libata-eh.c 69.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
/*
 *  libata-eh.c - libata error handling
 *
 *  Maintained by:  Jeff Garzik <jgarzik@pobox.com>
 *    		    Please ALWAYS copy linux-ide@vger.kernel.org
 *		    on emails.
 *
 *  Copyright 2006 Tejun Heo <htejun@gmail.com>
 *
 *
 *  This program is free software; you can redistribute it and/or
 *  modify it under the terms of the GNU General Public License as
 *  published by the Free Software Foundation; either version 2, or
 *  (at your option) any later version.
 *
 *  This program is distributed in the hope that it will be useful,
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 *  General Public License for more details.
 *
 *  You should have received a copy of the GNU General Public License
 *  along with this program; see the file COPYING.  If not, write to
 *  the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139,
 *  USA.
 *
 *
 *  libata documentation is available via 'make {ps|pdf}docs',
 *  as Documentation/DocBook/libata.*
 *
 *  Hardware documentation available from http://www.t13.org/ and
 *  http://www.sata-io.org/
 *
 */

#include <linux/kernel.h>
36
#include <linux/pci.h>
37 38 39 40 41
#include <scsi/scsi.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_cmnd.h>
J
Jeff Garzik 已提交
42
#include "../scsi/scsi_transport_api.h"
43 44 45 46 47

#include <linux/libata.h>

#include "libata.h"

48 49 50 51 52 53
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
54 55 56 57 58 59
/* 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
60
	ATA_EH_FASTDRAIN_INTERVAL	= 3 * HZ,
T
Tejun Heo 已提交
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
};

/* 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 */
};

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

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

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

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

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

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

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

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

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

154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
/**
 *	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 */

221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
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();
}

237
static void ata_ering_clear(struct ata_ering *ering)
238
{
239
	memset(ering, 0, sizeof(*ering));
240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
}

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;
}

263 264
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
T
Tejun Heo 已提交
265
	struct ata_eh_context *ehc = &dev->link->eh_context;
266 267 268 269

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

270
static void ata_eh_clear_action(struct ata_link *link, struct ata_device *dev,
271 272
				struct ata_eh_info *ehi, unsigned int action)
{
273
	struct ata_device *tdev;
274 275 276

	if (!dev) {
		ehi->action &= ~action;
277 278
		ata_link_for_each_dev(tdev, link)
			ehi->dev_action[tdev->devno] &= ~action;
279 280 281 282 283 284
	} 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) {
285 286 287
			ata_link_for_each_dev(tdev, link)
				ehi->dev_action[tdev->devno] |=
					ehi->action & action;
288 289 290 291 292 293 294 295
			ehi->action &= ~action;
		}

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

296 297 298 299 300 301 302 303 304 305 306
/**
 *	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.
 *
307 308
 *	TODO: kill this function once old EH is gone.
 *
309 310 311 312 313 314 315 316 317
 *	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;
318
	struct ata_port *ap = ata_shost_to_port(host);
319 320
	unsigned long flags;
	struct ata_queued_cmd *qc;
321
	enum scsi_eh_timer_return ret;
322 323 324

	DPRINTK("ENTER\n");

325 326 327 328 329 330
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
331
	spin_lock_irqsave(ap->lock, flags);
T
Tejun Heo 已提交
332
	qc = ata_qc_from_tag(ap, ap->link.active_tag);
333 334 335 336 337 338
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
339
	spin_unlock_irqrestore(ap->lock, flags);
340

341
 out:
342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357
	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.
 */
358
void ata_scsi_error(struct Scsi_Host *host)
359
{
360
	struct ata_port *ap = ata_shost_to_port(host);
361
	int i;
362
	unsigned long flags;
363 364 365

	DPRINTK("ENTER\n");

366
	/* synchronize with port task */
367 368
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
369
	/* synchronize with host lock and sort out timeouts */
370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386

	/* 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;

387
		spin_lock_irqsave(ap->lock, flags);
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425

		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);

426
		spin_unlock_irqrestore(ap->lock, flags);
427 428 429

		/* initialize eh_tries */
		ap->eh_tries = ATA_EH_MAX_TRIES;
430
	} else
431
		spin_unlock_wait(ap->lock);
432 433 434 435

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
436 437
		struct ata_link *link;

T
Tejun Heo 已提交
438 439 440
		/* kill fast drain timer */
		del_timer_sync(&ap->fastdrain_timer);

441 442 443
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

444
		/* fetch & clear EH info */
445
		spin_lock_irqsave(ap->lock, flags);
446

447 448 449 450 451
		__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));
		}
452

453 454
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
455
		ap->excl_link = NULL;	/* don't maintain exclusion over EH */
456

457
		spin_unlock_irqrestore(ap->lock, flags);
458

459 460
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
461 462 463
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
464

465 466 467
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

468 469 470 471
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
472
		spin_lock_irqsave(ap->lock, flags);
473

474
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
475
			if (--ap->eh_tries) {
476
				spin_unlock_irqrestore(ap->lock, flags);
477 478 479
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
480
					"tries, giving up\n", ATA_EH_MAX_TRIES);
T
Tejun Heo 已提交
481
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
482 483
		}

484
		/* this run is complete, make sure EH info is clear */
485 486
		__ata_port_for_each_link(link, ap)
			memset(&link->eh_info, 0, sizeof(link->eh_info));
487

488
		/* Clear host_eh_scheduled while holding ap->lock such
489 490 491 492 493 494
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

495
		spin_unlock_irqrestore(ap->lock, flags);
496
	} else {
T
Tejun Heo 已提交
497
		WARN_ON(ata_qc_from_tag(ap, ap->link.active_tag) == NULL);
498 499
		ap->ops->eng_timeout(ap);
	}
500

501
	/* finish or retry handled scmd's and clean up */
502 503 504 505
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

