btrfs: kill the subvol_srcu

Now that we have proper root ref counting everywhere we can kill the
subvol_srcu.

* removal of fs_info::subvol_srcu reduces size of fs_info by 1176 bytes

* the refcount_t used for the references checks for accidental 0->1
  in cases where the root lifetime would not be properly protected

* there's a leak detector for roots to catch unfreed roots at umount
  time

* SRCU served us well over the years but is was not a proper
  synchronization mechanism for some cases
Signed-off-by: NJosef Bacik <josef@toxicpanda.com>
Reviewed-by: NDavid Sterba <dsterba@suse.com>
[ update changelog ]
Signed-off-by: NDavid Sterba <dsterba@suse.com>

fs/btrfs/backref.c

@@ -542,24 +542,19 @@ static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	int ret = 0;
 	int root_level;
 	int level = ref->level;
-	int index;
 	struct btrfs_key search_key = ref->key_for_search;
 
 	root_key.objectid = ref->root_id;
 	root_key.type = BTRFS_ROOT_ITEM_KEY;
 	root_key.offset = (u64)-1;
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
-
 	root = btrfs_get_fs_root(fs_info, &root_key, false);
 	if (IS_ERR(root)) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		ret = PTR_ERR(root);
 		goto out_free;
 	}
 
 	if (btrfs_is_testing(fs_info)) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 		ret = -ENOENT;
 		goto out;
 	}
@@ -571,10 +566,8 @@ static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	else
 		root_level = btrfs_old_root_level(root, time_seq);
 
-	if (root_level + 1 == level) {
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
+	if (root_level + 1 == level)
 		goto out;
-	}
 
 	/*
 	 * We can often find data backrefs with an offset that is too large
@@ -604,9 +597,6 @@ static int resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	else
 		ret = btrfs_search_old_slot(root, &search_key, path, time_seq);
 
-	/* root node has been locked, we can release @subvol_srcu safely here */
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
-
 	btrfs_debug(fs_info,
 		"search slot in root %llu (level %d, ref count %d) returned %d for key (%llu %u %llu)",
 		 ref->root_id, level, ref->count, ret,

fs/btrfs/ctree.h

@@ -697,7 +697,6 @@ struct btrfs_fs_info {
 	struct rw_semaphore cleanup_work_sem;
 
 	struct rw_semaphore subvol_sem;
-	struct srcu_struct subvol_srcu;
 
 	spinlock_t trans_lock;
 	/*

fs/btrfs/disk-io.c

@@ -2757,46 +2757,33 @@ static int init_mount_fs_info(struct btrfs_fs_info *fs_info, struct super_block
 	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
 	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
 
-	ret = init_srcu_struct(&fs_info->subvol_srcu);
-	if (ret)
-		return ret;
-
 	ret = percpu_counter_init(&fs_info->dio_bytes, 0, GFP_KERNEL);
 	if (ret)
-		goto fail;
+		return ret;
 
 	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
 	if (ret)
-		goto fail;
+		return ret;
 
 	fs_info->dirty_metadata_batch = PAGE_SIZE *
 					(1 + ilog2(nr_cpu_ids));
 
 	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
 	if (ret)
-		goto fail;
+		return ret;
 
 	ret = percpu_counter_init(&fs_info->dev_replace.bio_counter, 0,
 			GFP_KERNEL);
 	if (ret)
-		goto fail;
+		return ret;
 
 	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
 					GFP_KERNEL);
-	if (!fs_info->delayed_root) {
-		ret = -ENOMEM;
-		goto fail;
-	}
+	if (!fs_info->delayed_root)
+		return -ENOMEM;
 	btrfs_init_delayed_root(fs_info->delayed_root);
 
