Hi,

Andreas Dilger wrote:

+static int ioctl_freeze(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If filesystem doesn't support freeze feature, return. */
+ if (sb->s_op->write_super_lockfs == NULL)
+ return -EINVAL;


Should this be EINVAL, or EOPNOTSUPP? Usually EINVAL means there is
something wrong with the passed ioctl parameters (e.g. bad value),
while EOPNOTSUPP is "operation not supported" and makes more sense.


Sounds good.
I will send new patch-set which is rebased to 2.6.26-rc3 and includes
this fix, in this weekend.

Cheers, Takashi

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

I have modified to set the suitable error number (EOPNOTSUPP)
in case the filesystem doesn't support the freeze feature.
This was pointed out by Andreas Dilger.

The ioctls for the generic freeze feature are below.
o Freeze the filesystem
int ioctl(int fd, int FIFREEZE, arg)
fd: The file descriptor of the mountpoint
FIFREEZE: request code for the freeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

o Unfreeze the filesystem
int ioctl(int fd, int FITHAW, arg)
fd: The file descriptor of the mountpoint
FITHAW: request code for unfreeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

Signed-off-by: Takashi Sato <t-sato@yk.jp.nec.com>
Signed-off-by: Masayuki Hamaguchi <m-hamaguchi@ys.jp.nec.com>
---
fs/buffer.c | 30 ++++++++++++++++++++++++
fs/ioctl.c | 53 ++++++++++++++++++++++++++++++++++++++++++++
fs/super.c | 32 +++++++++++++++++++++++++-
include/linux/buffer_head.h | 2 -
include/linux/fs.h | 14 +++++++++++
5 files changed, 128 insertions(+), 3 deletions(-)

diff -uprN -X /home/sho/pub/MC/freeze-set/dontdiff linux-2.6.26-rc3.org/fs/buffer.c linux-2.6.26-rc3-freeze/fs/buffer.c
--- linux-2.6.26-rc3.org/fs/buffer.c 2008-05-19 06:36:41.000000000 +0900
+++ linux-2.6.26-rc3-freeze/fs/buffer.c 2008-05-19 21:03:29.000000000 +0900
@@ -201,6 +201,21 @@ struct super_block *freeze_bdev(struct b
{
struct super_block *sb;

+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return ERR_PTR(-EBUSY);
+
+ sb = get_super_without_lock(bdev);
+
+ /* If super_block has been already frozen, return. */
+ if (sb && sb->s_frozen != SB_UNFROZEN) {
+ put_super(sb);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return sb;
+ }
+
+ if (sb)
+ put_super(sb);
+
down(&bdev->bd_mount_sem);
sb = get_super(bdev);
if (sb && !(sb->s_flags & MS_RDONLY)) {
@@ -219,6 +234,8 @@ struct super_block *freeze_bdev(struct b
}

sync_blockdev(bdev);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+
return sb; /* thaw_bdev releases s->s_umount and bd_mount_sem */
}
EXPORT_SYMBOL(freeze_bdev);
@@ -230,8 +247,17 @@ EXPORT_SYMBOL(freeze_bdev);
*
* Unlocks the filesystem and marks it writeable again after freeze_bdev().
*/
-void thaw_bdev(struct block_device *bdev, struct super_block *sb)
+int thaw_bdev(struct block_device *bdev, struct super_block *sb)
{
+
+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return -EBUSY;
+
+ if (sb && sb->s_frozen == SB_UNFROZEN) {
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
+ }
+
if (sb) {
BUG_ON(sb->s_bdev != bdev);

@@ -244,6 +270,8 @@ void thaw_bdev(struct block_device *bdev
}

up(&bdev->bd_mount_sem);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
}
EXPORT_SYMBOL(thaw_bdev);

diff -uprN -X /home/sho/pub/MC/freeze-set/dontdiff linux-2.6.26-rc3.org/fs/ioctl.c linux-2.6.26-rc3-freeze/fs/ioctl.c
--- linux-2.6.26-rc3.org/fs/ioctl.c 2008-05-19 06:36:41.000000000 +0900
+++ linux-2.6.26-rc3-freeze/fs/ioctl.c 2008-05-21 11:43:39.000000000 +0900
@@ -13,6 +13,7 @@
#include <linux/security.h>
#include <linux/module.h>
#include <linux/uaccess.h>
+#include <linux/buffer_head.h>