506
	/* clean up */
507
	spin_lock_irqsave(ap->lock, flags);
508

509
	if (ap->pflags & ATA_PFLAG_LOADING)
510
		ap->pflags &= ~ATA_PFLAG_LOADING;
511
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
512
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
513 514 515

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
516

517
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
518

519
	/* tell wait_eh that we're done */
520
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
521 522
	wake_up_all(&ap->eh_wait_q);

523
	spin_unlock_irqrestore(ap->lock, flags);
524

525 526 527
	DPRINTK("EXIT\n");
}

528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
/**
 *	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:
543
	spin_lock_irqsave(ap->lock, flags);
544

545
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
546
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
547
		spin_unlock_irqrestore(ap->lock, flags);
548
		schedule();
549
		spin_lock_irqsave(ap->lock, flags);
550
	}
551
	finish_wait(&ap->eh_wait_q, &wait);
552

553
	spin_unlock_irqrestore(ap->lock, flags);
554 555

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
556
	if (scsi_host_in_recovery(ap->scsi_host)) {
557 558 559 560 561
		msleep(10);
		goto retry;
	}
}

T
Tejun Heo 已提交
562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649
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);
}

650 651 652 653 654 655 656 657
/**
 *	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
658
 *	spin_lock_irqsave(host lock)
659 660 661 662 663 664 665 666
 */
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
667
	ata_eh_set_pending(ap, 1);
668 669 670 671 672 673 674 675 676

	/* 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);
}

677 678 679 680 681 682 683 684
/**
 *	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
685
 *	spin_lock_irqsave(host lock)
686 687 688 689 690
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

691 692 693
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

T
Tejun Heo 已提交
694
	ata_eh_set_pending(ap, 1);
J
Jeff Garzik 已提交
695
	scsi_schedule_eh(ap->scsi_host);
696 697 698 699

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

700
static int ata_do_link_abort(struct ata_port *ap, struct ata_link *link)
701 702 703 704 705
{
	int tag, nr_aborted = 0;

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

T
Tejun Heo 已提交
706 707 708
	/* we're gonna abort all commands, no need for fast drain */
	ata_eh_set_pending(ap, 0);

709 710 711
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = ata_qc_from_tag(ap, tag);

712
		if (qc && (!link || qc->dev->link == link)) {
713 714 715 716 717 718 719 720 721 722 723 724
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
/**
 *	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);
}

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

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

784
	ap->pflags |= ATA_PFLAG_FROZEN;
785

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

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
796
 *	spin_lock_irqsave(host lock)
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812
 *
 *	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;
}

813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885
/**
 *	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;
	}
}

886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
/**
 *	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;

902
	spin_lock_irqsave(ap->lock, flags);
903
	__ata_port_freeze(ap);
904
	spin_unlock_irqrestore(ap->lock, flags);
905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922
}

/**
 *	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;

923
	spin_lock_irqsave(ap->lock, flags);
924

925
	ap->pflags &= ~ATA_PFLAG_FROZEN;
926 927 928 929

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

930
	spin_unlock_irqrestore(ap->lock, flags);
931

T
Tejun Heo 已提交
932
	DPRINTK("ata%u port thawed\n", ap->print_id);
933 934
}

935 936 937 938 939 940 941 942 943 944 945
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;

946
	spin_lock_irqsave(ap->lock, flags);
947 948 949
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
950
	spin_unlock_irqrestore(ap->lock, flags);
951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986

	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
987

988 989 990 991 992 993 994 995 996
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
997
void ata_eh_detach_dev(struct ata_device *dev)
998
{
999 1000
	struct ata_link *link = dev->link;
	struct ata_port *ap = link->ap;
1001 1002 1003 1004
	unsigned long flags;

	ata_dev_disable(dev);

1005
	spin_lock_irqsave(ap->lock, flags);
1006 1007 1008 1009 1010

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
1011
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
1012 1013
	}

1014
	/* clear per-dev EH actions */
1015 1016
	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);
1017

1018
	spin_unlock_irqrestore(ap->lock, flags);
1019 1020
}

T
Tejun Heo 已提交
1021 1022
/**
 *	ata_eh_about_to_do - about to perform eh_action
T
Tejun Heo 已提交
1023
 *	@link: target ATA link
1024
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
1025 1026 1027
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
T
Tejun Heo 已提交
1028 1029
 *	in @link->eh_info such that eh actions are not unnecessarily
 *	repeated.
T
Tejun Heo 已提交
1030 1031 1032 1033
 *
 *	LOCKING:
 *	None.
 */
1034 1035
void ata_eh_about_to_do(struct ata_link *link, struct ata_device *dev,
			unsigned int action)
T
Tejun Heo 已提交
1036
{
T
Tejun Heo 已提交
1037 1038 1039
	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
1040 1041
	unsigned long flags;

1042
	spin_lock_irqsave(ap->lock, flags);
1043

1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
	/* 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
1057
	ata_eh_clear_action(link, dev, ehi, action);
1058

1059
	if (!(ehc->i.flags & ATA_EHI_QUIET))
1060 1061
		ap->pflags |= ATA_PFLAG_RECOVERED;

1062
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
1063 1064
}

1065 1066
/**
 *	ata_eh_done - EH action complete
T
Tejun Heo 已提交
1067
*	@ap: target ATA port
1068 1069 1070 1071
 *	@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
1072
 *	in @link->eh_context.
1073 1074 1075 1076
 *
 *	LOCKING:
 *	None.
 */
1077 1078
void ata_eh_done(struct ata_link *link, struct ata_device *dev,
		 unsigned int action)
1079
{
T
Tejun Heo 已提交
1080
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1081

1082 1083 1084
	/* 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
1085
		ehc->i.flags &= ~ATA_EHI_RESET_MODIFIER_MASK;
1086 1087
	}

T
Tejun Heo 已提交
1088
	ata_eh_clear_action(link, dev, &ehc->i, action);
1089 1090
}

