Btrfs: Seed device support

Seed device is a special btrfs with SEEDING super flag
set and can only be mounted in read-only mode. Seed
devices allow people to create new btrfs on top of it.

The new FS contains the same contents as the seed device,
but it can be mounted in read-write mode.

This patch does the following:

1) split code in btrfs_alloc_chunk into two parts. The first part does makes
the newly allocated chunk usable, but does not do any operation that modifies
the chunk tree. The second part does the the chunk tree modifications. This
division is for the bootstrap step of adding storage to the seed device.

2) Update device management code to handle seed device.
The basic idea is: For an FS grown from seed devices, its
seed devices are put into a list. Seed devices are
opened on demand at mounting time. If any seed device is
missing or has been changed, btrfs kernel module will
refuse to mount the FS.

3) make btrfs_find_block_group not return NULL when all
block groups are read-only.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>

fs/btrfs/ctree.c

@@ -185,6 +185,10 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans,
 	btrfs_set_header_owner(cow, new_root_objectid);
 	btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN);
 
+	write_extent_buffer(cow, root->fs_info->fsid,
+			    (unsigned long)btrfs_header_fsid(cow),
+			    BTRFS_FSID_SIZE);
+
 	WARN_ON(btrfs_header_generation(buf) > trans->transid);
 	ret = btrfs_inc_ref(trans, new_root, buf, cow, NULL);
 	kfree(new_root);
@@ -274,6 +278,10 @@ int noinline __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	btrfs_set_header_owner(cow, root->root_key.objectid);
 	btrfs_clear_header_flag(cow, BTRFS_HEADER_FLAG_WRITTEN);
 
+	write_extent_buffer(cow, root->fs_info->fsid,
+			    (unsigned long)btrfs_header_fsid(cow),
+			    BTRFS_FSID_SIZE);
+
 	WARN_ON(btrfs_header_generation(buf) > trans->transid);
 	if (btrfs_header_generation(buf) != trans->transid) {
 		u32 nr_extents;

fs/btrfs/ctree.h

@@ -177,6 +177,9 @@ struct btrfs_dev_item {
 	/* type and info about this device */
 	__le64 type;
 
+	/* expected generation for this device */
+	__le64 generation;
+
 	/* grouping information for allocation decisions */
 	__le32 dev_group;
 
@@ -188,6 +191,9 @@ struct btrfs_dev_item {
 
 	/* btrfs generated uuid for this device */
 	u8 uuid[BTRFS_UUID_SIZE];
+
+	/* uuid of FS who owns this device */
+	u8 fsid[BTRFS_UUID_SIZE];
 } __attribute__ ((__packed__));
 
 struct btrfs_stripe {
@@ -263,6 +269,7 @@ struct btrfs_header {
 					sizeof(struct btrfs_item) - \
 					sizeof(struct btrfs_file_extent_item))
 
+#define BTRFS_SUPER_FLAG_SEEDING (1ULL << 32)
 
 /*
  * this is a very generous portion of the super block, giving us
@@ -278,7 +285,7 @@ struct btrfs_header {
 struct btrfs_super_block {
 	u8 csum[BTRFS_CSUM_SIZE];
 	/* the first 4 fields must match struct btrfs_header */
-	u8 fsid[16];    /* FS specific uuid */
+	u8 fsid[BTRFS_FSID_SIZE];    /* FS specific uuid */
 	__le64 bytenr; /* this block number */
 	__le64 flags;
 
