From mboxrd@z Thu Jan  1 00:00:00 1970
From: Radim =?utf-8?B?S3LEjW3DocWZ?= <rkrcmar@redhat.com>
Subject: Re: [PATCH v1 06/11] KVM: x86: use hardware-compatible format for
 APIC ID register
Date: Fri, 1 Jul 2016 15:11:20 +0200
Message-ID: <20160701131119.GC27840@potion>
References: <20160630205429.16480-1-rkrcmar@redhat.com>
 <20160630205429.16480-7-rkrcmar@redhat.com>
 <733b706a-f30f-3854-5fc7-402d6b5fd79a@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
	"Lan, Tianyu" <tianyu.lan@intel.com>,
	Igor Mammedov <imammedo@redhat.com>,
	Jan Kiszka <jan.kiszka@web.de>, Peter Xu <peterx@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <733b706a-f30f-3854-5fc7-402d6b5fd79a@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org

2016-07-01 10:33+0200, Paolo Bonzini:
> On 30/06/2016 22:54, Radim Kr=C4=8Dm=C3=A1=C5=99 wrote:
>> +static void __kvm_apic_state_fixup(struct kvm_vcpu *vcpu,
>> +		struct kvm_lapic_state *s, bool set)
>> +{
>> +	if (apic_x2apic_mode(vcpu->arch.apic)) {
>> +		u32 *id =3D (u32 *)(s->regs + APIC_ID);
>> +		if (set)
>> +			*id >>=3D 24;
>> +		else
>> +			*id <<=3D 24;
>> +	}
>=20
> When setting, this should read from the apic_base being set.  So I th=
ink
> you cannot use apic_x2apic_mode.

apic_x2apic_mode uses apic_base MSR, so its value does not depend on
LAPIC_SET/GET.  I don't like the dependency much, but all combinations
of values/orders should work well.

> With this change, you don't need the struct kvm_vcpu argument now; ad=
d a
> struct kvm argument instead in patch 10.

I don't think we can use hardware-compatible format in API without
knowing the x2apic mode through vcpu.
(The big endian format could do without knowing apic mode.)

>> @@ -2780,6 +2780,7 @@ static int kvm_vcpu_ioctl_get_lapic(struct kvm=
_vcpu *vcpu,
>>  		kvm_x86_ops->sync_pir_to_irr(vcpu);
>> =20
>>  	memcpy(s->regs, vcpu->arch.apic->regs, sizeof *s);
>> +	kvm_apic_state_get_fixup(vcpu, s);
>=20
> Instead of kvm_apic_state_get/set_fixup, group the memcpy and
> __kvm_apic_state_fixup in a new function kvm_apic_get_state(struct
> kvm_lapic *apic, char *regs).

Ok.

>>  	return 0;
>>  }
>> @@ -2787,6 +2788,7 @@ static int kvm_vcpu_ioctl_get_lapic(struct kvm=
_vcpu *vcpu,
>>  static int kvm_vcpu_ioctl_set_lapic(struct kvm_vcpu *vcpu,
>>  				    struct kvm_lapic_state *s)
>>  {
>> +	kvm_apic_state_set_fixup(vcpu, s);
>>  	kvm_apic_post_state_restore(vcpu, s);
>=20
> ... and likewise merge these two in a refactored
> kvm_apic_post_state_restore called kvm_apic_set_state(struct kvm_lapi=
c
> *apic, char *regs), by calling __kvm_apic_state_fixup before
> kvm_apic_post_state_restore's memcpy.
>=20
> With these changes I guess there's no need for the underscores in
> __kvm_apic_state_fixup.

Yes, good idea.

>                          And you can also change the struct
> kvm_lapic_state pointer in the function to a char *.

Ok, I'll pass just s->regs.

Thanks.