Move libata to drivers/ata.

1 files changed, 2246 insertions, 0 deletions

diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c
new file mode 100644
index 000000000000..2c476eee463f
--- /dev/null
+++ b/drivers/ata/libata-eh.c
@@ -0,0 +1,2246 @@
+/*
+ *  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/config.h>
+#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>
+#include "../scsi/scsi_transport_api.h"
+
+#include <linux/libata.h>
+
+#include "libata.h"
+
+static void __ata_port_freeze(struct ata_port *ap);
+static void ata_eh_finish(struct ata_port *ap);
+static void ata_eh_handle_port_suspend(struct ata_port *ap);
+static void ata_eh_handle_port_resume(struct ata_port *ap);
+
+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();
+}
+
+static struct ata_ering_entry * ata_ering_top(struct ata_ering *ering)
+{
+	struct ata_ering_entry *ent = &ering->ring[ering->cursor];
+	if (!ent->err_mask)
+		return NULL;
+	return ent;
+}
+
+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;
+}
+
+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];
+}
+
+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;
+	}
+}
+
+/**
+ *	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.
+ *
+ *	TODO: kill this function once old EH is gone.
+ *
+ *	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;
+	struct ata_port *ap = ata_shost_to_port(host);
+	unsigned long flags;
+	struct ata_queued_cmd *qc;
+	enum scsi_eh_timer_return ret;
+
+	DPRINTK("ENTER\n");
+
+	if (ap->ops->error_handler) {
+		ret = EH_NOT_HANDLED;
+		goto out;
+	}
+
+	ret = EH_HANDLED;
+	spin_lock_irqsave(ap->lock, flags);
+	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;
+	}
+	spin_unlock_irqrestore(ap->lock, flags);
+
+ out:
+	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.
+ */
+void ata_scsi_error(struct Scsi_Host *host)
+{
+	struct ata_port *ap = ata_shost_to_port(host);
+	int i, repeat_cnt = ATA_EH_MAX_REPEAT;
+	unsigned long flags;
+
+	DPRINTK("ENTER\n");
+
+	/* synchronize with port task */
+	ata_port_flush_task(ap);
+
+	/* synchronize with host_set lock and sort out timeouts */
+
+	/* 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;
+
+		spin_lock_irqsave(ap->lock, flags);
+
+		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);
+
+		spin_unlock_irqrestore(ap->lock, flags);
+	} else
+		spin_unlock_wait(ap->lock);
+
+ repeat:
+	/* invoke error handler */
+	if (ap->ops->error_handler) {
+		/* process port resume request */
+		ata_eh_handle_port_resume(ap);
+
+		/* fetch & clear EH info */
+		spin_lock_irqsave(ap->lock, flags);
+
+		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));
+
+		ap->pflags |= ATA_PFLAG_EH_IN_PROGRESS;
+		ap->pflags &= ~ATA_PFLAG_EH_PENDING;
+
+		spin_unlock_irqrestore(ap->lock, flags);
+
+		/* invoke EH, skip if unloading or suspended */
+		if (!(ap->pflags & (ATA_PFLAG_UNLOADING | ATA_PFLAG_SUSPENDED)))
+			ap->ops->error_handler(ap);
+		else
+			ata_eh_finish(ap);
+
+		/* process port suspend request */
+		ata_eh_handle_port_suspend(ap);
+
+		/* Exception might have happend after ->error_handler
+		 * recovered the port but before this point.  Repeat
+		 * EH in such case.
+		 */
+		spin_lock_irqsave(ap->lock, flags);
+
+		if (ap->pflags & ATA_PFLAG_EH_PENDING) {
+			if (--repeat_cnt) {
+				ata_port_printk(ap, KERN_INFO,
+					"EH pending after completion, "
+					"repeating EH (cnt=%d)\n", repeat_cnt);
+				spin_unlock_irqrestore(ap->lock, flags);
+				goto repeat;
+			}
+			ata_port_printk(ap, KERN_ERR, "EH pending after %d "
+					"tries, giving up\n", ATA_EH_MAX_REPEAT);
+		}
+
+		/* this run is complete, make sure EH info is clear */
+		memset(&ap->eh_info, 0, sizeof(ap->eh_info));
+
+		/* Clear host_eh_scheduled while holding ap->lock such
+		 * that if exception occurs after this point but
+		 * before EH completion, SCSI midlayer will
+		 * re-initiate EH.
