// SPDX-License-Identifier: GPL-2.0-or-later
/*
 *  Linux MegaRAID driver for SAS based RAID controllers
 *
 *  Copyright (c) 2009-2013  LSI Corporation
 *  Copyright (c) 2013-2016  Avago Technologies
 *  Copyright (c) 2016-2018  Broadcom Inc.
 *
 *  FILE: megaraid_sas_fusion.c
 *
 *  Authors: Broadcom Inc.
 *           Sumant Patro
 *           Adam Radford
 *           Kashyap Desai <kashyap.desai@broadcom.com>
 *           Sumit Saxena <sumit.saxena@broadcom.com>
 *
 *  Send feedback to: megaraidlinux.pdl@broadcom.com
 */

#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/pci.h>
#include <linux/list.h>
#include <linux/moduleparam.h>
#include <linux/module.h>
#include <linux/spinlock.h>
#include <linux/interrupt.h>
#include <linux/delay.h>
#include <linux/uio.h>
#include <linux/uaccess.h>
#include <linux/fs.h>
#include <linux/compat.h>
#include <linux/blkdev.h>
#include <linux/mutex.h>
#include <linux/poll.h>
#include <linux/vmalloc.h>
#include <linux/workqueue.h>
#include <linux/irq_poll.h>

#include <scsi/scsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_dbg.h>
#include <linux/dmi.h>

#include "megaraid_sas_fusion.h"
#include "megaraid_sas.h"


extern void
megasas_complete_cmd(struct megasas_instance *instance,
                     struct megasas_cmd *cmd, u8 alt_status);
int
wait_and_poll(struct megasas_instance *instance, struct megasas_cmd *cmd,
              int seconds);

int
megasas_clear_intr_fusion(struct megasas_instance *instance);

int megasas_transition_to_ready(struct megasas_instance *instance, int ocr);

extern u32 megasas_dbg_lvl;
int megasas_sriov_start_heartbeat(struct megasas_instance *instance,
                                  int initial);
extern struct megasas_mgmt_info megasas_mgmt_info;
extern unsigned int resetwaittime;
extern unsigned int dual_qdepth_disable;
static void megasas_free_rdpq_fusion(struct megasas_instance *instance);
static void megasas_free_reply_fusion(struct megasas_instance *instance);
static inline
void megasas_configure_queue_sizes(struct megasas_instance *instance);
static void megasas_fusion_crash_dump(struct megasas_instance *instance);

/**
 * megasas_adp_reset_wait_for_ready -   initiate chip reset and wait for
 *                                      controller to come to ready state
 * @instance:                           adapter's soft state
 * @do_adp_reset:                       If true, do a chip reset
 * @ocr_context:                        If called from OCR context this will
 *                                      be set to 1, else 0
 *
 * This function initates a chip reset followed by a wait for controller to
 * transition to ready state.
 * During this, driver will block all access to PCI config space from userspace
 */
int
megasas_adp_reset_wait_for_ready(struct megasas_instance *instance,
                                 bool do_adp_reset,
                                 int ocr_context)
{
        int ret = FAILED;

        /*
         * Block access to PCI config space from userspace
         * when diag reset is initiated from driver
         */
        if (megasas_dbg_lvl & OCR_DEBUG)
                dev_info(&instance->pdev->dev,
                         "Block access to PCI config space %s %d\n",
                         __func__, __LINE__);

        pci_cfg_access_lock(instance->pdev);

        if (do_adp_reset) {
                if (instance->instancet->adp_reset
                        (instance, instance->reg_set))
                        goto out;
        }

        /* Wait for FW to become ready */
        if (megasas_transition_to_ready(instance, ocr_context)) {
                dev_warn(&instance->pdev->dev,
                         "Failed to transition controller to ready for scsi%d.\n",
                         instance->host->host_no);
                goto out;
        }

        ret = SUCCESS;
out:
        if (megasas_dbg_lvl & OCR_DEBUG)
                dev_info(&instance->pdev->dev,
                         "Unlock access to PCI config space %s %d\n",
                         __func__, __LINE__);

        pci_cfg_access_unlock(instance->pdev);

        return ret;
}

/**
 * megasas_check_same_4gb_region -      check if allocation
 *                                      crosses same 4GB boundary or not
 * @instance:                           adapter's soft instance
 * @start_addr:                         start address of DMA allocation
 * @size:                               size of allocation in bytes
 * @return:                             true : allocation does not cross same
 *                                      4GB boundary
 *                                      false: allocation crosses same
 *                                      4GB boundary
 */
static inline bool megasas_check_same_4gb_region
        (struct megasas_instance *instance, dma_addr_t start_addr, size_t size)
{
        dma_addr_t end_addr;

        end_addr = start_addr + size;

        if (upper_32_bits(start_addr) != upper_32_bits(end_addr)) {
                dev_err(&instance->pdev->dev,
                        "Failed to get same 4GB boundary: start_addr: 0x%llx end_addr: 0x%llx\n",
                        (unsigned long long)start_addr,
                        (unsigned long long)end_addr);
                return false;
        }

        return true;
}

/**
 * megasas_enable_intr_fusion - Enables interrupts
 * @instance:   adapter's soft instance
 */
static void
megasas_enable_intr_fusion(struct megasas_instance *instance)
{
        struct megasas_register_set __iomem *regs;
        regs = instance->reg_set;

        instance->mask_interrupts = 0;
        /* For Thunderbolt/Invader also clear intr on enable */
        writel(~0, &regs->outbound_intr_status);
        readl(&regs->outbound_intr_status);

        writel(~MFI_FUSION_ENABLE_INTERRUPT_MASK, &(regs)->outbound_intr_mask);

        /* Dummy readl to force pci flush */
        dev_info(&instance->pdev->dev, "%s is called outbound_intr_mask:0x%08x\n",
                 __func__, readl(&regs->outbound_intr_mask));
}

/**
 * megasas_disable_intr_fusion - Disables interrupt
 * @instance:   adapter's soft instance
 */
static void
megasas_disable_intr_fusion(struct megasas_instance *instance)
{
        u32 mask = 0xFFFFFFFF;
        struct megasas_register_set __iomem *regs;
        regs = instance->reg_set;
        instance->mask_interrupts = 1;

        writel(mask, &regs->outbound_intr_mask);
        /* Dummy readl to force pci flush */
        dev_info(&instance->pdev->dev, "%s is called outbound_intr_mask:0x%08x\n",
                 __func__, readl(&regs->outbound_intr_mask));
}

int
megasas_clear_intr_fusion(struct megasas_instance *instance)
{
        u32 status;
        struct megasas_register_set __iomem *regs;
        regs = instance->reg_set;
        /*
         * Check if it is our interrupt
         */
        status = megasas_readl(instance,
                               &regs->outbound_intr_status);

        if (status & 1) {
                writel(status, &regs->outbound_intr_status);
                readl(&regs->outbound_intr_status);
                return 1;
        }
        if (!(status & MFI_FUSION_ENABLE_INTERRUPT_MASK))
                return 0;

        return 1;
}

/**
 * megasas_get_cmd_fusion -     Get a command from the free pool
 * @instance:           Adapter soft state
 * @blk_tag:            Command tag
 *
 * Returns a blk_tag indexed mpt frame
 */
inline struct megasas_cmd_fusion *megasas_get_cmd_fusion(struct megasas_instance
                                                  *instance, u32 blk_tag)
{
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;
        return fusion->cmd_list[blk_tag];
}

/**
 * megasas_return_cmd_fusion -  Return a cmd to free command pool
 * @instance:           Adapter soft state
 * @cmd:                Command packet to be returned to free command pool
 */
inline void megasas_return_cmd_fusion(struct megasas_instance *instance,
        struct megasas_cmd_fusion *cmd)
{
        cmd->scmd = NULL;
        memset(cmd->io_request, 0, MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE);
        cmd->r1_alt_dev_handle = MR_DEVHANDLE_INVALID;
        cmd->cmd_completed = false;
}

/**
 * megasas_write_64bit_req_desc -       PCI writes 64bit request descriptor
 * @instance:                           Adapter soft state
 * @req_desc:                           64bit Request descriptor
 */
static void
megasas_write_64bit_req_desc(struct megasas_instance *instance,
                union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc)
{
#if defined(writeq) && defined(CONFIG_64BIT)
        u64 req_data = (((u64)le32_to_cpu(req_desc->u.high) << 32) |
                le32_to_cpu(req_desc->u.low));
        writeq(req_data, &instance->reg_set->inbound_low_queue_port);
#else
        unsigned long flags;
        spin_lock_irqsave(&instance->hba_lock, flags);
        writel(le32_to_cpu(req_desc->u.low),
                &instance->reg_set->inbound_low_queue_port);
        writel(le32_to_cpu(req_desc->u.high),
                &instance->reg_set->inbound_high_queue_port);
        spin_unlock_irqrestore(&instance->hba_lock, flags);
#endif
}

/**
 * megasas_fire_cmd_fusion -    Sends command to the FW
 * @instance:                   Adapter soft state
 * @req_desc:                   32bit or 64bit Request descriptor
 *
 * Perform PCI Write. AERO SERIES supports 32 bit Descriptor.
 * Prior to AERO_SERIES support 64 bit Descriptor.
 */
static void
megasas_fire_cmd_fusion(struct megasas_instance *instance,
                union MEGASAS_REQUEST_DESCRIPTOR_UNION *req_desc)
{
        if (instance->atomic_desc_support)
                writel(le32_to_cpu(req_desc->u.low),
                        &instance->reg_set->inbound_single_queue_port);
        else
                megasas_write_64bit_req_desc(instance, req_desc);
}

/**
 * megasas_fusion_update_can_queue -    Do all Adapter Queue depth related calculations here
 * @instance:           Adapter soft state
 * @fw_boot_context:    Whether this function called during probe or after OCR
 *
 * This function is only for fusion controllers.
 * Update host can queue, if firmware downgrade max supported firmware commands.
 * Firmware upgrade case will be skiped because underlying firmware has
 * more resource than exposed to the OS.
 *
 */
static void
megasas_fusion_update_can_queue(struct megasas_instance *instance, int fw_boot_context)
{
        u16 cur_max_fw_cmds = 0;
        u16 ldio_threshold = 0;

        /* ventura FW does not fill outbound_scratch_pad_2 with queue depth */
        if (instance->adapter_type < VENTURA_SERIES)
                cur_max_fw_cmds =
                megasas_readl(instance,
                              &instance->reg_set->outbound_scratch_pad_2) & 0x00FFFF;

        if (dual_qdepth_disable || !cur_max_fw_cmds)
                cur_max_fw_cmds = instance->instancet->read_fw_status_reg(instance) & 0x00FFFF;
        else
                ldio_threshold =
                        (instance->instancet->read_fw_status_reg(instance) & 0x00FFFF) - MEGASAS_FUSION_IOCTL_CMDS;

        dev_info(&instance->pdev->dev,
                 "Current firmware supports maximum commands: %d\t LDIO threshold: %d\n",
                 cur_max_fw_cmds, ldio_threshold);

        if (fw_boot_context == OCR_CONTEXT) {
                cur_max_fw_cmds = cur_max_fw_cmds - 1;
                if (cur_max_fw_cmds < instance->max_fw_cmds) {
                        instance->cur_can_queue =
                                cur_max_fw_cmds - (MEGASAS_FUSION_INTERNAL_CMDS +
                                                MEGASAS_FUSION_IOCTL_CMDS);
                        instance->host->can_queue = instance->cur_can_queue;
                        instance->ldio_threshold = ldio_threshold;
                }
        } else {
                instance->max_fw_cmds = cur_max_fw_cmds;
                instance->ldio_threshold = ldio_threshold;

                if (reset_devices)
                        instance->max_fw_cmds = min(instance->max_fw_cmds,
                                                (u16)MEGASAS_KDUMP_QUEUE_DEPTH);
                /*
                * Reduce the max supported cmds by 1. This is to ensure that the
                * reply_q_sz (1 more than the max cmd that driver may send)
                * does not exceed max cmds that the FW can support
                */
                instance->max_fw_cmds = instance->max_fw_cmds-1;
        }
}

static inline void
megasas_get_msix_index(struct megasas_instance *instance,
                       struct scsi_cmnd *scmd,
                       struct megasas_cmd_fusion *cmd,
                       u8 data_arms)
{
        int sdev_busy;

        /* nr_hw_queue = 1 for MegaRAID */
        struct blk_mq_hw_ctx *hctx =
                scmd->device->request_queue->queue_hw_ctx[0];

        sdev_busy = atomic_read(&hctx->nr_active);

        if (instance->perf_mode == MR_BALANCED_PERF_MODE &&
            sdev_busy > (data_arms * MR_DEVICE_HIGH_IOPS_DEPTH))
                cmd->request_desc->SCSIIO.MSIxIndex =
                        mega_mod64((atomic64_add_return(1, &instance->high_iops_outstanding) /
                                        MR_HIGH_IOPS_BATCH_COUNT), instance->low_latency_index_start);
        else if (instance->msix_load_balance)
                cmd->request_desc->SCSIIO.MSIxIndex =
                        (mega_mod64(atomic64_add_return(1, &instance->total_io_count),
                                instance->msix_vectors));
        else
                cmd->request_desc->SCSIIO.MSIxIndex =
                        instance->reply_map[raw_smp_processor_id()];
}

/**
 * megasas_free_cmds_fusion -   Free all the cmds in the free cmd pool
 * @instance:           Adapter soft state
 */
void
megasas_free_cmds_fusion(struct megasas_instance *instance)
{
        int i;
        struct fusion_context *fusion = instance->ctrl_context;
        struct megasas_cmd_fusion *cmd;

        if (fusion->sense)
                dma_pool_free(fusion->sense_dma_pool, fusion->sense,
                              fusion->sense_phys_addr);

        /* SG */
        if (fusion->cmd_list) {
                for (i = 0; i < instance->max_mpt_cmds; i++) {
                        cmd = fusion->cmd_list[i];
                        if (cmd) {
                                if (cmd->sg_frame)
                                        dma_pool_free(fusion->sg_dma_pool,
                                                      cmd->sg_frame,
                                                      cmd->sg_frame_phys_addr);
                        }
                        kfree(cmd);
                }
                kfree(fusion->cmd_list);
        }

        if (fusion->sg_dma_pool) {
                dma_pool_destroy(fusion->sg_dma_pool);
                fusion->sg_dma_pool = NULL;
        }
        if (fusion->sense_dma_pool) {
                dma_pool_destroy(fusion->sense_dma_pool);
                fusion->sense_dma_pool = NULL;
        }


