1 /****************************************************************************** 2 * blkif.h 3 * 4 * Unified block-device I/O interface for Xen guest OSes. 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a copy 7 * of this software and associated documentation files (the "Software"), to 8 * deal in the Software without restriction, including without limitation the 9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or 10 * sell copies of the Software, and to permit persons to whom the Software is 11 * furnished to do so, subject to the following conditions: 12 * 13 * The above copyright notice and this permission notice shall be included in 14 * all copies or substantial portions of the Software. 15 * 16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 21 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER 22 * DEALINGS IN THE SOFTWARE. 23 * 24 * Copyright (c) 2003-2004, Keir Fraser 25 * Copyright (c) 2012, Spectra Logic Corporation 26 */ 27 28 #ifndef __XEN_PUBLIC_IO_BLKIF_H__ 29 #define __XEN_PUBLIC_IO_BLKIF_H__ 30 31 #include "ring.h" 32 #include "../grant_table.h" 33 34 /* 35 * Front->back notifications: When enqueuing a new request, sending a 36 * notification can be made conditional on req_event (i.e., the generic 37 * hold-off mechanism provided by the ring macros). Backends must set 38 * req_event appropriately (e.g., using RING_FINAL_CHECK_FOR_REQUESTS()). 39 * 40 * Back->front notifications: When enqueuing a new response, sending a 41 * notification can be made conditional on rsp_event (i.e., the generic 42 * hold-off mechanism provided by the ring macros). Frontends must set 43 * rsp_event appropriately (e.g., using RING_FINAL_CHECK_FOR_RESPONSES()). 44 */ 45 46 #ifndef blkif_vdev_t 47 #define blkif_vdev_t uint16_t 48 #endif 49 #define blkif_sector_t uint64_t 50 51 /* 52 * Feature and Parameter Negotiation 53 * ================================= 54 * The two halves of a Xen block driver utilize nodes within the XenStore to 55 * communicate capabilities and to negotiate operating parameters. This 56 * section enumerates these nodes which reside in the respective front and 57 * backend portions of the XenStore, following the XenBus convention. 58 * 59 * All data in the XenStore is stored as strings. Nodes specifying numeric 60 * values are encoded in decimal. Integer value ranges listed below are 61 * expressed as fixed sized integer types capable of storing the conversion 62 * of a properly formated node string, without loss of information. 63 * 64 * Any specified default value is in effect if the corresponding XenBus node 65 * is not present in the XenStore. 66 * 67 * XenStore nodes in sections marked "PRIVATE" are solely for use by the 68 * driver side whose XenBus tree contains them. 69 * 70 * XenStore nodes marked "DEPRECATED" in their notes section should only be 71 * used to provide interoperability with legacy implementations. 72 * 73 * See the XenBus state transition diagram below for details on when XenBus 74 * nodes must be published and when they can be queried. 75 * 76 ***************************************************************************** 77 * Backend XenBus Nodes 78 ***************************************************************************** 79 * 80 *------------------ Backend Device Identification (PRIVATE) ------------------ 81 * 82 * mode 83 * Values: "r" (read only), "w" (writable) 84 * 85 * The read or write access permissions to the backing store to be 86 * granted to the frontend. 87 * 88 * params 89 * Values: string 90 * 91 * A free formatted string providing sufficient information for the 92 * hotplug script to attach the device and provide a suitable 93 * handler (ie: a block device) for blkback to use. 94 * 95 * physical-device 96 * Values: "MAJOR:MINOR" 97 * Notes: 11 98 * 99 * MAJOR and MINOR are the major number and minor number of the 100 * backing device respectively. 101 * 102 * physical-device-path 103 * Values: path string 104 * 105 * A string that contains the absolute path to the disk image. On 106 * NetBSD and Linux this is always a block device, while on FreeBSD 107 * it can be either a block device or a regular file. 108 * 109 * type 110 * Values: "file", "phy", "tap" 111 * 112 * The type of the backing device/object. 