auto mount.ntfs read-only when boot grml.iso which store on a ntfs partition

[dongshan3]

grml is useful for testing software installation. My PC just have one MBR partition and installed Windows 10 22H2. I am going to use grub-2.12-for-windows to load my grml.iso which stored as /dev/sda1/iso/grml.iso. My grub.cfg menuentry as:

menuentry "grml" {
          set root=(hd0,1)
          loopback loop /iso/grml.iso
          linux (loop)/boot/grml/vmlinuz boot=live live-media-path=/live/grml/ bootid=grml001 findiso=/iso/grml.iso toram=grml.squashfs
          initrd (loop)/boot/grml/initrd.img
}

grml booted success. But when I tried to run reboot or poweroff command, sometimes it not working, Crtl + Alt + Del not working too, have to press reset or power physics buttom . I saw some information on monitor:

Finished systemd-poweroff.service - System Power off
Reached target poweroff.target - system Power Off
I/O error, dev loop0, sector 0 op 0x1 write flags 0x800 phys_seg 0 prio class 2
......
I/O error, dev loop0, sector 0 op 0x1 write flags 0x800 phys_seg 0 prio class 2
blk_print_req_error: 1331 callbacks suppressed

Then I tried searched on Internel, as https://freedesktop.org/wiki/Software/systemd/Debugging/#index2h1 said, I can reboot or poweroff with --force option. I tried to enable systemd debug. There are some information on monitor after reboot or poweroff, such as:

Detaching loop devices.
Detaching loopback /dev/loop1
Could not detach loopback /dev/loop1: Device or resource busy
Detaching loopback /dev/loop0.
Not all loop devices detached, 1 left.
......
Detaching loop devices.
Detaching loopback /dev/loop1
Could not detach loopback /dev/loop1: Device or resource busy
Detaching loopback /dev/loop0.
Not all loop devices detached, 1 left.

I tried to do the same testing, but stored grml.iso on a USB stick(same PC) or a ext4 partition(another PC), reboot or poweroff success. So I listed grml mount device to compare:

a. grml.iso on a ntfs partition:

/dev/sda1 on /run/live/fromiso type fuseblk (rw,relatime,user_id=0,group_id=0,allow_other,blksize=4096)
/dev/loop0 on /run/live/medium type iso9660 (ro,noatime,nojoliet,check=s,map=n,blocksize=2048,iocharset=utf8)
/dev/loop1 on /run/live/rootfs/grml.squashfs type squashfs (ro,noatime,errors=continue)

b. grml.iso on a vfat pariton:

/dev/sdb1 on /run/live/medium type vfat (ro,noatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,utf8,errors=remount-ro)
/dev/loop0 on /run/live/rootfs/grml.squashfs type squashfs (ro,noatime,errors=continue)

c. grml.iso on a ext4 pariton:

/dev/sda2 on /run/live/findiso type ext4 (ro,noatime)
/dev/loop0 on /run/live/medium type iso9660 (ro,noatime,nojoliet,check=s,map=n,blocksize=2048,iocharset=utf8)
/dev/loop1 on /run/live/rootfs/grml.squashfs type squashfs (ro,noatime,errors=continue)

So, reboot/poweroff issue maybe happen when mount ntfs partition as read-write. I tried to find more information, about ntfs-3g, like this https://github.com/tuxera/ntfs-3g/wiki/NTFS-3G-FAQ . powercfg /h off looks like not working here. I do more testing.

a. when from Windows do restart action , ntfs-3g will consider /dev/sda1 is a clean file system, so grml auto mount it as read-write, then issues happen.

b. when from Windows do shutdown action, ntfs-3g will consider /dev/sda1 is a unclean file system, so grml mount it as read-only, then no problem

If I added toram=grml.squashfs or toram as kernel parameters as https://wiki.grml.org/doku.php?id=rescueboot said, grml will not auto mount /dev/sda1, so no problem

Describe the solution you'd like Auto mount.ntfs partition as read-only, no matter is clean filesystem or not.

Describe alternatives you've considered

• reboot or poweroff with --force option
• added toram=grml.squashfs kernel parameters to boot

Disk usage of the software I created debian 12 stable iso using grml64-full, grml-live -s stable -c GRMLBASE,GRML_SMALL,AMD64 -o /srv/grml-live

[mika]

Nice find, thanks!

Would it be possible that you boot the system when it's failing with added debug=1 boot option and provide us the resulting /var/log/live/boot.log? This would certainly be useful for looking further into this issue.

[dongshan3]

Nice find, thanks!

Would it be possible that you boot the system when it's failing with added debug=1 boot option and provide us the resulting /var/log/live/boot.log? This would certainly be useful for looking further into this issue.

I got journal.txt as freedesktop.org said journal.txt before, maybe this is what you want.

