1=============================== 2Adjunct Processor (AP) facility 3=============================== 4 5 6Introduction 7============ 8The Adjunct Processor (AP) facility is an IBM Z cryptographic facility comprised 9of three AP instructions and from 1 up to 256 PCIe cryptographic adapter cards. 10The AP devices provide cryptographic functions to all CPUs assigned to a 11linux system running in an IBM Z system LPAR. 12 13The AP adapter cards are exposed via the AP bus. The motivation for vfio-ap 14is to make AP cards available to KVM guests using the VFIO mediated device 15framework. This implementation relies considerably on the s390 virtualization 16facilities which do most of the hard work of providing direct access to AP 17devices. 18 19AP Architectural Overview 20========================= 21To facilitate the comprehension of the design, let's start with some 22definitions: 23 24* AP adapter 25 26 An AP adapter is an IBM Z adapter card that can perform cryptographic 27 functions. There can be from 0 to 256 adapters assigned to an LPAR. Adapters 28 assigned to the LPAR in which a linux host is running will be available to 29 the linux host. Each adapter is identified by a number from 0 to 255; however, 30 the maximum adapter number is determined by machine model and/or adapter type. 31 When installed, an AP adapter is accessed by AP instructions executed by any 32 CPU. 33 34 The AP adapter cards are assigned to a given LPAR via the system's Activation 35 Profile which can be edited via the HMC. When the linux host system is IPL'd 36 in the LPAR, the AP bus detects the AP adapter cards assigned to the LPAR and 37 creates a sysfs device for each assigned adapter. For example, if AP adapters 38 4 and 10 (0x0a) are assigned to the LPAR, the AP bus will create the following 39 sysfs device entries:: 40 41 /sys/devices/ap/card04 42 /sys/devices/ap/card0a 43 44 Symbolic links to these devices will also be created in the AP bus devices 45 sub-directory:: 46 47 /sys/bus/ap/devices/[card04] 48 /sys/bus/ap/devices/[card04] 49 50* AP domain 51 52 An adapter is partitioned into domains. An adapter can hold up to 256 domains 53 depending upon the adapter type and hardware configuration. A domain is 54 identified by a number from 0 to 255; however, the maximum domain number is 55 determined by machine model and/or adapter type.. A domain can be thought of 56 as a set of hardware registers and memory used for processing AP commands. A 57 domain can be configured with a secure private key used for clear key 58 encryption. A domain is classified in one of two ways depending upon how it 59 may be accessed: 60 61 * Usage domains are domains that are targeted by an AP instruction to 62 process an AP command. 63 64 * Control domains are domains that are changed by an AP command sent to a 65 usage domain; for example, to set the secure private key for the control 66 domain. 67 68 The AP usage and control domains are assigned to a given LPAR via the system's 69 Activation Profile which can be edited via the HMC. When a linux host system 70 is IPL'd in the LPAR, the AP bus module detects the AP usage and control 71 domains assigned to the LPAR. The domain number of each usage domain and 72 adapter number of each AP adapter are combined to create AP queue devices 73 (see AP Queue section below). The domain number of each control domain will be 74 represented in a bitmask and stored in a sysfs file 75 /sys/bus/ap/ap_control_domain_mask. The bits in the mask, from most to least 76 significant bit, correspond to domains 0-255. 77 78* AP Queue 79 80 An AP queue is the means by which an AP command is sent to a usage domain 81 inside a specific adapter. An AP queue is identified by a tuple 82 comprised of an AP adapter ID (APID) and an AP queue index (APQI). The 83 APQI corresponds to a given usage domain number within the adapter. This tuple 84 forms an AP Queue Number (APQN) uniquely identifying an AP queue. AP 85 instructions include a field containing the APQN to identify the AP queue to 86 which the AP command is to be sent for processing. 