diff options
Diffstat (limited to 'fs/btrfs/volumes.c')
| -rw-r--r-- | fs/btrfs/volumes.c | 5874 |
1 files changed, 3217 insertions, 2657 deletions
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 0d6e785bcb98..ae1742a35e76 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -4,39 +4,58 @@ */ #include <linux/sched.h> -#include <linux/bio.h> +#include <linux/sched/mm.h> #include <linux/slab.h> -#include <linux/blkdev.h> #include <linux/ratelimit.h> #include <linux/kthread.h> -#include <linux/raid/pq.h> #include <linux/semaphore.h> #include <linux/uuid.h> #include <linux/list_sort.h> +#include <linux/namei.h> #include "misc.h" -#include "ctree.h" -#include "extent_map.h" #include "disk-io.h" +#include "extent-tree.h" #include "transaction.h" -#include "print-tree.h" #include "volumes.h" #include "raid56.h" -#include "async-thread.h" -#include "check-integrity.h" -#include "rcu-string.h" #include "dev-replace.h" #include "sysfs.h" #include "tree-checker.h" #include "space-info.h" #include "block-group.h" #include "discard.h" +#include "zoned.h" +#include "fs.h" +#include "accessors.h" +#include "uuid-tree.h" +#include "ioctl.h" +#include "relocation.h" +#include "scrub.h" +#include "super.h" +#include "raid-stripe-tree.h" + +#define BTRFS_BLOCK_GROUP_STRIPE_MASK (BTRFS_BLOCK_GROUP_RAID0 | \ + BTRFS_BLOCK_GROUP_RAID10 | \ + BTRFS_BLOCK_GROUP_RAID56_MASK) + +struct btrfs_io_geometry { + u32 stripe_index; + u32 stripe_nr; + int mirror_num; + int num_stripes; + u64 stripe_offset; + u64 raid56_full_stripe_start; + int max_errors; + enum btrfs_map_op op; + bool use_rst; +}; const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = { [BTRFS_RAID_RAID10] = { .sub_stripes = 2, .dev_stripes = 1, .devs_max = 0, /* 0 == as many as possible */ - .devs_min = 4, + .devs_min = 2, .tolerated_failures = 1, .devs_increment = 2, .ncopies = 2, @@ -101,7 +120,7 @@ const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = { .sub_stripes = 1, .dev_stripes = 1, .devs_max = 0, - .devs_min = 2, + .devs_min = 1, .tolerated_failures = 0, .devs_increment = 1, .ncopies = 1, @@ -151,6 +170,20 @@ const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = { }, }; +/* + * Convert block group flags (BTRFS_BLOCK_GROUP_*) to btrfs_raid_types, which + * can be used as index to access btrfs_raid_array[]. + */ +enum btrfs_raid_types __attribute_const__ btrfs_bg_flags_to_raid_index(u64 flags) +{ + const u64 profile = (flags & BTRFS_BLOCK_GROUP_PROFILE_MASK); + + if (!profile) + return BTRFS_RAID_SINGLE; + + return BTRFS_BG_FLAG_TO_INDEX(profile); +} + const char *btrfs_bg_type_to_raid_name(u64 flags) { const int index = btrfs_bg_flags_to_raid_index(flags); @@ -161,6 +194,13 @@ const char *btrfs_bg_type_to_raid_name(u64 flags) return btrfs_raid_array[index].raid_name; } +int btrfs_nr_parity_stripes(u64 type) +{ + enum btrfs_raid_types index = btrfs_bg_flags_to_raid_index(type); + + return btrfs_raid_array[index].nparity; +} + /* * Fill @buf with textual description of @bg_flags, no more than @size_buf * bytes including terminating null byte. @@ -173,10 +213,8 @@ void btrfs_describe_block_groups(u64 bg_flags, char *buf, u32 size_buf) u64 flags = bg_flags; u32 size_bp = size_buf; - if (!flags) { - strcpy(bp, "NONE"); + if (!flags) return; - } #define DESCRIBE_FLAG(flag, desc) \ do { \ @@ -217,13 +255,7 @@ out_overflow:; static int init_first_rw_device(struct btrfs_trans_handle *trans); static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info); -static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev); static void btrfs_dev_stat_print_on_load(struct btrfs_device *device); -static int __btrfs_map_block(struct btrfs_fs_info *fs_info, - enum btrfs_map_op op, - u64 logical, u64 *length, - struct btrfs_bio **bbio_ret, - int mirror_num, int need_raid_map); /* * Device locking @@ -245,7 +277,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, * * global::fs_devs - add, remove, updates to the global list * - * does not protect: manipulation of the fs_devices::devices list! + * does not protect: manipulation of the fs_devices::devices list in general + * but in mount context it could be used to exclude list modifications by eg. + * scan ioctl * * btrfs_device::name - renames (write side), read is RCU * @@ -258,6 +292,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, * may be used to exclude some operations from running concurrently without any * modifications to the list (see write_all_supers) * + * Is not required at mount and close times, because our device list is + * protected by the uuid_mutex at that point. + * * balance_mutex * ------------- * protects balance structures (status, state) and context accessed from @@ -285,8 +322,8 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, * balance_mutex * * - * Exclusive operations, BTRFS_FS_EXCL_OP - * ====================================== + * Exclusive operations + * ==================== * * Maintains the exclusivity of the following operations that apply to the * whole filesystem and cannot run in parallel. @@ -312,11 +349,11 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, * - system power-cycle and filesystem mounted as read-only * - filesystem or device errors leading to forced read-only * - * BTRFS_FS_EXCL_OP flag is set and cleared using atomic operations. - * During the course of Paused state, the BTRFS_FS_EXCL_OP remains set. + * The status of exclusive operation is set and cleared atomically. + * During the course of Paused state, fs_info::exclusive_operation remains set. * A device operation in Paused or Running state can be canceled or resumed * either by ioctl (Balance only) or when remounted as read-write. - * BTRFS_FS_EXCL_OP flag is cleared when the device operation is canceled or + * The exclusive status is cleared when the device operation is canceled or * completed. */ @@ -328,16 +365,16 @@ struct list_head * __attribute_const__ btrfs_get_fs_uuids(void) } /* - * alloc_fs_devices - allocate struct btrfs_fs_devices - * @fsid: if not NULL, copy the UUID to fs_devices::fsid - * @metadata_fsid: if not NULL, copy the UUID to fs_devices::metadata_fsid + * Allocate new btrfs_fs_devices structure identified by a fsid. + * + * @fsid: if not NULL, copy the UUID to fs_devices::fsid and to + * fs_devices::metadata_fsid * * Return a pointer to a new struct btrfs_fs_devices on success, or ERR_PTR(). * The returned struct is not linked onto any lists and can be destroyed with * kfree() right away. */ -static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid, - const u8 *metadata_fsid) +static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid) { struct btrfs_fs_devices *fs_devs; @@ -350,33 +387,38 @@ static struct btrfs_fs_devices *alloc_fs_devices(const u8 *fsid, INIT_LIST_HEAD(&fs_devs->devices); INIT_LIST_HEAD(&fs_devs->alloc_list); INIT_LIST_HEAD(&fs_devs->fs_list); - if (fsid) - memcpy(fs_devs->fsid, fsid, BTRFS_FSID_SIZE); + INIT_LIST_HEAD(&fs_devs->seed_list); - if (metadata_fsid) - memcpy(fs_devs->metadata_uuid, metadata_fsid, BTRFS_FSID_SIZE); - else if (fsid) + if (fsid) { + memcpy(fs_devs->fsid, fsid, BTRFS_FSID_SIZE); memcpy(fs_devs->metadata_uuid, fsid, BTRFS_FSID_SIZE); + } return fs_devs; } -void btrfs_free_device(struct btrfs_device *device) +static void btrfs_free_device(struct btrfs_device *device) { WARN_ON(!list_empty(&device->post_commit_list)); - rcu_string_free(device->name); - extent_io_tree_release(&device->alloc_state); - bio_put(device->flush_bio); + /* + * No need to call kfree_rcu() nor do RCU lock/unlock, nothing is + * reading the device name. + */ + kfree(rcu_dereference_raw(device->name)); + btrfs_extent_io_tree_release(&device->alloc_state); + btrfs_destroy_dev_zone_info(device); kfree(device); } static void free_fs_devices(struct btrfs_fs_devices *fs_devices) { struct btrfs_device *device; + WARN_ON(fs_devices->opened); + WARN_ON(fs_devices->holding); while (!list_empty(&fs_devices->devices)) { - device = list_entry(fs_devices->devices.next, - struct btrfs_device, dev_list); + device = list_first_entry(&fs_devices->devices, + struct btrfs_device, dev_list); list_del(&device->dev_list); btrfs_free_device(device); } @@ -388,48 +430,26 @@ void __exit btrfs_cleanup_fs_uuids(void) struct btrfs_fs_devices *fs_devices; while (!list_empty(&fs_uuids)) { - fs_devices = list_entry(fs_uuids.next, - struct btrfs_fs_devices, fs_list); + fs_devices = list_first_entry(&fs_uuids, struct btrfs_fs_devices, + fs_list); list_del(&fs_devices->fs_list); free_fs_devices(fs_devices); } } -/* - * Returns a pointer to a new btrfs_device on success; ERR_PTR() on error. - * Returned struct is not linked onto any lists and must be destroyed using - * btrfs_free_device. - */ -static struct btrfs_device *__alloc_device(void) +static bool match_fsid_fs_devices(const struct btrfs_fs_devices *fs_devices, + const u8 *fsid, const u8 *metadata_fsid) { - struct btrfs_device *dev; - - dev = kzalloc(sizeof(*dev), GFP_KERNEL); - if (!dev) - return ERR_PTR(-ENOMEM); - - /* - * Preallocate a bio that's always going to be used for flushing device - * barriers and matches the device lifespan - */ - dev->flush_bio = bio_alloc_bioset(GFP_KERNEL, 0, NULL); - if (!dev->flush_bio) { - kfree(dev); - return ERR_PTR(-ENOMEM); - } + if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) != 0) + return false; - INIT_LIST_HEAD(&dev->dev_list); - INIT_LIST_HEAD(&dev->dev_alloc_list); - INIT_LIST_HEAD(&dev->post_commit_list); + if (!metadata_fsid) + return true; - atomic_set(&dev->reada_in_flight, 0); - atomic_set(&dev->dev_stats_ccnt, 0); - btrfs_device_data_ordered_init(dev); - INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM); - INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM); - extent_io_tree_init(NULL, &dev->alloc_state, 0, NULL); + if (memcmp(metadata_fsid, fs_devices->metadata_uuid, BTRFS_FSID_SIZE) != 0) + return false; - return dev; + return true; } static noinline struct btrfs_fs_devices *find_fsid( @@ -441,128 +461,79 @@ static noinline struct btrfs_fs_devices *find_fsid( /* Handle non-split brain cases */ list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - if (metadata_fsid) { - if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0 - && memcmp(metadata_fsid, fs_devices->metadata_uuid, - BTRFS_FSID_SIZE) == 0) - return fs_devices; - } else { - if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0) - return fs_devices; - } - } - return NULL; -} - -static struct btrfs_fs_devices *find_fsid_with_metadata_uuid( - struct btrfs_super_block *disk_super) -{ - - struct btrfs_fs_devices *fs_devices; - - /* - * Handle scanned device having completed its fsid change but - * belonging to a fs_devices that was created by first scanning - * a device which didn't have its fsid/metadata_uuid changed - * at all and the CHANGING_FSID_V2 flag set. - */ - list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - if (fs_devices->fsid_change && - memcmp(disk_super->metadata_uuid, fs_devices->fsid, - BTRFS_FSID_SIZE) == 0 && - memcmp(fs_devices->fsid, fs_devices->metadata_uuid, - BTRFS_FSID_SIZE) == 0) { - return fs_devices; - } - } - /* - * Handle scanned device having completed its fsid change but - * belonging to a fs_devices that was created by a device that - * has an outdated pair of fsid/metadata_uuid and - * CHANGING_FSID_V2 flag set. - */ - list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - if (fs_devices->fsid_change && - memcmp(fs_devices->metadata_uuid, - fs_devices->fsid, BTRFS_FSID_SIZE) != 0 && - memcmp(disk_super->metadata_uuid, fs_devices->metadata_uuid, - BTRFS_FSID_SIZE) == 0) { + if (match_fsid_fs_devices(fs_devices, fsid, metadata_fsid)) return fs_devices; - } } - - return find_fsid(disk_super->fsid, disk_super->metadata_uuid); + return NULL; } - static int -btrfs_get_bdev_and_sb(const char *device_path, fmode_t flags, void *holder, - int flush, struct block_device **bdev, +btrfs_get_bdev_and_sb(const char *device_path, blk_mode_t flags, void *holder, + int flush, struct file **bdev_file, struct btrfs_super_block **disk_super) { + struct block_device *bdev; int ret; - *bdev = blkdev_get_by_path(device_path, flags, holder); + *bdev_file = bdev_file_open_by_path(device_path, flags, holder, &fs_holder_ops); - if (IS_ERR(*bdev)) { - ret = PTR_ERR(*bdev); + if (IS_ERR(*bdev_file)) { + ret = PTR_ERR(*bdev_file); + btrfs_err(NULL, "failed to open device for path %s with flags 0x%x: %d", + device_path, flags, ret); goto error; } + bdev = file_bdev(*bdev_file); if (flush) - filemap_write_and_wait((*bdev)->bd_inode->i_mapping); - ret = set_blocksize(*bdev, BTRFS_BDEV_BLOCKSIZE); - if (ret) { - blkdev_put(*bdev, flags); - goto error; + sync_blockdev(bdev); + if (holder) { + ret = set_blocksize(*bdev_file, BTRFS_BDEV_BLOCKSIZE); + if (ret) { + bdev_fput(*bdev_file); + goto error; + } } - invalidate_bdev(*bdev); - *disk_super = btrfs_read_dev_super(*bdev); + invalidate_bdev(bdev); + *disk_super = btrfs_read_disk_super(bdev, 0, false); if (IS_ERR(*disk_super)) { ret = PTR_ERR(*disk_super); - blkdev_put(*bdev, flags); + bdev_fput(*bdev_file); goto error; } return 0; error: - *bdev = NULL; + *disk_super = NULL; + *bdev_file = NULL; return ret; } -static bool device_path_matched(const char *path, struct btrfs_device *device) -{ - int found; - - rcu_read_lock(); - found = strcmp(rcu_str_deref(device->name), path); - rcu_read_unlock(); - - return found == 0; -} - /* - * Search and remove all stale (devices which are not mounted) devices. - * When both inputs are NULL, it will search and release all stale devices. - * path: Optional. When provided will it release all unmounted devices - * matching this path only. - * skip_dev: Optional. Will skip this device when searching for the stale - * devices. - * Return: 0 for success or if @path is NULL. - * -EBUSY if @path is a mounted device. - * -ENOENT if @path does not match any device in the list. + * Search and remove all stale devices (which are not mounted). When both + * inputs are NULL, it will search and release all stale devices. + * + * @devt: Optional. When provided will it release all unmounted devices + * matching this devt only. + * @skip_device: Optional. Will skip this device when searching for the stale + * devices. + * + * Return: 0 for success or if @devt is 0. + * -EBUSY if @devt is a mounted device. + * -ENOENT if @devt does not match any device in the list. */ -static int btrfs_free_stale_devices(const char *path, - struct btrfs_device *skip_device) +static int btrfs_free_stale_devices(dev_t devt, struct btrfs_device *skip_device) { struct btrfs_fs_devices *fs_devices, *tmp_fs_devices; struct btrfs_device *device, *tmp_device; - int ret = 0; + int ret; + bool freed = false; - if (path) - ret = -ENOENT; + lockdep_assert_held(&uuid_mutex); + /* Return good status if there is no instance of devt. */ + ret = 0; list_for_each_entry_safe(fs_devices, tmp_fs_devices, &fs_uuids, fs_list) { mutex_lock(&fs_devices->device_list_mutex); @@ -570,13 +541,10 @@ static int btrfs_free_stale_devices(const char *path, &fs_devices->devices, dev_list) { if (skip_device && skip_device == device) continue; - if (path && !device->name) - continue; - if (path && !device_path_matched(path, device)) + if (devt && devt != device->devt) continue; - if (fs_devices->opened) { - /* for an already deleted device return 0 */ - if (path && ret != 0) + if (fs_devices->opened || fs_devices->holding) { + if (devt) ret = -EBUSY; break; } @@ -586,9 +554,7 @@ static int btrfs_free_stale_devices(const char *path, list_del(&device->dev_list); btrfs_free_device(device); - ret = 0; - if (fs_devices->num_devices == 0) - break; + freed = true; } mutex_unlock(&fs_devices->device_list_mutex); @@ -599,15 +565,91 @@ static int btrfs_free_stale_devices(const char *path, } } + /* If there is at least one freed device return 0. */ + if (freed) + return 0; + return ret; } +static struct btrfs_fs_devices *find_fsid_by_device( + struct btrfs_super_block *disk_super, + dev_t devt, bool *same_fsid_diff_dev) +{ + struct btrfs_fs_devices *fsid_fs_devices; + struct btrfs_fs_devices *devt_fs_devices; + const bool has_metadata_uuid = (btrfs_super_incompat_flags(disk_super) & + BTRFS_FEATURE_INCOMPAT_METADATA_UUID); + bool found_by_devt = false; + + /* Find the fs_device by the usual method, if found use it. */ + fsid_fs_devices = find_fsid(disk_super->fsid, + has_metadata_uuid ? disk_super->metadata_uuid : NULL); + + /* The temp_fsid feature is supported only with single device filesystem. */ + if (btrfs_super_num_devices(disk_super) != 1) + return fsid_fs_devices; + + /* + * A seed device is an integral component of the sprout device, which + * functions as a multi-device filesystem. So, temp-fsid feature is + * not supported. + */ + if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) + return fsid_fs_devices; + + /* Try to find a fs_devices by matching devt. */ + list_for_each_entry(devt_fs_devices, &fs_uuids, fs_list) { + struct btrfs_device *device; + + list_for_each_entry(device, &devt_fs_devices->devices, dev_list) { + if (device->devt == devt) { + found_by_devt = true; + break; + } + } + if (found_by_devt) + break; + } + + if (found_by_devt) { + /* Existing device. */ + if (fsid_fs_devices == NULL) { + if (devt_fs_devices->opened == 0) { + /* Stale device. */ + return NULL; + } else { + /* temp_fsid is mounting a subvol. */ + return devt_fs_devices; + } + } else { + /* Regular or temp_fsid device mounting a subvol. */ + return devt_fs_devices; + } + } else { + /* New device. */ + if (fsid_fs_devices == NULL) { + return NULL; + } else { + /* sb::fsid is already used create a new temp_fsid. */ + *same_fsid_diff_dev = true; + return NULL; + } + } + + /* Not reached. */ +} + +/* + * This is only used on mount, and we are protected from competing things + * messing with our fs_devices by the uuid_mutex, thus we do not need the + * fs_devices->device_list_mutex here. + */ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices, - struct btrfs_device *device, fmode_t flags, + struct btrfs_device *device, blk_mode_t flags, void *holder) { - struct request_queue *q; - struct block_device *bdev; + struct file *bdev_file; struct btrfs_super_block *disk_super; u64 devid; int ret; @@ -617,8 +659,8 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices, if (!device->name) return -EINVAL; - ret = btrfs_get_bdev_and_sb(device->name->str, flags, holder, 1, - &bdev, &disk_super); + ret = btrfs_get_bdev_and_sb(rcu_dereference_raw(device->name), flags, holder, 1, + &bdev_file, &disk_super); if (ret) return ret; @@ -634,27 +676,39 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices, if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING) { if (btrfs_super_incompat_flags(disk_super) & BTRFS_FEATURE_INCOMPAT_METADATA_UUID) { - pr_err( - "BTRFS: Invalid seeding and uuid-changed device detected\n"); + btrfs_err(NULL, + "invalid seeding and uuid-changed device detected"); goto error_free_page; } clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); fs_devices->seeding = true; } else { - if (bdev_read_only(bdev)) + if (bdev_read_only(file_bdev(bdev_file))) clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); else set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); } - q = bdev_get_queue(bdev); - if (!blk_queue_nonrot(q)) + if (!bdev_nonrot(file_bdev(bdev_file))) fs_devices->rotating = true; - device->bdev = bdev; + if (bdev_max_discard_sectors(file_bdev(bdev_file))) + fs_devices->discardable = true; + + device->bdev_file = bdev_file; + device->bdev = file_bdev(bdev_file); clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state); - device->mode = flags; + + if (device->devt != device->bdev->bd_dev) { + btrfs_warn(NULL, + "device %s maj:min changed from %d:%d to %d:%d", + rcu_dereference_raw(device->name), MAJOR(device->devt), + MINOR(device->devt), MAJOR(device->bdev->bd_dev), + MINOR(device->bdev->bd_dev)); + + device->devt = device->bdev->bd_dev; + } fs_devices->open_devices++; if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) && @@ -668,95 +722,54 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices, error_free_page: btrfs_release_disk_super(disk_super); - blkdev_put(bdev, flags); + bdev_fput(bdev_file); return -EINVAL; } -/* - * Handle scanned device having its CHANGING_FSID_V2 flag set and the fs_devices - * being created with a disk that has already completed its fsid change. Such - * disk can belong to an fs which has its FSID changed or to one which doesn't. - * Handle both cases here. - */ -static struct btrfs_fs_devices *find_fsid_inprogress( - struct btrfs_super_block *disk_super) +const u8 *btrfs_sb_fsid_ptr(const struct btrfs_super_block *sb) { - struct btrfs_fs_devices *fs_devices; + bool has_metadata_uuid = (btrfs_super_incompat_flags(sb) & + BTRFS_FEATURE_INCOMPAT_METADATA_UUID); - list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - if (memcmp(fs_devices->metadata_uuid, fs_devices->fsid, - BTRFS_FSID_SIZE) != 0 && - memcmp(fs_devices->metadata_uuid, disk_super->fsid, - BTRFS_FSID_SIZE) == 0 && !fs_devices->fsid_change) { - return fs_devices; - } - } - - return find_fsid(disk_super->fsid, NULL); + return has_metadata_uuid ? sb->metadata_uuid : sb->fsid; } - -static struct btrfs_fs_devices *find_fsid_changed( - struct btrfs_super_block *disk_super) +static bool is_same_device(struct btrfs_device *device, const char *new_path) { - struct btrfs_fs_devices *fs_devices; - - /* - * Handles the case where scanned device is part of an fs that had - * multiple successful changes of FSID but curently device didn't - * observe it. Meaning our fsid will be different than theirs. We need - * to handle two subcases : - * 1 - The fs still continues to have different METADATA/FSID uuids. - * 2 - The fs is switched back to its original FSID (METADATA/FSID - * are equal). - */ - list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - /* Changed UUIDs */ - if (memcmp(fs_devices->metadata_uuid, fs_devices->fsid, - BTRFS_FSID_SIZE) != 0 && - memcmp(fs_devices->metadata_uuid, disk_super->metadata_uuid, - BTRFS_FSID_SIZE) == 0 && - memcmp(fs_devices->fsid, disk_super->fsid, - BTRFS_FSID_SIZE) != 0) - return fs_devices; - - /* Unchanged UUIDs */ - if (memcmp(fs_devices->metadata_uuid, fs_devices->fsid, - BTRFS_FSID_SIZE) == 0 && - memcmp(fs_devices->fsid, disk_super->metadata_uuid, - BTRFS_FSID_SIZE) == 0) - return fs_devices; - } + struct path old = { .mnt = NULL, .dentry = NULL }; + struct path new = { .mnt = NULL, .dentry = NULL }; + char AUTO_KFREE(old_path); + bool is_same = false; + int ret; - return NULL; -} + if (!device->name) + goto out; -static struct btrfs_fs_devices *find_fsid_reverted_metadata( - struct btrfs_super_block *disk_super) -{ - struct btrfs_fs_devices *fs_devices; + old_path = kzalloc(PATH_MAX, GFP_NOFS); + if (!old_path) + goto out; - /* - * Handle the case where the scanned device is part of an fs whose last - * metadata UUID change reverted it to the original FSID. At the same - * time * fs_devices was first created by another constitutent device - * which didn't fully observe the operation. This results in an - * btrfs_fs_devices created with metadata/fsid different AND - * btrfs_fs_devices::fsid_change set AND the metadata_uuid of the - * fs_devices equal to the FSID of the disk. - */ - list_for_each_entry(fs_devices, &fs_uuids, fs_list) { - if (memcmp(fs_devices->fsid, fs_devices->metadata_uuid, - BTRFS_FSID_SIZE) != 0 && - memcmp(fs_devices->metadata_uuid, disk_super->fsid, - BTRFS_FSID_SIZE) == 0 && - fs_devices->fsid_change) - return fs_devices; - } + rcu_read_lock(); + ret = strscpy(old_path, rcu_dereference(device->name), PATH_MAX); + rcu_read_unlock(); + if (ret < 0) + goto out; - return NULL; + ret = kern_path(old_path, LOOKUP_FOLLOW, &old); + if (ret) + goto out; + ret = kern_path(new_path, LOOKUP_FOLLOW, &new); + if (ret) + goto out; + if (path_equal(&old, &new)) + is_same = true; +out: + path_put(&old); + path_put(&new); + return is_same; } + /* * Add new device to list of registered devices * @@ -770,92 +783,93 @@ static noinline struct btrfs_device *device_list_add(const char *path, { struct btrfs_device *device; struct btrfs_fs_devices *fs_devices = NULL; - struct rcu_string *name; + const char *name; u64 found_transid = btrfs_super_generation(disk_super); u64 devid = btrfs_stack_device_id(&disk_super->dev_item); + dev_t path_devt; + int ret; + bool same_fsid_diff_dev = false; bool has_metadata_uuid = (btrfs_super_incompat_flags(disk_super) & BTRFS_FEATURE_INCOMPAT_METADATA_UUID); - bool fsid_change_in_progress = (btrfs_super_flags(disk_super) & - BTRFS_SUPER_FLAG_CHANGING_FSID_V2); - if (fsid_change_in_progress) { - if (!has_metadata_uuid) - fs_devices = find_fsid_inprogress(disk_super); - else - fs_devices = find_fsid_changed(disk_super); - } else if (has_metadata_uuid) { - fs_devices = find_fsid_with_metadata_uuid(disk_super); - } else { - fs_devices = find_fsid_reverted_metadata(disk_super); - if (!fs_devices) - fs_devices = find_fsid(disk_super->fsid, NULL); + if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_CHANGING_FSID_V2) { + btrfs_err(NULL, +"device %s has incomplete metadata_uuid change, please use btrfstune to complete", + path); + return ERR_PTR(-EAGAIN); } + ret = lookup_bdev(path, &path_devt); + if (ret) { + btrfs_err(NULL, "failed to lookup block device for path %s: %d", + path, ret); + return ERR_PTR(ret); + } - if (!fs_devices) { - if (has_metadata_uuid) - fs_devices = alloc_fs_devices(disk_super->fsid, - disk_super->metadata_uuid); - else - fs_devices = alloc_fs_devices(disk_super->fsid, NULL); + fs_devices = find_fsid_by_device(disk_super, path_devt, &same_fsid_diff_dev); + if (!fs_devices) { + fs_devices = alloc_fs_devices(disk_super->fsid); if (IS_ERR(fs_devices)) return ERR_CAST(fs_devices); - fs_devices->fsid_change = fsid_change_in_progress; + if (has_metadata_uuid) + memcpy(fs_devices->metadata_uuid, + disk_super->metadata_uuid, BTRFS_FSID_SIZE); + + if (same_fsid_diff_dev) { + generate_random_uuid(fs_devices->fsid); + fs_devices->temp_fsid = true; + btrfs_info(NULL, "device %s (%d:%d) using temp-fsid %pU", + path, MAJOR(path_devt), MINOR(path_devt), + fs_devices->fsid); + } mutex_lock(&fs_devices->device_list_mutex); list_add(&fs_devices->fs_list, &fs_uuids); device = NULL; } else { + struct btrfs_dev_lookup_args args = { + .devid = devid, + .uuid = disk_super->dev_item.uuid, + }; + mutex_lock(&fs_devices->device_list_mutex); - device = btrfs_find_device(fs_devices, devid, - disk_super->dev_item.uuid, NULL, false); + device = btrfs_find_device(fs_devices, &args); - /* - * If this disk has been pulled into an fs devices created by - * a device which had the CHANGING_FSID_V2 flag then replace the - * metadata_uuid/fsid values of the fs_devices. - */ - if (fs_devices->fsid_change && - found_transid > fs_devices->latest_generation) { + if (found_transid > fs_devices->latest_generation) { memcpy(fs_devices->fsid, disk_super->fsid, BTRFS_FSID_SIZE); - - if (has_metadata_uuid) - memcpy(fs_devices->metadata_uuid, - disk_super->metadata_uuid, - BTRFS_FSID_SIZE); - else - memcpy(fs_devices->metadata_uuid, - disk_super->fsid, BTRFS_FSID_SIZE); - - fs_devices->fsid_change = false; + memcpy(fs_devices->metadata_uuid, + btrfs_sb_fsid_ptr(disk_super), BTRFS_FSID_SIZE); } } if (!device) { + unsigned int nofs_flag; + if (fs_devices->opened) { + btrfs_err(NULL, +"device %s (%d:%d) belongs to fsid %pU, and the fs is already mounted, scanned by %s (%d)", + path, MAJOR(path_devt), MINOR(path_devt), + fs_devices->fsid, current->comm, + task_pid_nr(current)); mutex_unlock(&fs_devices->device_list_mutex); return ERR_PTR(-EBUSY); } + nofs_flag = memalloc_nofs_save(); device = btrfs_alloc_device(NULL, &devid, - disk_super->dev_item.uuid); + disk_super->dev_item.uuid, path); + memalloc_nofs_restore(nofs_flag); if (IS_ERR(device)) { mutex_unlock(&fs_devices->device_list_mutex); /* we can safely leave the fs_devices entry around */ return device; } - name = rcu_string_strdup(path, GFP_NOFS); - if (!name) { - btrfs_free_device(device); - mutex_unlock(&fs_devices->device_list_mutex); - return ERR_PTR(-ENOMEM); - } - rcu_assign_pointer(device->name, name); + device->devt = path_devt; list_add_rcu(&device->dev_list, &fs_devices->devices); fs_devices->num_devices++; @@ -865,16 +879,20 @@ static noinline struct btrfs_device *device_list_add(const char *path, if (disk_super->label[0]) pr_info( - "BTRFS: device label %s devid %llu transid %llu %s scanned by %s (%d)\n", +"BTRFS: device label %s devid %llu transid %llu %s (%d:%d) scanned by %s (%d)\n", disk_super->label, devid, found_transid, path, + MAJOR(path_devt), MINOR(path_devt), current->comm, task_pid_nr(current)); else pr_info( - "BTRFS: device fsid %pU devid %llu transid %llu %s scanned by %s (%d)\n", +"BTRFS: device fsid %pU devid %llu transid %llu %s (%d:%d) scanned by %s (%d)\n", disk_super->fsid, devid, found_transid, path, + MAJOR(path_devt), MINOR(path_devt), current->comm, task_pid_nr(current)); - } else if (!device->name || strcmp(device->name->str, path)) { + } else if (!device->name || !is_same_device(device, path)) { + const char *old_name; + /* * When FS is already mounted. * 1. If you are here and if the device->name is NULL that @@ -910,49 +928,54 @@ static noinline struct btrfs_device *device_list_add(const char *path, * generation are equal. */ mutex_unlock(&fs_devices->device_list_mutex); + btrfs_err(NULL, +"device %s already registered with a higher generation, found %llu expect %llu", + path, found_transid, device->generation); return ERR_PTR(-EEXIST); } /* * We are going to replace the device path for a given devid, * make sure it's the same device if the device is mounted + * + * NOTE: the device->fs_info may not be reliable here so pass + * in a NULL to message helpers instead. This avoids a possible + * use-after-free when the fs_info and fs_info->sb are already + * torn down. */ if (device->bdev) { - struct block_device *path_bdev; - - path_bdev = lookup_bdev(path); - if (IS_ERR(path_bdev)) { - mutex_unlock(&fs_devices->device_list_mutex); - return ERR_CAST(path_bdev); - } - - if (device->bdev != path_bdev) { - bdput(path_bdev); + if (device->devt != path_devt) { mutex_unlock(&fs_devices->device_list_mutex); - btrfs_warn_in_rcu(device->fs_info, - "duplicate device fsid:devid for %pU:%llu old:%s new:%s", - disk_super->fsid, devid, - rcu_str_deref(device->name), path); + btrfs_warn(NULL, + "duplicate device %s devid %llu generation %llu scanned by %s (%d)", + path, devid, found_transid, + current->comm, + task_pid_nr(current)); return ERR_PTR(-EEXIST); } - bdput(path_bdev); - btrfs_info_in_rcu(device->fs_info, - "device fsid %pU devid %llu moved old:%s new:%s", - disk_super->fsid, devid, - rcu_str_deref(device->name), path); + btrfs_info(NULL, + "devid %llu device path %s changed to %s scanned by %s (%d)", + devid, btrfs_dev_name(device), + path, current->comm, + task_pid_nr(current)); } - name = rcu_string_strdup(path, GFP_NOFS); + name = kstrdup(path, GFP_NOFS); if (!name) { mutex_unlock(&fs_devices->device_list_mutex); return ERR_PTR(-ENOMEM); } - rcu_string_free(device->name); + rcu_read_lock(); + old_name = rcu_dereference(device->name); + rcu_read_unlock(); rcu_assign_pointer(device->name, name); + kfree_rcu_mightsleep(old_name); + if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) { fs_devices->missing_devices--; clear_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state); } + device->devt = path_devt; } /* @@ -980,115 +1003,113 @@ static struct btrfs_fs_devices *clone_fs_devices(struct btrfs_fs_devices *orig) struct btrfs_device *orig_dev; int ret = 0; - fs_devices = alloc_fs_devices(orig->fsid, NULL); + lockdep_assert_held(&uuid_mutex); + + fs_devices = alloc_fs_devices(orig->fsid); if (IS_ERR(fs_devices)) return fs_devices; - mutex_lock(&orig->device_list_mutex); fs_devices->total_devices = orig->total_devices; list_for_each_entry(orig_dev, &orig->devices, dev_list) { - struct rcu_string *name; + const char *dev_path = NULL; + + /* + * This is ok to do without RCU read locked because we hold the + * uuid mutex so nothing we touch in here is going to disappear. + */ + if (orig_dev->name) + dev_path = rcu_dereference_raw(orig_dev->name); device = btrfs_alloc_device(NULL, &orig_dev->devid, - orig_dev->uuid); + orig_dev->uuid, dev_path); if (IS_ERR(device)) { ret = PTR_ERR(device); goto error; } - /* - * This is ok to do without rcu read locked because we hold the - * uuid mutex so nothing we touch in here is going to disappear. - */ - if (orig_dev->name) { - name = rcu_string_strdup(orig_dev->name->str, - GFP_KERNEL); - if (!name) { + if (orig_dev->zone_info) { + struct btrfs_zoned_device_info *zone_info; + + zone_info = btrfs_clone_dev_zone_info(orig_dev); + if (!zone_info) { btrfs_free_device(device); ret = -ENOMEM; goto error; } - rcu_assign_pointer(device->name, name); + device->zone_info = zone_info; } list_add(&device->dev_list, &fs_devices->devices); device->fs_devices = fs_devices; fs_devices->num_devices++; } - mutex_unlock(&orig->device_list_mutex); return fs_devices; error: - mutex_unlock(&orig->device_list_mutex); free_fs_devices(fs_devices); return ERR_PTR(ret); } -/* - * After we have read the system tree and know devids belonging to - * this filesystem, remove the device which does not belong there. - */ -void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, int step) +static void __btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, + struct btrfs_device **latest_dev) { struct btrfs_device *device, *next; - struct btrfs_device *latest_dev = NULL; - mutex_lock(&uuid_mutex); -again: /* This is the initialized path, it is safe to release the devices. */ list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) { - if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, - &device->dev_state)) { + if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state)) { if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT, - &device->dev_state) && + &device->dev_state) && !test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state) && - (!latest_dev || - device->generation > latest_dev->generation)) { - latest_dev = device; + (!