extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \
export.o tree-log.o free-space-cache.o zlib.o lzo.o \
- compression.o delayed-ref.o relocation.o delayed-inode.o backref.o \
- scrub.o
+ compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o \
- reada.o
++ reada.o backref.o
btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o
static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
int verify)
{
- u16 csum_size =
- btrfs_super_csum_size(&root->fs_info->super_copy);
+ u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
char *result = NULL;
unsigned long len;
unsigned long cur_len;
clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
while (1) {
- ret = read_extent_buffer_pages(io_tree, eb, start, 1,
+ ret = read_extent_buffer_pages(io_tree, eb, start,
+ WAIT_COMPLETE,
btree_get_extent, mirror_num);
if (!ret &&
!verify_parent_transid(io_tree, eb, parent_transid))
end = min_t(u64, eb->len, PAGE_CACHE_SIZE);
end = eb->start + end - 1;
err:
+ if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
+ clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
+ btree_readahead_hook(root, eb, eb->start, ret);
+ }
+
free_extent_buffer(eb);
out:
return ret;
}
- struct extent_state *state)
+static int btree_io_failed_hook(struct bio *failed_bio,
+ struct page *page, u64 start, u64 end,
++ u64 mirror_num, struct extent_state *state)
+{
+ struct extent_io_tree *tree;
+ unsigned long len;
+ struct extent_buffer *eb;
+ struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
+
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
+ if (page->private == EXTENT_PAGE_PRIVATE)
+ goto out;
+ if (!page->private)
+ goto out;
+
+ len = page->private >> 2;
+ WARN_ON(len == 0);
+
+ eb = alloc_extent_buffer(tree, start, len, page);
+ if (eb == NULL)
+ goto out;
+
+ if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
+ clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
+ btree_readahead_hook(root, eb, eb->start, -EIO);
+ }
+
+out:
+ return -EIO; /* we fixed nothing */
+}
+
static void end_workqueue_bio(struct bio *bio, int err)
{
struct end_io_wq *end_io_wq = bio->bi_private;
{
struct extent_io_tree *tree;
tree = &BTRFS_I(page->mapping->host)->io_tree;
- return extent_read_full_page(tree, page, btree_get_extent);
+ return extent_read_full_page(tree, page, btree_get_extent, 0);
}
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
if (!buf)
return 0;
read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
- buf, 0, 0, btree_get_extent, 0);
+ buf, 0, WAIT_NONE, btree_get_extent, 0);
free_extent_buffer(buf);
return ret;
}
+int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize,
+ int mirror_num, struct extent_buffer **eb)
+{
+ struct extent_buffer *buf = NULL;
+ struct inode *btree_inode = root->fs_info->btree_inode;
+ struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
+ int ret;
+
+ buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
+ if (!buf)
+ return 0;
+
+ set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);
+
+ ret = read_extent_buffer_pages(io_tree, buf, 0, WAIT_PAGE_LOCK,
+ btree_get_extent, mirror_num);
+ if (ret) {
+ free_extent_buffer(buf);
+ return ret;
+ }
+
+ if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
+ free_extent_buffer(buf);
+ return -EIO;
+ } else if (extent_buffer_uptodate(io_tree, buf, NULL)) {
+ *eb = buf;
+ } else {
+ free_extent_buffer(buf);
+ }
+ return 0;
+}
+
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
u64 bytenr, u32 blocksize)
{
generation = btrfs_root_generation(&root->root_item);
blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
+ root->commit_root = NULL;
root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
blocksize, generation);
if (!root->node || !btrfs_buffer_uptodate(root->node, generation)) {
free_extent_buffer(root->node);
+ root->node = NULL;
return -EIO;
}
root->commit_root = btrfs_root_node(root);
return 0;
}
+/*
+ * this will find the highest generation in the array of
+ * root backups. The index of the highest array is returned,
+ * or -1 if we can't find anything.
+ *
+ * We check to make sure the array is valid by comparing the
+ * generation of the latest root in the array with the generation
+ * in the super block. If they don't match we pitch it.
+ */
+static int find_newest_super_backup(struct btrfs_fs_info *info, u64 newest_gen)
+{
+ u64 cur;
+ int newest_index = -1;
+ struct btrfs_root_backup *root_backup;
+ int i;
+
+ for (i = 0; i < BTRFS_NUM_BACKUP_ROOTS; i++) {
+ root_backup = info->super_copy->super_roots + i;
+ cur = btrfs_backup_tree_root_gen(root_backup);
+ if (cur == newest_gen)
+ newest_index = i;
+ }
+
+ /* check to see if we actually wrapped around */
+ if (newest_index == BTRFS_NUM_BACKUP_ROOTS - 1) {
+ root_backup = info->super_copy->super_roots;
+ cur = btrfs_backup_tree_root_gen(root_backup);
+ if (cur == newest_gen)
+ newest_index = 0;
+ }
+ return newest_index;
+}
+
+
+/*
+ * find the oldest backup so we know where to store new entries
+ * in the backup array. This will set the backup_root_index
+ * field in the fs_info struct
+ */
+static void find_oldest_super_backup(struct btrfs_fs_info *info,
+ u64 newest_gen)
+{
+ int newest_index = -1;
+
+ newest_index = find_newest_super_backup(info, newest_gen);
+ /* if there was garbage in there, just move along */
+ if (newest_index == -1) {
+ info->backup_root_index = 0;
+ } else {
+ info->backup_root_index = (newest_index + 1) % BTRFS_NUM_BACKUP_ROOTS;
+ }
+}
+
+/*
+ * copy all the root pointers into the super backup array.
+ * this will bump the backup pointer by one when it is
+ * done
+ */
+static void backup_super_roots(struct btrfs_fs_info *info)
+{
+ int next_backup;
+ struct btrfs_root_backup *root_backup;
+ int last_backup;
+
+ next_backup = info->backup_root_index;
+ last_backup = (next_backup + BTRFS_NUM_BACKUP_ROOTS - 1) %
+ BTRFS_NUM_BACKUP_ROOTS;
+
+ /*
+ * just overwrite the last backup if we're at the same generation
+ * this happens only at umount
+ */
+ root_backup = info->super_for_commit->super_roots + last_backup;
+ if (btrfs_backup_tree_root_gen(root_backup) ==
+ btrfs_header_generation(info->tree_root->node))
+ next_backup = last_backup;
+
+ root_backup = info->super_for_commit->super_roots + next_backup;
+
+ /*
+ * make sure all of our padding and empty slots get zero filled
+ * regardless of which ones we use today
+ */
+ memset(root_backup, 0, sizeof(*root_backup));
+
+ info->backup_root_index = (next_backup + 1) % BTRFS_NUM_BACKUP_ROOTS;
+
+ btrfs_set_backup_tree_root(root_backup, info->tree_root->node->start);
+ btrfs_set_backup_tree_root_gen(root_backup,
+ btrfs_header_generation(info->tree_root->node));
+
+ btrfs_set_backup_tree_root_level(root_backup,
+ btrfs_header_level(info->tree_root->node));
+
+ btrfs_set_backup_chunk_root(root_backup, info->chunk_root->node->start);
+ btrfs_set_backup_chunk_root_gen(root_backup,
+ btrfs_header_generation(info->chunk_root->node));
+ btrfs_set_backup_chunk_root_level(root_backup,
+ btrfs_header_level(info->chunk_root->node));
+
+ btrfs_set_backup_extent_root(root_backup, info->extent_root->node->start);
+ btrfs_set_backup_extent_root_gen(root_backup,
+ btrfs_header_generation(info->extent_root->node));
+ btrfs_set_backup_extent_root_level(root_backup,
+ btrfs_header_level(info->extent_root->node));
+
+ btrfs_set_backup_fs_root(root_backup, info->fs_root->node->start);
+ btrfs_set_backup_fs_root_gen(root_backup,
+ btrfs_header_generation(info->fs_root->node));
+ btrfs_set_backup_fs_root_level(root_backup,
+ btrfs_header_level(info->fs_root->node));
+
+ btrfs_set_backup_dev_root(root_backup, info->dev_root->node->start);
+ btrfs_set_backup_dev_root_gen(root_backup,
+ btrfs_header_generation(info->dev_root->node));
+ btrfs_set_backup_dev_root_level(root_backup,
+ btrfs_header_level(info->dev_root->node));
+
+ btrfs_set_backup_csum_root(root_backup, info->csum_root->node->start);
+ btrfs_set_backup_csum_root_gen(root_backup,
+ btrfs_header_generation(info->csum_root->node));
+ btrfs_set_backup_csum_root_level(root_backup,
+ btrfs_header_level(info->csum_root->node));
+
+ btrfs_set_backup_total_bytes(root_backup,
+ btrfs_super_total_bytes(info->super_copy));
+ btrfs_set_backup_bytes_used(root_backup,
+ btrfs_super_bytes_used(info->super_copy));
+ btrfs_set_backup_num_devices(root_backup,
+ btrfs_super_num_devices(info->super_copy));
+
+ /*
+ * if we don't copy this out to the super_copy, it won't get remembered
+ * for the next commit
+ */
+ memcpy(&info->super_copy->super_roots,
+ &info->super_for_commit->super_roots,
+ sizeof(*root_backup) * BTRFS_NUM_BACKUP_ROOTS);
+}
+
+/*
+ * this copies info out of the root backup array and back into
+ * the in-memory super block. It is meant to help iterate through
+ * the array, so you send it the number of backups you've already
+ * tried and the last backup index you used.
+ *
+ * this returns -1 when it has tried all the backups
+ */
+static noinline int next_root_backup(struct btrfs_fs_info *info,
+ struct btrfs_super_block *super,
+ int *num_backups_tried, int *backup_index)
+{
+ struct btrfs_root_backup *root_backup;
+ int newest = *backup_index;
+
+ if (*num_backups_tried == 0) {
+ u64 gen = btrfs_super_generation(super);
+
+ newest = find_newest_super_backup(info, gen);
+ if (newest == -1)
+ return -1;
+
+ *backup_index = newest;
+ *num_backups_tried = 1;
+ } else if (*num_backups_tried == BTRFS_NUM_BACKUP_ROOTS) {
+ /* we've tried all the backups, all done */
+ return -1;
+ } else {
+ /* jump to the next oldest backup */
+ newest = (*backup_index + BTRFS_NUM_BACKUP_ROOTS - 1) %
+ BTRFS_NUM_BACKUP_ROOTS;
+ *backup_index = newest;
+ *num_backups_tried += 1;
+ }
+ root_backup = super->super_roots + newest;
+
+ btrfs_set_super_generation(super,
+ btrfs_backup_tree_root_gen(root_backup));
+ btrfs_set_super_root(super, btrfs_backup_tree_root(root_backup));
+ btrfs_set_super_root_level(super,
+ btrfs_backup_tree_root_level(root_backup));
+ btrfs_set_super_bytes_used(super, btrfs_backup_bytes_used(root_backup));
+
+ /*
+ * fixme: the total bytes and num_devices need to match or we should
+ * need a fsck
+ */
+ btrfs_set_super_total_bytes(super, btrfs_backup_total_bytes(root_backup));
+ btrfs_set_super_num_devices(super, btrfs_backup_num_devices(root_backup));
+ return 0;
+}
+
+/* helper to cleanup tree roots */
+static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
+{
+ free_extent_buffer(info->tree_root->node);
+ free_extent_buffer(info->tree_root->commit_root);
+ free_extent_buffer(info->dev_root->node);
+ free_extent_buffer(info->dev_root->commit_root);
+ free_extent_buffer(info->extent_root->node);
+ free_extent_buffer(info->extent_root->commit_root);
+ free_extent_buffer(info->csum_root->node);
+ free_extent_buffer(info->csum_root->commit_root);
+
+ info->tree_root->node = NULL;
+ info->tree_root->commit_root = NULL;
+ info->dev_root->node = NULL;
+ info->dev_root->commit_root = NULL;
+ info->extent_root->node = NULL;
+ info->extent_root->commit_root = NULL;
+ info->csum_root->node = NULL;
+ info->csum_root->commit_root = NULL;
+
+ if (chunk_root) {
+ free_extent_buffer(info->chunk_root->node);
+ free_extent_buffer(info->chunk_root->commit_root);
+ info->chunk_root->node = NULL;
+ info->chunk_root->commit_root = NULL;
+ }
+}
+
+
struct btrfs_root *open_ctree(struct super_block *sb,
struct btrfs_fs_devices *fs_devices,
char *options)
int ret;
int err = -EINVAL;
+ int num_backups_tried = 0;
+ int backup_index = 0;
struct btrfs_super_block *disk_super;
spin_lock_init(&fs_info->fs_roots_radix_lock);
spin_lock_init(&fs_info->delayed_iput_lock);
spin_lock_init(&fs_info->defrag_inodes_lock);
+ spin_lock_init(&fs_info->free_chunk_lock);
mutex_init(&fs_info->reloc_mutex);
init_completion(&fs_info->kobj_unregister);
btrfs_init_block_rsv(&fs_info->trans_block_rsv);
btrfs_init_block_rsv(&fs_info->chunk_block_rsv);
btrfs_init_block_rsv(&fs_info->empty_block_rsv);
- INIT_LIST_HEAD(&fs_info->durable_block_rsv_list);
- mutex_init(&fs_info->durable_block_rsv_mutex);
+ btrfs_init_block_rsv(&fs_info->delayed_block_rsv);
atomic_set(&fs_info->nr_async_submits, 0);
atomic_set(&fs_info->async_delalloc_pages, 0);
atomic_set(&fs_info->async_submit_draining, 0);
fs_info->metadata_ratio = 0;
fs_info->defrag_inodes = RB_ROOT;
fs_info->trans_no_join = 0;
+ fs_info->free_chunk_space = 0;
+
+ /* readahead state */
+ INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT);
+ spin_lock_init(&fs_info->reada_lock);
fs_info->thread_pool_size = min_t(unsigned long,
num_online_cpus() + 2, 8);
goto fail_alloc;
}
- memcpy(&fs_info->super_copy, bh->b_data, sizeof(fs_info->super_copy));
- memcpy(&fs_info->super_for_commit, &fs_info->super_copy,
- sizeof(fs_info->super_for_commit));
+ memcpy(fs_info->super_copy, bh->b_data, sizeof(*fs_info->super_copy));
+ memcpy(fs_info->super_for_commit, fs_info->super_copy,
+ sizeof(*fs_info->super_for_commit));
brelse(bh);
- memcpy(fs_info->fsid, fs_info->super_copy.fsid, BTRFS_FSID_SIZE);
+ memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
- disk_super = &fs_info->super_copy;
+ disk_super = fs_info->super_copy;
if (!btrfs_super_root(disk_super))
goto fail_alloc;
btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
+ /*
+ * run through our array of backup supers and setup
+ * our ring pointer to the oldest one
+ */
+ generation = btrfs_super_generation(disk_super);
+ find_oldest_super_backup(fs_info, generation);
+
/*
* In the long term, we'll store the compression type in the super
* block, and it'll be used for per file compression control.
btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
fs_info->thread_pool_size,
&fs_info->generic_worker);
+ btrfs_init_workers(&fs_info->readahead_workers, "readahead",
+ fs_info->thread_pool_size,
+ &fs_info->generic_worker);
/*
* endios are largely parallel and should have a very
fs_info->endio_write_workers.idle_thresh = 2;
fs_info->endio_meta_write_workers.idle_thresh = 2;
+ fs_info->readahead_workers.idle_thresh = 2;
btrfs_start_workers(&fs_info->workers, 1);
btrfs_start_workers(&fs_info->generic_worker, 1);
btrfs_start_workers(&fs_info->endio_freespace_worker, 1);
btrfs_start_workers(&fs_info->delayed_workers, 1);
btrfs_start_workers(&fs_info->caching_workers, 1);
+ btrfs_start_workers(&fs_info->readahead_workers, 1);
fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
if (!test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
sb->s_id);
- goto fail_chunk_root;
+ goto fail_tree_roots;
}
btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
chunk_root->commit_root = btrfs_root_node(chunk_root);
if (ret) {
printk(KERN_WARNING "btrfs: failed to read chunk tree on %s\n",
sb->s_id);
- goto fail_chunk_root;
+ goto fail_tree_roots;
}
btrfs_close_extra_devices(fs_devices);
+retry_root_backup:
blocksize = btrfs_level_size(tree_root,
btrfs_super_root_level(disk_super));
generation = btrfs_super_generation(disk_super);
tree_root->node = read_tree_block(tree_root,
btrfs_super_root(disk_super),
blocksize, generation);
- if (!tree_root->node)
- goto fail_chunk_root;
- if (!test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
+ if (!tree_root->node ||
+ !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
printk(KERN_WARNING "btrfs: failed to read tree root on %s\n",
sb->s_id);
- goto fail_tree_root;
+
+ goto recovery_tree_root;
}
+
btrfs_set_root_node(&tree_root->root_item, tree_root->node);
tree_root->commit_root = btrfs_root_node(tree_root);
ret = find_and_setup_root(tree_root, fs_info,
BTRFS_EXTENT_TREE_OBJECTID, extent_root);
if (ret)
- goto fail_tree_root;
+ goto recovery_tree_root;
extent_root->track_dirty = 1;
ret = find_and_setup_root(tree_root, fs_info,
BTRFS_DEV_TREE_OBJECTID, dev_root);
if (ret)
- goto fail_extent_root;
+ goto recovery_tree_root;
dev_root->track_dirty = 1;
ret = find_and_setup_root(tree_root, fs_info,
BTRFS_CSUM_TREE_OBJECTID, csum_root);
if (ret)
- goto fail_dev_root;
+ goto recovery_tree_root;
csum_root->track_dirty = 1;
fail_block_groups:
btrfs_free_block_groups(fs_info);
- free_extent_buffer(csum_root->node);
- free_extent_buffer(csum_root->commit_root);
-fail_dev_root:
- free_extent_buffer(dev_root->node);
- free_extent_buffer(dev_root->commit_root);
-fail_extent_root:
- free_extent_buffer(extent_root->node);
- free_extent_buffer(extent_root->commit_root);
-fail_tree_root:
- free_extent_buffer(tree_root->node);
- free_extent_buffer(tree_root->commit_root);
-fail_chunk_root:
- free_extent_buffer(chunk_root->node);
- free_extent_buffer(chunk_root->commit_root);
+
+fail_tree_roots:
+ free_root_pointers(fs_info, 1);
+
fail_sb_buffer:
btrfs_stop_workers(&fs_info->generic_worker);
btrfs_stop_workers(&fs_info->fixup_workers);
btrfs_stop_workers(&fs_info->delayed_workers);
btrfs_stop_workers(&fs_info->caching_workers);
fail_alloc:
- kfree(fs_info->delayed_root);
fail_iput:
invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
iput(fs_info->btree_inode);
fail_srcu:
cleanup_srcu_struct(&fs_info->subvol_srcu);
fail:
- kfree(extent_root);
- kfree(tree_root);
- kfree(fs_info);
- kfree(chunk_root);
- kfree(dev_root);
- kfree(csum_root);
+ free_fs_info(fs_info);
return ERR_PTR(err);
+
+recovery_tree_root:
+
+ if (!btrfs_test_opt(tree_root, RECOVERY))
+ goto fail_tree_roots;
+
+ free_root_pointers(fs_info, 0);
+
+ /* don't use the log in recovery mode, it won't be valid */
+ btrfs_set_super_log_root(disk_super, 0);
+
+ /* we can't trust the free space cache either */
+ btrfs_set_opt(fs_info->mount_opt, CLEAR_CACHE);
+
+ ret = next_root_backup(fs_info, fs_info->super_copy,
+ &num_backups_tried, &backup_index);
+ if (ret == -1)
+ goto fail_block_groups;
+ goto retry_root_backup;
}
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
int total_errors = 0;
u64 flags;
- max_errors = btrfs_super_num_devices(&root->fs_info->super_copy) - 1;
+ max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
do_barriers = !btrfs_test_opt(root, NOBARRIER);
+ backup_super_roots(root->fs_info);
- sb = &root->fs_info->super_for_commit;
+ sb = root->fs_info->super_for_commit;
dev_item = &sb->dev_item;
mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
/* clear out the rbtree of defraggable inodes */
btrfs_run_defrag_inodes(root->fs_info);
- btrfs_put_block_group_cache(fs_info);
-
/*
* Here come 2 situations when btrfs is broken to flip readonly:
*
printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
}
+ btrfs_put_block_group_cache(fs_info);
+
kthread_stop(root->fs_info->transaction_kthread);
kthread_stop(root->fs_info->cleaner_kthread);
del_fs_roots(fs_info);
iput(fs_info->btree_inode);
- kfree(fs_info->delayed_root);
btrfs_stop_workers(&fs_info->generic_worker);
btrfs_stop_workers(&fs_info->fixup_workers);
btrfs_stop_workers(&fs_info->submit_workers);
btrfs_stop_workers(&fs_info->delayed_workers);
btrfs_stop_workers(&fs_info->caching_workers);
+ btrfs_stop_workers(&fs_info->readahead_workers);
btrfs_close_devices(fs_info->fs_devices);
btrfs_mapping_tree_free(&fs_info->mapping_tree);
bdi_destroy(&fs_info->bdi);
cleanup_srcu_struct(&fs_info->subvol_srcu);
- kfree(fs_info->extent_root);
- kfree(fs_info->tree_root);
- kfree(fs_info->chunk_root);
- kfree(fs_info->dev_root);
- kfree(fs_info->csum_root);
- kfree(fs_info);
+ free_fs_info(fs_info);
return 0;
}
return ret;
}
-int btree_lock_page_hook(struct page *page)
+static int btree_lock_page_hook(struct page *page, void *data,
+ void (*flush_fn)(void *))
{
struct inode *inode = page->mapping->host;
struct btrfs_root *root = BTRFS_I(inode)->root;
if (!eb)
goto out;
- btrfs_tree_lock(eb);
+ if (!btrfs_try_tree_write_lock(eb)) {
+ flush_fn(data);
+ btrfs_tree_lock(eb);
+ }
btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
btrfs_tree_unlock(eb);
free_extent_buffer(eb);
out:
- lock_page(page);
+ if (!trylock_page(page)) {
+ flush_fn(data);
+ lock_page(page);
+ }
return 0;
}
static struct extent_io_ops btree_extent_io_ops = {
.write_cache_pages_lock_hook = btree_lock_page_hook,
.readpage_end_io_hook = btree_readpage_end_io_hook,
+ .readpage_io_failed_hook = btree_io_failed_hook,
.submit_bio_hook = btree_submit_bio_hook,
/* note we're sharing with inode.c for the merge bio hook */
.merge_bio_hook = btrfs_merge_bio_hook,
#include <linux/rcupdate.h>
#include <linux/kthread.h>
#include <linux/slab.h>
+#include <linux/ratelimit.h>
#include "compat.h"
#include "hash.h"
#include "ctree.h"
CHUNK_ALLOC_LIMITED = 2,
};
+/*
+ * Control how reservations are dealt with.
+ *
+ * RESERVE_FREE - freeing a reservation.
+ * RESERVE_ALLOC - allocating space and we need to update bytes_may_use for
+ * ENOSPC accounting
+ * RESERVE_ALLOC_NO_ACCOUNT - allocating space and we should not update
+ * bytes_may_use as the ENOSPC accounting is done elsewhere
+ */
+enum {
+ RESERVE_FREE = 0,
+ RESERVE_ALLOC = 1,
+ RESERVE_ALLOC_NO_ACCOUNT = 2,
+};
+
static int update_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc);
struct btrfs_key *key);
static void dump_space_info(struct btrfs_space_info *info, u64 bytes,
int dump_block_groups);
+static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
+ u64 num_bytes, int reserve);
static noinline int
block_group_cache_done(struct btrfs_block_group_cache *cache)
if (atomic_dec_and_test(&cache->count)) {
WARN_ON(cache->pinned > 0);
WARN_ON(cache->reserved > 0);
- WARN_ON(cache->reserved_pinned > 0);
kfree(cache->free_space_ctl);
kfree(cache);
}
* we likely hold important locks.
