Re: [PATCH v7 3/4] s390: ap: implement PAPQ AQIC interception in kernel

[Pierre Morel <pmorel@linux.ibm.com>]

On 30/04/2019 10:18, Pierre Morel wrote:
> On 29/04/2019 18:50, Halil Pasic wrote:
>> On Fri, 26 Apr 2019 15:01:27 +0200
>> Pierre Morel <pmorel@linux.ibm.com> wrote:
>>
>>> +static struct ap_queue_status vfio_ap_setirq(struct vfio_ap_queue *q)
>>> +{
>>> +    struct ap_qirq_ctrl aqic_gisa = {};
>>> +    struct ap_queue_status status = {};
>>> +    struct kvm_s390_gisa *gisa;
>>> +    struct kvm *kvm;
>>> +    unsigned long h_nib, h_pfn;
>>> +    int ret;
>>> +
>>> +    q->a_pfn = q->a_nib >> PAGE_SHIFT;
>>> +    ret = vfio_pin_pages(mdev_dev(q->matrix_mdev->mdev), &q->a_pfn, 1,
>>> +                 IOMMU_READ | IOMMU_WRITE, &h_pfn);
>>> +    switch (ret) {
>>> +    case 1:
>>> +        break;
>>> +    case -EINVAL:
>>> +    case -E2BIG:
>>> +        status.response_code = AP_RESPONSE_INVALID_ADDRESS;
>>> +        /* Fallthrough */
>>> +    default:
>>> +        return status;
>>
>> Can we actually hit the default label? AFICT you would return an
>> all-zero status, i.e. status.response_code == 0 'Normal completion'.
> 
> hum right, the setting of AP_INVALID_ADDRESS should be in the default 
> and there is no need for the two error cases, they will match the default.
> 
> 
>>
>>> +    }
>>> +
>>> +    kvm = q->matrix_mdev->kvm;
>>> +    gisa = kvm->arch.gisa_int.origin;
>>> +
>>> +    h_nib = (h_pfn << PAGE_SHIFT) | (q->a_nib & ~PAGE_MASK);
>>> +    aqic_gisa.gisc = q->a_isc;
>>> +    aqic_gisa.isc = kvm_s390_gisc_register(kvm, q->a_isc);
>>> +    aqic_gisa.ir = 1;
>>> +    aqic_gisa.gisa = gisa->next_alert >> 4;
>>
>> Why gisa->next_alert? Isn't this supposed to get set to gisa origin
>> (without some bits on the left)?
> 
> Someone already asked this question.
> The answer is: look at the ap_qirq_ctrl structure, you will see that the 
> gisa field is 27 bits wide.
> 
>>
>>> +
>>> +    status = ap_aqic(q->apqn, aqic_gisa, (void *)h_nib);
>>> +    switch (status.response_code) {
>>> +    case AP_RESPONSE_NORMAL:
>>> +        /* See if we did clear older IRQ configuration */
>>> +        if (q->p_pfn)
>>> +            vfio_unpin_pages(mdev_dev(q->matrix_mdev->mdev),
>>> +                     &q->p_pfn, 1);
>>> +        if (q->p_isc != VFIO_AP_ISC_INVALID)
>>> +            kvm_s390_gisc_unregister(kvm, q->p_isc);
>>> +        q->p_pfn = q->a_pfn;
>>> +        q->p_isc = q->a_isc;
>>> +        break;
>>> +    case AP_RESPONSE_OTHERWISE_CHANGED:
>>> +        /* We could not modify IRQ setings: clear new configuration */
>>> +        vfio_unpin_pages(mdev_dev(q->matrix_mdev->mdev), &q->a_pfn, 1);
>>> +        kvm_s390_gisc_unregister(kvm, q->a_isc);
>>
>> Hm, see below. Wouldn't you want to set a_isc to VFIO_AP_ISC_INVALID?
> 
> grrr!!! when did I insert these 3 lines, it was OK in previous series!
> all 3 lines, vfio_unpin() / gisc_unregister / break must go away.

No it wasn't, I will correct this.

> 
>>
>>> +        break;
>>> +    default:    /* Fall Through */
>>
>> Is it 'break' or is it 'Fall Through'?
> 
> it is a fall through
> 
>>
>>> +        pr_warn("%s: apqn %04x: response: %02x\n", __func__, q->apqn,
>>> +            status.response_code);
>>> +        vfio_ap_free_irq_data(q);
>>> +        break;
>>> +    }
>>> +
>>> +    return status;
>>> +}
> 
> 


-- 
Pierre Morel
Linux/KVM/QEMU in Böblingen - Germany