*latest_dev || + device->generation > (*latest_dev)->generation)) { + *latest_dev = device; } continue; } - if (device->devid == BTRFS_DEV_REPLACE_DEVID) { - /* - * In the first step, keep the device which has - * the correct fsid and the devid that is used - * for the dev_replace procedure. - * In the second step, the dev_replace state is - * read from the device tree and it is known - * whether the procedure is really active or - * not, which means whether this device is - * used or whether it should be removed. - */ - if (step == 0 || test_bit(BTRFS_DEV_STATE_REPLACE_TGT, - &device->dev_state)) { - continue; - } - } - if (device->bdev) { - blkdev_put(device->bdev, device->mode); + /* + * We have already validated the presence of BTRFS_DEV_REPLACE_DEVID, + * in btrfs_init_dev_replace() so just continue. + */ + if (device->devid == BTRFS_DEV_REPLACE_DEVID) + continue; + + if (device->bdev_file) { + bdev_fput(device->bdev_file); device->bdev = NULL; + device->bdev_file = NULL; fs_devices->open_devices--; } if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { list_del_init(&device->dev_alloc_list); clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); - if (!test_bit(BTRFS_DEV_STATE_REPLACE_TGT, - &device->dev_state)) - fs_devices->rw_devices--; + fs_devices->rw_devices--; } list_del_init(&device->dev_list); fs_devices->num_devices--; btrfs_free_device(device); } - if (fs_devices->seed) { - fs_devices = fs_devices->seed; - goto again; - } +} + +/* + * After we have read the system tree and know devids belonging to this + * filesystem, remove the device which does not belong there. + */ +void btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices) +{ + struct btrfs_device *latest_dev = NULL; + struct btrfs_fs_devices *seed_dev; + + mutex_lock(&uuid_mutex); + __btrfs_free_extra_devids(fs_devices, &latest_dev); - fs_devices->latest_bdev = latest_dev->bdev; + list_for_each_entry(seed_dev, &fs_devices->seed_list, seed_list) + __btrfs_free_extra_devids(seed_dev, &latest_dev); + + fs_devices->latest_dev = latest_dev; mutex_unlock(&uuid_mutex); } @@ -1103,7 +1124,7 @@ static void btrfs_close_bdev(struct btrfs_device *device) invalidate_bdev(device->bdev); } - blkdev_put(device->bdev, device->mode); + bdev_fput(device->bdev_file); } static void btrfs_close_one_device(struct btrfs_device *device) @@ -1116,102 +1137,158 @@ static void btrfs_close_one_device(struct btrfs_device *device) fs_devices->rw_devices--; } - if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) + if (device->devid == BTRFS_DEV_REPLACE_DEVID) + clear_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state); + + if (test_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state)) { + clear_bit(BTRFS_DEV_STATE_MISSING, &device->dev_state); fs_devices->missing_devices--; + } btrfs_close_bdev(device); if (device->bdev) { fs_devices->open_devices--; device->bdev = NULL; + device->bdev_file = NULL; } clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); + btrfs_destroy_dev_zone_info(device); device->fs_info = NULL; atomic_set(&device->dev_stats_ccnt, 0); - extent_io_tree_release(&device->alloc_state); + btrfs_extent_io_tree_release(&device->alloc_state); + + /* + * Reset the flush error record. We might have a transient flush error + * in this mount, and if so we aborted the current transaction and set + * the fs to an error state, guaranteeing no super blocks can be further + * committed. However that error might be transient and if we unmount the + * filesystem and mount it again, we should allow the mount to succeed + * (btrfs_check_rw_degradable() should not fail) - if after mounting the + * filesystem again we still get flush errors, then we will again abort + * any transaction and set the error state, guaranteeing no commits of + * unsafe super blocks. + */ + device->last_flush_error = 0; /* Verify the device is back in a pristine state */ - ASSERT(!test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state)); - ASSERT(!test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)); - ASSERT(list_empty(&device->dev_alloc_list)); - ASSERT(list_empty(&device->post_commit_list)); - ASSERT(atomic_read(&device->reada_in_flight) == 0); + WARN_ON(test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state)); + WARN_ON(test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)); + WARN_ON(!list_empty(&device->dev_alloc_list)); + WARN_ON(!list_empty(&device->post_commit_list)); } -static int close_fs_devices(struct btrfs_fs_devices *fs_devices) +static void close_fs_devices(struct btrfs_fs_devices *fs_devices) { struct btrfs_device *device, *tmp; + lockdep_assert_held(&uuid_mutex); + if (--fs_devices->opened > 0) - return 0; + return; - mutex_lock(&fs_devices->device_list_mutex); - list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list) { + list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list) btrfs_close_one_device(device); - } - mutex_unlock(&fs_devices->device_list_mutex); WARN_ON(fs_devices->open_devices); WARN_ON(fs_devices->rw_devices); fs_devices->opened = 0; fs_devices->seeding = false; - - return 0; + fs_devices->fs_info = NULL; } -int btrfs_close_devices(struct btrfs_fs_devices *fs_devices) +void btrfs_close_devices(struct btrfs_fs_devices *fs_devices) { - struct btrfs_fs_devices *seed_devices = NULL; - int ret; + LIST_HEAD(list); + struct btrfs_fs_devices *tmp; mutex_lock(&uuid_mutex); - ret = close_fs_devices(fs_devices); - if (!fs_devices->opened) { - seed_devices = fs_devices->seed; - fs_devices->seed = NULL; + close_fs_devices(fs_devices); + if (!fs_devices->opened && !fs_devices->holding) { + list_splice_init(&fs_devices->seed_list, &list); + + /* + * If the struct btrfs_fs_devices is not assembled with any + * other device, it can be re-initialized during the next mount + * without the needing device-scan step. Therefore, it can be + * fully freed. + */ + if (fs_devices->num_devices == 1) { + list_del(&fs_devices->fs_list); + free_fs_devices(fs_devices); + } } - mutex_unlock(&uuid_mutex); - while (seed_devices) { - fs_devices = seed_devices; - seed_devices = fs_devices->seed; + + list_for_each_entry_safe(fs_devices, tmp, &list, seed_list) { close_fs_devices(fs_devices); + list_del(&fs_devices->seed_list); free_fs_devices(fs_devices); } - return ret; + mutex_unlock(&uuid_mutex); } static int open_fs_devices(struct btrfs_fs_devices *fs_devices, - fmode_t flags, void *holder) + blk_mode_t flags, void *holder) { struct btrfs_device *device; struct btrfs_device *latest_dev = NULL; + struct btrfs_device *tmp_device; + s64 __maybe_unused value = 0; + int ret = 0; - flags |= FMODE_EXCL; - - list_for_each_entry(device, &fs_devices->devices, dev_list) { - /* Just open everything we can; ignore failures here */ - if (btrfs_open_one_device(fs_devices, device, flags, holder)) - continue; + list_for_each_entry_safe(device, tmp_device, &fs_devices->devices, + dev_list) { + int ret2; - if (!latest_dev || - device->generation > latest_dev->generation) + ret2 = btrfs_open_one_device(fs_devices, device, flags, holder); + if (ret2 == 0 && + (!latest_dev || device->generation > latest_dev->generation)) { latest_dev = device; + } else if (ret2 == -ENODATA) { + fs_devices->num_devices--; + list_del(&device->dev_list); + btrfs_free_device(device); + } + if (ret == 0 && ret2 != 0) + ret = ret2; } - if (fs_devices->open_devices == 0) + + if (fs_devices->open_devices == 0) { + if (ret) + return ret; return -EINVAL; + } fs_devices->opened = 1; - fs_devices->latest_bdev = latest_dev->bdev; + fs_devices->latest_dev = latest_dev; fs_devices->total_rw_bytes = 0; fs_devices->chunk_alloc_policy = BTRFS_CHUNK_ALLOC_REGULAR; +#ifdef CONFIG_BTRFS_EXPERIMENTAL + fs_devices->rr_min_contig_read = BTRFS_DEFAULT_RR_MIN_CONTIG_READ; + fs_devices->read_devid = latest_dev->devid; + fs_devices->read_policy = btrfs_read_policy_to_enum(btrfs_get_mod_read_policy(), + &value); + if (fs_devices->read_policy == BTRFS_READ_POLICY_RR) + fs_devices->collect_fs_stats = true; + + if (value) { + if (fs_devices->read_policy == BTRFS_READ_POLICY_RR) + fs_devices->rr_min_contig_read = value; + if (fs_devices->read_policy == BTRFS_READ_POLICY_DEVID) + fs_devices->read_devid = value; + } +#else + fs_devices->read_policy = BTRFS_READ_POLICY_PID; +#endif return 0; } -static int devid_cmp(void *priv, struct list_head *a, struct list_head *b) +static int devid_cmp(void *priv, const struct list_head *a, + const struct list_head *b) { - struct btrfs_device *dev1, *dev2; + const struct btrfs_device *dev1, *dev2; dev1 = list_entry(a, struct btrfs_device, dev_list); dev2 = list_entry(b, struct btrfs_device, dev_list); @@ -1224,13 +1301,19 @@ static int devid_cmp(void *priv, struct list_head *a, struct list_head *b) } int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, - fmode_t flags, void *holder) + blk_mode_t flags, void *holder) { int ret; lockdep_assert_held(&uuid_mutex); + /* + * The device_list_mutex cannot be taken here in case opening the + * underlying device takes further locks like open_mutex. + * + * We also don't need the lock here as this is called during mount and + * exclusion is provided by uuid_mutex + */ - mutex_lock(&fs_devices->device_list_mutex); if (fs_devices->opened) { fs_devices->opened++; ret = 0; @@ -1238,7 +1321,6 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, list_sort(NULL, &fs_devices->devices, devid_cmp); ret = open_fs_devices(fs_devices, flags, holder); } - mutex_unlock(&fs_devices->device_list_mutex); return ret; } @@ -1250,106 +1332,172 @@ void btrfs_release_disk_super(struct btrfs_super_block *super) put_page(page); } -static struct btrfs_super_block *btrfs_read_disk_super(struct block_device *bdev, - u64 bytenr) +struct btrfs_super_block *btrfs_read_disk_super(struct block_device *bdev, + int copy_num, bool drop_cache) { - struct btrfs_super_block *disk_super; + struct btrfs_super_block *super; struct page *page; - void *p; - pgoff_t index; + u64 bytenr, bytenr_orig; + struct address_space *mapping = bdev->bd_mapping; + int ret; - /* make sure our super fits in the device */ - if (bytenr + PAGE_SIZE >= i_size_read(bdev->bd_inode)) - return ERR_PTR(-EINVAL); + bytenr_orig = btrfs_sb_offset(copy_num); + ret = btrfs_sb_log_location_bdev(bdev, copy_num, READ, &bytenr); + if (ret < 0) { + if (ret == -ENOENT) + ret = -EINVAL; + return ERR_PTR(ret); + } - /* make sure our super fits in the page */ - if (sizeof(*disk_super) > PAGE_SIZE) + if (bytenr + BTRFS_SUPER_INFO_SIZE >= bdev_nr_bytes(bdev)) return ERR_PTR(-EINVAL); - /* make sure our super doesn't straddle pages on disk */ - index = bytenr >> PAGE_SHIFT; - if ((bytenr + sizeof(*disk_super) - 1) >> PAGE_SHIFT != index) - return ERR_PTR(-EINVAL); + if (drop_cache) { + /* This should only be called with the primary sb. */ + ASSERT(copy_num == 0); - /* pull in the page with our super */ - page = read_cache_page_gfp(bdev->bd_inode->i_mapping, index, GFP_KERNEL); + /* + * Drop the page of the primary superblock, so later read will + * always read from the device. + */ + invalidate_inode_pages2_range(mapping, bytenr >> PAGE_SHIFT, + (bytenr + BTRFS_SUPER_INFO_SIZE) >> PAGE_SHIFT); + } + page = read_cache_page_gfp(mapping, bytenr >> PAGE_SHIFT, GFP_NOFS); if (IS_ERR(page)) return ERR_CAST(page); - p = page_address(page); - - /* align our pointer to the offset of the super block */ - disk_super = p + offset_in_page(bytenr); - - if (btrfs_super_bytenr(disk_super) != bytenr || - btrfs_super_magic(disk_super) != BTRFS_MAGIC) { - btrfs_release_disk_super(p); + super = page_address(page); + if (btrfs_super_magic(super) != BTRFS_MAGIC || + btrfs_super_bytenr(super) != bytenr_orig) { + btrfs_release_disk_super(super); return ERR_PTR(-EINVAL); } - if (disk_super->label[0] && disk_super->label[BTRFS_LABEL_SIZE - 1]) - disk_super->label[BTRFS_LABEL_SIZE - 1] = 0; + /* + * Make sure the last byte of label is properly NUL terminated. We use + * '%s' to print the label, if not properly NUL terminated we can access + * beyond the label. + */ + if (super->label[0] && super->label[BTRFS_LABEL_SIZE - 1]) + super->label[BTRFS_LABEL_SIZE - 1] = 0; - return disk_super; + return super; } -int btrfs_forget_devices(const char *path) +int btrfs_forget_devices(dev_t devt) { int ret; mutex_lock(&uuid_mutex); - ret = btrfs_free_stale_devices(strlen(path) ? path : NULL, NULL); + ret = btrfs_free_stale_devices(devt, NULL); mutex_unlock(&uuid_mutex); return ret; } +static bool btrfs_skip_registration(struct btrfs_super_block *disk_super, + const char *path, dev_t devt, + bool mount_arg_dev) +{ + struct btrfs_fs_devices *fs_devices; + + /* + * Do not skip device registration for mounted devices with matching + * maj:min but different paths. Booting without initrd relies on + * /dev/root initially, later replaced with the actual root device. + * A successful scan ensures grub2-probe selects the correct device. + */ + list_for_each_entry(fs_devices, &fs_uuids, fs_list) { + struct btrfs_device *device; + + mutex_lock(&fs_devices->device_list_mutex); + + if (!fs_devices->opened) { + mutex_unlock(&fs_devices->device_list_mutex); + continue; + } + + list_for_each_entry(device, &fs_devices->devices, dev_list) { + if (device->bdev && (device->bdev->bd_dev == devt) && + strcmp(rcu_dereference_raw(device->name), path) != 0) { + mutex_unlock(&fs_devices->device_list_mutex); + + /* Do not skip registration. */ + return false; + } + } + mutex_unlock(&fs_devices->device_list_mutex); + } + + if (!mount_arg_dev && btrfs_super_num_devices(disk_super) == 1 && + !(btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_SEEDING)) + return true; + + return false; +} + /* * Look for a btrfs signature on a device. This may be called out of the mount path * and we are not allowed to call set_blocksize during the scan. The superblock - * is read via pagecache + * is read via pagecache. + * + * With @mount_arg_dev it's a scan during mount time that will always register + * the device or return an error. Multi-device and seeding devices are registered + * in both cases. */ -struct btrfs_device *btrfs_scan_one_device(const char *path, fmode_t flags, - void *holder) +struct btrfs_device *btrfs_scan_one_device(const char *path, + bool mount_arg_dev) { struct btrfs_super_block *disk_super; bool new_device_added = false; struct btrfs_device *device = NULL; - struct block_device *bdev; - u64 bytenr; + struct file *bdev_file; + dev_t devt; lockdep_assert_held(&uuid_mutex); /* - * we would like to check all the supers, but that would make - * a btrfs mount succeed after a mkfs from a different FS. - * So, we need to add a special mount option to scan for - * later supers, using BTRFS_SUPER_MIRROR_MAX instead + * Avoid an exclusive open here, as the systemd-udev may initiate the + * device scan which may race with the user's mount or mkfs command, + * resulting in failure. + * Since the device scan is solely for reading purposes, there is no + * need for an exclusive open. Additionally, the devices are read again + * during the mount process. It is ok to get some inconsistent + * values temporarily, as the device paths of the fsid are the only + * required information for assembling the volume. */ - bytenr = btrfs_sb_offset(0); - flags |= FMODE_EXCL; - - bdev = blkdev_get_by_path(path, flags, holder); - if (IS_ERR(bdev)) - return ERR_CAST(bdev); + bdev_file = bdev_file_open_by_path(path, BLK_OPEN_READ, NULL, NULL); + if (IS_ERR(bdev_file)) + return ERR_CAST(bdev_file); - disk_super = btrfs_read_disk_super(bdev, bytenr); + disk_super = btrfs_read_disk_super(file_bdev(bdev_file), 0, false); if (IS_ERR(disk_super)) { device = ERR_CAST(disk_super); goto error_bdev_put; } - device = device_list_add(path, disk_super, &new_device_added); - if (!IS_ERR(device)) { - if (new_device_added) - btrfs_free_stale_devices(path, device); + devt = file_bdev(bdev_file)->bd_dev; + if (btrfs_skip_registration(disk_super, path, devt, mount_arg_dev)) { + btrfs_debug(NULL, "skip registering single non-seed device %s (%d:%d)", + path, MAJOR(devt), MINOR(devt)); + + btrfs_free_stale_devices(devt, NULL); + + device = NULL; + goto free_disk_super; } + device = device_list_add(path, disk_super, &new_device_added); + if (!IS_ERR(device) && new_device_added) + btrfs_free_stale_devices(device->devt, device); + +free_disk_super: btrfs_release_disk_super(disk_super); error_bdev_put: - blkdev_put(bdev, flags); + bdev_fput(bdev_file); return device; } @@ -1365,13 +1513,13 @@ static bool contains_pending_extent(struct btrfs_device *device, u64 *start, lockdep_assert_held(&device->fs_info->chunk_mutex); - if (!find_first_extent_bit(&device->alloc_state, *start, - &physical_start, &physical_end, - CHUNK_ALLOCATED, NULL)) { + if (btrfs_find_first_extent_bit(&device->alloc_state, *start, + &physical_start, &physical_end, + CHUNK_ALLOCATED, NULL)) { if (in_range(physical_start, *start, len) || in_range(*start, physical_start, - physical_end - physical_start)) { + physical_end + 1 - physical_start)) { *start = physical_end + 1; return true; } @@ -1379,29 +1527,78 @@ static bool contains_pending_extent(struct btrfs_device *device, u64 *start, return false; } -static u64 dev_extent_search_start(struct btrfs_device *device, u64 start) +static u64 dev_extent_search_start(struct btrfs_device *device) { switch (device->fs_devices->chunk_alloc_policy) { + default: + btrfs_warn_unknown_chunk_allocation(device->fs_devices->chunk_alloc_policy); + fallthrough; case BTRFS_CHUNK_ALLOC_REGULAR: + return BTRFS_DEVICE_RANGE_RESERVED; + case BTRFS_CHUNK_ALLOC_ZONED: /* - * We don't want to overwrite the superblock on the drive nor - * any area used by the boot loader (grub for example), so we - * make sure to start at an offset of at least 1MB. + * We don't care about the starting region like regular + * allocator, because we anyway use/reserve the first two zones + * for superblock logging. */ - return max_t(u64, start, SZ_1M); - default: - BUG(); + return 0; } } -/** - * dev_extent_hole_check - check if specified hole is suitable for allocation +static bool dev_extent_hole_check_zoned(struct btrfs_device *device, + u64 *hole_start, u64 *hole_size, + u64 num_bytes) +{ + u64 zone_size = device->zone_info->zone_size; + u64 pos; + int ret; + bool changed = false; + + ASSERT(IS_ALIGNED(*hole_start, zone_size), + "hole_start=%llu zone_size=%llu", *hole_start, zone_size); + + while (*hole_size > 0) { + pos = btrfs_find_allocatable_zones(device, *hole_start, + *hole_start + *hole_size, + num_bytes); + if (pos != *hole_start) { + *hole_size = *hole_start + *hole_size - pos; + *hole_start = pos; + changed = true; + if (*hole_size < num_bytes) + break; + } + + ret = btrfs_ensure_empty_zones(device, pos, num_bytes); + + /* Range is ensured to be empty */ + if (!ret) + return changed; + + /* Given hole range was invalid (outside of device) */ + if (ret == -ERANGE) { + *hole_start += *hole_size; + *hole_size = 0; + return true; + } + + *hole_start += zone_size; + *hole_size -= zone_size; + changed = true; + } + + return changed; +} + +/* + * Check if specified hole is suitable for allocation. + * * @device: the device which we have the hole * @hole_start: starting position of the hole * @hole_size: the size of the hole * @num_bytes: the size of the free space that we need * - * This function may modify @hole_start and @hole_end to reflect the suitable + * This function may modify @hole_start and @hole_size to reflect the suitable * position for allocation. Returns 1 if hole position is updated, 0 otherwise. */ static bool dev_extent_hole_check(struct btrfs_device *device, u64 *hole_start, @@ -1410,31 +1607,48 @@ static bool dev_extent_hole_check(struct btrfs_device *device, u64 *hole_start, bool changed = false; u64 hole_end = *hole_start + *hole_size; - /* - * Check before we set max_hole_start, otherwise we could end up - * sending back this offset anyway. - */ - if (contains_pending_extent(device, hole_start, *hole_size)) { - if (hole_end >= *hole_start) - *hole_size = hole_end - *hole_start; - else - *hole_size = 0; - changed = true; - } + for (;;) { + /* + * Check before we set max_hole_start, otherwise we could end up + * sending back this offset anyway. + */ + if (contains_pending_extent(device, hole_start, *hole_size)) { + if (hole_end >= *hole_start) + *hole_size = hole_end - *hole_start; + else + *hole_size = 0; + changed = true; + } + + switch (device->fs_devices->chunk_alloc_policy) { + default: + btrfs_warn_unknown_chunk_allocation(device->fs_devices->chunk_alloc_policy); + fallthrough; + case BTRFS_CHUNK_ALLOC_REGULAR: + /* No extra check */ + break; + case BTRFS_CHUNK_ALLOC_ZONED: + if (dev_extent_hole_check_zoned(device, hole_start, + hole_size, num_bytes)) { + changed = true; + /* + * The changed hole can contain pending extent. + * Loop again to check that. + */ + continue; + } + break; + } - switch (device->fs_devices->chunk_alloc_policy) { - case BTRFS_CHUNK_ALLOC_REGULAR: - /* No extra check */ break; - default: - BUG(); } return changed; } /* - * find_free_dev_extent_start - find free space in the specified device + * Find free space in the specified device. + * * @device: the device which we search the free space in * @num_bytes: the size of the free space that we need * @search_start: the position from which to begin the search @@ -1442,9 +1656,8 @@ static bool dev_extent_hole_check(struct btrfs_device *device, u64 *hole_start, * @len: the size of the free space. that we find, or the size * of the max free space if we don't find suitable free space * - * this uses a pretty simple search, the expectation is that it is - * called very infrequently and that a given device has a small number - * of extents + * This does a pretty simple search, the expectation is that it is called very + * infrequently and that a given device has a small number of extents. * * @start is used to store the start of the free space if we find. But if we * don't find suitable free space, it will be used to store the start position @@ -1458,35 +1671,37 @@ static bool dev_extent_hole_check(struct btrfs_device *device, u64 *hole_start, * check to ensure dev extents are not double allocated. * This makes the function safe to allocate dev extents but may not report * correct usable device space, as device extent freed in current transaction - * is not reported as avaiable. + * is not reported as available. */ -static int find_free_dev_extent_start(struct btrfs_device *device, - u64 num_bytes, u64 search_start, u64 *start, - u64 *len) +static int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes, + u64 *start, u64 *len) { struct btrfs_fs_info *fs_info = device->fs_info; struct btrfs_root *root = fs_info->dev_root; struct btrfs_key key; struct btrfs_dev_extent *dev_extent; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); + u64 search_start; u64 hole_size; u64 max_hole_start; - u64 max_hole_size; + u64 max_hole_size = 0; u64 extent_end; u64 search_end = device->total_bytes; int ret; int slot; struct extent_buffer *l; - search_start = dev_extent_search_start(device, search_start); - - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - + search_start = dev_extent_search_start(device); max_hole_start = search_start; - max_hole_size = 0; + WARN_ON(device->zone_info && + !IS_ALIGNED(num_bytes, device->zone_info->zone_size)); + + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out; + } again: if (search_start >= search_end || test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { @@ -1495,23 +1710,18 @@ again: } path->reada = READA_FORWARD; - path->search_commit_root = 1; - path->skip_locking = 1; + path->search_commit_root = true; + path->skip_locking = true; key.objectid = device->devid; - key.offset = search_start; key.type = BTRFS_DEV_EXTENT_KEY; + key.offset = search_start; - ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + ret = btrfs_search_backwards(root, &key, path); if (ret < 0) goto out; - if (ret > 0) { - ret = btrfs_previous_item(root, path, key.objectid, key.type); - if (ret < 0) - goto out; - } - while (1) { + while (search_start < search_end) { l = path->nodes[0]; slot = path->slots[0]; if (slot >= btrfs_header_nritems(l)) { @@ -1534,6 +1744,9 @@ again: if (key.type != BTRFS_DEV_EXTENT_KEY) goto next; + if (key.offset > search_end) + break; + if (key.offset > search_start) { hole_size = key.offset - search_start; dev_extent_hole_check(device, &search_start, &hole_size, @@ -1594,21 +1807,16 @@ next: else ret = 0; + ASSERT(max_hole_start + max_hole_size <= search_end, + "max_hole_start=%llu max_hole_size=%llu search_end=%llu", + max_hole_start, max_hole_size, search_end); out: - btrfs_free_path(path); *start = max_hole_start; if (len) *len = max_hole_size; return ret; } -int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes, - u64 *start, u64 *len) -{ - /* FIXME use last free of some kind */ - return find_free_dev_extent_start(device, num_bytes, 0, start, len); -} - static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans, struct btrfs_device *device, u64 start, u64 *dev_extent_len) @@ -1616,7 +1824,7 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans, struct btrfs_fs_info *fs_info = device->fs_info; struct btrfs_root *root = fs_info->dev_root; int ret; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; struct btrfs_key found_key; struct extent_buffer *leaf = NULL; @@ -1627,15 +1835,15 @@ static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans, return -ENOMEM; key.objectid = device->devid; - key.offset = start; key.type = BTRFS_DEV_EXTENT_KEY; + key.offset = start; again: ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret > 0) { ret = btrfs_previous_item(root, path, key.objectid, BTRFS_DEV_EXTENT_KEY); if (ret) - goto out; + return ret; leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); extent = btrfs_item_ptr(leaf, path->slots[0], @@ -1650,81 +1858,31 @@ again: extent = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_extent); } else { - btrfs_handle_fs_error(fs_info, ret, "Slot search failed"); - goto out; + return ret; } *dev_extent_len = btrfs_dev_extent_length(leaf, extent); ret = btrfs_del_item(trans, root, path); - if (ret) { - btrfs_handle_fs_error(fs_info, ret, - "Failed to remove dev extent item"); - } else { + if (ret == 0) set_bit(BTRFS_TRANS_HAVE_FREE_BGS, &trans->transaction->flags); - } -out: - btrfs_free_path(path); - return ret; -} - -static int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans, - struct btrfs_device *device, - u64 chunk_offset, u64 start, u64 num_bytes) -{ - int ret; - struct btrfs_path *path; - struct btrfs_fs_info *fs_info = device->fs_info; - struct btrfs_root *root = fs_info->dev_root; - struct btrfs_dev_extent *extent; - struct extent_buffer *leaf; - struct btrfs_key key; - - WARN_ON(!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state)); - WARN_ON(test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)); - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - - key.objectid = device->devid; - key.offset = start; - key.type = BTRFS_DEV_EXTENT_KEY; - ret = btrfs_insert_empty_item(trans, root, path, &key, - sizeof(*extent)); - if (ret) - goto out; - - leaf = path->nodes[0]; - extent = btrfs_item_ptr(leaf, path->slots[0], - struct btrfs_dev_extent); - btrfs_set_dev_extent_chunk_tree(leaf, extent, - BTRFS_CHUNK_TREE_OBJECTID); - btrfs_set_dev_extent_chunk_objectid(leaf, extent, - BTRFS_FIRST_CHUNK_TREE_OBJECTID); - btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset); - - btrfs_set_dev_extent_length(leaf, extent, num_bytes); - btrfs_mark_buffer_dirty(leaf); -out: - btrfs_free_path(path); return ret; } static u64 find_next_chunk(struct btrfs_fs_info *fs_info) { - struct extent_map_tree *em_tree; - struct extent_map *em; struct rb_node *n; u64 ret = 0; - em_tree = &fs_info->mapping_tree; - read_lock(&em_tree->lock); - n = rb_last(&em_tree->map.rb_root); + read_lock(&fs_info->mapping_tree_lock); + n = rb_last(&fs_info->mapping_tree.rb_root); if (n) { - em = rb_entry(n, struct extent_map, rb_node); - ret = em->start + em->len; + struct btrfs_chunk_map *map; + + map = rb_entry(n, struct btrfs_chunk_map, rb_node); + ret = map->start + map->chunk_len; } - read_unlock(&em_tree->lock); + read_unlock(&fs_info->mapping_tree_lock); return ret; } @@ -1735,7 +1893,7 @@ static noinline int find_next_devid(struct btrfs_fs_info *fs_info, int ret; struct btrfs_key key; struct btrfs_key found_key; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); path = btrfs_alloc_path(); if (!path) @@ -1747,13 +1905,12 @@ static noinline int find_next_devid(struct btrfs_fs_info *fs_info, ret = btrfs_search_slot(NULL, fs_info->chunk_root, &key, path, 0, 0); if (ret < 0) - goto error; + return ret; - if (ret == 0) { + if (unlikely(ret == 0)) { /* Corruption */ btrfs_err(fs_info, "corrupted chunk tree devid -1 matched"); - ret = -EUCLEAN; - goto error; + return -EUCLEAN; } ret = btrfs_previous_item(fs_info->chunk_root, path, @@ -1766,10 +1923,7 @@ static noinline int find_next_devid(struct btrfs_fs_info *fs_info, path->slots[0]); *devid_ret = found_key.offset + 1; } - ret = 0; -error: - btrfs_free_path(path); - return ret; + return 0; } /* @@ -1780,7 +1934,7 @@ static int btrfs_add_dev_item(struct btrfs_trans_handle *trans, struct btrfs_device *device) { int ret; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_dev_item *dev_item; struct extent_buffer *leaf; struct btrfs_key key; @@ -1794,10 +1948,12 @@ static int btrfs_add_dev_item(struct btrfs_trans_handle *trans, key.type = BTRFS_DEV_ITEM_KEY; key.offset = device->devid; + btrfs_reserve_chunk_metadata(trans, true); ret = btrfs_insert_empty_item(trans, trans->fs_info->chunk_root, path, &key, sizeof(*dev_item)); + btrfs_trans_release_chunk_metadata(trans); if (ret) - goto out; + return ret; leaf = path->nodes[0]; dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item); @@ -1822,70 +1978,51 @@ static int btrfs_add_dev_item(struct btrfs_trans_handle *trans, ptr = btrfs_device_fsid(dev_item); write_extent_buffer(leaf, trans->fs_info->fs_devices->metadata_uuid, ptr, BTRFS_FSID_SIZE); - btrfs_mark_buffer_dirty(leaf); - ret = 0; -out: - btrfs_free_path(path); - return ret; + return 0; } /* * Function to update ctime/mtime for a given device path. * Mainly used for ctime/mtime based probe like libblkid. + * + * We don't care about errors here, this is just to be kind to userspace. */ -static void update_dev_time(const char *path_name) +static void update_dev_time(const char *device_path) { - struct file *filp; + struct path path; - filp = filp_open(path_name, O_RDWR, 0); - if (IS_ERR(filp)) - return; - file_update_time(filp); - filp_close(filp, NULL); + if (!kern_path(device_path, LOOKUP_FOLLOW, &path)) { + vfs_utimes(&path, NULL); + path_put(&path); + } } -static int btrfs_rm_dev_item(struct btrfs_device *device) +static int btrfs_rm_dev_item(struct btrfs_trans_handle *trans, + struct btrfs_device *device) { struct btrfs_root *root = device->fs_info->chunk_root; int ret; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; - struct btrfs_trans_handle *trans; path = btrfs_alloc_path(); if (!path) return -ENOMEM; - trans = btrfs_start_transaction(root, 0); - if (IS_ERR(trans)) { - btrfs_free_path(path); - return PTR_ERR(trans); - } key.objectid = BTRFS_DEV_ITEMS_OBJECTID; key.type = BTRFS_DEV_ITEM_KEY; key.offset = device->devid; + btrfs_reserve_chunk_metadata(trans, false); ret = btrfs_search_slot(trans, root, &key, path, -1, 1); - if (ret) { - if (ret > 0) - ret = -ENOENT; - btrfs_abort_transaction(trans, ret); - btrfs_end_transaction(trans); - goto out; - } - - ret = btrfs_del_item(trans, root, path); - if (ret) { - btrfs_abort_transaction(trans, ret); - btrfs_end_transaction(trans); - } + btrfs_trans_release_chunk_metadata(trans); + if (ret > 0) + return -ENOENT; + if (ret < 0) + return ret; -out: - btrfs_free_path(path); - if (!ret) - ret = btrfs_commit_transaction(trans); - return ret; + return btrfs_del_item(trans, root, path); } /* @@ -1912,12 +2049,8 @@ static int btrfs_check_raid_min_devices(struct btrfs_fs_info *fs_info, if (!(all_avail & btrfs_raid_array[i].bg_flag)) continue; - if (num_devices < btrfs_raid_array[i].devs_min) { - int ret = btrfs_raid_array[i].mindev_error; - - if (ret) - return ret; - } + if (num_devices < btrfs_raid_array[i].devs_min) + return btrfs_raid_array[i].mindev_error; } return 0; @@ -1939,30 +2072,27 @@ static struct btrfs_device * btrfs_find_next_active_device( } /* - * Helper function to check if the given device is part of s_bdev / latest_bdev + * Helper function to check if the given device is part of s_bdev / latest_dev * and replace it with the provided or the next active device, in the context * where this function called, there should be always be another device (or * this_dev) which is active. */ void __cold btrfs_assign_next_active_device(struct btrfs_device *device, - struct btrfs_device *this_dev) + struct btrfs_device *next_device) { struct btrfs_fs_info *fs_info = device->fs_info; - struct btrfs_device *next_device; - if (this_dev) - next_device = this_dev; - else + if (!next_device) next_device = btrfs_find_next_active_device(fs_info->fs_devices, - device); + device); ASSERT(next_device); if (fs_info->sb->s_bdev && (fs_info->sb->s_bdev == device->bdev)) fs_info->sb->s_bdev = next_device->bdev; - if (fs_info->fs_devices->latest_bdev == device->bdev) - fs_info->fs_devices->latest_bdev = next_device->bdev; + if (fs_info->fs_devices->latest_dev->bdev == device->bdev) + fs_info->fs_devices->latest_dev = next_device; } /* @@ -1975,7 +2105,7 @@ static u64 btrfs_num_devices(struct btrfs_fs_info *fs_info) down_read(&fs_info->dev_replace.rwsem); if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace)) { - ASSERT(num_devices > 1); + ASSERT(num_devices > 1, "num_devices=%llu", num_devices); num_devices--; } up_read(&fs_info->dev_replace.rwsem); @@ -1983,92 +2113,99 @@ static u64 btrfs_num_devices(struct btrfs_fs_info *fs_info) return num_devices; } -static void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info, - struct block_device *bdev, - const char *device_path) +static void btrfs_scratch_superblock(struct btrfs_fs_info *fs_info, + struct block_device *bdev, int copy_num) { struct btrfs_super_block *disk_super; - int copy_num; + const size_t len = sizeof(disk_super->magic); + const u64 bytenr = btrfs_sb_offset(copy_num); + int ret; - if (!bdev) + disk_super = btrfs_read_disk_super(bdev, copy_num, false); + if (IS_ERR(disk_super)) return; - for (copy_num = 0; copy_num < BTRFS_SUPER_MIRROR_MAX; copy_num++) { - struct page *page; - int ret; + memset(&disk_super->magic, 0, len); + folio_mark_dirty(virt_to_folio(disk_super)); + btrfs_release_disk_super(disk_super); - disk_super = btrfs_read_dev_one_super(bdev, copy_num); - if (IS_ERR(disk_super)) - continue; + ret = sync_blockdev_range(bdev, bytenr, bytenr + len - 1); + if (ret) + btrfs_warn(fs_info, "error clearing superblock number %d (%d)", + copy_num, ret); +} - memset(&disk_super->magic, 0, sizeof(disk_super->magic)); +void btrfs_scratch_superblocks(struct btrfs_fs_info *fs_info, struct btrfs_device *device) +{ + int copy_num; + struct block_device *bdev = device->bdev; - page = virt_to_page(disk_super); - set_page_dirty(page); - lock_page(page); - /* write_on_page() unlocks the page */ - ret = write_one_page(page); - if (ret) - btrfs_warn(fs_info, - "error clearing superblock number %d (%d)", - copy_num, ret); - btrfs_release_disk_super(disk_super); + if (!bdev) + return; + for (copy_num = 0; copy_num < BTRFS_SUPER_MIRROR_MAX; copy_num++) { + if (bdev_is_zoned(bdev)) + btrfs_reset_sb_log_zones(bdev, copy_num); + else + btrfs_scratch_superblock(fs_info, bdev, copy_num); } /* Notify udev that device has changed */ btrfs_kobject_uevent(bdev, KOBJ_CHANGE); /* Update ctime/mtime for device path for libblkid */ - update_dev_time(device_path); + update_dev_time(rcu_dereference_raw(device->name)); } -int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, - u64 devid) +int btrfs_rm_device(struct btrfs_fs_info *fs_info, + struct btrfs_dev_lookup_args *args, + struct file **bdev_file) { + struct btrfs_trans_handle *trans; struct btrfs_device *device; struct btrfs_fs_devices *cur_devices; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; u64 num_devices; int ret = 0; - mutex_lock(&uuid_mutex); + if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) { + btrfs_err(fs_info, "device remove not supported on extent tree v2 yet"); + return -EINVAL; + } + /* + * The device list in fs_devices is accessed without locks (neither + * uuid_mutex nor device_list_mutex) as it won't change on a mounted + * filesystem and another device rm cannot run. + */ num_devices = btrfs_num_devices(fs_info); ret = btrfs_check_raid_min_devices(fs_info, num_devices - 1); if (ret) - goto out; - - device = btrfs_find_device_by_devspec(fs_info, devid, device_path); + return ret; - if (IS_ERR(device)) { - if (PTR_ERR(device) == -ENOENT && - strcmp(device_path, "missing") == 0) + device = btrfs_find_device(fs_info->fs_devices, args); + if (!device) { + if (args->missing) ret = BTRFS_ERROR_DEV_MISSING_NOT_FOUND; else - ret = PTR_ERR(device); - goto out; + ret = -ENOENT; + return ret; } if (btrfs_pinned_by_swapfile(fs_info, device)) { - btrfs_warn_in_rcu(fs_info, + btrfs_warn(fs_info, "cannot remove device %s (devid %llu) due to active swapfile", - rcu_str_deref(device->name), device->devid); - ret = -ETXTBSY; - goto out; + btrfs_dev_name(device), device->devid); + return -ETXTBSY; } - if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { - ret = BTRFS_ERROR_DEV_TGT_REPLACE; - goto out; - } + if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) + return BTRFS_ERROR_DEV_TGT_REPLACE; if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) && - fs_info->fs_devices->rw_devices == 1) { - ret = BTRFS_ERROR_DEV_ONLY_WRITABLE; - goto out; - } + fs_info->fs_devices->rw_devices == 1) + return BTRFS_ERROR_DEV_ONLY_WRITABLE; if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { mutex_lock(&fs_info->chunk_mutex); @@ -2077,20 +2214,26 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, mutex_unlock(&fs_info->chunk_mutex); } - mutex_unlock(&uuid_mutex); ret = btrfs_shrink_device(device, 0); - mutex_lock(&uuid_mutex); if (ret) goto error_undo; - /* - * TODO: the superblock still includes this device in its num_devices - * counter although write_all_supers() is not locked out. This - * could give a filesystem state which requires a degraded mount. - */ - ret = btrfs_rm_dev_item(device); - if (ret) + trans = btrfs_start_transaction(fs_info->chunk_root, 0); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); goto error_undo; + } + + ret = btrfs_rm_dev_item(trans, device); + if (unlikely(ret)) { + /* Any error in dev item removal is critical */ + btrfs_crit(fs_info, + "failed to remove device item for devid %llu: %d", + device->devid, ret); + btrfs_abort_transaction(trans, ret); + btrfs_end_transaction(trans); + return ret; + } clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state); btrfs_scrub_cancel_dev(device); @@ -2108,7 +2251,7 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, /* * In normal cases the cur_devices == fs_devices. But in case * of deleting a seed device, the cur_devices should point to - * its own fs_devices listed under the fs_devices->seed. + * its own fs_devices listed under the fs_devices->seed_list. */ cur_devices = device->fs_devices; mutex_lock(&fs_devices->device_list_mutex); @@ -2125,10 +2268,10 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, btrfs_assign_next_active_device(device, NULL); - if (device->bdev) { + if (device->bdev_file) { cur_devices->open_devices--; /* remove sysfs entry */ - btrfs_sysfs_remove_devices_dir(fs_devices, device); + btrfs_sysfs_remove_device(device); } num_devices = btrfs_super_num_devices(fs_info->super_copy) - 1; @@ -2136,33 +2279,43 @@ int btrfs_rm_device(struct btrfs_fs_info *fs_info, const char *device_path, mutex_unlock(&fs_devices->device_list_mutex); /* - * at this point, the device is zero sized and detached from - * the devices list. All that's left is to zero out the old - * supers and free the device. + * At this point, the device is zero sized and detached from the + * devices list. All that's left is to zero out the old supers and + * free the device. + * + * We cannot call btrfs_close_bdev() here because we're holding the sb + * write lock, and bdev_fput() on the block device will pull in the + * ->open_mutex on the block device and it's dependencies. Instead + * just flush the device and let the caller do the final bdev_release. */ - if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) - btrfs_scratch_superblocks(fs_info, device->bdev, - device->name->str); + if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { + btrfs_scratch_superblocks(fs_info, device); + if (device->bdev) { + sync_blockdev(device->bdev); + invalidate_bdev(device->bdev); + } + } - btrfs_close_bdev(device); + *bdev_file = device->bdev_file; synchronize_rcu(); btrfs_free_device(device); - if (cur_devices->open_devices == 0) { - while (fs_devices) { - if (fs_devices->seed == cur_devices) { - fs_devices->seed = cur_devices->seed; - break; - } - fs_devices = fs_devices->seed; - } - cur_devices->seed = NULL; - close_fs_devices(cur_devices); + /* + * This can happen if cur_devices is the private seed devices list. We + * cannot call close_fs_devices() here because it expects the uuid_mutex + * to be held, but in fact we don't need that for the private + * seed_devices, we can simply decrement cur_devices->opened and then + * remove it from our list and free the fs_devices. + */ + if (cur_devices->num_devices == 0) { + list_del_init(&cur_devices->seed_list); + ASSERT(cur_devices->opened == 1, "opened=%d", cur_devices->opened); + cur_devices->opened--; free_fs_devices(cur_devices); } -out: - mutex_unlock(&uuid_mutex); + ret = btrfs_commit_transaction(trans); + return ret; error_undo: @@ -2173,7 +2326,7 @@ error_undo: device->fs_devices->rw_devices++; mutex_unlock(&fs_info->chunk_mutex); } - goto out; + return ret; } void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev) @@ -2205,14 +2358,9 @@ void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_device *srcdev) void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev) { - struct btrfs_fs_info *fs_info = srcdev->fs_info; struct btrfs_fs_devices *fs_devices = srcdev->fs_devices; - if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &srcdev->dev_state)) { - /* zero out the old super if it is writable */ - btrfs_scratch_superblocks(fs_info, srcdev->bdev, - srcdev->name->str); - } + mutex_lock(&uuid_mutex); btrfs_close_bdev(srcdev); synchronize_rcu(); @@ -2220,8 +2368,6 @@ void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev) /* if this is no devs we rather delete the fs_devices */ if (!fs_devices->num_devices) { - struct btrfs_fs_devices *tmp_fs_devices; - /* * On a mounted FS, num_devices can't be zero unless it's a * seed. In case of a seed device being replaced, the replace @@ -2230,18 +2376,11 @@ void btrfs_rm_dev_replace_free_srcdev(struct btrfs_device *srcdev) */ ASSERT(fs_devices->seeding); - tmp_fs_devices = fs_info->fs_devices; - while (tmp_fs_devices) { - if (tmp_fs_devices->seed == fs_devices) { - tmp_fs_devices->seed = fs_devices->seed; - break; - } - tmp_fs_devices = tmp_fs_devices->seed; - } - fs_devices->seed = NULL; + list_del_init(&fs_devices->seed_list); close_fs_devices(fs_devices); free_fs_devices(fs_devices); } + mutex_unlock(&uuid_mutex); } void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev) @@ -2250,7 +2389,7 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev) mutex_lock(&fs_devices->device_list_mutex); - btrfs_sysfs_remove_devices_dir(fs_devices, tgtdev); + btrfs_sysfs_remove_device(tgtdev); if (tgtdev->bdev) fs_devices->open_devices--; @@ -2263,110 +2402,138 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_device *tgtdev) mutex_unlock(&fs_devices->device_list_mutex); - /* - * The update_dev_time() with in btrfs_scratch_superblocks() - * may lead to a call to btrfs_show_devname() which will try - * to hold device_list_mutex. And here this device - * is already out of device list, so we don't have to hold - * the device_list_mutex lock. - */ - btrfs_scratch_superblocks(tgtdev->fs_info, tgtdev->bdev, - tgtdev->name->str); + btrfs_scratch_superblocks(tgtdev->fs_info, tgtdev); btrfs_close_bdev(tgtdev); synchronize_rcu(); btrfs_free_device(tgtdev); } -static struct btrfs_device *btrfs_find_device_by_path( - struct btrfs_fs_info *fs_info, const char *device_path) +/* + * Populate args from device at path. + * + * @fs_info: the filesystem + * @args: the args to populate + * @path: the path to the device + * + * This will read the super block of the device at @path and populate @args with + * the devid, fsid, and uuid. This is meant to be used for ioctls that need to + * lookup a device to operate on, but need to do it before we take any locks. + * This properly handles the special case of "missing" that a user may pass in, + * and does some basic sanity checks. The caller must make sure that @path is + * properly NUL terminated before calling in, and must call + * btrfs_put_dev_args_from_path() in order to free up the temporary fsid and + * uuid buffers. + * + * Return: 0 for success, -errno for failure + */ +int btrfs_get_dev_args_from_path(struct btrfs_fs_info *fs_info, + struct btrfs_dev_lookup_args *args, + const char *path) { - int ret = 0; struct btrfs_super_block *disk_super; - u64 devid; - u8 *dev_uuid; - struct block_device *bdev; - struct btrfs_device *device; + struct file *bdev_file; + int ret; - ret = btrfs_get_bdev_and_sb(device_path, FMODE_READ, - fs_info->bdev_holder, 0, &bdev, &disk_super); - if (ret) - return ERR_PTR(ret); + if (!path || !path[0]) + return -EINVAL; + if (!strcmp(path, "missing")) { + args->missing = true; + return 0; + } - devid = btrfs_stack_device_id(&disk_super->dev_item); - dev_uuid = disk_super->dev_item.uuid; + args->uuid = kzalloc(BTRFS_UUID_SIZE, GFP_KERNEL); + args->fsid = kzalloc(BTRFS_FSID_SIZE, GFP_KERNEL); + if (!args->uuid || !args->fsid) { + btrfs_put_dev_args_from_path(args); + return -ENOMEM; + } + + ret = btrfs_get_bdev_and_sb(path, BLK_OPEN_READ, NULL, 0, + &bdev_file, &disk_super); + if (ret) { + btrfs_put_dev_args_from_path(args); + return ret; + } + + args->devid = btrfs_stack_device_id(&disk_super->dev_item); + memcpy(args->uuid, disk_super->dev_item.uuid, BTRFS_UUID_SIZE); if (btrfs_fs_incompat(fs_info, METADATA_UUID)) - device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid, - disk_super->metadata_uuid, true); + memcpy(args->fsid, disk_super->metadata_uuid, BTRFS_FSID_SIZE); else - device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid, - disk_super->fsid, true); - + memcpy(args->fsid, disk_super->fsid, BTRFS_FSID_SIZE); btrfs_release_disk_super(disk_super); - if (!device) - device = ERR_PTR(-ENOENT); - blkdev_put(bdev, FMODE_READ); - return device; + bdev_fput(bdev_file); + return 0; } /* - * Lookup a device given by device id, or the path if the id is 0. + * Only use this jointly with btrfs_get_dev_args_from_path() because we will + * allocate our ->uuid and ->fsid pointers, everybody else uses local variables + * that don't need to be freed. */ +void btrfs_put_dev_args_from_path(struct btrfs_dev_lookup_args *args) +{ + kfree(args->uuid); + kfree(args->fsid); + args->uuid = NULL; + args->fsid = NULL; +} + struct btrfs_device *btrfs_find_device_by_devspec( struct btrfs_fs_info *fs_info, u64 devid, const char *device_path) { + BTRFS_DEV_LOOKUP_ARGS(args); struct btrfs_device *device; + int ret; if (devid) { - device = btrfs_find_device(fs_info->fs_devices, devid, NULL, - NULL, true); + args.devid = devid; + device = btrfs_find_device(fs_info->fs_devices, &args); if (!device) return ERR_PTR(-ENOENT); return device; } - if (!device_path || !device_path[0]) - return ERR_PTR(-EINVAL); - - if (strcmp(device_path, "missing") == 0) { - /* Find first missing device */ - list_for_each_entry(device, &fs_info->fs_devices->devices, - dev_list) { - if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, - &device->dev_state) && !device->bdev) - return device; - } + ret = btrfs_get_dev_args_from_path(fs_info, &args, device_path); + if (ret) + return ERR_PTR(ret); + device = btrfs_find_device(fs_info->fs_devices, &args); + btrfs_put_dev_args_from_path(&args); + if (!device) return ERR_PTR(-ENOENT); - } - - return btrfs_find_device_by_path(fs_info, device_path); + return device; } -/* - * does all the dirty work required for changing file system's UUID. - */ -static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info) +static struct btrfs_fs_devices *btrfs_init_sprout(struct btrfs_fs_info *fs_info) { struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; struct btrfs_fs_devices *old_devices; struct btrfs_fs_devices *seed_devices; - struct btrfs_super_block *disk_super = fs_info->super_copy; - struct btrfs_device *device; - u64 super_flags; lockdep_assert_held(&uuid_mutex); if (!fs_devices->seeding) - return -EINVAL; + return ERR_PTR(-EINVAL); - seed_devices = alloc_fs_devices(NULL, NULL); + /* + * Private copy of the seed devices, anchored at + * fs_info->fs_devices->seed_list + */ + seed_devices = alloc_fs_devices(NULL); if (IS_ERR(seed_devices)) - return PTR_ERR(seed_devices); + return seed_devices; + /* + * It's necessary to retain a copy of the original seed fs_devices in + * fs_uuids so that filesystems which have been seeded can successfully + * reference the seed device from open_seed_devices. This also supports + * multiple fs seed. + */ old_devices = clone_fs_devices(fs_devices); if (IS_ERR(old_devices)) { kfree(seed_devices); - return PTR_ERR(old_devices); + return old_devices; } list_add(&old_devices->fs_list, &fs_uuids); @@ -2377,33 +2544,60 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info) INIT_LIST_HEAD(&seed_devices->alloc_list); mutex_init(&seed_devices->device_list_mutex); - mutex_lock(&fs_devices->device_list_mutex); + return seed_devices; +} + +/* + * Splice seed devices into the sprout fs_devices. + * Generate a new fsid for the sprouted read-write filesystem. + */ +static void btrfs_setup_sprout(struct btrfs_fs_info *fs_info, + struct btrfs_fs_devices *seed_devices) +{ + struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct btrfs_super_block *disk_super = fs_info->super_copy; + struct btrfs_device *device; + u64 super_flags; + + /* + * We are updating the fsid, the thread leading to device_list_add() + * could race, so uuid_mutex is needed. + */ + lockdep_assert_held(&uuid_mutex); + + /* + * The threads listed below may traverse dev_list but can do that without + * device_list_mutex: + * - All device ops and balance - as we are in btrfs_exclop_start. + * - Various dev_list readers - are using RCU. + * - btrfs_ioctl_fitrim() - is using RCU. + * + * For-read threads as below are using device_list_mutex: + * - Readonly scrub btrfs_scrub_dev() + * - Readonly scrub btrfs_scrub_progress() + * - btrfs_get_dev_stats() + */ + lockdep_assert_held(&fs_devices->device_list_mutex); + list_splice_init_rcu(&fs_devices->devices, &seed_devices->devices, synchronize_rcu); list_for_each_entry(device, &seed_devices->devices, dev_list) device->fs_devices = seed_devices; - mutex_lock(&fs_info->chunk_mutex); - list_splice_init(&fs_devices->alloc_list, &seed_devices->alloc_list); - mutex_unlock(&fs_info->chunk_mutex); - fs_devices->seeding = false; fs_devices->num_devices = 0; fs_devices->open_devices = 0; fs_devices->missing_devices = 0; fs_devices->rotating = false; - fs_devices->seed = seed_devices; + list_add(&seed_devices->seed_list, &fs_devices->seed_list); generate_random_uuid(fs_devices->fsid); memcpy(fs_devices->metadata_uuid, fs_devices->fsid, BTRFS_FSID_SIZE); memcpy(disk_super->fsid, fs_devices->fsid, BTRFS_FSID_SIZE); - mutex_unlock(&fs_devices->device_list_mutex); super_flags = btrfs_super_flags(disk_super) & ~BTRFS_SUPER_FLAG_SEEDING; btrfs_set_super_flags(disk_super, super_flags); - - return 0; } /* @@ -2411,16 +2605,16 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info) */ static int btrfs_finish_sprout(struct btrfs_trans_handle *trans) { + BTRFS_DEV_LOOKUP_ARGS(args); struct btrfs_fs_info *fs_info = trans->fs_info; struct btrfs_root *root = fs_info->chunk_root; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct extent_buffer *leaf; struct btrfs_dev_item *dev_item; struct btrfs_device *device; struct btrfs_key key; u8 fs_uuid[BTRFS_FSID_SIZE]; u8 dev_uuid[BTRFS_UUID_SIZE]; - u64 devid; int ret; path = btrfs_alloc_path(); @@ -2428,13 +2622,15 @@ static int btrfs_finish_sprout(struct btrfs_trans_handle *trans) return -ENOMEM; key.objectid = BTRFS_DEV_ITEMS_OBJECTID; - key.offset = 0; key.type = BTRFS_DEV_ITEM_KEY; + key.offset = 0; while (1) { + btrfs_reserve_chunk_metadata(trans, false); ret = btrfs_search_slot(trans, root, &key, path, 0, 1); + btrfs_trans_release_chunk_metadata(trans); if (ret < 0) - goto error; + return ret; leaf = path->nodes[0]; next_slot: @@ -2443,7 +2639,7 @@ next_slot: if (ret > 0) break; if (ret < 0) - goto error; + return ret; leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); btrfs_release_path(path); @@ -2457,123 +2653,135 @@ next_slot: dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item); - devid = btrfs_device_id(leaf, dev_item); + args.devid = btrfs_device_id(leaf, dev_item); read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item), BTRFS_UUID_SIZE); read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item), BTRFS_FSID_SIZE); - device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid, - fs_uuid, true); + args.uuid = dev_uuid; + args.fsid = fs_uuid; + device = btrfs_find_device(fs_info->fs_devices, &args); BUG_ON(!device); /* Logic error */ - if (device->fs_devices->seeding) { + if (device->fs_devices->seeding) btrfs_set_device_generation(leaf, dev_item, device->generation); - btrfs_mark_buffer_dirty(leaf); - } path->slots[0]++; goto next_slot; } - ret = 0; -error: - btrfs_free_path(path); - return ret; + return 0; } int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path) { struct btrfs_root *root = fs_info->dev_root; - struct request_queue *q; struct btrfs_trans_handle *trans; struct btrfs_device *device; - struct block_device *bdev; + struct file *bdev_file; struct super_block *sb = fs_info->sb; - struct rcu_string *name; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct btrfs_fs_devices *seed_devices = NULL; u64 orig_super_total_bytes; u64 orig_super_num_devices; - int seeding_dev = 0; int ret = 0; - bool unlocked = false; + bool seeding_dev = false; + bool locked = false; if (sb_rdonly(sb) && !fs_devices->seeding) return -EROFS; - bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL, - fs_info->bdev_holder); - if (IS_ERR(bdev)) - return PTR_ERR(bdev); + bdev_file = bdev_file_open_by_path(device_path, BLK_OPEN_WRITE, + fs_info->sb, &fs_holder_ops); + if (IS_ERR(bdev_file)) + return PTR_ERR(bdev_file); + + if (!btrfs_check_device_zone_type(fs_info, file_bdev(bdev_file))) { + ret = -EINVAL; + goto error; + } + + if (bdev_nr_bytes(file_bdev(bdev_file)) <= BTRFS_DEVICE_RANGE_RESERVED) { + ret = -EINVAL; + goto error; + } if (fs_devices->seeding) { - seeding_dev = 1; + seeding_dev = true; down_write(&sb->s_umount); mutex_lock(&uuid_mutex); + locked = true; } - filemap_write_and_wait(bdev->bd_inode->i_mapping); + sync_blockdev(file_bdev(bdev_file)); - mutex_lock(&fs_devices->device_list_mutex); - list_for_each_entry(device, &fs_devices->devices, dev_list) { - if (device->bdev == bdev) { + rcu_read_lock(); + list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { + if (device->bdev == file_bdev(bdev_file)) { ret = -EEXIST; - mutex_unlock( - &fs_devices->device_list_mutex); + rcu_read_unlock(); goto error; } } - mutex_unlock(&fs_devices->device_list_mutex); + rcu_read_unlock(); - device = btrfs_alloc_device(fs_info, NULL, NULL); + device = btrfs_alloc_device(fs_info, NULL, NULL, device_path); if (IS_ERR(device)) { /* we can safely leave the fs_devices entry around */ ret = PTR_ERR(device); goto error; } - name = rcu_string_strdup(device_path, GFP_KERNEL); - if (!name) { - ret = -ENOMEM; + device->fs_info = fs_info; + device->bdev_file = bdev_file; + device->bdev = file_bdev(bdev_file); + ret = lookup_bdev(device_path, &device->devt); + if (ret) + goto error_free_device; + + ret = btrfs_get_dev_zone_info(device, false); + if (ret) goto error_free_device; - } - rcu_assign_pointer(device->name, name); trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) { ret = PTR_ERR(trans); - goto error_free_device; + goto error_free_zone; } - q = bdev_get_queue(bdev); set_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); device->generation = trans->transid; device->io_width = fs_info->sectorsize; device->io_align = fs_info->sectorsize; device->sector_size = fs_info->sectorsize; - device->total_bytes = round_down(i_size_read(bdev->bd_inode), - fs_info->sectorsize); + device->total_bytes = + round_down(bdev_nr_bytes(device->bdev), fs_info->sectorsize); device->disk_total_bytes = device->total_bytes; device->commit_total_bytes = device->total_bytes; - device->fs_info = fs_info; - device->bdev = bdev; set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state); clear_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state); - device->mode = FMODE_EXCL; device->dev_stats_valid = 1; - set_blocksize(device->bdev, BTRFS_BDEV_BLOCKSIZE); + set_blocksize(device->bdev_file, BTRFS_BDEV_BLOCKSIZE); if (seeding_dev) { - sb->s_flags &= ~SB_RDONLY; - ret = btrfs_prepare_sprout(fs_info); - if (ret) { + /* GFP_KERNEL allocation must not be under device_list_mutex */ + seed_devices = btrfs_init_sprout(fs_info); + if (IS_ERR(seed_devices)) { + ret = PTR_ERR(seed_devices); btrfs_abort_transaction(trans, ret); goto error_trans; } } + mutex_lock(&fs_devices->device_list_mutex); + if (seeding_dev) { + btrfs_setup_sprout(fs_info, seed_devices); + btrfs_assign_next_active_device(fs_info->fs_devices->latest_dev, + device); + } + device->fs_devices = fs_devices; - mutex_lock(&fs_devices->device_list_mutex); mutex_lock(&fs_info->chunk_mutex); list_add_rcu(&device->dev_list, &fs_devices->devices); list_add(&device->dev_alloc_list, &fs_devices->alloc_list); @@ -2585,7 +2793,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path atomic64_add(device->total_bytes, &fs_info->free_chunk_space); - if (!blk_queue_nonrot(q)) + if (!bdev_nonrot(device->bdev)) fs_devices->rotating = true; orig_super_total_bytes = btrfs_super_total_bytes(fs_info->super_copy); @@ -2597,9 +2805,6 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path btrfs_set_super_num_devices(fs_info->super_copy, orig_super_num_devices + 1); - /* add sysfs device entry */ - btrfs_sysfs_add_devices_dir(fs_devices, device); - /* * we've got more storage, clear any full flags on the space * infos @@ -2607,33 +2812,40 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path btrfs_clear_space_info_full(fs_info); mutex_unlock(&fs_info->chunk_mutex); + + /* Add sysfs device entry */ + btrfs_sysfs_add_device(device); + mutex_unlock(&fs_devices->device_list_mutex); if (seeding_dev) { mutex_lock(&fs_info->chunk_mutex); ret = init_first_rw_device(trans); mutex_unlock(&fs_info->chunk_mutex); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto error_sysfs; } } ret = btrfs_add_dev_item(trans, device); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto error_sysfs; } if (seeding_dev) { ret = btrfs_finish_sprout(trans); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto error_sysfs; } - btrfs_sysfs_update_sprout_fsid(fs_devices, - fs_info->fs_devices->fsid); + /* + * fs_devices now represents the newly sprouted filesystem and + * its fsid has been changed by btrfs_sprout_splice(). + */ + btrfs_sysfs_update_sprout_fsid(fs_devices); } ret = btrfs_commit_transaction(trans); @@ -2641,7 +2853,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path if (seeding_dev) { mutex_unlock(&uuid_mutex); up_write(&sb->s_umount); - unlocked = true; + locked = false; if (ret) /* transaction commit */ return ret; @@ -2668,7 +2880,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path * We can ignore the return value as it typically returns -EINVAL and * only succeeds if the device was an alien. */ - btrfs_forget_devices(device_path); + btrfs_forget_devices(device->devt); /* Update ctime/mtime for blkid or udev */ update_dev_time(device_path); @@ -2676,7 +2888,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path return ret; error_sysfs: - btrfs_sysfs_remove_devices_dir(fs_devices, device); + btrfs_sysfs_remove_device(device); mutex_lock(&fs_info->fs_devices->device_list_mutex); mutex_lock(&fs_info->chunk_mutex); list_del_rcu(&device->dev_list); @@ -2694,15 +2906,15 @@ error_sysfs: mutex_unlock(&fs_info->chunk_mutex); mutex_unlock(&fs_info->fs_devices->device_list_mutex); error_trans: - if (seeding_dev) - sb->s_flags |= SB_RDONLY; if (trans) btrfs_end_transaction(trans); +error_free_zone: + btrfs_destroy_dev_zone_info(device); error_free_device: btrfs_free_device(device); error: - blkdev_put(bdev, FMODE_EXCL); - if (seeding_dev && !unlocked) { + bdev_fput(bdev_file); + if (locked) { mutex_unlock(&uuid_mutex); up_write(&sb->s_umount); } @@ -2713,7 +2925,7 @@ static noinline int btrfs_update_device(struct btrfs_trans_handle *trans, struct btrfs_device *device) { int ret; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_root *root = device->fs_info->chunk_root; struct btrfs_dev_item *dev_item; struct extent_buffer *leaf; @@ -2729,12 +2941,10 @@ static noinline int btrfs_update_device(struct btrfs_trans_handle *trans, ret = btrfs_search_slot(trans, root, &key, path, 0, 1); if (ret < 0) - goto out; + return ret; - if (ret > 0) { - ret = -ENOENT; - goto out; - } + if (ret > 0) + return -ENOENT; leaf = path->nodes[0]; dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item); @@ -2748,10 +2958,6 @@ static noinline int btrfs_update_device(struct btrfs_trans_handle *trans, btrfs_device_get_disk_total_bytes(device)); btrfs_set_device_bytes_used(leaf, dev_item, btrfs_device_get_bytes_used(device)); - btrfs_mark_buffer_dirty(leaf); - -out: - btrfs_free_path(path); return ret; } @@ -2762,6 +2968,7 @@ int btrfs_grow_device(struct btrfs_trans_handle *trans, struct btrfs_super_block *super_copy = fs_info->super_copy; u64 old_total; u64 diff; + int ret; if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) return -EACCES; @@ -2781,6 +2988,7 @@ int btrfs_grow_device(struct btrfs_trans_handle *trans, btrfs_set_super_total_bytes(super_copy, round_down(old_total + diff, fs_info->sectorsize)); device->fs_devices->total_rw_bytes += diff; + atomic64_add(diff, &fs_info->free_chunk_space); btrfs_device_set_total_bytes(device, new_size); btrfs_device_set_disk_total_bytes(device, new_size); @@ -2790,7 +2998,11 @@ int btrfs_grow_device(struct btrfs_trans_handle *trans, &trans->transaction->dev_update_list); mutex_unlock(&fs_info->chunk_mutex); - return btrfs_update_device(trans, device); + btrfs_reserve_chunk_metadata(trans, false); + ret = btrfs_update_device(trans, device); + btrfs_trans_release_chunk_metadata(trans); + + return ret; } static int btrfs_free_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) @@ -2798,7 +3010,7 @@ static int btrfs_free_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) struct btrfs_fs_info *fs_info = trans->fs_info; struct btrfs_root *root = fs_info->chunk_root; int ret; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; path = btrfs_alloc_path(); @@ -2806,25 +3018,26 @@ static int btrfs_free_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) return -ENOMEM; key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID; - key.offset = chunk_offset; key.type = BTRFS_CHUNK_ITEM_KEY; + key.offset = chunk_offset; ret = btrfs_search_slot(trans, root, &key, path, -1, 1); if (ret < 0) - goto out; - else if (ret > 0) { /* Logic error or corruption */ - btrfs_handle_fs_error(fs_info, -ENOENT, - "Failed lookup