Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6:
  jbd: change the field "b_cow_tid" of struct journal_head from type unsigned to tid_t
  ext3.txt: update the links in the section "useful links" to the latest ones
  ext3: Fix data corruption in inodes with journalled data
  ext2: check xattr name_len before acquiring xattr_sem in ext2_xattr_get
  ext3: Fix compilation with -DDX_DEBUG
  quota: Remove unused declaration
  jbd: Use WRITE_SYNC in journal checkpoint.
  jbd: Fix oops in journal_remove_journal_head()
  ext3: Return -EINVAL when start is beyond the end of fs in ext3_trim_fs()
  ext3/ioctl.c: silence sparse warnings about different address spaces
  ext3/ext4 Documentation: remove bh/nobh since it has been deprecated
  ext3: Improve truncate error handling
  ext3: use proper little-endian bitops
  ext2: include fs.h into ext2_fs.h
  ext3: Fix oops in ext3_try_to_allocate_with_rsv()
  jbd: fix a bug of leaking jh->b_jcount
  jbd: remove dependency on __GFP_NOFAIL
  ext3: Convert ext3 to new truncate calling convention
  jbd: Add fixed tracepoints
  ext3: Add fixed tracepoints

Resolve conflicts in fs/ext3/fsync.c due to fsync locking push-down and
new fixed tracepoints.

14 files changed, 339 insertions, 230 deletions

diff --git a/fs/ext2/xattr.c b/fs/ext2/xattr.c
index 5299706..d27b71f 100644
--- a/fs/ext2/xattr.c
+++ b/fs/ext2/xattr.c
@@ -161,6 +161,10 @@ ext2_xattr_get(struct inode *inode, int name_index, const char *name,
 
 	if (name == NULL)
 		return -EINVAL;
+	name_len = strlen(name);
+	if (name_len > 255)
+		return -ERANGE;
+
 	down_read(&EXT2_I(inode)->xattr_sem);
 	error = -ENODATA;
 	if (!EXT2_I(inode)->i_file_acl)
@@ -181,12 +185,8 @@ bad_block:	ext2_error(inode->i_sb, "ext2_xattr_get",
 		error = -EIO;
 		goto cleanup;
 	}
-	/* find named attribute */
-	name_len = strlen(name);
 
-	error = -ERANGE;
-	if (name_len > 255)
-		goto cleanup;
+	/* find named attribute */
 	entry = FIRST_ENTRY(bh);
 	while (!IS_LAST_ENTRY(entry)) {
 		struct ext2_xattr_entry *next =
diff --git a/fs/ext3/balloc.c b/fs/ext3/balloc.c
index fe52297..6386d76 100644
--- a/fs/ext3/balloc.c
+++ b/fs/ext3/balloc.c
@@ -21,6 +21,7 @@
 #include <linux/quotaops.h>
 #include <linux/buffer_head.h>
 #include <linux/blkdev.h>
+#include <trace/events/ext3.h>
 
 /*
  * balloc.c contains the blocks allocation and deallocation routines
@@ -161,6 +162,7 @@ read_block_bitmap(struct super_block *sb, unsigned int block_group)
 	desc = ext3_get_group_desc(sb, block_group, NULL);
 	if (!desc)
 		return NULL;
+	trace_ext3_read_block_bitmap(sb, block_group);
 	bitmap_blk = le32_to_cpu(desc->bg_block_bitmap);
 	bh = sb_getblk(sb, bitmap_blk);
 	if (unlikely(!bh)) {
@@ -351,6 +353,7 @@ void ext3_rsv_window_add(struct super_block *sb,
 	struct rb_node * parent = NULL;
 	struct ext3_reserve_window_node *this;
 
+	trace_ext3_rsv_window_add(sb, rsv);
 	while (*p)
 	{
 		parent = *p;
@@ -476,8 +479,10 @@ void ext3_discard_reservation(struct inode *inode)
 	rsv = &block_i->rsv_window_node;
 	if (!rsv_is_empty(&rsv->rsv_window)) {
 		spin_lock(rsv_lock);
-		if (!rsv_is_empty(&rsv->rsv_window))
+		if (!rsv_is_empty(&rsv->rsv_window)) {
+			trace_ext3_discard_reservation(inode, rsv);
 			rsv_window_remove(inode->i_sb, rsv);
+		}
 		spin_unlock(rsv_lock);
 	}
 }
@@ -683,14 +688,10 @@ error_return:
 void ext3_free_blocks(handle_t *handle, struct inode *inode,
 			ext3_fsblk_t block, unsigned long count)
 {
-	struct super_block * sb;
+	struct super_block *sb = inode->i_sb;
 	unsigned long dquot_freed_blocks;
 
-	sb = inode->i_sb;
-	if (!sb) {
-		printk ("ext3_free_blocks: nonexistent device");
-		return;
-	}
+	trace_ext3_free_blocks(inode, block, count);
 	ext3_free_blocks_sb(handle, sb, block, count, &dquot_freed_blocks);
 	if (dquot_freed_blocks)
 		dquot_free_block(inode, dquot_freed_blocks);
@@ -1136,6 +1137,7 @@ static int alloc_new_reservation(struct ext3_reserve_window_node *my_rsv,
 	else
 		start_block = grp_goal + group_first_block;
 
+	trace_ext3_alloc_new_reservation(sb, start_block);
 	size = my_rsv->rsv_goal_size;
 
 	if (!rsv_is_empty(&my_rsv->rsv_window)) {
@@ -1230,8 +1232,11 @@ retry:
 	 * check if the first free block is within the
 	 * free space we just reserved
 	 */
-	if (start_block >= my_rsv->rsv_start && start_block <= my_rsv->rsv_end)
+	if (start_block >= my_rsv->rsv_start &&
+	    start_block <= my_rsv->rsv_end) {
+		trace_ext3_reserved(sb, start_block, my_rsv);
 		return 0;		/* success */
+	}
 	/*
 	 * if the first free bit we found is out of the reservable space
 	 * continue search for next reservable space,
@@ -1514,10 +1519,6 @@ ext3_fsblk_t ext3_new_blocks(handle_t *handle, struct inode *inode,
 
 	*errp = -ENOSPC;
 	sb = inode->i_sb;
-	if (!sb) {
-		printk("ext3_new_block: nonexistent device");
-		return 0;
-	}
 
 	/*
 	 * Check quota for allocation of this block.
@@ -1528,8 +1529,10 @@ ext3_fsblk_t ext3_new_blocks(handle_t *handle, struct inode *inode,
 		return 0;
 	}
 
+	trace_ext3_request_blocks(inode, goal, num);
+
 	sbi = EXT3_SB(sb);
-	es = EXT3_SB(sb)->s_es;
+	es = sbi->s_es;
 	ext3_debug("goal=%lu.\n", goal);
 	/*
 	 * Allocate a block from reservation only when
@@ -1742,6 +1745,10 @@ allocated:
 	brelse(bitmap_bh);
 	dquot_free_block(inode, *count-num);
 	*count = num;
+
+	trace_ext3_allocate_blocks(inode, goal, num,
+				   (unsigned long long)ret_block);
+
 	return ret_block;
 
 io_error:
@@ -1996,6 +2003,7 @@ ext3_grpblk_t ext3_trim_all_free(struct super_block *sb, unsigned int group,
 		if ((next - start) < minblocks)
 			goto free_extent;
 
+		trace_ext3_discard_blocks(sb, discard_block, next - start);
 		 /* Send the TRIM command down to the device */
 		err = sb_issue_discard(sb, discard_block, next - start,
 				       GFP_NOFS, 0);
@@ -2100,7 +2108,7 @@ int ext3_trim_fs(struct super_block *sb, struct fstrim_range *range)
 	if (unlikely(minlen > EXT3_BLOCKS_PER_GROUP(sb)))
 		return -EINVAL;
 	if (start >= max_blks)
-		goto out;
+		return -EINVAL;
 	if (start + len > max_blks)
 		len = max_blks - start;
 
