diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index acf22ad6115c..7556f8319c60 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -37,7 +37,7 @@ extern struct kmem_cache *btrfs_transaction_cachep; extern struct kmem_cache *btrfs_bit_radix_cachep; extern struct kmem_cache *btrfs_path_cachep; -#define BTRFS_MAGIC "_B4RfS_M" +#define BTRFS_MAGIC "_B5RfS_M" #define BTRFS_MAX_LEVEL 8 @@ -238,6 +238,7 @@ struct btrfs_super_block { __le64 total_bytes; __le64 bytes_used; __le64 root_dir_objectid; + __le64 num_devices; __le32 sectorsize; __le32 nodesize; __le32 leafsize; @@ -440,6 +441,7 @@ struct btrfs_block_group_cache { }; struct btrfs_device; +struct btrfs_fs_devices; struct btrfs_fs_info { u8 fsid[BTRFS_FSID_SIZE]; struct btrfs_root *extent_root; @@ -489,7 +491,7 @@ struct btrfs_fs_info { u64 total_pinned; struct list_head dirty_cowonly_roots; - struct list_head devices; + struct btrfs_fs_devices *fs_devices; struct list_head space_info; spinlock_t delalloc_lock; spinlock_t new_trans_lock; @@ -677,6 +679,19 @@ BTRFS_SETGET_FUNCS(device_io_width, struct btrfs_dev_item, io_width, 32); BTRFS_SETGET_FUNCS(device_sector_size, struct btrfs_dev_item, sector_size, 32); BTRFS_SETGET_FUNCS(device_id, struct btrfs_dev_item, devid, 64); +BTRFS_SETGET_STACK_FUNCS(stack_device_type, struct btrfs_dev_item, type, 64); +BTRFS_SETGET_STACK_FUNCS(stack_device_total_bytes, struct btrfs_dev_item, + total_bytes, 64); +BTRFS_SETGET_STACK_FUNCS(stack_device_bytes_used, struct btrfs_dev_item, + bytes_used, 64); +BTRFS_SETGET_STACK_FUNCS(stack_device_io_align, struct btrfs_dev_item, + io_align, 32); +BTRFS_SETGET_STACK_FUNCS(stack_device_io_width, struct btrfs_dev_item, + io_width, 32); +BTRFS_SETGET_STACK_FUNCS(stack_device_sector_size, struct btrfs_dev_item, + sector_size, 32); +BTRFS_SETGET_STACK_FUNCS(stack_device_id, struct btrfs_dev_item, devid, 64); + static inline char *btrfs_device_uuid(struct btrfs_dev_item *d) { return (char *)d + offsetof(struct btrfs_dev_item, uuid); @@ -1106,6 +1121,8 @@ BTRFS_SETGET_STACK_FUNCS(super_stripesize, struct btrfs_super_block, stripesize, 32); BTRFS_SETGET_STACK_FUNCS(super_root_dir, struct btrfs_super_block, root_dir_objectid, 64); +BTRFS_SETGET_STACK_FUNCS(super_num_devices, struct btrfs_super_block, + num_devices, 64); static inline unsigned long btrfs_leaf_data(struct extent_buffer *l) { diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 4890151cd68d..f971a29e4f20 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -365,12 +365,12 @@ static int close_all_devices(struct btrfs_fs_info *fs_info) struct list_head *next; struct btrfs_device *device; - list = &fs_info->devices; - while(!list_empty(list)) { - next = list->next; - list_del(next); + list = &fs_info->fs_devices->devices; + list_for_each(next, list) { device = list_entry(next, struct btrfs_device, dev_list); - kfree(device); + if (device->bdev && device->bdev != fs_info->sb->s_bdev) + close_bdev_excl(device->bdev); + device->bdev = NULL; } return 0; } @@ -655,7 +655,8 @@ static int add_hasher(struct btrfs_fs_info *info, char *type) { return 0; } #endif -struct btrfs_root *open_ctree(struct super_block *sb) +struct btrfs_root *open_ctree(struct super_block *sb, + struct btrfs_fs_devices *fs_devices) { u32 sectorsize; u32 nodesize; @@ -697,8 +698,8 @@ struct btrfs_root *open_ctree(struct super_block *sb) fs_info->extent_root = extent_root; fs_info->chunk_root = chunk_root; fs_info->dev_root = dev_root; + fs_info->fs_devices = fs_devices; INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots); - INIT_LIST_HEAD(&fs_info->devices); INIT_LIST_HEAD(&fs_info->space_info); btrfs_mapping_init(&fs_info->mapping_tree); fs_info->sb = sb; @@ -779,6 +780,12 @@ struct btrfs_root *open_ctree(struct super_block *sb) if (!btrfs_super_root(disk_super)) goto fail_sb_buffer; + if (btrfs_super_num_devices(disk_super) != fs_devices->num_devices) { + printk("Btrfs: wanted %llu devices, but found %llu\n", + (unsigned long