			agno)
+	struct xfs_perag		*pag)
 {
 	struct xchk_bmap_check_rmap_info	sbcri;
 	struct xfs_btree_cur		*cur;
 	struct xfs_buf			*agf;
 	int				error;
 
-	error = xfs_alloc_read_agf(sc->mp, sc->tp, agno, 0, &agf);
+	error = xfs_alloc_read_agf(sc->mp, sc->tp, pag->pag_agno, 0, &agf);
 	if (error)
 		return error;
 
-	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, agf, agno);
+	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, agf, pag);
 
 	sbcri.sc = sc;
 	sbcri.whichfork = whichfork;
@@ -575,6 +576,7 @@ xchk_bmap_check_rmaps(
 	int			whichfork)
 {
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(sc->ip, whichfork);
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno;
 	bool			zero_size;
 	int			error;
@@ -607,15 +609,16 @@ xchk_bmap_check_rmaps(
 	    (zero_size || ifp->if_nextents > 0))
 		return 0;
 
-	for (agno = 0; agno < sc->mp->m_sb.sb_agcount; agno++) {
-		error = xchk_bmap_check_ag_rmaps(sc, whichfork, agno);
+	for_each_perag(sc->mp, agno, pag) {
+		error = xchk_bmap_check_ag_rmaps(sc, whichfork, pag);
 		if (error)
-			return error;
+			break;
 		if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
 			break;
 	}
-
-	return 0;
+	if (pag)
+		xfs_perag_put(pag);
+	return error;
 }
 
 /*
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index 6cc92291c46f..8558ca05e11d 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -12,7 +12,6 @@
 #include "xfs_btree.h"
 #include "xfs_log_format.h"
 #include "xfs_trans.h"
-#include "xfs_sb.h"
 #include "xfs_inode.h"
 #include "xfs_icache.h"
 #include "xfs_alloc.h"
@@ -26,6 +25,7 @@
 #include "xfs_trans_priv.h"
 #include "xfs_attr.h"
 #include "xfs_reflink.h"
+#include "xfs_ag.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -460,49 +460,48 @@ xchk_ag_btcur_init(
 	struct xchk_ag		*sa)
 {
 	struct xfs_mount	*mp = sc->mp;
-	xfs_agnumber_t		agno = sa->agno;
 
 	xchk_perag_get(sc->mp, sa);
 	if (sa->agf_bp &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_BNO)) {
 		/* Set up a bnobt cursor for cross-referencing. */
 		sa->bno_cur = xfs_allocbt_init_cursor(mp, sc->tp, sa->agf_bp,
-				agno, XFS_BTNUM_BNO);
+				sa->pag, XFS_BTNUM_BNO);
 	}
 
 	if (sa->agf_bp &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_CNT)) {
 		/* Set up a cntbt cursor for cross-referencing. */
 		sa->cnt_cur = xfs_allocbt_init_cursor(mp, sc->tp, sa->agf_bp,
-				agno, XFS_BTNUM_CNT);
+				sa->pag, XFS_BTNUM_CNT);
 	}
 
 	/* Set up a inobt cursor for cross-referencing. */
 	if (sa->agi_bp &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_INO)) {
 		sa->ino_cur = xfs_inobt_init_cursor(mp, sc->tp, sa->agi_bp,
-					agno, XFS_BTNUM_INO);
+				sa->pag, XFS_BTNUM_INO);
 	}
 
 	/* Set up a finobt cursor for cross-referencing. */
 	if (sa->agi_bp && xfs_sb_version_hasfinobt(&mp->m_sb) &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_FINO)) {
 		sa->fino_cur = xfs_inobt_init_cursor(mp, sc->tp, sa->agi_bp,
-				agno, XFS_BTNUM_FINO);
+				sa->pag, XFS_BTNUM_FINO);
 	}
 
 	/* Set up a rmapbt cursor for cross-referencing. */
 	if (sa->agf_bp && xfs_sb_version_hasrmapbt(&mp->m_sb) &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_RMAP)) {
 		sa->rmap_cur = xfs_rmapbt_init_cursor(mp, sc->tp, sa->agf_bp,
-				agno);
+				sa->pag);
 	}
 
 	/* Set up a refcountbt cursor for cross-referencing. */
 	if (sa->agf_bp && xfs_sb_version_hasreflink(&mp->m_sb) &&
 	    xchk_ag_btree_healthy_enough(sc, sa->pag, XFS_BTNUM_REFC)) {
 		sa->refc_cur = xfs_refcountbt_init_cursor(mp, sc->tp,
-				sa->agf_bp, agno);
+				sa->agf_bp, sa->pag);
 	}
 }
 
diff --git a/fs/xfs/scrub/fscounters.c b/fs/xfs/scrub/fscounters.c
index f1d1a8c58853..fd7941e04ae1 100644
--- a/fs/xfs/scrub/fscounters.c
+++ b/fs/xfs/scrub/fscounters.c
@@ -9,11 +9,11 @@
 #include "xfs_format.h"
 #include "xfs_trans_resv.h"
 #include "xfs_mount.h"
-#include "xfs_sb.h"
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_health.h"
 #include "xfs_btree.h"
+#include "xfs_ag.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -71,11 +71,11 @@ xchk_fscount_warmup(
 	xfs_agnumber_t		agno;
 	int			error = 0;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		pag = xfs_perag_get(mp, agno);
-
+	for_each_perag(mp, agno, pag) {
+		if (xchk_should_terminate(sc, &error))
+			break;
 		if (pag->pagi_init && pag->pagf_init)
-			goto next_loop_perag;
+			continue;
 
 		/* Lock both AG headers. */
 		error = xfs_ialloc_read_agi(mp, sc->tp, agno, &agi_bp);
@@ -89,21 +89,15 @@ xchk_fscount_warmup(
 		 * These are supposed to be initialized by the header read
 		 * function.
 		 */
-		error = -EFSCORRUPTED;
-		if (!pag->pagi_init || !pag->pagf_init)
+		if (!pag->pagi_init || !pag->pagf_init) {
+			error = -EFSCORRUPTED;
 			break;
+		}
 
 		xfs_buf_relse(agf_bp);
 		agf_bp = NULL;
 		xfs_buf_relse(agi_bp);
 		agi_bp = NULL;
-next_loop_perag:
-		xfs_perag_put(pag);
-		pag = NULL;
-		error = 0;
-
-		if (xchk_should_terminate(sc, &error))
-			break;
 	}
 
 	if (agf_bp)
@@ -196,13 +190,14 @@ retry:
 	fsc->ifree = 0;
 	fsc->fdblocks = 0;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		pag = xfs_perag_get(mp, agno);
+	for_each_perag(mp, agno, pag) {
+		if (xchk_should_terminate(sc, &error))
+			break;
 
 		/* This somehow got unset since the warmup? */
 		if (!pag->pagi_init || !pag->pagf_init) {
-			xfs_perag_put(pag);
-			return -EFSCORRUPTED;
+			error = -EFSCORRUPTED;
+			break;
 		}
 
 		/* Count all the inodes */
@@ -216,10 +211,8 @@ retry:
 			fsc->fdblocks += pag->pagf_btreeblks;
 		} else {
 			error = xchk_fscount_btreeblks(sc, fsc, agno);
-			if (error) {
-				xfs_perag_put(pag);
+			if (error)
 				break;
-			}
 		}
 
 		/*
@@ -229,12 +222,9 @@ retry:
 		fsc->fdblocks -= pag->pag_meta_resv.ar_reserved;
 		fsc->fdblocks -= pag->pag_rmapbt_resv.ar_orig_reserved;
 
-		xfs_perag_put(pag);
-
-		if (xchk_should_terminate(sc, &error))
-			break;
 	}
-
+	if (pag)
+		xfs_perag_put(pag);
 	if (error)
 		return error;
 
diff --git a/fs/xfs/scrub/health.c b/fs/xfs/scrub/health.c
index 3de59b5c2ce6..2e61df3bca83 100644
--- a/fs/xfs/scrub/health.c
+++ b/fs/xfs/scrub/health.c
@@ -8,7 +8,7 @@
 #include "xfs_shared.h"
 #include "xfs_format.h"
 #include "xfs_btree.h"
-#include "xfs_sb.h"
+#include "xfs_ag.h"
 #include "xfs_health.h"
 #include "scrub/scrub.h"
 #include "scrub/health.h"
diff --git a/fs/xfs/scrub/ialloc.c b/fs/xfs/scrub/ialloc.c
index 8d9f3fb0cd22..30e568596b79 100644
--- a/fs/xfs/scrub/ialloc.c
+++ b/fs/xfs/scrub/ialloc.c
@@ -21,6 +21,7 @@
 #include "scrub/common.h"
 #include "scrub/btree.h"
 #include "scrub/trace.h"
+#include "xfs_ag.h"
 
 /*
  * Set us up to scrub inode btrees.
@@ -103,7 +104,7 @@ xchk_iallocbt_chunk(
 	xfs_extlen_t			len)
 {
 	struct xfs_mount		*mp = bs->cur->bc_mp;
-	xfs_agnumber_t			agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t			agno = bs->cur->bc_ag.pag->pag_agno;
 	xfs_agblock_t			bno;
 
 	bno = XFS_AGINO_TO_AGBNO(mp, agino);
@@ -163,7 +164,7 @@ xchk_iallocbt_check_cluster_ifree(
 	 * the record, compute which fs inode we're talking about.
 	 */
 	agino = irec->ir_startino + irec_ino;
-	fsino = XFS_AGINO_TO_INO(mp, bs->cur->bc_ag.agno, agino);
+	fsino = XFS_AGINO_TO_INO(mp, bs->cur->bc_ag.pag->pag_agno, agino);
 	irec_free = (irec->ir_free & XFS_INOBT_MASK(irec_ino));
 
 	if (be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC ||
@@ -213,7 +214,7 @@ xchk_iallocbt_check_cluster(
 	struct xfs_mount		*mp = bs->cur->bc_mp;
 	struct xfs_buf			*cluster_bp;
 	unsigned int			nr_inodes;
-	xfs_agnumber_t			agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t			agno = bs->cur->bc_ag.pag->pag_agno;
 	xfs_agblock_t			agbno;
 	unsigned int			cluster_index;
 	uint16_t			cluster_mask = 0;
@@ -423,7 +424,7 @@ xchk_iallocbt_rec(
 	struct xchk_iallocbt		*iabt = bs->private;
 	struct xfs_inobt_rec_incore	irec;
 	uint64_t			holes;
-	xfs_agnumber_t			agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t			agno = bs->cur->bc_ag.pag->pag_agno;
 	xfs_agino_t			agino;
 	xfs_extlen_t			len;
 	int				holecount;
diff --git a/fs/xfs/scrub/refcount.c b/fs/xfs/scrub/refcount.c
index 744530a66c0c..7014b7408bad 100644
--- a/fs/xfs/scrub/refcount.c
+++ b/fs/xfs/scrub/refcount.c
@@ -13,6 +13,7 @@
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/btree.h"
+#include "xfs_ag.h"
 
 /*
  * Set us up to scrub reference count btrees.
@@ -333,7 +334,7 @@ xchk_refcountbt_rec(
 {
 	struct xfs_mount	*mp = bs->cur->bc_mp;
 	xfs_agblock_t		*cow_blocks = bs->private;
-	xfs_agnumber_t		agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t		agno = bs->cur->bc_ag.pag->pag_agno;
 	xfs_agblock_t		bno;
 	xfs_extlen_t		len;
 	xfs_nlink_t		refcount;
diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
index b8202dd08939..ebb0e245aa72 100644
--- a/fs/xfs/scrub/repair.c
+++ b/fs/xfs/scrub/repair.c
@@ -22,6 +22,7 @@
 #include "xfs_rmap_btree.h"
 #include "xfs_refcount_btree.h"
 #include "xfs_extent_busy.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 #include "xfs_quota.h"
 #include "scrub/scrub.h"
@@ -303,7 +304,7 @@ xrep_alloc_ag_block(
 			return error;
 		if (bno == NULLAGBLOCK)
 			return -ENOSPC;
-		xfs_extent_busy_reuse(sc->mp, sc->sa.agno, bno,
+		xfs_extent_busy_reuse(sc->mp, sc->sa.pag, bno,
 				1, false);
 		*fsbno = XFS_AGB_TO_FSB(sc->mp, sc->sa.agno, bno);
 		if (resv == XFS_AG_RESV_RMAPBT)
@@ -508,7 +509,7 @@ xrep_put_freelist(
 	 * create an rmap for the block prior to merging it or else other
 	 * parts will break.
 	 */
-	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
+	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.pag, agbno, 1,
 			&XFS_RMAP_OINFO_AG);
 	if (error)
 		return error;
@@ -518,7 +519,7 @@ xrep_put_freelist(
 			agbno, 0);
 	if (error)
 		return error;
-	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
+	xfs_extent_busy_insert(sc->tp, sc->sa.pag, agbno, 1,
 			XFS_EXTENT_BUSY_SKIP_DISCARD);
 
 	return 0;
@@ -554,7 +555,7 @@ xrep_reap_block(
 	} else {
 		agf_bp = sc->sa.agf_bp;
 	}
-	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, agf_bp, agno);
+	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, agf_bp, sc->sa.pag);
 
 	/* Can we find any other rmappings? */
 	error = xfs_rmap_has_other_keys(cur, agbno, 1, oinfo, &has_other_rmap);
@@ -576,7 +577,8 @@ xrep_reap_block(
 	 * to run xfs_repair.
 	 */
 	if (has_other_rmap)
-		error = xfs_rmap_free(sc->tp, agf_bp, agno, agbno, 1, oinfo);
+		error = xfs_rmap_free(sc->tp, agf_bp, sc->sa.pag, agbno,
+					1, oinfo);
 	else if (resv == XFS_AG_RESV_AGFL)
 		error = xrep_put_freelist(sc, agbno);
 	else
@@ -891,7 +893,7 @@ xrep_find_ag_btree_roots(
 		fab->height = 0;
 	}
 
-	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.pag);
 	error = xfs_rmap_query_all(cur, xrep_findroot_rmap, &ri);
 	xfs_btree_del_cursor(cur, error);
 
diff --git a/fs/xfs/scrub/rmap.c b/fs/xfs/scrub/rmap.c
index a4f17477c5d1..fc306573f0ac 100644
--- a/fs/xfs/scrub/rmap.c
+++ b/fs/xfs/scrub/rmap.c
@@ -15,6 +15,7 @@
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/btree.h"
+#include "xfs_ag.h"
 
 /*
  * Set us up to scrub reverse mapping btrees.
@@ -91,7 +92,7 @@ xchk_rmapbt_rec(
 {
 	struct xfs_mount	*mp = bs->cur->bc_mp;
 	struct xfs_rmap_irec	irec;
-	xfs_agnumber_t		agno = bs->cur->bc_ag.agno;
+	xfs_agnumber_t		agno = bs->cur->bc_ag.pag->pag_agno;
 	bool			non_inode;
 	bool			is_unwritten;
 	bool			is_bmbt;
diff --git a/fs/xfs/scrub/trace.c b/fs/xfs/scrub/trace.c
index 2c6c248be823..03882a605a3c 100644
--- a/fs/xfs/scrub/trace.c
+++ b/fs/xfs/scrub/trace.c
@@ -13,6 +13,7 @@
 #include "xfs_inode.h"
 #include "xfs_btree.h"
 #include "scrub/scrub.h"
+#include "xfs_ag.h"
 
 /* Figure out which block the btree cursor was pointing to. */
 static inline xfs_fsblock_t
@@ -26,7 +27,7 @@ xchk_btree_cur_fsbno(
 		 cur->bc_flags & XFS_BTREE_LONG_PTRS)
 		return XFS_INO_TO_FSB(cur->bc_mp, cur->bc_ino.ip->i_ino);
 	else if (!(cur->bc_flags & XFS_BTREE_LONG_PTRS))
-		return XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.agno, 0);
+		return XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_ag.pag->pag_agno, 0);
 	return NULLFSBLOCK;
 }
 
diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
index bfad669e6b2f..aaa7e66c42d7 100644
--- a/fs/xfs/xfs_attr_inactive.c
+++ b/fs/xfs/xfs_attr_inactive.c
@@ -15,10 +15,10 @@
 #include "xfs_da_format.h"
 #include "xfs_da_btree.h"
 #include "xfs_inode.h"
+#include "xfs_attr.h"
 #include "xfs_attr_remote.h"
 #include "xfs_trans.h"
 #include "xfs_bmap.h"
-#include "xfs_attr.h"
 #include "xfs_attr_leaf.h"
 #include "xfs_quota.h"
 #include "xfs_dir2.h"
diff --git a/fs/xfs/xfs_bio_io.c b/fs/xfs/xfs_bio_io.c
index 17f36db2f792..667e297f59b1 100644
--- a/fs/xfs/xfs_bio_io.c
+++ b/fs/xfs/xfs_bio_io.c
@@ -9,6 +9,41 @@ static inline unsigned int bio_max_vecs(unsigned int count)
 	return bio_max_segs(howmany(count, PAGE_SIZE));
 }
 
+static void
+xfs_flush_bdev_async_endio(
+	struct bio	*bio)
+{
+	complete(bio->bi_private);
+}
+
+/*
+ * Submit a request for an async cache flush to run. If the request queue does
+ * not require flush operations, just skip it altogether. If the caller needs
+ * to wait for the flush completion at a later point in time, they must supply a
+ * valid completion. This will be signalled when the flush completes.  The
+ * caller never sees the bio that is issued here.
+ */
+void
+xfs_flush_bdev_async(
+	struct bio		*bio,
+	struct block_device	*bdev,
+	struct completion	*done)
+{
+	struct request_queue	*q = bdev->bd_disk->queue;
+
+	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags)) {
+		complete(done);
+		return;
+	}
+
+	bio_init(bio, NULL, 0);
+	bio_set_dev(bio, bdev);
+	bio->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC;
+	bio->bi_private = done;
+	bio->bi_end_io = xfs_flush_bdev_async_endio;
+
+	submit_bio(bio);
+}
 int
 xfs_rw_bdev(
 	struct block_device	*bdev,
diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
index 2988efa97e14..213a97a921bb 100644
--- a/fs/xfs/xfs_bmap_util.c
+++ b/fs/xfs/xfs_bmap_util.c
@@ -945,7 +945,7 @@ xfs_flush_unmap_range(
 	xfs_off_t		rounding, start, end;
 	int			error;
 
-	rounding = max_t(xfs_off_t, 1 << mp->m_sb.sb_blocklog, PAGE_SIZE);
+	rounding = max_t(xfs_off_t, mp->m_sb.sb_blocksize, PAGE_SIZE);
 	start = round_down(offset, rounding);
 	end = round_up(offset + len, rounding) - 1;
 
@@ -1053,9 +1053,9 @@ xfs_prepare_shift(
 	 * extent (after split) during the shift and corrupt the file. Start
 	 * with the block just prior to the start to stabilize the boundary.
 	 */
-	offset = round_down(offset, 1 << mp->m_sb.sb_blocklog);
+	offset = round_down(offset, mp->m_sb.sb_blocksize);
 	if (offset)
-		offset -= (1 << mp->m_sb.sb_blocklog);
+		offset -= mp->m_sb.sb_blocksize;
 
 	/*
 	 * Writeback and invalidate cache for the remainder of the file as we're
diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
index 592800c8852f..8ff42b3585e0 100644
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -10,7 +10,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_trace.h"
 #include "xfs_log.h"
@@ -19,12 +18,10 @@
 #include "xfs_buf_item.h"
 #include "xfs_errortag.h"
 #include "xfs_error.h"
+#include "xfs_ag.h"
 
 static kmem_zone_t *xfs_buf_zone;
 
-#define xb_to_gfp(flags) \
-	((((flags) & XBF_READ_AHEAD) ? __GFP_NORETRY : GFP_NOFS) | __GFP_NOWARN)
-
 /*
  * Locking orders
  *
@@ -79,7 +76,7 @@ static inline int
 xfs_buf_vmap_len(
 	struct xfs_buf	*bp)
 {
-	return (bp->b_page_count * PAGE_SIZE) - bp->b_offset;
+	return (bp->b_page_count * PAGE_SIZE);
 }
 
 /*
@@ -272,51 +269,30 @@ _xfs_buf_alloc(
 	return 0;
 }
 
-/*
- *	Allocate a page array capable of holding a specified number
- *	of pages, and point the page buf at it.
- */
-STATIC int
-_xfs_buf_get_pages(
-	struct xfs_buf		*bp,
-	int			page_count)
+static void
+xfs_buf_free_pages(
+	struct xfs_buf	*bp)
 {
-	/* Make sure that we have a page list */
-	if (bp->b_pages == NULL) {
-		bp->b_page_count = page_count;
-		if (page_count <= XB_PAGES) {
-			bp->b_pages = bp->b_page_array;
-		} else {
-			bp->b_pages = kmem_alloc(sizeof(struct page *) *
-						 page_count, KM_NOFS);
-			if (bp->b_pages == NULL)
-				return -ENOMEM;
-		}
-		memset(bp->b_pages, 0, sizeof(struct page *) * page_count);
+	uint		i;
+
+	ASSERT(bp->b_flags & _XBF_PAGES);
+
+	if (xfs_buf_is_vmapped(bp))
+		vm_unmap_ram(bp->b_addr, bp->b_page_count);
+
+	for (i = 0; i < bp->b_page_count; i++) {
+		if (bp->b_pages[i])
+			__free_page(bp->b_pages[i]);
 	}
-	return 0;
-}
+	if (current->reclaim_state)
+		current->reclaim_state->reclaimed_slab += bp->b_page_count;
 
-/*
- *	Frees b_pages if it was allocated.
- */
-STATIC void
-_xfs_buf_free_pages(
-	struct xfs_buf	*bp)
-{
-	if (bp->b_pages != bp->b_page_array) {
+	if (bp->b_pages != bp->b_page_array)
 		kmem_free(bp->b_pages);
-		bp->b_pages = NULL;
-	}
+	bp->b_pages = NULL;
+	bp->b_flags &= ~_XBF_PAGES;
 }
 
-/*
- *	Releases the specified buffer.
- *
- * 	The modification state of any associated pages is left unchanged.
- * 	The buffer must not be on any hash - use xfs_buf_rele instead for
- * 	hashed and refcounted buffers
- */
 static void
 xfs_buf_free(
 	struct xfs_buf		*bp)
@@ -325,137 +301,103 @@ xfs_buf_free(
 
 	ASSERT(list_empty(&bp->b_lru));
 
-	if (bp->b_flags & _XBF_PAGES) {
-		uint		i;
-
-		if (xfs_buf_is_vmapped(bp))
-			vm_unmap_ram(bp->b_addr - bp->b_offset,
-					bp->b_page_count);
-
-		for (i = 0; i < bp->b_page_count; i++) {
-			struct page	*page = bp->b_pages[i];
-
-			__free_page(page);
-		}
-		if (current->reclaim_state)
-			current->reclaim_state->reclaimed_slab +=
-							bp->b_page_count;
-	} else if (bp->b_flags & _XBF_KMEM)
+	if (bp->b_flags & _XBF_PAGES)
+		xfs_buf_free_pages(bp);
+	else if (bp->b_flags & _XBF_KMEM)
 		kmem_free(bp->b_addr);
-	_xfs_buf_free_pages(bp);
+
 	xfs_buf_free_maps(bp);
 	kmem_cache_free(xfs_buf_zone, bp);
 }
 
-/*
- * Allocates all the pages for buffer in question and builds it's page list.
- */
-STATIC int
-xfs_buf_allocate_memory(
-	struct xfs_buf		*bp,
-	uint			flags)
+static int
+xfs_buf_alloc_kmem(
+	struct xfs_buf	*bp,
+	xfs_buf_flags_t	flags)
 {
-	size_t			size;
-	size_t			nbytes, offset;
-	gfp_t			gfp_mask = xb_to_gfp(flags);
-	unsigned short		page_count, i;
-	xfs_off_t		start, end;
-	int			error;
-	xfs_km_flags_t		kmflag_mask = 0;
+	int		align_mask = xfs_buftarg_dma_alignment(bp->b_target);
+	xfs_km_flags_t	kmflag_mask = KM_NOFS;
+	size_t		size = BBTOB(bp->b_length);
 
-	/*
-	 * assure zeroed buffer for non-read cases.
-	 */
-	if (!(flags & XBF_READ)) {
+	/* Assure zeroed buffer for non-read cases. */
+	if (!(flags & XBF_READ))
 		kmflag_mask |= KM_ZERO;
-		gfp_mask |= __GFP_ZERO;
-	}
 
-	/*
-	 * for buffers that are contained within a single page, just allocate
-	 * the memory from the heap - there's no need for the complexity of
-	 * page arrays to keep allocation down to order 0.
-	 */
-	size = BBTOB(bp->b_length);
-	if (size < PAGE_SIZE) {
-		int align_mask = xfs_buftarg_dma_alignment(bp->b_target);
-		bp->b_addr = kmem_alloc_io(size, align_mask,
-					   KM_NOFS | kmflag_mask);
-		if (!bp->b_addr) {
-			/* low memory - use alloc_page loop instead */
-			goto use_alloc_page;
-		}
+	bp->b_addr = kmem_alloc_io(size, align_mask, kmflag_mask);
+	if (!bp->b_addr)
+		return -ENOMEM;
 
-		if (((unsigned long)(bp->b_addr + size - 1) & PAGE_MASK) !=
-		    ((unsigned long)bp->b_addr & PAGE_MASK)) {
-			/* b_addr spans two pages - use alloc_page instead */
-			kmem_free(bp->b_addr);
-			bp->b_addr = NULL;
-			goto use_alloc_page;
-		}
-		bp->b_offset = offset_in_page(bp->b_addr);
-		bp->b_pages = bp->b_page_array;
-		bp->b_pages[0] = kmem_to_page(bp->b_addr);
-		bp->b_page_count = 1;
-		bp->b_flags |= _XBF_KMEM;
-		return 0;
+	if (((unsigned long)(bp->b_addr + size - 1) & PAGE_MASK) !=
+	    ((unsigned long)bp->b_addr & PAGE_MASK)) {
+		/* b_addr spans two pages - use alloc_page instead */
+		kmem_free(bp->b_addr);
+		bp->b_addr = NULL;
+		return -ENOMEM;
 	}
+	bp->b_offset = offset_in_page(bp->b_addr);
+	bp->b_pages = bp->b_page_array;
+	bp->b_pages[0] = kmem_to_page(bp->b_addr);
+	bp->b_page_count = 1;
+	bp->b_flags |= _XBF_KMEM;
+	return 0;
+}
 
-use_alloc_page:
-	start = BBTOB(bp->b_maps[0].bm_bn) >> PAGE_SHIFT;
-	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)
-								>> PAGE_SHIFT;
-	page_count = end - start;
-	error = _xfs_buf_get_pages(bp, page_count);
-	if (unlikely(error))
-		return error;
+static int
+xfs_buf_alloc_pages(
+	struct xfs_buf	*bp,
+	xfs_buf_flags_t	flags)
+{
+	gfp_t		gfp_mask = __GFP_NOWARN;
+	long		filled = 0;
 
-	offset = bp->b_offset;
+	if (flags & XBF_READ_AHEAD)
+		gfp_mask |= __GFP_NORETRY;
+	else
+		gfp_mask |= GFP_NOFS;
+
+	/* Make sure that we have a page list */
+	bp->b_page_count = DIV_ROUND_UP(BBTOB(bp->b_length), PAGE_SIZE);
+	if (bp->b_page_count <= XB_PAGES) {
+		bp->b_pages = bp->b_page_array;
+	} else {
+		bp->b_pages = kzalloc(sizeof(struct page *) * bp->b_page_count,
+					gfp_mask);
+		if (!bp->b_pages)
+			return -ENOMEM;
+	}
 	bp->b_flags |= _XBF_PAGES;
 
-	for (i = 0; i < bp->b_page_count; i++) {
-		struct page	*page;
-		uint		retries = 0;
-retry:
-		page = alloc_page(gfp_mask);
-		if (unlikely(page == NULL)) {
-			if (flags & XBF_READ_AHEAD) {
-				bp->b_page_count = i;
-				error = -ENOMEM;
-				goto out_free_pages;
-			}
+	/* Assure zeroed buffer for non-read cases. */
+	if (!(flags & XBF_READ))
+		gfp_mask |= __GFP_ZERO;
 
-			/*
-			 * This could deadlock.
-			 *
-			 * But until all the XFS lowlevel code is revamped to
-			 * handle buffer allocation failures we can't do much.
-			 */
-			if (!(++retries % 100))
-				xfs_err(NULL,
-		"%s(%u) possible memory allocation deadlock in %s (mode:0x%x)",
-					current->comm, current->pid,
-					__func__, gfp_mask);
-
-			XFS_STATS_INC(bp->b_mount, xb_page_retries);
-			congestion_wait(BLK_RW_ASYNC, HZ/50);
-			goto retry;
+	/*
+	 * Bulk filling of pages can take multiple calls. Not filling the entire
+	 * array is not an allocation failure, so don't back off if we get at
+	 * least one extra page.
+	 */
+	for (;;) {
+		long	last = filled;
+
+		filled = alloc_pages_bulk_array(gfp_mask, bp->b_page_count,
+						bp->b_pages);
+		if (filled == bp->b_page_count) {
+			XFS_STATS_INC(bp->b_mount, xb_page_found);
+			break;
 		}
 
-		XFS_STATS_INC(bp->b_mount, xb_page_found);
+		if (filled != last)
+			continue;
 
-		nbytes = min_t(size_t, size, PAGE_SIZE - offset);
-		size -= nbytes;
-		bp->b_pages[i] = page;
-		offset = 0;
+		if (flags & XBF_READ_AHEAD) {
+			xfs_buf_free_pages(bp);
+			return -ENOMEM;
+		}
+
+		XFS_STATS_INC(bp->b_mount, xb_page_retries);
+		congestion_wait(BLK_RW_ASYNC, HZ / 50);
 	}
 	return 0;
-
-out_free_pages:
-	for (i = 0; i < bp->b_page_count; i++)
-		__free_page(bp->b_pages[i]);
-	bp->b_flags &= ~_XBF_PAGES;
-	return error;
 }
 
 /*
@@ -469,7 +411,7 @@ _xfs_buf_map_pages(
 	ASSERT(bp->b_flags & _XBF_PAGES);
 	if (bp->b_page_count == 1) {
 		/* A single page buffer is always mappable */
-		bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset;
+		bp->b_addr = page_address(bp->b_pages[0]);
 	} else if (flags & XBF_UNMAPPED) {
 		bp->b_addr = NULL;
 	} else {
@@ -496,7 +438,6 @@ _xfs_buf_map_pages(
 
 		if (!bp->b_addr)
 			return -ENOMEM;
-		bp->b_addr += bp->b_offset;
 	}
 
 	return 0;
@@ -707,7 +648,7 @@ xfs_buf_get_map(
 {
 	struct xfs_buf		*bp;
 	struct xfs_buf		*new_bp;
-	int			error = 0;
+	int			error;
 
 	*bpp = NULL;
 	error = xfs_buf_find(target, map, nmaps, flags, NULL, &bp);
@@ -720,17 +661,22 @@ xfs_buf_get_map(
 	if (error)
 		return error;
 
-	error = xfs_buf_allocate_memory(new_bp, flags);
-	if (error) {
-		xfs_buf_free(new_bp);
-		return error;
+	/*
+	 * For buffers that fit entirely within a single page, first attempt to
+	 * allocate the memory from the heap to minimise memory usage. If we
+	 * can't get heap memory for these small buffers, we fall back to using
+	 * the page allocator.
+	 */
+	if (BBTOB(new_bp->b_length) >= PAGE_SIZE ||
+	    xfs_buf_alloc_kmem(new_bp, flags) < 0) {
+		error = xfs_buf_alloc_pages(new_bp, flags);
+		if (error)
+			goto out_free_buf;
 	}
 
 	error = xfs_buf_find(target, map, nmaps, flags, new_bp, &bp);
-	if (error) {
-		xfs_buf_free(new_bp);
-		return error;
-	}
+	if (error)
+		goto out_free_buf;
 
 	if (bp != new_bp)
 		xfs_buf_free(new_bp);
@@ -758,6 +704,9 @@ found:
 	trace_xfs_buf_get(bp, flags, _RET_IP_);
 	*bpp = bp;
 	return 0;
+out_free_buf:
+	xfs_buf_free(new_bp);
+	return error;
 }
 
 int
@@ -950,8 +899,7 @@ xfs_buf_get_uncached(
 	int			flags,
 	struct xfs_buf		**bpp)
 {
-	unsigned long		page_count;
-	int			error, i;
+	int			error;
 	struct xfs_buf		*bp;
 	DEFINE_SINGLE_BUF_MAP(map, XFS_BUF_DADDR_NULL, numblks);
 
@@ -960,41 +908,25 @@ xfs_buf_get_uncached(
 	/* flags might contain irrelevant bits, pass only what we care about */
 	error = _xfs_buf_alloc(target, &map, 1, flags & XBF_NO_IOACCT, &bp);
 	if (error)
-		goto fail;
+		return error;
 
-	page_count = PAGE_ALIGN(numblks << BBSHIFT) >> PAGE_SHIFT;
-	error = _xfs_buf_get_pages(bp, page_count);
+	error = xfs_buf_alloc_pages(bp, flags);
 	if (error)
 		goto fail_free_buf;
 