@@ -2148,8 +2156,6 @@ int ext3_trim_fs(struct super_block *sb, struct fstrim_range *range)
 
 	if (ret >= 0)
 		ret = 0;
-
-out:
 	range->len = trimmed * sb->s_blocksize;
 
 	return ret;
diff --git a/fs/ext3/file.c b/fs/ext3/file.c
index 2be5b99..724df69 100644
--- a/fs/ext3/file.c
+++ b/fs/ext3/file.c
@@ -71,7 +71,6 @@ const struct file_operations ext3_file_operations = {
 };
 
 const struct inode_operations ext3_file_inode_operations = {
-	.truncate	= ext3_truncate,
 	.setattr	= ext3_setattr,
 #ifdef CONFIG_EXT3_FS_XATTR
 	.setxattr	= generic_setxattr,
diff --git a/fs/ext3/fsync.c b/fs/ext3/fsync.c
index 0bcf63a..d494c55 100644
--- a/fs/ext3/fsync.c
+++ b/fs/ext3/fsync.c
@@ -30,6 +30,7 @@
 #include <linux/jbd.h>
 #include <linux/ext3_fs.h>
 #include <linux/ext3_jbd.h>
+#include <trace/events/ext3.h>
 
 /*
  * akpm: A new design for ext3_sync_file().
@@ -51,12 +52,14 @@ int ext3_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	int ret, needs_barrier = 0;
 	tid_t commit_tid;
 
+	trace_ext3_sync_file_enter(file, datasync);
+
 	if (inode->i_sb->s_flags & MS_RDONLY)
 		return 0;
 
 	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
 	if (ret)
-		return ret;
+		goto out;
 
 	/*
 	 * Taking the mutex here just to keep consistent with how fsync was
@@ -83,7 +86,8 @@ int ext3_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 */
 	if (ext3_should_journal_data(inode)) {
 		mutex_unlock(&inode->i_mutex);
-		return ext3_force_commit(inode->i_sb);
+		ret = ext3_force_commit(inode->i_sb);
+		goto out;
 	}
 
 	if (datasync)
@@ -104,6 +108,9 @@ int ext3_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 */
 	if (needs_barrier)
 		blkdev_issue_flush(inode->i_sb->s_bdev, GFP_KERNEL, NULL);
+
 	mutex_unlock(&inode->i_mutex);
+out:
+	trace_ext3_sync_file_exit(inode, ret);
 	return ret;
 }
diff --git a/fs/ext3/ialloc.c b/fs/ext3/ialloc.c
index bfc2dc4..bf09cbf 100644
--- a/fs/ext3/ialloc.c
+++ b/fs/ext3/ialloc.c
@@ -23,6 +23,7 @@
 #include <linux/buffer_head.h>
 #include <linux/random.h>
 #include <linux/bitops.h>
+#include <trace/events/ext3.h>
 
 #include <asm/byteorder.h>
 
@@ -118,6 +119,7 @@ void ext3_free_inode (handle_t *handle, struct inode * inode)
 
 	ino = inode->i_ino;
 	ext3_debug ("freeing inode %lu\n", ino);
+	trace_ext3_free_inode(inode);
 
 	is_directory = S_ISDIR(inode->i_mode);
 
@@ -426,6 +428,7 @@ struct inode *ext3_new_inode(handle_t *handle, struct inode * dir,
 		return ERR_PTR(-EPERM);
 
 	sb = dir->i_sb;
+	trace_ext3_request_inode(dir, mode);
 	inode = new_inode(sb);
 	if (!inode)
 		return ERR_PTR(-ENOMEM);
@@ -601,6 +604,7 @@ got:
 	}
 
 	ext3_debug("allocating inode %lu\n", inode->i_ino);
+	trace_ext3_allocate_inode(inode, dir, mode);
 	goto really_out;
 fail:
 	ext3_std_error(sb, err);
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index 2978a2a..04da6ac 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -38,10 +38,12 @@
 #include <linux/bio.h>
 #include <linux/fiemap.h>
 #include <linux/namei.h>
+#include <trace/events/ext3.h>
 #include "xattr.h"
 #include "acl.h"
 
 static int ext3_writepage_trans_blocks(struct inode *inode);
+static int ext3_block_truncate_page(struct inode *inode, loff_t from);
 
 /*
  * Test whether an inode is a fast symlink.
@@ -70,6 +72,7 @@ int ext3_forget(handle_t *handle, int is_metadata, struct inode *inode,
 
 	might_sleep();
 
+	trace_ext3_forget(inode, is_metadata, blocknr);
 	BUFFER_TRACE(bh, "enter");
 
 	jbd_debug(4, "forgetting bh %p: is_metadata = %d, mode %o, "
@@ -194,20 +197,47 @@ static int truncate_restart_transaction(handle_t *handle, struct inode *inode)
  */
 void ext3_evict_inode (struct inode *inode)
 {
+	struct ext3_inode_info *ei = EXT3_I(inode);
 	struct ext3_block_alloc_info *rsv;
 	handle_t *handle;
 	int want_delete = 0;
 
+	trace_ext3_evict_inode(inode);
 	if (!inode->i_nlink && !is_bad_inode(inode)) {
 		dquot_initialize(inode);
 		want_delete = 1;
 	}
 
+	/*
+	 * When journalling data dirty buffers are tracked only in the journal.
+	 * So although mm thinks everything is clean and ready for reaping the
+	 * inode might still have some pages to write in the running
+	 * transaction or waiting to be checkpointed. Thus calling
+	 * journal_invalidatepage() (via truncate_inode_pages()) to discard
+	 * these buffers can cause data loss. Also even if we did not discard
+	 * these buffers, we would have no way to find them after the inode
+	 * is reaped and thus user could see stale data if he tries to read
+	 * them before the transaction is checkpointed. So be careful and
+	 * force everything to disk here... We use ei->i_datasync_tid to
+	 * store the newest transaction containing inode's data.
+	 *
+	 * Note that directories do not have this problem because they don't
+	 * use page cache.
+	 */
+	if (inode->i_nlink && ext3_should_journal_data(inode) &&
+	    (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode))) {
+		tid_t commit_tid = atomic_read(&ei->i_datasync_tid);
+		journal_t *journal = EXT3_SB(inode->i_sb)->s_journal;
+
+		log_start_commit(journal, commit_tid);
+		log_wait_commit(journal, commit_tid);
+		filemap_write_and_wait(&inode->i_data);
+	}
 	truncate_inode_pages(&inode->i_data, 0);
 
 	ext3_discard_reservation(inode);
-	rsv = EXT3_I(inode)->i_block_alloc_info;
-	EXT3_I(inode)->i_block_alloc_info = NULL;
+	rsv = ei->i_block_alloc_info;
+	ei->i_block_alloc_info = NULL;
 	if (unlikely(rsv))
 		kfree(rsv);
 
@@ -231,15 +261,13 @@ void ext3_evict_inode (struct inode *inode)
 	if (inode->i_blocks)
 		ext3_truncate(inode);
 	/*
-	 * Kill off the orphan record which ext3_truncate created.
-	 * AKPM: I think this can be inside the above `if'.
-	 * Note that ext3_orphan_del() has to be able to cope with the
-	 * deletion of a non-existent orphan - this is because we don't
-	 * know if ext3_truncate() actually created an orphan record.
-	 * (Well, we could do this if we need to, but heck - it works)
+	 * Kill off the orphan record created when the inode lost the last
+	 * link.  Note that ext3_orphan_del() has to be able to cope with the
+	 * deletion of a non-existent orphan - ext3_truncate() could
+	 * have removed the record.
 	 */
 	ext3_orphan_del(handle, inode);
-	EXT3_I(inode)->i_dtime	= get_seconds();
+	ei->i_dtime = get_seconds();
 
 	/*
 	 * One subtle ordering requirement: if anything has gone wrong
@@ -842,6 +870,7 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
 	ext3_fsblk_t first_block = 0;
 
 
+	trace_ext3_get_blocks_enter(inode, iblock, maxblocks, create);
 	J_ASSERT(handle != NULL || create == 0);
 	depth = ext3_block_to_path(inode,iblock,offsets,&blocks_to_boundary);
 
@@ -886,6 +915,9 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
 	if (!create || err == -EIO)
 		goto cleanup;
 
+	/*
+	 * Block out ext3_truncate while we alter the tree
+	 */
 	mutex_lock(&ei->truncate_mutex);
 
 	/*
@@ -934,9 +966,6 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
 	 */
 	count = ext3_blks_to_allocate(partial, indirect_blks,
 					maxblocks, blocks_to_boundary);
-	/*
-	 * Block out ext3_truncate while we alter the tree
-	 */
 	err = ext3_alloc_branch(handle, inode, indirect_blks, &count, goal,
 				offsets + (partial - chain), partial);
 
@@ -970,6 +999,9 @@ cleanup:
 	}
 	BUFFER_TRACE(bh_result, "returned");
 out:
+	trace_ext3_get_blocks_exit(inode, iblock,
+				   depth ? le32_to_cpu(chain[depth-1].key) : 0,
+				   count, err);
 	return err;
 }
 
