scsi_error.c 56.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 *  scsi_error.c Copyright (C) 1997 Eric Youngdale
 *
 *  SCSI error/timeout handling
 *      Initial versions: Eric Youngdale.  Based upon conversations with
 *                        Leonard Zubkoff and David Miller at Linux Expo, 
 *                        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
 *	minor  cleanups.
 *	September 30, 2002 Mike Anderson (andmike@us.ibm.com)
 */

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

#include <scsi/scsi.h>
29
#include <scsi/scsi_cmnd.h>
L
Linus Torvalds 已提交
30 31 32
#include <scsi/scsi_dbg.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_eh.h>
33
#include <scsi/scsi_transport.h>
L
Linus Torvalds 已提交
34 35 36 37 38
#include <scsi/scsi_host.h>
#include <scsi/scsi_ioctl.h>

#include "scsi_priv.h"
#include "scsi_logging.h"
39
#include "scsi_transport_api.h"
L
Linus Torvalds 已提交
40 41 42 43 44 45 46 47 48 49 50 51 52 53

#define SENSE_TIMEOUT		(10*HZ)

/*
 * 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)

/* called with shost->host_lock held */
void scsi_eh_wakeup(struct Scsi_Host *shost)
{
	if (shost->host_busy == shost->host_failed) {
54
		wake_up_process(shost->ehandler);
L
Linus Torvalds 已提交
55 56 57 58
		SCSI_LOG_ERROR_RECOVERY(5,
				printk("Waking error handler thread\n"));
	}
}
59 60 61 62 63 64

/**
 * scsi_schedule_eh - schedule EH for SCSI host
 * @shost:	SCSI host to invoke error handling on.
 *
 * Schedule SCSI EH without scmd.
65
 */
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
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
81 82 83 84 85 86 87 88

/**
 * scsi_eh_scmd_add - add scsi cmd to error handling.
 * @scmd:	scmd to run eh on.
 * @eh_flag:	optional SCSI_EH flag.
 *
 * Return value:
 *	0 on failure.
89
 */
L
Linus Torvalds 已提交
90 91 92 93
int scsi_eh_scmd_add(struct scsi_cmnd *scmd, int eh_flag)
{
	struct Scsi_Host *shost = scmd->device->host;
	unsigned long flags;
94
	int ret = 0;
L
Linus Torvalds 已提交
95

96
	if (!shost->ehandler)
L
Linus Torvalds 已提交
97 98 99
		return 0;

	spin_lock_irqsave(shost->host_lock, flags);
100 101 102
	if (scsi_host_set_state(shost, SHOST_RECOVERY))
		if (scsi_host_set_state(shost, SHOST_CANCEL_RECOVERY))
			goto out_unlock;
L
Linus Torvalds 已提交
103

104
	ret = 1;
105
	scmd->eh_eflags |= eh_flag;
L
Linus Torvalds 已提交
106 107 108
	list_add_tail(&scmd->eh_entry, &shost->eh_cmd_q);
	shost->host_failed++;
	scsi_eh_wakeup(shost);
109
 out_unlock:
L
Linus Torvalds 已提交
110
	spin_unlock_irqrestore(shost->host_lock, flags);
111
	return ret;
L
Linus Torvalds 已提交
112 113 114 115
}

/**
 * scsi_times_out - Timeout function for normal scsi commands.
J
Jens Axboe 已提交
116
 * @req:	request that is timing out.
L
Linus Torvalds 已提交
117 118 119 120 121 122
 *
 * 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.
123
 */
J
Jens Axboe 已提交
124
enum blk_eh_timer_return scsi_times_out(struct request *req)
L
Linus Torvalds 已提交
125
{
J
Jens Axboe 已提交
126 127
	struct scsi_cmnd *scmd = req->special;
	enum blk_eh_timer_return rtn = BLK_EH_NOT_HANDLED;
128

L
Linus Torvalds 已提交
129 130
	scsi_log_completion(scmd, TIMEOUT_ERROR);

131
	if (scmd->device->host->transportt->eh_timed_out)
132
		rtn = scmd->device->host->transportt->eh_timed_out(scmd);
133
	else if (scmd->device->host->hostt->eh_timed_out)
134
		rtn = scmd->device->host->hostt->eh_timed_out(scmd);
L
Linus Torvalds 已提交
135

136 137
	if (unlikely(rtn == BLK_EH_NOT_HANDLED &&
		     !scsi_eh_scmd_add(scmd, SCSI_EH_CANCEL_CMD))) {
138
		scmd->result |= DID_TIME_OUT << 16;
139
		rtn = BLK_EH_HANDLED;
L
Linus Torvalds 已提交
140
	}
J
Jens Axboe 已提交
141

142
	return rtn;
L
Linus Torvalds 已提交
143 144 145 146 147 148 149 150 151 152 153 154
}

/**
 * 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.
155
 */
L
Linus Torvalds 已提交
156 157 158 159
int scsi_block_when_processing_errors(struct scsi_device *sdev)
{
	int online;

160
	wait_event(sdev->host->host_wait, !scsi_host_in_recovery(sdev->host));
L
Linus Torvalds 已提交
161 162 163

	online = scsi_device_online(sdev);

164
	SCSI_LOG_ERROR_RECOVERY(5, printk("%s: rtn: %d\n", __func__,
L
Linus Torvalds 已提交
165 166 167 168 169 170 171 172 173 174 175
					  online));

	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.
176
 */
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185 186 187 188 189 190
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;
191
				if (scmd->eh_eflags & SCSI_EH_CANCEL_CMD)
L
Linus Torvalds 已提交
192 193 194 195 196 197 198 199
					++cmd_cancel;
				else 
					++cmd_failed;
			}
		}

		if (cmd_cancel || cmd_failed) {
			SCSI_LOG_ERROR_RECOVERY(3,
200 201
				sdev_printk(KERN_INFO, sdev,
					    "%s: cmds failed: %d, cancel: %d\n",
202
					    __func__, cmd_failed,
203
					    cmd_cancel));
L
Linus Torvalds 已提交
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
			cmd_cancel = 0;
			cmd_failed = 0;
			++devices_failed;
		}
	}

	SCSI_LOG_ERROR_RECOVERY(2, printk("Total of %d commands on %d"
					  " devices require eh work\n",
				  total_failures, devices_failed));
}
#endif

/**
 * scsi_check_sense - Examine scsi cmd sense
 * @scmd:	Cmd to have sense checked.
 *
 * Return value:
 * 	SUCCESS or FAILED or NEEDS_RETRY
 *
 * Notes:
 *	When a deferred error is detected the current command has
 *	not been executed and needs retrying.
226
 */
L
Linus Torvalds 已提交
227 228
static int scsi_check_sense(struct scsi_cmnd *scmd)
{
229
	struct scsi_device *sdev = scmd->device;
L
Linus Torvalds 已提交
230 231 232 233 234 235 236 237
	struct scsi_sense_hdr sshdr;

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

	if (scsi_sense_is_deferred(&sshdr))
		return NEEDS_RETRY;

238 239 240 241 242 243 244 245 246 247
	if (sdev->scsi_dh_data && sdev->scsi_dh_data->scsi_dh &&
			sdev->scsi_dh_data->scsi_dh->check_sense) {
		int rc;

		rc = sdev->scsi_dh_data->scsi_dh->check_sense(sdev, &sshdr);
		if (rc != SCSI_RETURN_NOT_HANDLED)
			return rc;
		/* handler does not care. Drop down to default handling */
	}