        /* Reply Frame, Desc*/
        if (instance->is_rdpq)
                megasas_free_rdpq_fusion(instance);
        else
                megasas_free_reply_fusion(instance);

        /* Request Frame, Desc*/
        if (fusion->req_frames_desc)
                dma_free_coherent(&instance->pdev->dev,
                        fusion->request_alloc_sz, fusion->req_frames_desc,
                        fusion->req_frames_desc_phys);
        if (fusion->io_request_frames)
                dma_pool_free(fusion->io_request_frames_pool,
                        fusion->io_request_frames,
                        fusion->io_request_frames_phys);
        if (fusion->io_request_frames_pool) {
                dma_pool_destroy(fusion->io_request_frames_pool);
                fusion->io_request_frames_pool = NULL;
        }
}

/**
 * megasas_create_sg_sense_fusion -     Creates DMA pool for cmd frames
 * @instance:                   Adapter soft state
 *
 */
static int megasas_create_sg_sense_fusion(struct megasas_instance *instance)
{
        int i;
        u16 max_cmd;
        struct fusion_context *fusion;
        struct megasas_cmd_fusion *cmd;
        int sense_sz;
        u32 offset;

        fusion = instance->ctrl_context;
        max_cmd = instance->max_fw_cmds;
        sense_sz = instance->max_mpt_cmds * SCSI_SENSE_BUFFERSIZE;

        fusion->sg_dma_pool =
                        dma_pool_create("mr_sg", &instance->pdev->dev,
                                instance->max_chain_frame_sz,
                                MR_DEFAULT_NVME_PAGE_SIZE, 0);
        /* SCSI_SENSE_BUFFERSIZE  = 96 bytes */
        fusion->sense_dma_pool =
                        dma_pool_create("mr_sense", &instance->pdev->dev,
                                sense_sz, 64, 0);

        if (!fusion->sense_dma_pool || !fusion->sg_dma_pool) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        fusion->sense = dma_pool_alloc(fusion->sense_dma_pool,
                                       GFP_KERNEL, &fusion->sense_phys_addr);
        if (!fusion->sense) {
                dev_err(&instance->pdev->dev,
                        "failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        /* sense buffer, request frame and reply desc pool requires to be in
         * same 4 gb region. Below function will check this.
         * In case of failure, new pci pool will be created with updated
         * alignment.
         * Older allocation and pool will be destroyed.
         * Alignment will be used such a way that next allocation if success,
         * will always meet same 4gb region requirement.
         * Actual requirement is not alignment, but we need start and end of
         * DMA address must have same upper 32 bit address.
         */

        if (!megasas_check_same_4gb_region(instance, fusion->sense_phys_addr,
                                           sense_sz)) {
                dma_pool_free(fusion->sense_dma_pool, fusion->sense,
                              fusion->sense_phys_addr);
                fusion->sense = NULL;
                dma_pool_destroy(fusion->sense_dma_pool);

                fusion->sense_dma_pool =
                        dma_pool_create("mr_sense_align", &instance->pdev->dev,
                                        sense_sz, roundup_pow_of_two(sense_sz),
                                        0);
                if (!fusion->sense_dma_pool) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
                fusion->sense = dma_pool_alloc(fusion->sense_dma_pool,
                                               GFP_KERNEL,
                                               &fusion->sense_phys_addr);
                if (!fusion->sense) {
                        dev_err(&instance->pdev->dev,
                                "failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        /*
         * Allocate and attach a frame to each of the commands in cmd_list
         */
        for (i = 0; i < max_cmd; i++) {
                cmd = fusion->cmd_list[i];
                cmd->sg_frame = dma_pool_alloc(fusion->sg_dma_pool,
                                        GFP_KERNEL, &cmd->sg_frame_phys_addr);

                offset = SCSI_SENSE_BUFFERSIZE * i;
                cmd->sense = (u8 *)fusion->sense + offset;
                cmd->sense_phys_addr = fusion->sense_phys_addr + offset;

                if (!cmd->sg_frame) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        /* create sense buffer for the raid 1/10 fp */
        for (i = max_cmd; i < instance->max_mpt_cmds; i++) {
                cmd = fusion->cmd_list[i];
                offset = SCSI_SENSE_BUFFERSIZE * i;
                cmd->sense = (u8 *)fusion->sense + offset;
                cmd->sense_phys_addr = fusion->sense_phys_addr + offset;

        }

        return 0;
}

static int
megasas_alloc_cmdlist_fusion(struct megasas_instance *instance)
{
        u32 max_mpt_cmd, i, j;
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

        max_mpt_cmd = instance->max_mpt_cmds;

        /*
         * fusion->cmd_list is an array of struct megasas_cmd_fusion pointers.
         * Allocate the dynamic array first and then allocate individual
         * commands.
         */
        fusion->cmd_list =
                kcalloc(max_mpt_cmd, sizeof(struct megasas_cmd_fusion *),
                        GFP_KERNEL);
        if (!fusion->cmd_list) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        for (i = 0; i < max_mpt_cmd; i++) {
                fusion->cmd_list[i] = kzalloc(sizeof(struct megasas_cmd_fusion),
                                              GFP_KERNEL);
                if (!fusion->cmd_list[i]) {
                        for (j = 0; j < i; j++)
                                kfree(fusion->cmd_list[j]);
                        kfree(fusion->cmd_list);
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        return 0;
}

static int
megasas_alloc_request_fusion(struct megasas_instance *instance)
{
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

retry_alloc:
        fusion->io_request_frames_pool =
                        dma_pool_create("mr_ioreq", &instance->pdev->dev,
                                fusion->io_frames_alloc_sz, 16, 0);

        if (!fusion->io_request_frames_pool) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        fusion->io_request_frames =
                        dma_pool_alloc(fusion->io_request_frames_pool,
                                GFP_KERNEL | __GFP_NOWARN,
                                &fusion->io_request_frames_phys);
        if (!fusion->io_request_frames) {
                if (instance->max_fw_cmds >= (MEGASAS_REDUCE_QD_COUNT * 2)) {
                        instance->max_fw_cmds -= MEGASAS_REDUCE_QD_COUNT;
                        dma_pool_destroy(fusion->io_request_frames_pool);
                        megasas_configure_queue_sizes(instance);
                        goto retry_alloc;
                } else {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        if (!megasas_check_same_4gb_region(instance,
                                           fusion->io_request_frames_phys,
                                           fusion->io_frames_alloc_sz)) {
                dma_pool_free(fusion->io_request_frames_pool,
                              fusion->io_request_frames,
                              fusion->io_request_frames_phys);
                fusion->io_request_frames = NULL;
                dma_pool_destroy(fusion->io_request_frames_pool);

                fusion->io_request_frames_pool =
                        dma_pool_create("mr_ioreq_align",
                                        &instance->pdev->dev,
                                        fusion->io_frames_alloc_sz,
                                        roundup_pow_of_two(fusion->io_frames_alloc_sz),
                                        0);

                if (!fusion->io_request_frames_pool) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }

                fusion->io_request_frames =
                        dma_pool_alloc(fusion->io_request_frames_pool,
                                       GFP_KERNEL | __GFP_NOWARN,
                                       &fusion->io_request_frames_phys);

                if (!fusion->io_request_frames) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        fusion->req_frames_desc =
                dma_alloc_coherent(&instance->pdev->dev,
                                   fusion->request_alloc_sz,
                                   &fusion->req_frames_desc_phys, GFP_KERNEL);
        if (!fusion->req_frames_desc) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        return 0;
}

static int
megasas_alloc_reply_fusion(struct megasas_instance *instance)
{
        int i, count;
        struct fusion_context *fusion;
        union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
        fusion = instance->ctrl_context;

        count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;
        fusion->reply_frames_desc_pool =
                        dma_pool_create("mr_reply", &instance->pdev->dev,
                                fusion->reply_alloc_sz * count, 16, 0);

        if (!fusion->reply_frames_desc_pool) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        fusion->reply_frames_desc[0] =
                dma_pool_alloc(fusion->reply_frames_desc_pool,
                        GFP_KERNEL, &fusion->reply_frames_desc_phys[0]);
        if (!fusion->reply_frames_desc[0]) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        if (!megasas_check_same_4gb_region(instance,
                                           fusion->reply_frames_desc_phys[0],
                                           (fusion->reply_alloc_sz * count))) {
                dma_pool_free(fusion->reply_frames_desc_pool,
                              fusion->reply_frames_desc[0],
                              fusion->reply_frames_desc_phys[0]);
                fusion->reply_frames_desc[0] = NULL;
                dma_pool_destroy(fusion->reply_frames_desc_pool);

                fusion->reply_frames_desc_pool =
                        dma_pool_create("mr_reply_align",
                                        &instance->pdev->dev,
                                        fusion->reply_alloc_sz * count,
                                        roundup_pow_of_two(fusion->reply_alloc_sz * count),
                                        0);

                if (!fusion->reply_frames_desc_pool) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }

                fusion->reply_frames_desc[0] =
                        dma_pool_alloc(fusion->reply_frames_desc_pool,
                                       GFP_KERNEL,
                                       &fusion->reply_frames_desc_phys[0]);

                if (!fusion->reply_frames_desc[0]) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
        }

        reply_desc = fusion->reply_frames_desc[0];
        for (i = 0; i < fusion->reply_q_depth * count; i++, reply_desc++)
                reply_desc->Words = cpu_to_le64(ULLONG_MAX);

        /* This is not a rdpq mode, but driver still populate
         * reply_frame_desc array to use same msix index in ISR path.
         */
        for (i = 0; i < (count - 1); i++)
                fusion->reply_frames_desc[i + 1] =
                        fusion->reply_frames_desc[i] +
                        (fusion->reply_alloc_sz)/sizeof(union MPI2_REPLY_DESCRIPTORS_UNION);

        return 0;
}

static int
megasas_alloc_rdpq_fusion(struct megasas_instance *instance)
{
        int i, j, k, msix_count;
        struct fusion_context *fusion;
        union MPI2_REPLY_DESCRIPTORS_UNION *reply_desc;
        union MPI2_REPLY_DESCRIPTORS_UNION *rdpq_chunk_virt[RDPQ_MAX_CHUNK_COUNT];
        dma_addr_t rdpq_chunk_phys[RDPQ_MAX_CHUNK_COUNT];
        u8 dma_alloc_count, abs_index;
        u32 chunk_size, array_size, offset;

        fusion = instance->ctrl_context;
        chunk_size = fusion->reply_alloc_sz * RDPQ_MAX_INDEX_IN_ONE_CHUNK;
        array_size = sizeof(struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY) *
                     MAX_MSIX_QUEUES_FUSION;

        fusion->rdpq_virt = dma_alloc_coherent(&instance->pdev->dev,
                                               array_size, &fusion->rdpq_phys,
                                               GFP_KERNEL);
        if (!fusion->rdpq_virt) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

        msix_count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;

        fusion->reply_frames_desc_pool = dma_pool_create("mr_rdpq",
                                                         &instance->pdev->dev,
                                                         chunk_size, 16, 0);
        fusion->reply_frames_desc_pool_align =
                                dma_pool_create("mr_rdpq_align",
                                                &instance->pdev->dev,
                                                chunk_size,
                                                roundup_pow_of_two(chunk_size),
                                                0);

        if (!fusion->reply_frames_desc_pool ||
            !fusion->reply_frames_desc_pool_align) {
                dev_err(&instance->pdev->dev,
                        "Failed from %s %d\n",  __func__, __LINE__);
                return -ENOMEM;
        }

/*
 * For INVADER_SERIES each set of 8 reply queues(0-7, 8-15, ..) and
 * VENTURA_SERIES each set of 16 reply queues(0-15, 16-31, ..) should be
 * within 4GB boundary and also reply queues in a set must have same
 * upper 32-bits in their memory address. so here driver is allocating the
 * DMA'able memory for reply queues according. Driver uses limitation of
 * VENTURA_SERIES to manage INVADER_SERIES as well.
 */
        dma_alloc_count = DIV_ROUND_UP(msix_count, RDPQ_MAX_INDEX_IN_ONE_CHUNK);

        for (i = 0; i < dma_alloc_count; i++) {
                rdpq_chunk_virt[i] =
                        dma_pool_alloc(fusion->reply_frames_desc_pool,
                                       GFP_KERNEL, &rdpq_chunk_phys[i]);
                if (!rdpq_chunk_virt[i]) {
                        dev_err(&instance->pdev->dev,
                                "Failed from %s %d\n",  __func__, __LINE__);
                        return -ENOMEM;
                }
                /* reply desc pool requires to be in same 4 gb region.
                 * Below function will check this.
                 * In case of failure, new pci pool will be created with updated
                 * alignment.
                 * For RDPQ buffers, driver always allocate two separate pci pool.
                 * Alignment will be used such a way that next allocation if
                 * success, will always meet same 4gb region requirement.
                 * rdpq_tracker keep track of each buffer's physical,
                 * virtual address and pci pool descriptor. It will help driver
                 * while freeing the resources.
                 *
                 */
                if (!megasas_check_same_4gb_region(instance, rdpq_chunk_phys[i],
                                                   chunk_size)) {
                        dma_pool_free(fusion->reply_frames_desc_pool,
                                      rdpq_chunk_virt[i],
                                      rdpq_chunk_phys[i]);

                        rdpq_chunk_virt[i] =
                                dma_pool_alloc(fusion->reply_frames_desc_pool_align,
                                               GFP_KERNEL, &rdpq_chunk_phys[i]);
                        if (!rdpq_chunk_virt[i]) {
                                dev_err(&instance->pdev->dev,
                                        "Failed from %s %d\n",
                                        __func__, __LINE__);
                                return -ENOMEM;
                        }
                        fusion->rdpq_tracker[i].dma_pool_ptr =
                                        fusion->reply_frames_desc_pool_align;
                } else {
                        fusion->rdpq_tracker[i].dma_pool_ptr =
                                        fusion->reply_frames_desc_pool;
                }

                fusion->rdpq_tracker[i].pool_entry_phys = rdpq_chunk_phys[i];
                fusion->rdpq_tracker[i].pool_entry_virt = rdpq_chunk_virt[i];
        }

        for (k = 0; k < dma_alloc_count; k++) {
                for (i = 0; i < RDPQ_MAX_INDEX_IN_ONE_CHUNK; i++) {
                        abs_index = (k * RDPQ_MAX_INDEX_IN_ONE_CHUNK) + i;