@@ -1202,6 +1234,16 @@ static void ext3_truncate_failed_write(struct inode *inode)
 	ext3_truncate(inode);
 }
 
+/*
+ * Truncate blocks that were not used by direct IO write. We have to zero out
+ * the last file block as well because direct IO might have written to it.
+ */
+static void ext3_truncate_failed_direct_write(struct inode *inode)
+{
+	ext3_block_truncate_page(inode, inode->i_size);
+	ext3_truncate(inode);
+}
+
 static int ext3_write_begin(struct file *file, struct address_space *mapping,
 				loff_t pos, unsigned len, unsigned flags,
 				struct page **pagep, void **fsdata)
@@ -1217,6 +1259,8 @@ static int ext3_write_begin(struct file *file, struct address_space *mapping,
 	 * we allocate blocks but write fails for some reason */
 	int needed_blocks = ext3_writepage_trans_blocks(inode) + 1;
 
+	trace_ext3_write_begin(inode, pos, len, flags);
+
 	index = pos >> PAGE_CACHE_SHIFT;
 	from = pos & (PAGE_CACHE_SIZE - 1);
 	to = from + len;
@@ -1332,6 +1376,7 @@ static int ext3_ordered_write_end(struct file *file,
 	unsigned from, to;
 	int ret = 0, ret2;
 
+	trace_ext3_ordered_write_end(inode, pos, len, copied);
 	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);
 
 	from = pos & (PAGE_CACHE_SIZE - 1);
@@ -1367,6 +1412,7 @@ static int ext3_writeback_write_end(struct file *file,
 	struct inode *inode = file->f_mapping->host;
 	int ret;
 
+	trace_ext3_writeback_write_end(inode, pos, len, copied);
 	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);
 	update_file_sizes(inode, pos, copied);
 	/*
@@ -1391,10 +1437,12 @@ static int ext3_journalled_write_end(struct file *file,
 {
 	handle_t *handle = ext3_journal_current_handle();
 	struct inode *inode = mapping->host;
+	struct ext3_inode_info *ei = EXT3_I(inode);
 	int ret = 0, ret2;
 	int partial = 0;
 	unsigned from, to;
 
+	trace_ext3_journalled_write_end(inode, pos, len, copied);
 	from = pos & (PAGE_CACHE_SIZE - 1);
 	to = from + len;
 
@@ -1419,8 +1467,9 @@ static int ext3_journalled_write_end(struct file *file,
 	if (pos + len > inode->i_size && ext3_can_truncate(inode))
 		ext3_orphan_add(handle, inode);
 	ext3_set_inode_state(inode, EXT3_STATE_JDATA);
-	if (inode->i_size > EXT3_I(inode)->i_disksize) {
-		EXT3_I(inode)->i_disksize = inode->i_size;
+	atomic_set(&ei->i_datasync_tid, handle->h_transaction->t_tid);
+	if (inode->i_size > ei->i_disksize) {
+		ei->i_disksize = inode->i_size;
 		ret2 = ext3_mark_inode_dirty(handle, inode);
 		if (!ret)
 			ret = ret2;
@@ -1577,6 +1626,7 @@ static int ext3_ordered_writepage(struct page *page,
 	if (ext3_journal_current_handle())
 		goto out_fail;
 
+	trace_ext3_ordered_writepage(page);
 	if (!page_has_buffers(page)) {
 		create_empty_buffers(page, inode->i_sb->s_blocksize,
 				(1 << BH_Dirty)|(1 << BH_Uptodate));
@@ -1647,6 +1697,7 @@ static int ext3_writeback_writepage(struct page *page,
 	if (ext3_journal_current_handle())
 		goto out_fail;
 
+	trace_ext3_writeback_writepage(page);
 	if (page_has_buffers(page)) {
 		if (!walk_page_buffers(NULL, page_buffers(page), 0,
 				      PAGE_CACHE_SIZE, NULL, buffer_unmapped)) {
@@ -1689,6 +1740,7 @@ static int ext3_journalled_writepage(struct page *page,
 	if (ext3_journal_current_handle())
 		goto no_write;
 
+	trace_ext3_journalled_writepage(page);
 	handle = ext3_journal_start(inode, ext3_writepage_trans_blocks(inode));
 	if (IS_ERR(handle)) {
 		ret = PTR_ERR(handle);
@@ -1715,6 +1767,8 @@ static int ext3_journalled_writepage(struct page *page,
 		if (ret == 0)
 			ret = err;
 		ext3_set_inode_state(inode, EXT3_STATE_JDATA);
+		atomic_set(&EXT3_I(inode)->i_datasync_tid,
+			   handle->h_transaction->t_tid);
 		unlock_page(page);
 	} else {
 		/*
@@ -1739,6 +1793,7 @@ out_unlock:
 
 static int ext3_readpage(struct file *file, struct page *page)
 {
+	trace_ext3_readpage(page);
 	return mpage_readpage(page, ext3_get_block);
 }
 
@@ -1753,6 +1808,8 @@ static void ext3_invalidatepage(struct page *page, unsigned long offset)
 {
 	journal_t *journal = EXT3_JOURNAL(page->mapping->host);
 
+	trace_ext3_invalidatepage(page, offset);
+
 	/*
 	 * If it's a full truncate we just forget about the pending dirtying
 	 */
@@ -1766,6 +1823,7 @@ static int ext3_releasepage(struct page *page, gfp_t wait)
 {
 	journal_t *journal = EXT3_JOURNAL(page->mapping->host);
 
+	trace_ext3_releasepage(page);
 	WARN_ON(PageChecked(page));
 	if (!page_has_buffers(page))
 		return 0;
@@ -1794,6 +1852,8 @@ static ssize_t ext3_direct_IO(int rw, struct kiocb *iocb,
 	size_t count = iov_length(iov, nr_segs);
 	int retries = 0;
 
+	trace_ext3_direct_IO_enter(inode, offset, iov_length(iov, nr_segs), rw);
+
 	if (rw == WRITE) {
 		loff_t final_size = offset + count;
 
@@ -1827,7 +1887,7 @@ retry:
 		loff_t end = offset + iov_length(iov, nr_segs);
 
 		if (end > isize)
-			vmtruncate(inode, isize);
+			ext3_truncate_failed_direct_write(inode);
 	}
 	if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries))
 		goto retry;
@@ -1841,7 +1901,7 @@ retry:
 			/* This is really bad luck. We've written the data
 			 * but cannot extend i_size. Truncate allocated blocks
 			 * and pretend the write failed... */
-			ext3_truncate(inode);
+			ext3_truncate_failed_direct_write(inode);
 			ret = PTR_ERR(handle);
 			goto out;
 		}
@@ -1867,6 +1927,8 @@ retry:
 			ret = err;
 	}
 out:
+	trace_ext3_direct_IO_exit(inode, offset,
+				iov_length(iov, nr_segs), rw, ret);
 	return ret;
 }
 
@@ -1949,17 +2011,24 @@ void ext3_set_aops(struct inode *inode)
  * This required during truncate. We need to physically zero the tail end
  * of that block so it doesn't yield old data if the file is later grown.
  */
-static int ext3_block_truncate_page(handle_t *handle, struct page *page,
-		struct address_space *mapping, loff_t from)
+static int ext3_block_truncate_page(struct inode *inode, loff_t from)
 {
 	ext3_fsblk_t index = from >> PAGE_CACHE_SHIFT;
-	unsigned offset = from & (PAGE_CACHE_SIZE-1);
+	unsigned offset = from & (PAGE_CACHE_SIZE - 1);
 	unsigned blocksize, iblock, length, pos;
-	struct inode *inode = mapping->host;
+	struct page *page;
+	handle_t *handle = NULL;
 	struct buffer_head *bh;
 	int err = 0;
 
+	/* Truncated on block boundary - nothing to do */
 	blocksize = inode->i_sb->s_blocksize;
+	if ((from & (blocksize - 1)) == 0)
+		return 0;
+
+	page = grab_cache_page(inode->i_mapping, index);
+	if (!page)
+		return -ENOMEM;
 	length = blocksize - (offset & (blocksize - 1));
 	iblock = index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits);
 
@@ -2004,11 +2073,23 @@ static int ext3_block_truncate_page(handle_t *handle, struct page *page,
 			goto unlock;
 	}
 
