block: remove i_bdev

Switch the block device lookup interfaces to directly work with a dev_t
so that struct block_device references are only acquired by the
blkdev_get variants (and the blk-cgroup special case).  This means that
we now don't need an extra reference in the inode and can generally
simplify handling of struct block_device to keep the lookups contained
in the core block layer code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Coly Li <colyli@suse.de>		[bcache]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

9 files changed, 91 insertions, 208 deletions

diff --git a/fs/block_dev.c b/fs/block_dev.c
index 2b8c0586314f..6d6e4d50834c 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -883,7 +883,6 @@ static struct block_device *bdget(dev_t dev)
 		bdev->bd_dev = dev;
 		inode->i_mode = S_IFBLK;
 		inode->i_rdev = dev;
-		inode->i_bdev = bdev;
 		inode->i_data.a_ops = &def_blk_aops;
 		mapping_set_gfp_mask(&inode->i_data, GFP_USER);
 		unlock_new_inode(inode);
@@ -928,67 +927,8 @@ void bdput(struct block_device *bdev)
 {
 	iput(bdev->bd_inode);
 }
-
 EXPORT_SYMBOL(bdput);
  
-static struct block_device *bd_acquire(struct inode *inode)
-{
-	struct block_device *bdev;
-
-	spin_lock(&bdev_lock);
-	bdev = inode->i_bdev;
-	if (bdev && !inode_unhashed(bdev->bd_inode)) {
-		bdgrab(bdev);
-		spin_unlock(&bdev_lock);
-		return bdev;
-	}
-	spin_unlock(&bdev_lock);
-
-	/*
-	 * i_bdev references block device inode that was already shut down
-	 * (corresponding device got removed).  Remove the reference and look
-	 * up block device inode again just in case new device got
-	 * reestablished under the same device number.
-	 */
-	if (bdev)
-		bd_forget(inode);
-
-	bdev = bdget(inode->i_rdev);
-	if (bdev) {
-		spin_lock(&bdev_lock);
-		if (!inode->i_bdev) {
-			/*
-			 * We take an additional reference to bd_inode,
-			 * and it's released in clear_inode() of inode.
-			 * So, we can access it via ->i_mapping always
-			 * without igrab().
-			 */
-			bdgrab(bdev);
-			inode->i_bdev = bdev;
-			inode->i_mapping = bdev->bd_inode->i_mapping;
-		}
-		spin_unlock(&bdev_lock);
-	}
-	return bdev;
-}
-
-/* Call when you free inode */
-
-void bd_forget(struct inode *inode)
-{
-	struct block_device *bdev = NULL;
-
-	spin_lock(&bdev_lock);
-	if (!sb_is_blkdev_sb(inode->i_sb))
-		bdev = inode->i_bdev;
-	inode->i_bdev = NULL;
-	inode->i_mapping = &inode->i_data;
-	spin_unlock(&bdev_lock);
-
-	if (bdev)
-		bdput(bdev);
-}
-
 /**
  * bd_may_claim - test whether a block device can be claimed
  * @bdev: block device of interest
@@ -1497,38 +1437,45 @@ static int __blkdev_get(struct block_device *bdev, struct gendisk *disk,
 }
 
 /**
- * blkdev_get - open a block device
- * @bdev: block_device to open
+ * blkdev_get_by_dev - open a block device by device number
+ * @dev: device number of block device to open
  * @mode: FMODE_* mask
  * @holder: exclusive holder identifier
  *
- * Open @bdev with @mode.  If @mode includes %FMODE_EXCL, @bdev is
- * open with exclusive access.  Specifying %FMODE_EXCL with %NULL
- * @holder is invalid.  Exclusive opens may nest for the same @holder.
+ * Open the block device described by device number @dev. If @mode includes
+ * %FMODE_EXCL, the block device is opened with exclusive access.  Specifying
+ * %FMODE_EXCL with a %NULL @holder is invalid.  Exclusive opens may nest for
+ * the same @holder.
  *
- * On success, the reference count of @bdev is unchanged.  On failure,
- * @bdev is put.
+ * 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().
  *
  * CONTEXT:
  * Might sleep.
  *
  * RETURNS:
- * 0 on success, -errno on failure.
+ * Reference to the block_device on success, ERR_PTR(-errno) on failure.
  */
-static int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder)
+struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder)
 {
 	struct block_device *claiming;
 	bool unblock_events = true;
+	struct block_device *bdev;
 	struct gendisk *disk;
 	int partno;
 	int ret;
 
