scsi_dh_alua.c 30.6 KB
Newer Older
1 2 3
/*
 * Generic SCSI-3 ALUA SCSI Device Handler
 *
4
 * Copyright (C) 2007-2010 Hannes Reinecke, SUSE Linux Products GmbH.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
 * All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 *
 */
22
#include <linux/slab.h>
23
#include <linux/delay.h>
24
#include <linux/module.h>
25
#include <asm/unaligned.h>
26
#include <scsi/scsi.h>
27
#include <scsi/scsi_proto.h>
28
#include <scsi/scsi_dbg.h>
29 30 31 32
#include <scsi/scsi_eh.h>
#include <scsi/scsi_dh.h>

#define ALUA_DH_NAME "alua"
33
#define ALUA_DH_VER "2.0"
34 35 36 37 38 39

#define TPGS_SUPPORT_NONE		0x00
#define TPGS_SUPPORT_OPTIMIZED		0x01
#define TPGS_SUPPORT_NONOPTIMIZED	0x02
#define TPGS_SUPPORT_STANDBY		0x04
#define TPGS_SUPPORT_UNAVAILABLE	0x08
40
#define TPGS_SUPPORT_LBA_DEPENDENT	0x10
41 42 43
#define TPGS_SUPPORT_OFFLINE		0x40
#define TPGS_SUPPORT_TRANSITION		0x80

44 45 46
#define RTPG_FMT_MASK			0x70
#define RTPG_FMT_EXT_HDR		0x10

47 48 49 50 51
#define TPGS_MODE_UNINITIALIZED		 -1
#define TPGS_MODE_NONE			0x0
#define TPGS_MODE_IMPLICIT		0x1
#define TPGS_MODE_EXPLICIT		0x2

52
#define ALUA_RTPG_SIZE			128
53
#define ALUA_FAILOVER_TIMEOUT		60
54
#define ALUA_FAILOVER_RETRIES		5
55
#define ALUA_RTPG_DELAY_MSECS		5
56

57
/* device handler flags */
58 59
#define ALUA_OPTIMIZE_STPG		0x01
#define ALUA_RTPG_EXT_HDR_UNSUPP	0x02
60
#define ALUA_SYNC_STPG			0x04
61 62 63 64
/* State machine flags */
#define ALUA_PG_RUN_RTPG		0x10
#define ALUA_PG_RUN_STPG		0x20
#define ALUA_PG_RUNNING			0x40
65

66 67 68 69
static uint optimize_stpg;
module_param(optimize_stpg, uint, S_IRUGO|S_IWUSR);
MODULE_PARM_DESC(optimize_stpg, "Allow use of a non-optimized path, rather than sending a STPG, when implicit TPGS is supported (0=No,1=Yes). Default is 0.");

70 71
static LIST_HEAD(port_group_list);
static DEFINE_SPINLOCK(port_group_lock);
72
static struct workqueue_struct *kaluad_wq;
73
static struct workqueue_struct *kaluad_sync_wq;
74 75 76

struct alua_port_group {
	struct kref		kref;
77
	struct rcu_head		rcu;
78
	struct list_head	node;
79
	struct list_head	dh_list;
80 81
	unsigned char		device_id_str[256];
	int			device_id_len;
82 83 84
	int			group_id;
	int			tpgs;
	int			state;
85
	int			pref;
86
	unsigned		flags; /* used for optimizing STPG */
87
	unsigned char		transition_tmo;
88 89 90 91 92 93
	unsigned long		expiry;
	unsigned long		interval;
	struct delayed_work	rtpg_work;
	spinlock_t		lock;
	struct list_head	rtpg_list;
	struct scsi_device	*rtpg_sdev;
94 95 96
};

struct alua_dh_data {
97
	struct list_head	node;
98
	struct alua_port_group __rcu *pg;
99
	int			group_id;
100
	spinlock_t		pg_lock;
101
	struct scsi_device	*sdev;
102 103 104 105 106 107
	int			init_error;
	struct mutex		init_mutex;
};

struct alua_queue_data {
	struct list_head	entry;
108 109
	activate_complete	callback_fn;
	void			*callback_data;
110 111 112 113 114
};

#define ALUA_POLICY_SWITCH_CURRENT	0
#define ALUA_POLICY_SWITCH_ALL		1

115 116 117
static void alua_rtpg_work(struct work_struct *work);
static void alua_rtpg_queue(struct alua_port_group *pg,
			    struct scsi_device *sdev,
118 119
			    struct alua_queue_data *qdata, bool force);
static void alua_check(struct scsi_device *sdev, bool force);
120

121 122 123 124 125
static void release_port_group(struct kref *kref)
{
	struct alua_port_group *pg;

	pg = container_of(kref, struct alua_port_group, kref);
126 127
	if (pg->rtpg_sdev)
		flush_delayed_work(&pg->rtpg_work);
128 129 130
	spin_lock(&port_group_lock);
	list_del(&pg->node);
	spin_unlock(&port_group_lock);
131
	kfree_rcu(pg, rcu);
132 133
}

134 135 136 137
/*
 * submit_rtpg - Issue a REPORT TARGET GROUP STATES command
 * @sdev: sdev the command should be sent to
 */
138 139
static int submit_rtpg(struct scsi_device *sdev, unsigned char *buff,
		       int bufflen, struct scsi_sense_hdr *sshdr, int flags)
140
{
141 142 143
	u8 cdb[COMMAND_SIZE(MAINTENANCE_IN)];
	int req_flags = REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT |
		REQ_FAILFAST_DRIVER;
144 145

	/* Prepare the command. */
146 147
	memset(cdb, 0x0, COMMAND_SIZE(MAINTENANCE_IN));
	cdb[0] = MAINTENANCE_IN;
148
	if (!(flags & ALUA_RTPG_EXT_HDR_UNSUPP))
149
		cdb[1] = MI_REPORT_TARGET_PGS | MI_EXT_HDR_PARAM_FMT;
150
	else
151 152 153
		cdb[1] = MI_REPORT_TARGET_PGS;
	put_unaligned_be32(bufflen, &cdb[6]);