@@ -941,6 +948,7 @@ BTRFS_SETGET_FUNCS(device_id, struct btrfs_dev_item, devid, 64);
 BTRFS_SETGET_FUNCS(device_group, struct btrfs_dev_item, dev_group, 32);
 BTRFS_SETGET_FUNCS(device_seek_speed, struct btrfs_dev_item, seek_speed, 8);
 BTRFS_SETGET_FUNCS(device_bandwidth, struct btrfs_dev_item, bandwidth, 8);
+BTRFS_SETGET_FUNCS(device_generation, struct btrfs_dev_item, generation, 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,
@@ -960,12 +968,19 @@ BTRFS_SETGET_STACK_FUNCS(stack_device_seek_speed, struct btrfs_dev_item,
 			 seek_speed, 8);
 BTRFS_SETGET_STACK_FUNCS(stack_device_bandwidth, struct btrfs_dev_item,
 			 bandwidth, 8);
+BTRFS_SETGET_STACK_FUNCS(stack_device_generation, struct btrfs_dev_item,
+			 generation, 64);
 
 static inline char *btrfs_device_uuid(struct btrfs_dev_item *d)
 {
 	return (char *)d + offsetof(struct btrfs_dev_item, uuid);
 }
 
+static inline char *btrfs_device_fsid(struct btrfs_dev_item *d)
+{
+	return (char *)d + offsetof(struct btrfs_dev_item, fsid);
+}
+
 BTRFS_SETGET_FUNCS(chunk_length, struct btrfs_chunk, length, 64);
 BTRFS_SETGET_FUNCS(chunk_owner, struct btrfs_chunk, owner, 64);
 BTRFS_SETGET_FUNCS(chunk_stripe_len, struct btrfs_chunk, stripe_len, 64);
@@ -1661,6 +1676,7 @@ int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
 			       struct extent_buffer *buf, u64 orig_start);
 int btrfs_add_dead_reloc_root(struct btrfs_root *root);
 int btrfs_cleanup_reloc_trees(struct btrfs_root *root);
+u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags);
 /* ctree.c */
 int btrfs_previous_item(struct btrfs_root *root,
 			struct btrfs_path *path, u64 min_objectid,

fs/btrfs/disk-io.c

@@ -345,6 +345,25 @@ int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
 	return 0;
 }
 
+static int check_tree_block_fsid(struct btrfs_root *root,
+				 struct extent_buffer *eb)
+{
+	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
+	u8 fsid[BTRFS_UUID_SIZE];
+	int ret = 1;
+
+	read_extent_buffer(eb, fsid, (unsigned long)btrfs_header_fsid(eb),
+			   BTRFS_FSID_SIZE);
+	while (fs_devices) {
+		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
+			ret = 0;
+			break;
+		}
+		fs_devices = fs_devices->seed;
+	}
+	return ret;
+}
+
 int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 			       struct extent_state *state)
 {
@@ -382,9 +401,7 @@ int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		ret = -EIO;
 		goto err;
 	}
-	if (memcmp_extent_buffer(eb, root->fs_info->fsid,
-				 (unsigned long)btrfs_header_fsid(eb),
-				 BTRFS_FSID_SIZE)) {
+	if (check_tree_block_fsid(root, eb)) {
 		printk("bad fsid on block %Lu\n", eb->start);
 		ret = -EIO;
 		goto err;
@@ -1558,9 +1575,11 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	if (!btrfs_super_root(disk_super))
 		goto fail_sb_buffer;
 
-	err = btrfs_parse_options(tree_root, options);
-	if (err)
+	ret = btrfs_parse_options(tree_root, options);
+	if (ret) {
+		err = ret;
 		goto fail_sb_buffer;
+	}
 
 	/*
 	 * we need to start all the end_io workers up front because the
@@ -1610,18 +1629,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	btrfs_start_workers(&fs_info->endio_write_workers,
 			    fs_info->thread_pool_size);
 