#include <asm/ioctls.h>

@@ -141,6 +142,49 @@ static int ioctl_fioasync(unsigned int f
}

/*
+ * ioctl_freeze - Freeze the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call freeze_bdev() to freeze the filesystem.
+ */
+static int ioctl_freeze(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If filesystem doesn't support freeze feature, return. */
+ if (sb->s_op->write_super_lockfs == NULL)
+ return -EOPNOTSUPP;
+
+ /* Freeze */
+ sb = freeze_bdev(sb->s_bdev);
+ if (IS_ERR(sb))
+ return PTR_ERR(sb);
+ return 0;
+}
+
+/*
+ * ioctl_thaw - Thaw the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call thaw_bdev() to thaw the filesystem.
+ */
+static int ioctl_thaw(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* Thaw */
+ return thaw_bdev(sb->s_bdev, sb);
+}
+
+/*
* When you add any new common ioctls to the switches above and below
* please update compat_sys_ioctl() too.
*
@@ -181,6 +225,15 @@ int do_vfs_ioctl(struct file *filp, unsi
} else
error = -ENOTTY;
break;
+
+ case FIFREEZE:
+ error = ioctl_freeze(filp);
+ break;
+
+ case FITHAW:
+ error = ioctl_thaw(filp);
+ break;
+
default:
if (S_ISREG(filp->f_path.dentry->d_inode->i_mode))
error = file_ioctl(filp, cmd, arg);
diff -uprN -X /home/sho/pub/MC/freeze-set/dontdiff linux-2.6.26-rc3.org/fs/super.c linux-2.6.26-rc3-freeze/fs/super.c
--- linux-2.6.26-rc3.org/fs/super.c 2008-05-19 06:36:41.000000000 +0900
+++ linux-2.6.26-rc3-freeze/fs/super.c 2008-05-19 21:03:29.000000000 +0900
@@ -156,7 +156,7 @@ int __put_super_and_need_restart(struct
* Drops a temporary reference, frees superblock if there's no
* references left.
*/
-static void put_super(struct super_block *sb)
+void put_super(struct super_block *sb)
{
spin_lock(&sb_lock);
__put_super(sb);
@@ -509,6 +509,36 @@ rescan:

EXPORT_SYMBOL(get_super);

+/*
+ * get_super_without_lock - Get super_block from block_device without lock.
+ * @bdev: block device struct
+ *
+ * Scan the superblock list and finds the superblock of the file system
+ * mounted on the block device given. This doesn't lock anyone.
+ * %NULL is returned if no match is found.
+ */
+struct super_block *get_super_without_lock(struct block_device *bdev)
+{
+ struct super_block *sb;
+
+ if (!bdev)
+ return NULL;
+
+ spin_lock(&sb_lock);
+ list_for_each_entry(sb, &super_blocks, s_list) {
+ if (sb->s_bdev == bdev) {
+ if (sb->s_root) {
+ sb->s_count++;
+ spin_unlock(&sb_lock);
+ return sb;
+ }
+ }
+ }
+ spin_unlock(&sb_lock);
+ return NULL;
+}
+EXPORT_SYMBOL(get_super_without_lock);
+
struct super_block * user_get_super(dev_t dev)
{
struct super_block *sb;
diff -uprN -X /home/sho/pub/MC/freeze-set/dontdiff linux-2.6.26-rc3.org/include/linux/buffer_head.h linux-2.6.26-rc3-fre
eze/include/linux/buffer_head.h
--- linux-2.6.26-rc3.org/include/linux/buffer_head.h 2008-05-19 06:36:41.000000000 +0900
+++ linux-2.6.26-rc3-freeze/include/linux/buffer_head.h 2008-05-19 21:03:29.000000000 +0900
@@ -171,7 +171,7 @@ void __wait_on_buffer(struct buffer_head
wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
int fsync_bdev(struct block_device *);
struct super_block *freeze_bdev(struct block_device *);
-void thaw_bdev(struct block_device *, struct super_block *);
+int thaw_bdev(struct block_device *, struct super_block *);
int fsync_super(struct super_block *);
int fsync_no_super(struct block_device *);
struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
diff -uprN -X /home/sho/pub/MC/freeze-set/dontdiff linux-2.6.26-rc3.org/include/linux/fs.h linux-2.6.26-rc3-freeze/inclu
de/linux/fs.h
--- linux-2.6.26-rc3.org/include/linux/fs.h 2008-05-19 06:36:41.000000000 +0900
+++ linux-2.6.26-rc3-freeze/include/linux/fs.h 2008-05-19 21:03:29.000000000 +0900
@@ -223,6 +223,8 @@ extern int dir_notify_enable;
#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
#define FIBMAP _IO(0x00,1) /* bmap access */
#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
+#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
+#define FITHAW _IOWR('X', 120, int) /* Thaw */

#define FS_IOC_GETFLAGS _IOR('f', 1, long)
#define FS_IOC_SETFLAGS _IOW('f', 2, long)
@@ -494,6 +496,13 @@ int pagecache_write_end(struct file *, s
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata);

+/*
+ * Bits in block_device.bd_state.
+ */
+enum bd_state {
+ BD_FREEZE_OP /* In freeze operation */
+};
+
struct backing_dev_info;
struct address_space {
struct inode *host; /* owner: inode, block_device */
@@ -547,6 +556,9 @@ struct block_device {
* care to not mess up bd_private for that case.
*/
unsigned long bd_private;
+
+ /* State of the block device. (Used by freeze feature) */
+ unsigned long bd_state;
};

/*
@@ -1964,7 +1976,9 @@ extern int do_vfs_ioctl(struct file *fil
extern void get_filesystem(struct file_system_type *fs);
extern void put_filesystem(struct file_system_type *fs);
extern struct file_system_type *get_fs_type(const char *name);
+extern void put_super(struct super_block *sb);
extern struct super_block *get_super(struct block_device *);
+extern struct super_block *get_super_without_lock(struct block_device *);
extern struct super_block *user_get_super(dev_t);
extern void drop_super(struct super_block *sb);


--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

> + if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
> + return ERR_PTR(-EBUSY);
> +
> + sb = get_super_without_lock(bdev);
> +
> + /* If super_block has been already frozen, return. */
> + if (sb && sb->s_frozen != SB_UNFROZEN) {
> + put_super(sb);
> + clear_bit(BD_FREEZE_OP, &bdev->bd_state);
> + return sb;
> + }

The BD_FREEZE_OP flag in the block_device already prevents multiple
freezes for a singe block device, so there is no need for this
additional check and the get_super_without_lock helper.