long)btrfs_super_num_devices(disk_super), + (unsigned long long)fs_devices->num_devices); + goto fail_sb_buffer; + } nodesize = btrfs_super_nodesize(disk_super); leafsize = btrfs_super_leafsize(disk_super); sectorsize = btrfs_super_sectorsize(disk_super); @@ -799,8 +806,6 @@ struct btrfs_root *open_ctree(struct super_block *sb) } mutex_lock(&fs_info->fs_mutex); - ret = btrfs_read_super_device(tree_root, fs_info->sb_buffer); - BUG_ON(ret); ret = btrfs_read_sys_array(tree_root); BUG_ON(ret); @@ -859,6 +864,7 @@ fail_sb_buffer: fail_iput: iput(fs_info->btree_inode); fail: + close_all_devices(fs_info); kfree(extent_root); kfree(tree_root); kfree(fs_info); diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index 206cb48638f7..b7cbc58a5553 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -21,6 +21,7 @@ #define BTRFS_SUPER_INFO_OFFSET (16 * 1024) struct btrfs_device; +struct btrfs_fs_devices; struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize); @@ -29,7 +30,8 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize); int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf); -struct btrfs_root *open_ctree(struct super_block *sb); +struct btrfs_root *open_ctree(struct super_block *sb, + struct btrfs_fs_devices *fs_devices); int close_ctree(struct btrfs_root *root); int write_ctree_super(struct btrfs_trans_handle *trans, struct btrfs_root *root); diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index 8c6290665d49..4551e82013c8 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -22,8 +22,10 @@ #define BTRFS_IOCTL_MAGIC 0x94 #define BTRFS_VOL_NAME_MAX 255 +#define BTRFS_PATH_NAME_MAX 4095 + struct btrfs_ioctl_vol_args { - char name[BTRFS_VOL_NAME_MAX + 1]; + char name[BTRFS_PATH_NAME_MAX + 1]; }; #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ @@ -32,4 +34,6 @@ struct btrfs_ioctl_vol_args { struct btrfs_ioctl_vol_args) #define BTRFS_IOC_RESIZE _IOW(BTRFS_IOCTL_MAGIC, 3, \ struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_SCAN_DEV _IOW(BTRFS_IOCTL_MAGIC, 4, \ + struct btrfs_ioctl_vol_args) #endif diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 67ed216df475..9624923a33dc 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -44,6 +44,7 @@ #include "ioctl.h" #include "print-tree.h" #include "xattr.h" +#include "volumes.h" #define BTRFS_SUPER_MAGIC 0x9123683E @@ -216,7 +217,9 @@ static int parse_options (char * options, return 1; } -static int btrfs_fill_super(struct super_block * sb, void * data, int silent) +static int btrfs_fill_super(struct super_block * sb, + struct btrfs_fs_devices *fs_devices, + void * data, int silent) { struct inode * inode; struct dentry * root_dentry; @@ -231,7 +234,7 @@ static int btrfs_fill_super(struct super_block * sb, void * data, int silent) sb->s_xattr = btrfs_xattr_handlers; sb->s_time_gran = 1; - tree_root = open_ctree(sb); + tree_root = open_ctree(sb, fs_devices); if (!tree_root || IS_ERR(tree_root)) { printk("btrfs: open_ctree failed\n"); @@ -334,18 +337,23 @@ static int test_bdev_super(struct super_block *s, void *data) int btrfs_get_sb_bdev(struct file_system_type *fs_type, int flags, const char *dev_name, void *data, - int (*fill_super)(struct super_block *, void *, int), struct vfsmount *mnt, const char *subvol) { struct block_device *bdev = NULL; struct super_block *s; struct dentry *root; + struct btrfs_fs_devices *fs_devices = NULL; int error = 0; - bdev = open_bdev_excl(dev_name, flags, fs_type); - if (IS_ERR(bdev)) - return PTR_ERR(bdev); + error = btrfs_scan_one_device(dev_name, flags, fs_type, &fs_devices); + if (error) + return error; + error = btrfs_open_devices(fs_devices, flags, fs_type); + if (error) + return error; + + bdev = fs_devices->lowest_bdev; /* * once the super is inserted into the list by sget, s_umount * will protect the lockfs code from trying to start a snapshot @@ -372,7 +380,8 @@ int btrfs_get_sb_bdev(struct file_system_type *fs_type, s->s_flags = flags; strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id)); sb_set_blocksize(s, block_size(bdev)); - error = fill_super(s, data, flags & MS_SILENT ? 