87 88 The AP bus will create a sysfs device for each APQN that can be derived from 89 the cross product of the AP adapter and usage domain numbers detected when the 90 AP bus module is loaded. For example, if adapters 4 and 10 (0x0a) and usage 91 domains 6 and 71 (0x47) are assigned to the LPAR, the AP bus will create the 92 following sysfs entries:: 93 94 /sys/devices/ap/card04/04.0006 95 /sys/devices/ap/card04/04.0047 96 /sys/devices/ap/card0a/0a.0006 97 /sys/devices/ap/card0a/0a.0047 98 99 The following symbolic links to these devices will be created in the AP bus 100 devices subdirectory:: 101 102 /sys/bus/ap/devices/[04.0006] 103 /sys/bus/ap/devices/[04.0047] 104 /sys/bus/ap/devices/[0a.0006] 105 /sys/bus/ap/devices/[0a.0047] 106 107* AP Instructions: 108 109 There are three AP instructions: 110 111 * NQAP: to enqueue an AP command-request message to a queue 112 * DQAP: to dequeue an AP command-reply message from a queue 113 * PQAP: to administer the queues 114 115 AP instructions identify the domain that is targeted to process the AP 116 command; this must be one of the usage domains. An AP command may modify a 117 domain that is not one of the usage domains, but the modified domain 118 must be one of the control domains. 119 120AP and SIE 121========== 122Let's now take a look at how AP instructions executed on a guest are interpreted 123by the hardware. 124 125A satellite control block called the Crypto Control Block (CRYCB) is attached to 126our main hardware virtualization control block. The CRYCB contains three fields 127to identify the adapters, usage domains and control domains assigned to the KVM 128guest: 129 130* The AP Mask (APM) field is a bit mask that identifies the AP adapters assigned 131 to the KVM guest. Each bit in the mask, from left to right (i.e. from most 132 significant to least significant bit in big endian order), corresponds to 133 an APID from 0-255. If a bit is set, the corresponding adapter is valid for 134 use by the KVM guest. 135 136* The AP Queue Mask (AQM) field is a bit mask identifying the AP usage domains 137 assigned to the KVM guest. Each bit in the mask, from left to right (i.e. from 138 most significant to least significant bit in big endian order), corresponds to 139 an AP queue index (APQI) from 0-255. If a bit is set, the corresponding queue 140 is valid for use by the KVM guest. 141 142* The AP Domain Mask field is a bit mask that identifies the AP control domains 143 assigned to the KVM guest. The ADM bit mask controls which domains can be 144 changed by an AP command-request message sent to a usage domain from the 145 guest. Each bit in the mask, from left to right (i.e. from most significant to 146 least significant bit in big endian order), corresponds to a domain from 147 0-255. If a bit is set, the corresponding domain can be modified by an AP 148 command-request message sent to a usage domain. 149 150If you recall from the description of an AP Queue, AP instructions include 151an APQN to identify the AP queue to which an AP command-request message is to be 152sent (NQAP and PQAP instructions), or from which a command-reply message is to 153be received (DQAP instruction). The validity of an APQN is defined by the matrix 154calculated from the APM and AQM; it is the cross product of all assigned adapter 155numbers (APM) with all assigned queue indexes (AQM). For example, if adapters 1 156and 2 and usage domains 5 and 6 are assigned to a guest, the APQNs (1,5), (1,6), 157(2,5) and (2,6) will be valid for the guest. 158 159The APQNs can provide secure key functionality - i.e., a private key is stored 160on the adapter card for each of its domains - so each APQN must be assigned to 161at most one guest or to the linux host:: 162 163 Example 1: Valid configuration: 164 ------------------------------ 165 Guest1: adapters 1,2 domains 5,6 166 Guest2: adapter 1,2 domain 7 167 168 This is valid because both guests have a unique set of APQNs: 169 Guest1 has APQNs (1,5), (1,6), (2,5), (2,6); 170 Guest2 has APQNs (1,7), (2,7) 171 172 Example 2: Valid configuration: 173 ------------------------------ 174 Guest1: adapters 1,2 domains 5,6 175 Guest2: adapters 3,4 domains 5,6 176 177 This is also valid because both guests have a unique set of APQNs: 178 Guest1 has APQNs (1,5), (1,6), (2,5), (2,6); 179 Guest2 has APQNs (3,5), (3,6), (4,5), (4,6) 180 181 Example 3: Invalid configuration: 182 -------------------------------- 183 Guest1: adapters 1,2 domains 5,6 184 Guest2: adapter 1 domains 6,7 185 186 This is an invalid configuration because both guests have access to 187 APQN (1,6). 188 189The Design 190========== 191The design introduces three new objects: 192 1931. AP matrix device 1942. VFIO AP device driver (vfio_ap.ko) 1953. VFIO AP mediated matrix pass-through device 196 197The VFIO AP device driver 198------------------------- 199The VFIO AP (vfio_ap) device driver serves the following purposes: 200 2011. Provides the interfaces to secure APQNs for exclusive use of KVM guests. 