libata-eh.c 58.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
/*
 *  libata-eh.c - libata error handling
 *
 *  Maintained by:  Jeff Garzik <jgarzik@pobox.com>
 *    		    Please ALWAYS copy linux-ide@vger.kernel.org
 *		    on emails.
 *
 *  Copyright 2006 Tejun Heo <htejun@gmail.com>
 *
 *
 *  This program is free software; you can redistribute it and/or
 *  modify it under the terms of the GNU General Public License as
 *  published by the Free Software Foundation; either version 2, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 *  General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; see the file COPYING.  If not, write to
 *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139,
 *  USA.
 *
 *
 *  libata documentation is available via 'make {ps|pdf}docs',
 *  as Documentation/DocBook/libata.*
 *
 *  Hardware documentation available from http://www.t13.org/ and
 *  http://www.sata-io.org/
 *
 */

#include <linux/kernel.h>
#include <scsi/scsi.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_cmnd.h>
J
Jeff Garzik 已提交
41
#include "../scsi/scsi_transport_api.h"
42 43 44 45 46

#include <linux/libata.h>

#include "libata.h"

47 48 49 50 51 52
enum {
	ATA_EH_SPDN_NCQ_OFF		= (1 << 0),
	ATA_EH_SPDN_SPEED_DOWN		= (1 << 1),
	ATA_EH_SPDN_FALLBACK_TO_PIO	= (1 << 2),
};

T
Tejun Heo 已提交
53 54 55 56 57 58
/* Waiting in ->prereset can never be reliable.  It's sometimes nice
 * to wait there but it can't be depended upon; otherwise, we wouldn't
 * be resetting.  Just give it enough time for most drives to spin up.
 */
enum {
	ATA_EH_PRERESET_TIMEOUT		= 10 * HZ,
T
Tejun Heo 已提交
59
	ATA_EH_FASTDRAIN_INTERVAL	= 3 * HZ,
T
Tejun Heo 已提交
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
};

/* The following table determines how we sequence resets.  Each entry
 * represents timeout for that try.  The first try can be soft or
 * hardreset.  All others are hardreset if available.  In most cases
 * the first reset w/ 10sec timeout should succeed.  Following entries
 * are mostly for error handling, hotplug and retarded devices.
 */
static const unsigned long ata_eh_reset_timeouts[] = {
	10 * HZ,	/* most drives spin up by 10sec */
	10 * HZ,	/* > 99% working drives spin up before 20sec */
	35 * HZ,	/* give > 30 secs of idleness for retarded devices */
	5 * HZ,		/* and sweet one last chance */
	/* > 1 min has elapsed, give up */
};

76
static void __ata_port_freeze(struct ata_port *ap);
77
static void ata_eh_finish(struct ata_port *ap);
78
#ifdef CONFIG_PM
79 80
static void ata_eh_handle_port_suspend(struct ata_port *ap);
static void ata_eh_handle_port_resume(struct ata_port *ap);
81 82 83 84 85 86 87
#else /* CONFIG_PM */
static void ata_eh_handle_port_suspend(struct ata_port *ap)
{ }

static void ata_eh_handle_port_resume(struct ata_port *ap)
{ }
#endif /* CONFIG_PM */
88

T
Tejun Heo 已提交
89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153
static void __ata_ehi_pushv_desc(struct ata_eh_info *ehi, const char *fmt,
				 va_list args)
{
	ehi->desc_len += vscnprintf(ehi->desc + ehi->desc_len,
				     ATA_EH_DESC_LEN - ehi->desc_len,
				     fmt, args);
}

/**
 *	__ata_ehi_push_desc - push error description without adding separator
 *	@ehi: target EHI
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to @ehi->desc.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void __ata_ehi_push_desc(struct ata_eh_info *ehi, const char *fmt, ...)
{
	va_list args;

	va_start(args, fmt);
	__ata_ehi_pushv_desc(ehi, fmt, args);
	va_end(args);
}

/**
 *	ata_ehi_push_desc - push error description with separator
 *	@ehi: target EHI
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to @ehi->desc.
 *	If @ehi->desc is not empty, ", " is added in-between.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_ehi_push_desc(struct ata_eh_info *ehi, const char *fmt, ...)
{
	va_list args;

	if (ehi->desc_len)
		__ata_ehi_push_desc(ehi, ", ");

	va_start(args, fmt);
	__ata_ehi_pushv_desc(ehi, fmt, args);
	va_end(args);
}

/**
 *	ata_ehi_clear_desc - clean error description
 *	@ehi: target EHI
 *
 *	Clear @ehi->desc.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
void ata_ehi_clear_desc(struct ata_eh_info *ehi)
{
	ehi->desc[0] = '\0';
	ehi->desc_len = 0;
}

154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
static void ata_ering_record(struct ata_ering *ering, int is_io,
			     unsigned int err_mask)
{
	struct ata_ering_entry *ent;

	WARN_ON(!err_mask);

	ering->cursor++;
	ering->cursor %= ATA_ERING_SIZE;

	ent = &ering->ring[ering->cursor];
	ent->is_io = is_io;
	ent->err_mask = err_mask;
	ent->timestamp = get_jiffies_64();
}

170
static void ata_ering_clear(struct ata_ering *ering)
171
{
172
	memset(ering, 0, sizeof(*ering));
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
}

static int ata_ering_map(struct ata_ering *ering,
			 int (*map_fn)(struct ata_ering_entry *, void *),
			 void *arg)
{
	int idx, rc = 0;
	struct ata_ering_entry *ent;

	idx = ering->cursor;
	do {
		ent = &ering->ring[idx];
		if (!ent->err_mask)
			break;
		rc = map_fn(ent, arg);
		if (rc)
			break;
		idx = (idx - 1 + ATA_ERING_SIZE) % ATA_ERING_SIZE;
	} while (idx != ering->cursor);

	return rc;
}

196 197
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
198
	struct ata_eh_context *ehc = &dev->link->eh_context;
199 200 201 202

	return ehc->i.action | ehc->i.dev_action[dev->devno];
}

203
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
204 205
				struct ata_eh_info *ehi, unsigned int action)
{
206
	struct ata_device *tdev;
207 208 209

	if (!dev) {
		ehi->action &= ~action;
210 211
		ata_link_for_each_dev(tdev, link)
			ehi->dev_action[tdev->devno] &= ~action;
212 213 214 215 216 217
	} else {
		/* doesn't make sense for port-wide EH actions */
		WARN_ON(!(action & ATA_EH_PERDEV_MASK));

		/* break ehi->action into ehi->dev_action */
		if (ehi->action & action) {
218 219 220
			ata_link_for_each_dev(tdev, link)
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
221 222 223 224 225 226 227 228
			ehi->action &= ~action;
		}

		/* turn off the specified per-dev action */
		ehi->dev_action[dev->devno] &= ~action;
	}
}

229 230 231 232 233 234 235 236 237 238 239
/**
 *	ata_scsi_timed_out - SCSI layer time out callback
 *	@cmd: timed out SCSI command
 *
 *	Handles SCSI layer timeout.  We race with normal completion of
 *	the qc for @cmd.  If the qc is already gone, we lose and let
 *	the scsi command finish (EH_HANDLED).  Otherwise, the qc has
 *	timed out and EH should be invoked.  Prevent ata_qc_complete()
 *	from finishing it by setting EH_SCHEDULED and return
 *	EH_NOT_HANDLED.
 *
240 241
 *	TODO: kill this function once old EH is gone.
 *
242 243 244 245 246 247 248 249 250
 *	LOCKING:
 *	Called from timer context
 *
 *	RETURNS:
 *	EH_HANDLED or EH_NOT_HANDLED
 */
enum scsi_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
{
	struct Scsi_Host *host = cmd->device->host;
251
	struct ata_port *ap = ata_shost_to_port(host);
252 253
	unsigned long flags;
	struct ata_queued_cmd *qc;
254
	enum scsi_eh_timer_return ret;
255 256 257

	DPRINTK("ENTER\n");

258 259 260 261 262 263
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
264
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
265
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
266 267 268 269 270 271
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
272
	spin_unlock_irqrestore(ap->lock, flags);
273

274
 out:
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
	DPRINTK("EXIT, ret=%d\n", ret);
	return ret;
}

/**
 *	ata_scsi_error - SCSI layer error handler callback
 *	@host: SCSI host on which error occurred
 *
 *	Handles SCSI-layer-thrown error events.
 *
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 *
 *	RETURNS:
 *	Zero.
 */
291
void ata_scsi_error(struct Scsi_Host *host)
292
{
293
	struct ata_port *ap = ata_shost_to_port(host);
294 295
	int i, repeat_cnt = ATA_EH_MAX_REPEAT;
	unsigned long flags;
296 297 298

	DPRINTK("ENTER\n");

299
	/* synchronize with port task */
300 301
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
302
	/* synchronize with host lock and sort out timeouts */
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319

	/* For new EH, all qcs are finished in one of three ways -
	 * normal completion, error completion, and SCSI timeout.
	 * Both cmpletions can race against SCSI timeout.  When normal
	 * completion wins, the qc never reaches EH.  When error
	 * completion wins, the qc has ATA_QCFLAG_FAILED set.
	 *
	 * When SCSI timeout wins, things are a bit more complex.
	 * Normal or error completion can occur after the timeout but
	 * before this point.  In such cases, both types of
	 * completions are honored.  A scmd is determined to have
	 * timed out iff its associated qc is active and not failed.
	 */
	if (ap->ops->error_handler) {
		struct scsi_cmnd *scmd, *tmp;
		int nr_timedout = 0;

