Pileus Git - ~andy/linux/blob - drivers/block/xen-blkback/blkback.c

   1 /******************************************************************************
   2  *
   3  * Back-end of the driver for virtual block devices. This portion of the
   4  * driver exports a 'unified' block-device interface that can be accessed
   5  * by any operating system that implements a compatible front end. A
   6  * reference front-end implementation can be found in:
   7  *  drivers/block/xen-blkfront.c
   8  *
   9  * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
  10  * Copyright (c) 2005, Christopher Clark
  11  *
  12  * This program is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU General Public License version 2
  14  * as published by the Free Software Foundation; or, when distributed
  15  * separately from the Linux kernel or incorporated into other
  16  * software packages, subject to the following license:
  17  *
  18  * Permission is hereby granted, free of charge, to any person obtaining a copy
  19  * of this source file (the "Software"), to deal in the Software without
  20  * restriction, including without limitation the rights to use, copy, modify,
  21  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
  22  * and to permit persons to whom the Software is furnished to do so, subject to
  23  * the following conditions:
  24  *
  25  * The above copyright notice and this permission notice shall be included in
  26  * all copies or substantial portions of the Software.
  27  *
  28  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  29  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  30  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  31  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  32  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  33  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  34  * IN THE SOFTWARE.
  35  */
  36
  37 #include <linux/spinlock.h>
  38 #include <linux/kthread.h>
  39 #include <linux/list.h>
  40 #include <linux/delay.h>
  41 #include <linux/freezer.h>
  42
  43 #include <xen/events.h>
  44 #include <xen/page.h>
  45 #include <asm/xen/hypervisor.h>
  46 #include <asm/xen/hypercall.h>
  47 #include "common.h"
  48
  49 /*
  50  * These are rather arbitrary. They are fairly large because adjacent requests
  51  * pulled from a communication ring are quite likely to end up being part of
  52  * the same scatter/gather request at the disc.
  53  *
  54  * ** TRY INCREASING 'xen_blkif_reqs' IF WRITE SPEEDS SEEM TOO LOW **
  55  *
  56  * This will increase the chances of being able to write whole tracks.
  57  * 64 should be enough to keep us competitive with Linux.
  58  */
  59 static int xen_blkif_reqs = 64;
  60 module_param_named(reqs, xen_blkif_reqs, int, 0);
  61 MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate");
  62
  63 /* Run-time switchable: /sys/module/blkback/parameters/ */
  64 static unsigned int log_stats;
  65 module_param(log_stats, int, 0644);
  66
  67 /*
  68  * Each outstanding request that we've passed to the lower device layers has a
  69  * 'pending_req' allocated to it. Each buffer_head that completes decrements
  70  * the pendcnt towards zero. When it hits zero, the specified domain has a
  71  * response queued for it, with the saved 'id' passed back.
  72  */
  73 struct pending_req {
  74         struct xen_blkif        *blkif;
  75         u64                     id;
  76         int                     nr_pages;
  77         atomic_t                pendcnt;
  78         unsigned short          operation;
  79         int                     status;
  80         struct list_head        free_list;
  81 };
  82
  83 #define BLKBACK_INVALID_HANDLE (~0)
  84
  85 struct xen_blkbk {
  86         struct pending_req      *pending_reqs;
  87         /* List of all 'pending_req' available */
  88         struct list_head        pending_free;
  89         /* And its spinlock. */
  90         spinlock_t              pending_free_lock;
  91         wait_queue_head_t       pending_free_wq;
  92         /* The list of all pages that are available. */
  93         struct page             **pending_pages;
  94         /* And the grant handles that are available. */
  95         grant_handle_t          *pending_grant_handles;
  96 };
  97
  98 static struct xen_blkbk *blkbk;
  99
 100 /*
 101  * Little helpful macro to figure out the index and virtual address of the
 102  * pending_pages[..]. For each 'pending_req' we have have up to
 103  * BLKIF_MAX_SEGMENTS_PER_REQUEST (11) pages. The seg would be from 0 through
 104  * 10 and would index in the pending_pages[..].
