Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  jbd2: Fix oops in jbd2_journal_remove_journal_head()
  jbd2: Remove obsolete parameters in the comments for some jbd2 functions
  ext4: fixed tracepoints cleanup
  ext4: use FIEMAP_EXTENT_LAST flag for last extent in fiemap
  ext4: Fix max file size and logical block counting of extent format file
  ext4: correct comments for ext4_free_blocks()

10 files changed, 147 insertions, 160 deletions

diff --git a/fs/ext4/ext4_extents.h b/fs/ext4/ext4_extents.h
index 2e29abb30f7..095c36f3b61 100644
--- a/fs/ext4/ext4_extents.h
+++ b/fs/ext4/ext4_extents.h
@@ -125,7 +125,7 @@ struct ext4_ext_path {
  * positive retcode - signal for ext4_ext_walk_space(), see below
  * callback must return valid extent (passed or newly created)
  */
-typedef int (*ext_prepare_callback)(struct inode *, struct ext4_ext_path *,
+typedef int (*ext_prepare_callback)(struct inode *, ext4_lblk_t,
 					struct ext4_ext_cache *,
 					struct ext4_extent *, void *);
 
@@ -133,8 +133,11 @@ typedef int (*ext_prepare_callback)(struct inode *, struct ext4_ext_path *,
 #define EXT_BREAK      1
 #define EXT_REPEAT     2
 
-/* Maximum logical block in a file; ext4_extent's ee_block is __le32 */
-#define EXT_MAX_BLOCK	0xffffffff
+/*
+ * Maximum number of logical blocks in a file; ext4_extent's ee_block is
+ * __le32.
+ */
+#define EXT_MAX_BLOCKS	0xffffffff
 
 /*
  * EXT_INIT_MAX_LEN is the maximum number of blocks we can have in an
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 5199bac7fc6..f815cc81e7a 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -1408,7 +1408,7 @@ got_index:
 
 /*
  * ext4_ext_next_allocated_block:
- * returns allocated block in subsequent extent or EXT_MAX_BLOCK.
+ * returns allocated block in subsequent extent or EXT_MAX_BLOCKS.
  * NOTE: it considers block number from index entry as
  * allocated block. Thus, index entries have to be consistent
  * with leaves.
@@ -1422,7 +1422,7 @@ ext4_ext_next_allocated_block(struct ext4_ext_path *path)
 	depth = path->p_depth;
 
 	if (depth == 0 && path->p_ext == NULL)
-		return EXT_MAX_BLOCK;
+		return EXT_MAX_BLOCKS;
 
 	while (depth >= 0) {
 		if (depth == path->p_depth) {
@@ -1439,12 +1439,12 @@ ext4_ext_next_allocated_block(struct ext4_ext_path *path)
 		depth--;
 	}
 
-	return EXT_MAX_BLOCK;
+	return EXT_MAX_BLOCKS;
 }
 
 /*
  * ext4_ext_next_leaf_block:
- * returns first allocated block from next leaf or EXT_MAX_BLOCK
+ * returns first allocated block from next leaf or EXT_MAX_BLOCKS
  */
 static ext4_lblk_t ext4_ext_next_leaf_block(struct inode *inode,
 					struct ext4_ext_path *path)
@@ -1456,7 +1456,7 @@ static ext4_lblk_t ext4_ext_next_leaf_block(struct inode *inode,
 
 	/* zero-tree has no leaf blocks at all */
 	if (depth == 0)
-		return EXT_MAX_BLOCK;
+		return EXT_MAX_BLOCKS;
 
 	/* go to index block */
 	depth--;
@@ -1469,7 +1469,7 @@ static ext4_lblk_t ext4_ext_next_leaf_block(struct inode *inode,
 		depth--;
 	}
 
-	return EXT_MAX_BLOCK;
+	return EXT_MAX_BLOCKS;
 }
 
 /*
@@ -1677,13 +1677,13 @@ static unsigned int ext4_ext_check_overlap(struct inode *inode,
 	 */
 	if (b2 < b1) {
 		b2 = ext4_ext_next_allocated_block(path);
-		if (b2 == EXT_MAX_BLOCK)
+		if (b2 == EXT_MAX_BLOCKS)
 			goto out;
 	}
 
