scsi_error.c 66.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5
/*
 *  scsi_error.c Copyright (C) 1997 Eric Youngdale
 *
 *  SCSI error/timeout handling
 *      Initial versions: Eric Youngdale.  Based upon conversations with
6
 *                        Leonard Zubkoff and David Miller at Linux Expo,
L
Linus Torvalds 已提交
7 8 9 10 11 12
 *                        ideas originating from all over the place.
 *
 *	Restructured scsi_unjam_host and associated functions.
 *	September 04, 2002 Mike Anderson (andmike@us.ibm.com)
 *
 *	Forward port of Russell King's (rmk@arm.linux.org.uk) changes and
13
 *	minor cleanups.
L
Linus Torvalds 已提交
14 15 16 17 18
 *	September 30, 2002 Mike Anderson (andmike@us.ibm.com)
 */

#include <linux/module.h>
#include <linux/sched.h>
19
#include <linux/gfp.h>
L
Linus Torvalds 已提交
20 21 22
#include <linux/timer.h>
#include <linux/string.h>
#include <linux/kernel.h>
23
#include <linux/freezer.h>
24
#include <linux/kthread.h>
L
Linus Torvalds 已提交
25 26 27
#include <linux/interrupt.h>
#include <linux/blkdev.h>
#include <linux/delay.h>
28
#include <linux/jiffies.h>
L
Linus Torvalds 已提交
29 30

#include <scsi/scsi.h>
31
#include <scsi/scsi_cmnd.h>
L
Linus Torvalds 已提交
32 33
#include <scsi/scsi_dbg.h>
#include <scsi/scsi_device.h>
34
#include <scsi/scsi_driver.h>
L
Linus Torvalds 已提交
35
#include <scsi/scsi_eh.h>
36
#include <scsi/scsi_common.h>
37
#include <scsi/scsi_transport.h>
L
Linus Torvalds 已提交
38 39
#include <scsi/scsi_host.h>
#include <scsi/scsi_ioctl.h>
40
#include <scsi/scsi_dh.h>
41
#include <scsi/sg.h>
L
Linus Torvalds 已提交
42 43 44

#include "scsi_priv.h"
#include "scsi_logging.h"
45
#include "scsi_transport_api.h"
L
Linus Torvalds 已提交
46

47 48
#include <trace/events/scsi.h>

49 50
#include <asm/unaligned.h>

51 52
static void scsi_eh_done(struct scsi_cmnd *scmd);

L
Linus Torvalds 已提交
53 54 55 56 57 58 59
/*
 * These should *probably* be handled by the host itself.
 * Since it is allowed to sleep, it probably should.
 */
#define BUS_RESET_SETTLE_TIME   (10)
#define HOST_RESET_SETTLE_TIME  (10)

60
static int scsi_eh_try_stu(struct scsi_cmnd *scmd);
61 62
static int scsi_try_to_abort_cmd(struct scsi_host_template *,
				 struct scsi_cmnd *);
63

L
Linus Torvalds 已提交
64 65
void scsi_eh_wakeup(struct Scsi_Host *shost)
{
66 67
	lockdep_assert_held(shost->host_lock);

68
	if (atomic_read(&shost->host_busy) == shost->host_failed) {
69
		trace_scsi_eh_wakeup(shost);
70
		wake_up_process(shost->ehandler);
71 72
		SCSI_LOG_ERROR_RECOVERY(5, shost_printk(KERN_INFO, shost,
			"Waking error handler thread\n"));
L
Linus Torvalds 已提交
73 74
	}
}
75 76 77 78 79 80

/**
 * scsi_schedule_eh - schedule EH for SCSI host
 * @shost:	SCSI host to invoke error handling on.
 *
 * Schedule SCSI EH without scmd.
81
 */
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
void scsi_schedule_eh(struct Scsi_Host *shost)
{
	unsigned long flags;

	spin_lock_irqsave(shost->host_lock, flags);

	if (scsi_host_set_state(shost, SHOST_RECOVERY) == 0 ||
	    scsi_host_set_state(shost, SHOST_CANCEL_RECOVERY) == 0) {
		shost->host_eh_scheduled++;
		scsi_eh_wakeup(shost);
	}

	spin_unlock_irqrestore(shost->host_lock, flags);
}
EXPORT_SYMBOL_GPL(scsi_schedule_eh);
L
Linus Torvalds 已提交
97

98 99
static int scsi_host_eh_past_deadline(struct Scsi_Host *shost)
{
100
	if (!shost->last_reset || shost->eh_deadline == -1)
101 102
		return 0;

103 104 105 106
	/*
	 * 32bit accesses are guaranteed to be atomic
	 * (on all supported architectures), so instead
	 * of using a spinlock we can as well double check
107
	 * if eh_deadline has been set to 'off' during the
108 109 110
	 * time_before call.
	 */
	if (time_before(jiffies, shost->last_reset + shost->eh_deadline) &&
111
	    shost->eh_deadline > -1)
112 113 114 115 116
		return 0;

	return 1;
}

117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
/**
 * scmd_eh_abort_handler - Handle command aborts
 * @work:	command to be aborted.
 */
void
scmd_eh_abort_handler(struct work_struct *work)
{
	struct scsi_cmnd *scmd =
		container_of(work, struct scsi_cmnd, abort_work.work);
	struct scsi_device *sdev = scmd->device;
	int rtn;

	if (scsi_host_eh_past_deadline(sdev->host)) {
		SCSI_LOG_ERROR_RECOVERY(3,
			scmd_printk(KERN_INFO, scmd,
132
				    "eh timeout, not aborting\n"));
133 134 135
	} else {
		SCSI_LOG_ERROR_RECOVERY(3,
			scmd_printk(KERN_INFO, scmd,
136
				    "aborting command\n"));
137 138
		rtn = scsi_try_to_abort_cmd(sdev->host->hostt, scmd);
		if (rtn == SUCCESS) {
139
			set_host_byte(scmd, DID_TIME_OUT);
140 141 142
			if (scsi_host_eh_past_deadline(sdev->host)) {
				SCSI_LOG_ERROR_RECOVERY(3,
					scmd_printk(KERN_INFO, scmd,
143 144
						    "eh timeout, not retrying "
						    "aborted command\n"));
145
			} else if (!scsi_noretry_cmd(scmd) &&
146 147 148
			    (++scmd->retries <= scmd->allowed)) {
				SCSI_LOG_ERROR_RECOVERY(3,
					scmd_printk(KERN_WARNING, scmd,
149
						    "retry aborted command\n"));
150
				scsi_queue_insert(scmd, SCSI_MLQUEUE_EH_RETRY);
151
				return;
152 153 154
			} else {
				SCSI_LOG_ERROR_RECOVERY(3,
					scmd_printk(KERN_WARNING, scmd,
155
						    "finish aborted command\n"));
156
				scsi_finish_command(scmd);
157
				return;
158
			}
159 160 161
		} else {
			SCSI_LOG_ERROR_RECOVERY(3,
				scmd_printk(KERN_INFO, scmd,
162
					    "cmd abort %s\n",
163 164
					    (rtn == FAST_IO_FAIL) ?
					    "not send" : "failed"));
165 166 167
		}
	}

168
	scsi_eh_scmd_add(scmd);
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
}

/**
 * scsi_abort_command - schedule a command abort
 * @scmd:	scmd to abort.
 *
 * We only need to abort commands after a command timeout
 */
static int
scsi_abort_command(struct scsi_cmnd *scmd)
{
	struct scsi_device *sdev = scmd->device;
	struct Scsi_Host *shost = sdev->host;
	unsigned long flags;

	if (scmd->eh_eflags & SCSI_EH_ABORT_SCHEDULED) {
		/*
		 * Retry after abort failed, escalate to next level.
		 */
		SCSI_LOG_ERROR_RECOVERY(3,
			scmd_printk(KERN_INFO, scmd,
190
				    "previous abort failed\n"));
191
		BUG_ON(delayed_work_pending(&scmd->abort_work));
192 193 194 195
		return FAILED;
	}

	spin_lock_irqsave(shost->host_lock, flags);
196
	if (shost->eh_deadline != -1 && !shost->last_reset)
197 198 199 200 201
		shost->last_reset = jiffies;
	spin_unlock_irqrestore(shost->host_lock, flags);

	scmd->eh_eflags |= SCSI_EH_ABORT_SCHEDULED;
	SCSI_LOG_ERROR_RECOVERY(3,
202
		scmd_printk(KERN_INFO, scmd, "abort scheduled\n"));
203 204 205 206
	queue_delayed_work(shost->tmf_work_q, &scmd->abort_work, HZ / 100);
	return SUCCESS;
}

L
Linus Torvalds 已提交
207
/**
208
 * scsi_eh_reset - call into ->eh_action to reset internal counters
L
Linus Torvalds 已提交
209 210
 * @scmd:	scmd to run eh on.
 *
211 212 213
 * The scsi driver might be carrying internal state about the
 * devices, so we need to call into the driver to reset the
 * internal state once the error handler is started.
214
 */
215 216 217 218 219 220 221 222 223
static void scsi_eh_reset(struct scsi_cmnd *scmd)
{
	if (!blk_rq_is_passthrough(scmd->request)) {
		struct scsi_driver *sdrv = scsi_cmd_to_driver(scmd);
		if (sdrv->eh_reset)
			sdrv->eh_reset(scmd);
	}
}

224 225 226 227 228 229 230 231 232 233 234
static void scsi_eh_inc_host_failed(struct rcu_head *head)
{
	struct Scsi_Host *shost = container_of(head, typeof(*shost), rcu);
	unsigned long flags;

	spin_lock_irqsave(shost->host_lock, flags);
	shost->host_failed++;
	scsi_eh_wakeup(shost);
	spin_unlock_irqrestore(shost->host_lock, flags);
}

L
Linus Torvalds 已提交
235 236 237
/**
 * scsi_eh_scmd_add - add scsi cmd to error handling.
 * @scmd:	scmd to run eh on.
238
 */
239
void scsi_eh_scmd_add(struct scsi_cmnd *scmd)
L
Linus Torvalds 已提交
240 241 242
{
	struct Scsi_Host *shost = scmd->device->host;
	unsigned long flags;
243
	int ret;
L
Linus Torvalds 已提交
244

245
	WARN_ON_ONCE(!shost->ehandler);
L
Linus Torvalds 已提交
246 247

	spin_lock_irqsave(shost->host_lock, flags);
248 249 250 251
	if (scsi_host_set_state(shost, SHOST_RECOVERY)) {
		ret = scsi_host_set_state(shost, SHOST_CANCEL_RECOVERY);
		WARN_ON_ONCE(ret);
	}
252
	if (shost->eh_deadline != -1 && !shost->last_reset)
253 254
		shost->last_reset = jiffies;

255
	scsi_eh_reset(scmd);
L
Linus Torvalds 已提交
256 257
	list_add_tail(&scmd->eh_entry, &shost->eh_cmd_q);
	spin_unlock_irqrestore(shost->host_lock, flags);
258 259 260 261 262
	/*
	 * Ensure that all tasks observe the host state change before the
	 * host_failed change.
	 */
	call_rcu(&shost->rcu, scsi_eh_inc_host_failed);
L
Linus Torvalds 已提交
263 264 265 266
}

/**
 * scsi_times_out - Timeout function for normal scsi commands.
J
Jens Axboe 已提交
267
 * @req:	request that is timing out.
L
Linus Torvalds 已提交
268 269 270 271 272 273
 *
 * Notes:
 *     We do not need to lock this.  There is the potential for a race
 *     only in that the normal completion handling might run, but if the
 *     normal completion function determines that the timer has already
 *     fired, then it mustn't do anything.
274
 */
J
Jens Axboe 已提交
275
enum blk_eh_timer_return scsi_times_out(struct request *req)
L
Linus Torvalds 已提交
276
{
277
	struct scsi_cmnd *scmd = blk_mq_rq_to_pdu(req);
J
Jens Axboe 已提交
278
	enum blk_eh_timer_return rtn = BLK_EH_NOT_HANDLED;
279
	struct Scsi_Host *host = scmd->device->host;
280

281
	trace_scsi_dispatch_cmd_timeout(scmd);
L
Linus Torvalds 已提交
282 283
	scsi_log_completion(scmd, TIMEOUT_ERROR);

284
	if (host->eh_deadline != -1 && !host->last_reset)
285 286
		host->last_reset = jiffies;

287
	if (host->hostt->eh_timed_out)
288
		rtn = host->hostt->eh_timed_out(scmd);
L
Linus Torvalds 已提交
289

290
	if (rtn == BLK_EH_NOT_HANDLED) {
291
		if (scsi_abort_command(scmd) != SUCCESS) {
292
			set_host_byte(scmd, DID_TIME_OUT);
293
			scsi_eh_scmd_add(scmd);
294
		}
295
	}
J
Jens Axboe 已提交
296