-	for (i = 0; i < page_count; i++) {
-		bp->b_pages[i] = alloc_page(xb_to_gfp(flags));
-		if (!bp->b_pages[i]) {
-			error = -ENOMEM;
-			goto fail_free_mem;
-		}
-	}
-	bp->b_flags |= _XBF_PAGES;
-
 	error = _xfs_buf_map_pages(bp, 0);
 	if (unlikely(error)) {
 		xfs_warn(target->bt_mount,
 			"%s: failed to map pages", __func__);
-		goto fail_free_mem;
+		goto fail_free_buf;
 	}
 
 	trace_xfs_buf_get_uncached(bp, _RET_IP_);
 	*bpp = bp;
 	return 0;
 
- fail_free_mem:
-	while (--i >= 0)
-		__free_page(bp->b_pages[i]);
-	_xfs_buf_free_pages(bp);
- fail_free_buf:
-	xfs_buf_free_maps(bp);
-	kmem_cache_free(xfs_buf_zone, bp);
- fail:
+fail_free_buf:
+	xfs_buf_free(bp);
 	return error;
 }
 
@@ -1722,7 +1654,6 @@ xfs_buf_offset(
 	if (bp->b_addr)
 		return bp->b_addr + offset;
 
-	offset += bp->b_offset;
 	page = bp->b_pages[offset >> PAGE_SHIFT];
 	return page_address(page) + (offset & (PAGE_SIZE-1));
 }
@@ -1958,7 +1889,7 @@ xfs_free_buftarg(
 	percpu_counter_destroy(&btp->bt_io_count);
 	list_lru_destroy(&btp->bt_lru);
 
-	xfs_blkdev_issue_flush(btp);
+	blkdev_issue_flush(btp->bt_bdev);
 
 	kmem_free(btp);
 }
diff --git a/fs/xfs/xfs_buf.h b/fs/xfs/xfs_buf.h
index 459ca34f26f5..464dc548fa23 100644
--- a/fs/xfs/xfs_buf.h
+++ b/fs/xfs/xfs_buf.h
@@ -167,7 +167,8 @@ struct xfs_buf {
 	atomic_t		b_pin_count;	/* pin count */
 	atomic_t		b_io_remaining;	/* #outstanding I/O requests */
 	unsigned int		b_page_count;	/* size of page array */
-	unsigned int		b_offset;	/* page offset in first page */
+	unsigned int		b_offset;	/* page offset of b_addr,
+						   only for _XBF_KMEM buffers */
 	int			b_error;	/* error code on I/O */
 
 	/*
diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
index fb69879e4b2b..2828ce45b701 100644
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -74,14 +74,12 @@ xfs_buf_item_straddle(
 }
 
 /*
- * This returns the number of log iovecs needed to log the
- * given buf log item.
+ * Return the number of log iovecs and space needed to log the given buf log
+ * item segment.
  *
- * It calculates this as 1 iovec for the buf log format structure
- * and 1 for each stretch of non-contiguous chunks to be logged.
- * Contiguous chunks are logged in a single iovec.
- *
- * If the XFS_BLI_STALE flag has been set, then log nothing.
+ * It calculates this as 1 iovec for the buf log format structure and 1 for each
+ * stretch of non-contiguous chunks to be logged.  Contiguous chunks are logged
+ * in a single iovec.
  */
 STATIC void
 xfs_buf_item_size_segment(
@@ -168,11 +166,8 @@ slow_scan:
 }
 
 /*
- * This returns the number of log iovecs needed to log the given buf log item.
- *
- * It calculates this as 1 iovec for the buf log format structure and 1 for each
- * stretch of non-contiguous chunks to be logged.  Contiguous chunks are logged
- * in a single iovec.
+ * Return the number of log iovecs and space needed to log the given buf log
+ * item.
  *
  * Discontiguous buffers need a format structure per region that is being
  * logged. This makes the changes in the buffer appear to log recovery as though
@@ -182,7 +177,11 @@ slow_scan:
  * what ends up on disk.
  *
  * If the XFS_BLI_STALE flag has been set, then log nothing but the buf log
- * format structures.
+ * format structures. If the item has previously been logged and has dirty
+ * regions, we do not relog them in stale buffers. This has the effect of
+ * reducing the size of the relogged item by the amount of dirty data tracked
+ * by the log item. This can result in the committing transaction reducing the
+ * amount of space being consumed by the CIL.
  */
 STATIC void
 xfs_buf_item_size(
@@ -199,9 +198,9 @@ xfs_buf_item_size(
 	ASSERT(atomic_read(&bip->bli_refcount) > 0);
 	if (bip->bli_flags & XFS_BLI_STALE) {
 		/*
-		 * The buffer is stale, so all we need to log
-		 * is the buf log format structure with the
-		 * cancel flag in it.
+		 * The buffer is stale, so all we need to log is the buf log
+		 * format structure with the cancel flag in it as we are never
+		 * going to replay the changes tracked in the log item.
 		 */
 		trace_xfs_buf_item_size_stale(bip);
 		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
@@ -216,9 +215,9 @@ xfs_buf_item_size(
 
 	if (bip->bli_flags & XFS_BLI_ORDERED) {
 		/*
-		 * The buffer has been logged just to order it.
-		 * It is not being included in the transaction
-		 * commit, so no vectors are used at all.
+		 * The buffer has been logged just to order it. It is not being
+		 * included in the transaction commit, so no vectors are used at
+		 * all.
 		 */
 		trace_xfs_buf_item_size_ordered(bip);
 		*nvecs = XFS_LOG_VEC_ORDERED;
@@ -475,17 +474,8 @@ xfs_buf_item_pin(
 }
 
 /*
- * This is called to unpin the buffer associated with the buf log
- * item which was previously pinned with a call to xfs_buf_item_pin().
- *
- * Also drop the reference to the buf item for the current transaction.
- * If the XFS_BLI_STALE flag is set and we are the last reference,
- * then free up the buf log item and unlock the buffer.
- *
- * If the remove flag is set we are called from uncommit in the
- * forced-shutdown path.  If that is true and the reference count on
- * the log item is going to drop to zero we need to free the item's
- * descriptor in the transaction.
+ * This is called to unpin the buffer associated with the buf log item which
+ * was previously pinned with a call to xfs_buf_item_pin().
  */
 STATIC void
 xfs_buf_item_unpin(
@@ -502,38 +492,35 @@ xfs_buf_item_unpin(
 
 	trace_xfs_buf_item_unpin(bip);
 
+	/*
+	 * Drop the bli ref associated with the pin and grab the hold required
+	 * for the I/O simulation failure in the abort case. We have to do this
+	 * before the pin count drops because the AIL doesn't acquire a bli
+	 * reference. Therefore if the refcount drops to zero, the bli could
+	 * still be AIL resident and the buffer submitted for I/O (and freed on
+	 * completion) at any point before we return. This can be removed once
+	 * the AIL properly holds a reference on the bli.
+	 */
 	freed = atomic_dec_and_test(&bip->bli_refcount);
-
+	if (freed && !stale && remove)
+		xfs_buf_hold(bp);
 	if (atomic_dec_and_test(&bp->b_pin_count))
 		wake_up_all(&bp->b_waiters);
 
-	if (freed && stale) {
+	 /* nothing to do but drop the pin count if the bli is active */
+	if (!freed)
+		return;
+
+	if (stale) {
 		ASSERT(bip->bli_flags & XFS_BLI_STALE);
 		ASSERT(xfs_buf_islocked(bp));
 		ASSERT(bp->b_flags & XBF_STALE);
 		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(list_empty(&lip->li_trans));
+		ASSERT(!bp->b_transp);
 
 		trace_xfs_buf_item_unpin_stale(bip);
 
-		if (remove) {
-			/*
-			 * If we are in a transaction context, we have to
-			 * remove the log item from the transaction as we are
-			 * about to release our reference to the buffer.  If we
-			 * don't, the unlock that occurs later in
-			 * xfs_trans_uncommit() will try to reference the
-			 * buffer which we no longer have a hold on.
-			 */
-			if (!list_empty(&lip->li_trans))
-				xfs_trans_del_item(lip);
-
-			/*
-			 * Since the transaction no longer refers to the buffer,
-			 * the buffer should no longer refer to the transaction.
-			 */
-			bp->b_transp = NULL;
-		}
-
 		/*
 		 * If we get called here because of an IO error, we may or may
 		 * not have the item on the AIL. xfs_trans_ail_delete() will
@@ -550,13 +537,13 @@ xfs_buf_item_unpin(
 			ASSERT(bp->b_log_item == NULL);
 		}
 		xfs_buf_relse(bp);
-	} else if (freed && remove) {
+	} else if (remove) {
 		/*
 		 * The buffer must be locked and held by the caller to simulate
-		 * an async I/O failure.
+		 * an async I/O failure. We acquired the hold for this case
+		 * before the buffer was unpinned.
 		 */
 		xfs_buf_lock(bp);
-		xfs_buf_hold(bp);
 		bp->b_flags |= XBF_ASYNC;
 		xfs_buf_ioend_fail(bp);
 	}
@@ -714,7 +701,7 @@ xfs_buf_item_release(
 STATIC void
 xfs_buf_item_committing(
 	struct xfs_log_item	*lip,
-	xfs_lsn_t		commit_lsn)
+	xfs_csn_t		seq)
 {
 	return xfs_buf_item_release(lip);
 }
diff --git a/fs/xfs/xfs_discard.c b/fs/xfs/xfs_discard.c
index f979d0d7e6cd..736df5660f1f 100644
--- a/fs/xfs/xfs_discard.c
+++ b/fs/xfs/xfs_discard.c
@@ -8,7 +8,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_btree.h"
 #include "xfs_alloc_btree.h"
@@ -18,6 +17,7 @@
 #include "xfs_extent_busy.h"
 #include "xfs_trace.h"
 #include "xfs_log.h"
+#include "xfs_ag.h"
 
 STATIC int
 xfs_trim_extents(
@@ -50,7 +50,7 @@ xfs_trim_extents(
 		goto out_put_perag;
 	agf = agbp->b_addr;
 
-	cur = xfs_allocbt_init_cursor(mp, NULL, agbp, agno, XFS_BTNUM_CNT);
+	cur = xfs_allocbt_init_cursor(mp, NULL, agbp, pag, XFS_BTNUM_CNT);
 
 	/*
 	 * Look up the longest btree in the AGF and start with it.
@@ -108,7 +108,7 @@ xfs_trim_extents(
 		 * If any blocks in the range are still busy, skip the
 		 * discard and try again the next time.
 		 */
-		if (xfs_extent_busy_search(mp, agno, fbno, flen)) {
+		if (xfs_extent_busy_search(mp, pag, fbno, flen)) {
 			trace_xfs_discard_busy(mp, agno, fbno, flen);
 			goto next_extent;
 		}
diff --git a/fs/xfs/xfs_dquot_item.c b/fs/xfs/xfs_dquot_item.c
index 8c1fdf37ee8f..8ed47b739b6c 100644
--- a/fs/xfs/xfs_dquot_item.c
+++ b/fs/xfs/xfs_dquot_item.c
@@ -188,7 +188,7 @@ xfs_qm_dquot_logitem_release(
 STATIC void
 xfs_qm_dquot_logitem_committing(
 	struct xfs_log_item	*lip,
-	xfs_lsn_t		commit_lsn)
+	xfs_csn_t		seq)
 {
 	return xfs_qm_dquot_logitem_release(lip);
 }
diff --git a/fs/xfs/xfs_extent_busy.c b/fs/xfs/xfs_extent_busy.c
index a4075685d9eb..ad22a003f959 100644
--- a/fs/xfs/xfs_extent_busy.c
+++ b/fs/xfs/xfs_extent_busy.c
@@ -11,39 +11,37 @@
 #include "xfs_log_format.h"
 #include "xfs_shared.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_alloc.h"
 #include "xfs_extent_busy.h"
 #include "xfs_trace.h"
 #include "xfs_trans.h"
 #include "xfs_log.h"
+#include "xfs_ag.h"
 
 void
 xfs_extent_busy_insert(
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agblock_t		bno,
 	xfs_extlen_t		len,
 	unsigned int		flags)
 {
 	struct xfs_extent_busy	*new;
 	struct xfs_extent_busy	*busyp;
-	struct xfs_perag	*pag;
 	struct rb_node		**rbp;
 	struct rb_node		*parent = NULL;
 
 	new = kmem_zalloc(sizeof(struct xfs_extent_busy), 0);
-	new->agno = agno;
+	new->agno = pag->pag_agno;
 	new->bno = bno;
 	new->length = len;
 	INIT_LIST_HEAD(&new->list);
 	new->flags = flags;
 
 	/* trace before insert to be able to see failed inserts */
-	trace_xfs_extent_busy(tp->t_mountp, agno, bno, len);
+	trace_xfs_extent_busy(tp->t_mountp, pag->pag_agno, bno, len);
 
-	pag = xfs_perag_get(tp->t_mountp, new->agno);
 	spin_lock(&pag->pagb_lock);
 	rbp = &pag->pagb_tree.rb_node;
 	while (*rbp) {
@@ -66,7 +64,6 @@ xfs_extent_busy_insert(
 
 	list_add(&new->list, &tp->t_busy);
 	spin_unlock(&pag->pagb_lock);
-	xfs_perag_put(pag);
 }
 
 /*
@@ -81,21 +78,17 @@ xfs_extent_busy_insert(
 int
 xfs_extent_busy_search(
 	struct xfs_mount	*mp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agblock_t		bno,
 	xfs_extlen_t		len)
 {
-	struct xfs_perag	*pag;
 	struct rb_node		*rbp;
 	struct xfs_extent_busy	*busyp;
 	int			match = 0;
 
-	pag = xfs_perag_get(mp, agno);
+	/* find closest start bno overlap */
 	spin_lock(&pag->pagb_lock);
-
 	rbp = pag->pagb_tree.rb_node;
-
-	/* find closest start bno overlap */
 	while (rbp) {
 		busyp = rb_entry(rbp, struct xfs_extent_busy, rb_node);
 		if (bno < busyp->bno) {
@@ -115,7 +108,6 @@ xfs_extent_busy_search(
 		}
 	}
 	spin_unlock(&pag->pagb_lock);
-	xfs_perag_put(pag);
 	return match;
 }
 
@@ -281,17 +273,14 @@ out_force_log:
 void
 xfs_extent_busy_reuse(
 	struct xfs_mount	*mp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agblock_t		fbno,
 	xfs_extlen_t		flen,
 	bool			userdata)
 {
-	struct xfs_perag	*pag;
 	struct rb_node		*rbp;
 
 	ASSERT(flen > 0);
-
-	pag = xfs_perag_get(mp, agno);
 	spin_lock(&pag->pagb_lock);
 restart:
 	rbp = pag->pagb_tree.rb_node;
@@ -314,7 +303,6 @@ restart:
 			goto restart;
 	}
 	spin_unlock(&pag->pagb_lock);
-	xfs_perag_put(pag);
 }
 
 /*
@@ -605,12 +593,11 @@ void
 xfs_extent_busy_wait_all(
 	struct xfs_mount	*mp)
 {
+	struct xfs_perag	*pag;
 	DEFINE_WAIT		(wait);
 	xfs_agnumber_t		agno;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		struct xfs_perag *pag = xfs_perag_get(mp, agno);
-
+	for_each_perag(mp, agno, pag) {
 		do {
 			prepare_to_wait(&pag->pagb_wait, &wait, TASK_KILLABLE);
 			if  (RB_EMPTY_ROOT(&pag->pagb_tree))
@@ -618,8 +605,6 @@ xfs_extent_busy_wait_all(
 			schedule();
 		} while (1);
 		finish_wait(&pag->pagb_wait, &wait);
-
-		xfs_perag_put(pag);
 	}
 }
 
diff --git a/fs/xfs/xfs_extent_busy.h b/fs/xfs/xfs_extent_busy.h
index 8aea07100092..4a118131059f 100644
--- a/fs/xfs/xfs_extent_busy.h
+++ b/fs/xfs/xfs_extent_busy.h
@@ -9,6 +9,7 @@
 #define	__XFS_EXTENT_BUSY_H__
 
 struct xfs_mount;
+struct xfs_perag;
 struct xfs_trans;
 struct xfs_alloc_arg;
 
@@ -31,7 +32,7 @@ struct xfs_extent_busy {
 };
 
 void
-xfs_extent_busy_insert(struct xfs_trans *tp, xfs_agnumber_t agno,
+xfs_extent_busy_insert(struct xfs_trans *tp, struct xfs_perag *pag,
 	xfs_agblock_t bno, xfs_extlen_t len, unsigned int flags);
 
 void
@@ -39,11 +40,11 @@ xfs_extent_busy_clear(struct xfs_mount *mp, struct list_head *list,
 	bool do_discard);
 
 int
-xfs_extent_busy_search(struct xfs_mount *mp, xfs_agnumber_t agno,
+xfs_extent_busy_search(struct xfs_mount *mp, struct xfs_perag *pag,
 	xfs_agblock_t bno, xfs_extlen_t len);
 
 void
-xfs_extent_busy_reuse(struct xfs_mount *mp, xfs_agnumber_t agno,
+xfs_extent_busy_reuse(struct xfs_mount *mp, struct xfs_perag *pag,
 	xfs_agblock_t fbno, xfs_extlen_t flen, bool userdata);
 
 bool
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 3c0749ab9e40..cc3cfb12df53 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -119,8 +119,8 @@ xfs_dir_fsync(
 	return xfs_log_force_inode(ip);
 }
 
-static xfs_lsn_t
-xfs_fsync_lsn(
+static xfs_csn_t
+xfs_fsync_seq(
 	struct xfs_inode	*ip,
 	bool			datasync)
 {
@@ -128,7 +128,7 @@ xfs_fsync_lsn(
 		return 0;
 	if (datasync && !(ip->i_itemp->ili_fsync_fields & ~XFS_ILOG_TIMESTAMP))
 		return 0;
-	return ip->i_itemp->ili_last_lsn;
+	return ip->i_itemp->ili_commit_seq;
 }
 
 /*
@@ -151,12 +151,12 @@ xfs_fsync_flush_log(
 	int			*log_flushed)
 {
 	int			error = 0;
-	xfs_lsn_t		lsn;
+	xfs_csn_t		seq;
 
 	xfs_ilock(ip, XFS_ILOCK_SHARED);
-	lsn = xfs_fsync_lsn(ip, datasync);
-	if (lsn) {
-		error = xfs_log_force_lsn(ip->i_mount, lsn, XFS_LOG_SYNC,
+	seq = xfs_fsync_seq(ip, datasync);
+	if (seq) {
+		error = xfs_log_force_seq(ip->i_mount, seq, XFS_LOG_SYNC,
 					  log_flushed);
 
 		spin_lock(&ip->i_itemp->ili_lock);
@@ -197,9 +197,9 @@ xfs_file_fsync(
 	 * inode size in case of an extending write.
 	 */
 	if (XFS_IS_REALTIME_INODE(ip))
-		xfs_blkdev_issue_flush(mp->m_rtdev_targp);
+		blkdev_issue_flush(mp->m_rtdev_targp->bt_bdev);
 	else if (mp->m_logdev_targp != mp->m_ddev_targp)
-		xfs_blkdev_issue_flush(mp->m_ddev_targp);
+		blkdev_issue_flush(mp->m_ddev_targp->bt_bdev);
 
 	/*
 	 * Any inode that has dirty modifications in the log is pinned.  The
@@ -219,7 +219,7 @@ xfs_file_fsync(
 	 */
 	if (!log_flushed && !XFS_IS_REALTIME_INODE(ip) &&
 	    mp->m_logdev_targp == mp->m_ddev_targp)
-		xfs_blkdev_issue_flush(mp->m_ddev_targp);
+		blkdev_issue_flush(mp->m_ddev_targp->bt_bdev);
 
 	return error;
 }
@@ -384,21 +384,30 @@ restart:
 		}
 		goto restart;
 	}
+
 	/*
 	 * If the offset is beyond the size of the file, we need to zero any
 	 * blocks that fall between the existing EOF and the start of this
-	 * write.  If zeroing is needed and we are currently holding the
-	 * iolock shared, we need to update it to exclusive which implies
-	 * having to redo all checks before.
+	 * write.  If zeroing is needed and we are currently holding the iolock
+	 * shared, we need to update it to exclusive which implies having to
+	 * redo all checks before.
+	 *
+	 * We need to serialise against EOF updates that occur in IO completions
+	 * here. We want to make sure that nobody is changing the size while we
+	 * do this check until we have placed an IO barrier (i.e.  hold the
+	 * XFS_IOLOCK_EXCL) that prevents new IO from being dispatched.  The
+	 * spinlock effectively forms a memory barrier once we have the
+	 * XFS_IOLOCK_EXCL so we are guaranteed to see the latest EOF value and
+	 * hence be able to correctly determine if we need to run zeroing.
 	 *
-	 * We need to serialise against EOF updates that occur in IO
-	 * completions here. We want to make sure that nobody is changing the
-	 * size while we do this check until we have placed an IO barrier (i.e.
-	 * hold the XFS_IOLOCK_EXCL) that prevents new IO from being dispatched.
-	 * The spinlock effectively forms a memory barrier once we have the
-	 * XFS_IOLOCK_EXCL so we are guaranteed to see the latest EOF value
-	 * and hence be able to correctly determine if we need to run zeroing.
+	 * We can do an unlocked check here safely as IO completion can only
+	 * extend EOF. Truncate is locked out at this point, so the EOF can
+	 * not move backwards, only forwards. Hence we only need to take the
+	 * slow path and spin locks when we are at or beyond the current EOF.
 	 */
+	if (iocb->ki_pos <= i_size_read(inode))
+		goto out;
+
 	spin_lock(&ip->i_flags_lock);
 	isize = i_size_read(inode);
 	if (iocb->ki_pos > isize) {
@@ -426,7 +435,7 @@ restart:
 			drained_dio = true;
 			goto restart;
 		}
-	
+
 		trace_xfs_zero_eof(ip, isize, iocb->ki_pos - isize);
 		error = iomap_zero_range(inode, isize, iocb->ki_pos - isize,
 				NULL, &xfs_buffered_write_iomap_ops);
@@ -435,6 +444,7 @@ restart:
 	} else
 		spin_unlock(&ip->i_flags_lock);
 
+out:
 	return file_modified(file);
 }
 
@@ -500,7 +510,17 @@ xfs_dio_write_end_io(
 	 * other IO completions here to update the EOF. Failing to serialise
 	 * here can result in EOF moving backwards and Bad Things Happen when
 	 * that occurs.
+	 *
+	 * As IO completion only ever extends EOF, we can do an unlocked check
+	 * here to avoid taking the spinlock. If we land within the current EOF,
+	 * then we do not need to do an extending update at all, and we don't
+	 * need to take the lock to check this. If we race with an update moving
+	 * EOF, then we'll either still be beyond EOF and need to take the lock,
+	 * or we'll be within EOF and we don't need to take it at all.
 	 */
+	if (offset + size <= i_size_read(inode))
+		goto out;
+
 	spin_lock(&ip->i_flags_lock);
 	if (offset + size > i_size_read(inode)) {
 		i_size_write(inode, offset + size);
@@ -749,18 +769,18 @@ write_retry:
 	 */
 	if (ret == -EDQUOT && !cleared_space) {
 		xfs_iunlock(ip, iolock);
-		xfs_blockgc_free_quota(ip, XFS_EOF_FLAGS_SYNC);
+		xfs_blockgc_free_quota(ip, XFS_ICWALK_FLAG_SYNC);
 		cleared_space = true;
 		goto write_retry;
 	} else if (ret == -ENOSPC && !cleared_space) {
-		struct xfs_eofblocks eofb = {0};
+		struct xfs_icwalk	icw = {0};
 
 		cleared_space = true;
 		xfs_flush_inodes(ip->i_mount);
 
 		xfs_iunlock(ip, iolock);
-		eofb.eof_flags = XFS_EOF_FLAGS_SYNC;
-		xfs_blockgc_free_space(ip->i_mount, &eofb);
+		icw.icw_flags = XFS_ICWALK_FLAG_SYNC;
+		xfs_blockgc_free_space(ip->i_mount, &icw);
 		goto write_retry;
 	}
 
diff --git a/fs/xfs/xfs_filestream.c b/fs/xfs/xfs_filestream.c
index db23e455eb91..eed6ca5f8f91 100644
--- a/fs/xfs/xfs_filestream.c
+++ b/fs/xfs/xfs_filestream.c
@@ -9,13 +9,13 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_inode.h"
 #include "xfs_bmap.h"
 #include "xfs_alloc.h"
 #include "xfs_mru_cache.h"
 #include "xfs_trace.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 #include "xfs_trans.h"
 #include "xfs_filestream.h"
diff --git a/fs/xfs/xfs_fsmap.c b/fs/xfs/xfs_fsmap.c
index 34f2b971ce43..7d0b09c1366e 100644
--- a/fs/xfs/xfs_fsmap.c
+++ b/fs/xfs/xfs_fsmap.c
@@ -24,6 +24,7 @@
 #include "xfs_refcount_btree.h"
 #include "xfs_alloc_btree.h"
 #include "xfs_rtalloc.h"
+#include "xfs_ag.h"
 
 /* Convert an xfs_fsmap to an fsmap. */
 static void
@@ -157,10 +158,10 @@ struct xfs_getfsmap_info {
 	struct xfs_fsmap_head	*head;
 	struct fsmap		*fsmap_recs;	/* mapping records */
 	struct xfs_buf		*agf_bp;	/* AGF, for refcount queries */
+	struct xfs_perag	*pag;		/* AG info, if applicable */
 	xfs_daddr_t		next_daddr;	/* next daddr we expect */
 	u64			missing_owner;	/* owner of holes */
 	u32			dev;		/* device id */
-	xfs_agnumber_t		agno;		/* AG number, if applicable */
 	struct xfs_rmap_irec	low;		/* low rmap key */
 	struct xfs_rmap_irec	high;		/* high rmap key */
 	bool			last;		/* last extent? */
@@ -203,14 +204,13 @@ xfs_getfsmap_is_shared(
 	*stat = false;
 	if (!xfs_sb_version_hasreflink(&mp->m_sb))
 		return 0;
-	/* rt files will have agno set to NULLAGNUMBER */
-	if (info->agno == NULLAGNUMBER)
+	/* rt files will have no perag structure */
+	if (!info->pag)
 		return 0;
 
 	/* Are there any shared blocks here? */
 	flen = 0;
-	cur = xfs_refcountbt_init_cursor(mp, tp, info->agf_bp,
-			info->agno);
+	cur = xfs_refcountbt_init_cursor(mp, tp, info->agf_bp, info->pag);
 
 	error = xfs_refcount_find_shared(cur, rec->rm_startblock,
 			rec->rm_blockcount, &fbno, &flen, false);
@@ -311,7 +311,8 @@ xfs_getfsmap_helper(
 	if (info->head->fmh_entries >= info->head->fmh_count)
 		return -ECANCELED;
 
-	trace_xfs_fsmap_mapping(mp, info->dev, info->agno, rec);
+	trace_xfs_fsmap_mapping(mp, info->dev,
+			info->pag ? info->pag->pag_agno : NULLAGNUMBER, rec);
 
 	fmr.fmr_device = info->dev;
 	fmr.fmr_physical = rec_daddr;
@@ -354,7 +355,7 @@ xfs_getfsmap_datadev_helper(
 	xfs_fsblock_t			fsb;
 	xfs_daddr_t			rec_daddr;
 
-	fsb = XFS_AGB_TO_FSB(mp, cur->bc_ag.agno, rec->rm_startblock);
+	fsb = XFS_AGB_TO_FSB(mp, cur->bc_ag.pag->pag_agno, rec->rm_startblock);
 	rec_daddr = XFS_FSB_TO_DADDR(mp, fsb);
 
 	return xfs_getfsmap_helper(cur->bc_tp, info, rec, rec_daddr);
@@ -372,7 +373,7 @@ xfs_getfsmap_datadev_bnobt_helper(
 	struct xfs_rmap_irec		irec;
 	xfs_daddr_t			rec_daddr;
 
-	rec_daddr = XFS_AGB_TO_DADDR(mp, cur->bc_ag.agno,
+	rec_daddr = XFS_AGB_TO_DADDR(mp, cur->bc_ag.pag->pag_agno,
 			rec->ar_startblock);
 
 	irec.rm_startblock = rec->ar_startblock;
@@ -429,8 +430,8 @@ xfs_getfsmap_logdev(
 	info->high.rm_flags = XFS_RMAP_KEY_FLAGS | XFS_RMAP_REC_FLAGS;
 	info->missing_owner = XFS_FMR_OWN_FREE;
 
-	trace_xfs_fsmap_low_key(mp, info->dev, info->agno, &info->low);
-	trace_xfs_fsmap_high_key(mp, info->dev, info->agno, &info->high);
+	trace_xfs_fsmap_low_key(mp, info->dev, NULLAGNUMBER, &info->low);
+	trace_xfs_fsmap_high_key(mp, info->dev, NULLAGNUMBER, &info->high);
 
 	if (keys[0].fmr_physical > 0)
 		return 0;
@@ -508,8 +509,8 @@ __xfs_getfsmap_rtdev(
 	info->high.rm_blockcount = 0;
 	xfs_getfsmap_set_irec_flags(&info->high, &keys[1]);
 
-	trace_xfs_fsmap_low_key(mp, info->dev, info->agno, &info->low);
-	trace_xfs_fsmap_high_key(mp, info->dev, info->agno, &info->high);
+	trace_xfs_fsmap_low_key(mp, info->dev, NULLAGNUMBER, &info->low);
+	trace_xfs_fsmap_high_key(mp, info->dev, NULLAGNUMBER, &info->high);
 
 	return query_fn(tp, info);
 }
@@ -572,6 +573,7 @@ __xfs_getfsmap_datadev(
 	void				*priv)
 {
 	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_perag		*pag;
 	struct xfs_btree_cur		*bt_cur = NULL;
 	xfs_fsblock_t			start_fsb;
 	xfs_fsblock_t			end_fsb;
@@ -610,20 +612,20 @@ __xfs_getfsmap_datadev(
 	start_ag = XFS_FSB_TO_AGNO(mp, start_fsb);
 	end_ag = XFS_FSB_TO_AGNO(mp, end_fsb);
 
-	/* Query each AG */
-	for (info->agno = start_ag; info->agno <= end_ag; info->agno++) {
+	for_each_perag_range(mp, start_ag, end_ag, pag) {
 		/*
 		 * Set the AG high key from the fsmap high key if this
 		 * is the last AG that we're querying.
 		 */
-		if (info->agno == end_ag) {
+		info->pag = pag;
+		if (pag->pag_agno == end_ag) {
 			info->high.rm_startblock = XFS_FSB_TO_AGBNO(mp,
 					end_fsb);
 			info->high.rm_offset = XFS_BB_TO_FSBT(mp,
 					keys[1].fmr_offset);
 			error = xfs_fsmap_owner_to_rmap(&info->high, &keys[1]);
 			if (error)
-				goto err;
+				break;
 			xfs_getfsmap_set_irec_flags(&info->high, &keys[1]);
 		}
 
@@ -634,38 +636,45 @@ __xfs_getfsmap_datadev(
 			info->agf_bp = NULL;
 		}
 
-		error = xfs_alloc_read_agf(mp, tp, info->agno, 0,
+		error = xfs_alloc_read_agf(mp, tp, pag->pag_agno, 0,
 				&info->agf_bp);
 		if (error)
-			goto err;
+			break;
 
-		trace_xfs_fsmap_low_key(mp, info->dev, info->agno, &info->low);
-		trace_xfs_fsmap_high_key(mp, info->dev, info->agno,
+		trace_xfs_fsmap_low_key(mp, info->dev, pag->pag_agno,
+				&info->low);
+		trace_xfs_fsmap_high_key(mp, info->dev, pag->pag_agno,
 				&info->high);
 
 		error = query_fn(tp, info, &bt_cur, priv);
 		if (error)
-			goto err;
+			break;
 
 		/*
 		 * Set the AG low key to the start of the AG prior to
 		 * moving on to the next AG.
 		 */
-		if (info->agno == start_ag) {
+		if (pag->pag_agno == start_ag) {
 			info->low.rm_startblock = 0;
 			info->low.rm_owner = 0;
 			info->low.rm_offset = 0;
 			info->low.rm_flags = 0;
 		}
-	}
 
-	/* Report any gap at the end of the AG */
-	info->last = true;
-	error = query_fn(tp, info, &bt_cur, priv);
-	if (error)
-		goto err;
+		/*
+		 * If this is the last AG, report any gap at the end of it
+		 * before we drop the reference to the perag when the loop
+		 * terminates.
+		 */
+		if (pag->pag_agno == end_ag) {
+			info->last = true;
+			error = query_fn(tp, info, &bt_cur, priv);
+			if (error)
+				break;
+		}
+		info->pag = NULL;
+	}
 