 	/* check for wrap through zero on extent logical start block*/
 	if (b1 + len1 < b1) {
-		len1 = EXT_MAX_BLOCK - b1;
+		len1 = EXT_MAX_BLOCKS - b1;
 		newext->ee_len = cpu_to_le16(len1);
 		ret = 1;
 	}
@@ -1767,7 +1767,7 @@ repeat:
 	fex = EXT_LAST_EXTENT(eh);
 	next = ext4_ext_next_leaf_block(inode, path);
 	if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block)
-	    && next != EXT_MAX_BLOCK) {
+	    && next != EXT_MAX_BLOCKS) {
 		ext_debug("next leaf block - %d\n", next);
 		BUG_ON(npath != NULL);
 		npath = ext4_ext_find_extent(inode, next, NULL);
@@ -1887,7 +1887,7 @@ static int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
 	BUG_ON(func == NULL);
 	BUG_ON(inode == NULL);
 
-	while (block < last && block != EXT_MAX_BLOCK) {
+	while (block < last && block != EXT_MAX_BLOCKS) {
 		num = last - block;
 		/* find extent for this block */
 		down_read(&EXT4_I(inode)->i_data_sem);
@@ -1958,7 +1958,7 @@ static int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
 			err = -EIO;
 			break;
 		}
-		err = func(inode, path, &cbex, ex, cbdata);
+		err = func(inode, next, &cbex, ex, cbdata);
 		ext4_ext_drop_refs(path);
 
 		if (err < 0)
@@ -2020,7 +2020,7 @@ ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
 	if (ex == NULL) {
 		/* there is no extent yet, so gap is [0;-] */
 		lblock = 0;
-		len = EXT_MAX_BLOCK;
+		len = EXT_MAX_BLOCKS;
 		ext_debug("cache gap(whole file):");
 	} else if (block < le32_to_cpu(ex->ee_block)) {
 		lblock = block;
@@ -2350,7 +2350,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 			 * never happen because at least one of the end points
 			 * needs to be on the edge of the extent.
 			 */
-			if (end == EXT_MAX_BLOCK) {
+			if (end == EXT_MAX_BLOCKS - 1) {
 				ext_debug("  bad truncate %u:%u\n",
 						start, end);
 				block = 0;
@@ -2398,7 +2398,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 			 * If this is a truncate, this condition
 			 * should never happen
 			 */
-			if (end == EXT_MAX_BLOCK) {
+			if (end == EXT_MAX_BLOCKS - 1) {
 				ext_debug("  bad truncate %u:%u\n",
 					start, end);
 				err = -EIO;
@@ -2478,7 +2478,7 @@ ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 		 * we need to remove it from the leaf
 		 */
 		if (num == 0) {
-			if (end != EXT_MAX_BLOCK) {
+			if (end != EXT_MAX_BLOCKS - 1) {
 				/*
 				 * For hole punching, we need to scoot all the
 				 * extents up when an extent is removed so that
@@ -3699,7 +3699,7 @@ void ext4_ext_truncate(struct inode *inode)
 
 	last_block = (inode->i_size + sb->s_blocksize - 1)
 			>> EXT4_BLOCK_SIZE_BITS(sb);
-	err = ext4_ext_remove_space(inode, last_block, EXT_MAX_BLOCK);
+	err = ext4_ext_remove_space(inode, last_block, EXT_MAX_BLOCKS - 1);
 
 	/* In a multi-transaction truncate, we only make the final
 	 * transaction synchronous.
@@ -3914,14 +3914,13 @@ int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
 /*
  * Callback function called for each extent to gather FIEMAP information.
  */
-static int ext4_ext_fiemap_cb(struct inode *inode, struct ext4_ext_path *path,
+static int ext4_ext_fiemap_cb(struct inode *inode, ext4_lblk_t next,
 		       struct ext4_ext_cache *newex, struct ext4_extent *ex,
 		       void *data)
 {
 	__u64	logical;
 	__u64	physical;
 	__u64	length;
-	loff_t	size;
 	__u32	flags = 0;
 	int		ret = 0;
 	struct fiemap_extent_info *fieinfo = data;
@@ -4103,8 +4102,7 @@ found_delayed_extent:
 	if (ex && ext4_ext_is_uninitialized(ex))
 		flags |= FIEMAP_EXTENT_UNWRITTEN;
 
