1 /* 2 * Xen SCSI frontend driver 3 * 4 * Copyright (c) 2008, FUJITSU Limited 5 * 6 * This program is free software; you can redistribute it and/or 7 * modify it under the terms of the GNU General Public License version 2 8 * as published by the Free Software Foundation; or, when distributed 9 * separately from the Linux kernel or incorporated into other 10 * software packages, subject to the following license: 11 * 12 * Permission is hereby granted, free of charge, to any person obtaining a copy 13 * of this source file (the "Software"), to deal in the Software without 14 * restriction, including without limitation the rights to use, copy, modify, 15 * merge, publish, distribute, sublicense, and/or sell copies of the Software, 16 * and to permit persons to whom the Software is furnished to do so, subject to 17 * the following conditions: 18 * 19 * The above copyright notice and this permission notice shall be included in 20 * all copies or substantial portions of the Software. 21 * 22 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 23 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 24 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 25 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 26 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 27 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS 28 * IN THE SOFTWARE. 29 */ 30 31 #include <linux/module.h> 32 #include <linux/kernel.h> 33 #include <linux/device.h> 34 #include <linux/wait.h> 35 #include <linux/interrupt.h> 36 #include <linux/mutex.h> 37 #include <linux/spinlock.h> 38 #include <linux/sched.h> 39 #include <linux/blkdev.h> 40 #include <linux/pfn.h> 41 #include <linux/slab.h> 42 #include <linux/bitops.h> 43 44 #include <scsi/scsi_cmnd.h> 45 #include <scsi/scsi_device.h> 46 #include <scsi/scsi.h> 47 #include <scsi/scsi_host.h> 48 49 #include <xen/xen.h> 50 #include <xen/xenbus.h> 51 #include <xen/grant_table.h> 52 #include <xen/events.h> 53 #include <xen/page.h> 54 55 #include <xen/interface/grant_table.h> 56 #include <xen/interface/io/vscsiif.h> 57 #include <xen/interface/io/protocols.h> 58 59 #include <asm/xen/hypervisor.h> 60 61 62 #define GRANT_INVALID_REF 0 63 64 #define VSCSIFRONT_OP_ADD_LUN 1 65 #define VSCSIFRONT_OP_DEL_LUN 2 66 #define VSCSIFRONT_OP_READD_LUN 3 67 68 /* Tuning point. */ 69 #define VSCSIIF_DEFAULT_CMD_PER_LUN 10 70 #define VSCSIIF_MAX_TARGET 64 71 #define VSCSIIF_MAX_LUN 255 72 73 #define VSCSIIF_RING_SIZE __CONST_RING_SIZE(vscsiif, PAGE_SIZE) 74 #define VSCSIIF_MAX_REQS VSCSIIF_RING_SIZE 75 76 #define vscsiif_grants_sg(_sg) (PFN_UP((_sg) * \ 77 sizeof(struct scsiif_request_segment))) 78 79 struct vscsifrnt_shadow { 80 /* command between backend and frontend */ 81 unsigned char act; 82 uint8_t nr_segments; 83 uint16_t rqid; 84 uint16_t ref_rqid; 85 86 unsigned int nr_grants; /* number of grants in gref[] */ 87 struct scsiif_request_segment *sg; /* scatter/gather elements */ 88 struct scsiif_request_segment seg[VSCSIIF_SG_TABLESIZE]; 89 90 /* Do reset or abort function. */ 91 wait_queue_head_t wq_reset; /* reset work queue */ 92 int wait_reset; /* reset work queue condition */ 93 int32_t rslt_reset; /* reset response status: */ 94 /* SUCCESS or FAILED or: */ 95 #define RSLT_RESET_WAITING 0 96 #define RSLT_RESET_ERR -1 97 98 /* Requested struct scsi_cmnd is stored from kernel. */ 99 struct scsi_cmnd *sc; 100 int gref[vscsiif_grants_sg(SG_ALL) + SG_ALL]; 101 }; 102 103 struct vscsifrnt_info { 104 struct xenbus_device *dev; 105 106 struct Scsi_Host *host; 107 int host_active; 108 109 unsigned int evtchn; 110 unsigned int irq; 111 112 grant_ref_t ring_ref; 113 struct vscsiif_front_ring ring; 114 struct vscsiif_response ring_rsp; 115 116 spinlock_t shadow_lock; 117 DECLARE_BITMAP(shadow_free_bitmap, VSCSIIF_MAX_REQS); 118 struct vscsifrnt_shadow *shadow[VSCSIIF_MAX_REQS]; 119 120 /* Following items are protected by the host lock. */ 121 wait_queue_head_t wq_sync; 122 wait_queue_head_t wq_pause; 123 unsigned int wait_ring_available:1; 124 unsigned