1 files changed, 319 insertions, 245 deletions

diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c b/fs/xfs/libxfs/xfs_alloc_btree.c
index 4e59cc8a2802..fa1f03c1331e 100644
--- a/fs/xfs/libxfs/xfs_alloc_btree.c
+++ b/fs/xfs/libxfs/xfs_alloc_btree.c
@@ -9,62 +9,74 @@
 #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"
 #include "xfs_alloc_btree.h"
 #include "xfs_alloc.h"
 #include "xfs_extent_busy.h"
 #include "xfs_error.h"
+#include "xfs_health.h"
 #include "xfs_trace.h"
-#include "xfs_cksum.h"
 #include "xfs_trans.h"
+#include "xfs_ag.h"
 
+static struct kmem_cache	*xfs_allocbt_cur_cache;
 
 STATIC struct xfs_btree_cur *
-xfs_allocbt_dup_cursor(
+xfs_bnobt_dup_cursor(
 	struct xfs_btree_cur	*cur)
 {
-	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
-			cur->bc_private.a.agbp, cur->bc_private.a.agno,
-			cur->bc_btnum);
+	return xfs_bnobt_init_cursor(cur->bc_mp, cur->bc_tp, cur->bc_ag.agbp,
+			to_perag(cur->bc_group));
+}
+
+STATIC struct xfs_btree_cur *
+xfs_cntbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_cntbt_init_cursor(cur->bc_mp, cur->bc_tp, cur->bc_ag.agbp,
+			to_perag(cur->bc_group));
 }
 
 STATIC void
 xfs_allocbt_set_root(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_ptr	*ptr,
-	int			inc)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*ptr,
+	int				inc)
 {
-	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
-	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
-	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
-	int			btnum = cur->bc_btnum;
-	struct xfs_perag	*pag = xfs_perag_get(cur->bc_mp, seqno);
+	struct xfs_perag		*pag = to_perag(cur->bc_group);
+	struct xfs_buf			*agbp = cur->bc_ag.agbp;
+	struct xfs_agf			*agf = agbp->b_addr;
 
 	ASSERT(ptr->s != 0);
 
-	agf->agf_roots[btnum] = ptr->s;
-	be32_add_cpu(&agf->agf_levels[btnum], inc);
-	pag->pagf_levels[btnum] += inc;
-	xfs_perag_put(pag);
+	if (xfs_btree_is_bno(cur->bc_ops)) {
+		agf->agf_bno_root = ptr->s;
+		be32_add_cpu(&agf->agf_bno_level, inc);
+		pag->pagf_bno_level += inc;
+	} else {
+		agf->agf_cnt_root = ptr->s;
+		be32_add_cpu(&agf->agf_cnt_level, inc);
+		pag->pagf_cnt_level += inc;
+	}
 
 	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
 }
 
 STATIC int
 xfs_allocbt_alloc_block(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_ptr	*start,
-	union xfs_btree_ptr	*new,
-	int			*stat)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_ptr	*start,
+	union xfs_btree_ptr		*new,
+	int				*stat)
 {
 	int			error;
 	xfs_agblock_t		bno;
 
 	/* Allocate the new block from the freelist. If we can't, give up.  */
-	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
-				       &bno, 1);
+	error = xfs_alloc_get_freelist(to_perag(cur->bc_group), cur->bc_tp,
+			cur->bc_ag.agbp, &bno, 1);
 	if (error)
 		return error;
 
@@ -73,9 +85,9 @@ xfs_allocbt_alloc_block(
 		return 0;
 	}
 
-	xfs_extent_busy_reuse(cur->bc_mp, cur->bc_private.a.agno, bno, 1, false);
+	atomic64_inc(&cur->bc_mp->m_allocbt_blks);
+	xfs_extent_busy_reuse(cur->bc_group, bno, 1, false);
 
-	xfs_trans_agbtree_delta(cur->bc_tp, 1);
 	new->s = cpu_to_be32(bno);
 
 	*stat = 1;
@@ -87,85 +99,22 @@ xfs_allocbt_free_block(
 	struct xfs_btree_cur	*cur,
 	struct xfs_buf		*bp)
 {
-	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
-	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	struct xfs_buf		*agbp = cur->bc_ag.agbp;
 	xfs_agblock_t		bno;
 	int			error;
 
