#include "raid56.h"
#include "locking.h"
#include "free-space-cache.h"
+#include "free-space-tree.h"
#include "math.h"
#include "sysfs.h"
#include "qgroup.h"
return (cache->flags & bits) == bits;
}
-static void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
+void btrfs_get_block_group(struct btrfs_block_group_cache *cache)
{
atomic_inc(&cache->count);
}
* we need to check the pinned_extents for any extents that can't be used yet
* since their free space will be released as soon as the transaction commits.
*/
-static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
- struct btrfs_fs_info *info, u64 start, u64 end)
+u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
+ struct btrfs_fs_info *info, u64 start, u64 end)
{
u64 extent_start, extent_end, size, total_added = 0;
int ret;
return total_added;
}
-static noinline void caching_thread(struct btrfs_work *work)
+static int load_extent_tree_free(struct btrfs_caching_control *caching_ctl)
{
struct btrfs_block_group_cache *block_group;
struct btrfs_fs_info *fs_info;
- struct btrfs_caching_control *caching_ctl;
struct btrfs_root *extent_root;
struct btrfs_path *path;
struct extent_buffer *leaf;
u64 total_found = 0;
u64 last = 0;
u32 nritems;
- int ret = -ENOMEM;
+ int ret;
bool wakeup = true;
- caching_ctl = container_of(work, struct btrfs_caching_control, work);
block_group = caching_ctl->block_group;
fs_info = block_group->fs_info;
extent_root = fs_info->extent_root;
path = btrfs_alloc_path();
if (!path)
- goto out;
+ return -ENOMEM;
last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
key.objectid = last;
key.offset = 0;
key.type = BTRFS_EXTENT_ITEM_KEY;
-again:
- mutex_lock(&caching_ctl->mutex);
- /* need to make sure the commit_root doesn't disappear */
- down_read(&fs_info->commit_root_sem);
next:
ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
if (ret < 0)
- goto err;
+ goto out;
leaf = path->nodes[0];
nritems = btrfs_header_nritems(leaf);
up_read(&fs_info->commit_root_sem);
mutex_unlock(&caching_ctl->mutex);
cond_resched();
- goto again;
+ mutex_lock(&caching_ctl->mutex);
+ down_read(&fs_info->commit_root_sem);
+ goto next;
}
ret = btrfs_next_leaf(extent_root, path);
if (ret < 0)
- goto err;
+ goto out;
if (ret)
break;
leaf = path->nodes[0];
else
last = key.objectid + key.offset;
- if (total_found > (1024 * 1024 * 2)) {
+ if (total_found > CACHING_CTL_WAKE_UP) {
total_found = 0;
if (wakeup)
wake_up(&caching_ctl->wait);
total_found += add_new_free_space(block_group, fs_info, last,
block_group->key.objectid +
block_group->key.offset);
+ caching_ctl->progress = (u64)-1;
+
+out:
+ btrfs_free_path(path);
+ return ret;
+}
+
+static noinline void caching_thread(struct btrfs_work *work)
+{
+ struct btrfs_block_group_cache *block_group;
+ struct btrfs_fs_info *fs_info;
+ struct btrfs_caching_control *caching_ctl;
+ int ret;
+
+ caching_ctl = container_of(work, struct btrfs_caching_control, work);
+ block_group = caching_ctl->block_group;
+ fs_info = block_group->fs_info;
+
+ mutex_lock(&caching_ctl->mutex);
+ down_read(&fs_info->commit_root_sem);
+
+ if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE))
+ ret = load_free_space_tree(caching_ctl);
+ else
+ ret = load_extent_tree_free(caching_ctl);
+
spin_lock(&block_group->lock);
block_group->caching_ctl = NULL;
- block_group->cached = BTRFS_CACHE_FINISHED;
+ block_group->cached = ret ? BTRFS_CACHE_ERROR : BTRFS_CACHE_FINISHED;
spin_unlock(&block_group->lock);
#ifdef CONFIG_BTRFS_DEBUG
#endif
caching_ctl->progress = (u64)-1;
-err:
- btrfs_free_path(path);
- up_read(&fs_info->commit_root_sem);
-
- free_excluded_extents(extent_root, block_group);
+ up_read(&fs_info->commit_root_sem);
+ free_excluded_extents(fs_info->extent_root, block_group);
mutex_unlock(&caching_ctl->mutex);
-out:
- if (ret) {
- spin_lock(&block_group->lock);
- block_group->caching_ctl = NULL;
- block_group->cached = BTRFS_CACHE_ERROR;
- spin_unlock(&block_group->lock);
- }
+
wake_up(&caching_ctl->wait);
put_caching_control(caching_ctl);
}
} else {
/*
- * We are not going to do the fast caching, set cached to the
- * appropriate value and wakeup any waiters.