int waiting_pause:1; 125 unsigned int pause:1; 126 unsigned callers; 127 128 char dev_state_path[64]; 129 struct task_struct *curr; 130 }; 131 132 static DEFINE_MUTEX(scsifront_mutex); 133 134 static void scsifront_wake_up(struct vscsifrnt_info *info) 135 { 136 info->wait_ring_available = 0; 137 wake_up(&info->wq_sync); 138 } 139 140 static int scsifront_get_rqid(struct vscsifrnt_info *info) 141 { 142 unsigned long flags; 143 int free; 144 145 spin_lock_irqsave(&info->shadow_lock, flags); 146 147 free = find_first_bit(info->shadow_free_bitmap, VSCSIIF_MAX_REQS); 148 __clear_bit(free, info->shadow_free_bitmap); 149 150 spin_unlock_irqrestore(&info->shadow_lock, flags); 151 152 return free; 153 } 154 155 static int _scsifront_put_rqid(struct vscsifrnt_info *info, uint32_t id) 156 { 157 int empty = bitmap_empty(info->shadow_free_bitmap, VSCSIIF_MAX_REQS); 158 159 __set_bit(id, info->shadow_free_bitmap); 160 info->shadow[id] = NULL; 161 162 return empty || info->wait_ring_available; 163 } 164 165 static void scsifront_put_rqid(struct vscsifrnt_info *info, uint32_t id) 166 { 167 unsigned long flags; 168 int kick; 169 170 spin_lock_irqsave(&info->shadow_lock, flags); 171 kick = _scsifront_put_rqid(info, id); 172 spin_unlock_irqrestore(&info->shadow_lock, flags); 173 174 if (kick) 175 scsifront_wake_up(info); 176 } 177 178 static int scsifront_do_request(struct vscsifrnt_info *info, 179 struct vscsifrnt_shadow *shadow) 180 { 181 struct vscsiif_front_ring *ring = &(info->ring); 182 struct vscsiif_request *ring_req; 183 struct scsi_cmnd *sc = shadow->sc; 184 uint32_t id; 185 int i, notify; 186 187 if (RING_FULL(&info->ring)) 188 return -EBUSY; 189 190 id = scsifront_get_rqid(info); /* use id in response */ 191 if (id >= VSCSIIF_MAX_REQS) 192 return -EBUSY; 193 194 info->shadow[id] = shadow; 195 shadow->rqid = id; 196 197 ring_req = RING_GET_REQUEST(&(info->ring), ring->req_prod_pvt); 198 ring->req_prod_pvt++; 199 200 ring_req->rqid = id; 201 ring_req->act = shadow->act; 202 ring_req->ref_rqid = shadow->ref_rqid; 203 ring_req->nr_segments = shadow->nr_segments; 204 205 ring_req->id = sc->device->id; 206 ring_req->lun = sc->device->lun; 207 ring_req->channel = sc->device->channel; 208 ring_req->cmd_len = sc->cmd_len; 209 210 BUG_ON(sc->cmd_len > VSCSIIF_MAX_COMMAND_SIZE); 211 212 memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len); 213 214 ring_req->sc_data_direction = (uint8_t)sc->sc_data_direction; 215 ring_req->timeout_per_command = sc->request->timeout / HZ; 216 217 for (i = 0; i < (shadow->nr_segments & ~VSCSIIF_SG_GRANT); i++) 218 ring_req->seg[i] = shadow->seg[i]; 219 220 RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(ring, notify); 221 if (notify) 222 notify_remote_via_irq(info->irq); 223 224 return 0; 225 } 226 227 static void scsifront_gnttab_done(struct vscsifrnt_info *info, 228 struct vscsifrnt_shadow *shadow) 229 { 230 int i; 231 232 if (shadow->sc->sc_data_direction == DMA_NONE) 233 return; 234 235 for (i = 0; i < shadow->nr_grants; i++) { 236 if (unlikely(gnttab_query_foreign_access(shadow->gref[i]))) { 237 shost_printk(KERN_ALERT, info->host, KBUILD_MODNAME 238 "grant still in use by backend\n"); 239 BUG(); 240 } 241 gnttab_end_foreign_access(shadow->gref[i], 0, 0UL); 242 } 243 244 kfree(shadow->sg); 245 } 246 247 static void scsifront_cdb_cmd_done(struct vscsifrnt_info *info, 248 struct vscsiif_response *ring_rsp) 249 { 250 struct vscsifrnt_shadow *shadow; 251 struct scsi_cmnd *sc; 252 uint32_t id; 253 uint8_t sense_len; 254 int result; 255 256 id = ring_rsp->rqid; 257 shadow = info->shadow[id]; 258 sc = shadow->sc; 259 260 BUG_ON(sc == NULL); 261 262 scsifront_gnttab_done(info, shadow); 263 scsifront_put_rqid(info, id); 264 265 result = ring_rsp->rslt; 266 if (result >> 24) 267 set_host_byte(sc, DID_ERROR); 268 else 269 set_host_byte(sc, host_byte(result)); 270 set_status_byte(sc, result & 0xff); 271 scsi_set_resid(sc, ring_rsp->residual_len); 272 273 sense_len = min_t(uint8_t, VSCSIIF_SENSE_BUFFERSIZE, 274 ring_rsp->sense_len); 