libata-eh.c 54.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 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 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
/* 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,
};

/* 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 */
};

75
static void __ata_port_freeze(struct ata_port *ap);
76
static void ata_eh_finish(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

88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
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();
}

104
static void ata_ering_clear(struct ata_ering *ering)
105
{
106
	memset(ering, 0, sizeof(*ering));
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
}

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;
}

130 131 132 133 134 135 136
static unsigned int ata_eh_dev_action(struct ata_device *dev)
{
	struct ata_eh_context *ehc = &dev->ap->eh_context;

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

137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
static void ata_eh_clear_action(struct ata_device *dev,
				struct ata_eh_info *ehi, unsigned int action)
{
	int i;

	if (!dev) {
		ehi->action &= ~action;
		for (i = 0; i < ATA_MAX_DEVICES; i++)
			ehi->dev_action[i] &= ~action;
	} 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) {
			for (i = 0; i < ATA_MAX_DEVICES; i++)
				ehi->dev_action[i] |= ehi->action & action;
			ehi->action &= ~action;
		}

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

162 163 164 165 166 167 168 169 170 171 172
/**
 *	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.
 *
173 174
 *	TODO: kill this function once old EH is gone.
 *
175 176 177 178 179 180 181 182 183
 *	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;
184
	struct ata_port *ap = ata_shost_to_port(host);
185 186
	unsigned long flags;
	struct ata_queued_cmd *qc;
187
	enum scsi_eh_timer_return ret;
188 189 190

	DPRINTK("ENTER\n");

191 192 193 194 195 196
	if (ap->ops->error_handler) {
		ret = EH_NOT_HANDLED;
		goto out;
	}

	ret = EH_HANDLED;
197
	spin_lock_irqsave(ap->lock, flags);
198 199 200 201 202 203 204
	qc = ata_qc_from_tag(ap, ap->active_tag);
	if (qc) {
		WARN_ON(qc->scsicmd != cmd);
		qc->flags |= ATA_QCFLAG_EH_SCHEDULED;
		qc->err_mask |= AC_ERR_TIMEOUT;
		ret = EH_NOT_HANDLED;
	}
205
	spin_unlock_irqrestore(ap->lock, flags);
206

207
 out:
208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
	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.
 */
224
void ata_scsi_error(struct Scsi_Host *host)
225
{
226
	struct ata_port *ap = ata_shost_to_port(host);
227 228
	int i, repeat_cnt = ATA_EH_MAX_REPEAT;
	unsigned long flags;
229 230 231

	DPRINTK("ENTER\n");

232
	/* synchronize with port task */
233 234
	ata_port_flush_task(ap);

J
Jeff Garzik 已提交
235
	/* synchronize with host lock and sort out timeouts */
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252

	/* 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;

253
		spin_lock_irqsave(ap->lock, flags);
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291

		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);

292
		spin_unlock_irqrestore(ap->lock, flags);
293
	} else
294
		spin_unlock_wait(ap->lock);
295 296 297 298

 repeat:
	/* invoke error handler */
	if (ap->ops->error_handler) {
299 300 301
		/* process port resume request */
		ata_eh_handle_port_resume(ap);

302
		/* fetch & clear EH info */
303
		spin_lock_irqsave(ap->lock, flags);
304 305 306 307 308

		memset(&ap->eh_context, 0, sizeof(ap->eh_context));
		ap->eh_context.i = ap->eh_info;
		memset(&ap->eh_info, 0, sizeof(ap->eh_info));

309 310
		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
311

312
		spin_unlock_irqrestore(ap->lock, flags);
313

314 315
		/* invoke EH, skip if unloading or suspended */
		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
316 317 318
			ap->ops->error_handler(ap);
		else
			ata_eh_finish(ap);
319

320 321 322
		/* process port suspend request */
		ata_eh_handle_port_suspend(ap);

323 324 325 326
		/* Exception might have happend after ->error_handler
		 * recovered the port but before this point.  Repeat
		 * EH in such case.
		 */
327
		spin_lock_irqsave(ap->lock, flags);
328

329
		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
330 331 332 333
			if (--repeat_cnt) {
				ata_port_printk(ap, KERN_INFO,
					"EH pending after completion, "
					"repeating EH (cnt=%d)\n", repeat_cnt);
334
				spin_unlock_irqrestore(ap->lock, flags);
335 336 337 338
				goto repeat;
			}
			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
					"tries, giving up\n", ATA_EH_MAX_REPEAT);
T
Tejun Heo 已提交
339
			ap->pflags &= ~ATA_PFLAG_EH_PENDING;
340 341
		}

342 343 344
		/* this run is complete, make sure EH info is clear */
		memset(&ap->eh_info, 0, sizeof(ap->eh_info));

345
		/* Clear host_eh_scheduled while holding ap->lock such
346 347 348 349 350 351
		 * that if exception occurs after this point but
		 * before EH completion, SCSI midlayer will
		 * re-initiate EH.
		 */
		host->host_eh_scheduled = 0;

352
		spin_unlock_irqrestore(ap->lock, flags);
353 354 355 356
	} else {
		WARN_ON(ata_qc_from_tag(ap, ap->active_tag) == NULL);
		ap->ops->eng_timeout(ap);
	}
357

358
	/* finish or retry handled scmd's and clean up */
359 360 361 362
	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));

	scsi_eh_flush_done_q(&ap->eh_done_q);

363
	/* clean up */
364
	spin_lock_irqsave(ap->lock, flags);
365

366
	if (ap->pflags & ATA_PFLAG_LOADING)
367
		ap->pflags &= ~ATA_PFLAG_LOADING;
368
	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
369
		queue_delayed_work(ata_aux_wq, &ap->hotplug_task, 0);
370 371 372

	if (ap->pflags & ATA_PFLAG_RECOVERED)
		ata_port_printk(ap, KERN_INFO, "EH complete\n");
373

