Merge tag 'vfs-6.8.super' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

Pull vfs super updates from Christian Brauner:
 "This contains the super work for this cycle including the long-awaited
  series by Jan to make it possible to prevent writing to mounted block
  devices:

   - Writing to mounted devices is dangerous and can lead to filesystem
     corruption as well as crashes. Furthermore syzbot comes with more
     and more involved examples how to corrupt block device under a
     mounted filesystem leading to kernel crashes and reports we can do
     nothing about. Add tracking of writers to each block device and a
     kernel cmdline argument which controls whether other writeable
     opens to block devices open with BLK_OPEN_RESTRICT_WRITES flag are
     allowed.

     Note that this effectively only prevents modification of the
     particular block device's page cache by other writers. The actual
     device content can still be modified by other means - e.g. by
     issuing direct scsi commands, by doing writes through devices lower
     in the storage stack (e.g. in case loop devices, DM, or MD are
     involved) etc. But blocking direct modifications of the block
     device page cache is enough to give filesystems a chance to perform
     data validation when loading data from the underlying storage and
     thus prevent kernel crashes.

     Syzbot can use this cmdline argument option to avoid uninteresting
     crashes. Also users whose userspace setup does not need writing to
     mounted block devices can set this option for hardening. We expect
     that this will be interesting to quite a few workloads.

     Btrfs is currently opted out of this because they still haven't
     merged patches we require for this to work from three kernel
     releases ago.

   - Reimplement block device freezing and thawing as holder operations
     on the block device.

     This allows us to extend block device freezing to all devices
     associated with a superblock and not just the main device. It also
     allows us to remove get_active_super() and thus another function
     that scans the global list of superblocks.

     Freezing via additional block devices only works if the filesystem
     chooses to use @fs_holder_ops for these additional devices as well.
     That currently only includes ext4 and xfs.

     Earlier releases switched get_tree_bdev() and mount_bdev() to use
     @fs_holder_ops. The remaining nilfs2 open-coded version of
     mount_bdev() has been converted to rely on @fs_holder_ops as well.
     So block device freezing for the main block device will continue to
     work as before.

     There should be no regressions in functionality. The only special
     case is btrfs where block device freezing for the main block device
     never worked because sb->s_bdev isn't set. Block device freezing
     for btrfs can be fixed once they can switch to @fs_holder_ops but
     that can happen whenever they're ready"

* tag 'vfs-6.8.super' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: (27 commits)
  block: Fix a memory leak in bdev_open_by_dev()
  super: don't bother with WARN_ON_ONCE()
  super: massage wait event mechanism
  ext4: Block writes to journal device
  xfs: Block writes to log device
  fs: Block writes to mounted block devices
  btrfs: Do not restrict writes to btrfs devices
  block: Add config option to not allow writing to mounted devices
  block: Remove blkdev_get_by_*() functions
  bcachefs: Convert to bdev_open_by_path()
  fs: handle freezing from multiple devices
  fs: remove dead check
  nilfs2: simplify device handling
  fs: streamline thaw_super_locked
  ext4: simplify device handling
  xfs: simplify device handling
  fs: simplify setup_bdev_super() calls
  blkdev: comment fs_holder_ops
  porting: document block device freeze and thaw changes
  fs: remove unused helper
  ...

Documentation/filesystems/porting.rst

@@ -1061,3 +1061,15 @@ export_operations ->encode_fh() no longer has a default implementation to
 encode FILEID_INO32_GEN* file handles.
 Filesystems that used the default implementation may use the generic helper
 generic_encode_ino32_fh() explicitly.
+
+---
+
+**recommended**
+
+Block device freezing and thawing have been moved to holder operations.
+
+Before this change, get_active_super() would only be able to find the
+superblock of the main block device, i.e., the one stored in sb->s_bdev. Block
+device freezing now works for any block device owned by a given superblock, not
+just the main block device. The get_active_super() helper and bd_fsfreeze_sb
+pointer are gone.

block/Kconfig

@@ -78,6 +78,26 @@ config BLK_DEV_INTEGRITY_T10
 	select CRC_T10DIF
 	select CRC64_ROCKSOFT
 
+config BLK_DEV_WRITE_MOUNTED
+	bool "Allow writing to mounted block devices"
+	default y
+	help
+	When a block device is mounted, writing to its buffer cache is very
+	likely going to cause filesystem corruption. It is also rather easy to
+	crash the kernel in this way since the filesystem has no practical way
+	of detecting these writes to buffer cache and verifying its metadata
+	integrity. However there are some setups that need this capability
+	like running fsck on read-only mounted root device, modifying some
+	features on mounted ext4 filesystem, and similar. If you say N, the
+	kernel will prevent processes from writing to block devices that are
+	mounted by filesystems which provides some more protection from runaway
+	privileged processes and generally makes it much harder to crash
+	filesystem drivers. Note however that this does not prevent
+	underlying device(s) from being modified by other means, e.g. by
+	directly submitting SCSI commands or through access to lower layers of
+	storage stack. If in doubt, say Y. The configuration can be overridden
+	with the bdev_allow_write_mounted boot option.
+
 config BLK_DEV_ZONED
 	bool "Zoned block device support"
 	select MQ_IOSCHED_DEADLINE

block/bdev.c

@@ -30,6 +30,9 @@
 #include "../fs/internal.h"
 #include "blk.h"
 
+/* Should we allow writing to mounted block devices? */
+static bool bdev_allow_write_mounted = IS_ENABLED(CONFIG_BLK_DEV_WRITE_MOUNTED);
+
 struct bdev_inode {
 	struct block_device bdev;
 	struct inode vfs_inode;
@@ -207,85 +210,88 @@ int sync_blockdev_range(struct block_device *bdev, loff_t lstart, loff_t lend)
 EXPORT_SYMBOL(sync_blockdev_range);
 
 /**
- * freeze_bdev - lock a filesystem and force it into a consistent state
+ * bdev_freeze - lock a filesystem and force it into a consistent state
  * @bdev:	blockdevice to lock
  *
  * If a superblock is found on this device, we take the s_umount semaphore
  * on it to make sure nobody unmounts until the snapshot creation is done.
  * The reference counter (bd_fsfreeze_count) guarantees that only the last
  * unfreeze process can unfreeze the frozen filesystem actually when multiple
- * freeze requests arrive simultaneously. It counts up in freeze_bdev() and
- * count down in thaw_bdev(). When it becomes 0, thaw_bdev() will unfreeze
+ * freeze requests arrive simultaneously. It counts up in bdev_freeze() and
+ * count down in bdev_thaw(). When it becomes 0, thaw_bdev() will unfreeze
  * actually.
+ *
+ * Return: On success zero is returned, negative error code on failure.
  */
-int freeze_bdev(struct block_device *bdev)
+int bdev_freeze(struct block_device *bdev)
 {
-	struct super_block *sb;
 	int error = 0;
 
 	mutex_lock(&bdev->bd_fsfreeze_mutex);
-	if (++bdev->bd_fsfreeze_count > 1)
-		goto done;
-
-	sb = get_active_super(bdev);
-	if (!sb)
-		goto sync;
-	if (sb->s_op->freeze_super)
-		error = sb->s_op->freeze_super(sb, FREEZE_HOLDER_USERSPACE);
-	else
-		error = freeze_super(sb, FREEZE_HOLDER_USERSPACE);
-	deactivate_super(sb);
 
-	if (error) {
-		bdev->bd_fsfreeze_count--;
-		goto done;
+	if (atomic_inc_return(&bdev->bd_fsfreeze_count) > 1) {
+		mutex_unlock(&bdev->bd_fsfreeze_mutex);
+		return 0;
 	}
-	bdev->bd_fsfreeze_sb = sb;
 
-sync:
-	sync_blockdev(bdev);
-done:
+	mutex_lock(&bdev->bd_holder_lock);
+	if (bdev->bd_holder_ops && bdev->bd_holder_ops->freeze) {
+		error = bdev->bd_holder_ops->freeze(bdev);
+		lockdep_assert_not_held(&bdev->bd_holder_lock);
+	} else {
+		mutex_unlock(&bdev->bd_holder_lock);
+		error = sync_blockdev(bdev);
+	}
+
+	if (error)
+		atomic_dec(&bdev->bd_fsfreeze_count);
+
 	mutex_unlock(&bdev->bd_fsfreeze_mutex);
 	return error;
 }
-EXPORT_SYMBOL(freeze_bdev);
+EXPORT_SYMBOL(bdev_freeze);
 
