19 files changed, 318 insertions, 234 deletions

diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 15412fe..b552f81 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -852,8 +852,8 @@ xfs_convert_page(
 		SetPageUptodate(page);
 
 	if (count) {
-		wbc->nr_to_write--;
-		if (wbc->nr_to_write <= 0)
+		if (--wbc->nr_to_write <= 0 &&
+		    wbc->sync_mode == WB_SYNC_NONE)
 			done = 1;
 	}
 	xfs_start_page_writeback(page, !page_dirty, count);
@@ -1068,7 +1068,7 @@ xfs_vm_writepage(
 	 * by themselves.
 	 */
 	if ((current->flags & (PF_MEMALLOC|PF_KSWAPD)) == PF_MEMALLOC)
-		goto out_fail;
+		goto redirty;
 
 	/*
 	 * We need a transaction if there are delalloc or unwritten buffers
@@ -1080,7 +1080,7 @@ xfs_vm_writepage(
 	 */
 	xfs_count_page_state(page, &delalloc, &unwritten);
 	if ((current->flags & PF_FSTRANS) && (delalloc || unwritten))
-		goto out_fail;
+		goto redirty;
 
 	/* Is this page beyond the end of the file? */
 	offset = i_size_read(inode);
@@ -1245,12 +1245,15 @@ error:
 	if (iohead)
 		xfs_cancel_ioend(iohead);
 
+	if (err == -EAGAIN)
+		goto redirty;
+
 	xfs_aops_discard_page(page);
 	ClearPageUptodate(page);
 	unlock_page(page);
 	return err;
 
-out_fail:
+redirty:
 	redirty_page_for_writepage(wbc, page);
 	unlock_page(page);
 	return 0;
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index ea79072..286e36e 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -440,12 +440,7 @@ _xfs_buf_find(
 		ASSERT(btp == bp->b_target);
 		if (bp->b_file_offset == range_base &&
 		    bp->b_buffer_length == range_length) {
-			/*
-			 * If we look at something, bring it to the
-			 * front of the list for next time.
-			 */
 			atomic_inc(&bp->b_hold);
-			list_move(&bp->b_hash_list, &hash->bh_list);
 			goto found;
 		}
 	}
@@ -1443,8 +1438,7 @@ xfs_alloc_bufhash(
 {
 	unsigned int		i;
 
-	btp->bt_hashshift = external ? 3 : 8;	/* 8 or 256 buckets */
-	btp->bt_hashmask = (1 << btp->bt_hashshift) - 1;
+	btp->bt_hashshift = external ? 3 : 12;	/* 8 or 4096 buckets */
 	btp->bt_hash = kmem_zalloc_large((1 << btp->bt_hashshift) *
 					 sizeof(xfs_bufhash_t));
 	for (i = 0; i < (1 << btp->bt_hashshift); i++) {
@@ -1938,7 +1932,8 @@ xfs_buf_init(void)
 	if (!xfs_buf_zone)
 		goto out;
 
-	xfslogd_workqueue = create_workqueue("xfslogd");
+	xfslogd_workqueue = alloc_workqueue("xfslogd",
+					WQ_RESCUER | WQ_HIGHPRI, 1);
 	if (!xfslogd_workqueue)
 		goto out_free_buf_zone;
 
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index d072e5f..2a05614 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -137,7 +137,6 @@ typedef struct xfs_buftarg {
 	size_t			bt_smask;
 
 	/* per device buffer hash table */
-	uint			bt_hashmask;
 	uint			bt_hashshift;
 	xfs_bufhash_t		*bt_hash;
 
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index 237f5ff..3b9e626 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -785,6 +785,8 @@ xfs_ioc_fsgetxattr(
 {
 	struct fsxattr		fa;
 
+	memset(&fa, 0, sizeof(struct fsxattr));
+
 	xfs_ilock(ip, XFS_ILOCK_SHARED);
 	fa.fsx_xflags = xfs_ip2xflags(ip);
 	fa.fsx_extsize = ip->i_d.di_extsize << ip->i_mount->m_sb.sb_blocklog;
@@ -907,6 +909,13 @@ xfs_ioctl_setattr(
 		return XFS_ERROR(EIO);
 
 	/*
+	 * Disallow 32bit project ids because on-disk structure
+	 * is 16bit only.
+	 */
+	if ((mask & FSX_PROJID) && (fa->fsx_projid > (__uint16_t)-1))
+		return XFS_ERROR(EINVAL);
+
+	/*
 	 * If disk quotas is on, we make sure that the dquots do exist on disk,
 	 * before we start any other transactions. Trying to do this later
 	 * is messy. We don't care to take a readlock to look at the ids
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index 68be25d..b1fc2a6 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -664,7 +664,7 @@ xfs_vn_fiemap(
 					fieinfo->fi_extents_max + 1;
 	bm.bmv_count = min_t(__s32, bm.bmv_count,
 			     (PAGE_SIZE * 16 / sizeof(struct getbmapx)));
-	bm.bmv_iflags = BMV_IF_PREALLOC;
+	bm.bmv_iflags = BMV_IF_PREALLOC | BMV_IF_NO_HOLES;
 	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR)
 		bm.bmv_iflags |= BMV_IF_ATTRFORK;
 	if (!(fieinfo->fi_flags & FIEMAP_FLAG_SYNC))
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 15c35b6..a4e0797 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1226,6 +1226,7 @@ xfs_fs_statfs(
 	struct xfs_inode	*ip = XFS_I(dentry->d_inode);
 	__uint64_t		fakeinos, id;
 	xfs_extlen_t		lsize;
+	__int64_t		ffree;
 
