drivers/block/xen-blkfront.c

   1 /*
   2  * blkfront.c
   3  *
   4  * XenLinux virtual block device driver.
   5  *
   6  * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
   7  * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
   8  * Copyright (c) 2004, Christian Limpach
   9  * Copyright (c) 2004, Andrew Warfield
  10  * Copyright (c) 2005, Christopher Clark
  11  * Copyright (c) 2005, XenSource Ltd
  12  *
  13  * This program is free software; you can redistribute it and/or
  14  * modify it under the terms of the GNU General Public License version 2
  15  * as published by the Free Software Foundation; or, when distributed
  16  * separately from the Linux kernel or incorporated into other
  17  * software packages, subject to the following license:
  18  *
  19  * Permission is hereby granted, free of charge, to any person obtaining a copy
  20  * of this source file (the "Software"), to deal in the Software without
  21  * restriction, including without limitation the rights to use, copy, modify,
  22  * merge, publish, distribute, sublicense, and/or sell copies of the Software,
  23  * and to permit persons to whom the Software is furnished to do so, subject to
  24  * the following conditions:
  25  *
  26  * The above copyright notice and this permission notice shall be included in
  27  * all copies or substantial portions of the Software.
  28  *
  29  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  30  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  31  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  32  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  33  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  34  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  35  * IN THE SOFTWARE.
  36  */
  37
  38 #include <linux/interrupt.h>
  39 #include <linux/blkdev.h>
  40 #include <linux/hdreg.h>
  41 #include <linux/cdrom.h>
  42 #include <linux/module.h>
  43 #include <linux/slab.h>
  44 #include <linux/mutex.h>
  45 #include <linux/scatterlist.h>
  46
  47 #include <xen/xen.h>
  48 #include <xen/xenbus.h>
  49 #include <xen/grant_table.h>
  50 #include <xen/events.h>
  51 #include <xen/page.h>
  52 #include <xen/platform_pci.h>
  53
  54 #include <xen/interface/grant_table.h>
  55 #include <xen/interface/io/blkif.h>
  56 #include <xen/interface/io/protocols.h>
  57
  58 #include <asm/xen/hypervisor.h>
  59
  60 enum blkif_state {
  61         BLKIF_STATE_DISCONNECTED,
  62         BLKIF_STATE_CONNECTED,
  63         BLKIF_STATE_SUSPENDED,
  64 };
  65
  66 struct blk_shadow {
  67         struct blkif_request req;
  68         struct request *request;
  69         unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST];
  70 };
  71
  72 static DEFINE_MUTEX(blkfront_mutex);
  73 static const struct block_device_operations xlvbd_block_fops;
  74
  75 #define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE)
  76
  77 /*
  78  * We have one of these per vbd, whether ide, scsi or 'other'.  They
  79  * hang in private_data off the gendisk structure. We may end up
  80  * putting all kinds of interesting stuff here :-)
  81  */
  82 struct blkfront_info
  83 {
  84         struct mutex mutex;
  85         struct xenbus_device *xbdev;
  86         struct gendisk *gd;
  87         int vdevice;
  88         blkif_vdev_t handle;
  89         enum blkif_state connected;
  90         int ring_ref;
  91         struct blkif_front_ring ring;
  92         struct scatterlist sg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
  93         unsigned int evtchn, irq;
  94         struct request_queue *rq;
  95         struct work_struct work;
  96         struct gnttab_free_callback callback;
  97         struct blk_shadow shadow[BLK_RING_SIZE];
  98         unsigned long shadow_free;
  99         unsigned int feature_flush;
 100         int is_ready;
 101 };
 102
 103 static DEFINE_SPINLOCK(blkif_io_lock);
 104
 105 static unsigned int nr_minors;
 106 static unsigned long *minors;
 107 static DEFINE_SPINLOCK(minor_lock);
 108
 109 #define MAXIMUM_OUTSTANDING_BLOCK_REQS \
 110         (BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
 111 #define GRANT_INVALID_REF       0
 112
 113 #define PARTS_PER_DISK          16
 114 #define PARTS_PER_EXT_DISK      256
 115
 116 #define BLKIF_MAJOR(dev) ((dev)>>8)
 117 #define BLKIF_MINOR(dev) ((dev) & 0xff)
 118
 119 #define EXT_SHIFT 28
 120 #define EXTENDED (1<<EXT_SHIFT)
 121 #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
 122 #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
 123 #define EMULATED_HD_DISK_MINOR_OFFSET (0)
 124 #define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
 125 #define EMULATED_SD_DISK_MINOR_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET + (4 * 16))
 126 #define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_HD_DISK_NAME_OFFSET + 4)
 127
 128 #define DEV_NAME        "xvd"   /* name in /dev */
 129
 130 static int get_id_from_freelist(struct blkfront_info *info)
 131 {
 132         unsigned long free = info->shadow_free;
 133         BUG_ON(free >= BLK_RING_SIZE);
 134         info->shadow_free = info->shadow[free].req.id;
 135         info->shadow[free].req.id = 0x0fffffee; /* debug */
 136         return free;
 137 }
 138
 139 static void add_id_to_freelist(struct blkfront_info *info,
 140                                unsigned long id)
 141 {
 142         info->shadow[id].req.id  = info->shadow_free;
 143         info->shadow[id].request = NULL;
 144         info->shadow_free = id;
 145 }
 146
 147 static int xlbd_reserve_minors(unsigned int minor, unsigned int nr)
 148 {
 149         unsigned int end = minor + nr;
 150         int rc;
 151
 152         if (end > nr_minors) {
 153                 unsigned long *bitmap, *old;
