summaryrefslogtreecommitdiff
path: root/fs/bcachefs
diff options
context:
space:
mode:
authorKent Overstreet <kent.overstreet@gmail.com>2016-03-21 15:12:47 -0800
committerKent Overstreet <kent.overstreet@gmail.com>2016-08-28 19:15:19 -0800
commit26031fdf96022ec685b53daf4c1fc3981f8f0d25 (patch)
tree22bd7cfa595378b659805ec42ad9dc43bef26306 /fs/bcachefs
parentc46f9987fdfdcac553ca298c6feb4c108ee9ded3 (diff)
XXX acl thing
Diffstat (limited to 'fs/bcachefs')
-rw-r--r--fs/bcachefs/acl.c9
1 files changed, 9 insertions, 0 deletions
diff --git a/fs/bcachefs/acl.c b/fs/bcachefs/acl.c
index aaec01208042..29249c8aeb08 100644
--- a/fs/bcachefs/acl.c
+++ b/fs/bcachefs/acl.c
@@ -216,6 +216,15 @@ int bch_set_acl(struct inode *inode, struct posix_acl *acl, int type)
if (ret == -ERANGE)
ret = -E2BIG;
+ /*
+ * XXX
+ *
+ * viro> tytso: what's to prevent ext4_get_acl() vs. ext4_set_acl()
+ * race with the former reading the acl, then the latter setting and
+ * caching the new value, *and* former overwriting cached acl with the
+ * value _it_ has got earlier?
+ */
+
if (!ret)
set_cached_acl(inode, type, acl);