xen-blkfront.c 72.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39
/*
 * blkfront.c
 *
 * XenLinux virtual block device driver.
 *
 * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
 * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
 * Copyright (c) 2004, Christian Limpach
 * Copyright (c) 2004, Andrew Warfield
 * Copyright (c) 2005, Christopher Clark
 * Copyright (c) 2005, XenSource Ltd
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version 2
 * as published by the Free Software Foundation; or, when distributed
 * separately from the Linux kernel or incorporated into other
 * software packages, subject to the following license:
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this source file (the "Software"), to deal in the Software without
 * restriction, including without limitation the rights to use, copy, modify,
 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
 * and to permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 */

#include <linux/interrupt.h>
#include <linux/blkdev.h>
B
Bob Liu 已提交
40
#include <linux/blk-mq.h>
41
#include <linux/hdreg.h>
42
#include <linux/cdrom.h>
43
#include <linux/module.h>
44
#include <linux/slab.h>
45
#include <linux/mutex.h>
46
#include <linux/scatterlist.h>
47
#include <linux/bitmap.h>
48
#include <linux/list.h>
49

50
#include <xen/xen.h>
51 52 53 54
#include <xen/xenbus.h>
#include <xen/grant_table.h>
#include <xen/events.h>
#include <xen/page.h>
55
#include <xen/platform_pci.h>
56 57 58

#include <xen/interface/grant_table.h>
#include <xen/interface/io/blkif.h>
59
#include <xen/interface/io/protocols.h>
60 61 62

#include <asm/xen/hypervisor.h>

63 64 65 66 67 68 69 70 71 72 73 74 75 76
/*
 * The minimal size of segment supported by the block framework is PAGE_SIZE.
 * When Linux is using a different page size than Xen, it may not be possible
 * to put all the data in a single segment.
 * This can happen when the backend doesn't support indirect descriptor and
 * therefore the maximum amount of data that a request can carry is
 * BLKIF_MAX_SEGMENTS_PER_REQUEST * XEN_PAGE_SIZE = 44KB
 *
 * Note that we only support one extra request. So the Linux page size
 * should be <= ( 2 * BLKIF_MAX_SEGMENTS_PER_REQUEST * XEN_PAGE_SIZE) =
 * 88KB.
 */
#define HAS_EXTRA_REQ (BLKIF_MAX_SEGMENTS_PER_REQUEST < XEN_PFN_PER_PAGE)

77 78 79 80 81 82
enum blkif_state {
	BLKIF_STATE_DISCONNECTED,
	BLKIF_STATE_CONNECTED,
	BLKIF_STATE_SUSPENDED,
};

83 84
struct grant {
	grant_ref_t gref;
85
	struct page *page;
86
	struct list_head node;
87 88
};

89 90 91 92 93 94 95
enum blk_req_status {
	REQ_WAITING,
	REQ_DONE,
	REQ_ERROR,
	REQ_EOPNOTSUPP,
};

96 97
struct blk_shadow {
	struct blkif_request req;
98
	struct request *request;
99 100
	struct grant **grants_used;
	struct grant **indirect_grants;
101
	struct scatterlist *sg;
102
	unsigned int num_sg;
103 104 105 106 107 108 109 110
	enum blk_req_status status;

	#define NO_ASSOCIATED_ID ~0UL
	/*
	 * Id of the sibling if we ever need 2 requests when handling a
	 * block I/O request
	 */
	unsigned long associated_id;
111 112 113 114 115
};

struct split_bio {
	struct bio *bio;
	atomic_t pending;
116 117
};

118
static DEFINE_MUTEX(blkfront_mutex);
119
static const struct block_device_operations xlvbd_block_fops;
120

121 122 123 124 125 126 127
/*
 * Maximum number of segments in indirect requests, the actual value used by
 * the frontend driver is the minimum of this value and the value provided
 * by the backend driver.
 */

static unsigned int xen_blkif_max_segments = 32;
128 129 130 131
module_param_named(max_indirect_segments, xen_blkif_max_segments, uint,
		   S_IRUGO);
MODULE_PARM_DESC(max_indirect_segments,
		 "Maximum amount of segments in indirect requests (default is 32)");
132

133 134 135 136
static unsigned int xen_blkif_max_queues = 4;
module_param_named(max_queues, xen_blkif_max_queues, uint, S_IRUGO);
MODULE_PARM_DESC(max_queues, "Maximum number of hardware queues/rings used per virtual disk");

B
Bob Liu 已提交
137 138 139 140 141 142 143 144
/*
 * Maximum order of pages to be used for the shared ring between front and
 * backend, 4KB page granularity is used.
 */
static unsigned int xen_blkif_max_ring_order;
module_param_named(max_ring_page_order, xen_blkif_max_ring_order, int, S_IRUGO);
MODULE_PARM_DESC(max_ring_page_order, "Maximum order of pages to be used for the shared ring");

145 146 147 148
#define BLK_RING_SIZE(info)	\
	__CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * (info)->nr_ring_pages)

#define BLK_MAX_RING_SIZE	\
149
	__CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * XENBUS_MAX_RING_GRANTS)
150

B
Bob Liu 已提交
151
/*
152 153
 * ring-ref%u i=(-1UL) would take 11 characters + 'ring-ref' is 8, so 19
 * characters are enough. Define to 20 to keep consistent with backend.
B
Bob Liu 已提交
154 155
 */
#define RINGREF_NAME_LEN (20)
156 157 158 159
/*
 * queue-%u would take 7 + 10(UINT_MAX) = 17 characters.
 */
#define QUEUE_NAME_LEN (17)
160

161 162 163 164 165 166
/*
 *  Per-ring info.
 *  Every blkfront device can associate with one or more blkfront_ring_info,
 *  depending on how many hardware queues/rings to be used.
 */
struct blkfront_ring_info {
B
Bob Liu 已提交
167 168
	/* Lock to protect data in every ring buffer. */
	spinlock_t ring_lock;
169 170 171 172 173 174 175
	struct blkif_front_ring ring;
	unsigned int ring_ref[XENBUS_MAX_RING_GRANTS];
	unsigned int evtchn, irq;
	struct work_struct work;
	struct gnttab_free_callback callback;
	struct blk_shadow shadow[BLK_MAX_RING_SIZE];
	struct list_head indirect_pages;
176 177
	struct list_head grants;
	unsigned int persistent_gnts_c;
178 179 180 181
	unsigned long shadow_free;
	struct blkfront_info *dev_info;
};

182 183 184 185 186 187 188
/*
 * We have one of these per vbd, whether ide, scsi or 'other'.  They
 * hang in private_data off the gendisk structure. We may end up
 * putting all kinds of interesting stuff here :-)
 */
struct blkfront_info
{
189
	struct mutex mutex;
190 191
	struct xenbus_device *xbdev;
	struct gendisk *gd;
192 193
	u16 sector_size;
	unsigned int physical_sector_size;
194 195 196
	int vdevice;
	blkif_vdev_t handle;
	enum blkif_state connected;
197
	/* Number of pages per ring buffer. */
B
Bob Liu 已提交
198
	unsigned int nr_ring_pages;
199
	struct request_queue *rq;
200
	unsigned int feature_flush;
201
	unsigned int feature_fua;
202 203
	unsigned int feature_discard:1;
	unsigned int feature_secdiscard:1;
204 205
	unsigned int discard_granularity;
	unsigned int discard_alignment;
206
	unsigned int feature_persistent:1;
207
	/* Number of 4KB segments handled */
208
	unsigned int max_indirect_segments;
209
	int is_ready;
B
Bob Liu 已提交
210
	struct blk_mq_tag_set tag_set;
211 212
	struct blkfront_ring_info *rinfo;
	unsigned int nr_rings;
213 214 215
	/* Save uncomplete reqs and bios for migration. */
	struct list_head requests;
	struct bio_list bio_list;
216 217
};

218 219 220 221
static unsigned int nr_minors;
static unsigned long *minors;
static DEFINE_SPINLOCK(minor_lock);

222 223 224
#define GRANT_INVALID_REF	0

#define PARTS_PER_DISK		16
225
#define PARTS_PER_EXT_DISK      256
226 227 228 229

#define BLKIF_MAJOR(dev) ((dev)>>8)
#define BLKIF_MINOR(dev) ((dev) & 0xff)

230 231 232 233
#define EXT_SHIFT 28
#define EXTENDED (1<<EXT_SHIFT)
#define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
#define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
234 235
#define EMULATED_HD_DISK_MINOR_OFFSET (0)
#define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
236 237
#define EMULATED_SD_DISK_MINOR_OFFSET (0)
#define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_SD_DISK_MINOR_OFFSET / 256)
238

239
#define DEV_NAME	"xvd"	/* name in /dev */
240

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
/*
 * Grants are always the same size as a Xen page (i.e 4KB).
 * A physical segment is always the same size as a Linux page.
 * Number of grants per physical segment
 */
#define GRANTS_PER_PSEG	(PAGE_SIZE / XEN_PAGE_SIZE)

#define GRANTS_PER_INDIRECT_FRAME \
	(XEN_PAGE_SIZE / sizeof(struct blkif_request_segment))

#define PSEGS_PER_INDIRECT_FRAME	\
	(GRANTS_INDIRECT_FRAME / GRANTS_PSEGS)

#define INDIRECT_GREFS(_grants)		\
	DIV_ROUND_UP(_grants, GRANTS_PER_INDIRECT_FRAME)

#define GREFS(_psegs)	((_psegs) * GRANTS_PER_PSEG)
258

259
static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo);
260
static void blkfront_gather_backend_features(struct blkfront_info *info);
261

262
static int get_id_from_freelist(struct blkfront_ring_info *rinfo)
263
{
264 265 266 267 268
	unsigned long free = rinfo->shadow_free;

	BUG_ON(free >= BLK_RING_SIZE(rinfo->dev_info));
	rinfo->shadow_free = rinfo->shadow[free].req.u.rw.id;
	rinfo->shadow[free].req.u.rw.id = 0x0fffffee; /* debug */
269 270 271
	return free;
}

272
static int add_id_to_freelist(struct blkfront_ring_info *rinfo,
273
			      unsigned long id)
274
{
275
	if (rinfo->shadow[id].req.u.rw.id != id)
276
		return -EINVAL;
277
	if (rinfo->shadow[id].request == NULL)
278
		return -EINVAL;
279 280 281
	rinfo->shadow[id].req.u.rw.id  = rinfo->shadow_free;
	rinfo->shadow[id].request = NULL;
	rinfo->shadow_free = id;
282
	return 0;
283 284
}

285
static int fill_grant_buffer(struct blkfront_ring_info *rinfo, int num)
286
{
287
	struct blkfront_info *info = rinfo->dev_info;
288 289 290 291
	struct page *granted_page;
	struct grant *gnt_list_entry, *n;
	int i = 0;

292
	while (i < num) {
293 294 295 296
		gnt_list_entry = kzalloc(sizeof(struct grant), GFP_NOIO);
		if (!gnt_list_entry)
			goto out_of_memory;

297 298 299 300 301 302
		if (info->feature_persistent) {
			granted_page = alloc_page(GFP_NOIO);
			if (!granted_page) {
				kfree(gnt_list_entry);
				goto out_of_memory;
			}
303
			gnt_list_entry->page = granted_page;
304 305 306
		}

		gnt_list_entry->gref = GRANT_INVALID_REF;
307
		list_add(&gnt_list_entry->node, &rinfo->grants);
308 309 310 311 312 313 314
		i++;
	}

	return 0;

out_of_memory:
	list_for_each_entry_safe(gnt_list_entry, n,
315
	                         &rinfo->grants, node) {
316
		list_del(&gnt_list_entry->node);
317
		if (info->feature_persistent)
318
			__free_page(gnt_list_entry->page);
319 320 321 322 323 324 325
		kfree(gnt_list_entry);
		i--;
	}
	BUG_ON(i != 0);
	return -ENOMEM;
}

326
static struct grant *get_free_grant(struct blkfront_ring_info *rinfo)
327 328 329
{
	struct grant *gnt_list_entry;

330 331
	BUG_ON(list_empty(&rinfo->grants));
	gnt_list_entry = list_first_entry(&rinfo->grants, struct grant,
332
					  node);
333 334
	list_del(&gnt_list_entry->node);

335
	if (gnt_list_entry->gref != GRANT_INVALID_REF)
336
		rinfo->persistent_gnts_c--;
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351

	return gnt_list_entry;
}

static inline void grant_foreign_access(const struct grant *gnt_list_entry,
					const struct blkfront_info *info)
{
	gnttab_page_grant_foreign_access_ref_one(gnt_list_entry->gref,
						 info->xbdev->otherend_id,
						 gnt_list_entry->page,
						 0);
}

static struct grant *get_grant(grant_ref_t *gref_head,
			       unsigned long gfn,
352
			       struct blkfront_ring_info *rinfo)
353
{
354 355
	struct grant *gnt_list_entry = get_free_grant(rinfo);
	struct blkfront_info *info = rinfo->dev_info;
356 357

	if (gnt_list_entry->gref != GRANT_INVALID_REF)
358
		return gnt_list_entry;
359 360 361 362 363 364 365 366 367 368 369

