If switching RW->RO, then stop deferring timestamp writes...and possibly

other pending inode attribute changes.  We appear to be missing UFS_UPDATE()
calls in some paths with the result that bsd.rd remounting the newly
created /mnt to RO would lose the GID changes on device inodes there.
This only affected devices, as they're the only inodes where timestamp
writes are delayed.

ok deraadt@

diff --git a/sys/ufs/ffs/ffs_vfsops.c b/sys/ufs/ffs/ffs_vfsops.c
index 15c0722..a04cda2 100644 (file)
--- a/sys/ufs/ffs/ffs_vfsops.c
+++ b/sys/ufs/ffs/ffs_vfsops.c
@@ -1,4 +1,4 @@
-/*     $OpenBSD: ffs_vfsops.c,v 1.168 2017/12/11 17:13:34 deraadt Exp $        */
+/*     $OpenBSD: ffs_vfsops.c,v 1.169 2017/12/14 17:25:39 guenther Exp $       */
 /*     $NetBSD: ffs_vfsops.c,v 1.19 1996/02/09 22:22:26 christos Exp $ */
 
 /*
@@ -1151,11 +1151,23 @@ ffs_sync_vnode(struct vnode *vp, void *arg) {
        struct inode *ip;
        int error;
 
+       if (vp->v_type == VNON)
+               return (0);
+
        ip = VTOI(vp);
-       if (vp->v_type == VNON || 
-           ((ip->i_flag &
+
+       /*
+        * If unmounting or converting rw to ro, then stop deferring
+        * timestamp writes.
+        */
+       if (fsa->waitfor == MNT_WAIT && (ip->i_flag & IN_LAZYMOD)) {
+               ip->i_flag |= IN_MODIFIED;
+               UFS_UPDATE(ip, 1);
+       }
+
+       if ((ip->i_flag &
                (IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) == 0 &&
-               LIST_EMPTY(&vp->v_dirtyblkhd)) ) {
+               LIST_EMPTY(&vp->v_dirtyblkhd)) {
                return (0);
        }