113 * 114 * 115 * direct-io-safe 116 * Values: 0/1 (boolean) 117 * Default Value: 0 118 * 119 * The underlying storage is not affected by the direct IO memory 120 * lifetime bug. See: 121 * http://lists.xen.org/archives/html/xen-devel/2012-12/msg01154.html 122 * 123 * Therefore this option gives the backend permission to use 124 * O_DIRECT, notwithstanding that bug. 125 * 126 * That is, if this option is enabled, use of O_DIRECT is safe, 127 * in circumstances where we would normally have avoided it as a 128 * workaround for that bug. This option is not relevant for all 129 * backends, and even not necessarily supported for those for 130 * which it is relevant. A backend which knows that it is not 131 * affected by the bug can ignore this option. 132 * 133 * This option doesn't require a backend to use O_DIRECT, so it 134 * should not be used to try to control the caching behaviour. 135 * 136 *--------------------------------- Features --------------------------------- 137 * 138 * feature-barrier 139 * Values: 0/1 (boolean) 140 * Default Value: 0 141 * 142 * A value of "1" indicates that the backend can process requests 143 * containing the BLKIF_OP_WRITE_BARRIER request opcode. Requests 144 * of this type may still be returned at any time with the 145 * BLKIF_RSP_EOPNOTSUPP result code. 146 * 147 * feature-flush-cache 148 * Values: 0/1 (boolean) 149 * Default Value: 0 150 * 151 * A value of "1" indicates that the backend can process requests 152 * containing the BLKIF_OP_FLUSH_DISKCACHE request opcode. Requests 153 * of this type may still be returned at any time with the 154 * BLKIF_RSP_EOPNOTSUPP result code. 155 * 156 * feature-discard 157 * Values: 0/1 (boolean) 158 * Default Value: 0 159 * 160 * A value of "1" indicates that the backend can process requests 161 * containing the BLKIF_OP_DISCARD request opcode. Requests 162 * of this type may still be returned at any time with the 163 * BLKIF_RSP_EOPNOTSUPP result code. 164 * 165 * feature-persistent 166 * Values: 0/1 (boolean) 167 * Default Value: 0 168 * Notes: 7 169 * 170 * A value of "1" indicates that the backend can keep the grants used 171 * by the frontend driver mapped, so the same set of grants should be 172 * used in all transactions. The maximum number of grants the backend 173 * can map persistently depends on the implementation, but ideally it 174 * should be RING_SIZE * BLKIF_MAX_SEGMENTS_PER_REQUEST. Using this 175 * feature the backend doesn't need to unmap each grant, preventing 176 * costly TLB flushes. The backend driver should only map grants 177 * persistently if the frontend supports it. If a backend driver chooses 178 * to use the persistent protocol when the frontend doesn't support it, 179 * it will probably hit the maximum number of persistently mapped grants 180 * (due to the fact that the frontend won't be reusing the same grants), 181 * and fall back to non-persistent mode. Backend implementations may 182 * shrink or expand the number of persistently mapped grants without 183 * notifying the frontend depending on memory constraints (this might 184 * cause a performance degradation). 185 * 186 * If a backend driver wants to limit the maximum number of persistently 187 * mapped grants to a value less than RING_SIZE * 188 * BLKIF_MAX_SEGMENTS_PER_REQUEST a LRU strategy should be used to 189 * discard the grants that are less commonly used. Using a LRU in the 190 * backend driver paired with a LIFO queue in the frontend will 191 * allow us to have better performance in this scenario. 192 * 193 *----------------------- Request Transport Parameters ------------------------ 194 * 195 * max-ring-page-order 196 * Values: <uint32_t> 197 * Default Value: 0 198 * Notes: 1, 3 199 * 200 * The maximum supported size of the request ring buffer in units of 201 * lb(machine pages). (e.g. 0 == 1 page, 1 = 2 pages, 2 == 4 pages, 202 * etc.). 203 * 204 * max-ring-pages 205 * Values: <uint32_t> 206 * Default Value: 1 207 * Notes: DEPRECATED, 2, 3 208 * 209 * The maximum supported size of the request ring buffer in units of 210 * machine pages. The value must be a power of 2. 