 	ret = devcgroup_check_permission(DEVCG_DEV_BLOCK,
-			imajor(bdev->bd_inode), iminor(bdev->bd_inode),
+			MAJOR(dev), MINOR(dev),
 			((mode & FMODE_READ) ? DEVCG_ACC_READ : 0) |
 			((mode & FMODE_WRITE) ? DEVCG_ACC_WRITE : 0));
 	if (ret)
-		goto bdput;
+		return ERR_PTR(ret);
+
+	bdev = bdget(dev);
+	if (!bdev)
+		return ERR_PTR(-ENOMEM);
 
 	/*
 	 * If we lost a race with 'disk' being deleted, try again.  See md.c.
@@ -1589,10 +1536,13 @@ put_disk:
 	if (ret == -ERESTARTSYS)
 		goto retry;
 bdput:
-	if (ret)
+	if (ret) {
 		bdput(bdev);
-	return ret;
+		return ERR_PTR(ret);
+	}
+	return bdev;
 }
+EXPORT_SYMBOL(blkdev_get_by_dev);
 
 /**
  * blkdev_get_by_path - open a block device by name
@@ -1600,32 +1550,30 @@ bdput:
  * @mode: FMODE_* mask
  * @holder: exclusive holder identifier
  *
- * Open the blockdevice described by the device file at @path.  @mode
- * and @holder are identical to blkdev_get().
- *
- * On success, the returned block_device has reference count of one.
+ * Open the block device described by the device file at @path.  If @mode
+ * includes %FMODE_EXCL, the block device is opened with exclusive access.
+ * Specifying %FMODE_EXCL with a %NULL @holder is invalid.  Exclusive opens may
+ * nest for the same @holder.
  *
  * CONTEXT:
  * Might sleep.
  *
  * RETURNS:
- * Pointer to block_device on success, ERR_PTR(-errno) on failure.
+ * Reference to the block_device on success, ERR_PTR(-errno) on failure.
  */
 struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
 					void *holder)
 {
 	struct block_device *bdev;
-	int err;
-
-	bdev = lookup_bdev(path);
-	if (IS_ERR(bdev))
-		return bdev;
+	dev_t dev;
+	int error;
 
-	err = blkdev_get(bdev, mode, holder);
-	if (err)
-		return ERR_PTR(err);
+	error = lookup_bdev(path, &dev);
+	if (error)
+		return ERR_PTR(error);
 
-	if ((mode & FMODE_WRITE) && bdev_read_only(bdev)) {
+	bdev = blkdev_get_by_dev(dev, mode, holder);
+	if (!IS_ERR(bdev) && (mode & FMODE_WRITE) && bdev_read_only(bdev)) {
 		blkdev_put(bdev, mode);
 		return ERR_PTR(-EACCES);
 	}
@@ -1634,45 +1582,6 @@ struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
 }
 EXPORT_SYMBOL(blkdev_get_by_path);
 
-/**
- * blkdev_get_by_dev - open a block device by device number
- * @dev: device number of block device to open
- * @mode: FMODE_* mask
- * @holder: exclusive holder identifier
- *
- * Open the blockdevice described by device number @dev.  @mode and
- * @holder are identical to blkdev_get().
- *
- * Use it 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.  _Never_ to be used for internal purposes.  If you
- * ever need it - reconsider your API.
- *
- * On success, the returned block_device has reference count of one.
- *
- * CONTEXT:
- * Might sleep.
- *
- * RETURNS:
- * Pointer to block_device on success, ERR_PTR(-errno) on failure.
- */
-struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder)
-{
-	struct block_device *bdev;
-	int err;
-
-	bdev = bdget(dev);
-	if (!bdev)
-		return ERR_PTR(-ENOMEM);
-
-	err = blkdev_get(bdev, mode, holder);
-	if (err)
-		return ERR_PTR(err);
-
-	return bdev;
-}
-EXPORT_SYMBOL(blkdev_get_by_dev);
-
 static int blkdev_open(struct inode * inode, struct file * filp)
 {
 	struct block_device *bdev;
@@ -1694,14 +1603,12 @@ static int blkdev_open(struct inode * inode, struct file * filp)
 	if ((filp->f_flags & O_ACCMODE) == 3)
 		filp->f_mode |= FMODE_WRITE_IOCTL;
 