374
	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
375

376
	/* tell wait_eh that we're done */
377
	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
378 379
	wake_up_all(&ap->eh_wait_q);

380
	spin_unlock_irqrestore(ap->lock, flags);
381

382 383 384
	DPRINTK("EXIT\n");
}

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
/**
 *	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:
400
	spin_lock_irqsave(ap->lock, flags);
401

402
	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
403
		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
404
		spin_unlock_irqrestore(ap->lock, flags);
405
		schedule();
406
		spin_lock_irqsave(ap->lock, flags);
407
	}
408
	finish_wait(&ap->eh_wait_q, &wait);
409

410
	spin_unlock_irqrestore(ap->lock, flags);
411 412

	/* make sure SCSI EH is complete */
J
Jeff Garzik 已提交
413
	if (scsi_host_in_recovery(ap->scsi_host)) {
414 415 416 417 418
		msleep(10);
		goto retry;
	}
}

419 420 421 422 423 424 425 426 427 428 429 430 431 432 433
/**
 *	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.
 *
434 435
 *	TODO: kill this function once old EH is gone.
 *
436 437 438 439 440 441 442 443 444 445 446 447 448
 *	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;

449
	spin_lock_irqsave(ap->lock, flags);
450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468

	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);

469 470 471
		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);
472 473

		/* complete taskfile transaction */
474
		qc->err_mask |= AC_ERR_TIMEOUT;
475 476 477
		break;
	}

478
	spin_unlock_irqrestore(ap->lock, flags);
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499

	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.
 *
500 501
 *	TODO: kill this function once old EH is gone.
 *
502 503 504 505 506 507 508 509 510 511 512 513
 *	LOCKING:
 *	Inherited from SCSI layer (none, can sleep)
 */
void ata_eng_timeout(struct ata_port *ap)
{
	DPRINTK("ENTER\n");

	ata_qc_timeout(ata_qc_from_tag(ap, ap->active_tag));

	DPRINTK("EXIT\n");
}

514 515 516 517 518 519 520 521
/**
 *	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
522
 *	spin_lock_irqsave(host lock)
523 524 525 526 527 528 529 530
 */
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;
531
	qc->ap->pflags |= ATA_PFLAG_EH_PENDING;
532 533 534 535 536 537 538 539 540

	/* 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);
}

541 542 543 544 545 546 547 548
/**
 *	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
549
 *	spin_lock_irqsave(host lock)
550 551 552 553 554
 */
void ata_port_schedule_eh(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

555 556 557
	if (ap->pflags & ATA_PFLAG_INITIALIZING)
		return;

558
	ap->pflags |= ATA_PFLAG_EH_PENDING;
J
Jeff Garzik 已提交
559
	scsi_schedule_eh(ap->scsi_host);
560 561 562 563 564 565 566 567 568 569 570

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

/**
 *	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:
J
Jeff Garzik 已提交
571
 *	spin_lock_irqsave(host lock)
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
 *
 *	RETURNS:
 *	Number of aborted qc's.
 */
int ata_port_abort(struct ata_port *ap)
{
	int tag, nr_aborted = 0;

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

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

		if (qc) {
			qc->flags |= ATA_QCFLAG_FAILED;
			ata_qc_complete(qc);
			nr_aborted++;
		}
	}

	if (!nr_aborted)
		ata_port_schedule_eh(ap);

	return nr_aborted;
}

598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613
/**
 *	__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
614
 *	spin_lock_irqsave(host lock)
615 616 617 618 619 620 621 622
 */
static void __ata_port_freeze(struct ata_port *ap)
{
	WARN_ON(!ap->ops->error_handler);

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

623
	ap->pflags |= ATA_PFLAG_FROZEN;
624

T
Tejun Heo 已提交
625
	DPRINTK("ata%u port frozen\n", ap->print_id);
626 627 628 629 630 631 632 633 634
}

/**
 *	ata_port_freeze - abort & freeze port
 *	@ap: ATA port to freeze
 *
 *	Abort and freeze @ap.
 *
 *	LOCKING:
J
Jeff Garzik 已提交
635
 *	spin_lock_irqsave(host lock)
636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667
 *
 *	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;
}

/**
 *	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;

668
	spin_lock_irqsave(ap->lock, flags);
669
	__ata_port_freeze(ap);
670
	spin_unlock_irqrestore(ap->lock, flags);
671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
}

/**
 *	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;

689
	spin_lock_irqsave(ap->lock, flags);
690

691
	ap->pflags &= ~ATA_PFLAG_FROZEN;
692 693 694 695

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

696
	spin_unlock_irqrestore(ap->lock, flags);
697

T
Tejun Heo 已提交
698
	DPRINTK("ata%u port thawed\n", ap->print_id);
699 700
}

701 702 703 704 705 706 707 708 709 710 711
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;

712
	spin_lock_irqsave(ap->lock, flags);
713 714 715
	qc->scsidone = ata_eh_scsidone;
	__ata_qc_complete(qc);
	WARN_ON(ata_tag_valid(qc->tag));
716
	spin_unlock_irqrestore(ap->lock, flags);
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752

	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
753

754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769
/**
 *	ata_eh_detach_dev - detach ATA device
 *	@dev: ATA device to detach
 *
 *	Detach @dev.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_detach_dev(struct ata_device *dev)
{
	struct ata_port *ap = dev->ap;
	unsigned long flags;

	ata_dev_disable(dev);

770
	spin_lock_irqsave(ap->lock, flags);
771 772 773 774 775

	dev->flags &= ~ATA_DFLAG_DETACH;

	if (ata_scsi_offline_dev(dev)) {
		dev->flags |= ATA_DFLAG_DETACHED;
776
		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
777 778
	}

779 780 781 782
	/* clear per-dev EH actions */
	ata_eh_clear_action(dev, &ap->eh_info, ATA_EH_PERDEV_MASK);
	ata_eh_clear_action(dev, &ap->eh_context.i, ATA_EH_PERDEV_MASK);

