From f3a608827d1f8de0dd12813e8d9c6803fe64e119 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 8 Feb 2024 18:47:35 +0100 Subject: bdev: open block device as files Add two new helpers to allow opening block devices as files. This is not the final infrastructure. This still opens the block device before opening a struct a file. Until we have removed all references to struct bdev_handle we can't switch the order: * Introduce blk_to_file_flags() to translate from block specific to flags usable to pen a new file. * Introduce bdev_file_open_by_{dev,path}(). * Introduce temporary sb_bdev_handle() helper to retrieve a struct bdev_handle from a block device file and update places that directly reference struct bdev_handle to rely on it. * Don't count block device openes against the number of open files. A bdev_file_open_by_{dev,path}() file is never installed into any file descriptor table. One idea that came to mind was to use kernel_tmpfile_open() which would require us to pass a path and it would then call do_dentry_open() going through the regular fops->open::blkdev_open() path. But then we're back to the problem of routing block specific flags such as BLK_OPEN_RESTRICT_WRITES through the open path and would have to waste FMODE_* flags every time we add a new one. With this we can avoid using a flag bit and we have more leeway in how we open block devices from bdev_open_by_{dev,path}(). Link: https://lore.kernel.org/r/20240123-vfs-bdev-file-v2-1-adbd023e19cc@kernel.org Signed-off-by: Christian Brauner --- include/linux/fs.h | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'include/linux/fs.h') diff --git a/include/linux/fs.h b/include/linux/fs.h index ed5966a70495..e9291e27cc47 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1228,8 +1228,8 @@ struct super_block { #endif struct hlist_bl_head s_roots; /* alternate root dentries for NFS */ struct list_head s_mounts; /* list of mounts; _not_ for fs use */ - struct block_device *s_bdev; - struct bdev_handle *s_bdev_handle; + struct block_device *s_bdev; /* can go away once we use an accessor for @s_bdev_file */ + struct file *s_bdev_file; struct backing_dev_info *s_bdi; struct mtd_info *s_mtd; struct hlist_node s_instances; @@ -1327,6 +1327,12 @@ struct super_block { struct list_head s_inodes_wb; /* writeback inodes */ } __randomize_layout; +/* Temporary helper that will go away. */ +static inline struct bdev_handle *sb_bdev_handle(struct super_block *sb) +{ + return sb->s_bdev_file->private_data; +} + static inline struct user_namespace *i_user_ns(const struct inode *inode) { return inode->i_sb->s_user_ns; -- cgit v1.2.3 From a56aefca8d386181415a1fb7cfec2f72b0404797 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Tue, 23 Jan 2024 14:26:46 +0100 Subject: bdev: make struct bdev_handle private to the block layer Link: https://lore.kernel.org/r/20240123-vfs-bdev-file-v2-29-adbd023e19cc@kernel.org Reviewed-by: Christoph Hellwig Signed-off-by: Christian Brauner --- block/bdev.c | 119 +++++++++++++++++++++++++------------------------ block/blk.h | 12 +++-- block/fops.c | 37 +++++++-------- include/linux/blkdev.h | 7 --- include/linux/fs.h | 6 --- 5 files changed, 86 insertions(+), 95 deletions(-) (limited to 'include/linux/fs.h') diff --git a/block/bdev.c b/block/bdev.c index e6e46f24a89a..8f33f160e923 100644 --- a/block/bdev.c +++ b/block/bdev.c @@ -703,6 +703,24 @@ out_blkdev_put: return ret; } +int bdev_permission(dev_t dev, blk_mode_t mode, void *holder) +{ + int ret; + + 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 ret; + + /* Blocking writes requires exclusive opener */ + if (mode & BLK_OPEN_RESTRICT_WRITES && !holder) + return -EINVAL; + + return 0; +} + static void blkdev_put_part(struct block_device *part) { struct block_device *whole = bdev_whole(part); @@ -795,69 +813,43 @@ static void