Bug 10851

Summary:	vmalloc problem with DEBUG_PAGEALLOC config option
Product:	Memory Management	Reporter:	Zdenek Kabelac (zdenek.kabelac)
Component:	Page Allocator	Assignee:	Andrew Morton (akpm)
Status:	CLOSED CODE_FIX
Severity:	normal	CC:	tglx
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.26-rc3	Subsystem:
Regression:	---	Bisected commit-id:

Description Zdenek Kabelac 2008-06-03 05:12:44 UTC

Latest working kernel version:
Earliest failing kernel version:
Distribution:
Hardware Environment:
Software Environment:
Problem Description:
http://lkml.org/lkml/2008/5/27/51

Steps to reproduce:
Basicaly my test case looks like two shells - one is running loop of make install and another  vgchage -an/y - this sometimes makes the nash crashing - and possibly during this segmentation fault something bad happens (the 'make install' loop must be restarted couple time as the first crash might not be the one where the problem starts)

Comment 1 Zdenek Kabelac 2008-06-09 06:01:36 UTC

adding short comment:  with recent kernel version 2.6.26-rc5 commit: aab2545fdd6641b76af0ae96456c4ca9d1e50dad it looks like I'm unable to easily hit my problem again - thus either the bug was fixed or the problem is much harder to hit. I'll keep an eye on this issue and I've started to use compiled kernel with this flag again.

Comment 2 Thomas Gleixner 2008-07-08 23:29:28 UTC

Any update to this ?

Comment 3 Zdenek Kabelac 2008-07-09 00:45:00 UTC

looks like the problem/race was fixed in the -rc5 build - I've not seen this  problem for that moment.  I guess the bug could be closed.