libata-eh.c 68.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
/*
 *  libata-eh.c - libata error handling
 *
 *  Maintained by:  Jeff Garzik <jgarzik@pobox.com>
 *    		    Please ALWAYS copy linux-ide@vger.kernel.org
 *		    on emails.
 *
 *  Copyright 2006 Tejun Heo <htejun@gmail.com>
 *
 *
 *  This program is free software; you can redistribute it and/or
 *  modify it under the terms of the GNU General Public License as
 *  published by the Free Software Foundation; either version 2, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 *  General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; see the file COPYING.  If not, write to
 *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139,
 *  USA.
 *
 *
 *  libata documentation is available via 'make {ps|pdf}docs',
 *  as Documentation/DocBook/libata.*
 *
 *  Hardware documentation available from http://www.t13.org/ and
 *  http://www.sata-io.org/
 *
 */

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

#include <linux/libata.h>

#include "libata.h"

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

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

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

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

T
Tejun Heo 已提交
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
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;
}

153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
/**
 *	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
		ata_port_desc(ap, "%s 0x%llx", name, start + offset);
}

#endif /* CONFIG_PCI */

220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
static void ata_ering_record(struct ata_ering *ering, int is_io,
			     unsigned int err_mask)
{
	struct ata_ering_entry *ent;

	WARN_ON(!err_mask);

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

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

236
static void ata_ering_clear(struct ata_ering *ering)
237
{
238
	memset(ering, 0, sizeof(*ering));
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
}

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

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

	return rc;
}

262 263
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
264
	struct ata_eh_context *ehc = &dev->link->eh_context;
265 266 267 268

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

269
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
270 271
				struct ata_eh_info *ehi, unsigned int action)
{
272
	struct ata_device *tdev;
273 274 275

	if (!dev) {
		ehi->action &= ~action;
276 277
		ata_link_for_each_dev(tdev, link)
			ehi->dev_action[tdev->devno] &= ~action;
278 279 280 281 282 283
	} 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) {
284 285 286
			ata_link_for_each_dev(tdev, link)
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
287 288 289 290 291 292 293 294
			ehi->action &= ~action;
		}

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

295 296 297 298 299 300 301 302 303 304 305
/**
 *	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.
 *
306 307
 *	TODO: kill this function once old EH is gone.
 *
308 309 310 311 312 313 314 315 316
 *	LOCKING:
 *	Called from timer context
 *
 *	RETURNS:
 *	EH_HANDLED or EH_NOT_HANDLED
 */
enum scsi_eh_timer_return ata_scsi_timed_out(struct scsi_cmnd *cmd)
{
	struct Scsi_Host *host = cmd->device->host;
317
	struct ata_port *ap = ata_shost_to_port(host);
318 319
	unsigned long flags;
	struct ata_queued_cmd *qc;
320
	enum scsi_eh_timer_return ret;
321 322 323

	DPRINTK("ENTER\n");

324 325 326 327 328 329
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
330
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
331
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
332 333 334 335 336 337
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
338
	spin_unlock_irqrestore(ap->lock, flags);
339

340
 out:
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
	DPRINTK("EXIT, ret=%d\n", ret);
	return ret;
}

/**
 *	ata_scsi_error - SCSI layer error handler callback
 *	@host: SCSI host on which error occurred
 *
 *	Handles SCSI-layer-thrown error events.
 *
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 *
 *	RETURNS:
 *	Zero.
 */
357
void ata_scsi_error(struct Scsi_Host *host)
358
{
359
	struct ata_port *ap = ata_shost_to_port(host);
360
	int i;
361
	unsigned long flags;
362 363 364

	DPRINTK("ENTER\n");

365
	/* synchronize with port task */
366 367
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
368
	/* synchronize with host lock and sort out timeouts */
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385

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

386
		spin_lock_irqsave(ap->lock, flags);
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

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

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

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

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

425
		spin_unlock_irqrestore(ap->lock, flags);
426 427 428

		/* initialize eh_tries */
		ap->eh_tries = ATA_EH_MAX_TRIES;
429
	} else
430
		spin_unlock_wait(ap->lock);
431 432 433 434

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
435 436
		struct ata_link *link;

T
Tejun Heo 已提交
437 438 439
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

440 441 442
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

443
		/* fetch & clear EH info */
444
		spin_lock_irqsave(ap->lock, flags);
445

446 447 448 449 450
		__ata_port_for_each_link(link, ap) {
			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));
		}
451

452 453
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
454
		ap->excl_link = NULL;	/* don't maintain exclusion over EH */
455

456
		spin_unlock_irqrestore(ap->lock, flags);
457

458 459
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
460 461 462
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
463

464 465 466
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

467 468 469 470
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
471
		spin_lock_irqsave(ap->lock, flags);
472

473
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
474
			if (--ap->eh_tries) {
475
				spin_unlock_irqrestore(ap->lock, flags);
476 477 478
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
479
					"tries, giving up\n", ATA_EH_MAX_TRIES);
T
Tejun Heo 已提交
480
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
481 482
		}

483
		/* this run is complete, make sure EH info is clear */
484 485
		__ata_port_for_each_link(link, ap)
			memset(&link->eh_info, 0, sizeof(link->eh_info));
486

487
		/* Clear host_eh_scheduled while holding ap->lock such
488 489 490 491 492 493
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

494
		spin_unlock_irqrestore(ap->lock, flags);
495
	} else {
T
Tejun Heo 已提交
496
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
497 498
		ap->ops->eng_timeout(ap);
	}
499

500
	/* finish or retry handled scmd's and clean up */
501 502 503 504
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

505
	/* clean up */
506
	spin_lock_irqsave(ap->lock, flags);
507

508
	if (ap->pflags & ATA_PFLAG_LOADING)
509
		ap->pflags &= ~ATA_PFLAG_LOADING;
510
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
511
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
512 513 514

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
515

516
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
517

518
	/* tell wait_eh that we're done */
519
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
520 521
	wake_up_all(&ap->eh_wait_q);

522
	spin_unlock_irqrestore(ap->lock, flags);
523

524 525 526
	DPRINTK("EXIT\n");
}

527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
/**
 *	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:
542
	spin_lock_irqsave(ap->lock, flags);
543

544
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
545
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
546
		spin_unlock_irqrestore(ap->lock, flags);
547
		schedule();
548
		spin_lock_irqsave(ap->lock, flags);
549
	}
550
	finish_wait(&ap->eh_wait_q, &wait);
551

552
	spin_unlock_irqrestore(ap->lock, flags);
553 554

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
555
	if (scsi_host_in_recovery(ap->scsi_host)) {
556 557 558 559 560
		msleep(10);
		goto retry;
	}
}

561 562 563 564 565 566 567 568 569 570 571 572 573 574 575
/**
 *	ata_qc_timeout - Handle timeout of queued command
 *	@qc: Command that timed out
 *
 *	Some part of the kernel (currently, only the SCSI layer)
 *	has noticed that the active command on port @ap has not
 *	completed after a specified length of time.  Handle this
 *	condition by disabling DMA (if necessary) and completing
 *	transactions, with error if necessary.
 *
 *	This also handles the case of the "lost interrupt", where
 *	for some reason (possibly hardware bug, possibly driver bug)
 *	an interrupt was not delivered to the driver, even though the
 *	transaction completed successfully.
 *
576 577
 *	TODO: kill this function once old EH is gone.
 *
578 579 580 581 582 583 584 585 586 587 588 589 590
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 */
static void ata_qc_timeout(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;
	u8 host_stat = 0, drv_stat;
	unsigned long flags;