+ * We're either using the free space tree or no caching at all.
+ * Set cached to the appropriate value and wakeup any waiters.
*/
spin_lock(&cache->lock);
if (load_cache_only) {
return -ENOMEM;
/*
- * We don't need the lock here since we are protected by the transaction
- * commit. We want to do the cache_save_setup first and then run the
+ * Even though we are in the critical section of the transaction commit,
+ * we can still have concurrent tasks adding elements to this
+ * transaction's list of dirty block groups. These tasks correspond to
+ * endio free space workers started when writeback finishes for a
+ * space cache, which run inode.c:btrfs_finish_ordered_io(), and can
+ * allocate new block groups as a result of COWing nodes of the root
+ * tree when updating the free space inode. The writeback for the space
+ * caches is triggered by an earlier call to
+ * btrfs_start_dirty_block_groups() and iterations of the following
+ * loop.
+ * Also we want to do the cache_save_setup first and then run the
* delayed refs to make sure we have the best chance at doing this all
* in one shot.
*/
+ spin_lock(&cur_trans->dirty_bgs_lock);
while (!list_empty(&cur_trans->dirty_bgs)) {
cache = list_first_entry(&cur_trans->dirty_bgs,
struct btrfs_block_group_cache,
* finish and then do it all again
*/
if (!list_empty(&cache->io_list)) {
+ spin_unlock(&cur_trans->dirty_bgs_lock);
list_del_init(&cache->io_list);
btrfs_wait_cache_io(root, trans, cache,
&cache->io_ctl, path,
cache->key.objectid);
btrfs_put_block_group(cache);
+ spin_lock(&cur_trans->dirty_bgs_lock);
}
/*
* on any pending IO
*/
list_del_init(&cache->dirty_list);
+ spin_unlock(&cur_trans->dirty_bgs_lock);
should_put = 1;
cache_save_setup(cache, trans, path);
/* if its not on the io list, we need to put the block group */
if (should_put)
btrfs_put_block_group(cache);
+ spin_lock(&cur_trans->dirty_bgs_lock);
}
+ spin_unlock(&cur_trans->dirty_bgs_lock);
while (!list_empty(io)) {
cache = list_first_entry(io, struct btrfs_block_group_cache,
set_extent_dirty(info->pinned_extents,
bytenr, bytenr + num_bytes - 1,
GFP_NOFS | __GFP_NOFAIL);
- /*
- * No longer have used bytes in this block group, queue
- * it for deletion.
- */
- if (old_val == 0) {
- spin_lock(&info->unused_bgs_lock);
- if (list_empty(&cache->bg_list)) {
- btrfs_get_block_group(cache);
- list_add_tail(&cache->bg_list,
- &info->unused_bgs);
- }
- spin_unlock(&info->unused_bgs_lock);
- }
}
spin_lock(&trans->transaction->dirty_bgs_lock);
}
spin_unlock(&trans->transaction->dirty_bgs_lock);
+ /*
+ * No longer have used bytes in this block group, queue it for
+ * deletion. We do this after adding the block group to the
+ * dirty list to avoid races between cleaner kthread and space
+ * cache writeout.
