/*
 *    Disk Array driver for HP Smart Array controllers.
 *    (C) Copyright 2000, 2007 Hewlett-Packard Development Company, L.P.
 *
 *    This program is free software; you can redistribute it and/or modify
 *    it under the terms of the GNU General Public License as published by
 *    the Free Software Foundation; version 2 of the License.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. See the GNU
 *    General Public License for more details.
 *
 *    You should have received a copy of the GNU General Public License
 *    along with this program; if not, write to the Free Software
 *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
 *    02111-1307, USA.
 *
 *    Questions/Comments/Bugfixes to iss_storagedev@hp.com
 *
 */

#include <linux/module.h>
#include <linux/interrupt.h>
#include <linux/types.h>
#include <linux/pci.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/major.h>
#include <linux/fs.h>
#include <linux/bio.h>
#include <linux/blkpg.h>
#include <linux/timer.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/init.h>
#include <linux/jiffies.h>
#include <linux/hdreg.h>
#include <linux/spinlock.h>
#include <linux/compat.h>
#include <linux/mutex.h>
#include <asm/uaccess.h>
#include <asm/io.h>

#include <linux/dma-mapping.h>
#include <linux/blkdev.h>
#include <linux/genhd.h>
#include <linux/completion.h>
#include <scsi/scsi.h>
#include <scsi/sg.h>
#include <scsi/scsi_ioctl.h>
#include <linux/cdrom.h>
#include <linux/scatterlist.h>
#include <linux/kthread.h>

#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin))
#define DRIVER_NAME "HP CISS Driver (v 3.6.26)"
#define DRIVER_VERSION CCISS_DRIVER_VERSION(3, 6, 26)

/* Embedded module documentation macros - see modules.h */
MODULE_AUTHOR("Hewlett-Packard Company");
MODULE_DESCRIPTION("Driver for HP Smart Array Controllers");
MODULE_SUPPORTED_DEVICE("HP Smart Array Controllers");
MODULE_VERSION("3.6.26");
MODULE_LICENSE("GPL");

static DEFINE_MUTEX(cciss_mutex);
static struct proc_dir_entry *proc_cciss;

#include "cciss_cmd.h"
#include "cciss.h"
#include <linux/cciss_ioctl.h>

/* define the PCI info for the cards we can control */
static const struct pci_device_id cciss_pci_device_id[] = {
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS,  0x0E11, 0x4070},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4080},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4082},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSB, 0x0E11, 0x4083},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x4091},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409A},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409B},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409C},
        {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISSC, 0x0E11, 0x409D},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSA,     0x103C, 0x3225},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSC,     0x103C, 0x3223},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSC,     0x103C, 0x3234},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSC,     0x103C, 0x3235},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSD,     0x103C, 0x3211},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSD,     0x103C, 0x3212},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSD,     0x103C, 0x3213},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSD,     0x103C, 0x3214},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSD,     0x103C, 0x3215},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSC,     0x103C, 0x3237},
        {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSC,     0x103C, 0x323D},
        {0,}
};

MODULE_DEVICE_TABLE(pci, cciss_pci_device_id);

/*  board_id = Subsystem Device ID & Vendor ID
 *  product = Marketing Name for the board
 *  access = Address of the struct of function pointers
 */
static struct board_type products[] = {
        {0x40700E11, "Smart Array 5300", &SA5_access},
        {0x40800E11, "Smart Array 5i", &SA5B_access},
        {0x40820E11, "Smart Array 532", &SA5B_access},
        {0x40830E11, "Smart Array 5312", &SA5B_access},
        {0x409A0E11, "Smart Array 641", &SA5_access},
        {0x409B0E11, "Smart Array 642", &SA5_access},
        {0x409C0E11, "Smart Array 6400", &SA5_access},
        {0x409D0E11, "Smart Array 6400 EM", &SA5_access},
        {0x40910E11, "Smart Array 6i", &SA5_access},
        {0x3225103C, "Smart Array P600", &SA5_access},
        {0x3223103C, "Smart Array P800", &SA5_access},
        {0x3234103C, "Smart Array P400", &SA5_access},
        {0x3235103C, "Smart Array P400i", &SA5_access},
        {0x3211103C, "Smart Array E200i", &SA5_access},
        {0x3212103C, "Smart Array E200", &SA5_access},
        {0x3213103C, "Smart Array E200i", &SA5_access},
        {0x3214103C, "Smart Array E200i", &SA5_access},
        {0x3215103C, "Smart Array E200i", &SA5_access},
        {0x3237103C, "Smart Array E500", &SA5_access},
        {0x3223103C, "Smart Array P800", &SA5_access},
        {0x3234103C, "Smart Array P400", &SA5_access},
        {0x323D103C, "Smart Array P700m", &SA5_access},
};

/* How long to wait (in milliseconds) for board to go into simple mode */
#define MAX_CONFIG_WAIT 30000
#define MAX_IOCTL_CONFIG_WAIT 1000

/*define how many times we will try a command because of bus resets */
#define MAX_CMD_RETRIES 3

#define MAX_CTLR        32

/* Originally cciss driver only supports 8 major numbers */
#define MAX_CTLR_ORIG   8

static ctlr_info_t *hba[MAX_CTLR];

static struct task_struct *cciss_scan_thread;
static DEFINE_MUTEX(scan_mutex);
static LIST_HEAD(scan_q);

static void do_cciss_request(struct request_queue *q);
static irqreturn_t do_cciss_intx(int irq, void *dev_id);
static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id);
static int cciss_open(struct block_device *bdev, fmode_t mode);
static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode);
static int cciss_release(struct gendisk *disk, fmode_t mode);
static int do_ioctl(struct block_device *bdev, fmode_t mode,
                    unsigned int cmd, unsigned long arg);
static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
                       unsigned int cmd, unsigned long arg);
static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo);

static int cciss_revalidate(struct gendisk *disk);
static int rebuild_lun_table(ctlr_info_t *h, int first_time, int via_ioctl);
static int deregister_disk(ctlr_info_t *h, int drv_index,
                           int clear_all, int via_ioctl);

static void cciss_read_capacity(ctlr_info_t *h, int logvol,
                        sector_t *total_size, unsigned int *block_size);
static void cciss_read_capacity_16(ctlr_info_t *h, int logvol,
                        sector_t *total_size, unsigned int *block_size);
static void cciss_geometry_inquiry(ctlr_info_t *h, int logvol,
                        sector_t total_size,
                        unsigned int block_size, InquiryData_struct *inq_buff,
                                   drive_info_struct *drv);
static void __devinit cciss_interrupt_mode(ctlr_info_t *);
static void start_io(ctlr_info_t *h);
static int sendcmd_withirq(ctlr_info_t *h, __u8 cmd, void *buff, size_t size,
                        __u8 page_code, unsigned char scsi3addr[],
                        int cmd_type);
static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c,
        int attempt_retry);
static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c);

static int add_to_scan_list(struct ctlr_info *h);
static int scan_thread(void *data);
static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
static void cciss_hba_release(struct device *dev);
static void cciss_device_release(struct device *dev);
static void cciss_free_gendisk(ctlr_info_t *h, int drv_index);
static void cciss_free_drive_info(ctlr_info_t *h, int drv_index);
static inline u32 next_command(ctlr_info_t *h);
static int __devinit cciss_find_cfg_addrs(struct pci_dev *pdev,
        void __iomem *vaddr, u32 *cfg_base_addr, u64 *cfg_base_addr_index,
        u64 *cfg_offset);
static int __devinit cciss_pci_find_memory_BAR(struct pci_dev *pdev,
        unsigned long *memory_bar);


/* performant mode helper functions */
static void  calc_bucket_map(int *bucket, int num_buckets, int nsgs,
                                int *bucket_map);
static void cciss_put_controller_into_performant_mode(ctlr_info_t *h);

#ifdef CONFIG_PROC_FS
static void cciss_procinit(ctlr_info_t *h);
#else
static void cciss_procinit(ctlr_info_t *h)
{
}
#endif                          /* CONFIG_PROC_FS */

#ifdef CONFIG_COMPAT
static int cciss_compat_ioctl(struct block_device *, fmode_t,
                              unsigned, unsigned long);
#endif

static const struct block_device_operations cciss_fops = {
        .owner = THIS_MODULE,
        .open = cciss_unlocked_open,
        .release = cciss_release,
        .ioctl = do_ioctl,
        .getgeo = cciss_getgeo,
#ifdef CONFIG_COMPAT
        .compat_ioctl = cciss_compat_ioctl,
#endif
        .revalidate_disk = cciss_revalidate,
};

/* set_performant_mode: Modify the tag for cciss performant
 * set bit 0 for pull model, bits 3-1 for block fetch
 * register number
 */
static void set_performant_mode(ctlr_info_t *h, CommandList_struct *c)
{
        if (likely(h->transMethod == CFGTBL_Trans_Performant))
                c->busaddr |= 1 | (h->blockFetchTable[c->Header.SGList] << 1);
}

/*
 * Enqueuing and dequeuing functions for cmdlists.
 */
static inline void addQ(struct hlist_head *list, CommandList_struct *c)
{
        hlist_add_head(&c->list, list);
}

static inline void removeQ(CommandList_struct *c)
{
        /*
         * After kexec/dump some commands might still
         * be in flight, which the firmware will try
         * to complete. Resetting the firmware doesn't work
         * with old fw revisions, so we have to mark
         * them off as 'stale' to prevent the driver from
         * falling over.
         */
        if (WARN_ON(hlist_unhashed(&c->list))) {
                c->cmd_type = CMD_MSG_STALE;
                return;
        }

        hlist_del_init(&c->list);
}

static void enqueue_cmd_and_start_io(ctlr_info_t *h,
        CommandList_struct *c)
{
        unsigned long flags;
        set_performant_mode(h, c);
        spin_lock_irqsave(&h->lock, flags);
        addQ(&h->reqQ, c);
        h->Qdepth++;
        if (h->Qdepth > h->maxQsinceinit)
                h->maxQsinceinit = h->Qdepth;
        start_io(h);
        spin_unlock_irqrestore(&h->lock, flags);
}

static void cciss_free_sg_chain_blocks(SGDescriptor_struct **cmd_sg_list,
        int nr_cmds)
{
        int i;

        if (!cmd_sg_list)
                return;
        for (i = 0; i < nr_cmds; i++) {
                kfree(cmd_sg_list[i]);
                cmd_sg_list[i] = NULL;
        }
        kfree(cmd_sg_list);
}

static SGDescriptor_struct **cciss_allocate_sg_chain_blocks(
        ctlr_info_t *h, int chainsize, int nr_cmds)
{
        int j;
        SGDescriptor_struct **cmd_sg_list;

        if (chainsize <= 0)
                return NULL;

        cmd_sg_list = kmalloc(sizeof(*cmd_sg_list) * nr_cmds, GFP_KERNEL);
        if (!cmd_sg_list)
                return NULL;

        /* Build up chain blocks for each command */
        for (j = 0; j < nr_cmds; j++) {
                /* Need a block of chainsized s/g elements. */
                cmd_sg_list[j] = kmalloc((chainsize *
                        sizeof(*cmd_sg_list[j])), GFP_KERNEL);
                if (!cmd_sg_list[j]) {
                        dev_err(&h->pdev->dev, "Cannot get memory "
                                "for s/g chains.\n");
                        goto clean;
                }
        }
        return cmd_sg_list;
clean:
        cciss_free_sg_chain_blocks(cmd_sg_list, nr_cmds);
        return NULL;
}

static void cciss_unmap_sg_chain_block(ctlr_info_t *h, CommandList_struct *c)
{
        SGDescriptor_struct *chain_sg;
        u64bit temp64;

        if (c->Header.SGTotal <= h->max_cmd_sgentries)
                return;

        chain_sg = &c->SG[h->max_cmd_sgentries - 1];
        temp64.val32.lower = chain_sg->Addr.lower;
        temp64.val32.upper = chain_sg->Addr.upper;
        pci_unmap_single(h->pdev, temp64.val, chain_sg->Len, PCI_DMA_TODEVICE);
}

static void cciss_map_sg_chain_block(ctlr_info_t *h, CommandList_struct *c,
        SGDescriptor_struct *chain_block, int len)
{
        SGDescriptor_struct *chain_sg;
        u64bit temp64;

        chain_sg = &c->SG[h->max_cmd_sgentries - 1];
        chain_sg->Ext = CCISS_SG_CHAIN;
        chain_sg->Len = len;
        temp64.val = pci_map_single(h->pdev, chain_block, len,
                                PCI_DMA_TODEVICE);
        chain_sg->Addr.lower = temp64.val32.lower;
        chain_sg->Addr.upper = temp64.val32.upper;
}