*/
if (trans && (!trans->transaction->in_commit) &&
- (root && root != root->fs_info->tree_root)) {
+ (root && root != root->fs_info->tree_root) &&
+ btrfs_test_opt(root, SPACE_CACHE)) {
spin_lock(&cache->lock);
if (cache->cached != BTRFS_CACHE_NO) {
spin_unlock(&cache->lock);
{
int ret;
u64 discarded_bytes = 0;
- struct btrfs_multi_bio *multi = NULL;
+ struct btrfs_bio *bbio = NULL;
/* Tell the block device(s) that the sectors can be discarded */
ret = btrfs_map_block(&root->fs_info->mapping_tree, REQ_DISCARD,
- bytenr, &num_bytes, &multi, 0);
+ bytenr, &num_bytes, &bbio, 0);
if (!ret) {
- struct btrfs_bio_stripe *stripe = multi->stripes;
+ struct btrfs_bio_stripe *stripe = bbio->stripes;
int i;
- for (i = 0; i < multi->num_stripes; i++, stripe++) {
+ for (i = 0; i < bbio->num_stripes; i++, stripe++) {
if (!stripe->dev->can_discard)
continue;
*/
ret = 0;
}
- kfree(multi);
+ kfree(bbio);
}
if (actual_bytes)
goto again;
}
+ /* We've already setup this transaction, go ahead and exit */
+ if (block_group->cache_generation == trans->transid &&
+ i_size_read(inode)) {
+ dcs = BTRFS_DC_SETUP;
+ goto out_put;
+ }
+
/*
* We want to set the generation to 0, that way if anything goes wrong
* from here on out we know not to trust this cache when we load up next
if (!ret)
dcs = BTRFS_DC_SETUP;
btrfs_free_reserved_data_space(inode, num_pages);
+
out_put:
iput(inode);
out_free:
btrfs_release_path(path);
out:
spin_lock(&block_group->lock);
+ if (!ret)
+ block_group->cache_generation = trans->transid;
block_group->disk_cache_state = dcs;
spin_unlock(&block_group->lock);
return -ENOSPC;
}
data_sinfo->bytes_may_use += bytes;
- BTRFS_I(inode)->reserved_bytes += bytes;
spin_unlock(&data_sinfo->lock);
return 0;
}
/*
- * called when we are clearing an delalloc extent from the
- * inode's io_tree or there was an error for whatever reason
- * after calling btrfs_check_data_free_space
+ * Called if we need to clear a data reservation for this inode.
*/
void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes)
{
data_sinfo = BTRFS_I(inode)->space_info;
spin_lock(&data_sinfo->lock);
data_sinfo->bytes_may_use -= bytes;
- BTRFS_I(inode)->reserved_bytes -= bytes;
spin_unlock(&data_sinfo->lock);
}
struct btrfs_space_info *sinfo, u64 alloc_bytes,
int force)
{
+ struct btrfs_block_rsv *global_rsv = &root->fs_info->global_block_rsv;
u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly;
u64 num_allocated = sinfo->bytes_used + sinfo->bytes_reserved;
u64 thresh;
if (force == CHUNK_ALLOC_FORCE)
return 1;
+ /*
+ * We need to take into account the global rsv because for all intents
+ * and purposes it's used space. Don't worry about locking the
+ * global_rsv, it doesn't change except when the transaction commits.
+ */
+ num_allocated += global_rsv->size;
+
/*
* in limited mode, we want to have some free space up to
* about 1% of the FS size.
*/
if (force == CHUNK_ALLOC_LIMITED) {
- thresh = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ thresh = btrfs_super_total_bytes(root->fs_info->super_copy);
thresh = max_t(u64, 64 * 1024 * 1024,
div_factor_fine(thresh, 1));
if (num_allocated + alloc_bytes < div_factor(num_bytes, 8))
return 0;
- thresh = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ thresh = btrfs_super_total_bytes(root->fs_info->super_copy);
/* 256MB or 5% of the FS */
thresh = max_t(u64, 256 * 1024 * 1024, div_factor_fine(thresh, 5));
/*
* shrink metadata reservation for delalloc
*/
-static int shrink_delalloc(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 to_reclaim, int sync)
+static int shrink_delalloc(struct btrfs_root *root, u64 to_reclaim,
+ bool wait_ordered)
{
struct btrfs_block_rsv *block_rsv;
struct btrfs_space_info *space_info;
+ struct btrfs_trans_handle *trans;
u64 reserved;
u64 max_reclaim;
u64 reclaimed = 0;
long time_left;
- int nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT;
+ unsigned long nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT;
int loops = 0;
unsigned long progress;
+ trans = (struct btrfs_trans_handle *)current->journal_info;
block_rsv = &root->fs_info->delalloc_block_rsv;
space_info = block_rsv->space_info;
smp_mb();
- reserved = space_info->bytes_reserved;
+ reserved = space_info->bytes_may_use;
progress = space_info->reservation_progress;
if (reserved == 0)
}
max_reclaim = min(reserved, to_reclaim);
-
+ nr_pages = max_t(unsigned long, nr_pages,
+ max_reclaim >> PAGE_CACHE_SHIFT);
while (loops < 1024) {
/* have the flusher threads jump in and do some IO */
smp_mb();
writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages);
spin_lock(&space_info->lock);
- if (reserved > space_info->bytes_reserved)
- reclaimed += reserved - space_info->bytes_reserved;
- reserved = space_info->bytes_reserved;
+ if (reserved > space_info->bytes_may_use)
+ reclaimed += reserved - space_info->bytes_may_use;
+ reserved = space_info->bytes_may_use;
spin_unlock(&space_info->lock);
loops++;
if (trans && trans->transaction->blocked)
return -EAGAIN;
- time_left = schedule_timeout_interruptible(1);
+ if (wait_ordered && !trans) {
+ btrfs_wait_ordered_extents(root, 0, 0);
+ } else {
+ time_left = schedule_timeout_interruptible(1);
- /* We were interrupted, exit */
- if (time_left)
- break;
+ /* We were interrupted, exit */
+ if (time_left)
+ break;
+ }
/* we've kicked the IO a few times, if anything has been freed,
* exit. There is no sense in looping here for a long time
}
}
- if (reclaimed >= to_reclaim && !trans)
- btrfs_wait_ordered_extents(root, 0, 0);
+
return reclaimed >= to_reclaim;
}
-/*
- * Retries tells us how many times we've called reserve_metadata_bytes. The
- * idea is if this is the first call (retries == 0) then we will add to our
- * reserved count if we can't make the allocation in order to hold our place
- * while we go and try and free up space. That way for retries > 1 we don't try
- * and add space, we just check to see if the amount of unused space is >= the
- * total space, meaning that our reservation is valid.
+/**
+ * maybe_commit_transaction - possibly commit the transaction if its ok to
+ * @root - the root we're allocating for
+ * @bytes - the number of bytes we want to reserve
+ * @force - force the commit
*
- * However if we don't intend to retry this reservation, pass -1 as retries so
- * that it short circuits this logic.
+ * This will check to make sure that committing the transaction will actually
+ * get us somewhere and then commit the transaction if it does. Otherwise it
+ * will return -ENOSPC.
*/
-static int reserve_metadata_bytes(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
+static int may_commit_transaction(struct btrfs_root *root,
+ struct btrfs_space_info *space_info,
+ u64 bytes, int force)
+{
+ struct btrfs_block_rsv *delayed_rsv = &root->fs_info->delayed_block_rsv;
+ struct btrfs_trans_handle *trans;
+
+ trans = (struct btrfs_trans_handle *)current->journal_info;
+ if (trans)
+ return -EAGAIN;
+
+ if (force)
+ goto commit;
+
+ /* See if there is enough pinned space to make this reservation */
+ spin_lock(&space_info->lock);
+ if (space_info->bytes_pinned >= bytes) {
+ spin_unlock(&space_info->lock);
+ goto commit;
+ }
+ spin_unlock(&space_info->lock);
+
+ /*
+ * See if there is some space in the delayed insertion reservation for
+ * this reservation.
+ */
+ if (space_info != delayed_rsv->space_info)
+ return -ENOSPC;
+
+ spin_lock(&delayed_rsv->lock);
+ if (delayed_rsv->size < bytes) {
+ spin_unlock(&delayed_rsv->lock);
+ return -ENOSPC;
+ }
+ spin_unlock(&delayed_rsv->lock);
+
+commit:
+ trans = btrfs_join_transaction(root);
+ if (IS_ERR(trans))
+ return -ENOSPC;
+
+ return btrfs_commit_transaction(trans, root);
+}
+
+/**
+ * reserve_metadata_bytes - try to reserve bytes from the block_rsv's space
+ * @root - the root we're allocating for
+ * @block_rsv - the block_rsv we're allocating for
+ * @orig_bytes - the number of bytes we want
+ * @flush - wether or not we can flush to make our reservation
+ *
+ * This will reserve orgi_bytes number of bytes from the space info associated
+ * with the block_rsv. If there is not enough space it will make an attempt to
+ * flush out space to make room. It will do this by flushing delalloc if
+ * possible or committing the transaction. If flush is 0 then no attempts to
+ * regain reservations will be made and this will fail if there is not enough
+ * space already.
+ */
+static int reserve_metadata_bytes(struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv,
u64 orig_bytes, int flush)
{
struct btrfs_space_info *space_info = block_rsv->space_info;
- u64 unused;
+ u64 used;
u64 num_bytes = orig_bytes;
int retries = 0;
int ret = 0;
bool committed = false;
bool flushing = false;
+ bool wait_ordered = false;
again:
ret = 0;
* deadlock since we are waiting for the flusher to finish, but
* hold the current transaction open.
*/
- if (trans)
+ if (current->journal_info)
return -EAGAIN;
ret = wait_event_interruptible(space_info->wait,
!space_info->flush);
}
ret = -ENOSPC;
- unused = space_info->bytes_used + space_info->bytes_reserved +
- space_info->bytes_pinned + space_info->bytes_readonly +
- space_info->bytes_may_use;
+ used = space_info->bytes_used + space_info->bytes_reserved +
+ space_info->bytes_pinned + space_info->bytes_readonly +
+ space_info->bytes_may_use;
/*
* The idea here is that we've not already over-reserved the block group
* lets start flushing stuff first and then come back and try to make
* our reservation.
*/
- if (unused <= space_info->total_bytes) {
- unused = space_info->total_bytes - unused;
- if (unused >= num_bytes) {
- space_info->bytes_reserved += orig_bytes;
+ if (used <= space_info->total_bytes) {
+ if (used + orig_bytes <= space_info->total_bytes) {
+ space_info->bytes_may_use += orig_bytes;
ret = 0;
} else {
/*
* amount plus the amount of bytes that we need for this
* reservation.
*/
- num_bytes = unused - space_info->total_bytes +
+ wait_ordered = true;
+ num_bytes = used - space_info->total_bytes +
(orig_bytes * (retries + 1));
}
+ if (ret) {
+ u64 profile = btrfs_get_alloc_profile(root, 0);
+ u64 avail;
+
+ /*
+ * If we have a lot of space that's pinned, don't bother doing
+ * the overcommit dance yet and just commit the transaction.
+ */
+ avail = (space_info->total_bytes - space_info->bytes_used) * 8;
+ do_div(avail, 10);
+ if (space_info->bytes_pinned >= avail && flush && !committed) {
+ space_info->flush = 1;
+ flushing = true;
+ spin_unlock(&space_info->lock);
+ ret = may_commit_transaction(root, space_info,
+ orig_bytes, 1);
+ if (ret)
+ goto out;
+ committed = true;
+ goto again;
+ }
+
+ spin_lock(&root->fs_info->free_chunk_lock);
+ avail = root->fs_info->free_chunk_space;
+
+ /*
+ * If we have dup, raid1 or raid10 then only half of the free
+ * space is actually useable.
+ */
+ if (profile & (BTRFS_BLOCK_GROUP_DUP |
+ BTRFS_BLOCK_GROUP_RAID1 |
+ BTRFS_BLOCK_GROUP_RAID10))
+ avail >>= 1;
+
+ /*
+ * If we aren't flushing don't let us overcommit too much, say
+ * 1/8th of the space. If we can flush, let it overcommit up to
+ * 1/2 of the space.
+ */
+ if (flush)
+ avail >>= 3;
+ else
+ avail >>= 1;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+
+ if (used + num_bytes < space_info->total_bytes + avail) {
+ space_info->bytes_may_use += orig_bytes;
+ ret = 0;
+ } else {
+ wait_ordered = true;
+ }
+ }
+
/*
* Couldn't make our reservation, save our place so while we're trying
* to reclaim space we can actually use it instead of somebody else
* We do synchronous shrinking since we don't actually unreserve
* metadata until after the IO is completed.
*/
- ret = shrink_delalloc(trans, root, num_bytes, 1);
+ ret = shrink_delalloc(root, num_bytes, wait_ordered);
if (ret < 0)
goto out;
* so go back around and try again.
*/
if (retries < 2) {
+ wait_ordered = true;
retries++;
goto again;
}
- /*
- * Not enough space to be reclaimed, don't bother committing the
- * transaction.
- */
- spin_lock(&space_info->lock);
- if (space_info->bytes_pinned < orig_bytes)
- ret = -ENOSPC;
- spin_unlock(&space_info->lock);
- if (ret)
- goto out;
-
- ret = -EAGAIN;
- if (trans)
- goto out;
-
ret = -ENOSPC;
if (committed)
goto out;
- trans = btrfs_join_transaction(root);
- if (IS_ERR(trans))
- goto out;
- ret = btrfs_commit_transaction(trans, root);
+ ret = may_commit_transaction(root, space_info, orig_bytes, 0);
if (!ret) {
- trans = NULL;
committed = true;
goto again;
}
static struct btrfs_block_rsv *get_block_rsv(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
- struct btrfs_block_rsv *block_rsv;
- if (root->ref_cows)
+ struct btrfs_block_rsv *block_rsv = NULL;
+
+ if (root->ref_cows || root == root->fs_info->csum_root)
block_rsv = trans->block_rsv;
- else
+
+ if (!block_rsv)
block_rsv = root->block_rsv;
if (!block_rsv)
}
if (num_bytes) {
spin_lock(&space_info->lock);
- space_info->bytes_reserved -= num_bytes;
+ space_info->bytes_may_use -= num_bytes;
space_info->reservation_progress++;
spin_unlock(&space_info->lock);
}
{
memset(rsv, 0, sizeof(*rsv));
spin_lock_init(&rsv->lock);
- atomic_set(&rsv->usage, 1);
- rsv->priority = 6;
- INIT_LIST_HEAD(&rsv->list);
}
struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_root *root)
void btrfs_free_block_rsv(struct btrfs_root *root,
struct btrfs_block_rsv *rsv)
{
- if (rsv && atomic_dec_and_test(&rsv->usage)) {
- btrfs_block_rsv_release(root, rsv, (u64)-1);
- if (!rsv->durable)
- kfree(rsv);
- }
+ btrfs_block_rsv_release(root, rsv, (u64)-1);
+ kfree(rsv);
}
-/*
- * make the block_rsv struct be able to capture freed space.
- * the captured space will re-add to the the block_rsv struct
- * after transaction commit
- */
-void btrfs_add_durable_block_rsv(struct btrfs_fs_info *fs_info,
- struct btrfs_block_rsv *block_rsv)
+int btrfs_block_rsv_add(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv,
+ u64 num_bytes)
{
- block_rsv->durable = 1;
- mutex_lock(&fs_info->durable_block_rsv_mutex);
- list_add_tail(&block_rsv->list, &fs_info->durable_block_rsv_list);
- mutex_unlock(&fs_info->durable_block_rsv_mutex);
+ int ret;
+
+ if (num_bytes == 0)
+ return 0;
+
+ ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 1);
+ if (!ret) {
+ block_rsv_add_bytes(block_rsv, num_bytes, 1);
+ return 0;
+ }
+
+ return ret;
}
-int btrfs_block_rsv_add(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 num_bytes)
+int btrfs_block_rsv_add_noflush(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv,
+ u64 num_bytes)
{
int ret;
if (num_bytes == 0)
return 0;
- ret = reserve_metadata_bytes(trans, root, block_rsv, num_bytes, 1);
+ ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 0);
if (!ret) {
block_rsv_add_bytes(block_rsv, num_bytes, 1);
return 0;
return ret;
}
-int btrfs_block_rsv_check(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 min_reserved, int min_factor)
+int btrfs_block_rsv_check(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv, int min_factor)
{
u64 num_bytes = 0;
- int commit_trans = 0;
int ret = -ENOSPC;
if (!block_rsv)
return 0;
spin_lock(&block_rsv->lock);
- if (min_factor > 0)
- num_bytes = div_factor(block_rsv->size, min_factor);
- if (min_reserved > num_bytes)
- num_bytes = min_reserved;
+ num_bytes = div_factor(block_rsv->size, min_factor);
+ if (block_rsv->reserved >= num_bytes)
+ ret = 0;
+ spin_unlock(&block_rsv->lock);
- if (block_rsv->reserved >= num_bytes) {
+ return ret;
+}
+
+int btrfs_block_rsv_refill(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv,
+ u64 min_reserved)
+{
+ u64 num_bytes = 0;
+ int ret = -ENOSPC;
+
+ if (!block_rsv)
+ return 0;
+
+ spin_lock(&block_rsv->lock);
+ num_bytes = min_reserved;
+ if (block_rsv->reserved >= num_bytes)
ret = 0;
- } else {
+ else
num_bytes -= block_rsv->reserved;
- if (block_rsv->durable &&
- block_rsv->freed[0] + block_rsv->freed[1] >= num_bytes)
- commit_trans = 1;
- }
spin_unlock(&block_rsv->lock);
+
if (!ret)
return 0;
- if (block_rsv->refill_used) {
- ret = reserve_metadata_bytes(trans, root, block_rsv,
- num_bytes, 0);
- if (!ret) {
- block_rsv_add_bytes(block_rsv, num_bytes, 0);
- return 0;
- }
- }
-
- if (commit_trans) {
- if (trans)
- return -EAGAIN;
- trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
- ret = btrfs_commit_transaction(trans, root);
+ ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 1);
+ if (!ret) {
+ block_rsv_add_bytes(block_rsv, num_bytes, 0);
return 0;
}
- return -ENOSPC;
+ return ret;
}
int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv,
u64 num_bytes;
u64 meta_used;
u64 data_used;
- int csum_size = btrfs_super_csum_size(&fs_info->super_copy);
+ int csum_size = btrfs_super_csum_size(fs_info->super_copy);
sinfo = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_DATA);
spin_lock(&sinfo->lock);
if (sinfo->total_bytes > num_bytes) {
num_bytes = sinfo->total_bytes - num_bytes;
block_rsv->reserved += num_bytes;
- sinfo->bytes_reserved += num_bytes;
+ sinfo->bytes_may_use += num_bytes;
}
if (block_rsv->reserved >= block_rsv->size) {
num_bytes = block_rsv->reserved - block_rsv->size;
- sinfo->bytes_reserved -= num_bytes;
+ sinfo->bytes_may_use -= num_bytes;
sinfo->reservation_progress++;
block_rsv->reserved = block_rsv->size;
block_rsv->full = 1;
space_info = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
fs_info->chunk_block_rsv.space_info = space_info;
- fs_info->chunk_block_rsv.priority = 10;
space_info = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA);
fs_info->global_block_rsv.space_info = space_info;
- fs_info->global_block_rsv.priority = 10;
- fs_info->global_block_rsv.refill_used = 1;
fs_info->delalloc_block_rsv.space_info = space_info;
fs_info->trans_block_rsv.space_info = space_info;
fs_info->empty_block_rsv.space_info = space_info;
- fs_info->empty_block_rsv.priority = 10;
+ fs_info->delayed_block_rsv.space_info = space_info;
fs_info->extent_root->block_rsv = &fs_info->global_block_rsv;
fs_info->csum_root->block_rsv = &fs_info->global_block_rsv;
fs_info->tree_root->block_rsv = &fs_info->global_block_rsv;
fs_info->chunk_root->block_rsv = &fs_info->chunk_block_rsv;
- btrfs_add_durable_block_rsv(fs_info, &fs_info->global_block_rsv);
-
- btrfs_add_durable_block_rsv(fs_info, &fs_info->delalloc_block_rsv);
-
update_global_block_rsv(fs_info);
}
WARN_ON(fs_info->trans_block_rsv.reserved > 0);
WARN_ON(fs_info->chunk_block_rsv.size > 0);
WARN_ON(fs_info->chunk_block_rsv.reserved > 0);
-}
-
-int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_block_rsv *rsv)
-{
- struct btrfs_block_rsv *trans_rsv = &root->fs_info->trans_block_rsv;
- u64 num_bytes;
- int ret;
-
- /*
- * Truncate should be freeing data, but give us 2 items just in case it
- * needs to use some space. We may want to be smarter about this in the
- * future.
- */
- num_bytes = btrfs_calc_trans_metadata_size(root, 2);
-
- /* We already have enough bytes, just return */
- if (rsv->reserved >= num_bytes)
- return 0;
-
- num_bytes -= rsv->reserved;
-
- /*
- * You should have reserved enough space before hand to do this, so this
- * should not fail.
- */
- ret = block_rsv_migrate_bytes(trans_rsv, rsv, num_bytes);
- BUG_ON(ret);
-
- return 0;
+ WARN_ON(fs_info->delayed_block_rsv.size > 0);
+ WARN_ON(fs_info->delayed_block_rsv.reserved > 0);
}
void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
if (!trans->bytes_reserved)
return;
- BUG_ON(trans->block_rsv != &root->fs_info->trans_block_rsv);
- btrfs_block_rsv_release(root, trans->block_rsv,
- trans->bytes_reserved);
+ btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved);
trans->bytes_reserved = 0;
}
return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes);
}
+/**
+ * drop_outstanding_extent - drop an outstanding extent
+ * @inode: the inode we're dropping the extent for
+ *
+ * This is called when we are freeing up an outstanding extent, either called
+ * after an error or after an extent is written. This will return the number of
+ * reserved extents that need to be freed. This must be called with
+ * BTRFS_I(inode)->lock held.
+ */
static unsigned drop_outstanding_extent(struct inode *inode)
{
unsigned dropped_extents = 0;
- spin_lock(&BTRFS_I(inode)->lock);
BUG_ON(!BTRFS_I(inode)->outstanding_extents);
BTRFS_I(inode)->outstanding_extents--;
*/
if (BTRFS_I(inode)->outstanding_extents >=
BTRFS_I(inode)->reserved_extents)
- goto out;
+ return 0;
dropped_extents = BTRFS_I(inode)->reserved_extents -
BTRFS_I(inode)->outstanding_extents;
BTRFS_I(inode)->reserved_extents -= dropped_extents;
-out:
- spin_unlock(&BTRFS_I(inode)->lock);
return dropped_extents;
}
-static u64 calc_csum_metadata_size(struct inode *inode, u64 num_bytes)
+/**
+ * calc_csum_metadata_size - return the amount of metada space that must be
+ * reserved/free'd for the given bytes.
+ * @inode: the inode we're manipulating
+ * @num_bytes: the number of bytes in question
+ * @reserve: 1 if we are reserving space, 0 if we are freeing space
+ *
+ * This adjusts the number of csum_bytes in the inode and then returns the
+ * correct amount of metadata that must either be reserved or freed. We
+ * calculate how many checksums we can fit into one leaf and then divide the
+ * number of bytes that will need to be checksumed by this value to figure out
+ * how many checksums will be required. If we are adding bytes then the number
+ * may go up and we will return the number of additional bytes that must be
+ * reserved. If it is going down we will return the number of bytes that must
+ * be freed.
+ *
+ * This must be called with BTRFS_I(inode)->lock held.
+ */
+static u64 calc_csum_metadata_size(struct inode *inode, u64 num_bytes,
+ int reserve)
{
- return num_bytes >>= 3;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ u64 csum_size;
+ int num_csums_per_leaf;
+ int num_csums;
+ int old_csums;
+
+ if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM &&
+ BTRFS_I(inode)->csum_bytes == 0)
+ return 0;
+
+ old_csums = (int)div64_u64(BTRFS_I(inode)->csum_bytes, root->sectorsize);
+ if (reserve)
+ BTRFS_I(inode)->csum_bytes += num_bytes;
+ else
+ BTRFS_I(inode)->csum_bytes -= num_bytes;
+ csum_size = BTRFS_LEAF_DATA_SIZE(root) - sizeof(struct btrfs_item);
+ num_csums_per_leaf = (int)div64_u64(csum_size,
+ sizeof(struct btrfs_csum_item) +
+ sizeof(struct btrfs_disk_key));
+ num_csums = (int)div64_u64(BTRFS_I(inode)->csum_bytes, root->sectorsize);
+ num_csums = num_csums + num_csums_per_leaf - 1;
+ num_csums = num_csums / num_csums_per_leaf;
+
+ old_csums = old_csums + num_csums_per_leaf - 1;
+ old_csums = old_csums / num_csums_per_leaf;
+
+ /* No change, no need to reserve more */
+ if (old_csums == num_csums)
+ return 0;
+
+ if (reserve)
+ return btrfs_calc_trans_metadata_size(root,
+ num_csums - old_csums);
+
+ return btrfs_calc_trans_metadata_size(root, old_csums - num_csums);
}
int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes)
struct btrfs_block_rsv *block_rsv = &root->fs_info->delalloc_block_rsv;
u64 to_reserve = 0;
unsigned nr_extents = 0;
+ int flush = 1;
int ret;
- if (btrfs_transaction_in_commit(root->fs_info))
+ if (btrfs_is_free_space_inode(root, inode))
+ flush = 0;
+
+ if (flush && btrfs_transaction_in_commit(root->fs_info))
schedule_timeout(1);
num_bytes = ALIGN(num_bytes, root->sectorsize);
to_reserve = btrfs_calc_trans_metadata_size(root, nr_extents);
}
+ to_reserve += calc_csum_metadata_size(inode, num_bytes, 1);
spin_unlock(&BTRFS_I(inode)->lock);
- to_reserve += calc_csum_metadata_size(inode, num_bytes);
- ret = reserve_metadata_bytes(NULL, root, block_rsv, to_reserve, 1);
+ ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush);
if (ret) {
+ u64 to_free = 0;
unsigned dropped;
+
+ spin_lock(&BTRFS_I(inode)->lock);
+ dropped = drop_outstanding_extent(inode);
+ to_free = calc_csum_metadata_size(inode, num_bytes, 0);
+ spin_unlock(&BTRFS_I(inode)->lock);
+ to_free += btrfs_calc_trans_metadata_size(root, dropped);
+
/*
- * We don't need the return value since our reservation failed,
- * we just need to clean up our counter.
