@@ -3458,6 +3458,7 @@ int btrfs_check_data_free_space(struct inode *inode, u64 bytes, u64 write_bytes)
void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes);
void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
+void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans);
int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
struct inode *inode);
void btrfs_orphan_release_metadata(struct inode *inode);
@@ -4116,11 +4116,19 @@ static void check_system_chunk(struct btrfs_trans_handle *trans,
struct btrfs_space_info *info;
u64 left;
u64 thresh;
+ int ret = 0;
+
+ /*
+ * Needed because we can end up allocating a system chunk and for an
+ * atomic and race free space reservation in the chunk block reserve.
+ */
+ ASSERT(mutex_is_locked(&root->fs_info->chunk_mutex));
info = __find_space_info(root->fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
spin_lock(&info->lock);
left = info->total_bytes - info->bytes_used - info->bytes_pinned -
- info->bytes_reserved - info->bytes_readonly;
+ info->bytes_reserved - info->bytes_readonly -
+ info->bytes_may_use;
spin_unlock(&info->lock);
thresh = get_system_chunk_thresh(root, type);
@@ -4134,7 +4142,21 @@ static void check_system_chunk(struct btrfs_trans_handle *trans,
u64 flags;
flags = btrfs_get_alloc_profile(root->fs_info->chunk_root, 0);
- btrfs_alloc_chunk(trans, root, flags);
+ /*
+ * Ignore failure to create system chunk. We might end up not
+ * needing it, as we might not need to COW all nodes/leafs from
+ * the paths we visit in the chunk tree (they were already COWed
+ * or created in the current transaction for example).
+ */
+ ret = btrfs_alloc_chunk(trans, root, flags);
+ }
+
+ if (!ret) {
+ ret = btrfs_block_rsv_add(root->fs_info->chunk_root,
+ &root->fs_info->chunk_block_rsv,
+ thresh, BTRFS_RESERVE_NO_FLUSH);
+ if (!ret)
+ trans->chunk_bytes_reserved += thresh;
}
}
@@ -5192,6 +5214,24 @@ void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
trans->bytes_reserved = 0;
}
+/*
+ * To be called after all the new block groups attached to the transaction
+ * handle have been created (btrfs_create_pending_block_groups()).
+ */
+void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans)
+{
+ struct btrfs_fs_info *fs_info = trans->root->fs_info;
+
+ if (!trans->chunk_bytes_reserved)
+ return;
+
+ WARN_ON_ONCE(!list_empty(&trans->new_bgs));
+
+ block_rsv_release_bytes(fs_info, &fs_info->chunk_block_rsv, NULL,
+ trans->chunk_bytes_reserved);
+ trans->chunk_bytes_reserved = 0;
+}
+
/* Can only return 0 or -ENOSPC */
int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
struct inode *inode)
@@ -509,6 +509,7 @@ again:
h->transaction = cur_trans;
h->blocks_used = 0;
h->bytes_reserved = 0;
+ h->chunk_bytes_reserved = 0;
h->root = root;
h->delayed_ref_updates = 0;
h->use_count = 1;
@@ -792,6 +793,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
if (!list_empty(&trans->new_bgs))
btrfs_create_pending_block_groups(trans, root);
+ btrfs_trans_release_chunk_metadata(trans);
+
if (lock && !atomic_read(&root->fs_info->open_ioctl_trans) &&
should_end_transaction(trans, root) &&
ACCESS_ONCE(cur_trans->state) == TRANS_STATE_RUNNING) {
@@ -2054,6 +2057,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
clear_bit(BTRFS_INODE_BTREE_LOG1_ERR, &btree_ino->runtime_flags);
clear_bit(BTRFS_INODE_BTREE_LOG2_ERR, &btree_ino->runtime_flags);
+ btrfs_trans_release_chunk_metadata(trans);
+
spin_lock(&root->fs_info->trans_lock);
cur_trans->state = TRANS_STATE_UNBLOCKED;