	DPRINTK("ENTER\n");

	ap->hsm_task_state = HSM_ST_IDLE;

591
	spin_lock_irqsave(ap->lock, flags);
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610

	switch (qc->tf.protocol) {

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

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

		/* fall through */

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

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

611 612 613
		ata_dev_printk(qc->dev, KERN_ERR, "command 0x%x timeout, "
			       "stat 0x%x host_stat 0x%x\n",
			       qc->tf.command, drv_stat, host_stat);
614 615

		/* complete taskfile transaction */
616
		qc->err_mask |= AC_ERR_TIMEOUT;
617 618 619
		break;
	}

620
	spin_unlock_irqrestore(ap->lock, flags);
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641

	ata_eh_qc_complete(qc);

	DPRINTK("EXIT\n");
}

/**
 *	ata_eng_timeout - Handle timeout of queued command
 *	@ap: Port on which timed-out command is active
 *
 *	Some part of the kernel (currently, only the SCSI layer)
 *	has noticed that the active command on port @ap has not
 *	completed after a specified length of time.  Handle this
 *	condition by disabling DMA (if necessary) and completing
 *	transactions, with error if necessary.
 *
 *	This also handles the case of the "lost interrupt", where
 *	for some reason (possibly hardware bug, possibly driver bug)
 *	an interrupt was not delivered to the driver, even though the
 *	transaction completed successfully.
 *
642 643
 *	TODO: kill this function once old EH is gone.
 *
644 645 646 647 648 649 650
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 */
void ata_eng_timeout(struct ata_port *ap)
{
	DPRINTK("ENTER\n");

T
Tejun Heo 已提交
651
	ata_qc_timeout(ata_qc_from_tag(ap, ap->link.active_tag));
652 653 654 655

	DPRINTK("EXIT\n");
}

T
Tejun Heo 已提交
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
static int ata_eh_nr_in_flight(struct ata_port *ap)
{
	unsigned int tag;
	int nr = 0;

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

	return nr;
}

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

	spin_lock_irqsave(ap->lock, flags);

	cnt = ata_eh_nr_in_flight(ap);

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

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

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

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

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

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

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

	ap->pflags |= ATA_PFLAG_EH_PENDING;

	if (!fastdrain)
		return;

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

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

744 745 746 747 748 749 750 751
/**
 *	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
752
 *	spin_lock_irqsave(host lock)
753 754 755 756 757 758 759 760
 */
void ata_qc_schedule_eh(struct ata_queued_cmd *qc)
{
	struct ata_port *ap = qc->ap;

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

	qc->flags |= ATA_QCFLAG_FAILED;
T
Tejun Heo 已提交
761
	ata_eh_set_pending(ap, 1);
762 763 764 765 766 767 768 769 770

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

771 772 773 774 775 776 777 778
/**
 *	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
779
 *	spin_lock_irqsave(host lock)
780 781 782 783 784
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

785 786 787
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
788
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
789
	scsi_schedule_eh(ap->scsi_host);
790 791 792 793

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

794
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
795 796 797 798 799
{
	int tag, nr_aborted = 0;

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

T
Tejun Heo 已提交
800 801 802
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

803 804 805
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

806
		if (qc && (!link || qc->dev->link == link)) {
807 808 809 810 811 812 813 814 815 816 817 818
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852
/**
 *	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);
}

853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
/**
 *	__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
869
 *	spin_lock_irqsave(host lock)
870 871 872 873 874 875 876 877
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

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

878
	ap->pflags |= ATA_PFLAG_FROZEN;
879

T
Tejun Heo 已提交
880
	DPRINTK("ata%u port frozen\n", ap->print_id);
881 882 883 884 885 886 887 888 889
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
890
 *	spin_lock_irqsave(host lock)
891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906
 *
 *	RETURNS:
 *	Number of aborted commands.
 */
int ata_port_freeze(struct ata_port *ap)
{
	int nr_aborted;

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

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

	return nr_aborted;
}

907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
/**
 *	sata_async_notification - SATA async notification handler
 *	@ap: ATA port where async notification is received
 *
 *	Handler to be called when async notification via SDB FIS is
 *	received.  This function schedules EH if necessary.
 *
 *	LOCKING:
 *	spin_lock_irqsave(host lock)
 *
 *	RETURNS:
 *	1 if EH is scheduled, 0 otherwise.
 */
int sata_async_notification(struct ata_port *ap)
{
	u32 sntf;
	int rc;

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

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

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

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

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

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

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

		return 0;
	}
}

980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995
/**
 *	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;

996
	spin_lock_irqsave(ap->lock, flags);
997
	__ata_port_freeze(ap);
998
	spin_unlock_irqrestore(ap->lock, flags);
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016
}

/**
 *	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;

1017
	spin_lock_irqsave(ap->lock, flags);
1018

1019
	ap->pflags &= ~ATA_PFLAG_FROZEN;
1020 1021 1022 1023

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

1024
	spin_unlock_irqrestore(ap->lock, flags);
1025

T
Tejun Heo 已提交
1026
	DPRINTK("ata%u port thawed\n", ap->print_id);
1027 1028
}

1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
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;

1040
	spin_lock_irqsave(ap->lock, flags);
1041 1042 1043
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
1044
	spin_unlock_irqrestore(ap->lock, flags);
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 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080

	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
1081

1082 1083 1084 1085 1086 1087 1088 1089 1090
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
1091
void ata_eh_detach_dev(struct ata_device *dev)
1092
{
1093 1094
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1095 1096 1097 1098
	unsigned long flags;

	ata_dev_disable(dev);

1099
	spin_lock_irqsave(ap->lock, flags);
1100 1101 1102 1103 1104

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1105
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1106 1107
	}

1108
	/* clear per-dev EH actions */
1109 1110
	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);
1111

1112
	spin_unlock_irqrestore(ap->lock, flags);
1113 1114
}

T
Tejun Heo 已提交
1115 1116
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
1117
 *	@link: target ATA link
1118
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
1119 1120 1121
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
1122 1123
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
1124 1125 1126 1127
 *
 *	LOCKING:
 *	None.
 */
1128 1129
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
T
Tejun Heo 已提交
1130
{
T
Tejun Heo 已提交
1131 1132 1133
	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
1134 1135
	unsigned long flags;

1136
	spin_lock_irqsave(ap->lock, flags);
1137

1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
	/* Reset is represented by combination of actions and EHI
	 * flags.  Suck in all related bits before clearing eh_info to
	 * avoid losing requested action.
	 */
	if (action & ATA_EH_RESET_MASK) {
		ehc->i.action |= ehi->action & ATA_EH_RESET_MASK;
		ehc->i.flags |= ehi->flags & ATA_EHI_RESET_MODIFIER_MASK;

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

T
Tejun Heo 已提交
1151
	ata_eh_clear_action(link, dev, ehi, action);
1152

1153
	if (!(ehc->i.flags & ATA_EHI_QUIET))
1154 1155
		ap->pflags |= ATA_PFLAG_RECOVERED;

1156
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
1157 1158
}

1159 1160
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
1161
*	@ap: target ATA port
1162 1163 1164 1165
 *	@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
1166
 *	in @link->eh_context.
1167 1168 1169 1170
 *
 *	LOCKING:
 *	None.
 */
1171 1172
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1173
{
T
Tejun Heo 已提交
1174
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1175

