Re: [PATCH] x86/kvm: refine condition for checking vCPU preempted

[Sean Christopherson <seanjc@google.com>]

+Paolo, Wanpeng, and Vitaly

In the future, use get_maintainers.pl to build To: and Cc: so that the right folks
see the patch.  Not everyone habitually scours the KVM list. :-)

On Wed, Feb 15, 2023, lirongqing@baidu.com wrote:
> From: Li RongQing <lirongqing@baidu.com>
> 
> Check whether vcpu is preempted or not when HLT is trapped or there
> is not realtime hint.

Please explain _why_ there's no need to check for preemption in this setup.  What
may be obvious to you isn't necessarily obvious to reviewers or readers.

> In other words, it is unnecessary to check preemption if HLT is not
> intercepted and guest has realtime hint
> 
> Signed-off-by: Li RongQing <lirongqing@baidu.com>
> ---
>  arch/x86/kernel/kvm.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index 1cceac5..1a2744d 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -820,8 +820,10 @@ static void __init kvm_guest_init(void)
>  		has_steal_clock = 1;
>  		static_call_update(pv_steal_clock, kvm_steal_clock);
>  
> -		pv_ops.lock.vcpu_is_preempted =
> -			PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
> +		if (kvm_para_has_feature(KVM_FEATURE_PV_UNHALT) ||

Rather than have the guest rely on host KVM behavior clearing PV_UNHALT when HLT
is passed through), would it make sense to add something like KVM_HINTS_HLT_PASSTHROUGH
to more explicitly tell the guest that HLT isn't intercepted?

> +		     !kvm_para_has_hint(KVM_HINTS_REALTIME))

Misaligned indentation (one too many spaces).

> +			pv_ops.lock.vcpu_is_preempted =
> +				PV_CALLEE_SAVE(__kvm_vcpu_is_preempted);
>  	}
>  
>  	if (kvm_para_has_feature(KVM_FEATURE_PV_EOI))
> -- 
> 2.9.4
>