#include "cciss_scsi.c"         /* For SCSI tape support */

static const char *raid_label[] = { "0", "4", "1(1+0)", "5", "5+1", "ADG",
        "UNKNOWN"
};
#define RAID_UNKNOWN (ARRAY_SIZE(raid_label)-1)

#ifdef CONFIG_PROC_FS

/*
 * Report information about this controller.
 */
#define ENG_GIG 1000000000
#define ENG_GIG_FACTOR (ENG_GIG/512)
#define ENGAGE_SCSI     "engage scsi"

static void cciss_seq_show_header(struct seq_file *seq)
{
        ctlr_info_t *h = seq->private;

        seq_printf(seq, "%s: HP %s Controller\n"
                "Board ID: 0x%08lx\n"
                "Firmware Version: %c%c%c%c\n"
                "IRQ: %d\n"
                "Logical drives: %d\n"
                "Current Q depth: %d\n"
                "Current # commands on controller: %d\n"
                "Max Q depth since init: %d\n"
                "Max # commands on controller since init: %d\n"
                "Max SG entries since init: %d\n",
                h->devname,
                h->product_name,
                (unsigned long)h->board_id,
                h->firm_ver[0], h->firm_ver[1], h->firm_ver[2],
                h->firm_ver[3], (unsigned int)h->intr[PERF_MODE_INT],
                h->num_luns,
                h->Qdepth, h->commands_outstanding,
                h->maxQsinceinit, h->max_outstanding, h->maxSG);

#ifdef CONFIG_CISS_SCSI_TAPE
        cciss_seq_tape_report(seq, h);
#endif /* CONFIG_CISS_SCSI_TAPE */
}

static void *cciss_seq_start(struct seq_file *seq, loff_t *pos)
{
        ctlr_info_t *h = seq->private;
        unsigned long flags;

        /* prevent displaying bogus info during configuration
         * or deconfiguration of a logical volume
         */
        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(&h->lock, flags);
                return ERR_PTR(-EBUSY);
        }
        h->busy_configuring = 1;
        spin_unlock_irqrestore(&h->lock, flags);

        if (*pos == 0)
                cciss_seq_show_header(seq);

        return pos;
}

static int cciss_seq_show(struct seq_file *seq, void *v)
{
        sector_t vol_sz, vol_sz_frac;
        ctlr_info_t *h = seq->private;
        unsigned ctlr = h->ctlr;
        loff_t *pos = v;
        drive_info_struct *drv = h->drv[*pos];

        if (*pos > h->highest_lun)
                return 0;

        if (drv == NULL) /* it's possible for h->drv[] to have holes. */
                return 0;

        if (drv->heads == 0)
                return 0;

        vol_sz = drv->nr_blocks;
        vol_sz_frac = sector_div(vol_sz, ENG_GIG_FACTOR);
        vol_sz_frac *= 100;
        sector_div(vol_sz_frac, ENG_GIG_FACTOR);

        if (drv->raid_level < 0 || drv->raid_level > RAID_UNKNOWN)
                drv->raid_level = RAID_UNKNOWN;
        seq_printf(seq, "cciss/c%dd%d:"
                        "\t%4u.%02uGB\tRAID %s\n",
                        ctlr, (int) *pos, (int)vol_sz, (int)vol_sz_frac,
                        raid_label[drv->raid_level]);
        return 0;
}

static void *cciss_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
        ctlr_info_t *h = seq->private;

        if (*pos > h->highest_lun)
                return NULL;
        *pos += 1;

        return pos;
}

static void cciss_seq_stop(struct seq_file *seq, void *v)
{
        ctlr_info_t *h = seq->private;

        /* Only reset h->busy_configuring if we succeeded in setting
         * it during cciss_seq_start. */
        if (v == ERR_PTR(-EBUSY))
                return;

        h->busy_configuring = 0;
}

static const struct seq_operations cciss_seq_ops = {
        .start = cciss_seq_start,
        .show  = cciss_seq_show,
        .next  = cciss_seq_next,
        .stop  = cciss_seq_stop,
};

static int cciss_seq_open(struct inode *inode, struct file *file)
{
        int ret = seq_open(file, &cciss_seq_ops);
        struct seq_file *seq = file->private_data;

        if (!ret)
                seq->private = PDE(inode)->data;

        return ret;
}

static ssize_t
cciss_proc_write(struct file *file, const char __user *buf,
                 size_t length, loff_t *ppos)
{
        int err;
        char *buffer;

#ifndef CONFIG_CISS_SCSI_TAPE
        return -EINVAL;
#endif

        if (!buf || length > PAGE_SIZE - 1)
                return -EINVAL;

        buffer = (char *)__get_free_page(GFP_KERNEL);
        if (!buffer)
                return -ENOMEM;

        err = -EFAULT;
        if (copy_from_user(buffer, buf, length))
                goto out;
        buffer[length] = '\0';

#ifdef CONFIG_CISS_SCSI_TAPE
        if (strncmp(ENGAGE_SCSI, buffer, sizeof ENGAGE_SCSI - 1) == 0) {
                struct seq_file *seq = file->private_data;
                ctlr_info_t *h = seq->private;

                err = cciss_engage_scsi(h);
                if (err == 0)
                        err = length;
        } else
#endif /* CONFIG_CISS_SCSI_TAPE */
                err = -EINVAL;
        /* might be nice to have "disengage" too, but it's not
           safely possible. (only 1 module use count, lock issues.) */

out:
        free_page((unsigned long)buffer);
        return err;
}

static const struct file_operations cciss_proc_fops = {
        .owner   = THIS_MODULE,
        .open    = cciss_seq_open,
        .read    = seq_read,
        .llseek  = seq_lseek,
        .release = seq_release,
        .write   = cciss_proc_write,
};

static void __devinit cciss_procinit(ctlr_info_t *h)
{
        struct proc_dir_entry *pde;

        if (proc_cciss == NULL)
                proc_cciss = proc_mkdir("driver/cciss", NULL);
        if (!proc_cciss)
                return;
        pde = proc_create_data(h->devname, S_IWUSR | S_IRUSR | S_IRGRP |
                                        S_IROTH, proc_cciss,
                                        &cciss_proc_fops, h);
}
#endif                          /* CONFIG_PROC_FS */

#define MAX_PRODUCT_NAME_LEN 19

#define to_hba(n) container_of(n, struct ctlr_info, dev)
#define to_drv(n) container_of(n, drive_info_struct, dev)

static ssize_t host_store_rescan(struct device *dev,
                                 struct device_attribute *attr,
                                 const char *buf, size_t count)
{
        struct ctlr_info *h = to_hba(dev);

        add_to_scan_list(h);
        wake_up_process(cciss_scan_thread);
        wait_for_completion_interruptible(&h->scan_wait);

        return count;
}
static DEVICE_ATTR(rescan, S_IWUSR, NULL, host_store_rescan);

static ssize_t dev_show_unique_id(struct device *dev,
                                 struct device_attribute *attr,
                                 char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        __u8 sn[16];
        unsigned long flags;
        int ret = 0;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring)
                ret = -EBUSY;
        else
                memcpy(sn, drv->serial_no, sizeof(sn));
        spin_unlock_irqrestore(&h->lock, flags);

        if (ret)
                return ret;
        else
                return snprintf(buf, 16 * 2 + 2,
                                "%02X%02X%02X%02X%02X%02X%02X%02X"
                                "%02X%02X%02X%02X%02X%02X%02X%02X\n",
                                sn[0], sn[1], sn[2], sn[3],
                                sn[4], sn[5], sn[6], sn[7],
                                sn[8], sn[9], sn[10], sn[11],
                                sn[12], sn[13], sn[14], sn[15]);
}
static DEVICE_ATTR(unique_id, S_IRUGO, dev_show_unique_id, NULL);

static ssize_t dev_show_vendor(struct device *dev,
                               struct device_attribute *attr,
                               char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        char vendor[VENDOR_LEN + 1];
        unsigned long flags;
        int ret = 0;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring)
                ret = -EBUSY;
        else
                memcpy(vendor, drv->vendor, VENDOR_LEN + 1);
        spin_unlock_irqrestore(&h->lock, flags);

        if (ret)
                return ret;
        else
                return snprintf(buf, sizeof(vendor) + 1, "%s\n", drv->vendor);
}
static DEVICE_ATTR(vendor, S_IRUGO, dev_show_vendor, NULL);

static ssize_t dev_show_model(struct device *dev,
                              struct device_attribute *attr,
                              char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        char model[MODEL_LEN + 1];
        unsigned long flags;
        int ret = 0;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring)
                ret = -EBUSY;
        else
                memcpy(model, drv->model, MODEL_LEN + 1);
        spin_unlock_irqrestore(&h->lock, flags);

        if (ret)
                return ret;
        else
                return snprintf(buf, sizeof(model) + 1, "%s\n", drv->model);
}
static DEVICE_ATTR(model, S_IRUGO, dev_show_model, NULL);

static ssize_t dev_show_rev(struct device *dev,
                            struct device_attribute *attr,
                            char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        char rev[REV_LEN + 1];
        unsigned long flags;
        int ret = 0;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring)
                ret = -EBUSY;
        else
                memcpy(rev, drv->rev, REV_LEN + 1);
        spin_unlock_irqrestore(&h->lock, flags);

        if (ret)
                return ret;
        else
                return snprintf(buf, sizeof(rev) + 1, "%s\n", drv->rev);
}
static DEVICE_ATTR(rev, S_IRUGO, dev_show_rev, NULL);

static ssize_t cciss_show_lunid(struct device *dev,
                                struct device_attribute *attr, char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        unsigned long flags;
        unsigned char lunid[8];

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(&h->lock, flags);
                return -EBUSY;
        }
        if (!drv->heads) {
                spin_unlock_irqrestore(&h->lock, flags);
                return -ENOTTY;
        }
        memcpy(lunid, drv->LunID, sizeof(lunid));
        spin_unlock_irqrestore(&h->lock, flags);
        return snprintf(buf, 20, "0x%02x%02x%02x%02x%02x%02x%02x%02x\n",
                lunid[0], lunid[1], lunid[2], lunid[3],
                lunid[4], lunid[5], lunid[6], lunid[7]);
}
static DEVICE_ATTR(lunid, S_IRUGO, cciss_show_lunid, NULL);

static ssize_t cciss_show_raid_level(struct device *dev,
                                     struct device_attribute *attr, char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        int raid;
        unsigned long flags;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(&h->lock, flags);
                return -EBUSY;
        }
        raid = drv->raid_level;
        spin_unlock_irqrestore(&h->lock, flags);
        if (raid < 0 || raid > RAID_UNKNOWN)
                raid = RAID_UNKNOWN;

        return snprintf(buf, strlen(raid_label[raid]) + 7, "RAID %s\n",
                        raid_label[raid]);
}
static DEVICE_ATTR(raid_level, S_IRUGO, cciss_show_raid_level, NULL);

static ssize_t cciss_show_usage_count(struct device *dev,
                                      struct device_attribute *attr, char *buf)
{
        drive_info_struct *drv = to_drv(dev);
        struct ctlr_info *h = to_hba(drv->dev.parent);
        unsigned long flags;
        int count;

        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(&h->lock, flags);
                return -EBUSY;
        }
        count = drv->usage_count;
        spin_unlock_irqrestore(&h->lock, flags);
        return snprintf(buf, 20, "%d\n", count);
}
static DEVICE_ATTR(usage_count, S_IRUGO, cciss_show_usage_count, NULL);

static struct attribute *cciss_host_attrs[] = {
        &dev_attr_rescan.attr,
        NULL
};

static struct attribute_group cciss_host_attr_group = {
        .attrs = cciss_host_attrs,
};

static const struct attribute_group *cciss_host_attr_groups[] = {
        &cciss_host_attr_group,
        NULL
};

static struct device_type cciss_host_type = {
        .name           = "cciss_host",
        .groups         = cciss_host_attr_groups,
        .release        = cciss_hba_release,
};

static struct attribute *cciss_dev_attrs[] = {
        &dev_attr_unique_id.attr,
        &dev_attr_model.attr,
        &dev_attr_vendor.attr,
        &dev_attr_rev.attr,
        &dev_attr_lunid.attr,
        &dev_attr_raid_level.attr,
        &dev_attr_usage_count.attr,
        NULL
};

static struct attribute_group cciss_dev_attr_group = {
        .attrs = cciss_dev_attrs,
};

static const struct attribute_group *cciss_dev_attr_groups[] = {
        &cciss_dev_attr_group,
        NULL
};

static struct device_type cciss_dev_type = {
        .name           = "cciss_device",
        .groups         = cciss_dev_attr_groups,
        .release        = cciss_device_release,
};

static struct bus_type cciss_bus_type = {
        .name           = "cciss",
};