1176 1177 1178
	/* if reset is complete, clear all reset actions & reset modifier */
	if (action & ATA_EH_RESET_MASK) {
		action |= ATA_EH_RESET_MASK;
T
Tejun Heo 已提交
1179
		ehc->i.flags &= ~ATA_EHI_RESET_MODIFIER_MASK;
1180 1181
	}

T
Tejun Heo 已提交
1182
	ata_eh_clear_action(link, dev, &ehc->i, action);
1183 1184
}

T
Tejun Heo 已提交
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
/**
 *	ata_err_string - convert err_mask to descriptive string
 *	@err_mask: error mask to convert to string
 *
 *	Convert @err_mask to descriptive string.  Errors are
 *	prioritized according to severity and only the most severe
 *	error is reported.
 *
 *	LOCKING:
 *	None.
 *
 *	RETURNS:
 *	Descriptive string for @err_mask
 */
static const char * ata_err_string(unsigned int err_mask)
{
	if (err_mask & AC_ERR_HOST_BUS)
		return "host bus error";
	if (err_mask & AC_ERR_ATA_BUS)
		return "ATA bus error";
	if (err_mask & AC_ERR_TIMEOUT)
		return "timeout";
	if (err_mask & AC_ERR_HSM)
		return "HSM violation";
	if (err_mask & AC_ERR_SYSTEM)
		return "internal error";
	if (err_mask & AC_ERR_MEDIA)
		return "media error";
	if (err_mask & AC_ERR_INVALID)
		return "invalid argument";
	if (err_mask & AC_ERR_DEV)
		return "device error";
	return "unknown error";
}

1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275
/**
 *	ata_read_log_page - read a specific log page
 *	@dev: target device
 *	@page: page to read
 *	@buf: buffer to store read page
 *	@sectors: number of sectors to read
 *
 *	Read log page using READ_LOG_EXT command.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask otherwise.
 */
static unsigned int ata_read_log_page(struct ata_device *dev,
				      u8 page, void *buf, unsigned int sectors)
{
	struct ata_taskfile tf;
	unsigned int err_mask;

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

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

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

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

/**
 *	ata_eh_read_log_10h - Read log page 10h for NCQ error details
 *	@dev: Device to read log page 10h from
 *	@tag: Resulting tag of the failed command
 *	@tf: Resulting taskfile registers of the failed command
 *
 *	Read log page 10h to obtain NCQ error details and clear error
 *	condition.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno otherwise.
 */
static int ata_eh_read_log_10h(struct ata_device *dev,
			       int *tag, struct ata_taskfile *tf)
{
T
Tejun Heo 已提交
1276
	u8 *buf = dev->link->ap->sector_buf;
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
	unsigned int err_mask;
	u8 csum;
	int i;

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

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

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

	*tag = buf[0] & 0x1f;

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

	return 0;
}

T
Tejun Heo 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
/**
 *	atapi_eh_request_sense - perform ATAPI REQUEST_SENSE
 *	@dev: device to perform REQUEST_SENSE to
 *	@sense_buf: result sense data buffer (SCSI_SENSE_BUFFERSIZE bytes long)
 *
 *	Perform ATAPI REQUEST_SENSE after the device reported CHECK
 *	SENSE.  This function is EH helper.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, AC_ERR_* mask on failure
 */
1326
static unsigned int atapi_eh_request_sense(struct ata_queued_cmd *qc)
T
Tejun Heo 已提交
1327
{
1328 1329
	struct ata_device *dev = qc->dev;
	unsigned char *sense_buf = qc->scsicmd->sense_buffer;
T
Tejun Heo 已提交
1330
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1331 1332 1333 1334 1335 1336 1337 1338
	struct ata_taskfile tf;
	u8 cdb[ATAPI_CDB_LEN];

	DPRINTK("ATAPI request sense\n");

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

1339 1340 1341
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1342
	sense_buf[0] = 0x70;
1343 1344
	sense_buf[2] = qc->result_tf.feature >> 4;

J
Jeff Garzik 已提交
1345
	/* some devices time out if garbage left in tf */
1346
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370

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

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

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

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

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1371
 *	@link: ATA link to analyze SError for
T
Tejun Heo 已提交
1372 1373 1374 1375 1376 1377 1378
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1379
static void ata_eh_analyze_serror(struct ata_link *link)
T
Tejun Heo 已提交
1380
{
1381
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1382 1383
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1384
	u32 hotplug_mask;
T
Tejun Heo 已提交
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400

	if (serror & SERR_PERSISTENT) {
		err_mask |= AC_ERR_ATA_BUS;
		action |= ATA_EH_HARDRESET;
	}
	if (serror &
	    (SERR_DATA_RECOVERED | SERR_COMM_RECOVERED | SERR_DATA)) {
		err_mask |= AC_ERR_ATA_BUS;
		action |= ATA_EH_SOFTRESET;
	}
	if (serror & SERR_PROTOCOL) {
		err_mask |= AC_ERR_HSM;
		action |= ATA_EH_SOFTRESET;
	}
	if (serror & SERR_INTERNAL) {
		err_mask |= AC_ERR_SYSTEM;
T
Tejun Heo 已提交
1401
		action |= ATA_EH_HARDRESET;
T
Tejun Heo 已提交
1402
	}
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416

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

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

	if (serror & hotplug_mask)
T
Tejun Heo 已提交
1417
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1418 1419 1420 1421 1422

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

1423 1424
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
1425
 *	@link: ATA link to analyze NCQ error for
1426 1427 1428 1429 1430 1431 1432 1433 1434
 *
 *	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).
 */
1435
static void ata_eh_analyze_ncq_error(struct ata_link *link)
1436
{
1437 1438 1439
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
1440 1441 1442 1443 1444
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1445
	if (ap->pflags & ATA_PFLAG_FROZEN)
1446 1447 1448
		return;

	/* is it NCQ device error? */
1449
	if (!link->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
		return;

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

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

		if (qc->err_mask)
			return;
	}

	/* okay, this error is ours */
	rc = ata_eh_read_log_10h(dev, &tag, &tf);
	if (rc) {
1466
		ata_link_printk(link, KERN_ERR, "failed to read log page 10h "
1467 1468 1469 1470
				"(errno=%d)\n", rc);
		return;
	}

1471 1472
	if (!(link->sactive & (1 << tag))) {
		ata_link_printk(link, KERN_ERR, "log page 10h reported "
1473 1474 1475 1476 1477 1478 1479
				"inactive tag %d\n", tag);
		return;
	}

	/* we've got the perpetrator, condemn it */
	qc = __ata_qc_from_tag(ap, tag);
	memcpy(&qc->result_tf, &tf, sizeof(tf));
T
Tejun Heo 已提交
1480
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1481 1482 1483
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
/**
 *	ata_eh_analyze_tf - analyze taskfile of a failed qc
 *	@qc: qc to analyze
 *	@tf: Taskfile registers to analyze
 *
 *	Analyze taskfile of @qc and further determine cause of
 *	failure.  This function also requests ATAPI sense data if
 *	avaliable.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	Determined recovery action
 */
static unsigned int ata_eh_analyze_tf(struct ata_queued_cmd *qc,
				      const struct ata_taskfile *tf)
{
	unsigned int tmp, action = 0;
	u8 stat = tf->command, err = tf->feature;