297
	return rtn;
L
Linus Torvalds 已提交
298 299 300 301 302 303 304 305 306 307 308 309
}

/**
 * scsi_block_when_processing_errors - Prevent cmds from being queued.
 * @sdev:	Device on which we are performing recovery.
 *
 * Description:
 *     We block until the host is out of error recovery, and then check to
 *     see whether the host or the device is offline.
 *
 * Return value:
 *     0 when dev was taken offline by error recovery. 1 OK to proceed.
310
 */
L
Linus Torvalds 已提交
311 312 313 314
int scsi_block_when_processing_errors(struct scsi_device *sdev)
{
	int online;

315
	wait_event(sdev->host->host_wait, !scsi_host_in_recovery(sdev->host));
L
Linus Torvalds 已提交
316 317 318

	online = scsi_device_online(sdev);

319 320
	SCSI_LOG_ERROR_RECOVERY(5, sdev_printk(KERN_INFO, sdev,
		"%s: rtn: %d\n", __func__, online));
L
Linus Torvalds 已提交
321 322 323 324 325 326 327 328 329 330

	return online;
}
EXPORT_SYMBOL(scsi_block_when_processing_errors);

#ifdef CONFIG_SCSI_LOGGING
/**
 * scsi_eh_prt_fail_stats - Log info on failures.
 * @shost:	scsi host being recovered.
 * @work_q:	Queue of scsi cmds to process.
331
 */
L
Linus Torvalds 已提交
332 333 334 335 336 337 338 339 340 341 342 343 344 345
static inline void scsi_eh_prt_fail_stats(struct Scsi_Host *shost,
					  struct list_head *work_q)
{
	struct scsi_cmnd *scmd;
	struct scsi_device *sdev;
	int total_failures = 0;
	int cmd_failed = 0;
	int cmd_cancel = 0;
	int devices_failed = 0;

	shost_for_each_device(sdev, shost) {
		list_for_each_entry(scmd, work_q, eh_entry) {
			if (scmd->device == sdev) {
				++total_failures;
346
				if (scmd->eh_eflags & SCSI_EH_ABORT_SCHEDULED)
L
Linus Torvalds 已提交
347
					++cmd_cancel;
348
				else
L
Linus Torvalds 已提交
349 350 351 352 353 354
					++cmd_failed;
			}
		}

		if (cmd_cancel || cmd_failed) {
			SCSI_LOG_ERROR_RECOVERY(3,
355
				shost_printk(KERN_INFO, shost,
356
					    "%s: cmds failed: %d, cancel: %d\n",
357
					    __func__, cmd_failed,
358
					    cmd_cancel));
L
Linus Torvalds 已提交
359 360 361 362 363 364
			cmd_cancel = 0;
			cmd_failed = 0;
			++devices_failed;
		}
	}

365 366 367
	SCSI_LOG_ERROR_RECOVERY(2, shost_printk(KERN_INFO, shost,
				   "Total of %d commands on %d"
				   " devices require eh work\n",
368
				   total_failures, devices_failed));
L
Linus Torvalds 已提交
369 370 371
}
#endif

372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
 /**
 * scsi_report_lun_change - Set flag on all *other* devices on the same target
 *                          to indicate that a UNIT ATTENTION is expected.
 * @sdev:	Device reporting the UNIT ATTENTION
 */
static void scsi_report_lun_change(struct scsi_device *sdev)
{
	sdev->sdev_target->expecting_lun_change = 1;
}

/**
 * scsi_report_sense - Examine scsi sense information and log messages for
 *		       certain conditions, also issue uevents for some of them.
 * @sdev:	Device reporting the sense code
 * @sshdr:	sshdr to be examined
 */
static void scsi_report_sense(struct scsi_device *sdev,
			      struct scsi_sense_hdr *sshdr)
{
	enum scsi_device_event evt_type = SDEV_EVT_MAXBITS;	/* i.e. none */

	if (sshdr->sense_key == UNIT_ATTENTION) {
		if (sshdr->asc == 0x3f && sshdr->ascq == 0x03) {
			evt_type = SDEV_EVT_INQUIRY_CHANGE_REPORTED;
			sdev_printk(KERN_WARNING, sdev,
				    "Inquiry data has changed");
		} else if (sshdr->asc == 0x3f && sshdr->ascq == 0x0e) {
			evt_type = SDEV_EVT_LUN_CHANGE_REPORTED;
			scsi_report_lun_change(sdev);
			sdev_printk(KERN_WARNING, sdev,
				    "Warning! Received an indication that the "
				    "LUN assignments on this target have "
				    "changed. The Linux SCSI layer does not "
				    "automatically remap LUN assignments.\n");
		} else if (sshdr->asc == 0x3f)
			sdev_printk(KERN_WARNING, sdev,
				    "Warning! Received an indication that the "
				    "operating parameters on this target have "
				    "changed. The Linux SCSI layer does not "
				    "automatically adjust these parameters.\n");

		if (sshdr->asc == 0x38 && sshdr->ascq == 0x07) {
			evt_type = SDEV_EVT_SOFT_THRESHOLD_REACHED_REPORTED;
			sdev_printk(KERN_WARNING, sdev,
				    "Warning! Received an indication that the "
				    "LUN reached a thin provisioning soft "
				    "threshold.\n");
		}

421 422 423 424 425 426
		if (sshdr->asc == 0x29) {
			evt_type = SDEV_EVT_POWER_ON_RESET_OCCURRED;
			sdev_printk(KERN_WARNING, sdev,
				    "Power-on or device reset occurred\n");
		}

427 428 429 430
		if (sshdr->asc == 0x2a && sshdr->ascq == 0x01) {
			evt_type = SDEV_EVT_MODE_PARAMETER_CHANGE_REPORTED;
			sdev_printk(KERN_WARNING, sdev,
				    "Mode parameters changed");
431 432 433 434
		} else if (sshdr->asc == 0x2a && sshdr->ascq == 0x06) {
			evt_type = SDEV_EVT_ALUA_STATE_CHANGE_REPORTED;
			sdev_printk(KERN_WARNING, sdev,
				    "Asymmetric access state changed");
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
		} else if (sshdr->asc == 0x2a && sshdr->ascq == 0x09) {
			evt_type = SDEV_EVT_CAPACITY_CHANGE_REPORTED;
			sdev_printk(KERN_WARNING, sdev,
				    "Capacity data has changed");
		} else if (sshdr->asc == 0x2a)
			sdev_printk(KERN_WARNING, sdev,
				    "Parameters changed");
	}

	if (evt_type != SDEV_EVT_MAXBITS) {
		set_bit(evt_type, sdev->pending_events);
		schedule_work(&sdev->event_work);
	}
}

L
Linus Torvalds 已提交
450 451 452 453 454
/**
 * scsi_check_sense - Examine scsi cmd sense
 * @scmd:	Cmd to have sense checked.
 *
 * Return value:
455
 *	SUCCESS or FAILED or NEEDS_RETRY or ADD_TO_MLQUEUE
L
Linus Torvalds 已提交
456 457 458 459
 *
 * Notes:
 *	When a deferred error is detected the current command has
 *	not been executed and needs retrying.
460
 */
461
int scsi_check_sense(struct scsi_cmnd *scmd)
L
Linus Torvalds 已提交
462
{
463
	struct scsi_device *sdev = scmd->device;
L
Linus Torvalds 已提交
464 465 466 467 468
	struct scsi_sense_hdr sshdr;

	if (! scsi_command_normalize_sense(scmd, &sshdr))
		return FAILED;	/* no valid sense data */

469 470
	scsi_report_sense(sdev, &sshdr);

L
Linus Torvalds 已提交
471 472 473
	if (scsi_sense_is_deferred(&sshdr))
		return NEEDS_RETRY;

474
	if (sdev->handler && sdev->handler->check_sense) {
475 476
		int rc;

477
		rc = sdev->handler->check_sense(sdev, &sshdr);
478 479 480 481 482
		if (rc != SCSI_RETURN_NOT_HANDLED)
			return rc;
		/* handler does not care. Drop down to default handling */
	}

483 484 485 486 487 488 489 490
	if (scmd->cmnd[0] == TEST_UNIT_READY && scmd->scsi_done != scsi_eh_done)
		/*
		 * nasty: for mid-layer issued TURs, we need to return the
		 * actual sense data without any recovery attempt.  For eh
		 * issued ones, we need to try to recover and interpret
		 */
		return SUCCESS;

L
Linus Torvalds 已提交
491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
	/*
	 * Previous logic looked for FILEMARK, EOM or ILI which are
	 * mainly associated with tapes and returned SUCCESS.
	 */
	if (sshdr.response_code == 0x70) {
		/* fixed format */
		if (scmd->sense_buffer[2] & 0xe0)
			return SUCCESS;
	} else {
		/*
		 * descriptor format: look for "stream commands sense data
		 * descriptor" (see SSC-3). Assume single sense data
		 * descriptor. Ignore ILI from SBC-2 READ LONG and WRITE LONG.
		 */
		if ((sshdr.additional_length > 3) &&
		    (scmd->sense_buffer[8] == 0x4) &&
		    (scmd->sense_buffer[11] & 0xe0))
			return SUCCESS;
	}

	switch (sshdr.sense_key) {
	case NO_SENSE:
		return SUCCESS;
	case RECOVERED_ERROR:
		return /* soft_error */ SUCCESS;

	case ABORTED_COMMAND:
518 519 520
		if (sshdr.asc == 0x10) /* DIF */
			return SUCCESS;

L
Linus Torvalds 已提交
521 522 523 524 525 526 527 528 529 530
		return NEEDS_RETRY;
	case NOT_READY:
	case UNIT_ATTENTION:
		/*
		 * if we are expecting a cc/ua because of a bus reset that we
		 * performed, treat this just as a retry.  otherwise this is
		 * information that we should pass up to the upper-level driver
		 * so that we can deal with it there.
		 */
		if (scmd->device->expecting_cc_ua) {
531 532 533 534 535 536 537 538 539 540
			/*
			 * Because some device does not queue unit
			 * attentions correctly, we carefully check
			 * additional sense code and qualifier so as
			 * not to squash media change unit attention.
			 */
			if (sshdr.asc != 0x28 || sshdr.ascq != 0x00) {
				scmd->device->expecting_cc_ua = 0;
				return NEEDS_RETRY;
			}
L
Linus Torvalds 已提交
541
		}
542 543 544 545 546 547 548 549
		/*
		 * we might also expect a cc/ua if another LUN on the target
		 * reported a UA with an ASC/ASCQ of 3F 0E -
		 * REPORTED LUNS DATA HAS CHANGED.
		 */
		if (scmd->device->sdev_target->expecting_lun_change &&
		    sshdr.asc == 0x3f && sshdr.ascq == 0x0e)
			return NEEDS_RETRY;
L
Linus Torvalds 已提交
550
		/*
551
		 * if the device is in the process of becoming ready, we
L
Linus Torvalds 已提交
552 553 554 555 556 557 558 559 560 561 562
		 * should retry.
		 */
		if ((sshdr.asc == 0x04) && (sshdr.ascq == 0x01))
			return NEEDS_RETRY;
		/*
		 * if the device is not started, we need to wake
		 * the error handler to start the motor
		 */
		if (scmd->device->allow_restart &&
		    (sshdr.asc == 0x04) && (sshdr.ascq == 0x02))
			return FAILED;
C
Christoph Hellwig 已提交
563 564 565 566 567
		/*
		 * Pass the UA upwards for a determination in the completion
		 * functions.
		 */
		return SUCCESS;
L
Linus Torvalds 已提交
568

569
		/* these are not supported */
570 571 572 573 574 575
	case DATA_PROTECT:
		if (sshdr.asc == 0x27 && sshdr.ascq == 0x07) {
			/* Thin provisioning hard threshold reached */
			set_host_byte(scmd, DID_ALLOC_FAILURE);
			return SUCCESS;
		}
576
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
577 578 579
	case COPY_ABORTED:
	case VOLUME_OVERFLOW:
	case MISCOMPARE:
580
	case BLANK_CHECK:
581 582
		set_host_byte(scmd, DID_TARGET_FAILURE);
		return SUCCESS;
L
Linus Torvalds 已提交
583 584

	case MEDIUM_ERROR:
585 586 587
		if (sshdr.asc == 0x11 || /* UNRECOVERED READ ERR */
		    sshdr.asc == 0x13 || /* AMNF DATA FIELD */
		    sshdr.asc == 0x14) { /* RECORD NOT FOUND */
588
			set_host_byte(scmd, DID_MEDIUM_ERROR);
589
			return SUCCESS;
590
		}
L
Linus Torvalds 已提交
591 592 593 594
		return NEEDS_RETRY;

	case HARDWARE_ERROR:
		if (scmd->device->retry_hwerror)
595
			return ADD_TO_MLQUEUE;
L
Linus Torvalds 已提交
596
		else
597
			set_host_byte(scmd, DID_TARGET_FAILURE);
598
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
599 600

	case ILLEGAL_REQUEST:
601 602
		if (sshdr.asc == 0x20 || /* Invalid command operation code */
		    sshdr.asc == 0x21 || /* Logical block address out of range */
603
		    sshdr.asc == 0x22 || /* Invalid function */
604
		    sshdr.asc == 0x24 || /* Invalid field in cdb */
605 606
		    sshdr.asc == 0x26 || /* Parameter value invalid */
		    sshdr.asc == 0x27) { /* Write protected */
607
			set_host_byte(scmd, DID_TARGET_FAILURE);
608 609 610
		}
		return SUCCESS;