root->fs_info->running_transaction = NULL;
@@ -2123,6 +2128,7 @@ scrub_continue:
btrfs_scrub_continue(root);
cleanup_transaction:
btrfs_trans_release_metadata(trans, root);
+ btrfs_trans_release_chunk_metadata(trans);
trans->block_rsv = NULL;
if (trans->qgroup_reserved) {
btrfs_qgroup_free(root, trans->qgroup_reserved);
@@ -102,6 +102,7 @@ struct btrfs_transaction {
struct btrfs_trans_handle {
u64 transid;
u64 bytes_reserved;
+ u64 chunk_bytes_reserved;
u64 qgroup_reserved;
unsigned long use_count;
unsigned long blocks_reserved;
While creating a block group, we often end up getting ENOSPC while updating the chunk tree, which leads to a transaction abortion that produces a trace like the following: [30670.116368] WARNING: CPU: 4 PID: 20735 at fs/btrfs/super.c:260 __btrfs_abort_transaction+0x52/0x106 [btrfs]() [30670.117777] BTRFS: Transaction aborted (error -28) (...) [30670.163567] Call Trace: [30670.163906] [<ffffffff8142fa46>] dump_stack+0x4f/0x7b [30670.164522] [<ffffffff8108b6a2>] ? console_unlock+0x361/0x3ad [30670.165171] [<ffffffff81045ea5>] warn_slowpath_common+0xa1/0xbb [30670.166323] [<ffffffffa035daa7>] ? __btrfs_abort_transaction+0x52/0x106 [btrfs] [30670.167213] [<ffffffff81045f05>] warn_slowpath_fmt+0x46/0x48 [30670.167862] [<ffffffffa035daa7>] __btrfs_abort_transaction+0x52/0x106 [btrfs] [30670.169116] [<ffffffffa03743d7>] btrfs_create_pending_block_groups+0x101/0x130 [btrfs] [30670.170593] [<ffffffffa038426a>] __btrfs_end_transaction+0x84/0x366 [btrfs] [30670.171960] [<ffffffffa038455c>] btrfs_end_transaction+0x10/0x12 [btrfs] [30670.174649] [<ffffffffa036eb6b>] btrfs_check_data_free_space+0x11f/0x27c [btrfs] [30670.176092] [<ffffffffa039450d>] btrfs_fallocate+0x7c8/0xb96 [btrfs] [30670.177218] [<ffffffff812459f2>] ? __this_cpu_preempt_check+0x13/0x15 [30670.178622] [<ffffffff81152447>] vfs_fallocate+0x14c/0x1de [30670.179642] [<ffffffff8116b915>] ? __fget_light+0x2d/0x4f [30670.180692] [<ffffffff81152863>] SyS_fallocate+0x47/0x62 [30670.186737] [<ffffffff81435b32>] system_call_fastpath+0x12/0x17 [30670.187792] ---[ end trace 0373e6b491c4a8cc ]--- This is because we don't do proper space reservation for the chunk block reserve when we have multiple tasks allocating chunks in parallel. So block group creation has 2 phases, and the first phase essentially checks if there is enough space in the system space_info, allocating a new system chunk if there isn't, while the second phase updates the device, extent and chunk trees. However, because the updates to the chunk tree happen in the second phase, if we have N tasks, each with its own transaction handle, allocating new chunks in parallel and if there is only enough space in the system space_info to allocate M chunks, where M < N, none of the tasks ends up allocating a new system chunk in the first phase and N - M tasks will get -ENOSPC when attempting to update the chunk tree in phase 2 if they need to COW any nodes/leafs from the chunk tree. Fix this by doing proper reservation in the chunk block reserve. The issue could be reproduced by running fstests generic/038 in a loop, which eventually triggered the problem. Signed-off-by: Filipe Manana <fdmanana@suse.com> --- Applies on top of the recent patch titled: "Btrfs: fix racy system chunk allocation when setting block group ro" fs/btrfs/ctree.h | 1 + fs/btrfs/extent-tree.c | 44 ++++++++++++++++++++++++++++++++++++++++++-- fs/btrfs/transaction.c | 6 ++++++ fs/btrfs/transaction.h | 1 + 4 files changed, 50 insertions(+), 2 deletions(-)