Re: [PATCH v3 1/3] KVM: x86: kvm_x86_ops.get_exit_info should include the exit reason

From: David Edmondson <dme@dme.org>
To: Sean Christopherson <seanjc@google.com>
Cc: linux-kernel@vger.kernel.org,
	Thomas Gleixner <tglx@linutronix.de>,
	Joerg Roedel <joro@8bytes.org>, Ingo Molnar <mingo@redhat.com>,
	Jim Mattson <jmattson@google.com>,
	kvm@vger.kernel.org, Borislav Petkov <bp@alien8.de>,
	David Matlack <dmatlack@google.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	"H. Peter Anvin" <hpa@zytor.com>,
	x86@kernel.org, Wanpeng Li <wanpengli@tencent.com>,
	Vitaly Kuznetsov <vkuznets@redhat.com>
Subject: Re: [PATCH v3 1/3] KVM: x86: kvm_x86_ops.get_exit_info should include the exit reason
Date: Fri, 30 Jul 2021 08:29:53 +0100	[thread overview]
Message-ID: <cunsfzwmf7y.fsf@dme.org> (raw)
In-Reply-To: <YQMrUOjZMD1eiIeE@google.com>

On Thursday, 2021-07-29 at 22:27:28 GMT, Sean Christopherson wrote:

> Shortlog is a bit odd, "should" is subjective and makes this sound like a bug fix.
>
>   KVM: x86: Get exit_reason as part of kvm_x86_ops.get_exit_info

Okay.

> On Thu, Jul 29, 2021, David Edmondson wrote:
>> Extend the get_exit_info static call to provide the reason for the VM
>> exit. Modify relevant trace points to use this rather than extracting
>> the reason in the caller.
>> 
>> Signed-off-by: David Edmondson <david.edmondson@oracle.com>
>> ---
>> -static void svm_get_exit_info(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2,
>> +static void svm_get_exit_info(struct kvm_vcpu *vcpu, u64 *reason,
>> +			      u64 *info1, u64 *info2,
>>  			      u32 *intr_info, u32 *error_code)
>>  {
>>  	struct vmcb_control_area *control = &to_svm(vcpu)->vmcb->control;
>>  
>> +	*reason = control->exit_code;
>>  	*info1 = control->exit_info_1;
>>  	*info2 = control->exit_info_2;
>>  	*intr_info = control->exit_int_info;
>
> ...
>
>> diff --git a/arch/x86/kvm/trace.h b/arch/x86/kvm/trace.h
>> index b484141ea15b..2228565beda2 100644
>> --- a/arch/x86/kvm/trace.h
>> +++ b/arch/x86/kvm/trace.h
>> @@ -273,11 +273,11 @@ TRACE_EVENT(kvm_apic,
>>  
>>  #define TRACE_EVENT_KVM_EXIT(name)					     \
>>  TRACE_EVENT(name,							     \
>> -	TP_PROTO(unsigned int exit_reason, struct kvm_vcpu *vcpu, u32 isa),  \
>> -	TP_ARGS(exit_reason, vcpu, isa),				     \
>> +	TP_PROTO(struct kvm_vcpu *vcpu, u32 isa),			     \
>> +	TP_ARGS(vcpu, isa),						     \
>>  									     \
>>  	TP_STRUCT__entry(						     \
>> -		__field(	unsigned int,	exit_reason	)	     \
>> +		__field(	u64,		exit_reason	)	     \
>
> Converting to a u64 is unnecessary and misleading.  vmcs.EXIT_REASON and
> vmcb.EXIT_CODE are both u32s, a.k.a. unsigned ints.  There is vmcb.EXIT_CODE_HI,
> but that's not being included, and AFAICT isn't even sanity checked by KVM.

Thanks for pointing this out, I can only blame brain fade.

>>  		__field(	unsigned long,	guest_rip	)	     \
>>  		__field(	u32,	        isa             )	     \
>>  		__field(	u64,	        info1           )	     \