 /**
- * thaw_bdev - unlock filesystem
+ * bdev_thaw - unlock filesystem
  * @bdev:	blockdevice to unlock
  *
- * Unlocks the filesystem and marks it writeable again after freeze_bdev().
+ * Unlocks the filesystem and marks it writeable again after bdev_freeze().
+ *
+ * Return: On success zero is returned, negative error code on failure.
  */
-int thaw_bdev(struct block_device *bdev)
+int bdev_thaw(struct block_device *bdev)
 {
-	struct super_block *sb;
-	int error = -EINVAL;
+	int error = -EINVAL, nr_freeze;
 
 	mutex_lock(&bdev->bd_fsfreeze_mutex);
-	if (!bdev->bd_fsfreeze_count)
+
+	/*
+	 * If this returns < 0 it means that @bd_fsfreeze_count was
+	 * already 0 and no decrement was performed.
+	 */
+	nr_freeze = atomic_dec_if_positive(&bdev->bd_fsfreeze_count);
+	if (nr_freeze < 0)
 		goto out;
 
 	error = 0;
-	if (--bdev->bd_fsfreeze_count > 0)
+	if (nr_freeze > 0)
 		goto out;
 
-	sb = bdev->bd_fsfreeze_sb;
-	if (!sb)
-		goto out;
+	mutex_lock(&bdev->bd_holder_lock);
+	if (bdev->bd_holder_ops && bdev->bd_holder_ops->thaw) {
+		error = bdev->bd_holder_ops->thaw(bdev);
+		lockdep_assert_not_held(&bdev->bd_holder_lock);
+	} else {
+		mutex_unlock(&bdev->bd_holder_lock);
+	}
 
-	if (sb->s_op->thaw_super)
-		error = sb->s_op->thaw_super(sb, FREEZE_HOLDER_USERSPACE);
-	else
-		error = thaw_super(sb, FREEZE_HOLDER_USERSPACE);
 	if (error)
-		bdev->bd_fsfreeze_count++;
-	else
-		bdev->bd_fsfreeze_sb = NULL;
+		atomic_inc(&bdev->bd_fsfreeze_count);
 out:
 	mutex_unlock(&bdev->bd_fsfreeze_mutex);
 	return error;
 }
-EXPORT_SYMBOL(thaw_bdev);
+EXPORT_SYMBOL(bdev_thaw);
 
 /*
  * pseudo-fs
@@ -730,8 +736,59 @@ void blkdev_put_no_open(struct block_device *bdev)
 	put_device(&bdev->bd_device);
 }
 
+static bool bdev_writes_blocked(struct block_device *bdev)
+{
+	return bdev->bd_writers == -1;
+}
+
+static void bdev_block_writes(struct block_device *bdev)
+{
+	bdev->bd_writers = -1;
+}
+
+static void bdev_unblock_writes(struct block_device *bdev)
+{
+	bdev->bd_writers = 0;
+}
+
+static bool bdev_may_open(struct block_device *bdev, blk_mode_t mode)
+{
+	if (bdev_allow_write_mounted)
+		return true;
+	/* Writes blocked? */
+	if (mode & BLK_OPEN_WRITE && bdev_writes_blocked(bdev))
+		return false;
+	if (mode & BLK_OPEN_RESTRICT_WRITES && bdev->bd_writers > 0)
+		return false;
+	return true;
+}
+
+static void bdev_claim_write_access(struct block_device *bdev, blk_mode_t mode)
+{
+	if (bdev_allow_write_mounted)
+		return;
+
+	/* Claim exclusive or shared write access. */
+	if (mode & BLK_OPEN_RESTRICT_WRITES)
+		bdev_block_writes(bdev);
+	else if (mode & BLK_OPEN_WRITE)
+		bdev->bd_writers++;
+}
+
+static void bdev_yield_write_access(struct block_device *bdev, blk_mode_t mode)
+{
+	if (bdev_allow_write_mounted)
+		return;
+
+	/* Yield exclusive or shared write access. */
+	if (mode & BLK_OPEN_RESTRICT_WRITES)
+		bdev_unblock_writes(bdev);
+	else if (mode & BLK_OPEN_WRITE)
+		bdev->bd_writers--;
+}
+
 /**
- * blkdev_get_by_dev - open a block device by device number
+ * bdev_open_by_dev - open a block device by device number
  * @dev: device number of block device to open
  * @mode: open mode (BLK_OPEN_*)
  * @holder: exclusive holder identifier
@@ -743,32 +800,46 @@ void blkdev_put_no_open(struct block_device *bdev)
  *
  * Use this interface ONLY if you really do not have anything better - i.e. when
  * you are behind a truly sucky interface and all you are given is a device
- * number.  Everything else should use blkdev_get_by_path().
+ * number.  Everything else should use bdev_open_by_path().
  *
  * CONTEXT:
  * Might sleep.
  *
  * RETURNS:
- * Reference to the block_device on success, ERR_PTR(-errno) on failure.
+ * Handle with a reference to the block_device on success, ERR_PTR(-errno) on
+ * failure.
  */
-struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder,
+struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
 				     const struct blk_holder_ops *hops)
 {
-	bool unblock_events = true;
+	struct bdev_handle *handle = kmalloc(sizeof(struct bdev_handle),
+					     GFP_KERNEL);
 	struct block_device *bdev;
+	bool unblock_events = true;
 	struct gendisk *disk;
 	int ret;
 
+	if (!handle)
+		return ERR_PTR(-ENOMEM);
+
 	ret = devcgroup_check_permission(DEVCG_DEV_BLOCK,
 			MAJOR(dev), MINOR(dev),
 			((mode & BLK_OPEN_READ) ? DEVCG_ACC_READ : 0) |
 			((mode & BLK_OPEN_WRITE) ? DEVCG_ACC_WRITE : 0));
 	if (ret)
-		return ERR_PTR(ret);
+		goto free_handle;
+
+	/* Blocking writes requires exclusive opener */
+	if (mode & BLK_OPEN_RESTRICT_WRITES && !holder) {
+		ret = -EINVAL;
+		goto free_handle;
+	}
 
 	bdev = blkdev_get_no_open(dev);
-	if (!bdev)
-		return ERR_PTR(-ENXIO);
+	if (!bdev) {
+		ret = -ENXIO;
+		goto free_handle;
+	}
 	disk = bdev->bd_disk;
 
 	if (holder) {
@@ -791,12 +862,16 @@ struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder,
 		goto abort_claiming;
 	if (!try_module_get(disk->fops->owner))
 		goto abort_claiming;
+	ret = -EBUSY;
+	if (!bdev_may_open(bdev, mode))
+		goto abort_claiming;
 	if (bdev_is_partition(bdev))
 		ret = blkdev_get_part(bdev, mode);
 	else
 		ret = blkdev_get_whole(bdev, mode);
 	if (ret)
 		goto put_module;
+	bdev_claim_write_access(bdev, mode);
 	if (holder) {
 		bd_finish_claiming(bdev, holder, hops);
 
@@ -817,7 +892,10 @@ struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder,
 
 	if (unblock_events)
 		disk_unblock_events(disk);
-	return bdev;
+	handle->bdev = bdev;
+	handle->holder = holder;
+	handle->mode = mode;
+	return handle;
 put_module:
 	module_put(disk->fops->owner);
 abort_claiming:
@@ -827,34 +905,14 @@ struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder,
 	disk_unblock_events(disk);
 put_blkdev:
 	blkdev_put_no_open(bdev);
-	return ERR_PTR(ret);
-}
-EXPORT_SYMBOL(blkdev_get_by_dev);
-
-struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
-				     const struct blk_holder_ops *hops)
-{
-	struct bdev_handle *handle = kmalloc(sizeof(*handle), GFP_KERNEL);
-	struct block_device *bdev;
-
-	if (!handle)
-		return ERR_PTR(-ENOMEM);
-	bdev = blkdev_get_by_dev(dev, mode, holder, hops);
-	if (IS_ERR(bdev)) {
+free_handle:
 	kfree(handle);
-		return ERR_CAST(bdev);
-	}
-	handle->bdev = bdev;
-	handle->holder = holder;
-	if (holder)
-		mode |= BLK_OPEN_EXCL;
-	handle->mode = mode;
-	return handle;
+	return ERR_PTR(ret);
 }
 EXPORT_SYMBOL(bdev_open_by_dev);
 