> down(&bdev->bd_mount_sem);

And with that flag bd_mount_sem is also obsolete for preventing the
multiple freeze operations. We still need investigate what
synchronization we need vs unmount which also takes bd_mount_sem without
every having document what it exactly protects.

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Hi,


+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return ERR_PTR(-EBUSY);
+
+ sb = get_super_without_lock(bdev);
+
+ /* If super_block has been already frozen, return. */
+ if (sb && sb->s_frozen != SB_UNFROZEN) {
+ put_super(sb);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return sb;
+ }


The BD_FREEZE_OP flag in the block_device already prevents multiple
freezes for a singe block device, so there is no need for this
additional check and the get_super_without_lock helper.


I think the above check is needed because BD_FREEZE_OP is used to
guarantee that the following two operations are done atomically.
1. Check if the filesystem has already been frozen.
(sb->s_frozen != SB_UNFROZEN)
2. Do the filesystem specific freeze operation
(write_super_lockfs)

The freeze operation can be done again for the frozen filesystem
unless the filesystem is checked.
And if write_super_lockfs doesn't consider that the filesystem
has already been frozen, a problem might occur.


down(&bdev->bd_mount_sem);


And with that flag bd_mount_sem is also obsolete for preventing the
multiple freeze operations.


bd_mount_sem is needed because it is used to guarantee that the freeze
operation never run while mounting (get_sb_bdev()).


We still need investigate what
synchronization we need vs unmount which also takes bd_mount_sem without
every having document what it exactly protects.


And s_umount (semaphore) is acquired in get_user() (called by freeze_bdev())
to guarantee that unmount never run while the filesystem is frozen.

Cheers, Takashi

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

I have modified to set the suitable error number (EOPNOTSUPP)
in case the filesystem doesn't support the freeze feature.
This was pointed out by Andreas Dilger.

The ioctls for the generic freeze feature are below.
o Freeze the filesystem
int ioctl(int fd, int FIFREEZE, arg)
fd: The file descriptor of the mountpoint
FIFREEZE: request code for the freeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

o Unfreeze the filesystem
int ioctl(int fd, int FITHAW, arg)
fd: The file descriptor of the mountpoint
FITHAW: request code for unfreeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

Signed-off-by: Takashi Sato <t-sato@yk.jp.nec.com>
Signed-off-by: Masayuki Hamaguchi <m-hamaguchi@ys.jp.nec.com>
---
fs/buffer.c | 30 ++++++++++++++++++++++++
fs/ioctl.c | 53 ++++++++++++++++++++++++++++++++++++++++++++
fs/super.c | 32 +++++++++++++++++++++++++-
include/linux/buffer_head.h | 2 -
include/linux/fs.h | 14 +++++++++++
5 files changed, 128 insertions(+), 3 deletions(-)

diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/fs/buffer.c linux-2.6.26-rc7-freeze/fs/bu
ffer.c
--- linux-2.6.26-rc7.org/fs/buffer.c 2008-06-21 08:19:44.000000000 +0900
+++ linux-2.6.26-rc7-freeze/fs/buffer.c 2008-06-23 11:54:48.000000000 +0900
@@ -201,6 +201,21 @@ struct super_block *freeze_bdev(struct b
{
struct super_block *sb;

+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return ERR_PTR(-EBUSY);
+
+ sb = get_super_without_lock(bdev);
+
+ /* If super_block has been already frozen, return. */
+ if (sb && sb->s_frozen != SB_UNFROZEN) {
+ put_super(sb);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return sb;
+ }
+
+ if (sb)
+ put_super(sb);
+
down(&bdev->bd_mount_sem);
sb = get_super(bdev);
if (sb && !(sb->s_flags & MS_RDONLY)) {
@@ -219,6 +234,8 @@ struct super_block *freeze_bdev(struct b
}

sync_blockdev(bdev);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+
return sb; /* thaw_bdev releases s->s_umount and bd_mount_sem */
}
EXPORT_SYMBOL(freeze_bdev);
@@ -230,8 +247,17 @@ EXPORT_SYMBOL(freeze_bdev);
*
* Unlocks the filesystem and marks it writeable again after freeze_bdev().
*/
-void thaw_bdev(struct block_device *bdev, struct super_block *sb)
+int thaw_bdev(struct block_device *bdev, struct super_block *sb)
{
+
+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return -EBUSY;
+
+ if (sb && sb->s_frozen == SB_UNFROZEN) {
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
+ }
+
if (sb) {
BUG_ON(sb->s_bdev != bdev);

@@ -244,6 +270,8 @@ void thaw_bdev(struct block_device *bdev
}

up(&bdev->bd_mount_sem);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
}
EXPORT_SYMBOL(thaw_bdev);

diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/fs/ioctl.c linux-2.6.26-rc7-freeze/fs/ioc
tl.c
--- linux-2.6.26-rc7.org/fs/ioctl.c 2008-06-21 08:19:44.000000000 +0900
+++ linux-2.6.26-rc7-freeze/fs/ioctl.c 2008-06-23 11:54:48.000000000 +0900
@@ -13,6 +13,7 @@
#include <linux/security.h>
#include <linux/module.h>
#include <linux/uaccess.h>
+#include <linux/buffer_head.h>