L
Linus Torvalds 已提交
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
	/*
	 * 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:
275 276 277
		if (sshdr.asc == 0x10) /* DIF */
			return SUCCESS;

L
Linus Torvalds 已提交
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
		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) {
			scmd->device->expecting_cc_ua = 0;
			return NEEDS_RETRY;
		}
		/*
		 * if the device is in the process of becoming ready, we 
		 * 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;
		return SUCCESS;

		/* these three are not supported */
	case COPY_ABORTED:
	case VOLUME_OVERFLOW:
	case MISCOMPARE:
		return SUCCESS;

	case MEDIUM_ERROR:
313 314 315 316 317
		if (sshdr.asc == 0x11 || /* UNRECOVERED READ ERR */
		    sshdr.asc == 0x13 || /* AMNF DATA FIELD */
		    sshdr.asc == 0x14) { /* RECORD NOT FOUND */
			return SUCCESS;
		}
L
Linus Torvalds 已提交
318 319 320 321
		return NEEDS_RETRY;

	case HARDWARE_ERROR:
		if (scmd->device->retry_hwerror)
322
			return ADD_TO_MLQUEUE;
L
Linus Torvalds 已提交
323 324 325 326 327 328 329 330 331 332 333
		else
			return SUCCESS;

	case ILLEGAL_REQUEST:
	case BLANK_CHECK:
	case DATA_PROTECT:
	default:
		return SUCCESS;
	}
}

V
Vasu Dev 已提交
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
static void scsi_handle_queue_ramp_up(struct scsi_device *sdev)
{
	struct scsi_host_template *sht = sdev->host->hostt;
	struct scsi_device *tmp_sdev;

	if (!sht->change_queue_depth ||
	    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;
		/*
		 * call back into LLD to increase queue_depth by one
		 * with ramp up reason code.
		 */
		sht->change_queue_depth(tmp_sdev, tmp_sdev->queue_depth + 1,
					SCSI_QDEPTH_RAMP_UP);
		sdev->last_queue_ramp_up = jiffies;
	}
}

370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
static void scsi_handle_queue_full(struct scsi_device *sdev)
{
	struct scsi_host_template *sht = sdev->host->hostt;
	struct scsi_device *tmp_sdev;

	if (!sht->change_queue_depth)
		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.
		 */
		sht->change_queue_depth(tmp_sdev, tmp_sdev->queue_depth - 1,
					SCSI_QDEPTH_QFULL);
	}
}

L
Linus Torvalds 已提交
392 393 394 395 396 397 398 399 400
/**
 * 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.
401
 */
L
Linus Torvalds 已提交
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431
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
432
		scsi_handle_queue_ramp_up(scmd->device);
L
Linus Torvalds 已提交
433 434 435 436 437 438 439 440 441 442 443
	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;
444 445 446 447 448
	case RESERVATION_CONFLICT:
		/*
		 * let issuer deal with this, it could be just fine
		 */
		return SUCCESS;
L
Linus Torvalds 已提交
449
	case QUEUE_FULL:
450 451 452
		scsi_handle_queue_full(scmd->device);
		/* fall through */
	case BUSY:
L
Linus Torvalds 已提交
453 454 455 456 457 458 459 460 461
	default:
		return FAILED;
	}
	return FAILED;
}

/**
 * scsi_eh_done - Completion function for error handling.
 * @scmd:	Cmd that is done.
462
 */
L
Linus Torvalds 已提交
463 464
static void scsi_eh_done(struct scsi_cmnd *scmd)
{
465 466
	struct completion     *eh_action;

467 468
	SCSI_LOG_ERROR_RECOVERY(3,
		printk("%s scmd: %p result: %x\n",
469
			__func__, scmd, scmd->result));
470 471 472 473

	eh_action = scmd->device->host->eh_action;
	if (eh_action)
		complete(eh_action);
L
Linus Torvalds 已提交
474 475
}

476 477 478
/**
 * scsi_try_host_reset - ask host adapter to reset itself
 * @scmd:	SCSI cmd to send hsot reset.
479
 */
480 481 482 483 484 485
static int scsi_try_host_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;

	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Snd Host RST\n",
486
					  __func__));
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507

	if (!scmd->device->host->hostt->eh_host_reset_handler)
		return FAILED;

	rtn = scmd->device->host->hostt->eh_host_reset_handler(scmd);

	if (rtn == SUCCESS) {
		if (!scmd->device->host->hostt->skip_settle_delay)
			ssleep(HOST_RESET_SETTLE_TIME);
		spin_lock_irqsave(scmd->device->host->host_lock, flags);
		scsi_report_bus_reset(scmd->device->host,
				      scmd_channel(scmd));
		spin_unlock_irqrestore(scmd->device->host->host_lock, flags);
	}

	return rtn;
}

/**
 * scsi_try_bus_reset - ask host to perform a bus reset
 * @scmd:	SCSI cmd to send bus reset.
508
 */
509 510 511 512 513 514
static int scsi_try_bus_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;

	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Snd Bus RST\n",
515
					  __func__));
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533

	if (!scmd->device->host->hostt->eh_bus_reset_handler)
		return FAILED;

	rtn = scmd->device->host->hostt->eh_bus_reset_handler(scmd);

	if (rtn == SUCCESS) {
		if (!scmd->device->host->hostt->skip_settle_delay)
			ssleep(BUS_RESET_SETTLE_TIME);
		spin_lock_irqsave(scmd->device->host->host_lock, flags);
		scsi_report_bus_reset(scmd->device->host,
				      scmd_channel(scmd));
		spin_unlock_irqrestore(scmd->device->host->host_lock, flags);
	}

	return rtn;
}

534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
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;

	if (!scmd->device->host->hostt->eh_target_reset_handler)
		return FAILED;

	rtn = scmd->device->host->hostt->eh_target_reset_handler(scmd);
	if (rtn == SUCCESS) {
		spin_lock_irqsave(scmd->device->host->host_lock, flags);
		__starget_for_each_device(scsi_target(scmd->device), NULL,
					  __scsi_report_device_reset);
		spin_unlock_irqrestore(scmd->device->host->host_lock, flags);
	}

	return rtn;
}

569 570 571 572 573 574 575 576 577
/**
 * 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.
578
 */
579 580 581 582 583 584 585 586
static int scsi_try_bus_device_reset(struct scsi_cmnd *scmd)
{
	int rtn;

	if (!scmd->device->host->hostt->eh_device_reset_handler)
		return FAILED;

	rtn = scmd->device->host->hostt->eh_device_reset_handler(scmd);
587 588
	if (rtn == SUCCESS)
		__scsi_report_device_reset(scmd->device, NULL);
589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609
	return rtn;
}

static int __scsi_try_to_abort_cmd(struct scsi_cmnd *scmd)
{
	if (!scmd->device->host->hostt->eh_abort_handler)
		return FAILED;

	return scmd->device->host->hostt->eh_abort_handler(scmd);
}

/**
 * scsi_try_to_abort_cmd - Ask host to abort a running command.
 * @scmd:	SCSI cmd to abort from Lower Level.
 *
 * Notes:
 *    This function will not return until the user's completion function
 *    has been called.  there is no timeout on this operation.  if the
 *    author of the low-level driver wishes this operation to be timed,
 *    they can provide this facility themselves.  helper functions in
 *    scsi_error.c can be supplied to make this easier to do.
610
 */
611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
static int scsi_try_to_abort_cmd(struct scsi_cmnd *scmd)
{
	/*
	 * scsi_done was called just after the command timed out and before
	 * we had a chance to process it. (db)
	 */
	if (scmd->serial_number == 0)
		return SUCCESS;
	return __scsi_try_to_abort_cmd(scmd);
}

static void scsi_abort_eh_cmnd(struct scsi_cmnd *scmd)
{
	if (__scsi_try_to_abort_cmd(scmd) != SUCCESS)
		if (scsi_try_bus_device_reset(scmd) != SUCCESS)
626 627 628
			if (scsi_try_target_reset(scmd) != SUCCESS)
				if (scsi_try_bus_reset(scmd) != SUCCESS)
					scsi_try_host_reset(scmd);
629 630
}

