Another Rather Strange Boot Problem

[Java Jive]

I still have an old PC dual-booting (X)Ubunutu 22 & flavours of Windows.
Currently, it triple boots via GRUB between Ubuntu 22, Windows XP
32-bit, & Windows 10, where the OS partitions are all primaries, while
the two data partitions are logical drives in an extended partition:

Win 10 Pro | Win XP-32 | Ubuntu 22 Root || Win Data | Ubuntu 22 Home ||

However, now I wish to add another bootable partition, so, instead of
booting from a new volume in the extended partition by shuffling the
existing partitions around to make room for it, I've decided to reformat
the entire disk as GPT/BIOS.

Accordingly, as a proof of concept and to act as a working Linux system
to boot from while accomplishing this, and to preserve the Windows Data
partition which holds all the necessary Windows images for repopulating
the original disk, I've formatted a spare disk in an external USB dock
as follows ...

Boot | Win7 | Win 10 | Win 32-bit | Ub 22 Root | Win Data | Ub 22 Home |

... have copied the two Ubuntu partitions and the Windows Data
partition, have successfully installed GRUB, which found not just the
Ubuntu on the external disk but also the still existing options on the
internal hard disk and can boot into either Ubuntu. However, while
Ubuntu on the internal HD still boots normally even though GRUB is now
on the external disk, with the Ubuntu on the latter there is a problem
in that the GUI won't start, and messages concerning trying to start it
disrupt anything that you try to type into the <Alt-F1> text console by
constantly switching you back to the primary console, and when you
switch back to the one you're trying to type into, you have just a
couple of seconds or so to type something in before being switched away
again.

The messages are, where nn is an number incrementing by 1 on each
occurrence (beware possible unintended line wrap) ...

Starting Light Display Manager...
Started Light Display Manager.
Started session Cnn of User lightdm [ OK ] Stopped Light Display Manager.
Starting Detect the available GPUs and deal with any system changes...
[ 1296.065219*] nouveau 0000:01:00.0 fb: trapped read at 01004b0034 on
channel -1 [0fee0000 unknown] engine 06 [BAR] client 08 [PFIFO_READ]
subclient 01 [IN] reascon 00000002 [PAGE_NOT_PRESENT]
Finished Detect the available GPUs and deal with any system changes.
[Repeat ad nauseam]

* this number also increments, I presume it's a time since boot up or
something like that.

... which I have reduced to just the nouveau messages by typing ...

systemctl stop gpu-manager.service

... interspersed with ...

Starting CUPS scheduler.
Stopped CUPS scheduler.
Stopping CUPS scheduler...
Starting CUPS scheduler.
[FAILED] Failed to start CUPS scheduler.
See 'systemctl status cups.service' for details

... though I think I've managed to stop these latter by typing ...

systemctl stop cups.service

I'm guessing that Ubuntu is trying to find a display chip in the
hardware of the dock, which of course is bound to fail, and to fix it I
will have temporarily to swap the disks around, which is what I will try
next, but can anyone explain exactly what is going wrong, and suggest
how to make it work? For example, how does a suck-it-and-see Linux
distro USB stick avoid this situation?

[Java Jive]