	/* Assign a gref to this page */
	gnt_list_entry->gref = gnttab_claim_grant_reference(gref_head);
	BUG_ON(gnt_list_entry->gref == -ENOSPC);
	if (info->feature_persistent)
		grant_foreign_access(gnt_list_entry, info);
	else {
		/* Grant access to the GFN passed by the caller */
		gnttab_grant_foreign_access_ref(gnt_list_entry->gref,
						info->xbdev->otherend_id,
						gfn, 0);
370 371
	}

372 373 374 375
	return gnt_list_entry;
}

static struct grant *get_indirect_grant(grant_ref_t *gref_head,
376
					struct blkfront_ring_info *rinfo)
377
{
378 379
	struct grant *gnt_list_entry = get_free_grant(rinfo);
	struct blkfront_info *info = rinfo->dev_info;
380 381 382 383

	if (gnt_list_entry->gref != GRANT_INVALID_REF)
		return gnt_list_entry;

384 385 386
	/* Assign a gref to this page */
	gnt_list_entry->gref = gnttab_claim_grant_reference(gref_head);
	BUG_ON(gnt_list_entry->gref == -ENOSPC);
387
	if (!info->feature_persistent) {
388 389 390
		struct page *indirect_page;

		/* Fetch a pre-allocated page to use for indirect grefs */
391 392
		BUG_ON(list_empty(&rinfo->indirect_pages));
		indirect_page = list_first_entry(&rinfo->indirect_pages,
393 394 395
						 struct page, lru);
		list_del(&indirect_page->lru);
		gnt_list_entry->page = indirect_page;
396
	}
397 398
	grant_foreign_access(gnt_list_entry, info);

399 400 401
	return gnt_list_entry;
}

402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
static const char *op_name(int op)
{
	static const char *const names[] = {
		[BLKIF_OP_READ] = "read",
		[BLKIF_OP_WRITE] = "write",
		[BLKIF_OP_WRITE_BARRIER] = "barrier",
		[BLKIF_OP_FLUSH_DISKCACHE] = "flush",
		[BLKIF_OP_DISCARD] = "discard" };

	if (op < 0 || op >= ARRAY_SIZE(names))
		return "unknown";

	if (!names[op])
		return "reserved";

	return names[op];
}
419 420 421 422 423 424 425 426
static int xlbd_reserve_minors(unsigned int minor, unsigned int nr)
{
	unsigned int end = minor + nr;
	int rc;

	if (end > nr_minors) {
		unsigned long *bitmap, *old;

427
		bitmap = kcalloc(BITS_TO_LONGS(end), sizeof(*bitmap),
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446
				 GFP_KERNEL);
		if (bitmap == NULL)
			return -ENOMEM;

		spin_lock(&minor_lock);
		if (end > nr_minors) {
			old = minors;
			memcpy(bitmap, minors,
			       BITS_TO_LONGS(nr_minors) * sizeof(*bitmap));
			minors = bitmap;
			nr_minors = BITS_TO_LONGS(end) * BITS_PER_LONG;
		} else
			old = bitmap;
		spin_unlock(&minor_lock);
		kfree(old);
	}

	spin_lock(&minor_lock);
	if (find_next_bit(minors, end, minor) >= end) {
447
		bitmap_set(minors, minor, nr);
448 449 450 451 452 453 454 455 456 457 458 459 460 461
		rc = 0;
	} else
		rc = -EBUSY;
	spin_unlock(&minor_lock);

	return rc;
}

static void xlbd_release_minors(unsigned int minor, unsigned int nr)
{
	unsigned int end = minor + nr;

	BUG_ON(end > nr_minors);
	spin_lock(&minor_lock);
462
	bitmap_clear(minors,  minor, nr);
463 464 465
	spin_unlock(&minor_lock);
}

466 467
static void blkif_restart_queue_callback(void *arg)
{
468 469
	struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)arg;
	schedule_work(&rinfo->work);
470 471
}

H
Harvey Harrison 已提交
472
static int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
{
	/* We don't have real geometry info, but let's at least return
	   values consistent with the size of the device */
	sector_t nsect = get_capacity(bd->bd_disk);
	sector_t cylinders = nsect;

	hg->heads = 0xff;
	hg->sectors = 0x3f;
	sector_div(cylinders, hg->heads * hg->sectors);
	hg->cylinders = cylinders;
	if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
		hg->cylinders = 0xffff;
	return 0;
}

A
Al Viro 已提交
488
static int blkif_ioctl(struct block_device *bdev, fmode_t mode,
489
		       unsigned command, unsigned long argument)
490
{
A
Al Viro 已提交
491
	struct blkfront_info *info = bdev->bd_disk->private_data;
492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
	int i;

	dev_dbg(&info->xbdev->dev, "command: 0x%x, argument: 0x%lx\n",
		command, (long)argument);

	switch (command) {
	case CDROMMULTISESSION:
		dev_dbg(&info->xbdev->dev, "FIXME: support multisession CDs later\n");
		for (i = 0; i < sizeof(struct cdrom_multisession); i++)
			if (put_user(0, (char __user *)(argument + i)))
				return -EFAULT;
		return 0;

	case CDROM_GET_CAPABILITY: {
		struct gendisk *gd = info->gd;
		if (gd->flags & GENHD_FL_CD)
			return 0;
		return -EINVAL;
	}

	default:
		/*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
		  command);*/
		return -EINVAL; /* same return as native Linux */
	}

	return 0;
}

521 522 523 524 525 526 527 528 529 530 531
static unsigned long blkif_ring_get_request(struct blkfront_ring_info *rinfo,
					    struct request *req,
					    struct blkif_request **ring_req)
{
	unsigned long id;

	*ring_req = RING_GET_REQUEST(&rinfo->ring, rinfo->ring.req_prod_pvt);
	rinfo->ring.req_prod_pvt++;

	id = get_id_from_freelist(rinfo);
	rinfo->shadow[id].request = req;
532 533
	rinfo->shadow[id].status = REQ_WAITING;
	rinfo->shadow[id].associated_id = NO_ASSOCIATED_ID;
534 535 536 537 538 539

	(*ring_req)->u.rw.id = id;

	return id;
}

540
static int blkif_queue_discard_req(struct request *req, struct blkfront_ring_info *rinfo)
541
{
542
	struct blkfront_info *info = rinfo->dev_info;
543 544
	struct blkif_request *ring_req;
	unsigned long id;
545 546

	/* Fill out a communications ring structure. */
547
	id = blkif_ring_get_request(rinfo, req, &ring_req);
548 549 550 551 552

	ring_req->operation = BLKIF_OP_DISCARD;
	ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
	ring_req->u.discard.id = id;
	ring_req->u.discard.sector_number = (blkif_sector_t)blk_rq_pos(req);
553
	if (req_op(req) == REQ_OP_SECURE_ERASE && info->feature_secdiscard)
554 555 556 557 558
		ring_req->u.discard.flag = BLKIF_DISCARD_SECURE;
	else
		ring_req->u.discard.flag = 0;

	/* Keep a private copy so we can reissue requests when recovering. */
559
	rinfo->shadow[id].req = *ring_req;
560 561 562 563

	return 0;
}

564 565 566
struct setup_rw_req {
	unsigned int grant_idx;
	struct blkif_request_segment *segments;
567
	struct blkfront_ring_info *rinfo;
568 569 570 571 572 573 574
	struct blkif_request *ring_req;
	grant_ref_t gref_head;
	unsigned int id;
	/* Only used when persistent grant is used and it's a read request */
	bool need_copy;
	unsigned int bvec_off;
	char *bvec_data;
575 576 577

	bool require_extra_req;
	struct blkif_request *extra_ring_req;
578 579 580 581 582 583 584 585
};

static void blkif_setup_rw_req_grant(unsigned long gfn, unsigned int offset,
				     unsigned int len, void *data)
{
	struct setup_rw_req *setup = data;
	int n, ref;
	struct grant *gnt_list_entry;
586
	unsigned int fsect, lsect;
587 588 589
	/* Convenient aliases */
	unsigned int grant_idx = setup->grant_idx;
	struct blkif_request *ring_req = setup->ring_req;
590
	struct blkfront_ring_info *rinfo = setup->rinfo;
591 592 593 594 595 596
	/*
	 * We always use the shadow of the first request to store the list
	 * of grant associated to the block I/O request. This made the
	 * completion more easy to handle even if the block I/O request is
	 * split.
	 */
597
	struct blk_shadow *shadow = &rinfo->shadow[setup->id];
598

599 600 601 602 603 604 605 606 607 608
	if (unlikely(setup->require_extra_req &&
		     grant_idx >= BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
		/*
		 * We are using the second request, setup grant_idx
		 * to be the index of the segment array.
		 */
		grant_idx -= BLKIF_MAX_SEGMENTS_PER_REQUEST;
		ring_req = setup->extra_ring_req;
	}

609 610 611 612 613 614
	if ((ring_req->operation == BLKIF_OP_INDIRECT) &&
	    (grant_idx % GRANTS_PER_INDIRECT_FRAME == 0)) {
		if (setup->segments)
			kunmap_atomic(setup->segments);

		n = grant_idx / GRANTS_PER_INDIRECT_FRAME;
615
		gnt_list_entry = get_indirect_grant(&setup->gref_head, rinfo);
616 617 618 619 620
		shadow->indirect_grants[n] = gnt_list_entry;
		setup->segments = kmap_atomic(gnt_list_entry->page);
		ring_req->u.indirect.indirect_grefs[n] = gnt_list_entry->gref;
	}

621
	gnt_list_entry = get_grant(&setup->gref_head, gfn, rinfo);
622
	ref = gnt_list_entry->gref;
623 624 625 626 627
	/*
	 * All the grants are stored in the shadow of the first
	 * request. Therefore we have to use the global index.
	 */
	shadow->grants_used[setup->grant_idx] = gnt_list_entry;
628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668

	if (setup->need_copy) {
		void *shared_data;

		shared_data = kmap_atomic(gnt_list_entry->page);
		/*
		 * this does not wipe data stored outside the
		 * range sg->offset..sg->offset+sg->length.
		 * Therefore, blkback *could* see data from
		 * previous requests. This is OK as long as
		 * persistent grants are shared with just one
		 * domain. It may need refactoring if this
		 * changes
		 */
		memcpy(shared_data + offset,
		       setup->bvec_data + setup->bvec_off,
		       len);

		kunmap_atomic(shared_data);
		setup->bvec_off += len;
	}

	fsect = offset >> 9;
	lsect = fsect + (len >> 9) - 1;
	if (ring_req->operation != BLKIF_OP_INDIRECT) {
		ring_req->u.rw.seg[grant_idx] =
			(struct blkif_request_segment) {
				.gref       = ref,
				.first_sect = fsect,
				.last_sect  = lsect };
	} else {
		setup->segments[grant_idx % GRANTS_PER_INDIRECT_FRAME] =
			(struct blkif_request_segment) {
				.gref       = ref,
				.first_sect = fsect,
				.last_sect  = lsect };
	}

	(setup->grant_idx)++;
}

669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687
static void blkif_setup_extra_req(struct blkif_request *first,
				  struct blkif_request *second)
{
	uint16_t nr_segments = first->u.rw.nr_segments;

	/*
	 * The second request is only present when the first request uses
	 * all its segments. It's always the continuity of the first one.
	 */
	first->u.rw.nr_segments = BLKIF_MAX_SEGMENTS_PER_REQUEST;

	second->u.rw.nr_segments = nr_segments - BLKIF_MAX_SEGMENTS_PER_REQUEST;
	second->u.rw.sector_number = first->u.rw.sector_number +
		(BLKIF_MAX_SEGMENTS_PER_REQUEST * XEN_PAGE_SIZE) / 512;

	second->u.rw.handle = first->u.rw.handle;
	second->operation = first->operation;
}

688
static int blkif_queue_rw_req(struct request *req, struct blkfront_ring_info *rinfo)
689
{
690
	struct blkfront_info *info = rinfo->dev_info;
691 692 693
	struct blkif_request *ring_req, *extra_ring_req = NULL;
	unsigned long id, extra_id = NO_ASSOCIATED_ID;
	bool require_extra_req = false;
694 695 696 697
	int i;
	struct setup_rw_req setup = {
		.grant_idx = 0,
		.segments = NULL,
698
		.rinfo = rinfo,
699 700
		.need_copy = rq_data_dir(req) && info->feature_persistent,
	};
701 702 703 704 705 706

