Re: HURD not unmounting disks cleanly (not the old problem)

[Mark Kettenis <kettenis@wins.uva.nl>]

   Date: Mon, 23 Aug 1999 13:18:02 +0200
   From: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

   On Mon, Aug 23, 1999 at 09:58:18AM +0200, Pontus Lidman wrote:
   > HURD doesn't seem to unmount my disks cleanly when I reboot using the
   > "reboot" command. I saw this covered in a previous message on this list to
   > which Marcus replied that the newer versions of the Debian packages will
   > indeed unmount disks cleanly unless sysvinit is installed.

   I don't remember saying something like this, but I may have thought so for a
   while.

   Fact is that nobody knows under which circumstances disks are not unmounted
   cleanly. I *know* that when you fsck under linux, you get a clean disk at
   next reboot, and if you don't screw up, it stays clean. But when you fsck
   under Hurd during boot time, it does not end up clean. I am not sure what
   happens if you e2fsck under Hurd manually (forcing).

Since I'm using split-init, my root filesystem is never unmounted
cleanly.  Other filesystems sometimes are unmounted cleanly and
sometimes are not unmounted cleanly (just like the root filesystem
before split-init).

I usually do

   # settrans -a /mnt/*

before rebooting.  This way I can avoid having to fsck the filesystems
that I MIGHT have used under the Hurd.  Of course this only works if
your mount points live in the /mnt/ directory.

AFAICT fsck works perfectly under the Hurd.  Don't execute e2fsck
manually though.  You risk checking a mounted filesystem that way,
which is not a good thing to do.  The fsck wrapper makes sure this
won't happen (although accessing the partition while fsck is running
is still a bad idea).

If somebody is annoyed enough by this problem he/she should try to
debug it.  Using gdb is probably not possible (although debugging a
sub-hurd might work), so try to add some printf's at strategous places
in the init and ext2fs code.

Mark