+ */
+ if (!alloc && old_val == 0) {
+ spin_lock(&info->unused_bgs_lock);
+ if (list_empty(&cache->bg_list)) {
+ btrfs_get_block_group(cache);
+ list_add_tail(&cache->bg_list,
+ &info->unused_bgs);
+ }
+ spin_unlock(&info->unused_bgs_lock);
+ }
+
btrfs_put_block_group(cache);
total -= num_bytes;
bytenr += num_bytes;
}
}
+ ret = add_to_free_space_tree(trans, root->fs_info, bytenr,
+ num_bytes);
+ if (ret) {
+ btrfs_abort_transaction(trans, extent_root, ret);
+ goto out;
+ }
+
ret = update_block_group(trans, root, bytenr, num_bytes, 0);
if (ret) {
btrfs_abort_transaction(trans, extent_root, ret);
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_free_path(path);
+ ret = remove_from_free_space_tree(trans, fs_info, ins->objectid,
+ ins->offset);
+ if (ret)
+ return ret;
+
ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
if (ret) { /* -ENOENT, logic error */
btrfs_err(fs_info, "update block group failed for %llu %llu",
btrfs_mark_buffer_dirty(leaf);
btrfs_free_path(path);
+ ret = remove_from_free_space_tree(trans, fs_info, ins->objectid,
+ num_bytes);
+ if (ret)
+ return ret;
+
ret = update_block_group(trans, root, ins->objectid, root->nodesize,
1);
if (ret) { /* -ENOENT, logic error */
clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
btrfs_set_lock_blocking(buf);
- btrfs_set_buffer_uptodate(buf);
+ set_extent_buffer_uptodate(buf);
if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
buf->log_index = root->log_transid % 2;
}
/*
- * TODO: Modify related function to add related node/leaf to dirty_extent_root,
- * for later qgroup accounting.
- *
- * Current, this function does nothing.
+ * These may not be seen by the usual inc/dec ref code so we have to
+ * add them here.
*/
+static int record_one_subtree_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, u64 bytenr,
+ u64 num_bytes)
+{
+ struct btrfs_qgroup_extent_record *qrecord;
+ struct btrfs_delayed_ref_root *delayed_refs;
+
+ qrecord = kmalloc(sizeof(*qrecord), GFP_NOFS);
+ if (!qrecord)
+ return -ENOMEM;
+
+ qrecord->bytenr = bytenr;
+ qrecord->num_bytes = num_bytes;
+ qrecord->old_roots = NULL;
+
+ delayed_refs = &trans->transaction->delayed_refs;
+ spin_lock(&delayed_refs->lock);
+ if (btrfs_qgroup_insert_dirty_extent(delayed_refs, qrecord))
+ kfree(qrecord);
+ spin_unlock(&delayed_refs->lock);
+
+ return 0;
+}
+
static int account_leaf_items(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *eb)
{
int nr = btrfs_header_nritems(eb);
- int i, extent_type;
+ int i, extent_type, ret;
struct btrfs_key key;
struct btrfs_file_extent_item *fi;
u64 bytenr, num_bytes;
+ /* We can be called directly from walk_up_proc() */
+ if (!root->fs_info->quota_enabled)
+ return 0;
+
for (i = 0; i < nr; i++) {
btrfs_item_key_to_cpu(eb, &key, i);
continue;
num_bytes = btrfs_file_extent_disk_num_bytes(eb, fi);
+
+ ret = record_one_subtree_extent(trans, root, bytenr, num_bytes);
+ if (ret)
+ return ret;
}
return 0;
}
/*
* root_eb is the subtree root and is locked before this function is called.
- * TODO: Modify this function to mark all (including complete shared node)
- * to dirty_extent_root to allow it get accounted in qgroup.
