[lantiq] cleanup patches
[openwrt.git] / target / linux / lantiq / files / net / ipv4 / svip_nat.c
diff --git a/target/linux/lantiq/files/net/ipv4/svip_nat.c b/target/linux/lantiq/files/net/ipv4/svip_nat.c
new file mode 100644 (file)
index 0000000..04a0d22
--- /dev/null
@@ -0,0 +1,1569 @@
+/******************************************************************************
+
+                               Copyright (c) 2009
+                            Lantiq Deutschland GmbH
+                     Am Campeon 3; 81726 Munich, Germany
+
+  THE DELIVERY OF THIS SOFTWARE AS WELL AS THE HEREBY GRANTED NON-EXCLUSIVE,
+  WORLDWIDE LICENSE TO USE, COPY, MODIFY, DISTRIBUTE AND SUBLICENSE THIS
+  SOFTWARE IS FREE OF CHARGE.
+
+  THE LICENSED SOFTWARE IS PROVIDED "AS IS" AND INFINEON EXPRESSLY DISCLAIMS
+  ALL REPRESENTATIONS AND WARRANTIES, WHETHER EXPRESS OR IMPLIED, INCLUDING
+  WITHOUT LIMITATION, WARRANTIES OR REPRESENTATIONS OF WORKMANSHIP,
+  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, DURABILITY, THAT THE
+  OPERATING OF THE LICENSED SOFTWARE WILL BE ERROR FREE OR FREE OF ANY THIRD
+  PARTY CLAIMS, INCLUDING WITHOUT LIMITATION CLAIMS OF THIRD PARTY INTELLECTUAL
+  PROPERTY INFRINGEMENT.
+
+  EXCEPT FOR ANY LIABILITY DUE TO WILFUL ACTS OR GROSS NEGLIGENCE AND EXCEPT
+  FOR ANY PERSONAL INJURY INFINEON SHALL IN NO EVENT BE LIABLE FOR ANY CLAIM
+  OR DAMAGES OF ANY KIND, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+  ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+  DEALINGS IN THE SOFTWARE.
+
+ ****************************************************************************
+
+Description : This file contains implementation of Custom NAT function
+for Infineon's VINETIC-SVIP16
+ *******************************************************************************/
+
+#include <linux/module.h>
+#include <linux/netfilter_ipv4.h>
+#include <linux/if_ether.h>
+#include <linux/netdevice.h>
+#include <linux/inetdevice.h>
+#include <linux/in.h>
+#include <linux/ip.h>
+#include <linux/if_vlan.h>
+#include <linux/udp.h>
+#include <linux/kernel.h>
+#include <linux/version.h>
+#include <linux/proc_fs.h>
+#include <linux/in6.h> /* just to shut up a warning */
+#include <linux/miscdevice.h>
+#include <asm/checksum.h>
+
+#include <linux/svip_nat.h>
+
+MODULE_AUTHOR("Lantiq Deutschland GmbH");
+MODULE_DESCRIPTION("SVIP Network Address Translation module");
+MODULE_LICENSE("GPL");
+
+#define SVIP_NAT_INFO_STR "@(#)SVIP NAT, version "SVIP_NAT_VERSION
+
+/** maximum voice packet channels possible on the SVIP LC system
+  (equals maximum number of Codec channels possible) */
+#define SVIP_SYS_CODEC_NUM    ((SVIP_SYS_NUM) * (SVIP_CODEC_NUM))
+
+/** end UDP port number of the SVIP Linecard System */
+#define SVIP_UDP_TO           ((SVIP_UDP_FROM) + (SVIP_SYS_CODEC_NUM) - 1)
+
+/** end UDP port number of the Master SVIP in SVIP Linecard System */
+#define SVIP_UDP_TO_VOFW0     ((SVIP_UDP_FROM) + (SVIP_CODEC_NUM) - 1)
+
+#define SVIP_PORT_INRANGE(nPort) \
+       ((nPort) >= (SVIP_UDP_FROM) && (nPort) <= (SVIP_UDP_TO))
+
+#define SVIP_PORT_INDEX(nPort)   (nPort - SVIP_UDP_FROM)
+
+#define SVIP_NET_DEV_ETH0_IDX       0
+#define SVIP_NET_DEV_VETH0_IDX      1
+#define SVIP_NET_DEV_LO_IDX         2
+
+#define SVIP_NET_DEV_ETH0_NAME      "eth0"
+#define SVIP_NET_DEV_ETH1_NAME      "eth1"
+#define SVIP_NET_DEV_VETH1_NAME     "veth0"
+#define SVIP_NET_DEV_LO_NAME        "lo"
+
+#define SVIP_NAT_STATS_LOC2REM   0
+#define SVIP_NAT_STATS_REM2LOC   1
+#define SVIP_NAT_STATS_TYPES     2
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24)
+#define SVIP_NAT_FOR_EACH_NETDEV(d) for_each_netdev(&init_net, dev)
+#define SVIP_NAT_IP_HDR(ethhdr) ip_hdr(ethhdr)
+#else
+#define SVIP_NAT_FOR_EACH_NETDEV(d) for(d=dev_base; dev; dev = dev->next)
+#define SVIP_NAT_IP_HDR(ethhdr) (ethhdr)->nh.iph
+#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24) */
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+#define SVIP_NAT_SKB_MAC_HEADER(ethhdr) (ethhdr)->mac.ethernet
+#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
+#define SVIP_NAT_SKB_MAC_HEADER(ethhdr) (ethhdr)->mac.raw
+#else
+#define SVIP_NAT_SKB_MAC_HEADER(ethhdr) skb_mac_header(ethhdr)
+#endif
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24)
+#define VLAN_DEV_REAL_DEV(dev)      vlan_dev_real_dev(dev)
+#define VLAN_DEV_VLAN_ID(dev)       vlan_dev_vlan_id(dev)
+#else
+#define VLAN_DEV_REAL_DEV(dev)      (VLAN_DEV_INFO(dev)->real_dev)
+#define VLAN_DEV_VLAN_ID(dev)       (VLAN_DEV_INFO(dev)->vlan_id)
+#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24) */
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0))
+#define MOD_INC_USE_COUNT
+#define MOD_DEC_USE_COUNT
+#endif
+
+#if ! ((LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && \
+       (defined(CONFIG_VLAN_8021Q) || defined(CONFIG_VLAN_8021Q_MODULE)))
+#define VLAN_8021Q_UNUSED
+#endif
+
+
+extern spinlock_t vlan_group_lock;
+extern struct net_device *__vlan_find_dev_deep(struct net_device *real_dev, unsigned short VID);
+
+typedef struct SVIP_NAT_stats
+{
+       unsigned long        inPackets;
+       unsigned long        outPackets;
+       unsigned long        outErrors;
+} SVIP_NAT_stats_t;
+
+typedef struct SVIP_NAT_table_entry
+{
+       SVIP_NAT_IO_Rule_t   natRule;
+       SVIP_NAT_stats_t     natStats[SVIP_NAT_STATS_TYPES];
+} SVIP_NAT_table_entry_t;
+
+/* pointer to the SVIP NAT table */
+static SVIP_NAT_table_entry_t *pNatTable = NULL;
+
+struct net_device *net_devs[3];
+static u32 *paddr_eth0;
