Update Linux to v5.4.148

Sourced from [1]

[1] https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.4.148.tar.gz

Change-Id: Ib3d26c5ba9b022e2e03533005c4fed4d7c30b61b
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 5cd42b6..ba68b0b 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -517,6 +517,34 @@
 	return 1;
 }
 
+static bool reloc_root_is_dead(struct btrfs_root *root)
+{
+	/*
+	 * Pair with set_bit/clear_bit in clean_dirty_subvols and
+	 * btrfs_update_reloc_root. We need to see the updated bit before
+	 * trying to access reloc_root
+	 */
+	smp_rmb();
+	if (test_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state))
+		return true;
+	return false;
+}
+
+/*
+ * Check if this subvolume tree has valid reloc tree.
+ *
+ * Reloc tree after swap is considered dead, thus not considered as valid.
+ * This is enough for most callers, as they don't distinguish dead reloc root
+ * from no reloc root.  But should_ignore_root() below is a special case.
+ */
+static bool have_reloc_root(struct btrfs_root *root)
+{
+	if (reloc_root_is_dead(root))
+		return false;
+	if (!root->reloc_root)
+		return false;
+	return true;
+}
 
 static int should_ignore_root(struct btrfs_root *root)
 {
@@ -525,12 +553,16 @@
 	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
 		return 0;
 
+	/* This root has been merged with its reloc tree, we can ignore it */
+	if (reloc_root_is_dead(root))
+		return 1;
+
 	reloc_root = root->reloc_root;
 	if (!reloc_root)
 		return 0;
 
-	if (btrfs_root_last_snapshot(&reloc_root->root_item) ==
-	    root->fs_info->running_transaction->transid - 1)
+	if (btrfs_header_generation(reloc_root->commit_root) ==
+	    root->fs_info->running_transaction->transid)
 		return 0;
 	/*
 	 * if there is reloc tree and it was created in previous
@@ -1154,7 +1186,7 @@
 			free_backref_node(cache, lower);
 		}
 
-		free_backref_node(cache, node);
+		remove_backref_node(cache, node);
 		return ERR_PTR(err);
 	}
 	ASSERT(!node || !node->detached);
@@ -1266,7 +1298,7 @@
 	if (!node)
 		return -ENOMEM;
 
-	node->bytenr = root->node->start;
+	node->bytenr = root->commit_root->start;
 	node->data = root;
 
 	spin_lock(&rc->reloc_root_tree.lock);
@@ -1297,15 +1329,14 @@
 	if (rc && root->node) {
 		spin_lock(&rc->reloc_root_tree.lock);
 		rb_node = tree_search(&rc->reloc_root_tree.rb_root,
-				      root->node->start);
+				      root->commit_root->start);
 		if (rb_node) {
 			node = rb_entry(rb_node, struct mapping_node, rb_node);
 			rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
+			RB_CLEAR_NODE(&node->rb_node);
 		}
 		spin_unlock(&rc->reloc_root_tree.lock);
-		if (!node)
-			return;
-		BUG_ON((struct btrfs_root *)node->data != root);
+		ASSERT(!node || (struct btrfs_root *)node->data == root);
 	}
 
 	spin_lock(&fs_info->trans_lock);
@@ -1318,7 +1349,7 @@
  * helper to update the 'address of tree root -> reloc tree'
  * mapping
  */
-static int __update_reloc_root(struct btrfs_root *root, u64 new_bytenr)
+static int __update_reloc_root(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct rb_node *rb_node;
@@ -1327,7 +1358,7 @@
 
 	spin_lock(&rc->reloc_root_tree.lock);
 	rb_node = tree_search(&rc->reloc_root_tree.rb_root,
-			      root->node->start);
+			      root->commit_root->start);
 	if (rb_node) {
 		node = rb_entry(rb_node, struct mapping_node, rb_node);
 		rb_erase(&node->rb_node, &rc->reloc_root_tree.rb_root);
@@ -1339,7 +1370,7 @@
 	BUG_ON((struct btrfs_root *)node->data != root);
 
 	spin_lock(&rc->reloc_root_tree.lock);
-	node->bytenr = new_bytenr;
+	node->bytenr = root->node->start;
 	rb_node = tree_insert(&rc->reloc_root_tree.rb_root,
 			      node->bytenr, &node->rb_node);
 	spin_unlock(&rc->reloc_root_tree.lock);
@@ -1435,20 +1466,35 @@
 	int clear_rsv = 0;
 	int ret;
 
