The VFIO AP device driver¶

The VFIO AP (vfio_ap) device driver serves the following purposes:

1. Provides the interfaces to secure APQNs for exclusive use of KVM guests.
2. Sets up the VFIO mediated device interfaces to manage a mediated matrixdevice and creates the sysfs interfaces for assigning adapters, usagedomains, and control domains comprising the matrix for a KVM guest.
3. Configures the APM, AQM and ADM in the CRYCB referenced by a KVM guest’sSIE state description to grant the guest access to a matrix of AP devices

Reserve APQNs for exclusive use of KVM guests¶

The following block diagram illustrates the mechanism by which APQNs arereserved:

                              +------------------+               7 remove       |                  |         +--------------------> cex4queue driver |         |                    |                  |         |                    +------------------+         |         |         |                    +------------------+          +----------------+         |  5 register driver |                  | 3 create |                |         |   +---------------->   Device core    +---------->  matrix device |         |   |                |                  |          |                |         |   |                +--------^---------+          +----------------+         |   |                         |         |   |                         +-------------------+         |   | +-----------------------------------+       |         |   | |      4 register AP driver         |       | 2 register device         |   | |                                   |       |+--------+---+-v---+                      +--------+-------+-+|                  |                      |                  ||      ap_bus      +--------------------- >  vfio_ap driver  ||                  |       8 probe        |                  |+--------^---------+                      +--^--^------------+6 edit   |                                   |  |  apmask |     +-----------------------------+  | 9 mdev create  aqmask |     |           1 modprobe           |+--------+-----+---+           +----------------+-+         +----------------+|                  |           |                  |8 create |     mediated   ||      admin       |           | VFIO device core |--------->     matrix     ||                  +           |                  |         |     device     |+------+-+---------+           +--------^---------+         +--------^-------+       | |                              |                            |       | | 9 create vfio_ap-passthrough |                            |       | +------------------------------+                            |       +-------------------------------------------------------------+                   10  assign adapter/domain/control domain

The process for reserving an AP queue for use by a KVM guest is:

1. The administrator loads the vfio_ap device driver
2. The vfio-ap driver during its initialization will register a single ‘matrix’device with the device core. This will serve as the parent device forall mediated matrix devices used to configure an AP matrix for a guest.
3. The /sys/devices/vfio_ap/matrix device is created by the device core
4. The vfio_ap device driver will register with the AP bus for AP queue devicesof type 10 and higher (CEX4 and newer). The driver will provide the vfio_apdriver’s probe and remove callback interfaces. Devices older than CEX4 queuesare not supported to simplify the implementation by not needlesslycomplicating the design by supporting older devices that will go out ofservice in the relatively near future, and for which there are few oldersystems around on which to test.
5. The AP bus registers the vfio_ap device driver with the device core
6. The administrator edits the AP adapter and queue masks to reserve AP queuesfor use by the vfio_ap device driver.
7. The AP bus removes the AP queues reserved for the vfio_ap driver from thedefault zcrypt cex4queue driver.
8. The AP bus probes the vfio_ap device driver to bind the queues reserved forit.
9. The administrator creates a passthrough type mediated matrix device to beused by a guest
10. The administrator assigns the adapters, usage domains and control domainsto be exclusively used by a guest.

Set up the VFIO mediated device interfaces¶

The VFIO AP device driver utilizes the common interface of the VFIO mediateddevice core driver to:

• Register an AP mediated bus driver to add a mediated matrix device to andremove it from a VFIO group.
• Create and destroy a mediated matrix device
• Add a mediated matrix device to and remove it from the AP mediated bus driver
• Add a mediated matrix device to and remove it from an IOMMU group

The following high-level block diagram shows the main components and interfacesof the VFIO AP mediated matrix device driver:

+-------------+|             || +---------+ | mdev_register_driver() +--------------+| |  Mdev   | +<-----------------------+              || |  bus    | |                        | vfio_mdev.ko || | driver  | +----------------------->+              |<-> VFIO user| +---------+ |    probe()/remove()    +--------------+    APIs|             ||  MDEV CORE  ||   MODULE    ||   mdev.ko   || +---------+ | mdev_register_device() +--------------+| |Physical | +<-----------------------+              || | device  | |                        |  vfio_ap.ko  |<-> matrix| |interface| +----------------------->+              |    device| +---------+ |       callback         +--------------++-------------+

During initialization of the vfio_ap module, the matrix device is registeredwith an ‘mdev_parent_ops’ structure that provides the sysfs attributestructures, mdev functions and callback interfaces for managing the mediatedmatrix device.