-	size = i_size_read(inode);
-	if (logical + length >= size)
+	if (next == EXT_MAX_BLOCKS)
 		flags |= FIEMAP_EXTENT_LAST;
 
 	ret = fiemap_fill_next_extent(fieinfo, logical, physical,
@@ -4347,8 +4345,8 @@ int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 
 		start_blk = start >> inode->i_sb->s_blocksize_bits;
 		last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
-		if (last_blk >= EXT_MAX_BLOCK)
-			last_blk = EXT_MAX_BLOCK-1;
+		if (last_blk >= EXT_MAX_BLOCKS)
+			last_blk = EXT_MAX_BLOCKS-1;
 		len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
 
 		/*
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index a5763e3505b..e3126c05100 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2634,7 +2634,7 @@ static int ext4_writepage(struct page *page,
 	struct buffer_head *page_bufs = NULL;
 	struct inode *inode = page->mapping->host;
 
-	trace_ext4_writepage(inode, page);
+	trace_ext4_writepage(page);
 	size = i_size_read(inode);
 	if (page->index == size >> PAGE_CACHE_SHIFT)
 		len = size & ~PAGE_CACHE_MASK;
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 859f2ae8864..6ed859d5685 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -3578,8 +3578,8 @@ ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh,
 		free += next - bit;
 
 		trace_ext4_mballoc_discard(sb, NULL, group, bit, next - bit);
-		trace_ext4_mb_release_inode_pa(sb, pa->pa_inode, pa,
-					       grp_blk_start + bit, next - bit);
+		trace_ext4_mb_release_inode_pa(pa, grp_blk_start + bit,
+					       next - bit);
 		mb_free_blocks(pa->pa_inode, e4b, bit, next - bit);
 		bit = next + 1;
 	}
@@ -3608,7 +3608,7 @@ ext4_mb_release_group_pa(struct ext4_buddy *e4b,
 	ext4_group_t group;
 	ext4_grpblk_t bit;
 
-	trace_ext4_mb_release_group_pa(sb, pa);
+	trace_ext4_mb_release_group_pa(pa);
 	BUG_ON(pa->pa_deleted == 0);
 	ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
 	BUG_ON(group != e4b->bd_group && pa->pa_len != 0);
@@ -4448,7 +4448,7 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
  * @inode:		inode
  * @block:		start physical block to free
  * @count:		number of blocks to count
- * @metadata: 		Are these metadata blocks
+ * @flags:		flags used by ext4_free_blocks
  */
 void ext4_free_blocks(handle_t *handle, struct inode *inode,
 		      struct buffer_head *bh, ext4_fsblk_t block,
diff --git a/fs/ext4/move_extent.c b/fs/ext4/move_extent.c
index 2b8304bf3c5..f57455a1b1b 100644
--- a/fs/ext4/move_extent.c
+++ b/fs/ext4/move_extent.c
@@ -1002,12 +1002,12 @@ mext_check_arguments(struct inode *orig_inode,
 		return -EINVAL;
 	}
 