-	ret = btrfs_alloc_stripe_hash_table(fs_info);
-	if (ret)
-		goto fail;
-
-	return 0;
-fail:
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
-	return ret;
+	return btrfs_alloc_stripe_hash_table(fs_info);
 }
 
 static int btrfs_uuid_rescan_kthread(void *data)
@@ -2870,13 +2857,13 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	fs_info->chunk_root = chunk_root;
 	if (!tree_root || !chunk_root) {
 		err = -ENOMEM;
-		goto fail_srcu;
+		goto fail;
 	}
 
 	fs_info->btree_inode = new_inode(sb);
 	if (!fs_info->btree_inode) {
 		err = -ENOMEM;
-		goto fail_srcu;
+		goto fail;
 	}
 	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
 	btrfs_init_btree_inode(fs_info);
@@ -3398,8 +3385,6 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 
 	iput(fs_info->btree_inode);
-fail_srcu:
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
 fail:
 	btrfs_close_devices(fs_info->fs_devices);
 	return err;
@@ -3902,9 +3887,6 @@ void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info,
 		drop_ref = true;
 	spin_unlock(&fs_info->fs_roots_radix_lock);
 
-	if (btrfs_root_refs(&root->root_item) == 0)
-		synchronize_srcu(&fs_info->subvol_srcu);
-
 	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
 		btrfs_free_log(NULL, root);
 		if (root->reloc_root) {
@@ -4116,7 +4098,6 @@ void __cold close_ctree(struct btrfs_fs_info *fs_info)
 
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 	btrfs_close_devices(fs_info->fs_devices);
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
 }
 
 int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,

fs/btrfs/export.c

@@ -65,8 +65,6 @@ struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
 	struct btrfs_root *root;
 	struct inode *inode;
 	struct btrfs_key key;
-	int index;
-	int err = 0;
 
 	if (objectid < BTRFS_FIRST_FREE_OBJECTID)
 		return ERR_PTR(-ESTALE);
@@ -75,13 +73,9 @@ struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
 	key.type = BTRFS_ROOT_ITEM_KEY;
 	key.offset = (u64)-1;
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
-
 	root = btrfs_get_fs_root(fs_info, &key, true);
-	if (IS_ERR(root)) {
-		err = PTR_ERR(root);
-		goto fail;
-	}
+	if (IS_ERR(root))
+		return ERR_CAST(root);
 
 	key.objectid = objectid;
 	key.type = BTRFS_INODE_ITEM_KEY;
@@ -89,12 +83,8 @@ struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
 
 	inode = btrfs_iget(sb, &key, root);
 	btrfs_put_root(root);
-	if (IS_ERR(inode)) {
-		err = PTR_ERR(inode);
-		goto fail;
-	}
-
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
 
 	if (check_generation && generation != inode->i_generation) {
 		iput(inode);
@@ -102,9 +92,6 @@ struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid,
 	}
 
 	return d_obtain_alias(inode);
-fail:
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
-	return ERR_PTR(err);
 }
 
 static struct dentry *btrfs_fh_to_parent(struct super_block *sb, struct fid *fh,

fs/btrfs/file.c

@@ -278,7 +278,6 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 	struct btrfs_key key;
 	struct btrfs_ioctl_defrag_range_args range;
 	int num_defrag;
-	int index;
 	int ret;
 
 	/* get the inode */
@@ -286,8 +285,6 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 	key.type = BTRFS_ROOT_ITEM_KEY;
 	key.offset = (u64)-1;
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
-
 	inode_root = btrfs_get_fs_root(fs_info, &key, true);
 	if (IS_ERR(inode_root)) {
 		ret = PTR_ERR(inode_root);
@@ -303,7 +300,6 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 		ret = PTR_ERR(inode);
 		goto cleanup;
 	}
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
 
 	/* do a chunk of defrag */
 	clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags);
@@ -339,7 +335,6 @@ static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 	iput(inode);
 	return 0;
 cleanup:
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
 	kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 	return ret;
 }

fs/btrfs/inode.c

@@ -5364,7 +5364,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 	struct btrfs_root *sub_root = root;
 	struct btrfs_key location;
 	u8 di_type = 0;
-	int index;
 	int ret = 0;
 
 	if (dentry->d_name.len > BTRFS_NAME_LEN)
@@ -5391,7 +5390,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 		return inode;
 	}
 
