Merge branch 'master' into for-linus-merged

This merge pulls the XFS master branch into the latest Linus master.
This results in a merge conflict whose best fix is not obvious.
I manually fixed the conflict, in "fs/xfs/xfs_iget.c".

Dave Chinner had done work that resulted in RCU freeing of inodes
separate from what Nick Piggin had done, and their results differed
slightly in xfs_inode_free().  The fix updates Nick's call_rcu()
with the use of VFS_I(), while incorporating needed updates to some
XFS inode fields implemented in Dave's series.  Dave's RCU callback
function has also been removed.

Signed-off-by: Alex Elder <aelder@sgi.com>

fs/xfs/linux-2.6/sv.h

@@ -1,59 +0,0 @@
-/*
- * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#ifndef __XFS_SUPPORT_SV_H__
-#define __XFS_SUPPORT_SV_H__
-
-#include <linux/wait.h>
-#include <linux/sched.h>
-#include <linux/spinlock.h>
-
-/*
- * Synchronisation variables.
- *
- * (Parameters "pri", "svf" and "rts" are not implemented)
- */
-
-typedef struct sv_s {
-	wait_queue_head_t waiters;
-} sv_t;
-
-static inline void _sv_wait(sv_t *sv, spinlock_t *lock)
-{
-	DECLARE_WAITQUEUE(wait, current);
-
-	add_wait_queue_exclusive(&sv->waiters, &wait);
-	__set_current_state(TASK_UNINTERRUPTIBLE);
-	spin_unlock(lock);
-
-	schedule();
-
-	remove_wait_queue(&sv->waiters, &wait);
-}
-
-#define sv_init(sv,flag,name) \
-	init_waitqueue_head(&(sv)->waiters)
-#define sv_destroy(sv) \
-	/*NOTHING*/
-#define sv_wait(sv, pri, lock, s) \
-	_sv_wait(sv, lock)
-#define sv_signal(sv) \
-	wake_up(&(sv)->waiters)
-#define sv_broadcast(sv) \
-	wake_up_all(&(sv)->waiters)
-
-#endif /* __XFS_SUPPORT_SV_H__ */

fs/xfs/linux-2.6/xfs_aops.c

@@ -38,15 +38,6 @@
 #include <linux/pagevec.h>
 #include <linux/writeback.h>
 
-/*
- * Types of I/O for bmap clustering and I/O completion tracking.
- */
-enum {
-	IO_READ,	/* mapping for a read */
-	IO_DELAY,	/* mapping covers delalloc region */
-	IO_UNWRITTEN,	/* mapping covers allocated but uninitialized data */
-	IO_NEW		/* just allocated */
-};
 
 /*
  * Prime number of hash buckets since address is used as the key.
@@ -182,9 +173,6 @@ xfs_setfilesize(
 	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
 	xfs_fsize_t		isize;
 
-	ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
-	ASSERT(ioend->io_type != IO_READ);
-
 	if (unlikely(ioend->io_error))
 		return 0;
 
@@ -244,10 +232,8 @@ xfs_end_io(
 	 * We might have to update the on-disk file size after extending
 	 * writes.
 	 */
-	if (ioend->io_type != IO_READ) {
-		error = xfs_setfilesize(ioend);
-		ASSERT(!error || error == EAGAIN);
-	}
+	error = xfs_setfilesize(ioend);
+	ASSERT(!error || error == EAGAIN);
 
 	/*
 	 * If we didn't complete processing of the ioend, requeue it to the
@@ -318,14 +304,63 @@ STATIC int
 xfs_map_blocks(
 	struct inode		*inode,
 	loff_t			offset,
-	ssize_t			count,
 	struct xfs_bmbt_irec	*imap,
-	int			flags)
+	int			type,
+	int			nonblocking)
 {
-	int			nmaps = 1;
-	int			new = 0;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	ssize_t			count = 1 << inode->i_blkbits;
+	xfs_fileoff_t		offset_fsb, end_fsb;
+	int			error = 0;
+	int			bmapi_flags = XFS_BMAPI_ENTIRE;
+	int			nimaps = 1;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	if (type == IO_UNWRITTEN)
+		bmapi_flags |= XFS_BMAPI_IGSTATE;
+
+	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED)) {
+		if (nonblocking)
+			return -XFS_ERROR(EAGAIN);
+		xfs_ilock(ip, XFS_ILOCK_SHARED);
+	}
 
-	return -xfs_iomap(XFS_I(inode), offset, count, flags, imap, &nmaps, &new);
+	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
+	       (ip->i_df.if_flags & XFS_IFEXTENTS));
+	ASSERT(offset <= mp->m_maxioffset);
+
+	if (offset + count > mp->m_maxioffset)
+		count = mp->m_maxioffset - offset;
+	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+			  bmapi_flags,  NULL, 0, imap, &nimaps, NULL);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	if (error)
+		return -XFS_ERROR(error);
+
+	if (type == IO_DELALLOC &&
+	    (!nimaps || isnullstartblock(imap->br_startblock))) {
+		error = xfs_iomap_write_allocate(ip, offset, count, imap);
+		if (!error)
+			trace_xfs_map_blocks_alloc(ip, offset, count, type, imap);
+		return -XFS_ERROR(error);
+	}
+
+#ifdef DEBUG
+	if (type == IO_UNWRITTEN) {
+		ASSERT(nimaps);
+		ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+		ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
+	}
+#endif
+	if (nimaps)
+		trace_xfs_map_blocks_found(ip, offset, count, type, imap);
+	return 0;
 }
 
 STATIC int
@@ -380,26 +415,18 @@ xfs_submit_ioend_bio(
 
 	submit_bio(wbc->sync_mode == WB_SYNC_ALL ?
 		   WRITE_SYNC_PLUG : WRITE, bio);
-	ASSERT(!bio_flagged(bio, BIO_EOPNOTSUPP));
-	bio_put(bio);
 }
 
 STATIC struct bio *
 xfs_alloc_ioend_bio(
 	struct buffer_head	*bh)
 {
-	struct bio		*bio;
 	int			nvecs = bio_get_nr_vecs(bh->b_bdev);
-
-	do {
-		bio = bio_alloc(GFP_NOIO, nvecs);
-		nvecs >>= 1;
-	} while (!bio);
+	struct bio		*bio = bio_alloc(GFP_NOIO, nvecs);
 
 	ASSERT(bio->bi_private == NULL);
 	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
 	bio->bi_bdev = bh->b_bdev;
-	bio_get(bio);
 	return bio;
 }
 
@@ -470,9 +497,8 @@ xfs_submit_ioend(
 	/* Pass 1 - start writeback */
 	do {
 		next = ioend->io_list;
-		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
+		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private)
 			xfs_start_buffer_writeback(bh);
-		}
 	} while ((ioend = next) != NULL);
 
 	/* Pass 2 - submit I/O */
@@ -600,116 +626,12 @@ xfs_map_at_offset(
 	ASSERT(imap->br_startblock != HOLESTARTBLOCK);
 	ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
 
-	lock_buffer(bh);
 	xfs_map_buffer(inode, bh, imap, offset);
-	bh->b_bdev = xfs_find_bdev_for_inode(inode);
 	set_buffer_mapped(bh);
 	clear_buffer_delay(bh);
 	clear_buffer_unwritten(bh);
 }
 
-/*
- * Look for a page at index that is suitable for clustering.
- */
-STATIC unsigned int
-xfs_probe_page(
-	struct page		*page,
-	unsigned int		pg_offset)
-{
-	struct buffer_head	*bh, *head;
-	int			ret = 0;
-
-	if (PageWriteback(page))
-		return 0;
-	if (!PageDirty(page))
-		return 0;
-	if (!page->mapping)
-		return 0;
-	if (!page_has_buffers(page))
-		return 0;
-
-	bh = head = page_buffers(page);
-	do {
-		if (!buffer_uptodate(bh))
-			break;
-		if (!buffer_mapped(bh))
-			break;
-		ret += bh->b_size;
-		if (ret >= pg_offset)
-			break;
-	} while ((bh = bh->b_this_page) != head);
-
-	return ret;
-}
-
-STATIC size_t
-xfs_probe_cluster(
-	struct inode		*inode,
-	struct page		*startpage,
-	struct buffer_head	*bh,
-	struct buffer_head	*head)
-{
-	struct pagevec		pvec;
-	pgoff_t			tindex, tlast, tloff;
-	size_t			total = 0;
-	int			done = 0, i;
-
-	/* First sum forwards in this page */
-	do {
-		if (!buffer_uptodate(bh) || !buffer_mapped(bh))
-			return total;
-		total += bh->b_size;
-	} while ((bh = bh->b_this_page) != head);
-
-	/* if we reached the end of the page, sum forwards in following pages */
-	tlast = i_size_read(inode) >> PAGE_CACHE_SHIFT;
-	tindex = startpage->index + 1;
-
-	/* Prune this back to avoid pathological behavior */
-	tloff = min(tlast, startpage->index + 64);
-
-	pagevec_init(&pvec, 0);
-	while (!done && tindex <= tloff) {
-		unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
-
-		if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
-			break;
-
-		for (i = 0; i < pagevec_count(&pvec); i++) {
-			struct page *page = pvec.pages[i];
-			size_t pg_offset, pg_len = 0;
-
-			if (tindex == tlast) {
-				pg_offset =
-				    i_size_read(inode) & (PAGE_CACHE_SIZE - 1);
-				if (!pg_offset) {
-					done = 1;
-					break;
-				}
-			} else
-				pg_offset = PAGE_CACHE_SIZE;
-
-			if (page->index == tindex && trylock_page(page)) {
-				pg_len = xfs_probe_page(page, pg_offset);
-				unlock_page(page);
-			}
-
-			if (!pg_len) {
-				done = 1;
-				break;
-			}
-
-			total += pg_len;
-			tindex++;
-		}
-
-		pagevec_release(&pvec);
-		cond_resched();
-	}
-
-	return total;
-}
-
 /*
  * Test if a given page is suitable for writing as part of an unwritten
  * or delayed allocate extent.
@@ -731,9 +653,9 @@ xfs_is_delayed_page(
 			if (buffer_unwritten(bh))
 				acceptable = (type == IO_UNWRITTEN);
 			else if (buffer_delay(bh))
-				acceptable = (type == IO_DELAY);
+				acceptable = (type == IO_DELALLOC);
 			else if (buffer_dirty(bh) && buffer_mapped(bh))
-				acceptable = (type == IO_NEW);
+				acceptable = (type == IO_OVERWRITE);
 			else
 				break;
 		} while ((bh = bh->b_this_page) != head);
@@ -758,8 +680,7 @@ xfs_convert_page(
 	loff_t			tindex,
 	struct xfs_bmbt_irec	*imap,
 	xfs_ioend_t		**ioendp,
-	struct writeback_control *wbc,
-	int			all_bh)
+	struct writeback_control *wbc)
 {
 	struct buffer_head	*bh, *head;
 	xfs_off_t		end_offset;
@@ -814,37 +735,30 @@ xfs_convert_page(
 			continue;
 		}
 
-		if (buffer_unwritten(bh) || buffer_delay(bh)) {
+		if (buffer_unwritten(bh) || buffer_delay(bh) ||
+		    buffer_mapped(bh)) {
 			if (buffer_unwritten(bh))
 				type = IO_UNWRITTEN;
+			else if (buffer_delay(bh))
+				type = IO_DELALLOC;
 			else
-				type = IO_DELAY;
+				type = IO_OVERWRITE;
 
 			if (!xfs_imap_valid(inode, imap, offset)) {
 				done = 1;
 				continue;
 			}
 
-			ASSERT(imap->br_startblock != HOLESTARTBLOCK);
-			ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
-
-			xfs_map_at_offset(inode, bh, imap, offset);
+			lock_buffer(bh);
+			if (type != IO_OVERWRITE)
+				xfs_map_at_offset(inode, bh, imap, offset);
 			xfs_add_to_ioend(inode, bh, offset, type,
 					 ioendp, done);
 
 			page_dirty--;
 			count++;
 		} else {
-			type = IO_NEW;
-			if (buffer_mapped(bh) && all_bh) {
-				lock_buffer(bh);
-				xfs_add_to_ioend(inode, bh, offset,
-						type, ioendp, done);
-				count++;
-				page_dirty--;
-			} else {
-				done = 1;
-			}
+			done = 1;
 		}
 	} while (offset += len, (bh = bh->b_this_page) != head);
 
@@ -876,7 +790,6 @@ xfs_cluster_write(
 	struct xfs_bmbt_irec	*imap,
 	xfs_ioend_t		**ioendp,
 	struct writeback_control *wbc,
-	int			all_bh,
 	pgoff_t			tlast)
 {
 	struct pagevec		pvec;
@@ -891,7 +804,7 @@ xfs_cluster_write(
 
 		for (i = 0; i < pagevec_count(&pvec); i++) {
 			done = xfs_convert_page(inode, pvec.pages[i], tindex++,
-					imap, ioendp, wbc, all_bh);
+					imap, ioendp, wbc);
 			if (done)
 				break;
 		}
@@ -935,7 +848,7 @@ xfs_aops_discard_page(
 	struct buffer_head	*bh, *head;
 	loff_t			offset = page_offset(page);
 
-	if (!xfs_is_delayed_page(page, IO_DELAY))
+	if (!xfs_is_delayed_page(page, IO_DELALLOC))
 		goto out_invalidate;
 
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
@@ -1002,10 +915,10 @@ xfs_vm_writepage(
 	unsigned int		type;
 	__uint64_t              end_offset;
 	pgoff_t                 end_index, last_index;
-	ssize_t			size, len;
-	int			flags, err, imap_valid = 0, uptodate = 1;
+	ssize_t			len;
+	int			err, imap_valid = 0, uptodate = 1;
 	int			count = 0;
-	int			all_bh = 0;
+	int			nonblocking = 0;
 
 	trace_xfs_writepage(inode, page, 0);
 
@@ -1056,10 +969,14 @@ xfs_vm_writepage(
 
 	bh = head = page_buffers(page);
 	offset = page_offset(page);
-	flags = BMAPI_READ;
-	type = IO_NEW;
+	type = IO_OVERWRITE;
+
+	if (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking)
+		nonblocking = 1;
 
 	do {
+		int new_ioend = 0;
+
 		if (offset >= end_offset)
 			break;
 		if (!buffer_uptodate(bh))
@@ -1076,90 +993,54 @@ xfs_vm_writepage(
 			continue;
 		}
 
-		if (imap_valid)
-			imap_valid = xfs_imap_valid(inode, &imap, offset);
-
-		if (buffer_unwritten(bh) || buffer_delay(bh)) {
-			int new_ioend = 0;
-
-			/*
-			 * Make sure we don't use a read-only iomap
-			 */
-			if (flags == BMAPI_READ)
-				imap_valid = 0;
-
-			if (buffer_unwritten(bh)) {
+		if (buffer_unwritten(bh)) {
+			if (type != IO_UNWRITTEN) {
 				type = IO_UNWRITTEN;
-				flags = BMAPI_WRITE | BMAPI_IGNSTATE;
-			} else if (buffer_delay(bh)) {
-				type = IO_DELAY;
-				flags = BMAPI_ALLOCATE;
-
-				if (wbc->sync_mode == WB_SYNC_NONE)
-					flags |= BMAPI_TRYLOCK;
-			}
-
-			if (!imap_valid) {
-				/*
-				 * If we didn't have a valid mapping then we
-				 * need to ensure that we put the new mapping
-				 * in a new ioend structure. This needs to be
-				 * done to ensure that the ioends correctly
-				 * reflect the block mappings at io completion
-				 * for unwritten extent conversion.
-				 */
-				new_ioend = 1;
-				err = xfs_map_blocks(inode, offset, len,
-						&imap, flags);
-				if (err)
-					goto error;
-				imap_valid = xfs_imap_valid(inode, &imap,
-							    offset);
+				imap_valid = 0;
 			}
-			if (imap_valid) {
-				xfs_map_at_offset(inode, bh, &imap, offset);
-				xfs_add_to_ioend(inode, bh, offset, type,
-						 &ioend, new_ioend);
-				count++;
+		} else if (buffer_delay(bh)) {
+			if (type != IO_DELALLOC) {
+				type = IO_DELALLOC;
+				imap_valid = 0;
 			}
 		} else if (buffer_uptodate(bh)) {
-			/*
-			 * we got here because the buffer is already mapped.
-			 * That means it must already have extents allocated
-			 * underneath it. Map the extent by reading it.
-			 */
-			if (!imap_valid || flags != BMAPI_READ) {
-				flags = BMAPI_READ;
-				size = xfs_probe_cluster(inode, page, bh, head);
-				err = xfs_map_blocks(inode, offset, size,
-						&imap, flags);
-				if (err)
-					goto error;
-				imap_valid = xfs_imap_valid(inode, &imap,
-							    offset);
+			if (type != IO_OVERWRITE) {
+				type = IO_OVERWRITE;
+				imap_valid = 0;
 			}
+		} else {
+			if (PageUptodate(page)) {
+				ASSERT(buffer_mapped(bh));
+				imap_valid = 0;
+			}
+			continue;
+		}
 
+		if (imap_valid)
+			imap_valid = xfs_imap_valid(inode, &imap, offset);
+		if (!imap_valid) {
 			/*
-			 * We set the type to IO_NEW in case we are doing a
-			 * small write at EOF that is extending the file but
-			 * without needing an allocation. We need to update the
-			 * file size on I/O completion in this case so it is
-			 * the same case as having just allocated a new extent
-			 * that we are writing into for the first time.
+			 * If we didn't have a valid mapping then we need to
+			 * put the new mapping into a separate ioend structure.
+			 * This ensures non-contiguous extents always have
+			 * separate ioends, which is particularly important
+			 * for unwritten extent conversion at I/O completion
+			 * time.
 			 */
-			type = IO_NEW;
-			if (trylock_buffer(bh)) {
-				if (imap_valid)
-					all_bh = 1;
-				xfs_add_to_ioend(inode, bh, offset, type,
-						&ioend, !imap_valid);
-				count++;
-			} else {
-				imap_valid = 0;
-			}
-		} else if (PageUptodate(page)) {
-			ASSERT(buffer_mapped(bh));
-			imap_valid = 0;
+			new_ioend = 1;
+			err = xfs_map_blocks(inode, offset, &imap, type,
+					     nonblocking);
+			if (err)
+				goto error;
+			imap_valid = xfs_imap_valid(inode, &imap, offset);
+		}
+		if (imap_valid) {
+			lock_buffer(bh);
+			if (type != IO_OVERWRITE)
+				xfs_map_at_offset(inode, bh, &imap, offset);
+			xfs_add_to_ioend(inode, bh, offset, type, &ioend,
+					 new_ioend);
+			count++;
 		}
 
 		if (!iohead)
@@ -1188,7 +1069,7 @@ xfs_vm_writepage(
 			end_index = last_index;
 
 		xfs_cluster_write(inode, page->index + 1, &imap, &ioend,
-					wbc, all_bh, end_index);
+				  wbc, end_index);
 	}
 
 	if (iohead)
@@ -1257,13 +1138,19 @@ __xfs_get_blocks(
 	int			create,
 	int			direct)
 {
-	int			flags = create ? BMAPI_WRITE : BMAPI_READ;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_fileoff_t		offset_fsb, end_fsb;
+	int			error = 0;
+	int			lockmode = 0;
 	struct xfs_bmbt_irec	imap;
+	int			nimaps = 1;
 	xfs_off_t		offset;
 	ssize_t			size;
-	int			nimap = 1;
 	int			new = 0;
-	int			error;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
 
 	offset = (xfs_off_t)iblock << inode->i_blkbits;
 	ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
@@ -1272,15 +1159,45 @@ __xfs_get_blocks(
 	if (!create && direct && offset >= i_size_read(inode))
 		return 0;
 
-	if (direct && create)
-		flags |= BMAPI_DIRECT;
+	if (create) {
+		lockmode = XFS_ILOCK_EXCL;
+		xfs_ilock(ip, lockmode);
+	} else {
+		lockmode = xfs_ilock_map_shared(ip);
+	}
+
+	ASSERT(offset <= mp->m_maxioffset);
+	if (offset + size > mp->m_maxioffset)
+		size = mp->m_maxioffset - offset;
+	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + size);
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 
-	error = xfs_iomap(XFS_I(inode), offset, size, flags, &imap, &nimap,
-			  &new);
+	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+			  XFS_BMAPI_ENTIRE,  NULL, 0, &imap, &nimaps, NULL);
 	if (error)
-		return -error;
-	if (nimap == 0)
-		return 0;
+		goto out_unlock;
+
+	if (create &&
+	    (!nimaps ||
+	     (imap.br_startblock == HOLESTARTBLOCK ||
+	      imap.br_startblock == DELAYSTARTBLOCK))) {
+		if (direct) {
+			error = xfs_iomap_write_direct(ip, offset, size,
+						       &imap, nimaps);
+		} else {
+			error = xfs_iomap_write_delay(ip, offset, size, &imap);
+		}
+		if (error)
+			goto out_unlock;
+
+		trace_xfs_get_blocks_alloc(ip, offset, size, 0, &imap);
+	} else if (nimaps) {
+		trace_xfs_get_blocks_found(ip, offset, size, 0, &imap);
+	} else {
+		trace_xfs_get_blocks_notfound(ip, offset, size);
+		goto out_unlock;
+	}
+	xfs_iunlock(ip, lockmode);
 
 	if (imap.br_startblock != HOLESTARTBLOCK &&
 	    imap.br_startblock != DELAYSTARTBLOCK) {
@@ -1347,6 +1264,10 @@ __xfs_get_blocks(
 	}
 
 	return 0;
+
+out_unlock:
+	xfs_iunlock(ip, lockmode);
+	return -error;
 }
 
 int
@@ -1434,7 +1355,7 @@ xfs_vm_direct_IO(
 	ssize_t			ret;
 
 	if (rw & WRITE) {
-		iocb->private = xfs_alloc_ioend(inode, IO_NEW);
+		iocb->private = xfs_alloc_ioend(inode, IO_DIRECT);
 
 		ret = __blockdev_direct_IO(rw, iocb, inode, bdev, iov,
 					    offset, nr_segs,

fs/xfs/linux-2.6/xfs_aops.h

@@ -22,6 +22,22 @@ extern struct workqueue_struct *xfsdatad_workqueue;
 extern struct workqueue_struct *xfsconvertd_workqueue;
 extern mempool_t *xfs_ioend_pool;
 
+/*
+ * Types of I/O for bmap clustering and I/O completion tracking.
+ */
+enum {
+	IO_DIRECT = 0,	/* special case for direct I/O ioends */
+	IO_DELALLOC,	/* mapping covers delalloc region */
+	IO_UNWRITTEN,	/* mapping covers allocated but uninitialized data */
+	IO_OVERWRITE,	/* mapping covers already allocated extent */
+};
+
+#define XFS_IO_TYPES \
+	{ 0,			"" }, \
+	{ IO_DELALLOC,		"delalloc" }, \
+	{ IO_UNWRITTEN,		"unwritten" }, \
+	{ IO_OVERWRITE,		"overwrite" }
+
 /*
  * xfs_ioend struct manages large extent writes for XFS.
  * It can manage several multi-page bio's at once.

fs/xfs/linux-2.6/xfs_buf.c

@@ -44,12 +44,7 @@
 
 static kmem_zone_t *xfs_buf_zone;
 STATIC int xfsbufd(void *);
-STATIC int xfsbufd_wakeup(struct shrinker *, int, gfp_t);
 STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
-static struct shrinker xfs_buf_shake = {
-	.shrink = xfsbufd_wakeup,
-	.seeks = DEFAULT_SEEKS,
-};
 
 static struct workqueue_struct *xfslogd_workqueue;
 struct workqueue_struct *xfsdatad_workqueue;
@@ -168,8 +163,79 @@ test_page_region(
 }
 
 /*
- *	Internal xfs_buf_t object manipulation
+ * xfs_buf_lru_add - add a buffer to the LRU.
+ *
+ * The LRU takes a new reference to the buffer so that it will only be freed
+ * once the shrinker takes the buffer off the LRU.
  */
+STATIC void
+xfs_buf_lru_add(
+	struct xfs_buf	*bp)
+{
+	struct xfs_buftarg *btp = bp->b_target;
+
+	spin_lock(&btp->bt_lru_lock);
+	if (list_empty(&bp->b_lru)) {
+		atomic_inc(&bp->b_hold);
+		list_add_tail(&bp->b_lru, &btp->bt_lru);
+		btp->bt_lru_nr++;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+}
+
+/*
+ * xfs_buf_lru_del - remove a buffer from the LRU
+ *
+ * The unlocked check is safe here because it only occurs when there are not
+ * b_lru_ref counts left on the inode under the pag->pag_buf_lock. it is there
+ * to optimise the shrinker removing the buffer from the LRU and calling
+ * xfs_buf_free(). i.e. it removes an unneccessary round trip on the
+ * bt_lru_lock.
+ */
+STATIC void
+xfs_buf_lru_del(
+	struct xfs_buf	*bp)
+{
+	struct xfs_buftarg *btp = bp->b_target;
+
+	if (list_empty(&bp->b_lru))
+		return;
+
+	spin_lock(&btp->bt_lru_lock);
+	if (!list_empty(&bp->b_lru)) {
+		list_del_init(&bp->b_lru);
+		btp->bt_lru_nr--;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+}
+
+/*
+ * When we mark a buffer stale, we remove the buffer from the LRU and clear the
+ * b_lru_ref count so that the buffer is freed immediately when the buffer
+ * reference count falls to zero. If the buffer is already on the LRU, we need
+ * to remove the reference that LRU holds on the buffer.
+ *
+ * This prevents build-up of stale buffers on the LRU.
+ */
+void
+xfs_buf_stale(
+	struct xfs_buf	*bp)
+{
+	bp->b_flags |= XBF_STALE;
+	atomic_set(&(bp)->b_lru_ref, 0);
+	if (!list_empty(&bp->b_lru)) {
+		struct xfs_buftarg *btp = bp->b_target;
+
+		spin_lock(&btp->bt_lru_lock);
+		if (!list_empty(&bp->b_lru)) {
+			list_del_init(&bp->b_lru);
+			btp->bt_lru_nr--;
+			atomic_dec(&bp->b_hold);
+		}
+		spin_unlock(&btp->bt_lru_lock);
+	}
+	ASSERT(atomic_read(&bp->b_hold) >= 1);
+}
 
 STATIC void
 _xfs_buf_initialize(
@@ -186,7 +252,9 @@ _xfs_buf_initialize(
 
 	memset(bp, 0, sizeof(xfs_buf_t));
 	atomic_set(&bp->b_hold, 1);
+	atomic_set(&bp->b_lru_ref, 1);
 	init_completion(&bp->b_iowait);
+	INIT_LIST_HEAD(&bp->b_lru);
 	INIT_LIST_HEAD(&bp->b_list);
 	RB_CLEAR_NODE(&bp->b_rbnode);
 	sema_init(&bp->b_sema, 0); /* held, no waiters */
@@ -262,6 +330,8 @@ xfs_buf_free(
 {
 	trace_xfs_buf_free(bp, _RET_IP_);
 
+	ASSERT(list_empty(&bp->b_lru));
+
 	if (bp->b_flags & (_XBF_PAGE_CACHE|_XBF_PAGES)) {
 		uint		i;
 
@@ -337,7 +407,6 @@ _xfs_buf_lookup_pages(
 					__func__, gfp_mask);
 
 			XFS_STATS_INC(xb_page_retries);
-			xfsbufd_wakeup(NULL, 0, gfp_mask);
 			congestion_wait(BLK_RW_ASYNC, HZ/50);
 			goto retry;
 		}
@@ -828,6 +897,7 @@ xfs_buf_rele(
 
 	if (!pag) {
 		ASSERT(!bp->b_relse);
+		ASSERT(list_empty(&bp->b_lru));
 		ASSERT(RB_EMPTY_NODE(&bp->b_rbnode));
 		if (atomic_dec_and_test(&bp->b_hold))
 			xfs_buf_free(bp);
@@ -835,13 +905,19 @@ xfs_buf_rele(
 	}
 
 	ASSERT(!RB_EMPTY_NODE(&bp->b_rbnode));
+
 	ASSERT(atomic_read(&bp->b_hold) > 0);
 	if (atomic_dec_and_lock(&bp->b_hold, &pag->pag_buf_lock)) {
 		if (bp->b_relse) {
 			atomic_inc(&bp->b_hold);
 			spin_unlock(&pag->pag_buf_lock);
 			bp->b_relse(bp);
+		} else if (!(bp->b_flags & XBF_STALE) &&
+			   atomic_read(&bp->b_lru_ref)) {
+			xfs_buf_lru_add(bp);
+			spin_unlock(&pag->pag_buf_lock);
 		} else {
+			xfs_buf_lru_del(bp);
 			ASSERT(!(bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)));
 			rb_erase(&bp->b_rbnode, &pag->pag_buf_tree);
 			spin_unlock(&pag->pag_buf_lock);
@@ -1438,51 +1514,84 @@ xfs_buf_iomove(
  */
 
 /*
- *	Wait for any bufs with callbacks that have been submitted but
- *	have not yet returned... walk the hash list for the target.
+ * Wait for any bufs with callbacks that have been submitted but have not yet
+ * returned. These buffers will have an elevated hold count, so wait on those
+ * while freeing all the buffers only held by the LRU.
  */
 void
 xfs_wait_buftarg(
 	struct xfs_buftarg	*btp)
 {
-	struct xfs_perag	*pag;
-	uint			i;
+	struct xfs_buf		*bp;
 
-	for (i = 0; i < btp->bt_mount->m_sb.sb_agcount; i++) {
-		pag = xfs_perag_get(btp->bt_mount, i);
-		spin_lock(&pag->pag_buf_lock);
-		while (rb_first(&pag->pag_buf_tree)) {
-			spin_unlock(&pag->pag_buf_lock);
+restart:
+	spin_lock(&btp->bt_lru_lock);
+	while (!list_empty(&btp->bt_lru)) {
+		bp = list_first_entry(&btp->bt_lru, struct xfs_buf, b_lru);
+		if (atomic_read(&bp->b_hold) > 1) {
+			spin_unlock(&btp->bt_lru_lock);
 			delay(100);
-			spin_lock(&pag->pag_buf_lock);
+			goto restart;
 		}
-		spin_unlock(&pag->pag_buf_lock);
-		xfs_perag_put(pag);
+		/*
+		 * clear the LRU reference count so the bufer doesn't get
+		 * ignored in xfs_buf_rele().
+		 */
+		atomic_set(&bp->b_lru_ref, 0);
+		spin_unlock(&btp->bt_lru_lock);
+		xfs_buf_rele(bp);
+		spin_lock(&btp->bt_lru_lock);
 	}
+	spin_unlock(&btp->bt_lru_lock);
 }
 