202 2032. Sets up the VFIO mediated device interfaces to manage a mediated matrix 204 device and creates the sysfs interfaces for assigning adapters, usage 205 domains, and control domains comprising the matrix for a KVM guest. 206 2073. Configures the APM, AQM and ADM in the CRYCB referenced by a KVM guest's 208 SIE state description to grant the guest access to a matrix of AP devices 209 210Reserve APQNs for exclusive use of KVM guests 211--------------------------------------------- 212The following block diagram illustrates the mechanism by which APQNs are 213reserved:: 214 215 +------------------+ 216 7 remove | | 217 +--------------------> cex4queue driver | 218 | | | 219 | +------------------+ 220 | 221 | 222 | +------------------+ +----------------+ 223 | 5 register driver | | 3 create | | 224 | +----------------> Device core +----------> matrix device | 225 | | | | | | 226 | | +--------^---------+ +----------------+ 227 | | | 228 | | +-------------------+ 229 | | +-----------------------------------+ | 230 | | | 4 register AP driver | | 2 register device 231 | | | | | 232 +--------+---+-v---+ +--------+-------+-+ 233 | | | | 234 | ap_bus +--------------------- > vfio_ap driver | 235 | | 8 probe | | 236 +--------^---------+ +--^--^------------+ 237 6 edit | | | 238 apmask | +-----------------------------+ | 9 mdev create 239 aqmask | | 1 modprobe | 240 +--------+-----+---+ +----------------+-+ +----------------+ 241 | | | |8 create | mediated | 242 | admin | | VFIO device core |---------> matrix | 243 | + | | | device | 244 +------+-+---------+ +--------^---------+ +--------^-------+ 245 | | | | 246 | | 9 create vfio_ap-passthrough | | 247 | +------------------------------+ | 248 +-------------------------------------------------------------+ 249 10 assign adapter/domain/control domain 250 251The process for reserving an AP queue for use by a KVM guest is: 252 2531. The administrator loads the vfio_ap device driver 2542. The vfio-ap driver during its initialization will register a single 'matrix' 255 device with the device core. This will serve as the parent device for 256 all mediated matrix devices used to configure an AP matrix for a guest. 2573. The /sys/devices/vfio_ap/matrix device is created by the device core 2584. The vfio_ap device driver will register with the AP bus for AP queue devices 259 of type 10 and higher (CEX4 and newer). The driver will provide the vfio_ap 260 driver's probe and remove callback interfaces. Devices older than CEX4 queues 261 are not supported to simplify the implementation by not needlessly 262 complicating the design by supporting older devices that will go out of 263 service in the relatively near future, and for which there are few older 264 systems around on which to test. 2655. The AP bus registers the vfio_ap device driver with the device core 2666. The administrator edits the AP adapter and queue masks to reserve AP queues 267 for use by the vfio_ap device driver. 2687. The AP bus removes the AP queues reserved for the vfio_ap driver from the 269 default zcrypt cex4queue driver. 2708. The AP bus probes the vfio_ap device driver to bind the queues reserved for 271 it. 2729. The administrator creates a passthrough type mediated matrix device to be 273 used by a guest 27410. The administrator assigns the adapters, usage domains and control domains 275 to be exclusively used by a guest. 276 277Set up the VFIO mediated device interfaces 278------------------------------------------ 279The VFIO AP device driver utilizes the common interface of the VFIO mediated 280device core driver to: 281 282* Register an AP mediated bus driver to add a mediated matrix device to and 283 remove it from a VFIO group. 284* Create and destroy a mediated matrix device 285* Add a mediated matrix device to and remove it from the AP mediated bus driver 286* Add a mediated matrix device to and remove it from an IOMMU group 287 288The following high-level block diagram shows the main components and interfaces 289of the VFIO AP mediated matrix device driver:: 290 291 +-------------+ 292 | | 293 | +---------+ | mdev_register_driver() +--------------+ 294 | | Mdev | +<-----------------------+ | 295 | | bus | | | vfio_mdev.ko | 296 | | driver | +----------------------->+ |<-> VFIO user 297 | +---------+ | probe()/remove() +--------------+ APIs 298 | | 299 | MDEV CORE | 300 | MODULE | 301 | mdev.ko | 302 | +---------+ | mdev_register_device() +--------------+ 303 | |Physical | +<-----------------------+ | 304 | | device | | | vfio_ap.ko |<-> matrix 305 | |interface| +----------------------->+ | device 306 | +---------+ | callback +--------------+ 307 +-------------+ 308 309During initialization of the vfio_ap module, the matrix device is registered 310with an 'mdev_parent_ops' structure that provides the sysfs attribute 311structures, mdev functions and callback interfaces for managing the mediated 312matrix device. 