-	if ((orig_start > EXT_MAX_BLOCK) ||
-	    (donor_start > EXT_MAX_BLOCK) ||
-	    (*len > EXT_MAX_BLOCK) ||
-	    (orig_start + *len > EXT_MAX_BLOCK))  {
+	if ((orig_start >= EXT_MAX_BLOCKS) ||
+	    (donor_start >= EXT_MAX_BLOCKS) ||
+	    (*len > EXT_MAX_BLOCKS) ||
+	    (orig_start + *len >= EXT_MAX_BLOCKS))  {
 		ext4_debug("ext4 move extent: Can't handle over [%u] blocks "
-			"[ino:orig %lu, donor %lu]\n", EXT_MAX_BLOCK,
+			"[ino:orig %lu, donor %lu]\n", EXT_MAX_BLOCKS,
 			orig_inode->i_ino, donor_inode->i_ino);
 		return -EINVAL;
 	}
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index cc5c157aa11..9ea71aa864b 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -2243,6 +2243,12 @@ static void ext4_orphan_cleanup(struct super_block *sb,
  * in the vfs.  ext4 inode has 48 bits of i_block in fsblock units,
  * so that won't be a limiting factor.
  *
+ * However there is other limiting factor. We do store extents in the form
+ * of starting block and length, hence the resulting length of the extent
+ * covering maximum file size must fit into on-disk format containers as
+ * well. Given that length is always by 1 unit bigger than max unit (because
+ * we count 0 as well) we have to lower the s_maxbytes by one fs block.
+ *
  * Note, this does *not* consider any metadata overhead for vfs i_blocks.
  */
 static loff_t ext4_max_size(int blkbits, int has_huge_files)
@@ -2264,10 +2270,13 @@ static loff_t ext4_max_size(int blkbits, int has_huge_files)
 		upper_limit <<= blkbits;
 	}
 
-	/* 32-bit extent-start container, ee_block */
-	res = 1LL << 32;
+	/*
+	 * 32-bit extent-start container, ee_block. We lower the maxbytes
+	 * by one fs block, so ee_len can cover the extent of maximum file
+	 * size
+	 */
+	res = (1LL << 32) - 1;
 	res <<= blkbits;
-	res -= 1;
 
 	/* Sanity check against vm- & vfs- imposed limits */
 	if (res > upper_limit)
diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c
index 6a79fd0a1a3..2c62c5aae82 100644
--- a/fs/jbd2/checkpoint.c
+++ b/fs/jbd2/checkpoint.c
@@ -97,10 +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 = __jbd2_journal_remove_checkpoint(jh) + 1;
 		jbd_unlock_bh_state(bh);
-		jbd2_journal_remove_journal_head(bh);
 		BUFFER_TRACE(bh, "release");
 		__brelse(bh);
 	} else {
@@ -223,8 +227,8 @@ restart:
 			spin_lock(&journal->j_list_lock);
 			goto restart;
 		}
+		get_bh(bh);
 		if (buffer_locked(bh)) {
-			atomic_inc(&bh->b_count);
 			spin_unlock(&journal->j_list_lock);
 			jbd_unlock_bh_state(bh);
 			wait_on_buffer(bh);
@@ -243,7 +247,6 @@ restart:
 		 */
 		released = __jbd2_journal_remove_checkpoint(jh);
 		jbd_unlock_bh_state(bh);
-		jbd2_journal_remove_journal_head(bh);
 		__brelse(bh);
 	}
 
@@ -284,7 +287,7 @@ static int __process_buffer(journal_t *journal, struct journal_head *jh,
 	int ret = 0;
 
 	if (buffer_locked(bh)) {
-		atomic_inc(&bh->b_count);
+		get_bh(bh);
 		spin_unlock(&journal->j_list_lock);
 		jbd_unlock_bh_state(bh);
 		wait_on_buffer(bh);
@@ -316,12 +319,12 @@ 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");
 		__jbd2_journal_remove_checkpoint(jh);
 		spin_unlock(&journal->j_list_lock);
 		jbd_unlock_bh_state(bh);
-		jbd2_journal_remove_journal_head(bh);
 		__brelse(bh);
 	} else {
 		/*
@@ -554,7 +557,8 @@ int jbd2_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.
@@ -663,8 +667,8 @@ 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))
  */
@@ -684,13 +688,14 @@ int __jbd2_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;
+	jbd2_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
@@ -701,10 +706,8 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
 	 * The locking here around t_state is a bit sleazy.
 	 * See the comment at the end of jbd2_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 */
@@ -723,7 +726,6 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
 	wake_up(&journal->j_wait_logspace);
 	ret = 1;
 out:
-	JBUFFER_TRACE(jh, "exit");
 	return ret;
 }
 