 105  */
 106 static inline int vaddr_pagenr(struct pending_req *req, int seg)
 107 {
 108         return (req - blkbk->pending_reqs) *
 109                 BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
 110 }
 111
 112 #define pending_page(req, seg) pending_pages[vaddr_pagenr(req, seg)]
 113
 114 static inline unsigned long vaddr(struct pending_req *req, int seg)
 115 {
 116         unsigned long pfn = page_to_pfn(blkbk->pending_page(req, seg));
 117         return (unsigned long)pfn_to_kaddr(pfn);
 118 }
 119
 120 #define pending_handle(_req, _seg) \
 121         (blkbk->pending_grant_handles[vaddr_pagenr(_req, _seg)])
 122
 123
 124 static int do_block_io_op(struct xen_blkif *blkif);
 125 static int dispatch_rw_block_io(struct xen_blkif *blkif,
 126                                 struct blkif_request *req,
 127                                 struct pending_req *pending_req);
 128 static void make_response(struct xen_blkif *blkif, u64 id,
 129                           unsigned short op, int st);
 130
 131 /*
 132  * Retrieve from the 'pending_reqs' a free pending_req structure to be used.
 133  */
 134 static struct pending_req *alloc_req(void)
 135 {
 136         struct pending_req *req = NULL;
 137         unsigned long flags;
 138
 139         spin_lock_irqsave(&blkbk->pending_free_lock, flags);
 140         if (!list_empty(&blkbk->pending_free)) {
 141                 req = list_entry(blkbk->pending_free.next, struct pending_req,
 142                                  free_list);
 143                 list_del(&req->free_list);
 144         }
 145         spin_unlock_irqrestore(&blkbk->pending_free_lock, flags);
 146         return req;
 147 }
 148
 149 /*
 150  * Return the 'pending_req' structure back to the freepool. We also
 151  * wake up the thread if it was waiting for a free page.
 152  */
 153 static void free_req(struct pending_req *req)
 154 {
 155         unsigned long flags;
 156         int was_empty;
 157
 158         spin_lock_irqsave(&blkbk->pending_free_lock, flags);
 159         was_empty = list_empty(&blkbk->pending_free);
 160         list_add(&req->free_list, &blkbk->pending_free);
 161         spin_unlock_irqrestore(&blkbk->pending_free_lock, flags);
 162         if (was_empty)
 163                 wake_up(&blkbk->pending_free_wq);
 164 }
 165
 166 /*
 167  * Routines for managing virtual block devices (vbds).
 168  */
 169 static int xen_vbd_translate(struct phys_req *req, struct xen_blkif *blkif,
 170                              int operation)
 171 {
 172         struct xen_vbd *vbd = &blkif->vbd;
 173         int rc = -EACCES;
 174
 175         if ((operation != READ) && vbd->readonly)
 176                 goto out;
 177
 178         if (likely(req->nr_sects)) {
 179                 blkif_sector_t end = req->sector_number + req->nr_sects;
 180
 181                 if (unlikely(end < req->sector_number))
 182                         goto out;
 183                 if (unlikely(end > vbd_sz(vbd)))
 184                         goto out;
 185         }
 186
 187         req->dev  = vbd->pdevice;
 188         req->bdev = vbd->bdev;
 189         rc = 0;
 190
 191  out:
 192         return rc;
 193 }
 194
 195 static void xen_vbd_resize(struct xen_blkif *blkif)
 196 {
 197         struct xen_vbd *vbd = &blkif->vbd;
 198         struct xenbus_transaction xbt;
 199         int err;
 200         struct xenbus_device *dev = xen_blkbk_xenbus(blkif->be);
 201         unsigned long long new_size = vbd_sz(vbd);
 202
 203         pr_info(DRV_PFX "VBD Resize: Domid: %d, Device: (%d, %d)\n",
 204                 blkif->domid, MAJOR(vbd->pdevice), MINOR(vbd->pdevice));
 205         pr_info(DRV_PFX "VBD Resize: new size %llu\n", new_size);
 206         vbd->size = new_size;
 207 again:
 208         err = xenbus_transaction_start(&xbt);
 209         if (err) {
 210                 pr_warn(DRV_PFX "Error starting transaction");
 211                 return;
 212         }
 213         err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
 214                             (unsigned long long)vbd_sz(vbd));
 215         if (err) {
 216                 pr_warn(DRV_PFX "Error writing new size");
 217                 goto abort;
 218         }
 219         /*
 220          * Write the current state; we will use this to synchronize
 221          * the front-end. If the current state is "connected" the
 222          * front-end will get the new size information online.