 	statp->f_type = XFS_SB_MAGIC;
 	statp->f_namelen = MAXNAMELEN - 1;
@@ -1249,7 +1250,11 @@ xfs_fs_statfs(
 		statp->f_files = min_t(typeof(statp->f_files),
 					statp->f_files,
 					mp->m_maxicount);
-	statp->f_ffree = statp->f_files - (sbp->sb_icount - sbp->sb_ifree);
+
+	/* make sure statp->f_ffree does not underflow */
+	ffree = statp->f_files - (sbp->sb_icount - sbp->sb_ifree);
+	statp->f_ffree = max_t(__int64_t, ffree, 0);
+
 	spin_unlock(&mp->m_sb_lock);
 
 	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) ||
@@ -1402,7 +1407,7 @@ xfs_fs_freeze(
 
 	xfs_save_resvblks(mp);
 	xfs_quiesce_attr(mp);
-	return -xfs_fs_log_dummy(mp);
+	return -xfs_fs_log_dummy(mp, SYNC_WAIT);
 }
 
 STATIC int
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
index dfcbd98..d59c4a6 100644
--- a/fs/xfs/linux-2.6/xfs_sync.c
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -34,6 +34,7 @@
 #include "xfs_inode_item.h"
 #include "xfs_quota.h"
 #include "xfs_trace.h"
+#include "xfs_fsops.h"
 
 #include <linux/kthread.h>
 #include <linux/freezer.h>
@@ -341,38 +342,6 @@ xfs_sync_attr(
 }
 
 STATIC int
-xfs_commit_dummy_trans(
-	struct xfs_mount	*mp,
-	uint			flags)
-{
-	struct xfs_inode	*ip = mp->m_rootip;
-	struct xfs_trans	*tp;
-	int			error;
-
-	/*
-	 * Put a dummy transaction in the log to tell recovery
-	 * that all others are OK.
-	 */
-	tp = xfs_trans_alloc(mp, XFS_TRANS_DUMMY1);
-	error = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
-	if (error) {
-		xfs_trans_cancel(tp, 0);
-		return error;
-	}
-
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-
-	xfs_trans_ijoin(tp, ip);
-	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	error = xfs_trans_commit(tp, 0);
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-
-	/* the log force ensures this transaction is pushed to disk */
-	xfs_log_force(mp, (flags & SYNC_WAIT) ? XFS_LOG_SYNC : 0);
-	return error;
-}
-
-STATIC int
 xfs_sync_fsdata(
 	struct xfs_mount	*mp)
 {
@@ -432,7 +401,7 @@ xfs_quiesce_data(
 
 	/* mark the log as covered if needed */
 	if (xfs_log_need_covered(mp))
-		error2 = xfs_commit_dummy_trans(mp, SYNC_WAIT);
+		error2 = xfs_fs_log_dummy(mp, SYNC_WAIT);
 
 	/* flush data-only devices */
 	if (mp->m_rtdev_targp)
@@ -563,7 +532,7 @@ xfs_flush_inodes(
 /*
  * Every sync period we need to unpin all items, reclaim inodes and sync
  * disk quotas.  We might need to cover the log to indicate that the
- * filesystem is idle.
+ * filesystem is idle and not frozen.
  */
 STATIC void
 xfs_sync_worker(
@@ -577,8 +546,9 @@ xfs_sync_worker(
 		xfs_reclaim_inodes(mp, 0);
 		/* dgc: errors ignored here */
 		error = xfs_qm_sync(mp, SYNC_TRYLOCK);
-		if (xfs_log_need_covered(mp))
-			error = xfs_commit_dummy_trans(mp, 0);
+		if (mp->m_super->s_frozen == SB_UNFROZEN &&
+		    xfs_log_need_covered(mp))
+			error = xfs_fs_log_dummy(mp, 0);
 	}
 	mp->m_sync_seq++;
 	wake_up(&mp->m_wait_single_sync_task);
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 23f14e5..f90dadd 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5533,12 +5533,24 @@ xfs_getbmap(
 					map[i].br_startblock))
 				goto out_free_map;
 
-			nexleft--;
 			bmv->bmv_offset =
 				out[cur_ext].bmv_offset +
 				out[cur_ext].bmv_length;
 			bmv->bmv_length =
 				max_t(__int64_t, 0, bmvend - bmv->bmv_offset);
+
+			/*
+			 * In case we don't want to return the hole,
+			 * don't increase cur_ext so that we can reuse
+			 * it in the next loop.
+			 */
+			if ((iflags & BMV_IF_NO_HOLES) &&
+			    map[i].br_startblock == HOLESTARTBLOCK) {
+				memset(&out[cur_ext], 0, sizeof(out[cur_ext]));
+				continue;
+			}
+
+			nexleft--;
 			bmv->bmv_entries++;
 			cur_ext++;
 		}
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
index 7cf7220..87c2e9d 100644
--- a/fs/xfs/xfs_fs.h
+++ b/fs/xfs/xfs_fs.h
@@ -114,8 +114,10 @@ struct getbmapx {
 #define BMV_IF_NO_DMAPI_READ	0x2	/* Do not generate DMAPI read event  */
 #define BMV_IF_PREALLOC		0x4	/* rtn status BMV_OF_PREALLOC if req */
 #define BMV_IF_DELALLOC		0x8	/* rtn status BMV_OF_DELALLOC if req */
+#define BMV_IF_NO_HOLES		0x10	/* Do not return holes */
 #define BMV_IF_VALID	\
-	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|BMV_IF_DELALLOC)
+	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|	\
+	 BMV_IF_DELALLOC|BMV_IF_NO_HOLES)
 