 /**
- * blkdev_get_by_path - open a block device by name
+ * bdev_open_by_path - open a block device by name
  * @path: path to the block device to open
  * @mode: open mode (BLK_OPEN_*)
  * @holder: exclusive holder identifier
@@ -868,29 +926,9 @@ EXPORT_SYMBOL(bdev_open_by_dev);
  * Might sleep.
  *
  * RETURNS:
- * Reference to the block_device on success, ERR_PTR(-errno) on failure.
+ * Handle with a reference to the block_device on success, ERR_PTR(-errno) on
+ * failure.
  */
-struct block_device *blkdev_get_by_path(const char *path, blk_mode_t mode,
-		void *holder, const struct blk_holder_ops *hops)
-{
-	struct block_device *bdev;
-	dev_t dev;
-	int error;
-
-	error = lookup_bdev(path, &dev);
-	if (error)
-		return ERR_PTR(error);
-
-	bdev = blkdev_get_by_dev(dev, mode, holder, hops);
-	if (!IS_ERR(bdev) && (mode & BLK_OPEN_WRITE) && bdev_read_only(bdev)) {
-		blkdev_put(bdev, holder);
-		return ERR_PTR(-EACCES);
-	}
-
-	return bdev;
-}
-EXPORT_SYMBOL(blkdev_get_by_path);
-
 struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode,
 		void *holder, const struct blk_holder_ops *hops)
 {
@@ -913,8 +951,9 @@ struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode,
 }
 EXPORT_SYMBOL(bdev_open_by_path);
 
-void blkdev_put(struct block_device *bdev, void *holder)
+void bdev_release(struct bdev_handle *handle)
 {
+	struct block_device *bdev = handle->bdev;
 	struct gendisk *disk = bdev->bd_disk;
 
 	/*
@@ -928,8 +967,10 @@ void blkdev_put(struct block_device *bdev, void *holder)
 		sync_blockdev(bdev);
 
 	mutex_lock(&disk->open_mutex);
-	if (holder)
-		bd_end_claim(bdev, holder);
+	bdev_yield_write_access(bdev, handle->mode);
+
+	if (handle->holder)
+		bd_end_claim(bdev, handle->holder);
 
 	/*
 	 * Trigger event checking and tell drivers to flush MEDIA_CHANGE
@@ -946,12 +987,6 @@ void blkdev_put(struct block_device *bdev, void *holder)
 
 	module_put(disk->fops->owner);
 	blkdev_put_no_open(bdev);
-}
-EXPORT_SYMBOL(blkdev_put);
-
-void bdev_release(struct bdev_handle *handle)
-{
-	blkdev_put(handle->bdev, handle->holder);
 	kfree(handle);
 }
 EXPORT_SYMBOL(bdev_release);
@@ -1102,3 +1137,12 @@ void bdev_statx_dioalign(struct inode *inode, struct kstat *stat)
 
 	blkdev_put_no_open(bdev);
 }
+
+static int __init setup_bdev_allow_write_mounted(char *str)
+{
+	if (kstrtobool(str, &bdev_allow_write_mounted))
+		pr_warn("Invalid option string for bdev_allow_write_mounted:"
+			" '%s'\n", str);
+	return 1;
+}
+__setup("bdev_allow_write_mounted=", setup_bdev_allow_write_mounted);

drivers/md/dm.c

@@ -2675,7 +2675,7 @@ static int lock_fs(struct mapped_device *md)
 
 	WARN_ON(test_bit(DMF_FROZEN, &md->flags));
 
-	r = freeze_bdev(md->disk->part0);
+	r = bdev_freeze(md->disk->part0);
 	if (!r)
 		set_bit(DMF_FROZEN, &md->flags);
 	return r;
@@ -2685,7 +2685,7 @@ static void unlock_fs(struct mapped_device *md)
 {
 	if (!test_bit(DMF_FROZEN, &md->flags))
 		return;
-	thaw_bdev(md->disk->part0);
+	bdev_thaw(md->disk->part0);
 	clear_bit(DMF_FROZEN, &md->flags);
 }
 

fs/bcachefs/fs-ioctl.c

@@ -289,14 +289,14 @@ static int bch2_ioc_goingdown(struct bch_fs *c, u32 __user *arg)
 
 	switch (flags) {
 	case FSOP_GOING_FLAGS_DEFAULT:
-		ret = freeze_bdev(c->vfs_sb->s_bdev);
+		ret = bdev_freeze(c->vfs_sb->s_bdev);
 		if (ret)
 			goto err;
 
 		bch2_journal_flush(&c->journal);
 		c->vfs_sb->s_flags |= SB_RDONLY;
 		bch2_fs_emergency_read_only(c);
-		thaw_bdev(c->vfs_sb->s_bdev);
+		bdev_thaw(c->vfs_sb->s_bdev);
 		break;
 
 	case FSOP_GOING_FLAGS_LOGFLUSH:

fs/bcachefs/super-io.c

@@ -164,8 +164,8 @@ void bch2_sb_field_delete(struct bch_sb_handle *sb,
 void bch2_free_super(struct bch_sb_handle *sb)
 {
 	kfree(sb->bio);
-	if (!IS_ERR_OR_NULL(sb->bdev))
-		blkdev_put(sb->bdev, sb->holder);
+	if (!IS_ERR_OR_NULL(sb->bdev_handle))
+		bdev_release(sb->bdev_handle);
 	kfree(sb->holder);
 	kfree(sb->sb_name);
 
@@ -725,21 +725,22 @@ int bch2_read_super(const char *path, struct bch_opts *opts,
 	if (!opt_get(*opts, nochanges))
 		sb->mode |= BLK_OPEN_WRITE;
 
-	sb->bdev = blkdev_get_by_path(path, sb->mode, sb->holder, &bch2_sb_handle_bdev_ops);
-	if (IS_ERR(sb->bdev) &&
-	    PTR_ERR(sb->bdev) == -EACCES &&
+	sb->bdev_handle = bdev_open_by_path(path, sb->mode, sb->holder, &bch2_sb_handle_bdev_ops);
+	if (IS_ERR(sb->bdev_handle) &&
+	    PTR_ERR(sb->bdev_handle) == -EACCES &&
 	    opt_get(*opts, read_only)) {
 		sb->mode &= ~BLK_OPEN_WRITE;
 
-		sb->bdev = blkdev_get_by_path(path, sb->mode, sb->holder, &bch2_sb_handle_bdev_ops);
-		if (!IS_ERR(sb->bdev))
+		sb->bdev_handle = bdev_open_by_path(path, sb->mode, sb->holder, &bch2_sb_handle_bdev_ops);
+		if (!IS_ERR(sb->bdev_handle))
 			opt_set(*opts, nochanges, true);
 	}
 
-	if (IS_ERR(sb->bdev)) {
-		ret = PTR_ERR(sb->bdev);
+	if (IS_ERR(sb->bdev_handle)) {
+		ret = PTR_ERR(sb->bdev_handle);
 		goto out;
 	}
+	sb->bdev = sb->bdev_handle->bdev;
 
 	ret = bch2_sb_realloc(sb, 0);
 	if (ret) {

fs/bcachefs/super_types.h

@@ -4,6 +4,7 @@
 
 struct bch_sb_handle {
 	struct bch_sb		*sb;
+	struct bdev_handle	*bdev_handle;
 	struct block_device	*bdev;
 	char			*sb_name;
 	struct bio		*bio;

fs/btrfs/super.c

@@ -1406,6 +1406,8 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type,
 			return ERR_PTR(error);
 	}
 
+	/* No support for restricting writes to btrfs devices yet... */
+	mode &= ~BLK_OPEN_RESTRICT_WRITES;
 	/*
 	 * Setup a dummy root and fs_info for test/set super.  This is because
 	 * we don't actually fill this stuff out until open_ctree, but we need

fs/ext4/ioctl.c

@@ -819,11 +819,11 @@ int ext4_force_shutdown(struct super_block *sb, u32 flags)
 
 	switch (flags) {
 	case EXT4_GOING_FLAGS_DEFAULT:
-		ret = freeze_bdev(sb->s_bdev);
+		ret = bdev_freeze(sb->s_bdev);
 		if (ret)
 			return ret;
 		set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
-		thaw_bdev(sb->s_bdev);
+		bdev_thaw(sb->s_bdev);
 		break;
 	case EXT4_GOING_FLAGS_LOGFLUSH:
 		set_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);

fs/ext4/super.c

@@ -5864,11 +5864,9 @@ static struct bdev_handle *ext4_get_journal_blkdev(struct super_block *sb,
 	struct ext4_super_block *es;
 	int errno;
 