+	/* data=writeback mode doesn't need transaction to zero-out data */
+	if (!ext3_should_writeback_data(inode)) {
+		/* We journal at most one block */
+		handle = ext3_journal_start(inode, 1);
+		if (IS_ERR(handle)) {
+			clear_highpage(page);
+			flush_dcache_page(page);
+			err = PTR_ERR(handle);
+			goto unlock;
+		}
+	}
+
 	if (ext3_should_journal_data(inode)) {
 		BUFFER_TRACE(bh, "get write access");
 		err = ext3_journal_get_write_access(handle, bh);
 		if (err)
-			goto unlock;
+			goto stop;
 	}
 
 	zero_user(page, offset, length);
@@ -2022,6 +2103,9 @@ static int ext3_block_truncate_page(handle_t *handle, struct page *page,
 			err = ext3_journal_dirty_data(handle, bh);
 		mark_buffer_dirty(bh);
 	}
+stop:
+	if (handle)
+		ext3_journal_stop(handle);
 
 unlock:
 	unlock_page(page);
@@ -2390,8 +2474,6 @@ static void ext3_free_branches(handle_t *handle, struct inode *inode,
 
 int ext3_can_truncate(struct inode *inode)
 {
-	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
-		return 0;
 	if (S_ISREG(inode->i_mode))
 		return 1;
 	if (S_ISDIR(inode->i_mode))
@@ -2435,7 +2517,6 @@ void ext3_truncate(struct inode *inode)
 	struct ext3_inode_info *ei = EXT3_I(inode);
 	__le32 *i_data = ei->i_data;
 	int addr_per_block = EXT3_ADDR_PER_BLOCK(inode->i_sb);
-	struct address_space *mapping = inode->i_mapping;
 	int offsets[4];
 	Indirect chain[4];
 	Indirect *partial;
@@ -2443,7 +2524,8 @@ void ext3_truncate(struct inode *inode)
 	int n;
 	long last_block;
 	unsigned blocksize = inode->i_sb->s_blocksize;
-	struct page *page;
+
+	trace_ext3_truncate_enter(inode);
 
 	if (!ext3_can_truncate(inode))
 		goto out_notrans;
@@ -2451,37 +2533,12 @@ void ext3_truncate(struct inode *inode)
 	if (inode->i_size == 0 && ext3_should_writeback_data(inode))
 		ext3_set_inode_state(inode, EXT3_STATE_FLUSH_ON_CLOSE);
 
-	/*
-	 * We have to lock the EOF page here, because lock_page() nests
-	 * outside journal_start().
-	 */
-	if ((inode->i_size & (blocksize - 1)) == 0) {
-		/* Block boundary? Nothing to do */
-		page = NULL;
-	} else {
-		page = grab_cache_page(mapping,
-				inode->i_size >> PAGE_CACHE_SHIFT);
-		if (!page)
-			goto out_notrans;
-	}
-
 	handle = start_transaction(inode);
-	if (IS_ERR(handle)) {
-		if (page) {
-			clear_highpage(page);
-			flush_dcache_page(page);
-			unlock_page(page);
-			page_cache_release(page);
-		}
+	if (IS_ERR(handle))
 		goto out_notrans;
-	}
 
 	last_block = (inode->i_size + blocksize-1)
 					>> EXT3_BLOCK_SIZE_BITS(inode->i_sb);
-
-	if (page)
-		ext3_block_truncate_page(handle, page, mapping, inode->i_size);
-
 	n = ext3_block_to_path(inode, last_block, offsets, NULL);
 	if (n == 0)
 		goto out_stop;	/* error */
@@ -2596,6 +2653,7 @@ out_stop:
 		ext3_orphan_del(handle, inode);
 
 	ext3_journal_stop(handle);
+	trace_ext3_truncate_exit(inode);
 	return;
 out_notrans:
 	/*
@@ -2604,6 +2662,7 @@ out_notrans:
 	 */
 	if (inode->i_nlink)
 		ext3_orphan_del(NULL, inode);
+	trace_ext3_truncate_exit(inode);
 }
 
 static ext3_fsblk_t ext3_get_inode_block(struct super_block *sb,
@@ -2745,6 +2804,7 @@ make_io:
 		 * has in-inode xattrs, or we don't have this inode in memory.
 		 * Read the block from disk.
 		 */
+		trace_ext3_load_inode(inode);
 		get_bh(bh);
 		bh->b_end_io = end_buffer_read_sync;
 		submit_bh(READ_META, bh);
@@ -3229,18 +3289,36 @@ int ext3_setattr(struct dentry *dentry, struct iattr *attr)
 		}
 
 		error = ext3_orphan_add(handle, inode);
+		if (error) {
+			ext3_journal_stop(handle);
+			goto err_out;
+		}
 		EXT3_I(inode)->i_disksize = attr->ia_size;
-		rc = ext3_mark_inode_dirty(handle, inode);
-		if (!error)
-			error = rc;
+		error = ext3_mark_inode_dirty(handle, inode);
 		ext3_journal_stop(handle);
+		if (error) {
+			/* Some hard fs error must have happened. Bail out. */
+			ext3_orphan_del(NULL, inode);
+			goto err_out;
+		}
+		rc = ext3_block_truncate_page(inode, attr->ia_size);
+		if (rc) {
+			/* Cleanup orphan list and exit */
+			handle = ext3_journal_start(inode, 3);
+			if (IS_ERR(handle)) {
+				ext3_orphan_del(NULL, inode);
+				goto err_out;
+			}
+			ext3_orphan_del(handle, inode);
+			ext3_journal_stop(handle);
+			goto err_out;
+		}
 	}
 
 	if ((attr->ia_valid & ATTR_SIZE) &&
 	    attr->ia_size != i_size_read(inode)) {
-		rc = vmtruncate(inode, attr->ia_size);
-		if (rc)
-			goto err_out;
+		truncate_setsize(inode, attr->ia_size);
+		ext3_truncate(inode);
 	}
 
 	setattr_copy(inode, attr);
@@ -3374,6 +3452,7 @@ int ext3_mark_inode_dirty(handle_t *handle, struct inode *inode)
 	int err;
 
 	might_sleep();
+	trace_ext3_mark_inode_dirty(inode, _RET_IP_);
 	err = ext3_reserve_inode_write(handle, inode, &iloc);
 	if (!err)
 		err = ext3_mark_iloc_dirty(handle, inode, &iloc);
diff --git a/fs/ext3/ioctl.c b/fs/ext3/ioctl.c
index f4090bd..c7f4394 100644
--- a/fs/ext3/ioctl.c
+++ b/fs/ext3/ioctl.c
@@ -285,7 +285,7 @@ group_add_out:
 		if (!capable(CAP_SYS_ADMIN))
 			return -EPERM;
 
-		if (copy_from_user(&range, (struct fstrim_range *)arg,
+		if (copy_from_user(&range, (struct fstrim_range __user *)arg,
 				   sizeof(range)))
 			return -EFAULT;
 
@@ -293,7 +293,7 @@ group_add_out:
 		if (ret < 0)
 			return ret;
 
-		if (copy_to_user((struct fstrim_range *)arg, &range,
+		if (copy_to_user((struct fstrim_range __user *)arg, &range,
 				 sizeof(range)))
 			return -EFAULT;
 
diff --git a/fs/ext3/namei.c b/fs/ext3/namei.c
index 3b57230..6e18a0b 100644
--- a/fs/ext3/namei.c
+++ b/fs/ext3/namei.c
@@ -36,6 +36,7 @@
 #include <linux/quotaops.h>
 #include <linux/buffer_head.h>
 #include <linux/bio.h>
+#include <trace/events/ext3.h>
 
 #include "namei.h"
 #include "xattr.h"
@@ -287,7 +288,7 @@ static struct stats dx_show_leaf(struct dx_hash_info *hinfo, struct ext3_dir_ent
 				while (len--) printk("%c", *name++);
 				ext3fs_dirhash(de->name, de->name_len, &h);
 				printk(":%x.%u ", h.hash,
-				       ((char *) de - base));
+				       (unsigned) ((char *) de - base));
 			}
 			space += EXT3_DIR_REC_LEN(de->name_len);
 			names++;
@@ -1013,7 +1014,7 @@ static struct buffer_head * ext3_dx_find_entry(struct inode *dir,
 