+static u32 *paddr_eth0_0;
+static u32 *paddr_veth0;
+static u32 *pmask_veth0;
+
+static struct semaphore *sem_nat_tbl_access;
+static int proc_read_in_progress = 0;
+
+static int nDeviceOpen = 0;
+
+/* saves the NAT table index between subsequent invocation */
+static int nProcReadIdx = 0;
+
+static long SVIP_NAT_device_ioctl(struct file *,unsigned int ,unsigned long);
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,2,0)
+static int  SVIP_NAT_device_release (struct inode *,struct file *);
+#else
+static void SVIP_NAT_device_release (struct inode *,struct file *);
+#endif
+static int  SVIP_NAT_device_open    (struct inode *,struct file *);
+
+/* This structure holds the interface functions supported by
+   the SVIP NAT configuration device. */
+struct file_operations SVIP_NAT_Fops = {
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,0)
+owner:      THIS_MODULE,
+#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,0) */
+           llseek:  NULL,                      /* seek */
+           read:    NULL,
+           write:   NULL,
+           readdir: NULL,                      /* readdir */
+           poll:    NULL,                      /* select */
+           unlocked_ioctl:   SVIP_NAT_device_ioctl,     /* ioctl */
+           mmap:    NULL,                      /* mmap */
+           open:    SVIP_NAT_device_open,      /* open, */
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,2,0)
+           flush:   NULL,                      /* flush */
+#endif
+           release: SVIP_NAT_device_release    /* close */
+};
+
+/** Structure holding MISC module operations */
+static struct miscdevice SVIP_NAT_miscdev =
+{
+minor:   MINOR_NUM_SVIP_NAT,
+        name:    SVIP_NAT_DEVICE_NAME,
+        fops:    &SVIP_NAT_Fops
+};
+
+#ifdef CONFIG_SVIP_FW_PKT_SNIFFER
+int nSVIP_NAT_Sniffer;
+unsigned char pSVIP_NAT_SnifferMAC[ETH_ALEN];
+int nSVIP_NAT_SnifferMacSet;
+#endif
+
+/******************************************************************************/
+/**
+  Function to read /proc/net/svip_nat/nat proc entry
+
+  \arguments
+  page     - pointer to page buffer
+  start    - pointer to start address pointer
+  off      - offset
+  count    - maximum data length to read
+  eof      - end of file flag
+  data     - proc read data (provided by the function
+  pointed to by data)
+
+  \return
+  length of read data
+
+  \remarks:
+  Each call of this routine forces a copy_to_user of the data returned by
+  'fn'. This routine will be called by the user until 'len = 0'.
+ ****************************************************************************/
+static int SVIP_NAT_ProcRead (char *page, char **start, off_t off,
+                             int count, int *eof, void *data)
+{
+       unsigned long flags;
+       int (*fn)(char *buf, int size);
+       int len;
+
+       /* If the NAT table index is negative, the reading has completed */
+       if (nProcReadIdx < 0)
+       {
+               nProcReadIdx = 0;
+               *eof = 1;
+               proc_read_in_progress = 0;
+               up(sem_nat_tbl_access);
+               return 0;
+       }
+
+       local_irq_save(flags);
+       if (!proc_read_in_progress)
+       {
+               proc_read_in_progress = 1;
+               local_irq_restore(flags);
+               /* we use this semaphore in order to ensure no other party(could be ioctl
+                  FIO_SVIP_NAT_RULE_LIST), uses function SVIP_NAT_ProcReadNAT(), during
+                  the time read of the proc file takes place */
+               down(sem_nat_tbl_access);
+       }
+       else
+       {
+               local_irq_restore(flags);
+       }
+
+       if (data != NULL)
+       {
+               fn = data;
+               len = fn (page, count);
+               /* In this setup each read of the proc entries returns the read data by
+                  'fn' to the user. The user keeps issuing read requests as long as the
+                  returned value of 'len' is greater than zero. */
+               *eof = 1;
+               *start = page;
+       }
+       else
+       {
+               len = 0;
+       }
+
+       return len;
+}
+
+#ifdef CONFIG_SVIP_FW_PKT_SNIFFER
+/**
+  Function to read remaining proc entries
+  */
+static int SVIP_NAT_ProcReadGen (char *page, char **start, off_t off,
+                                int count, int *eof, void *data)
+{
+       int (*fn)(char *buf, int size);
+       int len = 0;
+
+       MOD_INC_USE_COUNT;
+
+       if (data == NULL)
+       {
+               MOD_DEC_USE_COUNT;
+               return 0;
+       }
+
+       fn = data;
+       len = fn (page, count);
+
+       if (len <= off + count)
+       {
+               *eof = 1;
+       }
+       *start = page + off;
+       len -= off;
+       if (len > count)
+       {
+               len = count;
+       }
+       if (len < 0)
+       {
+               len = 0;
+       }
+
+       MOD_DEC_USE_COUNT;
+
+       return len;
+}
+#endif
+
+/******************************************************************************/
+/**
+  Function for setting up /proc/net/svip_nat read data
+
+  \arguments
+  buf      - pointer to read buffer
+  count    - size of read buffer
+
+  \return
+  length of read data into buffer
+
+  \remarks:
+  The global variable 'nProcReadIdx' is used to save the table index where
+  the reading of the NAT table stopped. Reading is stopped when the end of
+  the read buffer is approached. On the next itteration the reading continues
+  from the saved index.