-/*
- *	buftarg list for delwrite queue processing
- */
-static LIST_HEAD(xfs_buftarg_list);
-static DEFINE_SPINLOCK(xfs_buftarg_lock);
-
-STATIC void
-xfs_register_buftarg(
-	xfs_buftarg_t           *btp)
+int
+xfs_buftarg_shrink(
+	struct shrinker		*shrink,
+	int			nr_to_scan,
+	gfp_t			mask)
 {
-	spin_lock(&xfs_buftarg_lock);
-	list_add(&btp->bt_list, &xfs_buftarg_list);
-	spin_unlock(&xfs_buftarg_lock);
-}
+	struct xfs_buftarg	*btp = container_of(shrink,
+					struct xfs_buftarg, bt_shrinker);
+	struct xfs_buf		*bp;
+	LIST_HEAD(dispose);
 
-STATIC void
-xfs_unregister_buftarg(
-	xfs_buftarg_t           *btp)
-{
-	spin_lock(&xfs_buftarg_lock);
-	list_del(&btp->bt_list);
-	spin_unlock(&xfs_buftarg_lock);
+	if (!nr_to_scan)
+		return btp->bt_lru_nr;
+
+	spin_lock(&btp->bt_lru_lock);
+	while (!list_empty(&btp->bt_lru)) {
+		if (nr_to_scan-- <= 0)
+			break;
+
+		bp = list_first_entry(&btp->bt_lru, struct xfs_buf, b_lru);
+
+		/*
+		 * Decrement the b_lru_ref count unless the value is already
+		 * zero. If the value is already zero, we need to reclaim the
+		 * buffer, otherwise it gets another trip through the LRU.
+		 */
+		if (!atomic_add_unless(&bp->b_lru_ref, -1, 0)) {
+			list_move_tail(&bp->b_lru, &btp->bt_lru);
+			continue;
+		}
+
+		/*
+		 * remove the buffer from the LRU now to avoid needing another
+		 * lock round trip inside xfs_buf_rele().
+		 */
+		list_move(&bp->b_lru, &dispose);
+		btp->bt_lru_nr--;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+
+	while (!list_empty(&dispose)) {
+		bp = list_first_entry(&dispose, struct xfs_buf, b_lru);
+		list_del_init(&bp->b_lru);
+		xfs_buf_rele(bp);
+	}
+
+	return btp->bt_lru_nr;
 }
 
 void
@@ -1490,17 +1599,14 @@ xfs_free_buftarg(
 	struct xfs_mount	*mp,
 	struct xfs_buftarg	*btp)
 {
+	unregister_shrinker(&btp->bt_shrinker);
+
 	xfs_flush_buftarg(btp, 1);
 	if (mp->m_flags & XFS_MOUNT_BARRIER)
 		xfs_blkdev_issue_flush(btp);
 	iput(btp->bt_mapping->host);
 
-	/* Unregister the buftarg first so that we don't get a
-	 * wakeup finding a non-existent task
-	 */
-	xfs_unregister_buftarg(btp);
 	kthread_stop(btp->bt_task);
-
 	kmem_free(btp);
 }
 
@@ -1597,20 +1703,13 @@ xfs_alloc_delwrite_queue(
 	xfs_buftarg_t		*btp,
 	const char		*fsname)
 {
-	int	error = 0;
-
-	INIT_LIST_HEAD(&btp->bt_list);
 	INIT_LIST_HEAD(&btp->bt_delwrite_queue);
 	spin_lock_init(&btp->bt_delwrite_lock);
 	btp->bt_flags = 0;
 	btp->bt_task = kthread_run(xfsbufd, btp, "xfsbufd/%s", fsname);
-	if (IS_ERR(btp->bt_task)) {
-		error = PTR_ERR(btp->bt_task);
-		goto out_error;
-	}
-	xfs_register_buftarg(btp);
-out_error:
-	return error;
+	if (IS_ERR(btp->bt_task))
+		return PTR_ERR(btp->bt_task);
+	return 0;
 }
 
 xfs_buftarg_t *
@@ -1627,12 +1726,17 @@ xfs_alloc_buftarg(
 	btp->bt_mount = mp;
 	btp->bt_dev =  bdev->bd_dev;
 	btp->bt_bdev = bdev;
+	INIT_LIST_HEAD(&btp->bt_lru);
+	spin_lock_init(&btp->bt_lru_lock);
 	if (xfs_setsize_buftarg_early(btp, bdev))
 		goto error;
 	if (xfs_mapping_buftarg(btp, bdev))
 		goto error;
 	if (xfs_alloc_delwrite_queue(btp, fsname))
 		goto error;
+	btp->bt_shrinker.shrink = xfs_buftarg_shrink;
+	btp->bt_shrinker.seeks = DEFAULT_SEEKS;
+	register_shrinker(&btp->bt_shrinker);
 	return btp;
 
 error:
@@ -1737,27 +1841,6 @@ xfs_buf_runall_queues(
 	flush_workqueue(queue);
 }
 
-STATIC int
-xfsbufd_wakeup(
-	struct shrinker		*shrink,
-	int			priority,
-	gfp_t			mask)
-{
-	xfs_buftarg_t		*btp;
-
-	spin_lock(&xfs_buftarg_lock);
-	list_for_each_entry(btp, &xfs_buftarg_list, bt_list) {
-		if (test_bit(XBT_FORCE_SLEEP, &btp->bt_flags))
-			continue;
-		if (list_empty(&btp->bt_delwrite_queue))
-			continue;
-		set_bit(XBT_FORCE_FLUSH, &btp->bt_flags);
-		wake_up_process(btp->bt_task);
-	}
-	spin_unlock(&xfs_buftarg_lock);
-	return 0;
-}
-
 /*
  * Move as many buffers as specified to the supplied list
  * idicating if we skipped any buffers to prevent deadlocks.
@@ -1952,7 +2035,6 @@ xfs_buf_init(void)
 	if (!xfsconvertd_workqueue)
 		goto out_destroy_xfsdatad_workqueue;
 
-	register_shrinker(&xfs_buf_shake);
 	return 0;
 
  out_destroy_xfsdatad_workqueue:
@@ -1968,7 +2050,6 @@ xfs_buf_init(void)
 void
 xfs_buf_terminate(void)
 {
-	unregister_shrinker(&xfs_buf_shake);
 	destroy_workqueue(xfsconvertd_workqueue);
 	destroy_workqueue(xfsdatad_workqueue);
 	destroy_workqueue(xfslogd_workqueue);

fs/xfs/linux-2.6/xfs_buf.h

@@ -128,10 +128,15 @@ typedef struct xfs_buftarg {
 
 	/* per device delwri queue */
 	struct task_struct	*bt_task;
-	struct list_head	bt_list;
 	struct list_head	bt_delwrite_queue;
 	spinlock_t		bt_delwrite_lock;
 	unsigned long		bt_flags;
+
+	/* LRU control structures */
+	struct shrinker		bt_shrinker;
+	struct list_head	bt_lru;
+	spinlock_t		bt_lru_lock;
+	unsigned int		bt_lru_nr;
 } xfs_buftarg_t;
 
 /*
@@ -164,9 +169,11 @@ typedef struct xfs_buf {
 	xfs_off_t		b_file_offset;	/* offset in file */
 	size_t			b_buffer_length;/* size of buffer in bytes */
 	atomic_t		b_hold;		/* reference count */
+	atomic_t		b_lru_ref;	/* lru reclaim ref count */
 	xfs_buf_flags_t		b_flags;	/* status flags */
 	struct semaphore	b_sema;		/* semaphore for lockables */
 
+	struct list_head	b_lru;		/* lru list */
 	wait_queue_head_t	b_waiters;	/* unpin waiters */
 	struct list_head	b_list;
 	struct xfs_perag	*b_pag;		/* contains rbtree root */
@@ -264,7 +271,8 @@ extern void xfs_buf_terminate(void);
 #define XFS_BUF_ZEROFLAGS(bp)	((bp)->b_flags &= \
 		~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED))
 
-#define XFS_BUF_STALE(bp)	((bp)->b_flags |= XBF_STALE)
+void xfs_buf_stale(struct xfs_buf *bp);
+#define XFS_BUF_STALE(bp)	xfs_buf_stale(bp);
 #define XFS_BUF_UNSTALE(bp)	((bp)->b_flags &= ~XBF_STALE)
 #define XFS_BUF_ISSTALE(bp)	((bp)->b_flags & XBF_STALE)
 #define XFS_BUF_SUPER_STALE(bp)	do {				\
@@ -328,9 +336,15 @@ extern void xfs_buf_terminate(void);
 #define XFS_BUF_SIZE(bp)		((bp)->b_buffer_length)
 #define XFS_BUF_SET_SIZE(bp, cnt)	((bp)->b_buffer_length = (cnt))
 
-#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)	do { } while (0)
+static inline void
+xfs_buf_set_ref(
+	struct xfs_buf	*bp,
+	int		lru_ref)
+{
+	atomic_set(&bp->b_lru_ref, lru_ref);
+}
+#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)	xfs_buf_set_ref(bp, ref)
 #define XFS_BUF_SET_VTYPE(bp, type)		do { } while (0)
-#define XFS_BUF_SET_REF(bp, ref)		do { } while (0)
 
 #define XFS_BUF_ISPINNED(bp)	atomic_read(&((bp)->b_pin_count))
 

fs/xfs/linux-2.6/xfs_export.c

@@ -70,8 +70,16 @@ xfs_fs_encode_fh(
 	else
 		fileid_type = FILEID_INO32_GEN_PARENT;
 
-	/* filesystem may contain 64bit inode numbers */
-	if (!(XFS_M(inode->i_sb)->m_flags & XFS_MOUNT_SMALL_INUMS))
+	/*
+	 * If the the filesystem may contain 64bit inode numbers, we need
+	 * to use larger file handles that can represent them.
+	 *
+	 * While we only allocate inodes that do not fit into 32 bits any
+	 * large enough filesystem may contain them, thus the slightly
+	 * confusing looking conditional below.
+	 */
+	if (!(XFS_M(inode->i_sb)->m_flags & XFS_MOUNT_SMALL_INUMS) ||
+	    (XFS_M(inode->i_sb)->m_flags & XFS_MOUNT_32BITINODES))
 		fileid_type |= XFS_FILEID_TYPE_64FLAG;
 
 	/*

fs/xfs/linux-2.6/xfs_linux.h

@@ -37,7 +37,6 @@
 
 #include <kmem.h>
 #include <mrlock.h>
-#include <sv.h>
 #include <time.h>
 
 #include <support/debug.h>

fs/xfs/linux-2.6/xfs_super.c

@@ -834,8 +834,11 @@ xfsaild_wakeup(
 	struct xfs_ail		*ailp,
 	xfs_lsn_t		threshold_lsn)
 {
-	ailp->xa_target = threshold_lsn;
-	wake_up_process(ailp->xa_task);
+	/* only ever move the target forwards */
+	if (XFS_LSN_CMP(threshold_lsn, ailp->xa_target) > 0) {
+		ailp->xa_target = threshold_lsn;
+		wake_up_process(ailp->xa_task);
+	}
 }
 
 STATIC int
@@ -847,8 +850,17 @@ xfsaild(
 	long		tout = 0; /* milliseconds */
 
 	while (!kthread_should_stop()) {
-		schedule_timeout_interruptible(tout ?
-				msecs_to_jiffies(tout) : MAX_SCHEDULE_TIMEOUT);
+		/*
+		 * for short sleeps indicating congestion, don't allow us to
+		 * get woken early. Otherwise all we do is bang on the AIL lock
+		 * without making progress.
+		 */
+		if (tout && tout <= 20)
+			__set_current_state(TASK_KILLABLE);
+		else
+			__set_current_state(TASK_INTERRUPTIBLE);
+		schedule_timeout(tout ?
+				 msecs_to_jiffies(tout) : MAX_SCHEDULE_TIMEOUT);
 
 		/* swsusp */
 		try_to_freeze();
@@ -1118,6 +1130,8 @@ xfs_fs_evict_inode(
 	 */
 	ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));
 	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+	lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+			&xfs_iolock_reclaimable, "xfs_iolock_reclaimable");
 
 	xfs_inactive(ip);
 }

fs/xfs/linux-2.6/xfs_sync.c

@@ -53,14 +53,30 @@ xfs_inode_ag_walk_grab(
 {
 	struct inode		*inode = VFS_I(ip);
 
+	ASSERT(rcu_read_lock_held());
+
+	/*
+	 * check for stale RCU freed inode
+	 *
+	 * If the inode has been reallocated, it doesn't matter if it's not in
+	 * the AG we are walking - we are walking for writeback, so if it
+	 * passes all the "valid inode" checks and is dirty, then we'll write
+	 * it back anyway.  If it has been reallocated and still being
+	 * initialised, the XFS_INEW check below will catch it.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	if (!ip->i_ino)
+		goto out_unlock_noent;
+
+	/* avoid new or reclaimable inodes. Leave for reclaim code to flush */
+	if (__xfs_iflags_test(ip, XFS_INEW | XFS_IRECLAIMABLE | XFS_IRECLAIM))
+		goto out_unlock_noent;
+	spin_unlock(&ip->i_flags_lock);
+
 	/* nothing to sync during shutdown */
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
 		return EFSCORRUPTED;
 
-	/* avoid new or reclaimable inodes. Leave for reclaim code to flush */
-	if (xfs_iflags_test(ip, XFS_INEW | XFS_IRECLAIMABLE | XFS_IRECLAIM))
-		return ENOENT;
-
 	/* If we can't grab the inode, it must on it's way to reclaim. */
 	if (!igrab(inode))
 		return ENOENT;
@@ -72,6 +88,10 @@ xfs_inode_ag_walk_grab(
 
 	/* inode is valid */
 	return 0;
+
+out_unlock_noent:
+	spin_unlock(&ip->i_flags_lock);
+	return ENOENT;
 }
 
 STATIC int
@@ -98,12 +118,12 @@ restart:
 		int		error = 0;
 		int		i;
 
-		read_lock(&pag->pag_ici_lock);
+		rcu_read_lock();
 		nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
 					(void **)batch, first_index,
 					XFS_LOOKUP_BATCH);
 		if (!nr_found) {
-			read_unlock(&pag->pag_ici_lock);
+			rcu_read_unlock();
 			break;
 		}
 
@@ -118,18 +138,26 @@ restart:
 				batch[i] = NULL;
 
 			/*
-			 * Update the index for the next lookup. Catch overflows
-			 * into the next AG range which can occur if we have inodes
-			 * in the last block of the AG and we are currently
-			 * pointing to the last inode.
+			 * Update the index for the next lookup. Catch
+			 * overflows into the next AG range which can occur if
+			 * we have inodes in the last block of the AG and we
+			 * are currently pointing to the last inode.
+			 *
+			 * Because we may see inodes that are from the wrong AG
+			 * due to RCU freeing and reallocation, only update the
+			 * index if it lies in this AG. It was a race that lead
+			 * us to see this inode, so another lookup from the
+			 * same index will not find it again.
 			 */
+			if (XFS_INO_TO_AGNO(mp, ip->i_ino) != pag->pag_agno)
+				continue;
 			first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
 			if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
 				done = 1;
 		}
 
 		/* unlock now we've grabbed the inodes. */
-		read_unlock(&pag->pag_ici_lock);
+		rcu_read_unlock();
 
 		for (i = 0; i < nr_found; i++) {
 			if (!batch[i])
@@ -592,12 +620,12 @@ xfs_inode_set_reclaim_tag(
 	struct xfs_perag *pag;
 
 	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
-	write_lock(&pag->pag_ici_lock);
+	spin_lock(&pag->pag_ici_lock);
 	spin_lock(&ip->i_flags_lock);
 	__xfs_inode_set_reclaim_tag(pag, ip);
 	__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
 	spin_unlock(&ip->i_flags_lock);
-	write_unlock(&pag->pag_ici_lock);
+	spin_unlock(&pag->pag_ici_lock);
 	xfs_perag_put(pag);
 }
 
@@ -639,9 +667,14 @@ xfs_reclaim_inode_grab(
 	struct xfs_inode	*ip,
 	int			flags)
 {
+	ASSERT(rcu_read_lock_held());
+
+	/* quick check for stale RCU freed inode */
+	if (!ip->i_ino)
+		return 1;
 
 	/*
-	 * do some unlocked checks first to avoid unnecceary lock traffic.
+	 * do some unlocked checks first to avoid unnecessary lock traffic.
 	 * The first is a flush lock check, the second is a already in reclaim
 	 * check. Only do these checks if we are not going to block on locks.
 	 */
@@ -654,11 +687,16 @@ xfs_reclaim_inode_grab(
 	 * The radix tree lock here protects a thread in xfs_iget from racing
 	 * with us starting reclaim on the inode.  Once we have the
 	 * XFS_IRECLAIM flag set it will not touch us.
+	 *
+	 * Due to RCU lookup, we may find inodes that have been freed and only
+	 * have XFS_IRECLAIM set.  Indeed, we may see reallocated inodes that
+	 * aren't candidates for reclaim at all, so we must check the
+	 * XFS_IRECLAIMABLE is set first before proceeding to reclaim.
 	 */
 	spin_lock(&ip->i_flags_lock);
-	ASSERT_ALWAYS(__xfs_iflags_test(ip, XFS_IRECLAIMABLE));
-	if (__xfs_iflags_test(ip, XFS_IRECLAIM)) {
-		/* ignore as it is already under reclaim */
+	if (!__xfs_iflags_test(ip, XFS_IRECLAIMABLE) ||
+	    __xfs_iflags_test(ip, XFS_IRECLAIM)) {
+		/* not a reclaim candidate. */
 		spin_unlock(&ip->i_flags_lock);
 		return 1;
 	}
@@ -795,12 +833,12 @@ reclaim:
 	 * added to the tree assert that it's been there before to catch
 	 * problems with the inode life time early on.
 	 */
-	write_lock(&pag->pag_ici_lock);
+	spin_lock(&pag->pag_ici_lock);
 	if (!radix_tree_delete(&pag->pag_ici_root,
 				XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino)))
 		ASSERT(0);
 	__xfs_inode_clear_reclaim(pag, ip);
-	write_unlock(&pag->pag_ici_lock);
+	spin_unlock(&pag->pag_ici_lock);
 
 	/*
 	 * Here we do an (almost) spurious inode lock in order to coordinate
@@ -864,14 +902,14 @@ restart:
 			struct xfs_inode *batch[XFS_LOOKUP_BATCH];
 			int	i;
 
-			write_lock(&pag->pag_ici_lock);
+			rcu_read_lock();
 			nr_found = radix_tree_gang_lookup_tag(
 					&pag->pag_ici_root,
 					(void **)batch, first_index,
 					XFS_LOOKUP_BATCH,
 					XFS_ICI_RECLAIM_TAG);
 			if (!nr_found) {
-				write_unlock(&pag->pag_ici_lock);
+				rcu_read_unlock();
 				break;
 			}
 
@@ -891,14 +929,24 @@ restart:
 				 * occur if we have inodes in the last block of
 				 * the AG and we are currently pointing to the
 				 * last inode.
+				 *
+				 * Because we may see inodes that are from the
+				 * wrong AG due to RCU freeing and
+				 * reallocation, only update the index if it
+				 * lies in this AG. It was a race that lead us
+				 * to see this inode, so another lookup from
+				 * the same index will not find it again.
 				 */
+				if (XFS_INO_TO_AGNO(mp, ip->i_ino) !=
+								pag->pag_agno)
+					continue;
 				first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
 				if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
 					done = 1;
 			}
 
 			/* unlock now we've grabbed the inodes. */
-			write_unlock(&pag->pag_ici_lock);
+			rcu_read_unlock();
 
 			for (i = 0; i < nr_found; i++) {
 				if (!batch[i])

fs/xfs/linux-2.6/xfs_trace.h

@@ -766,8 +766,8 @@ DECLARE_EVENT_CLASS(xfs_loggrant_class,
 		__field(int, curr_res)
 		__field(int, unit_res)
 		__field(unsigned int, flags)
-		__field(void *, reserve_headq)
-		__field(void *, write_headq)
+		__field(int, reserveq)
+		__field(int, writeq)
 		__field(int, grant_reserve_cycle)
 		__field(int, grant_reserve_bytes)
 		__field(int, grant_write_cycle)
@@ -784,19 +784,21 @@ DECLARE_EVENT_CLASS(xfs_loggrant_class,
 		__entry->curr_res = tic->t_curr_res;
 		__entry->unit_res = tic->t_unit_res;
 		__entry->flags = tic->t_flags;
-		__entry->reserve_headq = log->l_reserve_headq;
-		__entry->write_headq = log->l_write_headq;
-		__entry->grant_reserve_cycle = log->l_grant_reserve_cycle;
-		__entry->grant_reserve_bytes = log->l_grant_reserve_bytes;
-		__entry->grant_write_cycle = log->l_grant_write_cycle;
-		__entry->grant_write_bytes = log->l_grant_write_bytes;
+		__entry->reserveq = list_empty(&log->l_reserveq);
+		__entry->writeq = list_empty(&log->l_writeq);
+		xlog_crack_grant_head(&log->l_grant_reserve_head,
+				&__entry->grant_reserve_cycle,
+				&__entry->grant_reserve_bytes);
+		xlog_crack_grant_head(&log->l_grant_write_head,
+				&__entry->grant_write_cycle,
+				&__entry->grant_write_bytes);
 		__entry->curr_cycle = log->l_curr_cycle;
 		__entry->curr_block = log->l_curr_block;
-		__entry->tail_lsn = log->l_tail_lsn;
+		__entry->tail_lsn = atomic64_read(&log->l_tail_lsn);
 	),
 	TP_printk("dev %d:%d type %s t_ocnt %u t_cnt %u t_curr_res %u "
-		  "t_unit_res %u t_flags %s reserve_headq 0x%p "
-		  "write_headq 0x%p grant_reserve_cycle %d "
+		  "t_unit_res %u t_flags %s reserveq %s "
+		  "writeq %s grant_reserve_cycle %d "
 		  "grant_reserve_bytes %d grant_write_cycle %d "
 		  "grant_write_bytes %d curr_cycle %d curr_block %d "
 		  "tail_cycle %d tail_block %d",
@@ -807,8 +809,8 @@ DECLARE_EVENT_CLASS(xfs_loggrant_class,
 		  __entry->curr_res,
 		  __entry->unit_res,
 		  __print_flags(__entry->flags, "|", XLOG_TIC_FLAGS),
-		  __entry->reserve_headq,
-		  __entry->write_headq,
+		  __entry->reserveq ? "empty" : "active",
+		  __entry->writeq ? "empty" : "active",
 		  __entry->grant_reserve_cycle,
 		  __entry->grant_reserve_bytes,
 		  __entry->grant_write_cycle,
@@ -835,6 +837,7 @@ DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep1);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake1);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep2);
 DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
@@ -842,6 +845,7 @@ DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep1);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake1);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep2);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
 DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_sub);
@@ -935,10 +939,10 @@ DEFINE_PAGE_EVENT(xfs_writepage);
 DEFINE_PAGE_EVENT(xfs_releasepage);
 DEFINE_PAGE_EVENT(xfs_invalidatepage);
 
-DECLARE_EVENT_CLASS(xfs_iomap_class,
+DECLARE_EVENT_CLASS(xfs_imap_class,
 	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
-		 int flags, struct xfs_bmbt_irec *irec),
-	TP_ARGS(ip, offset, count, flags, irec),
+		 int type, struct xfs_bmbt_irec *irec),
+	TP_ARGS(ip, offset, count, type, irec),
 	TP_STRUCT__entry(
 		__field(dev_t, dev)
 		__field(xfs_ino_t, ino)
@@ -946,7 +950,7 @@ DECLARE_EVENT_CLASS(xfs_iomap_class,
 		__field(loff_t, new_size)
 		__field(loff_t, offset)
 		__field(size_t, count)
-		__field(int, flags)
+		__field(int, type)
 		__field(xfs_fileoff_t, startoff)
 		__field(xfs_fsblock_t, startblock)
 		__field(xfs_filblks_t, blockcount)
@@ -958,13 +962,13 @@ DECLARE_EVENT_CLASS(xfs_iomap_class,
 		__entry->new_size = ip->i_new_size;
 		__entry->offset = offset;
 		__entry->count = count;
-		__entry->flags = flags;
+		__entry->type = type;
 		__entry->startoff = irec ? irec->br_startoff : 0;
 		__entry->startblock = irec ? irec->br_startblock : 0;
 		__entry->blockcount = irec ? irec->br_blockcount : 0;
 	),
 	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
-		  "offset 0x%llx count %zd flags %s "
+		  "offset 0x%llx count %zd type %s "
 		  "startoff 0x%llx startblock %lld blockcount 0x%llx",
 		  MAJOR(__entry->dev), MINOR(__entry->dev),
 		  __entry->ino,
@@ -972,20 +976,21 @@ DECLARE_EVENT_CLASS(xfs_iomap_class,
 		  __entry->new_size,
 		  __entry->offset,
 		  __entry->count,
-		  __print_flags(__entry->flags, "|", BMAPI_FLAGS),
+		  __print_symbolic(__entry->type, XFS_IO_TYPES),
 		  __entry->startoff,
 		  (__int64_t)__entry->startblock,
 		  __entry->blockcount)
 )
 
 #define DEFINE_IOMAP_EVENT(name)	\
-DEFINE_EVENT(xfs_iomap_class, name,	\
+DEFINE_EVENT(xfs_imap_class, name,	\
 	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,	\
-		 int flags, struct xfs_bmbt_irec *irec),		\
-	TP_ARGS(ip, offset, count, flags, irec))
-DEFINE_IOMAP_EVENT(xfs_iomap_enter);
-DEFINE_IOMAP_EVENT(xfs_iomap_found);
-DEFINE_IOMAP_EVENT(xfs_iomap_alloc);
+		 int type, struct xfs_bmbt_irec *irec),		\
+	TP_ARGS(ip, offset, count, type, irec))
+DEFINE_IOMAP_EVENT(xfs_map_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_map_blocks_alloc);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_alloc);
 
 DECLARE_EVENT_CLASS(xfs_simple_io_class,
 	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count),
@@ -1022,6 +1027,7 @@ DEFINE_EVENT(xfs_simple_io_class, name,	\
 	TP_ARGS(ip, offset, count))
 DEFINE_SIMPLE_IO_EVENT(xfs_delalloc_enospc);
 DEFINE_SIMPLE_IO_EVENT(xfs_unwritten_convert);
+DEFINE_SIMPLE_IO_EVENT(xfs_get_blocks_notfound);
 
 
 TRACE_EVENT(xfs_itruncate_start,
@@ -1420,6 +1426,7 @@ DEFINE_EVENT(xfs_alloc_class, name, \
 	TP_PROTO(struct xfs_alloc_arg *args), \
 	TP_ARGS(args))
 DEFINE_ALLOC_EVENT(xfs_alloc_exact_done);
+DEFINE_ALLOC_EVENT(xfs_alloc_exact_notfound);
 DEFINE_ALLOC_EVENT(xfs_alloc_exact_error);
 DEFINE_ALLOC_EVENT(xfs_alloc_near_nominleft);
 DEFINE_ALLOC_EVENT(xfs_alloc_near_first);

fs/xfs/quota/xfs_dquot.c

@@ -149,7 +149,6 @@ xfs_qm_dqdestroy(
 	ASSERT(list_empty(&dqp->q_freelist));
 
 	mutex_destroy(&dqp->q_qlock);
-	sv_destroy(&dqp->q_pinwait);
 	kmem_zone_free(xfs_Gqm->qm_dqzone, dqp);
 
 	atomic_dec(&xfs_Gqm->qm_totaldquots);

fs/xfs/xfs_ag.h

@@ -227,7 +227,7 @@ typedef struct xfs_perag {
 
 	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
 
-	rwlock_t	pag_ici_lock;	/* incore inode lock */
+	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
 	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
 	int		pag_ici_reclaimable;	/* reclaimable inodes */
 	struct mutex	pag_ici_reclaim_lock;	/* serialisation point */

fs/xfs/xfs_alloc.c

@@ -577,61 +577,58 @@ xfs_alloc_ag_vextent_exact(
 	xfs_extlen_t	rlen;	/* length of returned extent */
 
 	ASSERT(args->alignment == 1);
+
 	/*
 	 * Allocate/initialize a cursor for the by-number freespace btree.
 	 */
 	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_BNO);
+					  args->agno, XFS_BTNUM_BNO);
+
 	/*
 	 * Lookup bno and minlen in the btree (minlen is irrelevant, really).
 	 * Look for the closest free block <= bno, it must contain bno
 	 * if any free block does.
 	 */
-	if ((error = xfs_alloc_lookup_le(bno_cur, args->agbno, args->minlen, &i)))
+	error = xfs_alloc_lookup_le(bno_cur, args->agbno, args->minlen, &i);
+	if (error)
 		goto error0;
-	if (!i) {
-		/*
-		 * Didn't find it, return null.
-		 */
-		xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
-		args->agbno = NULLAGBLOCK;
-		return 0;
-	}
+	if (!i)
+		goto not_found;
+
 	/*
 	 * Grab the freespace record.
 	 */
-	if ((error = xfs_alloc_get_rec(bno_cur, &fbno, &flen, &i)))
+	error = xfs_alloc_get_rec(bno_cur, &fbno, &flen, &i);
+	if (error)
 		goto error0;
 	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 	ASSERT(fbno <= args->agbno);
 	minend = args->agbno + args->minlen;
 	maxend = args->agbno + args->maxlen;
 	fend = fbno + flen;
+
 	/*
 	 * Give up if the freespace isn't long enough for the minimum request.
 	 */
-	if (fend < minend) {
-		xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
-		args->agbno = NULLAGBLOCK;
-		return 0;
-	}
+	if (fend < minend)
+		goto not_found;
+
 	/*
 	 * End of extent will be smaller of the freespace end and the
 	 * maximal requested end.
-	 */
-	end = XFS_AGBLOCK_MIN(fend, maxend);
-	/*
+	 *
 	 * Fix the length according to mod and prod if given.
 	 */
+	end = XFS_AGBLOCK_MIN(fend, maxend);
 	args->len = end - args->agbno;
 	xfs_alloc_fix_len(args);
-	if (!xfs_alloc_fix_minleft(args)) {
-		xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
-		return 0;
-	}
+	if (!xfs_alloc_fix_minleft(args))
+		goto not_found;
+
 	rlen = args->len;
 	ASSERT(args->agbno + rlen <= fend);
 	end = args->agbno + rlen;
+
 	/*
 	 * We are allocating agbno for rlen [agbno .. end]
 	 * Allocate/initialize a cursor for the by-size btree.
@@ -640,16 +637,25 @@ xfs_alloc_ag_vextent_exact(
 		args->agno, XFS_BTNUM_CNT);
 	ASSERT(args->agbno + args->len <=
 		be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length));
-	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
-			args->agbno, args->len, XFSA_FIXUP_BNO_OK))) {
+	error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen, args->agbno,
+				      args->len, XFSA_FIXUP_BNO_OK);
+	if (error) {
 		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
 		goto error0;
 	}
+
 	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
 	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
 