211 * 212 *------------------------- Backend Device Properties ------------------------- 213 * 214 * discard-enable 215 * Values: 0/1 (boolean) 216 * Default Value: 1 217 * 218 * This optional property, set by the toolstack, instructs the backend 219 * to offer (or not to offer) discard to the frontend. If the property 220 * is missing the backend should offer discard if the backing storage 221 * actually supports it. 222 * 223 * discard-alignment 224 * Values: <uint32_t> 225 * Default Value: 0 226 * Notes: 4, 5 227 * 228 * The offset, in bytes from the beginning of the virtual block device, 229 * to the first, addressable, discard extent on the underlying device. 230 * 231 * discard-granularity 232 * Values: <uint32_t> 233 * Default Value: <"sector-size"> 234 * Notes: 4 235 * 236 * The size, in bytes, of the individually addressable discard extents 237 * of the underlying device. 238 * 239 * discard-secure 240 * Values: 0/1 (boolean) 241 * Default Value: 0 242 * Notes: 10 243 * 244 * A value of "1" indicates that the backend can process BLKIF_OP_DISCARD 245 * requests with the BLKIF_DISCARD_SECURE flag set. 246 * 247 * info 248 * Values: <uint32_t> (bitmap) 249 * 250 * A collection of bit flags describing attributes of the backing 251 * device. The VDISK_* macros define the meaning of each bit 252 * location. 253 * 254 * sector-size 255 * Values: <uint32_t> 256 * 257 * The logical block size, in bytes, of the underlying storage. This 258 * must be a power of two with a minimum value of 512. 259 * 260 * NOTE: Because of implementation bugs in some frontends this must be 261 * set to 512, unless the frontend advertizes a non-zero value 262 * in its "feature-large-sector-size" xenbus node. (See below). 263 * 264 * physical-sector-size 265 * Values: <uint32_t> 266 * Default Value: <"sector-size"> 267 * 268 * The physical block size, in bytes, of the backend storage. This 269 * must be an integer multiple of "sector-size". 270 * 271 * sectors 272 * Values: <uint64_t> 273 * 274 * The size of the backend device, expressed in units of "sector-size". 275 * The product of "sector-size" and "sectors" must also be an integer 276 * multiple of "physical-sector-size", if that node is present. 277 * 278 ***************************************************************************** 279 * Frontend XenBus Nodes 280 ***************************************************************************** 281 * 282 *----------------------- Request Transport Parameters ----------------------- 283 * 284 * event-channel 285 * Values: <uint32_t> 286 * 287 * The identifier of the Xen event channel used to signal activity 288 * in the ring buffer. 289 * 290 * ring-ref 291 * Values: <uint32_t> 292 * Notes: 6 293 * 294 * The Xen grant reference granting permission for the backend to map 295 * the sole page in a single page sized ring buffer. 296 * 297 * ring-ref%u 298 * Values: <uint32_t> 299 * Notes: 6 300 * 301 * For a frontend providing a multi-page ring, a "number of ring pages" 302 * sized list of nodes, each containing a Xen grant reference granting 303 * permission for the backend to map the page of the ring located 304 * at page index "%u". Page indexes are zero based. 305 * 306 * protocol 307 * Values: string (XEN_IO_PROTO_ABI_*) 308 * Default Value: XEN_IO_PROTO_ABI_NATIVE 309 * 310 * The machine ABI rules governing the format of all ring request and 311 * response structures. 312 * 313 * ring-page-order 314 * Values: <uint32_t> 315 * Default Value: 0 316 * Maximum Value: MAX(ffs(max-ring-pages) - 1, max-ring-page-order) 317 * Notes: 1, 3 318 * 319 * The size of the frontend allocated request ring buffer in units 320 * of lb(machine pages). (e.g. 0 == 1 page, 1 = 2 pages, 2 == 4 pages, 321 * etc.). 322 * 323 * num-ring-pages 324 * Values: <uint32_t> 325 * Default Value: 1 326 * Maximum Value: MAX(max-ring-pages,(0x1 << max-ring-page-order)) 327 * Notes: DEPRECATED, 2, 3 328 * 329 * The size of the frontend allocated request ring buffer in units of 330 * machine pages. The value must be a power of 2. 