Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH v4 05/15] KVM: s390: enable/disable AP interpretive
 execution
To:     Pierre Morel <pmorel@linux.vnet.ibm.com>,
        linux-s390@vger.kernel.org, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org, buendgen@de.ibm.com
Cc:     freude@de.ibm.com, schwidefsky@de.ibm.com,
        heiko.carstens@de.ibm.com, borntraeger@de.ibm.com,
        cohuck@redhat.com, kwankhede@nvidia.com,
        bjsdjshi@linux.vnet.ibm.com, pbonzini@redhat.com,
        alex.williamson@redhat.com, alifm@linux.vnet.ibm.com,
        mjrosato@linux.vnet.ibm.com, jjherne@linux.vnet.ibm.com,
        thuth@redhat.com, pasic@linux.vnet.ibm.com, berrange@redhat.com,
        fiuczy@linux.vnet.ibm.com, akrowiak@linux.vnet.ibm.com
References: <1523827345-11600-1-git-send-email-akrowiak@linux.vnet.ibm.com>
 <1523827345-11600-6-git-send-email-akrowiak@linux.vnet.ibm.com>
 <15b60572-40d6-0f03-11f9-c50cb7eb00e8@linux.vnet.ibm.com>
 <cf928702-db64-93c7-9b89-bf1c66a33cfa@linux.vnet.ibm.com>
 <6cba35df-b0de-8000-bb39-c4cec8622c57@linux.vnet.ibm.com>
 <2b053349-071e-17ed-6ebd-a37bcfd2f330@linux.vnet.ibm.com>
 <f01eb8fa-8a63-92f2-c15f-2c9a4e198f65@linux.vnet.ibm.com>
 <0d18aeef-dfe7-89b6-f91e-022a71fb5ed9@linux.vnet.ibm.com>
 <faef5768-8eaa-a5c5-f2d5-10a0a67c2aa0@linux.vnet.ibm.com>
From:   Tony Krowiak <akrowiak@linux.vnet.ibm.com>
Date:   Thu, 19 Apr 2018 10:28:50 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.0
MIME-Version: 1.0
In-Reply-To: <faef5768-8eaa-a5c5-f2d5-10a0a67c2aa0@linux.vnet.ibm.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Message-Id: <8c917b82-15cc-ad95-4195-f70cd67ddd6d@linux.vnet.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 04/18/2018 04:31 AM, Pierre Morel wrote:
> On 17/04/2018 20:11, Tony Krowiak wrote:
>> On 04/17/2018 12:55 PM, Pierre Morel wrote:
>>> On 17/04/2018 18:22, Tony Krowiak wrote:
>>>> On 04/17/2018 12:13 PM, Pierre Morel wrote:
>>>>> On 17/04/2018 17:02, Tony Krowiak wrote:
>>>>>> On 04/16/2018 06:51 AM, Pierre Morel wrote:
>>>>>>> On 15/04/2018 23:22, Tony Krowiak wrote:
>>>>>>>> The VFIO AP device model exploits interpretive execution of AP
>>>>>>>> instructions (APIE) to provide guests passthrough access to AP
>>>>>>>> devices. This patch introduces a new interface to enable and
>>>>>>>> disable APIE.
>>>>>>>>
>>>>>>>> Signed-off-by: Tony Krowiak <akrowiak@linux.vnet.ibm.com>
>>>>>>>> ---
>>>>>>>>   arch/s390/include/asm/kvm-ap.h   |   16 ++++++++++++++++
>>>>>>>>   arch/s390/include/asm/kvm_host.h |    1 +
>>>>>>>>   arch/s390/kvm/kvm-ap.c           |   20 ++++++++++++++++++++
>>>>>>>>   arch/s390/kvm/kvm-s390.c         |    9 +++++++++
>>>>>>>>   4 files changed, 46 insertions(+), 0 deletions(-)
>>>>>>>>
>>>>>>>> diff --git a/arch/s390/include/asm/kvm-ap.h 
>>>>>>>> b/arch/s390/include/asm/kvm-ap.h
>>>>>>>> index 736e93e..a6c092e 100644
>>>>>>>> --- a/arch/s390/include/asm/kvm-ap.h
>>>>>>>> +++ b/arch/s390/include/asm/kvm-ap.h
>>>>>>>> @@ -35,4 +35,20 @@
>>>>>>>>    */
>>>>>>>>   void kvm_ap_build_crycbd(struct kvm *kvm);
>>>>>>>>
>>>>>>>> +/**
>>>>>>>> + * kvm_ap_interpret_instructions
>>>>>>>> + *
>>>>>>>> + * Indicate whether AP instructions shall be interpreted. If 
>>>>>>>> they are not
>>>>>>>> + * interpreted, all AP instructions will be intercepted and 
>>>>>>>> routed back to
>>>>>>>> + * userspace.