-	trace_xfs_alloc_exact_done(args);
 	args->wasfromfl = 0;
+	trace_xfs_alloc_exact_done(args);
+	return 0;
+
+not_found:
+	/* Didn't find it, return null. */
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	args->agbno = NULLAGBLOCK;
+	trace_xfs_alloc_exact_notfound(args);
 	return 0;
 
 error0:
@@ -658,6 +664,95 @@ error0:
 	return error;
 }
 
+/*
+ * Search the btree in a given direction via the search cursor and compare
+ * the records found against the good extent we've already found.
+ */
+STATIC int
+xfs_alloc_find_best_extent(
+	struct xfs_alloc_arg	*args,	/* allocation argument structure */
+	struct xfs_btree_cur	**gcur,	/* good cursor */
+	struct xfs_btree_cur	**scur,	/* searching cursor */
+	xfs_agblock_t		gdiff,	/* difference for search comparison */
+	xfs_agblock_t		*sbno,	/* extent found by search */
+	xfs_extlen_t		*slen,
+	xfs_extlen_t		*slena,	/* aligned length */
+	int			dir)	/* 0 = search right, 1 = search left */
+{
+	xfs_agblock_t		bno;
+	xfs_agblock_t		new;
+	xfs_agblock_t		sdiff;
+	int			error;
+	int			i;
+
+	/* The good extent is perfect, no need to  search. */
+	if (!gdiff)
+		goto out_use_good;
+
+	/*
+	 * Look until we find a better one, run out of space or run off the end.
+	 */
+	do {
+		error = xfs_alloc_get_rec(*scur, sbno, slen, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		xfs_alloc_compute_aligned(*sbno, *slen, args->alignment,
+					  args->minlen, &bno, slena);
+
+		/*
+		 * The good extent is closer than this one.
+		 */
+		if (!dir) {
+			if (bno >= args->agbno + gdiff)
+				goto out_use_good;
+		} else {
+			if (bno <= args->agbno - gdiff)
+				goto out_use_good;
+		}
+
+		/*
+		 * Same distance, compare length and pick the best.
+		 */
+		if (*slena >= args->minlen) {
+			args->len = XFS_EXTLEN_MIN(*slena, args->maxlen);
+			xfs_alloc_fix_len(args);
+
+			sdiff = xfs_alloc_compute_diff(args->agbno, args->len,
+						       args->alignment, *sbno,
+						       *slen, &new);
+
+			/*
+			 * Choose closer size and invalidate other cursor.
+			 */
+			if (sdiff < gdiff)
+				goto out_use_search;
+			goto out_use_good;
+		}
+
+		if (!dir)
+			error = xfs_btree_increment(*scur, 0, &i);
+		else
+			error = xfs_btree_decrement(*scur, 0, &i);
+		if (error)
+			goto error0;
+	} while (i);
+
+out_use_good:
+	xfs_btree_del_cursor(*scur, XFS_BTREE_NOERROR);
+	*scur = NULL;
+	return 0;
+
+out_use_search:
+	xfs_btree_del_cursor(*gcur, XFS_BTREE_NOERROR);
+	*gcur = NULL;
+	return 0;
+
+error0:
+	/* caller invalidates cursors */
+	return error;
+}
+
 /*
  * Allocate a variable extent near bno in the allocation group agno.
  * Extent's length (returned in len) will be between minlen and maxlen,
@@ -925,203 +1020,45 @@ xfs_alloc_ag_vextent_near(
 			}
 		}
 	} while (bno_cur_lt || bno_cur_gt);
+
 	/*
 	 * Got both cursors still active, need to find better entry.
 	 */
 	if (bno_cur_lt && bno_cur_gt) {
-		/*
-		 * Left side is long enough, look for a right side entry.
-		 */
 		if (ltlena >= args->minlen) {
 			/*
-			 * Fix up the length.
+			 * Left side is good, look for a right side entry.
 			 */
 			args->len = XFS_EXTLEN_MIN(ltlena, args->maxlen);
 			xfs_alloc_fix_len(args);
-			rlen = args->len;
-			ltdiff = xfs_alloc_compute_diff(args->agbno, rlen,
+			ltdiff = xfs_alloc_compute_diff(args->agbno, args->len,
 				args->alignment, ltbno, ltlen, &ltnew);
+
+			error = xfs_alloc_find_best_extent(args,
+						&bno_cur_lt, &bno_cur_gt,
+						ltdiff, &gtbno, &gtlen, &gtlena,
+						0 /* search right */);
+		} else {
+			ASSERT(gtlena >= args->minlen);
+
 			/*
-			 * Not perfect.
-			 */
-			if (ltdiff) {
-				/*
-				 * Look until we find a better one, run out of
-				 * space, or run off the end.
-				 */
-				while (bno_cur_lt && bno_cur_gt) {
-					if ((error = xfs_alloc_get_rec(
-							bno_cur_gt, &gtbno,
-							&gtlen, &i)))
-						goto error0;
-					XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-					xfs_alloc_compute_aligned(gtbno, gtlen,
-						args->alignment, args->minlen,
-						&gtbnoa, &gtlena);
-					/*
-					 * The left one is clearly better.
-					 */
-					if (gtbnoa >= args->agbno + ltdiff) {
-						xfs_btree_del_cursor(
-							bno_cur_gt,
-							XFS_BTREE_NOERROR);
-						bno_cur_gt = NULL;
-						break;
-					}
-					/*
-					 * If we reach a big enough entry,
-					 * compare the two and pick the best.
-					 */
-					if (gtlena >= args->minlen) {
-						args->len =
-							XFS_EXTLEN_MIN(gtlena,
-								args->maxlen);
-						xfs_alloc_fix_len(args);
-						rlen = args->len;
-						gtdiff = xfs_alloc_compute_diff(
-							args->agbno, rlen,
-							args->alignment,
-							gtbno, gtlen, &gtnew);
-						/*
-						 * Right side is better.
-						 */
-						if (gtdiff < ltdiff) {
-							xfs_btree_del_cursor(
-								bno_cur_lt,
-								XFS_BTREE_NOERROR);
-							bno_cur_lt = NULL;
-						}
-						/*
-						 * Left side is better.
-						 */
-						else {
-							xfs_btree_del_cursor(
-								bno_cur_gt,
-								XFS_BTREE_NOERROR);
-							bno_cur_gt = NULL;
-						}
-						break;
-					}
-					/*
-					 * Fell off the right end.
-					 */
-					if ((error = xfs_btree_increment(
-							bno_cur_gt, 0, &i)))
-						goto error0;
-					if (!i) {
-						xfs_btree_del_cursor(
-							bno_cur_gt,
-							XFS_BTREE_NOERROR);
-						bno_cur_gt = NULL;
-						break;
-					}
-				}
-			}
-			/*
-			 * The left side is perfect, trash the right side.
-			 */
-			else {
-				xfs_btree_del_cursor(bno_cur_gt,
-						     XFS_BTREE_NOERROR);
-				bno_cur_gt = NULL;
-			}
-		}
-		/*
-		 * It's the right side that was found first, look left.
-		 */
-		else {
-			/*
-			 * Fix up the length.
+			 * Right side is good, look for a left side entry.
 			 */
 			args->len = XFS_EXTLEN_MIN(gtlena, args->maxlen);
 			xfs_alloc_fix_len(args);
-			rlen = args->len;
-			gtdiff = xfs_alloc_compute_diff(args->agbno, rlen,
+			gtdiff = xfs_alloc_compute_diff(args->agbno, args->len,
 				args->alignment, gtbno, gtlen, &gtnew);
-			/*
-			 * Right side entry isn't perfect.
-			 */
-			if (gtdiff) {
-				/*
-				 * Look until we find a better one, run out of
-				 * space, or run off the end.
-				 */
-				while (bno_cur_lt && bno_cur_gt) {
-					if ((error = xfs_alloc_get_rec(
-							bno_cur_lt, &ltbno,
-							&ltlen, &i)))
-						goto error0;
-					XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-					xfs_alloc_compute_aligned(ltbno, ltlen,
-						args->alignment, args->minlen,
-						&ltbnoa, &ltlena);
-					/*
-					 * The right one is clearly better.
-					 */
-					if (ltbnoa <= args->agbno - gtdiff) {
-						xfs_btree_del_cursor(
-							bno_cur_lt,
-							XFS_BTREE_NOERROR);
-						bno_cur_lt = NULL;
-						break;
-					}
-					/*
-					 * If we reach a big enough entry,
-					 * compare the two and pick the best.
-					 */
-					if (ltlena >= args->minlen) {
-						args->len = XFS_EXTLEN_MIN(
-							ltlena, args->maxlen);
-						xfs_alloc_fix_len(args);
-						rlen = args->len;
-						ltdiff = xfs_alloc_compute_diff(
-							args->agbno, rlen,
-							args->alignment,
-							ltbno, ltlen, &ltnew);
-						/*
-						 * Left side is better.
-						 */
-						if (ltdiff < gtdiff) {
-							xfs_btree_del_cursor(
-								bno_cur_gt,
-								XFS_BTREE_NOERROR);
-							bno_cur_gt = NULL;
-						}
-						/*
-						 * Right side is better.
-						 */
-						else {
-							xfs_btree_del_cursor(
-								bno_cur_lt,
-								XFS_BTREE_NOERROR);
-							bno_cur_lt = NULL;
-						}
-						break;
-					}
-					/*
-					 * Fell off the left end.
-					 */
-					if ((error = xfs_btree_decrement(
-							bno_cur_lt, 0, &i)))
-						goto error0;
-					if (!i) {
-						xfs_btree_del_cursor(bno_cur_lt,
-							XFS_BTREE_NOERROR);
-						bno_cur_lt = NULL;
-						break;
-					}
-				}
-			}
-			/*
-			 * The right side is perfect, trash the left side.
-			 */
-			else {
-				xfs_btree_del_cursor(bno_cur_lt,
-					XFS_BTREE_NOERROR);
-				bno_cur_lt = NULL;
-			}
+
+			error = xfs_alloc_find_best_extent(args,
+						&bno_cur_gt, &bno_cur_lt,
+						gtdiff, &ltbno, &ltlen, &ltlena,
+						1 /* search left */);
 		}
+
+		if (error)
+			goto error0;
 	}
+
 	/*
 	 * If we couldn't get anything, give up.
 	 */
@@ -1130,6 +1067,7 @@ xfs_alloc_ag_vextent_near(
 		args->agbno = NULLAGBLOCK;
 		return 0;
 	}
+
 	/*
 	 * At this point we have selected a freespace entry, either to the
 	 * left or to the right.  If it's on the right, copy all the
@@ -1146,6 +1084,7 @@ xfs_alloc_ag_vextent_near(
 		j = 1;
 	} else
 		j = 0;
+
 	/*
 	 * Fix up the length and compute the useful address.
 	 */

fs/xfs/xfs_attr_leaf.c

@@ -637,7 +637,7 @@ xfs_attr_shortform_list(xfs_attr_list_context_t *context)
 	 * It didn't all fit, so we have to sort everything on hashval.
 	 */
 	sbsize = sf->hdr.count * sizeof(*sbuf);
-	sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP);
+	sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP | KM_NOFS);
 
 	/*
 	 * Scan the attribute list for the rest of the entries, storing
@@ -2386,7 +2386,7 @@ xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
 				args.dp = context->dp;
 				args.whichfork = XFS_ATTR_FORK;
 				args.valuelen = valuelen;
-				args.value = kmem_alloc(valuelen, KM_SLEEP);
+				args.value = kmem_alloc(valuelen, KM_SLEEP | KM_NOFS);
 				args.rmtblkno = be32_to_cpu(name_rmt->valueblk);
 				args.rmtblkcnt = XFS_B_TO_FSB(args.dp->i_mount, valuelen);
 				retval = xfs_attr_rmtval_get(&args);

fs/xfs/xfs_btree.c

@@ -634,9 +634,8 @@ xfs_btree_read_bufl(
 		return error;
 	}
 	ASSERT(!bp || !XFS_BUF_GETERROR(bp));
-	if (bp != NULL) {
+	if (bp)
 		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, refval);
-	}
 	*bpp = bp;
 	return 0;
 }
@@ -944,13 +943,13 @@ xfs_btree_set_refs(
 	switch (cur->bc_btnum) {
 	case XFS_BTNUM_BNO:
 	case XFS_BTNUM_CNT:
-		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_MAP, XFS_ALLOC_BTREE_REF);
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, XFS_ALLOC_BTREE_REF);
 		break;
 	case XFS_BTNUM_INO:
-		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_INOMAP, XFS_INO_BTREE_REF);
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_INOMAP, XFS_INO_BTREE_REF);
 		break;
 	case XFS_BTNUM_BMAP:
-		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_MAP, XFS_BMAP_BTREE_REF);
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, XFS_BMAP_BTREE_REF);
 		break;
 	default:
 		ASSERT(0);

fs/xfs/xfs_buf_item.c

@@ -142,7 +142,7 @@ xfs_buf_item_log_check(
 #endif
 
 STATIC void	xfs_buf_error_relse(xfs_buf_t *bp);
-STATIC void	xfs_buf_do_callbacks(xfs_buf_t *bp, xfs_log_item_t *lip);
+STATIC void	xfs_buf_do_callbacks(struct xfs_buf *bp);
 
 /*
  * This returns the number of log iovecs needed to log the
@@ -450,7 +450,7 @@ xfs_buf_item_unpin(
 		 * xfs_trans_ail_delete() drops the AIL lock.
 		 */
 		if (bip->bli_flags & XFS_BLI_STALE_INODE) {
-			xfs_buf_do_callbacks(bp, (xfs_log_item_t *)bip);
+			xfs_buf_do_callbacks(bp);
 			XFS_BUF_SET_FSPRIVATE(bp, NULL);
 			XFS_BUF_CLR_IODONE_FUNC(bp);
 		} else {
@@ -918,15 +918,26 @@ xfs_buf_attach_iodone(
 	XFS_BUF_SET_IODONE_FUNC(bp, xfs_buf_iodone_callbacks);
 }
 
+/*
+ * We can have many callbacks on a buffer. Running the callbacks individually
+ * can cause a lot of contention on the AIL lock, so we allow for a single
+ * callback to be able to scan the remaining lip->li_bio_list for other items
+ * of the same type and callback to be processed in the first call.
+ *
+ * As a result, the loop walking the callback list below will also modify the
+ * list. it removes the first item from the list and then runs the callback.
+ * The loop then restarts from the new head of the list. This allows the
+ * callback to scan and modify the list attached to the buffer and we don't
+ * have to care about maintaining a next item pointer.
+ */
 STATIC void
 xfs_buf_do_callbacks(
-	xfs_buf_t	*bp,
-	xfs_log_item_t	*lip)
+	struct xfs_buf		*bp)
 {
-	xfs_log_item_t	*nlip;
+	struct xfs_log_item	*lip;
 
-	while (lip != NULL) {
-		nlip = lip->li_bio_list;
+	while ((lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *)) != NULL) {
+		XFS_BUF_SET_FSPRIVATE(bp, lip->li_bio_list);
 		ASSERT(lip->li_cb != NULL);
 		/*
 		 * Clear the next pointer so we don't have any
@@ -936,7 +947,6 @@ xfs_buf_do_callbacks(
 		 */
 		lip->li_bio_list = NULL;
 		lip->li_cb(bp, lip);
-		lip = nlip;
 	}
 }
 
@@ -970,7 +980,7 @@ xfs_buf_iodone_callbacks(
 			ASSERT(XFS_BUF_TARGET(bp) == mp->m_ddev_targp);
 			XFS_BUF_SUPER_STALE(bp);
 			trace_xfs_buf_item_iodone(bp, _RET_IP_);
-			xfs_buf_do_callbacks(bp, lip);
+			xfs_buf_do_callbacks(bp);
 			XFS_BUF_SET_FSPRIVATE(bp, NULL);
 			XFS_BUF_CLR_IODONE_FUNC(bp);
 			xfs_buf_ioend(bp, 0);
@@ -1029,7 +1039,7 @@ xfs_buf_iodone_callbacks(
 		return;
 	}
 
-	xfs_buf_do_callbacks(bp, lip);
+	xfs_buf_do_callbacks(bp);
 	XFS_BUF_SET_FSPRIVATE(bp, NULL);
 	XFS_BUF_CLR_IODONE_FUNC(bp);
 	xfs_buf_ioend(bp, 0);
@@ -1063,7 +1073,7 @@ xfs_buf_error_relse(
 	 * We have to unpin the pinned buffers so do the
 	 * callbacks.
 	 */
-	xfs_buf_do_callbacks(bp, lip);
+	xfs_buf_do_callbacks(bp);
 	XFS_BUF_SET_FSPRIVATE(bp, NULL);
 	XFS_BUF_CLR_IODONE_FUNC(bp);
 	XFS_BUF_SET_BRELSE_FUNC(bp,NULL);

fs/xfs/xfs_buf_item.h

@@ -105,17 +105,6 @@ typedef struct xfs_buf_log_item {
 	xfs_buf_log_format_t	bli_format;	/* in-log header */
 } xfs_buf_log_item_t;
 
-/*
- * This structure is used during recovery to record the buf log
- * items which have been canceled and should not be replayed.
- */
-typedef struct xfs_buf_cancel {
-	xfs_daddr_t		bc_blkno;
-	uint			bc_len;
-	int			bc_refcount;
-	struct xfs_buf_cancel	*bc_next;
-} xfs_buf_cancel_t;
-
 void	xfs_buf_item_init(struct xfs_buf *, struct xfs_mount *);
 void	xfs_buf_item_relse(struct xfs_buf *);
 void	xfs_buf_item_log(xfs_buf_log_item_t *, uint, uint);

fs/xfs/xfs_extfree_item.c

@@ -47,6 +47,28 @@ xfs_efi_item_free(
 		kmem_zone_free(xfs_efi_zone, efip);
 }
 
+/*
+ * Freeing the efi requires that we remove it from the AIL if it has already
+ * been placed there. However, the EFI may not yet have been placed in the AIL
+ * when called by xfs_efi_release() from EFD processing due to the ordering of
+ * committed vs unpin operations in bulk insert operations. Hence the
+ * test_and_clear_bit(XFS_EFI_COMMITTED) to ensure only the last caller frees
+ * the EFI.
+ */
+STATIC void
+__xfs_efi_release(
+	struct xfs_efi_log_item	*efip)
+{
+	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
+
+	if (!test_and_clear_bit(XFS_EFI_COMMITTED, &efip->efi_flags)) {
+		spin_lock(&ailp->xa_lock);
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		xfs_trans_ail_delete(ailp, &efip->efi_item);
+		xfs_efi_item_free(efip);
+	}
+}
+
 /*
  * This returns the number of iovecs needed to log the given efi item.
  * We only need 1 iovec for an efi item.  It just logs the efi_log_format
@@ -74,7 +96,8 @@ xfs_efi_item_format(
 	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
 	uint			size;
 
-	ASSERT(efip->efi_next_extent == efip->efi_format.efi_nextents);
+	ASSERT(atomic_read(&efip->efi_next_extent) ==
+				efip->efi_format.efi_nextents);
 
 	efip->efi_format.efi_type = XFS_LI_EFI;
 
@@ -99,10 +122,12 @@ xfs_efi_item_pin(
 }
 
 /*
- * While EFIs cannot really be pinned, the unpin operation is the
- * last place at which the EFI is manipulated during a transaction.
- * Here we coordinate with xfs_efi_cancel() to determine who gets to
- * free the EFI.
+ * While EFIs cannot really be pinned, the unpin operation is the last place at
+ * which the EFI is manipulated during a transaction.  If we are being asked to
+ * remove the EFI it's because the transaction has been cancelled and by
+ * definition that means the EFI cannot be in the AIL so remove it from the
+ * transaction and free it.  Otherwise coordinate with xfs_efi_release() (via
+ * XFS_EFI_COMMITTED) to determine who gets to free the EFI.
  */
 STATIC void
 xfs_efi_item_unpin(
@@ -110,20 +135,14 @@ xfs_efi_item_unpin(
 	int			remove)
 {
 	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
-	struct xfs_ail		*ailp = lip->li_ailp;
-
-	spin_lock(&ailp->xa_lock);
-	if (efip->efi_flags & XFS_EFI_CANCELED) {
-		if (remove)
-			xfs_trans_del_item(lip);
 
-		/* xfs_trans_ail_delete() drops the AIL lock. */
-		xfs_trans_ail_delete(ailp, lip);
+	if (remove) {
+		ASSERT(!(lip->li_flags & XFS_LI_IN_AIL));
+		xfs_trans_del_item(lip);
 		xfs_efi_item_free(efip);
-	} else {
-		efip->efi_flags |= XFS_EFI_COMMITTED;
-		spin_unlock(&ailp->xa_lock);
+		return;
 	}
+	__xfs_efi_release(efip);
 }
 
 /*
@@ -152,16 +171,20 @@ xfs_efi_item_unlock(
 }
 
 /*
- * The EFI is logged only once and cannot be moved in the log, so
- * simply return the lsn at which it's been logged.  The canceled
- * flag is not paid any attention here.  Checking for that is delayed
- * until the EFI is unpinned.
+ * The EFI is logged only once and cannot be moved in the log, so simply return
+ * the lsn at which it's been logged.  For bulk transaction committed
+ * processing, the EFI may be processed but not yet unpinned prior to the EFD
+ * being processed. Set the XFS_EFI_COMMITTED flag so this case can be detected
+ * when processing the EFD.
  */
 STATIC xfs_lsn_t
 xfs_efi_item_committed(
 	struct xfs_log_item	*lip,
 	xfs_lsn_t		lsn)
 {
+	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
+
+	set_bit(XFS_EFI_COMMITTED, &efip->efi_flags);
 	return lsn;
 }
 
@@ -230,6 +253,7 @@ xfs_efi_init(
 	xfs_log_item_init(mp, &efip->efi_item, XFS_LI_EFI, &xfs_efi_item_ops);
 	efip->efi_format.efi_nextents = nextents;
 	efip->efi_format.efi_id = (__psint_t)(void*)efip;
+	atomic_set(&efip->efi_next_extent, 0);
 
 	return efip;
 }
@@ -289,37 +313,18 @@ xfs_efi_copy_format(xfs_log_iovec_t *buf, xfs_efi_log_format_t *dst_efi_fmt)
 }
 
 /*
- * This is called by the efd item code below to release references to
- * the given efi item.  Each efd calls this with the number of
- * extents that it has logged, and when the sum of these reaches
- * the total number of extents logged by this efi item we can free
- * the efi item.
- *
- * Freeing the efi item requires that we remove it from the AIL.
- * We'll use the AIL lock to protect our counters as well as
- * the removal from the AIL.
+ * This is called by the efd item code below to release references to the given
+ * efi item.  Each efd calls this with the number of extents that it has
+ * logged, and when the sum of these reaches the total number of extents logged
+ * by this efi item we can free the efi item.
  */
 void
 xfs_efi_release(xfs_efi_log_item_t	*efip,
 		uint			nextents)
 {
-	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
-	int			extents_left;
-
-	ASSERT(efip->efi_next_extent > 0);
-	ASSERT(efip->efi_flags & XFS_EFI_COMMITTED);
-
-	spin_lock(&ailp->xa_lock);
-	ASSERT(efip->efi_next_extent >= nextents);
-	efip->efi_next_extent -= nextents;
-	extents_left = efip->efi_next_extent;
-	if (extents_left == 0) {
-		/* xfs_trans_ail_delete() drops the AIL lock. */
-		xfs_trans_ail_delete(ailp, (xfs_log_item_t *)efip);
-		xfs_efi_item_free(efip);
-	} else {
-		spin_unlock(&ailp->xa_lock);
-	}
+	ASSERT(atomic_read(&efip->efi_next_extent) >= nextents);
+	if (atomic_sub_and_test(nextents, &efip->efi_next_extent))
+		__xfs_efi_release(efip);
 }
 
 static inline struct xfs_efd_log_item *EFD_ITEM(struct xfs_log_item *lip)

fs/xfs/xfs_extfree_item.h

@@ -111,11 +111,10 @@ typedef struct xfs_efd_log_format_64 {
 #define	XFS_EFI_MAX_FAST_EXTENTS	16
 
 /*
- * Define EFI flags.
+ * Define EFI flag bits. Manipulated by set/clear/test_bit operators.
  */
-#define	XFS_EFI_RECOVERED	0x1
-#define	XFS_EFI_COMMITTED	0x2
-#define	XFS_EFI_CANCELED	0x4
+#define	XFS_EFI_RECOVERED	1
+#define	XFS_EFI_COMMITTED	2
 
 /*
  * This is the "extent free intention" log item.  It is used
@@ -125,8 +124,8 @@ typedef struct xfs_efd_log_format_64 {
  */
 typedef struct xfs_efi_log_item {
 	xfs_log_item_t		efi_item;
-	uint			efi_flags;	/* misc flags */
-	uint			efi_next_extent;
+	atomic_t		efi_next_extent;
+	unsigned long		efi_flags;	/* misc flags */
 	xfs_efi_log_format_t	efi_format;
 } xfs_efi_log_item_t;
 

fs/xfs/xfs_fsops.c

@@ -374,6 +374,7 @@ xfs_growfs_data_private(
 		mp->m_maxicount = icount << mp->m_sb.sb_inopblog;
 	} else
 		mp->m_maxicount = 0;
+	xfs_set_low_space_thresholds(mp);
 
 	/* update secondary superblocks. */
 	for (agno = 1; agno < nagcount; agno++) {

fs/xfs/xfs_iget.c

@@ -42,6 +42,17 @@
 #include "xfs_trace.h"
 
 
+/*
+ * Define xfs inode iolock lockdep classes. We need to ensure that all active
+ * inodes are considered the same for lockdep purposes, including inodes that
+ * are recycled through the XFS_IRECLAIMABLE state. This is the the only way to
+ * guarantee the locks are considered the same when there are multiple lock
+ * initialisation siteѕ. Also, define a reclaimable inode class so it is
+ * obvious in lockdep reports which class the report is against.
+ */
+static struct lock_class_key xfs_iolock_active;
+struct lock_class_key xfs_iolock_reclaimable;
+
 /*
  * Allocate and initialise an xfs_inode.
  */
@@ -69,8 +80,11 @@ xfs_inode_alloc(
 	ASSERT(atomic_read(&ip->i_pincount) == 0);
 	ASSERT(!spin_is_locked(&ip->i_flags_lock));
 	ASSERT(completion_done(&ip->i_flush));
+	ASSERT(ip->i_ino == 0);
 
 	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+	lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+			&xfs_iolock_active, "xfs_iolock_active");
 
 	/* initialise the xfs inode */
 	ip->i_ino = ino;
@@ -85,9 +99,6 @@ xfs_inode_alloc(
 	ip->i_size = 0;
 	ip->i_new_size = 0;
 
-	/* prevent anyone from using this yet */
-	VFS_I(ip)->i_state = I_NEW;
-
 	return ip;
 }
 
@@ -145,7 +156,18 @@ xfs_inode_free(
 	ASSERT(!spin_is_locked(&ip->i_flags_lock));
 	ASSERT(completion_done(&ip->i_flush));
 
-	call_rcu(&ip->i_vnode.i_rcu, xfs_inode_free_callback);
+	/*
+	 * Because we use RCU freeing we need to ensure the inode always
+	 * appears to be reclaimed with an invalid inode number when in the
+	 * free state. The ip->i_flags_lock provides the barrier against lookup
+	 * races.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	ip->i_flags = XFS_IRECLAIM;
+	ip->i_ino = 0;
+	spin_unlock(&ip->i_flags_lock);
+
+	call_rcu(&VFS_I(ip)->i_rcu, xfs_inode_free_callback);
 }
 
 /*
@@ -155,14 +177,29 @@ static int
 xfs_iget_cache_hit(
 	struct xfs_perag	*pag,
 	struct xfs_inode	*ip,
+	xfs_ino_t		ino,
 	int			flags,
-	int			lock_flags) __releases(pag->pag_ici_lock)
+	int			lock_flags) __releases(RCU)
 {
 	struct inode		*inode = VFS_I(ip);
 	struct xfs_mount	*mp = ip->i_mount;
 	int			error;
 
+	/*
+	 * check for re-use of an inode within an RCU grace period due to the
+	 * radix tree nodes not being updated yet. We monitor for this by
+	 * setting the inode number to zero before freeing the inode structure.
+	 * If the inode has been reallocated and set up, then the inode number
+	 * will not match, so check for that, too.
+	 */
 	spin_lock(&ip->i_flags_lock);
+	if (ip->i_ino != ino) {
+		trace_xfs_iget_skip(ip);
+		XFS_STATS_INC(xs_ig_frecycle);
+		error = EAGAIN;
+		goto out_error;
+	}
+
 
 	/*
 	 * If we are racing with another cache hit that is currently
@@ -205,7 +242,7 @@ xfs_iget_cache_hit(
 		ip->i_flags |= XFS_IRECLAIM;
 
 		spin_unlock(&ip->i_flags_lock);
-		read_unlock(&pag->pag_ici_lock);
+		rcu_read_unlock();
 
 		error = -inode_init_always(mp->m_super, inode);
 		if (error) {
@@ -213,7 +250,7 @@ xfs_iget_cache_hit(
 			 * Re-initializing the inode failed, and we are in deep
 			 * trouble.  Try to re-add it to the reclaim list.
 			 */
-			read_lock(&pag->pag_ici_lock);
+			rcu_read_lock();
 			spin_lock(&ip->i_flags_lock);
 
 			ip->i_flags &= ~XFS_INEW;
@@ -223,14 +260,20 @@ xfs_iget_cache_hit(
 			goto out_error;
 		}
 
-		write_lock(&pag->pag_ici_lock);
+		spin_lock(&pag->pag_ici_lock);
 		spin_lock(&ip->i_flags_lock);
 		ip->i_flags &= ~(XFS_IRECLAIMABLE | XFS_IRECLAIM);
 		ip->i_flags |= XFS_INEW;
 		__xfs_inode_clear_reclaim_tag(mp, pag, ip);
 		inode->i_state = I_NEW;
+
+		ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));
+		mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+		lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+				&xfs_iolock_active, "xfs_iolock_active");
+
 		spin_unlock(&ip->i_flags_lock);
-		write_unlock(&pag->pag_ici_lock);
+		spin_unlock(&pag->pag_ici_lock);
 	} else {
 		/* If the VFS inode is being torn down, pause and try again. */
 		if (!igrab(inode)) {
@@ -241,7 +284,7 @@ xfs_iget_cache_hit(
 
 		/* We've got a live one. */
 		spin_unlock(&ip->i_flags_lock);
-		read_unlock(&pag->pag_ici_lock);
+		rcu_read_unlock();
 		trace_xfs_iget_hit(ip);
 	}
 