                        if (abs_index == msix_count)
                                break;
                        offset = fusion->reply_alloc_sz * i;
                        fusion->rdpq_virt[abs_index].RDPQBaseAddress =
                                        cpu_to_le64(rdpq_chunk_phys[k] + offset);
                        fusion->reply_frames_desc_phys[abs_index] =
                                        rdpq_chunk_phys[k] + offset;
                        fusion->reply_frames_desc[abs_index] =
                                        (union MPI2_REPLY_DESCRIPTORS_UNION *)((u8 *)rdpq_chunk_virt[k] + offset);

                        reply_desc = fusion->reply_frames_desc[abs_index];
                        for (j = 0; j < fusion->reply_q_depth; j++, reply_desc++)
                                reply_desc->Words = ULLONG_MAX;
                }
        }

        return 0;
}

static void
megasas_free_rdpq_fusion(struct megasas_instance *instance) {

        int i;
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

        for (i = 0; i < RDPQ_MAX_CHUNK_COUNT; i++) {
                if (fusion->rdpq_tracker[i].pool_entry_virt)
                        dma_pool_free(fusion->rdpq_tracker[i].dma_pool_ptr,
                                      fusion->rdpq_tracker[i].pool_entry_virt,
                                      fusion->rdpq_tracker[i].pool_entry_phys);

        }

        dma_pool_destroy(fusion->reply_frames_desc_pool);
        dma_pool_destroy(fusion->reply_frames_desc_pool_align);

        if (fusion->rdpq_virt)
                dma_free_coherent(&instance->pdev->dev,
                        sizeof(struct MPI2_IOC_INIT_RDPQ_ARRAY_ENTRY) * MAX_MSIX_QUEUES_FUSION,
                        fusion->rdpq_virt, fusion->rdpq_phys);
}

static void
megasas_free_reply_fusion(struct megasas_instance *instance) {

        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

        if (fusion->reply_frames_desc[0])
                dma_pool_free(fusion->reply_frames_desc_pool,
                        fusion->reply_frames_desc[0],
                        fusion->reply_frames_desc_phys[0]);

        dma_pool_destroy(fusion->reply_frames_desc_pool);

}


/**
 * megasas_alloc_cmds_fusion -  Allocates the command packets
 * @instance:           Adapter soft state
 *
 *
 * Each frame has a 32-bit field called context. This context is used to get
 * back the megasas_cmd_fusion from the frame when a frame gets completed
 * In this driver, the 32 bit values are the indices into an array cmd_list.
 * This array is used only to look up the megasas_cmd_fusion given the context.
 * The free commands themselves are maintained in a linked list called cmd_pool.
 *
 * cmds are formed in the io_request and sg_frame members of the
 * megasas_cmd_fusion. The context field is used to get a request descriptor
 * and is used as SMID of the cmd.
 * SMID value range is from 1 to max_fw_cmds.
 */
static int
megasas_alloc_cmds_fusion(struct megasas_instance *instance)
{
        int i;
        struct fusion_context *fusion;
        struct megasas_cmd_fusion *cmd;
        u32 offset;
        dma_addr_t io_req_base_phys;
        u8 *io_req_base;


        fusion = instance->ctrl_context;

        if (megasas_alloc_request_fusion(instance))
                goto fail_exit;

        if (instance->is_rdpq) {
                if (megasas_alloc_rdpq_fusion(instance))
                        goto fail_exit;
        } else
                if (megasas_alloc_reply_fusion(instance))
                        goto fail_exit;

        if (megasas_alloc_cmdlist_fusion(instance))
                goto fail_exit;

        dev_info(&instance->pdev->dev, "Configured max firmware commands: %d\n",
                 instance->max_fw_cmds);

        /* The first 256 bytes (SMID 0) is not used. Don't add to the cmd list */
        io_req_base = fusion->io_request_frames + MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;
        io_req_base_phys = fusion->io_request_frames_phys + MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE;

        /*
         * Add all the commands to command pool (fusion->cmd_pool)
         */

        /* SMID 0 is reserved. Set SMID/index from 1 */
        for (i = 0; i < instance->max_mpt_cmds; i++) {
                cmd = fusion->cmd_list[i];
                offset = MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE * i;
                memset(cmd, 0, sizeof(struct megasas_cmd_fusion));
                cmd->index = i + 1;
                cmd->scmd = NULL;
                cmd->sync_cmd_idx =
                (i >= instance->max_scsi_cmds && i < instance->max_fw_cmds) ?
                                (i - instance->max_scsi_cmds) :
                                (u32)ULONG_MAX; /* Set to Invalid */
                cmd->instance = instance;
                cmd->io_request =
                        (struct MPI2_RAID_SCSI_IO_REQUEST *)
                  (io_req_base + offset);
                memset(cmd->io_request, 0,
                       sizeof(struct MPI2_RAID_SCSI_IO_REQUEST));
                cmd->io_request_phys_addr = io_req_base_phys + offset;
                cmd->r1_alt_dev_handle = MR_DEVHANDLE_INVALID;
        }

        if (megasas_create_sg_sense_fusion(instance))
                goto fail_exit;

        return 0;

fail_exit:
        megasas_free_cmds_fusion(instance);
        return -ENOMEM;
}

/**
 * wait_and_poll -      Issues a polling command
 * @instance:                   Adapter soft state
 * @cmd:                        Command packet to be issued
 * @seconds:                    Maximum poll time
 *
 * For polling, MFI requires the cmd_status to be set to 0xFF before posting.
 */
int
wait_and_poll(struct megasas_instance *instance, struct megasas_cmd *cmd,
        int seconds)
{
        int i;
        struct megasas_header *frame_hdr = &cmd->frame->hdr;
        u32 status_reg;

        u32 msecs = seconds * 1000;

        /*
         * Wait for cmd_status to change
         */
        for (i = 0; (i < msecs) && (frame_hdr->cmd_status == 0xff); i += 20) {
                rmb();
                msleep(20);
                if (!(i % 5000)) {
                        status_reg = instance->instancet->read_fw_status_reg(instance)
                                        & MFI_STATE_MASK;
                        if (status_reg == MFI_STATE_FAULT)
                                break;
                }
        }

        if (frame_hdr->cmd_status == MFI_STAT_INVALID_STATUS)
                return DCMD_TIMEOUT;
        else if (frame_hdr->cmd_status == MFI_STAT_OK)
                return DCMD_SUCCESS;
        else
                return DCMD_FAILED;
}

/**
 * megasas_ioc_init_fusion -    Initializes the FW
 * @instance:           Adapter soft state
 *
 * Issues the IOC Init cmd
 */
int
megasas_ioc_init_fusion(struct megasas_instance *instance)
{
        struct megasas_init_frame *init_frame;
        struct MPI2_IOC_INIT_REQUEST *IOCInitMessage = NULL;
        dma_addr_t      ioc_init_handle;
        struct megasas_cmd *cmd;
        u8 ret, cur_rdpq_mode;
        struct fusion_context *fusion;
        union MEGASAS_REQUEST_DESCRIPTOR_UNION req_desc;
        int i;
        struct megasas_header *frame_hdr;
        const char *sys_info;
        MFI_CAPABILITIES *drv_ops;
        u32 scratch_pad_1;
        ktime_t time;
        bool cur_fw_64bit_dma_capable;
        bool cur_intr_coalescing;

        fusion = instance->ctrl_context;

        ioc_init_handle = fusion->ioc_init_request_phys;
        IOCInitMessage = fusion->ioc_init_request;

        cmd = fusion->ioc_init_cmd;

        scratch_pad_1 = megasas_readl
                (instance, &instance->reg_set->outbound_scratch_pad_1);

        cur_rdpq_mode = (scratch_pad_1 & MR_RDPQ_MODE_OFFSET) ? 1 : 0;

        if (instance->adapter_type == INVADER_SERIES) {
                cur_fw_64bit_dma_capable =
                        (scratch_pad_1 & MR_CAN_HANDLE_64_BIT_DMA_OFFSET) ? true : false;

                if (instance->consistent_mask_64bit && !cur_fw_64bit_dma_capable) {
                        dev_err(&instance->pdev->dev, "Driver was operating on 64bit "
                                "DMA mask, but upcoming FW does not support 64bit DMA mask\n");
                        megaraid_sas_kill_hba(instance);
                        ret = 1;
                        goto fail_fw_init;
                }
        }

        if (instance->is_rdpq && !cur_rdpq_mode) {
                dev_err(&instance->pdev->dev, "Firmware downgrade *NOT SUPPORTED*"
                        " from RDPQ mode to non RDPQ mode\n");
                ret = 1;
                goto fail_fw_init;
        }

        cur_intr_coalescing = (scratch_pad_1 & MR_INTR_COALESCING_SUPPORT_OFFSET) ?
                                                        true : false;

        if ((instance->low_latency_index_start ==
                MR_HIGH_IOPS_QUEUE_COUNT) && cur_intr_coalescing)
                instance->perf_mode = MR_BALANCED_PERF_MODE;

        dev_info(&instance->pdev->dev, "Performance mode :%s\n",
                MEGASAS_PERF_MODE_2STR(instance->perf_mode));

        instance->fw_sync_cache_support = (scratch_pad_1 &
                MR_CAN_HANDLE_SYNC_CACHE_OFFSET) ? 1 : 0;
        dev_info(&instance->pdev->dev, "FW supports sync cache\t: %s\n",
                 instance->fw_sync_cache_support ? "Yes" : "No");

        memset(IOCInitMessage, 0, sizeof(struct MPI2_IOC_INIT_REQUEST));

        IOCInitMessage->Function = MPI2_FUNCTION_IOC_INIT;
        IOCInitMessage->WhoInit = MPI2_WHOINIT_HOST_DRIVER;
        IOCInitMessage->MsgVersion = cpu_to_le16(MPI2_VERSION);
        IOCInitMessage->HeaderVersion = cpu_to_le16(MPI2_HEADER_VERSION);
        IOCInitMessage->SystemRequestFrameSize = cpu_to_le16(MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE / 4);

        IOCInitMessage->ReplyDescriptorPostQueueDepth = cpu_to_le16(fusion->reply_q_depth);
        IOCInitMessage->ReplyDescriptorPostQueueAddress = instance->is_rdpq ?
                        cpu_to_le64(fusion->rdpq_phys) :
                        cpu_to_le64(fusion->reply_frames_desc_phys[0]);
        IOCInitMessage->MsgFlags = instance->is_rdpq ?
                        MPI2_IOCINIT_MSGFLAG_RDPQ_ARRAY_MODE : 0;
        IOCInitMessage->SystemRequestFrameBaseAddress = cpu_to_le64(fusion->io_request_frames_phys);
        IOCInitMessage->SenseBufferAddressHigh = cpu_to_le32(upper_32_bits(fusion->sense_phys_addr));
        IOCInitMessage->HostMSIxVectors = instance->msix_vectors;
        IOCInitMessage->HostPageSize = MR_DEFAULT_NVME_PAGE_SHIFT;

        time = ktime_get_real();
        /* Convert to milliseconds as per FW requirement */
        IOCInitMessage->TimeStamp = cpu_to_le64(ktime_to_ms(time));

        init_frame = (struct megasas_init_frame *)cmd->frame;
        memset(init_frame, 0, IOC_INIT_FRAME_SIZE);

        frame_hdr = &cmd->frame->hdr;
        frame_hdr->cmd_status = 0xFF;
        frame_hdr->flags |= cpu_to_le16(MFI_FRAME_DONT_POST_IN_REPLY_QUEUE);

        init_frame->cmd = MFI_CMD_INIT;
        init_frame->cmd_status = 0xFF;

        drv_ops = (MFI_CAPABILITIES *) &(init_frame->driver_operations);

        /* driver support Extended MSIX */
        if (instance->adapter_type >= INVADER_SERIES)
                drv_ops->mfi_capabilities.support_additional_msix = 1;
        /* driver supports HA / Remote LUN over Fast Path interface */
        drv_ops->mfi_capabilities.support_fp_remote_lun = 1;

        drv_ops->mfi_capabilities.support_max_255lds = 1;
        drv_ops->mfi_capabilities.support_ndrive_r1_lb = 1;
        drv_ops->mfi_capabilities.security_protocol_cmds_fw = 1;

        if (instance->max_chain_frame_sz > MEGASAS_CHAIN_FRAME_SZ_MIN)
                drv_ops->mfi_capabilities.support_ext_io_size = 1;

        drv_ops->mfi_capabilities.support_fp_rlbypass = 1;
        if (!dual_qdepth_disable)
                drv_ops->mfi_capabilities.support_ext_queue_depth = 1;

        drv_ops->mfi_capabilities.support_qd_throttling = 1;
        drv_ops->mfi_capabilities.support_pd_map_target_id = 1;
        drv_ops->mfi_capabilities.support_nvme_passthru = 1;
        drv_ops->mfi_capabilities.support_fw_exposed_dev_list = 1;

        if (instance->consistent_mask_64bit)
                drv_ops->mfi_capabilities.support_64bit_mode = 1;

        /* Convert capability to LE32 */
        cpu_to_le32s((u32 *)&init_frame->driver_operations.mfi_capabilities);

        sys_info = dmi_get_system_info(DMI_PRODUCT_UUID);
        if (instance->system_info_buf && sys_info) {
                memcpy(instance->system_info_buf->systemId, sys_info,
                        strlen(sys_info) > 64 ? 64 : strlen(sys_info));
                instance->system_info_buf->systemIdLength =
                        strlen(sys_info) > 64 ? 64 : strlen(sys_info);
                init_frame->system_info_lo = cpu_to_le32(lower_32_bits(instance->system_info_h));
                init_frame->system_info_hi = cpu_to_le32(upper_32_bits(instance->system_info_h));
        }

        init_frame->queue_info_new_phys_addr_hi =
                cpu_to_le32(upper_32_bits(ioc_init_handle));
        init_frame->queue_info_new_phys_addr_lo =
                cpu_to_le32(lower_32_bits(ioc_init_handle));
        init_frame->data_xfer_len = cpu_to_le32(sizeof(struct MPI2_IOC_INIT_REQUEST));