 223          */
 224         err = xenbus_printf(xbt, dev->nodename, "state", "%d", dev->state);
 225         if (err) {
 226                 pr_warn(DRV_PFX "Error writing the state");
 227                 goto abort;
 228         }
 229
 230         err = xenbus_transaction_end(xbt, 0);
 231         if (err == -EAGAIN)
 232                 goto again;
 233         if (err)
 234                 pr_warn(DRV_PFX "Error ending transaction");
 235         return;
 236 abort:
 237         xenbus_transaction_end(xbt, 1);
 238 }
 239
 240 /*
 241  * Notification from the guest OS.
 242  */
 243 static void blkif_notify_work(struct xen_blkif *blkif)
 244 {
 245         blkif->waiting_reqs = 1;
 246         wake_up(&blkif->wq);
 247 }
 248
 249 irqreturn_t xen_blkif_be_int(int irq, void *dev_id)
 250 {
 251         blkif_notify_work(dev_id);
 252         return IRQ_HANDLED;
 253 }
 254
 255 /*
 256  * SCHEDULER FUNCTIONS
 257  */
 258
 259 static void print_stats(struct xen_blkif *blkif)
 260 {
 261         pr_info("xen-blkback (%s): oo %3d  |  rd %4d  |  wr %4d  |  f %4d"
 262                  "  |  ds %4d\n",
 263                  current->comm, blkif->st_oo_req,
 264                  blkif->st_rd_req, blkif->st_wr_req,
 265                  blkif->st_f_req, blkif->st_ds_req);
 266         blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
 267         blkif->st_rd_req = 0;
 268         blkif->st_wr_req = 0;
 269         blkif->st_oo_req = 0;
 270         blkif->st_ds_req = 0;
 271 }
 272
 273 int xen_blkif_schedule(void *arg)
 274 {
 275         struct xen_blkif *blkif = arg;
 276         struct xen_vbd *vbd = &blkif->vbd;
 277
 278         xen_blkif_get(blkif);
 279
 280         while (!kthread_should_stop()) {
 281                 if (try_to_freeze())
 282                         continue;
 283                 if (unlikely(vbd->size != vbd_sz(vbd)))
 284                         xen_vbd_resize(blkif);
 285
 286                 wait_event_interruptible(
 287                         blkif->wq,
 288                         blkif->waiting_reqs || kthread_should_stop());
 289                 wait_event_interruptible(
 290                         blkbk->pending_free_wq,
 291                         !list_empty(&blkbk->pending_free) ||
 292                         kthread_should_stop());
 293
 294                 blkif->waiting_reqs = 0;
 295                 smp_mb(); /* clear flag *before* checking for work */
 296
 297                 if (do_block_io_op(blkif))
 298                         blkif->waiting_reqs = 1;
 299
 300                 if (log_stats && time_after(jiffies, blkif->st_print))
 301                         print_stats(blkif);
 302         }
 303
 304         if (log_stats)
 305                 print_stats(blkif);
 306
 307         blkif->xenblkd = NULL;
 308         xen_blkif_put(blkif);
 309
 310         return 0;
 311 }
 312
 313 struct seg_buf {
 314         unsigned long buf;
 315         unsigned int nsec;
 316 };
 317 /*
 318  * Unmap the grant references, and also remove the M2P over-rides
 319  * used in the 'pending_req'.
 320  */
 321 static void xen_blkbk_unmap(struct pending_req *req)
 322 {
 323         struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 324         unsigned int i, invcount = 0;
 325         grant_handle_t handle;
 326         int ret;
 327
 328         for (i = 0; i < req->nr_pages; i++) {
 329                 handle = pending_handle(req, i);
 330                 if (handle == BLKBACK_INVALID_HANDLE)
 331                         continue;
 332                 gnttab_set_unmap_op(&unmap[invcount], vaddr(req, i),
 333                                     GNTMAP_host_map, handle);
 334                 pending_handle(req, i) = BLKBACK_INVALID_HANDLE;
 335                 invcount++;
 336         }
 337
 338         ret = HYPERVISOR_grant_table_op(
 339                 GNTTABOP_unmap_grant_ref, unmap, invcount);
 340         BUG_ON(ret);
 341         /*
 342          * Note, we use invcount, so nr->pages, so we can't index
 343          * using vaddr(req, i).