#include <asm/ioctls.h>

@@ -141,6 +142,49 @@ static int ioctl_fioasync(unsigned int f
}

/*
+ * ioctl_freeze - Freeze the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call freeze_bdev() to freeze the filesystem.
+ */
+static int ioctl_freeze(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If filesystem doesn't support freeze feature, return. */
+ if (sb->s_op->write_super_lockfs == NULL)
+ return -EOPNOTSUPP;
+
+ /* Freeze */
+ sb = freeze_bdev(sb->s_bdev);
+ if (IS_ERR(sb))
+ return PTR_ERR(sb);
+ return 0;
+}
+
+/*
+ * ioctl_thaw - Thaw the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call thaw_bdev() to thaw the filesystem.
+ */
+static int ioctl_thaw(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* Thaw */
+ return thaw_bdev(sb->s_bdev, sb);
+}
+
+/*
* When you add any new common ioctls to the switches above and below
* please update compat_sys_ioctl() too.
*
@@ -181,6 +225,15 @@ int do_vfs_ioctl(struct file *filp, unsi
} else
error = -ENOTTY;
break;
+
+ case FIFREEZE:
+ error = ioctl_freeze(filp);
+ break;
+
+ case FITHAW:
+ error = ioctl_thaw(filp);
+ break;
+
default:
if (S_ISREG(filp->f_path.dentry->d_inode->i_mode))
error = file_ioctl(filp, cmd, arg);
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/fs/super.c linux-2.6.26-rc7-freeze/fs/sup
er.c
--- linux-2.6.26-rc7.org/fs/super.c 2008-06-21 08:19:44.000000000 +0900
+++ linux-2.6.26-rc7-freeze/fs/super.c 2008-06-23 11:54:48.000000000 +0900
@@ -156,7 +156,7 @@ int __put_super_and_need_restart(struct
* Drops a temporary reference, frees superblock if there's no
* references left.
*/
-static void put_super(struct super_block *sb)
+void put_super(struct super_block *sb)
{
spin_lock(&sb_lock);
__put_super(sb);
@@ -509,6 +509,36 @@ rescan:

EXPORT_SYMBOL(get_super);

+/*
+ * get_super_without_lock - Get super_block from block_device without lock.
+ * @bdev: block device struct
+ *
+ * Scan the superblock list and finds the superblock of the file system
+ * mounted on the block device given. This doesn't lock anyone.
+ * %NULL is returned if no match is found.
+ */
+struct super_block *get_super_without_lock(struct block_device *bdev)
+{
+ struct super_block *sb;
+
+ if (!bdev)
+ return NULL;
+
+ spin_lock(&sb_lock);
+ list_for_each_entry(sb, &super_blocks, s_list) {
+ if (sb->s_bdev == bdev) {
+ if (sb->s_root) {
+ sb->s_count++;
+ spin_unlock(&sb_lock);
+ return sb;
+ }
+ }
+ }
+ spin_unlock(&sb_lock);
+ return NULL;
+}
+EXPORT_SYMBOL(get_super_without_lock);
+
struct super_block * user_get_super(dev_t dev)
{
struct super_block *sb;
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/include/linux/buffer_head.h linux-2.6.26-
rc7-freeze/include/linux/buffer_head.h
--- linux-2.6.26-rc7.org/include/linux/buffer_head.h 2008-06-21 08:19:44.000000000 +0900
+++ linux-2.6.26-rc7-freeze/include/linux/buffer_head.h 2008-06-23 11:54:48.000000000 +0900
@@ -171,7 +171,7 @@ void __wait_on_buffer(struct buffer_head
wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
int fsync_bdev(struct block_device *);
struct super_block *freeze_bdev(struct block_device *);
-void thaw_bdev(struct block_device *, struct super_block *);
+int thaw_bdev(struct block_device *, struct super_block *);
int fsync_super(struct super_block *);
int fsync_no_super(struct block_device *);
struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/include/linux/fs.h linux-2.6.26-rc7-freez
e/include/linux/fs.h
--- linux-2.6.26-rc7.org/include/linux/fs.h 2008-06-21 08:19:44.000000000 +0900
+++ linux-2.6.26-rc7-freeze/include/linux/fs.h 2008-06-23 11:54:48.000000000 +0900
@@ -223,6 +223,8 @@ extern int dir_notify_enable;
#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
#define FIBMAP _IO(0x00,1) /* bmap access */
#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
+#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
+#define FITHAW _IOWR('X', 120, int) /* Thaw */

#define FS_IOC_GETFLAGS _IOR('f', 1, long)
#define FS_IOC_SETFLAGS _IOW('f', 2, long)
@@ -494,6 +496,13 @@ int pagecache_write_end(struct file *, s
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata);

+/*
+ * Bits in block_device.bd_state.
+ */
+enum bd_state {
+ BD_FREEZE_OP /* In freeze operation */
+};
+
struct backing_dev_info;
struct address_space {
struct inode *host; /* owner: inode, block_device */
@@ -547,6 +556,9 @@ struct block_device {
* care to not mess up bd_private for that case.
*/
unsigned long bd_private;
+
+ /* State of the block device. (Used by freeze feature) */
+ unsigned long bd_state;
};

