Bug 2283058 - Upgrade to pam-1.6.1-1.fc40 rendered systems unusable: only single-user was functional.
Summary: Upgrade to pam-1.6.1-1.fc40 rendered systems unusable: only single-user was ...
Keywords:
Status: CLOSED DUPLICATE of bug 2280896
Alias: None
Product: Fedora
Classification: Fedora
Component: pam
Version: 40
Hardware: x86_64
OS: Linux
unspecified
urgent
Target Milestone: ---
Assignee: Iker Pedrosa
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2024-05-23 20:08 UTC by Jay Guerette
Modified: 2024-05-30 08:25 UTC (History)
3 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2024-05-30 08:25:07 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Core files from +8y old system (499.71 KB, application/octet-stream)
2024-05-23 20:34 UTC, Jay Guerette
no flags Details
Core files from +8y old system (317.11 KB, application/octet-stream)
2024-05-23 20:35 UTC, Jay Guerette
no flags Details
Core files from +8y old system (499.71 KB, application/octet-stream)
2024-05-23 20:38 UTC, Jay Guerette
no flags Details
Core files from -1y old system (336.53 KB, application/octet-stream)
2024-05-23 20:39 UTC, Jay Guerette
no flags Details
Core files from -1y old system (330.96 KB, application/octet-stream)
2024-05-23 20:40 UTC, Jay Guerette
no flags Details

Description Jay Guerette 2024-05-23 20:08:59 UTC
Upgraded 2 very different Fedora 40 systems using dnf to pam-1.6.1-1.fc40. In both cases, after reboot the graphical target wouldn't load and failing processes created 1000's of core files. Switching to a virtual console, I was unable to log in. I rebooted to single mode, enabled the multi-user target, and rebooted again. I was still unable to log in. I also verified that SSH would fail & core when attempting to log in.

After inspecting the logs, it was obvious that pam was the problem. I ran dnf downgrade pam, rebooted, and everything worked as expected on both systems. One is +8y old and the other is -1y old.


Reproducible: Always

Steps to Reproduce:
1.Upgrade to pam-1.6.1-1.fc40
2.Reboot

Actual Results:  
System completely unusable and core files being generated. Graphical target fails. No logins via any method succeed.

Expected Results:  
System operated normally as previously: graphical target loaded without errors, and logins via DM, virtual console, or SSH worked.

Comment 1 Jay Guerette 2024-05-23 20:10:11 UTC
Possibly related to bug 2280896, but that only mentions the graphical target.

Comment 2 Jay Guerette 2024-05-23 20:34:57 UTC
Created attachment 2034867 [details]
Core files from +8y old system

I initially thought the DM was the source of the problem, so I tried different DMs, and they all failed the same way. I'll attach 3 cores files from this system.

Comment 3 Jay Guerette 2024-05-23 20:35:20 UTC
Created attachment 2034868 [details]
Core files from +8y old system

Comment 4 Jay Guerette 2024-05-23 20:38:37 UTC
Created attachment 2034869 [details]
Core files from +8y old system

Comment 5 Jay Guerette 2024-05-23 20:39:51 UTC
Created attachment 2034870 [details]
Core files from -1y old system

Comment 6 Jay Guerette 2024-05-23 20:40:26 UTC
Created attachment 2034871 [details]
Core files from -1y old system

Comment 7 Iker Pedrosa 2024-05-27 08:30:25 UTC
I think you are hitting https://bugzilla.redhat.com/show_bug.cgi?id=2280896. Can you check whether /etc/environment contains any NULL value?

If it's the same issue I think I have a fix for it. Would you be willing to test it?

Comment 8 Jay Guerette 2024-05-27 13:38:10 UTC
In /etc/environment there is one line that is: DESKTOP_AUTOSTART_ID=
I assume that be a NULL value?
I don't remember why that's there, it could be to solve a problem in the past 20y that is no longer relevant.

I see the COPR link in the other bug, I'll test it.

Comment 9 Iker Pedrosa 2024-05-27 13:43:29 UTC
Yes, that's a NULL value

Comment 10 Jay Guerette 2024-05-28 12:41:24 UTC
I also looked at the conversation in https://bugzilla.redhat.com/show_bug.cgi?id=2280896

I verified:
- the issue is still not present in 1.6.0-2
- the issue still exists in 1.6.1-1
- the issue goes away when I remove the NULL assignment in /etc/environment
- the issue goes away when I change the NULL assignment to ""
- the issue returns when I restore the NULL assignment
- the issue is NOT present in 1.6.1-2test
But as noted in the other bug:
- in 1.6.1-2test a NULL assignment results in the variable NOT getting set, which is a change in behavior

Comment 11 Iker Pedrosa 2024-05-30 08:25:07 UTC

*** This bug has been marked as a duplicate of bug 2280896 ***


Note You need to log in before you can comment on or make changes to this bug.