Bug 2080133

Summary:	suggestion: default x86_64 cpu to x86_64-v2 (or -cpu host)
Product:	[Fedora] Fedora	Reporter:	Jens Petersen <petersen>
Component:	qemu	Assignee:	Fedora Virtualization Maintainers <virt-maint>
Status:	CLOSED WONTFIX	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	rawhide	CC:	berrange, cfergeau, crobinso, ondrejj, pbonzini, philmd, rjones, virt-maint
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2022-05-17 15:55:12 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:

Description Jens Petersen 2022-04-29 05:54:57 UTC

Description of problem:
Centos Stream 9 and RHEL9 now assumes a x86_64 cpu supporting x86_64-v2.
https://developers.redhat.com/blog/2021/01/05/building-red-hat-enterprise-linux-9-for-the-x86-64-v2-microarchitecture-level

This means that for example (bug 2079915) that booting a RHEL9 iso with 
"qemu-kvm -m 2048 -boot d -cdrom ..." panics, which is a little surprising.

As noted in the above bug a workaround is to use "-cpu host",
but that is probably not immediately obvious to most users.

So I am wonder if it would be better to try to default qemu-kvm to "-cpu host"
or x86_64-v2 for a qemu-system-x86 vm?

Comment 1 Cole Robinson 2022-05-17 15:55:12 UTC

Thanks for the report. It's already generated a lot of bugs and discussions elsewhere, See this upstream qemu thread for example from almost a year ago: https://lore.kernel.org/qemu-devel/20210607135843.196595-1-berrange@redhat.com/

But we won't change Fedora to differ from upstream here. Until upstream changes, best to consider `-cpu host` a mandatory option, like `-m XXX` is to get a usable ram mount.

Comment 2 Jens Petersen 2022-05-18 04:14:20 UTC

Okay thanks, Cole

((I still feel from my limited perspective that `-cpu host` would be a pretty sane default anyway.))

Comment 3 Daniel Berrangé 2022-05-18 08:20:19 UTC

Using '-cpu host' would mean that live migration is significantly restricted in a default scenario, so that's non-trivial change.  At the libvirt level we have a 'host-model' which is close to '-cpu host' but with improve live migration support - some apps using libvirt explicitly set this as a default, such that the QEMU default is irrelevant. Overall we expect the majority of people to be using apps built using libvirt, rather than invoking QEMU directly, because there is sooo much more to configuring a well performing VM than the few CLI args shown aove.

Also note it would actually have to be '-cpu max', because '-cpu host' doesn't exist as a model for TCG (emulation), only for KVM.  '-cpu max' is equivalent to '-cpu host'.