• sysfs attribute structures:

  supported_type_groups

  The VFIO mediated device framework supports creation of user-definedmediated device types. These mediated device types are specifiedvia the ‘supported_type_groups’ structure when a device is registeredwith the mediated device framework. The registration process creates thesysfs structures for each mediated device type specified in the‘mdev_supported_types’ sub-directory of the device being registered. Alongwith the device type, the sysfs attributes of the mediated device type areprovided.

  The VFIO AP device driver will register one mediated device type forpassthrough devices:

  /sys/devices/vfio_ap/matrix/mdev_supported_types/vfio_ap-passthrough

  Only the read-only attributes required by the VFIO mdev framework willbe provided:

  ... name... device_api... available_instances... device_api

  Where:

  • name:

    specifies the name of the mediated device type

  • device_api:

    the mediated device type’s API

  • available_instances:

    the number of mediated matrix passthrough devicesthat can be created

  • device_api:

    specifies the VFIO API

  mdev_attr_groups

  This attribute group identifies the user-defined sysfs attributes of themediated device. When a device is registered with the VFIO mediated deviceframework, the sysfs attribute files identified in the ‘mdev_attr_groups’structure will be created in the mediated matrix device’s directory. Thesysfs attributes for a mediated matrix device are:

  assign_adapter / unassign_adapter:

  Write-only attributes for assigning/unassigning an AP adapter to/from themediated matrix device. To assign/unassign an adapter, the APID of theadapter is echoed to the respective attribute file.

  assign_domain / unassign_domain:

  Write-only attributes for assigning/unassigning an AP usage domain to/fromthe mediated matrix device. To assign/unassign a domain, the domainnumber of the usage domain is echoed to the respective attributefile.

  matrix:

  A read-only file for displaying the APQNs derived from the cross productof the adapter and domain numbers assigned to the mediated matrix device.

  assign_control_domain / unassign_control_domain:

  Write-only attributes for assigning/unassigning an AP control domainto/from the mediated matrix device. To assign/unassign a control domain,the ID of the domain to be assigned/unassigned is echoed to the respectiveattribute file.

  control_domains:

  A read-only file for displaying the control domain numbers assigned to themediated matrix device.

• functions:

  create:

  allocates the ap_matrix_mdev structure used by the vfio_ap driver to:

  • Store the reference to the KVM structure for the guest using the mdev
  • Store the AP matrix configuration for the adapters, domains, and controldomains assigned via the corresponding sysfs attributes files

  remove:

  deallocates the mediated matrix device’s ap_matrix_mdev structure. This willbe allowed only if a running guest is not using the mdev.

• callback interfaces

  open:

  The vfio_ap driver uses this callback to register aVFIO_GROUP_NOTIFY_SET_KVM notifier callback function for the mdev matrixdevice. The open is invoked when QEMU connects the VFIO iommu groupfor the mdev matrix device to the MDEV bus. Access to the KVM structure usedto configure the KVM guest is provided via this callback. The KVM structure,is used to configure the guest’s access to the AP matrix defined via themediated matrix device’s sysfs attribute files.

  release:

  unregisters the VFIO_GROUP_NOTIFY_SET_KVM notifier callback function for themdev matrix device and deconfigures the guest’s AP matrix.

Configure the APM, AQM and ADM in the CRYCB¶

Configuring the AP matrix for a KVM guest will be performed when theVFIO_GROUP_NOTIFY_SET_KVM notifier callback is invoked. The notifierfunction is called when QEMU connects to KVM. The guest’s AP matrix isconfigured via it’s CRYCB by:

• Setting the bits in the APM corresponding to the APIDs assigned to themediated matrix device via its ‘assign_adapter’ interface.
• Setting the bits in the AQM corresponding to the domains assigned to themediated matrix device via its ‘assign_domain’ interface.
• Setting the bits in the ADM corresponding to the domain dIDs assigned to themediated matrix device via its ‘assign_control_domains’ interface.