 /*	bmv_oflags values - returned for each non-header segment */
 #define BMV_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
index dbca5f5..43b1d56 100644
--- a/fs/xfs/xfs_fsops.c
+++ b/fs/xfs/xfs_fsops.c
@@ -604,31 +604,36 @@ out:
 	return 0;
 }
 
+/*
+ * Dump a transaction into the log that contains no real change. This is needed
+ * to be able to make the log dirty or stamp the current tail LSN into the log
+ * during the covering operation.
+ *
+ * We cannot use an inode here for this - that will push dirty state back up
+ * into the VFS and then periodic inode flushing will prevent log covering from
+ * making progress. Hence we log a field in the superblock instead.
+ */
 int
 xfs_fs_log_dummy(
-	xfs_mount_t	*mp)
+	xfs_mount_t	*mp,
+	int		flags)
 {
 	xfs_trans_t	*tp;
-	xfs_inode_t	*ip;
 	int		error;
 
 	tp = _xfs_trans_alloc(mp, XFS_TRANS_DUMMY1, KM_SLEEP);
-	error = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
+	error = xfs_trans_reserve(tp, 0, mp->m_sb.sb_sectsize + 128, 0, 0,
+					XFS_DEFAULT_LOG_COUNT);
 	if (error) {
 		xfs_trans_cancel(tp, 0);
 		return error;
 	}
 
-	ip = mp->m_rootip;
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-
-	xfs_trans_ijoin(tp, ip);
-	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	xfs_trans_set_sync(tp);
-	error = xfs_trans_commit(tp, 0);
-
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-	return error;
+	/* log the UUID because it is an unchanging field */
+	xfs_mod_sb(tp, XFS_SB_UUID);
+	if (flags & SYNC_WAIT)
+		xfs_trans_set_sync(tp);
+	return xfs_trans_commit(tp, 0);
 }
 
 int
diff --git a/fs/xfs/xfs_fsops.h b/fs/xfs/xfs_fsops.h
index 88435e0..a786c52 100644
--- a/fs/xfs/xfs_fsops.h
+++ b/fs/xfs/xfs_fsops.h
@@ -25,6 +25,6 @@ extern int xfs_fs_counts(xfs_mount_t *mp, xfs_fsop_counts_t *cnt);
 extern int xfs_reserve_blocks(xfs_mount_t *mp, __uint64_t *inval,
 				xfs_fsop_resblks_t *outval);
 extern int xfs_fs_goingdown(xfs_mount_t *mp, __uint32_t inflags);
-extern int xfs_fs_log_dummy(xfs_mount_t *mp);
+extern int xfs_fs_log_dummy(xfs_mount_t *mp, int flags);
 
 #endif	/* __XFS_FSOPS_H__ */
diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
index abf80ae..5371d2d 100644
--- a/fs/xfs/xfs_ialloc.c
+++ b/fs/xfs/xfs_ialloc.c
@@ -1213,7 +1213,6 @@ xfs_imap_lookup(
 	struct xfs_inobt_rec_incore rec;
 	struct xfs_btree_cur	*cur;
 	struct xfs_buf		*agbp;
-	xfs_agino_t		startino;
 	int			error;
 	int			i;
 
@@ -1227,13 +1226,13 @@ xfs_imap_lookup(
 	}
 
 	/*
-	 * derive and lookup the exact inode record for the given agino. If the
-	 * record cannot be found, then it's an invalid inode number and we
-	 * should abort.
+	 * Lookup the inode record for the given agino. If the record cannot be
+	 * found, then it's an invalid inode number and we should abort. Once
+	 * 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);
-	startino = agino & ~(XFS_IALLOC_INODES(mp) - 1);
-	error = xfs_inobt_lookup(cur, startino, XFS_LOOKUP_EQ, &i);
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i);
 	if (!error) {
 		if (i)
 			error = xfs_inobt_get_rec(cur, &rec, &i);
@@ -1246,6 +1245,11 @@ xfs_imap_lookup(
 	if (error)
 		return error;
 
+	/* check that the returned record contains the required inode */
+	if (rec.ir_startino > agino ||
+	    rec.ir_startino + XFS_IALLOC_INODES(mp) <= agino)
+		return EINVAL;
+
 	/* for untrusted inodes check it is allocated first */
 	if ((flags & XFS_IGET_UNTRUSTED) &&
 	    (rec.ir_free & XFS_INOBT_MASK(agino - rec.ir_startino)))
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 68415cb..34798f3 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -1914,6 +1914,11 @@ xfs_iunlink_remove(
 	return 0;
 }
 