@@ -255,7 +298,7 @@ xfs_iget_cache_hit(
 
 out_error:
 	spin_unlock(&ip->i_flags_lock);
-	read_unlock(&pag->pag_ici_lock);
+	rcu_read_unlock();
 	return error;
 }
 
@@ -308,7 +351,7 @@ xfs_iget_cache_miss(
 			BUG();
 	}
 
-	write_lock(&pag->pag_ici_lock);
+	spin_lock(&pag->pag_ici_lock);
 
 	/* insert the new inode */
 	error = radix_tree_insert(&pag->pag_ici_root, agino, ip);
@@ -323,14 +366,14 @@ xfs_iget_cache_miss(
 	ip->i_udquot = ip->i_gdquot = NULL;
 	xfs_iflags_set(ip, XFS_INEW);
 
-	write_unlock(&pag->pag_ici_lock);
+	spin_unlock(&pag->pag_ici_lock);
 	radix_tree_preload_end();
 
 	*ipp = ip;
 	return 0;
 
 out_preload_end:
-	write_unlock(&pag->pag_ici_lock);
+	spin_unlock(&pag->pag_ici_lock);
 	radix_tree_preload_end();
 	if (lock_flags)
 		xfs_iunlock(ip, lock_flags);
@@ -377,7 +420,7 @@ xfs_iget(
 	xfs_agino_t	agino;
 
 	/* reject inode numbers outside existing AGs */
-	if (XFS_INO_TO_AGNO(mp, ino) >= mp->m_sb.sb_agcount)
+	if (!ino || XFS_INO_TO_AGNO(mp, ino) >= mp->m_sb.sb_agcount)
 		return EINVAL;
 
 	/* get the perag structure and ensure that it's inode capable */
@@ -386,15 +429,15 @@ xfs_iget(
 
 again:
 	error = 0;
-	read_lock(&pag->pag_ici_lock);
+	rcu_read_lock();
 	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
 
 	if (ip) {
-		error = xfs_iget_cache_hit(pag, ip, flags, lock_flags);
+		error = xfs_iget_cache_hit(pag, ip, ino, flags, lock_flags);
 		if (error)
 			goto out_error_or_again;
 	} else {
-		read_unlock(&pag->pag_ici_lock);
+		rcu_read_unlock();
 		XFS_STATS_INC(xs_ig_missed);
 
 		error = xfs_iget_cache_miss(mp, pag, tp, ino, &ip,

fs/xfs/xfs_inode.c

@@ -887,7 +887,7 @@ xfs_iread(
 	 * around for a while.  This helps to keep recently accessed
 	 * meta-data in-core longer.
 	 */
-	XFS_BUF_SET_REF(bp, XFS_INO_REF);
+	xfs_buf_set_ref(bp, XFS_INO_REF);
 
 	/*
 	 * Use xfs_trans_brelse() to release the buffer containing the
@@ -2000,16 +2000,32 @@ xfs_ifree_cluster(
 		 */
 		for (i = 0; i < ninodes; i++) {
 retry:
-			read_lock(&pag->pag_ici_lock);
+			rcu_read_lock();
 			ip = radix_tree_lookup(&pag->pag_ici_root,
 					XFS_INO_TO_AGINO(mp, (inum + i)));
 
-			/* Inode not in memory or stale, nothing to do */
-			if (!ip || xfs_iflags_test(ip, XFS_ISTALE)) {
-				read_unlock(&pag->pag_ici_lock);
+			/* Inode not in memory, nothing to do */
+			if (!ip) {
+				rcu_read_unlock();
 				continue;
 			}
 
+			/*
+			 * because this is an RCU protected lookup, we could
+			 * find a recently freed or even reallocated inode
+			 * during the lookup. We need to check under the
+			 * i_flags_lock for a valid inode here. Skip it if it
+			 * is not valid, the wrong inode or stale.
+			 */
+			spin_lock(&ip->i_flags_lock);
+			if (ip->i_ino != inum + i ||
+			    __xfs_iflags_test(ip, XFS_ISTALE)) {
+				spin_unlock(&ip->i_flags_lock);
+				rcu_read_unlock();
+				continue;
+			}
+			spin_unlock(&ip->i_flags_lock);
+
 			/*
 			 * Don't try to lock/unlock the current inode, but we
 			 * _cannot_ skip the other inodes that we did not find
@@ -2019,11 +2035,11 @@ retry:
 			 */
 			if (ip != free_ip &&
 			    !xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
-				read_unlock(&pag->pag_ici_lock);
+				rcu_read_unlock();
 				delay(1);
 				goto retry;
 			}
-			read_unlock(&pag->pag_ici_lock);
+			rcu_read_unlock();
 
 			xfs_iflock(ip);
 			xfs_iflags_set(ip, XFS_ISTALE);
@@ -2629,7 +2645,7 @@ xfs_iflush_cluster(
 
 	mask = ~(((XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog)) - 1);
 	first_index = XFS_INO_TO_AGINO(mp, ip->i_ino) & mask;
-	read_lock(&pag->pag_ici_lock);
+	rcu_read_lock();
 	/* really need a gang lookup range call here */
 	nr_found = radix_tree_gang_lookup(&pag->pag_ici_root, (void**)ilist,
 					first_index, inodes_per_cluster);
@@ -2640,9 +2656,21 @@ xfs_iflush_cluster(
 		iq = ilist[i];
 		if (iq == ip)
 			continue;
-		/* if the inode lies outside this cluster, we're done. */
-		if ((XFS_INO_TO_AGINO(mp, iq->i_ino) & mask) != first_index)
-			break;
+
+		/*
+		 * because this is an RCU protected lookup, we could find a
+		 * recently freed or even reallocated inode during the lookup.
+		 * We need to check under the i_flags_lock for a valid inode
+		 * here. Skip it if it is not valid or the wrong inode.
+		 */
+		spin_lock(&ip->i_flags_lock);
+		if (!ip->i_ino ||
+		    (XFS_INO_TO_AGINO(mp, iq->i_ino) & mask) != first_index) {
+			spin_unlock(&ip->i_flags_lock);
+			continue;
+		}
+		spin_unlock(&ip->i_flags_lock);
+
 		/*
 		 * Do an un-protected check to see if the inode is dirty and
 		 * is a candidate for flushing.  These checks will be repeated
@@ -2692,7 +2720,7 @@ xfs_iflush_cluster(
 	}
 
 out_free:
-	read_unlock(&pag->pag_ici_lock);
+	rcu_read_unlock();
 	kmem_free(ilist);
 out_put:
 	xfs_perag_put(pag);
@@ -2704,7 +2732,7 @@ cluster_corrupt_out:
 	 * Corruption detected in the clustering loop.  Invalidate the
 	 * inode buffer and shut down the filesystem.
 	 */
-	read_unlock(&pag->pag_ici_lock);
+	rcu_read_unlock();
 	/*
 	 * Clean up the buffer.  If it was B_DELWRI, just release it --
 	 * brelse can handle it with no problems.  If not, shut down the

fs/xfs/xfs_inode.h

@@ -376,12 +376,13 @@ static inline void xfs_ifunlock(xfs_inode_t *ip)
 /*
  * In-core inode flags.
  */
-#define XFS_IRECLAIM    0x0001  /* we have started reclaiming this inode    */
-#define XFS_ISTALE	0x0002	/* inode has been staled */
-#define XFS_IRECLAIMABLE 0x0004 /* inode can be reclaimed */
-#define XFS_INEW	0x0008	/* inode has just been allocated */
-#define XFS_IFILESTREAM	0x0010	/* inode is in a filestream directory */
-#define XFS_ITRUNCATED	0x0020	/* truncated down so flush-on-close */
+#define XFS_IRECLAIM		0x0001  /* started reclaiming this inode */
+#define XFS_ISTALE		0x0002	/* inode has been staled */
+#define XFS_IRECLAIMABLE	0x0004	/* inode can be reclaimed */
+#define XFS_INEW		0x0008	/* inode has just been allocated */
+#define XFS_IFILESTREAM		0x0010	/* inode is in a filestream directory */
+#define XFS_ITRUNCATED		0x0020	/* truncated down so flush-on-close */
+#define XFS_IDIRTY_RELEASE	0x0040	/* dirty release already seen */
 
 /*
  * Flags for inode locking.
@@ -438,6 +439,8 @@ static inline void xfs_ifunlock(xfs_inode_t *ip)
 #define XFS_IOLOCK_DEP(flags)	(((flags) & XFS_IOLOCK_DEP_MASK) >> XFS_IOLOCK_SHIFT)
 #define XFS_ILOCK_DEP(flags)	(((flags) & XFS_ILOCK_DEP_MASK) >> XFS_ILOCK_SHIFT)
 
+extern struct lock_class_key xfs_iolock_reclaimable;
+
 /*
  * Flags for xfs_itruncate_start().
  */

fs/xfs/xfs_inode_item.c

@@ -842,15 +842,64 @@ xfs_inode_item_destroy(
  * flushed to disk.  It is responsible for removing the inode item
  * from the AIL if it has not been re-logged, and unlocking the inode's
  * flush lock.
+ *
+ * To reduce AIL lock traffic as much as possible, we scan the buffer log item
+ * list for other inodes that will run this function. We remove them from the
+ * buffer list so we can process all the inode IO completions in one AIL lock
+ * traversal.
  */
 void
 xfs_iflush_done(
 	struct xfs_buf		*bp,
 	struct xfs_log_item	*lip)
 {
-	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
-	xfs_inode_t		*ip = iip->ili_inode;
+	struct xfs_inode_log_item *iip;
+	struct xfs_log_item	*blip;
+	struct xfs_log_item	*next;
+	struct xfs_log_item	*prev;
 	struct xfs_ail		*ailp = lip->li_ailp;
+	int			need_ail = 0;
+
+	/*
+	 * Scan the buffer IO completions for other inodes being completed and
+	 * attach them to the current inode log item.
+	 */
+	blip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+	prev = NULL;
+	while (blip != NULL) {
+		if (lip->li_cb != xfs_iflush_done) {
+			prev = blip;
+			blip = blip->li_bio_list;
+			continue;
+		}
+
+		/* remove from list */
+		next = blip->li_bio_list;
+		if (!prev) {
+			XFS_BUF_SET_FSPRIVATE(bp, next);
+		} else {
+			prev->li_bio_list = next;
+		}
+
+		/* add to current list */
+		blip->li_bio_list = lip->li_bio_list;
+		lip->li_bio_list = blip;
+
+		/*
+		 * while we have the item, do the unlocked check for needing
+		 * the AIL lock.
+		 */
+		iip = INODE_ITEM(blip);
+		if (iip->ili_logged && blip->li_lsn == iip->ili_flush_lsn)
+			need_ail++;
+
+		blip = next;
+	}
+
+	/* make sure we capture the state of the initial inode. */
+	iip = INODE_ITEM(lip);
+	if (iip->ili_logged && lip->li_lsn == iip->ili_flush_lsn)
+		need_ail++;
 
 	/*
 	 * We only want to pull the item from the AIL if it is
@@ -861,28 +910,37 @@ xfs_iflush_done(
 	 * the lock since it's cheaper, and then we recheck while
 	 * holding the lock before removing the inode from the AIL.
 	 */
-	if (iip->ili_logged && lip->li_lsn == iip->ili_flush_lsn) {
+	if (need_ail) {
+		struct xfs_log_item *log_items[need_ail];
+		int i = 0;
 		spin_lock(&ailp->xa_lock);
-		if (lip->li_lsn == iip->ili_flush_lsn) {
-			/* xfs_trans_ail_delete() drops the AIL lock. */
-			xfs_trans_ail_delete(ailp, lip);
-		} else {
-			spin_unlock(&ailp->xa_lock);
+		for (blip = lip; blip; blip = blip->li_bio_list) {
+			iip = INODE_ITEM(blip);
+			if (iip->ili_logged &&
+			    blip->li_lsn == iip->ili_flush_lsn) {
+				log_items[i++] = blip;
+			}
+			ASSERT(i <= need_ail);
 		}
+		/* xfs_trans_ail_delete_bulk() drops the AIL lock. */
+		xfs_trans_ail_delete_bulk(ailp, log_items, i);
 	}
 
-	iip->ili_logged = 0;
 
 	/*
-	 * Clear the ili_last_fields bits now that we know that the
-	 * data corresponding to them is safely on disk.
+	 * clean up and unlock the flush lock now we are done. We can clear the
+	 * ili_last_fields bits now that we know that the data corresponding to
+	 * them is safely on disk.
 	 */
-	iip->ili_last_fields = 0;
+	for (blip = lip; blip; blip = next) {
+		next = blip->li_bio_list;
+		blip->li_bio_list = NULL;
 
-	/*
-	 * Release the inode's flush lock since we're done with it.
-	 */
-	xfs_ifunlock(ip);
+		iip = INODE_ITEM(blip);
+		iip->ili_logged = 0;
+		iip->ili_last_fields = 0;
+		xfs_ifunlock(iip->ili_inode);
+	}
 }
 
 /*

fs/xfs/xfs_iomap.c

@@ -47,127 +47,8 @@
 
 #define XFS_WRITEIO_ALIGN(mp,off)	(((off) >> mp->m_writeio_log) \
 						<< mp->m_writeio_log)
-#define XFS_STRAT_WRITE_IMAPS	2
 #define XFS_WRITE_IMAPS		XFS_BMAP_MAX_NMAP
 
-STATIC int xfs_iomap_write_direct(struct xfs_inode *, xfs_off_t, size_t,
-				  int, struct xfs_bmbt_irec *, int *);
-STATIC int xfs_iomap_write_delay(struct xfs_inode *, xfs_off_t, size_t, int,
-				 struct xfs_bmbt_irec *, int *);
-STATIC int xfs_iomap_write_allocate(struct xfs_inode *, xfs_off_t, size_t,
-				struct xfs_bmbt_irec *, int *);
-
-int
-xfs_iomap(
-	struct xfs_inode	*ip,
-	xfs_off_t		offset,
-	ssize_t			count,
-	int			flags,
-	struct xfs_bmbt_irec	*imap,
-	int			*nimaps,
-	int			*new)
-{
-	struct xfs_mount	*mp = ip->i_mount;
-	xfs_fileoff_t		offset_fsb, end_fsb;
-	int			error = 0;
-	int			lockmode = 0;
-	int			bmapi_flags = 0;
-
-	ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
-
-	*new = 0;
-
-	if (XFS_FORCED_SHUTDOWN(mp))
-		return XFS_ERROR(EIO);
-
-	trace_xfs_iomap_enter(ip, offset, count, flags, NULL);
-
-	switch (flags & (BMAPI_READ | BMAPI_WRITE | BMAPI_ALLOCATE)) {
-	case BMAPI_READ:
-		lockmode = xfs_ilock_map_shared(ip);
-		bmapi_flags = XFS_BMAPI_ENTIRE;
-		break;
-	case BMAPI_WRITE:
-		lockmode = XFS_ILOCK_EXCL;
-		if (flags & BMAPI_IGNSTATE)
-			bmapi_flags |= XFS_BMAPI_IGSTATE|XFS_BMAPI_ENTIRE;
-		xfs_ilock(ip, lockmode);
-		break;
-	case BMAPI_ALLOCATE:
-		lockmode = XFS_ILOCK_SHARED;
-		bmapi_flags = XFS_BMAPI_ENTIRE;
-
-		/* Attempt non-blocking lock */
-		if (flags & BMAPI_TRYLOCK) {
-			if (!xfs_ilock_nowait(ip, lockmode))
-				return XFS_ERROR(EAGAIN);
-		} else {
-			xfs_ilock(ip, lockmode);
-		}
-		break;
-	default:
-		BUG();
-	}
-
-	ASSERT(offset <= mp->m_maxioffset);
-	if ((xfs_fsize_t)offset + count > mp->m_maxioffset)
-		count = mp->m_maxioffset - offset;
-	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
-	offset_fsb = XFS_B_TO_FSBT(mp, offset);
-
-	error = xfs_bmapi(NULL, ip, offset_fsb,
-			(xfs_filblks_t)(end_fsb - offset_fsb),
-			bmapi_flags,  NULL, 0, imap,
-			nimaps, NULL);
-
-	if (error)
-		goto out;
-
-	switch (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)) {
-	case BMAPI_WRITE:
-		/* If we found an extent, return it */
-		if (*nimaps &&
-		    (imap->br_startblock != HOLESTARTBLOCK) &&
-		    (imap->br_startblock != DELAYSTARTBLOCK)) {
-			trace_xfs_iomap_found(ip, offset, count, flags, imap);
-			break;
-		}
-
-		if (flags & BMAPI_DIRECT) {
-			error = xfs_iomap_write_direct(ip, offset, count, flags,
-						       imap, nimaps);
-		} else {
-			error = xfs_iomap_write_delay(ip, offset, count, flags,
-						      imap, nimaps);
-		}
-		if (!error) {
-			trace_xfs_iomap_alloc(ip, offset, count, flags, imap);
-		}
-		*new = 1;
-		break;
-	case BMAPI_ALLOCATE:
-		/* If we found an extent, return it */
-		xfs_iunlock(ip, lockmode);
-		lockmode = 0;
-
-		if (*nimaps && !isnullstartblock(imap->br_startblock)) {
-			trace_xfs_iomap_found(ip, offset, count, flags, imap);
-			break;
-		}
-
-		error = xfs_iomap_write_allocate(ip, offset, count,
-						 imap, nimaps);
-		break;
-	}
-
-	ASSERT(*nimaps <= 1);
-
-out:
-	if (lockmode)
-		xfs_iunlock(ip, lockmode);
-	return XFS_ERROR(error);
-}
-
 STATIC int
 xfs_iomap_eof_align_last_fsb(
 	xfs_mount_t	*mp,
@@ -236,14 +117,13 @@ xfs_cmn_err_fsblock_zero(
 	return EFSCORRUPTED;
 }
 
-STATIC int
+int
 xfs_iomap_write_direct(
 	xfs_inode_t	*ip,
 	xfs_off_t	offset,
 	size_t		count,
-	int		flags,
 	xfs_bmbt_irec_t *imap,
-	int		*nmaps)
+	int		nmaps)
 {
 	xfs_mount_t	*mp = ip->i_mount;
 	xfs_fileoff_t	offset_fsb;
@@ -279,7 +159,7 @@ xfs_iomap_write_direct(
 		if (error)
 			goto error_out;
 	} else {
-		if (*nmaps && (imap->br_startblock == HOLESTARTBLOCK))
+		if (nmaps && (imap->br_startblock == HOLESTARTBLOCK))
 			last_fsb = MIN(last_fsb, (xfs_fileoff_t)
 					imap->br_blockcount +
 					imap->br_startoff);
@@ -331,7 +211,7 @@ xfs_iomap_write_direct(
 	xfs_trans_ijoin(tp, ip);
 
 	bmapi_flag = XFS_BMAPI_WRITE;
-	if ((flags & BMAPI_DIRECT) && (offset < ip->i_size || extsz))
+	if (offset < ip->i_size || extsz)
 		bmapi_flag |= XFS_BMAPI_PREALLOC;
 
 	/*
@@ -370,7 +250,6 @@ xfs_iomap_write_direct(
 		goto error_out;
 	}
 
-	*nmaps = 1;
 	return 0;
 
 error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
@@ -379,7 +258,6 @@ error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
 
 error1:	/* Just cancel transaction */
 	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
-	*nmaps = 0;	/* nothing set-up here */
 
 error_out:
 	return XFS_ERROR(error);
@@ -389,6 +267,9 @@ error_out:
  * If the caller is doing a write at the end of the file, then extend the
  * allocation out to the file system's write iosize.  We clean up any extra
  * space left over when the file is closed in xfs_inactive().
+ *
+ * If we find we already have delalloc preallocation beyond EOF, don't do more
+ * preallocation as it it not needed.
  */
 STATIC int
 xfs_iomap_eof_want_preallocate(
@@ -396,7 +277,6 @@ xfs_iomap_eof_want_preallocate(
 	xfs_inode_t	*ip,
 	xfs_off_t	offset,
 	size_t		count,
-	int		ioflag,
 	xfs_bmbt_irec_t *imap,
 	int		nimaps,
 	int		*prealloc)
@@ -405,6 +285,7 @@ xfs_iomap_eof_want_preallocate(
 	xfs_filblks_t   count_fsb;
 	xfs_fsblock_t	firstblock;
 	int		n, error, imaps;
+	int		found_delalloc = 0;
 
 	*prealloc = 0;
 	if ((offset + count) <= ip->i_size)
@@ -429,20 +310,66 @@ xfs_iomap_eof_want_preallocate(
 				return 0;
 			start_fsb += imap[n].br_blockcount;
 			count_fsb -= imap[n].br_blockcount;
+
+			if (imap[n].br_startblock == DELAYSTARTBLOCK)
+				found_delalloc = 1;
 		}
 	}
-	*prealloc = 1;
+	if (!found_delalloc)
+		*prealloc = 1;
 	return 0;
 }
 
-STATIC int
+/*
+ * If we don't have a user specified preallocation size, dynamically increase
+ * the preallocation size as the size of the file grows. Cap the maximum size
+ * at a single extent or less if the filesystem is near full. The closer the
+ * filesystem is to full, the smaller the maximum prealocation.
+ */
+STATIC xfs_fsblock_t
+xfs_iomap_prealloc_size(
+	struct xfs_mount	*mp,
+	struct xfs_inode	*ip)
+{
+	xfs_fsblock_t		alloc_blocks = 0;
+
+	if (!(mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)) {
+		int shift = 0;
+		int64_t freesp;
+
+		alloc_blocks = XFS_B_TO_FSB(mp, ip->i_size);
+		alloc_blocks = XFS_FILEOFF_MIN(MAXEXTLEN,
+					rounddown_pow_of_two(alloc_blocks));
+
+		xfs_icsb_sync_counters(mp, XFS_ICSB_LAZY_COUNT);
+		freesp = mp->m_sb.sb_fdblocks;
+		if (freesp < mp->m_low_space[XFS_LOWSP_5_PCNT]) {
+			shift = 2;
+			if (freesp < mp->m_low_space[XFS_LOWSP_4_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_3_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_2_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_1_PCNT])
+				shift++;
+		}
+		if (shift)
+			alloc_blocks >>= shift;
+	}
+
+	if (alloc_blocks < mp->m_writeio_blocks)
+		alloc_blocks = mp->m_writeio_blocks;
+
+	return alloc_blocks;
+}
+
+int
 xfs_iomap_write_delay(
 	xfs_inode_t	*ip,
 	xfs_off_t	offset,
 	size_t		count,
-	int		ioflag,
-	xfs_bmbt_irec_t *ret_imap,
-	int		*nmaps)
+	xfs_bmbt_irec_t *ret_imap)
 {
 	xfs_mount_t	*mp = ip->i_mount;
 	xfs_fileoff_t	offset_fsb;
@@ -469,16 +396,19 @@ xfs_iomap_write_delay(
 	extsz = xfs_get_extsz_hint(ip);
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 
+
 	error = xfs_iomap_eof_want_preallocate(mp, ip, offset, count,
-				ioflag, imap, XFS_WRITE_IMAPS, &prealloc);
+				imap, XFS_WRITE_IMAPS, &prealloc);
 	if (error)
 		return error;
 
 retry:
 	if (prealloc) {
+		xfs_fsblock_t	alloc_blocks = xfs_iomap_prealloc_size(mp, ip);
+
 		aligned_offset = XFS_WRITEIO_ALIGN(mp, (offset + count - 1));
 		ioalign = XFS_B_TO_FSBT(mp, aligned_offset);
-		last_fsb = ioalign + mp->m_writeio_blocks;
+		last_fsb = ioalign + alloc_blocks;
 	} else {
 		last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
 	}
@@ -496,22 +426,31 @@ retry:
 			  XFS_BMAPI_DELAY | XFS_BMAPI_WRITE |
 			  XFS_BMAPI_ENTIRE, &firstblock, 1, imap,
 			  &nimaps, NULL);
-	if (error && (error != ENOSPC))
+	switch (error) {
+	case 0:
+	case ENOSPC:
+	case EDQUOT:
+		break;
+	default:
 		return XFS_ERROR(error);
+	}
 
 	/*
-	 * If bmapi returned us nothing, and if we didn't get back EDQUOT,
-	 * then we must have run out of space - flush all other inodes with
-	 * delalloc blocks and retry without EOF preallocation.
+	 * If bmapi returned us nothing, we got either ENOSPC or EDQUOT.  For
+	 * ENOSPC, * flush all other inodes with delalloc blocks to free up
+	 * some of the excess reserved metadata space. For both cases, retry
+	 * without EOF preallocation.
 	 */
 	if (nimaps == 0) {
 		trace_xfs_delalloc_enospc(ip, offset, count);
 		if (flushed)
-			return XFS_ERROR(ENOSPC);
+			return XFS_ERROR(error ? error : ENOSPC);
 
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		xfs_flush_inodes(ip);
-		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		if (error == ENOSPC) {
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			xfs_flush_inodes(ip);
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
+		}
 
 		flushed = 1;
 		error = 0;
@@ -523,8 +462,6 @@ retry:
 		return xfs_cmn_err_fsblock_zero(ip, &imap[0]);
 
 	*ret_imap = imap[0];
-	*nmaps = 1;
-
 	return 0;
 }
 
@@ -538,13 +475,12 @@ retry:
  * We no longer bother to look at the incoming map - all we have to
  * guarantee is that whatever we allocate fills the required range.
  */
-STATIC int
+int
 xfs_iomap_write_allocate(
 	xfs_inode_t	*ip,
 	xfs_off_t	offset,
 	size_t		count,
-	xfs_bmbt_irec_t *imap,
-	int		*retmap)
+	xfs_bmbt_irec_t *imap)
 {
 	xfs_mount_t	*mp = ip->i_mount;
 	xfs_fileoff_t	offset_fsb, last_block;
@@ -557,8 +493,6 @@ xfs_iomap_write_allocate(
 	int		error = 0;
 	int		nres;
 
-	*retmap = 0;
-
 	/*
 	 * Make sure that the dquots are there.
 	 */
@@ -680,7 +614,6 @@ xfs_iomap_write_allocate(
 		if ((offset_fsb >= imap->br_startoff) &&
 		    (offset_fsb < (imap->br_startoff +
 				   imap->br_blockcount))) {
-			*retmap = 1;
 			XFS_STATS_INC(xs_xstrat_quick);
 			return 0;
 		}

fs/xfs/xfs_iomap.h

@@ -18,30 +18,15 @@
 #ifndef __XFS_IOMAP_H__
 #define __XFS_IOMAP_H__
 
-/* base extent manipulation calls */
-#define BMAPI_READ	(1 << 0)	/* read extents */
-#define BMAPI_WRITE	(1 << 1)	/* create extents */
-#define BMAPI_ALLOCATE	(1 << 2)	/* delayed allocate to real extents */
-
-/* modifiers */
-#define BMAPI_IGNSTATE	(1 << 4)	/* ignore unwritten state on read */
-#define BMAPI_DIRECT	(1 << 5)	/* direct instead of buffered write */
-#define BMAPI_MMA	(1 << 6)	/* allocate for mmap write */
-#define BMAPI_TRYLOCK	(1 << 7)	/* non-blocking request */
-
-#define BMAPI_FLAGS \
-	{ BMAPI_READ,		"READ" }, \
-	{ BMAPI_WRITE,		"WRITE" }, \
-	{ BMAPI_ALLOCATE,	"ALLOCATE" }, \
-	{ BMAPI_IGNSTATE,	"IGNSTATE" }, \
-	{ BMAPI_DIRECT,		"DIRECT" }, \
-	{ BMAPI_TRYLOCK,	"TRYLOCK" }
-
 struct xfs_inode;
 struct xfs_bmbt_irec;
 
-extern int xfs_iomap(struct xfs_inode *, xfs_off_t, ssize_t, int,
-		     struct xfs_bmbt_irec *, int *, int *);
+extern int xfs_iomap_write_direct(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *, int);
+extern int xfs_iomap_write_delay(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *);
+extern int xfs_iomap_write_allocate(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *);
 extern int xfs_iomap_write_unwritten(struct xfs_inode *, xfs_off_t, size_t);
 
 #endif /* __XFS_IOMAP_H__*/

fs/xfs/xfs_log.c

@@ -47,7 +47,7 @@ STATIC xlog_t *  xlog_alloc_log(xfs_mount_t	*mp,
 				xfs_buftarg_t	*log_target,
 				xfs_daddr_t	blk_offset,
 				int		num_bblks);
-STATIC int	 xlog_space_left(xlog_t *log, int cycle, int bytes);
+STATIC int	 xlog_space_left(struct log *log, atomic64_t *head);
 STATIC int	 xlog_sync(xlog_t *log, xlog_in_core_t *iclog);
 STATIC void	 xlog_dealloc_log(xlog_t *log);
 
@@ -70,7 +70,7 @@ STATIC void xlog_state_want_sync(xlog_t	*log, xlog_in_core_t *iclog);
 /* local functions to manipulate grant head */
 STATIC int  xlog_grant_log_space(xlog_t		*log,
 				 xlog_ticket_t	*xtic);
-STATIC void xlog_grant_push_ail(xfs_mount_t	*mp,
+STATIC void xlog_grant_push_ail(struct log	*log,
 				int		need_bytes);
 STATIC void xlog_regrant_reserve_log_space(xlog_t	 *log,
 					   xlog_ticket_t *ticket);
@@ -81,98 +81,73 @@ STATIC void xlog_ungrant_log_space(xlog_t	 *log,
 
 #if defined(DEBUG)
 STATIC void	xlog_verify_dest_ptr(xlog_t *log, char *ptr);
-STATIC void	xlog_verify_grant_head(xlog_t *log, int equals);
+STATIC void	xlog_verify_grant_tail(struct log *log);
 STATIC void	xlog_verify_iclog(xlog_t *log, xlog_in_core_t *iclog,
 				  int count, boolean_t syncing);
 STATIC void	xlog_verify_tail_lsn(xlog_t *log, xlog_in_core_t *iclog,
 				     xfs_lsn_t tail_lsn);
 #else
 #define xlog_verify_dest_ptr(a,b)
-#define xlog_verify_grant_head(a,b)
+#define xlog_verify_grant_tail(a)
 #define xlog_verify_iclog(a,b,c,d)
 #define xlog_verify_tail_lsn(a,b,c)
 #endif
 
 STATIC int	xlog_iclogs_empty(xlog_t *log);
 
-
 static void
-xlog_ins_ticketq(struct xlog_ticket **qp, struct xlog_ticket *tic)
+xlog_grant_sub_space(
+	struct log	*log,
+	atomic64_t	*head,
+	int		bytes)
 {
-	if (*qp) {
-		tic->t_next	    = (*qp);
-		tic->t_prev	    = (*qp)->t_prev;
-		(*qp)->t_prev->t_next = tic;
-		(*qp)->t_prev	    = tic;
-	} else {
-		tic->t_prev = tic->t_next = tic;
-		*qp = tic;
-	}
+	int64_t	head_val = atomic64_read(head);
+	int64_t new, old;
 