L
Linus Torvalds 已提交
631
/**
632
 * scsi_eh_prep_cmnd  - Save a scsi command info as part of error recory
633
 * @scmd:       SCSI command structure to hijack
634
 * @ses:        structure to save restore information
635
 * @cmnd:       CDB to send. Can be NULL if no new cmnd is needed
636
 * @cmnd_size:  size in bytes of @cmnd (must be <= BLK_MAX_CDB)
637
 * @sense_bytes: size of sense data to copy. or 0 (if != 0 @cmnd is ignored)
638
 *
639
 * This function is used to save a scsi command information before re-execution
640 641 642 643
 * 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.
644
 */
645 646
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
647
{
已提交
648
	struct scsi_device *sdev = scmd->device;
L
Linus Torvalds 已提交
649

650 651 652 653 654 655 656
	/*
	 * 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.
	 */
657
	ses->cmd_len = scmd->cmd_len;
658
	ses->cmnd = scmd->cmnd;
659
	ses->data_direction = scmd->sc_data_direction;
B
Boaz Harrosh 已提交
660
	ses->sdb = scmd->sdb;
661
	ses->next_rq = scmd->request->next_rq;
662
	ses->result = scmd->result;
663
	ses->underflow = scmd->underflow;
664
	ses->prot_op = scmd->prot_op;
665

666
	scmd->prot_op = SCSI_PROT_NORMAL;
667 668
	scmd->cmnd = ses->eh_cmnd;
	memset(scmd->cmnd, 0, BLK_MAX_CDB);
B
Boaz Harrosh 已提交
669
	memset(&scmd->sdb, 0, sizeof(scmd->sdb));
670
	scmd->request->next_rq = NULL;
B
Boaz Harrosh 已提交
671

672
	if (sense_bytes) {
B
Boaz Harrosh 已提交
673 674
		scmd->sdb.length = min_t(unsigned, SCSI_SENSE_BUFFERSIZE,
					 sense_bytes);
675
		sg_init_one(&ses->sense_sgl, scmd->sense_buffer,
B
Boaz Harrosh 已提交
676 677
			    scmd->sdb.length);
		scmd->sdb.table.sgl = &ses->sense_sgl;
678
		scmd->sc_data_direction = DMA_FROM_DEVICE;
B
Boaz Harrosh 已提交
679
		scmd->sdb.table.nents = 1;
680
		scmd->cmnd[0] = REQUEST_SENSE;
B
Boaz Harrosh 已提交
681
		scmd->cmnd[4] = scmd->sdb.length;
682
		scmd->cmd_len = COMMAND_SIZE(scmd->cmnd[0]);
683 684
	} else {
		scmd->sc_data_direction = DMA_NONE;
685
		if (cmnd) {
686
			BUG_ON(cmnd_size > BLK_MAX_CDB);
687 688 689
			memcpy(scmd->cmnd, cmnd, cmnd_size);
			scmd->cmd_len = COMMAND_SIZE(scmd->cmnd[0]);
		}
690 691 692 693
	}

	scmd->underflow = 0;

694
	if (sdev->scsi_level <= SCSI_2 && sdev->scsi_level != SCSI_UNKNOWN)
L
Linus Torvalds 已提交
695
		scmd->cmnd[1] = (scmd->cmnd[1] & 0x1f) |
已提交
696
			(sdev->lun << 5 & 0xe0);
L
Linus Torvalds 已提交
697

698 699 700 701
	/*
	 * Zero the sense buffer.  The scsi spec mandates that any
	 * untransferred sense data should be interpreted as being zero.
	 */
702
	memset(scmd->sense_buffer, 0, SCSI_SENSE_BUFFERSIZE);
703 704 705 706 707 708
}
EXPORT_SYMBOL(scsi_eh_prep_cmnd);

/**
 * scsi_eh_restore_cmnd  - Restore a scsi command info as part of error recory
 * @scmd:       SCSI command structure to restore
709
 * @ses:        saved information from a coresponding call to scsi_eh_prep_cmnd
710
 *
711
 * Undo any damage done by above scsi_eh_prep_cmnd().
712
 */
713 714 715 716 717 718
void scsi_eh_restore_cmnd(struct scsi_cmnd* scmd, struct scsi_eh_save *ses)
{
	/*
	 * Restore original data
	 */
	scmd->cmd_len = ses->cmd_len;
719
	scmd->cmnd = ses->cmnd;
720
	scmd->sc_data_direction = ses->data_direction;
B
Boaz Harrosh 已提交
721
	scmd->sdb = ses->sdb;
722
	scmd->request->next_rq = ses->next_rq;
723
	scmd->result = ses->result;
724
	scmd->underflow = ses->underflow;
725
	scmd->prot_op = ses->prot_op;
726 727
}
EXPORT_SYMBOL(scsi_eh_restore_cmnd);
728

729 730 731 732 733 734 735 736 737 738 739 740 741
/**
 * scsi_send_eh_cmnd  - submit a scsi command as part of error recory
 * @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
742
 */
743 744 745 746 747 748 749 750 751 752 753 754
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);
	unsigned long timeleft;
	unsigned long flags;
	struct scsi_eh_save ses;
	int rtn;

	scsi_eh_prep_cmnd(scmd, &ses, cmnd, cmnd_size, sense_bytes);
755
	shost->eh_action = &done;
L
Linus Torvalds 已提交
756

已提交
757
	spin_lock_irqsave(shost->host_lock, flags);
L
Linus Torvalds 已提交
758
	scsi_log_send(scmd);
已提交
759 760
	shost->hostt->queuecommand(scmd, scsi_eh_done);
	spin_unlock_irqrestore(shost->host_lock, flags);
L
Linus Torvalds 已提交
761

762
	timeleft = wait_for_completion_timeout(&done, timeout);
L
Linus Torvalds 已提交
763

已提交
764
	shost->eh_action = NULL;
L
Linus Torvalds 已提交
765

766
	scsi_log_completion(scmd, SUCCESS);
L
Linus Torvalds 已提交
767

768 769
	SCSI_LOG_ERROR_RECOVERY(3,
		printk("%s: scmd: %p, timeleft: %ld\n",
770
			__func__, scmd, timeleft));
L
Linus Torvalds 已提交
771 772

	/*
773 774 775 776
	 * 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 tell the host to forget
	 * about this command.
L
Linus Torvalds 已提交
777
	 */
778
	if (timeleft) {
L
Linus Torvalds 已提交
779 780 781
		rtn = scsi_eh_completed_normally(scmd);
		SCSI_LOG_ERROR_RECOVERY(3,
			printk("%s: scsi_eh_completed_normally %x\n",
782
			       __func__, rtn));
783