+/*
+ * A big issue when freeing the inode cluster is is that we _cannot_ skip any
+ * inodes that are in memory - they all must be marked stale and attached to
+ * the cluster buffer.
+ */
 STATIC void
 xfs_ifree_cluster(
 	xfs_inode_t	*free_ip,
@@ -1945,8 +1950,6 @@ xfs_ifree_cluster(
 	}
 
 	for (j = 0; j < nbufs; j++, inum += ninodes) {
-		int	found = 0;
-
 		blkno = XFS_AGB_TO_DADDR(mp, XFS_INO_TO_AGNO(mp, inum),
 					 XFS_INO_TO_AGBNO(mp, inum));
 
@@ -1965,7 +1968,9 @@ xfs_ifree_cluster(
 		/*
 		 * Walk the inodes already attached to the buffer and mark them
 		 * stale. These will all have the flush locks held, so an
-		 * in-memory inode walk can't lock them.
+		 * in-memory inode walk can't lock them. By marking them all
+		 * stale first, we will not attempt to lock them in the loop
+		 * below as the XFS_ISTALE flag will be set.
 		 */
 		lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
 		while (lip) {
@@ -1977,11 +1982,11 @@ xfs_ifree_cluster(
 							&iip->ili_flush_lsn,
 							&iip->ili_item.li_lsn);
 				xfs_iflags_set(iip->ili_inode, XFS_ISTALE);
-				found++;
 			}
 			lip = lip->li_bio_list;
 		}
 
+
 		/*
 		 * For each inode in memory attempt to add it to the inode
 		 * buffer and set it up for being staled on buffer IO
@@ -1993,6 +1998,7 @@ xfs_ifree_cluster(
 		 * even trying to lock them.
 		 */
 		for (i = 0; i < ninodes; i++) {
+retry:
 			read_lock(&pag->pag_ici_lock);
 			ip = radix_tree_lookup(&pag->pag_ici_root,
 					XFS_INO_TO_AGINO(mp, (inum + i)));
@@ -2003,38 +2009,36 @@ xfs_ifree_cluster(
 				continue;
 			}
 
-			/* don't try to lock/unlock the current inode */
+			/*
+			 * Don't try to lock/unlock the current inode, but we
+			 * _cannot_ skip the other inodes that we did not find
+			 * in the list attached to the buffer and are not
+			 * already marked stale. If we can't lock it, back off
+			 * and retry.
+			 */
 			if (ip != free_ip &&
 			    !xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
 				read_unlock(&pag->pag_ici_lock);
-				continue;
+				delay(1);
+				goto retry;
 			}
 			read_unlock(&pag->pag_ici_lock);
 
-			if (!xfs_iflock_nowait(ip)) {
-				if (ip != free_ip)
-					xfs_iunlock(ip, XFS_ILOCK_EXCL);
-				continue;
-			}
-
+			xfs_iflock(ip);
 			xfs_iflags_set(ip, XFS_ISTALE);
-			if (xfs_inode_clean(ip)) {
-				ASSERT(ip != free_ip);
-				xfs_ifunlock(ip);
-				xfs_iunlock(ip, XFS_ILOCK_EXCL);
-				continue;
-			}
 
+			/*
+			 * we don't need to attach clean inodes or those only
+			 * with unlogged changes (which we throw away, anyway).
+			 */
 			iip = ip->i_itemp;
-			if (!iip) {
-				/* inode with unlogged changes only */
+			if (!iip || xfs_inode_clean(ip)) {
 				ASSERT(ip != free_ip);
 				ip->i_update_core = 0;
 				xfs_ifunlock(ip);
 				xfs_iunlock(ip, XFS_ILOCK_EXCL);
 				continue;
 			}
-			found++;
 
 			iip->ili_last_fields = iip->ili_format.ilf_fields;
 			iip->ili_format.ilf_fields = 0;
@@ -2049,8 +2053,7 @@ xfs_ifree_cluster(
 				xfs_iunlock(ip, XFS_ILOCK_EXCL);
 		}
 
-		if (found)
-			xfs_trans_stale_inode_buf(tp, bp);
+		xfs_trans_stale_inode_buf(tp, bp);
 		xfs_trans_binval(tp, bp);
 	}
 
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 925d572..33f718f 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -3015,7 +3015,8 @@ _xfs_log_force(
 
 	XFS_STATS_INC(xs_log_force);
 
-	xlog_cil_push(log, 1);
+	if (log->l_cilp)
+		xlog_cil_force(log);
 
 	spin_lock(&log->l_icloglock);
 
@@ -3167,7 +3168,7 @@ _xfs_log_force_lsn(
 	XFS_STATS_INC(xs_log_force);
 
 	if (log->l_cilp) {
-		lsn = xlog_cil_push_lsn(log, lsn);
+		lsn = xlog_cil_force_lsn(log, lsn);
 		if (lsn == NULLCOMMITLSN)
 			return 0;
 	}
@@ -3724,7 +3725,7 @@ xfs_log_force_umount(
 	 * call below.
 	 */
 	if (!logerror && (mp->m_flags & XFS_MOUNT_DELAYLOG))
-		xlog_cil_push(log, 1);
+		xlog_cil_force(log);
 
