drivers/scsi/xen-scsifront.c

   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         uint16_t rqid;
  83
  84         unsigned int nr_grants;         /* number of grants in gref[] */
  85         struct scsiif_request_segment *sg;      /* scatter/gather elements */
  86
  87         /* Do reset or abort function. */
  88         wait_queue_head_t wq_reset;     /* reset work queue           */
  89         int wait_reset;                 /* reset work queue condition */
  90         int32_t rslt_reset;             /* reset response status:     */
  91                                         /* SUCCESS or FAILED or:      */
  92 #define RSLT_RESET_WAITING      0
  93 #define RSLT_RESET_ERR          -1
  94
  95         /* Requested struct scsi_cmnd is stored from kernel. */
  96         struct scsi_cmnd *sc;
  97         int gref[vscsiif_grants_sg(SG_ALL) + SG_ALL];
  98 };
  99
 100 struct vscsifrnt_info {
 101         struct xenbus_device *dev;
 102
 103         struct Scsi_Host *host;
 104         int host_active;
 105
 106         unsigned int evtchn;
 107         unsigned int irq;
 108
 109         grant_ref_t ring_ref;
 110         struct vscsiif_front_ring ring;
 111         struct vscsiif_response ring_rsp;
 112
 113         spinlock_t shadow_lock;
 114         DECLARE_BITMAP(shadow_free_bitmap, VSCSIIF_MAX_REQS);
 115         struct vscsifrnt_shadow *shadow[VSCSIIF_MAX_REQS];
 116
 117         /* Following items are protected by the host lock. */
 118         wait_queue_head_t wq_sync;
 119         wait_queue_head_t wq_pause;
 120         unsigned int wait_ring_available:1;
 121         unsigned int waiting_pause:1;
 122         unsigned int pause:1;
 123         unsigned callers;
 124
 125         char dev_state_path[64];
 126         struct task_struct *curr;
 127 };
 128
 129 static DEFINE_MUTEX(scsifront_mutex);
 130
 131 static void scsifront_wake_up(struct vscsifrnt_info *info)
 132 {
 133         info->wait_ring_available = 0;
 134         wake_up(&info->wq_sync);
 135 }
 136
 137 static int scsifront_get_rqid(struct vscsifrnt_info *info)
 138 {
 139         unsigned long flags;
 140         int free;
 141
 142         spin_lock_irqsave(&info->shadow_lock, flags);
 143
 144         free = find_first_bit(info->shadow_free_bitmap, VSCSIIF_MAX_REQS);
 145         __clear_bit(free, info->shadow_free_bitmap);
 146
 147         spin_unlock_irqrestore(&info->shadow_lock, flags);
 148
 149         return free;
 150 }
 151
 152 static int _scsifront_put_rqid(struct vscsifrnt_info *info, uint32_t id)
 153 {
 154         int empty = bitmap_empty(info->shadow_free_bitmap, VSCSIIF_MAX_REQS);
 155
 156         __set_bit(id, info->shadow_free_bitmap);
 157         info->shadow[id] = NULL;
 158
 159         return empty || info->wait_ring_available;
 160 }
 161
 162 static void scsifront_put_rqid(struct vscsifrnt_info *info, uint32_t id)
 163 {
 164         unsigned long flags;
 165         int kick;
 166
 167         spin_lock_irqsave(&info->shadow_lock, flags);
 168         kick = _scsifront_put_rqid(info, id);
 169         spin_unlock_irqrestore(&info->shadow_lock, flags);
 170
 171         if (kick)
 172                 scsifront_wake_up(info);
 173 }
 174
 175 static struct vscsiif_request *scsifront_pre_req(struct vscsifrnt_info *info)
 176 {
 177         struct vscsiif_front_ring *ring = &(info->ring);
 178         struct vscsiif_request *ring_req;
 179         uint32_t id;
 180
 181         id = scsifront_get_rqid(info);  /* use id in response */
 182         if (id >= VSCSIIF_MAX_REQS)
 183                 return NULL;
 184
 185         ring_req = RING_GET_REQUEST(&(info->ring), ring->req_prod_pvt);
 186
 187         ring->req_prod_pvt++;
 188
 189         ring_req->rqid = (uint16_t)id;
 190
 191         return ring_req;
 192 }
 193
 194 static void scsifront_do_request(struct vscsifrnt_info *info)
 195 {
 196         struct vscsiif_front_ring *ring = &(info->ring);
 197         int notify;
 198
 199         RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(ring, notify);
 200         if (notify)
 201                 notify_remote_via_irq(info->irq);
 202 }
 203
 204 static void scsifront_gnttab_done(struct vscsifrnt_info *info, uint32_t id)
 205 {
 206         struct vscsifrnt_shadow *s = info->shadow[id];
 207         int i;
 208
 209         if (s->sc->sc_data_direction == DMA_NONE)
 210                 return;
 211
 212         for (i = 0; i < s->nr_grants; i++) {
 213                 if (unlikely(gnttab_query_foreign_access(s->gref[i]) != 0)) {