-	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
-	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
+	bno = xfs_daddr_to_agbno(cur->bc_mp, xfs_buf_daddr(bp));
+	error = xfs_alloc_put_freelist(to_perag(cur->bc_group), cur->bc_tp,
+			agbp, NULL, bno, 1);
 	if (error)
 		return error;
 
-	xfs_extent_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
+	atomic64_dec(&cur->bc_mp->m_allocbt_blks);
+	xfs_extent_busy_insert(cur->bc_tp, pag_group(agbp->b_pag), bno, 1,
 			      XFS_EXTENT_BUSY_SKIP_DISCARD);
-	xfs_trans_agbtree_delta(cur->bc_tp, -1);
 	return 0;
 }
 
-/*
- * Update the longest extent in the AGF
- */
-STATIC void
-xfs_allocbt_update_lastrec(
-	struct xfs_btree_cur	*cur,
-	struct xfs_btree_block	*block,
-	union xfs_btree_rec	*rec,
-	int			ptr,
-	int			reason)
-{
-	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
-	struct xfs_perag	*pag;
-	__be32			len;
-	int			numrecs;
-
-	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
-
-	switch (reason) {
-	case LASTREC_UPDATE:
-		/*
-		 * If this is the last leaf block and it's the last record,
-		 * then update the size of the longest extent in the AG.
-		 */
-		if (ptr != xfs_btree_get_numrecs(block))
-			return;
-		len = rec->alloc.ar_blockcount;
-		break;
-	case LASTREC_INSREC:
-		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
-		    be32_to_cpu(agf->agf_longest))
-			return;
-		len = rec->alloc.ar_blockcount;
-		break;
-	case LASTREC_DELREC:
-		numrecs = xfs_btree_get_numrecs(block);
-		if (ptr <= numrecs)
-			return;
-		ASSERT(ptr == numrecs + 1);
-
-		if (numrecs) {
-			xfs_alloc_rec_t *rrp;
-
-			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
-			len = rrp->ar_blockcount;
-		} else {
-			len = 0;
-		}
-
-		break;
-	default:
-		ASSERT(0);
-		return;
-	}
-
-	agf->agf_longest = len;
-	pag = xfs_perag_get(cur->bc_mp, seqno);
-	pag->pagf_longest = be32_to_cpu(len);
-	xfs_perag_put(pag);
-	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
-}
-
 STATIC int
 xfs_allocbt_get_minrecs(
 	struct xfs_btree_cur	*cur,
@@ -184,8 +133,8 @@ xfs_allocbt_get_maxrecs(
 
 STATIC void
 xfs_allocbt_init_key_from_rec(
-	union xfs_btree_key	*key,
-	union xfs_btree_rec	*rec)
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
 {
 	key->alloc.ar_startblock = rec->alloc.ar_startblock;
 	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
@@ -193,10 +142,10 @@ xfs_allocbt_init_key_from_rec(
 
 STATIC void
 xfs_bnobt_init_high_key_from_rec(
-	union xfs_btree_key	*key,
-	union xfs_btree_rec	*rec)
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
 {
-	__u32			x;
+	__u32				x;
 
 	x = be32_to_cpu(rec->alloc.ar_startblock);
 	x += be32_to_cpu(rec->alloc.ar_blockcount) - 1;
@@ -206,8 +155,8 @@ xfs_bnobt_init_high_key_from_rec(
 
 STATIC void
 xfs_cntbt_init_high_key_from_rec(
-	union xfs_btree_key	*key,
-	union xfs_btree_rec	*rec)
+	union xfs_btree_key		*key,
+	const union xfs_btree_rec	*rec)
 {
 	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
 	key->alloc.ar_startblock = 0;
@@ -227,120 +176,124 @@ xfs_allocbt_init_ptr_from_cur(
 	struct xfs_btree_cur	*cur,
 	union xfs_btree_ptr	*ptr)
 {
-	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
+	struct xfs_agf		*agf = cur->bc_ag.agbp->b_addr;
 
-	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(cur->bc_group->xg_gno == be32_to_cpu(agf->agf_seqno));
 