/*
 * cciss_hba_release is called when the reference count
 * of h->dev goes to zero.
 */
static void cciss_hba_release(struct device *dev)
{
        /*
         * nothing to do, but need this to avoid a warning
         * about not having a release handler from lib/kref.c.
         */
}

/*
 * Initialize sysfs entry for each controller.  This sets up and registers
 * the 'cciss#' directory for each individual controller under
 * /sys/bus/pci/devices/<dev>/.
 */
static int cciss_create_hba_sysfs_entry(struct ctlr_info *h)
{
        device_initialize(&h->dev);
        h->dev.type = &cciss_host_type;
        h->dev.bus = &cciss_bus_type;
        dev_set_name(&h->dev, "%s", h->devname);
        h->dev.parent = &h->pdev->dev;

        return device_add(&h->dev);
}

/*
 * Remove sysfs entries for an hba.
 */
static void cciss_destroy_hba_sysfs_entry(struct ctlr_info *h)
{
        device_del(&h->dev);
        put_device(&h->dev); /* final put. */
}

/* cciss_device_release is called when the reference count
 * of h->drv[x]dev goes to zero.
 */
static void cciss_device_release(struct device *dev)
{
        drive_info_struct *drv = to_drv(dev);
        kfree(drv);
}

/*
 * Initialize sysfs for each logical drive.  This sets up and registers
 * the 'c#d#' directory for each individual logical drive under
 * /sys/bus/pci/devices/<dev/ccis#/. We also create a link from
 * /sys/block/cciss!c#d# to this entry.
 */
static long cciss_create_ld_sysfs_entry(struct ctlr_info *h,
                                       int drv_index)
{
        struct device *dev;

        if (h->drv[drv_index]->device_initialized)
                return 0;

        dev = &h->drv[drv_index]->dev;
        device_initialize(dev);
        dev->type = &cciss_dev_type;
        dev->bus = &cciss_bus_type;
        dev_set_name(dev, "c%dd%d", h->ctlr, drv_index);
        dev->parent = &h->dev;
        h->drv[drv_index]->device_initialized = 1;
        return device_add(dev);
}

/*
 * Remove sysfs entries for a logical drive.
 */
static void cciss_destroy_ld_sysfs_entry(struct ctlr_info *h, int drv_index,
        int ctlr_exiting)
{
        struct device *dev = &h->drv[drv_index]->dev;

        /* special case for c*d0, we only destroy it on controller exit */
        if (drv_index == 0 && !ctlr_exiting)
                return;

        device_del(dev);
        put_device(dev); /* the "final" put. */
        h->drv[drv_index] = NULL;
}

/*
 * For operations that cannot sleep, a command block is allocated at init,
 * and managed by cmd_alloc() and cmd_free() using a simple bitmap to track
 * which ones are free or in use.
 */
static CommandList_struct *cmd_alloc(ctlr_info_t *h)
{
        CommandList_struct *c;
        int i;
        u64bit temp64;
        dma_addr_t cmd_dma_handle, err_dma_handle;

        do {
                i = find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds);
                if (i == h->nr_cmds)
                        return NULL;
        } while (test_and_set_bit(i & (BITS_PER_LONG - 1),
                  h->cmd_pool_bits + (i / BITS_PER_LONG)) != 0);
        c = h->cmd_pool + i;
        memset(c, 0, sizeof(CommandList_struct));
        cmd_dma_handle = h->cmd_pool_dhandle + i * sizeof(CommandList_struct);
        c->err_info = h->errinfo_pool + i;
        memset(c->err_info, 0, sizeof(ErrorInfo_struct));
        err_dma_handle = h->errinfo_pool_dhandle
            + i * sizeof(ErrorInfo_struct);
        h->nr_allocs++;

        c->cmdindex = i;

        INIT_HLIST_NODE(&c->list);
        c->busaddr = (__u32) cmd_dma_handle;
        temp64.val = (__u64) err_dma_handle;
        c->ErrDesc.Addr.lower = temp64.val32.lower;
        c->ErrDesc.Addr.upper = temp64.val32.upper;
        c->ErrDesc.Len = sizeof(ErrorInfo_struct);

        c->ctlr = h->ctlr;
        return c;
}

/* allocate a command using pci_alloc_consistent, used for ioctls,
 * etc., not for the main i/o path.
 */
static CommandList_struct *cmd_special_alloc(ctlr_info_t *h)
{
        CommandList_struct *c;
        u64bit temp64;
        dma_addr_t cmd_dma_handle, err_dma_handle;

        c = (CommandList_struct *) pci_alloc_consistent(h->pdev,
                sizeof(CommandList_struct), &cmd_dma_handle);
        if (c == NULL)
                return NULL;
        memset(c, 0, sizeof(CommandList_struct));

        c->cmdindex = -1;

        c->err_info = (ErrorInfo_struct *)
            pci_alloc_consistent(h->pdev, sizeof(ErrorInfo_struct),
                    &err_dma_handle);

        if (c->err_info == NULL) {
                pci_free_consistent(h->pdev,
                        sizeof(CommandList_struct), c, cmd_dma_handle);
                return NULL;
        }
        memset(c->err_info, 0, sizeof(ErrorInfo_struct));

        INIT_HLIST_NODE(&c->list);
        c->busaddr = (__u32) cmd_dma_handle;
        temp64.val = (__u64) err_dma_handle;
        c->ErrDesc.Addr.lower = temp64.val32.lower;
        c->ErrDesc.Addr.upper = temp64.val32.upper;
        c->ErrDesc.Len = sizeof(ErrorInfo_struct);

        c->ctlr = h->ctlr;
        return c;
}

static void cmd_free(ctlr_info_t *h, CommandList_struct *c)
{
        int i;

        i = c - h->cmd_pool;
        clear_bit(i & (BITS_PER_LONG - 1),
                  h->cmd_pool_bits + (i / BITS_PER_LONG));
        h->nr_frees++;
}

static void cmd_special_free(ctlr_info_t *h, CommandList_struct *c)
{
        u64bit temp64;

        temp64.val32.lower = c->ErrDesc.Addr.lower;
        temp64.val32.upper = c->ErrDesc.Addr.upper;
        pci_free_consistent(h->pdev, sizeof(ErrorInfo_struct),
                            c->err_info, (dma_addr_t) temp64.val);
        pci_free_consistent(h->pdev, sizeof(CommandList_struct),
                            c, (dma_addr_t) c->busaddr);
}

static inline ctlr_info_t *get_host(struct gendisk *disk)
{
        return disk->queue->queuedata;
}

static inline drive_info_struct *get_drv(struct gendisk *disk)
{
        return disk->private_data;
}

/*
 * Open.  Make sure the device is really there.
 */
static int cciss_open(struct block_device *bdev, fmode_t mode)
{
        ctlr_info_t *h = get_host(bdev->bd_disk);
        drive_info_struct *drv = get_drv(bdev->bd_disk);

        dev_dbg(&h->pdev->dev, "cciss_open %s\n", bdev->bd_disk->disk_name);
        if (drv->busy_configuring)
                return -EBUSY;

        /*
         * Root is allowed to open raw volume zero even if it's not configured
         * so array config can still work. Root is also allowed to open any
         * volume that has a LUN ID, so it can issue IOCTL to reread the
         * disk information.  I don't think I really like this
         * but I'm already using way to many device nodes to claim another one
         * for "raw controller".
         */
        if (drv->heads == 0) {
                if (MINOR(bdev->bd_dev) != 0) { /* not node 0? */
                        /* if not node 0 make sure it is a partition = 0 */
                        if (MINOR(bdev->bd_dev) & 0x0f) {
                                return -ENXIO;
                                /* if it is, make sure we have a LUN ID */
                        } else if (memcmp(drv->LunID, CTLR_LUNID,
                                sizeof(drv->LunID))) {
                                return -ENXIO;
                        }
                }
                if (!capable(CAP_SYS_ADMIN))
                        return -EPERM;
        }
        drv->usage_count++;
        h->usage_count++;
        return 0;
}

static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode)
{
        int ret;

        mutex_lock(&cciss_mutex);
        ret = cciss_open(bdev, mode);
        mutex_unlock(&cciss_mutex);

        return ret;
}

/*
 * Close.  Sync first.
 */
static int cciss_release(struct gendisk *disk, fmode_t mode)
{
        ctlr_info_t *h;
        drive_info_struct *drv;

        mutex_lock(&cciss_mutex);
        h = get_host(disk);
        drv = get_drv(disk);
        dev_dbg(&h->pdev->dev, "cciss_release %s\n", disk->disk_name);
        drv->usage_count--;
        h->usage_count--;
        mutex_unlock(&cciss_mutex);
        return 0;
}

static int do_ioctl(struct block_device *bdev, fmode_t mode,
                    unsigned cmd, unsigned long arg)
{
        int ret;
        mutex_lock(&cciss_mutex);
        ret = cciss_ioctl(bdev, mode, cmd, arg);
        mutex_unlock(&cciss_mutex);
        return ret;
}

#ifdef CONFIG_COMPAT

static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode,
                                  unsigned cmd, unsigned long arg);
static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode,
                                      unsigned cmd, unsigned long arg);

static int cciss_compat_ioctl(struct block_device *bdev, fmode_t mode,
                              unsigned cmd, unsigned long arg)
{
        switch (cmd) {
        case CCISS_GETPCIINFO:
        case CCISS_GETINTINFO:
        case CCISS_SETINTINFO:
        case CCISS_GETNODENAME:
        case CCISS_SETNODENAME:
        case CCISS_GETHEARTBEAT:
        case CCISS_GETBUSTYPES:
        case CCISS_GETFIRMVER:
        case CCISS_GETDRIVVER:
        case CCISS_REVALIDVOLS:
        case CCISS_DEREGDISK:
        case CCISS_REGNEWDISK:
        case CCISS_REGNEWD:
        case CCISS_RESCANDISK:
        case CCISS_GETLUNINFO:
                return do_ioctl(bdev, mode, cmd, arg);

        case CCISS_PASSTHRU32:
                return cciss_ioctl32_passthru(bdev, mode, cmd, arg);
        case CCISS_BIG_PASSTHRU32:
                return cciss_ioctl32_big_passthru(bdev, mode, cmd, arg);

        default:
                return -ENOIOCTLCMD;
        }
}

static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode,
                                  unsigned cmd, unsigned long arg)
{
        IOCTL32_Command_struct __user *arg32 =
            (IOCTL32_Command_struct __user *) arg;
        IOCTL_Command_struct arg64;
        IOCTL_Command_struct __user *p = compat_alloc_user_space(sizeof(arg64));
        int err;
        u32 cp;

        err = 0;
        err |=
            copy_from_user(&arg64.LUN_info, &arg32->LUN_info,
                           sizeof(arg64.LUN_info));
        err |=
            copy_from_user(&arg64.Request, &arg32->Request,
                           sizeof(arg64.Request));
        err |=
            copy_from_user(&arg64.error_info, &arg32->error_info,
                           sizeof(arg64.error_info));
        err |= get_user(arg64.buf_size, &arg32->buf_size);
        err |= get_user(cp, &arg32->buf);
        arg64.buf = compat_ptr(cp);
        err |= copy_to_user(p, &arg64, sizeof(arg64));

        if (err)
                return -EFAULT;

        err = do_ioctl(bdev, mode, CCISS_PASSTHRU, (unsigned long)p);
        if (err)
                return err;
        err |=
            copy_in_user(&arg32->error_info, &p->error_info,
                         sizeof(arg32->error_info));
        if (err)
                return -EFAULT;
        return err;
}

static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode,
                                      unsigned cmd, unsigned long arg)
{
        BIG_IOCTL32_Command_struct __user *arg32 =
            (BIG_IOCTL32_Command_struct __user *) arg;
        BIG_IOCTL_Command_struct arg64;
        BIG_IOCTL_Command_struct __user *p =
            compat_alloc_user_space(sizeof(arg64));
        int err;
        u32 cp;

        memset(&arg64, 0, sizeof(arg64));
        err = 0;
        err |=
            copy_from_user(&arg64.LUN_info, &arg32->LUN_info,
                           sizeof(arg64.LUN_info));
        err |=
            copy_from_user(&arg64.Request, &arg32->Request,
                           sizeof(arg64.Request));
        err |=
            copy_from_user(&arg64.error_info, &arg32->error_info,
                           sizeof(arg64.error_info));
        err |= get_user(arg64.buf_size, &arg32->buf_size);
        err |= get_user(arg64.malloc_size, &arg32->malloc_size);
        err |= get_user(cp, &arg32->buf);
        arg64.buf = compat_ptr(cp);
        err |= copy_to_user(p, &arg64, sizeof(arg64));

        if (err)
                return -EFAULT;