783
	spin_unlock_irqrestore(ap->lock, flags);
784 785
}

T
Tejun Heo 已提交
786 787 788
/**
 *	ata_eh_about_to_do - about to perform eh_action
 *	@ap: target ATA port
789
 *	@dev: target ATA dev for per-dev action (can be NULL)
T
Tejun Heo 已提交
790 791 792 793 794 795 796 797 798
 *	@action: action about to be performed
 *
 *	Called just before performing EH actions to clear related bits
 *	in @ap->eh_info such that eh actions are not unnecessarily
 *	repeated.
 *
 *	LOCKING:
 *	None.
 */
799 800
static void ata_eh_about_to_do(struct ata_port *ap, struct ata_device *dev,
			       unsigned int action)
T
Tejun Heo 已提交
801 802
{
	unsigned long flags;
803 804
	struct ata_eh_info *ehi = &ap->eh_info;
	struct ata_eh_context *ehc = &ap->eh_context;
T
Tejun Heo 已提交
805

806
	spin_lock_irqsave(ap->lock, flags);
807

808 809 810 811 812 813 814 815 816 817 818 819 820 821
	/* 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;
	}

	ata_eh_clear_action(dev, ehi, action);
822

823
	if (!(ehc->i.flags & ATA_EHI_QUIET))
824 825
		ap->pflags |= ATA_PFLAG_RECOVERED;

826
	spin_unlock_irqrestore(ap->lock, flags);
T
Tejun Heo 已提交
827 828
}

829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
/**
 *	ata_eh_done - EH action complete
 *	@ap: target ATA port
 *	@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
 *	in @ap->eh_context.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_done(struct ata_port *ap, struct ata_device *dev,
			unsigned int action)
{
844 845 846 847 848 849
	/* if reset is complete, clear all reset actions & reset modifier */
	if (action & ATA_EH_RESET_MASK) {
		action |= ATA_EH_RESET_MASK;
		ap->eh_context.i.flags &= ~ATA_EHI_RESET_MODIFIER_MASK;
	}

850 851 852
	ata_eh_clear_action(dev, &ap->eh_context.i, action);
}

T
Tejun Heo 已提交
853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887
/**
 *	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";
}

888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979
/**
 *	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)
{
	u8 *buf = dev->ap->sector_buf;
	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
980 981 982 983 984 985 986 987 988 989 990 991 992 993
/**
 *	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
 */
994
static unsigned int atapi_eh_request_sense(struct ata_queued_cmd *qc)
T
Tejun Heo 已提交
995
{
996 997
	struct ata_device *dev = qc->dev;
	unsigned char *sense_buf = qc->scsicmd->sense_buffer;
T
Tejun Heo 已提交
998 999 1000 1001 1002 1003 1004 1005 1006
	struct ata_port *ap = dev->ap;
	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);

1007 1008 1009
	/* initialize sense_buf with the error register,
	 * for the case where they are -not- overwritten
	 */
T
Tejun Heo 已提交
1010
	sense_buf[0] = 0x70;
1011 1012
	sense_buf[2] = qc->result_tf.feature >> 4;

J
Jeff Garzik 已提交
1013
	/* some devices time out if garbage left in tf */
1014
	ata_tf_init(dev, &tf);
T
Tejun Heo 已提交
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067

	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
 *	@ap: ATA port to analyze SError for
 *
 *	Analyze SError if available and further determine cause of
 *	failure.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_analyze_serror(struct ata_port *ap)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	u32 serror = ehc->i.serror;
	unsigned int err_mask = 0, action = 0;

	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
1068
		action |= ATA_EH_HARDRESET;
T
Tejun Heo 已提交
1069
	}
T
Tejun Heo 已提交
1070 1071
	if (serror & (SERR_PHYRDY_CHG | SERR_DEV_XCHG))
		ata_ehi_hotplugged(&ehc->i);
T
Tejun Heo 已提交
1072 1073 1074 1075 1076

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

1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
/**
 *	ata_eh_analyze_ncq_error - analyze NCQ error
 *	@ap: ATA port to analyze NCQ error for
 *
 *	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).
 */
static void ata_eh_analyze_ncq_error(struct ata_port *ap)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	struct ata_device *dev = ap->device;
	struct ata_queued_cmd *qc;
	struct ata_taskfile tf;
	int tag, rc;

	/* if frozen, we can't do much */
1098
	if (ap->pflags & ATA_PFLAG_FROZEN)
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
		return;

	/* is it NCQ device error? */
	if (!ap->sactive || !(ehc->i.err_mask & AC_ERR_DEV))
		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) {
		ata_port_printk(ap, KERN_ERR, "failed to read log page 10h "
				"(errno=%d)\n", rc);
		return;
	}

	if (!(ap->sactive & (1 << tag))) {
		ata_port_printk(ap, KERN_ERR, "log page 10h reported "
				"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));
	qc->err_mask |= AC_ERR_DEV;
	ehc->i.err_mask &= ~AC_ERR_DEV;
}

T
Tejun Heo 已提交
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
/**
 *	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;
	}

1163 1164 1165
	if (stat & (ATA_ERR | ATA_DF))
		qc->err_mask |= AC_ERR_DEV;
	else
T
Tejun Heo 已提交
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
		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:
1179
		if (!(qc->ap->pflags & ATA_PFLAG_FROZEN)) {
1180
			tmp = atapi_eh_request_sense(qc);
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192
			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
1193 1194 1195 1196 1197 1198 1199 1200
	}

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

	return action;
}

1201
static int ata_eh_categorize_error(int is_io, unsigned int err_mask)
T
Tejun Heo 已提交
1202
{
1203
	if (err_mask & AC_ERR_ATA_BUS)
T
Tejun Heo 已提交
1204 1205
		return 1;