 	/*
 	 * We must hold both the GRANT lock and the LOG lock,
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
index 31e4ea2..7e206fc 100644
--- a/fs/xfs/xfs_log_cil.c
+++ b/fs/xfs/xfs_log_cil.c
@@ -68,6 +68,7 @@ xlog_cil_init(
 	ctx->sequence = 1;
 	ctx->cil = cil;
 	cil->xc_ctx = ctx;
+	cil->xc_current_sequence = ctx->sequence;
 
 	cil->xc_log = log;
 	log->l_cilp = cil;
@@ -269,15 +270,10 @@ xlog_cil_insert(
 static void
 xlog_cil_format_items(
 	struct log		*log,
-	struct xfs_log_vec	*log_vector,
-	struct xlog_ticket	*ticket,
-	xfs_lsn_t		*start_lsn)
+	struct xfs_log_vec	*log_vector)
 {
 	struct xfs_log_vec *lv;
 
-	if (start_lsn)
-		*start_lsn = log->l_cilp->xc_ctx->sequence;
-
 	ASSERT(log_vector);
 	for (lv = log_vector; lv; lv = lv->lv_next) {
 		void	*ptr;
@@ -301,9 +297,24 @@ xlog_cil_format_items(
 			ptr += vec->i_len;
 		}
 		ASSERT(ptr == lv->lv_buf + lv->lv_buf_len);
+	}
+}
 
+static void
+xlog_cil_insert_items(
+	struct log		*log,
+	struct xfs_log_vec	*log_vector,
+	struct xlog_ticket	*ticket,
+	xfs_lsn_t		*start_lsn)
+{
+	struct xfs_log_vec *lv;
+
+	if (start_lsn)
+		*start_lsn = log->l_cilp->xc_ctx->sequence;
+
+	ASSERT(log_vector);
+	for (lv = log_vector; lv; lv = lv->lv_next)
 		xlog_cil_insert(log, ticket, lv->lv_item, lv);
-	}
 }
 
 static void
@@ -321,80 +332,6 @@ xlog_cil_free_logvec(
 }
 
 /*
- * Commit a transaction with the given vector to the Committed Item List.
- *
- * To do this, we need to format the item, pin it in memory if required and
- * account for the space used by the transaction. Once we have done that we
- * need to release the unused reservation for the transaction, attach the
- * transaction to the checkpoint context so we carry the busy extents through
- * to checkpoint completion, and then unlock all the items in the transaction.
- *
- * For more specific information about the order of operations in
- * xfs_log_commit_cil() please refer to the comments in
- * xfs_trans_commit_iclog().
- *
- * Called with the context lock already held in read mode to lock out
- * background commit, returns without it held once background commits are
- * allowed again.
- */
-int
-xfs_log_commit_cil(
-	struct xfs_mount	*mp,
-	struct xfs_trans	*tp,
-	struct xfs_log_vec	*log_vector,
-	xfs_lsn_t		*commit_lsn,
-	int			flags)
-{
-	struct log		*log = mp->m_log;
-	int			log_flags = 0;
-	int			push = 0;
-
-	if (flags & XFS_TRANS_RELEASE_LOG_RES)
-		log_flags = XFS_LOG_REL_PERM_RESERV;
-
-	if (XLOG_FORCED_SHUTDOWN(log)) {
-		xlog_cil_free_logvec(log_vector);
-		return XFS_ERROR(EIO);
-	}
-
-	/* lock out background commit */
-	down_read(&log->l_cilp->xc_ctx_lock);
-	xlog_cil_format_items(log, log_vector, tp->t_ticket, commit_lsn);
-
-	/* check we didn't blow the reservation */
-	if (tp->t_ticket->t_curr_res < 0)
-		xlog_print_tic_res(log->l_mp, tp->t_ticket);
-
-	/* attach the transaction to the CIL if it has any busy extents */
-	if (!list_empty(&tp->t_busy)) {
-		spin_lock(&log->l_cilp->xc_cil_lock);
-		list_splice_init(&tp->t_busy,
-					&log->l_cilp->xc_ctx->busy_extents);
-		spin_unlock(&log->l_cilp->xc_cil_lock);
-	}
-
-	tp->t_commit_lsn = *commit_lsn;
-	xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
-	xfs_trans_unreserve_and_mod_sb(tp);
-
-	/* check for background commit before unlock */
-	if (log->l_cilp->xc_ctx->space_used > XLOG_CIL_SPACE_LIMIT(log))
-		push = 1;
-	up_read(&log->l_cilp->xc_ctx_lock);
-
-	/*
-	 * We need to push CIL every so often so we don't cache more than we
-	 * can fit in the log. The limit really is that a checkpoint can't be
-	 * more than half the log (the current checkpoint is not allowed to
-	 * overwrite the previous checkpoint), but commit latency and memory
-	 * usage limit this to a smaller size in most cases.
-	 */
-	if (push)
-		xlog_cil_push(log, 0);
-	return 0;
-}
-
-/*
  * Mark all items committed and clear busy extents. We free the log vector
  * chains in a separate pass so that we unpin the log items as quickly as
  * possible.
@@ -427,13 +364,23 @@ xlog_cil_committed(
 }
 
