diff mbox

[v3,2/3] ext4: fix races between changing inode journal mode and ext4_writepages

Message ID 1457510987-31726-2-git-send-email-daeho.jeong@samsung.com
State New, archived
Headers show

Commit Message

Daeho Jeong March 9, 2016, 8:09 a.m. UTC 
Now, in ext4, there is a race condition between changing inode journal
mode and ext4_writepages(). While ext4_writepages() is executed on
a non-journalled mode inode, the inode's journal mode could be enabled
by ioctl() and then, some pages dirtied after switching the journal
mode will be still exposed to ext4_writepages() in non-journaled mode.
To resolve this problem, we use fs-wide per-cpu rw semaphore by
Jan Kara's suggestion because we don't want to waste ext4_inode_info's
space for this extra rare case.

Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
Signed-off-by: Jan Kara <jack@suse.cz>
---
 fs/ext4/ext4.h                |    4 ++++
 fs/ext4/inode.c               |    7 +++++++
 fs/ext4/super.c               |    4 ++++
 kernel/locking/percpu-rwsem.c |    1 +
 4 files changed, 16 insertions(+)

Comments

Jan Kara March 10, 2016, 10:04 a.m. UTC | #1 
On Wed 09-03-16 17:09:46, Daeho Jeong wrote:
> Now, in ext4, there is a race condition between changing inode journal
> mode and ext4_writepages(). While ext4_writepages() is executed on
> a non-journalled mode inode, the inode's journal mode could be enabled
> by ioctl() and then, some pages dirtied after switching the journal
> mode will be still exposed to ext4_writepages() in non-journaled mode.
> To resolve this problem, we use fs-wide per-cpu rw semaphore by
> Jan Kara's suggestion because we don't want to waste ext4_inode_info's
> space for this extra rare case.
> 
> Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
> Signed-off-by: Jan Kara <jack@suse.cz>

The patch is almost fine except for one small issue:

> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 9ecfb76..1176142 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -2476,6 +2476,7 @@ static int ext4_writepages(struct address_space *mapping,
>  	struct blk_plug plug;
>  	bool give_up_on_write = false;
>  
> +	percpu_down_read(&sbi->s_journal_flag_rwsem);
>  	trace_ext4_writepages(inode, wbc);

You need to change how dax_writeback_mapping_range() is called a few lines
below so that it also exits via out_writepages: and not directly.

								Honza
diff mbox

Patch

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 157b458..c757a3d 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -33,6 +33,7 @@ 
 #include <linux/ratelimit.h>
 #include <crypto/hash.h>
 #include <linux/falloc.h>
+#include <linux/percpu-rwsem.h>
 #ifdef __KERNEL__
 #include <linux/compat.h>
 #endif
@@ -1475,6 +1476,9 @@  struct ext4_sb_info {
 	struct ratelimit_state s_err_ratelimit_state;
 	struct ratelimit_state s_warning_ratelimit_state;
 	struct ratelimit_state s_msg_ratelimit_state;
+
+	/* Barrier between changing inodes' journal flags and writepages ops. */
+	struct percpu_rw_semaphore s_journal_flag_rwsem;
 };
 
 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 9ecfb76..1176142 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2476,6 +2476,7 @@  static int ext4_writepages(struct address_space *mapping,
 	struct blk_plug plug;
 	bool give_up_on_write = false;
 
+	percpu_down_read(&sbi->s_journal_flag_rwsem);
 	trace_ext4_writepages(inode, wbc);
 
 	/*
@@ -2646,6 +2647,7 @@  retry:
 out_writepages:
 	trace_ext4_writepages_result(inode, wbc, ret,
 				     nr_to_write - wbc->nr_to_write);
+	percpu_up_read(&sbi->s_journal_flag_rwsem);
 	return ret;
 }
 
@@ -5362,6 +5364,7 @@  int ext4_change_inode_journal_flag(struct inode *inode, int val)
 	journal_t *journal;
 	handle_t *handle;
 	int err;
+	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
 
 	/*
 	 * We have to be very careful here: changing a data block's
@@ -5401,6 +5404,7 @@  int ext4_change_inode_journal_flag(struct inode *inode, int val)
 		}
 	}
 
+	percpu_down_write(&sbi->s_journal_flag_rwsem);
 	jbd2_journal_lock_updates(journal);
 
 	/*
@@ -5417,6 +5421,7 @@  int ext4_change_inode_journal_flag(struct inode *inode, int val)
 		err = jbd2_journal_flush(journal);
 		if (err < 0) {
 			jbd2_journal_unlock_updates(journal);
+			percpu_up_write(&sbi->s_journal_flag_rwsem);
 			ext4_inode_resume_unlocked_dio(inode);
 			return err;
 		}
@@ -5425,6 +5430,8 @@  int ext4_change_inode_journal_flag(struct inode *inode, int val)
 	ext4_set_aops(inode);
 
 	jbd2_journal_unlock_updates(journal);
+	percpu_up_write(&sbi->s_journal_flag_rwsem);
+
 	if (val)
 		up_write(&EXT4_I(inode)->i_mmap_sem);
 	ext4_inode_resume_unlocked_dio(inode);
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 3ed01ec..a12950d 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -861,6 +861,7 @@  static void ext4_put_super(struct super_block *sb)
 	percpu_counter_destroy(&sbi->s_freeinodes_counter);
 	percpu_counter_destroy(&sbi->s_dirs_counter);
 	percpu_counter_destroy(&sbi->s_dirtyclusters_counter);
+	percpu_free_rwsem(&sbi->s_journal_flag_rwsem);
 	brelse(sbi->s_sbh);
 #ifdef CONFIG_QUOTA
 	for (i = 0; i < EXT4_MAXQUOTAS; i++)
@@ -3926,6 +3927,9 @@  no_journal:
 	if (!err)
 		err = percpu_counter_init(&sbi->s_dirtyclusters_counter, 0,
 					  GFP_KERNEL);
+	if (!err)
+		err = percpu_init_rwsem(&sbi->s_journal_flag_rwsem);
+
 	if (err) {
 		ext4_msg(sb, KERN_ERR, "insufficient memory");
 		goto failed_mount6;
diff --git a/kernel/locking/percpu-rwsem.c b/kernel/locking/percpu-rwsem.c
index f231e0b..bec0b64 100644
--- a/kernel/locking/percpu-rwsem.c
+++ b/kernel/locking/percpu-rwsem.c
@@ -37,6 +37,7 @@  void percpu_free_rwsem(struct percpu_rw_semaphore *brw)
 	free_percpu(brw->fast_read_ctr);
 	brw->fast_read_ctr = NULL; /* catch use after free bugs */
 }
+EXPORT_SYMBOL_GPL(percpu_free_rwsem);
 
 /*
  * This is the fast-path for down_read/up_read. If it succeeds we rely