-err:
 	if (bt_cur)
 		xfs_btree_del_cursor(bt_cur, error < 0 ? XFS_BTREE_ERROR :
 							 XFS_BTREE_NOERROR);
@@ -673,6 +682,13 @@ err:
 		xfs_trans_brelse(tp, info->agf_bp);
 		info->agf_bp = NULL;
 	}
+	if (info->pag) {
+		xfs_perag_put(info->pag);
+		info->pag = NULL;
+	} else if (pag) {
+		/* loop termination case */
+		xfs_perag_put(pag);
+	}
 
 	return error;
 }
@@ -691,7 +707,7 @@ xfs_getfsmap_datadev_rmapbt_query(
 
 	/* Allocate cursor for this AG and query_range it. */
 	*curpp = xfs_rmapbt_init_cursor(tp->t_mountp, tp, info->agf_bp,
-			info->agno);
+			info->pag);
 	return xfs_rmap_query_range(*curpp, &info->low, &info->high,
 			xfs_getfsmap_datadev_helper, info);
 }
@@ -724,7 +740,7 @@ xfs_getfsmap_datadev_bnobt_query(
 
 	/* Allocate cursor for this AG and query_range it. */
 	*curpp = xfs_allocbt_init_cursor(tp->t_mountp, tp, info->agf_bp,
-			info->agno, XFS_BTNUM_BNO);
+			info->pag, XFS_BTNUM_BNO);
 	key->ar_startblock = info->low.rm_startblock;
 	key[1].ar_startblock = info->high.rm_startblock;
 	return xfs_alloc_query_range(*curpp, key, &key[1],
@@ -937,7 +953,7 @@ xfs_getfsmap(
 
 		info.dev = handlers[i].dev;
 		info.last = false;
-		info.agno = NULLAGNUMBER;
+		info.pag = NULL;
 		error = handlers[i].fn(tp, dkeys, &info);
 		if (error)
 			break;
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
index be9cf88d2ad7..6ed29b158312 100644
--- a/fs/xfs/xfs_fsops.c
+++ b/fs/xfs/xfs_fsops.c
@@ -538,25 +538,25 @@ xfs_do_force_shutdown(
 
 	if (flags & SHUTDOWN_FORCE_UMOUNT) {
 		xfs_alert(mp,
-"User initiated shutdown received. Shutting down filesystem");
+"User initiated shutdown (0x%x) received. Shutting down filesystem",
+				flags);
 		return;
 	}
 
-	xfs_notice(mp,
-"%s(0x%x) called from line %d of file %s. Return address = "PTR_FMT,
-		__func__, flags, lnnum, fname, __return_address);
-
 	if (flags & SHUTDOWN_CORRUPT_INCORE) {
 		xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_CORRUPT,
-"Corruption of in-memory data detected.  Shutting down filesystem");
+"Corruption of in-memory data (0x%x) detected at %pS (%s:%d).  Shutting down filesystem",
+				flags, __return_address, fname, lnnum);
 		if (XFS_ERRLEVEL_HIGH <= xfs_error_level)
 			xfs_stack_trace();
 	} else if (logerror) {
 		xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_LOGERROR,
-			"Log I/O Error Detected. Shutting down filesystem");
+"Log I/O error (0x%x) detected at %pS (%s:%d). Shutting down filesystem",
+				flags, __return_address, fname, lnnum);
 	} else {
 		xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_IOERROR,
-			"I/O Error Detected. Shutting down filesystem");
+"I/O error (0x%x) detected at %pS (%s:%d). Shutting down filesystem",
+				flags, __return_address, fname, lnnum);
 	}
 
 	xfs_alert(mp,
@@ -576,10 +576,8 @@ xfs_fs_reserve_ag_blocks(
 	int			err2;
 
 	mp->m_finobt_nores = false;
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		pag = xfs_perag_get(mp, agno);
+	for_each_perag(mp, agno, pag) {
 		err2 = xfs_ag_resv_init(pag, NULL);
-		xfs_perag_put(pag);
 		if (err2 && !error)
 			error = err2;
 	}
@@ -605,10 +603,8 @@ xfs_fs_unreserve_ag_blocks(
 	int			error = 0;
 	int			err2;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		pag = xfs_perag_get(mp, agno);
+	for_each_perag(mp, agno, pag) {
 		err2 = xfs_ag_resv_free(pag);
-		xfs_perag_put(pag);
 		if (err2 && !error)
 			error = err2;
 	}
diff --git a/fs/xfs/xfs_health.c b/fs/xfs/xfs_health.c
index 8e0cb05a7142..eb10eacabc8f 100644
--- a/fs/xfs/xfs_health.c
+++ b/fs/xfs/xfs_health.c
@@ -9,11 +9,11 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_inode.h"
 #include "xfs_trace.h"
 #include "xfs_health.h"
+#include "xfs_ag.h"
 
 /*
  * Warn about metadata corruption that we detected but haven't fixed, and
@@ -34,14 +34,12 @@ xfs_health_unmount(
 		return;
 
 	/* Measure AG corruption levels. */
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		pag = xfs_perag_get(mp, agno);
+	for_each_perag(mp, agno, pag) {
 		xfs_ag_measure_sickness(pag, &sick, &checked);
 		if (sick) {
 			trace_xfs_ag_unfixed_corruption(mp, agno, sick);
 			warn = true;
 		}
-		xfs_perag_put(pag);
 	}
 
 	/* Measure realtime volume corruption levels. */
@@ -231,6 +229,15 @@ xfs_inode_mark_sick(
 	ip->i_sick |= mask;
 	ip->i_checked |= mask;
 	spin_unlock(&ip->i_flags_lock);
+
+	/*
+	 * Keep this inode around so we don't lose the sickness report.  Scrub
+	 * grabs inodes with DONTCACHE assuming that most inode are ok, which
+	 * is not the case here.
+	 */
+	spin_lock(&VFS_I(ip)->i_lock);
+	VFS_I(ip)->i_state &= ~I_DONTCACHE;
+	spin_unlock(&VFS_I(ip)->i_lock);
 }
 
 /* Mark parts of an inode healed. */
diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index 3c81daca0e9a..6007683482c6 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -9,7 +9,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_inode.h"
 #include "xfs_trans.h"
@@ -23,9 +22,65 @@
 #include "xfs_dquot.h"
 #include "xfs_reflink.h"
 #include "xfs_ialloc.h"
+#include "xfs_ag.h"
 
 #include <linux/iversion.h>
 
+/* Radix tree tags for incore inode tree. */
+
+/* inode is to be reclaimed */
+#define XFS_ICI_RECLAIM_TAG	0
+/* Inode has speculative preallocations (posteof or cow) to clean. */
+#define XFS_ICI_BLOCKGC_TAG	1
+
+/*
+ * The goal for walking incore inodes.  These can correspond with incore inode
+ * radix tree tags when convenient.  Avoid existing XFS_IWALK namespace.
+ */
+enum xfs_icwalk_goal {
+	/* Goals that are not related to tags; these must be < 0. */
+	XFS_ICWALK_DQRELE	= -1,
+
+	/* Goals directly associated with tagged inodes. */
+	XFS_ICWALK_BLOCKGC	= XFS_ICI_BLOCKGC_TAG,
+	XFS_ICWALK_RECLAIM	= XFS_ICI_RECLAIM_TAG,
+};
+
+#define XFS_ICWALK_NULL_TAG	(-1U)
+
+/* Compute the inode radix tree tag for this goal. */
+static inline unsigned int
+xfs_icwalk_tag(enum xfs_icwalk_goal goal)
+{
+	return goal < 0 ? XFS_ICWALK_NULL_TAG : goal;
+}
+
+static int xfs_icwalk(struct xfs_mount *mp,
+		enum xfs_icwalk_goal goal, struct xfs_icwalk *icw);
+static int xfs_icwalk_ag(struct xfs_perag *pag,
+		enum xfs_icwalk_goal goal, struct xfs_icwalk *icw);
+
+/*
+ * Private inode cache walk flags for struct xfs_icwalk.  Must not
+ * coincide with XFS_ICWALK_FLAGS_VALID.
+ */
+#define XFS_ICWALK_FLAG_DROP_UDQUOT	(1U << 31)
+#define XFS_ICWALK_FLAG_DROP_GDQUOT	(1U << 30)
+#define XFS_ICWALK_FLAG_DROP_PDQUOT	(1U << 29)
+
+/* Stop scanning after icw_scan_limit inodes. */
+#define XFS_ICWALK_FLAG_SCAN_LIMIT	(1U << 28)
+
+#define XFS_ICWALK_FLAG_RECLAIM_SICK	(1U << 27)
+#define XFS_ICWALK_FLAG_UNION		(1U << 26) /* union filter algorithm */
+
+#define XFS_ICWALK_PRIVATE_FLAGS	(XFS_ICWALK_FLAG_DROP_UDQUOT | \
+					 XFS_ICWALK_FLAG_DROP_GDQUOT | \
+					 XFS_ICWALK_FLAG_DROP_PDQUOT | \
+					 XFS_ICWALK_FLAG_SCAN_LIMIT | \
+					 XFS_ICWALK_FLAG_RECLAIM_SICK | \
+					 XFS_ICWALK_FLAG_UNION)
+
 /*
  * Allocate and initialise an xfs_inode.
  */
@@ -157,46 +212,94 @@ xfs_reclaim_work_queue(
 	rcu_read_unlock();
 }
 
-static void
-xfs_perag_set_reclaim_tag(
+/*
+ * Background scanning to trim preallocated space. This is queued based on the
+ * 'speculative_prealloc_lifetime' tunable (5m by default).
+ */
+static inline void
+xfs_blockgc_queue(
 	struct xfs_perag	*pag)
 {
+	rcu_read_lock();
+	if (radix_tree_tagged(&pag->pag_ici_root, XFS_ICI_BLOCKGC_TAG))
+		queue_delayed_work(pag->pag_mount->m_gc_workqueue,
+				   &pag->pag_blockgc_work,
+				   msecs_to_jiffies(xfs_blockgc_secs * 1000));
+	rcu_read_unlock();
+}
+
+/* Set a tag on both the AG incore inode tree and the AG radix tree. */
+static void
+xfs_perag_set_inode_tag(
+	struct xfs_perag	*pag,
+	xfs_agino_t		agino,
+	unsigned int		tag)
+{
 	struct xfs_mount	*mp = pag->pag_mount;
+	bool			was_tagged;
 
 	lockdep_assert_held(&pag->pag_ici_lock);
-	if (pag->pag_ici_reclaimable++)
+
+	was_tagged = radix_tree_tagged(&pag->pag_ici_root, tag);
+	radix_tree_tag_set(&pag->pag_ici_root, agino, tag);
+
+	if (tag == XFS_ICI_RECLAIM_TAG)
+		pag->pag_ici_reclaimable++;
+
+	if (was_tagged)
 		return;
 
-	/* propagate the reclaim tag up into the perag radix tree */
+	/* propagate the tag up into the perag radix tree */
 	spin_lock(&mp->m_perag_lock);
-	radix_tree_tag_set(&mp->m_perag_tree, pag->pag_agno,
-			   XFS_ICI_RECLAIM_TAG);
+	radix_tree_tag_set(&mp->m_perag_tree, pag->pag_agno, tag);
 	spin_unlock(&mp->m_perag_lock);
 
-	/* schedule periodic background inode reclaim */
-	xfs_reclaim_work_queue(mp);
+	/* start background work */
+	switch (tag) {
+	case XFS_ICI_RECLAIM_TAG:
+		xfs_reclaim_work_queue(mp);
+		break;
+	case XFS_ICI_BLOCKGC_TAG:
+		xfs_blockgc_queue(pag);
+		break;
+	}
 
-	trace_xfs_perag_set_reclaim(mp, pag->pag_agno, -1, _RET_IP_);
+	trace_xfs_perag_set_inode_tag(mp, pag->pag_agno, tag, _RET_IP_);
 }
 
+/* Clear a tag on both the AG incore inode tree and the AG radix tree. */
 static void
-xfs_perag_clear_reclaim_tag(
-	struct xfs_perag	*pag)
+xfs_perag_clear_inode_tag(
+	struct xfs_perag	*pag,
+	xfs_agino_t		agino,
+	unsigned int		tag)
 {
 	struct xfs_mount	*mp = pag->pag_mount;
 
 	lockdep_assert_held(&pag->pag_ici_lock);
-	if (--pag->pag_ici_reclaimable)
+
+	/*
+	 * Reclaim can signal (with a null agino) that it cleared its own tag
+	 * by removing the inode from the radix tree.
+	 */
+	if (agino != NULLAGINO)
+		radix_tree_tag_clear(&pag->pag_ici_root, agino, tag);
+	else
+		ASSERT(tag == XFS_ICI_RECLAIM_TAG);
+
+	if (tag == XFS_ICI_RECLAIM_TAG)
+		pag->pag_ici_reclaimable--;
+
+	if (radix_tree_tagged(&pag->pag_ici_root, tag))
 		return;
 
-	/* clear the reclaim tag from the perag radix tree */
+	/* clear the tag from the perag radix tree */
 	spin_lock(&mp->m_perag_lock);
-	radix_tree_tag_clear(&mp->m_perag_tree, pag->pag_agno,
-			     XFS_ICI_RECLAIM_TAG);
+	radix_tree_tag_clear(&mp->m_perag_tree, pag->pag_agno, tag);
 	spin_unlock(&mp->m_perag_lock);
-	trace_xfs_perag_clear_reclaim(mp, pag->pag_agno, -1, _RET_IP_);
-}
 
+	trace_xfs_perag_clear_inode_tag(mp, pag->pag_agno, tag, _RET_IP_);
+}
 
 /*
  * We set the inode flag atomically with the radix tree tag.
@@ -204,7 +307,7 @@ xfs_perag_clear_reclaim_tag(
  * can go away.
  */
 void
-xfs_inode_set_reclaim_tag(
+xfs_inode_mark_reclaimable(
 	struct xfs_inode	*ip)
 {
 	struct xfs_mount	*mp = ip->i_mount;
@@ -214,9 +317,8 @@ xfs_inode_set_reclaim_tag(
 	spin_lock(&pag->pag_ici_lock);
 	spin_lock(&ip->i_flags_lock);
 
-	radix_tree_tag_set(&pag->pag_ici_root, XFS_INO_TO_AGINO(mp, ip->i_ino),
-			   XFS_ICI_RECLAIM_TAG);
-	xfs_perag_set_reclaim_tag(pag);
+	xfs_perag_set_inode_tag(pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
+			XFS_ICI_RECLAIM_TAG);
 	__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
 
 	spin_unlock(&ip->i_flags_lock);
@@ -224,18 +326,7 @@ xfs_inode_set_reclaim_tag(
 	xfs_perag_put(pag);
 }
 
-STATIC void
-xfs_inode_clear_reclaim_tag(
-	struct xfs_perag	*pag,
-	xfs_ino_t		ino)
-{
-	radix_tree_tag_clear(&pag->pag_ici_root,
-			     XFS_INO_TO_AGINO(pag->pag_mount, ino),
-			     XFS_ICI_RECLAIM_TAG);
-	xfs_perag_clear_reclaim_tag(pag);
-}
-
-static void
+static inline void
 xfs_inew_wait(
 	struct xfs_inode	*ip)
 {
@@ -264,14 +355,14 @@ xfs_reinit_inode(
 	struct xfs_mount	*mp,
 	struct inode		*inode)
 {
-	int		error;
-	uint32_t	nlink = inode->i_nlink;
-	uint32_t	generation = inode->i_generation;
-	uint64_t	version = inode_peek_iversion(inode);
-	umode_t		mode = inode->i_mode;
-	dev_t		dev = inode->i_rdev;
-	kuid_t		uid = inode->i_uid;
-	kgid_t		gid = inode->i_gid;
+	int			error;
+	uint32_t		nlink = inode->i_nlink;
+	uint32_t		generation = inode->i_generation;
+	uint64_t		version = inode_peek_iversion(inode);
+	umode_t			mode = inode->i_mode;
+	dev_t			dev = inode->i_rdev;
+	kuid_t			uid = inode->i_uid;
+	kgid_t			gid = inode->i_gid;
 
 	error = inode_init_always(mp->m_super, inode);
 
@@ -286,6 +377,74 @@ xfs_reinit_inode(
 }
 
 /*
+ * Carefully nudge an inode whose VFS state has been torn down back into a
+ * usable state.  Drops the i_flags_lock and the rcu read lock.
+ */
+static int
+xfs_iget_recycle(
+	struct xfs_perag	*pag,
+	struct xfs_inode	*ip) __releases(&ip->i_flags_lock)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct inode		*inode = VFS_I(ip);
+	int			error;
+
+	trace_xfs_iget_recycle(ip);
+
+	/*
+	 * We need to make it look like the inode is being reclaimed to prevent
+	 * the actual reclaim workers from stomping over us while we recycle
+	 * the inode.  We can't clear the radix tree tag yet as it requires
+	 * pag_ici_lock to be held exclusive.
+	 */
+	ip->i_flags |= XFS_IRECLAIM;
+
+	spin_unlock(&ip->i_flags_lock);
+	rcu_read_unlock();
+
+	ASSERT(!rwsem_is_locked(&inode->i_rwsem));
+	error = xfs_reinit_inode(mp, inode);
+	if (error) {
+		bool	wake;
+
+		/*
+		 * Re-initializing the inode failed, and we are in deep
+		 * trouble.  Try to re-add it to the reclaim list.
+		 */
+		rcu_read_lock();
+		spin_lock(&ip->i_flags_lock);
+		wake = !!__xfs_iflags_test(ip, XFS_INEW);
+		ip->i_flags &= ~(XFS_INEW | XFS_IRECLAIM);
+		if (wake)
+			wake_up_bit(&ip->i_flags, __XFS_INEW_BIT);
+		ASSERT(ip->i_flags & XFS_IRECLAIMABLE);
+		spin_unlock(&ip->i_flags_lock);
+		rcu_read_unlock();
+
+		trace_xfs_iget_recycle_fail(ip);
+		return error;
+	}
+
+	spin_lock(&pag->pag_ici_lock);
+	spin_lock(&ip->i_flags_lock);
+
+	/*
+	 * Clear the per-lifetime state in the inode as we are now effectively
+	 * a new inode and need to return to the initial state before reuse
+	 * occurs.
+	 */
+	ip->i_flags &= ~XFS_IRECLAIM_RESET_FLAGS;
+	ip->i_flags |= XFS_INEW;
+	xfs_perag_clear_inode_tag(pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
+			XFS_ICI_RECLAIM_TAG);
+	inode->i_state = I_NEW;
+	spin_unlock(&ip->i_flags_lock);
+	spin_unlock(&pag->pag_ici_lock);
+
+	return 0;
+}
+
+/*
  * If we are allocating a new inode, then check what was returned is
  * actually a free, empty inode. If we are not allocating an inode,
  * then check we didn't find a free inode.
@@ -348,30 +507,21 @@ xfs_iget_cache_hit(
 	 * will not match, so check for that, too.
 	 */
 	spin_lock(&ip->i_flags_lock);
-	if (ip->i_ino != ino) {
-		trace_xfs_iget_skip(ip);
-		XFS_STATS_INC(mp, xs_ig_frecycle);
-		error = -EAGAIN;
-		goto out_error;
-	}
-
+	if (ip->i_ino != ino)
+		goto out_skip;
 
 	/*
 	 * If we are racing with another cache hit that is currently
 	 * instantiating this inode or currently recycling it out of
-	 * reclaimabe state, wait for the initialisation to complete
+	 * reclaimable state, wait for the initialisation to complete
 	 * before continuing.
 	 *
 	 * XXX(hch): eventually we should do something equivalent to
 	 *	     wait_on_inode to wait for these flags to be cleared
 	 *	     instead of polling for it.
 	 */
-	if (ip->i_flags & (XFS_INEW|XFS_IRECLAIM)) {
-		trace_xfs_iget_skip(ip);
-		XFS_STATS_INC(mp, xs_ig_frecycle);
-		error = -EAGAIN;
-		goto out_error;
-	}
+	if (ip->i_flags & (XFS_INEW | XFS_IRECLAIM))
+		goto out_skip;
 
 	/*
 	 * Check the inode free state is valid. This also detects lookup
@@ -381,72 +531,21 @@ xfs_iget_cache_hit(
 	if (error)
 		goto out_error;
 
-	/*
-	 * If IRECLAIMABLE is set, we've torn down the VFS inode already.
-	 * Need to carefully get it back into useable state.
-	 */
-	if (ip->i_flags & XFS_IRECLAIMABLE) {
-		trace_xfs_iget_reclaim(ip);
-
-		if (flags & XFS_IGET_INCORE) {
-			error = -EAGAIN;
-			goto out_error;
-		}
-
-		/*
-		 * We need to set XFS_IRECLAIM to prevent xfs_reclaim_inode
-		 * from stomping over us while we recycle the inode.  We can't
-		 * clear the radix tree reclaimable tag yet as it requires
-		 * pag_ici_lock to be held exclusive.
-		 */
-		ip->i_flags |= XFS_IRECLAIM;
-
-		spin_unlock(&ip->i_flags_lock);
-		rcu_read_unlock();
+	/* Skip inodes that have no vfs state. */
+	if ((flags & XFS_IGET_INCORE) &&
+	    (ip->i_flags & XFS_IRECLAIMABLE))
+		goto out_skip;
 
-		ASSERT(!rwsem_is_locked(&inode->i_rwsem));
-		error = xfs_reinit_inode(mp, inode);
-		if (error) {
-			bool wake;
-			/*
-			 * Re-initializing the inode failed, and we are in deep
-			 * trouble.  Try to re-add it to the reclaim list.
-			 */
-			rcu_read_lock();
-			spin_lock(&ip->i_flags_lock);
-			wake = !!__xfs_iflags_test(ip, XFS_INEW);
-			ip->i_flags &= ~(XFS_INEW | XFS_IRECLAIM);
-			if (wake)
-				wake_up_bit(&ip->i_flags, __XFS_INEW_BIT);
-			ASSERT(ip->i_flags & XFS_IRECLAIMABLE);
-			trace_xfs_iget_reclaim_fail(ip);
-			goto out_error;
-		}
-
-		spin_lock(&pag->pag_ici_lock);
-		spin_lock(&ip->i_flags_lock);
-
-		/*
-		 * Clear the per-lifetime state in the inode as we are now
-		 * effectively a new inode and need to return to the initial
-		 * state before reuse occurs.
-		 */
-		ip->i_flags &= ~XFS_IRECLAIM_RESET_FLAGS;
-		ip->i_flags |= XFS_INEW;
-		xfs_inode_clear_reclaim_tag(pag, ip->i_ino);
-		inode->i_state = I_NEW;
-		ip->i_sick = 0;
-		ip->i_checked = 0;
-
-		spin_unlock(&ip->i_flags_lock);
-		spin_unlock(&pag->pag_ici_lock);
+	/* The inode fits the selection criteria; process it. */
+	if (ip->i_flags & XFS_IRECLAIMABLE) {
+		/* Drops i_flags_lock and RCU read lock. */
+		error = xfs_iget_recycle(pag, ip);
+		if (error)
+			return error;
 	} else {
 		/* If the VFS inode is being torn down, pause and try again. */
-		if (!igrab(inode)) {
-			trace_xfs_iget_skip(ip);
-			error = -EAGAIN;
-			goto out_error;
-		}
+		if (!igrab(inode))
+			goto out_skip;
 
 		/* We've got a live one. */
 		spin_unlock(&ip->i_flags_lock);
@@ -463,13 +562,16 @@ xfs_iget_cache_hit(
 
 	return 0;
 
+out_skip:
+	trace_xfs_iget_skip(ip);
+	XFS_STATS_INC(mp, xs_ig_frecycle);
+	error = -EAGAIN;
 out_error:
 	spin_unlock(&ip->i_flags_lock);
 	rcu_read_unlock();
 	return error;
 }
 
-
 static int
 xfs_iget_cache_miss(
 	struct xfs_mount	*mp,
@@ -715,207 +817,96 @@ xfs_icache_inode_is_allocated(
 	return 0;
 }
 
-/*
- * The inode lookup is done in batches to keep the amount of lock traffic and
- * radix tree lookups to a minimum. The batch size is a trade off between
- * lookup reduction and stack usage. This is in the reclaim path, so we can't
- * be too greedy.
- */
-#define XFS_LOOKUP_BATCH	32
-
-/*
- * Decide if the given @ip is eligible to be a part of the inode walk, and
- * grab it if so.  Returns true if it's ready to go or false if we should just
- * ignore it.
- */
-STATIC bool
-xfs_inode_walk_ag_grab(
-	struct xfs_inode	*ip,
-	int			flags)
+#ifdef CONFIG_XFS_QUOTA
+/* Decide if we want to grab this inode to drop its dquots. */
+static bool
+xfs_dqrele_igrab(
+	struct xfs_inode	*ip)
 {
-	struct inode		*inode = VFS_I(ip);
-	bool			newinos = !!(flags & XFS_INODE_WALK_INEW_WAIT);
+	bool			ret = false;
 
 	ASSERT(rcu_read_lock_held());
 
 	/* Check for stale RCU freed inode */
 	spin_lock(&ip->i_flags_lock);
 	if (!ip->i_ino)
-		goto out_unlock_noent;
-
-	/* avoid new or reclaimable inodes. Leave for reclaim code to flush */
-	if ((!newinos && __xfs_iflags_test(ip, XFS_INEW)) ||
-	    __xfs_iflags_test(ip, XFS_IRECLAIMABLE | XFS_IRECLAIM))
-		goto out_unlock_noent;
-	spin_unlock(&ip->i_flags_lock);
-
-	/* nothing to sync during shutdown */
-	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
-		return false;
+		goto out_unlock;
 
-	/* If we can't grab the inode, it must on it's way to reclaim. */
-	if (!igrab(inode))
-		return false;
+	/*
+	 * Skip inodes that are anywhere in the reclaim machinery because we
+	 * drop dquots before tagging an inode for reclamation.
+	 */
+	if (ip->i_flags & (XFS_IRECLAIM | XFS_IRECLAIMABLE))
+		goto out_unlock;
 
-	/* inode is valid */
-	return true;
+	/*
+	 * The inode looks alive; try to grab a VFS reference so that it won't
+	 * get destroyed.  If we got the reference, return true to say that
+	 * we grabbed the inode.
+	 *
+	 * If we can't get the reference, then we know the inode had its VFS
+	 * state torn down and hasn't yet entered the reclaim machinery.  Since
+	 * we also know that dquots are detached from an inode before it enters
+	 * reclaim, we can skip the inode.
+	 */
+	ret = igrab(VFS_I(ip)) != NULL;
 
-out_unlock_noent:
+out_unlock:
 	spin_unlock(&ip->i_flags_lock);
-	return false;
+	return ret;
 }
 
-/*
- * For a given per-AG structure @pag, grab, @execute, and rele all incore
- * inodes with the given radix tree @tag.
- */
-STATIC int
-xfs_inode_walk_ag(
-	struct xfs_perag	*pag,
-	int			iter_flags,
-	int			(*execute)(struct xfs_inode *ip, void *args),
-	void			*args,
-	int			tag)
+/* Drop this inode's dquots. */
+static void
+xfs_dqrele_inode(
+	struct xfs_inode	*ip,
+	struct xfs_icwalk	*icw)
 {
-	struct xfs_mount	*mp = pag->pag_mount;
-	uint32_t		first_index;
-	int			last_error = 0;
-	int			skipped;
-	bool			done;
-	int			nr_found;
-
-restart:
-	done = false;
-	skipped = 0;
-	first_index = 0;
-	nr_found = 0;
-	do {
-		struct xfs_inode *batch[XFS_LOOKUP_BATCH];
-		int		error = 0;
-		int		i;
-
-		rcu_read_lock();
-
-		if (tag == XFS_ICI_NO_TAG)
-			nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
-					(void **)batch, first_index,
-					XFS_LOOKUP_BATCH);
-		else
-			nr_found = radix_tree_gang_lookup_tag(
-					&pag->pag_ici_root,
-					(void **) batch, first_index,
-					XFS_LOOKUP_BATCH, tag);
-
-		if (!nr_found) {
-			rcu_read_unlock();
-			break;
-		}
-
-		/*
-		 * Grab the inodes before we drop the lock. if we found
-		 * nothing, nr == 0 and the loop will be skipped.
-		 */
-		for (i = 0; i < nr_found; i++) {
-			struct xfs_inode *ip = batch[i];
-
-			if (done || !xfs_inode_walk_ag_grab(ip, iter_flags))
-				batch[i] = NULL;
-
-			/*
-			 * Update the index for the next lookup. Catch
-			 * overflows into the next AG range which can occur if
-			 * we have inodes in the last block of the AG and we
-			 * are currently pointing to the last inode.
-			 *
-			 * Because we may see inodes that are from the wrong AG
-			 * due to RCU freeing and reallocation, only update the
-			 * index if it lies in this AG. It was a race that lead
-			 * us to see this inode, so another lookup from the
-			 * same index will not find it again.
-			 */
-			if (XFS_INO_TO_AGNO(mp, ip->i_ino) != pag->pag_agno)
-				continue;
-			first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
-			if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
-				done = true;
-		}
-
-		/* unlock now we've grabbed the inodes. */
-		rcu_read_unlock();
-
-		for (i = 0; i < nr_found; i++) {
-			if (!batch[i])
-				continue;
-			if ((iter_flags & XFS_INODE_WALK_INEW_WAIT) &&
-			    xfs_iflags_test(batch[i], XFS_INEW))
-				xfs_inew_wait(batch[i]);
-			error = execute(batch[i], args);
-			xfs_irele(batch[i]);
-			if (error == -EAGAIN) {
-				skipped++;
-				continue;
-			}
-			if (error && last_error != -EFSCORRUPTED)
-				last_error = error;
-		}
-
-		/* bail out if the filesystem is corrupted.  */
-		if (error == -EFSCORRUPTED)
-			break;
+	if (xfs_iflags_test(ip, XFS_INEW))
+		xfs_inew_wait(ip);
 
-		cond_resched();
-
-	} while (nr_found && !done);
-
-	if (skipped) {
-		delay(1);
-		goto restart;
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if (icw->icw_flags & XFS_ICWALK_FLAG_DROP_UDQUOT) {
+		xfs_qm_dqrele(ip->i_udquot);
+		ip->i_udquot = NULL;
 	}
-	return last_error;
-}
-
-/* Fetch the next (possibly tagged) per-AG structure. */
-static inline struct xfs_perag *
-xfs_inode_walk_get_perag(
-	struct xfs_mount	*mp,
-	xfs_agnumber_t		agno,
-	int			tag)
-{
-	if (tag == XFS_ICI_NO_TAG)
-		return xfs_perag_get(mp, agno);
-	return xfs_perag_get_tag(mp, agno, tag);
+	if (icw->icw_flags & XFS_ICWALK_FLAG_DROP_GDQUOT) {
+		xfs_qm_dqrele(ip->i_gdquot);
+		ip->i_gdquot = NULL;
+	}
+	if (icw->icw_flags & XFS_ICWALK_FLAG_DROP_PDQUOT) {
+		xfs_qm_dqrele(ip->i_pdquot);
+		ip->i_pdquot = NULL;
+	}
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	xfs_irele(ip);
 }
 
 /*
- * Call the @execute function on all incore inodes matching the radix tree
- * @tag.
+ * Detach all dquots from incore inodes if we can.  The caller must already
+ * have dropped the relevant XFS_[UGP]QUOTA_ACTIVE flags so that dquots will
+ * not get reattached.
  */
 int
-xfs_inode_walk(
+xfs_dqrele_all_inodes(
 	struct xfs_mount	*mp,
-	int			iter_flags,
-	int			(*execute)(struct xfs_inode *ip, void *args),
-	void			*args,
-	int			tag)
+	unsigned int		qflags)
 {
-	struct xfs_perag	*pag;
-	int			error = 0;
-	int			last_error = 0;
-	xfs_agnumber_t		ag;
+	struct xfs_icwalk	icw = { .icw_flags = 0 };
 
-	ag = 0;
-	while ((pag = xfs_inode_walk_get_perag(mp, ag, tag))) {
-		ag = pag->pag_agno + 1;
-		error = xfs_inode_walk_ag(pag, iter_flags, execute, args, tag);
-		xfs_perag_put(pag);
-		if (error) {
-			last_error = error;
-			if (error == -EFSCORRUPTED)
-				break;
-		}
-	}
-	return last_error;
+	if (qflags & XFS_UQUOTA_ACCT)
+		icw.icw_flags |= XFS_ICWALK_FLAG_DROP_UDQUOT;
+	if (qflags & XFS_GQUOTA_ACCT)
+		icw.icw_flags |= XFS_ICWALK_FLAG_DROP_GDQUOT;
+	if (qflags & XFS_PQUOTA_ACCT)
+		icw.icw_flags |= XFS_ICWALK_FLAG_DROP_PDQUOT;
+
+	return xfs_icwalk(mp, XFS_ICWALK_DQRELE, &icw);
 }
+#else
+# define xfs_dqrele_igrab(ip)		(false)
+# define xfs_dqrele_inode(ip, priv)	((void)0)
+#endif /* CONFIG_XFS_QUOTA */
 
 /*
  * Grab the inode for reclaim exclusively.
@@ -935,8 +926,9 @@ xfs_inode_walk(
  * Return true if we grabbed it, false otherwise.
  */
 static bool
-xfs_reclaim_inode_grab(
-	struct xfs_inode	*ip)
+xfs_reclaim_igrab(
+	struct xfs_inode	*ip,
+	struct xfs_icwalk	*icw)
 {
 	ASSERT(rcu_read_lock_held());
 