The CPU model features for AP¶

The AP stack relies on the presence of the AP instructions as well as twofacilities: The AP Facilities Test (APFT) facility; and the AP QueryConfiguration Information (QCI) facility. These features/facilities are madeavailable to a KVM guest via the following CPU model features:

1. ap: Indicates whether the AP instructions are installed on the guest. Thisfeature will be enabled by KVM only if the AP instructions are installedon the host.
2. apft: Indicates the APFT facility is available on the guest. This facilitycan be made available to the guest only if it is available on the host (i.e.,facility bit 15 is set).
3. apqci: Indicates the AP QCI facility is available on the guest. This facilitycan be made available to the guest only if it is available on the host (i.e.,facility bit 12 is set).

Note: If the user chooses to specify a CPU model different than the ‘host’model to QEMU, the CPU model features and facilities need to be turned onexplicitly; for example:

/usr/bin/qemu-system-s390x ... -cpu z13,ap=on,apqci=on,apft=on

A guest can be precluded from using AP features/facilities by turning them offexplicitly; for example:

/usr/bin/qemu-system-s390x ... -cpu host,ap=off,apqci=off,apft=off

Note: If the APFT facility is turned off (apft=off) for the guest, the guestwill not see any AP devices. The zcrypt device drivers that register for type 10and newer AP devices - i.e., the cex4card and cex4queue device drivers - needthe APFT facility to ascertain the facilities installed on a given AP device. Ifthe APFT facility is not installed on the guest, then the probe of devicedrivers will fail since only type 10 and newer devices can be configured forguest use.

Guest1¶

Guest2¶

Guest3¶

These are the steps:

1. Install the vfio_ap module on the linux host. The dependency chain for thevfio_ap module is:* iommu* s390* zcrypt* vfio* vfio_mdev* vfio_mdev_device* KVM

   To build the vfio_ap module, the kernel build must be configured with thefollowing Kconfig elements selected:* IOMMU_SUPPORT* S390* ZCRYPT* S390_AP_IOMMU* VFIO* VFIO_MDEV* VFIO_MDEV_DEVICE* KVM

   If using make menuconfig select the following to build the vfio_ap module:

   -> Device Drivers   -> IOMMU Hardware Support      select S390 AP IOMMU Support   -> VFIO Non-Privileged userspace driver framework      -> Mediated device driver frramework         -> VFIO driver for Mediated devices-> I/O subsystem   -> VFIO support for AP devices

2. Secure the AP queues to be used by the three guests so that the host can notaccess them. To secure them, there are two sysfs files that specifybitmasks marking a subset of the APQN range as ‘usable by the default APqueue device drivers’ or ‘not usable by the default device drivers’ and thusavailable for use by the vfio_ap device driver’. The location of the sysfsfiles containing the masks are:

   /sys/bus/ap/apmask/sys/bus/ap/aqmask

   The ‘apmask’ is a 256-bit mask that identifies a set of AP adapter IDs(APID). Each bit in the mask, from left to right (i.e., from most significantto least significant bit in big endian order), corresponds to an APID from0-255. If a bit is set, the APID is marked as usable only by the default APqueue device drivers; otherwise, the APID is usable by the vfio_apdevice driver.

   The ‘aqmask’ is a 256-bit mask that identifies a set of AP queue indexes(APQI). Each bit in the mask, from left to right (i.e., from most significantto least significant bit in big endian order), corresponds to an APQI from0-255. If a bit is set, the APQI is marked as usable only by the default APqueue device drivers; otherwise, the APQI is usable by the vfio_ap devicedriver.

   Take, for example, the following mask:

     0x7dffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffIt indicates:  1, 2, 3, 4, 5, and 7-255 belong to the default drivers' pool, and 0 and 6  belong to the vfio_ap device driver's pool.

   The APQN of each AP queue device assigned to the linux host is checked by theAP bus against the set of APQNs derived from the cross product of APIDsand APQIs marked as usable only by the default AP queue device drivers. If amatch is detected, only the default AP queue device drivers will be probed;otherwise, the vfio_ap device driver will be probed.

