[PATCH v3 2/2] ext4: fix race between writepages and enabling EXT4_EXTENTS_FL

[Eric Biggers <ebiggers@kernel.org>]

From: Eric Biggers <ebiggers@google.com>

If EXT4_EXTENTS_FL is set on an inode while ext4_writepages() is running
on it, the following warning in ext4_add_complete_io() can be hit:

WARNING: CPU: 1 PID: 0 at fs/ext4/page-io.c:234 ext4_put_io_end_defer+0xf0/0x120

Here's a minimal reproducer (not 100% reliable) (root isn't required):

        while true; do
                sync
        done &
        while true; do
                rm -f file
                touch file
                chattr -e file
                echo X >> file
                chattr +e file
        done

The problem is that in ext4_writepages(), ext4_should_dioread_nolock()
(which only returns true on extent-based files) is checked once to set
the number of reserved journal credits, and also again later to select
the flags for ext4_map_blocks() and copy the reserved journal handle to
ext4_io_end::handle.  But if EXT4_EXTENTS_FL is being concurrently set,
the first check can see dioread_nolock disabled while the later one can
see it enabled, causing the reserved handle to unexpectedly be NULL.

Since changing EXT4_EXTENTS_FL is uncommon, and there may be other races
related to doing so as well, fix this by synchronizing changing
EXT4_EXTENTS_FL with ext4_writepages() via the existing
s_writepages_rwsem (previously called s_journal_flag_rwsem).

This was originally reported by syzbot without a reproducer at
https://syzkaller.appspot.com/bug?extid=2202a584a00fffd19fbf,
but now that dioread_nolock is the default I also started seeing this
when running syzkaller locally.

Reported-by: syzbot+2202a584a00fffd19fbf@syzkaller.appspotmail.com
Fixes: 6b523df4fb5a ("ext4: use transaction reservation for extent conversion in ext4_end_io")
Cc: stable@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
---
 fs/ext4/ext4.h    |  5 ++++-
 fs/ext4/migrate.c | 27 +++++++++++++++++++--------
 2 files changed, 23 insertions(+), 9 deletions(-)

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 487a7b430b9dd..0a59006c621a0 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1552,7 +1552,10 @@ struct ext4_sb_info {
 	struct ratelimit_state s_warning_ratelimit_state;
 	struct ratelimit_state s_msg_ratelimit_state;
 
-	/* Barrier between changing inodes' journal flags and writepages ops. */
+	/*
+	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
+	 * or EXTENTS flag.
+	 */
 	struct percpu_rw_semaphore s_writepages_rwsem;
 	struct dax_device *s_daxdev;
 #ifdef CONFIG_EXT4_DEBUG
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
index 89725fa425732..fb6520f371355 100644
--- a/fs/ext4/migrate.c
+++ b/fs/ext4/migrate.c
@@ -407,6 +407,7 @@ static int free_ext_block(handle_t *handle, struct inode *inode)
 
 int ext4_ext_migrate(struct inode *inode)
 {
+	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
 	handle_t *handle;
 	int retval = 0, i;
 	__le32 *i_data;
@@ -431,6 +432,8 @@ int ext4_ext_migrate(struct inode *inode)
 		 */
 		return retval;
 
+	percpu_down_write(&sbi->s_writepages_rwsem);
+
 	/*
 	 * Worst case we can touch the allocation bitmaps, a bgd
 	 * block, and a block to link in the orphan list.  We do need
@@ -441,7 +444,7 @@ int ext4_ext_migrate(struct inode *inode)
 
 	if (IS_ERR(handle)) {
 		retval = PTR_ERR(handle);
-		return retval;
+		goto out_unlock;
 	}
 	goal = (((inode->i_ino - 1) / EXT4_INODES_PER_GROUP(inode->i_sb)) *
 		EXT4_INODES_PER_GROUP(inode->i_sb)) + 1;
@@ -452,7 +455,7 @@ int ext4_ext_migrate(struct inode *inode)
 	if (IS_ERR(tmp_inode)) {
 		retval = PTR_ERR(tmp_inode);
 		ext4_journal_stop(handle);
-		return retval;
+		goto out_unlock;
 	}
 	i_size_write(tmp_inode, i_size_read(inode));
 	/*
@@ -494,7 +497,7 @@ int ext4_ext_migrate(struct inode *inode)
 		 */
 		ext4_orphan_del(NULL, tmp_inode);
 		retval = PTR_ERR(handle);
-		goto out;
+		goto out_tmp_inode;
 	}
 
 	ei = EXT4_I(inode);
@@ -576,10 +579,11 @@ int ext4_ext_migrate(struct inode *inode)
 	ext4_ext_tree_init(handle, tmp_inode);
 out_stop:
 	ext4_journal_stop(handle);
-out:
+out_tmp_inode:
 	unlock_new_inode(tmp_inode);
 	iput(tmp_inode);
-
+out_unlock:
+	percpu_up_write(&sbi->s_writepages_rwsem);
 	return retval;
 }
 
@@ -589,7 +593,8 @@ int ext4_ext_migrate(struct inode *inode)
 int ext4_ind_migrate(struct inode *inode)
 {
 	struct ext4_extent_header	*eh;
-	struct ext4_super_block		*es = EXT4_SB(inode->i_sb)->s_es;
+	struct ext4_sb_info		*sbi = EXT4_SB(inode->i_sb);
+	struct ext4_super_block		*es = sbi->s_es;
 	struct ext4_inode_info		*ei = EXT4_I(inode);
 	struct ext4_extent		*ex;
 	unsigned int			i, len;
@@ -613,9 +618,13 @@ int ext4_ind_migrate(struct inode *inode)
 	if (test_opt(inode->i_sb, DELALLOC))
 		ext4_alloc_da_blocks(inode);
 
+	percpu_down_write(&sbi->s_writepages_rwsem);
+
 	handle = ext4_journal_start(inode, EXT4_HT_MIGRATE, 1);
-	if (IS_ERR(handle))
-		return PTR_ERR(handle);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		goto out_unlock;
+	}
 
 	down_write(&EXT4_I(inode)->i_data_sem);
 	ret = ext4_ext_check_inode(inode);
@@ -650,5 +659,7 @@ int ext4_ind_migrate(struct inode *inode)
 errout:
 	ext4_journal_stop(handle);
 	up_write(&EXT4_I(inode)->i_data_sem);
+out_unlock:
+	percpu_up_write(&sbi->s_writepages_rwsem);
 	return ret;
 }
-- 
2.25.0