313 314* sysfs attribute structures: 315 316 supported_type_groups 317 The VFIO mediated device framework supports creation of user-defined 318 mediated device types. These mediated device types are specified 319 via the 'supported_type_groups' structure when a device is registered 320 with the mediated device framework. The registration process creates the 321 sysfs structures for each mediated device type specified in the 322 'mdev_supported_types' sub-directory of the device being registered. Along 323 with the device type, the sysfs attributes of the mediated device type are 324 provided. 325 326 The VFIO AP device driver will register one mediated device type for 327 passthrough devices: 328 329 /sys/devices/vfio_ap/matrix/mdev_supported_types/vfio_ap-passthrough 330 331 Only the read-only attributes required by the VFIO mdev framework will 332 be provided:: 333 334 ... name 335 ... device_api 336 ... available_instances 337 ... device_api 338 339 Where: 340 341 * name: 342 specifies the name of the mediated device type 343 * device_api: 344 the mediated device type's API 345 * available_instances: 346 the number of mediated matrix passthrough devices 347 that can be created 348 * device_api: 349 specifies the VFIO API 350 mdev_attr_groups 351 This attribute group identifies the user-defined sysfs attributes of the 352 mediated device. When a device is registered with the VFIO mediated device 353 framework, the sysfs attribute files identified in the 'mdev_attr_groups' 354 structure will be created in the mediated matrix device's directory. The 355 sysfs attributes for a mediated matrix device are: 356 357 assign_adapter / unassign_adapter: 358 Write-only attributes for assigning/unassigning an AP adapter to/from the 359 mediated matrix device. To assign/unassign an adapter, the APID of the 360 adapter is echoed to the respective attribute file. 361 assign_domain / unassign_domain: 362 Write-only attributes for assigning/unassigning an AP usage domain to/from 363 the mediated matrix device. To assign/unassign a domain, the domain 364 number of the usage domain is echoed to the respective attribute 365 file. 366 matrix: 367 A read-only file for displaying the APQNs derived from the cross product 368 of the adapter and domain numbers assigned to the mediated matrix device. 369 assign_control_domain / unassign_control_domain: 370 Write-only attributes for assigning/unassigning an AP control domain 371 to/from the mediated matrix device. To assign/unassign a control domain, 372 the ID of the domain to be assigned/unassigned is echoed to the respective 373 attribute file. 374 control_domains: 375 A read-only file for displaying the control domain numbers assigned to the 376 mediated matrix device. 377 378* functions: 379 380 create: 381 allocates the ap_matrix_mdev structure used by the vfio_ap driver to: 382 383 * Store the reference to the KVM structure for the guest using the mdev 384 * Store the AP matrix configuration for the adapters, domains, and control 385 domains assigned via the corresponding sysfs attributes files 386 387 remove: 388 deallocates the mediated matrix device's ap_matrix_mdev structure. This will 389 be allowed only if a running guest is not using the mdev. 390 391* callback interfaces 392 393 open: 394 The vfio_ap driver uses this callback to register a 395 VFIO_GROUP_NOTIFY_SET_KVM notifier callback function for the mdev matrix 396 device. The open is invoked when QEMU connects the VFIO iommu group 397 for the mdev matrix device to the MDEV bus. Access to the KVM structure used 398 to configure the KVM guest is provided via this callback. The KVM structure, 399 is used to configure the guest's access to the AP matrix defined via the 400 mediated matrix device's sysfs attribute files. 401 release: 402 unregisters the VFIO_GROUP_NOTIFY_SET_KVM notifier callback function for the 403 mdev matrix device and deconfigures the guest's AP matrix. 