L
Linus Torvalds 已提交
611 612 613 614
	default:
		return SUCCESS;
	}
}
615
EXPORT_SYMBOL_GPL(scsi_check_sense);
L
Linus Torvalds 已提交
616

V
Vasu Dev 已提交
617 618 619 620 621
static void scsi_handle_queue_ramp_up(struct scsi_device *sdev)
{
	struct scsi_host_template *sht = sdev->host->hostt;
	struct scsi_device *tmp_sdev;

622
	if (!sht->track_queue_depth ||
V
Vasu Dev 已提交
623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
	    sdev->queue_depth >= sdev->max_queue_depth)
		return;

	if (time_before(jiffies,
	    sdev->last_queue_ramp_up + sdev->queue_ramp_up_period))
		return;

	if (time_before(jiffies,
	    sdev->last_queue_full_time + sdev->queue_ramp_up_period))
		return;

	/*
	 * Walk all devices of a target and do
	 * ramp up on them.
	 */
	shost_for_each_device(tmp_sdev, sdev->host) {
		if (tmp_sdev->channel != sdev->channel ||
		    tmp_sdev->id != sdev->id ||
		    tmp_sdev->queue_depth == sdev->max_queue_depth)
			continue;
643

644
		scsi_change_queue_depth(tmp_sdev, tmp_sdev->queue_depth + 1);
V
Vasu Dev 已提交
645 646 647 648
		sdev->last_queue_ramp_up = jiffies;
	}
}

649 650 651 652 653
static void scsi_handle_queue_full(struct scsi_device *sdev)
{
	struct scsi_host_template *sht = sdev->host->hostt;
	struct scsi_device *tmp_sdev;

654
	if (!sht->track_queue_depth)
655 656 657 658 659 660 661 662 663 664 665
		return;

	shost_for_each_device(tmp_sdev, sdev->host) {
		if (tmp_sdev->channel != sdev->channel ||
		    tmp_sdev->id != sdev->id)
			continue;
		/*
		 * We do not know the number of commands that were at
		 * the device when we got the queue full so we start
		 * from the highest possible value and work our way down.
		 */
666
		scsi_track_queue_full(tmp_sdev, tmp_sdev->queue_depth - 1);
667 668 669
	}
}

L
Linus Torvalds 已提交
670 671 672 673 674 675 676 677 678
/**
 * scsi_eh_completed_normally - Disposition a eh cmd on return from LLD.
 * @scmd:	SCSI cmd to examine.
 *
 * Notes:
 *    This is *only* called when we are examining the status of commands
 *    queued during error recovery.  the main difference here is that we
 *    don't allow for the possibility of retries here, and we are a lot
 *    more restrictive about what we consider acceptable.
679
 */
L
Linus Torvalds 已提交
680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
static int scsi_eh_completed_normally(struct scsi_cmnd *scmd)
{
	/*
	 * first check the host byte, to see if there is anything in there
	 * that would indicate what we need to do.
	 */
	if (host_byte(scmd->result) == DID_RESET) {
		/*
		 * rats.  we are already in the error handler, so we now
		 * get to try and figure out what to do next.  if the sense
		 * is valid, we have a pretty good idea of what to do.
		 * if not, we mark it as FAILED.
		 */
		return scsi_check_sense(scmd);
	}
	if (host_byte(scmd->result) != DID_OK)
		return FAILED;

	/*
	 * next, check the message byte.
	 */
	if (msg_byte(scmd->result) != COMMAND_COMPLETE)
		return FAILED;

	/*
	 * now, check the status byte to see if this indicates
	 * anything special.
	 */
	switch (status_byte(scmd->result)) {
	case GOOD:
V
Vasu Dev 已提交
710
		scsi_handle_queue_ramp_up(scmd->device);
711
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
712 713 714 715 716 717 718 719 720 721 722
	case COMMAND_TERMINATED:
		return SUCCESS;
	case CHECK_CONDITION:
		return scsi_check_sense(scmd);
	case CONDITION_GOOD:
	case INTERMEDIATE_GOOD:
	case INTERMEDIATE_C_GOOD:
		/*
		 * who knows?  FIXME(eric)
		 */
		return SUCCESS;
723
	case RESERVATION_CONFLICT:
724 725 726 727 728 729
		if (scmd->cmnd[0] == TEST_UNIT_READY)
			/* it is a success, we probed the device and
			 * found it */
			return SUCCESS;
		/* otherwise, we failed to send the command */
		return FAILED;
L
Linus Torvalds 已提交
730
	case QUEUE_FULL:
731 732 733
		scsi_handle_queue_full(scmd->device);
		/* fall through */
	case BUSY:
734
		return NEEDS_RETRY;
L
Linus Torvalds 已提交
735 736 737 738 739 740 741 742 743
	default:
		return FAILED;
	}
	return FAILED;
}

/**
 * scsi_eh_done - Completion function for error handling.
 * @scmd:	Cmd that is done.
744
 */
L
Linus Torvalds 已提交
745 746
static void scsi_eh_done(struct scsi_cmnd *scmd)
{
747
	struct completion *eh_action;
748

749
	SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
750
			"%s result: %x\n", __func__, scmd->result));
751 752 753 754

	eh_action = scmd->device->host->eh_action;
	if (eh_action)
		complete(eh_action);
L
Linus Torvalds 已提交
755 756
}

757 758
/**
 * scsi_try_host_reset - ask host adapter to reset itself
G
Geert Uytterhoeven 已提交
759
 * @scmd:	SCSI cmd to send host reset.
760
 */
761 762 763 764
static int scsi_try_host_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;
765 766
	struct Scsi_Host *host = scmd->device->host;
	struct scsi_host_template *hostt = host->hostt;
767

768 769
	SCSI_LOG_ERROR_RECOVERY(3,
		shost_printk(KERN_INFO, host, "Snd Host RST\n"));
770

771
	if (!hostt->eh_host_reset_handler)
772 773
		return FAILED;

774
	rtn = hostt->eh_host_reset_handler(scmd);
775 776

	if (rtn == SUCCESS) {
777
		if (!hostt->skip_settle_delay)
778
			ssleep(HOST_RESET_SETTLE_TIME);
779 780 781
		spin_lock_irqsave(host->host_lock, flags);
		scsi_report_bus_reset(host, scmd_channel(scmd));
		spin_unlock_irqrestore(host->host_lock, flags);
782 783 784 785 786 787 788 789
	}

	return rtn;
}

/**
 * scsi_try_bus_reset - ask host to perform a bus reset
 * @scmd:	SCSI cmd to send bus reset.
790
 */
791 792 793 794
static int scsi_try_bus_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;
795 796
	struct Scsi_Host *host = scmd->device->host;
	struct scsi_host_template *hostt = host->hostt;
797

798 799
	SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
		"%s: Snd Bus RST\n", __func__));
800

801
	if (!hostt->eh_bus_reset_handler)
802 803
		return FAILED;

804
	rtn = hostt->eh_bus_reset_handler(scmd);
805 806

	if (rtn == SUCCESS) {
807
		if (!hostt->skip_settle_delay)
808
			ssleep(BUS_RESET_SETTLE_TIME);
809 810 811
		spin_lock_irqsave(host->host_lock, flags);
		scsi_report_bus_reset(host, scmd_channel(scmd));
		spin_unlock_irqrestore(host->host_lock, flags);
812 813 814 815 816
	}

	return rtn;
}

817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
static void __scsi_report_device_reset(struct scsi_device *sdev, void *data)
{
	sdev->was_reset = 1;
	sdev->expecting_cc_ua = 1;
}

/**
 * scsi_try_target_reset - Ask host to perform a target reset
 * @scmd:	SCSI cmd used to send a target reset
 *
 * Notes:
 *    There is no timeout for this operation.  if this operation is
 *    unreliable for a given host, then the host itself needs to put a
 *    timer on it, and set the host back to a consistent state prior to
 *    returning.
 */
static int scsi_try_target_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;
837 838
	struct Scsi_Host *host = scmd->device->host;
	struct scsi_host_template *hostt = host->hostt;
839

840
	if (!hostt->eh_target_reset_handler)
841 842
		return FAILED;

843
	rtn = hostt->eh_target_reset_handler(scmd);
844
	if (rtn == SUCCESS) {
845
		spin_lock_irqsave(host->host_lock, flags);
846 847
		__starget_for_each_device(scsi_target(scmd->device), NULL,
					  __scsi_report_device_reset);
848
		spin_unlock_irqrestore(host->host_lock, flags);
849 850 851 852 853
	}

	return rtn;
}

854 855 856 857 858 859 860 861 862
/**
 * scsi_try_bus_device_reset - Ask host to perform a BDR on a dev
 * @scmd:	SCSI cmd used to send BDR
 *
 * Notes:
 *    There is no timeout for this operation.  if this operation is
 *    unreliable for a given host, then the host itself needs to put a
 *    timer on it, and set the host back to a consistent state prior to
 *    returning.
863
 */
864 865 866
static int scsi_try_bus_device_reset(struct scsi_cmnd *scmd)
{
	int rtn;
867
	struct scsi_host_template *hostt = scmd->device->host->hostt;
868

869
	if (!hostt->eh_device_reset_handler)
870 871
		return FAILED;

872
	rtn = hostt->eh_device_reset_handler(scmd);
873 874
	if (rtn == SUCCESS)
		__scsi_report_device_reset(scmd->device, NULL);
875 876 877
	return rtn;
}

878 879
/**
 * scsi_try_to_abort_cmd - Ask host to abort a SCSI command
880
 * @hostt:	SCSI driver host template
881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896
 * @scmd:	SCSI cmd used to send a target reset
 *
 * Return value:
 *	SUCCESS, FAILED, or FAST_IO_FAIL
 *
 * Notes:
 *    SUCCESS does not necessarily indicate that the command
 *    has been aborted; it only indicates that the LLDDs
 *    has cleared all references to that command.
 *    LLDDs should return FAILED only if an abort was required
 *    but could not be executed. LLDDs should return FAST_IO_FAIL
 *    if the device is temporarily unavailable (eg due to a
 *    link down on FibreChannel)
 */
static int scsi_try_to_abort_cmd(struct scsi_host_template *hostt,
				 struct scsi_cmnd *scmd)
897
{
898
	if (!hostt->eh_abort_handler)
899 900
		return FAILED;

901
	return hostt->eh_abort_handler(scmd);
902 903 904 905
}

static void scsi_abort_eh_cmnd(struct scsi_cmnd *scmd)
{
906
	if (scsi_try_to_abort_cmd(scmd->device->host->hostt, scmd) != SUCCESS)
907
		if (scsi_try_bus_device_reset(scmd) != SUCCESS)
908 909 910
			if (scsi_try_target_reset(scmd) != SUCCESS)
				if (scsi_try_bus_reset(scmd) != SUCCESS)
					scsi_try_host_reset(scmd);
911 912
}

L
Linus Torvalds 已提交
913
/**
914
 * scsi_eh_prep_cmnd  - Save a scsi command info as part of error recovery
915
 * @scmd:       SCSI command structure to hijack
916
 * @ses:        structure to save restore information
917
 * @cmnd:       CDB to send. Can be NULL if no new cmnd is needed
918
 * @cmnd_size:  size in bytes of @cmnd (must be <= BLK_MAX_CDB)
919
 * @sense_bytes: size of sense data to copy. or 0 (if != 0 @cmnd is ignored)
920
 *
921
 * This function is used to save a scsi command information before re-execution
922 923 924 925
 * as part of the error recovery process.  If @sense_bytes is 0 the command
 * sent must be one that does not transfer any data.  If @sense_bytes != 0
 * @cmnd is ignored and this functions sets up a REQUEST_SENSE command
 * and cmnd buffers to read @sense_bytes into @scmd->sense_buffer.
926
 */
927 928
void scsi_eh_prep_cmnd(struct scsi_cmnd *scmd, struct scsi_eh_save *ses,
			unsigned char *cmnd, int cmnd_size, unsigned sense_bytes)
L
Linus Torvalds 已提交
929
{
已提交
930
	struct scsi_device *sdev = scmd->device;
L
Linus Torvalds 已提交
931