-	tic->t_flags |= XLOG_TIC_IN_Q;
-}
+	do {
+		int	cycle, space;
 
-static void
-xlog_del_ticketq(struct xlog_ticket **qp, struct xlog_ticket *tic)
-{
-	if (tic == tic->t_next) {
-		*qp = NULL;
-	} else {
-		*qp = tic->t_next;
-		tic->t_next->t_prev = tic->t_prev;
-		tic->t_prev->t_next = tic->t_next;
-	}
+		xlog_crack_grant_head_val(head_val, &cycle, &space);
 
-	tic->t_next = tic->t_prev = NULL;
-	tic->t_flags &= ~XLOG_TIC_IN_Q;
+		space -= bytes;
+		if (space < 0) {
+			space += log->l_logsize;
+			cycle--;
+		}
+
+		old = head_val;
+		new = xlog_assign_grant_head_val(cycle, space);
+		head_val = atomic64_cmpxchg(head, old, new);
+	} while (head_val != old);
 }
 
 static void
-xlog_grant_sub_space(struct log *log, int bytes)
+xlog_grant_add_space(
+	struct log	*log,
+	atomic64_t	*head,
+	int		bytes)
 {
-	log->l_grant_write_bytes -= bytes;
-	if (log->l_grant_write_bytes < 0) {
-		log->l_grant_write_bytes += log->l_logsize;
-		log->l_grant_write_cycle--;
-	}
-
-	log->l_grant_reserve_bytes -= bytes;
-	if ((log)->l_grant_reserve_bytes < 0) {
-		log->l_grant_reserve_bytes += log->l_logsize;
-		log->l_grant_reserve_cycle--;
-	}
+	int64_t	head_val = atomic64_read(head);
+	int64_t new, old;
 
-}
+	do {
+		int		tmp;
+		int		cycle, space;
 
-static void
-xlog_grant_add_space_write(struct log *log, int bytes)
-{
-	int tmp = log->l_logsize - log->l_grant_write_bytes;
-	if (tmp > bytes)
-		log->l_grant_write_bytes += bytes;
-	else {
-		log->l_grant_write_cycle++;
-		log->l_grant_write_bytes = bytes - tmp;
-	}
-}
+		xlog_crack_grant_head_val(head_val, &cycle, &space);
 
-static void
-xlog_grant_add_space_reserve(struct log *log, int bytes)
-{
-	int tmp = log->l_logsize - log->l_grant_reserve_bytes;
-	if (tmp > bytes)
-		log->l_grant_reserve_bytes += bytes;
-	else {
-		log->l_grant_reserve_cycle++;
-		log->l_grant_reserve_bytes = bytes - tmp;
-	}
-}
+		tmp = log->l_logsize - space;
+		if (tmp > bytes)
+			space += bytes;
+		else {
+			space = bytes - tmp;
+			cycle++;
+		}
 
-static inline void
-xlog_grant_add_space(struct log *log, int bytes)
-{
-	xlog_grant_add_space_write(log, bytes);
-	xlog_grant_add_space_reserve(log, bytes);
+		old = head_val;
+		new = xlog_assign_grant_head_val(cycle, space);
+		head_val = atomic64_cmpxchg(head, old, new);
+	} while (head_val != old);
 }
 
 static void
@@ -355,7 +330,7 @@ xfs_log_reserve(
 
 		trace_xfs_log_reserve(log, internal_ticket);
 
-		xlog_grant_push_ail(mp, internal_ticket->t_unit_res);
+		xlog_grant_push_ail(log, internal_ticket->t_unit_res);
 		retval = xlog_regrant_write_log_space(log, internal_ticket);
 	} else {
 		/* may sleep if need to allocate more tickets */
@@ -369,7 +344,7 @@ xfs_log_reserve(
 
 		trace_xfs_log_reserve(log, internal_ticket);
 
-		xlog_grant_push_ail(mp,
+		xlog_grant_push_ail(log,
 				    (internal_ticket->t_unit_res *
 				     internal_ticket->t_cnt));
 		retval = xlog_grant_log_space(log, internal_ticket);
@@ -584,8 +559,8 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		if (!(iclog->ic_state == XLOG_STATE_ACTIVE ||
 		      iclog->ic_state == XLOG_STATE_DIRTY)) {
 			if (!XLOG_FORCED_SHUTDOWN(log)) {
-				sv_wait(&iclog->ic_force_wait, PMEM,
-					&log->l_icloglock, s);
+				xlog_wait(&iclog->ic_force_wait,
+							&log->l_icloglock);
 			} else {
 				spin_unlock(&log->l_icloglock);
 			}
@@ -625,8 +600,8 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 			|| iclog->ic_state == XLOG_STATE_DIRTY
 			|| iclog->ic_state == XLOG_STATE_IOERROR) ) {
 
-				sv_wait(&iclog->ic_force_wait, PMEM,
-					&log->l_icloglock, s);
+				xlog_wait(&iclog->ic_force_wait,
+							&log->l_icloglock);
 		} else {
 			spin_unlock(&log->l_icloglock);
 		}
@@ -703,55 +678,46 @@ xfs_log_move_tail(xfs_mount_t	*mp,
 {
 	xlog_ticket_t	*tic;
 	xlog_t		*log = mp->m_log;
-	int		need_bytes, free_bytes, cycle, bytes;
+	int		need_bytes, free_bytes;
 
 	if (XLOG_FORCED_SHUTDOWN(log))
 		return;
 
-	if (tail_lsn == 0) {
-		/* needed since sync_lsn is 64 bits */
-		spin_lock(&log->l_icloglock);
-		tail_lsn = log->l_last_sync_lsn;
-		spin_unlock(&log->l_icloglock);
-	}
-
-	spin_lock(&log->l_grant_lock);
+	if (tail_lsn == 0)
+		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
 
-	/* Also an invalid lsn.  1 implies that we aren't passing in a valid
-	 * tail_lsn.
-	 */
-	if (tail_lsn != 1) {
-		log->l_tail_lsn = tail_lsn;
-	}
+	/* tail_lsn == 1 implies that we weren't passed a valid value.  */
+	if (tail_lsn != 1)
+		atomic64_set(&log->l_tail_lsn, tail_lsn);
 
-	if ((tic = log->l_write_headq)) {
+	if (!list_empty_careful(&log->l_writeq)) {
 #ifdef DEBUG
 		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
 			panic("Recovery problem");
 #endif
-		cycle = log->l_grant_write_cycle;
-		bytes = log->l_grant_write_bytes;
-		free_bytes = xlog_space_left(log, cycle, bytes);
-		do {
+		spin_lock(&log->l_grant_write_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+		list_for_each_entry(tic, &log->l_writeq, t_queue) {
 			ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
 
 			if (free_bytes < tic->t_unit_res && tail_lsn != 1)
 				break;
 			tail_lsn = 0;
 			free_bytes -= tic->t_unit_res;
-			sv_signal(&tic->t_wait);
-			tic = tic->t_next;
-		} while (tic != log->l_write_headq);
+			trace_xfs_log_regrant_write_wake_up(log, tic);
+			wake_up(&tic->t_wait);
+		}
+		spin_unlock(&log->l_grant_write_lock);
 	}
-	if ((tic = log->l_reserve_headq)) {
+
+	if (!list_empty_careful(&log->l_reserveq)) {
 #ifdef DEBUG
 		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
 			panic("Recovery problem");
 #endif
-		cycle = log->l_grant_reserve_cycle;
-		bytes = log->l_grant_reserve_bytes;
-		free_bytes = xlog_space_left(log, cycle, bytes);
-		do {
+		spin_lock(&log->l_grant_reserve_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+		list_for_each_entry(tic, &log->l_reserveq, t_queue) {
 			if (tic->t_flags & XLOG_TIC_PERM_RESERV)
 				need_bytes = tic->t_unit_res*tic->t_cnt;
 			else
@@ -760,12 +726,12 @@ xfs_log_move_tail(xfs_mount_t	*mp,
 				break;
 			tail_lsn = 0;
 			free_bytes -= need_bytes;
-			sv_signal(&tic->t_wait);
-			tic = tic->t_next;
-		} while (tic != log->l_reserve_headq);
+			trace_xfs_log_grant_wake_up(log, tic);
+			wake_up(&tic->t_wait);
+		}
+		spin_unlock(&log->l_grant_reserve_lock);
 	}
-	spin_unlock(&log->l_grant_lock);
-}	/* xfs_log_move_tail */
+}
 
 /*
  * Determine if we have a transaction that has gone to disk
@@ -831,23 +797,19 @@ xfs_log_need_covered(xfs_mount_t *mp)
  * We may be holding the log iclog lock upon entering this routine.
  */
 xfs_lsn_t
-xlog_assign_tail_lsn(xfs_mount_t *mp)
+xlog_assign_tail_lsn(
+	struct xfs_mount	*mp)
 {
-	xfs_lsn_t tail_lsn;
-	xlog_t	  *log = mp->m_log;
+	xfs_lsn_t		tail_lsn;
+	struct log		*log = mp->m_log;
 
 	tail_lsn = xfs_trans_ail_tail(mp->m_ail);
-	spin_lock(&log->l_grant_lock);
-	if (tail_lsn != 0) {
-		log->l_tail_lsn = tail_lsn;
-	} else {
-		tail_lsn = log->l_tail_lsn = log->l_last_sync_lsn;
-	}
-	spin_unlock(&log->l_grant_lock);
+	if (!tail_lsn)
+		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
 
+	atomic64_set(&log->l_tail_lsn, tail_lsn);
 	return tail_lsn;
-}	/* xlog_assign_tail_lsn */
-
+}
 
 /*
  * Return the space in the log between the tail and the head.  The head
@@ -864,21 +826,26 @@ xlog_assign_tail_lsn(xfs_mount_t *mp)
  * result is that we return the size of the log as the amount of space left.
  */
 STATIC int
-xlog_space_left(xlog_t *log, int cycle, int bytes)
-{
-	int free_bytes;
-	int tail_bytes;
-	int tail_cycle;
-
-	tail_bytes = BBTOB(BLOCK_LSN(log->l_tail_lsn));
-	tail_cycle = CYCLE_LSN(log->l_tail_lsn);
-	if ((tail_cycle == cycle) && (bytes >= tail_bytes)) {
-		free_bytes = log->l_logsize - (bytes - tail_bytes);
-	} else if ((tail_cycle + 1) < cycle) {
+xlog_space_left(
+	struct log	*log,
+	atomic64_t	*head)
+{
+	int		free_bytes;
+	int		tail_bytes;
+	int		tail_cycle;
+	int		head_cycle;
+	int		head_bytes;
+
+	xlog_crack_grant_head(head, &head_cycle, &head_bytes);
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &tail_cycle, &tail_bytes);
+	tail_bytes = BBTOB(tail_bytes);
+	if (tail_cycle == head_cycle && head_bytes >= tail_bytes)
+		free_bytes = log->l_logsize - (head_bytes - tail_bytes);
+	else if (tail_cycle + 1 < head_cycle)
 		return 0;
-	} else if (tail_cycle < cycle) {
-		ASSERT(tail_cycle == (cycle - 1));
-		free_bytes = tail_bytes - bytes;
+	else if (tail_cycle < head_cycle) {
+		ASSERT(tail_cycle == (head_cycle - 1));
+		free_bytes = tail_bytes - head_bytes;
 	} else {
 		/*
 		 * The reservation head is behind the tail.
@@ -889,12 +856,12 @@ xlog_space_left(xlog_t *log, int cycle, int bytes)
 			"xlog_space_left: head behind tail\n"
 			"  tail_cycle = %d, tail_bytes = %d\n"
 			"  GH   cycle = %d, GH   bytes = %d",
-			tail_cycle, tail_bytes, cycle, bytes);
+			tail_cycle, tail_bytes, head_cycle, head_bytes);
 		ASSERT(0);
 		free_bytes = log->l_logsize;
 	}
 	return free_bytes;
-}	/* xlog_space_left */
+}
 
 
 /*
@@ -1047,12 +1014,16 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	log->l_flags	   |= XLOG_ACTIVE_RECOVERY;
 
 	log->l_prev_block  = -1;
-	log->l_tail_lsn	   = xlog_assign_lsn(1, 0);
 	/* log->l_tail_lsn = 0x100000000LL; cycle = 1; current block = 0 */
-	log->l_last_sync_lsn = log->l_tail_lsn;
+	xlog_assign_atomic_lsn(&log->l_tail_lsn, 1, 0);
+	xlog_assign_atomic_lsn(&log->l_last_sync_lsn, 1, 0);
 	log->l_curr_cycle  = 1;	    /* 0 is bad since this is initial value */
-	log->l_grant_reserve_cycle = 1;
-	log->l_grant_write_cycle = 1;
+	xlog_assign_grant_head(&log->l_grant_reserve_head, 1, 0);
+	xlog_assign_grant_head(&log->l_grant_write_head, 1, 0);
+	INIT_LIST_HEAD(&log->l_reserveq);
+	INIT_LIST_HEAD(&log->l_writeq);
+	spin_lock_init(&log->l_grant_reserve_lock);
+	spin_lock_init(&log->l_grant_write_lock);
 
 	error = EFSCORRUPTED;
 	if (xfs_sb_version_hassector(&mp->m_sb)) {
@@ -1094,8 +1065,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 	log->l_xbuf = bp;
 
 	spin_lock_init(&log->l_icloglock);
-	spin_lock_init(&log->l_grant_lock);
-	sv_init(&log->l_flush_wait, 0, "flush_wait");
+	init_waitqueue_head(&log->l_flush_wait);
 
 	/* log record size must be multiple of BBSIZE; see xlog_rec_header_t */
 	ASSERT((XFS_BUF_SIZE(bp) & BBMASK) == 0);
@@ -1151,8 +1121,8 @@ xlog_alloc_log(xfs_mount_t	*mp,
 
 		ASSERT(XFS_BUF_ISBUSY(iclog->ic_bp));
 		ASSERT(XFS_BUF_VALUSEMA(iclog->ic_bp) <= 0);
-		sv_init(&iclog->ic_force_wait, SV_DEFAULT, "iclog-force");
-		sv_init(&iclog->ic_write_wait, SV_DEFAULT, "iclog-write");
+		init_waitqueue_head(&iclog->ic_force_wait);
+		init_waitqueue_head(&iclog->ic_write_wait);
 
 		iclogp = &iclog->ic_next;
 	}
@@ -1167,15 +1137,11 @@ xlog_alloc_log(xfs_mount_t	*mp,
 out_free_iclog:
 	for (iclog = log->l_iclog; iclog; iclog = prev_iclog) {
 		prev_iclog = iclog->ic_next;
-		if (iclog->ic_bp) {
-			sv_destroy(&iclog->ic_force_wait);
-			sv_destroy(&iclog->ic_write_wait);
+		if (iclog->ic_bp)
 			xfs_buf_free(iclog->ic_bp);
-		}
 		kmem_free(iclog);
 	}
 	spinlock_destroy(&log->l_icloglock);
-	spinlock_destroy(&log->l_grant_lock);
 	xfs_buf_free(log->l_xbuf);
 out_free_log:
 	kmem_free(log);
@@ -1223,61 +1189,60 @@ xlog_commit_record(
  * water mark.  In this manner, we would be creating a low water mark.
  */
 STATIC void
-xlog_grant_push_ail(xfs_mount_t	*mp,
-		    int		need_bytes)
+xlog_grant_push_ail(
+	struct log	*log,
+	int		need_bytes)
 {
-    xlog_t	*log = mp->m_log;	/* pointer to the log */
-    xfs_lsn_t	tail_lsn;		/* lsn of the log tail */
-    xfs_lsn_t	threshold_lsn = 0;	/* lsn we'd like to be at */
-    int		free_blocks;		/* free blocks left to write to */
-    int		free_bytes;		/* free bytes left to write to */
-    int		threshold_block;	/* block in lsn we'd like to be at */
-    int		threshold_cycle;	/* lsn cycle we'd like to be at */
-    int		free_threshold;
-
-    ASSERT(BTOBB(need_bytes) < log->l_logBBsize);
-
-    spin_lock(&log->l_grant_lock);
-    free_bytes = xlog_space_left(log,
-				 log->l_grant_reserve_cycle,
-				 log->l_grant_reserve_bytes);
-    tail_lsn = log->l_tail_lsn;
-    free_blocks = BTOBBT(free_bytes);
-
-    /*
-     * Set the threshold for the minimum number of free blocks in the
-     * log to the maximum of what the caller needs, one quarter of the
-     * log, and 256 blocks.
-     */
-    free_threshold = BTOBB(need_bytes);
-    free_threshold = MAX(free_threshold, (log->l_logBBsize >> 2));
-    free_threshold = MAX(free_threshold, 256);
-    if (free_blocks < free_threshold) {
-	threshold_block = BLOCK_LSN(tail_lsn) + free_threshold;
-	threshold_cycle = CYCLE_LSN(tail_lsn);
+	xfs_lsn_t	threshold_lsn = 0;
+	xfs_lsn_t	last_sync_lsn;
+	int		free_blocks;
+	int		free_bytes;
+	int		threshold_block;
+	int		threshold_cycle;
+	int		free_threshold;
+
+	ASSERT(BTOBB(need_bytes) < log->l_logBBsize);
+
+	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+	free_blocks = BTOBBT(free_bytes);
+
+	/*
+	 * Set the threshold for the minimum number of free blocks in the
+	 * log to the maximum of what the caller needs, one quarter of the
+	 * log, and 256 blocks.
+	 */
+	free_threshold = BTOBB(need_bytes);
+	free_threshold = MAX(free_threshold, (log->l_logBBsize >> 2));
+	free_threshold = MAX(free_threshold, 256);
+	if (free_blocks >= free_threshold)
+		return;
+
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &threshold_cycle,
+						&threshold_block);
+	threshold_block += free_threshold;
 	if (threshold_block >= log->l_logBBsize) {
-	    threshold_block -= log->l_logBBsize;
-	    threshold_cycle += 1;
+		threshold_block -= log->l_logBBsize;
+		threshold_cycle += 1;
 	}
-	threshold_lsn = xlog_assign_lsn(threshold_cycle, threshold_block);
+	threshold_lsn = xlog_assign_lsn(threshold_cycle,
+					threshold_block);
+	/*
+	 * Don't pass in an lsn greater than the lsn of the last
+	 * log record known to be on disk. Use a snapshot of the last sync lsn
+	 * so that it doesn't change between the compare and the set.
+	 */
+	last_sync_lsn = atomic64_read(&log->l_last_sync_lsn);
+	if (XFS_LSN_CMP(threshold_lsn, last_sync_lsn) > 0)
+		threshold_lsn = last_sync_lsn;
 
-	/* Don't pass in an lsn greater than the lsn of the last
-	 * log record known to be on disk.
+	/*
+	 * Get the transaction layer to kick the dirty buffers out to
+	 * disk asynchronously. No point in trying to do this if
+	 * the filesystem is shutting down.
 	 */
-	if (XFS_LSN_CMP(threshold_lsn, log->l_last_sync_lsn) > 0)
-	    threshold_lsn = log->l_last_sync_lsn;
-    }
-    spin_unlock(&log->l_grant_lock);
-
-    /*
-     * Get the transaction layer to kick the dirty buffers out to
-     * disk asynchronously. No point in trying to do this if
-     * the filesystem is shutting down.
-     */
-    if (threshold_lsn &&
-	!XLOG_FORCED_SHUTDOWN(log))
-	    xfs_trans_ail_push(log->l_ailp, threshold_lsn);
-}	/* xlog_grant_push_ail */
+	if (!XLOG_FORCED_SHUTDOWN(log))
+		xfs_trans_ail_push(log->l_ailp, threshold_lsn);
+}
 
 /*
  * The bdstrat callback function for log bufs. This gives us a central
@@ -1372,9 +1337,8 @@ xlog_sync(xlog_t		*log,
 		 roundoff < BBTOB(1)));
 
 	/* move grant heads by roundoff in sync */
-	spin_lock(&log->l_grant_lock);
-	xlog_grant_add_space(log, roundoff);
-	spin_unlock(&log->l_grant_lock);
+	xlog_grant_add_space(log, &log->l_grant_reserve_head, roundoff);
+	xlog_grant_add_space(log, &log->l_grant_write_head, roundoff);
 
 	/* put cycle number in every block */
 	xlog_pack_data(log, iclog, roundoff); 
@@ -1489,15 +1453,12 @@ xlog_dealloc_log(xlog_t *log)
 
 	iclog = log->l_iclog;
 	for (i=0; i<log->l_iclog_bufs; i++) {
-		sv_destroy(&iclog->ic_force_wait);
-		sv_destroy(&iclog->ic_write_wait);
 		xfs_buf_free(iclog->ic_bp);
 		next_iclog = iclog->ic_next;
 		kmem_free(iclog);
 		iclog = next_iclog;
 	}
 	spinlock_destroy(&log->l_icloglock);
-	spinlock_destroy(&log->l_grant_lock);
 
 	xfs_buf_free(log->l_xbuf);
 	log->l_mp->m_log = NULL;
@@ -2232,7 +2193,7 @@ xlog_state_do_callback(
 				lowest_lsn = xlog_get_lowest_lsn(log);
 				if (lowest_lsn &&
 				    XFS_LSN_CMP(lowest_lsn,
-				    		be64_to_cpu(iclog->ic_header.h_lsn)) < 0) {
+						be64_to_cpu(iclog->ic_header.h_lsn)) < 0) {
 					iclog = iclog->ic_next;
 					continue; /* Leave this iclog for
 						   * another thread */
@@ -2240,23 +2201,21 @@ xlog_state_do_callback(
 
 				iclog->ic_state = XLOG_STATE_CALLBACK;
 
-				spin_unlock(&log->l_icloglock);
 
-				/* l_last_sync_lsn field protected by
-				 * l_grant_lock. Don't worry about iclog's lsn.
-				 * No one else can be here except us.
+				/*
+				 * update the last_sync_lsn before we drop the
+				 * icloglock to ensure we are the only one that
+				 * can update it.
 				 */
-				spin_lock(&log->l_grant_lock);
-				ASSERT(XFS_LSN_CMP(log->l_last_sync_lsn,
-				       be64_to_cpu(iclog->ic_header.h_lsn)) <= 0);
-				log->l_last_sync_lsn =
-					be64_to_cpu(iclog->ic_header.h_lsn);
-				spin_unlock(&log->l_grant_lock);
+				ASSERT(XFS_LSN_CMP(atomic64_read(&log->l_last_sync_lsn),
+					be64_to_cpu(iclog->ic_header.h_lsn)) <= 0);
+				atomic64_set(&log->l_last_sync_lsn,
+					be64_to_cpu(iclog->ic_header.h_lsn));
 
-			} else {
-				spin_unlock(&log->l_icloglock);
+			} else
 				ioerrors++;
-			}
+
+			spin_unlock(&log->l_icloglock);
 
 			/*
 			 * Keep processing entries in the callback list until
@@ -2297,7 +2256,7 @@ xlog_state_do_callback(
 			xlog_state_clean_log(log);
 
 			/* wake up threads waiting in xfs_log_force() */
-			sv_broadcast(&iclog->ic_force_wait);
+			wake_up_all(&iclog->ic_force_wait);
 
 			iclog = iclog->ic_next;
 		} while (first_iclog != iclog);
@@ -2344,7 +2303,7 @@ xlog_state_do_callback(
 	spin_unlock(&log->l_icloglock);
 
 	if (wake)
-		sv_broadcast(&log->l_flush_wait);
+		wake_up_all(&log->l_flush_wait);
 }
 
 
@@ -2395,7 +2354,7 @@ xlog_state_done_syncing(
 	 * iclog buffer, we wake them all, one will get to do the
 	 * I/O, the others get to wait for the result.
 	 */
-	sv_broadcast(&iclog->ic_write_wait);
+	wake_up_all(&iclog->ic_write_wait);
 	spin_unlock(&log->l_icloglock);
 	xlog_state_do_callback(log, aborted, iclog);	/* also cleans log */
 }	/* xlog_state_done_syncing */
@@ -2444,7 +2403,7 @@ restart:
 		XFS_STATS_INC(xs_log_noiclogs);
 
 		/* Wait for log writes to have flushed */
-		sv_wait(&log->l_flush_wait, 0, &log->l_icloglock, 0);
+		xlog_wait(&log->l_flush_wait, &log->l_icloglock);
 		goto restart;
 	}
 
@@ -2527,6 +2486,18 @@ restart:
  *
  * Once a ticket gets put onto the reserveq, it will only return after
  * the needed reservation is satisfied.
+ *
+ * This function is structured so that it has a lock free fast path. This is
+ * necessary because every new transaction reservation will come through this
+ * path. Hence any lock will be globally hot if we take it unconditionally on
+ * every pass.
+ *
+ * As tickets are only ever moved on and off the reserveq under the
+ * l_grant_reserve_lock, we only need to take that lock if we are going
+ * to add the ticket to the queue and sleep. We can avoid taking the lock if the
+ * ticket was never added to the reserveq because the t_queue list head will be
+ * empty and we hold the only reference to it so it can safely be checked
+ * unlocked.
  */
 STATIC int
 xlog_grant_log_space(xlog_t	   *log,
@@ -2534,24 +2505,27 @@ xlog_grant_log_space(xlog_t	   *log,
 {
 	int		 free_bytes;
 	int		 need_bytes;
-#ifdef DEBUG
-	xfs_lsn_t	 tail_lsn;
-#endif
-
 
 #ifdef DEBUG
 	if (log->l_flags & XLOG_ACTIVE_RECOVERY)
 		panic("grant Recovery problem");
 #endif
 
-	/* Is there space or do we need to sleep? */
-	spin_lock(&log->l_grant_lock);
-
 	trace_xfs_log_grant_enter(log, tic);
 
+	need_bytes = tic->t_unit_res;
+	if (tic->t_flags & XFS_LOG_PERM_RESERV)
+		need_bytes *= tic->t_ocnt;
+
 	/* something is already sleeping; insert new transaction at end */
-	if (log->l_reserve_headq) {
-		xlog_ins_ticketq(&log->l_reserve_headq, tic);
+	if (!list_empty_careful(&log->l_reserveq)) {
+		spin_lock(&log->l_grant_reserve_lock);
+		/* recheck the queue now we are locked */
+		if (list_empty(&log->l_reserveq)) {
+			spin_unlock(&log->l_grant_reserve_lock);
+			goto redo;
+		}
+		list_add_tail(&tic->t_queue, &log->l_reserveq);
 
 		trace_xfs_log_grant_sleep1(log, tic);
 
@@ -2563,72 +2537,57 @@ xlog_grant_log_space(xlog_t	   *log,
 			goto error_return;
 
 		XFS_STATS_INC(xs_sleep_logspace);
-		sv_wait(&tic->t_wait, PINOD|PLTWAIT, &log->l_grant_lock, s);
+		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
+
 		/*
 		 * If we got an error, and the filesystem is shutting down,
 		 * we'll catch it down below. So just continue...
 		 */
 		trace_xfs_log_grant_wake1(log, tic);
-		spin_lock(&log->l_grant_lock);
 	}
-	if (tic->t_flags & XFS_LOG_PERM_RESERV)
-		need_bytes = tic->t_unit_res*tic->t_ocnt;
-	else
-		need_bytes = tic->t_unit_res;
 
 redo:
 	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return;
+		goto error_return_unlocked;
 
-	free_bytes = xlog_space_left(log, log->l_grant_reserve_cycle,
-				     log->l_grant_reserve_bytes);
+	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
 	if (free_bytes < need_bytes) {
-		if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-			xlog_ins_ticketq(&log->l_reserve_headq, tic);
+		spin_lock(&log->l_grant_reserve_lock);
+		if (list_empty(&tic->t_queue))
+			list_add_tail(&tic->t_queue, &log->l_reserveq);
 
 		trace_xfs_log_grant_sleep2(log, tic);
 
-		spin_unlock(&log->l_grant_lock);
-		xlog_grant_push_ail(log->l_mp, need_bytes);
-		spin_lock(&log->l_grant_lock);
-
-		XFS_STATS_INC(xs_sleep_logspace);
-		sv_wait(&tic->t_wait, PINOD|PLTWAIT, &log->l_grant_lock, s);
-
-		spin_lock(&log->l_grant_lock);
 		if (XLOG_FORCED_SHUTDOWN(log))
 			goto error_return;
 
-		trace_xfs_log_grant_wake2(log, tic);
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
 
+		trace_xfs_log_grant_wake2(log, tic);
 		goto redo;
-	} else if (tic->t_flags & XLOG_TIC_IN_Q)
-		xlog_del_ticketq(&log->l_reserve_headq, tic);
+	}
 
-	/* we've got enough space */
-	xlog_grant_add_space(log, need_bytes);
-#ifdef DEBUG
-	tail_lsn = log->l_tail_lsn;
-	/*
-	 * Check to make sure the grant write head didn't just over lap the
-	 * tail.  If the cycles are the same, we can't be overlapping.
-	 * Otherwise, make sure that the cycles differ by exactly one and
-	 * check the byte count.
-	 */
-	if (CYCLE_LSN(tail_lsn) != log->l_grant_write_cycle) {
-		ASSERT(log->l_grant_write_cycle-1 == CYCLE_LSN(tail_lsn));
-		ASSERT(log->l_grant_write_bytes <= BBTOB(BLOCK_LSN(tail_lsn)));
+	if (!list_empty(&tic->t_queue)) {
+		spin_lock(&log->l_grant_reserve_lock);
+		list_del_init(&tic->t_queue);
+		spin_unlock(&log->l_grant_reserve_lock);
 	}
-#endif
+
+	/* we've got enough space */
+	xlog_grant_add_space(log, &log->l_grant_reserve_head, need_bytes);
+	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
 	trace_xfs_log_grant_exit(log, tic);
-	xlog_verify_grant_head(log, 1);
-	spin_unlock(&log->l_grant_lock);
+	xlog_verify_grant_tail(log);
 	return 0;
 
- error_return:
-	if (tic->t_flags & XLOG_TIC_IN_Q)
-		xlog_del_ticketq(&log->l_reserve_headq, tic);
-
+error_return_unlocked:
+	spin_lock(&log->l_grant_reserve_lock);
+error_return:
+	list_del_init(&tic->t_queue);
+	spin_unlock(&log->l_grant_reserve_lock);
 	trace_xfs_log_grant_error(log, tic);
 
 	/*
@@ -2638,7 +2597,6 @@ redo:
 	 */
 	tic->t_curr_res = 0;
 	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
-	spin_unlock(&log->l_grant_lock);
 	return XFS_ERROR(EIO);
 }	/* xlog_grant_log_space */
 