        /*
         * Each bit in replyqueue_mask represents one group of MSI-x vectors
         * (each group has 8 vectors)
         */
        switch (instance->perf_mode) {
        case MR_BALANCED_PERF_MODE:
                init_frame->replyqueue_mask =
                       cpu_to_le16(~(~0 << instance->low_latency_index_start/8));
                break;
        case MR_IOPS_PERF_MODE:
                init_frame->replyqueue_mask =
                       cpu_to_le16(~(~0 << instance->msix_vectors/8));
                break;
        }


        req_desc.u.low = cpu_to_le32(lower_32_bits(cmd->frame_phys_addr));
        req_desc.u.high = cpu_to_le32(upper_32_bits(cmd->frame_phys_addr));
        req_desc.MFAIo.RequestFlags =
                (MEGASAS_REQ_DESCRIPT_FLAGS_MFA <<
                MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);

        /*
         * disable the intr before firing the init frame
         */
        instance->instancet->disable_intr(instance);

        for (i = 0; i < (10 * 1000); i += 20) {
                if (megasas_readl(instance, &instance->reg_set->doorbell) & 1)
                        msleep(20);
                else
                        break;
        }

        /* For AERO also, IOC_INIT requires 64 bit descriptor write */
        megasas_write_64bit_req_desc(instance, &req_desc);

        wait_and_poll(instance, cmd, MFI_IO_TIMEOUT_SECS);

        frame_hdr = &cmd->frame->hdr;
        if (frame_hdr->cmd_status != 0) {
                ret = 1;
                goto fail_fw_init;
        }

        if (instance->adapter_type >= AERO_SERIES) {
                scratch_pad_1 = megasas_readl
                        (instance, &instance->reg_set->outbound_scratch_pad_1);

                instance->atomic_desc_support =
                        (scratch_pad_1 & MR_ATOMIC_DESCRIPTOR_SUPPORT_OFFSET) ? 1 : 0;

                dev_info(&instance->pdev->dev, "FW supports atomic descriptor\t: %s\n",
                        instance->atomic_desc_support ? "Yes" : "No");
        }

        return 0;

fail_fw_init:
        dev_err(&instance->pdev->dev,
                "Init cmd return status FAILED for SCSI host %d\n",
                instance->host->host_no);

        return ret;
}

/**
 * megasas_sync_pd_seq_num -    JBOD SEQ MAP
 * @instance:           Adapter soft state
 * @pend:               set to 1, if it is pended jbod map.
 *
 * Issue Jbod map to the firmware. If it is pended command,
 * issue command and return. If it is first instance of jbod map
 * issue and receive command.
 */
int
megasas_sync_pd_seq_num(struct megasas_instance *instance, bool pend) {
        int ret = 0;
        size_t pd_seq_map_sz;
        struct megasas_cmd *cmd;
        struct megasas_dcmd_frame *dcmd;
        struct fusion_context *fusion = instance->ctrl_context;
        struct MR_PD_CFG_SEQ_NUM_SYNC *pd_sync;
        dma_addr_t pd_seq_h;

        pd_sync = (void *)fusion->pd_seq_sync[(instance->pd_seq_map_id & 1)];
        pd_seq_h = fusion->pd_seq_phys[(instance->pd_seq_map_id & 1)];
        pd_seq_map_sz = struct_size(pd_sync, seq, MAX_PHYSICAL_DEVICES - 1);

        cmd = megasas_get_cmd(instance);
        if (!cmd) {
                dev_err(&instance->pdev->dev,
                        "Could not get mfi cmd. Fail from %s %d\n",
                        __func__, __LINE__);
                return -ENOMEM;
        }

        dcmd = &cmd->frame->dcmd;

        memset(pd_sync, 0, pd_seq_map_sz);
        memset(dcmd->mbox.b, 0, MFI_MBOX_SIZE);

        if (pend) {
                dcmd->mbox.b[0] = MEGASAS_DCMD_MBOX_PEND_FLAG;
                dcmd->flags = MFI_FRAME_DIR_WRITE;
                instance->jbod_seq_cmd = cmd;
        } else {
                dcmd->flags = MFI_FRAME_DIR_READ;
        }

        dcmd->cmd = MFI_CMD_DCMD;
        dcmd->cmd_status = 0xFF;
        dcmd->sge_count = 1;
        dcmd->timeout = 0;
        dcmd->pad_0 = 0;
        dcmd->data_xfer_len = cpu_to_le32(pd_seq_map_sz);
        dcmd->opcode = cpu_to_le32(MR_DCMD_SYSTEM_PD_MAP_GET_INFO);

        megasas_set_dma_settings(instance, dcmd, pd_seq_h, pd_seq_map_sz);

        if (pend) {
                instance->instancet->issue_dcmd(instance, cmd);
                return 0;
        }

        /* Below code is only for non pended DCMD */
        if (!instance->mask_interrupts)
                ret = megasas_issue_blocked_cmd(instance, cmd,
                        MFI_IO_TIMEOUT_SECS);
        else
                ret = megasas_issue_polled(instance, cmd);

        if (le32_to_cpu(pd_sync->count) > MAX_PHYSICAL_DEVICES) {
                dev_warn(&instance->pdev->dev,
                        "driver supports max %d JBOD, but FW reports %d\n",
                        MAX_PHYSICAL_DEVICES, le32_to_cpu(pd_sync->count));
                ret = -EINVAL;
        }

        if (ret == DCMD_TIMEOUT)
                dev_warn(&instance->pdev->dev,
                         "%s DCMD timed out, continue without JBOD sequence map\n",
                         __func__);

        if (ret == DCMD_SUCCESS)
                instance->pd_seq_map_id++;

        megasas_return_cmd(instance, cmd);
        return ret;
}

/*
 * megasas_get_ld_map_info -    Returns FW's ld_map structure
 * @instance:                           Adapter soft state
 * @pend:                               Pend the command or not
 * Issues an internal command (DCMD) to get the FW's controller PD
 * list structure.  This information is mainly used to find out SYSTEM
 * supported by the FW.
 * dcmd.mbox value setting for MR_DCMD_LD_MAP_GET_INFO
 * dcmd.mbox.b[0]       - number of LDs being sync'd
 * dcmd.mbox.b[1]       - 0 - complete command immediately.
 *                      - 1 - pend till config change
 * dcmd.mbox.b[2]       - 0 - supports max 64 lds and uses legacy MR_FW_RAID_MAP
 *                      - 1 - supports max MAX_LOGICAL_DRIVES_EXT lds and
 *                              uses extended struct MR_FW_RAID_MAP_EXT
 */
static int
megasas_get_ld_map_info(struct megasas_instance *instance)
{
        int ret = 0;
        struct megasas_cmd *cmd;
        struct megasas_dcmd_frame *dcmd;
        void *ci;
        dma_addr_t ci_h = 0;
        u32 size_map_info;
        struct fusion_context *fusion;

        cmd = megasas_get_cmd(instance);

        if (!cmd) {
                dev_printk(KERN_DEBUG, &instance->pdev->dev, "Failed to get cmd for map info\n");
                return -ENOMEM;
        }

        fusion = instance->ctrl_context;

        if (!fusion) {
                megasas_return_cmd(instance, cmd);
                return -ENXIO;
        }

        dcmd = &cmd->frame->dcmd;

        size_map_info = fusion->current_map_sz;

        ci = (void *) fusion->ld_map[(instance->map_id & 1)];
        ci_h = fusion->ld_map_phys[(instance->map_id & 1)];

        if (!ci) {
                dev_printk(KERN_DEBUG, &instance->pdev->dev, "Failed to alloc mem for ld_map_info\n");
                megasas_return_cmd(instance, cmd);
                return -ENOMEM;
        }

        memset(ci, 0, fusion->max_map_sz);
        memset(dcmd->mbox.b, 0, MFI_MBOX_SIZE);
        dcmd->cmd = MFI_CMD_DCMD;
        dcmd->cmd_status = 0xFF;
        dcmd->sge_count = 1;
        dcmd->flags = MFI_FRAME_DIR_READ;
        dcmd->timeout = 0;
        dcmd->pad_0 = 0;
        dcmd->data_xfer_len = cpu_to_le32(size_map_info);
        dcmd->opcode = cpu_to_le32(MR_DCMD_LD_MAP_GET_INFO);

        megasas_set_dma_settings(instance, dcmd, ci_h, size_map_info);

        if (!instance->mask_interrupts)
                ret = megasas_issue_blocked_cmd(instance, cmd,
                        MFI_IO_TIMEOUT_SECS);
        else
                ret = megasas_issue_polled(instance, cmd);

        if (ret == DCMD_TIMEOUT)
                dev_warn(&instance->pdev->dev,
                         "%s DCMD timed out, RAID map is disabled\n",
                         __func__);

        megasas_return_cmd(instance, cmd);

        return ret;
}

u8
megasas_get_map_info(struct megasas_instance *instance)
{
        struct fusion_context *fusion = instance->ctrl_context;

        fusion->fast_path_io = 0;
        if (!megasas_get_ld_map_info(instance)) {
                if (MR_ValidateMapInfo(instance, instance->map_id)) {
                        fusion->fast_path_io = 1;
                        return 0;
                }
        }
        return 1;
}

/*
 * megasas_sync_map_info -      Returns FW's ld_map structure
 * @instance:                           Adapter soft state
 *
 * Issues an internal command (DCMD) to get the FW's controller PD
 * list structure.  This information is mainly used to find out SYSTEM
 * supported by the FW.
 */
int
megasas_sync_map_info(struct megasas_instance *instance)
{
        int i;
        struct megasas_cmd *cmd;
        struct megasas_dcmd_frame *dcmd;
        u16 num_lds;
        struct fusion_context *fusion;
        struct MR_LD_TARGET_SYNC *ci = NULL;
        struct MR_DRV_RAID_MAP_ALL *map;
        struct MR_LD_RAID  *raid;
        struct MR_LD_TARGET_SYNC *ld_sync;
        dma_addr_t ci_h = 0;
        u32 size_map_info;

        cmd = megasas_get_cmd(instance);

        if (!cmd) {
                dev_printk(KERN_DEBUG, &instance->pdev->dev, "Failed to get cmd for sync info\n");
                return -ENOMEM;
        }

        fusion = instance->ctrl_context;

        if (!fusion) {
                megasas_return_cmd(instance, cmd);
                return 1;
        }

        map = fusion->ld_drv_map[instance->map_id & 1];

        num_lds = le16_to_cpu(map->raidMap.ldCount);

        dcmd = &cmd->frame->dcmd;

        memset(dcmd->mbox.b, 0, MFI_MBOX_SIZE);

        ci = (struct MR_LD_TARGET_SYNC *)
          fusion->ld_map[(instance->map_id - 1) & 1];
        memset(ci, 0, fusion->max_map_sz);

        ci_h = fusion->ld_map_phys[(instance->map_id - 1) & 1];

        ld_sync = (struct MR_LD_TARGET_SYNC *)ci;

        for (i = 0; i < num_lds; i++, ld_sync++) {
                raid = MR_LdRaidGet(i, map);
                ld_sync->targetId = MR_GetLDTgtId(i, map);
                ld_sync->seqNum = raid->seqNum;
        }

        size_map_info = fusion->current_map_sz;

        dcmd->cmd = MFI_CMD_DCMD;
        dcmd->cmd_status = 0xFF;
        dcmd->sge_count = 1;
        dcmd->flags = MFI_FRAME_DIR_WRITE;
        dcmd->timeout = 0;
        dcmd->pad_0 = 0;
        dcmd->data_xfer_len = cpu_to_le32(size_map_info);
        dcmd->mbox.b[0] = num_lds;
        dcmd->mbox.b[1] = MEGASAS_DCMD_MBOX_PEND_FLAG;
        dcmd->opcode = cpu_to_le32(MR_DCMD_LD_MAP_GET_INFO);

        megasas_set_dma_settings(instance, dcmd, ci_h, size_map_info);

        instance->map_update_cmd = cmd;

        instance->instancet->issue_dcmd(instance, cmd);

        return 0;
}

/*
 * meagasas_display_intel_branding - Display branding string
 * @instance: per adapter object
 *
 * Return nothing.
 */
static void
megasas_display_intel_branding(struct megasas_instance *instance)
{
        if (instance->pdev->subsystem_vendor != PCI_VENDOR_ID_INTEL)
                return;

        switch (instance->pdev->device) {
        case PCI_DEVICE_ID_LSI_INVADER:
                switch (instance->pdev->subsystem_device) {
                case MEGARAID_INTEL_RS3DC080_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3DC080_BRANDING);
                        break;
                case MEGARAID_INTEL_RS3DC040_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3DC040_BRANDING);
                        break;
                case MEGARAID_INTEL_RS3SC008_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3SC008_BRANDING);
                        break;
                case MEGARAID_INTEL_RS3MC044_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3MC044_BRANDING);
                        break;
                default:
                        break;
                }
                break;
        case PCI_DEVICE_ID_LSI_FURY:
                switch (instance->pdev->subsystem_device) {
                case MEGARAID_INTEL_RS3WC080_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3WC080_BRANDING);
                        break;
                case MEGARAID_INTEL_RS3WC040_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RS3WC040_BRANDING);
                        break;
                default:
                        break;
                }
                break;
        case PCI_DEVICE_ID_LSI_CUTLASS_52:
        case PCI_DEVICE_ID_LSI_CUTLASS_53:
                switch (instance->pdev->subsystem_device) {
                case MEGARAID_INTEL_RMS3BC160_SSDID:
                        dev_info(&instance->pdev->dev, "scsi host %d: %s\n",
                                instance->host->host_no,
                                MEGARAID_INTEL_RMS3BC160_BRANDING);
                        break;
                default:
                        break;
                }
                break;
        default:
                break;
        }
}

/**
 * megasas_allocate_raid_maps - Allocate memory for RAID maps
 * @instance:                           Adapter soft state
 *
 * return:                              if success: return 0
 *                                      failed:  return -ENOMEM
 */
static inline int megasas_allocate_raid_maps(struct megasas_instance *instance)
{
        struct fusion_context *fusion;
        int i = 0;

        fusion = instance->ctrl_context;

        fusion->drv_map_pages = get_order(fusion->drv_map_sz);

        for (i = 0; i < 2; i++) {
                fusion->ld_map[i] = NULL;

                fusion->ld_drv_map[i] = (void *)
                        __get_free_pages(__GFP_ZERO | GFP_KERNEL,
                                         fusion->drv_map_pages);

                if (!fusion->ld_drv_map[i]) {
                        fusion->ld_drv_map[i] = vzalloc(fusion->drv_map_sz);