 154
 155                 bitmap = kzalloc(BITS_TO_LONGS(end) * sizeof(*bitmap),
 156                                  GFP_KERNEL);
 157                 if (bitmap == NULL)
 158                         return -ENOMEM;
 159
 160                 spin_lock(&minor_lock);
 161                 if (end > nr_minors) {
 162                         old = minors;
 163                         memcpy(bitmap, minors,
 164                                BITS_TO_LONGS(nr_minors) * sizeof(*bitmap));
 165                         minors = bitmap;
 166                         nr_minors = BITS_TO_LONGS(end) * BITS_PER_LONG;
 167                 } else
 168                         old = bitmap;
 169                 spin_unlock(&minor_lock);
 170                 kfree(old);
 171         }
 172
 173         spin_lock(&minor_lock);
 174         if (find_next_bit(minors, end, minor) >= end) {
 175                 for (; minor < end; ++minor)
 176                         __set_bit(minor, minors);
 177                 rc = 0;
 178         } else
 179                 rc = -EBUSY;
 180         spin_unlock(&minor_lock);
 181
 182         return rc;
 183 }
 184
 185 static void xlbd_release_minors(unsigned int minor, unsigned int nr)
 186 {
 187         unsigned int end = minor + nr;
 188
 189         BUG_ON(end > nr_minors);
 190         spin_lock(&minor_lock);
 191         for (; minor < end; ++minor)
 192                 __clear_bit(minor, minors);
 193         spin_unlock(&minor_lock);
 194 }
 195
 196 static void blkif_restart_queue_callback(void *arg)
 197 {
 198         struct blkfront_info *info = (struct blkfront_info *)arg;
 199         schedule_work(&info->work);
 200 }
 201
 202 static int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
 203 {
 204         /* We don't have real geometry info, but let's at least return
 205            values consistent with the size of the device */
 206         sector_t nsect = get_capacity(bd->bd_disk);
 207         sector_t cylinders = nsect;
 208
 209         hg->heads = 0xff;
 210         hg->sectors = 0x3f;
 211         sector_div(cylinders, hg->heads * hg->sectors);
 212         hg->cylinders = cylinders;
 213         if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
 214                 hg->cylinders = 0xffff;
 215         return 0;
 216 }
 217
 218 static int blkif_ioctl(struct block_device *bdev, fmode_t mode,
 219                        unsigned command, unsigned long argument)
 220 {
 221         struct blkfront_info *info = bdev->bd_disk->private_data;
 222         int i;
 223
 224         dev_dbg(&info->xbdev->dev, "command: 0x%x, argument: 0x%lx\n",
 225                 command, (long)argument);
 226
 227         switch (command) {
 228         case CDROMMULTISESSION:
 229                 dev_dbg(&info->xbdev->dev, "FIXME: support multisession CDs later\n");
 230                 for (i = 0; i < sizeof(struct cdrom_multisession); i++)
 231                         if (put_user(0, (char __user *)(argument + i)))
 232                                 return -EFAULT;
 233                 return 0;
 234
 235         case CDROM_GET_CAPABILITY: {
 236                 struct gendisk *gd = info->gd;
 237                 if (gd->flags & GENHD_FL_CD)
 238                         return 0;
 239                 return -EINVAL;
 240         }
 241
 242         default:
 243                 /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
 244                   command);*/
 245                 return -EINVAL; /* same return as native Linux */
 246         }
 247
 248         return 0;
 249 }
 250
 251 /*
 252  * Generate a Xen blkfront IO request from a blk layer request.  Reads
 253  * and writes are handled as expected.  Since we lack a loose flush
 254  * request, we map flushes into a full ordered barrier.
 255  *
 256  * @req: a request struct
 257  */
 258 static int blkif_queue_request(struct request *req)
 259 {
 260         struct blkfront_info *info = req->rq_disk->private_data;
 261         unsigned long buffer_mfn;
 262         struct blkif_request *ring_req;
 263         unsigned long id;
 264         unsigned int fsect, lsect;
 265         int i, ref;
 266         grant_ref_t gref_head;
 267         struct scatterlist *sg;
 268
 269         if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
 270                 return 1;
 271
 272         if (gnttab_alloc_grant_references(
 273                 BLKIF_MAX_SEGMENTS_PER_REQUEST, &gref_head) < 0) {
 274                 gnttab_request_free_callback(
 275                         &info->callback,
 276                         blkif_restart_queue_callback,
 277                         info,
 278                         BLKIF_MAX_SEGMENTS_PER_REQUEST);
 279                 return 1;
 280         }
 281
 282         /* Fill out a communications ring structure. */
 283         ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
 284         id = get_id_from_freelist(info);
 285         info->shadow[id].request = req;
 286
 287         ring_req->id = id;
 288         ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req);
 289         ring_req->handle = info->handle;
 290
 291         ring_req->operation = rq_data_dir(req) ?
 292                 BLKIF_OP_WRITE : BLKIF_OP_READ;
 293
 294         if (req->cmd_flags & (REQ_FLUSH | REQ_FUA)) {
 295                 /*
 296                  * Ideally we could just do an unordered
 297                  * flush-to-disk, but all we have is a full write
 298                  * barrier at the moment.  However, a barrier write is
 299                  * a superset of FUA, so we can implement it the same
 300                  * way.  (It's also a FLUSH+FUA, since it is
 301                  * guaranteed ordered WRT previous writes.)