+	if (!rc)
+		return 0;
+
 	/*
 	 * The subvolume has reloc tree but the swap is finished, no need to
 	 * create/update the dead reloc tree
 	 */
-	if (test_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state))
+	if (reloc_root_is_dead(root))
 		return 0;
 
+	/*
+	 * This is subtle but important.  We do not do
+	 * record_root_in_transaction for reloc roots, instead we record their
+	 * corresponding fs root, and then here we update the last trans for the
+	 * reloc root.  This means that we have to do this for the entire life
+	 * of the reloc root, regardless of which stage of the relocation we are
+	 * in.
+	 */
 	if (root->reloc_root) {
 		reloc_root = root->reloc_root;
 		reloc_root->last_trans = trans->transid;
 		return 0;
 	}
 
-	if (!rc || !rc->create_reloc_tree ||
+	/*
+	 * We are merging reloc roots, we do not need new reloc trees.  Also
+	 * reloc trees never need their own reloc tree.
+	 */
+	if (!rc->create_reloc_tree ||
 	    root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID)
 		return 0;
 
@@ -1478,8 +1524,7 @@
 	struct btrfs_root_item *root_item;
 	int ret;
 
-	if (test_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state) ||
-	    !root->reloc_root)
+	if (!have_reloc_root(root))
 		goto out;
 
 	reloc_root = root->reloc_root;
@@ -1489,10 +1534,16 @@
 	if (fs_info->reloc_ctl->merge_reloc_tree &&
 	    btrfs_root_refs(root_item) == 0) {
 		set_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
+		/*
+		 * Mark the tree as dead before we change reloc_root so
+		 * have_reloc_root will not touch it from now on.
+		 */
+		smp_wmb();
 		__del_reloc_root(reloc_root);
 	}
 
 	if (reloc_root->commit_root != reloc_root->node) {
+		__update_reloc_root(reloc_root);
 		btrfs_set_root_node(root_item, reloc_root->node);
 		free_extent_buffer(reloc_root->commit_root);
 		reloc_root->commit_root = btrfs_root_node(reloc_root);
@@ -1785,8 +1836,8 @@
 	int ret;
 	int slot;
 
-	BUG_ON(src->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
-	BUG_ON(dest->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID);
+	ASSERT(src->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID);
+	ASSERT(dest->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID);
 
 	last_snapshot = btrfs_root_last_snapshot(&src->root_item);
 again:
@@ -1820,7 +1871,7 @@
 		struct btrfs_key first_key;
 
 		level = btrfs_header_level(parent);
-		BUG_ON(level < lowest_level);
+		ASSERT(level >= lowest_level);
 
 		ret = btrfs_bin_search(parent, &key, level, &slot);
 		if (ret < 0)
@@ -2202,6 +2253,11 @@
 				if (ret2 < 0 && !ret)
 					ret = ret2;
 			}
+			/*
+			 * Need barrier to ensure clear_bit() only happens after
+			 * root->reloc_root = NULL. Pairs with have_reloc_root.
+			 */
+			smp_wmb();
 			clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
 			btrfs_put_fs_root(root);
 		} else {
@@ -2229,6 +2285,7 @@
 	struct btrfs_root_item *root_item;
 	struct btrfs_path *path;
 	struct extent_buffer *leaf;
+	int reserve_level;
 	int level;
 	int max_level;
 	int replaced = 0;
@@ -2269,12 +2326,21 @@
 		btrfs_unlock_up_safe(path, 0);
 	}
 