 214                         shost_printk(KERN_ALERT, info->host, KBUILD_MODNAME
 215                                      "grant still in use by backend\n");
 216                         BUG();
 217                 }
 218                 gnttab_end_foreign_access(s->gref[i], 0, 0UL);
 219         }
 220
 221         kfree(s->sg);
 222 }
 223
 224 static void scsifront_cdb_cmd_done(struct vscsifrnt_info *info,
 225                                    struct vscsiif_response *ring_rsp)
 226 {
 227         struct scsi_cmnd *sc;
 228         uint32_t id;
 229         uint8_t sense_len;
 230
 231         id = ring_rsp->rqid;
 232         sc = info->shadow[id]->sc;
 233
 234         BUG_ON(sc == NULL);
 235
 236         scsifront_gnttab_done(info, id);
 237         scsifront_put_rqid(info, id);
 238
 239         sc->result = ring_rsp->rslt;
 240         scsi_set_resid(sc, ring_rsp->residual_len);
 241
 242         sense_len = min_t(uint8_t, VSCSIIF_SENSE_BUFFERSIZE,
 243                           ring_rsp->sense_len);
 244
 245         if (sense_len)
 246                 memcpy(sc->sense_buffer, ring_rsp->sense_buffer, sense_len);
 247
 248         sc->scsi_done(sc);
 249 }
 250
 251 static void scsifront_sync_cmd_done(struct vscsifrnt_info *info,
 252                                     struct vscsiif_response *ring_rsp)
 253 {
 254         uint16_t id = ring_rsp->rqid;
 255         unsigned long flags;
 256         struct vscsifrnt_shadow *shadow = info->shadow[id];
 257         int kick;
 258
 259         spin_lock_irqsave(&info->shadow_lock, flags);
 260         shadow->wait_reset = 1;
 261         switch (shadow->rslt_reset) {
 262         case RSLT_RESET_WAITING:
 263                 shadow->rslt_reset = ring_rsp->rslt;
 264                 break;
 265         case RSLT_RESET_ERR:
 266                 kick = _scsifront_put_rqid(info, id);
 267                 spin_unlock_irqrestore(&info->shadow_lock, flags);
 268                 kfree(shadow);
 269                 if (kick)
 270                         scsifront_wake_up(info);
 271                 return;
 272         default:
 273                 shost_printk(KERN_ERR, info->host, KBUILD_MODNAME
 274                              "bad reset state %d, possibly leaking %u\n",
 275                              shadow->rslt_reset, id);
 276                 break;
 277         }
 278         spin_unlock_irqrestore(&info->shadow_lock, flags);
 279
 280         wake_up(&shadow->wq_reset);
 281 }
 282
 283 static void scsifront_do_response(struct vscsifrnt_info *info,
 284                                   struct vscsiif_response *ring_rsp)
 285 {
 286         if (WARN(ring_rsp->rqid >= VSCSIIF_MAX_REQS ||
 287                  test_bit(ring_rsp->rqid, info->shadow_free_bitmap),
 288                  "illegal rqid %u returned by backend!\n", ring_rsp->rqid))
 289                 return;
 290
 291         if (info->shadow[ring_rsp->rqid]->act == VSCSIIF_ACT_SCSI_CDB)
 292                 scsifront_cdb_cmd_done(info, ring_rsp);
 293         else
 294                 scsifront_sync_cmd_done(info, ring_rsp);
 295 }
 296
 297 static int scsifront_ring_drain(struct vscsifrnt_info *info)
 298 {
 299         struct vscsiif_response *ring_rsp;
 300         RING_IDX i, rp;
 301         int more_to_do = 0;
 302
 303         rp = info->ring.sring->rsp_prod;
 304         rmb();  /* ordering required respective to dom0 */
 305         for (i = info->ring.rsp_cons; i != rp; i++) {
 306                 ring_rsp = RING_GET_RESPONSE(&info->ring, i);
 307                 scsifront_do_response(info, ring_rsp);
 308         }
 309
 310         info->ring.rsp_cons = i;
 311
 312         if (i != info->ring.req_prod_pvt)
 313                 RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
 314         else
 315                 info->ring.sring->rsp_event = i + 1;
 316
 317         return more_to_do;
 318 }
 319
 320 static int scsifront_cmd_done(struct vscsifrnt_info *info)
 321 {
 322         int more_to_do;
 323         unsigned long flags;
 324
 325         spin_lock_irqsave(info->host->host_lock, flags);
 326
 327         more_to_do = scsifront_ring_drain(info);
 328
 329         info->wait_ring_available = 0;
 330
 331         spin_unlock_irqrestore(info->host->host_lock, flags);
 332
 333         wake_up(&info->wq_sync);
 334
 335         return more_to_do;
 336 }
 337
 338 static irqreturn_t scsifront_irq_fn(int irq, void *dev_id)
 339 {
 340         struct vscsifrnt_info *info = dev_id;
 341
 342         while (scsifront_cmd_done(info))