L
Linus Torvalds 已提交
784 785 786 787 788
		switch (rtn) {
		case SUCCESS:
		case NEEDS_RETRY:
		case FAILED:
			break;
789 790 791
		case ADD_TO_MLQUEUE:
			rtn = NEEDS_RETRY;
			break;
L
Linus Torvalds 已提交
792 793 794 795
		default:
			rtn = FAILED;
			break;
		}
796
	} else {
797
		scsi_abort_eh_cmnd(scmd);
798
		rtn = FAILED;
L
Linus Torvalds 已提交
799 800
	}

801
	scsi_eh_restore_cmnd(scmd, &ses);
L
Linus Torvalds 已提交
802 803 804 805 806 807 808 809 810 811 812
	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.
813
 */
L
Linus Torvalds 已提交
814 815
static int scsi_request_sense(struct scsi_cmnd *scmd)
{
816
	return scsi_send_eh_cmnd(scmd, NULL, 0, SENSE_TIMEOUT, ~0);
L
Linus Torvalds 已提交
817 818 819 820 821 822 823 824 825 826
}

/**
 * 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,
827
 *    and that would disturb what we are doing.  Thus we really want to
L
Linus Torvalds 已提交
828 829
 *    keep a list of pending commands for final completion, and once we
 *    are ready to leave error handling we handle completion for real.
830
 */
831
void scsi_eh_finish_cmd(struct scsi_cmnd *scmd, struct list_head *done_q)
L
Linus Torvalds 已提交
832 833
{
	scmd->device->host->host_failed--;
834
	scmd->eh_eflags = 0;
L
Linus Torvalds 已提交
835 836
	list_move_tail(&scmd->eh_entry, done_q);
}
837
EXPORT_SYMBOL(scsi_eh_finish_cmd);
L
Linus Torvalds 已提交
838 839 840 841

/**
 * scsi_eh_get_sense - Get device sense data.
 * @work_q:	Queue of commands to process.
842
 * @done_q:	Queue of processed commands.
L
Linus Torvalds 已提交
843 844 845 846 847 848 849
 *
 * Description:
 *    See if we need to request sense information.  if so, then get it
 *    now, so we have a better idea of what to do.  
 *
 * Notes:
 *    This has the unfortunate side effect that if a shost adapter does
850
 *    not automatically request sense information, we end up shutting
L
Linus Torvalds 已提交
851 852 853 854 855 856 857
 *    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.
858
 */
859 860
int scsi_eh_get_sense(struct list_head *work_q,
		      struct list_head *done_q)
L
Linus Torvalds 已提交
861
{
862
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
863 864
	int rtn;

865
	list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
866
		if ((scmd->eh_eflags & SCSI_EH_CANCEL_CMD) ||
L
Linus Torvalds 已提交
867 868 869
		    SCSI_SENSE_VALID(scmd))
			continue;

870 871 872
		SCSI_LOG_ERROR_RECOVERY(2, scmd_printk(KERN_INFO, scmd,
						  "%s: requesting sense\n",
						  current->comm));
L
Linus Torvalds 已提交
873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
		rtn = scsi_request_sense(scmd);
		if (rtn != SUCCESS)
			continue;

		SCSI_LOG_ERROR_RECOVERY(3, printk("sense requested for %p"
						  " result %x\n", scmd,
						  scmd->result));
		SCSI_LOG_ERROR_RECOVERY(3, scsi_print_sense("bh", scmd));

		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);
}
902
EXPORT_SYMBOL_GPL(scsi_eh_get_sense);
L
Linus Torvalds 已提交
903 904 905

/**
 * scsi_eh_tur - Send TUR to device.
906
 * @scmd:	&scsi_cmnd to send TUR
L
Linus Torvalds 已提交
907 908 909
 *
 * Return value:
 *    0 - Device is ready. 1 - Device NOT ready.
910
 */
L
Linus Torvalds 已提交
911 912 913 914 915 916
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:
917
	rtn = scsi_send_eh_cmnd(scmd, tur_command, 6, SENSE_TIMEOUT, 0);
L
Linus Torvalds 已提交
918 919

	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: scmd %p rtn %x\n",
920
		__func__, scmd, rtn));
921 922 923

	switch (rtn) {
	case NEEDS_RETRY:
L
Linus Torvalds 已提交
924 925
		if (retry_cnt--)
			goto retry_tur;
926 927
		/*FALLTHRU*/
	case SUCCESS:
928
		return 0;
929 930
	default:
		return 1;
931
	}
L
Linus Torvalds 已提交
932 933 934
}

/**
935 936 937
 * scsi_eh_abort_cmds - abort pending commands.
 * @work_q:	&list_head for pending commands.
 * @done_q:	&list_head for processed commands.
L
Linus Torvalds 已提交
938 939 940
 *
 * Decription:
 *    Try and see whether or not it makes sense to try and abort the
941 942
 *    running command.  This only works out to be the case if we have one
 *    command that has timed out.  If the command simply failed, it makes
L
Linus Torvalds 已提交
943 944
 *    no sense to try and abort the command, since as far as the shost
 *    adapter is concerned, it isn't running.
945
 */
L
Linus Torvalds 已提交
946 947 948
static int scsi_eh_abort_cmds(struct list_head *work_q,
			      struct list_head *done_q)
{
949
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
950 951
	int rtn;

952
	list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
953
		if (!(scmd->eh_eflags & SCSI_EH_CANCEL_CMD))
L
Linus Torvalds 已提交
954 955 956 957 958 959
			continue;
		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: aborting cmd:"
						  "0x%p\n", current->comm,
						  scmd));
		rtn = scsi_try_to_abort_cmd(scmd);
		if (rtn == SUCCESS) {
960
			scmd->eh_eflags &= ~SCSI_EH_CANCEL_CMD;
L
Linus Torvalds 已提交
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978
			if (!scsi_device_online(scmd->device) ||
			    !scsi_eh_tur(scmd)) {
				scsi_eh_finish_cmd(scmd, done_q);
			}
				
		} else
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: aborting"
							  " cmd failed:"
							  "0x%p\n",
							  current->comm,
							  scmd));
	}

	return list_empty(work_q);
}

/**
 * scsi_eh_try_stu - Send START_UNIT to device.
979
 * @scmd:	&scsi_cmnd to send START_UNIT
L
Linus Torvalds 已提交
980 981 982
 *
 * Return value:
 *    0 - Device is ready. 1 - Device NOT ready.
983
 */
L
Linus Torvalds 已提交
984 985 986 987
static int scsi_eh_try_stu(struct scsi_cmnd *scmd)
{
	static unsigned char stu_command[6] = {START_STOP, 0, 0, 0, 1, 0};

988
	if (scmd->device->allow_restart) {
989 990 991
		int i, rtn = NEEDS_RETRY;

		for (i = 0; rtn == NEEDS_RETRY && i < 2; i++)
992
			rtn = scsi_send_eh_cmnd(scmd, stu_command, 6, scmd->device->request_queue->rq_timeout, 0);
L
Linus Torvalds 已提交
993