-	ptr->s = agf->agf_roots[cur->bc_btnum];
+	if (xfs_btree_is_bno(cur->bc_ops))
+		ptr->s = agf->agf_bno_root;
+	else
+		ptr->s = agf->agf_cnt_root;
 }
 
-STATIC int64_t
-xfs_bnobt_key_diff(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*key)
+STATIC int
+xfs_bnobt_cmp_key_with_cur(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
 {
-	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
-	xfs_alloc_key_t		*kp = &key->alloc;
+	struct xfs_alloc_rec_incore	*rec = &cur->bc_rec.a;
+	const struct xfs_alloc_rec	*kp = &key->alloc;
 
-	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
+	return cmp_int(be32_to_cpu(kp->ar_startblock),
+		       rec->ar_startblock);
 }
 
-STATIC int64_t
-xfs_cntbt_key_diff(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*key)
+STATIC int
+xfs_cntbt_cmp_key_with_cur(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key)
 {
-	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
-	xfs_alloc_key_t		*kp = &key->alloc;
-	int64_t			diff;
-
-	diff = (int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
-	if (diff)
-		return diff;
+	struct xfs_alloc_rec_incore	*rec = &cur->bc_rec.a;
+	const struct xfs_alloc_rec	*kp = &key->alloc;
 
-	return (int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
+	return cmp_int(be32_to_cpu(kp->ar_blockcount), rec->ar_blockcount) ?:
+	       cmp_int(be32_to_cpu(kp->ar_startblock), rec->ar_startblock);
 }
 
-STATIC int64_t
-xfs_bnobt_diff_two_keys(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*k1,
-	union xfs_btree_key	*k2)
+STATIC int
+xfs_bnobt_cmp_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2,
+	const union xfs_btree_key	*mask)
 {
-	return (int64_t)be32_to_cpu(k1->alloc.ar_startblock) -
-			  be32_to_cpu(k2->alloc.ar_startblock);
+	ASSERT(!mask || mask->alloc.ar_startblock);
+
+	return cmp_int(be32_to_cpu(k1->alloc.ar_startblock),
+		       be32_to_cpu(k2->alloc.ar_startblock));
 }
 
-STATIC int64_t
-xfs_cntbt_diff_two_keys(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*k1,
-	union xfs_btree_key	*k2)
+STATIC int
+xfs_cntbt_cmp_two_keys(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2,
+	const union xfs_btree_key	*mask)
 {
-	int64_t			diff;
-
-	diff =  be32_to_cpu(k1->alloc.ar_blockcount) -
-		be32_to_cpu(k2->alloc.ar_blockcount);
-	if (diff)
-		return diff;
+	ASSERT(!mask || (mask->alloc.ar_blockcount &&
+			 mask->alloc.ar_startblock));
 
-	return  be32_to_cpu(k1->alloc.ar_startblock) -
-		be32_to_cpu(k2->alloc.ar_startblock);
+	return cmp_int(be32_to_cpu(k1->alloc.ar_blockcount),
+		       be32_to_cpu(k2->alloc.ar_blockcount)) ?:
+	       cmp_int(be32_to_cpu(k1->alloc.ar_startblock),
+		       be32_to_cpu(k2->alloc.ar_startblock));
 }
 
 static xfs_failaddr_t
 xfs_allocbt_verify(
 	struct xfs_buf		*bp)
 {
-	struct xfs_mount	*mp = bp->b_target->bt_mount;
+	struct xfs_mount	*mp = bp->b_mount;
 	struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
 	struct xfs_perag	*pag = bp->b_pag;
 	xfs_failaddr_t		fa;
 	unsigned int		level;
 