320
		spin_lock_irqsave(ap->lock, flags);
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358

		list_for_each_entry_safe(scmd, tmp, &host->eh_cmd_q, eh_entry) {
			struct ata_queued_cmd *qc;

			for (i = 0; i < ATA_MAX_QUEUE; i++) {
				qc = __ata_qc_from_tag(ap, i);
				if (qc->flags & ATA_QCFLAG_ACTIVE &&
				    qc->scsicmd == scmd)
					break;
			}

			if (i < ATA_MAX_QUEUE) {
				/* the scmd has an associated qc */
				if (!(qc->flags & ATA_QCFLAG_FAILED)) {
					/* which hasn't failed yet, timeout */
					qc->err_mask |= AC_ERR_TIMEOUT;
					qc->flags |= ATA_QCFLAG_FAILED;
					nr_timedout++;
				}
			} else {
				/* Normal completion occurred after
				 * SCSI timeout but before this point.
				 * Successfully complete it.
				 */
				scmd->retries = scmd->allowed;
				scsi_eh_finish_cmd(scmd, &ap->eh_done_q);
			}
		}

		/* If we have timed out qcs.  They belong to EH from
		 * this point but the state of the controller is
		 * unknown.  Freeze the port to make sure the IRQ
		 * handler doesn't diddle with those qcs.  This must
		 * be done atomically w.r.t. setting QCFLAG_FAILED.
		 */
		if (nr_timedout)
			__ata_port_freeze(ap);

359
		spin_unlock_irqrestore(ap->lock, flags);
360
	} else
361
		spin_unlock_wait(ap->lock);
362 363 364 365

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
T
Tejun Heo 已提交
366 367 368
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

369 370 371
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

372
		/* fetch & clear EH info */
373
		spin_lock_irqsave(ap->lock, flags);
374

T
Tejun Heo 已提交
375 376 377
		memset(&ap->link.eh_context, 0, sizeof(ap->link.eh_context));
		ap->link.eh_context.i = ap->link.eh_info;
		memset(&ap->link.eh_info, 0, sizeof(ap->link.eh_info));
378

379 380
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
381

382
		spin_unlock_irqrestore(ap->lock, flags);
383

384 385
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
386 387 388
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
389

390 391 392
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

393 394 395 396
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
397
		spin_lock_irqsave(ap->lock, flags);
398

399
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
400 401 402 403
			if (--repeat_cnt) {
				ata_port_printk(ap, KERN_INFO,
					"EH pending after completion, "
					"repeating EH (cnt=%d)\n", repeat_cnt);
404
				spin_unlock_irqrestore(ap->lock, flags);
405 406 407 408
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
					"tries, giving up\n", ATA_EH_MAX_REPEAT);
T
Tejun Heo 已提交
409
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
410 411
		}

412
		/* this run is complete, make sure EH info is clear */
T
Tejun Heo 已提交
413
		memset(&ap->link.eh_info, 0, sizeof(ap->link.eh_info));
414

415
		/* Clear host_eh_scheduled while holding ap->lock such
416 417 418 419 420 421
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

422
		spin_unlock_irqrestore(ap->lock, flags);
423
	} else {
T
Tejun Heo 已提交
424
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
425 426
		ap->ops->eng_timeout(ap);
	}
427

428
	/* finish or retry handled scmd's and clean up */
429 430 431 432
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

433
	/* clean up */
434
	spin_lock_irqsave(ap->lock, flags);
435

436
	if (ap->pflags & ATA_PFLAG_LOADING)
437
		ap->pflags &= ~ATA_PFLAG_LOADING;
438
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
439
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
440 441 442

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
443

444
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
445

446
	/* tell wait_eh that we're done */
447
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
448 449
	wake_up_all(&ap->eh_wait_q);

450
	spin_unlock_irqrestore(ap->lock, flags);
451

452 453 454
	DPRINTK("EXIT\n");
}

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
/**
 *	ata_port_wait_eh - Wait for the currently pending EH to complete
 *	@ap: Port to wait EH for
 *
 *	Wait until the currently pending EH is complete.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
void ata_port_wait_eh(struct ata_port *ap)
{
	unsigned long flags;
	DEFINE_WAIT(wait);

 retry:
470
	spin_lock_irqsave(ap->lock, flags);
471

472
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
473
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
474
		spin_unlock_irqrestore(ap->lock, flags);
475
		schedule();
476
		spin_lock_irqsave(ap->lock, flags);
477
	}
478
	finish_wait(&ap->eh_wait_q, &wait);
479

480
	spin_unlock_irqrestore(ap->lock, flags);
481 482

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
483
	if (scsi_host_in_recovery(ap->scsi_host)) {
484 485 486 487 488
		msleep(10);
		goto retry;
	}
}

489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
/**
 *	ata_qc_timeout - Handle timeout of queued command
 *	@qc: Command that timed out
 *
 *	Some part of the kernel (currently, only the SCSI layer)
 *	has noticed that the active command on port @ap has not
 *	completed after a specified length of time.  Handle this
 *	condition by disabling DMA (if necessary) and completing
 *	transactions, with error if necessary.
 *
 *	This also handles the case of the "lost interrupt", where
 *	for some reason (possibly hardware bug, possibly driver bug)
 *	an interrupt was not delivered to the driver, even though the
 *	transaction completed successfully.
 *
504 505
 *	TODO: kill this function once old EH is gone.
 *
506 507 508 509 510 511 512 513 514 515 516 517 518
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 */
static void ata_qc_timeout(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;
	u8 host_stat = 0, drv_stat;
	unsigned long flags;

	DPRINTK("ENTER\n");

	ap->hsm_task_state = HSM_ST_IDLE;

519
	spin_lock_irqsave(ap->lock, flags);
520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538

	switch (qc->tf.protocol) {

	case ATA_PROT_DMA:
	case ATA_PROT_ATAPI_DMA:
		host_stat = ap->ops->bmdma_status(ap);

		/* before we do anything else, clear DMA-Start bit */
		ap->ops->bmdma_stop(qc);

		/* fall through */

	default:
		ata_altstatus(ap);
		drv_stat = ata_chk_status(ap);

		/* ack bmdma irq events */
		ap->ops->irq_clear(ap);

539 540 541
		ata_dev_printk(qc->dev, KERN_ERR, "command 0x%x timeout, "
			       "stat 0x%x host_stat 0x%x\n",
			       qc->tf.command, drv_stat, host_stat);
542 543

		/* complete taskfile transaction */
544
		qc->err_mask |= AC_ERR_TIMEOUT;
545 546 547
		break;
	}

548
	spin_unlock_irqrestore(ap->lock, flags);
549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569

	ata_eh_qc_complete(qc);

	DPRINTK("EXIT\n");
}

/**
 *	ata_eng_timeout - Handle timeout of queued command
 *	@ap: Port on which timed-out command is active
 *
 *	Some part of the kernel (currently, only the SCSI layer)
 *	has noticed that the active command on port @ap has not
 *	completed after a specified length of time.  Handle this
 *	condition by disabling DMA (if necessary) and completing
 *	transactions, with error if necessary.
 *
 *	This also handles the case of the "lost interrupt", where
 *	for some reason (possibly hardware bug, possibly driver bug)
 *	an interrupt was not delivered to the driver, even though the
 *	transaction completed successfully.
 *
570 571
 *	TODO: kill this function once old EH is gone.
 *
572 573 574 575 576 577 578
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 */
void ata_eng_timeout(struct ata_port *ap)
{
	DPRINTK("ENTER\n");

T
Tejun Heo 已提交
579
	ata_qc_timeout(ata_qc_from_tag(ap, ap->link.active_tag));
580 581 582 583

	DPRINTK("EXIT\n");
}

T
Tejun Heo 已提交
584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
static int ata_eh_nr_in_flight(struct ata_port *ap)
{
	unsigned int tag;
	int nr = 0;

	/* count only non-internal commands */
	for (tag = 0; tag < ATA_MAX_QUEUE - 1; tag++)
		if (ata_qc_from_tag(ap, tag))
			nr++;

	return nr;
}

void ata_eh_fastdrain_timerfn(unsigned long arg)
{
	struct ata_port *ap = (void *)arg;
	unsigned long flags;
	int cnt;

	spin_lock_irqsave(ap->lock, flags);

	cnt = ata_eh_nr_in_flight(ap);

	/* are we done? */
	if (!cnt)
		goto out_unlock;

	if (cnt == ap->fastdrain_cnt) {
		unsigned int tag;

		/* No progress during the last interval, tag all
		 * in-flight qcs as timed out and freeze the port.
		 */
		for (tag = 0; tag < ATA_MAX_QUEUE - 1; tag++) {
			struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);
			if (qc)
				qc->err_mask |= AC_ERR_TIMEOUT;
		}

		ata_port_freeze(ap);
	} else {
		/* some qcs have finished, give it another chance */
		ap->fastdrain_cnt = cnt;
		ap->fastdrain_timer.expires =
			jiffies + ATA_EH_FASTDRAIN_INTERVAL;
		add_timer(&ap->fastdrain_timer);
	}

 out_unlock:
	spin_unlock_irqrestore(ap->lock, flags);
}

/**
 *	ata_eh_set_pending - set ATA_PFLAG_EH_PENDING and activate fast drain
 *	@ap: target ATA port
 *	@fastdrain: activate fast drain
 *
 *	Set ATA_PFLAG_EH_PENDING and activate fast drain if @fastdrain
 *	is non-zero and EH wasn't pending before.  Fast drain ensures
 *	that EH kicks in in timely manner.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 */
static void ata_eh_set_pending(struct ata_port *ap, int fastdrain)
{
	int cnt;

	/* already scheduled? */
	if (ap->pflags & ATA_PFLAG_EH_PENDING)
		return;

	ap->pflags |= ATA_PFLAG_EH_PENDING;

	if (!fastdrain)
		return;

	/* do we have in-flight qcs? */
	cnt = ata_eh_nr_in_flight(ap);
	if (!cnt)
		return;

	/* activate fast drain */
	ap->fastdrain_cnt = cnt;
	ap->fastdrain_timer.expires = jiffies + ATA_EH_FASTDRAIN_INTERVAL;
	add_timer(&ap->fastdrain_timer);
}

672 673 674 675 676 677 678 679
/**
 *	ata_qc_schedule_eh - schedule qc for error handling
 *	@qc: command to schedule error handling for
 *
 *	Schedule error handling for @qc.  EH will kick in as soon as
 *	other commands are drained.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
680
 *	spin_lock_irqsave(host lock)
681 682 683 684 685 686 687 688
 */
void ata_qc_schedule_eh(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;

	WARN_ON(!ap->ops->error_handler);

	qc->flags |= ATA_QCFLAG_FAILED;
T
Tejun Heo 已提交
689
	ata_eh_set_pending(ap, 1);
690 691 692 693 694 695 696 697 698

	/* The following will fail if timeout has already expired.
	 * ata_scsi_error() takes care of such scmds on EH entry.
	 * Note that ATA_QCFLAG_FAILED is unconditionally set after
	 * this function completes.
	 */
	scsi_req_abort_cmd(qc->scsicmd);
}

699 700 701 702 703 704 705 706
/**
 *	ata_port_schedule_eh - schedule error handling without a qc
 *	@ap: ATA port to schedule EH for
 *
 *	Schedule error handling for @ap.  EH will kick in as soon as
 *	all commands are drained.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
707
 *	spin_lock_irqsave(host lock)
708 709 710 711 712
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

713 714 715
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
716
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
717
	scsi_schedule_eh(ap->scsi_host);
718 719 720 721 722 723 724 725 726 727 728

	DPRINTK("port EH scheduled\n");
}

/**
 *	ata_port_abort - abort all qc's on the port
 *	@ap: ATA port to abort qc's for
 *
 *	Abort all active qc's of @ap and schedule EH.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
729
 *	spin_lock_irqsave(host lock)
730 731 732 733 734 735 736 737 738 739
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_port_abort(struct ata_port *ap)
{
	int tag, nr_aborted = 0;

	WARN_ON(!ap->ops->error_handler);

T
Tejun Heo 已提交
740 741 742
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

		if (qc) {
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774
/**
 *	__ata_port_freeze - freeze port
 *	@ap: ATA port to freeze
 *
 *	This function is called when HSM violation or some other
 *	condition disrupts normal operation of the port.  Frozen port
 *	is not allowed to perform any operation until the port is
 *	thawed, which usually follows a successful reset.
 *
 *	ap->ops->freeze() callback can be used for freezing the port
 *	hardware-wise (e.g. mask interrupt and stop DMA engine).  If a
 *	port cannot be frozen hardware-wise, the interrupt handler
 *	must ack and clear interrupts unconditionally while the port
 *	is frozen.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
775
 *	spin_lock_irqsave(host lock)
776 777 778 779 780 781 782 783
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

	if (ap->ops->freeze)
		ap->ops->freeze(ap);

784
	ap->pflags |= ATA_PFLAG_FROZEN;
785

T
Tejun Heo 已提交
786
	DPRINTK("ata%u port frozen\n", ap->print_id);
787 788 789 790 791 792 793 794 795
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
796
 *	spin_lock_irqsave(host lock)
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828
 *
 *	RETURNS:
 *	Number of aborted commands.
 */
int ata_port_freeze(struct ata_port *ap)
{
	int nr_aborted;

	WARN_ON(!ap->ops->error_handler);

	nr_aborted = ata_port_abort(ap);
	__ata_port_freeze(ap);

	return nr_aborted;
}

/**
 *	ata_eh_freeze_port - EH helper to freeze port
 *	@ap: ATA port to freeze
 *
 *	Freeze @ap.
 *
 *	LOCKING:
 *	None.
 */
void ata_eh_freeze_port(struct ata_port *ap)
{
	unsigned long flags;

	if (!ap->ops->error_handler)
		return;

829
	spin_lock_irqsave(ap->lock, flags);
830
	__ata_port_freeze(ap);
831
	spin_unlock_irqrestore(ap->lock, flags);
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849
}

/**
 *	ata_port_thaw_port - EH helper to thaw port
 *	@ap: ATA port to thaw
 *
 *	Thaw frozen port @ap.
 *
 *	LOCKING:
 *	None.
 */
void ata_eh_thaw_port(struct ata_port *ap)
{
	unsigned long flags;

	if (!ap->ops->error_handler)
		return;

850
	spin_lock_irqsave(ap->lock, flags);
851

852
	ap->pflags &= ~ATA_PFLAG_FROZEN;
853 854 855 856

	if (ap->ops->thaw)
		ap->ops->thaw(ap);

857
	spin_unlock_irqrestore(ap->lock, flags);
858

T
Tejun Heo 已提交
859
	DPRINTK("ata%u port thawed\n", ap->print_id);
860 861
}

862 863 864 865 866 867 868 869 870 871 872
static void ata_eh_scsidone(struct scsi_cmnd *scmd)
{
	/* nada */
}

static void __ata_eh_qc_complete(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;
	struct scsi_cmnd *scmd = qc->scsicmd;
	unsigned long flags;

873
	spin_lock_irqsave(ap->lock, flags);
874 875 876
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
877
	spin_unlock_irqrestore(ap->lock, flags);
878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913

	scsi_eh_finish_cmd(scmd, &ap->eh_done_q);
}

/**
 *	ata_eh_qc_complete - Complete an active ATA command from EH
 *	@qc: Command to complete
 *
 *	Indicate to the mid and upper layers that an ATA command has
 *	completed.  To be used from EH.
 */
void ata_eh_qc_complete(struct ata_queued_cmd *qc)
{
	struct scsi_cmnd *scmd = qc->scsicmd;
	scmd->retries = scmd->allowed;
	__ata_eh_qc_complete(qc);
}

/**
 *	ata_eh_qc_retry - Tell midlayer to retry an ATA command after EH
 *	@qc: Command to retry
 *
 *	Indicate to the mid and upper layers that an ATA command
 *	should be retried.  To be used from EH.
 *
 *	SCSI midlayer limits the number of retries to scmd->allowed.
 *	scmd->retries is decremented for commands which get retried
 *	due to unrelated failures (qc->err_mask is zero).
 */
void ata_eh_qc_retry(struct ata_queued_cmd *qc)
{
	struct scsi_cmnd *scmd = qc->scsicmd;
	if (!qc->err_mask && scmd->retries)
		scmd->retries--;
	__ata_eh_qc_complete(qc);
}
T
Tejun Heo 已提交
914

915 916 917 918 919 920 921 922 923 924 925
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_detach_dev(struct ata_device *dev)
{
926 927
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
928 929 930 931
	unsigned long flags;

	ata_dev_disable(dev);

932
	spin_lock_irqsave(ap->lock, flags);
933 934 935 936 937

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
938
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
939 940
	}

941
	/* clear per-dev EH actions */
942 943
	ata_eh_clear_action(link, dev, &link->eh_info, ATA_EH_PERDEV_MASK);
	ata_eh_clear_action(link, dev, &link->eh_context.i, ATA_EH_PERDEV_MASK);
944

945
	spin_unlock_irqrestore(ap->lock, flags);
946 947
}

T
Tejun Heo 已提交
948 949
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
950
 *	@link: target ATA link
951
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
952 953 954
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
955 956
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
957 958 959 960
 *
 *	LOCKING:
 *	None.
 */
T
Tejun Heo 已提交
961
static void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
962
			       unsigned int action)
T
Tejun Heo 已提交
963
{
T
Tejun Heo 已提交
964 965 966
	struct ata_port *ap = link->ap;
	struct ata_eh_info *ehi = &link->eh_info;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
967 968
	unsigned long flags;

969
	spin_lock_irqsave(ap->lock, flags);
970

971 972 973 974 975 976 977 978 979 980 981 982 983
	/* Reset is represented by combination of actions and EHI
	 * flags.  Suck in all related bits before clearing eh_info to
	 * avoid losing requested action.
	 */
	if (action & ATA_EH_RESET_MASK) {
		ehc->i.action |= ehi->action & ATA_EH_RESET_MASK;
		ehc->i.flags |= ehi->flags & ATA_EHI_RESET_MODIFIER_MASK;

		/* make sure all reset actions are cleared & clear EHI flags */
		action |= ATA_EH_RESET_MASK;
		ehi->flags &= ~ATA_EHI_RESET_MODIFIER_MASK;
	}

T
Tejun Heo 已提交
984
	ata_eh_clear_action(link, dev, ehi, action);
985

986
	if (!(ehc->i.flags & ATA_EHI_QUIET))
987 988
		ap->pflags |= ATA_PFLAG_RECOVERED;

989
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
990 991
}

992 993
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
994
*	@ap: target ATA port
995 996 997 998
 *	@dev: target ATA dev for per-dev action (can be NULL)
 *	@action: action just completed
 *
 *	Called right after performing EH actions to clear related bits
T
Tejun Heo 已提交
999
 *	in @link->eh_context.
1000 1001 1002 1003
 *
 *	LOCKING:
 *	None.
 */
T
Tejun Heo 已提交
1004
static void ata_eh_done(struct ata_link *link, struct ata_device *dev,
1005 1006
			unsigned int action)
{
T
Tejun Heo 已提交
1007
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1008

1009 1010 1011
	/* if reset is complete, clear all reset actions & reset modifier */
	if (action & ATA_EH_RESET_MASK) {
		action |= ATA_EH_RESET_MASK;
T
Tejun Heo 已提交
1012
		ehc->i.flags &= ~ATA_EHI_RESET_MODIFIER_MASK;
1013 1014
	}

T
Tejun Heo 已提交
1015
	ata_eh_clear_action(link, dev, &ehc->i, action);
1016 1017
}

T
Tejun Heo 已提交
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
/**
 *	ata_err_string - convert err_mask to descriptive string
 *	@err_mask: error mask to convert to string
 *
 *	Convert @err_mask to descriptive string.  Errors are
 *	prioritized according to severity and only the most severe
 *	error is reported.
 *
 *	LOCKING:
 *	None.
 *
 *	RETURNS:
 *	Descriptive string for @err_mask
 */
static const char * ata_err_string(unsigned int err_mask)
{
	if (err_mask & AC_ERR_HOST_BUS)
		return "host bus error";
	if (err_mask & AC_ERR_ATA_BUS)
		return "ATA bus error";
	if (err_mask & AC_ERR_TIMEOUT)
		return "timeout";
	if (err_mask & AC_ERR_HSM)
		return "HSM violation";
	if (err_mask & AC_ERR_SYSTEM)
		return "internal error";
	if (err_mask & AC_ERR_MEDIA)
		return "media error";
	if (err_mask & AC_ERR_INVALID)
		return "invalid argument";
	if (err_mask & AC_ERR_DEV)
		return "device error";
	return "unknown error";
}

1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
/**
 *	ata_read_log_page - read a specific log page
 *	@dev: target device
 *	@page: page to read
 *	@buf: buffer to store read page
 *	@sectors: number of sectors to read
 *
 *	Read log page using READ_LOG_EXT command.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask otherwise.
 */
static unsigned int ata_read_log_page(struct ata_device *dev,
				      u8 page, void *buf, unsigned int sectors)
{
	struct ata_taskfile tf;
	unsigned int err_mask;

	DPRINTK("read log page - page %d\n", page);

	ata_tf_init(dev, &tf);
	tf.command = ATA_CMD_READ_LOG_EXT;
	tf.lbal = page;
	tf.nsect = sectors;
	tf.hob_nsect = sectors >> 8;
	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_LBA48 | ATA_TFLAG_DEVICE;
	tf.protocol = ATA_PROT_PIO;

	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_FROM_DEVICE,
				     buf, sectors * ATA_SECT_SIZE);

	DPRINTK("EXIT, err_mask=%x\n", err_mask);
	return err_mask;
}

/**
 *	ata_eh_read_log_10h - Read log page 10h for NCQ error details
 *	@dev: Device to read log page 10h from
 *	@tag: Resulting tag of the failed command
 *	@tf: Resulting taskfile registers of the failed command
 *
 *	Read log page 10h to obtain NCQ error details and clear error
 *	condition.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno otherwise.
 */
static int ata_eh_read_log_10h(struct ata_device *dev,
			       int *tag, struct ata_taskfile *tf)
{
T
Tejun Heo 已提交
1109
	u8 *buf = dev->link->ap->sector_buf;
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
	unsigned int err_mask;
	u8 csum;
	int i;

	err_mask = ata_read_log_page(dev, ATA_LOG_SATA_NCQ, buf, 1);
	if (err_mask)
		return -EIO;

	csum = 0;
	for (i = 0; i < ATA_SECT_SIZE; i++)
		csum += buf[i];
	if (csum)
		ata_dev_printk(dev, KERN_WARNING,
			       "invalid checksum 0x%x on log page 10h\n", csum);

	if (buf[0] & 0x80)
		return -ENOENT;

	*tag = buf[0] & 0x1f;

	tf->command = buf[2];
	tf->feature = buf[3];
	tf->lbal = buf[4];
	tf->lbam = buf[5];
	tf->lbah = buf[6];
	tf->device = buf[7];
	tf->hob_lbal = buf[8];
	tf->hob_lbam = buf[9];
	tf->hob_lbah = buf[10];
	tf->nsect = buf[12];
	tf->hob_nsect = buf[13];

	return 0;
}

T
Tejun Heo 已提交
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
/**
 *	atapi_eh_request_sense - perform ATAPI REQUEST_SENSE
 *	@dev: device to perform REQUEST_SENSE to
 *	@sense_buf: result sense data buffer (SCSI_SENSE_BUFFERSIZE bytes long)
 *
 *	Perform ATAPI REQUEST_SENSE after the device reported CHECK
 *	SENSE.  This function is EH helper.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask on failure
 */
1159
static unsigned int atapi_eh_request_sense(struct ata_queued_cmd *qc)
T
Tejun Heo 已提交
1160
{
1161 1162
	struct ata_device *dev = qc->dev;
	unsigned char *sense_buf = qc->scsicmd->sense_buffer;
T
Tejun Heo 已提交
1163
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1164 1165 1166 1167 1168 1169 1170 1171
	struct ata_taskfile tf;
	u8 cdb[ATAPI_CDB_LEN];

	DPRINTK("ATAPI request sense\n");

	/* FIXME: is this needed? */
	memset(sense_buf, 0, SCSI_SENSE_BUFFERSIZE);

1172 1173 1174
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1175
	sense_buf[0] = 0x70;
1176 1177
	sense_buf[2] = qc->result_tf.feature >> 4;

J
Jeff Garzik 已提交
1178
	/* some devices time out if garbage left in tf */
1179
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213

	memset(cdb, 0, ATAPI_CDB_LEN);
	cdb[0] = REQUEST_SENSE;
	cdb[4] = SCSI_SENSE_BUFFERSIZE;

	tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
	tf.command = ATA_CMD_PACKET;

	/* is it pointless to prefer PIO for "safety reasons"? */
	if (ap->flags & ATA_FLAG_PIO_DMA) {
		tf.protocol = ATA_PROT_ATAPI_DMA;
		tf.feature |= ATAPI_PKT_DMA;
	} else {
		tf.protocol = ATA_PROT_ATAPI;
		tf.lbam = (8 * 1024) & 0xff;
		tf.lbah = (8 * 1024) >> 8;
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
				 sense_buf, SCSI_SENSE_BUFFERSIZE);
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
 *	@ap: ATA port to analyze SError for
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_analyze_serror(struct ata_port *ap)
{
T
Tejun Heo 已提交
1214
	struct ata_eh_context *ehc = &ap->link.eh_context;
T
Tejun Heo 已提交
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;

	if (serror & SERR_PERSISTENT) {
		err_mask |= AC_ERR_ATA_BUS;
		action |= ATA_EH_HARDRESET;
	}
	if (serror &
	    (SERR_DATA_RECOVERED | SERR_COMM_RECOVERED | SERR_DATA)) {
		err_mask |= AC_ERR_ATA_BUS;
		action |= ATA_EH_SOFTRESET;
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
		action |= ATA_EH_SOFTRESET;
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
T
Tejun Heo 已提交
1233
		action |= ATA_EH_HARDRESET;
T
Tejun Heo 已提交
1234
	}
T
Tejun Heo 已提交
1235 1236
	if (serror & (SERR_PHYRDY_CHG | SERR_DEV_XCHG))
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1237 1238 1239 1240 1241

	ehc->i.err_mask |= err_mask;
	ehc->i.action |= action;
}

1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
 *	@ap: ATA port to analyze NCQ error for
 *
 *	Read log page 10h, determine the offending qc and acquire
 *	error status TF.  For NCQ device errors, all LLDDs have to do
 *	is setting AC_ERR_DEV in ehi->err_mask.  This function takes
 *	care of the rest.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_analyze_ncq_error(struct ata_port *ap)
{
T
Tejun Heo 已提交
1256 1257
	struct ata_eh_context *ehc = &ap->link.eh_context;
	struct ata_device *dev = ap->link.device;
1258 1259 1260 1261 1262
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1263
	if (ap->pflags & ATA_PFLAG_FROZEN)
1264 1265 1266
		return;

	/* is it NCQ device error? */
T
Tejun Heo 已提交
1267
	if (!ap->link.sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288
		return;

	/* has LLDD analyzed already? */
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		qc = __ata_qc_from_tag(ap, tag);

		if (!(qc->flags & ATA_QCFLAG_FAILED))
			continue;

		if (qc->err_mask)
			return;
	}

	/* okay, this error is ours */
	rc = ata_eh_read_log_10h(dev, &tag, &tf);
	if (rc) {
		ata_port_printk(ap, KERN_ERR, "failed to read log page 10h "
				"(errno=%d)\n", rc);
		return;
	}

T
Tejun Heo 已提交
1289
	if (!(ap->link.sactive & (1 << tag))) {
1290 1291 1292 1293 1294 1295 1296 1297
		ata_port_printk(ap, KERN_ERR, "log page 10h reported "
				"inactive tag %d\n", tag);
		return;
	}

	/* we've got the perpetrator, condemn it */
	qc = __ata_qc_from_tag(ap, tag);
	memcpy(&qc->result_tf, &tf, sizeof(tf));
T
Tejun Heo 已提交
1298
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1299 1300 1301
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
/**
 *	ata_eh_analyze_tf - analyze taskfile of a failed qc
 *	@qc: qc to analyze
 *	@tf: Taskfile registers to analyze
 *
 *	Analyze taskfile of @qc and further determine cause of
 *	failure.  This function also requests ATAPI sense data if
 *	avaliable.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	Determined recovery action
 */
static unsigned int ata_eh_analyze_tf(struct ata_queued_cmd *qc,
				      const struct ata_taskfile *tf)
{
	unsigned int tmp, action = 0;
	u8 stat = tf->command, err = tf->feature;

	if ((stat & (ATA_BUSY | ATA_DRQ | ATA_DRDY)) != ATA_DRDY) {
		qc->err_mask |= AC_ERR_HSM;
		return ATA_EH_SOFTRESET;
	}

1328 1329 1330
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
		return 0;

	switch (qc->dev->class) {
	case ATA_DEV_ATA:
		if (err & ATA_ICRC)
			qc->err_mask |= AC_ERR_ATA_BUS;
		if (err & ATA_UNC)
			qc->err_mask |= AC_ERR_MEDIA;
		if (err & ATA_IDNF)
			qc->err_mask |= AC_ERR_INVALID;
		break;

	case ATA_DEV_ATAPI:
1344
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1345
			tmp = atapi_eh_request_sense(qc);
1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357
			if (!tmp) {
				/* ATA_QCFLAG_SENSE_VALID is used to
				 * tell atapi_qc_complete() that sense
				 * data is already valid.
				 *
				 * TODO: interpret sense data and set
				 * appropriate err_mask.
				 */
				qc->flags |= ATA_QCFLAG_SENSE_VALID;
			} else
				qc->err_mask |= tmp;
		}
T
Tejun Heo 已提交
1358 1359 1360 1361 1362 1363 1364 1365
	}

	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
		action |= ATA_EH_SOFTRESET;

	return action;
}

1366
static int ata_eh_categorize_error(int is_io, unsigned int err_mask)
T
Tejun Heo 已提交
1367
{
1368
	if (err_mask & AC_ERR_ATA_BUS)
T
Tejun Heo 已提交
1369 1370
		return 1;

1371 1372 1373 1374 1375
	if (err_mask & AC_ERR_TIMEOUT)
		return 2;

	if (is_io) {
		if (err_mask & AC_ERR_HSM)
T
Tejun Heo 已提交
1376
			return 2;
1377 1378 1379
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
			return 3;
T
Tejun Heo 已提交
1380 1381 1382 1383 1384
	}

	return 0;
}

1385
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1386
	u64 since;
1387
	int nr_errors[4];
T
Tejun Heo 已提交
1388 1389
};

1390
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1391
{
1392 1393
	struct speed_down_verdict_arg *arg = void_arg;
	int cat = ata_eh_categorize_error(ent->is_io, ent->err_mask);
T
Tejun Heo 已提交
1394 1395 1396 1397

	if (ent->timestamp < arg->since)
		return -1;

1398
	arg->nr_errors[cat]++;
T
Tejun Heo 已提交
1399 1400 1401 1402
	return 0;
}

/**
1403
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1404 1405 1406
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1407 1408
 *	whether NCQ needs to be turned off, transfer speed should be
 *	stepped down, or falling back to PIO is necessary.
T
Tejun Heo 已提交
1409
 *
1410
 *	Cat-1 is ATA_BUS error for any command.
T
Tejun Heo 已提交
1411
 *
1412 1413 1414 1415
 *	Cat-2 is TIMEOUT for any command or HSM violation for known
 *	supported commands.
 *
 *	Cat-3 is is unclassified DEV error for known supported
T
Tejun Heo 已提交
1416 1417
 *	command.
 *
1418 1419 1420 1421 1422 1423 1424 1425 1426
 *	NCQ needs to be turned off if there have been more than 3
 *	Cat-2 + Cat-3 errors during last 10 minutes.
 *
 *	Speed down is necessary if there have been more than 3 Cat-1 +
 *	Cat-2 errors or 10 Cat-3 errors during last 10 minutes.
 *
 *	Falling back to PIO mode is necessary if there have been more
 *	than 10 Cat-1 + Cat-2 + Cat-3 errors during last 5 minutes.
 *
T
Tejun Heo 已提交
1427 1428 1429 1430
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1431
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1432
 */
1433
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1434
{
1435 1436 1437 1438
	const u64 j5mins = 5LLU * 60 * HZ, j10mins = 10LLU * 60 * HZ;
	u64 j64 = get_jiffies_64();
	struct speed_down_verdict_arg arg;
	unsigned int verdict = 0;
T
Tejun Heo 已提交
1439

1440 1441 1442 1443
	/* scan past 10 mins of error history */
	memset(&arg, 0, sizeof(arg));
	arg.since = j64 - min(j64, j10mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1444

1445 1446 1447 1448
	if (arg.nr_errors[2] + arg.nr_errors[3] > 3)
		verdict |= ATA_EH_SPDN_NCQ_OFF;
	if (arg.nr_errors[1] + arg.nr_errors[2] > 3 || arg.nr_errors[3] > 10)
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
T
Tejun Heo 已提交
1449

1450
	/* scan past 3 mins of error history */
T
Tejun Heo 已提交
1451
	memset(&arg, 0, sizeof(arg));
1452 1453
	arg.since = j64 - min(j64, j5mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1454

1455 1456
	if (arg.nr_errors[1] + arg.nr_errors[2] + arg.nr_errors[3] > 10)
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
T
Tejun Heo 已提交
1457

1458
	return verdict;
T
Tejun Heo 已提交
1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
 *	@is_io: Did the device fail during normal IO?
 *	@err_mask: err_mask of the error
 *
 *	Record error and examine error history to determine whether
 *	adjusting transmission speed is necessary.  It also sets
 *	transmission limits appropriately if such adjustment is
 *	necessary.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
1476
 *	Determined recovery action.
T
Tejun Heo 已提交
1477
 */
1478 1479
static unsigned int ata_eh_speed_down(struct ata_device *dev, int is_io,
				      unsigned int err_mask)
T
Tejun Heo 已提交
1480
{
1481 1482 1483 1484 1485
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
	if (ata_eh_categorize_error(is_io, err_mask) == 0)
T
Tejun Heo 已提交
1486 1487 1488 1489
		return 0;

	/* record error and determine whether speed down is necessary */
	ata_ering_record(&dev->ering, is_io, err_mask);
1490
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1491

1492 1493 1494 1495 1496 1497 1498 1499 1500
	/* turn off NCQ? */
	if ((verdict & ATA_EH_SPDN_NCQ_OFF) &&
	    (dev->flags & (ATA_DFLAG_PIO | ATA_DFLAG_NCQ |
			   ATA_DFLAG_NCQ_OFF)) == ATA_DFLAG_NCQ) {
		dev->flags |= ATA_DFLAG_NCQ_OFF;
		ata_dev_printk(dev, KERN_WARNING,
			       "NCQ disabled due to excessive errors\n");
		goto done;
	}
T
Tejun Heo 已提交
1501

1502 1503 1504
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1505
		if (sata_down_spd_limit(dev->link) == 0) {
1506 1507 1508
			action |= ATA_EH_HARDRESET;
			goto done;
		}
T
Tejun Heo 已提交
1509

1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
		/* lower transfer mode */
		if (dev->spdn_cnt < 2) {
			static const int dma_dnxfer_sel[] =
				{ ATA_DNXFER_DMA, ATA_DNXFER_40C };
			static const int pio_dnxfer_sel[] =
				{ ATA_DNXFER_PIO, ATA_DNXFER_FORCE_PIO0 };
			int sel;

			if (dev->xfer_shift != ATA_SHIFT_PIO)
				sel = dma_dnxfer_sel[dev->spdn_cnt];
			else
				sel = pio_dnxfer_sel[dev->spdn_cnt];

			dev->spdn_cnt++;

			if (ata_down_xfermask_limit(dev, sel) == 0) {
				action |= ATA_EH_SOFTRESET;
				goto done;
			}
		}
	}

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
	 * SATA.  Consider it only for PATA.
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
T
Tejun Heo 已提交
1536
	    (dev->link->ap->cbl != ATA_CBL_SATA) &&
1537 1538 1539 1540 1541 1542 1543
	    (dev->xfer_shift != ATA_SHIFT_PIO)) {
		if (ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO) == 0) {
			dev->spdn_cnt = 0;
			action |= ATA_EH_SOFTRESET;
			goto done;
		}
	}
T
Tejun Heo 已提交
1544 1545

	return 0;
1546 1547 1548 1549
 done:
	/* device has been slowed down, blow error history */
	ata_ering_clear(&dev->ering);
	return action;
T
Tejun Heo 已提交
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
}

/**
 *	ata_eh_autopsy - analyze error and determine recovery action
 *	@ap: ATA port to perform autopsy on
 *
 *	Analyze why @ap failed and determine which recovery action is
 *	needed.  This function also sets more detailed AC_ERR_* values
 *	and fills sense data for ATAPI CHECK SENSE.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_autopsy(struct ata_port *ap)
{
1565 1566
	struct ata_link *link = &ap->link;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1567 1568 1569 1570 1571 1572 1573
	unsigned int all_err_mask = 0;
	int tag, is_io = 0;
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1574 1575 1576
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1577
	/* obtain and analyze SError */
1578
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
1579 1580 1581
	if (rc == 0) {
		ehc->i.serror |= serror;
		ata_eh_analyze_serror(ap);
1582 1583 1584
	} else if (rc != -EOPNOTSUPP) {
		/* SError read failed, force hardreset and probing */
		ata_ehi_schedule_probe(&ehc->i);
1585
		ehc->i.action |= ATA_EH_HARDRESET;
1586 1587
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
1588

1589 1590 1591
	/* analyze NCQ failure */
	ata_eh_analyze_ncq_error(ap);

T
Tejun Heo 已提交
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
	/* any real error trumps AC_ERR_OTHER */
	if (ehc->i.err_mask & ~AC_ERR_OTHER)
		ehc->i.err_mask &= ~AC_ERR_OTHER;

	all_err_mask |= ehc->i.err_mask;

	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);

		if (!(qc->flags & ATA_QCFLAG_FAILED))
			continue;

		/* inherit upper level err_mask */
		qc->err_mask |= ehc->i.err_mask;

		/* analyze TF */
1608
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621

		/* DEV errors are probably spurious in case of ATA_BUS error */
		if (qc->err_mask & AC_ERR_ATA_BUS)
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_MEDIA |
					  AC_ERR_INVALID);

		/* any real error trumps unknown error */
		if (qc->err_mask & ~AC_ERR_OTHER)
			qc->err_mask &= ~AC_ERR_OTHER;

		/* SENSE_VALID trumps dev/unknown error and revalidation */
		if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);
1622
			ehc->i.action &= ~ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1623 1624 1625
		}

		/* accumulate error info */
1626
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1627 1628 1629 1630 1631
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
			is_io = 1;
	}

1632
	/* enforce default EH actions */
1633
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1634
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
1635
		ehc->i.action |= ATA_EH_SOFTRESET;
1636
	else if (all_err_mask)
1637
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1638

1639
	/* if we have offending qcs and the associated failed device */
1640
	if (ehc->i.dev) {
1641
		/* speed down */
1642 1643
		ehc->i.action |= ata_eh_speed_down(ehc->i.dev, is_io,
						   all_err_mask);
1644 1645

		/* perform per-dev EH action only on the offending device */
1646 1647 1648
		ehc->i.dev_action[ehc->i.dev->devno] |=
			ehc->i.action & ATA_EH_PERDEV_MASK;
		ehc->i.action &= ~ATA_EH_PERDEV_MASK;
1649 1650
	}

T
Tejun Heo 已提交
1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
	DPRINTK("EXIT\n");
}

/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port EH is going on
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_report(struct ata_port *ap)
{
T
Tejun Heo 已提交
1665
	struct ata_eh_context *ehc = &ap->link.eh_context;
T
Tejun Heo 已提交
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687
	const char *frozen, *desc;
	int tag, nr_failed = 0;

	desc = NULL;
	if (ehc->i.desc[0] != '\0')
		desc = ehc->i.desc;

	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);

		if (!(qc->flags & ATA_QCFLAG_FAILED))
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

	if (!nr_failed && !ehc->i.err_mask)
		return;

	frozen = "";
1688
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
1689 1690 1691
		frozen = " frozen";

	if (ehc->i.dev) {
1692 1693
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
			       "SAct 0x%x SErr 0x%x action 0x%x%s\n",
T
Tejun Heo 已提交
1694 1695
			       ehc->i.err_mask, ap->link.sactive,
			       ehc->i.serror, ehc->i.action, frozen);
T
Tejun Heo 已提交
1696
		if (desc)
T
Tejun Heo 已提交
1697
			ata_dev_printk(ehc->i.dev, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1698
	} else {
1699 1700
		ata_port_printk(ap, KERN_ERR, "exception Emask 0x%x "
				"SAct 0x%x SErr 0x%x action 0x%x%s\n",
T
Tejun Heo 已提交
1701 1702
				ehc->i.err_mask, ap->link.sactive,
				ehc->i.serror, ehc->i.action, frozen);
T
Tejun Heo 已提交
1703
		if (desc)
T
Tejun Heo 已提交
1704
			ata_port_printk(ap, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1705 1706 1707
	}

	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
1708 1709 1710 1711 1712 1713
		static const char *dma_str[] = {
			[DMA_BIDIRECTIONAL]	= "bidi",
			[DMA_TO_DEVICE]		= "out",
			[DMA_FROM_DEVICE]	= "in",
			[DMA_NONE]		= "",
		};
T
Tejun Heo 已提交
1714
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
1715
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
T
Tejun Heo 已提交
1716 1717 1718 1719

		if (!(qc->flags & ATA_QCFLAG_FAILED) || !qc->err_mask)
			continue;

1720 1721
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
1722
			"tag %d cdb 0x%x data %u %s\n         "
1723
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
1724
			"Emask 0x%x (%s)%s\n",
1725 1726 1727 1728
			cmd->command, cmd->feature, cmd->nsect,
			cmd->lbal, cmd->lbam, cmd->lbah,
			cmd->hob_feature, cmd->hob_nsect,
			cmd->hob_lbal, cmd->hob_lbam, cmd->hob_lbah,
T
Tejun Heo 已提交
1729
			cmd->device, qc->tag, qc->cdb[0], qc->nbytes,
1730
			dma_str[qc->dma_dir],
1731 1732 1733 1734
			res->command, res->feature, res->nsect,
			res->lbal, res->lbam, res->lbah,
			res->hob_feature, res->hob_nsect,
			res->hob_lbal, res->hob_lbam, res->hob_lbah,
T
Tejun Heo 已提交
1735 1736
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
T
Tejun Heo 已提交
1737 1738 1739
	}
}

1740
static int ata_do_reset(struct ata_port *ap, ata_reset_fn_t reset,
1741
			unsigned int *classes, unsigned long deadline)
1742
{
1743 1744
	struct ata_device *dev;
	int rc;
1745

1746 1747
	ata_link_for_each_dev(dev, &ap->link)
		classes[dev->devno] = ATA_DEV_UNKNOWN;
1748

1749
	rc = reset(ap, classes, deadline);
1750 1751 1752 1753 1754 1755 1756
	if (rc)
		return rc;

	/* If any class isn't ATA_DEV_UNKNOWN, consider classification
	 * is complete and convert all ATA_DEV_UNKNOWN to
	 * ATA_DEV_NONE.
	 */
1757 1758
	ata_link_for_each_dev(dev, &ap->link)
		if (classes[dev->devno] != ATA_DEV_UNKNOWN)
1759 1760
			break;

1761 1762 1763 1764 1765 1766
	if (dev) {
		ata_link_for_each_dev(dev, &ap->link) {
			if (classes[dev->devno] == ATA_DEV_UNKNOWN)
				classes[dev->devno] = ATA_DEV_NONE;
		}
	}
1767 1768 1769 1770

	return 0;
}

1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
static int ata_eh_followup_srst_needed(int rc, int classify,
				       const unsigned int *classes)
{
	if (rc == -EAGAIN)
		return 1;
	if (rc != 0)
		return 0;
	if (classify && classes[0] == ATA_DEV_UNKNOWN)
		return 1;
	return 0;
}

static int ata_eh_reset(struct ata_port *ap, int classify,
1784
			ata_prereset_fn_t prereset, ata_reset_fn_t softreset,
T
Tejun Heo 已提交
1785 1786
			ata_reset_fn_t hardreset, ata_postreset_fn_t postreset)
{
1787 1788
	struct ata_link *link = &ap->link;
	struct ata_eh_context *ehc = &link->eh_context;
1789
	unsigned int *classes = ehc->classes;
1790
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
T
Tejun Heo 已提交
1791
	int try = 0;
1792
	struct ata_device *dev;
T
Tejun Heo 已提交
1793
	unsigned long deadline;
1794
	unsigned int action;
T
Tejun Heo 已提交
1795
	ata_reset_fn_t reset;
1796
	int rc;
T
Tejun Heo 已提交
1797

1798
	/* about to reset */
T
Tejun Heo 已提交
1799
	ata_eh_about_to_do(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
1800

1801 1802 1803 1804 1805
	/* Determine which reset to use and record in ehc->i.action.
	 * prereset() may examine and modify it.
	 */
	action = ehc->i.action;
	ehc->i.action &= ~ATA_EH_RESET_MASK;
1806
	if (softreset && (!hardreset || (!sata_set_spd_needed(link) &&
1807 1808
					 !(action & ATA_EH_HARDRESET))))
		ehc->i.action |= ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
1809
	else
1810 1811 1812
		ehc->i.action |= ATA_EH_HARDRESET;

	if (prereset) {
T
Tejun Heo 已提交
1813
		rc = prereset(ap, jiffies + ATA_EH_PRERESET_TIMEOUT);
1814
		if (rc) {
1815
			if (rc == -ENOENT) {
1816 1817
				ata_port_printk(ap, KERN_DEBUG,
						"port disabled. ignoring.\n");
T
Tejun Heo 已提交
1818
				ehc->i.action &= ~ATA_EH_RESET_MASK;
1819

1820
				ata_link_for_each_dev(dev, link)
1821
					classes[dev->devno] = ATA_DEV_NONE;
1822 1823

				rc = 0;
1824 1825
			} else
				ata_port_printk(ap, KERN_ERR,
1826
					"prereset failed (errno=%d)\n", rc);
1827
			goto out;
1828 1829 1830 1831 1832
		}
	}

	/* prereset() might have modified ehc->i.action */
	if (ehc->i.action & ATA_EH_HARDRESET)
T
Tejun Heo 已提交
1833
		reset = hardreset;
1834 1835 1836 1837
	else if (ehc->i.action & ATA_EH_SOFTRESET)
		reset = softreset;
	else {
		/* prereset told us not to reset, bang classes and return */
1838
		ata_link_for_each_dev(dev, link)
1839
			classes[dev->devno] = ATA_DEV_NONE;
1840 1841
		rc = 0;
		goto out;
1842 1843 1844 1845 1846 1847 1848 1849 1850
	}

	/* did prereset() screw up?  if so, fix up to avoid oopsing */
	if (!reset) {
		if (softreset)
			reset = softreset;
		else
			reset = hardreset;
	}
T
Tejun Heo 已提交
1851 1852

 retry:
T
Tejun Heo 已提交
1853 1854
	deadline = jiffies + ata_eh_reset_timeouts[try++];

1855 1856 1857 1858
	/* shut up during boot probing */
	if (verbose)
		ata_port_printk(ap, KERN_INFO, "%s resetting port\n",
				reset == softreset ? "soft" : "hard");
T
Tejun Heo 已提交
1859

1860
	/* mark that this EH session started with reset */
1861 1862 1863 1864
	if (reset == hardreset)
		ehc->i.flags |= ATA_EHI_DID_HARDRESET;
	else
		ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
1865

T
Tejun Heo 已提交
1866
	rc = ata_do_reset(ap, reset, classes, deadline);
T
Tejun Heo 已提交
1867

1868 1869 1870 1871 1872 1873 1874 1875 1876
	if (reset == hardreset &&
	    ata_eh_followup_srst_needed(rc, classify, classes)) {
		/* okay, let's do follow-up softreset */
		reset = softreset;

		if (!reset) {
			ata_port_printk(ap, KERN_ERR,
					"follow-up softreset required "
					"but no softreset avaliable\n");
1877 1878
			rc = -EINVAL;
			goto out;
1879 1880
		}

T
Tejun Heo 已提交
1881
		ata_eh_about_to_do(link, NULL, ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
1882
		rc = ata_do_reset(ap, reset, classes, deadline);
1883 1884 1885 1886 1887

		if (rc == 0 && classify &&
		    classes[0] == ATA_DEV_UNKNOWN) {
			ata_port_printk(ap, KERN_ERR,
					"classification failed\n");
1888 1889
			rc = -EINVAL;
			goto out;
1890 1891 1892
		}
	}

T
Tejun Heo 已提交
1893 1894
	if (rc && try < ARRAY_SIZE(ata_eh_reset_timeouts)) {
		unsigned long now = jiffies;
1895

T
Tejun Heo 已提交
1896 1897
		if (time_before(now, deadline)) {
			unsigned long delta = deadline - jiffies;
1898

T
Tejun Heo 已提交
1899 1900 1901 1902 1903 1904
			ata_port_printk(ap, KERN_WARNING, "reset failed "
				"(errno=%d), retrying in %u secs\n",
				rc, (jiffies_to_msecs(delta) + 999) / 1000);

			schedule_timeout_uninterruptible(delta);
		}
T
Tejun Heo 已提交
1905

1906
		if (rc == -EPIPE ||
T
Tejun Heo 已提交
1907
		    try == ARRAY_SIZE(ata_eh_reset_timeouts) - 1)
1908
			sata_down_spd_limit(link);
T
Tejun Heo 已提交
1909 1910 1911 1912 1913 1914
		if (hardreset)
			reset = hardreset;
		goto retry;
	}

	if (rc == 0) {
1915 1916
		u32 sstatus;

1917 1918 1919
		/* After the reset, the device state is PIO 0 and the
		 * controller state is undefined.  Record the mode.
		 */
1920
		ata_link_for_each_dev(dev, link)
1921
			dev->pio_mode = XFER_PIO_0;
1922

1923
		/* record current link speed */
1924 1925
		if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
			link->sata_spd = (sstatus >> 4) & 0xf;
1926

T
Tejun Heo 已提交
1927 1928 1929 1930
		if (postreset)
			postreset(ap, classes);

		/* reset successful, schedule revalidation */
T
Tejun Heo 已提交
1931
		ata_eh_done(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
1932 1933
		ehc->i.action |= ATA_EH_REVALIDATE;
	}
1934 1935 1936
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
1937 1938 1939
	return rc;
}

T
Tejun Heo 已提交
1940 1941
static int ata_eh_revalidate_and_attach(struct ata_port *ap,
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
1942
{
T
Tejun Heo 已提交
1943
	struct ata_eh_context *ehc = &ap->link.eh_context;
T
Tejun Heo 已提交
1944
	struct ata_device *dev;
1945
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
1946
	unsigned long flags;
1947
	int rc = 0;
T
Tejun Heo 已提交
1948 1949 1950

	DPRINTK("ENTER\n");

1951 1952 1953 1954
	/* For PATA drive side cable detection to work, IDENTIFY must
	 * be done backwards such that PDIAG- is released by the slave
	 * device before the master device is identified.
	 */
1955 1956 1957
	ata_link_for_each_dev_reverse(dev, &ap->link) {
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
1958

1959 1960 1961
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

1962
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
1963
			if (ata_link_offline(&ap->link)) {
T
Tejun Heo 已提交
1964
				rc = -EIO;
1965
				goto err;
T
Tejun Heo 已提交
1966 1967
			}

T
Tejun Heo 已提交
1968
			ata_eh_about_to_do(&ap->link, dev, ATA_EH_REVALIDATE);
1969
			rc = ata_dev_revalidate(dev, readid_flags);
T
Tejun Heo 已提交
1970
			if (rc)
1971
				goto err;
T
Tejun Heo 已提交
1972

T
Tejun Heo 已提交
1973
			ata_eh_done(&ap->link, dev, ATA_EH_REVALIDATE);
1974

1975 1976 1977 1978 1979
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

1980 1981
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
1982 1983 1984 1985 1986
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

1987 1988
			rc = ata_dev_read_id(dev, &dev->class, readid_flags,
					     dev->id);
1989 1990
			switch (rc) {
			case 0:
1991
				new_mask |= 1 << dev->devno;
1992 1993
				break;
			case -ENOENT:
1994 1995 1996 1997 1998
				/* IDENTIFY was issued to non-existent
				 * device.  No need to reset.  Just
				 * thaw and kill the device.
				 */
				ata_eh_thaw_port(ap);
T
Tejun Heo 已提交
1999 2000
				dev->class = ATA_DEV_UNKNOWN;
				break;
2001 2002 2003
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
2004
			}
2005 2006
		}
	}
T
Tejun Heo 已提交
2007

2008 2009 2010 2011
	/* PDIAG- should have been released, ask cable type if post-reset */
	if ((ehc->i.flags & ATA_EHI_DID_RESET) && ap->ops->cable_detect)
		ap->cbl = ap->ops->cable_detect(ap);

2012 2013 2014
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
2015 2016
	ata_link_for_each_dev(dev, &ap->link) {
		if (!(new_mask & (1 << dev->devno)))
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030
			continue;

		ehc->i.flags |= ATA_EHI_PRINTINFO;
		rc = ata_dev_configure(dev);
		ehc->i.flags &= ~ATA_EHI_PRINTINFO;
		if (rc)
			goto err;

		spin_lock_irqsave(ap->lock, flags);
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
		spin_unlock_irqrestore(ap->lock, flags);

		/* new device discovered, configure xfermode */
		ehc->i.flags |= ATA_EHI_SETMODE;
T
Tejun Heo 已提交
2031 2032
	}

2033
	return 0;
T
Tejun Heo 已提交
2034

2035 2036 2037
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
2038 2039 2040 2041 2042
	return rc;
}

static int ata_port_nr_enabled(struct ata_port *ap)
{
2043 2044
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2045

2046 2047
	ata_link_for_each_dev(dev, &ap->link)
		if (ata_dev_enabled(dev))
T
Tejun Heo 已提交
2048 2049 2050 2051
			cnt++;
	return cnt;
}

T
Tejun Heo 已提交
2052 2053
static int ata_port_nr_vacant(struct ata_port *ap)
{
2054 2055
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2056

2057 2058
	ata_link_for_each_dev(dev, &ap->link)
		if (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
2059 2060 2061 2062 2063 2064
			cnt++;
	return cnt;
}

static int ata_eh_skip_recovery(struct ata_port *ap)
{
T
Tejun Heo 已提交
2065
	struct ata_eh_context *ehc = &ap->link.eh_context;
2066
	struct ata_device *dev;
T
Tejun Heo 已提交
2067

2068 2069 2070
	/* thaw frozen port, resume link and recover failed devices */
	if ((ap->pflags & ATA_PFLAG_FROZEN) ||
	    (ehc->i.flags & ATA_EHI_RESUME_LINK) || ata_port_nr_enabled(ap))
T
Tejun Heo 已提交
2071 2072 2073
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
2074
	ata_link_for_each_dev(dev, &ap->link) {
T
Tejun Heo 已提交
2075 2076 2077 2078 2079 2080 2081 2082
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

2083 2084
static void ata_eh_handle_dev_fail(struct ata_device *dev, int err)
{
T
Tejun Heo 已提交
2085
	struct ata_eh_context *ehc = &dev->link->eh_context;
2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100

	ehc->tries[dev->devno]--;

	switch (err) {
	case -ENODEV:
		/* device missing or wrong IDENTIFY data, schedule probing */
		ehc->i.probe_mask |= (1 << dev->devno);
	case -EINVAL:
		/* give it just one more chance */
		ehc->tries[dev->devno] = min(ehc->tries[dev->devno], 1);
	case -EIO:
		if (ehc->tries[dev->devno] == 1) {
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
2101
			sata_down_spd_limit(dev->link);
2102 2103 2104 2105 2106 2107 2108 2109 2110
			ata_down_xfermask_limit(dev, ATA_DNXFER_PIO);
		}
	}

	if (ata_dev_enabled(dev) && !ehc->tries[dev->devno]) {
		/* disable device if it has used up all its chances */
		ata_dev_disable(dev);

		/* detach if offline */
2111
		if (ata_link_offline(dev->link))
2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
			ata_eh_detach_dev(dev);

		/* probe if requested */
		if ((ehc->i.probe_mask & (1 << dev->devno)) &&
		    !(ehc->did_probe_mask & (1 << dev->devno))) {
			ata_eh_detach_dev(dev);
			ata_dev_init(dev);

			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
			ehc->did_probe_mask |= (1 << dev->devno);
			ehc->i.action |= ATA_EH_SOFTRESET;
		}
	} else {
		/* soft didn't work?  be haaaaard */
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			ehc->i.action |= ATA_EH_HARDRESET;
		else
			ehc->i.action |= ATA_EH_SOFTRESET;
	}
}

T
Tejun Heo 已提交
2133 2134 2135
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
2136
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2137 2138 2139 2140 2141 2142
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
T
Tejun Heo 已提交
2143 2144 2145 2146
 *	recover the port and hotplug requests are recorded in
 *	eh_context.  This function executes all the operations with
 *	appropriate retrials and fallbacks to resurrect failed
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
2147 2148 2149 2150 2151 2152 2153
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
2154 2155
static int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset,
			  ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
T
Tejun Heo 已提交
2156 2157
			  ata_postreset_fn_t postreset)
{
T
Tejun Heo 已提交
2158
	struct ata_eh_context *ehc = &ap->link.eh_context;
T
Tejun Heo 已提交
2159
	struct ata_device *dev;
2160
	int rc;
T
Tejun Heo 已提交
2161 2162 2163 2164

	DPRINTK("ENTER\n");

	/* prep for recovery */
2165
	ata_link_for_each_dev(dev, &ap->link) {
T
Tejun Heo 已提交
2166
		ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
T
Tejun Heo 已提交
2167

2168
		/* collect port action mask recorded in dev actions */
2169 2170 2171
		ehc->i.action |=
			ehc->i.dev_action[dev->devno] & ~ATA_EH_PERDEV_MASK;
		ehc->i.dev_action[dev->devno] &= ATA_EH_PERDEV_MASK;
2172

T
Tejun Heo 已提交
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184
		/* process hotplug request */
		if (dev->flags & ATA_DFLAG_DETACH)
			ata_eh_detach_dev(dev);

		if (!ata_dev_enabled(dev) &&
		    ((ehc->i.probe_mask & (1 << dev->devno)) &&
		     !(ehc->did_probe_mask & (1 << dev->devno)))) {
			ata_eh_detach_dev(dev);
			ata_dev_init(dev);
			ehc->did_probe_mask |= (1 << dev->devno);
			ehc->i.action |= ATA_EH_SOFTRESET;
		}
T
Tejun Heo 已提交
2185 2186 2187 2188 2189
	}

 retry:
	rc = 0;

2190
	/* if UNLOADING, finish immediately */
2191
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2192 2193
		goto out;

T
Tejun Heo 已提交
2194
	/* skip EH if possible. */
T
Tejun Heo 已提交
2195
	if (ata_eh_skip_recovery(ap))
T
Tejun Heo 已提交
2196 2197
		ehc->i.action = 0;

2198 2199
	ata_link_for_each_dev(dev, &ap->link)
		ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
T
Tejun Heo 已提交
2200

T
Tejun Heo 已提交
2201 2202 2203 2204
	/* reset */
	if (ehc->i.action & ATA_EH_RESET_MASK) {
		ata_eh_freeze_port(ap);

T
Tejun Heo 已提交
2205 2206
		rc = ata_eh_reset(ap, ata_port_nr_vacant(ap), prereset,
				  softreset, hardreset, postreset);
T
Tejun Heo 已提交
2207 2208 2209 2210 2211 2212 2213 2214 2215
		if (rc) {
			ata_port_printk(ap, KERN_ERR,
					"reset failed, giving up\n");
			goto out;
		}

		ata_eh_thaw_port(ap);
	}

T
Tejun Heo 已提交
2216 2217
	/* revalidate existing devices and attach new ones */
	rc = ata_eh_revalidate_and_attach(ap, &dev);
T
Tejun Heo 已提交
2218 2219 2220
	if (rc)
		goto dev_fail;

2221 2222
	/* configure transfer mode if necessary */
	if (ehc->i.flags & ATA_EHI_SETMODE) {
T
Tejun Heo 已提交
2223
		rc = ata_set_mode(ap, &dev);
2224
		if (rc)
T
Tejun Heo 已提交
2225
			goto dev_fail;
2226
		ehc->i.flags &= ~ATA_EHI_SETMODE;
T
Tejun Heo 已提交
2227 2228 2229 2230 2231
	}

	goto out;

 dev_fail:
2232
	ata_eh_handle_dev_fail(dev, rc);
T
Tejun Heo 已提交
2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246

	if (ata_port_nr_enabled(ap)) {
		ata_port_printk(ap, KERN_WARNING, "failed to recover some "
				"devices, retrying in 5 secs\n");
		ssleep(5);
	} else {
		/* no device left, repeat fast */
		msleep(500);
	}

	goto retry;

 out:
	if (rc) {
2247 2248
		ata_link_for_each_dev(dev, &ap->link);
			ata_dev_disable(dev);
T
Tejun Heo 已提交
2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299
	}

	DPRINTK("EXIT, rc=%d\n", rc);
	return rc;
}

/**
 *	ata_eh_finish - finish up EH
 *	@ap: host port to finish EH for
 *
 *	Recovery is complete.  Clean up EH states and retry or finish
 *	failed qcs.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_finish(struct ata_port *ap)
{
	int tag;

	/* retry or finish qcs */
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);

		if (!(qc->flags & ATA_QCFLAG_FAILED))
			continue;

		if (qc->err_mask) {
			/* FIXME: Once EH migration is complete,
			 * generate sense data in this function,
			 * considering both err_mask and tf.
			 */
			if (qc->err_mask & AC_ERR_INVALID)
				ata_eh_qc_complete(qc);
			else
				ata_eh_qc_retry(qc);
		} else {
			if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
				ata_eh_qc_complete(qc);
			} else {
				/* feed zero TF to sense generation */
				memset(&qc->result_tf, 0, sizeof(qc->result_tf));
				ata_eh_qc_retry(qc);
			}
		}
	}
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2300
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2301 2302 2303 2304 2305 2306 2307 2308 2309
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
 *
 *	Perform standard error handling sequence.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
2310 2311 2312
void ata_do_eh(struct ata_port *ap, ata_prereset_fn_t prereset,
	       ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
	       ata_postreset_fn_t postreset)
T
Tejun Heo 已提交
2313
{
2314 2315
	ata_eh_autopsy(ap);
	ata_eh_report(ap);
2316
	ata_eh_recover(ap, prereset, softreset, hardreset, postreset);
T
Tejun Heo 已提交
2317 2318
	ata_eh_finish(ap);
}
2319

2320
#ifdef CONFIG_PM
2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345
/**
 *	ata_eh_handle_port_suspend - perform port suspend operation
 *	@ap: port to suspend
 *
 *	Suspend @ap.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_handle_port_suspend(struct ata_port *ap)
{
	unsigned long flags;
	int rc = 0;

	/* are we suspending? */
	spin_lock_irqsave(ap->lock, flags);
	if (!(ap->pflags & ATA_PFLAG_PM_PENDING) ||
	    ap->pm_mesg.event == PM_EVENT_ON) {
		spin_unlock_irqrestore(ap->lock, flags);
		return;
	}
	spin_unlock_irqrestore(ap->lock, flags);

	WARN_ON(ap->pflags & ATA_PFLAG_SUSPENDED);

2346 2347 2348 2349 2350
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

2351 2352 2353 2354 2355 2356
	/* suspend */
	ata_eh_freeze_port(ap);

	if (ap->ops->port_suspend)
		rc = ap->ops->port_suspend(ap, ap->pm_mesg);

2357
 out:
2358 2359 2360 2361 2362 2363
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
2364
	else if (ap->pflags & ATA_PFLAG_FROZEN)
2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388
		ata_port_schedule_eh(ap);

	if (ap->pm_result) {
		*ap->pm_result = rc;
		ap->pm_result = NULL;
	}

	spin_unlock_irqrestore(ap->lock, flags);

	return;
}

/**
 *	ata_eh_handle_port_resume - perform port resume operation
 *	@ap: port to resume
 *
 *	Resume @ap.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_handle_port_resume(struct ata_port *ap)
{
	unsigned long flags;
2389
	int rc = 0;
2390 2391 2392 2393 2394 2395 2396 2397 2398 2399

	/* are we resuming? */
	spin_lock_irqsave(ap->lock, flags);
	if (!(ap->pflags & ATA_PFLAG_PM_PENDING) ||
	    ap->pm_mesg.event != PM_EVENT_ON) {
		spin_unlock_irqrestore(ap->lock, flags);
		return;
	}
	spin_unlock_irqrestore(ap->lock, flags);

2400
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
2401 2402 2403 2404

	if (ap->ops->port_resume)
		rc = ap->ops->port_resume(ap);

T
Tejun Heo 已提交
2405 2406 2407
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

2408
	/* report result */
2409 2410 2411 2412 2413 2414 2415 2416
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags &= ~(ATA_PFLAG_PM_PENDING | ATA_PFLAG_SUSPENDED);
	if (ap->pm_result) {
		*ap->pm_result = rc;
		ap->pm_result = NULL;
	}
	spin_unlock_irqrestore(ap->lock, flags);
}
2417
#endif /* CONFIG_PM */