while freeing chunk."); - ret = -ENOENT; - goto out; + return ret; + if (unlikely(ret > 0)) { + /* Logic error or corruption */ + btrfs_err(fs_info, "failed to lookup chunk %llu when freeing", + chunk_offset); + btrfs_abort_transaction(trans, -ENOENT); + return -EUCLEAN; } ret = btrfs_del_item(trans, root, path); - if (ret < 0) - btrfs_handle_fs_error(fs_info, ret, - "Failed to delete chunk item."); -out: - btrfs_free_path(path); + if (unlikely(ret < 0)) { + btrfs_err(fs_info, "failed to delete chunk %llu item", chunk_offset); + btrfs_abort_transaction(trans, ret); + return ret; + } return ret; } @@ -2841,7 +3054,7 @@ static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset) u32 cur; struct btrfs_key key; - mutex_lock(&fs_info->chunk_mutex); + lockdep_assert_held(&fs_info->chunk_mutex); array_size = btrfs_super_sys_array_size(super_copy); ptr = super_copy->sys_chunk_array; @@ -2871,74 +3084,171 @@ static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset) cur += len; } } - mutex_unlock(&fs_info->chunk_mutex); return ret; } +struct btrfs_chunk_map *btrfs_find_chunk_map_nolock(struct btrfs_fs_info *fs_info, + u64 logical, u64 length) +{ + struct rb_node *node = fs_info->mapping_tree.rb_root.rb_node; + struct rb_node *prev = NULL; + struct rb_node *orig_prev; + struct btrfs_chunk_map *map; + struct btrfs_chunk_map *prev_map = NULL; + + while (node) { + map = rb_entry(node, struct btrfs_chunk_map, rb_node); + prev = node; + prev_map = map; + + if (logical < map->start) { + node = node->rb_left; + } else if (logical >= map->start + map->chunk_len) { + node = node->rb_right; + } else { + refcount_inc(&map->refs); + return map; + } + } + + if (!prev) + return NULL; + + orig_prev = prev; + while (prev && logical >= prev_map->start + prev_map->chunk_len) { + prev = rb_next(prev); + prev_map = rb_entry(prev, struct btrfs_chunk_map, rb_node); + } + + if (!prev) { + prev = orig_prev; + prev_map = rb_entry(prev, struct btrfs_chunk_map, rb_node); + while (prev && logical < prev_map->start) { + prev = rb_prev(prev); + prev_map = rb_entry(prev, struct btrfs_chunk_map, rb_node); + } + } + + if (prev) { + u64 end = logical + length; + + /* + * Caller can pass a U64_MAX length when it wants to get any + * chunk starting at an offset of 'logical' or higher, so deal + * with underflow by resetting the end offset to U64_MAX. + */ + if (end < logical) + end = U64_MAX; + + if (end > prev_map->start && + logical < prev_map->start + prev_map->chunk_len) { + refcount_inc(&prev_map->refs); + return prev_map; + } + } + + return NULL; +} + +struct btrfs_chunk_map *btrfs_find_chunk_map(struct btrfs_fs_info *fs_info, + u64 logical, u64 length) +{ + struct btrfs_chunk_map *map; + + read_lock(&fs_info->mapping_tree_lock); + map = btrfs_find_chunk_map_nolock(fs_info, logical, length); + read_unlock(&fs_info->mapping_tree_lock); + + return map; +} + /* - * btrfs_get_chunk_map() - Find the mapping containing the given logical extent. + * Find the mapping containing the given logical extent. + * * @logical: Logical block offset in bytes. * @length: Length of extent in bytes. * * Return: Chunk mapping or ERR_PTR. */ -struct extent_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info, - u64 logical, u64 length) +struct btrfs_chunk_map *btrfs_get_chunk_map(struct btrfs_fs_info *fs_info, + u64 logical, u64 length) { - struct extent_map_tree *em_tree; - struct extent_map *em; + struct btrfs_chunk_map *map; - em_tree = &fs_info->mapping_tree; - read_lock(&em_tree->lock); - em = lookup_extent_mapping(em_tree, logical, length); - read_unlock(&em_tree->lock); + map = btrfs_find_chunk_map(fs_info, logical, length); - if (!em) { - btrfs_crit(fs_info, "unable to find logical %llu length %llu", + if (unlikely(!map)) { + btrfs_crit(fs_info, + "unable to find chunk map for logical %llu length %llu", logical, length); return ERR_PTR(-EINVAL); } - if (em->start > logical || em->start + em->len < logical) { + if (unlikely(map->start > logical || map->start + map->chunk_len <= logical)) { btrfs_crit(fs_info, - "found a bad mapping, wanted %llu-%llu, found %llu-%llu", - logical, length, em->start, em->start + em->len); - free_extent_map(em); + "found a bad chunk map, wanted %llu-%llu, found %llu-%llu", + logical, logical + length, map->start, + map->start + map->chunk_len); + btrfs_free_chunk_map(map); return ERR_PTR(-EINVAL); } - /* callers are responsible for dropping em's ref. */ - return em; + /* Callers are responsible for dropping the reference. */ + return map; +} + +static int remove_chunk_item(struct btrfs_trans_handle *trans, + struct btrfs_chunk_map *map, u64 chunk_offset) +{ + int i; + + /* + * Removing chunk items and updating the device items in the chunks btree + * requires holding the chunk_mutex. + * See the comment at btrfs_chunk_alloc() for the details. + */ + lockdep_assert_held(&trans->fs_info->chunk_mutex); + + for (i = 0; i < map->num_stripes; i++) { + int ret; + + ret = btrfs_update_device(trans, map->stripes[i].dev); + if (ret) + return ret; + } + + return btrfs_free_chunk(trans, chunk_offset); } int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) { struct btrfs_fs_info *fs_info = trans->fs_info; - struct extent_map *em; - struct map_lookup *map; + struct btrfs_chunk_map *map; u64 dev_extent_len = 0; int i, ret = 0; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; - em = btrfs_get_chunk_map(fs_info, chunk_offset, 1); - if (IS_ERR(em)) { + map = btrfs_get_chunk_map(fs_info, chunk_offset, 1); + if (IS_ERR(map)) { /* * This is a logic error, but we don't want to just rely on the * user having built with ASSERT enabled, so if ASSERT doesn't * do anything we still error out. */ - ASSERT(0); - return PTR_ERR(em); + DEBUG_WARN("errr %ld reading chunk map at offset %llu", + PTR_ERR(map), chunk_offset); + return PTR_ERR(map); } - map = em->map_lookup; - mutex_lock(&fs_info->chunk_mutex); - check_system_chunk(trans, map->type); - mutex_unlock(&fs_info->chunk_mutex); /* - * Take the device list mutex to prevent races with the final phase of - * a device replace operation that replaces the device object associated - * with map stripes (dev-replace.c:btrfs_dev_replace_finishing()). + * First delete the device extent items from the devices btree. + * We take the device_list_mutex to avoid racing with the finishing phase + * of a device replace operation. See the comment below before acquiring + * fs_info->chunk_mutex. Note that here we do not acquire the chunk_mutex + * because that can result in a deadlock when deleting the device extent + * items from the devices btree - COWing an extent buffer from the btree + * may result in allocating a new metadata chunk, which would attempt to + * lock again fs_info->chunk_mutex. */ mutex_lock(&fs_devices->device_list_mutex); for (i = 0; i < map->num_stripes; i++) { @@ -2946,7 +3256,7 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) ret = btrfs_free_dev_extent(trans, device, map->stripes[i].physical, &dev_extent_len); - if (ret) { + if (unlikely(ret)) { mutex_unlock(&fs_devices->device_list_mutex); btrfs_abort_transaction(trans, ret); goto out; @@ -2958,53 +3268,143 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans, u64 chunk_offset) device->bytes_used - dev_extent_len); atomic64_add(dev_extent_len, &fs_info->free_chunk_space); btrfs_clear_space_info_full(fs_info); + + if (list_empty(&device->post_commit_list)) { + list_add_tail(&device->post_commit_list, + &trans->transaction->dev_update_list); + } + mutex_unlock(&fs_info->chunk_mutex); } + } + mutex_unlock(&fs_devices->device_list_mutex); - ret = btrfs_update_device(trans, device); - if (ret) { - mutex_unlock(&fs_devices->device_list_mutex); + /* + * We acquire fs_info->chunk_mutex for 2 reasons: + * + * 1) Just like with the first phase of the chunk allocation, we must + * reserve system space, do all chunk btree updates and deletions, and + * update the system chunk array in the superblock while holding this + * mutex. This is for similar reasons as explained on the comment at + * the top of btrfs_chunk_alloc(); + * + * 2) Prevent races with the final phase of a device replace operation + * that replaces the device object associated with the map's stripes, + * because the device object's id can change at any time during that + * final phase of the device replace operation + * (dev-replace.c:btrfs_dev_replace_finishing()), so we could grab the + * replaced device and then see it with an ID of + * BTRFS_DEV_REPLACE_DEVID, which would cause a failure when updating + * the device item, which does not exists on the chunk btree. + * The finishing phase of device replace acquires both the + * device_list_mutex and the chunk_mutex, in that order, so we are + * safe by just acquiring the chunk_mutex. + */ + trans->removing_chunk = true; + mutex_lock(&fs_info->chunk_mutex); + + check_system_chunk(trans, map->type); + + ret = remove_chunk_item(trans, map, chunk_offset); + /* + * Normally we should not get -ENOSPC since we reserved space before + * through the call to check_system_chunk(). + * + * Despite our system space_info having enough free space, we may not + * be able to allocate extents from its block groups, because all have + * an incompatible profile, which will force us to allocate a new system + * block group with the right profile, or right after we called + * check_system_space() above, a scrub turned the only system block group + * with enough free space into RO mode. + * This is explained with more detail at do_chunk_alloc(). + * + * So if we get -ENOSPC, allocate a new system chunk and retry once. + */ + if (ret == -ENOSPC) { + const u64 sys_flags = btrfs_system_alloc_profile(fs_info); + struct btrfs_block_group *sys_bg; + struct btrfs_space_info *space_info; + + space_info = btrfs_find_space_info(fs_info, sys_flags); + if (unlikely(!space_info)) { + ret = -EINVAL; btrfs_abort_transaction(trans, ret); goto out; } - } - mutex_unlock(&fs_devices->device_list_mutex); - ret = btrfs_free_chunk(trans, chunk_offset); - if (ret) { + sys_bg = btrfs_create_chunk(trans, space_info, sys_flags); + if (IS_ERR(sys_bg)) { + ret = PTR_ERR(sys_bg); + btrfs_abort_transaction(trans, ret); + goto out; + } + + ret = btrfs_chunk_alloc_add_chunk_item(trans, sys_bg); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out; + } + + ret = remove_chunk_item(trans, map, chunk_offset); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out; + } + } else if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto out; } - trace_btrfs_chunk_free(fs_info, map, chunk_offset, em->len); + trace_btrfs_chunk_free(fs_info, map, chunk_offset, map->chunk_len); if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) { ret = btrfs_del_sys_chunk(fs_info, chunk_offset); - if (ret) { + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto out; } } - ret = btrfs_remove_block_group(trans, chunk_offset, em); - if (ret) { + mutex_unlock(&fs_info->chunk_mutex); + trans->removing_chunk = false; + + /* + * We are done with chunk btree updates and deletions, so release the + * system space we previously reserved (with check_system_chunk()). + */ + btrfs_trans_release_chunk_metadata(trans); + + ret = btrfs_remove_block_group(trans, map); + if (unlikely(ret)) { btrfs_abort_transaction(trans, ret); goto out; } out: + if (trans->removing_chunk) { + mutex_unlock(&fs_info->chunk_mutex); + trans->removing_chunk = false; + } /* once for us */ - free_extent_map(em); + btrfs_free_chunk_map(map); return ret; } -static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset) +int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset, + bool verbose) { struct btrfs_root *root = fs_info->chunk_root; struct btrfs_trans_handle *trans; struct btrfs_block_group *block_group; + u64 length; int ret; + if (btrfs_fs_incompat(fs_info, EXTENT_TREE_V2)) { + btrfs_err(fs_info, + "relocate: not supported on extent tree v2 yet"); + return -EINVAL; + } + /* * Prevent races with automatic removal of unused block groups. * After we relocate and before we remove the chunk with offset @@ -3017,21 +3417,43 @@ static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset) * we release the path used to search the chunk/dev tree and before * the current task acquires this mutex and calls us. */ - lockdep_assert_held(&fs_info->delete_unused_bgs_mutex); + lockdep_assert_held(&fs_info->reclaim_bgs_lock); /* step one, relocate all the extents inside this chunk */ btrfs_scrub_pause(fs_info); - ret = btrfs_relocate_block_group(fs_info, chunk_offset); + ret = btrfs_relocate_block_group(fs_info, chunk_offset, true); btrfs_scrub_continue(fs_info); - if (ret) + if (ret) { + /* + * If we had a transaction abort, stop all running scrubs. + * See transaction.c:cleanup_transaction() why we do it here. + */ + if (BTRFS_FS_ERROR(fs_info)) + btrfs_scrub_cancel(fs_info); return ret; + } block_group = btrfs_lookup_block_group(fs_info, chunk_offset); if (!block_group) return -ENOENT; btrfs_discard_cancel_work(&fs_info->discard_ctl, block_group); + length = block_group->length; btrfs_put_block_group(block_group); + /* + * On a zoned file system, discard the whole block group, this will + * trigger a REQ_OP_ZONE_RESET operation on the device zone. If + * resetting the zone fails, don't treat it as a fatal problem from the + * filesystem's point of view. + */ + if (btrfs_is_zoned(fs_info)) { + ret = btrfs_discard_extent(fs_info, chunk_offset, length, NULL); + if (ret) + btrfs_info(fs_info, + "failed to reset zone %llu after relocation", + chunk_offset); + } + trans = btrfs_start_trans_remove_block_group(root->fs_info, chunk_offset); if (IS_ERR(trans)) { @@ -3052,7 +3474,7 @@ static int btrfs_relocate_chunk(struct btrfs_fs_info *fs_info, u64 chunk_offset) static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info) { struct btrfs_root *chunk_root = fs_info->chunk_root; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct extent_buffer *leaf; struct btrfs_chunk *chunk; struct btrfs_key key; @@ -3068,24 +3490,34 @@ static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info) again: key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID; - key.offset = (u64)-1; key.type = BTRFS_CHUNK_ITEM_KEY; + key.offset = (u64)-1; while (1) { - mutex_lock(&fs_info->delete_unused_bgs_mutex); + mutex_lock(&fs_info->reclaim_bgs_lock); ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0); if (ret < 0) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); - goto error; + mutex_unlock(&fs_info->reclaim_bgs_lock); + return ret; + } + if (unlikely(ret == 0)) { + /* + * On the first search we would find chunk tree with + * offset -1, which is not possible. On subsequent + * loops this would find an existing item on an invalid + * offset (one less than the previous one, wrong + * alignment and size). + */ + mutex_unlock(&fs_info->reclaim_bgs_lock); + return -EUCLEAN; } - BUG_ON(ret == 0); /* Corruption */ ret = btrfs_previous_item(chunk_root, path, key.objectid, key.type); if (ret) - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); if (ret < 0) - goto error; + return ret; if (ret > 0) break; @@ -3098,13 +3530,14 @@ again: btrfs_release_path(path); if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) { - ret = btrfs_relocate_chunk(fs_info, found_key.offset); + ret = btrfs_relocate_chunk(fs_info, found_key.offset, + true); if (ret == -ENOSPC) failed++; else BUG_ON(ret); } - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); if (found_key.offset == 0) break; @@ -3118,8 +3551,6 @@ again: } else if (WARN_ON(failed && retried)) { ret = -ENOSPC; } -error: - btrfs_free_path(path); return ret; } @@ -3165,6 +3596,44 @@ static int btrfs_may_alloc_data_chunk(struct btrfs_fs_info *fs_info, return 0; } +static void btrfs_disk_balance_args_to_cpu(struct btrfs_balance_args *cpu, + const struct btrfs_disk_balance_args *disk) +{ + memset(cpu, 0, sizeof(*cpu)); + + cpu->profiles = le64_to_cpu(disk->profiles); + cpu->usage = le64_to_cpu(disk->usage); + cpu->devid = le64_to_cpu(disk->devid); + cpu->pstart = le64_to_cpu(disk->pstart); + cpu->pend = le64_to_cpu(disk->pend); + cpu->vstart = le64_to_cpu(disk->vstart); + cpu->vend = le64_to_cpu(disk->vend); + cpu->target = le64_to_cpu(disk->target); + cpu->flags = le64_to_cpu(disk->flags); + cpu->limit = le64_to_cpu(disk->limit); + cpu->stripes_min = le32_to_cpu(disk->stripes_min); + cpu->stripes_max = le32_to_cpu(disk->stripes_max); +} + +static void btrfs_cpu_balance_args_to_disk(struct btrfs_disk_balance_args *disk, + const struct btrfs_balance_args *cpu) +{ + memset(disk, 0, sizeof(*disk)); + + disk->profiles = cpu_to_le64(cpu->profiles); + disk->usage = cpu_to_le64(cpu->usage); + disk->devid = cpu_to_le64(cpu->devid); + disk->pstart = cpu_to_le64(cpu->pstart); + disk->pend = cpu_to_le64(cpu->pend); + disk->vstart = cpu_to_le64(cpu->vstart); + disk->vend = cpu_to_le64(cpu->vend); + disk->target = cpu_to_le64(cpu->target); + disk->flags = cpu_to_le64(cpu->flags); + disk->limit = cpu_to_le64(cpu->limit); + disk->stripes_min = cpu_to_le32(cpu->stripes_min); + disk->stripes_max = cpu_to_le32(cpu->stripes_max); +} + static int insert_balance_item(struct btrfs_fs_info *fs_info, struct btrfs_balance_control *bctl) { @@ -3207,10 +3676,7 @@ static int insert_balance_item(struct btrfs_fs_info *fs_info, btrfs_set_balance_meta(leaf, item, &disk_bargs); btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->sys); btrfs_set_balance_sys(leaf, item, &disk_bargs); - btrfs_set_balance_flags(leaf, item, bctl->flags); - - btrfs_mark_buffer_dirty(leaf); out: btrfs_free_path(path); err = btrfs_commit_transaction(trans); @@ -3231,7 +3697,7 @@ static int del_balance_item(struct btrfs_fs_info *fs_info) if (!path) return -ENOMEM; - trans = btrfs_start_transaction(root, 0); + trans = btrfs_start_transaction_fallback_global_rsv(root, 0); if (IS_ERR(trans)) { btrfs_free_path(path); return PTR_ERR(trans); @@ -3309,7 +3775,7 @@ static void reset_balance_state(struct btrfs_fs_info *fs_info) struct btrfs_balance_control *bctl = fs_info->balance_ctl; int ret; - BUG_ON(!fs_info->balance_ctl); + ASSERT(fs_info->balance_ctl); spin_lock(&fs_info->balance_lock); fs_info->balance_ctl = NULL; @@ -3325,26 +3791,25 @@ static void reset_balance_state(struct btrfs_fs_info *fs_info) * Balance filters. Return 1 if chunk should be filtered out * (should not be balanced). */ -static int chunk_profiles_filter(u64 chunk_type, - struct btrfs_balance_args *bargs) +static bool chunk_profiles_filter(u64 chunk_type, struct btrfs_balance_args *bargs) { chunk_type = chunk_to_extended(chunk_type) & BTRFS_EXTENDED_PROFILE_MASK; if (bargs->profiles & chunk_type) - return 0; + return false; - return 1; + return true; } -static int chunk_usage_range_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset, - struct btrfs_balance_args *bargs) +static bool chunk_usage_range_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset, + struct btrfs_balance_args *bargs) { struct btrfs_block_group *cache; u64 chunk_used; u64 user_thresh_min; u64 user_thresh_max; - int ret = 1; + bool ret = true; cache = btrfs_lookup_block_group(fs_info, chunk_offset); chunk_used = cache->used; @@ -3352,30 +3817,28 @@ static int chunk_usage_range_filter(struct btrfs_fs_info *fs_info, u64 chunk_off if (bargs->usage_min == 0) user_thresh_min = 0; else - user_thresh_min = div_factor_fine(cache->length, - bargs->usage_min); + user_thresh_min = mult_perc(cache->length, bargs->usage_min); if (bargs->usage_max == 0) user_thresh_max = 1; else if (bargs->usage_max > 100) user_thresh_max = cache->length; else - user_thresh_max = div_factor_fine(cache->length, - bargs->usage_max); + user_thresh_max = mult_perc(cache->length, bargs->usage_max); if (user_thresh_min <= chunk_used && chunk_used < user_thresh_max) - ret = 0; + ret = false; btrfs_put_block_group(cache); return ret; } -static int chunk_usage_filter(struct btrfs_fs_info *fs_info, - u64 chunk_offset, struct btrfs_balance_args *bargs) +static bool chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset, + struct btrfs_balance_args *bargs) { struct btrfs_block_group *cache; u64 chunk_used, user_thresh; - int ret = 1; + bool ret = true; cache = btrfs_lookup_block_group(fs_info, chunk_offset); chunk_used = cache->used; @@ -3385,18 +3848,17 @@ static int chunk_usage_filter(struct btrfs_fs_info *fs_info, else if (bargs->usage > 100) user_thresh = cache->length; else - user_thresh = div_factor_fine(cache->length, bargs->usage); + user_thresh = mult_perc(cache->length, bargs->usage); if (chunk_used < user_thresh) - ret = 0; + ret = false; btrfs_put_block_group(cache); return ret; } -static int chunk_devid_filter(struct extent_buffer *leaf, - struct btrfs_chunk *chunk, - struct btrfs_balance_args *bargs) +static bool chunk_devid_filter(struct extent_buffer *leaf, struct btrfs_chunk *chunk, + struct btrfs_balance_args *bargs) { struct btrfs_stripe *stripe; int num_stripes = btrfs_chunk_num_stripes(leaf, chunk); @@ -3405,10 +3867,10 @@ static int chunk_devid_filter(struct extent_buffer *leaf, for (i = 0; i < num_stripes; i++) { stripe = btrfs_stripe_nr(chunk, i); if (btrfs_stripe_devid(leaf, stripe) == bargs->devid) - return 0; + return false; } - return 1; + return true; } static u64 calc_data_stripes(u64 type, int num_stripes) @@ -3417,16 +3879,12 @@ static u64 calc_data_stripes(u64 type, int num_stripes) const int ncopies = btrfs_raid_array[index].ncopies; const int nparity = btrfs_raid_array[index].nparity; - if (nparity) - return num_stripes - nparity; - else - return num_stripes / ncopies; + return (num_stripes - nparity) / ncopies; } /* [pstart, pend) */ -static int chunk_drange_filter(struct extent_buffer *leaf, - struct btrfs_chunk *chunk, - struct btrfs_balance_args *bargs) +static bool chunk_drange_filter(struct extent_buffer *leaf, struct btrfs_chunk *chunk, + struct btrfs_balance_args *bargs) { struct btrfs_stripe *stripe; int num_stripes = btrfs_chunk_num_stripes(leaf, chunk); @@ -3437,7 +3895,7 @@ static int chunk_drange_filter(struct extent_buffer *leaf, int i; if (!(bargs->flags & BTRFS_BALANCE_ARGS_DEVID)) - return 0; + return false; type = btrfs_chunk_type(leaf, chunk); factor = calc_data_stripes(type, num_stripes); @@ -3453,56 +3911,53 @@ static int chunk_drange_filter(struct extent_buffer *leaf, if (stripe_offset < bargs->pend && stripe_offset + stripe_length > bargs->pstart) - return 0; + return false; } - return 1; + return true; } /* [vstart, vend) */ -static int chunk_vrange_filter(struct extent_buffer *leaf, - struct btrfs_chunk *chunk, - u64 chunk_offset, - struct btrfs_balance_args *bargs) +static bool chunk_vrange_filter(struct extent_buffer *leaf, struct btrfs_chunk *chunk, + u64 chunk_offset, struct btrfs_balance_args *bargs) { if (chunk_offset < bargs->vend && chunk_offset + btrfs_chunk_length(leaf, chunk) > bargs->vstart) /* at least part of the chunk is inside this vrange */ - return 0; + return false; - return 1; + return true; } -static int chunk_stripes_range_filter(struct extent_buffer *leaf, - struct btrfs_chunk *chunk, - struct btrfs_balance_args *bargs) +static bool chunk_stripes_range_filter(struct extent_buffer *leaf, + struct btrfs_chunk *chunk, + struct btrfs_balance_args *bargs) { int num_stripes = btrfs_chunk_num_stripes(leaf, chunk); if (bargs->stripes_min <= num_stripes && num_stripes <= bargs->stripes_max) - return 0; + return false; - return 1; + return true; } -static int chunk_soft_convert_filter(u64 chunk_type, - struct btrfs_balance_args *bargs) +static bool chunk_soft_convert_filter(u64 chunk_type, struct btrfs_balance_args *bargs) { if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT)) - return 0; + return false; chunk_type = chunk_to_extended(chunk_type) & BTRFS_EXTENDED_PROFILE_MASK; if (bargs->target == chunk_type) - return 1; + return true; - return 0; + return false; } -static int should_balance_chunk(struct extent_buffer *leaf, - struct btrfs_chunk *chunk, u64 chunk_offset) +static bool should_balance_chunk(struct extent_buffer *leaf, struct btrfs_chunk *chunk, + u64 chunk_offset) { struct btrfs_fs_info *fs_info = leaf->fs_info; struct btrfs_balance_control *bctl = fs_info->balance_ctl; @@ -3512,7 +3967,7 @@ static int should_balance_chunk(struct extent_buffer *leaf, /* type filter */ if (!((chunk_type & BTRFS_BLOCK_GROUP_TYPE_MASK) & (bctl->flags & BTRFS_BALANCE_TYPE_MASK))) { - return 0; + return false; } if (chunk_type & BTRFS_BLOCK_GROUP_DATA) @@ -3525,46 +3980,46 @@ static int should_balance_chunk(struct extent_buffer *leaf, /* profiles filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) && chunk_profiles_filter(chunk_type, bargs)) { - return 0; + return false; } /* usage filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE) && chunk_usage_filter(fs_info, chunk_offset, bargs)) { - return 0; + return false; } else if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE_RANGE) && chunk_usage_range_filter(fs_info, chunk_offset, bargs)) { - return 0; + return false; } /* devid filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) && chunk_devid_filter(leaf, chunk, bargs)) { - return 0; + return false; } /* drange filter, makes sense only with devid filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_DRANGE) && chunk_drange_filter(leaf, chunk, bargs)) { - return 0; + return false; } /* vrange filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_VRANGE) && chunk_vrange_filter(leaf, chunk, chunk_offset, bargs)) { - return 0; + return false; } /* stripes filter */ if ((bargs->flags & BTRFS_BALANCE_ARGS_STRIPES_RANGE) && chunk_stripes_range_filter(leaf, chunk, bargs)) { - return 0; + return false; } /* soft profile changing mode */ if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) && chunk_soft_convert_filter(chunk_type, bargs)) { - return 0; + return false; } /* @@ -3572,7 +4027,7 @@ static int should_balance_chunk(struct extent_buffer *leaf, */ if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT)) { if (bargs->limit == 0) - return 0; + return false; else bargs->limit--; } else if ((bargs->flags & BTRFS_BALANCE_ARGS_LIMIT_RANGE)) { @@ -3582,12 +4037,12 @@ static int should_balance_chunk(struct extent_buffer *leaf, * about the count of all chunks that satisfy the filters. */ if (bargs->limit_max == 0) - return 0; + return false; else bargs->limit_max--; } - return 1; + return true; } static int __btrfs_balance(struct btrfs_fs_info *fs_info) @@ -3596,7 +4051,7 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info) struct btrfs_root *chunk_root = fs_info->chunk_root; u64 chunk_type; struct btrfs_chunk *chunk; - struct btrfs_path *path = NULL; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; struct btrfs_key found_key; struct extent_buffer *leaf; @@ -3634,8 +4089,8 @@ again: bctl->sys.limit = limit_sys; } key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID; - key.offset = (u64)-1; key.type = BTRFS_CHUNK_ITEM_KEY; + key.offset = (u64)-1; while (1) { if ((!counting && atomic_read(&fs_info->balance_pause_req)) || @@ -3644,10 +4099,10 @@ again: goto error; } - mutex_lock(&fs_info->delete_unused_bgs_mutex); + mutex_lock(&fs_info->reclaim_bgs_lock); ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0); if (ret < 0) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto error; } @@ -3661,7 +4116,7 @@ again: ret = btrfs_previous_item(chunk_root, path, 0, BTRFS_CHUNK_ITEM_KEY); if (ret) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); ret = 0; break; } @@ -3671,7 +4126,7 @@ again: btrfs_item_key_to_cpu(leaf, &found_key, slot); if (found_key.objectid != key.objectid) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); break; } @@ -3688,12 +4143,12 @@ again: btrfs_release_path(path); if (!ret) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto loop; } if (counting) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); spin_lock(&fs_info->balance_lock); bctl->stat.expected++; spin_unlock(&fs_info->balance_lock); @@ -3718,7 +4173,7 @@ again: count_meta < bctl->meta.limit_min) || ((chunk_type & BTRFS_BLOCK_GROUP_SYSTEM) && count_sys < bctl->sys.limit_min)) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto loop; } @@ -3732,15 +4187,15 @@ again: ret = btrfs_may_alloc_data_chunk(fs_info, found_key.offset); if (ret < 0) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto error; } else if (ret == 1) { chunk_reserved = 1; } } - ret = btrfs_relocate_chunk(fs_info, found_key.offset); - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + ret = btrfs_relocate_chunk(fs_info, found_key.offset, true); + mutex_unlock(&fs_info->reclaim_bgs_lock); if (ret == -ENOSPC) { enospc_errors++; } else if (ret == -ETXTBSY) { @@ -3767,7 +4222,6 @@ loop: goto again; } error: - btrfs_free_path(path); if (enospc_errors) { btrfs_info(fs_info, "%d enospc errors during balance", enospc_errors); @@ -3778,12 +4232,13 @@ error: return ret; } -/** - * alloc_profile_is_valid - see if a given profile is valid and reduced - * @flags: profile to validate - * @extended: if true @flags is treated as an extended profile +/* + * See if a given profile is valid and reduced. + * + * @flags: profile to validate + * @extended: if true @flags is treated as an extended profile */ -static int alloc_profile_is_valid(u64 flags, int extended) +static int alloc_profile_is_valid(u64 flags, bool extended) { u64 mask = (extended ? BTRFS_EXTENDED_PROFILE_MASK : BTRFS_BLOCK_GROUP_PROFILE_MASK); @@ -3801,14 +4256,6 @@ static int alloc_profile_is_valid(u64 flags, int extended) return has_single_bit_set(flags); } -static inline int balance_need_close(struct btrfs_fs_info *fs_info) -{ - /* cancel requested || normal exit path */ - return atomic_read(&fs_info->balance_cancel_req) || - (atomic_read(&fs_info->balance_pause_req) == 0 && - atomic_read(&fs_info->balance_cancel_req) == 0); -} - /* * Validate target profile against allowed profiles and return true if it's OK. * Otherwise print the error message and return false. @@ -3932,7 +4379,7 @@ static void describe_balance_start_or_resume(struct btrfs_fs_info *fs_info) { u32 size_buf = 1024; char tmp_buf[192] = {'\0'}; - char *buf; + char AUTO_KFREE(buf); char *bp; u32 size_bp = size_buf; int ret; @@ -3980,12 +4427,10 @@ out_overflow: btrfs_info(fs_info, "balance: %s %s", (bctl->flags & BTRFS_BALANCE_RESUME) ? "resume" : "start", buf); - - kfree(buf); } /* - * Should be called with balance mutexe held + * Should be called with balance mutex held */ int btrfs_balance(struct btrfs_fs_info *fs_info, struct btrfs_balance_control *bctl, @@ -3998,6 +4443,7 @@ int btrfs_balance(struct btrfs_fs_info *fs_info, u64 num_devices; unsigned seq; bool reducing_redundancy; + bool paused = false; int i; if (btrfs_fs_closing(fs_info) || @@ -4029,7 +4475,7 @@ int btrfs_balance(struct btrfs_fs_info *fs_info, /* * rw_devices will not change at the moment, device add/delete/replace - * are excluded by EXCL_OP + * are exclusive */ num_devices = fs_info->fs_devices->rw_devices; @@ -4100,14 +4546,6 @@ int btrfs_balance(struct btrfs_fs_info *fs_info, btrfs_bg_type_to_raid_name(data_target)); } - if (fs_info->send_in_progress) { - btrfs_warn_rl(fs_info, -"cannot run balance while send operations are in progress (%d in progress)", - fs_info->send_in_progress); - ret = -EAGAIN; - goto out; - } - ret = insert_balance_item(fs_info, bctl); if (ret && ret != -EEXIST) goto out; @@ -4133,9 +4571,27 @@ int btrfs_balance(struct btrfs_fs_info *fs_info, ret = __btrfs_balance(fs_info); mutex_lock(&fs_info->balance_mutex); - if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req)) + if (ret == -ECANCELED && atomic_read(&fs_info->balance_pause_req)) { btrfs_info(fs_info, "balance: paused"); - else if (ret == -ECANCELED && atomic_read(&fs_info->balance_cancel_req)) + btrfs_exclop_balance(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED); + paused = true; + } + /* + * Balance can be canceled by: + * + * - Regular cancel request + * Then ret == -ECANCELED and balance_cancel_req > 0 + * + * - Fatal signal to "btrfs" process + * Either the signal caught by wait_reserve_ticket() and callers + * got -EINTR, or caught by btrfs_should_cancel_balance() and + * got -ECANCELED. + * Either way, in this case balance_cancel_req = 0, and + * ret == -EINTR or ret == -ECANCELED. + * + * So here we only check the return value to catch canceled balance. + */ + else if (ret == -ECANCELED || ret == -EINTR) btrfs_info(fs_info, "balance: canceled"); else btrfs_info(fs_info, "balance: ended with status: %d", ret); @@ -4147,10 +4603,10 @@ int btrfs_balance(struct btrfs_fs_info *fs_info, btrfs_update_ioctl_balance_args(fs_info, bargs); } - if ((ret && ret != -ECANCELED && ret != -ENOSPC) || - balance_need_close(fs_info)) { + /* We didn't pause, we can clean everything up. */ + if (!paused) { reset_balance_state(fs_info); - clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags); + btrfs_exclop_finish(fs_info); } wake_up(&fs_info->balance_wait_q); @@ -4161,7 +4617,7 @@ out: reset_balance_state(fs_info); else kfree(bctl); - clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags); + btrfs_exclop_finish(fs_info); return ret; } @@ -4171,6 +4627,8 @@ static int balance_kthread(void *data) struct btrfs_fs_info *fs_info = data; int ret = 0; + guard(super_write)(fs_info->sb); + mutex_lock(&fs_info->balance_mutex); if (fs_info->balance_ctl) ret = btrfs_balance(fs_info, fs_info->balance_ctl, NULL); @@ -4195,6 +4653,11 @@ int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info) return 0; } + spin_lock(&fs_info->super_lock); + ASSERT(fs_info->exclusive_operation == BTRFS_EXCLOP_BALANCE_PAUSED, + "exclusive_operation=%d", fs_info->exclusive_operation); + fs_info->exclusive_operation = BTRFS_EXCLOP_BALANCE; + spin_unlock(&fs_info->super_lock); /* * A ro->rw remount sequence should continue with the paused balance * regardless of who pauses it, system or the user as of now, so set @@ -4213,7 +4676,7 @@ int btrfs_recover_balance(struct btrfs_fs_info *fs_info) struct btrfs_balance_control *bctl; struct btrfs_balance_item *item; struct btrfs_disk_balance_args disk_bargs; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct extent_buffer *leaf; struct btrfs_key key; int ret; @@ -4228,17 +4691,14 @@ int btrfs_recover_balance(struct btrfs_fs_info *fs_info) ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); if (ret < 0) - goto out; + return ret; if (ret > 0) { /* ret = -ENOENT; */ - ret = 0; - goto out; + return 0; } bctl = kzalloc(sizeof(*bctl), GFP_NOFS); - if (!bctl) { - ret = -ENOMEM; - goto out; - } + if (!bctl) + return -ENOMEM; leaf = path->nodes[0]; item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_balance_item); @@ -4263,18 +4723,18 @@ int btrfs_recover_balance(struct btrfs_fs_info *fs_info) * is in a paused state and must have fs_info::balance_ctl properly * set up. */ - if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) + if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE_PAUSED)) btrfs_warn(fs_info, "balance: cannot set exclusive op status, resume manually"); + btrfs_release_path(path); + mutex_lock(&fs_info->balance_mutex); BUG_ON(fs_info->balance_ctl); spin_lock(&fs_info->balance_lock); fs_info->balance_ctl = bctl; spin_unlock(&fs_info->balance_lock); mutex_unlock(&fs_info->balance_mutex); -out: - btrfs_free_path(path); return ret; } @@ -4345,194 +4805,17 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info) if (fs_info->balance_ctl) { reset_balance_state(fs_info); - clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags); + btrfs_exclop_finish(fs_info); btrfs_info(fs_info, "balance: canceled"); } } - BUG_ON(fs_info->balance_ctl || - test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)); + ASSERT(!