1 files changed, 215 insertions, 303 deletions

diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 828da4ac4316..474f5a04ec63 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -25,6 +25,8 @@
 #include "xfs_dquot.h"
 #include "xfs_icache.h"
 #include "xfs_rtbitmap.h"
+#include "xfs_rtgroup.h"
+#include "xfs_sb.h"
 
 struct kmem_cache	*xfs_trans_cache;
 
@@ -67,7 +69,7 @@ xfs_trans_free(
 	struct xfs_trans	*tp)
 {
 	xfs_extent_busy_sort(&tp->t_busy);
-	xfs_extent_busy_clear(tp->t_mountp, &tp->t_busy, false);
+	xfs_extent_busy_clear(&tp->t_busy, false);
 
 	trace_xfs_trans_free(tp, _RET_IP_);
 	xfs_trans_clear_context(tp);
@@ -98,7 +100,6 @@ xfs_trans_dup(
 	/*
 	 * Initialize the new transaction structure.
 	 */
-	ntp->t_magic = XFS_TRANS_HEADER_MAGIC;
 	ntp->t_mountp = tp->t_mountp;
 	INIT_LIST_HEAD(&ntp->t_items);
 	INIT_LIST_HEAD(&ntp->t_busy);
@@ -133,18 +134,14 @@ xfs_trans_dup(
 }
 
 /*
- * This is called to reserve free disk blocks and log space for the
- * given transaction.  This must be done before allocating any resources
- * within the transaction.
+ * This is called to reserve free disk blocks and log space for the given
+ * transaction before allocating any resources within the transaction.
  *
  * This will return ENOSPC if there are not enough blocks available.
  * It will sleep waiting for available log space.
- * The only valid value for the flags parameter is XFS_RES_LOG_PERM, which
- * is used by long running transactions.  If any one of the reservations
- * fails then they will all be backed out.
  *
- * This does not do quota reservations. That typically is done by the
- * caller afterwards.
+ * This does not do quota reservations. That typically is done by the caller
+ * afterwards.
  */
 static int
 xfs_trans_reserve(
@@ -157,10 +154,12 @@ xfs_trans_reserve(
 	int			error = 0;
 	bool			rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
 
+	ASSERT(resp->tr_logres > 0);
+
 	/*
-	 * Attempt to reserve the needed disk blocks by decrementing
-	 * the number needed from the number available.  This will
-	 * fail if the count would go below zero.
+	 * Attempt to reserve the needed disk blocks by decrementing the number
+	 * needed from the number available.  This will fail if the count would
+	 * go below zero.
 	 */
 	if (blocks > 0) {
 		error = xfs_dec_fdblocks(mp, blocks, rsvd);
@@ -172,42 +171,20 @@ xfs_trans_reserve(
 	/*
 	 * Reserve the log space needed for this transaction.
 	 */
-	if (resp->tr_logres > 0) {
-		bool	permanent = false;
-
-		ASSERT(tp->t_log_res == 0 ||
-		       tp->t_log_res == resp->tr_logres);
-		ASSERT(tp->t_log_count == 0 ||
-		       tp->t_log_count == resp->tr_logcount);
-
-		if (resp->tr_logflags & XFS_TRANS_PERM_LOG_RES) {
-			tp->t_flags |= XFS_TRANS_PERM_LOG_RES;
-			permanent = true;
-		} else {
-			ASSERT(tp->t_ticket == NULL);
-			ASSERT(!(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
-		}
-
-		if (tp->t_ticket != NULL) {
-			ASSERT(resp->tr_logflags & XFS_TRANS_PERM_LOG_RES);
-			error = xfs_log_regrant(mp, tp->t_ticket);
-		} else {
-			error = xfs_log_reserve(mp, resp->tr_logres,
-						resp->tr_logcount,
-						&tp->t_ticket, permanent);
-		}
-
-		if (error)
-			goto undo_blocks;
+	if (resp->tr_logflags & XFS_TRANS_PERM_LOG_RES)
+		tp->t_flags |= XFS_TRANS_PERM_LOG_RES;
+	error = xfs_log_reserve(mp, resp->tr_logres, resp->tr_logcount,
+			&tp->t_ticket, (tp->t_flags & XFS_TRANS_PERM_LOG_RES));
+	if (error)
+		goto undo_blocks;
 