@@ -947,6 +939,14 @@ xfs_reclaim_inode_grab(
 		spin_unlock(&ip->i_flags_lock);
 		return false;
 	}
+
+	/* Don't reclaim a sick inode unless the caller asked for it. */
+	if (ip->i_sick &&
+	    (!icw || !(icw->icw_flags & XFS_ICWALK_FLAG_RECLAIM_SICK))) {
+		spin_unlock(&ip->i_flags_lock);
+		return false;
+	}
+
 	__xfs_iflags_set(ip, XFS_IRECLAIM);
 	spin_unlock(&ip->i_flags_lock);
 	return true;
@@ -1002,6 +1002,8 @@ reclaim:
 	spin_lock(&ip->i_flags_lock);
 	ip->i_flags = XFS_IRECLAIM;
 	ip->i_ino = 0;
+	ip->i_sick = 0;
+	ip->i_checked = 0;
 	spin_unlock(&ip->i_flags_lock);
 
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
@@ -1018,7 +1020,7 @@ reclaim:
 	if (!radix_tree_delete(&pag->pag_ici_root,
 				XFS_INO_TO_AGINO(ip->i_mount, ino)))
 		ASSERT(0);
-	xfs_perag_clear_reclaim_tag(pag);
+	xfs_perag_clear_inode_tag(pag, NULLAGINO, XFS_ICI_RECLAIM_TAG);
 	spin_unlock(&pag->pag_ici_lock);
 
 	/*
@@ -1030,7 +1032,7 @@ reclaim:
 	 * unlocked after the lookup before we go ahead and free it.
 	 */
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	xfs_qm_dqdetach(ip);
+	ASSERT(!ip->i_udquot && !ip->i_gdquot && !ip->i_pdquot);
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	ASSERT(xfs_inode_clean(ip));
 
@@ -1045,108 +1047,30 @@ out:
 	xfs_iflags_clear(ip, XFS_IRECLAIM);
 }
 
-/*
- * Walk the AGs and reclaim the inodes in them. Even if the filesystem is
- * corrupted, we still want to try to reclaim all the inodes. If we don't,
- * then a shut down during filesystem unmount reclaim walk leak all the
- * unreclaimed inodes.
- *
- * Returns non-zero if any AGs or inodes were skipped in the reclaim pass
- * so that callers that want to block until all dirty inodes are written back
- * and reclaimed can sanely loop.
- */
-static void
-xfs_reclaim_inodes_ag(
-	struct xfs_mount	*mp,
-	int			*nr_to_scan)
+/* Reclaim sick inodes if we're unmounting or the fs went down. */
+static inline bool
+xfs_want_reclaim_sick(
+	struct xfs_mount	*mp)
 {
-	struct xfs_perag	*pag;
-	xfs_agnumber_t		ag = 0;
-
-	while ((pag = xfs_perag_get_tag(mp, ag, XFS_ICI_RECLAIM_TAG))) {
-		unsigned long	first_index = 0;
-		int		done = 0;
-		int		nr_found = 0;
-
-		ag = pag->pag_agno + 1;
-
-		first_index = READ_ONCE(pag->pag_ici_reclaim_cursor);
-		do {
-			struct xfs_inode *batch[XFS_LOOKUP_BATCH];
-			int	i;
-
-			rcu_read_lock();
-			nr_found = radix_tree_gang_lookup_tag(
-					&pag->pag_ici_root,
-					(void **)batch, first_index,
-					XFS_LOOKUP_BATCH,
-					XFS_ICI_RECLAIM_TAG);
-			if (!nr_found) {
-				done = 1;
-				rcu_read_unlock();
-				break;
-			}
-
-			/*
-			 * Grab the inodes before we drop the lock. if we found
-			 * nothing, nr == 0 and the loop will be skipped.
-			 */
-			for (i = 0; i < nr_found; i++) {
-				struct xfs_inode *ip = batch[i];
-
-				if (done || !xfs_reclaim_inode_grab(ip))
-					batch[i] = NULL;
-
-				/*
-				 * Update the index for the next lookup. Catch
-				 * overflows into the next AG range which can
-				 * occur if we have inodes in the last block of
-				 * the AG and we are currently pointing to the
-				 * last inode.
-				 *
-				 * Because we may see inodes that are from the
-				 * wrong AG due to RCU freeing and
-				 * reallocation, only update the index if it
-				 * lies in this AG. It was a race that lead us
-				 * to see this inode, so another lookup from
-				 * the same index will not find it again.
-				 */
-				if (XFS_INO_TO_AGNO(mp, ip->i_ino) !=
-								pag->pag_agno)
-					continue;
-				first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
-				if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
-					done = 1;
-			}
-
-			/* unlock now we've grabbed the inodes. */
-			rcu_read_unlock();
-
-			for (i = 0; i < nr_found; i++) {
-				if (batch[i])
-					xfs_reclaim_inode(batch[i], pag);
-			}
-
-			*nr_to_scan -= XFS_LOOKUP_BATCH;
-			cond_resched();
-		} while (nr_found && !done && *nr_to_scan > 0);
-
-		if (done)
-			first_index = 0;
-		WRITE_ONCE(pag->pag_ici_reclaim_cursor, first_index);
-		xfs_perag_put(pag);
-	}
+	return (mp->m_flags & XFS_MOUNT_UNMOUNTING) ||
+	       (mp->m_flags & XFS_MOUNT_NORECOVERY) ||
+	       XFS_FORCED_SHUTDOWN(mp);
 }
 
 void
 xfs_reclaim_inodes(
 	struct xfs_mount	*mp)
 {
-	int		nr_to_scan = INT_MAX;
+	struct xfs_icwalk	icw = {
+		.icw_flags	= 0,
+	};
+
+	if (xfs_want_reclaim_sick(mp))
+		icw.icw_flags |= XFS_ICWALK_FLAG_RECLAIM_SICK;
 
 	while (radix_tree_tagged(&mp->m_perag_tree, XFS_ICI_RECLAIM_TAG)) {
 		xfs_ail_push_all_sync(mp->m_ail);
-		xfs_reclaim_inodes_ag(mp, &nr_to_scan);
+		xfs_icwalk(mp, XFS_ICWALK_RECLAIM, &icw);
 	}
 }
 
@@ -1160,13 +1084,21 @@ xfs_reclaim_inodes(
 long
 xfs_reclaim_inodes_nr(
 	struct xfs_mount	*mp,
-	int			nr_to_scan)
+	unsigned long		nr_to_scan)
 {
+	struct xfs_icwalk	icw = {
+		.icw_flags	= XFS_ICWALK_FLAG_SCAN_LIMIT,
+		.icw_scan_limit	= min_t(unsigned long, LONG_MAX, nr_to_scan),
+	};
+
+	if (xfs_want_reclaim_sick(mp))
+		icw.icw_flags |= XFS_ICWALK_FLAG_RECLAIM_SICK;
+
 	/* kick background reclaimer and push the AIL */
 	xfs_reclaim_work_queue(mp);
 	xfs_ail_push_all(mp->m_ail);
 
-	xfs_reclaim_inodes_ag(mp, &nr_to_scan);
+	xfs_icwalk(mp, XFS_ICWALK_RECLAIM, &icw);
 	return 0;
 }
 
@@ -1174,13 +1106,13 @@ xfs_reclaim_inodes_nr(
  * Return the number of reclaimable inodes in the filesystem for
  * the shrinker to determine how much to reclaim.
  */
-int
+long
 xfs_reclaim_inodes_count(
 	struct xfs_mount	*mp)
 {
 	struct xfs_perag	*pag;
 	xfs_agnumber_t		ag = 0;
-	int			reclaimable = 0;
+	long			reclaimable = 0;
 
 	while ((pag = xfs_perag_get_tag(mp, ag, XFS_ICI_RECLAIM_TAG))) {
 		ag = pag->pag_agno + 1;
@@ -1191,20 +1123,20 @@ xfs_reclaim_inodes_count(
 }
 
 STATIC bool
-xfs_inode_match_id(
+xfs_icwalk_match_id(
 	struct xfs_inode	*ip,
-	struct xfs_eofblocks	*eofb)
+	struct xfs_icwalk	*icw)
 {
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_UID) &&
-	    !uid_eq(VFS_I(ip)->i_uid, eofb->eof_uid))
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_UID) &&
+	    !uid_eq(VFS_I(ip)->i_uid, icw->icw_uid))
 		return false;
 
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_GID) &&
-	    !gid_eq(VFS_I(ip)->i_gid, eofb->eof_gid))
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_GID) &&
+	    !gid_eq(VFS_I(ip)->i_gid, icw->icw_gid))
 		return false;
 
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_PRID) &&
-	    ip->i_projid != eofb->eof_prid)
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_PRID) &&
+	    ip->i_projid != icw->icw_prid)
 		return false;
 
 	return true;
@@ -1215,20 +1147,20 @@ xfs_inode_match_id(
  * criteria match. This is for global/internal scans only.
  */
 STATIC bool
-xfs_inode_match_id_union(
+xfs_icwalk_match_id_union(
 	struct xfs_inode	*ip,
-	struct xfs_eofblocks	*eofb)
+	struct xfs_icwalk	*icw)
 {
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_UID) &&
-	    uid_eq(VFS_I(ip)->i_uid, eofb->eof_uid))
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_UID) &&
+	    uid_eq(VFS_I(ip)->i_uid, icw->icw_uid))
 		return true;
 
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_GID) &&
-	    gid_eq(VFS_I(ip)->i_gid, eofb->eof_gid))
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_GID) &&
+	    gid_eq(VFS_I(ip)->i_gid, icw->icw_gid))
 		return true;
 
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_PRID) &&
-	    ip->i_projid == eofb->eof_prid)
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_PRID) &&
+	    ip->i_projid == icw->icw_prid)
 		return true;
 
 	return false;
@@ -1236,29 +1168,29 @@ xfs_inode_match_id_union(
 
 /*
  * Is this inode @ip eligible for eof/cow block reclamation, given some
- * filtering parameters @eofb?  The inode is eligible if @eofb is null or
+ * filtering parameters @icw?  The inode is eligible if @icw is null or
  * if the predicate functions match.
  */
 static bool
-xfs_inode_matches_eofb(
+xfs_icwalk_match(
 	struct xfs_inode	*ip,
-	struct xfs_eofblocks	*eofb)
+	struct xfs_icwalk	*icw)
 {
 	bool			match;
 
-	if (!eofb)
+	if (!icw)
 		return true;
 
-	if (eofb->eof_flags & XFS_EOF_FLAGS_UNION)
-		match = xfs_inode_match_id_union(ip, eofb);
+	if (icw->icw_flags & XFS_ICWALK_FLAG_UNION)
+		match = xfs_icwalk_match_id_union(ip, icw);
 	else
-		match = xfs_inode_match_id(ip, eofb);
+		match = xfs_icwalk_match_id(ip, icw);
 	if (!match)
 		return false;
 
 	/* skip the inode if the file size is too small */
-	if ((eofb->eof_flags & XFS_EOF_FLAGS_MINFILESIZE) &&
-	    XFS_ISIZE(ip) < eofb->eof_min_file_size)
+	if ((icw->icw_flags & XFS_ICWALK_FLAG_MINFILESIZE) &&
+	    XFS_ISIZE(ip) < icw->icw_min_file_size)
 		return false;
 
 	return true;
@@ -1276,22 +1208,20 @@ xfs_reclaim_worker(
 {
 	struct xfs_mount *mp = container_of(to_delayed_work(work),
 					struct xfs_mount, m_reclaim_work);
-	int		nr_to_scan = INT_MAX;
 
-	xfs_reclaim_inodes_ag(mp, &nr_to_scan);
+	xfs_icwalk(mp, XFS_ICWALK_RECLAIM, NULL);
 	xfs_reclaim_work_queue(mp);
 }
 
 STATIC int
 xfs_inode_free_eofblocks(
 	struct xfs_inode	*ip,
-	void			*args,
+	struct xfs_icwalk	*icw,
 	unsigned int		*lockflags)
 {
-	struct xfs_eofblocks	*eofb = args;
 	bool			wait;
 
-	wait = eofb && (eofb->eof_flags & XFS_EOF_FLAGS_SYNC);
+	wait = icw && (icw->icw_flags & XFS_ICWALK_FLAG_SYNC);
 
 	if (!xfs_iflags_test(ip, XFS_IEOFBLOCKS))
 		return 0;
@@ -1303,7 +1233,7 @@ xfs_inode_free_eofblocks(
 	if (!wait && mapping_tagged(VFS_I(ip)->i_mapping, PAGECACHE_TAG_DIRTY))
 		return 0;
 
-	if (!xfs_inode_matches_eofb(ip, eofb))
+	if (!xfs_icwalk_match(ip, icw))
 		return 0;
 
 	/*
@@ -1326,22 +1256,6 @@ xfs_inode_free_eofblocks(
 	return 0;
 }
 
-/*
- * Background scanning to trim preallocated space. This is queued based on the
- * 'speculative_prealloc_lifetime' tunable (5m by default).
- */
-static inline void
-xfs_blockgc_queue(
-	struct xfs_perag	*pag)
-{
-	rcu_read_lock();
-	if (radix_tree_tagged(&pag->pag_ici_root, XFS_ICI_BLOCKGC_TAG))
-		queue_delayed_work(pag->pag_mount->m_gc_workqueue,
-				   &pag->pag_blockgc_work,
-				   msecs_to_jiffies(xfs_blockgc_secs * 1000));
-	rcu_read_unlock();
-}
-
 static void
 xfs_blockgc_set_iflag(
 	struct xfs_inode	*ip,
@@ -1349,7 +1263,6 @@ xfs_blockgc_set_iflag(
 {
 	struct xfs_mount	*mp = ip->i_mount;
 	struct xfs_perag	*pag;
-	int			tagged;
 
 	ASSERT((iflag & ~(XFS_IEOFBLOCKS | XFS_ICOWBLOCKS)) == 0);
 
@@ -1366,24 +1279,8 @@ xfs_blockgc_set_iflag(
 	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
 	spin_lock(&pag->pag_ici_lock);
 
-	tagged = radix_tree_tagged(&pag->pag_ici_root, XFS_ICI_BLOCKGC_TAG);
-	radix_tree_tag_set(&pag->pag_ici_root,
-			   XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino),
-			   XFS_ICI_BLOCKGC_TAG);
-	if (!tagged) {
-		/* propagate the blockgc tag up into the perag radix tree */
-		spin_lock(&ip->i_mount->m_perag_lock);
-		radix_tree_tag_set(&ip->i_mount->m_perag_tree,
-				   XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino),
-				   XFS_ICI_BLOCKGC_TAG);
-		spin_unlock(&ip->i_mount->m_perag_lock);
-
-		/* kick off background trimming */
-		xfs_blockgc_queue(pag);
-
-		trace_xfs_perag_set_blockgc(ip->i_mount, pag->pag_agno, -1,
-				_RET_IP_);
-	}
+	xfs_perag_set_inode_tag(pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
+			XFS_ICI_BLOCKGC_TAG);
 
 	spin_unlock(&pag->pag_ici_lock);
 	xfs_perag_put(pag);
@@ -1419,19 +1316,8 @@ xfs_blockgc_clear_iflag(
 	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
 	spin_lock(&pag->pag_ici_lock);
 
-	radix_tree_tag_clear(&pag->pag_ici_root,
-			     XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino),
-			     XFS_ICI_BLOCKGC_TAG);
-	if (!radix_tree_tagged(&pag->pag_ici_root, XFS_ICI_BLOCKGC_TAG)) {
-		/* clear the blockgc tag from the perag radix tree */
-		spin_lock(&ip->i_mount->m_perag_lock);
-		radix_tree_tag_clear(&ip->i_mount->m_perag_tree,
-				     XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino),
-				     XFS_ICI_BLOCKGC_TAG);
-		spin_unlock(&ip->i_mount->m_perag_lock);
-		trace_xfs_perag_clear_blockgc(ip->i_mount, pag->pag_agno, -1,
-				_RET_IP_);
-	}
+	xfs_perag_clear_inode_tag(pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
+			XFS_ICI_BLOCKGC_TAG);
 
 	spin_unlock(&pag->pag_ici_lock);
 	xfs_perag_put(pag);
@@ -1492,14 +1378,13 @@ xfs_prep_free_cowblocks(
 STATIC int
 xfs_inode_free_cowblocks(
 	struct xfs_inode	*ip,
-	void			*args,
+	struct xfs_icwalk	*icw,
 	unsigned int		*lockflags)
 {
-	struct xfs_eofblocks	*eofb = args;
 	bool			wait;
 	int			ret = 0;
 
-	wait = eofb && (eofb->eof_flags & XFS_EOF_FLAGS_SYNC);
+	wait = icw && (icw->icw_flags & XFS_ICWALK_FLAG_SYNC);
 
 	if (!xfs_iflags_test(ip, XFS_ICOWBLOCKS))
 		return 0;
@@ -1507,7 +1392,7 @@ xfs_inode_free_cowblocks(
 	if (!xfs_prep_free_cowblocks(ip))
 		return 0;
 
-	if (!xfs_inode_matches_eofb(ip, eofb))
+	if (!xfs_icwalk_match(ip, icw))
 		return 0;
 
 	/*
@@ -1554,14 +1439,6 @@ xfs_inode_clear_cowblocks_tag(
 	return xfs_blockgc_clear_iflag(ip, XFS_ICOWBLOCKS);
 }
 
-#define for_each_perag_tag(mp, next_agno, pag, tag) \
-	for ((next_agno) = 0, (pag) = xfs_perag_get_tag((mp), 0, (tag)); \
-		(pag) != NULL; \
-		(next_agno) = (pag)->pag_agno + 1, \
-		xfs_perag_put(pag), \
-		(pag) = xfs_perag_get_tag((mp), (next_agno), (tag)))
-
-
 /* Disable post-EOF and CoW block auto-reclamation. */
 void
 xfs_blockgc_stop(
@@ -1586,23 +1463,66 @@ xfs_blockgc_start(
 		xfs_blockgc_queue(pag);
 }
 
+/* Don't try to run block gc on an inode that's in any of these states. */
+#define XFS_BLOCKGC_NOGRAB_IFLAGS	(XFS_INEW | \
+					 XFS_IRECLAIMABLE | \
+					 XFS_IRECLAIM)
+/*
+ * Decide if the given @ip is eligible for garbage collection of speculative
+ * preallocations, and grab it if so.  Returns true if it's ready to go or
+ * false if we should just ignore it.
+ */
+static bool
+xfs_blockgc_igrab(
+	struct xfs_inode	*ip)
+{
+	struct inode		*inode = VFS_I(ip);
+
+	ASSERT(rcu_read_lock_held());
+
+	/* Check for stale RCU freed inode */
+	spin_lock(&ip->i_flags_lock);
+	if (!ip->i_ino)
+		goto out_unlock_noent;
+
+	if (ip->i_flags & XFS_BLOCKGC_NOGRAB_IFLAGS)
+		goto out_unlock_noent;
+	spin_unlock(&ip->i_flags_lock);
+
+	/* nothing to sync during shutdown */
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return false;
+
+	/* If we can't grab the inode, it must on it's way to reclaim. */
+	if (!igrab(inode))
+		return false;
+
+	/* inode is valid */
+	return true;
+
+out_unlock_noent:
+	spin_unlock(&ip->i_flags_lock);
+	return false;
+}
+
 /* Scan one incore inode for block preallocations that we can remove. */
 static int
 xfs_blockgc_scan_inode(
 	struct xfs_inode	*ip,
-	void			*args)
+	struct xfs_icwalk	*icw)
 {
 	unsigned int		lockflags = 0;
 	int			error;
 
-	error = xfs_inode_free_eofblocks(ip, args, &lockflags);
+	error = xfs_inode_free_eofblocks(ip, icw, &lockflags);
 	if (error)
 		goto unlock;
 
-	error = xfs_inode_free_cowblocks(ip, args, &lockflags);
+	error = xfs_inode_free_cowblocks(ip, icw, &lockflags);
 unlock:
 	if (lockflags)
 		xfs_iunlock(ip, lockflags);
+	xfs_irele(ip);
 	return error;
 }
 
@@ -1618,8 +1538,7 @@ xfs_blockgc_worker(
 
 	if (!sb_start_write_trylock(mp->m_super))
 		return;
-	error = xfs_inode_walk_ag(pag, 0, xfs_blockgc_scan_inode, NULL,
-			XFS_ICI_BLOCKGC_TAG);
+	error = xfs_icwalk_ag(pag, XFS_ICWALK_BLOCKGC, NULL);
 	if (error)
 		xfs_info(mp, "AG %u preallocation gc worker failed, err=%d",
 				pag->pag_agno, error);
@@ -1633,12 +1552,11 @@ xfs_blockgc_worker(
 int
 xfs_blockgc_free_space(
 	struct xfs_mount	*mp,
-	struct xfs_eofblocks	*eofb)
+	struct xfs_icwalk	*icw)
 {
-	trace_xfs_blockgc_free_space(mp, eofb, _RET_IP_);
+	trace_xfs_blockgc_free_space(mp, icw, _RET_IP_);
 
-	return xfs_inode_walk(mp, 0, xfs_blockgc_scan_inode, eofb,
-			XFS_ICI_BLOCKGC_TAG);
+	return xfs_icwalk(mp, XFS_ICWALK_BLOCKGC, icw);
 }
 
 /*
@@ -1648,7 +1566,7 @@ xfs_blockgc_free_space(
  * scan.
  *
  * Callers must not hold any inode's ILOCK.  If requesting a synchronous scan
- * (XFS_EOF_FLAGS_SYNC), the caller also must not hold any inode's IOLOCK or
+ * (XFS_ICWALK_FLAG_SYNC), the caller also must not hold any inode's IOLOCK or
  * MMAPLOCK.
  */
 int
@@ -1657,9 +1575,9 @@ xfs_blockgc_free_dquots(
 	struct xfs_dquot	*udqp,
 	struct xfs_dquot	*gdqp,
 	struct xfs_dquot	*pdqp,
-	unsigned int		eof_flags)
+	unsigned int		iwalk_flags)
 {
-	struct xfs_eofblocks	eofb = {0};
+	struct xfs_icwalk	icw = {0};
 	bool			do_work = false;
 
 	if (!udqp && !gdqp && !pdqp)
@@ -1669,40 +1587,260 @@ xfs_blockgc_free_dquots(
 	 * Run a scan to free blocks using the union filter to cover all
 	 * applicable quotas in a single scan.
 	 */
-	eofb.eof_flags = XFS_EOF_FLAGS_UNION | eof_flags;
+	icw.icw_flags = XFS_ICWALK_FLAG_UNION | iwalk_flags;
 
 	if (XFS_IS_UQUOTA_ENFORCED(mp) && udqp && xfs_dquot_lowsp(udqp)) {
-		eofb.eof_uid = make_kuid(mp->m_super->s_user_ns, udqp->q_id);
-		eofb.eof_flags |= XFS_EOF_FLAGS_UID;
+		icw.icw_uid = make_kuid(mp->m_super->s_user_ns, udqp->q_id);
+		icw.icw_flags |= XFS_ICWALK_FLAG_UID;
 		do_work = true;
 	}
 
 	if (XFS_IS_UQUOTA_ENFORCED(mp) && gdqp && xfs_dquot_lowsp(gdqp)) {
-		eofb.eof_gid = make_kgid(mp->m_super->s_user_ns, gdqp->q_id);
-		eofb.eof_flags |= XFS_EOF_FLAGS_GID;
+		icw.icw_gid = make_kgid(mp->m_super->s_user_ns, gdqp->q_id);
+		icw.icw_flags |= XFS_ICWALK_FLAG_GID;
 		do_work = true;
 	}
 
 	if (XFS_IS_PQUOTA_ENFORCED(mp) && pdqp && xfs_dquot_lowsp(pdqp)) {
-		eofb.eof_prid = pdqp->q_id;
-		eofb.eof_flags |= XFS_EOF_FLAGS_PRID;
+		icw.icw_prid = pdqp->q_id;
+		icw.icw_flags |= XFS_ICWALK_FLAG_PRID;
 		do_work = true;
 	}
 
 	if (!do_work)
 		return 0;
 
-	return xfs_blockgc_free_space(mp, &eofb);
+	return xfs_blockgc_free_space(mp, &icw);
 }
 
 /* Run cow/eofblocks scans on the quotas attached to the inode. */
 int
 xfs_blockgc_free_quota(
 	struct xfs_inode	*ip,
-	unsigned int		eof_flags)
+	unsigned int		iwalk_flags)
 {
 	return xfs_blockgc_free_dquots(ip->i_mount,
 			xfs_inode_dquot(ip, XFS_DQTYPE_USER),
 			xfs_inode_dquot(ip, XFS_DQTYPE_GROUP),
-			xfs_inode_dquot(ip, XFS_DQTYPE_PROJ), eof_flags);
+			xfs_inode_dquot(ip, XFS_DQTYPE_PROJ), iwalk_flags);
+}
+
+/* XFS Inode Cache Walking Code */
+
+/*
+ * The inode lookup is done in batches to keep the amount of lock traffic and
+ * radix tree lookups to a minimum. The batch size is a trade off between
+ * lookup reduction and stack usage. This is in the reclaim path, so we can't
+ * be too greedy.
+ */
+#define XFS_LOOKUP_BATCH	32
+
+
+/*
+ * Decide if we want to grab this inode in anticipation of doing work towards
+ * the goal.
+ */
+static inline bool
+xfs_icwalk_igrab(
+	enum xfs_icwalk_goal	goal,
+	struct xfs_inode	*ip,
+	struct xfs_icwalk	*icw)
+{
+	switch (goal) {
+	case XFS_ICWALK_DQRELE:
+		return xfs_dqrele_igrab(ip);
+	case XFS_ICWALK_BLOCKGC:
+		return xfs_blockgc_igrab(ip);
+	case XFS_ICWALK_RECLAIM:
+		return xfs_reclaim_igrab(ip, icw);
+	default:
+		return false;
+	}
+}
+
+/*
+ * Process an inode.  Each processing function must handle any state changes
+ * made by the icwalk igrab function.  Return -EAGAIN to skip an inode.
+ */
+static inline int
+xfs_icwalk_process_inode(
+	enum xfs_icwalk_goal	goal,
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	struct xfs_icwalk	*icw)
+{
+	int			error = 0;
+
+	switch (goal) {
+	case XFS_ICWALK_DQRELE:
+		xfs_dqrele_inode(ip, icw);
+		break;
+	case XFS_ICWALK_BLOCKGC:
+		error = xfs_blockgc_scan_inode(ip, icw);
+		break;
+	case XFS_ICWALK_RECLAIM:
+		xfs_reclaim_inode(ip, pag);
+		break;
+	}
+	return error;
+}
+
+/*
+ * For a given per-AG structure @pag and a goal, grab qualifying inodes and
+ * process them in some manner.
+ */
+static int
+xfs_icwalk_ag(
+	struct xfs_perag	*pag,
+	enum xfs_icwalk_goal	goal,
+	struct xfs_icwalk	*icw)
+{
+	struct xfs_mount	*mp = pag->pag_mount;
+	uint32_t		first_index;
+	int			last_error = 0;
+	int			skipped;
+	bool			done;
+	int			nr_found;
+
+restart:
+	done = false;
+	skipped = 0;
+	if (goal == XFS_ICWALK_RECLAIM)
+		first_index = READ_ONCE(pag->pag_ici_reclaim_cursor);
+	else
+		first_index = 0;
+	nr_found = 0;
+	do {
+		struct xfs_inode *batch[XFS_LOOKUP_BATCH];
+		unsigned int	tag = xfs_icwalk_tag(goal);
+		int		error = 0;
+		int		i;
+
+		rcu_read_lock();
+
+		if (tag == XFS_ICWALK_NULL_TAG)
+			nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
+					(void **)batch, first_index,
+					XFS_LOOKUP_BATCH);
+		else
+			nr_found = radix_tree_gang_lookup_tag(
+					&pag->pag_ici_root,
+					(void **) batch, first_index,
+					XFS_LOOKUP_BATCH, tag);
+
+		if (!nr_found) {
+			done = true;
+			rcu_read_unlock();
+			break;
+		}
+
+		/*
+		 * Grab the inodes before we drop the lock. if we found
+		 * nothing, nr == 0 and the loop will be skipped.
+		 */
+		for (i = 0; i < nr_found; i++) {
+			struct xfs_inode *ip = batch[i];
+
+			if (done || !xfs_icwalk_igrab(goal, ip, icw))
+				batch[i] = NULL;
+
+			/*
+			 * Update the index for the next lookup. Catch
+			 * overflows into the next AG range which can occur if
+			 * we have inodes in the last block of the AG and we
+			 * are currently pointing to the last inode.
+			 *
+			 * Because we may see inodes that are from the wrong AG
+			 * due to RCU freeing and reallocation, only update the
+			 * index if it lies in this AG. It was a race that lead
+			 * us to see this inode, so another lookup from the
+			 * same index will not find it again.
+			 */
+			if (XFS_INO_TO_AGNO(mp, ip->i_ino) != pag->pag_agno)
+				continue;
+			first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+			if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
+				done = true;
+		}
+
+		/* unlock now we've grabbed the inodes. */
+		rcu_read_unlock();
+
+		for (i = 0; i < nr_found; i++) {
+			if (!batch[i])
+				continue;
+			error = xfs_icwalk_process_inode(goal, batch[i], pag,
+					icw);
+			if (error == -EAGAIN) {
+				skipped++;
+				continue;
+			}
+			if (error && last_error != -EFSCORRUPTED)
+				last_error = error;
+		}
+
+		/* bail out if the filesystem is corrupted.  */
+		if (error == -EFSCORRUPTED)
+			break;
+
+		cond_resched();
+
+		if (icw && (icw->icw_flags & XFS_ICWALK_FLAG_SCAN_LIMIT)) {
+			icw->icw_scan_limit -= XFS_LOOKUP_BATCH;
+			if (icw->icw_scan_limit <= 0)
+				break;
+		}
+	} while (nr_found && !done);
+
+	if (goal == XFS_ICWALK_RECLAIM) {
+		if (done)
+			first_index = 0;
+		WRITE_ONCE(pag->pag_ici_reclaim_cursor, first_index);
+	}
+
+	if (skipped) {
+		delay(1);
+		goto restart;
+	}
+	return last_error;
+}
+
+/* Fetch the next (possibly tagged) per-AG structure. */
+static inline struct xfs_perag *
+xfs_icwalk_get_perag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	enum xfs_icwalk_goal	goal)
+{
+	unsigned int		tag = xfs_icwalk_tag(goal);
+
+	if (tag == XFS_ICWALK_NULL_TAG)
+		return xfs_perag_get(mp, agno);
+	return xfs_perag_get_tag(mp, agno, tag);
+}
+
+/* Walk all incore inodes to achieve a given goal. */
+static int
+xfs_icwalk(
+	struct xfs_mount	*mp,
+	enum xfs_icwalk_goal	goal,
+	struct xfs_icwalk	*icw)
+{
+	struct xfs_perag	*pag;
+	int			error = 0;
+	int			last_error = 0;
+	xfs_agnumber_t		agno = 0;
+
+	while ((pag = xfs_icwalk_get_perag(mp, agno, goal))) {
+		agno = pag->pag_agno + 1;
+		error = xfs_icwalk_ag(pag, goal, icw);
+		xfs_perag_put(pag);
+		if (error) {
+			last_error = error;
+			if (error == -EFSCORRUPTED)
+				break;
+		}
+	}
+	return last_error;
+	BUILD_BUG_ON(XFS_ICWALK_PRIVATE_FLAGS & XFS_ICWALK_FLAGS_VALID);
 }
diff --git a/fs/xfs/xfs_icache.h b/fs/xfs/xfs_icache.h
index d1fddb152420..c751cc32dc46 100644
--- a/fs/xfs/xfs_icache.h
+++ b/fs/xfs/xfs_icache.h
@@ -9,22 +9,27 @@
 struct xfs_mount;
 struct xfs_perag;
 
-struct xfs_eofblocks {
-	__u32		eof_flags;
-	kuid_t		eof_uid;
-	kgid_t		eof_gid;
-	prid_t		eof_prid;
-	__u64		eof_min_file_size;
+struct xfs_icwalk {
+	__u32		icw_flags;
+	kuid_t		icw_uid;
+	kgid_t		icw_gid;
+	prid_t		icw_prid;
+	__u64		icw_min_file_size;
+	long		icw_scan_limit;
 };
 
-/*
- * tags for inode radix tree
- */
-#define XFS_ICI_NO_TAG		(-1)	/* special flag for an untagged lookup
-					   in xfs_inode_walk */
-#define XFS_ICI_RECLAIM_TAG	0	/* inode is to be reclaimed */
-/* Inode has speculative preallocations (posteof or cow) to clean. */
-#define XFS_ICI_BLOCKGC_TAG	1
+/* Flags that reflect xfs_fs_eofblocks functionality. */
+#define XFS_ICWALK_FLAG_SYNC		(1U << 0) /* sync/wait mode scan */
+#define XFS_ICWALK_FLAG_UID		(1U << 1) /* filter by uid */
+#define XFS_ICWALK_FLAG_GID		(1U << 2) /* filter by gid */
+#define XFS_ICWALK_FLAG_PRID		(1U << 3) /* filter by project id */
+#define XFS_ICWALK_FLAG_MINFILESIZE	(1U << 4) /* filter by min file size */
+
+#define XFS_ICWALK_FLAGS_VALID		(XFS_ICWALK_FLAG_SYNC | \
+					 XFS_ICWALK_FLAG_UID | \
+					 XFS_ICWALK_FLAG_GID | \
+					 XFS_ICWALK_FLAG_PRID | \
+					 XFS_ICWALK_FLAG_MINFILESIZE)
 