 302                  */
 303                 ring_req->operation = BLKIF_OP_WRITE_BARRIER;
 304         }
 305
 306         ring_req->nr_segments = blk_rq_map_sg(req->q, req, info->sg);
 307         BUG_ON(ring_req->nr_segments > BLKIF_MAX_SEGMENTS_PER_REQUEST);
 308
 309         for_each_sg(info->sg, sg, ring_req->nr_segments, i) {
 310                 buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg)));
 311                 fsect = sg->offset >> 9;
 312                 lsect = fsect + (sg->length >> 9) - 1;
 313                 /* install a grant reference. */
 314                 ref = gnttab_claim_grant_reference(&gref_head);
 315                 BUG_ON(ref == -ENOSPC);
 316
 317                 gnttab_grant_foreign_access_ref(
 318                                 ref,
 319                                 info->xbdev->otherend_id,
 320                                 buffer_mfn,
 321                                 rq_data_dir(req) );
 322
 323                 info->shadow[id].frame[i] = mfn_to_pfn(buffer_mfn);
 324                 ring_req->u.rw.seg[i] =
 325                                 (struct blkif_request_segment) {
 326                                         .gref       = ref,
 327                                         .first_sect = fsect,
 328                                         .last_sect  = lsect };
 329         }
 330
 331         info->ring.req_prod_pvt++;
 332
 333         /* Keep a private copy so we can reissue requests when recovering. */
 334         info->shadow[id].req = *ring_req;
 335
 336         gnttab_free_grant_references(gref_head);
 337
 338         return 0;
 339 }
 340
 341
 342 static inline void flush_requests(struct blkfront_info *info)
 343 {
 344         int notify;
 345
 346         RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&info->ring, notify);
 347
 348         if (notify)
 349                 notify_remote_via_irq(info->irq);
 350 }
 351
 352 /*
 353  * do_blkif_request
 354  *  read a block; request is in a request queue
 355  */
 356 static void do_blkif_request(struct request_queue *rq)
 357 {
 358         struct blkfront_info *info = NULL;
 359         struct request *req;
 360         int queued;
 361
 362         pr_debug("Entered do_blkif_request\n");
 363
 364         queued = 0;
 365
 366         while ((req = blk_peek_request(rq)) != NULL) {
 367                 info = req->rq_disk->private_data;
 368
 369                 if (RING_FULL(&info->ring))
 370                         goto wait;
 371
 372                 blk_start_request(req);
 373
 374                 if (req->cmd_type != REQ_TYPE_FS) {
 375                         __blk_end_request_all(req, -EIO);
 376                         continue;
 377                 }
 378
 379                 pr_debug("do_blk_req %p: cmd %p, sec %lx, "
 380                          "(%u/%u) buffer:%p [%s]\n",
 381                          req, req->cmd, (unsigned long)blk_rq_pos(req),
 382                          blk_rq_cur_sectors(req), blk_rq_sectors(req),
 383                          req->buffer, rq_data_dir(req) ? "write" : "read");
 384
 385                 if (blkif_queue_request(req)) {
 386                         blk_requeue_request(rq, req);
 387 wait:
 388                         /* Avoid pointless unplugs. */
 389                         blk_stop_queue(rq);
 390                         break;
 391                 }
 392
 393                 queued++;
 394         }
 395
 396         if (queued != 0)
 397                 flush_requests(info);
 398 }
 399
 400 static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
 401 {
 402         struct request_queue *rq;
 403
 404         rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
 405         if (rq == NULL)
 406                 return -1;
 407
 408         queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq);
 409
 410         /* Hard sector size and max sectors impersonate the equiv. hardware. */
 411         blk_queue_logical_block_size(rq, sector_size);
 412         blk_queue_max_hw_sectors(rq, 512);
 413
 414         /* Each segment in a request is up to an aligned page in size. */
 415         blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
 416         blk_queue_max_segment_size(rq, PAGE_SIZE);
 417
 418         /* Ensure a merged request will fit in a single I/O ring slot. */
 419         blk_queue_max_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
 420
 421         /* Make sure buffer addresses are sector-aligned. */
 422         blk_queue_dma_alignment(rq, 511);
 423
 424         /* Make sure we don't use bounce buffers. */
 425         blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY);
 426
 427         gd->queue = rq;
 428
 429         return 0;
 430 }
 431
 432
 433 static void xlvbd_flush(struct blkfront_info *info)
 434 {
 435         blk_queue_flush(info->rq, info->feature_flush);
 436         printk(KERN_INFO "blkfront: %s: barriers %s\n",
 437                info->gd->disk_name,
 438                info->feature_flush ? "enabled" : "disabled");
 439 }
 440
 441 static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
 442 {
 443         int major;
 444         major = BLKIF_MAJOR(vdevice);
 445         *minor = BLKIF_MINOR(vdevice);
 446         switch (major) {
 447                 case XEN_IDE0_MAJOR:
 448                         *offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET;
 449                         *minor = ((*minor / 64) * PARTS_PER_DISK) +
 450                                 EMULATED_HD_DISK_MINOR_OFFSET;
 451                         break;
 452                 case XEN_IDE1_MAJOR:
 453                         *offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET;
 454                         *minor = (((*minor / 64) + 2) * PARTS_PER_DISK) +
 455                                 EMULATED_HD_DISK_MINOR_OFFSET;
 456                         break;
 457                 case XEN_SCSI_DISK0_MAJOR:
 458                         *offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET;
 459                         *minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET;
 460                         break;
 461                 case XEN_SCSI_DISK1_MAJOR:
 462                 case XEN_SCSI_DISK2_MAJOR:
 463                 case XEN_SCSI_DISK3_MAJOR:
 464                 case XEN_SCSI_DISK4_MAJOR:
 465                 case XEN_SCSI_DISK5_MAJOR:
 466                 case XEN_SCSI_DISK6_MAJOR:
 467                 case XEN_SCSI_DISK7_MAJOR:
 468                         *offset = (*minor / PARTS_PER_DISK) +
 469                                 ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) +
 470                                 EMULATED_SD_DISK_NAME_OFFSET;
 471                         *minor = *minor +
 472                                 ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) +
 473                                 EMULATED_SD_DISK_MINOR_OFFSET;
 474                         break;
 475                 case XEN_SCSI_DISK8_MAJOR:
 476                 case XEN_SCSI_DISK9_MAJOR:
 477                 case XEN_SCSI_DISK10_MAJOR:
 478                 case XEN_SCSI_DISK11_MAJOR:
 479                 case XEN_SCSI_DISK12_MAJOR:
 480                 case XEN_SCSI_DISK13_MAJOR:
 481                 case XEN_SCSI_DISK14_MAJOR:
 482                 case XEN_SCSI_DISK15_MAJOR:
 483                         *offset = (*minor / PARTS_PER_DISK) +
 484                                 ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) +
 485                                 EMULATED_SD_DISK_NAME_OFFSET;
 486                         *minor = *minor +
 487                                 ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) +
 488                                 EMULATED_SD_DISK_MINOR_OFFSET;
 489                         break;
 490                 case XENVBD_MAJOR:
 491                         *offset = *minor / PARTS_PER_DISK;
 492                         break;
 493                 default:
 494                         printk(KERN_WARNING "blkfront: your disk configuration is "
 495                                         "incorrect, please use an xvd device instead\n");
 496                         return -ENODEV;
 497         }
 498         return 0;
 499 }
 500
 501 static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 502                                struct blkfront_info *info,
 503                                u16 vdisk_info, u16 sector_size)
 504 {
 505         struct gendisk *gd;
 506         int nr_minors = 1;
 507         int err;
 508         unsigned int offset;
 509         int minor;
 510         int nr_parts;
 511
 512         BUG_ON(info->gd != NULL);
 513         BUG_ON(info->rq != NULL);
 514
 515         if ((info->vdevice>>EXT_SHIFT) > 1) {
 516                 /* this is above the extended range; something is wrong */
 517                 printk(KERN_WARNING "blkfront: vdevice 0x%x is above the extended range; ignoring\n", info->vdevice);
 518                 return -ENODEV;
 519         }
 520
 521         if (!VDEV_IS_EXTENDED(info->vdevice)) {
 522                 err = xen_translate_vdev(info->vdevice, &minor, &offset);
 523                 if (err)
 524                         return err;
 525                 nr_parts = PARTS_PER_DISK;
 526         } else {
 527                 minor = BLKIF_MINOR_EXT(info->vdevice);
 528                 nr_parts = PARTS_PER_EXT_DISK;
 529                 offset = minor / nr_parts;
 530                 if (xen_hvm_domain() && offset <= EMULATED_HD_DISK_NAME_OFFSET + 4)
 531                         printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with "
 532                                         "emulated IDE disks,\n\t choose an xvd device name"
 533                                         "from xvde on\n", info->vdevice);
 534         }
 535         err = -ENODEV;
 536
 537         if ((minor % nr_parts) == 0)
 538                 nr_minors = nr_parts;
 539
 540         err = xlbd_reserve_minors(minor, nr_minors);
 541         if (err)
 542                 goto out;
 543         err = -ENODEV;
 544
 545         gd = alloc_disk(nr_minors);
 546         if (gd == NULL)
 547                 goto release;
 548
 549         if (nr_minors > 1) {
 550                 if (offset < 26)
 551                         sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset);
 552                 else
 553                         sprintf(gd->disk_name, "%s%c%c", DEV_NAME,
 554                                 'a' + ((offset / 26)-1), 'a' + (offset % 26));
 555         } else {
 556                 if (offset < 26)
 557                         sprintf(gd->disk_name, "%s%c%d", DEV_NAME,
 558                                 'a' + offset,
 559                                 minor & (nr_parts - 1));
 560                 else
 561                         sprintf(gd->disk_name, "%s%c%c%d", DEV_NAME,
 562                                 'a' + ((offset / 26) - 1),
 563                                 'a' + (offset % 26),
 564                                 minor & (nr_parts - 1));
 565         }
 566
 567         gd->major = XENVBD_MAJOR;
 568         gd->first_minor = minor;
 569         gd->fops = &xlvbd_block_fops;
 570         gd->private_data = info;
 571         gd->driverfs_dev = &(info->xbdev->dev);
 572         set_capacity(gd, capacity);
 573
 574         if (xlvbd_init_blk_queue(gd, sector_size)) {
 575                 del_gendisk(gd);
 576                 goto release;
 577         }
 578
 579         info->rq = gd->queue;
 580         info->gd = gd;
 581
 582         xlvbd_flush(info);
 583
 584         if (vdisk_info & VDISK_READONLY)
 585                 set_disk_ro(gd, 1);
 586
 587         if (vdisk_info & VDISK_REMOVABLE)
 588                 gd->flags |= GENHD_FL_REMOVABLE;
 589
 590         if (vdisk_info & VDISK_CDROM)
 591                 gd->flags |= GENHD_FL_CD;
 592
 593         return 0;
 594
 595  release:
 596         xlbd_release_minors(minor, nr_minors);
 597  out:
 598         return err;
 599 }
 600
 601 static void xlvbd_release_gendisk(struct blkfront_info *info)
 602 {
 603         unsigned int minor, nr_minors;
 604         unsigned long flags;
 605
 606         if (info->rq == NULL)
 607                 return;
 608
 609         spin_lock_irqsave(&blkif_io_lock, flags);
 610
 611         /* No more blkif_request(). */
 612         blk_stop_queue(info->rq);
 613
 614         /* No more gnttab callback work. */
 615         gnttab_cancel_free_callback(&info->callback);
 616         spin_unlock_irqrestore(&blkif_io_lock, flags);
 617
 618         /* Flush gnttab callback work. Must be done with no locks held. */
 619         flush_work_sync(&info->work);
 620
 621         del_gendisk(info->gd);
 622
 623         minor = info->gd->first_minor;
 624         nr_minors = info->gd->minors;
 625         xlbd_release_minors(minor, nr_minors);
 626
 627         blk_cleanup_queue(info->rq);
 628         info->rq = NULL;
 629
 630         put_disk(info->gd);
 631         info->gd = NULL;
 632 }
 633
 634 static void kick_pending_request_queues(struct blkfront_info *info)
 635 {
 636         if (!RING_FULL(&info->ring)) {
 637                 /* Re-enable calldowns. */
 638                 blk_start_queue(info->rq);
 639                 /* Kick things off immediately. */
 640                 do_blkif_request(info->rq);
 641         }
 642 }
 643
 644 static void blkif_restart_queue(struct work_struct *work)
 645 {
 646         struct blkfront_info *info = container_of(work, struct blkfront_info, work);
 647
 648         spin_lock_irq(&blkif_io_lock);
 649         if (info->connected == BLKIF_STATE_CONNECTED)
 650                 kick_pending_request_queues(info);
 651         spin_unlock_irq(&blkif_io_lock);
 652 }
 653
 654 static void blkif_free(struct blkfront_info *info, int suspend)
 655 {
 656         /* Prevent new requests being issued until we fix things up. */
 657         spin_lock_irq(&blkif_io_lock);
 658         info->connected = suspend ?