-	/* see get_tree_bdev why this is needed and safe */
-	up_write(&sb->s_umount);
-	bdev_handle = bdev_open_by_dev(j_dev, BLK_OPEN_READ | BLK_OPEN_WRITE,
+	bdev_handle = bdev_open_by_dev(j_dev,
+		BLK_OPEN_READ | BLK_OPEN_WRITE | BLK_OPEN_RESTRICT_WRITES,
 		sb, &fs_holder_ops);
-	down_write(&sb->s_umount);
 	if (IS_ERR(bdev_handle)) {
 		ext4_msg(sb, KERN_ERR,
 			 "failed to open journal device unknown-block(%u,%u) %ld",

fs/f2fs/file.c

@@ -2239,11 +2239,11 @@ static int f2fs_ioc_shutdown(struct file *filp, unsigned long arg)
 
 	switch (in) {
 	case F2FS_GOING_DOWN_FULLSYNC:
-		ret = freeze_bdev(sb->s_bdev);
+		ret = bdev_freeze(sb->s_bdev);
 		if (ret)
 			goto out;
 		f2fs_stop_checkpoint(sbi, false, STOP_CP_REASON_SHUTDOWN);
-		thaw_bdev(sb->s_bdev);
+		bdev_thaw(sb->s_bdev);
 		break;
 	case F2FS_GOING_DOWN_METASYNC:
 		/* do checkpoint only */

fs/nilfs2/super.c

@@ -1314,15 +1314,7 @@ nilfs_mount(struct file_system_type *fs_type, int flags,
 		return ERR_CAST(s);
 
 	if (!s->s_root) {
-		/*
-		 * We drop s_umount here because we need to open the bdev and
-		 * bdev->open_mutex ranks above s_umount (blkdev_put() ->
-		 * __invalidate_device()). It is safe because we have active sb
-		 * reference and SB_BORN is not set yet.
-		 */
-		up_write(&s->s_umount);
 		err = setup_bdev_super(s, flags, NULL);
-		down_write(&s->s_umount);
 		if (!err)
 			err = nilfs_fill_super(s, data,
 					       flags & SB_SILENT ? 1 : 0);

fs/super.c

@@ -81,16 +81,13 @@ static inline void super_unlock_shared(struct super_block *sb)
 	super_unlock(sb, false);
 }
 
-static inline bool wait_born(struct super_block *sb)
+static bool super_flags(const struct super_block *sb, unsigned int flags)
 {
-	unsigned int flags;
-
 	/*
 	 * Pairs with smp_store_release() in super_wake() and ensures
-	 * that we see SB_BORN or SB_DYING after we're woken.
+	 * that we see @flags after we're woken.
 	 */
-	flags = smp_load_acquire(&sb->s_flags);
-	return flags & (SB_BORN | SB_DYING);
+	return smp_load_acquire(&sb->s_flags) & flags;
 }
 
 /**
@@ -105,15 +102,21 @@ static inline bool wait_born(struct super_block *sb)
  *
  * The caller must have acquired a temporary reference on @sb->s_count.
  *
- * Return: This returns true if SB_BORN was set, false if SB_DYING was
- *         set. The function acquires s_umount and returns with it held.
+ * Return: The function returns true if SB_BORN was set and with
+ *         s_umount held. The function returns false if SB_DYING was
+ *         set and without s_umount held.
  */
 static __must_check bool super_lock(struct super_block *sb, bool excl)
 {
-
 	lockdep_assert_not_held(&sb->s_umount);
 
-relock:
+	/* wait until the superblock is ready or dying */
+	wait_var_event(&sb->s_flags, super_flags(sb, SB_BORN | SB_DYING));
+
+	/* Don't pointlessly acquire s_umount. */
+	if (super_flags(sb, SB_DYING))
+		return false;
+
 	__super_lock(sb, excl);
 
 	/*
@@ -121,32 +124,22 @@ static __must_check bool super_lock(struct super_block *sb, bool excl)
 	 * @sb->s_root is NULL and @sb->s_active is 0. No one needs to
 	 * grab a reference to this. Tell them so.
 	 */
-	if (sb->s_flags & SB_DYING)
+	if (sb->s_flags & SB_DYING) {
+		super_unlock(sb, excl);
 		return false;
+	}
 
-	/* Has called ->get_tree() successfully. */
-	if (sb->s_flags & SB_BORN)
+	WARN_ON_ONCE(!(sb->s_flags & SB_BORN));
 	return true;
-
-	super_unlock(sb, excl);
-
-	/* wait until the superblock is ready or dying */
-	wait_var_event(&sb->s_flags, wait_born(sb));
-
-	/*
-	 * Neither SB_BORN nor SB_DYING are ever unset so we never loop.
-	 * Just reacquire @sb->s_umount for the caller.
-	 */
-	goto relock;
 }
 
-/* wait and acquire read-side of @sb->s_umount */
+/* wait and try to acquire read-side of @sb->s_umount */
 static inline bool super_lock_shared(struct super_block *sb)
 {
 	return super_lock(sb, false);
 }
 
-/* wait and acquire write-side of @sb->s_umount */
+/* wait and try to acquire write-side of @sb->s_umount */
 static inline bool super_lock_excl(struct super_block *sb)
 {
 	return super_lock(sb, true);
@@ -521,48 +514,7 @@ void deactivate_super(struct super_block *s)
 EXPORT_SYMBOL(deactivate_super);
 
 /**
- *	grab_super - acquire an active reference
- *	@s: reference we are trying to make active
- *
- *	Tries to acquire an active reference.  grab_super() is used when we
- * 	had just found a superblock in super_blocks or fs_type->fs_supers
- *	and want to turn it into a full-blown active reference.  grab_super()
- *	is called with sb_lock held and drops it.  Returns 1 in case of
- *	success, 0 if we had failed (superblock contents was already dead or
- *	dying when grab_super() had been called).  Note that this is only
- *	called for superblocks not in rundown mode (== ones still on ->fs_supers
- *	of their type), so increment of ->s_count is OK here.
- */
-static int grab_super(struct super_block *s) __releases(sb_lock)
-{
-	bool born;
-
-	s->s_count++;
-	spin_unlock(&sb_lock);
-	born = super_lock_excl(s);
-	if (born && atomic_inc_not_zero(&s->s_active)) {
-		put_super(s);
-		return 1;
-	}
-	super_unlock_excl(s);
-	put_super(s);
-	return 0;
-}
-
-static inline bool wait_dead(struct super_block *sb)
-{
-	unsigned int flags;
-
-	/*
-	 * Pairs with memory barrier in super_wake() and ensures
-	 * that we see SB_DEAD after we're woken.
-	 */
-	flags = smp_load_acquire(&sb->s_flags);
-	return flags & SB_DEAD;
-}
-
-/**
- * grab_super_dead - acquire an active reference to a superblock
+ * grab_super - acquire an active reference to a superblock
  * @sb: superblock to acquire
  *
  * Acquire a temporary reference on a superblock and try to trade it for
@@ -573,17 +525,21 @@ static inline bool wait_dead(struct super_block *sb)
  * Return: This returns true if an active reference could be acquired,
  *         false if not.
  */
-static bool grab_super_dead(struct super_block *sb)
+static bool grab_super(struct super_block *sb)
 {
+	bool locked;
 
 	sb->s_count++;
-	if (grab_super(sb)) {
+	spin_unlock(&sb_lock);
+	locked = super_lock_excl(sb);
+	if (locked) {
+		if (atomic_inc_not_zero(&sb->s_active)) {
 			put_super(sb);
-		lockdep_assert_held(&sb->s_umount);
 			return true;
 		}
-	wait_var_event(&sb->s_flags, wait_dead(sb));
-	lockdep_assert_not_held(&sb->s_umount);
+		super_unlock_excl(sb);
+	}
+	wait_var_event(&sb->s_flags, super_flags(sb, SB_DEAD));
 	put_super(sb);
 	return false;
 }
@@ -834,7 +790,7 @@ struct super_block *sget_fc(struct fs_context *fc,
 			warnfc(fc, "reusing existing filesystem in another namespace not allowed");
 		return ERR_PTR(-EBUSY);
 	}
-	if (!grab_super_dead(old))
+	if (!grab_super(old))
 		goto retry;
 	destroy_unused_super(s);
 	return old;
@@ -878,7 +834,7 @@ struct super_block *sget(struct file_system_type *type,
 				destroy_unused_super(s);
 				return ERR_PTR(-EBUSY);
 			}
-			if (!grab_super_dead(old))
+			if (!grab_super(old))
 				goto retry;
 			destroy_unused_super(s);
 			return old;
@@ -930,8 +886,7 @@ static void __iterate_supers(void (*f)(struct super_block *))
 
 	spin_lock(&sb_lock);
 	list_for_each_entry(sb, &super_blocks, s_list) {
-		/* Pairs with memory marrier in super_wake(). */
-		if (smp_load_acquire(&sb->s_flags) & SB_DYING)
+		if (super_flags(sb, SB_DYING))
 			continue;
 		sb->s_count++;
 		spin_unlock(&sb_lock);
@@ -961,15 +916,17 @@ void iterate_supers(void (*f)(struct super_block *, void *), void *arg)
 
 	spin_lock(&sb_lock);
 	list_for_each_entry(sb, &super_blocks, s_list) {
-		bool born;
+		bool locked;
 
 		sb->s_count++;
 		spin_unlock(&sb_lock);
 