 /*
  * Flags for xfs_iget()
@@ -34,11 +39,6 @@ struct xfs_eofblocks {
 #define XFS_IGET_DONTCACHE	0x4
 #define XFS_IGET_INCORE		0x8	/* don't read from disk or reinit */
 
-/*
- * flags for AG inode iterator
- */
-#define XFS_INODE_WALK_INEW_WAIT	0x1	/* wait on new inodes */
-
 int xfs_iget(struct xfs_mount *mp, struct xfs_trans *tp, xfs_ino_t ino,
 	     uint flags, uint lock_flags, xfs_inode_t **ipp);
 
@@ -49,16 +49,16 @@ void xfs_inode_free(struct xfs_inode *ip);
 void xfs_reclaim_worker(struct work_struct *work);
 
 void xfs_reclaim_inodes(struct xfs_mount *mp);
-int xfs_reclaim_inodes_count(struct xfs_mount *mp);
-long xfs_reclaim_inodes_nr(struct xfs_mount *mp, int nr_to_scan);
+long xfs_reclaim_inodes_count(struct xfs_mount *mp);
+long xfs_reclaim_inodes_nr(struct xfs_mount *mp, unsigned long nr_to_scan);
 
-void xfs_inode_set_reclaim_tag(struct xfs_inode *ip);
+void xfs_inode_mark_reclaimable(struct xfs_inode *ip);
 
 int xfs_blockgc_free_dquots(struct xfs_mount *mp, struct xfs_dquot *udqp,
 		struct xfs_dquot *gdqp, struct xfs_dquot *pdqp,
-		unsigned int eof_flags);
-int xfs_blockgc_free_quota(struct xfs_inode *ip, unsigned int eof_flags);
-int xfs_blockgc_free_space(struct xfs_mount *mp, struct xfs_eofblocks *eofb);
+		unsigned int iwalk_flags);
+int xfs_blockgc_free_quota(struct xfs_inode *ip, unsigned int iwalk_flags);
+int xfs_blockgc_free_space(struct xfs_mount *mp, struct xfs_icwalk *icm);
 
 void xfs_inode_set_eofblocks_tag(struct xfs_inode *ip);
 void xfs_inode_clear_eofblocks_tag(struct xfs_inode *ip);
@@ -68,9 +68,11 @@ void xfs_inode_clear_cowblocks_tag(struct xfs_inode *ip);
 
 void xfs_blockgc_worker(struct work_struct *work);
 
-int xfs_inode_walk(struct xfs_mount *mp, int iter_flags,
-	int (*execute)(struct xfs_inode *ip, void *args),
-	void *args, int tag);
+#ifdef CONFIG_XFS_QUOTA
+int xfs_dqrele_all_inodes(struct xfs_mount *mp, unsigned int qflags);
+#else
+# define xfs_dqrele_all_inodes(mp, qflags)	(0)
+#endif
 
 int xfs_icache_inode_is_allocated(struct xfs_mount *mp, struct xfs_trans *tp,
 				  xfs_ino_t ino, bool *inuse);
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 1db99a909b23..a835ceb79ba5 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -11,7 +11,6 @@
 #include "xfs_format.h"
 #include "xfs_log_format.h"
 #include "xfs_trans_resv.h"
-#include "xfs_sb.h"
 #include "xfs_mount.h"
 #include "xfs_defer.h"
 #include "xfs_inode.h"
@@ -35,6 +34,7 @@
 #include "xfs_log.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_reflink.h"
+#include "xfs_ag.h"
 
 kmem_zone_t *xfs_inode_zone;
 
@@ -45,7 +45,8 @@ kmem_zone_t *xfs_inode_zone;
 #define	XFS_ITRUNC_MAX_EXTENTS	2
 
 STATIC int xfs_iunlink(struct xfs_trans *, struct xfs_inode *);
-STATIC int xfs_iunlink_remove(struct xfs_trans *, struct xfs_inode *);
+STATIC int xfs_iunlink_remove(struct xfs_trans *tp, struct xfs_perag *pag,
+	struct xfs_inode *);
 
 /*
  * helper function to extract extent size hint from inode
@@ -778,7 +779,7 @@ xfs_inode_inherit_flags2(
  * Initialise a newly allocated inode and return the in-core inode to the
  * caller locked exclusively.
  */
-static int
+int
 xfs_init_new_inode(
 	struct user_namespace	*mnt_userns,
 	struct xfs_trans	*tp,
@@ -915,57 +916,6 @@ xfs_init_new_inode(
 }
 
 /*
- * Allocates a new inode from disk and return a pointer to the incore copy. This
- * routine will internally commit the current transaction and allocate a new one
- * if we needed to allocate more on-disk free inodes to perform the requested
- * operation.
- *
- * If we are allocating quota inodes, we do not have a parent inode to attach to
- * or associate with (i.e. dp == NULL) because they are not linked into the
- * directory structure - they are attached directly to the superblock - and so
- * have no parent.
- */
-int
-xfs_dir_ialloc(
-	struct user_namespace	*mnt_userns,
-	struct xfs_trans	**tpp,
-	struct xfs_inode	*dp,
-	umode_t			mode,
-	xfs_nlink_t		nlink,
-	dev_t			rdev,
-	prid_t			prid,
-	bool			init_xattrs,
-	struct xfs_inode	**ipp)
-{
-	struct xfs_buf		*agibp;
-	xfs_ino_t		parent_ino = dp ? dp->i_ino : 0;
-	xfs_ino_t		ino;
-	int			error;
-
-	ASSERT((*tpp)->t_flags & XFS_TRANS_PERM_LOG_RES);
-
-	/*
-	 * Call the space management code to pick the on-disk inode to be
-	 * allocated.
-	 */
-	error = xfs_dialloc_select_ag(tpp, parent_ino, mode, &agibp);
-	if (error)
-		return error;
-
-	if (!agibp)
-		return -ENOSPC;
-
-	/* Allocate an inode from the selected AG */
-	error = xfs_dialloc_ag(*tpp, agibp, parent_ino, &ino);
-	if (error)
-		return error;
-	ASSERT(ino != NULLFSINO);
-
-	return xfs_init_new_inode(mnt_userns, *tpp, dp, ino, mode, nlink, rdev,
-				  prid, init_xattrs, ipp);
-}
-
-/*
  * Decrement the link count on an inode & log the change.  If this causes the
  * link count to go to zero, move the inode to AGI unlinked list so that it can
  * be freed when the last active reference goes away via xfs_inactive().
@@ -1022,6 +972,7 @@ xfs_create(
 	struct xfs_dquot	*pdqp = NULL;
 	struct xfs_trans_res	*tres;
 	uint			resblks;
+	xfs_ino_t		ino;
 
 	trace_xfs_create(dp, name);
 
@@ -1078,14 +1029,16 @@ xfs_create(
 	 * entry pointing to them, but a directory also the "." entry
 	 * pointing to itself.
 	 */
-	error = xfs_dir_ialloc(mnt_userns, &tp, dp, mode, is_dir ? 2 : 1, rdev,
-			       prid, init_xattrs, &ip);
+	error = xfs_dialloc(&tp, dp->i_ino, mode, &ino);
+	if (!error)
+		error = xfs_init_new_inode(mnt_userns, tp, dp, ino, mode,
+				is_dir ? 2 : 1, rdev, prid, init_xattrs, &ip);
 	if (error)
 		goto out_trans_cancel;
 
 	/*
 	 * Now we join the directory inode to the transaction.  We do not do it
-	 * earlier because xfs_dir_ialloc might commit the previous transaction
+	 * earlier because xfs_dialloc might commit the previous transaction
 	 * (and release all the locks).  An error from here on will result in
 	 * the transaction cancel unlocking dp so don't do it explicitly in the
 	 * error path.
@@ -1175,6 +1128,7 @@ xfs_create_tmpfile(
 	struct xfs_dquot	*pdqp = NULL;
 	struct xfs_trans_res	*tres;
 	uint			resblks;
+	xfs_ino_t		ino;
 
 	if (XFS_FORCED_SHUTDOWN(mp))
 		return -EIO;
@@ -1199,8 +1153,10 @@ xfs_create_tmpfile(
 	if (error)
 		goto out_release_dquots;
 
-	error = xfs_dir_ialloc(mnt_userns, &tp, dp, mode, 0, 0, prid,
-				false, &ip);
+	error = xfs_dialloc(&tp, dp->i_ino, mode, &ino);
+	if (!error)
+		error = xfs_init_new_inode(mnt_userns, tp, dp, ino, mode,
+				0, 0, prid, false, &ip);
 	if (error)
 		goto out_trans_cancel;
 
@@ -1315,7 +1271,11 @@ xfs_link(
 	 * Handle initial link state of O_TMPFILE inode
 	 */
 	if (VFS_I(sip)->i_nlink == 0) {
-		error = xfs_iunlink_remove(tp, sip);
+		struct xfs_perag	*pag;
+
+		pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, sip->i_ino));
+		error = xfs_iunlink_remove(tp, pag, sip);
+		xfs_perag_put(pag);
 		if (error)
 			goto error_return;
 	}
@@ -1716,7 +1676,7 @@ xfs_inactive(
 	 */
 	if (VFS_I(ip)->i_mode == 0) {
 		ASSERT(ip->i_df.if_broot_bytes == 0);
-		return;
+		goto out;
 	}
 
 	mp = ip->i_mount;
@@ -1724,11 +1684,11 @@ xfs_inactive(
 
 	/* If this is a read-only mount, don't do this (would generate I/O) */
 	if (mp->m_flags & XFS_MOUNT_RDONLY)
-		return;
+		goto out;
 
 	/* Metadata inodes require explicit resource cleanup. */
 	if (xfs_is_metadata_inode(ip))
-		return;
+		goto out;
 
 	/* Try to clean out the cow blocks if there are any. */
 	if (xfs_inode_has_cow_data(ip))
@@ -1747,7 +1707,7 @@ xfs_inactive(
 		if (xfs_can_free_eofblocks(ip, true))
 			xfs_free_eofblocks(ip);
 
-		return;
+		goto out;
 	}
 
 	if (S_ISREG(VFS_I(ip)->i_mode) &&
@@ -1757,14 +1717,14 @@ xfs_inactive(
 
 	error = xfs_qm_dqattach(ip);
 	if (error)
-		return;
+		goto out;
 
 	if (S_ISLNK(VFS_I(ip)->i_mode))
 		error = xfs_inactive_symlink(ip);
 	else if (truncate)
 		error = xfs_inactive_truncate(ip);
 	if (error)
-		return;
+		goto out;
 
 	/*
 	 * If there are attributes associated with the file then blow them away
@@ -1774,7 +1734,7 @@ xfs_inactive(
 	if (XFS_IFORK_Q(ip)) {
 		error = xfs_attr_inactive(ip);
 		if (error)
-			return;
+			goto out;
 	}
 
 	ASSERT(!ip->i_afp);
@@ -1783,12 +1743,12 @@ xfs_inactive(
 	/*
 	 * Free the inode.
 	 */
-	error = xfs_inactive_ifree(ip);
-	if (error)
-		return;
+	xfs_inactive_ifree(ip);
 
+out:
 	/*
-	 * Release the dquots held by inode, if any.
+	 * We're done making metadata updates for this inode, so we can release
+	 * the attached dquots.
 	 */
 	xfs_qm_dqdetach(ip);
 }
@@ -2008,7 +1968,7 @@ xfs_iunlink_destroy(
 STATIC int
 xfs_iunlink_update_bucket(
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	struct xfs_buf		*agibp,
 	unsigned int		bucket_index,
 	xfs_agino_t		new_agino)
@@ -2017,10 +1977,10 @@ xfs_iunlink_update_bucket(
 	xfs_agino_t		old_value;
 	int			offset;
 
-	ASSERT(xfs_verify_agino_or_null(tp->t_mountp, agno, new_agino));
+	ASSERT(xfs_verify_agino_or_null(tp->t_mountp, pag->pag_agno, new_agino));
 
 	old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]);
-	trace_xfs_iunlink_update_bucket(tp->t_mountp, agno, bucket_index,
+	trace_xfs_iunlink_update_bucket(tp->t_mountp, pag->pag_agno, bucket_index,
 			old_value, new_agino);
 
 	/*
@@ -2044,7 +2004,7 @@ xfs_iunlink_update_bucket(
 STATIC void
 xfs_iunlink_update_dinode(
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agino_t		agino,
 	struct xfs_buf		*ibp,
 	struct xfs_dinode	*dip,
@@ -2054,9 +2014,9 @@ xfs_iunlink_update_dinode(
 	struct xfs_mount	*mp = tp->t_mountp;
 	int			offset;
 
-	ASSERT(xfs_verify_agino_or_null(mp, agno, next_agino));
+	ASSERT(xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino));
 
-	trace_xfs_iunlink_update_dinode(mp, agno, agino,
+	trace_xfs_iunlink_update_dinode(mp, pag->pag_agno, agino,
 			be32_to_cpu(dip->di_next_unlinked), next_agino);
 
 	dip->di_next_unlinked = cpu_to_be32(next_agino);
@@ -2074,7 +2034,7 @@ STATIC int
 xfs_iunlink_update_inode(
 	struct xfs_trans	*tp,
 	struct xfs_inode	*ip,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agino_t		next_agino,
 	xfs_agino_t		*old_next_agino)
 {
@@ -2084,7 +2044,7 @@ xfs_iunlink_update_inode(
 	xfs_agino_t		old_value;
 	int			error;
 
-	ASSERT(xfs_verify_agino_or_null(mp, agno, next_agino));
+	ASSERT(xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino));
 
 	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &ibp);
 	if (error)
@@ -2093,7 +2053,7 @@ xfs_iunlink_update_inode(
 
 	/* Make sure the old pointer isn't garbage. */
 	old_value = be32_to_cpu(dip->di_next_unlinked);
-	if (!xfs_verify_agino_or_null(mp, agno, old_value)) {
+	if (!xfs_verify_agino_or_null(mp, pag->pag_agno, old_value)) {
 		xfs_inode_verifier_error(ip, -EFSCORRUPTED, __func__, dip,
 				sizeof(*dip), __this_address);
 		error = -EFSCORRUPTED;
@@ -2116,7 +2076,7 @@ xfs_iunlink_update_inode(
 	}
 
 	/* Ok, update the new pointer. */
-	xfs_iunlink_update_dinode(tp, agno, XFS_INO_TO_AGINO(mp, ip->i_ino),
+	xfs_iunlink_update_dinode(tp, pag, XFS_INO_TO_AGINO(mp, ip->i_ino),
 			ibp, dip, &ip->i_imap, next_agino);
 	return 0;
 out:
@@ -2137,10 +2097,10 @@ xfs_iunlink(
 	struct xfs_inode	*ip)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_perag	*pag;
 	struct xfs_agi		*agi;
 	struct xfs_buf		*agibp;
 	xfs_agino_t		next_agino;
-	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
 	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
 	short			bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
 	int			error;
@@ -2149,10 +2109,12 @@ xfs_iunlink(
 	ASSERT(VFS_I(ip)->i_mode != 0);
 	trace_xfs_iunlink(ip);
 
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+
 	/* Get the agi buffer first.  It ensures lock ordering on the list. */
-	error = xfs_read_agi(mp, tp, agno, &agibp);
+	error = xfs_read_agi(mp, tp, pag->pag_agno, &agibp);
 	if (error)
-		return error;
+		goto out;
 	agi = agibp->b_addr;
 
 	/*
@@ -2162,9 +2124,10 @@ xfs_iunlink(
 	 */
 	next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
 	if (next_agino == agino ||
-	    !xfs_verify_agino_or_null(mp, agno, next_agino)) {
+	    !xfs_verify_agino_or_null(mp, pag->pag_agno, next_agino)) {
 		xfs_buf_mark_corrupt(agibp);
-		return -EFSCORRUPTED;
+		error = -EFSCORRUPTED;
+		goto out;
 	}
 
 	if (next_agino != NULLAGINO) {
@@ -2174,23 +2137,26 @@ xfs_iunlink(
 		 * There is already another inode in the bucket, so point this
 		 * inode to the current head of the list.
 		 */
-		error = xfs_iunlink_update_inode(tp, ip, agno, next_agino,
+		error = xfs_iunlink_update_inode(tp, ip, pag, next_agino,
 				&old_agino);
 		if (error)
-			return error;
+			goto out;
 		ASSERT(old_agino == NULLAGINO);
 
 		/*
 		 * agino has been unlinked, add a backref from the next inode
 		 * back to agino.
 		 */
-		error = xfs_iunlink_add_backref(agibp->b_pag, agino, next_agino);
+		error = xfs_iunlink_add_backref(pag, agino, next_agino);
 		if (error)
-			return error;
+			goto out;
 	}
 
 	/* Point the head of the list to point to this inode. */
-	return xfs_iunlink_update_bucket(tp, agno, agibp, bucket_index, agino);
+	error = xfs_iunlink_update_bucket(tp, pag, agibp, bucket_index, agino);
+out:
+	xfs_perag_put(pag);
+	return error;
 }
 
 /* Return the imap, dinode pointer, and buffer for an inode. */
@@ -2238,14 +2204,13 @@ xfs_iunlink_map_ino(
 STATIC int
 xfs_iunlink_map_prev(
 	struct xfs_trans	*tp,
-	xfs_agnumber_t		agno,
+	struct xfs_perag	*pag,
 	xfs_agino_t		head_agino,
 	xfs_agino_t		target_agino,
 	xfs_agino_t		*agino,
 	struct xfs_imap		*imap,
 	struct xfs_dinode	**dipp,
-	struct xfs_buf		**bpp,
-	struct xfs_perag	*pag)
+	struct xfs_buf		**bpp)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
 	xfs_agino_t		next_agino;
@@ -2257,7 +2222,8 @@ xfs_iunlink_map_prev(
 	/* See if our backref cache can find it faster. */
 	*agino = xfs_iunlink_lookup_backref(pag, target_agino);
 	if (*agino != NULLAGINO) {
-		error = xfs_iunlink_map_ino(tp, agno, *agino, imap, dipp, bpp);
+		error = xfs_iunlink_map_ino(tp, pag->pag_agno, *agino, imap,
+				dipp, bpp);
 		if (error)
 			return error;
 
@@ -2273,7 +2239,7 @@ xfs_iunlink_map_prev(
 		WARN_ON_ONCE(1);
 	}
 
-	trace_xfs_iunlink_map_prev_fallback(mp, agno);
+	trace_xfs_iunlink_map_prev_fallback(mp, pag->pag_agno);
 
 	/* Otherwise, walk the entire bucket until we find it. */
 	next_agino = head_agino;
@@ -2284,8 +2250,8 @@ xfs_iunlink_map_prev(
 			xfs_trans_brelse(tp, *bpp);
 
 		*agino = next_agino;
-		error = xfs_iunlink_map_ino(tp, agno, next_agino, imap, dipp,
-				bpp);
+		error = xfs_iunlink_map_ino(tp, pag->pag_agno, next_agino, imap,
+				dipp, bpp);
 		if (error)
 			return error;
 
@@ -2294,7 +2260,7 @@ xfs_iunlink_map_prev(
 		 * Make sure this pointer is valid and isn't an obvious
 		 * infinite loop.
 		 */
-		if (!xfs_verify_agino(mp, agno, unlinked_agino) ||
+		if (!xfs_verify_agino(mp, pag->pag_agno, unlinked_agino) ||
 		    next_agino == unlinked_agino) {
 			XFS_CORRUPTION_ERROR(__func__,
 					XFS_ERRLEVEL_LOW, mp,
@@ -2314,6 +2280,7 @@ xfs_iunlink_map_prev(
 STATIC int
 xfs_iunlink_remove(
 	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
 	struct xfs_inode	*ip)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
@@ -2321,7 +2288,6 @@ xfs_iunlink_remove(
 	struct xfs_buf		*agibp;
 	struct xfs_buf		*last_ibp;
 	struct xfs_dinode	*last_dip = NULL;
-	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
 	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
 	xfs_agino_t		next_agino;
 	xfs_agino_t		head_agino;
@@ -2331,7 +2297,7 @@ xfs_iunlink_remove(
 	trace_xfs_iunlink_remove(ip);
 
 	/* Get the agi buffer first.  It ensures lock ordering on the list. */
-	error = xfs_read_agi(mp, tp, agno, &agibp);
+	error = xfs_read_agi(mp, tp, pag->pag_agno, &agibp);
 	if (error)
 		return error;
 	agi = agibp->b_addr;
@@ -2341,7 +2307,7 @@ xfs_iunlink_remove(
 	 * go on.  Make sure the head pointer isn't garbage.
 	 */
 	head_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
-	if (!xfs_verify_agino(mp, agno, head_agino)) {
+	if (!xfs_verify_agino(mp, pag->pag_agno, head_agino)) {
 		XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp,
 				agi, sizeof(*agi));
 		return -EFSCORRUPTED;
@@ -2352,7 +2318,7 @@ xfs_iunlink_remove(
 	 * the old pointer value so that we can update whatever was previous
 	 * to us in the list to point to whatever was next in the list.
 	 */
-	error = xfs_iunlink_update_inode(tp, ip, agno, NULLAGINO, &next_agino);
+	error = xfs_iunlink_update_inode(tp, ip, pag, NULLAGINO, &next_agino);
 	if (error)
 		return error;
 
@@ -2364,8 +2330,7 @@ xfs_iunlink_remove(
 	 * this inode's backref to point from the next inode.
 	 */
 	if (next_agino != NULLAGINO) {
-		error = xfs_iunlink_change_backref(agibp->b_pag, next_agino,
-				NULLAGINO);
+		error = xfs_iunlink_change_backref(pag, next_agino, NULLAGINO);
 		if (error)
 			return error;
 	}
@@ -2375,14 +2340,13 @@ xfs_iunlink_remove(
 		xfs_agino_t	prev_agino;
 
 		/* We need to search the list for the inode being freed. */
-		error = xfs_iunlink_map_prev(tp, agno, head_agino, agino,
-				&prev_agino, &imap, &last_dip, &last_ibp,
-				agibp->b_pag);
+		error = xfs_iunlink_map_prev(tp, pag, head_agino, agino,
+				&prev_agino, &imap, &last_dip, &last_ibp);
 		if (error)
 			return error;
 
 		/* Point the previous inode on the list to the next inode. */
-		xfs_iunlink_update_dinode(tp, agno, prev_agino, last_ibp,
+		xfs_iunlink_update_dinode(tp, pag, prev_agino, last_ibp,
 				last_dip, &imap, next_agino);
 
 		/*
@@ -2398,7 +2362,7 @@ xfs_iunlink_remove(
 	}
 
 	/* Point the head of the list to the next unlinked inode. */
-	return xfs_iunlink_update_bucket(tp, agno, agibp, bucket_index,
+	return xfs_iunlink_update_bucket(tp, pag, agibp, bucket_index,
 			next_agino);
 }
 
@@ -2409,12 +2373,11 @@ xfs_iunlink_remove(
  */
 static void
 xfs_ifree_mark_inode_stale(
-	struct xfs_buf		*bp,
+	struct xfs_perag	*pag,
 	struct xfs_inode	*free_ip,
 	xfs_ino_t		inum)
 {
-	struct xfs_mount	*mp = bp->b_mount;
-	struct xfs_perag	*pag = bp->b_pag;
+	struct xfs_mount	*mp = pag->pag_mount;
 	struct xfs_inode_log_item *iip;
 	struct xfs_inode	*ip;
 
@@ -2504,10 +2467,11 @@ out_iflags_unlock:
  * inodes that are in memory - they all must be marked stale and attached to
  * the cluster buffer.
  */
-STATIC int
+static int
 xfs_ifree_cluster(
-	struct xfs_inode	*free_ip,
 	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	struct xfs_inode	*free_ip,
 	struct xfs_icluster	*xic)
 {
 	struct xfs_mount	*mp = free_ip->i_mount;
@@ -2569,7 +2533,7 @@ xfs_ifree_cluster(
 		 * already marked XFS_ISTALE.
 		 */
 		for (i = 0; i < igeo->inodes_per_cluster; i++)
-			xfs_ifree_mark_inode_stale(bp, free_ip, inum + i);
+			xfs_ifree_mark_inode_stale(pag, free_ip, inum + i);
 
 		xfs_trans_stale_inode_buf(tp, bp);
 		xfs_trans_binval(tp, bp);
@@ -2592,9 +2556,11 @@ xfs_ifree(
 	struct xfs_trans	*tp,
 	struct xfs_inode	*ip)
 {
-	int			error;
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_perag	*pag;
 	struct xfs_icluster	xic = { 0 };
 	struct xfs_inode_log_item *iip = ip->i_itemp;
+	int			error;
 
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
 	ASSERT(VFS_I(ip)->i_nlink == 0);
@@ -2602,16 +2568,18 @@ xfs_ifree(
 	ASSERT(ip->i_disk_size == 0 || !S_ISREG(VFS_I(ip)->i_mode));
 	ASSERT(ip->i_nblocks == 0);
 
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+
 	/*
 	 * Pull the on-disk inode from the AGI unlinked list.
 	 */
-	error = xfs_iunlink_remove(tp, ip);
+	error = xfs_iunlink_remove(tp, pag, ip);
 	if (error)
-		return error;
+		goto out;
 
-	error = xfs_difree(tp, ip->i_ino, &xic);
+	error = xfs_difree(tp, pag, ip->i_ino, &xic);
 	if (error)
-		return error;
+		goto out;
 
 	/*
 	 * Free any local-format data sitting around before we reset the
@@ -2626,7 +2594,7 @@ xfs_ifree(
 
 	VFS_I(ip)->i_mode = 0;		/* mark incore inode as free */
 	ip->i_diflags = 0;
-	ip->i_diflags2 = ip->i_mount->m_ino_geo.new_diflags2;
+	ip->i_diflags2 = mp->m_ino_geo.new_diflags2;
 	ip->i_forkoff = 0;		/* mark the attr fork not in use */
 	ip->i_df.if_format = XFS_DINODE_FMT_EXTENTS;
 	if (xfs_iflags_test(ip, XFS_IPRESERVE_DM_FIELDS))
@@ -2645,8 +2613,9 @@ xfs_ifree(
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
 	if (xic.deleted)
-		error = xfs_ifree_cluster(ip, tp, &xic);
-
+		error = xfs_ifree_cluster(tp, pag, ip, &xic);
+out:
+	xfs_perag_put(pag);
 	return error;
 }
 
@@ -2664,7 +2633,7 @@ xfs_iunpin(
 	trace_xfs_inode_unpin_nowait(ip, _RET_IP_);
 
 	/* Give the log a push to start the unpinning I/O */
-	xfs_log_force_lsn(ip->i_mount, ip->i_itemp->ili_last_lsn, 0, NULL);
+	xfs_log_force_seq(ip->i_mount, ip->i_itemp->ili_commit_seq, 0, NULL);
 
 }
 
@@ -3250,8 +3219,13 @@ xfs_rename(
 	 * in future.
 	 */
 	if (wip) {
+		struct xfs_perag	*pag;
+
 		ASSERT(VFS_I(wip)->i_nlink == 0);
-		error = xfs_iunlink_remove(tp, wip);
+
+		pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, wip->i_ino));
+		error = xfs_iunlink_remove(tp, pag, wip);
+		xfs_perag_put(pag);
 		if (error)
 			goto out_trans_cancel;
 
@@ -3673,16 +3647,16 @@ int
 xfs_log_force_inode(
 	struct xfs_inode	*ip)
 {
-	xfs_lsn_t		lsn = 0;
+	xfs_csn_t		seq = 0;
 
 	xfs_ilock(ip, XFS_ILOCK_SHARED);
 	if (xfs_ipincount(ip))
-		lsn = ip->i_itemp->ili_last_lsn;
+		seq = ip->i_itemp->ili_commit_seq;
 	xfs_iunlock(ip, XFS_ILOCK_SHARED);
 
-	if (!lsn)
+	if (!seq)
 		return 0;
-	return xfs_log_force_lsn(ip->i_mount, lsn, XFS_LOG_SYNC, NULL);
+	return xfs_log_force_seq(ip->i_mount, seq, XFS_LOG_SYNC, NULL);
 }
 
 /*
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index ca826cfba91c..4b6703dbffb8 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -431,11 +431,10 @@ void		xfs_lock_two_inodes(struct xfs_inode *ip0, uint ip0_mode,
 xfs_extlen_t	xfs_get_extsz_hint(struct xfs_inode *ip);
 xfs_extlen_t	xfs_get_cowextsz_hint(struct xfs_inode *ip);
 
-int		xfs_dir_ialloc(struct user_namespace *mnt_userns,
-			       struct xfs_trans **tpp, struct xfs_inode *dp,
-			       umode_t mode, xfs_nlink_t nlink, dev_t dev,
-			       prid_t prid, bool need_xattr,
-			       struct xfs_inode **ipp);
+int xfs_init_new_inode(struct user_namespace *mnt_userns, struct xfs_trans *tp,
+		struct xfs_inode *pip, xfs_ino_t ino, umode_t mode,
+		xfs_nlink_t nlink, dev_t rdev, prid_t prid, bool init_xattrs,
+		struct xfs_inode **ipp);
 
 static inline int
 xfs_itruncate_extents(
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 6764d12342da..35de30849fcc 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -28,6 +28,20 @@ static inline struct xfs_inode_log_item *INODE_ITEM(struct xfs_log_item *lip)
 	return container_of(lip, struct xfs_inode_log_item, ili_item);
 }
 
+/*
+ * The logged size of an inode fork is always the current size of the inode
+ * fork. This means that when an inode fork is relogged, the size of the logged
+ * region is determined by the current state, not the combination of the
+ * previously logged state + the current state. This is different relogging
+ * behaviour to most other log items which will retain the size of the
+ * previously logged changes when smaller regions are relogged.
+ *
+ * Hence operations that remove data from the inode fork (e.g. shortform
+ * dir/attr remove, extent form extent removal, etc), the size of the relogged
+ * inode gets -smaller- rather than stays the same size as the previously logged
+ * size and this can result in the committing transaction reducing the amount of
+ * space being consumed by the CIL.
+ */
 STATIC void
 xfs_inode_item_data_fork_size(
 	struct xfs_inode_log_item *iip,
@@ -629,9 +643,9 @@ xfs_inode_item_committed(
 STATIC void
 xfs_inode_item_committing(
 	struct xfs_log_item	*lip,
-	xfs_lsn_t		commit_lsn)
+	xfs_csn_t		seq)
 {
-	INODE_ITEM(lip)->ili_last_lsn = commit_lsn;
+	INODE_ITEM(lip)->ili_commit_seq = seq;
 	return xfs_inode_item_release(lip);
 }
 
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
index 4b926e32831c..403b45ab9aa2 100644
--- a/fs/xfs/xfs_inode_item.h
+++ b/fs/xfs/xfs_inode_item.h
@@ -33,7 +33,7 @@ struct xfs_inode_log_item {
 	unsigned int		ili_fields;	   /* fields to be logged */
 	unsigned int		ili_fsync_fields;  /* logged since last fsync */
 	xfs_lsn_t		ili_flush_lsn;	   /* lsn at last flush */
-	xfs_lsn_t		ili_last_lsn;	   /* lsn at last transaction */
+	xfs_csn_t		ili_commit_seq;	   /* last transaction commit */
 };
 
 static inline int xfs_inode_clean(struct xfs_inode *ip)
diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
index 7b1796cede10..65270e63c032 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -1875,7 +1875,7 @@ out:
 static inline int
 xfs_fs_eofblocks_from_user(
 	struct xfs_fs_eofblocks		*src,
-	struct xfs_eofblocks		*dst)
+	struct xfs_icwalk		*dst)
 {
 	if (src->eof_version != XFS_EOFBLOCKS_VERSION)
 		return -EINVAL;
@@ -1887,21 +1887,32 @@ xfs_fs_eofblocks_from_user(
 	    memchr_inv(src->pad64, 0, sizeof(src->pad64)))
 		return -EINVAL;
 
-	dst->eof_flags = src->eof_flags;
-	dst->eof_prid = src->eof_prid;
-	dst->eof_min_file_size = src->eof_min_file_size;
-
-	dst->eof_uid = INVALID_UID;
+	dst->icw_flags = 0;
+	if (src->eof_flags & XFS_EOF_FLAGS_SYNC)
+		dst->icw_flags |= XFS_ICWALK_FLAG_SYNC;
+	if (src->eof_flags & XFS_EOF_FLAGS_UID)
+		dst->icw_flags |= XFS_ICWALK_FLAG_UID;
+	if (src->eof_flags & XFS_EOF_FLAGS_GID)
+		dst->icw_flags |= XFS_ICWALK_FLAG_GID;
+	if (src->eof_flags & XFS_EOF_FLAGS_PRID)
+		dst->icw_flags |= XFS_ICWALK_FLAG_PRID;
+	if (src->eof_flags & XFS_EOF_FLAGS_MINFILESIZE)
+		dst->icw_flags |= XFS_ICWALK_FLAG_MINFILESIZE;
+
+	dst->icw_prid = src->eof_prid;
+	dst->icw_min_file_size = src->eof_min_file_size;
+
+	dst->icw_uid = INVALID_UID;
 	if (src->eof_flags & XFS_EOF_FLAGS_UID) {
-		dst->eof_uid = make_kuid(current_user_ns(), src->eof_uid);
-		if (!uid_valid(dst->eof_uid))
+		dst->icw_uid = make_kuid(current_user_ns(), src->eof_uid);
+		if (!uid_valid(dst->icw_uid))
 			return -EINVAL;
 	}
 