 659                 BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
 660         /* No more blkif_request(). */
 661         if (info->rq)
 662                 blk_stop_queue(info->rq);
 663         /* No more gnttab callback work. */
 664         gnttab_cancel_free_callback(&info->callback);
 665         spin_unlock_irq(&blkif_io_lock);
 666
 667         /* Flush gnttab callback work. Must be done with no locks held. */
 668         flush_work_sync(&info->work);
 669
 670         /* Free resources associated with old device channel. */
 671         if (info->ring_ref != GRANT_INVALID_REF) {
 672                 gnttab_end_foreign_access(info->ring_ref, 0,
 673                                           (unsigned long)info->ring.sring);
 674                 info->ring_ref = GRANT_INVALID_REF;
 675                 info->ring.sring = NULL;
 676         }
 677         if (info->irq)
 678                 unbind_from_irqhandler(info->irq, info);
 679         info->evtchn = info->irq = 0;
 680
 681 }
 682
 683 static void blkif_completion(struct blk_shadow *s)
 684 {
 685         int i;
 686         for (i = 0; i < s->req.nr_segments; i++)
 687                 gnttab_end_foreign_access(s->req.u.rw.seg[i].gref, 0, 0UL);
 688 }
 689
 690 static irqreturn_t blkif_interrupt(int irq, void *dev_id)
 691 {
 692         struct request *req;
 693         struct blkif_response *bret;
 694         RING_IDX i, rp;
 695         unsigned long flags;
 696         struct blkfront_info *info = (struct blkfront_info *)dev_id;
 697         int error;
 698
 699         spin_lock_irqsave(&blkif_io_lock, flags);
 700
 701         if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) {
 702                 spin_unlock_irqrestore(&blkif_io_lock, flags);
 703                 return IRQ_HANDLED;
 704         }
 705
 706  again:
 707         rp = info->ring.sring->rsp_prod;
 708         rmb(); /* Ensure we see queued responses up to 'rp'. */
 709
 710         for (i = info->ring.rsp_cons; i != rp; i++) {
 711                 unsigned long id;
 712
 713                 bret = RING_GET_RESPONSE(&info->ring, i);
 714                 id   = bret->id;
 715                 req  = info->shadow[id].request;
 716
 717                 blkif_completion(&info->shadow[id]);
 718
 719                 add_id_to_freelist(info, id);
 720
 721                 error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO;
 722                 switch (bret->operation) {
 723                 case BLKIF_OP_WRITE_BARRIER:
 724                         if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
 725                                 printk(KERN_WARNING "blkfront: %s: write barrier op failed\n",
 726                                        info->gd->disk_name);
 727                                 error = -EOPNOTSUPP;
 728                         }
 729                         if (unlikely(bret->status == BLKIF_RSP_ERROR &&
 730                                      info->shadow[id].req.nr_segments == 0)) {
 731                                 printk(KERN_WARNING "blkfront: %s: empty write barrier op failed\n",
 732                                        info->gd->disk_name);
 733                                 error = -EOPNOTSUPP;
 734                         }
 735                         if (unlikely(error)) {
 736                                 if (error == -EOPNOTSUPP)
 737                                         error = 0;
 738                                 info->feature_flush = 0;
 739                                 xlvbd_flush(info);
 740                         }
 741                         /* fall through */
 742                 case BLKIF_OP_READ:
 743                 case BLKIF_OP_WRITE:
 744                         if (unlikely(bret->status != BLKIF_RSP_OKAY))
 745                                 dev_dbg(&info->xbdev->dev, "Bad return from blkdev data "
 746                                         "request: %x\n", bret->status);
 747
 748                         __blk_end_request_all(req, error);
 749                         break;
 750                 default:
 751                         BUG();
 752                 }
 753         }
 754
 755         info->ring.rsp_cons = i;
 756
 757         if (i != info->ring.req_prod_pvt) {
 758                 int more_to_do;
 759                 RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
 760                 if (more_to_do)
 761                         goto again;
 762         } else
 763                 info->ring.sring->rsp_event = i + 1;
 764
 765         kick_pending_request_queues(info);
 766
 767         spin_unlock_irqrestore(&blkif_io_lock, flags);
 768
 769         return IRQ_HANDLED;
 770 }
 771
 772
 773 static int setup_blkring(struct xenbus_device *dev,
 774                          struct blkfront_info *info)
 775 {
 776         struct blkif_sring *sring;
 777         int err;
 778
 779         info->ring_ref = GRANT_INVALID_REF;
 780
 781         sring = (struct blkif_sring *)__get_free_page(GFP_NOIO | __GFP_HIGH);
 782         if (!sring) {
 783                 xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
 784                 return -ENOMEM;
 785         }
 786         SHARED_RING_INIT(sring);
 787         FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
 788
 789         sg_init_table(info->sg, BLKIF_MAX_SEGMENTS_PER_REQUEST);
 790
 791         err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
 792         if (err < 0) {
 793                 free_page((unsigned long)sring);
 794                 info->ring.sring = NULL;
 795                 goto fail;
 796         }
 797         info->ring_ref = err;
 798
 799         err = xenbus_alloc_evtchn(dev, &info->evtchn);
 800         if (err)
 801                 goto fail;
 802
 803         err = bind_evtchn_to_irqhandler(info->evtchn,
 804                                         blkif_interrupt,
 805                                         IRQF_SAMPLE_RANDOM, "blkif", info);
 806         if (err <= 0) {
 807                 xenbus_dev_fatal(dev, err,
 808                                  "bind_evtchn_to_irqhandler failed");
 809                 goto fail;
 810         }
 811         info->irq = err;
 812
 813         return 0;
 814 fail:
 815         blkif_free(info, 0);
 816         return err;
 817 }
 818
 819
 820 /* Common code used when first setting up, and when resuming. */
 821 static int talk_to_blkback(struct xenbus_device *dev,
 822                            struct blkfront_info *info)
 823 {
 824         const char *message = NULL;
 825         struct xenbus_transaction xbt;
 826         int err;
 827
 828         /* Create shared ring, alloc event channel. */
 829         err = setup_blkring(dev, info);
 830         if (err)
 831                 goto out;
 832
 833 again:
 834         err = xenbus_transaction_start(&xbt);
 835         if (err) {
 836                 xenbus_dev_fatal(dev, err, "starting transaction");
 837                 goto destroy_blkring;
 838         }
 839
 840         err = xenbus_printf(xbt, dev->nodename,
 841                             "ring-ref", "%u", info->ring_ref);
 842         if (err) {
 843                 message = "writing ring-ref";
 844                 goto abort_transaction;
 845         }
 846         err = xenbus_printf(xbt, dev->nodename,
 847                             "event-channel", "%u", info->evtchn);
 848         if (err) {
 849                 message = "writing event-channel";
 850                 goto abort_transaction;
 851         }
 852         err = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
 853                             XEN_IO_PROTO_ABI_NATIVE);
 854         if (err) {
 855                 message = "writing protocol";
 856                 goto abort_transaction;
 857         }
 858
 859         err = xenbus_transaction_end(xbt, 0);
 860         if (err) {
 861                 if (err == -EAGAIN)
 862                         goto again;
 863                 xenbus_dev_fatal(dev, err, "completing transaction");
 864                 goto destroy_blkring;
 865         }
 866
 867         xenbus_switch_state(dev, XenbusStateInitialised);
 868
 869         return 0;
 870
 871  abort_transaction:
 872         xenbus_transaction_end(xbt, 1);
 873         if (message)
 874                 xenbus_dev_fatal(dev, err, "%s", message);
 875  destroy_blkring:
 876         blkif_free(info, 0);
 877  out:
 878         return err;
 879 }
 880
 881 /**
 882  * Entry point to this code when a new device is created.  Allocate the basic
 883  * structures and the ring buffer for communication with the backend, and
 884  * inform the backend of the appropriate details for those.  Switch to
 885  * Initialised state.