 /*
- * Push the Committed Item List to the log. If the push_now flag is not set,
- * then it is a background flush and so we can chose to ignore it.
+ * Push the Committed Item List to the log. If @push_seq flag is zero, then it
+ * is a background flush and so we can chose to ignore it. Otherwise, if the
+ * current sequence is the same as @push_seq we need to do a flush. If
+ * @push_seq is less than the current sequence, then it has already been
+ * flushed and we don't need to do anything - the caller will wait for it to
+ * complete if necessary.
+ *
+ * @push_seq is a value rather than a flag because that allows us to do an
+ * unlocked check of the sequence number for a match. Hence we can allows log
+ * forces to run racily and not issue pushes for the same sequence twice. If we
+ * get a race between multiple pushes for the same sequence they will block on
+ * the first one and then abort, hence avoiding needless pushes.
  */
-int
+STATIC int
 xlog_cil_push(
 	struct log		*log,
-	int			push_now)
+	xfs_lsn_t		push_seq)
 {
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_log_vec	*lv;
@@ -453,12 +400,20 @@ xlog_cil_push(
 	if (!cil)
 		return 0;
 
+	ASSERT(!push_seq || push_seq <= cil->xc_ctx->sequence);
+
 	new_ctx = kmem_zalloc(sizeof(*new_ctx), KM_SLEEP|KM_NOFS);
 	new_ctx->ticket = xlog_cil_ticket_alloc(log);
 
-	/* lock out transaction commit, but don't block on background push */
+	/*
+	 * Lock out transaction commit, but don't block for background pushes
+	 * unless we are well over the CIL space limit. See the definition of
+	 * XLOG_CIL_HARD_SPACE_LIMIT() for the full explanation of the logic
+	 * used here.
+	 */
 	if (!down_write_trylock(&cil->xc_ctx_lock)) {
-		if (!push_now)
+		if (!push_seq &&
+		    cil->xc_ctx->space_used < XLOG_CIL_HARD_SPACE_LIMIT(log))
 			goto out_free_ticket;
 		down_write(&cil->xc_ctx_lock);
 	}
@@ -469,7 +424,11 @@ xlog_cil_push(
 		goto out_skip;
 
 	/* check for spurious background flush */
-	if (!push_now && cil->xc_ctx->space_used < XLOG_CIL_SPACE_LIMIT(log))
+	if (!push_seq && cil->xc_ctx->space_used < XLOG_CIL_SPACE_LIMIT(log))
+		goto out_skip;
+
+	/* check for a previously pushed seqeunce */
+	if (push_seq && push_seq < cil->xc_ctx->sequence)
 		goto out_skip;
 
 	/*
@@ -515,6 +474,13 @@ xlog_cil_push(
 	cil->xc_ctx = new_ctx;
 
 	/*
+	 * mirror the new sequence into the cil structure so that we can do
+	 * unlocked checks against the current sequence in log forces without
+	 * risking deferencing a freed context pointer.
+	 */
+	cil->xc_current_sequence = new_ctx->sequence;
+
+	/*
 	 * The switch is now done, so we can drop the context lock and move out
 	 * of a shared context. We can't just go straight to the commit record,
 	 * though - we need to synchronise with previous and future commits so
@@ -626,6 +592,102 @@ out_abort:
 }
 