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)); atomic_dec(&fs_info->balance_cancel_req); mutex_unlock(&fs_info->balance_mutex); return 0; } -int btrfs_uuid_scan_kthread(void *data) -{ - struct btrfs_fs_info *fs_info = data; - struct btrfs_root *root = fs_info->tree_root; - struct btrfs_key key; - struct btrfs_path *path = NULL; - int ret = 0; - struct extent_buffer *eb; - int slot; - struct btrfs_root_item root_item; - u32 item_size; - struct btrfs_trans_handle *trans = NULL; - bool closing = false; - - path = btrfs_alloc_path(); - if (!path) { - ret = -ENOMEM; - goto out; - } - - key.objectid = 0; - key.type = BTRFS_ROOT_ITEM_KEY; - key.offset = 0; - - while (1) { - if (btrfs_fs_closing(fs_info)) { - closing = true; - break; - } - ret = btrfs_search_forward(root, &key, path, - BTRFS_OLDEST_GENERATION); - if (ret) { - if (ret > 0) - ret = 0; - break; - } - - if (key.type != BTRFS_ROOT_ITEM_KEY || - (key.objectid < BTRFS_FIRST_FREE_OBJECTID && - key.objectid != BTRFS_FS_TREE_OBJECTID) || - key.objectid > BTRFS_LAST_FREE_OBJECTID) - goto skip; - - eb = path->nodes[0]; - slot = path->slots[0]; - item_size = btrfs_item_size_nr(eb, slot); - if (item_size < sizeof(root_item)) - goto skip; - - read_extent_buffer(eb, &root_item, - btrfs_item_ptr_offset(eb, slot), - (int)sizeof(root_item)); - if (btrfs_root_refs(&root_item) == 0) - goto skip; - - if (!btrfs_is_empty_uuid(root_item.uuid) || - !btrfs_is_empty_uuid(root_item.received_uuid)) { - if (trans) - goto update_tree; - - btrfs_release_path(path); - /* - * 1 - subvol uuid item - * 1 - received_subvol uuid item - */ - trans = btrfs_start_transaction(fs_info->uuid_root, 2); - if (IS_ERR(trans)) { - ret = PTR_ERR(trans); - break; - } - continue; - } else { - goto skip; - } -update_tree: - if (!btrfs_is_empty_uuid(root_item.uuid)) { - ret = btrfs_uuid_tree_add(trans, root_item.uuid, - BTRFS_UUID_KEY_SUBVOL, - key.objectid); - if (ret < 0) { - btrfs_warn(fs_info, "uuid_tree_add failed %d", - ret); - break; - } - } - - if (!btrfs_is_empty_uuid(root_item.received_uuid)) { - ret = btrfs_uuid_tree_add(trans, - root_item.received_uuid, - BTRFS_UUID_KEY_RECEIVED_SUBVOL, - key.objectid); - if (ret < 0) { - btrfs_warn(fs_info, "uuid_tree_add failed %d", - ret); - break; - } - } - -skip: - if (trans) { - ret = btrfs_end_transaction(trans); - trans = NULL; - if (ret) - break; - } - - btrfs_release_path(path); - if (key.offset < (u64)-1) { - key.offset++; - } else if (key.type < BTRFS_ROOT_ITEM_KEY) { - key.offset = 0; - key.type = BTRFS_ROOT_ITEM_KEY; - } else if (key.objectid < (u64)-1) { - key.offset = 0; - key.type = BTRFS_ROOT_ITEM_KEY; - key.objectid++; - } else { - break; - } - cond_resched(); - } - -out: - btrfs_free_path(path); - if (trans && !IS_ERR(trans)) - btrfs_end_transaction(trans); - if (ret) - btrfs_warn(fs_info, "btrfs_uuid_scan_kthread failed %d", ret); - else if (!closing) - set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags); - up(&fs_info->uuid_tree_rescan_sem); - return 0; -} - -int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info) -{ - struct btrfs_trans_handle *trans; - struct btrfs_root *tree_root = fs_info->tree_root; - struct btrfs_root *uuid_root; - struct task_struct *task; - int ret; - - /* - * 1 - root node - * 1 - root item - */ - trans = btrfs_start_transaction(tree_root, 2); - if (IS_ERR(trans)) - return PTR_ERR(trans); - - uuid_root = btrfs_create_tree(trans, BTRFS_UUID_TREE_OBJECTID); - if (IS_ERR(uuid_root)) { - ret = PTR_ERR(uuid_root); - btrfs_abort_transaction(trans, ret); - btrfs_end_transaction(trans); - return ret; - } - - fs_info->uuid_root = uuid_root; - - ret = btrfs_commit_transaction(trans); - if (ret) - return ret; - - down(&fs_info->uuid_tree_rescan_sem); - task = kthread_run(btrfs_uuid_scan_kthread, fs_info, "btrfs-uuid"); - if (IS_ERR(task)) { - /* fs_info->update_uuid_tree_gen remains 0 in all error case */ - btrfs_warn(fs_info, "failed to start uuid_scan task"); - up(&fs_info->uuid_tree_rescan_sem); - return PTR_ERR(task); - } - - return 0; -} - /* * shrinking a device means finding all of the device extents past * the new size, and then following the back refs to the chunks. @@ -4558,6 +4841,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size) u64 old_size = btrfs_device_get_total_bytes(device); u64 diff; u64 start; + u64 free_diff = 0; new_size = round_down(new_size, fs_info->sectorsize); start = new_size; @@ -4583,7 +4867,19 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size) btrfs_device_set_total_bytes(device, new_size); if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { device->fs_devices->total_rw_bytes -= diff; - atomic64_sub(diff, &fs_info->free_chunk_space); + + /* + * The new free_chunk_space is new_size - used, so we have to + * subtract the delta of the old free_chunk_space which included + * old_size - used. If used > new_size then just subtract this + * entire device's free space. + */ + if (device->bytes_used < new_size) + free_diff = (old_size - device->bytes_used) - + (new_size - device->bytes_used); + else + free_diff = old_size - device->bytes_used; + atomic64_sub(free_diff, &fs_info->free_chunk_space); } /* @@ -4603,23 +4899,22 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size) again: key.objectid = device->devid; - key.offset = (u64)-1; key.type = BTRFS_DEV_EXTENT_KEY; + key.offset = (u64)-1; do { - mutex_lock(&fs_info->delete_unused_bgs_mutex); + mutex_lock(&fs_info->reclaim_bgs_lock); ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto done; } ret = btrfs_previous_item(root, path, 0, key.type); - if (ret) - mutex_unlock(&fs_info->delete_unused_bgs_mutex); - if (ret < 0) - goto done; if (ret) { + mutex_unlock(&fs_info->reclaim_bgs_lock); + if (ret < 0) + goto done; ret = 0; btrfs_release_path(path); break; @@ -4630,7 +4925,7 @@ again: btrfs_item_key_to_cpu(l, &key, path->slots[0]); if (key.objectid != device->devid) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); btrfs_release_path(path); break; } @@ -4639,7 +4934,7 @@ again: length = btrfs_dev_extent_length(l, dev_extent); if (key.offset + length <= new_size) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); btrfs_release_path(path); break; } @@ -4655,12 +4950,12 @@ again: */ ret = btrfs_may_alloc_data_chunk(fs_info, chunk_offset); if (ret < 0) { - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + mutex_unlock(&fs_info->reclaim_bgs_lock); goto done; } - ret = btrfs_relocate_chunk(fs_info, chunk_offset); - mutex_unlock(&fs_info->delete_unused_bgs_mutex); + ret = btrfs_relocate_chunk(fs_info, chunk_offset, true); + mutex_unlock(&fs_info->reclaim_bgs_lock); if (ret == -ENOSPC) { failed++; } else if (ret) { @@ -4690,6 +4985,10 @@ again: } mutex_lock(&fs_info->chunk_mutex); + /* Clear all state bits beyond the shrunk device size */ + btrfs_clear_extent_bit(&device->alloc_state, new_size, (u64)-1, + CHUNK_STATE_MASK, NULL); + btrfs_device_set_disk_total_bytes(device, new_size); if (list_empty(&device->post_commit_list)) list_add_tail(&device->post_commit_list, @@ -4700,9 +4999,11 @@ again: round_down(old_total - diff, fs_info->sectorsize)); mutex_unlock(&fs_info->chunk_mutex); + btrfs_reserve_chunk_metadata(trans, false); /* Now btrfs_update_device() will change the on-disk size. */ ret = btrfs_update_device(trans, device); - if (ret < 0) { + btrfs_trans_release_chunk_metadata(trans); + if (unlikely(ret < 0)) { btrfs_abort_transaction(trans, ret); btrfs_end_transaction(trans); } else { @@ -4713,9 +5014,10 @@ done: if (ret) { mutex_lock(&fs_info->chunk_mutex); btrfs_device_set_total_bytes(device, old_size); - if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) + if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { device->fs_devices->total_rw_bytes += diff; - atomic64_add(diff, &fs_info->free_chunk_space); + atomic64_add(free_diff, &fs_info->free_chunk_space); + } mutex_unlock(&fs_info->chunk_mutex); } return ret; @@ -4730,13 +5032,12 @@ static int btrfs_add_system_chunk(struct btrfs_fs_info *fs_info, u32 array_size; u8 *ptr; - mutex_lock(&fs_info->chunk_mutex); + lockdep_assert_held(&fs_info->chunk_mutex); + array_size = btrfs_super_sys_array_size(super_copy); if (array_size + item_size + sizeof(disk_key) - > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) { - mutex_unlock(&fs_info->chunk_mutex); + > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) return -EFBIG; - } ptr = super_copy->sys_chunk_array + array_size; btrfs_cpu_key_to_disk(&disk_key, key); @@ -4745,7 +5046,6 @@ static int btrfs_add_system_chunk(struct btrfs_fs_info *fs_info, memcpy(ptr, chunk, item_size); item_size += sizeof(disk_key); btrfs_set_super_sys_array_size(super_copy, array_size + item_size); - mutex_unlock(&fs_info->chunk_mutex); return 0; } @@ -4786,7 +5086,7 @@ static void check_raid1c34_incompat_flag(struct btrfs_fs_info *info, u64 type) } /* - * Structure used internally for __btrfs_alloc_chunk() function. + * Structure used internally for btrfs_create_chunk() function. * Wraps needed parameters. */ struct alloc_chunk_ctl { @@ -4814,26 +5114,49 @@ struct alloc_chunk_ctl { u64 stripe_size; u64 chunk_size; int ndevs; + /* Space_info the block group is going to belong. */ + struct btrfs_space_info *space_info; }; static void init_alloc_chunk_ctl_policy_regular( struct btrfs_fs_devices *fs_devices, struct alloc_chunk_ctl *ctl) { + struct btrfs_space_info *space_info; + + space_info = btrfs_find_space_info(fs_devices->fs_info, ctl->type); + ASSERT(space_info); + + ctl->max_chunk_size = READ_ONCE(space_info->chunk_size); + ctl->max_stripe_size = min_t(u64, ctl->max_chunk_size, SZ_1G); + + if (ctl->type & BTRFS_BLOCK_GROUP_SYSTEM) + ctl->devs_max = min_t(int, ctl->devs_max, BTRFS_MAX_DEVS_SYS_CHUNK); + + /* We don't want a chunk larger than 10% of writable space */ + ctl->max_chunk_size = min(mult_perc(fs_devices->total_rw_bytes, 10), + ctl->max_chunk_size); + ctl->dev_extent_min = btrfs_stripe_nr_to_offset(ctl->dev_stripes); +} + +static void init_alloc_chunk_ctl_policy_zoned( + struct btrfs_fs_devices *fs_devices, + struct alloc_chunk_ctl *ctl) +{ + u64 zone_size = fs_devices->fs_info->zone_size; + u64 limit; + int min_num_stripes = ctl->devs_min * ctl->dev_stripes; + int min_data_stripes = (min_num_stripes - ctl->nparity) / ctl->ncopies; + u64 min_chunk_size = min_data_stripes * zone_size; u64 type = ctl->type; + ctl->max_stripe_size = zone_size; if (type & BTRFS_BLOCK_GROUP_DATA) { - ctl->max_stripe_size = SZ_1G; - ctl->max_chunk_size = BTRFS_MAX_DATA_CHUNK_SIZE; + ctl->max_chunk_size = round_down(BTRFS_MAX_DATA_CHUNK_SIZE, + zone_size); } else if (type & BTRFS_BLOCK_GROUP_METADATA) { - /* For larger filesystems, use larger metadata chunks */ - if (fs_devices->total_rw_bytes > 50ULL * SZ_1G) - ctl->max_stripe_size = SZ_1G; - else - ctl->max_stripe_size = SZ_256M; ctl->max_chunk_size = ctl->max_stripe_size; } else if (type & BTRFS_BLOCK_GROUP_SYSTEM) { - ctl->max_stripe_size = SZ_32M; ctl->max_chunk_size = 2 * ctl->max_stripe_size; ctl->devs_max = min_t(int, ctl->devs_max, BTRFS_MAX_DEVS_SYS_CHUNK); @@ -4842,9 +5165,11 @@ static void init_alloc_chunk_ctl_policy_regular( } /* We don't want a chunk larger than 10% of writable space */ - ctl->max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1), - ctl->max_chunk_size); - ctl->dev_extent_min = BTRFS_STRIPE_LEN * ctl->dev_stripes; + limit = max(round_down(mult_perc(fs_devices->total_rw_bytes, 10), + zone_size), + min_chunk_size); + ctl->max_chunk_size = min(limit, ctl->max_chunk_size); + ctl->dev_extent_min = zone_size * ctl->dev_stripes; } static void init_alloc_chunk_ctl(struct btrfs_fs_devices *fs_devices, @@ -4864,11 +5189,15 @@ static void init_alloc_chunk_ctl(struct btrfs_fs_devices *fs_devices, ctl->ndevs = 0; switch (fs_devices->chunk_alloc_policy) { + default: + btrfs_warn_unknown_chunk_allocation(fs_devices->chunk_alloc_policy); + fallthrough; case BTRFS_CHUNK_ALLOC_REGULAR: init_alloc_chunk_ctl_policy_regular(fs_devices, ctl); break; - default: - BUG(); + case BTRFS_CHUNK_ALLOC_ZONED: + init_alloc_chunk_ctl_policy_zoned(fs_devices, ctl); + break; } } @@ -4986,6 +5315,9 @@ static int decide_stripe_size_regular(struct alloc_chunk_ctl *ctl, ctl->stripe_size); } + /* Stripe size should not go beyond 1G. */ + ctl->stripe_size = min_t(u64, ctl->stripe_size, SZ_1G); + /* Align to BTRFS_STRIPE_LEN */ ctl->stripe_size = round_down(ctl->stripe_size, BTRFS_STRIPE_LEN); ctl->chunk_size = ctl->stripe_size * data_stripes; @@ -4993,6 +5325,42 @@ static int decide_stripe_size_regular(struct alloc_chunk_ctl *ctl, return 0; } +static int decide_stripe_size_zoned(struct alloc_chunk_ctl *ctl, + struct btrfs_device_info *devices_info) +{ + u64 zone_size = devices_info[0].dev->zone_info->zone_size; + /* Number of stripes that count for block group size */ + int data_stripes; + + /* + * It should hold because: + * dev_extent_min == dev_extent_want == zone_size * dev_stripes + */ + ASSERT(devices_info[ctl->ndevs - 1].max_avail == ctl->dev_extent_min, + "ndevs=%d max_avail=%llu dev_extent_min=%llu", ctl->ndevs, + devices_info[ctl->ndevs - 1].max_avail, ctl->dev_extent_min); + + ctl->stripe_size = zone_size; + ctl->num_stripes = ctl->ndevs * ctl->dev_stripes; + data_stripes = (ctl->num_stripes - ctl->nparity) / ctl->ncopies; + + /* stripe_size is fixed in zoned filesystem. Reduce ndevs instead. */ + if (ctl->stripe_size * data_stripes > ctl->max_chunk_size) { + ctl->ndevs = div_u64(div_u64(ctl->max_chunk_size * ctl->ncopies, + ctl->stripe_size) + ctl->nparity, + ctl->dev_stripes); + ctl->num_stripes = ctl->ndevs * ctl->dev_stripes; + data_stripes = (ctl->num_stripes - ctl->nparity) / ctl->ncopies; + ASSERT(ctl->stripe_size * data_stripes <= ctl->max_chunk_size, + "stripe_size=%llu data_stripes=%d max_chunk_size=%llu", + ctl->stripe_size, data_stripes, ctl->max_chunk_size); + } + + ctl->chunk_size = ctl->stripe_size * data_stripes; + + return 0; +} + static int decide_stripe_size(struct btrfs_fs_devices *fs_devices, struct alloc_chunk_ctl *ctl, struct btrfs_device_info *devices_info) @@ -5018,76 +5386,151 @@ static int decide_stripe_size(struct btrfs_fs_devices *fs_devices, ctl->ndevs = min(ctl->ndevs, ctl->devs_max); switch (fs_devices->chunk_alloc_policy) { + default: + btrfs_warn_unknown_chunk_allocation(fs_devices->chunk_alloc_policy); + fallthrough; case BTRFS_CHUNK_ALLOC_REGULAR: return decide_stripe_size_regular(ctl, devices_info); - default: - BUG(); + case BTRFS_CHUNK_ALLOC_ZONED: + return decide_stripe_size_zoned(ctl, devices_info); + } +} + +static void chunk_map_device_set_bits(struct btrfs_chunk_map *map, unsigned int bits) +{ + for (int i = 0; i < map->num_stripes; i++) { + struct btrfs_io_stripe *stripe = &map->stripes[i]; + struct btrfs_device *device = stripe->dev; + + btrfs_set_extent_bit(&device->alloc_state, stripe->physical, + stripe->physical + map->stripe_size - 1, + bits | EXTENT_NOWAIT, NULL); } } -static int create_chunk(struct btrfs_trans_handle *trans, +static void chunk_map_device_clear_bits(struct btrfs_chunk_map *map, unsigned int bits) +{ + for (int i = 0; i < map->num_stripes; i++) { + struct btrfs_io_stripe *stripe = &map->stripes[i]; + struct btrfs_device *device = stripe->dev; + + btrfs_clear_extent_bit(&device->alloc_state, stripe->physical, + stripe->physical + map->stripe_size - 1, + bits | EXTENT_NOWAIT, NULL); + } +} + +void btrfs_remove_chunk_map(struct btrfs_fs_info *fs_info, struct btrfs_chunk_map *map) +{ + write_lock(&fs_info->mapping_tree_lock); + rb_erase_cached(&map->rb_node, &fs_info->mapping_tree); + RB_CLEAR_NODE(&map->rb_node); + chunk_map_device_clear_bits(map, CHUNK_ALLOCATED); + write_unlock(&fs_info->mapping_tree_lock); + + /* Once for the tree reference. */ + btrfs_free_chunk_map(map); +} + +static int btrfs_chunk_map_cmp(const struct rb_node *new, + const struct rb_node *exist) +{ + const struct btrfs_chunk_map *new_map = + rb_entry(new, struct btrfs_chunk_map, rb_node); + const struct btrfs_chunk_map *exist_map = + rb_entry(exist, struct btrfs_chunk_map, rb_node); + + if (new_map->start == exist_map->start) + return 0; + if (new_map->start < exist_map->start) + return -1; + return 1; +} + +EXPORT_FOR_TESTS +int btrfs_add_chunk_map(struct btrfs_fs_info *fs_info, struct btrfs_chunk_map *map) +{ + struct rb_node *exist; + + write_lock(&fs_info->mapping_tree_lock); + exist = rb_find_add_cached(&map->rb_node, &fs_info->mapping_tree, + btrfs_chunk_map_cmp); + + if (exist) { + write_unlock(&fs_info->mapping_tree_lock); + return -EEXIST; + } + chunk_map_device_set_bits(map, CHUNK_ALLOCATED); + chunk_map_device_clear_bits(map, CHUNK_TRIMMED); + write_unlock(&fs_info->mapping_tree_lock); + + return 0; +} + +EXPORT_FOR_TESTS +struct btrfs_chunk_map *btrfs_alloc_chunk_map(int num_stripes, gfp_t gfp) +{ + struct btrfs_chunk_map *map; + + map = kmalloc(btrfs_chunk_map_size(num_stripes), gfp); + if (!map) + return NULL; + + refcount_set(&map->refs, 1); + RB_CLEAR_NODE(&map->rb_node); + + return map; +} + +static struct btrfs_block_group *create_chunk(struct btrfs_trans_handle *trans, struct alloc_chunk_ctl *ctl, struct btrfs_device_info *devices_info) { struct btrfs_fs_info *info = trans->fs_info; - struct map_lookup *map = NULL; - struct extent_map_tree *em_tree; - struct extent_map *em; + struct btrfs_chunk_map *map; + struct btrfs_block_group *block_group; u64 start = ctl->start; u64 type = ctl->type; int ret; - int i; - int j; - map = kmalloc(map_lookup_size(ctl->num_stripes), GFP_NOFS); + map = btrfs_alloc_chunk_map(ctl->num_stripes, GFP_NOFS); if (!map) - return -ENOMEM; + return ERR_PTR(-ENOMEM); + + map->start = start; + map->chunk_len = ctl->chunk_size; + map->stripe_size = ctl->stripe_size; + map->type = type; + map->io_align = BTRFS_STRIPE_LEN; + map->io_width = BTRFS_STRIPE_LEN; + map->sub_stripes = ctl->sub_stripes; map->num_stripes = ctl->num_stripes; - for (i = 0; i < ctl->ndevs; ++i) { - for (j = 0; j < ctl->dev_stripes; ++j) { + for (int i = 0; i < ctl->ndevs; i++) { + for (int j = 0; j < ctl->dev_stripes; j++) { int s = i * ctl->dev_stripes + j; map->stripes[s].dev = devices_info[i].dev; map->stripes[s].physical = devices_info[i].dev_offset + j * ctl->stripe_size; } } - map->stripe_len = BTRFS_STRIPE_LEN; - map->io_align = BTRFS_STRIPE_LEN; - map->io_width = BTRFS_STRIPE_LEN; - map->type = type; - map->sub_stripes = ctl->sub_stripes; trace_btrfs_chunk_alloc(info, map, start, ctl->chunk_size); - em = alloc_extent_map(); - if (!em) { - kfree(map); - return -ENOMEM; - } - set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags); - em->map_lookup = map; - em->start = start; - em->len = ctl->chunk_size; - em->block_start = 0; - em->block_len = em->len; - em->orig_block_len = ctl->stripe_size; - - em_tree = &info->mapping_tree; - write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, em, 0); + ret = btrfs_add_chunk_map(info, map); if (ret) { - write_unlock(&em_tree->lock); - free_extent_map(em); - return ret; + btrfs_free_chunk_map(map); + return ERR_PTR(ret); } - write_unlock(&em_tree->lock); - ret = btrfs_make_block_group(trans, 0, type, start, ctl->chunk_size); - if (ret) - goto error_del_extent; + block_group = btrfs_make_block_group(trans, ctl->space_info, type, start, + ctl->chunk_size); + if (IS_ERR(block_group)) { + btrfs_remove_chunk_map(info, map); + return block_group; + } - for (i = 0; i < map->num_stripes; i++) { + for (int i = 0; i < map->num_stripes; i++) { struct btrfs_device *dev = map->stripes[i].dev; btrfs_device_set_bytes_used(dev, @@ -5100,178 +5543,171 @@ static int create_chunk(struct btrfs_trans_handle *trans, atomic64_sub(ctl->stripe_size * map->num_stripes, &info->free_chunk_space); - free_extent_map(em); check_raid56_incompat_flag(info, type); check_raid1c34_incompat_flag(info, type); - return 0; - -error_del_extent: - write_lock(&em_tree->lock); - remove_extent_mapping(em_tree, em); - write_unlock(&em_tree->lock); - - /* One for our allocation */ - free_extent_map(em); - /* One for the tree reference */ - free_extent_map(em); - - return ret; + return block_group; } -int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, u64 type) +struct btrfs_block_group *btrfs_create_chunk(struct btrfs_trans_handle *trans, + struct btrfs_space_info *space_info, + u64 type) { struct btrfs_fs_info *info = trans->fs_info; struct btrfs_fs_devices *fs_devices = info->fs_devices; - struct btrfs_device_info *devices_info = NULL; + struct btrfs_device_info AUTO_KFREE(devices_info); struct alloc_chunk_ctl ctl; int ret; lockdep_assert_held(&info->chunk_mutex); if (!alloc_profile_is_valid(type, 0)) { - ASSERT(0); - return -EINVAL; + DEBUG_WARN("invalid alloc profile for type %llu", type); + return ERR_PTR(-EINVAL); } if (list_empty(&fs_devices->alloc_list)) { if (btrfs_test_opt(info, ENOSPC_DEBUG)) btrfs_debug(info, "%s: no writable device", __func__); - return -ENOSPC; + return ERR_PTR(-ENOSPC); } if (!(type & BTRFS_BLOCK_GROUP_TYPE_MASK)) { btrfs_err(info, "invalid chunk type 0x%llx requested", type); - ASSERT(0); - return -EINVAL; + DEBUG_WARN(); + return ERR_PTR(-EINVAL); } ctl.start = find_next_chunk(info); ctl.type = type; + ctl.space_info = space_info; init_alloc_chunk_ctl(fs_devices, &ctl); devices_info = kcalloc(fs_devices->rw_devices, sizeof(*devices_info), GFP_NOFS); if (!devices_info) - return -ENOMEM; + return ERR_PTR(-ENOMEM); ret = gather_device_info(fs_devices, &ctl, devices_info); if (ret < 0) - goto out; + return ERR_PTR(ret); ret = decide_stripe_size(fs_devices, &ctl, devices_info); if (ret < 0) - goto out; - - ret = create_chunk(trans, &ctl, devices_info); + return ERR_PTR(ret); -out: - kfree(devices_info); - return ret; + return create_chunk(trans, &ctl, devices_info); } /* - * Chunk allocation falls into two parts. The first part does work - * that makes the new allocated chunk usable, but does not do any operation - * that modifies the chunk tree. The second part does the work that - * requires modifying the chunk tree. This division is important for the - * bootstrap process of adding storage to a seed btrfs. + * This function, btrfs_chunk_alloc_add_chunk_item(), typically belongs to the + * phase 1 of chunk allocation. It belongs to phase 2 only when allocating system + * chunks. + * + * See the comment at btrfs_chunk_alloc() for details about the chunk allocation + * phases. */ -int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans, - u64 chunk_offset, u64 chunk_size) +int btrfs_chunk_alloc_add_chunk_item(struct btrfs_trans_handle *trans, + struct btrfs_block_group *bg) { struct btrfs_fs_info *fs_info = trans->fs_info; - struct btrfs_root *extent_root = fs_info->extent_root; struct btrfs_root *chunk_root = fs_info->chunk_root; struct btrfs_key key; - struct btrfs_device *device; struct btrfs_chunk *chunk; struct btrfs_stripe *stripe; - struct extent_map *em; - struct map_lookup *map; + struct btrfs_chunk_map *map; size_t item_size; - u64 dev_offset; - u64 stripe_size; - int i = 0; - int ret = 0; + int i; + int ret; + + /* + * We take the chunk_mutex for 2 reasons: + * + * 1) Updates and insertions in the chunk btree must be done while holding + * the chunk_mutex, as well as updating the system chunk array in the + * superblock. See the comment on top of btrfs_chunk_alloc() for the + * details; + * + * 2) To prevent races with the final phase of a device replace operation + * that replaces the device object associated with the map's stripes, + * because the device object's id can change at any time during that + * final phase of the device replace operation + * (dev-replace.c:btrfs_dev_replace_finishing()), so we could grab the + * replaced device and then see it with an ID of BTRFS_DEV_REPLACE_DEVID, + * which would cause a failure when updating the device item, which does + * not exists, or persisting a stripe of the chunk item with such ID. + * Here we can't use the device_list_mutex because our caller already + * has locked the chunk_mutex, and the final phase of device replace + * acquires both mutexes - first the device_list_mutex and then the + * chunk_mutex. Using any of those two mutexes protects us from a + * concurrent device replace. + */ + lockdep_assert_held(&fs_info->chunk_mutex); - em = btrfs_get_chunk_map(fs_info, chunk_offset, chunk_size); - if (IS_ERR(em)) - return PTR_ERR(em); + map = btrfs_get_chunk_map(fs_info, bg->start, bg->length); + if (IS_ERR(map)) { + ret = PTR_ERR(map); + btrfs_abort_transaction(trans, ret); + return ret; + } - map = em->map_lookup; item_size = btrfs_chunk_item_size(map->num_stripes); - stripe_size = em->orig_block_len; chunk = kzalloc(item_size, GFP_NOFS); - if (!chunk) { + if (unlikely(!chunk)) { ret = -ENOMEM; + btrfs_abort_transaction(trans, ret); goto out; } - /* - * Take the device list mutex to prevent races with the final phase of - * a device replace operation that replaces the device object associated - * with the map's stripes, because the device object's id can change - * at any time during that final phase of the device replace operation - * (dev-replace.c:btrfs_dev_replace_finishing()). - */ - mutex_lock(&fs_info->fs_devices->device_list_mutex); for (i = 0; i < map->num_stripes; i++) { - device = map->stripes[i].dev; - dev_offset = map->stripes[i].physical; + struct btrfs_device *device = map->stripes[i].dev; ret = btrfs_update_device(trans, device); if (ret) - break; - ret = btrfs_alloc_dev_extent(trans, device, chunk_offset, - dev_offset, stripe_size); - if (ret) - break; - } - if (ret) { - mutex_unlock(&fs_info->fs_devices->device_list_mutex); - goto out; + goto out; } stripe = &chunk->stripe; for (i = 0; i < map->num_stripes; i++) { - device = map->stripes[i].dev; - dev_offset = map->stripes[i].physical; + struct btrfs_device *device = map->stripes[i].dev; + const u64 dev_offset = map->stripes[i].physical; btrfs_set_stack_stripe_devid(stripe, device->devid); btrfs_set_stack_stripe_offset(stripe, dev_offset); memcpy(stripe->dev_uuid, device->uuid, BTRFS_UUID_SIZE); stripe++; } - mutex_unlock(&fs_info->fs_devices->device_list_mutex); - btrfs_set_stack_chunk_length(chunk, chunk_size); - btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid); - btrfs_set_stack_chunk_stripe_len(chunk, map->stripe_len); + btrfs_set_stack_chunk_length(chunk, bg->length); + btrfs_set_stack_chunk_owner(chunk, BTRFS_EXTENT_TREE_OBJECTID); + btrfs_set_stack_chunk_stripe_len(chunk, BTRFS_STRIPE_LEN); btrfs_set_stack_chunk_type(chunk, map->type); btrfs_set_stack_chunk_num_stripes(chunk, map->num_stripes); - btrfs_set_stack_chunk_io_align(chunk, map->stripe_len); - btrfs_set_stack_chunk_io_width(chunk, map->stripe_len); + btrfs_set_stack_chunk_io_align(chunk, BTRFS_STRIPE_LEN); + btrfs_set_stack_chunk_io_width(chunk, BTRFS_STRIPE_LEN); btrfs_set_stack_chunk_sector_size(chunk, fs_info->sectorsize); btrfs_set_stack_chunk_sub_stripes(chunk, map->sub_stripes); key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID; key.type = BTRFS_CHUNK_ITEM_KEY; - key.offset = chunk_offset; + key.offset = bg->start; ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size); - if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) { - /* - * TODO: Cleanup of inserted chunk root in case of - * failure. - */ + if (ret) + goto out; + + set_bit(BLOCK_GROUP_FLAG_CHUNK_ITEM_INSERTED, &bg->runtime_flags); + + if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) { ret = btrfs_add_system_chunk(fs_info, &key, chunk, item_size); + if (ret) + goto out; } out: kfree(chunk); - free_extent_map(em); + btrfs_free_chunk_map(map); return ret; } @@ -5279,38 +5715,73 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans) { struct btrfs_fs_info *fs_info = trans->fs_info; u64 alloc_profile; - int ret; + struct btrfs_block_group *meta_bg; + struct btrfs_space_info *meta_space_info; + struct btrfs_block_group *sys_bg; + struct btrfs_space_info *sys_space_info; + + /* + * When adding a new device for sprouting, the seed device is read-only + * so we must first allocate a metadata and a system chunk. But before + * adding the block group items to the extent, device and chunk btrees, + * we must first: + * + * 1) Create both chunks without doing any changes to the btrees, as + * otherwise we would get -ENOSPC since the block groups from the + * seed device are read-only; + * + * 2) Add the device item for the new sprout device - finishing the setup + * of a new block group requires updating the device item in the chunk + * btree, so it must exist when we attempt to do it. The previous step + * ensures this does not fail with -ENOSPC. + * + * After that we can add the block group items to their btrees: + * update existing device item in the chunk btree, add a new block group + * item to the extent btree, add a new chunk item to the chunk btree and + * finally add the new device extent items to the devices btree. + */ alloc_profile = btrfs_metadata_alloc_profile(fs_info); - ret = btrfs_alloc_chunk(trans, alloc_profile); - if (ret) - return ret; + meta_space_info = btrfs_find_space_info(fs_info, alloc_profile); + if (!meta_space_info) { + DEBUG_WARN(); + return -EINVAL; + } + meta_bg = btrfs_create_chunk(trans, meta_space_info, alloc_profile); + if (IS_ERR(meta_bg)) + return PTR_ERR(meta_bg); alloc_profile = btrfs_system_alloc_profile(fs_info); - ret = btrfs_alloc_chunk(trans, alloc_profile); - return ret; + sys_space_info = btrfs_find_space_info(fs_info, alloc_profile); + if (!sys_space_info) { + DEBUG_WARN(); + return -EINVAL; + } + sys_bg = btrfs_create_chunk(trans, sys_space_info, alloc_profile); + if (IS_ERR(sys_bg)) + return PTR_ERR(sys_bg); + + return 0; } -static inline int btrfs_chunk_max_errors(struct map_lookup *map) +static inline int btrfs_chunk_max_errors(struct btrfs_chunk_map *map) { const int index = btrfs_bg_flags_to_raid_index(map->type); return btrfs_raid_array[index].tolerated_failures; } -int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset) +bool