We should not just invalidate the ACL when setting the underlying attribute, but also when removing it. The ioctl interface gets that right, but the normal xattr inteface skipped the xfs_forget_acl due to an early return. Signed-off-by: Christoph Hellwig <hch@xxxxxx> --- fs/xfs/xfs_xattr.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/fs/xfs/xfs_xattr.c b/fs/xfs/xfs_xattr.c index 383f0203d103..2288f20ae282 100644 --- a/fs/xfs/xfs_xattr.c +++ b/fs/xfs/xfs_xattr.c @@ -74,10 +74,11 @@ xfs_xattr_set(const struct xattr_handler *handler, struct dentry *unused, if (flags & XATTR_REPLACE) xflags |= ATTR_REPLACE; - if (!value) - return xfs_attr_remove(ip, (unsigned char *)name, xflags); - error = xfs_attr_set(ip, (unsigned char *)name, + if (value) + error = xfs_attr_set(ip, (unsigned char *)name, (void *)value, size, xflags); + else + error = xfs_attr_remove(ip, (unsigned char *)name, xflags); if (!error) xfs_forget_acl(inode, name, xflags); -- 2.20.1