275 276 if (sense_len) 277 memcpy(sc->sense_buffer, ring_rsp->sense_buffer, sense_len); 278 279 sc->scsi_done(sc); 280 } 281 282 static void scsifront_sync_cmd_done(struct vscsifrnt_info *info, 283 struct vscsiif_response *ring_rsp) 284 { 285 uint16_t id = ring_rsp->rqid; 286 unsigned long flags; 287 struct vscsifrnt_shadow *shadow = info->shadow[id]; 288 int kick; 289 290 spin_lock_irqsave(&info->shadow_lock, flags); 291 shadow->wait_reset = 1; 292 switch (shadow->rslt_reset) { 293 case RSLT_RESET_WAITING: 294 shadow->rslt_reset = ring_rsp->rslt; 295 break; 296 case RSLT_RESET_ERR: 297 kick = _scsifront_put_rqid(info, id); 298 spin_unlock_irqrestore(&info->shadow_lock, flags); 299 kfree(shadow); 300 if (kick) 301 scsifront_wake_up(info); 302 return; 303 default: 304 shost_printk(KERN_ERR, info->host, KBUILD_MODNAME 305 "bad reset state %d, possibly leaking %u\n", 306 shadow->rslt_reset, id); 307 break; 308 } 309 spin_unlock_irqrestore(&info->shadow_lock, flags); 310 311 wake_up(&shadow->wq_reset); 312 } 313 314 static void scsifront_do_response(struct vscsifrnt_info *info, 315 struct vscsiif_response *ring_rsp) 316 { 317 if (WARN(ring_rsp->rqid >= VSCSIIF_MAX_REQS || 318 test_bit(ring_rsp->rqid, info->shadow_free_bitmap), 319 "illegal rqid %u returned by backend!\n", ring_rsp->rqid)) 320 return; 321 322 if (info->shadow[ring_rsp->rqid]->act == VSCSIIF_ACT_SCSI_CDB) 323 scsifront_cdb_cmd_done(info, ring_rsp); 324 else 325 scsifront_sync_cmd_done(info, ring_rsp); 326 } 327 328 static int scsifront_ring_drain(struct vscsifrnt_info *info) 329 { 330 struct vscsiif_response *ring_rsp; 331 RING_IDX i, rp; 332 int more_to_do = 0; 333 334 rp = info->ring.sring->rsp_prod; 335 rmb(); /* ordering required respective to dom0 */ 336 for (i = info->ring.rsp_cons; i != rp; i++) { 337 ring_rsp = RING_GET_RESPONSE(&info->ring, i); 338 scsifront_do_response(info, ring_rsp); 339 } 340 341 info->ring.rsp_cons = i; 342 343 if (i != info->ring.req_prod_pvt) 344 RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do); 345 else 346 info->ring.sring->rsp_event = i + 1; 347 348 return more_to_do; 349 } 350 351 static int scsifront_cmd_done(struct vscsifrnt_info *info) 352 { 353 int more_to_do; 354 unsigned long flags; 355 356 spin_lock_irqsave(info->host->host_lock, flags); 357 358 more_to_do = scsifront_ring_drain(info); 359 360 info->wait_ring_available = 0; 361 362 spin_unlock_irqrestore(info->host->host_lock, flags); 363 364 wake_up(&info->wq_sync); 365 366 return more_to_do; 367 } 368 369 static irqreturn_t scsifront_irq_fn(int irq, void *dev_id) 370 { 371 struct vscsifrnt_info *info = dev_id; 372 373 while (scsifront_cmd_done(info)) 374 /* Yield point for this unbounded loop. */ 375 cond_resched(); 376 377 return IRQ_HANDLED; 378 } 379 380 static void scsifront_finish_all(struct vscsifrnt_info *info) 381 { 382 unsigned i; 383 struct vscsiif_response resp; 384 385 scsifront_ring_drain(info); 386 387 for (i = 0; i < VSCSIIF_MAX_REQS; i++) { 388 if (test_bit(i, info->shadow_free_bitmap)) 389 continue; 390 resp.rqid = i; 391 resp.sense_len = 0; 392 resp.rslt = DID_RESET << 16; 393 resp.residual_len = 0; 394 scsifront_do_response(info, &resp); 395 } 396 } 397 398 static int map_data_for_request(struct vscsifrnt_info *info, 399 struct scsi_cmnd *sc, 400 struct vscsifrnt_shadow *shadow) 401 { 402 grant_ref_t gref_head; 403 struct page *page; 404 int err, ref, ref_cnt = 0; 405 int grant_ro = (sc->sc_data_direction == DMA_TO_DEVICE); 406 unsigned int i, off, len, bytes; 407 unsigned int data_len = scsi_bufflen(sc); 408 unsigned int data_grants = 0, seg_grants = 0; 409 struct scatterlist *sg; 410 struct scsiif_request_segment *seg; 411 412 if (sc->sc_data_direction == DMA_NONE || !data_len) 413 return 0; 414 415 scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i) 416 data_grants += PFN_UP(sg->offset + sg->length); 417 418 if (data_grants > VSCSIIF_SG_TABLESIZE) { 419 if (data_grants > info->host->sg_tablesize) { 420 shost_printk(KERN_ERR, info->host, KBUILD_MODNAME 421 "Unable to map request_buffer for command!