331 * 332 *--------------------------------- Features --------------------------------- 333 * 334 * feature-persistent 335 * Values: 0/1 (boolean) 336 * Default Value: 0 337 * Notes: 7, 8, 9 338 * 339 * A value of "1" indicates that the frontend will reuse the same grants 340 * for all transactions, allowing the backend to map them with write 341 * access (even when it should be read-only). If the frontend hits the 342 * maximum number of allowed persistently mapped grants, it can fallback 343 * to non persistent mode. This will cause a performance degradation, 344 * since the backend driver will still try to map those grants 345 * persistently. Since the persistent grants protocol is compatible with 346 * the previous protocol, a frontend driver can choose to work in 347 * persistent mode even when the backend doesn't support it. 348 * 349 * It is recommended that the frontend driver stores the persistently 350 * mapped grants in a LIFO queue, so a subset of all persistently mapped 351 * grants gets used commonly. This is done in case the backend driver 352 * decides to limit the maximum number of persistently mapped grants 353 * to a value less than RING_SIZE * BLKIF_MAX_SEGMENTS_PER_REQUEST. 354 * 355 * feature-large-sector-size 356 * Values: 0/1 (boolean) 357 * Default Value: 0 358 * 359 * A value of "1" indicates that the frontend will correctly supply and 360 * interpret all sector-based quantities in terms of the "sector-size" 361 * value supplied in the backend info, whatever that may be set to. 362 * If this node is not present or its value is "0" then it is assumed 363 * that the frontend requires that the logical block size is 512 as it 364 * is hardcoded (which is the case in some frontend implementations). 365 * 366 *------------------------- Virtual Device Properties ------------------------- 367 * 368 * device-type 369 * Values: "disk", "cdrom", "floppy", etc. 370 * 371 * virtual-device 372 * Values: <uint32_t> 373 * 374 * A value indicating the physical device to virtualize within the 375 * frontend's domain. (e.g. "The first ATA disk", "The third SCSI 376 * disk", etc.) 377 * 378 * See docs/misc/vbd-interface.txt for details on the format of this 379 * value. 380 * 381 * Notes 382 * ----- 383 * (1) Multi-page ring buffer scheme first developed in the Citrix XenServer 384 * PV drivers. 385 * (2) Multi-page ring buffer scheme first used in some RedHat distributions 386 * including a distribution deployed on certain nodes of the Amazon 387 * EC2 cluster. 388 * (3) Support for multi-page ring buffers was implemented independently, 389 * in slightly different forms, by both Citrix and RedHat/Amazon. 390 * For full interoperability, block front and backends should publish 391 * identical ring parameters, adjusted for unit differences, to the 392 * XenStore nodes used in both schemes. 393 * (4) Devices that support discard functionality may internally allocate space 394 * (discardable extents) in units that are larger than the exported logical 395 * block size. If the backing device has such discardable extents the 396 * backend should provide both discard-granularity and discard-alignment. 397 * Providing just one of the two may be considered an error by the frontend. 398 * Backends supporting discard should include discard-granularity and 399 * discard-alignment even if it supports discarding individual sectors. 400 * Frontends should assume discard-alignment == 0 and discard-granularity 401 * == sector size if these keys are missing. 402 * (5) The discard-alignment parameter allows a physical device to be 403 * partitioned into virtual devices that do not necessarily begin or 404 * end on a discardable extent boundary. 405 * (6) When there is only a single page allocated to the request ring, 406 * 'ring-ref' is used to communicate the grant reference for this 407 * page to the backend. When using a multi-page ring, the 'ring-ref' 408 * node is not created. Instead 'ring-ref0' - 'ring-refN' are used. 409 * (7) When using persistent grants data has to be copied from/to the page 410 * where the grant is currently mapped. The overhead of doing this copy 411 * however doesn't suppress the speed improvement of not having to unmap 412 * the grants. 413 * (8) The frontend driver has to allow the backend driver to map all grants 414 * with write access, even when they should be mapped read-only, since 415 * further requests may reuse these grants and require write permissions. 