bdev_yield_write_access(struct block_device *bdev, blk_mode_t mode) } /** - * bdev_open_by_dev - open a block device by device number - * @dev: device number of block device to open + * bdev_open - open a block device + * @bdev: block device to open * @mode: open mode (BLK_OPEN_*) * @holder: exclusive holder identifier * @hops: holder operations + * @bdev_file: file for the block device * - * Open the block device described by device number @dev. If @holder is not - * %NULL, the block device is opened with exclusive access. Exclusive opens may - * nest for the same @holder. - * - * 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 bdev_open_by_path(). + * Open the block device. If @holder is not %NULL, the block device is opened + * with exclusive access. Exclusive opens may nest for the same @holder. * * CONTEXT: * Might sleep. * * RETURNS: - * Handle with a reference to the block_device on success, ERR_PTR(-errno) on - * failure. + * zero on success, -errno on failure. */ -struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, - const struct blk_holder_ops *hops) +int bdev_open(struct block_device *bdev, blk_mode_t mode, void *holder, + const struct blk_holder_ops *hops, struct file *bdev_file) { - struct bdev_handle *handle = kmalloc(sizeof(struct bdev_handle), - GFP_KERNEL); - struct block_device *bdev; + struct bdev_handle *handle; bool unblock_events = true; - struct gendisk *disk; + struct gendisk *disk = bdev->bd_disk; int ret; + handle = kmalloc(sizeof(struct bdev_handle), GFP_KERNEL); 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) - 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) { - ret = -ENXIO; - goto free_handle; - } - disk = bdev->bd_disk; + return -ENOMEM; if (holder) { mode |= BLK_OPEN_EXCL; ret = bd_prepare_to_claim(bdev, holder, hops); if (ret) - goto put_blkdev; + goto free_handle; } else { if (WARN_ON_ONCE(mode & BLK_OPEN_EXCL)) { ret = -EIO; - goto put_blkdev; + goto free_handle; } } @@ -902,7 +894,16 @@ struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, handle->bdev = bdev; handle->holder = holder; handle->mode = mode; - return handle; + + bdev_file->f_flags |= O_LARGEFILE; + bdev_file->f_mode |= FMODE_BUF_RASYNC | FMODE_CAN_ODIRECT; + if (bdev_nowait(bdev)) + bdev_file->f_mode |= FMODE_NOWAIT; + bdev_file->f_mapping = handle->bdev->bd_inode->i_mapping; + bdev_file->f_wb_err = filemap_sample_wb_err(bdev_file->f_mapping); + bdev_file->private_data = handle; + + return 0; put_module: module_put(disk->fops->owner); abort_claiming: @@ -910,11 +911,9 @@ abort_claiming: bd_abort_claiming(bdev, holder); mutex_unlock(&disk->open_mutex); disk_unblock_events(disk); -put_blkdev: - blkdev_put_no_open(bdev); free_handle: kfree(handle); - return ERR_PTR(ret); + return ret; } /* @@ -951,29 +950,33 @@ struct file *bdev_file_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, const struct blk_holder_ops *hops) { struct file *bdev_file; - struct bdev_handle *handle; + struct block_device *bdev; unsigned int flags; + int ret; - handle = bdev_open_by_dev(dev, mode, holder, hops); - if (IS_ERR(handle)) - return ERR_CAST(handle); + ret = bdev_permission(dev, mode, holder); + if (ret) + return ERR_PTR(ret); + + bdev = blkdev_get_no_open(dev); + if (!bdev) + return ERR_PTR(-ENXIO); flags = blk_to_file_flags(mode); - bdev_file = alloc_file_pseudo_noaccount(handle->bdev->bd_inode, + bdev_file = alloc_file_pseudo_noaccount(bdev->bd_inode, blockdev_mnt, "", flags | O_LARGEFILE, &def_blk_fops); if (IS_ERR(bdev_file)) { - bdev_release(handle); + blkdev_put_no_open(bdev); return bdev_file; } - ihold(handle->bdev->bd_inode); + ihold(bdev->bd_inode); - bdev_file->f_mode |= FMODE_BUF_RASYNC | FMODE_CAN_ODIRECT; - if (bdev_nowait(handle->bdev)) - bdev_file->f_mode |= FMODE_NOWAIT; - - bdev_file->f_mapping = handle->bdev->bd_inode->i_mapping; - bdev_file->f_wb_err = filemap_sample_wb_err(bdev_file->f_mapping); - bdev_file->private_data = handle; + ret = bdev_open(bdev, mode, holder, hops, bdev_file); + if (ret) { + blkdev_put_no_open(bdev); + fput(bdev_file); + return ERR_PTR(ret); + } return bdev_file; } EXPORT_SYMBOL(bdev_file_open_by_dev); diff --git a/block/blk.h b/block/blk.h index c9630774767d..19b15870284f 100644 --- a/block/blk.h +++ b/block/blk.h @@ -25,6 +25,12 @@ struct blk_flush_queue { struct request *flush_rq; }; +struct bdev_handle { + struct block_device *bdev; + void *holder; + blk_mode_t mode; +}; + bool is_flush_rq(struct request *req); struct blk_flush_queue *blk_alloc_flush_queue(int node, int cmd_size, @@ -517,7 +523,7 @@ static inline int req_ref_read(struct request *req) } void bdev_release(struct bdev_handle *handle); -struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, - const struct blk_holder_ops *hops); - +int bdev_open(struct block_device *bdev, blk_mode_t mode, void *holder, + const struct blk_holder_ops *hops, struct file *bdev_file); +int bdev_permission(dev_t dev, blk_mode_t mode, void *holder); #endif /* BLK_INTERNAL_H */ diff --git a/block/fops.c b/block/fops.c index 0cf8cf72cdfa..a1ba1a50ae77 100644 --- a/block/fops.c +++ b/block/fops.c @@ -599,36 +599,31 @@ blk_mode_t file_to_blk_mode(struct file *file) static int blkdev_open(struct inode *inode, struct file *filp) { - struct bdev_handle *handle; + struct block_device *bdev; blk_mode_t mode; - - /* - * Preserve backwards compatibility and allow large file access - * even if userspace doesn't ask for it explicitly. Some mkfs - * binary needs it. We might want to drop this workaround - * during an unstable branch. - */ - filp->f_flags |= O_LARGEFILE; - filp->f_mode |= FMODE_BUF_RASYNC | FMODE_CAN_ODIRECT; + void *holder; + int ret; mode = file_to_blk_mode(filp); - handle = bdev_open_by_dev(inode->i_rdev, mode, - mode & BLK_OPEN_EXCL ? filp : NULL, NULL); - if (IS_ERR(handle)) - return PTR_ERR(handle); + holder = mode & BLK_OPEN_EXCL ? filp : NULL; + ret = bdev_permission(inode->i_rdev, mode, holder); + if (ret) + return ret; - if (bdev_nowait(handle->bdev)) - filp->f_mode |= FMODE_NOWAIT; + bdev = blkdev_get_no_open(inode->i_rdev); + if (!bdev) + return -ENXIO; - filp->f_mapping = handle->bdev->bd_inode->i_mapping; - filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping); - filp->private_data = handle; - return 0; + ret = bdev_open(bdev, mode, holder, NULL, filp); + if (ret) + blkdev_put_no_open(bdev); + return ret; } static int blkdev_release(struct inode *inode, struct file *filp) { - bdev_release(filp->private_data); + if (filp->private_data) + bdev_release(filp->private_data); return 0; } diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 495f55587207..2f5dbde23094 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -1475,13 +1475,6 @@ extern const struct blk_holder_ops fs_holder_ops; (BLK_OPEN_READ | BLK_OPEN_RESTRICT_WRITES | \ (((flags) & SB_RDONLY) ? 0 : BLK_OPEN_WRITE)) -/* @bdev_handle will be removed soon. */ -struct bdev_handle { - struct block_device *bdev; - void *holder; - blk_mode_t mode; -}; - struct file *bdev_file_open_by_dev(dev_t dev, blk_mode_t mode, void *holder, const struct blk_holder_ops *hops); struct file *bdev_file_open_by_path(const char *path, blk_mode_t mode, diff --git a/include/linux/fs.h b/include/linux/fs.h index e9291e27cc47..6e0714d35d9b 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1327,12 +1327,6 @@ struct super_block { struct list_head s_inodes_wb; /* writeback inodes */ } __randomize_layout; -/* Temporary helper that will go away. */ -static inline struct bdev_handle *sb_bdev_handle(struct super_block *sb) -{ - return sb->s_bdev_file->private_data; -} - static inline struct user_namespace *i_user_ns(const struct inode *inode) { return inode->i_sb->s_user_ns; -- cgit v1.2.3