Post by Java Jive
I still have an old PC dual-booting (X)Ubunutu 22 & flavours of Windows.
 Currently, it triple boots via GRUB between Ubuntu 22, Windows XP
32-bit, & Windows 10, where the OS partitions are all primaries, while
Win 10 Pro | Win XP-32 | Ubuntu 22 Root || Win Data | Ubuntu 22 Home ||
However, now I wish to add another bootable partition, so, instead of
booting from a new volume in the extended partition by shuffling the
existing partitions around to make room for it, I've decided to reformat
the entire disk as GPT/BIOS.
Accordingly, as a proof of concept and to act as a working Linux system
to boot from while accomplishing this, and to preserve the Windows Data
partition which holds all the necessary Windows images for repopulating
the original disk, I've formatted a spare disk in an external USB dock
as follows ...
Boot | Win7 | Win 10 | Win 32-bit | Ub 22 Root | Win Data | Ub 22 Home |
... have copied the two Ubuntu partitions and the Windows Data
partition, have successfully installed GRUB, which found not just the
Ubuntu on the external disk but also the still existing options on the
internal hard disk and can boot into either Ubuntu.  However, while
Ubuntu on the internal HD still boots normally even though GRUB is now
on the external disk, with the Ubuntu on the latter there is a problem
in that the GUI won't start, and messages concerning trying to start it
disrupt anything that you try to type into the <Alt-F1> text console by
constantly switching you back to the primary console, and when you
switch back to the one you're trying to type into, you have just a
couple of seconds or so to type something in before being switched away
again.
The messages are, where nn is an number incrementing by 1 on each
occurrence (beware possible unintended line wrap) ...
Starting Light Display Manager...
Started Light Display Manager.
Started session Cnn of User lightdm [ OK ] Stopped Light Display Manager.
Starting Detect the available GPUs and deal with any system changes...
[ 1296.065219*] nouveau 0000:01:00.0 fb: trapped read at 01004b0034 on
channel -1 [0fee0000 unknown]  engine 06 [BAR] client 08 [PFIFO_READ]
subclient 01 [IN] reascon 00000002 [PAGE_NOT_PRESENT]
Finished Detect the available GPUs and deal with any system changes.
[Repeat ad nauseam]
* this number also increments, I presume it's a time since boot up or
something like that.
... which I have reduced to just the nouveau messages by typing ...
    systemctl stop gpu-manager.service
... interspersed with ...
Starting CUPS scheduler.
Stopped CUPS scheduler.
Stopping CUPS scheduler...
Starting CUPS scheduler.
[FAILED] Failed to start CUPS scheduler.
See 'systemctl status cups.service' for details
... though I think I've managed to stop these latter by typing ...
    systemctl stop cups.service
I'm guessing that Ubuntu is trying to find a display chip in the
hardware of the dock, which of course is bound to fail, and to fix it I
will have temporarily to swap the disks around, which is what I will try
next, but can anyone explain exactly what is going wrong, and suggest
how to make it work?  For example, how does a suck-it-and-see Linux
distro USB stick avoid this situation?

No, swapping the disks hasn't helped. I now have only the nouveau
messages, but still no GUI.

[Theo]

Post by Java Jive
Starting Light Display Manager...
Started Light Display Manager.
Started session Cnn of User lightdm [ OK ] Stopped Light Display Manager.
Starting Detect the available GPUs and deal with any system changes...
[ 1296.065219*] nouveau 0000:01:00.0 fb: trapped read at 01004b0034 on
channel -1 [0fee0000 unknown] engine 06 [BAR] client 08 [PFIFO_READ]
subclient 01 [IN] reascon 00000002 [PAGE_NOT_PRESENT]
Finished Detect the available GPUs and deal with any system changes.
[Repeat ad nauseam]
* this number also increments, I presume it's a time since boot up or
something like that.
... which I have reduced to just the nouveau messages by typing ...
systemctl stop gpu-manager.service
... interspersed with ...
Starting CUPS scheduler.
Stopped CUPS scheduler.
Stopping CUPS scheduler...
Starting CUPS scheduler.
[FAILED] Failed to start CUPS scheduler.
See 'systemctl status cups.service' for details
... though I think I've managed to stop these latter by typing ...
systemctl stop cups.service
I'm guessing that Ubuntu is trying to find a display chip in the
hardware of the dock, which of course is bound to fail, and to fix it I
will have temporarily to swap the disks around, which is what I will try
next, but can anyone explain exactly what is going wrong, and suggest
how to make it work? For example, how does a suck-it-and-see Linux
distro USB stick avoid this situation?

It looks like it's trying to set up the open-source reverse-engineered
nouveau drivers for an NVIDIA GPU. These drivers aren't very good, and if
you have such a GPU I'd recommend installing nvidia's closed source drivers
which are much more reliable.

Some laptops have both an integrated Intel and discrete nvidia GPU in them,
and you can select either one, so even with no nvidia drivers the Intel side
still works. I don't remember which laptop you have but I think it's too
old to do that? In which case you need working nvidia drivers to get a
desktop display.

For USB sticks they either use a baseline VGA-style driver (which works with
nvidia hardware, only with no acceleration and limited screen modes), and/or
you can append 'nomodeset' to the kernel boot commandline which gets you a
basic display with which you can install the proper drivers.

For your setup, I'd get into the GRUB boot menu. There can be a '(safe
graphics)' boot option, perhaps in 'Advanced options'. If you have that,
select it to boot.

