GDM Freeze [SOLVED]

riggt · 2015-10-21 01:23:08

Hi. I've already solved this problem for my own purposes, but I want to make as much information available about it in case someone else is having this problem.

The Problem

The trouble I was having was that when I booted my computer it would go through initial boot sequence and then just stop, leaving me with an unusable computer. At first I thought that the issue was in the boot sequence so I started out by debugging it following this process: Boot Debugging all the way to the Heavy Debug section. I also followed this guide to edit the kernel parameters to run in single user mode. Still assuming that the problem was with the boot sequence I went through the journalctl output, trying to fix every issue I could in the log. Here is the output of my journalctl run that I took action on:

Oct 20 19:34:25 brick kernel: xhci_hcd 0000:00:14.0: can't derive routing for PCI INT A

Oct 20 19:34:25 brick kernel: xhci_hcd 0000:00:14.0: PCI INT A: no GSI

Oct 20 19:34:26 brick systemd[1]: laptop-mode.service: Cannot add dependency job, ignoring: Unit laptop-mode.service failed to load: No such file or directory.

Oct 20 19:34:26 brick systemd[1]: systemd-modules-load.service: Main process exited, code=exited, status=1/FAILURE
Oct 20 19:34:26 brick systemd[1]: Failed to start Load Kernel Modules.
Oct 20 19:34:26 brick systemd[1]: systemd-modules-load.service: Unit entered failed state.
Oct 20 19:34:26 brick systemd[1]: systemd-modules-load.service: Failed with result 'exit-code'.

Oct 20 19:34:26 brick kernel: ACPI Exception: AE_BAD_PARAMETER, Returned by Handler for [EmbeddedControl] (20150410/evregion-297)
Oct 20 19:34:26 brick kernel: ACPI Error: Method parse/execution failed [\_TZ_.TZ01._TMP] (Node ffff88011a8b0eb0), AE_BAD_PARAMETER (20150410/psparse-536)
Oct 20 19:34:26 brick kernel: ACPI Exception: AE_BAD_PARAMETER, Returned by Handler for [EmbeddedControl] (20150410/evregion-297)
Oct 20 19:34:26 brick kernel: ACPI Error: Method parse/execution failed [\_TZ_.TZ01._TMP] (Node ffff88011a8b0eb0), AE_BAD_PARAMETER (20150410/psparse-536)

Oct 20 19:34:26 brick kernel: kvm: disabled by bios

Oct 20 19:34:26 brick kernel: kvm: disabled by bios
Oct 20 19:34:26 brick kernel: kvm: disabled by bios
Oct 20 19:34:26 brick kernel: kvm: disabled by bios

Oct 20 19:34:26 brick systemd-modules-load[168]: Failed to find module 'acpi_call'

Oct 20 19:34:24 brick systemd-udevd[218]: Process '/usr/bin/set-wireless-regdom' failed with exit code 1.

Oct 20 19:34:44 brick nm-dispatcher[676]: Script '/etc/NetworkManager/dispatcher.d/10-chrony' exited with error status 1.

Oct 20 19:34:45 brick NetworkManager[444]: <warn>  (2) 10-chrony failed (failed): Script '/etc/NetworkManager/dispatcher.d/10-chrony' exited with error status 1.

Acting on these errors got me nowhere, though during this time I started working on the pet theory that the problem wasn't happening during the boot but was happening when GDM was being started. So, I rebooted again and used

systemctl disable gdm.service

Boot! Finally!

Finally I was able to log in without having to use the single user parameter. The first thing I tried was downgrading gdm, gnome shell, and gnome session. However, I found that I've only ever used one version of gdm, even back when things were working just fine. So gnome shell and gnome session were the only things I ended up messing with. The versions to and from or each of those are as follows:

gnome-shell          3.18.1-2     -> 3.18.1-1
gnome-session     3.18.1.2-1  -> 3.18.1.1-1

However, when I ran

systemctl default

things worked about the same as they did before. I also tried downgrading all of my networking packages (they were being a little wonky), but I put them right back immediately after that did nothing but make it even harder to connect to the internet (they went from a little wonky to completely unusable).