-	min_reserved = fs_info->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
+	/*
+	 * In merge_reloc_root(), we modify the upper level pointer to swap the
+	 * tree blocks between reloc tree and subvolume tree.  Thus for tree
+	 * block COW, we COW at most from level 1 to root level for each tree.
+	 *
+	 * Thus the needed metadata size is at most root_level * nodesize,
+	 * and * 2 since we have two trees to COW.
+	 */
+	reserve_level = max_t(int, 1, btrfs_root_level(root_item));
+	min_reserved = fs_info->nodesize * reserve_level * 2;
 	memset(&next_key, 0, sizeof(next_key));
 
 	while (1) {
 		ret = btrfs_block_rsv_refill(root, rc->block_rsv, min_reserved,
-					     BTRFS_RESERVE_FLUSH_ALL);
+					     BTRFS_RESERVE_FLUSH_LIMIT);
 		if (ret) {
 			err = ret;
 			goto out;
@@ -2285,6 +2351,18 @@
 			trans = NULL;
 			goto out;
 		}
+
+		/*
+		 * At this point we no longer have a reloc_control, so we can't
+		 * depend on btrfs_init_reloc_root to update our last_trans.
+		 *
+		 * But that's ok, we started the trans handle on our
+		 * corresponding fs_root, which means it's been added to the
+		 * dirty list.  At commit time we'll still call
+		 * btrfs_update_reloc_root() and update our root item
+		 * appropriately.
+		 */
+		reloc_root->last_trans = trans->transid;
 		trans->block_rsv = rc->block_rsv;
 
 		replaced = 0;
@@ -2482,12 +2560,10 @@
 		reloc_root = list_entry(reloc_roots.next,
 					struct btrfs_root, root_list);
 
+		root = read_fs_root(fs_info, reloc_root->root_key.offset);
 		if (btrfs_root_refs(&reloc_root->root_item) > 0) {
-			root = read_fs_root(fs_info,
-					    reloc_root->root_key.offset);
 			BUG_ON(IS_ERR(root));
 			BUG_ON(root->reloc_root != reloc_root);
-
 			ret = merge_reloc_root(rc, root);
 			if (ret) {
 				if (list_empty(&reloc_root->root_list))
@@ -2496,6 +2572,13 @@
 				goto out;
 			}
 		} else {
+			if (!IS_ERR(root)) {
+				if (root->reloc_root == reloc_root)
+					root->reloc_root = NULL;
+				clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE,
+					  &root->state);
+			}
+
 			list_del_init(&reloc_root->root_list);
 			/* Don't forget to queue this reloc root for cleanup */
 			list_add_tail(&reloc_root->reloc_dirty_list,
@@ -2521,7 +2604,21 @@
 			free_reloc_roots(&reloc_roots);
 	}
 
-	BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	/*
+	 * We used to have
+	 *
+	 * BUG_ON(!RB_EMPTY_ROOT(&rc->reloc_root_tree.rb_root));
+	 *
+	 * here, but it's wrong.  If we fail to start the transaction in
+	 * prepare_to_merge() we will have only 0 ref reloc roots, none of which
+	 * have actually been removed from the reloc_root_tree rb tree.  This is
+	 * fine because we're bailing here, and we hold a reference on the root
+	 * for the list that holds it, so these roots will be cleaned up when we
+	 * do the reloc_dirty_list afterwards.  Meanwhile the root->reloc_root
+	 * will be cleaned up on unmount.
+	 *
+	 * The remaining nodes will be cleaned up by free_reloc_control.
+	 */
 }
 
 static void free_block_list(struct rb_root *blocks)
@@ -3121,9 +3218,8 @@
 		ret = relocate_tree_block(trans, rc, node, &block->key,
 					  path);
 		if (ret < 0) {
-			if (ret != -EAGAIN || &block->rb_node == rb_first(blocks))
-				err = ret;
-			goto out;
+			err = ret;
+			break;
 		}
 	}
 out:
@@ -4099,12 +4195,6 @@
 		if (!RB_EMPTY_ROOT(&blocks)) {
 			ret = relocate_tree_blocks(trans, rc, &blocks);
 			if (ret < 0) {
-				/*
-				 * if we fail to relocate tree blocks, force to update
-				 * backref cache when committing transaction.
-				 */
-				rc->backref_cache.last_trans = trans->transid - 1;
-
 				if (ret != -EAGAIN) {
 					err = ret;
 					break;
@@ -4174,10 +4264,10 @@
 		goto out_free;
 	}
 	btrfs_commit_transaction(trans);
+out_free:
 	ret = clean_dirty_subvols(rc);
 	if (ret < 0 && !err)
 		err = ret;
-out_free:
 	btrfs_free_block_rsv(fs_info, rc->block_rsv);
 	btrfs_free_path(path);
 	return err;
@@ -4279,6 +4369,18 @@
 	return rc;
 }
 