+ * Somebody could have come in and twiddled with the
+ * reservation, so if we have to free more than we would have
+ * reserved from this reservation go ahead and release those
+ * bytes.
*/
- dropped = drop_outstanding_extent(inode);
- WARN_ON(dropped > 1);
+ to_free -= to_reserve;
+ if (to_free)
+ btrfs_block_rsv_release(root, block_rsv, to_free);
return ret;
}
return 0;
}
+/**
+ * btrfs_delalloc_release_metadata - release a metadata reservation for an inode
+ * @inode: the inode to release the reservation for
+ * @num_bytes: the number of bytes we're releasing
+ *
+ * This will release the metadata reservation for an inode. This can be called
+ * once we complete IO for a given set of bytes to release their metadata
+ * reservations.
+ */
void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
unsigned dropped;
num_bytes = ALIGN(num_bytes, root->sectorsize);
+ spin_lock(&BTRFS_I(inode)->lock);
dropped = drop_outstanding_extent(inode);
- to_free = calc_csum_metadata_size(inode, num_bytes);
+ to_free = calc_csum_metadata_size(inode, num_bytes, 0);
+ spin_unlock(&BTRFS_I(inode)->lock);
if (dropped > 0)
to_free += btrfs_calc_trans_metadata_size(root, dropped);
to_free);
}
+/**
+ * btrfs_delalloc_reserve_space - reserve data and metadata space for delalloc
+ * @inode: inode we're writing to
+ * @num_bytes: the number of bytes we want to allocate
+ *
+ * This will do the following things
+ *
+ * o reserve space in the data space info for num_bytes
+ * o reserve space in the metadata space info based on number of outstanding
+ * extents and how much csums will be needed
+ * o add to the inodes ->delalloc_bytes
+ * o add it to the fs_info's delalloc inodes list.
+ *
+ * This will return 0 for success and -ENOSPC if there is no space left.
+ */
int btrfs_delalloc_reserve_space(struct inode *inode, u64 num_bytes)
{
int ret;
return 0;
}
+/**
+ * btrfs_delalloc_release_space - release data and metadata space for delalloc
+ * @inode: inode we're releasing space for
+ * @num_bytes: the number of bytes we want to free up
+ *
+ * This must be matched with a call to btrfs_delalloc_reserve_space. This is
+ * called in the case that we don't need the metadata AND data reservations
+ * anymore. So if there is an error or we insert an inline extent.
+ *
+ * This function will release the metadata space that was not used and will
+ * decrement ->delalloc_bytes and remove it from the fs_info delalloc_inodes
+ * list if there are no delalloc bytes left.
+ */
void btrfs_delalloc_release_space(struct inode *inode, u64 num_bytes)
{
btrfs_delalloc_release_metadata(inode, num_bytes);
/* block accounting for super block */
spin_lock(&info->delalloc_lock);
- old_val = btrfs_super_bytes_used(&info->super_copy);
+ old_val = btrfs_super_bytes_used(info->super_copy);
if (alloc)
old_val += num_bytes;
else
old_val -= num_bytes;
- btrfs_set_super_bytes_used(&info->super_copy, old_val);
+ btrfs_set_super_bytes_used(info->super_copy, old_val);
spin_unlock(&info->delalloc_lock);
while (total) {
spin_lock(&cache->space_info->lock);
spin_lock(&cache->lock);
- if (btrfs_super_cache_generation(&info->super_copy) != 0 &&
+ if (btrfs_test_opt(root, SPACE_CACHE) &&
cache->disk_cache_state < BTRFS_DC_CLEAR)
cache->disk_cache_state = BTRFS_DC_CLEAR;
btrfs_set_block_group_used(&cache->item, old_val);
cache->reserved -= num_bytes;
cache->space_info->bytes_reserved -= num_bytes;
- cache->space_info->reservation_progress++;
cache->space_info->bytes_used += num_bytes;
cache->space_info->disk_used += num_bytes * factor;
spin_unlock(&cache->lock);
if (reserved) {
cache->reserved -= num_bytes;
cache->space_info->bytes_reserved -= num_bytes;
- cache->space_info->reservation_progress++;
}
spin_unlock(&cache->lock);
spin_unlock(&cache->space_info->lock);
}
/*
- * update size of reserved extents. this function may return -EAGAIN
- * if 'reserve' is true or 'sinfo' is false.
+ * this function must be called within transaction
+ */
+int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ u64 bytenr, u64 num_bytes)
+{
+ struct btrfs_block_group_cache *cache;
+
+ cache = btrfs_lookup_block_group(root->fs_info, bytenr);
+ BUG_ON(!cache);
+
+ /*
+ * pull in the free space cache (if any) so that our pin
+ * removes the free space from the cache. We have load_only set
+ * to one because the slow code to read in the free extents does check
+ * the pinned extents.
+ */
+ cache_block_group(cache, trans, root, 1);
+
+ pin_down_extent(root, cache, bytenr, num_bytes, 0);
+
+ /* remove us from the free space cache (if we're there at all) */
+ btrfs_remove_free_space(cache, bytenr, num_bytes);
+ btrfs_put_block_group(cache);
+ return 0;
+}
+
+/**
+ * btrfs_update_reserved_bytes - update the block_group and space info counters
+ * @cache: The cache we are manipulating
+ * @num_bytes: The number of bytes in question
+ * @reserve: One of the reservation enums
+ *
+ * This is called by the allocator when it reserves space, or by somebody who is
+ * freeing space that was never actually used on disk. For example if you
+ * reserve some space for a new leaf in transaction A and before transaction A
+ * commits you free that leaf, you call this with reserve set to 0 in order to
+ * clear the reservation.
+ *
+ * Metadata reservations should be called with RESERVE_ALLOC so we do the proper
+ * ENOSPC accounting. For data we handle the reservation through clearing the
+ * delalloc bits in the io_tree. We have to do this since we could end up
+ * allocating less disk space for the amount of data we have reserved in the
+ * case of compression.
+ *
+ * If this is a reservation and the block group has become read only we cannot
+ * make the reservation and return -EAGAIN, otherwise this function always
+ * succeeds.
*/
-int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
- u64 num_bytes, int reserve, int sinfo)
+static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
+ u64 num_bytes, int reserve)
{
+ struct btrfs_space_info *space_info = cache->space_info;
int ret = 0;
- if (sinfo) {
- struct btrfs_space_info *space_info = cache->space_info;
- spin_lock(&space_info->lock);
- spin_lock(&cache->lock);
- if (reserve) {
- if (cache->ro) {
- ret = -EAGAIN;
- } else {
- cache->reserved += num_bytes;
- space_info->bytes_reserved += num_bytes;
- }
- } else {
- if (cache->ro)
- space_info->bytes_readonly += num_bytes;
- cache->reserved -= num_bytes;
- space_info->bytes_reserved -= num_bytes;
- space_info->reservation_progress++;
- }
- spin_unlock(&cache->lock);
- spin_unlock(&space_info->lock);
- } else {
- spin_lock(&cache->lock);
+ spin_lock(&space_info->lock);
+ spin_lock(&cache->lock);
+ if (reserve != RESERVE_FREE) {
if (cache->ro) {
ret = -EAGAIN;
} else {
- if (reserve)
- cache->reserved += num_bytes;
- else
- cache->reserved -= num_bytes;
+ cache->reserved += num_bytes;
+ space_info->bytes_reserved += num_bytes;
+ if (reserve == RESERVE_ALLOC) {
+ BUG_ON(space_info->bytes_may_use < num_bytes);
+ space_info->bytes_may_use -= num_bytes;
+ }
}
- spin_unlock(&cache->lock);
+ } else {
+ if (cache->ro)
+ space_info->bytes_readonly += num_bytes;
+ cache->reserved -= num_bytes;
+ space_info->bytes_reserved -= num_bytes;
+ space_info->reservation_progress++;
}
+ spin_unlock(&cache->lock);
+ spin_unlock(&space_info->lock);
return ret;
}
spin_lock(&cache->lock);
cache->pinned -= len;
cache->space_info->bytes_pinned -= len;
- if (cache->ro) {
+ if (cache->ro)
cache->space_info->bytes_readonly += len;
- } else if (cache->reserved_pinned > 0) {
- len = min(len, cache->reserved_pinned);
- cache->reserved_pinned -= len;
- cache->space_info->bytes_reserved += len;
- }
spin_unlock(&cache->lock);
spin_unlock(&cache->space_info->lock);
}
{
struct btrfs_fs_info *fs_info = root->fs_info;
struct extent_io_tree *unpin;
- struct btrfs_block_rsv *block_rsv;
- struct btrfs_block_rsv *next_rsv;
u64 start;
u64 end;
- int idx;
int ret;
if (fs_info->pinned_extents == &fs_info->freed_extents[0])
cond_resched();
}
- mutex_lock(&fs_info->durable_block_rsv_mutex);
- list_for_each_entry_safe(block_rsv, next_rsv,
- &fs_info->durable_block_rsv_list, list) {
-
- idx = trans->transid & 0x1;
- if (block_rsv->freed[idx] > 0) {
- block_rsv_add_bytes(block_rsv,
- block_rsv->freed[idx], 0);
- block_rsv->freed[idx] = 0;
- }
- if (atomic_read(&block_rsv->usage) == 0) {
- btrfs_block_rsv_release(root, block_rsv, (u64)-1);
-
- if (block_rsv->freed[0] == 0 &&
- block_rsv->freed[1] == 0) {
- list_del_init(&block_rsv->list);
- kfree(block_rsv);
- }
- } else {
- btrfs_block_rsv_release(root, block_rsv, 0);
- }
- }
- mutex_unlock(&fs_info->durable_block_rsv_mutex);
-
return 0;
}
struct extent_buffer *buf,
u64 parent, int last_ref)
{
- struct btrfs_block_rsv *block_rsv;
struct btrfs_block_group_cache *cache = NULL;
int ret;
if (!last_ref)
return;
- block_rsv = get_block_rsv(trans, root);
cache = btrfs_lookup_block_group(root->fs_info, buf->start);
- if (block_rsv->space_info != cache->space_info)
- goto out;
if (btrfs_header_generation(buf) == trans->transid) {
if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
ret = check_ref_cleanup(trans, root, buf->start);
if (!ret)
- goto pin;
+ goto out;
}
if (btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN)) {
pin_down_extent(root, cache, buf->start, buf->len, 1);
- goto pin;
+ goto out;
}
WARN_ON(test_bit(EXTENT_BUFFER_DIRTY, &buf->bflags));
btrfs_add_free_space(cache, buf->start, buf->len);
- ret = btrfs_update_reserved_bytes(cache, buf->len, 0, 0);
- if (ret == -EAGAIN) {
- /* block group became read-only */
- btrfs_update_reserved_bytes(cache, buf->len, 0, 1);
- goto out;
- }
-
- ret = 1;
- spin_lock(&block_rsv->lock);
- if (block_rsv->reserved < block_rsv->size) {
- block_rsv->reserved += buf->len;
- ret = 0;
- }
- spin_unlock(&block_rsv->lock);
-
- if (ret) {
- spin_lock(&cache->space_info->lock);
- cache->space_info->bytes_reserved -= buf->len;
- cache->space_info->reservation_progress++;
- spin_unlock(&cache->space_info->lock);
- }
- goto out;
- }
-pin:
- if (block_rsv->durable && !cache->ro) {
- ret = 0;
- spin_lock(&cache->lock);
- if (!cache->ro) {
- cache->reserved_pinned += buf->len;
- ret = 1;
- }
- spin_unlock(&cache->lock);
-
- if (ret) {
- spin_lock(&block_rsv->lock);
- block_rsv->freed[trans->transid & 0x1] += buf->len;
- spin_unlock(&block_rsv->lock);
- }
+ btrfs_update_reserved_bytes(cache, buf->len, RESERVE_FREE);
}
out:
/*
int last_ptr_loop = 0;
int loop = 0;
int index = 0;
+ int alloc_type = (data & BTRFS_BLOCK_GROUP_DATA) ?
+ RESERVE_ALLOC_NO_ACCOUNT : RESERVE_ALLOC;
bool found_uncached_bg = false;
bool failed_cluster_refill = false;
bool failed_alloc = false;
bool use_cluster = true;
+ bool have_caching_bg = false;
u64 ideal_cache_percent = 0;
u64 ideal_cache_offset = 0;
}
}
search:
+ have_caching_bg = false;
down_read(&space_info->groups_sem);
list_for_each_entry(block_group, &space_info->block_groups[index],
list) {
failed_alloc = true;
goto have_block_group;
} else if (!offset) {
+ if (!cached)
+ have_caching_bg = true;
goto loop;
}
checks:
search_start - offset);
BUG_ON(offset > search_start);
- ret = btrfs_update_reserved_bytes(block_group, num_bytes, 1,
- (data & BTRFS_BLOCK_GROUP_DATA));
+ ret = btrfs_update_reserved_bytes(block_group, num_bytes,
+ alloc_type);
if (ret == -EAGAIN) {
btrfs_add_free_space(block_group, offset, num_bytes);
goto loop;
}
up_read(&space_info->groups_sem);
+ if (!ins->objectid && loop >= LOOP_CACHING_WAIT && have_caching_bg)
+ goto search;
+
if (!ins->objectid && ++index < BTRFS_NR_RAID_TYPES)
goto search;
int index = 0;
spin_lock(&info->lock);
- printk(KERN_INFO "space_info has %llu free, is %sfull\n",
+ printk(KERN_INFO "space_info %llu has %llu free, is %sfull\n",
+ (unsigned long long)info->flags,
(unsigned long long)(info->total_bytes - info->bytes_used -
info->bytes_pinned - info->bytes_reserved -
info->bytes_readonly),
return ret;
}
-int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len)
+static int __btrfs_free_reserved_extent(struct btrfs_root *root,
+ u64 start, u64 len, int pin)
{
struct btrfs_block_group_cache *cache;
int ret = 0;
if (btrfs_test_opt(root, DISCARD))
ret = btrfs_discard_extent(root, start, len, NULL);
- btrfs_add_free_space(cache, start, len);
- btrfs_update_reserved_bytes(cache, len, 0, 1);
+ if (pin)
+ pin_down_extent(root, cache, start, len, 1);
+ else {
+ btrfs_add_free_space(cache, start, len);
+ btrfs_update_reserved_bytes(cache, len, RESERVE_FREE);
+ }
btrfs_put_block_group(cache);
trace_btrfs_reserved_extent_free(root, start, len);
return ret;
}
+int btrfs_free_reserved_extent(struct btrfs_root *root,
+ u64 start, u64 len)
+{
+ return __btrfs_free_reserved_extent(root, start, len, 0);
+}
+
+int btrfs_free_and_pin_reserved_extent(struct btrfs_root *root,
+ u64 start, u64 len)
+{
+ return __btrfs_free_reserved_extent(root, start, len, 1);
+}
+
static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 parent, u64 root_objectid,
put_caching_control(caching_ctl);
}
- ret = btrfs_update_reserved_bytes(block_group, ins->offset, 1, 1);
+ ret = btrfs_update_reserved_bytes(block_group, ins->offset,
+ RESERVE_ALLOC_NO_ACCOUNT);
BUG_ON(ret);
btrfs_put_block_group(block_group);
ret = alloc_reserved_file_extent(trans, root, 0, root_objectid,
block_rsv = get_block_rsv(trans, root);
if (block_rsv->size == 0) {
- ret = reserve_metadata_bytes(trans, root, block_rsv,
- blocksize, 0);
+ ret = reserve_metadata_bytes(root, block_rsv, blocksize, 0);
/*
* If we couldn't reserve metadata bytes try and use some from
* the global reserve.
if (!ret)
return block_rsv;
if (ret) {
- WARN_ON(1);
- ret = reserve_metadata_bytes(trans, root, block_rsv, blocksize,
- 0);
+ static DEFINE_RATELIMIT_STATE(_rs,
+ DEFAULT_RATELIMIT_INTERVAL,
+ /*DEFAULT_RATELIMIT_BURST*/ 2);
+ if (__ratelimit(&_rs)) {
+ printk(KERN_DEBUG "btrfs: block rsv returned %d\n", ret);
+ WARN_ON(1);
+ }
+ ret = reserve_metadata_bytes(root, block_rsv, blocksize, 0);
if (!ret) {
- spin_lock(&block_rsv->lock);
- block_rsv->size += blocksize;
- spin_unlock(&block_rsv->lock);
return block_rsv;
} else if (ret && block_rsv != global_rsv) {
ret = block_rsv_use_bytes(global_rsv, blocksize);
cache->bytes_super - btrfs_block_group_used(&cache->item);
if (sinfo->bytes_used + sinfo->bytes_reserved + sinfo->bytes_pinned +
- sinfo->bytes_may_use + sinfo->bytes_readonly +
- cache->reserved_pinned + num_bytes + min_allocable_bytes <=
- sinfo->total_bytes) {
+ sinfo->bytes_may_use + sinfo->bytes_readonly + num_bytes +
+ min_allocable_bytes <= sinfo->total_bytes) {
sinfo->bytes_readonly += num_bytes;
- sinfo->bytes_reserved += cache->reserved_pinned;
- cache->reserved_pinned = 0;
cache->ro = 1;
ret = 0;
}
struct btrfs_space_info,
list);
if (space_info->bytes_pinned > 0 ||
- space_info->bytes_reserved > 0) {
+ space_info->bytes_reserved > 0 ||
+ space_info->bytes_may_use > 0) {
WARN_ON(1);
dump_space_info(space_info, 0, 0);
}
return -ENOMEM;
path->reada = 1;
- cache_gen = btrfs_super_cache_generation(&root->fs_info->super_copy);
- if (cache_gen != 0 &&
- btrfs_super_generation(&root->fs_info->super_copy) != cache_gen)
+ cache_gen = btrfs_super_cache_generation(root->fs_info->super_copy);
+ if (btrfs_test_opt(root, SPACE_CACHE) &&
+ btrfs_super_generation(root->fs_info->super_copy) != cache_gen)
need_clear = 1;
if (btrfs_test_opt(root, CLEAR_CACHE))
need_clear = 1;
- if (!btrfs_test_opt(root, SPACE_CACHE) && cache_gen)
- printk(KERN_INFO "btrfs: disk space caching is enabled\n");
while (1) {
ret = find_first_block_group(root, path, &key);
goto out;
}
- inode = lookup_free_space_inode(root, block_group, path);
+ inode = lookup_free_space_inode(tree_root, block_group, path);
if (!IS_ERR(inode)) {
ret = btrfs_orphan_add(trans, inode);
BUG_ON(ret);
spin_unlock(&block_group->lock);
}
/* One for our lookup ref */
- iput(inode);
+ btrfs_add_delayed_iput(inode);
}
key.objectid = BTRFS_FREE_SPACE_OBJECTID;
int mixed = 0;
int ret;
- disk_super = &fs_info->super_copy;
+ disk_super = fs_info->super_copy;
if (!btrfs_super_root(disk_super))
return 1;
#include "compat.h"
#include "ctree.h"
#include "btrfs_inode.h"
+ #include "volumes.h"
static struct kmem_cache *extent_state_cache;
static struct kmem_cache *extent_buffer_cache;
goto again;
}
+/**
+ * convert_extent - convert all bits in a given range from one bit to another
+ * @tree: the io tree to search
+ * @start: the start offset in bytes
+ * @end: the end offset in bytes (inclusive)
+ * @bits: the bits to set in this range
+ * @clear_bits: the bits to clear in this range
+ * @mask: the allocation mask
+ *
+ * This will go through and set bits for the given range. If any states exist
+ * already in this range they are set with the given bit and cleared of the
+ * clear_bits. This is only meant to be used by things that are mergeable, ie
+ * converting from say DELALLOC to DIRTY. This is not meant to be used with
+ * boundary bits like LOCK.
+ */
+int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+ int bits, int clear_bits, gfp_t mask)
+{
+ struct extent_state *state;
+ struct extent_state *prealloc = NULL;
+ struct rb_node *node;
+ int err = 0;
+ u64 last_start;
+ u64 last_end;
+
+again:
+ if (!prealloc && (mask & __GFP_WAIT)) {
+ prealloc = alloc_extent_state(mask);
+ if (!prealloc)
+ return -ENOMEM;
+ }
+
+ spin_lock(&tree->lock);
+ /*
+ * this search will find all the extents that end after
+ * our range starts.
+ */
+ node = tree_search(tree, start);
+ if (!node) {
+ prealloc = alloc_extent_state_atomic(prealloc);
+ if (!prealloc)
+ return -ENOMEM;
+ err = insert_state(tree, prealloc, start, end, &bits);
+ prealloc = NULL;
+ BUG_ON(err == -EEXIST);
+ goto out;
+ }
+ state = rb_entry(node, struct extent_state, rb_node);
+hit_next:
+ last_start = state->start;
+ last_end = state->end;
+
+ /*
+ * | ---- desired range ---- |
+ * | state |
+ *
+ * Just lock what we found and keep going
+ */
+ if (state->start == start && state->end <= end) {
+ struct rb_node *next_node;
+
+ set_state_bits(tree, state, &bits);
+ clear_state_bit(tree, state, &clear_bits, 0);
+
+ merge_state(tree, state);
+ if (last_end == (u64)-1)
+ goto out;
+
+ start = last_end + 1;
+ next_node = rb_next(&state->rb_node);
+ if (next_node && start < end && prealloc && !need_resched()) {
+ state = rb_entry(next_node, struct extent_state,
+ rb_node);
+ if (state->start == start)
+ goto hit_next;
+ }
+ goto search_again;
+ }
+
+ /*
+ * | ---- desired range ---- |
+ * | state |
+ * or
+ * | ------------- state -------------- |
+ *
+ * We need to split the extent we found, and may flip bits on
+ * second half.
+ *
+ * If the extent we found extends past our
+ * range, we just split and search again. It'll get split
+ * again the next time though.
+ *
+ * If the extent we found is inside our range, we set the
+ * desired bit on it.
+ */
+ if (state->start < start) {
+ prealloc = alloc_extent_state_atomic(prealloc);
+ if (!prealloc)
+ return -ENOMEM;
+ err = split_state(tree, state, prealloc, start);
+ BUG_ON(err == -EEXIST);
+ prealloc = NULL;
+ if (err)
+ goto out;
+ if (state->end <= end) {
+ set_state_bits(tree, state, &bits);
+ clear_state_bit(tree, state, &clear_bits, 0);
+ merge_state(tree, state);
+ if (last_end == (u64)-1)
+ goto out;
+ start = last_end + 1;
+ }
+ goto search_again;
+ }
+ /*
+ * | ---- desired range ---- |
+ * | state | or | state |
+ *
+ * There's a hole, we need to insert something in it and
+ * ignore the extent we found.
+ */
+ if (state->start > start) {
+ u64 this_end;
+ if (end < last_start)
+ this_end = end;
+ else
+ this_end = last_start - 1;
+
+ prealloc = alloc_extent_state_atomic(prealloc);
+ if (!prealloc)
+ return -ENOMEM;
+
+ /*
+ * Avoid to free 'prealloc' if it can be merged with
+ * the later extent.
+ */
+ err = insert_state(tree, prealloc, start, this_end,
+ &bits);
+ BUG_ON(err == -EEXIST);
+ if (err) {
+ free_extent_state(prealloc);
+ prealloc = NULL;
+ goto out;
+ }
+ prealloc = NULL;
+ start = this_end + 1;
+ goto search_again;
+ }
+ /*
+ * | ---- desired range ---- |
+ * | state |
+ * We need to split the extent, and set the bit
+ * on the first half
+ */
+ if (state->start <= end && state->end > end) {
+ prealloc = alloc_extent_state_atomic(prealloc);
+ if (!prealloc)
+ return -ENOMEM;
+
+ err = split_state(tree, state, prealloc, end + 1);
+ BUG_ON(err == -EEXIST);
+
+ set_state_bits(tree, prealloc, &bits);
+ clear_state_bit(tree, prealloc, &clear_bits, 0);
+
+ merge_state(tree, prealloc);
+ prealloc = NULL;
+ goto out;
+ }
+
+ goto search_again;
+
+out:
+ spin_unlock(&tree->lock);
+ if (prealloc)
+ free_extent_state(prealloc);
+
+ return err;
+
+search_again:
+ if (start > end)
+ goto out;
+ spin_unlock(&tree->lock);
+ if (mask & __GFP_WAIT)
+ cond_resched();
+ goto again;
+}
+
/* wrappers around set/clear extent bit */
int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask)
struct extent_state **cached_state, gfp_t mask)
{
return set_extent_bit(tree, start, end,
- EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE,
+ EXTENT_DELALLOC | EXTENT_UPTODATE,
0, NULL, cached_state, mask);
}
return 0;
}
+ /*
+ * When IO fails, either with EIO or csum verification fails, we
+ * try other mirrors that might have a good copy of the data. This
+ * io_failure_record is used to record state as we go through all the
+ * mirrors. If another mirror has good data, the page is set up to date
+ * and things continue. If a good mirror can't be found, the original
+ * bio end_io callback is called to indicate things have failed.