\n"); 422 return -E2BIG; 423 } 424 seg_grants = vscsiif_grants_sg(data_grants); 425 shadow->sg = kcalloc(data_grants, 426 sizeof(struct scsiif_request_segment), GFP_ATOMIC); 427 if (!shadow->sg) 428 return -ENOMEM; 429 } 430 seg = shadow->sg ? : shadow->seg; 431 432 err = gnttab_alloc_grant_references(seg_grants + data_grants, 433 &gref_head); 434 if (err) { 435 kfree(shadow->sg); 436 shost_printk(KERN_ERR, info->host, KBUILD_MODNAME 437 "gnttab_alloc_grant_references() error\n"); 438 return -ENOMEM; 439 } 440 441 if (seg_grants) { 442 page = virt_to_page(seg); 443 off = offset_in_page(seg); 444 len = sizeof(struct scsiif_request_segment) * data_grants; 445 while (len > 0) { 446 bytes = min_t(unsigned int, len, PAGE_SIZE - off); 447 448 ref = gnttab_claim_grant_reference(&gref_head); 449 BUG_ON(ref == -ENOSPC); 450 451 gnttab_grant_foreign_access_ref(ref, 452 info->dev->otherend_id, 453 xen_page_to_gfn(page), 1); 454 shadow->gref[ref_cnt] = ref; 455 shadow->seg[ref_cnt].gref = ref; 456 shadow->seg[ref_cnt].offset = (uint16_t)off; 457 shadow->seg[ref_cnt].length = (uint16_t)bytes; 458 459 page++; 460 len -= bytes; 461 off = 0; 462 ref_cnt++; 463 } 464 BUG_ON(seg_grants < ref_cnt); 465 seg_grants = ref_cnt; 466 } 467 468 scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i) { 469 page = sg_page(sg); 470 off = sg->offset; 471 len = sg->length; 472 473 while (len > 0 && data_len > 0) { 474 /* 475 * sg sends a scatterlist that is larger than 476 * the data_len it wants transferred for certain 477 * IO sizes. 478 */ 479 bytes = min_t(unsigned int, len, PAGE_SIZE - off); 480 bytes = min(bytes, data_len); 481 482 ref = gnttab_claim_grant_reference(&gref_head); 483 BUG_ON(ref == -ENOSPC); 484 485 gnttab_grant_foreign_access_ref(ref, 486 info->dev->otherend_id, 487 xen_page_to_gfn(page), 488 grant_ro); 489 490 shadow->gref[ref_cnt] = ref; 491 seg->gref = ref; 492 seg->offset = (uint16_t)off; 493 seg->length = (uint16_t)bytes; 494 495 page++; 496 seg++; 497 len -= bytes; 498 data_len -= bytes; 499 off = 0; 500 ref_cnt++; 501 } 502 } 503 504 if (seg_grants) 505 shadow->nr_segments = VSCSIIF_SG_GRANT | seg_grants; 506 else 507 shadow->nr_segments = (uint8_t)ref_cnt; 508 shadow->nr_grants = ref_cnt; 509 510 return 0; 511 } 512 513 static int scsifront_enter(struct vscsifrnt_info *info) 514 { 515 if (info->pause) 516 return 1; 517 info->callers++; 518 return 0; 519 } 520 521 static void scsifront_return(struct vscsifrnt_info *info) 522 { 523 info->callers--; 524 if (info->callers) 525 return; 526 527 if (!info->waiting_pause) 528 return; 529 530 info->waiting_pause = 0; 531 wake_up(&info->wq_pause); 532 } 533 534 static int scsifront_queuecommand(struct Scsi_Host *shost, 535 struct scsi_cmnd *sc) 536 { 537 struct vscsifrnt_info *info = shost_priv(shost); 538 struct vscsifrnt_shadow *shadow = scsi_cmd_priv(sc); 539 unsigned long flags; 540 int err; 541 542 sc->result = 0; 543 544 shadow->sc = sc; 545 shadow->act = VSCSIIF_ACT_SCSI_CDB; 546 547 spin_lock_irqsave(shost->host_lock, flags); 548 if (scsifront_enter(info)) { 549 spin_unlock_irqrestore(shost->host_lock, flags); 550 return SCSI_MLQUEUE_HOST_BUSY; 551 } 552 553 err = map_data_for_request(info, sc, shadow); 554 if (err < 0) { 555 pr_debug("%s: err %d\n", __func__, err); 556 scsifront_return(info); 557 spin_unlock_irqrestore(shost->host_lock, flags); 558 if (err == -ENOMEM) 559 return SCSI_MLQUEUE_HOST_BUSY; 560 sc->result = DID_ERROR << 16; 561 sc->scsi_done(sc); 562 return 0; 563 } 564 565 if (scsifront_do_request(info, shadow)) { 566 scsifront_gnttab_done(info, shadow); 567 goto busy; 568 } 569 570 scsifront_return(info); 571 spin_unlock_irqrestore(shost->host_lock, flags); 572 573 return 0; 574 575 busy: 576 scsifront_return(info); 577 spin_unlock_irqrestore(shost->host_lock, flags); 578 pr_debug("%s: busy\n", __func__); 579 return SCSI_MLQUEUE_HOST_BUSY; 580 } 581 582 /* 583 * Any exception handling (reset or abort) must be forwarded to the backend. 