994 995 996
		if (rtn == SUCCESS)
			return 0;
	}
L
Linus Torvalds 已提交
997 998 999 1000 1001 1002

	return 1;
}

 /**
 * scsi_eh_stu - send START_UNIT if needed
1003 1004 1005
 * @shost:	&scsi host being recovered.
 * @work_q:     &list_head for pending commands.
 * @done_q:	&list_head for processed commands.
L
Linus Torvalds 已提交
1006 1007 1008 1009
 *
 * Notes:
 *    If commands are failing due to not ready, initializing command required,
 *	try revalidating the device, which will end up sending a start unit. 
1010
 */
L
Linus Torvalds 已提交
1011 1012 1013 1014
static int scsi_eh_stu(struct Scsi_Host *shost,
			      struct list_head *work_q,
			      struct list_head *done_q)
{
1015
	struct scsi_cmnd *scmd, *stu_scmd, *next;
L
Linus Torvalds 已提交
1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
	struct scsi_device *sdev;

	shost_for_each_device(sdev, shost) {
		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;

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending START_UNIT to sdev:"
						  " 0x%p\n", current->comm, sdev));

		if (!scsi_eh_try_stu(stu_scmd)) {
			if (!scsi_device_online(sdev) ||
			    !scsi_eh_tur(stu_scmd)) {
1036 1037
				list_for_each_entry_safe(scmd, next,
							  work_q, eh_entry) {
L
Linus Torvalds 已提交
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
					if (scmd->device == sdev)
						scsi_eh_finish_cmd(scmd, done_q);
				}
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3,
						printk("%s: START_UNIT failed to sdev:"
						       " 0x%p\n", current->comm, sdev));
		}
	}

	return list_empty(work_q);
}


/**
 * scsi_eh_bus_device_reset - send bdr if needed
 * @shost:	scsi host being recovered.
1056 1057
 * @work_q:     &list_head for pending commands.
 * @done_q:	&list_head for processed commands.
L
Linus Torvalds 已提交
1058 1059
 *
 * Notes:
1060
 *    Try a bus device reset.  Still, look to see whether we have multiple
L
Linus Torvalds 已提交
1061 1062 1063
 *    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
 *    a bus_reset instead. 
1064
 */
L
Linus Torvalds 已提交
1065 1066 1067 1068
static int scsi_eh_bus_device_reset(struct Scsi_Host *shost,
				    struct list_head *work_q,
				    struct list_head *done_q)
{
1069
	struct scsi_cmnd *scmd, *bdr_scmd, *next;
L
Linus Torvalds 已提交
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
	struct scsi_device *sdev;
	int rtn;

	shost_for_each_device(sdev, shost) {
		bdr_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry)
			if (scmd->device == sdev) {
				bdr_scmd = scmd;
				break;
			}

		if (!bdr_scmd)
			continue;

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending BDR sdev:"
						  " 0x%p\n", current->comm,
						  sdev));
		rtn = scsi_try_bus_device_reset(bdr_scmd);
		if (rtn == SUCCESS) {
			if (!scsi_device_online(sdev) ||
			    !scsi_eh_tur(bdr_scmd)) {
1091 1092
				list_for_each_entry_safe(scmd, next,
							 work_q, eh_entry) {
L
Linus Torvalds 已提交
1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
					if (scmd->device == sdev)
						scsi_eh_finish_cmd(scmd,
								   done_q);
				}
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: BDR"
							  " failed sdev:"
							  "0x%p\n",
							  current->comm,
							   sdev));
		}
	}

	return list_empty(work_q);
}

1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
/**
 * scsi_eh_target_reset - send target reset if needed
 * @shost:	scsi host being recovered.
 * @work_q:     &list_head for pending commands.
 * @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)
{
	struct scsi_cmnd *scmd, *tgtr_scmd, *next;
1124
	unsigned int id = 0;
1125 1126
	int rtn;

1127
	do {
1128 1129 1130 1131 1132 1133 1134
		tgtr_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry) {
			if (id == scmd_id(scmd)) {
				tgtr_scmd = scmd;
				break;
			}
		}
1135 1136 1137 1138 1139 1140 1141 1142 1143
		if (!tgtr_scmd) {
			/* not one exactly equal; find the next highest */
			list_for_each_entry(scmd, work_q, eh_entry) {
				if (scmd_id(scmd) > id &&
				    (!tgtr_scmd ||
				     scmd_id(tgtr_scmd) > scmd_id(scmd)))
						tgtr_scmd = scmd;
			}
		}
1144
		if (!tgtr_scmd)
1145 1146
			/* no more commands, that's it */
			break;
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending target reset "
						  "to target %d\n",
						  current->comm, id));
		rtn = scsi_try_target_reset(tgtr_scmd);
		if (rtn == SUCCESS) {
			list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
				if (id == scmd_id(scmd))
					if (!scsi_device_online(scmd->device) ||
					    !scsi_eh_tur(tgtr_scmd))
						scsi_eh_finish_cmd(scmd,
								   done_q);
			}
		} else
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Target reset"
							  " failed target: "
							  "%d\n",
							  current->comm, id));
1165 1166
		id++;
	} while(id != 0);
1167 1168 1169 1170

	return list_empty(work_q);
}

L
Linus Torvalds 已提交
1171 1172
/**
 * scsi_eh_bus_reset - send a bus reset 
1173 1174 1175
 * @shost:	&scsi host being recovered.
 * @work_q:     &list_head for pending commands.
 * @done_q:	&list_head for processed commands.
1176
 */
L
Linus Torvalds 已提交
1177 1178 1179 1180
static int scsi_eh_bus_reset(struct Scsi_Host *shost,
			     struct list_head *work_q,
			     struct list_head *done_q)
{
1181
	struct scsi_cmnd *scmd, *chan_scmd, *next;
L
Linus Torvalds 已提交
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
	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
	 * the reset.  
	 */

	for (channel = 0; channel <= shost->max_channel; channel++) {
		chan_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry) {
1195
			if (channel == scmd_channel(scmd)) {
L
Linus Torvalds 已提交
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
				chan_scmd = scmd;
				break;
				/*
				 * FIXME add back in some support for
				 * soft_reset devices.
				 */
			}
		}

		if (!chan_scmd)
			continue;
		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending BRST chan:"
						  " %d\n", current->comm,
						  channel));
		rtn = scsi_try_bus_reset(chan_scmd);
		if (rtn == SUCCESS) {
1212
			list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
1213
				if (channel == scmd_channel(scmd))
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232
					if (!scsi_device_online(scmd->device) ||
					    !scsi_eh_tur(scmd))
						scsi_eh_finish_cmd(scmd,
								   done_q);
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: BRST"
							  " failed chan: %d\n",
							  current->comm,
							  channel));
		}
	}
	return list_empty(work_q);
}

/**
 * scsi_eh_host_reset - send a host reset 
 * @work_q:	list_head for processed commands.
 * @done_q:	list_head for processed commands.
1233
 */
L
Linus Torvalds 已提交
1234 1235 1236
static int scsi_eh_host_reset(struct list_head *work_q,
			      struct list_head *done_q)
{
1237
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
	int rtn;

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

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending HRST\n"
						  , current->comm));

		rtn = scsi_try_host_reset(scmd);
		if (rtn == SUCCESS) {
1249
			list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
L
Linus Torvalds 已提交
1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
				if (!scsi_device_online(scmd->device) ||
				    (!scsi_eh_try_stu(scmd) && !scsi_eh_tur(scmd)) ||
				    !scsi_eh_tur(scmd))
					scsi_eh_finish_cmd(scmd, done_q);
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: HRST"
							  " failed\n",
							  current->comm));
		}
	}
	return list_empty(work_q);
}