-	dst->eof_gid = INVALID_GID;
+	dst->icw_gid = INVALID_GID;
 	if (src->eof_flags & XFS_EOF_FLAGS_GID) {
-		dst->eof_gid = make_kgid(current_user_ns(), src->eof_gid);
-		if (!gid_valid(dst->eof_gid))
+		dst->icw_gid = make_kgid(current_user_ns(), src->eof_gid);
+		if (!gid_valid(dst->icw_gid))
 			return -EINVAL;
 	}
 	return 0;
@@ -2164,8 +2175,8 @@ xfs_file_ioctl(
 		return xfs_errortag_clearall(mp);
 
 	case XFS_IOC_FREE_EOFBLOCKS: {
-		struct xfs_fs_eofblocks eofb;
-		struct xfs_eofblocks keofb;
+		struct xfs_fs_eofblocks	eofb;
+		struct xfs_icwalk	icw;
 
 		if (!capable(CAP_SYS_ADMIN))
 			return -EPERM;
@@ -2176,14 +2187,14 @@ xfs_file_ioctl(
 		if (copy_from_user(&eofb, arg, sizeof(eofb)))
 			return -EFAULT;
 
-		error = xfs_fs_eofblocks_from_user(&eofb, &keofb);
+		error = xfs_fs_eofblocks_from_user(&eofb, &icw);
 		if (error)
 			return error;
 
-		trace_xfs_ioc_free_eofblocks(mp, &keofb, _RET_IP_);
+		trace_xfs_ioc_free_eofblocks(mp, &icw, _RET_IP_);
 
 		sb_start_write(mp->m_super);
-		error = xfs_blockgc_free_space(mp, &keofb);
+		error = xfs_blockgc_free_space(mp, &icw);
 		sb_end_write(mp->m_super);
 		return error;
 	}
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index dfe24b7f26e5..93c082db04b7 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -543,7 +543,7 @@ xfs_stat_blksize(
 	 * always return the realtime extent size.
 	 */
 	if (XFS_IS_REALTIME_INODE(ip))
-		return xfs_get_extsz_hint(ip) << mp->m_sb.sb_blocklog;
+		return XFS_FSB_TO_B(mp, xfs_get_extsz_hint(ip));
 
 	/*
 	 * Allow large block sizes to be reported to userspace programs if the
@@ -560,7 +560,7 @@ xfs_stat_blksize(
 	 */
 	if (mp->m_flags & XFS_MOUNT_LARGEIO) {
 		if (mp->m_swidth)
-			return mp->m_swidth << mp->m_sb.sb_blocklog;
+			return XFS_FSB_TO_B(mp, mp->m_swidth);
 		if (mp->m_flags & XFS_MOUNT_ALLOCSIZE)
 			return 1U << mp->m_allocsize_log;
 	}
diff --git a/fs/xfs/xfs_iwalk.c b/fs/xfs/xfs_iwalk.c
index c4a340f1f1e1..917d51eefee3 100644
--- a/fs/xfs/xfs_iwalk.c
+++ b/fs/xfs/xfs_iwalk.c
@@ -21,6 +21,7 @@
 #include "xfs_health.h"
 #include "xfs_trans.h"
 #include "xfs_pwork.h"
+#include "xfs_ag.h"
 
 /*
  * Walking Inodes in the Filesystem
@@ -51,6 +52,7 @@ struct xfs_iwalk_ag {
 
 	struct xfs_mount		*mp;
 	struct xfs_trans		*tp;
+	struct xfs_perag		*pag;
 
 	/* Where do we start the traversal? */
 	xfs_ino_t			startino;
@@ -90,7 +92,7 @@ struct xfs_iwalk_ag {
 STATIC void
 xfs_iwalk_ichunk_ra(
 	struct xfs_mount		*mp,
-	xfs_agnumber_t			agno,
+	struct xfs_perag		*pag,
 	struct xfs_inobt_rec_incore	*irec)
 {
 	struct xfs_ino_geometry		*igeo = M_IGEO(mp);
@@ -106,7 +108,7 @@ xfs_iwalk_ichunk_ra(
 
 		imask = xfs_inobt_maskn(i, igeo->inodes_per_cluster);
 		if (imask & ~irec->ir_free) {
-			xfs_btree_reada_bufs(mp, agno, agbno,
+			xfs_btree_reada_bufs(mp, pag->pag_agno, agbno,
 					igeo->blocks_per_cluster,
 					&xfs_inode_buf_ops);
 		}
@@ -174,26 +176,25 @@ xfs_iwalk_free(
 /* For each inuse inode in each cached inobt record, call our function. */
 STATIC int
 xfs_iwalk_ag_recs(
-	struct xfs_iwalk_ag		*iwag)
+	struct xfs_iwalk_ag	*iwag)
 {
-	struct xfs_mount		*mp = iwag->mp;
-	struct xfs_trans		*tp = iwag->tp;
-	xfs_ino_t			ino;
-	unsigned int			i, j;
-	xfs_agnumber_t			agno;
-	int				error;
+	struct xfs_mount	*mp = iwag->mp;
+	struct xfs_trans	*tp = iwag->tp;
+	struct xfs_perag	*pag = iwag->pag;
+	xfs_ino_t		ino;
+	unsigned int		i, j;
+	int			error;
 
-	agno = XFS_INO_TO_AGNO(mp, iwag->startino);
 	for (i = 0; i < iwag->nr_recs; i++) {
 		struct xfs_inobt_rec_incore	*irec = &iwag->recs[i];
 
-		trace_xfs_iwalk_ag_rec(mp, agno, irec);
+		trace_xfs_iwalk_ag_rec(mp, pag->pag_agno, irec);
 
 		if (xfs_pwork_want_abort(&iwag->pwork))
 			return 0;
 
 		if (iwag->inobt_walk_fn) {
-			error = iwag->inobt_walk_fn(mp, tp, agno, irec,
+			error = iwag->inobt_walk_fn(mp, tp, pag->pag_agno, irec,
 					iwag->data);
 			if (error)
 				return error;
@@ -211,7 +212,8 @@ xfs_iwalk_ag_recs(
 				continue;
 
 			/* Otherwise call our function. */
-			ino = XFS_AGINO_TO_INO(mp, agno, irec->ir_startino + j);
+			ino = XFS_AGINO_TO_INO(mp, pag->pag_agno,
+						irec->ir_startino + j);
 			error = iwag->iwalk_fn(mp, tp, ino, iwag->data);
 			if (error)
 				return error;
@@ -257,7 +259,6 @@ xfs_iwalk_del_inobt(
 STATIC int
 xfs_iwalk_ag_start(
 	struct xfs_iwalk_ag	*iwag,
-	xfs_agnumber_t		agno,
 	xfs_agino_t		agino,
 	struct xfs_btree_cur	**curpp,
 	struct xfs_buf		**agi_bpp,
@@ -265,12 +266,13 @@ xfs_iwalk_ag_start(
 {
 	struct xfs_mount	*mp = iwag->mp;
 	struct xfs_trans	*tp = iwag->tp;
+	struct xfs_perag	*pag = iwag->pag;
 	struct xfs_inobt_rec_incore *irec;
 	int			error;
 
 	/* Set up a fresh cursor and empty the inobt cache. */
 	iwag->nr_recs = 0;
-	error = xfs_inobt_cur(mp, tp, agno, XFS_BTNUM_INO, curpp, agi_bpp);
+	error = xfs_inobt_cur(mp, tp, pag, XFS_BTNUM_INO, curpp, agi_bpp);
 	if (error)
 		return error;
 
@@ -304,7 +306,7 @@ xfs_iwalk_ag_start(
 	if (XFS_IS_CORRUPT(mp, *has_more != 1))
 		return -EFSCORRUPTED;
 
-	iwag->lastino = XFS_AGINO_TO_INO(mp, agno,
+	iwag->lastino = XFS_AGINO_TO_INO(mp, pag->pag_agno,
 				irec->ir_startino + XFS_INODES_PER_CHUNK - 1);
 
 	/*
@@ -345,7 +347,6 @@ out_advance:
 STATIC int
 xfs_iwalk_run_callbacks(
 	struct xfs_iwalk_ag		*iwag,
-	xfs_agnumber_t			agno,
 	struct xfs_btree_cur		**curpp,
 	struct xfs_buf			**agi_bpp,
 	int				*has_more)
@@ -376,7 +377,7 @@ xfs_iwalk_run_callbacks(
 		return 0;
 
 	/* ...and recreate the cursor just past where we left off. */
-	error = xfs_inobt_cur(mp, tp, agno, XFS_BTNUM_INO, curpp, agi_bpp);
+	error = xfs_inobt_cur(mp, tp, iwag->pag, XFS_BTNUM_INO, curpp, agi_bpp);
 	if (error)
 		return error;
 
@@ -390,17 +391,17 @@ xfs_iwalk_ag(
 {
 	struct xfs_mount		*mp = iwag->mp;
 	struct xfs_trans		*tp = iwag->tp;
+	struct xfs_perag		*pag = iwag->pag;
 	struct xfs_buf			*agi_bp = NULL;
 	struct xfs_btree_cur		*cur = NULL;
-	xfs_agnumber_t			agno;
 	xfs_agino_t			agino;
 	int				has_more;
 	int				error = 0;
 
 	/* Set up our cursor at the right place in the inode btree. */
-	agno = XFS_INO_TO_AGNO(mp, iwag->startino);
+	ASSERT(pag->pag_agno == XFS_INO_TO_AGNO(mp, iwag->startino));
 	agino = XFS_INO_TO_AGINO(mp, iwag->startino);
-	error = xfs_iwalk_ag_start(iwag, agno, agino, &cur, &agi_bp, &has_more);
+	error = xfs_iwalk_ag_start(iwag, agino, &cur, &agi_bp, &has_more);
 
 	while (!error && has_more) {
 		struct xfs_inobt_rec_incore	*irec;
@@ -417,7 +418,7 @@ xfs_iwalk_ag(
 			break;
 
 		/* Make sure that we always move forward. */
-		rec_fsino = XFS_AGINO_TO_INO(mp, agno, irec->ir_startino);
+		rec_fsino = XFS_AGINO_TO_INO(mp, pag->pag_agno, irec->ir_startino);
 		if (iwag->lastino != NULLFSINO &&
 		    XFS_IS_CORRUPT(mp, iwag->lastino >= rec_fsino)) {
 			error = -EFSCORRUPTED;
@@ -438,7 +439,7 @@ xfs_iwalk_ag(
 		 * walking the inodes.
 		 */
 		if (iwag->iwalk_fn)
-			xfs_iwalk_ichunk_ra(mp, agno, irec);
+			xfs_iwalk_ichunk_ra(mp, pag, irec);
 
 		/*
 		 * If there's space in the buffer for more records, increment
@@ -458,15 +459,14 @@ xfs_iwalk_ag(
 		 * we would be if we had been able to increment like above.
 		 */
 		ASSERT(has_more);
-		error = xfs_iwalk_run_callbacks(iwag, agno, &cur, &agi_bp,
-				&has_more);
+		error = xfs_iwalk_run_callbacks(iwag, &cur, &agi_bp, &has_more);
 	}
 
 	if (iwag->nr_recs == 0 || error)
 		goto out;
 
 	/* Walk the unprocessed records in the cache. */
-	error = xfs_iwalk_run_callbacks(iwag, agno, &cur, &agi_bp, &has_more);
+	error = xfs_iwalk_run_callbacks(iwag, &cur, &agi_bp, &has_more);
 
 out:
 	xfs_iwalk_del_inobt(tp, &cur, &agi_bp, error);
@@ -555,6 +555,7 @@ xfs_iwalk(
 		.pwork		= XFS_PWORK_SINGLE_THREADED,
 		.lastino	= NULLFSINO,
 	};
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, startino);
 	int			error;
 
@@ -565,15 +566,19 @@ xfs_iwalk(
 	if (error)
 		return error;
 
-	for (; agno < mp->m_sb.sb_agcount; agno++) {
+	for_each_perag_from(mp, agno, pag) {
+		iwag.pag = pag;
 		error = xfs_iwalk_ag(&iwag);
 		if (error)
 			break;
 		iwag.startino = XFS_AGINO_TO_INO(mp, agno + 1, 0);
 		if (flags & XFS_INOBT_WALK_SAME_AG)
 			break;
+		iwag.pag = NULL;
 	}
 
+	if (iwag.pag)
+		xfs_perag_put(pag);
 	xfs_iwalk_free(&iwag);
 	return error;
 }
@@ -598,6 +603,7 @@ xfs_iwalk_ag_work(
 	error = xfs_iwalk_ag(iwag);
 	xfs_iwalk_free(iwag);
 out:
+	xfs_perag_put(iwag->pag);
 	kmem_free(iwag);
 	return error;
 }
@@ -617,6 +623,7 @@ xfs_iwalk_threaded(
 	void			*data)
 {
 	struct xfs_pwork_ctl	pctl;
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, startino);
 	int			error;
 
@@ -627,7 +634,7 @@ xfs_iwalk_threaded(
 	if (error)
 		return error;
 
-	for (; agno < mp->m_sb.sb_agcount; agno++) {
+	for_each_perag_from(mp, agno, pag) {
 		struct xfs_iwalk_ag	*iwag;
 
 		if (xfs_pwork_ctl_want_abort(&pctl))
@@ -635,17 +642,25 @@ xfs_iwalk_threaded(
 
 		iwag = kmem_zalloc(sizeof(struct xfs_iwalk_ag), 0);
 		iwag->mp = mp;
+
+		/*
+		 * perag is being handed off to async work, so take another
+		 * reference for the async work to release.
+		 */
+		atomic_inc(&pag->pag_ref);
+		iwag->pag = pag;
 		iwag->iwalk_fn = iwalk_fn;
 		iwag->data = data;
 		iwag->startino = startino;
 		iwag->sz_recs = xfs_iwalk_prefetch(inode_records);
 		iwag->lastino = NULLFSINO;
 		xfs_pwork_queue(&pctl, &iwag->pwork);
-		startino = XFS_AGINO_TO_INO(mp, agno + 1, 0);
+		startino = XFS_AGINO_TO_INO(mp, pag->pag_agno + 1, 0);
 		if (flags & XFS_INOBT_WALK_SAME_AG)
 			break;
 	}
-
+	if (pag)
+		xfs_perag_put(pag);
 	if (polled)
 		xfs_pwork_poll(&pctl);
 	return xfs_pwork_destroy(&pctl);
@@ -715,6 +730,7 @@ xfs_inobt_walk(
 		.pwork		= XFS_PWORK_SINGLE_THREADED,
 		.lastino	= NULLFSINO,
 	};
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno = XFS_INO_TO_AGNO(mp, startino);
 	int			error;
 
@@ -725,15 +741,19 @@ xfs_inobt_walk(
 	if (error)
 		return error;
 
-	for (; agno < mp->m_sb.sb_agcount; agno++) {
+	for_each_perag_from(mp, agno, pag) {
+		iwag.pag = pag;
 		error = xfs_iwalk_ag(&iwag);
 		if (error)
 			break;
-		iwag.startino = XFS_AGINO_TO_INO(mp, agno + 1, 0);
+		iwag.startino = XFS_AGINO_TO_INO(mp, pag->pag_agno + 1, 0);
 		if (flags & XFS_INOBT_WALK_SAME_AG)
 			break;
+		iwag.pag = NULL;
 	}
 
+	if (iwag.pag)
+		xfs_perag_put(pag);
 	xfs_iwalk_free(&iwag);
 	return error;
 }
diff --git a/fs/xfs/xfs_linux.h b/fs/xfs/xfs_linux.h
index 7688663b9773..c174262a074e 100644
--- a/fs/xfs/xfs_linux.h
+++ b/fs/xfs/xfs_linux.h
@@ -196,6 +196,8 @@ static inline uint64_t howmany_64(uint64_t x, uint32_t y)
 
 int xfs_rw_bdev(struct block_device *bdev, sector_t sector, unsigned int count,
 		char *data, unsigned int op);
+void xfs_flush_bdev_async(struct bio *bio, struct block_device *bdev,
+		struct completion *done);
 
 #define ASSERT_ALWAYS(expr)	\
 	(likely(expr) ? (void)0 : assfail(NULL, #expr, __FILE__, __LINE__))
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index a002425377b5..36fa2650b081 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -502,6 +502,7 @@ __xlog_state_release_iclog(
 		iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn);
 		xlog_verify_tail_lsn(log, iclog, tail_lsn);
 		/* cycle incremented when incrementing curr_block */
+		trace_xlog_iclog_syncing(iclog, _RET_IP_);
 		return true;
 	}
 
@@ -513,13 +514,14 @@ __xlog_state_release_iclog(
  * Flush iclog to disk if this is the last reference to the given iclog and the
  * it is in the WANT_SYNC state.
  */
-static int
+int
 xlog_state_release_iclog(
 	struct xlog		*log,
 	struct xlog_in_core	*iclog)
 {
 	lockdep_assert_held(&log->l_icloglock);
 
+	trace_xlog_iclog_release(iclog, _RET_IP_);
 	if (iclog->ic_state == XLOG_STATE_IOERROR)
 		return -EIO;
 
@@ -533,23 +535,6 @@ xlog_state_release_iclog(
 	return 0;
 }
 
-void
-xfs_log_release_iclog(
-	struct xlog_in_core	*iclog)
-{
-	struct xlog		*log = iclog->ic_log;
-	bool			sync = false;
-
-	if (atomic_dec_and_lock(&iclog->ic_refcnt, &log->l_icloglock)) {
-		if (iclog->ic_state != XLOG_STATE_IOERROR)
-			sync = __xlog_state_release_iclog(log, iclog);
-		spin_unlock(&log->l_icloglock);
-	}
-
-	if (sync)
-		xlog_sync(log, iclog);
-}
-
 /*
  * Mount a log filesystem
  *
@@ -770,6 +755,9 @@ xfs_log_mount_finish(
 	if (readonly)
 		mp->m_flags |= XFS_MOUNT_RDONLY;
 
+	/* Make sure the log is dead if we're returning failure. */
+	ASSERT(!error || (mp->m_log->l_flags & XLOG_IO_ERROR));
+
 	return error;
 }
 
@@ -786,16 +774,19 @@ xfs_log_mount_cancel(
 }
 
 /*
- * Wait for the iclog to be written disk, or return an error if the log has been
- * shut down.
+ * Wait for the iclog and all prior iclogs to be written disk as required by the
+ * log force state machine. Waiting on ic_force_wait ensures iclog completions
+ * have been ordered and callbacks run before we are woken here, hence
+ * guaranteeing that all the iclogs up to this one are on stable storage.
  */
-static int
+int
 xlog_wait_on_iclog(
 	struct xlog_in_core	*iclog)
 		__releases(iclog->ic_log->l_icloglock)
 {
 	struct xlog		*log = iclog->ic_log;
 
+	trace_xlog_iclog_wait_on(iclog, _RET_IP_);
 	if (!XLOG_FORCED_SHUTDOWN(log) &&
 	    iclog->ic_state != XLOG_STATE_ACTIVE &&
 	    iclog->ic_state != XLOG_STATE_DIRTY) {
@@ -818,9 +809,7 @@ xlog_wait_on_iclog(
 static int
 xlog_write_unmount_record(
 	struct xlog		*log,
-	struct xlog_ticket	*ticket,
-	xfs_lsn_t		*lsn,
-	uint			flags)
+	struct xlog_ticket	*ticket)
 {
 	struct xfs_unmount_log_format ulf = {
 		.magic = XLOG_UNMOUNT_TYPE,
@@ -837,7 +826,15 @@ xlog_write_unmount_record(
 
 	/* account for space used by record data */
 	ticket->t_curr_res -= sizeof(ulf);
-	return xlog_write(log, &vec, ticket, lsn, NULL, flags, false);
+
+	/*
+	 * For external log devices, we need to flush the data device cache
+	 * first to ensure all metadata writeback is on stable storage before we
+	 * stamp the tail LSN into the unmount record.
+	 */
+	if (log->l_targ != log->l_mp->m_ddev_targp)
+		blkdev_issue_flush(log->l_targ->bt_bdev);
+	return xlog_write(log, &vec, ticket, NULL, NULL, XLOG_UNMOUNT_TRANS);
 }
 
 /*
@@ -851,15 +848,13 @@ xlog_unmount_write(
 	struct xfs_mount	*mp = log->l_mp;
 	struct xlog_in_core	*iclog;
 	struct xlog_ticket	*tic = NULL;
-	xfs_lsn_t		lsn;
-	uint			flags = XLOG_UNMOUNT_TRANS;
 	int			error;
 
 	error = xfs_log_reserve(mp, 600, 1, &tic, XFS_LOG, 0);
 	if (error)
 		goto out_err;
 
-	error = xlog_write_unmount_record(log, tic, &lsn, flags);
+	error = xlog_write_unmount_record(log, tic);
 	/*
 	 * At this point, we're umounting anyway, so there's no point in
 	 * transitioning log state to IOERROR. Just continue...
@@ -876,6 +871,11 @@ out_err:
 	else
 		ASSERT(iclog->ic_state == XLOG_STATE_WANT_SYNC ||
 		       iclog->ic_state == XLOG_STATE_IOERROR);
+	/*
+	 * Ensure the journal is fully flushed and on stable storage once the
+	 * iclog containing the unmount record is written.
+	 */
+	iclog->ic_flags |= (XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA);
 	error = xlog_state_release_iclog(log, iclog);
 	xlog_wait_on_iclog(iclog);
 
@@ -1401,6 +1401,11 @@ xlog_alloc_log(
 	xlog_assign_atomic_lsn(&log->l_last_sync_lsn, 1, 0);
 	log->l_curr_cycle  = 1;	    /* 0 is bad since this is initial value */
 
+	if (xfs_sb_version_haslogv2(&mp->m_sb) && mp->m_sb.sb_logsunit > 1)
+		log->l_iclog_roundoff = mp->m_sb.sb_logsunit;
+	else
+		log->l_iclog_roundoff = BBSIZE;
+
 	xlog_grant_head_init(&log->l_reserve_head);
 	xlog_grant_head_init(&log->l_write_head);
 
@@ -1479,7 +1484,6 @@ xlog_alloc_log(
 		iclog->ic_state = XLOG_STATE_ACTIVE;
 		iclog->ic_log = log;
 		atomic_set(&iclog->ic_refcnt, 0);
-		spin_lock_init(&iclog->ic_callback_lock);
 		INIT_LIST_HEAD(&iclog->ic_callbacks);
 		iclog->ic_datap = (char *)iclog->ic_data + log->l_iclog_hsize;
 
@@ -1546,8 +1550,7 @@ xlog_commit_record(
 	if (XLOG_FORCED_SHUTDOWN(log))
 		return -EIO;
 
-	error = xlog_write(log, &vec, ticket, lsn, iclog, XLOG_COMMIT_TRANS,
-			   false);
+	error = xlog_write(log, &vec, ticket, lsn, iclog, XLOG_COMMIT_TRANS);
 	if (error)
 		xfs_force_shutdown(log->l_mp, SHUTDOWN_LOG_IO_ERROR);
 	return error;
@@ -1753,10 +1756,10 @@ xlog_write_iclog(
 	struct xlog		*log,
 	struct xlog_in_core	*iclog,
 	uint64_t		bno,
-	unsigned int		count,
-	bool			need_flush)
+	unsigned int		count)
 {
 	ASSERT(bno < log->l_logBBsize);
+	trace_xlog_iclog_write(iclog, _RET_IP_);
 
 	/*
 	 * We lock the iclogbufs here so that we can serialise against I/O
@@ -1792,10 +1795,12 @@ xlog_write_iclog(
 	 * writeback throttle from throttling log writes behind background
 	 * metadata writeback and causing priority inversions.
 	 */
-	iclog->ic_bio.bi_opf = REQ_OP_WRITE | REQ_META | REQ_SYNC |
-				REQ_IDLE | REQ_FUA;
-	if (need_flush)
+	iclog->ic_bio.bi_opf = REQ_OP_WRITE | REQ_META | REQ_SYNC | REQ_IDLE;
+	if (iclog->ic_flags & XLOG_ICL_NEED_FLUSH)
 		iclog->ic_bio.bi_opf |= REQ_PREFLUSH;
+	if (iclog->ic_flags & XLOG_ICL_NEED_FUA)
+		iclog->ic_bio.bi_opf |= REQ_FUA;
+	iclog->ic_flags &= ~(XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA);
 
 	if (xlog_map_iclog_data(&iclog->ic_bio, iclog->ic_data, count)) {
 		xfs_force_shutdown(log->l_mp, SHUTDOWN_LOG_IO_ERROR);
@@ -1854,29 +1859,15 @@ xlog_calc_iclog_size(
 	uint32_t		*roundoff)
 {
 	uint32_t		count_init, count;
-	bool			use_lsunit;
-
-	use_lsunit = xfs_sb_version_haslogv2(&log->l_mp->m_sb) &&
-			log->l_mp->m_sb.sb_logsunit > 1;
 
 	/* Add for LR header */
 	count_init = log->l_iclog_hsize + iclog->ic_offset;
+	count = roundup(count_init, log->l_iclog_roundoff);
 
-	/* Round out the log write size */
-	if (use_lsunit) {
-		/* we have a v2 stripe unit to use */
-		count = XLOG_LSUNITTOB(log, XLOG_BTOLSUNIT(log, count_init));
-	} else {
-		count = BBTOB(BTOBB(count_init));
-	}
-
-	ASSERT(count >= count_init);
 	*roundoff = count - count_init;
 
-	if (use_lsunit)
-		ASSERT(*roundoff < log->l_mp->m_sb.sb_logsunit);
-	else
-		ASSERT(*roundoff < BBTOB(1));
+	ASSERT(count >= count_init);
+	ASSERT(*roundoff < log->l_iclog_roundoff);
 	return count;
 }
 
@@ -1912,9 +1903,9 @@ xlog_sync(
 	unsigned int		roundoff;       /* roundoff to BB or stripe */
 	uint64_t		bno;
 	unsigned int		size;
-	bool			need_flush = true, split = false;
 
 	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
+	trace_xlog_iclog_sync(iclog, _RET_IP_);
 
 	count = xlog_calc_iclog_size(log, iclog, &roundoff);
 
@@ -1937,10 +1928,8 @@ xlog_sync(
 	bno = BLOCK_LSN(be64_to_cpu(iclog->ic_header.h_lsn));
 
 	/* Do we need to split this write into 2 parts? */
-	if (bno + BTOBB(count) > log->l_logBBsize) {
+	if (bno + BTOBB(count) > log->l_logBBsize)
 		xlog_split_iclog(log, &iclog->ic_header, bno, count);
-		split = true;
-	}
 
 	/* calculcate the checksum */
 	iclog->ic_header.h_crc = xlog_cksum(log, &iclog->ic_header,
@@ -1961,22 +1950,8 @@ xlog_sync(
 			 be64_to_cpu(iclog->ic_header.h_lsn));
 	}
 #endif
-
-	/*
-	 * Flush the data device before flushing the log to make sure all meta
-	 * data written back from the AIL actually made it to disk before
-	 * stamping the new log tail LSN into the log buffer.  For an external
-	 * log we need to issue the flush explicitly, and unfortunately
-	 * synchronously here; for an internal log we can simply use the block
-	 * layer state machine for preflushes.
-	 */
-	if (log->l_targ != log->l_mp->m_ddev_targp || split) {
-		xfs_blkdev_issue_flush(log->l_mp->m_ddev_targp);
-		need_flush = false;
-	}
-
 	xlog_verify_iclog(log, iclog, count);
-	xlog_write_iclog(log, iclog, bno, count, need_flush);
+	xlog_write_iclog(log, iclog, bno, count);
 }
 
 /*
@@ -2158,13 +2133,16 @@ static int
 xlog_write_calc_vec_length(
 	struct xlog_ticket	*ticket,
 	struct xfs_log_vec	*log_vector,
-	bool			need_start_rec)
+	uint			optype)
 {
 	struct xfs_log_vec	*lv;
-	int			headers = need_start_rec ? 1 : 0;
+	int			headers = 0;
 	int			len = 0;
 	int			i;
 
+	if (optype & XLOG_START_TRANS)
+		headers++;
+
 	for (lv = log_vector; lv; lv = lv->lv_next) {
 		/* we don't write ordered log vectors */
 		if (lv->lv_buf_len == XFS_LOG_VEC_ORDERED)
@@ -2384,8 +2362,7 @@ xlog_write(
 	struct xlog_ticket	*ticket,
 	xfs_lsn_t		*start_lsn,
 	struct xlog_in_core	**commit_iclog,
-	uint			flags,
-	bool			need_start_rec)
+	uint			optype)
 {
 	struct xlog_in_core	*iclog = NULL;
 	struct xfs_log_vec	*lv = log_vector;
@@ -2413,8 +2390,9 @@ xlog_write(
 		xfs_force_shutdown(log->l_mp, SHUTDOWN_LOG_IO_ERROR);
 	}
 
-	len = xlog_write_calc_vec_length(ticket, log_vector, need_start_rec);
-	*start_lsn = 0;
+	len = xlog_write_calc_vec_length(ticket, log_vector, optype);
+	if (start_lsn)
+		*start_lsn = 0;
 	while (lv && (!lv->lv_niovecs || index < lv->lv_niovecs)) {
 		void		*ptr;
 		int		log_offset;
@@ -2427,8 +2405,8 @@ xlog_write(
 		ASSERT(log_offset <= iclog->ic_size - 1);
 		ptr = iclog->ic_datap + log_offset;
 
-		/* start_lsn is the first lsn written to. That's all we need. */
-		if (!*start_lsn)
+		/* Start_lsn is the first lsn written to. */
+		if (start_lsn && !*start_lsn)
 			*start_lsn = be64_to_cpu(iclog->ic_header.h_lsn);
 
 		/*
@@ -2441,6 +2419,7 @@ xlog_write(
 			int			copy_len;
 			int			copy_off;
 			bool			ordered = false;
+			bool			wrote_start_rec = false;
 
 			/* ordered log vectors have no regions to write */
 			if (lv->lv_buf_len == XFS_LOG_VEC_ORDERED) {
@@ -2458,13 +2437,15 @@ xlog_write(
 			 * write a start record. Only do this for the first
 			 * iclog we write to.
 			 */
-			if (need_start_rec) {
+			if (optype & XLOG_START_TRANS) {
 				xlog_write_start_rec(ptr, ticket);
 				xlog_write_adv_cnt(&ptr, &len, &log_offset,
 						sizeof(struct xlog_op_header));
+				optype &= ~XLOG_START_TRANS;
+				wrote_start_rec = true;
 			}
 
-			ophdr = xlog_write_setup_ophdr(log, ptr, ticket, flags);
+			ophdr = xlog_write_setup_ophdr(log, ptr, ticket, optype);
 			if (!ophdr)
 				return -EIO;
 
@@ -2495,14 +2476,13 @@ xlog_write(
 			}
 			copy_len += sizeof(struct xlog_op_header);
 			record_cnt++;
-			if (need_start_rec) {
+			if (wrote_start_rec) {
 				copy_len += sizeof(struct xlog_op_header);
 				record_cnt++;
-				need_start_rec = false;
 			}
 			data_cnt += contwr ? copy_len : 0;
 
-			error = xlog_write_copy_finish(log, iclog, flags,
+			error = xlog_write_copy_finish(log, iclog, optype,
 						       &record_cnt, &data_cnt,
 						       &partial_copy,
 						       &partial_copy_len,
@@ -2546,7 +2526,7 @@ next_lv:
 	spin_lock(&log->l_icloglock);
 	xlog_state_finish_copy(log, iclog, record_cnt, data_cnt);
 	if (commit_iclog) {
-		ASSERT(flags & XLOG_COMMIT_TRANS);
+		ASSERT(optype & XLOG_COMMIT_TRANS);
 		*commit_iclog = iclog;
 	} else {
 		error = xlog_state_release_iclog(log, iclog);
@@ -2562,6 +2542,7 @@ xlog_state_activate_iclog(
 	int			*iclogs_changed)
 {
 	ASSERT(list_empty_careful(&iclog->ic_callbacks));
+	trace_xlog_iclog_activate(iclog, _RET_IP_);
 
 	/*
 	 * If the number of ops in this iclog indicate it just contains the
@@ -2652,6 +2633,8 @@ xlog_state_clean_iclog(
 {
 	int			iclogs_changed = 0;
 
+	trace_xlog_iclog_clean(dirty_iclog, _RET_IP_);
+
 	dirty_iclog->ic_state = XLOG_STATE_DIRTY;
 
 	xlog_state_activate_iclogs(log, &iclogs_changed);
@@ -2711,6 +2694,7 @@ xlog_state_set_callback(
 	struct xlog_in_core	*iclog,
 	xfs_lsn_t		header_lsn)
 {
+	trace_xlog_iclog_callback(iclog, _RET_IP_);
 	iclog->ic_state = XLOG_STATE_CALLBACK;
 
 	ASSERT(XFS_LSN_CMP(atomic64_read(&log->l_last_sync_lsn),
@@ -2776,43 +2760,6 @@ xlog_state_iodone_process_iclog(
 	}
 }
 