@@ -742,6 +744,8 @@ void __jbd2_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 */
+	jbd2_journal_grab_journal_head(jh2bh(jh));
 	jh->b_cp_transaction = transaction;
 
 	if (!transaction->t_checkpoint_list) {
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 7f21cf3aaf9..eef6979821a 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -848,10 +848,16 @@ 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);
 
@@ -914,28 +920,27 @@ restart_loop:
 			__jbd2_journal_insert_checkpoint(jh, commit_transaction);
 			if (is_journal_aborted(journal))
 				clear_buffer_jbddirty(bh);
-			JBUFFER_TRACE(jh, "refile for checkpoint writeback");
-			__jbd2_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");
-			__jbd2_journal_refile_buffer(jh);
-			if (!jh->b_transaction) {
-				jbd_unlock_bh_state(bh);
-				 /* needs a brelse */
-				jbd2_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 buffer");
+		__jbd2_journal_refile_buffer(jh);
+		jbd_unlock_bh_state(bh);
+		if (try_to_free)
+			release_buffer_page(bh);	/* Drops bh reference */
+		else
+			__brelse(bh);
 		cond_resched_lock(&journal->j_list_lock);
 	}
 	spin_unlock(&journal->j_list_lock);
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index 9a782699030..0dfa5b598e6 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -2078,10 +2078,9 @@ 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 jbd2_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
@@ -2094,17 +2093,16 @@ static void journal_free_journal_head(struct journal_head *jh)
  *	(Attach a journal_head if needed.  Increments b_jcount)
  *	struct journal_head *jh = jbd2_journal_add_journal_head(bh);
  *	...
+ *      (Get another reference for transaction)
+ *	jbd2_journal_grab_journal_head(bh);
  *	jh->b_transaction = xxx;
+ *	(Put original reference)
  *	jbd2_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.
  */
 
 /*
  * Give a buffer_head a journal_head.
  *
- * Doesn't need the journal lock.
  * May sleep.
  */
 struct journal_head *jbd2_journal_add_journal_head(struct buffer_head *bh)
@@ -2168,61 +2166,29 @@ 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__);
-				jbd2_free(jh->b_frozen_data, bh->b_size);
-			}
-			if (jh->b_committed_data) {
-				printk(KERN_WARNING "%s: freeing "
-						"b_committed_data\n",
-						__func__);
-				jbd2_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__);
+		jbd2_free(jh->b_frozen_data, bh->b_size);
 	}
+	if (jh->b_committed_data) {
+		printk(KERN_WARNING "%s: freeing b_committed_data\n", __func__);
+		jbd2_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);
 }
 
 /*
- * jbd2_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 jbd2_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 jbd2_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 jbd2_journal_put_journal_head(struct journal_head *jh)
@@ -2232,11 +2198,12 @@ void jbd2_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/jbd2/transaction.c b/fs/jbd2/transaction.c
index 3eec82d32fd..2d7109414cd 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -30,6 +30,7 @@
 #include <linux/module.h>
 
 static void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh);
+static void __jbd2_journal_unfile_buffer(struct journal_head *jh);
 
 /*
  * jbd2_get_transaction: obtain a new transaction_t object.
@@ -764,7 +765,6 @@ 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);
 		__jbd2_journal_file_buffer(jh, transaction, BJ_Reserved);
@@ -814,7 +814,6 @@ out:
  * int jbd2_journal_get_write_access() - notify intent to modify a buffer for metadata (not data) update.
  * @handle: transaction to add buffer modifications to
  * @bh:     bh to be used for metadata writes
- * @credits: variable that will receive credits for the buffer
  *
  * Returns an error code or 0 on success.
  *
@@ -896,8 +895,6 @@ int jbd2_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;
 