154 155 156
	return scsi_execute(sdev, cdb, DMA_FROM_DEVICE, buff, bufflen, NULL,
			sshdr, ALUA_FAILOVER_TIMEOUT * HZ,
			ALUA_FAILOVER_RETRIES, req_flags, 0, NULL);
157 158
}

159
/*
160
 * submit_stpg - Issue a SET TARGET PORT GROUP command
161 162 163 164 165
 *
 * Currently we're only setting the current target port group state
 * to 'active/optimized' and let the array firmware figure out
 * the states of the remaining groups.
 */
166 167
static int submit_stpg(struct scsi_device *sdev, int group_id,
		       struct scsi_sense_hdr *sshdr)
168
{
169
	u8 cdb[COMMAND_SIZE(MAINTENANCE_OUT)];
170
	unsigned char stpg_data[8];
171
	int stpg_len = 8;
172 173
	int req_flags = REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT |
		REQ_FAILFAST_DRIVER;
174 175

	/* Prepare the data buffer */
176
	memset(stpg_data, 0, stpg_len);
177
	stpg_data[4] = SCSI_ACCESS_STATE_OPTIMAL;
178
	put_unaligned_be16(group_id, &stpg_data[6]);
179 180

	/* Prepare the command. */
181 182 183 184 185
	memset(cdb, 0x0, COMMAND_SIZE(MAINTENANCE_OUT));
	cdb[0] = MAINTENANCE_OUT;
	cdb[1] = MO_SET_TARGET_PGS;
	put_unaligned_be32(stpg_len, &cdb[6]);

186 187 188
	return scsi_execute(sdev, cdb, DMA_TO_DEVICE, stpg_data, stpg_len, NULL,
			sshdr, ALUA_FAILOVER_TIMEOUT * HZ,
			ALUA_FAILOVER_RETRIES, req_flags, 0, NULL);
189 190
}

191 192
static struct alua_port_group *alua_find_get_pg(char *id_str, size_t id_size,
						int group_id)
193 194 195
{
	struct alua_port_group *pg;

196 197 198
	if (!id_str || !id_size || !strlen(id_str))
		return NULL;

199 200 201
	list_for_each_entry(pg, &port_group_list, node) {
		if (pg->group_id != group_id)
			continue;
202
		if (!pg->device_id_len || pg->device_id_len != id_size)
203 204 205 206 207 208 209 210 211 212 213
			continue;
		if (strncmp(pg->device_id_str, id_str, id_size))
			continue;
		if (!kref_get_unless_zero(&pg->kref))
			continue;
		return pg;
	}

	return NULL;
}

214 215 216 217 218 219 220 221 222
/*
 * alua_alloc_pg - Allocate a new port_group structure
 * @sdev: scsi device
 * @h: alua device_handler data
 * @group_id: port group id
 *
 * Allocate a new port_group structure for a given
 * device.
 */
223 224
static struct alua_port_group *alua_alloc_pg(struct scsi_device *sdev,
					     int group_id, int tpgs)
225
{
226
	struct alua_port_group *pg, *tmp_pg;
227 228 229

	pg = kzalloc(sizeof(struct alua_port_group), GFP_KERNEL);
	if (!pg)
230
		return ERR_PTR(-ENOMEM);
231

232 233 234 235
	pg->device_id_len = scsi_vpd_lun_id(sdev, pg->device_id_str,
					    sizeof(pg->device_id_str));
	if (pg->device_id_len <= 0) {
		/*
236 237
		 * TPGS supported but no device identification found.
		 * Generate private device identification.
238 239 240 241
		 */
		sdev_printk(KERN_INFO, sdev,
			    "%s: No device descriptors found\n",
			    ALUA_DH_NAME);
242 243
		pg->device_id_str[0] = '\0';
		pg->device_id_len = 0;
244
	}
245 246
	pg->group_id = group_id;
	pg->tpgs = tpgs;
247
	pg->state = SCSI_ACCESS_STATE_OPTIMAL;
248 249
	if (optimize_stpg)
		pg->flags |= ALUA_OPTIMIZE_STPG;
250
	kref_init(&pg->kref);
251 252 253
	INIT_DELAYED_WORK(&pg->rtpg_work, alua_rtpg_work);
	INIT_LIST_HEAD(&pg->rtpg_list);
	INIT_LIST_HEAD(&pg->node);
254
	INIT_LIST_HEAD(&pg->dh_list);
255
	spin_lock_init(&pg->lock);
256

257
	spin_lock(&port_group_lock);
258 259 260 261 262 263 264 265
	tmp_pg = alua_find_get_pg(pg->device_id_str, pg->device_id_len,
				  group_id);
	if (tmp_pg) {
		spin_unlock(&port_group_lock);
		kfree(pg);
		return tmp_pg;
	}

266 267 268 269 270 271
	list_add(&pg->node, &port_group_list);
	spin_unlock(&port_group_lock);

	return pg;
}

272
/*
273
 * alua_check_tpgs - Evaluate TPGS setting
274 275
 * @sdev: device to be checked
 *
276
 * Examine the TPGS setting of the sdev to find out if ALUA
277 278
 * is supported.
 */
279
static int alua_check_tpgs(struct scsi_device *sdev)
280
{
281
	int tpgs = TPGS_MODE_NONE;
282

283 284 285 286 287 288 289 290
	/*
	 * ALUA support for non-disk devices is fraught with
	 * difficulties, so disable it for now.
	 */
	if (sdev->type != TYPE_DISK) {
		sdev_printk(KERN_INFO, sdev,
			    "%s: disable for non-disk devices\n",
			    ALUA_DH_NAME);
291
		return tpgs;
292 293
	}