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

1510 1511 1512
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525
		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:
1526
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1527
			tmp = atapi_eh_request_sense(qc);
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
			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
1540 1541 1542 1543 1544 1545 1546 1547
	}

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

	return action;
}

1548
static int ata_eh_categorize_error(int is_io, unsigned int err_mask)
T
Tejun Heo 已提交
1549
{
1550
	if (err_mask & AC_ERR_ATA_BUS)
T
Tejun Heo 已提交
1551 1552
		return 1;

1553 1554 1555 1556 1557
	if (err_mask & AC_ERR_TIMEOUT)
		return 2;

	if (is_io) {
		if (err_mask & AC_ERR_HSM)
T
Tejun Heo 已提交
1558
			return 2;
1559 1560 1561
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
			return 3;
T
Tejun Heo 已提交
1562 1563 1564 1565 1566
	}

	return 0;
}

1567
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1568
	u64 since;
1569
	int nr_errors[4];
T
Tejun Heo 已提交
1570 1571
};

1572
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1573
{
1574 1575
	struct speed_down_verdict_arg *arg = void_arg;
	int cat = ata_eh_categorize_error(ent->is_io, ent->err_mask);
T
Tejun Heo 已提交
1576 1577 1578 1579

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

1580
	arg->nr_errors[cat]++;
T
Tejun Heo 已提交
1581 1582 1583 1584
	return 0;
}

/**
1585
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1586 1587 1588
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1589 1590
 *	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
1591
 *
1592
 *	Cat-1 is ATA_BUS error for any command.
T
Tejun Heo 已提交
1593
 *
1594 1595 1596 1597
 *	Cat-2 is TIMEOUT for any command or HSM violation for known
 *	supported commands.
 *
 *	Cat-3 is is unclassified DEV error for known supported
T
Tejun Heo 已提交
1598 1599
 *	command.
 *
1600 1601 1602 1603 1604 1605 1606 1607 1608
 *	NCQ needs to be turned off if there have been more than 3
 *	Cat-2 + Cat-3 errors during last 10 minutes.
 *
 *	Speed down is necessary if there have been more than 3 Cat-1 +
 *	Cat-2 errors or 10 Cat-3 errors during last 10 minutes.
 *
 *	Falling back to PIO mode is necessary if there have been more
 *	than 10 Cat-1 + Cat-2 + Cat-3 errors during last 5 minutes.
 *
T
Tejun Heo 已提交
1609 1610 1611 1612
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1613
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1614
 */
1615
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1616
{
1617 1618 1619 1620
	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
1621

1622 1623 1624 1625
	/* scan past 10 mins of error history */
	memset(&arg, 0, sizeof(arg));
	arg.since = j64 - min(j64, j10mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1626

1627 1628 1629 1630
	if (arg.nr_errors[2] + arg.nr_errors[3] > 3)
		verdict |= ATA_EH_SPDN_NCQ_OFF;
	if (arg.nr_errors[1] + arg.nr_errors[2] > 3 || arg.nr_errors[3] > 10)
		verdict |= ATA_EH_SPDN_SPEED_DOWN;
T
Tejun Heo 已提交
1631

1632
	/* scan past 3 mins of error history */
T
Tejun Heo 已提交
1633
	memset(&arg, 0, sizeof(arg));
1634 1635
	arg.since = j64 - min(j64, j5mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1636

1637 1638
	if (arg.nr_errors[1] + arg.nr_errors[2] + arg.nr_errors[3] > 10)
		verdict |= ATA_EH_SPDN_FALLBACK_TO_PIO;
T
Tejun Heo 已提交
1639

1640
	return verdict;
T
Tejun Heo 已提交
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
}

/**
 *	ata_eh_speed_down - record error and speed down if necessary
 *	@dev: Failed device
 *	@is_io: Did the device fail during normal IO?
 *	@err_mask: err_mask of the error
 *
 *	Record error and examine error history to determine whether
 *	adjusting transmission speed is necessary.  It also sets
 *	transmission limits appropriately if such adjustment is
 *	necessary.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
1658
 *	Determined recovery action.
T
Tejun Heo 已提交
1659
 */
1660 1661
static unsigned int ata_eh_speed_down(struct ata_device *dev, int is_io,
				      unsigned int err_mask)
T
Tejun Heo 已提交
1662
{
1663 1664 1665 1666 1667
	unsigned int verdict;
	unsigned int action = 0;

	/* don't bother if Cat-0 error */
	if (ata_eh_categorize_error(is_io, err_mask) == 0)
T
Tejun Heo 已提交
1668 1669 1670 1671
		return 0;

	/* record error and determine whether speed down is necessary */
	ata_ering_record(&dev->ering, is_io, err_mask);
1672
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1673

1674 1675 1676 1677 1678 1679 1680 1681 1682
	/* turn off NCQ? */
	if ((verdict & ATA_EH_SPDN_NCQ_OFF) &&
	    (dev->flags & (ATA_DFLAG_PIO | ATA_DFLAG_NCQ |
			   ATA_DFLAG_NCQ_OFF)) == ATA_DFLAG_NCQ) {
		dev->flags |= ATA_DFLAG_NCQ_OFF;
		ata_dev_printk(dev, KERN_WARNING,
			       "NCQ disabled due to excessive errors\n");
		goto done;
	}
T
Tejun Heo 已提交
1683

1684 1685 1686
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1687
		if (sata_down_spd_limit(dev->link) == 0) {
1688 1689 1690
			action |= ATA_EH_HARDRESET;
			goto done;
		}
T
Tejun Heo 已提交
1691

1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
		/* lower transfer mode */
		if (dev->spdn_cnt < 2) {
			static const int dma_dnxfer_sel[] =
				{ ATA_DNXFER_DMA, ATA_DNXFER_40C };
			static const int pio_dnxfer_sel[] =
				{ ATA_DNXFER_PIO, ATA_DNXFER_FORCE_PIO0 };
			int sel;

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

			dev->spdn_cnt++;

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

	/* Fall back to PIO?  Slowing down to PIO is meaningless for
	 * SATA.  Consider it only for PATA.
	 */
	if ((verdict & ATA_EH_SPDN_FALLBACK_TO_PIO) && (dev->spdn_cnt >= 2) &&
T
Tejun Heo 已提交
1718
	    (dev->link->ap->cbl != ATA_CBL_SATA) &&
1719 1720 1721 1722 1723 1724 1725
	    (dev->xfer_shift != ATA_SHIFT_PIO)) {
		if (ata_down_xfermask_limit(dev, ATA_DNXFER_FORCE_PIO) == 0) {
			dev->spdn_cnt = 0;
			action |= ATA_EH_SOFTRESET;
			goto done;
		}
	}
T
Tejun Heo 已提交
1726 1727

	return 0;
1728 1729 1730 1731
 done:
	/* device has been slowed down, blow error history */
	ata_ering_clear(&dev->ering);
	return action;
T
Tejun Heo 已提交
1732 1733 1734
}