+ */
+ struct io_failure_record {
+ struct page *page;
+ u64 start;
+ u64 len;
+ u64 logical;
+ unsigned long bio_flags;
+ int this_mirror;
+ int failed_mirror;
+ int in_validation;
+ };
+
+ static int free_io_failure(struct inode *inode, struct io_failure_record *rec,
+ int did_repair)
+ {
+ int ret;
+ int err = 0;
+ struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
+
+ set_state_private(failure_tree, rec->start, 0);
+ ret = clear_extent_bits(failure_tree, rec->start,
+ rec->start + rec->len - 1,
+ EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
+ if (ret)
+ err = ret;
+
+ if (did_repair) {
+ ret = clear_extent_bits(&BTRFS_I(inode)->io_tree, rec->start,
+ rec->start + rec->len - 1,
+ EXTENT_DAMAGED, GFP_NOFS);
+ if (ret && !err)
+ err = ret;
+ }
+
+ kfree(rec);
+ return err;
+ }
+
+ static void repair_io_failure_callback(struct bio *bio, int err)
+ {
+ complete(bio->bi_private);
+ }
+
+ /*
+ * this bypasses the standard btrfs submit functions deliberately, as
+ * the standard behavior is to write all copies in a raid setup. here we only
+ * want to write the one bad copy. so we do the mapping for ourselves and issue
+ * submit_bio directly.
+ * to avoid any synchonization issues, wait for the data after writing, which
+ * actually prevents the read that triggered the error from finishing.
+ * currently, there can be no more than two copies of every data bit. thus,
+ * exactly one rewrite is required.
+ */
+ int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start,
+ u64 length, u64 logical, struct page *page,
+ int mirror_num)
+ {
+ struct bio *bio;
+ struct btrfs_device *dev;
+ DECLARE_COMPLETION_ONSTACK(compl);
+ u64 map_length = 0;
+ u64 sector;
+ struct btrfs_bio *bbio = NULL;
+ int ret;
+
+ BUG_ON(!mirror_num);
+
+ bio = bio_alloc(GFP_NOFS, 1);
+ if (!bio)
+ return -EIO;
+ bio->bi_private = &compl;
+ bio->bi_end_io = repair_io_failure_callback;
+ bio->bi_size = 0;
+ map_length = length;
+
+ ret = btrfs_map_block(map_tree, WRITE, logical,
+ &map_length, &bbio, mirror_num);
+ if (ret) {
+ bio_put(bio);
+ return -EIO;
+ }
+ BUG_ON(mirror_num != bbio->mirror_num);
+ sector = bbio->stripes[mirror_num-1].physical >> 9;
+ bio->bi_sector = sector;
+ dev = bbio->stripes[mirror_num-1].dev;
+ kfree(bbio);
+ if (!dev || !dev->bdev || !dev->writeable) {
+ bio_put(bio);
+ return -EIO;
+ }
+ bio->bi_bdev = dev->bdev;
+ bio_add_page(bio, page, length, start-page_offset(page));
+ submit_bio(WRITE_SYNC, bio);
+ wait_for_completion(&compl);
+
+ if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) {
+ /* try to remap that extent elsewhere? */
+ bio_put(bio);
+ return -EIO;
+ }
+
+ printk(KERN_INFO "btrfs read error corrected: ino %lu off %llu (dev %s "
+ "sector %llu)\n", page->mapping->host->i_ino, start,
+ dev->name, sector);
+
+ bio_put(bio);
+ return 0;
+ }
+
+ /*
+ * each time an IO finishes, we do a fast check in the IO failure tree
+ * to see if we need to process or clean up an io_failure_record
+ */
+ static int clean_io_failure(u64 start, struct page *page)
+ {
+ u64 private;
+ u64 private_failure;
+ struct io_failure_record *failrec;
+ struct btrfs_mapping_tree *map_tree;
+ struct extent_state *state;
+ int num_copies;
+ int did_repair = 0;
+ int ret;
+ struct inode *inode = page->mapping->host;
+
+ private = 0;
+ ret = count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
+ (u64)-1, 1, EXTENT_DIRTY, 0);
+ if (!ret)
+ return 0;
+
+ ret = get_state_private(&BTRFS_I(inode)->io_failure_tree, start,
+ &private_failure);
+ if (ret)
+ return 0;
+
+ failrec = (struct io_failure_record *)(unsigned long) private_failure;
+ BUG_ON(!failrec->this_mirror);
+
+ if (failrec->in_validation) {
+ /* there was no real error, just free the record */
+ pr_debug("clean_io_failure: freeing dummy error at %llu\n",
+ failrec->start);
+ did_repair = 1;
+ goto out;
+ }
+
+ spin_lock(&BTRFS_I(inode)->io_tree.lock);
+ state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
+ failrec->start,
+ EXTENT_LOCKED);
+ spin_unlock(&BTRFS_I(inode)->io_tree.lock);
+
+ if (state && state->start == failrec->start) {
+ map_tree = &BTRFS_I(inode)->root->fs_info->mapping_tree;
+ num_copies = btrfs_num_copies(map_tree, failrec->logical,
+ failrec->len);
+ if (num_copies > 1) {
+ ret = repair_io_failure(map_tree, start, failrec->len,
+ failrec->logical, page,
+ failrec->failed_mirror);
+ did_repair = !ret;
+ }
+ }
+
+ out:
+ if (!ret)
+ ret = free_io_failure(inode, failrec, did_repair);
+
+ return ret;
+ }
+
+ /*
+ * this is a generic handler for readpage errors (default
+ * readpage_io_failed_hook). if other copies exist, read those and write back
+ * good data to the failed position. does not investigate in remapping the
+ * failed extent elsewhere, hoping the device will be smart enough to do this as
+ * needed
+ */
+
+ static int bio_readpage_error(struct bio *failed_bio, struct page *page,
+ u64 start, u64 end, int failed_mirror,
+ struct extent_state *state)
+ {
+ struct io_failure_record *failrec = NULL;
+ u64 private;
+ struct extent_map *em;
+ struct inode *inode = page->mapping->host;
+ struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
+ struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
+ struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct bio *bio;
+ int num_copies;
+ int ret;
+ int read_mode;
+ u64 logical;
+
+ BUG_ON(failed_bio->bi_rw & REQ_WRITE);
+
+ ret = get_state_private(failure_tree, start, &private);
+ if (ret) {
+ failrec = kzalloc(sizeof(*failrec), GFP_NOFS);
+ if (!failrec)
+ return -ENOMEM;
+ failrec->start = start;
+ failrec->len = end - start + 1;
+ failrec->this_mirror = 0;
+ failrec->bio_flags = 0;
+ failrec->in_validation = 0;
+
+ read_lock(&em_tree->lock);
+ em = lookup_extent_mapping(em_tree, start, failrec->len);
+ if (!em) {
+ read_unlock(&em_tree->lock);
+ kfree(failrec);
+ return -EIO;
+ }
+
+ if (em->start > start || em->start + em->len < start) {
+ free_extent_map(em);
+ em = NULL;
+ }
+ read_unlock(&em_tree->lock);
+
+ if (!em || IS_ERR(em)) {
+ kfree(failrec);
+ return -EIO;
+ }
+ logical = start - em->start;
+ logical = em->block_start + logical;
+ if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
+ logical = em->block_start;
+ failrec->bio_flags = EXTENT_BIO_COMPRESSED;
+ extent_set_compress_type(&failrec->bio_flags,
+ em->compress_type);
+ }
+ pr_debug("bio_readpage_error: (new) logical=%llu, start=%llu, "
+ "len=%llu\n", logical, start, failrec->len);
+ failrec->logical = logical;
+ free_extent_map(em);
+
+ /* set the bits in the private failure tree */
+ ret = set_extent_bits(failure_tree, start, end,
+ EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
+ if (ret >= 0)
+ ret = set_state_private(failure_tree, start,
+ (u64)(unsigned long)failrec);
+ /* set the bits in the inode's tree */
+ if (ret >= 0)
+ ret = set_extent_bits(tree, start, end, EXTENT_DAMAGED,
+ GFP_NOFS);
+ if (ret < 0) {
+ kfree(failrec);
+ return ret;
+ }
+ } else {
+ failrec = (struct io_failure_record *)(unsigned long)private;
+ pr_debug("bio_readpage_error: (found) logical=%llu, "
+ "start=%llu, len=%llu, validation=%d\n",
+ failrec->logical, failrec->start, failrec->len,
+ failrec->in_validation);
+ /*
+ * when data can be on disk more than twice, add to failrec here
+ * (e.g. with a list for failed_mirror) to make
+ * clean_io_failure() clean all those errors at once.
+ */
+ }
+ num_copies = btrfs_num_copies(
+ &BTRFS_I(inode)->root->fs_info->mapping_tree,
+ failrec->logical, failrec->len);
+ if (num_copies == 1) {
+ /*
+ * we only have a single copy of the data, so don't bother with
+ * all the retry and error correction code that follows. no
+ * matter what the error is, it is very likely to persist.
+ */
+ pr_debug("bio_readpage_error: cannot repair, num_copies == 1. "
+ "state=%p, num_copies=%d, next_mirror %d, "
+ "failed_mirror %d\n", state, num_copies,
+ failrec->this_mirror, failed_mirror);
+ free_io_failure(inode, failrec, 0);
+ return -EIO;
+ }
+
+ if (!state) {
+ spin_lock(&tree->lock);
+ state = find_first_extent_bit_state(tree, failrec->start,
+ EXTENT_LOCKED);
+ if (state && state->start != failrec->start)
+ state = NULL;
+ spin_unlock(&tree->lock);
+ }
+
+ /*
+ * there are two premises:
+ * a) deliver good data to the caller
+ * b) correct the bad sectors on disk
+ */
+ if (failed_bio->bi_vcnt > 1) {
+ /*
+ * to fulfill b), we need to know the exact failing sectors, as
+ * we don't want to rewrite any more than the failed ones. thus,
+ * we need separate read requests for the failed bio
+ *
+ * if the following BUG_ON triggers, our validation request got
+ * merged. we need separate requests for our algorithm to work.
+ */
+ BUG_ON(failrec->in_validation);
+ failrec->in_validation = 1;
+ failrec->this_mirror = failed_mirror;
+ read_mode = READ_SYNC | REQ_FAILFAST_DEV;
+ } else {
+ /*
+ * we're ready to fulfill a) and b) alongside. get a good copy
+ * of the failed sector and if we succeed, we have setup
+ * everything for repair_io_failure to do the rest for us.
+ */
+ if (failrec->in_validation) {
+ BUG_ON(failrec->this_mirror != failed_mirror);
+ failrec->in_validation = 0;
+ failrec->this_mirror = 0;
+ }
+ failrec->failed_mirror = failed_mirror;
+ failrec->this_mirror++;
+ if (failrec->this_mirror == failed_mirror)
+ failrec->this_mirror++;
+ read_mode = READ_SYNC;
+ }
+
+ if (!state || failrec->this_mirror > num_copies) {
+ pr_debug("bio_readpage_error: (fail) state=%p, num_copies=%d, "
+ "next_mirror %d, failed_mirror %d\n", state,
+ num_copies, failrec->this_mirror, failed_mirror);
+ free_io_failure(inode, failrec, 0);
+ return -EIO;
+ }
+
+ bio = bio_alloc(GFP_NOFS, 1);
+ bio->bi_private = state;
+ bio->bi_end_io = failed_bio->bi_end_io;
+ bio->bi_sector = failrec->logical >> 9;
+ bio->bi_bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
+ bio->bi_size = 0;
+
+ bio_add_page(bio, page, failrec->len, start - page_offset(page));
+
+ pr_debug("bio_readpage_error: submitting new read[%#x] to "
+ "this_mirror=%d, num_copies=%d, in_validation=%d\n", read_mode,
+ failrec->this_mirror, num_copies, failrec->in_validation);
+
+ tree->ops->submit_bio_hook(inode, read_mode, bio, failrec->this_mirror,
+ failrec->bio_flags, 0);
+ return 0;
+ }
+
/* lots and lots of room for performance fixes in the end_bio funcs */
/*
struct extent_state *cached = NULL;
struct extent_state *state;
+ pr_debug("end_bio_extent_readpage: bi_vcnt=%d, idx=%d, err=%d, "
+ "mirror=%ld\n", bio->bi_vcnt, bio->bi_idx, err,
+ (long int)bio->bi_bdev);
tree = &BTRFS_I(page->mapping->host)->io_tree;
start = ((u64)page->index << PAGE_CACHE_SHIFT) +
state);
if (ret)
uptodate = 0;
+ else
+ clean_io_failure(start, page);
}
- if (!uptodate && tree->ops &&
- tree->ops->readpage_io_failed_hook) {
- ret = tree->ops->readpage_io_failed_hook(bio, page,
- start, end, state);
+ if (!uptodate) {
+ u64 failed_mirror;
+ failed_mirror = (u64)bio->bi_bdev;
+ if (tree->ops && tree->ops->readpage_io_failed_hook)
+ ret = tree->ops->readpage_io_failed_hook(
+ bio, page, start, end,
- failed_mirror, NULL);
++ failed_mirror, state);
+ else
+ ret = bio_readpage_error(bio, page, start, end,
+ failed_mirror, NULL);
if (ret == 0) {
uptodate =
test_bit(BIO_UPTODATE, &bio->bi_flags);
mirror_num, bio_flags, start);
else
submit_bio(rw, bio);
+
if (bio_flagged(bio, BIO_EOPNOTSUPP))
ret = -EOPNOTSUPP;
bio_put(bio);
}
int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
- get_extent_t *get_extent)
+ get_extent_t *get_extent, int mirror_num)
{
struct bio *bio = NULL;
unsigned long bio_flags = 0;
int ret;
- ret = __extent_read_full_page(tree, page, get_extent, &bio, 0,
+ ret = __extent_read_full_page(tree, page, get_extent, &bio, mirror_num,
&bio_flags);
if (bio)
- ret = submit_one_bio(READ, bio, 0, bio_flags);
+ ret = submit_one_bio(READ, bio, mirror_num, bio_flags);
return ret;
}
int compressed;
int write_flags;
unsigned long nr_written = 0;
+ bool fill_delalloc = true;
if (wbc->sync_mode == WB_SYNC_ALL)
write_flags = WRITE_SYNC;
trace___extent_writepage(page, inode, wbc);
WARN_ON(!PageLocked(page));
+
+ ClearPageError(page);
+
pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
if (page->index > end_index ||
(page->index == end_index && !pg_offset)) {
set_page_extent_mapped(page);
+ if (!tree->ops || !tree->ops->fill_delalloc)
+ fill_delalloc = false;
+
delalloc_start = start;
delalloc_end = 0;
page_started = 0;
- if (!epd->extent_locked) {
+ if (!epd->extent_locked && fill_delalloc) {
u64 delalloc_to_write = 0;
/*
* make sure the wbc mapping index is at least updated
* swizzled back from swapper_space to tmpfs file
* mapping
*/
- if (tree->ops && tree->ops->write_cache_pages_lock_hook)
- tree->ops->write_cache_pages_lock_hook(page);
- else
- lock_page(page);
+ if (tree->ops &&
+ tree->ops->write_cache_pages_lock_hook) {
+ tree->ops->write_cache_pages_lock_hook(page,
+ data, flush_fn);
+ } else {
+ if (!trylock_page(page)) {
+ flush_fn(data);
+ lock_page(page);
+ }
+ }
if (unlikely(page->mapping != mapping)) {
unlock_page(page);
return ret;
}
- static inline struct page *extent_buffer_page(struct extent_buffer *eb,
+ inline struct page *extent_buffer_page(struct extent_buffer *eb,
unsigned long i)
{
struct page *p;
return p;
}
- static inline unsigned long num_extent_pages(u64 start, u64 len)
+ inline unsigned long num_extent_pages(u64 start, u64 len)
{
return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
(start >> PAGE_CACHE_SHIFT);
PAGECACHE_TAG_DIRTY);
}
spin_unlock_irq(&page->mapping->tree_lock);
+ ClearPageError(page);
unlock_page(page);
}
return 0;
}
int read_extent_buffer_pages(struct extent_io_tree *tree,
- struct extent_buffer *eb,
- u64 start, int wait,
+ struct extent_buffer *eb, u64 start, int wait,
get_extent_t *get_extent, int mirror_num)
{
unsigned long i;
num_pages = num_extent_pages(eb->start, eb->len);
for (i = start_i; i < num_pages; i++) {
page = extent_buffer_page(eb, i);
- if (!wait) {
+ if (wait == WAIT_NONE) {
if (!trylock_page(page))
goto unlock_exit;
} else {
if (bio)
submit_one_bio(READ, bio, mirror_num, bio_flags);
- if (ret || !wait)
+ if (ret || wait != WAIT_COMPLETE)
return ret;
for (i = start_i; i < num_pages; i++) {
#define EXTENT_NODATASUM (1 << 10)
#define EXTENT_DO_ACCOUNTING (1 << 11)
#define EXTENT_FIRST_DELALLOC (1 << 12)
-#define EXTENT_DAMAGED (1 << 13)
+#define EXTENT_NEED_WAIT (1 << 13)
++#define EXTENT_DAMAGED (1 << 14)
#define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
#define EXTENT_CTLBITS (EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC)
#define EXTENT_BUFFER_BLOCKING 1
#define EXTENT_BUFFER_DIRTY 2
#define EXTENT_BUFFER_CORRUPT 3
+#define EXTENT_BUFFER_READAHEAD 4 /* this got triggered by readahead */
/* these are flags for extent_clear_unlock_delalloc */
#define EXTENT_CLEAR_UNLOCK_PAGE 0x1
unsigned long bio_flags);
int (*readpage_io_hook)(struct page *page, u64 start, u64 end);
int (*readpage_io_failed_hook)(struct bio *bio, struct page *page,
- u64 start, u64 end,
+ u64 start, u64 end, u64 failed_mirror,
struct extent_state *state);
int (*writepage_io_failed_hook)(struct bio *bio, struct page *page,
u64 start, u64 end,
struct extent_state *other);
void (*split_extent_hook)(struct inode *inode,
struct extent_state *orig, u64 split);
- int (*write_cache_pages_lock_hook)(struct page *page);
+ int (*write_cache_pages_lock_hook)(struct page *page, void *data,
+ void (*flush_fn)(void *));
};
struct extent_io_tree {
int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask);
int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
- get_extent_t *get_extent);
+ get_extent_t *get_extent, int mirror_num);
int __init extent_io_init(void);
void extent_io_exit(void);
gfp_t mask);
int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
gfp_t mask);
+int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+ int bits, int clear_bits, gfp_t mask);
int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
struct extent_state **cached_state, gfp_t mask);
int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
u64 start, unsigned long len);
void free_extent_buffer(struct extent_buffer *eb);
+#define WAIT_NONE 0
+#define WAIT_COMPLETE 1
+#define WAIT_PAGE_LOCK 2
int read_extent_buffer_pages(struct extent_io_tree *tree,
struct extent_buffer *eb, u64 start, int wait,
get_extent_t *get_extent, int mirror_num);
+ unsigned long num_extent_pages(u64 start, u64 len);
+ struct page *extent_buffer_page(struct extent_buffer *eb, unsigned long i);
static inline void extent_buffer_get(struct extent_buffer *eb)
{
struct bio *
btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
gfp_t gfp_flags);
+
+ struct btrfs_mapping_tree;
+
+ int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start,
+ u64 length, u64 logical, struct page *page,
+ int mirror_num);
#endif
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
- #include "volumes.h"
#include "ordered-data.h"
#include "xattr.h"
#include "tree-log.h"
+ #include "volumes.h"
#include "compression.h"
#include "locking.h"
#include "free-space-cache.h"
(BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS))) {
WARN_ON(pages);
pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS);
- BUG_ON(!pages);
+ if (!pages) {
+ /* just bail out to the uncompressed code */
+ goto cont;
+ }
if (BTRFS_I(inode)->force_compress)
compress_type = BTRFS_I(inode)->force_compress;
will_compress = 1;
}
}
+cont:
if (start == 0) {
trans = btrfs_join_transaction(root);
BUG_ON(IS_ERR(trans));
}
BUG_ON(disk_num_bytes >
- btrfs_super_total_bytes(&root->fs_info->super_copy));
+ btrfs_super_total_bytes(root->fs_info->super_copy));
alloc_hint = get_extent_allocation_hint(inode, start, num_bytes);
btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
}
ret = 0;
out:
- if (nolock) {
- if (trans)
- btrfs_end_transaction_nolock(trans, root);
- } else {
+ if (root != root->fs_info->tree_root)
btrfs_delalloc_release_metadata(inode, ordered_extent->len);
- if (trans)
+ if (trans) {
+ if (nolock)
+ btrfs_end_transaction_nolock(trans, root);
+ else
btrfs_end_transaction(trans, root);
}
return btrfs_finish_ordered_io(page->mapping->host, start, end);
}
- /*
- * When IO fails, either with EIO or csum verification fails, we
- * try other mirrors that might have a good copy of the data. This
- * io_failure_record is used to record state as we go through all the
- * mirrors. If another mirror has good data, the page is set up to date
- * and things continue. If a good mirror can't be found, the original
- * bio end_io callback is called to indicate things have failed.
- */
- struct io_failure_record {
- struct page *page;
- u64 start;
- u64 len;
- u64 logical;
- unsigned long bio_flags;
- int last_mirror;
- };
-
- static int btrfs_io_failed_hook(struct bio *failed_bio,
- struct page *page, u64 start, u64 end,
- struct extent_state *state)
- {
- struct io_failure_record *failrec = NULL;
- u64 private;
- struct extent_map *em;
- struct inode *inode = page->mapping->host;
- struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- struct bio *bio;
- int num_copies;
- int ret;
- int rw;
- u64 logical;
-
- ret = get_state_private(failure_tree, start, &private);
- if (ret) {
- failrec = kmalloc(sizeof(*failrec), GFP_NOFS);
- if (!failrec)
- return -ENOMEM;
- failrec->start = start;
- failrec->len = end - start + 1;
- failrec->last_mirror = 0;
- failrec->bio_flags = 0;
-
- read_lock(&em_tree->lock);
- em = lookup_extent_mapping(em_tree, start, failrec->len);
- if (em->start > start || em->start + em->len < start) {
- free_extent_map(em);
- em = NULL;
- }
- read_unlock(&em_tree->lock);
-
- if (IS_ERR_OR_NULL(em)) {
- kfree(failrec);
- return -EIO;
- }
- logical = start - em->start;
- logical = em->block_start + logical;
- if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
- logical = em->block_start;
- failrec->bio_flags = EXTENT_BIO_COMPRESSED;
- extent_set_compress_type(&failrec->bio_flags,
- em->compress_type);
- }
- failrec->logical = logical;
- free_extent_map(em);
- set_extent_bits(failure_tree, start, end, EXTENT_LOCKED |
- EXTENT_DIRTY, GFP_NOFS);
- set_state_private(failure_tree, start,
- (u64)(unsigned long)failrec);
- } else {
- failrec = (struct io_failure_record *)(unsigned long)private;
- }
- num_copies = btrfs_num_copies(
- &BTRFS_I(inode)->root->fs_info->mapping_tree,
- failrec->logical, failrec->len);
- failrec->last_mirror++;
- if (!state) {
- spin_lock(&BTRFS_I(inode)->io_tree.lock);
- state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
- failrec->start,
- EXTENT_LOCKED);
- if (state && state->start != failrec->start)
- state = NULL;
- spin_unlock(&BTRFS_I(inode)->io_tree.lock);
- }
- if (!state || failrec->last_mirror > num_copies) {
- set_state_private(failure_tree, failrec->start, 0);
- clear_extent_bits(failure_tree, failrec->start,
- failrec->start + failrec->len - 1,
- EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
- kfree(failrec);
- return -EIO;
- }
- bio = bio_alloc(GFP_NOFS, 1);
- bio->bi_private = state;
- bio->bi_end_io = failed_bio->bi_end_io;
- bio->bi_sector = failrec->logical >> 9;
- bio->bi_bdev = failed_bio->bi_bdev;
- bio->bi_size = 0;
-
- bio_add_page(bio, page, failrec->len, start - page_offset(page));
- if (failed_bio->bi_rw & REQ_WRITE)
- rw = WRITE;
- else
- rw = READ;
-
- ret = BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
- failrec->last_mirror,
- failrec->bio_flags, 0);
- return ret;
- }
-
- /*
- * each time an IO finishes, we do a fast check in the IO failure tree
- * to see if we need to process or clean up an io_failure_record
- */
- static int btrfs_clean_io_failures(struct inode *inode, u64 start)
- {
- u64 private;
- u64 private_failure;
- struct io_failure_record *failure;
- int ret;
-
- private = 0;
- if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
- (u64)-1, 1, EXTENT_DIRTY, 0)) {
- ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
- start, &private_failure);
- if (ret == 0) {
- failure = (struct io_failure_record *)(unsigned long)
- private_failure;
- set_state_private(&BTRFS_I(inode)->io_failure_tree,
- failure->start, 0);
- clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
- failure->start,
- failure->start + failure->len - 1,
- EXTENT_DIRTY | EXTENT_LOCKED,
- GFP_NOFS);
- kfree(failure);
- }
- }
- return 0;
- }
-
/*
* when reads are done, we need to check csums to verify the data is correct
- * if there's a match, we allow the bio to finish. If not, we go through
- * the io_failure_record routines to find good copies
+ * if there's a match, we allow the bio to finish. If not, the code in
+ * extent_io.c will try to find good copies for us.