-	bdev = bd_acquire(inode);
-	if (bdev == NULL)
-		return -ENOMEM;
-
+	bdev = blkdev_get_by_dev(inode->i_rdev, filp->f_mode, filp);
+	if (IS_ERR(bdev))
+		return PTR_ERR(bdev);
 	filp->f_mapping = bdev->bd_inode->i_mapping;
 	filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping);
-
-	return blkdev_get(bdev, filp->f_mode, filp);
+	return 0;
 }
 
 static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part)
@@ -2010,37 +1917,32 @@ const struct file_operations def_blk_fops = {
  * namespace if possible and return it.  Return ERR_PTR(error)
  * otherwise.
  */
-struct block_device *lookup_bdev(const char *pathname)
+int lookup_bdev(const char *pathname, dev_t *dev)
 {
-	struct block_device *bdev;
 	struct inode *inode;
 	struct path path;
 	int error;
 
 	if (!pathname || !*pathname)
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	error = kern_path(pathname, LOOKUP_FOLLOW, &path);
 	if (error)
-		return ERR_PTR(error);
+		return error;
 
 	inode = d_backing_inode(path.dentry);
 	error = -ENOTBLK;
 	if (!S_ISBLK(inode->i_mode))
-		goto fail;
+		goto out_path_put;
 	error = -EACCES;
 	if (!may_open_dev(&path))
-		goto fail;
-	error = -ENOMEM;
-	bdev = bd_acquire(inode);
-	if (!bdev)
-		goto fail;
-out:
+		goto out_path_put;
+
+	*dev = inode->i_rdev;
+	error = 0;
+out_path_put:
 	path_put(&path);
-	return bdev;
-fail:
-	bdev = ERR_PTR(error);
-	goto out;
+	return error;
 }
 EXPORT_SYMBOL(lookup_bdev);
 
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index a6406b3b8c2b..fbc4b58228f7 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -929,16 +929,16 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 		 * make sure it's the same device if the device is mounted
 		 */
 		if (device->bdev) {
-			struct block_device *path_bdev;
+			int error;
+			dev_t path_dev;
 
-			path_bdev = lookup_bdev(path);
-			if (IS_ERR(path_bdev)) {
+			error = lookup_bdev(path, &path_dev);
+			if (error) {
 				mutex_unlock(&fs_devices->device_list_mutex);
-				return ERR_CAST(path_bdev);
+				return ERR_PTR(error);
 			}
 
-			if (device->bdev != path_bdev) {
-				bdput(path_bdev);
+			if (device->bdev->bd_dev != path_dev) {
 				mutex_unlock(&fs_devices->device_list_mutex);
 				btrfs_warn_in_rcu(device->fs_info,
 	"duplicate device %s devid %llu generation %llu scanned by %s (%d)",
@@ -947,7 +947,6 @@ static noinline struct btrfs_device *device_list_add(const char *path,
 						  task_pid_nr(current));
 				return ERR_PTR(-EEXIST);
 			}
-			bdput(path_bdev);
 			btrfs_info_in_rcu(device->fs_info,
 	"devid %llu device path %s changed to %s scanned by %s (%d)",
 					  devid, rcu_str_deref(device->name),
diff --git a/fs/inode.c b/fs/inode.c
index 9d78c37b00b8..cb008acf0efd 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -155,7 +155,6 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
 	inode->i_bytes = 0;
 	inode->i_generation = 0;
 	inode->i_pipe = NULL;
-	inode->i_bdev = NULL;
 	inode->i_cdev = NULL;
 	inode->i_link = NULL;
 	inode->i_dir_seq = 0;
@@ -580,8 +579,6 @@ static void evict(struct inode *inode)
 		truncate_inode_pages_final(&inode->i_data);
 		clear_inode(inode);
 	}
-	if (S_ISBLK(inode->i_mode) && inode->i_bdev)
-		bd_forget(inode);
 	if (S_ISCHR(inode->i_mode) && inode->i_cdev)
 		cd_forget(inode);
 
