1 files changed, 286 insertions, 82 deletions

diff --git a/fs/xfs/libxfs/xfs_defer.c b/fs/xfs/libxfs/xfs_defer.c
index 22557527cfdb..3a78a189ea01 100644
--- a/fs/xfs/libxfs/xfs_defer.c
+++ b/fs/xfs/libxfs/xfs_defer.c
@@ -16,6 +16,8 @@
 #include "xfs_inode.h"
 #include "xfs_inode_item.h"
 #include "xfs_trace.h"
+#include "xfs_icache.h"
+#include "xfs_log.h"
 
 /*
  * Deferred Operations in XFS
@@ -178,6 +180,19 @@ static const struct xfs_defer_op_type *defer_op_types[] = {
 	[XFS_DEFER_OPS_TYPE_AGFL_FREE]	= &xfs_agfl_free_defer_type,
 };
 
+static void
+xfs_defer_create_intent(
+	struct xfs_trans		*tp,
+	struct xfs_defer_pending	*dfp,
+	bool				sort)
+{
+	const struct xfs_defer_op_type	*ops = defer_op_types[dfp->dfp_type];
+
+	if (!dfp->dfp_intent)
+		dfp->dfp_intent = ops->create_intent(tp, &dfp->dfp_work,
+						     dfp->dfp_count, sort);
+}
+
 /*
  * For each pending item in the intake list, log its intent item and the
  * associated extents, then add the entire intake list to the end of
@@ -187,17 +202,11 @@ STATIC void
 xfs_defer_create_intents(
 	struct xfs_trans		*tp)
 {
-	struct list_head		*li;
 	struct xfs_defer_pending	*dfp;
-	const struct xfs_defer_op_type	*ops;
 
 	list_for_each_entry(dfp, &tp->t_dfops, dfp_list) {
-		ops = defer_op_types[dfp->dfp_type];
-		dfp->dfp_intent = ops->create_intent(tp, dfp->dfp_count);
 		trace_xfs_defer_create_intent(tp->t_mountp, dfp);
-		list_sort(tp->t_mountp, &dfp->dfp_work, ops->diff_items);
-		list_for_each(li, &dfp->dfp_work)
-			ops->log_item(tp, dfp->dfp_intent, li);
+		xfs_defer_create_intent(tp, dfp, true);
 	}
 }
 
@@ -234,10 +243,13 @@ xfs_defer_trans_roll(
 	struct xfs_log_item		*lip;
 	struct xfs_buf			*bplist[XFS_DEFER_OPS_NR_BUFS];
 	struct xfs_inode		*iplist[XFS_DEFER_OPS_NR_INODES];
+	unsigned int			ordered = 0; /* bitmap */
 	int				bpcount = 0, ipcount = 0;
 	int				i;
 	int				error;
 
+	BUILD_BUG_ON(NBBY * sizeof(ordered) < XFS_DEFER_OPS_NR_BUFS);
+
 	list_for_each_entry(lip, &tp->t_items, li_trans) {
 		switch (lip->li_type) {
 		case XFS_LI_BUF:
@@ -248,7 +260,10 @@ xfs_defer_trans_roll(
 					ASSERT(0);
 					return -EFSCORRUPTED;
 				}
-				xfs_trans_dirty_buf(tp, bli->bli_buf);
+				if (bli->bli_flags & XFS_BLI_ORDERED)
+					ordered |= (1U << bpcount);
+				else
+					xfs_trans_dirty_buf(tp, bli->bli_buf);
 				bplist[bpcount++] = bli->bli_buf;
 			}
 			break;
@@ -289,6 +304,8 @@ xfs_defer_trans_roll(
 	/* Rejoin the buffers and dirty them so the log moves forward. */
 	for (i = 0; i < bpcount; i++) {
 		xfs_trans_bjoin(tp, bplist[i]);
+		if (ordered & (1U << i))
+			xfs_trans_ordered_buf(tp, bplist[i]);
 		xfs_trans_bhold(tp, bplist[i]);
 	}
 
