If switching RW->RO, then stop deferring timestamp writes...and possibly
authorguenther <guenther@openbsd.org>
Thu, 14 Dec 2017 17:25:39 +0000 (17:25 +0000)
committerguenther <guenther@openbsd.org>
Thu, 14 Dec 2017 17:25:39 +0000 (17:25 +0000)
commit44961df1dceab8b67f900b362f44ce7d89e9fa6a
tree1547df23c18186a33d0c2e7ed778f959a4cbc526
parentcfd60b50b925dbd1f966fb2bd4ff813c47c0f606
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@
sys/ufs/ffs/ffs_vfsops.c