/*
@@ -1964,7 +1976,9 @@ extern int do_vfs_ioctl(struct file *fil
extern void get_filesystem(struct file_system_type *fs);
extern void put_filesystem(struct file_system_type *fs);
extern struct file_system_type *get_fs_type(const char *name);
+extern void put_super(struct super_block *sb);
extern struct super_block *get_super(struct block_device *);
+extern struct super_block *get_super_without_lock(struct block_device *);
extern struct super_block *user_get_super(dev_t);
extern void drop_super(struct super_block *sb);


--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

On Tue, 24 Jun 2008 15:59:50 +0900
Takashi Sato <t-sato@yk.jp.nec.com> wrote:

> I have modified to set the suitable error number (EOPNOTSUPP)
> in case the filesystem doesn't support the freeze feature.
> This was pointed out by Andreas Dilger.
>
> The ioctls for the generic freeze feature are below.
> o Freeze the filesystem
> int ioctl(int fd, int FIFREEZE, arg)
> fd: The file descriptor of the mountpoint
> FIFREEZE: request code for the freeze
> arg: Ignored
> Return value: 0 if the operation succeeds. Otherwise, -1
>
> o Unfreeze the filesystem
> int ioctl(int fd, int FITHAW, arg)
> fd: The file descriptor of the mountpoint
> FITHAW: request code for unfreeze
> arg: Ignored
> Return value: 0 if the operation succeeds. Otherwise, -1
>
> ...
>
> +/*
> + * get_super_without_lock - Get super_block from block_device without lock.
> + * @bdev: block device struct
> + *
> + * Scan the superblock list and finds the superblock of the file system
> + * mounted on the block device given. This doesn't lock anyone.
> + * %NULL is returned if no match is found.
> + */

This is not a terribly good comment.

Which lock are we not taking? I _assume_ that it's referring to
s_umount? If so, the text should describe that.

It should also go to some lengths explaining why this dangerous-looking
and rather nasty-looking function exists.

Look at it this way: there is no way in which the reviewer of this
patch (ie: me) can work out why this function exists. Hence there will
be no way in which future readers of this code will be able to work out
why this function exists either. This is bad. These things should be
described in code comments and in the changelog (whichever is most
appropriate).

> +struct super_block *get_super_without_lock(struct block_device *bdev)
> +{
> + struct super_block *sb;
> +
> + if (!bdev)
> + return NULL;
> +
> + spin_lock(&sb_lock);
> + list_for_each_entry(sb, &super_blocks, s_list) {
> + if (sb->s_bdev == bdev) {
> + if (sb->s_root) {
> + sb->s_count++;
> + spin_unlock(&sb_lock);
> + return sb;
> + }
> + }
> + }
> + spin_unlock(&sb_lock);
> + return NULL;
> +}
> +EXPORT_SYMBOL(get_super_without_lock);

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Hi,


+/*
+ * get_super_without_lock - Get super_block from block_device without lock.
+ * @bdev: block device struct
+ *
+ * Scan the superblock list and finds the superblock of the file system
+ * mounted on the block device given. This doesn't lock anyone.
+ * %NULL is returned if no match is found.
+ */


This is not a terribly good comment.

Which lock are we not taking? I _assume_ that it's referring to
s_umount? If so, the text should describe that.

It should also go to some lengths explaining why this dangerous-looking
and rather nasty-looking function exists.

Look at it this way: there is no way in which the reviewer of this
patch (ie: me) can work out why this function exists. Hence there will
be no way in which future readers of this code will be able to work out
why this function exists either. This is bad. These things should be
described in code comments and in the changelog (whichever is most
appropriate).


Thank you for your comment. I will write comments appropriately.

I was wrong. I thought we didn't need to lock s_umount because
this ioctl required to open a regular file or a directory and we cannot
unmount a target filesystem.
So I created get_super_without_lock() used in freeze_bdev().

But, I have found that the ioctl (DM_DEV_SUSPEND_CMD in
drivers/md/dm-ioctl.c) requires to open a logical volume
(not a file or a directory) and calls freeze_bdev(), so we can unmount
a filesystem.
So I will replace get_super_without_lock with get_super to get s_umount.

Cheers, Takashi

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

The ioctls for the generic freeze feature are below.
o Freeze the filesystem
int ioctl(int fd, int FIFREEZE, arg)
fd: The file descriptor of the mountpoint
FIFREEZE: request code for the freeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

o Unfreeze the filesystem
int ioctl(int fd, int FITHAW, arg)
fd: The file descriptor of the mountpoint
FITHAW: request code for unfreeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

Signed-off-by: Takashi Sato <t-sato@yk.jp.nec.com>
Signed-off-by: Masayuki Hamaguchi <m-hamaguchi@ys.jp.nec.com>
---
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/fs/buffer.c linux-2.6.26-rc7-freeze/fs/bu
ffer.c
--- linux-2.6.26-rc7.org/fs/buffer.c 2008-06-25 12:08:10.000000000 +0900
+++ linux-2.6.26-rc7-freeze/fs/buffer.c 2008-06-27 09:18:19.000000000 +0900
@@ -201,6 +201,21 @@ struct super_block *freeze_bdev(struct b
{
struct super_block *sb;