 343                 /* Yield point for this unbounded loop. */
 344                 cond_resched();
 345
 346         return IRQ_HANDLED;
 347 }
 348
 349 static void scsifront_finish_all(struct vscsifrnt_info *info)
 350 {
 351         unsigned i;
 352         struct vscsiif_response resp;
 353
 354         scsifront_ring_drain(info);
 355
 356         for (i = 0; i < VSCSIIF_MAX_REQS; i++) {
 357                 if (test_bit(i, info->shadow_free_bitmap))
 358                         continue;
 359                 resp.rqid = i;
 360                 resp.sense_len = 0;
 361                 resp.rslt = DID_RESET << 16;
 362                 resp.residual_len = 0;
 363                 scsifront_do_response(info, &resp);
 364         }
 365 }
 366
 367 static int map_data_for_request(struct vscsifrnt_info *info,
 368                                 struct scsi_cmnd *sc,
 369                                 struct vscsiif_request *ring_req,
 370                                 struct vscsifrnt_shadow *shadow)
 371 {
 372         grant_ref_t gref_head;
 373         struct page *page;
 374         int err, ref, ref_cnt = 0;
 375         int grant_ro = (sc->sc_data_direction == DMA_TO_DEVICE);
 376         unsigned int i, off, len, bytes;
 377         unsigned int data_len = scsi_bufflen(sc);
 378         unsigned int data_grants = 0, seg_grants = 0;
 379         struct scatterlist *sg;
 380         struct scsiif_request_segment *seg;
 381
 382         ring_req->nr_segments = 0;
 383         if (sc->sc_data_direction == DMA_NONE || !data_len)
 384                 return 0;
 385
 386         scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i)
 387                 data_grants += PFN_UP(sg->offset + sg->length);
 388
 389         if (data_grants > VSCSIIF_SG_TABLESIZE) {
 390                 if (data_grants > info->host->sg_tablesize) {
 391                         shost_printk(KERN_ERR, info->host, KBUILD_MODNAME
 392                              "Unable to map request_buffer for command!\n");
 393                         return -E2BIG;
 394                 }
 395                 seg_grants = vscsiif_grants_sg(data_grants);
 396                 shadow->sg = kcalloc(data_grants,
 397                         sizeof(struct scsiif_request_segment), GFP_ATOMIC);
 398                 if (!shadow->sg)
 399                         return -ENOMEM;
 400         }
 401         seg = shadow->sg ? : ring_req->seg;
 402
 403         err = gnttab_alloc_grant_references(seg_grants + data_grants,
 404                                             &gref_head);
 405         if (err) {
 406                 kfree(shadow->sg);
 407                 shost_printk(KERN_ERR, info->host, KBUILD_MODNAME
 408                              "gnttab_alloc_grant_references() error\n");
 409                 return -ENOMEM;
 410         }
 411
 412         if (seg_grants) {
 413                 page = virt_to_page(seg);
 414                 off = (unsigned long)seg & ~PAGE_MASK;
 415                 len = sizeof(struct scsiif_request_segment) * data_grants;
 416                 while (len > 0) {
 417                         bytes = min_t(unsigned int, len, PAGE_SIZE - off);
 418
 419                         ref = gnttab_claim_grant_reference(&gref_head);
 420                         BUG_ON(ref == -ENOSPC);
 421
 422                         gnttab_grant_foreign_access_ref(ref,
 423                                 info->dev->otherend_id,
 424                                 xen_page_to_gfn(page), 1);
 425                         shadow->gref[ref_cnt] = ref;
 426                         ring_req->seg[ref_cnt].gref   = ref;
 427                         ring_req->seg[ref_cnt].offset = (uint16_t)off;
 428                         ring_req->seg[ref_cnt].length = (uint16_t)bytes;
 429
 430                         page++;
 431                         len -= bytes;
 432                         off = 0;
 433                         ref_cnt++;
 434                 }
 435                 BUG_ON(seg_grants < ref_cnt);
 436                 seg_grants = ref_cnt;
 437         }
 438
 439         scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i) {
 440                 page = sg_page(sg);
 441                 off = sg->offset;
 442                 len = sg->length;
 443
 444                 while (len > 0 && data_len > 0) {
 445                         /*
 446                          * sg sends a scatterlist that is larger than
 447                          * the data_len it wants transferred for certain
 448                          * IO sizes.