 344          */
 345         for (i = 0; i < invcount; i++) {
 346                 ret = m2p_remove_override(
 347                         virt_to_page(unmap[i].host_addr), false);
 348                 if (ret) {
 349                         pr_alert(DRV_PFX "Failed to remove M2P override for %lx\n",
 350                                  (unsigned long)unmap[i].host_addr);
 351                         continue;
 352                 }
 353         }
 354 }
 355
 356 static int xen_blkbk_map(struct blkif_request *req,
 357                          struct pending_req *pending_req,
 358                          struct seg_buf seg[])
 359 {
 360         struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 361         int i;
 362         int nseg = req->u.rw.nr_segments;
 363         int ret = 0;
 364
 365         /*
 366          * Fill out preq.nr_sects with proper amount of sectors, and setup
 367          * assign map[..] with the PFN of the page in our domain with the
 368          * corresponding grant reference for each page.
 369          */
 370         for (i = 0; i < nseg; i++) {
 371                 uint32_t flags;
 372
 373                 flags = GNTMAP_host_map;
 374                 if (pending_req->operation != BLKIF_OP_READ)
 375                         flags |= GNTMAP_readonly;
 376                 gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
 377                                   req->u.rw.seg[i].gref,
 378                                   pending_req->blkif->domid);
 379         }
 380
 381         ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg);
 382         BUG_ON(ret);
 383
 384         /*
 385          * Now swizzle the MFN in our domain with the MFN from the other domain
 386          * so that when we access vaddr(pending_req,i) it has the contents of
 387          * the page from the other domain.
 388          */
 389         for (i = 0; i < nseg; i++) {
 390                 if (unlikely(map[i].status != 0)) {
 391                         pr_debug(DRV_PFX "invalid buffer -- could not remap it\n");
 392                         map[i].handle = BLKBACK_INVALID_HANDLE;
 393                         ret |= 1;
 394                 }
 395
 396                 pending_handle(pending_req, i) = map[i].handle;
 397
 398                 if (ret)
 399                         continue;
 400
 401                 ret = m2p_add_override(PFN_DOWN(map[i].dev_bus_addr),
 402                         blkbk->pending_page(pending_req, i), NULL);
 403                 if (ret) {
 404                         pr_alert(DRV_PFX "Failed to install M2P override for %lx (ret: %d)\n",
 405                                  (unsigned long)map[i].dev_bus_addr, ret);
 406                         /* We could switch over to GNTTABOP_copy */
 407                         continue;
 408                 }
 409
 410                 seg[i].buf  = map[i].dev_bus_addr |
 411                         (req->u.rw.seg[i].first_sect << 9);
 412         }
 413         return ret;
 414 }
 415
 416 static int dispatch_discard_io(struct xen_blkif *blkif,
 417                                 struct blkif_request *req)
 418 {
 419         int err = 0;
 420         int status = BLKIF_RSP_OKAY;
 421         struct block_device *bdev = blkif->vbd.bdev;
 422
 423         blkif->st_ds_req++;
 424
 425         xen_blkif_get(blkif);
 426         if (blkif->blk_backend_type == BLKIF_BACKEND_PHY ||
 427             blkif->blk_backend_type == BLKIF_BACKEND_FILE) {
 428                 unsigned long secure = (blkif->vbd.discard_secure &&
 429                         (req->u.discard.flag & BLKIF_DISCARD_SECURE)) ?