@@ -2646,17 +2604,14 @@ redo:
 /*
  * Replenish the byte reservation required by moving the grant write head.
  *
- *
+ * Similar to xlog_grant_log_space, the function is structured to have a lock
+ * free fast path.
  */
 STATIC int
 xlog_regrant_write_log_space(xlog_t	   *log,
 			     xlog_ticket_t *tic)
 {
 	int		free_bytes, need_bytes;
-	xlog_ticket_t	*ntic;
-#ifdef DEBUG
-	xfs_lsn_t	tail_lsn;
-#endif
 
 	tic->t_curr_res = tic->t_unit_res;
 	xlog_tic_reset_res(tic);
@@ -2669,12 +2624,9 @@ xlog_regrant_write_log_space(xlog_t	   *log,
 		panic("regrant Recovery problem");
 #endif
 
-	spin_lock(&log->l_grant_lock);
-
 	trace_xfs_log_regrant_write_enter(log, tic);
-
 	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return;
+		goto error_return_unlocked;
 
 	/* If there are other waiters on the queue then give them a
 	 * chance at logspace before us. Wake up the first waiters,
@@ -2683,92 +2635,76 @@ xlog_regrant_write_log_space(xlog_t	   *log,
 	 * this transaction.
 	 */
 	need_bytes = tic->t_unit_res;
-	if ((ntic = log->l_write_headq)) {
-		free_bytes = xlog_space_left(log, log->l_grant_write_cycle,
-					     log->l_grant_write_bytes);
-		do {
+	if (!list_empty_careful(&log->l_writeq)) {
+		struct xlog_ticket *ntic;
+
+		spin_lock(&log->l_grant_write_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+		list_for_each_entry(ntic, &log->l_writeq, t_queue) {
 			ASSERT(ntic->t_flags & XLOG_TIC_PERM_RESERV);
 
 			if (free_bytes < ntic->t_unit_res)
 				break;
 			free_bytes -= ntic->t_unit_res;
-			sv_signal(&ntic->t_wait);
-			ntic = ntic->t_next;
-		} while (ntic != log->l_write_headq);
-
-		if (ntic != log->l_write_headq) {
-			if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-				xlog_ins_ticketq(&log->l_write_headq, tic);
+			wake_up(&ntic->t_wait);
+		}
 
+		if (ntic != list_first_entry(&log->l_writeq,
+						struct xlog_ticket, t_queue)) {
+			if (list_empty(&tic->t_queue))
+				list_add_tail(&tic->t_queue, &log->l_writeq);
 			trace_xfs_log_regrant_write_sleep1(log, tic);
 
-			spin_unlock(&log->l_grant_lock);
-			xlog_grant_push_ail(log->l_mp, need_bytes);
-			spin_lock(&log->l_grant_lock);
+			xlog_grant_push_ail(log, need_bytes);
 
 			XFS_STATS_INC(xs_sleep_logspace);
-			sv_wait(&tic->t_wait, PINOD|PLTWAIT,
-				&log->l_grant_lock, s);
-
-			/* If we're shutting down, this tic is already
-			 * off the queue */
-			spin_lock(&log->l_grant_lock);
-			if (XLOG_FORCED_SHUTDOWN(log))
-				goto error_return;
-
+			xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
 			trace_xfs_log_regrant_write_wake1(log, tic);
-		}
+		} else
+			spin_unlock(&log->l_grant_write_lock);
 	}
 
 redo:
 	if (XLOG_FORCED_SHUTDOWN(log))
-		goto error_return;
+		goto error_return_unlocked;
 
-	free_bytes = xlog_space_left(log, log->l_grant_write_cycle,
-				     log->l_grant_write_bytes);
+	free_bytes = xlog_space_left(log, &log->l_grant_write_head);
 	if (free_bytes < need_bytes) {
-		if ((tic->t_flags & XLOG_TIC_IN_Q) == 0)
-			xlog_ins_ticketq(&log->l_write_headq, tic);
-		spin_unlock(&log->l_grant_lock);
-		xlog_grant_push_ail(log->l_mp, need_bytes);
-		spin_lock(&log->l_grant_lock);
-
-		XFS_STATS_INC(xs_sleep_logspace);
-		trace_xfs_log_regrant_write_sleep2(log, tic);
-
-		sv_wait(&tic->t_wait, PINOD|PLTWAIT, &log->l_grant_lock, s);
+		spin_lock(&log->l_grant_write_lock);
+		if (list_empty(&tic->t_queue))
+			list_add_tail(&tic->t_queue, &log->l_writeq);
 
-		/* If we're shutting down, this tic is already off the queue */
-		spin_lock(&log->l_grant_lock);
 		if (XLOG_FORCED_SHUTDOWN(log))
 			goto error_return;
 
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		trace_xfs_log_regrant_write_sleep2(log, tic);
+		xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
+
 		trace_xfs_log_regrant_write_wake2(log, tic);
 		goto redo;
-	} else if (tic->t_flags & XLOG_TIC_IN_Q)
-		xlog_del_ticketq(&log->l_write_headq, tic);
+	}
 
-	/* we've got enough space */
-	xlog_grant_add_space_write(log, need_bytes);
-#ifdef DEBUG
-	tail_lsn = log->l_tail_lsn;
-	if (CYCLE_LSN(tail_lsn) != log->l_grant_write_cycle) {
-		ASSERT(log->l_grant_write_cycle-1 == CYCLE_LSN(tail_lsn));
-		ASSERT(log->l_grant_write_bytes <= BBTOB(BLOCK_LSN(tail_lsn)));
+	if (!list_empty(&tic->t_queue)) {
+		spin_lock(&log->l_grant_write_lock);
+		list_del_init(&tic->t_queue);
+		spin_unlock(&log->l_grant_write_lock);
 	}
-#endif
 
+	/* we've got enough space */
+	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
 	trace_xfs_log_regrant_write_exit(log, tic);
-
-	xlog_verify_grant_head(log, 1);
-	spin_unlock(&log->l_grant_lock);
+	xlog_verify_grant_tail(log);
 	return 0;
 
 
+ error_return_unlocked:
+	spin_lock(&log->l_grant_write_lock);
  error_return:
-	if (tic->t_flags & XLOG_TIC_IN_Q)
-		xlog_del_ticketq(&log->l_reserve_headq, tic);
-
+	list_del_init(&tic->t_queue);
+	spin_unlock(&log->l_grant_write_lock);
 	trace_xfs_log_regrant_write_error(log, tic);
 
 	/*
@@ -2778,7 +2714,6 @@ redo:
 	 */
 	tic->t_curr_res = 0;
 	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
-	spin_unlock(&log->l_grant_lock);
 	return XFS_ERROR(EIO);
 }	/* xlog_regrant_write_log_space */
 
@@ -2799,27 +2734,24 @@ xlog_regrant_reserve_log_space(xlog_t	     *log,
 	if (ticket->t_cnt > 0)
 		ticket->t_cnt--;
 
-	spin_lock(&log->l_grant_lock);
-	xlog_grant_sub_space(log, ticket->t_curr_res);
+	xlog_grant_sub_space(log, &log->l_grant_reserve_head,
+					ticket->t_curr_res);
+	xlog_grant_sub_space(log, &log->l_grant_write_head,
+					ticket->t_curr_res);
 	ticket->t_curr_res = ticket->t_unit_res;
 	xlog_tic_reset_res(ticket);
 
 	trace_xfs_log_regrant_reserve_sub(log, ticket);
 
-	xlog_verify_grant_head(log, 1);
-
 	/* just return if we still have some of the pre-reserved space */
-	if (ticket->t_cnt > 0) {
-		spin_unlock(&log->l_grant_lock);
+	if (ticket->t_cnt > 0)
 		return;
-	}
 
-	xlog_grant_add_space_reserve(log, ticket->t_unit_res);
+	xlog_grant_add_space(log, &log->l_grant_reserve_head,
+					ticket->t_unit_res);
 
 	trace_xfs_log_regrant_reserve_exit(log, ticket);
 
-	xlog_verify_grant_head(log, 0);
-	spin_unlock(&log->l_grant_lock);
 	ticket->t_curr_res = ticket->t_unit_res;
 	xlog_tic_reset_res(ticket);
 }	/* xlog_regrant_reserve_log_space */
@@ -2843,28 +2775,29 @@ STATIC void
 xlog_ungrant_log_space(xlog_t	     *log,
 		       xlog_ticket_t *ticket)
 {
+	int	bytes;
+
 	if (ticket->t_cnt > 0)
 		ticket->t_cnt--;
 
-	spin_lock(&log->l_grant_lock);
 	trace_xfs_log_ungrant_enter(log, ticket);
-
-	xlog_grant_sub_space(log, ticket->t_curr_res);
-
 	trace_xfs_log_ungrant_sub(log, ticket);
 
-	/* If this is a permanent reservation ticket, we may be able to free
+	/*
+	 * If this is a permanent reservation ticket, we may be able to free
 	 * up more space based on the remaining count.
 	 */
+	bytes = ticket->t_curr_res;
 	if (ticket->t_cnt > 0) {
 		ASSERT(ticket->t_flags & XLOG_TIC_PERM_RESERV);
-		xlog_grant_sub_space(log, ticket->t_unit_res*ticket->t_cnt);
+		bytes += ticket->t_unit_res*ticket->t_cnt;
 	}
 
+	xlog_grant_sub_space(log, &log->l_grant_reserve_head, bytes);
+	xlog_grant_sub_space(log, &log->l_grant_write_head, bytes);
+
 	trace_xfs_log_ungrant_exit(log, ticket);
 
-	xlog_verify_grant_head(log, 1);
-	spin_unlock(&log->l_grant_lock);
 	xfs_log_move_tail(log->l_mp, 1);
 }	/* xlog_ungrant_log_space */
 
@@ -2901,11 +2834,11 @@ xlog_state_release_iclog(
 
 	if (iclog->ic_state == XLOG_STATE_WANT_SYNC) {
 		/* update tail before writing to iclog */
-		xlog_assign_tail_lsn(log->l_mp);
+		xfs_lsn_t tail_lsn = xlog_assign_tail_lsn(log->l_mp);
 		sync++;
 		iclog->ic_state = XLOG_STATE_SYNCING;
-		iclog->ic_header.h_tail_lsn = cpu_to_be64(log->l_tail_lsn);
-		xlog_verify_tail_lsn(log, iclog, log->l_tail_lsn);
+		iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn);
+		xlog_verify_tail_lsn(log, iclog, tail_lsn);
 		/* cycle incremented when incrementing curr_block */
 	}
 	spin_unlock(&log->l_icloglock);
@@ -3088,7 +3021,7 @@ maybe_sleep:
 			return XFS_ERROR(EIO);
 		}
 		XFS_STATS_INC(xs_log_force_sleep);
-		sv_wait(&iclog->ic_force_wait, PINOD, &log->l_icloglock, s);
+		xlog_wait(&iclog->ic_force_wait, &log->l_icloglock);
 		/*
 		 * No need to grab the log lock here since we're
 		 * only deciding whether or not to return EIO
@@ -3206,8 +3139,8 @@ try_again:
 
 				XFS_STATS_INC(xs_log_force_sleep);
 
-				sv_wait(&iclog->ic_prev->ic_write_wait,
-					PSWP, &log->l_icloglock, s);
+				xlog_wait(&iclog->ic_prev->ic_write_wait,
+							&log->l_icloglock);
 				if (log_flushed)
 					*log_flushed = 1;
 				already_slept = 1;
@@ -3235,7 +3168,7 @@ try_again:
 				return XFS_ERROR(EIO);
 			}
 			XFS_STATS_INC(xs_log_force_sleep);
-			sv_wait(&iclog->ic_force_wait, PSWP, &log->l_icloglock, s);
+			xlog_wait(&iclog->ic_force_wait, &log->l_icloglock);
 			/*
 			 * No need to grab the log lock here since we're
 			 * only deciding whether or not to return EIO
@@ -3310,10 +3243,8 @@ xfs_log_ticket_put(
 	xlog_ticket_t	*ticket)
 {
 	ASSERT(atomic_read(&ticket->t_ref) > 0);
-	if (atomic_dec_and_test(&ticket->t_ref)) {
-		sv_destroy(&ticket->t_wait);
+	if (atomic_dec_and_test(&ticket->t_ref))
 		kmem_zone_free(xfs_log_ticket_zone, ticket);
-	}
 }
 
 xlog_ticket_t *
@@ -3435,6 +3366,7 @@ xlog_ticket_alloc(
         }
 
 	atomic_set(&tic->t_ref, 1);
+	INIT_LIST_HEAD(&tic->t_queue);
 	tic->t_unit_res		= unit_bytes;
 	tic->t_curr_res		= unit_bytes;
 	tic->t_cnt		= cnt;
@@ -3445,7 +3377,7 @@ xlog_ticket_alloc(
 	tic->t_trans_type	= 0;
 	if (xflags & XFS_LOG_PERM_RESERV)
 		tic->t_flags |= XLOG_TIC_PERM_RESERV;
-	sv_init(&tic->t_wait, SV_DEFAULT, "logtick");
+	init_waitqueue_head(&tic->t_wait);
 
 	xlog_tic_reset_res(tic);
 
@@ -3484,18 +3416,25 @@ xlog_verify_dest_ptr(
 }
 
 STATIC void
-xlog_verify_grant_head(xlog_t *log, int equals)
+xlog_verify_grant_tail(
+	struct log	*log)
 {
-    if (log->l_grant_reserve_cycle == log->l_grant_write_cycle) {
-	if (equals)
-	    ASSERT(log->l_grant_reserve_bytes >= log->l_grant_write_bytes);
-	else
-	    ASSERT(log->l_grant_reserve_bytes > log->l_grant_write_bytes);
-    } else {
-	ASSERT(log->l_grant_reserve_cycle-1 == log->l_grant_write_cycle);
-	ASSERT(log->l_grant_write_bytes >= log->l_grant_reserve_bytes);
-    }
-}	/* xlog_verify_grant_head */
+	int		tail_cycle, tail_blocks;
+	int		cycle, space;
+
+	/*
+	 * Check to make sure the grant write head didn't just over lap the
+	 * tail.  If the cycles are the same, we can't be overlapping.
+	 * Otherwise, make sure that the cycles differ by exactly one and
+	 * check the byte count.
+	 */
+	xlog_crack_grant_head(&log->l_grant_write_head, &cycle, &space);
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &tail_cycle, &tail_blocks);
+	if (tail_cycle != cycle) {
+		ASSERT(cycle - 1 == tail_cycle);
+		ASSERT(space <= BBTOB(tail_blocks));
+	}
+}
 
 /* check if it will fit */
 STATIC void
@@ -3716,12 +3655,10 @@ xfs_log_force_umount(
 		xlog_cil_force(log);
 
 	/*
-	 * We must hold both the GRANT lock and the LOG lock,
-	 * before we mark the filesystem SHUTDOWN and wake
-	 * everybody up to tell the bad news.
+	 * mark the filesystem and the as in a shutdown state and wake
+	 * everybody up to tell them the bad news.
 	 */
 	spin_lock(&log->l_icloglock);
-	spin_lock(&log->l_grant_lock);
 	mp->m_flags |= XFS_MOUNT_FS_SHUTDOWN;
 	if (mp->m_sb_bp)
 		XFS_BUF_DONE(mp->m_sb_bp);
@@ -3742,27 +3679,21 @@ xfs_log_force_umount(
 	spin_unlock(&log->l_icloglock);
 
 	/*
-	 * We don't want anybody waiting for log reservations
-	 * after this. That means we have to wake up everybody
-	 * queued up on reserve_headq as well as write_headq.
-	 * In addition, we make sure in xlog_{re}grant_log_space
-	 * that we don't enqueue anything once the SHUTDOWN flag
-	 * is set, and this action is protected by the GRANTLOCK.
+	 * We don't want anybody waiting for log reservations after this. That
+	 * means we have to wake up everybody queued up on reserveq as well as
+	 * writeq.  In addition, we make sure in xlog_{re}grant_log_space that
+	 * we don't enqueue anything once the SHUTDOWN flag is set, and this
+	 * action is protected by the grant locks.
 	 */
-	if ((tic = log->l_reserve_headq)) {
-		do {
-			sv_signal(&tic->t_wait);
-			tic = tic->t_next;
-		} while (tic != log->l_reserve_headq);
-	}
-
-	if ((tic = log->l_write_headq)) {
-		do {
-			sv_signal(&tic->t_wait);
-			tic = tic->t_next;
-		} while (tic != log->l_write_headq);
-	}
-	spin_unlock(&log->l_grant_lock);
+	spin_lock(&log->l_grant_reserve_lock);
+	list_for_each_entry(tic, &log->l_reserveq, t_queue)
+		wake_up(&tic->t_wait);
+	spin_unlock(&log->l_grant_reserve_lock);
+
+	spin_lock(&log->l_grant_write_lock);
+	list_for_each_entry(tic, &log->l_writeq, t_queue)
+		wake_up(&tic->t_wait);
+	spin_unlock(&log->l_grant_write_lock);
 
 	if (!(log->l_iclog->ic_state & XLOG_STATE_IOERROR)) {
 		ASSERT(!logerror);

fs/xfs/xfs_log_cil.c

@@ -61,7 +61,7 @@ xlog_cil_init(
 	INIT_LIST_HEAD(&cil->xc_committing);
 	spin_lock_init(&cil->xc_cil_lock);
 	init_rwsem(&cil->xc_ctx_lock);
-	sv_init(&cil->xc_commit_wait, SV_DEFAULT, "cilwait");
+	init_waitqueue_head(&cil->xc_commit_wait);
 
 	INIT_LIST_HEAD(&ctx->committing);
 	INIT_LIST_HEAD(&ctx->busy_extents);
@@ -361,15 +361,10 @@ xlog_cil_committed(
 	int	abort)
 {
 	struct xfs_cil_ctx	*ctx = args;
-	struct xfs_log_vec	*lv;
-	int			abortflag = abort ? XFS_LI_ABORTED : 0;
 	struct xfs_busy_extent	*busyp, *n;
 
-	/* unpin all the log items */
-	for (lv = ctx->lv_chain; lv; lv = lv->lv_next ) {
-		xfs_trans_item_committed(lv->lv_item, ctx->start_lsn,
-							abortflag);
-	}
+	xfs_trans_committed_bulk(ctx->cil->xc_log->l_ailp, ctx->lv_chain,
+					ctx->start_lsn, abort);
 
 	list_for_each_entry_safe(busyp, n, &ctx->busy_extents, list)
 		xfs_alloc_busy_clear(ctx->cil->xc_log->l_mp, busyp);
@@ -568,7 +563,7 @@ restart:
 			 * It is still being pushed! Wait for the push to
 			 * complete, then start again from the beginning.
 			 */
-			sv_wait(&cil->xc_commit_wait, 0, &cil->xc_cil_lock, 0);
+			xlog_wait(&cil->xc_commit_wait, &cil->xc_cil_lock);
 			goto restart;
 		}
 	}
@@ -592,7 +587,7 @@ restart:
 	 */
 	spin_lock(&cil->xc_cil_lock);
 	ctx->commit_lsn = commit_lsn;
-	sv_broadcast(&cil->xc_commit_wait);
+	wake_up_all(&cil->xc_commit_wait);
 	spin_unlock(&cil->xc_cil_lock);
 
 	/* release the hounds! */
@@ -757,7 +752,7 @@ restart:
 			 * It is still being pushed! Wait for the push to
 			 * complete, then start again from the beginning.
 			 */
-			sv_wait(&cil->xc_commit_wait, 0, &cil->xc_cil_lock, 0);
+			xlog_wait(&cil->xc_commit_wait, &cil->xc_cil_lock);
 			goto restart;
 		}
 		if (ctx->sequence != sequence)

fs/xfs/xfs_log_priv.h

@@ -21,7 +21,6 @@
 struct xfs_buf;
 struct log;
 struct xlog_ticket;
-struct xfs_buf_cancel;
 struct xfs_mount;
 
 /*
@@ -54,7 +53,6 @@ struct xfs_mount;
 	BTOBB(XLOG_MAX_ICLOGS << (xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? \
 	 XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
 
-
 static inline xfs_lsn_t xlog_assign_lsn(uint cycle, uint block)
 {
 	return ((xfs_lsn_t)cycle << 32) | block;
@@ -133,12 +131,10 @@ static inline uint xlog_get_client_id(__be32 i)
  */
 #define XLOG_TIC_INITED		0x1	/* has been initialized */
 #define XLOG_TIC_PERM_RESERV	0x2	/* permanent reservation */
-#define XLOG_TIC_IN_Q		0x4
 
 #define XLOG_TIC_FLAGS \
 	{ XLOG_TIC_INITED,	"XLOG_TIC_INITED" }, \
-	{ XLOG_TIC_PERM_RESERV,	"XLOG_TIC_PERM_RESERV" }, \
-	{ XLOG_TIC_IN_Q,	"XLOG_TIC_IN_Q" }
+	{ XLOG_TIC_PERM_RESERV,	"XLOG_TIC_PERM_RESERV" }
 
 #endif	/* __KERNEL__ */
 
@@ -244,9 +240,8 @@ typedef struct xlog_res {
 } xlog_res_t;
 
 typedef struct xlog_ticket {
-	sv_t		   t_wait;	 /* ticket wait queue            : 20 */
-	struct xlog_ticket *t_next;	 /*			         :4|8 */
-	struct xlog_ticket *t_prev;	 /*				 :4|8 */
+	wait_queue_head_t  t_wait;	 /* ticket wait queue */
+	struct list_head   t_queue;	 /* reserve/write queue */
 	xlog_tid_t	   t_tid;	 /* transaction identifier	 : 4  */
 	atomic_t	   t_ref;	 /* ticket reference count       : 4  */
 	int		   t_curr_res;	 /* current reservation in bytes : 4  */
@@ -353,8 +348,8 @@ typedef union xlog_in_core2 {
  * and move everything else out to subsequent cachelines.
  */
 typedef struct xlog_in_core {
-	sv_t			ic_force_wait;
-	sv_t			ic_write_wait;
+	wait_queue_head_t	ic_force_wait;
+	wait_queue_head_t	ic_write_wait;
 	struct xlog_in_core	*ic_next;
 	struct xlog_in_core	*ic_prev;
 	struct xfs_buf		*ic_bp;
@@ -421,7 +416,7 @@ struct xfs_cil {
 	struct xfs_cil_ctx	*xc_ctx;
 	struct rw_semaphore	xc_ctx_lock;
 	struct list_head	xc_committing;
-	sv_t			xc_commit_wait;
+	wait_queue_head_t	xc_commit_wait;
 	xfs_lsn_t		xc_current_sequence;
 };
 
@@ -491,7 +486,7 @@ typedef struct log {
 	struct xfs_buftarg	*l_targ;        /* buftarg of log */
 	uint			l_flags;
 	uint			l_quotaoffs_flag; /* XFS_DQ_*, for QUOTAOFFs */
-	struct xfs_buf_cancel	**l_buf_cancel_table;
+	struct list_head	*l_buf_cancel_table;
 	int			l_iclog_hsize;  /* size of iclog header */
 	int			l_iclog_heads;  /* # of iclog header sectors */
 	uint			l_sectBBsize;   /* sector size in BBs (2^n) */
@@ -503,29 +498,40 @@ typedef struct log {
 	int			l_logBBsize;    /* size of log in BB chunks */
 
 	/* The following block of fields are changed while holding icloglock */
-	sv_t			l_flush_wait ____cacheline_aligned_in_smp;
+	wait_queue_head_t	l_flush_wait ____cacheline_aligned_in_smp;
 						/* waiting for iclog flush */
 	int			l_covered_state;/* state of "covering disk
 						 * log entries" */
 	xlog_in_core_t		*l_iclog;       /* head log queue	*/
 	spinlock_t		l_icloglock;    /* grab to change iclog state */
-	xfs_lsn_t		l_tail_lsn;     /* lsn of 1st LR with unflushed
-						 * buffers */
-	xfs_lsn_t		l_last_sync_lsn;/* lsn of last LR on disk */
 	int			l_curr_cycle;   /* Cycle number of log writes */
 	int			l_prev_cycle;   /* Cycle number before last
 						 * block increment */
 	int			l_curr_block;   /* current logical log block */
 	int			l_prev_block;   /* previous logical log block */
 
-	/* The following block of fields are changed while holding grant_lock */
-	spinlock_t		l_grant_lock ____cacheline_aligned_in_smp;
-	xlog_ticket_t		*l_reserve_headq;
-	xlog_ticket_t		*l_write_headq;
-	int			l_grant_reserve_cycle;
-	int			l_grant_reserve_bytes;
-	int			l_grant_write_cycle;
-	int			l_grant_write_bytes;
+	/*
+	 * l_last_sync_lsn and l_tail_lsn are atomics so they can be set and
+	 * read without needing to hold specific locks. To avoid operations
+	 * contending with other hot objects, place each of them on a separate
+	 * cacheline.
+	 */
+	/* lsn of last LR on disk */
+	atomic64_t		l_last_sync_lsn ____cacheline_aligned_in_smp;
+	/* lsn of 1st LR with unflushed * buffers */
+	atomic64_t		l_tail_lsn ____cacheline_aligned_in_smp;
+
+	/*
+	 * ticket grant locks, queues and accounting have their own cachlines
+	 * as these are quite hot and can be operated on concurrently.
+	 */
+	spinlock_t		l_grant_reserve_lock ____cacheline_aligned_in_smp;
+	struct list_head	l_reserveq;
+	atomic64_t		l_grant_reserve_head;
+
+	spinlock_t		l_grant_write_lock ____cacheline_aligned_in_smp;
+	struct list_head	l_writeq;
+	atomic64_t		l_grant_write_head;
 
 	/* The following field are used for debugging; need to hold icloglock */
 #ifdef DEBUG
@@ -534,6 +540,9 @@ typedef struct log {
 
 } xlog_t;
 
+#define XLOG_BUF_CANCEL_BUCKET(log, blkno) \
+	((log)->l_buf_cancel_table + ((__uint64_t)blkno % XLOG_BC_TABLE_SIZE))
+
 #define XLOG_FORCED_SHUTDOWN(log)	((log)->l_flags & XLOG_IO_ERROR)
 
 /* common routines */
@@ -561,6 +570,61 @@ int	xlog_write(struct log *log, struct xfs_log_vec *log_vector,
 				struct xlog_ticket *tic, xfs_lsn_t *start_lsn,
 				xlog_in_core_t **commit_iclog, uint flags);
 
+/*
+ * When we crack an atomic LSN, we sample it first so that the value will not
+ * change while we are cracking it into the component values. This means we
+ * will always get consistent component values to work from. This should always
+ * be used to smaple and crack LSNs taht are stored and updated in atomic
+ * variables.
+ */
+static inline void
+xlog_crack_atomic_lsn(atomic64_t *lsn, uint *cycle, uint *block)
+{
+	xfs_lsn_t val = atomic64_read(lsn);
+
+	*cycle = CYCLE_LSN(val);
+	*block = BLOCK_LSN(val);
+}
+
+/*
+ * Calculate and assign a value to an atomic LSN variable from component pieces.
+ */
+static inline void
+xlog_assign_atomic_lsn(atomic64_t *lsn, uint cycle, uint block)
+{
+	atomic64_set(lsn, xlog_assign_lsn(cycle, block));
+}
+
+/*
+ * When we crack the grant head, we sample it first so that the value will not
+ * change while we are cracking it into the component values. This means we
+ * will always get consistent component values to work from.
+ */
+static inline void
+xlog_crack_grant_head_val(int64_t val, int *cycle, int *space)
+{
+	*cycle = val >> 32;
+	*space = val & 0xffffffff;
+}
+
+static inline void
+xlog_crack_grant_head(atomic64_t *head, int *cycle, int *space)
+{
+	xlog_crack_grant_head_val(atomic64_read(head), cycle, space);
+}
+
+static inline int64_t
+xlog_assign_grant_head_val(int cycle, int space)
+{
+	return ((int64_t)cycle << 32) | space;
+}
+
+static inline void
+xlog_assign_grant_head(atomic64_t *head, int cycle, int space)
+{
+	atomic64_set(head, xlog_assign_grant_head_val(cycle, space));
+}
+
 /*
  * Committed Item List interfaces
  */
@@ -585,6 +649,21 @@ xlog_cil_force(struct log *log)
  */
 #define XLOG_UNMOUNT_REC_TYPE	(-1U)
 
+/*
+ * Wrapper function for waiting on a wait queue serialised against wakeups
+ * by a spinlock. This matches the semantics of all the wait queues used in the
+ * log code.
+ */
+static inline void xlog_wait(wait_queue_head_t *wq, spinlock_t *lock)
+{
+	DECLARE_WAITQUEUE(wait, current);
+
+	add_wait_queue_exclusive(wq, &wait);
+	__set_current_state(TASK_UNINTERRUPTIBLE);
+	spin_unlock(lock);
+	schedule();
+	remove_wait_queue(wq, &wait);
+}
 #endif	/* __KERNEL__ */
 
 #endif	/* __XFS_LOG_PRIV_H__ */

fs/xfs/xfs_log_recover.c

@@ -52,6 +52,17 @@ STATIC void	xlog_recover_check_summary(xlog_t *);
 #define	xlog_recover_check_summary(log)
 #endif
 
+/*
+ * This structure is used during recovery to record the buf log items which
+ * have been canceled and should not be replayed.
+ */
+struct xfs_buf_cancel {
+	xfs_daddr_t		bc_blkno;
+	uint			bc_len;
+	int			bc_refcount;
+	struct list_head	bc_list;
+};
+
 /*
  * Sector aligned buffer routines for buffer create/read/write/access
  */
@@ -925,12 +936,12 @@ xlog_find_tail(
 	log->l_curr_cycle = be32_to_cpu(rhead->h_cycle);
 	if (found == 2)
 		log->l_curr_cycle++;
-	log->l_tail_lsn = be64_to_cpu(rhead->h_tail_lsn);
-	log->l_last_sync_lsn = be64_to_cpu(rhead->h_lsn);
-	log->l_grant_reserve_cycle = log->l_curr_cycle;
-	log->l_grant_reserve_bytes = BBTOB(log->l_curr_block);
-	log->l_grant_write_cycle = log->l_curr_cycle;
-	log->l_grant_write_bytes = BBTOB(log->l_curr_block);
+	atomic64_set(&log->l_tail_lsn, be64_to_cpu(rhead->h_tail_lsn));
+	atomic64_set(&log->l_last_sync_lsn, be64_to_cpu(rhead->h_lsn));
+	xlog_assign_grant_head(&log->l_grant_reserve_head, log->l_curr_cycle,
+					BBTOB(log->l_curr_block));
+	xlog_assign_grant_head(&log->l_grant_write_head, log->l_curr_cycle,
+					BBTOB(log->l_curr_block));
 