*/
static int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
struct extent_state *state)
kunmap_atomic(kaddr, KM_USER0);
good:
- /* if the io failure tree for this inode is non-empty,
- * check to see if we've recovered from a failed IO
- */
- btrfs_clean_io_failures(inode, start);
return 0;
zeroit:
up_read(&root->fs_info->cleanup_work_sem);
}
-/*
- * calculate extra metadata reservation when snapshotting a subvolume
- * contains orphan files.
- */
-void btrfs_orphan_pre_snapshot(struct btrfs_trans_handle *trans,
- struct btrfs_pending_snapshot *pending,
- u64 *bytes_to_reserve)
-{
- struct btrfs_root *root;
- struct btrfs_block_rsv *block_rsv;
- u64 num_bytes;
- int index;
-
- root = pending->root;
- if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
- return;
-
- block_rsv = root->orphan_block_rsv;
-
- /* orphan block reservation for the snapshot */
- num_bytes = block_rsv->size;
-
- /*
- * after the snapshot is created, COWing tree blocks may use more
- * space than it frees. So we should make sure there is enough
- * reserved space.
- */
- index = trans->transid & 0x1;
- if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
- num_bytes += block_rsv->size -
- (block_rsv->reserved + block_rsv->freed[index]);
- }
-
- *bytes_to_reserve += num_bytes;
-}
-
-void btrfs_orphan_post_snapshot(struct btrfs_trans_handle *trans,
- struct btrfs_pending_snapshot *pending)
-{
- struct btrfs_root *root = pending->root;
- struct btrfs_root *snap = pending->snap;
- struct btrfs_block_rsv *block_rsv;
- u64 num_bytes;
- int index;
- int ret;
-
- if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
- return;
-
- /* refill source subvolume's orphan block reservation */
- block_rsv = root->orphan_block_rsv;
- index = trans->transid & 0x1;
- if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
- num_bytes = block_rsv->size -
- (block_rsv->reserved + block_rsv->freed[index]);
- ret = btrfs_block_rsv_migrate(&pending->block_rsv,
- root->orphan_block_rsv,
- num_bytes);
- BUG_ON(ret);
- }
-
- /* setup orphan block reservation for the snapshot */
- block_rsv = btrfs_alloc_block_rsv(snap);
- BUG_ON(!block_rsv);
-
- btrfs_add_durable_block_rsv(root->fs_info, block_rsv);
- snap->orphan_block_rsv = block_rsv;
-
- num_bytes = root->orphan_block_rsv->size;
- ret = btrfs_block_rsv_migrate(&pending->block_rsv,
- block_rsv, num_bytes);
- BUG_ON(ret);
-
-#if 0
- /* insert orphan item for the snapshot */
- WARN_ON(!root->orphan_item_inserted);
- ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
- snap->root_key.objectid);
- BUG_ON(ret);
- snap->orphan_item_inserted = 1;
-#endif
-}
-
enum btrfs_orphan_cleanup_state {
ORPHAN_CLEANUP_STARTED = 1,
ORPHAN_CLEANUP_DONE = 2,
}
spin_unlock(&root->orphan_lock);
- if (block_rsv)
- btrfs_add_durable_block_rsv(root->fs_info, block_rsv);
-
/* grab metadata reservation from transaction handle */
if (reserve) {
ret = btrfs_orphan_reserve_metadata(trans, inode);
struct btrfs_key key, found_key;
struct btrfs_trans_handle *trans;
struct inode *inode;
+ u64 last_objectid = 0;
int ret = 0, nr_unlink = 0, nr_truncate = 0;
if (cmpxchg(&root->orphan_cleanup_state, 0, ORPHAN_CLEANUP_STARTED))
* crossing root thing. we store the inode number in the
* offset of the orphan item.
*/
+
+ if (found_key.offset == last_objectid) {
+ printk(KERN_ERR "btrfs: Error removing orphan entry, "
+ "stopping orphan cleanup\n");
+ ret = -EINVAL;
+ goto out;
+ }
+
+ last_objectid = found_key.offset;
+
found_key.objectid = found_key.offset;
found_key.type = BTRFS_INODE_ITEM_KEY;
found_key.offset = 0;
inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL);
- if (IS_ERR(inode)) {
- ret = PTR_ERR(inode);
+ ret = PTR_RET(inode);
+ if (ret && ret != -ESTALE)
goto out;
- }
/*
- * add this inode to the orphan list so btrfs_orphan_del does
- * the proper thing when we hit it
- */
- spin_lock(&root->orphan_lock);
- list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
- spin_unlock(&root->orphan_lock);
-
- /*
- * if this is a bad inode, means we actually succeeded in
- * removing the inode, but not the orphan record, which means
- * we need to manually delete the orphan since iput will just
- * do a destroy_inode
+ * Inode is already gone but the orphan item is still there,
+ * kill the orphan item.
*/
- if (is_bad_inode(inode)) {
- trans = btrfs_start_transaction(root, 0);
+ if (ret == -ESTALE) {
+ trans = btrfs_start_transaction(root, 1);
if (IS_ERR(trans)) {
ret = PTR_ERR(trans);
goto out;
}
- btrfs_orphan_del(trans, inode);
+ ret = btrfs_del_orphan_item(trans, root,
+ found_key.objectid);
+ BUG_ON(ret);
btrfs_end_transaction(trans, root);
- iput(inode);
continue;
}
+ /*
+ * add this inode to the orphan list so btrfs_orphan_del does
+ * the proper thing when we hit it
+ */
+ spin_lock(&root->orphan_lock);
+ list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
+ spin_unlock(&root->orphan_lock);
+
/* if we have links, this was a truncate, lets do that */
if (inode->i_nlink) {
if (!S_ISREG(inode->i_mode)) {
u64 ino = btrfs_ino(inode);
u64 dir_ino = btrfs_ino(dir);
- trans = btrfs_start_transaction(root, 10);
+ /*
+ * 1 for the possible orphan item
+ * 1 for the dir item
+ * 1 for the dir index
+ * 1 for the inode ref
+ * 1 for the inode ref in the tree log
+ * 2 for the dir entries in the log
+ * 1 for the inode
+ */
+ trans = btrfs_start_transaction(root, 8);
if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
return trans;
return ERR_PTR(-ENOMEM);
}
- trans = btrfs_start_transaction(root, 0);
+ /* 1 for the orphan item */
+ trans = btrfs_start_transaction(root, 1);
if (IS_ERR(trans)) {
btrfs_free_path(path);
root->fs_info->enospc_unlink = 0;
err = 0;
out:
btrfs_free_path(path);
+ /* Migrate the orphan reservation over */
+ if (!err)
+ err = btrfs_block_rsv_migrate(trans->block_rsv,
+ &root->fs_info->global_block_rsv,
+ trans->bytes_reserved);
+
if (err) {
btrfs_end_transaction(trans, root);
root->fs_info->enospc_unlink = 0;
struct btrfs_root *root)
{
if (trans->block_rsv == &root->fs_info->global_block_rsv) {
+ btrfs_block_rsv_release(root, trans->block_rsv,
+ trans->bytes_reserved);
+ trans->block_rsv = &root->fs_info->trans_block_rsv;
BUG_ON(!root->fs_info->enospc_unlink);
root->fs_info->enospc_unlink = 0;
}
pgoff_t index = from >> PAGE_CACHE_SHIFT;
unsigned offset = from & (PAGE_CACHE_SIZE-1);
struct page *page;
+ gfp_t mask = btrfs_alloc_write_mask(mapping);
int ret = 0;
u64 page_start;
u64 page_end;
ret = -ENOMEM;
again:
- page = find_or_create_page(mapping, index, GFP_NOFS);
+ page = find_or_create_page(mapping, index, mask);
if (!page) {
btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE);
goto out;
{
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_block_rsv *rsv, *global_rsv;
+ u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
unsigned long nr;
int ret;
goto no_delete;
}
+ rsv = btrfs_alloc_block_rsv(root);
+ if (!rsv) {
+ btrfs_orphan_del(NULL, inode);
+ goto no_delete;
+ }
+ rsv->size = min_size;
+ global_rsv = &root->fs_info->global_block_rsv;
+
btrfs_i_size_write(inode, 0);
+ /*
+ * This is a bit simpler than btrfs_truncate since
+ *
+ * 1) We've already reserved our space for our orphan item in the
+ * unlink.
+ * 2) We're going to delete the inode item, so we don't need to update
+ * it at all.
+ *
+ * So we just need to reserve some slack space in case we add bytes when
+ * doing the truncate.
+ */
while (1) {
- trans = btrfs_join_transaction(root);
- BUG_ON(IS_ERR(trans));
- trans->block_rsv = root->orphan_block_rsv;
+ ret = btrfs_block_rsv_refill(root, rsv, min_size);
+
+ /*
+ * Try and steal from the global reserve since we will
+ * likely not use this space anyway, we want to try as
+ * hard as possible to get this to work.
+ */
+ if (ret)
+ ret = btrfs_block_rsv_migrate(global_rsv, rsv, min_size);
- ret = btrfs_block_rsv_check(trans, root,
- root->orphan_block_rsv, 0, 5);
if (ret) {
- BUG_ON(ret != -EAGAIN);
- ret = btrfs_commit_transaction(trans, root);
- BUG_ON(ret);
- continue;
+ printk(KERN_WARNING "Could not get space for a "
+ "delete, will truncate on mount %d\n", ret);
+ btrfs_orphan_del(NULL, inode);
+ btrfs_free_block_rsv(root, rsv);
+ goto no_delete;
+ }
+
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ btrfs_orphan_del(NULL, inode);
+ btrfs_free_block_rsv(root, rsv);
+ goto no_delete;
}
+ trans->block_rsv = rsv;
+
ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
if (ret != -EAGAIN)
break;
btrfs_end_transaction(trans, root);
trans = NULL;
btrfs_btree_balance_dirty(root, nr);
-
}
+ btrfs_free_block_rsv(root, rsv);
+
if (ret == 0) {
+ trans->block_rsv = root->orphan_block_rsv;
ret = btrfs_orphan_del(trans, inode);
BUG_ON(ret);
}
+ trans->block_rsv = &root->fs_info->trans_block_rsv;
if (!(root == root->fs_info->tree_root ||
root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID))
btrfs_return_ino(root, btrfs_ino(inode));
if (test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags)) {
ret = btrfs_ordered_update_i_size(inode, 0, ordered);
if (!ret)
- ret = btrfs_update_inode(trans, root, inode);
- err = ret;
+ err = btrfs_update_inode(trans, root, inode);
goto out;
}
{
struct extent_io_tree *tree;
tree = &BTRFS_I(page->mapping->host)->io_tree;
- return extent_read_full_page(tree, page, btrfs_get_extent);
+ return extent_read_full_page(tree, page, btrfs_get_extent, 0);
}
static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
struct btrfs_trans_handle *trans;
unsigned long nr;
u64 mask = root->sectorsize - 1;
+ u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
ret = btrfs_truncate_page(inode->i_mapping, inode->i_size);
if (ret)
rsv = btrfs_alloc_block_rsv(root);
if (!rsv)
return -ENOMEM;
- btrfs_add_durable_block_rsv(root->fs_info, rsv);
+ rsv->size = min_size;
+ /*
+ * 1 for the truncate slack space
+ * 1 for the orphan item we're going to add
+ * 1 for the orphan item deletion
+ * 1 for updating the inode.
+ */
trans = btrfs_start_transaction(root, 4);
if (IS_ERR(trans)) {
err = PTR_ERR(trans);
goto out;
}
- /*
- * Reserve space for the truncate process. Truncate should be adding
- * space, but if there are snapshots it may end up using space.
- */
- ret = btrfs_truncate_reserve_metadata(trans, root, rsv);
+ /* Migrate the slack space for the truncate to our reserve */
+ ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv,
+ min_size);
BUG_ON(ret);
ret = btrfs_orphan_add(trans, inode);
goto out;
}
- nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- btrfs_btree_balance_dirty(root, nr);
-
- /*
- * Ok so we've already migrated our bytes over for the truncate, so here
- * just reserve the one slot we need for updating the inode.
- */
- trans = btrfs_start_transaction(root, 1);
- if (IS_ERR(trans)) {
- err = PTR_ERR(trans);
- goto out;
- }
- trans->block_rsv = rsv;
-
/*
* setattr is responsible for setting the ordered_data_close flag,
* but that is only tested during the last file release. That
btrfs_add_ordered_operation(trans, root, inode);
while (1) {
+ ret = btrfs_block_rsv_refill(root, rsv, min_size);
+ if (ret) {
+ /*
+ * This can only happen with the original transaction we
+ * started above, every other time we shouldn't have a
+ * transaction started yet.
+ */
+ if (ret == -EAGAIN)
+ goto end_trans;
+ err = ret;
+ break;
+ }
+
if (!trans) {
- trans = btrfs_start_transaction(root, 3);
+ /* Just need the 1 for updating the inode */
+ trans = btrfs_start_transaction(root, 1);
if (IS_ERR(trans)) {
err = PTR_ERR(trans);
goto out;
}
-
- ret = btrfs_truncate_reserve_metadata(trans, root,
- rsv);
- BUG_ON(ret);
-
- trans->block_rsv = rsv;
}
+ trans->block_rsv = rsv;
+
ret = btrfs_truncate_inode_items(trans, root, inode,
inode->i_size,
BTRFS_EXTENT_DATA_KEY);
err = ret;
break;
}
-
+end_trans:
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
trans = NULL;
ei->last_sub_trans = 0;
ei->logged_trans = 0;
ei->delalloc_bytes = 0;
- ei->reserved_bytes = 0;
ei->disk_i_size = 0;
ei->flags = 0;
+ ei->csum_bytes = 0;
ei->index_cnt = (u64)-1;
ei->last_unlink_trans = 0;
WARN_ON(inode->i_data.nrpages);
WARN_ON(BTRFS_I(inode)->outstanding_extents);
WARN_ON(BTRFS_I(inode)->reserved_extents);
+ WARN_ON(BTRFS_I(inode)->delalloc_bytes);
+ WARN_ON(BTRFS_I(inode)->csum_bytes);
/*
* This can happen where we create an inode, but somebody else also
.readpage_end_io_hook = btrfs_readpage_end_io_hook,
.writepage_end_io_hook = btrfs_writepage_end_io_hook,
.writepage_start_hook = btrfs_writepage_start_hook,
- .readpage_io_failed_hook = btrfs_io_failed_hook,
.set_bit_hook = btrfs_set_bit_hook,
.clear_bit_hook = btrfs_clear_bit_hook,
.merge_extent_hook = btrfs_merge_extent_hook,
#include "volumes.h"
#include "locking.h"
#include "inode-map.h"
+ #include "backref.h"
/* Mask out flags that are inappropriate for the given type of inode. */
static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
/*
* Inherit flags from the parent inode.
*
- * Unlike extN we don't have any flags we don't want to inherit currently.
+ * Currently only the compression flags and the cow flags are inherited.
*/
void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
{
flags = BTRFS_I(dir)->flags;
- if (S_ISREG(inode->i_mode))
- flags &= ~BTRFS_INODE_DIRSYNC;
- else if (!S_ISDIR(inode->i_mode))
- flags &= (BTRFS_INODE_NODUMP | BTRFS_INODE_NOATIME);
+ if (flags & BTRFS_INODE_NOCOMPRESS) {
+ BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
+ BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
+ } else if (flags & BTRFS_INODE_COMPRESS) {
+ BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS;
+ BTRFS_I(inode)->flags |= BTRFS_INODE_COMPRESS;
+ }
+
+ if (flags & BTRFS_INODE_NODATACOW)
+ BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
- BTRFS_I(inode)->flags = flags;
btrfs_update_iflags(inode);
}
struct fstrim_range range;
u64 minlen = ULLONG_MAX;
u64 num_devices = 0;
+ u64 total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
int ret;
if (!capable(CAP_SYS_ADMIN))
}
}
rcu_read_unlock();
+
if (!num_devices)
return -EOPNOTSUPP;
-
if (copy_from_user(&range, arg, sizeof(range)))
return -EFAULT;
+ if (range.start > total_bytes)
+ return -EINVAL;
+ range.len = min(range.len, total_bytes - range.start);
range.minlen = max(range.minlen, minlen);
ret = btrfs_trim_fs(root, &range);
if (ret < 0)
int ret = 1;
/*
- * make sure that once we start defragging and extent, we keep on
+ * make sure that once we start defragging an extent, we keep on
* defragging it
*/
if (start < *defrag_end)
* extent will force at least part of that big extent to be defragged.
*/
if (ret) {
- *last_len += len;
*defrag_end = extent_map_end(em);
} else {
*last_len = 0;
int i_done;
struct btrfs_ordered_extent *ordered;
struct extent_state *cached_state = NULL;
+ gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
if (isize == 0)
return 0;
for (i = 0; i < num_pages; i++) {
struct page *page;
page = find_or_create_page(inode->i_mapping,
- start_index + i, GFP_NOFS);
+ start_index + i, mask);
if (!page)
break;
struct btrfs_super_block *disk_super;
struct file_ra_state *ra = NULL;
unsigned long last_index;
+ u64 isize = i_size_read(inode);
u64 features;
u64 last_len = 0;
u64 skip = 0;
u64 defrag_end = 0;
u64 newer_off = range->start;
- int newer_left = 0;
unsigned long i;
+ unsigned long ra_index = 0;
int ret;
int defrag_count = 0;
int compress_type = BTRFS_COMPRESS_ZLIB;
int extent_thresh = range->extent_thresh;
- int newer_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
+ int max_cluster = (256 * 1024) >> PAGE_CACHE_SHIFT;
+ int cluster = max_cluster;
u64 new_align = ~((u64)128 * 1024 - 1);
struct page **pages = NULL;
compress_type = range->compress_type;
}
- if (inode->i_size == 0)
+ if (isize == 0)
return 0;
/*
ra = &file->f_ra;
}
- pages = kmalloc(sizeof(struct page *) * newer_cluster,
+ pages = kmalloc(sizeof(struct page *) * max_cluster,
GFP_NOFS);
if (!pages) {
ret = -ENOMEM;
/* find the last page to defrag */
if (range->start + range->len > range->start) {
- last_index = min_t(u64, inode->i_size - 1,
+ last_index = min_t(u64, isize - 1,
range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
} else {
- last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
+ last_index = (isize - 1) >> PAGE_CACHE_SHIFT;
}
if (newer_than) {
* the extents in the file evenly spaced
*/
i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
- newer_left = newer_cluster;
} else
goto out_ra;
} else {
i = range->start >> PAGE_CACHE_SHIFT;
}
if (!max_to_defrag)
- max_to_defrag = last_index - 1;
+ max_to_defrag = last_index;
+
+ /*
+ * make writeback starts from i, so the defrag range can be
+ * written sequentially.
+ */
+ if (i < inode->i_mapping->writeback_index)
+ inode->i_mapping->writeback_index = i;
- while (i <= last_index && defrag_count < max_to_defrag) {
+ while (i <= last_index && defrag_count < max_to_defrag &&
+ (i < (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
+ PAGE_CACHE_SHIFT)) {
/*
* make sure we stop running if someone unmounts
* the FS
i = max(i + 1, next);
continue;
}
+
+ if (!newer_than) {
+ cluster = (PAGE_CACHE_ALIGN(defrag_end) >>
+ PAGE_CACHE_SHIFT) - i;
+ cluster = min(cluster, max_cluster);
+ } else {
+ cluster = max_cluster;
+ }
+
if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
BTRFS_I(inode)->force_compress = compress_type;
- btrfs_force_ra(inode->i_mapping, ra, file, i, newer_cluster);
+ if (i + cluster > ra_index) {
+ ra_index = max(i, ra_index);
+ btrfs_force_ra(inode->i_mapping, ra, file, ra_index,
+ cluster);
+ ra_index += max_cluster;
+ }
- ret = cluster_pages_for_defrag(inode, pages, i, newer_cluster);
+ ret = cluster_pages_for_defrag(inode, pages, i, cluster);
if (ret < 0)
goto out_ra;
defrag_count += ret;
balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
- i += ret;
if (newer_than) {
if (newer_off == (u64)-1)
if (!ret) {
range->start = newer_off;
i = (newer_off & new_align) >> PAGE_CACHE_SHIFT;
- newer_left = newer_cluster;
} else {
break;
}
} else {
- i++;
+ if (ret > 0) {
+ i += ret;
+ last_len += ret << PAGE_CACHE_SHIFT;
+ } else {
+ i++;
+ last_len = 0;
+ }
}
}
mutex_unlock(&inode->i_mutex);
}
- disk_super = &root->fs_info->super_copy;
+ disk_super = root->fs_info->super_copy;
features = btrfs_super_incompat_flags(disk_super);
if (range->compress_type == BTRFS_COMPRESS_LZO) {
features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
btrfs_set_super_incompat_flags(disk_super, features);
}
- if (!file)
- kfree(ra);
- return defrag_count;
+ ret = defrag_count;
out_ra:
if (!file)
return PTR_ERR(trans);
}
- dir_id = btrfs_super_root_dir(&root->fs_info->super_copy);
+ dir_id = btrfs_super_root_dir(root->fs_info->super_copy);
di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path,
dir_id, "default", 7, 1);
if (IS_ERR_OR_NULL(di)) {
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_free_path(path);
- disk_super = &root->fs_info->super_copy;
+ disk_super = root->fs_info->super_copy;
features = btrfs_super_incompat_flags(disk_super);
if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) {
features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL;
return ret;
}
- struct btrfs_ioctl_ino_path_args *ipa;
+ static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
+ {
+ int ret = 0;
+ int i;
+ unsigned long rel_ptr;
+ int size;
++ struct btrfs_ioctl_ino_path_args *ipa = NULL;
+ struct inode_fs_paths *ipath = NULL;
+ struct btrfs_path *path;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ path = btrfs_alloc_path();
+ if (!path) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ ipa = memdup_user(arg, sizeof(*ipa));
+ if (IS_ERR(ipa)) {
+ ret = PTR_ERR(ipa);
+ ipa = NULL;
+ goto out;
+ }
+
+ size = min_t(u32, ipa->size, 4096);
+ ipath = init_ipath(size, root, path);
+ if (IS_ERR(ipath)) {
+ ret = PTR_ERR(ipath);
+ ipath = NULL;
+ goto out;
+ }
+
+ ret = paths_from_inode(ipa->inum, ipath);
+ if (ret < 0)
+ goto out;
+
+ for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
+ rel_ptr = ipath->fspath->str[i] - (char *)ipath->fspath->str;
+ ipath->fspath->str[i] = (void *)rel_ptr;
+ }
+
+ ret = copy_to_user(ipa->fspath, ipath->fspath, size);
+ if (ret) {
+ ret = -EFAULT;
+ goto out;
+ }
+
+ out:
+ btrfs_free_path(path);
+ free_ipath(ipath);
+ kfree(ipa);
+
+ return ret;
+ }
+
+ static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
+ {
+ struct btrfs_data_container *inodes = ctx;
+ const size_t c = 3 * sizeof(u64);
+
+ if (inodes->bytes_left >= c) {
+ inodes->bytes_left -= c;
+ inodes->val[inodes->elem_cnt] = inum;
+ inodes->val[inodes->elem_cnt + 1] = offset;
+ inodes->val[inodes->elem_cnt + 2] = root;
+ inodes->elem_cnt += 3;
+ } else {
+ inodes->bytes_missing += c - inodes->bytes_left;
+ inodes->bytes_left = 0;
+ inodes->elem_missed += 3;
+ }
+
+ return 0;
+ }
+
+ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
+ void __user *arg)
+ {
+ int ret = 0;
+ int size;
+ u64 extent_offset;
+ struct btrfs_ioctl_logical_ino_args *loi;
+ struct btrfs_data_container *inodes = NULL;
+ struct btrfs_path *path = NULL;
+ struct btrfs_key key;
+
+ if (!capable(CAP_SYS_ADMIN))
+ return -EPERM;
+
+ loi = memdup_user(arg, sizeof(*loi));
+ if (IS_ERR(loi)) {
+ ret = PTR_ERR(loi);
+ loi = NULL;
+ goto out;
+ }
+
+ path = btrfs_alloc_path();
+ if (!path) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ size = min_t(u32, loi->size, 4096);
+ inodes = init_data_container(size);
+ if (IS_ERR(inodes)) {
+ ret = PTR_ERR(inodes);
+ inodes = NULL;
+ goto out;
+ }
+
+ ret = extent_from_logical(root->fs_info, loi->logical, path, &key);
+
+ if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+ ret = -ENOENT;
+ if (ret < 0)
+ goto out;
+
+ extent_offset = loi->logical - key.objectid;
+ ret = iterate_extent_inodes(root->fs_info, path, key.objectid,
+ extent_offset, build_ino_list, inodes);
+
+ if (ret < 0)
+ goto out;
+
+ ret = copy_to_user(loi->inodes, inodes, size);
+ if (ret)
+ ret = -EFAULT;
+
+ out:
+ btrfs_free_path(path);
+ kfree(inodes);
+ kfree(loi);
+
+ return ret;
+ }
+
long btrfs_ioctl(struct file *file, unsigned int
cmd, unsigned long arg)
{
return btrfs_ioctl_tree_search(file, argp);
case BTRFS_IOC_INO_LOOKUP:
return btrfs_ioctl_ino_lookup(file, argp);
+ case BTRFS_IOC_INO_PATHS:
+ return btrfs_ioctl_ino_to_path(root, argp);
+ case BTRFS_IOC_LOGICAL_INO:
+ return btrfs_ioctl_logical_to_ino(root, argp);
case BTRFS_IOC_SPACE_INFO:
return btrfs_ioctl_space_info(root, argp);
case BTRFS_IOC_SYNC:
--- /dev/null
- struct btrfs_multi_bio *multi)
+/*
+ * Copyright (C) 2011 STRATO. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include <linux/sched.h>
+#include <linux/pagemap.h>
+#include <linux/writeback.h>
+#include <linux/blkdev.h>
+#include <linux/rbtree.h>
+#include <linux/slab.h>
+#include <linux/workqueue.h>
+#include "ctree.h"
+#include "volumes.h"
+#include "disk-io.h"
+#include "transaction.h"
+
+#undef DEBUG
+
+/*
+ * This is the implementation for the generic read ahead framework.