932 933 934 935 936 937 938
	/*
	 * We need saved copies of a number of fields - this is because
	 * error handling may need to overwrite these with different values
	 * to run different commands, and once error handling is complete,
	 * we will need to restore these values prior to running the actual
	 * command.
	 */
939
	ses->cmd_len = scmd->cmd_len;
940
	ses->cmnd = scmd->cmnd;
941
	ses->data_direction = scmd->sc_data_direction;
B
Boaz Harrosh 已提交
942
	ses->sdb = scmd->sdb;
943
	ses->next_rq = scmd->request->next_rq;
944
	ses->result = scmd->result;
945
	ses->underflow = scmd->underflow;
946
	ses->prot_op = scmd->prot_op;
947
	ses->eh_eflags = scmd->eh_eflags;
948

949
	scmd->prot_op = SCSI_PROT_NORMAL;
J
James Bottomley 已提交
950
	scmd->eh_eflags = 0;
951 952
	scmd->cmnd = ses->eh_cmnd;
	memset(scmd->cmnd, 0, BLK_MAX_CDB);
B
Boaz Harrosh 已提交
953
	memset(&scmd->sdb, 0, sizeof(scmd->sdb));
954
	scmd->request->next_rq = NULL;
955
	scmd->result = 0;
B
Boaz Harrosh 已提交
956

957
	if (sense_bytes) {
B
Boaz Harrosh 已提交
958 959
		scmd->sdb.length = min_t(unsigned, SCSI_SENSE_BUFFERSIZE,
					 sense_bytes);
960
		sg_init_one(&ses->sense_sgl, scmd->sense_buffer,
B
Boaz Harrosh 已提交
961 962
			    scmd->sdb.length);
		scmd->sdb.table.sgl = &ses->sense_sgl;
963
		scmd->sc_data_direction = DMA_FROM_DEVICE;
964
		scmd->sdb.table.nents = scmd->sdb.table.orig_nents = 1;
965
		scmd->cmnd[0] = REQUEST_SENSE;
B
Boaz Harrosh 已提交
966
		scmd->cmnd[4] = scmd->sdb.length;
967
		scmd->cmd_len = COMMAND_SIZE(scmd->cmnd[0]);
968 969
	} else {
		scmd->sc_data_direction = DMA_NONE;
970
		if (cmnd) {
971
			BUG_ON(cmnd_size > BLK_MAX_CDB);
972 973 974
			memcpy(scmd->cmnd, cmnd, cmnd_size);
			scmd->cmd_len = COMMAND_SIZE(scmd->cmnd[0]);
		}
975 976 977 978
	}

	scmd->underflow = 0;

979
	if (sdev->scsi_level <= SCSI_2 && sdev->scsi_level != SCSI_UNKNOWN)
L
Linus Torvalds 已提交
980
		scmd->cmnd[1] = (scmd->cmnd[1] & 0x1f) |
已提交
981
			(sdev->lun << 5 & 0xe0);
L
Linus Torvalds 已提交
982

983 984 985 986
	/*
	 * Zero the sense buffer.  The scsi spec mandates that any
	 * untransferred sense data should be interpreted as being zero.
	 */
987
	memset(scmd->sense_buffer, 0, SCSI_SENSE_BUFFERSIZE);
988 989 990 991
}
EXPORT_SYMBOL(scsi_eh_prep_cmnd);

/**
992
 * scsi_eh_restore_cmnd  - Restore a scsi command info as part of error recovery
993
 * @scmd:       SCSI command structure to restore
994
 * @ses:        saved information from a coresponding call to scsi_eh_prep_cmnd
995
 *
996
 * Undo any damage done by above scsi_eh_prep_cmnd().
997
 */
998 999 1000 1001 1002 1003
void scsi_eh_restore_cmnd(struct scsi_cmnd* scmd, struct scsi_eh_save *ses)
{
	/*
	 * Restore original data
	 */
	scmd->cmd_len = ses->cmd_len;
1004
	scmd->cmnd = ses->cmnd;
1005
	scmd->sc_data_direction = ses->data_direction;
B
Boaz Harrosh 已提交
1006
	scmd->sdb = ses->sdb;
1007
	scmd->request->next_rq = ses->next_rq;
1008
	scmd->result = ses->result;
1009
	scmd->underflow = ses->underflow;
1010
	scmd->prot_op = ses->prot_op;
1011
	scmd->eh_eflags = ses->eh_eflags;
1012 1013
}
EXPORT_SYMBOL(scsi_eh_restore_cmnd);
1014

1015
/**
1016
 * scsi_send_eh_cmnd  - submit a scsi command as part of error recovery
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027
 * @scmd:       SCSI command structure to hijack
 * @cmnd:       CDB to send
 * @cmnd_size:  size in bytes of @cmnd
 * @timeout:    timeout for this request
 * @sense_bytes: size of sense data to copy or 0
 *
 * This function is used to send a scsi command down to a target device
 * as part of the error recovery process. See also scsi_eh_prep_cmnd() above.
 *
 * Return value:
 *    SUCCESS or FAILED or NEEDS_RETRY
1028
 */
1029 1030 1031 1032 1033 1034
static int scsi_send_eh_cmnd(struct scsi_cmnd *scmd, unsigned char *cmnd,
			     int cmnd_size, int timeout, unsigned sense_bytes)
{
	struct scsi_device *sdev = scmd->device;
	struct Scsi_Host *shost = sdev->host;
	DECLARE_COMPLETION_ONSTACK(done);
1035
	unsigned long timeleft = timeout;
1036
	struct scsi_eh_save ses;
1037
	const unsigned long stall_for = msecs_to_jiffies(100);
1038 1039
	int rtn;

1040
retry:
1041
	scsi_eh_prep_cmnd(scmd, &ses, cmnd, cmnd_size, sense_bytes);
1042
	shost->eh_action = &done;
L
Linus Torvalds 已提交
1043 1044

	scsi_log_send(scmd);
J
Jeff Garzik 已提交
1045
	scmd->scsi_done = scsi_eh_done;
1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
	rtn = shost->hostt->queuecommand(shost, scmd);
	if (rtn) {
		if (timeleft > stall_for) {
			scsi_eh_restore_cmnd(scmd, &ses);
			timeleft -= stall_for;
			msleep(jiffies_to_msecs(stall_for));
			goto retry;
		}
		/* signal not to enter either branch of the if () below */
		timeleft = 0;
1056
		rtn = FAILED;
1057 1058
	} else {
		timeleft = wait_for_completion_timeout(&done, timeout);
1059
		rtn = SUCCESS;
1060
	}
L
Linus Torvalds 已提交
1061

已提交
1062
	shost->eh_action = NULL;
L
Linus Torvalds 已提交
1063

1064
	scsi_log_completion(scmd, rtn);
L
Linus Torvalds 已提交
1065

1066
	SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
1067 1068
			"%s timeleft: %ld\n",
			__func__, timeleft));
L
Linus Torvalds 已提交
1069 1070

	/*
1071 1072 1073 1074 1075 1076 1077
	 * If there is time left scsi_eh_done got called, and we will examine
	 * the actual status codes to see whether the command actually did
	 * complete normally, else if we have a zero return and no time left,
	 * the command must still be pending, so abort it and return FAILED.
	 * If we never actually managed to issue the command, because
	 * ->queuecommand() kept returning non zero, use the rtn = FAILED
	 * value above (so don't execute either branch of the if)
L
Linus Torvalds 已提交
1078
	 */
1079
	if (timeleft) {
L
Linus Torvalds 已提交
1080
		rtn = scsi_eh_completed_normally(scmd);
1081 1082
		SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
			"%s: scsi_eh_completed_normally %x\n", __func__, rtn));
1083

L
Linus Torvalds 已提交
1084 1085 1086 1087 1088
		switch (rtn) {
		case SUCCESS:
		case NEEDS_RETRY:
		case FAILED:
			break;
1089 1090 1091
		case ADD_TO_MLQUEUE:
			rtn = NEEDS_RETRY;
			break;
L
Linus Torvalds 已提交
1092 1093 1094 1095
		default:
			rtn = FAILED;
			break;
		}
1096
	} else if (rtn != FAILED) {
1097
		scsi_abort_eh_cmnd(scmd);
1098
		rtn = FAILED;
L
Linus Torvalds 已提交
1099 1100
	}

1101
	scsi_eh_restore_cmnd(scmd, &ses);
1102

L
Linus Torvalds 已提交
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
	return rtn;
}

/**
 * scsi_request_sense - Request sense data from a particular target.
 * @scmd:	SCSI cmd for request sense.
 *
 * Notes:
 *    Some hosts automatically obtain this information, others require
 *    that we obtain it on our own. This function will *not* return until
 *    the command either times out, or it completes.
1114
 */
L
Linus Torvalds 已提交
1115 1116
static int scsi_request_sense(struct scsi_cmnd *scmd)
{
1117
	return scsi_send_eh_cmnd(scmd, NULL, 0, scmd->device->eh_timeout, ~0);
L
Linus Torvalds 已提交
1118 1119
}

1120 1121
static int scsi_eh_action(struct scsi_cmnd *scmd, int rtn)
{
1122
	if (!blk_rq_is_passthrough(scmd->request)) {
1123 1124 1125 1126 1127 1128 1129
		struct scsi_driver *sdrv = scsi_cmd_to_driver(scmd);
		if (sdrv->eh_action)
			rtn = sdrv->eh_action(scmd, rtn);
	}
	return rtn;
}

L
Linus Torvalds 已提交
1130 1131 1132 1133 1134 1135 1136 1137
/**
 * scsi_eh_finish_cmd - Handle a cmd that eh is finished with.
 * @scmd:	Original SCSI cmd that eh has finished.
 * @done_q:	Queue for processed commands.
 *
 * Notes:
 *    We don't want to use the normal command completion while we are are
 *    still handling errors - it may cause other commands to be queued,
1138
 *    and that would disturb what we are doing.  Thus we really want to
L
Linus Torvalds 已提交
1139 1140
 *    keep a list of pending commands for final completion, and once we
 *    are ready to leave error handling we handle completion for real.
1141
 */
1142
void scsi_eh_finish_cmd(struct scsi_cmnd *scmd, struct list_head *done_q)
L
Linus Torvalds 已提交
1143 1144 1145
{
	list_move_tail(&scmd->eh_entry, done_q);
}
1146
EXPORT_SYMBOL(scsi_eh_finish_cmd);
L
Linus Torvalds 已提交
1147 1148 1149 1150

/**
 * scsi_eh_get_sense - Get device sense data.
 * @work_q:	Queue of commands to process.
1151
 * @done_q:	Queue of processed commands.
L
Linus Torvalds 已提交
1152 1153 1154
 *
 * Description:
 *    See if we need to request sense information.  if so, then get it
1155
 *    now, so we have a better idea of what to do.
L
Linus Torvalds 已提交
1156 1157 1158
 *
 * Notes:
 *    This has the unfortunate side effect that if a shost adapter does
1159
 *    not automatically request sense information, we end up shutting
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164 1165 1166
 *    it down before we request it.
 *
 *    All drivers should request sense information internally these days,
 *    so for now all I have to say is tough noogies if you end up in here.
 *
 *    XXX: Long term this code should go away, but that needs an audit of
 *         all LLDDs first.
1167
 */
1168 1169
int scsi_eh_get_sense(struct list_head *work_q,
		      struct list_head *done_q)
L
Linus Torvalds 已提交
1170
{
1171
	struct scsi_cmnd *scmd, *next;
1172
	struct Scsi_Host *shost;
L
Linus Torvalds 已提交
1173 1174
	int rtn;

1175 1176 1177 1178
	/*
	 * If SCSI_EH_ABORT_SCHEDULED has been set, it is timeout IO,
	 * should not get sense.
	 */
1179
	list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
1180
		if ((scmd->eh_eflags & SCSI_EH_ABORT_SCHEDULED) ||
L
Linus Torvalds 已提交
1181 1182 1183
		    SCSI_SENSE_VALID(scmd))
			continue;

1184 1185 1186
		shost = scmd->device->host;
		if (scsi_host_eh_past_deadline(shost)) {
			SCSI_LOG_ERROR_RECOVERY(3,
1187 1188 1189
				scmd_printk(KERN_INFO, scmd,
					    "%s: skip request sense, past eh deadline\n",
					     current->comm));
1190 1191
			break;
		}
1192 1193 1194 1195 1196 1197 1198 1199 1200
		if (status_byte(scmd->result) != CHECK_CONDITION)
			/*
			 * don't request sense if there's no check condition
			 * status because the error we're processing isn't one
			 * that has a sense code (and some devices get
			 * confused by sense requests out of the blue)
			 */
			continue;

1201 1202 1203
		SCSI_LOG_ERROR_RECOVERY(2, scmd_printk(KERN_INFO, scmd,
						  "%s: requesting sense\n",
						  current->comm));
L
Linus Torvalds 已提交
1204 1205 1206 1207
		rtn = scsi_request_sense(scmd);
		if (rtn != SUCCESS)
			continue;

1208
		SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
1209
			"sense requested, result %x\n", scmd->result));
1210
		SCSI_LOG_ERROR_RECOVERY(3, scsi_print_sense(scmd));
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231

		rtn = scsi_decide_disposition(scmd);