+ *******************************************************************************/
+static int SVIP_NAT_ProcReadNAT(char *buf, int count)
+{
+       int i, j;
+       int len = 0;
+       SVIP_NAT_IO_Rule_t *pNatRule;
+
+       if (nProcReadIdx == -1)
+       {
+               nProcReadIdx = 0;
+               return 0;
+       }
+
+       if (nProcReadIdx == 0)
+       {
+               len = sprintf(buf+len,
+                             "Remote host IP  "         /* 16 char */
+                             "Remote host MAC    "      /* 19 char */
+                             "Local host IP  "          /* 15 char */
+                             "Local host MAC     "      /* 19 char */
+                             "Local host UDP  "         /* 16 char */
+                             "Loc->Rem(in/out/err)  "   /* 22 char */
+                             "Rem->Loc(in/out/err)\n\r");
+       }
+
+       for (i = nProcReadIdx; i < SVIP_SYS_CODEC_NUM; i++)
+       {
+               int slen;
+
+               pNatRule = &pNatTable[i].natRule;
+
+               if (pNatRule->remIP != 0)
+               {
+                       /* make sure not to overwrite the buffer */
+                       if (count < len+120)
+                               break;
+
+                       /* remIP */
+                       slen = sprintf(buf+len, "%d.%d.%d.%d",
+                                      (int)((pNatRule->remIP >> 24) & 0xff),
+                                      (int)((pNatRule->remIP >> 16) & 0xff),
+                                      (int)((pNatRule->remIP >> 8) & 0xff),
+                                      (int)((pNatRule->remIP >> 0) & 0xff));
+                       len += slen;
+                       for (j = 0; j < (16-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* remMAC */
+                       slen = 0;
+                       for (j = 0; j < ETH_ALEN; j++)
+                       {
+                               slen += sprintf(buf+len+slen, "%02x%s",
+                                               pNatRule->remMAC[j], j < ETH_ALEN-1 ? ":" : " ");
+                       }
+                       len += slen;
+                       for (j = 0; j < (19-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* locIP */
+                       slen = sprintf(buf+len, "%d.%d.%d.%d",
+                                      (int)((pNatRule->locIP >> 24) & 0xff),
+                                      (int)((pNatRule->locIP >> 16) & 0xff),
+                                      (int)((pNatRule->locIP >> 8) & 0xff),
+                                      (int)((pNatRule->locIP >> 0) & 0xff));
+                       len += slen;
+                       for (j = 0; j < (15-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* locMAC */
+                       slen = 0;
+                       for (j = 0; j < ETH_ALEN; j++)
+                       {
+                               slen += sprintf(buf+len+slen, "%02x%s",
+                                               pNatRule->locMAC[j], j < ETH_ALEN-1 ? ":" : " ");
+                       }
+                       len += slen;
+                       for (j = 0; j < (19-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* locUDP */
+                       slen = sprintf(buf+len, "%d", pNatRule->locUDP);
+                       len += slen;
+                       for (j = 0; j < (16-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* NAT statistics, Local to Remote translation */
+                       slen = sprintf(buf+len, "(%ld/%ld/%ld)",
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_LOC2REM].inPackets,
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_LOC2REM].outPackets,
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_LOC2REM].outErrors);
+                       len += slen;
+                       for (j = 0; j < (22-slen); j++)
+                               len += sprintf(buf+len, " ");
+
+                       /* NAT statistics, Remote to Local translation */
+                       len += sprintf(buf+len, "(%ld/%ld/%ld)\n\r",
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_REM2LOC].inPackets,
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_REM2LOC].outPackets,
+                                      pNatTable[i].natStats[SVIP_NAT_STATS_REM2LOC].outErrors);
+               }
+       }
+       if (i == SVIP_SYS_CODEC_NUM)
+               nProcReadIdx = -1;   /* reading completed */
+       else
+               nProcReadIdx = i;    /* reading still in process, buffer was full */
+
+       return len;
+}
+
+#ifdef CONFIG_SVIP_FW_PKT_SNIFFER
+/**
+  Converts MAC address from ascii to hex respesentaion
+  */
+static int SVIP_NAT_MacAsciiToHex(const char *pMacStr, unsigned char *pMacHex)
+{
+       int i=0, c=0, b=0, n=0;
+
+       memset(pMacHex, 0, ETH_ALEN);
+       while (pMacStr[i] != '\0')
+       {
+               if (n >= 0)
+               {
+                       unsigned char nToHex = 0;
+
+                       /* check for hex digit */
+                       if (pMacStr[i] >= '0' && pMacStr[i] <= '9')
+                               nToHex = 0x30;
+                       else if (pMacStr[i] >= 'a' && pMacStr[i] <= 'f')
+                               nToHex = 0x57;
+                       else if (pMacStr[i] >= 'A' && pMacStr[i] <= 'F')
+                               nToHex = 0x37;
+                       else
+                       {
+                               if (n != 0)
+                               {
+                                       printk(KERN_ERR "SVIP NAT: invalid MAC address format[%s]\n", pMacStr);
+                                       return -1;
+                               }
+                               i++;
+                               continue;
+                       }
+                       n^=1;
+                       pMacHex[b] |= ((pMacStr[i] - nToHex)&0xf) << (4*n);
+                       if (n == 0)
+                       {
+                               /* advance to next byte, check if complete */
+                               if (++b >= ETH_ALEN)
+                                       return 0;
+                               /* byte completed, next we expect a colon... */
+                               c = 1;
+                               /* and, do not check for hex digit */
+                               n = -1;
+                       }
+                       i++;
+                       continue;
+               }
+               if (c == 1)
+               {
+                       if (pMacStr[i] == ':')
+                       {
+                               /* next we expect hex digit, again */
+                               n = 0;
+                       }
+                       else
+                       {
+                               printk(KERN_ERR "SVIP NAT: invalid MAC address format[%s]\n", pMacStr);
+                               return -1;
+                       }
+               }
+               i++;
+       }
+       return 0;
+}
+
+/**
+  Used to set the destination MAC address of a host where incoming
+  SVIP VoFW packets are to be addressed. In case the address is set
+  to 00:00:00:00:00:00 (the default case), the packets will written
+  out to eth0 with its original MAC addess.
+
+  \remark
+usage: 'echo "00:03:19:00:15:D1" > cat /proc/net/svip_nat/snifferMAC'
+*/
+int SVIP_NAT_ProcWriteSnifferMAC (struct file *file, const char *buffer,
+                                 unsigned long count, void *data)
+{
+       /* at least strlen("xx:xx:xx:xx:xx:xx") characters, followed by '\0' */
+       if (count >= 18)
+       {
+               int ret;
+
+               ret = SVIP_NAT_MacAsciiToHex(buffer, pSVIP_NAT_SnifferMAC);
+
+               if (ret != 0)
+                       return 0;
+
+               if (!(pSVIP_NAT_SnifferMAC[0]==0 && pSVIP_NAT_SnifferMAC[1]==0 &&
+                     pSVIP_NAT_SnifferMAC[2]==0 && pSVIP_NAT_SnifferMAC[3]==0 &&