1 : 0); + error = btrfs_fill_super(s, fs_devices, data, + flags & MS_SILENT ? 1 : 0); if (error) { up_write(&s->s_umount); deactivate_super(s); @@ -408,7 +417,7 @@ int btrfs_get_sb_bdev(struct file_system_type *fs_type, error_s: error = PTR_ERR(s); error_bdev: - close_bdev_excl(bdev); + btrfs_close_devices(fs_devices); error: return error; } @@ -421,8 +430,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type, char *subvol_name = NULL; parse_options((char *)data, NULL, &subvol_name); - ret = btrfs_get_sb_bdev(fs_type, flags, dev_name, data, - btrfs_fill_super, mnt, + ret = btrfs_get_sb_bdev(fs_type, flags, dev_name, data, mnt, subvol_name ? subvol_name : "default"); if (subvol_name) kfree(subvol_name); @@ -445,13 +453,6 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) return 0; } -static long btrfs_control_ioctl(struct file *file, unsigned int cmd, - unsigned long arg) -{ - printk("btrfs control ioctl %d\n", cmd); - return 0; -} - static struct file_system_type btrfs_fs_type = { .owner = THIS_MODULE, .name = "btrfs", @@ -460,6 +461,31 @@ static struct file_system_type btrfs_fs_type = { .fs_flags = FS_REQUIRES_DEV, }; +static long btrfs_control_ioctl(struct file *file, unsigned int cmd, + unsigned long arg) +{ + struct btrfs_ioctl_vol_args *vol; + struct btrfs_fs_devices *fs_devices; + int ret; + int len; + + vol = kmalloc(sizeof(*vol), GFP_KERNEL); + if (copy_from_user(vol, (void __user *)arg, sizeof(*vol))) { + ret = -EFAULT; + goto out; + } + len = strnlen(vol->name, BTRFS_PATH_NAME_MAX); + switch (cmd) { + case BTRFS_IOC_SCAN_DEV: + ret = btrfs_scan_one_device(vol->name, MS_RDONLY, + &btrfs_fs_type, &fs_devices); + break; + } +out: + kfree(vol); + return 0; +} + static void btrfs_write_super_lockfs(struct super_block *sb) { struct btrfs_root *root = btrfs_sb(sb); @@ -567,6 +593,7 @@ static void __exit exit_btrfs_fs(void) btrfs_interface_exit(); unregister_filesystem(&btrfs_fs_type); btrfs_exit_sysfs(); + btrfs_cleanup_fs_uuids(); } module_init(init_btrfs_fs) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 16fb6bbe6e28..263f01cc3db4 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -17,6 +17,7 @@ */ #include #include +#include #include "ctree.h" #include "extent_map.h" #include "disk-io.h" @@ -28,6 +29,215 @@ struct map_lookup { struct btrfs_device *dev; u64 physical; }; +static DEFINE_MUTEX(uuid_mutex); +static LIST_HEAD(fs_uuids); + +int btrfs_cleanup_fs_uuids(void) +{ + struct btrfs_fs_devices *fs_devices; + struct list_head *uuid_cur; + struct list_head *devices_cur; + struct btrfs_device *dev; + + list_for_each(uuid_cur, &fs_uuids) { + fs_devices = list_entry(uuid_cur, struct btrfs_fs_devices, + list); + while(!list_empty(&fs_devices->devices)) { + devices_cur = fs_devices->devices.next; + dev = list_entry(devices_cur, struct btrfs_device, + dev_list); + printk("uuid cleanup finds %s\n", dev->name); + if (dev->bdev) { + printk("closing\n"); + close_bdev_excl(dev->bdev); + } + list_del(&dev->dev_list); + kfree(dev); + } + } + return 0; +} + +static struct btrfs_device *__find_device(struct list_head *head, u64 devid) +{ + struct btrfs_device *dev; + struct list_head *cur; + + list_for_each(cur, head) { + dev = list_entry(cur, struct btrfs_device, dev_list); + if (dev->devid == devid) + return dev; + } + return NULL; +} + +static struct btrfs_fs_devices *find_fsid(u8 *fsid) +{ + struct list_head *cur; + struct btrfs_fs_devices *fs_devices; + + list_for_each(cur, &fs_uuids) { + fs_devices = list_entry(cur, struct btrfs_fs_devices, list); + if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0) + return fs_devices; + } + return NULL; +} + +static int device_list_add(const char *path, + struct btrfs_super_block *disk_super, + u64 devid, struct btrfs_fs_devices **fs_devices_ret) +{ + struct btrfs_device *device; + struct btrfs_fs_devices *fs_devices; + u64 found_transid = btrfs_super_generation(disk_super); + + fs_devices = find_fsid(disk_super->fsid); + if (!fs_devices) { + fs_devices = kmalloc(sizeof(*fs_devices), GFP_NOFS); + if (!fs_devices) + return -ENOMEM; + INIT_LIST_HEAD(&fs_devices->devices); + list_add(&fs_devices->list, &fs_uuids); + memcpy(fs_devices->fsid, disk_super->fsid, BTRFS_FSID_SIZE); + fs_devices->latest_devid = devid; + fs_devices->latest_trans = found_transid; + fs_devices->lowest_devid = (u64)-1; + fs_devices->num_devices = 0; + device = NULL; + } else { + device = __find_device(&fs_devices->devices, devid); + } + if (!device) { + device = kzalloc(sizeof(*device), GFP_NOFS); + if (!device) { + /* we can safely leave the fs_devices entry around */ + return -ENOMEM; + } + device->devid = devid; + device->name = kstrdup(path, GFP_NOFS); + if (!device->name) { + kfree(device); + return -ENOMEM; + } + list_add(&device->dev_list, &fs_devices->devices); + fs_devices->num_devices++; + } + + if (found_transid > fs_devices->latest_trans) { + fs_devices->latest_devid = devid; + fs_devices->latest_trans = found_transid; + } + if (fs_devices->lowest_devid > devid) { + fs_devices->lowest_devid = devid; + printk("lowest devid now %Lu\n", devid); + } + *fs_devices_ret = fs_devices; + return 0; +} + +int btrfs_close_devices(struct btrfs_fs_devices *fs_devices) +{ + struct list_head *head = &fs_devices->devices; + struct list_head *cur; + struct btrfs_device *device; + + mutex_lock(&uuid_mutex); + list_for_each(cur, head) { + device = list_entry(cur, struct btrfs_device, dev_list); + if (device->bdev) { + close_bdev_excl(device->bdev); + printk("close devices closes %s\n", device->name); + } + device->bdev = NULL; + } + mutex_unlock(&uuid_mutex); + return 0; +} + +int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, + int flags, void *holder) +{ + struct block_device *bdev; + struct list_head *head = &fs_devices->devices; + struct list_head *cur; + struct btrfs_device *device; + int ret; + + mutex_lock(&uuid_mutex); + list_for_each(cur, head) { + device = list_entry(cur, struct btrfs_device, dev_list); + bdev = open_bdev_excl(device->name, flags, holder); +printk("opening %s devid %Lu\n", device->name, device->devid); + if (IS_ERR(bdev)) { + printk("open %s failed\n", device->name); + ret = PTR_ERR(bdev); + goto fail; + } + if (device->devid == fs_devices->latest_devid) + fs_devices->latest_bdev = bdev; + if (device->devid == fs_devices->lowest_devid) { + fs_devices->lowest_bdev = bdev; +printk("lowest bdev %s\n", device->name); + } + device->bdev = bdev; + } + mutex_unlock(&uuid_mutex); + return 0; +fail: + mutex_unlock(&uuid_mutex); + btrfs_close_devices(fs_devices); + return ret; +} + +int btrfs_scan_one_device(const char *path, int flags, void *holder, + struct btrfs_fs_devices **fs_devices_ret) +{ + struct btrfs_super_block *disk_super; + struct block_device *bdev; + struct buffer_head *bh; + int ret; + u64 devid; + + mutex_lock(&uuid_mutex); + + printk("scan one opens %s\n", path); + bdev = open_bdev_excl(path, flags, holder); + + if (IS_ERR(bdev)) { + printk("open failed\n"); + ret = PTR_ERR(bdev); + goto error; + } + + ret = set_blocksize(bdev, 4096); + if (ret) + goto error_close; + bh = __bread(bdev, BTRFS_SUPER_INFO_OFFSET / 4096, 4096); + if (!bh) { + ret = -EIO; + goto error_close; + } + disk_super = (struct btrfs_super_block *)bh->b_data; + if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC, + sizeof(disk_super->magic))) { + printk("no btrfs found on %s\n", path); + ret = -ENOENT; + goto error_brelse; + } + devid = le64_to_cpu(disk_super->dev_item.devid); + printk("found device %Lu on %s\n", devid, path); + ret = device_list_add(path, disk_super, devid, fs_devices_ret); + +error_brelse: + brelse(bh); +error_close: + close_bdev_excl(bdev); + printk("scan one closes bdev %s\n", path); +error: + mutex_unlock(&uuid_mutex); + return ret; +} /* * this uses a pretty simple search, the expectation is that it is @@ -56,6 +266,10 @@ static int find_free_dev_extent(struct btrfs_trans_handle *trans, /* FIXME