-		born = super_lock_shared(sb);
-		if (born && sb->s_root)
+		locked = super_lock_shared(sb);
+		if (locked) {
+			if (sb->s_root)
 				f(sb, arg);
 			super_unlock_shared(sb);
+		}
 
 		spin_lock(&sb_lock);
 		if (p)
@@ -997,15 +954,17 @@ void iterate_supers_type(struct file_system_type *type,
 
 	spin_lock(&sb_lock);
 	hlist_for_each_entry(sb, &type->fs_supers, s_instances) {
-		bool born;
+		bool locked;
 
 		sb->s_count++;
 		spin_unlock(&sb_lock);
 
-		born = super_lock_shared(sb);
-		if (born && sb->s_root)
+		locked = super_lock_shared(sb);
+		if (locked) {
+			if (sb->s_root)
 				f(sb, arg);
 			super_unlock_shared(sb);
+		}
 
 		spin_lock(&sb_lock);
 		if (p)
@@ -1019,34 +978,6 @@ void iterate_supers_type(struct file_system_type *type,
 
 EXPORT_SYMBOL(iterate_supers_type);
 
-/**
- * get_active_super - get an active reference to the superblock of a device
- * @bdev: device to get the superblock for
- *
- * Scans the superblock list and finds the superblock of the file system
- * mounted on the device given.  Returns the superblock with an active
- * reference or %NULL if none was found.
- */
-struct super_block *get_active_super(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 (!grab_super(sb))
-				return NULL;
-			super_unlock_excl(sb);
-			return sb;
-		}
-	}
-	spin_unlock(&sb_lock);
-	return NULL;
-}
-
 struct super_block *user_get_super(dev_t dev, bool excl)
 {
 	struct super_block *sb;
@@ -1054,15 +985,17 @@ struct super_block *user_get_super(dev_t dev, bool excl)
 	spin_lock(&sb_lock);
 	list_for_each_entry(sb, &super_blocks, s_list) {
 		if (sb->s_dev ==  dev) {
-			bool born;
+			bool locked;
 
 			sb->s_count++;
 			spin_unlock(&sb_lock);
 			/* still alive? */
-			born = super_lock(sb, excl);
-			if (born && sb->s_root)
+			locked = super_lock(sb, excl);
+			if (locked) {
+				if (sb->s_root)
 					return sb;
 				super_unlock(sb, excl);
+			}
 			/* nope, got unmounted */
 			spin_lock(&sb_lock);
 			__put_super(sb);
@@ -1173,9 +1106,9 @@ int reconfigure_super(struct fs_context *fc)
 
 static void do_emergency_remount_callback(struct super_block *sb)
 {
-	bool born = super_lock_excl(sb);
+	bool locked = super_lock_excl(sb);
 
-	if (born && sb->s_root && sb->s_bdev && !sb_rdonly(sb)) {
+	if (locked && sb->s_root && sb->s_bdev && !sb_rdonly(sb)) {
 		struct fs_context *fc;
 
 		fc = fs_context_for_reconfigure(sb->s_root,
@@ -1186,6 +1119,7 @@ static void do_emergency_remount_callback(struct super_block *sb)
 			put_fs_context(fc);
 		}
 	}
+	if (locked)
 		super_unlock_excl(sb);
 }
 
@@ -1209,16 +1143,17 @@ void emergency_remount(void)
 
 static void do_thaw_all_callback(struct super_block *sb)
 {
-	bool born = super_lock_excl(sb);
+	bool locked = super_lock_excl(sb);
 
-	if (born && sb->s_root) {
+	if (locked && sb->s_root) {
 		if (IS_ENABLED(CONFIG_BLOCK))
-			while (sb->s_bdev && !thaw_bdev(sb->s_bdev))
+			while (sb->s_bdev && !bdev_thaw(sb->s_bdev))
 				pr_warn("Emergency Thaw on %pg\n", sb->s_bdev);
 		thaw_super_locked(sb, FREEZE_HOLDER_USERSPACE);
-	} else {
-		super_unlock_excl(sb);
+		return;
 	}
+	if (locked)
+		super_unlock_excl(sb);
 }
 
 static void do_thaw_all(struct work_struct *work)
@@ -1428,11 +1363,11 @@ EXPORT_SYMBOL(sget_dev);
  *
  * The function must be called with bdev->bd_holder_lock and releases it.
  */
-static struct super_block *bdev_super_lock_shared(struct block_device *bdev)
+static struct super_block *bdev_super_lock(struct block_device *bdev, bool excl)
 	__releases(&bdev->bd_holder_lock)
 {
 	struct super_block *sb = bdev->bd_holder;
-	bool born;
+	bool locked;
 
 	lockdep_assert_held(&bdev->bd_holder_lock);
 	lockdep_assert_not_held(&sb->s_umount);
@@ -1442,19 +1377,25 @@ static struct super_block *bdev_super_lock_shared(struct block_device *bdev)
 	spin_lock(&sb_lock);
 	sb->s_count++;
 	spin_unlock(&sb_lock);
+
 	mutex_unlock(&bdev->bd_holder_lock);
 
-	born = super_lock_shared(sb);
-	if (!born || !sb->s_root || !(sb->s_flags & SB_ACTIVE)) {
-		super_unlock_shared(sb);
-		put_super(sb);
-		return NULL;
-	}
+	locked = super_lock(sb, excl);
+
 	/*
-	 * The superblock is active and we hold s_umount, we can drop our
-	 * temporary reference now.
+	 * If the superblock wasn't already SB_DYING then we hold
+	 * s_umount and can safely drop our temporary reference.
          */
 	put_super(sb);
+
+	if (!locked)
+		return NULL;
+
+	if (!sb->s_root || !(sb->s_flags & SB_ACTIVE)) {
+		super_unlock(sb, excl);
+		return NULL;
+	}
+
 	return sb;
 }
 
@@ -1462,7 +1403,7 @@ static void fs_bdev_mark_dead(struct block_device *bdev, bool surprise)
 {
 	struct super_block *sb;
 
-	sb = bdev_super_lock_shared(bdev);
+	sb = bdev_super_lock(bdev, false);
 	if (!sb)
 		return;
 
@@ -1480,16 +1421,110 @@ static void fs_bdev_sync(struct block_device *bdev)
 {
 	struct super_block *sb;
 
-	sb = bdev_super_lock_shared(bdev);
+	sb = bdev_super_lock(bdev, false);
 	if (!sb)
 		return;
+
 	sync_filesystem(sb);
 	super_unlock_shared(sb);
 }
 