 449                          */
 450                         bytes = min_t(unsigned int, len, PAGE_SIZE - off);
 451                         bytes = min(bytes, data_len);
 452
 453                         ref = gnttab_claim_grant_reference(&gref_head);
 454                         BUG_ON(ref == -ENOSPC);
 455
 456                         gnttab_grant_foreign_access_ref(ref,
 457                                 info->dev->otherend_id,
 458                                 xen_page_to_gfn(page),
 459                                 grant_ro);
 460
 461                         shadow->gref[ref_cnt] = ref;
 462                         seg->gref   = ref;
 463                         seg->offset = (uint16_t)off;
 464                         seg->length = (uint16_t)bytes;
 465
 466                         page++;
 467                         seg++;
 468                         len -= bytes;
 469                         data_len -= bytes;
 470                         off = 0;
 471                         ref_cnt++;
 472                 }
 473         }
 474
 475         if (seg_grants)
 476                 ring_req->nr_segments = VSCSIIF_SG_GRANT | seg_grants;
 477         else
 478                 ring_req->nr_segments = (uint8_t)ref_cnt;
 479         shadow->nr_grants = ref_cnt;
 480
 481         return 0;
 482 }
 483
 484 static struct vscsiif_request *scsifront_command2ring(
 485                 struct vscsifrnt_info *info, struct scsi_cmnd *sc,
 486                 struct vscsifrnt_shadow *shadow)
 487 {
 488         struct vscsiif_request *ring_req;
 489
 490         memset(shadow, 0, sizeof(*shadow));
 491
 492         ring_req = scsifront_pre_req(info);
 493         if (!ring_req)
 494                 return NULL;
 495
 496         info->shadow[ring_req->rqid] = shadow;
 497         shadow->rqid = ring_req->rqid;
 498
 499         ring_req->id      = sc->device->id;
 500         ring_req->lun     = sc->device->lun;
 501         ring_req->channel = sc->device->channel;
 502         ring_req->cmd_len = sc->cmd_len;
 503
 504         BUG_ON(sc->cmd_len > VSCSIIF_MAX_COMMAND_SIZE);
 505
 506         memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len);
 507
 508         ring_req->sc_data_direction   = (uint8_t)sc->sc_data_direction;
 509         ring_req->timeout_per_command = sc->request->timeout / HZ;
 510
 511         return ring_req;
 512 }
 513
 514 static int scsifront_enter(struct vscsifrnt_info *info)
 515 {
 516         if (info->pause)
 517                 return 1;
 518         info->callers++;
 519         return 0;
 520 }
 521
 522 static void scsifront_return(struct vscsifrnt_info *info)
 523 {
 524         info->callers--;
 525         if (info->callers)
 526                 return;
 527
 528         if (!info->waiting_pause)
 529                 return;
 530
 531         info->waiting_pause = 0;
 532         wake_up(&info->wq_pause);
 533 }
 534
 535 static int scsifront_queuecommand(struct Scsi_Host *shost,
 536                                   struct scsi_cmnd *sc)
 537 {
 538         struct vscsifrnt_info *info = shost_priv(shost);
 539         struct vscsiif_request *ring_req;
 540         struct vscsifrnt_shadow *shadow = scsi_cmd_priv(sc);
 541         unsigned long flags;
 542         int err;
 543         uint16_t rqid;
 544
 545         spin_lock_irqsave(shost->host_lock, flags);
 546         if (scsifront_enter(info)) {
 547                 spin_unlock_irqrestore(shost->host_lock, flags);
 548                 return SCSI_MLQUEUE_HOST_BUSY;
 549         }
 550         if (RING_FULL(&info->ring))
 551                 goto busy;
 552
 553         ring_req = scsifront_command2ring(info, sc, shadow);
 554         if (!ring_req)
 555                 goto busy;
 556
 557         sc->result = 0;
 558
 559         rqid = ring_req->rqid;
 560         ring_req->act = VSCSIIF_ACT_SCSI_CDB;
 561
 562         shadow->sc  = sc;
 563         shadow->act = VSCSIIF_ACT_SCSI_CDB;
 564
 565         err = map_data_for_request(info, sc, ring_req, shadow);
 566         if (err < 0) {
 567                 pr_debug("%s: err %d\n", __func__, err);
 568                 scsifront_put_rqid(info, rqid);
 569                 scsifront_return(info);
 570                 spin_unlock_irqrestore(shost->host_lock, flags);
 571                 if (err == -ENOMEM)
 572                         return SCSI_MLQUEUE_HOST_BUSY;
 573                 sc->result = DID_ERROR << 16;
 574                 sc->scsi_done(sc);
 575                 return 0;
 576         }
 577
 578         scsifront_do_request(info);
 579         scsifront_return(info);
 580         spin_unlock_irqrestore(shost->host_lock, flags);
 581
 582         return 0;
 583
 584 busy:
 585         scsifront_return(info);
 586         spin_unlock_irqrestore(shost->host_lock, flags);
 587         pr_debug("%s: busy\n", __func__);
 588         return SCSI_MLQUEUE_HOST_BUSY;
 589 }
 590
 591 /*
 592  * Any exception handling (reset or abort) must be forwarded to the backend.
 593  * We have to wait until an answer is returned. This answer contains the
 594  * result to be returned to the requestor.