   By default, the two masks are set to reserve all APQNs for use by the defaultAP queue device drivers. There are two ways the default masks can be changed:

   1. The sysfs mask files can be edited by echoing a string into therespective sysfs mask file in one of two formats:

      • An absolute hex string starting with 0x - like “0x12345678” - setsthe mask. If the given string is shorter than the mask, it is paddedwith 0s on the right; for example, specifying a mask value of 0x41 isthe same as specifying:

        0x4100000000000000000000000000000000000000000000000000000000000000

        Keep in mind that the mask reads from left to right (i.e., mostsignificant to least significant bit in big endian order), so the maskabove identifies device numbers 1 and 7 (01000001).

        If the string is longer than the mask, the operation is terminated withan error (EINVAL).

      • Individual bits in the mask can be switched on and off by specifyingeach bit number to be switched in a comma separated list. Each bitnumber string must be prepended with a (‘+’) or minus (‘-‘) to indicatethe corresponding bit is to be switched on (‘+’) or off (‘-‘). Somevalid values are:

        • “+0” switches bit 0 on
        • “-13” switches bit 13 off
        • “+0x41” switches bit 65 on
        • “-0xff” switches bit 255 off

        The following example:

        +0,-6,+0x47,-0xf0

        Switches bits 0 and 71 (0x47) on

        Switches bits 6 and 240 (0xf0) off

        Note that the bits not specified in the list remain as they were beforethe operation.

   2. The masks can also be changed at boot time via parameters on the kernelcommand line like this:

      ap.apmask=0xffff ap.aqmask=0x40

      This would create the following masks:

      apmask:0xffff000000000000000000000000000000000000000000000000000000000000aqmask:0x4000000000000000000000000000000000000000000000000000000000000000

      Resulting in these two pools:

      default drivers pool:    adapter 0-15, domain 1alternate drivers pool:  adapter 16-255, domains 0, 2-255

Securing the APQNs for our example¶

To secure the AP queues 05.0004, 05.0047, 05.00ab, 05.00ff, 06.0004, 06.0047,06.00ab, and 06.00ff for use by the vfio_ap device driver, the correspondingAPQNs can either be removed from the default masks:

echo -5,-6 > /sys/bus/ap/apmaskecho -4,-0x47,-0xab,-0xff > /sys/bus/ap/aqmask

Or the masks can be set as follows:

echo 0xf9ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff \> apmaskecho 0xf7fffffffffffffffeffffffffffffffffffffffffeffffffffffffffffffffe \> aqmask

This will result in AP queues 05.0004, 05.0047, 05.00ab, 05.00ff, 06.0004,06.0047, 06.00ab, and 06.00ff getting bound to the vfio_ap device driver. Thesysfs directory for the vfio_ap device driver will now contain symbolic linksto the AP queue devices bound to it:

/sys/bus/ap... [drivers]...... [vfio_ap]......... [05.0004]......... [05.0047]......... [05.00ab]......... [05.00ff]......... [06.0004]......... [06.0047]......... [06.00ab]......... [06.00ff]

Keep in mind that only type 10 and newer adapters (i.e., CEX4 and later)can be bound to the vfio_ap device driver. The reason for this is tosimplify the implementation by not needlessly complicating the design bysupporting older devices that will go out of service in the relatively nearfuture and for which there are few older systems on which to test.

The administrator, therefore, must take care to secure only AP queues thatcan be bound to the vfio_ap device driver. The device type for a given APqueue device can be read from the parent card’s sysfs directory. For example,to see the hardware type of the queue 05.0004:

cat /sys/bus/ap/devices/card05/hwtype

The hwtype must be 10 or higher (CEX4 or newer) in order to be bound to thevfio_ap device driver.

3. Create the mediated devices needed to configure the AP matrixes for thethree guests and to provide an interface to the vfio_ap driver foruse by the guests:

   /sys/devices/vfio_ap/matrix/--- [mdev_supported_types]------ [vfio_ap-passthrough] (passthrough mediated matrix device type)--------- create--------- [devices]

   To create the mediated devices for the three guests:

   uuidgen > createuuidgen > createuuidgen > createorecho $uuid1 > createecho $uuid2 > createecho $uuid3 > create

   This will create three mediated devices in the [devices] subdirectory namedafter the UUID written to the create attribute file. We call them $uuid1,$uuid2 and $uuid3 and this is the sysfs directory structure after creation:

   /sys/devices/vfio_ap/matrix/--- [mdev_supported_types]------ [vfio_ap-passthrough]--------- [devices]------------ [$uuid1]--------------- assign_adapter--------------- assign_control_domain--------------- assign_domain--------------- matrix--------------- unassign_adapter--------------- unassign_control_domain--------------- unassign_domain------------ [$uuid2]--------------- assign_adapter--------------- assign_control_domain--------------- assign_domain--------------- matrix--------------- unassign_adapter----------------unassign_control_domain----------------unassign_domain------------ [$uuid3]--------------- assign_adapter--------------- assign_control_domain--------------- assign_domain--------------- matrix--------------- unassign_adapter----------------unassign_control_domain----------------unassign_domain

4. The administrator now needs to configure the matrixes for the mediateddevices $uuid1 (for Guest1), $uuid2 (for Guest2) and $uuid3 (for Guest3).

   This is how the matrix is configured for Guest1:

   echo 5 > assign_adapterecho 6 > assign_adapterecho 4 > assign_domainecho 0xab > assign_domain

   Control domains can similarly be assigned using the assign_control_domainsysfs file.

   If a mistake is made configuring an adapter, domain or control domain,you can use the unassign_xxx files to unassign the adapter, domain orcontrol domain.

   To display the matrix configuration for Guest1:

   cat matrix

   This is how the matrix is configured for Guest2:

   echo 5 > assign_adapterecho 0x47 > assign_domainecho 0xff > assign_domain

   This is how the matrix is configured for Guest3:

   echo 6 > assign_adapterecho 0x47 > assign_domainecho 0xff > assign_domain

   In order to successfully assign an adapter:

   • The adapter number specified must represent a value from 0 up to themaximum adapter number configured for the system. If an adapter numberhigher than the maximum is specified, the operation will terminate withan error (ENODEV).

   • All APQNs that can be derived from the adapter ID and the IDs ofthe previously assigned domains must be bound to the vfio_ap devicedriver. If no domains have yet been assigned, then there must be at leastone APQN with the specified APID bound to the vfio_ap driver. If no suchAPQNs are bound to the driver, the operation will terminate with anerror (EADDRNOTAVAIL).

     No APQN that can be derived from the adapter ID and the IDs of thepreviously assigned domains can be assigned to another mediated matrixdevice. If an APQN is assigned to another mediated matrix device, theoperation will terminate with an error (EADDRINUSE).

   In order to successfully assign a domain:

   • The domain number specified must represent a value from 0 up to themaximum domain number configured for the system. If a domain numberhigher than the maximum is specified, the operation will terminate withan error (ENODEV).

   • All APQNs that can be derived from the domain ID and the IDs ofthe previously assigned adapters must be bound to the vfio_ap devicedriver. If no domains have yet been assigned, then there must be at leastone APQN with the specified APQI bound to the vfio_ap driver. If no suchAPQNs are bound to the driver, the operation will terminate with anerror (EADDRNOTAVAIL).

     No APQN that can be derived from the domain ID and the IDs of thepreviously assigned adapters can be assigned to another mediated matrixdevice. If an APQN is assigned to another mediated matrix device, theoperation will terminate with an error (EADDRINUSE).

   In order to successfully assign a control domain, the domain numberspecified must represent a value from 0 up to the maximum domain numberconfigured for the system. If a control domain number higher than the maximumis specified, the operation will terminate with an error (ENODEV).

5. Start Guest1:

   /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \   -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid1 ...

7. Start Guest2:

   /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \   -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid2 ...

7. Start Guest3:

   /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \   -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid3 ...

When the guest is shut down, the mediated matrix devices may be removed.

Using our example again, to remove the mediated matrix device $uuid1:

/sys/devices/vfio_ap/matrix/   --- [mdev_supported_types]   ------ [vfio_ap-passthrough]   --------- [devices]   ------------ [$uuid1]   --------------- remove

echo 1 > remove

This will remove all of the mdev matrix device’s sysfs structures includingthe mdev device itself. To recreate and reconfigure the mdev matrix device,all of the steps starting with step 3 will have to be performed again. Notethat the remove will fail if a guest using the mdev is still running.

It is not necessary to remove an mdev matrix device, but one may want toremove it if no guest will use it during the remaining lifetime of the linuxhost. If the mdev matrix device is removed, one may want to also reconfigurethe pool of adapters and queues reserved for use by the default drivers.