Merge tag 'md/4.4-rc5-fixes' of git://neil.brown.name/md
[firefly-linux-kernel-4.4.55.git] / fs / btrfs / extent-tree.c
index acf3ed11cfb60e95b685aeb009e4d72fbdba3c3b..c4661db2b72ae4c412adaaecb7cbc236fbe1e7cf 100644 (file)
@@ -124,7 +124,7 @@ static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
        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);
 }
@@ -5915,19 +5915,6 @@ static int update_block_group(struct btrfs_trans_handle *trans,
                        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);
@@ -5939,6 +5926,22 @@ static int update_block_group(struct btrfs_trans_handle *trans,
                }
                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;
@@ -8105,21 +8108,47 @@ reada:
 }
 
 /*
- * 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);
 
@@ -8138,6 +8167,10 @@ static int account_leaf_items(struct btrfs_trans_handle *trans,
                        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;
 }
@@ -8206,8 +8239,6 @@ static int adjust_slots_upwards(struct btrfs_root *root,
 
 /*
  * 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,
@@ -8285,6 +8316,11 @@ walk_down:
                        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) {
@@ -10256,6 +10292,47 @@ out:
        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.
@@ -10322,8 +10399,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
                 * 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);
@@ -10403,11 +10480,15 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
                 * 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: