block: remove some includings of blktrace_api.h
[safe/jmp/linux-2.6] / drivers / block / cciss.c
index d719a5d..c7a527c 100644 (file)
@@ -1,20 +1,20 @@
 /*
- *    Disk Array driver for HP SA 5xxx and 6xxx Controllers
- *    Copyright 2000, 2006 Hewlett-Packard Development Company, L.P.
+ *    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; either version 2 of the License, or
- *    (at your option) any later version.
+ *    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, GOOD TITLE or
- *    NON INFRINGEMENT.  See the GNU General Public License for more details.
+ *    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., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+ *    02111-1307, USA.
  *
  *    Questions/Comments/Bugfixes to iss_storagedev@hp.com
  *
 #include <linux/blkpg.h>
 #include <linux/timer.h>
 #include <linux/proc_fs.h>
+#include <linux/seq_file.h>
 #include <linux/init.h>
 #include <linux/hdreg.h>
 #include <linux/spinlock.h>
 #include <linux/compat.h>
-#include <linux/blktrace_api.h>
 #include <asm/uaccess.h>
 #include <asm/io.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 3.6.14)"
-#define DRIVER_VERSION CCISS_DRIVER_VERSION(3,6,14)
+#define DRIVER_NAME "HP CISS Driver (v 3.6.20)"
+#define DRIVER_VERSION CCISS_DRIVER_VERSION(3, 6, 20)
 
 /* Embedded module documentation macros - see modules.h */
 MODULE_AUTHOR("Hewlett-Packard Company");
-MODULE_DESCRIPTION("Driver for HP Controller SA5xxx SA6xxx version 3.6.14");
+MODULE_DESCRIPTION("Driver for HP Smart Array Controllers");
 MODULE_SUPPORTED_DEVICE("HP SA5i SA5i+ SA532 SA5300 SA5312 SA641 SA642 SA6400"
-                       " SA6i P600 P800 P400 P400i E200 E200i E500");
-MODULE_VERSION("3.6.14");
+                       " SA6i P600 P800 P400 P400i E200 E200i E500 P700m"
+                       " Smart Array G2 Series SAS/SATA Controllers");
+MODULE_VERSION("3.6.20");
 MODULE_LICENSE("GPL");
 
 #include "cciss_cmd.h"
@@ -83,6 +90,14 @@ static const struct pci_device_id cciss_pci_device_id[] = {
        {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},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3241},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3243},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3245},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3247},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x3249},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x324A},
+       {PCI_VENDOR_ID_HP,     PCI_DEVICE_ID_HP_CISSE,     0x103C, 0x324B},
        {PCI_VENDOR_ID_HP,     PCI_ANY_ID,      PCI_ANY_ID, PCI_ANY_ID,
                PCI_CLASS_STORAGE_RAID << 8, 0xffff << 8, 0},
        {0,}
@@ -93,29 +108,36 @@ 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
- *  nr_cmds = Number of commands supported by controller
  */
 static struct board_type products[] = {
-       {0x40700E11, "Smart Array 5300", &SA5_access, 512},
-       {0x40800E11, "Smart Array 5i", &SA5B_access, 512},
-       {0x40820E11, "Smart Array 532", &SA5B_access, 512},
-       {0x40830E11, "Smart Array 5312", &SA5B_access, 512},
-       {0x409A0E11, "Smart Array 641", &SA5_access, 512},
-       {0x409B0E11, "Smart Array 642", &SA5_access, 512},
-       {0x409C0E11, "Smart Array 6400", &SA5_access, 512},
-       {0x409D0E11, "Smart Array 6400 EM", &SA5_access, 512},
-       {0x40910E11, "Smart Array 6i", &SA5_access, 512},
-       {0x3225103C, "Smart Array P600", &SA5_access, 512},
-       {0x3223103C, "Smart Array P800", &SA5_access, 512},
-       {0x3234103C, "Smart Array P400", &SA5_access, 512},
-       {0x3235103C, "Smart Array P400i", &SA5_access, 512},
-       {0x3211103C, "Smart Array E200i", &SA5_access, 120},
-       {0x3212103C, "Smart Array E200", &SA5_access, 120},
-       {0x3213103C, "Smart Array E200i", &SA5_access, 120},
-       {0x3214103C, "Smart Array E200i", &SA5_access, 120},
-       {0x3215103C, "Smart Array E200i", &SA5_access, 120},
-       {0x3237103C, "Smart Array E500", &SA5_access, 512},
-       {0xFFFF103C, "Unknown Smart Array", &SA5_access, 120},
+       {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},
+       {0x323D103C, "Smart Array P700m", &SA5_access},
+       {0x3241103C, "Smart Array P212", &SA5_access},
+       {0x3243103C, "Smart Array P410", &SA5_access},
+       {0x3245103C, "Smart Array P410i", &SA5_access},
+       {0x3247103C, "Smart Array P411", &SA5_access},
+       {0x3249103C, "Smart Array P812", &SA5_access},
+       {0x324A103C, "Smart Array P712m", &SA5_access},
+       {0x324B103C, "Smart Array P711m", &SA5_access},
+       {0xFFFF103C, "Unknown Smart Array", &SA5_access},
 };
 
 /* How long to wait (in milliseconds) for board to go into simple mode */
@@ -125,7 +147,6 @@ static struct board_type products[] = {
 /*define how many times we will try a command because of bus resets */
 #define MAX_CMD_RETRIES 3
 
-#define READ_AHEAD      1024
 #define MAX_CTLR       32
 
 /* Originally cciss driver only supports 8 major numbers */
@@ -133,17 +154,17 @@ static struct board_type products[] = {
 
 static ctlr_info_t *hba[MAX_CTLR];
 
-static void do_cciss_request(request_queue_t *q);
+static void do_cciss_request(struct request_queue *q);
 static irqreturn_t do_cciss_intr(int irq, void *dev_id);
-static int cciss_open(struct inode *inode, struct file *filep);
-static int cciss_release(struct inode *inode, struct file *filep);
-static int cciss_ioctl(struct inode *inode, struct file *filep,
+static int cciss_open(struct block_device *bdev, fmode_t mode);
+static int cciss_release(struct gendisk *disk, fmode_t mode);
+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, struct gendisk *del_disk);
-static int deregister_disk(struct gendisk *disk, drive_info_struct *drv,
+static int rebuild_lun_table(ctlr_info_t *h, int first_time);
+static int deregister_disk(ctlr_info_t *h, int drv_index,
                           int clear_all);
 
 static void cciss_read_capacity(int ctlr, int logvol, int withirq,
@@ -154,22 +175,23 @@ static void cciss_geometry_inquiry(int ctlr, int logvol,
                        int withirq, sector_t total_size,
                        unsigned int block_size, InquiryData_struct *inq_buff,
                                   drive_info_struct *drv);
-static void cciss_getgeometry(int cntl_num);
 static void __devinit cciss_interrupt_mode(ctlr_info_t *, struct pci_dev *,
                                           __u32);
 static void start_io(ctlr_info_t *h);
 static int sendcmd(__u8 cmd, int ctlr, void *buff, size_t size,
-                  unsigned int use_unit_num, unsigned int log_unit,
                   __u8 page_code, unsigned char *scsi3addr, int cmd_type);
 static int sendcmd_withirq(__u8 cmd, int ctlr, void *buff, size_t size,
-                          unsigned int use_unit_num, unsigned int log_unit,
-                          __u8 page_code, int cmd_type);
+                       __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 void fail_all_cmds(unsigned long ctlr);
+static int scan_thread(void *data);
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
 
 #ifdef CONFIG_PROC_FS
-static int cciss_proc_get_info(char *buffer, char **start, off_t offset,
-                              int length, int *eof, void *data);
 static void cciss_procinit(int i);
 #else
 static void cciss_procinit(int i)
@@ -178,14 +200,15 @@ static void cciss_procinit(int i)
 #endif                         /* CONFIG_PROC_FS */
 
 #ifdef CONFIG_COMPAT
-static long cciss_compat_ioctl(struct file *f, unsigned cmd, unsigned long arg);
+static int cciss_compat_ioctl(struct block_device *, fmode_t,
+                             unsigned, unsigned long);
 #endif
 
 static struct block_device_operations cciss_fops = {
        .owner = THIS_MODULE,
        .open = cciss_open,
        .release = cciss_release,
-       .ioctl = cciss_ioctl,
+       .locked_ioctl = cciss_ioctl,
        .getgeo = cciss_getgeo,
 #ifdef CONFIG_COMPAT
        .compat_ioctl = cciss_compat_ioctl,
@@ -196,35 +219,23 @@ static struct block_device_operations cciss_fops = {
 /*
  * Enqueuing and dequeuing functions for cmdlists.
  */
-static inline void addQ(CommandList_struct **Qptr, CommandList_struct *c)
+static inline void addQ(struct hlist_head *list, CommandList_struct *c)
 {
-       if (*Qptr == NULL) {
-               *Qptr = c;
-               c->next = c->prev = c;
-       } else {
-               c->prev = (*Qptr)->prev;
-               c->next = (*Qptr);
-               (*Qptr)->prev->next = c;
-               (*Qptr)->prev = c;
-       }
+       hlist_add_head(&c->list, list);
 }
 
-static inline CommandList_struct *removeQ(CommandList_struct **Qptr,
-                                         CommandList_struct *c)
+static inline void removeQ(CommandList_struct *c)
 {
-       if (c && c->next != c) {
-               if (*Qptr == c)
-                       *Qptr = c->next;
-               c->prev->next = c->next;
-               c->next->prev = c->prev;
-       } else {
-               *Qptr = NULL;
-       }
-       return c;
+       if (WARN_ON(hlist_unhashed(&c->list)))
+               return;
+
+       hlist_del_init(&c->list);
 }
 
 #include "cciss_scsi.c"                /* For SCSI tape support */
 
+#define RAID_UNKNOWN 6
+
 #ifdef CONFIG_PROC_FS
 
 /*
@@ -232,25 +243,46 @@ static inline CommandList_struct *removeQ(CommandList_struct **Qptr,
  */
 #define ENG_GIG 1000000000
 #define ENG_GIG_FACTOR (ENG_GIG/512)
-#define RAID_UNKNOWN 6
+#define ENGAGE_SCSI    "engage scsi"
 static const char *raid_label[] = { "0", "4", "1(1+0)", "5", "5+1", "ADG",
        "UNKNOWN"
 };
 
 static struct proc_dir_entry *proc_cciss;
 
-static int cciss_proc_get_info(char *buffer, char **start, off_t offset,
-                              int length, int *eof, void *data)
+static void cciss_seq_show_header(struct seq_file *seq)
 {
-       off_t pos = 0;
-       off_t len = 0;
-       int size, i, ctlr;
-       ctlr_info_t *h = (ctlr_info_t *) data;
-       drive_info_struct *drv;
-       unsigned long flags;
-       sector_t vol_sz, vol_sz_frac;
+       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[SIMPLE_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->ctlr);
+#endif /* CONFIG_CISS_SCSI_TAPE */
+}
 
-       ctlr = h->ctlr;
+static void *cciss_seq_start(struct seq_file *seq, loff_t *pos)
+{
+       ctlr_info_t *h = seq->private;
+       unsigned ctlr = h->ctlr;
+       unsigned long flags;
 
        /* prevent displaying bogus info during configuration
         * or deconfiguration of a logical volume
@@ -258,118 +290,342 @@ static int cciss_proc_get_info(char *buffer, char **start, off_t offset,
        spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
        if (h->busy_configuring) {
                spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
-               return -EBUSY;
+               return ERR_PTR(-EBUSY);
        }
        h->busy_configuring = 1;
        spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
 
-       size = sprintf(buffer, "%s: HP %s Controller\n"
-                      "Board ID: 0x%08lx\n"
-                      "Firmware Version: %c%c%c%c\n"
-                      "IRQ: %d\n"
-                      "Logical drives: %d\n"
-                      "Max sectors: %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\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[SIMPLE_MODE_INT],
-                      h->num_luns,
-                      h->cciss_max_sectors,
-                      h->Qdepth, h->commands_outstanding,
-                      h->maxQsinceinit, h->max_outstanding, h->maxSG);
-
-       pos += size;
-       len += size;
-       cciss_proc_tape_report(ctlr, buffer, &pos, &len);
-       for (i = 0; i <= h->highest_lun; i++) {
+       if (*pos == 0)
+               cciss_seq_show_header(seq);
 
-               drv = &h->drv[i];
-               if (drv->heads == 0)
-                       continue;
+       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];
 
-               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 (*pos > h->highest_lun)
+               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 > 5)
+               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;
 
-               if (drv->raid_level > 5)
-                       drv->raid_level = RAID_UNKNOWN;
-               size = sprintf(buffer + len, "cciss/c%dd%d:"
-                              "\t%4u.%02uGB\tRAID %s\n",
-                              ctlr, i, (int)vol_sz, (int)vol_sz_frac,
-                              raid_label[drv->raid_level]);
-               pos += size;
-               len += size;
-       }
-
-       *eof = 1;
-       *start = buffer + offset;
-       len -= offset;
-       if (len > length)
-               len = length;
        h->busy_configuring = 0;
-       return len;
 }
 
-static int
-cciss_proc_write(struct file *file, const char __user *buffer,
-                unsigned long count, void *data)
+static 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)
 {
-       unsigned char cmd[80];
-       int len;
-#ifdef CONFIG_CISS_SCSI_TAPE
-       ctlr_info_t *h = (ctlr_info_t *) data;
-       int rc;
+       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 (count > sizeof(cmd) - 1)
+       if (!buf || length > PAGE_SIZE - 1)
                return -EINVAL;
-       if (copy_from_user(cmd, buffer, count))
-               return -EFAULT;
-       cmd[count] = '\0';
-       len = strlen(cmd);      // above 3 lines ensure safety
-       if (len && cmd[len - 1] == '\n')
-               cmd[--len] = '\0';
-#      ifdef CONFIG_CISS_SCSI_TAPE
-       if (strcmp("engage scsi", cmd) == 0) {
+
+       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;
+               int rc;
+
                rc = cciss_engage_scsi(h->ctlr);
                if (rc != 0)
-                       return -rc;
-               return count;
-       }
+                       err = -rc;
+               else
+                       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.) */
-#      endif
-       return -EINVAL;
+
+out:
+       free_page((unsigned long)buffer);
+       return err;
 }
 
-/*
- * Get us a file in /proc/cciss that says something about each controller.
- * Create /proc/cciss if it doesn't exist yet.
- */
+static 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(int i)
 {
        struct proc_dir_entry *pde;
 
-       if (proc_cciss == NULL) {
-               proc_cciss = proc_mkdir("cciss", proc_root_driver);
-               if (!proc_cciss)
-                       return;
-       }
-
-       pde = create_proc_read_entry(hba[i]->devname,
-                                    S_IWUSR | S_IRUSR | S_IRGRP | S_IROTH,
-                                    proc_cciss, cciss_proc_get_info, hba[i]);
-       pde->write_proc = cciss_proc_write;
+       if (proc_cciss == NULL)
+               proc_cciss = proc_mkdir("driver/cciss", NULL);
+       if (!proc_cciss)
+               return;
+       pde = proc_create_data(hba[i]->devname, S_IWUSR | S_IRUSR | S_IRGRP |
+                                       S_IROTH, proc_cciss,
+                                       &cciss_proc_fops, hba[i]);
 }
 #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 struct device_type cciss_host_type = {
+       .name           = "cciss_host",
+};
+
+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(CCISS_LOCK(h->ctlr), flags);
+       if (h->busy_configuring)
+               ret = -EBUSY;
+       else
+               memcpy(sn, drv->serial_no, sizeof(sn));
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), 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]);
+}
+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(CCISS_LOCK(h->ctlr), flags);
+       if (h->busy_configuring)
+               ret = -EBUSY;
+       else
+               memcpy(vendor, drv->vendor, VENDOR_LEN + 1);
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+
+       if (ret)
+               return ret;
+       else
+               return snprintf(buf, sizeof(vendor) + 1, "%s\n", drv->vendor);
+}
+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(CCISS_LOCK(h->ctlr), flags);
+       if (h->busy_configuring)
+               ret = -EBUSY;
+       else
+               memcpy(model, drv->model, MODEL_LEN + 1);
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+
+       if (ret)
+               return ret;
+       else
+               return snprintf(buf, sizeof(model) + 1, "%s\n", drv->model);
+}
+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(CCISS_LOCK(h->ctlr), flags);
+       if (h->busy_configuring)
+               ret = -EBUSY;
+       else
+               memcpy(rev, drv->rev, REV_LEN + 1);
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+
+       if (ret)
+               return ret;
+       else
+               return snprintf(buf, sizeof(rev) + 1, "%s\n", drv->rev);
+}
+DEVICE_ATTR(rev, S_IRUGO, dev_show_rev, NULL);
+
+static struct attribute *cciss_dev_attrs[] = {
+       &dev_attr_unique_id.attr,
+       &dev_attr_model.attr,
+       &dev_attr_vendor.attr,
+       &dev_attr_rev.attr,
+       NULL
+};
+
+static struct attribute_group cciss_dev_attr_group = {
+       .attrs = cciss_dev_attrs,
+};
+
+static 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,
+};
+
+static struct bus_type cciss_bus_type = {
+       .name           = "cciss",
+};
+
+
+/*
+ * 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);
+}
+
+/*
+ * 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 int cciss_create_ld_sysfs_entry(struct ctlr_info *h,
+                                      drive_info_struct *drv,
+                                      int drv_index)
+{
+       device_initialize(&drv->dev);
+       drv->dev.type = &cciss_dev_type;
+       drv->dev.bus = &cciss_bus_type;
+       dev_set_name(&drv->dev, "c%dd%d", h->ctlr, drv_index);
+       drv->dev.parent = &h->dev;
+       return device_add(&drv->dev);
+}
+
+/*
+ * Remove sysfs entries for a logical drive.
+ */
+static void cciss_destroy_ld_sysfs_entry(drive_info_struct *drv)
+{
+       device_del(&drv->dev);
+}
+
 /*
  * 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
@@ -428,6 +684,7 @@ static CommandList_struct *cmd_alloc(ctlr_info_t *h, int get_from_pool)
                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;
@@ -474,13 +731,13 @@ static inline drive_info_struct *get_drv(struct gendisk *disk)
 /*
  * Open.  Make sure the device is really there.
  */
-static int cciss_open(struct inode *inode, struct file *filep)
+static int cciss_open(struct block_device *bdev, fmode_t mode)
 {
-       ctlr_info_t *host = get_host(inode->i_bdev->bd_disk);
-       drive_info_struct *drv = get_drv(inode->i_bdev->bd_disk);
+       ctlr_info_t *host = get_host(bdev->bd_disk);
+       drive_info_struct *drv = get_drv(bdev->bd_disk);
 
 #ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "cciss_open %s\n", inode->i_bdev->bd_disk->disk_name);
+       printk(KERN_DEBUG "cciss_open %s\n", bdev->bd_disk->disk_name);
 #endif                         /* CCISS_DEBUG */
 
        if (host->busy_initializing || drv->busy_configuring)
@@ -494,9 +751,9 @@ static int cciss_open(struct inode *inode, struct file *filep)
         * for "raw controller".
         */
        if (drv->heads == 0) {
-               if (iminor(inode) != 0) {       /* not node 0? */
+               if (MINOR(bdev->bd_dev) != 0) { /* not node 0? */
                        /* if not node 0 make sure it is a partition = 0 */
-                       if (iminor(inode) & 0x0f) {
+                       if (MINOR(bdev->bd_dev) & 0x0f) {
                                return -ENXIO;
                                /* if it is, make sure we have a LUN ID */
                        } else if (drv->LunID == 0) {
@@ -514,14 +771,13 @@ static int cciss_open(struct inode *inode, struct file *filep)
 /*
  * Close.  Sync first.
  */
-static int cciss_release(struct inode *inode, struct file *filep)
+static int cciss_release(struct gendisk *disk, fmode_t mode)
 {
-       ctlr_info_t *host = get_host(inode->i_bdev->bd_disk);
-       drive_info_struct *drv = get_drv(inode->i_bdev->bd_disk);
+       ctlr_info_t *host = get_host(disk);
+       drive_info_struct *drv = get_drv(disk);
 
 #ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "cciss_release %s\n",
-              inode->i_bdev->bd_disk->disk_name);
+       printk(KERN_DEBUG "cciss_release %s\n", disk->disk_name);
 #endif                         /* CCISS_DEBUG */
 
        drv->usage_count--;
@@ -531,21 +787,23 @@ static int cciss_release(struct inode *inode, struct file *filep)
 
 #ifdef CONFIG_COMPAT
 
-static int do_ioctl(struct file *f, unsigned cmd, unsigned long arg)
+static int do_ioctl(struct block_device *bdev, fmode_t mode,
+                   unsigned cmd, unsigned long arg)
 {
        int ret;
        lock_kernel();
-       ret = cciss_ioctl(f->f_path.dentry->d_inode, f, cmd, arg);
+       ret = cciss_ioctl(bdev, mode, cmd, arg);
        unlock_kernel();
        return ret;
 }
 
-static int cciss_ioctl32_passthru(struct file *f, unsigned cmd,
-                                 unsigned long arg);
-static int cciss_ioctl32_big_passthru(struct file *f, unsigned cmd,
-                                     unsigned long arg);
+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 long cciss_compat_ioctl(struct file *f, 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:
@@ -563,20 +821,20 @@ static long cciss_compat_ioctl(struct file *f, unsigned cmd, unsigned long arg)
        case CCISS_REGNEWD:
        case CCISS_RESCANDISK:
        case CCISS_GETLUNINFO:
-               return do_ioctl(f, cmd, arg);
+               return do_ioctl(bdev, mode, cmd, arg);
 
        case CCISS_PASSTHRU32:
-               return cciss_ioctl32_passthru(f, cmd, arg);
+               return cciss_ioctl32_passthru(bdev, mode, cmd, arg);
        case CCISS_BIG_PASSTHRU32:
-               return cciss_ioctl32_big_passthru(f, cmd, arg);
+               return cciss_ioctl32_big_passthru(bdev, mode, cmd, arg);
 
        default:
                return -ENOIOCTLCMD;
        }
 }
 
-static int cciss_ioctl32_passthru(struct file *f, unsigned cmd,
-                                 unsigned long arg)
+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;
@@ -603,7 +861,7 @@ static int cciss_ioctl32_passthru(struct file *f, unsigned cmd,
        if (err)
                return -EFAULT;
 
-       err = do_ioctl(f, CCISS_PASSTHRU, (unsigned long)p);
+       err = do_ioctl(bdev, mode, CCISS_PASSTHRU, (unsigned long)p);
        if (err)
                return err;
        err |=
@@ -614,8 +872,8 @@ static int cciss_ioctl32_passthru(struct file *f, unsigned cmd,
        return err;
 }
 
-static int cciss_ioctl32_big_passthru(struct file *file, unsigned cmd,
-                                     unsigned long arg)
+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;
@@ -644,7 +902,7 @@ static int cciss_ioctl32_big_passthru(struct file *file, unsigned cmd,
        if (err)
                return -EFAULT;
 
-       err = do_ioctl(file, CCISS_BIG_PASSTHRU, (unsigned long)p);
+       err = do_ioctl(bdev, mode, CCISS_BIG_PASSTHRU, (unsigned long)p);
        if (err)
                return err;
        err |=
@@ -669,13 +927,18 @@ static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo)
        return 0;
 }
 
+static void check_ioctl_unit_attention(ctlr_info_t *host, CommandList_struct *c)
+{
+       if (c->err_info->CommandStatus == CMD_TARGET_STATUS &&
+                       c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION)
+               (void)check_for_unit_attention(host, c);
+}
 /*
  * ioctl
  */
-static int cciss_ioctl(struct inode *inode, struct file *filep,
+static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
                       unsigned int cmd, unsigned long arg)
 {
-       struct block_device *bdev = inode->i_bdev;
        struct gendisk *disk = bdev->bd_disk;
        ctlr_info_t *host = get_host(disk);
        drive_info_struct *drv = get_drv(disk);
@@ -855,8 +1118,10 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                        return 0;
                }
 
+       case CCISS_DEREGDISK:
+       case CCISS_REGNEWD:
        case CCISS_REVALIDVOLS:
-               return rebuild_lun_table(host, NULL);
+               return rebuild_lun_table(host, 0);
 
        case CCISS_GETLUNINFO:{
                        LogvolInfo_struct luninfo;
@@ -869,12 +1134,6 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                                return -EFAULT;
                        return 0;
                }
-       case CCISS_DEREGDISK:
-               return rebuild_lun_table(host, disk);
-
-       case CCISS_REGNEWD:
-               return rebuild_lun_table(host, NULL);
-
        case CCISS_PASSTHRU:
                {
                        IOCTL_Command_struct iocommand;
@@ -968,6 +1227,8 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                                         iocommand.buf_size,
                                         PCI_DMA_BIDIRECTIONAL);
 
+                       check_ioctl_unit_attention(host, c);
+
                        /* Copy the error information out */
                        iocommand.error_info = *(c->err_info);
                        if (copy_to_user
@@ -1060,7 +1321,7 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                                if (ioc->Request.Type.Direction == XFER_WRITE) {
                                        if (copy_from_user
                                            (buff[sg_used], data_ptr, sz)) {
-                                               status = -ENOMEM;
+                                               status = -EFAULT;
                                                goto cleanup1;
                                        }
                                } else {
@@ -1119,6 +1380,7 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                                        (dma_addr_t) temp64.val, buff_size[i],
                                        PCI_DMA_BIDIRECTIONAL);
                        }
+                       check_ioctl_unit_attention(host, c);
                        /* Copy the error information out */
                        ioc->error_info = *(c->err_info);
                        if (copy_to_user(argp, ioc, sizeof(*ioc))) {
@@ -1151,23 +1413,35 @@ static int cciss_ioctl(struct inode *inode, struct file *filep,
                        kfree(ioc);
                        return status;
                }
+
+       /* 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 inline void complete_buffers(struct bio *bio, int status)
-{
-       while (bio) {
-               struct bio *xbh = bio->bi_next;
-               int nr_sectors = bio_sectors(bio);
-
-               bio->bi_next = NULL;
-               bio_endio(bio, nr_sectors << 9, status ? 0 : -EIO);
-               bio = xbh;
-       }
-}
-
 static void cciss_check_queues(ctlr_info_t *h)
 {
        int start_queue = h->next_to_run;
@@ -1206,8 +1480,6 @@ static void cciss_check_queues(ctlr_info_t *h)
                                h->next_to_run = curr_queue;
                                break;
                        }
-               } else {
-                       curr_queue = (curr_queue + 1) % (h->highest_lun + 1);
                }
        }
 }
@@ -1233,35 +1505,153 @@ static void cciss_softirq_done(struct request *rq)
                pci_unmap_page(h->pdev, temp64.val, cmd->SG[i].Len, ddir);
        }
 
-       complete_buffers(rq->bio, rq->errors);
-
-       if (blk_fs_request(rq)) {
-               const int rw = rq_data_dir(rq);
-
-               disk_stat_add(rq->rq_disk, sectors[rw], rq->nr_sectors);
-       }
-
 #ifdef CCISS_DEBUG
        printk("Done with %p\n", rq);
 #endif                         /* CCISS_DEBUG */
 
-       add_disk_randomness(rq->rq_disk);
+       /* set the residual count for pc requests */
+       if (blk_pc_request(rq))
+               rq->resid_len = cmd->err_info->ResidualCnt;
+
+       blk_end_request_all(rq, (rq->errors == 0) ? 0 : -EIO);
+
        spin_lock_irqsave(&h->lock, flags);
-       end_that_request_last(rq, rq->errors);
        cmd_free(h, cmd, 1);
        cciss_check_queues(h);
        spin_unlock_irqrestore(&h->lock, flags);
 }
 
+static void log_unit_to_scsi3addr(ctlr_info_t *h, unsigned char scsi3addr[],
+       uint32_t log_unit)
+{
+       log_unit = h->drv[log_unit].LunID & 0x03fff;
+       memset(&scsi3addr[4], 0, 4);
+       memcpy(&scsi3addr[0], &log_unit, 4);
+       scsi3addr[3] |= 0x40;
+}
+
+/* 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(int ctlr, int logvol, int withirq,
+                                  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(hba[ctlr], scsi3addr, logvol);
+       if (withirq)
+               rc = sendcmd_withirq(CISS_INQUIRY, ctlr, inq_buf,
+                            sizeof(InquiryData_struct), 0,
+                               scsi3addr, TYPE_CMD);
+       else
+               rc = sendcmd(CISS_INQUIRY, ctlr, inq_buf,
+                            sizeof(InquiryData_struct), 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(int ctlr, int logvol, int withirq,
+                               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(hba[ctlr], scsi3addr, logvol);
+       if (withirq)
+               rc = sendcmd_withirq(CISS_INQUIRY, ctlr, buf,
+                       PAGE_83_INQ_BYTES, 0x83, scsi3addr, TYPE_CMD);
+       else
+               rc = sendcmd(CISS_INQUIRY, ctlr, buf,
+                       PAGE_83_INQ_BYTES, 0x83, scsi3addr, TYPE_CMD);
+       if (rc == IO_OK)
+               memcpy(serial_no, &buf[8], buflen);
+       kfree(buf);
+       return;
+}
+
+static void cciss_add_disk(ctlr_info_t *h, struct gendisk *disk,
+                               int drv_index)
+{
+       disk->queue = blk_init_queue(do_cciss_request, &h->lock);
+       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;
+       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_hw_segments(disk->queue, MAXSGENTRIES);
+
+       /* This is a limit in the driver and could be eliminated. */
+       blk_queue_max_phys_segments(disk->queue, MAXSGENTRIES);
+
+       blk_queue_max_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);
+}
+
 /* This function will check the usage_count of the drive to be updated/added.
- * If the usage_count is zero then the drive information will be updated and
- * the disk will be re-registered with the kernel.  If not 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.
+ * 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(int ctlr, int drv_index)
+static void cciss_update_drive_info(int ctlr, int drv_index, int first_time)
 {
        ctlr_info_t *h = hba[ctlr];
        struct gendisk *disk;
@@ -1270,108 +1660,224 @@ static void cciss_update_drive_info(int ctlr, int drv_index)
        sector_t total_size;
        unsigned long flags = 0;
        int ret = 0;
+       drive_info_struct *drvinfo;
+       int was_only_controller_node;
+
+       /* Get information about the disk and modify the driver structure */
+       inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL);
+       drvinfo = kmalloc(sizeof(*drvinfo), GFP_KERNEL);
+       if (inq_buff == NULL || drvinfo == NULL)
+               goto mem_msg;
+
+       /* See if we're trying to update the "controller node"
+        * this will happen the when the first logical drive gets
+        * created by ACU.
+        */
+       was_only_controller_node = (drv_index == 0 &&
+                               h->drv[0].raid_level == -1);
+
+       /* testing to see if 16-byte CDBs are already being used */
+       if (h->cciss_read == CCISS_READ_16) {
+               cciss_read_capacity_16(h->ctlr, drv_index, 1,
+                       &total_size, &block_size);
+
+       } else {
+               cciss_read_capacity(ctlr, drv_index, 1,
+                                   &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(ctlr, drv_index, 1,
+                       &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(ctlr, drv_index, 1, total_size, block_size,
+                              inq_buff, drvinfo);
+       drvinfo->block_size = block_size;
+       drvinfo->nr_blocks = total_size + 1;
+
+       cciss_get_device_descr(ctlr, drv_index, 1, drvinfo->vendor,
+                               drvinfo->model, drvinfo->rev);
+       cciss_get_serial_no(ctlr, drv_index, 1, drvinfo->serial_no,
+                       sizeof(drvinfo->serial_no));
+
+       /* 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 the disk already exists then deregister it before proceeding */
-       if (h->drv[drv_index].raid_level != -1) {
+       /* 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) {
+               printk(KERN_WARNING "disk %d has changed.\n", drv_index);
                spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
                h->drv[drv_index].busy_configuring = 1;
                spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
-               ret = deregister_disk(h->gendisk[drv_index],
-                                     &h->drv[drv_index], 0);
+
+               /* 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);
                h->drv[drv_index].busy_configuring = 0;
        }
 
        /* If the disk is in use return */
        if (ret)
-               return;
-
-       /* Get information about the disk and modify the driver structure */
-       inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL);
-       if (inq_buff == NULL)
-               goto mem_msg;
+               goto freeret;
 
-       cciss_read_capacity(ctlr, drv_index, 1,
-                           &total_size, &block_size);
-
-       /* total size = last LBA + 1 */
-       /* FFFFFFFF + 1 = 0, cannot have a logical volume of size 0 */
-       /* so we assume this volume this must be >2TB in size */
-       if (total_size == (__u32) 0) {
-               cciss_read_capacity_16(ctlr, drv_index, 1,
-               &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(ctlr, drv_index, 1, total_size, block_size,
-                              inq_buff, &h->drv[drv_index]);
+       /* Save the new information from cciss_geometry_inquiry
+        * and serial number inquiry.
+        */
+       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 the controller it's already added */
-       if (drv_index) {
-               disk->queue = blk_init_queue(do_cciss_request, &h->lock);
-               sprintf(disk->disk_name, "cciss/c%dd%d", ctlr, drv_index);
-               disk->major = h->major;
-               disk->first_minor = drv_index << NWD_SHIFT;
-               disk->fops = &cciss_fops;
-               disk->private_data = &h->drv[drv_index];
+       /* 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)
+               cciss_add_disk(h, disk, drv_index);
 
-               /* Set up queue information */
-               disk->queue->backing_dev_info.ra_pages = READ_AHEAD;
-               blk_queue_bounce_limit(disk->queue, hba[ctlr]->pdev->dma_mask);
-
-               /* This is a hardware imposed limit. */
-               blk_queue_max_hw_segments(disk->queue, MAXSGENTRIES);
-
-               /* This is a limit in the driver and could be eliminated. */
-               blk_queue_max_phys_segments(disk->queue, MAXSGENTRIES);
-
-               blk_queue_max_sectors(disk->queue, h->cciss_max_sectors);
-
-               blk_queue_softirq_done(disk->queue, cciss_softirq_done);
-
-               disk->queue->queuedata = hba[ctlr];
-
-               blk_queue_hardsect_size(disk->queue,
-                                       hba[ctlr]->drv[drv_index].block_size);
-
-               h->drv[drv_index].queue = disk->queue;
-               add_disk(disk);
-       }
-
-      freeret:
-       kfree(inq_buff);
-       return;
-      mem_msg:
-       printk(KERN_ERR "cciss: out of memory\n");
-       goto freeret;
-}
+freeret:
+       kfree(inq_buff);
+       kfree(drvinfo);
+       return;
+mem_msg:
+       printk(KERN_ERR "cciss: out of memory\n");
+       goto freeret;
+}
 
 /* This function will find the first index of the controllers drive array
  * that has a -1 for the raid_level 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 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_find_free_drive_index(int ctlr)
+static int cciss_find_free_drive_index(int ctlr, int controller_node)
 {
        int i;
 
        for (i = 0; i < CISS_MAX_LUN; i++) {
                if (hba[ctlr]->drv[i].raid_level == -1) {
                        if (i > hba[ctlr]->highest_lun)
-                               hba[ctlr]->highest_lun = i;
+                               if (!controller_node)
+                                       hba[ctlr]->highest_lun = i;
                        return i;
                }
        }
        return -1;
 }
 
+/* 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, __u32 lunid, int controller_node)
+{
+       int drv_index;
+
+       drv_index = cciss_find_free_drive_index(h->ctlr, 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]) {
+                       printk(KERN_ERR "cciss%d: could not "
+                               "allocate a new disk %d\n",
+                               h->ctlr, drv_index);
+                       return -1;
+               }
+       }
+       h->drv[drv_index].LunID = lunid;
+       if (cciss_create_ld_sysfs_entry(h, &h->drv[drv_index], 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:
+       put_disk(h->gendisk[drv_index]);
+       h->gendisk[drv_index] = NULL;
+       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, 0, 1);
+       if (drv_index == -1) {
+               printk(KERN_WARNING "cciss%d: could not "
+                       "add disk 0.\n", h->ctlr);
+               return;
+       }
+       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];
+       cciss_add_disk(h, disk, drv_index);
+}
+
 /* 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
@@ -1379,15 +1885,12 @@ static int cciss_find_free_drive_index(int ctlr)
  * without a re-ordering of those drives.
  * INPUT
  * h           = The controller to perform the operations on
- * del_disk    = The disk to remove if specified.  If the value given
- *               is NULL then no disk is removed.
  */
-static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
+static int rebuild_lun_table(ctlr_info_t *h, int first_time)
 {
        int ctlr = h->ctlr;
        int num_luns;
        ReportLunData_struct *ld_buff = NULL;
-       drive_info_struct *drv = NULL;
        int return_code;
        int listlength = 0;
        int i;
@@ -1396,6 +1899,9 @@ static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
        __u32 lunid = 0;
        unsigned long flags;
 
+       if (!capable(CAP_SYS_RAWIO))
+               return -EPERM;
+
        /* Set busy_configuring flag for this operation */
        spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
        if (h->busy_configuring) {
@@ -1403,100 +1909,105 @@ static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
                return -EBUSY;
        }
        h->busy_configuring = 1;
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+
+       ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL);
+       if (ld_buff == NULL)
+               goto mem_msg;
+
+       return_code = sendcmd_withirq(CISS_REPORT_LOG, ctlr, ld_buff,
+                                     sizeof(ReportLunData_struct),
+                                     0, CTLR_LUNID, TYPE_CMD);
 
-       /* if del_disk is NULL then we are being called to add a new disk
-        * and update the logical drive table.  If it is not NULL then
-        * we will check if the disk is in use or not.
+       if (return_code == IO_OK)
+               listlength = be32_to_cpu(*(__be32 *) ld_buff->LUNListLength);
+       else {  /* reading number of logical volumes failed */
+               printk(KERN_WARNING "cciss: 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;
+               printk(KERN_WARNING "cciss: 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.
         */
-       if (del_disk != NULL) {
-               drv = get_drv(del_disk);
-               drv->busy_configuring = 1;
-               spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
-               return_code = deregister_disk(del_disk, drv, 1);
-               drv->busy_configuring = 0;
-               h->busy_configuring = 0;
-               return return_code;
-       } else {
-               spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
-               if (!capable(CAP_SYS_RAWIO))
-                       return -EPERM;
+       for (i = 0; i <= h->highest_lun; i++) {
+               int j;
+               drv_found = 0;
 
-               ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL);
-               if (ld_buff == NULL)
-                       goto mem_msg;
-
-               return_code = sendcmd_withirq(CISS_REPORT_LOG, ctlr, ld_buff,
-                                             sizeof(ReportLunData_struct), 0,
-                                             0, 0, TYPE_CMD);
-
-               if (return_code == IO_OK) {
-                       listlength =
-                               be32_to_cpu(*(__u32 *) ld_buff->LUNListLength);
-               } else {        /* reading number of logical volumes failed */
-                       printk(KERN_WARNING "cciss: report logical volume"
-                              " command failed\n");
-                       listlength = 0;
-                       goto freeret;
-               }
+               /* skip holes in the array from already deleted drives */
+               if (h->drv[i].raid_level == -1)
+                       continue;
 
-               num_luns = listlength / 8;      /* 8 bytes per entry */
-               if (num_luns > CISS_MAX_LUN) {
-                       num_luns = CISS_MAX_LUN;
-                       printk(KERN_WARNING "cciss: more luns configured"
-                              " on controller than can be handled by"
-                              " this driver.\n");
+               for (j = 0; j < num_luns; j++) {
+                       memcpy(&lunid, &ld_buff->LUN[j][0], 4);
+                       lunid = le32_to_cpu(lunid);
+                       if (h->drv[i].LunID == lunid) {
+                               drv_found = 1;
+                               break;
+                       }
+               }
+               if (!drv_found) {
+                       /* Deregister it from the OS, it's gone. */
+                       spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
+                       h->drv[i].busy_configuring = 1;
+                       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+                       return_code = deregister_disk(h, i, 1);
+                       cciss_destroy_ld_sysfs_entry(&h->drv[i]);
+                       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], 4);
+               lunid = le32_to_cpu(lunid);
 
-               /* Compare controller drive array to drivers drive array.
-                * Check for updates in the drive information and any new drives
-                * on the controller.
+               /* 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 (i = 0; i < num_luns; i++) {
-                       int j;
-
-                       drv_found = 0;
-
-                       lunid = (0xff &
-                                (unsigned int)(ld_buff->LUN[i][3])) << 24;
-                       lunid |= (0xff &
-                                 (unsigned int)(ld_buff->LUN[i][2])) << 16;
-                       lunid |= (0xff &
-                                 (unsigned int)(ld_buff->LUN[i][1])) << 8;
-                       lunid |= 0xff & (unsigned int)(ld_buff->LUN[i][0]);
-
-                       /* Find if the LUN is already in the drive array
-                        * of the controller.  If so then update its info
-                        * if not is 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].LunID == lunid) {
-                                       drv_index = j;
-                                       drv_found = 1;
-                               }
+               for (j = 0; j <= h->highest_lun; j++) {
+                       if (h->drv[j].raid_level != -1 &&
+                               h->drv[j].LunID == lunid) {
+                               drv_index = j;
+                               drv_found = 1;
+                               break;
                        }
+               }
 
-                       /* check if the drive was found already in the array */
-                       if (!drv_found) {
-                               drv_index = cciss_find_free_drive_index(ctlr);
-                               if (drv_index == -1)
-                                       goto freeret;
-
-                               /*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]){
-                                               printk(KERN_ERR "cciss: could not allocate new disk %d\n", drv_index);
-                                               goto mem_msg;
-                                       }
-                               }
-                       }
-                       h->drv[drv_index].LunID = lunid;
-                       cciss_update_drive_info(ctlr, drv_index);
-               }               /* end for */
-       }                       /* end else */
+               /* 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(ctlr, drv_index, first_time);
+       }               /* end for */
 
-      freeret:
+freeret:
        kfree(ld_buff);
        h->busy_configuring = 0;
        /* We return -1 here to tell the ACU that we have registered/updated
@@ -1504,8 +2015,9 @@ static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
         * additional times.
         */
        return -1;
-      mem_msg:
+mem_msg:
        printk(KERN_ERR "cciss: out of memory\n");
+       h->busy_configuring = 0;
        goto freeret;
 }
 
@@ -1524,15 +2036,19 @@ static int rebuild_lun_table(ctlr_info_t *h, struct gendisk *del_disk)
  *             the highest_lun should be left unchanged and the LunID
  *             should not be cleared.
 */
-static int deregister_disk(struct gendisk *disk, drive_info_struct *drv,
+static int deregister_disk(ctlr_info_t *h, int drv_index,
                           int clear_all)
 {
        int i;
-       ctlr_info_t *h = get_host(disk);
+       struct gendisk *disk;
+       drive_info_struct *drv;
 
        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 > 1)
@@ -1545,38 +2061,36 @@ static int deregister_disk(struct gendisk *disk, drive_info_struct *drv,
         * allows us to delete disk zero but keep the controller registered.
         */
        if (h->gendisk[0] != disk) {
-               if (disk) {
-                       request_queue_t *q = disk->queue;
-                       if (disk->flags & GENHD_FL_UP)
-                               del_gendisk(disk);
-                       if (q) {
-                               blk_cleanup_queue(q);
-                               /* Set drv->queue to NULL so that we do not try
-                                * to call blk_start_queue on this queue in the
-                                * interrupt handler
-                                */
-                               drv->queue = NULL;
-                       }
-                       /* 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;
-                                       }
+               struct request_queue *q = disk->queue;
+               if (disk->flags & GENHD_FL_UP)
+                       del_gendisk(disk);
+               if (q) {
+                       blk_cleanup_queue(q);
+                       /* Set drv->queue to NULL so that we do not try
+                        * to call blk_start_queue on this queue in the
+                        * interrupt handler
+                        */
+                       drv->queue = NULL;
+               }
+               /* 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);
                        }
+                       put_disk(disk);
                }
        } else {
                set_capacity(disk, 0);
@@ -1599,7 +2113,7 @@ static int deregister_disk(struct gendisk *disk, drive_info_struct *drv,
                if (drv == h->drv + h->highest_lun) {
                        /* if so, find the new hightest lun */
                        int i, newhighest = -1;
-                       for (i = 0; i < h->highest_lun; i++) {
+                       for (i = 0; i <= h->highest_lun; i++) {
                                /* if the disk has size > 0, it is available */
                                if (h->drv[i].heads)
                                        newhighest = i;
@@ -1612,11 +2126,9 @@ static int deregister_disk(struct gendisk *disk, drive_info_struct *drv,
        return 0;
 }
 
-static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_t size, unsigned int use_unit_num,     /* 0: address the controller,
-                                                                                                                          1: address logical volume log_unit,
-                                                                                                                          2: periph device address is scsi3addr */
-                   unsigned int log_unit, __u8 page_code,
-                   unsigned char *scsi3addr, int cmd_type)
+static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff,
+               size_t size, __u8 page_code, unsigned char *scsi3addr,
+               int cmd_type)
 {
        ctlr_info_t *h = hba[ctlr];
        u64bit buff_dma_handle;
@@ -1632,27 +2144,12 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
                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:
-                       /* If the logical unit number is 0 then, this is going
-                          to controller so It's a physical command
-                          mode = 0 target = 0.  So we have nothing to write.
-                          otherwise, if use_unit_num == 1,
-                          mode = 1(volume set addressing) target = LUNID
-                          otherwise, if use_unit_num == 2,
-                          mode = 0(periph dev addr) target = scsi3addr */
-                       if (use_unit_num == 1) {
-                               c->Header.LUN.LogDev.VolId =
-                                   h->drv[log_unit].LunID;
-                               c->Header.LUN.LogDev.Mode = 1;
-                       } else if (use_unit_num == 2) {
-                               memcpy(c->Header.LUN.LunAddrBytes, scsi3addr,
-                                      8);
-                               c->Header.LUN.LogDev.Mode = 0;
-                       }
                        /* are we trying to read a vital product page */
                        if (page_code != 0) {
                                c->Request.CDB[1] = 0x01;
@@ -1682,8 +2179,6 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
                        break;
 
                case CCISS_READ_CAPACITY:
-                       c->Header.LUN.LogDev.VolId = h->drv[log_unit].LunID;
-                       c->Header.LUN.LogDev.Mode = 1;
                        c->Request.CDBLen = 10;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
@@ -1691,8 +2186,6 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
                        c->Request.CDB[0] = cmd;
                        break;
                case CCISS_READ_CAPACITY_16:
-                       c->Header.LUN.LogDev.VolId = h->drv[log_unit].LunID;
-                       c->Header.LUN.LogDev.Mode = 1;
                        c->Request.CDBLen = 16;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
                        c->Request.Type.Direction = XFER_READ;
@@ -1714,6 +2207,12 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
                        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:
                        printk(KERN_WARNING
                               "cciss%d:  Unknown Command 0x%c\n", ctlr, cmd);
@@ -1732,13 +2231,13 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
                        memcpy(&c->Request.CDB[4], buff, 8);
                        break;
                case 1: /* RESET message */
-                       c->Request.CDBLen = 12;
+                       c->Request.CDBLen = 16;
                        c->Request.Type.Attribute = ATTR_SIMPLE;
-                       c->Request.Type.Direction = XFER_WRITE;
+                       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] = 0x04;       /* reset a LUN */
+                       c->Request.CDB[1] = 0x03;       /* reset a target */
                        break;
                case 3: /* No-Op message */
                        c->Request.CDBLen = 1;
@@ -1770,114 +2269,152 @@ static int fill_cmd(CommandList_struct *c, __u8 cmd, int ctlr, void *buff, size_
        return status;
 }
 
-static int sendcmd_withirq(__u8 cmd,
-                          int ctlr,
-                          void *buff,
-                          size_t size,
-                          unsigned int use_unit_num,
-                          unsigned int log_unit, __u8 page_code, int cmd_type)
+static int check_target_status(ctlr_info_t *h, CommandList_struct *c)
 {
-       ctlr_info_t *h = hba[ctlr];
-       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:
+                       printk(KERN_WARNING "cciss%d: cmd 0x%02x "
+                               "check condition, sense key = 0x%02x\n",
+                               h->ctlr, c->Request.CDB[0],
+                               c->err_info->SenseInfo[2]);
+               }
+               break;
+       default:
+               printk(KERN_WARNING "cciss%d: cmd 0x%02x"
+                       "scsi status = 0x%02x\n", h->ctlr,
+                       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:
+               printk(KERN_WARNING "cciss: cmd 0x%02x is "
+                      "reported invalid\n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_PROTOCOL_ERR:
+               printk(KERN_WARNING "cciss: cmd 0x%02x has "
+                      "protocol error \n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_HARDWARE_ERR:
+               printk(KERN_WARNING "cciss: cmd 0x%02x had "
+                      " hardware error\n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_CONNECTION_LOST:
+               printk(KERN_WARNING "cciss: cmd 0x%02x had "
+                      "connection lost\n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_ABORTED:
+               printk(KERN_WARNING "cciss: cmd 0x%02x was "
+                      "aborted\n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_ABORT_FAILED:
+               printk(KERN_WARNING "cciss: cmd 0x%02x reports "
+                      "abort failed\n", c->Request.CDB[0]);
+               return_status = IO_ERROR;
+               break;
+       case CMD_UNSOLICITED_ABORT:
+               printk(KERN_WARNING
+                      "cciss%d: unsolicited abort 0x%02x\n", h->ctlr,
+                       c->Request.CDB[0]);
+               return_status = IO_NEEDS_RETRY;
+               break;
+       default:
+               printk(KERN_WARNING "cciss: 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;
        unsigned long flags;
-       int return_status;
-       DECLARE_COMPLETION_ONSTACK(wait);
+       int return_status = IO_OK;
 
-       if ((c = cmd_alloc(h, 0)) == NULL)
-               return -ENOMEM;
-       return_status = fill_cmd(c, cmd, ctlr, buff, size, use_unit_num,
-                                log_unit, page_code, NULL, cmd_type);
-       if (return_status != IO_OK) {
-               cmd_free(h, c, 0);
-               return return_status;
-       }
-      resend_cmd2:
+resend_cmd2:
        c->waiting = &wait;
-
        /* Put the request on the tail of the queue and send it */
-       spin_lock_irqsave(CCISS_LOCK(ctlr), flags);
+       spin_lock_irqsave(CCISS_LOCK(h->ctlr), flags);
        addQ(&h->reqQ, c);
        h->Qdepth++;
        start_io(h);
-       spin_unlock_irqrestore(CCISS_LOCK(ctlr), flags);
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
 
        wait_for_completion(&wait);
 
-       if (c->err_info->CommandStatus != 0) {  /* an error has occurred */
-               switch (c->err_info->CommandStatus) {
-               case CMD_TARGET_STATUS:
-                       printk(KERN_WARNING "cciss: cmd %p has "
-                              " completed with errors\n", c);
-                       if (c->err_info->ScsiStatus) {
-                               printk(KERN_WARNING "cciss: cmd %p "
-                                      "has SCSI Status = %x\n",
-                                      c, c->err_info->ScsiStatus);
-                       }
+       if (c->err_info->CommandStatus == 0 || !attempt_retry)
+               goto command_done;
 
-                       break;
-               case CMD_DATA_UNDERRUN:
-               case CMD_DATA_OVERRUN:
-                       /* expected for inquire and report lun commands */
-                       break;
-               case CMD_INVALID:
-                       printk(KERN_WARNING "cciss: Cmd %p is "
-                              "reported invalid\n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_PROTOCOL_ERR:
-                       printk(KERN_WARNING "cciss: cmd %p has "
-                              "protocol error \n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_HARDWARE_ERR:
-                       printk(KERN_WARNING "cciss: cmd %p had "
-                              " hardware error\n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_CONNECTION_LOST:
-                       printk(KERN_WARNING "cciss: cmd %p had "
-                              "connection lost\n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_ABORTED:
-                       printk(KERN_WARNING "cciss: cmd %p was "
-                              "aborted\n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_ABORT_FAILED:
-                       printk(KERN_WARNING "cciss: cmd %p reports "
-                              "abort failed\n", c);
-                       return_status = IO_ERROR;
-                       break;
-               case CMD_UNSOLICITED_ABORT:
-                       printk(KERN_WARNING
-                              "cciss%d: unsolicited abort %p\n", ctlr, c);
-                       if (c->retry_count < MAX_CMD_RETRIES) {
-                               printk(KERN_WARNING
-                                      "cciss%d: retrying %p\n", ctlr, c);
-                               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;
-                       }
-                       return_status = IO_ERROR;
-                       break;
-               default:
-                       printk(KERN_WARNING "cciss: cmd %p returned "
-                              "unknown status %x\n", c,
-                              c->err_info->CommandStatus);
-                       return_status = IO_ERROR;
-               }
+       return_status = process_sendcmd_error(h, c);
+
+       if (return_status == IO_NEEDS_RETRY &&
+               c->retry_count < MAX_CMD_RETRIES) {
+               printk(KERN_WARNING "cciss%d: retrying 0x%02x\n", h->ctlr,
+                       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(__u8 cmd, int ctlr, void *buff, size_t size,
+                          __u8 page_code, unsigned char scsi3addr[],
+                       int cmd_type)
+{
+       ctlr_info_t *h = hba[ctlr];
+       CommandList_struct *c;
+       int return_status;
+
+       c = cmd_alloc(h, 0);
+       if (!c)
+               return -ENOMEM;
+       return_status = fill_cmd(c, cmd, ctlr, buff, size, page_code,
+               scsi3addr, cmd_type);
+       if (return_status == IO_OK)
+               return_status = sendcmd_withirq_core(h, c, 1);
+
        cmd_free(h, c, 0);
        return return_status;
 }
@@ -1890,15 +2427,17 @@ static void cciss_geometry_inquiry(int ctlr, int logvol,
 {
        int return_code;
        unsigned long t;
+       unsigned char scsi3addr[8];
 
        memset(inq_buff, 0, sizeof(InquiryData_struct));
+       log_unit_to_scsi3addr(hba[ctlr], scsi3addr, logvol);
        if (withirq)
                return_code = sendcmd_withirq(CISS_INQUIRY, ctlr,
-                                             inq_buff, sizeof(*inq_buff), 1,
-                                             logvol, 0xC1, TYPE_CMD);
+                                             inq_buff, sizeof(*inq_buff),
+                                             0xC1, scsi3addr, TYPE_CMD);
        else
                return_code = sendcmd(CISS_INQUIRY, ctlr, inq_buff,
-                                     sizeof(*inq_buff), 1, logvol, 0xC1, NULL,
+                                     sizeof(*inq_buff), 0xC1, scsi3addr,
                                      TYPE_CMD);
        if (return_code == IO_OK) {
                if (inq_buff->data_byte[8] == 0xFF) {
@@ -1907,6 +2446,8 @@ static void cciss_geometry_inquiry(int ctlr, int logvol,
                               "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];
@@ -1915,13 +2456,14 @@ static void cciss_geometry_inquiry(int ctlr, int logvol,
                        drv->raid_level = inq_buff->data_byte[8];
                }
                drv->block_size = block_size;
-               drv->nr_blocks = total_size;
+               drv->nr_blocks = total_size + 1;
                t = drv->heads * drv->sectors;
                if (t > 1) {
-                       unsigned rem = sector_div(total_size, t);
+                       sector_t real_size = total_size + 1;
+                       unsigned long rem = sector_div(real_size, t);
                        if (rem)
-                               total_size++;
-                       drv->cylinders = total_size;
+                               real_size++;
+                       drv->cylinders = real_size;
                }
        } else {                /* Get geometry failed */
                printk(KERN_WARNING "cciss: reading geometry failed\n");
@@ -1936,33 +2478,35 @@ cciss_read_capacity(int ctlr, int logvol, int withirq, sector_t *total_size,
 {
        ReadCapdata_struct *buf;
        int return_code;
-       buf = kmalloc(sizeof(ReadCapdata_struct), GFP_KERNEL);
-       if (buf == NULL) {
+       unsigned char scsi3addr[8];
+
+       buf = kzalloc(sizeof(ReadCapdata_struct), GFP_KERNEL);
+       if (!buf) {
                printk(KERN_WARNING "cciss: out of memory\n");
                return;
        }
-       memset(buf, 0, sizeof(ReadCapdata_struct));
+
+       log_unit_to_scsi3addr(hba[ctlr], scsi3addr, logvol);
        if (withirq)
                return_code = sendcmd_withirq(CCISS_READ_CAPACITY,
                                ctlr, buf, sizeof(ReadCapdata_struct),
-                                       1, logvol, 0, TYPE_CMD);
+                                       0, scsi3addr, TYPE_CMD);
        else
                return_code = sendcmd(CCISS_READ_CAPACITY,
                                ctlr, buf, sizeof(ReadCapdata_struct),
-                                       1, logvol, 0, NULL, TYPE_CMD);
+                                       0, scsi3addr, TYPE_CMD);
        if (return_code == IO_OK) {
-               *total_size = be32_to_cpu(*(__u32 *) buf->total_size)+1;
-               *block_size = be32_to_cpu(*(__u32 *) buf->block_size);
+               *total_size = be32_to_cpu(*(__be32 *) buf->total_size);
+               *block_size = be32_to_cpu(*(__be32 *) buf->block_size);
        } else {                /* read capacity command failed */
                printk(KERN_WARNING "cciss: read capacity failed\n");
                *total_size = 0;
                *block_size = BLOCK_SIZE;
        }
-       if (*total_size != (__u32) 0)
+       if (*total_size != 0)
                printk(KERN_INFO "      blocks= %llu block_size= %d\n",
-               (unsigned long long)*total_size, *block_size);
+               (unsigned long long)*total_size+1, *block_size);
        kfree(buf);
-       return;
 }
 
 static void
@@ -1970,34 +2514,36 @@ cciss_read_capacity_16(int ctlr, int logvol, int withirq, sector_t *total_size,
 {
        ReadCapdata_struct_16 *buf;
        int return_code;
-       buf = kmalloc(sizeof(ReadCapdata_struct_16), GFP_KERNEL);
-       if (buf == NULL) {
+       unsigned char scsi3addr[8];
+
+       buf = kzalloc(sizeof(ReadCapdata_struct_16), GFP_KERNEL);
+       if (!buf) {
                printk(KERN_WARNING "cciss: out of memory\n");
                return;
        }
-       memset(buf, 0, sizeof(ReadCapdata_struct_16));
+
+       log_unit_to_scsi3addr(hba[ctlr], scsi3addr, logvol);
        if (withirq) {
                return_code = sendcmd_withirq(CCISS_READ_CAPACITY_16,
                        ctlr, buf, sizeof(ReadCapdata_struct_16),
-                               1, logvol, 0, TYPE_CMD);
+                               0, scsi3addr, TYPE_CMD);
        }
        else {
                return_code = sendcmd(CCISS_READ_CAPACITY_16,
                        ctlr, buf, sizeof(ReadCapdata_struct_16),
-                               1, logvol, 0, NULL, TYPE_CMD);
+                               0, scsi3addr, TYPE_CMD);
        }
        if (return_code == IO_OK) {
-               *total_size = be64_to_cpu(*(__u64 *) buf->total_size)+1;
-               *block_size = be32_to_cpu(*(__u32 *) buf->block_size);
+               *total_size = be64_to_cpu(*(__be64 *) buf->total_size);
+               *block_size = be32_to_cpu(*(__be32 *) buf->block_size);
        } else {                /* read capacity command failed */
                printk(KERN_WARNING "cciss: read capacity failed\n");
                *total_size = 0;
                *block_size = BLOCK_SIZE;
        }
        printk(KERN_INFO "      blocks= %llu block_size= %d\n",
-              (unsigned long long)*total_size, *block_size);
+              (unsigned long long)*total_size+1, *block_size);
        kfree(buf);
-       return;
 }
 
 static int cciss_revalidate(struct gendisk *disk)
@@ -2035,7 +2581,7 @@ static int cciss_revalidate(struct gendisk *disk)
        cciss_geometry_inquiry(h->ctlr, logvol, 1, total_size, block_size,
                               inq_buff, drv);
 
-       blk_queue_hardsect_size(drv->queue, drv->block_size);
+       blk_queue_logical_block_size(drv->queue, drv->block_size);
        set_capacity(disk, drv->nr_blocks);
 
        kfree(inq_buff);
@@ -2065,86 +2611,21 @@ static unsigned long pollcomplete(int ctlr)
        return 1;
 }
 
-static int add_sendcmd_reject(__u8 cmd, int ctlr, unsigned long complete)
-{
-       /* We get in here if sendcmd() is polling for completions
-          and gets some command back that it wasn't expecting --
-          something other than that which it just sent down.
-          Ordinarily, that shouldn't happen, but it can happen when
-          the scsi tape stuff gets into error handling mode, and
-          starts using sendcmd() to try to abort commands and
-          reset tape drives.  In that case, sendcmd may pick up
-          completions of commands that were sent to logical drives
-          through the block i/o system, or cciss ioctls completing, etc.
-          In that case, we need to save those completions for later
-          processing by the interrupt handler.
-        */
-
-#ifdef CONFIG_CISS_SCSI_TAPE
-       struct sendcmd_reject_list *srl = &hba[ctlr]->scsi_rejects;
-
-       /* If it's not the scsi tape stuff doing error handling, (abort */
-       /* or reset) then we don't expect anything weird. */
-       if (cmd != CCISS_RESET_MSG && cmd != CCISS_ABORT_MSG) {
-#endif
-               printk(KERN_WARNING "cciss cciss%d: SendCmd "
-                      "Invalid command list address returned! (%lx)\n",
-                      ctlr, complete);
-               /* not much we can do. */
-#ifdef CONFIG_CISS_SCSI_TAPE
-               return 1;
-       }
-
-       /* We've sent down an abort or reset, but something else
-          has completed */
-       if (srl->ncompletions >= (hba[ctlr]->nr_cmds + 2)) {
-               /* Uh oh.  No room to save it for later... */
-               printk(KERN_WARNING "cciss%d: Sendcmd: Invalid command addr, "
-                      "reject list overflow, command lost!\n", ctlr);
-               return 1;
-       }
-       /* Save it for later */
-       srl->complete[srl->ncompletions] = complete;
-       srl->ncompletions++;
-#endif
-       return 0;
-}
-
-/*
- * Send a command to the controller, and wait for it to complete.
- * Only used at init time.
+/* Send command c to controller h and poll for it to complete.
+ * Turns interrupts off on the board.  Used at driver init time
+ * and during SCSI error recovery.
  */
-static int sendcmd(__u8 cmd, int ctlr, void *buff, size_t size, unsigned int use_unit_num,     /* 0: address the controller,
-                                                                                                  1: address logical volume log_unit,
-                                                                                                  2: periph device address is scsi3addr */
-                  unsigned int log_unit,
-                  __u8 page_code, unsigned char *scsi3addr, int cmd_type)
+static int sendcmd_core(ctlr_info_t *h, CommandList_struct *c)
 {
-       CommandList_struct *c;
        int i;
        unsigned long complete;
-       ctlr_info_t *info_p = hba[ctlr];
+       int status = IO_ERROR;
        u64bit buff_dma_handle;
-       int status, done = 0;
 
-       if ((c = cmd_alloc(info_p, 1)) == NULL) {
-               printk(KERN_WARNING "cciss: unable to get memory");
-               return IO_ERROR;
-       }
-       status = fill_cmd(c, cmd, ctlr, buff, size, use_unit_num,
-                         log_unit, page_code, scsi3addr, cmd_type);
-       if (status != IO_OK) {
-               cmd_free(info_p, c, 1);
-               return status;
-       }
-      resend_cmd1:
-       /*
-        * Disable interrupt
-        */
-#ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "cciss: turning intr off\n");
-#endif                         /* CCISS_DEBUG */
-       info_p->access.set_intr_mask(info_p, CCISS_INTR_OFF);
+resend_cmd1:
+
+       /* Disable interrupt on the board. */
+       h->access.set_intr_mask(h, CCISS_INTR_OFF);
 
        /* Make sure there is room in the command FIFO */
        /* Actually it should be completely empty at this time */
@@ -2152,21 +2633,15 @@ static int sendcmd(__u8 cmd, int ctlr, void *buff, size_t size, unsigned int use
        /* tape side of the driver. */
        for (i = 200000; i > 0; i--) {
                /* if fifo isn't full go */
-               if (!(info_p->access.fifo_full(info_p))) {
-
+               if (!(h->access.fifo_full(h)))
                        break;
-               }
                udelay(10);
                printk(KERN_WARNING "cciss cciss%d: SendCmd FIFO full,"
-                      " waiting!\n", ctlr);
+                      " waiting!\n", h->ctlr);
        }
-       /*
-        * Send the cmd
-        */
-       info_p->access.submit_command(info_p, c);
-       done = 0;
+       h->access.submit_command(h, c); /* Send the cmd */
        do {
-               complete = pollcomplete(ctlr);
+               complete = pollcomplete(h->ctlr);
 
 #ifdef CCISS_DEBUG
                printk(KERN_DEBUG "cciss: command completed\n");
@@ -2175,97 +2650,102 @@ static int sendcmd(__u8 cmd, int ctlr, void *buff, size_t size, unsigned int use
                if (complete == 1) {
                        printk(KERN_WARNING
                               "cciss cciss%d: SendCmd Timeout out, "
-                              "No command list address returned!\n", ctlr);
+                              "No command list address returned!\n", h->ctlr);
                        status = IO_ERROR;
-                       done = 1;
                        break;
                }
 
-               /* This will need to change for direct lookup completions */
-               if ((complete & CISS_ERROR_BIT)
-                   && (complete & ~CISS_ERROR_BIT) == c->busaddr) {
-                       /* if data overrun or underun on Report command
-                          ignore it
-                        */
-                       if (((c->Request.CDB[0] == CISS_REPORT_LOG) ||
-                            (c->Request.CDB[0] == CISS_REPORT_PHYS) ||
-                            (c->Request.CDB[0] == CISS_INQUIRY)) &&
-                           ((c->err_info->CommandStatus ==
-                             CMD_DATA_OVERRUN) ||
-                            (c->err_info->CommandStatus == CMD_DATA_UNDERRUN)
-                           )) {
-                               complete = c->busaddr;
-                       } else {
-                               if (c->err_info->CommandStatus ==
-                                   CMD_UNSOLICITED_ABORT) {
-                                       printk(KERN_WARNING "cciss%d: "
-                                              "unsolicited abort %p\n",
-                                              ctlr, c);
-                                       if (c->retry_count < MAX_CMD_RETRIES) {
-                                               printk(KERN_WARNING
-                                                      "cciss%d: retrying %p\n",
-                                                      ctlr, c);
-                                               c->retry_count++;
-                                               /* erase the old error */
-                                               /* information */
-                                               memset(c->err_info, 0,
-                                                      sizeof
-                                                      (ErrorInfo_struct));
-                                               goto resend_cmd1;
-                                       } else {
-                                               printk(KERN_WARNING
-                                                      "cciss%d: retried %p too "
-                                                      "many times\n", ctlr, c);
-                                               status = IO_ERROR;
-                                               goto cleanup1;
-                                       }
-                               } else if (c->err_info->CommandStatus ==
-                                          CMD_UNABORTABLE) {
-                                       printk(KERN_WARNING
-                                              "cciss%d: command could not be aborted.\n",
-                                              ctlr);
-                                       status = IO_ERROR;
-                                       goto cleanup1;
-                               }
-                               printk(KERN_WARNING "ciss ciss%d: sendcmd"
-                                      " Error %x \n", ctlr,
-                                      c->err_info->CommandStatus);
-                               printk(KERN_WARNING "ciss ciss%d: sendcmd"
-                                      " offensive info\n"
-                                      "  size %x\n   num %x   value %x\n",
-                                      ctlr,
-                                      c->err_info->MoreErrInfo.Invalid_Cmd.
-                                      offense_size,
-                                      c->err_info->MoreErrInfo.Invalid_Cmd.
-                                      offense_num,
-                                      c->err_info->MoreErrInfo.Invalid_Cmd.
-                                      offense_value);
-                               status = IO_ERROR;
-                               goto cleanup1;
-                       }
+               /* Make sure it's the command we're expecting. */
+               if ((complete & ~CISS_ERROR_BIT) != c->busaddr) {
+                       printk(KERN_WARNING "cciss%d: Unexpected command "
+                               "completion.\n", h->ctlr);
+                       continue;
                }
-               /* This will need changing for direct lookup completions */
-               if (complete != c->busaddr) {
-                       if (add_sendcmd_reject(cmd, ctlr, complete) != 0) {
-                               BUG();  /* we are pretty much hosed if we get here. */
+
+               /* It is our command.  If no error, we're done. */
+               if (!(complete & CISS_ERROR_BIT)) {
+                       status = IO_OK;
+                       break;
+               }
+
+               /* There is an error... */
+
+               /* if data overrun or underun on Report command ignore it */
+               if (((c->Request.CDB[0] == CISS_REPORT_LOG) ||
+                    (c->Request.CDB[0] == CISS_REPORT_PHYS) ||
+                    (c->Request.CDB[0] == CISS_INQUIRY)) &&
+                       ((c->err_info->CommandStatus == CMD_DATA_OVERRUN) ||
+                        (c->err_info->CommandStatus == CMD_DATA_UNDERRUN))) {
+                       complete = c->busaddr;
+                       status = IO_OK;
+                       break;
+               }
+
+               if (c->err_info->CommandStatus == CMD_UNSOLICITED_ABORT) {
+                       printk(KERN_WARNING "cciss%d: unsolicited abort %p\n",
+                               h->ctlr, c);
+                       if (c->retry_count < MAX_CMD_RETRIES) {
+                               printk(KERN_WARNING "cciss%d: retrying %p\n",
+                                  h->ctlr, c);
+                               c->retry_count++;
+                               /* erase the old error information */
+                               memset(c->err_info, 0, sizeof(c->err_info));
+                               goto resend_cmd1;
                        }
-                       continue;
-               } else
-                       done = 1;
-       } while (!done);
+                       printk(KERN_WARNING "cciss%d: retried %p too many "
+                               "times\n", h->ctlr, c);
+                       status = IO_ERROR;
+                       break;
+               }
+
+               if (c->err_info->CommandStatus == CMD_UNABORTABLE) {
+                       printk(KERN_WARNING "cciss%d: command could not be "
+                               "aborted.\n", h->ctlr);
+                       status = IO_ERROR;
+                       break;
+               }
+
+               if (c->err_info->CommandStatus == CMD_TARGET_STATUS) {
+                       status = check_target_status(h, c);
+                       break;
+               }
+
+               printk(KERN_WARNING "cciss%d: sendcmd error\n", h->ctlr);
+               printk(KERN_WARNING "cmd = 0x%02x, CommandStatus = 0x%02x\n",
+                       c->Request.CDB[0], c->err_info->CommandStatus);
+               status = IO_ERROR;
+               break;
+
+       } while (1);
 
-      cleanup1:
        /* unlock the data buffer 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(info_p->pdev, (dma_addr_t) buff_dma_handle.val,
+       pci_unmap_single(h->pdev, (dma_addr_t) buff_dma_handle.val,
                         c->SG[0].Len, PCI_DMA_BIDIRECTIONAL);
-#ifdef CONFIG_CISS_SCSI_TAPE
-       /* if we saved some commands for later, process them now. */
-       if (info_p->scsi_rejects.ncompletions > 0)
-               do_cciss_intr(0, info_p);
-#endif
-       cmd_free(info_p, c, 1);
+       return status;
+}
+
+/*
+ * Send a command to the controller, and wait for it to complete.
+ * Used at init time, and during SCSI error recovery.
+ */
+static int sendcmd(__u8 cmd, int ctlr, void *buff, size_t size,
+       __u8 page_code, unsigned char *scsi3addr, int cmd_type)
+{
+       CommandList_struct *c;
+       int status;
+
+       c = cmd_alloc(hba[ctlr], 1);
+       if (!c) {
+               printk(KERN_WARNING "cciss: unable to get memory");
+               return IO_ERROR;
+       }
+       status = fill_cmd(c, cmd, ctlr, buff, size, page_code,
+               scsi3addr, cmd_type);
+       if (status == IO_OK)
+               status = sendcmd_core(hba[ctlr], c);
+       cmd_free(hba[ctlr], c, 1);
        return status;
 }
 
@@ -2289,7 +2769,8 @@ static void start_io(ctlr_info_t *h)
 {
        CommandList_struct *c;
 
-       while ((c = h->reqQ) != NULL) {
+       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))) {
                        printk(KERN_WARNING "cciss: fifo full\n");
@@ -2297,14 +2778,14 @@ static void start_io(ctlr_info_t *h)
                }
 
                /* Get the first entry from the Request Q */
-               removeQ(&(h->reqQ), c);
+               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);
+               addQ(&h->cmpQ, c);
        }
 }
 
@@ -2317,7 +2798,7 @@ static inline void resend_cciss_cmd(ctlr_info_t *h, CommandList_struct *c)
        memset(c->err_info, 0, sizeof(ErrorInfo_struct));
 
        /* add it to software queue and then send it to the controller */
-       addQ(&(h->reqQ), c);
+       addQ(&h->reqQ, c);
        h->Qdepth++;
        if (h->Qdepth > h->maxQsinceinit)
                h->maxQsinceinit = h->Qdepth;
@@ -2325,6 +2806,76 @@ static inline void resend_cciss_cmd(ctlr_info_t *h, CommandList_struct *c)
        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 (blk_pc_request(cmd->rq))
+               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 (!blk_pc_request(cmd->rq))
+                       printk(KERN_WARNING "cciss: 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)) && !blk_pc_request(cmd->rq))
+               error_value = 0;
+
+       if (check_for_unit_attention(h, cmd)) {
+               *retry_cmd = !blk_pc_request(cmd->rq);
+               return 0;
+       }
+
+       if (!blk_pc_request(cmd->rq)) { /* Not SG_IO or similar? */
+               if (error_value != 0)
+                       printk(KERN_WARNING "cciss: 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
  * buffers for the completed job. Note that this function does not need
  * to hold the hba/queue lock.
@@ -2332,117 +2883,123 @@ static inline void resend_cciss_cmd(ctlr_info_t *h, CommandList_struct *c)
 static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
                                    int timeout)
 {
-       int status = 1;
        int retry_cmd = 0;
+       struct request *rq = cmd->rq;
+
+       rq->errors = 0;
 
        if (timeout)
-               status = 0;
+               rq->errors = make_status_bytes(0, 0, 0, DRIVER_TIMEOUT);
 
-       if (cmd->err_info->CommandStatus != 0) {        /* an error has occurred */
-               switch (cmd->err_info->CommandStatus) {
-                       unsigned char sense_key;
-               case CMD_TARGET_STATUS:
-                       status = 0;
+       if (cmd->err_info->CommandStatus == 0)  /* no error has occurred */
+               goto after_error_processing;
 
-                       if (cmd->err_info->ScsiStatus == 0x02) {
-                               printk(KERN_WARNING "cciss: cmd %p "
-                                      "has CHECK CONDITION "
-                                      " byte 2 = 0x%x\n", cmd,
-                                      cmd->err_info->SenseInfo[2]
-                                   );
-                               /* check the sense key */
-                               sense_key = 0xf & cmd->err_info->SenseInfo[2];
-                               /* no status or recovered error */
-                               if ((sense_key == 0x0) || (sense_key == 0x1)) {
-                                       status = 1;
-                               }
-                       } else {
-                               printk(KERN_WARNING "cciss: cmd %p "
-                                      "has SCSI Status 0x%x\n",
-                                      cmd, cmd->err_info->ScsiStatus);
-                       }
-                       break;
-               case CMD_DATA_UNDERRUN:
+       switch (cmd->err_info->CommandStatus) {
+       case CMD_TARGET_STATUS:
+               rq->errors = evaluate_target_status(h, cmd, &retry_cmd);
+               break;
+       case CMD_DATA_UNDERRUN:
+               if (blk_fs_request(cmd->rq)) {
                        printk(KERN_WARNING "cciss: cmd %p has"
                               " completed with data underrun "
                               "reported\n", cmd);
-                       break;
-               case CMD_DATA_OVERRUN:
+                       cmd->rq->resid_len = cmd->err_info->ResidualCnt;
+               }
+               break;
+       case CMD_DATA_OVERRUN:
+               if (blk_fs_request(cmd->rq))
                        printk(KERN_WARNING "cciss: cmd %p has"
                               " completed with data overrun "
                               "reported\n", cmd);
-                       break;
-               case CMD_INVALID:
-                       printk(KERN_WARNING "cciss: cmd %p is "
-                              "reported invalid\n", cmd);
-                       status = 0;
-                       break;
-               case CMD_PROTOCOL_ERR:
-                       printk(KERN_WARNING "cciss: cmd %p has "
-                              "protocol error \n", cmd);
-                       status = 0;
-                       break;
-               case CMD_HARDWARE_ERR:
-                       printk(KERN_WARNING "cciss: cmd %p had "
-                              " hardware error\n", cmd);
-                       status = 0;
-                       break;
-               case CMD_CONNECTION_LOST:
-                       printk(KERN_WARNING "cciss: cmd %p had "
-                              "connection lost\n", cmd);
-                       status = 0;
-                       break;
-               case CMD_ABORTED:
-                       printk(KERN_WARNING "cciss: cmd %p was "
-                              "aborted\n", cmd);
-                       status = 0;
-                       break;
-               case CMD_ABORT_FAILED:
-                       printk(KERN_WARNING "cciss: cmd %p reports "
-                              "abort failed\n", cmd);
-                       status = 0;
-                       break;
-               case CMD_UNSOLICITED_ABORT:
-                       printk(KERN_WARNING "cciss%d: unsolicited "
-                              "abort %p\n", h->ctlr, cmd);
-                       if (cmd->retry_count < MAX_CMD_RETRIES) {
-                               retry_cmd = 1;
-                               printk(KERN_WARNING
-                                      "cciss%d: retrying %p\n", h->ctlr, cmd);
-                               cmd->retry_count++;
-                       } else
-                               printk(KERN_WARNING
-                                      "cciss%d: %p retried too "
-                                      "many times\n", h->ctlr, cmd);
-                       status = 0;
-                       break;
-               case CMD_TIMEOUT:
-                       printk(KERN_WARNING "cciss: cmd %p timedout\n", cmd);
-                       status = 0;
-                       break;
-               default:
-                       printk(KERN_WARNING "cciss: cmd %p returned "
-                              "unknown status %x\n", cmd,
-                              cmd->err_info->CommandStatus);
-                       status = 0;
-               }
+               break;
+       case CMD_INVALID:
+               printk(KERN_WARNING "cciss: cmd %p is "
+                      "reported invalid\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       case CMD_PROTOCOL_ERR:
+               printk(KERN_WARNING "cciss: cmd %p has "
+                      "protocol error \n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       case CMD_HARDWARE_ERR:
+               printk(KERN_WARNING "cciss: cmd %p had "
+                      " hardware error\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       case CMD_CONNECTION_LOST:
+               printk(KERN_WARNING "cciss: cmd %p had "
+                      "connection lost\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       case CMD_ABORTED:
+               printk(KERN_WARNING "cciss: cmd %p was "
+                      "aborted\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ABORT);
+               break;
+       case CMD_ABORT_FAILED:
+               printk(KERN_WARNING "cciss: cmd %p reports "
+                      "abort failed\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       case CMD_UNSOLICITED_ABORT:
+               printk(KERN_WARNING "cciss%d: unsolicited "
+                      "abort %p\n", h->ctlr, cmd);
+               if (cmd->retry_count < MAX_CMD_RETRIES) {
+                       retry_cmd = 1;
+                       printk(KERN_WARNING
+                              "cciss%d: retrying %p\n", h->ctlr, cmd);
+                       cmd->retry_count++;
+               } else
+                       printk(KERN_WARNING
+                              "cciss%d: %p retried too "
+                              "many times\n", h->ctlr, cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ABORT);
+               break;
+       case CMD_TIMEOUT:
+               printk(KERN_WARNING "cciss: cmd %p timedout\n", cmd);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
+               break;
+       default:
+               printk(KERN_WARNING "cciss: cmd %p returned "
+                      "unknown status %x\n", cmd,
+                      cmd->err_info->CommandStatus);
+               rq->errors = make_status_bytes(SAM_STAT_GOOD,
+                       cmd->err_info->CommandStatus, DRIVER_OK,
+                       blk_pc_request(cmd->rq) ? DID_PASSTHROUGH : DID_ERROR);
        }
+
+after_error_processing:
+
        /* We need to return this command */
        if (retry_cmd) {
                resend_cciss_cmd(h, cmd);
                return;
        }
-
        cmd->rq->completion_data = cmd;
-       cmd->rq->errors = status;
-       blk_add_trace_rq(cmd->rq->q, cmd->rq, BLK_TA_COMPLETE);
        blk_complete_request(cmd->rq);
 }
 
 /*
  * Get a request and submit it to the controller.
  */
-static void do_cciss_request(request_queue_t *q)
+static void do_cciss_request(struct request_queue *q)
 {
        ctlr_info_t *h = q->queuedata;
        CommandList_struct *c;
@@ -2461,7 +3018,7 @@ static void do_cciss_request(request_queue_t *q)
                goto startio;
 
       queue:
-       creq = elv_next_request(q);
+       creq = blk_peek_request(q);
        if (!creq)
                goto startio;
 
@@ -2470,7 +3027,7 @@ static void do_cciss_request(request_queue_t *q)
        if ((c = cmd_alloc(h, 1)) == NULL)
                goto full;
 
-       blkdev_dequeue_request(creq);
+       blk_start_request(creq);
 
        spin_unlock_irq(q->queue_lock);
 
@@ -2491,16 +3048,17 @@ static void do_cciss_request(request_queue_t *q)
        c->Request.Type.Type = TYPE_CMD;        // It is a command.
        c->Request.Type.Attribute = ATTR_SIMPLE;
        c->Request.Type.Direction =
-           (rq_data_dir(creq) == READ) ? h->cciss_read : h->cciss_write;
+           (rq_data_dir(creq) == READ) ? XFER_READ : XFER_WRITE;
        c->Request.Timeout = 0; // Don't time out
        c->Request.CDB[0] =
            (rq_data_dir(creq) == READ) ? h->cciss_read : h->cciss_write;
-       start_blk = creq->sector;
+       start_blk = blk_rq_pos(creq);
 #ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "ciss: sector =%d nr_sectors=%d\n", (int)creq->sector,
-              (int)creq->nr_sectors);
+       printk(KERN_DEBUG "ciss: sector =%d nr_sectors=%d\n",
+              (int)blk_rq_pos(creq), (int)blk_rq_sectors(creq));
 #endif                         /* CCISS_DEBUG */
 
+       sg_init_table(tmp_sg, MAXSGENTRIES);
        seg = blk_rq_map_sg(q, creq, tmp_sg);
 
        /* get the DMA records for the setup */
@@ -2511,7 +3069,7 @@ static void do_cciss_request(request_queue_t *q)
 
        for (i = 0; i < seg; i++) {
                c->SG[i].Len = tmp_sg[i].length;
-               temp64.val = (__u64) pci_map_page(h->pdev, tmp_sg[i].page,
+               temp64.val = (__u64) pci_map_page(h->pdev, sg_page(&tmp_sg[i]),
                                                  tmp_sg[i].offset,
                                                  tmp_sg[i].length, dir);
                c->SG[i].Addr.lower = temp64.val32.lower;
@@ -2523,42 +3081,52 @@ static void do_cciss_request(request_queue_t *q)
                h->maxSG = seg;
 
 #ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "cciss: Submitting %d sectors in %d segments\n",
-              creq->nr_sectors, seg);
+       printk(KERN_DEBUG "cciss: Submitting %u sectors in %d segments\n",
+              blk_rq_sectors(creq), seg);
 #endif                         /* CCISS_DEBUG */
 
        c->Header.SGList = c->Header.SGTotal = seg;
-       if(h->cciss_read == CCISS_READ_10) {
-               c->Request.CDB[1] = 0;
-               c->Request.CDB[2] = (start_blk >> 24) & 0xff;   //MSB
-               c->Request.CDB[3] = (start_blk >> 16) & 0xff;
-               c->Request.CDB[4] = (start_blk >> 8) & 0xff;
-               c->Request.CDB[5] = start_blk & 0xff;
-               c->Request.CDB[6] = 0;  // (sect >> 24) & 0xff; MSB
-               c->Request.CDB[7] = (creq->nr_sectors >> 8) & 0xff;
-               c->Request.CDB[8] = creq->nr_sectors & 0xff;
-               c->Request.CDB[9] = c->Request.CDB[11] = c->Request.CDB[12] = 0;
+       if (likely(blk_fs_request(creq))) {
+               if(h->cciss_read == CCISS_READ_10) {
+                       c->Request.CDB[1] = 0;
+                       c->Request.CDB[2] = (start_blk >> 24) & 0xff;   //MSB
+                       c->Request.CDB[3] = (start_blk >> 16) & 0xff;
+                       c->Request.CDB[4] = (start_blk >> 8) & 0xff;
+                       c->Request.CDB[5] = start_blk & 0xff;
+                       c->Request.CDB[6] = 0;  // (sect >> 24) & 0xff; MSB
+                       c->Request.CDB[7] = (blk_rq_sectors(creq) >> 8) & 0xff;
+                       c->Request.CDB[8] = blk_rq_sectors(creq) & 0xff;
+                       c->Request.CDB[9] = c->Request.CDB[11] = c->Request.CDB[12] = 0;
+               } else {
+                       u32 upper32 = upper_32_bits(start_blk);
+
+                       c->Request.CDBLen = 16;
+                       c->Request.CDB[1]= 0;
+                       c->Request.CDB[2]= (upper32 >> 24) & 0xff;      //MSB
+                       c->Request.CDB[3]= (upper32 >> 16) & 0xff;
+                       c->Request.CDB[4]= (upper32 >>  8) & 0xff;
+                       c->Request.CDB[5]= upper32 & 0xff;
+                       c->Request.CDB[6]= (start_blk >> 24) & 0xff;
+                       c->Request.CDB[7]= (start_blk >> 16) & 0xff;
+                       c->Request.CDB[8]= (start_blk >>  8) & 0xff;
+                       c->Request.CDB[9]= start_blk & 0xff;
+                       c->Request.CDB[10]= (blk_rq_sectors(creq) >> 24) & 0xff;
+                       c->Request.CDB[11]= (blk_rq_sectors(creq) >> 16) & 0xff;
+                       c->Request.CDB[12]= (blk_rq_sectors(creq) >>  8) & 0xff;
+                       c->Request.CDB[13]= blk_rq_sectors(creq) & 0xff;
+                       c->Request.CDB[14] = c->Request.CDB[15] = 0;
+               }
+       } else if (blk_pc_request(creq)) {
+               c->Request.CDBLen = creq->cmd_len;
+               memcpy(c->Request.CDB, creq->cmd, BLK_MAX_CDB);
        } else {
-               c->Request.CDBLen = 16;
-               c->Request.CDB[1]= 0;
-               c->Request.CDB[2]= (start_blk >> 56) & 0xff;    //MSB
-               c->Request.CDB[3]= (start_blk >> 48) & 0xff;
-               c->Request.CDB[4]= (start_blk >> 40) & 0xff;
-               c->Request.CDB[5]= (start_blk >> 32) & 0xff;
-               c->Request.CDB[6]= (start_blk >> 24) & 0xff;
-               c->Request.CDB[7]= (start_blk >> 16) & 0xff;
-               c->Request.CDB[8]= (start_blk >>  8) & 0xff;
-               c->Request.CDB[9]= start_blk & 0xff;
-               c->Request.CDB[10]= (creq->nr_sectors >>  24) & 0xff;
-               c->Request.CDB[11]= (creq->nr_sectors >>  16) & 0xff;
-               c->Request.CDB[12]= (creq->nr_sectors >>  8) & 0xff;
-               c->Request.CDB[13]= creq->nr_sectors & 0xff;
-               c->Request.CDB[14] = c->Request.CDB[15] = 0;
+               printk(KERN_WARNING "cciss%d: bad request type %d\n", h->ctlr, creq->cmd_type);
+               BUG();
        }
 
        spin_lock_irq(q->queue_lock);
 
-       addQ(&(h->reqQ), c);
+       addQ(&h->reqQ, c);
        h->Qdepth++;
        if (h->Qdepth > h->maxQsinceinit)
                h->maxQsinceinit = h->Qdepth;
@@ -2575,44 +3143,18 @@ startio:
 
 static inline unsigned long get_next_completion(ctlr_info_t *h)
 {
-#ifdef CONFIG_CISS_SCSI_TAPE
-       /* Any rejects from sendcmd() lying around? Process them first */
-       if (h->scsi_rejects.ncompletions == 0)
-               return h->access.command_completed(h);
-       else {
-               struct sendcmd_reject_list *srl;
-               int n;
-               srl = &h->scsi_rejects;
-               n = --srl->ncompletions;
-               /* printk("cciss%d: processing saved reject\n", h->ctlr); */
-               printk("p");
-               return srl->complete[n];
-       }
-#else
        return h->access.command_completed(h);
-#endif
 }
 
 static inline int interrupt_pending(ctlr_info_t *h)
 {
-#ifdef CONFIG_CISS_SCSI_TAPE
-       return (h->access.intr_pending(h)
-               || (h->scsi_rejects.ncompletions > 0));
-#else
        return h->access.intr_pending(h);
-#endif
 }
 
 static inline long interrupt_not_for_us(ctlr_info_t *h)
 {
-#ifdef CONFIG_CISS_SCSI_TAPE
-       return (((h->access.intr_pending(h) == 0) ||
-                (h->interrupts_enabled == 0))
-               && (h->scsi_rejects.ncompletions == 0));
-#else
        return (((h->access.intr_pending(h) == 0) ||
                 (h->interrupts_enabled == 0)));
-#endif
 }
 
 static irqreturn_t do_cciss_intr(int irq, void *dev_id)
@@ -2646,16 +3188,12 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
                                a = c->busaddr;
 
                        } else {
+                               struct hlist_node *tmp;
+
                                a &= ~3;
-                               if ((c = h->cmpQ) == NULL) {
-                                       printk(KERN_WARNING
-                                              "cciss: Completion of %08x ignored\n",
-                                              a1);
-                                       continue;
-                               }
-                               while (c->busaddr != a) {
-                                       c = c->next;
-                                       if (c == h->cmpQ)
+                               c = NULL;
+                               hlist_for_each_entry(c, tmp, &h->cmpQ, list) {
+                                       if (c->busaddr == a)
                                                break;
                                }
                        }
@@ -2663,8 +3201,8 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
                         * If we've found the command, take it off the
                         * completion Q and free it
                         */
-                       if (c->busaddr == a) {
-                               removeQ(&h->cmpQ, c);
+                       if (c && c->busaddr == a) {
+                               removeQ(c);
                                if (c->cmd_type == CMD_RWREQ) {
                                        complete_command(h, c, 0);
                                } else if (c->cmd_type == CMD_IOCTL_PEND) {
@@ -2679,8 +3217,65 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
                }
        }
 
-       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
-       return IRQ_HANDLED;
+       spin_unlock_irqrestore(CCISS_LOCK(h->ctlr), flags);
+       return IRQ_HANDLED;
+}
+
+static int scan_thread(void *data)
+{
+       ctlr_info_t *h = data;
+       int rc;
+       DECLARE_COMPLETION_ONSTACK(wait);
+       h->rescan_wait = &wait;
+
+       for (;;) {
+               rc = wait_for_completion_interruptible(&wait);
+               if (kthread_should_stop())
+                       break;
+               if (!rc)
+                       rebuild_lun_table(h, 0);
+       }
+       return 0;
+}
+
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c)
+{
+       if (c->err_info->SenseInfo[2] != UNIT_ATTENTION)
+               return 0;
+
+       switch (c->err_info->SenseInfo[12]) {
+       case STATE_CHANGED:
+               printk(KERN_WARNING "cciss%d: a state change "
+                       "detected, command retried\n", h->ctlr);
+               return 1;
+       break;
+       case LUN_FAILED:
+               printk(KERN_WARNING "cciss%d: LUN failure "
+                       "detected, action required\n", h->ctlr);
+               return 1;
+       break;
+       case REPORT_LUNS_CHANGED:
+               printk(KERN_WARNING "cciss%d: report LUN data "
+                       "changed\n", h->ctlr);
+               if (h->rescan_wait)
+                       complete(h->rescan_wait);
+               return 1;
+       break;
+       case POWER_OR_RESET:
+               printk(KERN_WARNING "cciss%d: a power on "
+                       "or device reset detected\n", h->ctlr);
+               return 1;
+       break;
+       case UNIT_ATTENTION_CLEARED:
+               printk(KERN_WARNING "cciss%d: unit attention "
+                   "cleared by another initiator\n", h->ctlr);
+               return 1;
+       break;
+       default:
+               printk(KERN_WARNING "cciss%d: unknown "
+                       "unit attention detected\n", h->ctlr);
+                               return 1;
+       }
 }
 
 /*
@@ -2809,7 +3404,7 @@ default_int_mode:
        return;
 }
 
-static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
+static int __devinit cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 {
        ushort subsystem_vendor_id, subsystem_device_id, command;
        __u32 board_id, scratchpad = 0;
@@ -2856,14 +3451,23 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
  */
        cciss_interrupt_mode(c, pdev, board_id);
 
-       /*
-        * Memory base addr is first addr , the second points to the config
-        *   table
-        */
+       /* find the memory BAR */
+       for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
+               if (pci_resource_flags(pdev, i) & IORESOURCE_MEM)
+                       break;
+       }
+       if (i == DEVICE_COUNT_RESOURCE) {
+               printk(KERN_WARNING "cciss: No memory BAR found\n");
+               err = -ENODEV;
+               goto err_out_free_res;
+       }
+
+       c->paddr = pci_resource_start(pdev, i); /* addressing mode bits
+                                                * already removed
+                                                */
 
-       c->paddr = pci_resource_start(pdev, 0); /* addressing mode bits already removed */
 #ifdef CCISS_DEBUG
-       printk("address 0 = %x\n", c->paddr);
+       printk("address 0 = %lx\n", c->paddr);
 #endif                         /* CCISS_DEBUG */
        c->vaddr = remap_pci_mem(c->paddr, 0x250);
 
@@ -2890,7 +3494,8 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 #endif                         /* CCISS_DEBUG */
        cfg_base_addr_index = find_PCI_BAR_index(pdev, cfg_base_addr);
 #ifdef CCISS_DEBUG
-       printk("cfg base address index = %x\n", cfg_base_addr_index);
+       printk("cfg base address index = %llx\n",
+               (unsigned long long)cfg_base_addr_index);
 #endif                         /* CCISS_DEBUG */
        if (cfg_base_addr_index == -1) {
                printk(KERN_WARNING "cciss: Cannot find cfg_base_addr_index\n");
@@ -2900,7 +3505,7 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
 
        cfg_offset = readl(c->vaddr + SA5_CTMEM_OFFSET);
 #ifdef CCISS_DEBUG
-       printk("cfg offset = %x\n", cfg_offset);
+       printk("cfg offset = %llx\n", (unsigned long long)cfg_offset);
 #endif                         /* CCISS_DEBUG */
        c->cfgtable = remap_pci_mem(pci_resource_start(pdev,
                                                       cfg_base_addr_index) +
@@ -2911,11 +3516,20 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
        print_cfg_table(c->cfgtable);
 #endif                         /* CCISS_DEBUG */
 
+       /* Some controllers support Zero Memory Raid (ZMR).
+        * When configured in ZMR mode the number of supported
+        * commands drops to 64. So instead of just setting an
+        * arbitrary value we make the driver a little smarter.
+        * We read the config table to tell us how many commands
+        * are supported on the controller then subtract 4 to
+        * leave a little room for ioctl calls.
+        */
+       c->max_commands = readl(&(c->cfgtable->CmdsOutMax));
        for (i = 0; i < ARRAY_SIZE(products); i++) {
                if (board_id == products[i].board_id) {
                        c->product_name = products[i].product_name;
                        c->access = *(products[i].access);
-                       c->nr_cmds = products[i].nr_cmds;
+                       c->nr_cmds = c->max_commands - 4;
                        break;
                }
        }
@@ -2935,7 +3549,7 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
                if (subsystem_vendor_id == PCI_VENDOR_ID_HP) {
                        c->product_name = products[i-1].product_name;
                        c->access = *(products[i-1].access);
-                       c->nr_cmds = products[i-1].nr_cmds;
+                       c->nr_cmds = c->max_commands - 4;
                        printk(KERN_WARNING "cciss: This is an unknown "
                                "Smart Array controller.\n"
                                "cciss: Please update to the latest driver "
@@ -2958,15 +3572,20 @@ static int cciss_pci_init(ctlr_info_t *c, struct pci_dev *pdev)
        }
 #endif
 
-       /* Disabling DMA prefetch for the P600
-        * An ASIC bug may result in a prefetch beyond
-        * physical memory.
+       /* Disabling DMA prefetch and refetch for the P600.
+        * An ASIC bug may result in accesses to invalid memory addresses.
+        * We've disabled prefetch for some time now. Testing with XEN
+        * kernels revealed a bug in the refetch if dom0 resides on a P600.
         */
        if(board_id == 0x3225103C) {
                __u32 dma_prefetch;
+               __u32 dma_refetch;
                dma_prefetch = readl(c->vaddr + I2O_DMA1_CFG);
                dma_prefetch |= 0x8000;
                writel(dma_prefetch, c->vaddr + I2O_DMA1_CFG);
+               pci_read_config_dword(pdev, PCI_COMMAND_PARITY, &dma_refetch);
+               dma_refetch |= 0x1;
+               pci_write_config_dword(pdev, PCI_COMMAND_PARITY, dma_refetch);
        }
 
 #ifdef CCISS_DEBUG
@@ -3013,135 +3632,9 @@ err_out_free_res:
        return err;
 }
 
-/*
- * Gets information about the local volumes attached to the controller.
+/* Function to find the first free pointer into our hba[] array
+ * Returns -1 if no free entries are left.
  */
-static void cciss_getgeometry(int cntl_num)
-{
-       ReportLunData_struct *ld_buff;
-       InquiryData_struct *inq_buff;
-       int return_code;
-       int i;
-       int listlength = 0;
-       __u32 lunid = 0;
-       int block_size;
-       sector_t total_size;
-
-       ld_buff = kzalloc(sizeof(ReportLunData_struct), GFP_KERNEL);
-       if (ld_buff == NULL) {
-               printk(KERN_ERR "cciss: out of memory\n");
-               return;
-       }
-       inq_buff = kmalloc(sizeof(InquiryData_struct), GFP_KERNEL);
-       if (inq_buff == NULL) {
-               printk(KERN_ERR "cciss: out of memory\n");
-               kfree(ld_buff);
-               return;
-       }
-       /* Get the firmware version */
-       return_code = sendcmd(CISS_INQUIRY, cntl_num, inq_buff,
-                             sizeof(InquiryData_struct), 0, 0, 0, NULL,
-                             TYPE_CMD);
-       if (return_code == IO_OK) {
-               hba[cntl_num]->firm_ver[0] = inq_buff->data_byte[32];
-               hba[cntl_num]->firm_ver[1] = inq_buff->data_byte[33];
-               hba[cntl_num]->firm_ver[2] = inq_buff->data_byte[34];
-               hba[cntl_num]->firm_ver[3] = inq_buff->data_byte[35];
-       } else {                /* send command failed */
-
-               printk(KERN_WARNING "cciss: unable to determine firmware"
-                      " version of controller\n");
-       }
-       /* Get the number of logical volumes */
-       return_code = sendcmd(CISS_REPORT_LOG, cntl_num, ld_buff,
-                             sizeof(ReportLunData_struct), 0, 0, 0, NULL,
-                             TYPE_CMD);
-
-       if (return_code == IO_OK) {
-#ifdef CCISS_DEBUG
-               printk("LUN Data\n--------------------------\n");
-#endif                         /* CCISS_DEBUG */
-
-               listlength |=
-                   (0xff & (unsigned int)(ld_buff->LUNListLength[0])) << 24;
-               listlength |=
-                   (0xff & (unsigned int)(ld_buff->LUNListLength[1])) << 16;
-               listlength |=
-                   (0xff & (unsigned int)(ld_buff->LUNListLength[2])) << 8;
-               listlength |= 0xff & (unsigned int)(ld_buff->LUNListLength[3]);
-       } else {                /* reading number of logical volumes failed */
-
-               printk(KERN_WARNING "cciss: report logical volume"
-                      " command failed\n");
-               listlength = 0;
-       }
-       hba[cntl_num]->num_luns = listlength / 8;       // 8 bytes pre entry
-       if (hba[cntl_num]->num_luns > CISS_MAX_LUN) {
-               printk(KERN_ERR
-                      "ciss:  only %d number of logical volumes supported\n",
-                      CISS_MAX_LUN);
-               hba[cntl_num]->num_luns = CISS_MAX_LUN;
-       }
-#ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "Length = %x %x %x %x = %d\n",
-              ld_buff->LUNListLength[0], ld_buff->LUNListLength[1],
-              ld_buff->LUNListLength[2], ld_buff->LUNListLength[3],
-              hba[cntl_num]->num_luns);
-#endif                         /* CCISS_DEBUG */
-
-       hba[cntl_num]->highest_lun = hba[cntl_num]->num_luns - 1;
-       for (i = 0; i < CISS_MAX_LUN; i++) {
-               if (i < hba[cntl_num]->num_luns) {
-                       lunid = (0xff & (unsigned int)(ld_buff->LUN[i][3]))
-                           << 24;
-                       lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][2]))
-                           << 16;
-                       lunid |= (0xff & (unsigned int)(ld_buff->LUN[i][1]))
-                           << 8;
-                       lunid |= 0xff & (unsigned int)(ld_buff->LUN[i][0]);
-
-                       hba[cntl_num]->drv[i].LunID = lunid;
-
-#ifdef CCISS_DEBUG
-                       printk(KERN_DEBUG "LUN[%d]:  %x %x %x %x = %x\n", i,
-                              ld_buff->LUN[i][0], ld_buff->LUN[i][1],
-                              ld_buff->LUN[i][2], ld_buff->LUN[i][3],
-                              hba[cntl_num]->drv[i].LunID);
-#endif                         /* CCISS_DEBUG */
-
-               /* testing to see if 16-byte CDBs are already being used */
-               if(hba[cntl_num]->cciss_read == CCISS_READ_16) {
-                       cciss_read_capacity_16(cntl_num, i, 0,
-                                           &total_size, &block_size);
-                       goto geo_inq;
-               }
-               cciss_read_capacity(cntl_num, i, 0, &total_size, &block_size);
-
-               /* total_size = last LBA + 1 */
-               if(total_size == (__u32) 0) {
-                       cciss_read_capacity_16(cntl_num, i, 0,
-                       &total_size, &block_size);
-                       hba[cntl_num]->cciss_read = CCISS_READ_16;
-                       hba[cntl_num]->cciss_write = CCISS_WRITE_16;
-               } else {
-                       hba[cntl_num]->cciss_read = CCISS_READ_10;
-                       hba[cntl_num]->cciss_write = CCISS_WRITE_10;
-               }
-geo_inq:
-                       cciss_geometry_inquiry(cntl_num, i, 0, total_size,
-                                              block_size, inq_buff,
-                                              &hba[cntl_num]->drv[i]);
-               } else {
-                       /* initialize raid_level to indicate a free space */
-                       hba[cntl_num]->drv[i].raid_level = -1;
-               }
-       }
-       kfree(ld_buff);
-       kfree(inq_buff);
-}
-
-/* Function to find the first free pointer into our hba[] array */
-/* Returns -1 if no free entries are left.  */
 static int alloc_cciss_hba(void)
 {
        int i;
@@ -3149,12 +3642,10 @@ static int alloc_cciss_hba(void)
        for (i = 0; i < MAX_CTLR; i++) {
                if (!hba[i]) {
                        ctlr_info_t *p;
+
                        p = kzalloc(sizeof(ctlr_info_t), GFP_KERNEL);
                        if (!p)
                                goto Enomem;
-                       p->gendisk[0] = alloc_disk(1 << NWD_SHIFT);
-                       if (!p->gendisk[0])
-                               goto Enomem;
                        hba[i] = p;
                        return i;
                }
@@ -3178,6 +3669,203 @@ static void free_hba(int i)
        kfree(p);
 }
 
+/* Send a message CDB to the firmware. */
+static __devinit int cciss_message(struct pci_dev *pdev, unsigned char opcode, unsigned char type)
+{
+       typedef struct {
+               CommandListHeader_struct CommandHeader;
+               RequestBlock_struct Request;
+               ErrDescriptor_struct ErrorDescriptor;
+       } Command;
+       static const size_t cmd_sz = sizeof(Command) + sizeof(ErrorInfo_struct);
+       Command *cmd;
+       dma_addr_t paddr64;
+       uint32_t paddr32, tag;
+       void __iomem *vaddr;
+       int i, err;
+
+       vaddr = ioremap_nocache(pci_resource_start(pdev, 0), pci_resource_len(pdev, 0));
+       if (vaddr == NULL)
+               return -ENOMEM;
+
+       /* The Inbound Post Queue only accepts 32-bit physical addresses for the
+          CCISS commands, so they must be allocated from the lower 4GiB of
+          memory. */
+       err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32));
+       if (err) {
+               iounmap(vaddr);
+               return -ENOMEM;
+       }
+
+       cmd = pci_alloc_consistent(pdev, cmd_sz, &paddr64);
+       if (cmd == NULL) {
+               iounmap(vaddr);
+               return -ENOMEM;
+       }
+
+       /* This must fit, because of the 32-bit consistent DMA mask.  Also,
+          although there's no guarantee, we assume that the address is at
+          least 4-byte aligned (most likely, it's page-aligned). */
+       paddr32 = paddr64;
+
+       cmd->CommandHeader.ReplyQueue = 0;
+       cmd->CommandHeader.SGList = 0;
+       cmd->CommandHeader.SGTotal = 0;
+       cmd->CommandHeader.Tag.lower = paddr32;
+       cmd->CommandHeader.Tag.upper = 0;
+       memset(&cmd->CommandHeader.LUN.LunAddrBytes, 0, 8);
+
+       cmd->Request.CDBLen = 16;
+       cmd->Request.Type.Type = TYPE_MSG;
+       cmd->Request.Type.Attribute = ATTR_HEADOFQUEUE;
+       cmd->Request.Type.Direction = XFER_NONE;
+       cmd->Request.Timeout = 0; /* Don't time out */
+       cmd->Request.CDB[0] = opcode;
+       cmd->Request.CDB[1] = type;
+       memset(&cmd->Request.CDB[2], 0, 14); /* the rest of the CDB is reserved */
+
+       cmd->ErrorDescriptor.Addr.lower = paddr32 + sizeof(Command);
+       cmd->ErrorDescriptor.Addr.upper = 0;
+       cmd->ErrorDescriptor.Len = sizeof(ErrorInfo_struct);
+
+       writel(paddr32, vaddr + SA5_REQUEST_PORT_OFFSET);
+
+       for (i = 0; i < 10; i++) {
+               tag = readl(vaddr + SA5_REPLY_PORT_OFFSET);
+               if ((tag & ~3) == paddr32)
+                       break;
+               schedule_timeout_uninterruptible(HZ);
+       }
+
+       iounmap(vaddr);
+
+       /* we leak the DMA buffer here ... no choice since the controller could
+          still complete the command. */
+       if (i == 10) {
+               printk(KERN_ERR "cciss: controller message %02x:%02x timed out\n",
+                       opcode, type);
+               return -ETIMEDOUT;
+       }
+
+       pci_free_consistent(pdev, cmd_sz, cmd, paddr64);
+
+       if (tag & 2) {
+               printk(KERN_ERR "cciss: controller message %02x:%02x failed\n",
+                       opcode, type);
+               return -EIO;
+       }
+
+       printk(KERN_INFO "cciss: controller message %02x:%02x succeeded\n",
+               opcode, type);
+       return 0;
+}
+
+#define cciss_soft_reset_controller(p) cciss_message(p, 1, 0)
+#define cciss_noop(p) cciss_message(p, 3, 0)
+
+static __devinit int cciss_reset_msi(struct pci_dev *pdev)
+{
+/* the #defines are stolen from drivers/pci/msi.h. */
+#define msi_control_reg(base)          (base + PCI_MSI_FLAGS)
+#define PCI_MSIX_FLAGS_ENABLE          (1 << 15)
+
+       int pos;
+       u16 control = 0;
+
+       pos = pci_find_capability(pdev, PCI_CAP_ID_MSI);
+       if (pos) {
+               pci_read_config_word(pdev, msi_control_reg(pos), &control);
+               if (control & PCI_MSI_FLAGS_ENABLE) {
+                       printk(KERN_INFO "cciss: resetting MSI\n");
+                       pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSI_FLAGS_ENABLE);
+               }
+       }
+
+       pos = pci_find_capability(pdev, PCI_CAP_ID_MSIX);
+       if (pos) {
+               pci_read_config_word(pdev, msi_control_reg(pos), &control);
+               if (control & PCI_MSIX_FLAGS_ENABLE) {
+                       printk(KERN_INFO "cciss: resetting MSI-X\n");
+                       pci_write_config_word(pdev, msi_control_reg(pos), control & ~PCI_MSIX_FLAGS_ENABLE);
+               }
+       }
+
+       return 0;
+}
+
+/* This does a hard reset of the controller using PCI power management
+ * states. */
+static __devinit int cciss_hard_reset_controller(struct pci_dev *pdev)
+{
+       u16 pmcsr, saved_config_space[32];
+       int i, pos;
+
+       printk(KERN_INFO "cciss: using PCI PM to reset controller\n");
+
+       /* This is very nearly the same thing as
+
+          pci_save_state(pci_dev);
+          pci_set_power_state(pci_dev, PCI_D3hot);
+          pci_set_power_state(pci_dev, PCI_D0);
+          pci_restore_state(pci_dev);
+
+          but we can't use these nice canned kernel routines on
+          kexec, because they also check the MSI/MSI-X state in PCI
+          configuration space and do the wrong thing when it is
+          set/cleared.  Also, the pci_save/restore_state functions
+          violate the ordering requirements for restoring the
+          configuration space from the CCISS document (see the
+          comment below).  So we roll our own .... */
+
+       for (i = 0; i < 32; i++)
+               pci_read_config_word(pdev, 2*i, &saved_config_space[i]);
+
+       pos = pci_find_capability(pdev, PCI_CAP_ID_PM);
+       if (pos == 0) {
+               printk(KERN_ERR "cciss_reset_controller: PCI PM not supported\n");
+               return -ENODEV;
+       }
+
+       /* Quoting from the Open CISS Specification: "The Power
+        * Management Control/Status Register (CSR) controls the power
+        * state of the device.  The normal operating state is D0,
+        * CSR=00h.  The software off state is D3, CSR=03h.  To reset
+        * the controller, place the interface device in D3 then to
+        * D0, this causes a secondary PCI reset which will reset the
+        * controller." */
+
+       /* enter the D3hot power management state */
+       pci_read_config_word(pdev, pos + PCI_PM_CTRL, &pmcsr);
+       pmcsr &= ~PCI_PM_CTRL_STATE_MASK;
+       pmcsr |= PCI_D3hot;
+       pci_write_config_word(pdev, pos + PCI_PM_CTRL, pmcsr);
+
+       schedule_timeout_uninterruptible(HZ >> 1);
+
+       /* enter the D0 power management state */
+       pmcsr &= ~PCI_PM_CTRL_STATE_MASK;
+       pmcsr |= PCI_D0;
+       pci_write_config_word(pdev, pos + PCI_PM_CTRL, pmcsr);
+
+       schedule_timeout_uninterruptible(HZ >> 1);
+
+       /* Restore the PCI configuration space.  The Open CISS
+        * Specification says, "Restore the PCI Configuration
+        * Registers, offsets 00h through 60h. It is important to
+        * restore the command register, 16-bits at offset 04h,
+        * last. Do not restore the configuration status register,
+        * 16-bits at offset 06h."  Note that the offset is 2*i. */
+       for (i = 0; i < 32; i++) {
+               if (i == 2 || i == 3)
+                       continue;
+               pci_write_config_word(pdev, 2*i, saved_config_space[i]);
+       }
+       wmb();
+       pci_write_config_word(pdev, 4, saved_config_space[2]);
+
+       return 0;
+}
+
 /*
  *  This is it.  Find all the controllers and register them.  I really hate
  *  stealing all these major device numbers.
@@ -3189,25 +3877,51 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
        int i;
        int j = 0;
        int rc;
-       int dac;
+       int dac, return_code;
+       InquiryData_struct *inq_buff = NULL;
+
+       if (reset_devices) {
+               /* Reset the controller with a PCI power-cycle */
+               if (cciss_hard_reset_controller(pdev) || cciss_reset_msi(pdev))
+                       return -ENODEV;
+
+               /* Now try to get the controller to respond to a no-op. Some
+                  devices (notably the HP Smart Array 5i Controller) need
+                  up to 30 seconds to respond. */
+               for (i=0; i<30; i++) {
+                       if (cciss_noop(pdev) == 0)
+                               break;
+
+                       schedule_timeout_uninterruptible(HZ);
+               }
+               if (i == 30) {
+                       printk(KERN_ERR "cciss: controller seems dead\n");
+                       return -EBUSY;
+               }
+       }
 
        i = alloc_cciss_hba();
        if (i < 0)
                return -1;
 
        hba[i]->busy_initializing = 1;
+       INIT_HLIST_HEAD(&hba[i]->cmpQ);
+       INIT_HLIST_HEAD(&hba[i]->reqQ);
 
        if (cciss_pci_init(hba[i], pdev) != 0)
-               goto clean1;
+               goto clean0;
 
        sprintf(hba[i]->devname, "cciss%d", i);
        hba[i]->ctlr = i;
        hba[i]->pdev = pdev;
 
+       if (cciss_create_hba_sysfs_entry(hba[i]))
+               goto clean0;
+
        /* configure PCI DMA stuff */
-       if (!pci_set_dma_mask(pdev, DMA_64BIT_MASK))
+       if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(64)))
                dac = 1;
-       else if (!pci_set_dma_mask(pdev, DMA_32BIT_MASK))
+       else if (!pci_set_dma_mask(pdev, DMA_BIT_MASK(32)))
                dac = 0;
        else {
                printk(KERN_ERR "cciss: no suitable DMA available\n");
@@ -3246,8 +3960,8 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
               hba[i]->intr[SIMPLE_MODE_INT], dac ? "" : " not");
 
        hba[i]->cmd_pool_bits =
-           kmalloc(((hba[i]->nr_cmds + BITS_PER_LONG -
-                     1) / BITS_PER_LONG) * sizeof(unsigned long), GFP_KERNEL);
+           kmalloc(DIV_ROUND_UP(hba[i]->nr_cmds, BITS_PER_LONG)
+                       * sizeof(unsigned long), GFP_KERNEL);
        hba[i]->cmd_pool = (CommandList_struct *)
            pci_alloc_consistent(hba[i]->pdev,
                    hba[i]->nr_cmds * sizeof(CommandList_struct),
@@ -3262,15 +3976,6 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
                printk(KERN_ERR "cciss: out of memory");
                goto clean4;
        }
-#ifdef CONFIG_CISS_SCSI_TAPE
-       hba[i]->scsi_rejects.complete =
-           kmalloc(sizeof(hba[i]->scsi_rejects.complete[0]) *
-                   (hba[i]->nr_cmds + 5), GFP_KERNEL);
-       if (hba[i]->scsi_rejects.complete == NULL) {
-               printk(KERN_ERR "cciss: out of memory");
-               goto clean4;
-       }
-#endif
        spin_lock_init(&hba[i]->lock);
 
        /* Initialize the pdev driver private data.
@@ -3279,89 +3984,57 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
        /* command and error info recs zeroed out before
           they are used */
        memset(hba[i]->cmd_pool_bits, 0,
-              ((hba[i]->nr_cmds + BITS_PER_LONG -
-                1) / BITS_PER_LONG) * sizeof(unsigned long));
-
-#ifdef CCISS_DEBUG
-       printk(KERN_DEBUG "Scanning for drives on controller cciss%d\n", i);
-#endif                         /* CCISS_DEBUG */
+              DIV_ROUND_UP(hba[i]->nr_cmds, BITS_PER_LONG)
+                       * sizeof(unsigned long));
 
-       cciss_getgeometry(i);
+       hba[i]->num_luns = 0;
+       hba[i]->highest_lun = -1;
+       for (j = 0; j < CISS_MAX_LUN; j++) {
+               hba[i]->drv[j].raid_level = -1;
+               hba[i]->drv[j].queue = NULL;
+               hba[i]->gendisk[j] = NULL;
+       }
 
        cciss_scsi_setup(i);
 
        /* Turn the interrupts on so we can service requests */
        hba[i]->access.set_intr_mask(hba[i], CCISS_INTR_ON);
 
+       /* Get the firmware version */
+       inq_buff = kzalloc(sizeof(InquiryData_struct), GFP_KERNEL);
+       if (inq_buff == NULL) {
+               printk(KERN_ERR "cciss: out of memory\n");
+               goto clean4;
+       }
+
+       return_code = sendcmd_withirq(CISS_INQUIRY, i, inq_buff,
+               sizeof(InquiryData_struct), 0, CTLR_LUNID, TYPE_CMD);
+       if (return_code == IO_OK) {
+               hba[i]->firm_ver[0] = inq_buff->data_byte[32];
+               hba[i]->firm_ver[1] = inq_buff->data_byte[33];
+               hba[i]->firm_ver[2] = inq_buff->data_byte[34];
+               hba[i]->firm_ver[3] = inq_buff->data_byte[35];
+       } else {         /* send command failed */
+               printk(KERN_WARNING "cciss: unable to determine firmware"
+                       " version of controller\n");
+       }
+
        cciss_procinit(i);
 
        hba[i]->cciss_max_sectors = 2048;
 
        hba[i]->busy_initializing = 0;
 
-       do {
-               drive_info_struct *drv = &(hba[i]->drv[j]);
-               struct gendisk *disk = hba[i]->gendisk[j];
-               request_queue_t *q;
-
-               /* Check if the disk was allocated already */
-               if (!disk){
-                       hba[i]->gendisk[j] = alloc_disk(1 << NWD_SHIFT);
-                       disk = hba[i]->gendisk[j];
-               }
-
-               /* Check that the disk was able to be allocated */
-               if (!disk) {
-                       printk(KERN_ERR "cciss: unable to allocate memory for disk %d\n", j);
-                       goto clean4;
-               }
-
-               q = blk_init_queue(do_cciss_request, &hba[i]->lock);
-               if (!q) {
-                       printk(KERN_ERR
-                              "cciss:  unable to allocate queue for disk %d\n",
-                              j);
-                       goto clean4;
-               }
-               drv->queue = q;
-
-               q->backing_dev_info.ra_pages = READ_AHEAD;
-               blk_queue_bounce_limit(q, hba[i]->pdev->dma_mask);
-
-               /* This is a hardware imposed limit. */
-               blk_queue_max_hw_segments(q, MAXSGENTRIES);
-
-               /* This is a limit in the driver and could be eliminated. */
-               blk_queue_max_phys_segments(q, MAXSGENTRIES);
-
-               blk_queue_max_sectors(q, hba[i]->cciss_max_sectors);
-
-               blk_queue_softirq_done(q, cciss_softirq_done);
-
-               q->queuedata = hba[i];
-               sprintf(disk->disk_name, "cciss/c%dd%d", i, j);
-               disk->major = hba[i]->major;
-               disk->first_minor = j << NWD_SHIFT;
-               disk->fops = &cciss_fops;
-               disk->queue = q;
-               disk->private_data = drv;
-               disk->driverfs_dev = &pdev->dev;
-               /* we must register the controller even if no disks exist */
-               /* this is for the online array utilities */
-               if (!drv->heads && j)
-                       continue;
-               blk_queue_hardsect_size(q, drv->block_size);
-               set_capacity(disk, drv->nr_blocks);
-               add_disk(disk);
-               j++;
-       } while (j <= hba[i]->highest_lun);
+       rebuild_lun_table(hba[i], 1);
+       hba[i]->cciss_scan_thread = kthread_run(scan_thread, hba[i],
+                               "cciss_scan%02d", i);
+       if (IS_ERR(hba[i]->cciss_scan_thread))
+               return PTR_ERR(hba[i]->cciss_scan_thread);
 
        return 1;
 
-      clean4:
-#ifdef CONFIG_CISS_SCSI_TAPE
-       kfree(hba[i]->scsi_rejects.complete);
-#endif
+clean4:
+       kfree(inq_buff);
        kfree(hba[i]->cmd_pool_bits);
        if (hba[i]->cmd_pool)
                pci_free_consistent(hba[i]->pdev,
@@ -3373,9 +4046,11 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
                                    hba[i]->errinfo_pool,
                                    hba[i]->errinfo_pool_dhandle);
        free_irq(hba[i]->intr[SIMPLE_MODE_INT], hba[i]);
-      clean2:
+clean2:
        unregister_blkdev(hba[i]->major, hba[i]->devname);
-      clean1:
+clean1:
+       cciss_destroy_hba_sysfs_entry(hba[i]);
+clean0:
        hba[i]->busy_initializing = 0;
        /* cleanup any queues that may have been initialized */
        for (j=0; j <= hba[i]->highest_lun; j++){
@@ -3393,17 +4068,44 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
        return -1;
 }
 
-static void __devexit cciss_remove_one(struct pci_dev *pdev)
+static void cciss_shutdown(struct pci_dev *pdev)
 {
        ctlr_info_t *tmp_ptr;
-       int i, j;
+       int i;
        char flush_buf[4];
        int return_code;
 
+       tmp_ptr = pci_get_drvdata(pdev);
+       if (tmp_ptr == NULL)
+               return;
+       i = tmp_ptr->ctlr;
+       if (hba[i] == NULL)
+               return;
+
+       /* Turn board interrupts off  and send the flush cache command */
+       /* sendcmd will turn off interrupt, and send the flush...
+        * To write all data in the battery backed cache to disks */
+       memset(flush_buf, 0, 4);
+       return_code = sendcmd(CCISS_CACHE_FLUSH, i, flush_buf, 4, 0,
+               CTLR_LUNID, TYPE_CMD);
+       if (return_code == IO_OK) {
+               printk(KERN_INFO "Completed flushing cache on controller %d\n", i);
+       } else {
+               printk(KERN_WARNING "Error flushing cache on controller %d\n", i);
+       }
+       free_irq(hba[i]->intr[2], hba[i]);
+}
+
+static void __devexit cciss_remove_one(struct pci_dev *pdev)
+{
+       ctlr_info_t *tmp_ptr;
+       int i, j;
+
        if (pci_get_drvdata(pdev) == NULL) {
                printk(KERN_ERR "cciss: Unable to remove device \n");
                return;
        }
+
        tmp_ptr = pci_get_drvdata(pdev);
        i = tmp_ptr->ctlr;
        if (hba[i] == NULL) {
@@ -3411,35 +4113,17 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
                       "already be removed \n");
                return;
        }
-       /* Turn board interrupts off  and send the flush cache command */
-       /* sendcmd will turn off interrupt, and send the flush...
-        * To write all data in the battery backed cache to disks */
-       memset(flush_buf, 0, 4);
-       return_code = sendcmd(CCISS_CACHE_FLUSH, i, flush_buf, 4, 0, 0, 0, NULL,
-                             TYPE_CMD);
-       if (return_code != IO_OK) {
-               printk(KERN_WARNING "Error Flushing cache on controller %d\n",
-                      i);
-       }
-       free_irq(hba[i]->intr[2], hba[i]);
 
-#ifdef CONFIG_PCI_MSI
-       if (hba[i]->msix_vector)
-               pci_disable_msix(hba[i]->pdev);
-       else if (hba[i]->msi_vector)
-               pci_disable_msi(hba[i]->pdev);
-#endif                         /* CONFIG_PCI_MSI */
+       kthread_stop(hba[i]->cciss_scan_thread);
 
-       iounmap(hba[i]->vaddr);
-       cciss_unregister_scsi(i);       /* unhook from SCSI subsystem */
-       unregister_blkdev(hba[i]->major, hba[i]->devname);
        remove_proc_entry(hba[i]->devname, proc_cciss);
+       unregister_blkdev(hba[i]->major, hba[i]->devname);
 
        /* remove it from the disk list */
        for (j = 0; j < CISS_MAX_LUN; j++) {
                struct gendisk *disk = hba[i]->gendisk[j];
                if (disk) {
-                       request_queue_t *q = disk->queue;
+                       struct request_queue *q = disk->queue;
 
                        if (disk->flags & GENHD_FL_UP)
                                del_gendisk(disk);
@@ -3448,20 +4132,33 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
                }
        }
 
+#ifdef CONFIG_CISS_SCSI_TAPE
+       cciss_unregister_scsi(i);       /* unhook from SCSI subsystem */
+#endif
+
+       cciss_shutdown(pdev);
+
+#ifdef CONFIG_PCI_MSI
+       if (hba[i]->msix_vector)
+               pci_disable_msix(hba[i]->pdev);
+       else if (hba[i]->msi_vector)
+               pci_disable_msi(hba[i]->pdev);
+#endif                         /* CONFIG_PCI_MSI */
+
+       iounmap(hba[i]->vaddr);
+
        pci_free_consistent(hba[i]->pdev, hba[i]->nr_cmds * sizeof(CommandList_struct),
                            hba[i]->cmd_pool, hba[i]->cmd_pool_dhandle);
        pci_free_consistent(hba[i]->pdev, hba[i]->nr_cmds * sizeof(ErrorInfo_struct),
                            hba[i]->errinfo_pool, hba[i]->errinfo_pool_dhandle);
        kfree(hba[i]->cmd_pool_bits);
-#ifdef CONFIG_CISS_SCSI_TAPE
-       kfree(hba[i]->scsi_rejects.complete);
-#endif
        /*
         * Deliberately omit pci_disable_device(): it does something nasty to
         * Smart Array controllers that pci_enable_device does not undo
         */
        pci_release_regions(pdev);
        pci_set_drvdata(pdev, NULL);
+       cciss_destroy_hba_sysfs_entry(hba[i]);
        free_hba(i);
 }
 
@@ -3470,6 +4167,7 @@ static struct pci_driver cciss_pci_driver = {
        .probe = cciss_init_one,
        .remove = __devexit_p(cciss_remove_one),
        .id_table = cciss_pci_device_id,        /* id_table */
+       .shutdown = cciss_shutdown,
 };
 
 /*
@@ -3478,10 +4176,31 @@ static struct pci_driver cciss_pci_driver = {
  */
 static int __init cciss_init(void)
 {
+       int err;
+
+       /*
+        * The hardware requires that commands are aligned on a 64-bit
+        * boundary. Given that we use pci_alloc_consistent() to allocate an
+        * array of them, the size must be a multiple of 8 bytes.
+        */
+       BUILD_BUG_ON(sizeof(CommandList_struct) % 8);
+
        printk(KERN_INFO DRIVER_NAME "\n");
 
+       err = bus_register(&cciss_bus_type);
+       if (err)
+               return err;
+
        /* Register for our PCI devices */
-       return pci_register_driver(&cciss_pci_driver);
+       err = pci_register_driver(&cciss_pci_driver);
+       if (err)
+               goto err_bus_register;
+
+       return 0;
+
+err_bus_register:
+       bus_unregister(&cciss_bus_type);
+       return err;
 }
 
 static void __exit cciss_cleanup(void)
@@ -3497,7 +4216,8 @@ static void __exit cciss_cleanup(void)
                        cciss_remove_one(hba[i]->pdev);
                }
        }
-       remove_proc_entry("cciss", proc_root_driver);
+       remove_proc_entry("driver/cciss", NULL);
+       bus_unregister(&cciss_bus_type);
 }
 
 static void fail_all_cmds(unsigned long ctlr)
@@ -3515,15 +4235,17 @@ static void fail_all_cmds(unsigned long ctlr)
        pci_disable_device(h->pdev);    /* Make sure it is really dead. */
 
        /* move everything off the request queue onto the completed queue */
-       while ((c = h->reqQ) != NULL) {
-               removeQ(&(h->reqQ), c);
+       while (!hlist_empty(&h->reqQ)) {
+               c = hlist_entry(h->reqQ.first, CommandList_struct, list);
+               removeQ(c);
                h->Qdepth--;
-               addQ(&(h->cmpQ), c);
+               addQ(&h->cmpQ, c);
        }
 
        /* Now, fail everything on the completed queue with a HW error */
-       while ((c = h->cmpQ) != NULL) {
-               removeQ(&h->cmpQ, c);
+       while (!hlist_empty(&h->cmpQ)) {
+               c = hlist_entry(h->cmpQ.first, CommandList_struct, list);
+               removeQ(c);
                c->err_info->CommandStatus = CMD_HARDWARE_ERR;
                if (c->cmd_type == CMD_RWREQ) {
                        complete_command(h, c, 0);