294 295
	tpgs = scsi_device_tpgs(sdev);
	switch (tpgs) {
296 297 298 299 300 301 302 303 304 305 306 307 308
	case TPGS_MODE_EXPLICIT|TPGS_MODE_IMPLICIT:
		sdev_printk(KERN_INFO, sdev,
			    "%s: supports implicit and explicit TPGS\n",
			    ALUA_DH_NAME);
		break;
	case TPGS_MODE_EXPLICIT:
		sdev_printk(KERN_INFO, sdev, "%s: supports explicit TPGS\n",
			    ALUA_DH_NAME);
		break;
	case TPGS_MODE_IMPLICIT:
		sdev_printk(KERN_INFO, sdev, "%s: supports implicit TPGS\n",
			    ALUA_DH_NAME);
		break;
H
Hannes Reinecke 已提交
309
	case TPGS_MODE_NONE:
310 311 312
		sdev_printk(KERN_INFO, sdev, "%s: not supported\n",
			    ALUA_DH_NAME);
		break;
H
Hannes Reinecke 已提交
313 314 315
	default:
		sdev_printk(KERN_INFO, sdev,
			    "%s: unsupported TPGS setting %d\n",
316 317
			    ALUA_DH_NAME, tpgs);
		tpgs = TPGS_MODE_NONE;
H
Hannes Reinecke 已提交
318
		break;
319 320
	}

321
	return tpgs;
322 323 324
}

/*
325
 * alua_check_vpd - Evaluate INQUIRY vpd page 0x83
326 327 328 329 330
 * @sdev: device to be checked
 *
 * Extract the relative target port and the target port group
 * descriptor from the list of identificators.
 */
331 332
static int alua_check_vpd(struct scsi_device *sdev, struct alua_dh_data *h,
			  int tpgs)
333
{
334
	int rel_port = -1, group_id;
335
	struct alua_port_group *pg, *old_pg = NULL;
336
	bool pg_updated = false;
337
	unsigned long flags;
338

339 340
	group_id = scsi_vpd_tpg_id(sdev, &rel_port);
	if (group_id < 0) {
341 342 343 344 345 346 347 348
		/*
		 * Internal error; TPGS supported but required
		 * VPD identification descriptors not present.
		 * Disable ALUA support
		 */
		sdev_printk(KERN_INFO, sdev,
			    "%s: No target port descriptors found\n",
			    ALUA_DH_NAME);
349
		return SCSI_DH_DEV_UNSUPP;
350
	}
351

352 353 354
	pg = alua_alloc_pg(sdev, group_id, tpgs);
	if (IS_ERR(pg)) {
		if (PTR_ERR(pg) == -ENOMEM)
355 356 357
			return SCSI_DH_NOMEM;
		return SCSI_DH_DEV_UNSUPP;
	}
358 359 360 361 362 363 364 365 366
	if (pg->device_id_len)
		sdev_printk(KERN_INFO, sdev,
			    "%s: device %s port group %x rel port %x\n",
			    ALUA_DH_NAME, pg->device_id_str,
			    group_id, rel_port);
	else
		sdev_printk(KERN_INFO, sdev,
			    "%s: port group %x rel port %x\n",
			    ALUA_DH_NAME, group_id, rel_port);
367 368 369

	/* Check for existing port group references */
	spin_lock(&h->pg_lock);
370
	old_pg = rcu_dereference_protected(h->pg, lockdep_is_held(&h->pg_lock));
371 372
	if (old_pg != pg) {
		/* port group has changed. Update to new port group */
373 374 375 376 377
		if (h->pg) {
			spin_lock_irqsave(&old_pg->lock, flags);
			list_del_rcu(&h->node);
			spin_unlock_irqrestore(&old_pg->lock, flags);
		}
378
		rcu_assign_pointer(h->pg, pg);
379
		pg_updated = true;
380
	}
381 382

	spin_lock_irqsave(&pg->lock, flags);
383 384
	if (sdev->synchronous_alua)
		pg->flags |= ALUA_SYNC_STPG;
385 386 387 388
	if (pg_updated)
		list_add_rcu(&h->node, &pg->dh_list);
	spin_unlock_irqrestore(&pg->lock, flags);

389 390 391
	alua_rtpg_queue(rcu_dereference_protected(h->pg,
						  lockdep_is_held(&h->pg_lock)),
			sdev, NULL, true);
392 393 394 395
	spin_unlock(&h->pg_lock);

	if (old_pg)
		kref_put(&old_pg->kref, release_port_group);
396

397
	return SCSI_DH_OK;
398 399
}

400
static char print_alua_state(unsigned char state)
401 402
{
	switch (state) {
403
	case SCSI_ACCESS_STATE_OPTIMAL:
404
		return 'A';
405
	case SCSI_ACCESS_STATE_ACTIVE:
406
		return 'N';
407
	case SCSI_ACCESS_STATE_STANDBY:
408
		return 'S';
409
	case SCSI_ACCESS_STATE_UNAVAILABLE:
410
		return 'U';
411
	case SCSI_ACCESS_STATE_LBA:
412
		return 'L';
413
	case SCSI_ACCESS_STATE_OFFLINE:
414
		return 'O';
415
	case SCSI_ACCESS_STATE_TRANSITIONING:
416 417 418 419 420 421 422 423 424 425 426
		return 'T';
	default:
		return 'X';
	}
}

static int alua_check_sense(struct scsi_device *sdev,
			    struct scsi_sense_hdr *sense_hdr)
{
	switch (sense_hdr->sense_key) {
	case NOT_READY:
427
		if (sense_hdr->asc == 0x04 && sense_hdr->ascq == 0x0a) {
428 429 430
			/*
			 * LUN Not Accessible - ALUA state transition
			 */
431 432 433
			alua_check(sdev, false);
			return NEEDS_RETRY;
		}
434 435
		break;
	case UNIT_ATTENTION:
436
		if (sense_hdr->asc == 0x29 && sense_hdr->ascq == 0x00) {
437
			/*
438 439 440
			 * Power On, Reset, or Bus Device Reset.
			 * Might have obscured a state transition,
			 * so schedule a recheck.
441
			 */
442
			alua_check(sdev, true);
443
			return ADD_TO_MLQUEUE;
444
		}
445 446 447 448 449
		if (sense_hdr->asc == 0x29 && sense_hdr->ascq == 0x04)
			/*
			 * Device internal reset
			 */
			return ADD_TO_MLQUEUE;
450 451 452 453 454
		if (sense_hdr->asc == 0x2a && sense_hdr->ascq == 0x01)
			/*
			 * Mode Parameters Changed
			 */
			return ADD_TO_MLQUEUE;
455
		if (sense_hdr->asc == 0x2a && sense_hdr->ascq == 0x06) {
456 457 458
			/*
			 * ALUA state changed
			 */
459
			alua_check(sdev, true);
460
			return ADD_TO_MLQUEUE;
461 462
		}
		if (sense_hdr->asc == 0x2a && sense_hdr->ascq == 0x07) {
463 464 465
			/*
			 * Implicit ALUA state transition failed
			 */
466
			alua_check(sdev, true);
467
			return ADD_TO_MLQUEUE;
468
		}
469 470 471 472 473 474
		if (sense_hdr->asc == 0x3f && sense_hdr->ascq == 0x03)
			/*
			 * Inquiry data has changed
			 */
			return ADD_TO_MLQUEUE;
		if (sense_hdr->asc == 0x3f && sense_hdr->ascq == 0x0e)
475 476 477 478 479 480
			/*
			 * REPORTED_LUNS_DATA_HAS_CHANGED is reported
			 * when switching controllers on targets like
			 * Intel Multi-Flex. We can just retry.
			 */
			return ADD_TO_MLQUEUE;
481 482 483 484 485 486
		break;
	}