404 405Configure the APM, AQM and ADM in the CRYCB 406------------------------------------------- 407Configuring the AP matrix for a KVM guest will be performed when the 408VFIO_GROUP_NOTIFY_SET_KVM notifier callback is invoked. The notifier 409function is called when QEMU connects to KVM. The guest's AP matrix is 410configured via it's CRYCB by: 411 412* Setting the bits in the APM corresponding to the APIDs assigned to the 413 mediated matrix device via its 'assign_adapter' interface. 414* Setting the bits in the AQM corresponding to the domains assigned to the 415 mediated matrix device via its 'assign_domain' interface. 416* Setting the bits in the ADM corresponding to the domain dIDs assigned to the 417 mediated matrix device via its 'assign_control_domains' interface. 418 419The CPU model features for AP 420----------------------------- 421The AP stack relies on the presence of the AP instructions as well as two 422facilities: The AP Facilities Test (APFT) facility; and the AP Query 423Configuration Information (QCI) facility. These features/facilities are made 424available to a KVM guest via the following CPU model features: 425 4261. ap: Indicates whether the AP instructions are installed on the guest. This 427 feature will be enabled by KVM only if the AP instructions are installed 428 on the host. 429 4302. apft: Indicates the APFT facility is available on the guest. This facility 431 can be made available to the guest only if it is available on the host (i.e., 432 facility bit 15 is set). 433 4343. apqci: Indicates the AP QCI facility is available on the guest. This facility 435 can be made available to the guest only if it is available on the host (i.e., 436 facility bit 12 is set). 437 438Note: If the user chooses to specify a CPU model different than the 'host' 439model to QEMU, the CPU model features and facilities need to be turned on 440explicitly; for example:: 441 442 /usr/bin/qemu-system-s390x ... -cpu z13,ap=on,apqci=on,apft=on 443 444A guest can be precluded from using AP features/facilities by turning them off 445explicitly; for example:: 446 447 /usr/bin/qemu-system-s390x ... -cpu host,ap=off,apqci=off,apft=off 448 449Note: If the APFT facility is turned off (apft=off) for the guest, the guest 450will not see any AP devices. The zcrypt device drivers that register for type 10 451and newer AP devices - i.e., the cex4card and cex4queue device drivers - need 452the APFT facility to ascertain the facilities installed on a given AP device. If 453the APFT facility is not installed on the guest, then the probe of device 454drivers will fail since only type 10 and newer devices can be configured for 455guest use. 456 457Example 458======= 459Let's now provide an example to illustrate how KVM guests may be given 460access to AP facilities. For this example, we will show how to configure 461three guests such that executing the lszcrypt command on the guests would 462look like this: 463 464Guest1 465------ 466=========== ===== ============ 467CARD.DOMAIN TYPE MODE 468=========== ===== ============ 46905 CEX5C CCA-Coproc 47005.0004 CEX5C CCA-Coproc 47105.00ab CEX5C CCA-Coproc 47206 CEX5A Accelerator 47306.0004 CEX5A Accelerator 47406.00ab CEX5C CCA-Coproc 475=========== ===== ============ 476 477Guest2 478------ 479=========== ===== ============ 480CARD.DOMAIN TYPE MODE 481=========== ===== ============ 48205 CEX5A Accelerator 48305.0047 CEX5A Accelerator 48405.00ff CEX5A Accelerator 485=========== ===== ============ 486 487Guest3 488------ 489=========== ===== ============ 490CARD.DOMAIN TYPE MODE 491=========== ===== ============ 49206 CEX5A Accelerator 49306.0047 CEX5A Accelerator 49406.00ff CEX5A Accelerator 495=========== ===== ============ 496 497These are the steps: 498 4991. Install the vfio_ap module on the linux host. The dependency chain for the 500 vfio_ap module is: 501 * iommu 502 * s390 503 * zcrypt 504 * vfio 505 * vfio_mdev 506 * vfio_mdev_device 507 * KVM 508 509 To build the vfio_ap module, the kernel build must be configured with the 510 following Kconfig elements selected: 511 * IOMMU_SUPPORT 512 * S390 513 * ZCRYPT 514 * S390_AP_IOMMU 515 * VFIO 516 * VFIO_MDEV 517 * VFIO_MDEV_DEVICE 518 * KVM 519 520 If using make menuconfig select the following to build the vfio_ap module:: 521 522 -> Device Drivers 523 -> IOMMU Hardware Support 524 select S390 AP IOMMU Support 525 -> VFIO Non-Privileged userspace driver framework 526 -> Mediated device driver frramework 