+ *
+ * To trigger a readahead, btrfs_reada_add must be called. It will start
+ * a read ahead for the given range [start, end) on tree root. The returned
+ * handle can either be used to wait on the readahead to finish
+ * (btrfs_reada_wait), or to send it to the background (btrfs_reada_detach).
+ *
+ * The read ahead works as follows:
+ * On btrfs_reada_add, the root of the tree is inserted into a radix_tree.
+ * reada_start_machine will then search for extents to prefetch and trigger
+ * some reads. When a read finishes for a node, all contained node/leaf
+ * pointers that lie in the given range will also be enqueued. The reads will
+ * be triggered in sequential order, thus giving a big win over a naive
+ * enumeration. It will also make use of multi-device layouts. Each disk
+ * will have its on read pointer and all disks will by utilized in parallel.
+ * Also will no two disks read both sides of a mirror simultaneously, as this
+ * would waste seeking capacity. Instead both disks will read different parts
+ * of the filesystem.
+ * Any number of readaheads can be started in parallel. The read order will be
+ * determined globally, i.e. 2 parallel readaheads will normally finish faster
+ * than the 2 started one after another.
+ */
+
+#define MAX_MIRRORS 2
+#define MAX_IN_FLIGHT 6
+
+struct reada_extctl {
+ struct list_head list;
+ struct reada_control *rc;
+ u64 generation;
+};
+
+struct reada_extent {
+ u64 logical;
+ struct btrfs_key top;
+ u32 blocksize;
+ int err;
+ struct list_head extctl;
+ struct kref refcnt;
+ spinlock_t lock;
+ struct reada_zone *zones[MAX_MIRRORS];
+ int nzones;
+ struct btrfs_device *scheduled_for;
+};
+
+struct reada_zone {
+ u64 start;
+ u64 end;
+ u64 elems;
+ struct list_head list;
+ spinlock_t lock;
+ int locked;
+ struct btrfs_device *device;
+ struct btrfs_device *devs[MAX_MIRRORS]; /* full list, incl self */
+ int ndevs;
+ struct kref refcnt;
+};
+
+struct reada_machine_work {
+ struct btrfs_work work;
+ struct btrfs_fs_info *fs_info;
+};
+
+static void reada_extent_put(struct btrfs_fs_info *, struct reada_extent *);
+static void reada_control_release(struct kref *kref);
+static void reada_zone_release(struct kref *kref);
+static void reada_start_machine(struct btrfs_fs_info *fs_info);
+static void __reada_start_machine(struct btrfs_fs_info *fs_info);
+
+static int reada_add_block(struct reada_control *rc, u64 logical,
+ struct btrfs_key *top, int level, u64 generation);
+
+/* recurses */
+/* in case of err, eb might be NULL */
+static int __readahead_hook(struct btrfs_root *root, struct extent_buffer *eb,
+ u64 start, int err)
+{
+ int level = 0;
+ int nritems;
+ int i;
+ u64 bytenr;
+ u64 generation;
+ struct reada_extent *re;
+ struct btrfs_fs_info *fs_info = root->fs_info;
+ struct list_head list;
+ unsigned long index = start >> PAGE_CACHE_SHIFT;
+ struct btrfs_device *for_dev;
+
+ if (eb)
+ level = btrfs_header_level(eb);
+
+ /* find extent */
+ spin_lock(&fs_info->reada_lock);
+ re = radix_tree_lookup(&fs_info->reada_tree, index);
+ if (re)
+ kref_get(&re->refcnt);
+ spin_unlock(&fs_info->reada_lock);
+
+ if (!re)
+ return -1;
+
+ spin_lock(&re->lock);
+ /*
+ * just take the full list from the extent. afterwards we
+ * don't need the lock anymore
+ */
+ list_replace_init(&re->extctl, &list);
+ for_dev = re->scheduled_for;
+ re->scheduled_for = NULL;
+ spin_unlock(&re->lock);
+
+ if (err == 0) {
+ nritems = level ? btrfs_header_nritems(eb) : 0;
+ generation = btrfs_header_generation(eb);
+ /*
+ * FIXME: currently we just set nritems to 0 if this is a leaf,
+ * effectively ignoring the content. In a next step we could
+ * trigger more readahead depending from the content, e.g.
+ * fetch the checksums for the extents in the leaf.
+ */
+ } else {
+ /*
+ * this is the error case, the extent buffer has not been
+ * read correctly. We won't access anything from it and
+ * just cleanup our data structures. Effectively this will
+ * cut the branch below this node from read ahead.
+ */
+ nritems = 0;
+ generation = 0;
+ }
+
+ for (i = 0; i < nritems; i++) {
+ struct reada_extctl *rec;
+ u64 n_gen;
+ struct btrfs_key key;
+ struct btrfs_key next_key;
+
+ btrfs_node_key_to_cpu(eb, &key, i);
+ if (i + 1 < nritems)
+ btrfs_node_key_to_cpu(eb, &next_key, i + 1);
+ else
+ next_key = re->top;
+ bytenr = btrfs_node_blockptr(eb, i);
+ n_gen = btrfs_node_ptr_generation(eb, i);
+
+ list_for_each_entry(rec, &list, list) {
+ struct reada_control *rc = rec->rc;
+
+ /*
+ * if the generation doesn't match, just ignore this
+ * extctl. This will probably cut off a branch from
+ * prefetch. Alternatively one could start a new (sub-)
+ * prefetch for this branch, starting again from root.
+ * FIXME: move the generation check out of this loop
+ */
+#ifdef DEBUG
+ if (rec->generation != generation) {
+ printk(KERN_DEBUG "generation mismatch for "
+ "(%llu,%d,%llu) %llu != %llu\n",
+ key.objectid, key.type, key.offset,
+ rec->generation, generation);
+ }
+#endif
+ if (rec->generation == generation &&
+ btrfs_comp_cpu_keys(&key, &rc->key_end) < 0 &&
+ btrfs_comp_cpu_keys(&next_key, &rc->key_start) > 0)
+ reada_add_block(rc, bytenr, &next_key,
+ level - 1, n_gen);
+ }
+ }
+ /*
+ * free extctl records
+ */
+ while (!list_empty(&list)) {
+ struct reada_control *rc;
+ struct reada_extctl *rec;
+
+ rec = list_first_entry(&list, struct reada_extctl, list);
+ list_del(&rec->list);
+ rc = rec->rc;
+ kfree(rec);
+
+ kref_get(&rc->refcnt);
+ if (atomic_dec_and_test(&rc->elems)) {
+ kref_put(&rc->refcnt, reada_control_release);
+ wake_up(&rc->wait);
+ }
+ kref_put(&rc->refcnt, reada_control_release);
+
+ reada_extent_put(fs_info, re); /* one ref for each entry */
+ }
+ reada_extent_put(fs_info, re); /* our ref */
+ if (for_dev)
+ atomic_dec(&for_dev->reada_in_flight);
+
+ return 0;
+}
+
+/*
+ * start is passed separately in case eb in NULL, which may be the case with
+ * failed I/O
+ */
+int btree_readahead_hook(struct btrfs_root *root, struct extent_buffer *eb,
+ u64 start, int err)
+{
+ int ret;
+
+ ret = __readahead_hook(root, eb, start, err);
+
+ reada_start_machine(root->fs_info);
+
+ return ret;
+}
+
+static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
+ struct btrfs_device *dev, u64 logical,
- struct btrfs_multi_bio *multi = NULL;
++ struct btrfs_bio *multi)
+{
+ int ret;
+ int looped = 0;
+ struct reada_zone *zone;
+ struct btrfs_block_group_cache *cache = NULL;
+ u64 start;
+ u64 end;
+ int i;
+
+again:
+ zone = NULL;
+ spin_lock(&fs_info->reada_lock);
+ ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
+ logical >> PAGE_CACHE_SHIFT, 1);
+ if (ret == 1)
+ kref_get(&zone->refcnt);
+ spin_unlock(&fs_info->reada_lock);
+
+ if (ret == 1) {
+ if (logical >= zone->start && logical < zone->end)
+ return zone;
+ spin_lock(&fs_info->reada_lock);
+ kref_put(&zone->refcnt, reada_zone_release);
+ spin_unlock(&fs_info->reada_lock);
+ }
+
+ if (looped)
+ return NULL;
+
+ cache = btrfs_lookup_block_group(fs_info, logical);
+ if (!cache)
+ return NULL;
+
+ start = cache->key.objectid;
+ end = start + cache->key.offset - 1;
+ btrfs_put_block_group(cache);
+
+ zone = kzalloc(sizeof(*zone), GFP_NOFS);
+ if (!zone)
+ return NULL;
+
+ zone->start = start;
+ zone->end = end;
+ INIT_LIST_HEAD(&zone->list);
+ spin_lock_init(&zone->lock);
+ zone->locked = 0;
+ kref_init(&zone->refcnt);
+ zone->elems = 0;
+ zone->device = dev; /* our device always sits at index 0 */
+ for (i = 0; i < multi->num_stripes; ++i) {
+ /* bounds have already been checked */
+ zone->devs[i] = multi->stripes[i].dev;
+ }
+ zone->ndevs = multi->num_stripes;
+
+ spin_lock(&fs_info->reada_lock);
+ ret = radix_tree_insert(&dev->reada_zones,
+ (unsigned long)zone->end >> PAGE_CACHE_SHIFT,
+ zone);
+ spin_unlock(&fs_info->reada_lock);
+
+ if (ret) {
+ kfree(zone);
+ looped = 1;
+ goto again;
+ }
+
+ return zone;
+}
+
+static struct reada_extent *reada_find_extent(struct btrfs_root *root,
+ u64 logical,
+ struct btrfs_key *top, int level)
+{
+ int ret;
+ int looped = 0;
+ struct reada_extent *re = NULL;
+ struct btrfs_fs_info *fs_info = root->fs_info;
+ struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
++ struct btrfs_bio *multi = NULL;
+ struct btrfs_device *dev;
+ u32 blocksize;
+ u64 length;
+ int nzones = 0;
+ int i;
+ unsigned long index = logical >> PAGE_CACHE_SHIFT;
+
+again:
+ spin_lock(&fs_info->reada_lock);
+ re = radix_tree_lookup(&fs_info->reada_tree, index);
+ if (re)
+ kref_get(&re->refcnt);
+ spin_unlock(&fs_info->reada_lock);
+
+ if (re || looped)
+ return re;
+
+ re = kzalloc(sizeof(*re), GFP_NOFS);
+ if (!re)
+ return NULL;
+
+ blocksize = btrfs_level_size(root, level);
+ re->logical = logical;
+ re->blocksize = blocksize;
+ re->top = *top;
+ INIT_LIST_HEAD(&re->extctl);
+ spin_lock_init(&re->lock);
+ kref_init(&re->refcnt);
+
+ /*
+ * map block
+ */
+ length = blocksize;
+ ret = btrfs_map_block(map_tree, REQ_WRITE, logical, &length, &multi, 0);
+ if (ret || !multi || length < blocksize)
+ goto error;
+
+ if (multi->num_stripes > MAX_MIRRORS) {
+ printk(KERN_ERR "btrfs readahead: more than %d copies not "
+ "supported", MAX_MIRRORS);
+ goto error;
+ }
+
+ for (nzones = 0; nzones < multi->num_stripes; ++nzones) {
+ struct reada_zone *zone;
+
+ dev = multi->stripes[nzones].dev;
+ zone = reada_find_zone(fs_info, dev, logical, multi);
+ if (!zone)
+ break;
+
+ re->zones[nzones] = zone;
+ spin_lock(&zone->lock);
+ if (!zone->elems)
+ kref_get(&zone->refcnt);
+ ++zone->elems;
+ spin_unlock(&zone->lock);
+ spin_lock(&fs_info->reada_lock);
+ kref_put(&zone->refcnt, reada_zone_release);
+ spin_unlock(&fs_info->reada_lock);
+ }
+ re->nzones = nzones;
+ if (nzones == 0) {
+ /* not a single zone found, error and out */
+ goto error;
+ }
+
+ /* insert extent in reada_tree + all per-device trees, all or nothing */
+ spin_lock(&fs_info->reada_lock);
+ ret = radix_tree_insert(&fs_info->reada_tree, index, re);
+ if (ret) {
+ spin_unlock(&fs_info->reada_lock);
+ if (ret != -ENOMEM) {
+ /* someone inserted the extent in the meantime */
+ looped = 1;
+ }
+ goto error;
+ }
+ for (i = 0; i < nzones; ++i) {
+ dev = multi->stripes[i].dev;
+ ret = radix_tree_insert(&dev->reada_extents, index, re);
+ if (ret) {
+ while (--i >= 0) {
+ dev = multi->stripes[i].dev;
+ BUG_ON(dev == NULL);
+ radix_tree_delete(&dev->reada_extents, index);
+ }
+ BUG_ON(fs_info == NULL);
+ radix_tree_delete(&fs_info->reada_tree, index);
+ spin_unlock(&fs_info->reada_lock);
+ goto error;
+ }
+ }
+ spin_unlock(&fs_info->reada_lock);
+
+ return re;
+
+error:
+ while (nzones) {
+ struct reada_zone *zone;
+
+ --nzones;
+ zone = re->zones[nzones];
+ kref_get(&zone->refcnt);
+ spin_lock(&zone->lock);
+ --zone->elems;
+ if (zone->elems == 0) {
+ /*
+ * no fs_info->reada_lock needed, as this can't be
+ * the last ref
+ */
+ kref_put(&zone->refcnt, reada_zone_release);
+ }
+ spin_unlock(&zone->lock);
+
+ spin_lock(&fs_info->reada_lock);
+ kref_put(&zone->refcnt, reada_zone_release);
+ spin_unlock(&fs_info->reada_lock);
+ }
+ kfree(re);
+ if (looped)
+ goto again;
+ return NULL;
+}
+
+static void reada_kref_dummy(struct kref *kr)
+{
+}
+
+static void reada_extent_put(struct btrfs_fs_info *fs_info,
+ struct reada_extent *re)
+{
+ int i;
+ unsigned long index = re->logical >> PAGE_CACHE_SHIFT;
+
+ spin_lock(&fs_info->reada_lock);
+ if (!kref_put(&re->refcnt, reada_kref_dummy)) {
+ spin_unlock(&fs_info->reada_lock);
+ return;
+ }
+
+ radix_tree_delete(&fs_info->reada_tree, index);
+ for (i = 0; i < re->nzones; ++i) {
+ struct reada_zone *zone = re->zones[i];
+
+ radix_tree_delete(&zone->device->reada_extents, index);
+ }
+
+ spin_unlock(&fs_info->reada_lock);
+
+ for (i = 0; i < re->nzones; ++i) {
+ struct reada_zone *zone = re->zones[i];
+
+ kref_get(&zone->refcnt);
+ spin_lock(&zone->lock);
+ --zone->elems;
+ if (zone->elems == 0) {
+ /* no fs_info->reada_lock needed, as this can't be
+ * the last ref */
+ kref_put(&zone->refcnt, reada_zone_release);
+ }
+ spin_unlock(&zone->lock);
+
+ spin_lock(&fs_info->reada_lock);
+ kref_put(&zone->refcnt, reada_zone_release);
+ spin_unlock(&fs_info->reada_lock);
+ }
+ if (re->scheduled_for)
+ atomic_dec(&re->scheduled_for->reada_in_flight);
+
+ kfree(re);
+}
+
+static void reada_zone_release(struct kref *kref)
+{
+ struct reada_zone *zone = container_of(kref, struct reada_zone, refcnt);
+
+ radix_tree_delete(&zone->device->reada_zones,
+ zone->end >> PAGE_CACHE_SHIFT);
+
+ kfree(zone);
+}
+
+static void reada_control_release(struct kref *kref)
+{
+ struct reada_control *rc = container_of(kref, struct reada_control,
+ refcnt);
+
+ kfree(rc);
+}
+
+static int reada_add_block(struct reada_control *rc, u64 logical,
+ struct btrfs_key *top, int level, u64 generation)
+{
+ struct btrfs_root *root = rc->root;
+ struct reada_extent *re;
+ struct reada_extctl *rec;
+
+ re = reada_find_extent(root, logical, top, level); /* takes one ref */
+ if (!re)
+ return -1;
+
+ rec = kzalloc(sizeof(*rec), GFP_NOFS);
+ if (!rec) {
+ reada_extent_put(root->fs_info, re);
+ return -1;
+ }
+
+ rec->rc = rc;
+ rec->generation = generation;
+ atomic_inc(&rc->elems);
+
+ spin_lock(&re->lock);
+ list_add_tail(&rec->list, &re->extctl);
+ spin_unlock(&re->lock);
+
+ /* leave the ref on the extent */
+
+ return 0;
+}
+
+/*
+ * called with fs_info->reada_lock held
+ */
+static void reada_peer_zones_set_lock(struct reada_zone *zone, int lock)
+{
+ int i;
+ unsigned long index = zone->end >> PAGE_CACHE_SHIFT;
+
+ for (i = 0; i < zone->ndevs; ++i) {
+ struct reada_zone *peer;
+ peer = radix_tree_lookup(&zone->devs[i]->reada_zones, index);
+ if (peer && peer->device != zone->device)
+ peer->locked = lock;
+ }
+}
+
+/*
+ * called with fs_info->reada_lock held
+ */
+static int reada_pick_zone(struct btrfs_device *dev)
+{
+ struct reada_zone *top_zone = NULL;
+ struct reada_zone *top_locked_zone = NULL;
+ u64 top_elems = 0;
+ u64 top_locked_elems = 0;
+ unsigned long index = 0;
+ int ret;
+
+ if (dev->reada_curr_zone) {
+ reada_peer_zones_set_lock(dev->reada_curr_zone, 0);
+ kref_put(&dev->reada_curr_zone->refcnt, reada_zone_release);
+ dev->reada_curr_zone = NULL;
+ }
+ /* pick the zone with the most elements */
+ while (1) {
+ struct reada_zone *zone;
+
+ ret = radix_tree_gang_lookup(&dev->reada_zones,
+ (void **)&zone, index, 1);
+ if (ret == 0)
+ break;
+ index = (zone->end >> PAGE_CACHE_SHIFT) + 1;
+ if (zone->locked) {
+ if (zone->elems > top_locked_elems) {
+ top_locked_elems = zone->elems;
+ top_locked_zone = zone;
+ }
+ } else {
+ if (zone->elems > top_elems) {
+ top_elems = zone->elems;
+ top_zone = zone;
+ }
+ }
+ }
+ if (top_zone)
+ dev->reada_curr_zone = top_zone;
+ else if (top_locked_zone)
+ dev->reada_curr_zone = top_locked_zone;
+ else
+ return 0;
+
+ dev->reada_next = dev->reada_curr_zone->start;
+ kref_get(&dev->reada_curr_zone->refcnt);
+ reada_peer_zones_set_lock(dev->reada_curr_zone, 1);
+
+ return 1;
+}
+
+static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
+ struct btrfs_device *dev)
+{
+ struct reada_extent *re = NULL;
+ int mirror_num = 0;
+ struct extent_buffer *eb = NULL;
+ u64 logical;
+ u32 blocksize;
+ int ret;
+ int i;
+ int need_kick = 0;
+
+ spin_lock(&fs_info->reada_lock);
+ if (dev->reada_curr_zone == NULL) {
+ ret = reada_pick_zone(dev);
+ if (!ret) {
+ spin_unlock(&fs_info->reada_lock);
+ return 0;
+ }
+ }
+ /*
+ * FIXME currently we issue the reads one extent at a time. If we have
+ * a contiguous block of extents, we could also coagulate them or use
+ * plugging to speed things up
+ */
+ ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
+ dev->reada_next >> PAGE_CACHE_SHIFT, 1);
+ if (ret == 0 || re->logical >= dev->reada_curr_zone->end) {
+ ret = reada_pick_zone(dev);
+ if (!ret) {
+ spin_unlock(&fs_info->reada_lock);
+ return 0;
+ }
+ re = NULL;
+ ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
+ dev->reada_next >> PAGE_CACHE_SHIFT, 1);
+ }
+ if (ret == 0) {
+ spin_unlock(&fs_info->reada_lock);
+ return 0;
+ }
+ dev->reada_next = re->logical + re->blocksize;
+ kref_get(&re->refcnt);
+
+ spin_unlock(&fs_info->reada_lock);
+
+ /*
+ * find mirror num
+ */
+ for (i = 0; i < re->nzones; ++i) {
+ if (re->zones[i]->device == dev) {
+ mirror_num = i + 1;
+ break;
+ }
+ }
+ logical = re->logical;
+ blocksize = re->blocksize;
+
+ spin_lock(&re->lock);
+ if (re->scheduled_for == NULL) {
+ re->scheduled_for = dev;
+ need_kick = 1;
+ }
+ spin_unlock(&re->lock);
+
+ reada_extent_put(fs_info, re);
+
+ if (!need_kick)
+ return 0;
+
+ atomic_inc(&dev->reada_in_flight);
+ ret = reada_tree_block_flagged(fs_info->extent_root, logical, blocksize,
+ mirror_num, &eb);
+ if (ret)
+ __readahead_hook(fs_info->extent_root, NULL, logical, ret);
+ else if (eb)
+ __readahead_hook(fs_info->extent_root, eb, eb->start, ret);
+
+ if (eb)
+ free_extent_buffer(eb);
+
+ return 1;
+
+}
+
+static void reada_start_machine_worker(struct btrfs_work *work)
+{
+ struct reada_machine_work *rmw;
+ struct btrfs_fs_info *fs_info;
+
+ rmw = container_of(work, struct reada_machine_work, work);
+ fs_info = rmw->fs_info;
+
+ kfree(rmw);
+
+ __reada_start_machine(fs_info);
+}
+
+static void __reada_start_machine(struct btrfs_fs_info *fs_info)
+{
+ struct btrfs_device *device;
+ struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+ u64 enqueued;
+ u64 total = 0;
+ int i;
+
+ do {
+ enqueued = 0;
+ list_for_each_entry(device, &fs_devices->devices, dev_list) {
+ if (atomic_read(&device->reada_in_flight) <
+ MAX_IN_FLIGHT)
+ enqueued += reada_start_machine_dev(fs_info,
+ device);
+ }
+ total += enqueued;
+ } while (enqueued && total < 10000);
+
+ if (enqueued == 0)
+ return;
+
+ /*
+ * If everything is already in the cache, this is effectively single
+ * threaded. To a) not hold the caller for too long and b) to utilize
+ * more cores, we broke the loop above after 10000 iterations and now
+ * enqueue to workers to finish it. This will distribute the load to
+ * the cores.
