libata-eh.c 77.6 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 71
	ATA_EH_CMD_DFL_TIMEOUT		=  5000,

72 73 74
	/* always put at least this amount of time between resets */
	ATA_EH_RESET_COOL_DOWN		=  5000,

75 76 77 78 79 80 81
	/* 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.
	 */
	ATA_EH_PRERESET_TIMEOUT		= 10000,
	ATA_EH_FASTDRAIN_INTERVAL	=  3000,
T
Tejun Heo 已提交
82 83 84 85 86 87 88 89 90
};

/* 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[] = {
91 92 93 94
	10000,	/* most drives spin up by 10sec */
	10000,	/* > 99% working drives spin up before 20sec */
	35000,	/* give > 30 secs of idleness for retarded devices */
	 5000,	/* and sweet one last chance */
95
	ULONG_MAX, /* > 1 min has elapsed, give up */
T
Tejun Heo 已提交
96 97
};

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
static const unsigned long ata_eh_identify_timeouts[] = {
	 5000,	/* covers > 99% of successes and not too boring on failures */
	10000,  /* combined time till here is enough even for media access */
	30000,	/* for true idiots */
	ULONG_MAX,
};

static const unsigned long ata_eh_other_timeouts[] = {
	 5000,	/* same rationale as identify timeout */
	10000,	/* ditto */
	/* but no merciful 30sec for other commands, it just isn't worth it */
	ULONG_MAX,
};

struct ata_eh_cmd_timeout_ent {
	const u8		*commands;
	const unsigned long	*timeouts;
};

/* The following table determines timeouts to use for EH internal
 * commands.  Each table entry is a command class and matches the
 * commands the entry applies to and the timeout table to use.
 *
 * On the retry after a command timed out, the next timeout value from
 * the table is used.  If the table doesn't contain further entries,
 * the last value is used.
 *
 * ehc->cmd_timeout_idx keeps track of which timeout to use per
 * command class, so if SET_FEATURES times out on the first try, the
 * next try will use the second timeout value only for that class.
 */
#define CMDS(cmds...)	(const u8 []){ cmds, 0 }
static const struct ata_eh_cmd_timeout_ent
ata_eh_cmd_timeout_table[ATA_EH_CMD_TIMEOUT_TABLE_SIZE] = {
	{ .commands = CMDS(ATA_CMD_ID_ATA, ATA_CMD_ID_ATAPI),
	  .timeouts = ata_eh_identify_timeouts, },
	{ .commands = CMDS(ATA_CMD_READ_NATIVE_MAX, ATA_CMD_READ_NATIVE_MAX_EXT),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_SET_MAX, ATA_CMD_SET_MAX_EXT),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_SET_FEATURES),
	  .timeouts = ata_eh_other_timeouts, },
	{ .commands = CMDS(ATA_CMD_INIT_DEV_PARAMS),
	  .timeouts = ata_eh_other_timeouts, },
};
#undef CMDS

145
static void __ata_port_freeze(struct ata_port *ap);
146
#ifdef CONFIG_PM
147 148
static void ata_eh_handle_port_suspend(struct ata_port *ap);
static void ata_eh_handle_port_resume(struct ata_port *ap);
149 150 151 152 153 154 155
#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 */
156

T
Tejun Heo 已提交
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
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;
}

222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
/**
 *	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
284 285
		ata_port_desc(ap, "%s 0x%llx", name,
				start + (unsigned long long)offset);
286 287 288 289
}

#endif /* CONFIG_PCI */

290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
static int ata_lookup_timeout_table(u8 cmd)
{
	int i;

	for (i = 0; i < ATA_EH_CMD_TIMEOUT_TABLE_SIZE; i++) {
		const u8 *cur;

		for (cur = ata_eh_cmd_timeout_table[i].commands; *cur; cur++)
			if (*cur == cmd)
				return i;
	}

	return -1;
}

/**
 *	ata_internal_cmd_timeout - determine timeout for an internal command
 *	@dev: target device
 *	@cmd: internal command to be issued
 *
 *	Determine timeout for internal command @cmd for @dev.
 *
 *	LOCKING:
 *	EH context.
 *
 *	RETURNS:
 *	Determined timeout.
 */
unsigned long ata_internal_cmd_timeout(struct ata_device *dev, u8 cmd)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	int ent = ata_lookup_timeout_table(cmd);
	int idx;

	if (ent < 0)
		return ATA_EH_CMD_DFL_TIMEOUT;

	idx = ehc->cmd_timeout_idx[dev->devno][ent];
	return ata_eh_cmd_timeout_table[ent].timeouts[idx];
}

/**
 *	ata_internal_cmd_timed_out - notification for internal command timeout
 *	@dev: target device
 *	@cmd: internal command which timed out
 *
 *	Notify EH that internal command @cmd for @dev timed out.  This
 *	function should be called only for commands whose timeouts are
 *	determined using ata_internal_cmd_timeout().
 *
 *	LOCKING:
 *	EH context.
 */
void ata_internal_cmd_timed_out(struct ata_device *dev, u8 cmd)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	int ent = ata_lookup_timeout_table(cmd);
	int idx;

	if (ent < 0)
		return;

	idx = ehc->cmd_timeout_idx[dev->devno][ent];
	if (ata_eh_cmd_timeout_table[ent].timeouts[idx + 1] != ULONG_MAX)
		ehc->cmd_timeout_idx[dev->devno][ent]++;
}

357
static void ata_ering_record(struct ata_ering *ering, unsigned int eflags,
358 359 360 361 362 363 364 365 366 367
			     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];
368
	ent->eflags = eflags;
369 370 371 372
	ent->err_mask = err_mask;
	ent->timestamp = get_jiffies_64();
}

373 374 375 376 377 378 379 380 381
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;
}

382
static void ata_ering_clear(struct ata_ering *ering)
383
{
384
	memset(ering, 0, sizeof(*ering));
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
}

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;
}

408 409
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
410
	struct ata_eh_context *ehc = &dev->link->eh_context;
411 412 413 414

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

415
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
416 417
				struct ata_eh_info *ehi, unsigned int action)
{
418
	struct ata_device *tdev;
419 420 421

	if (!dev) {
		ehi->action &= ~action;
422 423
		ata_link_for_each_dev(tdev, link)
			ehi->dev_action[tdev->devno] &= ~action;
424 425 426 427 428 429
	} 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) {
430 431 432
			ata_link_for_each_dev(tdev, link)
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
433 434 435 436 437 438 439 440
			ehi->action &= ~action;
		}

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

441 442 443 444 445 446 447 448 449 450 451
/**
 *	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.
 *
452 453
 *	TODO: kill this function once old EH is gone.
 *
454 455 456 457 458 459 460 461 462
 *	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;
463
	struct ata_port *ap = ata_shost_to_port(host);
464 465
	unsigned long flags;
	struct ata_queued_cmd *qc;
466
	enum scsi_eh_timer_return ret;
467 468 469

	DPRINTK("ENTER\n");

470 471 472 473 474 475
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
476
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
477
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
478 479 480 481 482 483
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
484
	spin_unlock_irqrestore(ap->lock, flags);
485

486
 out:
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502
	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.
 */
503
void ata_scsi_error(struct Scsi_Host *host)
504
{
505
	struct ata_port *ap = ata_shost_to_port(host);
506
	int i;
507
	unsigned long flags;
508 509 510

	DPRINTK("ENTER\n");

511
	/* synchronize with port task */
512 513
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
514
	/* synchronize with host lock and sort out timeouts */
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531

	/* 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;

532
		spin_lock_irqsave(ap->lock, flags);
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570

		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);

571
		spin_unlock_irqrestore(ap->lock, flags);
572 573 574

		/* initialize eh_tries */
		ap->eh_tries = ATA_EH_MAX_TRIES;
575
	} else
576
		spin_unlock_wait(ap->lock);
577 578 579 580

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
581 582
		struct ata_link *link;

T
Tejun Heo 已提交
583 584 585
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

586 587 588
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

589
		/* fetch & clear EH info */
590
		spin_lock_irqsave(ap->lock, flags);
591

592
		__ata_port_for_each_link(link, ap) {
593 594 595
			struct ata_eh_context *ehc = &link->eh_context;
			struct ata_device *dev;

596 597 598
			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));
599 600 601 602 603 604 605 606

			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;
			}
607 608 609

			/* set last reset timestamp to some time in the past */
			ehc->last_reset = jiffies - 60 * HZ;
610
		}
611

612 613
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
614
		ap->excl_link = NULL;	/* don't maintain exclusion over EH */
615

616
		spin_unlock_irqrestore(ap->lock, flags);
617

618 619
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
620 621 622
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
623

624 625 626
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

627 628 629 630
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
631
		spin_lock_irqsave(ap->lock, flags);
632

633
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
634
			if (--ap->eh_tries) {
635
				spin_unlock_irqrestore(ap->lock, flags);
636 637 638
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
639
					"tries, giving up\n", ATA_EH_MAX_TRIES);
T
Tejun Heo 已提交
640
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
641 642
		}

643
		/* this run is complete, make sure EH info is clear */
644 645
		__ata_port_for_each_link(link, ap)
			memset(&link->eh_info, 0, sizeof(link->eh_info));
646

647
		/* Clear host_eh_scheduled while holding ap->lock such
648 649 650 651 652 653
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

654
		spin_unlock_irqrestore(ap->lock, flags);
655
	} else {
T
Tejun Heo 已提交
656
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
657 658
		ap->ops->eng_timeout(ap);
	}
659

660
	/* finish or retry handled scmd's and clean up */
661 662 663 664
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

665
	/* clean up */
666
	spin_lock_irqsave(ap->lock, flags);
667

668
	if (ap->pflags & ATA_PFLAG_LOADING)
669
		ap->pflags &= ~ATA_PFLAG_LOADING;
670
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
671
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
672 673 674

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
675

676
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
677

678
	/* tell wait_eh that we're done */
679
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
680 681
	wake_up_all(&ap->eh_wait_q);

682
	spin_unlock_irqrestore(ap->lock, flags);
683

684 685 686
	DPRINTK("EXIT\n");
}

687 688 689 690 691 692 693 694 695 696 697 698 699 700 701
/**
 *	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:
702
	spin_lock_irqsave(ap->lock, flags);
703

704
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
705
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
706
		spin_unlock_irqrestore(ap->lock, flags);
707
		schedule();
708
		spin_lock_irqsave(ap->lock, flags);
709
	}
710
	finish_wait(&ap->eh_wait_q, &wait);
711

712
	spin_unlock_irqrestore(ap->lock, flags);
713 714

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
715
	if (scsi_host_in_recovery(ap->scsi_host)) {
716 717 718 719 720
		msleep(10);
		goto retry;
	}
}

T
Tejun Heo 已提交
721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
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 =
765
			ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
T
Tejun Heo 已提交
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
		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;
805 806
	ap->fastdrain_timer.expires =
		ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
T
Tejun Heo 已提交
807 808 809
	add_timer(&ap->fastdrain_timer);
}

810 811 812 813 814 815 816 817
/**
 *	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
818
 *	spin_lock_irqsave(host lock)
819 820 821 822 823 824 825 826
 */
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
827
	ata_eh_set_pending(ap, 1);
828 829 830 831 832 833 834 835 836

	/* 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);
}

837 838 839 840 841 842 843 844
/**
 *	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
845
 *	spin_lock_irqsave(host lock)
846 847 848 849 850
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

851 852 853
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
854
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
855
	scsi_schedule_eh(ap->scsi_host);
856 857 858 859

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

860
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
861 862 863 864 865
{
	int tag, nr_aborted = 0;

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

T
Tejun Heo 已提交
866 867 868
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

869 870 871
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

872
		if (qc && (!link || qc->dev->link == link)) {
873 874 875 876 877 878 879 880 881 882 883 884
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

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
/**
 *	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);
}

919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
/**
 *	__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
935
 *	spin_lock_irqsave(host lock)
936 937 938 939 940 941 942 943
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

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

944
	ap->pflags |= ATA_PFLAG_FROZEN;
945

T
Tejun Heo 已提交
946
	DPRINTK("ata%u port frozen\n", ap->print_id);
947 948 949 950 951 952 953 954 955
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
956
 *	spin_lock_irqsave(host lock)
957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972
 *
 *	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;
}

973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997
/**
 *	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);

T
Tejun Heo 已提交
998
	if (!sata_pmp_attached(ap) || rc) {
999
		/* PMP is not attached or SNTF is not available */
T
Tejun Heo 已提交
1000
		if (!sata_pmp_attached(ap)) {
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
			/* 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;
	}
}

1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
/**
 *	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;

1062
	spin_lock_irqsave(ap->lock, flags);
1063
	__ata_port_freeze(ap);
1064
	spin_unlock_irqrestore(ap->lock, flags);
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
}

/**
 *	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;

1083
	spin_lock_irqsave(ap->lock, flags);
1084

1085
	ap->pflags &= ~ATA_PFLAG_FROZEN;
1086 1087 1088 1089

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

1090
	spin_unlock_irqrestore(ap->lock, flags);
1091

T
Tejun Heo 已提交
1092
	DPRINTK("ata%u port thawed\n", ap->print_id);
1093 1094
}

1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105
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;

1106
	spin_lock_irqsave(ap->lock, flags);
1107 1108 1109
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
1110
	spin_unlock_irqrestore(ap->lock, flags);
1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146

	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
1147

1148 1149 1150 1151 1152 1153 1154 1155 1156
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
1157
void ata_eh_detach_dev(struct ata_device *dev)
1158
{
1159 1160
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1161 1162 1163 1164
	unsigned long flags;

	ata_dev_disable(dev);

1165
	spin_lock_irqsave(ap->lock, flags);
1166 1167 1168 1169 1170

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1171
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1172 1173
	}

1174
	/* clear per-dev EH actions */
1175 1176
	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);
1177

1178
	spin_unlock_irqrestore(ap->lock, flags);
1179 1180
}

T
Tejun Heo 已提交
1181 1182
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
1183
 *	@link: target ATA link
1184
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
1185 1186 1187
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
1188 1189
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
1190 1191 1192 1193
 *
 *	LOCKING:
 *	None.
 */
1194 1195
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
T
Tejun Heo 已提交
1196
{
T
Tejun Heo 已提交
1197 1198 1199
	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
1200 1201
	unsigned long flags;

1202
	spin_lock_irqsave(ap->lock, flags);
1203

T
Tejun Heo 已提交
1204
	ata_eh_clear_action(link, dev, ehi, action);
1205

1206
	if (!(ehc->i.flags & ATA_EHI_QUIET))
1207 1208
		ap->pflags |= ATA_PFLAG_RECOVERED;

1209
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
1210 1211
}

1212 1213
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
1214
*	@ap: target ATA port
1215 1216 1217 1218
 *	@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
1219
 *	in @link->eh_context.
1220 1221 1222 1223
 *
 *	LOCKING:
 *	None.
 */
1224 1225
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1226
{
T
Tejun Heo 已提交
1227
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1228

T
Tejun Heo 已提交
1229
	ata_eh_clear_action(link, dev, &ehc->i, action);
1230 1231
}

T
Tejun Heo 已提交
1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245
/**
 *	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
 */
1246
static const char *ata_err_string(unsigned int err_mask)
T
Tejun Heo 已提交
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
{
	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";
}

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298
/**
 *	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,
1299
				     buf, sectors * ATA_SECT_SIZE, 0);
1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322

	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
1323
	u8 *buf = dev->link->ap->sector_buf;
1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
	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
1359 1360 1361 1362
/**
 *	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)
1363
 *	@dfl_sense_key: default sense key to use
T
Tejun Heo 已提交
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
 *
 *	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
 */
1374 1375
static unsigned int atapi_eh_request_sense(struct ata_device *dev,
					   u8 *sense_buf, u8 dfl_sense_key)
T
Tejun Heo 已提交
1376
{
1377 1378
	u8 cdb[ATAPI_CDB_LEN] =
		{ REQUEST_SENSE, 0, 0, 0, SCSI_SENSE_BUFFERSIZE, 0 };
T
Tejun Heo 已提交
1379
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1380 1381 1382 1383 1384 1385 1386
	struct ata_taskfile tf;

	DPRINTK("ATAPI request sense\n");

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

1387 1388 1389
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1390
	sense_buf[0] = 0x70;
1391
	sense_buf[2] = dfl_sense_key;
1392

J
Jeff Garzik 已提交
1393
	/* some devices time out if garbage left in tf */
1394
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1395 1396 1397 1398 1399 1400

	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) {
1401
		tf.protocol = ATAPI_PROT_DMA;
T
Tejun Heo 已提交
1402 1403
		tf.feature |= ATAPI_PKT_DMA;
	} else {
1404
		tf.protocol = ATAPI_PROT_PIO;
1405 1406
		tf.lbam = SCSI_SENSE_BUFFERSIZE;
		tf.lbah = 0;
T
Tejun Heo 已提交
1407 1408 1409
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
1410
				 sense_buf, SCSI_SENSE_BUFFERSIZE, 0);
T
Tejun Heo 已提交
1411 1412 1413 1414
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1415
 *	@link: ATA link to analyze SError for
T
Tejun Heo 已提交
1416 1417 1418 1419 1420 1421 1422
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1423
static void ata_eh_analyze_serror(struct ata_link *link)
T
Tejun Heo 已提交
1424
{
1425
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1426 1427
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1428
	u32 hotplug_mask;
T
Tejun Heo 已提交
1429

T
Tejun Heo 已提交
1430
	if (serror & (SERR_PERSISTENT | SERR_DATA)) {
T
Tejun Heo 已提交
1431
		err_mask |= AC_ERR_ATA_BUS;
T
Tejun Heo 已提交
1432
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1433 1434 1435
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
T
Tejun Heo 已提交
1436
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1437 1438 1439
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
T
Tejun Heo 已提交
1440
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1441
	}
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455

	/* 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
1456
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1457 1458 1459 1460 1461

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

1462 1463
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
1464
 *	@link: ATA link to analyze NCQ error for
1465 1466 1467 1468 1469 1470 1471 1472 1473
 *
 *	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).
 */
1474
void ata_eh_analyze_ncq_error(struct ata_link *link)
1475
{
1476 1477 1478
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
1479 1480 1481 1482 1483
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1484
	if (ap->pflags & ATA_PFLAG_FROZEN)
1485 1486 1487
		return;

	/* is it NCQ device error? */
1488
	if (!link->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504
		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) {
1505
		ata_link_printk(link, KERN_ERR, "failed to read log page 10h "
1506 1507 1508 1509
				"(errno=%d)\n", rc);
		return;
	}

1510 1511
	if (!(link->sactive & (1 << tag))) {
		ata_link_printk(link, KERN_ERR, "log page 10h reported "
1512 1513 1514 1515 1516 1517 1518
				"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));
1519
	qc->result_tf.flags = ATA_TFLAG_ISADDR | ATA_TFLAG_LBA | ATA_TFLAG_LBA48;
T
Tejun Heo 已提交
1520
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1521 1522 1523
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546
/**
 *	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
1547
		return ATA_EH_RESET;
T
Tejun Heo 已提交
1548 1549
	}

1550 1551 1552
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565
		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:
1566
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1567 1568 1569
			tmp = atapi_eh_request_sense(qc->dev,
						qc->scsicmd->sense_buffer,
						qc->result_tf.feature >> 4);
1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581
			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
1582 1583 1584
	}

	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
T
Tejun Heo 已提交
1585
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1586 1587 1588 1589

	return action;
}

1590 1591
static int ata_eh_categorize_error(unsigned int eflags, unsigned int err_mask,
				   int *xfer_ok)
T
Tejun Heo 已提交
1592
{
1593 1594 1595 1596 1597 1598
	int base = 0;

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

	if (!*xfer_ok)
1599
		base = ATA_ECAT_DUBIOUS_NONE;
1600

1601
	if (err_mask & AC_ERR_ATA_BUS)
1602
		return base + ATA_ECAT_ATA_BUS;
T
Tejun Heo 已提交
1603

1604
	if (err_mask & AC_ERR_TIMEOUT)
1605
		return base + ATA_ECAT_TOUT_HSM;
1606

1607
	if (eflags & ATA_EFLAG_IS_IO) {
1608
		if (err_mask & AC_ERR_HSM)
1609
			return base + ATA_ECAT_TOUT_HSM;
1610 1611
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
1612
			return base + ATA_ECAT_UNK_DEV;
T
Tejun Heo 已提交
1613 1614 1615 1616 1617
	}

	return 0;
}

1618
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1619
	u64 since;
1620
	int xfer_ok;
1621
	int nr_errors[ATA_ECAT_NR];
T
Tejun Heo 已提交
1622 1623
};

1624
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1625
{
1626
	struct speed_down_verdict_arg *arg = void_arg;
1627
	int cat;
T
Tejun Heo 已提交
1628 1629 1630 1631

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

1632 1633
	cat = ata_eh_categorize_error(ent->eflags, ent->err_mask,
				      &arg->xfer_ok);
1634
	arg->nr_errors[cat]++;
1635

T
Tejun Heo 已提交
1636 1637 1638 1639
	return 0;
}

/**
1640
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1641 1642 1643
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1644 1645
 *	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
1646
 *
1647 1648 1649 1650 1651 1652 1653
 *	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
 *
1654 1655 1656
 *	ECAT_DUBIOUS_*	: Identical to above three but occurred while
 *			  data transfer hasn't been verified.
 *
1657 1658 1659
 *	Verdicts are
 *
 *	NCQ_OFF		: Turn off NCQ.
T
Tejun Heo 已提交
1660
 *
1661 1662
 *	SPEED_DOWN	: Speed down transfer speed but don't fall back
 *			  to PIO.
1663
 *
1664
 *	FALLBACK_TO_PIO	: Fall back to PIO.
T
Tejun Heo 已提交
1665
 *
1666
 *	Even if multiple verdicts are returned, only one action is
1667 1668 1669 1670 1671 1672 1673
 *	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.
1674
 *
1675 1676 1677 1678 1679 1680 1681
 *	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
1682
 *	   ocurred during last 5 mins, FALLBACK_TO_PIO
1683
 *
1684
 *	4. If more than 3 TOUT_HSM or UNK_DEV errors occurred
1685 1686
 *	   during last 10 mins, NCQ_OFF.
 *
1687
 *	5. If more than 3 ATA_BUS or TOUT_HSM errors, or more than 6
1688
 *	   UNK_DEV errors occurred during last 10 mins, SPEED_DOWN.
1689
 *
T
Tejun Heo 已提交
1690 1691 1692 1693
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1694
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1695
 */
1696
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1697
{
1698 1699 1700 1701
	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
1702

1703
	/* scan past 5 mins of error history */
1704
	memset(&arg, 0, sizeof(arg));
1705
	arg.since = j64 - min(j64, j5mins);
1706
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1707

1708 1709 1710 1711 1712 1713 1714 1715 1716
	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;

1717 1718
	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] +
T
Tejun Heo 已提交
1719
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1720
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
T
Tejun Heo 已提交
1721

1722
	/* scan past 10 mins of error history */
T
Tejun Heo 已提交
1723
	memset(&arg, 0, sizeof(arg));
1724
	arg.since = j64 - min(j64, j10mins);
1725
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1726

1727 1728 1729 1730 1731 1732
	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
1733
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1734
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
T
Tejun Heo 已提交
1735

1736
	return verdict;
T
Tejun Heo 已提交
1737 1738 1739 1740 1741
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
1742
 *	@eflags: mask of ATA_EFLAG_* flags
T
Tejun Heo 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
 *	@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:
1754
 *	Determined recovery action.
T
Tejun Heo 已提交
1755
 */
1756 1757
static unsigned int ata_eh_speed_down(struct ata_device *dev,
				unsigned int eflags, unsigned int err_mask)
T
Tejun Heo 已提交
1758
{
1759
	struct ata_link *link = dev->link;
1760
	int xfer_ok = 0;
1761 1762 1763 1764
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
1765
	if (ata_eh_categorize_error(eflags, err_mask, &xfer_ok) == 0)
T
Tejun Heo 已提交
1766 1767 1768
		return 0;

	/* record error and determine whether speed down is necessary */
1769
	ata_ering_record(&dev->ering, eflags, err_mask);
1770
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1771

1772 1773 1774 1775 1776 1777 1778 1779 1780
	/* 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
1781

1782 1783 1784
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1785
		if (sata_down_spd_limit(link) == 0) {
T
Tejun Heo 已提交
1786
			action |= ATA_EH_RESET;
1787 1788
			goto done;
		}
T
Tejun Heo 已提交
1789

1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
		/* 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
1806
				action |= ATA_EH_RESET;
1807 1808 1809 1810 1811 1812
				goto done;
			}
		}
	}

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
T
Tejun Heo 已提交
1813
	 * SATA ATA devices.  Consider it only for PATA and SATAPI.
1814 1815
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
T
Tejun Heo 已提交
1816
	    (link->ap->cbl != ATA_CBL_SATA || dev->class == ATA_DEV_ATAPI) &&
1817 1818 1819
	    (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
1820
			action |= ATA_EH_RESET;
1821 1822 1823
			goto done;
		}
	}
T
Tejun Heo 已提交
1824 1825

	return 0;
1826 1827
 done:
	/* device has been slowed down, blow error history */
1828 1829
	if (!(verdict & ATA_EH_SPDN_KEEP_ERRORS))
		ata_ering_clear(&dev->ering);
1830
	return action;
T
Tejun Heo 已提交
1831 1832 1833
}

/**
1834 1835
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
T
Tejun Heo 已提交
1836
 *
1837 1838 1839
 *	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
1840 1841 1842 1843
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1844
static void ata_eh_link_autopsy(struct ata_link *link)
T
Tejun Heo 已提交
1845
{
1846
	struct ata_port *ap = link->ap;
1847
	struct ata_eh_context *ehc = &link->eh_context;
1848
	struct ata_device *dev;
1849 1850
	unsigned int all_err_mask = 0, eflags = 0;
	int tag;
T
Tejun Heo 已提交
1851 1852 1853 1854 1855
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1856 1857 1858
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1859
	/* obtain and analyze SError */
1860
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
1861 1862
	if (rc == 0) {
		ehc->i.serror |= serror;
1863
		ata_eh_analyze_serror(link);
1864
	} else if (rc != -EOPNOTSUPP) {
T
Tejun Heo 已提交
1865
		/* SError read failed, force reset and probing */
1866
		ehc->i.probe_mask |= ATA_ALL_DEVICES;
T
Tejun Heo 已提交
1867
		ehc->i.action |= ATA_EH_RESET;
1868 1869
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
1870

1871
	/* analyze NCQ failure */
1872
	ata_eh_analyze_ncq_error(link);
1873

T
Tejun Heo 已提交
1874 1875 1876 1877 1878 1879 1880 1881 1882
	/* 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);

1883
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link)
T
Tejun Heo 已提交
1884 1885 1886 1887 1888 1889
			continue;

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

		/* analyze TF */
1890
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901

		/* 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 */
1902
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
T
Tejun Heo 已提交
1903 1904
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);

T
Tejun Heo 已提交
1905 1906 1907 1908 1909
		/* 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
1910
		/* accumulate error info */
1911
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1912 1913
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
1914
			eflags |= ATA_EFLAG_IS_IO;
T
Tejun Heo 已提交
1915 1916
	}

1917
	/* enforce default EH actions */
1918
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1919
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
T
Tejun Heo 已提交
1920
		ehc->i.action |= ATA_EH_RESET;
1921 1922
	else if (((eflags & ATA_EFLAG_IS_IO) && all_err_mask) ||
		 (!(eflags & ATA_EFLAG_IS_IO) && (all_err_mask & ~AC_ERR_DEV)))
1923
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1924

1925 1926 1927
	/* If we have offending qcs and the associated failed device,
	 * perform per-dev EH action only on the offending device.
	 */
1928 1929 1930 1931
	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;
1932 1933
	}

1934 1935 1936 1937 1938
	/* 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 */
1939
	dev = ehc->i.dev;
1940 1941 1942
	if (!dev && ((ata_link_max_devices(link) == 1 &&
		      ata_dev_enabled(link->device))))
	    dev = link->device;
1943

1944 1945 1946
	if (dev) {
		if (dev->flags & ATA_DFLAG_DUBIOUS_XFER)
			eflags |= ATA_EFLAG_DUBIOUS_XFER;
1947
		ehc->i.action |= ata_eh_speed_down(dev, eflags, all_err_mask);
1948
	}
1949

T
Tejun Heo 已提交
1950 1951 1952 1953
	DPRINTK("EXIT\n");
}

/**
1954 1955 1956 1957 1958 1959 1960 1961 1962
 *	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).
 */
1963
void ata_eh_autopsy(struct ata_port *ap)
1964 1965 1966
{
	struct ata_link *link;

1967
	ata_port_for_each_link(link, ap)
1968
		ata_eh_link_autopsy(link);
1969 1970 1971 1972

	/* Autopsy of fanout ports can affect host link autopsy.
	 * Perform host link autopsy last.
	 */
T
Tejun Heo 已提交
1973
	if (sata_pmp_attached(ap))
1974
		ata_eh_link_autopsy(&ap->link);
1975 1976 1977 1978
}

/**
 *	ata_eh_link_report - report error handling to user
1979
 *	@link: ATA link EH is going on
T
Tejun Heo 已提交
1980 1981 1982 1983 1984 1985
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
1986
static void ata_eh_link_report(struct ata_link *link)
T
Tejun Heo 已提交
1987
{
1988 1989
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1990
	const char *frozen, *desc;
1991
	char tries_buf[6];
T
Tejun Heo 已提交
1992 1993
	int tag, nr_failed = 0;

1994 1995 1996
	if (ehc->i.flags & ATA_EHI_QUIET)
		return;

T
Tejun Heo 已提交
1997 1998 1999 2000 2001 2002 2003
	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);

2004 2005 2006
		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
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

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

	frozen = "";
2018
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
2019 2020
		frozen = " frozen";

2021 2022 2023 2024 2025
	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
2026
	if (ehc->i.dev) {
2027
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
2028 2029 2030
			       "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
2031
		if (desc)
T
Tejun Heo 已提交
2032
			ata_dev_printk(ehc->i.dev, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
2033
	} else {
2034
		ata_link_printk(link, KERN_ERR, "exception Emask 0x%x "
2035 2036 2037
				"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
2038
		if (desc)
2039
			ata_link_printk(link, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
2040 2041
	}

2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
	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 " : "",
2061
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "");
2062

T
Tejun Heo 已提交
2063 2064
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
2065
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
2066 2067 2068
		const u8 *cdb = qc->cdb;
		char data_buf[20] = "";
		char cdb_buf[70] = "";
T
Tejun Heo 已提交
2069

2070 2071
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    qc->dev->link != link || !qc->err_mask)
T
Tejun Heo 已提交
2072 2073
			continue;

2074 2075 2076 2077 2078 2079 2080 2081 2082 2083
		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",
2084 2085
				[ATAPI_PROT_PIO]	= "pio",
				[ATAPI_PROT_DMA]	= "dma",
2086 2087 2088 2089 2090 2091 2092
			};

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

2093
		if (ata_is_atapi(qc->tf.protocol))
2094 2095 2096 2097 2098 2099 2100 2101
			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]);

2102 2103
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
2104
			"tag %d%s\n         %s"
2105
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
2106
			"Emask 0x%x (%s)%s\n",
2107 2108 2109 2110
			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,
2111
			cmd->device, qc->tag, data_buf, cdb_buf,
2112 2113 2114 2115
			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
2116 2117
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
2118 2119

		if (res->command & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
2120
				    ATA_ERR)) {
2121 2122
			if (res->command & ATA_BUSY)
				ata_dev_printk(qc->dev, KERN_ERR,
2123
				  "status: { Busy }\n");
2124 2125 2126 2127 2128 2129
			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 " : "",
2130
				  res->command & ATA_ERR ? "ERR " : "");
2131 2132 2133 2134 2135 2136 2137 2138 2139 2140
		}

		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 " : "",
2141
			  res->feature & ATA_ABORTED ? "ABRT " : "");
T
Tejun Heo 已提交
2142 2143 2144
	}
}

2145 2146 2147 2148 2149 2150 2151 2152 2153
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2154
void ata_eh_report(struct ata_port *ap)
2155 2156 2157 2158 2159 2160 2161
{
	struct ata_link *link;

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

T
Tejun Heo 已提交
2162
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
2163
			unsigned int *classes, unsigned long deadline)
2164
{
2165
	struct ata_device *dev;
2166

T
Tejun Heo 已提交
2167
	ata_link_for_each_dev(dev, link)
2168
		classes[dev->devno] = ATA_DEV_UNKNOWN;
2169

2170
	return reset(link, classes, deadline);
2171 2172
}

2173 2174
static int ata_eh_followup_srst_needed(struct ata_link *link,
				       int rc, int classify,
2175 2176
				       const unsigned int *classes)
{
2177
	if ((link->flags & ATA_LFLAG_NO_SRST) || ata_link_offline(link))
2178
		return 0;
2179 2180 2181 2182 2183
	if (rc == -EAGAIN) {
		if (classify)
			return 1;
		rc = 0;
	}
2184 2185
	if (rc != 0)
		return 0;
T
Tejun Heo 已提交
2186
	if (sata_pmp_supported(link->ap) && ata_is_host_link(link))
2187
		return 1;
2188 2189 2190
	return 0;
}

2191 2192 2193
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
2194
{
T
Tejun Heo 已提交
2195
	struct ata_port *ap = link->ap;
2196
	struct ata_eh_context *ehc = &link->eh_context;
2197
	unsigned int *classes = ehc->classes;
2198
	unsigned int lflags = link->flags;
2199
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
2200
	int max_tries = 0, try = 0;
2201
	struct ata_device *dev;
2202
	unsigned long deadline, now;
T
Tejun Heo 已提交
2203
	ata_reset_fn_t reset;
T
Tejun Heo 已提交
2204
	unsigned long flags;
2205
	u32 sstatus;
2206
	int nr_known, rc;
T
Tejun Heo 已提交
2207

2208 2209 2210
	/*
	 * Prepare to reset
	 */
2211 2212
	while (ata_eh_reset_timeouts[max_tries] != ULONG_MAX)
		max_tries++;
2213 2214 2215 2216
	if (link->flags & ATA_LFLAG_NO_HRST)
		hardreset = NULL;
	if (link->flags & ATA_LFLAG_NO_SRST)
		softreset = NULL;
2217

2218 2219 2220 2221 2222
	now = jiffies;
	deadline = ata_deadline(ehc->last_reset, ATA_EH_RESET_COOL_DOWN);
	if (time_before(now, deadline))
		schedule_timeout_uninterruptible(deadline - now);

T
Tejun Heo 已提交
2223 2224 2225 2226
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags |= ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

T
Tejun Heo 已提交
2227
	ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
2228
	ehc->last_reset = jiffies;
2229

T
Tejun Heo 已提交
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248
	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);
	}

T
Tejun Heo 已提交
2249
	/* prefer hardreset */
2250
	reset = NULL;
T
Tejun Heo 已提交
2251 2252 2253
	ehc->i.action &= ~ATA_EH_RESET;
	if (hardreset) {
		reset = hardreset;
2254
		ehc->i.action |= ATA_EH_HARDRESET;
2255
	} else if (softreset) {
T
Tejun Heo 已提交
2256
		reset = softreset;
2257
		ehc->i.action |= ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
2258
	}
2259 2260

	if (prereset) {
2261 2262
		rc = prereset(link,
			      ata_deadline(jiffies, ATA_EH_PRERESET_TIMEOUT));
2263
		if (rc) {
2264
			if (rc == -ENOENT) {
T
Tejun Heo 已提交
2265
				ata_link_printk(link, KERN_DEBUG,
2266
						"port disabled. ignoring.\n");
T
Tejun Heo 已提交
2267
				ehc->i.action &= ~ATA_EH_RESET;
2268

2269
				ata_link_for_each_dev(dev, link)
2270
					classes[dev->devno] = ATA_DEV_NONE;
2271 2272

				rc = 0;
2273
			} else
T
Tejun Heo 已提交
2274
				ata_link_printk(link, KERN_ERR,
2275
					"prereset failed (errno=%d)\n", rc);
2276
			goto out;
2277 2278
		}

2279 2280 2281 2282 2283 2284 2285 2286 2287
		/* prereset() might have cleared ATA_EH_RESET.  If so,
		 * bang classes and return.
		 */
		if (reset && !(ehc->i.action & ATA_EH_RESET)) {
			ata_link_for_each_dev(dev, link)
				classes[dev->devno] = ATA_DEV_NONE;
			rc = 0;
			goto out;
		}
2288 2289
	}

T
Tejun Heo 已提交
2290
 retry:
2291 2292 2293
	/*
	 * Perform reset
	 */
2294
	ehc->last_reset = jiffies;
2295 2296 2297
	if (ata_is_host_link(link))
		ata_eh_freeze_port(ap);

2298
	deadline = ata_deadline(jiffies, ata_eh_reset_timeouts[try++]);
T
Tejun Heo 已提交
2299

2300 2301 2302 2303 2304 2305 2306 2307 2308 2309
	if (reset) {
		if (verbose)
			ata_link_printk(link, KERN_INFO, "%s resetting link\n",
					reset == softreset ? "soft" : "hard");

		/* mark that this EH session started with reset */
		if (reset == hardreset)
			ehc->i.flags |= ATA_EHI_DID_HARDRESET;
		else
			ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
2310

2311
		rc = ata_do_reset(link, reset, classes, deadline);
T
Tejun Heo 已提交
2312

2313 2314 2315 2316
		if (reset == hardreset &&
		    ata_eh_followup_srst_needed(link, rc, classify, classes)) {
			/* okay, let's do follow-up softreset */
			reset = softreset;
T
Tejun Heo 已提交
2317

2318 2319 2320 2321 2322 2323 2324
			if (!reset) {
				ata_link_printk(link, KERN_ERR,
						"follow-up softreset required "
						"but no softreset avaliable\n");
				rc = -EINVAL;
				goto fail;
			}
2325

2326 2327
			ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
			rc = ata_do_reset(link, reset, classes, deadline);
2328 2329
		}

2330 2331 2332 2333 2334 2335 2336 2337 2338
		/* -EAGAIN can happen if we skipped followup SRST */
		if (rc && rc != -EAGAIN)
			goto fail;
	} else {
		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;
2339 2340
	}

2341 2342 2343
	/*
	 * Post-reset processing
	 */
2344 2345 2346 2347 2348 2349 2350
	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
2351

2352 2353
		if (ata_link_offline(link))
			continue;
T
Tejun Heo 已提交
2354

2355
		/* apply class override */
2356 2357 2358 2359
		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
2360 2361
	}

2362 2363 2364
	/* record current link speed */
	if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
		link->sata_spd = (sstatus >> 4) & 0xf;
2365

2366 2367 2368 2369
	/* thaw the port */
	if (ata_is_host_link(link))
		ata_eh_thaw_port(ap);

2370 2371 2372 2373 2374 2375 2376
	/* postreset() should clear hardware SError.  Although SError
	 * is cleared during link resume, clearing SError here is
	 * necessary as some PHYs raise hotplug events after SRST.
	 * This introduces race condition where hotplug occurs between
	 * reset and here.  This race is mediated by cross checking
	 * link onlineness and classification result later.
	 */
2377 2378
	if (postreset)
		postreset(link, classes);
2379

2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412
	/* clear cached SError */
	spin_lock_irqsave(link->ap->lock, flags);
	link->eh_info.serror = 0;
	spin_unlock_irqrestore(link->ap->lock, flags);

	/* Make sure onlineness and classification result correspond.
	 * Hotplug could have happened during reset and some
	 * controllers fail to wait while a drive is spinning up after
	 * being hotplugged causing misdetection.  By cross checking
	 * link onlineness and classification result, those conditions
	 * can be reliably detected and retried.
	 */
	nr_known = 0;
	ata_link_for_each_dev(dev, link) {
		/* convert all ATA_DEV_UNKNOWN to ATA_DEV_NONE */
		if (classes[dev->devno] == ATA_DEV_UNKNOWN)
			classes[dev->devno] = ATA_DEV_NONE;
		else
			nr_known++;
	}

	if (classify && !nr_known && ata_link_online(link)) {
		if (try < max_tries) {
			ata_link_printk(link, KERN_WARNING, "link online but "
				       "device misclassified, retrying\n");
			rc = -EAGAIN;
			goto fail;
		}
		ata_link_printk(link, KERN_WARNING,
			       "link online but device misclassified, "
			       "device detection might fail\n");
	}

2413
	/* reset successful, schedule revalidation */
T
Tejun Heo 已提交
2414
	ata_eh_done(link, NULL, ATA_EH_RESET);
2415
	ehc->last_reset = jiffies;
2416
	ehc->i.action |= ATA_EH_REVALIDATE;
2417

2418
	rc = 0;
2419 2420 2421
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2422 2423 2424 2425 2426

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

T
Tejun Heo 已提交
2427
	return rc;
2428 2429

 fail:
2430 2431 2432 2433 2434
	/* if SCR isn't accessible on a fan-out port, PMP needs to be reset */
	if (!ata_is_host_link(link) &&
	    sata_scr_read(link, SCR_STATUS, &sstatus))
		rc = -ERESTART;

2435 2436 2437 2438 2439 2440 2441
	if (rc == -ERESTART || try >= max_tries)
		goto out;

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

2442 2443 2444
		ata_link_printk(link, KERN_WARNING,
			"reset failed (errno=%d), retrying in %u secs\n",
			rc, DIV_ROUND_UP(jiffies_to_msecs(delta), 1000));
2445 2446 2447 2448 2449 2450 2451 2452 2453 2454

		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
2455 2456
}

2457
static int ata_eh_revalidate_and_attach(struct ata_link *link,
T
Tejun Heo 已提交
2458
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
2459
{
2460 2461
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2462
	struct ata_device *dev;
2463
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
2464
	unsigned long flags;
2465
	int rc = 0;
T
Tejun Heo 已提交
2466 2467 2468

	DPRINTK("ENTER\n");

2469 2470 2471 2472
	/* 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.
	 */
2473
	ata_link_for_each_dev_reverse(dev, link) {
2474 2475
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
2476

2477 2478 2479
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

2480
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
2481 2482
			WARN_ON(dev->class == ATA_DEV_PMP);

2483
			if (ata_link_offline(link)) {
T
Tejun Heo 已提交
2484
				rc = -EIO;
2485
				goto err;
T
Tejun Heo 已提交
2486 2487
			}

2488
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
2489 2490
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
T
Tejun Heo 已提交
2491
			if (rc)
2492
				goto err;
T
Tejun Heo 已提交
2493

2494
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
2495

2496 2497 2498 2499 2500
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

2501 2502
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
2503 2504 2505 2506 2507
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

2508 2509 2510 2511 2512
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
2513 2514
			switch (rc) {
			case 0:
2515
				new_mask |= 1 << dev->devno;
2516 2517
				break;
			case -ENOENT:
2518 2519 2520 2521 2522
				/* 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
2523 2524
				dev->class = ATA_DEV_UNKNOWN;
				break;
2525 2526 2527
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
2528
			}
2529 2530
		}
	}
T
Tejun Heo 已提交
2531

2532
	/* PDIAG- should have been released, ask cable type if post-reset */
2533 2534 2535 2536 2537
	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);
	}
2538

2539 2540 2541
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
2542
	ata_link_for_each_dev(dev, link) {
2543 2544
		if (!(new_mask & (1 << dev->devno)) ||
		    dev->class == ATA_DEV_PMP)
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558
			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
2559 2560
	}

2561
	return 0;
T
Tejun Heo 已提交
2562

2563 2564 2565
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
2566 2567 2568
	return rc;
}

2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586
/**
 *	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;
2587 2588
	struct ata_device *dev;
	int rc;
2589

2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600
	/* 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;
		}
	}

2601 2602
	/* has private set_mode? */
	if (ap->ops->set_mode)
2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618
		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;
2619 2620
}

2621
static int ata_link_nr_enabled(struct ata_link *link)
T
Tejun Heo 已提交
2622
{
2623 2624
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2625

2626
	ata_link_for_each_dev(dev, link)
2627
		if (ata_dev_enabled(dev))
T
Tejun Heo 已提交
2628 2629 2630 2631
			cnt++;
	return cnt;
}

2632
static int ata_link_nr_vacant(struct ata_link *link)
T
Tejun Heo 已提交
2633
{
2634 2635
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2636

2637
	ata_link_for_each_dev(dev, link)
2638
		if (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
2639 2640 2641 2642
			cnt++;
	return cnt;
}

2643
static int ata_eh_skip_recovery(struct ata_link *link)
T
Tejun Heo 已提交
2644
{
T
Tejun Heo 已提交
2645
	struct ata_port *ap = link->ap;
2646
	struct ata_eh_context *ehc = &link->eh_context;
2647
	struct ata_device *dev;
T
Tejun Heo 已提交
2648

2649 2650 2651 2652
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

T
Tejun Heo 已提交
2653 2654 2655 2656 2657 2658 2659
	/* 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
2660 2661 2662
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
2663
	ata_link_for_each_dev(dev, link) {
T
Tejun Heo 已提交
2664 2665 2666 2667 2668 2669 2670 2671
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682
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
2683
	ehc->i.action |= ATA_EH_RESET;
2684 2685
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
2686 2687 2688 2689

	return 1;
}

2690
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
2691
{
T
Tejun Heo 已提交
2692
	struct ata_eh_context *ehc = &dev->link->eh_context;
2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703

	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:
2704
		if (ehc->tries[dev->devno] == 1 && dev->pio_mode > XFER_PIO_0) {
2705 2706 2707
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
2708
			sata_down_spd_limit(dev->link);
2709 2710 2711 2712 2713 2714 2715 2716 2717
			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 */
2718
		if (ata_link_offline(dev->link))
2719 2720
			ata_eh_detach_dev(dev);

2721
		/* schedule probe if necessary */
2722
		if (ata_eh_schedule_probe(dev)) {
2723
			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
2724 2725 2726
			memset(ehc->cmd_timeout_idx[dev->devno], 0,
			       sizeof(ehc->cmd_timeout_idx[dev->devno]));
		}
2727 2728

		return 1;
2729
	} else {
T
Tejun Heo 已提交
2730
		ehc->i.action |= ATA_EH_RESET;
2731
		return 0;
2732 2733 2734
	}
}

T
Tejun Heo 已提交
2735 2736 2737
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
2738
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2739 2740 2741
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
2742
 *	@r_failed_link: out parameter for failed link
T
Tejun Heo 已提交
2743 2744 2745
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
2746 2747 2748
 *	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
2749
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
2750 2751 2752 2753 2754 2755 2756
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
2757 2758 2759 2760
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
2761
{
2762
	struct ata_link *link;
T
Tejun Heo 已提交
2763
	struct ata_device *dev;
2764
	int nr_failed_devs;
2765
	int rc;
2766
	unsigned long flags;
T
Tejun Heo 已提交
2767 2768 2769 2770

	DPRINTK("ENTER\n");

	/* prep for recovery */
2771 2772
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2773

2774 2775 2776 2777 2778 2779 2780 2781 2782
		/* 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);
		}

2783
		ata_link_for_each_dev(dev, link) {
2784 2785 2786 2787
			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
2788

2789 2790 2791 2792 2793 2794 2795 2796 2797
			/* 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);

2798 2799 2800
			/* schedule probe if necessary */
			if (!ata_dev_enabled(dev))
				ata_eh_schedule_probe(dev);
T
Tejun Heo 已提交
2801
		}
T
Tejun Heo 已提交
2802 2803 2804 2805
	}

 retry:
	rc = 0;
2806
	nr_failed_devs = 0;
T
Tejun Heo 已提交
2807

2808
	/* if UNLOADING, finish immediately */
2809
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2810 2811
		goto out;

2812 2813 2814
	/* prep for EH */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2815

2816 2817 2818 2819 2820 2821 2822
		/* skip EH if possible. */
		if (ata_eh_skip_recovery(link))
			ehc->i.action = 0;

		ata_link_for_each_dev(dev, link)
			ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
	}
T
Tejun Heo 已提交
2823

T
Tejun Heo 已提交
2824
	/* reset */
2825 2826
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
2827

2828 2829
		if (!(ehc->i.action & ATA_EH_RESET))
			continue;
2830

2831 2832 2833 2834 2835 2836
		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
2837 2838 2839
		}
	}

2840 2841 2842
	/* the rest */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2843

2844 2845
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
2846
		if (rc)
T
Tejun Heo 已提交
2847 2848
			goto dev_fail;

2849 2850 2851 2852 2853 2854
		/* 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;
		}

2855 2856 2857 2858 2859 2860 2861 2862
		/* 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;
		}

2863
		if (ehc->i.action & ATA_EH_LPM)
2864 2865 2866
			ata_link_for_each_dev(dev, link)
				ata_dev_enable_pm(dev, ap->pm_policy);

2867 2868 2869
		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
T
Tejun Heo 已提交
2870

2871
dev_fail:
2872
		nr_failed_devs++;
2873
		ata_eh_handle_dev_fail(dev, rc);
T
Tejun Heo 已提交
2874

2875 2876 2877 2878
		if (ap->pflags & ATA_PFLAG_FROZEN) {
			/* PMP reset requires working host port.
			 * Can't retry if it's frozen.
			 */
T
Tejun Heo 已提交
2879
			if (sata_pmp_attached(ap))
2880
				goto out;
2881
			break;
2882
		}
T
Tejun Heo 已提交
2883 2884
	}

2885
	if (nr_failed_devs)
2886
		goto retry;
T
Tejun Heo 已提交
2887

2888 2889 2890 2891
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

T
Tejun Heo 已提交
2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905
	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.
 */
2906
void ata_eh_finish(struct ata_port *ap)
T
Tejun Heo 已提交
2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921
{
	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
2922
			if (qc->flags & ATA_QCFLAG_RETRY)
T
Tejun Heo 已提交
2923
				ata_eh_qc_retry(qc);
T
Tejun Heo 已提交
2924 2925
			else
				ata_eh_qc_complete(qc);
T
Tejun Heo 已提交
2926 2927 2928 2929 2930 2931 2932 2933 2934 2935
		} 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);
			}
		}
	}
2936 2937 2938 2939

	/* 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
2940 2941 2942 2943 2944
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2945
 *
2946
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2947 2948 2949 2950 2951 2952 2953 2954 2955
 *	@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).
 */
2956 2957 2958
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
2959
{
2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972
	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
2973 2974
	ata_eh_finish(ap);
}
2975

2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989
/**
 *	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;

2990 2991
	/* ignore built-in hardreset if SCR access is not available */
	if (ata_is_builtin_hardreset(hardreset) && !sata_scr_valid(&ap->link))
2992 2993 2994 2995 2996
		hardreset = NULL;

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

2997
#ifdef CONFIG_PM
2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022
/**
 *	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);

3023 3024 3025 3026 3027
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

3028 3029 3030 3031 3032 3033
	/* 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
3034
	ata_acpi_set_state(ap, PMSG_SUSPEND);
3035
 out:
3036 3037 3038 3039 3040 3041
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
3042
	else if (ap->pflags & ATA_PFLAG_FROZEN)
3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066
		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;
3067
	int rc = 0;
3068 3069 3070 3071 3072 3073 3074 3075 3076 3077

	/* 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);

3078
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
3079

S
Shaohua Li 已提交
3080 3081
	ata_acpi_set_state(ap, PMSG_ON);

3082 3083 3084
	if (ap->ops->port_resume)
		rc = ap->ops->port_resume(ap);

T
Tejun Heo 已提交
3085 3086 3087
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

3088
	/* report result */
3089 3090 3091 3092 3093 3094 3095 3096
	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);
}
3097
#endif /* CONFIG_PM */