	return SCSI_RETURN_NOT_HANDLED;
}

487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
/*
 * alua_tur - Send a TEST UNIT READY
 * @sdev: device to which the TEST UNIT READY command should be send
 *
 * Send a TEST UNIT READY to @sdev to figure out the device state
 * Returns SCSI_DH_RETRY if the sense code is NOT READY/ALUA TRANSITIONING,
 * SCSI_DH_OK if no error occurred, and SCSI_DH_IO otherwise.
 */
static int alua_tur(struct scsi_device *sdev)
{
	struct scsi_sense_hdr sense_hdr;
	int retval;

	retval = scsi_test_unit_ready(sdev, ALUA_FAILOVER_TIMEOUT * HZ,
				      ALUA_FAILOVER_RETRIES, &sense_hdr);
	if (sense_hdr.sense_key == NOT_READY &&
	    sense_hdr.asc == 0x04 && sense_hdr.ascq == 0x0a)
		return SCSI_DH_RETRY;
	else if (retval)
		return SCSI_DH_IO;
	else
		return SCSI_DH_OK;
}

511 512 513 514 515 516
/*
 * alua_rtpg - Evaluate REPORT TARGET GROUP STATES
 * @sdev: the device to be evaluated.
 *
 * Evaluate the Target Port Group State.
 * Returns SCSI_DH_DEV_OFFLINED if the path is
L
Lucas De Marchi 已提交
517
 * found to be unusable.
518
 */
519
static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
520 521
{
	struct scsi_sense_hdr sense_hdr;
522
	struct alua_port_group *tmp_pg;
523
	int len, k, off, valid_states = 0, bufflen = ALUA_RTPG_SIZE;
524
	unsigned char *desc, *buff;
525
	unsigned err, retval;
526 527
	unsigned int tpg_desc_tbl_off;
	unsigned char orig_transition_tmo;
528
	unsigned long flags;
529

530 531 532 533 534 535 536 537
	if (!pg->expiry) {
		unsigned long transition_tmo = ALUA_FAILOVER_TIMEOUT * HZ;

		if (pg->transition_tmo)
			transition_tmo = pg->transition_tmo * HZ;

		pg->expiry = round_jiffies_up(jiffies + transition_tmo);
	}
538

539 540 541 542
	buff = kzalloc(bufflen, GFP_KERNEL);
	if (!buff)
		return SCSI_DH_DEV_TEMP_BUSY;

543
 retry:
544
	err = 0;
545
	retval = submit_rtpg(sdev, buff, bufflen, &sense_hdr, pg->flags);
546

547
	if (retval) {
548
		if (!scsi_sense_valid(&sense_hdr)) {
549 550 551
			sdev_printk(KERN_INFO, sdev,
				    "%s: rtpg failed, result %d\n",
				    ALUA_DH_NAME, retval);
552
			kfree(buff);
553
			if (driver_byte(retval) == DRIVER_ERROR)
554
				return SCSI_DH_DEV_TEMP_BUSY;
555
			return SCSI_DH_IO;
556
		}
557

558 559 560 561 562 563 564 565
		/*
		 * submit_rtpg() has failed on existing arrays
		 * when requesting extended header info, and
		 * the array doesn't support extended headers,
		 * even though it shouldn't according to T10.
		 * The retry without rtpg_ext_hdr_req set
		 * handles this.
		 */
566
		if (!(pg->flags & ALUA_RTPG_EXT_HDR_UNSUPP) &&
567 568
		    sense_hdr.sense_key == ILLEGAL_REQUEST &&
		    sense_hdr.asc == 0x24 && sense_hdr.ascq == 0) {
569
			pg->flags |= ALUA_RTPG_EXT_HDR_UNSUPP;
570 571
			goto retry;
		}
572 573 574 575 576 577 578 579 580
		/*
		 * Retry on ALUA state transition or if any
		 * UNIT ATTENTION occurred.
		 */
		if (sense_hdr.sense_key == NOT_READY &&
		    sense_hdr.asc == 0x04 && sense_hdr.ascq == 0x0a)
			err = SCSI_DH_RETRY;
		else if (sense_hdr.sense_key == UNIT_ATTENTION)
			err = SCSI_DH_RETRY;
581 582
		if (err == SCSI_DH_RETRY &&
		    pg->expiry != 0 && time_before(jiffies, pg->expiry)) {
583 584 585
			sdev_printk(KERN_ERR, sdev, "%s: rtpg retry\n",
				    ALUA_DH_NAME);
			scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
586
			kfree(buff);
587
			return err;
588 589 590 591
		}
		sdev_printk(KERN_ERR, sdev, "%s: rtpg failed\n",
			    ALUA_DH_NAME);
		scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
592
		kfree(buff);
593
		pg->expiry = 0;
594
		return SCSI_DH_IO;
595 596
	}