+                     pSVIP_NAT_SnifferMAC[4]==0 && pSVIP_NAT_SnifferMAC[5]==0))
+               {
+                       nSVIP_NAT_SnifferMacSet = 1;
+               }
+       }
+       return count;
+}
+
+/**
+  Used to read the destination MAC address of a sniffer host
+  */
+int SVIP_NAT_ProcReadSnifferMAC (char *buf, int count)
+{
+       int len = 0;
+
+       len = snprintf(buf, count, "%02x:%02x:%02x:%02x:%02x:%02x\n",
+                      pSVIP_NAT_SnifferMAC[0], pSVIP_NAT_SnifferMAC[1],
+                      pSVIP_NAT_SnifferMAC[2], pSVIP_NAT_SnifferMAC[3],
+                      pSVIP_NAT_SnifferMAC[4], pSVIP_NAT_SnifferMAC[5]);
+
+       if (len > count)
+       {
+               printk(KERN_ERR "SVIP NAT: Only part of the text could be put into the buffer\n");
+               return count;
+       }
+
+       return len;
+}
+
+/**
+  Used to switch VoFW message sniffer on/off
+
+  \remark
+usage: 'echo "1" > cat /proc/net/svip_nat/snifferOnOff'
+*/
+int SVIP_NAT_ProcWriteSnifferOnOff (struct file *file, const char *buffer,
+                                   unsigned long count, void *data)
+{
+       /* at least one digit expected, followed by '\0' */
+       if (count >= 2)
+       {
+               int ret, nSnifferOnOff;
+
+               ret = sscanf(buffer, "%d", &nSnifferOnOff);
+
+               if (ret != 1)
+                       return count;
+
+               if (nSnifferOnOff > 0)
+                       nSnifferOnOff = 1;
+
+               nSVIP_NAT_Sniffer = nSnifferOnOff;
+       }
+       return count;
+}
+
+/**
+  Used to read the VoFW message sniffer configuration (on/off)
+  */
+int SVIP_NAT_ProcReadSnifferOnOff (char *buf, int count)
+{
+       int len = 0;
+
+       len = snprintf(buf, count, "%d\n", nSVIP_NAT_Sniffer);
+
+       if (len > count)
+       {
+               printk(KERN_ERR "SVIP NAT: Only part of the text could be put into the buffer\n");
+               return count;
+       }
+
+       return len;
+}
+#endif
+
+/******************************************************************************/
+/**
+  Creates proc read/write entries
+
+  \return
+  0 on success, -1 on error
+  */
+/******************************************************************************/
+static int SVIP_NAT_ProcInstall(void)
+{
+       struct proc_dir_entry *pProcParentDir, *pProcDir;
+       struct proc_dir_entry *pProcNode;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
+       pProcParentDir = proc_net;
+#else
+       pProcParentDir = init_net.proc_net;
+#endif
+       pProcDir = proc_mkdir(SVIP_NAT_DEVICE_NAME, pProcParentDir);
+       if (pProcDir == NULL)
+       {
+               printk(KERN_ERR "SVIP NAT: cannot create proc dir %s/%s\n\r",
+                      pProcParentDir->name, SVIP_NAT_DEVICE_NAME);
+               return -1;
+       }
+
+       pProcNode = create_proc_read_entry("nat", S_IFREG|S_IRUGO, pProcDir,
+                                          SVIP_NAT_ProcRead, (void *)SVIP_NAT_ProcReadNAT);
+       if (pProcNode == NULL)
+       {
+               printk(KERN_ERR "SVIP NAT: cannot create proc entry %s/%s",
+                      pProcDir->name, "nat");
+               return -1;
+       }
+
+#ifdef CONFIG_SVIP_FW_PKT_SNIFFER
+       nSVIP_NAT_Sniffer = 0;
+       /* creates proc entry for switching on/off sniffer to VoFW messages */
+       pProcNode = create_proc_read_entry("snifferOnOff", S_IFREG|S_IRUGO|S_IWUGO,
+                                          pProcDir, SVIP_NAT_ProcReadGen, (void *)SVIP_NAT_ProcReadSnifferOnOff);
+       if (pProcNode == NULL)
+       {
+               printk(KERN_ERR "SVIP NAT: cannot create proc entry %s/%s\n\r",
+                      pProcDir->name, "snifferOnOff");
+               return -1;
+       }
+       pProcNode->write_proc = SVIP_NAT_ProcWriteSnifferOnOff;
+
+       memset (pSVIP_NAT_SnifferMAC, 0, ETH_ALEN);
+       nSVIP_NAT_SnifferMacSet = 0;
+       /* creates proc entry for setting MAC address of sniffer host to VoFW messages */
+       pProcNode = create_proc_read_entry("snifferMAC", S_IFREG|S_IRUGO|S_IWUGO,
+                                          pProcDir, SVIP_NAT_ProcReadGen, (void *)SVIP_NAT_ProcReadSnifferMAC);
+       if (pProcNode == NULL)
+       {
+               printk(KERN_ERR "SVIP NAT: cannot create proc entry %s/%s\n\r",
+                      pProcDir->name, "snifferMAC");
+               return -1;
+       }
+       pProcNode->write_proc = SVIP_NAT_ProcWriteSnifferMAC;
+#endif
+
+       return 0;
+}
+
+/******************************************************************************/
+/**
+  No actions done here, simply a check is performed if an open has already
+  been performed. Currently only a single open is allowed as it is a sufficient
+  to have hat a single process configuring the SVIP NAT at one time.
+
+  \arguments
+  inode       - pointer to disk file data
+  file        - pointer to device file data
+
+  \return
+  0 on success, else -1
+  */
+/******************************************************************************/
+static int SVIP_NAT_device_open(struct inode *inode, struct file *file)
+{
+       unsigned long flags;
+       struct in_device *in_dev;
+       struct in_ifaddr *ifa;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+       local_irq_save(flags);
+#else
+       local_save_flags(flags);
+#endif
+
+       if (nDeviceOpen)
+       {
+               MOD_INC_USE_COUNT;
+               local_irq_restore(flags);
+               nDeviceOpen++;
+               return 0;
+       }
+
+       /* find pointer to IP address of eth0 */
+       if ((in_dev=in_dev_get(net_devs[SVIP_NET_DEV_ETH0_IDX])) != NULL)
+       {
+               for (ifa = in_dev->ifa_list; ifa != NULL; ifa = ifa->ifa_next)
+               {
+                       if (!paddr_eth0 && ifa->ifa_address != 0)
+                       {
+                               paddr_eth0 = &ifa->ifa_address;
+                               continue;
+                       }
+                       if (paddr_eth0 && ifa->ifa_address != 0)
+                       {
+                               paddr_eth0_0 = &ifa->ifa_address;
+                               break;
+                       }
+               }
+               in_dev_put(in_dev);
+       }
+       if (paddr_eth0 == NULL || paddr_eth0_0 == NULL)
+       {
+               local_irq_restore(flags);
+               return -ENODATA;
+       }
+
+       /* find pointer to IP address of veth0 */
+       if ((in_dev=in_dev_get(net_devs[SVIP_NET_DEV_VETH0_IDX])) != NULL)
+       {
+               for (ifa = in_dev->ifa_list; ifa != NULL; ifa = ifa->ifa_next)
+               {
+                       if (ifa->ifa_address != 0)
+                       {
+                               paddr_veth0 = &ifa->ifa_address;
+                               pmask_veth0 = &ifa->ifa_mask;
+                               break;
+                       }
+               }
+               in_dev_put(in_dev);
+       }
+       if (paddr_veth0 == NULL)
+       {
+               local_irq_restore(flags);
+               return -ENODATA;
+       }
+
+       MOD_INC_USE_COUNT;
+       nDeviceOpen++;
+       local_irq_restore(flags);
+
+       return 0;
+}
+
+
+/******************************************************************************/
+/**
+  This function is called when a process closes the SVIP NAT device file
+
+  \arguments
+  inode       - pointer to disk file data
+  file        - pointer to device file data
+
+  \return
+  0 on success, else -1
+
+*/
+/******************************************************************************/
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,2,0)
+static int SVIP_NAT_device_release(struct inode *inode,
+                                  struct file *file)
+#else
+static void SVIP_NAT_device_release(struct inode *inode,
+                                   struct file *file)
+#endif
+{
+       unsigned long flags;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+       save_flags(flags);
+       cli();
+#else
+       local_save_flags(flags);
+#endif
+
+       /* The device can now be openned by the next caller */
+       nDeviceOpen--;
+
+       MOD_DEC_USE_COUNT;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+       restore_flags(flags);
+#else
+       local_irq_restore(flags);
+#endif
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,2,0)
+       return 0;
+#endif
+}
+
+
+/******************************************************************************/
+/**
+  This function is called when a process closes the SVIP NAT device file
+
+  \arguments
+  inode       - pointer to disk file data
+  file        - pointer to device file data
+  ioctl_num   - ioctl number requested
+  ioctl_param - pointer to data related to the ioctl number
+
+  \return
+  0 on success, else -1
+
+*/
+/******************************************************************************/
+long SVIP_NAT_device_ioctl (struct file *file,
+                          unsigned int ioctl_num, unsigned long ioctl_param)
+{
+       int ret = 0;
+       SVIP_NAT_IO_Rule_t *pNatRule, *pNatRuleIn;
+       SVIP_UDP_PORT_t nPort;
+       int nNatIdx;
+       int bWrite = 0;
+       int bRead = 0;
+       unsigned char *pData = 0;
+       int nSize;
+
+       if (_IOC_DIR(ioctl_num) & _IOC_WRITE)
+               bWrite = 1;
+       if (_IOC_DIR(ioctl_num) & _IOC_READ)
+               bRead = 1;
+       nSize = _IOC_SIZE(ioctl_num);
+
+       if (nSize > sizeof(int))
+       {
+               if (bRead || bWrite)
+               {
+                       pData = kmalloc (nSize, GFP_KERNEL);
+                       if (bWrite)
+                       {
+                               if (copy_from_user ((void *)pData, (void *)ioctl_param, nSize) != 0)
+                               {
+                                       printk(KERN_ERR "SVIP NAT: ioctl %x: copy_from_user() failed!\n", ioctl_num);
+                                       ret = -1;
+                                       goto error;
+                               }
+                       }
+               }
+       }
+
+       switch (ioctl_num)
+       {
+       case FIO_SVIP_NAT_RULE_ADD:
+
+               pNatRuleIn = (SVIP_NAT_IO_Rule_t *)pData;
+
+               /* check if destination UDP port is within range */
+               nPort = ntohs(pNatRuleIn->locUDP);
+
+               if (!SVIP_PORT_INRANGE(nPort))
+               {
+                       printk(KERN_ERR "SVIP NAT: Error, UDP port(%d) is out of range(%d..%d)\n",
+                              nPort, SVIP_UDP_FROM, SVIP_UDP_TO);
+                       ret = -1;
+                       goto error;
+               }
+               nNatIdx = SVIP_PORT_INDEX(nPort);
+
+               down(sem_nat_tbl_access);
+               pNatRule = &pNatTable[nNatIdx].natRule;
+
+               /* add rule to the NAT table */
+               pNatRule->remIP  = pNatRuleIn->remIP;
+               memcpy((char *)pNatRule->remMAC, (char *)pNatRuleIn->remMAC, ETH_ALEN);
+               pNatRule->locIP  = pNatRuleIn->locIP;
+               memcpy((char *)pNatRule->locMAC, (char *)pNatRuleIn->locMAC, ETH_ALEN);
+               pNatRule->locUDP = pNatRuleIn->locUDP;
+
+               memset(pNatTable[nNatIdx].natStats, 0,
+                      sizeof(SVIP_NAT_stats_t)*SVIP_NAT_STATS_TYPES);
+               up(sem_nat_tbl_access);
+               break;
+
+       case FIO_SVIP_NAT_RULE_REMOVE:
+
+               pNatRuleIn = (SVIP_NAT_IO_Rule_t *)pData;
+
+               /* check if destination UDP port is within range */
+               nPort = ntohs(pNatRuleIn->locUDP);
+               if (!SVIP_PORT_INRANGE(nPort))
+               {
+                       printk(KERN_ERR "SVIP NAT: Error, UDP port(%d) is out of range(%d..%d)\n",
+                              nPort, SVIP_UDP_FROM, SVIP_UDP_TO);
+                       ret = -1;
+                       goto error;
+               }
+               nNatIdx = SVIP_PORT_INDEX(nPort);
+               down(sem_nat_tbl_access);
+               /* remove rule from the NAT table */
+               memset(&pNatTable[nNatIdx], 0, sizeof(SVIP_NAT_table_entry_t));
+               up(sem_nat_tbl_access);
+               break;
+
+       case FIO_SVIP_NAT_RULE_LIST:
+               {
+                       int len;
+                       char buf[256];
+
+                       down(sem_nat_tbl_access);
+                       while (nProcReadIdx != -1)
+                       {
+                               len = SVIP_NAT_ProcReadNAT(buf, 256);
+                               if (len > 0)
+                                       printk("%s", buf);
+                       }
+                       nProcReadIdx = 0;
+                       up(sem_nat_tbl_access);
+                       break;
+               }
+
+       default:
+               printk(KERN_ERR "SVIP NAT: unsupported ioctl (%x) command for device %s\n",
+                      ioctl_num, PATH_SVIP_NAT_DEVICE_NAME);
+               ret = -1;
+               goto error;
+       }
+
+       if (nSize > sizeof(int))
+       {
+               if (bRead)
+               {
+                       if (copy_to_user ((void *)ioctl_param, (void *)pData, nSize) != 0)
+                       {
+                               printk(KERN_ERR "SVIP NAT: ioctl %x: copy_to_user() failed!\n", ioctl_num);
+                               ret = -1;
+                               goto error;
+                       }
+               }
+       }
+
+error:
+       if (pData)
+               kfree(pData);
+
+       return ret;
+}
+
+#if 0
+void dump_msg(unsigned char *pData, unsigned int nLen)
+{
+       int i;
+
+       for (i=0; i<nLen; i++)
+       {
+               if (!i || !(i%16))
+                       printk("\n    ");
+               else if (i && !(i%4))
+                       printk(" ");
+               printk("%02x", pData[i]);
+       }
+       if (--i%16)
+               printk("\n");
+}
+#endif
+
+/******************************************************************************/
+/**
+  Used to recalculate IP/UDP checksum using the original IP/UDP checksum
+  coming with the packet. The original source and destination IP addresses
+  are accounted for, and, the checksum is updated using the new source and
+  destination IP addresses.
+
+  \arguments
+  skb         - pointer to the receiving socket buffer
+  csum_old    - original checksum
+  saddr_old   - pointer to original source IP address
+  saddr_new   - pointer to new source IP address
+  daddr_old   - pointer to original destination IP address
+  daddr_new   - pointer to new destination IP address
+
+  \return
+  recalculated IP/UDP checksum
+  */
+/******************************************************************************/
+static inline u16 ip_udp_quick_csum(u16 csum_old, u16 *saddr_old, u16 *saddr_new,
+                                   u16 *daddr_old, u16 *daddr_new)
+{
+       u32 sum;
+
+       sum = csum_old;
+
+       /* convert back from one's complement */
+       sum = ~sum & 0xffff;
+
+       if (sum < saddr_old[0]) sum += 0xffff;
+       sum -= saddr_old[0];
+       if (sum < saddr_old[1]) sum += 0xffff;
+       sum -= saddr_old[1];
+       if (sum < daddr_old[0]) sum += 0xffff;
+       sum -= daddr_old[0];
+       if (sum < daddr_old[1]) sum += 0xffff;
+       sum -= daddr_old[1];
+
+       sum += saddr_new[0];
+       sum += saddr_new[1];
+       sum += daddr_new[0];
+       sum += daddr_new[1];
+
+       /* take only 16 bits out of the 32 bit sum and add up the carries */
+       while (sum >> 16)
+               sum = (sum & 0xffff)+((sum >> 16) & 0xffff);
+
+       /* one's complement the result */
+       sum = ~sum;
+
+       return (u16)(sum & 0xffff);
+}
+
+
+/******************************************************************************/
+/**
+  Returns a pointer to an ipv4 address assigned to device dev. The ipv4
+  instance checked is pointed to by ifa_start. The function is suited for
+  itterative calls.