	/*
	 * Used to store if we are able to queue the request by just using
	 * existing persistent grants, or if we have to get new grants,
	 * as there are not sufficiently many free.
	 */
707
	struct scatterlist *sg;
708
	int num_sg, max_grefs, num_grant;
709

710
	max_grefs = req->nr_phys_segments * GRANTS_PER_PSEG;
711 712 713 714 715
	if (max_grefs > BLKIF_MAX_SEGMENTS_PER_REQUEST)
		/*
		 * If we are using indirect segments we need to account
		 * for the indirect grefs used in the request.
		 */
716
		max_grefs += INDIRECT_GREFS(max_grefs);
717

718 719 720 721 722 723
	/*
	 * We have to reserve 'max_grefs' grants because persistent
	 * grants are shared by all rings.
	 */
	if (max_grefs > 0)
		if (gnttab_alloc_grant_references(max_grefs, &setup.gref_head) < 0) {
724
			gnttab_request_free_callback(
725
				&rinfo->callback,
726
				blkif_restart_queue_callback,
727
				rinfo,
728
				max_grefs);
729 730
			return 1;
		}
731 732

	/* Fill out a communications ring structure. */
733
	id = blkif_ring_get_request(rinfo, req, &ring_req);
734

735
	num_sg = blk_rq_map_sg(req->q, req, rinfo->shadow[id].sg);
736 737
	num_grant = 0;
	/* Calculate the number of grant used */
738
	for_each_sg(rinfo->shadow[id].sg, sg, num_sg, i)
739 740
	       num_grant += gnttab_count_grant(sg->offset, sg->length);

741 742 743 744
	require_extra_req = info->max_indirect_segments == 0 &&
		num_grant > BLKIF_MAX_SEGMENTS_PER_REQUEST;
	BUG_ON(!HAS_EXTRA_REQ && require_extra_req);

745
	rinfo->shadow[id].num_sg = num_sg;
746 747
	if (num_grant > BLKIF_MAX_SEGMENTS_PER_REQUEST &&
	    likely(!require_extra_req)) {
748 749 750 751
		/*
		 * The indirect operation can only be a BLKIF_OP_READ or
		 * BLKIF_OP_WRITE
		 */
752
		BUG_ON(req_op(req) == REQ_OP_FLUSH || req->cmd_flags & REQ_FUA);
753 754 755 756 757
		ring_req->operation = BLKIF_OP_INDIRECT;
		ring_req->u.indirect.indirect_op = rq_data_dir(req) ?
			BLKIF_OP_WRITE : BLKIF_OP_READ;
		ring_req->u.indirect.sector_number = (blkif_sector_t)blk_rq_pos(req);
		ring_req->u.indirect.handle = info->handle;
758
		ring_req->u.indirect.nr_segments = num_grant;
759
	} else {
760 761 762 763
		ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req);
		ring_req->u.rw.handle = info->handle;
		ring_req->operation = rq_data_dir(req) ?
			BLKIF_OP_WRITE : BLKIF_OP_READ;
764
		if (req_op(req) == REQ_OP_FLUSH || req->cmd_flags & REQ_FUA) {
765
			/*
766 767 768 769 770
			 * Ideally we can do an unordered flush-to-disk.
			 * In case the backend onlysupports barriers, use that.
			 * A barrier request a superset of FUA, so we can
			 * implement it the same way.  (It's also a FLUSH+FUA,
			 * since it is guaranteed ordered WRT previous writes.)
771
			 */
772
			if (info->feature_flush && info->feature_fua)
773 774
				ring_req->operation =
					BLKIF_OP_WRITE_BARRIER;
775
			else if (info->feature_flush)
776 777
				ring_req->operation =
					BLKIF_OP_FLUSH_DISKCACHE;
778
			else
779
				ring_req->operation = 0;
780
		}
781
		ring_req->u.rw.nr_segments = num_grant;
782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
		if (unlikely(require_extra_req)) {
			extra_id = blkif_ring_get_request(rinfo, req,
							  &extra_ring_req);
			/*
			 * Only the first request contains the scatter-gather
			 * list.
			 */
			rinfo->shadow[extra_id].num_sg = 0;

			blkif_setup_extra_req(ring_req, extra_ring_req);

			/* Link the 2 requests together */
			rinfo->shadow[extra_id].associated_id = id;
			rinfo->shadow[id].associated_id = extra_id;
		}
797
	}
798

799 800
	setup.ring_req = ring_req;
	setup.id = id;
801 802 803 804 805

	setup.require_extra_req = require_extra_req;
	if (unlikely(require_extra_req))
		setup.extra_ring_req = extra_ring_req;

806
	for_each_sg(rinfo->shadow[id].sg, sg, num_sg, i) {
807
		BUG_ON(sg->offset + sg->length > PAGE_SIZE);
808

809 810 811 812
		if (setup.need_copy) {
			setup.bvec_off = sg->offset;
			setup.bvec_data = kmap_atomic(sg_page(sg));
		}
813

814 815 816 817 818
		gnttab_foreach_grant_in_range(sg_page(sg),
					      sg->offset,
					      sg->length,
					      blkif_setup_rw_req_grant,
					      &setup);
819

820 821
		if (setup.need_copy)
			kunmap_atomic(setup.bvec_data);
822
	}
823 824
	if (setup.segments)
		kunmap_atomic(setup.segments);
825 826

	/* Keep a private copy so we can reissue requests when recovering. */
827
	rinfo->shadow[id].req = *ring_req;
828 829
	if (unlikely(require_extra_req))
		rinfo->shadow[extra_id].req = *extra_ring_req;
830

831
	if (max_grefs > 0)
832
		gnttab_free_grant_references(setup.gref_head);
833 834 835 836

	return 0;
}

837 838 839 840 841 842
/*
 * Generate a Xen blkfront IO request from a blk layer request.  Reads
 * and writes are handled as expected.
 *
 * @req: a request struct
 */
843
static int blkif_queue_request(struct request *req, struct blkfront_ring_info *rinfo)
844
{
845
	if (unlikely(rinfo->dev_info->connected != BLKIF_STATE_CONNECTED))
846 847
		return 1;

M
Mike Christie 已提交
848
	if (unlikely(req_op(req) == REQ_OP_DISCARD ||
849
		     req_op(req) == REQ_OP_SECURE_ERASE))
850
		return blkif_queue_discard_req(req, rinfo);
851
	else
852
		return blkif_queue_rw_req(req, rinfo);
853
}
854

855
static inline void flush_requests(struct blkfront_ring_info *rinfo)
856 857 858
{
	int notify;

859
	RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&rinfo->ring, notify);
860 861

	if (notify)
862
		notify_remote_via_irq(rinfo->irq);
863 864
}

865 866
static inline bool blkif_request_flush_invalid(struct request *req,
					       struct blkfront_info *info)
867 868
{
	return ((req->cmd_type != REQ_TYPE_FS) ||
869
		((req_op(req) == REQ_OP_FLUSH) &&
870
		 !info->feature_flush) ||
871
		((req->cmd_flags & REQ_FUA) &&
872
		 !info->feature_fua));
873 874
}

B
Bob Liu 已提交
875
static int blkif_queue_rq(struct blk_mq_hw_ctx *hctx,
876
			  const struct blk_mq_queue_data *qd)
877
{
B
Bob Liu 已提交
878
	unsigned long flags;
879 880 881
	int qid = hctx->queue_num;
	struct blkfront_info *info = hctx->queue->queuedata;
	struct blkfront_ring_info *rinfo = NULL;
882

883 884
	BUG_ON(info->nr_rings <= qid);
	rinfo = &info->rinfo[qid];
B
Bob Liu 已提交
885
	blk_mq_start_request(qd->rq);
B
Bob Liu 已提交
886
	spin_lock_irqsave(&rinfo->ring_lock, flags);
887
	if (RING_FULL(&rinfo->ring))
B
Bob Liu 已提交
888
		goto out_busy;
889

890
	if (blkif_request_flush_invalid(qd->rq, rinfo->dev_info))
B
Bob Liu 已提交
891
		goto out_err;
892

893
	if (blkif_queue_request(qd->rq, rinfo))
B
Bob Liu 已提交
894
		goto out_busy;
895

896
	flush_requests(rinfo);
B
Bob Liu 已提交
897
	spin_unlock_irqrestore(&rinfo->ring_lock, flags);
B
Bob Liu 已提交
898
	return BLK_MQ_RQ_QUEUE_OK;
899

B
Bob Liu 已提交
900
out_err:
B
Bob Liu 已提交
901
	spin_unlock_irqrestore(&rinfo->ring_lock, flags);
B
Bob Liu 已提交
902
	return BLK_MQ_RQ_QUEUE_ERROR;
903

B
Bob Liu 已提交
904
out_busy:
B
Bob Liu 已提交
905
	spin_unlock_irqrestore(&rinfo->ring_lock, flags);
B
Bob Liu 已提交
906 907
	blk_mq_stop_hw_queue(hctx);
	return BLK_MQ_RQ_QUEUE_BUSY;
908 909
}

B
Bob Liu 已提交
910 911 912 913
static struct blk_mq_ops blkfront_mq_ops = {
	.queue_rq = blkif_queue_rq,
};

914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
static void blkif_set_queue_limits(struct blkfront_info *info)
{
	struct request_queue *rq = info->rq;
	struct gendisk *gd = info->gd;
	unsigned int segments = info->max_indirect_segments ? :
				BLKIF_MAX_SEGMENTS_PER_REQUEST;

	queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq);

	if (info->feature_discard) {
		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, rq);
		blk_queue_max_discard_sectors(rq, get_capacity(gd));
		rq->limits.discard_granularity = info->discard_granularity;
		rq->limits.discard_alignment = info->discard_alignment;
		if (info->feature_secdiscard)
			queue_flag_set_unlocked(QUEUE_FLAG_SECERASE, rq);
	}

	/* Hard sector size and max sectors impersonate the equiv. hardware. */
	blk_queue_logical_block_size(rq, info->sector_size);
	blk_queue_physical_block_size(rq, info->physical_sector_size);
	blk_queue_max_hw_sectors(rq, (segments * XEN_PAGE_SIZE) / 512);

	/* Each segment in a request is up to an aligned page in size. */
	blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
	blk_queue_max_segment_size(rq, PAGE_SIZE);

	/* Ensure a merged request will fit in a single I/O ring slot. */
	blk_queue_max_segments(rq, segments / GRANTS_PER_PSEG);

	/* Make sure buffer addresses are sector-aligned. */
	blk_queue_dma_alignment(rq, 511);

	/* Make sure we don't use bounce buffers. */
	blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY);
}

951
static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size,
952
				unsigned int physical_sector_size)
953
{
954
	struct request_queue *rq;
955
	struct blkfront_info *info = gd->private_data;
956

B
Bob Liu 已提交
957 958
	memset(&info->tag_set, 0, sizeof(info->tag_set));
	info->tag_set.ops = &blkfront_mq_ops;
959
	info->tag_set.nr_hw_queues = info->nr_rings;
960 961 962 963 964 965 966 967 968 969
	if (HAS_EXTRA_REQ && info->max_indirect_segments == 0) {
		/*
		 * When indirect descriptior is not supported, the I/O request
		 * will be split between multiple request in the ring.
		 * To avoid problems when sending the request, divide by
		 * 2 the depth of the queue.
		 */
		info->tag_set.queue_depth =  BLK_RING_SIZE(info) / 2;
	} else
		info->tag_set.queue_depth = BLK_RING_SIZE(info);
B
Bob Liu 已提交
970 971 972 973 974 975
	info->tag_set.numa_node = NUMA_NO_NODE;
	info->tag_set.flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_SG_MERGE;
	info->tag_set.cmd_size = 0;
	info->tag_set.driver_data = info;

	if (blk_mq_alloc_tag_set(&info->tag_set))
976
		return -EINVAL;
B
Bob Liu 已提交
977 978 979
	rq = blk_mq_init_queue(&info->tag_set);
	if (IS_ERR(rq)) {
		blk_mq_free_tag_set(&info->tag_set);
980
		return PTR_ERR(rq);
B
Bob Liu 已提交
981
	}
982

983
	rq->queuedata = info;
984 985 986 987 988
	info->rq = gd->queue = rq;
	info->gd = gd;
	info->sector_size = sector_size;
	info->physical_sector_size = physical_sector_size;
	blkif_set_queue_limits(info);
989 990 991 992

	return 0;
}

993
static const char *flush_info(struct blkfront_info *info)
994
{
995
	if (info->feature_flush && info->feature_fua)
996
		return "barrier: enabled;";
997
	else if (info->feature_flush)
998
		return "flush diskcache: enabled;";
999
	else
1000 1001
		return "barrier or flush: disabled;";
}
1002