-		tp->t_log_res = resp->tr_logres;
-		tp->t_log_count = resp->tr_logcount;
-	}
+	tp->t_log_res = resp->tr_logres;
+	tp->t_log_count = resp->tr_logcount;
 
 	/*
-	 * Attempt to reserve the needed realtime extents by decrementing
-	 * the number needed from the number available.  This will
-	 * fail if the count would go below zero.
+	 * Attempt to reserve the needed realtime extents by decrementing the
+	 * number needed from the number available.  This will fail if the
+	 * count would go below zero.
 	 */
 	if (rtextents > 0) {
 		error = xfs_dec_frextents(mp, rtextents);
@@ -220,18 +197,11 @@ xfs_trans_reserve(
 
 	return 0;
 
-	/*
-	 * Error cases jump to one of these labels to undo any
-	 * reservations which have already been performed.
-	 */
 undo_log:
-	if (resp->tr_logres > 0) {
-		xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
-		tp->t_ticket = NULL;
-		tp->t_log_res = 0;
-		tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES;
-	}
-
+	xfs_log_ticket_ungrant(mp->m_log, tp->t_ticket);
+	tp->t_ticket = NULL;
+	tp->t_log_res = 0;
+	tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES;
 undo_blocks:
 	if (blocks > 0) {
 		xfs_add_fdblocks(mp, blocks);
@@ -240,6 +210,28 @@ undo_blocks:
 	return error;
 }
 
+static struct xfs_trans *
+__xfs_trans_alloc(
+	struct xfs_mount	*mp,
+	uint			flags)
+{
+	struct xfs_trans	*tp;
+
+	ASSERT(!(flags & XFS_TRANS_RES_FDBLKS) || xfs_has_lazysbcount(mp));
+
+	tp = kmem_cache_zalloc(xfs_trans_cache, GFP_KERNEL | __GFP_NOFAIL);
+	if (!(flags & XFS_TRANS_NO_WRITECOUNT))
+		sb_start_intwrite(mp->m_super);
+	xfs_trans_set_context(tp);
+	tp->t_flags = flags;
+	tp->t_mountp = mp;
+	INIT_LIST_HEAD(&tp->t_items);
+	INIT_LIST_HEAD(&tp->t_busy);
+	INIT_LIST_HEAD(&tp->t_dfops);
+	tp->t_highest_agno = NULLAGNUMBER;
+	return tp;
+}
+
 int
 xfs_trans_alloc(
 	struct xfs_mount	*mp,
@@ -253,34 +245,16 @@ xfs_trans_alloc(
 	bool			want_retry = true;
 	int			error;
 
+	ASSERT(resp->tr_logres > 0);
+
 	/*
 	 * Allocate the handle before we do our freeze accounting and setting up
 	 * GFP_NOFS allocation context so that we avoid lockdep false positives
 	 * by doing GFP_KERNEL allocations inside sb_start_intwrite().
 	 */
 retry:
-	tp = kmem_cache_zalloc(xfs_trans_cache, GFP_KERNEL | __GFP_NOFAIL);
-	if (!(flags & XFS_TRANS_NO_WRITECOUNT))
-		sb_start_intwrite(mp->m_super);
-	xfs_trans_set_context(tp);
-
-	/*
-	 * Zero-reservation ("empty") transactions can't modify anything, so
-	 * they're allowed to run while we're frozen.
-	 */
-	WARN_ON(resp->tr_logres > 0 &&
-		mp->m_super->s_writers.frozen == SB_FREEZE_COMPLETE);
-	ASSERT(!(flags & XFS_TRANS_RES_FDBLKS) ||
-	       xfs_has_lazysbcount(mp));
-
-	tp->t_magic = XFS_TRANS_HEADER_MAGIC;
-	tp->t_flags = flags;
-	tp->t_mountp = mp;
-	INIT_LIST_HEAD(&tp->t_items);
-	INIT_LIST_HEAD(&tp->t_busy);
-	INIT_LIST_HEAD(&tp->t_dfops);
-	tp->t_highest_agno = NULLAGNUMBER;
-
+	tp = __xfs_trans_alloc(mp, flags);
+	WARN_ON(mp->m_super->s_writers.frozen == SB_FREEZE_COMPLETE);
 	error = xfs_trans_reserve(tp, resp, blocks, rtextents);
 	if (error == -ENOSPC && want_retry) {
 		xfs_trans_cancel(tp);
@@ -324,14 +298,11 @@ retry:
  * where we can be grabbing buffers at the same time that freeze is trying to
  * drain the buffer LRU list.
  */
-int
+struct xfs_trans *
 xfs_trans_alloc_empty(
-	struct xfs_mount		*mp,
-	struct xfs_trans		**tpp)
+	struct xfs_mount		*mp)
 {
-	struct xfs_trans_res		resv = {0};
-
-	return xfs_trans_alloc(mp, &resv, 0, 0, XFS_TRANS_NO_WRITECOUNT, tpp);
+	return __xfs_trans_alloc(mp, XFS_TRANS_NO_WRITECOUNT);
 }
 
