Bug 824991

Summary:	per-machine-type CPU models for safe migration
Product:	Red Hat Enterprise Linux 7	Reporter:	Eduardo Habkost <ehabkost>
Component:	qemu-kvm	Assignee:	Eduardo Habkost <ehabkost>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Virtualization Bugs <virt-bugs>
Severity:	high	Docs Contact:
Priority:	high
Version:	7.1	CC:	areis, hui.xiao, jamorgan, jane.lv, juzhang, jvillalo, jwilleford, qzhang, ruwang, virt-maint, xiaolong.wang
Target Milestone:	rc
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	upstream qemu-1.4.0	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:
Clones:	833152 (view as bug list)		Environment:
Last Closed:	2014-06-13 12:22:54 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	728045, 824987

Description Eduardo Habkost 2012-05-24 18:36:44 UTC

Sometimes we want to update an existing QEMU CPU model, because of a new feature, or because a bug needs to be fixed. But existing virtual machines using the CPU model can't change under the guest's feet when live-migrated. To make this work, we need to implement:

- Versioned CPU models, so new CPU models can get the fixes or new features while the existing CPU models keep the same.
- Per-machine-type CPU model aliases, so users can get the latest version of a CPU model automatically, but keep the same CPU when live-migrating to a newer QEMU version.

Comment 1 Eduardo Habkost 2013-02-28 15:55:55 UTC

An ad-hoc solution is already available upstream, based on using different machine-type init functions that will set different defaults on different machine-types. The alias system won't be implemented.

A better solution based on global properties should be implemented soon, but the basic functionality of "per-machine-type CPU model definitions" already exists.

To test this bug, just check if the kvm_pv_eoi feature is enabled by default only on pc-1.3 and newer.

Comment 3 Qunfang Zhang 2014-01-16 08:55:53 UTC

Verified this bug with kernel-3.10.0-72.el7.x86_64 and qemu-kvm-1.5.3-37.el7.x86_64 packages on host. 

Boot up guest and do migration on SandyBridge host. 

(1) Boot up guest with "-M pc". And check guest cpu flags.

Result: There's "tsc_deadline_timer" flag inside guest. 

(2) Boot up guest with "-M q35" and check guest cpu flags. 

Result: There's "tsc_deadline_timer" flag inside guest. 

(3) Boot up guest with "-M rhel6.5.0" and "-M rhel6.4.0" separately and check guest cpu flags.

Result: There's "tsc_deadline_timer" flag inside guest when boot with rhel6.4.0 or rhel6.5.0 machine types. 

(4) Boot up guest with "-M rhel6.3.0"/"-M rhel6.2.0"/"-M rhel6.1.0"/"-M rhel6.0.0" machine types one by one. Check guest cpu flags each time after guest boots up. 

Result: There's NO "tsc_deadline_timer" flag inside guest. 

(5) Migrate guest from rhel6.5 host to rhel7.0 host with "-M rhel6.3.0" machine type, check the guest cpu flags before and after migration. 

Result: There's NO "tsc_deadline_timer" flag inside guest before and after migration. Migration could finish successfully. 

(6) Migrate guest from rhel6.5 host to rhel7.0 host with "-M rhel6.5.0" machine type, check the guest cpu flags before and after migration. 

Result: There's "tsc_deadline_timer" flag inside guest before and after migration. Migration could finish successfully. 

Based on above, this bug is fixed.

Comment 4 Ludek Smid 2014-06-13 12:22:54 UTC

This request was resolved in Red Hat Enterprise Linux 7.0.

Contact your manager or support representative in case you have further questions about the request.