Skip to contentSkip to conversation

ext3: remove lock/unlock super

PATCH In review Marco Stornelli 3 participants 6 messages Aug 16, 2012 – Aug 18, 2012

Subsystems (2)
VFS ext4 file system
lore
Share & tools
lore

Marco Stornelli

author
#0 lore raw

PATCH 3/8 ext3: remove lock/unlock super

From: Marco Stornelli <[email protected]>

Replaced lock and unlock super with a new fs mutex s_lock.

Signed-off-by Marco Stornelli <[email protected]>

---

fs/ext3/ext3.h +1 -0
diff -Nurp linux-3.6-rc1-orig/fs/ext3/ext3.h linux-3.6-rc1/fs/ext3/ext3.h
@@ -672,6 +672,7 @@ struct ext3_sb_info {
 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
 	int s_jquota_fmt;			/* Format of quota to use */
 #endif
+	struct mutex s_lock;
 };
  static inline spinlock_t *
fs/ext3/super.c +6 -5
diff -Nurp linux-3.6-rc1-orig/fs/ext3/super.c linux-3.6-rc1/fs/ext3/super.c
@@ -1939,6 +1939,7 @@ static int ext3_fill_super (struct super
 	sbi->s_gdb_count = db_count;
 	get_random_bytes(&sbi->s_next_generation, sizeof(u32));
 	spin_lock_init(&sbi->s_next_gen_lock);
+	mutex_init(&sbi->s_lock);
  	/* per fileystem reservation list head & lock */
 	spin_lock_init(&sbi->s_rsv_window_lock);

@@ -2582,11 +2583,11 @@ out:
 static int ext3_unfreeze(struct super_block *sb)
 {
 	if (!(sb->s_flags & MS_RDONLY)) {
-		lock_super(sb);
+		mutex_lock(&EXT3_SB(sb)->s_lock);
 		/* Reser the needs_recovery flag before the fs is unlocked. */
 		EXT3_SET_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER);
 		ext3_commit_super(sb, EXT3_SB(sb)->s_es, 1);
-		unlock_super(sb);
+		mutex_unlock(&EXT3_SB(sb)->s_lock);
 		journal_unlock_updates(EXT3_SB(sb)->s_journal);
 	}
 	return 0;

@@ -2606,7 +2607,7 @@ static int ext3_remount (struct super_bl
 #endif
  	/* Store the original options */
-	lock_super(sb);
+	mutex_lock(&EXT3_SB(sb)->s_lock);
 	old_sb_flags = sb->s_flags;
 	old_opts.s_mount_opt = sbi->s_mount_opt;
 	old_opts.s_resuid = sbi->s_resuid;

@@ -2712,7 +2713,7 @@ static int ext3_remount (struct super_bl
 		    old_opts.s_qf_names[i] != sbi->s_qf_names[i])
 			kfree(old_opts.s_qf_names[i]);
 #endif
-	unlock_super(sb);
+	mutex_unlock(&EXT3_SB(sb)->s_lock);
  	if (enable_quota)
 		dquot_resume(sb, -1);

@@ -2732,7 +2733,7 @@ restore_opts:
 		sbi->s_qf_names[i] = old_opts.s_qf_names[i];
 	}
 #endif
-	unlock_super(sb);
+	mutex_unlock(&EXT3_SB(sb)->s_lock);
 	return err;
 }
reviewer
#1 lore raw
Quoted 3 lines · Marco Stornelli

From: Marco Stornelli <[email protected]>

Replaced lock and unlock super with a new fs mutex s_lock.

Hum, is the lock needed at all? Remount & unfreeze both run with s_umount
held for writing. Thus we already have exclusion between these two calls. The same seems to hold for ext4 BTW.

