Date:   Tue, 11 Jan 2022 16:19:06 -0500
From:   Tony Krowiak <akrowiak@...ux.ibm.com>
To:     Halil Pasic <pasic@...ux.ibm.com>
Cc:     linux-s390@...r.kernel.org, linux-kernel@...r.kernel.org,
        kvm@...r.kernel.org, jjherne@...ux.ibm.com, freude@...ux.ibm.com,
        borntraeger@...ibm.com, cohuck@...hat.com, mjrosato@...ux.ibm.com,
        alex.williamson@...hat.com, kwankhede@...dia.com,
        fiuczy@...ux.ibm.com
Subject: Re: [PATCH v17 06/15] s390/vfio-ap: refresh guest's APCB by filtering
 APQNs assigned to mdev



On 12/27/21 03:53, Halil Pasic wrote:
> On Thu, 21 Oct 2021 11:23:23 -0400
> Tony Krowiak <akrowiak@...ux.ibm.com> wrote:
>
>> Refresh the guest's APCB by filtering the APQNs assigned to the matrix mdev
>> that do not reference an AP queue device bound to the vfio_ap device
>> driver. The mdev's APQNs will be filtered according to the following rules:
>>
>> * The APID of each adapter and the APQI of each domain that is not in the
>> host's AP configuration is filtered out.
>>
>> * The APID of each adapter comprising an APQN that does not reference a
>> queue device bound to the vfio_ap device driver is filtered. The APQNs
>> are derived from the Cartesian product of the APID of each adapter and
>> APQI of each domain assigned to the mdev.
>>
>> The control domains that are not assigned to the host's AP configuration
>> will also be filtered before assigning them to the guest's APCB.
> The v16 version used to filer on queue removal from vfio_ap, which makes
> a ton of sense.
>
> This version will "filter back" the queues once these become bound, but
> if a queue is removed form vfio_ap, we don't seem to care to filter. Is
> this intentional?

See patch the changes to the vfio_ap_mdev_remove_queue function in patch 
09/15,
s390/vfio-ap: allow hot plug/unplug of AP resources using mdev device

Now that I look at that patch, I should probably rearrange this patch to 
also do
filtering on queue removal, but only do the hotplug stuff in patch 09.

>
> Also we could probably do the filtering incrementally. In a sense that
> at a time only so much changes, and we know that the invariant was
> preserved without that change. But that would probably end up trading
> complexity for cycles. I will trust your judgment and your tests on this
> matter.

I am not entirely clear on what you are suggesting. I think you are
suggesting that there may not be a need to look at every APQN
assigned to the mdev when an adapter or domain is assigned or
unassigned or a queue is probed or removed. Maybe you can clarify
what you are suggesting here.