 /*
@@ -420,6 +391,8 @@ xfs_trans_mod_sb(
 			ASSERT(tp->t_rtx_res_used <= tp->t_rtx_res);
 		}
 		tp->t_frextents_delta += delta;
+		if (xfs_has_rtgroups(mp))
+			flags &= ~XFS_TRANS_SB_DIRTY;
 		break;
 	case XFS_TRANS_SB_RES_FREXTENTS:
 		/*
@@ -429,6 +402,8 @@ xfs_trans_mod_sb(
 		 */
 		ASSERT(delta < 0);
 		tp->t_res_frextents_delta += delta;
+		if (xfs_has_rtgroups(mp))
+			flags &= ~XFS_TRANS_SB_DIRTY;
 		break;
 	case XFS_TRANS_SB_DBLOCKS:
 		tp->t_dblocks_delta += delta;
@@ -455,6 +430,10 @@ xfs_trans_mod_sb(
 	case XFS_TRANS_SB_REXTSLOG:
 		tp->t_rextslog_delta += delta;
 		break;
+	case XFS_TRANS_SB_RGCOUNT:
+		ASSERT(delta > 0);
+		tp->t_rgcount_delta += delta;
+		break;
 	default:
 		ASSERT(0);
 		return;
@@ -473,19 +452,17 @@ xfs_trans_mod_sb(
  */
 STATIC void
 xfs_trans_apply_sb_deltas(
-	xfs_trans_t	*tp)
+	struct xfs_trans	*tp)
 {
-	struct xfs_dsb	*sbp;
-	struct xfs_buf	*bp;
-	int		whole = 0;
-
-	bp = xfs_trans_getsb(tp);
-	sbp = bp->b_addr;
+	struct xfs_mount	*mp = tp->t_mountp;
+	struct xfs_buf		*bp = xfs_trans_getsb(tp);
+	struct xfs_dsb		*sbp = bp->b_addr;
+	int			whole = 0;
 
 	/*
 	 * Only update the superblock counters if we are logging them
 	 */
-	if (!xfs_has_lazysbcount((tp->t_mountp))) {
+	if (!xfs_has_lazysbcount(mp)) {
 		if (tp->t_icount_delta)
 			be64_add_cpu(&sbp->sb_icount, tp->t_icount_delta);
 		if (tp->t_ifree_delta)
@@ -497,21 +474,22 @@ xfs_trans_apply_sb_deltas(
 	}
 
 	/*
-	 * Updating frextents requires careful handling because it does not
-	 * behave like the lazysb counters because we cannot rely on log
-	 * recovery in older kenels to recompute the value from the rtbitmap.
-	 * This means that the ondisk frextents must be consistent with the
-	 * rtbitmap.
+	 * sb_frextents was added to the lazy sb counters when the rt groups
+	 * feature was introduced.  This is possible because we know that all
+	 * kernels supporting rtgroups will also recompute frextents from the
+	 * realtime bitmap.
+	 *
+	 * For older file systems, updating frextents requires careful handling
+	 * because we cannot rely on log recovery in older kernels to recompute
+	 * the value from the rtbitmap.  This means that the ondisk frextents
+	 * must be consistent with the rtbitmap.
 	 *
 	 * Therefore, log the frextents change to the ondisk superblock and
 	 * update the incore superblock so that future calls to xfs_log_sb
 	 * write the correct value ondisk.
-	 *
-	 * Don't touch m_frextents because it includes incore reservations,
-	 * and those are handled by the unreserve function.
 	 */
-	if (tp->t_frextents_delta || tp->t_res_frextents_delta) {
-		struct xfs_mount	*mp = tp->t_mountp;
+	if ((tp->t_frextents_delta || tp->t_res_frextents_delta) &&
+	    !xfs_has_rtgroups(mp)) {
 		int64_t			rtxdelta;
 