1003
static void xlvbd_flush(struct blkfront_info *info)
1004
{
1005 1006
	blk_queue_write_cache(info->rq, info->feature_flush ? true : false,
			      info->feature_fua ? true : false);
1007
	pr_info("blkfront: %s: %s %s %s %s %s\n",
1008
		info->gd->disk_name, flush_info(info),
1009 1010 1011
		"persistent grants:", info->feature_persistent ?
		"enabled;" : "disabled;", "indirect descriptors:",
		info->max_indirect_segments ? "enabled;" : "disabled;");
1012 1013
}

1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
{
	int major;
	major = BLKIF_MAJOR(vdevice);
	*minor = BLKIF_MINOR(vdevice);
	switch (major) {
		case XEN_IDE0_MAJOR:
			*offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET;
			*minor = ((*minor / 64) * PARTS_PER_DISK) +
				EMULATED_HD_DISK_MINOR_OFFSET;
			break;
		case XEN_IDE1_MAJOR:
			*offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET;
			*minor = (((*minor / 64) + 2) * PARTS_PER_DISK) +
				EMULATED_HD_DISK_MINOR_OFFSET;
			break;
		case XEN_SCSI_DISK0_MAJOR:
			*offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET;
			*minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET;
			break;
		case XEN_SCSI_DISK1_MAJOR:
		case XEN_SCSI_DISK2_MAJOR:
		case XEN_SCSI_DISK3_MAJOR:
		case XEN_SCSI_DISK4_MAJOR:
		case XEN_SCSI_DISK5_MAJOR:
		case XEN_SCSI_DISK6_MAJOR:
		case XEN_SCSI_DISK7_MAJOR:
			*offset = (*minor / PARTS_PER_DISK) + 
				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) +
				EMULATED_SD_DISK_NAME_OFFSET;
			*minor = *minor +
				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) +
				EMULATED_SD_DISK_MINOR_OFFSET;
			break;
		case XEN_SCSI_DISK8_MAJOR:
		case XEN_SCSI_DISK9_MAJOR:
		case XEN_SCSI_DISK10_MAJOR:
		case XEN_SCSI_DISK11_MAJOR:
		case XEN_SCSI_DISK12_MAJOR:
		case XEN_SCSI_DISK13_MAJOR:
		case XEN_SCSI_DISK14_MAJOR:
		case XEN_SCSI_DISK15_MAJOR:
			*offset = (*minor / PARTS_PER_DISK) + 
				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) +
				EMULATED_SD_DISK_NAME_OFFSET;
			*minor = *minor +
				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) +
				EMULATED_SD_DISK_MINOR_OFFSET;
			break;
		case XENVBD_MAJOR:
			*offset = *minor / PARTS_PER_DISK;
			break;
		default:
			printk(KERN_WARNING "blkfront: your disk configuration is "
					"incorrect, please use an xvd device instead\n");
			return -ENODEV;
	}
	return 0;
}
1073

1074 1075 1076 1077 1078 1079 1080 1081
static char *encode_disk_name(char *ptr, unsigned int n)
{
	if (n >= 26)
		ptr = encode_disk_name(ptr, n / 26 - 1);
	*ptr = 'a' + n % 26;
	return ptr + 1;
}

1082 1083
static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
			       struct blkfront_info *info,
1084 1085
			       u16 vdisk_info, u16 sector_size,
			       unsigned int physical_sector_size)
1086 1087 1088
{
	struct gendisk *gd;
	int nr_minors = 1;
1089
	int err;
1090 1091 1092
	unsigned int offset;
	int minor;
	int nr_parts;
1093
	char *ptr;
1094 1095 1096 1097

	BUG_ON(info->gd != NULL);
	BUG_ON(info->rq != NULL);

1098 1099 1100 1101 1102 1103 1104
	if ((info->vdevice>>EXT_SHIFT) > 1) {
		/* this is above the extended range; something is wrong */
		printk(KERN_WARNING "blkfront: vdevice 0x%x is above the extended range; ignoring\n", info->vdevice);
		return -ENODEV;
	}

	if (!VDEV_IS_EXTENDED(info->vdevice)) {
1105 1106 1107 1108
		err = xen_translate_vdev(info->vdevice, &minor, &offset);
		if (err)
			return err;		
 		nr_parts = PARTS_PER_DISK;
1109 1110 1111
	} else {
		minor = BLKIF_MINOR_EXT(info->vdevice);
		nr_parts = PARTS_PER_EXT_DISK;
1112
		offset = minor / nr_parts;
1113
		if (xen_hvm_domain() && offset < EMULATED_HD_DISK_NAME_OFFSET + 4)
1114 1115 1116
			printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with "
					"emulated IDE disks,\n\t choose an xvd device name"
					"from xvde on\n", info->vdevice);
1117
	}
1118 1119 1120 1121 1122
	if (minor >> MINORBITS) {
		pr_warn("blkfront: %#x's minor (%#x) out of range; ignoring\n",
			info->vdevice, minor);
		return -ENODEV;
	}
1123 1124 1125

	if ((minor % nr_parts) == 0)
		nr_minors = nr_parts;
1126

1127 1128 1129 1130 1131
	err = xlbd_reserve_minors(minor, nr_minors);
	if (err)
		goto out;
	err = -ENODEV;

1132 1133
	gd = alloc_disk(nr_minors);
	if (gd == NULL)
1134
		goto release;
1135

1136 1137 1138 1139 1140 1141 1142 1143
	strcpy(gd->disk_name, DEV_NAME);
	ptr = encode_disk_name(gd->disk_name + sizeof(DEV_NAME) - 1, offset);
	BUG_ON(ptr >= gd->disk_name + DISK_NAME_LEN);
	if (nr_minors > 1)
		*ptr = 0;
	else
		snprintf(ptr, gd->disk_name + DISK_NAME_LEN - ptr,
			 "%d", minor & (nr_parts - 1));
1144 1145 1146 1147 1148 1149 1150

	gd->major = XENVBD_MAJOR;
	gd->first_minor = minor;
	gd->fops = &xlvbd_block_fops;
	gd->private_data = info;
	set_capacity(gd, capacity);

1151
	if (xlvbd_init_blk_queue(gd, sector_size, physical_sector_size)) {
1152
		del_gendisk(gd);
1153
		goto release;
1154 1155
	}

1156
	xlvbd_flush(info);
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168

	if (vdisk_info & VDISK_READONLY)
		set_disk_ro(gd, 1);

	if (vdisk_info & VDISK_REMOVABLE)
		gd->flags |= GENHD_FL_REMOVABLE;

	if (vdisk_info & VDISK_CDROM)
		gd->flags |= GENHD_FL_CD;

	return 0;

1169 1170
 release:
	xlbd_release_minors(minor, nr_minors);
1171 1172 1173 1174
 out:
	return err;
}

D
Daniel Stodden 已提交
1175 1176
static void xlvbd_release_gendisk(struct blkfront_info *info)
{
1177
	unsigned int minor, nr_minors, i;
D
Daniel Stodden 已提交
1178 1179 1180 1181 1182

	if (info->rq == NULL)
		return;

	/* No more blkif_request(). */
B
Bob Liu 已提交
1183
	blk_mq_stop_hw_queues(info->rq);
D
Daniel Stodden 已提交
1184

1185 1186
	for (i = 0; i < info->nr_rings; i++) {
		struct blkfront_ring_info *rinfo = &info->rinfo[i];
D
Daniel Stodden 已提交
1187

1188 1189 1190 1191 1192 1193
		/* No more gnttab callback work. */
		gnttab_cancel_free_callback(&rinfo->callback);

		/* Flush gnttab callback work. Must be done with no locks held. */
		flush_work(&rinfo->work);
	}
D
Daniel Stodden 已提交
1194 1195 1196 1197 1198 1199 1200 1201

	del_gendisk(info->gd);

	minor = info->gd->first_minor;
	nr_minors = info->gd->minors;
	xlbd_release_minors(minor, nr_minors);

	blk_cleanup_queue(info->rq);
B
Bob Liu 已提交
1202
	blk_mq_free_tag_set(&info->tag_set);
D
Daniel Stodden 已提交
1203 1204 1205 1206 1207 1208
	info->rq = NULL;

	put_disk(info->gd);
	info->gd = NULL;
}

B
Bob Liu 已提交
1209 1210
/* Already hold rinfo->ring_lock. */
static inline void kick_pending_request_queues_locked(struct blkfront_ring_info *rinfo)
1211
{
1212 1213
	if (!RING_FULL(&rinfo->ring))
		blk_mq_start_stopped_hw_queues(rinfo->dev_info->rq, true);
1214 1215
}

B
Bob Liu 已提交
1216 1217 1218 1219 1220 1221 1222 1223 1224
static void kick_pending_request_queues(struct blkfront_ring_info *rinfo)
{
	unsigned long flags;

	spin_lock_irqsave(&rinfo->ring_lock, flags);
	kick_pending_request_queues_locked(rinfo);
	spin_unlock_irqrestore(&rinfo->ring_lock, flags);
}

1225 1226
static void blkif_restart_queue(struct work_struct *work)
{
1227
	struct blkfront_ring_info *rinfo = container_of(work, struct blkfront_ring_info, work);
1228

1229 1230
	if (rinfo->dev_info->connected == BLKIF_STATE_CONNECTED)
		kick_pending_request_queues(rinfo);
1231 1232
}

1233
static void blkif_free_ring(struct blkfront_ring_info *rinfo)
1234
{
1235
	struct grant *persistent_gnt, *n;
1236
	struct blkfront_info *info = rinfo->dev_info;
1237
	int i, j, segs;
1238

1239 1240 1241 1242
	/*
	 * Remove indirect pages, this only happens when using indirect
	 * descriptors but not persistent grants
	 */
1243
	if (!list_empty(&rinfo->indirect_pages)) {
1244 1245 1246
		struct page *indirect_page, *n;

		BUG_ON(info->feature_persistent);
1247
		list_for_each_entry_safe(indirect_page, n, &rinfo->indirect_pages, lru) {
1248 1249 1250 1251 1252
			list_del(&indirect_page->lru);
			__free_page(indirect_page);
		}
	}

1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
	/* Remove all persistent grants. */
	if (!list_empty(&rinfo->grants)) {
		list_for_each_entry_safe(persistent_gnt, n,
					 &rinfo->grants, node) {
			list_del(&persistent_gnt->node);
			if (persistent_gnt->gref != GRANT_INVALID_REF) {
				gnttab_end_foreign_access(persistent_gnt->gref,
							  0, 0UL);
				rinfo->persistent_gnts_c--;
			}
			if (info->feature_persistent)
				__free_page(persistent_gnt->page);
			kfree(persistent_gnt);
		}
	}
	BUG_ON(rinfo->persistent_gnts_c != 0);

B
Bob Liu 已提交
1270
	for (i = 0; i < BLK_RING_SIZE(info); i++) {
1271 1272 1273 1274
		/*
		 * Clear persistent grants present in requests already
		 * on the shared ring
		 */
1275
		if (!rinfo->shadow[i].request)
1276 1277
			goto free_shadow;

1278 1279 1280
		segs = rinfo->shadow[i].req.operation == BLKIF_OP_INDIRECT ?
		       rinfo->shadow[i].req.u.indirect.nr_segments :
		       rinfo->shadow[i].req.u.rw.nr_segments;
1281
		for (j = 0; j < segs; j++) {
1282
			persistent_gnt = rinfo->shadow[i].grants_used[j];
1283
			gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL);
1284
			if (info->feature_persistent)
1285
				__free_page(persistent_gnt->page);
1286 1287 1288
			kfree(persistent_gnt);
		}

1289
		if (rinfo->shadow[i].req.operation != BLKIF_OP_INDIRECT)
1290 1291 1292 1293 1294 1295 1296
			/*
			 * If this is not an indirect operation don't try to
			 * free indirect segments
			 */
			goto free_shadow;

		for (j = 0; j < INDIRECT_GREFS(segs); j++) {
1297
			persistent_gnt = rinfo->shadow[i].indirect_grants[j];
1298
			gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL);
1299
			__free_page(persistent_gnt->page);
1300 1301 1302 1303
			kfree(persistent_gnt);
		}

free_shadow:
1304 1305 1306 1307 1308 1309
		kfree(rinfo->shadow[i].grants_used);
		rinfo->shadow[i].grants_used = NULL;
		kfree(rinfo->shadow[i].indirect_grants);
		rinfo->shadow[i].indirect_grants = NULL;
		kfree(rinfo->shadow[i].sg);
		rinfo->shadow[i].sg = NULL;
1310 1311
	}

1312
	/* No more gnttab callback work. */
1313
	gnttab_cancel_free_callback(&rinfo->callback);
1314 1315

	/* Flush gnttab callback work. Must be done with no locks held. */
1316
	flush_work(&rinfo->work);
1317 1318