                        if (!fusion->ld_drv_map[i]) {
                                dev_err(&instance->pdev->dev,
                                        "Could not allocate memory for local map"
                                        " size requested: %d\n",
                                        fusion->drv_map_sz);
                                goto ld_drv_map_alloc_fail;
                        }
                }
        }

        for (i = 0; i < 2; i++) {
                fusion->ld_map[i] = dma_alloc_coherent(&instance->pdev->dev,
                                                       fusion->max_map_sz,
                                                       &fusion->ld_map_phys[i],
                                                       GFP_KERNEL);
                if (!fusion->ld_map[i]) {
                        dev_err(&instance->pdev->dev,
                                "Could not allocate memory for map info %s:%d\n",
                                __func__, __LINE__);
                        goto ld_map_alloc_fail;
                }
        }

        return 0;

ld_map_alloc_fail:
        for (i = 0; i < 2; i++) {
                if (fusion->ld_map[i])
                        dma_free_coherent(&instance->pdev->dev,
                                          fusion->max_map_sz,
                                          fusion->ld_map[i],
                                          fusion->ld_map_phys[i]);
        }

ld_drv_map_alloc_fail:
        for (i = 0; i < 2; i++) {
                if (fusion->ld_drv_map[i]) {
                        if (is_vmalloc_addr(fusion->ld_drv_map[i]))
                                vfree(fusion->ld_drv_map[i]);
                        else
                                free_pages((ulong)fusion->ld_drv_map[i],
                                           fusion->drv_map_pages);
                }
        }

        return -ENOMEM;
}

/**
 * megasas_configure_queue_sizes -      Calculate size of request desc queue,
 *                                      reply desc queue,
 *                                      IO request frame queue, set can_queue.
 * @instance:                           Adapter soft state
 * @return:                             void
 */
static inline
void megasas_configure_queue_sizes(struct megasas_instance *instance)
{
        struct fusion_context *fusion;
        u16 max_cmd;

        fusion = instance->ctrl_context;
        max_cmd = instance->max_fw_cmds;

        if (instance->adapter_type >= VENTURA_SERIES)
                instance->max_mpt_cmds = instance->max_fw_cmds * RAID_1_PEER_CMDS;
        else
                instance->max_mpt_cmds = instance->max_fw_cmds;

        instance->max_scsi_cmds = instance->max_fw_cmds - instance->max_mfi_cmds;
        instance->cur_can_queue = instance->max_scsi_cmds;
        instance->host->can_queue = instance->cur_can_queue;

        fusion->reply_q_depth = 2 * ((max_cmd + 1 + 15) / 16) * 16;

        fusion->request_alloc_sz = sizeof(union MEGASAS_REQUEST_DESCRIPTOR_UNION) *
                                          instance->max_mpt_cmds;
        fusion->reply_alloc_sz = sizeof(union MPI2_REPLY_DESCRIPTORS_UNION) *
                                        (fusion->reply_q_depth);
        fusion->io_frames_alloc_sz = MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE +
                (MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE
                 * (instance->max_mpt_cmds + 1)); /* Extra 1 for SMID 0 */
}

static int megasas_alloc_ioc_init_frame(struct megasas_instance *instance)
{
        struct fusion_context *fusion;
        struct megasas_cmd *cmd;

        fusion = instance->ctrl_context;

        cmd = kzalloc(sizeof(struct megasas_cmd), GFP_KERNEL);

        if (!cmd) {
                dev_err(&instance->pdev->dev, "Failed from func: %s line: %d\n",
                        __func__, __LINE__);
                return -ENOMEM;
        }

        cmd->frame = dma_alloc_coherent(&instance->pdev->dev,
                                        IOC_INIT_FRAME_SIZE,
                                        &cmd->frame_phys_addr, GFP_KERNEL);

        if (!cmd->frame) {
                dev_err(&instance->pdev->dev, "Failed from func: %s line: %d\n",
                        __func__, __LINE__);
                kfree(cmd);
                return -ENOMEM;
        }

        fusion->ioc_init_cmd = cmd;
        return 0;
}

/**
 * megasas_free_ioc_init_cmd -  Free IOC INIT command frame
 * @instance:           Adapter soft state
 */
static inline void megasas_free_ioc_init_cmd(struct megasas_instance *instance)
{
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

        if (fusion->ioc_init_cmd && fusion->ioc_init_cmd->frame)
                dma_free_coherent(&instance->pdev->dev,
                                  IOC_INIT_FRAME_SIZE,
                                  fusion->ioc_init_cmd->frame,
                                  fusion->ioc_init_cmd->frame_phys_addr);

        kfree(fusion->ioc_init_cmd);
}

/**
 * megasas_init_adapter_fusion -        Initializes the FW
 * @instance:           Adapter soft state
 *
 * This is the main function for initializing firmware.
 */
static u32
megasas_init_adapter_fusion(struct megasas_instance *instance)
{
        struct fusion_context *fusion;
        u32 scratch_pad_1;
        int i = 0, count;
        u32 status_reg;

        fusion = instance->ctrl_context;

        megasas_fusion_update_can_queue(instance, PROBE_CONTEXT);

        /*
         * Only Driver's internal DCMDs and IOCTL DCMDs needs to have MFI frames
         */
        instance->max_mfi_cmds =
                MEGASAS_FUSION_INTERNAL_CMDS + MEGASAS_FUSION_IOCTL_CMDS;

        megasas_configure_queue_sizes(instance);

        scratch_pad_1 = megasas_readl(instance,
                                      &instance->reg_set->outbound_scratch_pad_1);
        /* If scratch_pad_1 & MEGASAS_MAX_CHAIN_SIZE_UNITS_MASK is set,
         * Firmware support extended IO chain frame which is 4 times more than
         * legacy Firmware.
         * Legacy Firmware - Frame size is (8 * 128) = 1K
         * 1M IO Firmware  - Frame size is (8 * 128 * 4)  = 4K
         */
        if (scratch_pad_1 & MEGASAS_MAX_CHAIN_SIZE_UNITS_MASK)
                instance->max_chain_frame_sz =
                        ((scratch_pad_1 & MEGASAS_MAX_CHAIN_SIZE_MASK) >>
                        MEGASAS_MAX_CHAIN_SHIFT) * MEGASAS_1MB_IO;
        else
                instance->max_chain_frame_sz =
                        ((scratch_pad_1 & MEGASAS_MAX_CHAIN_SIZE_MASK) >>
                        MEGASAS_MAX_CHAIN_SHIFT) * MEGASAS_256K_IO;

        if (instance->max_chain_frame_sz < MEGASAS_CHAIN_FRAME_SZ_MIN) {
                dev_warn(&instance->pdev->dev, "frame size %d invalid, fall back to legacy max frame size %d\n",
                        instance->max_chain_frame_sz,
                        MEGASAS_CHAIN_FRAME_SZ_MIN);
                instance->max_chain_frame_sz = MEGASAS_CHAIN_FRAME_SZ_MIN;
        }

        fusion->max_sge_in_main_msg =
                (MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE
                        - offsetof(struct MPI2_RAID_SCSI_IO_REQUEST, SGL))/16;

        fusion->max_sge_in_chain =
                instance->max_chain_frame_sz
                        / sizeof(union MPI2_SGE_IO_UNION);

        instance->max_num_sge =
                rounddown_pow_of_two(fusion->max_sge_in_main_msg
                        + fusion->max_sge_in_chain - 2);

        /* Used for pass thru MFI frame (DCMD) */
        fusion->chain_offset_mfi_pthru =
                offsetof(struct MPI2_RAID_SCSI_IO_REQUEST, SGL)/16;

        fusion->chain_offset_io_request =
                (MEGA_MPI2_RAID_DEFAULT_IO_FRAME_SIZE -
                 sizeof(union MPI2_SGE_IO_UNION))/16;

        count = instance->msix_vectors > 0 ? instance->msix_vectors : 1;
        for (i = 0 ; i < count; i++)
                fusion->last_reply_idx[i] = 0;

        /*
         * For fusion adapters, 3 commands for IOCTL and 8 commands
         * for driver's internal DCMDs.
         */
        instance->max_scsi_cmds = instance->max_fw_cmds -
                                (MEGASAS_FUSION_INTERNAL_CMDS +
                                MEGASAS_FUSION_IOCTL_CMDS);
        sema_init(&instance->ioctl_sem, MEGASAS_FUSION_IOCTL_CMDS);

        if (megasas_alloc_ioc_init_frame(instance))
                return 1;

        /*
         * Allocate memory for descriptors
         * Create a pool of commands
         */
        if (megasas_alloc_cmds(instance))
                goto fail_alloc_mfi_cmds;
        if (megasas_alloc_cmds_fusion(instance))
                goto fail_alloc_cmds;

        if (megasas_ioc_init_fusion(instance)) {
                status_reg = instance->instancet->read_fw_status_reg(instance);
                if (((status_reg & MFI_STATE_MASK) == MFI_STATE_FAULT) &&
                    (status_reg & MFI_RESET_ADAPTER)) {
                        /* Do a chip reset and then retry IOC INIT once */
                        if (megasas_adp_reset_wait_for_ready
                                (instance, true, 0) == FAILED)
                                goto fail_ioc_init;

                        if (megasas_ioc_init_fusion(instance))
                                goto fail_ioc_init;
                } else {
                        goto fail_ioc_init;
                }
        }

        megasas_display_intel_branding(instance);
        if (megasas_get_ctrl_info(instance)) {
                dev_err(&instance->pdev->dev,
                        "Could not get controller info. Fail from %s %d\n",
                        __func__, __LINE__);
                goto fail_ioc_init;
        }

        instance->flag_ieee = 1;
        instance->r1_ldio_hint_default =  MR_R1_LDIO_PIGGYBACK_DEFAULT;
        instance->threshold_reply_count = instance->max_fw_cmds / 4;
        fusion->fast_path_io = 0;

        if (megasas_allocate_raid_maps(instance))
                goto fail_ioc_init;

        if (!megasas_get_map_info(instance))
                megasas_sync_map_info(instance);

        return 0;

fail_ioc_init:
        megasas_free_cmds_fusion(instance);
fail_alloc_cmds:
        megasas_free_cmds(instance);
fail_alloc_mfi_cmds:
        megasas_free_ioc_init_cmd(instance);
        return 1;
}

/**
 * megasas_fault_detect_work    -       Worker function of
 *                                      FW fault handling workqueue.
 * @work:       FW fault work struct
 */
static void
megasas_fault_detect_work(struct work_struct *work)
{
        struct megasas_instance *instance =
                container_of(work, struct megasas_instance,
                             fw_fault_work.work);
        u32 fw_state, dma_state, status;

        /* Check the fw state */
        fw_state = instance->instancet->read_fw_status_reg(instance) &
                        MFI_STATE_MASK;

        if (fw_state == MFI_STATE_FAULT) {
                dma_state = instance->instancet->read_fw_status_reg(instance) &
                                MFI_STATE_DMADONE;
                /* Start collecting crash, if DMA bit is done */
                if (instance->crash_dump_drv_support &&
                    instance->crash_dump_app_support && dma_state) {
                        megasas_fusion_crash_dump(instance);
                } else {
                        if (instance->unload == 0) {
                                status = megasas_reset_fusion(instance->host, 0);
                                if (status != SUCCESS) {
                                        dev_err(&instance->pdev->dev,
                                                "Failed from %s %d, do not re-arm timer\n",
                                                __func__, __LINE__);
                                        return;
                                }
                        }
                }
        }

        if (instance->fw_fault_work_q)
                queue_delayed_work(instance->fw_fault_work_q,
                        &instance->fw_fault_work,
                        msecs_to_jiffies(MEGASAS_WATCHDOG_THREAD_INTERVAL));
}

int
megasas_fusion_start_watchdog(struct megasas_instance *instance)
{
        /* Check if the Fault WQ is already started */
        if (instance->fw_fault_work_q)
                return SUCCESS;

        INIT_DELAYED_WORK(&instance->fw_fault_work, megasas_fault_detect_work);

        snprintf(instance->fault_handler_work_q_name,
                 sizeof(instance->fault_handler_work_q_name),
                 "poll_megasas%d_status", instance->host->host_no);

        instance->fw_fault_work_q =
                create_singlethread_workqueue(instance->fault_handler_work_q_name);
        if (!instance->fw_fault_work_q) {
                dev_err(&instance->pdev->dev, "Failed from %s %d\n",
                        __func__, __LINE__);
                return FAILED;
        }

        queue_delayed_work(instance->fw_fault_work_q,
                           &instance->fw_fault_work,
                           msecs_to_jiffies(MEGASAS_WATCHDOG_THREAD_INTERVAL));

        return SUCCESS;
}

void
megasas_fusion_stop_watchdog(struct megasas_instance *instance)
{
        struct workqueue_struct *wq;

        if (instance->fw_fault_work_q) {
                wq = instance->fw_fault_work_q;
                instance->fw_fault_work_q = NULL;
                if (!cancel_delayed_work_sync(&instance->fw_fault_work))
                        flush_workqueue(wq);
                destroy_workqueue(wq);
        }
}

/**
 * map_cmd_status -     Maps FW cmd status to OS cmd status
 * @fusion:             fusion context
 * @scmd:               Pointer to cmd
 * @status:             status of cmd returned by FW
 * @ext_status:         ext status of cmd returned by FW
 * @data_length:        command data length
 * @sense:              command sense data
 */
static void
map_cmd_status(struct fusion_context *fusion,
                struct scsi_cmnd *scmd, u8 status, u8 ext_status,
                u32 data_length, u8 *sense)
{
        u8 cmd_type;
        int resid;

        cmd_type = megasas_cmd_type(scmd);
        switch (status) {

        case MFI_STAT_OK:
                scmd->result = DID_OK << 16;
                break;

        case MFI_STAT_SCSI_IO_FAILED:
        case MFI_STAT_LD_INIT_IN_PROGRESS:
                scmd->result = (DID_ERROR << 16) | ext_status;
                break;

        case MFI_STAT_SCSI_DONE_WITH_ERROR:

                scmd->result = (DID_OK << 16) | ext_status;
                if (ext_status == SAM_STAT_CHECK_CONDITION) {
                        memset(scmd->sense_buffer, 0,
                               SCSI_SENSE_BUFFERSIZE);
                        memcpy(scmd->sense_buffer, sense,
                               SCSI_SENSE_BUFFERSIZE);
                        scmd->result |= DRIVER_SENSE << 24;
                }