-/*
- * Keep processing entries in the iclog callback list until we come around and
- * it is empty.  We need to atomically see that the list is empty and change the
- * state to DIRTY so that we don't miss any more callbacks being added.
- *
- * This function is called with the icloglock held and returns with it held. We
- * drop it while running callbacks, however, as holding it over thousands of
- * callbacks is unnecessary and causes excessive contention if we do.
- */
-static void
-xlog_state_do_iclog_callbacks(
-	struct xlog		*log,
-	struct xlog_in_core	*iclog)
-		__releases(&log->l_icloglock)
-		__acquires(&log->l_icloglock)
-{
-	spin_unlock(&log->l_icloglock);
-	spin_lock(&iclog->ic_callback_lock);
-	while (!list_empty(&iclog->ic_callbacks)) {
-		LIST_HEAD(tmp);
-
-		list_splice_init(&iclog->ic_callbacks, &tmp);
-
-		spin_unlock(&iclog->ic_callback_lock);
-		xlog_cil_process_committed(&tmp);
-		spin_lock(&iclog->ic_callback_lock);
-	}
-
-	/*
-	 * Pick up the icloglock while still holding the callback lock so we
-	 * serialise against anyone trying to add more callbacks to this iclog
-	 * now we've finished processing.
-	 */
-	spin_lock(&log->l_icloglock);
-	spin_unlock(&iclog->ic_callback_lock);
-}
-
 STATIC void
 xlog_state_do_callback(
 	struct xlog		*log)
@@ -2841,6 +2788,8 @@ xlog_state_do_callback(
 		repeats++;
 
 		do {
+			LIST_HEAD(cb_list);
+
 			if (xlog_state_iodone_process_iclog(log, iclog,
 							&ioerror))
 				break;
@@ -2850,13 +2799,15 @@ xlog_state_do_callback(
 				iclog = iclog->ic_next;
 				continue;
 			}
+			list_splice_init(&iclog->ic_callbacks, &cb_list);
+			spin_unlock(&log->l_icloglock);
 
-			/*
-			 * Running callbacks will drop the icloglock which means
-			 * we'll have to run at least one more complete loop.
-			 */
+			trace_xlog_iclog_callbacks_start(iclog, _RET_IP_);
+			xlog_cil_process_committed(&cb_list);
+			trace_xlog_iclog_callbacks_done(iclog, _RET_IP_);
 			cycled_icloglock = true;
-			xlog_state_do_iclog_callbacks(log, iclog);
+
+			spin_lock(&log->l_icloglock);
 			if (XLOG_FORCED_SHUTDOWN(log))
 				wake_up_all(&iclog->ic_force_wait);
 			else
@@ -2902,6 +2853,7 @@ xlog_state_done_syncing(
 
 	spin_lock(&log->l_icloglock);
 	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
+	trace_xlog_iclog_sync_done(iclog, _RET_IP_);
 
 	/*
 	 * If we got an error, either on the first buffer, or in the case of
@@ -2975,6 +2927,8 @@ restart:
 	atomic_inc(&iclog->ic_refcnt);	/* prevents sync */
 	log_offset = iclog->ic_offset;
 
+	trace_xlog_iclog_get_space(iclog, _RET_IP_);
+
 	/* On the 1st write to an iclog, figure out lsn.  This works
 	 * if iclogs marked XLOG_STATE_WANT_SYNC always write out what they are
 	 * committing to.  If the offset is set, that's how many blocks
@@ -3140,6 +3094,7 @@ xlog_state_switch_iclogs(
 {
 	ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE);
 	assert_spin_locked(&log->l_icloglock);
+	trace_xlog_iclog_switch(iclog, _RET_IP_);
 
 	if (!eventual_size)
 		eventual_size = iclog->ic_offset;
@@ -3152,9 +3107,8 @@ xlog_state_switch_iclogs(
 	log->l_curr_block += BTOBB(eventual_size)+BTOBB(log->l_iclog_hsize);
 
 	/* Round up to next log-sunit */
-	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb) &&
-	    log->l_mp->m_sb.sb_logsunit > 1) {
-		uint32_t sunit_bb = BTOBB(log->l_mp->m_sb.sb_logsunit);
+	if (log->l_iclog_roundoff > BBSIZE) {
+		uint32_t sunit_bb = BTOBB(log->l_iclog_roundoff);
 		log->l_curr_block = roundup(log->l_curr_block, sunit_bb);
 	}
 
@@ -3223,6 +3177,8 @@ xfs_log_force(
 	if (iclog->ic_state == XLOG_STATE_IOERROR)
 		goto out_error;
 
+	trace_xlog_iclog_force(iclog, _RET_IP_);
+
 	if (iclog->ic_state == XLOG_STATE_DIRTY ||
 	    (iclog->ic_state == XLOG_STATE_ACTIVE &&
 	     atomic_read(&iclog->ic_refcnt) == 0 && iclog->ic_offset == 0)) {
@@ -3281,14 +3237,13 @@ out_error:
 }
 
 static int
-__xfs_log_force_lsn(
-	struct xfs_mount	*mp,
+xlog_force_lsn(
+	struct xlog		*log,
 	xfs_lsn_t		lsn,
 	uint			flags,
 	int			*log_flushed,
 	bool			already_slept)
 {
-	struct xlog		*log = mp->m_log;
 	struct xlog_in_core	*iclog;
 
 	spin_lock(&log->l_icloglock);
@@ -3297,6 +3252,7 @@ __xfs_log_force_lsn(
 		goto out_error;
 
 	while (be64_to_cpu(iclog->ic_header.h_lsn) != lsn) {
+		trace_xlog_iclog_force_lsn(iclog, _RET_IP_);
 		iclog = iclog->ic_next;
 		if (iclog == log->l_iclog)
 			goto out_unlock;
@@ -3321,8 +3277,6 @@ __xfs_log_force_lsn(
 		if (!already_slept &&
 		    (iclog->ic_prev->ic_state == XLOG_STATE_WANT_SYNC ||
 		     iclog->ic_prev->ic_state == XLOG_STATE_SYNCING)) {
-			XFS_STATS_INC(mp, xs_log_force_sleep);
-
 			xlog_wait(&iclog->ic_prev->ic_write_wait,
 					&log->l_icloglock);
 			return -EAGAIN;
@@ -3360,25 +3314,29 @@ out_error:
  * to disk, that thread will wake up all threads waiting on the queue.
  */
 int
-xfs_log_force_lsn(
+xfs_log_force_seq(
 	struct xfs_mount	*mp,
-	xfs_lsn_t		lsn,
+	xfs_csn_t		seq,
 	uint			flags,
 	int			*log_flushed)
 {
+	struct xlog		*log = mp->m_log;
+	xfs_lsn_t		lsn;
 	int			ret;
-	ASSERT(lsn != 0);
+	ASSERT(seq != 0);
 
 	XFS_STATS_INC(mp, xs_log_force);
-	trace_xfs_log_force(mp, lsn, _RET_IP_);
+	trace_xfs_log_force(mp, seq, _RET_IP_);
 
-	lsn = xlog_cil_force_lsn(mp->m_log, lsn);
+	lsn = xlog_cil_force_seq(log, seq);
 	if (lsn == NULLCOMMITLSN)
 		return 0;
 
-	ret = __xfs_log_force_lsn(mp, lsn, flags, log_flushed, false);
-	if (ret == -EAGAIN)
-		ret = __xfs_log_force_lsn(mp, lsn, flags, log_flushed, true);
+	ret = xlog_force_lsn(log, lsn, flags, log_flushed, false);
+	if (ret == -EAGAIN) {
+		XFS_STATS_INC(mp, xs_log_force_sleep);
+		ret = xlog_force_lsn(log, lsn, flags, log_flushed, true);
+	}
 	return ret;
 }
 
@@ -3407,12 +3365,11 @@ xfs_log_ticket_get(
  * Figure out the total log space unit (in bytes) that would be
  * required for a log ticket.
  */
-int
-xfs_log_calc_unit_res(
-	struct xfs_mount	*mp,
+static int
+xlog_calc_unit_res(
+	struct xlog		*log,
 	int			unit_bytes)
 {
-	struct xlog		*log = mp->m_log;
 	int			iclog_space;
 	uint			num_headers;
 
@@ -3488,18 +3445,20 @@ xfs_log_calc_unit_res(
 	/* for commit-rec LR header - note: padding will subsume the ophdr */
 	unit_bytes += log->l_iclog_hsize;
 
-	/* for roundoff padding for transaction data and one for commit record */
-	if (xfs_sb_version_haslogv2(&mp->m_sb) && mp->m_sb.sb_logsunit > 1) {
-		/* log su roundoff */
-		unit_bytes += 2 * mp->m_sb.sb_logsunit;
-	} else {
-		/* BB roundoff */
-		unit_bytes += 2 * BBSIZE;
-        }
+	/* roundoff padding for transaction data and one for commit record */
+	unit_bytes += 2 * log->l_iclog_roundoff;
 
 	return unit_bytes;
 }
 
+int
+xfs_log_calc_unit_res(
+	struct xfs_mount	*mp,
+	int			unit_bytes)
+{
+	return xlog_calc_unit_res(mp->m_log, unit_bytes);
+}
+
 /*
  * Allocate and initialise a new log ticket.
  */
@@ -3516,7 +3475,7 @@ xlog_ticket_alloc(
 
 	tic = kmem_cache_zalloc(xfs_log_ticket_zone, GFP_NOFS | __GFP_NOFAIL);
 
-	unit_res = xfs_log_calc_unit_res(log->l_mp, unit_bytes);
+	unit_res = xlog_calc_unit_res(log, unit_bytes);
 
 	atomic_set(&tic->t_ref, 1);
 	tic->t_task		= current;
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index 044e02cb8921..813b972e9788 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -106,7 +106,7 @@ struct xfs_item_ops;
 struct xfs_trans;
 
 int	  xfs_log_force(struct xfs_mount *mp, uint flags);
-int	  xfs_log_force_lsn(struct xfs_mount *mp, xfs_lsn_t lsn, uint flags,
+int	  xfs_log_force_seq(struct xfs_mount *mp, xfs_csn_t seq, uint flags,
 		int *log_forced);
 int	  xfs_log_mount(struct xfs_mount	*mp,
 			struct xfs_buftarg	*log_target,
@@ -117,7 +117,6 @@ void	xfs_log_mount_cancel(struct xfs_mount *);
 xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
 xfs_lsn_t xlog_assign_tail_lsn_locked(struct xfs_mount *mp);
 void	  xfs_log_space_wake(struct xfs_mount *mp);
-void	  xfs_log_release_iclog(struct xlog_in_core *iclog);
 int	  xfs_log_reserve(struct xfs_mount *mp,
 			  int		   length,
 			  int		   count,
@@ -132,8 +131,6 @@ bool	xfs_log_writable(struct xfs_mount *mp);
 struct xlog_ticket *xfs_log_ticket_get(struct xlog_ticket *ticket);
 void	  xfs_log_ticket_put(struct xlog_ticket *ticket);
 
-void	xfs_log_commit_cil(struct xfs_mount *mp, struct xfs_trans *tp,
-				xfs_lsn_t *commit_lsn, bool regrant);
 void	xlog_cil_process_committed(struct list_head *list);
 bool	xfs_log_item_in_current_chkpt(struct xfs_log_item *lip);
 
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
index b0ef071b3cb5..b128aaa9b870 100644
--- a/fs/xfs/xfs_log_cil.c
+++ b/fs/xfs/xfs_log_cil.c
@@ -656,6 +656,8 @@ xlog_cil_push_work(
 	struct xfs_log_vec	lvhdr = { NULL };
 	xfs_lsn_t		commit_lsn;
 	xfs_lsn_t		push_seq;
+	struct bio		bio;
+	DECLARE_COMPLETION_ONSTACK(bdev_flush);
 
 	new_ctx = kmem_zalloc(sizeof(*new_ctx), KM_NOFS);
 	new_ctx->ticket = xlog_cil_ticket_alloc(log);
@@ -668,9 +670,14 @@ xlog_cil_push_work(
 	ASSERT(push_seq <= ctx->sequence);
 
 	/*
-	 * Wake up any background push waiters now this context is being pushed.
+	 * As we are about to switch to a new, empty CIL context, we no longer
+	 * need to throttle tasks on CIL space overruns. Wake any waiters that
+	 * the hard push throttle may have caught so they can start committing
+	 * to the new context. The ctx->xc_push_lock provides the serialisation
+	 * necessary for safely using the lockless waitqueue_active() check in
+	 * this context.
 	 */
-	if (ctx->space_used >= XLOG_CIL_BLOCKING_SPACE_LIMIT(log))
+	if (waitqueue_active(&cil->xc_push_wait))
 		wake_up_all(&cil->xc_push_wait);
 
 	/*
@@ -719,10 +726,19 @@ xlog_cil_push_work(
 	spin_unlock(&cil->xc_push_lock);
 
 	/*
-	 * pull all the log vectors off the items in the CIL, and
-	 * remove the items from the CIL. We don't need the CIL lock
-	 * here because it's only needed on the transaction commit
-	 * side which is currently locked out by the flush lock.
+	 * The CIL is stable at this point - nothing new will be added to it
+	 * because we hold the flush lock exclusively. Hence we can now issue
+	 * a cache flush to ensure all the completed metadata in the journal we
+	 * are about to overwrite is on stable storage.
+	 */
+	xfs_flush_bdev_async(&bio, log->l_mp->m_ddev_targp->bt_bdev,
+				&bdev_flush);
+
+	/*
+	 * Pull all the log vectors off the items in the CIL, and remove the
+	 * items from the CIL. We don't need the CIL lock here because it's only
+	 * needed on the transaction commit side which is currently locked out
+	 * by the flush lock.
 	 */
 	lv = NULL;
 	num_iovecs = 0;
@@ -772,7 +788,7 @@ xlog_cil_push_work(
 	 * that higher sequences will wait for us to write out a commit record
 	 * before they do.
 	 *
-	 * xfs_log_force_lsn requires us to mirror the new sequence into the cil
+	 * xfs_log_force_seq requires us to mirror the new sequence into the cil
 	 * structure atomically with the addition of this sequence to the
 	 * committing list. This also ensures that we can do unlocked checks
 	 * against the current sequence in log forces without risking
@@ -806,7 +822,14 @@ xlog_cil_push_work(
 	lvhdr.lv_iovecp = &lhdr;
 	lvhdr.lv_next = ctx->lv_chain;
 
-	error = xlog_write(log, &lvhdr, tic, &ctx->start_lsn, NULL, 0, true);
+	/*
+	 * Before we format and submit the first iclog, we have to ensure that
+	 * the metadata writeback ordering cache flush is complete.
+	 */
+	wait_for_completion(&bdev_flush);
+
+	error = xlog_write(log, &lvhdr, tic, &ctx->start_lsn, NULL,
+				XLOG_START_TRANS);
 	if (error)
 		goto out_abort_free_ticket;
 
@@ -850,15 +873,21 @@ restart:
 
 	xfs_log_ticket_ungrant(log, tic);
 
-	spin_lock(&commit_iclog->ic_callback_lock);
+	/*
+	 * Once we attach the ctx to the iclog, a shutdown can process the
+	 * iclog, run the callbacks and free the ctx. The only thing preventing
+	 * this potential UAF situation here is that we are holding the
+	 * icloglock. Hence we cannot access the ctx once we have attached the
+	 * callbacks and dropped the icloglock.
+	 */
+	spin_lock(&log->l_icloglock);
 	if (commit_iclog->ic_state == XLOG_STATE_IOERROR) {
-		spin_unlock(&commit_iclog->ic_callback_lock);
+		spin_unlock(&log->l_icloglock);
 		goto out_abort;
 	}
 	ASSERT_ALWAYS(commit_iclog->ic_state == XLOG_STATE_ACTIVE ||
 		      commit_iclog->ic_state == XLOG_STATE_WANT_SYNC);
 	list_add_tail(&ctx->iclog_entry, &commit_iclog->ic_callbacks);
-	spin_unlock(&commit_iclog->ic_callback_lock);
 
 	/*
 	 * now the checkpoint commit is complete and we've attached the
@@ -870,8 +899,50 @@ restart:
 	wake_up_all(&cil->xc_commit_wait);
 	spin_unlock(&cil->xc_push_lock);
 
-	/* release the hounds! */
-	xfs_log_release_iclog(commit_iclog);
+	/*
+	 * If the checkpoint spans multiple iclogs, wait for all previous iclogs
+	 * to complete before we submit the commit_iclog. We can't use state
+	 * checks for this - ACTIVE can be either a past completed iclog or a
+	 * future iclog being filled, while WANT_SYNC through SYNC_DONE can be a
+	 * past or future iclog awaiting IO or ordered IO completion to be run.
+	 * In the latter case, if it's a future iclog and we wait on it, the we
+	 * will hang because it won't get processed through to ic_force_wait
+	 * wakeup until this commit_iclog is written to disk.  Hence we use the
+	 * iclog header lsn and compare it to the commit lsn to determine if we
+	 * need to wait on iclogs or not.
+	 *
+	 * NOTE: It is not safe to reference the ctx after this check as we drop
+	 * the icloglock if we have to wait for completion of other iclogs.
+	 */
+	if (ctx->start_lsn != commit_lsn) {
+		xfs_lsn_t	plsn;
+
+		plsn = be64_to_cpu(commit_iclog->ic_prev->ic_header.h_lsn);
+		if (plsn && XFS_LSN_CMP(plsn, commit_lsn) < 0) {
+			/*
+			 * Waiting on ic_force_wait orders the completion of
+			 * iclogs older than ic_prev. Hence we only need to wait
+			 * on the most recent older iclog here.
+			 */
+			xlog_wait_on_iclog(commit_iclog->ic_prev);
+			spin_lock(&log->l_icloglock);
+		}
+
+		/*
+		 * We need to issue a pre-flush so that the ordering for this
+		 * checkpoint is correctly preserved down to stable storage.
+		 */
+		commit_iclog->ic_flags |= XLOG_ICL_NEED_FLUSH;
+	}
+
+	/*
+	 * The commit iclog must be written to stable storage to guarantee
+	 * journal IO vs metadata writeback IO is correctly ordered on stable
+	 * storage.
+	 */
+	commit_iclog->ic_flags |= XLOG_ICL_NEED_FUA;
+	xlog_state_release_iclog(log, commit_iclog);
+	spin_unlock(&log->l_icloglock);
 	return;
 
 out_skip:
@@ -907,7 +978,7 @@ xlog_cil_push_background(
 	ASSERT(!list_empty(&cil->xc_cil));
 
 	/*
-	 * don't do a background push if we haven't used up all the
+	 * Don't do a background push if we haven't used up all the
 	 * space available yet.
 	 */
 	if (cil->xc_ctx->space_used < XLOG_CIL_SPACE_LIMIT(log)) {
@@ -931,9 +1002,16 @@ xlog_cil_push_background(
 
 	/*
 	 * If we are well over the space limit, throttle the work that is being
-	 * done until the push work on this context has begun.
+	 * done until the push work on this context has begun. Enforce the hard
+	 * throttle on all transaction commits once it has been activated, even
+	 * if the committing transactions have resulted in the space usage
+	 * dipping back down under the hard limit.
+	 *
+	 * The ctx->xc_push_lock provides the serialisation necessary for safely
+	 * using the lockless waitqueue_active() check in this context.
 	 */
-	if (cil->xc_ctx->space_used >= XLOG_CIL_BLOCKING_SPACE_LIMIT(log)) {
+	if (cil->xc_ctx->space_used >= XLOG_CIL_BLOCKING_SPACE_LIMIT(log) ||
+	    waitqueue_active(&cil->xc_push_wait)) {
 		trace_xfs_log_cil_wait(log, cil->xc_ctx->ticket);
 		ASSERT(cil->xc_ctx->space_used < log->l_logsize);
 		xlog_wait(&cil->xc_push_wait, &cil->xc_push_lock);
@@ -1008,16 +1086,14 @@ xlog_cil_empty(
  * allowed again.
  */
 void
-xfs_log_commit_cil(
-	struct xfs_mount	*mp,
+xlog_cil_commit(
+	struct xlog		*log,
 	struct xfs_trans	*tp,
-	xfs_lsn_t		*commit_lsn,
+	xfs_csn_t		*commit_seq,
 	bool			regrant)
 {
-	struct xlog		*log = mp->m_log;
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_log_item	*lip, *next;
-	xfs_lsn_t		xc_commit_lsn;
 
 	/*
 	 * Do all necessary memory allocation before we lock the CIL.
@@ -1031,10 +1107,6 @@ xfs_log_commit_cil(
 
 	xlog_cil_insert_items(log, tp);
 
-	xc_commit_lsn = cil->xc_ctx->sequence;
-	if (commit_lsn)
-		*commit_lsn = xc_commit_lsn;
-
 	if (regrant && !XLOG_FORCED_SHUTDOWN(log))
 		xfs_log_ticket_regrant(log, tp->t_ticket);
 	else
@@ -1057,8 +1129,10 @@ xfs_log_commit_cil(
 	list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
 		xfs_trans_del_item(lip);
 		if (lip->li_ops->iop_committing)
-			lip->li_ops->iop_committing(lip, xc_commit_lsn);
+			lip->li_ops->iop_committing(lip, cil->xc_ctx->sequence);
 	}
+	if (commit_seq)
+		*commit_seq = cil->xc_ctx->sequence;
 
 	/* xlog_cil_push_background() releases cil->xc_ctx_lock */
 	xlog_cil_push_background(log);
@@ -1075,9 +1149,9 @@ xfs_log_commit_cil(
  * iclog flush is necessary following this call.
  */
 xfs_lsn_t
-xlog_cil_force_lsn(
+xlog_cil_force_seq(
 	struct xlog	*log,
-	xfs_lsn_t	sequence)
+	xfs_csn_t	sequence)
 {
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_cil_ctx	*ctx;
@@ -1173,21 +1247,17 @@ bool
 xfs_log_item_in_current_chkpt(
 	struct xfs_log_item *lip)
 {
-	struct xfs_cil_ctx *ctx;
+	struct xfs_cil_ctx *ctx = lip->li_mountp->m_log->l_cilp->xc_ctx;
 
 	if (list_empty(&lip->li_cil))
 		return false;
 
-	ctx = lip->li_mountp->m_log->l_cilp->xc_ctx;
-
 	/*
 	 * li_seq is written on the first commit of a log item to record the
 	 * first checkpoint it is written to. Hence if it is different to the
 	 * current sequence, we're in a new checkpoint.
 	 */
-	if (XFS_LSN_CMP(lip->li_seq, ctx->sequence) != 0)
-		return false;
-	return true;
+	return lip->li_seq == ctx->sequence;
 }
 
 /*
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 1c6fdbf3d506..4c41bbfa33b0 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -50,6 +50,16 @@ enum xlog_iclog_state {
 	XLOG_STATE_IOERROR,	/* IO error happened in sync'ing log */
 };
 
+#define XLOG_STATE_STRINGS \
+	{ XLOG_STATE_ACTIVE,	"XLOG_STATE_ACTIVE" }, \
+	{ XLOG_STATE_WANT_SYNC,	"XLOG_STATE_WANT_SYNC" }, \
+	{ XLOG_STATE_SYNCING,	"XLOG_STATE_SYNCING" }, \
+	{ XLOG_STATE_DONE_SYNC,	"XLOG_STATE_DONE_SYNC" }, \
+	{ XLOG_STATE_CALLBACK,	"XLOG_STATE_CALLBACK" }, \
+	{ XLOG_STATE_DIRTY,	"XLOG_STATE_DIRTY" }, \
+	{ XLOG_STATE_IOERROR,	"XLOG_STATE_IOERROR" }
+
+
 /*
  * Log ticket flags
  */
@@ -133,6 +143,9 @@ enum xlog_iclog_state {
 
 #define XLOG_COVER_OPS		5
 
+#define XLOG_ICL_NEED_FLUSH	(1 << 0)	/* iclog needs REQ_PREFLUSH */
+#define XLOG_ICL_NEED_FUA	(1 << 1)	/* iclog needs REQ_FUA */
+
 /* Ticket reservation region accounting */ 
 #define XLOG_TIC_LEN_MAX	15
 
@@ -201,10 +214,8 @@ typedef struct xlog_in_core {
 	u32			ic_size;
 	u32			ic_offset;
 	enum xlog_iclog_state	ic_state;
+	unsigned int		ic_flags;
 	char			*ic_datap;	/* pointer to iclog data */
-
-	/* Callback structures need their own cacheline */
-	spinlock_t		ic_callback_lock ____cacheline_aligned_in_smp;
 	struct list_head	ic_callbacks;
 
 	/* reference counts need their own cacheline */
@@ -230,7 +241,7 @@ struct xfs_cil;
 
 struct xfs_cil_ctx {
 	struct xfs_cil		*cil;
-	xfs_lsn_t		sequence;	/* chkpt sequence # */
+	xfs_csn_t		sequence;	/* chkpt sequence # */
 	xfs_lsn_t		start_lsn;	/* first LSN of chkpt commit */
 	xfs_lsn_t		commit_lsn;	/* chkpt commit record lsn */
 	struct xlog_ticket	*ticket;	/* chkpt ticket */
@@ -268,10 +279,10 @@ struct xfs_cil {
 	struct xfs_cil_ctx	*xc_ctx;
 
 	spinlock_t		xc_push_lock ____cacheline_aligned_in_smp;
-	xfs_lsn_t		xc_push_seq;
+	xfs_csn_t		xc_push_seq;
 	struct list_head	xc_committing;
 	wait_queue_head_t	xc_commit_wait;
-	xfs_lsn_t		xc_current_sequence;
+	xfs_csn_t		xc_current_sequence;
 	struct work_struct	xc_push_work;
 	wait_queue_head_t	xc_push_wait;	/* background push throttle */
 } ____cacheline_aligned_in_smp;
@@ -436,6 +447,8 @@ struct xlog {
 #endif
 	/* log recovery lsn tracking (for buffer submission */
 	xfs_lsn_t		l_recovery_lsn;
+
+	uint32_t		l_iclog_roundoff;/* padding roundoff */
 };
 
 #define XLOG_BUF_CANCEL_BUCKET(log, blkno) \
@@ -478,13 +491,14 @@ void	xlog_print_tic_res(struct xfs_mount *mp, struct xlog_ticket *ticket);
 void	xlog_print_trans(struct xfs_trans *);
 int	xlog_write(struct xlog *log, struct xfs_log_vec *log_vector,
 		struct xlog_ticket *tic, xfs_lsn_t *start_lsn,
-		struct xlog_in_core **commit_iclog, uint flags,
-		bool need_start_rec);
+		struct xlog_in_core **commit_iclog, uint optype);
 int	xlog_commit_record(struct xlog *log, struct xlog_ticket *ticket,
 		struct xlog_in_core **iclog, xfs_lsn_t *lsn);
 void	xfs_log_ticket_ungrant(struct xlog *log, struct xlog_ticket *ticket);
 void	xfs_log_ticket_regrant(struct xlog *log, struct xlog_ticket *ticket);
 
+int xlog_state_release_iclog(struct xlog *log, struct xlog_in_core *iclog);
+
 /*
  * When we crack an atomic LSN, we sample it first so that the value will not
  * change while we are cracking it into the component values. This means we
@@ -547,19 +561,18 @@ int	xlog_cil_init(struct xlog *log);
 void	xlog_cil_init_post_recovery(struct xlog *log);
 void	xlog_cil_destroy(struct xlog *log);
 bool	xlog_cil_empty(struct xlog *log);
+void	xlog_cil_commit(struct xlog *log, struct xfs_trans *tp,
+			xfs_csn_t *commit_seq, bool regrant);
 
 /*
  * CIL force routines
  */
-xfs_lsn_t
-xlog_cil_force_lsn(
-	struct xlog *log,
-	xfs_lsn_t sequence);
+xfs_lsn_t xlog_cil_force_seq(struct xlog *log, xfs_csn_t sequence);
 
 static inline void
 xlog_cil_force(struct xlog *log)
 {
-	xlog_cil_force_lsn(log, log->l_cilp->xc_current_sequence);
+	xlog_cil_force_seq(log, log->l_cilp->xc_current_sequence);
 }
 
 /*
@@ -582,6 +595,8 @@ xlog_wait(
 	remove_wait_queue(wq, &wait);
 }
 
+int xlog_wait_on_iclog(struct xlog_in_core *iclog);
+
 /*
  * The LSN is valid so long as it is behind the current LSN. If it isn't, this
  * means that the next log record that includes this metadata could have a
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index e5dd1c0c2f03..1721fce2ec94 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -25,6 +25,7 @@
 #include "xfs_icache.h"
 #include "xfs_error.h"
 #include "xfs_buf_item.h"
+#include "xfs_ag.h"
 
 #define BLK_AVG(blk1, blk2)	((blk1+blk2) >> 1)
 
@@ -2457,8 +2458,10 @@ xlog_finish_defer_ops(
 
 		error = xfs_trans_alloc(mp, &resv, dfc->dfc_blkres,
 				dfc->dfc_rtxres, XFS_TRANS_RESERVE, &tp);
-		if (error)
+		if (error) {
+			xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
 			return error;
+		}
 
 		/*
 		 * Transfer to this new transaction all the dfops we captured
@@ -2741,21 +2744,17 @@ STATIC void
 xlog_recover_process_iunlinks(
 	struct xlog	*log)
 {
-	xfs_mount_t	*mp;
-	xfs_agnumber_t	agno;
-	xfs_agi_t	*agi;
-	struct xfs_buf	*agibp;
-	xfs_agino_t	agino;
-	int		bucket;
-	int		error;
-
-	mp = log->l_mp;
+	struct xfs_mount	*mp = log->l_mp;
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		agno;
+	struct xfs_agi		*agi;
+	struct xfs_buf		*agibp;
+	xfs_agino_t		agino;
+	int			bucket;
+	int			error;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		/*
-		 * Find the agi for this ag.
-		 */
-		error = xfs_read_agi(mp, NULL, agno, &agibp);
+	for_each_perag(mp, agno, pag) {
+		error = xfs_read_agi(mp, NULL, pag->pag_agno, &agibp);
 		if (error) {
 			/*
 			 * AGI is b0rked. Don't process it.
@@ -2781,7 +2780,7 @@ xlog_recover_process_iunlinks(
 			agino = be32_to_cpu(agi->agi_unlinked[bucket]);
 			while (agino != NULLAGINO) {
 				agino = xlog_recover_process_one_iunlink(mp,
-							agno, agino, bucket);
+						pag->pag_agno, agino, bucket);
 				cond_resched();
 			}
 		}
@@ -3452,6 +3451,7 @@ xlog_recover_finish(
 			 * this) before we get around to xfs_log_mount_cancel.
 			 */
 			xlog_recover_cancel_intents(log);
+			xfs_force_shutdown(log->l_mp, SHUTDOWN_LOG_IO_ERROR);
 			xfs_alert(log->l_mp, "Failed to recover intents");
 			return error;
 		}
@@ -3493,27 +3493,28 @@ xlog_recover_cancel(
  */
 STATIC void
 xlog_recover_check_summary(
-	struct xlog	*log)
+	struct xlog		*log)
 {
-	xfs_mount_t	*mp;
-	struct xfs_buf	*agfbp;
-	struct xfs_buf	*agibp;
-	xfs_agnumber_t	agno;
-	uint64_t	freeblks;
-	uint64_t	itotal;
-	uint64_t	ifree;
-	int		error;
+	struct xfs_mount	*mp = log->l_mp;
+	struct xfs_perag	*pag;
+	struct xfs_buf		*agfbp;
+	struct xfs_buf		*agibp;
+	xfs_agnumber_t		agno;
+	uint64_t		freeblks;
+	uint64_t		itotal;
+	uint64_t		ifree;
+	int			error;
 
 	mp = log->l_mp;
 
 	freeblks = 0LL;
 	itotal = 0LL;
 	ifree = 0LL;
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		error = xfs_read_agf(mp, NULL, agno, 0, &agfbp);
+	for_each_perag(mp, agno, pag) {
+		error = xfs_read_agf(mp, NULL, pag->pag_agno, 0, &agfbp);
 		if (error) {
 			xfs_alert(mp, "%s agf read failed agno %d error %d",
-						__func__, agno, error);
+						__func__, pag->pag_agno, error);
 		} else {
 			struct xfs_agf	*agfp = agfbp->b_addr;
 
@@ -3522,10 +3523,10 @@ xlog_recover_check_summary(
 			xfs_buf_relse(agfbp);
 		}
 
-		error = xfs_read_agi(mp, NULL, agno, &agibp);
+		error = xfs_read_agi(mp, NULL, pag->pag_agno, &agibp);
 		if (error) {
 			xfs_alert(mp, "%s agi read failed agno %d error %d",
-						__func__, agno, error);
+						__func__, pag->pag_agno, error);
 		} else {
 			struct xfs_agi	*agi = agibp->b_addr;
 
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index bdfee1943796..d0755494597f 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -32,6 +32,7 @@
 #include "xfs_extent_busy.h"
 #include "xfs_health.h"
 #include "xfs_trace.h"
+#include "xfs_ag.h"
 
 static DEFINE_MUTEX(xfs_uuid_table_mutex);
 static int xfs_uuid_table_size;
@@ -119,41 +120,6 @@ xfs_uuid_unmount(
 	mutex_unlock(&xfs_uuid_table_mutex);
 }
 