The journal (the output of journalctl -b > journal.txt) ideally after booting with systemd.log_level=debug systemd.log_target=kmsg log_buf_len=1M printk.devkmsg=on

There is such:

Jan 29 08:08:18 grml systemd-fstab-generator[1606]: Found entry what=/dev/disk/by-label/新加卷 where=/media/新加卷 type=ntfs makefs=no growfs=no noauto=yes nofail=no

it should be /dev/sda1, ntfs partition.

[mika]

Nice find, thanks! Would it be possible that you boot the system when it's failing with added debug=1 boot option and provide us the resulting /var/log/live/boot.log? This would certainly be useful for looking further into this issue.

I got journal.txt as freedesktop.org said journal.txt before, maybe this is what you want.

The journal (the output of journalctl -b > journal.txt) ideally after booting with systemd.log_level=debug systemd.log_target=kmsg log_buf_len=1M printk.devkmsg=on

Thanks, that's also good to have! But with debug=1 the code triggered during bootup would also be logged (and available as /var/log/live/boot.log after booting) and that would be useful to see what exactly is going on in your situation. Thx

[dongshan3]

Thanks, that's also good to have! But with debug=1 the code triggered during bootup would also be logged (and available as /var/log/live/boot.log after booting) and that would be useful to see what exactly is going on in your situation. Thx

Okay, I got /var/log/live/boot.log now.
boot.log

There is mount -t ntfs -o ro,noatime /dev/sda1 /run/live/medium in boot.log, but Shell show rw as below output:

root@grml ~ # df
Filesystem     1K-blocks     Used Available Use% Mounted on
udev             3979948        0   3979948   0% /dev
tmpfs             800736      952    799784   1% /run
/dev/sda1      125032444 76365112  48667332  62% /run/live/findiso
/dev/loop0        501760   501760         0 100% /run/live/medium
/dev/loop1        441344   441344         0 100% /run/live/rootfs/grml.squashfs
tmpfs            4003676     8468   3995208   1% /run/live/overlay
overlay          4003676     8468   3995208   1% /
tmpfs            4003676        0   4003676   0% /dev/shm
tmpfs               5120        0      5120   0% /run/lock
tmpfs            4003676        4   4003672   1% /tmp
tmpfs             800732        0    800732   0% /run/user/0
root@grml ~ # mount | grep /dev/sda1
/dev/sda1 on /run/live/findiso type fuseblk (rw,relatime,user_id=0,group_id=0,allow_other,blksize=4096)
/dev/sda1 on /lib/live/mount/findiso type fuseblk (rw,relatime,user_id=0,group_id=0,allow_other,blksize=4096)
root@grml ~ # ps aux | grep mount.ntfs
root         296  0.0  0.0   3544  2000 ?        Ss   20:19   0:00 mount.ntfs /dev/sda1 /run/live/findiso
root        2417  0.0  0.0   8088  2188 pts/0    S+   20:25   0:00 grep --color=auto mount.ntfs
root@grml ~ #

If I run reboot, then screen scroll with:

systemd-shutdown[1]: Not all file system unmounted. 2 left.
systemd-shutdown[1]: Detaching loop devices.
systemd-shutdown[1]: Detaching loopback /dev/loop1
systemd-shutdown[1]: Could not detach loopback /dev/loop1: Device or reource busy
systemd-shutdown[1]: Detaching loopback /dev/loop0
systemd-shutdown[1]: Not all loop devices detached. 1 left.
systemd-shutdown[1]: Unmounting file systems.
(sd-umount) [58311]: Unmounting '/run/live/medium'
(sd-umount) [58311]: Failed to umount /run/live/medium: Device or resource busy
(sd-reount) [58312]:  Remounting '/run/live/findiso' ready-only with options 'user_id=0,group_id=0,allow_other...
(sd-reount) [58312]:  Failed to remount '/run/live/findiso' read-only: Device or resource busy
(sd-umount) [58313]:  Unmounting '/run/live/findiso'
(sd-umount) [58313]: Failed to umount /run/live/findis: Device or resource busy

[mika]

Thanks, yeah the log says that all relevant mounts take place in r/o mode, this is what I expected also from the code perspective, so that's definitely interesting what's going on here. :)

You could try booting with an additional break=init boot option, which will drop you into the initramfs during bootup before handing over to main user-space. There you can verify whether the NTFS partition is actually used r/o or r/w at that stage (check with mount and creating files at the accordingly please with e.g. touch /root/run/...).

[dongshan3]

Hi, mika, thanks for your guidance.

a. Windows -> restart, mount | grep sda show as /dev/sda1 as rw in initrams, then continute to user-space, also rw. reboot or shutdown not working.

b. Windows -> shutdow, mount | grep sda show as /dev/sda1 as ro in initrams, then continute to user-space, also ro. reboot or shutdown working.

Not sure why Windows restart / shutdown will result grml livecd reboot success or not.