        err = do_ioctl(bdev, mode, CCISS_BIG_PASSTHRU, (unsigned long)p);
        if (err)
                return err;
        err |=
            copy_in_user(&arg32->error_info, &p->error_info,
                         sizeof(arg32->error_info));
        if (err)
                return -EFAULT;
        return err;
}
#endif

static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo)
{
        drive_info_struct *drv = get_drv(bdev->bd_disk);

        if (!drv->cylinders)
                return -ENXIO;

        geo->heads = drv->heads;
        geo->sectors = drv->sectors;
        geo->cylinders = drv->cylinders;
        return 0;
}

static void check_ioctl_unit_attention(ctlr_info_t *h, CommandList_struct *c)
{
        if (c->err_info->CommandStatus == CMD_TARGET_STATUS &&
                        c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION)
                (void)check_for_unit_attention(h, c);
}

static int cciss_getpciinfo(ctlr_info_t *h, void __user *argp)
{
        cciss_pci_info_struct pciinfo;

        if (!argp)
                return -EINVAL;
        pciinfo.domain = pci_domain_nr(h->pdev->bus);
        pciinfo.bus = h->pdev->bus->number;
        pciinfo.dev_fn = h->pdev->devfn;
        pciinfo.board_id = h->board_id;
        if (copy_to_user(argp, &pciinfo, sizeof(cciss_pci_info_struct)))
                return -EFAULT;
        return 0;
}

static int cciss_getintinfo(ctlr_info_t *h, void __user *argp)
{
        cciss_coalint_struct intinfo;

        if (!argp)
                return -EINVAL;
        intinfo.delay = readl(&h->cfgtable->HostWrite.CoalIntDelay);
        intinfo.count = readl(&h->cfgtable->HostWrite.CoalIntCount);
        if (copy_to_user
            (argp, &intinfo, sizeof(cciss_coalint_struct)))
                return -EFAULT;
        return 0;
}

static int cciss_setintinfo(ctlr_info_t *h, void __user *argp)
{
        cciss_coalint_struct intinfo;
        unsigned long flags;
        int i;

        if (!argp)
                return -EINVAL;
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
        if (copy_from_user(&intinfo, argp, sizeof(intinfo)))
                return -EFAULT;
        if ((intinfo.delay == 0) && (intinfo.count == 0))
                return -EINVAL;
        spin_lock_irqsave(&h->lock, flags);
        /* Update the field, and then ring the doorbell */
        writel(intinfo.delay, &(h->cfgtable->HostWrite.CoalIntDelay));
        writel(intinfo.count, &(h->cfgtable->HostWrite.CoalIntCount));
        writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL);

        for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) {
                if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq))
                        break;
                udelay(1000); /* delay and try again */
        }
        spin_unlock_irqrestore(&h->lock, flags);
        if (i >= MAX_IOCTL_CONFIG_WAIT)
                return -EAGAIN;
        return 0;
}

static int cciss_getnodename(ctlr_info_t *h, void __user *argp)
{
        NodeName_type NodeName;
        int i;

        if (!argp)
                return -EINVAL;
        for (i = 0; i < 16; i++)
                NodeName[i] = readb(&h->cfgtable->ServerName[i]);
        if (copy_to_user(argp, NodeName, sizeof(NodeName_type)))
                return -EFAULT;
        return 0;
}

static int cciss_setnodename(ctlr_info_t *h, void __user *argp)
{
        NodeName_type NodeName;
        unsigned long flags;
        int i;

        if (!argp)
                return -EINVAL;
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
        if (copy_from_user(NodeName, argp, sizeof(NodeName_type)))
                return -EFAULT;
        spin_lock_irqsave(&h->lock, flags);
        /* Update the field, and then ring the doorbell */
        for (i = 0; i < 16; i++)
                writeb(NodeName[i], &h->cfgtable->ServerName[i]);
        writel(CFGTBL_ChangeReq, h->vaddr + SA5_DOORBELL);
        for (i = 0; i < MAX_IOCTL_CONFIG_WAIT; i++) {
                if (!(readl(h->vaddr + SA5_DOORBELL) & CFGTBL_ChangeReq))
                        break;
                udelay(1000); /* delay and try again */
        }
        spin_unlock_irqrestore(&h->lock, flags);
        if (i >= MAX_IOCTL_CONFIG_WAIT)
                return -EAGAIN;
        return 0;
}

static int cciss_getheartbeat(ctlr_info_t *h, void __user *argp)
{
        Heartbeat_type heartbeat;

        if (!argp)
                return -EINVAL;
        heartbeat = readl(&h->cfgtable->HeartBeat);
        if (copy_to_user(argp, &heartbeat, sizeof(Heartbeat_type)))
                return -EFAULT;
        return 0;
}

static int cciss_getbustypes(ctlr_info_t *h, void __user *argp)
{
        BusTypes_type BusTypes;

        if (!argp)
                return -EINVAL;
        BusTypes = readl(&h->cfgtable->BusTypes);
        if (copy_to_user(argp, &BusTypes, sizeof(BusTypes_type)))
                return -EFAULT;
        return 0;
}

static int cciss_getfirmver(ctlr_info_t *h, void __user *argp)
{
        FirmwareVer_type firmware;

        if (!argp)
                return -EINVAL;
        memcpy(firmware, h->firm_ver, 4);

        if (copy_to_user
            (argp, firmware, sizeof(FirmwareVer_type)))
                return -EFAULT;
        return 0;
}

static int cciss_getdrivver(ctlr_info_t *h, void __user *argp)
{
        DriverVer_type DriverVer = DRIVER_VERSION;

        if (!argp)
                return -EINVAL;
        if (copy_to_user(argp, &DriverVer, sizeof(DriverVer_type)))
                return -EFAULT;
        return 0;
}

static int cciss_getluninfo(ctlr_info_t *h,
        struct gendisk *disk, void __user *argp)
{
        LogvolInfo_struct luninfo;
        drive_info_struct *drv = get_drv(disk);

        if (!argp)
                return -EINVAL;
        memcpy(&luninfo.LunID, drv->LunID, sizeof(luninfo.LunID));
        luninfo.num_opens = drv->usage_count;
        luninfo.num_parts = 0;
        if (copy_to_user(argp, &luninfo, sizeof(LogvolInfo_struct)))
                return -EFAULT;
        return 0;
}

static int cciss_passthru(ctlr_info_t *h, void __user *argp)
{
        IOCTL_Command_struct iocommand;
        CommandList_struct *c;
        char *buff = NULL;
        u64bit temp64;
        DECLARE_COMPLETION_ONSTACK(wait);

        if (!argp)
                return -EINVAL;

        if (!capable(CAP_SYS_RAWIO))
                return -EPERM;

        if (copy_from_user
            (&iocommand, argp, sizeof(IOCTL_Command_struct)))
                return -EFAULT;
        if ((iocommand.buf_size < 1) &&
            (iocommand.Request.Type.Direction != XFER_NONE)) {
                return -EINVAL;
        }
        if (iocommand.buf_size > 0) {
                buff = kmalloc(iocommand.buf_size, GFP_KERNEL);
                if (buff == NULL)
                        return -EFAULT;
        }
        if (iocommand.Request.Type.Direction == XFER_WRITE) {
                /* Copy the data into the buffer we created */
                if (copy_from_user(buff, iocommand.buf, iocommand.buf_size)) {
                        kfree(buff);
                        return -EFAULT;
                }
        } else {
                memset(buff, 0, iocommand.buf_size);
        }
        c = cmd_special_alloc(h);
        if (!c) {
                kfree(buff);
                return -ENOMEM;
        }
        /* Fill in the command type */
        c->cmd_type = CMD_IOCTL_PEND;
        /* Fill in Command Header */
        c->Header.ReplyQueue = 0;   /* unused in simple mode */
        if (iocommand.buf_size > 0) { /* buffer to fill */
                c->Header.SGList = 1;
                c->Header.SGTotal = 1;
        } else { /* no buffers to fill */
                c->Header.SGList = 0;
                c->Header.SGTotal = 0;
        }
        c->Header.LUN = iocommand.LUN_info;
        /* use the kernel address the cmd block for tag */
        c->Header.Tag.lower = c->busaddr;

        /* Fill in Request block */
        c->Request = iocommand.Request;

        /* Fill in the scatter gather information */
        if (iocommand.buf_size > 0) {
                temp64.val = pci_map_single(h->pdev, buff,
                        iocommand.buf_size, PCI_DMA_BIDIRECTIONAL);
                c->SG[0].Addr.lower = temp64.val32.lower;
                c->SG[0].Addr.upper = temp64.val32.upper;
                c->SG[0].Len = iocommand.buf_size;
                c->SG[0].Ext = 0;  /* we are not chaining */
        }
        c->waiting = &wait;

        enqueue_cmd_and_start_io(h, c);
        wait_for_completion(&wait);

        /* unlock the buffers from DMA */
        temp64.val32.lower = c->SG[0].Addr.lower;
        temp64.val32.upper = c->SG[0].Addr.upper;
        pci_unmap_single(h->pdev, (dma_addr_t) temp64.val, iocommand.buf_size,
                         PCI_DMA_BIDIRECTIONAL);
        check_ioctl_unit_attention(h, c);

        /* Copy the error information out */
        iocommand.error_info = *(c->err_info);
        if (copy_to_user(argp, &iocommand, sizeof(IOCTL_Command_struct))) {
                kfree(buff);
                cmd_special_free(h, c);
                return -EFAULT;
        }

        if (iocommand.Request.Type.Direction == XFER_READ) {
                /* Copy the data out of the buffer we created */
                if (copy_to_user(iocommand.buf, buff, iocommand.buf_size)) {
                        kfree(buff);
                        cmd_special_free(h, c);
                        return -EFAULT;
                }
        }
        kfree(buff);
        cmd_special_free(h, c);
        return 0;
}

static int cciss_bigpassthru(ctlr_info_t *h, void __user *argp)
{
        BIG_IOCTL_Command_struct *ioc;
        CommandList_struct *c;
        unsigned char **buff = NULL;
        int *buff_size = NULL;
        u64bit temp64;
        BYTE sg_used = 0;
        int status = 0;
        int i;
        DECLARE_COMPLETION_ONSTACK(wait);
        __u32 left;
        __u32 sz;
        BYTE __user *data_ptr;

        if (!argp)
                return -EINVAL;
        if (!capable(CAP_SYS_RAWIO))
                return -EPERM;
        ioc = (BIG_IOCTL_Command_struct *)
            kmalloc(sizeof(*ioc), GFP_KERNEL);
        if (!ioc) {
                status = -ENOMEM;
                goto cleanup1;
        }
        if (copy_from_user(ioc, argp, sizeof(*ioc))) {
                status = -EFAULT;
                goto cleanup1;
        }
        if ((ioc->buf_size < 1) &&
            (ioc->Request.Type.Direction != XFER_NONE)) {
                status = -EINVAL;
                goto cleanup1;
        }
        /* Check kmalloc limits  using all SGs */
        if (ioc->malloc_size > MAX_KMALLOC_SIZE) {
                status = -EINVAL;
                goto cleanup1;
        }
        if (ioc->buf_size > ioc->malloc_size * MAXSGENTRIES) {
                status = -EINVAL;
                goto cleanup1;
        }
        buff = kzalloc(MAXSGENTRIES * sizeof(char *), GFP_KERNEL);
        if (!buff) {
                status = -ENOMEM;
                goto cleanup1;
        }
        buff_size = kmalloc(MAXSGENTRIES * sizeof(int), GFP_KERNEL);
        if (!buff_size) {
                status = -ENOMEM;
                goto cleanup1;
        }
        left = ioc->buf_size;
        data_ptr = ioc->buf;
        while (left) {
                sz = (left > ioc->malloc_size) ? ioc->malloc_size : left;
                buff_size[sg_used] = sz;
                buff[sg_used] = kmalloc(sz, GFP_KERNEL);
                if (buff[sg_used] == NULL) {
                        status = -ENOMEM;
                        goto cleanup1;
                }
                if (ioc->Request.Type.Direction == XFER_WRITE) {
                        if (copy_from_user(buff[sg_used], data_ptr, sz)) {
                                status = -EFAULT;
                                goto cleanup1;
                        }
                } else {
                        memset(buff[sg_used], 0, sz);
                }
                left -= sz;
                data_ptr += sz;
                sg_used++;
        }
        c = cmd_special_alloc(h);
        if (!c) {
                status = -ENOMEM;
                goto cleanup1;
        }
        c->cmd_type = CMD_IOCTL_PEND;
        c->Header.ReplyQueue = 0;
        c->Header.SGList = sg_used;
        c->Header.SGTotal = sg_used;
        c->Header.LUN = ioc->LUN_info;
        c->Header.Tag.lower = c->busaddr;