T
Tejun Heo 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
/**
 *	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
 */
1105
static const char *ata_err_string(unsigned int err_mask)
T
Tejun Heo 已提交
1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
{
	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";
}

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 1151 1152 1153 1154 1155 1156 1157
/**
 *	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,
1158
				     buf, sectors * ATA_SECT_SIZE, 0);
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181

	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
1182
	u8 *buf = dev->link->ap->sector_buf;
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
	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
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
/**
 *	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
 */
1232
static unsigned int atapi_eh_request_sense(struct ata_queued_cmd *qc)
T
Tejun Heo 已提交
1233
{
1234 1235
	struct ata_device *dev = qc->dev;
	unsigned char *sense_buf = qc->scsicmd->sense_buffer;
T
Tejun Heo 已提交
1236
	struct ata_port *ap = dev->link->ap;
T
Tejun Heo 已提交
1237 1238 1239 1240 1241 1242 1243 1244
	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);

1245 1246 1247
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1248
	sense_buf[0] = 0x70;
1249 1250
	sense_buf[2] = qc->result_tf.feature >> 4;

J
Jeff Garzik 已提交
1251
	/* some devices time out if garbage left in tf */
1252
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266

	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;
1267 1268
		tf.lbam = SCSI_SENSE_BUFFERSIZE;
		tf.lbah = 0;
T
Tejun Heo 已提交
1269 1270 1271
	}

	return ata_exec_internal(dev, &tf, cdb, DMA_FROM_DEVICE,
1272
				 sense_buf, SCSI_SENSE_BUFFERSIZE, 0);
T
Tejun Heo 已提交
1273 1274 1275 1276
}

/**
 *	ata_eh_analyze_serror - analyze SError for a failed port
1277
 *	@link: ATA link to analyze SError for
T
Tejun Heo 已提交
1278 1279 1280 1281 1282 1283 1284
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
1285
static void ata_eh_analyze_serror(struct ata_link *link)
T
Tejun Heo 已提交
1286
{
1287
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1288 1289
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;
1290
	u32 hotplug_mask;
T
Tejun Heo 已提交
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

	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
1307
		action |= ATA_EH_HARDRESET;
T
Tejun Heo 已提交
1308
	}
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322

	/* 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
1323
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1324 1325 1326 1327 1328

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

1329 1330
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
1331
 *	@link: ATA link to analyze NCQ error for
1332 1333 1334 1335 1336 1337 1338 1339 1340
 *
 *	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).
 */
1341
static void ata_eh_analyze_ncq_error(struct ata_link *link)
1342
{
1343 1344 1345
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
	struct ata_device *dev = link->device;
1346 1347 1348 1349 1350
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1351
	if (ap->pflags & ATA_PFLAG_FROZEN)
1352 1353 1354
		return;

	/* is it NCQ device error? */
1355
	if (!link->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371
		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) {
1372
		ata_link_printk(link, KERN_ERR, "failed to read log page 10h "
1373 1374 1375 1376
				"(errno=%d)\n", rc);
		return;
	}

1377 1378
	if (!(link->sactive & (1 << tag))) {
		ata_link_printk(link, KERN_ERR, "log page 10h reported "
1379 1380 1381 1382 1383 1384 1385
				"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
1386
	qc->err_mask |= AC_ERR_DEV | AC_ERR_NCQ;
1387 1388 1389
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415
/**
 *	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;
	}

1416 1417 1418
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
		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:
1432
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1433
			tmp = atapi_eh_request_sense(qc);
1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445
			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
1446 1447 1448 1449 1450 1451 1452 1453
	}

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

	return action;
}

1454
static int ata_eh_categorize_error(int is_io, unsigned int err_mask)
T
Tejun Heo 已提交
1455
{
1456
	if (err_mask & AC_ERR_ATA_BUS)
T
Tejun Heo 已提交
1457 1458
		return 1;

1459 1460 1461 1462 1463
	if (err_mask & AC_ERR_TIMEOUT)
		return 2;

	if (is_io) {
		if (err_mask & AC_ERR_HSM)
T
Tejun Heo 已提交
1464
			return 2;
1465 1466 1467
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
			return 3;
T
Tejun Heo 已提交
1468 1469 1470 1471 1472
	}

	return 0;
}

1473
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1474
	u64 since;
1475
	int nr_errors[4];
T
Tejun Heo 已提交
1476 1477
};

1478
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1479
{
1480 1481
	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
1482 1483 1484 1485

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

1486
	arg->nr_errors[cat]++;
T
Tejun Heo 已提交
1487 1488 1489 1490
	return 0;
}

/**
1491
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1492 1493 1494
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1495 1496
 *	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
1497
 *
1498
 *	Cat-1 is ATA_BUS error for any command.
T
Tejun Heo 已提交
1499
 *
1500 1501 1502 1503
 *	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
1504 1505
 *	command.
 *
1506 1507 1508 1509 1510 1511 1512 1513 1514
 *	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
1515 1516 1517 1518
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1519
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1520
 */
1521
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1522
{
1523 1524 1525 1526
	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
1527

1528 1529 1530 1531
	/* 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
1532

1533 1534 1535 1536
	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
1537

1538
	/* scan past 3 mins of error history */
T
Tejun Heo 已提交
1539
	memset(&arg, 0, sizeof(arg));
1540 1541
	arg.since = j64 - min(j64, j5mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1542

1543 1544
	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
1545

1546
	return verdict;
T
Tejun Heo 已提交
1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
}

/**
 *	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:
1564
 *	Determined recovery action.
T
Tejun Heo 已提交
1565
 */
1566 1567
static unsigned int ata_eh_speed_down(struct ata_device *dev, int is_io,
				      unsigned int err_mask)
T
Tejun Heo 已提交
1568
{
1569 1570 1571 1572 1573
	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
1574 1575 1576 1577
		return 0;

	/* record error and determine whether speed down is necessary */
	ata_ering_record(&dev->ering, is_io, err_mask);
1578
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1579