	/* Free resources associated with old device channel. */
B
Bob Liu 已提交
1319
	for (i = 0; i < info->nr_ring_pages; i++) {
1320 1321 1322
		if (rinfo->ring_ref[i] != GRANT_INVALID_REF) {
			gnttab_end_foreign_access(rinfo->ring_ref[i], 0, 0);
			rinfo->ring_ref[i] = GRANT_INVALID_REF;
B
Bob Liu 已提交
1323
		}
1324
	}
1325
	free_pages((unsigned long)rinfo->ring.sring, get_order(info->nr_ring_pages * XEN_PAGE_SIZE));
1326
	rinfo->ring.sring = NULL;
B
Bob Liu 已提交
1327

1328 1329 1330
	if (rinfo->irq)
		unbind_from_irqhandler(rinfo->irq, rinfo);
	rinfo->evtchn = rinfo->irq = 0;
1331
}
1332

1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349
static void blkif_free(struct blkfront_info *info, int suspend)
{
	unsigned int i;

	/* Prevent new requests being issued until we fix things up. */
	info->connected = suspend ?
		BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
	/* No more blkif_request(). */
	if (info->rq)
		blk_mq_stop_hw_queues(info->rq);

	for (i = 0; i < info->nr_rings; i++)
		blkif_free_ring(&info->rinfo[i]);

	kfree(info->rinfo);
	info->rinfo = NULL;
	info->nr_rings = 0;
1350 1351
}

1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377
struct copy_from_grant {
	const struct blk_shadow *s;
	unsigned int grant_idx;
	unsigned int bvec_offset;
	char *bvec_data;
};

static void blkif_copy_from_grant(unsigned long gfn, unsigned int offset,
				  unsigned int len, void *data)
{
	struct copy_from_grant *info = data;
	char *shared_data;
	/* Convenient aliases */
	const struct blk_shadow *s = info->s;

	shared_data = kmap_atomic(s->grants_used[info->grant_idx]->page);

	memcpy(info->bvec_data + info->bvec_offset,
	       shared_data + offset, len);

	info->bvec_offset += len;
	info->grant_idx++;

	kunmap_atomic(shared_data);
}

1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
static enum blk_req_status blkif_rsp_to_req_status(int rsp)
{
	switch (rsp)
	{
	case BLKIF_RSP_OKAY:
		return REQ_DONE;
	case BLKIF_RSP_EOPNOTSUPP:
		return REQ_EOPNOTSUPP;
	case BLKIF_RSP_ERROR:
		/* Fallthrough. */
	default:
		return REQ_ERROR;
	}
}

/*
 * Get the final status of the block request based on two ring response
 */
static int blkif_get_final_status(enum blk_req_status s1,
				  enum blk_req_status s2)
{
	BUG_ON(s1 == REQ_WAITING);
	BUG_ON(s2 == REQ_WAITING);

	if (s1 == REQ_ERROR || s2 == REQ_ERROR)
		return BLKIF_RSP_ERROR;
	else if (s1 == REQ_EOPNOTSUPP || s2 == REQ_EOPNOTSUPP)
		return BLKIF_RSP_EOPNOTSUPP;
	return BLKIF_RSP_OKAY;
}

static bool blkif_completion(unsigned long *id,
			     struct blkfront_ring_info *rinfo,
1411
			     struct blkif_response *bret)
1412
{
1413
	int i = 0;
1414
	struct scatterlist *sg;
1415
	int num_sg, num_grant;
1416
	struct blkfront_info *info = rinfo->dev_info;
1417
	struct blk_shadow *s = &rinfo->shadow[*id];
1418 1419 1420
	struct copy_from_grant data = {
		.grant_idx = 0,
	};
1421

1422
	num_grant = s->req.operation == BLKIF_OP_INDIRECT ?
1423
		s->req.u.indirect.nr_segments : s->req.u.rw.nr_segments;
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464

	/* The I/O request may be split in two. */
	if (unlikely(s->associated_id != NO_ASSOCIATED_ID)) {
		struct blk_shadow *s2 = &rinfo->shadow[s->associated_id];

		/* Keep the status of the current response in shadow. */
		s->status = blkif_rsp_to_req_status(bret->status);

		/* Wait the second response if not yet here. */
		if (s2->status == REQ_WAITING)
			return 0;

		bret->status = blkif_get_final_status(s->status,
						      s2->status);

		/*
		 * All the grants is stored in the first shadow in order
		 * to make the completion code simpler.
		 */
		num_grant += s2->req.u.rw.nr_segments;

		/*
		 * The two responses may not come in order. Only the
		 * first request will store the scatter-gather list.
		 */
		if (s2->num_sg != 0) {
			/* Update "id" with the ID of the first response. */
			*id = s->associated_id;
			s = s2;
		}

		/*
		 * We don't need anymore the second request, so recycling
		 * it now.
		 */
		if (add_id_to_freelist(rinfo, s->associated_id))
			WARN(1, "%s: can't recycle the second part (id = %ld) of the request\n",
			     info->gd->disk_name, s->associated_id);
	}

	data.s = s;
1465
	num_sg = s->num_sg;
1466

1467
	if (bret->operation == BLKIF_OP_READ && info->feature_persistent) {
1468
		for_each_sg(s->sg, sg, num_sg, i) {
1469
			BUG_ON(sg->offset + sg->length > PAGE_SIZE);
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480

			data.bvec_offset = sg->offset;
			data.bvec_data = kmap_atomic(sg_page(sg));

			gnttab_foreach_grant_in_range(sg_page(sg),
						      sg->offset,
						      sg->length,
						      blkif_copy_from_grant,
						      &data);

			kunmap_atomic(data.bvec_data);
1481 1482 1483
		}
	}
	/* Add the persistent grant into the list of free grants */
1484
	for (i = 0; i < num_grant; i++) {
1485 1486 1487 1488 1489 1490 1491
		if (gnttab_query_foreign_access(s->grants_used[i]->gref)) {
			/*
			 * If the grant is still mapped by the backend (the
			 * backend has chosen to make this grant persistent)
			 * we add it at the head of the list, so it will be
			 * reused first.
			 */
1492 1493 1494
			if (!info->feature_persistent)
				pr_alert_ratelimited("backed has not unmapped grant: %u\n",
						     s->grants_used[i]->gref);
1495 1496
			list_add(&s->grants_used[i]->node, &rinfo->grants);
			rinfo->persistent_gnts_c++;
1497 1498 1499 1500 1501 1502 1503 1504 1505
		} else {
			/*
			 * If the grant is not mapped by the backend we end the
			 * foreign access and add it to the tail of the list,
			 * so it will not be picked again unless we run out of
			 * persistent grants.
			 */
			gnttab_end_foreign_access(s->grants_used[i]->gref, 0, 0UL);
			s->grants_used[i]->gref = GRANT_INVALID_REF;
1506
			list_add_tail(&s->grants_used[i]->node, &rinfo->grants);
1507
		}
1508
	}
1509
	if (s->req.operation == BLKIF_OP_INDIRECT) {
1510
		for (i = 0; i < INDIRECT_GREFS(num_grant); i++) {
1511
			if (gnttab_query_foreign_access(s->indirect_grants[i]->gref)) {
1512 1513 1514
				if (!info->feature_persistent)
					pr_alert_ratelimited("backed has not unmapped grant: %u\n",
							     s->indirect_grants[i]->gref);
1515 1516
				list_add(&s->indirect_grants[i]->node, &rinfo->grants);
				rinfo->persistent_gnts_c++;
1517
			} else {
1518 1519
				struct page *indirect_page;

1520
				gnttab_end_foreign_access(s->indirect_grants[i]->gref, 0, 0UL);
1521 1522 1523 1524
				/*
				 * Add the used indirect page back to the list of
				 * available pages for indirect grefs.
				 */
1525
				if (!info->feature_persistent) {
1526
					indirect_page = s->indirect_grants[i]->page;
1527
					list_add(&indirect_page->lru, &rinfo->indirect_pages);
1528
				}
1529
				s->indirect_grants[i]->gref = GRANT_INVALID_REF;
1530
				list_add_tail(&s->indirect_grants[i]->node, &rinfo->grants);
1531
			}
1532 1533
		}
	}
1534 1535

	return 1;
1536 1537 1538 1539 1540 1541 1542 1543
}

static irqreturn_t blkif_interrupt(int irq, void *dev_id)
{
	struct request *req;
	struct blkif_response *bret;
	RING_IDX i, rp;
	unsigned long flags;
1544 1545
	struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)dev_id;
	struct blkfront_info *info = rinfo->dev_info;
1546
	int error;
1547

B
Bob Liu 已提交
1548
	if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
1549 1550
		return IRQ_HANDLED;

B
Bob Liu 已提交
1551
	spin_lock_irqsave(&rinfo->ring_lock, flags);
1552
 again:
1553
	rp = rinfo->ring.sring->rsp_prod;
1554 1555
	rmb(); /* Ensure we see queued responses up to 'rp'. */

1556
	for (i = rinfo->ring.rsp_cons; i != rp; i++) {
1557 1558
		unsigned long id;

1559
		bret = RING_GET_RESPONSE(&rinfo->ring, i);
1560
		id   = bret->id;
1561 1562 1563 1564 1565
		/*
		 * The backend has messed up and given us an id that we would
		 * never have given to it (we stamp it up to BLK_RING_SIZE -
		 * look in get_id_from_freelist.
		 */
B
Bob Liu 已提交
1566
		if (id >= BLK_RING_SIZE(info)) {
1567 1568 1569 1570 1571 1572
			WARN(1, "%s: response to %s has incorrect id (%ld)\n",
			     info->gd->disk_name, op_name(bret->operation), id);
			/* We can't safely get the 'struct request' as
			 * the id is busted. */
			continue;
		}
1573
		req  = rinfo->shadow[id].request;
1574

1575 1576 1577 1578 1579 1580 1581 1582
		if (bret->operation != BLKIF_OP_DISCARD) {
			/*
			 * We may need to wait for an extra response if the
			 * I/O request is split in 2
			 */
			if (!blkif_completion(&id, rinfo, bret))
				continue;
		}
1583

1584
		if (add_id_to_freelist(rinfo, id)) {
1585 1586 1587 1588
			WARN(1, "%s: response to %s (id %ld) couldn't be recycled!\n",
			     info->gd->disk_name, op_name(bret->operation), id);
			continue;
		}
1589

1590
		error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO;
1591
		switch (bret->operation) {
1592 1593 1594
		case BLKIF_OP_DISCARD:
			if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
				struct request_queue *rq = info->rq;
1595 1596
				printk(KERN_WARNING "blkfront: %s: %s op failed\n",
					   info->gd->disk_name, op_name(bret->operation));
1597
				error = -EOPNOTSUPP;
1598
				info->feature_discard = 0;
1599
				info->feature_secdiscard = 0;
1600
				queue_flag_clear(QUEUE_FLAG_DISCARD, rq);
1601
				queue_flag_clear(QUEUE_FLAG_SECERASE, rq);
1602
			}
1603
			blk_mq_complete_request(req, error);
1604
			break;
1605
		case BLKIF_OP_FLUSH_DISKCACHE:
1606 1607
		case BLKIF_OP_WRITE_BARRIER:
			if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
1608 1609
				printk(KERN_WARNING "blkfront: %s: %s op failed\n",
				       info->gd->disk_name, op_name(bret->operation));
1610
				error = -EOPNOTSUPP;
1611 1612
			}
			if (unlikely(bret->status == BLKIF_RSP_ERROR &&
1613
				     rinfo->shadow[id].req.u.rw.nr_segments == 0)) {
1614 1615
				printk(KERN_WARNING "blkfront: %s: empty %s op failed\n",
				       info->gd->disk_name, op_name(bret->operation));
1616
				error = -EOPNOTSUPP;
1617
			}
1618 1619 1620
			if (unlikely(error)) {
				if (error == -EOPNOTSUPP)
					error = 0;
1621
				info->feature_fua = 0;
1622 1623
				info->feature_flush = 0;
				xlvbd_flush(info);
1624 1625 1626 1627 1628 1629 1630 1631
			}
			/* fall through */
		case BLKIF_OP_READ:
		case BLKIF_OP_WRITE:
			if (unlikely(bret->status != BLKIF_RSP_OKAY))
				dev_dbg(&info->xbdev->dev, "Bad return from blkdev data "
					"request: %x\n", bret->status);

1632
			blk_mq_complete_request(req, error);
1633 1634 1635 1636 1637 1638
			break;
		default:
			BUG();
		}
	}

1639
	rinfo->ring.rsp_cons = i;
1640

1641
	if (i != rinfo->ring.req_prod_pvt) {
1642
		int more_to_do;
1643
		RING_FINAL_CHECK_FOR_RESPONSES(&rinfo->ring, more_to_do);
1644 1645 1646
		if (more_to_do)
			goto again;
	} else
1647
		rinfo->ring.sring->rsp_event = i + 1;
1648

