add siit for stateless ip and icmp translation to ipv6
authornbd <nbd@3c298f89-4303-0410-b956-a3cf2f4a3e73>
Mon, 22 Dec 2008 23:33:10 +0000 (23:33 +0000)
committernbd <nbd@3c298f89-4303-0410-b956-a3cf2f4a3e73>
Mon, 22 Dec 2008 23:33:10 +0000 (23:33 +0000)
git-svn-id: svn://svn.openwrt.org/openwrt/trunk@13729 3c298f89-4303-0410-b956-a3cf2f4a3e73

package/siit/Makefile [new file with mode: 0644]
package/siit/src/Makefile [new file with mode: 0644]
package/siit/src/siit.c [new file with mode: 0644]
package/siit/src/siit.h [new file with mode: 0644]

diff --git a/package/siit/Makefile b/package/siit/Makefile
new file mode 100644 (file)
index 0000000..0636301
--- /dev/null
@@ -0,0 +1,43 @@
+# 
+# Copyright (C) 2006 OpenWrt.org
+#
+# This is free software, licensed under the GNU General Public License v2.
+# See /LICENSE for more information.
+#
+# $Id: Makefile 10203 2008-01-15 03:25:11Z matteo $
+
+include $(TOPDIR)/rules.mk
+include $(INCLUDE_DIR)/kernel.mk
+
+PKG_NAME:=siit
+PKG_VERSION:=1.1
+
+include $(INCLUDE_DIR)/package.mk
+
+define KernelPackage/siit
+  SUBMENU:=Network Devices
+  TITLE:=Stateless IP ICMP Translation Algorithm
+  FILES:=$(PKG_BUILD_DIR)/siit.$(LINUX_KMOD_SUFFIX)
+  AUTOLOAD:=$(call AutoLoad,50,siit)
+endef
+
+define KernelPackage/siit/description
+ Stateless IP ICMP Translation Algorithm
+endef
+
+define Build/Prepare
+       mkdir -p $(PKG_BUILD_DIR)
+       cp src/Makefile src/siit.h src/siit.c $(PKG_BUILD_DIR)/
+endef
+
+define Build/Compile
+       $(MAKE) -C $(LINUX_DIR) \
+               CROSS_COMPILE="$(TARGET_CROSS)" \
+               ARCH="$(LINUX_KARCH)" \
+               SUBDIRS="$(PKG_BUILD_DIR)" \
+               KERNELDIR=$(LINUX_DIR) \
+               CC="$(TARGET_CC)" \
+               modules
+endef
+
+$(eval $(call KernelPackage,siit))
diff --git a/package/siit/src/Makefile b/package/siit/src/Makefile
new file mode 100644 (file)
index 0000000..06c36f1
--- /dev/null
@@ -0,0 +1,5 @@
+obj-m   := siit.o
+ifeq ($(MAKING_MODULES),1)
+-include $(TOPDIR)/Rules.make
+endif
+
diff --git a/package/siit/src/siit.c b/package/siit/src/siit.c
new file mode 100644 (file)
index 0000000..828df22
--- /dev/null
@@ -0,0 +1,1457 @@
+/*
+ * siit.c: the Stateless IP/ICMP Translator (SIIT) module for Linux.
+ *
+ *
+ */
+
+#include <linux/autoconf.h>
+#include <linux/module.h>
+#include <linux/version.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>       /* printk() */
+#include <linux/slab.h>
+
+#include <linux/errno.h>        /* error codes */
+#include <linux/types.h>        /* size_t */
+#include <linux/interrupt.h>    /* mark_bh */
+#include <linux/random.h>
+#include <linux/in.h>
+#include <linux/netdevice.h>    /* struct device, and other headers */
+#include <linux/etherdevice.h>  /* eth_type_trans */
+#include <net/ip.h>             /* struct iphdr */
+#include <net/icmp.h>           /* struct icmphdr */
+#include <net/ipv6.h>
+#include <net/udp.h>
+#include <linux/skbuff.h>
+#include <linux/in6.h>
+#include <linux/init.h>
+#include <asm/uaccess.h>
+#include <asm/checksum.h>
+
+#include <linux/in6.h>
+#include "siit.h"
+
+MODULE_AUTHOR("Dmitriy Moscalev, Grigory Klyuchnikov, Felix Fietkau");
+
+/*
+ * If tos_ignore_flag != 0, we don't copy TOS and Traffic Class
+ * from origin paket and set it to 0
+ */
+int tos_ignore_flag = 0;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+static inline void
+skb_reset_mac_header(struct sk_buff *skb)
+{
+       skb->mac.raw=skb->data;
+}
+
+static struct net_device_stats *
+siit_get_stats(struct net_device *dev)
+{
+       return netdev_priv(dev);
+}
+
+static inline void random_ether_addr(u8 *addr)
+{
+       get_random_bytes (addr, ETH_ALEN);
+       addr [0] &= 0xfe;       /* clear multicast bit */
+       addr [0] |= 0x02;       /* set local assignment bit (IEEE802) */
+}
+
+
+#define siit_stats(_dev) ((struct net_device_stats *)netdev_priv(_dev))
+#else
+#define siit_stats(_dev) (&(_dev)->stats)
+#endif
+
+/*
+ * The Utility  stuff
+ */
+
+#ifdef SIIT_DEBUG
+/* print dump bytes (data point data area sizeof len and message
+ * before dump.