+		 */
+		host->host_eh_scheduled = 0;
+
+		spin_unlock_irqrestore(ap->lock, flags);
+	} else {
+		WARN_ON(ata_qc_from_tag(ap, ap->active_tag) == NULL);
+		ap->ops->eng_timeout(ap);
+	}
+
+	/* finish or retry handled scmd's and clean up */
+	WARN_ON(host->host_failed || !list_empty(&host->eh_cmd_q));
+
+	scsi_eh_flush_done_q(&ap->eh_done_q);
+
+	/* clean up */
+	spin_lock_irqsave(ap->lock, flags);
+
+	if (ap->pflags & ATA_PFLAG_LOADING)
+		ap->pflags &= ~ATA_PFLAG_LOADING;
+	else if (ap->pflags & ATA_PFLAG_SCSI_HOTPLUG)
+		queue_work(ata_aux_wq, &ap->hotplug_task);
+
+	if (ap->pflags & ATA_PFLAG_RECOVERED)
+		ata_port_printk(ap, KERN_INFO, "EH complete\n");
+
+	ap->pflags &= ~(ATA_PFLAG_SCSI_HOTPLUG | ATA_PFLAG_RECOVERED);
+
+	/* tell wait_eh that we're done */
+	ap->pflags &= ~ATA_PFLAG_EH_IN_PROGRESS;
+	wake_up_all(&ap->eh_wait_q);
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	DPRINTK("EXIT\n");
+}
+
+/**
+ *	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:
+	spin_lock_irqsave(ap->lock, flags);
+
+	while (ap->pflags & (ATA_PFLAG_EH_PENDING | ATA_PFLAG_EH_IN_PROGRESS)) {
+		prepare_to_wait(&ap->eh_wait_q, &wait, TASK_UNINTERRUPTIBLE);
+		spin_unlock_irqrestore(ap->lock, flags);
+		schedule();
+		spin_lock_irqsave(ap->lock, flags);
+	}
+	finish_wait(&ap->eh_wait_q, &wait);
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	/* make sure SCSI EH is complete */
+	if (scsi_host_in_recovery(ap->host)) {
+		msleep(10);
+		goto retry;
+	}
+}
+
+/**
+ *	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.
+ *
+ *	TODO: kill this function once old EH is gone.
+ *
+ *	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;
+
+	spin_lock_irqsave(ap->lock, flags);
+
+	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);
+
+		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);
+
+		/* complete taskfile transaction */
+		qc->err_mask |= AC_ERR_TIMEOUT;
+		break;
+	}
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	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.
+ *
+ *	TODO: kill this function once old EH is gone.
+ *
+ *	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");
+}
+
+/**
+ *	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:
+ *	spin_lock_irqsave(host_set lock)
+ */
+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;
+	qc->ap->pflags |= ATA_PFLAG_EH_PENDING;
+
+	/* 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);
+}
+
+/**
+ *	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:
+ *	spin_lock_irqsave(host_set lock)
+ */
+void ata_port_schedule_eh(struct ata_port *ap)
+{
+	WARN_ON(!ap->ops->error_handler);
+
+	ap->pflags |= ATA_PFLAG_EH_PENDING;
+	scsi_schedule_eh(ap->host);
+
+	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:
+ *	spin_lock_irqsave(host_set lock)
+ *
+ *	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;
+}
+
+/**
+ *	__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:
+ *	spin_lock_irqsave(host_set lock)
+ */
+static void __ata_port_freeze(struct ata_port *ap)
+{
+	WARN_ON(!ap->ops->error_handler);
+
+	if (ap->ops->freeze)
+		ap->ops->freeze(ap);
+
+	ap->pflags |= ATA_PFLAG_FROZEN;
+
+	DPRINTK("ata%u port frozen\n", ap->id);
+}
+
+/**
+ *	ata_port_freeze - abort & freeze port
+ *	@ap: ATA port to freeze
+ *
+ *	Abort and freeze @ap.
+ *
+ *	LOCKING:
+ *	spin_lock_irqsave(host_set lock)
+ *
+ *	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;
+
+	spin_lock_irqsave(ap->lock, flags);
+	__ata_port_freeze(ap);
+	spin_unlock_irqrestore(ap->lock, flags);
+}
+
+/**
+ *	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;
+
+	spin_lock_irqsave(ap->lock, flags);
+
+	ap->pflags &= ~ATA_PFLAG_FROZEN;
+
+	if (ap->ops->thaw)
+		ap->ops->thaw(ap);
+
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	DPRINTK("ata%u port thawed\n", ap->id);
+}
+
+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;
+
+	spin_lock_irqsave(ap->lock, flags);
+	qc->scsidone = ata_eh_scsidone;
+	__ata_qc_complete(qc);
+	WARN_ON(ata_tag_valid(qc->tag));
+	spin_unlock_irqrestore(ap->lock, flags);
+
+	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);
+}
+
+/**
+ *	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);
+
+	spin_lock_irqsave(ap->lock, flags);
+
+	dev->flags &= ~ATA_DFLAG_DETACH;
+
+	if (ata_scsi_offline_dev(dev)) {
+		dev->flags |= ATA_DFLAG_DETACHED;
+		ap->pflags |= ATA_PFLAG_SCSI_HOTPLUG;
+	}
+
+	/* 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);
+
+	spin_unlock_irqrestore(ap->lock, flags);
+}
+
+/**
+ *	ata_eh_about_to_do - about to perform eh_action
+ *	@ap: target ATA port
+ *	@dev: target ATA dev for per-dev action (can be NULL)
+ *	@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.
