[v5,1/3] btrfs: Add WARN_ON for qgroup reserved underflow

Commit Message

Qu Wenruo Nov. 30, 2016, 7:36 a.m. UTC

Goldwyn Rodrigues has exposed and fixed a bug which underflows btrfs
qgroup reserved space, and leads to non-writable fs.

This reminds us that we don't have enough underflow check for qgroup
reserved space.

For underflow case, we should not really underflow the numbers but warn
and keeps qgroup still work.

So add more check on qgroup reserved space and add WARN_ON() and
btrfs_warn() for any underflow case.

Signed-off-by: Qu Wenruo <[e-mail]>
Reviewed-by: David Sterba <[e-mail]>
---
v2:
  Add btrfs_warn() output for fsid, qgroupid, original reserved space and
  num_bytes to reduce, for end-user to locate the subvolume causing the
  problem. Suggested by David.
v3:
  None
v4:
  None
v5:
  Fix a copy-n-paste error, which leads to reserved space error.
  Ironically, the patch will detect the problem caused by itself.

To David:
  This is only part of the fix, test case like btrfs/104 will still
  reproduce kernel warning.
  Fix will come soon.
---
 fs/btrfs/qgroup.c | 32 +++++++++++++++++++++++++++-----
 1 file changed, 27 insertions(+), 5 deletions(-)

Patch

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 11f4fff..368cdaf 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1031,6 +1031,15 @@  static void qgroup_dirty(struct btrfs_fs_info *fs_info,
 		list_add(&qgroup->dirty, &fs_info->dirty_qgroups);
 }
 
+static void report_reserved_underflow(struct btrfs_fs_info *fs_info,
+				      struct btrfs_qgroup *qgroup,
+				      u64 num_bytes)
+{
+	btrfs_warn(fs_info,
+		"qgroup %llu reserved space underflow, have: %llu, to free: %llu",
+		qgroup->qgroupid, qgroup->reserved, num_bytes);
+	qgroup->reserved = 0;
+}
 /*
  * The easy accounting, if we are adding/removing the only ref for an extent
  * then this qgroup and all of the parent qgroups get their reference and
@@ -1058,8 +1067,13 @@  static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
 	WARN_ON(sign < 0 && qgroup->excl < num_bytes);
 	qgroup->excl += sign * num_bytes;
 	qgroup->excl_cmpr += sign * num_bytes;
-	if (sign > 0)
-		qgroup->reserved -= num_bytes;
+	if (sign > 0) {
+		if (WARN_ON(qgroup->reserved < num_bytes))
+			report_reserved_underflow(fs_info, qgroup,
+						  num_bytes);
+		else
+			qgroup->reserved -= num_bytes;
+	}
 
 	qgroup_dirty(fs_info, qgroup);
 
@@ -1079,8 +1093,13 @@  static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
 		qgroup->rfer_cmpr += sign * num_bytes;
 		WARN_ON(sign < 0 && qgroup->excl < num_bytes);
 		qgroup->excl += sign * num_bytes;
-		if (sign > 0)
-			qgroup->reserved -= num_bytes;
+		if (sign > 0) {
+			if (WARN_ON(qgroup->reserved < num_bytes))
+				report_reserved_underflow(fs_info, qgroup,
+							  num_bytes);
+			else
+				qgroup->reserved -= num_bytes;
+		}
 		qgroup->excl_cmpr += sign * num_bytes;
 		qgroup_dirty(fs_info, qgroup);
 
@@ -2204,7 +2223,10 @@  void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 
 		qg = u64_to_ptr(unode->aux);
 
-		qg->reserved -= num_bytes;
+		if (WARN_ON(qg->reserved < num_bytes))
+			report_reserved_underflow(fs_info, qg, num_bytes);
+		else
+			qg->reserved -= num_bytes;
 
 		list_for_each_entry(glist, &qg->groups, next_group) {
 			ret = ulist_add(fs_info->qgroup_ulist,