Merge tag 'for-5.0-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:

 - regression fix: transaction commit can run away due to delayed ref
   waiting heuristic, this is not necessary now because of the proper
   reservation mechanism introduced in 5.0

 - regression fix: potential crash due to use-before-check of an ERR_PTR
   return value

 - fix for transaction abort during transaction commit that needs to
   properly clean up pending block groups

 - fix deadlock during b-tree node/leaf splitting, when this happens on
   some of the fundamental trees, we must prevent new tree block
   allocation to re-enter indirectly via the block group flushing path

 - potential memory leak after errors during mount

* tag 'for-5.0-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: On error always free subvol_name in btrfs_mount
  btrfs: clean up pending block groups when transaction commit aborts
  btrfs: fix potential oops in device_list_add
  btrfs: don't end the transaction for delayed refs in throttle
  Btrfs: fix deadlock when allocating tree block during leaf/node split

4 files changed, 71 insertions, 38 deletions

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index f64aad613727..5a6c39b44c84 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -968,6 +968,48 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
         return 0;
 }
 
+static struct extent_buffer *alloc_tree_block_no_bg_flush(
+                                          struct btrfs_trans_handle *trans,
+                                          struct btrfs_root *root,
+                                          u64 parent_start,
+                                          const struct btrfs_disk_key *disk_key,
+                                          int level,
+                                          u64 hint,
+                                          u64 empty_size)
+{
+        struct btrfs_fs_info *fs_info = root->fs_info;
+        struct extent_buffer *ret;
+
+        /*
+         * If we are COWing a node/leaf from the extent, chunk, device or free
+         * space trees, make sure that we do not finish block group creation of
+         * pending block groups. We do this to avoid a deadlock.
+         * COWing can result in allocation of a new chunk, and flushing pending
+         * block groups (btrfs_create_pending_block_groups()) can be triggered
+         * when finishing allocation of a new chunk. Creation of a pending block
+         * group modifies the extent, chunk, device and free space trees,
+         * therefore we could deadlock with ourselves since we are holding a
+         * lock on an extent buffer that btrfs_create_pending_block_groups() may
+         * try to COW later.
+         * For similar reasons, we also need to delay flushing pending block
+         * groups when splitting a leaf or node, from one of those trees, since
+         * we are holding a write lock on it and its parent or when inserting a
+         * new root node for one of those trees.
+         */
+        if (root == fs_info->extent_root ||
+            root == fs_info->chunk_root ||
+            root == fs_info->dev_root ||
+            root == fs_info->free_space_root)
+                trans->can_flush_pending_bgs = false;
+
+        ret = btrfs_alloc_tree_block(trans, root, parent_start,
+                                     root->root_key.objectid, disk_key, level,
+                                     hint, empty_size);
+        trans->can_flush_pending_bgs = true;
+
+        return ret;
+}
+
 /*
  * does the dirty work in cow of a single block.  The parent block (if
  * supplied) is updated to point to the new cow copy.  The new buffer is marked
@@ -1015,28 +1057,8 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
         if ((root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) && parent)
                 parent_start = parent->start;
 
-        /*
-         * If we are COWing a node/leaf from the extent, chunk, device or free
-         * space trees, make sure that we do not finish block group creation of
-         * pending block groups. We do this to avoid a deadlock.
-         * COWing can result in allocation of a new chunk, and flushing pending
-         * block groups (btrfs_create_pending_block_groups()) can be triggered
-         * when finishing allocation of a new chunk. Creation of a pending block
-         * group modifies the extent, chunk, device and free space trees,
-         * therefore we could deadlock with ourselves since we are holding a
-         * lock on an extent buffer that btrfs_create_pending_block_groups() may
-         * try to COW later.
-         */
-        if (root == fs_info->extent_root ||
-            root == fs_info->chunk_root ||
-            root == fs_info->dev_root ||
-            root == fs_info->free_space_root)
-                trans->can_flush_pending_bgs = false;
-
-        cow = btrfs_alloc_tree_block(trans, root, parent_start,
-                        root->root_key.objectid, &disk_key, level,
-                        search_start, empty_size);
-        trans->can_flush_pending_bgs = true;
+        cow = alloc_tree_block_no_bg_flush(trans, root, parent_start, &disk_key,
+                                           level, search_start, empty_size);
         if (IS_ERR(cow))
                 return PTR_ERR(cow);
 
