Re: [PATCH v7 06/15] s390/vfio-ap: sysfs attribute to display the guest CRYCB

[Tony Krowiak <akrowiak@linux.ibm.com>]

On 4/8/20 12:46 PM, Cornelia Huck wrote:
> On Wed, 8 Apr 2020 12:38:49 -0400
> Tony Krowiak <akrowiak@linux.ibm.com> wrote:
>
>> On 4/8/20 6:33 AM, Cornelia Huck wrote:
>>> On Tue,  7 Apr 2020 15:20:06 -0400
>>> Tony Krowiak <akrowiak@linux.ibm.com> wrote:
>>>   
>>>> The matrix of adapters and domains configured in a guest's CRYCB may
>>>> differ from the matrix of adapters and domains assigned to the matrix mdev,
>>>> so this patch introduces a sysfs attribute to display the CRYCB of a guest
>>>> using the matrix mdev. For a matrix mdev denoted by $uuid, the crycb for a
>>>> guest using the matrix mdev can be displayed as follows:
>>>>
>>>>      cat /sys/devices/vfio_ap/matrix/$uuid/guest_matrix
>>>>
>>>> If a guest is not using the matrix mdev at the time the crycb is displayed,
>>>> an error (ENODEV) will be returned.
>>>>
>>>> Signed-off-by: Tony Krowiak <akrowiak@linux.ibm.com>
>>>> ---
>>>>    drivers/s390/crypto/vfio_ap_ops.c | 58 +++++++++++++++++++++++++++++++
>>>>    1 file changed, 58 insertions(+)
>>>> +static DEVICE_ATTR_RO(guest_matrix);
>>> Hm... should information like the guest configuration be readable by
>>> everyone? Or should it be restricted a bit more?
>> Why? The matrix attribute already displays the APQNs of the queues
>> assigned to the matrix mdev. The guest_matrix attribute merely displays
>> a subset of the matrix (i.e., the APQNs assigned to the mdev that reference
>> queue devices bound to the vfio_ap device driver).
>>
>> How can this be restricted?
> I was thinking of using e.g. 400 instead of 444 for the permissions.
>
> I'm not sure if the info about what subset of the queues the guest
> actually uses is all that interesting (except for managing guests); but
> if I see guest information being exposed, I get a little wary, so I
> just stumbled over this.
>
> Maybe I'll come back to that once I have looked at the series in more
> detail, but this might not be a problem at all.

I created this patch primarily because I found it very helpful when
testing. It saved me from logging on to the guest and executing
lszcrypt to verify the AP configuration for the guest. I thought this
might also add value for a system administrator responsible for
KVM guests. It would not be a problem to remove this patch or
change the permissions for the attribute. Let me know when you
complete the series review. Thanks.

>
>>>   
>>>> +
>>>>    static struct attribute *vfio_ap_mdev_attrs[] = {
>>>>    	&dev_attr_assign_adapter.attr,
>>>>    	&dev_attr_unassign_adapter.attr,
>>>> @@ -1050,6 +1107,7 @@ static struct attribute *vfio_ap_mdev_attrs[] = {
>>>>    	&dev_attr_unassign_control_domain.attr,
>>>>    	&dev_attr_control_domains.attr,
>>>>    	&dev_attr_matrix.attr,
>>>> +	&dev_attr_guest_matrix.attr,
>>>>    	NULL,
>>>>    };
>>>>