xref: /openbmc/linux/drivers/scsi/isci/host.c (revision 2396a2650a5a39634e3ad6b29e1104944e5ab88f)

/*
 * This file is provided under a dual BSD/GPLv2 license.  When using or
 * redistributing this file, you may do so under either license.
 *
 * GPL LICENSE SUMMARY
 *
 * Copyright(c) 2008 - 2011 Intel Corporation. All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * 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; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
 * The full GNU General Public License is included in this distribution
 * in the file called LICENSE.GPL.
 *
 * BSD LICENSE
 *
 * Copyright(c) 2008 - 2011 Intel Corporation. All rights reserved.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 *   * Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *   * Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in
 *     the documentation and/or other materials provided with the
 *     distribution.
 *   * Neither the name of Intel Corporation nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
#include <linux/circ_buf.h>
#include <linux/device.h>
#include <scsi/sas.h>
#include "host.h"
#include "isci.h"
#include "port.h"
#include "probe_roms.h"
#include "remote_device.h"
#include "request.h"
#include "scu_completion_codes.h"
#include "scu_event_codes.h"
#include "registers.h"
#include "scu_remote_node_context.h"
#include "scu_task_context.h"

#define SCU_CONTEXT_RAM_INIT_STALL_TIME      200

#define smu_max_ports(dcc_value) \
	(\
		(((dcc_value) & SMU_DEVICE_CONTEXT_CAPACITY_MAX_LP_MASK) \
		 >> SMU_DEVICE_CONTEXT_CAPACITY_MAX_LP_SHIFT) + 1 \
	)

#define smu_max_task_contexts(dcc_value)	\
	(\
		(((dcc_value) & SMU_DEVICE_CONTEXT_CAPACITY_MAX_TC_MASK) \
		 >> SMU_DEVICE_CONTEXT_CAPACITY_MAX_TC_SHIFT) + 1 \
	)

#define smu_max_rncs(dcc_value) \
	(\
		(((dcc_value) & SMU_DEVICE_CONTEXT_CAPACITY_MAX_RNC_MASK) \
		 >> SMU_DEVICE_CONTEXT_CAPACITY_MAX_RNC_SHIFT) + 1 \
	)

#define SCIC_SDS_CONTROLLER_PHY_START_TIMEOUT      100

/**
 *
 *
 * The number of milliseconds to wait while a given phy is consuming power
 * before allowing another set of phys to consume power. Ultimately, this will
 * be specified by OEM parameter.
 */
#define SCIC_SDS_CONTROLLER_POWER_CONTROL_INTERVAL 500

/**
 * NORMALIZE_PUT_POINTER() -
 *
 * This macro will normalize the completion queue put pointer so its value can
 * be used as an array inde
 */
#define NORMALIZE_PUT_POINTER(x) \
	((x) & SMU_COMPLETION_QUEUE_PUT_POINTER_MASK)


/**
 * NORMALIZE_EVENT_POINTER() -
 *
 * This macro will normalize the completion queue event entry so its value can
 * be used as an index.
 */
#define NORMALIZE_EVENT_POINTER(x) \
	(\
		((x) & SMU_COMPLETION_QUEUE_GET_EVENT_POINTER_MASK) \
		>> SMU_COMPLETION_QUEUE_GET_EVENT_POINTER_SHIFT	\
	)

/**
 * NORMALIZE_GET_POINTER() -
 *
 * This macro will normalize the completion queue get pointer so its value can
 * be used as an index into an array
 */
#define NORMALIZE_GET_POINTER(x) \
	((x) & SMU_COMPLETION_QUEUE_GET_POINTER_MASK)

/**
 * NORMALIZE_GET_POINTER_CYCLE_BIT() -
 *
 * This macro will normalize the completion queue cycle pointer so it matches
 * the completion queue cycle bit
 */
#define NORMALIZE_GET_POINTER_CYCLE_BIT(x) \
	((SMU_CQGR_CYCLE_BIT & (x)) << (31 - SMU_COMPLETION_QUEUE_GET_CYCLE_BIT_SHIFT))

/**
 * COMPLETION_QUEUE_CYCLE_BIT() -
 *
 * This macro will return the cycle bit of the completion queue entry
 */
#define COMPLETION_QUEUE_CYCLE_BIT(x) ((x) & 0x80000000)

/* Init the state machine and call the state entry function (if any) */
void sci_init_sm(struct sci_base_state_machine *sm,
		 const struct sci_base_state *state_table, u32 initial_state)
{
	sci_state_transition_t handler;

	sm->initial_state_id    = initial_state;
	sm->previous_state_id   = initial_state;
	sm->current_state_id    = initial_state;
	sm->state_table         = state_table;

	handler = sm->state_table[initial_state].enter_state;
	if (handler)
		handler(sm);
}

/* Call the state exit fn, update the current state, call the state entry fn */
void sci_change_state(struct sci_base_state_machine *sm, u32 next_state)
{
	sci_state_transition_t handler;

	handler = sm->state_table[sm->current_state_id].exit_state;
	if (handler)
		handler(sm);

	sm->previous_state_id = sm->current_state_id;
	sm->current_state_id = next_state;

	handler = sm->state_table[sm->current_state_id].enter_state;
	if (handler)
		handler(sm);
}

static bool sci_controller_completion_queue_has_entries(struct isci_host *ihost)
{
	u32 get_value = ihost->completion_queue_get;
	u32 get_index = get_value & SMU_COMPLETION_QUEUE_GET_POINTER_MASK;

	if (NORMALIZE_GET_POINTER_CYCLE_BIT(get_value) ==
	    COMPLETION_QUEUE_CYCLE_BIT(ihost->completion_queue[get_index]))
		return true;

	return false;
}

static bool sci_controller_isr(struct isci_host *ihost)
{
	if (sci_controller_completion_queue_has_entries(ihost))
		return true;

	/* we have a spurious interrupt it could be that we have already
	 * emptied the completion queue from a previous interrupt
	 * FIXME: really!?
	 */
	writel(SMU_ISR_COMPLETION, &ihost->smu_registers->interrupt_status);

	/* There is a race in the hardware that could cause us not to be
	 * notified of an interrupt completion if we do not take this
	 * step.  We will mask then unmask the interrupts so if there is
	 * another interrupt pending the clearing of the interrupt
	 * source we get the next interrupt message.
	 */
	spin_lock(&ihost->scic_lock);
	if (test_bit(IHOST_IRQ_ENABLED, &ihost->flags)) {
		writel(0xFF000000, &ihost->smu_registers->interrupt_mask);
		writel(0, &ihost->smu_registers->interrupt_mask);
	}
	spin_unlock(&ihost->scic_lock);

	return false;
}

irqreturn_t isci_msix_isr(int vec, void *data)
{
	struct isci_host *ihost = data;

	if (sci_controller_isr(ihost))
		tasklet_schedule(&ihost->completion_tasklet);

	return IRQ_HANDLED;
}

static bool sci_controller_error_isr(struct isci_host *ihost)
{
	u32 interrupt_status;

	interrupt_status =
		readl(&ihost->smu_registers->interrupt_status);
	interrupt_status &= (SMU_ISR_QUEUE_ERROR | SMU_ISR_QUEUE_SUSPEND);

	if (interrupt_status != 0) {
		/*
		 * There is an error interrupt pending so let it through and handle
		 * in the callback */
		return true;
	}

	/*
	 * There is a race in the hardware that could cause us not to be notified
	 * of an interrupt completion if we do not take this step.  We will mask
	 * then unmask the error interrupts so if there was another interrupt
	 * pending we will be notified.
	 * Could we write the value of (SMU_ISR_QUEUE_ERROR | SMU_ISR_QUEUE_SUSPEND)? */
	writel(0xff, &ihost->smu_registers->interrupt_mask);
	writel(0, &ihost->smu_registers->interrupt_mask);

	return false;
}

static void sci_controller_task_completion(struct isci_host *ihost, u32 ent)
{
	u32 index = SCU_GET_COMPLETION_INDEX(ent);
	struct isci_request *ireq = ihost->reqs[index];

	/* Make sure that we really want to process this IO request */
	if (test_bit(IREQ_ACTIVE, &ireq->flags) &&
	    ireq->io_tag != SCI_CONTROLLER_INVALID_IO_TAG &&
	    ISCI_TAG_SEQ(ireq->io_tag) == ihost->io_request_sequence[index])
		/* Yep this is a valid io request pass it along to the
		 * io request handler
		 */
		sci_io_request_tc_completion(ireq, ent);
}

static void sci_controller_sdma_completion(struct isci_host *ihost, u32 ent)
{
	u32 index;
	struct isci_request *ireq;
	struct isci_remote_device *idev;

	index = SCU_GET_COMPLETION_INDEX(ent);

	switch (scu_get_command_request_type(ent)) {
	case SCU_CONTEXT_COMMAND_REQUEST_TYPE_POST_TC:
	case SCU_CONTEXT_COMMAND_REQUEST_TYPE_DUMP_TC:
		ireq = ihost->reqs[index];
		dev_warn(&ihost->pdev->dev, "%s: %x for io request %p\n",
			 __func__, ent, ireq);
		/* @todo For a post TC operation we need to fail the IO
		 * request
		 */
		break;
	case SCU_CONTEXT_COMMAND_REQUEST_TYPE_DUMP_RNC:
	case SCU_CONTEXT_COMMAND_REQUEST_TYPE_OTHER_RNC:
	case SCU_CONTEXT_COMMAND_REQUEST_TYPE_POST_RNC:
		idev = ihost->device_table[index];
		dev_warn(&ihost->pdev->dev, "%s: %x for device %p\n",
			 __func__, ent, idev);
		/* @todo For a port RNC operation we need to fail the
		 * device
		 */
		break;
	default:
		dev_warn(&ihost->pdev->dev, "%s: unknown completion type %x\n",
			 __func__, ent);
		break;
	}
}

static void sci_controller_unsolicited_frame(struct isci_host *ihost, u32 ent)
{
	u32 index;
	u32 frame_index;

	struct scu_unsolicited_frame_header *frame_header;
	struct isci_phy *iphy;
	struct isci_remote_device *idev;

	enum sci_status result = SCI_FAILURE;

	frame_index = SCU_GET_FRAME_INDEX(ent);

	frame_header = ihost->uf_control.buffers.array[frame_index].header;
	ihost->uf_control.buffers.array[frame_index].state = UNSOLICITED_FRAME_IN_USE;

	if (SCU_GET_FRAME_ERROR(ent)) {
		/*
		 * / @todo If the IAF frame or SIGNATURE FIS frame has an error will
		 * /       this cause a problem? We expect the phy initialization will
		 * /       fail if there is an error in the frame. */
		sci_controller_release_frame(ihost, frame_index);
		return;
	}

	if (frame_header->is_address_frame) {
		index = SCU_GET_PROTOCOL_ENGINE_INDEX(ent);
		iphy = &ihost->phys[index];
		result = sci_phy_frame_handler(iphy, frame_index);
	} else {

		index = SCU_GET_COMPLETION_INDEX(ent);

		if (index == SCIC_SDS_REMOTE_NODE_CONTEXT_INVALID_INDEX) {
			/*
			 * This is a signature fis or a frame from a direct attached SATA
			 * device that has not yet been created.  In either case forwared
			 * the frame to the PE and let it take care of the frame data. */
			index = SCU_GET_PROTOCOL_ENGINE_INDEX(ent);
			iphy = &ihost->phys[index];
			result = sci_phy_frame_handler(iphy, frame_index);
		} else {
			if (index < ihost->remote_node_entries)
				idev = ihost->device_table[index];
			else
				idev = NULL;

			if (idev != NULL)
				result = sci_remote_device_frame_handler(idev, frame_index);
			else
				sci_controller_release_frame(ihost, frame_index);
		}
	}

	if (result != SCI_SUCCESS) {
		/*
		 * / @todo Is there any reason to report some additional error message
		 * /       when we get this failure notifiction? */
	}
}

static void sci_controller_event_completion(struct isci_host *ihost, u32 ent)
{
	struct isci_remote_device *idev;
	struct isci_request *ireq;
	struct isci_phy *iphy;
	u32 index;

	index = SCU_GET_COMPLETION_INDEX(ent);