@@ -3345,8 +3367,8 @@ static noinline int insert_new_root(struct btrfs_trans_handle *trans,
         else
                 btrfs_node_key(lower, &lower_key, 0);
 
-        c = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
-                                   &lower_key, level, root->node->start, 0);
+        c = alloc_tree_block_no_bg_flush(trans, root, 0, &lower_key, level,
+                                         root->node->start, 0);
         if (IS_ERR(c))
                 return PTR_ERR(c);
 
@@ -3475,8 +3497,8 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
         mid = (c_nritems + 1) / 2;
         btrfs_node_key(c, &disk_key, mid);
 
-        split = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
-                        &disk_key, level, c->start, 0);
+        split = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, level,
+                                             c->start, 0);
         if (IS_ERR(split))
                 return PTR_ERR(split);
 
@@ -4260,8 +4282,8 @@ again:
         else
                 btrfs_item_key(l, &disk_key, mid);
 
-        right = btrfs_alloc_tree_block(trans, root, 0, root->root_key.objectid,
-                        &disk_key, 0, l->start, 0);
+        right = alloc_tree_block_no_bg_flush(trans, root, 0, &disk_key, 0,
+                                             l->start, 0);
         if (IS_ERR(right))
                 return PTR_ERR(right);
 
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index c5586ffd1426..0a3f122dd61f 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1621,6 +1621,7 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
                                 flags | SB_RDONLY, device_name, data);
                         if (IS_ERR(mnt_root)) {
                                 root = ERR_CAST(mnt_root);
+                                kfree(subvol_name);
                                 goto out;
                         }
 
@@ -1630,12 +1631,14 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
                         if (error < 0) {
                                 root = ERR_PTR(error);
                                 mntput(mnt_root);
+                                kfree(subvol_name);
                                 goto out;
                         }
                 }
         }
         if (IS_ERR(mnt_root)) {
                 root = ERR_CAST(mnt_root);
+                kfree(subvol_name);
                 goto out;
         }
 
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 127fa1535f58..4ec2b660d014 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -850,14 +850,6 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 
         btrfs_trans_release_chunk_metadata(trans);
 
-        if (lock && should_end_transaction(trans) &&
-            READ_ONCE(cur_trans->state) == TRANS_STATE_RUNNING) {
-                spin_lock(&info->trans_lock);
-                if (cur_trans->state == TRANS_STATE_RUNNING)
-                        cur_trans->state = TRANS_STATE_BLOCKED;
-                spin_unlock(&info->trans_lock);
-        }
-
         if (lock && READ_ONCE(cur_trans->state) == TRANS_STATE_BLOCKED) {
                 if (throttle)
                         return btrfs_commit_transaction(trans);
@@ -1879,6 +1871,21 @@ static void cleanup_transaction(struct btrfs_trans_handle *trans, int err)
         kmem_cache_free(btrfs_trans_handle_cachep, trans);
 }
 
+/*
+ * Release reserved delayed ref space of all pending block groups of the
+ * transaction and remove them from the list
+ */
+static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans)
+{
+       struct btrfs_fs_info *fs_info = trans->fs_info;
+       struct btrfs_block_group_cache *block_group, *tmp;
+
+       list_for_each_entry_safe(block_group, tmp, &trans->new_bgs, bg_list) {
+               btrfs_delayed_refs_rsv_release(fs_info, 1);
+               list_del_init(&block_group->bg_list);
+       }
+}
+
 static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info)
 {
         /*
@@ -2270,6 +2277,7 @@ scrub_continue:
         btrfs_scrub_continue(fs_info);
 cleanup_transaction:
         btrfs_trans_release_metadata(trans);
+        btrfs_cleanup_pending_block_groups(trans);
         btrfs_trans_release_chunk_metadata(trans);
         trans->block_rsv = NULL;
         btrfs_warn(fs_info, "Skipping commit of aborted transaction.");
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 3e4f8f88353e..15561926ab32 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -957,11 +957,11 @@ static noinline struct btrfs_device *device_list_add(const char *path,
                 else
                         fs_devices = alloc_fs_devices(disk_super->fsid, NULL);
 
-                fs_devices->fsid_change = fsid_change_in_progress;
-
                 if (IS_ERR(fs_devices))
                         return ERR_CAST(fs_devices);
 
+                fs_devices->fsid_change = fsid_change_in_progress;
+
                 mutex_lock(&fs_devices->device_list_mutex);
                 list_add(&fs_devices->fs_list, &fs_uuids);