1580 1581 1582 1583 1584 1585 1586 1587 1588
	/* 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
1589

1590 1591 1592
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
1593
		if (sata_down_spd_limit(dev->link) == 0) {
1594 1595 1596
			action |= ATA_EH_HARDRESET;
			goto done;
		}
T
Tejun Heo 已提交
1597

1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623
		/* 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
1624
	    (dev->link->ap->cbl != ATA_CBL_SATA) &&
1625 1626 1627 1628 1629 1630 1631
	    (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
1632 1633

	return 0;
1634 1635 1636 1637
 done:
	/* device has been slowed down, blow error history */
	ata_ering_clear(&dev->ering);
	return action;
T
Tejun Heo 已提交
1638 1639 1640
}

/**
1641 1642
 *	ata_eh_link_autopsy - analyze error and determine recovery action
 *	@link: host link to perform autopsy on
T
Tejun Heo 已提交
1643
 *
1644 1645 1646
 *	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
1647 1648 1649 1650
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
1651
static void ata_eh_link_autopsy(struct ata_link *link)
T
Tejun Heo 已提交
1652
{
1653
	struct ata_port *ap = link->ap;
1654
	struct ata_eh_context *ehc = &link->eh_context;
1655
	struct ata_device *dev;
T
Tejun Heo 已提交
1656 1657 1658 1659 1660 1661 1662
	unsigned int all_err_mask = 0;
	int tag, is_io = 0;
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1663 1664 1665
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1666
	/* obtain and analyze SError */
1667
	rc = sata_scr_read(link, SCR_ERROR, &serror);
T
Tejun Heo 已提交
1668 1669
	if (rc == 0) {
		ehc->i.serror |= serror;
1670
		ata_eh_analyze_serror(link);
1671 1672 1673
	} else if (rc != -EOPNOTSUPP) {
		/* SError read failed, force hardreset and probing */
		ata_ehi_schedule_probe(&ehc->i);
1674
		ehc->i.action |= ATA_EH_HARDRESET;
1675 1676
		ehc->i.err_mask |= AC_ERR_OTHER;
	}
T
Tejun Heo 已提交
1677

1678
	/* analyze NCQ failure */
1679
	ata_eh_analyze_ncq_error(link);
1680

T
Tejun Heo 已提交
1681 1682 1683 1684 1685 1686 1687 1688 1689
	/* 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);

1690
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link)
T
Tejun Heo 已提交
1691 1692 1693 1694 1695 1696
			continue;

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

		/* analyze TF */
1697
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708

		/* 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 */
1709
		if (qc->flags & ATA_QCFLAG_SENSE_VALID)
T
Tejun Heo 已提交
1710 1711 1712
			qc->err_mask &= ~(AC_ERR_DEV | AC_ERR_OTHER);

		/* accumulate error info */
1713
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1714 1715 1716 1717 1718
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
			is_io = 1;
	}

1719
	/* enforce default EH actions */
1720
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1721
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
1722
		ehc->i.action |= ATA_EH_SOFTRESET;
1723 1724
	else if ((is_io && all_err_mask) ||
		 (!is_io && (all_err_mask & ~AC_ERR_DEV)))
1725
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1726

1727 1728 1729
	/* If we have offending qcs and the associated failed device,
	 * perform per-dev EH action only on the offending device.
	 */
1730 1731 1732 1733
	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;
1734 1735
	}

1736 1737 1738 1739 1740
	/* 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 */
1741
	dev = ehc->i.dev;
1742 1743 1744
	if (!dev && ((ata_link_max_devices(link) == 1 &&
		      ata_dev_enabled(link->device))))
	    dev = link->device;
1745 1746 1747 1748

	if (dev)
		ehc->i.action |= ata_eh_speed_down(dev, is_io, all_err_mask);

T
Tejun Heo 已提交
1749 1750 1751 1752
	DPRINTK("EXIT\n");
}

/**
1753 1754 1755 1756 1757 1758 1759 1760 1761
 *	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).
 */
1762
void ata_eh_autopsy(struct ata_port *ap)
1763 1764 1765
{
	struct ata_link *link;

1766
	ata_port_for_each_link(link, ap)
1767
		ata_eh_link_autopsy(link);
1768 1769 1770 1771 1772 1773

	/* Autopsy of fanout ports can affect host link autopsy.
	 * Perform host link autopsy last.
	 */
	if (ap->nr_pmp_links)
		ata_eh_link_autopsy(&ap->link);
1774 1775 1776 1777
}

/**
 *	ata_eh_link_report - report error handling to user
1778
 *	@link: ATA link EH is going on
T
Tejun Heo 已提交
1779 1780 1781 1782 1783 1784
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
1785
static void ata_eh_link_report(struct ata_link *link)
T
Tejun Heo 已提交
1786
{
1787 1788
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
1789
	const char *frozen, *desc;
1790
	char tries_buf[6];
T
Tejun Heo 已提交
1791 1792
	int tag, nr_failed = 0;

1793 1794 1795
	if (ehc->i.flags & ATA_EHI_QUIET)
		return;

T
Tejun Heo 已提交
1796 1797 1798 1799 1800 1801 1802
	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);

1803 1804 1805
		if (!(qc->flags & ATA_QCFLAG_FAILED) || qc->dev->link != link ||
		    ((qc->flags & ATA_QCFLAG_QUIET) &&
		     qc->err_mask == AC_ERR_DEV))
T
Tejun Heo 已提交
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816
			continue;
		if (qc->flags & ATA_QCFLAG_SENSE_VALID && !qc->err_mask)
			continue;

		nr_failed++;
	}

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

	frozen = "";
1817
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
1818 1819
		frozen = " frozen";