584 * We have to wait until an answer is returned. This answer contains the 585 * result to be returned to the requestor. 586 */ 587 static int scsifront_action_handler(struct scsi_cmnd *sc, uint8_t act) 588 { 589 struct Scsi_Host *host = sc->device->host; 590 struct vscsifrnt_info *info = shost_priv(host); 591 struct vscsifrnt_shadow *shadow, *s = scsi_cmd_priv(sc); 592 int err = 0; 593 594 shadow = kzalloc(sizeof(*shadow), GFP_NOIO); 595 if (!shadow) 596 return FAILED; 597 598 shadow->act = act; 599 shadow->rslt_reset = RSLT_RESET_WAITING; 600 shadow->sc = sc; 601 shadow->ref_rqid = s->rqid; 602 init_waitqueue_head(&shadow->wq_reset); 603 604 spin_lock_irq(host->host_lock); 605 606 for (;;) { 607 if (scsifront_enter(info)) 608 goto fail; 609 610 if (!scsifront_do_request(info, shadow)) 611 break; 612 613 scsifront_return(info); 614 if (err) 615 goto fail; 616 info->wait_ring_available = 1; 617 spin_unlock_irq(host->host_lock); 618 err = wait_event_interruptible(info->wq_sync, 619 !info->wait_ring_available); 620 spin_lock_irq(host->host_lock); 621 } 622 623 spin_unlock_irq(host->host_lock); 624 err = wait_event_interruptible(shadow->wq_reset, shadow->wait_reset); 625 spin_lock_irq(host->host_lock); 626 627 if (!err) { 628 err = shadow->rslt_reset; 629 scsifront_put_rqid(info, shadow->rqid); 630 kfree(shadow); 631 } else { 632 spin_lock(&info->shadow_lock); 633 shadow->rslt_reset = RSLT_RESET_ERR; 634 spin_unlock(&info->shadow_lock); 635 err = FAILED; 636 } 637 638 scsifront_return(info); 639 spin_unlock_irq(host->host_lock); 640 return err; 641 642 fail: 643 spin_unlock_irq(host->host_lock); 644 kfree(shadow); 645 return FAILED; 646 } 647 648 static int scsifront_eh_abort_handler(struct scsi_cmnd *sc) 649 { 650 pr_debug("%s\n", __func__); 651 return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_ABORT); 652 } 653 654 static int scsifront_dev_reset_handler(struct scsi_cmnd *sc) 655 { 656 pr_debug("%s\n", __func__); 657 return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_RESET); 658 } 659 660 static int scsifront_sdev_configure(struct scsi_device *sdev) 661 { 662 struct vscsifrnt_info *info = shost_priv(sdev->host); 663 int err; 664 665 if (info && current == info->curr) { 666 err = xenbus_printf(XBT_NIL, info->dev->nodename, 667 info->dev_state_path, "%d", XenbusStateConnected); 668 if (err) { 669 xenbus_dev_error(info->dev, err, 670 "%s: writing dev_state_path", __func__); 671 return err; 672 } 673 } 674 675 return 0; 676 } 677 678 static void scsifront_sdev_destroy(struct scsi_device *sdev) 679 { 680 struct vscsifrnt_info *info = shost_priv(sdev->host); 681 int err; 682 683 if (info && current == info->curr) { 684 err = xenbus_printf(XBT_NIL, info->dev->nodename, 685 info->dev_state_path, "%d", XenbusStateClosed); 686 if (err) 687 xenbus_dev_error(info->dev, err, 688 "%s: writing dev_state_path", __func__); 689 } 690 } 691 692 static struct scsi_host_template scsifront_sht = { 693 .module = THIS_MODULE, 694 .name = "Xen SCSI frontend driver", 695 .queuecommand = scsifront_queuecommand, 696 .eh_abort_handler = scsifront_eh_abort_handler, 697 .eh_device_reset_handler = scsifront_dev_reset_handler, 698 .slave_configure = scsifront_sdev_configure, 699 .slave_destroy = scsifront_sdev_destroy, 700 .cmd_per_lun = VSCSIIF_DEFAULT_CMD_PER_LUN, 701 .can_queue = VSCSIIF_MAX_REQS, 702 .this_id = -1, 703 .cmd_size = sizeof(struct vscsifrnt_shadow), 704 .sg_tablesize = VSCSIIF_SG_TABLESIZE, 705 .proc_name = "scsifront", 706 }; 707 708 static int scsifront_alloc_ring(struct vscsifrnt_info *info) 709 { 710 struct xenbus_device *dev = info->dev; 711 struct vscsiif_sring *sring; 712 grant_ref_t gref; 713 int err = -ENOMEM; 714 715 /***** Frontend to Backend ring start *****/ 716 sring = (struct vscsiif_sring *)__get_free_page(GFP_KERNEL); 717 if (!sring) { 718 xenbus_dev_fatal(dev, err, 719 "fail to allocate shared ring (Front to Back)"); 720 return err; 721 } 722 SHARED_RING_INIT(sring); 723 FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE); 724 725 