		/*
		 * if the result was normal, then just pass it along to the
		 * upper level.
		 */
		if (rtn == SUCCESS)
			/* we don't want this command reissued, just
			 * finished with the sense data, so set
			 * retries to the max allowed to ensure it
			 * won't get reissued */
			scmd->retries = scmd->allowed;
		else if (rtn != NEEDS_RETRY)
			continue;

		scsi_eh_finish_cmd(scmd, done_q);
	}

	return list_empty(work_q);
}
1232
EXPORT_SYMBOL_GPL(scsi_eh_get_sense);
L
Linus Torvalds 已提交
1233 1234 1235

/**
 * scsi_eh_tur - Send TUR to device.
1236
 * @scmd:	&scsi_cmnd to send TUR
L
Linus Torvalds 已提交
1237 1238 1239
 *
 * Return value:
 *    0 - Device is ready. 1 - Device NOT ready.
1240
 */
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246
static int scsi_eh_tur(struct scsi_cmnd *scmd)
{
	static unsigned char tur_command[6] = {TEST_UNIT_READY, 0, 0, 0, 0, 0};
	int retry_cnt = 1, rtn;

retry_tur:
1247 1248
	rtn = scsi_send_eh_cmnd(scmd, tur_command, 6,
				scmd->device->eh_timeout, 0);
L
Linus Torvalds 已提交
1249

1250
	SCSI_LOG_ERROR_RECOVERY(3, scmd_printk(KERN_INFO, scmd,
1251
		"%s return: %x\n", __func__, rtn));
1252 1253 1254

	switch (rtn) {
	case NEEDS_RETRY:
L
Linus Torvalds 已提交
1255 1256
		if (retry_cnt--)
			goto retry_tur;
1257 1258
		/*FALLTHRU*/
	case SUCCESS:
1259
		return 0;
1260 1261
	default:
		return 1;
1262
	}
L
Linus Torvalds 已提交
1263 1264
}

1265 1266 1267
/**
 * scsi_eh_test_devices - check if devices are responding from error recovery.
 * @cmd_list:	scsi commands in error recovery.
1268 1269 1270
 * @work_q:	queue for commands which still need more error recovery
 * @done_q:	queue for commands which are finished
 * @try_stu:	boolean on if a STU command should be tried in addition to TUR.
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
 *
 * Decription:
 *    Tests if devices are in a working state.  Commands to devices now in
 *    a working state are sent to the done_q while commands to devices which
 *    are still failing to respond are returned to the work_q for more
 *    processing.
 **/
static int scsi_eh_test_devices(struct list_head *cmd_list,
				struct list_head *work_q,
				struct list_head *done_q, int try_stu)
{
	struct scsi_cmnd *scmd, *next;
	struct scsi_device *sdev;
	int finish_cmds;

	while (!list_empty(cmd_list)) {
		scmd = list_entry(cmd_list->next, struct scsi_cmnd, eh_entry);
		sdev = scmd->device;

1290 1291 1292 1293 1294
		if (!try_stu) {
			if (scsi_host_eh_past_deadline(sdev->host)) {
				/* Push items back onto work_q */
				list_splice_init(cmd_list, work_q);
				SCSI_LOG_ERROR_RECOVERY(3,
1295 1296 1297
					sdev_printk(KERN_INFO, sdev,
						    "%s: skip test device, past eh deadline",
						    current->comm));
1298 1299 1300 1301
				break;
			}
		}

1302 1303 1304 1305 1306 1307 1308
		finish_cmds = !scsi_device_online(scmd->device) ||
			(try_stu && !scsi_eh_try_stu(scmd) &&
			 !scsi_eh_tur(scmd)) ||
			!scsi_eh_tur(scmd);

		list_for_each_entry_safe(scmd, next, cmd_list, eh_entry)
			if (scmd->device == sdev) {
1309 1310 1311
				if (finish_cmds &&
				    (try_stu ||
				     scsi_eh_action(scmd, SUCCESS) == SUCCESS))
1312 1313 1314 1315 1316 1317 1318 1319
					scsi_eh_finish_cmd(scmd, done_q);
				else
					list_move_tail(&scmd->eh_entry, work_q);
			}
	}
	return list_empty(work_q);
}

L
Linus Torvalds 已提交
1320 1321
/**
 * scsi_eh_try_stu - Send START_UNIT to device.
1322
 * @scmd:	&scsi_cmnd to send START_UNIT
L
Linus Torvalds 已提交
1323 1324 1325
 *
 * Return value:
 *    0 - Device is ready. 1 - Device NOT ready.
1326
 */
L
Linus Torvalds 已提交
1327 1328 1329 1330
static int scsi_eh_try_stu(struct scsi_cmnd *scmd)
{
	static unsigned char stu_command[6] = {START_STOP, 0, 0, 0, 1, 0};

1331
	if (scmd->device->allow_restart) {
1332 1333 1334
		int i, rtn = NEEDS_RETRY;

		for (i = 0; rtn == NEEDS_RETRY && i < 2; i++)
1335
			rtn = scsi_send_eh_cmnd(scmd, stu_command, 6, scmd->device->request_queue->rq_timeout, 0);
L
Linus Torvalds 已提交
1336

1337 1338 1339
		if (rtn == SUCCESS)
			return 0;
	}
L
Linus Torvalds 已提交
1340 1341 1342 1343 1344 1345

	return 1;
}

 /**
 * scsi_eh_stu - send START_UNIT if needed
1346
 * @shost:	&scsi host being recovered.
1347
 * @work_q:	&list_head for pending commands.
1348
 * @done_q:	&list_head for processed commands.
L
Linus Torvalds 已提交
1349 1350 1351
 *
 * Notes:
 *    If commands are failing due to not ready, initializing command required,
1352
 *	try revalidating the device, which will end up sending a start unit.
1353
 */
L
Linus Torvalds 已提交
1354 1355 1356 1357
static int scsi_eh_stu(struct Scsi_Host *shost,
			      struct list_head *work_q,
			      struct list_head *done_q)
{
1358
	struct scsi_cmnd *scmd, *stu_scmd, *next;
L
Linus Torvalds 已提交
1359 1360 1361
	struct scsi_device *sdev;

	shost_for_each_device(sdev, shost) {
1362 1363
		if (scsi_host_eh_past_deadline(shost)) {
			SCSI_LOG_ERROR_RECOVERY(3,
1364 1365 1366
				sdev_printk(KERN_INFO, sdev,
					    "%s: skip START_UNIT, past eh deadline\n",
					    current->comm));
1367 1368
			break;
		}
L
Linus Torvalds 已提交
1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379
		stu_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry)
			if (scmd->device == sdev && SCSI_SENSE_VALID(scmd) &&
			    scsi_check_sense(scmd) == FAILED ) {
				stu_scmd = scmd;
				break;
			}

		if (!stu_scmd)
			continue;

1380
		SCSI_LOG_ERROR_RECOVERY(3,
1381 1382 1383
			sdev_printk(KERN_INFO, sdev,
				     "%s: Sending START_UNIT\n",
				    current->comm));
L
Linus Torvalds 已提交
1384 1385 1386 1387

		if (!scsi_eh_try_stu(stu_scmd)) {
			if (!scsi_device_online(sdev) ||
			    !scsi_eh_tur(stu_scmd)) {
1388 1389
				list_for_each_entry_safe(scmd, next,
							  work_q, eh_entry) {
1390 1391
					if (scmd->device == sdev &&
					    scsi_eh_action(scmd, SUCCESS) == SUCCESS)
L
Linus Torvalds 已提交
1392 1393 1394 1395 1396
						scsi_eh_finish_cmd(scmd, done_q);
				}
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3,
1397 1398 1399
				sdev_printk(KERN_INFO, sdev,
					    "%s: START_UNIT failed\n",
					    current->comm));
L
Linus Torvalds 已提交
1400 1401 1402 1403 1404 1405 1406 1407 1408 1409
		}
	}

	return list_empty(work_q);
}


/**
 * scsi_eh_bus_device_reset - send bdr if needed
 * @shost:	scsi host being recovered.
1410
 * @work_q:	&list_head for pending commands.
1411
 * @done_q:	&list_head for processed commands.
L
Linus Torvalds 已提交
1412 1413
 *
 * Notes:
1414
 *    Try a bus device reset.  Still, look to see whether we have multiple
L
Linus Torvalds 已提交
1415 1416
 *    devices that are jammed or not - if we have multiple devices, it
 *    makes no sense to try bus_device_reset - we really would need to try
1417
 *    a bus_reset instead.
1418
 */
L
Linus Torvalds 已提交
1419 1420 1421 1422
static int scsi_eh_bus_device_reset(struct Scsi_Host *shost,
				    struct list_head *work_q,
				    struct list_head *done_q)
{
1423
	struct scsi_cmnd *scmd, *bdr_scmd, *next;
L
Linus Torvalds 已提交
1424 1425 1426 1427
	struct scsi_device *sdev;
	int rtn;

	shost_for_each_device(sdev, shost) {
1428 1429
		if (scsi_host_eh_past_deadline(shost)) {
			SCSI_LOG_ERROR_RECOVERY(3,
1430 1431 1432
				sdev_printk(KERN_INFO, sdev,
					    "%s: skip BDR, past eh deadline\n",
					     current->comm));
1433 1434
			break;
		}
L
Linus Torvalds 已提交
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
		bdr_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry)
			if (scmd->device == sdev) {
				bdr_scmd = scmd;
				break;
			}

		if (!bdr_scmd)
			continue;

1445
		SCSI_LOG_ERROR_RECOVERY(3,
1446 1447
			sdev_printk(KERN_INFO, sdev,
				     "%s: Sending BDR\n", current->comm));
L
Linus Torvalds 已提交
1448
		rtn = scsi_try_bus_device_reset(bdr_scmd);
1449
		if (rtn == SUCCESS || rtn == FAST_IO_FAIL) {
L
Linus Torvalds 已提交
1450
			if (!scsi_device_online(sdev) ||
1451
			    rtn == FAST_IO_FAIL ||
L
Linus Torvalds 已提交
1452
			    !scsi_eh_tur(bdr_scmd)) {
1453 1454
				list_for_each_entry_safe(scmd, next,
							 work_q, eh_entry) {
1455 1456
					if (scmd->device == sdev &&
					    scsi_eh_action(scmd, rtn) != FAILED)
L
Linus Torvalds 已提交
1457 1458 1459 1460 1461
						scsi_eh_finish_cmd(scmd,
								   done_q);
				}
			}
		} else {
1462
			SCSI_LOG_ERROR_RECOVERY(3,
1463 1464
				sdev_printk(KERN_INFO, sdev,
					    "%s: BDR failed\n", current->comm));
L
Linus Torvalds 已提交
1465 1466 1467 1468 1469 1470
		}
	}

	return list_empty(work_q);
}

1471 1472 1473
/**
 * scsi_eh_target_reset - send target reset if needed
 * @shost:	scsi host being recovered.
1474
 * @work_q:	&list_head for pending commands.
1475 1476 1477 1478 1479 1480 1481 1482 1483
 * @done_q:	&list_head for processed commands.
 *
 * Notes:
 *    Try a target reset.
 */
static int scsi_eh_target_reset(struct Scsi_Host *shost,
				struct list_head *work_q,
				struct list_head *done_q)
{
1484
	LIST_HEAD(tmp_list);
1485
	LIST_HEAD(check_list);
1486

1487 1488 1489 1490 1491 1492
	list_splice_init(work_q, &tmp_list);

	while (!list_empty(&tmp_list)) {
		struct scsi_cmnd *next, *scmd;
		int rtn;
		unsigned int id;
1493 1494 1495 1496 1497 1498 1499

		if (scsi_host_eh_past_deadline(shost)) {
			/* push back on work queue for further processing */
			list_splice_init(&check_list, work_q);
			list_splice_init(&tmp_list, work_q);
			SCSI_LOG_ERROR_RECOVERY(3,
				shost_printk(KERN_INFO, shost,
1500 1501
					    "%s: Skip target reset, past eh deadline\n",
					     current->comm));
1502 1503
			return list_empty(work_q);
		}
1504 1505 1506

		scmd = list_entry(tmp_list.next, struct scsi_cmnd, eh_entry);
		id = scmd_id(scmd);
1507

1508 1509 1510 1511
		SCSI_LOG_ERROR_RECOVERY(3,
			shost_printk(KERN_INFO, shost,
				     "%s: Sending target reset to target %d\n",
				     current->comm, id));
1512 1513
		rtn = scsi_try_target_reset(scmd);
		if (rtn != SUCCESS && rtn != FAST_IO_FAIL)
1514 1515 1516 1517 1518
			SCSI_LOG_ERROR_RECOVERY(3,
				shost_printk(KERN_INFO, shost,
					     "%s: Target reset failed"
					     " target: %d\n",
					     current->comm, id));
1519 1520 1521 1522
		list_for_each_entry_safe(scmd, next, &tmp_list, eh_entry) {
			if (scmd_id(scmd) != id)
				continue;