597
	len = get_unaligned_be32(&buff[0]) + 4;
598

599
	if (len > bufflen) {
600
		/* Resubmit with the correct length */
601 602 603 604
		kfree(buff);
		bufflen = len;
		buff = kmalloc(bufflen, GFP_KERNEL);
		if (!buff) {
605
			sdev_printk(KERN_WARNING, sdev,
606
				    "%s: kmalloc buffer failed\n",__func__);
607
			/* Temporary failure, bypass */
608
			pg->expiry = 0;
609 610 611 612 613
			return SCSI_DH_DEV_TEMP_BUSY;
		}
		goto retry;
	}

614
	orig_transition_tmo = pg->transition_tmo;
615
	if ((buff[4] & RTPG_FMT_MASK) == RTPG_FMT_EXT_HDR && buff[5] != 0)
616
		pg->transition_tmo = buff[5];
617
	else
618
		pg->transition_tmo = ALUA_FAILOVER_TIMEOUT;
619

620
	if (orig_transition_tmo != pg->transition_tmo) {
621 622
		sdev_printk(KERN_INFO, sdev,
			    "%s: transition timeout set to %d seconds\n",
623
			    ALUA_DH_NAME, pg->transition_tmo);
624
		pg->expiry = jiffies + pg->transition_tmo * HZ;
625 626
	}

627
	if ((buff[4] & RTPG_FMT_MASK) == RTPG_FMT_EXT_HDR)
628 629 630 631
		tpg_desc_tbl_off = 8;
	else
		tpg_desc_tbl_off = 4;

632
	for (k = tpg_desc_tbl_off, desc = buff + tpg_desc_tbl_off;
633
	     k < len;
634 635 636 637 638 639 640 641 642 643 644
	     k += off, desc += off) {
		u16 group_id = get_unaligned_be16(&desc[2]);

		spin_lock_irqsave(&port_group_lock, flags);
		tmp_pg = alua_find_get_pg(pg->device_id_str, pg->device_id_len,
					  group_id);
		spin_unlock_irqrestore(&port_group_lock, flags);
		if (tmp_pg) {
			if (spin_trylock_irqsave(&tmp_pg->lock, flags)) {
				if ((tmp_pg == pg) ||
				    !(tmp_pg->flags & ALUA_PG_RUNNING)) {
645 646
					struct alua_dh_data *h;

647 648
					tmp_pg->state = desc[0] & 0x0f;
					tmp_pg->pref = desc[0] >> 7;
649 650 651 652 653 654 655 656
					rcu_read_lock();
					list_for_each_entry_rcu(h,
						&tmp_pg->dh_list, node) {
						/* h->sdev should always be valid */
						BUG_ON(!h->sdev);
						h->sdev->access_state = desc[0];
					}
					rcu_read_unlock();
657 658 659 660 661 662
				}
				if (tmp_pg == pg)
					valid_states = desc[1];
				spin_unlock_irqrestore(&tmp_pg->lock, flags);
			}
			kref_put(&tmp_pg->kref, release_port_group);
663
		}
664
		off = 8 + (desc[7] * 4);
665 666
	}

667
	spin_lock_irqsave(&pg->lock, flags);
668
	sdev_printk(KERN_INFO, sdev,
669
		    "%s: port group %02x state %c %s supports %c%c%c%c%c%c%c\n",
670 671
		    ALUA_DH_NAME, pg->group_id, print_alua_state(pg->state),
		    pg->pref ? "preferred" : "non-preferred",
672 673
		    valid_states&TPGS_SUPPORT_TRANSITION?'T':'t',
		    valid_states&TPGS_SUPPORT_OFFLINE?'O':'o',
674
		    valid_states&TPGS_SUPPORT_LBA_DEPENDENT?'L':'l',
675 676 677 678 679
		    valid_states&TPGS_SUPPORT_UNAVAILABLE?'U':'u',
		    valid_states&TPGS_SUPPORT_STANDBY?'S':'s',
		    valid_states&TPGS_SUPPORT_NONOPTIMIZED?'N':'n',
		    valid_states&TPGS_SUPPORT_OPTIMIZED?'A':'a');

680
	switch (pg->state) {
681
	case SCSI_ACCESS_STATE_TRANSITIONING:
682
		if (time_before(jiffies, pg->expiry)) {
683
			/* State transition, retry */
684 685 686
			pg->interval = 2;
			err = SCSI_DH_RETRY;
		} else {
687 688
			struct alua_dh_data *h;

689 690
			/* Transitioning time exceeded, set port to standby */
			err = SCSI_DH_IO;
691
			pg->state = SCSI_ACCESS_STATE_STANDBY;
692
			pg->expiry = 0;
693 694 695 696 697 698 699 700 701 702
			rcu_read_lock();
			list_for_each_entry_rcu(h, &pg->dh_list, node) {
				BUG_ON(!h->sdev);
				h->sdev->access_state =
					(pg->state & SCSI_ACCESS_STATE_MASK);
				if (pg->pref)
					h->sdev->access_state |=
						SCSI_ACCESS_STATE_PREFERRED;
			}
			rcu_read_unlock();
703
		}
704
		break;
705
	case SCSI_ACCESS_STATE_OFFLINE:
706
		/* Path unusable */
707
		err = SCSI_DH_DEV_OFFLINED;
708
		pg->expiry = 0;
709 710 711 712
		break;
	default:
		/* Useable path if active */
		err = SCSI_DH_OK;
713
		pg->expiry = 0;
714
		break;
715
	}
716
	spin_unlock_irqrestore(&pg->lock, flags);
717
	kfree(buff);
718 719 720
	return err;
}

721 722 723 724
/*
 * alua_stpg - Issue a SET TARGET PORT GROUP command
 *
 * Issue a SET TARGET PORT GROUP command and evaluate the
725 726 727
 * response. Returns SCSI_DH_RETRY per default to trigger
 * a re-evaluation of the target group state or SCSI_DH_OK
 * if no further action needs to be taken.
728
 */
729
static unsigned alua_stpg(struct scsi_device *sdev, struct alua_port_group *pg)
730
{
731 732
	int retval;
	struct scsi_sense_hdr sense_hdr;
733