527 -> VFIO driver for Mediated devices 528 -> I/O subsystem 529 -> VFIO support for AP devices 530 5312. Secure the AP queues to be used by the three guests so that the host can not 532 access them. To secure them, there are two sysfs files that specify 533 bitmasks marking a subset of the APQN range as 'usable by the default AP 534 queue device drivers' or 'not usable by the default device drivers' and thus 535 available for use by the vfio_ap device driver'. The location of the sysfs 536 files containing the masks are:: 537 538 /sys/bus/ap/apmask 539 /sys/bus/ap/aqmask 540 541 The 'apmask' is a 256-bit mask that identifies a set of AP adapter IDs 542 (APID). Each bit in the mask, from left to right (i.e., from most significant 543 to least significant bit in big endian order), corresponds to an APID from 544 0-255. If a bit is set, the APID is marked as usable only by the default AP 545 queue device drivers; otherwise, the APID is usable by the vfio_ap 546 device driver. 547 548 The 'aqmask' is a 256-bit mask that identifies a set of AP queue indexes 549 (APQI). Each bit in the mask, from left to right (i.e., from most significant 550 to least significant bit in big endian order), corresponds to an APQI from 551 0-255. If a bit is set, the APQI is marked as usable only by the default AP 552 queue device drivers; otherwise, the APQI is usable by the vfio_ap device 553 driver. 554 555 Take, for example, the following mask:: 556 557 0x7dffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff 558 559 It indicates: 560 561 1, 2, 3, 4, 5, and 7-255 belong to the default drivers' pool, and 0 and 6 562 belong to the vfio_ap device driver's pool. 563 564 The APQN of each AP queue device assigned to the linux host is checked by the 565 AP bus against the set of APQNs derived from the cross product of APIDs 566 and APQIs marked as usable only by the default AP queue device drivers. If a 567 match is detected, only the default AP queue device drivers will be probed; 568 otherwise, the vfio_ap device driver will be probed. 569 570 By default, the two masks are set to reserve all APQNs for use by the default 571 AP queue device drivers. There are two ways the default masks can be changed: 572 573 1. The sysfs mask files can be edited by echoing a string into the 574 respective sysfs mask file in one of two formats: 575 576 * An absolute hex string starting with 0x - like "0x12345678" - sets 577 the mask. If the given string is shorter than the mask, it is padded 578 with 0s on the right; for example, specifying a mask value of 0x41 is 579 the same as specifying:: 580 581 0x4100000000000000000000000000000000000000000000000000000000000000 582 583 Keep in mind that the mask reads from left to right (i.e., most 584 significant to least significant bit in big endian order), so the mask 585 above identifies device numbers 1 and 7 (01000001). 586 587 If the string is longer than the mask, the operation is terminated with 588 an error (EINVAL). 589 590 * Individual bits in the mask can be switched on and off by specifying 591 each bit number to be switched in a comma separated list. Each bit 592 number string must be prepended with a ('+') or minus ('-') to indicate 593 the corresponding bit is to be switched on ('+') or off ('-'). Some 594 valid values are: 595 596 - "+0" switches bit 0 on 597 - "-13" switches bit 13 off 598 - "+0x41" switches bit 65 on 599 - "-0xff" switches bit 255 off 600 601 The following example: 602 603 +0,-6,+0x47,-0xf0 604 605 Switches bits 0 and 71 (0x47) on 606 607 Switches bits 6 and 240 (0xf0) off 608 609 Note that the bits not specified in the list remain as they were before 610 the operation. 611 612 2. The masks can also be changed at boot time via parameters on the kernel 613 command line like this: 614 615 ap.apmask=0xffff ap.aqmask=0x40 616 617 This would create the following masks:: 618 619 apmask: 620 0xffff000000000000000000000000000000000000000000000000000000000000 621 622 aqmask: 623 0x4000000000000000000000000000000000000000000000000000000000000000 624 625 Resulting in these two pools:: 626 627 default drivers pool: adapter 0-15, domain 1 628 alternate drivers pool: adapter 16-255, domains 0, 2-255 629 630Securing the APQNs for our example 631---------------------------------- 632 To secure the AP queues 05.0004, 05.0047, 05.00ab, 05.00ff, 06.0004, 06.0047, 633 06.00ab, and 06.00ff for use by the vfio_ap device driver, the corresponding 634 APQNs can either be removed from the default