1523 1524 1525
			if (rtn == SUCCESS)
				list_move_tail(&scmd->eh_entry, &check_list);
			else if (rtn == FAST_IO_FAIL)
1526 1527 1528 1529 1530 1531
				scsi_eh_finish_cmd(scmd, done_q);
			else
				/* push back on work queue for further processing */
				list_move(&scmd->eh_entry, work_q);
		}
	}
1532

1533
	return scsi_eh_test_devices(&check_list, work_q, done_q, 0);
1534 1535
}

L
Linus Torvalds 已提交
1536
/**
1537
 * scsi_eh_bus_reset - send a bus reset
1538
 * @shost:	&scsi host being recovered.
1539
 * @work_q:	&list_head for pending commands.
1540
 * @done_q:	&list_head for processed commands.
1541
 */
L
Linus Torvalds 已提交
1542 1543 1544 1545
static int scsi_eh_bus_reset(struct Scsi_Host *shost,
			     struct list_head *work_q,
			     struct list_head *done_q)
{
1546
	struct scsi_cmnd *scmd, *chan_scmd, *next;
1547
	LIST_HEAD(check_list);
L
Linus Torvalds 已提交
1548 1549 1550 1551 1552 1553 1554
	unsigned int channel;
	int rtn;

	/*
	 * we really want to loop over the various channels, and do this on
	 * a channel by channel basis.  we should also check to see if any
	 * of the failed commands are on soft_reset devices, and if so, skip
1555
	 * the reset.
L
Linus Torvalds 已提交
1556 1557 1558
	 */

	for (channel = 0; channel <= shost->max_channel; channel++) {
1559 1560 1561 1562
		if (scsi_host_eh_past_deadline(shost)) {
			list_splice_init(&check_list, work_q);
			SCSI_LOG_ERROR_RECOVERY(3,
				shost_printk(KERN_INFO, shost,
1563 1564
					    "%s: skip BRST, past eh deadline\n",
					     current->comm));
1565 1566 1567
			return list_empty(work_q);
		}

L
Linus Torvalds 已提交
1568 1569
		chan_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry) {
1570
			if (channel == scmd_channel(scmd)) {
L
Linus Torvalds 已提交
1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581
				chan_scmd = scmd;
				break;
				/*
				 * FIXME add back in some support for
				 * soft_reset devices.
				 */
			}
		}

		if (!chan_scmd)
			continue;
1582 1583 1584 1585
		SCSI_LOG_ERROR_RECOVERY(3,
			shost_printk(KERN_INFO, shost,
				     "%s: Sending BRST chan: %d\n",
				     current->comm, channel));
L
Linus Torvalds 已提交
1586
		rtn = scsi_try_bus_reset(chan_scmd);
1587
		if (rtn == SUCCESS || rtn == FAST_IO_FAIL) {
1588
			list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
1589 1590
				if (channel == scmd_channel(scmd)) {
					if (rtn == FAST_IO_FAIL)
L
Linus Torvalds 已提交
1591 1592
						scsi_eh_finish_cmd(scmd,
								   done_q);
1593 1594 1595 1596
					else
						list_move_tail(&scmd->eh_entry,
							       &check_list);
				}
L
Linus Torvalds 已提交
1597 1598
			}
		} else {
1599 1600 1601 1602
			SCSI_LOG_ERROR_RECOVERY(3,
				shost_printk(KERN_INFO, shost,
					     "%s: BRST failed chan: %d\n",
					     current->comm, channel));
L
Linus Torvalds 已提交
1603 1604
		}
	}
1605
	return scsi_eh_test_devices(&check_list, work_q, done_q, 0);
L
Linus Torvalds 已提交
1606 1607 1608
}

/**
1609
 * scsi_eh_host_reset - send a host reset
1610 1611 1612
 * @shost:	host to be reset.
 * @work_q:	&list_head for pending commands.
 * @done_q:	&list_head for processed commands.
1613
 */
1614 1615
static int scsi_eh_host_reset(struct Scsi_Host *shost,
			      struct list_head *work_q,
L
Linus Torvalds 已提交
1616 1617
			      struct list_head *done_q)
{
1618
	struct scsi_cmnd *scmd, *next;
1619
	LIST_HEAD(check_list);
L
Linus Torvalds 已提交
1620 1621 1622 1623 1624 1625
	int rtn;

	if (!list_empty(work_q)) {
		scmd = list_entry(work_q->next,
				  struct scsi_cmnd, eh_entry);

1626 1627 1628 1629
		SCSI_LOG_ERROR_RECOVERY(3,
			shost_printk(KERN_INFO, shost,
				     "%s: Sending HRST\n",
				     current->comm));
L
Linus Torvalds 已提交
1630 1631

		rtn = scsi_try_host_reset(scmd);
1632 1633 1634
		if (rtn == SUCCESS) {
			list_splice_init(work_q, &check_list);
		} else if (rtn == FAST_IO_FAIL) {
1635
			list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
L
Linus Torvalds 已提交
1636 1637 1638
					scsi_eh_finish_cmd(scmd, done_q);
			}
		} else {
1639 1640 1641 1642
			SCSI_LOG_ERROR_RECOVERY(3,
				shost_printk(KERN_INFO, shost,
					     "%s: HRST failed\n",
					     current->comm));
L
Linus Torvalds 已提交
1643 1644
		}
	}
1645
	return scsi_eh_test_devices(&check_list, work_q, done_q, 1);
L
Linus Torvalds 已提交
1646 1647 1648 1649
}

/**
 * scsi_eh_offline_sdevs - offline scsi devices that fail to recover
1650 1651
 * @work_q:	&list_head for pending commands.
 * @done_q:	&list_head for processed commands.
1652
 */
L
Linus Torvalds 已提交
1653 1654 1655
static void scsi_eh_offline_sdevs(struct list_head *work_q,
				  struct list_head *done_q)
{
1656
	struct scsi_cmnd *scmd, *next;
1657
	struct scsi_device *sdev;
L
Linus Torvalds 已提交
1658

1659
	list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
1660 1661
		sdev_printk(KERN_INFO, scmd->device, "Device offlined - "
			    "not ready after error recovery\n");
1662 1663 1664 1665 1666 1667
		sdev = scmd->device;

		mutex_lock(&sdev->state_mutex);
		scsi_device_set_state(sdev, SDEV_OFFLINE);
		mutex_unlock(&sdev->state_mutex);

L
Linus Torvalds 已提交
1668 1669 1670 1671 1672
		scsi_eh_finish_cmd(scmd, done_q);
	}
	return;
}

1673
/**
1674
 * scsi_noretry_cmd - determine if command should be failed fast
1675 1676 1677 1678 1679 1680 1681
 * @scmd:	SCSI cmd to examine.
 */
int scsi_noretry_cmd(struct scsi_cmnd *scmd)
{
	switch (host_byte(scmd->result)) {
	case DID_OK:
		break;
1682 1683
	case DID_TIME_OUT:
		goto check_type;
1684
	case DID_BUS_BUSY:
1685
		return (scmd->request->cmd_flags & REQ_FAILFAST_TRANSPORT);
1686
	case DID_PARITY:
1687
		return (scmd->request->cmd_flags & REQ_FAILFAST_DEV);
1688 1689 1690 1691 1692 1693
	case DID_ERROR:
		if (msg_byte(scmd->result) == COMMAND_COMPLETE &&
		    status_byte(scmd->result) == RESERVATION_CONFLICT)
			return 0;
		/* fall through */
	case DID_SOFT_ERROR:
1694
		return (scmd->request->cmd_flags & REQ_FAILFAST_DRIVER);
1695 1696
	}

1697 1698
	if (status_byte(scmd->result) != CHECK_CONDITION)
		return 0;
1699

1700 1701 1702 1703 1704 1705
check_type:
	/*
	 * assume caller has checked sense and determined
	 * the check condition was retryable.
	 */
	if (scmd->request->cmd_flags & REQ_FAILFAST_DEV ||
1706
	    blk_rq_is_passthrough(scmd->request))
1707 1708 1709
		return 1;
	else
		return 0;
1710 1711
}

L
Linus Torvalds 已提交
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724
/**
 * scsi_decide_disposition - Disposition a cmd on return from LLD.
 * @scmd:	SCSI cmd to examine.
 *
 * Notes:
 *    This is *only* called when we are examining the status after sending
 *    out the actual data command.  any commands that are queued for error
 *    recovery (e.g. test_unit_ready) do *not* come through here.
 *
 *    When this routine returns failed, it means the error handler thread
 *    is woken.  In cases where the error code indicates an error that
 *    doesn't require the error handler read (i.e. we don't need to
 *    abort/reset), this function should return SUCCESS.
1725
 */
L
Linus Torvalds 已提交
1726 1727 1728 1729 1730 1731 1732 1733 1734
int scsi_decide_disposition(struct scsi_cmnd *scmd)
{
	int rtn;

	/*
	 * if the device is offline, then we clearly just pass the result back
	 * up to the top level.
	 */
	if (!scsi_device_online(scmd->device)) {
1735 1736
		SCSI_LOG_ERROR_RECOVERY(5, scmd_printk(KERN_INFO, scmd,
			"%s: device offline - report as SUCCESS\n", __func__));
L
Linus Torvalds 已提交
1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
		return SUCCESS;
	}

	/*
	 * first check the host byte, to see if there is anything in there
	 * that would indicate what we need to do.
	 */
	switch (host_byte(scmd->result)) {
	case DID_PASSTHROUGH:
		/*
		 * no matter what, pass this through to the upper layer.
		 * nuke this special code so that it looks like we are saying
		 * did_ok.
		 */
		scmd->result &= 0xff00ffff;
		return SUCCESS;
	case DID_OK:
		/*
		 * looks good.  drop through, and check the next byte.
		 */
		break;
1758 1759
	case DID_ABORT:
		if (scmd->eh_eflags & SCSI_EH_ABORT_SCHEDULED) {
1760
			set_host_byte(scmd, DID_TIME_OUT);
1761 1762
			return SUCCESS;
		}
1763
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
1764 1765 1766 1767 1768 1769 1770 1771
	case DID_NO_CONNECT:
	case DID_BAD_TARGET:
		/*
		 * note - this means that we just report the status back
		 * to the top level driver, not that we actually think
		 * that it indicates SUCCESS.
		 */
		return SUCCESS;
1772
	case DID_SOFT_ERROR:
L
Linus Torvalds 已提交
1773 1774
		/*
		 * when the low level driver returns did_soft_error,
1775
		 * it is responsible for keeping an internal retry counter
L
Linus Torvalds 已提交
1776 1777 1778 1779 1780 1781
		 * in order to avoid endless loops (db)
		 */
		goto maybe_retry;
	case DID_IMM_RETRY:
		return NEEDS_RETRY;

已提交
1782 1783
	case DID_REQUEUE:
		return ADD_TO_MLQUEUE;
1784 1785 1786 1787 1788
	case DID_TRANSPORT_DISRUPTED:
		/*
		 * LLD/transport was disrupted during processing of the IO.
		 * The transport class is now blocked/blocking,
		 * and the transport will decide what to do with the IO
1789 1790
		 * based on its timers and recovery capablilities if
		 * there are enough retries.
1791
		 */
1792
		goto maybe_retry;
1793 1794 1795 1796 1797 1798
	case DID_TRANSPORT_FAILFAST:
		/*
		 * The transport decided to failfast the IO (most likely
		 * the fast io fail tmo fired), so send IO directly upwards.
		 */
		return SUCCESS;
L
Linus Torvalds 已提交
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
	case DID_ERROR:
		if (msg_byte(scmd->result) == COMMAND_COMPLETE &&
		    status_byte(scmd->result) == RESERVATION_CONFLICT)
			/*
			 * execute reservation conflict processing code
			 * lower down
			 */
			break;
		/* fallthrough */
	case DID_BUS_BUSY:
	case DID_PARITY:
		goto maybe_retry;
	case DID_TIME_OUT:
		/*
		 * when we scan the bus, we get timeout messages for
		 * these commands if there is no device available.
		 * other hosts report did_no_connect for the same thing.
		 */
		if ((scmd->cmnd[0] == TEST_UNIT_READY ||
		     scmd->cmnd[0] == INQUIRY)) {
			return SUCCESS;
		} else {
			return FAILED;
		}
	case DID_RESET:
		return SUCCESS;
	default:
		return FAILED;
	}

	/*
	 * next, check the message byte.
	 */
	if (msg_byte(scmd->result) != COMMAND_COMPLETE)
		return FAILED;