+ */
+ for (i = 0; i < 2; ++i)
+ reada_start_machine(fs_info);
+}
+
+static void reada_start_machine(struct btrfs_fs_info *fs_info)
+{
+ struct reada_machine_work *rmw;
+
+ rmw = kzalloc(sizeof(*rmw), GFP_NOFS);
+ if (!rmw) {
+ /* FIXME we cannot handle this properly right now */
+ BUG();
+ }
+ rmw->work.func = reada_start_machine_worker;
+ rmw->fs_info = fs_info;
+
+ btrfs_queue_worker(&fs_info->readahead_workers, &rmw->work);
+}
+
+#ifdef DEBUG
+static void dump_devs(struct btrfs_fs_info *fs_info, int all)
+{
+ struct btrfs_device *device;
+ struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+ unsigned long index;
+ int ret;
+ int i;
+ int j;
+ int cnt;
+
+ spin_lock(&fs_info->reada_lock);
+ list_for_each_entry(device, &fs_devices->devices, dev_list) {
+ printk(KERN_DEBUG "dev %lld has %d in flight\n", device->devid,
+ atomic_read(&device->reada_in_flight));
+ index = 0;
+ while (1) {
+ struct reada_zone *zone;
+ ret = radix_tree_gang_lookup(&device->reada_zones,
+ (void **)&zone, index, 1);
+ if (ret == 0)
+ break;
+ printk(KERN_DEBUG " zone %llu-%llu elems %llu locked "
+ "%d devs", zone->start, zone->end, zone->elems,
+ zone->locked);
+ for (j = 0; j < zone->ndevs; ++j) {
+ printk(KERN_CONT " %lld",
+ zone->devs[j]->devid);
+ }
+ if (device->reada_curr_zone == zone)
+ printk(KERN_CONT " curr off %llu",
+ device->reada_next - zone->start);
+ printk(KERN_CONT "\n");
+ index = (zone->end >> PAGE_CACHE_SHIFT) + 1;
+ }
+ cnt = 0;
+ index = 0;
+ while (all) {
+ struct reada_extent *re = NULL;
+
+ ret = radix_tree_gang_lookup(&device->reada_extents,
+ (void **)&re, index, 1);
+ if (ret == 0)
+ break;
+ printk(KERN_DEBUG
+ " re: logical %llu size %u empty %d for %lld",
+ re->logical, re->blocksize,
+ list_empty(&re->extctl), re->scheduled_for ?
+ re->scheduled_for->devid : -1);
+
+ for (i = 0; i < re->nzones; ++i) {
+ printk(KERN_CONT " zone %llu-%llu devs",
+ re->zones[i]->start,
+ re->zones[i]->end);
+ for (j = 0; j < re->zones[i]->ndevs; ++j) {
+ printk(KERN_CONT " %lld",
+ re->zones[i]->devs[j]->devid);
+ }
+ }
+ printk(KERN_CONT "\n");
+ index = (re->logical >> PAGE_CACHE_SHIFT) + 1;
+ if (++cnt > 15)
+ break;
+ }
+ }
+
+ index = 0;
+ cnt = 0;
+ while (all) {
+ struct reada_extent *re = NULL;
+
+ ret = radix_tree_gang_lookup(&fs_info->reada_tree, (void **)&re,
+ index, 1);
+ if (ret == 0)
+ break;
+ if (!re->scheduled_for) {
+ index = (re->logical >> PAGE_CACHE_SHIFT) + 1;
+ continue;
+ }
+ printk(KERN_DEBUG
+ "re: logical %llu size %u list empty %d for %lld",
+ re->logical, re->blocksize, list_empty(&re->extctl),
+ re->scheduled_for ? re->scheduled_for->devid : -1);
+ for (i = 0; i < re->nzones; ++i) {
+ printk(KERN_CONT " zone %llu-%llu devs",
+ re->zones[i]->start,
+ re->zones[i]->end);
+ for (i = 0; i < re->nzones; ++i) {
+ printk(KERN_CONT " zone %llu-%llu devs",
+ re->zones[i]->start,
+ re->zones[i]->end);
+ for (j = 0; j < re->zones[i]->ndevs; ++j) {
+ printk(KERN_CONT " %lld",
+ re->zones[i]->devs[j]->devid);
+ }
+ }
+ }
+ printk(KERN_CONT "\n");
+ index = (re->logical >> PAGE_CACHE_SHIFT) + 1;
+ }
+ spin_unlock(&fs_info->reada_lock);
+}
+#endif
+
+/*
+ * interface
+ */
+struct reada_control *btrfs_reada_add(struct btrfs_root *root,
+ struct btrfs_key *key_start, struct btrfs_key *key_end)
+{
+ struct reada_control *rc;
+ u64 start;
+ u64 generation;
+ int level;
+ struct extent_buffer *node;
+ static struct btrfs_key max_key = {
+ .objectid = (u64)-1,
+ .type = (u8)-1,
+ .offset = (u64)-1
+ };
+
+ rc = kzalloc(sizeof(*rc), GFP_NOFS);
+ if (!rc)
+ return ERR_PTR(-ENOMEM);
+
+ rc->root = root;
+ rc->key_start = *key_start;
+ rc->key_end = *key_end;
+ atomic_set(&rc->elems, 0);
+ init_waitqueue_head(&rc->wait);
+ kref_init(&rc->refcnt);
+ kref_get(&rc->refcnt); /* one ref for having elements */
+
+ node = btrfs_root_node(root);
+ start = node->start;
+ level = btrfs_header_level(node);
+ generation = btrfs_header_generation(node);
+ free_extent_buffer(node);
+
+ reada_add_block(rc, start, &max_key, level, generation);
+
+ reada_start_machine(root->fs_info);
+
+ return rc;
+}
+
+#ifdef DEBUG
+int btrfs_reada_wait(void *handle)
+{
+ struct reada_control *rc = handle;
+
+ while (atomic_read(&rc->elems)) {
+ wait_event_timeout(rc->wait, atomic_read(&rc->elems) == 0,
+ 5 * HZ);
+ dump_devs(rc->root->fs_info, rc->elems < 10 ? 1 : 0);
+ }
+
+ dump_devs(rc->root->fs_info, rc->elems < 10 ? 1 : 0);
+
+ kref_put(&rc->refcnt, reada_control_release);
+
+ return 0;
+}
+#else
+int btrfs_reada_wait(void *handle)
+{
+ struct reada_control *rc = handle;
+
+ while (atomic_read(&rc->elems)) {
+ wait_event(rc->wait, atomic_read(&rc->elems) == 0);
+ }
+
+ kref_put(&rc->refcnt, reada_control_release);
+
+ return 0;
+}
+#endif
+
+void btrfs_reada_detach(void *handle)
+{
+ struct reada_control *rc = handle;
+
+ kref_put(&rc->refcnt, reada_control_release);
+}
*/
#include <linux/blkdev.h>
+ #include <linux/ratelimit.h>
#include "ctree.h"
#include "volumes.h"
#include "disk-io.h"
#include "ordered-data.h"
+ #include "transaction.h"
+ #include "backref.h"
+ #include "extent_io.h"
/*
* This is only the first step towards a full-features scrub. It reads all
* any can be found.
*
* Future enhancements:
- * - To enhance the performance, better read-ahead strategies for the
- * extent-tree can be employed.
* - In case an unrepairable extent is encountered, track which files are
* affected and report them
* - In case of a read error on files with nodatasum, map the file and read
* the extent to trigger a writeback of the good copy
* - track and record media errors, throw out bad devices
* - add a mode to also read unallocated space
- * - make the prefetch cancellable
*/
struct scrub_bio;
struct scrub_page {
u64 flags; /* extent flags */
u64 generation;
- u64 mirror_num;
+ int mirror_num;
int have_csum;
u8 csum[BTRFS_CSUM_SIZE];
};
int first_free;
int curr;
atomic_t in_flight;
+ atomic_t fixup_cnt;
spinlock_t list_lock;
wait_queue_head_t list_wait;
u16 csum_size;
spinlock_t stat_lock;
};
+ struct scrub_fixup_nodatasum {
+ struct scrub_dev *sdev;
+ u64 logical;
+ struct btrfs_root *root;
+ struct btrfs_work work;
+ int mirror_num;
+ };
+
+ struct scrub_warning {
+ struct btrfs_path *path;
+ u64 extent_item_size;
+ char *scratch_buf;
+ char *msg_buf;
+ const char *errstr;
+ sector_t sector;
+ u64 logical;
+ struct btrfs_device *dev;
+ int msg_bufsize;
+ int scratch_bufsize;
+ };
+
static void scrub_free_csums(struct scrub_dev *sdev)
{
while (!list_empty(&sdev->csum_list)) {
if (i != SCRUB_BIOS_PER_DEV-1)
sdev->bios[i]->next_free = i + 1;
- else
+ else
sdev->bios[i]->next_free = -1;
}
sdev->first_free = 0;
sdev->curr = -1;
atomic_set(&sdev->in_flight, 0);
+ atomic_set(&sdev->fixup_cnt, 0);
atomic_set(&sdev->cancel_req, 0);
- sdev->csum_size = btrfs_super_csum_size(&fs_info->super_copy);
+ sdev->csum_size = btrfs_super_csum_size(fs_info->super_copy);
INIT_LIST_HEAD(&sdev->csum_list);
spin_lock_init(&sdev->list_lock);
return ERR_PTR(-ENOMEM);
}
+ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root, void *ctx)
+ {
+ u64 isize;
+ u32 nlink;
+ int ret;
+ int i;
+ struct extent_buffer *eb;
+ struct btrfs_inode_item *inode_item;
+ struct scrub_warning *swarn = ctx;
+ struct btrfs_fs_info *fs_info = swarn->dev->dev_root->fs_info;
+ struct inode_fs_paths *ipath = NULL;
+ struct btrfs_root *local_root;
+ struct btrfs_key root_key;
+
+ root_key.objectid = root;
+ root_key.type = BTRFS_ROOT_ITEM_KEY;
+ root_key.offset = (u64)-1;
+ local_root = btrfs_read_fs_root_no_name(fs_info, &root_key);
+ if (IS_ERR(local_root)) {
+ ret = PTR_ERR(local_root);
+ goto err;
+ }
+
+ ret = inode_item_info(inum, 0, local_root, swarn->path);
+ if (ret) {
+ btrfs_release_path(swarn->path);
+ goto err;
+ }
+
+ eb = swarn->path->nodes[0];
+ inode_item = btrfs_item_ptr(eb, swarn->path->slots[0],
+ struct btrfs_inode_item);
+ isize = btrfs_inode_size(eb, inode_item);
+ nlink = btrfs_inode_nlink(eb, inode_item);
+ btrfs_release_path(swarn->path);
+
+ ipath = init_ipath(4096, local_root, swarn->path);
+ ret = paths_from_inode(inum, ipath);
+
+ if (ret < 0)
+ goto err;
+
+ /*
+ * we deliberately ignore the bit ipath might have been too small to
+ * hold all of the paths here
+ */
+ for (i = 0; i < ipath->fspath->elem_cnt; ++i)
+ printk(KERN_WARNING "btrfs: %s at logical %llu on dev "
+ "%s, sector %llu, root %llu, inode %llu, offset %llu, "
+ "length %llu, links %u (path: %s)\n", swarn->errstr,
+ swarn->logical, swarn->dev->name,
+ (unsigned long long)swarn->sector, root, inum, offset,
+ min(isize - offset, (u64)PAGE_SIZE), nlink,
+ ipath->fspath->str[i]);
+
+ free_ipath(ipath);
+ return 0;
+
+ err:
+ printk(KERN_WARNING "btrfs: %s at logical %llu on dev "
+ "%s, sector %llu, root %llu, inode %llu, offset %llu: path "
+ "resolving failed with ret=%d\n", swarn->errstr,
+ swarn->logical, swarn->dev->name,
+ (unsigned long long)swarn->sector, root, inum, offset, ret);
+
+ free_ipath(ipath);
+ return 0;
+ }
+
+ static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
+ int ix)
+ {
+ struct btrfs_device *dev = sbio->sdev->dev;
+ struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
+ struct btrfs_path *path;
+ struct btrfs_key found_key;
+ struct extent_buffer *eb;
+ struct btrfs_extent_item *ei;
+ struct scrub_warning swarn;
+ u32 item_size;
+ int ret;
+ u64 ref_root;
+ u8 ref_level;
+ unsigned long ptr = 0;
+ const int bufsize = 4096;
+ u64 extent_offset;
+
+ path = btrfs_alloc_path();
+
+ swarn.scratch_buf = kmalloc(bufsize, GFP_NOFS);
+ swarn.msg_buf = kmalloc(bufsize, GFP_NOFS);
+ swarn.sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+ swarn.logical = sbio->logical + ix * PAGE_SIZE;
+ swarn.errstr = errstr;
+ swarn.dev = dev;
+ swarn.msg_bufsize = bufsize;
+ swarn.scratch_bufsize = bufsize;
+
+ if (!path || !swarn.scratch_buf || !swarn.msg_buf)
+ goto out;
+
+ ret = extent_from_logical(fs_info, swarn.logical, path, &found_key);
+ if (ret < 0)
+ goto out;
+
+ extent_offset = swarn.logical - found_key.objectid;
+ swarn.extent_item_size = found_key.offset;
+
+ eb = path->nodes[0];
+ ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
+ item_size = btrfs_item_size_nr(eb, path->slots[0]);
+
+ if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+ do {
+ ret = tree_backref_for_extent(&ptr, eb, ei, item_size,
+ &ref_root, &ref_level);
+ printk(KERN_WARNING "%s at logical %llu on dev %s, "
+ "sector %llu: metadata %s (level %d) in tree "
+ "%llu\n", errstr, swarn.logical, dev->name,
+ (unsigned long long)swarn.sector,
+ ref_level ? "node" : "leaf",
+ ret < 0 ? -1 : ref_level,
+ ret < 0 ? -1 : ref_root);
+ } while (ret != 1);
+ } else {
+ swarn.path = path;
+ iterate_extent_inodes(fs_info, path, found_key.objectid,
+ extent_offset,
+ scrub_print_warning_inode, &swarn);
+ }
+
+ out:
+ btrfs_free_path(path);
+ kfree(swarn.scratch_buf);
+ kfree(swarn.msg_buf);
+ }
+
+ static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *ctx)
+ {
+ struct page *page = NULL;
+ unsigned long index;
+ struct scrub_fixup_nodatasum *fixup = ctx;
+ int ret;
+ int corrected = 0;
+ struct btrfs_key key;
+ struct inode *inode = NULL;
+ u64 end = offset + PAGE_SIZE - 1;
+ struct btrfs_root *local_root;
+
+ key.objectid = root;
+ key.type = BTRFS_ROOT_ITEM_KEY;
+ key.offset = (u64)-1;
+ local_root = btrfs_read_fs_root_no_name(fixup->root->fs_info, &key);
+ if (IS_ERR(local_root))
+ return PTR_ERR(local_root);
+
+ key.type = BTRFS_INODE_ITEM_KEY;
+ key.objectid = inum;
+ key.offset = 0;
+ inode = btrfs_iget(fixup->root->fs_info->sb, &key, local_root, NULL);
+ if (IS_ERR(inode))
+ return PTR_ERR(inode);
+
+ index = offset >> PAGE_CACHE_SHIFT;
+
+ page = find_or_create_page(inode->i_mapping, index, GFP_NOFS);
+ if (!page) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ if (PageUptodate(page)) {
+ struct btrfs_mapping_tree *map_tree;
+ if (PageDirty(page)) {
+ /*
+ * we need to write the data to the defect sector. the
+ * data that was in that sector is not in memory,
+ * because the page was modified. we must not write the
+ * modified page to that sector.
+ *
+ * TODO: what could be done here: wait for the delalloc
+ * runner to write out that page (might involve
+ * COW) and see whether the sector is still
+ * referenced afterwards.
+ *
+ * For the meantime, we'll treat this error
+ * incorrectable, although there is a chance that a
+ * later scrub will find the bad sector again and that
+ * there's no dirty page in memory, then.
+ */
+ ret = -EIO;
+ goto out;
+ }
+ map_tree = &BTRFS_I(inode)->root->fs_info->mapping_tree;
+ ret = repair_io_failure(map_tree, offset, PAGE_SIZE,
+ fixup->logical, page,
+ fixup->mirror_num);
+ unlock_page(page);
+ corrected = !ret;
+ } else {
+ /*
+ * we need to get good data first. the general readpage path
+ * will call repair_io_failure for us, we just have to make
+ * sure we read the bad mirror.
+ */
+ ret = set_extent_bits(&BTRFS_I(inode)->io_tree, offset, end,
+ EXTENT_DAMAGED, GFP_NOFS);
+ if (ret) {
+ /* set_extent_bits should give proper error */
+ WARN_ON(ret > 0);
+ if (ret > 0)
+ ret = -EFAULT;
+ goto out;
+ }
+
+ ret = extent_read_full_page(&BTRFS_I(inode)->io_tree, page,
+ btrfs_get_extent,
+ fixup->mirror_num);
+ wait_on_page_locked(page);
+
+ corrected = !test_range_bit(&BTRFS_I(inode)->io_tree, offset,
+ end, EXTENT_DAMAGED, 0, NULL);
+ if (!corrected)
+ clear_extent_bits(&BTRFS_I(inode)->io_tree, offset, end,
+ EXTENT_DAMAGED, GFP_NOFS);
+ }
+
+ out:
+ if (page)
+ put_page(page);
+ if (inode)
+ iput(inode);
+
+ if (ret < 0)
+ return ret;
+
+ if (ret == 0 && corrected) {
+ /*
+ * we only need to call readpage for one of the inodes belonging
+ * to this extent. so make iterate_extent_inodes stop
+ */
+ return 1;
+ }
+
+ return -EIO;
+ }
+
+ static void scrub_fixup_nodatasum(struct btrfs_work *work)
+ {
+ int ret;
+ struct scrub_fixup_nodatasum *fixup;
+ struct scrub_dev *sdev;
+ struct btrfs_trans_handle *trans = NULL;
+ struct btrfs_fs_info *fs_info;
+ struct btrfs_path *path;
+ int uncorrectable = 0;
+
+ fixup = container_of(work, struct scrub_fixup_nodatasum, work);
+ sdev = fixup->sdev;
+ fs_info = fixup->root->fs_info;
+
+ path = btrfs_alloc_path();
+ if (!path) {
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.malloc_errors;
+ spin_unlock(&sdev->stat_lock);
+ uncorrectable = 1;
+ goto out;
+ }
+
+ trans = btrfs_join_transaction(fixup->root);
+ if (IS_ERR(trans)) {
+ uncorrectable = 1;
+ goto out;
+ }
+
+ /*
+ * the idea is to trigger a regular read through the standard path. we
+ * read a page from the (failed) logical address by specifying the
+ * corresponding copynum of the failed sector. thus, that readpage is
+ * expected to fail.
+ * that is the point where on-the-fly error correction will kick in
+ * (once it's finished) and rewrite the failed sector if a good copy
+ * can be found.
+ */
+ ret = iterate_inodes_from_logical(fixup->logical, fixup->root->fs_info,
+ path, scrub_fixup_readpage,
+ fixup);
+ if (ret < 0) {
+ uncorrectable = 1;
+ goto out;
+ }
+ WARN_ON(ret != 1);
+
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.corrected_errors;
+ spin_unlock(&sdev->stat_lock);
+
+ out:
+ if (trans && !IS_ERR(trans))
+ btrfs_end_transaction(trans, fixup->root);
+ if (uncorrectable) {
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.uncorrectable_errors;
+ spin_unlock(&sdev->stat_lock);
+ printk_ratelimited(KERN_ERR "btrfs: unable to fixup "
+ "(nodatasum) error at logical %llu\n",
+ fixup->logical);
+ }
+
+ btrfs_free_path(path);
+ kfree(fixup);
+
+ /* see caller why we're pretending to be paused in the scrub counters */
+ mutex_lock(&fs_info->scrub_lock);
+ atomic_dec(&fs_info->scrubs_running);
+ atomic_dec(&fs_info->scrubs_paused);
+ mutex_unlock(&fs_info->scrub_lock);
+ atomic_dec(&sdev->fixup_cnt);
+ wake_up(&fs_info->scrub_pause_wait);
+ wake_up(&sdev->list_wait);
+ }
+
/*
* scrub_recheck_error gets called when either verification of the page
* failed or the bio failed to read, e.g. with EIO. In the latter case,
* recheck_error gets called for every page in the bio, even though only
* one may be bad
*/
- static void scrub_recheck_error(struct scrub_bio *sbio, int ix)
+ static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
{
+ struct scrub_dev *sdev = sbio->sdev;
+ u64 sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+ static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
+ DEFAULT_RATELIMIT_BURST);
+
if (sbio->err) {
- if (scrub_fixup_io(READ, sbio->sdev->dev->bdev,
- (sbio->physical + ix * PAGE_SIZE) >> 9,
+ if (scrub_fixup_io(READ, sbio->sdev->dev->bdev, sector,
sbio->bio->bi_io_vec[ix].bv_page) == 0) {
if (scrub_fixup_check(sbio, ix) == 0)
- return;
+ return 0;
}
+ if (__ratelimit(&_rs))
+ scrub_print_warning("i/o error", sbio, ix);
+ } else {
+ if (__ratelimit(&_rs))
+ scrub_print_warning("checksum error", sbio, ix);
}
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.read_errors;
+ spin_unlock(&sdev->stat_lock);
+
scrub_fixup(sbio, ix);
+ return 1;
}
static int scrub_fixup_check(struct scrub_bio *sbio, int ix)
struct scrub_dev *sdev = sbio->sdev;
struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
- struct btrfs_multi_bio *multi = NULL;
+ struct btrfs_bio *bbio = NULL;
+ struct scrub_fixup_nodatasum *fixup;
u64 logical = sbio->logical + ix * PAGE_SIZE;
u64 length;
int i;
if ((sbio->spag[ix].flags & BTRFS_EXTENT_FLAG_DATA) &&
(sbio->spag[ix].have_csum == 0)) {
+ fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
+ if (!fixup)
+ goto uncorrectable;
+ fixup->sdev = sdev;
+ fixup->logical = logical;
+ fixup->root = fs_info->extent_root;
+ fixup->mirror_num = sbio->spag[ix].mirror_num;
/*
- * nodatasum, don't try to fix anything
- * FIXME: we can do better, open the inode and trigger a
- * writeback
+ * increment scrubs_running to prevent cancel requests from
+ * completing as long as a fixup worker is running. we must also
+ * increment scrubs_paused to prevent deadlocking on pause
+ * requests used for transactions commits (as the worker uses a
+ * transaction context). it is safe to regard the fixup worker
+ * as paused for all matters practical. effectively, we only
+ * avoid cancellation requests from completing.