-	err = -EINVAL;
-	if (btrfs_super_num_devices(disk_super) > fs_devices->open_devices) {
-		printk("Btrfs: wanted %llu devices, but found %llu\n",
-		       (unsigned long long)btrfs_super_num_devices(disk_super),
-		       (unsigned long long)fs_devices->open_devices);
-		if (btrfs_test_opt(tree_root, DEGRADED))
-			printk("continuing in degraded mode\n");
-		else {
-			goto fail_sb_buffer;
-		}
-	}
-
 	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
 	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
 				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
@@ -1672,7 +1679,10 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	mutex_lock(&fs_info->chunk_mutex);
 	ret = btrfs_read_chunk_tree(chunk_root);
 	mutex_unlock(&fs_info->chunk_mutex);
-	BUG_ON(ret);
+	if (ret) {
+		printk("btrfs: failed to read chunk tree on %s\n", sb->s_id);
+		goto fail_chunk_root;
+	}
 
 	btrfs_close_extra_devices(fs_devices);
 
@@ -1684,7 +1694,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 					  btrfs_super_root(disk_super),
 					  blocksize, generation);
 	if (!tree_root->node)
-		goto fail_sb_buffer;
+		goto fail_chunk_root;
 
 
 	ret = find_and_setup_root(tree_root, fs_info,
@@ -1753,6 +1763,8 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	free_extent_buffer(extent_root->node);
 fail_tree_root:
 	free_extent_buffer(tree_root->node);
+fail_chunk_root:
+	free_extent_buffer(chunk_root->node);
 fail_sys_array:
 fail_sb_buffer:
 	btrfs_stop_workers(&fs_info->fixup_workers);
@@ -1823,9 +1835,10 @@ int write_all_supers(struct btrfs_root *root)
 			total_errors++;
 			continue;
 		}
-		if (!dev->in_fs_metadata)
+		if (!dev->in_fs_metadata || !dev->writeable)
 			continue;
 
+		btrfs_set_stack_device_generation(dev_item, 0);
 		btrfs_set_stack_device_type(dev_item, dev->type);
 		btrfs_set_stack_device_id(dev_item, dev->devid);
 		btrfs_set_stack_device_total_bytes(dev_item, dev->total_bytes);
@@ -1834,6 +1847,7 @@ int write_all_supers(struct btrfs_root *root)
 		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
 		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
 		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
+		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
 		flags = btrfs_super_flags(sb);
 		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);
 
@@ -1881,7 +1895,7 @@ int write_all_supers(struct btrfs_root *root)
 		dev = list_entry(cur, struct btrfs_device, dev_list);
 		if (!dev->bdev)
 			continue;
-		if (!dev->in_fs_metadata)
+		if (!dev->in_fs_metadata || !dev->writeable)
 			continue;
 
 		BUG_ON(!dev->pending_io);

fs/btrfs/extent-tree.c

@@ -355,7 +355,7 @@ __btrfs_find_block_group(struct btrfs_root *root,
 	if (search_start) {
 		struct btrfs_block_group_cache *shint;
 		shint = btrfs_lookup_first_block_group(info, search_start);
-		if (shint && block_group_bits(shint, data) && !shint->ro) {
+		if (shint && block_group_bits(shint, data)) {
 			spin_lock(&shint->lock);
 			used = btrfs_block_group_used(&shint->item);
 			if (used + shint->pinned + shint->reserved <
@@ -366,7 +366,7 @@ __btrfs_find_block_group(struct btrfs_root *root,
 			spin_unlock(&shint->lock);
 		}
 	}
-	if (hint && !hint->ro && block_group_bits(hint, data)) {
+	if (hint && block_group_bits(hint, data)) {
 		spin_lock(&hint->lock);
 		used = btrfs_block_group_used(&hint->item);
 		if (used + hint->pinned + hint->reserved <
@@ -392,7 +392,7 @@ __btrfs_find_block_group(struct btrfs_root *root,
 		last = cache->key.objectid + cache->key.offset;
 		used = btrfs_block_group_used(&cache->item);
 
-		if (!cache->ro && block_group_bits(cache, data)) {
+		if (block_group_bits(cache, data)) {
 			free_check = div_factor(cache->key.offset, factor);
 			if (used + cache->pinned + cache->reserved <
 			    free_check) {
@@ -1843,9 +1843,9 @@ static void set_block_group_readonly(struct btrfs_block_group_cache *cache)
 	spin_unlock(&cache->space_info->lock);
 }
 