Honza

Quoted 63 lines
Signed-off-by Marco Stornelli <[email protected]>

---

diff -Nurp linux-3.6-rc1-orig/fs/ext3/ext3.h linux-3.6-rc1/fs/ext3/ext3.h

fs/ext3/ext3.h +1 -0
@@ -672,6 +672,7 @@ struct ext3_sb_info {
 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
 	int s_jquota_fmt;			/* Format of quota to use */
 #endif
+	struct mutex s_lock;
 };
  static inline spinlock_t *

diff -Nurp linux-3.6-rc1-orig/fs/ext3/super.c linux-3.6-rc1/fs/ext3/super.c

fs/ext3/super.c +6 -5
@@ -1939,6 +1939,7 @@ static int ext3_fill_super (struct super
 	sbi->s_gdb_count = db_count;
 	get_random_bytes(&sbi->s_next_generation, sizeof(u32));
 	spin_lock_init(&sbi->s_next_gen_lock);
+	mutex_init(&sbi->s_lock);
  	/* per fileystem reservation list head & lock */
 	spin_lock_init(&sbi->s_rsv_window_lock);

@@ -2582,11 +2583,11 @@ out:
 static int ext3_unfreeze(struct super_block *sb)
 {
 	if (!(sb->s_flags & MS_RDONLY)) {
-		lock_super(sb);
+		mutex_lock(&EXT3_SB(sb)->s_lock);
 		/* Reser the needs_recovery flag before the fs is unlocked. */
 		EXT3_SET_INCOMPAT_FEATURE(sb, EXT3_FEATURE_INCOMPAT_RECOVER);
 		ext3_commit_super(sb, EXT3_SB(sb)->s_es, 1);
-		unlock_super(sb);
+		mutex_unlock(&EXT3_SB(sb)->s_lock);
 		journal_unlock_updates(EXT3_SB(sb)->s_journal);
 	}
 	return 0;

@@ -2606,7 +2607,7 @@ static int ext3_remount (struct super_bl
 #endif
  	/* Store the original options */
-	lock_super(sb);
+	mutex_lock(&EXT3_SB(sb)->s_lock);
 	old_sb_flags = sb->s_flags;
 	old_opts.s_mount_opt = sbi->s_mount_opt;
 	old_opts.s_resuid = sbi->s_resuid;

@@ -2712,7 +2713,7 @@ static int ext3_remount (struct super_bl
 		    old_opts.s_qf_names[i] != sbi->s_qf_names[i])
 			kfree(old_opts.s_qf_names[i]);
 #endif
-	unlock_super(sb);
+	mutex_unlock(&EXT3_SB(sb)->s_lock);
  	if (enable_quota)
 		dquot_resume(sb, -1);

@@ -2732,7 +2733,7 @@ restore_opts:
 		sbi->s_qf_names[i] = old_opts.s_qf_names[i];
 	}
 #endif
-	unlock_super(sb);
+	mutex_unlock(&EXT3_SB(sb)->s_lock);
 	return err;
 }
Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR
maintainer
#2 lore raw
Quoted 7 lines · Jan Kara
Quoted 3 lines · Marco Stornelli

From: Marco Stornelli <[email protected]>

Replaced lock and unlock super with a new fs mutex s_lock.

Hum, is the lock needed at all? Remount & unfreeze both run with s_umount
held for writing. Thus we already have exclusion between these two calls. The same seems to hold for ext4 BTW.

Agreed, it's not clear lock_super() is needed at all for ext4 at this point.

- Ted

author
#3 lore raw

Il 16/08/2012 21:19, Theodore Ts'o ha scritto:

Quoted 5 lines · Jan Kara
Quoted 4 lines · Marco Stornelli
Quoted 3 lines

From: Marco Stornelli <[email protected]>

Replaced lock and unlock super with a new fs mutex s_lock.

Quoted 9 lines
Quoted 3 lines

Hum, is the lock needed at all? Remount & unfreeze both run with s_umount
held for writing. Thus we already have exclusion between these two calls. The same seems to hold for ext4 BTW.

Agreed, it's not clear lock_super() is needed at all for ext4 at this point.

- Ted

Great. I'll remove the calls for ext3/ext4 when I'll submit the second version of the patch.

Thanks for your feedback,

Marco

maintainer
#4 lore raw
Quoted 2 lines · Marco Stornelli

Great. I'll remove the calls for ext3/ext4 when I'll submit the second version of the patch.

FYI, I have the following patch my ext4 tree, so I could do more intensive testing. I'll let folks know if anything goes horribly wrong, but I'm pretty sure this should be safe.

- Ted

Quoted 1 line

From 81f74e743344217487792e4190f9478963d1bb21 Mon Sep 17 00:00:00 2001

From: Theodore Ts'o <[email protected]>
Date: Fri, 17 Aug 2012 19:00:34 -0400
Subject: [PATCH] ext4: drop lock_super()/unlock_super()

We don't need lock_super()/unlock_super() any more, since the places where it is used, we are protected by the s_umount r/w semaphore.

