Re: Commit "x86/kvm: Move context tracking where it belongs" broke guest time accounting

[Sean Christopherson]

On Tue, Apr 06, 2021, Michael Tokarev wrote:
> Hi!
> 
> It looks like this commit:
> 
> commit 87fa7f3e98a1310ef1ac1900e7ee7f9610a038bc
> Author: Thomas Gleixner <tglx@linutronix.de>
> Date:   Wed Jul 8 21:51:54 2020 +0200
> 
>     x86/kvm: Move context tracking where it belongs
> 
>     Context tracking for KVM happens way too early in the vcpu_run()
>     code. Anything after guest_enter_irqoff() and before guest_exit_irqoff()
>     cannot use RCU and should also be not instrumented.
> 
>     The current way of doing this covers way too much code. Move it closer to
>     the actual vmenter/exit code.
> 
> broke kvm guest cpu time accounting - after this commit, when running
> qemu-system-x86_64 -enable-kvm, the guest time (in /proc/stat and
> elsewhere) is always 0.
> 
> I dunno why it happened, but it happened, and all kernels after 5.9
> are affected by this.
> 
> This commit is found in a (painful) git bisect between kernel 5.8 and 5.10.

Yes :-(

There's a bugzilla[1] and two proposed fixes[2][3].  I don't particularly like
either of the fixes, but an elegant solution hasn't presented itself.

Thomas/Paolo, can you please weigh in?

[1] https://bugzilla.kernel.org/show_bug.cgi?id=209831
[2] 
https://lkml.kernel.org/r/1617011036-11734-1-git-send-email-wanpengli@tencent.com
[3] https://lkml.kernel.org/r/20210206004218.312023-1-seanjc@google.com