Christian Brauner [off-list ref] wrote:
quoted hunk ↗ jump to hunk
diff --git a/fs/cachefiles/xattr.c b/fs/cachefiles/xattr.c
index 72e42438f3d7..a591b5e09637 100644
--- a/fs/cachefiles/xattr.c
+++ b/fs/cachefiles/xattr.c
@@ -39,8 +39,8 @@ int cachefiles_check_object_type(struct cachefiles_object *object)
_enter("%p{%s}", object, type);
/* attempt to install a type label directly */
- ret = vfs_setxattr(dentry, cachefiles_xattr_cache, type, 2,
- XATTR_CREATE);
+ ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache, type,
+ 2, XATTR_CREATE);
Actually, on further consideration, this might be the wrong thing to do in
cachefiles. The creds are (or should be) overridden when accesses to the
underlying filesystem are being made.
I wonder if this should be using current_cred()->user_ns or
cache->cache_cred->user_ns instead.
David