 	*err = -ENOENT;
 errout:
-	dxtrace(printk("%s not found\n", name));
+	dxtrace(printk("%s not found\n", entry->name));
 	dx_release (frames);
 	return NULL;
 }
@@ -2140,6 +2141,7 @@ static int ext3_unlink(struct inode * dir, struct dentry *dentry)
 	struct ext3_dir_entry_2 * de;
 	handle_t *handle;
 
+	trace_ext3_unlink_enter(dir, dentry);
 	/* Initialize quotas before so that eventual writes go
 	 * in separate transaction */
 	dquot_initialize(dir);
@@ -2185,6 +2187,7 @@ static int ext3_unlink(struct inode * dir, struct dentry *dentry)
 end_unlink:
 	ext3_journal_stop(handle);
 	brelse (bh);
+	trace_ext3_unlink_exit(dentry, retval);
 	return retval;
 }
 
diff --git a/fs/ext3/super.c b/fs/ext3/super.c
index b57ea2f..7beb69a 100644
--- a/fs/ext3/super.c
+++ b/fs/ext3/super.c
@@ -44,6 +44,9 @@
 #include "acl.h"
 #include "namei.h"
 
+#define CREATE_TRACE_POINTS
+#include <trace/events/ext3.h>
+
 #ifdef CONFIG_EXT3_DEFAULTS_TO_ORDERED
   #define EXT3_MOUNT_DEFAULT_DATA_MODE EXT3_MOUNT_ORDERED_DATA
 #else
@@ -497,6 +500,14 @@ static struct inode *ext3_alloc_inode(struct super_block *sb)
 	return &ei->vfs_inode;
 }
 
+static int ext3_drop_inode(struct inode *inode)
+{
+	int drop = generic_drop_inode(inode);
+
+	trace_ext3_drop_inode(inode, drop);
+	return drop;
+}
+
 static void ext3_i_callback(struct rcu_head *head)
 {
 	struct inode *inode = container_of(head, struct inode, i_rcu);
@@ -788,6 +799,7 @@ static const struct super_operations ext3_sops = {
 	.destroy_inode	= ext3_destroy_inode,
 	.write_inode	= ext3_write_inode,
 	.dirty_inode	= ext3_dirty_inode,
+	.drop_inode	= ext3_drop_inode,
 	.evict_inode	= ext3_evict_inode,
 	.put_super	= ext3_put_super,
 	.sync_fs	= ext3_sync_fs,
@@ -2509,6 +2521,7 @@ static int ext3_sync_fs(struct super_block *sb, int wait)
 {
 	tid_t target;
 
+	trace_ext3_sync_fs(sb, wait);
 	if (journal_start_commit(EXT3_SB(sb)->s_journal, &target)) {
 		if (wait)
 			log_wait_commit(EXT3_SB(sb)->s_journal, target);
diff --git a/fs/ext3/xattr.c b/fs/ext3/xattr.c
index 32e6cc2..d565759 100644
--- a/fs/ext3/xattr.c
+++ b/fs/ext3/xattr.c
@@ -803,8 +803,16 @@ inserted:
 			/* We need to allocate a new block */
 			ext3_fsblk_t goal = ext3_group_first_block_no(sb,
 						EXT3_I(inode)->i_block_group);
-			ext3_fsblk_t block = ext3_new_block(handle, inode,
-							goal, &error);
+			ext3_fsblk_t block;
+
+			/*
+			 * Protect us agaist concurrent allocations to the
+			 * same inode from ext3_..._writepage(). Reservation
+			 * code does not expect racing allocations.
+			 */
+			mutex_lock(&EXT3_I(inode)->truncate_mutex);
+			block = ext3_new_block(handle, inode, goal, &error);
+			mutex_unlock(&EXT3_I(inode)->truncate_mutex);
 			if (error)
 				goto cleanup;
 			ea_idebug(inode, "creating block %d", block);
diff --git a/fs/jbd/checkpoint.c b/fs/jbd/checkpoint.c
index e4b87bc..f94fc48 100644
--- a/fs/jbd/checkpoint.c
+++ b/fs/jbd/checkpoint.c
@@ -22,6 +22,8 @@
 #include <linux/jbd.h>
 #include <linux/errno.h>
 #include <linux/slab.h>
+#include <linux/blkdev.h>
+#include <trace/events/jbd.h>
 
 /*
  * Unlink a buffer from a transaction checkpoint list.
@@ -95,10 +97,14 @@ static int __try_to_free_cp_buf(struct journal_head *jh)
 
 	if (jh->b_jlist == BJ_None && !buffer_locked(bh) &&
 	    !buffer_dirty(bh) && !buffer_write_io_error(bh)) {
+		/*
+		 * Get our reference so that bh cannot be freed before
+		 * we unlock it
+		 */
+		get_bh(bh);
 		JBUFFER_TRACE(jh, "remove from checkpoint list");
 		ret = __journal_remove_checkpoint(jh) + 1;
 		jbd_unlock_bh_state(bh);
-		journal_remove_journal_head(bh);
 		BUFFER_TRACE(bh, "release");
 		__brelse(bh);
 	} else {
@@ -220,8 +226,8 @@ restart:
 			spin_lock(&journal->j_list_lock);
 			goto restart;
 		}
+		get_bh(bh);
 		if (buffer_locked(bh)) {
-			get_bh(bh);
 			spin_unlock(&journal->j_list_lock);
 			jbd_unlock_bh_state(bh);
 			wait_on_buffer(bh);
@@ -240,7 +246,6 @@ restart:
 		 */
 		released = __journal_remove_checkpoint(jh);
 		jbd_unlock_bh_state(bh);
-		journal_remove_journal_head(bh);
 		__brelse(bh);
 	}
 
@@ -253,9 +258,12 @@ static void
 __flush_batch(journal_t *journal, struct buffer_head **bhs, int *batch_count)
 {
 	int i;
+	struct blk_plug plug;
 
+	blk_start_plug(&plug);
 	for (i = 0; i < *batch_count; i++)
-		write_dirty_buffer(bhs[i], WRITE);
+		write_dirty_buffer(bhs[i], WRITE_SYNC);
+	blk_finish_plug(&plug);
 
 	for (i = 0; i < *batch_count; i++) {
 		struct buffer_head *bh = bhs[i];
@@ -304,12 +312,12 @@ static int __process_buffer(journal_t *journal, struct journal_head *jh,
 		ret = 1;
 		if (unlikely(buffer_write_io_error(bh)))
 			ret = -EIO;
+		get_bh(bh);
 		J_ASSERT_JH(jh, !buffer_jbddirty(bh));
 		BUFFER_TRACE(bh, "remove from checkpoint");
 		__journal_remove_checkpoint(jh);
 		spin_unlock(&journal->j_list_lock);
 		jbd_unlock_bh_state(bh);
-		journal_remove_journal_head(bh);
 		__brelse(bh);
 	} else {
 		/*
@@ -358,6 +366,7 @@ int log_do_checkpoint(journal_t *journal)
 	 * journal straight away.
 	 */
 	result = cleanup_journal_tail(journal);
+	trace_jbd_checkpoint(journal, result);
 	jbd_debug(1, "cleanup_journal_tail returned %d\n", result);
 	if (result <= 0)
 		return result;
@@ -503,6 +512,7 @@ int cleanup_journal_tail(journal_t *journal)
 	if (blocknr < journal->j_tail)
 		freed = freed + journal->j_last - journal->j_first;
 
+	trace_jbd_cleanup_journal_tail(journal, first_tid, blocknr, freed);
 	jbd_debug(1,
 		  "Cleaning journal tail from %d to %d (offset %u), "
 		  "freeing %u\n",
@@ -523,9 +533,9 @@ int cleanup_journal_tail(journal_t *journal)
 /*
  * journal_clean_one_cp_list
  *
- * Find all the written-back checkpoint buffers in the given list and release them.
+ * Find all the written-back checkpoint buffers in the given list and release
+ * them.
  *
- * Called with the journal locked.
  * Called with j_list_lock held.
  * Returns number of bufers reaped (for debug)
  */
@@ -632,8 +642,8 @@ out:
  * checkpoint lists.
  *
  * The function returns 1 if it frees the transaction, 0 otherwise.
+ * The function can free jh and bh.
  *
- * This function is called with the journal locked.
  * This function is called with j_list_lock held.
  * This function is called with jbd_lock_bh_state(jh2bh(jh))
  */
@@ -652,13 +662,14 @@ int __journal_remove_checkpoint(struct journal_head *jh)
 	}
 	journal = transaction->t_journal;
 