 430                         BLKDEV_DISCARD_SECURE : 0;
 431                 err = blkdev_issue_discard(bdev,
 432                                 req->u.discard.sector_number,
 433                                 req->u.discard.nr_sectors,
 434                                 GFP_KERNEL, secure);
 435         } else
 436                 err = -EOPNOTSUPP;
 437
 438         if (err == -EOPNOTSUPP) {
 439                 pr_debug(DRV_PFX "discard op failed, not supported\n");
 440                 status = BLKIF_RSP_EOPNOTSUPP;
 441         } else if (err)
 442                 status = BLKIF_RSP_ERROR;
 443
 444         make_response(blkif, req->u.discard.id, req->operation, status);
 445         xen_blkif_put(blkif);
 446         return err;
 447 }
 448
 449 static void xen_blk_drain_io(struct xen_blkif *blkif)
 450 {
 451         atomic_set(&blkif->drain, 1);
 452         do {
 453                 /* The initial value is one, and one refcnt taken at the
 454                  * start of the xen_blkif_schedule thread. */
 455                 if (atomic_read(&blkif->refcnt) <= 2)
 456                         break;
 457                 wait_for_completion_interruptible_timeout(
 458                                 &blkif->drain_complete, HZ);
 459
 460                 if (!atomic_read(&blkif->drain))
 461                         break;
 462         } while (!kthread_should_stop());
 463         atomic_set(&blkif->drain, 0);
 464 }
 465
 466 /*
 467  * Completion callback on the bio's. Called as bh->b_end_io()
 468  */
 469
 470 static void __end_block_io_op(struct pending_req *pending_req, int error)
 471 {
 472         /* An error fails the entire request. */
 473         if ((pending_req->operation == BLKIF_OP_FLUSH_DISKCACHE) &&
 474             (error == -EOPNOTSUPP)) {
 475                 pr_debug(DRV_PFX "flush diskcache op failed, not supported\n");
 476                 xen_blkbk_flush_diskcache(XBT_NIL, pending_req->blkif->be, 0);
 477                 pending_req->status = BLKIF_RSP_EOPNOTSUPP;
 478         } else if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) &&
 479                     (error == -EOPNOTSUPP)) {
 480                 pr_debug(DRV_PFX "write barrier op failed, not supported\n");
 481                 xen_blkbk_barrier(XBT_NIL, pending_req->blkif->be, 0);
 482                 pending_req->status = BLKIF_RSP_EOPNOTSUPP;
 483         } else if (error) {
 484                 pr_debug(DRV_PFX "Buffer not up-to-date at end of operation,"
 485                          " error=%d\n", error);
 486                 pending_req->status = BLKIF_RSP_ERROR;
 487         }
 488
 489         /*
 490          * If all of the bio's have completed it is time to unmap
 491          * the grant references associated with 'request' and provide
 492          * the proper response on the ring.
 493          */
 494         if (atomic_dec_and_test(&pending_req->pendcnt)) {
 495                 xen_blkbk_unmap(pending_req);
 496                 make_response(pending_req->blkif, pending_req->id,
 497                               pending_req->operation, pending_req->status);
 498                 xen_blkif_put(pending_req->blkif);
 499                 if (atomic_read(&pending_req->blkif->refcnt) <= 2) {
 500                         if (atomic_read(&pending_req->blkif->drain))
 501                                 complete(&pending_req->blkif->drain_complete);
 502                 }
 503                 free_req(pending_req);
 504         }
 505 }
 506
 507 /*
 508  * bio callback.
 509  */
 510 static void end_block_io_op(struct bio *bio, int error)
 511 {
 512         __end_block_io_op(bio->bi_private, error);
 513         bio_put(bio);
 514 }
 515
 516
 517
 518 /*
 519  * Function to copy the from the ring buffer the 'struct blkif_request'
 520  * (which has the sectors we want, number of them, grant references, etc),
 521  * and transmute  it to the block API to hand it over to the proper block disk.