/**
1735 1736
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
T
Tejun Heo 已提交
1737
 *
1738 1739 1740
 *	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
1741 1742 1743 1744
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1745
static void ata_eh_link_autopsy(struct ata_link *link)
T
Tejun Heo 已提交
1746
{
1747
	struct ata_port *ap = link->ap;
1748
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1749 1750 1751 1752 1753 1754 1755
	unsigned int all_err_mask = 0;
	int tag, is_io = 0;
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1756 1757 1758
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1759
	/* obtain and analyze SError */
1760
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
1761 1762
	if (rc == 0) {
		ehc->i.serror |= serror;
1763
		ata_eh_analyze_serror(link);
1764 1765 1766
	} else if (rc != -EOPNOTSUPP) {
		/* SError read failed, force hardreset and probing */
		ata_ehi_schedule_probe(&ehc->i);
1767
		ehc->i.action |= ATA_EH_HARDRESET;
1768 1769
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
1770

1771
	/* analyze NCQ failure */
1772
	ata_eh_analyze_ncq_error(link);
1773

T
Tejun Heo 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782
	/* 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);

1783
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link)
T
Tejun Heo 已提交
1784 1785 1786 1787 1788 1789
			continue;

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

		/* analyze TF */
1790
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803

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

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

		/* SENSE_VALID trumps dev/unknown error and revalidation */
		if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);
1804
			ehc->i.action &= ~ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1805 1806 1807
		}

		/* accumulate error info */
1808
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1809 1810 1811 1812 1813
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
			is_io = 1;
	}

1814
	/* enforce default EH actions */
1815
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1816
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
1817
		ehc->i.action |= ATA_EH_SOFTRESET;
1818
	else if (all_err_mask)
1819
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1820

1821
	/* if we have offending qcs and the associated failed device */
1822
	if (ehc->i.dev) {
1823
		/* speed down */
1824 1825
		ehc->i.action |= ata_eh_speed_down(ehc->i.dev, is_io,
						   all_err_mask);
1826 1827

		/* perform per-dev EH action only on the offending device */
1828 1829 1830
		ehc->i.dev_action[ehc->i.dev->devno] |=
			ehc->i.action & ATA_EH_PERDEV_MASK;
		ehc->i.action &= ~ATA_EH_PERDEV_MASK;
1831 1832
	}

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

/**
1837 1838 1839 1840 1841 1842 1843 1844 1845
 *	ata_eh_autopsy - analyze error and determine recovery action
 *	@ap: host port to perform autopsy on
 *
 *	Analyze all links of @ap and determine why they failed and
 *	which recovery actions are needed.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1846
void ata_eh_autopsy(struct ata_port *ap)
1847 1848 1849 1850 1851 1852 1853 1854 1855
{
	struct ata_link *link;

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

/**
 *	ata_eh_link_report - report error handling to user
1856
 *	@link: ATA link EH is going on
T
Tejun Heo 已提交
1857 1858 1859 1860 1861 1862
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
1863
static void ata_eh_link_report(struct ata_link *link)
T
Tejun Heo 已提交
1864
{
1865 1866
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1867
	const char *frozen, *desc;
1868
	char tries_buf[6];
T
Tejun Heo 已提交
1869 1870 1871 1872 1873 1874 1875 1876 1877
	int tag, nr_failed = 0;

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

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

1878
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link)
T
Tejun Heo 已提交
1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

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

	frozen = "";
1890
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
1891 1892
		frozen = " frozen";

1893 1894 1895 1896 1897
	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
1898
	if (ehc->i.dev) {
1899
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
1900 1901 1902
			       "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
1903
		if (desc)
T
Tejun Heo 已提交
1904
			ata_dev_printk(ehc->i.dev, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1905
	} else {
1906
		ata_link_printk(link, KERN_ERR, "exception Emask 0x%x "
1907 1908 1909
				"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
1910
		if (desc)
1911
			ata_link_printk(link, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1912 1913
	}

1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934
	if (ehc->i.serror)
		ata_port_printk(ap, KERN_ERR,
		  "SError: { %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s}\n",
		  ehc->i.serror & SERR_DATA_RECOVERED ? "RecovData " : "",
		  ehc->i.serror & SERR_COMM_RECOVERED ? "RecovComm " : "",
		  ehc->i.serror & SERR_DATA ? "UnrecovData " : "",
		  ehc->i.serror & SERR_PERSISTENT ? "Persist " : "",
		  ehc->i.serror & SERR_PROTOCOL ? "Proto " : "",
		  ehc->i.serror & SERR_INTERNAL ? "HostInt " : "",
		  ehc->i.serror & SERR_PHYRDY_CHG ? "PHYRdyChg " : "",
		  ehc->i.serror & SERR_PHY_INT_ERR ? "PHYInt " : "",
		  ehc->i.serror & SERR_COMM_WAKE ? "CommWake " : "",
		  ehc->i.serror & SERR_10B_8B_ERR ? "10B8B " : "",
		  ehc->i.serror & SERR_DISPARITY ? "Dispar " : "",
		  ehc->i.serror & SERR_CRC ? "BadCRC " : "",
		  ehc->i.serror & SERR_HANDSHAKE ? "Handshk " : "",
		  ehc->i.serror & SERR_LINK_SEQ_ERR ? "LinkSeq " : "",
		  ehc->i.serror & SERR_TRANS_ST_ERROR ? "TrStaTrns " : "",
		  ehc->i.serror & SERR_UNRECOG_FIS ? "UnrecFIS " : "",
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "" );

T
Tejun Heo 已提交
1935
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
1936 1937 1938 1939 1940 1941
		static const char *dma_str[] = {
			[DMA_BIDIRECTIONAL]	= "bidi",
			[DMA_TO_DEVICE]		= "out",
			[DMA_FROM_DEVICE]	= "in",
			[DMA_NONE]		= "",
		};
T
Tejun Heo 已提交
1942
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
1943
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
T
Tejun Heo 已提交
1944

1945 1946
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    qc->dev->link != link || !qc->err_mask)
T
Tejun Heo 已提交
1947 1948
			continue;

1949 1950
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
1951
			"tag %d cdb 0x%x data %u %s\n         "
1952
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
1953
			"Emask 0x%x (%s)%s\n",
1954 1955 1956 1957
			cmd->command, cmd->feature, cmd->nsect,
			cmd->lbal, cmd->lbam, cmd->lbah,
			cmd->hob_feature, cmd->hob_nsect,
			cmd->hob_lbal, cmd->hob_lbam, cmd->hob_lbah,
T
Tejun Heo 已提交
1958
			cmd->device, qc->tag, qc->cdb[0], qc->nbytes,
1959
			dma_str[qc->dma_dir],
1960 1961 1962 1963
			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
1964 1965
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

		if (res->command & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
				    ATA_ERR) ) {
			if (res->command & ATA_BUSY)
				ata_dev_printk(qc->dev, KERN_ERR,
				  "status: { Busy }\n" );
			else
				ata_dev_printk(qc->dev, KERN_ERR,
				  "status: { %s%s%s%s}\n",
				  res->command & ATA_DRDY ? "DRDY " : "",
				  res->command & ATA_DF ? "DF " : "",
				  res->command & ATA_DRQ ? "DRQ " : "",
				  res->command & ATA_ERR ? "ERR " : "" );
		}

		if (cmd->command != ATA_CMD_PACKET &&
		    (res->feature & (ATA_ICRC | ATA_UNC | ATA_IDNF |
				     ATA_ABORTED)))
			ata_dev_printk(qc->dev, KERN_ERR,
			  "error: { %s%s%s%s}\n",
			  res->feature & ATA_ICRC ? "ICRC " : "",
			  res->feature & ATA_UNC ? "UNC " : "",
			  res->feature & ATA_IDNF ? "IDNF " : "",
			  res->feature & ATA_ABORTED ? "ABRT " : "" );
T
Tejun Heo 已提交
1990 1991 1992
	}
}