The Solution that I found

Eventually I found a forum post referring to what sounded like what I was happening. It was this post from two years ago. The users in this thread installed SLiM as an alternative to GDM. However when I looked at the SLiM wiki page it sounded like it would not work for my needs ever since it was deprecated. I followed the link in the warning to the list of display managers, and selected LightDM which met my needs the best. After setting up LightDM, I was able to log into the graphical version of gnome, and was also able to reboot without issue.

Going Forward

There are two things that I want to keep track of going forward. I want to know if anyone else is having this problem so that we can start to pinpoint what specifically the problem is so that we can fix it. I also want to see if future updates to the gnome group fix this problem. I also what to see if there are anymore details that I personally can provide both so that I can learn and improve my contributions to the community, but also so that if there actually does turn out to be a bug somewhere it can be quashed.

Further Information (edits)

Wed Oct 21, 1:48 AM: I found a bug in the bug tracker that sounded similar to what I am experiencing here. The difference however is that my computer is an intel integrated graphics system. The bug linked to in the comments for the bug I have linked do not describe the same problem that I experienced. I am about to reboot to see if editing the GDM custom.conf causes it to run properly.

Wed Oct 21, 2:01 AM: Editing the GDM custom.conf allowed GDM to function properly without hanging the system.

Wed Oct 21, 2:15 AM: Because the only other place I can find a similar discussion of what is going on with me was related to nvidia drivers I went ahead and double checked that nouveu wasn't installed accidentally at some point. It was not.

Wed Oct 21, 1:53 PM: I realized that I could be a little clearer about what I did to the custom.conf file and also how to find it in case someone comes across this forum post looking for a solution to a problem they are having. The file that I edited was

/etc/gdm/custom.conf

and I edited it so that it appears as follows:

# GDM configuration storage

[daemon]
# Uncoment the line below to force the login screen to use Xorg
WaylandEnable=false

[security]

[xdmcp]

[greeter]

[chooser]

[debug]
# Uncomment the line below to turn on debugging
#Enable=true

All I did, truly, was uncomment the

WaylandEnable=false

line.

Last edited by riggt (2015-10-22 20:34:16)

jhenriquez · 2015-10-22 15:50:36

I had not upgraded the system in a long time (gnome-3.16 was my last version) until today. Everything seemed fine until I rebooted and encountered with GDM freezing. Now, maybe what happened to me is not exactly the same issue, because my computer reached GDM and then freezed.

I went into Single User Mode and tried to follow along your steps without much luck. There was not logs in /var/log/gdm (I'm still new to this linux stuff I don't know much on how to look for information on the system when this kind of stuff occurs).

Downgrading gnome-shell, gnome-session and GDM didn't solve my problem. In fact, when running systemctl default it didn't even reached the graphical interface. But it didn't freeze the whole system. Disabling WaylandEnable actually did freeze the system again.

So I tried to reinstall gnome bundle (3.16) maybe I was missing packages from the downgrade, I noticed I didn't have a network interface. So, I figured something worse than just gnome was going on. I noticed in the dmesg logs that the computer was not able to setup bluetooth either. So, (Cowboy style) I downgraded the kernel as well from 4.2.3-1 to 4.1.6-1. This fixed my network stuff, but at this point I had given up on gdm (didn't even tried it).

I started trying with lightdm and it didn't work either. It freezed too. Very disappointed, I tried reinstalling the gnome bundle again. It brought it back to 3.18. I did systemctl start gdm.service and everything was normal again. Honestly, I'm not sure what fixed my stuff, since I was kind of guessing around. But I thought leaving a comment here could possibly help someone.

Last edited by jhenriquez (2015-10-22 15:55:28)

loqs · 2015-10-22 17:53:08

Forgive the off topic as this is not responsive to the original poster

jhenriquez wrote:

So I tried to reinstall gnome bundle (3.16) maybe I was missing packages from the downgrade, I noticed I didn't have a network interface. So, I figured something worse than just gnome was going on. I noticed in the dmesg logs that the computer was not able to setup bluetooth either. So, (Cowboy style) I downgraded the kernel as well from 4.2.3-1 to 4.1.6-1. This fixed my network stuff, but at this point I had given up on gdm (didn't even tried it).

Those symptoms were /are most likely caused by a mismatch between the booted kernels version and modules provided by the kernel package (commonly caused by upgrading the kernel package without /boot being mounted).
The downgrade will have brought versions back into sync.

Arch Linux

#1 2015-10-21 01:23:08