	switch (scu_get_event_type(ent)) {
	case SCU_EVENT_TYPE_SMU_COMMAND_ERROR:
		/* / @todo The driver did something wrong and we need to fix the condtion. */
		dev_err(&ihost->pdev->dev,
			"%s: SCIC Controller 0x%p received SMU command error "
			"0x%x\n",
			__func__,
			ihost,
			ent);
		break;

	case SCU_EVENT_TYPE_SMU_PCQ_ERROR:
	case SCU_EVENT_TYPE_SMU_ERROR:
	case SCU_EVENT_TYPE_FATAL_MEMORY_ERROR:
		/*
		 * / @todo This is a hardware failure and its likely that we want to
		 * /       reset the controller. */
		dev_err(&ihost->pdev->dev,
			"%s: SCIC Controller 0x%p received fatal controller "
			"event  0x%x\n",
			__func__,
			ihost,
			ent);
		break;

	case SCU_EVENT_TYPE_TRANSPORT_ERROR:
		ireq = ihost->reqs[index];
		sci_io_request_event_handler(ireq, ent);
		break;

	case SCU_EVENT_TYPE_PTX_SCHEDULE_EVENT:
		switch (scu_get_event_specifier(ent)) {
		case SCU_EVENT_SPECIFIC_SMP_RESPONSE_NO_PE:
		case SCU_EVENT_SPECIFIC_TASK_TIMEOUT:
			ireq = ihost->reqs[index];
			if (ireq != NULL)
				sci_io_request_event_handler(ireq, ent);
			else
				dev_warn(&ihost->pdev->dev,
					 "%s: SCIC Controller 0x%p received "
					 "event 0x%x for io request object "
					 "that doesnt exist.\n",
					 __func__,
					 ihost,
					 ent);

			break;

		case SCU_EVENT_SPECIFIC_IT_NEXUS_TIMEOUT:
			idev = ihost->device_table[index];
			if (idev != NULL)
				sci_remote_device_event_handler(idev, ent);
			else
				dev_warn(&ihost->pdev->dev,
					 "%s: SCIC Controller 0x%p received "
					 "event 0x%x for remote device object "
					 "that doesnt exist.\n",
					 __func__,
					 ihost,
					 ent);

			break;
		}
		break;

	case SCU_EVENT_TYPE_BROADCAST_CHANGE:
	/*
	 * direct the broadcast change event to the phy first and then let
	 * the phy redirect the broadcast change to the port object */
	case SCU_EVENT_TYPE_ERR_CNT_EVENT:
	/*
	 * direct error counter event to the phy object since that is where
	 * we get the event notification.  This is a type 4 event. */
	case SCU_EVENT_TYPE_OSSP_EVENT:
		index = SCU_GET_PROTOCOL_ENGINE_INDEX(ent);
		iphy = &ihost->phys[index];
		sci_phy_event_handler(iphy, ent);
		break;

	case SCU_EVENT_TYPE_RNC_SUSPEND_TX:
	case SCU_EVENT_TYPE_RNC_SUSPEND_TX_RX:
	case SCU_EVENT_TYPE_RNC_OPS_MISC:
		if (index < ihost->remote_node_entries) {
			idev = ihost->device_table[index];

			if (idev != NULL)
				sci_remote_device_event_handler(idev, ent);
		} else
			dev_err(&ihost->pdev->dev,
				"%s: SCIC Controller 0x%p received event 0x%x "
				"for remote device object 0x%0x that doesnt "
				"exist.\n",
				__func__,
				ihost,
				ent,
				index);

		break;

	default:
		dev_warn(&ihost->pdev->dev,
			 "%s: SCIC Controller received unknown event code %x\n",
			 __func__,
			 ent);
		break;
	}
}

static void sci_controller_process_completions(struct isci_host *ihost)
{
	u32 completion_count = 0;
	u32 ent;
	u32 get_index;
	u32 get_cycle;
	u32 event_get;
	u32 event_cycle;

	dev_dbg(&ihost->pdev->dev,
		"%s: completion queue begining get:0x%08x\n",
		__func__,
		ihost->completion_queue_get);

	/* Get the component parts of the completion queue */
	get_index = NORMALIZE_GET_POINTER(ihost->completion_queue_get);
	get_cycle = SMU_CQGR_CYCLE_BIT & ihost->completion_queue_get;

	event_get = NORMALIZE_EVENT_POINTER(ihost->completion_queue_get);
	event_cycle = SMU_CQGR_EVENT_CYCLE_BIT & ihost->completion_queue_get;

	while (
		NORMALIZE_GET_POINTER_CYCLE_BIT(get_cycle)
		== COMPLETION_QUEUE_CYCLE_BIT(ihost->completion_queue[get_index])
		) {
		completion_count++;

		ent = ihost->completion_queue[get_index];

		/* increment the get pointer and check for rollover to toggle the cycle bit */
		get_cycle ^= ((get_index+1) & SCU_MAX_COMPLETION_QUEUE_ENTRIES) <<
			     (SMU_COMPLETION_QUEUE_GET_CYCLE_BIT_SHIFT - SCU_MAX_COMPLETION_QUEUE_SHIFT);
		get_index = (get_index+1) & (SCU_MAX_COMPLETION_QUEUE_ENTRIES-1);

		dev_dbg(&ihost->pdev->dev,
			"%s: completion queue entry:0x%08x\n",
			__func__,
			ent);

		switch (SCU_GET_COMPLETION_TYPE(ent)) {
		case SCU_COMPLETION_TYPE_TASK:
			sci_controller_task_completion(ihost, ent);
			break;

		case SCU_COMPLETION_TYPE_SDMA:
			sci_controller_sdma_completion(ihost, ent);
			break;

		case SCU_COMPLETION_TYPE_UFI:
			sci_controller_unsolicited_frame(ihost, ent);
			break;

		case SCU_COMPLETION_TYPE_EVENT:
			sci_controller_event_completion(ihost, ent);
			break;

		case SCU_COMPLETION_TYPE_NOTIFY: {
			event_cycle ^= ((event_get+1) & SCU_MAX_EVENTS) <<
				       (SMU_COMPLETION_QUEUE_GET_EVENT_CYCLE_BIT_SHIFT - SCU_MAX_EVENTS_SHIFT);
			event_get = (event_get+1) & (SCU_MAX_EVENTS-1);

			sci_controller_event_completion(ihost, ent);
			break;
		}
		default:
			dev_warn(&ihost->pdev->dev,
				 "%s: SCIC Controller received unknown "
				 "completion type %x\n",
				 __func__,
				 ent);
			break;
		}
	}

	/* Update the get register if we completed one or more entries */
	if (completion_count > 0) {
		ihost->completion_queue_get =
			SMU_CQGR_GEN_BIT(ENABLE) |
			SMU_CQGR_GEN_BIT(EVENT_ENABLE) |
			event_cycle |
			SMU_CQGR_GEN_VAL(EVENT_POINTER, event_get) |
			get_cycle |
			SMU_CQGR_GEN_VAL(POINTER, get_index);

		writel(ihost->completion_queue_get,
		       &ihost->smu_registers->completion_queue_get);

	}

	dev_dbg(&ihost->pdev->dev,
		"%s: completion queue ending get:0x%08x\n",
		__func__,
		ihost->completion_queue_get);

}

static void sci_controller_error_handler(struct isci_host *ihost)
{
	u32 interrupt_status;

	interrupt_status =
		readl(&ihost->smu_registers->interrupt_status);

	if ((interrupt_status & SMU_ISR_QUEUE_SUSPEND) &&
	    sci_controller_completion_queue_has_entries(ihost)) {

		sci_controller_process_completions(ihost);
		writel(SMU_ISR_QUEUE_SUSPEND, &ihost->smu_registers->interrupt_status);
	} else {
		dev_err(&ihost->pdev->dev, "%s: status: %#x\n", __func__,
			interrupt_status);

		sci_change_state(&ihost->sm, SCIC_FAILED);

		return;
	}

	/* If we dont process any completions I am not sure that we want to do this.
	 * We are in the middle of a hardware fault and should probably be reset.
	 */
	writel(0, &ihost->smu_registers->interrupt_mask);
}

irqreturn_t isci_intx_isr(int vec, void *data)
{
	irqreturn_t ret = IRQ_NONE;
	struct isci_host *ihost = data;

	if (sci_controller_isr(ihost)) {
		writel(SMU_ISR_COMPLETION, &ihost->smu_registers->interrupt_status);
		tasklet_schedule(&ihost->completion_tasklet);
		ret = IRQ_HANDLED;
	} else if (sci_controller_error_isr(ihost)) {
		spin_lock(&ihost->scic_lock);
		sci_controller_error_handler(ihost);
		spin_unlock(&ihost->scic_lock);
		ret = IRQ_HANDLED;
	}

	return ret;
}

irqreturn_t isci_error_isr(int vec, void *data)
{
	struct isci_host *ihost = data;

	if (sci_controller_error_isr(ihost))
		sci_controller_error_handler(ihost);

	return IRQ_HANDLED;
}

/**
 * isci_host_start_complete() - This function is called by the core library,
 *    through the ISCI Module, to indicate controller start status.
 * @isci_host: This parameter specifies the ISCI host object
 * @completion_status: This parameter specifies the completion status from the
 *    core library.
 *
 */
static void isci_host_start_complete(struct isci_host *ihost, enum sci_status completion_status)
{
	if (completion_status != SCI_SUCCESS)
		dev_info(&ihost->pdev->dev,
			"controller start timed out, continuing...\n");
	clear_bit(IHOST_START_PENDING, &ihost->flags);
	wake_up(&ihost->eventq);
}

int isci_host_scan_finished(struct Scsi_Host *shost, unsigned long time)
{
	struct sas_ha_struct *ha = SHOST_TO_SAS_HA(shost);
	struct isci_host *ihost = ha->lldd_ha;

	if (test_bit(IHOST_START_PENDING, &ihost->flags))
		return 0;

	sas_drain_work(ha);

	return 1;
}

/**
 * sci_controller_get_suggested_start_timeout() - This method returns the
 *    suggested sci_controller_start() timeout amount.  The user is free to
 *    use any timeout value, but this method provides the suggested minimum
 *    start timeout value.  The returned value is based upon empirical
 *    information determined as a result of interoperability testing.
 * @controller: the handle to the controller object for which to return the
 *    suggested start timeout.
 *
 * This method returns the number of milliseconds for the suggested start
 * operation timeout.
 */
static u32 sci_controller_get_suggested_start_timeout(struct isci_host *ihost)
{
	/* Validate the user supplied parameters. */
	if (!ihost)
		return 0;

	/*
	 * The suggested minimum timeout value for a controller start operation:
	 *
	 *     Signature FIS Timeout
	 *   + Phy Start Timeout
	 *   + Number of Phy Spin Up Intervals
	 *   ---------------------------------
	 *   Number of milliseconds for the controller start operation.
	 *
	 * NOTE: The number of phy spin up intervals will be equivalent
	 *       to the number of phys divided by the number phys allowed
	 *       per interval - 1 (once OEM parameters are supported).
	 *       Currently we assume only 1 phy per interval. */

	return SCIC_SDS_SIGNATURE_FIS_TIMEOUT
		+ SCIC_SDS_CONTROLLER_PHY_START_TIMEOUT
		+ ((SCI_MAX_PHYS - 1) * SCIC_SDS_CONTROLLER_POWER_CONTROL_INTERVAL);
}

static void sci_controller_enable_interrupts(struct isci_host *ihost)
{
	set_bit(IHOST_IRQ_ENABLED, &ihost->flags);
	writel(0, &ihost->smu_registers->interrupt_mask);
}

void sci_controller_disable_interrupts(struct isci_host *ihost)
{
	clear_bit(IHOST_IRQ_ENABLED, &ihost->flags);
	writel(0xffffffff, &ihost->smu_registers->interrupt_mask);
	readl(&ihost->smu_registers->interrupt_mask); /* flush */
}

static void sci_controller_enable_port_task_scheduler(struct isci_host *ihost)
{
	u32 port_task_scheduler_value;

	port_task_scheduler_value =
		readl(&ihost->scu_registers->peg0.ptsg.control);
	port_task_scheduler_value |=
		(SCU_PTSGCR_GEN_BIT(ETM_ENABLE) |
		 SCU_PTSGCR_GEN_BIT(PTSG_ENABLE));
	writel(port_task_scheduler_value,
	       &ihost->scu_registers->peg0.ptsg.control);
}

static void sci_controller_assign_task_entries(struct isci_host *ihost)
{
	u32 task_assignment;