+static void free_reloc_control(struct reloc_control *rc)
+{
+	struct mapping_node *node, *tmp;
+
+	free_reloc_roots(&rc->reloc_roots);
+	rbtree_postorder_for_each_entry_safe(node, tmp,
+			&rc->reloc_root_tree.rb_root, rb_node)
+		kfree(node);
+
+	kfree(rc);
+}
+
 /*
  * Print the block group being relocated
  */
@@ -4326,7 +4428,7 @@
 	rc->extent_root = extent_root;
 	rc->block_group = bg;
 
-	ret = btrfs_inc_block_group_ro(rc->block_group);
+	ret = btrfs_inc_block_group_ro(rc->block_group, true);
 	if (ret) {
 		err = ret;
 		goto out;
@@ -4411,7 +4513,7 @@
 		btrfs_dec_block_group_ro(rc->block_group);
 	iput(rc->data_inode);
 	btrfs_put_block_group(rc->block_group);
-	kfree(rc);
+	free_reloc_control(rc);
 	return err;
 }
 
@@ -4534,9 +4636,8 @@
 
 	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
-		unset_reloc_control(rc);
 		err = PTR_ERR(trans);
-		goto out_free;
+		goto out_unset;
 	}
 
 	rc->merge_reloc_tree = 1;
@@ -4555,7 +4656,9 @@
 		fs_root = read_fs_root(fs_info, reloc_root->root_key.offset);
 		if (IS_ERR(fs_root)) {
 			err = PTR_ERR(fs_root);
-			goto out_free;
+			list_add_tail(&reloc_root->root_list, &reloc_roots);
+			btrfs_end_transaction(trans);
+			goto out_unset;
 		}
 
 		err = __add_reloc_root(reloc_root);
@@ -4565,7 +4668,7 @@
 
 	err = btrfs_commit_transaction(trans);
 	if (err)
-		goto out_free;
+		goto out_unset;
 
 	merge_reloc_roots(rc);
 
@@ -4574,15 +4677,16 @@
 	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
 		err = PTR_ERR(trans);
-		goto out_free;
+		goto out_clean;
 	}
 	err = btrfs_commit_transaction(trans);
-
+out_clean:
 	ret = clean_dirty_subvols(rc);
 	if (ret < 0 && !err)
 		err = ret;
-out_free:
-	kfree(rc);
+out_unset:
+	unset_reloc_control(rc);
+	free_reloc_control(rc);
 out:
 	if (!list_empty(&reloc_roots))
 		free_reloc_roots(&reloc_roots);
@@ -4669,11 +4773,6 @@
 	BUG_ON(rc->stage == UPDATE_DATA_PTRS &&
 	       root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID);
 
-	if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) {
-		if (buf == root->node)
-			__update_reloc_root(root, cow->start);
-	}
-
 	level = btrfs_header_level(buf);
 	if (btrfs_header_generation(buf) <=
 	    btrfs_root_last_snapshot(&root->root_item))
@@ -4720,7 +4819,7 @@
 	struct btrfs_root *root = pending->root;
 	struct reloc_control *rc = root->fs_info->reloc_ctl;
 
-	if (!root->reloc_root || !rc)
+	if (!rc || !have_reloc_root(root))
 		return;
 
 	if (!rc->merge_reloc_tree)
@@ -4754,7 +4853,7 @@
 	struct reloc_control *rc = root->fs_info->reloc_ctl;
 	int ret;
 
-	if (!root->reloc_root || !rc)
+	if (!rc || !have_reloc_root(root))
 		return 0;
 
 	rc = root->fs_info->reloc_ctl;