416 * (9) Linux implementation doesn't have a limit on the maximum number of 417 * grants that can be persistently mapped in the frontend driver, but 418 * due to the frontent driver implementation it should never be bigger 419 * than RING_SIZE * BLKIF_MAX_SEGMENTS_PER_REQUEST. 420 *(10) The discard-secure property may be present and will be set to 1 if the 421 * backing device supports secure discard. 422 *(11) Only used by Linux and NetBSD. 423 */ 424 425 /* 426 * Multiple hardware queues/rings: 427 * If supported, the backend will write the key "multi-queue-max-queues" to 428 * the directory for that vbd, and set its value to the maximum supported 429 * number of queues. 430 * Frontends that are aware of this feature and wish to use it can write the 431 * key "multi-queue-num-queues" with the number they wish to use, which must be 432 * greater than zero, and no more than the value reported by the backend in 433 * "multi-queue-max-queues". 434 * 435 * For frontends requesting just one queue, the usual event-channel and 436 * ring-ref keys are written as before, simplifying the backend processing 437 * to avoid distinguishing between a frontend that doesn't understand the 438 * multi-queue feature, and one that does, but requested only one queue. 439 * 440 * Frontends requesting two or more queues must not write the toplevel 441 * event-channel and ring-ref keys, instead writing those keys under sub-keys 442 * having the name "queue-N" where N is the integer ID of the queue/ring for 443 * which those keys belong. Queues are indexed from zero. 444 * For example, a frontend with two queues must write the following set of 445 * queue-related keys: 446 * 447 * /local/domain/1/device/vbd/0/multi-queue-num-queues = "2" 448 * /local/domain/1/device/vbd/0/queue-0 = "" 449 * /local/domain/1/device/vbd/0/queue-0/ring-ref = "<ring-ref#0>" 450 * /local/domain/1/device/vbd/0/queue-0/event-channel = "<evtchn#0>" 451 * /local/domain/1/device/vbd/0/queue-1 = "" 452 * /local/domain/1/device/vbd/0/queue-1/ring-ref = "<ring-ref#1>" 453 * /local/domain/1/device/vbd/0/queue-1/event-channel = "<evtchn#1>" 454 * 455 * It is also possible to use multiple queues/rings together with 456 * feature multi-page ring buffer. 457 * For example, a frontend requests two queues/rings and the size of each ring 458 * buffer is two pages must write the following set of related keys: 459 * 460 * /local/domain/1/device/vbd/0/multi-queue-num-queues = "2" 461 * /local/domain/1/device/vbd/0/ring-page-order = "1" 462 * /local/domain/1/device/vbd/0/queue-0 = "" 463 * /local/domain/1/device/vbd/0/queue-0/ring-ref0 = "<ring-ref#0>" 464 * /local/domain/1/device/vbd/0/queue-0/ring-ref1 = "<ring-ref#1>" 465 * /local/domain/1/device/vbd/0/queue-0/event-channel = "<evtchn#0>" 466 * /local/domain/1/device/vbd/0/queue-1 = "" 467 * /local/domain/1/device/vbd/0/queue-1/ring-ref0 = "<ring-ref#2>" 468 * /local/domain/1/device/vbd/0/queue-1/ring-ref1 = "<ring-ref#3>" 469 * /local/domain/1/device/vbd/0/queue-1/event-channel = "<evtchn#1>" 470 * 471 */ 472 473 /* 474 * STATE DIAGRAMS 475 * 476 ***************************************************************************** 477 * Startup * 478 ***************************************************************************** 479 * 480 * Tool stack creates front and back nodes with state XenbusStateInitialising. 481 * 482 * Front Back 483 * ================================= ===================================== 484 * XenbusStateInitialising XenbusStateInitialising 485 * o Query virtual device o Query backend device identification 486 * properties. data. 487 * o Setup OS device instance. o Open and validate backend device. 488 * o Publish backend features and 489 * transport parameters. 490 * | 491 * | 492 * V 493 * XenbusStateInitWait 494 * 495 * o Query backend features and 496 * transport parameters. 497 * o Allocate and initialize the 498 * request ring. 499 * o Publish transport parameters 500 * that will be in effect during 501 * this connection. 502 * | 503 * | 504 * V 505 * XenbusStateInitialised 506 * 507 * o Query frontend transport parameters. 508 * o Connect to the request ring and 509 * event channel. 510 * o Publish backend device properties. 