 886  */
 887 static int blkfront_probe(struct xenbus_device *dev,
 888                           const struct xenbus_device_id *id)
 889 {
 890         int err, vdevice, i;
 891         struct blkfront_info *info;
 892
 893         /* FIXME: Use dynamic device id if this is not set. */
 894         err = xenbus_scanf(XBT_NIL, dev->nodename,
 895                            "virtual-device", "%i", &vdevice);
 896         if (err != 1) {
 897                 /* go looking in the extended area instead */
 898                 err = xenbus_scanf(XBT_NIL, dev->nodename, "virtual-device-ext",
 899                                    "%i", &vdevice);
 900                 if (err != 1) {
 901                         xenbus_dev_fatal(dev, err, "reading virtual-device");
 902                         return err;
 903                 }
 904         }
 905
 906         if (xen_hvm_domain()) {
 907                 char *type;
 908                 int len;
 909                 /* no unplug has been done: do not hook devices != xen vbds */
 910                 if (xen_platform_pci_unplug & XEN_UNPLUG_UNNECESSARY) {
 911                         int major;
 912
 913                         if (!VDEV_IS_EXTENDED(vdevice))
 914                                 major = BLKIF_MAJOR(vdevice);
 915                         else
 916                                 major = XENVBD_MAJOR;
 917
 918                         if (major != XENVBD_MAJOR) {
 919                                 printk(KERN_INFO
 920                                                 "%s: HVM does not support vbd %d as xen block device\n",
 921                                                 __FUNCTION__, vdevice);
 922                                 return -ENODEV;
 923                         }
 924                 }
 925                 /* do not create a PV cdrom device if we are an HVM guest */
 926                 type = xenbus_read(XBT_NIL, dev->nodename, "device-type", &len);
 927                 if (IS_ERR(type))
 928                         return -ENODEV;
 929                 if (strncmp(type, "cdrom", 5) == 0) {
 930                         kfree(type);
 931                         return -ENODEV;
 932                 }
 933                 kfree(type);
 934         }
 935         info = kzalloc(sizeof(*info), GFP_KERNEL);
 936         if (!info) {
 937                 xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
 938                 return -ENOMEM;
 939         }
 940
 941         mutex_init(&info->mutex);
 942         info->xbdev = dev;
 943         info->vdevice = vdevice;
 944         info->connected = BLKIF_STATE_DISCONNECTED;
 945         INIT_WORK(&info->work, blkif_restart_queue);
 946
 947         for (i = 0; i < BLK_RING_SIZE; i++)
 948                 info->shadow[i].req.id = i+1;
 949         info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
 950
 951         /* Front end dir is a number, which is used as the id. */
 952         info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
 953         dev_set_drvdata(&dev->dev, info);
 954
 955         err = talk_to_blkback(dev, info);
 956         if (err) {
 957                 kfree(info);
 958                 dev_set_drvdata(&dev->dev, NULL);
 959                 return err;
 960         }
 961
 962         return 0;
 963 }
 964
 965
 966 static int blkif_recover(struct blkfront_info *info)
 967 {
 968         int i;
 969         struct blkif_request *req;
 970         struct blk_shadow *copy;
 971         int j;
 972
 973         /* Stage 1: Make a safe copy of the shadow state. */
 974         copy = kmalloc(sizeof(info->shadow),
 975                        GFP_NOIO | __GFP_REPEAT | __GFP_HIGH);
 976         if (!copy)
 977                 return -ENOMEM;
 978         memcpy(copy, info->shadow, sizeof(info->shadow));
 979
 980         /* Stage 2: Set up free list. */
 981         memset(&info->shadow, 0, sizeof(info->shadow));
 982         for (i = 0; i < BLK_RING_SIZE; i++)
 983                 info->shadow[i].req.id = i+1;
 984         info->shadow_free = info->ring.req_prod_pvt;
 985         info->shadow[BLK_RING_SIZE-1].req.id = 0x0fffffff;
 986
 987         /* Stage 3: Find pending requests and requeue them. */
 988         for (i = 0; i < BLK_RING_SIZE; i++) {
 989                 /* Not in use? */
 990                 if (!copy[i].request)
 991                         continue;
 992
 993                 /* Grab a request slot and copy shadow state into it. */
 994                 req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
 995                 *req = copy[i].req;
 996
 997                 /* We get a new request id, and must reset the shadow state. */
 998                 req->id = get_id_from_freelist(info);
 999                 memcpy(&info->shadow[req->id], &copy[i], sizeof(copy[i]));
1000
1001                 /* Rewrite any grant references invalidated by susp/resume. */
1002                 for (j = 0; j < req->nr_segments; j++)
1003                         gnttab_grant_foreign_access_ref(
1004                                 req->u.rw.seg[j].gref,
1005                                 info->xbdev->otherend_id,
1006                                 pfn_to_mfn(info->shadow[req->id].frame[j]),
1007                                 rq_data_dir(info->shadow[req->id].request));
1008                 info->shadow[req->id].req = *req;
1009
1010                 info->ring.req_prod_pvt++;
1011         }
1012
1013         kfree(copy);
1014
1015         xenbus_switch_state(info->xbdev, XenbusStateConnected);
1016
1017         spin_lock_irq(&blkif_io_lock);
1018
1019         /* Now safe for us to use the shared ring */
1020         info->connected = BLKIF_STATE_CONNECTED;
1021
1022         /* Send off requeued requests */
1023         flush_requests(info);
1024
1025         /* Kick any other new requests queued since we resumed */
1026         kick_pending_request_queues(info);
1027
1028         spin_unlock_irq(&blkif_io_lock);
1029
1030         return 0;
1031 }
1032
1033 /**
1034  * We are reconnecting to the backend, due to a suspend/resume, or a backend
1035  * driver restart.  We tear down our blkif structure and recreate it, but
1036  * leave the device-layer structures intact so that this is transparent to the
1037  * rest of the kernel.