-static u64 reduce_alloc_profile(struct btrfs_root *root, u64 flags)
+u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
 {
-	u64 num_devices = root->fs_info->fs_devices->num_devices;
+	u64 num_devices = root->fs_info->fs_devices->rw_devices;
 
 	if (num_devices == 1)
 		flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0);
@@ -1877,13 +1877,11 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_space_info *space_info;
 	u64 thresh;
-	u64 start;
-	u64 num_bytes;
 	int ret = 0;
 
 	mutex_lock(&extent_root->fs_info->chunk_mutex);
 
-	flags = reduce_alloc_profile(extent_root, flags);
+	flags = btrfs_reduce_alloc_profile(extent_root, flags);
 
 	space_info = __find_space_info(extent_root->fs_info, flags);
 	if (!space_info) {
@@ -1913,16 +1911,11 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	}
 	spin_unlock(&space_info->lock);
 
-	ret = btrfs_alloc_chunk(trans, extent_root, &start, &num_bytes, flags);
+	ret = btrfs_alloc_chunk(trans, extent_root, flags);
 	if (ret) {
 printk("space info full %Lu\n", flags);
 		space_info->full = 1;
-		goto out;
 	}
-
-	ret = btrfs_make_block_group(trans, extent_root, 0, flags,
-		     BTRFS_FIRST_CHUNK_TREE_OBJECTID, start, num_bytes);
-	BUG_ON(ret);
 out:
 	mutex_unlock(&extent_root->fs_info->chunk_mutex);
 	return ret;
@@ -3040,7 +3033,7 @@ static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 		data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
 	}
 again:
-	data = reduce_alloc_profile(root, data);
+	data = btrfs_reduce_alloc_profile(root, data);
 	/*
 	 * the only place that sets empty_size is btrfs_realloc_node, which
 	 * is not called recursively on allocations
@@ -5136,7 +5129,8 @@ static int noinline relocate_one_path(struct btrfs_trans_handle *trans,
 			else
 				btrfs_node_key_to_cpu(eb, &keys[level], 0);
 		}
-		if (ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
+		if (nodes[0] &&
+		    ref_path->owner_objectid >= BTRFS_FIRST_FREE_OBJECTID) {
 			eb = path->nodes[0];
 			ret = replace_extents_in_leaf(trans, reloc_root, eb,
 						      group, reloc_inode);
@@ -5377,7 +5371,7 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 	u64 stripped = BTRFS_BLOCK_GROUP_RAID0 |
 		BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
 
-	num_devices = root->fs_info->fs_devices->num_devices;
+	num_devices = root->fs_info->fs_devices->rw_devices;
 	if (num_devices == 1) {
 		stripped |= BTRFS_BLOCK_GROUP_DUP;
 		stripped = flags & ~stripped;
@@ -5801,6 +5795,8 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		BUG_ON(ret);
 
 		set_avail_alloc_bits(root->fs_info, cache->flags);
+		if (btrfs_chunk_readonly(root, cache->key.objectid))
+			set_block_group_readonly(cache);
 	}
 	ret = 0;
 error:
@@ -5889,6 +5885,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	block_group->space_info->total_bytes -= block_group->key.offset;
 	block_group->space_info->bytes_readonly -= block_group->key.offset;
 	spin_unlock(&block_group->space_info->lock);
+	block_group->space_info->full = 0;
 