 		rtxdelta = tp->t_frextents_delta + tp->t_res_frextents_delta;
@@ -524,6 +502,8 @@ xfs_trans_apply_sb_deltas(
 
 	if (tp->t_dblocks_delta) {
 		be64_add_cpu(&sbp->sb_dblocks, tp->t_dblocks_delta);
+		mp->m_ddev_targp->bt_nr_sectors +=
+			XFS_FSB_TO_BB(mp, tp->t_dblocks_delta);
 		whole = 1;
 	}
 	if (tp->t_agcount_delta) {
@@ -536,6 +516,18 @@ xfs_trans_apply_sb_deltas(
 	}
 	if (tp->t_rextsize_delta) {
 		be32_add_cpu(&sbp->sb_rextsize, tp->t_rextsize_delta);
+
+		/*
+		 * Because the ondisk sb records rtgroup size in units of rt
+		 * extents, any time we update the rt extent size we have to
+		 * recompute the ondisk rtgroup block log.  The incore values
+		 * will be recomputed in xfs_trans_unreserve_and_mod_sb.
+		 */
+		if (xfs_has_rtgroups(mp)) {
+			sbp->sb_rgblklog = xfs_compute_rgblklog(
+						be32_to_cpu(sbp->sb_rgextents),
+						be32_to_cpu(sbp->sb_rextsize));
+		}
 		whole = 1;
 	}
 	if (tp->t_rbmblocks_delta) {
@@ -544,6 +536,8 @@ xfs_trans_apply_sb_deltas(
 	}
 	if (tp->t_rblocks_delta) {
 		be64_add_cpu(&sbp->sb_rblocks, tp->t_rblocks_delta);
+		mp->m_rtdev_targp->bt_nr_sectors +=
+			XFS_FSB_TO_BB(mp, tp->t_rblocks_delta);
 		whole = 1;
 	}
 	if (tp->t_rextents_delta) {
@@ -554,6 +548,10 @@ xfs_trans_apply_sb_deltas(
 		sbp->sb_rextslog += tp->t_rextslog_delta;
 		whole = 1;
 	}
+	if (tp->t_rgcount_delta) {
+		be32_add_cpu(&sbp->sb_rgcount, tp->t_rgcount_delta);
+		whole = 1;
+	}
 
 	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);
 	if (whole)
@@ -618,7 +616,7 @@ xfs_trans_unreserve_and_mod_sb(
 	}
 
 	ASSERT(tp->t_rtx_res || tp->t_frextents_delta >= 0);
-	if (tp->t_flags & XFS_TRANS_SB_DIRTY) {
+	if (xfs_has_rtgroups(mp) || (tp->t_flags & XFS_TRANS_SB_DIRTY)) {
 		rtxdelta += tp->t_frextents_delta;
 		ASSERT(rtxdelta >= 0);
 	}
@@ -651,23 +649,21 @@ xfs_trans_unreserve_and_mod_sb(
 	mp->m_sb.sb_icount += idelta;
 	mp->m_sb.sb_ifree += ifreedelta;
 	/*
-	 * Do not touch sb_frextents here because we are dealing with incore
-	 * reservation.  sb_frextents is not part of the lazy sb counters so it
-	 * must be consistent with the ondisk rtbitmap and must never include
-	 * incore reservations.
+	 * Do not touch sb_frextents here because it is handled in
+	 * xfs_trans_apply_sb_deltas for file systems where it isn't a lazy
+	 * counter anyway.
 	 */
 	mp->m_sb.sb_dblocks += tp->t_dblocks_delta;
 	mp->m_sb.sb_agcount += tp->t_agcount_delta;
 	mp->m_sb.sb_imax_pct += tp->t_imaxpct_delta;
-	mp->m_sb.sb_rextsize += tp->t_rextsize_delta;
-	if (tp->t_rextsize_delta) {
-		mp->m_rtxblklog = log2_if_power2(mp->m_sb.sb_rextsize);
-		mp->m_rtxblkmask = mask64_if_power2(mp->m_sb.sb_rextsize);
-	}
+	if (tp->t_rextsize_delta)
+		xfs_mount_sb_set_rextsize(mp, &mp->m_sb,
+				mp->m_sb.sb_rextsize + tp->t_rextsize_delta);
 	mp->m_sb.sb_rbmblocks += tp->t_rbmblocks_delta;
 	mp->m_sb.sb_rblocks += tp->t_rblocks_delta;
 	mp->m_sb.sb_rextents += tp->t_rextents_delta;
 	mp->m_sb.sb_rextslog += tp->t_rextslog_delta;
+	mp->m_sb.sb_rgcount += tp->t_rgcount_delta;
 	spin_unlock(&mp->m_sb_lock);
 
