libata-eh.c 73.3 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
/*
 *  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>
36
#include <linux/pci.h>
37 38 39 40 41
#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
42
#include "../scsi/scsi_transport_api.h"
43 44 45 46 47

#include <linux/libata.h>

#include "libata.h"

48
enum {
49
	/* speed down verdicts */
50 51 52
	ATA_EH_SPDN_NCQ_OFF		= (1 << 0),
	ATA_EH_SPDN_SPEED_DOWN		= (1 << 1),
	ATA_EH_SPDN_FALLBACK_TO_PIO	= (1 << 2),
53
	ATA_EH_SPDN_KEEP_ERRORS		= (1 << 3),
54 55 56

	/* error flags */
	ATA_EFLAG_IS_IO			= (1 << 0),
57
	ATA_EFLAG_DUBIOUS_XFER		= (1 << 1),
58 59 60 61 62 63

	/* error categories */
	ATA_ECAT_NONE			= 0,
	ATA_ECAT_ATA_BUS		= 1,
	ATA_ECAT_TOUT_HSM		= 2,
	ATA_ECAT_UNK_DEV		= 3,
64 65 66 67 68
	ATA_ECAT_DUBIOUS_NONE		= 4,
	ATA_ECAT_DUBIOUS_ATA_BUS	= 5,
	ATA_ECAT_DUBIOUS_TOUT_HSM	= 6,
	ATA_ECAT_DUBIOUS_UNK_DEV	= 7,
	ATA_ECAT_NR			= 8,
69 70
};

T
Tejun Heo 已提交
71 72 73 74 75 76
/* 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
77
	ATA_EH_FASTDRAIN_INTERVAL	= 3 * HZ,
T
Tejun Heo 已提交
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
};

/* 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 */
};

94
static void __ata_port_freeze(struct ata_port *ap);
95
#ifdef CONFIG_PM
96 97
static void ata_eh_handle_port_suspend(struct ata_port *ap);
static void ata_eh_handle_port_resume(struct ata_port *ap);
98 99 100 101 102 103 104
#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 */
105

T
Tejun Heo 已提交
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
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;
}

171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
/**
 *	ata_port_desc - append port description
 *	@ap: target ATA port
 *	@fmt: printf format string
 *
 *	Format string according to @fmt and append it to port
 *	description.  If port description is not empty, " " is added
 *	in-between.  This function is to be used while initializing
 *	ata_host.  The description is printed on host registration.
 *
 *	LOCKING:
 *	None.
 */
void ata_port_desc(struct ata_port *ap, const char *fmt, ...)
{
	va_list args;

	WARN_ON(!(ap->pflags & ATA_PFLAG_INITIALIZING));

	if (ap->link.eh_info.desc_len)
		__ata_ehi_push_desc(&ap->link.eh_info, " ");

	va_start(args, fmt);
	__ata_ehi_pushv_desc(&ap->link.eh_info, fmt, args);
	va_end(args);
}

#ifdef CONFIG_PCI

/**
 *	ata_port_pbar_desc - append PCI BAR description
 *	@ap: target ATA port
 *	@bar: target PCI BAR
 *	@offset: offset into PCI BAR
 *	@name: name of the area
 *
 *	If @offset is negative, this function formats a string which
 *	contains the name, address, size and type of the BAR and
 *	appends it to the port description.  If @offset is zero or
 *	positive, only name and offsetted address is appended.
 *
 *	LOCKING:
 *	None.
 */
void ata_port_pbar_desc(struct ata_port *ap, int bar, ssize_t offset,
			const char *name)
{
	struct pci_dev *pdev = to_pci_dev(ap->host->dev);
	char *type = "";
	unsigned long long start, len;

	if (pci_resource_flags(pdev, bar) & IORESOURCE_MEM)
		type = "m";
	else if (pci_resource_flags(pdev, bar) & IORESOURCE_IO)
		type = "i";

	start = (unsigned long long)pci_resource_start(pdev, bar);
	len = (unsigned long long)pci_resource_len(pdev, bar);

	if (offset < 0)
		ata_port_desc(ap, "%s %s%llu@0x%llx", name, type, len, start);
	else
233 234
		ata_port_desc(ap, "%s 0x%llx", name,
				start + (unsigned long long)offset);
235 236 237 238
}

#endif /* CONFIG_PCI */

239
static void ata_ering_record(struct ata_ering *ering, unsigned int eflags,
240 241 242 243 244 245 246 247 248 249
			     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];
250
	ent->eflags = eflags;
251 252 253 254
	ent->err_mask = err_mask;
	ent->timestamp = get_jiffies_64();
}

255 256 257 258 259 260 261 262 263
static struct ata_ering_entry *ata_ering_top(struct ata_ering *ering)
{
	struct ata_ering_entry *ent = &ering->ring[ering->cursor];

	if (ent->err_mask)
		return ent;
	return NULL;
}

264
static void ata_ering_clear(struct ata_ering *ering)
265
{
266
	memset(ering, 0, sizeof(*ering));
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
}

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;
}

290 291
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
292
	struct ata_eh_context *ehc = &dev->link->eh_context;
293 294 295 296

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

297
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
298 299
				struct ata_eh_info *ehi, unsigned int action)
{
300
	struct ata_device *tdev;
301 302 303

	if (!dev) {
		ehi->action &= ~action;
304 305
		ata_link_for_each_dev(tdev, link)
			ehi->dev_action[tdev->devno] &= ~action;
306 307 308 309 310 311
	} 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) {
312 313 314
			ata_link_for_each_dev(tdev, link)
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
315 316 317 318 319 320 321 322
			ehi->action &= ~action;
		}

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

323 324 325 326 327 328 329 330 331 332 333
/**
 *	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.
 *
334 335
 *	TODO: kill this function once old EH is gone.
 *
336 337 338 339 340 341 342 343 344
 *	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;
345
	struct ata_port *ap = ata_shost_to_port(host);
346 347
	unsigned long flags;
	struct ata_queued_cmd *qc;
348
	enum scsi_eh_timer_return ret;
349 350 351

	DPRINTK("ENTER\n");

352 353 354 355 356 357
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
358
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
359
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
360 361 362 363 364 365
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
366
	spin_unlock_irqrestore(ap->lock, flags);
367

368
 out:
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
	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.
 */
385
void ata_scsi_error(struct Scsi_Host *host)
386
{
387
	struct ata_port *ap = ata_shost_to_port(host);
388
	int i;
389
	unsigned long flags;
390 391 392

	DPRINTK("ENTER\n");

393
	/* synchronize with port task */
394 395
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
396
	/* synchronize with host lock and sort out timeouts */
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413

	/* 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;

414
		spin_lock_irqsave(ap->lock, flags);
415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452

		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);

453
		spin_unlock_irqrestore(ap->lock, flags);
454 455 456

		/* initialize eh_tries */
		ap->eh_tries = ATA_EH_MAX_TRIES;
457
	} else
458
		spin_unlock_wait(ap->lock);
459 460 461 462

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
463 464
		struct ata_link *link;

T
Tejun Heo 已提交
465 466 467
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

468 469 470
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

471
		/* fetch & clear EH info */
472
		spin_lock_irqsave(ap->lock, flags);
473

474
		__ata_port_for_each_link(link, ap) {
475 476 477
			struct ata_eh_context *ehc = &link->eh_context;
			struct ata_device *dev;

478 479 480
			memset(&link->eh_context, 0, sizeof(link->eh_context));
			link->eh_context.i = link->eh_info;
			memset(&link->eh_info, 0, sizeof(link->eh_info));
481 482 483 484 485 486 487 488

			ata_link_for_each_dev(dev, link) {
				int devno = dev->devno;

				ehc->saved_xfer_mode[devno] = dev->xfer_mode;
				if (ata_ncq_enabled(dev))
					ehc->saved_ncq_enabled |= 1 << devno;
			}
489
		}
490

491 492
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
493
		ap->excl_link = NULL;	/* don't maintain exclusion over EH */
494

495
		spin_unlock_irqrestore(ap->lock, flags);
496

497 498
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
499 500 501
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
502

503 504 505
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

506 507 508 509
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
510
		spin_lock_irqsave(ap->lock, flags);
511

512
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
513
			if (--ap->eh_tries) {
514
				spin_unlock_irqrestore(ap->lock, flags);
515 516 517
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
518
					"tries, giving up\n", ATA_EH_MAX_TRIES);
T
Tejun Heo 已提交
519
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
520 521
		}

522
		/* this run is complete, make sure EH info is clear */
523 524
		__ata_port_for_each_link(link, ap)
			memset(&link->eh_info, 0, sizeof(link->eh_info));
525

526
		/* Clear host_eh_scheduled while holding ap->lock such
527 528 529 530 531 532
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

533
		spin_unlock_irqrestore(ap->lock, flags);
534
	} else {
T
Tejun Heo 已提交
535
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
536 537
		ap->ops->eng_timeout(ap);
	}
538

539
	/* finish or retry handled scmd's and clean up */
540 541 542 543
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

544
	/* clean up */
545
	spin_lock_irqsave(ap->lock, flags);
546

547
	if (ap->pflags & ATA_PFLAG_LOADING)
548
		ap->pflags &= ~ATA_PFLAG_LOADING;
549
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
550
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
551 552 553

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
554

555
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
556

557
	/* tell wait_eh that we're done */
558
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
559 560
	wake_up_all(&ap->eh_wait_q);

561
	spin_unlock_irqrestore(ap->lock, flags);
562

563 564 565
	DPRINTK("EXIT\n");
}

566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
/**
 *	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:
581
	spin_lock_irqsave(ap->lock, flags);
582

583
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
584
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
585
		spin_unlock_irqrestore(ap->lock, flags);
586
		schedule();
587
		spin_lock_irqsave(ap->lock, flags);
588
	}
589
	finish_wait(&ap->eh_wait_q, &wait);
590

591
	spin_unlock_irqrestore(ap->lock, flags);
592 593

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
594
	if (scsi_host_in_recovery(ap->scsi_host)) {
595 596 597 598 599
		msleep(10);
		goto retry;
	}
}

T
Tejun Heo 已提交
600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687
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);
}

688 689 690 691 692 693 694 695
/**
 *	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
696
 *	spin_lock_irqsave(host lock)
697 698 699 700 701 702 703 704
 */
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
705
	ata_eh_set_pending(ap, 1);
706 707 708 709 710 711 712 713 714

	/* 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);
}

715 716 717 718 719 720 721 722
/**
 *	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
723
 *	spin_lock_irqsave(host lock)
724 725 726 727 728
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

729 730 731
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
732
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
733
	scsi_schedule_eh(ap->scsi_host);
734 735 736 737

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

738
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
739 740 741 742 743
{
	int tag, nr_aborted = 0;

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

T
Tejun Heo 已提交
744 745 746
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

747 748 749
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

750
		if (qc && (!link || qc->dev->link == link)) {
751 752 753 754 755 756 757 758 759 760 761 762
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796
/**
 *	ata_link_abort - abort all qc's on the link
 *	@link: ATA link to abort qc's for
 *
 *	Abort all active qc's active on @link and schedule EH.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_link_abort(struct ata_link *link)
{
	return ata_do_link_abort(link->ap, link);
}

/**
 *	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:
 *	spin_lock_irqsave(host_set lock)
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_port_abort(struct ata_port *ap)
{
	return ata_do_link_abort(ap, NULL);
}

797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
/**
 *	__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
813
 *	spin_lock_irqsave(host lock)
814 815 816 817 818 819 820 821
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

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

822
	ap->pflags |= ATA_PFLAG_FROZEN;
823

T
Tejun Heo 已提交
824
	DPRINTK("ata%u port frozen\n", ap->print_id);
825 826 827 828 829 830 831 832 833
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
834
 *	spin_lock_irqsave(host lock)
835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
 *
 *	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;
}

851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923
/**
 *	sata_async_notification - SATA async notification handler
 *	@ap: ATA port where async notification is received
 *
 *	Handler to be called when async notification via SDB FIS is
 *	received.  This function schedules EH if necessary.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 *
 *	RETURNS:
 *	1 if EH is scheduled, 0 otherwise.
 */
int sata_async_notification(struct ata_port *ap)
{
	u32 sntf;
	int rc;

	if (!(ap->flags & ATA_FLAG_AN))
		return 0;

	rc = sata_scr_read(&ap->link, SCR_NOTIFICATION, &sntf);
	if (rc == 0)
		sata_scr_write(&ap->link, SCR_NOTIFICATION, sntf);

	if (!ap->nr_pmp_links || rc) {
		/* PMP is not attached or SNTF is not available */
		if (!ap->nr_pmp_links) {
			/* PMP is not attached.  Check whether ATAPI
			 * AN is configured.  If so, notify media
			 * change.
			 */
			struct ata_device *dev = ap->link.device;

			if ((dev->class == ATA_DEV_ATAPI) &&
			    (dev->flags & ATA_DFLAG_AN))
				ata_scsi_media_change_notify(dev);
			return 0;
		} else {
			/* PMP is attached but SNTF is not available.
			 * ATAPI async media change notification is
			 * not used.  The PMP must be reporting PHY
			 * status change, schedule EH.
			 */
			ata_port_schedule_eh(ap);
			return 1;
		}
	} else {
		/* PMP is attached and SNTF is available */
		struct ata_link *link;

		/* check and notify ATAPI AN */
		ata_port_for_each_link(link, ap) {
			if (!(sntf & (1 << link->pmp)))
				continue;

			if ((link->device->class == ATA_DEV_ATAPI) &&
			    (link->device->flags & ATA_DFLAG_AN))
				ata_scsi_media_change_notify(link->device);
		}

		/* If PMP is reporting that PHY status of some
		 * downstream ports has changed, schedule EH.
		 */
		if (sntf & (1 << SATA_PMP_CTRL_PORT)) {
			ata_port_schedule_eh(ap);
			return 1;
		}

		return 0;
	}
}

924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939
/**
 *	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;

940
	spin_lock_irqsave(ap->lock, flags);
941
	__ata_port_freeze(ap);
942
	spin_unlock_irqrestore(ap->lock, flags);
943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
}

/**
 *	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;

961
	spin_lock_irqsave(ap->lock, flags);
962

963
	ap->pflags &= ~ATA_PFLAG_FROZEN;
964 965 966 967

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

968
	spin_unlock_irqrestore(ap->lock, flags);
969

T
Tejun Heo 已提交
970
	DPRINTK("ata%u port thawed\n", ap->print_id);
971 972
}

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

984
	spin_lock_irqsave(ap->lock, flags);
985 986 987
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
988
	spin_unlock_irqrestore(ap->lock, flags);
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024

	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
1025

1026 1027 1028 1029 1030 1031 1032 1033 1034
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
1035
void ata_eh_detach_dev(struct ata_device *dev)
1036
{
1037 1038
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1039 1040 1041 1042
	unsigned long flags;

	ata_dev_disable(dev);

1043
	spin_lock_irqsave(ap->lock, flags);
1044 1045 1046 1047 1048

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1049
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1050 1051
	}

1052
	/* clear per-dev EH actions */
1053 1054
	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);
1055

1056
	spin_unlock_irqrestore(ap->lock, flags);
1057 1058
}

T
Tejun Heo 已提交
1059 1060
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
1061
 *	@link: target ATA link
1062
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
1063 1064 1065
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
1066 1067
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
1068 1069 1070 1071
 *
 *	LOCKING:
 *	None.
 */
1072 1073
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
T
Tejun Heo 已提交
1074
{
T
Tejun Heo 已提交
1075 1076 1077
	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
1078 1079
	unsigned long flags;

1080
	spin_lock_irqsave(ap->lock, flags);
1081

T
Tejun Heo 已提交
1082
	ata_eh_clear_action(link, dev, ehi, action);
1083

1084
	if (!(ehc->i.flags & ATA_EHI_QUIET))
1085 1086
		ap->pflags |= ATA_PFLAG_RECOVERED;

1087
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
1088 1089
}

1090 1091
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
1092
*	@ap: target ATA port
1093 1094 1095 1096
 *	@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
1097
 *	in @link->eh_context.
1098 1099 1100 1101
 *
 *	LOCKING:
 *	None.
 */
1102 1103
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1104
{
T
Tejun Heo 已提交
1105
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1106

T
Tejun Heo 已提交
1107
	ata_eh_clear_action(link, dev, &ehc->i, action);
1108 1109
}

T
Tejun Heo 已提交
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
/**
 *	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
 */
1124
static const char *ata_err_string(unsigned int err_mask)
T
Tejun Heo 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
{
	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";
}

1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176
/**
 *	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,
1177
				     buf, sectors * ATA_SECT_SIZE, 0);
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200

	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
1201
	u8 *buf = dev->link->ap->sector_buf;
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
	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
1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
/**
 *	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
 */
1251
static unsigned int atapi_eh_request_sense(struct ata_queued_cmd *qc)
T
Tejun Heo 已提交
1252
{
1253 1254
	struct ata_device *dev = qc->dev;
	unsigned char *sense_buf = qc->scsicmd->sense_buffer;
T
Tejun Heo 已提交
1255
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1256 1257 1258 1259 1260 1261 1262 1263
	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);

1264 1265 1266
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1267
	sense_buf[0] = 0x70;
1268 1269
	sense_buf[2] = qc->result_tf.feature >> 4;

J
Jeff Garzik 已提交
1270
	/* some devices time out if garbage left in tf */
1271
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281

	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) {
1282
		tf.protocol = ATAPI_PROT_DMA;
T
Tejun Heo 已提交
1283 1284
		tf.feature |= ATAPI_PKT_DMA;
	} else {
1285
		tf.protocol = ATAPI_PROT_PIO;
1286 1287
		tf.lbam = SCSI_SENSE_BUFFERSIZE;
		tf.lbah = 0;
T
Tejun Heo 已提交
1288 1289 1290
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
1291
				 sense_buf, SCSI_SENSE_BUFFERSIZE, 0);
T
Tejun Heo 已提交
1292 1293 1294 1295
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1296
 *	@link: ATA link to analyze SError for
T
Tejun Heo 已提交
1297 1298 1299 1300 1301 1302 1303
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1304
static void ata_eh_analyze_serror(struct ata_link *link)
T
Tejun Heo 已提交
1305
{
1306
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1307 1308
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1309
	u32 hotplug_mask;
T
Tejun Heo 已提交
1310 1311 1312

	if (serror & SERR_PERSISTENT) {
		err_mask |= AC_ERR_ATA_BUS;
T
Tejun Heo 已提交
1313
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1314 1315 1316 1317
	}
	if (serror &
	    (SERR_DATA_RECOVERED | SERR_COMM_RECOVERED | SERR_DATA)) {
		err_mask |= AC_ERR_ATA_BUS;
T
Tejun Heo 已提交
1318
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1319 1320 1321
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
T
Tejun Heo 已提交
1322
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1323 1324 1325
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
T
Tejun Heo 已提交
1326
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1327
	}
1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341

	/* Determine whether a hotplug event has occurred.  Both
	 * SError.N/X are considered hotplug events for enabled or
	 * host links.  For disabled PMP links, only N bit is
	 * considered as X bit is left at 1 for link plugging.
	 */
	hotplug_mask = 0;

	if (!(link->flags & ATA_LFLAG_DISABLED) || ata_is_host_link(link))
		hotplug_mask = SERR_PHYRDY_CHG | SERR_DEV_XCHG;
	else
		hotplug_mask = SERR_PHYRDY_CHG;

	if (serror & hotplug_mask)
T
Tejun Heo 已提交
1342
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1343 1344 1345 1346 1347

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

1348 1349
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
1350
 *	@link: ATA link to analyze NCQ error for
1351 1352 1353 1354 1355 1356 1357 1358 1359
 *
 *	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).
 */
1360
static void ata_eh_analyze_ncq_error(struct ata_link *link)
1361
{
1362 1363 1364
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
1365 1366 1367 1368 1369
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1370
	if (ap->pflags & ATA_PFLAG_FROZEN)
1371 1372 1373
		return;

	/* is it NCQ device error? */
1374
	if (!link->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
		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) {
1391
		ata_link_printk(link, KERN_ERR, "failed to read log page 10h "
1392 1393 1394 1395
				"(errno=%d)\n", rc);
		return;
	}

1396 1397
	if (!(link->sactive & (1 << tag))) {
		ata_link_printk(link, KERN_ERR, "log page 10h reported "
1398 1399 1400 1401 1402 1403 1404
				"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
1405
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1406 1407 1408
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 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
/**
 *	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;
T
Tejun Heo 已提交
1432
		return ATA_EH_RESET;
T
Tejun Heo 已提交
1433 1434
	}

1435 1436 1437
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
		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:
1451
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1452
			tmp = atapi_eh_request_sense(qc);
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
			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
1465 1466 1467
	}

	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
T
Tejun Heo 已提交
1468
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1469 1470 1471 1472

	return action;
}

1473 1474
static int ata_eh_categorize_error(unsigned int eflags, unsigned int err_mask,
				   int *xfer_ok)
T
Tejun Heo 已提交
1475
{
1476 1477 1478 1479 1480 1481
	int base = 0;

	if (!(eflags & ATA_EFLAG_DUBIOUS_XFER))
		*xfer_ok = 1;

	if (!*xfer_ok)
1482
		base = ATA_ECAT_DUBIOUS_NONE;
1483

1484
	if (err_mask & AC_ERR_ATA_BUS)
1485
		return base + ATA_ECAT_ATA_BUS;
T
Tejun Heo 已提交
1486

1487
	if (err_mask & AC_ERR_TIMEOUT)
1488
		return base + ATA_ECAT_TOUT_HSM;
1489

1490
	if (eflags & ATA_EFLAG_IS_IO) {
1491
		if (err_mask & AC_ERR_HSM)
1492
			return base + ATA_ECAT_TOUT_HSM;
1493 1494
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
1495
			return base + ATA_ECAT_UNK_DEV;
T
Tejun Heo 已提交
1496 1497 1498 1499 1500
	}

	return 0;
}

1501
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1502
	u64 since;
1503
	int xfer_ok;
1504
	int nr_errors[ATA_ECAT_NR];
T
Tejun Heo 已提交
1505 1506
};

1507
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1508
{
1509
	struct speed_down_verdict_arg *arg = void_arg;
1510
	int cat;
T
Tejun Heo 已提交
1511 1512 1513 1514

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

1515 1516
	cat = ata_eh_categorize_error(ent->eflags, ent->err_mask,
				      &arg->xfer_ok);
1517
	arg->nr_errors[cat]++;
1518

T
Tejun Heo 已提交
1519 1520 1521 1522
	return 0;
}

/**
1523
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1524 1525 1526
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1527 1528
 *	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
1529
 *
1530 1531 1532 1533 1534 1535 1536
 *	ECAT_ATA_BUS	: ATA_BUS error for any command
 *
 *	ECAT_TOUT_HSM	: TIMEOUT for any command or HSM violation for
 *			  IO commands
 *
 *	ECAT_UNK_DEV	: Unknown DEV error for IO commands
 *
1537 1538 1539
 *	ECAT_DUBIOUS_*	: Identical to above three but occurred while
 *			  data transfer hasn't been verified.
 *
1540 1541 1542
 *	Verdicts are
 *
 *	NCQ_OFF		: Turn off NCQ.
T
Tejun Heo 已提交
1543
 *
1544 1545
 *	SPEED_DOWN	: Speed down transfer speed but don't fall back
 *			  to PIO.
1546
 *
1547
 *	FALLBACK_TO_PIO	: Fall back to PIO.
T
Tejun Heo 已提交
1548
 *
1549
 *	Even if multiple verdicts are returned, only one action is
1550 1551 1552 1553 1554 1555 1556
 *	taken per error.  An action triggered by non-DUBIOUS errors
 *	clears ering, while one triggered by DUBIOUS_* errors doesn't.
 *	This is to expedite speed down decisions right after device is
 *	initially configured.
 *
 *	The followings are speed down rules.  #1 and #2 deal with
 *	DUBIOUS errors.
1557
 *
1558 1559 1560 1561 1562 1563 1564
 *	1. If more than one DUBIOUS_ATA_BUS or DUBIOUS_TOUT_HSM errors
 *	   occurred during last 5 mins, SPEED_DOWN and FALLBACK_TO_PIO.
 *
 *	2. If more than one DUBIOUS_TOUT_HSM or DUBIOUS_UNK_DEV errors
 *	   occurred during last 5 mins, NCQ_OFF.
 *
 *	3. If more than 8 ATA_BUS, TOUT_HSM or UNK_DEV errors
1565
 *	   ocurred during last 5 mins, FALLBACK_TO_PIO
1566
 *
1567
 *	4. If more than 3 TOUT_HSM or UNK_DEV errors occurred
1568 1569
 *	   during last 10 mins, NCQ_OFF.
 *
1570
 *	5. If more than 3 ATA_BUS or TOUT_HSM errors, or more than 6
1571
 *	   UNK_DEV errors occurred during last 10 mins, SPEED_DOWN.
1572
 *
T
Tejun Heo 已提交
1573 1574 1575 1576
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1577
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1578
 */
1579
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1580
{
1581 1582 1583 1584
	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
1585

1586
	/* scan past 5 mins of error history */
1587
	memset(&arg, 0, sizeof(arg));
1588
	arg.since = j64 - min(j64, j5mins);
1589
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1590

1591 1592 1593 1594 1595 1596 1597 1598 1599
	if (arg.nr_errors[ATA_ECAT_DUBIOUS_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_DUBIOUS_TOUT_HSM] > 1)
		verdict |= ATA_EH_SPDN_SPEED_DOWN |
			ATA_EH_SPDN_FALLBACK_TO_PIO | ATA_EH_SPDN_KEEP_ERRORS;

	if (arg.nr_errors[ATA_ECAT_DUBIOUS_TOUT_HSM] +
	    arg.nr_errors[ATA_ECAT_DUBIOUS_UNK_DEV] > 1)
		verdict |= ATA_EH_SPDN_NCQ_OFF | ATA_EH_SPDN_KEEP_ERRORS;

1600 1601
	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] +
T
Tejun Heo 已提交
1602
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1603
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
T
Tejun Heo 已提交
1604

1605
	/* scan past 10 mins of error history */
T
Tejun Heo 已提交
1606
	memset(&arg, 0, sizeof(arg));
1607
	arg.since = j64 - min(j64, j10mins);
1608
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1609

1610 1611 1612 1613 1614 1615
	if (arg.nr_errors[ATA_ECAT_TOUT_HSM] +
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 3)
		verdict |= ATA_EH_SPDN_NCQ_OFF;

	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] > 3 ||
T
Tejun Heo 已提交
1616
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1617
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
T
Tejun Heo 已提交
1618

1619
	return verdict;
T
Tejun Heo 已提交
1620 1621 1622 1623 1624
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
1625
 *	@eflags: mask of ATA_EFLAG_* flags
T
Tejun Heo 已提交
1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636
 *	@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:
1637
 *	Determined recovery action.
T
Tejun Heo 已提交
1638
 */
1639 1640
static unsigned int ata_eh_speed_down(struct ata_device *dev,
				unsigned int eflags, unsigned int err_mask)
T
Tejun Heo 已提交
1641
{
1642
	struct ata_link *link = dev->link;
1643
	int xfer_ok = 0;
1644 1645 1646 1647
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
1648
	if (ata_eh_categorize_error(eflags, err_mask, &xfer_ok) == 0)
T
Tejun Heo 已提交
1649 1650 1651
		return 0;

	/* record error and determine whether speed down is necessary */
1652
	ata_ering_record(&dev->ering, eflags, err_mask);
1653
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1654

1655 1656 1657 1658 1659 1660 1661 1662 1663
	/* 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
1664

1665 1666 1667
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1668
		if (sata_down_spd_limit(link) == 0) {
T
Tejun Heo 已提交
1669
			action |= ATA_EH_RESET;
1670 1671
			goto done;
		}
T
Tejun Heo 已提交
1672

1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
		/* 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) {
T
Tejun Heo 已提交
1689
				action |= ATA_EH_RESET;
1690 1691 1692 1693 1694 1695
				goto done;
			}
		}
	}

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
T
Tejun Heo 已提交
1696
	 * SATA ATA devices.  Consider it only for PATA and SATAPI.
1697 1698
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
T
Tejun Heo 已提交
1699
	    (link->ap->cbl != ATA_CBL_SATA || dev->class == ATA_DEV_ATAPI) &&
1700 1701 1702
	    (dev->xfer_shift != ATA_SHIFT_PIO)) {
		if (ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO) == 0) {
			dev->spdn_cnt = 0;
T
Tejun Heo 已提交
1703
			action |= ATA_EH_RESET;
1704 1705 1706
			goto done;
		}
	}
T
Tejun Heo 已提交
1707 1708

	return 0;
1709 1710
 done:
	/* device has been slowed down, blow error history */
1711 1712
	if (!(verdict & ATA_EH_SPDN_KEEP_ERRORS))
		ata_ering_clear(&dev->ering);
1713
	return action;
T
Tejun Heo 已提交
1714 1715 1716
}

/**
1717 1718
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
T
Tejun Heo 已提交
1719
 *
1720 1721 1722
 *	Analyze why @link failed and determine which recovery actions
 *	are needed.  This function also sets more detailed AC_ERR_*
 *	values and fills sense data for ATAPI CHECK SENSE.
T
Tejun Heo 已提交
1723 1724 1725 1726
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1727
static void ata_eh_link_autopsy(struct ata_link *link)
T
Tejun Heo 已提交
1728
{
1729
	struct ata_port *ap = link->ap;
1730
	struct ata_eh_context *ehc = &link->eh_context;
1731
	struct ata_device *dev;
1732 1733
	unsigned int all_err_mask = 0, eflags = 0;
	int tag;
T
Tejun Heo 已提交
1734 1735 1736 1737 1738
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1739 1740 1741
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1742
	/* obtain and analyze SError */
1743
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
1744 1745
	if (rc == 0) {
		ehc->i.serror |= serror;
1746
		ata_eh_analyze_serror(link);
1747
	} else if (rc != -EOPNOTSUPP) {
T
Tejun Heo 已提交
1748
		/* SError read failed, force reset and probing */
1749
		ehc->i.probe_mask |= ATA_ALL_DEVICES;
T
Tejun Heo 已提交
1750
		ehc->i.action |= ATA_EH_RESET;
1751 1752
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
1753

1754
	/* analyze NCQ failure */
1755
	ata_eh_analyze_ncq_error(link);
1756

T
Tejun Heo 已提交
1757 1758 1759 1760 1761 1762 1763 1764 1765
	/* 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);

1766
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link)
T
Tejun Heo 已提交
1767 1768 1769 1770 1771 1772
			continue;

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

		/* analyze TF */
1773
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784

		/* 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 */
1785
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
T
Tejun Heo 已提交
1786 1787
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);

T
Tejun Heo 已提交
1788 1789 1790 1791 1792
		/* determine whether the command is worth retrying */
		if (!(qc->err_mask & AC_ERR_INVALID) &&
		    ((qc->flags & ATA_QCFLAG_IO) || qc->err_mask != AC_ERR_DEV))
			qc->flags |= ATA_QCFLAG_RETRY;

T
Tejun Heo 已提交
1793
		/* accumulate error info */
1794
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1795 1796
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
1797
			eflags |= ATA_EFLAG_IS_IO;
T
Tejun Heo 已提交
1798 1799
	}

1800
	/* enforce default EH actions */
1801
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1802
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
T
Tejun Heo 已提交
1803
		ehc->i.action |= ATA_EH_RESET;
1804 1805
	else if (((eflags & ATA_EFLAG_IS_IO) && all_err_mask) ||
		 (!(eflags & ATA_EFLAG_IS_IO) && (all_err_mask & ~AC_ERR_DEV)))
1806
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1807

1808 1809 1810
	/* If we have offending qcs and the associated failed device,
	 * perform per-dev EH action only on the offending device.
	 */
1811 1812 1813 1814
	if (ehc->i.dev) {
		ehc->i.dev_action[ehc->i.dev->devno] |=
			ehc->i.action & ATA_EH_PERDEV_MASK;
		ehc->i.action &= ~ATA_EH_PERDEV_MASK;
1815 1816
	}

1817 1818 1819 1820 1821
	/* propagate timeout to host link */
	if ((all_err_mask & AC_ERR_TIMEOUT) && !ata_is_host_link(link))
		ap->link.eh_context.i.err_mask |= AC_ERR_TIMEOUT;

	/* record error and consider speeding down */
1822
	dev = ehc->i.dev;
1823 1824 1825
	if (!dev && ((ata_link_max_devices(link) == 1 &&
		      ata_dev_enabled(link->device))))
	    dev = link->device;
1826

1827 1828 1829
	if (dev) {
		if (dev->flags & ATA_DFLAG_DUBIOUS_XFER)
			eflags |= ATA_EFLAG_DUBIOUS_XFER;
1830
		ehc->i.action |= ata_eh_speed_down(dev, eflags, all_err_mask);
1831
	}
1832

T
Tejun Heo 已提交
1833 1834 1835 1836
	DPRINTK("EXIT\n");
}

/**
1837 1838 1839 1840 1841 1842 1843 1844 1845
 *	ata_eh_autopsy - analyze error and determine recovery action
 *	@ap: host port to perform autopsy on
 *
 *	Analyze all links of @ap and determine why they failed and
 *	which recovery actions are needed.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1846
void ata_eh_autopsy(struct ata_port *ap)
1847 1848 1849
{
	struct ata_link *link;

1850
	ata_port_for_each_link(link, ap)
1851
		ata_eh_link_autopsy(link);
1852 1853 1854 1855 1856 1857

	/* Autopsy of fanout ports can affect host link autopsy.
	 * Perform host link autopsy last.
	 */
	if (ap->nr_pmp_links)
		ata_eh_link_autopsy(&ap->link);
1858 1859 1860 1861
}

/**
 *	ata_eh_link_report - report error handling to user
1862
 *	@link: ATA link EH is going on
T
Tejun Heo 已提交
1863 1864 1865 1866 1867 1868
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
1869
static void ata_eh_link_report(struct ata_link *link)
T
Tejun Heo 已提交
1870
{
1871 1872
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1873
	const char *frozen, *desc;
1874
	char tries_buf[6];
T
Tejun Heo 已提交
1875 1876
	int tag, nr_failed = 0;

1877 1878 1879
	if (ehc->i.flags & ATA_EHI_QUIET)
		return;

T
Tejun Heo 已提交
1880 1881 1882 1883 1884 1885 1886
	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);

1887 1888 1889
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link ||
		    ((qc->flags & ATA_QCFLAG_QUIET) &&
		     qc->err_mask == AC_ERR_DEV))
T
Tejun Heo 已提交
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

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

	frozen = "";
1901
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
1902 1903
		frozen = " frozen";

1904 1905 1906 1907 1908
	memset(tries_buf, 0, sizeof(tries_buf));
	if (ap->eh_tries < ATA_EH_MAX_TRIES)
		snprintf(tries_buf, sizeof(tries_buf) - 1, " t%d",
			 ap->eh_tries);

T
Tejun Heo 已提交
1909
	if (ehc->i.dev) {
1910
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
1911 1912 1913
			       "SAct 0x%x SErr 0x%x action 0x%x%s%s\n",
			       ehc->i.err_mask, link->sactive, ehc->i.serror,
			       ehc->i.action, frozen, tries_buf);
T
Tejun Heo 已提交
1914
		if (desc)
T
Tejun Heo 已提交
1915
			ata_dev_printk(ehc->i.dev, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1916
	} else {
1917
		ata_link_printk(link, KERN_ERR, "exception Emask 0x%x "
1918 1919 1920
				"SAct 0x%x SErr 0x%x action 0x%x%s%s\n",
				ehc->i.err_mask, link->sactive, ehc->i.serror,
				ehc->i.action, frozen, tries_buf);
T
Tejun Heo 已提交
1921
		if (desc)
1922
			ata_link_printk(link, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1923 1924
	}

1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943
	if (ehc->i.serror)
		ata_port_printk(ap, KERN_ERR,
		  "SError: { %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s}\n",
		  ehc->i.serror & SERR_DATA_RECOVERED ? "RecovData " : "",
		  ehc->i.serror & SERR_COMM_RECOVERED ? "RecovComm " : "",
		  ehc->i.serror & SERR_DATA ? "UnrecovData " : "",
		  ehc->i.serror & SERR_PERSISTENT ? "Persist " : "",
		  ehc->i.serror & SERR_PROTOCOL ? "Proto " : "",
		  ehc->i.serror & SERR_INTERNAL ? "HostInt " : "",
		  ehc->i.serror & SERR_PHYRDY_CHG ? "PHYRdyChg " : "",
		  ehc->i.serror & SERR_PHY_INT_ERR ? "PHYInt " : "",
		  ehc->i.serror & SERR_COMM_WAKE ? "CommWake " : "",
		  ehc->i.serror & SERR_10B_8B_ERR ? "10B8B " : "",
		  ehc->i.serror & SERR_DISPARITY ? "Dispar " : "",
		  ehc->i.serror & SERR_CRC ? "BadCRC " : "",
		  ehc->i.serror & SERR_HANDSHAKE ? "Handshk " : "",
		  ehc->i.serror & SERR_LINK_SEQ_ERR ? "LinkSeq " : "",
		  ehc->i.serror & SERR_TRANS_ST_ERROR ? "TrStaTrns " : "",
		  ehc->i.serror & SERR_UNRECOG_FIS ? "UnrecFIS " : "",
1944
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "");
1945

T
Tejun Heo 已提交
1946 1947
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
1948
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
1949 1950 1951
		const u8 *cdb = qc->cdb;
		char data_buf[20] = "";
		char cdb_buf[70] = "";
T
Tejun Heo 已提交
1952

1953 1954
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    qc->dev->link != link || !qc->err_mask)
T
Tejun Heo 已提交
1955 1956
			continue;

1957 1958 1959 1960 1961 1962 1963 1964 1965 1966
		if (qc->dma_dir != DMA_NONE) {
			static const char *dma_str[] = {
				[DMA_BIDIRECTIONAL]	= "bidi",
				[DMA_TO_DEVICE]		= "out",
				[DMA_FROM_DEVICE]	= "in",
			};
			static const char *prot_str[] = {
				[ATA_PROT_PIO]		= "pio",
				[ATA_PROT_DMA]		= "dma",
				[ATA_PROT_NCQ]		= "ncq",
1967 1968
				[ATAPI_PROT_PIO]	= "pio",
				[ATAPI_PROT_DMA]	= "dma",
1969 1970 1971 1972 1973 1974 1975
			};

			snprintf(data_buf, sizeof(data_buf), " %s %u %s",
				 prot_str[qc->tf.protocol], qc->nbytes,
				 dma_str[qc->dma_dir]);
		}

1976
		if (ata_is_atapi(qc->tf.protocol))
1977 1978 1979 1980 1981 1982 1983 1984
			snprintf(cdb_buf, sizeof(cdb_buf),
				 "cdb %02x %02x %02x %02x %02x %02x %02x %02x  "
				 "%02x %02x %02x %02x %02x %02x %02x %02x\n         ",
				 cdb[0], cdb[1], cdb[2], cdb[3],
				 cdb[4], cdb[5], cdb[6], cdb[7],
				 cdb[8], cdb[9], cdb[10], cdb[11],
				 cdb[12], cdb[13], cdb[14], cdb[15]);

1985 1986
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
1987
			"tag %d%s\n         %s"
1988
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
1989
			"Emask 0x%x (%s)%s\n",
1990 1991 1992 1993
			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,
1994
			cmd->device, qc->tag, data_buf, cdb_buf,
1995 1996 1997 1998
			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
1999 2000
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
2001 2002

		if (res->command & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
2003
				    ATA_ERR)) {
2004 2005
			if (res->command & ATA_BUSY)
				ata_dev_printk(qc->dev, KERN_ERR,
2006
				  "status: { Busy }\n");
2007 2008 2009 2010 2011 2012
			else
				ata_dev_printk(qc->dev, KERN_ERR,
				  "status: { %s%s%s%s}\n",
				  res->command & ATA_DRDY ? "DRDY " : "",
				  res->command & ATA_DF ? "DF " : "",
				  res->command & ATA_DRQ ? "DRQ " : "",
2013
				  res->command & ATA_ERR ? "ERR " : "");
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
		}

		if (cmd->command != ATA_CMD_PACKET &&
		    (res->feature & (ATA_ICRC | ATA_UNC | ATA_IDNF |
				     ATA_ABORTED)))
			ata_dev_printk(qc->dev, KERN_ERR,
			  "error: { %s%s%s%s}\n",
			  res->feature & ATA_ICRC ? "ICRC " : "",
			  res->feature & ATA_UNC ? "UNC " : "",
			  res->feature & ATA_IDNF ? "IDNF " : "",
2024
			  res->feature & ATA_ABORTED ? "ABRT " : "");
T
Tejun Heo 已提交
2025 2026 2027
	}
}

2028 2029 2030 2031 2032 2033 2034 2035 2036
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2037
void ata_eh_report(struct ata_port *ap)
2038 2039 2040 2041 2042 2043 2044
{
	struct ata_link *link;

	__ata_port_for_each_link(link, ap)
		ata_eh_link_report(link);
}

T
Tejun Heo 已提交
2045
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
2046
			unsigned int *classes, unsigned long deadline)
2047
{
2048 2049
	struct ata_device *dev;
	int rc;
2050

T
Tejun Heo 已提交
2051
	ata_link_for_each_dev(dev, link)
2052
		classes[dev->devno] = ATA_DEV_UNKNOWN;
2053

T
Tejun Heo 已提交
2054
	rc = reset(link, classes, deadline);
2055 2056 2057 2058 2059 2060 2061
	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.
	 */
T
Tejun Heo 已提交
2062
	ata_link_for_each_dev(dev, link)
2063
		if (classes[dev->devno] != ATA_DEV_UNKNOWN)
2064 2065
			break;

2066
	if (dev) {
T
Tejun Heo 已提交
2067
		ata_link_for_each_dev(dev, link) {
2068 2069 2070 2071
			if (classes[dev->devno] == ATA_DEV_UNKNOWN)
				classes[dev->devno] = ATA_DEV_NONE;
		}
	}
2072 2073 2074 2075

	return 0;
}

2076 2077
static int ata_eh_followup_srst_needed(struct ata_link *link,
				       int rc, int classify,
2078 2079
				       const unsigned int *classes)
{
2080 2081
	if (link->flags & ATA_LFLAG_NO_SRST)
		return 0;
2082 2083 2084 2085
	if (rc == -EAGAIN)
		return 1;
	if (rc != 0)
		return 0;
2086 2087
	if ((link->ap->flags & ATA_FLAG_PMP) && ata_is_host_link(link))
		return 1;
2088 2089
	if (classify && !(link->flags & ATA_LFLAG_ASSUME_CLASS) &&
	    classes[0] == ATA_DEV_UNKNOWN)
2090 2091 2092 2093
		return 1;
	return 0;
}

2094 2095 2096
int ata_eh_reset(struct ata_link *link, int classify,
		 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
2097
{
2098
	const int max_tries = ARRAY_SIZE(ata_eh_reset_timeouts);
T
Tejun Heo 已提交
2099
	struct ata_port *ap = link->ap;
2100
	struct ata_eh_context *ehc = &link->eh_context;
2101
	unsigned int *classes = ehc->classes;
2102
	unsigned int lflags = link->flags;
2103
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
T
Tejun Heo 已提交
2104
	int try = 0;
2105
	struct ata_device *dev;
2106
	unsigned long deadline, now;
T
Tejun Heo 已提交
2107
	ata_reset_fn_t reset;
T
Tejun Heo 已提交
2108
	unsigned long flags;
2109
	u32 sstatus;
2110
	int rc;
T
Tejun Heo 已提交
2111

2112
	/* about to reset */
T
Tejun Heo 已提交
2113 2114 2115 2116
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags |= ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

T
Tejun Heo 已提交
2117
	ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
2118

T
Tejun Heo 已提交
2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137
	ata_link_for_each_dev(dev, link) {
		/* If we issue an SRST then an ATA drive (not ATAPI)
		 * may change configuration and be in PIO0 timing. If
		 * we do a hard reset (or are coming from power on)
		 * this is true for ATA or ATAPI. Until we've set a
		 * suitable controller mode we should not touch the
		 * bus as we may be talking too fast.
		 */
		dev->pio_mode = XFER_PIO_0;

		/* If the controller has a pio mode setup function
		 * then use it to set the chipset to rights. Don't
		 * touch the DMA setup as that will be dealt with when
		 * configuring devices.
		 */
		if (ap->ops->set_piomode)
			ap->ops->set_piomode(ap, dev);
	}

2138 2139 2140 2141 2142 2143 2144 2145 2146
	if (!softreset && !hardreset) {
		if (verbose)
			ata_link_printk(link, KERN_INFO, "no reset method "
					"available, skipping reset\n");
		if (!(lflags & ATA_LFLAG_ASSUME_CLASS))
			lflags |= ATA_LFLAG_ASSUME_ATA;
		goto done;
	}

T
Tejun Heo 已提交
2147 2148 2149 2150 2151 2152 2153 2154 2155
	/* prefer hardreset */
	ehc->i.action &= ~ATA_EH_RESET;
	if (hardreset) {
		reset = hardreset;
		ehc->i.action = ATA_EH_HARDRESET;
	} else {
		reset = softreset;
		ehc->i.action = ATA_EH_SOFTRESET;
	}
2156 2157

	if (prereset) {
T
Tejun Heo 已提交
2158
		rc = prereset(link, jiffies + ATA_EH_PRERESET_TIMEOUT);
2159
		if (rc) {
2160
			if (rc == -ENOENT) {
T
Tejun Heo 已提交
2161
				ata_link_printk(link, KERN_DEBUG,
2162
						"port disabled. ignoring.\n");
T
Tejun Heo 已提交
2163
				ehc->i.action &= ~ATA_EH_RESET;
2164

2165
				ata_link_for_each_dev(dev, link)
2166
					classes[dev->devno] = ATA_DEV_NONE;
2167 2168

				rc = 0;
2169
			} else
T
Tejun Heo 已提交
2170
				ata_link_printk(link, KERN_ERR,
2171
					"prereset failed (errno=%d)\n", rc);
2172
			goto out;
2173 2174 2175
		}
	}

T
Tejun Heo 已提交
2176 2177
	/* prereset() might have cleared ATA_EH_RESET */
	if (!(ehc->i.action & ATA_EH_RESET)) {
2178
		/* prereset told us not to reset, bang classes and return */
2179
		ata_link_for_each_dev(dev, link)
2180
			classes[dev->devno] = ATA_DEV_NONE;
2181 2182
		rc = 0;
		goto out;
2183 2184
	}

T
Tejun Heo 已提交
2185
 retry:
T
Tejun Heo 已提交
2186 2187
	deadline = jiffies + ata_eh_reset_timeouts[try++];

2188 2189
	/* shut up during boot probing */
	if (verbose)
T
Tejun Heo 已提交
2190
		ata_link_printk(link, KERN_INFO, "%s resetting link\n",
2191
				reset == softreset ? "soft" : "hard");
T
Tejun Heo 已提交
2192

2193
	/* mark that this EH session started with reset */
2194 2195 2196 2197
	if (reset == hardreset)
		ehc->i.flags |= ATA_EHI_DID_HARDRESET;
	else
		ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
2198

T
Tejun Heo 已提交
2199
	rc = ata_do_reset(link, reset, classes, deadline);
T
Tejun Heo 已提交
2200

2201
	if (reset == hardreset &&
2202
	    ata_eh_followup_srst_needed(link, rc, classify, classes)) {
2203 2204 2205 2206
		/* okay, let's do follow-up softreset */
		reset = softreset;

		if (!reset) {
T
Tejun Heo 已提交
2207
			ata_link_printk(link, KERN_ERR,
2208 2209
					"follow-up softreset required "
					"but no softreset avaliable\n");
2210
			rc = -EINVAL;
2211
			goto fail;
2212 2213
		}

T
Tejun Heo 已提交
2214
		ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
T
Tejun Heo 已提交
2215
		rc = ata_do_reset(link, reset, classes, deadline);
2216 2217
	}

2218 2219 2220
	/* -EAGAIN can happen if we skipped followup SRST */
	if (rc && rc != -EAGAIN)
		goto fail;
2221

2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236
	/* was classification successful? */
	if (classify && classes[0] == ATA_DEV_UNKNOWN &&
	    !(lflags & ATA_LFLAG_ASSUME_CLASS)) {
		if (try < max_tries) {
			ata_link_printk(link, KERN_WARNING,
					"classification failed\n");
			rc = -EINVAL;
			goto fail;
		}

		ata_link_printk(link, KERN_WARNING,
				"classfication failed, assuming ATA\n");
		lflags |= ATA_LFLAG_ASSUME_ATA;
	}

2237
 done:
2238 2239 2240 2241 2242 2243 2244
	ata_link_for_each_dev(dev, link) {
		/* After the reset, the device state is PIO 0 and the
		 * controller state is undefined.  Reset also wakes up
		 * drives from sleeping mode.
		 */
		dev->pio_mode = XFER_PIO_0;
		dev->flags &= ~ATA_DFLAG_SLEEPING;
T
Tejun Heo 已提交
2245

2246 2247
		if (ata_link_offline(link))
			continue;
T
Tejun Heo 已提交
2248

2249
		/* apply class override */
2250 2251 2252 2253
		if (lflags & ATA_LFLAG_ASSUME_ATA)
			classes[dev->devno] = ATA_DEV_ATA;
		else if (lflags & ATA_LFLAG_ASSUME_SEMB)
			classes[dev->devno] = ATA_DEV_SEMB_UNSUP; /* not yet */
T
Tejun Heo 已提交
2254 2255
	}

2256 2257 2258
	/* record current link speed */
	if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
		link->sata_spd = (sstatus >> 4) & 0xf;
2259

2260 2261
	if (postreset)
		postreset(link, classes);
2262

2263
	/* reset successful, schedule revalidation */
T
Tejun Heo 已提交
2264
	ata_eh_done(link, NULL, ATA_EH_RESET);
2265
	ehc->i.action |= ATA_EH_REVALIDATE;
2266

2267
	rc = 0;
2268 2269 2270
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2271 2272 2273 2274 2275

	spin_lock_irqsave(ap->lock, flags);
	ap->pflags &= ~ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

T
Tejun Heo 已提交
2276
	return rc;
2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298

 fail:
	if (rc == -ERESTART || try >= max_tries)
		goto out;

	now = jiffies;
	if (time_before(now, deadline)) {
		unsigned long delta = deadline - now;

		ata_link_printk(link, KERN_WARNING, "reset failed "
				"(errno=%d), retrying in %u secs\n",
				rc, (jiffies_to_msecs(delta) + 999) / 1000);

		while (delta)
			delta = schedule_timeout_uninterruptible(delta);
	}

	if (rc == -EPIPE || try == max_tries - 1)
		sata_down_spd_limit(link);
	if (hardreset)
		reset = hardreset;
	goto retry;
T
Tejun Heo 已提交
2299 2300
}

2301
static int ata_eh_revalidate_and_attach(struct ata_link *link,
T
Tejun Heo 已提交
2302
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
2303
{
2304 2305
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2306
	struct ata_device *dev;
2307
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
2308
	unsigned long flags;
2309
	int rc = 0;
T
Tejun Heo 已提交
2310 2311 2312

	DPRINTK("ENTER\n");

2313 2314 2315 2316
	/* 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.
	 */
2317
	ata_link_for_each_dev_reverse(dev, link) {
2318 2319
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
2320

2321 2322 2323
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

2324
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
2325 2326
			WARN_ON(dev->class == ATA_DEV_PMP);

2327
			if (ata_link_offline(link)) {
T
Tejun Heo 已提交
2328
				rc = -EIO;
2329
				goto err;
T
Tejun Heo 已提交
2330 2331
			}

2332
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
2333 2334
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
T
Tejun Heo 已提交
2335
			if (rc)
2336
				goto err;
T
Tejun Heo 已提交
2337

2338
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
2339

2340 2341 2342 2343 2344
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

2345 2346
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
2347 2348 2349 2350 2351
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

2352 2353 2354 2355 2356
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
2357 2358
			switch (rc) {
			case 0:
2359
				new_mask |= 1 << dev->devno;
2360 2361
				break;
			case -ENOENT:
2362 2363 2364 2365 2366
				/* 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
2367 2368
				dev->class = ATA_DEV_UNKNOWN;
				break;
2369 2370 2371
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
2372
			}
2373 2374
		}
	}
T
Tejun Heo 已提交
2375

2376
	/* PDIAG- should have been released, ask cable type if post-reset */
2377 2378 2379 2380 2381
	if ((ehc->i.flags & ATA_EHI_DID_RESET) && ata_is_host_link(link)) {
		if (ap->ops->cable_detect)
			ap->cbl = ap->ops->cable_detect(ap);
		ata_force_cbl(ap);
	}
2382

2383 2384 2385
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
2386
	ata_link_for_each_dev(dev, link) {
2387 2388
		if (!(new_mask & (1 << dev->devno)) ||
		    dev->class == ATA_DEV_PMP)
2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402
			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
2403 2404
	}

2405
	return 0;
T
Tejun Heo 已提交
2406

2407 2408 2409
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
2410 2411 2412
	return rc;
}

2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
/**
 *	ata_set_mode - Program timings and issue SET FEATURES - XFER
 *	@link: link on which timings will be programmed
 *	@r_failed_dev: out paramter for failed device
 *
 *	Set ATA device disk transfer mode (PIO3, UDMA6, etc.).  If
 *	ata_set_mode() fails, pointer to the failing device is
 *	returned in @r_failed_dev.
 *
 *	LOCKING:
 *	PCI/etc. bus probe sem.
 *
 *	RETURNS:
 *	0 on success, negative errno otherwise
 */
int ata_set_mode(struct ata_link *link, struct ata_device **r_failed_dev)
{
	struct ata_port *ap = link->ap;
2431 2432
	struct ata_device *dev;
	int rc;
2433

2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
	/* if data transfer is verified, clear DUBIOUS_XFER on ering top */
	ata_link_for_each_dev(dev, link) {
		if (!(dev->flags & ATA_DFLAG_DUBIOUS_XFER)) {
			struct ata_ering_entry *ent;

			ent = ata_ering_top(&dev->ering);
			if (ent)
				ent->eflags &= ~ATA_EFLAG_DUBIOUS_XFER;
		}
	}

2445 2446
	/* has private set_mode? */
	if (ap->ops->set_mode)
2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462
		rc = ap->ops->set_mode(link, r_failed_dev);
	else
		rc = ata_do_set_mode(link, r_failed_dev);

	/* if transfer mode has changed, set DUBIOUS_XFER on device */
	ata_link_for_each_dev(dev, link) {
		struct ata_eh_context *ehc = &link->eh_context;
		u8 saved_xfer_mode = ehc->saved_xfer_mode[dev->devno];
		u8 saved_ncq = !!(ehc->saved_ncq_enabled & (1 << dev->devno));

		if (dev->xfer_mode != saved_xfer_mode ||
		    ata_ncq_enabled(dev) != saved_ncq)
			dev->flags |= ATA_DFLAG_DUBIOUS_XFER;
	}

	return rc;
2463 2464
}

2465
static int ata_link_nr_enabled(struct ata_link *link)
T
Tejun Heo 已提交
2466
{
2467 2468
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2469

2470
	ata_link_for_each_dev(dev, link)
2471
		if (ata_dev_enabled(dev))
T
Tejun Heo 已提交
2472 2473 2474 2475
			cnt++;
	return cnt;
}

2476
static int ata_link_nr_vacant(struct ata_link *link)
T
Tejun Heo 已提交
2477
{
2478 2479
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2480

2481
	ata_link_for_each_dev(dev, link)
2482
		if (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
2483 2484 2485 2486
			cnt++;
	return cnt;
}

2487
static int ata_eh_skip_recovery(struct ata_link *link)
T
Tejun Heo 已提交
2488
{
T
Tejun Heo 已提交
2489
	struct ata_port *ap = link->ap;
2490
	struct ata_eh_context *ehc = &link->eh_context;
2491
	struct ata_device *dev;
T
Tejun Heo 已提交
2492

2493 2494 2495 2496
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

T
Tejun Heo 已提交
2497 2498 2499 2500 2501 2502 2503
	/* thaw frozen port and recover failed devices */
	if ((ap->pflags & ATA_PFLAG_FROZEN) || ata_link_nr_enabled(link))
		return 0;

	/* reset at least once if reset is requested */
	if ((ehc->i.action & ATA_EH_RESET) &&
	    !(ehc->i.flags & ATA_EHI_DID_RESET))
T
Tejun Heo 已提交
2504 2505 2506
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
2507
	ata_link_for_each_dev(dev, link) {
T
Tejun Heo 已提交
2508 2509 2510 2511 2512 2513 2514 2515
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526
static int ata_eh_schedule_probe(struct ata_device *dev)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;

	if (!(ehc->i.probe_mask & (1 << dev->devno)) ||
	    (ehc->did_probe_mask & (1 << dev->devno)))
		return 0;

	ata_eh_detach_dev(dev);
	ata_dev_init(dev);
	ehc->did_probe_mask |= (1 << dev->devno);
T
Tejun Heo 已提交
2527
	ehc->i.action |= ATA_EH_RESET;
2528 2529
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
2530 2531 2532 2533

	return 1;
}

2534
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
2535
{
T
Tejun Heo 已提交
2536
	struct ata_eh_context *ehc = &dev->link->eh_context;
2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547

	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:
2548
		if (ehc->tries[dev->devno] == 1 && dev->pio_mode > XFER_PIO_0) {
2549 2550 2551
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
2552
			sata_down_spd_limit(dev->link);
2553 2554 2555 2556 2557 2558 2559 2560 2561
			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 */
2562
		if (ata_link_offline(dev->link))
2563 2564
			ata_eh_detach_dev(dev);

2565 2566
		/* schedule probe if necessary */
		if (ata_eh_schedule_probe(dev))
2567
			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
2568 2569

		return 1;
2570
	} else {
T
Tejun Heo 已提交
2571
		ehc->i.action |= ATA_EH_RESET;
2572
		return 0;
2573 2574 2575
	}
}

T
Tejun Heo 已提交
2576 2577 2578
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
2579
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2580 2581 2582
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
2583
 *	@r_failed_link: out parameter for failed link
T
Tejun Heo 已提交
2584 2585 2586
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
2587 2588 2589
 *	recover each link and hotplug requests are recorded in the
 *	link's eh_context.  This function executes all the operations
 *	with appropriate retrials and fallbacks to resurrect failed
T
Tejun Heo 已提交
2590
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
2591 2592 2593 2594 2595 2596 2597
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
2598 2599 2600 2601
int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset,
		   ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
		   ata_postreset_fn_t postreset,
		   struct ata_link **r_failed_link)
T
Tejun Heo 已提交
2602
{
2603
	struct ata_link *link;
T
Tejun Heo 已提交
2604
	struct ata_device *dev;
2605 2606
	int nr_failed_devs, nr_disabled_devs;
	int reset, rc;
2607
	unsigned long flags;
T
Tejun Heo 已提交
2608 2609 2610 2611

	DPRINTK("ENTER\n");

	/* prep for recovery */
2612 2613
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2614

2615 2616 2617 2618 2619 2620 2621 2622 2623
		/* re-enable link? */
		if (ehc->i.action & ATA_EH_ENABLE_LINK) {
			ata_eh_about_to_do(link, NULL, ATA_EH_ENABLE_LINK);
			spin_lock_irqsave(ap->lock, flags);
			link->flags &= ~ATA_LFLAG_DISABLED;
			spin_unlock_irqrestore(ap->lock, flags);
			ata_eh_done(link, NULL, ATA_EH_ENABLE_LINK);
		}

2624
		ata_link_for_each_dev(dev, link) {
2625 2626 2627 2628
			if (link->flags & ATA_LFLAG_NO_RETRY)
				ehc->tries[dev->devno] = 1;
			else
				ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
T
Tejun Heo 已提交
2629

2630 2631 2632 2633 2634 2635 2636 2637 2638
			/* collect port action mask recorded in dev actions */
			ehc->i.action |= ehc->i.dev_action[dev->devno] &
					 ~ATA_EH_PERDEV_MASK;
			ehc->i.dev_action[dev->devno] &= ATA_EH_PERDEV_MASK;

			/* process hotplug request */
			if (dev->flags & ATA_DFLAG_DETACH)
				ata_eh_detach_dev(dev);

2639 2640 2641
			/* schedule probe if necessary */
			if (!ata_dev_enabled(dev))
				ata_eh_schedule_probe(dev);
T
Tejun Heo 已提交
2642
		}
T
Tejun Heo 已提交
2643 2644 2645 2646
	}

 retry:
	rc = 0;
2647 2648 2649
	nr_failed_devs = 0;
	nr_disabled_devs = 0;
	reset = 0;
T
Tejun Heo 已提交
2650

2651
	/* if UNLOADING, finish immediately */
2652
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2653 2654
		goto out;

2655 2656 2657
	/* prep for EH */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2658

2659 2660 2661 2662 2663
		/* skip EH if possible. */
		if (ata_eh_skip_recovery(link))
			ehc->i.action = 0;

		/* do we need to reset? */
T
Tejun Heo 已提交
2664
		if (ehc->i.action & ATA_EH_RESET)
2665 2666 2667 2668 2669
			reset = 1;

		ata_link_for_each_dev(dev, link)
			ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
	}
T
Tejun Heo 已提交
2670

T
Tejun Heo 已提交
2671
	/* reset */
2672
	if (reset) {
2673 2674 2675 2676 2677
		/* if PMP is attached, this function only deals with
		 * downstream links, port should stay thawed.
		 */
		if (!ap->nr_pmp_links)
			ata_eh_freeze_port(ap);
T
Tejun Heo 已提交
2678

2679 2680 2681
		ata_port_for_each_link(link, ap) {
			struct ata_eh_context *ehc = &link->eh_context;

T
Tejun Heo 已提交
2682
			if (!(ehc->i.action & ATA_EH_RESET))
2683 2684 2685 2686 2687 2688 2689 2690 2691 2692
				continue;

			rc = ata_eh_reset(link, ata_link_nr_vacant(link),
					  prereset, softreset, hardreset,
					  postreset);
			if (rc) {
				ata_link_printk(link, KERN_ERR,
						"reset failed, giving up\n");
				goto out;
			}
T
Tejun Heo 已提交
2693 2694
		}

2695 2696
		if (!ap->nr_pmp_links)
			ata_eh_thaw_port(ap);
T
Tejun Heo 已提交
2697 2698
	}

2699 2700 2701
	/* the rest */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2702

2703 2704
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
2705
		if (rc)
T
Tejun Heo 已提交
2706 2707
			goto dev_fail;

2708 2709 2710 2711 2712 2713
		/* if PMP got attached, return, pmp EH will take care of it */
		if (link->device->class == ATA_DEV_PMP) {
			ehc->i.action = 0;
			return 0;
		}

2714 2715 2716 2717 2718 2719 2720 2721
		/* configure transfer mode if necessary */
		if (ehc->i.flags & ATA_EHI_SETMODE) {
			rc = ata_set_mode(link, &dev);
			if (rc)
				goto dev_fail;
			ehc->i.flags &= ~ATA_EHI_SETMODE;
		}

2722
		if (ehc->i.action & ATA_EH_LPM)
2723 2724 2725
			ata_link_for_each_dev(dev, link)
				ata_dev_enable_pm(dev, ap->pm_policy);

2726 2727 2728
		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
T
Tejun Heo 已提交
2729

2730
dev_fail:
2731 2732 2733
		nr_failed_devs++;
		if (ata_eh_handle_dev_fail(dev, rc))
			nr_disabled_devs++;
T
Tejun Heo 已提交
2734

2735 2736 2737 2738 2739 2740
		if (ap->pflags & ATA_PFLAG_FROZEN) {
			/* PMP reset requires working host port.
			 * Can't retry if it's frozen.
			 */
			if (ap->nr_pmp_links)
				goto out;
2741
			break;
2742
		}
T
Tejun Heo 已提交
2743 2744
	}

2745 2746 2747 2748 2749 2750 2751 2752 2753
	if (nr_failed_devs) {
		if (nr_failed_devs != nr_disabled_devs) {
			ata_port_printk(ap, KERN_WARNING, "failed to recover "
					"some devices, retrying in 5 secs\n");
			ssleep(5);
		} else {
			/* no device left to recover, repeat fast */
			msleep(500);
		}
T
Tejun Heo 已提交
2754

2755
		goto retry;
T
Tejun Heo 已提交
2756 2757
	}

2758 2759 2760 2761
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

T
Tejun Heo 已提交
2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775
	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.
 */
2776
void ata_eh_finish(struct ata_port *ap)
T
Tejun Heo 已提交
2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791
{
	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.
			 */
T
Tejun Heo 已提交
2792
			if (qc->flags & ATA_QCFLAG_RETRY)
T
Tejun Heo 已提交
2793
				ata_eh_qc_retry(qc);
T
Tejun Heo 已提交
2794 2795
			else
				ata_eh_qc_complete(qc);
T
Tejun Heo 已提交
2796 2797 2798 2799 2800 2801 2802 2803 2804 2805
		} 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);
			}
		}
	}
2806 2807 2808 2809

	/* make sure nr_active_links is zero after EH */
	WARN_ON(ap->nr_active_links);
	ap->nr_active_links = 0;
T
Tejun Heo 已提交
2810 2811 2812 2813 2814
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2815
 *
2816
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2817 2818 2819 2820 2821 2822 2823 2824 2825
 *	@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).
 */
2826 2827 2828
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
2829
{
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842
	struct ata_device *dev;
	int rc;

	ata_eh_autopsy(ap);
	ata_eh_report(ap);

	rc = ata_eh_recover(ap, prereset, softreset, hardreset, postreset,
			    NULL);
	if (rc) {
		ata_link_for_each_dev(dev, &ap->link)
			ata_dev_disable(dev);
	}

T
Tejun Heo 已提交
2843 2844
	ata_eh_finish(ap);
}
2845

2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859
/**
 *	ata_std_error_handler - standard error handler
 *	@ap: host port to handle error for
 *
 *	Standard error handler
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
void ata_std_error_handler(struct ata_port *ap)
{
	struct ata_port_operations *ops = ap->ops;
	ata_reset_fn_t hardreset = ops->hardreset;

2860 2861
	/* ignore built-in hardreset if SCR access is not available */
	if (ata_is_builtin_hardreset(hardreset) && !sata_scr_valid(&ap->link))
2862 2863 2864 2865 2866
		hardreset = NULL;

	ata_do_eh(ap, ops->prereset, ops->softreset, hardreset, ops->postreset);
}

2867
#ifdef CONFIG_PM
2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892
/**
 *	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);

2893 2894 2895 2896 2897
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

2898 2899 2900 2901 2902 2903
	/* suspend */
	ata_eh_freeze_port(ap);

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

S
Shaohua Li 已提交
2904
	ata_acpi_set_state(ap, PMSG_SUSPEND);
2905
 out:
2906 2907 2908 2909 2910 2911
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
2912
	else if (ap->pflags & ATA_PFLAG_FROZEN)
2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936
		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;
2937
	int rc = 0;
2938 2939 2940 2941 2942 2943 2944 2945 2946 2947

	/* 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);

2948
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
2949

S
Shaohua Li 已提交
2950 2951
	ata_acpi_set_state(ap, PMSG_ON);

2952 2953 2954
	if (ap->ops->port_resume)
		rc = ap->ops->port_resume(ap);

T
Tejun Heo 已提交
2955 2956 2957
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

2958
	/* report result */
2959 2960 2961 2962 2963 2964 2965 2966
	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);
}
2967
#endif /* CONFIG_PM */