        c->Request = ioc->Request;
        for (i = 0; i < sg_used; i++) {
                temp64.val = pci_map_single(h->pdev, buff[i], buff_size[i],
                                    PCI_DMA_BIDIRECTIONAL);
                c->SG[i].Addr.lower = temp64.val32.lower;
                c->SG[i].Addr.upper = temp64.val32.upper;
                c->SG[i].Len = buff_size[i];
                c->SG[i].Ext = 0;       /* we are not chaining */
        }
        c->waiting = &wait;
        enqueue_cmd_and_start_io(h, c);
        wait_for_completion(&wait);
        /* unlock the buffers from DMA */
        for (i = 0; i < sg_used; i++) {
                temp64.val32.lower = c->SG[i].Addr.lower;
                temp64.val32.upper = c->SG[i].Addr.upper;
                pci_unmap_single(h->pdev,
                        (dma_addr_t) temp64.val, buff_size[i],
                        PCI_DMA_BIDIRECTIONAL);
        }
        check_ioctl_unit_attention(h, c);
        /* Copy the error information out */
        ioc->error_info = *(c->err_info);
        if (copy_to_user(argp, ioc, sizeof(*ioc))) {
                cmd_special_free(h, c);
                status = -EFAULT;
                goto cleanup1;
        }
        if (ioc->Request.Type.Direction == XFER_READ) {
                /* Copy the data out of the buffer we created */
                BYTE __user *ptr = ioc->buf;
                for (i = 0; i < sg_used; i++) {
                        if (copy_to_user(ptr, buff[i], buff_size[i])) {
                                cmd_special_free(h, c);
                                status = -EFAULT;
                                goto cleanup1;
                        }
                        ptr += buff_size[i];
                }
        }
        cmd_special_free(h, c);
        status = 0;
cleanup1:
        if (buff) {
                for (i = 0; i < sg_used; i++)
                        kfree(buff[i]);
                kfree(buff);
        }
        kfree(buff_size);
        kfree(ioc);
        return status;
}

static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
        unsigned int cmd, unsigned long arg)
{
        struct gendisk *disk = bdev->bd_disk;
        ctlr_info_t *h = get_host(disk);
        void __user *argp = (void __user *)arg;

        dev_dbg(&h->pdev->dev, "cciss_ioctl: Called with cmd=%x %lx\n",
                cmd, arg);
        switch (cmd) {
        case CCISS_GETPCIINFO:
                return cciss_getpciinfo(h, argp);
        case CCISS_GETINTINFO:
                return cciss_getintinfo(h, argp);
        case CCISS_SETINTINFO:
                return cciss_setintinfo(h, argp);
        case CCISS_GETNODENAME:
                return cciss_getnodename(h, argp);
        case CCISS_SETNODENAME:
                return cciss_setnodename(h, argp);
        case CCISS_GETHEARTBEAT:
                return cciss_getheartbeat(h, argp);
        case CCISS_GETBUSTYPES:
                return cciss_getbustypes(h, argp);
        case CCISS_GETFIRMVER:
                return cciss_getfirmver(h, argp);
        case CCISS_GETDRIVVER:
                return cciss_getdrivver(h, argp);
        case CCISS_DEREGDISK:
        case CCISS_REGNEWD:
        case CCISS_REVALIDVOLS:
                return rebuild_lun_table(h, 0, 1);
        case CCISS_GETLUNINFO:
                return cciss_getluninfo(h, disk, argp);
        case CCISS_PASSTHRU:
                return cciss_passthru(h, argp);
        case CCISS_BIG_PASSTHRU:
                return cciss_bigpassthru(h, argp);

        /* scsi_cmd_ioctl handles these, below, though some are not */
        /* very meaningful for cciss.  SG_IO is the main one people want. */

        case SG_GET_VERSION_NUM:
        case SG_SET_TIMEOUT:
        case SG_GET_TIMEOUT:
        case SG_GET_RESERVED_SIZE:
        case SG_SET_RESERVED_SIZE:
        case SG_EMULATED_HOST:
        case SG_IO:
        case SCSI_IOCTL_SEND_COMMAND:
                return scsi_cmd_ioctl(disk->queue, disk, mode, cmd, argp);

        /* scsi_cmd_ioctl would normally handle these, below, but */
        /* they aren't a good fit for cciss, as CD-ROMs are */
        /* not supported, and we don't have any bus/target/lun */
        /* which we present to the kernel. */

        case CDROM_SEND_PACKET:
        case CDROMCLOSETRAY:
        case CDROMEJECT:
        case SCSI_IOCTL_GET_IDLUN:
        case SCSI_IOCTL_GET_BUS_NUMBER:
        default:
                return -ENOTTY;
        }
}

static void cciss_check_queues(ctlr_info_t *h)
{
        int start_queue = h->next_to_run;
        int i;

        /* check to see if we have maxed out the number of commands that can
         * be placed on the queue.  If so then exit.  We do this check here
         * in case the interrupt we serviced was from an ioctl and did not
         * free any new commands.
         */
        if ((find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds)) == h->nr_cmds)
                return;

        /* We have room on the queue for more commands.  Now we need to queue
         * them up.  We will also keep track of the next queue to run so
         * that every queue gets a chance to be started first.
         */
        for (i = 0; i < h->highest_lun + 1; i++) {
                int curr_queue = (start_queue + i) % (h->highest_lun + 1);
                /* make sure the disk has been added and the drive is real
                 * because this can be called from the middle of init_one.
                 */
                if (!h->drv[curr_queue])
                        continue;
                if (!(h->drv[curr_queue]->queue) ||
                        !(h->drv[curr_queue]->heads))
                        continue;
                blk_start_queue(h->gendisk[curr_queue]->queue);

                /* check to see if we have maxed out the number of commands
                 * that can be placed on the queue.
                 */
                if ((find_first_zero_bit(h->cmd_pool_bits, h->nr_cmds)) == h->nr_cmds) {
                        if (curr_queue == start_queue) {
                                h->next_to_run =
                                    (start_queue + 1) % (h->highest_lun + 1);
                                break;
                        } else {
                                h->next_to_run = curr_queue;
                                break;
                        }
                }
        }
}

static void cciss_softirq_done(struct request *rq)
{
        CommandList_struct *c = rq->completion_data;
        ctlr_info_t *h = hba[c->ctlr];
        SGDescriptor_struct *curr_sg = c->SG;
        u64bit temp64;
        unsigned long flags;
        int i, ddir;
        int sg_index = 0;

        if (c->Request.Type.Direction == XFER_READ)
                ddir = PCI_DMA_FROMDEVICE;
        else
                ddir = PCI_DMA_TODEVICE;

        /* command did not need to be retried */
        /* unmap the DMA mapping for all the scatter gather elements */
        for (i = 0; i < c->Header.SGList; i++) {
                if (curr_sg[sg_index].Ext == CCISS_SG_CHAIN) {
                        cciss_unmap_sg_chain_block(h, c);
                        /* Point to the next block */
                        curr_sg = h->cmd_sg_list[c->cmdindex];
                        sg_index = 0;
                }
                temp64.val32.lower = curr_sg[sg_index].Addr.lower;
                temp64.val32.upper = curr_sg[sg_index].Addr.upper;
                pci_unmap_page(h->pdev, temp64.val, curr_sg[sg_index].Len,
                                ddir);
                ++sg_index;
        }

        dev_dbg(&h->pdev->dev, "Done with %p\n", rq);

        /* set the residual count for pc requests */
        if (rq->cmd_type == REQ_TYPE_BLOCK_PC)
                rq->resid_len = c->err_info->ResidualCnt;

        blk_end_request_all(rq, (rq->errors == 0) ? 0 : -EIO);

        spin_lock_irqsave(&h->lock, flags);
        cmd_free(h, c);
        cciss_check_queues(h);
        spin_unlock_irqrestore(&h->lock, flags);
}

static inline void log_unit_to_scsi3addr(ctlr_info_t *h,
        unsigned char scsi3addr[], uint32_t log_unit)
{
        memcpy(scsi3addr, h->drv[log_unit]->LunID,
                sizeof(h->drv[log_unit]->LunID));
}

/* This function gets the SCSI vendor, model, and revision of a logical drive
 * via the inquiry page 0.  Model, vendor, and rev are set to empty strings if
 * they cannot be read.
 */
static void cciss_get_device_descr(ctlr_info_t *h, int logvol,
                                   char *vendor, char *model, char *rev)
{
        int rc;
        InquiryData_struct *inq_buf;
        unsigned char scsi3addr[8];

        *vendor = '\0';
        *model = '\0';
        *rev = '\0';

        inq_buf = kzalloc(sizeof(InquiryData_struct), GFP_KERNEL);
        if (!inq_buf)
                return;

        log_unit_to_scsi3addr(h, scsi3addr, logvol);
        rc = sendcmd_withirq(h, CISS_INQUIRY, inq_buf, sizeof(*inq_buf), 0,
                        scsi3addr, TYPE_CMD);
        if (rc == IO_OK) {
                memcpy(vendor, &inq_buf->data_byte[8], VENDOR_LEN);
                vendor[VENDOR_LEN] = '\0';
                memcpy(model, &inq_buf->data_byte[16], MODEL_LEN);
                model[MODEL_LEN] = '\0';
                memcpy(rev, &inq_buf->data_byte[32], REV_LEN);
                rev[REV_LEN] = '\0';
        }

        kfree(inq_buf);
        return;
}

/* This function gets the serial number of a logical drive via
 * inquiry page 0x83.  Serial no. is 16 bytes.  If the serial
 * number cannot be had, for whatever reason, 16 bytes of 0xff
 * are returned instead.
 */
static void cciss_get_serial_no(ctlr_info_t *h, int logvol,
                                unsigned char *serial_no, int buflen)
{
#define PAGE_83_INQ_BYTES 64
        int rc;
        unsigned char *buf;
        unsigned char scsi3addr[8];

        if (buflen > 16)
                buflen = 16;
        memset(serial_no, 0xff, buflen);
        buf = kzalloc(PAGE_83_INQ_BYTES, GFP_KERNEL);
        if (!buf)
                return;
        memset(serial_no, 0, buflen);
        log_unit_to_scsi3addr(h, scsi3addr, logvol);
        rc = sendcmd_withirq(h, CISS_INQUIRY, buf,
                PAGE_83_INQ_BYTES, 0x83, scsi3addr, TYPE_CMD);
        if (rc == IO_OK)
                memcpy(serial_no, &buf[8], buflen);
        kfree(buf);
        return;
}

/*
 * cciss_add_disk sets up the block device queue for a logical drive
 */
static int cciss_add_disk(ctlr_info_t *h, struct gendisk *disk,
                                int drv_index)
{
        disk->queue = blk_init_queue(do_cciss_request, &h->lock);
        if (!disk->queue)
                goto init_queue_failure;
        sprintf(disk->disk_name, "cciss/c%dd%d", h->ctlr, drv_index);
        disk->major = h->major;
        disk->first_minor = drv_index << NWD_SHIFT;
        disk->fops = &cciss_fops;
        if (cciss_create_ld_sysfs_entry(h, drv_index))
                goto cleanup_queue;
        disk->private_data = h->drv[drv_index];
        disk->driverfs_dev = &h->drv[drv_index]->dev;

        /* Set up queue information */
        blk_queue_bounce_limit(disk->queue, h->pdev->dma_mask);

        /* This is a hardware imposed limit. */
        blk_queue_max_segments(disk->queue, h->maxsgentries);

        blk_queue_max_hw_sectors(disk->queue, h->cciss_max_sectors);

        blk_queue_softirq_done(disk->queue, cciss_softirq_done);

        disk->queue->queuedata = h;

        blk_queue_logical_block_size(disk->queue,
                                     h->drv[drv_index]->block_size);

        /* Make sure all queue data is written out before */
        /* setting h->drv[drv_index]->queue, as setting this */
        /* allows the interrupt handler to start the queue */
        wmb();
        h->drv[drv_index]->queue = disk->queue;
        add_disk(disk);
        return 0;

cleanup_queue:
        blk_cleanup_queue(disk->queue);
        disk->queue = NULL;
init_queue_failure:
        return -1;
}

/* This function will check the usage_count of the drive to be updated/added.
 * If the usage_count is zero and it is a heretofore unknown drive, or,
 * the drive's capacity, geometry, or serial number has changed,
 * then the drive information will be updated and the disk will be
 * re-registered with the kernel.  If these conditions don't hold,
 * then it will be left alone for the next reboot.  The exception to this
 * is disk 0 which will always be left registered with the kernel since it
 * is also the controller node.  Any changes to disk 0 will show up on
 * the next reboot.
 */
static void cciss_update_drive_info(ctlr_info_t *h, int drv_index,
        int first_time, int via_ioctl)
{
        struct gendisk *disk;
        InquiryData_struct *inq_buff = NULL;
        unsigned int block_size;
        sector_t total_size;
        unsigned long flags = 0;
        int ret = 0;
        drive_info_struct *drvinfo;