If not, find an existing Ubuntu boot entry, edit it ('e' key) and look for
the line starting 'linux'. Edit to add 'nomodeset' to the end of the line.
(you can delete 'quiet splash $vt_handoff' to turn off pretty boot
animations and see boot messages) then boot it (Ctrl-X or F10).

Instructions for something completely different but it shows the GRUB boot
menus at step 6
https://www.digitalocean.com/community/tutorials/how-to-configure-multi-factor-authentication-on-ubuntu-18-04

Once you've booted into a GUI you can install the nvidia drivers:
https://itsfoss.com/install-additional-drivers-ubuntu/

and then you should have graphics working more reliably.

If you can't get into the GUI, you can also do the drivers install from the
command line (which might be easier than dealing with GRUB if unfamiliar):

$ sudo ubuntu-drivers install
should do it, see
https://ubuntu.com/server/docs/nvidia-drivers-installation

Theo

[Java Jive]

Post by Theo

Post by Java Jive
Starting Light Display Manager...
Started Light Display Manager.
Started session Cnn of User lightdm [ OK ] Stopped Light Display Manager.
Starting Detect the available GPUs and deal with any system changes...
[ 1296.065219*] nouveau 0000:01:00.0 fb: trapped read at 01004b0034 on
channel -1 [0fee0000 unknown] engine 06 [BAR] client 08 [PFIFO_READ]
subclient 01 [IN] reascon 00000002 [PAGE_NOT_PRESENT]
Finished Detect the available GPUs and deal with any system changes.
[Repeat ad nauseam]
* this number also increments, I presume it's a time since boot up or
something like that.
... which I have reduced to just the nouveau messages by typing ...
systemctl stop gpu-manager.service
... interspersed with ...
Starting CUPS scheduler.
Stopped CUPS scheduler.
Stopping CUPS scheduler...
Starting CUPS scheduler.
[FAILED] Failed to start CUPS scheduler.
See 'systemctl status cups.service' for details
... though I think I've managed to stop these latter by typing ...
systemctl stop cups.service
I'm guessing that Ubuntu is trying to find a display chip in the
hardware of the dock, which of course is bound to fail, and to fix it I
will have temporarily to swap the disks around, which is what I will try
next, but can anyone explain exactly what is going wrong, and suggest
how to make it work? For example, how does a suck-it-and-see Linux
distro USB stick avoid this situation?

It looks like it's trying to set up the open-source reverse-engineered
nouveau drivers for an NVIDIA GPU. These drivers aren't very good, and if
you have such a GPU I'd recommend installing nvidia's closed source drivers
which are much more reliable.
Some laptops have both an integrated Intel and discrete nvidia GPU in them,
and you can select either one, so even with no nvidia drivers the Intel side
still works. I don't remember which laptop you have but I think it's too
old to do that? In which case you need working nvidia drivers to get a
desktop display.
For USB sticks they either use a baseline VGA-style driver (which works with
nvidia hardware, only with no acceleration and limited screen modes), and/or
you can append 'nomodeset' to the kernel boot commandline which gets you a
basic display with which you can install the proper drivers.
For your setup, I'd get into the GRUB boot menu. There can be a '(safe
graphics)' boot option, perhaps in 'Advanced options'. If you have that,
select it to boot.
If not, find an existing Ubuntu boot entry, edit it ('e' key) and look for
the line starting 'linux'. Edit to add 'nomodeset' to the end of the line.
(you can delete 'quiet splash $vt_handoff' to turn off pretty boot
animations and see boot messages) then boot it (Ctrl-X or F10).
Instructions for something completely different but it shows the GRUB boot
menus at step 6
https://www.digitalocean.com/community/tutorials/how-to-configure-multi-factor-authentication-on-ubuntu-18-04
https://itsfoss.com/install-additional-drivers-ubuntu/
and then you should have graphics working more reliably.
If you can't get into the GUI, you can also do the drivers install from the
$ sudo ubuntu-drivers install
should do it, see
https://ubuntu.com/server/docs/nvidia-drivers-installation

Thanks for your help, but I think I've solved it anyway. It seems from
the copy from the old root to the new root didn't quite go according to
plan, so I erased the new root and tried again, and this time it worked,
and I was able to boot into Ubuntu, albeit rather slowly by comparison,
from the external hard disk. I've now repartitioned the internal hard
disk and am in the process of putting everything back onto it.