>>>>>>>> + *
>>>>>>>> + * @kvm: the virtual machine attributes
>>>>>>>> + * @enable: indicates whether AP instructions are to be 
>>>>>>>> interpreted (true) or
>>>>>>>> + *        or not (false).
>>>>>>>> + *
>>>>>>>> + * Returns 0 if completed successfully; otherwise, returns 
>>>>>>>> -EOPNOTSUPP
>>>>>>>> + * indicating that AP instructions are not installed on the 
>>>>>>>> guest.
>>>>>>>> + */
>>>>>>>> +int kvm_ap_interpret_instructions(struct kvm *kvm, bool enable);
>>>>>>>> +
>>>>>>>>   #endif /* _ASM_KVM_AP */
>>>>>>>> diff --git a/arch/s390/include/asm/kvm_host.h 
>>>>>>>> b/arch/s390/include/asm/kvm_host.h
>>>>>>>> index 3162783..5470685 100644
>>>>>>>> --- a/arch/s390/include/asm/kvm_host.h
>>>>>>>> +++ b/arch/s390/include/asm/kvm_host.h
>>>>>>>> @@ -715,6 +715,7 @@ struct kvm_s390_crypto {
>>>>>>>>       __u32 crycbd;
>>>>>>>>       __u8 aes_kw;
>>>>>>>>       __u8 dea_kw;
>>>>>>>> +    __u8 apie;
>>>>>>>>   };
>>>>>>>>
>>>>>>>>   #define APCB0_MASK_SIZE 1
>>>>>>>> diff --git a/arch/s390/kvm/kvm-ap.c b/arch/s390/kvm/kvm-ap.c
>>>>>>>> index 991bae4..55d11b5 100644
>>>>>>>> --- a/arch/s390/kvm/kvm-ap.c
>>>>>>>> +++ b/arch/s390/kvm/kvm-ap.c
>>>>>>>> @@ -58,3 +58,23 @@ void kvm_ap_build_crycbd(struct kvm *kvm)
>>>>>>>>       }
>>>>>>>>   }
>>>>>>>>   EXPORT_SYMBOL(kvm_ap_build_crycbd);
>>>>>>>> +
>>>>>>>> +int kvm_ap_interpret_instructions(struct kvm *kvm, bool enable)
>>>>>>>> +{
>>>>>>>> +    int ret = 0;
>>>>>>>> +
>>>>>>>> +    mutex_lock(&kvm->lock);
>>>>>>>> +
>>>>>>>> +    if (!test_kvm_cpu_feat(kvm, KVM_S390_VM_CPU_FEAT_AP)) {
>>>>>>>
>>>>>>> Do we really need to test CPU_FEAT_AP?
>>>>>>
>>>>>> Yes we do.
>>>>>
>>>>> really? why?
>>>>
>>>> The KVM_S390_VM_CPU_FEAT_AP will not be enabled by KVM if the AP
>>>> instructions are not installed on the host. I assume - but have
>>>> no way of verifying - that if the AP instructions are not installed
>>>> on the host, that interpretation would fail. Do you know what would
>>>> happen if AP instructions are interpreted when not installed on
>>>> the host?
>>>
>>> If the host has no AP instructions (his ECA.28=0) but it set ECA.28 
>>> for a guest,
>>> there will be no AP instructions available in the guest.
>>
>> Then there's the answer to your question; this is why we to test 
>> CPU_FEAT_AP.
>
> We can postpone this discussion when we discuss on VSIE.
> For this specific call I just wanted to point out that obviously this 
> function should not
> be called if the guest has no AP instructions.

I disagree, at least as far as the way things are currently designed. 
Whether AP instructions
are interpreted or not is determined by the vfio_ap device driver. The 
device driver should not
be required to have "knowledge" about how a guest is configured in KVM 
which is why I
encapsulated most of the AP guest configuration in kvm-ap.c. Besides, 
the function above allows
setting of AP interpretation only if CPU_FEAT_AP is enabled.