/**
 * scsi_eh_offline_sdevs - offline scsi devices that fail to recover
 * @work_q:	list_head for processed commands.
 * @done_q:	list_head for processed commands.
1268
 */
L
Linus Torvalds 已提交
1269 1270 1271
static void scsi_eh_offline_sdevs(struct list_head *work_q,
				  struct list_head *done_q)
{
1272
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
1273

1274
	list_for_each_entry_safe(scmd, next, work_q, eh_entry) {
1275 1276
		sdev_printk(KERN_INFO, scmd->device, "Device offlined - "
			    "not ready after error recovery\n");
L
Linus Torvalds 已提交
1277
		scsi_device_set_state(scmd->device, SDEV_OFFLINE);
1278
		if (scmd->eh_eflags & SCSI_EH_CANCEL_CMD) {
L
Linus Torvalds 已提交
1279 1280 1281 1282 1283 1284 1285 1286 1287
			/*
			 * FIXME: Handle lost cmds.
			 */
		}
		scsi_eh_finish_cmd(scmd, done_q);
	}
	return;
}

1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
/**
 * scsi_noretry_cmd - determinte if command should be failed fast
 * @scmd:	SCSI cmd to examine.
 */
int scsi_noretry_cmd(struct scsi_cmnd *scmd)
{
	switch (host_byte(scmd->result)) {
	case DID_OK:
		break;
	case DID_BUS_BUSY:
		return blk_failfast_transport(scmd->request);
	case DID_PARITY:
		return blk_failfast_dev(scmd->request);
	case DID_ERROR:
		if (msg_byte(scmd->result) == COMMAND_COMPLETE &&
		    status_byte(scmd->result) == RESERVATION_CONFLICT)
			return 0;
		/* fall through */
	case DID_SOFT_ERROR:
		return blk_failfast_driver(scmd->request);
	}

	switch (status_byte(scmd->result)) {
	case CHECK_CONDITION:
		/*
		 * assume caller has checked sense and determinted
		 * the check condition was retryable.
		 */
		return blk_failfast_dev(scmd->request);
	}

	return 0;
}

L
Linus Torvalds 已提交
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
/**
 * 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.
1335
 */
L
Linus Torvalds 已提交
1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
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)) {
		SCSI_LOG_ERROR_RECOVERY(5, printk("%s: device offline - report"
						  " as SUCCESS\n",
1347
						  __func__));
L
Linus Torvalds 已提交
1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391
		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;
	case DID_NO_CONNECT:
	case DID_BAD_TARGET:
	case DID_ABORT:
		/*
		 * 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;
		/*
		 * when the low level driver returns did_soft_error,
		 * it is responsible for keeping an internal retry counter 
		 * in order to avoid endless loops (db)
		 *
		 * actually this is a bug in this function here.  we should
		 * be mindful of the maximum number of retries specified
		 * and not get stuck in a loop.
		 */
	case DID_SOFT_ERROR:
		goto maybe_retry;
	case DID_IMM_RETRY:
		return NEEDS_RETRY;

已提交
1392 1393
	case DID_REQUEUE:
		return ADD_TO_MLQUEUE;
1394 1395 1396 1397 1398
	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
1399 1400
		 * based on its timers and recovery capablilities if
		 * there are enough retries.
1401
		 */
1402
		goto maybe_retry;
1403 1404 1405 1406 1407 1408
	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
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	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:
1451
		scsi_handle_queue_full(scmd->device);
L
Linus Torvalds 已提交
1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
		/*
		 * the case of trying to send too many commands to a
		 * tagged queueing device.
		 */
	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:
V
Vasu Dev 已提交
1465
		scsi_handle_queue_ramp_up(scmd->device);
L
Linus Torvalds 已提交
1466 1467
	case COMMAND_TERMINATED:
		return SUCCESS;
1468 1469
	case TASK_ABORTED:
		goto maybe_retry;
L
Linus Torvalds 已提交
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
	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:
1489 1490
		sdev_printk(KERN_INFO, scmd->device,
			    "reservation conflict\n");
L
Linus Torvalds 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
		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) */
1503
	if ((++scmd->retries) <= scmd->allowed
1504
	    && !scsi_noretry_cmd(scmd)) {
L
Linus Torvalds 已提交
1505 1506 1507 1508 1509 1510 1511 1512 1513
		return NEEDS_RETRY;
	} else {
		/*
		 * no more retries - report this one back to upper level.
		 */
		return SUCCESS;
	}
}

F
FUJITA Tomonori 已提交
1514 1515 1516 1517 1518
static void eh_lock_door_done(struct request *req, int uptodate)
{
	__blk_put_request(req->q, req);
}

L
Linus Torvalds 已提交
1519 1520 1521 1522 1523
/**
 * scsi_eh_lock_door - Prevent medium removal for the specified device
 * @sdev:	SCSI device to prevent medium removal
 *
 * Locking:
1524
 * 	We must be called from process context.
L
Linus Torvalds 已提交
1525 1526 1527 1528
 *
 * Notes:
 * 	We queue up an asynchronous "ALLOW MEDIUM REMOVAL" request on the
 * 	head of the devices request queue, and continue.
1529
 */
L
Linus Torvalds 已提交
1530 1531
static void scsi_eh_lock_door(struct scsi_device *sdev)
{
F
FUJITA Tomonori 已提交
1532
	struct request *req;
L
Linus Torvalds 已提交
1533

1534 1535 1536 1537
	/*
	 * blk_get_request with GFP_KERNEL (__GFP_WAIT) sleeps until a
	 * request becomes available
	 */
F
FUJITA Tomonori 已提交
1538
	req = blk_get_request(sdev->request_queue, READ, GFP_KERNEL);
L
Linus Torvalds 已提交
1539

F
FUJITA Tomonori 已提交
1540 1541 1542 1543 1544 1545
	req->cmd[0] = ALLOW_MEDIUM_REMOVAL;
	req->cmd[1] = 0;
	req->cmd[2] = 0;
	req->cmd[3] = 0;
	req->cmd[4] = SCSI_REMOVAL_PREVENT;
	req->cmd[5] = 0;
L
Linus Torvalds 已提交
1546

F
FUJITA Tomonori 已提交
1547 1548 1549 1550 1551 1552 1553 1554 1555
	req->cmd_len = COMMAND_SIZE(req->cmd[0]);

	req->cmd_type = REQ_TYPE_BLOCK_PC;
	req->cmd_flags |= REQ_QUIET;
	req->timeout = 10 * HZ;
	req->retries = 5;

	blk_execute_rq_nowait(req->q, NULL, req, 1, eh_lock_door_done);
}
L
Linus Torvalds 已提交
1556 1557 1558 1559 1560 1561 1562 1563

/**
 * 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.
1564
 */
L
Linus Torvalds 已提交
1565 1566 1567
static void scsi_restart_operations(struct Scsi_Host *shost)
{
	struct scsi_device *sdev;
1568
	unsigned long flags;
L
Linus Torvalds 已提交
1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585

	/*
	 * 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) {
		if (scsi_device_online(sdev) && sdev->locked)
			scsi_eh_lock_door(sdev);
	}

	/*
	 * 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.
	 */
	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: waking up host to restart\n",
1586
					  __func__));
L
Linus Torvalds 已提交
1587

1588 1589 1590 1591 1592
	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
1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607