511 * | 512 * | 513 * V 514 * XenbusStateConnected 515 * 516 * o Query backend device properties. 517 * o Finalize OS virtual device 518 * instance. 519 * | 520 * | 521 * V 522 * XenbusStateConnected 523 * 524 * Note: Drivers that do not support any optional features, or the negotiation 525 * of transport parameters, can skip certain states in the state machine: 526 * 527 * o A frontend may transition to XenbusStateInitialised without 528 * waiting for the backend to enter XenbusStateInitWait. In this 529 * case, default transport parameters are in effect and any 530 * transport parameters published by the frontend must contain 531 * their default values. 532 * 533 * o A backend may transition to XenbusStateInitialised, bypassing 534 * XenbusStateInitWait, without waiting for the frontend to first 535 * enter the XenbusStateInitialised state. In this case, default 536 * transport parameters are in effect and any transport parameters 537 * published by the backend must contain their default values. 538 * 539 * Drivers that support optional features and/or transport parameter 540 * negotiation must tolerate these additional state transition paths. 541 * In general this means performing the work of any skipped state 542 * transition, if it has not already been performed, in addition to the 543 * work associated with entry into the current state. 544 */ 545 546 /* 547 * REQUEST CODES. 548 */ 549 #define BLKIF_OP_READ 0 550 #define BLKIF_OP_WRITE 1 551 /* 552 * All writes issued prior to a request with the BLKIF_OP_WRITE_BARRIER 553 * operation code ("barrier request") must be completed prior to the 554 * execution of the barrier request. All writes issued after the barrier 555 * request must not execute until after the completion of the barrier request. 556 * 557 * Optional. See "feature-barrier" XenBus node documentation above. 558 */ 559 #define BLKIF_OP_WRITE_BARRIER 2 560 /* 561 * Commit any uncommitted contents of the backing device's volatile cache 562 * to stable storage. 563 * 564 * Optional. See "feature-flush-cache" XenBus node documentation above. 565 */ 566 #define BLKIF_OP_FLUSH_DISKCACHE 3 567 /* 568 * Used in SLES sources for device specific command packet 569 * contained within the request. Reserved for that purpose. 570 */ 571 #define BLKIF_OP_RESERVED_1 4 572 /* 573 * Indicate to the backend device that a region of storage is no longer in 574 * use, and may be discarded at any time without impact to the client. If 575 * the BLKIF_DISCARD_SECURE flag is set on the request, all copies of the 576 * discarded region on the device must be rendered unrecoverable before the 577 * command returns. 578 * 579 * This operation is analogous to performing a trim (ATA) or unamp (SCSI), 580 * command on a native device. 581 * 582 * More information about trim/unmap operations can be found at: 583 * http://t13.org/Documents/UploadedDocuments/docs2008/ 584 * e07154r6-Data_Set_Management_Proposal_for_ATA-ACS2.doc 585 * http://www.seagate.com/staticfiles/support/disc/manuals/ 586 * Interface%20manuals/100293068c.pdf 587 * 588 * Optional. See "feature-discard", "discard-alignment", 589 * "discard-granularity", and "discard-secure" in the XenBus node 590 * documentation above. 591 */ 592 #define BLKIF_OP_DISCARD 5 593 594 /* 595 * Recognized if "feature-max-indirect-segments" in present in the backend 596 * xenbus info. The "feature-max-indirect-segments" node contains the maximum 597 * number of segments allowed by the backend per request. If the node is 598 * present, the frontend might use blkif_request_indirect structs in order to 599 * issue requests with more than BLKIF_MAX_SEGMENTS_PER_REQUEST (11). The 600 * maximum number of indirect segments is fixed by the backend, but the 601 * frontend can issue requests with any number of indirect segments as long as 602 * it's less than the number provided by the backend. The indirect_grefs field 603 * in blkif_request_indirect should be filled by the frontend with the 604 * grant references of the pages that are holding the indirect segments. 