+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return ERR_PTR(-EBUSY);
+
+ sb = get_super(bdev);
+
+ /* If super_block has been already frozen, return. */
+ if (sb && sb->s_frozen != SB_UNFROZEN) {
+ drop_super(sb);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return sb;
+ }
+
+ if (sb)
+ drop_super(sb);
+
down(&bdev->bd_mount_sem);
sb = get_super(bdev);
if (sb && !(sb->s_flags & MS_RDONLY)) {
@@ -219,6 +234,8 @@ struct super_block *freeze_bdev(struct b
}

sync_blockdev(bdev);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+
return sb; /* thaw_bdev releases s->s_umount and bd_mount_sem */
}
EXPORT_SYMBOL(freeze_bdev);
@@ -230,8 +247,17 @@ EXPORT_SYMBOL(freeze_bdev);
*
* Unlocks the filesystem and marks it writeable again after freeze_bdev().
*/
-void thaw_bdev(struct block_device *bdev, struct super_block *sb)
+int thaw_bdev(struct block_device *bdev, struct super_block *sb)
{
+
+ if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
+ return -EBUSY;
+
+ if (sb && sb->s_frozen == SB_UNFROZEN) {
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
+ }
+
if (sb) {
BUG_ON(sb->s_bdev != bdev);

@@ -244,6 +270,8 @@ void thaw_bdev(struct block_device *bdev
}

up(&bdev->bd_mount_sem);
+ clear_bit(BD_FREEZE_OP, &bdev->bd_state);
+ return 0;
}
EXPORT_SYMBOL(thaw_bdev);

diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/fs/ioctl.c linux-2.6.26-rc7-freeze/fs/ioc
tl.c
--- linux-2.6.26-rc7.org/fs/ioctl.c 2008-06-25 12:08:14.000000000 +0900
+++ linux-2.6.26-rc7-freeze/fs/ioctl.c 2008-06-25 12:00:41.000000000 +0900
@@ -13,6 +13,7 @@
#include <linux/security.h>
#include <linux/module.h>
#include <linux/uaccess.h>
+#include <linux/buffer_head.h>

#include <asm/ioctls.h>

@@ -141,6 +142,49 @@ static int ioctl_fioasync(unsigned int f
}

/*
+ * ioctl_freeze - Freeze the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call freeze_bdev() to freeze the filesystem.
+ */
+static int ioctl_freeze(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If filesystem doesn't support freeze feature, return. */
+ if (sb->s_op->write_super_lockfs == NULL)
+ return -EOPNOTSUPP;
+
+ /* Freeze */
+ sb = freeze_bdev(sb->s_bdev);
+ if (IS_ERR(sb))
+ return PTR_ERR(sb);
+ return 0;
+}
+
+/*
+ * ioctl_thaw - Thaw the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call thaw_bdev() to thaw the filesystem.
+ */
+static int ioctl_thaw(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* Thaw */
+ return thaw_bdev(sb->s_bdev, sb);
+}
+
+/*
* When you add any new common ioctls to the switches above and below
* please update compat_sys_ioctl() too.
*
@@ -181,6 +225,15 @@ int do_vfs_ioctl(struct file *filp, unsi
} else
error = -ENOTTY;
break;
+
+ case FIFREEZE:
+ error = ioctl_freeze(filp);
+ break;
+
+ case FITHAW:
+ error = ioctl_thaw(filp);
+ break;
+
default:
if (S_ISREG(filp->f_path.dentry->d_inode->i_mode))
error = file_ioctl(filp, cmd, arg);
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/include/linux/buffer_head.h linux-2.6.26-
rc7-freeze/include/linux/buffer_head.h
--- linux-2.6.26-rc7.org/include/linux/buffer_head.h 2008-06-25 12:08:19.000000000 +0900
+++ linux-2.6.26-rc7-freeze/include/linux/buffer_head.h 2008-06-25 12:00:45.000000000 +0900
@@ -171,7 +171,7 @@ void __wait_on_buffer(struct buffer_head
wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
int fsync_bdev(struct block_device *);
struct super_block *freeze_bdev(struct block_device *);
-void thaw_bdev(struct block_device *, struct super_block *);
+int thaw_bdev(struct block_device *, struct super_block *);
int fsync_super(struct super_block *);
int fsync_no_super(struct block_device *);
struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
diff -uprN -X linux-2.6.26-rc7.org/Documentation/dontdiff linux-2.6.26-rc7.org/include/linux/fs.h linux-2.6.26-rc7-freez
e/include/linux/fs.h
--- linux-2.6.26-rc7.org/include/linux/fs.h 2008-06-25 12:08:19.000000000 +0900
+++ linux-2.6.26-rc7-freeze/include/linux/fs.h 2008-06-30 16:40:49.000000000 +0900
@@ -223,6 +223,8 @@ extern int dir_notify_enable;
#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
#define FIBMAP _IO(0x00,1) /* bmap access */
#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
+#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
+#define FITHAW _IOWR('X', 120, int) /* Thaw */

#define FS_IOC_GETFLAGS _IOR('f', 1, long)
#define FS_IOC_SETFLAGS _IOW('f', 2, long)
@@ -494,6 +496,13 @@ int pagecache_write_end(struct file *, s
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata);

+/*
+ * Bits in block_device.bd_state.
+ */
+enum bd_state {
+ BD_FREEZE_OP /* In freeze operation */
+};
+
struct backing_dev_info;
struct address_space {
struct inode *host; /* owner: inode, block_device */
@@ -547,6 +556,9 @@ struct block_device {
* care to not mess up bd_private for that case.
*/
unsigned long bd_private;
+
+ /* State of the block device. (Used by freeze feature) */
+ unsigned long bd_state;
};