	/*
	 * Assign all the TCs to function 0
	 * TODO: Do we actually need to read this register to write it back?
	 */

	task_assignment =
		readl(&ihost->smu_registers->task_context_assignment[0]);

	task_assignment |= (SMU_TCA_GEN_VAL(STARTING, 0)) |
		(SMU_TCA_GEN_VAL(ENDING,  ihost->task_context_entries - 1)) |
		(SMU_TCA_GEN_BIT(RANGE_CHECK_ENABLE));

	writel(task_assignment,
		&ihost->smu_registers->task_context_assignment[0]);

}

static void sci_controller_initialize_completion_queue(struct isci_host *ihost)
{
	u32 index;
	u32 completion_queue_control_value;
	u32 completion_queue_get_value;
	u32 completion_queue_put_value;

	ihost->completion_queue_get = 0;

	completion_queue_control_value =
		(SMU_CQC_QUEUE_LIMIT_SET(SCU_MAX_COMPLETION_QUEUE_ENTRIES - 1) |
		 SMU_CQC_EVENT_LIMIT_SET(SCU_MAX_EVENTS - 1));

	writel(completion_queue_control_value,
	       &ihost->smu_registers->completion_queue_control);


	/* Set the completion queue get pointer and enable the queue */
	completion_queue_get_value = (
		(SMU_CQGR_GEN_VAL(POINTER, 0))
		| (SMU_CQGR_GEN_VAL(EVENT_POINTER, 0))
		| (SMU_CQGR_GEN_BIT(ENABLE))
		| (SMU_CQGR_GEN_BIT(EVENT_ENABLE))
		);

	writel(completion_queue_get_value,
	       &ihost->smu_registers->completion_queue_get);

	/* Set the completion queue put pointer */
	completion_queue_put_value = (
		(SMU_CQPR_GEN_VAL(POINTER, 0))
		| (SMU_CQPR_GEN_VAL(EVENT_POINTER, 0))
		);

	writel(completion_queue_put_value,
	       &ihost->smu_registers->completion_queue_put);

	/* Initialize the cycle bit of the completion queue entries */
	for (index = 0; index < SCU_MAX_COMPLETION_QUEUE_ENTRIES; index++) {
		/*
		 * If get.cycle_bit != completion_queue.cycle_bit
		 * its not a valid completion queue entry
		 * so at system start all entries are invalid */
		ihost->completion_queue[index] = 0x80000000;
	}
}

static void sci_controller_initialize_unsolicited_frame_queue(struct isci_host *ihost)
{
	u32 frame_queue_control_value;
	u32 frame_queue_get_value;
	u32 frame_queue_put_value;

	/* Write the queue size */
	frame_queue_control_value =
		SCU_UFQC_GEN_VAL(QUEUE_SIZE, SCU_MAX_UNSOLICITED_FRAMES);

	writel(frame_queue_control_value,
	       &ihost->scu_registers->sdma.unsolicited_frame_queue_control);

	/* Setup the get pointer for the unsolicited frame queue */
	frame_queue_get_value = (
		SCU_UFQGP_GEN_VAL(POINTER, 0)
		|  SCU_UFQGP_GEN_BIT(ENABLE_BIT)
		);

	writel(frame_queue_get_value,
	       &ihost->scu_registers->sdma.unsolicited_frame_get_pointer);
	/* Setup the put pointer for the unsolicited frame queue */
	frame_queue_put_value = SCU_UFQPP_GEN_VAL(POINTER, 0);
	writel(frame_queue_put_value,
	       &ihost->scu_registers->sdma.unsolicited_frame_put_pointer);
}

void sci_controller_transition_to_ready(struct isci_host *ihost, enum sci_status status)
{
	if (ihost->sm.current_state_id == SCIC_STARTING) {
		/*
		 * We move into the ready state, because some of the phys/ports
		 * may be up and operational.
		 */
		sci_change_state(&ihost->sm, SCIC_READY);

		isci_host_start_complete(ihost, status);
	}
}

static bool is_phy_starting(struct isci_phy *iphy)
{
	enum sci_phy_states state;

	state = iphy->sm.current_state_id;
	switch (state) {
	case SCI_PHY_STARTING:
	case SCI_PHY_SUB_INITIAL:
	case SCI_PHY_SUB_AWAIT_SAS_SPEED_EN:
	case SCI_PHY_SUB_AWAIT_IAF_UF:
	case SCI_PHY_SUB_AWAIT_SAS_POWER:
	case SCI_PHY_SUB_AWAIT_SATA_POWER:
	case SCI_PHY_SUB_AWAIT_SATA_PHY_EN:
	case SCI_PHY_SUB_AWAIT_SATA_SPEED_EN:
	case SCI_PHY_SUB_AWAIT_OSSP_EN:
	case SCI_PHY_SUB_AWAIT_SIG_FIS_UF:
	case SCI_PHY_SUB_FINAL:
		return true;
	default:
		return false;
	}
}

bool is_controller_start_complete(struct isci_host *ihost)
{
	int i;

	for (i = 0; i < SCI_MAX_PHYS; i++) {
		struct isci_phy *iphy = &ihost->phys[i];
		u32 state = iphy->sm.current_state_id;

		/* in apc mode we need to check every phy, in
		 * mpc mode we only need to check phys that have
		 * been configured into a port
		 */
		if (is_port_config_apc(ihost))
			/* pass */;
		else if (!phy_get_non_dummy_port(iphy))
			continue;

		/* The controller start operation is complete iff:
		 * - all links have been given an opportunity to start
		 * - have no indication of a connected device
		 * - have an indication of a connected device and it has
		 *   finished the link training process.
		 */
		if ((iphy->is_in_link_training == false && state == SCI_PHY_INITIAL) ||
		    (iphy->is_in_link_training == false && state == SCI_PHY_STOPPED) ||
		    (iphy->is_in_link_training == true && is_phy_starting(iphy)) ||
		    (ihost->port_agent.phy_ready_mask != ihost->port_agent.phy_configured_mask))
			return false;
	}

	return true;
}

/**
 * sci_controller_start_next_phy - start phy
 * @scic: controller
 *
 * If all the phys have been started, then attempt to transition the
 * controller to the READY state and inform the user
 * (sci_cb_controller_start_complete()).
 */
static enum sci_status sci_controller_start_next_phy(struct isci_host *ihost)
{
	struct sci_oem_params *oem = &ihost->oem_parameters;
	struct isci_phy *iphy;
	enum sci_status status;

	status = SCI_SUCCESS;

	if (ihost->phy_startup_timer_pending)
		return status;

	if (ihost->next_phy_to_start >= SCI_MAX_PHYS) {
		if (is_controller_start_complete(ihost)) {
			sci_controller_transition_to_ready(ihost, SCI_SUCCESS);
			sci_del_timer(&ihost->phy_timer);
			ihost->phy_startup_timer_pending = false;
		}
	} else {
		iphy = &ihost->phys[ihost->next_phy_to_start];

		if (oem->controller.mode_type == SCIC_PORT_MANUAL_CONFIGURATION_MODE) {
			if (phy_get_non_dummy_port(iphy) == NULL) {
				ihost->next_phy_to_start++;

				/* Caution recursion ahead be forwarned
				 *
				 * The PHY was never added to a PORT in MPC mode
				 * so start the next phy in sequence This phy
				 * will never go link up and will not draw power
				 * the OEM parameters either configured the phy
				 * incorrectly for the PORT or it was never
				 * assigned to a PORT
				 */
				return sci_controller_start_next_phy(ihost);
			}
		}

		status = sci_phy_start(iphy);

		if (status == SCI_SUCCESS) {
			sci_mod_timer(&ihost->phy_timer,
				      SCIC_SDS_CONTROLLER_PHY_START_TIMEOUT);
			ihost->phy_startup_timer_pending = true;
		} else {
			dev_warn(&ihost->pdev->dev,
				 "%s: Controller stop operation failed "
				 "to stop phy %d because of status "
				 "%d.\n",
				 __func__,
				 ihost->phys[ihost->next_phy_to_start].phy_index,
				 status);
		}

		ihost->next_phy_to_start++;
	}

	return status;
}

static void phy_startup_timeout(unsigned long data)
{
	struct sci_timer *tmr = (struct sci_timer *)data;
	struct isci_host *ihost = container_of(tmr, typeof(*ihost), phy_timer);
	unsigned long flags;
	enum sci_status status;

	spin_lock_irqsave(&ihost->scic_lock, flags);

	if (tmr->cancel)
		goto done;

	ihost->phy_startup_timer_pending = false;

	do {
		status = sci_controller_start_next_phy(ihost);
	} while (status != SCI_SUCCESS);

done:
	spin_unlock_irqrestore(&ihost->scic_lock, flags);
}

static u16 isci_tci_active(struct isci_host *ihost)
{
	return CIRC_CNT(ihost->tci_head, ihost->tci_tail, SCI_MAX_IO_REQUESTS);
}

static enum sci_status sci_controller_start(struct isci_host *ihost,
					     u32 timeout)
{
	enum sci_status result;
	u16 index;

	if (ihost->sm.current_state_id != SCIC_INITIALIZED) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	/* Build the TCi free pool */
	BUILD_BUG_ON(SCI_MAX_IO_REQUESTS > 1 << sizeof(ihost->tci_pool[0]) * 8);
	ihost->tci_head = 0;
	ihost->tci_tail = 0;
	for (index = 0; index < ihost->task_context_entries; index++)
		isci_tci_free(ihost, index);

	/* Build the RNi free pool */
	sci_remote_node_table_initialize(&ihost->available_remote_nodes,
					 ihost->remote_node_entries);

	/*
	 * Before anything else lets make sure we will not be
	 * interrupted by the hardware.
	 */
	sci_controller_disable_interrupts(ihost);

	/* Enable the port task scheduler */
	sci_controller_enable_port_task_scheduler(ihost);

	/* Assign all the task entries to ihost physical function */
	sci_controller_assign_task_entries(ihost);

	/* Now initialize the completion queue */
	sci_controller_initialize_completion_queue(ihost);

	/* Initialize the unsolicited frame queue for use */
	sci_controller_initialize_unsolicited_frame_queue(ihost);

	/* Start all of the ports on this controller */
	for (index = 0; index < ihost->logical_port_entries; index++) {
		struct isci_port *iport = &ihost->ports[index];

		result = sci_port_start(iport);
		if (result)
			return result;
	}

	sci_controller_start_next_phy(ihost);

	sci_mod_timer(&ihost->timer, timeout);

	sci_change_state(&ihost->sm, SCIC_STARTING);

	return SCI_SUCCESS;
}

void isci_host_scan_start(struct Scsi_Host *shost)
{
	struct isci_host *ihost = SHOST_TO_SAS_HA(shost)->lldd_ha;
	unsigned long tmo = sci_controller_get_suggested_start_timeout(ihost);

	set_bit(IHOST_START_PENDING, &ihost->flags);

	spin_lock_irq(&ihost->scic_lock);
	sci_controller_start(ihost, tmo);
	sci_controller_enable_interrupts(ihost);
	spin_unlock_irq(&ihost->scic_lock);
}

static void isci_host_stop_complete(struct isci_host *ihost)
{
	sci_controller_disable_interrupts(ihost);
	clear_bit(IHOST_STOP_PENDING, &ihost->flags);
	wake_up(&ihost->eventq);
}

static void sci_controller_completion_handler(struct isci_host *ihost)
{
	/* Empty out the completion queue */
	if (sci_controller_completion_queue_has_entries(ihost))
		sci_controller_process_completions(ihost);

	/* Clear the interrupt and enable all interrupts again */
	writel(SMU_ISR_COMPLETION, &ihost->smu_registers->interrupt_status);
	/* Could we write the value of SMU_ISR_COMPLETION? */
	writel(0xFF000000, &ihost->smu_registers->interrupt_mask);
	writel(0, &ihost->smu_registers->interrupt_mask);
}