+	if (!xfs_verify_magic(bp, block->bb_magic))
+		return __this_address;
+
+	if (xfs_has_crc(mp)) {
+		fa = xfs_btree_agblock_v5hdr_verify(bp);
+		if (fa)
+			return fa;
+	}
+
 	/*
-	 * magic number and level verification
-	 *
-	 * During growfs operations, we can't verify the exact level or owner as
-	 * the perag is not fully initialised and hence not attached to the
-	 * buffer.  In this case, check against the maximum tree depth.
+	 * The perag may not be attached during grow operations or fully
+	 * initialized from the AGF during log recovery. Therefore we can only
+	 * check against maximum tree depth from those contexts.
 	 *
-	 * Similarly, during log recovery we will have a perag structure
-	 * attached, but the agf information will not yet have been initialised
-	 * from the on disk AGF. Again, we can only check against maximum limits
-	 * in this case.
+	 * Otherwise check against the per-tree limit. Peek at one of the
+	 * verifier magic values to determine the type of tree we're verifying
+	 * against.
 	 */
 	level = be16_to_cpu(block->bb_level);
-	switch (block->bb_magic) {
-	case cpu_to_be32(XFS_ABTB_CRC_MAGIC):
-		fa = xfs_btree_sblock_v5hdr_verify(bp);
-		if (fa)
-			return fa;
-		/* fall through */
-	case cpu_to_be32(XFS_ABTB_MAGIC):
-		if (pag && pag->pagf_init) {
-			if (level >= pag->pagf_levels[XFS_BTNUM_BNOi])
-				return __this_address;
-		} else if (level >= mp->m_ag_maxlevels)
-			return __this_address;
-		break;
-	case cpu_to_be32(XFS_ABTC_CRC_MAGIC):
-		fa = xfs_btree_sblock_v5hdr_verify(bp);
-		if (fa)
-			return fa;
-		/* fall through */
-	case cpu_to_be32(XFS_ABTC_MAGIC):
-		if (pag && pag->pagf_init) {
-			if (level >= pag->pagf_levels[XFS_BTNUM_CNTi])
-				return __this_address;
-		} else if (level >= mp->m_ag_maxlevels)
+	if (pag && xfs_perag_initialised_agf(pag)) {
+		unsigned int	maxlevel, repair_maxlevel = 0;
+
+		/*
+		 * Online repair could be rewriting the free space btrees, so
+		 * we'll validate against the larger of either tree while this
+		 * is going on.
+		 */
+		if (bp->b_ops->magic[0] == cpu_to_be32(XFS_ABTC_MAGIC)) {
+			maxlevel = pag->pagf_cnt_level;
+#ifdef CONFIG_XFS_ONLINE_REPAIR
+			repair_maxlevel = pag->pagf_repair_cnt_level;
+#endif
+		} else {
+			maxlevel = pag->pagf_bno_level;
+#ifdef CONFIG_XFS_ONLINE_REPAIR
+			repair_maxlevel = pag->pagf_repair_bno_level;
+#endif
+		}
+
+		if (level >= max(maxlevel, repair_maxlevel))
 			return __this_address;
-		break;
-	default:
+	} else if (level >= mp->m_alloc_maxlevels)
 		return __this_address;
-	}
 
-	return xfs_btree_sblock_verify(bp, mp->m_alloc_mxr[level != 0]);
+	return xfs_btree_agblock_verify(bp, mp->m_alloc_mxr[level != 0]);
 }
 
 static void
@@ -349,7 +302,7 @@ xfs_allocbt_read_verify(
 {
 	xfs_failaddr_t	fa;
 
-	if (!xfs_btree_sblock_verify_crc(bp))
+	if (!xfs_btree_agblock_verify_crc(bp))
 		xfs_verifier_error(bp, -EFSBADCRC, __this_address);
 	else {
 		fa = xfs_allocbt_verify(bp);
@@ -373,23 +326,33 @@ xfs_allocbt_write_verify(
 		xfs_verifier_error(bp, -EFSCORRUPTED, fa);
 		return;
 	}
-	xfs_btree_sblock_calc_crc(bp);
+	xfs_btree_agblock_calc_crc(bp);
 
 }
 
-const struct xfs_buf_ops xfs_allocbt_buf_ops = {
-	.name = "xfs_allocbt",
+const struct xfs_buf_ops xfs_bnobt_buf_ops = {
+	.name = "xfs_bnobt",
+	.magic = { cpu_to_be32(XFS_ABTB_MAGIC),
+		   cpu_to_be32(XFS_ABTB_CRC_MAGIC) },
 	.verify_read = xfs_allocbt_read_verify,
 	.verify_write = xfs_allocbt_write_verify,
 	.verify_struct = xfs_allocbt_verify,
 };
 