diff --git a/fs/internal.h b/fs/internal.h
index 47be21dfeebe..53f890446e75 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -25,7 +25,6 @@ extern void __init bdev_cache_init(void);
 extern int __sync_blockdev(struct block_device *bdev, int wait);
 void iterate_bdevs(void (*)(struct block_device *, void *), void *);
 void emergency_thaw_bdev(struct super_block *sb);
-void bd_forget(struct inode *inode);
 #else
 static inline void bdev_cache_init(void)
 {
@@ -43,9 +42,6 @@ static inline int emergency_thaw_bdev(struct super_block *sb)
 {
 	return 0;
 }
-static inline void bd_forget(struct inode *inode)
-{
-}
 #endif /* CONFIG_BLOCK */
 
 /*
@@ -114,8 +110,7 @@ extern struct file *alloc_empty_file_noaccount(int, const struct cred *);
  */
 extern int reconfigure_super(struct fs_context *);
 extern bool trylock_super(struct super_block *sb);
-struct super_block *__get_super(struct block_device *bdev, bool excl);
-extern struct super_block *user_get_super(dev_t);
+struct super_block *user_get_super(dev_t, bool excl);
 void put_super(struct super_block *sb);
 extern bool mount_capable(struct fs_context *);
 
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 4ead291b2976..8f13c0417f94 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2716,11 +2716,7 @@ static struct file *__io_file_get(struct io_submit_state *state, int fd)
 
 static bool io_bdev_nowait(struct block_device *bdev)
 {
-#ifdef CONFIG_BLOCK
 	return !bdev || blk_queue_nowait(bdev_get_queue(bdev));
-#else
-	return true;
-#endif
 }
 
 /*
@@ -2733,14 +2729,16 @@ static bool io_file_supports_async(struct file *file, int rw)
 	umode_t mode = file_inode(file)->i_mode;
 
 	if (S_ISBLK(mode)) {
-		if (io_bdev_nowait(file->f_inode->i_bdev))
+		if (IS_ENABLED(CONFIG_BLOCK) &&
+		    io_bdev_nowait(I_BDEV(file->f_mapping->host)))
 			return true;
 		return false;
 	}
 	if (S_ISCHR(mode) || S_ISSOCK(mode))
 		return true;
 	if (S_ISREG(mode)) {
-		if (io_bdev_nowait(file->f_inode->i_sb->s_bdev) &&
+		if (IS_ENABLED(CONFIG_BLOCK) &&
+		    io_bdev_nowait(file->f_inode->i_sb->s_bdev) &&
 		    file->f_op != &io_uring_fops)
 			return true;
 		return false;
diff --git a/fs/pipe.c b/fs/pipe.c
index 0ac197658a2d..c5989cfd564d 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -1342,9 +1342,8 @@ out_revert_acct:
 }
 
 /*
- * After the inode slimming patch, i_pipe/i_bdev/i_cdev share the same
- * location, so checking ->i_pipe is not enough to verify that this is a
- * pipe.
+ * Note that i_pipe and i_cdev share the same location, so checking ->i_pipe is
+ * not enough to verify that this is a pipe.
  */
 struct pipe_inode_info *get_pipe_info(struct file *file, bool for_splice)
 {
diff --git a/fs/quota/quota.c b/fs/quota/quota.c
index f3d32b0d9008..6d16b2be5ac4 100644
--- a/fs/quota/quota.c
+++ b/fs/quota/quota.c
@@ -866,17 +866,18 @@ static bool quotactl_cmd_onoff(int cmd)
 static struct super_block *quotactl_block(const char __user *special, int cmd)
 {
 #ifdef CONFIG_BLOCK
-	struct block_device *bdev;
 	struct super_block *sb;
 	struct filename *tmp = getname(special);
 	bool excl = false, thawed = false;
+	int error;
+	dev_t dev;
 
 	if (IS_ERR(tmp))
 		return ERR_CAST(tmp);
-	bdev = lookup_bdev(tmp->name);
+	error = lookup_bdev(tmp->name, &dev);
 	putname(tmp);
-	if (IS_ERR(bdev))
-		return ERR_CAST(bdev);
+	if (error)
+		return ERR_PTR(error);
 