+	JBUFFER_TRACE(jh, "removing from transaction");
 	__buffer_unlink(jh);
 	jh->b_cp_transaction = NULL;
+	journal_put_journal_head(jh);
 
 	if (transaction->t_checkpoint_list != NULL ||
 	    transaction->t_checkpoint_io_list != NULL)
 		goto out;
-	JBUFFER_TRACE(jh, "transaction has no more buffers");
 
 	/*
 	 * There is one special case to worry about: if we have just pulled the
@@ -669,10 +680,8 @@ int __journal_remove_checkpoint(struct journal_head *jh)
 	 * The locking here around t_state is a bit sleazy.
 	 * See the comment at the end of journal_commit_transaction().
 	 */
-	if (transaction->t_state != T_FINISHED) {
-		JBUFFER_TRACE(jh, "belongs to running/committing transaction");
+	if (transaction->t_state != T_FINISHED)
 		goto out;
-	}
 
 	/* OK, that was the last buffer for the transaction: we can now
 	   safely remove this transaction from the log */
@@ -684,7 +693,6 @@ int __journal_remove_checkpoint(struct journal_head *jh)
 	wake_up(&journal->j_wait_logspace);
 	ret = 1;
 out:
-	JBUFFER_TRACE(jh, "exit");
 	return ret;
 }
 
@@ -703,6 +711,8 @@ void __journal_insert_checkpoint(struct journal_head *jh,
 	J_ASSERT_JH(jh, buffer_dirty(jh2bh(jh)) || buffer_jbddirty(jh2bh(jh)));
 	J_ASSERT_JH(jh, jh->b_cp_transaction == NULL);
 
+	/* Get reference for checkpointing transaction */
+	journal_grab_journal_head(jh2bh(jh));
 	jh->b_cp_transaction = transaction;
 
 	if (!transaction->t_checkpoint_list) {
@@ -752,6 +762,7 @@ void __journal_drop_transaction(journal_t *journal, transaction_t *transaction)
 	J_ASSERT(journal->j_committing_transaction != transaction);
 	J_ASSERT(journal->j_running_transaction != transaction);
 
+	trace_jbd_drop_transaction(journal, transaction);
 	jbd_debug(1, "Dropping transaction %d, all done\n", transaction->t_tid);
 	kfree(transaction);
 }
diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c
index 72ffa97..8799207 100644
--- a/fs/jbd/commit.c
+++ b/fs/jbd/commit.c
@@ -21,6 +21,7 @@
 #include <linux/pagemap.h>
 #include <linux/bio.h>
 #include <linux/blkdev.h>
+#include <trace/events/jbd.h>
 
 /*
  * Default IO end handler for temporary BJ_IO buffer_heads.
@@ -204,6 +205,8 @@ write_out_data:
 			if (!trylock_buffer(bh)) {
 				BUFFER_TRACE(bh, "needs blocking lock");
 				spin_unlock(&journal->j_list_lock);
+				trace_jbd_do_submit_data(journal,
+						     commit_transaction);
 				/* Write out all data to prevent deadlocks */
 				journal_do_submit_data(wbuf, bufs, write_op);
 				bufs = 0;
@@ -236,6 +239,8 @@ write_out_data:
 			jbd_unlock_bh_state(bh);
 			if (bufs == journal->j_wbufsize) {
 				spin_unlock(&journal->j_list_lock);
+				trace_jbd_do_submit_data(journal,
+						     commit_transaction);
 				journal_do_submit_data(wbuf, bufs, write_op);
 				bufs = 0;
 				goto write_out_data;
@@ -253,10 +258,6 @@ write_out_data:
 			jbd_unlock_bh_state(bh);
 			if (locked)
 				unlock_buffer(bh);
-			journal_remove_journal_head(bh);
-			/* One for our safety reference, other for
-			 * journal_remove_journal_head() */
-			put_bh(bh);
 			release_data_buffer(bh);
 		}
 
@@ -266,6 +267,7 @@ write_out_data:
 		}
 	}
 	spin_unlock(&journal->j_list_lock);
+	trace_jbd_do_submit_data(journal, commit_transaction);
 	journal_do_submit_data(wbuf, bufs, write_op);
 
 	return err;
@@ -316,12 +318,14 @@ void journal_commit_transaction(journal_t *journal)
 	commit_transaction = journal->j_running_transaction;
 	J_ASSERT(commit_transaction->t_state == T_RUNNING);
 
+	trace_jbd_start_commit(journal, commit_transaction);
 	jbd_debug(1, "JBD: starting commit of transaction %d\n",
 			commit_transaction->t_tid);
 
 	spin_lock(&journal->j_state_lock);
 	commit_transaction->t_state = T_LOCKED;
 
+	trace_jbd_commit_locking(journal, commit_transaction);
 	spin_lock(&commit_transaction->t_handle_lock);
 	while (commit_transaction->t_updates) {
 		DEFINE_WAIT(wait);
@@ -392,6 +396,7 @@ void journal_commit_transaction(journal_t *journal)
 	 */
 	journal_switch_revoke_table(journal);
 
+	trace_jbd_commit_flushing(journal, commit_transaction);
 	commit_transaction->t_state = T_FLUSH;
 	journal->j_committing_transaction = commit_transaction;
 	journal->j_running_transaction = NULL;
@@ -446,14 +451,9 @@ void journal_commit_transaction(journal_t *journal)
 		}
 		if (buffer_jbd(bh) && bh2jh(bh) == jh &&
 		    jh->b_transaction == commit_transaction &&
-		    jh->b_jlist == BJ_Locked) {
+		    jh->b_jlist == BJ_Locked)
 			__journal_unfile_buffer(jh);
-			jbd_unlock_bh_state(bh);
-			journal_remove_journal_head(bh);
-			put_bh(bh);
-		} else {
-			jbd_unlock_bh_state(bh);
-		}
+		jbd_unlock_bh_state(bh);
 		release_data_buffer(bh);
 		cond_resched_lock(&journal->j_list_lock);
 	}
@@ -493,6 +493,7 @@ void journal_commit_transaction(journal_t *journal)
 	commit_transaction->t_state = T_COMMIT;
 	spin_unlock(&journal->j_state_lock);
 
+	trace_jbd_commit_logging(journal, commit_transaction);
 	J_ASSERT(commit_transaction->t_nr_buffers <=
 		 commit_transaction->t_outstanding_credits);
 
@@ -797,10 +798,16 @@ restart_loop:
 	while (commit_transaction->t_forget) {
 		transaction_t *cp_transaction;
 		struct buffer_head *bh;
+		int try_to_free = 0;
 
 		jh = commit_transaction->t_forget;
 		spin_unlock(&journal->j_list_lock);
 		bh = jh2bh(jh);
+		/*
+		 * Get a reference so that bh cannot be freed before we are
+		 * done with it.
+		 */
+		get_bh(bh);
 		jbd_lock_bh_state(bh);
 		J_ASSERT_JH(jh,	jh->b_transaction == commit_transaction ||
 			jh->b_transaction == journal->j_running_transaction);
@@ -858,28 +865,27 @@ restart_loop:
 			__journal_insert_checkpoint(jh, commit_transaction);
 			if (is_journal_aborted(journal))
 				clear_buffer_jbddirty(bh);
-			JBUFFER_TRACE(jh, "refile for checkpoint writeback");
-			__journal_refile_buffer(jh);
-			jbd_unlock_bh_state(bh);
 		} else {
 			J_ASSERT_BH(bh, !buffer_dirty(bh));
-			/* The buffer on BJ_Forget list and not jbddirty means
+			/*
+			 * The buffer on BJ_Forget list and not jbddirty means
 			 * it has been freed by this transaction and hence it
 			 * could not have been reallocated until this
 			 * transaction has committed. *BUT* it could be
 			 * reallocated once we have written all the data to
 			 * disk and before we process the buffer on BJ_Forget
-			 * list. */
-			JBUFFER_TRACE(jh, "refile or unfile freed buffer");
-			__journal_refile_buffer(jh);
-			if (!jh->b_transaction) {
-				jbd_unlock_bh_state(bh);
-				 /* needs a brelse */
-				journal_remove_journal_head(bh);
-				release_buffer_page(bh);
-			} else
-				jbd_unlock_bh_state(bh);
+			 * list.
+			 */
+			if (!jh->b_next_transaction)
+				try_to_free = 1;
 		}
+		JBUFFER_TRACE(jh, "refile or unfile freed buffer");
+		__journal_refile_buffer(jh);
+		jbd_unlock_bh_state(bh);
+		if (try_to_free)
+			release_buffer_page(bh);
+		else
+			__brelse(bh);
 		cond_resched_lock(&journal->j_list_lock);
 	}
 	spin_unlock(&journal->j_list_lock);
@@ -946,6 +952,7 @@ restart_loop:
 	}
 	spin_unlock(&journal->j_list_lock);
 