 	/*
@@ -718,142 +714,15 @@ xfs_trans_free_items(
 
 	list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
 		xfs_trans_del_item(lip);
-		if (abort)
+		if (abort) {
+			trace_xfs_trans_free_abort(lip);
 			set_bit(XFS_LI_ABORTED, &lip->li_flags);
+		}
 		if (lip->li_ops->iop_release)
 			lip->li_ops->iop_release(lip);
 	}
 }
 
-static inline void
-xfs_log_item_batch_insert(
-	struct xfs_ail		*ailp,
-	struct xfs_ail_cursor	*cur,
-	struct xfs_log_item	**log_items,
-	int			nr_items,
-	xfs_lsn_t		commit_lsn)
-{
-	int	i;
-
-	spin_lock(&ailp->ail_lock);
-	/* xfs_trans_ail_update_bulk drops ailp->ail_lock */
-	xfs_trans_ail_update_bulk(ailp, cur, log_items, nr_items, commit_lsn);
-
-	for (i = 0; i < nr_items; i++) {
-		struct xfs_log_item *lip = log_items[i];
-
-		if (lip->li_ops->iop_unpin)
-			lip->li_ops->iop_unpin(lip, 0);
-	}
-}
-
-/*
- * Bulk operation version of xfs_trans_committed that takes a log vector of
- * items to insert into the AIL. This uses bulk AIL insertion techniques to
- * minimise lock traffic.
- *
- * If we are called with the aborted flag set, it is because a log write during
- * a CIL checkpoint commit has failed. In this case, all the items in the
- * checkpoint have already gone through iop_committed and iop_committing, which
- * means that checkpoint commit abort handling is treated exactly the same
- * as an iclog write error even though we haven't started any IO yet. Hence in
- * this case all we need to do is iop_committed processing, followed by an
- * iop_unpin(aborted) call.
- *
- * The AIL cursor is used to optimise the insert process. If commit_lsn is not
- * at the end of the AIL, the insert cursor avoids the need to walk
- * the AIL to find the insertion point on every xfs_log_item_batch_insert()
- * call. This saves a lot of needless list walking and is a net win, even
- * though it slightly increases that amount of AIL lock traffic to set it up
- * and tear it down.
- */
-void
-xfs_trans_committed_bulk(
-	struct xfs_ail		*ailp,
-	struct list_head	*lv_chain,
-	xfs_lsn_t		commit_lsn,
-	bool			aborted)
-{
-#define LOG_ITEM_BATCH_SIZE	32
-	struct xfs_log_item	*log_items[LOG_ITEM_BATCH_SIZE];
-	struct xfs_log_vec	*lv;
-	struct xfs_ail_cursor	cur;
-	int			i = 0;
-
-	spin_lock(&ailp->ail_lock);
-	xfs_trans_ail_cursor_last(ailp, &cur, commit_lsn);
-	spin_unlock(&ailp->ail_lock);
-
-	/* unpin all the log items */
-	list_for_each_entry(lv, lv_chain, lv_list) {
-		struct xfs_log_item	*lip = lv->lv_item;
-		xfs_lsn_t		item_lsn;
-
-		if (aborted)
-			set_bit(XFS_LI_ABORTED, &lip->li_flags);
-
-		if (lip->li_ops->flags & XFS_ITEM_RELEASE_WHEN_COMMITTED) {
-			lip->li_ops->iop_release(lip);
-			continue;
-		}
-
-		if (lip->li_ops->iop_committed)
-			item_lsn = lip->li_ops->iop_committed(lip, commit_lsn);
-		else
-			item_lsn = commit_lsn;
-
-		/* item_lsn of -1 means the item needs no further processing */
-		if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
-			continue;
-
-		/*
-		 * if we are aborting the operation, no point in inserting the
-		 * object into the AIL as we are in a shutdown situation.
-		 */
-		if (aborted) {
-			ASSERT(xlog_is_shutdown(ailp->ail_log));
-			if (lip->li_ops->iop_unpin)
-				lip->li_ops->iop_unpin(lip, 1);
-			continue;
-		}
-
-		if (item_lsn != commit_lsn) {
-
-			/*
-			 * Not a bulk update option due to unusual item_lsn.
-			 * Push into AIL immediately, rechecking the lsn once
-			 * we have the ail lock. Then unpin the item. This does
-			 * not affect the AIL cursor the bulk insert path is
-			 * using.
-			 */
-			spin_lock(&ailp->ail_lock);
-			if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0)
-				xfs_trans_ail_update(ailp, lip, item_lsn);
-			else
-				spin_unlock(&ailp->ail_lock);
-			if (lip->li_ops->iop_unpin)
-				lip->li_ops->iop_unpin(lip, 0);
-			continue;
-		}
-
-		/* Item is a candidate for bulk AIL insert.  */
-		log_items[i++] = lv->lv_item;
-		if (i >= LOG_ITEM_BATCH_SIZE) {
-			xfs_log_item_batch_insert(ailp, &cur, log_items,
-					LOG_ITEM_BATCH_SIZE, commit_lsn);
-			i = 0;
-		}
-	}
-
-	/* make sure we insert the remainder! */
-	if (i)
-		xfs_log_item_batch_insert(ailp, &cur, log_items, i, commit_lsn);
-
-	spin_lock(&ailp->ail_lock);
-	xfs_trans_ail_cursor_done(&cur);
-	spin_unlock(&ailp->ail_lock);
-}
-
 /*
  * Sort transaction items prior to running precommit operations. This will
  * attempt to order the items such that they will always be locked in the same
@@ -963,29 +832,17 @@ __xfs_trans_commit(
 