1993 1994 1995 1996 1997 1998 1999 2000 2001
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
2002
void ata_eh_report(struct ata_port *ap)
2003 2004 2005 2006 2007 2008 2009
{
	struct ata_link *link;

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

T
Tejun Heo 已提交
2010
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
2011
			unsigned int *classes, unsigned long deadline)
2012
{
2013 2014
	struct ata_device *dev;
	int rc;
2015

T
Tejun Heo 已提交
2016
	ata_link_for_each_dev(dev, link)
2017
		classes[dev->devno] = ATA_DEV_UNKNOWN;
2018

T
Tejun Heo 已提交
2019
	rc = reset(link, classes, deadline);
2020 2021 2022 2023 2024 2025 2026
	if (rc)
		return rc;

	/* If any class isn't ATA_DEV_UNKNOWN, consider classification
	 * is complete and convert all ATA_DEV_UNKNOWN to
	 * ATA_DEV_NONE.
	 */
T
Tejun Heo 已提交
2027
	ata_link_for_each_dev(dev, link)
2028
		if (classes[dev->devno] != ATA_DEV_UNKNOWN)
2029 2030
			break;

2031
	if (dev) {
T
Tejun Heo 已提交
2032
		ata_link_for_each_dev(dev, link) {
2033 2034 2035 2036
			if (classes[dev->devno] == ATA_DEV_UNKNOWN)
				classes[dev->devno] = ATA_DEV_NONE;
		}
	}
2037 2038 2039 2040

	return 0;
}

2041 2042
static int ata_eh_followup_srst_needed(struct ata_link *link,
				       int rc, int classify,
2043 2044
				       const unsigned int *classes)
{
2045 2046
	if (link->flags & ATA_LFLAG_NO_SRST)
		return 0;
2047 2048 2049 2050
	if (rc == -EAGAIN)
		return 1;
	if (rc != 0)
		return 0;
2051 2052
	if ((link->ap->flags & ATA_FLAG_PMP) && ata_is_host_link(link))
		return 1;
2053 2054
	if (classify && !(link->flags & ATA_LFLAG_ASSUME_CLASS) &&
	    classes[0] == ATA_DEV_UNKNOWN)
2055 2056 2057 2058
		return 1;
	return 0;
}

2059 2060 2061
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
2062
{
2063
	struct ata_eh_context *ehc = &link->eh_context;
2064
	unsigned int *classes = ehc->classes;
2065
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
T
Tejun Heo 已提交
2066
	int try = 0;
2067
	struct ata_device *dev;
T
Tejun Heo 已提交
2068
	unsigned long deadline;
2069
	unsigned int action;
T
Tejun Heo 已提交
2070
	ata_reset_fn_t reset;
2071
	int rc;
T
Tejun Heo 已提交
2072

2073
	/* about to reset */
T
Tejun Heo 已提交
2074
	ata_eh_about_to_do(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
2075

2076 2077 2078 2079 2080
	/* Determine which reset to use and record in ehc->i.action.
	 * prereset() may examine and modify it.
	 */
	action = ehc->i.action;
	ehc->i.action &= ~ATA_EH_RESET_MASK;
2081 2082
	if (softreset && (!hardreset || (!(link->flags & ATA_LFLAG_NO_SRST) &&
					 !sata_set_spd_needed(link) &&
2083 2084
					 !(action & ATA_EH_HARDRESET))))
		ehc->i.action |= ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
2085
	else
2086 2087 2088
		ehc->i.action |= ATA_EH_HARDRESET;

	if (prereset) {
T
Tejun Heo 已提交
2089
		rc = prereset(link, jiffies + ATA_EH_PRERESET_TIMEOUT);
2090
		if (rc) {
2091
			if (rc == -ENOENT) {
T
Tejun Heo 已提交
2092
				ata_link_printk(link, KERN_DEBUG,
2093
						"port disabled. ignoring.\n");
T
Tejun Heo 已提交
2094
				ehc->i.action &= ~ATA_EH_RESET_MASK;
2095

2096
				ata_link_for_each_dev(dev, link)
2097
					classes[dev->devno] = ATA_DEV_NONE;
2098 2099

				rc = 0;
2100
			} else
T
Tejun Heo 已提交
2101
				ata_link_printk(link, KERN_ERR,
2102
					"prereset failed (errno=%d)\n", rc);
2103
			goto out;
2104 2105 2106 2107 2108
		}
	}

	/* prereset() might have modified ehc->i.action */
	if (ehc->i.action & ATA_EH_HARDRESET)
T
Tejun Heo 已提交
2109
		reset = hardreset;
2110 2111 2112 2113
	else if (ehc->i.action & ATA_EH_SOFTRESET)
		reset = softreset;
	else {
		/* prereset told us not to reset, bang classes and return */
2114
		ata_link_for_each_dev(dev, link)
2115
			classes[dev->devno] = ATA_DEV_NONE;
2116 2117
		rc = 0;
		goto out;
2118 2119 2120 2121 2122 2123 2124 2125 2126
	}

	/* did prereset() screw up?  if so, fix up to avoid oopsing */
	if (!reset) {
		if (softreset)
			reset = softreset;
		else
			reset = hardreset;
	}
T
Tejun Heo 已提交
2127 2128

 retry:
T
Tejun Heo 已提交
2129 2130
	deadline = jiffies + ata_eh_reset_timeouts[try++];

2131 2132
	/* shut up during boot probing */
	if (verbose)
T
Tejun Heo 已提交
2133
		ata_link_printk(link, KERN_INFO, "%s resetting link\n",
2134
				reset == softreset ? "soft" : "hard");
T
Tejun Heo 已提交
2135

2136
	/* mark that this EH session started with reset */
2137 2138 2139 2140
	if (reset == hardreset)
		ehc->i.flags |= ATA_EHI_DID_HARDRESET;
	else
		ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
2141

T
Tejun Heo 已提交
2142
	rc = ata_do_reset(link, reset, classes, deadline);
T
Tejun Heo 已提交
2143

2144
	if (reset == hardreset &&
2145
	    ata_eh_followup_srst_needed(link, rc, classify, classes)) {
2146 2147 2148 2149
		/* okay, let's do follow-up softreset */
		reset = softreset;

		if (!reset) {
T
Tejun Heo 已提交
2150
			ata_link_printk(link, KERN_ERR,
2151 2152
					"follow-up softreset required "
					"but no softreset avaliable\n");
2153 2154
			rc = -EINVAL;
			goto out;
2155 2156
		}

T
Tejun Heo 已提交
2157
		ata_eh_about_to_do(link, NULL, ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
2158
		rc = ata_do_reset(link, reset, classes, deadline);
2159

2160 2161
		if (rc == 0 && classify && classes[0] == ATA_DEV_UNKNOWN &&
		    !(link->flags & ATA_LFLAG_ASSUME_CLASS)) {
T
Tejun Heo 已提交
2162
			ata_link_printk(link, KERN_ERR,
2163
					"classification failed\n");
2164 2165
			rc = -EINVAL;
			goto out;
2166 2167 2168
		}
	}

2169 2170 2171 2172
	/* if we skipped follow-up srst, clear rc */
	if (rc == -EAGAIN)
		rc = 0;