 595  */
 596 static int scsifront_action_handler(struct scsi_cmnd *sc, uint8_t act)
 597 {
 598         struct Scsi_Host *host = sc->device->host;
 599         struct vscsifrnt_info *info = shost_priv(host);
 600         struct vscsifrnt_shadow *shadow, *s = scsi_cmd_priv(sc);
 601         struct vscsiif_request *ring_req;
 602         int err = 0;
 603
 604         shadow = kmalloc(sizeof(*shadow), GFP_NOIO);
 605         if (!shadow)
 606                 return FAILED;
 607
 608         spin_lock_irq(host->host_lock);
 609
 610         for (;;) {
 611                 if (!RING_FULL(&info->ring)) {
 612                         ring_req = scsifront_command2ring(info, sc, shadow);
 613                         if (ring_req)
 614                                 break;
 615                 }
 616                 if (err || info->pause) {
 617                         spin_unlock_irq(host->host_lock);
 618                         kfree(shadow);
 619                         return FAILED;
 620                 }
 621                 info->wait_ring_available = 1;
 622                 spin_unlock_irq(host->host_lock);
 623                 err = wait_event_interruptible(info->wq_sync,
 624                                                !info->wait_ring_available);
 625                 spin_lock_irq(host->host_lock);
 626         }
 627
 628         if (scsifront_enter(info)) {
 629                 spin_unlock_irq(host->host_lock);
 630                 return FAILED;
 631         }
 632
 633         ring_req->act = act;
 634         ring_req->ref_rqid = s->rqid;
 635
 636         shadow->act = act;
 637         shadow->rslt_reset = RSLT_RESET_WAITING;
 638         init_waitqueue_head(&shadow->wq_reset);
 639
 640         ring_req->nr_segments = 0;
 641
 642         scsifront_do_request(info);
 643
 644         spin_unlock_irq(host->host_lock);
 645         err = wait_event_interruptible(shadow->wq_reset, shadow->wait_reset);
 646         spin_lock_irq(host->host_lock);
 647
 648         if (!err) {
 649                 err = shadow->rslt_reset;
 650                 scsifront_put_rqid(info, shadow->rqid);
 651                 kfree(shadow);
 652         } else {
 653                 spin_lock(&info->shadow_lock);
 654                 shadow->rslt_reset = RSLT_RESET_ERR;
 655                 spin_unlock(&info->shadow_lock);
 656                 err = FAILED;
 657         }
 658
 659         scsifront_return(info);
 660         spin_unlock_irq(host->host_lock);
 661         return err;
 662 }
 663
 664 static int scsifront_eh_abort_handler(struct scsi_cmnd *sc)
 665 {
 666         pr_debug("%s\n", __func__);
 667         return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_ABORT);
 668 }
 669
 670 static int scsifront_dev_reset_handler(struct scsi_cmnd *sc)
 671 {
 672         pr_debug("%s\n", __func__);
 673         return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_RESET);
 674 }
 675
 676 static int scsifront_sdev_configure(struct scsi_device *sdev)
 677 {
 678         struct vscsifrnt_info *info = shost_priv(sdev->host);
 679
 680         if (info && current == info->curr)
 681                 xenbus_printf(XBT_NIL, info->dev->nodename,
 682                               info->dev_state_path, "%d", XenbusStateConnected);
 683
 684         return 0;
 685 }
 686
 687 static void scsifront_sdev_destroy(struct scsi_device *sdev)
 688 {
 689         struct vscsifrnt_info *info = shost_priv(sdev->host);
 690
 691         if (info && current == info->curr)
 692                 xenbus_printf(XBT_NIL, info->dev->nodename,
 693                               info->dev_state_path, "%d", XenbusStateClosed);
 694 }
 695
 696 static struct scsi_host_template scsifront_sht = {
 697         .module                 = THIS_MODULE,
 698         .name                   = "Xen SCSI frontend driver",
 699         .queuecommand           = scsifront_queuecommand,
 700         .eh_abort_handler       = scsifront_eh_abort_handler,
 701         .eh_device_reset_handler = scsifront_dev_reset_handler,
 702         .slave_configure        = scsifront_sdev_configure,
 703         .slave_destroy          = scsifront_sdev_destroy,
 704         .cmd_per_lun            = VSCSIIF_DEFAULT_CMD_PER_LUN,
 705         .can_queue              = VSCSIIF_MAX_REQS,
 706         .this_id                = -1,
 707         .cmd_size               = sizeof(struct vscsifrnt_shadow),
 708         .sg_tablesize           = VSCSIIF_SG_TABLESIZE,
 709         .use_clustering         = DISABLE_CLUSTERING,
 710         .proc_name              = "scsifront",
 711 };
 712
 713 static int scsifront_alloc_ring(struct vscsifrnt_info *info)
 714 {
 715         struct xenbus_device *dev = info->dev;
 716         struct vscsiif_sring *sring;
 717         grant_ref_t gref;
 718         int err = -ENOMEM;
 719
 720         /***** Frontend to Backend ring start *****/
 721         sring = (struct vscsiif_sring *)__get_free_page(GFP_KERNEL);
 722         if (!sring) {
 723                 xenbus_dev_fatal(dev, err,
 724                         "fail to allocate shared ring (Front to Back)");