 	trace_xfs_trans_commit(tp, _RET_IP_);
 
-	error = xfs_trans_run_precommits(tp);
-	if (error) {
-		if (tp->t_flags & XFS_TRANS_PERM_LOG_RES)
-			xfs_defer_cancel(tp);
-		goto out_unreserve;
-	}
-
 	/*
-	 * Finish deferred items on final commit. Only permanent transactions
-	 * should ever have deferred ops.
+	 * Commit per-transaction changes that are not already tracked through
+	 * log items.  This can add dirty log items to the transaction.
 	 */
-	WARN_ON_ONCE(!list_empty(&tp->t_dfops) &&
-		     !(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
-	if (!regrant && (tp->t_flags & XFS_TRANS_PERM_LOG_RES)) {
-		error = xfs_defer_finish_noroll(&tp);
-		if (error)
-			goto out_unreserve;
+	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
+		xfs_trans_apply_sb_deltas(tp);
+	xfs_trans_apply_dquot_deltas(tp);
 
-		/* Run precommits from final tx in defer chain. */
-		error = xfs_trans_run_precommits(tp);
-		if (error)
-			goto out_unreserve;
-	}
+	error = xfs_trans_run_precommits(tp);
+	if (error)
+		goto out_unreserve;
 
 	/*
 	 * If there is nothing to be logged by the transaction,
@@ -1010,13 +867,6 @@ __xfs_trans_commit(
 
 	ASSERT(tp->t_ticket != NULL);
 
-	/*
-	 * If we need to update the superblock, then do it now.
-	 */
-	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
-		xfs_trans_apply_sb_deltas(tp);
-	xfs_trans_apply_dquot_deltas(tp);
-
 	xlog_cil_commit(log, tp, &commit_seq, regrant);
 
 	xfs_trans_free(tp);
@@ -1042,7 +892,7 @@ out_unreserve:
 	 * the dqinfo portion to be.  All that means is that we have some
 	 * (non-persistent) quota reservations that need to be unreserved.
 	 */
-	xfs_trans_unreserve_and_mod_dquots(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp, true);
 	if (tp->t_ticket) {
 		if (regrant && !xlog_is_shutdown(log))
 			xfs_log_ticket_regrant(log, tp->t_ticket);
@@ -1061,6 +911,20 @@ int
 xfs_trans_commit(
 	struct xfs_trans	*tp)
 {
+	/*
+	 * Finish deferred items on final commit. Only permanent transactions
+	 * should ever have deferred ops.
+	 */
+	WARN_ON_ONCE(!list_empty(&tp->t_dfops) &&
+		     !(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
+	if (tp->t_flags & XFS_TRANS_PERM_LOG_RES) {
+		int error = xfs_defer_finish_noroll(&tp);
+		if (error) {
+			xfs_trans_cancel(tp);
+			return error;
+		}
+	}
+
 	return __xfs_trans_commit(tp, false);
 }
 