                /*
                 * If the  IO request is partially completed, then MR FW will
                 * update "io_request->DataLength" field with actual number of
                 * bytes transferred.Driver will set residual bytes count in
                 * SCSI command structure.
                 */
                resid = (scsi_bufflen(scmd) - data_length);
                scsi_set_resid(scmd, resid);

                if (resid &&
                        ((cmd_type == READ_WRITE_LDIO) ||
                        (cmd_type == READ_WRITE_SYSPDIO)))
                        scmd_printk(KERN_INFO, scmd, "BRCM Debug mfi stat 0x%x, data len"
                                " requested/completed 0x%x/0x%x\n",
                                status, scsi_bufflen(scmd), data_length);
                break;

        case MFI_STAT_LD_OFFLINE:
        case MFI_STAT_DEVICE_NOT_FOUND:
                scmd->result = DID_BAD_TARGET << 16;
                break;
        case MFI_STAT_CONFIG_SEQ_MISMATCH:
                scmd->result = DID_IMM_RETRY << 16;
                break;
        default:
                scmd->result = DID_ERROR << 16;
                break;
        }
}

/**
 * megasas_is_prp_possible -
 * Checks if native NVMe PRPs can be built for the IO
 *
 * @instance:           Adapter soft state
 * @scmd:               SCSI command from the mid-layer
 * @sge_count:          scatter gather element count.
 *
 * Returns:             true: PRPs can be built
 *                      false: IEEE SGLs needs to be built
 */
static bool
megasas_is_prp_possible(struct megasas_instance *instance,
                        struct scsi_cmnd *scmd, int sge_count)
{
        u32 data_length = 0;
        struct scatterlist *sg_scmd;
        bool build_prp = false;
        u32 mr_nvme_pg_size;

        mr_nvme_pg_size = max_t(u32, instance->nvme_page_size,
                                MR_DEFAULT_NVME_PAGE_SIZE);
        data_length = scsi_bufflen(scmd);
        sg_scmd = scsi_sglist(scmd);

        /*
         * NVMe uses one PRP for each page (or part of a page)
         * look at the data length - if 4 pages or less then IEEE is OK
         * if  > 5 pages then we need to build a native SGL
         * if > 4 and <= 5 pages, then check physical address of 1st SG entry
         * if this first size in the page is >= the residual beyond 4 pages
         * then use IEEE, otherwise use native SGL
         */

        if (data_length > (mr_nvme_pg_size * 5)) {
                build_prp = true;
        } else if ((data_length > (mr_nvme_pg_size * 4)) &&
                        (data_length <= (mr_nvme_pg_size * 5)))  {
                /* check if 1st SG entry size is < residual beyond 4 pages */
                if (sg_dma_len(sg_scmd) < (data_length - (mr_nvme_pg_size * 4)))
                        build_prp = true;
        }

        return build_prp;
}

/**
 * megasas_make_prp_nvme -
 * Prepare PRPs(Physical Region Page)- SGLs specific to NVMe drives only
 *
 * @instance:           Adapter soft state
 * @scmd:               SCSI command from the mid-layer
 * @sgl_ptr:            SGL to be filled in
 * @cmd:                Fusion command frame
 * @sge_count:          scatter gather element count.
 *
 * Returns:             true: PRPs are built
 *                      false: IEEE SGLs needs to be built
 */
static bool
megasas_make_prp_nvme(struct megasas_instance *instance, struct scsi_cmnd *scmd,
                      struct MPI25_IEEE_SGE_CHAIN64 *sgl_ptr,
                      struct megasas_cmd_fusion *cmd, int sge_count)
{
        int sge_len, offset, num_prp_in_chain = 0;
        struct MPI25_IEEE_SGE_CHAIN64 *main_chain_element, *ptr_first_sgl;
        u64 *ptr_sgl;
        dma_addr_t ptr_sgl_phys;
        u64 sge_addr;
        u32 page_mask, page_mask_result;
        struct scatterlist *sg_scmd;
        u32 first_prp_len;
        bool build_prp = false;
        int data_len = scsi_bufflen(scmd);
        u32 mr_nvme_pg_size = max_t(u32, instance->nvme_page_size,
                                        MR_DEFAULT_NVME_PAGE_SIZE);

        build_prp = megasas_is_prp_possible(instance, scmd, sge_count);

        if (!build_prp)
                return false;

        /*
         * Nvme has a very convoluted prp format.  One prp is required
         * for each page or partial page. Driver need to split up OS sg_list
         * entries if it is longer than one page or cross a page
         * boundary.  Driver also have to insert a PRP list pointer entry as
         * the last entry in each physical page of the PRP list.
         *
         * NOTE: The first PRP "entry" is actually placed in the first
         * SGL entry in the main message as IEEE 64 format.  The 2nd
         * entry in the main message is the chain element, and the rest
         * of the PRP entries are built in the contiguous pcie buffer.
         */
        page_mask = mr_nvme_pg_size - 1;
        ptr_sgl = (u64 *)cmd->sg_frame;
        ptr_sgl_phys = cmd->sg_frame_phys_addr;
        memset(ptr_sgl, 0, instance->max_chain_frame_sz);

        /* Build chain frame element which holds all prps except first*/
        main_chain_element = (struct MPI25_IEEE_SGE_CHAIN64 *)
            ((u8 *)sgl_ptr + sizeof(struct MPI25_IEEE_SGE_CHAIN64));

        main_chain_element->Address = cpu_to_le64(ptr_sgl_phys);
        main_chain_element->NextChainOffset = 0;
        main_chain_element->Flags = IEEE_SGE_FLAGS_CHAIN_ELEMENT |
                                        IEEE_SGE_FLAGS_SYSTEM_ADDR |
                                        MPI26_IEEE_SGE_FLAGS_NSF_NVME_PRP;

        /* Build first prp, sge need not to be page aligned*/
        ptr_first_sgl = sgl_ptr;
        sg_scmd = scsi_sglist(scmd);
        sge_addr = sg_dma_address(sg_scmd);
        sge_len = sg_dma_len(sg_scmd);

        offset = (u32)(sge_addr & page_mask);
        first_prp_len = mr_nvme_pg_size - offset;

        ptr_first_sgl->Address = cpu_to_le64(sge_addr);
        ptr_first_sgl->Length = cpu_to_le32(first_prp_len);

        data_len -= first_prp_len;

        if (sge_len > first_prp_len) {
                sge_addr += first_prp_len;
                sge_len -= first_prp_len;
        } else if (sge_len == first_prp_len) {
                sg_scmd = sg_next(sg_scmd);
                sge_addr = sg_dma_address(sg_scmd);
                sge_len = sg_dma_len(sg_scmd);
        }

        for (;;) {
                offset = (u32)(sge_addr & page_mask);

                /* Put PRP pointer due to page boundary*/
                page_mask_result = (uintptr_t)(ptr_sgl + 1) & page_mask;
                if (unlikely(!page_mask_result)) {
                        scmd_printk(KERN_NOTICE,
                                    scmd, "page boundary ptr_sgl: 0x%p\n",
                                    ptr_sgl);
                        ptr_sgl_phys += 8;
                        *ptr_sgl = cpu_to_le64(ptr_sgl_phys);
                        ptr_sgl++;
                        num_prp_in_chain++;
                }

                *ptr_sgl = cpu_to_le64(sge_addr);
                ptr_sgl++;
                ptr_sgl_phys += 8;
                num_prp_in_chain++;

                sge_addr += mr_nvme_pg_size;
                sge_len -= mr_nvme_pg_size;
                data_len -= mr_nvme_pg_size;

                if (data_len <= 0)
                        break;

                if (sge_len > 0)
                        continue;

                sg_scmd = sg_next(sg_scmd);
                sge_addr = sg_dma_address(sg_scmd);
                sge_len = sg_dma_len(sg_scmd);
        }

        main_chain_element->Length =
                        cpu_to_le32(num_prp_in_chain * sizeof(u64));

        return build_prp;
}

/**
 * megasas_make_sgl_fusion -    Prepares 32-bit SGL
 * @instance:           Adapter soft state
 * @scp:                SCSI command from the mid-layer
 * @sgl_ptr:            SGL to be filled in
 * @cmd:                cmd we are working on
 * @sge_count:          sge count
 *
 */
static void
megasas_make_sgl_fusion(struct megasas_instance *instance,
                        struct scsi_cmnd *scp,
                        struct MPI25_IEEE_SGE_CHAIN64 *sgl_ptr,
                        struct megasas_cmd_fusion *cmd, int sge_count)
{
        int i, sg_processed;
        struct scatterlist *os_sgl;
        struct fusion_context *fusion;

        fusion = instance->ctrl_context;

        if (instance->adapter_type >= INVADER_SERIES) {
                struct MPI25_IEEE_SGE_CHAIN64 *sgl_ptr_end = sgl_ptr;
                sgl_ptr_end += fusion->max_sge_in_main_msg - 1;
                sgl_ptr_end->Flags = 0;
        }

        scsi_for_each_sg(scp, os_sgl, sge_count, i) {
                sgl_ptr->Length = cpu_to_le32(sg_dma_len(os_sgl));
                sgl_ptr->Address = cpu_to_le64(sg_dma_address(os_sgl));
                sgl_ptr->Flags = 0;
                if (instance->adapter_type >= INVADER_SERIES)
                        if (i == sge_count - 1)
                                sgl_ptr->Flags = IEEE_SGE_FLAGS_END_OF_LIST;
                sgl_ptr++;
                sg_processed = i + 1;

                if ((sg_processed ==  (fusion->max_sge_in_main_msg - 1)) &&
                    (sge_count > fusion->max_sge_in_main_msg)) {

                        struct MPI25_IEEE_SGE_CHAIN64 *sg_chain;
                        if (instance->adapter_type >= INVADER_SERIES) {
                                if ((le16_to_cpu(cmd->io_request->IoFlags) &
                                        MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH) !=
                                        MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH)
                                        cmd->io_request->ChainOffset =
                                                fusion->
                                                chain_offset_io_request;
                                else
                                        cmd->io_request->ChainOffset = 0;
                        } else
                                cmd->io_request->ChainOffset =
                                        fusion->chain_offset_io_request;

                        sg_chain = sgl_ptr;
                        /* Prepare chain element */
                        sg_chain->NextChainOffset = 0;
                        if (instance->adapter_type >= INVADER_SERIES)
                                sg_chain->Flags = IEEE_SGE_FLAGS_CHAIN_ELEMENT;
                        else
                                sg_chain->Flags =
                                        (IEEE_SGE_FLAGS_CHAIN_ELEMENT |
                                         MPI2_IEEE_SGE_FLAGS_IOCPLBNTA_ADDR);
                        sg_chain->Length =  cpu_to_le32((sizeof(union MPI2_SGE_IO_UNION) * (sge_count - sg_processed)));
                        sg_chain->Address = cpu_to_le64(cmd->sg_frame_phys_addr);

                        sgl_ptr =
                          (struct MPI25_IEEE_SGE_CHAIN64 *)cmd->sg_frame;
                        memset(sgl_ptr, 0, instance->max_chain_frame_sz);
                }
        }
}

/**
 * megasas_make_sgl -   Build Scatter Gather List(SGLs)
 * @scp:                SCSI command pointer
 * @instance:           Soft instance of controller
 * @cmd:                Fusion command pointer
 *
 * This function will build sgls based on device type.
 * For nvme drives, there is different way of building sgls in nvme native
 * format- PRPs(Physical Region Page).
 *
 * Returns the number of sg lists actually used, zero if the sg lists
 * is NULL, or -ENOMEM if the mapping failed
 */
static
int megasas_make_sgl(struct megasas_instance *instance, struct scsi_cmnd *scp,
                     struct megasas_cmd_fusion *cmd)
{
        int sge_count;
        bool build_prp = false;
        struct MPI25_IEEE_SGE_CHAIN64 *sgl_chain64;

        sge_count = scsi_dma_map(scp);

        if ((sge_count > instance->max_num_sge) || (sge_count <= 0))
                return sge_count;

        sgl_chain64 = (struct MPI25_IEEE_SGE_CHAIN64 *)&cmd->io_request->SGL;
        if ((le16_to_cpu(cmd->io_request->IoFlags) &
            MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH) &&
            (cmd->pd_interface == NVME_PD))
                build_prp = megasas_make_prp_nvme(instance, scp, sgl_chain64,
                                                  cmd, sge_count);

        if (!build_prp)
                megasas_make_sgl_fusion(instance, scp, sgl_chain64,
                                        cmd, sge_count);

        return sge_count;
}

/**
 * megasas_set_pd_lba - Sets PD LBA
 * @io_request:         IO request
 * @cdb_len:            cdb length
 * @io_info:            IO information
 * @scp:                SCSI command
 * @local_map_ptr:      Raid map
 * @ref_tag:            Primary reference tag
 *
 * Used to set the PD LBA in CDB for FP IOs
 */
static void
megasas_set_pd_lba(struct MPI2_RAID_SCSI_IO_REQUEST *io_request, u8 cdb_len,
                   struct IO_REQUEST_INFO *io_info, struct scsi_cmnd *scp,
                   struct MR_DRV_RAID_MAP_ALL *local_map_ptr, u32 ref_tag)
{
        struct MR_LD_RAID *raid;
        u16 ld;
        u64 start_blk = io_info->pdBlock;
        u8 *cdb = io_request->CDB.CDB32;
        u32 num_blocks = io_info->numBlocks;
        u8 opcode = 0, flagvals = 0, groupnum = 0, control = 0;

        /* Check if T10 PI (DIF) is enabled for this LD */
        ld = MR_TargetIdToLdGet(io_info->ldTgtId, local_map_ptr);
        raid = MR_LdRaidGet(ld, local_map_ptr);
        if (raid->capability.ldPiMode == MR_PROT_INFO_TYPE_CONTROLLER) {
                memset(cdb, 0, sizeof(io_request->CDB.CDB32));
                cdb[0] =  MEGASAS_SCSI_VARIABLE_LENGTH_CMD;
                cdb[7] =  MEGASAS_SCSI_ADDL_CDB_LEN;

                if (scp->sc_data_direction == DMA_FROM_DEVICE)
                        cdb[9] = MEGASAS_SCSI_SERVICE_ACTION_READ32;
                else
                        cdb[9] = MEGASAS_SCSI_SERVICE_ACTION_WRITE32;
                cdb[10] = MEGASAS_RD_WR_PROTECT_CHECK_ALL;

                /* LBA */
                cdb[12] = (u8)((start_blk >> 56) & 0xff);
                cdb[13] = (u8)((start_blk >> 48) & 0xff);
                cdb[14] = (u8)((start_blk >> 40) & 0xff);
                cdb[15] = (u8)((start_blk >> 32) & 0xff);
                cdb[16] = (u8)((start_blk >> 24) & 0xff);
                cdb[17] = (u8)((start_blk >> 16) & 0xff);
                cdb[18] = (u8)((start_blk >> 8) & 0xff);
                cdb[19] = (u8)(start_blk & 0xff);