 725                 return err;
 726         }
 727         SHARED_RING_INIT(sring);
 728         FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
 729
 730         err = xenbus_grant_ring(dev, sring, 1, &gref);
 731         if (err < 0) {
 732                 free_page((unsigned long)sring);
 733                 xenbus_dev_fatal(dev, err,
 734                         "fail to grant shared ring (Front to Back)");
 735                 return err;
 736         }
 737         info->ring_ref = gref;
 738
 739         err = xenbus_alloc_evtchn(dev, &info->evtchn);
 740         if (err) {
 741                 xenbus_dev_fatal(dev, err, "xenbus_alloc_evtchn");
 742                 goto free_gnttab;
 743         }
 744
 745         err = bind_evtchn_to_irq(info->evtchn);
 746         if (err <= 0) {
 747                 xenbus_dev_fatal(dev, err, "bind_evtchn_to_irq");
 748                 goto free_gnttab;
 749         }
 750
 751         info->irq = err;
 752
 753         err = request_threaded_irq(info->irq, NULL, scsifront_irq_fn,
 754                                    IRQF_ONESHOT, "scsifront", info);
 755         if (err) {
 756                 xenbus_dev_fatal(dev, err, "request_threaded_irq");
 757                 goto free_irq;
 758         }
 759
 760         return 0;
 761
 762 /* free resource */
 763 free_irq:
 764         unbind_from_irqhandler(info->irq, info);
 765 free_gnttab:
 766         gnttab_end_foreign_access(info->ring_ref, 0,
 767                                   (unsigned long)info->ring.sring);
 768
 769         return err;
 770 }
 771
 772 static void scsifront_free_ring(struct vscsifrnt_info *info)
 773 {
 774         unbind_from_irqhandler(info->irq, info);
 775         gnttab_end_foreign_access(info->ring_ref, 0,
 776                                   (unsigned long)info->ring.sring);
 777 }
 778
 779 static int scsifront_init_ring(struct vscsifrnt_info *info)
 780 {
 781         struct xenbus_device *dev = info->dev;
 782         struct xenbus_transaction xbt;
 783         int err;
 784
 785         pr_debug("%s\n", __func__);
 786
 787         err = scsifront_alloc_ring(info);
 788         if (err)
 789                 return err;
 790         pr_debug("%s: %u %u\n", __func__, info->ring_ref, info->evtchn);
 791
 792 again:
 793         err = xenbus_transaction_start(&xbt);
 794         if (err)
 795                 xenbus_dev_fatal(dev, err, "starting transaction");
 796
 797         err = xenbus_printf(xbt, dev->nodename, "ring-ref", "%u",
 798                             info->ring_ref);
 799         if (err) {
 800                 xenbus_dev_fatal(dev, err, "%s", "writing ring-ref");
 801                 goto fail;
 802         }
 803
 804         err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
 805                             info->evtchn);
 806
 807         if (err) {
 808                 xenbus_dev_fatal(dev, err, "%s", "writing event-channel");
 809                 goto fail;
 810         }
 811
 812         err = xenbus_transaction_end(xbt, 0);
 813         if (err) {
 814                 if (err == -EAGAIN)
 815                         goto again;
 816                 xenbus_dev_fatal(dev, err, "completing transaction");
 817                 goto free_sring;
 818         }
 819
 820         return 0;
 821
 822 fail:
 823         xenbus_transaction_end(xbt, 1);
 824 free_sring:
 825         scsifront_free_ring(info);
 826
 827         return err;
 828 }
 829
 830
 831 static int scsifront_probe(struct xenbus_device *dev,
 832                            const struct xenbus_device_id *id)
 833 {
 834         struct vscsifrnt_info *info;
 835         struct Scsi_Host *host;
 836         int err = -ENOMEM;
 837         char name[TASK_COMM_LEN];
 838
 839         host = scsi_host_alloc(&scsifront_sht, sizeof(*info));
 840         if (!host) {
 841                 xenbus_dev_fatal(dev, err, "fail to allocate scsi host");
 842                 return err;
 843         }
 844         info = (struct vscsifrnt_info *)host->hostdata;
 845
 846         dev_set_drvdata(&dev->dev, info);
 847         info->dev = dev;
 848
 849         bitmap_fill(info->shadow_free_bitmap, VSCSIIF_MAX_REQS);
 850
 851         err = scsifront_init_ring(info);
 852         if (err) {
 853                 scsi_host_put(host);
 854                 return err;
 855         }
 856
 857         init_waitqueue_head(&info->wq_sync);
 858         init_waitqueue_head(&info->wq_pause);
 859         spin_lock_init(&info->shadow_lock);
 860
 861         snprintf(name, TASK_COMM_LEN, "vscsiif.%d", host->host_no);
 862
 863         host->max_id      = VSCSIIF_MAX_TARGET;
 864         host->max_channel = 0;
 865         host->max_lun     = VSCSIIF_MAX_LUN;
 866         host->max_sectors = (host->sg_tablesize - 1) * PAGE_SIZE / 512;
 867         host->max_cmd_len = VSCSIIF_MAX_COMMAND_SIZE;
 868
 869         err = scsi_add_host(host, &dev->dev);
 870         if (err) {
 871                 dev_err(&dev->dev, "fail to add scsi host %d\n", err);
 872                 goto free_sring;