-
-STATIC void
-__xfs_free_perag(
-	struct rcu_head	*head)
-{
-	struct xfs_perag *pag = container_of(head, struct xfs_perag, rcu_head);
-
-	ASSERT(!delayed_work_pending(&pag->pag_blockgc_work));
-	ASSERT(atomic_read(&pag->pag_ref) == 0);
-	kmem_free(pag);
-}
-
-/*
- * Free up the per-ag resources associated with the mount structure.
- */
-STATIC void
-xfs_free_perag(
-	xfs_mount_t	*mp)
-{
-	xfs_agnumber_t	agno;
-	struct xfs_perag *pag;
-
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		spin_lock(&mp->m_perag_lock);
-		pag = radix_tree_delete(&mp->m_perag_tree, agno);
-		spin_unlock(&mp->m_perag_lock);
-		ASSERT(pag);
-		ASSERT(atomic_read(&pag->pag_ref) == 0);
-		cancel_delayed_work_sync(&pag->pag_blockgc_work);
-		xfs_iunlink_destroy(pag);
-		xfs_buf_hash_destroy(pag);
-		call_rcu(&pag->rcu_head, __xfs_free_perag);
-	}
-}
-
 /*
  * Check size of device based on the (data/realtime) block count.
  * Note: this check is used by the growfs code as well as mount.
@@ -172,96 +138,6 @@ xfs_sb_validate_fsb_count(
 	return 0;
 }
 
-int
-xfs_initialize_perag(
-	xfs_mount_t	*mp,
-	xfs_agnumber_t	agcount,
-	xfs_agnumber_t	*maxagi)
-{
-	xfs_agnumber_t	index;
-	xfs_agnumber_t	first_initialised = NULLAGNUMBER;
-	xfs_perag_t	*pag;
-	int		error = -ENOMEM;
-
-	/*
-	 * Walk the current per-ag tree so we don't try to initialise AGs
-	 * that already exist (growfs case). Allocate and insert all the
-	 * AGs we don't find ready for initialisation.
-	 */
-	for (index = 0; index < agcount; index++) {
-		pag = xfs_perag_get(mp, index);
-		if (pag) {
-			xfs_perag_put(pag);
-			continue;
-		}
-
-		pag = kmem_zalloc(sizeof(*pag), KM_MAYFAIL);
-		if (!pag) {
-			error = -ENOMEM;
-			goto out_unwind_new_pags;
-		}
-		pag->pag_agno = index;
-		pag->pag_mount = mp;
-		spin_lock_init(&pag->pag_ici_lock);
-		INIT_DELAYED_WORK(&pag->pag_blockgc_work, xfs_blockgc_worker);
-		INIT_RADIX_TREE(&pag->pag_ici_root, GFP_ATOMIC);
-
-		error = xfs_buf_hash_init(pag);
-		if (error)
-			goto out_free_pag;
-		init_waitqueue_head(&pag->pagb_wait);
-		spin_lock_init(&pag->pagb_lock);
-		pag->pagb_count = 0;
-		pag->pagb_tree = RB_ROOT;
-
-		error = radix_tree_preload(GFP_NOFS);
-		if (error)
-			goto out_hash_destroy;
-
-		spin_lock(&mp->m_perag_lock);
-		if (radix_tree_insert(&mp->m_perag_tree, index, pag)) {
-			WARN_ON_ONCE(1);
-			spin_unlock(&mp->m_perag_lock);
-			radix_tree_preload_end();
-			error = -EEXIST;
-			goto out_hash_destroy;
-		}
-		spin_unlock(&mp->m_perag_lock);
-		radix_tree_preload_end();
-		/* first new pag is fully initialized */
-		if (first_initialised == NULLAGNUMBER)
-			first_initialised = index;
-		error = xfs_iunlink_init(pag);
-		if (error)
-			goto out_hash_destroy;
-		spin_lock_init(&pag->pag_state_lock);
-	}
-
-	index = xfs_set_inode_alloc(mp, agcount);
-
-	if (maxagi)
-		*maxagi = index;
-
-	mp->m_ag_prealloc_blocks = xfs_prealloc_blocks(mp);
-	return 0;
-
-out_hash_destroy:
-	xfs_buf_hash_destroy(pag);
-out_free_pag:
-	kmem_free(pag);
-out_unwind_new_pags:
-	/* unwind any prior newly initialized pags */
-	for (index = first_initialised; index < agcount; index++) {
-		pag = radix_tree_delete(&mp->m_perag_tree, index);
-		if (!pag)
-			break;
-		xfs_buf_hash_destroy(pag);
-		xfs_iunlink_destroy(pag);
-		kmem_free(pag);
-	}
-	return error;
-}
-
 /*
  * xfs_readsb
  *
@@ -983,9 +859,17 @@ xfs_mountfs(
 	/*
 	 * Finish recovering the file system.  This part needed to be delayed
 	 * until after the root and real-time bitmap inodes were consistently
-	 * read in.
+	 * read in.  Temporarily create per-AG space reservations for metadata
+	 * btree shape changes because space freeing transactions (for inode
+	 * inactivation) require the per-AG reservation in lieu of reserving
+	 * blocks.
 	 */
+	error = xfs_fs_reserve_ag_blocks(mp);
+	if (error && error == -ENOSPC)
+		xfs_warn(mp,
+	"ENOSPC reserving per-AG metadata pool, log recovery may fail.");
 	error = xfs_log_mount_finish(mp);
+	xfs_fs_unreserve_ag_blocks(mp);
 	if (error) {
 		xfs_warn(mp, "log mount finish failed");
 		goto out_rtunmount;
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index bb67274ee23f..c78b63fe779a 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -12,6 +12,7 @@ struct xfs_mru_cache;
 struct xfs_ail;
 struct xfs_quotainfo;
 struct xfs_da_geometry;
+struct xfs_perag;
 
 /* dynamic preallocation free space thresholds, 5% down to 1% */
 enum {
@@ -297,117 +298,12 @@ xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
 	return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
 }
 
-/* per-AG block reservation data structures*/
-struct xfs_ag_resv {
-	/* number of blocks originally reserved here */
-	xfs_extlen_t			ar_orig_reserved;
-	/* number of blocks reserved here */
-	xfs_extlen_t			ar_reserved;
-	/* number of blocks originally asked for */
-	xfs_extlen_t			ar_asked;
-};
-
-/*
- * Per-ag incore structure, copies of information in agf and agi, to improve the
- * performance of allocation group selection.
- */
-typedef struct xfs_perag {
-	struct xfs_mount *pag_mount;	/* owner filesystem */
-	xfs_agnumber_t	pag_agno;	/* AG this structure belongs to */
-	atomic_t	pag_ref;	/* perag reference count */
-	char		pagf_init;	/* this agf's entry is initialized */
-	char		pagi_init;	/* this agi's entry is initialized */
-	char		pagf_metadata;	/* the agf is preferred to be metadata */
-	char		pagi_inodeok;	/* The agi is ok for inodes */
-	uint8_t		pagf_levels[XFS_BTNUM_AGF];
-					/* # of levels in bno & cnt btree */
-	bool		pagf_agflreset; /* agfl requires reset before use */
-	uint32_t	pagf_flcount;	/* count of blocks in freelist */
-	xfs_extlen_t	pagf_freeblks;	/* total free blocks */
-	xfs_extlen_t	pagf_longest;	/* longest free space */
-	uint32_t	pagf_btreeblks;	/* # of blocks held in AGF btrees */
-	xfs_agino_t	pagi_freecount;	/* number of free inodes */
-	xfs_agino_t	pagi_count;	/* number of allocated inodes */
-
-	/*
-	 * Inode allocation search lookup optimisation.
-	 * If the pagino matches, the search for new inodes
-	 * doesn't need to search the near ones again straight away
-	 */
-	xfs_agino_t	pagl_pagino;
-	xfs_agino_t	pagl_leftrec;
-	xfs_agino_t	pagl_rightrec;
-
-	/*
-	 * Bitsets of per-ag metadata that have been checked and/or are sick.
-	 * Callers should hold pag_state_lock before accessing this field.
-	 */
-	uint16_t	pag_checked;
-	uint16_t	pag_sick;
-	spinlock_t	pag_state_lock;
-
-	spinlock_t	pagb_lock;	/* lock for pagb_tree */
-	struct rb_root	pagb_tree;	/* ordered tree of busy extents */
-	unsigned int	pagb_gen;	/* generation count for pagb_tree */
-	wait_queue_head_t pagb_wait;	/* woken when pagb_gen changes */
-
-	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
-
-	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
-	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
-	int		pag_ici_reclaimable;	/* reclaimable inodes */
-	unsigned long	pag_ici_reclaim_cursor;	/* reclaim restart point */
-
-	/* buffer cache index */
-	spinlock_t	pag_buf_lock;	/* lock for pag_buf_hash */
-	struct rhashtable pag_buf_hash;
-
-	/* for rcu-safe freeing */
-	struct rcu_head	rcu_head;
-	int		pagb_count;	/* pagb slots in use */
-
-	/* Blocks reserved for all kinds of metadata. */
-	struct xfs_ag_resv	pag_meta_resv;
-	/* Blocks reserved for the reverse mapping btree. */
-	struct xfs_ag_resv	pag_rmapbt_resv;
-
-	/* background prealloc block trimming */
-	struct delayed_work	pag_blockgc_work;
-
-	/* reference count */
-	uint8_t			pagf_refcount_level;
-
-	/*
-	 * Unlinked inode information.  This incore information reflects
-	 * data stored in the AGI, so callers must hold the AGI buffer lock
-	 * or have some other means to control concurrency.
-	 */
-	struct rhashtable	pagi_unlinked_hash;
-} xfs_perag_t;
-
-static inline struct xfs_ag_resv *
-xfs_perag_resv(
-	struct xfs_perag	*pag,
-	enum xfs_ag_resv_type	type)
-{
-	switch (type) {
-	case XFS_AG_RESV_METADATA:
-		return &pag->pag_meta_resv;
-	case XFS_AG_RESV_RMAPBT:
-		return &pag->pag_rmapbt_resv;
-	default:
-		return NULL;
-	}
-}
-
-int xfs_buf_hash_init(xfs_perag_t *pag);
-void xfs_buf_hash_destroy(xfs_perag_t *pag);
+int xfs_buf_hash_init(struct xfs_perag *pag);
+void xfs_buf_hash_destroy(struct xfs_perag *pag);
 
 extern void	xfs_uuid_table_free(void);
 extern uint64_t xfs_default_resblks(xfs_mount_t *mp);
 extern int	xfs_mountfs(xfs_mount_t *mp);
-extern int	xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
-				     xfs_agnumber_t *maxagi);
 extern void	xfs_unmountfs(xfs_mount_t *);
 
 extern int	xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
diff --git a/fs/xfs/xfs_qm.c b/fs/xfs/xfs_qm.c
index 4bf949a89d0d..fe341f3fd419 100644
--- a/fs/xfs/xfs_qm.c
+++ b/fs/xfs/xfs_qm.c
@@ -23,6 +23,8 @@
 #include "xfs_trace.h"
 #include "xfs_icache.h"
 #include "xfs_error.h"
+#include "xfs_ag.h"
+#include "xfs_ialloc.h"
 
 /*
  * The global quota manager. There is only one of these for the entire
@@ -787,8 +789,12 @@ xfs_qm_qino_alloc(
 		return error;
 
 	if (need_alloc) {
-		error = xfs_dir_ialloc(&init_user_ns, &tp, NULL, S_IFREG, 1, 0,
-				       0, false, ipp);
+		xfs_ino_t	ino;
+
+		error = xfs_dialloc(&tp, 0, S_IFREG, &ino);
+		if (!error)
+			error = xfs_init_new_inode(&init_user_ns, tp, NULL, ino,
+					S_IFREG, 1, 0, 0, false, ipp);
 		if (error) {
 			xfs_trans_cancel(tp);
 			return error;
diff --git a/fs/xfs/xfs_qm.h b/fs/xfs/xfs_qm.h
index e3dabab44097..ebbb484c49dc 100644
--- a/fs/xfs/xfs_qm.h
+++ b/fs/xfs/xfs_qm.h
@@ -142,7 +142,6 @@ extern void		xfs_qm_destroy_quotainfo(struct xfs_mount *);
 
 /* dquot stuff */
 extern void		xfs_qm_dqpurge_all(struct xfs_mount *, uint);
-extern void		xfs_qm_dqrele_all_inodes(struct xfs_mount *, uint);
 
 /* quota ops */
 extern int		xfs_qm_scall_trunc_qfiles(struct xfs_mount *, uint);
diff --git a/fs/xfs/xfs_qm_syscalls.c b/fs/xfs/xfs_qm_syscalls.c
index 11f1e2fbf22f..13a56e1ea15c 100644
--- a/fs/xfs/xfs_qm_syscalls.c
+++ b/fs/xfs/xfs_qm_syscalls.c
@@ -201,7 +201,8 @@ xfs_qm_scall_quotaoff(
 	 * depend on the quota inodes (and other things) being valid as long as
 	 * we keep the lock(s).
 	 */
-	xfs_qm_dqrele_all_inodes(mp, flags);
+	error = xfs_dqrele_all_inodes(mp, flags);
+	ASSERT(!error);
 
 	/*
 	 * Next we make the changes in the quota flag in the mount struct.
@@ -747,54 +748,3 @@ xfs_qm_scall_getquota_next(
 	xfs_qm_dqput(dqp);
 	return error;
 }
-
-STATIC int
-xfs_dqrele_inode(
-	struct xfs_inode	*ip,
-	void			*args)
-{
-	uint			*flags = args;
-
-	/* skip quota inodes */
-	if (ip == ip->i_mount->m_quotainfo->qi_uquotaip ||
-	    ip == ip->i_mount->m_quotainfo->qi_gquotaip ||
-	    ip == ip->i_mount->m_quotainfo->qi_pquotaip) {
-		ASSERT(ip->i_udquot == NULL);
-		ASSERT(ip->i_gdquot == NULL);
-		ASSERT(ip->i_pdquot == NULL);
-		return 0;
-	}
-
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	if ((*flags & XFS_UQUOTA_ACCT) && ip->i_udquot) {
-		xfs_qm_dqrele(ip->i_udquot);
-		ip->i_udquot = NULL;
-	}
-	if ((*flags & XFS_GQUOTA_ACCT) && ip->i_gdquot) {
-		xfs_qm_dqrele(ip->i_gdquot);
-		ip->i_gdquot = NULL;
-	}
-	if ((*flags & XFS_PQUOTA_ACCT) && ip->i_pdquot) {
-		xfs_qm_dqrele(ip->i_pdquot);
-		ip->i_pdquot = NULL;
-	}
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-	return 0;
-}
-
-
-/*
- * Go thru all the inodes in the file system, releasing their dquots.
- *
- * Note that the mount structure gets modified to indicate that quotas are off
- * AFTER this, in the case of quotaoff.
- */
-void
-xfs_qm_dqrele_all_inodes(
-	struct xfs_mount	*mp,
-	uint			flags)
-{
-	ASSERT(mp->m_quotainfo);
-	xfs_inode_walk(mp, XFS_INODE_WALK_INEW_WAIT, xfs_dqrele_inode,
-			&flags, XFS_ICI_NO_TAG);
-}
diff --git a/fs/xfs/xfs_reflink.c b/fs/xfs/xfs_reflink.c
index 060695d6d56a..c256104772cb 100644
--- a/fs/xfs/xfs_reflink.c
+++ b/fs/xfs/xfs_reflink.c
@@ -27,7 +27,7 @@
 #include "xfs_quota.h"
 #include "xfs_reflink.h"
 #include "xfs_iomap.h"
-#include "xfs_sb.h"
+#include "xfs_ag.h"
 #include "xfs_ag_resv.h"
 
 /*
@@ -144,7 +144,7 @@ xfs_reflink_find_shared(
 	if (error)
 		return error;
 
-	cur = xfs_refcountbt_init_cursor(mp, tp, agbp, agno);
+	cur = xfs_refcountbt_init_cursor(mp, tp, agbp, agbp->b_pag);
 
 	error = xfs_refcount_find_shared(cur, agbno, aglen, fbno, flen,
 			find_end_of_shared);
@@ -755,16 +755,19 @@ int
 xfs_reflink_recover_cow(
 	struct xfs_mount	*mp)
 {
+	struct xfs_perag	*pag;
 	xfs_agnumber_t		agno;
 	int			error = 0;
 
 	if (!xfs_sb_version_hasreflink(&mp->m_sb))
 		return 0;
 
-	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		error = xfs_refcount_recover_cow_leftovers(mp, agno);
-		if (error)
+	for_each_perag(mp, agno, pag) {
+		error = xfs_refcount_recover_cow_leftovers(mp, pag);
+		if (error) {
+			xfs_perag_put(pag);
 			break;
+		}
 	}
 
 	return error;
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index a2dab05332ac..2c9e26a44546 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -36,6 +36,7 @@
 #include "xfs_bmap_item.h"
 #include "xfs_reflink.h"
 #include "xfs_pwork.h"
+#include "xfs_ag.h"
 
 #include <linux/magic.h>
 #include <linux/fs_context.h>
@@ -339,13 +340,6 @@ xfs_blkdev_put(
 		blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
 }
 
-void
-xfs_blkdev_issue_flush(
-	xfs_buftarg_t		*buftarg)
-{
-	blkdev_issue_flush(buftarg->bt_bdev);
-}
-
 STATIC void
 xfs_close_devices(
 	struct xfs_mount	*mp)
@@ -667,7 +661,7 @@ xfs_fs_destroy_inode(
 	 * reclaim path handles this more efficiently than we can here, so
 	 * simply let background reclaim tear down all inodes.
 	 */
-	xfs_inode_set_reclaim_tag(ip);
+	xfs_inode_mark_reclaimable(ip);
 }
 
 static void
diff --git a/fs/xfs/xfs_super.h b/fs/xfs/xfs_super.h
index d2b40dc60dfc..167d23f92ffe 100644
--- a/fs/xfs/xfs_super.h
+++ b/fs/xfs/xfs_super.h
@@ -87,7 +87,6 @@ struct xfs_buftarg;
 struct block_device;
 
 extern void xfs_flush_inodes(struct xfs_mount *mp);
-extern void xfs_blkdev_issue_flush(struct xfs_buftarg *);
 extern xfs_agnumber_t xfs_set_inode_alloc(struct xfs_mount *,
 					   xfs_agnumber_t agcount);
 
diff --git a/fs/xfs/xfs_symlink.c b/fs/xfs/xfs_symlink.c
index 99fbec32c10a..1525636f4065 100644
--- a/fs/xfs/xfs_symlink.c
+++ b/fs/xfs/xfs_symlink.c
@@ -21,6 +21,7 @@
 #include "xfs_trans_space.h"
 #include "xfs_trace.h"
 #include "xfs_trans.h"
+#include "xfs_ialloc.h"
 
 /* ----- Kernel only functions below ----- */
 int
@@ -161,6 +162,7 @@ xfs_symlink(
 	struct xfs_dquot	*gdqp = NULL;
 	struct xfs_dquot	*pdqp = NULL;
 	uint			resblks;
+	xfs_ino_t		ino;
 
 	*ipp = NULL;
 
@@ -223,8 +225,11 @@ xfs_symlink(
 	/*
 	 * Allocate an inode for the symlink.
 	 */
-	error = xfs_dir_ialloc(mnt_userns, &tp, dp, S_IFLNK | (mode & ~S_IFMT),
-			       1, 0, prid, false, &ip);
+	error = xfs_dialloc(&tp, dp->i_ino, S_IFLNK, &ino);
+	if (!error)
+		error = xfs_init_new_inode(mnt_userns, tp, dp, ino,
+				S_IFLNK | (mode & ~S_IFMT), 1, 0, prid,
+				false, &ip);
 	if (error)
 		goto out_trans_cancel;
 
diff --git a/fs/xfs/xfs_trace.c b/fs/xfs/xfs_trace.c
index 9b8d703dc9fd..7e01e00550ac 100644
--- a/fs/xfs/xfs_trace.c
+++ b/fs/xfs/xfs_trace.c
@@ -30,6 +30,8 @@
 #include "xfs_fsmap.h"
 #include "xfs_btree_staging.h"
 #include "xfs_icache.h"
+#include "xfs_ag.h"
+#include "xfs_ag_resv.h"
 
 /*
  * We include this last to have the helpers above available for the trace
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 808ae337b222..f9d8d605f9b1 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -24,6 +24,7 @@ struct xlog_ticket;
 struct xlog_recover;
 struct xlog_recover_item;
 struct xlog_rec_header;
+struct xlog_in_core;
 struct xfs_buf_log_format;
 struct xfs_inode_log_format;
 struct xfs_bmbt_irec;
@@ -37,7 +38,7 @@ struct xfs_trans_res;
 struct xfs_inobt_rec_incore;
 union xfs_btree_ptr;
 struct xfs_dqtrx;
-struct xfs_eofblocks;
+struct xfs_icwalk;
 
 #define XFS_ATTR_FILTER_FLAGS \
 	{ XFS_ATTR_ROOT,	"ROOT" }, \
@@ -153,10 +154,8 @@ DEFINE_EVENT(xfs_perag_class, name,	\
 DEFINE_PERAG_REF_EVENT(xfs_perag_get);
 DEFINE_PERAG_REF_EVENT(xfs_perag_get_tag);
 DEFINE_PERAG_REF_EVENT(xfs_perag_put);
-DEFINE_PERAG_REF_EVENT(xfs_perag_set_reclaim);
-DEFINE_PERAG_REF_EVENT(xfs_perag_clear_reclaim);
-DEFINE_PERAG_REF_EVENT(xfs_perag_set_blockgc);
-DEFINE_PERAG_REF_EVENT(xfs_perag_clear_blockgc);
+DEFINE_PERAG_REF_EVENT(xfs_perag_set_inode_tag);
+DEFINE_PERAG_REF_EVENT(xfs_perag_clear_inode_tag);
 
 DECLARE_EVENT_CLASS(xfs_ag_class,
 	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno),
@@ -632,8 +631,8 @@ DEFINE_EVENT(xfs_inode_class, name, \
 	TP_PROTO(struct xfs_inode *ip), \
 	TP_ARGS(ip))
 DEFINE_INODE_EVENT(xfs_iget_skip);
-DEFINE_INODE_EVENT(xfs_iget_reclaim);
-DEFINE_INODE_EVENT(xfs_iget_reclaim_fail);
+DEFINE_INODE_EVENT(xfs_iget_recycle);
+DEFINE_INODE_EVENT(xfs_iget_recycle_fail);
 DEFINE_INODE_EVENT(xfs_iget_hit);
 DEFINE_INODE_EVENT(xfs_iget_miss);
 
@@ -1914,7 +1913,6 @@ DEFINE_ATTR_EVENT(xfs_attr_leaf_add);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_add_old);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_add_new);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_add_work);
-DEFINE_ATTR_EVENT(xfs_attr_leaf_addname);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_create);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_compact);
 DEFINE_ATTR_EVENT(xfs_attr_leaf_get);
@@ -1944,7 +1942,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate);
 
 DEFINE_ATTR_EVENT(xfs_attr_rmtval_get);
 DEFINE_ATTR_EVENT(xfs_attr_rmtval_set);
-DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove);
 
 #define DEFINE_DA_EVENT(name) \
 DEFINE_EVENT(xfs_da_class, name, \
@@ -3730,7 +3727,7 @@ TRACE_EVENT(xfs_btree_commit_afakeroot,
 	TP_fast_assign(
 		__entry->dev = cur->bc_mp->m_super->s_dev;
 		__entry->btnum = cur->bc_btnum;
-		__entry->agno = cur->bc_ag.agno;
+		__entry->agno = cur->bc_ag.pag->pag_agno;
 		__entry->agbno = cur->bc_ag.afake->af_root;
 		__entry->levels = cur->bc_ag.afake->af_levels;
 		__entry->blocks = cur->bc_ag.afake->af_blocks;
@@ -3845,7 +3842,7 @@ TRACE_EVENT(xfs_btree_bload_block,
 			__entry->agno = XFS_FSB_TO_AGNO(cur->bc_mp, fsb);
 			__entry->agbno = XFS_FSB_TO_AGBNO(cur->bc_mp, fsb);
 		} else {
-			__entry->agno = cur->bc_ag.agno;
+			__entry->agno = cur->bc_ag.pag->pag_agno;
 			__entry->agbno = be32_to_cpu(ptr->s);
 		}
 		__entry->nr_records = nr_records;
@@ -3887,10 +3884,10 @@ DEFINE_EVENT(xfs_timestamp_range_class, name, \
 DEFINE_TIMESTAMP_RANGE_EVENT(xfs_inode_timestamp_range);
 DEFINE_TIMESTAMP_RANGE_EVENT(xfs_quota_expiry_range);
 
-DECLARE_EVENT_CLASS(xfs_eofblocks_class,
-	TP_PROTO(struct xfs_mount *mp, struct xfs_eofblocks *eofb,
+DECLARE_EVENT_CLASS(xfs_icwalk_class,
+	TP_PROTO(struct xfs_mount *mp, struct xfs_icwalk *icw,
 		 unsigned long caller_ip),
-	TP_ARGS(mp, eofb, caller_ip),
+	TP_ARGS(mp, icw, caller_ip),
 	TP_STRUCT__entry(
 		__field(dev_t, dev)
 		__field(__u32, flags)
@@ -3898,35 +3895,97 @@ DECLARE_EVENT_CLASS(xfs_eofblocks_class,
 		__field(uint32_t, gid)
 		__field(prid_t, prid)
 		__field(__u64, min_file_size)
+		__field(long, scan_limit)
 		__field(unsigned long, caller_ip)
 	),
 	TP_fast_assign(
 		__entry->dev = mp->m_super->s_dev;
-		__entry->flags = eofb ? eofb->eof_flags : 0;
-		__entry->uid = eofb ? from_kuid(mp->m_super->s_user_ns,
-						eofb->eof_uid) : 0;
-		__entry->gid = eofb ? from_kgid(mp->m_super->s_user_ns,
-						eofb->eof_gid) : 0;
-		__entry->prid = eofb ? eofb->eof_prid : 0;
-		__entry->min_file_size = eofb ? eofb->eof_min_file_size : 0;
+		__entry->flags = icw ? icw->icw_flags : 0;
+		__entry->uid = icw ? from_kuid(mp->m_super->s_user_ns,
+						icw->icw_uid) : 0;
+		__entry->gid = icw ? from_kgid(mp->m_super->s_user_ns,
+						icw->icw_gid) : 0;
+		__entry->prid = icw ? icw->icw_prid : 0;
+		__entry->min_file_size = icw ? icw->icw_min_file_size : 0;
+		__entry->scan_limit = icw ? icw->icw_scan_limit : 0;
 		__entry->caller_ip = caller_ip;
 	),
-	TP_printk("dev %d:%d flags 0x%x uid %u gid %u prid %u minsize %llu caller %pS",
+	TP_printk("dev %d:%d flags 0x%x uid %u gid %u prid %u minsize %llu scan_limit %ld caller %pS",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->flags,
 		  __entry->uid,
 		  __entry->gid,
 		  __entry->prid,
 		  __entry->min_file_size,
+		  __entry->scan_limit,
 		  (char *)__entry->caller_ip)
 );
-#define DEFINE_EOFBLOCKS_EVENT(name)	\
-DEFINE_EVENT(xfs_eofblocks_class, name,	\
-	TP_PROTO(struct xfs_mount *mp, struct xfs_eofblocks *eofb, \
+#define DEFINE_ICWALK_EVENT(name)	\
+DEFINE_EVENT(xfs_icwalk_class, name,	\
+	TP_PROTO(struct xfs_mount *mp, struct xfs_icwalk *icw, \
 		 unsigned long caller_ip), \
-	TP_ARGS(mp, eofb, caller_ip))
-DEFINE_EOFBLOCKS_EVENT(xfs_ioc_free_eofblocks);
-DEFINE_EOFBLOCKS_EVENT(xfs_blockgc_free_space);
+	TP_ARGS(mp, icw, caller_ip))
+DEFINE_ICWALK_EVENT(xfs_ioc_free_eofblocks);
+DEFINE_ICWALK_EVENT(xfs_blockgc_free_space);
+
+TRACE_DEFINE_ENUM(XLOG_STATE_ACTIVE);
+TRACE_DEFINE_ENUM(XLOG_STATE_WANT_SYNC);
+TRACE_DEFINE_ENUM(XLOG_STATE_SYNCING);
+TRACE_DEFINE_ENUM(XLOG_STATE_DONE_SYNC);
+TRACE_DEFINE_ENUM(XLOG_STATE_CALLBACK);
+TRACE_DEFINE_ENUM(XLOG_STATE_DIRTY);
+TRACE_DEFINE_ENUM(XLOG_STATE_IOERROR);
+
+DECLARE_EVENT_CLASS(xlog_iclog_class,
+	TP_PROTO(struct xlog_in_core *iclog, unsigned long caller_ip),
+	TP_ARGS(iclog, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(uint32_t, state)
+		__field(int32_t, refcount)
+		__field(uint32_t, offset)
+		__field(unsigned long long, lsn)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = iclog->ic_log->l_mp->m_super->s_dev;
+		__entry->state = iclog->ic_state;
+		__entry->refcount = atomic_read(&iclog->ic_refcnt);
+		__entry->offset = iclog->ic_offset;
+		__entry->lsn = be64_to_cpu(iclog->ic_header.h_lsn);
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d state %s refcnt %d offset %u lsn 0x%llx caller %pS",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __print_symbolic(__entry->state, XLOG_STATE_STRINGS),
+		  __entry->refcount,
+		  __entry->offset,
+		  __entry->lsn,
+		  (char *)__entry->caller_ip)
+
+);
+
+#define DEFINE_ICLOG_EVENT(name)	\
+DEFINE_EVENT(xlog_iclog_class, name,	\
+	TP_PROTO(struct xlog_in_core *iclog, unsigned long caller_ip), \
+	TP_ARGS(iclog, caller_ip))
+
+DEFINE_ICLOG_EVENT(xlog_iclog_activate);
+DEFINE_ICLOG_EVENT(xlog_iclog_clean);
+DEFINE_ICLOG_EVENT(xlog_iclog_callback);
+DEFINE_ICLOG_EVENT(xlog_iclog_callbacks_start);
+DEFINE_ICLOG_EVENT(xlog_iclog_callbacks_done);
+DEFINE_ICLOG_EVENT(xlog_iclog_force);
+DEFINE_ICLOG_EVENT(xlog_iclog_force_lsn);
+DEFINE_ICLOG_EVENT(xlog_iclog_get_space);
+DEFINE_ICLOG_EVENT(xlog_iclog_release);
+DEFINE_ICLOG_EVENT(xlog_iclog_switch);
+DEFINE_ICLOG_EVENT(xlog_iclog_sync);
+DEFINE_ICLOG_EVENT(xlog_iclog_syncing);
+DEFINE_ICLOG_EVENT(xlog_iclog_sync_done);
+DEFINE_ICLOG_EVENT(xlog_iclog_want_sync);
+DEFINE_ICLOG_EVENT(xlog_iclog_wait_on);
+DEFINE_ICLOG_EVENT(xlog_iclog_write);
 
 #endif /* _TRACE_XFS_H */
 
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 586f2992b789..87bffd12c20c 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -839,7 +839,7 @@ __xfs_trans_commit(
 	bool			regrant)
 {
 	struct xfs_mount	*mp = tp->t_mountp;
-	xfs_lsn_t		commit_lsn = -1;
+	xfs_csn_t		commit_seq = 0;
 	int			error = 0;
 	int			sync = tp->t_flags & XFS_TRANS_SYNC;
 
@@ -881,7 +881,7 @@ __xfs_trans_commit(
 		xfs_trans_apply_sb_deltas(tp);
 	xfs_trans_apply_dquot_deltas(tp);
 
-	xfs_log_commit_cil(mp, tp, &commit_lsn, regrant);
+	xlog_cil_commit(mp->m_log, tp, &commit_seq, regrant);
 
 	xfs_trans_free(tp);
 
@@ -890,7 +890,7 @@ __xfs_trans_commit(
 	 * log out now and wait for it.
 	 */
 	if (sync) {
-		error = xfs_log_force_lsn(mp, commit_lsn, XFS_LOG_SYNC, NULL);
+		error = xfs_log_force_seq(mp, commit_seq, XFS_LOG_SYNC, NULL);
 		XFS_STATS_INC(mp, xs_trans_sync);
 	} else {
 		XFS_STATS_INC(mp, xs_trans_async);
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index ee42d98d9011..50da47f23a07 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -43,7 +43,7 @@ struct xfs_log_item {
 	struct list_head		li_cil;		/* CIL pointers */
 	struct xfs_log_vec		*li_lv;		/* active log vector */
 	struct xfs_log_vec		*li_lv_shadow;	/* standby vector */
-	xfs_lsn_t			li_seq;		/* CIL commit seq */
+	xfs_csn_t			li_seq;		/* CIL commit seq */
 };
 
 /*
@@ -69,7 +69,7 @@ struct xfs_item_ops {
 	void (*iop_pin)(struct xfs_log_item *);
 	void (*iop_unpin)(struct xfs_log_item *, int remove);
 	uint (*iop_push)(struct xfs_log_item *, struct list_head *);
-	void (*iop_committing)(struct xfs_log_item *, xfs_lsn_t commit_lsn);
+	void (*iop_committing)(struct xfs_log_item *lip, xfs_csn_t seq);
 	void (*iop_release)(struct xfs_log_item *);
 	xfs_lsn_t (*iop_committed)(struct xfs_log_item *, xfs_lsn_t);
 	int (*iop_recover)(struct xfs_log_item *lip,