	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);
}

/**
 * scsi_eh_ready_devs - check device ready state and recover if not.
 * @shost: 	host to be recovered.
1608 1609
 * @work_q:     &list_head for pending commands.
 * @done_q:	&list_head for processed commands.
1610
 */
1611 1612 1613
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
1614 1615 1616
{
	if (!scsi_eh_stu(shost, work_q, done_q))
		if (!scsi_eh_bus_device_reset(shost, work_q, done_q))
1617 1618 1619 1620 1621
			if (!scsi_eh_target_reset(shost, work_q, done_q))
				if (!scsi_eh_bus_reset(shost, work_q, done_q))
					if (!scsi_eh_host_reset(work_q, done_q))
						scsi_eh_offline_sdevs(work_q,
								      done_q);
L
Linus Torvalds 已提交
1622
}
1623
EXPORT_SYMBOL_GPL(scsi_eh_ready_devs);
L
Linus Torvalds 已提交
1624 1625 1626 1627

/**
 * scsi_eh_flush_done_q - finish processed commands or retry them.
 * @done_q:	list_head of processed commands.
1628
 */
1629
void scsi_eh_flush_done_q(struct list_head *done_q)
L
Linus Torvalds 已提交
1630
{
1631
	struct scsi_cmnd *scmd, *next;
L
Linus Torvalds 已提交
1632

1633 1634
	list_for_each_entry_safe(scmd, next, done_q, eh_entry) {
		list_del_init(&scmd->eh_entry);
L
Linus Torvalds 已提交
1635
		if (scsi_device_online(scmd->device) &&
1636
		    !scsi_noretry_cmd(scmd) &&
1637
		    (++scmd->retries <= scmd->allowed)) {
L
Linus Torvalds 已提交
1638 1639 1640 1641 1642 1643
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: flush"
							  " retry cmd: %p\n",
							  current->comm,
							  scmd));
				scsi_queue_insert(scmd, SCSI_MLQUEUE_EH_RETRY);
		} else {
1644 1645 1646 1647 1648
			/*
			 * 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
1649 1650 1651 1652 1653 1654 1655 1656 1657
			if (!scmd->result)
				scmd->result |= (DRIVER_TIMEOUT << 24);
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: flush finish"
							" cmd: %p\n",
							current->comm, scmd));
			scsi_finish_command(scmd);
		}
	}
}
1658
EXPORT_SYMBOL(scsi_eh_flush_done_q);
L
Linus Torvalds 已提交
1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681

/**
 * 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.
1682
 */
L
Linus Torvalds 已提交
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
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))
		if (!scsi_eh_abort_cmds(&eh_work_q, &eh_done_q))
			scsi_eh_ready_devs(shost, &eh_work_q, &eh_done_q);

	scsi_eh_flush_done_q(&eh_done_q);
}

/**
1703
 * scsi_error_handler - SCSI error handler thread
L
Linus Torvalds 已提交
1704 1705 1706
 * @data:	Host for which we are running.
 *
 * Notes:
1707 1708
 *    This is the main error handling loop.  This is run as a kernel thread
 *    for every SCSI host and handles all error handling activity.
1709
 */
L
Linus Torvalds 已提交
1710 1711
int scsi_error_handler(void *data)
{
1712
	struct Scsi_Host *shost = data;
L
Linus Torvalds 已提交
1713 1714

	/*
1715 1716 1717
	 * 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
1718
	 * disables signal delivery for the created thread.
L
Linus Torvalds 已提交
1719
	 */
1720 1721
	set_current_state(TASK_INTERRUPTIBLE);
	while (!kthread_should_stop()) {
1722
		if ((shost->host_failed == 0 && shost->host_eh_scheduled == 0) ||
1723
		    shost->host_failed != shost->host_busy) {
1724 1725 1726
			SCSI_LOG_ERROR_RECOVERY(1,
				printk("Error handler scsi_eh_%d sleeping\n",
					shost->host_no));
1727 1728 1729 1730
			schedule();
			set_current_state(TASK_INTERRUPTIBLE);
			continue;
		}
L
Linus Torvalds 已提交
1731

1732
		__set_current_state(TASK_RUNNING);
1733 1734 1735
		SCSI_LOG_ERROR_RECOVERY(1,
			printk("Error handler scsi_eh_%d waking up\n",
				shost->host_no));
L
Linus Torvalds 已提交
1736 1737 1738 1739 1740 1741

		/*
		 * 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.
		 */
1742 1743
		if (shost->transportt->eh_strategy_handler)
			shost->transportt->eh_strategy_handler(shost);
L
Linus Torvalds 已提交
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
		else
			scsi_unjam_host(shost);

		/*
		 * 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);
1755
		set_current_state(TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
1756
	}
1757 1758
	__set_current_state(TASK_RUNNING);

1759 1760
	SCSI_LOG_ERROR_RECOVERY(1,
		printk("Error handler scsi_eh_%d exiting\n", shost->host_no));
1761
	shost->ehandler = NULL;
L
Linus Torvalds 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
	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) {
1791 1792
		if (channel == sdev_channel(sdev))
			__scsi_report_device_reset(sdev, NULL);
L
Linus Torvalds 已提交
1793 1794 1795 1796 1797 1798 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
	}
}
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) {
1824
		if (channel == sdev_channel(sdev) &&
1825 1826
		    target == sdev_id(sdev))
			__scsi_report_device_reset(sdev, NULL);
L
Linus Torvalds 已提交
1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
	}
}
EXPORT_SYMBOL(scsi_report_device_reset);

static void
scsi_reset_provider_done_command(struct scsi_cmnd *scmd)
{
}

/*
 * Function:	scsi_reset_provider
 *
 * Purpose:	Send requested reset to a bus or device at any phase.
 *
 * Arguments:	device	- device to send reset to
 *		flag - reset type (see scsi.h)
 *
 * Returns:	SUCCESS/FAILURE.
 *
 * Notes:	This is used by the SCSI Generic driver to provide
 *		Bus/Device reset capability.
 */
int
scsi_reset_provider(struct scsi_device *dev, int flag)
{
	struct scsi_cmnd *scmd = scsi_get_command(dev, GFP_KERNEL);
1853
	struct Scsi_Host *shost = dev->host;
L
Linus Torvalds 已提交
1854
	struct request req;
1855
	unsigned long flags;
L
Linus Torvalds 已提交
1856 1857
	int rtn;

1858
	blk_rq_init(NULL, &req);
L
Linus Torvalds 已提交
1859
	scmd->request = &req;
1860

1861 1862
	scmd->cmnd = req.cmd;

L
Linus Torvalds 已提交
1863
	scmd->scsi_done		= scsi_reset_provider_done_command;
B
Boaz Harrosh 已提交
1864
	memset(&scmd->sdb, 0, sizeof(scmd->sdb));
L
Linus Torvalds 已提交
1865 1866 1867 1868 1869

	scmd->cmd_len			= 0;

	scmd->sc_data_direction		= DMA_BIDIRECTIONAL;

1870 1871 1872 1873
	spin_lock_irqsave(shost->host_lock, flags);
	shost->tmf_in_progress = 1;
	spin_unlock_irqrestore(shost->host_lock, flags);

L
Linus Torvalds 已提交
1874 1875 1876 1877 1878 1879
	switch (flag) {
	case SCSI_TRY_RESET_DEVICE:
		rtn = scsi_try_bus_device_reset(scmd);
		if (rtn == SUCCESS)
			break;
		/* FALLTHROUGH */
1880 1881 1882 1883 1884
	case SCSI_TRY_RESET_TARGET:
		rtn = scsi_try_target_reset(scmd);
		if (rtn == SUCCESS)
			break;
		/* FALLTHROUGH */
L
Linus Torvalds 已提交
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896
	case SCSI_TRY_RESET_BUS:
		rtn = scsi_try_bus_reset(scmd);
		if (rtn == SUCCESS)
			break;
		/* FALLTHROUGH */
	case SCSI_TRY_RESET_HOST:
		rtn = scsi_try_host_reset(scmd);
		break;
	default:
		rtn = FAILED;
	}

1897 1898 1899 1900 1901 1902 1903 1904 1905 1906
	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,
		printk("%s: waking up host to restart after TMF\n",
1907
		__func__));
1908 1909 1910 1911 1912

	wake_up(&shost->host_wait);

	scsi_run_host_queues(shost);

L
Linus Torvalds 已提交
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935
	scsi_next_command(scmd);
	return rtn;
}
EXPORT_SYMBOL(scsi_reset_provider);

/**
 * scsi_normalize_sense - normalize main elements from either fixed or
 *			descriptor sense data format into a common format.
 *
 * @sense_buffer:	byte array containing sense data returned by device
 * @sb_len:		number of valid bytes in sense_buffer
 * @sshdr:		pointer to instance of structure that common
 *			elements are written to.
 *
 * Notes:
 *	The "main elements" from sense data are: response_code, sense_key,
 *	asc, ascq and additional_length (only for descriptor format).
 *
 *	Typically this function can be called after a device has
 *	responded to a SCSI command with the CHECK_CONDITION status.
 *
 * Return value:
 *	1 if valid sense data information found, else 0;
1936
 */
L
Linus Torvalds 已提交
1937 1938 1939
int scsi_normalize_sense(const u8 *sense_buffer, int sb_len,
                         struct scsi_sense_hdr *sshdr)
{
1940
	if (!sense_buffer || !sb_len)
L
Linus Torvalds 已提交
1941 1942 1943 1944 1945
		return 0;

	memset(sshdr, 0, sizeof(struct scsi_sense_hdr));

	sshdr->response_code = (sense_buffer[0] & 0x7f);
1946 1947 1948 1949

	if (!scsi_sense_valid(sshdr))
		return 0;

L
Linus Torvalds 已提交
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
	if (sshdr->response_code >= 0x72) {
		/*
		 * descriptor format
		 */
		if (sb_len > 1)
			sshdr->sense_key = (sense_buffer[1] & 0xf);
		if (sb_len > 2)
			sshdr->asc = sense_buffer[2];
		if (sb_len > 3)
			sshdr->ascq = sense_buffer[3];
		if (sb_len > 7)
			sshdr->additional_length = sense_buffer[7];
	} else {
		/* 
		 * fixed format
		 */
		if (sb_len > 2)
			sshdr->sense_key = (sense_buffer[2] & 0xf);
		if (sb_len > 7) {
			sb_len = (sb_len < (sense_buffer[7] + 8)) ?
					 sb_len : (sense_buffer[7] + 8);
			if (sb_len > 12)
				sshdr->asc = sense_buffer[12];
			if (sb_len > 13)
				sshdr->ascq = sense_buffer[13];
		}
	}

	return 1;
}
EXPORT_SYMBOL(scsi_normalize_sense);

int scsi_command_normalize_sense(struct scsi_cmnd *cmd,
				 struct scsi_sense_hdr *sshdr)
{
	return scsi_normalize_sense(cmd->sense_buffer,
1986
			SCSI_SENSE_BUFFERSIZE, sshdr);
L
Linus Torvalds 已提交
1987 1988 1989 1990
}
EXPORT_SYMBOL(scsi_command_normalize_sense);

/**
1991
 * scsi_sense_desc_find - search for a given descriptor type in	descriptor sense data format.
L
Linus Torvalds 已提交
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
 * @sense_buffer:	byte array of descriptor format sense data
 * @sb_len:		number of valid bytes in sense_buffer
 * @desc_type:		value of descriptor type to find
 *			(e.g. 0 -> information)
 *
 * Notes:
 *	only valid when sense data is in descriptor format
 *
 * Return value:
 *	pointer to start of (first) descriptor if found else NULL
2002
 */
L
Linus Torvalds 已提交
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029
const u8 * scsi_sense_desc_find(const u8 * sense_buffer, int sb_len,
				int desc_type)
{
	int add_sen_len, add_len, desc_len, k;
	const u8 * descp;

	if ((sb_len < 8) || (0 == (add_sen_len = sense_buffer[7])))
		return NULL;
	if ((sense_buffer[0] < 0x72) || (sense_buffer[0] > 0x73))
		return NULL;
	add_sen_len = (add_sen_len < (sb_len - 8)) ?
			add_sen_len : (sb_len - 8);
	descp = &sense_buffer[8];
	for (desc_len = 0, k = 0; k < add_sen_len; k += desc_len) {
		descp += desc_len;
		add_len = (k < (add_sen_len - 1)) ? descp[1]: -1;
		desc_len = add_len + 2;
		if (descp[0] == desc_type)
			return descp;
		if (add_len < 0) // short descriptor ??
			break;
	}
	return NULL;
}
EXPORT_SYMBOL(scsi_sense_desc_find);

/**
2030
 * scsi_get_sense_info_fld - get information field from sense data (either fixed or descriptor format)
L
Linus Torvalds 已提交
2031 2032 2033 2034 2035 2036 2037
 * @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:
 *	1 if information field found, 0 if not found.
2038
 */
L
Linus Torvalds 已提交
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
int scsi_get_sense_info_fld(const u8 * sense_buffer, int sb_len,
			    u64 * info_out)
{
	int j;
	const u8 * ucp;
	u64 ull;

	if (sb_len < 7)
		return 0;
	switch (sense_buffer[0] & 0x7f) {
	case 0x70:
	case 0x71:
		if (sense_buffer[0] & 0x80) {
			*info_out = (sense_buffer[3] << 24) +
				    (sense_buffer[4] << 16) +
				    (sense_buffer[5] << 8) + sense_buffer[6];
			return 1;
		} else
			return 0;
	case 0x72:
	case 0x73:
		ucp = scsi_sense_desc_find(sense_buffer, sb_len,
					   0 /* info desc */);
		if (ucp && (0xa == ucp[1])) {
			ull = 0;
			for (j = 0; j < 8; ++j) {
				if (j > 0)
					ull <<= 8;
				ull |= ucp[4 + j];
			}
			*info_out = ull;
			return 1;
		} else
			return 0;
	default:
		return 0;
	}
}
EXPORT_SYMBOL(scsi_get_sense_info_fld);
2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105

/**
 * scsi_build_sense_buffer - build sense data in a buffer
 * @desc:	Sense format (non zero == descriptor format,
 * 		0 == fixed format)
 * @buf:	Where to build sense data
 * @key:	Sense key
 * @asc:	Additional sense code
 * @ascq:	Additional sense code qualifier
 *
 **/
void scsi_build_sense_buffer(int desc, u8 *buf, u8 key, u8 asc, u8 ascq)
{
	if (desc) {
		buf[0] = 0x72;	/* descriptor, current */
		buf[1] = key;
		buf[2] = asc;
		buf[3] = ascq;
		buf[7] = 0;
	} else {
		buf[0] = 0x70;	/* fixed, current */
		buf[2] = key;
		buf[7] = 0xa;
		buf[12] = asc;
		buf[13] = ascq;
	}
}
EXPORT_SYMBOL(scsi_build_sense_buffer);