+
+  \arguments
+  dev         - pointer to network interface
+  ifa_start   - pointer to ipv4 instance to return ipv4 address assigned
+  to, NULL for the first one
+  ppifa_addr   - output parameter
+
+  \return
+  pointer to the next ipv4 instance, which can be null if ifa_start was
+  the last instance present
+  */
+/******************************************************************************/
+static struct in_ifaddr *get_ifaddr(struct net_device *dev,
+                                   struct in_ifaddr *ifa_start, unsigned int **ppifa_addr)
+{
+       struct in_device *in_dev;
+       struct in_ifaddr *ifa = NULL;
+
+       if ((in_dev=in_dev_get(dev)) != NULL)
+       {
+               if (ifa_start == NULL)
+                       ifa = in_dev->ifa_list;
+               else
+                       ifa = ifa_start;
+               if (ifa)
+               {
+                       *ppifa_addr = &ifa->ifa_address;
+                       ifa = ifa->ifa_next;
+               }
+               in_dev_put(in_dev);
+               return ifa;
+       }
+       *ppifa_addr = NULL;
+       return NULL;
+}
+
+/******************************************************************************/
+/**
+  This function performs IP NAT for received packets satisfying the
+  following requirements:
+
+  - packet is destined to local IP host
+  - transport protocol type is UDP
+  - destination UDP port is within range
+
+  \arguments
+  skb         - pointer to the receiving socket buffer
+
+  \return
+  returns 1 on performed SVIP NAT, else returns 0
+
+  \remarks
+  When function returns 0, it indicates the caller to pass the
+  packet up the IP stack to make further decision about it
+  */
+/******************************************************************************/
+int do_SVIP_NAT (struct sk_buff *skb)
+{
+       struct net_device *real_dev;
+       struct iphdr *iph;
+       struct udphdr *udph;
+       SVIP_NAT_IO_Rule_t *pNatRule;
+       int nNatIdx, in_eth0, nDir;
+#ifndef VLAN_8021Q_UNUSED
+       int vlan;
+       unsigned short vid;
+#endif /* ! VLAN_8021Q_UNUSED */
+       SVIP_UDP_PORT_t nPort;
+       u32 orgSrcIp, orgDstIp, *pSrcIp, *pDstIp;
+       struct ethhdr *ethh;
+
+       /* do not consider if SVIP NAT device not open. */
+       if (!nDeviceOpen)
+       {
+               return 0;
+       }
+
+       /* consider only UDP packets. */
+       iph = SVIP_NAT_IP_HDR(skb);
+       if (iph->protocol != IPPROTO_UDP)
+       {
+               return 0;
+       }
+
+       udph = (struct udphdr *)((u_int32_t *)iph + iph->ihl);
+       /* consider only packets which UDP port numbers reside within
+          the predefined SVIP NAT UDP port range. */
+       if ((!SVIP_PORT_INRANGE(ntohs(udph->dest))) &&
+           (!SVIP_PORT_INRANGE(ntohs(udph->source))))
+       {
+               return 0;
+       }
+
+#ifndef VLAN_8021Q_UNUSED
+       /* check if packet delivered over VLAN. VLAN packets will be routed over
+          the VLAN interfaces of the respective real Ethernet interface, if one
+          exists(VIDs must match). Else, the packet will be send out as IEEE 802.3
+          Ethernet frame */
+       if (skb->dev->priv_flags & IFF_802_1Q_VLAN)
+       {
+               vlan = 1;
+               vid = VLAN_DEV_VLAN_ID(skb->dev);
+               real_dev = VLAN_DEV_REAL_DEV(skb->dev);
+       }
+       else
+       {
+               vlan = 0;
+               vid = 0;
+               real_dev = skb->dev;
+       }
+#endif /* ! VLAN_8021Q_UNUSED */
+
+#ifdef CONFIG_SVIP_FW_PKT_SNIFFER
+       /** Debugging feature which can be enabled by writing,
+         'echo 1 > /proc/net/svip_nat/snifferOnOff'.
+         It copies all packets received on veth0 and, sends them out over eth0.
+         When a destination MAC address is specified through
+         /proc/net/svip_nat/snifferMAC, this MAC addess will substitute the
+         original MAC address of the packet.
+         It is recommended to specify a MAC address of some host where Wireshark
+         runs and sniffs for this traffic, else you may flood your LAN with
+         undeliverable traffic.
+
+NOTE: In case of VLAN traffic the VLAN header information is lost. */
+       if (nSVIP_NAT_Sniffer)
+       {
+               if (real_dev == net_devs[SVIP_NET_DEV_VETH0_IDX])
+               {
+                       struct sk_buff *copied_skb;
+
+                       /* gain the Ethernet header from the skb */
+                       skb_push(skb, ETH_HLEN);
+
+                       copied_skb = skb_copy (skb, GFP_ATOMIC);
+
+                       if (nSVIP_NAT_SnifferMacSet == 1)
+                       {
+                               ethh = (struct ethhdr *)SVIP_NAT_SKB_MAC_HEADER(copied_skb);
+                               memcpy((char *)ethh->h_dest, (char *)pSVIP_NAT_SnifferMAC, ETH_ALEN);
+                       }
+                       copied_skb->dev = net_devs[SVIP_NET_DEV_ETH0_IDX];
+                       dev_queue_xmit(copied_skb);
+
+                       /* skip the ETH header again */
+                       skb_pull(skb, ETH_HLEN);
+               }
+       }
+#endif
+
+
+       /* check if packet arrived on eth0 */
+       if (real_dev == net_devs[SVIP_NET_DEV_ETH0_IDX])
+       {
+               /* check if destination IP address equals the primary assigned IP address
+                  of interface eth0. This is the case of packets originating from a
+                  remote peer that are to be delivered to a channel residing on THIS
+                  voice linecard system. This is typical SVIP NAT case, therefore this
+                  rule is placed on top. */
+               if (iph->daddr == *paddr_eth0)
+               {
+                       nPort = ntohs(udph->dest);
+                       nDir = SVIP_NAT_STATS_REM2LOC;
+               }
+               /* check if destination IP address equals the secondary assigned IP address
+                  of interface eth0. This is not a typical SVIP NAT case. It is basically
+                  there, as someone might like for debugging purpose to use the LCC to route
+                  Slave SVIP packets which are part of voice/fax streaming. */
+               else if (iph->daddr == *paddr_eth0_0)
+               {
+                       nPort = ntohs(udph->source);
+                       nDir = SVIP_NAT_STATS_LOC2REM;
+               }
+#ifndef VLAN_8021Q_UNUSED
+               /* when the packet did not hit the top two rules, here we check if the packet
+                  has addressed any of the IP addresses assigned to the VLAN interface attached
+                  to eth0. This is not recommended approach because of the CPU cost incurred. */
+               else if (vlan)
+               {
+                       unsigned int *pifa_addr;
+                       struct in_ifaddr *ifa_start = NULL;
+                       int i = 0;
+
+                       do
+                       {
+                               ifa_start = get_ifaddr(skb->dev, ifa_start, &pifa_addr);
+                               if (!pifa_addr)
+                               {
+                                       /* VLAN packet received on vlan interface attached to eth0,
+                                          however no IP address assigned to the interface.