+	trace_jbd_end_commit(journal, commit_transaction);
 	jbd_debug(1, "JBD: commit %d complete, head %d\n",
 		  journal->j_commit_sequence, journal->j_tail_sequence);
 
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index e2d4285..9fe061f 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -38,6 +38,9 @@
 #include <linux/debugfs.h>
 #include <linux/ratelimit.h>
 
+#define CREATE_TRACE_POINTS
+#include <trace/events/jbd.h>
+
 #include <asm/uaccess.h>
 #include <asm/page.h>
 
@@ -1065,6 +1068,7 @@ void journal_update_superblock(journal_t *journal, int wait)
 	} else
 		write_dirty_buffer(bh, WRITE);
 
+	trace_jbd_update_superblock_end(journal, wait);
 out:
 	/* If we have just flushed the log (by marking s_start==0), then
 	 * any future commit will have to be careful to update the
@@ -1799,10 +1803,9 @@ static void journal_free_journal_head(struct journal_head *jh)
  * When a buffer has its BH_JBD bit set it is immune from being released by
  * core kernel code, mainly via ->b_count.
  *
- * A journal_head may be detached from its buffer_head when the journal_head's
- * b_transaction, b_cp_transaction and b_next_transaction pointers are NULL.
- * Various places in JBD call journal_remove_journal_head() to indicate that the
- * journal_head can be dropped if needed.
+ * A journal_head is detached from its buffer_head when the journal_head's
+ * b_jcount reaches zero. Running transaction (b_transaction) and checkpoint
+ * transaction (b_cp_transaction) hold their references to b_jcount.
  *
  * Various places in the kernel want to attach a journal_head to a buffer_head
  * _before_ attaching the journal_head to a transaction.  To protect the
@@ -1815,17 +1818,16 @@ static void journal_free_journal_head(struct journal_head *jh)
  *	(Attach a journal_head if needed.  Increments b_jcount)
  *	struct journal_head *jh = journal_add_journal_head(bh);
  *	...
- *	jh->b_transaction = xxx;
- *	journal_put_journal_head(jh);
- *
- * Now, the journal_head's b_jcount is zero, but it is safe from being released
- * because it has a non-zero b_transaction.
+ *      (Get another reference for transaction)
+ *      journal_grab_journal_head(bh);
+ *      jh->b_transaction = xxx;
+ *      (Put original reference)
+ *      journal_put_journal_head(jh);
  */
 
 /*
  * Give a buffer_head a journal_head.
  *
- * Doesn't need the journal lock.
  * May sleep.
  */
 struct journal_head *journal_add_journal_head(struct buffer_head *bh)
@@ -1889,61 +1891,29 @@ static void __journal_remove_journal_head(struct buffer_head *bh)
 	struct journal_head *jh = bh2jh(bh);
 
 	J_ASSERT_JH(jh, jh->b_jcount >= 0);
-
-	get_bh(bh);
-	if (jh->b_jcount == 0) {
-		if (jh->b_transaction == NULL &&
-				jh->b_next_transaction == NULL &&
-				jh->b_cp_transaction == NULL) {
-			J_ASSERT_JH(jh, jh->b_jlist == BJ_None);
-			J_ASSERT_BH(bh, buffer_jbd(bh));
-			J_ASSERT_BH(bh, jh2bh(jh) == bh);
-			BUFFER_TRACE(bh, "remove journal_head");
-			if (jh->b_frozen_data) {
-				printk(KERN_WARNING "%s: freeing "
-						"b_frozen_data\n",
-						__func__);
-				jbd_free(jh->b_frozen_data, bh->b_size);
-			}
-			if (jh->b_committed_data) {
-				printk(KERN_WARNING "%s: freeing "
-						"b_committed_data\n",
-						__func__);
-				jbd_free(jh->b_committed_data, bh->b_size);
-			}
-			bh->b_private = NULL;
-			jh->b_bh = NULL;	/* debug, really */
-			clear_buffer_jbd(bh);
-			__brelse(bh);
-			journal_free_journal_head(jh);
-		} else {
-			BUFFER_TRACE(bh, "journal_head was locked");
-		}
+	J_ASSERT_JH(jh, jh->b_transaction == NULL);
+	J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
+	J_ASSERT_JH(jh, jh->b_cp_transaction == NULL);
+	J_ASSERT_JH(jh, jh->b_jlist == BJ_None);
+	J_ASSERT_BH(bh, buffer_jbd(bh));
+	J_ASSERT_BH(bh, jh2bh(jh) == bh);
+	BUFFER_TRACE(bh, "remove journal_head");
+	if (jh->b_frozen_data) {
+		printk(KERN_WARNING "%s: freeing b_frozen_data\n", __func__);
+		jbd_free(jh->b_frozen_data, bh->b_size);
 	}
+	if (jh->b_committed_data) {
+		printk(KERN_WARNING "%s: freeing b_committed_data\n", __func__);
+		jbd_free(jh->b_committed_data, bh->b_size);
+	}
+	bh->b_private = NULL;
+	jh->b_bh = NULL;	/* debug, really */
+	clear_buffer_jbd(bh);
+	journal_free_journal_head(jh);
 }
 
 /*
- * journal_remove_journal_head(): if the buffer isn't attached to a transaction
- * and has a zero b_jcount then remove and release its journal_head.   If we did
- * see that the buffer is not used by any transaction we also "logically"
- * decrement ->b_count.
- *
- * We in fact take an additional increment on ->b_count as a convenience,
- * because the caller usually wants to do additional things with the bh
- * after calling here.
- * The caller of journal_remove_journal_head() *must* run __brelse(bh) at some
- * time.  Once the caller has run __brelse(), the buffer is eligible for
- * reaping by try_to_free_buffers().
- */
-void journal_remove_journal_head(struct buffer_head *bh)
-{
-	jbd_lock_bh_journal_head(bh);
-	__journal_remove_journal_head(bh);
-	jbd_unlock_bh_journal_head(bh);
-}
-
-/*
- * Drop a reference on the passed journal_head.  If it fell to zero then try to
+ * Drop a reference on the passed journal_head.  If it fell to zero then
  * release the journal_head from the buffer_head.
  */
 void journal_put_journal_head(struct journal_head *jh)
@@ -1953,11 +1923,12 @@ void journal_put_journal_head(struct journal_head *jh)
 	jbd_lock_bh_journal_head(bh);
 	J_ASSERT_JH(jh, jh->b_jcount > 0);
 	--jh->b_jcount;
-	if (!jh->b_jcount && !jh->b_transaction) {
+	if (!jh->b_jcount) {
 		__journal_remove_journal_head(bh);
+		jbd_unlock_bh_journal_head(bh);
 		__brelse(bh);
-	}
-	jbd_unlock_bh_journal_head(bh);
+	} else
+		jbd_unlock_bh_journal_head(bh);
 }
 
 /*
diff --git a/fs/jbd/transaction.c b/fs/jbd/transaction.c
index f7ee81a..7e59c6e 100644
--- a/fs/jbd/transaction.c
+++ b/fs/jbd/transaction.c
@@ -26,6 +26,7 @@
 #include <linux/mm.h>
 #include <linux/highmem.h>
 #include <linux/hrtimer.h>
+#include <linux/backing-dev.h>
 
 static void __journal_temp_unlink_buffer(struct journal_head *jh);
 
@@ -99,11 +100,10 @@ static int start_this_handle(journal_t *journal, handle_t *handle)
 
 alloc_transaction:
 	if (!journal->j_running_transaction) {
-		new_transaction = kzalloc(sizeof(*new_transaction),
-						GFP_NOFS|__GFP_NOFAIL);
+		new_transaction = kzalloc(sizeof(*new_transaction), GFP_NOFS);
 		if (!new_transaction) {
-			ret = -ENOMEM;
-			goto out;
+			congestion_wait(BLK_RW_ASYNC, HZ/50);
+			goto alloc_transaction;
 		}
 	}
 