 /*
+ * Commit a transaction with the given vector to the Committed Item List.
+ *
+ * To do this, we need to format the item, pin it in memory if required and
+ * account for the space used by the transaction. Once we have done that we
+ * need to release the unused reservation for the transaction, attach the
+ * transaction to the checkpoint context so we carry the busy extents through
+ * to checkpoint completion, and then unlock all the items in the transaction.
+ *
+ * For more specific information about the order of operations in
+ * xfs_log_commit_cil() please refer to the comments in
+ * xfs_trans_commit_iclog().
+ *
+ * Called with the context lock already held in read mode to lock out
+ * background commit, returns without it held once background commits are
+ * allowed again.
+ */
+int
+xfs_log_commit_cil(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_log_vec	*log_vector,
+	xfs_lsn_t		*commit_lsn,
+	int			flags)
+{
+	struct log		*log = mp->m_log;
+	int			log_flags = 0;
+	int			push = 0;
+
+	if (flags & XFS_TRANS_RELEASE_LOG_RES)
+		log_flags = XFS_LOG_REL_PERM_RESERV;
+
+	if (XLOG_FORCED_SHUTDOWN(log)) {
+		xlog_cil_free_logvec(log_vector);
+		return XFS_ERROR(EIO);
+	}
+
+	/*
+	 * do all the hard work of formatting items (including memory
+	 * allocation) outside the CIL context lock. This prevents stalling CIL
+	 * pushes when we are low on memory and a transaction commit spends a
+	 * lot of time in memory reclaim.
+	 */
+	xlog_cil_format_items(log, log_vector);
+
+	/* lock out background commit */
+	down_read(&log->l_cilp->xc_ctx_lock);
+	xlog_cil_insert_items(log, log_vector, tp->t_ticket, commit_lsn);
+
+	/* check we didn't blow the reservation */
+	if (tp->t_ticket->t_curr_res < 0)
+		xlog_print_tic_res(log->l_mp, tp->t_ticket);
+
+	/* attach the transaction to the CIL if it has any busy extents */
+	if (!list_empty(&tp->t_busy)) {
+		spin_lock(&log->l_cilp->xc_cil_lock);
+		list_splice_init(&tp->t_busy,
+					&log->l_cilp->xc_ctx->busy_extents);
+		spin_unlock(&log->l_cilp->xc_cil_lock);
+	}
+
+	tp->t_commit_lsn = *commit_lsn;
+	xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
+	xfs_trans_unreserve_and_mod_sb(tp);
+
+	/*
+	 * Once all the items of the transaction have been copied to the CIL,
+	 * the items can be unlocked and freed.
+	 *
+	 * This needs to be done before we drop the CIL context lock because we
+	 * have to update state in the log items and unlock them before they go
+	 * to disk. If we don't, then the CIL checkpoint can race with us and
+	 * we can run checkpoint completion before we've updated and unlocked
+	 * the log items. This affects (at least) processing of stale buffers,
+	 * inodes and EFIs.
+	 */
+	xfs_trans_free_items(tp, *commit_lsn, 0);
+
+	/* check for background commit before unlock */
+	if (log->l_cilp->xc_ctx->space_used > XLOG_CIL_SPACE_LIMIT(log))
+		push = 1;
+
+	up_read(&log->l_cilp->xc_ctx_lock);
+
+	/*
+	 * We need to push CIL every so often so we don't cache more than we
+	 * can fit in the log. The limit really is that a checkpoint can't be
+	 * more than half the log (the current checkpoint is not allowed to
+	 * overwrite the previous checkpoint), but commit latency and memory
+	 * usage limit this to a smaller size in most cases.
+	 */
+	if (push)
+		xlog_cil_push(log, 0);
+	return 0;
+}
+
+/*
  * Conditionally push the CIL based on the sequence passed in.
  *
  * We only need to push if we haven't already pushed the sequence
@@ -639,39 +701,34 @@ out_abort:
  * commit lsn is there. It'll be empty, so this is broken for now.
  */
 xfs_lsn_t
-xlog_cil_push_lsn(
+xlog_cil_force_lsn(
 	struct log	*log,
-	xfs_lsn_t	push_seq)
+	xfs_lsn_t	sequence)
 {
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_cil_ctx	*ctx;
 	xfs_lsn_t		commit_lsn = NULLCOMMITLSN;
 
-restart:
-	down_write(&cil->xc_ctx_lock);
-	ASSERT(push_seq <= cil->xc_ctx->sequence);
-
-	/* check to see if we need to force out the current context */
-	if (push_seq == cil->xc_ctx->sequence) {
-		up_write(&cil->xc_ctx_lock);
-		xlog_cil_push(log, 1);
-		goto restart;
-	}
+	ASSERT(sequence <= cil->xc_current_sequence);
+
+	/*
+	 * check to see if we need to force out the current context.
+	 * xlog_cil_push() handles racing pushes for the same sequence,
+	 * so no need to deal with it here.
+	 */
+	if (sequence == cil->xc_current_sequence)
+		xlog_cil_push(log, sequence);
 
 	/*
 	 * See if we can find a previous sequence still committing.
-	 * We can drop the flush lock as soon as we have the cil lock
-	 * because we are now only comparing contexts protected by
-	 * the cil lock.
-	 *
 	 * We need to wait for all previous sequence commits to complete
 	 * before allowing the force of push_seq to go ahead. Hence block
 	 * on commits for those as well.
 	 */
+restart:
 	spin_lock(&cil->xc_cil_lock);
-	up_write(&cil->xc_ctx_lock);
 	list_for_each_entry(ctx, &cil->xc_committing, committing) {
-		if (ctx->sequence > push_seq)
+		if (ctx->sequence > sequence)
 			continue;
 		if (!ctx->commit_lsn) {
 			/*
@@ -681,7 +738,7 @@ restart:
 			sv_wait(&cil->xc_commit_wait, 0, &cil->xc_cil_lock, 0);
 			goto restart;
 		}
-		if (ctx->sequence != push_seq)
+		if (ctx->sequence != sequence)
 			continue;
 		/* found it! */
 		commit_lsn = ctx->commit_lsn;
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index 8c07261..edcdfe0 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -422,16 +422,17 @@ struct xfs_cil {
 	struct rw_semaphore	xc_ctx_lock;
 	struct list_head	xc_committing;
 	sv_t			xc_commit_wait;
+	xfs_lsn_t		xc_current_sequence;
 };
 