Having said that, I may remove this function since - as you pointed out 
earlier - AP instructions
are interpreted by default if CPU_FEAT_AP is enabled, so there will be 
no need to set this at
this time.

>
>>
>>>
>>>
>>>>
>>>>>
>>>>>
>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> I understand that KVM_S390_VM_CPU_FEAT_AP means AP instructions 
>>>>>>> are interpreted.
>>>>>>> shouldn't we add this information in the name?
>>>>>>> like KVM_S390_VM_CPU_FEAT_APIE
>>>>>>
>>>>>> KVM_S390_VM_CPU_FEAT_AP does NOT mean AP instructions are 
>>>>>> interpreted, it means
>>>>>> AP instructions are installed.
>>>>>
>>>>> Right same error I made all along this review.
>>>>> But AFAIK it means AP instructions are provided to the guest.
>>>>> Then should this function be called if the guest has no AP 
>>>>> instructions ?
>>>>>
>>>>>
>>>>>>
>>>>>>>
>>>>>>>> +        ret = -EOPNOTSUPP;
>>>>>>>> +        goto done;
>>>>>>>> +    }
>>>>>>>> +
>>>>>>>> +    kvm->arch.crypto.apie = enable;
>>>>>>>> +    kvm_s390_vcpu_crypto_reset_all(kvm);
>>>>>>>> +
>>>>>>>> +done:
>>>>>>>> +    mutex_unlock(&kvm->lock);
>>>>>>>> +    return ret;
>>>>>>>> +}
>>>>>>>> +EXPORT_SYMBOL(kvm_ap_interpret_instructions);
>>>>>>>> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
>>>>>>>> index 55cd897..1dc8566 100644
>>>>>>>> --- a/arch/s390/kvm/kvm-s390.c
>>>>>>>> +++ b/arch/s390/kvm/kvm-s390.c
>>>>>>>> @@ -1901,6 +1901,9 @@ static void kvm_s390_crypto_init(struct 
>>>>>>>> kvm *kvm)
>>>>>>>>       kvm->arch.crypto.crycb = &kvm->arch.sie_page2->crycb;
>>>>>>>>       kvm_ap_build_crycbd(kvm);
>>>>>>>>
>>>>>>>> +    /* Default setting indicating SIE shall interpret AP 
>>>>>>>> instructions */
>>>>>>>> +    kvm->arch.crypto.apie = 1;
>>>>>>>> +
>>>>>>>>       if (!test_kvm_facility(kvm, 76))
>>>>>>>>           return;
>>>>>>>>
>>>>>>>> @@ -2434,6 +2437,12 @@ static void 
>>>>>>>> kvm_s390_vcpu_crypto_setup(struct kvm_vcpu *vcpu)
>>>>>>>>   {
>>>>>>>>       vcpu->arch.sie_block->crycbd = 
>>>>>>>> vcpu->kvm->arch.crypto.crycbd;
>>>>>>>>
>>>>>>>> +    vcpu->arch.sie_block->eca &= ~ECA_APIE;
>>>>>>>> +    if (vcpu->kvm->arch.crypto.apie &&
>>>>>>>> +        test_kvm_cpu_feat(vcpu->kvm, KVM_S390_VM_CPU_FEAT_AP))
>>>>>>>
>>>>>>> Do we call xxx_crypto_setup() if KVM does not support AP 
>>>>>>> interpretation?
>>>>>>
>>>>>> Yes, kvm_s390_vcpu_crypto_setup(vcpu) is called by 
>>>>>> kvm_arch_vcpu_setup(vcpu)
>>>>>> as well as from kvm_s390_vcpu_crypto_reset_all(kvm). Calling it 
>>>>>> has nothing
>>>>>> to do with whether AP interpretation is supported or not as it 
>>>>>> does much
>>>>>> more than that, including setting up of wrapping keys and the 
>>>>>> CRYCBD.
>>>>>
>>>>> Sorry, still the same error I made about CPU_FEAT_AP meaning AP 
>>>>> instructions in the guest
>>>>> and not AP interpretation available.
>>>>> Could apie be set if AP instruction are not supported?
>>>>>
>>>>>>
>>>>>>>
>>>>>>>> + vcpu->arch.sie_block->eca |= ECA_APIE;
>>>>>>>> +
>>>>>>>> +
>>>>>>>>       if (!test_kvm_facility(vcpu->kvm, 76))
>>>>>>>>           return;
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>