+const struct xfs_buf_ops xfs_cntbt_buf_ops = {
+	.name = "xfs_cntbt",
+	.magic = { cpu_to_be32(XFS_ABTC_MAGIC),
+		   cpu_to_be32(XFS_ABTC_CRC_MAGIC) },
+	.verify_read = xfs_allocbt_read_verify,
+	.verify_write = xfs_allocbt_write_verify,
+	.verify_struct = xfs_allocbt_verify,
+};
 
 STATIC int
 xfs_bnobt_keys_inorder(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*k1,
-	union xfs_btree_key	*k2)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
 {
 	return be32_to_cpu(k1->alloc.ar_startblock) <
 	       be32_to_cpu(k2->alloc.ar_startblock);
@@ -397,9 +360,9 @@ xfs_bnobt_keys_inorder(
 
 STATIC int
 xfs_bnobt_recs_inorder(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_rec	*r1,
-	union xfs_btree_rec	*r2)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
 {
 	return be32_to_cpu(r1->alloc.ar_startblock) +
 		be32_to_cpu(r1->alloc.ar_blockcount) <=
@@ -408,9 +371,9 @@ xfs_bnobt_recs_inorder(
 
 STATIC int
 xfs_cntbt_keys_inorder(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_key	*k1,
-	union xfs_btree_key	*k2)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*k1,
+	const union xfs_btree_key	*k2)
 {
 	return be32_to_cpu(k1->alloc.ar_blockcount) <
 		be32_to_cpu(k2->alloc.ar_blockcount) ||
@@ -421,9 +384,9 @@ xfs_cntbt_keys_inorder(
 
 STATIC int
 xfs_cntbt_recs_inorder(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_rec	*r1,
-	union xfs_btree_rec	*r2)
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_rec	*r1,
+	const union xfs_btree_rec	*r2)
 {
 	return be32_to_cpu(r1->alloc.ar_blockcount) <
 		be32_to_cpu(r2->alloc.ar_blockcount) ||
@@ -432,107 +395,199 @@ xfs_cntbt_recs_inorder(
 		 be32_to_cpu(r2->alloc.ar_startblock));
 }
 
-static const struct xfs_btree_ops xfs_bnobt_ops = {
+STATIC enum xbtree_key_contig
+xfs_allocbt_keys_contiguous(
+	struct xfs_btree_cur		*cur,
+	const union xfs_btree_key	*key1,
+	const union xfs_btree_key	*key2,
+	const union xfs_btree_key	*mask)
+{
+	ASSERT(!mask || mask->alloc.ar_startblock);
+
+	return xbtree_key_contig(be32_to_cpu(key1->alloc.ar_startblock),
+				 be32_to_cpu(key2->alloc.ar_startblock));
+}
+
+const struct xfs_btree_ops xfs_bnobt_ops = {
+	.name			= "bno",
+	.type			= XFS_BTREE_TYPE_AG,
+
 	.rec_len		= sizeof(xfs_alloc_rec_t),
 	.key_len		= sizeof(xfs_alloc_key_t),
+	.ptr_len		= XFS_BTREE_SHORT_PTR_LEN,
 
-	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.lru_refs		= XFS_ALLOC_BTREE_REF,
+	.statoff		= XFS_STATS_CALC_INDEX(xs_abtb_2),
+	.sick_mask		= XFS_SICK_AG_BNOBT,
+
+	.dup_cursor		= xfs_bnobt_dup_cursor,
 	.set_root		= xfs_allocbt_set_root,
 	.alloc_block		= xfs_allocbt_alloc_block,
 	.free_block		= xfs_allocbt_free_block,
-	.update_lastrec		= xfs_allocbt_update_lastrec,
 	.get_minrecs		= xfs_allocbt_get_minrecs,
 	.get_maxrecs		= xfs_allocbt_get_maxrecs,
 	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
 	.init_high_key_from_rec	= xfs_bnobt_init_high_key_from_rec,
 	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
 	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
-	.key_diff		= xfs_bnobt_key_diff,
-	.buf_ops		= &xfs_allocbt_buf_ops,
-	.diff_two_keys		= xfs_bnobt_diff_two_keys,
+	.cmp_key_with_cur	= xfs_bnobt_cmp_key_with_cur,
+	.buf_ops		= &xfs_bnobt_buf_ops,
+	.cmp_two_keys		= xfs_bnobt_cmp_two_keys,
 	.keys_inorder		= xfs_bnobt_keys_inorder,
 	.recs_inorder		= xfs_bnobt_recs_inorder,
+	.keys_contiguous	= xfs_allocbt_keys_contiguous,
 };
 