1206 1207 1208 1209 1210
	if (err_mask & AC_ERR_TIMEOUT)
		return 2;

	if (is_io) {
		if (err_mask & AC_ERR_HSM)
T
Tejun Heo 已提交
1211
			return 2;
1212 1213 1214
		if ((err_mask &
		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
			return 3;
T
Tejun Heo 已提交
1215 1216 1217 1218 1219
	}

	return 0;
}

1220
struct speed_down_verdict_arg {
T
Tejun Heo 已提交
1221
	u64 since;
1222
	int nr_errors[4];
T
Tejun Heo 已提交
1223 1224
};

1225
static int speed_down_verdict_cb(struct ata_ering_entry *ent, void *void_arg)
T
Tejun Heo 已提交
1226
{
1227 1228
	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
1229 1230 1231 1232

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

1233
	arg->nr_errors[cat]++;
T
Tejun Heo 已提交
1234 1235 1236 1237
	return 0;
}

/**
1238
 *	ata_eh_speed_down_verdict - Determine speed down verdict
T
Tejun Heo 已提交
1239 1240 1241
 *	@dev: Device of interest
 *
 *	This function examines error ring of @dev and determines
1242 1243
 *	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
1244
 *
1245
 *	Cat-1 is ATA_BUS error for any command.
T
Tejun Heo 已提交
1246
 *
1247 1248 1249 1250
 *	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
1251 1252
 *	command.
 *
1253 1254 1255 1256 1257 1258 1259 1260 1261
 *	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
1262 1263 1264 1265
 *	LOCKING:
 *	Inherited from caller.
 *
 *	RETURNS:
1266
 *	OR of ATA_EH_SPDN_* flags.
T
Tejun Heo 已提交
1267
 */
1268
static unsigned int ata_eh_speed_down_verdict(struct ata_device *dev)
T
Tejun Heo 已提交
1269
{
1270 1271 1272 1273
	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
1274

1275 1276 1277 1278
	/* 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
1279

1280 1281 1282 1283
	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
1284

1285
	/* scan past 3 mins of error history */
T
Tejun Heo 已提交
1286
	memset(&arg, 0, sizeof(arg));
1287 1288
	arg.since = j64 - min(j64, j5mins);
	ata_ering_map(&dev->ering, speed_down_verdict_cb, &arg);
T
Tejun Heo 已提交
1289

1290 1291
	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
1292

1293
	return verdict;
T
Tejun Heo 已提交
1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310
}

/**
 *	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:
1311
 *	Determined recovery action.
T
Tejun Heo 已提交
1312
 */
1313 1314
static unsigned int ata_eh_speed_down(struct ata_device *dev, int is_io,
				      unsigned int err_mask)
T
Tejun Heo 已提交
1315
{
1316 1317 1318 1319 1320
	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
1321 1322 1323 1324
		return 0;

	/* record error and determine whether speed down is necessary */
	ata_ering_record(&dev->ering, is_io, err_mask);
1325
	verdict = ata_eh_speed_down_verdict(dev);
T
Tejun Heo 已提交
1326

1327 1328 1329 1330 1331 1332 1333 1334 1335
	/* 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
1336

1337 1338 1339 1340 1341 1342 1343
	/* speed down? */
	if (verdict & ATA_EH_SPDN_SPEED_DOWN) {
		/* speed down SATA link speed if possible */
		if (sata_down_spd_limit(dev->ap) == 0) {
			action |= ATA_EH_HARDRESET;
			goto done;
		}
T
Tejun Heo 已提交
1344

1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378
		/* 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) &&
	    (dev->ap->cbl != ATA_CBL_SATA) &&
	    (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
1379 1380

	return 0;
1381 1382 1383 1384
 done:
	/* device has been slowed down, blow error history */
	ata_ering_clear(&dev->ering);
	return action;
T
Tejun Heo 已提交
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
}

/**
 *	ata_eh_autopsy - analyze error and determine recovery action
 *	@ap: ATA port to perform autopsy on
 *
 *	Analyze why @ap failed and determine which recovery action is
 *	needed.  This function also sets more detailed AC_ERR_* values
 *	and fills sense data for ATAPI CHECK SENSE.
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 */
static void ata_eh_autopsy(struct ata_port *ap)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	unsigned int all_err_mask = 0;
	int tag, is_io = 0;
	u32 serror;
	int rc;

	DPRINTK("ENTER\n");

1408 1409 1410
	if (ehc->i.flags & ATA_EHI_NO_AUTOPSY)
		return;

T
Tejun Heo 已提交
1411 1412 1413 1414 1415 1416
	/* obtain and analyze SError */
	rc = sata_scr_read(ap, SCR_ERROR, &serror);
	if (rc == 0) {
		ehc->i.serror |= serror;
		ata_eh_analyze_serror(ap);
	} else if (rc != -EOPNOTSUPP)
1417
		ehc->i.action |= ATA_EH_HARDRESET;
T
Tejun Heo 已提交
1418

1419 1420 1421
	/* analyze NCQ failure */
	ata_eh_analyze_ncq_error(ap);

T
Tejun Heo 已提交
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
	/* 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);

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

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

		/* analyze TF */
1438
		ehc->i.action |= ata_eh_analyze_tf(qc, &qc->result_tf);
T
Tejun Heo 已提交
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451

		/* 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);
1452
			ehc->i.action &= ~ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1453 1454 1455
		}

		/* accumulate error info */
1456
		ehc->i.dev = qc->dev;
T
Tejun Heo 已提交
1457 1458 1459 1460 1461
		all_err_mask |= qc->err_mask;
		if (qc->flags & ATA_QCFLAG_IO)
			is_io = 1;
	}

1462
	/* enforce default EH actions */
1463
	if (ap->pflags & ATA_PFLAG_FROZEN ||
1464
	    all_err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT))
1465
		ehc->i.action |= ATA_EH_SOFTRESET;
1466
	else if (all_err_mask)
1467
		ehc->i.action |= ATA_EH_REVALIDATE;
T
Tejun Heo 已提交
1468