734
	if (!(pg->tpgs & TPGS_MODE_EXPLICIT)) {
735 736 737
		/* Only implicit ALUA supported, retry */
		return SCSI_DH_RETRY;
	}
738
	switch (pg->state) {
739
	case SCSI_ACCESS_STATE_OPTIMAL:
740
		return SCSI_DH_OK;
741
	case SCSI_ACCESS_STATE_ACTIVE:
742 743 744
		if ((pg->flags & ALUA_OPTIMIZE_STPG) &&
		    !pg->pref &&
		    (pg->tpgs & TPGS_MODE_IMPLICIT))
745
			return SCSI_DH_OK;
746
		break;
747 748
	case SCSI_ACCESS_STATE_STANDBY:
	case SCSI_ACCESS_STATE_UNAVAILABLE:
749
		break;
750
	case SCSI_ACCESS_STATE_OFFLINE:
751
		return SCSI_DH_IO;
752
	case SCSI_ACCESS_STATE_TRANSITIONING:
753 754
		break;
	default:
755 756
		sdev_printk(KERN_INFO, sdev,
			    "%s: stpg failed, unhandled TPGS state %d",
757
			    ALUA_DH_NAME, pg->state);
758
		return SCSI_DH_NOSYS;
759
	}
760
	retval = submit_stpg(sdev, pg->group_id, &sense_hdr);
761

762
	if (retval) {
763
		if (!scsi_sense_valid(&sense_hdr)) {
764 765 766
			sdev_printk(KERN_INFO, sdev,
				    "%s: stpg failed, result %d",
				    ALUA_DH_NAME, retval);
767
			if (driver_byte(retval) == DRIVER_ERROR)
768 769
				return SCSI_DH_DEV_TEMP_BUSY;
		} else {
770
			sdev_printk(KERN_INFO, sdev, "%s: stpg failed\n",
771 772 773
				    ALUA_DH_NAME);
			scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
		}
774
	}
775 776
	/* Retry RTPG */
	return SCSI_DH_RETRY;
777 778
}

779 780 781 782 783 784 785 786 787
static void alua_rtpg_work(struct work_struct *work)
{
	struct alua_port_group *pg =
		container_of(work, struct alua_port_group, rtpg_work.work);
	struct scsi_device *sdev;
	LIST_HEAD(qdata_list);
	int err = SCSI_DH_OK;
	struct alua_queue_data *qdata, *tmp;
	unsigned long flags;
788
	struct workqueue_struct *alua_wq = kaluad_wq;
789 790 791 792 793 794 795

	spin_lock_irqsave(&pg->lock, flags);
	sdev = pg->rtpg_sdev;
	if (!sdev) {
		WARN_ON(pg->flags & ALUA_PG_RUN_RTPG);
		WARN_ON(pg->flags & ALUA_PG_RUN_STPG);
		spin_unlock_irqrestore(&pg->lock, flags);
796
		kref_put(&pg->kref, release_port_group);
797 798
		return;
	}
799 800
	if (pg->flags & ALUA_SYNC_STPG)
		alua_wq = kaluad_sync_wq;
801 802
	pg->flags |= ALUA_PG_RUNNING;
	if (pg->flags & ALUA_PG_RUN_RTPG) {
803 804
		int state = pg->state;

805 806
		pg->flags &= ~ALUA_PG_RUN_RTPG;
		spin_unlock_irqrestore(&pg->lock, flags);
807
		if (state == SCSI_ACCESS_STATE_TRANSITIONING) {
808 809 810 811 812 813 814 815 816 817 818
			if (alua_tur(sdev) == SCSI_DH_RETRY) {
				spin_lock_irqsave(&pg->lock, flags);
				pg->flags &= ~ALUA_PG_RUNNING;
				pg->flags |= ALUA_PG_RUN_RTPG;
				spin_unlock_irqrestore(&pg->lock, flags);
				queue_delayed_work(alua_wq, &pg->rtpg_work,
						   pg->interval * HZ);
				return;
			}
			/* Send RTPG on failure or if TUR indicates SUCCESS */
		}
819 820
		err = alua_rtpg(sdev, pg);
		spin_lock_irqsave(&pg->lock, flags);
821
		if (err == SCSI_DH_RETRY || pg->flags & ALUA_PG_RUN_RTPG) {
822 823 824
			pg->flags &= ~ALUA_PG_RUNNING;
			pg->flags |= ALUA_PG_RUN_RTPG;
			spin_unlock_irqrestore(&pg->lock, flags);
825
			queue_delayed_work(alua_wq, &pg->rtpg_work,
826 827 828 829 830 831 832 833 834 835 836
					   pg->interval * HZ);
			return;
		}
		if (err != SCSI_DH_OK)
			pg->flags &= ~ALUA_PG_RUN_STPG;
	}
	if (pg->flags & ALUA_PG_RUN_STPG) {
		pg->flags &= ~ALUA_PG_RUN_STPG;
		spin_unlock_irqrestore(&pg->lock, flags);
		err = alua_stpg(sdev, pg);
		spin_lock_irqsave(&pg->lock, flags);
837
		if (err == SCSI_DH_RETRY || pg->flags & ALUA_PG_RUN_RTPG) {
838 839 840 841
			pg->flags |= ALUA_PG_RUN_RTPG;
			pg->interval = 0;
			pg->flags &= ~ALUA_PG_RUNNING;
			spin_unlock_irqrestore(&pg->lock, flags);
842
			queue_delayed_work(alua_wq, &pg->rtpg_work,
843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866
					   pg->interval * HZ);
			return;
		}
	}

	list_splice_init(&pg->rtpg_list, &qdata_list);
	pg->rtpg_sdev = NULL;
	spin_unlock_irqrestore(&pg->lock, flags);

	list_for_each_entry_safe(qdata, tmp, &qdata_list, entry) {
		list_del(&qdata->entry);
		if (qdata->callback_fn)
			qdata->callback_fn(qdata->callback_data, err);
		kfree(qdata);
	}
	spin_lock_irqsave(&pg->lock, flags);
	pg->flags &= ~ALUA_PG_RUNNING;
	spin_unlock_irqrestore(&pg->lock, flags);
	scsi_device_put(sdev);
	kref_put(&pg->kref, release_port_group);
}