        /* Get information about the disk and modify the driver structure */
        inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL);
        drvinfo = kzalloc(sizeof(*drvinfo), GFP_KERNEL);
        if (inq_buff == NULL || drvinfo == NULL)
                goto mem_msg;

        /* testing to see if 16-byte CDBs are already being used */
        if (h->cciss_read == CCISS_READ_16) {
                cciss_read_capacity_16(h, drv_index,
                        &total_size, &block_size);

        } else {
                cciss_read_capacity(h, drv_index, &total_size, &block_size);
                /* if read_capacity returns all F's this volume is >2TB */
                /* in size so we switch to 16-byte CDB's for all */
                /* read/write ops */
                if (total_size == 0xFFFFFFFFULL) {
                        cciss_read_capacity_16(h, drv_index,
                        &total_size, &block_size);
                        h->cciss_read = CCISS_READ_16;
                        h->cciss_write = CCISS_WRITE_16;
                } else {
                        h->cciss_read = CCISS_READ_10;
                        h->cciss_write = CCISS_WRITE_10;
                }
        }

        cciss_geometry_inquiry(h, drv_index, total_size, block_size,
                               inq_buff, drvinfo);
        drvinfo->block_size = block_size;
        drvinfo->nr_blocks = total_size + 1;

        cciss_get_device_descr(h, drv_index, drvinfo->vendor,
                                drvinfo->model, drvinfo->rev);
        cciss_get_serial_no(h, drv_index, drvinfo->serial_no,
                        sizeof(drvinfo->serial_no));
        /* Save the lunid in case we deregister the disk, below. */
        memcpy(drvinfo->LunID, h->drv[drv_index]->LunID,
                sizeof(drvinfo->LunID));

        /* Is it the same disk we already know, and nothing's changed? */
        if (h->drv[drv_index]->raid_level != -1 &&
                ((memcmp(drvinfo->serial_no,
                                h->drv[drv_index]->serial_no, 16) == 0) &&
                drvinfo->block_size == h->drv[drv_index]->block_size &&
                drvinfo->nr_blocks == h->drv[drv_index]->nr_blocks &&
                drvinfo->heads == h->drv[drv_index]->heads &&
                drvinfo->sectors == h->drv[drv_index]->sectors &&
                drvinfo->cylinders == h->drv[drv_index]->cylinders))
                        /* The disk is unchanged, nothing to update */
                        goto freeret;

        /* If we get here it's not the same disk, or something's changed,
         * so we need to * deregister it, and re-register it, if it's not
         * in use.
         * If the disk already exists then deregister it before proceeding
         * (unless it's the first disk (for the controller node).
         */
        if (h->drv[drv_index]->raid_level != -1 && drv_index != 0) {
                dev_warn(&h->pdev->dev, "disk %d has changed.\n", drv_index);
                spin_lock_irqsave(&h->lock, flags);
                h->drv[drv_index]->busy_configuring = 1;
                spin_unlock_irqrestore(&h->lock, flags);

                /* deregister_disk sets h->drv[drv_index]->queue = NULL
                 * which keeps the interrupt handler from starting
                 * the queue.
                 */
                ret = deregister_disk(h, drv_index, 0, via_ioctl);
        }

        /* If the disk is in use return */
        if (ret)
                goto freeret;

        /* Save the new information from cciss_geometry_inquiry
         * and serial number inquiry.  If the disk was deregistered
         * above, then h->drv[drv_index] will be NULL.
         */
        if (h->drv[drv_index] == NULL) {
                drvinfo->device_initialized = 0;
                h->drv[drv_index] = drvinfo;
                drvinfo = NULL; /* so it won't be freed below. */
        } else {
                /* special case for cxd0 */
                h->drv[drv_index]->block_size = drvinfo->block_size;
                h->drv[drv_index]->nr_blocks = drvinfo->nr_blocks;
                h->drv[drv_index]->heads = drvinfo->heads;
                h->drv[drv_index]->sectors = drvinfo->sectors;
                h->drv[drv_index]->cylinders = drvinfo->cylinders;

                h->drv[drv_index]->raid_level = drvinfo->raid_level;
                memcpy(h->drv[drv_index]->serial_no, drvinfo->serial_no, 16);
                memcpy(h->drv[drv_index]->vendor, drvinfo->vendor,
                        VENDOR_LEN + 1);
                memcpy(h->drv[drv_index]->model, drvinfo->model, MODEL_LEN + 1);
                memcpy(h->drv[drv_index]->rev, drvinfo->rev, REV_LEN + 1);
        }

        ++h->num_luns;
        disk = h->gendisk[drv_index];
        set_capacity(disk, h->drv[drv_index]->nr_blocks);

        /* If it's not disk 0 (drv_index != 0)
         * or if it was disk 0, but there was previously
         * no actual corresponding configured logical drive
         * (raid_leve == -1) then we want to update the
         * logical drive's information.
         */
        if (drv_index || first_time) {
                if (cciss_add_disk(h, disk, drv_index) != 0) {
                        cciss_free_gendisk(h, drv_index);
                        cciss_free_drive_info(h, drv_index);
                        dev_warn(&h->pdev->dev, "could not update disk %d\n",
                                drv_index);
                        --h->num_luns;
                }
        }

freeret:
        kfree(inq_buff);
        kfree(drvinfo);
        return;
mem_msg:
        dev_err(&h->pdev->dev, "out of memory\n");
        goto freeret;
}

/* This function will find the first index of the controllers drive array
 * that has a null drv pointer and allocate the drive info struct and
 * will return that index   This is where new drives will be added.
 * If the index to be returned is greater than the highest_lun index for
 * the controller then highest_lun is set * to this new index.
 * If there are no available indexes or if tha allocation fails, then -1
 * is returned.  * "controller_node" is used to know if this is a real
 * logical drive, or just the controller node, which determines if this
 * counts towards highest_lun.
 */
static int cciss_alloc_drive_info(ctlr_info_t *h, int controller_node)
{
        int i;
        drive_info_struct *drv;

        /* Search for an empty slot for our drive info */
        for (i = 0; i < CISS_MAX_LUN; i++) {

                /* if not cxd0 case, and it's occupied, skip it. */
                if (h->drv[i] && i != 0)
                        continue;
                /*
                 * If it's cxd0 case, and drv is alloc'ed already, and a
                 * disk is configured there, skip it.
                 */
                if (i == 0 && h->drv[i] && h->drv[i]->raid_level != -1)
                        continue;

                /*
                 * We've found an empty slot.  Update highest_lun
                 * provided this isn't just the fake cxd0 controller node.
                 */
                if (i > h->highest_lun && !controller_node)
                        h->highest_lun = i;

                /* If adding a real disk at cxd0, and it's already alloc'ed */
                if (i == 0 && h->drv[i] != NULL)
                        return i;

                /*
                 * Found an empty slot, not already alloc'ed.  Allocate it.
                 * Mark it with raid_level == -1, so we know it's new later on.
                 */
                drv = kzalloc(sizeof(*drv), GFP_KERNEL);
                if (!drv)
                        return -1;
                drv->raid_level = -1; /* so we know it's new */
                h->drv[i] = drv;
                return i;
        }
        return -1;
}

static void cciss_free_drive_info(ctlr_info_t *h, int drv_index)
{
        kfree(h->drv[drv_index]);
        h->drv[drv_index] = NULL;
}

static void cciss_free_gendisk(ctlr_info_t *h, int drv_index)
{
        put_disk(h->gendisk[drv_index]);
        h->gendisk[drv_index] = NULL;
}

/* cciss_add_gendisk finds a free hba[]->drv structure
 * and allocates a gendisk if needed, and sets the lunid
 * in the drvinfo structure.   It returns the index into
 * the ->drv[] array, or -1 if none are free.
 * is_controller_node indicates whether highest_lun should
 * count this disk, or if it's only being added to provide
 * a means to talk to the controller in case no logical
 * drives have yet been configured.
 */
static int cciss_add_gendisk(ctlr_info_t *h, unsigned char lunid[],
        int controller_node)
{
        int drv_index;

        drv_index = cciss_alloc_drive_info(h, controller_node);
        if (drv_index == -1)
                return -1;

        /*Check if the gendisk needs to be allocated */
        if (!h->gendisk[drv_index]) {
                h->gendisk[drv_index] =
                        alloc_disk(1 << NWD_SHIFT);
                if (!h->gendisk[drv_index]) {
                        dev_err(&h->pdev->dev,
                                "could not allocate a new disk %d\n",
                                drv_index);
                        goto err_free_drive_info;
                }
        }
        memcpy(h->drv[drv_index]->LunID, lunid,
                sizeof(h->drv[drv_index]->LunID));
        if (cciss_create_ld_sysfs_entry(h, drv_index))
                goto err_free_disk;
        /* Don't need to mark this busy because nobody */
        /* else knows about this disk yet to contend */
        /* for access to it. */
        h->drv[drv_index]->busy_configuring = 0;
        wmb();
        return drv_index;

err_free_disk:
        cciss_free_gendisk(h, drv_index);
err_free_drive_info:
        cciss_free_drive_info(h, drv_index);
        return -1;
}

/* This is for the special case of a controller which
 * has no logical drives.  In this case, we still need
 * to register a disk so the controller can be accessed
 * by the Array Config Utility.
 */
static void cciss_add_controller_node(ctlr_info_t *h)
{
        struct gendisk *disk;
        int drv_index;

        if (h->gendisk[0] != NULL) /* already did this? Then bail. */
                return;

        drv_index = cciss_add_gendisk(h, CTLR_LUNID, 1);
        if (drv_index == -1)
                goto error;
        h->drv[drv_index]->block_size = 512;
        h->drv[drv_index]->nr_blocks = 0;
        h->drv[drv_index]->heads = 0;
        h->drv[drv_index]->sectors = 0;
        h->drv[drv_index]->cylinders = 0;
        h->drv[drv_index]->raid_level = -1;
        memset(h->drv[drv_index]->serial_no, 0, 16);
        disk = h->gendisk[drv_index];
        if (cciss_add_disk(h, disk, drv_index) == 0)
                return;
        cciss_free_gendisk(h, drv_index);
        cciss_free_drive_info(h, drv_index);
error:
        dev_warn(&h->pdev->dev, "could not add disk 0.\n");
        return;
}

/* This function will add and remove logical drives from the Logical
 * drive array of the controller and maintain persistency of ordering
 * so that mount points are preserved until the next reboot.  This allows
 * for the removal of logical drives in the middle of the drive array
 * without a re-ordering of those drives.
 * INPUT
 * h            = The controller to perform the operations on
 */
static int rebuild_lun_table(ctlr_info_t *h, int first_time,
        int via_ioctl)
{
        int num_luns;
        ReportLunData_struct *ld_buff = NULL;
        int return_code;
        int listlength = 0;
        int i;
        int drv_found;
        int drv_index = 0;
        unsigned char lunid[8] = CTLR_LUNID;
        unsigned long flags;

        if (!capable(CAP_SYS_RAWIO))
                return -EPERM;

        /* Set busy_configuring flag for this operation */
        spin_lock_irqsave(&h->lock, flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(&h->lock, flags);
                return -EBUSY;
        }
        h->busy_configuring = 1;
        spin_unlock_irqrestore(&h->lock, flags);

        ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL);
        if (ld_buff == NULL)
                goto mem_msg;

        return_code = sendcmd_withirq(h, CISS_REPORT_LOG, ld_buff,
                                      sizeof(ReportLunData_struct),
                                      0, CTLR_LUNID, TYPE_CMD);

        if (return_code == IO_OK)
                listlength = be32_to_cpu(*(__be32 *) ld_buff->LUNListLength);
        else {  /* reading number of logical volumes failed */
                dev_warn(&h->pdev->dev,
                        "report logical volume command failed\n");
                listlength = 0;
                goto freeret;
        }

        num_luns = listlength / 8;      /* 8 bytes per entry */
        if (num_luns > CISS_MAX_LUN) {
                num_luns = CISS_MAX_LUN;
                dev_warn(&h->pdev->dev, "more luns configured"
                       " on controller than can be handled by"
                       " this driver.\n");
        }

        if (num_luns == 0)
                cciss_add_controller_node(h);

        /* Compare controller drive array to driver's drive array
         * to see if any drives are missing on the controller due
         * to action of Array Config Utility (user deletes drive)
         * and deregister logical drives which have disappeared.
         */
        for (i = 0; i <= h->highest_lun; i++) {
                int j;
                drv_found = 0;

