Bug 121829 - Kernel boot failure during resume from swap on RAID
Summary: Kernel boot failure during resume from swap on RAID
Status: RESOLVED DUPLICATE of bug 130693
: 120973 (view as bug list)
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: Kernel (show other bugs)
Version: Final
Hardware: i686 All
: P5 - None : Normal
Target Milestone: ---
Assignee: Pavel Machek
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-10-08 07:09 UTC by Krzysztof Lichota
Modified: 2005-10-30 11:43 UTC (History)
2 users (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Krzysztof Lichota 2005-10-08 07:09:58 UTC
Note: This Suse 10.0 has been installed from 1-CD image 
super-openSUSE-final-20051003-standard-kde.iso. 
 
I have a bit unordinary setup as my swap partition is on RAID0 partition 
(md4). I have Suse 9.3 installed and it uses this partition. 
I have installed Suse 10.0 on another paritition, it installed fine. The only 
glitch was that during partitioning it offered to create swap partition, 
although I already have swap (on md4). The strange thing is that it showed 
"Use swap: md4" in parititioning summary but it offered to create another swap 
anyway. But it is not a big problem. 
 
The major bug is that after installation, when I tried to boot, I got kernel 
panic. The messages were: 
Waiting for device /dev/md4 to appear: ok 
No record for 'md4' in database 
Attempting manual resume 
Kernel panic - not syncing: I/O error reading memory image 
 
I have tried booting in safe mode and it worked. I hoped that when swap is 
used by Suse 10.0 it will start working. But if I try to boot using normal 
boot configuration (not "safe mode") I get this error again.
Comment 1 Hubert Mantel 2005-10-10 13:46:01 UTC
I'm not sure I really understand that. Do you see the problem with a regular
boot or when resuming (as the summary suggests)?
Comment 2 Krzysztof Lichota 2005-10-10 14:15:07 UTC
(In reply to comment #1)
> I'm not sure I really understand that. Do you see the problem with a regular
> boot or when resuming (as the summary suggests)?

Right, I did not make it clear. It is normal boot, it is just checking for
possibility of resuming from swap. Please change summary to something more
appropriate.
Comment 3 Hubert Mantel 2005-10-20 14:25:46 UTC
I'm not totally sure where the problem really is. Pavel, can you please have a look?
Comment 4 Pavel Machek 2005-10-21 11:20:43 UTC
I don't know how to fix this one sanely. I could make read i/o error non-fatal, but that would be quite a big change, and would require a *lot* of testing. I think it is too late to do that.

I believe that we should mark suspend on raid as unsupported, and not attempt to set it up for resume if user wants to do that. [Or perhaps even disallow suspend on raid at all, it is pretty dangerous w.r.t. deadlocks, and probably not too tested].

Could you

boot
check that /dev/md4 is set up properly (cat /dev/md4 works)
sync
attempt manual resume?

It should tell you that there's no suspend signature in there... If yes, bug is in initrd, if no, bug is in kernel bug I'm not sure if we can fix it anytime soon.


Comment 5 Pavel Machek 2005-10-22 18:32:38 UTC
*** Bug 120973 has been marked as a duplicate of this bug. ***
Comment 6 Krzysztof Lichota 2005-10-23 19:35:52 UTC
(In reply to comment #4)
> Could you
> 
> boot
> check that /dev/md4 is set up properly (cat /dev/md4 works)
> sync
> attempt manual resume?
> 
> It should tell you that there's no suspend signature in there... If yes, bug is
> in initrd, if no, bug is in kernel bug I'm not sure if we can fix it anytime
> soon.

I have booted in safe mode, cat /dev/md4 worked OK. 
When I tried to suspend (using powersave -U), it hanged on "Freeing memory".
After reboot I have tried to boot in normal mode, but the same error as described in bug report happened.
Comment 7 Forgotten User ZhJd0F0L3x 2005-10-23 20:27:19 UTC
unfortunately it is a bit more complicated.
You should: (Pavel, correct me if i am wrong :-)

- boot "failsave", with the additional parameter "init=/bin/bash", this will give you a very minimal setup, only a shell. Beware: you will have US keyboard with that.
- "mount /proc"
- "mount /sys"
- "ls -l /dev/md4"
this gives (at my machine):
brw-rw----  1 root root 9, 4 2005-10-23 12:26 /dev/md4
- now "echo 9:4 > /sys/power/resume"
  9:4 are the two numbers after "root root" in ls -l output.

If this gives you the same error as always, it is a kernel bug. If it gives "could not find suspend signature", the bug is somewhere else.
Comment 8 Krzysztof Lichota 2005-10-26 11:13:15 UTC
(In reply to comment #7)
> unfortunately it is a bit more complicated.
> You should: (Pavel, correct me if i am wrong :-)
> 
> - boot "failsave", with the additional parameter "init=/bin/bash", this will
> give you a very minimal setup, only a shell. Beware: you will have US keyboard
> with that.

I have tried it and during boot it says it cannot find /bin/bash. I have tried also /bin/sh, but it's the same. Maybe it is not on initrd?
Comment 9 Forgotten User ZhJd0F0L3x 2005-10-26 12:36:39 UTC
hm, i am neither a master of md nor of initrd, maybe Hannes can give us a hint :-)

Hannes, anything special i need to do if i want a md-root system to boot with init=/bin/bash?
Comment 10 Pavel Machek 2005-10-30 11:43:26 UTC

*** This bug has been marked as a duplicate of 130693 ***