 522  */
 523 static int
 524 __do_block_io_op(struct xen_blkif *blkif)
 525 {
 526         union blkif_back_rings *blk_rings = &blkif->blk_rings;
 527         struct blkif_request req;
 528         struct pending_req *pending_req;
 529         RING_IDX rc, rp;
 530         int more_to_do = 0;
 531
 532         rc = blk_rings->common.req_cons;
 533         rp = blk_rings->common.sring->req_prod;
 534         rmb(); /* Ensure we see queued requests up to 'rp'. */
 535
 536         while (rc != rp) {
 537
 538                 if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc))
 539                         break;
 540
 541                 if (kthread_should_stop()) {
 542                         more_to_do = 1;
 543                         break;
 544                 }
 545
 546                 pending_req = alloc_req();
 547                 if (NULL == pending_req) {
 548                         blkif->st_oo_req++;
 549                         more_to_do = 1;
 550                         break;
 551                 }
 552
 553                 switch (blkif->blk_protocol) {
 554                 case BLKIF_PROTOCOL_NATIVE:
 555                         memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req));
 556                         break;
 557                 case BLKIF_PROTOCOL_X86_32:
 558                         blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
 559                         break;
 560                 case BLKIF_PROTOCOL_X86_64:
 561                         blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
 562                         break;
 563                 default:
 564                         BUG();
 565                 }
 566                 blk_rings->common.req_cons = ++rc; /* before make_response() */
 567
 568                 /* Apply all sanity checks to /private copy/ of request. */
 569                 barrier();
 570                 if (unlikely(req.operation == BLKIF_OP_DISCARD)) {
 571                         free_req(pending_req);
 572                         if (dispatch_discard_io(blkif, &req))
 573                                 break;
 574                 } else if (dispatch_rw_block_io(blkif, &req, pending_req))
 575                         break;
 576
 577                 /* Yield point for this unbounded loop. */
 578                 cond_resched();
 579         }
 580
 581         return more_to_do;
 582 }
 583
 584 static int
 585 do_block_io_op(struct xen_blkif *blkif)
 586 {
 587         union blkif_back_rings *blk_rings = &blkif->blk_rings;
 588         int more_to_do;
 589
 590         do {
 591                 more_to_do = __do_block_io_op(blkif);
 592                 if (more_to_do)
 593                         break;
 594
 595                 RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
 596         } while (more_to_do);
 597
 598         return more_to_do;
 599 }
 600 /*
 601  * Transmutation of the 'struct blkif_request' to a proper 'struct bio'
 602  * and call the 'submit_bio' to pass it to the underlying storage.
 603  */
 604 static int dispatch_rw_block_io(struct xen_blkif *blkif,
 605                                 struct blkif_request *req,
 606                                 struct pending_req *pending_req)
 607 {
 608         struct phys_req preq;
 609         struct seg_buf seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 610         unsigned int nseg;
 611         struct bio *bio = NULL;
 612         struct bio *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST];
 613         int i, nbio = 0;
 614         int operation;
 615         struct blk_plug plug;
 616         bool drain = false;
 617
 618         switch (req->operation) {
 619         case BLKIF_OP_READ:
 620                 blkif->st_rd_req++;
 621                 operation = READ;
 622                 break;
 623         case BLKIF_OP_WRITE:
 624                 blkif->st_wr_req++;
 625                 operation = WRITE_ODIRECT;
 626                 break;
 627         case BLKIF_OP_WRITE_BARRIER:
 628                 drain = true;
 629         case BLKIF_OP_FLUSH_DISKCACHE:
 630                 blkif->st_f_req++;
 631                 operation = WRITE_FLUSH;
 632                 break;
 633         default:
 634                 operation = 0; /* make gcc happy */
 635                 goto fail_response;
 636                 break;
 637         }
 638
 639         /* Check that the number of segments is sane. */
 640         nseg = req->u.rw.nr_segments;
 641
 642         if (unlikely(nseg == 0 && operation != WRITE_FLUSH) ||
 643             unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
 644                 pr_debug(DRV_PFX "Bad number of segments in request (%d)\n",
 645                          nseg);
 646                 /* Haven't submitted any bio's yet. */
 647                 goto fail_response;
 648         }
 649
 650         preq.dev           = req->u.rw.handle;
 651         preq.sector_number = req->u.rw.sector_number;
 652         preq.nr_sects      = 0;
 653
 654         pending_req->blkif     = blkif;
 655         pending_req->id        = req->u.rw.id;
 656         pending_req->operation = req->operation;
 657         pending_req->status    = BLKIF_RSP_OKAY;
 658         pending_req->nr_pages  = nseg;
 659
 660         for (i = 0; i < nseg; i++) {
 661                 seg[i].nsec = req->u.rw.seg[i].last_sect -
 662                         req->u.rw.seg[i].first_sect + 1;
 663                 if ((req->u.rw.seg[i].last_sect >= (PAGE_SIZE >> 9)) ||
 664                     (req->u.rw.seg[i].last_sect < req->u.rw.seg[i].first_sect))
 665                         goto fail_response;
 666                 preq.nr_sects += seg[i].nsec;
 667
 668         }
 669
 670         if (xen_vbd_translate(&preq, blkif, operation) != 0) {
 671                 pr_debug(DRV_PFX "access denied: %s of [%llu,%llu] on dev=%04x\n",
 672                          operation == READ ? "read" : "write",
 673                          preq.sector_number,
 674                          preq.sector_number + preq.nr_sects, preq.dev);
 675                 goto fail_response;
 676         }
 677
 678         /*
 679          * This check _MUST_ be done after xen_vbd_translate as the preq.bdev
 680          * is set there.