1820 1821 1822 1823 1824
	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
1825
	if (ehc->i.dev) {
1826
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
1827 1828 1829
			       "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
1830
		if (desc)
T
Tejun Heo 已提交
1831
			ata_dev_printk(ehc->i.dev, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1832
	} else {
1833
		ata_link_printk(link, KERN_ERR, "exception Emask 0x%x "
1834 1835 1836
				"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
1837
		if (desc)
1838
			ata_link_printk(link, KERN_ERR, "%s\n", desc);
T
Tejun Heo 已提交
1839 1840
	}

1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
	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 " : "",
1860
		  ehc->i.serror & SERR_DEV_XCHG ? "DevExch " : "");
1861

T
Tejun Heo 已提交
1862 1863
	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
1864
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
1865 1866 1867
		const u8 *cdb = qc->cdb;
		char data_buf[20] = "";
		char cdb_buf[70] = "";
T
Tejun Heo 已提交
1868

1869 1870
		if (!(qc->flags & ATA_QCFLAG_FAILED) ||
		    qc->dev->link != link || !qc->err_mask)
T
Tejun Heo 已提交
1871 1872
			continue;

1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900
		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",
				[ATA_PROT_ATAPI]	= "pio",
				[ATA_PROT_ATAPI_DMA]	= "dma",
			};

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

		if (is_atapi_taskfile(&qc->tf))
			snprintf(cdb_buf, sizeof(cdb_buf),
				 "cdb %02x %02x %02x %02x %02x %02x %02x %02x  "
				 "%02x %02x %02x %02x %02x %02x %02x %02x\n         ",
				 cdb[0], cdb[1], cdb[2], cdb[3],
				 cdb[4], cdb[5], cdb[6], cdb[7],
				 cdb[8], cdb[9], cdb[10], cdb[11],
				 cdb[12], cdb[13], cdb[14], cdb[15]);

1901 1902
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
1903
			"tag %d%s\n         %s"
1904
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
T
Tejun Heo 已提交
1905
			"Emask 0x%x (%s)%s\n",
1906 1907 1908 1909
			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,
1910
			cmd->device, qc->tag, data_buf, cdb_buf,
1911 1912 1913 1914
			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
1915 1916
			res->device, qc->err_mask, ata_err_string(qc->err_mask),
			qc->err_mask & AC_ERR_NCQ ? " <F>" : "");
1917 1918

		if (res->command & (ATA_BUSY | ATA_DRDY | ATA_DF | ATA_DRQ |
1919
				    ATA_ERR)) {
1920 1921
			if (res->command & ATA_BUSY)
				ata_dev_printk(qc->dev, KERN_ERR,
1922
				  "status: { Busy }\n");
1923 1924 1925 1926 1927 1928
			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 " : "",
1929
				  res->command & ATA_ERR ? "ERR " : "");
1930 1931 1932 1933 1934 1935 1936 1937 1938 1939
		}

		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 " : "",
1940
			  res->feature & ATA_ABORTED ? "ABRT " : "");
T
Tejun Heo 已提交
1941 1942 1943
	}
}

1944 1945 1946 1947 1948 1949 1950 1951 1952
/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port to report EH about
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
1953
void ata_eh_report(struct ata_port *ap)
1954 1955 1956 1957 1958 1959 1960
{
	struct ata_link *link;

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

T
Tejun Heo 已提交
1961
static int ata_do_reset(struct ata_link *link, ata_reset_fn_t reset,
1962
			unsigned int *classes, unsigned long deadline)
1963
{
1964 1965
	struct ata_device *dev;
	int rc;
1966

T
Tejun Heo 已提交
1967
	ata_link_for_each_dev(dev, link)
1968
		classes[dev->devno] = ATA_DEV_UNKNOWN;
1969

T
Tejun Heo 已提交
1970
	rc = reset(link, classes, deadline);
1971 1972 1973 1974 1975 1976 1977
	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
1978
	ata_link_for_each_dev(dev, link)
1979
		if (classes[dev->devno] != ATA_DEV_UNKNOWN)
1980 1981
			break;

1982
	if (dev) {
T
Tejun Heo 已提交
1983
		ata_link_for_each_dev(dev, link) {
1984 1985 1986 1987
			if (classes[dev->devno] == ATA_DEV_UNKNOWN)
				classes[dev->devno] = ATA_DEV_NONE;
		}
	}
1988 1989 1990 1991

	return 0;
}

1992 1993
static int ata_eh_followup_srst_needed(struct ata_link *link,
				       int rc, int classify,
1994 1995
				       const unsigned int *classes)
{
1996 1997
	if (link->flags & ATA_LFLAG_NO_SRST)
		return 0;
1998 1999 2000 2001
	if (rc == -EAGAIN)
		return 1;
	if (rc != 0)
		return 0;
2002 2003
	if ((link->ap->flags & ATA_FLAG_PMP) && ata_is_host_link(link))
		return 1;
2004 2005
	if (classify && !(link->flags & ATA_LFLAG_ASSUME_CLASS) &&
	    classes[0] == ATA_DEV_UNKNOWN)
2006 2007 2008 2009
		return 1;
	return 0;
}

2010 2011 2012
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
2013
{
2014
	const int max_tries = ARRAY_SIZE(ata_eh_reset_timeouts);
T
Tejun Heo 已提交
2015
	struct ata_port *ap = link->ap;
2016
	struct ata_eh_context *ehc = &link->eh_context;
2017
	unsigned int *classes = ehc->classes;
2018
	unsigned int lflags = link->flags;
2019
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
T
Tejun Heo 已提交
2020
	int try = 0;
2021
	struct ata_device *dev;
2022
	unsigned long deadline, now;
2023
	unsigned int tmp_action;
T
Tejun Heo 已提交
2024
	ata_reset_fn_t reset;
T
Tejun Heo 已提交
2025
	unsigned long flags;
2026
	u32 sstatus;
2027
	int rc;
T
Tejun Heo 已提交
2028

