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()
  ...

37 files changed, 2117 insertions, 1305 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 */