*/
static int account_shared_subtree(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
btrfs_tree_read_lock(eb);
btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
path->locks[level] = BTRFS_READ_LOCK_BLOCKING;
+
+ ret = record_one_subtree_extent(trans, root, child_bytenr,
+ root->nodesize);
+ if (ret)
+ goto out;
}
if (level == 0) {
cache->full_stripe_len = btrfs_full_stripe_len(root,
&root->fs_info->mapping_tree,
start);
+ set_free_space_tree_thresholds(cache);
+
atomic_set(&cache->count, 1);
spin_lock_init(&cache->lock);
init_rwsem(&cache->data_rwsem);
INIT_LIST_HEAD(&cache->io_list);
btrfs_init_free_space_ctl(cache);
atomic_set(&cache->trimming, 0);
+ mutex_init(&cache->free_space_lock);
return cache;
}
key.objectid, key.offset);
if (ret)
btrfs_abort_transaction(trans, extent_root, ret);
+ add_block_group_free_space(trans, root->fs_info, block_group);
+ /* already aborted the transaction if it failed. */
next:
list_del_init(&block_group->bg_list);
}
cache->flags = type;
cache->last_byte_to_unpin = (u64)-1;
cache->cached = BTRFS_CACHE_FINISHED;
+ cache->needs_free_space = 1;
ret = exclude_super_stripes(root, cache);
if (ret) {
/*
unlock_chunks(root);
+ ret = remove_block_group_free_space(trans, root->fs_info, block_group);
+ if (ret)
+ goto out;
+
btrfs_put_block_group(block_group);
btrfs_put_block_group(block_group);
return ret;
}
+struct btrfs_trans_handle *
+btrfs_start_trans_remove_block_group(struct btrfs_fs_info *fs_info,
+ const u64 chunk_offset)
+{
+ struct extent_map_tree *em_tree = &fs_info->mapping_tree.map_tree;
+ struct extent_map *em;
+ struct map_lookup *map;
+ unsigned int num_items;
+
+ read_lock(&em_tree->lock);
+ em = lookup_extent_mapping(em_tree, chunk_offset, 1);
+ read_unlock(&em_tree->lock);
+ ASSERT(em && em->start == chunk_offset);
+
+ /*
+ * We need to reserve 3 + N units from the metadata space info in order
+ * to remove a block group (done at btrfs_remove_chunk() and at
+ * btrfs_remove_block_group()), which are used for:
+ *
+ * 1 unit for adding the free space inode's orphan (located in the tree
+ * of tree roots).
+ * 1 unit for deleting the block group item (located in the extent
+ * tree).
+ * 1 unit for deleting the free space item (located in tree of tree
+ * roots).
+ * N units for deleting N device extent items corresponding to each
+ * stripe (located in the device tree).
+ *
+ * In order to remove a block group we also need to reserve units in the
+ * system space info in order to update the chunk tree (update one or
+ * more device items and remove one chunk item), but this is done at
+ * btrfs_remove_chunk() through a call to check_system_chunk().
+ */
+ map = (struct map_lookup *)em->bdev;
+ num_items = 3 + map->num_stripes;
+ free_extent_map(em);
+
+ return btrfs_start_transaction_fallback_global_rsv(fs_info->extent_root,
+ num_items, 1);
+}
+
/*
* Process the unused_bgs list and remove any that don't have any allocated
* space inside of them.
* Want to do this before we do anything else so we can recover
* properly if we fail to join the transaction.
*/
- /* 1 for btrfs_orphan_reserve_metadata() */
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_start_trans_remove_block_group(fs_info,
+ block_group->key.objectid);
if (IS_ERR(trans)) {
btrfs_dec_block_group_ro(root, block_group);
ret = PTR_ERR(trans);
* until transaction commit to do the actual discard.
*/
if (trimming) {
- WARN_ON(!list_empty(&block_group->bg_list));
- spin_lock(&trans->transaction->deleted_bgs_lock);
+ spin_lock(&fs_info->unused_bgs_lock);
+ /*
+ * A concurrent scrub might have added us to the list
+ * fs_info->unused_bgs, so use a list_move operation
+ * to add the block group to the deleted_bgs list.
+ */
list_move(&block_group->bg_list,
&trans->transaction->deleted_bgs);
- spin_unlock(&trans->transaction->deleted_bgs_lock);
+ spin_unlock(&fs_info->unused_bgs_lock);
btrfs_get_block_group(block_group);
}
end_trans:
projects / deliverable / linux.git / blobdiff