2173
	if (rc && rc != -ERESTART && try < ARRAY_SIZE(ata_eh_reset_timeouts)) {
T
Tejun Heo 已提交
2174
		unsigned long now = jiffies;
2175

T
Tejun Heo 已提交
2176 2177
		if (time_before(now, deadline)) {
			unsigned long delta = deadline - jiffies;
2178

T
Tejun Heo 已提交
2179
			ata_link_printk(link, KERN_WARNING, "reset failed "
T
Tejun Heo 已提交
2180 2181 2182 2183 2184
				"(errno=%d), retrying in %u secs\n",
				rc, (jiffies_to_msecs(delta) + 999) / 1000);

			schedule_timeout_uninterruptible(delta);
		}
T
Tejun Heo 已提交
2185

2186
		if (rc == -EPIPE ||
T
Tejun Heo 已提交
2187
		    try == ARRAY_SIZE(ata_eh_reset_timeouts) - 1)
2188
			sata_down_spd_limit(link);
T
Tejun Heo 已提交
2189 2190 2191 2192 2193 2194
		if (hardreset)
			reset = hardreset;
		goto retry;
	}

	if (rc == 0) {
2195 2196
		u32 sstatus;

2197 2198 2199 2200 2201
		ata_link_for_each_dev(dev, link) {
			/* After the reset, the device state is PIO 0
			 * and the controller state is undefined.
			 * Record the mode.
			 */
2202
			dev->pio_mode = XFER_PIO_0;
2203

2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215
			if (ata_link_offline(link))
				continue;

			/* apply class override and convert UNKNOWN to NONE */
			if (link->flags & ATA_LFLAG_ASSUME_ATA)
				classes[dev->devno] = ATA_DEV_ATA;
			else if (link->flags & ATA_LFLAG_ASSUME_SEMB)
				classes[dev->devno] = ATA_DEV_SEMB_UNSUP; /* not yet */
			else if (classes[dev->devno] == ATA_DEV_UNKNOWN)
				classes[dev->devno] = ATA_DEV_NONE;
		}

2216
		/* record current link speed */
2217 2218
		if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
			link->sata_spd = (sstatus >> 4) & 0xf;
2219

T
Tejun Heo 已提交
2220
		if (postreset)
T
Tejun Heo 已提交
2221
			postreset(link, classes);
T
Tejun Heo 已提交
2222 2223

		/* reset successful, schedule revalidation */
T
Tejun Heo 已提交
2224
		ata_eh_done(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
2225 2226
		ehc->i.action |= ATA_EH_REVALIDATE;
	}
2227 2228 2229
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2230 2231 2232
	return rc;
}

2233
static int ata_eh_revalidate_and_attach(struct ata_link *link,
T
Tejun Heo 已提交
2234
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
2235
{
2236 2237
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2238
	struct ata_device *dev;
2239
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
2240
	unsigned long flags;
2241
	int rc = 0;
T
Tejun Heo 已提交
2242 2243 2244

	DPRINTK("ENTER\n");

2245 2246 2247 2248
	/* 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.
	 */
2249
	ata_link_for_each_dev_reverse(dev, link) {
2250 2251
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
2252

2253 2254 2255
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

2256
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
2257 2258
			WARN_ON(dev->class == ATA_DEV_PMP);

2259
			if (ata_link_offline(link)) {
T
Tejun Heo 已提交
2260
				rc = -EIO;
2261
				goto err;
T
Tejun Heo 已提交
2262 2263
			}

2264
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
2265 2266
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
T
Tejun Heo 已提交
2267
			if (rc)
2268
				goto err;
T
Tejun Heo 已提交
2269

2270
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
2271

2272 2273 2274 2275 2276
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

2277 2278
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
2279 2280 2281 2282 2283
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

2284 2285 2286 2287 2288
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
2289 2290
			switch (rc) {
			case 0:
2291
				new_mask |= 1 << dev->devno;
2292 2293
				break;
			case -ENOENT:
2294 2295 2296 2297 2298
				/* IDENTIFY was issued to non-existent
				 * device.  No need to reset.  Just
				 * thaw and kill the device.
				 */
				ata_eh_thaw_port(ap);
T
Tejun Heo 已提交
2299 2300
				dev->class = ATA_DEV_UNKNOWN;
				break;
2301 2302 2303
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
2304
			}
2305 2306
		}
	}
T
Tejun Heo 已提交
2307

2308
	/* PDIAG- should have been released, ask cable type if post-reset */
2309 2310
	if (ata_is_host_link(link) && ap->ops->cable_detect &&
	    (ehc->i.flags & ATA_EHI_DID_RESET))
2311 2312
		ap->cbl = ap->ops->cable_detect(ap);

2313 2314 2315
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
2316
	ata_link_for_each_dev(dev, link) {
2317 2318
		if (!(new_mask & (1 << dev->devno)) ||
		    dev->class == ATA_DEV_PMP)
2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
			continue;

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

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

		/* new device discovered, configure xfermode */
		ehc->i.flags |= ATA_EHI_SETMODE;
T
Tejun Heo 已提交
2333 2334
	}

2335
	return 0;
T
Tejun Heo 已提交
2336

2337 2338 2339
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
2340 2341 2342
	return rc;
}

2343
static int ata_link_nr_enabled(struct ata_link *link)
T
Tejun Heo 已提交
2344
{
2345 2346
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2347

2348
	ata_link_for_each_dev(dev, link)
2349
		if (ata_dev_enabled(dev))
T
Tejun Heo 已提交
2350 2351 2352 2353
			cnt++;
	return cnt;
}

2354
static int ata_link_nr_vacant(struct ata_link *link)
T
Tejun Heo 已提交
2355
{
2356 2357
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2358

2359
	ata_link_for_each_dev(dev, link)
2360
		if (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
2361 2362 2363 2364
			cnt++;
	return cnt;
}

2365
static int ata_eh_skip_recovery(struct ata_link *link)
T
Tejun Heo 已提交
2366
{
2367
	struct ata_eh_context *ehc = &link->eh_context;
2368
	struct ata_device *dev;
T
Tejun Heo 已提交
2369

2370 2371 2372 2373
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

2374
	/* thaw frozen port, resume link and recover failed devices */
2375 2376
	if ((link->ap->pflags & ATA_PFLAG_FROZEN) ||
	    (ehc->i.flags & ATA_EHI_RESUME_LINK) || ata_link_nr_enabled(link))
T
Tejun Heo 已提交
2377 2378 2379
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
2380
	ata_link_for_each_dev(dev, link) {
T
Tejun Heo 已提交
2381 2382 2383 2384 2385 2386 2387 2388
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

2389
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
2390
{
T
Tejun Heo 已提交
2391
	struct ata_eh_context *ehc = &dev->link->eh_context;
2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406

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

	switch (err) {
	case -ENODEV:
		/* device missing or wrong IDENTIFY data, schedule probing */
		ehc->i.probe_mask |= (1 << dev->devno);
	case -EINVAL:
		/* give it just one more chance */
		ehc->tries[dev->devno] = min(ehc->tries[dev->devno], 1);
	case -EIO:
		if (ehc->tries[dev->devno] == 1) {
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
2407
			sata_down_spd_limit(dev->link);
2408 2409 2410 2411 2412 2413 2414 2415 2416
			ata_down_xfermask_limit(dev, ATA_DNXFER_PIO);
		}
	}