*/
- goto uncorrectable;
+ mutex_lock(&fs_info->scrub_lock);
+ atomic_inc(&fs_info->scrubs_running);
+ atomic_inc(&fs_info->scrubs_paused);
+ mutex_unlock(&fs_info->scrub_lock);
+ atomic_inc(&sdev->fixup_cnt);
+ fixup->work.func = scrub_fixup_nodatasum;
+ btrfs_queue_worker(&fs_info->scrub_workers, &fixup->work);
+ return;
}
length = PAGE_SIZE;
ret = btrfs_map_block(map_tree, REQ_WRITE, logical, &length,
- &multi, 0);
- if (ret || !multi || length < PAGE_SIZE) {
+ &bbio, 0);
+ if (ret || !bbio || length < PAGE_SIZE) {
printk(KERN_ERR
"scrub_fixup: btrfs_map_block failed us for %llu\n",
(unsigned long long)logical);
return;
}
- if (multi->num_stripes == 1)
+ if (bbio->num_stripes == 1)
/* there aren't any replicas */
goto uncorrectable;
/*
* first find a good copy
*/
- for (i = 0; i < multi->num_stripes; ++i) {
- if (i == sbio->spag[ix].mirror_num)
+ for (i = 0; i < bbio->num_stripes; ++i) {
+ if (i + 1 == sbio->spag[ix].mirror_num)
continue;
- if (scrub_fixup_io(READ, multi->stripes[i].dev->bdev,
- multi->stripes[i].physical >> 9,
+ if (scrub_fixup_io(READ, bbio->stripes[i].dev->bdev,
+ bbio->stripes[i].physical >> 9,
sbio->bio->bi_io_vec[ix].bv_page)) {
/* I/O-error, this is not a good copy */
continue;
if (scrub_fixup_check(sbio, ix) == 0)
break;
}
- if (i == multi->num_stripes)
+ if (i == bbio->num_stripes)
goto uncorrectable;
if (!sdev->readonly) {
}
}
- kfree(multi);
+ kfree(bbio);
spin_lock(&sdev->stat_lock);
++sdev->stat.corrected_errors;
spin_unlock(&sdev->stat_lock);
- if (printk_ratelimit())
- printk(KERN_ERR "btrfs: fixed up at %llu\n",
- (unsigned long long)logical);
+ printk_ratelimited(KERN_ERR "btrfs: fixed up error at logical %llu\n",
+ (unsigned long long)logical);
return;
uncorrectable:
- kfree(multi);
+ kfree(bbio);
spin_lock(&sdev->stat_lock);
++sdev->stat.uncorrectable_errors;
spin_unlock(&sdev->stat_lock);
- if (printk_ratelimit())
- printk(KERN_ERR "btrfs: unable to fixup at %llu\n",
- (unsigned long long)logical);
+ printk_ratelimited(KERN_ERR "btrfs: unable to fixup (regular) error at "
+ "logical %llu\n", (unsigned long long)logical);
}
static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
int ret;
if (sbio->err) {
+ ret = 0;
for (i = 0; i < sbio->count; ++i)
- scrub_recheck_error(sbio, i);
+ ret |= scrub_recheck_error(sbio, i);
+ if (!ret) {
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.unverified_errors;
+ spin_unlock(&sdev->stat_lock);
+ }
sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
bi->bv_offset = 0;
bi->bv_len = PAGE_SIZE;
}
-
- spin_lock(&sdev->stat_lock);
- ++sdev->stat.read_errors;
- spin_unlock(&sdev->stat_lock);
goto out;
}
for (i = 0; i < sbio->count; ++i) {
WARN_ON(1);
}
kunmap_atomic(buffer, KM_USER0);
- if (ret)
- scrub_recheck_error(sbio, i);
+ if (ret) {
+ ret = scrub_recheck_error(sbio, i);
+ if (!ret) {
+ spin_lock(&sdev->stat_lock);
+ ++sdev->stat.unverified_errors;
+ spin_unlock(&sdev->stat_lock);
+ }
+ }
}
out:
}
static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
- u64 physical, u64 flags, u64 gen, u64 mirror_num,
+ u64 physical, u64 flags, u64 gen, int mirror_num,
u8 *csum, int force)
{
struct scrub_bio *sbio;
/* scrub extent tries to collect up to 64 kB for each bio */
static int scrub_extent(struct scrub_dev *sdev, u64 logical, u64 len,
- u64 physical, u64 flags, u64 gen, u64 mirror_num)
+ u64 physical, u64 flags, u64 gen, int mirror_num)
{
int ret;
u8 csum[BTRFS_CSUM_SIZE];
int slot;
int i;
u64 nstripes;
- int start_stripe;
struct extent_buffer *l;
struct btrfs_key key;
u64 physical;
u64 logical;
u64 generation;
- u64 mirror_num;
+ int mirror_num;
+ struct reada_control *reada1;
+ struct reada_control *reada2;
+ struct btrfs_key key_start;
+ struct btrfs_key key_end;
u64 increment = map->stripe_len;
u64 offset;
if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
offset = map->stripe_len * num;
increment = map->stripe_len * map->num_stripes;
- mirror_num = 0;
+ mirror_num = 1;
} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
int factor = map->num_stripes / map->sub_stripes;
offset = map->stripe_len * (num / map->sub_stripes);
increment = map->stripe_len * factor;
- mirror_num = num % map->sub_stripes;
+ mirror_num = num % map->sub_stripes + 1;
} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
increment = map->stripe_len;
- mirror_num = num % map->num_stripes;
+ mirror_num = num % map->num_stripes + 1;
} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
increment = map->stripe_len;
- mirror_num = num % map->num_stripes;
+ mirror_num = num % map->num_stripes + 1;
} else {
increment = map->stripe_len;
- mirror_num = 0;
+ mirror_num = 1;
}
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
- path->reada = 2;
path->search_commit_root = 1;
path->skip_locking = 1;
/*
- * find all extents for each stripe and just read them to get
- * them into the page cache
- * FIXME: we can do better. build a more intelligent prefetching
+ * trigger the readahead for extent tree csum tree and wait for
+ * completion. During readahead, the scrub is officially paused
+ * to not hold off transaction commits
*/
logical = base + offset;
- physical = map->stripes[num].physical;
- ret = 0;
- for (i = 0; i < nstripes; ++i) {
- key.objectid = logical;
- key.type = BTRFS_EXTENT_ITEM_KEY;
- key.offset = (u64)0;
-
- ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
- if (ret < 0)
- goto out_noplug;
- /*
- * we might miss half an extent here, but that doesn't matter,
- * as it's only the prefetch
- */
- while (1) {
- l = path->nodes[0];
- slot = path->slots[0];
- if (slot >= btrfs_header_nritems(l)) {
- ret = btrfs_next_leaf(root, path);
- if (ret == 0)
- continue;
- if (ret < 0)
- goto out_noplug;
-
- break;
- }
- btrfs_item_key_to_cpu(l, &key, slot);
+ wait_event(sdev->list_wait,
+ atomic_read(&sdev->in_flight) == 0);
+ atomic_inc(&fs_info->scrubs_paused);
+ wake_up(&fs_info->scrub_pause_wait);
- if (key.objectid >= logical + map->stripe_len)
- break;
+ /* FIXME it might be better to start readahead at commit root */
+ key_start.objectid = logical;
+ key_start.type = BTRFS_EXTENT_ITEM_KEY;
+ key_start.offset = (u64)0;
+ key_end.objectid = base + offset + nstripes * increment;
+ key_end.type = BTRFS_EXTENT_ITEM_KEY;
+ key_end.offset = (u64)0;
+ reada1 = btrfs_reada_add(root, &key_start, &key_end);
+
+ key_start.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
+ key_start.type = BTRFS_EXTENT_CSUM_KEY;
+ key_start.offset = logical;
+ key_end.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
+ key_end.type = BTRFS_EXTENT_CSUM_KEY;
+ key_end.offset = base + offset + nstripes * increment;
+ reada2 = btrfs_reada_add(csum_root, &key_start, &key_end);
+
+ if (!IS_ERR(reada1))
+ btrfs_reada_wait(reada1);
+ if (!IS_ERR(reada2))
+ btrfs_reada_wait(reada2);
- path->slots[0]++;
- }
- btrfs_release_path(path);
- logical += increment;
- physical += map->stripe_len;
- cond_resched();
+ mutex_lock(&fs_info->scrub_lock);
+ while (atomic_read(&fs_info->scrub_pause_req)) {
+ mutex_unlock(&fs_info->scrub_lock);
+ wait_event(fs_info->scrub_pause_wait,
+ atomic_read(&fs_info->scrub_pause_req) == 0);
+ mutex_lock(&fs_info->scrub_lock);
}
+ atomic_dec(&fs_info->scrubs_paused);
+ mutex_unlock(&fs_info->scrub_lock);
+ wake_up(&fs_info->scrub_pause_wait);
/*
* collect all data csums for the stripe to avoid seeking during
* the scrub. This might currently (crc32) end up to be about 1MB
*/
- start_stripe = 0;
blk_start_plug(&plug);
-again:
- logical = base + offset + start_stripe * increment;
- for (i = start_stripe; i < nstripes; ++i) {
- ret = btrfs_lookup_csums_range(csum_root, logical,
- logical + map->stripe_len - 1,
- &sdev->csum_list, 1);
- if (ret)
- goto out;
- logical += increment;
- cond_resched();
- }
/*
* now find all extents for each stripe and scrub them
*/
- logical = base + offset + start_stripe * increment;
- physical = map->stripes[num].physical + start_stripe * map->stripe_len;
+ logical = base + offset;
+ physical = map->stripes[num].physical;
ret = 0;
- for (i = start_stripe; i < nstripes; ++i) {
+ for (i = 0; i < nstripes; ++i) {
/*
* canceled?
*/
atomic_dec(&fs_info->scrubs_paused);
mutex_unlock(&fs_info->scrub_lock);
wake_up(&fs_info->scrub_pause_wait);
- scrub_free_csums(sdev);
- start_stripe = i;
- goto again;
}
+ ret = btrfs_lookup_csums_range(csum_root, logical,
+ logical + map->stripe_len - 1,
+ &sdev->csum_list, 1);
+ if (ret)
+ goto out;
+
key.objectid = logical;
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = (u64)0;
out:
blk_finish_plug(&plug);
-out_noplug:
btrfs_free_path(path);
return ret < 0 ? ret : 0;
}
ret = scrub_enumerate_chunks(sdev, start, end);
wait_event(sdev->list_wait, atomic_read(&sdev->in_flight) == 0);
-
atomic_dec(&fs_info->scrubs_running);
wake_up(&fs_info->scrub_pause_wait);
+ wait_event(sdev->list_wait, atomic_read(&sdev->fixup_cnt) == 0);
+
if (progress)
memcpy(progress, &sdev->stat, sizeof(*progress));
}
INIT_LIST_HEAD(&device->dev_alloc_list);
+ /* init readahead state */
+ spin_lock_init(&device->reada_lock);
+ device->reada_curr_zone = NULL;
+ atomic_set(&device->reada_in_flight, 0);
+ device->reada_next = 0;
+ INIT_RADIX_TREE(&device->reada_zones, GFP_NOFS & ~__GFP_WAIT);
+ INIT_RADIX_TREE(&device->reada_extents, GFP_NOFS & ~__GFP_WAIT);
+
mutex_lock(&fs_devices->device_list_mutex);
list_add_rcu(&device->dev_list, &fs_devices->devices);
mutex_unlock(&fs_devices->device_list_mutex);
set_blocksize(bdev, 4096);
bh = btrfs_read_dev_super(bdev);
- if (!bh) {
- ret = -EINVAL;
+ if (!bh)
goto error_close;
- }
disk_super = (struct btrfs_super_block *)bh->b_data;
devid = btrfs_stack_device_id(&disk_super->dev_item);
continue;
}
if (fs_devices->open_devices == 0) {
- ret = -EIO;
+ ret = -EINVAL;
goto out;
}
fs_devices->seeding = seeding;
}
BUG_ON(ret);
- if (device->bytes_used > 0)
- device->bytes_used -= btrfs_dev_extent_length(leaf, extent);
+ if (device->bytes_used > 0) {
+ u64 len = btrfs_dev_extent_length(leaf, extent);
+ device->bytes_used -= len;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space += len;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+ }
ret = btrfs_del_item(trans, root, path);
out:
if (ret)
goto error_undo;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space = device->total_bytes -
+ device->bytes_used;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+
device->in_fs_metadata = 0;
btrfs_scrub_cancel_dev(root, device);
call_rcu(&device->rcu, free_device);
mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
- num_devices = btrfs_super_num_devices(&root->fs_info->super_copy) - 1;
- btrfs_set_super_num_devices(&root->fs_info->super_copy, num_devices);
+ num_devices = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
+ btrfs_set_super_num_devices(root->fs_info->super_copy, num_devices);
if (cur_devices->open_devices == 0) {
struct btrfs_fs_devices *fs_devices;
struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
struct btrfs_fs_devices *old_devices;
struct btrfs_fs_devices *seed_devices;
- struct btrfs_super_block *disk_super = &root->fs_info->super_copy;
+ struct btrfs_super_block *disk_super = root->fs_info->super_copy;
struct btrfs_device *device;
u64 super_flags;
root->fs_info->fs_devices->num_can_discard++;
root->fs_info->fs_devices->total_rw_bytes += device->total_bytes;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space += device->total_bytes;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+
if (!blk_queue_nonrot(bdev_get_queue(bdev)))
root->fs_info->fs_devices->rotating = 1;
- total_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
- btrfs_set_super_total_bytes(&root->fs_info->super_copy,
+ total_bytes = btrfs_super_total_bytes(root->fs_info->super_copy);
+ btrfs_set_super_total_bytes(root->fs_info->super_copy,
total_bytes + device->total_bytes);
- total_bytes = btrfs_super_num_devices(&root->fs_info->super_copy);
- btrfs_set_super_num_devices(&root->fs_info->super_copy,
+ total_bytes = btrfs_super_num_devices(root->fs_info->super_copy);
+ btrfs_set_super_num_devices(root->fs_info->super_copy,
total_bytes + 1);
mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
struct btrfs_device *device, u64 new_size)
{
struct btrfs_super_block *super_copy =
- &device->dev_root->fs_info->super_copy;
+ device->dev_root->fs_info->super_copy;
u64 old_total = btrfs_super_total_bytes(super_copy);
u64 diff = new_size - device->total_bytes;
static int btrfs_del_sys_chunk(struct btrfs_root *root, u64 chunk_objectid, u64
chunk_offset)
{
- struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
+ struct btrfs_super_block *super_copy = root->fs_info->super_copy;
struct btrfs_disk_key *disk_key;
struct btrfs_chunk *chunk;
u8 *ptr;
bool retried = false;
struct extent_buffer *l;
struct btrfs_key key;
- struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
+ struct btrfs_super_block *super_copy = root->fs_info->super_copy;
u64 old_total = btrfs_super_total_bytes(super_copy);
u64 old_size = device->total_bytes;
u64 diff = device->total_bytes - new_size;
lock_chunks(root);
device->total_bytes = new_size;
- if (device->writeable)
+ if (device->writeable) {
device->fs_devices->total_rw_bytes -= diff;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space -= diff;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+ }
unlock_chunks(root);
again:
device->total_bytes = old_size;
if (device->writeable)
device->fs_devices->total_rw_bytes += diff;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space += diff;
+ spin_unlock(&root->fs_info->free_chunk_lock);
unlock_chunks(root);
goto done;
}
struct btrfs_key *key,
struct btrfs_chunk *chunk, int item_size)
{
- struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
+ struct btrfs_super_block *super_copy = root->fs_info->super_copy;
struct btrfs_disk_key disk_key;
u32 array_size;
u8 *ptr;
index++;
}
+ spin_lock(&extent_root->fs_info->free_chunk_lock);
+ extent_root->fs_info->free_chunk_space -= (stripe_size *
+ map->num_stripes);
+ spin_unlock(&extent_root->fs_info->free_chunk_lock);
+
index = 0;
stripe = &chunk->stripe;
while (index < map->num_stripes) {
static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
u64 logical, u64 *length,
- struct btrfs_multi_bio **multi_ret,
+ struct btrfs_bio **bbio_ret,
int mirror_num)
{
struct extent_map *em;
int i;
int num_stripes;
int max_errors = 0;
- struct btrfs_multi_bio *multi = NULL;
+ struct btrfs_bio *bbio = NULL;
- if (multi_ret && !(rw & (REQ_WRITE | REQ_DISCARD)))
+ if (bbio_ret && !(rw & (REQ_WRITE | REQ_DISCARD)))
stripes_allocated = 1;
again:
- if (multi_ret) {
- multi = kzalloc(btrfs_multi_bio_size(stripes_allocated),
+ if (bbio_ret) {
+ bbio = kzalloc(btrfs_bio_size(stripes_allocated),
GFP_NOFS);
- if (!multi)
+ if (!bbio)
return -ENOMEM;
- atomic_set(&multi->error, 0);
+ atomic_set(&bbio->error, 0);
}
read_lock(&em_tree->lock);
if (mirror_num > map->num_stripes)
mirror_num = 0;
- /* if our multi bio struct is too small, back off and try again */
+ /* if our btrfs_bio struct is too small, back off and try again */
if (rw & REQ_WRITE) {
if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
BTRFS_BLOCK_GROUP_DUP)) {
stripes_required = map->num_stripes;
}
}
- if (multi_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
+ if (bbio_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
stripes_allocated < stripes_required) {
stripes_allocated = map->num_stripes;
free_extent_map(em);
- kfree(multi);
+ kfree(bbio);
goto again;
}
stripe_nr = offset;
*length = em->len - offset;
}
- if (!multi_ret)
+ if (!bbio_ret)
goto out;
num_stripes = 1;
stripe_index = find_live_mirror(map, 0,
map->num_stripes,
current->pid % map->num_stripes);
+ mirror_num = stripe_index + 1;
}
} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
- if (rw & (REQ_WRITE | REQ_DISCARD))
+ if (rw & (REQ_WRITE | REQ_DISCARD)) {
num_stripes = map->num_stripes;
- else if (mirror_num)
+ } else if (mirror_num) {
stripe_index = mirror_num - 1;
+ } else {
+ mirror_num = 1;
+ }
} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
int factor = map->num_stripes / map->sub_stripes;
stripe_index = find_live_mirror(map, stripe_index,
map->sub_stripes, stripe_index +
current->pid % map->sub_stripes);
+ mirror_num = stripe_index + 1;
}
} else {
/*
* stripe_index is the number of our device in the stripe array
*/
stripe_index = do_div(stripe_nr, map->num_stripes);
+ mirror_num = stripe_index + 1;
}
BUG_ON(stripe_index >= map->num_stripes);
if (rw & REQ_DISCARD) {
for (i = 0; i < num_stripes; i++) {
- multi->stripes[i].physical =
+ bbio->stripes[i].physical =
map->stripes[stripe_index].physical +
stripe_offset + stripe_nr * map->stripe_len;
- multi->stripes[i].dev = map->stripes[stripe_index].dev;
+ bbio->stripes[i].dev = map->stripes[stripe_index].dev;
if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
u64 stripes;
}
stripes = stripe_nr_end - 1 - j;
do_div(stripes, map->num_stripes);
- multi->stripes[i].length = map->stripe_len *
+ bbio->stripes[i].length = map->stripe_len *
(stripes - stripe_nr + 1);
if (i == 0) {
- multi->stripes[i].length -=
+ bbio->stripes[i].length -=
stripe_offset;
stripe_offset = 0;
}
if (stripe_index == last_stripe)
- multi->stripes[i].length -=
+ bbio->stripes[i].length -=
stripe_end_offset;
} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
u64 stripes;
}
stripes = stripe_nr_end - 1 - j;
do_div(stripes, factor);
- multi->stripes[i].length = map->stripe_len *
+ bbio->stripes[i].length = map->stripe_len *
(stripes - stripe_nr + 1);
if (i < map->sub_stripes) {
- multi->stripes[i].length -=
+ bbio->stripes[i].length -=
stripe_offset;
if (i == map->sub_stripes - 1)
stripe_offset = 0;
if (stripe_index >= last_stripe &&
stripe_index <= (last_stripe +
map->sub_stripes - 1)) {
- multi->stripes[i].length -=
+ bbio->stripes[i].length -=
stripe_end_offset;
}
} else
- multi->stripes[i].length = *length;
+ bbio->stripes[i].length = *length;
stripe_index++;
if (stripe_index == map->num_stripes) {
}
} else {
for (i = 0; i < num_stripes; i++) {
- multi->stripes[i].physical =
+ bbio->stripes[i].physical =
map->stripes[stripe_index].physical +
stripe_offset +
stripe_nr * map->stripe_len;
- multi->stripes[i].dev =
+ bbio->stripes[i].dev =
map->stripes[stripe_index].dev;
stripe_index++;
}
}
- if (multi_ret) {
- *multi_ret = multi;
- multi->num_stripes = num_stripes;
- multi->max_errors = max_errors;
+ if (bbio_ret) {
+ *bbio_ret = bbio;
+ bbio->num_stripes = num_stripes;
+ bbio->max_errors = max_errors;
+ bbio->mirror_num = mirror_num;
}
out:
free_extent_map(em);
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)
+ struct btrfs_bio **bbio_ret, int mirror_num)
{
- return __btrfs_map_block(map_tree, rw, logical, length, multi_ret,
+ return __btrfs_map_block(map_tree, rw, logical, length, bbio_ret,
mirror_num);
}
return 0;
}
- static void end_bio_multi_stripe(struct bio *bio, int err)
+ static void btrfs_end_bio(struct bio *bio, int err)
{
- struct btrfs_multi_bio *multi = bio->bi_private;
+ struct btrfs_bio *bbio = bio->bi_private;
int is_orig_bio = 0;
if (err)
- atomic_inc(&multi->error);
+ atomic_inc(&bbio->error);
- if (bio == multi->orig_bio)
+ if (bio == bbio->orig_bio)
is_orig_bio = 1;
- if (atomic_dec_and_test(&multi->stripes_pending)) {
+ if (atomic_dec_and_test(&bbio->stripes_pending)) {
if (!is_orig_bio) {
bio_put(bio);
- bio = multi->orig_bio;
+ bio = bbio->orig_bio;
}
- bio->bi_private = multi->private;
- bio->bi_end_io = multi->end_io;
+ bio->bi_private = bbio->private;
+ bio->bi_end_io = bbio->end_io;
+ bio->bi_bdev = (struct block_device *)
+ (unsigned long)bbio->mirror_num;
/* only send an error to the higher layers if it is
* beyond the tolerance of the multi-bio
*/
- if (atomic_read(&multi->error) > multi->max_errors) {
+ if (atomic_read(&bbio->error) > bbio->max_errors) {
err = -EIO;
} else if (err) {
/*
set_bit(BIO_UPTODATE, &bio->bi_flags);
err = 0;
}
- kfree(multi);
+ kfree(bbio);
bio_endio(bio, err);
} else if (!is_orig_bio) {
u64 logical = (u64)bio->bi_sector << 9;
u64 length = 0;
u64 map_length;
- struct btrfs_multi_bio *multi = NULL;
int ret;
int dev_nr = 0;
int total_devs = 1;
+ struct btrfs_bio *bbio = NULL;
length = bio->bi_size;
map_tree = &root->fs_info->mapping_tree;
map_length = length;
- ret = btrfs_map_block(map_tree, rw, logical, &map_length, &multi,
+ ret = btrfs_map_block(map_tree, rw, logical, &map_length, &bbio,
mirror_num);
BUG_ON(ret);
- total_devs = multi->num_stripes;
+ total_devs = bbio->num_stripes;
if (map_length < length) {
printk(KERN_CRIT "mapping failed logical %llu bio len %llu "
"len %llu\n", (unsigned long long)logical,
(unsigned long long)map_length);
BUG();
}
- multi->end_io = first_bio->bi_end_io;
- multi->private = first_bio->bi_private;
- multi->orig_bio = first_bio;
- atomic_set(&multi->stripes_pending, multi->num_stripes);
+
+ bbio->orig_bio = first_bio;
+ bbio->private = first_bio->bi_private;
+ bbio->end_io = first_bio->bi_end_io;
+ atomic_set(&bbio->stripes_pending, bbio->num_stripes);
while (dev_nr < total_devs) {
- if (total_devs > 1) {
- if (dev_nr < total_devs - 1) {
- bio = bio_clone(first_bio, GFP_NOFS);
- BUG_ON(!bio);
- } else {
- bio = first_bio;
- }
- bio->bi_private = multi;
- bio->bi_end_io = end_bio_multi_stripe;
+ if (dev_nr < total_devs - 1) {
+ bio = bio_clone(first_bio, GFP_NOFS);
+ BUG_ON(!bio);
+ } else {
+ bio = first_bio;
}
- bio->bi_sector = multi->stripes[dev_nr].physical >> 9;
- dev = multi->stripes[dev_nr].dev;
+ bio->bi_private = bbio;
+ bio->bi_end_io = btrfs_end_bio;
+ bio->bi_sector = bbio->stripes[dev_nr].physical >> 9;
+ dev = bbio->stripes[dev_nr].dev;
if (dev && dev->bdev && (rw != WRITE || dev->writeable)) {
+ pr_debug("btrfs_map_bio: rw %d, secor=%llu, dev=%lu "
+ "(%s id %llu), size=%u\n", rw,
+ (u64)bio->bi_sector, (u_long)dev->bdev->bd_dev,
+ dev->name, dev->devid, bio->bi_size);
bio->bi_bdev = dev->bdev;
if (async_submit)
schedule_bio(root, dev, rw, bio);
}
dev_nr++;
}
- if (total_devs == 1)
- kfree(multi);
return 0;
}
fill_device_from_item(leaf, dev_item, device);
device->dev_root = root->fs_info->dev_root;
device->in_fs_metadata = 1;
- if (device->writeable)
+ if (device->writeable) {
device->fs_devices->total_rw_bytes += device->total_bytes;
+ spin_lock(&root->fs_info->free_chunk_lock);
+ root->fs_info->free_chunk_space += device->total_bytes -
+ device->bytes_used;
+ spin_unlock(&root->fs_info->free_chunk_lock);
+ }
ret = 0;
return ret;
}
int btrfs_read_sys_array(struct btrfs_root *root)
{
- struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
+ struct btrfs_super_block *super_copy = root->fs_info->super_copy;
struct extent_buffer *sb;
struct btrfs_disk_key *disk_key;
struct btrfs_chunk *chunk;
struct btrfs_work work;
struct rcu_head rcu;
struct work_struct rcu_work;
+
+ /* readahead state */
+ spinlock_t reada_lock;
+ atomic_t reada_in_flight;
+ u64 reada_next;
+ struct reada_zone *reada_curr_zone;
+ struct radix_tree_root reada_zones;
+ struct radix_tree_root reada_extents;
};
struct btrfs_fs_devices {
u64 length; /* only used for discard mappings */
};
- struct btrfs_multi_bio {
+ struct btrfs_bio;
+ typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
+
+ struct btrfs_bio {
atomic_t stripes_pending;
bio_end_io_t *end_io;
struct bio *orig_bio;
atomic_t error;
int max_errors;
int num_stripes;
+ int mirror_num;
struct btrfs_bio_stripe stripes[];
};
int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
u64 end, u64 *length);
- #define btrfs_multi_bio_size(n) (sizeof(struct btrfs_multi_bio) + \
+ #define btrfs_bio_size(n) (sizeof(struct btrfs_bio) + \
(sizeof(struct btrfs_bio_stripe) * (n)))
int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
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);
+ struct btrfs_bio **bbio_ret, int mirror_num);
int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
u64 chunk_start, u64 physical, u64 devid,
u64 **logical, int *naddrs, int *stripe_len);
projects / linux-beck.git / commitdiff