masks:: 635 636 echo -5,-6 > /sys/bus/ap/apmask 637 638 echo -4,-0x47,-0xab,-0xff > /sys/bus/ap/aqmask 639 640 Or the masks can be set as follows:: 641 642 echo 0xf9ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff \ 643 > apmask 644 645 echo 0xf7fffffffffffffffeffffffffffffffffffffffffeffffffffffffffffffffe \ 646 > aqmask 647 648 This will result in AP queues 05.0004, 05.0047, 05.00ab, 05.00ff, 06.0004, 649 06.0047, 06.00ab, and 06.00ff getting bound to the vfio_ap device driver. The 650 sysfs directory for the vfio_ap device driver will now contain symbolic links 651 to the AP queue devices bound to it:: 652 653 /sys/bus/ap 654 ... [drivers] 655 ...... [vfio_ap] 656 ......... [05.0004] 657 ......... [05.0047] 658 ......... [05.00ab] 659 ......... [05.00ff] 660 ......... [06.0004] 661 ......... [06.0047] 662 ......... [06.00ab] 663 ......... [06.00ff] 664 665 Keep in mind that only type 10 and newer adapters (i.e., CEX4 and later) 666 can be bound to the vfio_ap device driver. The reason for this is to 667 simplify the implementation by not needlessly complicating the design by 668 supporting older devices that will go out of service in the relatively near 669 future and for which there are few older systems on which to test. 670 671 The administrator, therefore, must take care to secure only AP queues that 672 can be bound to the vfio_ap device driver. The device type for a given AP 673 queue device can be read from the parent card's sysfs directory. For example, 674 to see the hardware type of the queue 05.0004: 675 676 cat /sys/bus/ap/devices/card05/hwtype 677 678 The hwtype must be 10 or higher (CEX4 or newer) in order to be bound to the 679 vfio_ap device driver. 680 6813. Create the mediated devices needed to configure the AP matrixes for the 682 three guests and to provide an interface to the vfio_ap driver for 683 use by the guests:: 684 685 /sys/devices/vfio_ap/matrix/ 686 --- [mdev_supported_types] 687 ------ [vfio_ap-passthrough] (passthrough mediated matrix device type) 688 --------- create 689 --------- [devices] 690 691 To create the mediated devices for the three guests:: 692 693 uuidgen > create 694 uuidgen > create 695 uuidgen > create 696 697 or 698 699 echo $uuid1 > create 700 echo $uuid2 > create 701 echo $uuid3 > create 702 703 This will create three mediated devices in the [devices] subdirectory named 704 after the UUID written to the create attribute file. We call them $uuid1, 705 $uuid2 and $uuid3 and this is the sysfs directory structure after creation:: 706 707 /sys/devices/vfio_ap/matrix/ 708 --- [mdev_supported_types] 709 ------ [vfio_ap-passthrough] 710 --------- [devices] 711 ------------ [$uuid1] 712 --------------- assign_adapter 713 --------------- assign_control_domain 714 --------------- assign_domain 715 --------------- matrix 716 --------------- unassign_adapter 717 --------------- unassign_control_domain 718 --------------- unassign_domain 719 720 ------------ [$uuid2] 721 --------------- assign_adapter 722 --------------- assign_control_domain 723 --------------- assign_domain 724 --------------- matrix 725 --------------- unassign_adapter 726 ----------------unassign_control_domain 727 ----------------unassign_domain 728 729 ------------ [$uuid3] 730 --------------- assign_adapter 731 --------------- assign_control_domain 732 --------------- assign_domain 733 --------------- matrix 734 --------------- unassign_adapter 735 ----------------unassign_control_domain 736 ----------------unassign_domain 737 7384. The administrator now needs to configure the matrixes for the mediated 739 devices $uuid1 (for Guest1), $uuid2 (for Guest2) and $uuid3 (for Guest3). 740 741 This is how the matrix is configured for Guest1:: 742 743 echo 5 > assign_adapter 744 echo 6 > assign_adapter 745 echo 4 > assign_domain 746 echo 0xab > assign_domain 747 748 Control domains can similarly be assigned using the assign_control_domain 749 sysfs file. 750 751 If a mistake is made configuring an adapter, domain or control domain, 752 you can use the unassign_xxx files to unassign the adapter, domain or 753 control domain. 754 755 To display the matrix configuration for Guest1:: 756 757 cat matrix 758 759 This is how the matrix is configured for Guest2:: 760 761 echo 5 > assign_adapter 762 echo 0x47 > assign_domain 763 echo 0xff > assign_domain 764 765 This is how the matrix is configured for Guest3:: 766 767 echo 6 > assign_adapter 768 echo 0x47 > assign_domain 769 echo 0xff > assign_domain 770 771 In order to successfully assign an adapter: 772 773 * The adapter number specified must represent a value from 0 up to the 774 maximum adapter number configured for the system. If an adapter number 775 higher than the maximum is specified, the operation will terminate with 776 an error (ENODEV). 