2029
	/* about to reset */
T
Tejun Heo 已提交
2030 2031 2032 2033
	spin_lock_irqsave(ap->lock, flags);
	ap->pflags |= ATA_PFLAG_RESETTING;
	spin_unlock_irqrestore(ap->lock, flags);

T
Tejun Heo 已提交
2034
	ata_eh_about_to_do(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
2035

T
Tejun Heo 已提交
2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054
	ata_link_for_each_dev(dev, link) {
		/* If we issue an SRST then an ATA drive (not ATAPI)
		 * may change configuration and be in PIO0 timing. If
		 * we do a hard reset (or are coming from power on)
		 * this is true for ATA or ATAPI. Until we've set a
		 * suitable controller mode we should not touch the
		 * bus as we may be talking too fast.
		 */
		dev->pio_mode = XFER_PIO_0;

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

2055 2056 2057
	/* Determine which reset to use and record in ehc->i.action.
	 * prereset() may examine and modify it.
	 */
2058
	if (softreset && (!hardreset || (!(lflags & ATA_LFLAG_NO_SRST) &&
2059
					 !sata_set_spd_needed(link) &&
2060 2061
					 !(ehc->i.action & ATA_EH_HARDRESET))))
		tmp_action = ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
2062
	else
2063 2064 2065
		tmp_action = ATA_EH_HARDRESET;

	ehc->i.action = (ehc->i.action & ~ATA_EH_RESET_MASK) | tmp_action;
2066 2067

	if (prereset) {
T
Tejun Heo 已提交
2068
		rc = prereset(link, jiffies + ATA_EH_PRERESET_TIMEOUT);
2069
		if (rc) {
2070
			if (rc == -ENOENT) {
T
Tejun Heo 已提交
2071
				ata_link_printk(link, KERN_DEBUG,
2072
						"port disabled. ignoring.\n");
T
Tejun Heo 已提交
2073
				ehc->i.action &= ~ATA_EH_RESET_MASK;
2074

2075
				ata_link_for_each_dev(dev, link)
2076
					classes[dev->devno] = ATA_DEV_NONE;
2077 2078

				rc = 0;
2079
			} else
T
Tejun Heo 已提交
2080
				ata_link_printk(link, KERN_ERR,
2081
					"prereset failed (errno=%d)\n", rc);
2082
			goto out;
2083 2084 2085 2086 2087
		}
	}

	/* prereset() might have modified ehc->i.action */
	if (ehc->i.action & ATA_EH_HARDRESET)
T
Tejun Heo 已提交
2088
		reset = hardreset;
2089 2090 2091 2092
	else if (ehc->i.action & ATA_EH_SOFTRESET)
		reset = softreset;
	else {
		/* prereset told us not to reset, bang classes and return */
2093
		ata_link_for_each_dev(dev, link)
2094
			classes[dev->devno] = ATA_DEV_NONE;
2095 2096
		rc = 0;
		goto out;
2097 2098 2099 2100 2101 2102 2103 2104 2105
	}

	/* 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
2106 2107

 retry:
T
Tejun Heo 已提交
2108 2109
	deadline = jiffies + ata_eh_reset_timeouts[try++];

2110 2111
	/* shut up during boot probing */
	if (verbose)
T
Tejun Heo 已提交
2112
		ata_link_printk(link, KERN_INFO, "%s resetting link\n",
2113
				reset == softreset ? "soft" : "hard");
T
Tejun Heo 已提交
2114

2115
	/* mark that this EH session started with reset */
2116 2117 2118 2119
	if (reset == hardreset)
		ehc->i.flags |= ATA_EHI_DID_HARDRESET;
	else
		ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
2120

T
Tejun Heo 已提交
2121
	rc = ata_do_reset(link, reset, classes, deadline);
T
Tejun Heo 已提交
2122

2123
	if (reset == hardreset &&
2124
	    ata_eh_followup_srst_needed(link, rc, classify, classes)) {
2125 2126 2127 2128
		/* okay, let's do follow-up softreset */
		reset = softreset;

		if (!reset) {
T
Tejun Heo 已提交
2129
			ata_link_printk(link, KERN_ERR,
2130 2131
					"follow-up softreset required "
					"but no softreset avaliable\n");
2132
			rc = -EINVAL;
2133
			goto fail;
2134 2135
		}

T
Tejun Heo 已提交
2136
		ata_eh_about_to_do(link, NULL, ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
2137
		rc = ata_do_reset(link, reset, classes, deadline);
2138 2139
	}

2140 2141 2142
	/* -EAGAIN can happen if we skipped followup SRST */
	if (rc && rc != -EAGAIN)
		goto fail;
2143

2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158
	/* was classification successful? */
	if (classify && classes[0] == ATA_DEV_UNKNOWN &&
	    !(lflags & ATA_LFLAG_ASSUME_CLASS)) {
		if (try < max_tries) {
			ata_link_printk(link, KERN_WARNING,
					"classification failed\n");
			rc = -EINVAL;
			goto fail;
		}

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

2159 2160 2161 2162 2163 2164 2165
	ata_link_for_each_dev(dev, link) {
		/* After the reset, the device state is PIO 0 and the
		 * controller state is undefined.  Reset also wakes up
		 * drives from sleeping mode.
		 */
		dev->pio_mode = XFER_PIO_0;
		dev->flags &= ~ATA_DFLAG_SLEEPING;
T
Tejun Heo 已提交
2166

2167 2168
		if (ata_link_offline(link))
			continue;
T
Tejun Heo 已提交
2169

2170
		/* apply class override */
2171 2172 2173 2174
		if (lflags & ATA_LFLAG_ASSUME_ATA)
			classes[dev->devno] = ATA_DEV_ATA;
		else if (lflags & ATA_LFLAG_ASSUME_SEMB)
			classes[dev->devno] = ATA_DEV_SEMB_UNSUP; /* not yet */
T
Tejun Heo 已提交
2175 2176
	}