 681          */
 682         for (i = 0; i < nseg; i++) {
 683                 if (((int)preq.sector_number|(int)seg[i].nsec) &
 684                     ((bdev_logical_block_size(preq.bdev) >> 9) - 1)) {
 685                         pr_debug(DRV_PFX "Misaligned I/O request from domain %d",
 686                                  blkif->domid);
 687                         goto fail_response;
 688                 }
 689         }
 690
 691         /* Wait on all outstanding I/O's and once that has been completed
 692          * issue the WRITE_FLUSH.
 693          */
 694         if (drain)
 695                 xen_blk_drain_io(pending_req->blkif);
 696
 697         /*
 698          * If we have failed at this point, we need to undo the M2P override,
 699          * set gnttab_set_unmap_op on all of the grant references and perform
 700          * the hypercall to unmap the grants - that is all done in
 701          * xen_blkbk_unmap.
 702          */
 703         if (xen_blkbk_map(req, pending_req, seg))
 704                 goto fail_flush;
 705
 706         /*
 707          * This corresponding xen_blkif_put is done in __end_block_io_op, or
 708          * below (in "!bio") if we are handling a BLKIF_OP_DISCARD.
 709          */
 710         xen_blkif_get(blkif);
 711
 712         for (i = 0; i < nseg; i++) {
 713                 while ((bio == NULL) ||
 714                        (bio_add_page(bio,
 715                                      blkbk->pending_page(pending_req, i),
 716                                      seg[i].nsec << 9,
 717                                      seg[i].buf & ~PAGE_MASK) == 0)) {
 718
 719                         bio = bio_alloc(GFP_KERNEL, nseg-i);
 720                         if (unlikely(bio == NULL))
 721                                 goto fail_put_bio;
 722
 723                         biolist[nbio++] = bio;
 724                         bio->bi_bdev    = preq.bdev;
 725                         bio->bi_private = pending_req;
 726                         bio->bi_end_io  = end_block_io_op;
 727                         bio->bi_sector  = preq.sector_number;
 728                 }
 729
 730                 preq.sector_number += seg[i].nsec;
 731         }
 732
 733         /* This will be hit if the operation was a flush or discard. */
 734         if (!bio) {
 735                 BUG_ON(operation != WRITE_FLUSH);
 736
 737                 bio = bio_alloc(GFP_KERNEL, 0);
 738                 if (unlikely(bio == NULL))
 739                         goto fail_put_bio;
 740
 741                 biolist[nbio++] = bio;
 742                 bio->bi_bdev    = preq.bdev;
 743                 bio->bi_private = pending_req;
 744                 bio->bi_end_io  = end_block_io_op;
 745         }
 746
 747         /*
 748          * We set it one so that the last submit_bio does not have to call
 749          * atomic_inc.
 750          */
 751         atomic_set(&pending_req->pendcnt, nbio);
 752
 753         /* Get a reference count for the disk queue and start sending I/O */
 754         blk_start_plug(&plug);
 755
 756         for (i = 0; i < nbio; i++)
 757                 submit_bio(operation, biolist[i]);
 758
 759         /* Let the I/Os go.. */
 760         blk_finish_plug(&plug);
 761
 762         if (operation == READ)
 763                 blkif->st_rd_sect += preq.nr_sects;
 764         else if (operation & WRITE)
 765                 blkif->st_wr_sect += preq.nr_sects;
 766
 767         return 0;
 768
 769  fail_flush:
 770         xen_blkbk_unmap(pending_req);
 771  fail_response:
 772         /* Haven't submitted any bio's yet. */
 773         make_response(blkif, req->u.rw.id, req->operation, BLKIF_RSP_ERROR);
 774         free_req(pending_req);
 775         msleep(1); /* back off a bit */
 776         return -EIO;
 777
 778  fail_put_bio:
 779         for (i = 0; i < nbio; i++)
 780                 bio_put(biolist[i]);
 781         __end_block_io_op(pending_req, -EINVAL);
 782         msleep(1); /* back off a bit */
 783         return -EIO;
 784 }
 785
 786
 787
 788 /*
 789  * Put a response on the ring on how the operation fared.