777 778 * All APQNs that can be derived from the adapter ID and the IDs of 779 the previously assigned domains must be bound to the vfio_ap device 780 driver. If no domains have yet been assigned, then there must be at least 781 one APQN with the specified APID bound to the vfio_ap driver. If no such 782 APQNs are bound to the driver, the operation will terminate with an 783 error (EADDRNOTAVAIL). 784 785 No APQN that can be derived from the adapter ID and the IDs of the 786 previously assigned domains can be assigned to another mediated matrix 787 device. If an APQN is assigned to another mediated matrix device, the 788 operation will terminate with an error (EADDRINUSE). 789 790 In order to successfully assign a domain: 791 792 * The domain number specified must represent a value from 0 up to the 793 maximum domain number configured for the system. If a domain number 794 higher than the maximum is specified, the operation will terminate with 795 an error (ENODEV). 796 797 * All APQNs that can be derived from the domain ID and the IDs of 798 the previously assigned adapters must be bound to the vfio_ap device 799 driver. If no domains have yet been assigned, then there must be at least 800 one APQN with the specified APQI bound to the vfio_ap driver. If no such 801 APQNs are bound to the driver, the operation will terminate with an 802 error (EADDRNOTAVAIL). 803 804 No APQN that can be derived from the domain ID and the IDs of the 805 previously assigned adapters can be assigned to another mediated matrix 806 device. If an APQN is assigned to another mediated matrix device, the 807 operation will terminate with an error (EADDRINUSE). 808 809 In order to successfully assign a control domain, the domain number 810 specified must represent a value from 0 up to the maximum domain number 811 configured for the system. If a control domain number higher than the maximum 812 is specified, the operation will terminate with an error (ENODEV). 813 8145. Start Guest1:: 815 816 /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \ 817 -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid1 ... 818 8197. Start Guest2:: 820 821 /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \ 822 -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid2 ... 823 8247. Start Guest3:: 825 826 /usr/bin/qemu-system-s390x ... -cpu host,ap=on,apqci=on,apft=on \ 827 -device vfio-ap,sysfsdev=/sys/devices/vfio_ap/matrix/$uuid3 ... 828 829When the guest is shut down, the mediated matrix devices may be removed. 830 831Using our example again, to remove the mediated matrix device $uuid1:: 832 833 /sys/devices/vfio_ap/matrix/ 834 --- [mdev_supported_types] 835 ------ [vfio_ap-passthrough] 836 --------- [devices] 837 ------------ [$uuid1] 838 --------------- remove 839 840:: 841 842 echo 1 > remove 843 844This will remove all of the mdev matrix device's sysfs structures including 845the mdev device itself. To recreate and reconfigure the mdev matrix device, 846all of the steps starting with step 3 will have to be performed again. Note 847that the remove will fail if a guest using the mdev is still running. 848 849It is not necessary to remove an mdev matrix device, but one may want to 850remove it if no guest will use it during the remaining lifetime of the linux 851host. If the mdev matrix device is removed, one may want to also reconfigure 852the pool of adapters and queues reserved for use by the default drivers. 853 854Limitations 855=========== 856* The KVM/kernel interfaces do not provide a way to prevent restoring an APQN 857 to the default drivers pool of a queue that is still assigned to a mediated 858 device in use by a guest. It is incumbent upon the administrator to 859 ensure there is no mediated device in use by a guest to which the APQN is 860 assigned lest the host be given access to the private data of the AP queue 861 device such as a private key configured specifically for the guest. 862 863* Dynamically modifying the AP matrix for a running guest (which would amount to 864 hot(un)plug of AP devices for the guest) is currently not supported 865 866* Live guest migration is not supported for guests using AP devices. 867