+static struct super_block *get_bdev_super(struct block_device *bdev)
+{
+	bool active = false;
+	struct super_block *sb;
+
+	sb = bdev_super_lock(bdev, true);
+	if (sb) {
+		active = atomic_inc_not_zero(&sb->s_active);
+		super_unlock_excl(sb);
+	}
+	if (!active)
+		return NULL;
+	return sb;
+}
+
+/**
+ * fs_bdev_freeze - freeze owning filesystem of block device
+ * @bdev: block device
+ *
+ * Freeze the filesystem that owns this block device if it is still
+ * active.
+ *
+ * A filesystem that owns multiple block devices may be frozen from each
+ * block device and won't be unfrozen until all block devices are
+ * unfrozen. Each block device can only freeze the filesystem once as we
+ * nest freezes for block devices in the block layer.
+ *
+ * Return: If the freeze was successful zero is returned. If the freeze
+ *         failed a negative error code is returned.
+ */
+static int fs_bdev_freeze(struct block_device *bdev)
+{
+	struct super_block *sb;
+	int error = 0;
+
+	lockdep_assert_held(&bdev->bd_fsfreeze_mutex);
+
+	sb = get_bdev_super(bdev);
+	if (!sb)
+		return -EINVAL;
+
+	if (sb->s_op->freeze_super)
+		error = sb->s_op->freeze_super(sb,
+				FREEZE_MAY_NEST | FREEZE_HOLDER_USERSPACE);
+	else
+		error = freeze_super(sb,
+				FREEZE_MAY_NEST | FREEZE_HOLDER_USERSPACE);
+	if (!error)
+		error = sync_blockdev(bdev);
+	deactivate_super(sb);
+	return error;
+}
+
+/**
+ * fs_bdev_thaw - thaw owning filesystem of block device
+ * @bdev: block device
+ *
+ * Thaw the filesystem that owns this block device.
+ *
+ * A filesystem that owns multiple block devices may be frozen from each
+ * block device and won't be unfrozen until all block devices are
+ * unfrozen. Each block device can only freeze the filesystem once as we
+ * nest freezes for block devices in the block layer.
+ *
+ * Return: If the thaw was successful zero is returned. If the thaw
+ *         failed a negative error code is returned. If this function
+ *         returns zero it doesn't mean that the filesystem is unfrozen
+ *         as it may have been frozen multiple times (kernel may hold a
+ *         freeze or might be frozen from other block devices).
+ */
+static int fs_bdev_thaw(struct block_device *bdev)
+{
+	struct super_block *sb;
+	int error;
+
+	lockdep_assert_held(&bdev->bd_fsfreeze_mutex);
+
+	sb = get_bdev_super(bdev);
+	if (WARN_ON_ONCE(!sb))
+		return -EINVAL;
+
+	if (sb->s_op->thaw_super)
+		error = sb->s_op->thaw_super(sb,
+				FREEZE_MAY_NEST | FREEZE_HOLDER_USERSPACE);
+	else
+		error = thaw_super(sb,
+				FREEZE_MAY_NEST | FREEZE_HOLDER_USERSPACE);
+	deactivate_super(sb);
+	return error;
+}
+
 const struct blk_holder_ops fs_holder_ops = {
 	.mark_dead		= fs_bdev_mark_dead,
 	.sync			= fs_bdev_sync,
+	.freeze			= fs_bdev_freeze,
+	.thaw			= fs_bdev_thaw,
 };
 EXPORT_SYMBOL_GPL(fs_holder_ops);
 
@@ -1519,15 +1554,10 @@ int setup_bdev_super(struct super_block *sb, int sb_flags,
 	}
 
 	/*
-	 * Until SB_BORN flag is set, there can be no active superblock
-	 * references and thus no filesystem freezing. get_active_super() will
-	 * just loop waiting for SB_BORN so even freeze_bdev() cannot proceed.
-	 *
-	 * It is enough to check bdev was not frozen before we set s_bdev.
+	 * It is enough to check bdev was not frozen before we set
+	 * s_bdev as freezing will wait until SB_BORN is set.
 	 */
-	mutex_lock(&bdev->bd_fsfreeze_mutex);
-	if (bdev->bd_fsfreeze_count > 0) {
-		mutex_unlock(&bdev->bd_fsfreeze_mutex);
+	if (atomic_read(&bdev->bd_fsfreeze_count) > 0) {
 		if (fc)
 			warnf(fc, "%pg: Can't mount, blockdev is frozen", bdev);
 		bdev_release(bdev_handle);
@@ -1540,7 +1570,6 @@ int setup_bdev_super(struct super_block *sb, int sb_flags,
 	if (bdev_stable_writes(bdev))
 		sb->s_iflags |= SB_I_STABLE_WRITES;
 	spin_unlock(&sb_lock);
-	mutex_unlock(&bdev->bd_fsfreeze_mutex);
 
 	snprintf(sb->s_id, sizeof(sb->s_id), "%pg", bdev);
 	shrinker_debugfs_rename(sb->s_shrink, "sb-%s:%s", sb->s_type->name,
@@ -1585,15 +1614,7 @@ int get_tree_bdev(struct fs_context *fc,
 			return -EBUSY;
 		}
 	} else {
-		/*
-		 * We drop s_umount here because we need to open the bdev and
-		 * bdev->open_mutex ranks above s_umount (blkdev_put() ->
-		 * bdev_mark_dead()). It is safe because we have active sb
-		 * reference and SB_BORN is not set yet.
-		 */
-		super_unlock_excl(s);
 		error = setup_bdev_super(s, fc->sb_flags, fc);
-		__super_lock_excl(s);
 		if (!error)
 			error = fill_super(s, fc);
 		if (error) {
@@ -1637,15 +1658,7 @@ struct dentry *mount_bdev(struct file_system_type *fs_type,
 			return ERR_PTR(-EBUSY);
 		}
 	} else {
-		/*
-		 * We drop s_umount here because we need to open the bdev and
-		 * bdev->open_mutex ranks above s_umount (blkdev_put() ->
-		 * bdev_mark_dead()). It is safe because we have active sb
-		 * reference and SB_BORN is not set yet.
-		 */
-		super_unlock_excl(s);
 		error = setup_bdev_super(s, flags, NULL);
-		__super_lock_excl(s);
 		if (!error)
 			error = fill_super(s, data, flags & SB_SILENT ? 1 : 0);
 		if (error) {
@@ -1914,6 +1927,47 @@ static int wait_for_partially_frozen(struct super_block *sb)
 	return ret;
 }
 
+#define FREEZE_HOLDERS (FREEZE_HOLDER_KERNEL | FREEZE_HOLDER_USERSPACE)
+#define FREEZE_FLAGS (FREEZE_HOLDERS | FREEZE_MAY_NEST)
+
+static inline int freeze_inc(struct super_block *sb, enum freeze_holder who)
+{
+	WARN_ON_ONCE((who & ~FREEZE_FLAGS));
+	WARN_ON_ONCE(hweight32(who & FREEZE_HOLDERS) > 1);
+
+	if (who & FREEZE_HOLDER_KERNEL)
+		++sb->s_writers.freeze_kcount;
+	if (who & FREEZE_HOLDER_USERSPACE)
+		++sb->s_writers.freeze_ucount;
+	return sb->s_writers.freeze_kcount + sb->s_writers.freeze_ucount;
+}
+
+static inline int freeze_dec(struct super_block *sb, enum freeze_holder who)
+{
+	WARN_ON_ONCE((who & ~FREEZE_FLAGS));
+	WARN_ON_ONCE(hweight32(who & FREEZE_HOLDERS) > 1);
+
+	if ((who & FREEZE_HOLDER_KERNEL) && sb->s_writers.freeze_kcount)
+		--sb->s_writers.freeze_kcount;
+	if ((who & FREEZE_HOLDER_USERSPACE) && sb->s_writers.freeze_ucount)
+		--sb->s_writers.freeze_ucount;
+	return sb->s_writers.freeze_kcount + sb->s_writers.freeze_ucount;
+}
+
+static inline bool may_freeze(struct super_block *sb, enum freeze_holder who)
+{
+	WARN_ON_ONCE((who & ~FREEZE_FLAGS));
+	WARN_ON_ONCE(hweight32(who & FREEZE_HOLDERS) > 1);
+
+	if (who & FREEZE_HOLDER_KERNEL)
+		return (who & FREEZE_MAY_NEST) ||
+		       sb->s_writers.freeze_kcount == 0;
+	if (who & FREEZE_HOLDER_USERSPACE)
+		return (who & FREEZE_MAY_NEST) ||
+		       sb->s_writers.freeze_ucount == 0;
+	return false;
+}
+
 /**
  * freeze_super - lock the filesystem and force it into a consistent state
  * @sb: the super to lock
@@ -1926,6 +1980,7 @@ static int wait_for_partially_frozen(struct super_block *sb)
  * @who should be:
  * * %FREEZE_HOLDER_USERSPACE if userspace wants to freeze the fs;
  * * %FREEZE_HOLDER_KERNEL if the kernel wants to freeze the fs.
+ * * %FREEZE_MAY_NEST whether nesting freeze and thaw requests is allowed.
  *
  * The @who argument distinguishes between the kernel and userspace trying to
  * freeze the filesystem.  Although there cannot be multiple kernel freezes or
@@ -1933,6 +1988,13 @@ static int wait_for_partially_frozen(struct super_block *sb)
  * userspace can both hold a filesystem frozen.  The filesystem remains frozen
  * until there are no kernel or userspace freezes in effect.
  *
+ * A filesystem may hold multiple devices and thus a filesystems may be
+ * frozen through the block layer via multiple block devices. In this
+ * case the request is marked as being allowed to nest by passing
+ * FREEZE_MAY_NEST. The filesystem remains frozen until all block
+ * devices are unfrozen. If multiple freezes are attempted without
+ * FREEZE_MAY_NEST -EBUSY will be returned.
+ *
  * During this function, sb->s_writers.frozen goes through these values:
  *
  * SB_UNFROZEN: File system is normal, all writes progress as usual.
@@ -1957,31 +2019,29 @@ static int wait_for_partially_frozen(struct super_block *sb)
  * mostly auxiliary for filesystems to verify they do not modify frozen fs.
  *
  * sb->s_writers.frozen is protected by sb->s_umount.
+ *
+ * Return: If the freeze was successful zero is returned. If the freeze
+ *         failed a negative error code is returned.
  */
 int freeze_super(struct super_block *sb, enum freeze_holder who)
 {
 	int ret;
 
