]> git.karo-electronics.de Git - mv-sheeva.git/blobdiff - drivers/staging/sep/sep_driver.c
Merge tag 'v2.6.38' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6
[mv-sheeva.git] / drivers / staging / sep / sep_driver.c
diff --git a/drivers/staging/sep/sep_driver.c b/drivers/staging/sep/sep_driver.c
new file mode 100644 (file)
index 0000000..ac5d569
--- /dev/null
@@ -0,0 +1,3577 @@
+/*
+ *
+ *  sep_driver.c - Security Processor Driver main group of functions
+ *
+ *  Copyright(c) 2009,2010 Intel Corporation. All rights reserved.
+ *  Contributions(c) 2009,2010 Discretix. All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms of the GNU General Public License as published by the Free
+ *  Software Foundation; version 2 of the License.
+ *
+ *  This program is distributed in the hope that it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  You should have received a copy of the GNU General Public License along with
+ *  this program; if not, write to the Free Software Foundation, Inc., 59
+ *  Temple Place - Suite 330, Boston, MA  02111-1307, USA.
+ *
+ *  CONTACTS:
+ *
+ *  Mark Allyn         mark.a.allyn@intel.com
+ *  Jayant Mangalampalli jayant.mangalampalli@intel.com
+ *
+ *  CHANGES:
+ *
+ *  2009.06.26 Initial publish
+ *  2010.09.14  Upgrade to Medfield
+ *
+ */
+#define DEBUG
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/miscdevice.h>
+#include <linux/fs.h>
+#include <linux/cdev.h>
+#include <linux/kdev_t.h>
+#include <linux/mutex.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/poll.h>
+#include <linux/wait.h>
+#include <linux/pci.h>
+#include <linux/firmware.h>
+#include <linux/slab.h>
+#include <linux/ioctl.h>
+#include <asm/current.h>
+#include <linux/ioport.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/pagemap.h>
+#include <asm/cacheflush.h>
+#include <linux/sched.h>
+#include <linux/delay.h>
+#include <linux/jiffies.h>
+#include <linux/rar_register.h>
+
+#include "../memrar/memrar.h"
+
+#include "sep_driver_hw_defs.h"
+#include "sep_driver_config.h"
+#include "sep_driver_api.h"
+#include "sep_dev.h"
+
+/*----------------------------------------
+       DEFINES
+-----------------------------------------*/
+
+#define SEP_RAR_IO_MEM_REGION_SIZE 0x40000
+
+/*--------------------------------------------
+       GLOBAL variables
+--------------------------------------------*/
+
+/* Keep this a single static object for now to keep the conversion easy */
+
+static struct sep_device *sep_dev;
+
+/**
+ *     sep_load_firmware - copy firmware cache/resident
+ *     @sep: pointer to struct sep_device we are loading
+ *
+ *     This functions copies the cache and resident from their source
+ *     location into destination shared memory.
+ */
+static int sep_load_firmware(struct sep_device *sep)
+{
+       const struct firmware *fw;
+       char *cache_name = "cache.image.bin";
+       char *res_name = "resident.image.bin";
+       char *extapp_name = "extapp.image.bin";
+       int error ;
+       unsigned long work1, work2, work3;
+
+       /* Set addresses and load resident */
+       sep->resident_bus = sep->rar_bus;
+       sep->resident_addr = sep->rar_addr;
+
+       error = request_firmware(&fw, res_name, &sep->pdev->dev);
+       if (error) {
+               dev_warn(&sep->pdev->dev, "can't request resident fw\n");
+               return error;
+       }
+
+       memcpy(sep->resident_addr, (void *)fw->data, fw->size);
+       sep->resident_size = fw->size;
+       release_firmware(fw);
+
+       dev_dbg(&sep->pdev->dev, "resident virtual is %p\n",
+               sep->resident_addr);
+       dev_dbg(&sep->pdev->dev, "resident bus is %lx\n",
+               (unsigned long)sep->resident_bus);
+       dev_dbg(&sep->pdev->dev, "resident size is %08zx\n",
+               sep->resident_size);
+
+       /* Set addresses for dcache (no loading needed) */
+       work1 = (unsigned long)sep->resident_bus;
+       work2 = (unsigned long)sep->resident_size;
+       work3 = (work1 + work2 + (1024 * 4)) & 0xfffff000;
+       sep->dcache_bus = (dma_addr_t)work3;
+
+       work1 = (unsigned long)sep->resident_addr;
+       work2 = (unsigned long)sep->resident_size;
+       work3 = (work1 + work2 + (1024 * 4)) & 0xfffff000;
+       sep->dcache_addr = (void *)work3;
+
+       sep->dcache_size = 1024 * 128;
+
+       /* Set addresses and load cache */
+       sep->cache_bus = sep->dcache_bus + sep->dcache_size;
+       sep->cache_addr = sep->dcache_addr + sep->dcache_size;
+
+       error = request_firmware(&fw, cache_name, &sep->pdev->dev);
+       if (error) {
+               dev_warn(&sep->pdev->dev, "Unable to request cache firmware\n");
+               return error;
+       }
+
+       memcpy(sep->cache_addr, (void *)fw->data, fw->size);
+       sep->cache_size = fw->size;
+       release_firmware(fw);
+
+       dev_dbg(&sep->pdev->dev, "cache virtual is %p\n",
+               sep->cache_addr);
+       dev_dbg(&sep->pdev->dev, "cache bus is %08lx\n",
+               (unsigned long)sep->cache_bus);
+       dev_dbg(&sep->pdev->dev, "cache size is %08zx\n",
+               sep->cache_size);
+
+       /* Set addresses and load extapp */
+       sep->extapp_bus = sep->cache_bus + (1024 * 370);
+       sep->extapp_addr = sep->cache_addr + (1024 * 370);
+
+       error = request_firmware(&fw, extapp_name, &sep->pdev->dev);
+       if (error) {
+               dev_warn(&sep->pdev->dev, "Unable to request extapp firmware\n");
+               return error;
+       }
+
+       memcpy(sep->extapp_addr, (void *)fw->data, fw->size);
+       sep->extapp_size = fw->size;
+       release_firmware(fw);
+
+       dev_dbg(&sep->pdev->dev, "extapp virtual is %p\n",
+               sep->extapp_addr);
+       dev_dbg(&sep->pdev->dev, "extapp bus is %08llx\n",
+               (unsigned long long)sep->extapp_bus);
+       dev_dbg(&sep->pdev->dev, "extapp size is %08zx\n",
+               sep->extapp_size);
+
+       return error;
+}
+
+MODULE_FIRMWARE("sep/cache.image.bin");
+MODULE_FIRMWARE("sep/resident.image.bin");
+MODULE_FIRMWARE("sep/extapp.image.bin");
+
+/**
+ *     sep_dump_message - dump the message that is pending
+ *     @sep: SEP device
+ */
+static void sep_dump_message(struct sep_device *sep)
+{
+       int count;
+       u32 *p = sep->shared_addr;
+       for (count = 0; count < 12 * 4; count += 4)
+               dev_dbg(&sep->pdev->dev, "Word %d of the message is %x\n",
+                                                               count, *p++);
+}
+
+/**
+ *     sep_map_and_alloc_shared_area - allocate shared block
+ *     @sep: security processor
+ *     @size: size of shared area
+ */
+static int sep_map_and_alloc_shared_area(struct sep_device *sep)
+{
+       sep->shared_addr = dma_alloc_coherent(&sep->pdev->dev,
+               sep->shared_size,
+               &sep->shared_bus, GFP_KERNEL);
+
+       if (!sep->shared_addr) {
+               dev_warn(&sep->pdev->dev,
+                       "shared memory dma_alloc_coherent failed\n");
+               return -ENOMEM;
+       }
+       dev_dbg(&sep->pdev->dev,
+               "shared_addr %zx bytes @%p (bus %llx)\n",
+                               sep->shared_size, sep->shared_addr,
+                               (unsigned long long)sep->shared_bus);
+       return 0;
+}
+
+/**
+ *     sep_unmap_and_free_shared_area - free shared block
+ *     @sep: security processor
+ */
+static void sep_unmap_and_free_shared_area(struct sep_device *sep)
+{
+       dev_dbg(&sep->pdev->dev, "shared area unmap and free\n");
+       dma_free_coherent(&sep->pdev->dev, sep->shared_size,
+                               sep->shared_addr, sep->shared_bus);
+}
+
+/**
+ *     sep_shared_bus_to_virt - convert bus/virt addresses
+ *     @sep: pointer to struct sep_device
+ *     @bus_address: address to convert
+ *
+ *     Returns virtual address inside the shared area according
+ *     to the bus address.
+ */
+static void *sep_shared_bus_to_virt(struct sep_device *sep,
+                                               dma_addr_t bus_address)
+{
+       return sep->shared_addr + (bus_address - sep->shared_bus);
+}
+
+/**
+ *     open function for the singleton driver
+ *     @inode_ptr struct inode *
+ *     @file_ptr struct file *
+ *
+ *     Called when the user opens the singleton device interface
+ */
+static int sep_singleton_open(struct inode *inode_ptr, struct file *file_ptr)
+{
+       int error = 0;
+       struct sep_device *sep;
+
+       /*
+        * Get the SEP device structure and use it for the
+        * private_data field in filp for other methods
+        */
+       sep = sep_dev;
+
+       file_ptr->private_data = sep;
+
+       dev_dbg(&sep->pdev->dev, "Singleton open for pid %d\n", current->pid);
+
+       dev_dbg(&sep->pdev->dev, "calling test and set for singleton 0\n");
+       if (test_and_set_bit(0, &sep->singleton_access_flag)) {
+               error = -EBUSY;
+               goto end_function;
+       }
+
+       dev_dbg(&sep->pdev->dev, "sep_singleton_open end\n");
+end_function:
+       return error;
+}
+
+/**
+ *     sep_open - device open method
+ *     @inode: inode of SEP device
+ *     @filp: file handle to SEP device
+ *
+ *     Open method for the SEP device. Called when userspace opens
+ *     the SEP device node.
+ *
+ *     Returns zero on success otherwise an error code.
+ */
+static int sep_open(struct inode *inode, struct file *filp)
+{
+       struct sep_device *sep;
+
+       /*
+        * Get the SEP device structure and use it for the
+        * private_data field in filp for other methods
+        */
+       sep = sep_dev;
+       filp->private_data = sep;
+
+       dev_dbg(&sep->pdev->dev, "Open for pid %d\n", current->pid);
+
+       /* Anyone can open; locking takes place at transaction level */
+       return 0;
+}
+
+/**
+ *     sep_singleton_release - close a SEP singleton device
+ *     @inode: inode of SEP device
+ *     @filp: file handle being closed
+ *
+ *     Called on the final close of a SEP device. As the open protects against
+ *     multiple simultaenous opens that means this method is called when the
+ *     final reference to the open handle is dropped.
+ */
+static int sep_singleton_release(struct inode *inode, struct file *filp)
+{
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "Singleton release for pid %d\n",
+                                                       current->pid);
+       clear_bit(0, &sep->singleton_access_flag);
+       return 0;
+}
+
+/**
+ *     sep_request_daemonopen - request daemon open method
+ *     @inode: inode of SEP device
+ *     @filp: file handle to SEP device
+ *
+ *     Open method for the SEP request daemon. Called when
+ *     request daemon in userspace opens the SEP device node.
+ *
+ *     Returns zero on success otherwise an error code.
+ */
+static int sep_request_daemon_open(struct inode *inode, struct file *filp)
+{
+       struct sep_device *sep = sep_dev;
+       int error = 0;
+
+       filp->private_data = sep;
+
+       dev_dbg(&sep->pdev->dev, "Request daemon open for pid %d\n",
+               current->pid);
+
+       /* There is supposed to be only one request daemon */
+       dev_dbg(&sep->pdev->dev, "calling test and set for req_dmon open 0\n");
+       if (test_and_set_bit(0, &sep->request_daemon_open))
+               error = -EBUSY;
+       return error;
+}
+
+/**
+ *     sep_request_daemon_release - close a SEP daemon
+ *     @inode: inode of SEP device
+ *     @filp: file handle being closed
+ *
+ *     Called on the final close of a SEP daemon.
+ */
+static int sep_request_daemon_release(struct inode *inode, struct file *filp)
+{
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "Reques daemon release for pid %d\n",
+               current->pid);
+
+       /* Clear the request_daemon_open flag */
+       clear_bit(0, &sep->request_daemon_open);
+       return 0;
+}
+
+/**
+ *     sep_req_daemon_send_reply_command_handler - poke the SEP
+ *     @sep: struct sep_device *
+ *
+ *     This function raises interrupt to SEPm that signals that is has a
+ *     new command from HOST
+ */
+static int sep_req_daemon_send_reply_command_handler(struct sep_device *sep)
+{
+       unsigned long lck_flags;
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_req_daemon_send_reply_command_handler start\n");
+
+       sep_dump_message(sep);
+
+       /* Counters are lockable region */
+       spin_lock_irqsave(&sep->snd_rply_lck, lck_flags);
+       sep->send_ct++;
+       sep->reply_ct++;
+
+       /* Send the interrupt to SEP */
+       sep_write_reg(sep, HW_HOST_HOST_SEP_GPR2_REG_ADDR, sep->send_ct);
+       sep->send_ct++;
+
+       spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_req_daemon_send_reply send_ct %lx reply_ct %lx\n",
+               sep->send_ct, sep->reply_ct);
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_req_daemon_send_reply_command_handler end\n");
+
+       return 0;
+}
+
+
+/**
+ *     sep_free_dma_table_data_handler - free DMA table
+ *     @sep: pointere to struct sep_device
+ *
+ *     Handles the request to  free DMA table for synchronic actions
+ */
+static int sep_free_dma_table_data_handler(struct sep_device *sep)
+{
+       int count;
+       int dcb_counter;
+       /* Pointer to the current dma_resource struct */
+       struct sep_dma_resource *dma;
+
+       dev_dbg(&sep->pdev->dev, "sep_free_dma_table_data_handler start\n");
+
+       for (dcb_counter = 0; dcb_counter < sep->nr_dcb_creat; dcb_counter++) {
+               dma = &sep->dma_res_arr[dcb_counter];
+
+               /* Unmap and free input map array */
+               if (dma->in_map_array) {
+                       for (count = 0; count < dma->in_num_pages; count++) {
+                               dma_unmap_page(&sep->pdev->dev,
+                                       dma->in_map_array[count].dma_addr,
+                                       dma->in_map_array[count].size,
+                                       DMA_TO_DEVICE);
+                       }
+                       kfree(dma->in_map_array);
+               }
+
+               /* Unmap output map array, DON'T free it yet */
+               if (dma->out_map_array) {
+                       for (count = 0; count < dma->out_num_pages; count++) {
+                               dma_unmap_page(&sep->pdev->dev,
+                                       dma->out_map_array[count].dma_addr,
+                                       dma->out_map_array[count].size,
+                                       DMA_FROM_DEVICE);
+                       }
+                       kfree(dma->out_map_array);
+               }
+
+               /* Free page cache for output */
+               if (dma->in_page_array) {
+                       for (count = 0; count < dma->in_num_pages; count++) {
+                               flush_dcache_page(dma->in_page_array[count]);
+                               page_cache_release(dma->in_page_array[count]);
+                       }
+                       kfree(dma->in_page_array);
+               }
+
+               if (dma->out_page_array) {
+                       for (count = 0; count < dma->out_num_pages; count++) {
+                               if (!PageReserved(dma->out_page_array[count]))
+                                       SetPageDirty(dma->out_page_array[count]);
+                               flush_dcache_page(dma->out_page_array[count]);
+                               page_cache_release(dma->out_page_array[count]);
+                       }
+                       kfree(dma->out_page_array);
+               }
+
+               /* Reset all the values */
+               dma->in_page_array = NULL;
+               dma->out_page_array = NULL;
+               dma->in_num_pages = 0;
+               dma->out_num_pages = 0;
+               dma->in_map_array = NULL;
+               dma->out_map_array = NULL;
+               dma->in_map_num_entries = 0;
+               dma->out_map_num_entries = 0;
+       }
+
+       sep->nr_dcb_creat = 0;
+       sep->num_lli_tables_created = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_free_dma_table_data_handler end\n");
+       return 0;
+}
+
+/**
+ *     sep_request_daemon_mmap - maps the shared area to user space
+ *     @filp: pointer to struct file
+ *     @vma: pointer to vm_area_struct
+ *
+ *     Called by the kernel when the daemon attempts an mmap() syscall
+ *     using our handle.
+ */
+static int sep_request_daemon_mmap(struct file  *filp,
+       struct vm_area_struct  *vma)
+{
+       struct sep_device *sep = filp->private_data;
+       dma_addr_t bus_address;
+       int error = 0;
+
+       dev_dbg(&sep->pdev->dev, "daemon mmap start\n");
+
+       if ((vma->vm_end - vma->vm_start) > SEP_DRIVER_MMMAP_AREA_SIZE) {
+               error = -EINVAL;
+               goto end_function;
+       }
+
+       /* Get physical address */
+       bus_address = sep->shared_bus;
+
+       dev_dbg(&sep->pdev->dev, "bus_address is %08lx\n",
+                                       (unsigned long)bus_address);
+
+       if (remap_pfn_range(vma, vma->vm_start, bus_address >> PAGE_SHIFT,
+               vma->vm_end - vma->vm_start, vma->vm_page_prot)) {
+
+               dev_warn(&sep->pdev->dev, "remap_page_range failed\n");
+               error = -EAGAIN;
+               goto end_function;
+       }
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "daemon mmap end\n");
+       return error;
+}
+
+/**
+ *     sep_request_daemon_poll - poll implementation
+ *     @sep: struct sep_device * for current SEP device
+ *     @filp: struct file * for open file
+ *     @wait: poll_table * for poll
+ *
+ *     Called when our device is part of a poll() or select() syscall
+ */
+static unsigned int sep_request_daemon_poll(struct file *filp,
+       poll_table  *wait)
+{
+       u32     mask = 0;
+       /* GPR2 register */
+       u32     retval2;
+       unsigned long lck_flags;
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "daemon poll: start\n");
+
+       poll_wait(filp, &sep->event_request_daemon, wait);
+
+       dev_dbg(&sep->pdev->dev, "daemon poll: send_ct is %lx reply ct is %lx\n",
+                                               sep->send_ct, sep->reply_ct);
+
+       spin_lock_irqsave(&sep->snd_rply_lck, lck_flags);
+       /* Check if the data is ready */
+       if (sep->send_ct == sep->reply_ct) {
+               spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+
+               retval2 = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR2_REG_ADDR);
+               dev_dbg(&sep->pdev->dev,
+                       "daemon poll: data check (GPR2) is %x\n", retval2);
+
+               /* Check if PRINT request */
+               if ((retval2 >> 30) & 0x1) {
+                       dev_dbg(&sep->pdev->dev, "daemon poll: PRINTF request in\n");
+                       mask |= POLLIN;
+                       goto end_function;
+               }
+               /* Check if NVS request */
+               if (retval2 >> 31) {
+                       dev_dbg(&sep->pdev->dev, "daemon poll: NVS request in\n");
+                       mask |= POLLPRI | POLLWRNORM;
+               }
+       } else {
+               spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+               dev_dbg(&sep->pdev->dev,
+                       "daemon poll: no reply received; returning 0\n");
+               mask = 0;
+       }
+end_function:
+       dev_dbg(&sep->pdev->dev, "daemon poll: exit\n");
+       return mask;
+}
+
+/**
+ *     sep_release - close a SEP device
+ *     @inode: inode of SEP device
+ *     @filp: file handle being closed
+ *
+ *     Called on the final close of a SEP device.
+ */
+static int sep_release(struct inode *inode, struct file *filp)
+{
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "Release for pid %d\n", current->pid);
+
+       mutex_lock(&sep->sep_mutex);
+       /* Is this the process that has a transaction open?
+        * If so, lets reset pid_doing_transaction to 0 and
+        * clear the in use flags, and then wake up sep_event
+        * so that other processes can do transactions
+        */
+       dev_dbg(&sep->pdev->dev, "waking up event and mmap_event\n");
+       if (sep->pid_doing_transaction == current->pid) {
+               clear_bit(SEP_MMAP_LOCK_BIT, &sep->in_use_flags);
+               clear_bit(SEP_SEND_MSG_LOCK_BIT, &sep->in_use_flags);
+               sep_free_dma_table_data_handler(sep);
+               wake_up(&sep->event);
+               sep->pid_doing_transaction = 0;
+       }
+
+       mutex_unlock(&sep->sep_mutex);
+       return 0;
+}
+
+/**
+ *     sep_mmap -  maps the shared area to user space
+ *     @filp: pointer to struct file
+ *     @vma: pointer to vm_area_struct
+ *
+ *     Called on an mmap of our space via the normal SEP device
+ */
+static int sep_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+       dma_addr_t bus_addr;
+       struct sep_device *sep = filp->private_data;
+       unsigned long error = 0;
+
+       dev_dbg(&sep->pdev->dev, "mmap start\n");
+
+       /* Set the transaction busy (own the device) */
+       wait_event_interruptible(sep->event,
+               test_and_set_bit(SEP_MMAP_LOCK_BIT,
+               &sep->in_use_flags) == 0);
+
+       if (signal_pending(current)) {
+               error = -EINTR;
+               goto end_function_with_error;
+       }
+       /*
+        * The pid_doing_transaction indicates that this process
+        * now owns the facilities to performa a transaction with
+        * the SEP. While this process is performing a transaction,
+        * no other process who has the SEP device open can perform
+        * any transactions. This method allows more than one process
+        * to have the device open at any given time, which provides
+        * finer granularity for device utilization by multiple
+        * processes.
+        */
+       mutex_lock(&sep->sep_mutex);
+       sep->pid_doing_transaction = current->pid;
+       mutex_unlock(&sep->sep_mutex);
+
+       /* Zero the pools and the number of data pool alocation pointers */
+       sep->data_pool_bytes_allocated = 0;
+       sep->num_of_data_allocations = 0;
+
+       /*
+        * Check that the size of the mapped range is as the size of the message
+        * shared area
+        */
+       if ((vma->vm_end - vma->vm_start) > SEP_DRIVER_MMMAP_AREA_SIZE) {
+               error = -EINVAL;
+               goto end_function_with_error;
+       }
+
+       dev_dbg(&sep->pdev->dev, "shared_addr is %p\n", sep->shared_addr);
+
+       /* Get bus address */
+       bus_addr = sep->shared_bus;
+
+       dev_dbg(&sep->pdev->dev,
+               "bus_address is %lx\n", (unsigned long)bus_addr);
+
+       if (remap_pfn_range(vma, vma->vm_start, bus_addr >> PAGE_SHIFT,
+               vma->vm_end - vma->vm_start, vma->vm_page_prot)) {
+               dev_warn(&sep->pdev->dev, "remap_page_range failed\n");
+               error = -EAGAIN;
+               goto end_function_with_error;
+       }
+       dev_dbg(&sep->pdev->dev, "mmap end\n");
+       goto end_function;
+
+end_function_with_error:
+       /* Clear the bit */
+       clear_bit(SEP_MMAP_LOCK_BIT, &sep->in_use_flags);
+       mutex_lock(&sep->sep_mutex);
+       sep->pid_doing_transaction = 0;
+       mutex_unlock(&sep->sep_mutex);
+
+       /* Raise event for stuck contextes */
+
+       dev_warn(&sep->pdev->dev, "mmap error - waking up event\n");
+       wake_up(&sep->event);
+
+end_function:
+       return error;
+}
+
+/**
+ *     sep_poll - poll handler
+ *     @filp: pointer to struct file
+ *     @wait: pointer to poll_table
+ *
+ *     Called by the OS when the kernel is asked to do a poll on
+ *     a SEP file handle.
+ */
+static unsigned int sep_poll(struct file *filp, poll_table *wait)
+{
+       u32 mask = 0;
+       u32 retval = 0;
+       u32 retval2 = 0;
+       unsigned long lck_flags;
+
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "poll: start\n");
+
+       /* Am I the process that owns the transaction? */
+       mutex_lock(&sep->sep_mutex);
+       if (current->pid != sep->pid_doing_transaction) {
+               dev_warn(&sep->pdev->dev, "poll; wrong pid\n");
+               mask = POLLERR;
+               mutex_unlock(&sep->sep_mutex);
+               goto end_function;
+       }
+       mutex_unlock(&sep->sep_mutex);
+
+       /* Check if send command or send_reply were activated previously */
+       if (!test_bit(SEP_SEND_MSG_LOCK_BIT, &sep->in_use_flags)) {
+               dev_warn(&sep->pdev->dev, "poll; lock bit set\n");
+               mask = POLLERR;
+               goto end_function;
+       }
+
+       /* Add the event to the polling wait table */
+       dev_dbg(&sep->pdev->dev, "poll: calling wait sep_event\n");
+
+       poll_wait(filp, &sep->event, wait);
+
+       dev_dbg(&sep->pdev->dev, "poll: send_ct is %lx reply ct is %lx\n",
+               sep->send_ct, sep->reply_ct);
+
+       /* Check if error occured during poll */
+       retval2 = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR3_REG_ADDR);
+       if (retval2 != 0x0) {
+               dev_warn(&sep->pdev->dev, "poll; poll error %x\n", retval2);
+               mask |= POLLERR;
+               goto end_function;
+       }
+
+       spin_lock_irqsave(&sep->snd_rply_lck, lck_flags);
+
+       if (sep->send_ct == sep->reply_ct) {
+               spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+               retval = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR2_REG_ADDR);
+               dev_dbg(&sep->pdev->dev, "poll: data ready check (GPR2)  %x\n",
+                       retval);
+
+               /* Check if printf request  */
+               if ((retval >> 30) & 0x1) {
+                       dev_dbg(&sep->pdev->dev, "poll: SEP printf request\n");
+                       wake_up(&sep->event_request_daemon);
+                       goto end_function;
+               }
+
+               /* Check if the this is SEP reply or request */
+               if (retval >> 31) {
+                       dev_dbg(&sep->pdev->dev, "poll: SEP request\n");
+                       wake_up(&sep->event_request_daemon);
+               } else {
+                       dev_dbg(&sep->pdev->dev, "poll: normal return\n");
+                       /* In case it is again by send_reply_comand */
+                       clear_bit(SEP_SEND_MSG_LOCK_BIT, &sep->in_use_flags);
+                       sep_dump_message(sep);
+                       dev_dbg(&sep->pdev->dev,
+                               "poll; SEP reply POLLIN | POLLRDNORM\n");
+                       mask |= POLLIN | POLLRDNORM;
+               }
+       } else {
+               spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+               dev_dbg(&sep->pdev->dev,
+                       "poll; no reply received; returning mask of 0\n");
+               mask = 0;
+       }
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "poll: end\n");
+       return mask;
+}
+
+/**
+ *     sep_time_address - address in SEP memory of time
+ *     @sep: SEP device we want the address from
+ *
+ *     Return the address of the two dwords in memory used for time
+ *     setting.
+ */
+static u32 *sep_time_address(struct sep_device *sep)
+{
+       return sep->shared_addr + SEP_DRIVER_SYSTEM_TIME_MEMORY_OFFSET_IN_BYTES;
+}
+
+/**
+ *     sep_set_time - set the SEP time
+ *     @sep: the SEP we are setting the time for
+ *
+ *     Calculates time and sets it at the predefined address.
+ *     Called with the SEP mutex held.
+ */
+static unsigned long sep_set_time(struct sep_device *sep)
+{
+       struct timeval time;
+       u32 *time_addr; /* Address of time as seen by the kernel */
+
+
+       dev_dbg(&sep->pdev->dev, "sep_set_time start\n");
+
+       do_gettimeofday(&time);
+
+       /* Set value in the SYSTEM MEMORY offset */
+       time_addr = sep_time_address(sep);
+
+       time_addr[0] = SEP_TIME_VAL_TOKEN;
+       time_addr[1] = time.tv_sec;
+
+       dev_dbg(&sep->pdev->dev, "time.tv_sec is %lu\n", time.tv_sec);
+       dev_dbg(&sep->pdev->dev, "time_addr is %p\n", time_addr);
+       dev_dbg(&sep->pdev->dev, "sep->shared_addr is %p\n", sep->shared_addr);
+
+       return time.tv_sec;
+}
+
+/**
+ *     sep_set_caller_id_handler - insert caller id entry
+ *     @sep: SEP device
+ *     @arg: pointer to struct caller_id_struct
+ *
+ *     Inserts the data into the caller id table. Note that this function
+ *     falls under the ioctl lock
+ */
+static int sep_set_caller_id_handler(struct sep_device *sep, unsigned long arg)
+{
+       void __user *hash;
+       int   error = 0;
+       int   i;
+       struct caller_id_struct command_args;
+
+       dev_dbg(&sep->pdev->dev, "sep_set_caller_id_handler start\n");
+
+       for (i = 0; i < SEP_CALLER_ID_TABLE_NUM_ENTRIES; i++) {
+               if (sep->caller_id_table[i].pid == 0)
+                       break;
+       }
+
+       if (i == SEP_CALLER_ID_TABLE_NUM_ENTRIES) {
+               dev_warn(&sep->pdev->dev, "no more caller id entries left\n");
+               dev_warn(&sep->pdev->dev, "maximum number is %d\n",
+                                       SEP_CALLER_ID_TABLE_NUM_ENTRIES);
+               error = -EUSERS;
+               goto end_function;
+       }
+
+       /* Copy the data */
+       if (copy_from_user(&command_args, (void __user *)arg,
+               sizeof(command_args))) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       hash = (void __user *)(unsigned long)command_args.callerIdAddress;
+
+       if (!command_args.pid || !command_args.callerIdSizeInBytes) {
+               error = -EINVAL;
+               goto end_function;
+       }
+
+       dev_dbg(&sep->pdev->dev, "pid is %x\n", command_args.pid);
+       dev_dbg(&sep->pdev->dev, "callerIdSizeInBytes is %x\n",
+               command_args.callerIdSizeInBytes);
+
+       if (command_args.callerIdSizeInBytes >
+                                       SEP_CALLER_ID_HASH_SIZE_IN_BYTES) {
+               error = -EMSGSIZE;
+               goto end_function;
+       }
+
+       sep->caller_id_table[i].pid = command_args.pid;
+
+       if (copy_from_user(sep->caller_id_table[i].callerIdHash,
+               hash, command_args.callerIdSizeInBytes))
+               error = -EFAULT;
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_set_caller_id_handler end\n");
+       return error;
+}
+
+/**
+ *     sep_set_current_caller_id - set the caller id
+ *     @sep: pointer to struct_sep_device
+ *
+ *     Set the caller ID (if it exists) to the SEP. Note that this
+ *     function falls under the ioctl lock
+ */
+static int sep_set_current_caller_id(struct sep_device *sep)
+{
+       int i;
+       u32 *hash_buf_ptr;
+
+       dev_dbg(&sep->pdev->dev, "sep_set_current_caller_id start\n");
+       dev_dbg(&sep->pdev->dev, "current process is %d\n", current->pid);
+
+       /* Zero the previous value */
+       memset(sep->shared_addr + SEP_CALLER_ID_OFFSET_BYTES,
+                                       0, SEP_CALLER_ID_HASH_SIZE_IN_BYTES);
+
+       for (i = 0; i < SEP_CALLER_ID_TABLE_NUM_ENTRIES; i++) {
+               if (sep->caller_id_table[i].pid == current->pid) {
+                       dev_dbg(&sep->pdev->dev, "Caller Id found\n");
+
+                       memcpy(sep->shared_addr + SEP_CALLER_ID_OFFSET_BYTES,
+                               (void *)(sep->caller_id_table[i].callerIdHash),
+                               SEP_CALLER_ID_HASH_SIZE_IN_BYTES);
+                       break;
+               }
+       }
+       /* Ensure data is in little endian */
+       hash_buf_ptr = (u32 *)sep->shared_addr +
+               SEP_CALLER_ID_OFFSET_BYTES;
+
+       for (i = 0; i < SEP_CALLER_ID_HASH_SIZE_IN_WORDS; i++)
+               hash_buf_ptr[i] = cpu_to_le32(hash_buf_ptr[i]);
+
+       dev_dbg(&sep->pdev->dev, "sep_set_current_caller_id end\n");
+       return 0;
+}
+
+/**
+ *     sep_send_command_handler - kick off a command
+ *     @sep: SEP being signalled
+ *
+ *     This function raises interrupt to SEP that signals that is has a new
+ *     command from the host
+ *
+ *      Note that this function does fall under the ioctl lock
+ */
+static int sep_send_command_handler(struct sep_device *sep)
+{
+       unsigned long lck_flags;
+       int error = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_send_command_handler start\n");
+
+       if (test_and_set_bit(SEP_SEND_MSG_LOCK_BIT, &sep->in_use_flags)) {
+               error = -EPROTO;
+               goto end_function;
+       }
+       sep_set_time(sep);
+
+       sep_set_current_caller_id(sep);
+
+       sep_dump_message(sep);
+
+       /* Update counter */
+       spin_lock_irqsave(&sep->snd_rply_lck, lck_flags);
+       sep->send_ct++;
+       spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_send_command_handler send_ct %lx reply_ct %lx\n",
+                                               sep->send_ct, sep->reply_ct);
+
+       /* Send interrupt to SEP */
+       sep_write_reg(sep, HW_HOST_HOST_SEP_GPR0_REG_ADDR, 0x2);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_send_command_handler end\n");
+       return error;
+}
+
+/**
+ *     sep_allocate_data_pool_memory_handler -allocate pool memory
+ *     @sep: pointer to struct sep_device
+ *     @arg: pointer to struct alloc_struct
+ *
+ *     This function handles the allocate data pool memory request
+ *     This function returns calculates the bus address of the
+ *     allocated memory, and the offset of this area from the mapped address.
+ *     Therefore, the FVOs in user space can calculate the exact virtual
+ *     address of this allocated memory
+ */
+static int sep_allocate_data_pool_memory_handler(struct sep_device *sep,
+       unsigned long arg)
+{
+       int error = 0;
+       struct alloc_struct command_args;
+
+       /* Holds the allocated buffer address in the system memory pool */
+       u32 *token_addr;
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_allocate_data_pool_memory_handler start\n");
+
+       if (copy_from_user(&command_args, (void __user *)arg,
+                                       sizeof(struct alloc_struct))) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       /* Allocate memory */
+       if ((sep->data_pool_bytes_allocated + command_args.num_bytes) >
+               SEP_DRIVER_DATA_POOL_SHARED_AREA_SIZE_IN_BYTES) {
+               error = -ENOMEM;
+               goto end_function;
+       }
+
+       dev_dbg(&sep->pdev->dev,
+               "bytes_allocated: %x\n", (int)sep->data_pool_bytes_allocated);
+       dev_dbg(&sep->pdev->dev,
+               "offset: %x\n", SEP_DRIVER_DATA_POOL_AREA_OFFSET_IN_BYTES);
+       /* Set the virtual and bus address */
+       command_args.offset = SEP_DRIVER_DATA_POOL_AREA_OFFSET_IN_BYTES +
+               sep->data_pool_bytes_allocated;
+
+       dev_dbg(&sep->pdev->dev,
+               "command_args.offset: %x\n", command_args.offset);
+
+       /* Place in the shared area that is known by the SEP */
+       token_addr = (u32 *)(sep->shared_addr +
+               SEP_DRIVER_DATA_POOL_ALLOCATION_OFFSET_IN_BYTES +
+               (sep->num_of_data_allocations)*2*sizeof(u32));
+
+       dev_dbg(&sep->pdev->dev, "allocation offset: %x\n",
+               SEP_DRIVER_DATA_POOL_ALLOCATION_OFFSET_IN_BYTES);
+       dev_dbg(&sep->pdev->dev, "data pool token addr is %p\n", token_addr);
+
+       token_addr[0] = SEP_DATA_POOL_POINTERS_VAL_TOKEN;
+       token_addr[1] = (u32)sep->shared_bus +
+               SEP_DRIVER_DATA_POOL_AREA_OFFSET_IN_BYTES +
+               sep->data_pool_bytes_allocated;
+
+       dev_dbg(&sep->pdev->dev, "data pool token [0] %x\n", token_addr[0]);
+       dev_dbg(&sep->pdev->dev, "data pool token [1] %x\n", token_addr[1]);
+
+       /* Write the memory back to the user space */
+       error = copy_to_user((void *)arg, (void *)&command_args,
+               sizeof(struct alloc_struct));
+       if (error) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       /* Update the allocation */
+       sep->data_pool_bytes_allocated += command_args.num_bytes;
+       sep->num_of_data_allocations += 1;
+
+       dev_dbg(&sep->pdev->dev, "data_allocations %d\n",
+               sep->num_of_data_allocations);
+       dev_dbg(&sep->pdev->dev, "bytes allocated  %d\n",
+               (int)sep->data_pool_bytes_allocated);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_allocate_data_pool_memory_handler end\n");
+       return error;
+}
+
+/**
+ *     sep_lock_kernel_pages - map kernel pages for DMA
+ *     @sep: pointer to struct sep_device
+ *     @kernel_virt_addr: address of data buffer in kernel
+ *     @data_size: size of data
+ *     @lli_array_ptr: lli array
+ *     @in_out_flag: input into device or output from device
+ *
+ *     This function locks all the physical pages of the kernel virtual buffer
+ *     and construct a basic lli  array, where each entry holds the physical
+ *     page address and the size that application data holds in this page
+ *     This function is used only during kernel crypto mod calls from within
+ *     the kernel (when ioctl is not used)
+ */
+static int sep_lock_kernel_pages(struct sep_device *sep,
+       unsigned long kernel_virt_addr,
+       u32 data_size,
+       struct sep_lli_entry **lli_array_ptr,
+       int in_out_flag)
+
+{
+       int error = 0;
+       /* Array of lli */
+       struct sep_lli_entry *lli_array;
+       /* Map array */
+       struct sep_dma_map *map_array;
+
+       dev_dbg(&sep->pdev->dev, "sep_lock_kernel_pages start\n");
+       dev_dbg(&sep->pdev->dev, "kernel_virt_addr is %08lx\n",
+                               (unsigned long)kernel_virt_addr);
+       dev_dbg(&sep->pdev->dev, "data_size is %x\n", data_size);
+
+       lli_array = kmalloc(sizeof(struct sep_lli_entry), GFP_ATOMIC);
+       if (!lli_array) {
+               error = -ENOMEM;
+               goto end_function;
+       }
+       map_array = kmalloc(sizeof(struct sep_dma_map), GFP_ATOMIC);
+       if (!map_array) {
+               error = -ENOMEM;
+               goto end_function_with_error;
+       }
+
+       map_array[0].dma_addr =
+               dma_map_single(&sep->pdev->dev, (void *)kernel_virt_addr,
+               data_size, DMA_BIDIRECTIONAL);
+       map_array[0].size = data_size;
+
+
+       /*
+        * Set the start address of the first page - app data may start not at
+        * the beginning of the page
+        */
+       lli_array[0].bus_address = (u32)map_array[0].dma_addr;
+       lli_array[0].block_size = map_array[0].size;
+
+       dev_dbg(&sep->pdev->dev,
+       "lli_array[0].bus_address is %08lx, lli_array[0].block_size is %x\n",
+               (unsigned long)lli_array[0].bus_address,
+               lli_array[0].block_size);
+
+       /* Set the output parameters */
+       if (in_out_flag == SEP_DRIVER_IN_FLAG) {
+               *lli_array_ptr = lli_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages = 1;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_page_array = NULL;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_map_array = map_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_map_num_entries = 1;
+       } else {
+               *lli_array_ptr = lli_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_num_pages = 1;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_page_array = NULL;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_map_array = map_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_map_num_entries = 1;
+       }
+       goto end_function;
+
+end_function_with_error:
+       kfree(lli_array);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_lock_kernel_pages end\n");
+       return error;
+}
+
+/**
+ *     sep_lock_user_pages - lock and map user pages for DMA
+ *     @sep: pointer to struct sep_device
+ *     @app_virt_addr: user memory data buffer
+ *     @data_size: size of data buffer
+ *     @lli_array_ptr: lli array
+ *     @in_out_flag: input or output to device
+ *
+ *     This function locks all the physical pages of the application
+ *     virtual buffer and construct a basic lli  array, where each entry
+ *     holds the physical page address and the size that application
+ *     data holds in this physical pages
+ */
+static int sep_lock_user_pages(struct sep_device *sep,
+       u32 app_virt_addr,
+       u32 data_size,
+       struct sep_lli_entry **lli_array_ptr,
+       int in_out_flag)
+
+{
+       int error = 0;
+       u32 count;
+       int result;
+       /* The the page of the end address of the user space buffer */
+       u32 end_page;
+       /* The page of the start address of the user space buffer */
+       u32 start_page;
+       /* The range in pages */
+       u32 num_pages;
+       /* Array of pointers to page */
+       struct page **page_array;
+       /* Array of lli */
+       struct sep_lli_entry *lli_array;
+       /* Map array */
+       struct sep_dma_map *map_array;
+       /* Direction of the DMA mapping for locked pages */
+       enum dma_data_direction dir;
+
+       dev_dbg(&sep->pdev->dev, "sep_lock_user_pages start\n");
+
+       /* Set start and end pages  and num pages */
+       end_page = (app_virt_addr + data_size - 1) >> PAGE_SHIFT;
+       start_page = app_virt_addr >> PAGE_SHIFT;
+       num_pages = end_page - start_page + 1;
+
+       dev_dbg(&sep->pdev->dev, "app_virt_addr is %x\n", app_virt_addr);
+       dev_dbg(&sep->pdev->dev, "data_size is %x\n", data_size);
+       dev_dbg(&sep->pdev->dev, "start_page is %x\n", start_page);
+       dev_dbg(&sep->pdev->dev, "end_page is %x\n", end_page);
+       dev_dbg(&sep->pdev->dev, "num_pages is %x\n", num_pages);
+
+       dev_dbg(&sep->pdev->dev, "starting page_array malloc\n");
+
+       /* Allocate array of pages structure pointers */
+       page_array = kmalloc(sizeof(struct page *) * num_pages, GFP_ATOMIC);
+       if (!page_array) {
+               error = -ENOMEM;
+               goto end_function;
+       }
+       map_array = kmalloc(sizeof(struct sep_dma_map) * num_pages, GFP_ATOMIC);
+       if (!map_array) {
+               dev_warn(&sep->pdev->dev, "kmalloc for map_array failed\n");
+               error = -ENOMEM;
+               goto end_function_with_error1;
+       }
+
+       lli_array = kmalloc(sizeof(struct sep_lli_entry) * num_pages,
+               GFP_ATOMIC);
+
+       if (!lli_array) {
+               dev_warn(&sep->pdev->dev, "kmalloc for lli_array failed\n");
+               error = -ENOMEM;
+               goto end_function_with_error2;
+       }
+
+       dev_dbg(&sep->pdev->dev, "starting get_user_pages\n");
+
+       /* Convert the application virtual address into a set of physical */
+       down_read(&current->mm->mmap_sem);
+       result = get_user_pages(current, current->mm, app_virt_addr,
+               num_pages,
+               ((in_out_flag == SEP_DRIVER_IN_FLAG) ? 0 : 1),
+               0, page_array, NULL);
+
+       up_read(&current->mm->mmap_sem);
+
+       /* Check the number of pages locked - if not all then exit with error */
+       if (result != num_pages) {
+               dev_warn(&sep->pdev->dev,
+                       "not all pages locked by get_user_pages\n");
+               error = -ENOMEM;
+               goto end_function_with_error3;
+       }
+
+       dev_dbg(&sep->pdev->dev, "get_user_pages succeeded\n");
+
+       /* Set direction */
+       if (in_out_flag == SEP_DRIVER_IN_FLAG)
+               dir = DMA_TO_DEVICE;
+       else
+               dir = DMA_FROM_DEVICE;
+
+       /*
+        * Fill the array using page array data and
+        * map the pages - this action will also flush the cache as needed
+        */
+       for (count = 0; count < num_pages; count++) {
+               /* Fill the map array */
+               map_array[count].dma_addr =
+                       dma_map_page(&sep->pdev->dev, page_array[count],
+                       0, PAGE_SIZE, /*dir*/DMA_BIDIRECTIONAL);
+
+               map_array[count].size = PAGE_SIZE;
+
+               /* Fill the lli array entry */
+               lli_array[count].bus_address = (u32)map_array[count].dma_addr;
+               lli_array[count].block_size = PAGE_SIZE;
+
+               dev_warn(&sep->pdev->dev, "lli_array[%x].bus_address is %08lx, lli_array[%x].block_size is %x\n",
+                       count, (unsigned long)lli_array[count].bus_address,
+                       count, lli_array[count].block_size);
+       }
+
+       /* Check the offset for the first page */
+       lli_array[0].bus_address =
+               lli_array[0].bus_address + (app_virt_addr & (~PAGE_MASK));
+
+       /* Check that not all the data is in the first page only */
+       if ((PAGE_SIZE - (app_virt_addr & (~PAGE_MASK))) >= data_size)
+               lli_array[0].block_size = data_size;
+       else
+               lli_array[0].block_size =
+                       PAGE_SIZE - (app_virt_addr & (~PAGE_MASK));
+
+       dev_dbg(&sep->pdev->dev,
+               "lli_array[0].bus_address is %08lx, lli_array[0].block_size is %x\n",
+               (unsigned long)lli_array[count].bus_address,
+               lli_array[count].block_size);
+
+       /* Check the size of the last page */
+       if (num_pages > 1) {
+               lli_array[num_pages - 1].block_size =
+                       (app_virt_addr + data_size) & (~PAGE_MASK);
+
+               dev_warn(&sep->pdev->dev,
+                       "lli_array[%x].bus_address is %08lx, lli_array[%x].block_size is %x\n",
+                       num_pages - 1,
+                       (unsigned long)lli_array[count].bus_address,
+                       num_pages - 1,
+                       lli_array[count].block_size);
+       }
+
+       /* Set output params acording to the in_out flag */
+       if (in_out_flag == SEP_DRIVER_IN_FLAG) {
+               *lli_array_ptr = lli_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages = num_pages;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_page_array = page_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_map_array = map_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].in_map_num_entries =
+                                                               num_pages;
+       } else {
+               *lli_array_ptr = lli_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_num_pages = num_pages;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_page_array =
+                                                               page_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_map_array = map_array;
+               sep->dma_res_arr[sep->nr_dcb_creat].out_map_num_entries =
+                                                               num_pages;
+       }
+       goto end_function;
+
+end_function_with_error3:
+       /* Free lli array */
+       kfree(lli_array);
+
+end_function_with_error2:
+       kfree(map_array);
+
+end_function_with_error1:
+       /* Free page array */
+       kfree(page_array);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_lock_user_pages end\n");
+       return error;
+}
+
+/**
+ *     u32 sep_calculate_lli_table_max_size - size the LLI table
+ *     @sep: pointer to struct sep_device
+ *     @lli_in_array_ptr
+ *     @num_array_entries
+ *     @last_table_flag
+ *
+ *     This function calculates the size of data that can be inserted into
+ *     the lli table from this array, such that either the table is full
+ *     (all entries are entered), or there are no more entries in the
+ *     lli array
+ */
+static u32 sep_calculate_lli_table_max_size(struct sep_device *sep,
+       struct sep_lli_entry *lli_in_array_ptr,
+       u32 num_array_entries,
+       u32 *last_table_flag)
+{
+       u32 counter;
+       /* Table data size */
+       u32 table_data_size = 0;
+       /* Data size for the next table */
+       u32 next_table_data_size;
+
+       *last_table_flag = 0;
+
+       /*
+        * Calculate the data in the out lli table till we fill the whole
+        * table or till the data has ended
+        */
+       for (counter = 0;
+               (counter < (SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP - 1)) &&
+                       (counter < num_array_entries); counter++)
+               table_data_size += lli_in_array_ptr[counter].block_size;
+
+       /*
+        * Check if we reached the last entry,
+        * meaning this ia the last table to build,
+        * and no need to check the block alignment
+        */
+       if (counter == num_array_entries) {
+               /* Set the last table flag */
+               *last_table_flag = 1;
+               goto end_function;
+       }
+
+       /*
+        * Calculate the data size of the next table.
+        * Stop if no entries left or if data size is more the DMA restriction
+        */
+       next_table_data_size = 0;
+       for (; counter < num_array_entries; counter++) {
+               next_table_data_size += lli_in_array_ptr[counter].block_size;
+               if (next_table_data_size >= SEP_DRIVER_MIN_DATA_SIZE_PER_TABLE)
+                       break;
+       }
+
+       /*
+        * Check if the next table data size is less then DMA rstriction.
+        * if it is - recalculate the current table size, so that the next
+        * table data size will be adaquete for DMA
+        */
+       if (next_table_data_size &&
+               next_table_data_size < SEP_DRIVER_MIN_DATA_SIZE_PER_TABLE)
+
+               table_data_size -= (SEP_DRIVER_MIN_DATA_SIZE_PER_TABLE -
+                       next_table_data_size);
+
+       dev_dbg(&sep->pdev->dev, "table data size is %x\n",
+                                                       table_data_size);
+end_function:
+       return table_data_size;
+}
+
+/**
+ *     sep_build_lli_table - build an lli array for the given table
+ *     @sep: pointer to struct sep_device
+ *     @lli_array_ptr: pointer to lli array
+ *     @lli_table_ptr: pointer to lli table
+ *     @num_processed_entries_ptr: pointer to number of entries
+ *     @num_table_entries_ptr: pointer to number of tables
+ *     @table_data_size: total data size
+ *
+ *     Builds ant lli table from the lli_array according to
+ *     the given size of data
+ */
+static void sep_build_lli_table(struct sep_device *sep,
+       struct sep_lli_entry    *lli_array_ptr,
+       struct sep_lli_entry    *lli_table_ptr,
+       u32 *num_processed_entries_ptr,
+       u32 *num_table_entries_ptr,
+       u32 table_data_size)
+{
+       /* Current table data size */
+       u32 curr_table_data_size;
+       /* Counter of lli array entry */
+       u32 array_counter;
+
+       dev_dbg(&sep->pdev->dev, "sep_build_lli_table start\n");
+
+       /* Init currrent table data size and lli array entry counter */
+       curr_table_data_size = 0;
+       array_counter = 0;
+       *num_table_entries_ptr = 1;
+
+       dev_dbg(&sep->pdev->dev, "table_data_size is %x\n", table_data_size);
+
+       /* Fill the table till table size reaches the needed amount */
+       while (curr_table_data_size < table_data_size) {
+               /* Update the number of entries in table */
+               (*num_table_entries_ptr)++;
+
+               lli_table_ptr->bus_address =
+                       cpu_to_le32(lli_array_ptr[array_counter].bus_address);
+
+               lli_table_ptr->block_size =
+                       cpu_to_le32(lli_array_ptr[array_counter].block_size);
+
+               curr_table_data_size += lli_array_ptr[array_counter].block_size;
+
+               dev_dbg(&sep->pdev->dev, "lli_table_ptr is %p\n",
+                                                               lli_table_ptr);
+               dev_dbg(&sep->pdev->dev, "lli_table_ptr->bus_address is %08lx\n",
+                               (unsigned long)lli_table_ptr->bus_address);
+               dev_dbg(&sep->pdev->dev, "lli_table_ptr->block_size is %x\n",
+                       lli_table_ptr->block_size);
+
+               /* Check for overflow of the table data */
+               if (curr_table_data_size > table_data_size) {
+                       dev_dbg(&sep->pdev->dev,
+                               "curr_table_data_size too large\n");
+
+                       /* Update the size of block in the table */
+                       lli_table_ptr->block_size -=
+                       cpu_to_le32((curr_table_data_size - table_data_size));
+
+                       /* Update the physical address in the lli array */
+                       lli_array_ptr[array_counter].bus_address +=
+                       cpu_to_le32(lli_table_ptr->block_size);
+
+                       /* Update the block size left in the lli array */
+                       lli_array_ptr[array_counter].block_size =
+                               (curr_table_data_size - table_data_size);
+               } else
+                       /* Advance to the next entry in the lli_array */
+                       array_counter++;
+
+               dev_dbg(&sep->pdev->dev,
+                       "lli_table_ptr->bus_address is %08lx\n",
+                               (unsigned long)lli_table_ptr->bus_address);
+               dev_dbg(&sep->pdev->dev,
+                       "lli_table_ptr->block_size is %x\n",
+                       lli_table_ptr->block_size);
+
+               /* Move to the next entry in table */
+               lli_table_ptr++;
+       }
+
+       /* Set the info entry to default */
+       lli_table_ptr->bus_address = 0xffffffff;
+       lli_table_ptr->block_size = 0;
+
+       dev_dbg(&sep->pdev->dev, "lli_table_ptr is %p\n", lli_table_ptr);
+       dev_dbg(&sep->pdev->dev, "lli_table_ptr->bus_address is %08lx\n",
+                               (unsigned long)lli_table_ptr->bus_address);
+       dev_dbg(&sep->pdev->dev, "lli_table_ptr->block_size is %x\n",
+                                               lli_table_ptr->block_size);
+
+       /* Set the output parameter */
+       *num_processed_entries_ptr += array_counter;
+
+       dev_dbg(&sep->pdev->dev, "num_processed_entries_ptr is %x\n",
+               *num_processed_entries_ptr);
+
+       dev_dbg(&sep->pdev->dev, "sep_build_lli_table end\n");
+}
+
+/**
+ *     sep_shared_area_virt_to_bus - map shared area to bus address
+ *     @sep: pointer to struct sep_device
+ *     @virt_address: virtual address to convert
+ *
+ *     This functions returns the physical address inside shared area according
+ *     to the virtual address. It can be either on the externa RAM device
+ *     (ioremapped), or on the system RAM
+ *     This implementation is for the external RAM
+ */
+static dma_addr_t sep_shared_area_virt_to_bus(struct sep_device *sep,
+       void *virt_address)
+{
+       dev_dbg(&sep->pdev->dev, "sh virt to phys v %p\n", virt_address);
+       dev_dbg(&sep->pdev->dev, "sh virt to phys p %08lx\n",
+               (unsigned long)
+               sep->shared_bus + (virt_address - sep->shared_addr));
+
+       return sep->shared_bus + (size_t)(virt_address - sep->shared_addr);
+}
+
+/**
+ *     sep_shared_area_bus_to_virt - map shared area bus address to kernel
+ *     @sep: pointer to struct sep_device
+ *     @bus_address: bus address to convert
+ *
+ *     This functions returns the virtual address inside shared area
+ *     according to the physical address. It can be either on the
+ *     externa RAM device (ioremapped), or on the system RAM
+ *     This implementation is for the external RAM
+ */
+static void *sep_shared_area_bus_to_virt(struct sep_device *sep,
+       dma_addr_t bus_address)
+{
+       dev_dbg(&sep->pdev->dev, "shared bus to virt b=%lx v=%lx\n",
+               (unsigned long)bus_address, (unsigned long)(sep->shared_addr +
+                       (size_t)(bus_address - sep->shared_bus)));
+
+       return sep->shared_addr + (size_t)(bus_address - sep->shared_bus);
+}
+
+/**
+ *     sep_debug_print_lli_tables - dump LLI table
+ *     @sep: pointer to struct sep_device
+ *     @lli_table_ptr: pointer to sep_lli_entry
+ *     @num_table_entries: number of entries
+ *     @table_data_size: total data size
+ *
+ *     Walk the the list of the print created tables and print all the data
+ */
+static void sep_debug_print_lli_tables(struct sep_device *sep,
+       struct sep_lli_entry *lli_table_ptr,
+       unsigned long num_table_entries,
+       unsigned long table_data_size)
+{
+       unsigned long table_count = 1;
+       unsigned long entries_count = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_debug_print_lli_tables start\n");
+
+       while ((unsigned long) lli_table_ptr->bus_address != 0xffffffff) {
+               dev_dbg(&sep->pdev->dev,
+                       "lli table %08lx, table_data_size is %lu\n",
+                       table_count, table_data_size);
+               dev_dbg(&sep->pdev->dev, "num_table_entries is %lu\n",
+                                                       num_table_entries);
+
+               /* Print entries of the table (without info entry) */
+               for (entries_count = 0; entries_count < num_table_entries;
+                       entries_count++, lli_table_ptr++) {
+
+                       dev_dbg(&sep->pdev->dev,
+                               "lli_table_ptr address is %08lx\n",
+                               (unsigned long) lli_table_ptr);
+
+                       dev_dbg(&sep->pdev->dev,
+                               "phys address is %08lx block size is %x\n",
+                               (unsigned long)lli_table_ptr->bus_address,
+                               lli_table_ptr->block_size);
+               }
+               /* Point to the info entry */
+               lli_table_ptr--;
+
+               dev_dbg(&sep->pdev->dev,
+                       "phys lli_table_ptr->block_size is %x\n",
+                       lli_table_ptr->block_size);
+
+               dev_dbg(&sep->pdev->dev,
+                       "phys lli_table_ptr->physical_address is %08lu\n",
+                       (unsigned long)lli_table_ptr->bus_address);
+
+
+               table_data_size = lli_table_ptr->block_size & 0xffffff;
+               num_table_entries = (lli_table_ptr->block_size >> 24) & 0xff;
+
+               dev_dbg(&sep->pdev->dev,
+                       "phys table_data_size is %lu num_table_entries is"
+                       " %lu bus_address is%lu\n", table_data_size,
+                       num_table_entries, (unsigned long)lli_table_ptr->bus_address);
+
+               if ((unsigned long)lli_table_ptr->bus_address != 0xffffffff)
+                       lli_table_ptr = (struct sep_lli_entry *)
+                               sep_shared_bus_to_virt(sep,
+                               (unsigned long)lli_table_ptr->bus_address);
+
+               table_count++;
+       }
+       dev_dbg(&sep->pdev->dev, "sep_debug_print_lli_tables end\n");
+}
+
+
+/**
+ *     sep_prepare_empty_lli_table - create a blank LLI table
+ *     @sep: pointer to struct sep_device
+ *     @lli_table_addr_ptr: pointer to lli table
+ *     @num_entries_ptr: pointer to number of entries
+ *     @table_data_size_ptr: point to table data size
+ *
+ *     This function creates empty lli tables when there is no data
+ */
+static void sep_prepare_empty_lli_table(struct sep_device *sep,
+               dma_addr_t *lli_table_addr_ptr,
+               u32 *num_entries_ptr,
+               u32 *table_data_size_ptr)
+{
+       struct sep_lli_entry *lli_table_ptr;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_empty_lli_table start\n");
+
+       /* Find the area for new table */
+       lli_table_ptr =
+               (struct sep_lli_entry *)(sep->shared_addr +
+               SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+               sep->num_lli_tables_created * sizeof(struct sep_lli_entry) *
+                       SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP);
+
+       lli_table_ptr->bus_address = 0;
+       lli_table_ptr->block_size = 0;
+
+       lli_table_ptr++;
+       lli_table_ptr->bus_address = 0xFFFFFFFF;
+       lli_table_ptr->block_size = 0;
+
+       /* Set the output parameter value */
+       *lli_table_addr_ptr = sep->shared_bus +
+               SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+               sep->num_lli_tables_created *
+               sizeof(struct sep_lli_entry) *
+               SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP;
+
+       /* Set the num of entries and table data size for empty table */
+       *num_entries_ptr = 2;
+       *table_data_size_ptr = 0;
+
+       /* Update the number of created tables */
+       sep->num_lli_tables_created++;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_empty_lli_table start\n");
+
+}
+
+/**
+ *     sep_prepare_input_dma_table - prepare input DMA mappings
+ *     @sep: pointer to struct sep_device
+ *     @data_size:
+ *     @block_size:
+ *     @lli_table_ptr:
+ *     @num_entries_ptr:
+ *     @table_data_size_ptr:
+ *     @is_kva: set for kernel data (kernel cryptio call)
+ *
+ *     This function prepares only input DMA table for synhronic symmetric
+ *     operations (HASH)
+ *     Note that all bus addresses that are passed to the SEP
+ *     are in 32 bit format; the SEP is a 32 bit device
+ */
+static int sep_prepare_input_dma_table(struct sep_device *sep,
+       unsigned long app_virt_addr,
+       u32 data_size,
+       u32 block_size,
+       dma_addr_t *lli_table_ptr,
+       u32 *num_entries_ptr,
+       u32 *table_data_size_ptr,
+       bool is_kva)
+{
+       int error = 0;
+       /* Pointer to the info entry of the table - the last entry */
+       struct sep_lli_entry *info_entry_ptr;
+       /* Array of pointers to page */
+       struct sep_lli_entry *lli_array_ptr;
+       /* Points to the first entry to be processed in the lli_in_array */
+       u32 current_entry = 0;
+       /* Num entries in the virtual buffer */
+       u32 sep_lli_entries = 0;
+       /* Lli table pointer */
+       struct sep_lli_entry *in_lli_table_ptr;
+       /* The total data in one table */
+       u32 table_data_size = 0;
+       /* Flag for last table */
+       u32 last_table_flag = 0;
+       /* Number of entries in lli table */
+       u32 num_entries_in_table = 0;
+       /* Next table address */
+       void *lli_table_alloc_addr = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_input_dma_table start\n");
+       dev_dbg(&sep->pdev->dev, "data_size is %x\n", data_size);
+       dev_dbg(&sep->pdev->dev, "block_size is %x\n", block_size);
+
+       /* Initialize the pages pointers */
+       sep->dma_res_arr[sep->nr_dcb_creat].in_page_array = NULL;
+       sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages = 0;
+
+       /* Set the kernel address for first table to be allocated */
+       lli_table_alloc_addr = (void *)(sep->shared_addr +
+               SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+               sep->num_lli_tables_created * sizeof(struct sep_lli_entry) *
+               SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP);
+
+       if (data_size == 0) {
+               /* Special case  - create meptu table - 2 entries, zero data */
+               sep_prepare_empty_lli_table(sep, lli_table_ptr,
+                               num_entries_ptr, table_data_size_ptr);
+               goto update_dcb_counter;
+       }
+
+       /* Check if the pages are in Kernel Virtual Address layout */
+       if (is_kva == true)
+               /* Lock the pages in the kernel */
+               error = sep_lock_kernel_pages(sep, app_virt_addr,
+                       data_size, &lli_array_ptr, SEP_DRIVER_IN_FLAG);
+       else
+               /*
+                * Lock the pages of the user buffer
+                * and translate them to pages
+                */
+               error = sep_lock_user_pages(sep, app_virt_addr,
+                       data_size, &lli_array_ptr, SEP_DRIVER_IN_FLAG);
+
+       if (error)
+               goto end_function;
+
+       dev_dbg(&sep->pdev->dev, "output sep_in_num_pages is %x\n",
+               sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages);
+
+       current_entry = 0;
+       info_entry_ptr = NULL;
+
+       sep_lli_entries = sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages;
+
+       /* Loop till all the entries in in array are not processed */
+       while (current_entry < sep_lli_entries) {
+
+               /* Set the new input and output tables */
+               in_lli_table_ptr =
+                       (struct sep_lli_entry *)lli_table_alloc_addr;
+
+               lli_table_alloc_addr += sizeof(struct sep_lli_entry) *
+                       SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP;
+
+               if (lli_table_alloc_addr >
+                       ((void *)sep->shared_addr +
+                       SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+                       SYNCHRONIC_DMA_TABLES_AREA_SIZE_BYTES)) {
+
+                       error = -ENOMEM;
+                       goto end_function_error;
+
+               }
+
+               /* Update the number of created tables */
+               sep->num_lli_tables_created++;
+
+               /* Calculate the maximum size of data for input table */
+               table_data_size = sep_calculate_lli_table_max_size(sep,
+                       &lli_array_ptr[current_entry],
+                       (sep_lli_entries - current_entry),
+                       &last_table_flag);
+
+               /*
+                * If this is not the last table -
+                * then allign it to the block size
+                */
+               if (!last_table_flag)
+                       table_data_size =
+                               (table_data_size / block_size) * block_size;
+
+               dev_dbg(&sep->pdev->dev, "output table_data_size is %x\n",
+                                                       table_data_size);
+
+               /* Construct input lli table */
+               sep_build_lli_table(sep, &lli_array_ptr[current_entry],
+                       in_lli_table_ptr,
+                       &current_entry, &num_entries_in_table, table_data_size);
+
+               if (info_entry_ptr == NULL) {
+
+                       /* Set the output parameters to physical addresses */
+                       *lli_table_ptr = sep_shared_area_virt_to_bus(sep,
+                               in_lli_table_ptr);
+                       *num_entries_ptr = num_entries_in_table;
+                       *table_data_size_ptr = table_data_size;
+
+                       dev_dbg(&sep->pdev->dev,
+                               "output lli_table_in_ptr is %08lx\n",
+                               (unsigned long)*lli_table_ptr);
+
+               } else {
+                       /* Update the info entry of the previous in table */
+                       info_entry_ptr->bus_address =
+                               sep_shared_area_virt_to_bus(sep,
+                                                       in_lli_table_ptr);
+                       info_entry_ptr->block_size =
+                               ((num_entries_in_table) << 24) |
+                               (table_data_size);
+               }
+               /* Save the pointer to the info entry of the current tables */
+               info_entry_ptr = in_lli_table_ptr + num_entries_in_table - 1;
+       }
+       /* Print input tables */
+       sep_debug_print_lli_tables(sep, (struct sep_lli_entry *)
+               sep_shared_area_bus_to_virt(sep, *lli_table_ptr),
+               *num_entries_ptr, *table_data_size_ptr);
+       /* The array of the pages */
+       kfree(lli_array_ptr);
+
+update_dcb_counter:
+       /* Update DCB counter */
+       sep->nr_dcb_creat++;
+       goto end_function;
+
+end_function_error:
+       /* Free all the allocated resources */
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].in_map_array);
+       kfree(lli_array_ptr);
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].in_page_array);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_prepare_input_dma_table end\n");
+       return error;
+
+}
+/**
+ *     sep_construct_dma_tables_from_lli - prepare AES/DES mappings
+ *     @sep: pointer to struct sep_device
+ *     @lli_in_array:
+ *     @sep_in_lli_entries:
+ *     @lli_out_array:
+ *     @sep_out_lli_entries
+ *     @block_size
+ *     @lli_table_in_ptr
+ *     @lli_table_out_ptr
+ *     @in_num_entries_ptr
+ *     @out_num_entries_ptr
+ *     @table_data_size_ptr
+ *
+ *     This function creates the input and output DMA tables for
+ *     symmetric operations (AES/DES) according to the block
+ *     size from LLI arays
+ *     Note that all bus addresses that are passed to the SEP
+ *     are in 32 bit format; the SEP is a 32 bit device
+ */
+static int sep_construct_dma_tables_from_lli(
+       struct sep_device *sep,
+       struct sep_lli_entry *lli_in_array,
+       u32     sep_in_lli_entries,
+       struct sep_lli_entry *lli_out_array,
+       u32     sep_out_lli_entries,
+       u32     block_size,
+       dma_addr_t *lli_table_in_ptr,
+       dma_addr_t *lli_table_out_ptr,
+       u32     *in_num_entries_ptr,
+       u32     *out_num_entries_ptr,
+       u32     *table_data_size_ptr)
+{
+       /* Points to the area where next lli table can be allocated */
+       void *lli_table_alloc_addr = 0;
+       /* Input lli table */
+       struct sep_lli_entry *in_lli_table_ptr = NULL;
+       /* Output lli table */
+       struct sep_lli_entry *out_lli_table_ptr = NULL;
+       /* Pointer to the info entry of the table - the last entry */
+       struct sep_lli_entry *info_in_entry_ptr = NULL;
+       /* Pointer to the info entry of the table - the last entry */
+       struct sep_lli_entry *info_out_entry_ptr = NULL;
+       /* Points to the first entry to be processed in the lli_in_array */
+       u32 current_in_entry = 0;
+       /* Points to the first entry to be processed in the lli_out_array */
+       u32 current_out_entry = 0;
+       /* Max size of the input table */
+       u32 in_table_data_size = 0;
+       /* Max size of the output table */
+       u32 out_table_data_size = 0;
+       /* Flag te signifies if this is the last tables build */
+       u32 last_table_flag = 0;
+       /* The data size that should be in table */
+       u32 table_data_size = 0;
+       /* Number of etnries in the input table */
+       u32 num_entries_in_table = 0;
+       /* Number of etnries in the output table */
+       u32 num_entries_out_table = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_construct_dma_tables_from_lli start\n");
+
+       /* Initiate to point after the message area */
+       lli_table_alloc_addr = (void *)(sep->shared_addr +
+               SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+               (sep->num_lli_tables_created *
+               (sizeof(struct sep_lli_entry) *
+               SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP)));
+
+       /* Loop till all the entries in in array are not processed */
+       while (current_in_entry < sep_in_lli_entries) {
+               /* Set the new input and output tables */
+               in_lli_table_ptr =
+                       (struct sep_lli_entry *)lli_table_alloc_addr;
+
+               lli_table_alloc_addr += sizeof(struct sep_lli_entry) *
+                       SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP;
+
+               /* Set the first output tables */
+               out_lli_table_ptr =
+                       (struct sep_lli_entry *)lli_table_alloc_addr;
+
+               /* Check if the DMA table area limit was overrun */
+               if ((lli_table_alloc_addr + sizeof(struct sep_lli_entry) *
+                       SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP) >
+                       ((void *)sep->shared_addr +
+                       SYNCHRONIC_DMA_TABLES_AREA_OFFSET_BYTES +
+                       SYNCHRONIC_DMA_TABLES_AREA_SIZE_BYTES)) {
+
+                       dev_warn(&sep->pdev->dev, "dma table limit overrun\n");
+                       return -ENOMEM;
+               }
+
+               /* Update the number of the lli tables created */
+               sep->num_lli_tables_created += 2;
+
+               lli_table_alloc_addr += sizeof(struct sep_lli_entry) *
+                       SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP;
+
+               /* Calculate the maximum size of data for input table */
+               in_table_data_size =
+                       sep_calculate_lli_table_max_size(sep,
+                       &lli_in_array[current_in_entry],
+                       (sep_in_lli_entries - current_in_entry),
+                       &last_table_flag);
+
+               /* Calculate the maximum size of data for output table */
+               out_table_data_size =
+                       sep_calculate_lli_table_max_size(sep,
+                       &lli_out_array[current_out_entry],
+                       (sep_out_lli_entries - current_out_entry),
+                       &last_table_flag);
+
+               dev_dbg(&sep->pdev->dev,
+                       "in_table_data_size is %x\n",
+                       in_table_data_size);
+
+               dev_dbg(&sep->pdev->dev,
+                       "out_table_data_size is %x\n",
+                       out_table_data_size);
+
+               table_data_size = in_table_data_size;
+
+               if (!last_table_flag) {
+                       /*
+                        * If this is not the last table,
+                        * then must check where the data is smallest
+                        * and then align it to the block size
+                        */
+                       if (table_data_size > out_table_data_size)
+                               table_data_size = out_table_data_size;
+
+                       /*
+                        * Now calculate the table size so that
+                        * it will be module block size
+                        */
+                       table_data_size = (table_data_size / block_size) *
+                               block_size;
+               }
+
+               dev_dbg(&sep->pdev->dev, "table_data_size is %x\n",
+                                                       table_data_size);
+
+               /* Construct input lli table */
+               sep_build_lli_table(sep, &lli_in_array[current_in_entry],
+                       in_lli_table_ptr,
+                       &current_in_entry,
+                       &num_entries_in_table,
+                       table_data_size);
+
+               /* Construct output lli table */
+               sep_build_lli_table(sep, &lli_out_array[current_out_entry],
+                       out_lli_table_ptr,
+                       &current_out_entry,
+                       &num_entries_out_table,
+                       table_data_size);
+
+               /* If info entry is null - this is the first table built */
+               if (info_in_entry_ptr == NULL) {
+                       /* Set the output parameters to physical addresses */
+                       *lli_table_in_ptr =
+                       sep_shared_area_virt_to_bus(sep, in_lli_table_ptr);
+
+                       *in_num_entries_ptr = num_entries_in_table;
+
+                       *lli_table_out_ptr =
+                               sep_shared_area_virt_to_bus(sep,
+                               out_lli_table_ptr);
+
+                       *out_num_entries_ptr = num_entries_out_table;
+                       *table_data_size_ptr = table_data_size;
+
+                       dev_dbg(&sep->pdev->dev,
+                       "output lli_table_in_ptr is %08lx\n",
+                               (unsigned long)*lli_table_in_ptr);
+                       dev_dbg(&sep->pdev->dev,
+                       "output lli_table_out_ptr is %08lx\n",
+                               (unsigned long)*lli_table_out_ptr);
+               } else {
+                       /* Update the info entry of the previous in table */
+                       info_in_entry_ptr->bus_address =
+                               sep_shared_area_virt_to_bus(sep,
+                               in_lli_table_ptr);
+
+                       info_in_entry_ptr->block_size =
+                               ((num_entries_in_table) << 24) |
+                               (table_data_size);
+
+                       /* Update the info entry of the previous in table */
+                       info_out_entry_ptr->bus_address =
+                               sep_shared_area_virt_to_bus(sep,
+                               out_lli_table_ptr);
+
+                       info_out_entry_ptr->block_size =
+                               ((num_entries_out_table) << 24) |
+                               (table_data_size);
+
+                       dev_dbg(&sep->pdev->dev,
+                               "output lli_table_in_ptr:%08lx %08x\n",
+                               (unsigned long)info_in_entry_ptr->bus_address,
+                               info_in_entry_ptr->block_size);
+
+                       dev_dbg(&sep->pdev->dev,
+                               "output lli_table_out_ptr:%08lx  %08x\n",
+                               (unsigned long)info_out_entry_ptr->bus_address,
+                               info_out_entry_ptr->block_size);
+               }
+
+               /* Save the pointer to the info entry of the current tables */
+               info_in_entry_ptr = in_lli_table_ptr +
+                       num_entries_in_table - 1;
+               info_out_entry_ptr = out_lli_table_ptr +
+                       num_entries_out_table - 1;
+
+               dev_dbg(&sep->pdev->dev,
+                       "output num_entries_out_table is %x\n",
+                       (u32)num_entries_out_table);
+               dev_dbg(&sep->pdev->dev,
+                       "output info_in_entry_ptr is %lx\n",
+                       (unsigned long)info_in_entry_ptr);
+               dev_dbg(&sep->pdev->dev,
+                       "output info_out_entry_ptr is %lx\n",
+                       (unsigned long)info_out_entry_ptr);
+       }
+
+       /* Print input tables */
+       sep_debug_print_lli_tables(sep,
+       (struct sep_lli_entry *)
+       sep_shared_area_bus_to_virt(sep, *lli_table_in_ptr),
+       *in_num_entries_ptr,
+       *table_data_size_ptr);
+
+       /* Print output tables */
+       sep_debug_print_lli_tables(sep,
+       (struct sep_lli_entry *)
+       sep_shared_area_bus_to_virt(sep, *lli_table_out_ptr),
+       *out_num_entries_ptr,
+       *table_data_size_ptr);
+
+       dev_dbg(&sep->pdev->dev, "sep_construct_dma_tables_from_lli end\n");
+       return 0;
+}
+
+/**
+ *     sep_prepare_input_output_dma_table - prepare DMA I/O table
+ *     @app_virt_in_addr:
+ *     @app_virt_out_addr:
+ *     @data_size:
+ *     @block_size:
+ *     @lli_table_in_ptr:
+ *     @lli_table_out_ptr:
+ *     @in_num_entries_ptr:
+ *     @out_num_entries_ptr:
+ *     @table_data_size_ptr:
+ *     @is_kva: set for kernel data; used only for kernel crypto module
+ *
+ *     This function builds input and output DMA tables for synhronic
+ *     symmetric operations (AES, DES, HASH). It also checks that each table
+ *     is of the modular block size
+ *     Note that all bus addresses that are passed to the SEP
+ *     are in 32 bit format; the SEP is a 32 bit device
+ */
+static int sep_prepare_input_output_dma_table(struct sep_device *sep,
+       unsigned long app_virt_in_addr,
+       unsigned long app_virt_out_addr,
+       u32 data_size,
+       u32 block_size,
+       dma_addr_t *lli_table_in_ptr,
+       dma_addr_t *lli_table_out_ptr,
+       u32 *in_num_entries_ptr,
+       u32 *out_num_entries_ptr,
+       u32 *table_data_size_ptr,
+       bool is_kva)
+
+{
+       int error = 0;
+       /* Array of pointers of page */
+       struct sep_lli_entry *lli_in_array;
+       /* Array of pointers of page */
+       struct sep_lli_entry *lli_out_array;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_input_output_dma_table start\n");
+
+       if (data_size == 0) {
+               /* Prepare empty table for input and output */
+               sep_prepare_empty_lli_table(sep, lli_table_in_ptr,
+                       in_num_entries_ptr, table_data_size_ptr);
+
+               sep_prepare_empty_lli_table(sep, lli_table_out_ptr,
+                       out_num_entries_ptr, table_data_size_ptr);
+
+               goto update_dcb_counter;
+       }
+
+       /* Initialize the pages pointers */
+       sep->dma_res_arr[sep->nr_dcb_creat].in_page_array = NULL;
+       sep->dma_res_arr[sep->nr_dcb_creat].out_page_array = NULL;
+
+       /* Lock the pages of the buffer and translate them to pages */
+       if (is_kva == true) {
+               error = sep_lock_kernel_pages(sep, app_virt_in_addr,
+                       data_size, &lli_in_array, SEP_DRIVER_IN_FLAG);
+
+               if (error) {
+                       dev_warn(&sep->pdev->dev,
+                               "lock kernel for in failed\n");
+                       goto end_function;
+               }
+
+               error = sep_lock_kernel_pages(sep, app_virt_out_addr,
+                       data_size, &lli_out_array, SEP_DRIVER_OUT_FLAG);
+
+               if (error) {
+                       dev_warn(&sep->pdev->dev,
+                               "lock kernel for out failed\n");
+                       goto end_function;
+               }
+       }
+
+       else {
+               error = sep_lock_user_pages(sep, app_virt_in_addr,
+                               data_size, &lli_in_array, SEP_DRIVER_IN_FLAG);
+               if (error) {
+                       dev_warn(&sep->pdev->dev,
+                               "sep_lock_user_pages for input virtual buffer failed\n");
+                       goto end_function;
+               }
+
+               error = sep_lock_user_pages(sep, app_virt_out_addr,
+                       data_size, &lli_out_array, SEP_DRIVER_OUT_FLAG);
+
+               if (error) {
+                       dev_warn(&sep->pdev->dev,
+                               "sep_lock_user_pages for output virtual buffer failed\n");
+                       goto end_function_free_lli_in;
+               }
+       }
+
+       dev_dbg(&sep->pdev->dev, "sep_in_num_pages is %x\n",
+               sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages);
+       dev_dbg(&sep->pdev->dev, "sep_out_num_pages is %x\n",
+               sep->dma_res_arr[sep->nr_dcb_creat].out_num_pages);
+       dev_dbg(&sep->pdev->dev, "SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP is %x\n",
+               SEP_DRIVER_ENTRIES_PER_TABLE_IN_SEP);
+
+       /* Call the fucntion that creates table from the lli arrays */
+       error = sep_construct_dma_tables_from_lli(sep, lli_in_array,
+               sep->dma_res_arr[sep->nr_dcb_creat].in_num_pages,
+               lli_out_array,
+               sep->dma_res_arr[sep->nr_dcb_creat].out_num_pages,
+               block_size, lli_table_in_ptr, lli_table_out_ptr,
+               in_num_entries_ptr, out_num_entries_ptr, table_data_size_ptr);
+
+       if (error) {
+               dev_warn(&sep->pdev->dev,
+                       "sep_construct_dma_tables_from_lli failed\n");
+               goto end_function_with_error;
+       }
+
+       kfree(lli_out_array);
+       kfree(lli_in_array);
+
+update_dcb_counter:
+       /* Update DCB counter */
+       sep->nr_dcb_creat++;
+       /* Fall through - free the lli entry arrays */
+       dev_dbg(&sep->pdev->dev, "in_num_entries_ptr is %08x\n",
+                                               *in_num_entries_ptr);
+       dev_dbg(&sep->pdev->dev, "out_num_entries_ptr is %08x\n",
+                                               *out_num_entries_ptr);
+       dev_dbg(&sep->pdev->dev, "table_data_size_ptr is %08x\n",
+                                               *table_data_size_ptr);
+
+       goto end_function;
+
+end_function_with_error:
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].out_map_array);
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].out_page_array);
+       kfree(lli_out_array);
+
+
+end_function_free_lli_in:
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].in_map_array);
+       kfree(sep->dma_res_arr[sep->nr_dcb_creat].in_page_array);
+       kfree(lli_in_array);
+
+end_function:
+       dev_dbg(&sep->pdev->dev,
+               "sep_prepare_input_output_dma_table end result = %d\n", error);
+
+       return error;
+
+}
+
+/**
+ *     sep_prepare_input_output_dma_table_in_dcb - prepare control blocks
+ *     @app_in_address: unsigned long; for data buffer in (user space)
+ *     @app_out_address: unsigned long; for data buffer out (user space)
+ *     @data_in_size: u32; for size of data
+ *     @block_size: u32; for block size
+ *     @tail_block_size: u32; for size of tail block
+ *     @isapplet: bool; to indicate external app
+ *     @is_kva: bool; kernel buffer; only used for kernel crypto module
+ *
+ *     This function prepares the linked DMA tables and puts the
+ *     address for the linked list of tables inta a DCB (data control
+ *     block) the address of which is known by the SEP hardware
+ *     Note that all bus addresses that are passed to the SEP
+ *     are in 32 bit format; the SEP is a 32 bit device
+ */
+static int sep_prepare_input_output_dma_table_in_dcb(struct sep_device *sep,
+       unsigned long  app_in_address,
+       unsigned long  app_out_address,
+       u32  data_in_size,
+       u32  block_size,
+       u32  tail_block_size,
+       bool isapplet,
+       bool    is_kva)
+{
+       int error = 0;
+       /* Size of tail */
+       u32 tail_size = 0;
+       /* Address of the created DCB table */
+       struct sep_dcblock *dcb_table_ptr = NULL;
+       /* The physical address of the first input DMA table */
+       dma_addr_t in_first_mlli_address = 0;
+       /* Number of entries in the first input DMA table */
+       u32  in_first_num_entries = 0;
+       /* The physical address of the first output DMA table */
+       dma_addr_t  out_first_mlli_address = 0;
+       /* Number of entries in the first output DMA table */
+       u32  out_first_num_entries = 0;
+       /* Data in the first input/output table */
+       u32  first_data_size = 0;
+
+       dev_dbg(&sep->pdev->dev, "prepare_input_output_dma_table_in_dcb start\n");
+
+       if (sep->nr_dcb_creat == SEP_MAX_NUM_SYNC_DMA_OPS) {
+               /* No more DCBs to allocate */
+               dev_warn(&sep->pdev->dev, "no more DCBs available\n");
+               error = -ENOSPC;
+               goto end_function;
+       }
+
+       /* Allocate new DCB */
+       dcb_table_ptr = (struct sep_dcblock *)(sep->shared_addr +
+               SEP_DRIVER_SYSTEM_DCB_MEMORY_OFFSET_IN_BYTES +
+               (sep->nr_dcb_creat * sizeof(struct sep_dcblock)));
+
+       /* Set the default values in the DCB */
+       dcb_table_ptr->input_mlli_address = 0;
+       dcb_table_ptr->input_mlli_num_entries = 0;
+       dcb_table_ptr->input_mlli_data_size = 0;
+       dcb_table_ptr->output_mlli_address = 0;
+       dcb_table_ptr->output_mlli_num_entries = 0;
+       dcb_table_ptr->output_mlli_data_size = 0;
+       dcb_table_ptr->tail_data_size = 0;
+       dcb_table_ptr->out_vr_tail_pt = 0;
+
+       if (isapplet == true) {
+               tail_size = data_in_size % block_size;
+               if (tail_size) {
+                       if (data_in_size < tail_block_size) {
+                               dev_warn(&sep->pdev->dev, "data in size smaller than tail block size\n");
+                               error = -ENOSPC;
+                               goto end_function;
+                       }
+                       if (tail_block_size)
+                               /*
+                                * Case the tail size should be
+                                * bigger than the real block size
+                                */
+                               tail_size = tail_block_size +
+                                       ((data_in_size -
+                                               tail_block_size) % block_size);
+               }
+
+               /* Check if there is enough data for DMA operation */
+               if (data_in_size < SEP_DRIVER_MIN_DATA_SIZE_PER_TABLE) {
+                       if (is_kva == true) {
+                               memcpy(dcb_table_ptr->tail_data,
+                                       (void *)app_in_address, data_in_size);
+                       } else {
+                               if (copy_from_user(dcb_table_ptr->tail_data,
+                                       (void __user *)app_in_address,
+                                       data_in_size)) {
+                                       error = -EFAULT;
+                                       goto end_function;
+                               }
+                       }
+
+                       dcb_table_ptr->tail_data_size = data_in_size;
+
+                       /* Set the output user-space address for mem2mem op */
+                       if (app_out_address)
+                               dcb_table_ptr->out_vr_tail_pt =
+                                                       (u32)app_out_address;
+
+                       /*
+                        * Update both data length parameters in order to avoid
+                        * second data copy and allow building of empty mlli
+                        * tables
+                        */
+                       tail_size = 0x0;
+                       data_in_size = 0x0;
+               }
+               if (tail_size) {
+                       if (is_kva == true) {
+                               memcpy(dcb_table_ptr->tail_data,
+                                       (void *)(app_in_address + data_in_size -
+                                       tail_size), tail_size);
+                       } else {
+                               /* We have tail data - copy it to DCB */
+                               if (copy_from_user(dcb_table_ptr->tail_data,
+                                       (void *)(app_in_address +
+                                       data_in_size - tail_size), tail_size)) {
+                                       error = -EFAULT;
+                                       goto end_function;
+                               }
+                       }
+                       if (app_out_address)
+                               /*
+                                * Calculate the output address
+                                * according to tail data size
+                                */
+                               dcb_table_ptr->out_vr_tail_pt =
+                                       (u32)app_out_address + data_in_size
+                                       - tail_size;
+
+                       /* Save the real tail data size */
+                       dcb_table_ptr->tail_data_size = tail_size;
+                       /*
+                        * Update the data size without the tail
+                        * data size AKA data for the dma
+                        */
+                       data_in_size = (data_in_size - tail_size);
+               }
+       }
+       /* Check if we need to build only input table or input/output */
+       if (app_out_address) {
+               /* Prepare input/output tables */
+               error = sep_prepare_input_output_dma_table(sep,
+                       app_in_address,
+                       app_out_address,
+                       data_in_size,
+                       block_size,
+                       &in_first_mlli_address,
+                       &out_first_mlli_address,
+                       &in_first_num_entries,
+                       &out_first_num_entries,
+                       &first_data_size,
+                       is_kva);
+       } else {
+               /* Prepare input tables */
+               error = sep_prepare_input_dma_table(sep,
+                       app_in_address,
+                       data_in_size,
+                       block_size,
+                       &in_first_mlli_address,
+                       &in_first_num_entries,
+                       &first_data_size,
+                       is_kva);
+       }
+
+       if (error) {
+               dev_warn(&sep->pdev->dev, "prepare DMA table call failed from prepare DCB call\n");
+               goto end_function;
+       }
+
+       /* Set the DCB values */
+       dcb_table_ptr->input_mlli_address = in_first_mlli_address;
+       dcb_table_ptr->input_mlli_num_entries = in_first_num_entries;
+       dcb_table_ptr->input_mlli_data_size = first_data_size;
+       dcb_table_ptr->output_mlli_address = out_first_mlli_address;
+       dcb_table_ptr->output_mlli_num_entries = out_first_num_entries;
+       dcb_table_ptr->output_mlli_data_size = first_data_size;
+
+end_function:
+       dev_dbg(&sep->pdev->dev,
+               "sep_prepare_input_output_dma_table_in_dcb end\n");
+       return error;
+
+}
+
+
+/**
+ *     sep_create_sync_dma_tables_handler - create sync DMA tables
+ *     @sep: pointer to struct sep_device
+ *     @arg: pointer to struct bld_syn_tab_struct
+ *
+ *     Handle the request for creation of the DMA tables for the synchronic
+ *     symmetric operations (AES,DES). Note that all bus addresses that are
+ *     passed to the SEP are in 32 bit format; the SEP is a 32 bit device
+ */
+static int sep_create_sync_dma_tables_handler(struct sep_device *sep,
+                                               unsigned long arg)
+{
+       int error = 0;
+
+       /* Command arguments */
+       struct bld_syn_tab_struct command_args;
+
+       dev_dbg(&sep->pdev->dev,
+               "sep_create_sync_dma_tables_handler start\n");
+
+       if (copy_from_user(&command_args, (void __user *)arg,
+                                       sizeof(struct bld_syn_tab_struct))) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       dev_dbg(&sep->pdev->dev, "app_in_address is %08llx\n",
+                                               command_args.app_in_address);
+       dev_dbg(&sep->pdev->dev, "app_out_address is %08llx\n",
+                                               command_args.app_out_address);
+       dev_dbg(&sep->pdev->dev, "data_size is %u\n",
+                                               command_args.data_in_size);
+       dev_dbg(&sep->pdev->dev, "block_size is %u\n",
+                                               command_args.block_size);
+
+       /* Validate user parameters */
+       if (!command_args.app_in_address) {
+               error = -EINVAL;
+               goto end_function;
+       }
+
+       error = sep_prepare_input_output_dma_table_in_dcb(sep,
+               (unsigned long)command_args.app_in_address,
+               (unsigned long)command_args.app_out_address,
+               command_args.data_in_size,
+               command_args.block_size,
+               0x0,
+               false,
+               false);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_create_sync_dma_tables_handler end\n");
+       return error;
+}
+
+/**
+ *     sep_free_dma_tables_and_dcb - free DMA tables and DCBs
+ *     @sep: pointer to struct sep_device
+ *     @isapplet: indicates external application (used for kernel access)
+ *     @is_kva: indicates kernel addresses (only used for kernel crypto)
+ *
+ *     This function frees the DMA tables and DCB
+ */
+static int sep_free_dma_tables_and_dcb(struct sep_device *sep, bool isapplet,
+       bool is_kva)
+{
+       int i = 0;
+       int error = 0;
+       int error_temp = 0;
+       struct sep_dcblock *dcb_table_ptr;
+       unsigned long pt_hold;
+       void *tail_pt;
+
+       dev_dbg(&sep->pdev->dev, "sep_free_dma_tables_and_dcb start\n");
+
+       if (isapplet == true) {
+               /* Set pointer to first DCB table */
+               dcb_table_ptr = (struct sep_dcblock *)
+                       (sep->shared_addr +
+                       SEP_DRIVER_SYSTEM_DCB_MEMORY_OFFSET_IN_BYTES);
+
+               /* Go over each DCB and see if tail pointer must be updated */
+               for (i = 0; i < sep->nr_dcb_creat; i++, dcb_table_ptr++) {
+                       if (dcb_table_ptr->out_vr_tail_pt) {
+                               pt_hold = (unsigned long)dcb_table_ptr->out_vr_tail_pt;
+                               tail_pt = (void *)pt_hold;
+                               if (is_kva == true) {
+                                       memcpy(tail_pt,
+                                               dcb_table_ptr->tail_data,
+                                               dcb_table_ptr->tail_data_size);
+                               } else {
+                                       error_temp = copy_to_user(
+                                               tail_pt,
+                                               dcb_table_ptr->tail_data,
+                                               dcb_table_ptr->tail_data_size);
+                               }
+                               if (error_temp) {
+                                       /* Release the DMA resource */
+                                       error = -EFAULT;
+                                       break;
+                               }
+                       }
+               }
+       }
+       /* Free the output pages, if any */
+       sep_free_dma_table_data_handler(sep);
+
+       dev_dbg(&sep->pdev->dev, "sep_free_dma_tables_and_dcb end\n");
+       return error;
+}
+
+/**
+ *     sep_get_static_pool_addr_handler - get static pool address
+ *     @sep: pointer to struct sep_device
+ *
+ *     This function sets the bus and virtual addresses of the static pool
+ */
+static int sep_get_static_pool_addr_handler(struct sep_device *sep)
+{
+       u32 *static_pool_addr = NULL;
+
+       dev_dbg(&sep->pdev->dev, "sep_get_static_pool_addr_handler start\n");
+
+       static_pool_addr = (u32 *)(sep->shared_addr +
+               SEP_DRIVER_SYSTEM_RAR_MEMORY_OFFSET_IN_BYTES);
+
+       static_pool_addr[0] = SEP_STATIC_POOL_VAL_TOKEN;
+       static_pool_addr[1] = (u32)sep->shared_bus +
+               SEP_DRIVER_STATIC_AREA_OFFSET_IN_BYTES;
+
+       dev_dbg(&sep->pdev->dev, "static pool: physical %x\n",
+               (u32)static_pool_addr[1]);
+
+       dev_dbg(&sep->pdev->dev, "sep_get_static_pool_addr_handler end\n");
+
+       return 0;
+}
+
+/**
+ *     sep_start_handler - start device
+ *     @sep: pointer to struct sep_device
+ */
+static int sep_start_handler(struct sep_device *sep)
+{
+       unsigned long reg_val;
+       unsigned long error = 0;
+
+       dev_dbg(&sep->pdev->dev, "sep_start_handler start\n");
+
+       /* Wait in polling for message from SEP */
+       do {
+               reg_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR3_REG_ADDR);
+       } while (!reg_val);
+
+       /* Check the value */
+       if (reg_val == 0x1)
+               /* Fatal error - read error status from GPRO */
+               error = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR0_REG_ADDR);
+       dev_dbg(&sep->pdev->dev, "sep_start_handler end\n");
+       return error;
+}
+
+/**
+ *     ep_check_sum_calc - checksum messages
+ *     @data: buffer to checksum
+ *     @length: buffer size
+ *
+ *     This function performs a checksum for messages that are sent
+ *     to the SEP.
+ */
+static u32 sep_check_sum_calc(u8 *data, u32 length)
+{
+       u32 sum = 0;
+       u16 *Tdata = (u16 *)data;
+
+       while (length > 1) {
+               /*  This is the inner loop */
+               sum += *Tdata++;
+               length -= 2;
+       }
+
+       /*  Add left-over byte, if any */
+       if (length > 0)
+               sum += *(u8 *)Tdata;
+
+       /*  Fold 32-bit sum to 16 bits */
+       while (sum>>16)
+               sum = (sum & 0xffff) + (sum >> 16);
+
+       return ~sum & 0xFFFF;
+}
+
+/**
+ *     sep_init_handler -
+ *     @sep: pointer to struct sep_device
+ *     @arg: parameters from user space application
+ *
+ *     Handles the request for SEP initialization
+ *     Note that this will go away for Medfield once the SCU
+ *     SEP initialization is complete
+ *     Also note that the message to the SEP has components
+ *     from user space as well as components written by the driver
+ *     This is becuase the portions of the message that pertain to
+ *     physical addresses must be set by the driver after the message
+ *     leaves custody of the user space application for security
+ *     reasons.