static void alua_rtpg_queue(struct alua_port_group *pg,
			    struct scsi_device *sdev,
867
			    struct alua_queue_data *qdata, bool force)
868 869 870
{
	int start_queue = 0;
	unsigned long flags;
871
	struct workqueue_struct *alua_wq = kaluad_wq;
872 873 874 875 876 877 878 879

	if (!pg)
		return;

	spin_lock_irqsave(&pg->lock, flags);
	if (qdata) {
		list_add_tail(&qdata->entry, &pg->rtpg_list);
		pg->flags |= ALUA_PG_RUN_STPG;
880
		force = true;
881 882 883 884 885 886 887 888
	}
	if (pg->rtpg_sdev == NULL) {
		pg->interval = 0;
		pg->flags |= ALUA_PG_RUN_RTPG;
		kref_get(&pg->kref);
		pg->rtpg_sdev = sdev;
		scsi_device_get(sdev);
		start_queue = 1;
889 890 891 892 893
	} else if (!(pg->flags & ALUA_PG_RUN_RTPG) && force) {
		pg->flags |= ALUA_PG_RUN_RTPG;
		/* Do not queue if the worker is already running */
		if (!(pg->flags & ALUA_PG_RUNNING)) {
			kref_get(&pg->kref);
894
			sdev = NULL;
895 896
			start_queue = 1;
		}
897
	}
898

899 900
	if (pg->flags & ALUA_SYNC_STPG)
		alua_wq = kaluad_sync_wq;
901 902 903
	spin_unlock_irqrestore(&pg->lock, flags);

	if (start_queue &&
904
	    !queue_delayed_work(alua_wq, &pg->rtpg_work,
905
				msecs_to_jiffies(ALUA_RTPG_DELAY_MSECS))) {
906 907
		if (sdev)
			scsi_device_put(sdev);
908 909 910 911
		kref_put(&pg->kref, release_port_group);
	}
}

912 913 914 915 916 917 918 919 920
/*
 * alua_initialize - Initialize ALUA state
 * @sdev: the device to be initialized
 *
 * For the prep_fn to work correctly we have
 * to initialize the ALUA state for the device.
 */
static int alua_initialize(struct scsi_device *sdev, struct alua_dh_data *h)
{
921
	int err = SCSI_DH_DEV_UNSUPP, tpgs;
922

923
	mutex_lock(&h->init_mutex);
924
	tpgs = alua_check_tpgs(sdev);
925 926
	if (tpgs != TPGS_MODE_NONE)
		err = alua_check_vpd(sdev, h, tpgs);
927 928
	h->init_error = err;
	mutex_unlock(&h->init_mutex);
929 930
	return err;
}
931 932 933 934 935 936 937 938 939 940 941
/*
 * alua_set_params - set/unset the optimize flag
 * @sdev: device on the path to be activated
 * params - parameters in the following format
 *      "no_of_params\0param1\0param2\0param3\0...\0"
 * For example, to set the flag pass the following parameters
 * from multipath.conf
 *     hardware_handler        "2 alua 1"
 */
static int alua_set_params(struct scsi_device *sdev, const char *params)
{
942
	struct alua_dh_data *h = sdev->handler_data;
943
	struct alua_port_group *pg = NULL;
944 945 946
	unsigned int optimize = 0, argc;
	const char *p = params;
	int result = SCSI_DH_OK;
947
	unsigned long flags;
948 949 950 951 952 953 954 955 956

	if ((sscanf(params, "%u", &argc) != 1) || (argc != 1))
		return -EINVAL;

	while (*p++)
		;
	if ((sscanf(p, "%u", &optimize) != 1) || (optimize > 1))
		return -EINVAL;

957 958 959 960
	rcu_read_lock();
	pg = rcu_dereference(h->pg);
	if (!pg) {
		rcu_read_unlock();
961
		return -ENXIO;
962 963
	}
	spin_lock_irqsave(&pg->lock, flags);
964
	if (optimize)
965
		pg->flags |= ALUA_OPTIMIZE_STPG;
966
	else
967
		pg->flags &= ~ALUA_OPTIMIZE_STPG;
968 969
	spin_unlock_irqrestore(&pg->lock, flags);
	rcu_read_unlock();
970 971 972

	return result;
}
973 974 975 976 977 978 979 980 981 982 983

/*
 * alua_activate - activate a path
 * @sdev: device on the path to be activated
 *
 * We're currently switching the port group to be activated only and
 * let the array figure out the rest.
 * There may be other arrays which require us to switch all port groups
 * based on a certain policy. But until we actually encounter them it
 * should be okay.
 */
984 985
static int alua_activate(struct scsi_device *sdev,
			activate_complete fn, void *data)
986
{
987
	struct alua_dh_data *h = sdev->handler_data;
988
	int err = SCSI_DH_OK;
989
	struct alua_queue_data *qdata;
990
	struct alua_port_group *pg;
991

992 993 994
	qdata = kzalloc(sizeof(*qdata), GFP_KERNEL);
	if (!qdata) {
		err = SCSI_DH_RES_TEMP_UNAVAIL;
995
		goto out;
996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
	}
	qdata->callback_fn = fn;
	qdata->callback_data = data;

	mutex_lock(&h->init_mutex);
	rcu_read_lock();
	pg = rcu_dereference(h->pg);
	if (!pg || !kref_get_unless_zero(&pg->kref)) {
		rcu_read_unlock();
		kfree(qdata);
		err = h->init_error;
		mutex_unlock(&h->init_mutex);
1008 1009
		goto out;
	}
1010 1011 1012 1013
	fn = NULL;
	rcu_read_unlock();
	mutex_unlock(&h->init_mutex);

1014
	alua_rtpg_queue(pg, sdev, qdata, true);
1015
	kref_put(&pg->kref, release_port_group);
1016
out:
1017
	if (fn)
1018 1019
		fn(data, err);
	return 0;
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
/*
 * alua_check - check path status
 * @sdev: device on the path to be checked
 *
 * Check the device status
 */
static void alua_check(struct scsi_device *sdev, bool force)
{
	struct alua_dh_data *h = sdev->handler_data;
	struct alua_port_group *pg;

	rcu_read_lock();
	pg = rcu_dereference(h->pg);
	if (!pg || !kref_get_unless_zero(&pg->kref)) {
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();

	alua_rtpg_queue(pg, sdev, NULL, force);
	kref_put(&pg->kref, release_port_group);
}