/**
 * isci_host_completion_routine() - This function is the delayed service
 *    routine that calls the sci core library's completion handler. It's
 *    scheduled as a tasklet from the interrupt service routine when interrupts
 *    in use, or set as the timeout function in polled mode.
 * @data: This parameter specifies the ISCI host object
 *
 */
void isci_host_completion_routine(unsigned long data)
{
	struct isci_host *ihost = (struct isci_host *)data;
	struct list_head    completed_request_list;
	struct list_head    errored_request_list;
	struct list_head    *current_position;
	struct list_head    *next_position;
	struct isci_request *request;
	struct isci_request *next_request;
	struct sas_task     *task;
	u16 active;

	INIT_LIST_HEAD(&completed_request_list);
	INIT_LIST_HEAD(&errored_request_list);

	spin_lock_irq(&ihost->scic_lock);

	sci_controller_completion_handler(ihost);

	/* Take the lists of completed I/Os from the host. */

	list_splice_init(&ihost->requests_to_complete,
			 &completed_request_list);

	/* Take the list of errored I/Os from the host. */
	list_splice_init(&ihost->requests_to_errorback,
			 &errored_request_list);

	spin_unlock_irq(&ihost->scic_lock);

	/* Process any completions in the lists. */
	list_for_each_safe(current_position, next_position,
			   &completed_request_list) {

		request = list_entry(current_position, struct isci_request,
				     completed_node);
		task = isci_request_access_task(request);

		/* Normal notification (task_done) */
		dev_dbg(&ihost->pdev->dev,
			"%s: Normal - request/task = %p/%p\n",
			__func__,
			request,
			task);

		/* Return the task to libsas */
		if (task != NULL) {

			task->lldd_task = NULL;
			if (!(task->task_state_flags & SAS_TASK_STATE_ABORTED)) {

				/* If the task is already in the abort path,
				* the task_done callback cannot be called.
				*/
				task->task_done(task);
			}
		}

		spin_lock_irq(&ihost->scic_lock);
		isci_free_tag(ihost, request->io_tag);
		spin_unlock_irq(&ihost->scic_lock);
	}
	list_for_each_entry_safe(request, next_request, &errored_request_list,
				 completed_node) {

		task = isci_request_access_task(request);

		/* Use sas_task_abort */
		dev_warn(&ihost->pdev->dev,
			 "%s: Error - request/task = %p/%p\n",
			 __func__,
			 request,
			 task);

		if (task != NULL) {

			/* Put the task into the abort path if it's not there
			 * already.
			 */
			if (!(task->task_state_flags & SAS_TASK_STATE_ABORTED))
				sas_task_abort(task);

		} else {
			/* This is a case where the request has completed with a
			 * status such that it needed further target servicing,
			 * but the sas_task reference has already been removed
			 * from the request.  Since it was errored, it was not
			 * being aborted, so there is nothing to do except free
			 * it.
			 */

			spin_lock_irq(&ihost->scic_lock);
			/* Remove the request from the remote device's list
			* of pending requests.
			*/
			list_del_init(&request->dev_node);
			isci_free_tag(ihost, request->io_tag);
			spin_unlock_irq(&ihost->scic_lock);
		}
	}

	/* the coalesence timeout doubles at each encoding step, so
	 * update it based on the ilog2 value of the outstanding requests
	 */
	active = isci_tci_active(ihost);
	writel(SMU_ICC_GEN_VAL(NUMBER, active) |
	       SMU_ICC_GEN_VAL(TIMER, ISCI_COALESCE_BASE + ilog2(active)),
	       &ihost->smu_registers->interrupt_coalesce_control);
}

/**
 * sci_controller_stop() - This method will stop an individual controller
 *    object.This method will invoke the associated user callback upon
 *    completion.  The completion callback is called when the following
 *    conditions are met: -# the method return status is SCI_SUCCESS. -# the
 *    controller has been quiesced. This method will ensure that all IO
 *    requests are quiesced, phys are stopped, and all additional operation by
 *    the hardware is halted.
 * @controller: the handle to the controller object to stop.
 * @timeout: This parameter specifies the number of milliseconds in which the
 *    stop operation should complete.
 *
 * The controller must be in the STARTED or STOPPED state. Indicate if the
 * controller stop method succeeded or failed in some way. SCI_SUCCESS if the
 * stop operation successfully began. SCI_WARNING_ALREADY_IN_STATE if the
 * controller is already in the STOPPED state. SCI_FAILURE_INVALID_STATE if the
 * controller is not either in the STARTED or STOPPED states.
 */
static enum sci_status sci_controller_stop(struct isci_host *ihost, u32 timeout)
{
	if (ihost->sm.current_state_id != SCIC_READY) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	sci_mod_timer(&ihost->timer, timeout);
	sci_change_state(&ihost->sm, SCIC_STOPPING);
	return SCI_SUCCESS;
}

/**
 * sci_controller_reset() - This method will reset the supplied core
 *    controller regardless of the state of said controller.  This operation is
 *    considered destructive.  In other words, all current operations are wiped
 *    out.  No IO completions for outstanding devices occur.  Outstanding IO
 *    requests are not aborted or completed at the actual remote device.
 * @controller: the handle to the controller object to reset.
 *
 * Indicate if the controller reset method succeeded or failed in some way.
 * SCI_SUCCESS if the reset operation successfully started. SCI_FATAL_ERROR if
 * the controller reset operation is unable to complete.
 */
static enum sci_status sci_controller_reset(struct isci_host *ihost)
{
	switch (ihost->sm.current_state_id) {
	case SCIC_RESET:
	case SCIC_READY:
	case SCIC_STOPPING:
	case SCIC_FAILED:
		/*
		 * The reset operation is not a graceful cleanup, just
		 * perform the state transition.
		 */
		sci_change_state(&ihost->sm, SCIC_RESETTING);
		return SCI_SUCCESS;
	default:
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}
}

static enum sci_status sci_controller_stop_phys(struct isci_host *ihost)
{
	u32 index;
	enum sci_status status;
	enum sci_status phy_status;

	status = SCI_SUCCESS;

	for (index = 0; index < SCI_MAX_PHYS; index++) {
		phy_status = sci_phy_stop(&ihost->phys[index]);

		if (phy_status != SCI_SUCCESS &&
		    phy_status != SCI_FAILURE_INVALID_STATE) {
			status = SCI_FAILURE;

			dev_warn(&ihost->pdev->dev,
				 "%s: Controller stop operation failed to stop "
				 "phy %d because of status %d.\n",
				 __func__,
				 ihost->phys[index].phy_index, phy_status);
		}
	}

	return status;
}


/**
 * isci_host_deinit - shutdown frame reception and dma
 * @ihost: host to take down
 *
 * This is called in either the driver shutdown or the suspend path.  In
 * the shutdown case libsas went through port teardown and normal device
 * removal (i.e. physical links stayed up to service scsi_device removal
 * commands).  In the suspend case we disable the hardware without
 * notifying libsas of the link down events since we want libsas to
 * remember the domain across the suspend/resume cycle
 */
void isci_host_deinit(struct isci_host *ihost)
{
	int i;

	/* disable output data selects */
	for (i = 0; i < isci_gpio_count(ihost); i++)
		writel(SGPIO_HW_CONTROL, &ihost->scu_registers->peg0.sgpio.output_data_select[i]);

	set_bit(IHOST_STOP_PENDING, &ihost->flags);

	spin_lock_irq(&ihost->scic_lock);
	sci_controller_stop(ihost, SCIC_CONTROLLER_STOP_TIMEOUT);
	spin_unlock_irq(&ihost->scic_lock);

	wait_for_stop(ihost);

	/* phy stop is after controller stop to allow port and device to
	 * go idle before shutting down the phys, but the expectation is
	 * that i/o has been shut off well before we reach this
	 * function.
	 */
	sci_controller_stop_phys(ihost);

	/* disable sgpio: where the above wait should give time for the
	 * enclosure to sample the gpios going inactive
	 */
	writel(0, &ihost->scu_registers->peg0.sgpio.interface_control);

	spin_lock_irq(&ihost->scic_lock);
	sci_controller_reset(ihost);
	spin_unlock_irq(&ihost->scic_lock);

	/* Cancel any/all outstanding port timers */
	for (i = 0; i < ihost->logical_port_entries; i++) {
		struct isci_port *iport = &ihost->ports[i];
		del_timer_sync(&iport->timer.timer);
	}

	/* Cancel any/all outstanding phy timers */
	for (i = 0; i < SCI_MAX_PHYS; i++) {
		struct isci_phy *iphy = &ihost->phys[i];
		del_timer_sync(&iphy->sata_timer.timer);
	}

	del_timer_sync(&ihost->port_agent.timer.timer);

	del_timer_sync(&ihost->power_control.timer.timer);

	del_timer_sync(&ihost->timer.timer);

	del_timer_sync(&ihost->phy_timer.timer);
}

static void __iomem *scu_base(struct isci_host *isci_host)
{
	struct pci_dev *pdev = isci_host->pdev;
	int id = isci_host->id;

	return pcim_iomap_table(pdev)[SCI_SCU_BAR * 2] + SCI_SCU_BAR_SIZE * id;
}

static void __iomem *smu_base(struct isci_host *isci_host)
{
	struct pci_dev *pdev = isci_host->pdev;
	int id = isci_host->id;

	return pcim_iomap_table(pdev)[SCI_SMU_BAR * 2] + SCI_SMU_BAR_SIZE * id;
}

static void sci_controller_initial_state_enter(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	sci_change_state(&ihost->sm, SCIC_RESET);
}

static inline void sci_controller_starting_state_exit(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	sci_del_timer(&ihost->timer);
}

#define INTERRUPT_COALESCE_TIMEOUT_BASE_RANGE_LOWER_BOUND_NS 853
#define INTERRUPT_COALESCE_TIMEOUT_BASE_RANGE_UPPER_BOUND_NS 1280
#define INTERRUPT_COALESCE_TIMEOUT_MAX_US                    2700000
#define INTERRUPT_COALESCE_NUMBER_MAX                        256
#define INTERRUPT_COALESCE_TIMEOUT_ENCODE_MIN                7
#define INTERRUPT_COALESCE_TIMEOUT_ENCODE_MAX                28

/**
 * sci_controller_set_interrupt_coalescence() - This method allows the user to
 *    configure the interrupt coalescence.
 * @controller: This parameter represents the handle to the controller object
 *    for which its interrupt coalesce register is overridden.
 * @coalesce_number: Used to control the number of entries in the Completion
 *    Queue before an interrupt is generated. If the number of entries exceed
 *    this number, an interrupt will be generated. The valid range of the input
 *    is [0, 256]. A setting of 0 results in coalescing being disabled.
 * @coalesce_timeout: Timeout value in microseconds. The valid range of the
 *    input is [0, 2700000] . A setting of 0 is allowed and results in no
 *    interrupt coalescing timeout.
 *
 * Indicate if the user successfully set the interrupt coalesce parameters.
 * SCI_SUCCESS The user successfully updated the interrutp coalescence.
 * SCI_FAILURE_INVALID_PARAMETER_VALUE The user input value is out of range.
 */
static enum sci_status
sci_controller_set_interrupt_coalescence(struct isci_host *ihost,
					 u32 coalesce_number,
					 u32 coalesce_timeout)
{
	u8 timeout_encode = 0;
	u32 min = 0;
	u32 max = 0;

	/* Check if the input parameters fall in the range. */
	if (coalesce_number > INTERRUPT_COALESCE_NUMBER_MAX)
		return SCI_FAILURE_INVALID_PARAMETER_VALUE;