                /* Logical block reference tag */
                io_request->CDB.EEDP32.PrimaryReferenceTag =
                        cpu_to_be32(ref_tag);
                io_request->CDB.EEDP32.PrimaryApplicationTagMask = cpu_to_be16(0xffff);
                io_request->IoFlags = cpu_to_le16(32); /* Specify 32-byte cdb */

                /* Transfer length */
                cdb[28] = (u8)((num_blocks >> 24) & 0xff);
                cdb[29] = (u8)((num_blocks >> 16) & 0xff);
                cdb[30] = (u8)((num_blocks >> 8) & 0xff);
                cdb[31] = (u8)(num_blocks & 0xff);

                /* set SCSI IO EEDPFlags */
                if (scp->sc_data_direction == DMA_FROM_DEVICE) {
                        io_request->EEDPFlags = cpu_to_le16(
                                MPI2_SCSIIO_EEDPFLAGS_INC_PRI_REFTAG  |
                                MPI2_SCSIIO_EEDPFLAGS_CHECK_REFTAG |
                                MPI2_SCSIIO_EEDPFLAGS_CHECK_REMOVE_OP |
                                MPI2_SCSIIO_EEDPFLAGS_CHECK_APPTAG |
                                MPI25_SCSIIO_EEDPFLAGS_DO_NOT_DISABLE_MODE |
                                MPI2_SCSIIO_EEDPFLAGS_CHECK_GUARD);
                } else {
                        io_request->EEDPFlags = cpu_to_le16(
                                MPI2_SCSIIO_EEDPFLAGS_INC_PRI_REFTAG |
                                MPI2_SCSIIO_EEDPFLAGS_INSERT_OP);
                }
                io_request->Control |= cpu_to_le32((0x4 << 26));
                io_request->EEDPBlockSize = cpu_to_le32(scp->device->sector_size);
        } else {
                /* Some drives don't support 16/12 byte CDB's, convert to 10 */
                if (((cdb_len == 12) || (cdb_len == 16)) &&
                    (start_blk <= 0xffffffff)) {
                        if (cdb_len == 16) {
                                opcode = cdb[0] == READ_16 ? READ_10 : WRITE_10;
                                flagvals = cdb[1];
                                groupnum = cdb[14];
                                control = cdb[15];
                        } else {
                                opcode = cdb[0] == READ_12 ? READ_10 : WRITE_10;
                                flagvals = cdb[1];
                                groupnum = cdb[10];
                                control = cdb[11];
                        }

                        memset(cdb, 0, sizeof(io_request->CDB.CDB32));

                        cdb[0] = opcode;
                        cdb[1] = flagvals;
                        cdb[6] = groupnum;
                        cdb[9] = control;

                        /* Transfer length */
                        cdb[8] = (u8)(num_blocks & 0xff);
                        cdb[7] = (u8)((num_blocks >> 8) & 0xff);

                        io_request->IoFlags = cpu_to_le16(10); /* Specify 10-byte cdb */
                        cdb_len = 10;
                } else if ((cdb_len < 16) && (start_blk > 0xffffffff)) {
                        /* Convert to 16 byte CDB for large LBA's */
                        switch (cdb_len) {
                        case 6:
                                opcode = cdb[0] == READ_6 ? READ_16 : WRITE_16;
                                control = cdb[5];
                                break;
                        case 10:
                                opcode =
                                        cdb[0] == READ_10 ? READ_16 : WRITE_16;
                                flagvals = cdb[1];
                                groupnum = cdb[6];
                                control = cdb[9];
                                break;
                        case 12:
                                opcode =
                                        cdb[0] == READ_12 ? READ_16 : WRITE_16;
                                flagvals = cdb[1];
                                groupnum = cdb[10];
                                control = cdb[11];
                                break;
                        }

                        memset(cdb, 0, sizeof(io_request->CDB.CDB32));

                        cdb[0] = opcode;
                        cdb[1] = flagvals;
                        cdb[14] = groupnum;
                        cdb[15] = control;

                        /* Transfer length */
                        cdb[13] = (u8)(num_blocks & 0xff);
                        cdb[12] = (u8)((num_blocks >> 8) & 0xff);
                        cdb[11] = (u8)((num_blocks >> 16) & 0xff);
                        cdb[10] = (u8)((num_blocks >> 24) & 0xff);

                        io_request->IoFlags = cpu_to_le16(16); /* Specify 16-byte cdb */
                        cdb_len = 16;
                }

                /* Normal case, just load LBA here */
                switch (cdb_len) {
                case 6:
                {
                        u8 val = cdb[1] & 0xE0;
                        cdb[3] = (u8)(start_blk & 0xff);
                        cdb[2] = (u8)((start_blk >> 8) & 0xff);
                        cdb[1] = val | ((u8)(start_blk >> 16) & 0x1f);
                        break;
                }
                case 10:
                        cdb[5] = (u8)(start_blk & 0xff);
                        cdb[4] = (u8)((start_blk >> 8) & 0xff);
                        cdb[3] = (u8)((start_blk >> 16) & 0xff);
                        cdb[2] = (u8)((start_blk >> 24) & 0xff);
                        break;
                case 12:
                        cdb[5]    = (u8)(start_blk & 0xff);
                        cdb[4]    = (u8)((start_blk >> 8) & 0xff);
                        cdb[3]    = (u8)((start_blk >> 16) & 0xff);
                        cdb[2]    = (u8)((start_blk >> 24) & 0xff);
                        break;
                case 16:
                        cdb[9]    = (u8)(start_blk & 0xff);
                        cdb[8]    = (u8)((start_blk >> 8) & 0xff);
                        cdb[7]    = (u8)((start_blk >> 16) & 0xff);
                        cdb[6]    = (u8)((start_blk >> 24) & 0xff);
                        cdb[5]    = (u8)((start_blk >> 32) & 0xff);
                        cdb[4]    = (u8)((start_blk >> 40) & 0xff);
                        cdb[3]    = (u8)((start_blk >> 48) & 0xff);
                        cdb[2]    = (u8)((start_blk >> 56) & 0xff);
                        break;
                }
        }
}

/**
 * megasas_stream_detect -      stream detection on read and and write IOs
 * @instance:           Adapter soft state
 * @cmd:                    Command to be prepared
 * @io_info:            IO Request info
 *
 */

/** stream detection on read and and write IOs */
static void megasas_stream_detect(struct megasas_instance *instance,
                                  struct megasas_cmd_fusion *cmd,
                                  struct IO_REQUEST_INFO *io_info)
{
        struct fusion_context *fusion = instance->ctrl_context;
        u32 device_id = io_info->ldTgtId;
        struct LD_STREAM_DETECT *current_ld_sd
                = fusion->stream_detect_by_ld[device_id];
        u32 *track_stream = &current_ld_sd->mru_bit_map, stream_num;
        u32 shifted_values, unshifted_values;
        u32 index_value_mask, shifted_values_mask;
        int i;
        bool is_read_ahead = false;
        struct STREAM_DETECT *current_sd;
        /* find possible stream */
        for (i = 0; i < MAX_STREAMS_TRACKED; ++i) {
                stream_num = (*track_stream >>
                        (i * BITS_PER_INDEX_STREAM)) &
                        STREAM_MASK;
                current_sd = &current_ld_sd->stream_track[stream_num];
                /* if we found a stream, update the raid
                 *  context and also update the mruBitMap
                 */
                /*      boundary condition */
                if ((current_sd->next_seq_lba) &&
                    (io_info->ldStartBlock >= current_sd->next_seq_lba) &&
                    (io_info->ldStartBlock <= (current_sd->next_seq_lba + 32)) &&
                    (current_sd->is_read == io_info->isRead)) {

                        if ((io_info->ldStartBlock != current_sd->next_seq_lba) &&
                            ((!io_info->isRead) || (!is_read_ahead)))
                                /*
                                 * Once the API availible we need to change this.
                                 * At this point we are not allowing any gap
                                 */
                                continue;

                        SET_STREAM_DETECTED(cmd->io_request->RaidContext.raid_context_g35);
                        current_sd->next_seq_lba =
                        io_info->ldStartBlock + io_info->numBlocks;
                        /*
                         *      update the mruBitMap LRU
                         */
                        shifted_values_mask =
                                (1 <<  i * BITS_PER_INDEX_STREAM) - 1;
                        shifted_values = ((*track_stream & shifted_values_mask)
                                                << BITS_PER_INDEX_STREAM);
                        index_value_mask =
                                STREAM_MASK << i * BITS_PER_INDEX_STREAM;
                        unshifted_values =
                                *track_stream & ~(shifted_values_mask |
                                index_value_mask);
                        *track_stream =
                                unshifted_values | shifted_values | stream_num;
                        return;
                }
        }
        /*
         * if we did not find any stream, create a new one
         * from the least recently used
         */
        stream_num = (*track_stream >>
                ((MAX_STREAMS_TRACKED - 1) * BITS_PER_INDEX_STREAM)) &
                STREAM_MASK;
        current_sd = &current_ld_sd->stream_track[stream_num];
        current_sd->is_read = io_info->isRead;
        current_sd->next_seq_lba = io_info->ldStartBlock + io_info->numBlocks;
        *track_stream = (((*track_stream & ZERO_LAST_STREAM) << 4) | stream_num);
        return;
}

/**
 * megasas_set_raidflag_cpu_affinity - This function sets the cpu
 * affinity (cpu of the controller) and raid_flags in the raid context
 * based on IO type.
 *
 * @fusion:             Fusion context
 * @praid_context:      IO RAID context
 * @raid:               LD raid map
 * @fp_possible:        Is fast path possible?
 * @is_read:            Is read IO?
 * @scsi_buff_len:      SCSI command buffer length
 *
 */
static void
megasas_set_raidflag_cpu_affinity(struct fusion_context *fusion,
                                union RAID_CONTEXT_UNION *praid_context,
                                struct MR_LD_RAID *raid, bool fp_possible,
                                u8 is_read, u32 scsi_buff_len)
{
        u8 cpu_sel = MR_RAID_CTX_CPUSEL_0;
        struct RAID_CONTEXT_G35 *rctx_g35;

        rctx_g35 = &praid_context->raid_context_g35;
        if (fp_possible) {
                if (is_read) {
                        if ((raid->cpuAffinity.pdRead.cpu0) &&
                            (raid->cpuAffinity.pdRead.cpu1))
                                cpu_sel = MR_RAID_CTX_CPUSEL_FCFS;
                        else if (raid->cpuAffinity.pdRead.cpu1)
                                cpu_sel = MR_RAID_CTX_CPUSEL_1;
                } else {
                        if ((raid->cpuAffinity.pdWrite.cpu0) &&
                            (raid->cpuAffinity.pdWrite.cpu1))
                                cpu_sel = MR_RAID_CTX_CPUSEL_FCFS;
                        else if (raid->cpuAffinity.pdWrite.cpu1)
                                cpu_sel = MR_RAID_CTX_CPUSEL_1;
                        /* Fast path cache by pass capable R0/R1 VD */
                        if ((raid->level <= 1) &&
                            (raid->capability.fp_cache_bypass_capable)) {
                                rctx_g35->routing_flags |=
                                        (1 << MR_RAID_CTX_ROUTINGFLAGS_SLD_SHIFT);
                                rctx_g35->raid_flags =
                                        (MR_RAID_FLAGS_IO_SUB_TYPE_CACHE_BYPASS
                                        << MR_RAID_CTX_RAID_FLAGS_IO_SUB_TYPE_SHIFT);
                        }
                }
        } else {
                if (is_read) {
                        if ((raid->cpuAffinity.ldRead.cpu0) &&
                            (raid->cpuAffinity.ldRead.cpu1))
                                cpu_sel = MR_RAID_CTX_CPUSEL_FCFS;
                        else if (raid->cpuAffinity.ldRead.cpu1)
                                cpu_sel = MR_RAID_CTX_CPUSEL_1;
                } else {
                        if ((raid->cpuAffinity.ldWrite.cpu0) &&
                            (raid->cpuAffinity.ldWrite.cpu1))
                                cpu_sel = MR_RAID_CTX_CPUSEL_FCFS;
                        else if (raid->cpuAffinity.ldWrite.cpu1)
                                cpu_sel = MR_RAID_CTX_CPUSEL_1;

                        if (is_stream_detected(rctx_g35) &&
                            ((raid->level == 5) || (raid->level == 6)) &&
                            (raid->writeMode == MR_RL_WRITE_THROUGH_MODE) &&
                            (cpu_sel == MR_RAID_CTX_CPUSEL_FCFS))
                                cpu_sel = MR_RAID_CTX_CPUSEL_0;
                }
        }

        rctx_g35->routing_flags |=
                (cpu_sel << MR_RAID_CTX_ROUTINGFLAGS_CPUSEL_SHIFT);

        /* Always give priority to MR_RAID_FLAGS_IO_SUB_TYPE_LDIO_BW_LIMIT
         * vs MR_RAID_FLAGS_IO_SUB_TYPE_CACHE_BYPASS.
         * IO Subtype is not bitmap.
         */
        if ((fusion->pcie_bw_limitation) && (raid->level == 1) && (!is_read) &&
                        (scsi_buff_len > MR_LARGE_IO_MIN_SIZE)) {
                praid_context->raid_context_g35.raid_flags =
                        (MR_RAID_FLAGS_IO_SUB_TYPE_LDIO_BW_LIMIT
                        << MR_RAID_CTX_RAID_FLAGS_IO_SUB_TYPE_SHIFT);
        }
}