 	/*
 	 * Look for unmount record.  If we find it, then we know there
@@ -960,7 +971,7 @@ xlog_find_tail(
 	}
 	after_umount_blk = (i + hblks + (int)
 		BTOBB(be32_to_cpu(rhead->h_len))) % log->l_logBBsize;
-	tail_lsn = log->l_tail_lsn;
+	tail_lsn = atomic64_read(&log->l_tail_lsn);
 	if (*head_blk == after_umount_blk &&
 	    be32_to_cpu(rhead->h_num_logops) == 1) {
 		umount_data_blk = (i + hblks) % log->l_logBBsize;
@@ -975,12 +986,10 @@ xlog_find_tail(
 			 * log records will point recovery to after the
 			 * current unmount record.
 			 */
-			log->l_tail_lsn =
-				xlog_assign_lsn(log->l_curr_cycle,
-						after_umount_blk);
-			log->l_last_sync_lsn =
-				xlog_assign_lsn(log->l_curr_cycle,
-						after_umount_blk);
+			xlog_assign_atomic_lsn(&log->l_tail_lsn,
+					log->l_curr_cycle, after_umount_blk);
+			xlog_assign_atomic_lsn(&log->l_last_sync_lsn,
+					log->l_curr_cycle, after_umount_blk);
 			*tail_blk = after_umount_blk;
 
 			/*
@@ -1605,82 +1614,45 @@ xlog_recover_reorder_trans(
  * record in the table to tell us how many times we expect to see this
  * record during the second pass.
  */
-STATIC void
-xlog_recover_do_buffer_pass1(
-	xlog_t			*log,
-	xfs_buf_log_format_t	*buf_f)
+STATIC int
+xlog_recover_buffer_pass1(
+	struct log		*log,
+	xlog_recover_item_t	*item)
 {
-	xfs_buf_cancel_t	*bcp;
-	xfs_buf_cancel_t	*nextp;
-	xfs_buf_cancel_t	*prevp;
-	xfs_buf_cancel_t	**bucket;
-	xfs_daddr_t		blkno = 0;
-	uint			len = 0;
-	ushort			flags = 0;
-
-	switch (buf_f->blf_type) {
-	case XFS_LI_BUF:
-		blkno = buf_f->blf_blkno;
-		len = buf_f->blf_len;
-		flags = buf_f->blf_flags;
-		break;
-	}
+	xfs_buf_log_format_t	*buf_f = item->ri_buf[0].i_addr;
+	struct list_head	*bucket;
+	struct xfs_buf_cancel	*bcp;
 
 	/*
 	 * If this isn't a cancel buffer item, then just return.
 	 */
-	if (!(flags & XFS_BLF_CANCEL)) {
+	if (!(buf_f->blf_flags & XFS_BLF_CANCEL)) {
 		trace_xfs_log_recover_buf_not_cancel(log, buf_f);
-		return;
-	}
-
-	/*
-	 * Insert an xfs_buf_cancel record into the hash table of
-	 * them.  If there is already an identical record, bump
-	 * its reference count.
-	 */
-	bucket = &log->l_buf_cancel_table[(__uint64_t)blkno %
-					  XLOG_BC_TABLE_SIZE];
-	/*
-	 * If the hash bucket is empty then just insert a new record into
-	 * the bucket.
-	 */
-	if (*bucket == NULL) {
-		bcp = (xfs_buf_cancel_t *)kmem_alloc(sizeof(xfs_buf_cancel_t),
-						     KM_SLEEP);
-		bcp->bc_blkno = blkno;
-		bcp->bc_len = len;
-		bcp->bc_refcount = 1;
-		bcp->bc_next = NULL;
-		*bucket = bcp;
-		return;
+		return 0;
 	}
 
 	/*
-	 * The hash bucket is not empty, so search for duplicates of our
-	 * record.  If we find one them just bump its refcount.  If not
-	 * then add us at the end of the list.
+	 * Insert an xfs_buf_cancel record into the hash table of them.
+	 * If there is already an identical record, bump its reference count.
 	 */
-	prevp = NULL;
-	nextp = *bucket;
-	while (nextp != NULL) {
-		if (nextp->bc_blkno == blkno && nextp->bc_len == len) {
-			nextp->bc_refcount++;
+	bucket = XLOG_BUF_CANCEL_BUCKET(log, buf_f->blf_blkno);
+	list_for_each_entry(bcp, bucket, bc_list) {
+		if (bcp->bc_blkno == buf_f->blf_blkno &&
+		    bcp->bc_len == buf_f->blf_len) {
+			bcp->bc_refcount++;
 			trace_xfs_log_recover_buf_cancel_ref_inc(log, buf_f);
-			return;
+			return 0;
 		}
-		prevp = nextp;
-		nextp = nextp->bc_next;
-	}
-	ASSERT(prevp != NULL);
-	bcp = (xfs_buf_cancel_t *)kmem_alloc(sizeof(xfs_buf_cancel_t),
-					     KM_SLEEP);
-	bcp->bc_blkno = blkno;
-	bcp->bc_len = len;
+	}
+
+	bcp = kmem_alloc(sizeof(struct xfs_buf_cancel), KM_SLEEP);
+	bcp->bc_blkno = buf_f->blf_blkno;
+	bcp->bc_len = buf_f->blf_len;
 	bcp->bc_refcount = 1;
-	bcp->bc_next = NULL;
-	prevp->bc_next = bcp;
+	list_add_tail(&bcp->bc_list, bucket);
+
 	trace_xfs_log_recover_buf_cancel_add(log, buf_f);
+	return 0;
 }
 
 /*
@@ -1698,14 +1670,13 @@ xlog_recover_do_buffer_pass1(
  */
 STATIC int
 xlog_check_buffer_cancelled(
-	xlog_t			*log,
+	struct log		*log,
 	xfs_daddr_t		blkno,
 	uint			len,
 	ushort			flags)
 {
-	xfs_buf_cancel_t	*bcp;
-	xfs_buf_cancel_t	*prevp;
-	xfs_buf_cancel_t	**bucket;
+	struct list_head	*bucket;
+	struct xfs_buf_cancel	*bcp;
 
 	if (log->l_buf_cancel_table == NULL) {
 		/*
@@ -1716,128 +1687,70 @@ xlog_check_buffer_cancelled(
 		return 0;
 	}
 
-	bucket = &log->l_buf_cancel_table[(__uint64_t)blkno %
-					  XLOG_BC_TABLE_SIZE];
-	bcp = *bucket;
-	if (bcp == NULL) {
-		/*
-		 * There is no corresponding entry in the table built
-		 * in pass one, so this buffer has not been cancelled.
-		 */
-		ASSERT(!(flags & XFS_BLF_CANCEL));
-		return 0;
-	}
-
 	/*
-	 * Search for an entry in the buffer cancel table that
-	 * matches our buffer.
+	 * Search for an entry in the  cancel table that matches our buffer.
 	 */
-	prevp = NULL;
-	while (bcp != NULL) {
-		if (bcp->bc_blkno == blkno && bcp->bc_len == len) {
-			/*
-			 * We've go a match, so return 1 so that the
-			 * recovery of this buffer is cancelled.
-			 * If this buffer is actually a buffer cancel
-			 * log item, then decrement the refcount on the
-			 * one in the table and remove it if this is the
-			 * last reference.
-			 */
-			if (flags & XFS_BLF_CANCEL) {
-				bcp->bc_refcount--;
-				if (bcp->bc_refcount == 0) {
-					if (prevp == NULL) {
-						*bucket = bcp->bc_next;
-					} else {
-						prevp->bc_next = bcp->bc_next;
-					}
-					kmem_free(bcp);
-				}
-			}
-			return 1;
-		}
-		prevp = bcp;
-		bcp = bcp->bc_next;
+	bucket = XLOG_BUF_CANCEL_BUCKET(log, blkno);
+	list_for_each_entry(bcp, bucket, bc_list) {
+		if (bcp->bc_blkno == blkno && bcp->bc_len == len)
+			goto found;
 	}
+
 	/*
-	 * We didn't find a corresponding entry in the table, so
-	 * return 0 so that the buffer is NOT cancelled.
+	 * We didn't find a corresponding entry in the table, so return 0 so
+	 * that the buffer is NOT cancelled.
 	 */
 	ASSERT(!(flags & XFS_BLF_CANCEL));
 	return 0;
-}
 
-STATIC int
-xlog_recover_do_buffer_pass2(
-	xlog_t			*log,
-	xfs_buf_log_format_t	*buf_f)
-{
-	xfs_daddr_t		blkno = 0;
-	ushort			flags = 0;
-	uint			len = 0;
-
-	switch (buf_f->blf_type) {
-	case XFS_LI_BUF:
-		blkno = buf_f->blf_blkno;
-		flags = buf_f->blf_flags;
-		len = buf_f->blf_len;
-		break;
+found:
+	/*
+	 * We've go a match, so return 1 so that the recovery of this buffer
+	 * is cancelled.  If this buffer is actually a buffer cancel log
+	 * item, then decrement the refcount on the one in the table and
+	 * remove it if this is the last reference.
+	 */
+	if (flags & XFS_BLF_CANCEL) {
+		if (--bcp->bc_refcount == 0) {
+			list_del(&bcp->bc_list);
+			kmem_free(bcp);
+		}
 	}
-
-	return xlog_check_buffer_cancelled(log, blkno, len, flags);
+	return 1;
 }
 
 /*
- * Perform recovery for a buffer full of inodes.  In these buffers,
- * the only data which should be recovered is that which corresponds
- * to the di_next_unlinked pointers in the on disk inode structures.
- * The rest of the data for the inodes is always logged through the
- * inodes themselves rather than the inode buffer and is recovered
- * in xlog_recover_do_inode_trans().
+ * Perform recovery for a buffer full of inodes.  In these buffers, the only
+ * data which should be recovered is that which corresponds to the
+ * di_next_unlinked pointers in the on disk inode structures.  The rest of the
+ * data for the inodes is always logged through the inodes themselves rather
+ * than the inode buffer and is recovered in xlog_recover_inode_pass2().
  *
- * The only time when buffers full of inodes are fully recovered is
- * when the buffer is full of newly allocated inodes.  In this case
- * the buffer will not be marked as an inode buffer and so will be
- * sent to xlog_recover_do_reg_buffer() below during recovery.
+ * The only time when buffers full of inodes are fully recovered is when the
+ * buffer is full of newly allocated inodes.  In this case the buffer will
+ * not be marked as an inode buffer and so will be sent to
+ * xlog_recover_do_reg_buffer() below during recovery.
  */
 STATIC int
 xlog_recover_do_inode_buffer(
-	xfs_mount_t		*mp,
+	struct xfs_mount	*mp,
 	xlog_recover_item_t	*item,
-	xfs_buf_t		*bp,
+	struct xfs_buf		*bp,
 	xfs_buf_log_format_t	*buf_f)
 {
 	int			i;
-	int			item_index;
-	int			bit;
-	int			nbits;
-	int			reg_buf_offset;
-	int			reg_buf_bytes;
+	int			item_index = 0;
+	int			bit = 0;
+	int			nbits = 0;
+	int			reg_buf_offset = 0;
+	int			reg_buf_bytes = 0;
 	int			next_unlinked_offset;
 	int			inodes_per_buf;
 	xfs_agino_t		*logged_nextp;
 	xfs_agino_t		*buffer_nextp;
-	unsigned int		*data_map = NULL;
-	unsigned int		map_size = 0;
 
 	trace_xfs_log_recover_buf_inode_buf(mp->m_log, buf_f);
 
-	switch (buf_f->blf_type) {
-	case XFS_LI_BUF:
-		data_map = buf_f->blf_data_map;
-		map_size = buf_f->blf_map_size;
-		break;
-	}
-	/*
-	 * Set the variables corresponding to the current region to
-	 * 0 so that we'll initialize them on the first pass through
-	 * the loop.
-	 */
-	reg_buf_offset = 0;
-	reg_buf_bytes = 0;
-	bit = 0;
-	nbits = 0;
-	item_index = 0;
 	inodes_per_buf = XFS_BUF_COUNT(bp) >> mp->m_sb.sb_inodelog;
 	for (i = 0; i < inodes_per_buf; i++) {
 		next_unlinked_offset = (i * mp->m_sb.sb_inodesize) +
@@ -1852,18 +1765,18 @@ xlog_recover_do_inode_buffer(
 			 * the current di_next_unlinked field.
 			 */
 			bit += nbits;
-			bit = xfs_next_bit(data_map, map_size, bit);
+			bit = xfs_next_bit(buf_f->blf_data_map,
+					   buf_f->blf_map_size, bit);
 
 			/*
 			 * If there are no more logged regions in the
 			 * buffer, then we're done.
 			 */
-			if (bit == -1) {
+			if (bit == -1)
 				return 0;
-			}
 
-			nbits = xfs_contig_bits(data_map, map_size,
-							 bit);
+			nbits = xfs_contig_bits(buf_f->blf_data_map,
+						buf_f->blf_map_size, bit);
 			ASSERT(nbits > 0);
 			reg_buf_offset = bit << XFS_BLF_SHIFT;
 			reg_buf_bytes = nbits << XFS_BLF_SHIFT;
@@ -1875,9 +1788,8 @@ xlog_recover_do_inode_buffer(
 		 * di_next_unlinked field, then move on to the next
 		 * di_next_unlinked field.
 		 */
-		if (next_unlinked_offset < reg_buf_offset) {
+		if (next_unlinked_offset < reg_buf_offset)
 			continue;
-		}
 
 		ASSERT(item->ri_buf[item_index].i_addr != NULL);
 		ASSERT((item->ri_buf[item_index].i_len % XFS_BLF_CHUNK) == 0);
@@ -1913,36 +1825,29 @@ xlog_recover_do_inode_buffer(
  * given buffer.  The bitmap in the buf log format structure indicates
  * where to place the logged data.
  */
-/*ARGSUSED*/
 STATIC void
 xlog_recover_do_reg_buffer(
 	struct xfs_mount	*mp,
 	xlog_recover_item_t	*item,
-	xfs_buf_t		*bp,
+	struct xfs_buf		*bp,
 	xfs_buf_log_format_t	*buf_f)
 {
 	int			i;
 	int			bit;
 	int			nbits;
-	unsigned int		*data_map = NULL;
-	unsigned int		map_size = 0;
 	int                     error;
 
 	trace_xfs_log_recover_buf_reg_buf(mp->m_log, buf_f);
 
-	switch (buf_f->blf_type) {
-	case XFS_LI_BUF:
-		data_map = buf_f->blf_data_map;
-		map_size = buf_f->blf_map_size;
-		break;
-	}
 	bit = 0;
 	i = 1;  /* 0 is the buf format structure */
 	while (1) {
-		bit = xfs_next_bit(data_map, map_size, bit);
+		bit = xfs_next_bit(buf_f->blf_data_map,
+				   buf_f->blf_map_size, bit);
 		if (bit == -1)
 			break;
-		nbits = xfs_contig_bits(data_map, map_size, bit);
+		nbits = xfs_contig_bits(buf_f->blf_data_map,
+					buf_f->blf_map_size, bit);
 		ASSERT(nbits > 0);
 		ASSERT(item->ri_buf[i].i_addr != NULL);
 		ASSERT(item->ri_buf[i].i_len % XFS_BLF_CHUNK == 0);
@@ -2176,77 +2081,46 @@ xlog_recover_do_dquot_buffer(
  * for more details on the implementation of the table of cancel records.
  */
 STATIC int
-xlog_recover_do_buffer_trans(
+xlog_recover_buffer_pass2(
 	xlog_t			*log,
-	xlog_recover_item_t	*item,
-	int			pass)
+	xlog_recover_item_t	*item)
 {
 	xfs_buf_log_format_t	*buf_f = item->ri_buf[0].i_addr;
-	xfs_mount_t		*mp;
+	xfs_mount_t		*mp = log->l_mp;
 	xfs_buf_t		*bp;
 	int			error;
-	int			cancel;
-	xfs_daddr_t		blkno;
-	int			len;
-	ushort			flags;
 	uint			buf_flags;
 
-	if (pass == XLOG_RECOVER_PASS1) {
-		/*
-		 * In this pass we're only looking for buf items
-		 * with the XFS_BLF_CANCEL bit set.
-		 */
-		xlog_recover_do_buffer_pass1(log, buf_f);
+	/*
+	 * In this pass we only want to recover all the buffers which have
+	 * not been cancelled and are not cancellation buffers themselves.
+	 */
+	if (xlog_check_buffer_cancelled(log, buf_f->blf_blkno,
+			buf_f->blf_len, buf_f->blf_flags)) {
+		trace_xfs_log_recover_buf_cancel(log, buf_f);
 		return 0;
-	} else {
-		/*
-		 * In this pass we want to recover all the buffers
-		 * which have not been cancelled and are not
-		 * cancellation buffers themselves.  The routine
-		 * we call here will tell us whether or not to
-		 * continue with the replay of this buffer.
-		 */
-		cancel = xlog_recover_do_buffer_pass2(log, buf_f);
-		if (cancel) {
-			trace_xfs_log_recover_buf_cancel(log, buf_f);
-			return 0;
-		}
 	}
+
 	trace_xfs_log_recover_buf_recover(log, buf_f);
-	switch (buf_f->blf_type) {
-	case XFS_LI_BUF:
-		blkno = buf_f->blf_blkno;
-		len = buf_f->blf_len;
-		flags = buf_f->blf_flags;
-		break;
-	default:
-		xfs_fs_cmn_err(CE_ALERT, log->l_mp,
-			"xfs_log_recover: unknown buffer type 0x%x, logdev %s",
-			buf_f->blf_type, log->l_mp->m_logname ?
-			log->l_mp->m_logname : "internal");
-		XFS_ERROR_REPORT("xlog_recover_do_buffer_trans",
-				 XFS_ERRLEVEL_LOW, log->l_mp);
-		return XFS_ERROR(EFSCORRUPTED);
-	}
 
-	mp = log->l_mp;
 	buf_flags = XBF_LOCK;
-	if (!(flags & XFS_BLF_INODE_BUF))
+	if (!(buf_f->blf_flags & XFS_BLF_INODE_BUF))
 		buf_flags |= XBF_MAPPED;
 
-	bp = xfs_buf_read(mp->m_ddev_targp, blkno, len, buf_flags);
+	bp = xfs_buf_read(mp->m_ddev_targp, buf_f->blf_blkno, buf_f->blf_len,
+			  buf_flags);
 	if (XFS_BUF_ISERROR(bp)) {
-		xfs_ioerror_alert("xlog_recover_do..(read#1)", log->l_mp,
-				  bp, blkno);
+		xfs_ioerror_alert("xlog_recover_do..(read#1)", mp,
+				  bp, buf_f->blf_blkno);
 		error = XFS_BUF_GETERROR(bp);
 		xfs_buf_relse(bp);
 		return error;
 	}
 
 	error = 0;
-	if (flags & XFS_BLF_INODE_BUF) {
+	if (buf_f->blf_flags & XFS_BLF_INODE_BUF) {
 		error = xlog_recover_do_inode_buffer(mp, item, bp, buf_f);
-	} else if (flags &
+	} else if (buf_f->blf_flags &
 		  (XFS_BLF_UDQUOT_BUF|XFS_BLF_PDQUOT_BUF|XFS_BLF_GDQUOT_BUF)) {
 		xlog_recover_do_dquot_buffer(mp, log, item, bp, buf_f);
 	} else {
@@ -2286,16 +2160,14 @@ xlog_recover_do_buffer_trans(
 }
 
 STATIC int
-xlog_recover_do_inode_trans(
+xlog_recover_inode_pass2(
 	xlog_t			*log,
-	xlog_recover_item_t	*item,
-	int			pass)
+	xlog_recover_item_t	*item)
 {
 	xfs_inode_log_format_t	*in_f;
-	xfs_mount_t		*mp;
+	xfs_mount_t		*mp = log->l_mp;
 	xfs_buf_t		*bp;
 	xfs_dinode_t		*dip;
-	xfs_ino_t		ino;
 	int			len;
 	xfs_caddr_t		src;
 	xfs_caddr_t		dest;
@@ -2305,10 +2177,6 @@ xlog_recover_do_inode_trans(
 	xfs_icdinode_t		*dicp;
 	int			need_free = 0;
 
-	if (pass == XLOG_RECOVER_PASS1) {
-		return 0;
-	}
-
 	if (item->ri_buf[0].i_len == sizeof(xfs_inode_log_format_t)) {
 		in_f = item->ri_buf[0].i_addr;
 	} else {
@@ -2318,8 +2186,6 @@ xlog_recover_do_inode_trans(
 		if (error)
 			goto error;
 	}
-	ino = in_f->ilf_ino;
-	mp = log->l_mp;
 
 	/*
 	 * Inode buffers can be freed, look out for it,
@@ -2354,8 +2220,8 @@ xlog_recover_do_inode_trans(
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
 			"xfs_inode_recover: Bad inode magic number, dino ptr = 0x%p, dino bp = 0x%p, ino = %Ld",
-			dip, bp, ino);
-		XFS_ERROR_REPORT("xlog_recover_do_inode_trans(1)",
+			dip, bp, in_f->ilf_ino);
+		XFS_ERROR_REPORT("xlog_recover_inode_pass2(1)",
 				 XFS_ERRLEVEL_LOW, mp);
 		error = EFSCORRUPTED;
 		goto error;
@@ -2365,8 +2231,8 @@ xlog_recover_do_inode_trans(
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
 			"xfs_inode_recover: Bad inode log record, rec ptr 0x%p, ino %Ld",
-			item, ino);
-		XFS_ERROR_REPORT("xlog_recover_do_inode_trans(2)",
+			item, in_f->ilf_ino);
+		XFS_ERROR_REPORT("xlog_recover_inode_pass2(2)",
 				 XFS_ERRLEVEL_LOW, mp);
 		error = EFSCORRUPTED;
 		goto error;
@@ -2394,12 +2260,12 @@ xlog_recover_do_inode_trans(
 	if (unlikely((dicp->di_mode & S_IFMT) == S_IFREG)) {
 		if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
 		    (dicp->di_format != XFS_DINODE_FMT_BTREE)) {
-			XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(3)",
+			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(3)",
 					 XFS_ERRLEVEL_LOW, mp, dicp);
 			xfs_buf_relse(bp);
 			xfs_fs_cmn_err(CE_ALERT, mp,
 				"xfs_inode_recover: Bad regular inode log record, rec ptr 0x%p, ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
-				item, dip, bp, ino);
+				item, dip, bp, in_f->ilf_ino);
 			error = EFSCORRUPTED;
 			goto error;
 		}
@@ -2407,40 +2273,40 @@ xlog_recover_do_inode_trans(
 		if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
 		    (dicp->di_format != XFS_DINODE_FMT_BTREE) &&
 		    (dicp->di_format != XFS_DINODE_FMT_LOCAL)) {
-			XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(4)",
+			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(4)",
 					     XFS_ERRLEVEL_LOW, mp, dicp);
 			xfs_buf_relse(bp);
 			xfs_fs_cmn_err(CE_ALERT, mp,
 				"xfs_inode_recover: Bad dir inode log record, rec ptr 0x%p, ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
-				item, dip, bp, ino);
+				item, dip, bp, in_f->ilf_ino);
 			error = EFSCORRUPTED;
 			goto error;
 		}
 	}
 	if (unlikely(dicp->di_nextents + dicp->di_anextents > dicp->di_nblocks)){
-		XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(5)",
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(5)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
 			"xfs_inode_recover: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, dino bp 0x%p, ino %Ld, total extents = %d, nblocks = %Ld",
-			item, dip, bp, ino,
+			item, dip, bp, in_f->ilf_ino,
 			dicp->di_nextents + dicp->di_anextents,
 			dicp->di_nblocks);
 		error = EFSCORRUPTED;
 		goto error;
 	}
 	if (unlikely(dicp->di_forkoff > mp->m_sb.sb_inodesize)) {
-		XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(6)",
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(6)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
 			"xfs_inode_recover: Bad inode log rec ptr 0x%p, dino ptr 0x%p, dino bp 0x%p, ino %Ld, forkoff 0x%x",
-			item, dip, bp, ino, dicp->di_forkoff);
+			item, dip, bp, in_f->ilf_ino, dicp->di_forkoff);
 		error = EFSCORRUPTED;
 		goto error;
 	}
 	if (unlikely(item->ri_buf[1].i_len > sizeof(struct xfs_icdinode))) {
-		XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(7)",
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(7)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
@@ -2532,7 +2398,7 @@ xlog_recover_do_inode_trans(
 			break;
 
 		default:
-			xlog_warn("XFS: xlog_recover_do_inode_trans: Invalid flag");
+			xlog_warn("XFS: xlog_recover_inode_pass2: Invalid flag");
 			ASSERT(0);
 			xfs_buf_relse(bp);
 			error = EIO;
@@ -2556,18 +2422,11 @@ error:
  * of that type.
  */
 STATIC int
-xlog_recover_do_quotaoff_trans(
+xlog_recover_quotaoff_pass1(
 	xlog_t			*log,
-	xlog_recover_item_t	*item,
-	int			pass)
+	xlog_recover_item_t	*item)
 {
-	xfs_qoff_logformat_t	*qoff_f;
-
-	if (pass == XLOG_RECOVER_PASS2) {
-		return (0);
-	}
-
-	qoff_f = item->ri_buf[0].i_addr;
+	xfs_qoff_logformat_t	*qoff_f = item->ri_buf[0].i_addr;
 	ASSERT(qoff_f);
 
 	/*
@@ -2588,22 +2447,17 @@ xlog_recover_do_quotaoff_trans(
  * Recover a dquot record
  */
 STATIC int
-xlog_recover_do_dquot_trans(
+xlog_recover_dquot_pass2(
 	xlog_t			*log,
-	xlog_recover_item_t	*item,
-	int			pass)
+	xlog_recover_item_t	*item)
 {
-	xfs_mount_t		*mp;
+	xfs_mount_t		*mp = log->l_mp;
 	xfs_buf_t		*bp;
 	struct xfs_disk_dquot	*ddq, *recddq;
 	int			error;
 	xfs_dq_logformat_t	*dq_f;
 	uint			type;
 
-	if (pass == XLOG_RECOVER_PASS1) {
-		return 0;
-	}
-	mp = log->l_mp;
 
 	/*
 	 * Filesystems are required to send in quota flags at mount time.
@@ -2647,7 +2501,7 @@ xlog_recover_do_dquot_trans(
 	if ((error = xfs_qm_dqcheck(recddq,
 			   dq_f->qlf_id,
 			   0, XFS_QMOPT_DOWARN,
-			   "xlog_recover_do_dquot_trans (log copy)"))) {
+			   "xlog_recover_dquot_pass2 (log copy)"))) {
 		return XFS_ERROR(EIO);
 	}
 	ASSERT(dq_f->qlf_len == 1);
@@ -2670,7 +2524,7 @@ xlog_recover_do_dquot_trans(
 	 * minimal initialization then.
 	 */
 	if (xfs_qm_dqcheck(ddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
-			   "xlog_recover_do_dquot_trans")) {
+			   "xlog_recover_dquot_pass2")) {
 		xfs_buf_relse(bp);
 		return XFS_ERROR(EIO);
 	}
@@ -2693,38 +2547,31 @@ xlog_recover_do_dquot_trans(
  * LSN.
  */
 STATIC int
-xlog_recover_do_efi_trans(
+xlog_recover_efi_pass2(
 	xlog_t			*log,
 	xlog_recover_item_t	*item,
-	xfs_lsn_t		lsn,
-	int			pass)
+	xfs_lsn_t		lsn)
 {
 	int			error;
-	xfs_mount_t		*mp;
+	xfs_mount_t		*mp = log->l_mp;
 	xfs_efi_log_item_t	*efip;
 	xfs_efi_log_format_t	*efi_formatp;
 
-	if (pass == XLOG_RECOVER_PASS1) {
-		return 0;
-	}
-
 	efi_formatp = item->ri_buf[0].i_addr;
 
-	mp = log->l_mp;
 	efip = xfs_efi_init(mp, efi_formatp->efi_nextents);
 	if ((error = xfs_efi_copy_format(&(item->ri_buf[0]),
 					 &(efip->efi_format)))) {
 		xfs_efi_item_free(efip);
 		return error;
 	}
-	efip->efi_next_extent = efi_formatp->efi_nextents;
-	efip->efi_flags |= XFS_EFI_COMMITTED;
+	atomic_set(&efip->efi_next_extent, efi_formatp->efi_nextents);
 
 	spin_lock(&log->l_ailp->xa_lock);
 	/*
 	 * xfs_trans_ail_update() drops the AIL lock.
 	 */
-	xfs_trans_ail_update(log->l_ailp, (xfs_log_item_t *)efip, lsn);
+	xfs_trans_ail_update(log->l_ailp, &efip->efi_item, lsn);
 	return 0;
 }
 
@@ -2737,11 +2584,10 @@ xlog_recover_do_efi_trans(
  * efd format structure.  If we find it, we remove the efi from the
  * AIL and free it.
  */
-STATIC void
-xlog_recover_do_efd_trans(
+STATIC int
+xlog_recover_efd_pass2(
 	xlog_t			*log,
-	xlog_recover_item_t	*item,
-	int			pass)
+	xlog_recover_item_t	*item)
 {
 	xfs_efd_log_format_t	*efd_formatp;
 	xfs_efi_log_item_t	*efip = NULL;
@@ -2750,10 +2596,6 @@ xlog_recover_do_efd_trans(
 	struct xfs_ail_cursor	cur;
 	struct xfs_ail		*ailp = log->l_ailp;
 
-	if (pass == XLOG_RECOVER_PASS1) {
-		return;
-	}
-
 	efd_formatp = item->ri_buf[0].i_addr;
 	ASSERT((item->ri_buf[0].i_len == (sizeof(xfs_efd_log_format_32_t) +
 		((efd_formatp->efd_nextents - 1) * sizeof(xfs_extent_32_t)))) ||
@@ -2785,62 +2627,6 @@ xlog_recover_do_efd_trans(
 	}
 	xfs_trans_ail_cursor_done(ailp, &cur);
 	spin_unlock(&ailp->xa_lock);
-}
-
-/*
- * Perform the transaction
- *
- * If the transaction modifies a buffer or inode, do it now.  Otherwise,
- * EFIs and EFDs get queued up by adding entries into the AIL for them.
- */
-STATIC int
-xlog_recover_do_trans(
-	xlog_t			*log,
-	xlog_recover_t		*trans,
-	int			pass)
-{
-	int			error = 0;
-	xlog_recover_item_t	*item;
-
-	error = xlog_recover_reorder_trans(log, trans, pass);
-	if (error)
-		return error;
-
-	list_for_each_entry(item, &trans->r_itemq, ri_list) {
-		trace_xfs_log_recover_item_recover(log, trans, item, pass);
-		switch (ITEM_TYPE(item)) {
-		case XFS_LI_BUF:
-			error = xlog_recover_do_buffer_trans(log, item, pass);
-			break;
-		case XFS_LI_INODE:
-			error = xlog_recover_do_inode_trans(log, item, pass);
-			break;
-		case XFS_LI_EFI:
-			error = xlog_recover_do_efi_trans(log, item,
-							  trans->r_lsn, pass);
-			break;
-		case XFS_LI_EFD:
-			xlog_recover_do_efd_trans(log, item, pass);
-			error = 0;
-			break;
-		case XFS_LI_DQUOT:
-			error = xlog_recover_do_dquot_trans(log, item, pass);
-			break;
-		case XFS_LI_QUOTAOFF:
-			error = xlog_recover_do_quotaoff_trans(log, item,
-							       pass);
-			break;
-		default:
-			xlog_warn(
-	"XFS: invalid item type (%d) xlog_recover_do_trans", ITEM_TYPE(item));
-			ASSERT(0);
-			error = XFS_ERROR(EIO);
-			break;
-		}
-
-		if (error)
-			return error;
-	}
 
 	return 0;
 }
@@ -2852,7 +2638,7 @@ xlog_recover_do_trans(
  */
 STATIC void
 xlog_recover_free_trans(
-	xlog_recover_t		*trans)
+	struct xlog_recover	*trans)
 {
 	xlog_recover_item_t	*item, *n;
 	int			i;
@@ -2870,18 +2656,96 @@ xlog_recover_free_trans(
 	kmem_free(trans);
 }
 