1469
	/* if we have offending qcs and the associated failed device */
1470
	if (ehc->i.dev) {
1471
		/* speed down */
1472 1473
		ehc->i.action |= ata_eh_speed_down(ehc->i.dev, is_io,
						   all_err_mask);
1474 1475

		/* perform per-dev EH action only on the offending device */
1476 1477 1478
		ehc->i.dev_action[ehc->i.dev->devno] |=
			ehc->i.action & ATA_EH_PERDEV_MASK;
		ehc->i.action &= ~ATA_EH_PERDEV_MASK;
1479 1480
	}

T
Tejun Heo 已提交
1481 1482 1483 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 1510 1511 1512 1513 1514 1515 1516 1517
	DPRINTK("EXIT\n");
}

/**
 *	ata_eh_report - report error handling to user
 *	@ap: ATA port EH is going on
 *
 *	Report EH to user.
 *
 *	LOCKING:
 *	None.
 */
static void ata_eh_report(struct ata_port *ap)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	const char *frozen, *desc;
	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);

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

		nr_failed++;
	}

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

	frozen = "";
1518
	if (ap->pflags & ATA_PFLAG_FROZEN)
T
Tejun Heo 已提交
1519 1520 1521
		frozen = " frozen";

	if (ehc->i.dev) {
1522 1523 1524 1525
		ata_dev_printk(ehc->i.dev, KERN_ERR, "exception Emask 0x%x "
			       "SAct 0x%x SErr 0x%x action 0x%x%s\n",
			       ehc->i.err_mask, ap->sactive, ehc->i.serror,
			       ehc->i.action, frozen);
T
Tejun Heo 已提交
1526 1527 1528
		if (desc)
			ata_dev_printk(ehc->i.dev, KERN_ERR, "(%s)\n", desc);
	} else {
1529 1530 1531 1532
		ata_port_printk(ap, KERN_ERR, "exception Emask 0x%x "
				"SAct 0x%x SErr 0x%x action 0x%x%s\n",
				ehc->i.err_mask, ap->sactive, ehc->i.serror,
				ehc->i.action, frozen);
T
Tejun Heo 已提交
1533 1534 1535 1536 1537
		if (desc)
			ata_port_printk(ap, KERN_ERR, "(%s)\n", desc);
	}

	for (tag = 0; tag < ATA_MAX_QUEUE; tag++) {
1538 1539 1540 1541 1542 1543
		static const char *dma_str[] = {
			[DMA_BIDIRECTIONAL]	= "bidi",
			[DMA_TO_DEVICE]		= "out",
			[DMA_FROM_DEVICE]	= "in",
			[DMA_NONE]		= "",
		};
T
Tejun Heo 已提交
1544
		struct ata_queued_cmd *qc = __ata_qc_from_tag(ap, tag);
1545
		struct ata_taskfile *cmd = &qc->tf, *res = &qc->result_tf;
T
Tejun Heo 已提交
1546 1547 1548 1549

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

1550 1551
		ata_dev_printk(qc->dev, KERN_ERR,
			"cmd %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
1552
			"tag %d cdb 0x%x data %u %s\n         "
1553 1554 1555 1556 1557 1558
			"res %02x/%02x:%02x:%02x:%02x:%02x/%02x:%02x:%02x:%02x:%02x/%02x "
			"Emask 0x%x (%s)\n",
			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
1559
			cmd->device, qc->tag, qc->cdb[0], qc->nbytes,
1560
			dma_str[qc->dma_dir],
1561 1562 1563 1564 1565
			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,
			res->device, qc->err_mask, ata_err_string(qc->err_mask));
T
Tejun Heo 已提交
1566 1567 1568
	}
}

1569
static int ata_do_reset(struct ata_port *ap, ata_reset_fn_t reset,
1570
			unsigned int *classes, unsigned long deadline)
1571 1572 1573 1574 1575 1576
{
	int i, rc;

	for (i = 0; i < ATA_MAX_DEVICES; i++)
		classes[i] = ATA_DEV_UNKNOWN;

1577
	rc = reset(ap, classes, deadline);
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596
	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.
	 */
	for (i = 0; i < ATA_MAX_DEVICES; i++)
		if (classes[i] != ATA_DEV_UNKNOWN)
			break;

	if (i < ATA_MAX_DEVICES)
		for (i = 0; i < ATA_MAX_DEVICES; i++)
			if (classes[i] == ATA_DEV_UNKNOWN)
				classes[i] = ATA_DEV_NONE;

	return 0;
}

1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609
static int ata_eh_followup_srst_needed(int rc, int classify,
				       const unsigned int *classes)
{
	if (rc == -EAGAIN)
		return 1;
	if (rc != 0)
		return 0;
	if (classify && classes[0] == ATA_DEV_UNKNOWN)
		return 1;
	return 0;
}

static int ata_eh_reset(struct ata_port *ap, int classify,
1610
			ata_prereset_fn_t prereset, ata_reset_fn_t softreset,
T
Tejun Heo 已提交
1611 1612 1613
			ata_reset_fn_t hardreset, ata_postreset_fn_t postreset)
{
	struct ata_eh_context *ehc = &ap->eh_context;
1614
	unsigned int *classes = ehc->classes;
1615
	int verbose = !(ehc->i.flags & ATA_EHI_QUIET);
T
Tejun Heo 已提交
1616 1617
	int try = 0;
	unsigned long deadline;
1618
	unsigned int action;
T
Tejun Heo 已提交
1619
	ata_reset_fn_t reset;
1620
	int i, rc;
T
Tejun Heo 已提交
1621

1622 1623 1624
	/* about to reset */
	ata_eh_about_to_do(ap, NULL, ehc->i.action & ATA_EH_RESET_MASK);