/**
 * megasas_build_ldio_fusion -  Prepares IOs to devices
 * @instance:           Adapter soft state
 * @scp:                SCSI command
 * @cmd:                Command to be prepared
 *
 * Prepares the io_request and chain elements (sg_frame) for IO
 * The IO can be for PD (Fast Path) or LD
 */
static void
megasas_build_ldio_fusion(struct megasas_instance *instance,
                          struct scsi_cmnd *scp,
                          struct megasas_cmd_fusion *cmd)
{
        bool fp_possible;
        u16 ld;
        u32 start_lba_lo, start_lba_hi, device_id, datalength = 0;
        u32 scsi_buff_len;
        struct MPI2_RAID_SCSI_IO_REQUEST *io_request;
        struct IO_REQUEST_INFO io_info;
        struct fusion_context *fusion;
        struct MR_DRV_RAID_MAP_ALL *local_map_ptr;
        u8 *raidLUN;
        unsigned long spinlock_flags;
        struct MR_LD_RAID *raid = NULL;
        struct MR_PRIV_DEVICE *mrdev_priv;
        struct RAID_CONTEXT *rctx;
        struct RAID_CONTEXT_G35 *rctx_g35;

        device_id = MEGASAS_DEV_INDEX(scp);

        fusion = instance->ctrl_context;

        io_request = cmd->io_request;
        rctx = &io_request->RaidContext.raid_context;
        rctx_g35 = &io_request->RaidContext.raid_context_g35;

        rctx->virtual_disk_tgt_id = cpu_to_le16(device_id);
        rctx->status = 0;
        rctx->ex_status = 0;

        start_lba_lo = 0;
        start_lba_hi = 0;
        fp_possible = false;

        /*
         * 6-byte READ(0x08) or WRITE(0x0A) cdb
         */
        if (scp->cmd_len == 6) {
                datalength = (u32) scp->cmnd[4];
                start_lba_lo = ((u32) scp->cmnd[1] << 16) |
                        ((u32) scp->cmnd[2] << 8) | (u32) scp->cmnd[3];

                start_lba_lo &= 0x1FFFFF;
        }

        /*
         * 10-byte READ(0x28) or WRITE(0x2A) cdb
         */
        else if (scp->cmd_len == 10) {
                datalength = (u32) scp->cmnd[8] |
                        ((u32) scp->cmnd[7] << 8);
                start_lba_lo = ((u32) scp->cmnd[2] << 24) |
                        ((u32) scp->cmnd[3] << 16) |
                        ((u32) scp->cmnd[4] << 8) | (u32) scp->cmnd[5];
        }

        /*
         * 12-byte READ(0xA8) or WRITE(0xAA) cdb
         */
        else if (scp->cmd_len == 12) {
                datalength = ((u32) scp->cmnd[6] << 24) |
                        ((u32) scp->cmnd[7] << 16) |
                        ((u32) scp->cmnd[8] << 8) | (u32) scp->cmnd[9];
                start_lba_lo = ((u32) scp->cmnd[2] << 24) |
                        ((u32) scp->cmnd[3] << 16) |
                        ((u32) scp->cmnd[4] << 8) | (u32) scp->cmnd[5];
        }

        /*
         * 16-byte READ(0x88) or WRITE(0x8A) cdb
         */
        else if (scp->cmd_len == 16) {
                datalength = ((u32) scp->cmnd[10] << 24) |
                        ((u32) scp->cmnd[11] << 16) |
                        ((u32) scp->cmnd[12] << 8) | (u32) scp->cmnd[13];
                start_lba_lo = ((u32) scp->cmnd[6] << 24) |
                        ((u32) scp->cmnd[7] << 16) |
                        ((u32) scp->cmnd[8] << 8) | (u32) scp->cmnd[9];

                start_lba_hi = ((u32) scp->cmnd[2] << 24) |
                        ((u32) scp->cmnd[3] << 16) |
                        ((u32) scp->cmnd[4] << 8) | (u32) scp->cmnd[5];
        }

        memset(&io_info, 0, sizeof(struct IO_REQUEST_INFO));
        io_info.ldStartBlock = ((u64)start_lba_hi << 32) | start_lba_lo;
        io_info.numBlocks = datalength;
        io_info.ldTgtId = device_id;
        io_info.r1_alt_dev_handle = MR_DEVHANDLE_INVALID;
        scsi_buff_len = scsi_bufflen(scp);
        io_request->DataLength = cpu_to_le32(scsi_buff_len);
        io_info.data_arms = 1;

        if (scp->sc_data_direction == DMA_FROM_DEVICE)
                io_info.isRead = 1;

        local_map_ptr = fusion->ld_drv_map[(instance->map_id & 1)];
        ld = MR_TargetIdToLdGet(device_id, local_map_ptr);

        if (ld < instance->fw_supported_vd_count)
                raid = MR_LdRaidGet(ld, local_map_ptr);

        if (!raid || (!fusion->fast_path_io)) {
                rctx->reg_lock_flags  = 0;
                fp_possible = false;
        } else {
                if (MR_BuildRaidContext(instance, &io_info, rctx,
                                        local_map_ptr, &raidLUN))
                        fp_possible = (io_info.fpOkForIo > 0) ? true : false;
        }

        megasas_get_msix_index(instance, scp, cmd, io_info.data_arms);

        if (instance->adapter_type >= VENTURA_SERIES) {
                /* FP for Optimal raid level 1.
                 * All large RAID-1 writes (> 32 KiB, both WT and WB modes)
                 * are built by the driver as LD I/Os.
                 * All small RAID-1 WT writes (<= 32 KiB) are built as FP I/Os
                 * (there is never a reason to process these as buffered writes)
                 * All small RAID-1 WB writes (<= 32 KiB) are built as FP I/Os
                 * with the SLD bit asserted.
                 */
                if (io_info.r1_alt_dev_handle != MR_DEVHANDLE_INVALID) {
                        mrdev_priv = scp->device->hostdata;

                        if (atomic_inc_return(&instance->fw_outstanding) >
                                (instance->host->can_queue)) {
                                fp_possible = false;
                                atomic_dec(&instance->fw_outstanding);
                        } else if (fusion->pcie_bw_limitation &&
                                ((scsi_buff_len > MR_LARGE_IO_MIN_SIZE) ||
                                   (atomic_dec_if_positive(&mrdev_priv->r1_ldio_hint) > 0))) {
                                fp_possible = false;
                                atomic_dec(&instance->fw_outstanding);
                                if (scsi_buff_len > MR_LARGE_IO_MIN_SIZE)
                                        atomic_set(&mrdev_priv->r1_ldio_hint,
                                                   instance->r1_ldio_hint_default);
                        }
                }

                if (!fp_possible ||
                    (io_info.isRead && io_info.ra_capable)) {
                        spin_lock_irqsave(&instance->stream_lock,
                                          spinlock_flags);
                        megasas_stream_detect(instance, cmd, &io_info);
                        spin_unlock_irqrestore(&instance->stream_lock,
                                               spinlock_flags);
                        /* In ventura if stream detected for a read and it is
                         * read ahead capable make this IO as LDIO
                         */
                        if (is_stream_detected(rctx_g35))
                                fp_possible = false;
                }

                /* If raid is NULL, set CPU affinity to default CPU0 */
                if (raid)
                        megasas_set_raidflag_cpu_affinity(fusion, &io_request->RaidContext,
                                raid, fp_possible, io_info.isRead,
                                scsi_buff_len);
                else
                        rctx_g35->routing_flags |=
                                (MR_RAID_CTX_CPUSEL_0 << MR_RAID_CTX_ROUTINGFLAGS_CPUSEL_SHIFT);
        }

        if (fp_possible) {
                megasas_set_pd_lba(io_request, scp->cmd_len, &io_info, scp,
                                   local_map_ptr, start_lba_lo);
                io_request->Function = MPI2_FUNCTION_SCSI_IO_REQUEST;
                cmd->request_desc->SCSIIO.RequestFlags =
                        (MPI2_REQ_DESCRIPT_FLAGS_FP_IO
                         << MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
                if (instance->adapter_type == INVADER_SERIES) {
                        rctx->type = MPI2_TYPE_CUDA;
                        rctx->nseg = 0x1;
                        io_request->IoFlags |= cpu_to_le16(MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH);
                        rctx->reg_lock_flags |=
                          (MR_RL_FLAGS_GRANT_DESTINATION_CUDA |
                           MR_RL_FLAGS_SEQ_NUM_ENABLE);
                } else if (instance->adapter_type >= VENTURA_SERIES) {
                        rctx_g35->nseg_type |= (1 << RAID_CONTEXT_NSEG_SHIFT);
                        rctx_g35->nseg_type |= (MPI2_TYPE_CUDA << RAID_CONTEXT_TYPE_SHIFT);
                        rctx_g35->routing_flags |= (1 << MR_RAID_CTX_ROUTINGFLAGS_SQN_SHIFT);
                        io_request->IoFlags |=
                                cpu_to_le16(MPI25_SAS_DEVICE0_FLAGS_ENABLED_FAST_PATH);
                }
                if (fusion->load_balance_info &&
                        (fusion->load_balance_info[device_id].loadBalanceFlag) &&
                        (io_info.isRead)) {
                        io_info.devHandle =
                                get_updated_dev_handle(instance,
                                        &fusion->load_balance_info[device_id],
                                        &io_info, local_map_ptr);
                        scp->SCp.Status |= MEGASAS_LOAD_BALANCE_FLAG;
                        cmd->pd_r1_lb = io_info.pd_after_lb;
                        if (instance->adapter_type >= VENTURA_SERIES)
                                rctx_g35->span_arm = io_info.span_arm;
                        else
                                rctx->span_arm = io_info.span_arm;

                } else
                        scp->SCp.Status &= ~MEGASAS_LOAD_BALANCE_FLAG;

                if (instance->adapter_type >= VENTURA_SERIES)
                        cmd->r1_alt_dev_handle = io_info.r1_alt_dev_handle;
                else
                        cmd->r1_alt_dev_handle = MR_DEVHANDLE_INVALID;

                if ((raidLUN[0] == 1) &&
                        (local_map_ptr->raidMap.devHndlInfo[io_info.pd_after_lb].validHandles > 1)) {
                        instance->dev_handle = !(instance->dev_handle);
                        io_info.devHandle =
                                local_map_ptr->raidMap.devHndlInfo[io_info.pd_after_lb].devHandle[instance->dev_handle];
                }

                cmd->request_desc->SCSIIO.DevHandle = io_info.devHandle;
                io_request->DevHandle = io_info.devHandle;
                cmd->pd_interface = io_info.pd_interface;
                /* populate the LUN field */
                memcpy(io_request->LUN, raidLUN, 8);
        } else {
                rctx->timeout_value =
                        cpu_to_le16(local_map_ptr->raidMap.fpPdIoTimeoutSec);
                cmd->request_desc->SCSIIO.RequestFlags =
                        (MEGASAS_REQ_DESCRIPT_FLAGS_LD_IO
                         << MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
                if (instance->adapter_type == INVADER_SERIES) {
                        if (io_info.do_fp_rlbypass ||
                        (rctx->reg_lock_flags == REGION_TYPE_UNUSED))
                                cmd->request_desc->SCSIIO.RequestFlags =
                                        (MEGASAS_REQ_DESCRIPT_FLAGS_NO_LOCK <<
                                        MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
                        rctx->type = MPI2_TYPE_CUDA;
                        rctx->reg_lock_flags |=
                                (MR_RL_FLAGS_GRANT_DESTINATION_CPU0 |
                                        MR_RL_FLAGS_SEQ_NUM_ENABLE);
                        rctx->nseg = 0x1;
                } else if (instance->adapter_type >= VENTURA_SERIES) {
                        rctx_g35->routing_flags |= (1 << MR_RAID_CTX_ROUTINGFLAGS_SQN_SHIFT);
                        rctx_g35->nseg_type |= (1 << RAID_CONTEXT_NSEG_SHIFT);
                        rctx_g35->nseg_type |= (MPI2_TYPE_CUDA << RAID_CONTEXT_TYPE_SHIFT);
                }
                io_request->Function = MEGASAS_MPI2_FUNCTION_LD_IO_REQUEST;
                io_request->DevHandle = cpu_to_le16(device_id);

        } /* Not FP */
}

/**
 * megasas_build_ld_nonrw_fusion - prepares non rw ios for virtual disk
 * @instance:           Adapter soft state
 * @scmd:               SCSI command
 * @cmd:                Command to be prepared
 *
 * Prepares the io_request frame for non-rw io cmds for vd.
 */
static void megasas_build_ld_nonrw_fusion(struct megasas_instance *instance,
                          struct scsi_cmnd *scmd, struct megasas_cmd_fusion *cmd)
{
        u32 device_id;
        struct MPI2_RAID_SCSI_IO_REQUEST *io_request;
        u16 ld;
        struct MR_DRV_RAID_MAP_ALL *local_map_ptr;
        struct fusion_context *fusion = instance->ctrl_context;
        u8                          span, physArm;
        __le16                      devHandle;
        u32                         arRef, pd;
        struct MR_LD_RAID                  *raid;
        struct RAID_CONTEXT                *pRAID_Context;
        u8 fp_possible = 1;

        io_request = cmd->io_request;
        device_id = MEGASAS_DEV_INDEX(scmd);
        local_map_ptr = fusion->ld_drv_map[(instance->map_id & 1)];
        io_request->DataLength = cpu_to_le32(scsi_bufflen(scmd));
        /* get RAID_Context pointer */
        pRAID_Context = &io_request->RaidContext.raid_context;
        /* Check with FW team */
        pRAID_Context->virtual_disk_tgt_id = cpu_to_le16(device_id);
        pRAID_Context->reg_lock_row_lba    = 0;
        pRAID_Context->reg_lock_length    = 0;

        if (fusion->fast_path_io && (
                device_id < instance->fw_supported_vd_count)) {

                ld = MR_TargetIdToLdGet(device_id, local_map_ptr);
                if (ld >= instance->fw_supported_vd_count - 1)
                        fp_possible = 0;
                else {
                        raid = MR_LdRaidGet(ld, local_map_ptr);
                        if (!(raid->capability.fpNonRWCapable))
                                fp_possible = 0;
                }
        } else
                fp_possible = 0;

        if (!fp_possible) {
                io_request->Function  = MEGASAS_MPI2_FUNCTION_LD_IO_REQUEST;
                io_request->DevHandle = cpu_to_le16(device_id);
                io_request->LUN[1] = scmd->device->lun;
                pRAID_Context->timeout_value =
                        cpu_to_le16 (scmd->request->timeout / HZ);
                cmd->request_desc->SCSIIO.RequestFlags =
                        (MPI2_REQ_DESCRIPT_FLAGS_SCSI_IO <<
                        MEGASAS_REQ_DESCRIPT_FLAGS_TYPE_SHIFT);
        } else {

                /* set RAID context values */
                pRAID_Context->config_seq_num = raid->seqNum;
                if (instance->adapter_type < VENTURA_SERIES)
                        pRAID_Context->reg_lock_flags = REGION_TYPE_SHARED_READ;
                pRAID_Context->timeout_value =
                        cpu_to_le16(raid->fpIoTimeoutForLd);