605 * These pages are filled with an array of blkif_request_segment that hold the 606 * information about the segments. The number of indirect pages to use is 607 * determined by the number of segments an indirect request contains. Every 608 * indirect page can contain a maximum of 609 * (PAGE_SIZE / sizeof(struct blkif_request_segment)) segments, so to 610 * calculate the number of indirect pages to use we have to do 611 * ceil(indirect_segments / (PAGE_SIZE / sizeof(struct blkif_request_segment))). 612 * 613 * If a backend does not recognize BLKIF_OP_INDIRECT, it should *not* 614 * create the "feature-max-indirect-segments" node! 615 */ 616 #define BLKIF_OP_INDIRECT 6 617 618 /* 619 * Maximum scatter/gather segments per request. 620 * This is carefully chosen so that sizeof(blkif_ring_t) <= PAGE_SIZE. 621 * NB. This could be 12 if the ring indexes weren't stored in the same page. 622 */ 623 #define BLKIF_MAX_SEGMENTS_PER_REQUEST 11 624 625 /* 626 * Maximum number of indirect pages to use per request. 627 */ 628 #define BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST 8 629 630 /* 631 * NB. 'first_sect' and 'last_sect' in blkif_request_segment, as well as 632 * 'sector_number' in blkif_request, blkif_request_discard and 633 * blkif_request_indirect are sector-based quantities. See the description 634 * of the "feature-large-sector-size" frontend xenbus node above for 635 * more information. 636 */ 637 struct blkif_request_segment { 638 grant_ref_t gref; /* reference to I/O buffer frame */ 639 /* @first_sect: first sector in frame to transfer (inclusive). */ 640 /* @last_sect: last sector in frame to transfer (inclusive). */ 641 uint8_t first_sect, last_sect; 642 }; 643 644 /* 645 * Starting ring element for any I/O request. 646 */ 647 struct blkif_request { 648 uint8_t operation; /* BLKIF_OP_??? */ 649 uint8_t nr_segments; /* number of segments */ 650 blkif_vdev_t handle; /* only for read/write requests */ 651 uint64_t id; /* private guest value, echoed in resp */ 652 blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ 653 struct blkif_request_segment seg[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 654 }; 655 typedef struct blkif_request blkif_request_t; 656 657 /* 658 * Cast to this structure when blkif_request.operation == BLKIF_OP_DISCARD 659 * sizeof(struct blkif_request_discard) <= sizeof(struct blkif_request) 660 */ 661 struct blkif_request_discard { 662 uint8_t operation; /* BLKIF_OP_DISCARD */ 663 uint8_t flag; /* BLKIF_DISCARD_SECURE or zero */ 664 #define BLKIF_DISCARD_SECURE (1<<0) /* ignored if discard-secure=0 */ 665 blkif_vdev_t handle; /* same as for read/write requests */ 666 uint64_t id; /* private guest value, echoed in resp */ 667 blkif_sector_t sector_number;/* start sector idx on disk */ 668 uint64_t nr_sectors; /* number of contiguous sectors to discard*/ 669 }; 670 typedef struct blkif_request_discard blkif_request_discard_t; 671 672 struct blkif_request_indirect { 673 uint8_t operation; /* BLKIF_OP_INDIRECT */ 674 uint8_t indirect_op; /* BLKIF_OP_{READ/WRITE} */ 675 uint16_t nr_segments; /* number of segments */ 676 uint64_t id; /* private guest value, echoed in resp */ 677 blkif_sector_t sector_number;/* start sector idx on disk (r/w only) */ 678 blkif_vdev_t handle; /* same as for read/write requests */ 679 grant_ref_t indirect_grefs[BLKIF_MAX_INDIRECT_PAGES_PER_REQUEST]; 680 #ifdef __i386__ 681 uint64_t pad; /* Make it 64 byte aligned on i386 */ 682 #endif 683 }; 684 typedef struct blkif_request_indirect blkif_request_indirect_t; 685 686 struct blkif_response { 687 uint64_t id; /* copied from request */ 688 uint8_t operation; /* copied from request */ 689 int16_t status; /* BLKIF_RSP_??? */ 690 }; 691 typedef struct blkif_response blkif_response_t; 692 693 /* 694 * STATUS RETURN CODES. 695 */ 696 /* Operation not supported (only happens on barrier writes). */ 697 #define BLKIF_RSP_EOPNOTSUPP -2 698 /* Operation failed for some unspecified reason (-EIO). */ 699 #define BLKIF_RSP_ERROR -1 700 /* Operation completed successfully. */ 701 #define BLKIF_RSP_OKAY 0 702 703 /* 704 * Generate blkif ring structures and types. 705 */ 706 DEFINE_RING_TYPES(blkif, struct blkif_request, struct blkif_response); 707 708 #define VDISK_CDROM 0x1 709 #define VDISK_REMOVABLE 0x2 710 #define VDISK_READONLY 0x4 711 712 #endif /* __XEN_PUBLIC_IO_BLKIF_H__ */ 713