 873         }
 874         info->host = host;
 875         info->host_active = 1;
 876
 877         xenbus_switch_state(dev, XenbusStateInitialised);
 878
 879         return 0;
 880
 881 free_sring:
 882         scsifront_free_ring(info);
 883         scsi_host_put(host);
 884         return err;
 885 }
 886
 887 static int scsifront_resume(struct xenbus_device *dev)
 888 {
 889         struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
 890         struct Scsi_Host *host = info->host;
 891         int err;
 892
 893         spin_lock_irq(host->host_lock);
 894
 895         /* Finish all still pending commands. */
 896         scsifront_finish_all(info);
 897
 898         spin_unlock_irq(host->host_lock);
 899
 900         /* Reconnect to dom0. */
 901         scsifront_free_ring(info);
 902         err = scsifront_init_ring(info);
 903         if (err) {
 904                 dev_err(&dev->dev, "fail to resume %d\n", err);
 905                 scsi_host_put(host);
 906                 return err;
 907         }
 908
 909         xenbus_switch_state(dev, XenbusStateInitialised);
 910
 911         return 0;
 912 }
 913
 914 static int scsifront_suspend(struct xenbus_device *dev)
 915 {
 916         struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
 917         struct Scsi_Host *host = info->host;
 918         int err = 0;
 919
 920         /* No new commands for the backend. */
 921         spin_lock_irq(host->host_lock);
 922         info->pause = 1;
 923         while (info->callers && !err) {
 924                 info->waiting_pause = 1;
 925                 info->wait_ring_available = 0;
 926                 spin_unlock_irq(host->host_lock);
 927                 wake_up(&info->wq_sync);
 928                 err = wait_event_interruptible(info->wq_pause,
 929                                                !info->waiting_pause);
 930                 spin_lock_irq(host->host_lock);
 931         }
 932         spin_unlock_irq(host->host_lock);
 933         return err;
 934 }
 935
 936 static int scsifront_remove(struct xenbus_device *dev)
 937 {
 938         struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
 939
 940         pr_debug("%s: %s removed\n", __func__, dev->nodename);
 941
 942         mutex_lock(&scsifront_mutex);
 943         if (info->host_active) {
 944                 /* Scsi_host not yet removed */
 945                 scsi_remove_host(info->host);
 946                 info->host_active = 0;
 947         }
 948         mutex_unlock(&scsifront_mutex);
 949
 950         scsifront_free_ring(info);
 951         scsi_host_put(info->host);
 952
 953         return 0;
 954 }
 955
 956 static void scsifront_disconnect(struct vscsifrnt_info *info)
 957 {
 958         struct xenbus_device *dev = info->dev;
 959         struct Scsi_Host *host = info->host;
 960
 961         pr_debug("%s: %s disconnect\n", __func__, dev->nodename);
 962
 963         /*
 964          * When this function is executed, all devices of
 965          * Frontend have been deleted.
 966          * Therefore, it need not block I/O before remove_host.
 967          */
 968
 969         mutex_lock(&scsifront_mutex);
 970         if (info->host_active) {
 971                 scsi_remove_host(host);
 972                 info->host_active = 0;
 973         }
 974         mutex_unlock(&scsifront_mutex);
 975
 976         xenbus_frontend_closed(dev);
 977 }
 978
 979 static void scsifront_do_lun_hotplug(struct vscsifrnt_info *info, int op)
 980 {
 981         struct xenbus_device *dev = info->dev;
 982         int i, err = 0;
 983         char str[64];
 984         char **dir;
 985         unsigned int dir_n = 0;
 986         unsigned int device_state;
 987         unsigned int hst, chn, tgt, lun;
 988         struct scsi_device *sdev;
 989
 990         dir = xenbus_directory(XBT_NIL, dev->otherend, "vscsi-devs", &dir_n);
 991         if (IS_ERR(dir))
 992                 return;
 993
 994         /* mark current task as the one allowed to modify device states */
 995         BUG_ON(info->curr);
 996         info->curr = current;
 997
 998         for (i = 0; i < dir_n; i++) {
 999                 /* read status */
1000                 snprintf(str, sizeof(str), "vscsi-devs/%s/state", dir[i]);
1001                 err = xenbus_scanf(XBT_NIL, dev->otherend, str, "%u",
1002                                    &device_state);
1003                 if (XENBUS_EXIST_ERR(err))
1004                         continue;
1005
1006                 /* virtual SCSI device */
1007                 snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]);
1008                 err = xenbus_scanf(XBT_NIL, dev->otherend, str,
1009                                    "%u:%u:%u:%u", &hst, &chn, &tgt, &lun);
1010                 if (XENBUS_EXIST_ERR(err))
1011                         continue;
1012
1013                 /*
1014                  * Front device state path, used in slave_configure called
1015                  * on successfull scsi_add_device, and in slave_destroy called
1016                  * on remove of a device.