+	if (!super_lock_excl(sb)) {
+		WARN_ON_ONCE("Dying superblock while freezing!");
+		return -EINVAL;
+	}
 	atomic_inc(&sb->s_active);
-	if (!super_lock_excl(sb))
-		WARN(1, "Dying superblock while freezing!");
 
 retry:
 	if (sb->s_writers.frozen == SB_FREEZE_COMPLETE) {
-		if (sb->s_writers.freeze_holders & who) {
+		if (may_freeze(sb, who))
+			ret = !!WARN_ON_ONCE(freeze_inc(sb, who) == 1);
+		else
+			ret = -EBUSY;
+		/* All freezers share a single active reference. */
 		deactivate_locked_super(sb);
-			return -EBUSY;
-		}
-
-		WARN_ON(sb->s_writers.freeze_holders == 0);
-
-		/*
-		 * Someone else already holds this type of freeze; share the
-		 * freeze and assign the active ref to the freeze.
-		 */
-		sb->s_writers.freeze_holders |= who;
-		super_unlock_excl(sb);
-		return 0;
+		return ret;
 	}
 
 	if (sb->s_writers.frozen != SB_UNFROZEN) {
@@ -1994,14 +2054,9 @@ int freeze_super(struct super_block *sb, enum freeze_holder who)
 		goto retry;
 	}
 
-	if (!(sb->s_flags & SB_BORN)) {
-		super_unlock_excl(sb);
-		return 0;	/* sic - it's "nothing to do" */
-	}
-
 	if (sb_rdonly(sb)) {
 		/* Nothing to do really... */
-		sb->s_writers.freeze_holders |= who;
+		WARN_ON_ONCE(freeze_inc(sb, who) > 1);
 		sb->s_writers.frozen = SB_FREEZE_COMPLETE;
 		wake_up_var(&sb->s_writers.frozen);
 		super_unlock_excl(sb);
@@ -2012,8 +2067,7 @@ int freeze_super(struct super_block *sb, enum freeze_holder who)
 	/* Release s_umount to preserve sb_start_write -> s_umount ordering */
 	super_unlock_excl(sb);
 	sb_wait_write(sb, SB_FREEZE_WRITE);
-	if (!super_lock_excl(sb))
-		WARN(1, "Dying superblock while freezing!");
+	__super_lock_excl(sb);
 
 	/* Now we go and block page faults... */
 	sb->s_writers.frozen = SB_FREEZE_PAGEFAULT;
@@ -2049,7 +2103,7 @@ int freeze_super(struct super_block *sb, enum freeze_holder who)
 	 * For debugging purposes so that fs can warn if it sees write activity
 	 * when frozen is set to SB_FREEZE_COMPLETE, and for thaw_super().
 	 */
-	sb->s_writers.freeze_holders |= who;
+	WARN_ON_ONCE(freeze_inc(sb, who) > 1);
 	sb->s_writers.frozen = SB_FREEZE_COMPLETE;
 	wake_up_var(&sb->s_writers.frozen);
 	lockdep_sb_freeze_release(sb);
@@ -2066,34 +2120,22 @@ EXPORT_SYMBOL(freeze_super);
  */
 static int thaw_super_locked(struct super_block *sb, enum freeze_holder who)
 {
-	int error;
+	int error = -EINVAL;
 
-	if (sb->s_writers.frozen == SB_FREEZE_COMPLETE) {
-		if (!(sb->s_writers.freeze_holders & who)) {
-			super_unlock_excl(sb);
-			return -EINVAL;
-		}
+	if (sb->s_writers.frozen != SB_FREEZE_COMPLETE)
+		goto out_unlock;
 
 	/*
-		 * Freeze is shared with someone else.  Release our hold and
-		 * drop the active ref that freeze_super assigned to the
-		 * freezer.
+	 * All freezers share a single active reference.
+	 * So just unlock in case there are any left.
 	 */
-		if (sb->s_writers.freeze_holders & ~who) {
-			sb->s_writers.freeze_holders &= ~who;
-			deactivate_locked_super(sb);
-			return 0;
-		}
-	} else {
-		super_unlock_excl(sb);
-		return -EINVAL;
-	}
+	if (freeze_dec(sb, who))
+		goto out_unlock;
 
 	if (sb_rdonly(sb)) {
-		sb->s_writers.freeze_holders &= ~who;
 		sb->s_writers.frozen = SB_UNFROZEN;
 		wake_up_var(&sb->s_writers.frozen);
-		goto out;
+		goto out_deactivate;
 	}
 
 	lockdep_sb_freeze_acquire(sb);
@@ -2101,20 +2143,23 @@ static int thaw_super_locked(struct super_block *sb, enum freeze_holder who)
 	if (sb->s_op->unfreeze_fs) {
 		error = sb->s_op->unfreeze_fs(sb);
 		if (error) {
-			printk(KERN_ERR "VFS:Filesystem thaw failed\n");
+			pr_err("VFS: Filesystem thaw failed\n");
+			freeze_inc(sb, who);
 			lockdep_sb_freeze_release(sb);
-			super_unlock_excl(sb);
-			return error;
+			goto out_unlock;
 		}
 	}
 
-	sb->s_writers.freeze_holders &= ~who;
 	sb->s_writers.frozen = SB_UNFROZEN;
 	wake_up_var(&sb->s_writers.frozen);
 	sb_freeze_unlock(sb, SB_FREEZE_FS);
-out:
+out_deactivate:
 	deactivate_locked_super(sb);
 	return 0;
+
+out_unlock:
+	super_unlock_excl(sb);
+	return error;
 }
 
 /**
@@ -2128,11 +2173,18 @@ static int thaw_super_locked(struct super_block *sb, enum freeze_holder who)
  * @who should be:
  * * %FREEZE_HOLDER_USERSPACE if userspace wants to thaw the fs;
  * * %FREEZE_HOLDER_KERNEL if the kernel wants to thaw the fs.
+ * * %FREEZE_MAY_NEST whether nesting freeze and thaw requests is allowed
+ *
+ * A filesystem may hold multiple devices and thus a filesystems may
+ * have been frozen through the block layer via multiple block devices.
+ * The filesystem remains frozen until all block devices are unfrozen.
  */
 int thaw_super(struct super_block *sb, enum freeze_holder who)
 {
-	if (!super_lock_excl(sb))
-		WARN(1, "Dying superblock while thawing!");
+	if (!super_lock_excl(sb)) {
+		WARN_ON_ONCE("Dying superblock while thawing!");
+		return -EINVAL;
+	}
 	return thaw_super_locked(sb, who);
 }
 EXPORT_SYMBOL(thaw_super);

fs/xfs/xfs_fsops.c

@@ -482,9 +482,9 @@ xfs_fs_goingdown(
 {
 	switch (inflags) {
 	case XFS_FSOP_GOING_FLAGS_DEFAULT: {
-		if (!freeze_bdev(mp->m_super->s_bdev)) {
+		if (!bdev_freeze(mp->m_super->s_bdev)) {
 			xfs_force_shutdown(mp, SHUTDOWN_FORCE_UMOUNT);
-			thaw_bdev(mp->m_super->s_bdev);
+			bdev_thaw(mp->m_super->s_bdev);
 		}
 		break;
 	}

fs/xfs/xfs_super.c

@@ -366,7 +366,8 @@ xfs_blkdev_get(
 {
 	int			error = 0;
 