	/*
	 * check the status byte to see if this indicates anything special.
	 */
	switch (status_byte(scmd->result)) {
	case QUEUE_FULL:
1840
		scsi_handle_queue_full(scmd->device);
L
Linus Torvalds 已提交
1841 1842 1843 1844
		/*
		 * the case of trying to send too many commands to a
		 * tagged queueing device.
		 */
1845
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
1846 1847 1848 1849 1850 1851 1852 1853 1854
	case BUSY:
		/*
		 * device can't talk to us at the moment.  Should only
		 * occur (SAM-3) when the task queue is empty, so will cause
		 * the empty queue handling to trigger a stall in the
		 * device.
		 */
		return ADD_TO_MLQUEUE;
	case GOOD:
1855 1856
		if (scmd->cmnd[0] == REPORT_LUNS)
			scmd->device->sdev_target->expecting_lun_change = 0;
V
Vasu Dev 已提交
1857
		scsi_handle_queue_ramp_up(scmd->device);
1858
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
1859 1860
	case COMMAND_TERMINATED:
		return SUCCESS;
1861 1862
	case TASK_ABORTED:
		goto maybe_retry;
L
Linus Torvalds 已提交
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881
	case CHECK_CONDITION:
		rtn = scsi_check_sense(scmd);
		if (rtn == NEEDS_RETRY)
			goto maybe_retry;
		/* if rtn == FAILED, we have no sense information;
		 * returning FAILED will wake the error handler thread
		 * to collect the sense and redo the decide
		 * disposition */
		return rtn;
	case CONDITION_GOOD:
	case INTERMEDIATE_GOOD:
	case INTERMEDIATE_C_GOOD:
	case ACA_ACTIVE:
		/*
		 * who knows?  FIXME(eric)
		 */
		return SUCCESS;

	case RESERVATION_CONFLICT:
1882 1883
		sdev_printk(KERN_INFO, scmd->device,
			    "reservation conflict\n");
1884
		set_host_byte(scmd, DID_NEXUS_FAILURE);
L
Linus Torvalds 已提交
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896
		return SUCCESS; /* causes immediate i/o error */
	default:
		return FAILED;
	}
	return FAILED;

      maybe_retry:

	/* we requeue for retry because the error was retryable, and
	 * the request was not marked fast fail.  Note that above,
	 * even if the request is marked fast fail, we still requeue
	 * for queue congestion conditions (QUEUE_FULL or BUSY) */
1897
	if ((++scmd->retries) <= scmd->allowed
1898
	    && !scsi_noretry_cmd(scmd)) {
L
Linus Torvalds 已提交
1899 1900 1901 1902 1903 1904 1905 1906 1907
		return NEEDS_RETRY;
	} else {
		/*
		 * no more retries - report this one back to upper level.
		 */
		return SUCCESS;
	}
}

1908
static void eh_lock_door_done(struct request *req, blk_status_t status)
F
FUJITA Tomonori 已提交
1909 1910 1911 1912
{
	__blk_put_request(req->q, req);
}

L
Linus Torvalds 已提交
1913 1914 1915 1916 1917
/**
 * scsi_eh_lock_door - Prevent medium removal for the specified device
 * @sdev:	SCSI device to prevent medium removal
 *
 * Locking:
1918
 * 	We must be called from process context.
L
Linus Torvalds 已提交
1919 1920 1921 1922
 *
 * Notes:
 * 	We queue up an asynchronous "ALLOW MEDIUM REMOVAL" request on the
 * 	head of the devices request queue, and continue.
1923
 */
L
Linus Torvalds 已提交
1924 1925
static void scsi_eh_lock_door(struct scsi_device *sdev)
{
F
FUJITA Tomonori 已提交
1926
	struct request *req;
1927
	struct scsi_request *rq;
L
Linus Torvalds 已提交
1928

1929
	/*
1930
	 * blk_get_request with GFP_KERNEL (__GFP_RECLAIM) sleeps until a
1931 1932
	 * request becomes available
	 */
1933
	req = blk_get_request(sdev->request_queue, REQ_OP_SCSI_IN, GFP_KERNEL);
1934
	if (IS_ERR(req))
1935
		return;
1936
	rq = scsi_req(req);
L
Linus Torvalds 已提交
1937

1938 1939 1940 1941 1942 1943 1944
	rq->cmd[0] = ALLOW_MEDIUM_REMOVAL;
	rq->cmd[1] = 0;
	rq->cmd[2] = 0;
	rq->cmd[3] = 0;
	rq->cmd[4] = SCSI_REMOVAL_PREVENT;
	rq->cmd[5] = 0;
	rq->cmd_len = COMMAND_SIZE(rq->cmd[0]);
F
FUJITA Tomonori 已提交
1945

1946
	req->rq_flags |= RQF_QUIET;
F
FUJITA Tomonori 已提交
1947
	req->timeout = 10 * HZ;
1948
	rq->retries = 5;
F
FUJITA Tomonori 已提交
1949 1950 1951

	blk_execute_rq_nowait(req->q, NULL, req, 1, eh_lock_door_done);
}
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956 1957 1958 1959

/**
 * scsi_restart_operations - restart io operations to the specified host.
 * @shost:	Host we are restarting.
 *
 * Notes:
 *    When we entered the error handler, we blocked all further i/o to
 *    this device.  we need to 'reverse' this process.
1960
 */
L
Linus Torvalds 已提交
1961 1962 1963
static void scsi_restart_operations(struct Scsi_Host *shost)
{
	struct scsi_device *sdev;
1964
	unsigned long flags;
L
Linus Torvalds 已提交
1965 1966 1967 1968 1969 1970 1971

	/*
	 * If the door was locked, we need to insert a door lock request
	 * onto the head of the SCSI request queue for the device.  There
	 * is no point trying to lock the door of an off-line device.
	 */
	shost_for_each_device(sdev, shost) {
1972
		if (scsi_device_online(sdev) && sdev->was_reset && sdev->locked) {
L
Linus Torvalds 已提交
1973
			scsi_eh_lock_door(sdev);
1974 1975
			sdev->was_reset = 0;
		}
L
Linus Torvalds 已提交
1976 1977 1978 1979 1980 1981 1982
	}

	/*
	 * next free up anything directly waiting upon the host.  this
	 * will be requests for character device operations, and also for
	 * ioctls to queued block devices.
	 */
1983
	SCSI_LOG_ERROR_RECOVERY(3,
1984
		shost_printk(KERN_INFO, shost, "waking up host to restart\n"));
L
Linus Torvalds 已提交
1985

1986 1987 1988 1989 1990
	spin_lock_irqsave(shost->host_lock, flags);
	if (scsi_host_set_state(shost, SHOST_RUNNING))
		if (scsi_host_set_state(shost, SHOST_CANCEL))
			BUG_ON(scsi_host_set_state(shost, SHOST_DEL));
	spin_unlock_irqrestore(shost->host_lock, flags);
L
Linus Torvalds 已提交
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

	wake_up(&shost->host_wait);

	/*
	 * finally we need to re-initiate requests that may be pending.  we will
	 * have had everything blocked while error handling is taking place, and
	 * now that error recovery is done, we will need to ensure that these
	 * requests are started.
	 */
	scsi_run_host_queues(shost);
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

	/*
	 * if eh is active and host_eh_scheduled is pending we need to re-run
	 * recovery.  we do this check after scsi_run_host_queues() to allow
	 * everything pent up since the last eh run a chance to make forward
	 * progress before we sync again.  Either we'll immediately re-run
	 * recovery or scsi_device_unbusy() will wake us again when these
	 * pending commands complete.
	 */
	spin_lock_irqsave(shost->host_lock, flags);
	if (shost->host_eh_scheduled)
		if (scsi_host_set_state(shost, SHOST_RECOVERY))
			WARN_ON(scsi_host_set_state(shost, SHOST_CANCEL_RECOVERY));
	spin_unlock_irqrestore(shost->host_lock, flags);
L
Linus Torvalds 已提交
2015 2016 2017 2018
}

/**
 * scsi_eh_ready_devs - check device ready state and recover if not.
2019 2020
 * @shost:	host to be recovered.
 * @work_q:	&list_head for pending commands.
2021
 * @done_q:	&list_head for processed commands.
2022
 */
2023 2024 2025
void scsi_eh_ready_devs(struct Scsi_Host *shost,
			struct list_head *work_q,
			struct list_head *done_q)
L
Linus Torvalds 已提交
2026 2027 2028
{
	if (!scsi_eh_stu(shost, work_q, done_q))
		if (!scsi_eh_bus_device_reset(shost, work_q, done_q))
2029 2030
			if (!scsi_eh_target_reset(shost, work_q, done_q))
				if (!scsi_eh_bus_reset(shost, work_q, done_q))
2031
					if (!scsi_eh_host_reset(shost, work_q, done_q))
2032 2033
						scsi_eh_offline_sdevs(work_q,
								      done_q);
L
Linus Torvalds 已提交
2034
}
2035
EXPORT_SYMBOL_GPL(scsi_eh_ready_devs);
L
Linus Torvalds 已提交
2036 2037 2038 2039

/**
 * scsi_eh_flush_done_q - finish processed commands or retry them.
 * @done_q:	list_head of processed commands.
2040
 */
2041
void scsi_eh_flush_done_q(struct list_head *done_q)
L
Linus Torvalds 已提交
2042
{
2043
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
2044

2045 2046
	list_for_each_entry_safe(scmd, next, done_q, eh_entry) {
		list_del_init(&scmd->eh_entry);
L
Linus Torvalds 已提交
2047
		if (scsi_device_online(scmd->device) &&
2048
		    !scsi_noretry_cmd(scmd) &&
2049
		    (++scmd->retries <= scmd->allowed)) {
2050 2051
			SCSI_LOG_ERROR_RECOVERY(3,
				scmd_printk(KERN_INFO, scmd,
2052 2053
					     "%s: flush retry cmd\n",
					     current->comm));
L
Linus Torvalds 已提交
2054 2055
				scsi_queue_insert(scmd, SCSI_MLQUEUE_EH_RETRY);
		} else {
2056 2057 2058 2059 2060
			/*
			 * If just we got sense for the device (called
			 * scsi_eh_get_sense), scmd->result is already
			 * set, do not set DRIVER_TIMEOUT.
			 */
L
Linus Torvalds 已提交
2061 2062
			if (!scmd->result)
				scmd->result |= (DRIVER_TIMEOUT << 24);
2063 2064
			SCSI_LOG_ERROR_RECOVERY(3,
				scmd_printk(KERN_INFO, scmd,
2065 2066
					     "%s: flush finish cmd\n",
					     current->comm));
L
Linus Torvalds 已提交
2067 2068 2069 2070
			scsi_finish_command(scmd);
		}
	}
}
2071
EXPORT_SYMBOL(scsi_eh_flush_done_q);
L
Linus Torvalds 已提交
2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094

/**
 * scsi_unjam_host - Attempt to fix a host which has a cmd that failed.
 * @shost:	Host to unjam.
 *
 * Notes:
 *    When we come in here, we *know* that all commands on the bus have
 *    either completed, failed or timed out.  we also know that no further
 *    commands are being sent to the host, so things are relatively quiet
 *    and we have freedom to fiddle with things as we wish.
 *
 *    This is only the *default* implementation.  it is possible for
 *    individual drivers to supply their own version of this function, and
 *    if the maintainer wishes to do this, it is strongly suggested that
 *    this function be taken as a template and modified.  this function
 *    was designed to correctly handle problems for about 95% of the
 *    different cases out there, and it should always provide at least a
 *    reasonable amount of error recovery.
 *
 *    Any command marked 'failed' or 'timeout' must eventually have
 *    scsi_finish_cmd() called for it.  we do all of the retry stuff
 *    here, so when we restart the host after we return it should have an
 *    empty queue.
2095
 */
L
Linus Torvalds 已提交
2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108
static void scsi_unjam_host(struct Scsi_Host *shost)
{
	unsigned long flags;
	LIST_HEAD(eh_work_q);
	LIST_HEAD(eh_done_q);

	spin_lock_irqsave(shost->host_lock, flags);
	list_splice_init(&shost->eh_cmd_q, &eh_work_q);
	spin_unlock_irqrestore(shost->host_lock, flags);

	SCSI_LOG_ERROR_RECOVERY(1, scsi_eh_prt_fail_stats(shost, &eh_work_q));

	if (!scsi_eh_get_sense(&eh_work_q, &eh_done_q))
2109
		scsi_eh_ready_devs(shost, &eh_work_q, &eh_done_q);
L
Linus Torvalds 已提交
2110

2111
	spin_lock_irqsave(shost->host_lock, flags);
2112
	if (shost->eh_deadline != -1)
2113 2114
		shost->last_reset = 0;
	spin_unlock_irqrestore(shost->host_lock, flags);
L
Linus Torvalds 已提交
2115 2116 2117 2118
	scsi_eh_flush_done_q(&eh_done_q);
}