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

		/* detach if offline */
2417
		if (ata_link_offline(dev->link))
2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429
			ata_eh_detach_dev(dev);

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

			ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
			ehc->did_probe_mask |= (1 << dev->devno);
			ehc->i.action |= ATA_EH_SOFTRESET;
		}
2430 2431

		return 1;
2432 2433 2434 2435 2436 2437
	} else {
		/* soft didn't work?  be haaaaard */
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			ehc->i.action |= ATA_EH_HARDRESET;
		else
			ehc->i.action |= ATA_EH_SOFTRESET;
2438 2439

		return 0;
2440 2441 2442
	}
}

T
Tejun Heo 已提交
2443 2444 2445
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
2446
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2447 2448 2449
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
2450
 *	@r_failed_link: out parameter for failed link
T
Tejun Heo 已提交
2451 2452 2453
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
2454 2455 2456
 *	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
2457
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
2458 2459 2460 2461 2462 2463 2464
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
2465 2466 2467 2468
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
2469
{
2470
	struct ata_link *link;
T
Tejun Heo 已提交
2471
	struct ata_device *dev;
2472 2473
	int nr_failed_devs, nr_disabled_devs;
	int reset, rc;
2474
	unsigned long flags;
T
Tejun Heo 已提交
2475 2476 2477 2478

	DPRINTK("ENTER\n");

	/* prep for recovery */
2479 2480
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2481

2482 2483 2484 2485 2486 2487 2488 2489 2490
		/* 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);
		}

2491
		ata_link_for_each_dev(dev, link) {
2492 2493 2494 2495
			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
2496

2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513
			/* 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);

			if (!ata_dev_enabled(dev) &&
			    ((ehc->i.probe_mask & (1 << dev->devno)) &&
			     !(ehc->did_probe_mask & (1 << dev->devno)))) {
				ata_eh_detach_dev(dev);
				ata_dev_init(dev);
				ehc->did_probe_mask |= (1 << dev->devno);
				ehc->i.action |= ATA_EH_SOFTRESET;
			}
T
Tejun Heo 已提交
2514
		}
T
Tejun Heo 已提交
2515 2516 2517 2518
	}

 retry:
	rc = 0;
2519 2520 2521
	nr_failed_devs = 0;
	nr_disabled_devs = 0;
	reset = 0;
T
Tejun Heo 已提交
2522

2523
	/* if UNLOADING, finish immediately */
2524
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2525 2526
		goto out;

2527 2528 2529
	/* prep for EH */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2530

2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
		/* skip EH if possible. */
		if (ata_eh_skip_recovery(link))
			ehc->i.action = 0;

		/* do we need to reset? */
		if (ehc->i.action & ATA_EH_RESET_MASK)
			reset = 1;

		ata_link_for_each_dev(dev, link)
			ehc->classes[dev->devno] = ATA_DEV_UNKNOWN;
	}
T
Tejun Heo 已提交
2542

T
Tejun Heo 已提交
2543
	/* reset */
2544
	if (reset) {
T
Tejun Heo 已提交
2545 2546
		ata_eh_freeze_port(ap);

2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
		ata_port_for_each_link(link, ap) {
			struct ata_eh_context *ehc = &link->eh_context;

			if (!(ehc->i.action & ATA_EH_RESET_MASK))
				continue;

			rc = ata_eh_reset(link, ata_link_nr_vacant(link),
					  prereset, softreset, hardreset,
					  postreset);
			if (rc) {
				ata_link_printk(link, KERN_ERR,
						"reset failed, giving up\n");
				goto out;
			}
T
Tejun Heo 已提交
2561 2562 2563 2564 2565
		}

		ata_eh_thaw_port(ap);
	}

2566 2567 2568
	/* the rest */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2569

2570 2571
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
2572
		if (rc)
T
Tejun Heo 已提交
2573 2574
			goto dev_fail;

2575 2576 2577 2578 2579 2580
		/* 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;
		}

2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591
		/* configure transfer mode if necessary */
		if (ehc->i.flags & ATA_EHI_SETMODE) {
			rc = ata_set_mode(link, &dev);
			if (rc)
				goto dev_fail;
			ehc->i.flags &= ~ATA_EHI_SETMODE;
		}

		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
T
Tejun Heo 已提交
2592

2593 2594 2595 2596
	dev_fail:
		nr_failed_devs++;
		if (ata_eh_handle_dev_fail(dev, rc))
			nr_disabled_devs++;
T
Tejun Heo 已提交
2597

2598 2599
		if (ap->pflags & ATA_PFLAG_FROZEN)
			break;
T
Tejun Heo 已提交
2600 2601
	}

2602 2603 2604 2605 2606 2607 2608 2609 2610
	if (nr_failed_devs) {
		if (nr_failed_devs != nr_disabled_devs) {
			ata_port_printk(ap, KERN_WARNING, "failed to recover "
					"some devices, retrying in 5 secs\n");
			ssleep(5);
		} else {
			/* no device left to recover, repeat fast */
			msleep(500);
		}
T
Tejun Heo 已提交
2611

2612
		goto retry;
T
Tejun Heo 已提交
2613 2614
	}

2615 2616 2617 2618
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

T
Tejun Heo 已提交
2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632
	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.
 */
2633
void ata_eh_finish(struct ata_port *ap)
T
Tejun Heo 已提交
2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662
{
	int tag;

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

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

		if (qc->err_mask) {
			/* FIXME: Once EH migration is complete,
			 * generate sense data in this function,
			 * considering both err_mask and tf.
			 */
			if (qc->err_mask & AC_ERR_INVALID)
				ata_eh_qc_complete(qc);
			else
				ata_eh_qc_retry(qc);
		} else {
			if (qc->flags & ATA_QCFLAG_SENSE_VALID) {
				ata_eh_qc_complete(qc);
			} else {
				/* feed zero TF to sense generation */
				memset(&qc->result_tf, 0, sizeof(qc->result_tf));
				ata_eh_qc_retry(qc);
			}
		}
	}
2663 2664 2665 2666

	/* 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
2667 2668 2669 2670 2671
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2672
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2673 2674 2675 2676 2677 2678 2679 2680 2681
 *	@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).
 */
2682 2683 2684
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
2685
{
2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698
	struct ata_device *dev;
	int rc;

	ata_eh_autopsy(ap);
	ata_eh_report(ap);

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

T
Tejun Heo 已提交
2699 2700
	ata_eh_finish(ap);
}
2701

2702
#ifdef CONFIG_PM
2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727
/**
 *	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);

2728 2729 2730 2731 2732
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

2733 2734 2735 2736 2737 2738
	/* suspend */
	ata_eh_freeze_port(ap);

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

2739
 out:
2740 2741 2742 2743 2744 2745
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
2746
	else if (ap->pflags & ATA_PFLAG_FROZEN)
2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770
		ata_port_schedule_eh(ap);

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

	spin_unlock_irqrestore(ap->lock, flags);

	return;
}

/**
 *	ata_eh_handle_port_resume - perform port resume operation
 *	@ap: port to resume
 *
 *	Resume @ap.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_handle_port_resume(struct ata_port *ap)
{
	unsigned long flags;
2771
	int rc = 0;
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781

	/* 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);

2782
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
2783 2784 2785 2786

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

T
Tejun Heo 已提交
2787 2788 2789
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

2790
	/* report result */
2791 2792 2793 2794 2795 2796 2797 2798
	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);
}
2799
#endif /* CONFIG_PM */