+ */
+static int sep_init_handler(struct sep_device *sep, unsigned long arg)
+{
+       u32 message_buff[14];
+       u32 counter;
+       int error = 0;
+       u32 reg_val;
+       dma_addr_t new_base_addr;
+       unsigned long addr_hold;
+       struct init_struct command_args;
+
+       dev_dbg(&sep->pdev->dev, "sep_init_handler start\n");
+
+       /* Make sure that we have not initialized already */
+       reg_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR3_REG_ADDR);
+
+       if (reg_val != 0x2) {
+               error = SEP_ALREADY_INITIALIZED_ERR;
+               dev_warn(&sep->pdev->dev, "init; device already initialized\n");
+               goto end_function;
+       }
+
+       /* Only root can initialize */
+       if (!capable(CAP_SYS_ADMIN)) {
+               error = -EACCES;
+               goto end_function;
+       }
+
+       /* Copy in the parameters */
+       error = copy_from_user(&command_args, (void __user *)arg,
+               sizeof(struct init_struct));
+
+       if (error) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       /* Validate parameters */
+       if (!command_args.message_addr || !command_args.sep_sram_addr ||
+               command_args.message_size_in_words > 14) {
+               error = -EINVAL;
+               goto end_function;
+       }
+
+       /* Copy in the SEP init message */
+       addr_hold = (unsigned long)command_args.message_addr;
+       error = copy_from_user(message_buff,
+               (void __user *)addr_hold,
+               command_args.message_size_in_words*sizeof(u32));
+
+       if (error) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       /* Load resident, cache, and extapp firmware */
+       error = sep_load_firmware(sep);
+
+       if (error) {
+               dev_warn(&sep->pdev->dev,
+                       "init; copy SEP init message failed %x\n", error);
+               goto end_function;
+       }
+
+       /* Compute the base address */
+       new_base_addr = sep->shared_bus;
+
+       if (sep->resident_bus < new_base_addr)
+               new_base_addr = sep->resident_bus;
+
+       if (sep->cache_bus < new_base_addr)
+               new_base_addr = sep->cache_bus;
+
+       if (sep->dcache_bus < new_base_addr)
+               new_base_addr = sep->dcache_bus;
+
+       /* Put physical addresses in SEP message */
+       message_buff[3] = (u32)new_base_addr;
+       message_buff[4] = (u32)sep->shared_bus;
+       message_buff[6] = (u32)sep->resident_bus;
+       message_buff[7] = (u32)sep->cache_bus;
+       message_buff[8] = (u32)sep->dcache_bus;
+
+       message_buff[command_args.message_size_in_words - 1] = 0x0;
+       message_buff[command_args.message_size_in_words - 1] =
+               sep_check_sum_calc((u8 *)message_buff,
+               command_args.message_size_in_words*sizeof(u32));
+
+       /* Debug print of message */
+       for (counter = 0; counter < command_args.message_size_in_words;
+                                                               counter++)
+               dev_dbg(&sep->pdev->dev, "init; SEP message word %d is %x\n",
+                       counter, message_buff[counter]);
+
+       /* Tell the SEP the sram address */
+       sep_write_reg(sep, HW_SRAM_ADDR_REG_ADDR, command_args.sep_sram_addr);
+
+       /* Push the message to the SEP */
+       for (counter = 0; counter < command_args.message_size_in_words;
+                                                               counter++) {
+               sep_write_reg(sep, HW_SRAM_DATA_REG_ADDR,
+                                               message_buff[counter]);
+               sep_wait_sram_write(sep);
+       }
+
+       /* Signal SEP that message is ready and to init */
+       sep_write_reg(sep, HW_HOST_HOST_SEP_GPR0_REG_ADDR, 0x1);
+
+       /* Wait for acknowledge */
+       dev_dbg(&sep->pdev->dev, "init; waiting for msg response\n");
+
+       do {
+               reg_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR3_REG_ADDR);
+       } while (!(reg_val & 0xFFFFFFFD));
+
+       if (reg_val == 0x1) {
+               dev_warn(&sep->pdev->dev, "init; device int failed\n");
+               error = sep_read_reg(sep, 0x8060);
+               dev_warn(&sep->pdev->dev, "init; sw monitor is %x\n", error);
+               error = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR0_REG_ADDR);
+               dev_warn(&sep->pdev->dev, "init; error is %x\n", error);
+               goto end_function;
+       }
+       dev_dbg(&sep->pdev->dev, "init; end CC INIT, reg_val is %x\n", reg_val);
+
+       /* Signal SEP to zero the GPR3 */
+       sep_write_reg(sep, HW_HOST_HOST_SEP_GPR0_REG_ADDR, 0x10);
+
+       /* Wait for response */
+       dev_dbg(&sep->pdev->dev, "init; waiting for zero set response\n");
+
+       do {
+               reg_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR3_REG_ADDR);
+       } while (reg_val != 0);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "init is done\n");
+       return error;
+}
+
+/**
+ *     sep_end_transaction_handler - end transaction
+ *     @sep: pointer to struct sep_device
+ *
+ *     This API handles the end transaction request
+ */
+static int sep_end_transaction_handler(struct sep_device *sep)
+{
+       dev_dbg(&sep->pdev->dev, "sep_end_transaction_handler start\n");
+
+       /* Clear the data pool pointers Token */
+       memset((void *)(sep->shared_addr +
+               SEP_DRIVER_DATA_POOL_ALLOCATION_OFFSET_IN_BYTES),
+               0, sep->num_of_data_allocations*2*sizeof(u32));
+
+       /* Check that all the DMA resources were freed */
+       sep_free_dma_table_data_handler(sep);
+
+       clear_bit(SEP_MMAP_LOCK_BIT, &sep->in_use_flags);
+
+       /*
+        * We are now through with the transaction. Let's
+        * allow other processes who have the device open
+        * to perform transactions
+        */
+       mutex_lock(&sep->sep_mutex);
+       sep->pid_doing_transaction = 0;
+       mutex_unlock(&sep->sep_mutex);
+       /* Raise event for stuck contextes */
+       wake_up(&sep->event);
+
+       dev_dbg(&sep->pdev->dev, "waking up event\n");
+       dev_dbg(&sep->pdev->dev, "sep_end_transaction_handler end\n");
+
+       return 0;
+}
+
+/**
+ *     sep_prepare_dcb_handler - prepare a control block
+ *     @sep: pointer to struct sep_device
+ *     @arg: pointer to user parameters
+ *
+ *     This function will retrieve the RAR buffer physical addresses, type
+ *     & size corresponding to the RAR handles provided in the buffers vector.
+ */
+static int sep_prepare_dcb_handler(struct sep_device *sep, unsigned long arg)
+{
+       int error;
+       /* Command arguments */
+       struct build_dcb_struct command_args;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_dcb_handler start\n");
+
+       /* Get the command arguments */
+       if (copy_from_user(&command_args, (void __user *)arg,
+                                       sizeof(struct build_dcb_struct))) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       dev_dbg(&sep->pdev->dev, "app_in_address is %08llx\n",
+                                               command_args.app_in_address);
+       dev_dbg(&sep->pdev->dev, "app_out_address is %08llx\n",
+                                               command_args.app_out_address);
+       dev_dbg(&sep->pdev->dev, "data_size is %x\n",
+                                               command_args.data_in_size);
+       dev_dbg(&sep->pdev->dev, "block_size is %x\n",
+                                               command_args.block_size);
+       dev_dbg(&sep->pdev->dev, "tail block_size is %x\n",
+                                               command_args.tail_block_size);
+
+       error = sep_prepare_input_output_dma_table_in_dcb(sep,
+               (unsigned long)command_args.app_in_address,
+               (unsigned long)command_args.app_out_address,
+               command_args.data_in_size, command_args.block_size,
+               command_args.tail_block_size, true, false);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_prepare_dcb_handler end\n");
+       return error;
+
+}
+
+/**
+ *     sep_free_dcb_handler - free control block resources
+ *     @sep: pointer to struct sep_device
+ *
+ *     This function frees the DCB resources and updates the needed
+ *     user-space buffers.
+ */
+static int sep_free_dcb_handler(struct sep_device *sep)
+{
+       int error ;
+
+       dev_dbg(&sep->pdev->dev, "sep_prepare_dcb_handler start\n");
+       dev_dbg(&sep->pdev->dev, "num of DCBs %x\n", sep->nr_dcb_creat);
+
+       error = sep_free_dma_tables_and_dcb(sep, false, false);
+
+       dev_dbg(&sep->pdev->dev, "sep_free_dcb_handler end\n");
+       return error;
+}
+
+/**
+ *     sep_rar_prepare_output_msg_handler - prepare an output message
+ *     @sep: pointer to struct sep_device
+ *     @arg: pointer to user parameters
+ *
+ *     This function will retrieve the RAR buffer physical addresses, type
+ *     & size corresponding to the RAR handles provided in the buffers vector.
+ */
+static int sep_rar_prepare_output_msg_handler(struct sep_device *sep,
+       unsigned long arg)
+{
+       int error = 0;
+       /* Command args */
+       struct rar_hndl_to_bus_struct command_args;
+       struct RAR_buffer rar_buf;
+       /* Bus address */
+       dma_addr_t  rar_bus = 0;
+       /* Holds the RAR address in the system memory offset */
+       u32 *rar_addr;
+
+       dev_dbg(&sep->pdev->dev, "sep_rar_prepare_output_msg_handler start\n");
+
+       /* Copy the data */
+       if (copy_from_user(&command_args, (void __user *)arg,
+                                               sizeof(command_args))) {
+               error = -EFAULT;
+               goto end_function;
+       }
+
+       /* Call to translation function only if user handle is not NULL */
+       if (command_args.rar_handle) {
+               memset(&rar_buf, 0, sizeof(rar_buf));
+               rar_buf.info.handle = (u32)command_args.rar_handle;
+
+               if (rar_handle_to_bus(&rar_buf, 1) != 1) {
+                       dev_dbg(&sep->pdev->dev, "rar_handle_to_bus failure\n");
+                       error = -EFAULT;
+                       goto end_function;
+               }
+               rar_bus = rar_buf.bus_address;
+       }
+       dev_dbg(&sep->pdev->dev, "rar msg; rar_addr_bus = %x\n", (u32)rar_bus);
+
+       /* Set value in the SYSTEM MEMORY offset */
+       rar_addr = (u32 *)(sep->shared_addr +
+               SEP_DRIVER_SYSTEM_RAR_MEMORY_OFFSET_IN_BYTES);
+
+       /* Copy the physical address to the System Area for the SEP */
+       rar_addr[0] = SEP_RAR_VAL_TOKEN;
+       rar_addr[1] = rar_bus;
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "sep_rar_prepare_output_msg_handler start\n");
+       return error;
+}
+
+/**
+ *     sep_realloc_ext_cache_handler - report location of extcache
+ *     @sep: pointer to struct sep_device
+ *     @arg: pointer to user parameters
+ *
+ *     This function tells the SEP where the extapp is located
+ */
+static int sep_realloc_ext_cache_handler(struct sep_device *sep,
+       unsigned long arg)
+{
+       /* Holds the new ext cache address in the system memory offset */
+       u32 *system_addr;
+
+       /* Set value in the SYSTEM MEMORY offset */
+       system_addr = (u32 *)(sep->shared_addr +
+               SEP_DRIVER_SYSTEM_EXT_CACHE_ADDR_OFFSET_IN_BYTES);
+
+       /* Copy the physical address to the System Area for the SEP */
+       system_addr[0] = SEP_EXT_CACHE_ADDR_VAL_TOKEN;
+       dev_dbg(&sep->pdev->dev, "ext cache init; system addr 0 is %x\n",
+                                                       system_addr[0]);
+       system_addr[1] = sep->extapp_bus;
+       dev_dbg(&sep->pdev->dev, "ext cache init; system addr 1 is %x\n",
+                                                       system_addr[1]);
+
+       return 0;
+}
+
+/**
+ *     sep_ioctl - ioctl api
+ *     @filp: pointer to struct file
+ *     @cmd: command
+ *     @arg: pointer to argument structure
+ *
+ *     Implement the ioctl methods availble on the SEP device.
+ */
+static long sep_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
+{
+       int error = 0;
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "ioctl start\n");
+
+       dev_dbg(&sep->pdev->dev, "cmd is %x\n", cmd);
+
+       /* Make sure we own this device */
+       mutex_lock(&sep->sep_mutex);
+       if ((current->pid != sep->pid_doing_transaction) &&
+                               (sep->pid_doing_transaction != 0)) {
+               dev_dbg(&sep->pdev->dev, "ioctl pid is not owner\n");
+               mutex_unlock(&sep->sep_mutex);
+               error = -EACCES;
+               goto end_function;
+       }
+
+       mutex_unlock(&sep->sep_mutex);
+
+       /* Check that the command is for SEP device */
+       if (_IOC_TYPE(cmd) != SEP_IOC_MAGIC_NUMBER) {
+               error = -ENOTTY;
+               goto end_function;
+       }
+
+       /* Lock to prevent the daemon to interfere with operation */
+       mutex_lock(&sep->ioctl_mutex);
+
+       switch (cmd) {
+       case SEP_IOCSENDSEPCOMMAND:
+               /* Send command to SEP */
+               error = sep_send_command_handler(sep);
+               break;
+       case SEP_IOCALLOCDATAPOLL:
+               /* Allocate data pool */
+               error = sep_allocate_data_pool_memory_handler(sep, arg);
+               break;
+       case SEP_IOCCREATESYMDMATABLE:
+               /* Create DMA table for synhronic operation */
+               error = sep_create_sync_dma_tables_handler(sep, arg);
+               break;
+       case SEP_IOCFREEDMATABLEDATA:
+               /* Free the pages */
+               error = sep_free_dma_table_data_handler(sep);
+               break;
+       case SEP_IOCSEPSTART:
+               /* Start command to SEP */
+               if (sep->pdev->revision == 0) /* Only for old chip */
+                       error = sep_start_handler(sep);
+               else
+                       error = -EPERM; /* Not permitted on new chip */
+               break;
+       case SEP_IOCSEPINIT:
+               /* Init command to SEP */
+               if (sep->pdev->revision == 0) /* Only for old chip */
+                       error = sep_init_handler(sep, arg);
+               else
+                       error = -EPERM; /* Not permitted on new chip */
+               break;
+       case SEP_IOCGETSTATICPOOLADDR:
+               /* Inform the SEP the bus address of the static pool */
+               error = sep_get_static_pool_addr_handler(sep);
+               break;
+       case SEP_IOCENDTRANSACTION:
+               error = sep_end_transaction_handler(sep);
+               break;
+       case SEP_IOCREALLOCEXTCACHE:
+               if (sep->pdev->revision == 0) /* Only for old chip */
+                       error = sep_realloc_ext_cache_handler(sep, arg);
+               else
+                       error = -EPERM; /* Not permitted on new chip */
+               break;
+       case SEP_IOCRARPREPAREMESSAGE:
+               error = sep_rar_prepare_output_msg_handler(sep, arg);
+               break;
+       case SEP_IOCPREPAREDCB:
+               error = sep_prepare_dcb_handler(sep, arg);
+               break;
+       case SEP_IOCFREEDCB:
+               error = sep_free_dcb_handler(sep);
+               break;
+       default:
+               dev_dbg(&sep->pdev->dev, "invalid ioctl %x\n", cmd);
+               error = -ENOTTY;
+               break;
+       }
+       mutex_unlock(&sep->ioctl_mutex);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "ioctl end\n");
+       return error;
+}
+
+/**
+ *     sep_singleton_ioctl - ioctl api for singleton interface
+ *     @filp: pointer to struct file
+ *     @cmd: command
+ *     @arg: pointer to argument structure
+ *
+ *     Implement the additional ioctls for the singleton device
+ */
+static long sep_singleton_ioctl(struct file  *filp, u32 cmd, unsigned long arg)
+{
+       long error = 0;
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "singleton_ioctl start\n");
+       dev_dbg(&sep->pdev->dev, "cmd is %x\n", cmd);
+
+       /* Check that the command is for the SEP device */
+       if (_IOC_TYPE(cmd) != SEP_IOC_MAGIC_NUMBER) {
+               error =  -ENOTTY;
+               goto end_function;
+       }
+
+       /* Make sure we own this device */
+       mutex_lock(&sep->sep_mutex);
+       if ((current->pid != sep->pid_doing_transaction) &&
+                               (sep->pid_doing_transaction != 0)) {
+               dev_dbg(&sep->pdev->dev, "singleton ioctl pid is not owner\n");
+               mutex_unlock(&sep->sep_mutex);
+               error = -EACCES;
+               goto end_function;
+       }
+
+       mutex_unlock(&sep->sep_mutex);
+
+       switch (cmd) {
+       case SEP_IOCTLSETCALLERID:
+               mutex_lock(&sep->ioctl_mutex);
+               error = sep_set_caller_id_handler(sep, arg);
+               mutex_unlock(&sep->ioctl_mutex);
+               break;
+       default:
+               error = sep_ioctl(filp, cmd, arg);
+               break;
+       }
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "singleton ioctl end\n");
+       return error;
+}
+
+/**
+ *     sep_request_daemon_ioctl - ioctl for daemon
+ *     @filp: pointer to struct file
+ *     @cmd: command
+ *     @arg: pointer to argument structure
+ *
+ *     Called by the request daemon to perform ioctls on the daemon device
+ */
+static long sep_request_daemon_ioctl(struct file *filp, u32 cmd,
+       unsigned long arg)
+{
+
+       long error;
+       struct sep_device *sep = filp->private_data;
+
+       dev_dbg(&sep->pdev->dev, "daemon ioctl: start\n");
+       dev_dbg(&sep->pdev->dev, "daemon ioctl: cmd is %x\n", cmd);
+
+       /* Check that the command is for SEP device */
+       if (_IOC_TYPE(cmd) != SEP_IOC_MAGIC_NUMBER) {
+               error = -ENOTTY;
+               goto end_function;
+       }
+
+       /* Only one process can access ioctl at any given time */
+       mutex_lock(&sep->ioctl_mutex);
+
+       switch (cmd) {
+       case SEP_IOCSENDSEPRPLYCOMMAND:
+               /* Send reply command to SEP */
+               error = sep_req_daemon_send_reply_command_handler(sep);
+               break;
+       case SEP_IOCENDTRANSACTION:
+               /*
+                * End req daemon transaction, do nothing
+                * will be removed upon update in middleware
+                * API library
+                */
+               error = 0;
+               break;
+       default:
+               dev_dbg(&sep->pdev->dev, "daemon ioctl: no such IOCTL\n");
+               error = -ENOTTY;
+       }
+       mutex_unlock(&sep->ioctl_mutex);
+
+end_function:
+       dev_dbg(&sep->pdev->dev, "daemon ioctl: end\n");
+       return error;
+
+}
+
+/**
+ *     sep_inthandler - interrupt handler
+ *     @irq: interrupt
+ *     @dev_id: device id
+ */
+static irqreturn_t sep_inthandler(int irq, void *dev_id)
+{
+       irqreturn_t int_error = IRQ_HANDLED;
+       unsigned long lck_flags;
+       u32 reg_val, reg_val2 = 0;
+       struct sep_device *sep = dev_id;
+
+       /* Read the IRR register to check if this is SEP interrupt */
+       reg_val = sep_read_reg(sep, HW_HOST_IRR_REG_ADDR);
+       dev_dbg(&sep->pdev->dev, "SEP Interrupt - reg is %08x\n", reg_val);
+
+       if (reg_val & (0x1 << 13)) {
+               /* Lock and update the counter of reply messages */
+               spin_lock_irqsave(&sep->snd_rply_lck, lck_flags);
+               sep->reply_ct++;
+               spin_unlock_irqrestore(&sep->snd_rply_lck, lck_flags);
+
+               dev_dbg(&sep->pdev->dev, "sep int: send_ct %lx reply_ct %lx\n",
+                                       sep->send_ct, sep->reply_ct);
+
+               /* Is this printf or daemon request? */
+               reg_val2 = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR2_REG_ADDR);
+               dev_dbg(&sep->pdev->dev,
+                       "SEP Interrupt - reg2 is %08x\n", reg_val2);
+
+               if ((reg_val2 >> 30) & 0x1) {
+                       dev_dbg(&sep->pdev->dev, "int: printf request\n");
+                       wake_up(&sep->event_request_daemon);
+               } else if (reg_val2 >> 31) {
+                       dev_dbg(&sep->pdev->dev, "int: daemon request\n");
+                       wake_up(&sep->event_request_daemon);
+               } else {
+                       dev_dbg(&sep->pdev->dev, "int: SEP reply\n");
+                       wake_up(&sep->event);
+               }
+       } else {
+               dev_dbg(&sep->pdev->dev, "int: not SEP interrupt\n");
+               int_error = IRQ_NONE;
+       }
+       if (int_error == IRQ_HANDLED)
+               sep_write_reg(sep, HW_HOST_ICR_REG_ADDR, reg_val);
+
+       return int_error;
+}
+
+/**
+ *     sep_reconfig_shared_area - reconfigure shared area
+ *     @sep: pointer to struct sep_device
+ *
+ *     Reconfig the shared area between HOST and SEP - needed in case
+ *     the DX_CC_Init function was called before OS loading.