 790  */
 791 static void make_response(struct xen_blkif *blkif, u64 id,
 792                           unsigned short op, int st)
 793 {
 794         struct blkif_response  resp;
 795         unsigned long     flags;
 796         union blkif_back_rings *blk_rings = &blkif->blk_rings;
 797         int notify;
 798
 799         resp.id        = id;
 800         resp.operation = op;
 801         resp.status    = st;
 802
 803         spin_lock_irqsave(&blkif->blk_ring_lock, flags);
 804         /* Place on the response ring for the relevant domain. */
 805         switch (blkif->blk_protocol) {
 806         case BLKIF_PROTOCOL_NATIVE:
 807                 memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt),
 808                        &resp, sizeof(resp));
 809                 break;
 810         case BLKIF_PROTOCOL_X86_32:
 811                 memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt),
 812                        &resp, sizeof(resp));
 813                 break;
 814         case BLKIF_PROTOCOL_X86_64:
 815                 memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt),
 816                        &resp, sizeof(resp));
 817                 break;
 818         default:
 819                 BUG();
 820         }
 821         blk_rings->common.rsp_prod_pvt++;
 822         RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
 823         spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
 824         if (notify)
 825                 notify_remote_via_irq(blkif->irq);
 826 }
 827
 828 static int __init xen_blkif_init(void)
 829 {
 830         int i, mmap_pages;
 831         int rc = 0;
 832
 833         if (!xen_pv_domain())
 834                 return -ENODEV;
 835
 836         blkbk = kzalloc(sizeof(struct xen_blkbk), GFP_KERNEL);
 837         if (!blkbk) {
 838                 pr_alert(DRV_PFX "%s: out of memory!\n", __func__);
 839                 return -ENOMEM;
 840         }
 841
 842         mmap_pages = xen_blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
 843
 844         blkbk->pending_reqs          = kzalloc(sizeof(blkbk->pending_reqs[0]) *
 845                                         xen_blkif_reqs, GFP_KERNEL);
 846         blkbk->pending_grant_handles = kmalloc(sizeof(blkbk->pending_grant_handles[0]) *
 847                                         mmap_pages, GFP_KERNEL);
 848         blkbk->pending_pages         = kzalloc(sizeof(blkbk->pending_pages[0]) *
 849                                         mmap_pages, GFP_KERNEL);
 850
 851         if (!blkbk->pending_reqs || !blkbk->pending_grant_handles ||
 852             !blkbk->pending_pages) {
 853                 rc = -ENOMEM;
 854                 goto out_of_memory;
 855         }
 856
 857         for (i = 0; i < mmap_pages; i++) {
 858                 blkbk->pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
 859                 blkbk->pending_pages[i] = alloc_page(GFP_KERNEL);
 860                 if (blkbk->pending_pages[i] == NULL) {
 861                         rc = -ENOMEM;
 862                         goto out_of_memory;
 863                 }
 864         }
 865         rc = xen_blkif_interface_init();
 866         if (rc)
 867                 goto failed_init;
 868
 869         INIT_LIST_HEAD(&blkbk->pending_free);
 870         spin_lock_init(&blkbk->pending_free_lock);
 871         init_waitqueue_head(&blkbk->pending_free_wq);
 872
 873         for (i = 0; i < xen_blkif_reqs; i++)
 874                 list_add_tail(&blkbk->pending_reqs[i].free_list,
 875                               &blkbk->pending_free);
 876
 877         rc = xen_blkif_xenbus_init();
 878         if (rc)
 879                 goto failed_init;
 880
 881         return 0;
 882
 883  out_of_memory:
 884         pr_alert(DRV_PFX "%s: out of memory\n", __func__);
 885  failed_init:
 886         kfree(blkbk->pending_reqs);
 887         kfree(blkbk->pending_grant_handles);
 888         if (blkbk->pending_pages) {
 889                 for (i = 0; i < mmap_pages; i++) {
 890                         if (blkbk->pending_pages[i])
 891                                 __free_page(blkbk->pending_pages[i]);
 892                 }
 893                 kfree(blkbk->pending_pages);
 894         }
 895         kfree(blkbk);
 896         blkbk = NULL;
 897         return rc;
 898 }
 899
 900 module_init(xen_blkif_init);
 901
 902 MODULE_LICENSE("Dual BSD/GPL");
 903 MODULE_ALIAS("xen-backend:vbd");