@@ -1122,7 +986,7 @@ xfs_trans_cancel(
 	}
 #endif
 	xfs_trans_unreserve_and_mod_sb(tp);
-	xfs_trans_unreserve_and_mod_dquots(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp, false);
 
 	if (tp->t_ticket) {
 		xfs_log_ticket_ungrant(log, tp->t_ticket);
@@ -1134,51 +998,57 @@ xfs_trans_cancel(
 }
 
 /*
- * Roll from one trans in the sequence of PERMANENT transactions to
- * the next: permanent transactions are only flushed out when
- * committed with xfs_trans_commit(), but we still want as soon
- * as possible to let chunks of it go to the log. So we commit the
- * chunk we've been working on and get a new transaction to continue.
+ * Roll from one trans in the sequence of PERMANENT transactions to the next:
+ * permanent transactions are only flushed out when committed with
+ * xfs_trans_commit(), but we still want as soon as possible to let chunks of it
+ * go to the log.  So we commit the chunk we've been working on and get a new
+ * transaction to continue.
  */
 int
 xfs_trans_roll(
 	struct xfs_trans	**tpp)
 {
-	struct xfs_trans	*trans = *tpp;
-	struct xfs_trans_res	tres;
+	struct xfs_trans	*tp = *tpp;
+	unsigned int		log_res = tp->t_log_res;
+	unsigned int		log_count = tp->t_log_count;
 	int			error;
 
-	trace_xfs_trans_roll(trans, _RET_IP_);
+	trace_xfs_trans_roll(tp, _RET_IP_);
+
+	ASSERT(log_res > 0);
 
 	/*
 	 * Copy the critical parameters from one trans to the next.
 	 */
-	tres.tr_logres = trans->t_log_res;
-	tres.tr_logcount = trans->t_log_count;
-
-	*tpp = xfs_trans_dup(trans);
+	*tpp = xfs_trans_dup(tp);
 
 	/*
 	 * Commit the current transaction.
-	 * If this commit failed, then it'd just unlock those items that
-	 * are not marked ihold. That also means that a filesystem shutdown
-	 * is in progress. The caller takes the responsibility to cancel
-	 * the duplicate transaction that gets returned.
+	 *
+	 * If this commit failed, then it'd just unlock those items that are not
+	 * marked ihold. That also means that a filesystem shutdown is in
+	 * progress.  The caller takes the responsibility to cancel the
+	 * duplicate transaction that gets returned.
 	 */
-	error = __xfs_trans_commit(trans, true);
+	error = __xfs_trans_commit(tp, true);
 	if (error)
 		return error;
 
 	/*
 	 * Reserve space in the log for the next transaction.
-	 * This also pushes items in the "AIL", the list of logged items,
-	 * out to disk if they are taking up space at the tail of the log
-	 * that we want to use.  This requires that either nothing be locked
-	 * across this call, or that anything that is locked be logged in
-	 * the prior and the next transactions.
+	 *
+	 * This also pushes items in the AIL out to disk if they are taking up
+	 * space at the tail of the log that we want to use.  This requires that
+	 * either nothing be locked across this call, or that anything that is
+	 * locked be logged in the prior and the next transactions.
 	 */
-	tres.tr_logflags = XFS_TRANS_PERM_LOG_RES;
-	return xfs_trans_reserve(*tpp, &tres, 0, 0);
+	tp = *tpp;
+	error = xfs_log_regrant(tp->t_mountp, tp->t_ticket);
+	if (error)
+		return error;
+	tp->t_log_res = log_res;
+	tp->t_log_count = log_count;
+	return 0;
 }
 