btrfs_chunk_writeable(struct btrfs_fs_info *fs_info, u64 chunk_offset) { - struct extent_map *em; - struct map_lookup *map; - int readonly = 0; + struct btrfs_chunk_map *map; int miss_ndevs = 0; int i; + bool ret = true; - em = btrfs_get_chunk_map(fs_info, chunk_offset, 1); - if (IS_ERR(em)) - return 1; + map = btrfs_get_chunk_map(fs_info, chunk_offset, 1); + if (IS_ERR(map)) + return false; - map = em->map_lookup; for (i = 0; i < map->num_stripes; i++) { if (test_bit(BTRFS_DEV_STATE_MISSING, &map->stripes[i].dev->dev_state)) { @@ -5319,50 +5790,69 @@ int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset) } if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &map->stripes[i].dev->dev_state)) { - readonly = 1; + ret = false; goto end; } } /* - * If the number of missing devices is larger than max errors, - * we can not write the data into that chunk successfully, so - * set it readonly. + * If the number of missing devices is larger than max errors, we can + * not write the data into that chunk successfully. */ if (miss_ndevs > btrfs_chunk_max_errors(map)) - readonly = 1; + ret = false; end: - free_extent_map(em); - return readonly; + btrfs_free_chunk_map(map); + return ret; } -void btrfs_mapping_tree_free(struct extent_map_tree *tree) +void btrfs_mapping_tree_free(struct btrfs_fs_info *fs_info) { - struct extent_map *em; + write_lock(&fs_info->mapping_tree_lock); + while (!RB_EMPTY_ROOT(&fs_info->mapping_tree.rb_root)) { + struct btrfs_chunk_map *map; + struct rb_node *node; - while (1) { - write_lock(&tree->lock); - em = lookup_extent_mapping(tree, 0, (u64)-1); - if (em) - remove_extent_mapping(tree, em); - write_unlock(&tree->lock); - if (!em) - break; - /* once for us */ - free_extent_map(em); - /* once for the tree */ - free_extent_map(em); + node = rb_first_cached(&fs_info->mapping_tree); + map = rb_entry(node, struct btrfs_chunk_map, rb_node); + rb_erase_cached(&map->rb_node, &fs_info->mapping_tree); + RB_CLEAR_NODE(&map->rb_node); + chunk_map_device_clear_bits(map, CHUNK_ALLOCATED); + /* Once for the tree ref. */ + btrfs_free_chunk_map(map); + cond_resched_rwlock_write(&fs_info->mapping_tree_lock); } + write_unlock(&fs_info->mapping_tree_lock); +} + +static int btrfs_chunk_map_num_copies(const struct btrfs_chunk_map *map) +{ + enum btrfs_raid_types index = btrfs_bg_flags_to_raid_index(map->type); + + if (map->type & BTRFS_BLOCK_GROUP_RAID5) + return 2; + + /* + * There could be two corrupted data stripes, we need to loop retry in + * order to rebuild the correct data. + * + * Fail a stripe at a time on every retry except the stripe under + * reconstruction. + */ + if (map->type & BTRFS_BLOCK_GROUP_RAID6) + return map->num_stripes; + + /* Non-RAID56, use their ncopies from btrfs_raid_array. */ + return btrfs_raid_array[index].ncopies; } int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len) { - struct extent_map *em; - struct map_lookup *map; + struct btrfs_chunk_map *map; int ret; - em = btrfs_get_chunk_map(fs_info, logical, len); - if (IS_ERR(em)) + map = btrfs_get_chunk_map(fs_info, logical, len); + if (IS_ERR(map)) /* * We could return errors for these cases, but that could get * ugly and we'd probably do the same thing which is just not do @@ -5371,89 +5861,138 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len) */ return 1; - map = em->map_lookup; - if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1_MASK)) - ret = map->num_stripes; - else if (map->type & BTRFS_BLOCK_GROUP_RAID10) - ret = map->sub_stripes; - else if (map->type & BTRFS_BLOCK_GROUP_RAID5) - ret = 2; - else if (map->type & BTRFS_BLOCK_GROUP_RAID6) - /* - * There could be two corrupted data stripes, we need - * to loop retry in order to rebuild the correct data. - * - * Fail a stripe at a time on every retry except the - * stripe under reconstruction. - */ - ret = map->num_stripes; - else - ret = 1; - free_extent_map(em); - - down_read(&fs_info->dev_replace.rwsem); - if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace) && - fs_info->dev_replace.tgtdev) - ret++; - up_read(&fs_info->dev_replace.rwsem); - + ret = btrfs_chunk_map_num_copies(map); + btrfs_free_chunk_map(map); return ret; } unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info, u64 logical) { - struct extent_map *em; - struct map_lookup *map; + struct btrfs_chunk_map *map; unsigned long len = fs_info->sectorsize; - em = btrfs_get_chunk_map(fs_info, logical, len); + if (!btrfs_fs_incompat(fs_info, RAID56)) + return len; - if (!WARN_ON(IS_ERR(em))) { - map = em->map_lookup; + map = btrfs_get_chunk_map(fs_info, logical, len); + + if (!WARN_ON(IS_ERR(map))) { if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) - len = map->stripe_len * nr_data_stripes(map); - free_extent_map(em); + len = btrfs_stripe_nr_to_offset(nr_data_stripes(map)); + btrfs_free_chunk_map(map); } return len; } -int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info, u64 logical, u64 len) +#ifdef CONFIG_BTRFS_EXPERIMENTAL +static int btrfs_read_preferred(struct btrfs_chunk_map *map, int first, int num_stripes) { - struct extent_map *em; - struct map_lookup *map; - int ret = 0; + for (int index = first; index < first + num_stripes; index++) { + const struct btrfs_device *device = map->stripes[index].dev; - em = btrfs_get_chunk_map(fs_info, logical, len); + if (device->devid == READ_ONCE(device->fs_devices->read_devid)) + return index; + } - if(!WARN_ON(IS_ERR(em))) { - map = em->map_lookup; - if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) - ret = 1; - free_extent_map(em); + /* If no read-preferred device is set use the first stripe. */ + return first; +} + +struct stripe_mirror { + u64 devid; + int num; +}; + +static int btrfs_cmp_devid(const void *a, const void *b) +{ + const struct stripe_mirror *s1 = (const struct stripe_mirror *)a; + const struct stripe_mirror *s2 = (const struct stripe_mirror *)b; + + if (s1->devid < s2->devid) + return -1; + if (s1->devid > s2->devid) + return 1; + return 0; +} + +/* + * Select a stripe for reading using the round-robin algorithm. + * + * 1. Compute the read cycle as the total sectors read divided by the minimum + * sectors per device. + * 2. Determine the stripe number for the current read by taking the modulus + * of the read cycle with the total number of stripes: + * + * stripe index = (total sectors / min sectors per dev) % num stripes + * + * The calculated stripe index is then used to select the corresponding device + * from the list of devices, which is ordered by devid. + */ +static int btrfs_read_rr(const struct btrfs_chunk_map *map, int first, int num_stripes) +{ + struct stripe_mirror stripes[BTRFS_RAID1_MAX_MIRRORS] = { 0 }; + struct btrfs_device *device = map->stripes[first].dev; + struct btrfs_fs_info *fs_info = device->fs_devices->fs_info; + unsigned int read_cycle; + unsigned int total_reads; + unsigned int min_reads_per_dev; + + total_reads = percpu_counter_sum(&fs_info->stats_read_blocks); + min_reads_per_dev = READ_ONCE(fs_info->fs_devices->rr_min_contig_read) >> + fs_info->sectorsize_bits; + + for (int index = 0, i = first; i < first + num_stripes; i++) { + stripes[index].devid = map->stripes[i].dev->devid; + stripes[index].num = i; + index++; } - return ret; + sort(stripes, num_stripes, sizeof(struct stripe_mirror), + btrfs_cmp_devid, NULL); + + read_cycle = total_reads / min_reads_per_dev; + return stripes[read_cycle % num_stripes].num; } +#endif static int find_live_mirror(struct btrfs_fs_info *fs_info, - struct map_lookup *map, int first, - int dev_replace_is_ongoing) + struct btrfs_chunk_map *map, int first, + bool dev_replace_is_ongoing) { + const enum btrfs_read_policy policy = READ_ONCE(fs_info->fs_devices->read_policy); int i; int num_stripes; int preferred_mirror; int tolerance; struct btrfs_device *srcdev; - ASSERT((map->type & - (BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_RAID10))); + ASSERT((map->type & (BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_RAID10)), + "type=%llu", map->type); if (map->type & BTRFS_BLOCK_GROUP_RAID10) num_stripes = map->sub_stripes; else num_stripes = map->num_stripes; - preferred_mirror = first + current->pid % num_stripes; + switch (policy) { + default: + /* Shouldn't happen, just warn and use pid instead of failing */ + btrfs_warn_rl(fs_info, "unknown read_policy type %u, reset to pid", + policy); + WRITE_ONCE(fs_info->fs_devices->read_policy, BTRFS_READ_POLICY_PID); + fallthrough; + case BTRFS_READ_POLICY_PID: + preferred_mirror = first + (current->pid % num_stripes); + break; +#ifdef CONFIG_BTRFS_EXPERIMENTAL + case BTRFS_READ_POLICY_RR: + preferred_mirror = btrfs_read_rr(map, first, num_stripes); + break; + case BTRFS_READ_POLICY_DEVID: + preferred_mirror = btrfs_read_preferred(map, first, num_stripes); + break; +#endif + } if (dev_replace_is_ongoing && fs_info->dev_replace.cont_reading_from_srcdev_mode == @@ -5484,130 +6023,101 @@ static int find_live_mirror(struct btrfs_fs_info *fs_info, return preferred_mirror; } -/* Bubble-sort the stripe set to put the parity/syndrome stripes last */ -static void sort_parity_stripes(struct btrfs_bio *bbio, int num_stripes) +EXPORT_FOR_TESTS +struct btrfs_io_context *alloc_btrfs_io_context(struct btrfs_fs_info *fs_info, + u64 logical, u16 total_stripes) { - int i; - int again = 1; - - while (again) { - again = 0; - for (i = 0; i < num_stripes - 1; i++) { - /* Swap if parity is on a smaller index */ - if (bbio->raid_map[i] > bbio->raid_map[i + 1]) { - swap(bbio->stripes[i], bbio->stripes[i + 1]); - swap(bbio->raid_map[i], bbio->raid_map[i + 1]); - again = 1; - } - } - } -} + struct btrfs_io_context *bioc; -static struct btrfs_bio *alloc_btrfs_bio(int total_stripes, int real_stripes) -{ - struct btrfs_bio *bbio = kzalloc( - /* the size of the btrfs_bio */ - sizeof(struct btrfs_bio) + - /* plus the variable array for the stripes */ - sizeof(struct btrfs_bio_stripe) * (total_stripes) + - /* plus the variable array for the tgt dev */ - sizeof(int) * (real_stripes) + - /* - * plus the raid_map, which includes both the tgt dev - * and the stripes - */ - sizeof(u64) * (total_stripes), - GFP_NOFS|__GFP_NOFAIL); + bioc = kzalloc(struct_size(bioc, stripes, total_stripes), GFP_NOFS); + + if (!bioc) + return NULL; + + refcount_set(&bioc->refs, 1); - atomic_set(&bbio->error, 0); - refcount_set(&bbio->refs, 1); + bioc->fs_info = fs_info; + bioc->replace_stripe_src = -1; + bioc->full_stripe_logical = (u64)-1; + bioc->logical = logical; - return bbio; + return bioc; } -void btrfs_get_bbio(struct btrfs_bio *bbio) +void btrfs_get_bioc(struct btrfs_io_context *bioc) { - WARN_ON(!refcount_read(&bbio->refs)); - refcount_inc(&bbio->refs); + WARN_ON(!refcount_read(&bioc->refs)); + refcount_inc(&bioc->refs); } -void btrfs_put_bbio(struct btrfs_bio *bbio) +void btrfs_put_bioc(struct btrfs_io_context *bioc) { - if (!bbio) + if (!bioc) return; - if (refcount_dec_and_test(&bbio->refs)) - kfree(bbio); + if (refcount_dec_and_test(&bioc->refs)) + kfree(bioc); } -/* can REQ_OP_DISCARD be sent with other REQ like REQ_OP_WRITE? */ /* * Please note that, discard won't be sent to target device of device * replace. */ -static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info, - u64 logical, u64 *length_ret, - struct btrfs_bio **bbio_ret) +struct btrfs_discard_stripe *btrfs_map_discard(struct btrfs_fs_info *fs_info, + u64 logical, u64 *length_ret, + u32 *num_stripes) { - struct extent_map *em; - struct map_lookup *map; - struct btrfs_bio *bbio; + struct btrfs_chunk_map *map; + struct btrfs_discard_stripe *stripes; u64 length = *length_ret; u64 offset; - u64 stripe_nr; - u64 stripe_nr_end; + u32 stripe_nr; + u32 stripe_nr_end; + u32 stripe_cnt; u64 stripe_end_offset; - u64 stripe_cnt; - u64 stripe_len; u64 stripe_offset; - u64 num_stripes; u32 stripe_index; u32 factor = 0; u32 sub_stripes = 0; - u64 stripes_per_dev = 0; + u32 stripes_per_dev = 0; u32 remaining_stripes = 0; u32 last_stripe = 0; - int ret = 0; + int ret; int i; - /* discard always return a bbio */ - ASSERT(bbio_ret); - - em = btrfs_get_chunk_map(fs_info, logical, length); - if (IS_ERR(em)) - return PTR_ERR(em); + map = btrfs_get_chunk_map(fs_info, logical, length); + if (IS_ERR(map)) + return ERR_CAST(map); - map = em->map_lookup; /* we don't discard raid56 yet */ if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) { ret = -EOPNOTSUPP; - goto out; + goto out_free_map; } - offset = logical - em->start; - length = min_t(u64, em->start + em->len - logical, length); + offset = logical - map->start; + length = min_t(u64, map->start + map->chunk_len - logical, length); *length_ret = length; - stripe_len = map->stripe_len; /* * stripe_nr counts the total number of stripes we have to stride * to get to this block */ - stripe_nr = div64_u64(offset, stripe_len); + stripe_nr = offset >> BTRFS_STRIPE_LEN_SHIFT; /* stripe_offset is the offset of this block in its stripe */ - stripe_offset = offset - stripe_nr * stripe_len; + stripe_offset = offset - btrfs_stripe_nr_to_offset(stripe_nr); - stripe_nr_end = round_up(offset + length, map->stripe_len); - stripe_nr_end = div64_u64(stripe_nr_end, map->stripe_len); + stripe_nr_end = round_up(offset + length, BTRFS_STRIPE_LEN) >> + BTRFS_STRIPE_LEN_SHIFT; stripe_cnt = stripe_nr_end - stripe_nr; - stripe_end_offset = stripe_nr_end * map->stripe_len - + stripe_end_offset = btrfs_stripe_nr_to_offset(stripe_nr_end) - (offset + length); /* * after this, stripe_nr is the number of stripes on this * device we have to walk to find the data, and stripe_index is * the number of our device in the stripe array */ - num_stripes = 1; + *num_stripes = 1; stripe_index = 0; if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) { @@ -5617,42 +6127,41 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info, sub_stripes = map->sub_stripes; factor = map->num_stripes / sub_stripes; - num_stripes = min_t(u64, map->num_stripes, + *num_stripes = min_t(u64, map->num_stripes, sub_stripes * stripe_cnt); - stripe_nr = div_u64_rem(stripe_nr, factor, &stripe_index); + stripe_index = stripe_nr % factor; + stripe_nr /= factor; stripe_index *= sub_stripes; - stripes_per_dev = div_u64_rem(stripe_cnt, factor, - &remaining_stripes); - div_u64_rem(stripe_nr_end - 1, factor, &last_stripe); - last_stripe *= sub_stripes; + + remaining_stripes = stripe_cnt % factor; + stripes_per_dev = stripe_cnt / factor; + last_stripe = ((stripe_nr_end - 1) % factor) * sub_stripes; } else if (map->type & (BTRFS_BLOCK_GROUP_RAID1_MASK | BTRFS_BLOCK_GROUP_DUP)) { - num_stripes = map->num_stripes; + *num_stripes = map->num_stripes; } else { - stripe_nr = div_u64_rem(stripe_nr, map->num_stripes, - &stripe_index); + stripe_index = stripe_nr % map->num_stripes; + stripe_nr /= map->num_stripes; } - bbio = alloc_btrfs_bio(num_stripes, 0); - if (!bbio) { + stripes = kcalloc(*num_stripes, sizeof(*stripes), GFP_NOFS); + if (!stripes) { ret = -ENOMEM; - goto out; + goto out_free_map; } - for (i = 0; i < num_stripes; i++) { - bbio->stripes[i].physical = + for (i = 0; i < *num_stripes; i++) { + stripes[i].physical = map->stripes[stripe_index].physical + - stripe_offset + stripe_nr * map->stripe_len; - bbio->stripes[i].dev = map->stripes[stripe_index].dev; + stripe_offset + btrfs_stripe_nr_to_offset(stripe_nr); + stripes[i].dev = map->stripes[stripe_index].dev; if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) { - bbio->stripes[i].length = stripes_per_dev * - map->stripe_len; + stripes[i].length = btrfs_stripe_nr_to_offset(stripes_per_dev); if (i / sub_stripes < remaining_stripes) - bbio->stripes[i].length += - map->stripe_len; + stripes[i].length += BTRFS_STRIPE_LEN; /* * Special for the first stripe and @@ -5663,19 +6172,17 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info, * off end_off */ if (i < sub_stripes) - bbio->stripes[i].length -= - stripe_offset; + stripes[i].length -= stripe_offset; if (stripe_index >= last_stripe && stripe_index <= (last_stripe + sub_stripes - 1)) - bbio->stripes[i].length -= - stripe_end_offset; + stripes[i].length -= stripe_end_offset; if (i == sub_stripes - 1) stripe_offset = 0; } else { - bbio->stripes[i].length = length; + stripes[i].length = length; } stripe_index++; @@ -5685,746 +6192,583 @@ static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info, } } - *bbio_ret = bbio; - bbio->map_type = map->type; - bbio->num_stripes = num_stripes; -out: - free_extent_map(em); + btrfs_free_chunk_map(map); + return stripes; +out_free_map: + btrfs_free_chunk_map(map); + return ERR_PTR(ret); +} + +static bool is_block_group_to_copy(struct btrfs_fs_info *fs_info, u64 logical) +{ + struct btrfs_block_group *cache; + bool ret; + + /* Non zoned filesystem does not use "to_copy" flag */ + if (!btrfs_is_zoned(fs_info)) + return false; + + cache = btrfs_lookup_block_group(fs_info, logical); + + ret = test_bit(BLOCK_GROUP_FLAG_TO_COPY, &cache->runtime_flags); + + btrfs_put_block_group(cache); return ret; } -/* - * In dev-replace case, for repair case (that's the only case where the mirror - * is selected explicitly when calling btrfs_map_block), blocks left of the - * left cursor can also be read from the target drive. - * - * For REQ_GET_READ_MIRRORS, the target drive is added as the last one to the - * array of stripes. - * For READ, it also needs to be supported using the same mirror number. - * - * If the requested block is not left of the left cursor, EIO is returned. This - * can happen because btrfs_num_copies() returns one more in the dev-replace - * case. - */ -static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info, - u64 logical, u64 length, - u64 srcdev_devid, int *mirror_num, - u64 *physical) +static void handle_ops_on_dev_replace(struct btrfs_io_context *bioc, + struct btrfs_dev_replace *dev_replace, + u64 logical, + struct btrfs_io_geometry *io_geom) { - struct btrfs_bio *bbio = NULL; - int num_stripes; - int index_srcdev = 0; - int found = 0; - u64 physical_of_found = 0; + u64 srcdev_devid = dev_replace->srcdev->devid; + /* + * At this stage, num_stripes is still the real number of stripes, + * excluding the duplicated stripes. + */ + int num_stripes = io_geom->num_stripes; + int max_errors = io_geom->max_errors; + int nr_extra_stripes = 0; int i; - int ret = 0; - - ret = __btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, - logical, &length, &bbio, 0, 0); - if (ret) { - ASSERT(bbio == NULL); - return ret; - } - num_stripes = bbio->num_stripes; - if (*mirror_num > num_stripes) { - /* - * BTRFS_MAP_GET_READ_MIRRORS does not contain this mirror, - * that means that the requested area is not left of the left - * cursor - */ - btrfs_put_bbio(bbio); - return -EIO; - } + /* + * A block group which has "to_copy" set will eventually be copied by + * the dev-replace process. We can avoid cloning IO here. + */ + if (is_block_group_to_copy(dev_replace->srcdev->fs_info, logical)) + return; /* - * process the rest of the function using the mirror_num of the source - * drive. Therefore look it up first. At the end, patch the device - * pointer to the one of the target drive. + * Duplicate the write operations while the dev-replace procedure is + * running. Since the copying of the old disk to the new disk takes + * place at run time while the filesystem is mounted writable, the + * regular write operations to the old disk have to be duplicated to go + * to the new disk as well. + * + * Note that device->missing is handled by the caller, and that the + * write to the old disk is already set up in the stripes array. */ for (i = 0; i < num_stripes; i++) { - if (bbio->stripes[i].dev->devid != srcdev_devid) - continue; + struct btrfs_io_stripe *old = &bioc->stripes[i]; + struct btrfs_io_stripe *new = &bioc->stripes[num_stripes + nr_extra_stripes]; - /* - * In case of DUP, in order to keep it simple, only add the - * mirror with the lowest physical address - */ - if (found && - physical_of_found <= bbio->stripes[i].physical) + if (old->dev->devid != srcdev_devid) continue; - index_srcdev = i; - found = 1; - physical_of_found = bbio->stripes[i].physical; + new->physical = old->physical; + new->dev = dev_replace->tgtdev; + if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) + bioc->replace_stripe_src = i; + nr_extra_stripes++; } - btrfs_put_bbio(bbio); + /* We can only have at most 2 extra nr_stripes (for DUP). */ + ASSERT(nr_extra_stripes <= 2, "nr_extra_stripes=%d", nr_extra_stripes); + /* + * For GET_READ_MIRRORS, we can only return at most 1 extra stripe for + * replace. + * If we have 2 extra stripes, only choose the one with smaller physical. + */ + if (io_geom->op == BTRFS_MAP_GET_READ_MIRRORS && nr_extra_stripes == 2) { + struct btrfs_io_stripe *first = &bioc->stripes[num_stripes]; + struct btrfs_io_stripe *second = &bioc->stripes[num_stripes + 1]; - ASSERT(found); - if (!found) - return -EIO; + /* Only DUP can have two extra stripes. */ + ASSERT(bioc->map_type & BTRFS_BLOCK_GROUP_DUP, + "map_type=%llu", bioc->map_type); - *mirror_num = index_srcdev + 1; - *physical = physical_of_found; - return ret; + /* + * Swap the last stripe stripes and reduce @nr_extra_stripes. + * The extra stripe would still be there, but won't be accessed. + */ + if (first->physical > second->physical) { + swap(second->physical, first->physical); + swap(second->dev, first->dev); + nr_extra_stripes--; + } + } + + io_geom->num_stripes = num_stripes + nr_extra_stripes; + io_geom->max_errors = max_errors + nr_extra_stripes; + bioc->replace_nr_stripes = nr_extra_stripes; } -static void handle_ops_on_dev_replace(enum btrfs_map_op op, - struct btrfs_bio **bbio_ret, - struct btrfs_dev_replace *dev_replace, - int *num_stripes_ret, int *max_errors_ret) +static u64 btrfs_max_io_len(struct btrfs_chunk_map *map, u64 offset, + struct btrfs_io_geometry *io_geom) { - struct btrfs_bio *bbio = *bbio_ret; - u64 srcdev_devid = dev_replace->srcdev->devid; - int tgtdev_indexes = 0; - int num_stripes = *num_stripes_ret; - int max_errors = *max_errors_ret; - int i; + /* + * Stripe_nr is the stripe where this block falls. stripe_offset is + * the offset of this block in its stripe. + */ + io_geom->stripe_offset = offset & BTRFS_STRIPE_LEN_MASK; + io_geom->stripe_nr = offset >> BTRFS_STRIPE_LEN_SHIFT; + ASSERT(io_geom->stripe_offset < U32_MAX, + "stripe_offset=%llu", io_geom->stripe_offset); - if (op == BTRFS_MAP_WRITE) { - int index_where_to_add; + if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) { + unsigned long full_stripe_len = + btrfs_stripe_nr_to_offset(nr_data_stripes(map)); /* - * duplicate the write operations while the dev replace - * procedure is running. Since the copying of the old disk to - * the new disk takes place at run time while the filesystem is - * mounted writable, the regular write operations to the old - * disk have to be duplicated to go to the new disk as well. + * For full stripe start, we use previously calculated + * @stripe_nr. Align it to nr_data_stripes, then multiply with + * STRIPE_LEN. * - * Note that device->missing is handled by the caller, and that - * the write to the old disk is already set up in the stripes - * array. + * By this we can avoid u64 division completely. And we have + * to go rounddown(), not round_down(), as nr_data_stripes is + * not ensured to be power of 2. */ - index_where_to_add = num_stripes; - for (i = 0; i < num_stripes; i++) { - if (bbio->stripes[i].dev->devid == srcdev_devid) { - /* write to new disk, too */ - struct btrfs_bio_stripe *new = - bbio->stripes + index_where_to_add; - struct btrfs_bio_stripe *old = - bbio->stripes + i; - - new->physical = old->physical; - new->length = old->length; - new->dev = dev_replace->tgtdev; - bbio->tgtdev_map[i] = index_where_to_add; - index_where_to_add++; - max_errors++; - tgtdev_indexes++; - } - } - num_stripes = index_where_to_add; - } else if (op == BTRFS_MAP_GET_READ_MIRRORS) { - int index_srcdev = 0; - int found = 0; - u64 physical_of_found = 0; - + io_geom->raid56_full_stripe_start = btrfs_stripe_nr_to_offset( + rounddown(io_geom->stripe_nr, nr_data_stripes(map))); + + ASSERT(io_geom->raid56_full_stripe_start + full_stripe_len > offset, + "raid56_full_stripe_start=%llu full_stripe_len=%lu offset=%llu", + io_geom->raid56_full_stripe_start, full_stripe_len, offset); + ASSERT(io_geom->raid56_full_stripe_start <= offset, + "raid56_full_stripe_start=%llu offset=%llu", + io_geom->raid56_full_stripe_start, offset); /* - * During the dev-replace procedure, the target drive can also - * be used to read data in case it is needed to repair a corrupt - * block elsewhere. This is possible if the requested area is - * left of the left cursor. In this area, the target drive is a - * full copy of the source drive. + * For writes to RAID56, allow to write a full stripe set, but + * no straddling of stripe sets. */ - for (i = 0; i < num_stripes; i++) { - if (bbio->stripes[i].dev->devid == srcdev_devid) { - /* - * In case of DUP, in order to keep it simple, - * only add the mirror with the lowest physical - * address - */ - if (found && - physical_of_found <= - bbio->stripes[i].physical) - continue; - index_srcdev = i; - found = 1; - physical_of_found = bbio->stripes[i].physical; - } - } - if (found) { - struct btrfs_bio_stripe *tgtdev_stripe = - bbio->stripes + num_stripes; - - tgtdev_stripe->physical = physical_of_found; - tgtdev_stripe->length = - bbio->stripes[index_srcdev].length; - tgtdev_stripe->dev = dev_replace->tgtdev; - bbio->tgtdev_map[index_srcdev] = num_stripes; - - tgtdev_indexes++; - num_stripes++; - } + if (io_geom->op == BTRFS_MAP_WRITE) + return full_stripe_len - (offset - io_geom->raid56_full_stripe_start); } - *num_stripes_ret = num_stripes; - *max_errors_ret = max_errors; - bbio->num_tgtdevs = tgtdev_indexes; - *bbio_ret = bbio; + /* + * For other RAID types and for RAID56 reads, allow a single stripe (on + * a single disk). + */ + if (map->type & BTRFS_BLOCK_GROUP_STRIPE_MASK) + return BTRFS_STRIPE_LEN - io_geom->stripe_offset; + return U64_MAX; } -static bool need_full_stripe(enum btrfs_map_op op) +static int set_io_stripe(struct btrfs_fs_info *fs_info, u64 logical, + u64 *length, struct btrfs_io_stripe *dst, + struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom) { - return (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS); + dst->dev = map->stripes[io_geom->stripe_index].dev; + + if (io_geom->op == BTRFS_MAP_READ && io_geom->use_rst) + return btrfs_get_raid_extent_offset(fs_info, logical, length, + map->type, + io_geom->stripe_index, dst); + + dst->physical = map->stripes[io_geom->stripe_index].physical + + io_geom->stripe_offset + + btrfs_stripe_nr_to_offset(io_geom->stripe_nr); + return 0; } -/* - * btrfs_get_io_geometry - calculates the geomery of a particular (address, len) - * tuple. This information is used to calculate how big a - * particular bio can get before it straddles a stripe. - * - * @fs_info - the filesystem - * @logical - address that we want to figure out the geometry of - * @len - the length of IO we are going to perform, starting at @logical - * @op - type of operation - write or read - * @io_geom - pointer used to return values - * - * Returns < 0 in case a chunk for the given logical address cannot be found, - * usually shouldn't happen unless @logical is corrupted, 0 otherwise. - */ -int btrfs_get_io_geometry(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, - u64 logical, u64 len, struct btrfs_io_geometry *io_geom) +static bool is_single_device_io(struct btrfs_fs_info *fs_info, + const struct btrfs_io_stripe *smap, + const struct btrfs_chunk_map *map, + int num_alloc_stripes, + struct btrfs_io_geometry *io_geom) { - struct extent_map *em; - struct map_lookup *map; - u64 offset; - u64 stripe_offset; - u64 stripe_nr; - u64 stripe_len; - u64 raid56_full_stripe_start = (u64)-1; - int data_stripes; - int ret = 0; + if (!smap) + return false; - ASSERT(op != BTRFS_MAP_DISCARD); - - em = btrfs_get_chunk_map(fs_info, logical, len); - if (IS_ERR(em)) - return PTR_ERR(em); - - map = em->map_lookup; - /* Offset of this logical address in the chunk */ - offset = logical - em->start; - /* Len of a stripe in a chunk */ - stripe_len = map->stripe_len; - /* Stripe wher this block falls in */ - stripe_nr = div64_u64(offset, stripe_len); - /* Offset of stripe in the chunk */ - stripe_offset = stripe_nr * stripe_len; - if (offset < stripe_offset) { - btrfs_crit(fs_info, -"stripe math has gone wrong, stripe_offset=%llu offset=%llu start=%llu logical=%llu stripe_len=%llu", - stripe_offset, offset, em->start, logical, stripe_len); - ret = -EINVAL; - goto out; + if (num_alloc_stripes != 1) + return false; + + if (io_geom->use_rst && io_geom->op != BTRFS_MAP_READ) + return false; + + if ((map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) && io_geom->mirror_num > 1) + return false; + + return true; +} + +static void map_blocks_raid0(const struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom) +{ + io_geom->stripe_index = io_geom->stripe_nr % map->num_stripes; + io_geom->stripe_nr /= map->num_stripes; + if (io_geom->op == BTRFS_MAP_READ) + io_geom->mirror_num = 1; +} + +static void map_blocks_raid1(struct btrfs_fs_info *fs_info, + struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom, + bool dev_replace_is_ongoing) +{ + if (io_geom->op != BTRFS_MAP_READ) { + io_geom->num_stripes = map->num_stripes; + return; } - /* stripe_offset is the offset of this block in its stripe */ - stripe_offset = offset - stripe_offset; - data_stripes = nr_data_stripes(map); + if (io_geom->mirror_num) { + io_geom->stripe_index = io_geom->mirror_num - 1; + return; + } - if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) { - u64 max_len = stripe_len - stripe_offset; + io_geom->stripe_index = find_live_mirror(fs_info, map, 0, + dev_replace_is_ongoing); + io_geom->mirror_num = io_geom->stripe_index + 1; +} - /* - * In case of raid56, we need to know the stripe aligned start - */ - if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) { - unsigned long full_stripe_len = stripe_len * data_stripes; - raid56_full_stripe_start = offset; +static void map_blocks_dup(const struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom) +{ + if (io_geom->op != BTRFS_MAP_READ) { + io_geom->num_stripes = map->num_stripes; + return; + } - /* - * Allow a write of a full stripe, but make sure we - * don't allow straddling of stripes - */ - raid56_full_stripe_start = div64_u64(raid56_full_stripe_start, - full_stripe_len); - raid56_full_stripe_start *= full_stripe_len; + if (io_geom->mirror_num) { + io_geom->stripe_index = io_geom->mirror_num - 1; + return; + } - /* - * For writes to RAID[56], allow a full stripeset across - * all disks. For other RAID types and for RAID[56] - * reads, just allow a single stripe (on a single disk). - */ - if (op == BTRFS_MAP_WRITE) { - max_len = stripe_len * data_stripes - - (offset - raid56_full_stripe_start); - } - } - len = min_t(u64, em->len - offset, max_len); - } else { - len = em->len - offset; + io_geom->mirror_num = 1; +} + +static void map_blocks_raid10(struct btrfs_fs_info *fs_info, + struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom, + bool dev_replace_is_ongoing) +{ + u32 factor = map->num_stripes / map->sub_stripes; + int old_stripe_index; + + io_geom->stripe_index = (io_geom->stripe_nr % factor) * map->sub_stripes; + io_geom->stripe_nr /= factor; + + if (io_geom->op != BTRFS_MAP_READ) { + io_geom->num_stripes = map->sub_stripes; + return; } - io_geom->len = len; - io_geom->offset = offset; - io_geom->stripe_len = stripe_len; - io_geom->stripe_nr = stripe_nr; - io_geom->stripe_offset = stripe_offset; - io_geom->raid56_stripe_offset = raid56_full_stripe_start; + if (io_geom->mirror_num) { + io_geom->stripe_index += io_geom->mirror_num - 1; + return; + } -out: - /* once for us */ - free_extent_map(em); - return ret; + old_stripe_index = io_geom->stripe_index; + io_geom->stripe_index = find_live_mirror(fs_info, map, + io_geom->stripe_index, + dev_replace_is_ongoing); + io_geom->mirror_num = io_geom->stripe_index - old_stripe_index + 1; } -static int __btrfs_map_block(struct btrfs_fs_info *fs_info, - enum btrfs_map_op op, - u64 logical, u64 *length, - struct btrfs_bio **bbio_ret, - int mirror_num, int need_raid_map) +static void map_blocks_raid56_write(struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom, + u64 logical, u64 *length) { - struct extent_map *em; - struct map_lookup *map; - u64 stripe_offset; - u64 stripe_nr; - u64 stripe_len; - u32 stripe_index; - int data_stripes; - int i; + int data_stripes = nr_data_stripes(map); + + /* + * Needs full stripe mapping. + * + * Push stripe_nr back to the start of the full stripe For those cases + * needing a full stripe, @stripe_nr is the full stripe number. + * + * Originally we go raid56_full_stripe_start / full_stripe_len, but + * that can be expensive. Here we just divide @stripe_nr with + * @data_stripes. + */ + io_geom->stripe_nr /= data_stripes; + + /* RAID[56] write or recovery. Return all stripes */ + io_geom->num_stripes = map->num_stripes; + io_geom->max_errors = btrfs_chunk_max_errors(map); + + /* Return the length to the full stripe end. */ + *length = min(logical + *length, + io_geom->raid56_full_stripe_start + map->start + + btrfs_stripe_nr_to_offset(data_stripes)) - + logical; + io_geom->stripe_index = 0; + io_geom->stripe_offset = 0; +} + +static void map_blocks_raid56_read(struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom) +{ + int data_stripes = nr_data_stripes(map); + + ASSERT(io_geom->mirror_num <= 1, "mirror_num=%d", io_geom->mirror_num); + /* Just grab the data stripe directly. */ + io_geom->stripe_index = io_geom->stripe_nr % data_stripes; + io_geom->stripe_nr /= data_stripes; + + /* We distribute the parity blocks across stripes. */ + io_geom->stripe_index = + (io_geom->stripe_nr + io_geom->stripe_index) % map->num_stripes; + + if (io_geom->op == BTRFS_MAP_READ && io_geom->mirror_num < 1) + io_geom->mirror_num = 1; +} + +static void map_blocks_single(const struct btrfs_chunk_map *map, + struct btrfs_io_geometry *io_geom) +{ + io_geom->stripe_index = io_geom->stripe_nr % map->num_stripes; + io_geom->stripe_nr /= map->num_stripes; + io_geom->mirror_num = io_geom->stripe_index + 1; +} + +/* + * Map one logical range to one or more physical ranges. + * + * @length: (Mandatory) mapped length of this run. + * One logical range can be split into different segments + * due to factors like zones and RAID0/5/6/10 stripe + * boundaries. + * + * @bioc_ret: (Mandatory) returned btrfs_io_context structure. + * which has one or more physical ranges (btrfs_io_stripe) + * recorded inside. + * Caller should call btrfs_put_bioc() to free it after use. + * + * @smap: (Optional) single physical range optimization. + * If the map request can be fulfilled by one single + * physical range, and this is parameter is not NULL, + * then @bioc_ret would be NULL, and @smap would be + * updated. + * + * @mirror_num_ret: (Mandatory) returned mirror number if the original + * value is 0. + * + * Mirror number 0 means to choose any live mirrors. + * + * For non-RAID56 profiles, non-zero mirror_num means + * the Nth mirror. (e.g. mirror_num 1 means the first + * copy). + * + * For RAID56 profile, mirror 1 means rebuild from P and + * the remaining data stripes. + * + * For RAID6 profile, mirror > 2 means mark another + * data/P stripe error and rebuild from the remaining + * stripes.. + */ +int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, + u64 logical, u64 *length, + struct btrfs_io_context **bioc_ret, + struct btrfs_io_stripe *smap, int *mirror_num_ret) +{ + struct btrfs_chunk_map *map; + struct btrfs_io_geometry io_geom = { 0 }; + u64 map_offset; int ret = 0; - int num_stripes; - int max_errors = 0; - int tgtdev_indexes = 0; - struct btrfs_bio *bbio = NULL; + int num_copies; + struct btrfs_io_context *bioc = NULL; struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace; - int dev_replace_is_ongoing = 0; - int num_alloc_stripes; - int patch_the_first_stripe_for_dev_replace = 0; - u64 physical_to_patch_in_first_stripe = 0; - u64 raid56_full_stripe_start = (u64)-1; - struct btrfs_io_geometry geom; + bool dev_replace_is_ongoing = false; + u16 num_alloc_stripes; + u64 max_len; - ASSERT(bbio_ret); - ASSERT(op != BTRFS_MAP_DISCARD); + ASSERT(bioc_ret); - ret = btrfs_get_io_geometry(fs_info, op, logical, *length, &geom); - if (ret < 0) - return ret; + io_geom.mirror_num = (mirror_num_ret ? *mirror_num_ret : 0); + io_geom.num_stripes = 1; + io_geom.stripe_index = 0; + io_geom.op = op; + + map = btrfs_get_chunk_map(fs_info, logical, *length); + if (IS_ERR(map)) + return PTR_ERR(map); + + num_copies = btrfs_chunk_map_num_copies(map); + if (io_geom.mirror_num > num_copies) + return -EINVAL; - em = btrfs_get_chunk_map(fs_info, logical, *length); - ASSERT(!IS_ERR(em)); - map = em->map_lookup; + map_offset = logical - map->start; + io_geom.raid56_full_stripe_start = (u64)-1; + max_len = btrfs_max_io_len(map, map_offset, &io_geom); + *length = min_t(u64, map->chunk_len - map_offset, max_len); + io_geom.use_rst = btrfs_need_stripe_tree_update(fs_info, map->type); - *length = geom.len; - stripe_len = geom.stripe_len; - stripe_nr = geom.stripe_nr; - stripe_offset = geom.stripe_offset; - raid56_full_stripe_start = geom.raid56_stripe_offset; - data_stripes = nr_data_stripes(map); + if (dev_replace->replace_task != current) + down_read(&dev_replace->rwsem); - down_read(&dev_replace->rwsem); dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(dev_replace); /* * Hold the semaphore for read during the whole operation, write is * requested at commit time but must wait. */ - if (!dev_replace_is_ongoing) + if (!dev_replace_is_ongoing && dev_replace->replace_task != current) up_read(&dev_replace->rwsem); - if (dev_replace_is_ongoing && mirror_num == map->num_stripes + 1 && - !need_full_stripe(op) && dev_replace->tgtdev != NULL) { - ret = get_extra_mirror_from_replace(fs_info, logical, *length, - dev_replace->srcdev->devid, - &mirror_num, - &physical_to_patch_in_first_stripe); - if (ret) - goto out; + switch (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) { + case BTRFS_BLOCK_GROUP_RAID0: + map_blocks_raid0(map, &io_geom); + break; + case BTRFS_BLOCK_GROUP_RAID1: + case BTRFS_BLOCK_GROUP_RAID1C3: + case BTRFS_BLOCK_GROUP_RAID1C4: + map_blocks_raid1(fs_info, map, &io_geom, dev_replace_is_ongoing); + break; + case BTRFS_BLOCK_GROUP_DUP: + map_blocks_dup(map, &io_geom); + break; + case BTRFS_BLOCK_GROUP_RAID10: + map_blocks_raid10(fs_info, map, &io_geom, dev_replace_is_ongoing); + break; + case BTRFS_BLOCK_GROUP_RAID5: + case BTRFS_BLOCK_GROUP_RAID6: + if (op != BTRFS_MAP_READ || io_geom.mirror_num > 1) + map_blocks_raid56_write(map, &io_geom, logical, length); else - patch_the_first_stripe_for_dev_replace = 1; - } else if (mirror_num > map->num_stripes) { - mirror_num = 0; - } - - num_stripes = 1; - stripe_index = 0; - if (map->type & BTRFS_BLOCK_GROUP_RAID0) { - stripe_nr = div_u64_rem(stripe_nr, map->num_stripes, - &stripe_index); - if (!need_full_stripe(op)) - mirror_num = 1; - } else if (map->type & BTRFS_BLOCK_GROUP_RAID1_MASK) { - if (need_full_stripe(op)) - num_stripes = map->num_stripes; - else if (mirror_num) - stripe_index = mirror_num - 1; - else { - stripe_index = find_live_mirror(fs_info, map, 0, - dev_replace_is_ongoing); - mirror_num = stripe_index + 1; - } - - } else if (map->type & BTRFS_BLOCK_GROUP_DUP) { - if (need_full_stripe(op)) { - num_stripes = map->num_stripes; - } else if (mirror_num) { - stripe_index = mirror_num - 1; - } else { - mirror_num = 1; - } - - } else if (map->type & BTRFS_BLOCK_GROUP_RAID10) { - u32 factor = map->num_stripes / map->sub_stripes; - - stripe_nr = div_u64_rem(stripe_nr, factor, &stripe_index); - stripe_index *= map->sub_stripes; - - if (need_full_stripe(op)) - num_stripes = map->sub_stripes; - else if (mirror_num) - stripe_index += mirror_num - 1; - else { - int old_stripe_index = stripe_index; - stripe_index = find_live_mirror(fs_info, map, - stripe_index, - dev_replace_is_ongoing); - mirror_num = stripe_index - old_stripe_index + 1; - } - - } else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) { - if (need_raid_map && (need_full_stripe(op) || mirror_num > 1)) { - /* push stripe_nr back to the start of the full stripe */ - stripe_nr = div64_u64(raid56_full_stripe_start, - stripe_len * data_stripes); - - /* RAID[56] write or recovery. Return all stripes */ - num_stripes = map->num_stripes; - max_errors = nr_parity_stripes(map); - - *length = map->stripe_len; - stripe_index = 0; - stripe_offset = 0; - } else { - /* - * Mirror #0 or #1 means the original data block. - * Mirror #2 is RAID5 parity block. - * Mirror #3 is RAID6 Q block. - */ - stripe_nr = div_u64_rem(stripe_nr, - data_stripes, &stripe_index); - if (mirror_num > 1) - stripe_index = data_stripes + mirror_num - 2; - - /* We distribute the parity blocks across stripes */ - div_u64_rem(stripe_nr + stripe_index, map->num_stripes, - &stripe_index); - if (!need_full_stripe(op) && mirror_num <= 1) - mirror_num = 1; - } - } else { + map_blocks_raid56_read(map, &io_geom); + break; + default: /* - * after this, stripe_nr is the number of stripes on this + * After this, stripe_nr is the number of stripes on this * device we have to walk to find the data, and stripe_index is * the number of our device in the stripe array */ - stripe_nr = div_u64_rem(stripe_nr, map->num_stripes, - &stripe_index); - mirror_num = stripe_index + 1; + map_blocks_single(map, &io_geom); + break; } - if (stripe_index >= map->num_stripes) { + if (io_geom.stripe_index >= map->num_stripes) { btrfs_crit(fs_info, "stripe index math went horribly wrong, got stripe_index=%u, num_stripes=%u", - stripe_index, map->num_stripes); + io_geom.stripe_index, map->num_stripes); ret = -EINVAL; goto out; } - num_alloc_stripes = num_stripes; - if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL) { - if (op == BTRFS_MAP_WRITE) - num_alloc_stripes <<= 1; - if (op == BTRFS_MAP_GET_READ_MIRRORS) - num_alloc_stripes++; - tgtdev_indexes = num_stripes; + num_alloc_stripes = io_geom.num_stripes; + if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL && + op != BTRFS_MAP_READ) + /* + * For replace case, we need to add extra stripes for extra + * duplicated stripes. + * + * For both WRITE and GET_READ_MIRRORS, we may have at most + * 2 more stripes (DUP types, otherwise 1). + */ + num_alloc_stripes += 2; + + /* + * If this I/O maps to a single device, try to return the device and + * physical block information on the stack instead of allocating an + * I/O context structure. + */ + if (is_single_device_io(fs_info, smap, map, num_alloc_stripes, &io_geom)) { + ret = set_io_stripe(fs_info, logical, length, smap, map, &io_geom); + if (mirror_num_ret) + *mirror_num_ret = io_geom.mirror_num; + *bioc_ret = NULL; + goto out; } - bbio = alloc_btrfs_bio(num_alloc_stripes, tgtdev_indexes); - if (!bbio) { + bioc = alloc_btrfs_io_context(fs_info, logical, num_alloc_stripes); + if (!bioc) { ret = -ENOMEM; goto out; } - if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL) - bbio->tgtdev_map = (int *)(bbio->stripes + num_alloc_stripes); - - /* build raid_map */ - if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK && need_raid_map && - (need_full_stripe(op) || mirror_num > 1)) { - u64 tmp; - unsigned rot; - - bbio->raid_map = (u64 *)((void *)bbio->stripes + - sizeof(struct btrfs_bio_stripe) * - num_alloc_stripes + - sizeof(int) * tgtdev_indexes); - - /* Work out the disk rotation on this stripe-set */ - div_u64_rem(stripe_nr, num_stripes, &rot); + bioc->map_type = map->type; + bioc->use_rst = io_geom.use_rst; - /* Fill in the logical address of each stripe */ - tmp = stripe_nr * data_stripes; - for (i = 0; i < data_stripes; i++) - bbio->raid_map[(i+rot) % num_stripes] = - em->start + (tmp + i) * map->stripe_len; - - bbio->raid_map[(i+rot) % map->num_stripes] = RAID5_P_STRIPE; - if (map->type & BTRFS_BLOCK_GROUP_RAID6) - bbio->raid_map[(i+rot+1) % num_stripes] = - RAID6_Q_STRIPE; + /* + * For RAID56 full map, we need to make sure the stripes[] follows the + * rule that data stripes are all ordered, then followed with P and Q + * (if we have). + * + * It's still mostly the same as other profiles, just with extra rotation. + */ + if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK && + (op != BTRFS_MAP_READ || io_geom.mirror_num > 1)) { + /* + * For RAID56 @stripe_nr is already the number of full stripes + * before us, which is also the rotation value (needs to modulo + * with num_stripes). + * + * In this case, we just add @stripe_nr with @i, then do the + * modulo, to reduce one modulo call. + */ + bioc->full_stripe_logical = map->start + + btrfs_stripe_nr_to_offset(io_geom.stripe_nr * + nr_data_stripes(map)); + for (int i = 0; i < io_geom.num_stripes; i++) { + struct btrfs_io_stripe *dst = &bioc->stripes[i]; + u32 stripe_index; + + stripe_index = (i + io_geom.stripe_nr) % io_geom.num_stripes; + dst->dev = map->stripes[stripe_index].dev; + dst->physical = + map->stripes[stripe_index].physical + + io_geom.stripe_offset + + btrfs_stripe_nr_to_offset(io_geom.stripe_nr); + } + } else { + /* + * For all other non-RAID56 profiles, just copy the target + * stripe into the bioc. + */ + for (int i = 0; i < io_geom.num_stripes; i++) { + ret = set_io_stripe(fs_info, logical, length, + &bioc->stripes[i], map, &io_geom); + if (ret < 0) + break; + io_geom.stripe_index++; + } } - - for (i = 0; i < num_stripes; i++) { - bbio->stripes[i].physical = - map->stripes[stripe_index].physical + - stripe_offset + - stripe_nr * map->stripe_len; - bbio->stripes[i].dev = - map->stripes[stripe_index].dev; - stripe_index++; + if (ret) { + *bioc_ret = NULL; + btrfs_put_bioc(bioc); + goto out; } - if (need_full_stripe(op)) - max_errors = btrfs_chunk_max_errors(map); - - if (bbio->raid_map) - sort_parity_stripes(bbio, num_stripes); + if (op != BTRFS_MAP_READ) + io_geom.max_errors = btrfs_chunk_max_errors(map); if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL && - need_full_stripe(op)) { - handle_ops_on_dev_replace(op, &bbio, dev_replace, &num_stripes, - &max_errors); + op != BTRFS_MAP_READ) { + handle_ops_on_dev_replace(bioc, dev_replace, logical, &io_geom); } - *bbio_ret = bbio; - bbio->map_type = map->type; - bbio->num_stripes = num_stripes; - bbio->max_errors = max_errors; - bbio->mirror_num = mirror_num; + *bioc_ret = bioc; + bioc->num_stripes = io_geom.num_stripes; + bioc->max_errors = io_geom.max_errors; + bioc->mirror_num = io_geom.mirror_num; - /* - * this is the case that REQ_READ && dev_replace_is_ongoing && - * mirror_num == num_stripes + 1 && dev_replace target drive is - * available as a mirror - */ - if (patch_the_first_stripe_for_dev_replace && num_stripes > 0) { - WARN_ON(num_stripes > 1); - bbio->stripes[0].dev = dev_replace->tgtdev; - bbio->stripes[0].physical = physical_to_patch_in_first_stripe; - bbio->mirror_num = map->num_stripes + 1; - } out: - if (dev_replace_is_ongoing) { + if (dev_replace_is_ongoing && dev_replace->replace_task != current) { lockdep_assert_held(&dev_replace->rwsem); /* Unlock and let waiting writers proceed */ up_read(&dev_replace->rwsem); } - free_extent_map(em); + btrfs_free_chunk_map(map); return ret; } -int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, - u64 logical, u64 *length, - struct btrfs_bio **bbio_ret, int mirror_num) -{ - if (op == BTRFS_MAP_DISCARD) - return __btrfs_map_block_for_discard(fs_info, logical, - length, bbio_ret); - - return __btrfs_map_block(fs_info, op, logical, length, bbio_ret, - mirror_num, 0); -} - -/* For Scrub/replace */ -int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op, - u64 logical, u64 *length, - struct btrfs_bio **bbio_ret) -{ - return __btrfs_map_block(fs_info, op, logical, length, bbio_ret, 0, 1); -} - -static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio) -{ - bio->bi_private = bbio->private; - bio->bi_end_io = bbio->end_io; - bio_endio(bio); - - btrfs_put_bbio(bbio); -} - -static void btrfs_end_bio(struct bio *bio) -{ - struct btrfs_bio *bbio = bio->bi_private; - int is_orig_bio = 0; - - if (bio->bi_status) { - atomic_inc(&bbio->error); - if (bio->bi_status == BLK_STS_IOERR || - bio->bi_status == BLK_STS_TARGET) { - unsigned int stripe_index = - btrfs_io_bio(bio)->stripe_index; - struct btrfs_device *dev; - - BUG_ON(stripe_index >= bbio->num_stripes); - dev = bbio->stripes[stripe_index].dev; - if (dev->bdev) { - if (bio_op(bio) == REQ_OP_WRITE) - btrfs_dev_stat_inc_and_print(dev, - BTRFS_DEV_STAT_WRITE_ERRS); - else if (!(bio->bi_opf & REQ_RAHEAD)) - btrfs_dev_stat_inc_and_print(dev, - BTRFS_DEV_STAT_READ_ERRS); - if (bio->bi_opf & REQ_PREFLUSH) - btrfs_dev_stat_inc_and_print(dev, - BTRFS_DEV_STAT_FLUSH_ERRS); - } - } - } - - if (bio == bbio->orig_bio) - is_orig_bio = 1; - - btrfs_bio_counter_dec(bbio->fs_info); - - if (atomic_dec_and_test(&bbio->stripes_pending)) { - if (!is_orig_bio) { - bio_put(bio); - bio = bbio->orig_bio; - } - - btrfs_io_bio(bio)->mirror_num = bbio->mirror_num; - /* only send an error to the higher layers if it is - * beyond the tolerance of the btrfs bio - */ - if (atomic_read(&bbio->error) > bbio->max_errors) { - bio->bi_status = BLK_STS_IOERR; - } else { - /* - * this bio is actually up to date, we didn't - * go over the max number of errors - */ - bio->bi_status = BLK_STS_OK; - } - - btrfs_end_bbio(bbio, bio); - } else if (!is_orig_bio) { - bio_put(bio); - } -} - -static void submit_stripe_bio(struct btrfs_bio *bbio, struct bio *bio, - u64 physical, int dev_nr) -{ - struct btrfs_device *dev = bbio->stripes[dev_nr].dev; - struct btrfs_fs_info *fs_info = bbio->fs_info; - - bio->bi_private = bbio; - btrfs_io_bio(bio)->stripe_index = dev_nr; - bio->bi_end_io = btrfs_end_bio; - bio->bi_iter.bi_sector = physical >> 9; - btrfs_debug_in_rcu(fs_info, - "btrfs_map_bio: rw %d 0x%x, sector=%llu, dev=%lu (%s id %llu), size=%u", - bio_op(bio), bio->bi_opf, (u64)bio->bi_iter.bi_sector, - (unsigned long)dev->bdev->bd_dev, rcu_str_deref(dev->name), - dev->devid, bio->bi_iter.bi_size); - bio_set_dev(bio, dev->bdev); - - btrfs_bio_counter_inc_noblocked(fs_info); - - btrfsic_submit_bio(bio); -} - -static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical) +static bool dev_args_match_fs_devices(const struct btrfs_dev_lookup_args *args, + const struct btrfs_fs_devices *fs_devices) { - atomic_inc(&bbio->error); - if (atomic_dec_and_test(&bbio->stripes_pending)) { - /* Should be the original bio. */ - WARN_ON(bio != bbio->orig_bio); - - btrfs_io_bio(bio)->mirror_num = bbio->mirror_num; - bio->bi_iter.bi_sector = logical >> 9; - if (atomic_read(&bbio->error) > bbio->max_errors) - bio->bi_status = BLK_STS_IOERR; - else - bio->bi_status = BLK_STS_OK; - btrfs_end_bbio(bbio, bio); - } + if (args->fsid == NULL) + return true; + if (memcmp(fs_devices->metadata_uuid, args->fsid, BTRFS_FSID_SIZE) == 0) + return true; + return false; } -blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio, - int mirror_num) +static bool dev_args_match_device(const struct btrfs_dev_lookup_args *args, + const struct btrfs_device *device) { - struct btrfs_device *dev; - struct bio *first_bio = bio; - u64 logical = (u64)bio->bi_iter.bi_sector << 9; - u64 length = 0; - u64 map_length; - int ret; - int dev_nr; - int total_devs; - struct btrfs_bio *bbio = NULL; - - length = bio->bi_iter.bi_size; - map_length = length; - - btrfs_bio_counter_inc_blocked(fs_info); - ret = __btrfs_map_block(fs_info, btrfs_op(bio), logical, - &map_length, &bbio, mirror_num, 1); - if (ret) { - btrfs_bio_counter_dec(fs_info); - return errno_to_blk_status(ret); - } - - total_devs = bbio->num_stripes; - bbio->orig_bio = first_bio; - bbio->private = first_bio->bi_private; - bbio->end_io = first_bio->bi_end_io; - bbio->fs_info = fs_info; - atomic_set(&bbio->stripes_pending, bbio->num_stripes); - - if ((bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) && - ((bio_op(bio) == REQ_OP_WRITE) || (mirror_num > 1))) { - /* In this case, map_length has been set to the length of - a single stripe; not the whole write */ - if (bio_op(bio) == REQ_OP_WRITE) { - ret = raid56_parity_write(fs_info, bio, bbio, - map_length); - } else { - ret = raid56_parity_recover(fs_info, bio, bbio, - map_length, mirror_num, 1); - } - - btrfs_bio_counter_dec(fs_info); - return errno_to_blk_status(ret); - } - - if (map_length < length) { - btrfs_crit(fs_info, - "mapping failed logical %llu bio len %llu len %llu", - logical, length, map_length); - BUG(); + if (args->devt) + return device->devt == args->devt; + if (args->missing) { + if (test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state) && + !device->bdev) + return true; + return false; } - for (dev_nr = 0; dev_nr < total_devs; dev_nr++) { - dev = bbio->stripes[dev_nr].dev; - if (!dev || !dev->bdev || test_bit(BTRFS_DEV_STATE_MISSING, - &dev->dev_state) || - (bio_op(first_bio) == REQ_OP_WRITE && - !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))) { - bbio_error(bbio, first_bio, logical); - continue; - } - - if (dev_nr < total_devs - 1) - bio = btrfs_bio_clone(first_bio); - else - bio = first_bio; - - submit_stripe_bio(bbio, bio, bbio->stripes[dev_nr].physical, - dev_nr); - } - btrfs_bio_counter_dec(fs_info); - return BLK_STS_OK; + if (device->devid != args->devid) + return false; + if (args->uuid && memcmp(device->uuid, args->uuid, BTRFS_UUID_SIZE) != 0) + return false; + return true; } /* @@ -6433,31 +6777,29 @@ blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio, * * If devid and uuid are both specified, the match must be exact, otherwise * only devid is used. - * - * If @seed is true, traverse through the seed devices. */ -struct btrfs_device *btrfs_find_device(struct btrfs_fs_devices *fs_devices, - u64 devid, u8 *uuid, u8 *fsid, - bool seed) +struct btrfs_device *btrfs_find_device(const struct btrfs_fs_devices *fs_devices, + const struct btrfs_dev_lookup_args *args) { struct btrfs_device *device; + struct btrfs_fs_devices *seed_devs; - while (fs_devices) { - if (!fsid || - !memcmp(fs_devices->metadata_uuid, fsid, BTRFS_FSID_SIZE)) { - list_for_each_entry(device, &fs_devices->devices, - dev_list) { - if (device->devid == devid && - (!uuid || memcmp(device->uuid, uuid, - BTRFS_UUID_SIZE) == 0)) - return device; - } + if (dev_args_match_fs_devices(args, fs_devices)) { + list_for_each_entry(device, &fs_devices->devices, dev_list) { + if (dev_args_match_device(args, device)) + return device; } - if (seed) - fs_devices = fs_devices->seed; - else - return NULL; } + + list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) { + if (!dev_args_match_fs_devices(args, seed_devs)) + continue; + list_for_each_entry(device, &seed_devs->devices, dev_list) { + if (dev_args_match_device(args, device)) + return device; + } + } + return NULL; } @@ -6465,8 +6807,18 @@ static struct btrfs_device *add_missing_dev(struct btrfs_fs_devices *fs_devices, u64 devid, u8 *dev_uuid) { struct btrfs_device *device; + unsigned int nofs_flag; + + /* + * We call this under the chunk_mutex, so we want to use NOFS for this + * allocation, however we don't want to change btrfs_alloc_device() to + * always do NOFS because we use it in a lot of other GFP_KERNEL safe + * places. + */ - device = btrfs_alloc_device(NULL, &devid, dev_uuid); + nofs_flag = memalloc_nofs_save(); + device = btrfs_alloc_device(NULL, &devid, dev_uuid, NULL); + memalloc_nofs_restore(nofs_flag); if (IS_ERR(device)) return device; @@ -6480,22 +6832,24 @@ static struct btrfs_device *add_missing_dev(struct btrfs_fs_devices *fs_devices, return device; } -/** - * btrfs_alloc_device - allocate struct btrfs_device +/* + * Allocate new device struct, set up devid and UUID. + * * @fs_info: used only for generating a new devid, can be NULL if * devid is provided (i.e. @devid != NULL). * @devid: a pointer to devid for this device. If NULL a new devid * is generated. * @uuid: a pointer to UUID for this device. If NULL a new UUID * is generated. + * @path: a pointer to device path if available, NULL otherwise. * * Return: a pointer to a new &struct btrfs_device on success; ERR_PTR() * on error. Returned struct is not linked onto any lists and must be * destroyed with btrfs_free_device. */ struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info, - const u64 *devid, - const u8 *uuid) + const u64 *devid, const u8 *uuid, + const char *path) { struct btrfs_device *dev; u64 tmp; @@ -6503,9 +6857,17 @@ struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info, if (WARN_ON(!devid && !fs_info)) return ERR_PTR(-EINVAL); - dev = __alloc_device(); - if (IS_ERR(dev)) - return dev; + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + return ERR_PTR(-ENOMEM); + + INIT_LIST_HEAD(&dev->dev_list); + INIT_LIST_HEAD(&dev->dev_alloc_list); + INIT_LIST_HEAD(&dev->post_commit_list); + + atomic_set(&dev->dev_stats_ccnt, 0); + btrfs_device_data_ordered_init(dev); + btrfs_extent_io_tree_init(fs_info, &dev->alloc_state, IO_TREE_DEVICE_ALLOC_STATE); if (devid) tmp = *devid; @@ -6525,6 +6887,17 @@ struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info, else generate_random_uuid(dev->uuid); + if (path) { + const char *name; + + name = kstrdup(path, GFP_KERNEL); + if (!name) { + btrfs_free_device(dev); + return ERR_PTR(-ENOMEM); + } + rcu_assign_pointer(dev->name, name); + } + return dev; } @@ -6539,130 +6912,165 @@ static void btrfs_report_missing_device(struct btrfs_fs_info *fs_info, devid, uuid); } -static u64 calc_stripe_length(u64 type, u64 chunk_len, int num_stripes) +u64 btrfs_calc_stripe_length(const struct btrfs_chunk_map *map) { - int index = btrfs_bg_flags_to_raid_index(type); - int ncopies = btrfs_raid_array[index].ncopies; - const int nparity = btrfs_raid_array[index].nparity; - int data_stripes; + const int data_stripes = calc_data_stripes(map->type, map->num_stripes); - if (nparity) - data_stripes = num_stripes - nparity; - else - data_stripes = num_stripes / ncopies; + return div_u64(map->chunk_len, data_stripes); +} + +#if BITS_PER_LONG == 32 +/* + * Due to page cache limit, metadata beyond BTRFS_32BIT_MAX_FILE_SIZE + * can't be accessed on 32bit systems. + * + * This function do mount time check to reject the fs if it already has + * metadata chunk beyond that limit. + */ +static int check_32bit_meta_chunk(struct btrfs_fs_info *fs_info, + u64 logical, u64 length, u64 type) +{ + if (!(type & BTRFS_BLOCK_GROUP_METADATA)) + return 0; + + if (logical + length < MAX_LFS_FILESIZE) + return 0; + + btrfs_err_32bit_limit(fs_info); + return -EOVERFLOW; +} + +/* + * This is to give early warning for any metadata chunk reaching + * BTRFS_32BIT_EARLY_WARN_THRESHOLD. + * Although we can still access the metadata, it's not going to be possible + * once the limit is reached. + */ +static void warn_32bit_meta_chunk(struct btrfs_fs_info *fs_info, + u64 logical, u64 length, u64 type) +{ + if (!(type & BTRFS_BLOCK_GROUP_METADATA)) + return; + + if (logical + length < BTRFS_32BIT_EARLY_WARN_THRESHOLD) + return; + + btrfs_warn_32bit_limit(fs_info); +} +#endif + +static struct btrfs_device *handle_missing_device(struct btrfs_fs_info *fs_info, + u64 devid, u8 *uuid) +{ + struct btrfs_device *dev; - return div_u64(chunk_len, data_stripes); + if (!btrfs_test_opt(fs_info, DEGRADED)) { + btrfs_report_missing_device(fs_info, devid, uuid, true); + return ERR_PTR(-ENOENT); + } + + dev = add_missing_dev(fs_info->fs_devices, devid, uuid); + if (IS_ERR(dev)) { + btrfs_err(fs_info, "failed to init missing device %llu: %ld", + devid, PTR_ERR(dev)); + return dev; + } + btrfs_report_missing_device(fs_info, devid, uuid, false); + + return dev; } static int read_one_chunk(struct btrfs_key *key, struct extent_buffer *leaf, struct btrfs_chunk *chunk) { + BTRFS_DEV_LOOKUP_ARGS(args); struct btrfs_fs_info *fs_info = leaf->fs_info; - struct extent_map_tree *map_tree = &fs_info->mapping_tree; - struct map_lookup *map; - struct extent_map *em; + struct btrfs_chunk_map *map; u64 logical; u64 length; u64 devid; + u64 type; u8 uuid[BTRFS_UUID_SIZE]; + int index; int num_stripes; int ret; int i; logical = key->offset; length = btrfs_chunk_length(leaf, chunk); + type = btrfs_chunk_type(leaf, chunk); + index = btrfs_bg_flags_to_raid_index(type); num_stripes = btrfs_chunk_num_stripes(leaf, chunk); - /* - * Only need to verify chunk item if we're reading from sys chunk array, - * as chunk item in tree block is already verified by tree-checker. - */ - if (leaf->start == BTRFS_SUPER_INFO_OFFSET) { - ret = btrfs_check_chunk_valid(leaf, chunk, logical); - if (ret) - return ret; - } +#if BITS_PER_LONG == 32 + ret = check_32bit_meta_chunk(fs_info, logical, length, type); + if (ret < 0) + return ret; + warn_32bit_meta_chunk(fs_info, logical, length, type); +#endif - read_lock(&map_tree->lock); - em = lookup_extent_mapping(map_tree, logical, 1); - read_unlock(&map_tree->lock); + map = btrfs_find_chunk_map(fs_info, logical, 1); /* already mapped? */ - if (em && em->start <= logical && em->start + em->len > logical) { - free_extent_map(em); + if (map && map->start <= logical && map->start + map->chunk_len > logical) { + btrfs_free_chunk_map(map); return 0; - } else if (em) { - free_extent_map(em); + } else if (map) { + btrfs_free_chunk_map(map); } - em = alloc_extent_map(); - if (!em) - return -ENOMEM; - map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS); - if (!map) { - free_extent_map(em); + map = btrfs_alloc_chunk_map(num_stripes, GFP_NOFS); + if (!map) return -ENOMEM; - } - - set_bit(EXTENT_FLAG_FS_MAPPING, &em->flags); - em->map_lookup = map; - em->start = logical; - em->len = length; - em->orig_start = 0; - em->block_start = 0; - em->block_len = em->len; + map->start = logical; + map->chunk_len = length; map->num_stripes = num_stripes; map->io_width = btrfs_chunk_io_width(leaf, chunk); map->io_align = btrfs_chunk_io_align(leaf, chunk); - map->stripe_len = btrfs_chunk_stripe_len(leaf, chunk); - map->type = btrfs_chunk_type(leaf, chunk); - map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk); + map->type = type; + /* + * We can't use the sub_stripes value, as for profiles other than + * RAID10, they may have 0 as sub_stripes for filesystems created by + * older mkfs (<v5.4). + * In that case, it can cause divide-by-zero errors later. + * Since currently sub_stripes is fixed for each profile, let's + * use the trusted value instead. + */ + map->sub_stripes = btrfs_raid_array[index].sub_stripes; map->verified_stripes = 0; - em->orig_block_len = calc_stripe_length(map->type, em->len, - map->num_stripes); + map->stripe_size = btrfs_calc_stripe_length(map); for (i = 0; i < num_stripes; i++) { map->stripes[i].physical = btrfs_stripe_offset_nr(leaf, chunk, i); devid = btrfs_stripe_devid_nr(leaf, chunk, i); + args.devid = devid; read_extent_buffer(leaf, uuid, (unsigned long) btrfs_stripe_dev_uuid_nr(chunk, i), BTRFS_UUID_SIZE); - map->stripes[i].dev = btrfs_find_device(fs_info->fs_devices, - devid, uuid, NULL, true); - if (!map->stripes[i].dev && - !btrfs_test_opt(fs_info, DEGRADED)) { - free_extent_map(em); - btrfs_report_missing_device(fs_info, devid, uuid, true); - return -ENOENT; - } + args.uuid = uuid; + map->stripes[i].dev = btrfs_find_device(fs_info->fs_devices, &args); if (!map->stripes[i].dev) { - map->stripes[i].dev = - add_missing_dev(fs_info->fs_devices, devid, - uuid); + map->stripes[i].dev = handle_missing_device(fs_info, + devid, uuid); if (IS_ERR(map->stripes[i].dev)) { - free_extent_map(em); - btrfs_err(fs_info, - "failed to init missing dev %llu: %ld", - devid, PTR_ERR(map->stripes[i].dev)); - return PTR_ERR(map->stripes[i].dev); + ret = PTR_ERR(map->stripes[i].dev); + btrfs_free_chunk_map(map); + return ret; } - btrfs_report_missing_device(fs_info, devid, uuid, false); } + set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &(map->stripes[i].dev->dev_state)); - } - write_lock(&map_tree->lock); - ret = add_extent_mapping(map_tree, em, 0); - write_unlock(&map_tree->lock); + ret = btrfs_add_chunk_map(fs_info, map); if (ret < 0) { btrfs_err(fs_info, "failed to add chunk map, start=%llu len=%llu: %d", - em->start, em->len, ret); + map->start, map->chunk_len, ret); + btrfs_free_chunk_map(map); } - free_extent_map(em); return ret; } @@ -6699,20 +7107,22 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info, lockdep_assert_held(&uuid_mutex); ASSERT(fsid); - fs_devices = fs_info->fs_devices->seed; - while (fs_devices) { + /* This will match only for multi-device seed fs */ + list_for_each_entry(fs_devices, &fs_info->fs_devices->seed_list, seed_list) if (!memcmp(fs_devices->fsid, fsid, BTRFS_FSID_SIZE)) return fs_devices; - fs_devices = fs_devices->seed; - } fs_devices = find_fsid(fsid, NULL); if (!fs_devices) { - if (!btrfs_test_opt(fs_info, DEGRADED)) + if (!btrfs_test_opt(fs_info, DEGRADED)) { + btrfs_err(fs_info, + "failed to find fsid %pU when attempting to open seed devices", + fsid); return ERR_PTR(-ENOENT); + } - fs_devices = alloc_fs_devices(fsid, NULL); + fs_devices = alloc_fs_devices(fsid); if (IS_ERR(fs_devices)) return fs_devices; @@ -6721,33 +7131,35 @@ static struct btrfs_fs_devices *open_seed_devices(struct btrfs_fs_info *fs_info, return fs_devices; } + /* + * Upon first call for a seed fs fsid, just create a private copy of the + * respective fs_devices and anchor it at fs_info->fs_devices->seed_list + */ fs_devices = clone_fs_devices(fs_devices); if (IS_ERR(fs_devices)) return fs_devices; - ret = open_fs_devices(fs_devices, FMODE_READ, fs_info->bdev_holder); + ret = open_fs_devices(fs_devices, BLK_OPEN_READ, fs_info->sb); if (ret) { free_fs_devices(fs_devices); - fs_devices = ERR_PTR(ret); - goto out; + return ERR_PTR(ret); } if (!fs_devices->seeding) { close_fs_devices(fs_devices); free_fs_devices(fs_devices); - fs_devices = ERR_PTR(-EINVAL); - goto out; + return ERR_PTR(-EINVAL); } - fs_devices->seed = fs_info->fs_devices->seed; - fs_info->fs_devices->seed = fs_devices; -out: + list_add(&fs_devices->seed_list, &fs_info->fs_devices->seed_list); + return fs_devices; } static int read_one_dev(struct extent_buffer *leaf, struct btrfs_dev_item *dev_item) { + BTRFS_DEV_LOOKUP_ARGS(args); struct btrfs_fs_info *fs_info = leaf->fs_info; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; struct btrfs_device *device; @@ -6757,10 +7169,13 @@ static int read_one_dev(struct extent_buffer *leaf, u8 dev_uuid[BTRFS_UUID_SIZE]; devid = btrfs_device_id(leaf, dev_item); + args.devid = devid; read_extent_buffer(leaf, dev_uuid, btrfs_device_uuid(dev_item), BTRFS_UUID_SIZE); read_extent_buffer(leaf, fs_uuid, btrfs_device_fsid(dev_item), BTRFS_FSID_SIZE); + args.uuid = dev_uuid; + args.fsid = fs_uuid; if (memcmp(fs_uuid, fs_devices->metadata_uuid, BTRFS_FSID_SIZE)) { fs_devices = open_seed_devices(fs_info, fs_uuid); @@ -6768,8 +7183,7 @@ static int read_one_dev(struct extent_buffer *leaf, return PTR_ERR(fs_devices); } - device = btrfs_find_device(fs_info->fs_devices, devid, dev_uuid, - fs_uuid, true); + device = btrfs_find_device(fs_info->fs_devices, &args); if (!device) { if (!btrfs_test_opt(fs_info, DEGRADED)) { btrfs_report_missing_device(fs_info, devid, @@ -6832,6 +7246,16 @@ static int read_one_dev(struct extent_buffer *leaf, } fill_device_from_item(leaf, dev_item, device); + if (device->bdev) { + u64 max_total_bytes = bdev_nr_bytes(device->bdev); + + if (device->total_bytes > max_total_bytes) { + btrfs_err(fs_info, + "device total_bytes should be at most %llu but found %llu", + max_total_bytes, device->total_bytes); + return -EINVAL; + } + } set_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state); if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state) && !test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { @@ -6845,46 +7269,26 @@ static int read_one_dev(struct extent_buffer *leaf, int btrfs_read_sys_array(struct btrfs_fs_info *fs_info) { - struct btrfs_root *root = fs_info->tree_root; struct btrfs_super_block *super_copy = fs_info->super_copy; struct extent_buffer *sb; - struct btrfs_disk_key *disk_key; - struct btrfs_chunk *chunk; u8 *array_ptr; unsigned long sb_array_offset; int ret = 0; - u32 num_stripes; u32 array_size; - u32 len = 0; u32 cur_offset; - u64 type; struct btrfs_key key; ASSERT(BTRFS_SUPER_INFO_SIZE <= fs_info->nodesize); + /* - * This will create extent buffer of nodesize, superblock size is - * fixed to BTRFS_SUPER_INFO_SIZE. If nodesize > sb size, this will - * overallocate but we can keep it as-is, only the first page is used. + * We allocated a dummy extent, just to use extent buffer accessors. + * There will be unused space after BTRFS_SUPER_INFO_SIZE, but + * that's fine, we will not go beyond system chunk array anyway. */ - sb = btrfs_find_create_tree_block(fs_info, BTRFS_SUPER_INFO_OFFSET); - if (IS_ERR(sb)) - return PTR_ERR(sb); + sb = alloc_dummy_extent_buffer(fs_info, BTRFS_SUPER_INFO_OFFSET); + if (!sb) + return -ENOMEM; set_extent_buffer_uptodate(sb); - btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0); - /* - * The sb extent buffer is artificial and just used to read the system array. - * set_extent_buffer_uptodate() call does not properly mark all it's - * pages up-to-date when the page is larger: extent does not cover the - * whole page and consequently check_page_uptodate does not find all - * the page's extents up-to-date (the hole beyond sb), - * write_extent_buffer then triggers a WARN_ON. - * - * Regular short extents go through mark_extent_buffer_dirty/writeback cycle, - * but sb spans only this function. Add an explicit SetPageUptodate call - * to silence the warning eg. on PowerPC 64. - */ - if (PAGE_SIZE > BTRFS_SUPER_INFO_SIZE) - SetPageUptodate(sb->pages[0]); write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE); array_size = btrfs_super_sys_array_size(super_copy); @@ -6894,10 +7298,15 @@ int btrfs_read_sys_array(struct btrfs_fs_info *fs_info) cur_offset = 0; while (cur_offset < array_size) { - disk_key = (struct btrfs_disk_key *)array_ptr; - len = sizeof(*disk_key); - if (cur_offset + len > array_size) - goto out_short_read; + struct btrfs_chunk *chunk; + struct btrfs_disk_key *disk_key = (struct btrfs_disk_key *)array_ptr; + u32 len = sizeof(*disk_key); + + /* + * The sys_chunk_array has been already verified at super block + * read time. Only do ASSERT()s for basic checks. + */ + ASSERT(cur_offset + len <= array_size); btrfs_disk_key_to_cpu(&key, disk_key); @@ -6905,44 +7314,14 @@ int btrfs_read_sys_array(struct btrfs_fs_info *fs_info) sb_array_offset += len; cur_offset += len; - if (key.type != BTRFS_CHUNK_ITEM_KEY) { - btrfs_err(fs_info, - "unexpected item type %u in sys_array at offset %u", - (u32)key.type, cur_offset); - ret = -EIO; - break; - } + ASSERT(key.type == BTRFS_CHUNK_ITEM_KEY); chunk = (struct btrfs_chunk *)sb_array_offset; - /* - * At least one btrfs_chunk with one stripe must be present, - * exact stripe count check comes afterwards - */ - len = btrfs_chunk_item_size(1); - if (cur_offset + len > array_size) - goto out_short_read; + ASSERT(btrfs_chunk_type(sb, chunk) & BTRFS_BLOCK_GROUP_SYSTEM); - num_stripes = btrfs_chunk_num_stripes(sb, chunk); - if (!num_stripes) { - btrfs_err(fs_info, - "invalid number of stripes %u in sys_array at offset %u", - num_stripes, cur_offset); - ret = -EIO; - break; - } + len = btrfs_chunk_item_size(btrfs_chunk_num_stripes(sb, chunk)); - type = btrfs_chunk_type(sb, chunk); - if ((type & BTRFS_BLOCK_GROUP_SYSTEM) == 0) { - btrfs_err(fs_info, - "invalid chunk type %llu in sys_array at offset %u", - type, cur_offset); - ret = -EIO; - break; - } - - len = btrfs_chunk_item_size(num_stripes); - if (cur_offset + len > array_size) - goto out_short_read; + ASSERT(cur_offset + len <= array_size); ret = read_one_chunk(&key, sb, chunk); if (ret) @@ -6955,13 +7334,6 @@ int btrfs_read_sys_array(struct btrfs_fs_info *fs_info) clear_extent_buffer_uptodate(sb); free_extent_buffer_stale(sb); return ret; - -out_short_read: - btrfs_err(fs_info, "sys_array too short to read %u bytes at offset %u", - len, cur_offset); - clear_extent_buffer_uptodate(sb); - free_extent_buffer_stale(sb); - return -EIO; } /* @@ -6975,26 +7347,21 @@ out_short_read: bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info, struct btrfs_device *failing_dev) { - struct extent_map_tree *map_tree = &fs_info->mapping_tree; - struct extent_map *em; - u64 next_start = 0; + struct btrfs_chunk_map *map; + u64 next_start; bool ret = true; - read_lock(&map_tree->lock); - em = lookup_extent_mapping(map_tree, 0, (u64)-1); - read_unlock(&map_tree->lock); + map = btrfs_find_chunk_map(fs_info, 0, U64_MAX); /* No chunk at all? Return false anyway */ - if (!em) { + if (!map) { ret = false; goto out; } - while (em) { - struct map_lookup *map; + while (map) { int missing = 0; int max_tolerated; int i; - map = em->map_lookup; max_tolerated = btrfs_get_num_tolerated_disk_barrier_failures( map->type); @@ -7012,33 +7379,41 @@ bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info, if (!failing_dev) btrfs_warn(fs_info, "chunk %llu missing %d devices, max tolerance is %d for writable mount", - em->start, missing, max_tolerated); - free_extent_map(em); + map->start, missing, max_tolerated); + btrfs_free_chunk_map(map); ret = false; goto out; } - next_start = extent_map_end(em); - free_extent_map(em); + next_start = map->start + map->chunk_len; + btrfs_free_chunk_map(map); - read_lock(&map_tree->lock); - em = lookup_extent_mapping(map_tree, next_start, - (u64)(-1) - next_start); - read_unlock(&map_tree->lock); + map = btrfs_find_chunk_map(fs_info, next_start, U64_MAX - next_start); } out: return ret; } +static void readahead_tree_node_children(struct extent_buffer *node) +{ + int i; + const int nr_items = btrfs_header_nritems(node); + + for (i = 0; i < nr_items; i++) + btrfs_readahead_node_child(node, i); +} + int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) { struct btrfs_root *root = fs_info->chunk_root; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct extent_buffer *leaf; struct btrfs_key key; struct btrfs_key found_key; int ret; int slot; + int iter_ret = 0; u64 total_dev = 0; + u64 last_ra_node = 0; path = btrfs_alloc_path(); if (!path) @@ -7049,7 +7424,27 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) * otherwise we don't need it. */ mutex_lock(&uuid_mutex); - mutex_lock(&fs_info->chunk_mutex); + + /* + * It is possible for mount and umount to race in such a way that + * we execute this code path, but open_fs_devices failed to clear + * total_rw_bytes. We certainly want it cleared before reading the + * device items, so clear it here. + */ + fs_info->fs_devices->total_rw_bytes = 0; + + /* + * Lockdep complains about possible circular locking dependency between + * a disk's open_mutex (struct gendisk.open_mutex), the rw semaphores + * used for freeze protection of a fs (struct super_block.s_writers), + * which we take when starting a transaction, and extent buffers of the + * chunk tree if we call read_one_dev() while holding a lock on an + * extent buffer of the chunk tree. Since we are mounting the filesystem + * and at this point there can't be any concurrent task modifying the + * chunk tree, to keep it simple, just skip locking on the chunk tree. + */ + ASSERT(!test_bit(BTRFS_FS_OPEN, &fs_info->flags)); + path->skip_locking = true; /* * Read all device items, and then all the chunk items. All @@ -7058,23 +7453,20 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) * item - BTRFS_FIRST_CHUNK_TREE_OBJECTID). */ key.objectid = BTRFS_DEV_ITEMS_OBJECTID; - key.offset = 0; key.type = 0; - ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); - if (ret < 0) - goto error; - while (1) { + key.offset = 0; + btrfs_for_each_slot(root, &key, &found_key, path, iter_ret) { + struct extent_buffer *node = path->nodes[1]; + leaf = path->nodes[0]; slot = path->slots[0]; - if (slot >= btrfs_header_nritems(leaf)) { - ret = btrfs_next_leaf(root, path); - if (ret == 0) - continue; - if (ret < 0) - goto error; - break; + + if (node) { + if (last_ra_node != node->start) { + readahead_tree_node_children(node); + last_ra_node = node->start; + } } - btrfs_item_key_to_cpu(leaf, &found_key, slot); if (found_key.type == BTRFS_DEV_ITEM_KEY) { struct btrfs_dev_item *dev_item; dev_item = btrfs_item_ptr(leaf, slot, @@ -7085,12 +7477,25 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) total_dev++; } else if (found_key.type == BTRFS_CHUNK_ITEM_KEY) { struct btrfs_chunk *chunk; + + /* + * We are only called at mount time, so no need to take + * fs_info->chunk_mutex. Plus, to avoid lockdep warnings, + * we always lock first fs_info->chunk_mutex before + * acquiring any locks on the chunk tree. This is a + * requirement for chunk allocation, see the comment on + * top of btrfs_chunk_alloc() for details. + */ chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk); ret = read_one_chunk(&found_key, leaf, chunk); if (ret) goto error; } - path->slots[0]++; + } + /* Catch error found during iteration */ + if (iter_ret < 0) { + ret = iter_ret; + goto error; } /* @@ -7098,12 +7503,12 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) * do another round of validation checks. */ if (total_dev != fs_info->fs_devices->total_devices) { - btrfs_err(fs_info, - "super_num_devices %llu mismatch with num_devices %llu found here", + btrfs_warn(fs_info, +"super block num_devices %llu mismatch with DEV_ITEM count %llu, will be repaired on next transaction commit", btrfs_super_num_devices(fs_info->super_copy), total_dev); - ret = -EINVAL; - goto error; + fs_info->fs_devices->total_devices = total_dev; + btrfs_set_super_num_devices(fs_info->super_copy, total_dev); } if (btrfs_super_total_bytes(fs_info->super_copy) < fs_info->fs_devices->total_rw_bytes) { @@ -7116,26 +7521,33 @@ int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info) } ret = 0; error: - mutex_unlock(&fs_info->chunk_mutex); mutex_unlock(&uuid_mutex); - - btrfs_free_path(path); return ret; } -void btrfs_init_devices_late(struct btrfs_fs_info *fs_info) +int btrfs_init_devices_late(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct btrfs_fs_devices *fs_devices = fs_info->fs_devices, *seed_devs; struct btrfs_device *device; + int ret = 0; - while (fs_devices) { - mutex_lock(&fs_devices->device_list_mutex); - list_for_each_entry(device, &fs_devices->devices, dev_list) + mutex_lock(&fs_devices->device_list_mutex); + list_for_each_entry(device, &fs_devices->devices, dev_list) + device->fs_info = fs_info; + + list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) { + list_for_each_entry(device, &seed_devs->devices, dev_list) { device->fs_info = fs_info; - mutex_unlock(&fs_devices->device_list_mutex); + ret = btrfs_get_dev_zone_info(device, false); + if (ret) + break; + } - fs_devices = fs_devices->seed; + seed_devs->fs_info = fs_info; } + mutex_unlock(&fs_devices->device_list_mutex); + + return ret; } static u64 btrfs_dev_stats_value(const struct extent_buffer *eb, @@ -7161,17 +7573,56 @@ static void btrfs_set_dev_stats_value(struct extent_buffer *eb, sizeof(val)); } -int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info) +static int btrfs_device_init_dev_stats(struct btrfs_device *device, + struct btrfs_path *path) { - struct btrfs_key key; - struct btrfs_root *dev_root = fs_info->dev_root; - struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; + struct btrfs_dev_stats_item *ptr; struct extent_buffer *eb; - int slot; - int ret = 0; + struct btrfs_key key; + int item_size; + int i, ret, slot; + + if (!device->fs_info->dev_root) + return 0; + + key.objectid = BTRFS_DEV_STATS_OBJECTID; + key.type = BTRFS_PERSISTENT_ITEM_KEY; + key.offset = device->devid; + ret = btrfs_search_slot(NULL, device->fs_info->dev_root, &key, path, 0, 0); + if (ret) { + for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) + btrfs_dev_stat_set(device, i, 0); + device->dev_stats_valid = 1; + btrfs_release_path(path); + return ret < 0 ? ret : 0; + } + slot = path->slots[0]; + eb = path->nodes[0]; + item_size = btrfs_item_size(eb, slot); + + ptr = btrfs_item_ptr(eb, slot, struct btrfs_dev_stats_item); + + for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) { + if (item_size >= (1 + i) * sizeof(__le64)) + btrfs_dev_stat_set(device, i, + btrfs_dev_stats_value(eb, ptr, i)); + else + btrfs_dev_stat_set(device, i, 0); + } + + device->dev_stats_valid = 1; + btrfs_dev_stat_print_on_load(device); + btrfs_release_path(path); + + return 0; +} + +int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info) +{ + struct btrfs_fs_devices *fs_devices = fs_info->fs_devices, *seed_devs; struct btrfs_device *device; - struct btrfs_path *path = NULL; - int i; + BTRFS_PATH_AUTO_FREE(path); + int ret = 0; path = btrfs_alloc_path(); if (!path) @@ -7179,43 +7630,20 @@ int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info) mutex_lock(&fs_devices->device_list_mutex); list_for_each_entry(device, &fs_devices->devices, dev_list) { - int item_size; - struct btrfs_dev_stats_item *ptr; - - key.objectid = BTRFS_DEV_STATS_OBJECTID; - key.type = BTRFS_PERSISTENT_ITEM_KEY; - key.offset = device->devid; - ret = btrfs_search_slot(NULL, dev_root, &key, path, 0, 0); - if (ret) { - for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) - btrfs_dev_stat_set(device, i, 0); - device->dev_stats_valid = 1; - btrfs_release_path(path); - continue; - } - slot = path->slots[0]; - eb = path->nodes[0]; - item_size = btrfs_item_size_nr(eb, slot); - - ptr = btrfs_item_ptr(eb, slot, - struct btrfs_dev_stats_item); - - for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) { - if (item_size >= (1 + i) * sizeof(__le64)) - btrfs_dev_stat_set(device, i, - btrfs_dev_stats_value(eb, ptr, i)); - else - btrfs_dev_stat_set(device, i, 0); + ret = btrfs_device_init_dev_stats(device, path); + if (ret) + goto out; + } + list_for_each_entry(seed_devs, &fs_devices->seed_list, seed_list) { + list_for_each_entry(device, &seed_devs->devices, dev_list) { + ret = btrfs_device_init_dev_stats(device, path); + if (ret) + goto out; } - - device->dev_stats_valid = 1; - btrfs_dev_stat_print_on_load(device); - btrfs_release_path(path); } +out: mutex_unlock(&fs_devices->device_list_mutex); - - btrfs_free_path(path); - return ret < 0 ? ret : 0; + return ret; } static int update_dev_stat_item(struct btrfs_trans_handle *trans, @@ -7223,7 +7651,7 @@ static int update_dev_stat_item(struct btrfs_trans_handle *trans, { struct btrfs_fs_info *fs_info = trans->fs_info; struct btrfs_root *dev_root = fs_info->dev_root; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_key key; struct extent_buffer *eb; struct btrfs_dev_stats_item *ptr; @@ -7239,21 +7667,21 @@ static int update_dev_stat_item(struct btrfs_trans_handle *trans, return -ENOMEM; ret = btrfs_search_slot(trans, dev_root, &key, path, -1, 1); if (ret < 0) { - btrfs_warn_in_rcu(fs_info, + btrfs_warn(fs_info, "error %d while searching for dev_stats item for device %s", - ret, rcu_str_deref(device->name)); - goto out; + ret, btrfs_dev_name(device)); + return ret; } if (ret == 0 && - btrfs_item_size_nr(path->nodes[0], path->slots[0]) < sizeof(*ptr)) { + btrfs_item_size(path->nodes[0], path->slots[0]) < sizeof(*ptr)) { /* need to delete old one and insert a new one */ ret = btrfs_del_item(trans, dev_root, path); if (ret != 0) { - btrfs_warn_in_rcu(fs_info, + btrfs_warn(fs_info, "delete too small dev_stats item for device %s failed %d", - rcu_str_deref(device->name), ret); - goto out; + btrfs_dev_name(device), ret); + return ret; } ret = 1; } @@ -7264,10 +7692,10 @@ static int update_dev_stat_item(struct btrfs_trans_handle *trans, ret = btrfs_insert_empty_item(trans, dev_root, path, &key, sizeof(*ptr)); if (ret < 0) { - btrfs_warn_in_rcu(fs_info, + btrfs_warn(fs_info, "insert dev_stats item for device %s failed %d", - rcu_str_deref(device->name), ret); - goto out; + btrfs_dev_name(device), ret); + return ret; } } @@ -7276,10 +7704,6 @@ static int update_dev_stat_item(struct btrfs_trans_handle *trans, for (i = 0; i < BTRFS_DEV_STAT_VALUES_MAX; i++) btrfs_set_dev_stats_value(eb, ptr, i, btrfs_dev_stat_read(device, i)); - btrfs_mark_buffer_dirty(eb); - -out: - btrfs_free_path(path); return ret; } @@ -7326,16 +7750,12 @@ int btrfs_run_dev_stats(struct btrfs_trans_handle *trans) void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index) { btrfs_dev_stat_inc(dev, index); - btrfs_dev_stat_print_on_error(dev); -} -static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev) -{ if (!dev->dev_stats_valid) return; - btrfs_err_rl_in_rcu(dev->fs_info, + btrfs_err_rl(dev->fs_info, "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u", - rcu_str_deref(dev->name), + btrfs_dev_name(dev), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS), @@ -7353,9 +7773,9 @@ static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev) if (i == BTRFS_DEV_STAT_VALUES_MAX) return; /* all values == 0, suppress message */ - btrfs_info_in_rcu(dev->fs_info, + btrfs_info(dev->fs_info, "bdev %s errs: wr %u, rd %u, flush %u, corrupt %u, gen %u", - rcu_str_deref(dev->name), + btrfs_dev_name(dev), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_WRITE_ERRS), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_READ_ERRS), btrfs_dev_stat_read(dev, BTRFS_DEV_STAT_FLUSH_ERRS), @@ -7366,13 +7786,14 @@ static void btrfs_dev_stat_print_on_load(struct btrfs_device *dev) int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info, struct btrfs_ioctl_get_dev_stats *stats) { + BTRFS_DEV_LOOKUP_ARGS(args); struct btrfs_device *dev; struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; int i; mutex_lock(&fs_devices->device_list_mutex); - dev = btrfs_find_device(fs_info->fs_devices, stats->devid, NULL, NULL, - true); + args.devid = stats->devid; + dev = btrfs_find_device(fs_info->fs_devices, &args); mutex_unlock(&fs_devices->device_list_mutex); if (!dev) { @@ -7412,7 +7833,7 @@ void btrfs_commit_device_sizes(struct btrfs_transaction *trans) { struct btrfs_device *curr, *next; - ASSERT(trans->state == TRANS_STATE_COMMIT_DOING); + ASSERT(trans->state == TRANS_STATE_COMMIT_DOING, "state=%d" , trans->state); if (list_empty(&trans->dev_update_list)) return; @@ -7432,24 +7853,6 @@ void btrfs_commit_device_sizes(struct btrfs_transaction *trans) mutex_unlock(&trans->fs_info->chunk_mutex); } -void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info) -{ - struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; - while (fs_devices) { - fs_devices->fs_info = fs_info; - fs_devices = fs_devices->seed; - } -} - -void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info) -{ - struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; - while (fs_devices) { - fs_devices->fs_info = NULL; - fs_devices = fs_devices->seed; - } -} - /* * Multiplicity factor for simple profiles: DUP, RAID1-like and RAID10. */ @@ -7460,26 +7863,20 @@ int btrfs_bg_type_to_factor(u64 flags) return btrfs_raid_array[index].ncopies; } - - static int verify_one_dev_extent(struct btrfs_fs_info *fs_info, u64 chunk_offset, u64 devid, u64 physical_offset, u64 physical_len) { - struct extent_map_tree *em_tree = &fs_info->mapping_tree; - struct extent_map *em; - struct map_lookup *map; + struct btrfs_dev_lookup_args args = { .devid = devid }; + struct btrfs_chunk_map *map; struct btrfs_device *dev; u64 stripe_len; bool found = false; int ret = 0; int i; - read_lock(&em_tree->lock); - em = lookup_extent_mapping(em_tree, chunk_offset, 1); - read_unlock(&em_tree->lock); - - if (!em) { + map = btrfs_find_chunk_map(fs_info, chunk_offset, 1); + if (unlikely(!map)) { btrfs_err(fs_info, "dev extent physical offset %llu on devid %llu doesn't have corresponding chunk", physical_offset, devid); @@ -7487,25 +7884,34 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info, goto out; } - map = em->map_lookup; - stripe_len = calc_stripe_length(map->type, em->len, map->num_stripes); - if (physical_len != stripe_len) { + stripe_len = btrfs_calc_stripe_length(map); + if (unlikely(physical_len != stripe_len)) { btrfs_err(fs_info, "dev extent physical offset %llu on devid %llu length doesn't match chunk %llu, have %llu expect %llu", - physical_offset, devid, em->start, physical_len, + physical_offset, devid, map->start, physical_len, stripe_len); ret = -EUCLEAN; goto out; } + /* + * Very old mkfs.btrfs (before v4.15) will not respect the reserved + * space. Although kernel can handle it without problem, better to warn + * the users. + */ + if (physical_offset < BTRFS_DEVICE_RANGE_RESERVED) + btrfs_warn(fs_info, + "devid %llu physical %llu len %llu inside the reserved space", + devid, physical_offset, physical_len); + for (i = 0; i < map->num_stripes; i++) { - if (map->stripes[i].dev->devid == devid && - map->stripes[i].physical == physical_offset) { + if (unlikely(map->stripes[i].dev->devid == devid && + map->stripes[i].physical == physical_offset)) { found = true; if (map->verified_stripes >= map->num_stripes) { btrfs_err(fs_info, "too many dev extents for chunk %llu found", - em->start); + map->start); ret = -EUCLEAN; goto out; } @@ -7513,34 +7919,22 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info, break; } } - if (!found) { + if (unlikely(!found)) { btrfs_err(fs_info, "dev extent physical offset %llu devid %llu has no corresponding chunk", physical_offset, devid); ret = -EUCLEAN; } - /* Make sure no dev extent is beyond device bondary */ - dev = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true); - if (!dev) { + /* Make sure no dev extent is beyond device boundary */ + dev = btrfs_find_device(fs_info->fs_devices, &args); + if (unlikely(!dev)) { btrfs_err(fs_info, "failed to find devid %llu", devid); ret = -EUCLEAN; goto out; } - /* It's possible this device is a dummy for seed device */ - if (dev->disk_total_bytes == 0) { - dev = btrfs_find_device(fs_info->fs_devices->seed, devid, NULL, - NULL, false); - if (!dev) { - btrfs_err(fs_info, "failed to find seed devid %llu", - devid); - ret = -EUCLEAN; - goto out; - } - } - - if (physical_offset + physical_len > dev->disk_total_bytes) { + if (unlikely(physical_offset + physical_len > dev->disk_total_bytes)) { btrfs_err(fs_info, "dev extent devid %llu physical offset %llu len %llu is beyond device boundary %llu", devid, physical_offset, physical_len, @@ -7548,33 +7942,45 @@ static int verify_one_dev_extent(struct btrfs_fs_info *fs_info, ret = -EUCLEAN; goto out; } + + if (dev->zone_info) { + u64 zone_size = dev->zone_info->zone_size; + + if (unlikely(!IS_ALIGNED(physical_offset, zone_size) || + !IS_ALIGNED(physical_len, zone_size))) { + btrfs_err(fs_info, +"zoned: dev extent devid %llu physical offset %llu len %llu is not aligned to device zone", + devid, physical_offset, physical_len); + ret = -EUCLEAN; + goto out; + } + } + out: - free_extent_map(em); + btrfs_free_chunk_map(map); return ret; } static int verify_chunk_dev_extent_mapping(struct btrfs_fs_info *fs_info) { - struct extent_map_tree *em_tree = &fs_info->mapping_tree; - struct extent_map *em; struct rb_node *node; int ret = 0; - read_lock(&em_tree->lock); - for (node = rb_first_cached(&em_tree->map); node; node = rb_next(node)) { - em = rb_entry(node, struct extent_map, rb_node); - if (em->map_lookup->num_stripes != - em->map_lookup->verified_stripes) { + read_lock(&fs_info->mapping_tree_lock); + for (node = rb_first_cached(&fs_info->mapping_tree); node; node = rb_next(node)) { + struct btrfs_chunk_map *map; + + map = rb_entry(node, struct btrfs_chunk_map, rb_node); + if (unlikely(map->num_stripes != map->verified_stripes)) { btrfs_err(fs_info, "chunk %llu has missing dev extent, have %d expect %d", - em->start, em->map_lookup->verified_stripes, - em->map_lookup->num_stripes); + map->start, map->verified_stripes, map->num_stripes); ret = -EUCLEAN; goto out; } } out: - read_unlock(&em_tree->lock); + read_unlock(&fs_info->mapping_tree_lock); return ret; } @@ -7587,13 +7993,26 @@ out: */ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info) { - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_root *root = fs_info->dev_root; struct btrfs_key key; u64 prev_devid = 0; u64 prev_dev_ext_end = 0; int ret = 0; + /* + * We don't have a dev_root because we mounted with ignorebadroots and + * failed to load the root, so we want to skip the verification in this + * case for sure. + * + * However if the dev root is fine, but the tree itself is corrupted + * we'd still fail to mount. This verification is only to make sure + * writes can happen safely, so instead just bypass this check + * completely in the case of IGNOREBADROOTS. + */ + if (btrfs_test_opt(fs_info, IGNOREBADROOTS)) + return 0; + key.objectid = 1; key.type = BTRFS_DEV_EXTENT_KEY; key.offset = 0; @@ -7605,17 +8024,15 @@ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info) path->reada = READA_FORWARD; ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); if (ret < 0) - goto out; + return ret; if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) { - ret = btrfs_next_item(root, path); + ret = btrfs_next_leaf(root, path); if (ret < 0) - goto out; + return ret; /* No dev extents at all? Not good */ - if (ret > 0) { - ret = -EUCLEAN; - goto out; - } + if (unlikely(ret > 0)) + return -EUCLEAN; } while (1) { struct extent_buffer *leaf = path->nodes[0]; @@ -7637,24 +8054,23 @@ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info) physical_len = btrfs_dev_extent_length(leaf, dext); /* Check if this dev extent overlaps with the previous one */ - if (devid == prev_devid && physical_offset < prev_dev_ext_end) { + if (unlikely(devid == prev_devid && physical_offset < prev_dev_ext_end)) { btrfs_err(fs_info, "dev extent devid %llu physical offset %llu overlap with previous dev extent end %llu", devid, physical_offset, prev_dev_ext_end); - ret = -EUCLEAN; - goto out; + return -EUCLEAN; } ret = verify_one_dev_extent(fs_info, chunk_offset, devid, physical_offset, physical_len); if (ret < 0) - goto out; + return ret; prev_devid = devid; prev_dev_ext_end = physical_offset + physical_len; ret = btrfs_next_item(root, path); if (ret < 0) - goto out; + return ret; if (ret > 0) { ret = 0; break; @@ -7662,10 +8078,7 @@ int btrfs_verify_dev_extents(struct btrfs_fs_info *fs_info) } /* Ensure all chunks have corresponding dev extents */ - ret = verify_chunk_dev_extent_mapping(fs_info); -out: - btrfs_free_path(path); - return ret; + return verify_chunk_dev_extent_mapping(fs_info); } /* @@ -7691,3 +8104,150 @@ bool btrfs_pinned_by_swapfile(struct btrfs_fs_info *fs_info, void *ptr) spin_unlock(&fs_info->swapfile_pins_lock); return node != NULL; } + +static int relocating_repair_kthread(void *data) +{ + struct btrfs_block_group *cache = data; + struct btrfs_fs_info *fs_info = cache->fs_info; + u64 target; + int ret = 0; + + target = cache->start; + btrfs_put_block_group(cache); + + guard(super_write)(fs_info->sb); + + if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { + btrfs_info(fs_info, + "zoned: skip relocating block group %llu to repair: EBUSY", + target); + return -EBUSY; + } + + mutex_lock(&fs_info->reclaim_bgs_lock); + + /* Ensure block group still exists */ + cache = btrfs_lookup_block_group(fs_info, target); + if (!cache) + goto out; + + if (!test_bit(BLOCK_GROUP_FLAG_RELOCATING_REPAIR, &cache->runtime_flags)) + goto out; + + ret = btrfs_may_alloc_data_chunk(fs_info, target); + if (ret < 0) + goto out; + + btrfs_info(fs_info, + "zoned: relocating block group %llu to repair IO failure", + target); + ret = btrfs_relocate_chunk(fs_info, target, true); + +out: + if (cache) + btrfs_put_block_group(cache); + mutex_unlock(&fs_info->reclaim_bgs_lock); + btrfs_exclop_finish(fs_info); + + return ret; +} + +bool btrfs_repair_one_zone(struct btrfs_fs_info *fs_info, u64 logical) +{ + struct btrfs_block_group *cache; + + if (!btrfs_is_zoned(fs_info)) + return false; + + /* Do not attempt to repair in degraded state */ + if (btrfs_test_opt(fs_info, DEGRADED)) + return true; + + cache = btrfs_lookup_block_group(fs_info, logical); + if (!cache) + return true; + + if (test_and_set_bit(BLOCK_GROUP_FLAG_RELOCATING_REPAIR, &cache->runtime_flags)) { + btrfs_put_block_group(cache); + return true; + } + + kthread_run(relocating_repair_kthread, cache, + "btrfs-relocating-repair"); + + return true; +} + +static void map_raid56_repair_block(struct btrfs_io_context *bioc, + struct btrfs_io_stripe *smap, + u64 logical) +{ + int data_stripes = nr_bioc_data_stripes(bioc); + int i; + + for (i = 0; i < data_stripes; i++) { + u64 stripe_start = bioc->full_stripe_logical + + btrfs_stripe_nr_to_offset(i); + + if (logical >= stripe_start && + logical < stripe_start + BTRFS_STRIPE_LEN) + break; + } + ASSERT(i < data_stripes, "i=%d data_stripes=%d", i, data_stripes); + smap->dev = bioc->stripes[i].dev; + smap->physical = bioc->stripes[i].physical + + ((logical - bioc->full_stripe_logical) & + BTRFS_STRIPE_LEN_MASK); +} + +/* + * Map a repair write into a single device. + * + * A repair write is triggered by read time repair or scrub, which would only + * update the contents of a single device. + * Not update any other mirrors nor go through RMW path. + * + * Callers should ensure: + * + * - Call btrfs_bio_counter_inc_blocked() first + * - The range does not cross stripe boundary + * - Has a valid @mirror_num passed in. + */ +int btrfs_map_repair_block(struct btrfs_fs_info *fs_info, + struct btrfs_io_stripe *smap, u64 logical, + u32 length, int mirror_num) +{ + struct btrfs_io_context *bioc = NULL; + u64 map_length = length; + int mirror_ret = mirror_num; + int ret; + + ASSERT(mirror_num > 0, "mirror_num=%d", mirror_num); + + ret = btrfs_map_block(fs_info, BTRFS_MAP_WRITE, logical, &map_length, + &bioc, smap, &mirror_ret); + if (ret < 0) + return ret; + + /* The map range should not cross stripe boundary. */ + ASSERT(map_length >= length, "map_length=%llu length=%u", map_length, length); + + /* Already mapped to single stripe. */ + if (!bioc) + goto out; + + /* Map the RAID56 multi-stripe writes to a single one. */ + if (bioc->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) { + map_raid56_repair_block(bioc, smap, logical); + goto out; + } + + ASSERT(mirror_num <= bioc->num_stripes, + "mirror_num=%d num_stripes=%d", mirror_num, bioc->num_stripes); + smap->dev = bioc->stripes[mirror_num - 1].dev; + smap->physical = bioc->stripes[mirror_num - 1].physical; +out: + btrfs_put_bioc(bioc); + ASSERT(smap->dev); + return 0; +} |