@@ -932,7 +929,6 @@ out:
  *     non-rewindable consequences
  * @handle: transaction
  * @bh: buffer to undo
- * @credits: store the number of taken credits here (if not NULL)
  *
  * Sometimes there is a need to distinguish between metadata which has
  * been committed to disk and that which has not.  The ext3fs code uses
@@ -1232,8 +1228,6 @@ int jbd2_journal_forget (handle_t *handle, struct buffer_head *bh)
 			__jbd2_journal_file_buffer(jh, transaction, BJ_Forget);
 		} else {
 			__jbd2_journal_unfile_buffer(jh);
-			jbd2_journal_remove_journal_head(bh);
-			__brelse(bh);
 			if (!buffer_jbd(bh)) {
 				spin_unlock(&journal->j_list_lock);
 				jbd_unlock_bh_state(bh);
@@ -1556,19 +1550,32 @@ void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh)
 		mark_buffer_dirty(bh);	/* Expose it to the VM */
 }
 
-void __jbd2_journal_unfile_buffer(struct journal_head *jh)
+/*
+ * 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.
+ */
+static void __jbd2_journal_unfile_buffer(struct journal_head *jh)
 {
 	__jbd2_journal_temp_unlink_buffer(jh);
 	jh->b_transaction = NULL;
+	jbd2_journal_put_journal_head(jh);
 }
 
 void jbd2_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);
 	__jbd2_journal_unfile_buffer(jh);
 	spin_unlock(&journal->j_list_lock);
-	jbd_unlock_bh_state(jh2bh(jh));
+	jbd_unlock_bh_state(bh);
+	__brelse(bh);
 }
 
 /*
@@ -1595,8 +1602,6 @@ __journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh)
 		if (jh->b_jlist == BJ_None) {
 			JBUFFER_TRACE(jh, "remove from checkpoint list");
 			__jbd2_journal_remove_checkpoint(jh);
-			jbd2_journal_remove_journal_head(bh);
-			__brelse(bh);
 		}
 	}
 	spin_unlock(&journal->j_list_lock);
@@ -1659,7 +1664,6 @@ int jbd2_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
-		 * jbd2_journal_remove_journal_head() and
 		 * jbd2_journal_put_journal_head().
 		 */
 		jh = jbd2_journal_grab_journal_head(bh);
@@ -1697,10 +1701,9 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
 	int may_free = 1;
 	struct buffer_head *bh = jh2bh(jh);
 
-	__jbd2_journal_unfile_buffer(jh);
-
 	if (jh->b_cp_transaction) {
 		JBUFFER_TRACE(jh, "on running+cp transaction");
+		__jbd2_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
@@ -1711,8 +1714,7 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
 		may_free = 0;
 	} else {
 		JBUFFER_TRACE(jh, "on running transaction");
-		jbd2_journal_remove_journal_head(bh);
-		__brelse(bh);
+		__jbd2_journal_unfile_buffer(jh);
 	}
 	return may_free;
 }
@@ -1990,6 +1992,8 @@ void __jbd2_journal_file_buffer(struct journal_head *jh,
 
 	if (jh->b_transaction)
 		__jbd2_journal_temp_unlink_buffer(jh);
+	else
+		jbd2_journal_grab_journal_head(bh);
 	jh->b_transaction = transaction;
 
 	switch (jlist) {
@@ -2041,9 +2045,10 @@ void jbd2_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 __jbd2_journal_refile_buffer(struct journal_head *jh)
 {
@@ -2067,6 +2072,11 @@ void __jbd2_journal_refile_buffer(struct journal_head *jh)
 
 	was_dirty = test_clear_buffer_jbddirty(bh);
 	__jbd2_journal_temp_unlink_buffer(jh);
+	/*
+	 * We set b_transaction here because b_next_transaction will inherit
+	 * our jh reference and thus __jbd2_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))
@@ -2083,30 +2093,21 @@ void __jbd2_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.
- *
- * __jbd2_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 jbd2_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! ***
+ * __jbd2_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 jbd2_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);
-
 	__jbd2_journal_refile_buffer(jh);
 	jbd_unlock_bh_state(bh);
-	jbd2_journal_remove_journal_head(bh);
-
 	spin_unlock(&journal->j_list_lock);
 	__brelse(bh);
 }