Arthur Radley wrote: > On Thu, May 9, 2013 at 5:18 PM, Markku Pesonen <[email protected]> wrote: > >> >> I think this is caused by some change in util-linux-2.22 umount code (if >> I remember correctly). In /etc/rc.d/init.d/mountfs there is the >> following line: >> umount -a -d -r -t notmpfs,nosysfs,nodevtmpfs,noproc >/dev/null >> Despite using the -r option, umount does not remount the root filesystem >> in read-only mode during shutdown like it used to. Adding this command >> after that line fixed things for me: >> mount -n -o remount,ro / >> >> > Okay. I added that line to /etc/rc.d/init/mountfs. Recapping, the boot > messages before I did that... > > [ 3.412320] EXT4-fs (sda1): INFO: recovery required on readonly > filesystem > [ 3.412385] EXT4-fs (sda1): write access will be enabled during recovery > [ 3.688323] EXT4-fs (sda1): recovery complete > [ 3.704672] EXT4-fs (sda1): mounted filesystem with ordered data mode. > Opts: (null) > [ 3.704774] VFS: Mounted root (ext4 filesystem) readonly on device 8:1. > > And afterwards, no spew about recovery being required, completed, or write > access during recovery. Only... > > [ 3.429281] EXT4-fs (sda1): mounted filesystem with ordered data mode. > Opts: (null) > [ 3.429378] VFS: Mounted root (ext4 filesystem) readonly on device 8:1. > > I think I like that better. > > But a small amount of additional explanation of what was and is now > happening would be appreciated. I accept out-of-hand the explanation that > before, my root filesystem was not being remounted read-only during > shutdown or reboot. It's just that I still wonder what state or condition > required recovery before, what was being done about it, and why it is no > longer required after adding the command to remount the root filesystem ro > at shutdown and reboot.
OK, I'll look at adding that line to the stop portion of mountfs. What happens after the umount is running the localnet and halt scripts which amounts to ip link set lo down halt -d -f -i -p but localnet are does: log_info_msg "Bringing down the loopback interface..." which writes to /run/var/bootlog (a tmpfs that is not umounted). One thing I can think of that would 'corrupt' the root fs is the system trying to update an access time for the files used. The only other thing is that the umount is just not syncing status properly and just not working. The explicit remount read-only should ensure the disk is clean at shutdown. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