@@ -696,7 +696,6 @@ repeat:
 	if (!jh->b_transaction) {
 		JBUFFER_TRACE(jh, "no transaction");
 		J_ASSERT_JH(jh, !jh->b_next_transaction);
-		jh->b_transaction = transaction;
 		JBUFFER_TRACE(jh, "file as BJ_Reserved");
 		spin_lock(&journal->j_list_lock);
 		__journal_file_buffer(jh, transaction, BJ_Reserved);
@@ -818,7 +817,6 @@ int journal_get_create_access(handle_t *handle, struct buffer_head *bh)
 		 * committed and so it's safe to clear the dirty bit.
 		 */
 		clear_buffer_dirty(jh2bh(jh));
-		jh->b_transaction = transaction;
 
 		/* first access by this transaction */
 		jh->b_modified = 0;
@@ -844,8 +842,8 @@ int journal_get_create_access(handle_t *handle, struct buffer_head *bh)
 	 */
 	JBUFFER_TRACE(jh, "cancelling revoke");
 	journal_cancel_revoke(handle, jh);
-	journal_put_journal_head(jh);
 out:
+	journal_put_journal_head(jh);
 	return err;
 }
 
@@ -1069,8 +1067,9 @@ int journal_dirty_data(handle_t *handle, struct buffer_head *bh)
 				ret = -EIO;
 				goto no_journal;
 			}
-
-			if (jh->b_transaction != NULL) {
+			/* We might have slept so buffer could be refiled now */
+			if (jh->b_transaction != NULL &&
+			    jh->b_transaction != handle->h_transaction) {
 				JBUFFER_TRACE(jh, "unfile from commit");
 				__journal_temp_unlink_buffer(jh);
 				/* It still points to the committing
@@ -1091,8 +1090,6 @@ int journal_dirty_data(handle_t *handle, struct buffer_head *bh)
 		if (jh->b_jlist != BJ_SyncData && jh->b_jlist != BJ_Locked) {
 			JBUFFER_TRACE(jh, "not on correct data list: unfile");
 			J_ASSERT_JH(jh, jh->b_jlist != BJ_Shadow);
-			__journal_temp_unlink_buffer(jh);
-			jh->b_transaction = handle->h_transaction;
 			JBUFFER_TRACE(jh, "file as data");
 			__journal_file_buffer(jh, handle->h_transaction,
 						BJ_SyncData);
@@ -1300,8 +1297,6 @@ int journal_forget (handle_t *handle, struct buffer_head *bh)
 			__journal_file_buffer(jh, transaction, BJ_Forget);
 		} else {
 			__journal_unfile_buffer(jh);
-			journal_remove_journal_head(bh);
-			__brelse(bh);
 			if (!buffer_jbd(bh)) {
 				spin_unlock(&journal->j_list_lock);
 				jbd_unlock_bh_state(bh);
@@ -1622,19 +1617,32 @@ static void __journal_temp_unlink_buffer(struct journal_head *jh)
 		mark_buffer_dirty(bh);	/* Expose it to the VM */
 }
 
+/*
+ * Remove buffer from all transactions.
+ *
+ * Called with bh_state lock and j_list_lock
+ *
+ * jh and bh may be already freed when this function returns.
+ */
 void __journal_unfile_buffer(struct journal_head *jh)
 {
 	__journal_temp_unlink_buffer(jh);
 	jh->b_transaction = NULL;
+	journal_put_journal_head(jh);
 }
 
 void journal_unfile_buffer(journal_t *journal, struct journal_head *jh)
 {
-	jbd_lock_bh_state(jh2bh(jh));
+	struct buffer_head *bh = jh2bh(jh);
+
+	/* Get reference so that buffer cannot be freed before we unlock it */
+	get_bh(bh);
+	jbd_lock_bh_state(bh);
 	spin_lock(&journal->j_list_lock);
 	__journal_unfile_buffer(jh);
 	spin_unlock(&journal->j_list_lock);
-	jbd_unlock_bh_state(jh2bh(jh));
+	jbd_unlock_bh_state(bh);
+	__brelse(bh);
 }
 
 /*
@@ -1661,16 +1669,12 @@ __journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh)
 			/* A written-back ordered data buffer */
 			JBUFFER_TRACE(jh, "release data");
 			__journal_unfile_buffer(jh);
-			journal_remove_journal_head(bh);
-			__brelse(bh);
 		}
 	} else if (jh->b_cp_transaction != NULL && jh->b_transaction == NULL) {
 		/* written-back checkpointed metadata buffer */
 		if (jh->b_jlist == BJ_None) {
 			JBUFFER_TRACE(jh, "remove from checkpoint list");
 			__journal_remove_checkpoint(jh);
-			journal_remove_journal_head(bh);
-			__brelse(bh);
 		}
 	}
 	spin_unlock(&journal->j_list_lock);
@@ -1733,7 +1737,7 @@ int journal_try_to_free_buffers(journal_t *journal,
 		/*
 		 * We take our own ref against the journal_head here to avoid
 		 * having to add tons of locking around each instance of
-		 * journal_remove_journal_head() and journal_put_journal_head().
+		 * journal_put_journal_head().
 		 */
 		jh = journal_grab_journal_head(bh);
 		if (!jh)
@@ -1770,10 +1774,9 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
 	int may_free = 1;
 	struct buffer_head *bh = jh2bh(jh);
 
-	__journal_unfile_buffer(jh);
-
 	if (jh->b_cp_transaction) {
 		JBUFFER_TRACE(jh, "on running+cp transaction");
+		__journal_temp_unlink_buffer(jh);
 		/*
 		 * We don't want to write the buffer anymore, clear the
 		 * bit so that we don't confuse checks in
@@ -1784,8 +1787,7 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
 		may_free = 0;
 	} else {
 		JBUFFER_TRACE(jh, "on running transaction");
-		journal_remove_journal_head(bh);
-		__brelse(bh);
+		__journal_unfile_buffer(jh);
 	}
 	return may_free;
 }
@@ -2070,6 +2072,8 @@ void __journal_file_buffer(struct journal_head *jh,
 
 	if (jh->b_transaction)
 		__journal_temp_unlink_buffer(jh);
+	else
+		journal_grab_journal_head(bh);
 	jh->b_transaction = transaction;
 
 	switch (jlist) {
@@ -2127,9 +2131,10 @@ void journal_file_buffer(struct journal_head *jh,
  * already started to be used by a subsequent transaction, refile the
  * buffer on that transaction's metadata list.
  *
- * Called under journal->j_list_lock
- *
+ * Called under j_list_lock
  * Called under jbd_lock_bh_state(jh2bh(jh))
+ *
+ * jh and bh may be already free when this function returns
  */
 void __journal_refile_buffer(struct journal_head *jh)
 {
@@ -2153,6 +2158,11 @@ void __journal_refile_buffer(struct journal_head *jh)
 
 	was_dirty = test_clear_buffer_jbddirty(bh);
 	__journal_temp_unlink_buffer(jh);
+	/*
+	 * We set b_transaction here because b_next_transaction will inherit
+	 * our jh reference and thus __journal_file_buffer() must not take a
+	 * new one.
+	 */
 	jh->b_transaction = jh->b_next_transaction;
 	jh->b_next_transaction = NULL;
 	if (buffer_freed(bh))
@@ -2169,30 +2179,21 @@ void __journal_refile_buffer(struct journal_head *jh)
 }
 
 /*
- * For the unlocked version of this call, also make sure that any
- * hanging journal_head is cleaned up if necessary.
- *
- * __journal_refile_buffer is usually called as part of a single locked
- * operation on a buffer_head, in which the caller is probably going to
- * be hooking the journal_head onto other lists.  In that case it is up
- * to the caller to remove the journal_head if necessary.  For the
- * unlocked journal_refile_buffer call, the caller isn't going to be
- * doing anything else to the buffer so we need to do the cleanup
- * ourselves to avoid a jh leak.
- *
- * *** The journal_head may be freed by this call! ***
+ * __journal_refile_buffer() with necessary locking added. We take our bh
+ * reference so that we can safely unlock bh.
+ *
+ * The jh and bh may be freed by this call.
  */
 void journal_refile_buffer(journal_t *journal, struct journal_head *jh)
 {
 	struct buffer_head *bh = jh2bh(jh);
 
+	/* Get reference so that buffer cannot be freed before we unlock it */
+	get_bh(bh);
 	jbd_lock_bh_state(bh);
 	spin_lock(&journal->j_list_lock);
-
 	__journal_refile_buffer(jh);
 	jbd_unlock_bh_state(bh);
-	journal_remove_journal_head(bh);
-
 	spin_unlock(&journal->j_list_lock);
 	__brelse(bh);
 }