B
Bob Liu 已提交
1649
	kick_pending_request_queues_locked(rinfo);
1650

B
Bob Liu 已提交
1651
	spin_unlock_irqrestore(&rinfo->ring_lock, flags);
1652 1653 1654 1655 1656 1657

	return IRQ_HANDLED;
}


static int setup_blkring(struct xenbus_device *dev,
1658
			 struct blkfront_ring_info *rinfo)
1659 1660
{
	struct blkif_sring *sring;
B
Bob Liu 已提交
1661
	int err, i;
1662
	struct blkfront_info *info = rinfo->dev_info;
1663
	unsigned long ring_size = info->nr_ring_pages * XEN_PAGE_SIZE;
1664
	grant_ref_t gref[XENBUS_MAX_RING_GRANTS];
1665

B
Bob Liu 已提交
1666
	for (i = 0; i < info->nr_ring_pages; i++)
1667
		rinfo->ring_ref[i] = GRANT_INVALID_REF;
1668

B
Bob Liu 已提交
1669 1670
	sring = (struct blkif_sring *)__get_free_pages(GFP_NOIO | __GFP_HIGH,
						       get_order(ring_size));
1671 1672 1673 1674 1675
	if (!sring) {
		xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
		return -ENOMEM;
	}
	SHARED_RING_INIT(sring);
1676
	FRONT_RING_INIT(&rinfo->ring, sring, ring_size);
1677

1678
	err = xenbus_grant_ring(dev, rinfo->ring.sring, info->nr_ring_pages, gref);
1679
	if (err < 0) {
B
Bob Liu 已提交
1680
		free_pages((unsigned long)sring, get_order(ring_size));
1681
		rinfo->ring.sring = NULL;
1682 1683
		goto fail;
	}
B
Bob Liu 已提交
1684
	for (i = 0; i < info->nr_ring_pages; i++)
1685
		rinfo->ring_ref[i] = gref[i];
1686

1687
	err = xenbus_alloc_evtchn(dev, &rinfo->evtchn);
1688 1689 1690
	if (err)
		goto fail;

1691 1692
	err = bind_evtchn_to_irqhandler(rinfo->evtchn, blkif_interrupt, 0,
					"blkif", rinfo);
1693 1694 1695 1696 1697
	if (err <= 0) {
		xenbus_dev_fatal(dev, err,
				 "bind_evtchn_to_irqhandler failed");
		goto fail;
	}
1698
	rinfo->irq = err;
1699 1700 1701 1702 1703 1704 1705

	return 0;
fail:
	blkif_free(info, 0);
	return err;
}

1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752
/*
 * Write out per-ring/queue nodes including ring-ref and event-channel, and each
 * ring buffer may have multi pages depending on ->nr_ring_pages.
 */
static int write_per_ring_nodes(struct xenbus_transaction xbt,
				struct blkfront_ring_info *rinfo, const char *dir)
{
	int err;
	unsigned int i;
	const char *message = NULL;
	struct blkfront_info *info = rinfo->dev_info;

	if (info->nr_ring_pages == 1) {
		err = xenbus_printf(xbt, dir, "ring-ref", "%u", rinfo->ring_ref[0]);
		if (err) {
			message = "writing ring-ref";
			goto abort_transaction;
		}
	} else {
		for (i = 0; i < info->nr_ring_pages; i++) {
			char ring_ref_name[RINGREF_NAME_LEN];

			snprintf(ring_ref_name, RINGREF_NAME_LEN, "ring-ref%u", i);
			err = xenbus_printf(xbt, dir, ring_ref_name,
					    "%u", rinfo->ring_ref[i]);
			if (err) {
				message = "writing ring-ref";
				goto abort_transaction;
			}
		}
	}

	err = xenbus_printf(xbt, dir, "event-channel", "%u", rinfo->evtchn);
	if (err) {
		message = "writing event-channel";
		goto abort_transaction;
	}

	return 0;

abort_transaction:
	xenbus_transaction_end(xbt, 1);
	if (message)
		xenbus_dev_fatal(info->xbdev, err, "%s", message);

	return err;
}
1753 1754

/* Common code used when first setting up, and when resuming. */
1755
static int talk_to_blkback(struct xenbus_device *dev,
1756 1757 1758 1759
			   struct blkfront_info *info)
{
	const char *message = NULL;
	struct xenbus_transaction xbt;
1760 1761
	int err;
	unsigned int i, max_page_order = 0;
B
Bob Liu 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771
	unsigned int ring_page_order = 0;

	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "max-ring-page-order", "%u", &max_page_order);
	if (err != 1)
		info->nr_ring_pages = 1;
	else {
		ring_page_order = min(xen_blkif_max_ring_order, max_page_order);
		info->nr_ring_pages = 1 << ring_page_order;
	}
1772

1773
	for (i = 0; i < info->nr_rings; i++) {
1774 1775
		struct blkfront_ring_info *rinfo = &info->rinfo[i];

1776 1777 1778 1779 1780
		/* Create shared ring, alloc event channel. */
		err = setup_blkring(dev, rinfo);
		if (err)
			goto destroy_blkring;
	}
1781 1782 1783 1784 1785 1786 1787 1788

again:
	err = xenbus_transaction_start(&xbt);
	if (err) {
		xenbus_dev_fatal(dev, err, "starting transaction");
		goto destroy_blkring;
	}

1789 1790 1791 1792 1793 1794 1795 1796
	if (info->nr_ring_pages > 1) {
		err = xenbus_printf(xbt, dev->nodename, "ring-page-order", "%u",
				    ring_page_order);
		if (err) {
			message = "writing ring-page-order";
			goto abort_transaction;
		}
	}
1797

1798 1799 1800 1801 1802 1803 1804 1805
	/* We already got the number of queues/rings in _probe */
	if (info->nr_rings == 1) {
		err = write_per_ring_nodes(xbt, &info->rinfo[0], dev->nodename);
		if (err)
			goto destroy_blkring;
	} else {
		char *path;
		size_t pathsize;
1806

1807 1808
		err = xenbus_printf(xbt, dev->nodename, "multi-queue-num-queues", "%u",
				    info->nr_rings);
1809
		if (err) {
1810
			message = "writing multi-queue-num-queues";
1811 1812
			goto abort_transaction;
		}
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831

		pathsize = strlen(dev->nodename) + QUEUE_NAME_LEN;
		path = kmalloc(pathsize, GFP_KERNEL);
		if (!path) {
			err = -ENOMEM;
			message = "ENOMEM while writing ring references";
			goto abort_transaction;
		}

		for (i = 0; i < info->nr_rings; i++) {
			memset(path, 0, pathsize);
			snprintf(path, pathsize, "%s/queue-%u", dev->nodename, i);
			err = write_per_ring_nodes(xbt, &info->rinfo[i], path);
			if (err) {
				kfree(path);
				goto destroy_blkring;
			}
		}
		kfree(path);
1832
	}
1833 1834 1835 1836 1837 1838
	err = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
			    XEN_IO_PROTO_ABI_NATIVE);
	if (err) {
		message = "writing protocol";
		goto abort_transaction;
	}
1839
	err = xenbus_printf(xbt, dev->nodename,
1840
			    "feature-persistent", "%u", 1);
1841 1842 1843
	if (err)
		dev_warn(&dev->dev,
			 "writing persistent grants feature to xenbus");
1844 1845 1846 1847 1848 1849 1850 1851 1852

	err = xenbus_transaction_end(xbt, 0);
	if (err) {
		if (err == -EAGAIN)
			goto again;
		xenbus_dev_fatal(dev, err, "completing transaction");
		goto destroy_blkring;
	}

1853 1854
	for (i = 0; i < info->nr_rings; i++) {
		unsigned int j;
1855
		struct blkfront_ring_info *rinfo = &info->rinfo[i];
1856 1857 1858 1859 1860

		for (j = 0; j < BLK_RING_SIZE(info); j++)
			rinfo->shadow[j].req.u.rw.id = j + 1;
		rinfo->shadow[BLK_RING_SIZE(info)-1].req.u.rw.id = 0x0fffffff;
	}
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
	xenbus_switch_state(dev, XenbusStateInitialised);

	return 0;

 abort_transaction:
	xenbus_transaction_end(xbt, 1);
	if (message)
		xenbus_dev_fatal(dev, err, "%s", message);
 destroy_blkring:
	blkif_free(info, 0);
1871

1872 1873 1874
	kfree(info);
	dev_set_drvdata(&dev->dev, NULL);

1875 1876 1877
	return err;
}

1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
static int negotiate_mq(struct blkfront_info *info)
{
	unsigned int backend_max_queues = 0;
	int err;
	unsigned int i;

	BUG_ON(info->nr_rings);

	/* Check if backend supports multiple queues. */
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "multi-queue-max-queues", "%u", &backend_max_queues);
	if (err < 0)
		backend_max_queues = 1;

	info->nr_rings = min(backend_max_queues, xen_blkif_max_queues);
	/* We need at least one ring. */
	if (!info->nr_rings)
		info->nr_rings = 1;

	info->rinfo = kzalloc(sizeof(struct blkfront_ring_info) * info->nr_rings, GFP_KERNEL);
	if (!info->rinfo) {
		xenbus_dev_fatal(info->xbdev, -ENOMEM, "allocating ring_info structure");
		return -ENOMEM;
	}

	for (i = 0; i < info->nr_rings; i++) {
		struct blkfront_ring_info *rinfo;

		rinfo = &info->rinfo[i];
		INIT_LIST_HEAD(&rinfo->indirect_pages);
		INIT_LIST_HEAD(&rinfo->grants);
		rinfo->dev_info = info;
		INIT_WORK(&rinfo->work, blkif_restart_queue);
		spin_lock_init(&rinfo->ring_lock);
	}
	return 0;
}
1915 1916 1917 1918 1919 1920 1921 1922 1923
/**
 * Entry point to this code when a new device is created.  Allocate the basic
 * structures and the ring buffer for communication with the backend, and
 * inform the backend of the appropriate details for those.  Switch to
 * Initialised state.
 */
static int blkfront_probe(struct xenbus_device *dev,
			  const struct xenbus_device_id *id)
{
B
Bob Liu 已提交
1924
	int err, vdevice;
1925 1926 1927 1928 1929 1930
	struct blkfront_info *info;

	/* FIXME: Use dynamic device id if this is not set. */
	err = xenbus_scanf(XBT_NIL, dev->nodename,
			   "virtual-device", "%i", &vdevice);
	if (err != 1) {
1931 1932 1933 1934 1935 1936 1937
		/* go looking in the extended area instead */
		err = xenbus_scanf(XBT_NIL, dev->nodename, "virtual-device-ext",
				   "%i", &vdevice);
		if (err != 1) {
			xenbus_dev_fatal(dev, err, "reading virtual-device");
			return err;
		}
1938 1939
	}

1940 1941 1942 1943
	if (xen_hvm_domain()) {
		char *type;
		int len;
		/* no unplug has been done: do not hook devices != xen vbds */
1944
		if (xen_has_pv_and_legacy_disk_devices()) {
1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
			int major;

			if (!VDEV_IS_EXTENDED(vdevice))
				major = BLKIF_MAJOR(vdevice);
			else
				major = XENVBD_MAJOR;

			if (major != XENVBD_MAJOR) {
				printk(KERN_INFO
						"%s: HVM does not support vbd %d as xen block device\n",
1955
						__func__, vdevice);
1956 1957 1958 1959 1960 1961 1962 1963 1964
				return -ENODEV;
			}
		}
		/* do not create a PV cdrom device if we are an HVM guest */
		type = xenbus_read(XBT_NIL, dev->nodename, "device-type", &len);
		if (IS_ERR(type))
			return -ENODEV;
		if (strncmp(type, "cdrom", 5) == 0) {
			kfree(type);
1965 1966
			return -ENODEV;
		}
1967
		kfree(type);
1968
	}
1969 1970 1971 1972 1973 1974
	info = kzalloc(sizeof(*info), GFP_KERNEL);
	if (!info) {
		xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
		return -ENOMEM;
	}

1975
	info->xbdev = dev;
1976 1977
	err = negotiate_mq(info);
	if (err) {
1978
		kfree(info);
1979
		return err;
1980
	}
1981

1982
	mutex_init(&info->mutex);
1983 1984 1985 1986 1987
	info->vdevice = vdevice;
	info->connected = BLKIF_STATE_DISCONNECTED;

	/* Front end dir is a number, which is used as the id. */
	info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
1988
	dev_set_drvdata(&dev->dev, info);
1989 1990 1991 1992

	return 0;
}

1993
static void split_bio_end(struct bio *bio)
1994 1995 1996 1997 1998
{
	struct split_bio *split_bio = bio->bi_private;

	if (atomic_dec_and_test(&split_bio->pending)) {
		split_bio->bio->bi_phys_segments = 0;
1999 2000
		split_bio->bio->bi_error = bio->bi_error;
		bio_endio(split_bio->bio);
2001 2002 2003 2004
		kfree(split_bio);
	}
	bio_put(bio);
}
2005 2006 2007

static int blkif_recover(struct blkfront_info *info)
{
2008
	unsigned int i, r_index;
2009 2010 2011 2012 2013 2014 2015
	struct request *req, *n;
	int rc;
	struct bio *bio, *cloned_bio;
	unsigned int segs, offset;
	int pending, size;
	struct split_bio *split_bio;