 /*
@@ -1254,9 +1124,18 @@ xfs_trans_reserve_more(
 	unsigned int		blocks,
 	unsigned int		rtextents)
 {
-	struct xfs_trans_res	resv = { };
-
-	return xfs_trans_reserve(tp, &resv, blocks, rtextents);
+	bool			rsvd = tp->t_flags & XFS_TRANS_RESERVE;
+
+	if (blocks && xfs_dec_fdblocks(tp->t_mountp, blocks, rsvd))
+		return -ENOSPC;
+	if (rtextents && xfs_dec_frextents(tp->t_mountp, rtextents)) {
+		if (blocks)
+			xfs_add_fdblocks(tp->t_mountp, blocks);
+		return -ENOSPC;
+	}
+	tp->t_blk_res += blocks;
+	tp->t_rtx_res += rtextents;
+	return 0;
 }
 
 /*
@@ -1271,14 +1150,13 @@ xfs_trans_reserve_more_inode(
 	unsigned int		rblocks,
 	bool			force_quota)
 {
-	struct xfs_trans_res	resv = { };
 	struct xfs_mount	*mp = ip->i_mount;
 	unsigned int		rtx = xfs_extlen_to_rtxlen(mp, rblocks);
 	int			error;
 
 	xfs_assert_ilocked(ip, XFS_ILOCK_EXCL);
 
-	error = xfs_trans_reserve(tp, &resv, dblocks, rtx);
+	error = xfs_trans_reserve_more(tp, dblocks, rtx);
 	if (error)
 		return error;
 
@@ -1374,6 +1252,9 @@ retry:
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
 
+	if (xfs_is_metadir_inode(ip))
+		goto out;
+
 	error = xfs_qm_dqattach_locked(ip, false);
 	if (error) {
 		/* Caller should have allocated the dquots! */
@@ -1391,11 +1272,26 @@ retry:
 	gdqp = (new_gdqp != ip->i_gdquot) ? new_gdqp : NULL;
 	pdqp = (new_pdqp != ip->i_pdquot) ? new_pdqp : NULL;
 	if (udqp || gdqp || pdqp) {
+		xfs_filblks_t	dblocks, rblocks;
 		unsigned int	qflags = XFS_QMOPT_RES_REGBLKS;
+		bool		isrt = XFS_IS_REALTIME_INODE(ip);
 
 		if (force)
 			qflags |= XFS_QMOPT_FORCE_RES;
 
+		if (isrt) {
+			error = xfs_iread_extents(tp, ip, XFS_DATA_FORK);
+			if (error)
+				goto out_cancel;
+		}
+
+		xfs_inode_count_blocks(tp, ip, &dblocks, &rblocks);
+
+		if (isrt)
+			rblocks += ip->i_delayed_blks;
+		else
+			dblocks += ip->i_delayed_blks;
+
 		/*
 		 * Reserve enough quota to handle blocks on disk and reserved
 		 * for a delayed allocation.  We'll actually transfer the
@@ -1403,8 +1299,20 @@ retry:
 		 * though that part is only semi-transactional.
 		 */
 		error = xfs_trans_reserve_quota_bydquots(tp, mp, udqp, gdqp,
-				pdqp, ip->i_nblocks + ip->i_delayed_blks,
-				1, qflags);
+				pdqp, dblocks, 1, qflags);
+		if ((error == -EDQUOT || error == -ENOSPC) && !retried) {
+			xfs_trans_cancel(tp);
+			xfs_blockgc_free_dquots(mp, udqp, gdqp, pdqp, 0);
+			retried = true;
+			goto retry;
+		}
+		if (error)
+			goto out_cancel;
+
+		/* Do the same for realtime. */
+		qflags = XFS_QMOPT_RES_RTBLKS | (qflags & XFS_QMOPT_FORCE_RES);
+		error = xfs_trans_reserve_quota_bydquots(tp, mp, udqp, gdqp,
+				pdqp, rblocks, 0, qflags);
 		if ((error == -EDQUOT || error == -ENOSPC) && !retried) {
 			xfs_trans_cancel(tp);
 			xfs_blockgc_free_dquots(mp, udqp, gdqp, pdqp, 0);
@@ -1415,6 +1323,7 @@ retry:
 			goto out_cancel;
 	}
 
+out:
 	*tpp = tp;
 	return 0;
 
@@ -1511,5 +1420,8 @@ done:
 
 out_cancel:
 	xfs_trans_cancel(tp);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	if (dp != ip)
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	return error;
 }