1625 1626 1627 1628 1629
	/* 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;
T
Tejun Heo 已提交
1630
	if (softreset && (!hardreset || (!sata_set_spd_needed(ap) &&
1631 1632
					 !(action & ATA_EH_HARDRESET))))
		ehc->i.action |= ATA_EH_SOFTRESET;
T
Tejun Heo 已提交
1633
	else
1634 1635 1636
		ehc->i.action |= ATA_EH_HARDRESET;

	if (prereset) {
T
Tejun Heo 已提交
1637
		rc = prereset(ap, jiffies + ATA_EH_PRERESET_TIMEOUT);
1638
		if (rc) {
1639
			if (rc == -ENOENT) {
1640 1641
				ata_port_printk(ap, KERN_DEBUG,
						"port disabled. ignoring.\n");
1642
				ap->eh_context.i.action &= ~ATA_EH_RESET_MASK;
1643 1644 1645 1646 1647

				for (i = 0; i < ATA_MAX_DEVICES; i++)
					classes[i] = ATA_DEV_NONE;

				rc = 0;
1648 1649
			} else
				ata_port_printk(ap, KERN_ERR,
1650 1651 1652 1653 1654 1655 1656
					"prereset failed (errno=%d)\n", rc);
			return rc;
		}
	}

	/* prereset() might have modified ehc->i.action */
	if (ehc->i.action & ATA_EH_HARDRESET)
T
Tejun Heo 已提交
1657
		reset = hardreset;
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673
	else if (ehc->i.action & ATA_EH_SOFTRESET)
		reset = softreset;
	else {
		/* prereset told us not to reset, bang classes and return */
		for (i = 0; i < ATA_MAX_DEVICES; i++)
			classes[i] = ATA_DEV_NONE;
		return 0;
	}

	/* 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
1674 1675

 retry:
T
Tejun Heo 已提交
1676 1677
	deadline = jiffies + ata_eh_reset_timeouts[try++];

1678 1679 1680 1681
	/* shut up during boot probing */
	if (verbose)
		ata_port_printk(ap, KERN_INFO, "%s resetting port\n",
				reset == softreset ? "soft" : "hard");
T
Tejun Heo 已提交
1682

1683
	/* mark that this EH session started with reset */
1684 1685 1686 1687
	if (reset == hardreset)
		ehc->i.flags |= ATA_EHI_DID_HARDRESET;
	else
		ehc->i.flags |= ATA_EHI_DID_SOFTRESET;
T
Tejun Heo 已提交
1688

T
Tejun Heo 已提交
1689
	rc = ata_do_reset(ap, reset, classes, deadline);
T
Tejun Heo 已提交
1690

1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
	if (reset == hardreset &&
	    ata_eh_followup_srst_needed(rc, classify, classes)) {
		/* okay, let's do follow-up softreset */
		reset = softreset;

		if (!reset) {
			ata_port_printk(ap, KERN_ERR,
					"follow-up softreset required "
					"but no softreset avaliable\n");
			return -EINVAL;
		}

1703
		ata_eh_about_to_do(ap, NULL, ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
1704
		rc = ata_do_reset(ap, reset, classes, deadline);
1705 1706 1707 1708 1709 1710 1711 1712 1713

		if (rc == 0 && classify &&
		    classes[0] == ATA_DEV_UNKNOWN) {
			ata_port_printk(ap, KERN_ERR,
					"classification failed\n");
			return -EINVAL;
		}
	}

T
Tejun Heo 已提交
1714 1715
	if (rc && try < ARRAY_SIZE(ata_eh_reset_timeouts)) {
		unsigned long now = jiffies;
1716

T
Tejun Heo 已提交
1717 1718
		if (time_before(now, deadline)) {
			unsigned long delta = deadline - jiffies;
1719

T
Tejun Heo 已提交
1720 1721 1722 1723 1724 1725
			ata_port_printk(ap, KERN_WARNING, "reset failed "
				"(errno=%d), retrying in %u secs\n",
				rc, (jiffies_to_msecs(delta) + 999) / 1000);

			schedule_timeout_uninterruptible(delta);
		}
T
Tejun Heo 已提交
1726

T
Tejun Heo 已提交
1727 1728
		if (reset == hardreset &&
		    try == ARRAY_SIZE(ata_eh_reset_timeouts) - 1)
T
Tejun Heo 已提交
1729 1730 1731 1732 1733 1734 1735
			sata_down_spd_limit(ap);
		if (hardreset)
			reset = hardreset;
		goto retry;
	}

	if (rc == 0) {
1736 1737 1738 1739 1740 1741
		/* After the reset, the device state is PIO 0 and the
		 * controller state is undefined.  Record the mode.
		 */
		for (i = 0; i < ATA_MAX_DEVICES; i++)
			ap->device[i].pio_mode = XFER_PIO_0;

T
Tejun Heo 已提交
1742 1743 1744 1745
		if (postreset)
			postreset(ap, classes);

		/* reset successful, schedule revalidation */
1746
		ata_eh_done(ap, NULL, ehc->i.action & ATA_EH_RESET_MASK);
T
Tejun Heo 已提交
1747 1748 1749 1750 1751 1752
		ehc->i.action |= ATA_EH_REVALIDATE;
	}

	return rc;
}

T
Tejun Heo 已提交
1753 1754
static int ata_eh_revalidate_and_attach(struct ata_port *ap,
					struct ata_device **r_failed_dev)
T
Tejun Heo 已提交
1755 1756 1757
{
	struct ata_eh_context *ehc = &ap->eh_context;
	struct ata_device *dev;
1758
	unsigned int new_mask = 0;
T
Tejun Heo 已提交
1759
	unsigned long flags;
T
Tejun Heo 已提交
1760 1761 1762 1763
	int i, rc = 0;

	DPRINTK("ENTER\n");