 	if (quotactl_cmd_onoff(cmd)) {
 		excl = true;
@@ -886,8 +887,10 @@ static struct super_block *quotactl_block(const char __user *special, int cmd)
 	}
 
 retry:
-	sb = __get_super(bdev, excl);
-	if (thawed && sb && sb->s_writers.frozen != SB_UNFROZEN) {
+	sb = user_get_super(dev, excl);
+	if (!sb)
+		return ERR_PTR(-ENODEV);
+	if (thawed && sb->s_writers.frozen != SB_UNFROZEN) {
 		if (excl)
 			up_write(&sb->s_umount);
 		else
@@ -897,10 +900,6 @@ retry:
 		put_super(sb);
 		goto retry;
 	}
-
-	bdput(bdev);
-	if (!sb)
-		return ERR_PTR(-ENODEV);
 	return sb;
 
 #else
diff --git a/fs/statfs.c b/fs/statfs.c
index 59f33752c131..68cb07788750 100644
--- a/fs/statfs.c
+++ b/fs/statfs.c
@@ -235,7 +235,7 @@ SYSCALL_DEFINE3(fstatfs64, unsigned int, fd, size_t, sz, struct statfs64 __user
 
 static int vfs_ustat(dev_t dev, struct kstatfs *sbuf)
 {
-	struct super_block *s = user_get_super(dev);
+	struct super_block *s = user_get_super(dev, false);
 	int err;
 	if (!s)
 		return -EINVAL;
diff --git a/fs/super.c b/fs/super.c
index 343e5c1e538d..2c6cdea2ab2d 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -740,7 +740,14 @@ void iterate_supers_type(struct file_system_type *type,
 
 EXPORT_SYMBOL(iterate_supers_type);
 
-struct super_block *__get_super(struct block_device *bdev, bool excl)
+/**
+ * get_super - get 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. %NULL is returned if no match is found.
+ */
+struct super_block *get_super(struct block_device *bdev)
 {
 	struct super_block *sb;
 
@@ -755,17 +762,11 @@ rescan:
 		if (sb->s_bdev == bdev) {
 			sb->s_count++;
 			spin_unlock(&sb_lock);
-			if (!excl)
-				down_read(&sb->s_umount);
-			else
-				down_write(&sb->s_umount);
+			down_read(&sb->s_umount);
 			/* still alive? */
 			if (sb->s_root && (sb->s_flags & SB_BORN))
 				return sb;
-			if (!excl)
-				up_read(&sb->s_umount);
-			else
-				up_write(&sb->s_umount);
+			up_read(&sb->s_umount);
 			/* nope, got unmounted */
 			spin_lock(&sb_lock);
 			__put_super(sb);
@@ -777,19 +778,6 @@ rescan:
 }
 
 /**
- *	get_super - get 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. %NULL is returned if no match is found.
- */
-struct super_block *get_super(struct block_device *bdev)
-{
-	return __get_super(bdev, false);
-}
-EXPORT_SYMBOL(get_super);
-
-/**
  * get_active_super - get an active reference to the superblock of a device
  * @bdev: device to get the superblock for
  *
@@ -820,7 +808,7 @@ restart:
 	return NULL;
 }
 
-struct super_block *user_get_super(dev_t dev)
+struct super_block *user_get_super(dev_t dev, bool excl)
 {
 	struct super_block *sb;
 
@@ -832,11 +820,17 @@ rescan:
 		if (sb->s_dev ==  dev) {
 			sb->s_count++;
 			spin_unlock(&sb_lock);
-			down_read(&sb->s_umount);
+			if (excl)
+				down_write(&sb->s_umount);
+			else
+				down_read(&sb->s_umount);
 			/* still alive? */
 			if (sb->s_root && (sb->s_flags & SB_BORN))
 				return sb;
-			up_read(&sb->s_umount);
+			if (excl)
+				up_write(&sb->s_umount);
+			else
+				up_read(&sb->s_umount);
 			/* nope, got unmounted */
 			spin_lock(&sb_lock);
 			__put_super(sb);