Signed-off-by "Theodore Ts'o" <[email protected]>
Cc Marco Stornelli <[email protected]>
---
 fs/ext4/super.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)
fs/ext4/super.c +1 -10
index 3ab798d..bae4124 100644
@@ -861,7 +861,6 @@ static void ext4_put_super(struct super_block *sb)
 	flush_workqueue(sbi->dio_unwritten_wq);
 	destroy_workqueue(sbi->dio_unwritten_wq);
 
-	lock_super(sb);
 	if (sbi->s_journal) {
 		err = jbd2_journal_destroy(sbi->s_journal);
 		sbi->s_journal = NULL;

@@ -928,7 +927,6 @@ static void ext4_put_super(struct super_block *sb)
 	 * Now that we are completely done shutting down the
 	 * superblock, we need to actually destroy the kobject.
 	 */
-	unlock_super(sb);
 	kobject_put(&sbi->s_kobj);
 	wait_for_completion(&sbi->s_kobj_unregister);
 	if (sbi->s_chksum_driver)

@@ -4535,11 +4533,9 @@ static int ext4_unfreeze(struct super_block *sb)
 	if (sb->s_flags & MS_RDONLY)
 		return 0;
 
-	lock_super(sb);
 	/* Reset the needs_recovery flag before the fs is unlocked. */
 	EXT4_SET_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
 	ext4_commit_super(sb, 1);
-	unlock_super(sb);
 	return 0;
 }
 

@@ -4575,7 +4571,6 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 	char *orig_data = kstrdup(data, GFP_KERNEL);
 
 	/* Store the original options */
-	lock_super(sb);
 	old_sb_flags = sb->s_flags;
 	old_opts.s_mount_opt = sbi->s_mount_opt;
 	old_opts.s_mount_opt2 = sbi->s_mount_opt2;

@@ -4717,7 +4712,6 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 	if (sbi->s_journal == NULL)
 		ext4_commit_super(sb, 1);
 
-	unlock_super(sb);
 #ifdef CONFIG_QUOTA
 	/* Release old quota file names */
 	for (i = 0; i < MAXQUOTAS; i++)

@@ -4730,10 +4724,8 @@ static int ext4_remount(struct super_block *sb, int *flags, char *data)
 		else if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
 					EXT4_FEATURE_RO_COMPAT_QUOTA)) {
 			err = ext4_enable_quotas(sb);
-			if (err) {
-				lock_super(sb);
+			if (err)
 				goto restore_opts;
-			}
 		}
 	}
 #endif

@@ -4760,7 +4752,6 @@ restore_opts:
 		sbi->s_qf_names[i] = old_opts.s_qf_names[i];
 	}
 #endif
-	unlock_super(sb);
 	kfree(orig_data);
 	return err;
 }
Signature · 1 line
1.7.12.rc0.22.gcdd159b
author
#5 lore raw

Il 18/08/2012 01:08, Theodore Ts'o ha scritto:

Quoted 7 lines · Marco Stornelli
Quoted 2 lines

Great. I'll remove the calls for ext3/ext4 when I'll submit the second version of the patch.

FYI, I have the following patch my ext4 tree, so I could do more intensive testing. I'll let folks know if anything goes horribly wrong, but I'm pretty sure this should be safe.

Yeah, I'm agree. It should be safe.

Quoted 8 lines

- Ted

Quoted 1 line

From 81f74e743344217487792e4190f9478963d1bb21 Mon Sep 17 00:00:00 2001

From: Theodore Ts'o <[email protected]>
Date: Fri, 17 Aug 2012 19:00:34 -0400
Subject: [PATCH] ext4: drop lock_super()/unlock_super()

Quoted 6 lines

We don't need lock_super()/unlock_super() any more, since the places where it is used, we are protected by the s_umount r/w semaphore.

Signed-off-by "Theodore Ts'o" <[email protected]>
Cc Marco Stornelli <[email protected]>

---

I think the patches will go via Al's tree so if Al is agree I'll submit the patches without modify ext4 at this point. Al, let me know if we want to proceed in a different way.

Marco