2177 2178 2179
	/* record current link speed */
	if (sata_scr_read(link, SCR_STATUS, &sstatus) == 0)
		link->sata_spd = (sstatus >> 4) & 0xf;
2180

2181 2182
	if (postreset)
		postreset(link, classes);
2183

2184 2185 2186
	/* reset successful, schedule revalidation */
	ata_eh_done(link, NULL, ehc->i.action & ATA_EH_RESET_MASK);
	ehc->i.action |= ATA_EH_REVALIDATE;
2187

2188
	rc = 0;
2189 2190 2191
 out:
	/* clear hotplug flag */
	ehc->i.flags &= ~ATA_EHI_HOTPLUGGED;
T
Tejun Heo 已提交
2192 2193 2194 2195 2196

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

T
Tejun Heo 已提交
2197
	return rc;
2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219

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

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

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

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

	if (rc == -EPIPE || try == max_tries - 1)
		sata_down_spd_limit(link);
	if (hardreset)
		reset = hardreset;
	goto retry;
T
Tejun Heo 已提交
2220 2221
}

2222
static int ata_eh_revalidate_and_attach(struct ata_link *link,
T
Tejun Heo 已提交
2223
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
2224
{
2225 2226
	struct ata_port *ap = link->ap;
	struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2227
	struct ata_device *dev;
2228
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
2229
	unsigned long flags;
2230
	int rc = 0;
T
Tejun Heo 已提交
2231 2232 2233

	DPRINTK("ENTER\n");

2234 2235 2236 2237
	/* 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.
	 */
2238
	ata_link_for_each_dev_reverse(dev, link) {
2239 2240
		unsigned int action = ata_eh_dev_action(dev);
		unsigned int readid_flags = 0;
T
Tejun Heo 已提交
2241

2242 2243 2244
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

2245
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
2246 2247
			WARN_ON(dev->class == ATA_DEV_PMP);

2248
			if (ata_link_offline(link)) {
T
Tejun Heo 已提交
2249
				rc = -EIO;
2250
				goto err;
T
Tejun Heo 已提交
2251 2252
			}

2253
			ata_eh_about_to_do(link, dev, ATA_EH_REVALIDATE);
2254 2255
			rc = ata_dev_revalidate(dev, ehc->classes[dev->devno],
						readid_flags);
T
Tejun Heo 已提交
2256
			if (rc)
2257
				goto err;
T
Tejun Heo 已提交
2258

2259
			ata_eh_done(link, dev, ATA_EH_REVALIDATE);
2260

2261 2262 2263 2264 2265
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

2266 2267
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
2268 2269 2270 2271 2272
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

2273 2274 2275 2276 2277
			if (dev->class == ATA_DEV_PMP)
				rc = sata_pmp_attach(dev);
			else
				rc = ata_dev_read_id(dev, &dev->class,
						     readid_flags, dev->id);
2278 2279
			switch (rc) {
			case 0:
2280
				new_mask |= 1 << dev->devno;
2281 2282
				break;
			case -ENOENT:
2283 2284 2285 2286 2287
				/* 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
2288 2289
				dev->class = ATA_DEV_UNKNOWN;
				break;
2290 2291 2292
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
2293
			}
2294 2295
		}
	}
T
Tejun Heo 已提交
2296

2297
	/* PDIAG- should have been released, ask cable type if post-reset */
2298 2299
	if (ata_is_host_link(link) && ap->ops->cable_detect &&
	    (ehc->i.flags & ATA_EHI_DID_RESET))
2300 2301
		ap->cbl = ap->ops->cable_detect(ap);

2302 2303 2304
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
2305
	ata_link_for_each_dev(dev, link) {
2306 2307
		if (!(new_mask & (1 << dev->devno)) ||
		    dev->class == ATA_DEV_PMP)
2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321
			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
2322 2323
	}

2324
	return 0;
T
Tejun Heo 已提交
2325

2326 2327 2328
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
2329 2330 2331
	return rc;
}

2332
static int ata_link_nr_enabled(struct ata_link *link)
T
Tejun Heo 已提交
2333
{
2334 2335
	struct ata_device *dev;
	int cnt = 0;
T
Tejun Heo 已提交
2336

2337
	ata_link_for_each_dev(dev, link)
2338
		if (ata_dev_enabled(dev))
T
Tejun Heo 已提交
2339 2340 2341 2342
			cnt++;
	return cnt;
}

2343
static int ata_link_nr_vacant(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 (dev->class == ATA_DEV_UNKNOWN)
T
Tejun Heo 已提交
2350 2351 2352 2353
			cnt++;
	return cnt;
}

2354
static int ata_eh_skip_recovery(struct ata_link *link)
T
Tejun Heo 已提交
2355
{
2356
	struct ata_eh_context *ehc = &link->eh_context;
2357
	struct ata_device *dev;
T
Tejun Heo 已提交
2358

2359 2360 2361 2362
	/* skip disabled links */
	if (link->flags & ATA_LFLAG_DISABLED)
		return 1;

2363
	/* thaw frozen port, resume link and recover failed devices */
2364 2365
	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
2366 2367 2368
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
2369
	ata_link_for_each_dev(dev, link) {
T
Tejun Heo 已提交
2370 2371 2372 2373 2374 2375 2376 2377
		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

2378
static int ata_eh_handle_dev_fail(struct ata_device *dev, int err)
2379
{
T
Tejun Heo 已提交
2380
	struct ata_eh_context *ehc = &dev->link->eh_context;
2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391