 	/*
 	memset(shrink_block_group, 0, sizeof(*shrink_block_group));

fs/btrfs/ioctl.c

@@ -405,7 +405,7 @@ static int btrfs_ioctl_resize(struct btrfs_root *root, void __user *arg)
 		devid = simple_strtoull(devstr, &end, 10);
 		printk(KERN_INFO "resizing devid %llu\n", devid);
 	}
-	device = btrfs_find_device(root, devid, NULL);
+	device = btrfs_find_device(root, devid, NULL, NULL);
 	if (!device) {
 		printk(KERN_INFO "resizer unable to find device %llu\n", devid);
 		ret = -EINVAL;

fs/btrfs/super.c

@@ -431,6 +431,11 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
 	if (error)
 		goto error_free_subvol_name;
 
+	if (!(flags & MS_RDONLY) && fs_devices->rw_devices == 0) {
+		error = -EACCES;
+		goto error_close_devices;
+	}
+
 	bdev = fs_devices->latest_bdev;
 	s = sget(fs_type, btrfs_test_super, set_anon_super, fs_devices);
 	if (IS_ERR(s))
@@ -444,6 +449,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags,
 			goto error_close_devices;
 		}
 
+		btrfs_close_devices(fs_devices);
 	} else {
 		char b[BDEVNAME_SIZE];
 
@@ -512,6 +518,9 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 		ret =  btrfs_commit_super(root);
 		WARN_ON(ret);
 	} else {
+		if (root->fs_info->fs_devices->rw_devices == 0)
+			return -EACCES;
+
 		if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
 			return -EINVAL;
 

fs/btrfs/volumes.h

@@ -26,6 +26,7 @@ struct buffer_head;
 struct btrfs_device {
 	struct list_head dev_list;
 	struct list_head dev_alloc_list;
+	struct btrfs_fs_devices *fs_devices;
 	struct btrfs_root *dev_root;
 	struct buffer_head *pending_io;
 	struct bio *pending_bios;
@@ -34,6 +35,7 @@ struct btrfs_device {
 	u64 generation;
 
 	int barriers;
+	int writeable;
 	int in_fs_metadata;
 
 	spinlock_t io_lock;
@@ -77,6 +79,8 @@ struct btrfs_fs_devices {
 	u64 latest_trans;
 	u64 num_devices;
 	u64 open_devices;
+	u64 rw_devices;
+	u64 total_rw_bytes;
 	struct block_device *latest_bdev;
 	/* all of the devices in the FS */
 	struct list_head devices;
@@ -84,7 +88,12 @@ struct btrfs_fs_devices {
 	/* devices not currently being allocated */
 	struct list_head alloc_list;
 	struct list_head list;
-	int mounted;
+
+	struct btrfs_fs_devices *seed;
+	int seeding;
+	int sprouted;
+
+	int opened;
 };
 
 struct btrfs_bio_stripe {
@@ -109,16 +118,14 @@ struct btrfs_multi_bio {
 int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
 			   struct btrfs_device *device,
 			   u64 chunk_tree, u64 chunk_objectid,
-			   u64 chunk_offset,
-			   u64 num_bytes, u64 *start);
+			   u64 chunk_offset, u64 start, u64 num_bytes);
 int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 		    u64 logical, u64 *length,
 		    struct btrfs_multi_bio **multi_ret, int mirror_num);
 int btrfs_read_sys_array(struct btrfs_root *root);
 int btrfs_read_chunk_tree(struct btrfs_root *root);
 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
-		      struct btrfs_root *extent_root, u64 *start,
-		      u64 *num_bytes, u64 type);
+		      struct btrfs_root *extent_root, u64 type);
 void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
@@ -141,10 +148,11 @@ int btrfs_unplug_page(struct btrfs_mapping_tree *map_tree,
 int btrfs_grow_device(struct btrfs_trans_handle *trans,
 		      struct btrfs_device *device, u64 new_size);
 struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid,
-				       u8 *uuid);
+				       u8 *uuid, u8 *fsid);
 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
 int btrfs_init_new_device(struct btrfs_root *root, char *path);
 int btrfs_balance(struct btrfs_root *dev_root);
 void btrfs_unlock_volumes(void);
 void btrfs_lock_volumes(void);
+int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
 #endif