 /*
- * The amount of log space we should the CIL to aggregate is difficult to size.
- * Whatever we chose we have to make we can get a reservation for the log space
- * effectively, that it is large enough to capture sufficient relogging to
- * reduce log buffer IO significantly, but it is not too large for the log or
- * induces too much latency when writing out through the iclogs. We track both
- * space consumed and the number of vectors in the checkpoint context, so we
- * need to decide which to use for limiting.
+ * The amount of log space we allow the CIL to aggregate is difficult to size.
+ * Whatever we choose, we have to make sure we can get a reservation for the
+ * log space effectively, that it is large enough to capture sufficient
+ * relogging to reduce log buffer IO significantly, but it is not too large for
+ * the log or induces too much latency when writing out through the iclogs. We
+ * track both space consumed and the number of vectors in the checkpoint
+ * context, so we need to decide which to use for limiting.
  *
  * Every log buffer we write out during a push needs a header reserved, which
  * is at least one sector and more for v2 logs. Hence we need a reservation of
@@ -458,16 +459,21 @@ struct xfs_cil {
  * checkpoint transaction ticket is specific to the checkpoint context, rather
  * than the CIL itself.
  *
- * With dynamic reservations, we can basically make up arbitrary limits for the
- * checkpoint size so long as they don't violate any other size rules.  Hence
- * the initial maximum size for the checkpoint transaction will be set to a
- * quarter of the log or 8MB, which ever is smaller. 8MB is an arbitrary limit
- * right now based on the latency of writing out a large amount of data through
- * the circular iclog buffers.
+ * With dynamic reservations, we can effectively make up arbitrary limits for
+ * the checkpoint size so long as they don't violate any other size rules.
+ * Recovery imposes a rule that no transaction exceed half the log, so we are
+ * limited by that.  Furthermore, the log transaction reservation subsystem
+ * tries to keep 25% of the log free, so we need to keep below that limit or we
+ * risk running out of free log space to start any new transactions.
+ *
+ * In order to keep background CIL push efficient, we will set a lower
+ * threshold at which background pushing is attempted without blocking current
+ * transaction commits.  A separate, higher bound defines when CIL pushes are
+ * enforced to ensure we stay within our maximum checkpoint size bounds.
+ * threshold, yet give us plenty of space for aggregation on large logs.
  */
-
-#define XLOG_CIL_SPACE_LIMIT(log)	\
-	(min((log->l_logsize >> 2), (8 * 1024 * 1024)))
+#define XLOG_CIL_SPACE_LIMIT(log)	(log->l_logsize >> 3)
+#define XLOG_CIL_HARD_SPACE_LIMIT(log)	(3 * (log->l_logsize >> 4))
 
 /*
  * The reservation head lsn is not made up of a cycle number and block number.
@@ -562,8 +568,16 @@ int	xlog_cil_init(struct log *log);
 void	xlog_cil_init_post_recovery(struct log *log);
 void	xlog_cil_destroy(struct log *log);
 
-int	xlog_cil_push(struct log *log, int push_now);
-xfs_lsn_t xlog_cil_push_lsn(struct log *log, xfs_lsn_t push_sequence);
+/*
+ * CIL force routines
+ */
+xfs_lsn_t xlog_cil_force_lsn(struct log *log, xfs_lsn_t sequence);
+
+static inline void
+xlog_cil_force(struct log *log)
+{
+	xlog_cil_force_lsn(log, log->l_cilp->xc_current_sequence);
+}
 
 /*
  * Unmount record type is used as a pseudo transaction type for the ticket.
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index fdca741..1c47eda 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -1167,7 +1167,7 @@ xfs_trans_del_item(
  * Unlock all of the items of a transaction and free all the descriptors
  * of that transaction.
  */
-STATIC void
+void
 xfs_trans_free_items(
 	struct xfs_trans	*tp,
 	xfs_lsn_t		commit_lsn,
@@ -1653,9 +1653,6 @@ xfs_trans_commit_cil(
 		return error;
 
 	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
-
-	/* xfs_trans_free_items() unlocks them first */
-	xfs_trans_free_items(tp, *commit_lsn, 0);
 	xfs_trans_free(tp);
 	return 0;
 }
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
index e2d93d8..62da86c 100644
--- a/fs/xfs/xfs_trans_priv.h
+++ b/fs/xfs/xfs_trans_priv.h
@@ -25,7 +25,8 @@ struct xfs_trans;
 
 void	xfs_trans_add_item(struct xfs_trans *, struct xfs_log_item *);
 void	xfs_trans_del_item(struct xfs_log_item *);
-
+void	xfs_trans_free_items(struct xfs_trans *tp, xfs_lsn_t commit_lsn,
+				int flags);
 void	xfs_trans_item_committed(struct xfs_log_item *lip,
 				xfs_lsn_t commit_lsn, int aborted);
 void	xfs_trans_unreserve_and_mod_sb(struct xfs_trans *tp);
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 66d585c..4c7c7bf 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -2299,15 +2299,22 @@ xfs_alloc_file_space(
 			e = allocatesize_fsb;
 		}
 
+		/*
+		 * The transaction reservation is limited to a 32-bit block
+		 * count, hence we need to limit the number of blocks we are
+		 * trying to reserve to avoid an overflow. We can't allocate
+		 * more than @nimaps extents, and an extent is limited on disk
+		 * to MAXEXTLEN (21 bits), so use that to enforce the limit.
+		 */
+		resblks = min_t(xfs_fileoff_t, (e - s), (MAXEXTLEN * nimaps));
 		if (unlikely(rt)) {
-			resrtextents = qblocks = (uint)(e - s);
+			resrtextents = qblocks = resblks;
 			resrtextents /= mp->m_sb.sb_rextsize;
 			resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
 			quota_flag = XFS_QMOPT_RES_RTBLKS;
 		} else {
 			resrtextents = 0;
-			resblks = qblocks = \
-				XFS_DIOSTRAT_SPACE_RES(mp, (uint)(e - s));
+			resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resblks);
 			quota_flag = XFS_QMOPT_RES_REGBLKS;
 		}