Bug 16264 - boot failure possibly due to memory corruption
Summary: boot failure possibly due to memory corruption
Status: RESOLVED UNREPRODUCIBLE
Alias: None
Product: Memory Management
Classification: Unclassified
Component: Slab Allocator (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Andrew Morton
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-06-21 15:10 UTC by Denis Laxalde
Modified: 2011-06-26 21:13 UTC (History)
0 users

See Also:
Kernel Version: 2.6.32-10
Subsystem:
Regression: No
Bisected commit-id:

Attachments
dmesg from boot failure (68.04 KB, application/octet-stream)
2010-06-21 15:10 UTC, Denis Laxalde 		Details
another dmesg (72.49 KB, application/octet-stream)
2010-06-21 15:12 UTC, Denis Laxalde 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Denis Laxalde 2010-06-21 15:10:23 UTC 
Created attachment 26882 [details]
dmesg from boot failure

Original report in Debian BTS: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=575924

Boot sometimes fails on my laptop (see [1] for details).
Initially (with version 2.6.32-10) I could recover using SysRq key combinations and save the dmesg (attached) but now (from -13 I think, and now -15) it's no longer possible.

[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=575924#5
Comment 1 Denis Laxalde 2010-06-21 15:12:20 UTC 
Created attachment 26883 [details]
another dmesg
Comment 2 Andrew Morton 2010-06-22 21:04:26 UTC 
Geeze, messy.  Yes, it looks like either something is randomly scribbling on memory or your hardware is unreliable.

First thing to do is to enable every kernel debugging option you can get your hands on, see if that detects anything.  Documentation/SubmitChecklist section 12 has a list, but that's probably out of date.

If that doesn't help then all I can suggest is to perform a bisection.  http://landley.net/writing/git-quick.html has some tips.
Comment 3 Denis Laxalde 2010-12-14 17:33:08 UTC 
I'm afraid that the problem was solved by some upgrade (probably around 2.6.32.16) since I haven't experienced any failure since July (previously, this would happen once a week).
Note You need to log in before you can comment on or make changes to this bug.