+ */
+static int sep_reconfig_shared_area(struct sep_device *sep)
+{
+       int ret_val;
+
+       /* use to limit waiting for SEP */
+       unsigned long end_time;
+
+       dev_dbg(&sep->pdev->dev, "reconfig shared area start\n");
+
+       /* Send the new SHARED MESSAGE AREA to the SEP */
+       dev_dbg(&sep->pdev->dev, "sending %08llx to sep\n",
+                               (unsigned long long)sep->shared_bus);
+
+       sep_write_reg(sep, HW_HOST_HOST_SEP_GPR1_REG_ADDR, sep->shared_bus);
+
+       /* Poll for SEP response */
+       ret_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR1_REG_ADDR);
+
+       end_time = jiffies + (WAIT_TIME * HZ);
+
+       while ((time_before(jiffies, end_time)) && (ret_val != 0xffffffff) &&
+               (ret_val != sep->shared_bus))
+               ret_val = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR1_REG_ADDR);
+
+       /* Check the return value (register) */
+       if (ret_val != sep->shared_bus) {
+               dev_warn(&sep->pdev->dev, "could not reconfig shared area\n");
+               dev_warn(&sep->pdev->dev, "result was %x\n", ret_val);
+               ret_val = -ENOMEM;
+       } else
+               ret_val = 0;
+
+       dev_dbg(&sep->pdev->dev, "reconfig shared area end\n");
+       return ret_val;
+}
+
+/* File operation for singleton SEP operations */
+static const struct file_operations singleton_file_operations = {
+       .owner = THIS_MODULE,
+       .unlocked_ioctl = sep_singleton_ioctl,
+       .poll = sep_poll,
+       .open = sep_singleton_open,
+       .release = sep_singleton_release,
+       .mmap = sep_mmap,
+};
+
+/* File operation for daemon operations */
+static const struct file_operations daemon_file_operations = {
+       .owner = THIS_MODULE,
+       .unlocked_ioctl = sep_request_daemon_ioctl,
+       .poll = sep_request_daemon_poll,
+       .open = sep_request_daemon_open,
+       .release = sep_request_daemon_release,
+       .mmap = sep_request_daemon_mmap,
+};
+
+/* The files operations structure of the driver */
+static const struct file_operations sep_file_operations = {
+       .owner = THIS_MODULE,
+       .unlocked_ioctl = sep_ioctl,
+       .poll = sep_poll,
+       .open = sep_open,
+       .release = sep_release,
+       .mmap = sep_mmap,
+};
+
+/**
+ *     sep_register_driver_with_fs - register misc devices
+ *     @sep: pointer to struct sep_device
+ *
+ *     This function registers the driver with the file system
+ */
+static int sep_register_driver_with_fs(struct sep_device *sep)
+{
+       int ret_val;
+
+       sep->miscdev_sep.minor = MISC_DYNAMIC_MINOR;
+       sep->miscdev_sep.name = SEP_DEV_NAME;
+       sep->miscdev_sep.fops = &sep_file_operations;
+
+       sep->miscdev_singleton.minor = MISC_DYNAMIC_MINOR;
+       sep->miscdev_singleton.name = SEP_DEV_SINGLETON;
+       sep->miscdev_singleton.fops = &singleton_file_operations;
+
+       sep->miscdev_daemon.minor = MISC_DYNAMIC_MINOR;
+       sep->miscdev_daemon.name = SEP_DEV_DAEMON;
+       sep->miscdev_daemon.fops = &daemon_file_operations;
+
+       ret_val = misc_register(&sep->miscdev_sep);
+       if (ret_val) {
+               dev_warn(&sep->pdev->dev, "misc reg fails for SEP %x\n",
+                       ret_val);
+               return ret_val;
+       }
+
+       ret_val = misc_register(&sep->miscdev_singleton);
+       if (ret_val) {
+               dev_warn(&sep->pdev->dev, "misc reg fails for sing %x\n",
+                       ret_val);
+               misc_deregister(&sep->miscdev_sep);
+               return ret_val;
+       }
+
+       ret_val = misc_register(&sep->miscdev_daemon);
+       if (ret_val) {
+               dev_warn(&sep->pdev->dev, "misc reg fails for dmn %x\n",
+                       ret_val);
+               misc_deregister(&sep->miscdev_sep);
+               misc_deregister(&sep->miscdev_singleton);
+
+               return ret_val;
+       }
+       return ret_val;
+}
+
+
+/**
+ *     sep_probe - probe a matching PCI device
+ *     @pdev: pci_device
+ *     @end: pci_device_id
+ *
+ *     Attempt to set up and configure a SEP device that has been
+ *     discovered by the PCI layer.
+ */
+static int __devinit sep_probe(struct pci_dev *pdev,
+       const struct pci_device_id *ent)
+{
+       int error = 0;
+       struct sep_device *sep;
+
+       pr_debug("SEP pci probe starting\n");
+       if (sep_dev != NULL) {
+               dev_warn(&pdev->dev, "only one SEP supported.\n");
+               return -EBUSY;
+       }
+
+       /* Enable the device */
+       error = pci_enable_device(pdev);
+       if (error) {
+               dev_warn(&pdev->dev, "error enabling pci device\n");
+               goto end_function;
+       }
+
+       /* Allocate the sep_device structure for this device */
+       sep_dev = kzalloc(sizeof(struct sep_device), GFP_ATOMIC);
+       if (sep_dev == NULL) {
+               dev_warn(&pdev->dev,
+                       "can't kmalloc the sep_device structure\n");
+               error = -ENOMEM;
+               goto end_function_disable_device;
+       }
+
+       /*
+        * We're going to use another variable for actually
+        * working with the device; this way, if we have
+        * multiple devices in the future, it would be easier
+        * to make appropriate changes
+        */
+       sep = sep_dev;
+
+       sep->pdev = pci_dev_get(pdev);
+
+       init_waitqueue_head(&sep->event);
+       init_waitqueue_head(&sep->event_request_daemon);
+       spin_lock_init(&sep->snd_rply_lck);
+       mutex_init(&sep->sep_mutex);
+       mutex_init(&sep->ioctl_mutex);
+
+       dev_dbg(&sep->pdev->dev, "PCI obtained, device being prepared\n");
+       dev_dbg(&sep->pdev->dev, "revision is %d\n", sep->pdev->revision);
+
+       /* Set up our register area */
+       sep->reg_physical_addr = pci_resource_start(sep->pdev, 0);
+       if (!sep->reg_physical_addr) {
+               dev_warn(&sep->pdev->dev, "Error getting register start\n");
+               error = -ENODEV;
+               goto end_function_free_sep_dev;
+       }
+
+       sep->reg_physical_end = pci_resource_end(sep->pdev, 0);
+       if (!sep->reg_physical_end) {
+               dev_warn(&sep->pdev->dev, "Error getting register end\n");
+               error = -ENODEV;
+               goto end_function_free_sep_dev;
+       }
+
+       sep->reg_addr = ioremap_nocache(sep->reg_physical_addr,
+               (size_t)(sep->reg_physical_end - sep->reg_physical_addr + 1));
+       if (!sep->reg_addr) {
+               dev_warn(&sep->pdev->dev, "Error getting register virtual\n");
+               error = -ENODEV;
+               goto end_function_free_sep_dev;
+       }
+
+       dev_dbg(&sep->pdev->dev,
+               "Register area start %llx end %llx virtual %p\n",
+               (unsigned long long)sep->reg_physical_addr,
+               (unsigned long long)sep->reg_physical_end,
+               sep->reg_addr);
+
+       /* Allocate the shared area */
+       sep->shared_size = SEP_DRIVER_MESSAGE_SHARED_AREA_SIZE_IN_BYTES +
+               SYNCHRONIC_DMA_TABLES_AREA_SIZE_BYTES +
+               SEP_DRIVER_DATA_POOL_SHARED_AREA_SIZE_IN_BYTES +
+               SEP_DRIVER_STATIC_AREA_SIZE_IN_BYTES +
+               SEP_DRIVER_SYSTEM_DATA_MEMORY_SIZE_IN_BYTES;
+
+       if (sep_map_and_alloc_shared_area(sep)) {
+               error = -ENOMEM;
+               /* Allocation failed */
+               goto end_function_error;
+       }
+
+       sep->rar_size = FAKE_RAR_SIZE;
+       sep->rar_addr = dma_alloc_coherent(&sep->pdev->dev,
+               sep->rar_size, &sep->rar_bus, GFP_KERNEL);
+       if (sep->rar_addr == NULL) {
+               dev_warn(&sep->pdev->dev, "can't allocate mfld rar\n");
+               error = -ENOMEM;
+               goto end_function_deallocate_sep_shared_area;
+       }
+
+       dev_dbg(&sep->pdev->dev, "rar start is %p, phy is %llx,"
+               " size is %zx\n", sep->rar_addr,
+               (unsigned long long)sep->rar_bus,
+               sep->rar_size);
+
+       dev_dbg(&sep->pdev->dev, "about to write IMR and ICR REG_ADDR\n");
+
+       /* Clear ICR register */
+       sep_write_reg(sep, HW_HOST_ICR_REG_ADDR, 0xFFFFFFFF);
+
+       /* Set the IMR register - open only GPR 2 */
+       sep_write_reg(sep, HW_HOST_IMR_REG_ADDR, (~(0x1 << 13)));
+
+       /* Read send/receive counters from SEP */
+       sep->reply_ct = sep_read_reg(sep, HW_HOST_SEP_HOST_GPR2_REG_ADDR);
+       sep->reply_ct &= 0x3FFFFFFF;
+       sep->send_ct = sep->reply_ct;
+
+       dev_dbg(&sep->pdev->dev, "about to call request_irq\n");
+       /* Get the interrupt line */
+       error = request_irq(pdev->irq, sep_inthandler, IRQF_SHARED,
+               "sep_driver", sep);
+
+       if (error)
+               goto end_function_dealloc_rar;
+
+       /* The new chip requires ashared area reconfigure */
+       if (sep->pdev->revision == 4) { /* Only for new chip */
+               error = sep_reconfig_shared_area(sep);
+               if (error)
+                       goto end_function_free_irq;
+       }
+       /* Finally magic up the device nodes */
+       /* Register driver with the fs */
+       error = sep_register_driver_with_fs(sep);
+       if (error == 0)
+               /* Success */
+               return 0;
+
+end_function_free_irq:
+       free_irq(pdev->irq, sep);
+
+end_function_dealloc_rar:
+       if (sep->rar_addr)
+               dma_free_coherent(&sep->pdev->dev, sep->rar_size,
+                       sep->rar_addr, sep->rar_bus);
+       goto end_function;
+
+end_function_deallocate_sep_shared_area:
+       /* De-allocate shared area */
+       sep_unmap_and_free_shared_area(sep);
+
+end_function_error:
+       iounmap(sep->reg_addr);
+
+end_function_free_sep_dev:
+       pci_dev_put(sep_dev->pdev);
+       kfree(sep_dev);
+       sep_dev = NULL;
+
+end_function_disable_device:
+       pci_disable_device(pdev);
+
+end_function:
+       return error;
+}
+
+static void sep_remove(struct pci_dev *pdev)
+{
+       struct sep_device *sep = sep_dev;
+
+       /* Unregister from fs */
+       misc_deregister(&sep->miscdev_sep);
+       misc_deregister(&sep->miscdev_singleton);
+       misc_deregister(&sep->miscdev_daemon);
+
+       /* Free the irq */
+       free_irq(sep->pdev->irq, sep);
+
+       /* Free the shared area  */
+       sep_unmap_and_free_shared_area(sep_dev);
+       iounmap((void *) sep_dev->reg_addr);
+}
+
+static DEFINE_PCI_DEVICE_TABLE(sep_pci_id_tbl) = {
+       {PCI_DEVICE(PCI_VENDOR_ID_INTEL, MFLD_PCI_DEVICE_ID)},
+       {0}
+};
+
+MODULE_DEVICE_TABLE(pci, sep_pci_id_tbl);
+
+/* Field for registering driver to PCI device */
+static struct pci_driver sep_pci_driver = {
+       .name = "sep_sec_driver",
+       .id_table = sep_pci_id_tbl,
+       .probe = sep_probe,
+       .remove = sep_remove
+};
+
+
+/**
+ *     sep_init - init function
+ *
+ *     Module load time. Register the PCI device driver.
+ */
+static int __init sep_init(void)
+{
+       return pci_register_driver(&sep_pci_driver);
+}
+
+
+/**
+ *     sep_exit - called to unload driver
+ *
+ *     Drop the misc devices then remove and unmap the various resources
+ *     that are not released by the driver remove method.
+ */
+static void __exit sep_exit(void)
+{
+       pci_unregister_driver(&sep_pci_driver);
+}
+
+
+module_init(sep_init);
+module_exit(sep_exit);
+
+MODULE_LICENSE("GPL");