+ */
+static void ata_eh_about_to_do(struct ata_port *ap, struct ata_device *dev,
+			       unsigned int action)
+{
+	unsigned long flags;
+	struct ata_eh_info *ehi = &ap->eh_info;
+	struct ata_eh_context *ehc = &ap->eh_context;
+
+	spin_lock_irqsave(ap->lock, flags);
+
+	/* 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);
+
+	if (!(ehc->i.flags & ATA_EHI_QUIET))
+		ap->pflags |= ATA_PFLAG_RECOVERED;
+
+	spin_unlock_irqrestore(ap->lock, flags);
+}
+
+/**
+ *	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)
+{
+	/* 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;
+	}
+
+	ata_eh_clear_action(dev, &ap->eh_context.i, action);
+}
+
+/**
+ *	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";
+}
+
+/**
+ *	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;
+}
+
+/**
+ *	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
+ */
+static unsigned int atapi_eh_request_sense(struct ata_device *dev,
+					   unsigned char *sense_buf)
+{
+	struct ata_port *ap = dev->ap;
+	struct ata_taskfile tf;
+	u8 cdb[ATAPI_CDB_LEN];
+
+	DPRINTK("ATAPI request sense\n");
+
+	ata_tf_init(dev, &tf);
+
+	/* FIXME: is this needed? */
+	memset(sense_buf, 0, SCSI_SENSE_BUFFERSIZE);
+
+	/* XXX: why tf_read here? */
+	ap->ops->tf_read(ap, &tf);
+
+	/* fill these in, for the case where they are -not- overwritten */
+	sense_buf[0] = 0x70;
+	sense_buf[2] = tf.feature >> 4;
+
+	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;
+		action |= ATA_EH_SOFTRESET;
+	}
+	if (serror & (SERR_PHYRDY_CHG | SERR_DEV_XCHG))
+		ata_ehi_hotplugged(&ehc->i);
+
+	ehc->i.err_mask |= err_mask;
+	ehc->i.action |= action;
+}
+
+/**
+ *	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 */
+	if (ap->pflags & ATA_PFLAG_FROZEN)
+		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;
+}
+
+/**
+ *	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;
+	}
+
+	if (!(qc->err_mask & AC_ERR_DEV))
+		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:
+		tmp = atapi_eh_request_sense(qc->dev,
+					     qc->scsicmd->sense_buffer);
+		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;
+	}
+
+	if (qc->err_mask & (AC_ERR_HSM | AC_ERR_TIMEOUT | AC_ERR_ATA_BUS))
+		action |= ATA_EH_SOFTRESET;
+
+	return action;
+}
+
+static int ata_eh_categorize_ering_entry(struct ata_ering_entry *ent)
+{
+	if (ent->err_mask & (AC_ERR_ATA_BUS | AC_ERR_TIMEOUT))
+		return 1;
+
+	if (ent->is_io) {
+		if (ent->err_mask & AC_ERR_HSM)
+			return 1;
+		if ((ent->err_mask &
+		     (AC_ERR_DEV|AC_ERR_MEDIA|AC_ERR_INVALID)) == AC_ERR_DEV)
+			return 2;
+	}
+
+	return 0;
+}
+
+struct speed_down_needed_arg {
+	u64 since;
+	int nr_errors[3];
+};
+
+static int speed_down_needed_cb(struct ata_ering_entry *ent, void *void_arg)
+{
+	struct speed_down_needed_arg *arg = void_arg;
+
+	if (ent->timestamp < arg->since)
+		return -1;
+
+	arg->nr_errors[ata_eh_categorize_ering_entry(ent)]++;
+	return 0;
+}
+
+/**
+ *	ata_eh_speed_down_needed - Determine wheter speed down is necessary
+ *	@dev: Device of interest
+ *
+ *	This function examines error ring of @dev and determines
+ *	whether speed down is necessary.  Speed down is necessary if
+ *	there have been more than 3 of Cat-1 errors or 10 of Cat-2
+ *	errors during last 15 minutes.
+ *
+ *	Cat-1 errors are ATA_BUS, TIMEOUT for any command and HSM
+ *	violation for known supported commands.