	/*
	 *  Defined encoding for interrupt coalescing timeout:
	 *              Value   Min      Max     Units
	 *              -----   ---      ---     -----
	 *              0       -        -       Disabled
	 *              1       13.3     20.0    ns
	 *              2       26.7     40.0
	 *              3       53.3     80.0
	 *              4       106.7    160.0
	 *              5       213.3    320.0
	 *              6       426.7    640.0
	 *              7       853.3    1280.0
	 *              8       1.7      2.6     us
	 *              9       3.4      5.1
	 *              10      6.8      10.2
	 *              11      13.7     20.5
	 *              12      27.3     41.0
	 *              13      54.6     81.9
	 *              14      109.2    163.8
	 *              15      218.5    327.7
	 *              16      436.9    655.4
	 *              17      873.8    1310.7
	 *              18      1.7      2.6     ms
	 *              19      3.5      5.2
	 *              20      7.0      10.5
	 *              21      14.0     21.0
	 *              22      28.0     41.9
	 *              23      55.9     83.9
	 *              24      111.8    167.8
	 *              25      223.7    335.5
	 *              26      447.4    671.1
	 *              27      894.8    1342.2
	 *              28      1.8      2.7     s
	 *              Others Undefined */

	/*
	 * Use the table above to decide the encode of interrupt coalescing timeout
	 * value for register writing. */
	if (coalesce_timeout == 0)
		timeout_encode = 0;
	else{
		/* make the timeout value in unit of (10 ns). */
		coalesce_timeout = coalesce_timeout * 100;
		min = INTERRUPT_COALESCE_TIMEOUT_BASE_RANGE_LOWER_BOUND_NS / 10;
		max = INTERRUPT_COALESCE_TIMEOUT_BASE_RANGE_UPPER_BOUND_NS / 10;

		/* get the encode of timeout for register writing. */
		for (timeout_encode = INTERRUPT_COALESCE_TIMEOUT_ENCODE_MIN;
		      timeout_encode <= INTERRUPT_COALESCE_TIMEOUT_ENCODE_MAX;
		      timeout_encode++) {
			if (min <= coalesce_timeout &&  max > coalesce_timeout)
				break;
			else if (coalesce_timeout >= max && coalesce_timeout < min * 2
				 && coalesce_timeout <= INTERRUPT_COALESCE_TIMEOUT_MAX_US * 100) {
				if ((coalesce_timeout - max) < (2 * min - coalesce_timeout))
					break;
				else{
					timeout_encode++;
					break;
				}
			} else {
				max = max * 2;
				min = min * 2;
			}
		}

		if (timeout_encode == INTERRUPT_COALESCE_TIMEOUT_ENCODE_MAX + 1)
			/* the value is out of range. */
			return SCI_FAILURE_INVALID_PARAMETER_VALUE;
	}

	writel(SMU_ICC_GEN_VAL(NUMBER, coalesce_number) |
	       SMU_ICC_GEN_VAL(TIMER, timeout_encode),
	       &ihost->smu_registers->interrupt_coalesce_control);


	ihost->interrupt_coalesce_number = (u16)coalesce_number;
	ihost->interrupt_coalesce_timeout = coalesce_timeout / 100;

	return SCI_SUCCESS;
}


static void sci_controller_ready_state_enter(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);
	u32 val;

	/* enable clock gating for power control of the scu unit */
	val = readl(&ihost->smu_registers->clock_gating_control);
	val &= ~(SMU_CGUCR_GEN_BIT(REGCLK_ENABLE) |
		 SMU_CGUCR_GEN_BIT(TXCLK_ENABLE) |
		 SMU_CGUCR_GEN_BIT(XCLK_ENABLE));
	val |= SMU_CGUCR_GEN_BIT(IDLE_ENABLE);
	writel(val, &ihost->smu_registers->clock_gating_control);

	/* set the default interrupt coalescence number and timeout value. */
	sci_controller_set_interrupt_coalescence(ihost, 0, 0);
}

static void sci_controller_ready_state_exit(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	/* disable interrupt coalescence. */
	sci_controller_set_interrupt_coalescence(ihost, 0, 0);
}

static enum sci_status sci_controller_stop_ports(struct isci_host *ihost)
{
	u32 index;
	enum sci_status port_status;
	enum sci_status status = SCI_SUCCESS;

	for (index = 0; index < ihost->logical_port_entries; index++) {
		struct isci_port *iport = &ihost->ports[index];

		port_status = sci_port_stop(iport);

		if ((port_status != SCI_SUCCESS) &&
		    (port_status != SCI_FAILURE_INVALID_STATE)) {
			status = SCI_FAILURE;

			dev_warn(&ihost->pdev->dev,
				 "%s: Controller stop operation failed to "
				 "stop port %d because of status %d.\n",
				 __func__,
				 iport->logical_port_index,
				 port_status);
		}
	}

	return status;
}

static enum sci_status sci_controller_stop_devices(struct isci_host *ihost)
{
	u32 index;
	enum sci_status status;
	enum sci_status device_status;

	status = SCI_SUCCESS;

	for (index = 0; index < ihost->remote_node_entries; index++) {
		if (ihost->device_table[index] != NULL) {
			/* / @todo What timeout value do we want to provide to this request? */
			device_status = sci_remote_device_stop(ihost->device_table[index], 0);

			if ((device_status != SCI_SUCCESS) &&
			    (device_status != SCI_FAILURE_INVALID_STATE)) {
				dev_warn(&ihost->pdev->dev,
					 "%s: Controller stop operation failed "
					 "to stop device 0x%p because of "
					 "status %d.\n",
					 __func__,
					 ihost->device_table[index], device_status);
			}
		}
	}

	return status;
}

static void sci_controller_stopping_state_enter(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	sci_controller_stop_devices(ihost);
	sci_controller_stop_ports(ihost);

	if (!sci_controller_has_remote_devices_stopping(ihost))
		isci_host_stop_complete(ihost);
}

static void sci_controller_stopping_state_exit(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	sci_del_timer(&ihost->timer);
}

static void sci_controller_reset_hardware(struct isci_host *ihost)
{
	/* Disable interrupts so we dont take any spurious interrupts */
	sci_controller_disable_interrupts(ihost);

	/* Reset the SCU */
	writel(0xFFFFFFFF, &ihost->smu_registers->soft_reset_control);

	/* Delay for 1ms to before clearing the CQP and UFQPR. */
	udelay(1000);

	/* The write to the CQGR clears the CQP */
	writel(0x00000000, &ihost->smu_registers->completion_queue_get);

	/* The write to the UFQGP clears the UFQPR */
	writel(0, &ihost->scu_registers->sdma.unsolicited_frame_get_pointer);

	/* clear all interrupts */
	writel(~SMU_INTERRUPT_STATUS_RESERVED_MASK, &ihost->smu_registers->interrupt_status);
}

static void sci_controller_resetting_state_enter(struct sci_base_state_machine *sm)
{
	struct isci_host *ihost = container_of(sm, typeof(*ihost), sm);

	sci_controller_reset_hardware(ihost);
	sci_change_state(&ihost->sm, SCIC_RESET);
}

static const struct sci_base_state sci_controller_state_table[] = {
	[SCIC_INITIAL] = {
		.enter_state = sci_controller_initial_state_enter,
	},
	[SCIC_RESET] = {},
	[SCIC_INITIALIZING] = {},
	[SCIC_INITIALIZED] = {},
	[SCIC_STARTING] = {
		.exit_state  = sci_controller_starting_state_exit,
	},
	[SCIC_READY] = {
		.enter_state = sci_controller_ready_state_enter,
		.exit_state  = sci_controller_ready_state_exit,
	},
	[SCIC_RESETTING] = {
		.enter_state = sci_controller_resetting_state_enter,
	},
	[SCIC_STOPPING] = {
		.enter_state = sci_controller_stopping_state_enter,
		.exit_state = sci_controller_stopping_state_exit,
	},
	[SCIC_FAILED] = {}
};

static void controller_timeout(unsigned long data)
{
	struct sci_timer *tmr = (struct sci_timer *)data;
	struct isci_host *ihost = container_of(tmr, typeof(*ihost), timer);
	struct sci_base_state_machine *sm = &ihost->sm;
	unsigned long flags;

	spin_lock_irqsave(&ihost->scic_lock, flags);

	if (tmr->cancel)
		goto done;

	if (sm->current_state_id == SCIC_STARTING)
		sci_controller_transition_to_ready(ihost, SCI_FAILURE_TIMEOUT);
	else if (sm->current_state_id == SCIC_STOPPING) {
		sci_change_state(sm, SCIC_FAILED);
		isci_host_stop_complete(ihost);
	} else	/* / @todo Now what do we want to do in this case? */
		dev_err(&ihost->pdev->dev,
			"%s: Controller timer fired when controller was not "
			"in a state being timed.\n",
			__func__);

done:
	spin_unlock_irqrestore(&ihost->scic_lock, flags);
}

static enum sci_status sci_controller_construct(struct isci_host *ihost,
						void __iomem *scu_base,
						void __iomem *smu_base)
{
	u8 i;

	sci_init_sm(&ihost->sm, sci_controller_state_table, SCIC_INITIAL);

	ihost->scu_registers = scu_base;
	ihost->smu_registers = smu_base;

	sci_port_configuration_agent_construct(&ihost->port_agent);

	/* Construct the ports for this controller */
	for (i = 0; i < SCI_MAX_PORTS; i++)
		sci_port_construct(&ihost->ports[i], i, ihost);
	sci_port_construct(&ihost->ports[i], SCIC_SDS_DUMMY_PORT, ihost);

	/* Construct the phys for this controller */
	for (i = 0; i < SCI_MAX_PHYS; i++) {
		/* Add all the PHYs to the dummy port */
		sci_phy_construct(&ihost->phys[i],
				  &ihost->ports[SCI_MAX_PORTS], i);
	}

	ihost->invalid_phy_mask = 0;

	sci_init_timer(&ihost->timer, controller_timeout);

	return sci_controller_reset(ihost);
}

int sci_oem_parameters_validate(struct sci_oem_params *oem, u8 version)
{
	int i;

	for (i = 0; i < SCI_MAX_PORTS; i++)
		if (oem->ports[i].phy_mask > SCIC_SDS_PARM_PHY_MASK_MAX)
			return -EINVAL;

	for (i = 0; i < SCI_MAX_PHYS; i++)
		if (oem->phys[i].sas_address.high == 0 &&
		    oem->phys[i].sas_address.low == 0)
			return -EINVAL;

	if (oem->controller.mode_type == SCIC_PORT_AUTOMATIC_CONFIGURATION_MODE) {
		for (i = 0; i < SCI_MAX_PHYS; i++)
			if (oem->ports[i].phy_mask != 0)
				return -EINVAL;
	} else if (oem->controller.mode_type == SCIC_PORT_MANUAL_CONFIGURATION_MODE) {
		u8 phy_mask = 0;

		for (i = 0; i < SCI_MAX_PHYS; i++)
			phy_mask |= oem->ports[i].phy_mask;

		if (phy_mask == 0)
			return -EINVAL;
	} else
		return -EINVAL;

	if (oem->controller.max_concurr_spin_up > MAX_CONCURRENT_DEVICE_SPIN_UP_COUNT ||
	    oem->controller.max_concurr_spin_up < 1)
		return -EINVAL;

	if (oem->controller.do_enable_ssc) {
		if (version < ISCI_ROM_VER_1_1 && oem->controller.do_enable_ssc != 1)
			return -EINVAL;

		if (version >= ISCI_ROM_VER_1_1) {
			u8 test = oem->controller.ssc_sata_tx_spread_level;

			switch (test) {
			case 0:
			case 2:
			case 3:
			case 6:
			case 7:
				break;
			default:
				return -EINVAL;
			}

			test = oem->controller.ssc_sas_tx_spread_level;
			if (oem->controller.ssc_sas_tx_type == 0) {
				switch (test) {
				case 0:
				case 2:
				case 3:
					break;
				default:
					return -EINVAL;
				}
			} else if (oem->controller.ssc_sas_tx_type == 1) {
				switch (test) {
				case 0:
				case 3:
				case 6:
					break;
				default:
					return -EINVAL;
				}
			}
		}
	}

	return 0;
}

static u8 max_spin_up(struct isci_host *ihost)
{
	if (ihost->user_parameters.max_concurr_spinup)
		return min_t(u8, ihost->user_parameters.max_concurr_spinup,
			     MAX_CONCURRENT_DEVICE_SPIN_UP_COUNT);
	else
		return min_t(u8, ihost->oem_parameters.controller.max_concurr_spin_up,
			     MAX_CONCURRENT_DEVICE_SPIN_UP_COUNT);
}

static void power_control_timeout(unsigned long data)
{
	struct sci_timer *tmr = (struct sci_timer *)data;
	struct isci_host *ihost = container_of(tmr, typeof(*ihost), power_control.timer);
	struct isci_phy *iphy;
	unsigned long flags;
	u8 i;

	spin_lock_irqsave(&ihost->scic_lock, flags);

	if (tmr->cancel)
		goto done;

	ihost->power_control.phys_granted_power = 0;

	if (ihost->power_control.phys_waiting == 0) {
		ihost->power_control.timer_started = false;
		goto done;
	}

	for (i = 0; i < SCI_MAX_PHYS; i++) {

		if (ihost->power_control.phys_waiting == 0)
			break;

		iphy = ihost->power_control.requesters[i];
		if (iphy == NULL)
			continue;

		if (ihost->power_control.phys_granted_power >= max_spin_up(ihost))
			break;

		ihost->power_control.requesters[i] = NULL;
		ihost->power_control.phys_waiting--;
		ihost->power_control.phys_granted_power++;
		sci_phy_consume_power_handler(iphy);

		if (iphy->protocol == SAS_PROTOCOL_SSP) {
			u8 j;

			for (j = 0; j < SCI_MAX_PHYS; j++) {
				struct isci_phy *requester = ihost->power_control.requesters[j];