1045 1046 1047 1048 1049 1050 1051 1052
/*
 * alua_prep_fn - request callback
 *
 * Fail I/O to all paths not in state
 * active/optimized or active/non-optimized.
 */
static int alua_prep_fn(struct scsi_device *sdev, struct request *req)
{
1053
	struct alua_dh_data *h = sdev->handler_data;
1054
	struct alua_port_group *pg;
1055
	unsigned char state = SCSI_ACCESS_STATE_OPTIMAL;
1056 1057
	int ret = BLKPREP_OK;

1058 1059 1060 1061 1062
	rcu_read_lock();
	pg = rcu_dereference(h->pg);
	if (pg)
		state = pg->state;
	rcu_read_unlock();
1063
	if (state == SCSI_ACCESS_STATE_TRANSITIONING)
1064
		ret = BLKPREP_DEFER;
1065 1066 1067
	else if (state != SCSI_ACCESS_STATE_OPTIMAL &&
		 state != SCSI_ACCESS_STATE_ACTIVE &&
		 state != SCSI_ACCESS_STATE_LBA) {
1068
		ret = BLKPREP_KILL;
1069
		req->rq_flags |= RQF_QUIET;
1070 1071 1072 1073 1074
	}
	return ret;

}

H
Hannes Reinecke 已提交
1075 1076 1077 1078 1079 1080 1081
static void alua_rescan(struct scsi_device *sdev)
{
	struct alua_dh_data *h = sdev->handler_data;

	alua_initialize(sdev, h);
}

1082 1083 1084 1085
/*
 * alua_bus_attach - Attach device handler
 * @sdev: device to be attached to
 */
1086
static int alua_bus_attach(struct scsi_device *sdev)
1087 1088
{
	struct alua_dh_data *h;
1089
	int err, ret = -EINVAL;
1090

1091
	h = kzalloc(sizeof(*h) , GFP_KERNEL);
1092
	if (!h)
1093
		return -ENOMEM;
1094 1095 1096
	spin_lock_init(&h->pg_lock);
	rcu_assign_pointer(h->pg, NULL);
	h->init_error = SCSI_DH_OK;
1097
	h->sdev = sdev;
1098
	INIT_LIST_HEAD(&h->node);
1099

1100
	mutex_init(&h->init_mutex);
1101
	err = alua_initialize(sdev, h);
1102 1103
	if (err == SCSI_DH_NOMEM)
		ret = -ENOMEM;
1104
	if (err != SCSI_DH_OK && err != SCSI_DH_DEV_OFFLINED)
1105 1106
		goto failed;

1107 1108
	sdev->handler_data = h;
	return 0;
1109
failed:
1110
	kfree(h);
1111
	return ret;
1112 1113 1114 1115 1116 1117 1118 1119
}

/*
 * alua_bus_detach - Detach device handler
 * @sdev: device to be detached from
 */
static void alua_bus_detach(struct scsi_device *sdev)
{
1120
	struct alua_dh_data *h = sdev->handler_data;
1121 1122 1123
	struct alua_port_group *pg;

	spin_lock(&h->pg_lock);
1124
	pg = rcu_dereference_protected(h->pg, lockdep_is_held(&h->pg_lock));
1125 1126 1127
	rcu_assign_pointer(h->pg, NULL);
	h->sdev = NULL;
	spin_unlock(&h->pg_lock);
1128
	if (pg) {
1129
		spin_lock_irq(&pg->lock);
1130
		list_del_rcu(&h->node);
1131
		spin_unlock_irq(&pg->lock);
1132
		kref_put(&pg->kref, release_port_group);
1133
	}
1134
	sdev->handler_data = NULL;
1135
	kfree(h);
1136 1137
}

1138 1139 1140 1141 1142 1143 1144 1145
static struct scsi_device_handler alua_dh = {
	.name = ALUA_DH_NAME,
	.module = THIS_MODULE,
	.attach = alua_bus_attach,
	.detach = alua_bus_detach,
	.prep_fn = alua_prep_fn,
	.check_sense = alua_check_sense,
	.activate = alua_activate,
H
Hannes Reinecke 已提交
1146
	.rescan = alua_rescan,
1147 1148 1149
	.set_params = alua_set_params,
};

1150 1151 1152 1153
static int __init alua_init(void)
{
	int r;

1154 1155 1156 1157 1158
	kaluad_wq = alloc_workqueue("kaluad", WQ_MEM_RECLAIM, 0);
	if (!kaluad_wq) {
		/* Temporary failure, bypass */
		return SCSI_DH_DEV_TEMP_BUSY;
	}
1159 1160 1161 1162 1163
	kaluad_sync_wq = create_workqueue("kaluad_sync");
	if (!kaluad_sync_wq) {
		destroy_workqueue(kaluad_wq);
		return SCSI_DH_DEV_TEMP_BUSY;
	}
1164
	r = scsi_register_device_handler(&alua_dh);
1165
	if (r != 0) {
1166 1167
		printk(KERN_ERR "%s: Failed to register scsi device handler",
			ALUA_DH_NAME);
1168
		destroy_workqueue(kaluad_sync_wq);
1169 1170
		destroy_workqueue(kaluad_wq);
	}
1171 1172 1173 1174 1175 1176
	return r;
}

static void __exit alua_exit(void)
{
	scsi_unregister_device_handler(&alua_dh);
1177
	destroy_workqueue(kaluad_sync_wq);
1178
	destroy_workqueue(kaluad_wq);
1179 1180 1181 1182 1183 1184 1185 1186 1187
}

module_init(alua_init);
module_exit(alua_exit);

MODULE_DESCRIPTION("DM Multipath ALUA support");
MODULE_AUTHOR("Hannes Reinecke <hare@suse.de>");
MODULE_LICENSE("GPL");
MODULE_VERSION(ALUA_DH_VER);