err = xenbus_grant_ring(dev, sring, 1, &gref); 726 if (err < 0) { 727 free_page((unsigned long)sring); 728 xenbus_dev_fatal(dev, err, 729 "fail to grant shared ring (Front to Back)"); 730 return err; 731 } 732 info->ring_ref = gref; 733 734 err = xenbus_alloc_evtchn(dev, &info->evtchn); 735 if (err) { 736 xenbus_dev_fatal(dev, err, "xenbus_alloc_evtchn"); 737 goto free_gnttab; 738 } 739 740 err = bind_evtchn_to_irq(info->evtchn); 741 if (err <= 0) { 742 xenbus_dev_fatal(dev, err, "bind_evtchn_to_irq"); 743 goto free_gnttab; 744 } 745 746 info->irq = err; 747 748 err = request_threaded_irq(info->irq, NULL, scsifront_irq_fn, 749 IRQF_ONESHOT, "scsifront", info); 750 if (err) { 751 xenbus_dev_fatal(dev, err, "request_threaded_irq"); 752 goto free_irq; 753 } 754 755 return 0; 756 757 /* free resource */ 758 free_irq: 759 unbind_from_irqhandler(info->irq, info); 760 free_gnttab: 761 gnttab_end_foreign_access(info->ring_ref, 0, 762 (unsigned long)info->ring.sring); 763 764 return err; 765 } 766 767 static void scsifront_free_ring(struct vscsifrnt_info *info) 768 { 769 unbind_from_irqhandler(info->irq, info); 770 gnttab_end_foreign_access(info->ring_ref, 0, 771 (unsigned long)info->ring.sring); 772 } 773 774 static int scsifront_init_ring(struct vscsifrnt_info *info) 775 { 776 struct xenbus_device *dev = info->dev; 777 struct xenbus_transaction xbt; 778 int err; 779 780 pr_debug("%s\n", __func__); 781 782 err = scsifront_alloc_ring(info); 783 if (err) 784 return err; 785 pr_debug("%s: %u %u\n", __func__, info->ring_ref, info->evtchn); 786 787 again: 788 err = xenbus_transaction_start(&xbt); 789 if (err) 790 xenbus_dev_fatal(dev, err, "starting transaction"); 791 792 err = xenbus_printf(xbt, dev->nodename, "ring-ref", "%u", 793 info->ring_ref); 794 if (err) { 795 xenbus_dev_fatal(dev, err, "%s", "writing ring-ref"); 796 goto fail; 797 } 798 799 err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u", 800 info->evtchn); 801 802 if (err) { 803 xenbus_dev_fatal(dev, err, "%s", "writing event-channel"); 804 goto fail; 805 } 806 807 err = xenbus_transaction_end(xbt, 0); 808 if (err) { 809 if (err == -EAGAIN) 810 goto again; 811 xenbus_dev_fatal(dev, err, "completing transaction"); 812 goto free_sring; 813 } 814 815 return 0; 816 817 fail: 818 xenbus_transaction_end(xbt, 1); 819 free_sring: 820 scsifront_free_ring(info); 821 822 return err; 823 } 824 825 826 static int scsifront_probe(struct xenbus_device *dev, 827 const struct xenbus_device_id *id) 828 { 829 struct vscsifrnt_info *info; 830 struct Scsi_Host *host; 831 int err = -ENOMEM; 832 char name[TASK_COMM_LEN]; 833 834 host = scsi_host_alloc(&scsifront_sht, sizeof(*info)); 835 if (!host) { 836 xenbus_dev_fatal(dev, err, "fail to allocate scsi host"); 837 return err; 838 } 839 info = (struct vscsifrnt_info *)host->hostdata; 840 841 dev_set_drvdata(&dev->dev, info); 842 info->dev = dev; 843 844 bitmap_fill(info->shadow_free_bitmap, VSCSIIF_MAX_REQS); 845 846 err = scsifront_init_ring(info); 847 if (err) { 848 scsi_host_put(host); 849 return err; 850 } 851 852 init_waitqueue_head(&info->wq_sync); 853 init_waitqueue_head(&info->wq_pause); 854 spin_lock_init(&info->shadow_lock); 855 856 snprintf(name, TASK_COMM_LEN, "vscsiif.%d", host->host_no); 857 858 host->max_id = VSCSIIF_MAX_TARGET; 859 host->max_channel = 0; 860 host->max_lun = VSCSIIF_MAX_LUN; 861 host->max_sectors = (host->sg_tablesize - 1) * PAGE_SIZE / 512; 862 host->max_cmd_len = VSCSIIF_MAX_COMMAND_SIZE; 863 864 err = scsi_add_host(host, &dev->dev); 865 if (err) { 866 dev_err(&dev->dev, "fail to add scsi host %d\n", err); 867 goto free_sring; 868 } 869 info->host = host; 870 info->host_active = 1; 871 872 xenbus_switch_state(dev, XenbusStateInitialised); 873 874 return 0; 875 876 free_sring: 877 scsifront_free_ring(info); 878 scsi_host_put(host); 879 return err; 880 } 881 882 static int scsifront_resume(struct xenbus_device *dev) 883 { 884 struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev); 885 struct Scsi_Host *host = info->host; 886 