	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:
2392
		if (ehc->tries[dev->devno] == 1 && dev->pio_mode > XFER_PIO_0) {
2393 2394 2395
			/* This is the last chance, better to slow
			 * down than lose it.
			 */
2396
			sata_down_spd_limit(dev->link);
2397 2398 2399 2400 2401 2402 2403 2404 2405
			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 */
2406
		if (ata_link_offline(dev->link))
2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418
			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;
		}
2419 2420

		return 1;
2421 2422 2423 2424 2425 2426
	} 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;
2427 2428

		return 0;
2429 2430 2431
	}
}

T
Tejun Heo 已提交
2432 2433 2434
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
2435
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2436 2437 2438
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
2439
 *	@r_failed_link: out parameter for failed link
T
Tejun Heo 已提交
2440 2441 2442
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
2443 2444 2445
 *	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
2446
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
2447 2448 2449 2450 2451 2452 2453
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
2454 2455 2456 2457
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
2458
{
2459
	struct ata_link *link;
T
Tejun Heo 已提交
2460
	struct ata_device *dev;
2461 2462
	int nr_failed_devs, nr_disabled_devs;
	int reset, rc;
2463
	unsigned long flags;
T
Tejun Heo 已提交
2464 2465 2466 2467

	DPRINTK("ENTER\n");

	/* prep for recovery */
2468 2469
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2470

2471 2472 2473 2474 2475 2476 2477 2478 2479
		/* 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);
		}

2480
		ata_link_for_each_dev(dev, link) {
2481 2482 2483 2484
			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
2485

2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502
			/* 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
2503
		}
T
Tejun Heo 已提交
2504 2505 2506 2507
	}

 retry:
	rc = 0;
2508 2509 2510
	nr_failed_devs = 0;
	nr_disabled_devs = 0;
	reset = 0;
T
Tejun Heo 已提交
2511

2512
	/* if UNLOADING, finish immediately */
2513
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2514 2515
		goto out;

2516 2517 2518
	/* prep for EH */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2519

2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530
		/* 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
2531

T
Tejun Heo 已提交
2532
	/* reset */
2533
	if (reset) {
2534 2535 2536 2537 2538
		/* if PMP is attached, this function only deals with
		 * downstream links, port should stay thawed.
		 */
		if (!ap->nr_pmp_links)
			ata_eh_freeze_port(ap);
T
Tejun Heo 已提交
2539

2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553
		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
2554 2555
		}

2556 2557
		if (!ap->nr_pmp_links)
			ata_eh_thaw_port(ap);
T
Tejun Heo 已提交
2558 2559
	}

2560 2561 2562
	/* the rest */
	ata_port_for_each_link(link, ap) {
		struct ata_eh_context *ehc = &link->eh_context;
T
Tejun Heo 已提交
2563

2564 2565
		/* revalidate existing devices and attach new ones */
		rc = ata_eh_revalidate_and_attach(link, &dev);
2566
		if (rc)
T
Tejun Heo 已提交
2567 2568
			goto dev_fail;

2569 2570 2571 2572 2573 2574
		/* 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;
		}

2575 2576 2577 2578 2579 2580 2581 2582
		/* 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;
		}

2583 2584 2585 2586
		if (ehc->i.action & ATA_EHI_LPM)
			ata_link_for_each_dev(dev, link)
				ata_dev_enable_pm(dev, ap->pm_policy);

2587 2588 2589
		/* this link is okay now */
		ehc->i.flags = 0;
		continue;
T
Tejun Heo 已提交
2590

2591
dev_fail:
2592 2593 2594
		nr_failed_devs++;
		if (ata_eh_handle_dev_fail(dev, rc))
			nr_disabled_devs++;
T
Tejun Heo 已提交
2595

2596 2597 2598 2599 2600 2601
		if (ap->pflags & ATA_PFLAG_FROZEN) {
			/* PMP reset requires working host port.
			 * Can't retry if it's frozen.
			 */
			if (ap->nr_pmp_links)
				goto out;
2602
			break;
2603
		}
T
Tejun Heo 已提交
2604 2605
	}

2606 2607 2608 2609 2610 2611 2612 2613 2614
	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
2615

2616
		goto retry;
T
Tejun Heo 已提交
2617 2618
	}

2619 2620 2621 2622
 out:
	if (rc && r_failed_link)
		*r_failed_link = link;

T
Tejun Heo 已提交
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636
	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.
 */
2637
void ata_eh_finish(struct ata_port *ap)
T
Tejun Heo 已提交
2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651
{
	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.
2652 2653 2654 2655 2656
			 *
			 * There's no point in retrying invalid
			 * (detected by libata) and non-IO device
			 * errors (rejected by device).  Finish them
			 * immediately.
T
Tejun Heo 已提交
2657
			 */
2658 2659 2660
			if ((qc->err_mask & AC_ERR_INVALID) ||
			    (!(qc->flags & ATA_QCFLAG_IO) &&
			     qc->err_mask == AC_ERR_DEV))
T
Tejun Heo 已提交
2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673
				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);
			}
		}
	}
2674 2675 2676 2677

	/* 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
2678 2679 2680 2681 2682
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2683
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2684 2685 2686 2687 2688 2689 2690 2691 2692
 *	@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).
 */
2693 2694 2695
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
2696
{
2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709
	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
2710 2711
	ata_eh_finish(ap);
}
2712

2713
#ifdef CONFIG_PM
2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
/**
 *	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);

2739 2740 2741 2742 2743
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

2744 2745 2746 2747 2748 2749
	/* 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
2750
	ata_acpi_set_state(ap, PMSG_SUSPEND);
2751
 out:
2752 2753 2754 2755 2756 2757
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
2758
	else if (ap->pflags & ATA_PFLAG_FROZEN)
2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782
		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;
2783
	int rc = 0;
2784 2785 2786 2787 2788 2789 2790 2791 2792 2793

	/* 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);

2794
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
2795

S
Shaohua Li 已提交
2796 2797
	ata_acpi_set_state(ap, PMSG_ON);

2798 2799 2800
	if (ap->ops->port_resume)
		rc = ap->ops->port_resume(ap);

T
Tejun Heo 已提交
2801 2802 2803
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

2804
	/* report result */
2805 2806 2807 2808 2809 2810 2811 2812
	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);
}
2813
#endif /* CONFIG_PM */