-static const struct xfs_btree_ops xfs_cntbt_ops = {
+const struct xfs_btree_ops xfs_cntbt_ops = {
+	.name			= "cnt",
+	.type			= XFS_BTREE_TYPE_AG,
+
 	.rec_len		= sizeof(xfs_alloc_rec_t),
 	.key_len		= sizeof(xfs_alloc_key_t),
+	.ptr_len		= XFS_BTREE_SHORT_PTR_LEN,
+
+	.lru_refs		= XFS_ALLOC_BTREE_REF,
+	.statoff		= XFS_STATS_CALC_INDEX(xs_abtc_2),
+	.sick_mask		= XFS_SICK_AG_CNTBT,
 
-	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.dup_cursor		= xfs_cntbt_dup_cursor,
 	.set_root		= xfs_allocbt_set_root,
 	.alloc_block		= xfs_allocbt_alloc_block,
 	.free_block		= xfs_allocbt_free_block,
-	.update_lastrec		= xfs_allocbt_update_lastrec,
 	.get_minrecs		= xfs_allocbt_get_minrecs,
 	.get_maxrecs		= xfs_allocbt_get_maxrecs,
 	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
 	.init_high_key_from_rec	= xfs_cntbt_init_high_key_from_rec,
 	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
 	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
-	.key_diff		= xfs_cntbt_key_diff,
-	.buf_ops		= &xfs_allocbt_buf_ops,
-	.diff_two_keys		= xfs_cntbt_diff_two_keys,
+	.cmp_key_with_cur	= xfs_cntbt_cmp_key_with_cur,
+	.buf_ops		= &xfs_cntbt_buf_ops,
+	.cmp_two_keys		= xfs_cntbt_cmp_two_keys,
 	.keys_inorder		= xfs_cntbt_keys_inorder,
 	.recs_inorder		= xfs_cntbt_recs_inorder,
+	.keys_contiguous	= NULL, /* not needed right now */
 };
 
 /*
- * Allocate a new allocation btree cursor.
+ * Allocate a new bnobt cursor.
+ *
+ * For staging cursors tp and agbp are NULL.
  */
-struct xfs_btree_cur *			/* new alloc btree cursor */
-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 */
-	xfs_btnum_t		btnum)		/* btree identifier */
-{
-	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+struct xfs_btree_cur *
+xfs_bnobt_init_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
 	struct xfs_btree_cur	*cur;
 
-	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
+	cur = xfs_btree_alloc_cursor(mp, tp, &xfs_bnobt_ops,
+			mp->m_alloc_maxlevels, xfs_allocbt_cur_cache);
+	cur->bc_group = xfs_group_hold(pag_group(pag));
+	cur->bc_ag.agbp = agbp;
+	if (agbp) {
+		struct xfs_agf		*agf = agbp->b_addr;
 
-	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_NOFS);
+		cur->bc_nlevels = be32_to_cpu(agf->agf_bno_level);
+	}
+	return cur;
+}
 
-	cur->bc_tp = tp;
-	cur->bc_mp = mp;
-	cur->bc_btnum = btnum;
-	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+/*
+ * Allocate a new cntbt cursor.
+ *
+ * For staging cursors tp and agbp are NULL.
+ */
+struct xfs_btree_cur *
+xfs_cntbt_init_cursor(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp,
+	struct xfs_perag	*pag)
+{
+	struct xfs_btree_cur	*cur;
 