2016
	blkfront_gather_backend_features(info);
2017 2018
	/* Reset limits changed by blk_mq_update_nr_hw_queues(). */
	blkif_set_queue_limits(info);
2019
	segs = info->max_indirect_segments ? : BLKIF_MAX_SEGMENTS_PER_REQUEST;
2020
	blk_queue_max_segments(info->rq, segs / GRANTS_PER_PSEG);
2021

2022
	for (r_index = 0; r_index < info->nr_rings; r_index++) {
2023
		struct blkfront_ring_info *rinfo = &info->rinfo[r_index];
2024 2025

		rc = blkfront_setup_indirect(rinfo);
2026
		if (rc)
2027 2028
			return rc;
	}
2029 2030 2031 2032 2033
	xenbus_switch_state(info->xbdev, XenbusStateConnected);

	/* Now safe for us to use the shared ring */
	info->connected = BLKIF_STATE_CONNECTED;

2034 2035 2036 2037 2038 2039 2040
	for (r_index = 0; r_index < info->nr_rings; r_index++) {
		struct blkfront_ring_info *rinfo;

		rinfo = &info->rinfo[r_index];
		/* Kick any other new requests queued since we resumed */
		kick_pending_request_queues(rinfo);
	}
2041

2042
	list_for_each_entry_safe(req, n, &info->requests, queuelist) {
2043 2044 2045
		/* Requeue pending requests (flush or discard) */
		list_del_init(&req->queuelist);
		BUG_ON(req->nr_phys_segments > segs);
B
Bob Liu 已提交
2046
		blk_mq_requeue_request(req);
2047
	}
B
Bob Liu 已提交
2048
	blk_mq_kick_requeue_list(info->rq);
2049

2050
	while ((bio = bio_list_pop(&info->bio_list)) != NULL) {
2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
		/* Traverse the list of pending bios and re-queue them */
		if (bio_segments(bio) > segs) {
			/*
			 * This bio has more segments than what we can
			 * handle, we have to split it.
			 */
			pending = (bio_segments(bio) + segs - 1) / segs;
			split_bio = kzalloc(sizeof(*split_bio), GFP_NOIO);
			BUG_ON(split_bio == NULL);
			atomic_set(&split_bio->pending, pending);
			split_bio->bio = bio;
			for (i = 0; i < pending; i++) {
2063 2064
				offset = (i * segs * XEN_PAGE_SIZE) >> 9;
				size = min((unsigned int)(segs * XEN_PAGE_SIZE) >> 9,
2065
					   (unsigned int)bio_sectors(bio) - offset);
2066 2067
				cloned_bio = bio_clone(bio, GFP_NOIO);
				BUG_ON(cloned_bio == NULL);
2068
				bio_trim(cloned_bio, offset, size);
2069 2070
				cloned_bio->bi_private = split_bio;
				cloned_bio->bi_end_io = split_bio_end;
2071
				submit_bio(cloned_bio);
2072 2073 2074 2075 2076 2077 2078 2079
			}
			/*
			 * Now we have to wait for all those smaller bios to
			 * end, so we can also end the "parent" bio.
			 */
			continue;
		}
		/* We don't need to split this bio */
2080
		submit_bio(bio);
2081 2082
	}

2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093
	return 0;
}

/**
 * We are reconnecting to the backend, due to a suspend/resume, or a backend
 * driver restart.  We tear down our blkif structure and recreate it, but
 * leave the device-layer structures intact so that this is transparent to the
 * rest of the kernel.
 */
static int blkfront_resume(struct xenbus_device *dev)
{
2094
	struct blkfront_info *info = dev_get_drvdata(&dev->dev);
2095
	int err = 0;
2096
	unsigned int i, j;
2097 2098 2099

	dev_dbg(&dev->dev, "blkfront_resume: %s\n", dev->nodename);

2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114
	bio_list_init(&info->bio_list);
	INIT_LIST_HEAD(&info->requests);
	for (i = 0; i < info->nr_rings; i++) {
		struct blkfront_ring_info *rinfo = &info->rinfo[i];
		struct bio_list merge_bio;
		struct blk_shadow *shadow = rinfo->shadow;

		for (j = 0; j < BLK_RING_SIZE(info); j++) {
			/* Not in use? */
			if (!shadow[j].request)
				continue;

			/*
			 * Get the bios in the request so we can re-queue them.
			 */
2115 2116
			if (req_op(shadow[i].request) == REQ_OP_FLUSH ||
			    req_op(shadow[i].request) == REQ_OP_DISCARD ||
2117 2118
			    req_op(shadow[i].request) == REQ_OP_SECURE_ERASE ||
			    shadow[j].request->cmd_flags & REQ_FUA) {
2119 2120 2121
				/*
				 * Flush operations don't contain bios, so
				 * we need to requeue the whole request
2122 2123 2124
				 *
				 * XXX: but this doesn't make any sense for a
				 * write with the FUA flag set..
2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136
				 */
				list_add(&shadow[j].request->queuelist, &info->requests);
				continue;
			}
			merge_bio.head = shadow[j].request->bio;
			merge_bio.tail = shadow[j].request->biotail;
			bio_list_merge(&info->bio_list, &merge_bio);
			shadow[j].request->bio = NULL;
			blk_mq_end_request(shadow[j].request, 0);
		}
	}

2137 2138
	blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);

2139 2140 2141 2142
	err = negotiate_mq(info);
	if (err)
		return err;

2143
	err = talk_to_blkback(dev, info);
2144 2145
	if (!err)
		blk_mq_update_nr_hw_queues(&info->tag_set, info->nr_rings);
2146 2147 2148 2149 2150 2151

	/*
	 * We have to wait for the backend to switch to
	 * connected state, since we want to read which
	 * features it supports.
	 */
2152 2153 2154 2155

	return err;
}

2156
static void blkfront_closing(struct blkfront_info *info)
2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179
{
	struct xenbus_device *xbdev = info->xbdev;
	struct block_device *bdev = NULL;

	mutex_lock(&info->mutex);

	if (xbdev->state == XenbusStateClosing) {
		mutex_unlock(&info->mutex);
		return;
	}

	if (info->gd)
		bdev = bdget_disk(info->gd, 0);

	mutex_unlock(&info->mutex);

	if (!bdev) {
		xenbus_frontend_closed(xbdev);
		return;
	}

	mutex_lock(&bdev->bd_mutex);

2180
	if (bdev->bd_openers) {
2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
		xenbus_dev_error(xbdev, -EBUSY,
				 "Device in use; refusing to close");
		xenbus_switch_state(xbdev, XenbusStateClosing);
	} else {
		xlvbd_release_gendisk(info);
		xenbus_frontend_closed(xbdev);
	}

	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
}
2192

2193 2194 2195 2196 2197
static void blkfront_setup_discard(struct blkfront_info *info)
{
	int err;
	unsigned int discard_granularity;
	unsigned int discard_alignment;
2198
	unsigned int discard_secure;
2199

2200 2201 2202 2203 2204 2205 2206 2207 2208
	info->feature_discard = 1;
	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
		"discard-granularity", "%u", &discard_granularity,
		"discard-alignment", "%u", &discard_alignment,
		NULL);
	if (!err) {
		info->discard_granularity = discard_granularity;
		info->discard_alignment = discard_alignment;
	}
2209 2210 2211
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "discard-secure", "%u", &discard_secure);
	if (err > 0)
2212
		info->feature_secdiscard = !!discard_secure;
2213 2214
}

2215
static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo)
2216
{
2217
	unsigned int psegs, grants;
2218
	int err, i;
2219
	struct blkfront_info *info = rinfo->dev_info;
2220

2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
	if (info->max_indirect_segments == 0) {
		if (!HAS_EXTRA_REQ)
			grants = BLKIF_MAX_SEGMENTS_PER_REQUEST;
		else {
			/*
			 * When an extra req is required, the maximum
			 * grants supported is related to the size of the
			 * Linux block segment.
			 */
			grants = GRANTS_PER_PSEG;
		}
	}
2233
	else
2234 2235
		grants = info->max_indirect_segments;
	psegs = grants / GRANTS_PER_PSEG;
2236

2237
	err = fill_grant_buffer(rinfo,
2238
				(grants + INDIRECT_GREFS(grants)) * BLK_RING_SIZE(info));
2239 2240 2241
	if (err)
		goto out_of_memory;

2242 2243 2244 2245 2246 2247
	if (!info->feature_persistent && info->max_indirect_segments) {
		/*
		 * We are using indirect descriptors but not persistent
		 * grants, we need to allocate a set of pages that can be
		 * used for mapping indirect grefs
		 */
2248
		int num = INDIRECT_GREFS(grants) * BLK_RING_SIZE(info);
2249

2250
		BUG_ON(!list_empty(&rinfo->indirect_pages));
2251 2252 2253 2254
		for (i = 0; i < num; i++) {
			struct page *indirect_page = alloc_page(GFP_NOIO);
			if (!indirect_page)
				goto out_of_memory;
2255
			list_add(&indirect_page->lru, &rinfo->indirect_pages);
2256 2257 2258
		}
	}

B
Bob Liu 已提交
2259
	for (i = 0; i < BLK_RING_SIZE(info); i++) {
2260 2261
		rinfo->shadow[i].grants_used = kzalloc(
			sizeof(rinfo->shadow[i].grants_used[0]) * grants,
2262
			GFP_NOIO);
2263
		rinfo->shadow[i].sg = kzalloc(sizeof(rinfo->shadow[i].sg[0]) * psegs, GFP_NOIO);
2264
		if (info->max_indirect_segments)
2265 2266
			rinfo->shadow[i].indirect_grants = kzalloc(
				sizeof(rinfo->shadow[i].indirect_grants[0]) *
2267
				INDIRECT_GREFS(grants),
2268
				GFP_NOIO);
2269 2270
		if ((rinfo->shadow[i].grants_used == NULL) ||
			(rinfo->shadow[i].sg == NULL) ||
2271
		     (info->max_indirect_segments &&
2272
		     (rinfo->shadow[i].indirect_grants == NULL)))
2273
			goto out_of_memory;
2274
		sg_init_table(rinfo->shadow[i].sg, psegs);
2275 2276 2277 2278 2279 2280
	}


	return 0;

out_of_memory:
B
Bob Liu 已提交
2281
	for (i = 0; i < BLK_RING_SIZE(info); i++) {
2282 2283 2284 2285 2286 2287
		kfree(rinfo->shadow[i].grants_used);
		rinfo->shadow[i].grants_used = NULL;
		kfree(rinfo->shadow[i].sg);
		rinfo->shadow[i].sg = NULL;
		kfree(rinfo->shadow[i].indirect_grants);
		rinfo->shadow[i].indirect_grants = NULL;
2288
	}
2289
	if (!list_empty(&rinfo->indirect_pages)) {
2290
		struct page *indirect_page, *n;
2291
		list_for_each_entry_safe(indirect_page, n, &rinfo->indirect_pages, lru) {
2292 2293 2294 2295
			list_del(&indirect_page->lru);
			__free_page(indirect_page);
		}
	}
2296 2297 2298
	return -ENOMEM;
}

2299 2300 2301
/*
 * Gather all backend feature-*
 */
2302
static void blkfront_gather_backend_features(struct blkfront_info *info)
2303 2304 2305 2306 2307 2308
{
	int err;
	int barrier, flush, discard, persistent;
	unsigned int indirect_segments;

	info->feature_flush = 0;
2309
	info->feature_fua = 0;
2310

2311 2312
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "feature-barrier", "%d", &barrier);
2313 2314 2315 2316 2317 2318 2319 2320

	/*
	 * If there's no "feature-barrier" defined, then it means
	 * we're dealing with a very old backend which writes
	 * synchronously; nothing to do.
	 *
	 * If there are barriers, then we use flush.
	 */
2321
	if (err > 0 && barrier) {
2322 2323 2324 2325
		info->feature_flush = 1;
		info->feature_fua = 1;
	}

2326 2327 2328 2329
	/*
	 * And if there is "feature-flush-cache" use that above
	 * barriers.
	 */
2330 2331
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "feature-flush-cache", "%d", &flush);
2332

2333
	if (err > 0 && flush) {
2334 2335 2336
		info->feature_flush = 1;
		info->feature_fua = 0;
	}
2337

2338 2339
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "feature-discard", "%d", &discard);
2340

2341
	if (err > 0 && discard)
2342 2343
		blkfront_setup_discard(info);