/*

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

> {
> struct super_block *sb;
>
> + if (test_and_set_bit(BD_FREEZE_OP, &bdev->bd_state))
> + return ERR_PTR(-EBUSY);
> +
> + sb = get_super(bdev);
> +
> + /* If super_block has been already frozen, return. */
> + if (sb && sb->s_frozen != SB_UNFROZEN) {
> + drop_super(sb);
> + clear_bit(BD_FREEZE_OP, &bdev->bd_state);
> + return sb;
> + }
> +
> + if (sb)
> + drop_super(sb);
> +
> down(&bdev->bd_mount_sem);
> sb = get_super(bdev);
> if (sb && !(sb->s_flags & MS_RDONLY)) {
> @@ -219,6 +234,8 @@ struct super_block *freeze_bdev(struct b
> }
>
> sync_blockdev(bdev);
> + clear_bit(BD_FREEZE_OP, &bdev->bd_state);
> +

Please only clear BD_FREEZE_OP in thaw_bdev, that way you can also get
rid of the frozen check above, and the double-get_super. Also
bd_mount_sem could be removed that way by checking for BD_FREEZE_OP
in the unmount path.

> /*
> + * ioctl_freeze - Freeze the filesystem.
> + *
> + * @filp: target file
> + *
> + * Call freeze_bdev() to freeze the filesystem.
> + */

This is not a kerneldoc comment. But I think it can be simply removed
anyway, as it's a quite trivial function with static scope.

> +/*
> + * ioctl_thaw - Thaw the filesystem.
> + *
> + * @filp: target file
> + *
> + * Call thaw_bdev() to thaw the filesystem.
> + */

Same here.

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

When multiple freeze requests arrive simultaneously, only the last
unfreeze process should unfreeze the frozen filesystem actually
(as Dave Chinner, Eric Sandeen and Alasdair G Kergon commented).
So I've added the reference counter to the freeze feature.
It counts up in freeze_bdev() and counts down in thaw_bdev().
When it becomes "0", thaw_bdev() will unfreeze actually.

The ioctls for the generic freeze feature are below.
o Freeze the filesystem
int ioctl(int fd, int FIFREEZE, arg)
fd: The file descriptor of the mountpoint
FIFREEZE: request code for the freeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

o Unfreeze the filesystem
int ioctl(int fd, int FITHAW, arg)
fd: The file descriptor of the mountpoint
FITHAW: request code for unfreeze
arg: Ignored
Return value: 0 if the operation succeeds. Otherwise, -1

Signed-off-by: Takashi Sato <t-sato@yk.jp.nec.com>
Signed-off-by: Masayuki Hamaguchi <m-hamaguchi@ys.jp.nec.com>
---
fs/block_dev.c | 2 +
fs/buffer.c | 27 ++++++++++++++++++-
fs/ioctl.c | 61 ++++++++++++++++++++++++++++++++++++++++++++
include/linux/buffer_head.h | 2 -
include/linux/fs.h | 6 ++++
5 files changed, 96 insertions(+), 2 deletions(-)

diff -uprN -X linux-2.6.26.org/Documentation/dontdiff linux-2.6.26.org/fs/block_dev.c linux-2.6.26-freeze/fs/block_dev.c
--- linux-2.6.26.org/fs/block_dev.c 2008-07-14 06:51:29.000000000 +0900
+++ linux-2.6.26-freeze/fs/block_dev.c 2008-07-18 15:38:03.000000000 +0900
@@ -285,6 +285,8 @@ static void init_once(struct kmem_cache
INIT_LIST_HEAD(&bdev->bd_holder_list);
#endif
inode_init_once(&ei->vfs_inode);
+ /* Initialize semaphore for freeze. */
+ sema_init(&bdev->bd_freeze_sem, 1);
}

static inline void __bd_forget(struct inode *inode)
diff -uprN -X linux-2.6.26.org/Documentation/dontdiff linux-2.6.26.org/fs/buffer.c linux-2.6.26-freeze/fs/buffer.c
--- linux-2.6.26.org/fs/buffer.c 2008-07-14 06:51:29.000000000 +0900
+++ linux-2.6.26-freeze/fs/buffer.c 2008-07-18 15:45:17.000000000 +0900
@@ -201,6 +201,15 @@ struct super_block *freeze_bdev(struct b
{
struct super_block *sb;

+ down(&bdev->bd_freeze_sem);
+ bdev->bd_freeze_count++;
+ if (bdev->bd_freeze_count > 1) {
+ sb = get_super(bdev);
+ drop_super(sb);
+ up(&bdev->bd_freeze_sem);
+ return sb;
+ }
+
down(&bdev->bd_mount_sem);
sb = get_super(bdev);
if (sb && !(sb->s_flags & MS_RDONLY)) {
@@ -219,6 +228,8 @@ struct super_block *freeze_bdev(struct b
}

sync_blockdev(bdev);
+ up(&bdev->bd_freeze_sem);
+
return sb; /* thaw_bdev releases s->s_umount and bd_mount_sem */
}
EXPORT_SYMBOL(freeze_bdev);
@@ -230,8 +241,20 @@ EXPORT_SYMBOL(freeze_bdev);
*
* Unlocks the filesystem and marks it writeable again after freeze_bdev().
*/
-void thaw_bdev(struct block_device *bdev, struct super_block *sb)
+int thaw_bdev(struct block_device *bdev, struct super_block *sb)
{
+
+ down(&bdev->bd_freeze_sem);
+ if (!bdev->bd_freeze_count) {
+ up(&bdev->bd_freeze_sem);
+ return 0;
+ }
+ bdev->bd_freeze_count--;
+ if (bdev->bd_freeze_count > 0) {
+ up(&bdev->bd_freeze_sem);
+ return 0;
+ }
+
if (sb) {
BUG_ON(sb->s_bdev != bdev);