1764 1765 1766 1767 1768
	/* 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.
	 */
	for (i = ATA_MAX_DEVICES - 1; i >= 0; i--) {
1769
		unsigned int action, readid_flags = 0;
1770

T
Tejun Heo 已提交
1771
		dev = &ap->device[i];
1772
		action = ata_eh_dev_action(dev);
T
Tejun Heo 已提交
1773

1774 1775 1776
		if (ehc->i.flags & ATA_EHI_DID_RESET)
			readid_flags |= ATA_READID_POSTRESET;

1777
		if ((action & ATA_EH_REVALIDATE) && ata_dev_enabled(dev)) {
T
Tejun Heo 已提交
1778 1779
			if (ata_port_offline(ap)) {
				rc = -EIO;
1780
				goto err;
T
Tejun Heo 已提交
1781 1782
			}

1783
			ata_eh_about_to_do(ap, dev, ATA_EH_REVALIDATE);
1784
			rc = ata_dev_revalidate(dev, readid_flags);
T
Tejun Heo 已提交
1785
			if (rc)
1786
				goto err;
T
Tejun Heo 已提交
1787

1788 1789
			ata_eh_done(ap, dev, ATA_EH_REVALIDATE);

1790 1791 1792 1793 1794
			/* Configuration may have changed, reconfigure
			 * transfer mode.
			 */
			ehc->i.flags |= ATA_EHI_SETMODE;

1795 1796
			/* schedule the scsi_rescan_device() here */
			queue_work(ata_aux_wq, &(ap->scsi_rescan_task));
T
Tejun Heo 已提交
1797 1798 1799 1800 1801
		} else if (dev->class == ATA_DEV_UNKNOWN &&
			   ehc->tries[dev->devno] &&
			   ata_class_enabled(ehc->classes[dev->devno])) {
			dev->class = ehc->classes[dev->devno];

1802 1803
			rc = ata_dev_read_id(dev, &dev->class, readid_flags,
					     dev->id);
1804 1805 1806 1807 1808
			switch (rc) {
			case 0:
				new_mask |= 1 << i;
				break;
			case -ENOENT:
1809 1810 1811 1812 1813
				/* 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
1814 1815
				dev->class = ATA_DEV_UNKNOWN;
				break;
1816 1817 1818
			default:
				dev->class = ATA_DEV_UNKNOWN;
				goto err;
T
Tejun Heo 已提交
1819
			}
1820 1821
		}
	}
T
Tejun Heo 已提交
1822

1823 1824 1825 1826
	/* PDIAG- should have been released, ask cable type if post-reset */
	if ((ehc->i.flags & ATA_EHI_DID_RESET) && ap->ops->cable_detect)
		ap->cbl = ap->ops->cable_detect(ap);

1827 1828 1829 1830 1831
	/* Configure new devices forward such that user doesn't see
	 * device detection messages backwards.
	 */
	for (i = 0; i < ATA_MAX_DEVICES; i++) {
		dev = &ap->device[i];
1832

1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
		if (!(new_mask & (1 << i)))
			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
1848 1849
	}

1850
	return 0;
T
Tejun Heo 已提交
1851

1852 1853 1854
 err:
	*r_failed_dev = dev;
	DPRINTK("EXIT rc=%d\n", rc);
T
Tejun Heo 已提交
1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
	return rc;
}

static int ata_port_nr_enabled(struct ata_port *ap)
{
	int i, cnt = 0;

	for (i = 0; i < ATA_MAX_DEVICES; i++)
		if (ata_dev_enabled(&ap->device[i]))
			cnt++;
	return cnt;
}

T
Tejun Heo 已提交
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
static int ata_port_nr_vacant(struct ata_port *ap)
{
	int i, cnt = 0;

	for (i = 0; i < ATA_MAX_DEVICES; i++)
		if (ap->device[i].class == ATA_DEV_UNKNOWN)
			cnt++;
	return cnt;
}

static int ata_eh_skip_recovery(struct ata_port *ap)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	int i;

1883 1884 1885
	/* thaw frozen port, resume link and recover failed devices */
	if ((ap->pflags & ATA_PFLAG_FROZEN) ||
	    (ehc->i.flags & ATA_EHI_RESUME_LINK) || ata_port_nr_enabled(ap))
T
Tejun Heo 已提交
1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
		return 0;

	/* skip if class codes for all vacant slots are ATA_DEV_NONE */
	for (i = 0; i < ATA_MAX_DEVICES; i++) {
		struct ata_device *dev = &ap->device[i];

		if (dev->class == ATA_DEV_UNKNOWN &&
		    ehc->classes[dev->devno] != ATA_DEV_NONE)
			return 0;
	}

	return 1;
}

1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
static void ata_eh_handle_dev_fail(struct ata_device *dev, int err)
{
	struct ata_port *ap = dev->ap;
	struct ata_eh_context *ehc = &ap->eh_context;

	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.
			 */
			sata_down_spd_limit(ap);
			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 */
		if (ata_port_offline(ap))
			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;
		}
	} 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;
	}
}

T
Tejun Heo 已提交
1951 1952 1953
/**
 *	ata_eh_recover - recover host port after error
 *	@ap: host port to recover
1954
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
1955 1956 1957 1958 1959 1960
 *	@softreset: softreset method (can be NULL)
 *	@hardreset: hardreset method (can be NULL)
 *	@postreset: postreset method (can be NULL)
 *
 *	This is the alpha and omega, eum and yang, heart and soul of
 *	libata exception handling.  On entry, actions required to
T
Tejun Heo 已提交
1961 1962 1963 1964
 *	recover the port and hotplug requests are recorded in
 *	eh_context.  This function executes all the operations with
 *	appropriate retrials and fallbacks to resurrect failed
 *	devices, detach goners and greet newcomers.
T
Tejun Heo 已提交
1965 1966 1967 1968 1969 1970 1971
 *
 *	LOCKING:
 *	Kernel thread context (may sleep).
 *
 *	RETURNS:
 *	0 on success, -errno on failure.
 */
1972 1973
static int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset,
			  ata_reset_fn_t softreset, ata_reset_fn_t hardreset,
T
Tejun Heo 已提交
1974 1975 1976 1977
			  ata_postreset_fn_t postreset)
{
	struct ata_eh_context *ehc = &ap->eh_context;
	struct ata_device *dev;
1978
	int i, rc;
T
Tejun Heo 已提交
1979 1980 1981 1982 1983 1984 1985 1986