@@ -346,6 +363,106 @@ xfs_defer_cancel_list(
 }
 
 /*
+ * Prevent a log intent item from pinning the tail of the log by logging a
+ * done item to release the intent item; and then log a new intent item.
+ * The caller should provide a fresh transaction and roll it after we're done.
+ */
+static int
+xfs_defer_relog(
+	struct xfs_trans		**tpp,
+	struct list_head		*dfops)
+{
+	struct xlog			*log = (*tpp)->t_mountp->m_log;
+	struct xfs_defer_pending	*dfp;
+	xfs_lsn_t			threshold_lsn = NULLCOMMITLSN;
+
+
+	ASSERT((*tpp)->t_flags & XFS_TRANS_PERM_LOG_RES);
+
+	list_for_each_entry(dfp, dfops, dfp_list) {
+		/*
+		 * If the log intent item for this deferred op is not a part of
+		 * the current log checkpoint, relog the intent item to keep
+		 * the log tail moving forward.  We're ok with this being racy
+		 * because an incorrect decision means we'll be a little slower
+		 * at pushing the tail.
+		 */
+		if (dfp->dfp_intent == NULL ||
+		    xfs_log_item_in_current_chkpt(dfp->dfp_intent))
+			continue;
+
+		/*
+		 * Figure out where we need the tail to be in order to maintain
+		 * the minimum required free space in the log.  Only sample
+		 * the log threshold once per call.
+		 */
+		if (threshold_lsn == NULLCOMMITLSN) {
+			threshold_lsn = xlog_grant_push_threshold(log, 0);
+			if (threshold_lsn == NULLCOMMITLSN)
+				break;
+		}
+		if (XFS_LSN_CMP(dfp->dfp_intent->li_lsn, threshold_lsn) >= 0)
+			continue;
+
+		trace_xfs_defer_relog_intent((*tpp)->t_mountp, dfp);
+		XFS_STATS_INC((*tpp)->t_mountp, defer_relog);
+		dfp->dfp_intent = xfs_trans_item_relog(dfp->dfp_intent, *tpp);
+	}
+
+	if ((*tpp)->t_flags & XFS_TRANS_DIRTY)
+		return xfs_defer_trans_roll(tpp);
+	return 0;
+}
+
+/*
+ * Log an intent-done item for the first pending intent, and finish the work
+ * items.
+ */
+static int
+xfs_defer_finish_one(
+	struct xfs_trans		*tp,
+	struct xfs_defer_pending	*dfp)
+{
+	const struct xfs_defer_op_type	*ops = defer_op_types[dfp->dfp_type];
+	void				*state = NULL;
+	struct list_head		*li, *n;
+	int				error;
+
+	trace_xfs_defer_pending_finish(tp->t_mountp, dfp);
+
+	dfp->dfp_done = ops->create_done(tp, dfp->dfp_intent, dfp->dfp_count);
+	list_for_each_safe(li, n, &dfp->dfp_work) {
+		list_del(li);
+		dfp->dfp_count--;
+		error = ops->finish_item(tp, li, dfp->dfp_done, &state);
+		if (error == -EAGAIN) {
+			/*
+			 * Caller wants a fresh transaction; put the work item
+			 * back on the list and log a new log intent item to
+			 * replace the old one.  See "Requesting a Fresh
+			 * Transaction while Finishing Deferred Work" above.
+			 */
+			list_add(li, &dfp->dfp_work);
+			dfp->dfp_count++;
+			dfp->dfp_done = NULL;
+			dfp->dfp_intent = NULL;
+			xfs_defer_create_intent(tp, dfp, false);
+		}
+
+		if (error)
+			goto out;
+	}
+
+	/* Done with the dfp, free it. */
+	list_del(&dfp->dfp_list);
+	kmem_free(dfp);
+out:
+	if (ops->finish_cleanup)
+		ops->finish_cleanup(tp, state, error);
+	return error;
+}
+
+/*
  * Finish all the pending work.  This involves logging intent items for
  * any work items that wandered in since the last transaction roll (if
  * one has even happened), rolling the transaction, and finishing the
@@ -358,11 +475,7 @@ xfs_defer_finish_noroll(
 	struct xfs_trans		**tp)
 {
 	struct xfs_defer_pending	*dfp;
-	struct list_head		*li;
-	struct list_head		*n;
-	void				*state;
 	int				error = 0;
-	const struct xfs_defer_op_type	*ops;
 	LIST_HEAD(dop_pending);
 
 	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
@@ -371,87 +484,44 @@ xfs_defer_finish_noroll(
 