@@ -244,6 +267,8 @@ void thaw_bdev(struct block_device *bdev
}

up(&bdev->bd_mount_sem);
+ up(&bdev->bd_freeze_sem);
+ return 0;
}
EXPORT_SYMBOL(thaw_bdev);

diff -uprN -X linux-2.6.26.org/Documentation/dontdiff linux-2.6.26.org/fs/ioctl.c linux-2.6.26-freeze/fs/ioctl.c
--- linux-2.6.26.org/fs/ioctl.c 2008-07-14 06:51:29.000000000 +0900
+++ linux-2.6.26-freeze/fs/ioctl.c 2008-07-18 21:24:42.000000000 +0900
@@ -13,6 +13,7 @@
#include <linux/security.h>
#include <linux/module.h>
#include <linux/uaccess.h>
+#include <linux/buffer_head.h>

#include <asm/ioctls.h>

@@ -141,6 +142,57 @@ static int ioctl_fioasync(unsigned int f
}

/*
+ * ioctl_freeze - Freeze the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call freeze_bdev() to freeze the filesystem.
+ */
+static int ioctl_freeze(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If filesystem doesn't support freeze feature, return. */
+ if (sb->s_op->write_super_lockfs == NULL)
+ return -EOPNOTSUPP;
+
+ /* If a regular file or a directory isn't specified, return. */
+ if (sb->s_bdev == NULL)
+ return -EINVAL;
+
+ /* Freeze */
+ sb = freeze_bdev(sb->s_bdev);
+ if (IS_ERR(sb))
+ return PTR_ERR(sb);
+ return 0;
+}
+
+/*
+ * ioctl_thaw - Thaw the filesystem.
+ *
+ * @filp: target file
+ *
+ * Call thaw_bdev() to thaw the filesystem.
+ */
+static int ioctl_thaw(struct file *filp)
+{
+ struct super_block *sb = filp->f_path.dentry->d_inode->i_sb;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ /* If a regular file or a directory isn't specified, return EINVAL. */
+ if (sb->s_bdev == NULL)
+ return -EINVAL;
+
+ /* Thaw */
+ return thaw_bdev(sb->s_bdev, sb);
+}
+
+/*
* When you add any new common ioctls to the switches above and below
* please update compat_sys_ioctl() too.
*
@@ -181,6 +233,15 @@ int do_vfs_ioctl(struct file *filp, unsi
} else
error = -ENOTTY;
break;
+
+ case FIFREEZE:
+ error = ioctl_freeze(filp);
+ break;
+
+ case FITHAW:
+ error = ioctl_thaw(filp);
+ break;
+
default:
if (S_ISREG(filp->f_path.dentry->d_inode->i_mode))
error = file_ioctl(filp, cmd, arg);
diff -uprN -X linux-2.6.26.org/Documentation/dontdiff linux-2.6.26.org/include/linux/buffer_head.h linux-2.6.26-freeze/i
nclude/linux/buffer_head.h
--- linux-2.6.26.org/include/linux/buffer_head.h 2008-07-14 06:51:29.000000000 +0900
+++ linux-2.6.26-freeze/include/linux/buffer_head.h 2008-07-17 11:32:37.000000000 +0900
@@ -171,7 +171,7 @@ void __wait_on_buffer(struct buffer_head
wait_queue_head_t *bh_waitq_head(struct buffer_head *bh);
int fsync_bdev(struct block_device *);
struct super_block *freeze_bdev(struct block_device *);
-void thaw_bdev(struct block_device *, struct super_block *);
+int thaw_bdev(struct block_device *, struct super_block *);
int fsync_super(struct super_block *);
int fsync_no_super(struct block_device *);
struct buffer_head *__find_get_block(struct block_device *bdev, sector_t block,
diff -uprN -X linux-2.6.26.org/Documentation/dontdiff linux-2.6.26.org/include/linux/fs.h linux-2.6.26-freeze/include/li
nux/fs.h
--- linux-2.6.26.org/include/linux/fs.h 2008-07-14 06:51:29.000000000 +0900
+++ linux-2.6.26-freeze/include/linux/fs.h 2008-07-18 15:34:46.000000000 +0900
@@ -224,6 +224,8 @@ extern int dir_notify_enable;
#define BMAP_IOCTL 1 /* obsolete - kept for compatibility */
#define FIBMAP _IO(0x00,1) /* bmap access */
#define FIGETBSZ _IO(0x00,2) /* get the block size used for bmap */
+#define FIFREEZE _IOWR('X', 119, int) /* Freeze */
+#define FITHAW _IOWR('X', 120, int) /* Thaw */

#define FS_IOC_GETFLAGS _IOR('f', 1, long)
#define FS_IOC_SETFLAGS _IOW('f', 2, long)
@@ -548,6 +550,10 @@ struct block_device {
* care to not mess up bd_private for that case.
*/
unsigned long bd_private;
+ /* The counter of freeze processes */
+ int bd_freeze_count;
+ /* Semaphore for freeze */
+ struct semaphore bd_freeze_sem;
};

/*

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel