libata-eh.c 105.7 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>
J
Jens Axboe 已提交
36
#include <linux/blkdev.h>
37
#include <linux/export.h>
38
#include <linux/pci.h>
39 40 41 42 43
#include <scsi/scsi.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_cmnd.h>
44
#include <scsi/scsi_dbg.h>
J
Jeff Garzik 已提交
45
#include "../scsi/scsi_transport_api.h"
46 47 48 49 50

#include <linux/libata.h>

#include "libata.h"

51
enum {
52
	/* speed down verdicts */
53 54 55
	ATA_EH_SPDN_NCQ_OFF		= (1 << 0),
	ATA_EH_SPDN_SPEED_DOWN		= (1 << 1),
	ATA_EH_SPDN_FALLBACK_TO_PIO	= (1 << 2),
56
	ATA_EH_SPDN_KEEP_ERRORS		= (1 << 3),
57 58 59

	/* error flags */
	ATA_EFLAG_IS_IO			= (1 << 0),
60
	ATA_EFLAG_DUBIOUS_XFER		= (1 << 1),
G
Gwendal Grignou 已提交
61
	ATA_EFLAG_OLD_ER                = (1 << 31),
62 63 64 65 66 67

	/* error categories */
	ATA_ECAT_NONE			= 0,
	ATA_ECAT_ATA_BUS		= 1,
	ATA_ECAT_TOUT_HSM		= 2,
	ATA_ECAT_UNK_DEV		= 3,
68 69 70 71 72
	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,
73

74 75
	ATA_EH_CMD_DFL_TIMEOUT		=  5000,

76 77 78
	/* always put at least this amount of time between resets */
	ATA_EH_RESET_COOL_DOWN		=  5000,

79 80 81 82 83 84 85
	/* 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,
86 87

	ATA_EH_UA_TRIES			= 5,
88 89 90 91

	/* probe speed down parameters, see ata_eh_schedule_probe() */
	ATA_EH_PROBE_TRIAL_INTERVAL	= 60000,	/* 1 min */
	ATA_EH_PROBE_TRIALS		= 2,
T
Tejun Heo 已提交
92 93 94 95 96 97 98 99 100
};

/* 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[] = {
101 102 103 104
	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 */
105
	ULONG_MAX, /* > 1 min has elapsed, give up */
T
Tejun Heo 已提交
106 107
};

108 109 110 111 112 113 114
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,
};

115 116 117 118 119 120 121
static const unsigned long ata_eh_flush_timeouts[] = {
	15000,	/* be generous with flush */
	15000,  /* ditto */
	30000,	/* and even more generous */
	ULONG_MAX,
};

122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
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, },
159 160
	{ .commands = CMDS(ATA_CMD_FLUSH, ATA_CMD_FLUSH_EXT),
	  .timeouts = ata_eh_flush_timeouts },
161 162 163
};
#undef CMDS

164
static void __ata_port_freeze(struct ata_port *ap);
165
#ifdef CONFIG_PM
166 167
static void ata_eh_handle_port_suspend(struct ata_port *ap);
static void ata_eh_handle_port_resume(struct ata_port *ap);
168 169 170 171 172 173 174
#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 */
175

T
Tejun Heo 已提交
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
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;
}

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
/**
 *	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
303 304
		ata_port_desc(ap, "%s 0x%llx", name,
				start + (unsigned long long)offset);
305 306 307 308
}

#endif /* CONFIG_PCI */

309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
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]++;
}

376
static void ata_ering_record(struct ata_ering *ering, unsigned int eflags,
377 378 379 380 381 382 383 384 385 386
			     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];
387
	ent->eflags = eflags;
388 389 390 391
	ent->err_mask = err_mask;
	ent->timestamp = get_jiffies_64();
}

392 393 394 395 396 397 398 399 400
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;
}

G
Gwendal Grignou 已提交
401 402 403
int ata_ering_map(struct ata_ering *ering,
		  int (*map_fn)(struct ata_ering_entry *, void *),
		  void *arg)
404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
{
	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;
}

G
Gwendal Grignou 已提交
422 423 424 425 426 427 428 429 430 431 432
int ata_ering_clear_cb(struct ata_ering_entry *ent, void *void_arg)
{
	ent->eflags |= ATA_EFLAG_OLD_ER;
	return 0;
}

static void ata_ering_clear(struct ata_ering *ering)
{
	ata_ering_map(ering, ata_ering_clear_cb, NULL);
}

433 434
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
435
	struct ata_eh_context *ehc = &dev->link->eh_context;
436 437 438 439

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

440
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
441 442
				struct ata_eh_info *ehi, unsigned int action)
{
443
	struct ata_device *tdev;
444 445 446

	if (!dev) {
		ehi->action &= ~action;
T
Tejun Heo 已提交
447
		ata_for_each_dev(tdev, link, ALL)
448
			ehi->dev_action[tdev->devno] &= ~action;
449 450 451 452 453 454
	} 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) {
T
Tejun Heo 已提交
455
			ata_for_each_dev(tdev, link, ALL)
456 457
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
458 459 460 461 462 463 464 465
			ehi->action &= ~action;
		}

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

466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500
/**
 *	ata_eh_acquire - acquire EH ownership
 *	@ap: ATA port to acquire EH ownership for
 *
 *	Acquire EH ownership for @ap.  This is the basic exclusion
 *	mechanism for ports sharing a host.  Only one port hanging off
 *	the same host can claim the ownership of EH.
 *
 *	LOCKING:
 *	EH context.
 */
void ata_eh_acquire(struct ata_port *ap)
{
	mutex_lock(&ap->host->eh_mutex);
	WARN_ON_ONCE(ap->host->eh_owner);
	ap->host->eh_owner = current;
}

/**
 *	ata_eh_release - release EH ownership
 *	@ap: ATA port to release EH ownership for
 *
 *	Release EH ownership for @ap if the caller.  The caller must
 *	have acquired EH ownership using ata_eh_acquire() previously.
 *
 *	LOCKING:
 *	EH context.
 */
void ata_eh_release(struct ata_port *ap)
{
	WARN_ON_ONCE(ap->host->eh_owner != current);
	ap->host->eh_owner = NULL;
	mutex_unlock(&ap->host->eh_mutex);
}

501 502 503 504 505 506 507 508 509 510 511
/**
 *	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.
 *
512 513
 *	TODO: kill this function once old EH is gone.
 *
514 515 516 517 518 519
 *	LOCKING:
 *	Called from timer context
 *
 *	RETURNS:
 *	EH_HANDLED or EH_NOT_HANDLED
 */
J
Jens Axboe 已提交
520
enum blk_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
521 522
{
	struct Scsi_Host *host = cmd->device->host;
523
	struct ata_port *ap = ata_shost_to_port(host);
524 525
	unsigned long flags;
	struct ata_queued_cmd *qc;
J
Jens Axboe 已提交
526
	enum blk_eh_timer_return ret;
527 528 529

	DPRINTK("ENTER\n");

530
	if (ap->ops->error_handler) {
J
Jens Axboe 已提交
531
		ret = BLK_EH_NOT_HANDLED;
532 533 534
		goto out;
	}

J
Jens Axboe 已提交
535
	ret = BLK_EH_HANDLED;
536
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
537
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
538 539 540 541
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
J
Jens Axboe 已提交
542
		ret = BLK_EH_NOT_HANDLED;
543
	}
544
	spin_unlock_irqrestore(ap->lock, flags);
545

546
 out:
547 548 549 550
	DPRINTK("EXIT, ret=%d\n", ret);
	return ret;
}

T
Tejun Heo 已提交
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575
static void ata_eh_unload(struct ata_port *ap)
{
	struct ata_link *link;
	struct ata_device *dev;
	unsigned long flags;

	/* Restore SControl IPM and SPD for the next driver and
	 * disable attached devices.
	 */
	ata_for_each_link(link, ap, PMP_FIRST) {
		sata_scr_write(link, SCR_CONTROL, link->saved_scontrol & 0xff0);
		ata_for_each_dev(dev, link, ALL)
			ata_dev_disable(dev);
	}

	/* freeze and set UNLOADED */
	spin_lock_irqsave(ap->lock, flags);

	ata_port_freeze(ap);			/* won't be thawed */
	ap->pflags &= ~ATA_PFLAG_EH_PENDING;	/* clear pending from freeze */
	ap->pflags |= ATA_PFLAG_UNLOADED;

	spin_unlock_irqrestore(ap->lock, flags);
}

576 577 578 579 580 581 582 583 584 585 586 587
/**
 *	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.
 */
588
void ata_scsi_error(struct Scsi_Host *host)
589
{
590
	struct ata_port *ap = ata_shost_to_port(host);
591
	unsigned long flags;
J
James Bottomley 已提交
592
	LIST_HEAD(eh_work_q);
593 594 595

	DPRINTK("ENTER\n");

J
James Bottomley 已提交
596 597 598 599
	spin_lock_irqsave(host->host_lock, flags);
	list_splice_init(&host->eh_cmd_q, &eh_work_q);
	spin_unlock_irqrestore(host->host_lock, flags);

600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
	ata_scsi_cmd_error_handler(host, ap, &eh_work_q);

	/* If we timed raced normal completion and there is nothing to
	   recover nr_timedout == 0 why exactly are we doing error recovery ? */
	ata_scsi_port_error_handler(host, ap);

	/* finish or retry handled scmd's and clean up */
	WARN_ON(host->host_failed || !list_empty(&eh_work_q));

	DPRINTK("EXIT\n");
}

/**
 * ata_scsi_cmd_error_handler - error callback for a list of commands
 * @host:	scsi host containing the port
 * @ap:		ATA port within the host
 * @eh_work_q:	list of commands to process
 *
 * process the given list of commands and return those finished to the
 * ap->eh_done_q.  This function is the first part of the libata error
 * handler which processes a given list of failed commands.
 */
void ata_scsi_cmd_error_handler(struct Scsi_Host *host, struct ata_port *ap,
				struct list_head *eh_work_q)
{
	int i;
	unsigned long flags;

628 629
	/* make sure sff pio task is not running */
	ata_sff_flush_pio_task(ap);
630

J
Jeff Garzik 已提交
631
	/* synchronize with host lock and sort out timeouts */
632 633 634

	/* For new EH, all qcs are finished in one of three ways -
	 * normal completion, error completion, and SCSI timeout.
A
Alan Cox 已提交
635
	 * Both completions can race against SCSI timeout.  When normal
636 637 638 639 640 641 642 643 644 645 646 647 648
	 * 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;

649
		spin_lock_irqsave(ap->lock, flags);
G
Gwendal Grignou 已提交
650

A
Alan Cox 已提交
651 652
		/* This must occur under the ap->lock as we don't want
		   a polled recovery to race the real interrupt handler
G
Gwendal Grignou 已提交
653

A
Alan Cox 已提交
654 655
		   The lost_interrupt handler checks for any completed but
		   non-notified command and completes much like an IRQ handler.
G
Gwendal Grignou 已提交
656

A
Alan Cox 已提交
657 658 659 660 661
		   We then fall into the error recovery code which will treat
		   this as if normal completion won the race */

		if (ap->ops->lost_interrupt)
			ap->ops->lost_interrupt(ap);
G
Gwendal Grignou 已提交
662

663
		list_for_each_entry_safe(scmd, tmp, eh_work_q, eh_entry) {
664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
			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);

700
		spin_unlock_irqrestore(ap->lock, flags);
701 702 703

		/* initialize eh_tries */
		ap->eh_tries = ATA_EH_MAX_TRIES;
704
	} else
705
		spin_unlock_wait(ap->lock);
G
Gwendal Grignou 已提交
706

707 708 709 710 711 712 713 714 715 716 717 718 719 720
}
EXPORT_SYMBOL(ata_scsi_cmd_error_handler);

/**
 * ata_scsi_port_error_handler - recover the port after the commands
 * @host:	SCSI host containing the port
 * @ap:		the ATA port
 *
 * Handle the recovery of the port @ap after all the commands
 * have been recovered.
 */
void ata_scsi_port_error_handler(struct Scsi_Host *host, struct ata_port *ap)
{
	unsigned long flags;
721 722 723

	/* invoke error handler */
	if (ap->ops->error_handler) {
724 725
		struct ata_link *link;

726 727 728
		/* acquire EH ownership */
		ata_eh_acquire(ap);
 repeat:
T
Tejun Heo 已提交
729 730 731
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

732 733 734
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

735
		/* fetch & clear EH info */
736
		spin_lock_irqsave(ap->lock, flags);
737

T
Tejun Heo 已提交
738
		ata_for_each_link(link, ap, HOST_FIRST) {
739 740 741
			struct ata_eh_context *ehc = &link->eh_context;
			struct ata_device *dev;

742 743 744
			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));
745

T
Tejun Heo 已提交
746
			ata_for_each_dev(dev, link, ENABLED) {
747 748 749 750 751 752
				int devno = dev->devno;

				ehc->saved_xfer_mode[devno] = dev->xfer_mode;
				if (ata_ncq_enabled(dev))
					ehc->saved_ncq_enabled |= 1 << devno;
			}
753
		}
754

755 756
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
757
		ap->excl_link = NULL;	/* don't maintain exclusion over EH */
758

759
		spin_unlock_irqrestore(ap->lock, flags);
760

761 762
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
763
			ap->ops->error_handler(ap);
T
Tejun Heo 已提交
764 765 766 767 768
		else {
			/* if unloading, commence suicide */
			if ((ap->pflags & ATA_PFLAG_UNLOADING) &&
			    !(ap->pflags & ATA_PFLAG_UNLOADED))
				ata_eh_unload(ap);
769
			ata_eh_finish(ap);
T
Tejun Heo 已提交
770
		}
771

772 773 774
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

L
Lucas De Marchi 已提交
775
		/* Exception might have happened after ->error_handler
776 777 778
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
779
		spin_lock_irqsave(ap->lock, flags);
780

781
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
782
			if (--ap->eh_tries) {
783
				spin_unlock_irqrestore(ap->lock, flags);
784 785
				goto repeat;
			}
786 787 788
			ata_port_err(ap,
				     "EH pending after %d tries, giving up\n",
				     ATA_EH_MAX_TRIES);
T
Tejun Heo 已提交
789
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
790 791
		}

792
		/* this run is complete, make sure EH info is clear */
T
Tejun Heo 已提交
793
		ata_for_each_link(link, ap, HOST_FIRST)
794
			memset(&link->eh_info, 0, sizeof(link->eh_info));
795

796
		/* Clear host_eh_scheduled while holding ap->lock such
797 798 799 800 801 802
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

803
		spin_unlock_irqrestore(ap->lock, flags);
804
		ata_eh_release(ap);
805
	} else {
T
Tejun Heo 已提交
806
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
807 808
		ap->ops->eng_timeout(ap);
	}
809 810 811

	scsi_eh_flush_done_q(&ap->eh_done_q);

812
	/* clean up */
813
	spin_lock_irqsave(ap->lock, flags);
814

815
	if (ap->pflags & ATA_PFLAG_LOADING)
816
		ap->pflags &= ~ATA_PFLAG_LOADING;
817
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
818
		schedule_delayed_work(&ap->hotplug_task, 0);
819 820

	if (ap->pflags & ATA_PFLAG_RECOVERED)
821
		ata_port_info(ap, "EH complete\n");
822

823
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
824

825
	/* tell wait_eh that we're done */
826
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
827 828
	wake_up_all(&ap->eh_wait_q);

829
	spin_unlock_irqrestore(ap->lock, flags);
830
}
831
EXPORT_SYMBOL_GPL(ata_scsi_port_error_handler);
832

833 834 835 836 837 838 839 840 841 842 843 844 845 846 847
/**
 *	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:
848
	spin_lock_irqsave(ap->lock, flags);
849

850
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
851
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
852
		spin_unlock_irqrestore(ap->lock, flags);
853
		schedule();
854
		spin_lock_irqsave(ap->lock, flags);
855
	}
856
	finish_wait(&ap->eh_wait_q, &wait);
857

858
	spin_unlock_irqrestore(ap->lock, flags);
859 860

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
861
	if (scsi_host_in_recovery(ap->scsi_host)) {
862
		ata_msleep(ap, 10);
863 864 865
		goto retry;
	}
}
866
EXPORT_SYMBOL_GPL(ata_port_wait_eh);
867

T
Tejun Heo 已提交
868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911
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 =
912
			ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
T
Tejun Heo 已提交
913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951
		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;
952 953
	ap->fastdrain_timer.expires =
		ata_deadline(jiffies, ATA_EH_FASTDRAIN_INTERVAL);
T
Tejun Heo 已提交
954 955 956
	add_timer(&ap->fastdrain_timer);
}

957 958 959 960 961 962 963 964
/**
 *	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
965
 *	spin_lock_irqsave(host lock)
966 967 968 969
 */
void ata_qc_schedule_eh(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;
970 971
	struct request_queue *q = qc->scsicmd->device->request_queue;
	unsigned long flags;
972 973 974 975

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

	qc->flags |= ATA_QCFLAG_FAILED;
T
Tejun Heo 已提交
976
	ata_eh_set_pending(ap, 1);
977 978 979 980 981 982

	/* 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.
	 */
983
	spin_lock_irqsave(q->queue_lock, flags);
J
Jens Axboe 已提交
984
	blk_abort_request(qc->scsicmd->request);
985
	spin_unlock_irqrestore(q->queue_lock, flags);
986 987
}

988 989 990 991 992 993 994 995
/**
 *	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
996
 *	spin_lock_irqsave(host lock)
997 998 999 1000 1001
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

1002 1003 1004
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
1005
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
1006
	scsi_schedule_eh(ap->scsi_host);
1007 1008 1009 1010

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

1011
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
1012 1013 1014 1015 1016
{
	int tag, nr_aborted = 0;

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

T
Tejun Heo 已提交
1017 1018 1019
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

1020 1021 1022
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

1023
		if (qc && (!link || qc->dev->link == link)) {
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
/**
 *	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);
}

1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
/**
 *	__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
1086
 *	spin_lock_irqsave(host lock)
1087 1088 1089 1090 1091 1092 1093 1094
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

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

1095
	ap->pflags |= ATA_PFLAG_FROZEN;
1096

T
Tejun Heo 已提交
1097
	DPRINTK("ata%u port frozen\n", ap->print_id);
1098 1099 1100 1101 1102 1103
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
1104 1105 1106
 *	Abort and freeze @ap.  The freeze operation must be called
 *	first, because some hardware requires special operations
 *	before the taskfile registers are accessible.
1107 1108
 *
 *	LOCKING:
J
Jeff Garzik 已提交
1109
 *	spin_lock_irqsave(host lock)
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
 *
 *	RETURNS:
 *	Number of aborted commands.
 */
int ata_port_freeze(struct ata_port *ap)
{
	int nr_aborted;

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

	__ata_port_freeze(ap);
1121
	nr_aborted = ata_port_abort(ap);
1122 1123 1124 1125

	return nr_aborted;
}

1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
/**
 *	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
1151
	if (!sata_pmp_attached(ap) || rc) {
1152
		/* PMP is not attached or SNTF is not available */
T
Tejun Heo 已提交
1153
		if (!sata_pmp_attached(ap)) {
1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177
			/* 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 */
T
Tejun Heo 已提交
1178
		ata_for_each_link(link, ap, EDGE) {
1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
			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;
	}
}

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214
/**
 *	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;

1215
	spin_lock_irqsave(ap->lock, flags);
1216
	__ata_port_freeze(ap);
1217
	spin_unlock_irqrestore(ap->lock, flags);
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
}

/**
 *	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;

1236
	spin_lock_irqsave(ap->lock, flags);
1237

1238
	ap->pflags &= ~ATA_PFLAG_FROZEN;
1239 1240 1241 1242

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

1243
	spin_unlock_irqrestore(ap->lock, flags);
1244

T
Tejun Heo 已提交
1245
	DPRINTK("ata%u port thawed\n", ap->print_id);
1246 1247
}

1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258
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;

1259
	spin_lock_irqsave(ap->lock, flags);
1260 1261 1262
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
1263
	spin_unlock_irqrestore(ap->lock, flags);
1264 1265 1266 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 1299

	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
1300

1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315
/**
 *	ata_dev_disable - disable ATA device
 *	@dev: ATA device to disable
 *
 *	Disable @dev.
 *
 *	Locking:
 *	EH context.
 */
void ata_dev_disable(struct ata_device *dev)
{
	if (!ata_dev_enabled(dev))
		return;

	if (ata_msg_drv(dev->link->ap))
1316
		ata_dev_warn(dev, "disabled\n");
1317 1318 1319
	ata_acpi_on_disable(dev);
	ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO0 | ATA_DNXFER_QUIET);
	dev->class++;
1320 1321 1322 1323 1324

	/* From now till the next successful probe, ering is used to
	 * track probe failures.  Clear accumulated device error info.
	 */
	ata_ering_clear(&dev->ering);
1325 1326
}

1327 1328 1329 1330 1331 1332 1333 1334 1335
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
1336
void ata_eh_detach_dev(struct ata_device *dev)
1337
{
1338 1339
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1340
	struct ata_eh_context *ehc = &link->eh_context;
1341 1342 1343 1344
	unsigned long flags;

	ata_dev_disable(dev);

1345
	spin_lock_irqsave(ap->lock, flags);
1346 1347 1348 1349 1350

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1351
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1352 1353
	}

1354
	/* clear per-dev EH info */
1355 1356
	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);
1357 1358
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
1359

1360
	spin_unlock_irqrestore(ap->lock, flags);
1361 1362
}

T
Tejun Heo 已提交
1363 1364
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
1365
 *	@link: target ATA link
1366
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
1367 1368 1369
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
1370 1371
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
1372 1373 1374 1375
 *
 *	LOCKING:
 *	None.
 */
1376 1377
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
T
Tejun Heo 已提交
1378
{
T
Tejun Heo 已提交
1379 1380 1381
	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
1382 1383
	unsigned long flags;

1384
	spin_lock_irqsave(ap->lock, flags);
1385

T
Tejun Heo 已提交
1386
	ata_eh_clear_action(link, dev, ehi, action);
1387

1388 1389 1390 1391
	/* About to take EH action, set RECOVERED.  Ignore actions on
	 * slave links as master will do them again.
	 */
	if (!(ehc->i.flags & ATA_EHI_QUIET) && link != ap->slave_link)
1392 1393
		ap->pflags |= ATA_PFLAG_RECOVERED;

1394
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
1395 1396
}

1397 1398
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
1399
*	@ap: target ATA port
1400 1401 1402 1403
 *	@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
1404
 *	in @link->eh_context.
1405 1406 1407 1408
 *
 *	LOCKING:
 *	None.
 */
1409 1410
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1411
{
T
Tejun Heo 已提交
1412
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1413

T
Tejun Heo 已提交
1414
	ata_eh_clear_action(link, dev, &ehc->i, action);
1415 1416
}

T
Tejun Heo 已提交
1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
/**
 *	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
 */
1431
static const char *ata_err_string(unsigned int err_mask)
T
Tejun Heo 已提交
1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
{
	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";
}

1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
/**
 *	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,
1484
				     buf, sectors * ATA_SECT_SIZE, 0);
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507

	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
1508
	u8 *buf = dev->link->ap->sector_buf;
1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
	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)
1521 1522
		ata_dev_warn(dev, "invalid checksum 0x%x on log page 10h\n",
			     csum);
1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543

	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;
}

1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
/**
 *	atapi_eh_tur - perform ATAPI TEST_UNIT_READY
 *	@dev: target ATAPI device
 *	@r_sense_key: out parameter for sense_key
 *
 *	Perform ATAPI TEST_UNIT_READY.
 *
 *	LOCKING:
 *	EH context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask on failure.
 */
static unsigned int atapi_eh_tur(struct ata_device *dev, u8 *r_sense_key)
{
	u8 cdb[ATAPI_CDB_LEN] = { TEST_UNIT_READY, 0, 0, 0, 0, 0 };
	struct ata_taskfile tf;
	unsigned int err_mask;

	ata_tf_init(dev, &tf);

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

	err_mask = ata_exec_internal(dev, &tf, cdb, DMA_NONE, NULL, 0, 0);
	if (err_mask == AC_ERR_DEV)
		*r_sense_key = tf.feature >> 4;
	return err_mask;
}

T
Tejun Heo 已提交
1575 1576 1577 1578
/**
 *	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)
1579
 *	@dfl_sense_key: default sense key to use
T
Tejun Heo 已提交
1580 1581 1582 1583 1584 1585 1586 1587 1588 1589
 *
 *	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
 */
1590 1591
static unsigned int atapi_eh_request_sense(struct ata_device *dev,
					   u8 *sense_buf, u8 dfl_sense_key)
T
Tejun Heo 已提交
1592
{
1593 1594
	u8 cdb[ATAPI_CDB_LEN] =
		{ REQUEST_SENSE, 0, 0, 0, SCSI_SENSE_BUFFERSIZE, 0 };
T
Tejun Heo 已提交
1595
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1596 1597 1598 1599 1600 1601 1602
	struct ata_taskfile tf;

	DPRINTK("ATAPI request sense\n");

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

1603 1604 1605
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1606
	sense_buf[0] = 0x70;
1607
	sense_buf[2] = dfl_sense_key;
1608

J
Jeff Garzik 已提交
1609
	/* some devices time out if garbage left in tf */
1610
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1611 1612 1613 1614 1615 1616

	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) {
1617
		tf.protocol = ATAPI_PROT_DMA;
T
Tejun Heo 已提交
1618 1619
		tf.feature |= ATAPI_PKT_DMA;
	} else {
1620
		tf.protocol = ATAPI_PROT_PIO;
1621 1622
		tf.lbam = SCSI_SENSE_BUFFERSIZE;
		tf.lbah = 0;
T
Tejun Heo 已提交
1623 1624 1625
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
1626
				 sense_buf, SCSI_SENSE_BUFFERSIZE, 0);
T
Tejun Heo 已提交
1627 1628 1629 1630
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1631
 *	@link: ATA link to analyze SError for
T
Tejun Heo 已提交
1632 1633 1634 1635 1636 1637 1638
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1639
static void ata_eh_analyze_serror(struct ata_link *link)
T
Tejun Heo 已提交
1640
{
1641
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1642 1643
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1644
	u32 hotplug_mask;
T
Tejun Heo 已提交
1645

T
Tejun Heo 已提交
1646
	if (serror & (SERR_PERSISTENT | SERR_DATA)) {
T
Tejun Heo 已提交
1647
		err_mask |= AC_ERR_ATA_BUS;
T
Tejun Heo 已提交
1648
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1649 1650 1651
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
T
Tejun Heo 已提交
1652
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1653 1654 1655
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
T
Tejun Heo 已提交
1656
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1657
	}
1658 1659 1660 1661 1662 1663

	/* 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.
	 */
1664
	if (link->lpm_policy > ATA_LPM_MAX_POWER)
1665 1666
		hotplug_mask = 0;	/* hotplug doesn't work w/ LPM */
	else if (!(link->flags & ATA_LFLAG_DISABLED) || ata_is_host_link(link))
1667 1668 1669 1670 1671
		hotplug_mask = SERR_PHYRDY_CHG | SERR_DEV_XCHG;
	else
		hotplug_mask = SERR_PHYRDY_CHG;

	if (serror & hotplug_mask)
T
Tejun Heo 已提交
1672
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1673 1674 1675 1676 1677

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

1678 1679
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
1680
 *	@link: ATA link to analyze NCQ error for
1681 1682 1683 1684 1685 1686 1687 1688 1689
 *
 *	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).
 */
1690
void ata_eh_analyze_ncq_error(struct ata_link *link)
1691
{
1692 1693 1694
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
1695 1696 1697 1698 1699
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1700
	if (ap->pflags & ATA_PFLAG_FROZEN)
1701 1702 1703
		return;

	/* is it NCQ device error? */
1704
	if (!link->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
		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 */
1719
	memset(&tf, 0, sizeof(tf));
1720 1721
	rc = ata_eh_read_log_10h(dev, &tag, &tf);
	if (rc) {
1722 1723
		ata_link_err(link, "failed to read log page 10h (errno=%d)\n",
			     rc);
1724 1725 1726
		return;
	}

1727
	if (!(link->sactive & (1 << tag))) {
1728 1729
		ata_link_err(link, "log page 10h reported inactive tag %d\n",
			     tag);
1730 1731 1732 1733 1734 1735
		return;
	}

	/* we've got the perpetrator, condemn it */
	qc = __ata_qc_from_tag(ap, tag);
	memcpy(&qc->result_tf, &tf, sizeof(tf));
1736
	qc->result_tf.flags = ATA_TFLAG_ISADDR | ATA_TFLAG_LBA | ATA_TFLAG_LBA48;
T
Tejun Heo 已提交
1737
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1738 1739 1740
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1741 1742 1743 1744 1745 1746 1747
/**
 *	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
L
Lucas De Marchi 已提交
1748
 *	available.
T
Tejun Heo 已提交
1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
 *
 *	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
1764
		return ATA_EH_RESET;
T
Tejun Heo 已提交
1765 1766
	}

1767 1768 1769
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
		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:
1783
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1784 1785 1786
			tmp = atapi_eh_request_sense(qc->dev,
						qc->scsicmd->sense_buffer,
						qc->result_tf.feature >> 4);
1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
			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
1799 1800 1801
	}

	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
T
Tejun Heo 已提交
1802
		action |= ATA_EH_RESET;
T
Tejun Heo 已提交
1803 1804 1805 1806

	return action;
}

1807 1808
static int ata_eh_categorize_error(unsigned int eflags, unsigned int err_mask,
				   int *xfer_ok)
T
Tejun Heo 已提交
1809
{
1810 1811 1812 1813 1814 1815
	int base = 0;

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

	if (!*xfer_ok)
1816
		base = ATA_ECAT_DUBIOUS_NONE;
1817

1818
	if (err_mask & AC_ERR_ATA_BUS)
1819
		return base + ATA_ECAT_ATA_BUS;
T
Tejun Heo 已提交
1820

1821
	if (err_mask & AC_ERR_TIMEOUT)
1822
		return base + ATA_ECAT_TOUT_HSM;
1823

1824
	if (eflags & ATA_EFLAG_IS_IO) {
1825
		if (err_mask & AC_ERR_HSM)
1826
			return base + ATA_ECAT_TOUT_HSM;
1827 1828
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
1829
			return base + ATA_ECAT_UNK_DEV;
T
Tejun Heo 已提交
1830 1831 1832 1833 1834
	}

	return 0;
}

1835
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1836
	u64 since;
1837
	int xfer_ok;
1838
	int nr_errors[ATA_ECAT_NR];
T
Tejun Heo 已提交
1839 1840
};

1841
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1842
{
1843
	struct speed_down_verdict_arg *arg = void_arg;
1844
	int cat;
T
Tejun Heo 已提交
1845

G
Gwendal Grignou 已提交
1846
	if ((ent->eflags & ATA_EFLAG_OLD_ER) || (ent->timestamp < arg->since))
T
Tejun Heo 已提交
1847 1848
		return -1;

1849 1850
	cat = ata_eh_categorize_error(ent->eflags, ent->err_mask,
				      &arg->xfer_ok);
1851
	arg->nr_errors[cat]++;
1852

T
Tejun Heo 已提交
1853 1854 1855 1856
	return 0;
}

/**
1857
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1858 1859 1860
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1861 1862
 *	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
1863
 *
1864 1865 1866 1867 1868 1869 1870
 *	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
 *
1871 1872 1873
 *	ECAT_DUBIOUS_*	: Identical to above three but occurred while
 *			  data transfer hasn't been verified.
 *
1874 1875 1876
 *	Verdicts are
 *
 *	NCQ_OFF		: Turn off NCQ.
T
Tejun Heo 已提交
1877
 *
1878 1879
 *	SPEED_DOWN	: Speed down transfer speed but don't fall back
 *			  to PIO.
1880
 *
1881
 *	FALLBACK_TO_PIO	: Fall back to PIO.
T
Tejun Heo 已提交
1882
 *
1883
 *	Even if multiple verdicts are returned, only one action is
1884 1885 1886 1887 1888 1889 1890
 *	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.
1891
 *
1892 1893 1894 1895 1896 1897 1898
 *	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
L
Lucas De Marchi 已提交
1899
 *	   occurred during last 5 mins, FALLBACK_TO_PIO
1900
 *
1901
 *	4. If more than 3 TOUT_HSM or UNK_DEV errors occurred
1902 1903
 *	   during last 10 mins, NCQ_OFF.
 *
1904
 *	5. If more than 3 ATA_BUS or TOUT_HSM errors, or more than 6
1905
 *	   UNK_DEV errors occurred during last 10 mins, SPEED_DOWN.
1906
 *
T
Tejun Heo 已提交
1907 1908 1909 1910
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1911
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1912
 */
1913
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1914
{
1915 1916 1917 1918
	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
1919

1920
	/* scan past 5 mins of error history */
1921
	memset(&arg, 0, sizeof(arg));
1922
	arg.since = j64 - min(j64, j5mins);
1923
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1924

1925 1926 1927 1928 1929 1930 1931 1932 1933
	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;

1934 1935
	if (arg.nr_errors[ATA_ECAT_ATA_BUS] +
	    arg.nr_errors[ATA_ECAT_TOUT_HSM] +
T
Tejun Heo 已提交
1936
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1937
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
T
Tejun Heo 已提交
1938

1939
	/* scan past 10 mins of error history */
T
Tejun Heo 已提交
1940
	memset(&arg, 0, sizeof(arg));
1941
	arg.since = j64 - min(j64, j10mins);
1942
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1943

1944 1945 1946 1947 1948 1949
	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
1950
	    arg.nr_errors[ATA_ECAT_UNK_DEV] > 6)
1951
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
T
Tejun Heo 已提交
1952

1953
	return verdict;
T
Tejun Heo 已提交
1954 1955 1956 1957 1958
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
1959
 *	@eflags: mask of ATA_EFLAG_* flags
T
Tejun Heo 已提交
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970
 *	@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:
1971
 *	Determined recovery action.
T
Tejun Heo 已提交
1972
 */
1973 1974
static unsigned int ata_eh_speed_down(struct ata_device *dev,
				unsigned int eflags, unsigned int err_mask)
T
Tejun Heo 已提交
1975
{
T
Tejun Heo 已提交
1976
	struct ata_link *link = ata_dev_phys_link(dev);
1977
	int xfer_ok = 0;
1978 1979 1980 1981
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
1982
	if (ata_eh_categorize_error(eflags, err_mask, &xfer_ok) == 0)
T
Tejun Heo 已提交
1983 1984 1985
		return 0;

	/* record error and determine whether speed down is necessary */
1986
	ata_ering_record(&dev->ering, eflags, err_mask);
1987
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1988

1989 1990 1991 1992 1993
	/* 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;
1994
		ata_dev_warn(dev, "NCQ disabled due to excessive errors\n");
1995 1996
		goto done;
	}
T
Tejun Heo 已提交
1997

1998 1999 2000
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
2001
		if (sata_down_spd_limit(link, 0) == 0) {
T
Tejun Heo 已提交
2002
			action |= ATA_EH_RESET;
2003 2004
			goto done;
		}
T
Tejun Heo 已提交
2005

2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
		/* 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
2022
				action |= ATA_EH_RESET;
2023 2024 2025 2026 2027 2028
				goto done;
			}
		}
	}

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
T
Tejun Heo 已提交
2029
	 * SATA ATA devices.  Consider it only for PATA and SATAPI.
2030 2031
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
T
Tejun Heo 已提交
2032
	    (link->ap->cbl != ATA_CBL_SATA || dev->class == ATA_DEV_ATAPI) &&
2033 2034 2035
	    (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
2036
			action |= ATA_EH_RESET;
2037 2038 2039
			goto done;
		}
	}
T
Tejun Heo 已提交
2040 2041

	return 0;
2042 2043
 done:
	/* device has been slowed down, blow error history */
2044 2045
	if (!(verdict & ATA_EH_SPDN_KEEP_ERRORS))
		ata_ering_clear(&dev->ering);
2046
	return action;
T
Tejun Heo 已提交
2047 2048 2049
}

/**
2050 2051
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
T
Tejun Heo 已提交
2052
 *
2053 2054 2055
 *	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
2056 2057 2058 2059
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
2060
static void ata_eh_link_autopsy(struct ata_link *link)
T
Tejun Heo 已提交
2061
{
2062
	struct ata_port *ap = link->ap;
2063
	struct ata_eh_context *ehc = &link->eh_context;
2064
	struct ata_device *dev;
2065 2066
	unsigned int all_err_mask = 0, eflags = 0;
	int tag;
T
Tejun Heo 已提交
2067 2068 2069 2070 2071
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

2072 2073 2074
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
2075
	/* obtain and analyze SError */
2076
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
2077 2078
	if (rc == 0) {
		ehc->i.serror |= serror;
2079
		ata_eh_analyze_serror(link);
2080
	} else if (rc != -EOPNOTSUPP) {
T
Tejun Heo 已提交
2081
		/* SError read failed, force reset and probing */
2082
		ehc->i.probe_mask |= ATA_ALL_DEVICES;
T
Tejun Heo 已提交
2083
		ehc->i.action |= ATA_EH_RESET;
2084 2085
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
2086

2087
	/* analyze NCQ failure */
2088
	ata_eh_analyze_ncq_error(link);
2089

T
Tejun Heo 已提交
2090 2091 2092 2093 2094 2095 2096 2097 2098
	/* 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);

T
Tejun Heo 已提交
2099 2100
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    ata_dev_phys_link(qc->dev) != link)
T
Tejun Heo 已提交
2101 2102 2103 2104 2105 2106
			continue;

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

		/* analyze TF */
2107
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118

		/* 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 */
2119
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
T
Tejun Heo 已提交
2120 2121
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);

T
Tejun Heo 已提交
2122
		/* determine whether the command is worth retrying */
2123 2124 2125
		if (qc->flags & ATA_QCFLAG_IO ||
		    (!(qc->err_mask & AC_ERR_INVALID) &&
		     qc->err_mask != AC_ERR_DEV))
T
Tejun Heo 已提交
2126 2127
			qc->flags |= ATA_QCFLAG_RETRY;

T
Tejun Heo 已提交
2128
		/* accumulate error info */
2129
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
2130 2131
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
2132
			eflags |= ATA_EFLAG_IS_IO;
T
Tejun Heo 已提交
2133 2134
	}

2135
	/* enforce default EH actions */
2136
	if (ap->pflags & ATA_PFLAG_FROZEN ||
2137
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
T
Tejun Heo 已提交
2138
		ehc->i.action |= ATA_EH_RESET;
2139 2140
	else if (((eflags & ATA_EFLAG_IS_IO) && all_err_mask) ||
		 (!(eflags & ATA_EFLAG_IS_IO) && (all_err_mask & ~AC_ERR_DEV)))
2141
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
2142

2143 2144 2145
	/* If we have offending qcs and the associated failed device,
	 * perform per-dev EH action only on the offending device.
	 */
2146 2147 2148 2149
	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;
2150 2151
	}

2152 2153 2154 2155 2156
	/* 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 */
2157
	dev = ehc->i.dev;
2158 2159 2160
	if (!dev && ((ata_link_max_devices(link) == 1 &&
		      ata_dev_enabled(link->device))))
	    dev = link->device;
2161

2162 2163 2164
	if (dev) {
		if (dev->flags & ATA_DFLAG_DUBIOUS_XFER)
			eflags |= ATA_EFLAG_DUBIOUS_XFER;
2165
		ehc->i.action |= ata_eh_speed_down(dev, eflags, all_err_mask);
2166
	}
2167

T
Tejun Heo 已提交
2168 2169 2170 2171
	DPRINTK("EXIT\n");
}

/**
2172 2173 2174 2175 2176 2177 2178 2179 2180
 *	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).
 */
2181
void ata_eh_autopsy(struct ata_port *ap)
2182 2183 2184
{
	struct ata_link *link;

T
Tejun Heo 已提交
2185
	ata_for_each_link(link, ap, EDGE)
2186
		ata_eh_link_autopsy(link);
2187

T
Tejun Heo 已提交
2188 2189 2190 2191 2192 2193 2194 2195
	/* Handle the frigging slave link.  Autopsy is done similarly
	 * but actions and flags are transferred over to the master
	 * link and handled from there.
	 */
	if (ap->slave_link) {
		struct ata_eh_context *mehc = &ap->link.eh_context;
		struct ata_eh_context *sehc = &ap->slave_link->eh_context;

2196 2197 2198 2199
		/* transfer control flags from master to slave */
		sehc->i.flags |= mehc->i.flags & ATA_EHI_TO_SLAVE_MASK;

		/* perform autopsy on the slave link */
T
Tejun Heo 已提交
2200 2201
		ata_eh_link_autopsy(ap->slave_link);

2202
		/* transfer actions from slave to master and clear slave */
T
Tejun Heo 已提交
2203 2204 2205 2206 2207 2208 2209
		ata_eh_about_to_do(ap->slave_link, NULL, ATA_EH_ALL_ACTIONS);
		mehc->i.action		|= sehc->i.action;
		mehc->i.dev_action[1]	|= sehc->i.dev_action[1];
		mehc->i.flags		|= sehc->i.flags;
		ata_eh_done(ap->slave_link, NULL, ATA_EH_ALL_ACTIONS);
	}

2210 2211 2212
	/* Autopsy of fanout ports can affect host link autopsy.
	 * Perform host link autopsy last.
	 */
T
Tejun Heo 已提交
2213
	if (sata_pmp_attached(ap))
2214
		ata_eh_link_autopsy(&ap->link);
2215 2216
}

2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303
/**
 *	ata_get_cmd_descript - get description for ATA command
 *	@command: ATA command code to get description for
 *
 *	Return a textual description of the given command, or NULL if the
 *	command is not known.
 *
 *	LOCKING:
 *	None
 */
const char *ata_get_cmd_descript(u8 command)
{
#ifdef CONFIG_ATA_VERBOSE_ERROR
	static const struct
	{
		u8 command;
		const char *text;
	} cmd_descr[] = {
		{ ATA_CMD_DEV_RESET,		"DEVICE RESET" },
		{ ATA_CMD_CHK_POWER, 		"CHECK POWER MODE" },
		{ ATA_CMD_STANDBY, 		"STANDBY" },
		{ ATA_CMD_IDLE, 		"IDLE" },
		{ ATA_CMD_EDD, 			"EXECUTE DEVICE DIAGNOSTIC" },
		{ ATA_CMD_DOWNLOAD_MICRO,   	"DOWNLOAD MICROCODE" },
		{ ATA_CMD_NOP,			"NOP" },
		{ ATA_CMD_FLUSH, 		"FLUSH CACHE" },
		{ ATA_CMD_FLUSH_EXT, 		"FLUSH CACHE EXT" },
		{ ATA_CMD_ID_ATA,  		"IDENTIFY DEVICE" },
		{ ATA_CMD_ID_ATAPI, 		"IDENTIFY PACKET DEVICE" },
		{ ATA_CMD_SERVICE, 		"SERVICE" },
		{ ATA_CMD_READ, 		"READ DMA" },
		{ ATA_CMD_READ_EXT, 		"READ DMA EXT" },
		{ ATA_CMD_READ_QUEUED, 		"READ DMA QUEUED" },
		{ ATA_CMD_READ_STREAM_EXT, 	"READ STREAM EXT" },
		{ ATA_CMD_READ_STREAM_DMA_EXT,  "READ STREAM DMA EXT" },
		{ ATA_CMD_WRITE, 		"WRITE DMA" },
		{ ATA_CMD_WRITE_EXT, 		"WRITE DMA EXT" },
		{ ATA_CMD_WRITE_QUEUED, 	"WRITE DMA QUEUED EXT" },
		{ ATA_CMD_WRITE_STREAM_EXT, 	"WRITE STREAM EXT" },
		{ ATA_CMD_WRITE_STREAM_DMA_EXT, "WRITE STREAM DMA EXT" },
		{ ATA_CMD_WRITE_FUA_EXT,	"WRITE DMA FUA EXT" },
		{ ATA_CMD_WRITE_QUEUED_FUA_EXT, "WRITE DMA QUEUED FUA EXT" },
		{ ATA_CMD_FPDMA_READ,		"READ FPDMA QUEUED" },
		{ ATA_CMD_FPDMA_WRITE,		"WRITE FPDMA QUEUED" },
		{ ATA_CMD_PIO_READ,		"READ SECTOR(S)" },
		{ ATA_CMD_PIO_READ_EXT,		"READ SECTOR(S) EXT" },
		{ ATA_CMD_PIO_WRITE,		"WRITE SECTOR(S)" },
		{ ATA_CMD_PIO_WRITE_EXT,	"WRITE SECTOR(S) EXT" },
		{ ATA_CMD_READ_MULTI,		"READ MULTIPLE" },
		{ ATA_CMD_READ_MULTI_EXT,	"READ MULTIPLE EXT" },
		{ ATA_CMD_WRITE_MULTI,		"WRITE MULTIPLE" },
		{ ATA_CMD_WRITE_MULTI_EXT,	"WRITE MULTIPLE EXT" },
		{ ATA_CMD_WRITE_MULTI_FUA_EXT, 	"WRITE MULTIPLE FUA EXT" },
		{ ATA_CMD_SET_FEATURES,		"SET FEATURES" },
		{ ATA_CMD_SET_MULTI,		"SET MULTIPLE MODE" },
		{ ATA_CMD_VERIFY,		"READ VERIFY SECTOR(S)" },
		{ ATA_CMD_VERIFY_EXT,		"READ VERIFY SECTOR(S) EXT" },
		{ ATA_CMD_WRITE_UNCORR_EXT,	"WRITE UNCORRECTABLE EXT" },
		{ ATA_CMD_STANDBYNOW1,		"STANDBY IMMEDIATE" },
		{ ATA_CMD_IDLEIMMEDIATE,	"IDLE IMMEDIATE" },
		{ ATA_CMD_SLEEP,		"SLEEP" },
		{ ATA_CMD_INIT_DEV_PARAMS,	"INITIALIZE DEVICE PARAMETERS" },
		{ ATA_CMD_READ_NATIVE_MAX,	"READ NATIVE MAX ADDRESS" },
		{ ATA_CMD_READ_NATIVE_MAX_EXT,	"READ NATIVE MAX ADDRESS EXT" },
		{ ATA_CMD_SET_MAX,		"SET MAX ADDRESS" },
		{ ATA_CMD_SET_MAX_EXT,		"SET MAX ADDRESS EXT" },
		{ ATA_CMD_READ_LOG_EXT,		"READ LOG EXT" },
		{ ATA_CMD_WRITE_LOG_EXT,	"WRITE LOG EXT" },
		{ ATA_CMD_READ_LOG_DMA_EXT,	"READ LOG DMA EXT" },
		{ ATA_CMD_WRITE_LOG_DMA_EXT, 	"WRITE LOG DMA EXT" },
		{ ATA_CMD_TRUSTED_RCV,		"TRUSTED RECEIVE" },
		{ ATA_CMD_TRUSTED_RCV_DMA, 	"TRUSTED RECEIVE DMA" },
		{ ATA_CMD_TRUSTED_SND,		"TRUSTED SEND" },
		{ ATA_CMD_TRUSTED_SND_DMA, 	"TRUSTED SEND DMA" },
		{ ATA_CMD_PMP_READ,		"READ BUFFER" },
		{ ATA_CMD_PMP_WRITE,		"WRITE BUFFER" },
		{ ATA_CMD_CONF_OVERLAY,		"DEVICE CONFIGURATION OVERLAY" },
		{ ATA_CMD_SEC_SET_PASS,		"SECURITY SET PASSWORD" },
		{ ATA_CMD_SEC_UNLOCK,		"SECURITY UNLOCK" },
		{ ATA_CMD_SEC_ERASE_PREP,	"SECURITY ERASE PREPARE" },
		{ ATA_CMD_SEC_ERASE_UNIT,	"SECURITY ERASE UNIT" },
		{ ATA_CMD_SEC_FREEZE_LOCK,	"SECURITY FREEZE LOCK" },
		{ ATA_CMD_SEC_DISABLE_PASS,	"SECURITY DISABLE PASSWORD" },
		{ ATA_CMD_CONFIG_STREAM,	"CONFIGURE STREAM" },
		{ ATA_CMD_SMART,		"SMART" },
		{ ATA_CMD_MEDIA_LOCK,		"DOOR LOCK" },
		{ ATA_CMD_MEDIA_UNLOCK,		"DOOR UNLOCK" },
2304
		{ ATA_CMD_DSM,			"DATA SET MANAGEMENT" },
2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
		{ ATA_CMD_CHK_MED_CRD_TYP, 	"CHECK MEDIA CARD TYPE" },
		{ ATA_CMD_CFA_REQ_EXT_ERR, 	"CFA REQUEST EXTENDED ERROR" },
		{ ATA_CMD_CFA_WRITE_NE,		"CFA WRITE SECTORS WITHOUT ERASE" },
		{ ATA_CMD_CFA_TRANS_SECT,	"CFA TRANSLATE SECTOR" },
		{ ATA_CMD_CFA_ERASE,		"CFA ERASE SECTORS" },
		{ ATA_CMD_CFA_WRITE_MULT_NE, 	"CFA WRITE MULTIPLE WITHOUT ERASE" },
		{ ATA_CMD_READ_LONG,		"READ LONG (with retries)" },
		{ ATA_CMD_READ_LONG_ONCE,	"READ LONG (without retries)" },
		{ ATA_CMD_WRITE_LONG,		"WRITE LONG (with retries)" },
		{ ATA_CMD_WRITE_LONG_ONCE,	"WRITE LONG (without retries)" },
		{ ATA_CMD_RESTORE,		"RECALIBRATE" },
		{ 0,				NULL } /* terminate list */
	};

	unsigned int i;
	for (i = 0; cmd_descr[i].text; i++)
		if (cmd_descr[i].command == command)
			return cmd_descr[i].text;
#endif

	return NULL;
}

2328 2329
/**
 *	ata_eh_link_report - report error handling to user
2330
 *	@link: ATA link EH is going on
T
Tejun Heo 已提交
2331 2332 2333 2334 2335 2336
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2337
static void ata_eh_link_report(struct ata_link *link)
T
Tejun Heo 已提交
2338
{
2339 2340
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2341
	const char *frozen, *desc;
2342
	char tries_buf[6];
T
Tejun Heo 已提交
2343 2344
	int tag, nr_failed = 0;

2345 2346 2347
	if (ehc->i.flags & ATA_EHI_QUIET)
		return;

T
Tejun Heo 已提交
2348 2349 2350 2351 2352 2353 2354
	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);

T
Tejun Heo 已提交
2355 2356
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    ata_dev_phys_link(qc->dev) != link ||
2357 2358
		    ((qc->flags & ATA_QCFLAG_QUIET) &&
		     qc->err_mask == AC_ERR_DEV))
T
Tejun Heo 已提交
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

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

	frozen = "";
2370
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
2371 2372
		frozen = " frozen";

2373 2374 2375 2376 2377
	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
2378
	if (ehc->i.dev) {
2379 2380 2381 2382
		ata_dev_err(ehc->i.dev, "exception Emask 0x%x "
			    "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
2383
		if (desc)
2384
			ata_dev_err(ehc->i.dev, "%s\n", desc);
T
Tejun Heo 已提交
2385
	} else {
2386 2387 2388 2389
		ata_link_err(link, "exception Emask 0x%x "
			     "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
2390
		if (desc)
2391
			ata_link_err(link, "%s\n", desc);
T
Tejun Heo 已提交
2392 2393
	}

2394
#ifdef CONFIG_ATA_VERBOSE_ERROR
2395
	if (ehc->i.serror)
2396
		ata_link_err(link,
2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413
		  "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 " : "",
2414
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "");
2415
#endif
2416

T
Tejun Heo 已提交
2417 2418
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
2419
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
2420 2421 2422
		const u8 *cdb = qc->cdb;
		char data_buf[20] = "";
		char cdb_buf[70] = "";
T
Tejun Heo 已提交
2423

2424
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
T
Tejun Heo 已提交
2425
		    ata_dev_phys_link(qc->dev) != link || !qc->err_mask)
T
Tejun Heo 已提交
2426 2427
			continue;

2428 2429 2430 2431 2432 2433 2434 2435 2436 2437
		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",
2438 2439
				[ATAPI_PROT_PIO]	= "pio",
				[ATAPI_PROT_DMA]	= "dma",
2440 2441 2442 2443 2444 2445 2446
			};

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

2447 2448 2449 2450 2451
		if (ata_is_atapi(qc->tf.protocol)) {
			if (qc->scsicmd)
				scsi_print_command(qc->scsicmd);
			else
				snprintf(cdb_buf, sizeof(cdb_buf),
2452 2453 2454 2455 2456 2457
				 "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]);
2458 2459 2460
		} else {
			const char *descr = ata_get_cmd_descript(cmd->command);
			if (descr)
2461 2462
				ata_dev_err(qc->dev, "failed command: %s\n",
					    descr);
2463
		}
2464

2465
		ata_dev_err(qc->dev,
2466
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
2467
			"tag %d%s\n         %s"
2468
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
2469
			"Emask 0x%x (%s)%s\n",
2470 2471 2472 2473
			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,
2474
			cmd->device, qc->tag, data_buf, cdb_buf,
2475 2476 2477 2478
			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
2479 2480
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
2481

2482
#ifdef CONFIG_ATA_VERBOSE_ERROR
2483
		if (res->command & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
2484
				    ATA_ERR)) {
2485
			if (res->command & ATA_BUSY)
2486
				ata_dev_err(qc->dev, "status: { Busy }\n");
2487
			else
2488
				ata_dev_err(qc->dev, "status: { %s%s%s%s}\n",
2489 2490 2491
				  res->command & ATA_DRDY ? "DRDY " : "",
				  res->command & ATA_DF ? "DF " : "",
				  res->command & ATA_DRQ ? "DRQ " : "",
2492
				  res->command & ATA_ERR ? "ERR " : "");
2493 2494 2495 2496 2497
		}

		if (cmd->command != ATA_CMD_PACKET &&
		    (res->feature & (ATA_ICRC | ATA_UNC | ATA_IDNF |
				     ATA_ABORTED)))
2498
			ata_dev_err(qc->dev, "error: { %s%s%s%s}\n",
2499 2500 2501
			  res->feature & ATA_ICRC ? "ICRC " : "",
			  res->feature & ATA_UNC ? "UNC " : "",
			  res->feature & ATA_IDNF ? "IDNF " : "",
2502
			  res->feature & ATA_ABORTED ? "ABRT " : "");
2503
#endif
T
Tejun Heo 已提交
2504 2505 2506
	}
}

2507 2508 2509 2510 2511 2512 2513 2514 2515
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2516
void ata_eh_report(struct ata_port *ap)
2517 2518 2519
{
	struct ata_link *link;

T
Tejun Heo 已提交
2520
	ata_for_each_link(link, ap, HOST_FIRST)
2521 2522 2523
		ata_eh_link_report(link);
}

T
Tejun Heo 已提交
2524
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
T
Tejun Heo 已提交
2525 2526
			unsigned int *classes, unsigned long deadline,
			bool clear_classes)
2527
{
2528
	struct ata_device *dev;
2529

T
Tejun Heo 已提交
2530
	if (clear_classes)
T
Tejun Heo 已提交
2531
		ata_for_each_dev(dev, link, ALL)
T
Tejun Heo 已提交
2532
			classes[dev->devno] = ATA_DEV_UNKNOWN;
2533

2534
	return reset(link, classes, deadline);
2535 2536
}

2537
static int ata_eh_followup_srst_needed(struct ata_link *link, int rc)
2538
{
2539
	if ((link->flags & ATA_LFLAG_NO_SRST) || ata_link_offline(link))
2540
		return 0;
2541 2542
	if (rc == -EAGAIN)
		return 1;
T
Tejun Heo 已提交
2543
	if (sata_pmp_supported(link->ap) && ata_is_host_link(link))
2544
		return 1;
2545 2546 2547
	return 0;
}

2548 2549 2550
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
2551
{
T
Tejun Heo 已提交
2552
	struct ata_port *ap = link->ap;
T
Tejun Heo 已提交
2553
	struct ata_link *slave = ap->slave_link;
2554
	struct ata_eh_context *ehc = &link->eh_context;
2555
	struct ata_eh_context *sehc = slave ? &slave->eh_context : NULL;
2556
	unsigned int *classes = ehc->classes;
2557
	unsigned int lflags = link->flags;
2558
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
2559
	int max_tries = 0, try = 0;
T
Tejun Heo 已提交
2560
	struct ata_link *failed_link;
2561
	struct ata_device *dev;
2562
	unsigned long deadline, now;
T
Tejun Heo 已提交
2563
	ata_reset_fn_t reset;
T
Tejun Heo 已提交
2564
	unsigned long flags;
2565
	u32 sstatus;
T
Tejun Heo 已提交
2566
	int nr_unknown, rc;
T
Tejun Heo 已提交
2567

2568 2569 2570
	/*
	 * Prepare to reset
	 */
2571 2572
	while (ata_eh_reset_timeouts[max_tries] != ULONG_MAX)
		max_tries++;
2573 2574 2575 2576
	if (link->flags & ATA_LFLAG_NO_HRST)
		hardreset = NULL;
	if (link->flags & ATA_LFLAG_NO_SRST)
		softreset = NULL;
2577

L
Lucas De Marchi 已提交
2578
	/* make sure each reset attempt is at least COOL_DOWN apart */
2579 2580 2581 2582 2583 2584 2585 2586
	if (ehc->i.flags & ATA_EHI_DID_RESET) {
		now = jiffies;
		WARN_ON(time_after(ehc->last_reset, now));
		deadline = ata_deadline(ehc->last_reset,
					ATA_EH_RESET_COOL_DOWN);
		if (time_before(now, deadline))
			schedule_timeout_uninterruptible(deadline - now);
	}
2587

T
Tejun Heo 已提交
2588 2589 2590 2591
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags |= ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

T
Tejun Heo 已提交
2592
	ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
2593

T
Tejun Heo 已提交
2594
	ata_for_each_dev(dev, link, ALL) {
T
Tejun Heo 已提交
2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612
		/* 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
2613
	/* prefer hardreset */
2614
	reset = NULL;
T
Tejun Heo 已提交
2615 2616 2617
	ehc->i.action &= ~ATA_EH_RESET;
	if (hardreset) {
		reset = hardreset;
2618
		ehc->i.action |= ATA_EH_HARDRESET;
2619
	} else if (softreset) {
T
Tejun Heo 已提交
2620
		reset = softreset;
2621
		ehc->i.action |= ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
2622
	}
2623 2624

	if (prereset) {
T
Tejun Heo 已提交
2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648
		unsigned long deadline = ata_deadline(jiffies,
						      ATA_EH_PRERESET_TIMEOUT);

		if (slave) {
			sehc->i.action &= ~ATA_EH_RESET;
			sehc->i.action |= ehc->i.action;
		}

		rc = prereset(link, deadline);

		/* If present, do prereset on slave link too.  Reset
		 * is skipped iff both master and slave links report
		 * -ENOENT or clear ATA_EH_RESET.
		 */
		if (slave && (rc == 0 || rc == -ENOENT)) {
			int tmp;

			tmp = prereset(slave, deadline);
			if (tmp != -ENOENT)
				rc = tmp;

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

2649
		if (rc) {
2650
			if (rc == -ENOENT) {
2651
				ata_link_dbg(link, "port disabled--ignoring\n");
T
Tejun Heo 已提交
2652
				ehc->i.action &= ~ATA_EH_RESET;
2653

T
Tejun Heo 已提交
2654
				ata_for_each_dev(dev, link, ALL)
2655
					classes[dev->devno] = ATA_DEV_NONE;
2656 2657

				rc = 0;
2658
			} else
2659 2660 2661
				ata_link_err(link,
					     "prereset failed (errno=%d)\n",
					     rc);
2662
			goto out;
2663 2664
		}

2665
		/* prereset() might have cleared ATA_EH_RESET.  If so,
2666
		 * bang classes, thaw and return.
2667 2668
		 */
		if (reset && !(ehc->i.action & ATA_EH_RESET)) {
T
Tejun Heo 已提交
2669
			ata_for_each_dev(dev, link, ALL)
2670
				classes[dev->devno] = ATA_DEV_NONE;
2671 2672 2673
			if ((ap->pflags & ATA_PFLAG_FROZEN) &&
			    ata_is_host_link(link))
				ata_eh_thaw_port(ap);
2674 2675 2676
			rc = 0;
			goto out;
		}
2677 2678
	}

T
Tejun Heo 已提交
2679
 retry:
2680 2681 2682
	/*
	 * Perform reset
	 */
2683 2684 2685
	if (ata_is_host_link(link))
		ata_eh_freeze_port(ap);

2686
	deadline = ata_deadline(jiffies, ata_eh_reset_timeouts[try++]);
T
Tejun Heo 已提交
2687

2688 2689
	if (reset) {
		if (verbose)
2690 2691
			ata_link_info(link, "%s resetting link\n",
				      reset == softreset ? "soft" : "hard");
2692 2693

		/* mark that this EH session started with reset */
2694
		ehc->last_reset = jiffies;
2695 2696 2697 2698
		if (reset == hardreset)
			ehc->i.flags |= ATA_EHI_DID_HARDRESET;
		else
			ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
2699

T
Tejun Heo 已提交
2700 2701 2702
		rc = ata_do_reset(link, reset, classes, deadline, true);
		if (rc && rc != -EAGAIN) {
			failed_link = link;
2703
			goto fail;
T
Tejun Heo 已提交
2704 2705 2706 2707 2708 2709 2710
		}

		/* hardreset slave link if existent */
		if (slave && reset == hardreset) {
			int tmp;

			if (verbose)
2711
				ata_link_info(slave, "hard resetting link\n");
T
Tejun Heo 已提交
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726

			ata_eh_about_to_do(slave, NULL, ATA_EH_RESET);
			tmp = ata_do_reset(slave, reset, classes, deadline,
					   false);
			switch (tmp) {
			case -EAGAIN:
				rc = -EAGAIN;
			case 0:
				break;
			default:
				failed_link = slave;
				rc = tmp;
				goto fail;
			}
		}
T
Tejun Heo 已提交
2727

T
Tejun Heo 已提交
2728
		/* perform follow-up SRST if necessary */
2729
		if (reset == hardreset &&
2730
		    ata_eh_followup_srst_needed(link, rc)) {
2731
			reset = softreset;
T
Tejun Heo 已提交
2732

2733
			if (!reset) {
2734 2735
				ata_link_err(link,
	     "follow-up softreset required but no softreset available\n");
T
Tejun Heo 已提交
2736
				failed_link = link;
2737 2738 2739
				rc = -EINVAL;
				goto fail;
			}
2740

2741
			ata_eh_about_to_do(link, NULL, ATA_EH_RESET);
T
Tejun Heo 已提交
2742
			rc = ata_do_reset(link, reset, classes, deadline, true);
2743 2744 2745 2746
			if (rc) {
				failed_link = link;
				goto fail;
			}
2747
		}
2748 2749
	} else {
		if (verbose)
2750 2751
			ata_link_info(link,
	"no reset method available, skipping reset\n");
2752 2753
		if (!(lflags & ATA_LFLAG_ASSUME_CLASS))
			lflags |= ATA_LFLAG_ASSUME_ATA;
2754 2755
	}

2756 2757 2758
	/*
	 * Post-reset processing
	 */
T
Tejun Heo 已提交
2759
	ata_for_each_dev(dev, link, ALL) {
2760 2761 2762 2763 2764 2765
		/* 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
2766

2767 2768 2769 2770 2771 2772 2773 2774
		if (ata_phys_link_offline(ata_dev_phys_link(dev)))
			continue;

		/* apply class override */
		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;
T
Tejun Heo 已提交
2775 2776
	}

2777 2778 2779
	/* record current link speed */
	if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
		link->sata_spd = (sstatus >> 4) & 0xf;
T
Tejun Heo 已提交
2780 2781
	if (slave && sata_scr_read(slave, SCR_STATUS, &sstatus) == 0)
		slave->sata_spd = (sstatus >> 4) & 0xf;
2782

2783 2784 2785 2786
	/* thaw the port */
	if (ata_is_host_link(link))
		ata_eh_thaw_port(ap);

2787 2788 2789 2790 2791 2792 2793
	/* 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.
	 */
T
Tejun Heo 已提交
2794
	if (postreset) {
2795
		postreset(link, classes);
T
Tejun Heo 已提交
2796 2797 2798
		if (slave)
			postreset(slave, classes);
	}
2799

2800
	/*
2801 2802 2803 2804 2805
	 * Some controllers can't be frozen very well and may set spurious
	 * error conditions during reset.  Clear accumulated error
	 * information and re-thaw the port if frozen.  As reset is the
	 * final recovery action and we cross check link onlineness against
	 * device classification later, no hotplug event is lost by this.
2806
	 */
2807
	spin_lock_irqsave(link->ap->lock, flags);
2808
	memset(&link->eh_info, 0, sizeof(link->eh_info));
T
Tejun Heo 已提交
2809
	if (slave)
2810 2811
		memset(&slave->eh_info, 0, sizeof(link->eh_info));
	ap->pflags &= ~ATA_PFLAG_EH_PENDING;
2812 2813
	spin_unlock_irqrestore(link->ap->lock, flags);

2814 2815 2816
	if (ap->pflags & ATA_PFLAG_FROZEN)
		ata_eh_thaw_port(ap);

2817 2818
	/*
	 * Make sure onlineness and classification result correspond.
2819 2820 2821
	 * 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
2822 2823
	 * link on/offlineness and classification result, those
	 * conditions can be reliably detected and retried.
2824
	 */
T
Tejun Heo 已提交
2825
	nr_unknown = 0;
T
Tejun Heo 已提交
2826
	ata_for_each_dev(dev, link, ALL) {
2827 2828
		if (ata_phys_link_online(ata_dev_phys_link(dev))) {
			if (classes[dev->devno] == ATA_DEV_UNKNOWN) {
2829
				ata_dev_dbg(dev, "link online but device misclassified\n");
2830
				classes[dev->devno] = ATA_DEV_NONE;
T
Tejun Heo 已提交
2831
				nr_unknown++;
2832 2833 2834
			}
		} else if (ata_phys_link_offline(ata_dev_phys_link(dev))) {
			if (ata_class_enabled(classes[dev->devno]))
2835 2836 2837
				ata_dev_dbg(dev,
					    "link offline, clearing class %d to NONE\n",
					    classes[dev->devno]);
2838 2839
			classes[dev->devno] = ATA_DEV_NONE;
		} else if (classes[dev->devno] == ATA_DEV_UNKNOWN) {
2840 2841
			ata_dev_dbg(dev,
				    "link status unknown, clearing UNKNOWN to NONE\n");
2842
			classes[dev->devno] = ATA_DEV_NONE;
T
Tejun Heo 已提交
2843
		}
2844 2845
	}

T
Tejun Heo 已提交
2846
	if (classify && nr_unknown) {
2847
		if (try < max_tries) {
2848 2849 2850
			ata_link_warn(link,
				      "link online but %d devices misclassified, retrying\n",
				      nr_unknown);
T
Tejun Heo 已提交
2851
			failed_link = link;
2852 2853 2854
			rc = -EAGAIN;
			goto fail;
		}
2855 2856 2857
		ata_link_warn(link,
			      "link online but %d devices misclassified, "
			      "device detection might fail\n", nr_unknown);
2858 2859
	}

2860
	/* reset successful, schedule revalidation */
T
Tejun Heo 已提交
2861
	ata_eh_done(link, NULL, ATA_EH_RESET);
T
Tejun Heo 已提交
2862 2863
	if (slave)
		ata_eh_done(slave, NULL, ATA_EH_RESET);
2864
	ehc->last_reset = jiffies;		/* update to completion time */
2865
	ehc->i.action |= ATA_EH_REVALIDATE;
2866
	link->lpm_policy = ATA_LPM_UNKNOWN;	/* reset LPM state */
2867

2868
	rc = 0;
2869 2870 2871
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2872 2873
	if (slave)
		sehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2874 2875 2876 2877 2878

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

T
Tejun Heo 已提交
2879
	return rc;
2880 2881

 fail:
2882 2883 2884 2885 2886
	/* 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;

2887
	if (try >= max_tries) {
2888 2889 2890 2891 2892 2893 2894 2895
		/*
		 * Thaw host port even if reset failed, so that the port
		 * can be retried on the next phy event.  This risks
		 * repeated EH runs but seems to be a better tradeoff than
		 * shutting down a port after a botched hotplug attempt.
		 */
		if (ata_is_host_link(link))
			ata_eh_thaw_port(ap);
2896
		goto out;
2897
	}
2898 2899 2900 2901 2902

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

2903
		ata_link_warn(failed_link,
2904 2905
			"reset failed (errno=%d), retrying in %u secs\n",
			rc, DIV_ROUND_UP(jiffies_to_msecs(delta), 1000));
2906

2907
		ata_eh_release(ap);
2908 2909
		while (delta)
			delta = schedule_timeout_uninterruptible(delta);
2910
		ata_eh_acquire(ap);
2911 2912
	}

2913 2914 2915 2916 2917 2918 2919 2920 2921 2922
	/*
	 * While disks spinup behind PMP, some controllers fail sending SRST.
	 * They need to be reset - as well as the PMP - before retrying.
	 */
	if (rc == -ERESTART) {
		if (ata_is_host_link(link))
			ata_eh_thaw_port(ap);
		goto out;
	}

T
Tejun Heo 已提交
2923
	if (try == max_tries - 1) {
2924
		sata_down_spd_limit(link, 0);
T
Tejun Heo 已提交
2925
		if (slave)
2926
			sata_down_spd_limit(slave, 0);
T
Tejun Heo 已提交
2927
	} else if (rc == -EPIPE)
2928
		sata_down_spd_limit(failed_link, 0);
T
Tejun Heo 已提交
2929

2930 2931 2932
	if (hardreset)
		reset = hardreset;
	goto retry;
T
Tejun Heo 已提交
2933 2934
}

2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968
static inline void ata_eh_pull_park_action(struct ata_port *ap)
{
	struct ata_link *link;
	struct ata_device *dev;
	unsigned long flags;

	/*
	 * This function can be thought of as an extended version of
	 * ata_eh_about_to_do() specially crafted to accommodate the
	 * requirements of ATA_EH_PARK handling. Since the EH thread
	 * does not leave the do {} while () loop in ata_eh_recover as
	 * long as the timeout for a park request to *one* device on
	 * the port has not expired, and since we still want to pick
	 * up park requests to other devices on the same port or
	 * timeout updates for the same device, we have to pull
	 * ATA_EH_PARK actions from eh_info into eh_context.i
	 * ourselves at the beginning of each pass over the loop.
	 *
	 * Additionally, all write accesses to &ap->park_req_pending
	 * through INIT_COMPLETION() (see below) or complete_all()
	 * (see ata_scsi_park_store()) are protected by the host lock.
	 * As a result we have that park_req_pending.done is zero on
	 * exit from this function, i.e. when ATA_EH_PARK actions for
	 * *all* devices on port ap have been pulled into the
	 * respective eh_context structs. If, and only if,
	 * park_req_pending.done is non-zero by the time we reach
	 * wait_for_completion_timeout(), another ATA_EH_PARK action
	 * has been scheduled for at least one of the devices on port
	 * ap and we have to cycle over the do {} while () loop in
	 * ata_eh_recover() again.
	 */

	spin_lock_irqsave(ap->lock, flags);
	INIT_COMPLETION(ap->park_req_pending);
T
Tejun Heo 已提交
2969 2970
	ata_for_each_link(link, ap, EDGE) {
		ata_for_each_dev(dev, link, ALL) {
2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003
			struct ata_eh_info *ehi = &link->eh_info;

			link->eh_context.i.dev_action[dev->devno] |=
				ehi->dev_action[dev->devno] & ATA_EH_PARK;
			ata_eh_clear_action(link, dev, ehi, ATA_EH_PARK);
		}
	}
	spin_unlock_irqrestore(ap->lock, flags);
}

static void ata_eh_park_issue_cmd(struct ata_device *dev, int park)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
	struct ata_taskfile tf;
	unsigned int err_mask;

	ata_tf_init(dev, &tf);
	if (park) {
		ehc->unloaded_mask |= 1 << dev->devno;
		tf.command = ATA_CMD_IDLEIMMEDIATE;
		tf.feature = 0x44;
		tf.lbal = 0x4c;
		tf.lbam = 0x4e;
		tf.lbah = 0x55;
	} else {
		ehc->unloaded_mask &= ~(1 << dev->devno);
		tf.command = ATA_CMD_CHK_POWER;
	}

	tf.flags |= ATA_TFLAG_DEVICE | ATA_TFLAG_ISADDR;
	tf.protocol |= ATA_PROT_NODATA;
	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_NONE, NULL, 0, 0);
	if (park && (err_mask || tf.lbal != 0xc4)) {
3004
		ata_dev_err(dev, "head unload failed!\n");
3005 3006 3007 3008
		ehc->unloaded_mask &= ~(1 << dev->devno);
	}
}

3009
static int ata_eh_revalidate_and_attach(struct ata_link *link,
T
Tejun Heo 已提交
3010
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
3011
{
3012 3013
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
3014
	struct ata_device *dev;
3015
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
3016
	unsigned long flags;
3017
	int rc = 0;
T
Tejun Heo 已提交
3018 3019 3020

	DPRINTK("ENTER\n");

3021 3022 3023 3024
	/* 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.
	 */
T
Tejun Heo 已提交
3025
	ata_for_each_dev(dev, link, ALL_REVERSE) {
3026 3027
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
3028

3029 3030 3031
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

3032
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
3033 3034
			WARN_ON(dev->class == ATA_DEV_PMP);

T
Tejun Heo 已提交
3035
			if (ata_phys_link_offline(ata_dev_phys_link(dev))) {
T
Tejun Heo 已提交
3036
				rc = -EIO;
3037
				goto err;
T
Tejun Heo 已提交
3038 3039
			}

3040
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
3041 3042
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
T
Tejun Heo 已提交
3043
			if (rc)
3044
				goto err;
T
Tejun Heo 已提交
3045

3046
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
3047

3048 3049 3050 3051 3052
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

3053
			/* schedule the scsi_rescan_device() here */
3054
			schedule_work(&(ap->scsi_rescan_task));
T
Tejun Heo 已提交
3055 3056 3057
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
T
Tejun Heo 已提交
3058 3059 3060 3061 3062 3063
			/* Temporarily set dev->class, it will be
			 * permanently set once all configurations are
			 * complete.  This is necessary because new
			 * device configuration is done in two
			 * separate loops.
			 */
T
Tejun Heo 已提交
3064 3065
			dev->class = ehc->classes[dev->devno];

3066 3067 3068 3069 3070
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
T
Tejun Heo 已提交
3071 3072 3073 3074 3075

			/* read_id might have changed class, store and reset */
			ehc->classes[dev->devno] = dev->class;
			dev->class = ATA_DEV_UNKNOWN;

3076 3077
			switch (rc) {
			case 0:
3078 3079
				/* clear error info accumulated during probe */
				ata_ering_clear(&dev->ering);
3080
				new_mask |= 1 << dev->devno;
3081 3082
				break;
			case -ENOENT:
3083 3084
				/* IDENTIFY was issued to non-existent
				 * device.  No need to reset.  Just
T
Tejun Heo 已提交
3085
				 * thaw and ignore the device.
3086 3087
				 */
				ata_eh_thaw_port(ap);
T
Tejun Heo 已提交
3088
				break;
3089 3090
			default:
				goto err;
T
Tejun Heo 已提交
3091
			}
3092 3093
		}
	}
T
Tejun Heo 已提交
3094

3095
	/* PDIAG- should have been released, ask cable type if post-reset */
3096 3097 3098 3099 3100
	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);
	}
3101

3102 3103 3104
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
T
Tejun Heo 已提交
3105
	ata_for_each_dev(dev, link, ALL) {
T
Tejun Heo 已提交
3106
		if (!(new_mask & (1 << dev->devno)))
3107 3108
			continue;

T
Tejun Heo 已提交
3109 3110
		dev->class = ehc->classes[dev->devno];

T
Tejun Heo 已提交
3111 3112 3113
		if (dev->class == ATA_DEV_PMP)
			continue;

3114 3115 3116
		ehc->i.flags |= ATA_EHI_PRINTINFO;
		rc = ata_dev_configure(dev);
		ehc->i.flags &= ~ATA_EHI_PRINTINFO;
T
Tejun Heo 已提交
3117 3118
		if (rc) {
			dev->class = ATA_DEV_UNKNOWN;
3119
			goto err;
T
Tejun Heo 已提交
3120
		}
3121 3122 3123 3124 3125 3126 3127

		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
3128 3129
	}

3130
	return 0;
T
Tejun Heo 已提交
3131

3132 3133 3134
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
3135 3136 3137
	return rc;
}

3138 3139 3140
/**
 *	ata_set_mode - Program timings and issue SET FEATURES - XFER
 *	@link: link on which timings will be programmed
3141
 *	@r_failed_dev: out parameter for failed device
3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155
 *
 *	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;
3156 3157
	struct ata_device *dev;
	int rc;
3158

3159
	/* if data transfer is verified, clear DUBIOUS_XFER on ering top */
T
Tejun Heo 已提交
3160
	ata_for_each_dev(dev, link, ENABLED) {
3161 3162 3163 3164 3165 3166 3167 3168 3169
		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;
		}
	}

3170 3171
	/* has private set_mode? */
	if (ap->ops->set_mode)
3172 3173 3174 3175 3176
		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 */
T
Tejun Heo 已提交
3177
	ata_for_each_dev(dev, link, ENABLED) {
3178 3179 3180 3181 3182 3183 3184 3185 3186 3187
		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;
3188 3189
}

3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
/**
 *	atapi_eh_clear_ua - Clear ATAPI UNIT ATTENTION after reset
 *	@dev: ATAPI device to clear UA for
 *
 *	Resets and other operations can make an ATAPI device raise
 *	UNIT ATTENTION which causes the next operation to fail.  This
 *	function clears UA.
 *
 *	LOCKING:
 *	EH context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
static int atapi_eh_clear_ua(struct ata_device *dev)
{
	int i;

	for (i = 0; i < ATA_EH_UA_TRIES; i++) {
3209
		u8 *sense_buffer = dev->link->ap->sector_buf;
3210 3211 3212 3213 3214
		u8 sense_key = 0;
		unsigned int err_mask;

		err_mask = atapi_eh_tur(dev, &sense_key);
		if (err_mask != 0 && err_mask != AC_ERR_DEV) {
3215 3216 3217
			ata_dev_warn(dev,
				     "TEST_UNIT_READY failed (err_mask=0x%x)\n",
				     err_mask);
3218 3219 3220 3221 3222 3223 3224 3225
			return -EIO;
		}

		if (!err_mask || sense_key != UNIT_ATTENTION)
			return 0;

		err_mask = atapi_eh_request_sense(dev, sense_buffer, sense_key);
		if (err_mask) {
3226
			ata_dev_warn(dev, "failed to clear "
3227 3228 3229 3230 3231
				"UNIT ATTENTION (err_mask=0x%x)\n", err_mask);
			return -EIO;
		}
	}

3232 3233
	ata_dev_warn(dev, "UNIT ATTENTION persists after %d tries\n",
		     ATA_EH_UA_TRIES);
3234 3235 3236 3237

	return 0;
}

3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
/**
 *	ata_eh_maybe_retry_flush - Retry FLUSH if necessary
 *	@dev: ATA device which may need FLUSH retry
 *
 *	If @dev failed FLUSH, it needs to be reported upper layer
 *	immediately as it means that @dev failed to remap and already
 *	lost at least a sector and further FLUSH retrials won't make
 *	any difference to the lost sector.  However, if FLUSH failed
 *	for other reasons, for example transmission error, FLUSH needs
 *	to be retried.
 *
 *	This function determines whether FLUSH failure retry is
 *	necessary and performs it if so.
 *
 *	RETURNS:
 *	0 if EH can continue, -errno if EH needs to be repeated.
 */
static int ata_eh_maybe_retry_flush(struct ata_device *dev)
{
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	unsigned int err_mask;
	int rc = 0;

	/* did flush fail for this device? */
	if (!ata_tag_valid(link->active_tag))
		return 0;

	qc = __ata_qc_from_tag(ap, link->active_tag);
	if (qc->dev != dev || (qc->tf.command != ATA_CMD_FLUSH_EXT &&
			       qc->tf.command != ATA_CMD_FLUSH))
		return 0;

	/* if the device failed it, it should be reported to upper layers */
	if (qc->err_mask & AC_ERR_DEV)
		return 0;

	/* flush failed for some other reason, give it another shot */
	ata_tf_init(dev, &tf);

	tf.command = qc->tf.command;
	tf.flags |= ATA_TFLAG_DEVICE;
	tf.protocol = ATA_PROT_NODATA;

3284
	ata_dev_warn(dev, "retrying FLUSH 0x%x Emask 0x%x\n",
3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298
		       tf.command, qc->err_mask);

	err_mask = ata_exec_internal(dev, &tf, NULL, DMA_NONE, NULL, 0, 0);
	if (!err_mask) {
		/*
		 * FLUSH is complete but there's no way to
		 * successfully complete a failed command from EH.
		 * Making sure retry is allowed at least once and
		 * retrying it should do the trick - whatever was in
		 * the cache is already on the platter and this won't
		 * cause infinite loop.
		 */
		qc->scsicmd->allowed = max(qc->scsicmd->allowed, 1);
	} else {
3299
		ata_dev_warn(dev, "FLUSH failed Emask 0x%x\n",
3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313
			       err_mask);
		rc = -EIO;

		/* if device failed it, report it to upper layers */
		if (err_mask & AC_ERR_DEV) {
			qc->err_mask |= AC_ERR_DEV;
			qc->result_tf = tf;
			if (!(ap->pflags & ATA_PFLAG_FROZEN))
				rc = 0;
		}
	}
	return rc;
}

3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
/**
 *	ata_eh_set_lpm - configure SATA interface power management
 *	@link: link to configure power management
 *	@policy: the link power management policy
 *	@r_failed_dev: out parameter for failed device
 *
 *	Enable SATA Interface power management.  This will enable
 *	Device Interface Power Management (DIPM) for min_power
 * 	policy, and then call driver specific callbacks for
 *	enabling Host Initiated Power management.
 *
 *	LOCKING:
 *	EH context.
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
static int ata_eh_set_lpm(struct ata_link *link, enum ata_lpm_policy policy,
			  struct ata_device **r_failed_dev)
{
3334
	struct ata_port *ap = ata_is_host_link(link) ? link->ap : NULL;
3335 3336
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev, *link_dev = NULL, *lpm_dev = NULL;
3337
	enum ata_lpm_policy old_policy = link->lpm_policy;
3338
	bool no_dipm = link->ap->flags & ATA_FLAG_NO_DIPM;
3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354
	unsigned int hints = ATA_LPM_EMPTY | ATA_LPM_HIPM;
	unsigned int err_mask;
	int rc;

	/* if the link or host doesn't do LPM, noop */
	if ((link->flags & ATA_LFLAG_NO_LPM) || (ap && !ap->ops->set_lpm))
		return 0;

	/*
	 * DIPM is enabled only for MIN_POWER as some devices
	 * misbehave when the host NACKs transition to SLUMBER.  Order
	 * device and link configurations such that the host always
	 * allows DIPM requests.
	 */
	ata_for_each_dev(dev, link, ENABLED) {
		bool hipm = ata_id_has_hipm(dev->id);
3355
		bool dipm = ata_id_has_dipm(dev->id) && !no_dipm;
3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372

		/* find the first enabled and LPM enabled devices */
		if (!link_dev)
			link_dev = dev;

		if (!lpm_dev && (hipm || dipm))
			lpm_dev = dev;

		hints &= ~ATA_LPM_EMPTY;
		if (!hipm)
			hints &= ~ATA_LPM_HIPM;

		/* disable DIPM before changing link config */
		if (policy != ATA_LPM_MIN_POWER && dipm) {
			err_mask = ata_dev_set_feature(dev,
					SETFEATURES_SATA_DISABLE, SATA_DIPM);
			if (err_mask && err_mask != AC_ERR_DEV) {
3373 3374 3375
				ata_dev_warn(dev,
					     "failed to disable DIPM, Emask 0x%x\n",
					     err_mask);
3376 3377 3378 3379 3380 3381
				rc = -EIO;
				goto fail;
			}
		}
	}

3382 3383 3384 3385 3386 3387
	if (ap) {
		rc = ap->ops->set_lpm(link, policy, hints);
		if (!rc && ap->slave_link)
			rc = ap->ops->set_lpm(ap->slave_link, policy, hints);
	} else
		rc = sata_pmp_set_lpm(link, policy, hints);
3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401

	/*
	 * Attribute link config failure to the first (LPM) enabled
	 * device on the link.
	 */
	if (rc) {
		if (rc == -EOPNOTSUPP) {
			link->flags |= ATA_LFLAG_NO_LPM;
			return 0;
		}
		dev = lpm_dev ? lpm_dev : link_dev;
		goto fail;
	}

3402 3403 3404 3405 3406 3407 3408 3409
	/*
	 * Low level driver acked the transition.  Issue DIPM command
	 * with the new policy set.
	 */
	link->lpm_policy = policy;
	if (ap && ap->slave_link)
		ap->slave_link->lpm_policy = policy;

3410 3411
	/* host config updated, enable DIPM if transitioning to MIN_POWER */
	ata_for_each_dev(dev, link, ENABLED) {
3412 3413
		if (policy == ATA_LPM_MIN_POWER && !no_dipm &&
		    ata_id_has_dipm(dev->id)) {
3414 3415 3416
			err_mask = ata_dev_set_feature(dev,
					SETFEATURES_SATA_ENABLE, SATA_DIPM);
			if (err_mask && err_mask != AC_ERR_DEV) {
3417
				ata_dev_warn(dev,
3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428
					"failed to enable DIPM, Emask 0x%x\n",
					err_mask);
				rc = -EIO;
				goto fail;
			}
		}
	}

	return 0;

fail:
3429 3430 3431 3432 3433
	/* restore the old policy */
	link->lpm_policy = old_policy;
	if (ap && ap->slave_link)
		ap->slave_link->lpm_policy = old_policy;

3434 3435
	/* if no device or only one more chance is left, disable LPM */
	if (!dev || ehc->tries[dev->devno] <= 2) {
3436
		ata_link_warn(link, "disabling LPM on the link\n");
3437 3438 3439 3440 3441 3442 3443
		link->flags |= ATA_LFLAG_NO_LPM;
	}
	if (r_failed_dev)
		*r_failed_dev = dev;
	return rc;
}

3444
int ata_link_nr_enabled(struct ata_link *link)
T
Tejun Heo 已提交
3445
{
3446 3447
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
3448

T
Tejun Heo 已提交
3449 3450
	ata_for_each_dev(dev, link, ENABLED)
		cnt++;
T
Tejun Heo 已提交
3451 3452 3453
	return cnt;
}

3454
static int ata_link_nr_vacant(struct ata_link *link)
T
Tejun Heo 已提交
3455
{
3456 3457
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
3458

T
Tejun Heo 已提交
3459
	ata_for_each_dev(dev, link, ALL)
3460
		if (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
3461 3462 3463 3464
			cnt++;
	return cnt;
}

3465
static int ata_eh_skip_recovery(struct ata_link *link)
T
Tejun Heo 已提交
3466
{
T
Tejun Heo 已提交
3467
	struct ata_port *ap = link->ap;
3468
	struct ata_eh_context *ehc = &link->eh_context;
3469
	struct ata_device *dev;
T
Tejun Heo 已提交
3470

3471 3472 3473 3474
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

3475 3476 3477 3478
	/* skip if explicitly requested */
	if (ehc->i.flags & ATA_EHI_NO_RECOVERY)
		return 1;

T
Tejun Heo 已提交
3479 3480 3481 3482 3483 3484 3485
	/* 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
3486 3487 3488
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
T
Tejun Heo 已提交
3489
	ata_for_each_dev(dev, link, ALL) {
T
Tejun Heo 已提交
3490 3491 3492 3493 3494 3495 3496 3497
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510
static int ata_count_probe_trials_cb(struct ata_ering_entry *ent, void *void_arg)
{
	u64 interval = msecs_to_jiffies(ATA_EH_PROBE_TRIAL_INTERVAL);
	u64 now = get_jiffies_64();
	int *trials = void_arg;

	if (ent->timestamp < now - min(now, interval))
		return -1;

	(*trials)++;
	return 0;
}

3511 3512 3513
static int ata_eh_schedule_probe(struct ata_device *dev)
{
	struct ata_eh_context *ehc = &dev->link->eh_context;
3514 3515
	struct ata_link *link = ata_dev_phys_link(dev);
	int trials = 0;
3516 3517 3518 3519 3520 3521 3522 3523

	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
3524
	ehc->i.action |= ATA_EH_RESET;
3525 3526
	ehc->saved_xfer_mode[dev->devno] = 0;
	ehc->saved_ncq_enabled &= ~(1 << dev->devno);
3527

3528
	/* the link maybe in a deep sleep, wake it up */
3529 3530 3531 3532 3533 3534 3535 3536
	if (link->lpm_policy > ATA_LPM_MAX_POWER) {
		if (ata_is_host_link(link))
			link->ap->ops->set_lpm(link, ATA_LPM_MAX_POWER,
					       ATA_LPM_EMPTY);
		else
			sata_pmp_set_lpm(link, ATA_LPM_MAX_POWER,
					 ATA_LPM_EMPTY);
	}
3537

3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556
	/* Record and count probe trials on the ering.  The specific
	 * error mask used is irrelevant.  Because a successful device
	 * detection clears the ering, this count accumulates only if
	 * there are consecutive failed probes.
	 *
	 * If the count is equal to or higher than ATA_EH_PROBE_TRIALS
	 * in the last ATA_EH_PROBE_TRIAL_INTERVAL, link speed is
	 * forced to 1.5Gbps.
	 *
	 * This is to work around cases where failed link speed
	 * negotiation results in device misdetection leading to
	 * infinite DEVXCHG or PHRDY CHG events.
	 */
	ata_ering_record(&dev->ering, 0, AC_ERR_OTHER);
	ata_ering_map(&dev->ering, ata_count_probe_trials_cb, &trials);

	if (trials > ATA_EH_PROBE_TRIALS)
		sata_down_spd_limit(link, 1);

3557 3558 3559
	return 1;
}

3560
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
3561
{
T
Tejun Heo 已提交
3562
	struct ata_eh_context *ehc = &dev->link->eh_context;
3563

3564 3565 3566 3567 3568
	/* -EAGAIN from EH routine indicates retry without prejudice.
	 * The requester is responsible for ensuring forward progress.
	 */
	if (err != -EAGAIN)
		ehc->tries[dev->devno]--;
3569 3570 3571 3572 3573 3574 3575 3576 3577

	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:
3578
		if (ehc->tries[dev->devno] == 1) {
3579 3580 3581
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
3582
			sata_down_spd_limit(ata_dev_phys_link(dev), 0);
3583 3584
			if (dev->pio_mode > XFER_PIO_0)
				ata_down_xfermask_limit(dev, ATA_DNXFER_PIO);
3585 3586 3587 3588 3589 3590 3591 3592
		}
	}

	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 */
T
Tejun Heo 已提交
3593
		if (ata_phys_link_offline(ata_dev_phys_link(dev)))
3594 3595
			ata_eh_detach_dev(dev);

3596
		/* schedule probe if necessary */
3597
		if (ata_eh_schedule_probe(dev)) {
3598
			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
3599 3600 3601
			memset(ehc->cmd_timeout_idx[dev->devno], 0,
			       sizeof(ehc->cmd_timeout_idx[dev->devno]));
		}
3602 3603

		return 1;
3604
	} else {
T
Tejun Heo 已提交
3605
		ehc->i.action |= ATA_EH_RESET;
3606
		return 0;
3607 3608 3609
	}
}

T
Tejun Heo 已提交
3610 3611 3612
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
3613
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
3614 3615 3616
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
3617
 *	@r_failed_link: out parameter for failed link
T
Tejun Heo 已提交
3618 3619 3620
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
3621 3622 3623
 *	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
3624
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
3625 3626 3627 3628 3629 3630 3631
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
3632 3633 3634 3635
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
3636
{
3637
	struct ata_link *link;
T
Tejun Heo 已提交
3638
	struct ata_device *dev;
3639
	int rc, nr_fails;
3640
	unsigned long flags, deadline;
T
Tejun Heo 已提交
3641 3642 3643 3644

	DPRINTK("ENTER\n");

	/* prep for recovery */
T
Tejun Heo 已提交
3645
	ata_for_each_link(link, ap, EDGE) {
3646
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
3647

3648 3649 3650 3651 3652 3653 3654 3655 3656
		/* 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);
		}

T
Tejun Heo 已提交
3657
		ata_for_each_dev(dev, link, ALL) {
3658 3659 3660 3661
			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
3662

3663 3664 3665 3666 3667 3668 3669 3670 3671
			/* 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);

3672 3673 3674
			/* schedule probe if necessary */
			if (!ata_dev_enabled(dev))
				ata_eh_schedule_probe(dev);
T
Tejun Heo 已提交
3675
		}
T
Tejun Heo 已提交
3676 3677 3678 3679 3680
	}

 retry:
	rc = 0;

3681
	/* if UNLOADING, finish immediately */
3682
	if (ap->pflags & ATA_PFLAG_UNLOADING)
3683 3684
		goto out;

3685
	/* prep for EH */
T
Tejun Heo 已提交
3686
	ata_for_each_link(link, ap, EDGE) {
3687
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
3688

3689 3690 3691 3692
		/* skip EH if possible. */
		if (ata_eh_skip_recovery(link))
			ehc->i.action = 0;

T
Tejun Heo 已提交
3693
		ata_for_each_dev(dev, link, ALL)
3694 3695
			ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
	}
T
Tejun Heo 已提交
3696

T
Tejun Heo 已提交
3697
	/* reset */
T
Tejun Heo 已提交
3698
	ata_for_each_link(link, ap, EDGE) {
3699
		struct ata_eh_context *ehc = &link->eh_context;
3700

3701 3702
		if (!(ehc->i.action & ATA_EH_RESET))
			continue;
3703

3704 3705 3706
		rc = ata_eh_reset(link, ata_link_nr_vacant(link),
				  prereset, softreset, hardreset, postreset);
		if (rc) {
3707
			ata_link_err(link, "reset failed, giving up\n");
3708
			goto out;
T
Tejun Heo 已提交
3709 3710 3711
		}
	}

3712 3713 3714 3715 3716 3717 3718 3719 3720 3721
	do {
		unsigned long now;

		/*
		 * clears ATA_EH_PARK in eh_info and resets
		 * ap->park_req_pending
		 */
		ata_eh_pull_park_action(ap);

		deadline = jiffies;
T
Tejun Heo 已提交
3722 3723
		ata_for_each_link(link, ap, EDGE) {
			ata_for_each_dev(dev, link, ALL) {
3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747
				struct ata_eh_context *ehc = &link->eh_context;
				unsigned long tmp;

				if (dev->class != ATA_DEV_ATA)
					continue;
				if (!(ehc->i.dev_action[dev->devno] &
				      ATA_EH_PARK))
					continue;
				tmp = dev->unpark_deadline;
				if (time_before(deadline, tmp))
					deadline = tmp;
				else if (time_before_eq(tmp, jiffies))
					continue;
				if (ehc->unloaded_mask & (1 << dev->devno))
					continue;

				ata_eh_park_issue_cmd(dev, 1);
			}
		}

		now = jiffies;
		if (time_before_eq(deadline, now))
			break;

3748
		ata_eh_release(ap);
3749 3750
		deadline = wait_for_completion_timeout(&ap->park_req_pending,
						       deadline - now);
3751
		ata_eh_acquire(ap);
3752
	} while (deadline);
T
Tejun Heo 已提交
3753 3754
	ata_for_each_link(link, ap, EDGE) {
		ata_for_each_dev(dev, link, ALL) {
3755 3756 3757 3758 3759 3760 3761 3762 3763
			if (!(link->eh_context.unloaded_mask &
			      (1 << dev->devno)))
				continue;

			ata_eh_park_issue_cmd(dev, 0);
			ata_eh_done(link, dev, ATA_EH_PARK);
		}
	}

3764
	/* the rest */
3765 3766
	nr_fails = 0;
	ata_for_each_link(link, ap, PMP_FIRST) {
3767
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
3768

3769 3770 3771
		if (sata_pmp_attached(ap) && ata_is_host_link(link))
			goto config_lpm;

3772 3773
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
3774
		if (rc)
3775
			goto rest_fail;
T
Tejun Heo 已提交
3776

3777 3778 3779 3780 3781 3782
		/* 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;
		}

3783 3784 3785 3786
		/* configure transfer mode if necessary */
		if (ehc->i.flags & ATA_EHI_SETMODE) {
			rc = ata_set_mode(link, &dev);
			if (rc)
3787
				goto rest_fail;
3788 3789 3790
			ehc->i.flags &= ~ATA_EHI_SETMODE;
		}

3791 3792 3793 3794
		/* If reset has been issued, clear UA to avoid
		 * disrupting the current users of the device.
		 */
		if (ehc->i.flags & ATA_EHI_DID_RESET) {
T
Tejun Heo 已提交
3795
			ata_for_each_dev(dev, link, ALL) {
3796 3797 3798 3799
				if (dev->class != ATA_DEV_ATAPI)
					continue;
				rc = atapi_eh_clear_ua(dev);
				if (rc)
3800
					goto rest_fail;
3801 3802 3803
			}
		}

3804 3805 3806 3807 3808 3809
		/* retry flush if necessary */
		ata_for_each_dev(dev, link, ALL) {
			if (dev->class != ATA_DEV_ATA)
				continue;
			rc = ata_eh_maybe_retry_flush(dev);
			if (rc)
3810
				goto rest_fail;
3811 3812
		}

3813
	config_lpm:
3814
		/* configure link power saving */
3815 3816 3817 3818 3819
		if (link->lpm_policy != ap->target_lpm_policy) {
			rc = ata_eh_set_lpm(link, ap->target_lpm_policy, &dev);
			if (rc)
				goto rest_fail;
		}
3820

3821 3822 3823
		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
T
Tejun Heo 已提交
3824

3825 3826 3827 3828
	rest_fail:
		nr_fails++;
		if (dev)
			ata_eh_handle_dev_fail(dev, rc);
T
Tejun Heo 已提交
3829

3830 3831 3832 3833
		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
3834
			if (sata_pmp_attached(ap))
3835
				goto out;
3836
			break;
3837
		}
T
Tejun Heo 已提交
3838 3839
	}

3840
	if (nr_fails)
3841
		goto retry;
T
Tejun Heo 已提交
3842

3843 3844 3845 3846
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

T
Tejun Heo 已提交
3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860
	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.
 */
3861
void ata_eh_finish(struct ata_port *ap)
T
Tejun Heo 已提交
3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876
{
	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
3877
			if (qc->flags & ATA_QCFLAG_RETRY)
T
Tejun Heo 已提交
3878
				ata_eh_qc_retry(qc);
T
Tejun Heo 已提交
3879 3880
			else
				ata_eh_qc_complete(qc);
T
Tejun Heo 已提交
3881 3882 3883 3884 3885 3886 3887 3888 3889 3890
		} 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);
			}
		}
	}
3891 3892 3893 3894

	/* 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
3895 3896 3897 3898 3899
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
3900
 *
3901
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
3902 3903 3904 3905 3906 3907 3908 3909 3910
 *	@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).
 */
3911 3912 3913
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
3914
{
3915 3916 3917 3918 3919 3920 3921 3922 3923
	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) {
T
Tejun Heo 已提交
3924
		ata_for_each_dev(dev, &ap->link, ALL)
3925 3926 3927
			ata_dev_disable(dev);
	}

T
Tejun Heo 已提交
3928 3929
	ata_eh_finish(ap);
}
3930

3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944
/**
 *	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;

3945
	/* ignore built-in hardreset if SCR access is not available */
T
Tejun Heo 已提交
3946
	if (hardreset == sata_std_hardreset && !sata_scr_valid(&ap->link))
3947 3948 3949 3950 3951
		hardreset = NULL;

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

3952
#ifdef CONFIG_PM
3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
/**
 *	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);

3978 3979 3980 3981 3982
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

3983 3984 3985 3986 3987 3988
	/* 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
3989
	ata_acpi_set_state(ap, PMSG_SUSPEND);
3990
 out:
3991 3992 3993 3994 3995 3996
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
3997
	else if (ap->pflags & ATA_PFLAG_FROZEN)
3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020
		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)
{
T
Tejun Heo 已提交
4021 4022
	struct ata_link *link;
	struct ata_device *dev;
4023
	unsigned long flags;
4024
	int rc = 0;
4025 4026 4027 4028 4029 4030 4031 4032 4033 4034

	/* 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);

4035
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
4036

T
Tejun Heo 已提交
4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047
	/*
	 * Error timestamps are in jiffies which doesn't run while
	 * suspended and PHY events during resume isn't too uncommon.
	 * When the two are combined, it can lead to unnecessary speed
	 * downs if the machine is suspended and resumed repeatedly.
	 * Clear error history.
	 */
	ata_for_each_link(link, ap, HOST_FIRST)
		ata_for_each_dev(dev, link, ALL)
			ata_ering_clear(&dev->ering);

S
Shaohua Li 已提交
4048 4049
	ata_acpi_set_state(ap, PMSG_ON);

4050 4051 4052
	if (ap->ops->port_resume)
		rc = ap->ops->port_resume(ap);

T
Tejun Heo 已提交
4053 4054 4055
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

4056
	/* report result */
4057 4058 4059 4060 4061 4062 4063 4064
	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);
}
4065
#endif /* CONFIG_PM */