Re: [PATCH v6 4/7] vfio: ap: register IOMMU VFIO notifier

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

On 28/03/2019 21:46, Tony Krowiak wrote:
> On 3/22/19 10:43 AM, Pierre Morel wrote:
>> To be able to use the VFIO interface to facilitate the
>> mediated device memory pinning/unpinning we need to register
>> a notifier for IOMMU.
>>
>> While we will start to pin one guest page for the interrupt indicator
>> byte, this is still ok with ballooning as this page will never be
>> used by the guest virtio-balloon driver.
>> So the pinned page will never be freed. And even a broken guest does
>> so, that would not impact the host as the original page is still
>> in control by vfio.
> 
> I apologize, but I do not understand what you are saying in the second
> sentence of the paragraph above. Why will the pinned page never be 
> freed?
Because it is in use by the guest's kernel as a notification information 
byte for the original PQAP AQIC.

  I understand that the pinned page is under the control of vfio
> until it is freed, but have no idea what you mean by "and even a broken
> guest does so"? A broken guest does what? Can you please reword this so
> it makes more sense?

A broken guest could free the page used for the NIB. What is obviously 
wrong.

> 
>>
>> Signed-off-by: Pierre Morel <pmorel@linux.ibm.com>
>> Reviewed-by: Cornelia Huck <cohuck@redhat.com>
>> ---
>>   drivers/s390/crypto/vfio_ap_ops.c     | 38 
>> +++++++++++++++++++++++++++++++++++
>>   drivers/s390/crypto/vfio_ap_private.h |  2 ++
>>   2 files changed, 40 insertions(+)
>>
>> diff --git a/drivers/s390/crypto/vfio_ap_ops.c 
>> b/drivers/s390/crypto/vfio_ap_ops.c
>> index bdb36e0..3478499 100644
>> --- a/drivers/s390/crypto/vfio_ap_ops.c
>> +++ b/drivers/s390/crypto/vfio_ap_ops.c
>> @@ -787,6 +787,35 @@ static const struct attribute_group 
>> *vfio_ap_mdev_attr_groups[] = {
>>       NULL
>>   };
>> +/**
>> + * vfio_ap_mdev_iommu_notifier: IOMMU notifier callback
>> + *
>> + * @nb: The notifier block
>> + * @action: Action to be taken
>> + * @data: data associated with the request
>> + *
>> + * For an UNMAP request, unpin the guest IOVA (the NIB guest address we
>> + * pinned before). Other requests are ignored.
>> + *
>> + */
>> +static int vfio_ap_mdev_iommu_notifier(struct notifier_block *nb,
>> +                       unsigned long action, void *data)
>> +{
>> +    struct ap_matrix_mdev *matrix_mdev;
>> +
>> +    matrix_mdev = container_of(nb, struct ap_matrix_mdev, 
>> iommu_notifier);
>> +
> 
> I don't understand why we registered this notifier. I may be wrong, but
> AFAIU, this notifier will be invoked only when the VFIO_IOMMU_UNMAP_DMA
> ioctl is called from userspace. I did an experiment and inserted some
> printf's to see if this ever gets called and verified it does not. Maybe
> you have a good reason of which I'm not aware. Can you enlighten me
> here?

The vfio_iommu_type1 pin page requires a notifier.

Regards,
Pierre

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