                /* skip holes in the array from already deleted drives */
                if (h->drv[i] == NULL)
                        continue;

                for (j = 0; j < num_luns; j++) {
                        memcpy(lunid, &ld_buff->LUN[j][0], sizeof(lunid));
                        if (memcmp(h->drv[i]->LunID, lunid,
                                sizeof(lunid)) == 0) {
                                drv_found = 1;
                                break;
                        }
                }
                if (!drv_found) {
                        /* Deregister it from the OS, it's gone. */
                        spin_lock_irqsave(&h->lock, flags);
                        h->drv[i]->busy_configuring = 1;
                        spin_unlock_irqrestore(&h->lock, flags);
                        return_code = deregister_disk(h, i, 1, via_ioctl);
                        if (h->drv[i] != NULL)
                                h->drv[i]->busy_configuring = 0;
                }
        }

        /* Compare controller drive array to driver's drive array.
         * Check for updates in the drive information and any new drives
         * on the controller due to ACU adding logical drives, or changing
         * a logical drive's size, etc.  Reregister any new/changed drives
         */
        for (i = 0; i < num_luns; i++) {
                int j;

                drv_found = 0;

                memcpy(lunid, &ld_buff->LUN[i][0], sizeof(lunid));
                /* Find if the LUN is already in the drive array
                 * of the driver.  If so then update its info
                 * if not in use.  If it does not exist then find
                 * the first free index and add it.
                 */
                for (j = 0; j <= h->highest_lun; j++) {
                        if (h->drv[j] != NULL &&
                                memcmp(h->drv[j]->LunID, lunid,
                                        sizeof(h->drv[j]->LunID)) == 0) {
                                drv_index = j;
                                drv_found = 1;
                                break;
                        }
                }

                /* check if the drive was found already in the array */
                if (!drv_found) {
                        drv_index = cciss_add_gendisk(h, lunid, 0);
                        if (drv_index == -1)
                                goto freeret;
                }
                cciss_update_drive_info(h, drv_index, first_time, via_ioctl);
        }               /* end for */

freeret:
        kfree(ld_buff);
        h->busy_configuring = 0;
        /* We return -1 here to tell the ACU that we have registered/updated
         * all of the drives that we can and to keep it from calling us
         * additional times.
         */
        return -1;
mem_msg:
        dev_err(&h->pdev->dev, "out of memory\n");
        h->busy_configuring = 0;
        goto freeret;
}

static void cciss_clear_drive_info(drive_info_struct *drive_info)
{
        /* zero out the disk size info */
        drive_info->nr_blocks = 0;
        drive_info->block_size = 0;
        drive_info->heads = 0;
        drive_info->sectors = 0;
        drive_info->cylinders = 0;
        drive_info->raid_level = -1;
        memset(drive_info->serial_no, 0, sizeof(drive_info->serial_no));
        memset(drive_info->model, 0, sizeof(drive_info->model));
        memset(drive_info->rev, 0, sizeof(drive_info->rev));
        memset(drive_info->vendor, 0, sizeof(drive_info->vendor));
        /*
         * don't clear the LUNID though, we need to remember which
         * one this one is.
         */
}

/* This function will deregister the disk and it's queue from the
 * kernel.  It must be called with the controller lock held and the
 * drv structures busy_configuring flag set.  It's parameters are:
 *
 * disk = This is the disk to be deregistered
 * drv  = This is the drive_info_struct associated with the disk to be
 *        deregistered.  It contains information about the disk used
 *        by the driver.
 * clear_all = This flag determines whether or not the disk information
 *             is going to be completely cleared out and the highest_lun
 *             reset.  Sometimes we want to clear out information about
 *             the disk in preparation for re-adding it.  In this case
 *             the highest_lun should be left unchanged and the LunID
 *             should not be cleared.
 * via_ioctl
 *    This indicates whether we've reached this path via ioctl.
 *    This affects the maximum usage count allowed for c0d0 to be messed with.
 *    If this path is reached via ioctl(), then the max_usage_count will
 *    be 1, as the process calling ioctl() has got to have the device open.
 *    If we get here via sysfs, then the max usage count will be zero.
*/
static int deregister_disk(ctlr_info_t *h, int drv_index,
                           int clear_all, int via_ioctl)
{
        int i;
        struct gendisk *disk;
        drive_info_struct *drv;
        int recalculate_highest_lun;

        if (!capable(CAP_SYS_RAWIO))
                return -EPERM;

        drv = h->drv[drv_index];
        disk = h->gendisk[drv_index];

        /* make sure logical volume is NOT is use */
        if (clear_all || (h->gendisk[0] == disk)) {
                if (drv->usage_count > via_ioctl)
                        return -EBUSY;
        } else if (drv->usage_count > 0)
                return -EBUSY;

        recalculate_highest_lun = (drv == h->drv[h->highest_lun]);

        /* invalidate the devices and deregister the disk.  If it is disk
         * zero do not deregister it but just zero out it's values.  This
         * allows us to delete disk zero but keep the controller registered.
         */
        if (h->gendisk[0] != disk) {
                struct request_queue *q = disk->queue;
                if (disk->flags & GENHD_FL_UP) {
                        cciss_destroy_ld_sysfs_entry(h, drv_index, 0);
                        del_gendisk(disk);
                }
                if (q)
                        blk_cleanup_queue(q);
                /* If clear_all is set then we are deleting the logical
                 * drive, not just refreshing its info.  For drives
                 * other than disk 0 we will call put_disk.  We do not
                 * do this for disk 0 as we need it to be able to
                 * configure the controller.
                 */
                if (clear_all){
                        /* This isn't pretty, but we need to find the
                         * disk in our array and NULL our the pointer.
                         * This is so that we will call alloc_disk if
                         * this index is used again later.
                         */
                        for (i=0; i < CISS_MAX_LUN; i++){
                                if (h->gendisk[i] == disk) {
                                        h->gendisk[i] = NULL;
                                        break;
                                }
                        }
                        put_disk(disk);
                }
        } else {
                set_capacity(disk, 0);
                cciss_clear_drive_info(drv);
        }

        --h->num_luns;

        /* if it was the last disk, find the new hightest lun */
        if (clear_all && recalculate_highest_lun) {
                int newhighest = -1;
                for (i = 0; i <= h->highest_lun; i++) {
                        /* if the disk has size > 0, it is available */
                        if (h->drv[i] && h->drv[i]->heads)
                                newhighest = i;
                }
                h->highest_lun = newhighest;
        }
        return 0;
}

static int fill_cmd(ctlr_info_t *h, CommandList_struct *c, __u8 cmd, void *buff,
                size_t size, __u8 page_code, unsigned char *scsi3addr,
                int cmd_type)
{
        u64bit buff_dma_handle;
        int status = IO_OK;

        c->cmd_type = CMD_IOCTL_PEND;
        c->Header.ReplyQueue = 0;
        if (buff != NULL) {
                c->Header.SGList = 1;
                c->Header.SGTotal = 1;
        } else {
                c->Header.SGList = 0;
                c->Header.SGTotal = 0;
        }
        c->Header.Tag.lower = c->busaddr;
        memcpy(c->Header.LUN.LunAddrBytes, scsi3addr, 8);

        c->Request.Type.Type = cmd_type;
        if (cmd_type == TYPE_CMD) {
                switch (cmd) {
                case CISS_INQUIRY:
                        /* are we trying to read a vital product page */
                        if (page_code != 0) {
                                c->Request.CDB[1] = 0x01;
                                c->Request.CDB[2] = page_code;
                        }
                        c->Request.CDBLen = 6;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = CISS_INQUIRY;
                        c->Request.CDB[4] = size & 0xFF;
                        break;
                case CISS_REPORT_LOG:
                case CISS_REPORT_PHYS:
                        /* Talking to controller so It's a physical command
                           mode = 00 target = 0.  Nothing to write.
                         */
                        c->Request.CDBLen = 12;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;
                        c->Request.CDB[6] = (size >> 24) & 0xFF; /* MSB */
                        c->Request.CDB[7] = (size >> 16) & 0xFF;
                        c->Request.CDB[8] = (size >> 8) & 0xFF;
                        c->Request.CDB[9] = size & 0xFF;
                        break;

                case CCISS_READ_CAPACITY:
                        c->Request.CDBLen = 10;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;
                        break;
                case CCISS_READ_CAPACITY_16:
                        c->Request.CDBLen = 16;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;
                        c->Request.CDB[1] = 0x10;
                        c->Request.CDB[10] = (size >> 24) & 0xFF;
                        c->Request.CDB[11] = (size >> 16) & 0xFF;
                        c->Request.CDB[12] = (size >> 8) & 0xFF;
                        c->Request.CDB[13] = size & 0xFF;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;
                        break;
                case CCISS_CACHE_FLUSH:
                        c->Request.CDBLen = 12;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_WRITE;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = BMIC_WRITE;
                        c->Request.CDB[6] = BMIC_CACHE_FLUSH;
                        break;
                case TEST_UNIT_READY:
                        c->Request.CDBLen = 6;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_NONE;
                        c->Request.Timeout = 0;
                        break;
                default:
                        dev_warn(&h->pdev->dev, "Unknown Command 0x%c\n", cmd);
                        return IO_ERROR;
                }
        } else if (cmd_type == TYPE_MSG) {
                switch (cmd) {
                case 0: /* ABORT message */
                        c->Request.CDBLen = 12;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_WRITE;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;        /* abort */
                        c->Request.CDB[1] = 0;  /* abort a command */
                        /* buff contains the tag of the command to abort */
                        memcpy(&c->Request.CDB[4], buff, 8);
                        break;
                case 1: /* RESET message */
                        c->Request.CDBLen = 16;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_NONE;
                        c->Request.Timeout = 0;
                        memset(&c->Request.CDB[0], 0, sizeof(c->Request.CDB));
                        c->Request.CDB[0] = cmd;        /* reset */
                        c->Request.CDB[1] = 0x03;       /* reset a target */
                        break;
                case 3: /* No-Op message */
                        c->Request.CDBLen = 1;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_WRITE;
                        c->Request.Timeout = 0;
                        c->Request.CDB[0] = cmd;
                        break;
                default:
                        dev_warn(&h->pdev->dev,
                                "unknown message type %d\n", cmd);
                        return IO_ERROR;
                }
        } else {
                dev_warn(&h->pdev->dev, "unknown command type %d\n", cmd_type);
                return IO_ERROR;
        }
        /* Fill in the scatter gather information */
        if (size > 0) {
                buff_dma_handle.val = (__u64) pci_map_single(h->pdev,
                                                             buff, size,
                                                             PCI_DMA_BIDIRECTIONAL);
                c->SG[0].Addr.lower = buff_dma_handle.val32.lower;
                c->SG[0].Addr.upper = buff_dma_handle.val32.upper;
                c->SG[0].Len = size;
                c->SG[0].Ext = 0;       /* we are not chaining */
        }
        return status;
}

static int check_target_status(ctlr_info_t *h, CommandList_struct *c)
{
        switch (c->err_info->ScsiStatus) {
        case SAM_STAT_GOOD:
                return IO_OK;
        case SAM_STAT_CHECK_CONDITION:
                switch (0xf & c->err_info->SenseInfo[2]) {
                case 0: return IO_OK; /* no sense */
                case 1: return IO_OK; /* recovered error */
                default:
                        if (check_for_unit_attention(h, c))
                                return IO_NEEDS_RETRY;
                        dev_warn(&h->pdev->dev, "cmd 0x%02x "
                                "check condition, sense key = 0x%02x\n",
                                c->Request.CDB[0], c->err_info->SenseInfo[2]);
                }
                break;
        default:
                dev_warn(&h->pdev->dev, "cmd 0x%02x"
                        "scsi status = 0x%02x\n",
                        c->Request.CDB[0], c->err_info->ScsiStatus);
                break;
        }
        return IO_ERROR;
}

static int process_sendcmd_error(ctlr_info_t *h, CommandList_struct *c)
{
        int return_status = IO_OK;

        if (c->err_info->CommandStatus == CMD_SUCCESS)
                return IO_OK;