	DPRINTK("ENTER\n");

	/* prep for recovery */
	for (i = 0; i < ATA_MAX_DEVICES; i++) {
		dev = &ap->device[i];

		ehc->tries[dev->devno] = ATA_EH_DEV_TRIES;
T
Tejun Heo 已提交
1987

1988 1989 1990 1991
		/* collect port action mask recorded in dev actions */
		ehc->i.action |= ehc->i.dev_action[i] & ~ATA_EH_PERDEV_MASK;
		ehc->i.dev_action[i] &= ATA_EH_PERDEV_MASK;

T
Tejun Heo 已提交
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
		/* 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
2004 2005 2006 2007 2008
	}

 retry:
	rc = 0;

2009
	/* if UNLOADING, finish immediately */
2010
	if (ap->pflags & ATA_PFLAG_UNLOADING)
2011 2012
		goto out;

T
Tejun Heo 已提交
2013
	/* skip EH if possible. */
T
Tejun Heo 已提交
2014
	if (ata_eh_skip_recovery(ap))
T
Tejun Heo 已提交
2015 2016
		ehc->i.action = 0;

T
Tejun Heo 已提交
2017 2018 2019
	for (i = 0; i < ATA_MAX_DEVICES; i++)
		ehc->classes[i] = ATA_DEV_UNKNOWN;

T
Tejun Heo 已提交
2020 2021 2022 2023
	/* reset */
	if (ehc->i.action & ATA_EH_RESET_MASK) {
		ata_eh_freeze_port(ap);

T
Tejun Heo 已提交
2024 2025
		rc = ata_eh_reset(ap, ata_port_nr_vacant(ap), prereset,
				  softreset, hardreset, postreset);
T
Tejun Heo 已提交
2026 2027 2028 2029 2030 2031 2032 2033 2034
		if (rc) {
			ata_port_printk(ap, KERN_ERR,
					"reset failed, giving up\n");
			goto out;
		}

		ata_eh_thaw_port(ap);
	}

T
Tejun Heo 已提交
2035 2036
	/* revalidate existing devices and attach new ones */
	rc = ata_eh_revalidate_and_attach(ap, &dev);
T
Tejun Heo 已提交
2037 2038 2039
	if (rc)
		goto dev_fail;

2040 2041
	/* configure transfer mode if necessary */
	if (ehc->i.flags & ATA_EHI_SETMODE) {
T
Tejun Heo 已提交
2042
		rc = ata_set_mode(ap, &dev);
2043
		if (rc)
T
Tejun Heo 已提交
2044
			goto dev_fail;
2045
		ehc->i.flags &= ~ATA_EHI_SETMODE;
T
Tejun Heo 已提交
2046 2047 2048 2049 2050
	}

	goto out;

 dev_fail:
2051
	ata_eh_handle_dev_fail(dev, rc);
T
Tejun Heo 已提交
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118

	if (ata_port_nr_enabled(ap)) {
		ata_port_printk(ap, KERN_WARNING, "failed to recover some "
				"devices, retrying in 5 secs\n");
		ssleep(5);
	} else {
		/* no device left, repeat fast */
		msleep(500);
	}

	goto retry;

 out:
	if (rc) {
		for (i = 0; i < ATA_MAX_DEVICES; i++)
			ata_dev_disable(&ap->device[i]);
	}

	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.
 */
static void ata_eh_finish(struct ata_port *ap)
{
	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);
			}
		}
	}
}

/**
 *	ata_do_eh - do standard error handling
 *	@ap: host port to handle error for
2119
 *	@prereset: prereset method (can be NULL)
T
Tejun Heo 已提交
2120 2121 2122 2123 2124 2125 2126 2127 2128
 *	@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).
 */
2129 2130 2131
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
2132
{
2133 2134
	ata_eh_autopsy(ap);
	ata_eh_report(ap);
2135
	ata_eh_recover(ap, prereset, softreset, hardreset, postreset);
T
Tejun Heo 已提交
2136 2137
	ata_eh_finish(ap);
}
2138

2139
#ifdef CONFIG_PM
2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164
/**
 *	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);

2165 2166 2167 2168 2169
	/* tell ACPI we're suspending */
	rc = ata_acpi_on_suspend(ap);
	if (rc)
		goto out;

2170 2171 2172 2173 2174 2175
	/* suspend */
	ata_eh_freeze_port(ap);

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

2176
 out:
2177 2178 2179 2180 2181 2182
	/* report result */
	spin_lock_irqsave(ap->lock, flags);

	ap->pflags &= ~ATA_PFLAG_PM_PENDING;
	if (rc == 0)
		ap->pflags |= ATA_PFLAG_SUSPENDED;
2183
	else if (ap->pflags & ATA_PFLAG_FROZEN)
2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207
		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;
2208
	int rc = 0;
2209 2210 2211 2212 2213 2214 2215 2216 2217 2218

	/* 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);

2219
	WARN_ON(!(ap->pflags & ATA_PFLAG_SUSPENDED));
2220 2221 2222 2223

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

T
Tejun Heo 已提交
2224 2225 2226
	/* tell ACPI that we're resuming */
	ata_acpi_on_resume(ap);

2227
	/* report result */
2228 2229 2230 2231 2232 2233 2234 2235
	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);
}
2236
#endif /* CONFIG_PM */