2344 2345 2346
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "feature-persistent", "%d", &persistent);
	if (err <= 0)
2347 2348 2349 2350
		info->feature_persistent = 0;
	else
		info->feature_persistent = persistent;

2351 2352 2353 2354
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "feature-max-indirect-segments", "%u",
			   &indirect_segments);
	if (err <= 0)
2355 2356 2357 2358 2359 2360
		info->max_indirect_segments = 0;
	else
		info->max_indirect_segments = min(indirect_segments,
						  xen_blkif_max_segments);
}

2361 2362 2363 2364 2365 2366 2367 2368
/*
 * Invoked when the backend is finally 'ready' (and has told produced
 * the details about the physical device - #sectors, size, etc).
 */
static void blkfront_connect(struct blkfront_info *info)
{
	unsigned long long sectors;
	unsigned long sector_size;
2369
	unsigned int physical_sector_size;
2370
	unsigned int binfo;
2371
	int err, i;
2372

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
	switch (info->connected) {
	case BLKIF_STATE_CONNECTED:
		/*
		 * Potentially, the back-end may be signalling
		 * a capacity change; update the capacity.
		 */
		err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
				   "sectors", "%Lu", &sectors);
		if (XENBUS_EXIST_ERR(err))
			return;
		printk(KERN_INFO "Setting capacity to %Lu\n",
		       sectors);
		set_capacity(info->gd, sectors);
2386
		revalidate_disk(info->gd);
2387

2388
		return;
2389
	case BLKIF_STATE_SUSPENDED:
2390 2391 2392 2393 2394 2395 2396
		/*
		 * If we are recovering from suspension, we need to wait
		 * for the backend to announce it's features before
		 * reconnecting, at least we need to know if the backend
		 * supports indirect descriptors, and how many.
		 */
		blkif_recover(info);
2397 2398
		return;

2399 2400
	default:
		break;
2401
	}
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417

	dev_dbg(&info->xbdev->dev, "%s:%s.\n",
		__func__, info->xbdev->otherend);

	err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
			    "sectors", "%llu", &sectors,
			    "info", "%u", &binfo,
			    "sector-size", "%lu", &sector_size,
			    NULL);
	if (err) {
		xenbus_dev_fatal(info->xbdev, err,
				 "reading backend fields at %s",
				 info->xbdev->otherend);
		return;
	}

2418 2419 2420 2421 2422 2423 2424 2425 2426 2427
	/*
	 * physcial-sector-size is a newer field, so old backends may not
	 * provide this. Assume physical sector size to be the same as
	 * sector_size in that case.
	 */
	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
			   "physical-sector-size", "%u", &physical_sector_size);
	if (err != 1)
		physical_sector_size = sector_size;

2428 2429 2430 2431 2432 2433 2434 2435 2436
	blkfront_gather_backend_features(info);
	for (i = 0; i < info->nr_rings; i++) {
		err = blkfront_setup_indirect(&info->rinfo[i]);
		if (err) {
			xenbus_dev_fatal(info->xbdev, err, "setup_indirect at %s",
					 info->xbdev->otherend);
			blkif_free(info, 0);
			break;
		}
2437 2438
	}

2439 2440
	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size,
				  physical_sector_size);
2441 2442 2443
	if (err) {
		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
				 info->xbdev->otherend);
2444
		goto fail;
2445 2446 2447 2448 2449 2450
	}

	xenbus_switch_state(info->xbdev, XenbusStateConnected);

	/* Kick pending requests. */
	info->connected = BLKIF_STATE_CONNECTED;
2451 2452
	for (i = 0; i < info->nr_rings; i++)
		kick_pending_request_queues(&info->rinfo[i]);
2453

2454
	device_add_disk(&info->xbdev->dev, info->gd);
2455 2456

	info->is_ready = 1;
2457 2458 2459 2460 2461
	return;

fail:
	blkif_free(info, 0);
	return;
2462 2463 2464 2465 2466
}

/**
 * Callback received when the backend's state changes.
 */
2467
static void blkback_changed(struct xenbus_device *dev,
2468 2469
			    enum xenbus_state backend_state)
{
2470
	struct blkfront_info *info = dev_get_drvdata(&dev->dev);
2471

2472
	dev_dbg(&dev->dev, "blkfront:blkback_changed to state %d.\n", backend_state);
2473 2474 2475

	switch (backend_state) {
	case XenbusStateInitWait:
2476 2477
		if (dev->state != XenbusStateInitialising)
			break;
2478
		if (talk_to_blkback(dev, info))
2479 2480
			break;
	case XenbusStateInitialising:
2481
	case XenbusStateInitialised:
2482 2483
	case XenbusStateReconfiguring:
	case XenbusStateReconfigured:
2484 2485 2486 2487
	case XenbusStateUnknown:
		break;

	case XenbusStateConnected:
2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
		/*
		 * talk_to_blkback sets state to XenbusStateInitialised
		 * and blkfront_connect sets it to XenbusStateConnected
		 * (if connection went OK).
		 *
		 * If the backend (or toolstack) decides to poke at backend
		 * state (and re-trigger the watch by setting the state repeatedly
		 * to XenbusStateConnected (4)) we need to deal with this.
		 * This is allowed as this is used to communicate to the guest
		 * that the size of disk has changed!
		 */
		if ((dev->state != XenbusStateInitialised) &&
		    (dev->state != XenbusStateConnected)) {
2501 2502 2503
			if (talk_to_blkback(dev, info))
				break;
		}
2504

2505 2506 2507
		blkfront_connect(info);
		break;

2508 2509 2510 2511
	case XenbusStateClosed:
		if (dev->state == XenbusStateClosed)
			break;
		/* Missed the backend's Closing state -- fallthrough */
2512
	case XenbusStateClosing:
2513 2514
		if (info)
			blkfront_closing(info);
2515 2516 2517 2518
		break;
	}
}

2519
static int blkfront_remove(struct xenbus_device *xbdev)
2520
{
2521 2522 2523
	struct blkfront_info *info = dev_get_drvdata(&xbdev->dev);
	struct block_device *bdev = NULL;
	struct gendisk *disk;
2524

2525
	dev_dbg(&xbdev->dev, "%s removed", xbdev->nodename);
2526 2527 2528

	blkif_free(info, 0);

2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551
	mutex_lock(&info->mutex);

	disk = info->gd;
	if (disk)
		bdev = bdget_disk(disk, 0);

	info->xbdev = NULL;
	mutex_unlock(&info->mutex);

	if (!bdev) {
		kfree(info);
		return 0;
	}

	/*
	 * The xbdev was removed before we reached the Closed
	 * state. See if it's safe to remove the disk. If the bdev
	 * isn't closed yet, we let release take care of it.
	 */

	mutex_lock(&bdev->bd_mutex);
	info = disk->private_data;

2552 2553 2554 2555
	dev_warn(disk_to_dev(disk),
		 "%s was hot-unplugged, %d stale handles\n",
		 xbdev->nodename, bdev->bd_openers);

2556
	if (info && !bdev->bd_openers) {
2557 2558
		xlvbd_release_gendisk(info);
		disk->private_data = NULL;
2559
		kfree(info);
2560 2561 2562 2563
	}

	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
2564 2565 2566 2567

	return 0;
}

2568 2569
static int blkfront_is_ready(struct xenbus_device *dev)
{
2570
	struct blkfront_info *info = dev_get_drvdata(&dev->dev);
2571

2572
	return info->is_ready && info->xbdev;
2573 2574
}

A
Al Viro 已提交
2575
static int blkif_open(struct block_device *bdev, fmode_t mode)
2576
{
2577 2578 2579
	struct gendisk *disk = bdev->bd_disk;
	struct blkfront_info *info;
	int err = 0;
2580

2581
	mutex_lock(&blkfront_mutex);
2582

2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598
	info = disk->private_data;
	if (!info) {
		/* xbdev gone */
		err = -ERESTARTSYS;
		goto out;
	}

	mutex_lock(&info->mutex);

	if (!info->gd)
		/* xbdev is closed */
		err = -ERESTARTSYS;

	mutex_unlock(&info->mutex);

out:
2599
	mutex_unlock(&blkfront_mutex);
2600
	return err;
2601 2602
}

2603
static void blkif_release(struct gendisk *disk, fmode_t mode)
2604
{
A
Al Viro 已提交
2605
	struct blkfront_info *info = disk->private_data;
2606 2607 2608
	struct block_device *bdev;
	struct xenbus_device *xbdev;

2609
	mutex_lock(&blkfront_mutex);
2610 2611 2612

	bdev = bdget_disk(disk, 0);

2613 2614 2615 2616
	if (!bdev) {
		WARN(1, "Block device %s yanked out from us!\n", disk->disk_name);
		goto out_mutex;
	}
2617 2618 2619
	if (bdev->bd_openers)
		goto out;

2620 2621 2622 2623 2624 2625 2626 2627 2628 2629
	/*
	 * Check if we have been instructed to close. We will have
	 * deferred this request, because the bdev was still open.
	 */

	mutex_lock(&info->mutex);
	xbdev = info->xbdev;

	if (xbdev && xbdev->state == XenbusStateClosing) {
		/* pending switch to state closed */
2630
		dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
2631 2632 2633 2634 2635 2636 2637 2638
		xlvbd_release_gendisk(info);
		xenbus_frontend_closed(info->xbdev);
 	}

	mutex_unlock(&info->mutex);

	if (!xbdev) {
		/* sudden device removal */
2639
		dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
2640 2641 2642
		xlvbd_release_gendisk(info);
		disk->private_data = NULL;
		kfree(info);
2643
	}
2644

J
Jens Axboe 已提交
2645
out:
2646
	bdput(bdev);
2647
out_mutex:
2648
	mutex_unlock(&blkfront_mutex);
2649 2650
}

2651
static const struct block_device_operations xlvbd_block_fops =
2652 2653
{
	.owner = THIS_MODULE,
A
Al Viro 已提交
2654 2655
	.open = blkif_open,
	.release = blkif_release,
2656
	.getgeo = blkif_getgeo,
2657
	.ioctl = blkif_ioctl,
2658 2659 2660
};


2661
static const struct xenbus_device_id blkfront_ids[] = {
2662 2663 2664 2665
	{ "vbd" },
	{ "" }
};

2666 2667
static struct xenbus_driver blkfront_driver = {
	.ids  = blkfront_ids,
2668 2669 2670
	.probe = blkfront_probe,
	.remove = blkfront_remove,
	.resume = blkfront_resume,
2671
	.otherend_changed = blkback_changed,
2672
	.is_ready = blkfront_is_ready,
2673
};
2674 2675 2676

static int __init xlblk_init(void)
{
2677
	int ret;
2678
	int nr_cpus = num_online_cpus();
2679

2680
	if (!xen_domain())
2681 2682
		return -ENODEV;

2683
	if (xen_blkif_max_ring_order > XENBUS_MAX_RING_GRANT_ORDER) {
B
Bob Liu 已提交
2684
		pr_info("Invalid max_ring_order (%d), will use default max: %d.\n",
2685
			xen_blkif_max_ring_order, XENBUS_MAX_RING_GRANT_ORDER);
2686
		xen_blkif_max_ring_order = XENBUS_MAX_RING_GRANT_ORDER;
B
Bob Liu 已提交
2687 2688
	}

2689 2690 2691 2692 2693 2694
	if (xen_blkif_max_queues > nr_cpus) {
		pr_info("Invalid max_queues (%d), will use default max: %d.\n",
			xen_blkif_max_queues, nr_cpus);
		xen_blkif_max_queues = nr_cpus;
	}

2695
	if (!xen_has_pv_disk_devices())
2696 2697
		return -ENODEV;

2698 2699 2700 2701 2702 2703
	if (register_blkdev(XENVBD_MAJOR, DEV_NAME)) {
		printk(KERN_WARNING "xen_blk: can't get major %d with name %s\n",
		       XENVBD_MAJOR, DEV_NAME);
		return -ENODEV;
	}

2704
	ret = xenbus_register_frontend(&blkfront_driver);
2705 2706 2707 2708 2709 2710
	if (ret) {
		unregister_blkdev(XENVBD_MAJOR, DEV_NAME);
		return ret;
	}

	return 0;
2711 2712 2713 2714
}
module_init(xlblk_init);


2715
static void __exit xlblk_exit(void)
2716
{
2717 2718 2719
	xenbus_unregister_driver(&blkfront_driver);
	unregister_blkdev(XENVBD_MAJOR, DEV_NAME);
	kfree(minors);
2720 2721 2722 2723 2724 2725
}
module_exit(xlblk_exit);

MODULE_DESCRIPTION("Xen virtual block device frontend");
MODULE_LICENSE("GPL");
MODULE_ALIAS_BLOCKDEV_MAJOR(XENVBD_MAJOR);
2726
MODULE_ALIAS("xen:vbd");
2727
MODULE_ALIAS("xenblk");