+STATIC int
+xlog_recover_commit_pass1(
+	struct log		*log,
+	struct xlog_recover	*trans,
+	xlog_recover_item_t	*item)
+{
+	trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS1);
+
+	switch (ITEM_TYPE(item)) {
+	case XFS_LI_BUF:
+		return xlog_recover_buffer_pass1(log, item);
+	case XFS_LI_QUOTAOFF:
+		return xlog_recover_quotaoff_pass1(log, item);
+	case XFS_LI_INODE:
+	case XFS_LI_EFI:
+	case XFS_LI_EFD:
+	case XFS_LI_DQUOT:
+		/* nothing to do in pass 1 */
+		return 0;
+	default:
+		xlog_warn(
+	"XFS: invalid item type (%d) xlog_recover_commit_pass1",
+			ITEM_TYPE(item));
+		ASSERT(0);
+		return XFS_ERROR(EIO);
+	}
+}
+
+STATIC int
+xlog_recover_commit_pass2(
+	struct log		*log,
+	struct xlog_recover	*trans,
+	xlog_recover_item_t	*item)
+{
+	trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS2);
+
+	switch (ITEM_TYPE(item)) {
+	case XFS_LI_BUF:
+		return xlog_recover_buffer_pass2(log, item);
+	case XFS_LI_INODE:
+		return xlog_recover_inode_pass2(log, item);
+	case XFS_LI_EFI:
+		return xlog_recover_efi_pass2(log, item, trans->r_lsn);
+	case XFS_LI_EFD:
+		return xlog_recover_efd_pass2(log, item);
+	case XFS_LI_DQUOT:
+		return xlog_recover_dquot_pass2(log, item);
+	case XFS_LI_QUOTAOFF:
+		/* nothing to do in pass2 */
+		return 0;
+	default:
+		xlog_warn(
+	"XFS: invalid item type (%d) xlog_recover_commit_pass2",
+			ITEM_TYPE(item));
+		ASSERT(0);
+		return XFS_ERROR(EIO);
+	}
+}
+
+/*
+ * Perform the transaction.
+ *
+ * If the transaction modifies a buffer or inode, do it now.  Otherwise,
+ * EFIs and EFDs get queued up by adding entries into the AIL for them.
+ */
 STATIC int
 xlog_recover_commit_trans(
-	xlog_t			*log,
-	xlog_recover_t		*trans,
+	struct log		*log,
+	struct xlog_recover	*trans,
 	int			pass)
 {
-	int			error;
+	int			error = 0;
+	xlog_recover_item_t	*item;
 
 	hlist_del(&trans->r_list);
-	if ((error = xlog_recover_do_trans(log, trans, pass)))
+
+	error = xlog_recover_reorder_trans(log, trans, pass);
+	if (error)
 		return error;
-	xlog_recover_free_trans(trans);			/* no error */
+
+	list_for_each_entry(item, &trans->r_itemq, ri_list) {
+		if (pass == XLOG_RECOVER_PASS1)
+			error = xlog_recover_commit_pass1(log, trans, item);
+		else
+			error = xlog_recover_commit_pass2(log, trans, item);
+		if (error)
+			return error;
+	}
+
+	xlog_recover_free_trans(trans);
 	return 0;
 }
 
@@ -3011,7 +2875,7 @@ xlog_recover_process_efi(
 	xfs_extent_t		*extp;
 	xfs_fsblock_t		startblock_fsb;
 
-	ASSERT(!(efip->efi_flags & XFS_EFI_RECOVERED));
+	ASSERT(!test_bit(XFS_EFI_RECOVERED, &efip->efi_flags));
 
 	/*
 	 * First check the validity of the extents described by the
@@ -3050,7 +2914,7 @@ xlog_recover_process_efi(
 					 extp->ext_len);
 	}
 
-	efip->efi_flags |= XFS_EFI_RECOVERED;
+	set_bit(XFS_EFI_RECOVERED, &efip->efi_flags);
 	error = xfs_trans_commit(tp, 0);
 	return error;
 
@@ -3107,7 +2971,7 @@ xlog_recover_process_efis(
 		 * Skip EFIs that we've already processed.
 		 */
 		efip = (xfs_efi_log_item_t *)lip;
-		if (efip->efi_flags & XFS_EFI_RECOVERED) {
+		if (test_bit(XFS_EFI_RECOVERED, &efip->efi_flags)) {
 			lip = xfs_trans_ail_cursor_next(ailp, &cur);
 			continue;
 		}
@@ -3724,7 +3588,7 @@ xlog_do_log_recovery(
 	xfs_daddr_t	head_blk,
 	xfs_daddr_t	tail_blk)
 {
-	int		error;
+	int		error, i;
 
 	ASSERT(head_blk != tail_blk);
 
@@ -3732,10 +3596,12 @@ xlog_do_log_recovery(
 	 * First do a pass to find all of the cancelled buf log items.
 	 * Store them in the buf_cancel_table for use in the second pass.
 	 */
-	log->l_buf_cancel_table =
-		(xfs_buf_cancel_t **)kmem_zalloc(XLOG_BC_TABLE_SIZE *
-						 sizeof(xfs_buf_cancel_t*),
+	log->l_buf_cancel_table = kmem_zalloc(XLOG_BC_TABLE_SIZE *
+						 sizeof(struct list_head),
 						 KM_SLEEP);
+	for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
+		INIT_LIST_HEAD(&log->l_buf_cancel_table[i]);
+
 	error = xlog_do_recovery_pass(log, head_blk, tail_blk,
 				      XLOG_RECOVER_PASS1);
 	if (error != 0) {
@@ -3754,7 +3620,7 @@ xlog_do_log_recovery(
 		int	i;
 
 		for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
-			ASSERT(log->l_buf_cancel_table[i] == NULL);
+			ASSERT(list_empty(&log->l_buf_cancel_table[i]));
 	}
 #endif	/* DEBUG */
 

fs/xfs/xfs_mount.c

@@ -472,7 +472,7 @@ xfs_initialize_perag(
 			goto out_unwind;
 		pag->pag_agno = index;
 		pag->pag_mount = mp;
-		rwlock_init(&pag->pag_ici_lock);
+		spin_lock_init(&pag->pag_ici_lock);
 		mutex_init(&pag->pag_ici_reclaim_lock);
 		INIT_RADIX_TREE(&pag->pag_ici_root, GFP_ATOMIC);
 		spin_lock_init(&pag->pag_buf_lock);
@@ -974,6 +974,24 @@ xfs_set_rw_sizes(xfs_mount_t *mp)
 	mp->m_writeio_blocks = 1 << (mp->m_writeio_log - sbp->sb_blocklog);
 }
 
+/*
+ * precalculate the low space thresholds for dynamic speculative preallocation.
+ */
+void
+xfs_set_low_space_thresholds(
+	struct xfs_mount	*mp)
+{
+	int i;
+
+	for (i = 0; i < XFS_LOWSP_MAX; i++) {
+		__uint64_t space = mp->m_sb.sb_dblocks;
+
+		do_div(space, 100);
+		mp->m_low_space[i] = space * (i + 1);
+	}
+}
+
+
 /*
  * Set whether we're using inode alignment.
  */
@@ -1196,6 +1214,9 @@ xfs_mountfs(
 	 */
 	xfs_set_rw_sizes(mp);
 
+	/* set the low space thresholds for dynamic preallocation */
+	xfs_set_low_space_thresholds(mp);
+
 	/*
 	 * Set the inode cluster size.
 	 * This may still be overridden by the file system

fs/xfs/xfs_mount.h

@@ -103,6 +103,16 @@ extern int	xfs_icsb_modify_counters(struct xfs_mount *, xfs_sb_field_t,
 	xfs_mod_incore_sb(mp, field, delta, rsvd)
 #endif
 
+/* dynamic preallocation free space thresholds, 5% down to 1% */
+enum {
+	XFS_LOWSP_1_PCNT = 0,
+	XFS_LOWSP_2_PCNT,
+	XFS_LOWSP_3_PCNT,
+	XFS_LOWSP_4_PCNT,
+	XFS_LOWSP_5_PCNT,
+	XFS_LOWSP_MAX,
+};
+
 typedef struct xfs_mount {
 	struct super_block	*m_super;
 	xfs_tid_t		m_tid;		/* next unused tid for fs */
@@ -202,6 +212,8 @@ typedef struct xfs_mount {
 	__int64_t		m_update_flags;	/* sb flags we need to update
 						   on the next remount,rw */
 	struct shrinker		m_inode_shrink;	/* inode reclaim shrinker */
+	int64_t			m_low_space[XFS_LOWSP_MAX];
+						/* low free space thresholds */
 } xfs_mount_t;
 
 /*
@@ -379,6 +391,8 @@ extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
 
 extern int	xfs_dev_is_read_only(struct xfs_mount *, char *);
 
+extern void	xfs_set_low_space_thresholds(struct xfs_mount *);
+
 #endif	/* __KERNEL__ */
 
 extern void	xfs_mod_sb(struct xfs_trans *, __int64_t);

fs/xfs/xfs_trans.c

@@ -1350,7 +1350,7 @@ xfs_trans_fill_vecs(
  * they could be immediately flushed and we'd have to race with the flusher
  * trying to pull the item from the AIL as we add it.
  */
-void
+static void
 xfs_trans_item_committed(
 	struct xfs_log_item	*lip,
 	xfs_lsn_t		commit_lsn,
@@ -1425,6 +1425,83 @@ xfs_trans_committed(
 	xfs_trans_free(tp);
 }
 
+static inline void
+xfs_log_item_batch_insert(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	**log_items,
+	int			nr_items,
+	xfs_lsn_t		commit_lsn)
+{
+	int	i;
+
+	spin_lock(&ailp->xa_lock);
+	/* xfs_trans_ail_update_bulk drops ailp->xa_lock */
+	xfs_trans_ail_update_bulk(ailp, log_items, nr_items, commit_lsn);
+
+	for (i = 0; i < nr_items; i++)
+		IOP_UNPIN(log_items[i], 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.
+ */
+void
+xfs_trans_committed_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_log_vec	*log_vector,
+	xfs_lsn_t		commit_lsn,
+	int			aborted)
+{
+#define LOG_ITEM_BATCH_SIZE	32
+	struct xfs_log_item	*log_items[LOG_ITEM_BATCH_SIZE];
+	struct xfs_log_vec	*lv;
+	int			i = 0;
+
+	/* unpin all the log items */
+	for (lv = log_vector; lv; lv = lv->lv_next ) {
+		struct xfs_log_item	*lip = lv->lv_item;
+		xfs_lsn_t		item_lsn;
+
+		if (aborted)
+			lip->li_flags |= XFS_LI_ABORTED;
+		item_lsn = IOP_COMMITTED(lip, commit_lsn);
+
+		/* item_lsn of -1 means the item was freed */
+		if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
+			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.
+			 */
+			spin_lock(&ailp->xa_lock);
+			if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0)
+				xfs_trans_ail_update(ailp, lip, item_lsn);
+			else
+				spin_unlock(&ailp->xa_lock);
+			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, log_items,
+					LOG_ITEM_BATCH_SIZE, commit_lsn);
+			i = 0;
+		}
+	}
+
+	/* make sure we insert the remainder! */
+	if (i)
+		xfs_log_item_batch_insert(ailp, log_items, i, commit_lsn);
+}
+
 /*
  * Called from the trans_commit code when we notice that
  * the filesystem is in the middle of a forced shutdown.

fs/xfs/xfs_trans.h

@@ -294,8 +294,8 @@ struct xfs_log_item_desc {
 #define	XFS_ALLOC_BTREE_REF	2
 #define	XFS_BMAP_BTREE_REF	2
 #define	XFS_DIR_BTREE_REF	2
+#define	XFS_INO_REF		2
 #define	XFS_ATTR_BTREE_REF	1
-#define	XFS_INO_REF		1
 #define	XFS_DQUOT_REF		1
 
 #ifdef __KERNEL__

fs/xfs/xfs_trans_ail.c

@@ -28,8 +28,8 @@
 #include "xfs_trans_priv.h"
 #include "xfs_error.h"
 
-STATIC void xfs_ail_insert(struct xfs_ail *, xfs_log_item_t *);
-STATIC xfs_log_item_t * xfs_ail_delete(struct xfs_ail *, xfs_log_item_t *);
+STATIC void xfs_ail_splice(struct xfs_ail *, struct list_head *, xfs_lsn_t);
+STATIC void xfs_ail_delete(struct xfs_ail *, xfs_log_item_t *);
 STATIC xfs_log_item_t * xfs_ail_min(struct xfs_ail *);
 STATIC xfs_log_item_t * xfs_ail_next(struct xfs_ail *, xfs_log_item_t *);
 
@@ -449,129 +449,152 @@ xfs_trans_unlocked_item(
 		xfs_log_move_tail(ailp->xa_mount, 1);
 }	/* xfs_trans_unlocked_item */
 
-
 /*
- * Update the position of the item in the AIL with the new
- * lsn.  If it is not yet in the AIL, add it.  Otherwise, move
- * it to its new position by removing it and re-adding it.
+ * xfs_trans_ail_update - bulk AIL insertion operation.
+ *
+ * @xfs_trans_ail_update takes an array of log items that all need to be
+ * positioned at the same LSN in the AIL. If an item is not in the AIL, it will
+ * be added.  Otherwise, it will be repositioned  by removing it and re-adding
+ * it to the AIL. If we move the first item in the AIL, update the log tail to
+ * match the new minimum LSN in the AIL.
  *
- * Wakeup anyone with an lsn less than the item's lsn.  If the item
- * we move in the AIL is the minimum one, update the tail lsn in the
- * log manager.
+ * This function takes the AIL lock once to execute the update operations on
+ * all the items in the array, and as such should not be called with the AIL
+ * lock held. As a result, once we have the AIL lock, we need to check each log
+ * item LSN to confirm it needs to be moved forward in the AIL.
  *
- * This function must be called with the AIL lock held.  The lock
- * is dropped before returning.
+ * To optimise the insert operation, we delete all the items from the AIL in
+ * the first pass, moving them into a temporary list, then splice the temporary
+ * list into the correct position in the AIL. This avoids needing to do an
+ * insert operation on every item.
+ *
+ * This function must be called with the AIL lock held.  The lock is dropped
+ * before returning.
  */
 void
-xfs_trans_ail_update(
-	struct xfs_ail	*ailp,
-	xfs_log_item_t	*lip,
-	xfs_lsn_t	lsn) __releases(ailp->xa_lock)
+xfs_trans_ail_update_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	**log_items,
+	int			nr_items,
+	xfs_lsn_t		lsn) __releases(ailp->xa_lock)
 {
-	xfs_log_item_t		*dlip = NULL;
-	xfs_log_item_t		*mlip;	/* ptr to minimum lip */
+	xfs_log_item_t		*mlip;
 	xfs_lsn_t		tail_lsn;
+	int			mlip_changed = 0;
+	int			i;
+	LIST_HEAD(tmp);
 
 	mlip = xfs_ail_min(ailp);
 
-	if (lip->li_flags & XFS_LI_IN_AIL) {
-		dlip = xfs_ail_delete(ailp, lip);
-		ASSERT(dlip == lip);
-		xfs_trans_ail_cursor_clear(ailp, dlip);
-	} else {
-		lip->li_flags |= XFS_LI_IN_AIL;
+	for (i = 0; i < nr_items; i++) {
+		struct xfs_log_item *lip = log_items[i];
+		if (lip->li_flags & XFS_LI_IN_AIL) {
+			/* check if we really need to move the item */
+			if (XFS_LSN_CMP(lsn, lip->li_lsn) <= 0)
+				continue;
+
+			xfs_ail_delete(ailp, lip);
+			if (mlip == lip)
+				mlip_changed = 1;
+		} else {
+			lip->li_flags |= XFS_LI_IN_AIL;
+		}
+		lip->li_lsn = lsn;
+		list_add(&lip->li_ail, &tmp);
 	}
 
-	lip->li_lsn = lsn;
-	xfs_ail_insert(ailp, lip);
+	xfs_ail_splice(ailp, &tmp, lsn);
 
-	if (mlip == dlip) {
-		mlip = xfs_ail_min(ailp);
-		/*
-		 * It is not safe to access mlip after the AIL lock is
-		 * dropped, so we must get a copy of li_lsn before we do
-		 * so.  This is especially important on 32-bit platforms
-		 * where accessing and updating 64-bit values like li_lsn
-		 * is not atomic.
-		 */
-		tail_lsn = mlip->li_lsn;
-		spin_unlock(&ailp->xa_lock);
-		xfs_log_move_tail(ailp->xa_mount, tail_lsn);
-	} else {
+	if (!mlip_changed) {
 		spin_unlock(&ailp->xa_lock);
+		return;
 	}
 
-
-}	/* xfs_trans_update_ail */
+	/*
+	 * It is not safe to access mlip after the AIL lock is dropped, so we
+	 * must get a copy of li_lsn before we do so.  This is especially
+	 * important on 32-bit platforms where accessing and updating 64-bit
+	 * values like li_lsn is not atomic.
+	 */
+	mlip = xfs_ail_min(ailp);
+	tail_lsn = mlip->li_lsn;
+	spin_unlock(&ailp->xa_lock);
+	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
+}
 
 /*
- * Delete the given item from the AIL.  It must already be in
- * the AIL.
+ * xfs_trans_ail_delete_bulk - remove multiple log items from the AIL
  *
- * Wakeup anyone with an lsn less than item's lsn.    If the item
- * we delete in the AIL is the minimum one, update the tail lsn in the
- * log manager.
+ * @xfs_trans_ail_delete_bulk takes an array of log items that all need to
+ * removed from the AIL. The caller is already holding the AIL lock, and done
+ * all the checks necessary to ensure the items passed in via @log_items are
+ * ready for deletion. This includes checking that the items are in the AIL.
  *
- * Clear the IN_AIL flag from the item, reset its lsn to 0, and
- * bump the AIL's generation count to indicate that the tree
- * has changed.
+ * For each log item to be removed, unlink it  from the AIL, clear the IN_AIL
+ * flag from the item and reset the item's lsn to 0. If we remove the first
+ * item in the AIL, update the log tail to match the new minimum LSN in the
+ * AIL.
  *
- * This function must be called with the AIL lock held.  The lock
- * is dropped before returning.
+ * This function will not drop the AIL lock until all items are removed from
+ * the AIL to minimise the amount of lock traffic on the AIL. This does not
+ * greatly increase the AIL hold time, but does significantly reduce the amount
+ * of traffic on the lock, especially during IO completion.
+ *
+ * This function must be called with the AIL lock held.  The lock is dropped
+ * before returning.
  */
 void
-xfs_trans_ail_delete(
-	struct xfs_ail	*ailp,
-	xfs_log_item_t	*lip) __releases(ailp->xa_lock)
+xfs_trans_ail_delete_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	**log_items,
+	int			nr_items) __releases(ailp->xa_lock)
 {
-	xfs_log_item_t		*dlip;
 	xfs_log_item_t		*mlip;
 	xfs_lsn_t		tail_lsn;
+	int			mlip_changed = 0;
+	int			i;
 
-	if (lip->li_flags & XFS_LI_IN_AIL) {
-		mlip = xfs_ail_min(ailp);
-		dlip = xfs_ail_delete(ailp, lip);
-		ASSERT(dlip == lip);
-		xfs_trans_ail_cursor_clear(ailp, dlip);
-
+	mlip = xfs_ail_min(ailp);
 
-		lip->li_flags &= ~XFS_LI_IN_AIL;
-		lip->li_lsn = 0;
+	for (i = 0; i < nr_items; i++) {
+		struct xfs_log_item *lip = log_items[i];
+		if (!(lip->li_flags & XFS_LI_IN_AIL)) {
+			struct xfs_mount	*mp = ailp->xa_mount;
 
-		if (mlip == dlip) {
-			mlip = xfs_ail_min(ailp);
-			/*
-			 * It is not safe to access mlip after the AIL lock
-			 * is dropped, so we must get a copy of li_lsn
-			 * before we do so.  This is especially important
-			 * on 32-bit platforms where accessing and updating
-			 * 64-bit values like li_lsn is not atomic.
-			 */
-			tail_lsn = mlip ? mlip->li_lsn : 0;
-			spin_unlock(&ailp->xa_lock);
-			xfs_log_move_tail(ailp->xa_mount, tail_lsn);
-		} else {
 			spin_unlock(&ailp->xa_lock);
+			if (!XFS_FORCED_SHUTDOWN(mp)) {
+				xfs_cmn_err(XFS_PTAG_AILDELETE, CE_ALERT, mp,
+		"%s: attempting to delete a log item that is not in the AIL",
+						__func__);
+				xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+			}
+			return;
 		}
+
+		xfs_ail_delete(ailp, lip);
+		lip->li_flags &= ~XFS_LI_IN_AIL;
+		lip->li_lsn = 0;
+		if (mlip == lip)
+			mlip_changed = 1;
 	}
-	else {
-		/*
-		 * If the file system is not being shutdown, we are in
-		 * serious trouble if we get to this stage.
-		 */
-		struct xfs_mount	*mp = ailp->xa_mount;
 
+	if (!mlip_changed) {
 		spin_unlock(&ailp->xa_lock);
-		if (!XFS_FORCED_SHUTDOWN(mp)) {
-			xfs_cmn_err(XFS_PTAG_AILDELETE, CE_ALERT, mp,
-		"%s: attempting to delete a log item that is not in the AIL",
-					__func__);
-			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
-		}
+		return;
 	}
-}
-
 
+	/*
+	 * It is not safe to access mlip after the AIL lock is dropped, so we
+	 * must get a copy of li_lsn before we do so.  This is especially
+	 * important on 32-bit platforms where accessing and updating 64-bit
+	 * values like li_lsn is not atomic. It is possible we've emptied the
+	 * AIL here, so if that is the case, pass an LSN of 0 to the tail move.
+	 */
+	mlip = xfs_ail_min(ailp);
+	tail_lsn = mlip ? mlip->li_lsn : 0;
+	spin_unlock(&ailp->xa_lock);
+	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
+}
 
 /*
  * The active item list (AIL) is a doubly linked list of log
@@ -623,16 +646,13 @@ xfs_trans_ail_destroy(
 }
 
 /*
- * Insert the given log item into the AIL.
- * We almost always insert at the end of the list, so on inserts
- * we search from the end of the list to find where the
- * new item belongs.
+ * splice the log item list into the AIL at the given LSN.
  */
 STATIC void
-xfs_ail_insert(
+xfs_ail_splice(
 	struct xfs_ail	*ailp,
-	xfs_log_item_t	*lip)
-/* ARGSUSED */
+	struct list_head *list,
+	xfs_lsn_t	lsn)
 {
 	xfs_log_item_t	*next_lip;
 
@@ -640,39 +660,33 @@ xfs_ail_insert(
 	 * If the list is empty, just insert the item.
 	 */
 	if (list_empty(&ailp->xa_ail)) {
-		list_add(&lip->li_ail, &ailp->xa_ail);
+		list_splice(list, &ailp->xa_ail);
 		return;
 	}
 
 	list_for_each_entry_reverse(next_lip, &ailp->xa_ail, li_ail) {
-		if (XFS_LSN_CMP(next_lip->li_lsn, lip->li_lsn) <= 0)
+		if (XFS_LSN_CMP(next_lip->li_lsn, lsn) <= 0)
 			break;
 	}
 
 	ASSERT((&next_lip->li_ail == &ailp->xa_ail) ||
-	       (XFS_LSN_CMP(next_lip->li_lsn, lip->li_lsn) <= 0));
-
-	list_add(&lip->li_ail, &next_lip->li_ail);
+	       (XFS_LSN_CMP(next_lip->li_lsn, lsn) <= 0));
 
-	xfs_ail_check(ailp, lip);
+	list_splice_init(list, &next_lip->li_ail);
 	return;
 }
 
 /*
  * Delete the given item from the AIL.  Return a pointer to the item.
  */
-/*ARGSUSED*/
-STATIC xfs_log_item_t *
+STATIC void
 xfs_ail_delete(
 	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
-/* ARGSUSED */
 {
 	xfs_ail_check(ailp, lip);
-
 	list_del(&lip->li_ail);
-
-	return lip;
+	xfs_trans_ail_cursor_clear(ailp, lip);
 }
 
 /*
@@ -682,7 +696,6 @@ xfs_ail_delete(
 STATIC xfs_log_item_t *
 xfs_ail_min(
 	struct xfs_ail	*ailp)
-/* ARGSUSED */
 {
 	if (list_empty(&ailp->xa_ail))
 		return NULL;
@@ -699,7 +712,6 @@ STATIC xfs_log_item_t *
 xfs_ail_next(
 	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
-/* ARGSUSED */
 {
 	if (lip->li_ail.next == &ailp->xa_ail)
 		return NULL;

fs/xfs/xfs_trans_extfree.c

@@ -69,12 +69,16 @@ xfs_trans_log_efi_extent(xfs_trans_t		*tp,
 	tp->t_flags |= XFS_TRANS_DIRTY;
 	efip->efi_item.li_desc->lid_flags |= XFS_LID_DIRTY;
 
-	next_extent = efip->efi_next_extent;
+	/*
+	 * atomic_inc_return gives us the value after the increment;
+	 * we want to use it as an array index so we need to subtract 1 from
+	 * it.
+	 */
+	next_extent = atomic_inc_return(&efip->efi_next_extent) - 1;
 	ASSERT(next_extent < efip->efi_format.efi_nextents);
 	extp = &(efip->efi_format.efi_extents[next_extent]);
 	extp->ext_start = start_block;
 	extp->ext_len = ext_len;
-	efip->efi_next_extent++;
 }
 
 

fs/xfs/xfs_trans_priv.h

@@ -22,15 +22,17 @@ struct xfs_log_item;
 struct xfs_log_item_desc;
 struct xfs_mount;
 struct xfs_trans;
+struct xfs_ail;
+struct xfs_log_vec;
 
 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);
 
+void	xfs_trans_committed_bulk(struct xfs_ail *ailp, struct xfs_log_vec *lv,
+				xfs_lsn_t commit_lsn, int aborted);
 /*
  * AIL traversal cursor.
  *
@@ -73,12 +75,29 @@ struct xfs_ail {
 /*
  * From xfs_trans_ail.c
  */
-void			xfs_trans_ail_update(struct xfs_ail *ailp,
-					struct xfs_log_item *lip, xfs_lsn_t lsn)
-					__releases(ailp->xa_lock);
-void			xfs_trans_ail_delete(struct xfs_ail *ailp,
-					struct xfs_log_item *lip)
-					__releases(ailp->xa_lock);
+void	xfs_trans_ail_update_bulk(struct xfs_ail *ailp,
+				struct xfs_log_item **log_items, int nr_items,
+				xfs_lsn_t lsn) __releases(ailp->xa_lock);
+static inline void
+xfs_trans_ail_update(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn) __releases(ailp->xa_lock)
+{
+	xfs_trans_ail_update_bulk(ailp, &lip, 1, lsn);
+}
+
+void	xfs_trans_ail_delete_bulk(struct xfs_ail *ailp,
+				struct xfs_log_item **log_items, int nr_items)
+				__releases(ailp->xa_lock);
+static inline void
+xfs_trans_ail_delete(
+	struct xfs_ail	*ailp,
+	xfs_log_item_t	*lip) __releases(ailp->xa_lock)
+{
+	xfs_trans_ail_delete_bulk(ailp, &lip, 1);
+}
+
 void			xfs_trans_ail_push(struct xfs_ail *, xfs_lsn_t);
 void			xfs_trans_unlocked_item(struct xfs_ail *,
 					xfs_log_item_t *);

fs/xfs/xfs_vnodeops.c

@@ -964,29 +964,48 @@ xfs_release(
 			xfs_flush_pages(ip, 0, -1, XBF_ASYNC, FI_NONE);
 	}
 
-	if (ip->i_d.di_nlink != 0) {
-		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
-		     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
-		       ip->i_delayed_blks > 0)) &&
-		     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
-		    (!(ip->i_d.di_flags &
-				(XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
+	if (ip->i_d.di_nlink == 0)
+		return 0;
 
-			/*
-			 * If we can't get the iolock just skip truncating
-			 * the blocks past EOF because we could deadlock
-			 * with the mmap_sem otherwise.  We'll get another
-			 * chance to drop them once the last reference to
-			 * the inode is dropped, so we'll never leak blocks
-			 * permanently.
-			 */
-			error = xfs_free_eofblocks(mp, ip,
-						   XFS_FREE_EOF_TRYLOCK);
-			if (error)
-				return error;
-		}
-	}
+	if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
+	     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
+	       ip->i_delayed_blks > 0)) &&
+	     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
+	    (!(ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
 
+		/*
+		 * If we can't get the iolock just skip truncating the blocks
+		 * past EOF because we could deadlock with the mmap_sem
+		 * otherwise.  We'll get another chance to drop them once the
+		 * last reference to the inode is dropped, so we'll never leak
+		 * blocks permanently.
+		 *
+		 * Further, check if the inode is being opened, written and
+		 * closed frequently and we have delayed allocation blocks
+		 * oustanding (e.g. streaming writes from the NFS server),
+		 * truncating the blocks past EOF will cause fragmentation to
+		 * occur.
+		 *
+		 * In this case don't do the truncation, either, but we have to
+		 * be careful how we detect this case. Blocks beyond EOF show
+		 * up as i_delayed_blks even when the inode is clean, so we
+		 * need to truncate them away first before checking for a dirty
+		 * release. Hence on the first dirty close we will still remove
+		 * the speculative allocation, but after that we will leave it
+		 * in place.
+		 */
+		if (xfs_iflags_test(ip, XFS_IDIRTY_RELEASE))
+			return 0;
+
+		error = xfs_free_eofblocks(mp, ip,
+					   XFS_FREE_EOF_TRYLOCK);
+		if (error)
+			return error;
+
+		/* delalloc blocks after truncation means it really is dirty */
+		if (ip->i_delayed_blks)
+			xfs_iflags_set(ip, XFS_IDIRTY_RELEASE);
+	}
 	return 0;
 }