1038  */
1039 static int blkfront_resume(struct xenbus_device *dev)
1040 {
1041         struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1042         int err;
1043
1044         dev_dbg(&dev->dev, "blkfront_resume: %s\n", dev->nodename);
1045
1046         blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
1047
1048         err = talk_to_blkback(dev, info);
1049         if (info->connected == BLKIF_STATE_SUSPENDED && !err)
1050                 err = blkif_recover(info);
1051
1052         return err;
1053 }
1054
1055 static void
1056 blkfront_closing(struct blkfront_info *info)
1057 {
1058         struct xenbus_device *xbdev = info->xbdev;
1059         struct block_device *bdev = NULL;
1060
1061         mutex_lock(&info->mutex);
1062
1063         if (xbdev->state == XenbusStateClosing) {
1064                 mutex_unlock(&info->mutex);
1065                 return;
1066         }
1067
1068         if (info->gd)
1069                 bdev = bdget_disk(info->gd, 0);
1070
1071         mutex_unlock(&info->mutex);
1072
1073         if (!bdev) {
1074                 xenbus_frontend_closed(xbdev);
1075                 return;
1076         }
1077
1078         mutex_lock(&bdev->bd_mutex);
1079
1080         if (bdev->bd_openers) {
1081                 xenbus_dev_error(xbdev, -EBUSY,
1082                                  "Device in use; refusing to close");
1083                 xenbus_switch_state(xbdev, XenbusStateClosing);
1084         } else {
1085                 xlvbd_release_gendisk(info);
1086                 xenbus_frontend_closed(xbdev);
1087         }
1088
1089         mutex_unlock(&bdev->bd_mutex);
1090         bdput(bdev);
1091 }
1092
1093 /*
1094  * Invoked when the backend is finally 'ready' (and has told produced
1095  * the details about the physical device - #sectors, size, etc).
1096  */
1097 static void blkfront_connect(struct blkfront_info *info)
1098 {
1099         unsigned long long sectors;
1100         unsigned long sector_size;
1101         unsigned int binfo;
1102         int err;
1103         int barrier;
1104
1105         switch (info->connected) {
1106         case BLKIF_STATE_CONNECTED:
1107                 /*
1108                  * Potentially, the back-end may be signalling
1109                  * a capacity change; update the capacity.
1110                  */
1111                 err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
1112                                    "sectors", "%Lu", &sectors);
1113                 if (XENBUS_EXIST_ERR(err))
1114                         return;
1115                 printk(KERN_INFO "Setting capacity to %Lu\n",
1116                        sectors);
1117                 set_capacity(info->gd, sectors);
1118                 revalidate_disk(info->gd);
1119
1120                 /* fall through */
1121         case BLKIF_STATE_SUSPENDED:
1122                 return;
1123
1124         default:
1125                 break;
1126         }
1127
1128         dev_dbg(&info->xbdev->dev, "%s:%s.\n",
1129                 __func__, info->xbdev->otherend);
1130
1131         err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1132                             "sectors", "%llu", &sectors,
1133                             "info", "%u", &binfo,
1134                             "sector-size", "%lu", &sector_size,
1135                             NULL);
1136         if (err) {
1137                 xenbus_dev_fatal(info->xbdev, err,
1138                                  "reading backend fields at %s",
1139                                  info->xbdev->otherend);
1140                 return;
1141         }
1142
1143         err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1144                             "feature-barrier", "%lu", &barrier,
1145                             NULL);
1146
1147         /*
1148          * If there's no "feature-barrier" defined, then it means
1149          * we're dealing with a very old backend which writes
1150          * synchronously; nothing to do.
1151          *
1152          * If there are barriers, then we use flush.
1153          */
1154         info->feature_flush = 0;
1155
1156         if (!err && barrier)
1157                 info->feature_flush = REQ_FLUSH | REQ_FUA;
1158
1159         err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
1160         if (err) {
1161                 xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
1162                                  info->xbdev->otherend);
1163                 return;
1164         }
1165
1166         xenbus_switch_state(info->xbdev, XenbusStateConnected);
1167
1168         /* Kick pending requests. */
1169         spin_lock_irq(&blkif_io_lock);
1170         info->connected = BLKIF_STATE_CONNECTED;
1171         kick_pending_request_queues(info);
1172         spin_unlock_irq(&blkif_io_lock);
1173
1174         add_disk(info->gd);
1175
1176         info->is_ready = 1;
1177 }
1178
1179 /**
1180  * Callback received when the backend's state changes.