        switch (c->err_info->CommandStatus) {
        case CMD_TARGET_STATUS:
                return_status = check_target_status(h, c);
                break;
        case CMD_DATA_UNDERRUN:
        case CMD_DATA_OVERRUN:
                /* expected for inquiry and report lun commands */
                break;
        case CMD_INVALID:
                dev_warn(&h->pdev->dev, "cmd 0x%02x is "
                       "reported invalid\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_PROTOCOL_ERR:
                dev_warn(&h->pdev->dev, "cmd 0x%02x has "
                       "protocol error\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_HARDWARE_ERR:
                dev_warn(&h->pdev->dev, "cmd 0x%02x had "
                       " hardware error\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_CONNECTION_LOST:
                dev_warn(&h->pdev->dev, "cmd 0x%02x had "
                       "connection lost\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_ABORTED:
                dev_warn(&h->pdev->dev, "cmd 0x%02x was "
                       "aborted\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_ABORT_FAILED:
                dev_warn(&h->pdev->dev, "cmd 0x%02x reports "
                       "abort failed\n", c->Request.CDB[0]);
                return_status = IO_ERROR;
                break;
        case CMD_UNSOLICITED_ABORT:
                dev_warn(&h->pdev->dev, "unsolicited abort 0x%02x\n",
                        c->Request.CDB[0]);
                return_status = IO_NEEDS_RETRY;
                break;
        default:
                dev_warn(&h->pdev->dev, "cmd 0x%02x returned "
                       "unknown status %x\n", c->Request.CDB[0],
                       c->err_info->CommandStatus);
                return_status = IO_ERROR;
        }
        return return_status;
}

static int sendcmd_withirq_core(ctlr_info_t *h, CommandList_struct *c,
        int attempt_retry)
{
        DECLARE_COMPLETION_ONSTACK(wait);
        u64bit buff_dma_handle;
        int return_status = IO_OK;

resend_cmd2:
        c->waiting = &wait;
        enqueue_cmd_and_start_io(h, c);

        wait_for_completion(&wait);

        if (c->err_info->CommandStatus == 0 || !attempt_retry)
                goto command_done;

        return_status = process_sendcmd_error(h, c);

        if (return_status == IO_NEEDS_RETRY &&
                c->retry_count < MAX_CMD_RETRIES) {
                dev_warn(&h->pdev->dev, "retrying 0x%02x\n",
                        c->Request.CDB[0]);
                c->retry_count++;
                /* erase the old error information */
                memset(c->err_info, 0, sizeof(ErrorInfo_struct));
                return_status = IO_OK;
                INIT_COMPLETION(wait);
                goto resend_cmd2;
        }

command_done:
        /* unlock the buffers from DMA */
        buff_dma_handle.val32.lower = c->SG[0].Addr.lower;
        buff_dma_handle.val32.upper = c->SG[0].Addr.upper;
        pci_unmap_single(h->pdev, (dma_addr_t) buff_dma_handle.val,
                         c->SG[0].Len, PCI_DMA_BIDIRECTIONAL);
        return return_status;
}

static int sendcmd_withirq(ctlr_info_t *h, __u8 cmd, void *buff, size_t size,
                           __u8 page_code, unsigned char scsi3addr[],
                        int cmd_type)
{
        CommandList_struct *c;
        int return_status;

        c = cmd_special_alloc(h);
        if (!c)
                return -ENOMEM;
        return_status = fill_cmd(h, c, cmd, buff, size, page_code,
                scsi3addr, cmd_type);
        if (return_status == IO_OK)
                return_status = sendcmd_withirq_core(h, c, 1);

        cmd_special_free(h, c);
        return return_status;
}

static void cciss_geometry_inquiry(ctlr_info_t *h, int logvol,
                                   sector_t total_size,
                                   unsigned int block_size,
                                   InquiryData_struct *inq_buff,
                                   drive_info_struct *drv)
{
        int return_code;
        unsigned long t;
        unsigned char scsi3addr[8];

        memset(inq_buff, 0, sizeof(InquiryData_struct));
        log_unit_to_scsi3addr(h, scsi3addr, logvol);
        return_code = sendcmd_withirq(h, CISS_INQUIRY, inq_buff,
                        sizeof(*inq_buff), 0xC1, scsi3addr, TYPE_CMD);
        if (return_code == IO_OK) {
                if (inq_buff->data_byte[8] == 0xFF) {
                        dev_warn(&h->pdev->dev,
                               "reading geometry failed, volume "
                               "does not support reading geometry\n");
                        drv->heads = 255;
                        drv->sectors = 32;      /* Sectors per track */
                        drv->cylinders = total_size + 1;
                        drv->raid_level = RAID_UNKNOWN;
                } else {
                        drv->heads = inq_buff->data_byte[6];
                        drv->sectors = inq_buff->data_byte[7];
                        drv->cylinders = (inq_buff->data_byte[4] & 0xff) << 8;
                        drv->cylinders += inq_buff->data_byte[5];
                        drv->raid_level = inq_buff->data_byte[8];
                }
                drv->block_size = block_size;
                drv->nr_blocks = total_size + 1;
                t = drv->heads * drv->sectors;
                if (t > 1) {
                        sector_t real_size = total_size + 1;
                        unsigned long rem = sector_div(real_size, t);
                        if (rem)
                                real_size++;
                        drv->cylinders = real_size;
                }
        } else {                /* Get geometry failed */
                dev_warn(&h->pdev->dev, "reading geometry failed\n");
        }
}

static void
cciss_read_capacity(ctlr_info_t *h, int logvol, sector_t *total_size,
                    unsigned int *block_size)
{
        ReadCapdata_struct *buf;
        int return_code;
        unsigned char scsi3addr[8];

        buf = kzalloc(sizeof(ReadCapdata_struct), GFP_KERNEL);
        if (!buf) {
                dev_warn(&h->pdev->dev, "out of memory\n");
                return;
        }

        log_unit_to_scsi3addr(h, scsi3addr, logvol);
        return_code = sendcmd_withirq(h, CCISS_READ_CAPACITY, buf,
                sizeof(ReadCapdata_struct), 0, scsi3addr, TYPE_CMD);
        if (return_code == IO_OK) {
                *total_size = be32_to_cpu(*(__be32 *) buf->total_size);
                *block_size = be32_to_cpu(*(__be32 *) buf->block_size);
        } else {                /* read capacity command failed */
                dev_warn(&h->pdev->dev, "read capacity failed\n");
                *total_size = 0;
                *block_size = BLOCK_SIZE;
        }
        kfree(buf);
}

static void cciss_read_capacity_16(ctlr_info_t *h, int logvol,
        sector_t *total_size, unsigned int *block_size)
{
        ReadCapdata_struct_16 *buf;
        int return_code;
        unsigned char scsi3addr[8];

        buf = kzalloc(sizeof(ReadCapdata_struct_16), GFP_KERNEL);
        if (!buf) {
                dev_warn(&h->pdev->dev, "out of memory\n");
                return;
        }

        log_unit_to_scsi3addr(h, scsi3addr, logvol);
        return_code = sendcmd_withirq(h, CCISS_READ_CAPACITY_16,
                buf, sizeof(ReadCapdata_struct_16),
                        0, scsi3addr, TYPE_CMD);
        if (return_code == IO_OK) {
                *total_size = be64_to_cpu(*(__be64 *) buf->total_size);
                *block_size = be32_to_cpu(*(__be32 *) buf->block_size);
        } else {                /* read capacity command failed */
                dev_warn(&h->pdev->dev, "read capacity failed\n");
                *total_size = 0;
                *block_size = BLOCK_SIZE;
        }
        dev_info(&h->pdev->dev, "      blocks= %llu block_size= %d\n",
               (unsigned long long)*total_size+1, *block_size);
        kfree(buf);
}

static int cciss_revalidate(struct gendisk *disk)
{
        ctlr_info_t *h = get_host(disk);
        drive_info_struct *drv = get_drv(disk);
        int logvol;
        int FOUND = 0;
        unsigned int block_size;
        sector_t total_size;
        InquiryData_struct *inq_buff = NULL;

        for (logvol = 0; logvol < CISS_MAX_LUN; logvol++) {
                if (!h->drv[logvol])
                        continue;
                if (memcmp(h->drv[logvol]->LunID, drv->LunID,
                        sizeof(drv->LunID)) == 0) {
                        FOUND = 1;
                        break;
                }
        }

        if (!FOUND)
                return 1;

        inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL);
        if (inq_buff == NULL) {
                dev_warn(&h->pdev->dev, "out of memory\n");
                return 1;
        }
        if (h->cciss_read == CCISS_READ_10) {
                cciss_read_capacity(h, logvol,
                                        &total_size, &block_size);
        } else {
                cciss_read_capacity_16(h, logvol,
                                        &total_size, &block_size);
        }
        cciss_geometry_inquiry(h, logvol, total_size, block_size,
                               inq_buff, drv);

        blk_queue_logical_block_size(drv->queue, drv->block_size);
        set_capacity(disk, drv->nr_blocks);

        kfree(inq_buff);
        return 0;
}

/*
 * Map (physical) PCI mem into (virtual) kernel space
 */
static void __iomem *remap_pci_mem(ulong base, ulong size)
{
        ulong page_base = ((ulong) base) & PAGE_MASK;
        ulong page_offs = ((ulong) base) - page_base;
        void __iomem *page_remapped = ioremap(page_base, page_offs + size);

        return page_remapped ? (page_remapped + page_offs) : NULL;
}

/*
 * Takes jobs of the Q and sends them to the hardware, then puts it on
 * the Q to wait for completion.
 */
static void start_io(ctlr_info_t *h)
{
        CommandList_struct *c;

        while (!hlist_empty(&h->reqQ)) {
                c = hlist_entry(h->reqQ.first, CommandList_struct, list);
                /* can't do anything if fifo is full */
                if ((h->access.fifo_full(h))) {
                        dev_warn(&h->pdev->dev, "fifo full\n");
                        break;
                }

                /* Get the first entry from the Request Q */
                removeQ(c);
                h->Qdepth--;

                /* Tell the controller execute command */
                h->access.submit_command(h, c);

                /* Put job onto the completed Q */
                addQ(&h->cmpQ, c);
        }
}

/* Assumes that h->lock is held. */
/* Zeros out the error record and then resends the command back */
/* to the controller */
static inline void resend_cciss_cmd(ctlr_info_t *h, CommandList_struct *c)
{
        /* erase the old error information */
        memset(c->err_info, 0, sizeof(ErrorInfo_struct));

        /* add it to software queue and then send it to the controller */
        addQ(&h->reqQ, c);
        h->Qdepth++;
        if (h->Qdepth > h->maxQsinceinit)
                h->maxQsinceinit = h->Qdepth;

        start_io(h);
}

static inline unsigned int make_status_bytes(unsigned int scsi_status_byte,
        unsigned int msg_byte, unsigned int host_byte,
        unsigned int driver_byte)
{
        /* inverse of macros in scsi.h */
        return (scsi_status_byte & 0xff) |
                ((msg_byte & 0xff) << 8) |
                ((host_byte & 0xff) << 16) |
                ((driver_byte & 0xff) << 24);
}

static inline int evaluate_target_status(ctlr_info_t *h,
                        CommandList_struct *cmd, int *retry_cmd)
{
        unsigned char sense_key;
        unsigned char status_byte, msg_byte, host_byte, driver_byte;
        int error_value;

        *retry_cmd = 0;
        /* If we get in here, it means we got "target status", that is, scsi status */
        status_byte = cmd->err_info->ScsiStatus;
        driver_byte = DRIVER_OK;
        msg_byte = cmd->err_info->CommandStatus; /* correct?  seems too device specific */

        if (cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC)
                host_byte = DID_PASSTHROUGH;
        else
                host_byte = DID_OK;

        error_value = make_status_bytes(status_byte, msg_byte,
                host_byte, driver_byte);

        if (cmd->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION) {
                if (cmd->rq->cmd_type != REQ_TYPE_BLOCK_PC)
                        dev_warn(&h->pdev->dev, "cmd %p "
                               "has SCSI Status 0x%x\n",
                               cmd, cmd->err_info->ScsiStatus);
                return error_value;
        }

        /* check the sense key */
        sense_key = 0xf & cmd->err_info->SenseInfo[2];
        /* no status or recovered error */
        if (((sense_key == 0x0) || (sense_key == 0x1)) &&
            (cmd->rq->cmd_type != REQ_TYPE_BLOCK_PC))
                error_value = 0;

        if (check_for_unit_attention(h, cmd)) {
                *retry_cmd = !(cmd->rq->cmd_type == REQ_TYPE_BLOCK_PC);
                return 0;
        }

        /* Not SG_IO or similar? */
        if (cmd->rq->cmd_type != REQ_TYPE_BLOCK_PC) {
                if (error_value != 0)
                        dev_warn(&h->pdev->dev, "cmd %p has CHECK CONDITION"
                               " sense key = 0x%x\n", cmd, sense_key);
                return error_value;
        }

        /* SG_IO or similar, copy sense data back */
        if (cmd->rq->sense) {
                if (cmd->rq->sense_len > cmd->err_info->SenseLen)
                        cmd->rq->sense_len = cmd->err_info->SenseLen;
                memcpy(cmd->rq->sense, cmd->err_info->SenseInfo,
                        cmd->rq->sense_len);
        } else
                cmd->rq->sense_len = 0;

        return error_value;
}

/* checks the status of the job and calls complete buffers to mark all