int err; 887 888 spin_lock_irq(host->host_lock); 889 890 /* Finish all still pending commands. */ 891 scsifront_finish_all(info); 892 893 spin_unlock_irq(host->host_lock); 894 895 /* Reconnect to dom0. */ 896 scsifront_free_ring(info); 897 err = scsifront_init_ring(info); 898 if (err) { 899 dev_err(&dev->dev, "fail to resume %d\n", err); 900 scsi_host_put(host); 901 return err; 902 } 903 904 xenbus_switch_state(dev, XenbusStateInitialised); 905 906 return 0; 907 } 908 909 static int scsifront_suspend(struct xenbus_device *dev) 910 { 911 struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev); 912 struct Scsi_Host *host = info->host; 913 int err = 0; 914 915 /* No new commands for the backend. */ 916 spin_lock_irq(host->host_lock); 917 info->pause = 1; 918 while (info->callers && !err) { 919 info->waiting_pause = 1; 920 info->wait_ring_available = 0; 921 spin_unlock_irq(host->host_lock); 922 wake_up(&info->wq_sync); 923 err = wait_event_interruptible(info->wq_pause, 924 !info->waiting_pause); 925 spin_lock_irq(host->host_lock); 926 } 927 spin_unlock_irq(host->host_lock); 928 return err; 929 } 930 931 static int scsifront_remove(struct xenbus_device *dev) 932 { 933 struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev); 934 935 pr_debug("%s: %s removed\n", __func__, dev->nodename); 936 937 mutex_lock(&scsifront_mutex); 938 if (info->host_active) { 939 /* Scsi_host not yet removed */ 940 scsi_remove_host(info->host); 941 info->host_active = 0; 942 } 943 mutex_unlock(&scsifront_mutex); 944 945 scsifront_free_ring(info); 946 scsi_host_put(info->host); 947 948 return 0; 949 } 950 951 static void scsifront_disconnect(struct vscsifrnt_info *info) 952 { 953 struct xenbus_device *dev = info->dev; 954 struct Scsi_Host *host = info->host; 955 956 pr_debug("%s: %s disconnect\n", __func__, dev->nodename); 957 958 /* 959 * When this function is executed, all devices of 960 * Frontend have been deleted. 961 * Therefore, it need not block I/O before remove_host. 962 */ 963 964 mutex_lock(&scsifront_mutex); 965 if (info->host_active) { 966 scsi_remove_host(host); 967 info->host_active = 0; 968 } 969 mutex_unlock(&scsifront_mutex); 970 971 xenbus_frontend_closed(dev); 972 } 973 974 static void scsifront_do_lun_hotplug(struct vscsifrnt_info *info, int op) 975 { 976 struct xenbus_device *dev = info->dev; 977 int i, err = 0; 978 char str[64]; 979 char **dir; 980 unsigned int dir_n = 0; 981 unsigned int device_state; 982 unsigned int hst, chn, tgt, lun; 983 struct scsi_device *sdev; 984 985 dir = xenbus_directory(XBT_NIL, dev->otherend, "vscsi-devs", &dir_n); 986 if (IS_ERR(dir)) 987 return; 988 989 /* mark current task as the one allowed to modify device states */ 990 BUG_ON(info->curr); 991 info->curr = current; 992 993 for (i = 0; i < dir_n; i++) { 994 /* read status */ 995 snprintf(str, sizeof(str), "vscsi-devs/%s/state", dir[i]); 996 err = xenbus_scanf(XBT_NIL, dev->otherend, str, "%u", 997 &device_state); 998 if (XENBUS_EXIST_ERR(err)) 999 continue; 1000 1001 /* virtual SCSI device */ 1002 snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]); 1003 err = xenbus_scanf(XBT_NIL, dev->otherend, str, 1004 "%u:%u:%u:%u", &hst, &chn, &tgt, &lun); 1005 if (XENBUS_EXIST_ERR(err)) 1006 continue; 1007 1008 /* 1009 * Front device state path, used in slave_configure called 1010 * on successfull scsi_add_device, and in slave_destroy called 1011 * on remove of a device. 1012 */ 1013 snprintf(info->dev_state_path, sizeof(info->dev_state_path), 1014 "vscsi-devs/%s/state", dir[i]); 1015 1016 switch (op) { 1017 case VSCSIFRONT_OP_ADD_LUN: 1018 if (device_state != XenbusStateInitialised) 1019 break; 1020 1021 if (scsi_add_device(info->host, chn, tgt, lun)) { 1022 dev_err(&dev->dev, "scsi_add_device\n"); 1023 err = xenbus_printf(XBT_NIL, dev->nodename, 1024 info->dev_state_path, 1025 "%d", XenbusStateClosed); 1026 if (err) 1027 xenbus_dev_error(dev, err, 1028 "%s: writing dev_state_path", __func__); 1029 } 1030 break; 1031 case VSCSIFRONT_OP_DEL_LUN: 1032 if (device_state != XenbusStateClosing) 1033 break; 1034 1035 sdev = scsi_device_lookup(info->host, chn, tgt, lun); 1036 if (sdev) { 1037 scsi_remove_device(sdev); 1038 scsi_device_put(sdev); 1039 } 1040 break; 1041 case VSCSIFRONT_OP_READD_LUN: 1042 if (device_state == XenbusStateConnected) { 1043 err = xenbus_printf(XBT_NIL, dev->nodename, 1044 info->dev_state_path, 1045 "%d", XenbusStateConnected); 1046 if (err) 1047 xenbus_dev_error(dev, err, 1048 "%s: writing dev_state_path", __func__); 1049 } 1050 break; 1051 default: 1052 break; 1053 } 1054 } 1055 1056 info->curr = NULL; 1057 1058 kfree(dir); 1059 } 1060 1061 static void scsifront_read_backend_params(struct xenbus_device *dev, 1062 struct vscsifrnt_info *info) 1063 { 1064 unsigned int sg_grant, nr_segs; 1065 struct Scsi_Host *host = info->host; 1066 1067 sg_grant = xenbus_read_unsigned(dev->otherend, "feature-sg-grant", 0); 1068 nr_segs = min_t(unsigned int, sg_grant, SG_ALL); 1069 nr_segs = max_t(unsigned int, nr_segs, VSCSIIF_SG_TABLESIZE); 1070 nr_segs = min_t(unsigned int, nr_segs, 1071 VSCSIIF_SG_TABLESIZE * PAGE_SIZE / 1072 sizeof(struct scsiif_request_segment)); 1073 1074 if (!info->pause && sg_grant) 1075 dev_info(&dev->dev, "using up to %d SG entries\n", nr_segs); 1076 else if (info->pause && nr_segs < host->sg_tablesize) 1077 dev_warn(&dev->dev, 1078 "SG entries decreased from %d to %u - device may not work properly anymore\n", 1079 host->sg_tablesize, nr_segs); 1080 1081 host->sg_tablesize = nr_segs; 1082 host->max_sectors = (nr_segs - 1) * PAGE_SIZE / 512; 1083 } 1084 1085 static void scsifront_backend_changed(struct xenbus_device *dev, 1086 enum xenbus_state backend_state) 1087 { 1088 struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev); 1089 1090 pr_debug("%s: %p %u %u\n", __func__, dev, dev->state, backend_state); 1091 1092 switch (backend_state) { 1093 case XenbusStateUnknown: 1094 case XenbusStateInitialising: 1095 case XenbusStateInitWait: 1096 case XenbusStateInitialised: 1097 break; 1098 1099 case XenbusStateConnected: 1100 scsifront_read_backend_params(dev, info); 1101 1102 if (info->pause) { 1103 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_READD_LUN); 1104 xenbus_switch_state(dev, XenbusStateConnected); 1105 info->pause = 0; 1106 return; 1107 } 1108 1109 if (xenbus_read_driver_state(dev->nodename) == 1110 XenbusStateInitialised) 1111 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN); 1112 1113 if (dev->state != XenbusStateConnected) 1114 xenbus_switch_state(dev, XenbusStateConnected); 1115 break; 1116 1117 case XenbusStateClosed: 1118 if (dev->state == XenbusStateClosed) 1119 break; 1120 fallthrough; /* Missed the backend's Closing state */ 1121 case XenbusStateClosing: 1122 scsifront_disconnect(info); 1123 break; 1124 1125 case XenbusStateReconfiguring: 1126 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_DEL_LUN); 1127 xenbus_switch_state(dev, XenbusStateReconfiguring); 1128 break; 1129 1130 case XenbusStateReconfigured: 1131 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN); 1132 xenbus_switch_state(dev, XenbusStateConnected); 1133 break; 1134 } 1135 } 1136 1137 static const struct xenbus_device_id scsifront_ids[] = { 1138 { "vscsi" }, 1139 { "" } 1140 }; 1141 1142 static struct xenbus_driver scsifront_driver = { 1143 .ids = scsifront_ids, 1144 .probe = scsifront_probe, 1145 .remove = scsifront_remove, 1146 .resume = scsifront_resume, 1147 .suspend = scsifront_suspend, 1148 .otherend_changed = scsifront_backend_changed, 1149 }; 1150 1151 static int __init scsifront_init(void) 1152 { 1153 if (!xen_domain()) 1154 return -ENODEV; 1155 1156 return xenbus_register_frontend(&scsifront_driver); 1157 } 1158 module_init(scsifront_init); 1159 1160 static void __exit scsifront_exit(void) 1161 { 1162 xenbus_unregister_driver(&scsifront_driver); 1163 } 1164 module_exit(scsifront_exit); 1165 1166 MODULE_DESCRIPTION("Xen SCSI frontend driver"); 1167 MODULE_LICENSE("GPL"); 1168 MODULE_ALIAS("xen:vscsi"); 1169 MODULE_AUTHOR("Juergen Gross <jgross@suse.com>"); 1170