>
>> Signed-off-by: Tony Krowiak <akrowiak@...ux.ibm.com>
>> ---
>>   drivers/s390/crypto/vfio_ap_ops.c | 66 ++++++++++++++++++++++++++++++-
>>   1 file changed, 64 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c
>> index 4305177029bf..46c179363aca 100644
>> --- a/drivers/s390/crypto/vfio_ap_ops.c
>> +++ b/drivers/s390/crypto/vfio_ap_ops.c
>> @@ -314,6 +314,62 @@ static void vfio_ap_matrix_init(struct ap_config_info *info,
>>   	matrix->adm_max = info->apxa ? info->Nd : 15;
>>   }
>>   
>> +static void vfio_ap_mdev_filter_cdoms(struct ap_matrix_mdev *matrix_mdev)
>> +{
>> +	bitmap_and(matrix_mdev->shadow_apcb.adm, matrix_mdev->matrix.adm,
>> +		   (unsigned long *)matrix_dev->info.adm, AP_DOMAINS);
>> +}
>> +
>> +/*
>> + * vfio_ap_mdev_filter_matrix - copy the mdev's AP configuration to the KVM
>> + *				guest's APCB then filter the APIDs that do not
>> + *				comprise at least one APQN that references a
>> + *				queue device bound to the vfio_ap device driver.
>> + *
>> + * @matrix_mdev: the mdev whose AP configuration is to be filtered.
>> + */
>> +static void vfio_ap_mdev_filter_matrix(struct ap_matrix_mdev *matrix_mdev)
>> +{
>> +	int ret;
>> +	unsigned long apid, apqi, apqn;
>> +
>> +	ret = ap_qci(&matrix_dev->info);
>> +	if (ret)
>> +		return;
>> +
>> +	vfio_ap_matrix_init(&matrix_dev->info, &matrix_mdev->shadow_apcb);
>> +
>> +	/*
>> +	 * Copy the adapters, domains and control domains to the shadow_apcb
>> +	 * from the matrix mdev, but only those that are assigned to the host's
>> +	 * AP configuration.
>> +	 */
>> +	bitmap_and(matrix_mdev->shadow_apcb.apm, matrix_mdev->matrix.apm,
>> +		   (unsigned long *)matrix_dev->info.apm, AP_DEVICES);
>> +	bitmap_and(matrix_mdev->shadow_apcb.aqm, matrix_mdev->matrix.aqm,
>> +		   (unsigned long *)matrix_dev->info.aqm, AP_DOMAINS);
>> +
>> +	for_each_set_bit_inv(apid, matrix_mdev->shadow_apcb.apm, AP_DEVICES) {
>> +		for_each_set_bit_inv(apqi, matrix_mdev->shadow_apcb.aqm,
>> +				     AP_DOMAINS) {
>> +			/*
>> +			 * If the APQN is not bound to the vfio_ap device
>> +			 * driver, then we can't assign it to the guest's
>> +			 * AP configuration. The AP architecture won't
>> +			 * allow filtering of a single APQN, so if we're
>> +			 * filtering APIDs, then filter the APID; otherwise,
>> +			 * filter the APQI.
>> +			 */
>> +			apqn = AP_MKQID(apid, apqi);
>> +			if (!vfio_ap_mdev_get_queue(matrix_mdev, apqn)) {
>> +				clear_bit_inv(apid,
>> +					      matrix_mdev->shadow_apcb.apm);
>> +				break;
>> +			}
>> +		}
>> +	}
>> +}
>> +
>>   static int vfio_ap_mdev_probe(struct mdev_device *mdev)
>>   {
>>   	struct ap_matrix_mdev *matrix_mdev;
>> @@ -703,6 +759,7 @@ static ssize_t assign_adapter_store(struct device *dev,
>>   		goto share_err;
>>   
>>   	vfio_ap_mdev_link_adapter(matrix_mdev, apid);
>> +	vfio_ap_mdev_filter_matrix(matrix_mdev);
>>   	ret = count;
>>   	goto done;
>>   
>> @@ -771,6 +828,7 @@ static ssize_t unassign_adapter_store(struct device *dev,
>>   
>>   	clear_bit_inv((unsigned long)apid, matrix_mdev->matrix.apm);
>>   	vfio_ap_mdev_unlink_adapter(matrix_mdev, apid);
>> +	vfio_ap_mdev_filter_matrix(matrix_mdev);
>>   	ret = count;
>>   done:
>>   	mutex_unlock(&matrix_dev->lock);
>> @@ -874,6 +932,7 @@ static ssize_t assign_domain_store(struct device *dev,
>>   		goto share_err;
>>   
>>   	vfio_ap_mdev_link_domain(matrix_mdev, apqi);
>> +	vfio_ap_mdev_filter_matrix(matrix_mdev);
>>   	ret = count;
>>   	goto done;
>>   
>> @@ -942,6 +1001,7 @@ static ssize_t unassign_domain_store(struct device *dev,
>>   
>>   	clear_bit_inv((unsigned long)apqi, matrix_mdev->matrix.aqm);
>>   	vfio_ap_mdev_unlink_domain(matrix_mdev, apqi);
>> +	vfio_ap_mdev_filter_matrix(matrix_mdev);
>>   	ret = count;
>>   
>>   done:
>> @@ -995,6 +1055,7 @@ static ssize_t assign_control_domain_store(struct device *dev,
>>   	 * number of control domains that can be assigned.
>>   	 */
>>   	set_bit_inv(id, matrix_mdev->matrix.adm);
>> +	vfio_ap_mdev_filter_cdoms(matrix_mdev);
>>   	ret = count;
>>   done:
>>   	mutex_unlock(&matrix_dev->lock);
>> @@ -1042,6 +1103,7 @@ static ssize_t unassign_control_domain_store(struct device *dev,
>>   	}
>>   
>>   	clear_bit_inv(domid, matrix_mdev->matrix.adm);
>> +	clear_bit_inv(domid, matrix_mdev->shadow_apcb.adm);
>>   	ret = count;
>>   done:
>>   	mutex_unlock(&matrix_dev->lock);
>> @@ -1179,8 +1241,6 @@ static int vfio_ap_mdev_set_kvm(struct ap_matrix_mdev *matrix_mdev,
>>   		kvm_get_kvm(kvm);
>>   		matrix_mdev->kvm = kvm;
>>   		kvm->arch.crypto.data = matrix_mdev;
>> -		memcpy(&matrix_mdev->shadow_apcb, &matrix_mdev->matrix,
>> -		       sizeof(struct ap_matrix));
>>   		kvm_arch_crypto_set_masks(kvm, matrix_mdev->shadow_apcb.apm,
>>   					  matrix_mdev->shadow_apcb.aqm,
>>   					  matrix_mdev->shadow_apcb.adm);
>> @@ -1536,6 +1596,8 @@ int vfio_ap_mdev_probe_queue(struct ap_device *apdev)
>>   	q->apqn = to_ap_queue(&apdev->device)->qid;
>>   	q->saved_isc = VFIO_AP_ISC_INVALID;
>>   	vfio_ap_queue_link_mdev(q);
>> +	if (q->matrix_mdev)
>> +		vfio_ap_mdev_filter_matrix(q->matrix_mdev);
>>   	dev_set_drvdata(&apdev->device, q);
>>   	mutex_unlock(&matrix_dev->lock);
>>   

Powered by blists - more mailing lists