				/*
				 * Search the power_control queue to see if there are other phys
				 * attached to the same remote device. If found, take all of
				 * them out of await_sas_power state.
				 */
				if (requester != NULL && requester != iphy) {
					u8 other = memcmp(requester->frame_rcvd.iaf.sas_addr,
							  iphy->frame_rcvd.iaf.sas_addr,
							  sizeof(requester->frame_rcvd.iaf.sas_addr));

					if (other == 0) {
						ihost->power_control.requesters[j] = NULL;
						ihost->power_control.phys_waiting--;
						sci_phy_consume_power_handler(requester);
					}
				}
			}
		}
	}

	/*
	 * It doesn't matter if the power list is empty, we need to start the
	 * timer in case another phy becomes ready.
	 */
	sci_mod_timer(tmr, SCIC_SDS_CONTROLLER_POWER_CONTROL_INTERVAL);
	ihost->power_control.timer_started = true;

done:
	spin_unlock_irqrestore(&ihost->scic_lock, flags);
}

void sci_controller_power_control_queue_insert(struct isci_host *ihost,
					       struct isci_phy *iphy)
{
	BUG_ON(iphy == NULL);

	if (ihost->power_control.phys_granted_power < max_spin_up(ihost)) {
		ihost->power_control.phys_granted_power++;
		sci_phy_consume_power_handler(iphy);

		/*
		 * stop and start the power_control timer. When the timer fires, the
		 * no_of_phys_granted_power will be set to 0
		 */
		if (ihost->power_control.timer_started)
			sci_del_timer(&ihost->power_control.timer);

		sci_mod_timer(&ihost->power_control.timer,
				 SCIC_SDS_CONTROLLER_POWER_CONTROL_INTERVAL);
		ihost->power_control.timer_started = true;

	} else {
		/*
		 * There are phys, attached to the same sas address as this phy, are
		 * already in READY state, this phy don't need wait.
		 */
		u8 i;
		struct isci_phy *current_phy;

		for (i = 0; i < SCI_MAX_PHYS; i++) {
			u8 other;
			current_phy = &ihost->phys[i];

			other = memcmp(current_phy->frame_rcvd.iaf.sas_addr,
				       iphy->frame_rcvd.iaf.sas_addr,
				       sizeof(current_phy->frame_rcvd.iaf.sas_addr));

			if (current_phy->sm.current_state_id == SCI_PHY_READY &&
			    current_phy->protocol == SAS_PROTOCOL_SSP &&
			    other == 0) {
				sci_phy_consume_power_handler(iphy);
				break;
			}
		}

		if (i == SCI_MAX_PHYS) {
			/* Add the phy in the waiting list */
			ihost->power_control.requesters[iphy->phy_index] = iphy;
			ihost->power_control.phys_waiting++;
		}
	}
}

void sci_controller_power_control_queue_remove(struct isci_host *ihost,
					       struct isci_phy *iphy)
{
	BUG_ON(iphy == NULL);

	if (ihost->power_control.requesters[iphy->phy_index])
		ihost->power_control.phys_waiting--;

	ihost->power_control.requesters[iphy->phy_index] = NULL;
}

static int is_long_cable(int phy, unsigned char selection_byte)
{
	return !!(selection_byte & (1 << phy));
}

static int is_medium_cable(int phy, unsigned char selection_byte)
{
	return !!(selection_byte & (1 << (phy + 4)));
}

static enum cable_selections decode_selection_byte(
	int phy,
	unsigned char selection_byte)
{
	return ((selection_byte & (1 << phy)) ? 1 : 0)
		+ (selection_byte & (1 << (phy + 4)) ? 2 : 0);
}

static unsigned char *to_cable_select(struct isci_host *ihost)
{
	if (is_cable_select_overridden())
		return ((unsigned char *)&cable_selection_override)
			+ ihost->id;
	else
		return &ihost->oem_parameters.controller.cable_selection_mask;
}

enum cable_selections decode_cable_selection(struct isci_host *ihost, int phy)
{
	return decode_selection_byte(phy, *to_cable_select(ihost));
}

char *lookup_cable_names(enum cable_selections selection)
{
	static char *cable_names[] = {
		[short_cable]     = "short",
		[long_cable]      = "long",
		[medium_cable]    = "medium",
		[undefined_cable] = "<undefined, assumed long>" /* bit 0==1 */
	};
	return (selection <= undefined_cable) ? cable_names[selection]
					      : cable_names[undefined_cable];
}

#define AFE_REGISTER_WRITE_DELAY 10

static void sci_controller_afe_initialization(struct isci_host *ihost)
{
	struct scu_afe_registers __iomem *afe = &ihost->scu_registers->afe;
	const struct sci_oem_params *oem = &ihost->oem_parameters;
	struct pci_dev *pdev = ihost->pdev;
	u32 afe_status;
	u32 phy_id;
	unsigned char cable_selection_mask = *to_cable_select(ihost);

	/* Clear DFX Status registers */
	writel(0x0081000f, &afe->afe_dfx_master_control0);
	udelay(AFE_REGISTER_WRITE_DELAY);

	if (is_b0(pdev) || is_c0(pdev) || is_c1(pdev)) {
		/* PM Rx Equalization Save, PM SPhy Rx Acknowledgement
		 * Timer, PM Stagger Timer
		 */
		writel(0x0007FFFF, &afe->afe_pmsn_master_control2);
		udelay(AFE_REGISTER_WRITE_DELAY);
	}

	/* Configure bias currents to normal */
	if (is_a2(pdev))
		writel(0x00005A00, &afe->afe_bias_control);
	else if (is_b0(pdev) || is_c0(pdev))
		writel(0x00005F00, &afe->afe_bias_control);
	else if (is_c1(pdev))
		writel(0x00005500, &afe->afe_bias_control);

	udelay(AFE_REGISTER_WRITE_DELAY);

	/* Enable PLL */
	if (is_a2(pdev))
		writel(0x80040908, &afe->afe_pll_control0);
	else if (is_b0(pdev) || is_c0(pdev))
		writel(0x80040A08, &afe->afe_pll_control0);
	else if (is_c1(pdev)) {
		writel(0x80000B08, &afe->afe_pll_control0);
		udelay(AFE_REGISTER_WRITE_DELAY);
		writel(0x00000B08, &afe->afe_pll_control0);
		udelay(AFE_REGISTER_WRITE_DELAY);
		writel(0x80000B08, &afe->afe_pll_control0);
	}

	udelay(AFE_REGISTER_WRITE_DELAY);

	/* Wait for the PLL to lock */
	do {
		afe_status = readl(&afe->afe_common_block_status);
		udelay(AFE_REGISTER_WRITE_DELAY);
	} while ((afe_status & 0x00001000) == 0);

	if (is_a2(pdev)) {
		/* Shorten SAS SNW lock time (RxLock timer value from 76
		 * us to 50 us)
		 */
		writel(0x7bcc96ad, &afe->afe_pmsn_master_control0);
		udelay(AFE_REGISTER_WRITE_DELAY);
	}

	for (phy_id = 0; phy_id < SCI_MAX_PHYS; phy_id++) {
		struct scu_afe_transceiver *xcvr = &afe->scu_afe_xcvr[phy_id];
		const struct sci_phy_oem_params *oem_phy = &oem->phys[phy_id];
		int cable_length_long =
			is_long_cable(phy_id, cable_selection_mask);
		int cable_length_medium =
			is_medium_cable(phy_id, cable_selection_mask);

		if (is_a2(pdev)) {
			/* All defaults, except the Receive Word
			 * Alignament/Comma Detect Enable....(0xe800)
			 */
			writel(0x00004512, &xcvr->afe_xcvr_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(0x0050100F, &xcvr->afe_xcvr_control1);
			udelay(AFE_REGISTER_WRITE_DELAY);
		} else if (is_b0(pdev)) {
			/* Configure transmitter SSC parameters */
			writel(0x00030000, &xcvr->afe_tx_ssc_control);
			udelay(AFE_REGISTER_WRITE_DELAY);
		} else if (is_c0(pdev)) {
			/* Configure transmitter SSC parameters */
			writel(0x00010202, &xcvr->afe_tx_ssc_control);
			udelay(AFE_REGISTER_WRITE_DELAY);

			/* All defaults, except the Receive Word
			 * Alignament/Comma Detect Enable....(0xe800)
			 */
			writel(0x00014500, &xcvr->afe_xcvr_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);
		} else if (is_c1(pdev)) {
			/* Configure transmitter SSC parameters */
			writel(0x00010202, &xcvr->afe_tx_ssc_control);
			udelay(AFE_REGISTER_WRITE_DELAY);

			/* All defaults, except the Receive Word
			 * Alignament/Comma Detect Enable....(0xe800)
			 */
			writel(0x0001C500, &xcvr->afe_xcvr_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);
		}

		/* Power up TX and RX out from power down (PWRDNTX and
		 * PWRDNRX) & increase TX int & ext bias 20%....(0xe85c)
		 */
		if (is_a2(pdev))
			writel(0x000003F0, &xcvr->afe_channel_control);
		else if (is_b0(pdev)) {
			writel(0x000003D7, &xcvr->afe_channel_control);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(0x000003D4, &xcvr->afe_channel_control);
		} else if (is_c0(pdev)) {
			writel(0x000001E7, &xcvr->afe_channel_control);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(0x000001E4, &xcvr->afe_channel_control);
		} else if (is_c1(pdev)) {
			writel(cable_length_long ? 0x000002F7 : 0x000001F7,
			       &xcvr->afe_channel_control);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(cable_length_long ? 0x000002F4 : 0x000001F4,
			       &xcvr->afe_channel_control);
		}
		udelay(AFE_REGISTER_WRITE_DELAY);

		if (is_a2(pdev)) {
			/* Enable TX equalization (0xe824) */
			writel(0x00040000, &xcvr->afe_tx_control);
			udelay(AFE_REGISTER_WRITE_DELAY);
		}

		if (is_a2(pdev) || is_b0(pdev))
			/* RDPI=0x0(RX Power On), RXOOBDETPDNC=0x0,
			 * TPD=0x0(TX Power On), RDD=0x0(RX Detect
			 * Enabled) ....(0xe800)
			 */
			writel(0x00004100, &xcvr->afe_xcvr_control0);
		else if (is_c0(pdev))
			writel(0x00014100, &xcvr->afe_xcvr_control0);
		else if (is_c1(pdev))
			writel(0x0001C100, &xcvr->afe_xcvr_control0);
		udelay(AFE_REGISTER_WRITE_DELAY);

		/* Leave DFE/FFE on */
		if (is_a2(pdev))
			writel(0x3F11103F, &xcvr->afe_rx_ssc_control0);
		else if (is_b0(pdev)) {
			writel(0x3F11103F, &xcvr->afe_rx_ssc_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);
			/* Enable TX equalization (0xe824) */
			writel(0x00040000, &xcvr->afe_tx_control);
		} else if (is_c0(pdev)) {
			writel(0x01400C0F, &xcvr->afe_rx_ssc_control1);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(0x3F6F103F, &xcvr->afe_rx_ssc_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);