1181  */
1182 static void blkback_changed(struct xenbus_device *dev,
1183                             enum xenbus_state backend_state)
1184 {
1185         struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1186
1187         dev_dbg(&dev->dev, "blkfront:blkback_changed to state %d.\n", backend_state);
1188
1189         switch (backend_state) {
1190         case XenbusStateInitialising:
1191         case XenbusStateInitWait:
1192         case XenbusStateInitialised:
1193         case XenbusStateReconfiguring:
1194         case XenbusStateReconfigured:
1195         case XenbusStateUnknown:
1196         case XenbusStateClosed:
1197                 break;
1198
1199         case XenbusStateConnected:
1200                 blkfront_connect(info);
1201                 break;
1202
1203         case XenbusStateClosing:
1204                 blkfront_closing(info);
1205                 break;
1206         }
1207 }
1208
1209 static int blkfront_remove(struct xenbus_device *xbdev)
1210 {
1211         struct blkfront_info *info = dev_get_drvdata(&xbdev->dev);
1212         struct block_device *bdev = NULL;
1213         struct gendisk *disk;
1214
1215         dev_dbg(&xbdev->dev, "%s removed", xbdev->nodename);
1216
1217         blkif_free(info, 0);
1218
1219         mutex_lock(&info->mutex);
1220
1221         disk = info->gd;
1222         if (disk)
1223                 bdev = bdget_disk(disk, 0);
1224
1225         info->xbdev = NULL;
1226         mutex_unlock(&info->mutex);
1227
1228         if (!bdev) {
1229                 kfree(info);
1230                 return 0;
1231         }
1232
1233         /*
1234          * The xbdev was removed before we reached the Closed
1235          * state. See if it's safe to remove the disk. If the bdev
1236          * isn't closed yet, we let release take care of it.
1237          */
1238
1239         mutex_lock(&bdev->bd_mutex);
1240         info = disk->private_data;
1241
1242         dev_warn(disk_to_dev(disk),
1243                  "%s was hot-unplugged, %d stale handles\n",
1244                  xbdev->nodename, bdev->bd_openers);
1245
1246         if (info && !bdev->bd_openers) {
1247                 xlvbd_release_gendisk(info);
1248                 disk->private_data = NULL;
1249                 kfree(info);
1250         }
1251
1252         mutex_unlock(&bdev->bd_mutex);
1253         bdput(bdev);
1254
1255         return 0;
1256 }
1257
1258 static int blkfront_is_ready(struct xenbus_device *dev)
1259 {
1260         struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1261
1262         return info->is_ready && info->xbdev;
1263 }
1264
1265 static int blkif_open(struct block_device *bdev, fmode_t mode)
1266 {
1267         struct gendisk *disk = bdev->bd_disk;
1268         struct blkfront_info *info;
1269         int err = 0;
1270
1271         mutex_lock(&blkfront_mutex);
1272
1273         info = disk->private_data;
1274         if (!info) {
1275                 /* xbdev gone */
1276                 err = -ERESTARTSYS;
1277                 goto out;
1278         }
1279
1280         mutex_lock(&info->mutex);
1281
1282         if (!info->gd)
1283                 /* xbdev is closed */
1284                 err = -ERESTARTSYS;
1285
1286         mutex_unlock(&info->mutex);
1287
1288 out:
1289         mutex_unlock(&blkfront_mutex);
1290         return err;
1291 }
1292
1293 static int blkif_release(struct gendisk *disk, fmode_t mode)
1294 {
1295         struct blkfront_info *info = disk->private_data;
1296         struct block_device *bdev;
1297         struct xenbus_device *xbdev;
1298
1299         mutex_lock(&blkfront_mutex);
1300
1301         bdev = bdget_disk(disk, 0);
1302         bdput(bdev);
1303
1304         if (bdev->bd_openers)
1305                 goto out;
1306
1307         /*
1308          * Check if we have been instructed to close. We will have
1309          * deferred this request, because the bdev was still open.
1310          */
1311
1312         mutex_lock(&info->mutex);
1313         xbdev = info->xbdev;
1314
1315         if (xbdev && xbdev->state == XenbusStateClosing) {
1316                 /* pending switch to state closed */
1317                 dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
1318                 xlvbd_release_gendisk(info);
1319                 xenbus_frontend_closed(info->xbdev);
1320         }
1321
1322         mutex_unlock(&info->mutex);
1323
1324         if (!xbdev) {
1325                 /* sudden device removal */
1326                 dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
1327                 xlvbd_release_gendisk(info);
1328                 disk->private_data = NULL;
1329                 kfree(info);
1330         }
1331
1332 out:
1333         mutex_unlock(&blkfront_mutex);
1334         return 0;
1335 }
1336
1337 static const struct block_device_operations xlvbd_block_fops =
1338 {
1339         .owner = THIS_MODULE,
1340         .open = blkif_open,
1341         .release = blkif_release,
1342         .getgeo = blkif_getgeo,
1343         .ioctl = blkif_ioctl,
1344 };
1345
1346
1347 static const struct xenbus_device_id blkfront_ids[] = {
1348         { "vbd" },
1349         { "" }
1350 };
1351
1352 static struct xenbus_driver blkfront = {
1353         .name = "vbd",
1354         .owner = THIS_MODULE,
1355         .ids = blkfront_ids,
1356         .probe = blkfront_probe,
1357         .remove = blkfront_remove,
1358         .resume = blkfront_resume,
1359         .otherend_changed = blkback_changed,
1360         .is_ready = blkfront_is_ready,
1361 };
1362
1363 static int __init xlblk_init(void)
1364 {
1365         if (!xen_domain())
1366                 return -ENODEV;
1367
1368         if (register_blkdev(XENVBD_MAJOR, DEV_NAME)) {
1369                 printk(KERN_WARNING "xen_blk: can't get major %d with name %s\n",
1370                        XENVBD_MAJOR, DEV_NAME);
1371                 return -ENODEV;
1372         }
1373
1374         return xenbus_register_frontend(&blkfront);
1375 }
1376 module_init(xlblk_init);
1377
1378
1379 static void __exit xlblk_exit(void)
1380 {
1381         return xenbus_unregister_driver(&blkfront);
1382 }
1383 module_exit(xlblk_exit);
1384
1385 MODULE_DESCRIPTION("Xen virtual block device frontend");
1386 MODULE_LICENSE("GPL");
1387 MODULE_ALIAS_BLOCKDEV_MAJOR(XENVBD_MAJOR);
1388 MODULE_ALIAS("xen:vbd");
1389 MODULE_ALIAS("xenblk");