use last free of some kind */ + /* we don't want to overwrite the superblock on the drive, + * so we make sure to start at an offset of at least 1MB + */ + search_start = max((u64)1024 * 1024, search_start); key.objectid = device->devid; key.offset = search_start; key.type = BTRFS_DEV_EXTENT_KEY; @@ -285,6 +499,7 @@ int btrfs_add_device(struct btrfs_trans_handle *trans, leaf = path->nodes[0]; dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item); + device->devid = free_devid; btrfs_set_device_id(leaf, dev_item, device->devid); btrfs_set_device_type(leaf, dev_item, device->type); btrfs_set_device_io_align(leaf, dev_item, device->io_align); @@ -382,7 +597,7 @@ int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, struct btrfs_device *device = NULL; struct btrfs_chunk *chunk; struct list_head private_devs; - struct list_head *dev_list = &extent_root->fs_info->devices; + struct list_head *dev_list = &extent_root->fs_info->fs_devices->devices; struct list_head *cur; struct extent_map_tree *em_tree; struct map_lookup *map; @@ -449,7 +664,7 @@ again: key.objectid, calc_size, &dev_offset); BUG_ON(ret); - +printk("alloc chunk size %Lu from dev %Lu\n", calc_size, device->devid); device->bytes_used += calc_size; ret = btrfs_update_device(trans, device); BUG_ON(ret); @@ -592,17 +807,9 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio) struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid) { - struct btrfs_device *dev; - struct list_head *cur = root->fs_info->devices.next; - struct list_head *head = &root->fs_info->devices; + struct list_head *head = &root->fs_info->fs_devices->devices; - while(cur != head) { - dev = list_entry(cur, struct btrfs_device, dev_list); - if (dev->devid == devid) - return dev; - cur = cur->next; - } - return NULL; + return __find_device(head, devid); } static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key, @@ -699,15 +906,16 @@ static int read_one_dev(struct btrfs_root *root, devid = btrfs_device_id(leaf, dev_item); device = btrfs_find_device(root, devid); if (!device) { + printk("warning devid %Lu not found already\n", devid); device = kmalloc(sizeof(*device), GFP_NOFS); if (!device) return -ENOMEM; - list_add(&device->dev_list, &root->fs_info->devices); + list_add(&device->dev_list, + &root->fs_info->fs_devices->devices); } fill_device_from_item(leaf, dev_item, device); device->dev_root = root->fs_info->dev_root; - device->bdev = root->fs_info->sb->s_bdev; ret = 0; #if 0 ret = btrfs_open_device(device); diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index 20259128152e..12f297eb0559 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -24,6 +24,8 @@ struct btrfs_device { struct block_device *bdev; + char *name; + /* the internal btrfs device id */ u64 devid; @@ -49,6 +51,20 @@ struct btrfs_device { u8 uuid[BTRFS_DEV_UUID_SIZE]; }; +struct btrfs_fs_devices { + u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */ + + /* the device with this id has the most recent coyp of the super */ + u64 latest_devid; + u64 latest_trans; + u64 lowest_devid; + u64 num_devices; + struct block_device *latest_bdev; + struct block_device *lowest_bdev; + struct list_head devices; + struct list_head list; +}; + int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans, struct btrfs_device *device, u64 owner, u64 num_bytes, u64 *start); @@ -67,4 +83,13 @@ int btrfs_read_super_device(struct btrfs_root *root, struct extent_buffer *buf); int btrfs_map_block(struct btrfs_mapping_tree *map_tree, u64 logical, u64 *phys, u64 *length, struct btrfs_device **dev); +int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, + int flags, void *holder); +int btrfs_scan_one_device(const char *path, int flags, void *holder, + struct btrfs_fs_devices **fs_devices_ret); +int btrfs_close_devices(struct btrfs_fs_devices *fs_devices); +int btrfs_add_device(struct btrfs_trans_handle *trans, + struct btrfs_root *root, + struct btrfs_device *device); +int btrfs_cleanup_fs_uuids(void); #endif