Bug 684126

Summary:	cpu1 can not be up on s390x machine: The virtual machine is placed in CP mode due to a SIGP stop from CPU 01
Product:	Red Hat Enterprise Linux 6	Reporter:	Zhang Kexin <kzhang>
Component:	kernel	Assignee:	Hendrik Brueckner <brueckner>
Status:	CLOSED NOTABUG	QA Contact:	Red Hat Kernel QE team <kernel-qe>
Severity:	high	Docs Contact:
Priority:	high
Version:	6.2	CC:	arozansk, emcnabb, jjarvis, sbest
Target Milestone:	rc
Target Release:	---
Hardware:	s390x
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-03-23 15:48:55 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Zhang Kexin 2011-03-11 08:29:22 UTC

Description of problem:
When I do test on s390x machine, the console printed following messages:
--------------------------------
ibm-z10-25 login:
01: HCPGSP2627I The virtual machine is placed in CP mode due to a SIGP initial C
PU reset from CPU 00.
cpu: Processor 1 started, address 0, identification 32C5C2
01: HCPGSP2629I The virtual machine is placed in CP mode due to a SIGP stop from
 CPU 01.
cpu: Processor 1 stopped
-------------------------

the same console output when I do "echo 1 > /sys/devices/system/cpu/cpu1/online"


Version-Release number of selected component (if applicable):
2.6.32-118

How reproducible:
always

Steps to Reproduce:
1. install RHEL6.1-20110224.2 on s390x machine
2. run some test, eg, /kernel/power-management/gettimeofday
3.
  
Actual results:
above console output is printed, and cpu1 can not be up.

Expected results:


Additional info:
installed an older kernel 2.6.32-94, the same result.

Comment 5 Hendrik Brueckner 2011-03-23 15:48:55 UTC

With RHEL6.0, a program called "cpuplugd" manages the available CPUs depending on the work load of the system.  The cpuplug daemon is enabled by default.  Whenever a CPUs is switched online/offline, you can see a message with ID "HCPGSP2629I". This is an information message from the z/VM control program (CP).

If the system is idle, the cpuplugd switches off additional CPUs, for example, to reduce z/CM CP scheduling efforts. It also allows to run with a certain overcommitment.

Regarding your issue, if you manually switch an CPU online, cpuplugd will turn it offline on the next check interval.

For your tests, you can stop cpuplugd with "/etc/init.d/cpuplugd stop", or you can use chkconfig to switch off cpuplugd permanently.