+                                          The packet is ignored. */
+                                       return 0;
+                               }
+                               if (iph->daddr == *pifa_addr)
+                               {
+                                       /* packet destined to... */
+                                       break;
+                               }
+                               if (!ifa_start)
+                               {
+                                       return 0;
+                               }
+                               i++;
+                       } while (ifa_start);
+                       if (!i)
+                       {
+                               /* ...primary assigned IP address to the VLAN interface. */
+                               nPort = ntohs(udph->dest);
+                               nDir = SVIP_NAT_STATS_REM2LOC;
+                       }
+                       else
+                       {
+                               /* ...secondary assigned IP address to the VLAN interface. */
+                               nPort = ntohs(udph->source);
+                               nDir = SVIP_NAT_STATS_LOC2REM;
+                       }
+               }
+#endif /* ! VLAN_8021Q_UNUSED */
+               else
+               {
+                       return 0;
+               }
+               in_eth0 = 1;
+       }
+       /* check if packet arrived on veth0 */
+       else if (real_dev == net_devs[SVIP_NET_DEV_VETH0_IDX])
+       {
+               nPort = ntohs(udph->source);
+               nDir = SVIP_NAT_STATS_LOC2REM;
+               in_eth0 = 0;
+       }
+       else
+       {
+               /* packet arrived neither on eth0, nor veth0 */
+               return 0;
+       }
+
+       /* calculate the respective index of the NAT table */
+       nNatIdx = SVIP_PORT_INDEX(nPort);
+       /* process the packet if a respective NAT rule exists */
+       pNatRule = &pNatTable[nNatIdx].natRule;
+
+       ethh = (struct ethhdr *)SVIP_NAT_SKB_MAC_HEADER(skb);
+
+       /* copy packet's original source and destination IP addresses to use
+          later on to perform efficient checksum recalculation */
+       orgSrcIp = iph->saddr;
+       orgDstIp = iph->daddr;
+
+       if (in_eth0)
+       {
+               u8 *pDstMac;
+
+               /* Process packet arrived on eth0 */
+
+               if (nDir == SVIP_NAT_STATS_REM2LOC && iph->saddr == pNatRule->remIP)
+               {
+                       pDstIp = &pNatRule->locIP;
+                       pDstMac = pNatRule->locMAC;
+               }
+               else if (nDir == SVIP_NAT_STATS_LOC2REM && iph->saddr == pNatRule->locIP)
+               {
+                       pDstIp = &pNatRule->remIP;
+                       pDstMac = pNatRule->remMAC;
+               }
+               else
+               {
+                       /* Rule check failed. The packet is passed up the layers,
+                          it will be dropped by UDP */
+                       return 0;
+               }
+
+               if ((*pDstIp & *pmask_veth0) == (*paddr_veth0 & *pmask_veth0))
+               {
+#ifndef VLAN_8021Q_UNUSED
+                       if (vlan)
+                       {
+                               struct net_device *vlan_dev;
+
+                               spin_lock_bh(&vlan_group_lock);
+                               vlan_dev = __vlan_find_dev_deep(net_devs[SVIP_NET_DEV_VETH0_IDX], vid);
+                               spin_unlock_bh(&vlan_group_lock);
+                               if (vlan_dev)
+                               {
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+                                       struct vlan_ethhdr *vethh;
+
+                                       skb_push(skb, VLAN_ETH_HLEN);
+                                       /* reconstruct the VLAN header.
+NOTE: priority information is lost */
+                                       vethh = (struct vlan_ethhdr *)skb->data;
+                                       vethh->h_vlan_proto = htons(ETH_P_8021Q);
+                                       vethh->h_vlan_TCI = htons(vid);
+                                       vethh->h_vlan_encapsulated_proto = htons(ETH_P_IP);
+                                       ethh = (struct ethhdr *)vethh;
+#else
+                                       skb_push(skb, ETH_HLEN);
+#endif
+                                       skb->dev = vlan_dev;
+                               }
+                               else
+                               {
+                                       skb->dev = net_devs[SVIP_NET_DEV_VETH0_IDX];
+                                       skb_push(skb, ETH_HLEN);
+                               }
+                       }
+                       else
+#endif /* ! VLAN_8021Q_UNUSED */
+                       {
+                               skb->dev = net_devs[SVIP_NET_DEV_VETH0_IDX];
+                               skb_push(skb, ETH_HLEN);
+                       }
+                       pSrcIp = paddr_veth0;
+               }
+               else
+               {
+#ifndef VLAN_8021Q_UNUSED
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+                       if (vlan)
+                       {
+                               struct vlan_ethhdr *vethh;
+
+                               /* reconstruct the VLAN header.
+NOTE: priority information is lost */
+                               skb_push(skb, VLAN_ETH_HLEN);
+                               vethh = (struct vlan_ethhdr *)skb->data;
+                               vethh->h_vlan_proto = htons(ETH_P_8021Q);
+                               vethh->h_vlan_TCI = htons(vid);
+                               vethh->h_vlan_encapsulated_proto = htons(ETH_P_IP);
+                               ethh = (struct ethhdr *)vethh;
+                       }
+                       else
+#endif
+#endif /* ! VLAN_8021Q_UNUSED */
+                       {
+                               skb_push(skb, ETH_HLEN);
+                       }
+                       /* source IP address equals the destination IP address
+                          of the incoming packet */
+                       pSrcIp = &iph->daddr;
+               }
+               iph->saddr = *pSrcIp;
+               memcpy((char *)ethh->h_source, (char *)skb->dev->dev_addr, ETH_ALEN);
+               iph->daddr = *pDstIp;
+               memcpy((char *)ethh->h_dest, (char *)pDstMac, ETH_ALEN);
+       }
+       else
+       {
+               /* Process packet arrived on veth0 */
+
+               if (iph->saddr != pNatRule->locIP)
+               {
+                       /* Rule check failed. The packet is passed up the layers,
+                          it will be dropped by UDP */
+                       return 0;
+               }
+
+               if (!((pNatRule->remIP & *pmask_veth0) == (*paddr_veth0 & *pmask_veth0)))
+               {
+#ifndef VLAN_8021Q_UNUSED
+                       if (vlan)
+                       {
+                               struct net_device *vlan_dev;
+
+                               spin_lock_bh(&vlan_group_lock);
+                               vlan_dev = __vlan_find_dev_deep(net_devs[SVIP_NET_DEV_ETH0_IDX], vid);
+                               spin_unlock_bh(&vlan_group_lock);
+                               if (vlan_dev)
+                               {
+                                       unsigned int *pifa_addr;
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+                                       struct vlan_ethhdr *vethh;
+
+                                       skb_push(skb, VLAN_ETH_HLEN);
+                                       /* construct the VLAN header, note priority information is lost */
+                                       vethh = (struct vlan_ethhdr *)skb->data;
+                                       vethh->h_vlan_proto = htons(ETH_P_8021Q);
+                                       vethh->h_vlan_TCI = htons(vid);
+                                       vethh->h_vlan_encapsulated_proto = htons(ETH_P_IP);
+                                       ethh = (struct ethhdr *)vethh;
+#else
+                                       skb_push(skb, ETH_HLEN);
+#endif
+                                       skb->dev = vlan_dev;
+
+                                       get_ifaddr(skb->dev, NULL, &pifa_addr);
+                                       if (pifa_addr)
+                                       {
+                                               pSrcIp = pifa_addr;
+                                       }
+                                       else
+                                       {
+                                               pSrcIp = paddr_eth0;
+                                       }
+                               }
+                               else
+                               {
+                                       skb->dev = net_devs[SVIP_NET_DEV_ETH0_IDX];
+                                       pSrcIp = paddr_eth0;
+                                       skb_push(skb, ETH_HLEN);
+                               }
+                       }
+                       else
+#endif /* ! VLAN_8021Q_UNUSED */
+                       {
+                               skb->dev = net_devs[SVIP_NET_DEV_ETH0_IDX];
+                               pSrcIp = paddr_eth0;
+                               skb_push(skb, ETH_HLEN);
+                       }
+               }
+               else
+               {
+                       pSrcIp = paddr_veth0;
+#ifndef VLAN_8021Q_UNUSED
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+                       if (vlan)
+                       {
+                               struct vlan_ethhdr *vethh;
+
+                               skb_push(skb, VLAN_ETH_HLEN);
+                               /* reconstruct the VLAN header.