1017                  */
1018                 snprintf(info->dev_state_path, sizeof(info->dev_state_path),
1019                          "vscsi-devs/%s/state", dir[i]);
1020
1021                 switch (op) {
1022                 case VSCSIFRONT_OP_ADD_LUN:
1023                         if (device_state != XenbusStateInitialised)
1024                                 break;
1025
1026                         if (scsi_add_device(info->host, chn, tgt, lun)) {
1027                                 dev_err(&dev->dev, "scsi_add_device\n");
1028                                 xenbus_printf(XBT_NIL, dev->nodename,
1029                                               info->dev_state_path,
1030                                               "%d", XenbusStateClosed);
1031                         }
1032                         break;
1033                 case VSCSIFRONT_OP_DEL_LUN:
1034                         if (device_state != XenbusStateClosing)
1035                                 break;
1036
1037                         sdev = scsi_device_lookup(info->host, chn, tgt, lun);
1038                         if (sdev) {
1039                                 scsi_remove_device(sdev);
1040                                 scsi_device_put(sdev);
1041                         }
1042                         break;
1043                 case VSCSIFRONT_OP_READD_LUN:
1044                         if (device_state == XenbusStateConnected)
1045                                 xenbus_printf(XBT_NIL, dev->nodename,
1046                                               info->dev_state_path,
1047                                               "%d", XenbusStateConnected);
1048                         break;
1049                 default:
1050                         break;
1051                 }
1052         }
1053
1054         info->curr = NULL;
1055
1056         kfree(dir);
1057 }
1058
1059 static void scsifront_read_backend_params(struct xenbus_device *dev,
1060                                           struct vscsifrnt_info *info)
1061 {
1062         unsigned int sg_grant, nr_segs;
1063         int ret;
1064         struct Scsi_Host *host = info->host;
1065
1066         ret = xenbus_scanf(XBT_NIL, dev->otherend, "feature-sg-grant", "%u",
1067                            &sg_grant);
1068         if (ret != 1)
1069                 sg_grant = 0;
1070         nr_segs = min_t(unsigned int, sg_grant, SG_ALL);
1071         nr_segs = max_t(unsigned int, nr_segs, VSCSIIF_SG_TABLESIZE);
1072         nr_segs = min_t(unsigned int, nr_segs,
1073                         VSCSIIF_SG_TABLESIZE * PAGE_SIZE /
1074                         sizeof(struct scsiif_request_segment));
1075
1076         if (!info->pause && sg_grant)
1077                 dev_info(&dev->dev, "using up to %d SG entries\n", nr_segs);
1078         else if (info->pause && nr_segs < host->sg_tablesize)
1079                 dev_warn(&dev->dev,
1080                          "SG entries decreased from %d to %u - device may not work properly anymore\n",
1081                          host->sg_tablesize, nr_segs);
1082
1083         host->sg_tablesize = nr_segs;
1084         host->max_sectors = (nr_segs - 1) * PAGE_SIZE / 512;
1085 }
1086
1087 static void scsifront_backend_changed(struct xenbus_device *dev,
1088                                       enum xenbus_state backend_state)
1089 {
1090         struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
1091
1092         pr_debug("%s: %p %u %u\n", __func__, dev, dev->state, backend_state);
1093
1094         switch (backend_state) {
1095         case XenbusStateUnknown:
1096         case XenbusStateInitialising:
1097         case XenbusStateInitWait:
1098         case XenbusStateInitialised:
1099                 break;
1100
1101         case XenbusStateConnected:
1102                 scsifront_read_backend_params(dev, info);
1103
1104                 if (info->pause) {
1105                         scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_READD_LUN);
1106                         xenbus_switch_state(dev, XenbusStateConnected);
1107                         info->pause = 0;
1108                         return;
1109                 }
1110
1111                 if (xenbus_read_driver_state(dev->nodename) ==
1112                     XenbusStateInitialised)
1113                         scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
1114
1115                 if (dev->state != XenbusStateConnected)
1116                         xenbus_switch_state(dev, XenbusStateConnected);
1117                 break;
1118
1119         case XenbusStateClosed:
1120                 if (dev->state == XenbusStateClosed)
1121                         break;
1122                 /* Missed the backend's Closing state -- fallthrough */
1123         case XenbusStateClosing:
1124                 scsifront_disconnect(info);
1125                 break;
1126
1127         case XenbusStateReconfiguring:
1128                 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_DEL_LUN);
1129                 xenbus_switch_state(dev, XenbusStateReconfiguring);
1130                 break;
1131
1132         case XenbusStateReconfigured:
1133                 scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
1134                 xenbus_switch_state(dev, XenbusStateConnected);
1135                 break;
1136         }
1137 }
1138
1139 static const struct xenbus_device_id scsifront_ids[] = {
1140         { "vscsi" },
1141         { "" }
1142 };
1143
1144 static struct xenbus_driver scsifront_driver = {
1145         .ids                    = scsifront_ids,
1146         .probe                  = scsifront_probe,
1147         .remove                 = scsifront_remove,
1148         .resume                 = scsifront_resume,
1149         .suspend                = scsifront_suspend,
1150         .otherend_changed       = scsifront_backend_changed,
1151 };
1152
1153 static int __init scsifront_init(void)
1154 {
1155         if (!xen_domain())
1156                 return -ENODEV;
1157
1158         return xenbus_register_frontend(&scsifront_driver);
1159 }
1160 module_init(scsifront_init);
1161
1162 static void __exit scsifront_exit(void)
1163 {
1164         xenbus_unregister_driver(&scsifront_driver);
1165 }
1166 module_exit(scsifront_exit);
1167
1168 MODULE_DESCRIPTION("Xen SCSI frontend driver");
1169 MODULE_LICENSE("GPL");
1170 MODULE_ALIAS("xen:vscsi");
1171 MODULE_AUTHOR("Juergen Gross <jgross@suse.com>");