 	/* Until we run out of pending work to finish... */
 	while (!list_empty(&dop_pending) || !list_empty(&(*tp)->t_dfops)) {
-		/* log intents and pull in intake items */
-		xfs_defer_create_intents(*tp);
-		list_splice_tail_init(&(*tp)->t_dfops, &dop_pending);
-
 		/*
-		 * Roll the transaction.
+		 * Deferred items that are created in the process of finishing
+		 * other deferred work items should be queued at the head of
+		 * the pending list, which puts them ahead of the deferred work
+		 * that was created by the caller.  This keeps the number of
+		 * pending work items to a minimum, which decreases the amount
+		 * of time that any one intent item can stick around in memory,
+		 * pinning the log tail.
 		 */
+		xfs_defer_create_intents(*tp);
+		list_splice_init(&(*tp)->t_dfops, &dop_pending);
+
 		error = xfs_defer_trans_roll(tp);
 		if (error)
-			goto out;
+			goto out_shutdown;
+
+		/* Possibly relog intent items to keep the log moving. */
+		error = xfs_defer_relog(tp, &dop_pending);
+		if (error)
+			goto out_shutdown;
 
-		/* Log an intent-done item for the first pending item. */
 		dfp = list_first_entry(&dop_pending, struct xfs_defer_pending,
 				       dfp_list);
-		ops = defer_op_types[dfp->dfp_type];
-		trace_xfs_defer_pending_finish((*tp)->t_mountp, dfp);
-		dfp->dfp_done = ops->create_done(*tp, dfp->dfp_intent,
-				dfp->dfp_count);
-
-		/* Finish the work items. */
-		state = NULL;
-		list_for_each_safe(li, n, &dfp->dfp_work) {
-			list_del(li);
-			dfp->dfp_count--;
-			error = ops->finish_item(*tp, li, dfp->dfp_done,
-					&state);
-			if (error == -EAGAIN) {
-				/*
-				 * Caller wants a fresh transaction;
-				 * put the work item back on the list
-				 * and jump out.
-				 */
-				list_add(li, &dfp->dfp_work);
-				dfp->dfp_count++;
-				break;
-			} else if (error) {
-				/*
-				 * Clean up after ourselves and jump out.
-				 * xfs_defer_cancel will take care of freeing
-				 * all these lists and stuff.
-				 */
-				if (ops->finish_cleanup)
-					ops->finish_cleanup(*tp, state, error);
-				goto out;
-			}
-		}
-		if (error == -EAGAIN) {
-			/*
-			 * Caller wants a fresh transaction, so log a
-			 * new log intent item to replace the old one
-			 * and roll the transaction.  See "Requesting
-			 * a Fresh Transaction while Finishing
-			 * Deferred Work" above.
-			 */
-			dfp->dfp_intent = ops->create_intent(*tp,
-					dfp->dfp_count);
-			dfp->dfp_done = NULL;
-			list_for_each(li, &dfp->dfp_work)
-				ops->log_item(*tp, dfp->dfp_intent, li);
-		} else {
-			/* Done with the dfp, free it. */
-			list_del(&dfp->dfp_list);
-			kmem_free(dfp);
-		}
-
-		if (ops->finish_cleanup)
-			ops->finish_cleanup(*tp, state, error);
-	}
-
-out:
-	if (error) {
-		xfs_defer_trans_abort(*tp, &dop_pending);
-		xfs_force_shutdown((*tp)->t_mountp, SHUTDOWN_CORRUPT_INCORE);
-		trace_xfs_defer_finish_error(*tp, error);
-		xfs_defer_cancel_list((*tp)->t_mountp, &dop_pending);
-		xfs_defer_cancel(*tp);
-		return error;
+		error = xfs_defer_finish_one(*tp, dfp);
+		if (error && error != -EAGAIN)
+			goto out_shutdown;
 	}
 
 	trace_xfs_defer_finish_done(*tp, _RET_IP_);
 	return 0;
+
+out_shutdown:
+	xfs_defer_trans_abort(*tp, &dop_pending);
+	xfs_force_shutdown((*tp)->t_mountp, SHUTDOWN_CORRUPT_INCORE);
+	trace_xfs_defer_finish_error(*tp, error);
+	xfs_defer_cancel_list((*tp)->t_mountp, &dop_pending);
+	xfs_defer_cancel(*tp);
+	return error;
 }
 