			/* Enable TX equalization (0xe824) */
			writel(0x00040000, &xcvr->afe_tx_control);
		} else if (is_c1(pdev)) {
			writel(cable_length_long ? 0x01500C0C :
			       cable_length_medium ? 0x01400C0D : 0x02400C0D,
			       &xcvr->afe_xcvr_control1);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(0x000003E0, &xcvr->afe_dfx_rx_control1);
			udelay(AFE_REGISTER_WRITE_DELAY);

			writel(cable_length_long ? 0x33091C1F :
			       cable_length_medium ? 0x3315181F : 0x2B17161F,
			       &xcvr->afe_rx_ssc_control0);
			udelay(AFE_REGISTER_WRITE_DELAY);

			/* Enable TX equalization (0xe824) */
			writel(0x00040000, &xcvr->afe_tx_control);
		}

		udelay(AFE_REGISTER_WRITE_DELAY);

		writel(oem_phy->afe_tx_amp_control0, &xcvr->afe_tx_amp_control0);
		udelay(AFE_REGISTER_WRITE_DELAY);

		writel(oem_phy->afe_tx_amp_control1, &xcvr->afe_tx_amp_control1);
		udelay(AFE_REGISTER_WRITE_DELAY);

		writel(oem_phy->afe_tx_amp_control2, &xcvr->afe_tx_amp_control2);
		udelay(AFE_REGISTER_WRITE_DELAY);

		writel(oem_phy->afe_tx_amp_control3, &xcvr->afe_tx_amp_control3);
		udelay(AFE_REGISTER_WRITE_DELAY);
	}

	/* Transfer control to the PEs */
	writel(0x00010f00, &afe->afe_dfx_master_control0);
	udelay(AFE_REGISTER_WRITE_DELAY);
}

static void sci_controller_initialize_power_control(struct isci_host *ihost)
{
	sci_init_timer(&ihost->power_control.timer, power_control_timeout);

	memset(ihost->power_control.requesters, 0,
	       sizeof(ihost->power_control.requesters));

	ihost->power_control.phys_waiting = 0;
	ihost->power_control.phys_granted_power = 0;
}

static enum sci_status sci_controller_initialize(struct isci_host *ihost)
{
	struct sci_base_state_machine *sm = &ihost->sm;
	enum sci_status result = SCI_FAILURE;
	unsigned long i, state, val;

	if (ihost->sm.current_state_id != SCIC_RESET) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	sci_change_state(sm, SCIC_INITIALIZING);

	sci_init_timer(&ihost->phy_timer, phy_startup_timeout);

	ihost->next_phy_to_start = 0;
	ihost->phy_startup_timer_pending = false;

	sci_controller_initialize_power_control(ihost);

	/*
	 * There is nothing to do here for B0 since we do not have to
	 * program the AFE registers.
	 * / @todo The AFE settings are supposed to be correct for the B0 but
	 * /       presently they seem to be wrong. */
	sci_controller_afe_initialization(ihost);


	/* Take the hardware out of reset */
	writel(0, &ihost->smu_registers->soft_reset_control);

	/*
	 * / @todo Provide meaningfull error code for hardware failure
	 * result = SCI_FAILURE_CONTROLLER_HARDWARE; */
	for (i = 100; i >= 1; i--) {
		u32 status;

		/* Loop until the hardware reports success */
		udelay(SCU_CONTEXT_RAM_INIT_STALL_TIME);
		status = readl(&ihost->smu_registers->control_status);

		if ((status & SCU_RAM_INIT_COMPLETED) == SCU_RAM_INIT_COMPLETED)
			break;
	}
	if (i == 0)
		goto out;

	/*
	 * Determine what are the actaul device capacities that the
	 * hardware will support */
	val = readl(&ihost->smu_registers->device_context_capacity);

	/* Record the smaller of the two capacity values */
	ihost->logical_port_entries = min(smu_max_ports(val), SCI_MAX_PORTS);
	ihost->task_context_entries = min(smu_max_task_contexts(val), SCI_MAX_IO_REQUESTS);
	ihost->remote_node_entries = min(smu_max_rncs(val), SCI_MAX_REMOTE_DEVICES);

	/*
	 * Make all PEs that are unassigned match up with the
	 * logical ports
	 */
	for (i = 0; i < ihost->logical_port_entries; i++) {
		struct scu_port_task_scheduler_group_registers __iomem
			*ptsg = &ihost->scu_registers->peg0.ptsg;

		writel(i, &ptsg->protocol_engine[i]);
	}

	/* Initialize hardware PCI Relaxed ordering in DMA engines */
	val = readl(&ihost->scu_registers->sdma.pdma_configuration);
	val |= SCU_PDMACR_GEN_BIT(PCI_RELAXED_ORDERING_ENABLE);
	writel(val, &ihost->scu_registers->sdma.pdma_configuration);

	val = readl(&ihost->scu_registers->sdma.cdma_configuration);
	val |= SCU_CDMACR_GEN_BIT(PCI_RELAXED_ORDERING_ENABLE);
	writel(val, &ihost->scu_registers->sdma.cdma_configuration);

	/*
	 * Initialize the PHYs before the PORTs because the PHY registers
	 * are accessed during the port initialization.
	 */
	for (i = 0; i < SCI_MAX_PHYS; i++) {
		result = sci_phy_initialize(&ihost->phys[i],
					    &ihost->scu_registers->peg0.pe[i].tl,
					    &ihost->scu_registers->peg0.pe[i].ll);
		if (result != SCI_SUCCESS)
			goto out;
	}

	for (i = 0; i < ihost->logical_port_entries; i++) {
		struct isci_port *iport = &ihost->ports[i];

		iport->port_task_scheduler_registers = &ihost->scu_registers->peg0.ptsg.port[i];
		iport->port_pe_configuration_register = &ihost->scu_registers->peg0.ptsg.protocol_engine[0];
		iport->viit_registers = &ihost->scu_registers->peg0.viit[i];
	}

	result = sci_port_configuration_agent_initialize(ihost, &ihost->port_agent);

 out:
	/* Advance the controller state machine */
	if (result == SCI_SUCCESS)
		state = SCIC_INITIALIZED;
	else
		state = SCIC_FAILED;
	sci_change_state(sm, state);

	return result;
}

static int sci_controller_dma_alloc(struct isci_host *ihost)
{
	struct device *dev = &ihost->pdev->dev;
	size_t size;
	int i;

	/* detect re-initialization */
	if (ihost->completion_queue)
		return 0;

	size = SCU_MAX_COMPLETION_QUEUE_ENTRIES * sizeof(u32);
	ihost->completion_queue = dmam_alloc_coherent(dev, size, &ihost->cq_dma,
						      GFP_KERNEL);
	if (!ihost->completion_queue)
		return -ENOMEM;

	size = ihost->remote_node_entries * sizeof(union scu_remote_node_context);
	ihost->remote_node_context_table = dmam_alloc_coherent(dev, size, &ihost->rnc_dma,
							       GFP_KERNEL);

	if (!ihost->remote_node_context_table)
		return -ENOMEM;

	size = ihost->task_context_entries * sizeof(struct scu_task_context),
	ihost->task_context_table = dmam_alloc_coherent(dev, size, &ihost->tc_dma,
							GFP_KERNEL);
	if (!ihost->task_context_table)
		return -ENOMEM;

	size = SCI_UFI_TOTAL_SIZE;
	ihost->ufi_buf = dmam_alloc_coherent(dev, size, &ihost->ufi_dma, GFP_KERNEL);
	if (!ihost->ufi_buf)
		return -ENOMEM;

	for (i = 0; i < SCI_MAX_IO_REQUESTS; i++) {
		struct isci_request *ireq;
		dma_addr_t dma;

		ireq = dmam_alloc_coherent(dev, sizeof(*ireq), &dma, GFP_KERNEL);
		if (!ireq)
			return -ENOMEM;

		ireq->tc = &ihost->task_context_table[i];
		ireq->owning_controller = ihost;
		spin_lock_init(&ireq->state_lock);
		ireq->request_daddr = dma;
		ireq->isci_host = ihost;
		ihost->reqs[i] = ireq;
	}

	return 0;
}

static int sci_controller_mem_init(struct isci_host *ihost)
{
	int err = sci_controller_dma_alloc(ihost);

	if (err)
		return err;

	writel(lower_32_bits(ihost->cq_dma), &ihost->smu_registers->completion_queue_lower);
	writel(upper_32_bits(ihost->cq_dma), &ihost->smu_registers->completion_queue_upper);

	writel(lower_32_bits(ihost->rnc_dma), &ihost->smu_registers->remote_node_context_lower);
	writel(upper_32_bits(ihost->rnc_dma), &ihost->smu_registers->remote_node_context_upper);

	writel(lower_32_bits(ihost->tc_dma), &ihost->smu_registers->host_task_table_lower);
	writel(upper_32_bits(ihost->tc_dma), &ihost->smu_registers->host_task_table_upper);

	sci_unsolicited_frame_control_construct(ihost);

	/*
	 * Inform the silicon as to the location of the UF headers and
	 * address table.
	 */
	writel(lower_32_bits(ihost->uf_control.headers.physical_address),
		&ihost->scu_registers->sdma.uf_header_base_address_lower);
	writel(upper_32_bits(ihost->uf_control.headers.physical_address),
		&ihost->scu_registers->sdma.uf_header_base_address_upper);

	writel(lower_32_bits(ihost->uf_control.address_table.physical_address),
		&ihost->scu_registers->sdma.uf_address_table_lower);
	writel(upper_32_bits(ihost->uf_control.address_table.physical_address),
		&ihost->scu_registers->sdma.uf_address_table_upper);

	return 0;
}

/**
 * isci_host_init - (re-)initialize hardware and internal (private) state
 * @ihost: host to init
 *
 * Any public facing objects (like asd_sas_port, and asd_sas_phys), or
 * one-time initialization objects like locks and waitqueues, are
 * not touched (they are initialized in isci_host_alloc)
 */
int isci_host_init(struct isci_host *ihost)
{
	int i, err;
	enum sci_status status;

	spin_lock_irq(&ihost->scic_lock);
	status = sci_controller_construct(ihost, scu_base(ihost), smu_base(ihost));
	spin_unlock_irq(&ihost->scic_lock);
	if (status != SCI_SUCCESS) {
		dev_err(&ihost->pdev->dev,
			"%s: sci_controller_construct failed - status = %x\n",
			__func__,
			status);
		return -ENODEV;
	}

	spin_lock_irq(&ihost->scic_lock);
	status = sci_controller_initialize(ihost);
	spin_unlock_irq(&ihost->scic_lock);
	if (status != SCI_SUCCESS) {
		dev_warn(&ihost->pdev->dev,
			 "%s: sci_controller_initialize failed -"
			 " status = 0x%x\n",
			 __func__, status);
		return -ENODEV;
	}

	err = sci_controller_mem_init(ihost);
	if (err)
		return err;