-	if (btnum == XFS_BTNUM_CNT) {
-		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtc_2);
-		cur->bc_ops = &xfs_cntbt_ops;
-		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]);
-		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
-	} else {
-		cur->bc_statoff = XFS_STATS_CALC_INDEX(xs_abtb_2);
-		cur->bc_ops = &xfs_bnobt_ops;
-		cur->bc_nlevels = be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]);
+	cur = xfs_btree_alloc_cursor(mp, tp, &xfs_cntbt_ops,
+			mp->m_alloc_maxlevels, xfs_allocbt_cur_cache);
+	cur->bc_group = xfs_group_hold(pag_group(pag));
+	cur->bc_ag.agbp = agbp;
+	if (agbp) {
+		struct xfs_agf		*agf = agbp->b_addr;
+
+		cur->bc_nlevels = be32_to_cpu(agf->agf_cnt_level);
 	}
+	return cur;
+}
+
+/*
+ * Install a new free space btree root.  Caller is responsible for invalidating
+ * and freeing the old btree blocks.
+ */
+void
+xfs_allocbt_commit_staged_btree(
+	struct xfs_btree_cur	*cur,
+	struct xfs_trans	*tp,
+	struct xfs_buf		*agbp)
+{
+	struct xfs_agf		*agf = agbp->b_addr;
+	struct xbtree_afakeroot	*afake = cur->bc_ag.afake;
 
-	cur->bc_private.a.agbp = agbp;
-	cur->bc_private.a.agno = agno;
+	ASSERT(cur->bc_flags & XFS_BTREE_STAGING);
 
-	if (xfs_sb_version_hascrc(&mp->m_sb))
-		cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+	if (xfs_btree_is_bno(cur->bc_ops)) {
+		agf->agf_bno_root = cpu_to_be32(afake->af_root);
+		agf->agf_bno_level = cpu_to_be32(afake->af_levels);
+	} else {
+		agf->agf_cnt_root = cpu_to_be32(afake->af_root);
+		agf->agf_cnt_level = cpu_to_be32(afake->af_levels);
+	}
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
 
-	return cur;
+	xfs_btree_commit_afakeroot(cur, tp, agbp);
+}
+
+/* Calculate number of records in an alloc btree block. */
+static inline unsigned int
+xfs_allocbt_block_maxrecs(
+	unsigned int		blocklen,
+	bool			leaf)
+{
+	if (leaf)
+		return blocklen / sizeof(xfs_alloc_rec_t);
+	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
 }
 
 /*
  * Calculate number of records in an alloc btree block.
  */
-int
+unsigned int
 xfs_allocbt_maxrecs(
 	struct xfs_mount	*mp,
-	int			blocklen,
-	int			leaf)
+	unsigned int		blocklen,
+	bool			leaf)
 {
 	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
+	return xfs_allocbt_block_maxrecs(blocklen, leaf);
+}
 
-	if (leaf)
-		return blocklen / sizeof(xfs_alloc_rec_t);
-	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
+/* Free space btrees are at their largest when every other block is free. */
+#define XFS_MAX_FREESP_RECORDS	((XFS_MAX_AG_BLOCKS + 1) / 2)
+
+/* Compute the max possible height for free space btrees. */
+unsigned int
+xfs_allocbt_maxlevels_ondisk(void)
+{
+	unsigned int		minrecs[2];
+	unsigned int		blocklen;
+
+	blocklen = min(XFS_MIN_BLOCKSIZE - XFS_BTREE_SBLOCK_LEN,
+		       XFS_MIN_CRC_BLOCKSIZE - XFS_BTREE_SBLOCK_CRC_LEN);
+
+	minrecs[0] = xfs_allocbt_block_maxrecs(blocklen, true) / 2;
+	minrecs[1] = xfs_allocbt_block_maxrecs(blocklen, false) / 2;
+
+	return xfs_btree_compute_maxlevels(minrecs, XFS_MAX_FREESP_RECORDS);
 }
 
 /* Calculate the freespace btree size for some records. */
@@ -543,3 +598,22 @@ xfs_allocbt_calc_size(
 {
 	return xfs_btree_calc_size(mp->m_alloc_mnr, len);
 }
+
+int __init
+xfs_allocbt_init_cur_cache(void)
+{
+	xfs_allocbt_cur_cache = kmem_cache_create("xfs_bnobt_cur",
+			xfs_btree_cur_sizeof(xfs_allocbt_maxlevels_ondisk()),
+			0, 0, NULL);
+
+	if (!xfs_allocbt_cur_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_allocbt_destroy_cur_cache(void)
+{
+	kmem_cache_destroy(xfs_allocbt_cur_cache);
+	xfs_allocbt_cur_cache = NULL;
+}