Bug 130998

Summary: suspend sometimes fails with BUG at swsusp.c:905
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Martin Vidner <mvidner>
Component: KernelAssignee: Pavel Machek <pavel>
Status: RESOLVED FIXED QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None    
Version: Final   
Target Milestone: ---   
Hardware: i586   
OS: Other   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: BUG_swsusp_c_905.jpg - kernel BUG screenshot

Description Martin Vidner 2005-10-27 13:19:01 UTC
On Toshiba Portege 3020CT, suspend to disk via powersave -U sometimes fails. Then I have to reset the machine and hope that fs journaling does its job. 

I haven't determined yet the exact circumstances. It worked with the same hardware on 9.3.

The kenel trace is not unfortunately written to disk so I attach its picture.
Let me know what other information I should provide.
Comment 1 Martin Vidner 2005-10-27 13:20:18 UTC
Created attachment 55705 [details]
BUG_swsusp_c_905.jpg - kernel BUG screenshot
Comment 2 Pavel Machek 2005-10-30 11:10:38 UTC
I've seen that BUG() before, but I do not know how to reproduce it; it only happens rarely :-(. Having reproducible way to trigger it would help a lot.
Comment 3 Martin Vidner 2005-11-03 19:50:02 UTC
On my machine it happens quite often, but I haven't found a way to reproduce it reliably yet. Can I do anything to help tracking it down in the meantime?
Comment 4 Pavel Machek 2005-11-07 09:19:58 UTC
The code was completely rewritten in 2.6.14-git... you can switch to that to get rid of the bug (but perhaps get some new one). You may also just remove the BUG_ON(). It is condition I did not expect and condition I do not understand, but it is probably not harmfull.
Comment 5 Pavel Machek 2006-01-11 10:19:09 UTC
SLES10 should have 2.6.15, so it should be fixed at least there.
Comment 6 Martin Vidner 2006-01-11 10:24:53 UTC
Yes.
On SL 10.0 I seem to have found a workaround of swapoff -a; swapon -a before suspend.
On 2.6.15 I think I saw the "0 pages freed" problem too, which made the suspend fail (but did not oops), but after the swap trick it worked.
Comment 7 Pavel Machek 2006-01-11 10:28:06 UTC
Thanks for testing. I'll try to remember swapoff/swapon trick.