	/* enable sgpio */
	writel(1, &ihost->scu_registers->peg0.sgpio.interface_control);
	for (i = 0; i < isci_gpio_count(ihost); i++)
		writel(SGPIO_HW_CONTROL, &ihost->scu_registers->peg0.sgpio.output_data_select[i]);
	writel(0, &ihost->scu_registers->peg0.sgpio.vendor_specific_code);

	return 0;
}

void sci_controller_link_up(struct isci_host *ihost, struct isci_port *iport,
			    struct isci_phy *iphy)
{
	switch (ihost->sm.current_state_id) {
	case SCIC_STARTING:
		sci_del_timer(&ihost->phy_timer);
		ihost->phy_startup_timer_pending = false;
		ihost->port_agent.link_up_handler(ihost, &ihost->port_agent,
						  iport, iphy);
		sci_controller_start_next_phy(ihost);
		break;
	case SCIC_READY:
		ihost->port_agent.link_up_handler(ihost, &ihost->port_agent,
						  iport, iphy);
		break;
	default:
		dev_dbg(&ihost->pdev->dev,
			"%s: SCIC Controller linkup event from phy %d in "
			"unexpected state %d\n", __func__, iphy->phy_index,
			ihost->sm.current_state_id);
	}
}

void sci_controller_link_down(struct isci_host *ihost, struct isci_port *iport,
			      struct isci_phy *iphy)
{
	switch (ihost->sm.current_state_id) {
	case SCIC_STARTING:
	case SCIC_READY:
		ihost->port_agent.link_down_handler(ihost, &ihost->port_agent,
						   iport, iphy);
		break;
	default:
		dev_dbg(&ihost->pdev->dev,
			"%s: SCIC Controller linkdown event from phy %d in "
			"unexpected state %d\n",
			__func__,
			iphy->phy_index,
			ihost->sm.current_state_id);
	}
}

bool sci_controller_has_remote_devices_stopping(struct isci_host *ihost)
{
	u32 index;

	for (index = 0; index < ihost->remote_node_entries; index++) {
		if ((ihost->device_table[index] != NULL) &&
		   (ihost->device_table[index]->sm.current_state_id == SCI_DEV_STOPPING))
			return true;
	}

	return false;
}

void sci_controller_remote_device_stopped(struct isci_host *ihost,
					  struct isci_remote_device *idev)
{
	if (ihost->sm.current_state_id != SCIC_STOPPING) {
		dev_dbg(&ihost->pdev->dev,
			"SCIC Controller 0x%p remote device stopped event "
			"from device 0x%p in unexpected state %d\n",
			ihost, idev,
			ihost->sm.current_state_id);
		return;
	}

	if (!sci_controller_has_remote_devices_stopping(ihost))
		isci_host_stop_complete(ihost);
}

void sci_controller_post_request(struct isci_host *ihost, u32 request)
{
	dev_dbg(&ihost->pdev->dev, "%s[%d]: %#x\n",
		__func__, ihost->id, request);

	writel(request, &ihost->smu_registers->post_context_port);
}

struct isci_request *sci_request_by_tag(struct isci_host *ihost, u16 io_tag)
{
	u16 task_index;
	u16 task_sequence;

	task_index = ISCI_TAG_TCI(io_tag);

	if (task_index < ihost->task_context_entries) {
		struct isci_request *ireq = ihost->reqs[task_index];

		if (test_bit(IREQ_ACTIVE, &ireq->flags)) {
			task_sequence = ISCI_TAG_SEQ(io_tag);

			if (task_sequence == ihost->io_request_sequence[task_index])
				return ireq;
		}
	}

	return NULL;
}

/**
 * This method allocates remote node index and the reserves the remote node
 *    context space for use. This method can fail if there are no more remote
 *    node index available.
 * @scic: This is the controller object which contains the set of
 *    free remote node ids
 * @sci_dev: This is the device object which is requesting the a remote node
 *    id
 * @node_id: This is the remote node id that is assinged to the device if one
 *    is available
 *
 * enum sci_status SCI_FAILURE_OUT_OF_RESOURCES if there are no available remote
 * node index available.
 */
enum sci_status sci_controller_allocate_remote_node_context(struct isci_host *ihost,
							    struct isci_remote_device *idev,
							    u16 *node_id)
{
	u16 node_index;
	u32 remote_node_count = sci_remote_device_node_count(idev);

	node_index = sci_remote_node_table_allocate_remote_node(
		&ihost->available_remote_nodes, remote_node_count
		);

	if (node_index != SCIC_SDS_REMOTE_NODE_CONTEXT_INVALID_INDEX) {
		ihost->device_table[node_index] = idev;

		*node_id = node_index;

		return SCI_SUCCESS;
	}

	return SCI_FAILURE_INSUFFICIENT_RESOURCES;
}

void sci_controller_free_remote_node_context(struct isci_host *ihost,
					     struct isci_remote_device *idev,
					     u16 node_id)
{
	u32 remote_node_count = sci_remote_device_node_count(idev);

	if (ihost->device_table[node_id] == idev) {
		ihost->device_table[node_id] = NULL;

		sci_remote_node_table_release_remote_node_index(
			&ihost->available_remote_nodes, remote_node_count, node_id
			);
	}
}

void sci_controller_copy_sata_response(void *response_buffer,
				       void *frame_header,
				       void *frame_buffer)
{
	/* XXX type safety? */
	memcpy(response_buffer, frame_header, sizeof(u32));

	memcpy(response_buffer + sizeof(u32),
	       frame_buffer,
	       sizeof(struct dev_to_host_fis) - sizeof(u32));
}

void sci_controller_release_frame(struct isci_host *ihost, u32 frame_index)
{
	if (sci_unsolicited_frame_control_release_frame(&ihost->uf_control, frame_index))
		writel(ihost->uf_control.get,
			&ihost->scu_registers->sdma.unsolicited_frame_get_pointer);
}

void isci_tci_free(struct isci_host *ihost, u16 tci)
{
	u16 tail = ihost->tci_tail & (SCI_MAX_IO_REQUESTS-1);

	ihost->tci_pool[tail] = tci;
	ihost->tci_tail = tail + 1;
}

static u16 isci_tci_alloc(struct isci_host *ihost)
{
	u16 head = ihost->tci_head & (SCI_MAX_IO_REQUESTS-1);
	u16 tci = ihost->tci_pool[head];

	ihost->tci_head = head + 1;
	return tci;
}

static u16 isci_tci_space(struct isci_host *ihost)
{
	return CIRC_SPACE(ihost->tci_head, ihost->tci_tail, SCI_MAX_IO_REQUESTS);
}

u16 isci_alloc_tag(struct isci_host *ihost)
{
	if (isci_tci_space(ihost)) {
		u16 tci = isci_tci_alloc(ihost);
		u8 seq = ihost->io_request_sequence[tci];

		return ISCI_TAG(seq, tci);
	}

	return SCI_CONTROLLER_INVALID_IO_TAG;
}

enum sci_status isci_free_tag(struct isci_host *ihost, u16 io_tag)
{
	u16 tci = ISCI_TAG_TCI(io_tag);
	u16 seq = ISCI_TAG_SEQ(io_tag);

	/* prevent tail from passing head */
	if (isci_tci_active(ihost) == 0)
		return SCI_FAILURE_INVALID_IO_TAG;

	if (seq == ihost->io_request_sequence[tci]) {
		ihost->io_request_sequence[tci] = (seq+1) & (SCI_MAX_SEQ-1);

		isci_tci_free(ihost, tci);

		return SCI_SUCCESS;
	}
	return SCI_FAILURE_INVALID_IO_TAG;
}

enum sci_status sci_controller_start_io(struct isci_host *ihost,
					struct isci_remote_device *idev,
					struct isci_request *ireq)
{
	enum sci_status status;

	if (ihost->sm.current_state_id != SCIC_READY) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	status = sci_remote_device_start_io(ihost, idev, ireq);
	if (status != SCI_SUCCESS)
		return status;

	set_bit(IREQ_ACTIVE, &ireq->flags);
	sci_controller_post_request(ihost, ireq->post_context);
	return SCI_SUCCESS;
}

enum sci_status sci_controller_terminate_request(struct isci_host *ihost,
						 struct isci_remote_device *idev,
						 struct isci_request *ireq)
{
	/* terminate an ongoing (i.e. started) core IO request.  This does not
	 * abort the IO request at the target, but rather removes the IO
	 * request from the host controller.
	 */
	enum sci_status status;

	if (ihost->sm.current_state_id != SCIC_READY) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	status = sci_io_request_terminate(ireq);
	if (status != SCI_SUCCESS)
		return status;

	/*
	 * Utilize the original post context command and or in the POST_TC_ABORT
	 * request sub-type.
	 */
	sci_controller_post_request(ihost,
				    ireq->post_context | SCU_CONTEXT_COMMAND_REQUEST_POST_TC_ABORT);
	return SCI_SUCCESS;
}

/**
 * sci_controller_complete_io() - This method will perform core specific
 *    completion operations for an IO request.  After this method is invoked,
 *    the user should consider the IO request as invalid until it is properly
 *    reused (i.e. re-constructed).
 * @ihost: The handle to the controller object for which to complete the
 *    IO request.
 * @idev: The handle to the remote device object for which to complete
 *    the IO request.
 * @ireq: the handle to the io request object to complete.
 */
enum sci_status sci_controller_complete_io(struct isci_host *ihost,
					   struct isci_remote_device *idev,
					   struct isci_request *ireq)
{
	enum sci_status status;
	u16 index;

	switch (ihost->sm.current_state_id) {
	case SCIC_STOPPING:
		/* XXX: Implement this function */
		return SCI_FAILURE;
	case SCIC_READY:
		status = sci_remote_device_complete_io(ihost, idev, ireq);
		if (status != SCI_SUCCESS)
			return status;

		index = ISCI_TAG_TCI(ireq->io_tag);
		clear_bit(IREQ_ACTIVE, &ireq->flags);
		return SCI_SUCCESS;
	default:
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

}

enum sci_status sci_controller_continue_io(struct isci_request *ireq)
{
	struct isci_host *ihost = ireq->owning_controller;

	if (ihost->sm.current_state_id != SCIC_READY) {
		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
			 __func__, ihost->sm.current_state_id);
		return SCI_FAILURE_INVALID_STATE;
	}

	set_bit(IREQ_ACTIVE, &ireq->flags);
	sci_controller_post_request(ihost, ireq->post_context);
	return SCI_SUCCESS;
}

/**
 * sci_controller_start_task() - This method is called by the SCIC user to
 *    send/start a framework task management request.
 * @controller: the handle to the controller object for which to start the task
 *    management request.
 * @remote_device: the handle to the remote device object for which to start
 *    the task management request.
 * @task_request: the handle to the task request object to start.
 */
enum sci_task_status sci_controller_start_task(struct isci_host *ihost,
					       struct isci_remote_device *idev,
					       struct isci_request *ireq)
{
	enum sci_status status;

	if (ihost->sm.current_state_id != SCIC_READY) {
		dev_warn(&ihost->pdev->dev,
			 "%s: SCIC Controller starting task from invalid "
			 "state\n",
			 __func__);
		return SCI_TASK_FAILURE_INVALID_STATE;
	}

	status = sci_remote_device_start_task(ihost, idev, ireq);
	switch (status) {
	case SCI_FAILURE_RESET_DEVICE_PARTIAL_SUCCESS:
		set_bit(IREQ_ACTIVE, &ireq->flags);

		/*
		 * We will let framework know this task request started successfully,
		 * although core is still woring on starting the request (to post tc when
		 * RNC is resumed.)
		 */
		return SCI_SUCCESS;
	case SCI_SUCCESS:
		set_bit(IREQ_ACTIVE, &ireq->flags);
		sci_controller_post_request(ihost, ireq->post_context);
		break;
	default:
		break;
	}

	return status;
}

static int sci_write_gpio_tx_gp(struct isci_host *ihost, u8 reg_index, u8 reg_count, u8 *write_data)
{
	int d;

	/* no support for TX_GP_CFG */
	if (reg_index == 0)
		return -EINVAL;

	for (d = 0; d < isci_gpio_count(ihost); d++) {
		u32 val = 0x444; /* all ODx.n clear */
		int i;

		for (i = 0; i < 3; i++) {
			int bit = (i << 2) + 2;

			bit = try_test_sas_gpio_gp_bit(to_sas_gpio_od(d, i),
						       write_data, reg_index,
						       reg_count);
			if (bit < 0)
				break;

			/* if od is set, clear the 'invert' bit */
			val &= ~(bit << ((i << 2) + 2));
		}

		if (i < 3)
			break;
		writel(val, &ihost->scu_registers->peg0.sgpio.output_data_select[d]);
	}

	/* unless reg_index is > 1, we should always be able to write at
	 * least one register
	 */
	return d > 0;
}

int isci_gpio_write(struct sas_ha_struct *sas_ha, u8 reg_type, u8 reg_index,
		    u8 reg_count, u8 *write_data)
{
	struct isci_host *ihost = sas_ha->lldd_ha;
	int written;

	switch (reg_type) {
	case SAS_GPIO_REG_TX_GP:
		written = sci_write_gpio_tx_gp(ihost, reg_index, reg_count, write_data);
		break;
	default:
		written = -EINVAL;
	}

	return written;
}