 int
@@ -552,3 +622,137 @@ xfs_defer_move(
 
 	xfs_defer_reset(stp);
 }
+
+/*
+ * Prepare a chain of fresh deferred ops work items to be completed later.  Log
+ * recovery requires the ability to put off until later the actual finishing
+ * work so that it can process unfinished items recovered from the log in
+ * correct order.
+ *
+ * Create and log intent items for all the work that we're capturing so that we
+ * can be assured that the items will get replayed if the system goes down
+ * before log recovery gets a chance to finish the work it put off.  The entire
+ * deferred ops state is transferred to the capture structure and the
+ * transaction is then ready for the caller to commit it.  If there are no
+ * intent items to capture, this function returns NULL.
+ *
+ * If capture_ip is not NULL, the capture structure will obtain an extra
+ * reference to the inode.
+ */
+static struct xfs_defer_capture *
+xfs_defer_ops_capture(
+	struct xfs_trans		*tp,
+	struct xfs_inode		*capture_ip)
+{
+	struct xfs_defer_capture	*dfc;
+
+	if (list_empty(&tp->t_dfops))
+		return NULL;
+
+	/* Create an object to capture the defer ops. */
+	dfc = kmem_zalloc(sizeof(*dfc), KM_NOFS);
+	INIT_LIST_HEAD(&dfc->dfc_list);
+	INIT_LIST_HEAD(&dfc->dfc_dfops);
+
+	xfs_defer_create_intents(tp);
+
+	/* Move the dfops chain and transaction state to the capture struct. */
+	list_splice_init(&tp->t_dfops, &dfc->dfc_dfops);
+	dfc->dfc_tpflags = tp->t_flags & XFS_TRANS_LOWMODE;
+	tp->t_flags &= ~XFS_TRANS_LOWMODE;
+
+	/* Capture the remaining block reservations along with the dfops. */
+	dfc->dfc_blkres = tp->t_blk_res - tp->t_blk_res_used;
+	dfc->dfc_rtxres = tp->t_rtx_res - tp->t_rtx_res_used;
+
+	/* Preserve the log reservation size. */
+	dfc->dfc_logres = tp->t_log_res;
+
+	/*
+	 * Grab an extra reference to this inode and attach it to the capture
+	 * structure.
+	 */
+	if (capture_ip) {
+		ihold(VFS_I(capture_ip));
+		dfc->dfc_capture_ip = capture_ip;
+	}
+
+	return dfc;
+}
+
+/* Release all resources that we used to capture deferred ops. */
+void
+xfs_defer_ops_release(
+	struct xfs_mount		*mp,
+	struct xfs_defer_capture	*dfc)
+{
+	xfs_defer_cancel_list(mp, &dfc->dfc_dfops);
+	if (dfc->dfc_capture_ip)
+		xfs_irele(dfc->dfc_capture_ip);
+	kmem_free(dfc);
+}
+
+/*
+ * Capture any deferred ops and commit the transaction.  This is the last step
+ * needed to finish a log intent item that we recovered from the log.  If any
+ * of the deferred ops operate on an inode, the caller must pass in that inode
+ * so that the reference can be transferred to the capture structure.  The
+ * caller must hold ILOCK_EXCL on the inode, and must unlock it before calling
+ * xfs_defer_ops_continue.
+ */
+int
+xfs_defer_ops_capture_and_commit(
+	struct xfs_trans		*tp,
+	struct xfs_inode		*capture_ip,
+	struct list_head		*capture_list)
+{
+	struct xfs_mount		*mp = tp->t_mountp;
+	struct xfs_defer_capture	*dfc;
+	int				error;
+
+	ASSERT(!capture_ip || xfs_isilocked(capture_ip, XFS_ILOCK_EXCL));
+
+	/* If we don't capture anything, commit transaction and exit. */
+	dfc = xfs_defer_ops_capture(tp, capture_ip);
+	if (!dfc)
+		return xfs_trans_commit(tp);
+
+	/* Commit the transaction and add the capture structure to the list. */
+	error = xfs_trans_commit(tp);
+	if (error) {
+		xfs_defer_ops_release(mp, dfc);
+		return error;
+	}
+
+	list_add_tail(&dfc->dfc_list, capture_list);
+	return 0;
+}
+
+/*
+ * Attach a chain of captured deferred ops to a new transaction and free the
+ * capture structure.  If an inode was captured, it will be passed back to the
+ * caller with ILOCK_EXCL held and joined to the transaction with lockflags==0.
+ * The caller now owns the inode reference.
+ */
+void
+xfs_defer_ops_continue(
+	struct xfs_defer_capture	*dfc,
+	struct xfs_trans		*tp,
+	struct xfs_inode		**captured_ipp)
+{
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+	ASSERT(!(tp->t_flags & XFS_TRANS_DIRTY));
+
+	/* Lock and join the captured inode to the new transaction. */
+	if (dfc->dfc_capture_ip) {
+		xfs_ilock(dfc->dfc_capture_ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, dfc->dfc_capture_ip, 0);
+	}
+	*captured_ipp = dfc->dfc_capture_ip;
+
+	/* Move captured dfops chain and state to the transaction. */
+	list_splice_init(&dfc->dfc_dfops, &tp->t_dfops);
+	tp->t_flags |= dfc->dfc_tpflags;
+
+	kmem_free(dfc);
+}