-	index = srcu_read_lock(&fs_info->subvol_srcu);
 	ret = fixup_tree_root_location(fs_info, dir, dentry,
 				       &location, &sub_root);
 	if (ret < 0) {
@@ -5404,7 +5402,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 	}
 	if (root != sub_root)
 		btrfs_put_root(sub_root);
-	srcu_read_unlock(&fs_info->subvol_srcu, index);
 
 	if (!IS_ERR(inode) && root != sub_root) {
 		down_read(&fs_info->cleanup_work_sem);

fs/btrfs/send.c

@@ -7028,7 +7028,6 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 	int clone_sources_to_rollback = 0;
 	unsigned alloc_size;
 	int sort_clone_roots = 0;
-	int index;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
@@ -7155,11 +7154,8 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 			key.type = BTRFS_ROOT_ITEM_KEY;
 			key.offset = (u64)-1;
 
-			index = srcu_read_lock(&fs_info->subvol_srcu);
-
 			clone_root = btrfs_get_fs_root(fs_info, &key, true);
 			if (IS_ERR(clone_root)) {
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
 				ret = PTR_ERR(clone_root);
 				goto out;
 			}
@@ -7168,7 +7164,6 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 			    btrfs_root_dead(clone_root)) {
 				spin_unlock(&clone_root->root_item_lock);
 				btrfs_put_root(clone_root);
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
 				ret = -EPERM;
 				goto out;
 			}
@@ -7176,13 +7171,11 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 				dedupe_in_progress_warn(clone_root);
 				spin_unlock(&clone_root->root_item_lock);
 				btrfs_put_root(clone_root);
-				srcu_read_unlock(&fs_info->subvol_srcu, index);
 				ret = -EAGAIN;
 				goto out;
 			}
 			clone_root->send_in_progress++;
 			spin_unlock(&clone_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 
 			sctx->clone_roots[i].root = clone_root;
 			clone_sources_to_rollback = i + 1;
@@ -7196,11 +7189,8 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 		key.type = BTRFS_ROOT_ITEM_KEY;
 		key.offset = (u64)-1;
 
-		index = srcu_read_lock(&fs_info->subvol_srcu);
-
 		sctx->parent_root = btrfs_get_fs_root(fs_info, &key, true);
 		if (IS_ERR(sctx->parent_root)) {
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = PTR_ERR(sctx->parent_root);
 			goto out;
 		}
@@ -7210,20 +7200,16 @@ long btrfs_ioctl_send(struct file *mnt_file, struct btrfs_ioctl_send_args *arg)
 		if (!btrfs_root_readonly(sctx->parent_root) ||
 				btrfs_root_dead(sctx->parent_root)) {
 			spin_unlock(&sctx->parent_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = -EPERM;
 			goto out;
 		}
 		if (sctx->parent_root->dedupe_in_progress) {
 			dedupe_in_progress_warn(sctx->parent_root);
 			spin_unlock(&sctx->parent_root->root_item_lock);
-			srcu_read_unlock(&fs_info->subvol_srcu, index);
 			ret = -EAGAIN;
 			goto out;
 		}
 		spin_unlock(&sctx->parent_root->root_item_lock);
-
-		srcu_read_unlock(&fs_info->subvol_srcu, index);
 	}
 
 	/*

fs/btrfs/tests/btrfs-tests.c

@@ -134,14 +134,6 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize)
 
 	fs_info->nodesize = nodesize;
 	fs_info->sectorsize = sectorsize;
-
-	if (init_srcu_struct(&fs_info->subvol_srcu)) {
-		kfree(fs_info->fs_devices);
-		kfree(fs_info->super_copy);
-		kfree(fs_info);
-		return NULL;
-	}
-
 	set_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
 
 	test_mnt->mnt_sb->s_fs_info = fs_info;
@@ -191,7 +183,6 @@ void btrfs_free_dummy_fs_info(struct btrfs_fs_info *fs_info)
 	}
 	btrfs_free_qgroup_config(fs_info);
 	btrfs_free_fs_roots(fs_info);
-	cleanup_srcu_struct(&fs_info->subvol_srcu);
 	kfree(fs_info->super_copy);
 	btrfs_check_leaked_roots(fs_info);
 	btrfs_extent_buffer_leak_debug_check(fs_info);