Date:   Tue, 8 Oct 2019 14:57:07 +0200
From:   Pierre Morel <pmorel@...ux.ibm.com>
To:     Halil Pasic <pasic@...ux.ibm.com>,
        Tony Krowiak <akrowiak@...ux.ibm.com>
Cc:     linux-s390@...r.kernel.org, linux-kernel@...r.kernel.org,
        kvm@...r.kernel.org, freude@...ux.ibm.com, borntraeger@...ibm.com,
        cohuck@...hat.com, mjrosato@...ux.ibm.com,
        alex.williamson@...hat.com, kwankhede@...dia.com,
        jjherne@...ux.ibm.com
Subject: Re: [PATCH v6 00/10] s390: vfio-ap: dynamic configuration support


On 10/8/19 12:48 PM, Halil Pasic wrote:
> On Fri, 13 Sep 2019 17:26:48 -0400
> Tony Krowiak <akrowiak@...ux.ibm.com> wrote:
>
>> The current design for AP pass-through does not support making dynamic
>> changes to the AP matrix of a running guest resulting in three deficiencies
>> this patch series is intended to mitigate:
>>
>> 1. Adapters, domains and control domains can not be added to or removed
>>     from a running guest. In order to modify a guest's AP configuration,
>>     the guest must be terminated; only then can AP resources be assigned
>>     to or unassigned from the guest's matrix mdev. The new AP configuration
>>     becomes available to the guest when it is subsequently restarted.
>>
>> 2. The AP bus's /sys/bus/ap/apmask and /sys/bus/ap/aqmask interfaces can
>>     be modified by a root user without any restrictions. A change to either
>>     mask can result in AP queue devices being unbound from the vfio_ap
>>     device driver and bound to a zcrypt device driver even if a guest is
>>     using the queues, thus giving the host access to the guest's private
>>     crypto data and vice versa.
>>
>> 3. The APQNs derived from the Cartesian product of the APIDs of the
>>     adapters and APQIs of the domains assigned to a matrix mdev must
>>     reference an AP queue device bound to the vfio_ap device driver.
>>
>> This patch series introduces the following changes to the current design
>> to alleviate the shortcomings described above as well as to implement more
>> of the AP architecture:
>>
>> 1. A root user will be prevented from making changes to the AP bus's
>>     /sys/bus/ap/apmask or /sys/bus/ap/aqmask if the ownership of an APQN
>>     changes from the vfio_ap device driver to a zcrypt driver when the APQN
>>     is assigned to a matrix mdev.
>>
>> 2. The sysfs bind/unbind interfaces will be disabled for the vfio_ap
>>     device driver.
>>
> Doesn't this have the potential of breaking some userspace stuff that
> might be out there?
>
>> 3. Allow AP resources to be assigned to or removed from a matrix mdev
>>     while a guest is using it and hot plug the resource into or hot unplug
>>     the resource from the running guest.
> This looks like a natural extension of the interface -- i.e. should not
> break any userspace.
>
>> 4. Allow assignment of an AP adapter or domain to a matrix mdev even if it
>>     results in assignment of an APQN that does not reference an AP queue
>>     device bound to the vfio_ap device driver, as long as the APQN is owned
>>     by the vfio_ap driver. Allowing over-provisioning of AP resources
>>     better models the architecture which does not preclude assigning AP
>>     resources that are not yet available in the system. If/when the queue
>>     becomes available to the host, it will immediately also become available
>>     to the guest.
> Same here -- I don't think this change breaks any userspace.
>
>> 1. Rationale for changes to AP bus's apmask/aqmask interfaces:
>> ----------------------------------------------------------
>> Due to the extremely sensitive nature of cryptographic data, it is
>> imperative that great care be taken to ensure that such data is secured.
>> Allowing a root user, either inadvertently or maliciously, to configure
>> these masks such that a queue is shared between the host and a guest is
>> not only avoidable, it is advisable.

Just curious: how is it possible to do such a configuration?


>>   It was suggested that this scenario
>> is better handled in user space with management software, but that does
>> not preclude a malicious administrator from using the sysfs interfaces
>> to gain access to a guest's crypto data. It was also suggested that this
>> scenario could be avoided by taking access to the adapter away from the
>> guest and zeroing out the queues prior to the vfio_ap driver releasing the
>> device; however, stealing an adapter in use from a guest as a by-product
>> of an operation is bad and will likely cause problems for the guest
>> unnecessarily. It was decided that the most effective solution with the
>> least number of negative side effects is to prevent the situation at the
>> source.


Stealing an adapter in use by a guest, insn't it what is done if we 
allow to unassign an AP/Domain using the unassign sysfs interface when 
the mediated device is in use by the guest?


>>
>> 2. Rationale for disabling bind/unbind interfaces for vfio_ap driver:
>> -----------------------------------------------------------------
>> By disabling the bind/unbind interfaces for the vfio_ap device driver,
>> the user is forced to use the AP bus's apmask/aqmask interfaces to control
>> the probing and removing of AP queues. There are two primary reasons for
>> disabling the bind/unbind interfaces for the vfio_ap device driver:
>>
>> * The device architecture does not provide a means to prevent unbinding
>>    a device from a device driver, so an AP queue device can be unbound
>>    from the vfio_ap driver even when the queue is in use by a guest. By
>>    disabling the unbind interface, the user is forced to use the AP bus's
>>    apmask/aqmask interfaces which will prevent this.
>>
> Isn't this fixed by your filtering (if implemented correctly)? BTW I believe
> it solves the problem regardless whether the unbind was triggered by the
> drivers unbind attribute or by a[pq]mask
>
>> * Binding of AP queues is controlled by the AP bus /sys/bus/ap/apmask and
>>    /sys/bus/ap/aqmask interfaces. If the masks indicate that an APQN is
>>    owned by zcrypt, trying to bind it to the vfio_ap device driver will
>>    fail; therefore, the bind interface is somewhat redundant and certainly
>>    unnecessary.
>>    
>>
>>
>> Tony Krowiak (10):
>>    s390: vfio-ap: Refactor vfio_ap driver probe and remove callbacks
>>    s390: vfio-ap: allow assignment of unavailable AP resources to mdev
>>      device
>>    s390: vfio-ap: allow hot plug/unplug of AP resources using mdev device
>>    s390: vfio-ap: filter CRYCB bits for unavailable queue devices
>>    s390: vfio-ap: sysfs attribute to display the guest CRYCB
>>    s390: vfio-ap: update guest CRYCB in vfio_ap probe and remove
>>      callbacks
>>    s390: zcrypt: driver callback to indicate resource in use
>>    s390: vfio-ap: implement in-use callback for vfio_ap driver
>>    s390: vfio-ap: added versioning to vfio_ap module
>>    s390: vfio-ap: update documentation
> I believe it would be worthwhile to reorder the patches (fixes and
> re-factoring first, the features).
>
> Regards,
> Halil
>
>>   Documentation/s390/vfio-ap.rst        | 899 +++++++++++++++++++++++++---------
>>   drivers/s390/crypto/ap_bus.c          | 144 +++++-
>>   drivers/s390/crypto/ap_bus.h          |   4 +
>>   drivers/s390/crypto/vfio_ap_drv.c     |  27 +-
>>   drivers/s390/crypto/vfio_ap_ops.c     | 610 ++++++++++++++---------
>>   drivers/s390/crypto/vfio_ap_private.h |  12 +-
>>   6 files changed, 1200 insertions(+), 496 deletions(-)
>>
-- 
Pierre Morel
IBM Lab Boeblingen

Powered by blists - more mailing lists