+NOTE: priority information is lost. */
+                               vethh = (struct vlan_ethhdr *)skb->data;
+                               vethh->h_vlan_proto = htons(ETH_P_8021Q);
+                               vethh->h_vlan_TCI = htons(vid);
+                               vethh->h_vlan_encapsulated_proto = htons(ETH_P_IP);
+                               ethh = (struct ethhdr *)vethh;
+                       }
+                       else
+#endif
+#endif /* ! VLAN_8021Q_UNUSED */
+                       {
+                               skb_push(skb, ETH_HLEN);
+                       }
+               }
+               iph->saddr = *pSrcIp;
+               memcpy((char *)ethh->h_source, (char *)skb->dev->dev_addr, ETH_ALEN);
+               iph->daddr = pNatRule->remIP;
+               memcpy((char *)ethh->h_dest, (char *)pNatRule->remMAC, ETH_ALEN);
+       }
+       pNatTable[nNatIdx].natStats[nDir].inPackets++;
+
+       iph->check = ip_udp_quick_csum(iph->check, (u16 *)&orgSrcIp, (u16 *)&iph->saddr,
+                                      (u16 *)&orgDstIp, (u16 *)&iph->daddr);
+       if (udph->check != 0)
+       {
+               udph->check = ip_udp_quick_csum(udph->check, (u16 *)&orgSrcIp, (u16 *)&iph->saddr,
+                                               (u16 *)&orgDstIp, (u16 *)&iph->daddr);
+       }
+
+       /* write the packet out, directly to the network device */
+       if (dev_queue_xmit(skb) < 0)
+               pNatTable[nNatIdx].natStats[nDir].outErrors++;
+       else
+               pNatTable[nNatIdx].natStats[nDir].outPackets++;
+
+       return 1;
+}
+
+/******************************************************************************/
+/**
+  Function executed upon unloading of the SVIP NAT module. It unregisters the
+  SVIP NAT configuration device and frees the memory used for the NAT table.
+
+  \remarks:
+  Currently the SVIP NAT module is statically linked into the Linux kernel
+  therefore this routine cannot be executed.
+ *******************************************************************************/
+static int __init init(void)
+{
+       int ret = 0;
+       struct net_device *dev;
+
+       if (misc_register(&SVIP_NAT_miscdev) != 0)
+       {
+               printk(KERN_ERR "%s: cannot register SVIP NAT device node.\n",
+                      SVIP_NAT_miscdev.name);
+               return -EIO;
+       }
+
+       /* allocation of memory for NAT table */
+       pNatTable = (SVIP_NAT_table_entry_t *)kmalloc(
+                                                     sizeof(SVIP_NAT_table_entry_t) * SVIP_SYS_CODEC_NUM, GFP_ATOMIC);
+       if (pNatTable == NULL)
+       {
+               printk (KERN_ERR "SVIP NAT: Error(%d), allocating memory for NAT table\n", ret);
+               return -1;
+       }
+
+       /* clear the NAT table */
+       memset((void *)pNatTable, 0, sizeof(SVIP_NAT_table_entry_t) * SVIP_SYS_CODEC_NUM);
+
+       if ((sem_nat_tbl_access = kmalloc(sizeof(struct semaphore), GFP_KERNEL)))
+       {
+               sema_init(sem_nat_tbl_access, 1);
+       }
+
+       SVIP_NAT_ProcInstall();
+
+       /* find pointers to 'struct net_device' of eth0 and veth0, respectevely */
+       read_lock(&dev_base_lock);
+       SVIP_NAT_FOR_EACH_NETDEV(dev)
+       {
+               if (!strcmp(dev->name, SVIP_NET_DEV_ETH0_NAME))
+               {
+                       net_devs[SVIP_NET_DEV_ETH0_IDX] = dev;
+               }
+               if (!strcmp(dev->name, SVIP_NET_DEV_VETH1_NAME))
+               {
+                       net_devs[SVIP_NET_DEV_VETH0_IDX] = dev;
+               }
+               else if (!strcmp(dev->name, SVIP_NET_DEV_ETH1_NAME))
+               {
+                       net_devs[SVIP_NET_DEV_VETH0_IDX] = dev;
+               }
+       }
+       read_unlock(&dev_base_lock);
+
+       if (net_devs[SVIP_NET_DEV_ETH0_IDX] == NULL ||
+           net_devs[SVIP_NET_DEV_VETH0_IDX] == NULL)
+       {
+               printk (KERN_ERR "SVIP NAT: Error, unable to locate eth0 and veth0 interfaces\n");
+               return -1;
+       }
+
+       printk ("%s, (c) 2009, Lantiq Deutschland GmbH\n", &SVIP_NAT_INFO_STR[4]);
+
+       return ret;
+}
+
+/******************************************************************************/
+/**
+  Function executed upon unloading of the SVIP NAT module. It unregisters the
+  SVIP NAT configuration device and frees the memory used for the NAT table.
+
+  \remarks:
+  Currently the SVIP NAT module is statically linked into the Linux kernel
+  therefore this routine cannot be executed.
+ *******************************************************************************/
+static void __exit fini(void)
+{
+       MOD_DEC_USE_COUNT;
+
+       /* unregister SVIP NAT configuration device */
+       misc_deregister(&SVIP_NAT_miscdev);
+
+       /* release memory of SVIP NAT table */
+       if (pNatTable != NULL)
+       {
+               kfree (pNatTable);
+       }
+}
+
+module_init(init);
+module_exit(fini);