-	*handlep = bdev_open_by_path(name, BLK_OPEN_READ | BLK_OPEN_WRITE,
+	*handlep = bdev_open_by_path(name,
+		BLK_OPEN_READ | BLK_OPEN_WRITE | BLK_OPEN_RESTRICT_WRITES,
 		mp->m_super, &fs_holder_ops);
 	if (IS_ERR(*handlep)) {
 		error = PTR_ERR(*handlep);
@@ -438,19 +439,13 @@ xfs_open_devices(
 	struct bdev_handle	*logdev_handle = NULL, *rtdev_handle = NULL;
 	int			error;
 
-	/*
-	 * blkdev_put() can't be called under s_umount, see the comment
-	 * in get_tree_bdev() for more details
-	 */
-	up_write(&sb->s_umount);
-
 	/*
 	 * Open real time and log devices - order is important.
 	 */
 	if (mp->m_logname) {
 		error = xfs_blkdev_get(mp, mp->m_logname, &logdev_handle);
 		if (error)
-			goto out_relock;
+			return error;
 	}
 
 	if (mp->m_rtname) {
@@ -493,10 +488,7 @@ xfs_open_devices(
 			bdev_release(logdev_handle);
 	}
 
-	error = 0;
-out_relock:
-	down_write(&sb->s_umount);
-	return error;
+	return 0;
 
  out_free_rtdev_targ:
 	if (mp->m_rtdev_targp)
@@ -509,7 +501,7 @@ xfs_open_devices(
  out_close_logdev:
 	if (logdev_handle)
 		bdev_release(logdev_handle);
-	goto out_relock;
+	return error;
 }
 
 /*
@@ -759,10 +751,6 @@ static void
 xfs_mount_free(
 	struct xfs_mount	*mp)
 {
-	/*
-	 * Free the buftargs here because blkdev_put needs to be called outside
-	 * of sb->s_umount, which is held around the call to ->put_super.
-	 */
 	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp)
 		xfs_free_buftarg(mp->m_logdev_targp);
 	if (mp->m_rtdev_targp)

include/linux/blk_types.h

@@ -57,20 +57,18 @@ struct block_device {
 	void *			bd_holder;
 	const struct blk_holder_ops *bd_holder_ops;
 	struct mutex		bd_holder_lock;
-	/* The counter of freeze processes */
-	int			bd_fsfreeze_count;
 	int			bd_holders;
 	struct kobject		*bd_holder_dir;
 
-	/* Mutex for freeze */
-	struct mutex		bd_fsfreeze_mutex;
-	struct super_block	*bd_fsfreeze_sb;
+	atomic_t		bd_fsfreeze_count; /* number of freeze requests */
+	struct mutex		bd_fsfreeze_mutex; /* serialize freeze/thaw */
 
 	struct partition_meta_info *bd_meta_info;
 #ifdef CONFIG_FAIL_MAKE_REQUEST
 	bool			bd_make_it_fail;
 #endif
 	bool			bd_ro_warned;
+	int			bd_writers;
 	/*
 	 * keep this out-of-line as it's both big and not needed in the fast
 	 * path

include/linux/blkdev.h

@@ -124,6 +124,8 @@ typedef unsigned int __bitwise blk_mode_t;
 #define BLK_OPEN_NDELAY		((__force blk_mode_t)(1 << 3))
 /* open for "writes" only for ioctls (specialy hack for floppy.c) */
 #define BLK_OPEN_WRITE_IOCTL	((__force blk_mode_t)(1 << 4))
+/* open is exclusive wrt all other BLK_OPEN_WRITE opens to the device */
+#define BLK_OPEN_RESTRICT_WRITES	((__force blk_mode_t)(1 << 5))
 
 struct gendisk {
 	/*
@@ -1468,8 +1470,23 @@ struct blk_holder_ops {
 	 * Sync the file system mounted on the block device.
 	 */
 	void (*sync)(struct block_device *bdev);
+
+	/*
+	 * Freeze the file system mounted on the block device.
+	 */
+	int (*freeze)(struct block_device *bdev);
+
+	/*
+	 * Thaw the file system mounted on the block device.
+	 */
+	int (*thaw)(struct block_device *bdev);
 };
 
+/*
+ * For filesystems using @fs_holder_ops, the @holder argument passed to
+ * helpers used to open and claim block devices via
+ * bd_prepare_to_claim() must point to a superblock.
+ */
 extern const struct blk_holder_ops fs_holder_ops;
 
 /*
@@ -1477,7 +1494,8 @@ extern const struct blk_holder_ops fs_holder_ops;
  * as stored in sb->s_flags.
  */
 #define sb_open_mode(flags) \
-	(BLK_OPEN_READ | (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE))
+	(BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \
+	 (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE))
 
 struct bdev_handle {
 	struct block_device *bdev;
@@ -1485,10 +1503,6 @@ struct bdev_handle {
 	blk_mode_t mode;
 };
 
-struct block_device *blkdev_get_by_dev(dev_t dev, blk_mode_t mode, void *holder,
-		const struct blk_holder_ops *hops);
-struct block_device *blkdev_get_by_path(const char *path, blk_mode_t mode,
-		void *holder, const struct blk_holder_ops *hops);
 struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
 		const struct blk_holder_ops *hops);
 struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode,
@@ -1496,7 +1510,6 @@ struct bdev_handle *bdev_open_by_path(const char *path, blk_mode_t mode,
 int bd_prepare_to_claim(struct block_device *bdev, void *holder,
 		const struct blk_holder_ops *hops);
 void bd_abort_claiming(struct block_device *bdev, void *holder);
-void blkdev_put(struct block_device *bdev, void *holder);
 void bdev_release(struct bdev_handle *handle);
 
 /* just for blk-cgroup, don't use elsewhere */
@@ -1541,8 +1554,8 @@ static inline int early_lookup_bdev(const char *pathname, dev_t *dev)
 }
 #endif /* CONFIG_BLOCK */
 
-int freeze_bdev(struct block_device *bdev);
-int thaw_bdev(struct block_device *bdev);
+int bdev_freeze(struct block_device *bdev);
+int bdev_thaw(struct block_device *bdev);
 
 struct io_comp_batch {
 	struct request *req_list;

include/linux/fs.h

@@ -1187,7 +1187,8 @@ enum {
 
 struct sb_writers {
 	unsigned short			frozen;		/* Is sb frozen? */
-	unsigned short			freeze_holders;	/* Who froze fs? */
+	int				freeze_kcount;	/* How many kernel freeze requests? */
+	int				freeze_ucount;	/* How many userspace freeze requests? */
 	struct percpu_rw_semaphore	rw_sem[SB_FREEZE_LEVELS];
 };
 
@@ -2053,9 +2054,24 @@ extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
 					struct file *dst_file, loff_t dst_pos,
 					loff_t len, unsigned int remap_flags);
 
+/**
+ * enum freeze_holder - holder of the freeze
+ * @FREEZE_HOLDER_KERNEL: kernel wants to freeze or thaw filesystem
+ * @FREEZE_HOLDER_USERSPACE: userspace wants to freeze or thaw filesystem
+ * @FREEZE_MAY_NEST: whether nesting freeze and thaw requests is allowed
+ *
+ * Indicate who the owner of the freeze or thaw request is and whether
+ * the freeze needs to be exclusive or can nest.
+ * Without @FREEZE_MAY_NEST, multiple freeze and thaw requests from the
+ * same holder aren't allowed. It is however allowed to hold a single
+ * @FREEZE_HOLDER_USERSPACE and a single @FREEZE_HOLDER_KERNEL freeze at
+ * the same time. This is relied upon by some filesystems during online
+ * repair or similar.
+ */
 enum freeze_holder {
 	FREEZE_HOLDER_KERNEL	= (1U << 0),
 	FREEZE_HOLDER_USERSPACE	= (1U << 1),
+	FREEZE_MAY_NEST		= (1U << 2),
 };
 
 struct super_operations {
@@ -3131,7 +3147,6 @@ extern int vfs_readlink(struct dentry *, char __user *, int);
 extern struct file_system_type *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 struct super_block *get_active_super(struct block_device *bdev);
 extern void drop_super(struct super_block *sb);
 extern void drop_super_exclusive(struct super_block *sb);
 extern void iterate_supers(void (*)(struct super_block *, void *), void *);