/**
2119
 * scsi_error_handler - SCSI error handler thread
L
Linus Torvalds 已提交
2120 2121 2122
 * @data:	Host for which we are running.
 *
 * Notes:
2123 2124
 *    This is the main error handling loop.  This is run as a kernel thread
 *    for every SCSI host and handles all error handling activity.
2125
 */
L
Linus Torvalds 已提交
2126 2127
int scsi_error_handler(void *data)
{
2128
	struct Scsi_Host *shost = data;
L
Linus Torvalds 已提交
2129 2130

	/*
2131 2132 2133
	 * We use TASK_INTERRUPTIBLE so that the thread is not
	 * counted against the load average as a running process.
	 * We never actually get interrupted because kthread_run
2134
	 * disables signal delivery for the created thread.
L
Linus Torvalds 已提交
2135
	 */
2136 2137 2138 2139 2140 2141 2142
	while (true) {
		/*
		 * The sequence in kthread_stop() sets the stop flag first
		 * then wakes the process.  To avoid missed wakeups, the task
		 * should always be in a non running state before the stop
		 * flag is checked
		 */
2143
		set_current_state(TASK_INTERRUPTIBLE);
2144 2145 2146
		if (kthread_should_stop())
			break;

2147
		if ((shost->host_failed == 0 && shost->host_eh_scheduled == 0) ||
2148
		    shost->host_failed != atomic_read(&shost->host_busy)) {
2149
			SCSI_LOG_ERROR_RECOVERY(1,
2150 2151 2152
				shost_printk(KERN_INFO, shost,
					     "scsi_eh_%d: sleeping\n",
					     shost->host_no));
2153 2154 2155
			schedule();
			continue;
		}
L
Linus Torvalds 已提交
2156

2157
		__set_current_state(TASK_RUNNING);
2158
		SCSI_LOG_ERROR_RECOVERY(1,
2159 2160 2161
			shost_printk(KERN_INFO, shost,
				     "scsi_eh_%d: waking up %d/%d/%d\n",
				     shost->host_no, shost->host_eh_scheduled,
2162 2163
				     shost->host_failed,
				     atomic_read(&shost->host_busy)));
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169

		/*
		 * We have a host that is failing for some reason.  Figure out
		 * what we need to do to get it up and online again (if we can).
		 * If we fail, we end up taking the thing offline.
		 */
2170
		if (!shost->eh_noresume && scsi_autopm_get_host(shost) != 0) {
2171
			SCSI_LOG_ERROR_RECOVERY(1,
2172 2173 2174
				shost_printk(KERN_ERR, shost,
					     "scsi_eh_%d: unable to autoresume\n",
					     shost->host_no));
2175 2176 2177
			continue;
		}

2178 2179
		if (shost->transportt->eh_strategy_handler)
			shost->transportt->eh_strategy_handler(shost);
L
Linus Torvalds 已提交
2180 2181 2182
		else
			scsi_unjam_host(shost);

2183 2184 2185
		/* All scmds have been handled */
		shost->host_failed = 0;

L
Linus Torvalds 已提交
2186 2187 2188 2189 2190 2191 2192 2193
		/*
		 * Note - if the above fails completely, the action is to take
		 * individual devices offline and flush the queue of any
		 * outstanding requests that may have been pending.  When we
		 * restart, we restart any I/O to any other devices on the bus
		 * which are still online.
		 */
		scsi_restart_operations(shost);
2194 2195
		if (!shost->eh_noresume)
			scsi_autopm_put_host(shost);
L
Linus Torvalds 已提交
2196
	}
2197 2198
	__set_current_state(TASK_RUNNING);

2199
	SCSI_LOG_ERROR_RECOVERY(1,
2200 2201 2202
		shost_printk(KERN_INFO, shost,
			     "Error handler scsi_eh_%d exiting\n",
			     shost->host_no));
2203
	shost->ehandler = NULL;
L
Linus Torvalds 已提交
2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232
	return 0;
}

/*
 * Function:    scsi_report_bus_reset()
 *
 * Purpose:     Utility function used by low-level drivers to report that
 *		they have observed a bus reset on the bus being handled.
 *
 * Arguments:   shost       - Host in question
 *		channel     - channel on which reset was observed.
 *
 * Returns:     Nothing
 *
 * Lock status: Host lock must be held.
 *
 * Notes:       This only needs to be called if the reset is one which
 *		originates from an unknown location.  Resets originated
 *		by the mid-level itself don't need to call this, but there
 *		should be no harm.
 *
 *		The main purpose of this is to make sure that a CHECK_CONDITION
 *		is properly treated.
 */
void scsi_report_bus_reset(struct Scsi_Host *shost, int channel)
{
	struct scsi_device *sdev;

	__shost_for_each_device(sdev, shost) {
2233 2234
		if (channel == sdev_channel(sdev))
			__scsi_report_device_reset(sdev, NULL);
L
Linus Torvalds 已提交
2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
	}
}
EXPORT_SYMBOL(scsi_report_bus_reset);

/*
 * Function:    scsi_report_device_reset()
 *
 * Purpose:     Utility function used by low-level drivers to report that
 *		they have observed a device reset on the device being handled.
 *
 * Arguments:   shost       - Host in question
 *		channel     - channel on which reset was observed
 *		target	    - target on which reset was observed
 *
 * Returns:     Nothing
 *
 * Lock status: Host lock must be held
 *
 * Notes:       This only needs to be called if the reset is one which
 *		originates from an unknown location.  Resets originated
 *		by the mid-level itself don't need to call this, but there
 *		should be no harm.
 *
 *		The main purpose of this is to make sure that a CHECK_CONDITION
 *		is properly treated.
 */
void scsi_report_device_reset(struct Scsi_Host *shost, int channel, int target)
{
	struct scsi_device *sdev;

	__shost_for_each_device(sdev, shost) {
2266
		if (channel == sdev_channel(sdev) &&
2267 2268
		    target == sdev_id(sdev))
			__scsi_report_device_reset(sdev, NULL);
L
Linus Torvalds 已提交
2269 2270 2271 2272 2273 2274 2275 2276 2277
	}
}
EXPORT_SYMBOL(scsi_report_device_reset);

static void
scsi_reset_provider_done_command(struct scsi_cmnd *scmd)
{
}

2278 2279 2280 2281
/**
 * scsi_ioctl_reset: explicitly reset a host/bus/target/device
 * @dev:	scsi_device to operate on
 * @arg:	reset type (see sg.h)
L
Linus Torvalds 已提交
2282 2283
 */
int
2284
scsi_ioctl_reset(struct scsi_device *dev, int __user *arg)
L
Linus Torvalds 已提交
2285
{
2286
	struct scsi_cmnd *scmd;
2287
	struct Scsi_Host *shost = dev->host;
2288
	struct request *rq;
2289
	unsigned long flags;
2290 2291 2292 2293 2294 2295 2296 2297
	int error = 0, rtn, val;

	if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SYS_RAWIO))
		return -EACCES;

	error = get_user(val, arg);
	if (error)
		return error;
L
Linus Torvalds 已提交
2298

2299
	if (scsi_autopm_get_host(shost) < 0)
2300
		return -EIO;
2301

2302
	error = -EIO;
2303 2304 2305
	rq = kzalloc(sizeof(struct request) + sizeof(struct scsi_cmnd) +
			shost->hostt->cmd_size, GFP_KERNEL);
	if (!rq)
2306
		goto out_put_autopm_host;
2307
	blk_rq_init(NULL, rq);
2308

2309 2310 2311
	scmd = (struct scsi_cmnd *)(rq + 1);
	scsi_init_command(dev, scmd);
	scmd->request = rq;
2312
	scmd->cmnd = scsi_req(rq)->cmd;
2313

L
Linus Torvalds 已提交
2314
	scmd->scsi_done		= scsi_reset_provider_done_command;
B
Boaz Harrosh 已提交
2315
	memset(&scmd->sdb, 0, sizeof(scmd->sdb));
L
Linus Torvalds 已提交
2316 2317 2318 2319 2320

	scmd->cmd_len			= 0;

	scmd->sc_data_direction		= DMA_BIDIRECTIONAL;

2321 2322 2323 2324
	spin_lock_irqsave(shost->host_lock, flags);
	shost->tmf_in_progress = 1;
	spin_unlock_irqrestore(shost->host_lock, flags);

2325 2326 2327 2328 2329
	switch (val & ~SG_SCSI_RESET_NO_ESCALATE) {
	case SG_SCSI_RESET_NOTHING:
		rtn = SUCCESS;
		break;
	case SG_SCSI_RESET_DEVICE:
L
Linus Torvalds 已提交
2330
		rtn = scsi_try_bus_device_reset(scmd);
2331
		if (rtn == SUCCESS || (val & SG_SCSI_RESET_NO_ESCALATE))
L
Linus Torvalds 已提交
2332 2333
			break;
		/* FALLTHROUGH */
2334
	case SG_SCSI_RESET_TARGET:
2335
		rtn = scsi_try_target_reset(scmd);
2336
		if (rtn == SUCCESS || (val & SG_SCSI_RESET_NO_ESCALATE))
2337 2338
			break;
		/* FALLTHROUGH */
2339
	case SG_SCSI_RESET_BUS:
L
Linus Torvalds 已提交
2340
		rtn = scsi_try_bus_reset(scmd);
2341
		if (rtn == SUCCESS || (val & SG_SCSI_RESET_NO_ESCALATE))
L
Linus Torvalds 已提交
2342 2343
			break;
		/* FALLTHROUGH */
2344
	case SG_SCSI_RESET_HOST:
L
Linus Torvalds 已提交
2345
		rtn = scsi_try_host_reset(scmd);
2346 2347 2348
		if (rtn == SUCCESS)
			break;
		/* FALLTHROUGH */
2349
	default:
L
Linus Torvalds 已提交
2350
		rtn = FAILED;
2351
		break;
L
Linus Torvalds 已提交
2352 2353
	}

2354 2355
	error = (rtn == SUCCESS) ? 0 : -EIO;

2356 2357 2358 2359 2360 2361 2362 2363 2364
	spin_lock_irqsave(shost->host_lock, flags);
	shost->tmf_in_progress = 0;
	spin_unlock_irqrestore(shost->host_lock, flags);

	/*
	 * be sure to wake up anyone who was sleeping or had their queue
	 * suspended while we performed the TMF.
	 */
	SCSI_LOG_ERROR_RECOVERY(3,
2365 2366
		shost_printk(KERN_INFO, shost,
			     "waking up host to restart after TMF\n"));
2367 2368 2369 2370

	wake_up(&shost->host_wait);
	scsi_run_host_queues(shost);

2371
	scsi_put_command(scmd);
2372
	kfree(rq);
2373

2374
out_put_autopm_host:
2375
	scsi_autopm_put_host(shost);
2376
	return error;
L
Linus Torvalds 已提交
2377
}
2378
EXPORT_SYMBOL(scsi_ioctl_reset);
L
Linus Torvalds 已提交
2379

2380 2381
bool scsi_command_normalize_sense(const struct scsi_cmnd *cmd,
				  struct scsi_sense_hdr *sshdr)
L
Linus Torvalds 已提交
2382 2383
{
	return scsi_normalize_sense(cmd->sense_buffer,
2384
			SCSI_SENSE_BUFFERSIZE, sshdr);
L
Linus Torvalds 已提交
2385 2386 2387 2388
}
EXPORT_SYMBOL(scsi_command_normalize_sense);

/**
2389
 * scsi_get_sense_info_fld - get information field from sense data (either fixed or descriptor format)
L
Linus Torvalds 已提交
2390 2391 2392 2393 2394 2395
 * @sense_buffer:	byte array of sense data
 * @sb_len:		number of valid bytes in sense_buffer
 * @info_out:		pointer to 64 integer where 8 or 4 byte information
 *			field will be placed if found.
 *
 * Return value:
2396
 *	true if information field found, false if not found.
2397
 */
2398 2399
bool scsi_get_sense_info_fld(const u8 *sense_buffer, int sb_len,
			     u64 *info_out)
L
Linus Torvalds 已提交
2400 2401 2402 2403
{
	const u8 * ucp;

	if (sb_len < 7)
2404
		return false;
L
Linus Torvalds 已提交
2405 2406 2407 2408
	switch (sense_buffer[0] & 0x7f) {
	case 0x70:
	case 0x71:
		if (sense_buffer[0] & 0x80) {
2409 2410 2411 2412
			*info_out = get_unaligned_be32(&sense_buffer[3]);
			return true;
		}
		return false;
L
Linus Torvalds 已提交
2413 2414 2415 2416 2417
	case 0x72:
	case 0x73:
		ucp = scsi_sense_desc_find(sense_buffer, sb_len,
					   0 /* info desc */);
		if (ucp && (0xa == ucp[1])) {
2418 2419 2420 2421
			*info_out = get_unaligned_be64(&ucp[4]);
			return true;
		}
		return false;
L
Linus Torvalds 已提交
2422
	default:
2423
		return false;
L
Linus Torvalds 已提交
2424 2425 2426
	}
}
EXPORT_SYMBOL(scsi_get_sense_info_fld);