+ */
+static int siit_print_dump(char *data, int len, char *message)
+{
+       int i;
+       int j = 0, k = 1;
+
+       len = len > BUFF_SIZE ? BUFF_SIZE : len;
+       printk("%s:\n", message);
+       for (i=0; i < len; i++, k++) {
+               if( i == len-1 || k == 16) {
+                       printk("%02x\n", (~(~0 << 8) & *(data+i)));
+                       j = 0;
+                       k = 0;
+               }
+               else if (j) {
+                       printk("%02x ", (~(~0 << 8) & *(data+i)));
+                       j--;
+               }
+               else {
+                       printk("%02x", (~(~0 << 8) & *(data+i)));
+                       j++;
+               }
+       }
+       return 0;
+}
+#endif
+
+/*
+ * Open and close
+ */
+static int siit_open(struct net_device *dev)
+{
+       netif_start_queue(dev);
+       return 0;
+}
+
+
+static int siit_release(struct net_device *dev)
+{
+       netif_stop_queue(dev); /* can't transmit any more */
+       return 0;
+}
+
+/*
+ * Translation IPv4 to IPv6 stuff
+ *
+ * ip4_ip6 (src, len, dst, include_flag)
+ *
+ * where
+ * src - buffer with original IPv4 packet,
+ * len - size of original packet,
+ * dst - new buffer for IPv6 packet,
+ * include_flag - if = 1, dst point to IPv4 packet that is ICMP error
+ *                included IP packet, else = 0
+ */
+
+static int ip4_ip6(char *src, int len, char *dst, int include_flag)
+{
+       struct iphdr *ih4 = (struct iphdr *) src; /* point to current IPv4 header struct */
+       struct icmphdr *icmp_hdr;   /* point to current ICMPv4 header struct */
+       struct udphdr *udp_hdr;     /* point to current IPv4 UDP header struct */
+
+       struct ipv6hdr *ih6 = (struct ipv6hdr *) dst; /* point to current IPv6 header struct */
+       struct frag_hdr *ih6_frag = (struct frag_hdr *)(dst+sizeof(struct ipv6hdr));
+                                                                                     /* point to current IPv6 fragment header struct */
+       struct icmp6hdr *icmp6_hdr; /* point to current ICMPv6 header */
+
+       int hdr_len = (int)(ih4->ihl * 4); /* IPv4 header length */
+       int icmp_len;               /* ICMPv4 packet length */
+       int plen;                   /* payload length */
+
+       unsigned int csum;          /* need to calculate ICMPv6 and UDP checksum */
+       int fl_csum = 0;            /* flag to calculate UDP checksum */
+       int icmperr = 1;            /* flag to indicate ICMP error message and to need
+                                                                  translate ICMP included IP packet */
+       int fr_flag = 0;            /* fragment flag, if = 0 - don't add
+                                                                  fragment header */
+       __u16 new_tot_len;          /* need to calculate IPv6 total length */
+       __u8 new_nexthdr;           /* next header code */
+       __u16 icmp_ptr = 0;         /* Pointer field in ICMP_PARAMETERPROB */
+
+#ifdef SIIT_DEBUG              /* print IPv4 header dump */
+       siit_print_dump(src, hdr_len, "siit: ip4_ip6() (in) ip4 header dump");
+#endif
+
+       /* If DF == 1 && MF == 0 && Fragment Offset == 0
+        * or this packet is ICMP included IP packet
+        * we don't need fragment header */
+       if (ntohs(ih4->frag_off) == IP_DF || include_flag ) {
+               /* not fragment and we need not to add Fragment
+                * Header to IPv6 packet. */
+               /* total length = total length from IPv4 packet */
+               new_tot_len = ntohs(ih4->tot_len);
+
+               if (ih4->protocol == IPPROTO_ICMP)
+                       new_nexthdr = NEXTHDR_ICMP;
+               else
+                       new_nexthdr = ih4->protocol;
+       }
+       else {
+               /* need to add Fragment Header */
+               fr_flag = 1;
+               /* total length = total length from IPv4 packet +
+                  length of Fragment Header */
+               new_tot_len = ntohs(ih4->tot_len) + sizeof(struct frag_hdr);
+               /* IPv6 Header NextHeader = NEXTHDR_FRAGMENT */
+               new_nexthdr = NEXTHDR_FRAGMENT;
+               /* Fragment Header NextHeader copy from IPv4 packet */
+               if (ih4->protocol == IPPROTO_ICMP)
+                       ih6_frag->nexthdr = NEXTHDR_ICMP;
+               else
+                       ih6_frag->nexthdr = ih4->protocol;
+
+               /* copy frag offset from IPv4 packet */
+               ih6_frag->frag_off = htons((ntohs(ih4->frag_off) & IP_OFFSET) << 3);
+               /* copy MF flag from IPv4 packet */
+               ih6_frag->frag_off = htons((ntohs(ih6_frag->frag_off) |
+                                                                       ((ntohs(ih4->frag_off) & IP_MF) >> 13)));
+               /* copy Identification field from IPv4 packet */
+               ih6_frag->identification = htonl(ntohs(ih4->id));
+               /* reserved field initialized to zero */
+               ih6_frag->reserved = 0;
+       }
+
+       /* Form rest IPv6 fields */
+
+       /*
+        * At this point we need to add checking of unxpired source
+        * route optin and if it is, send ICMPv4 "destination
+        * unreacheble/source route failes" Type 3/Code 5 and
+        * drop packet. (NOT RELEASED YET)
+        */
+
+       /* IP version = 6 */
+       ih6->version = 6;
+
+       if (tos_ignore_flag) {
+               ih6->priority = 0;
+               ih6->flow_lbl[0] = 0;
+       } else {
+               ih6->priority = (ih4->tos & 0xf0) >> 4;
+               ih6->flow_lbl[0] = (ih4->tos & 0x0f) << 4;
+       }
+       ih6->flow_lbl[1] = 0;
+       ih6->flow_lbl[2] = 0;
+
+       /* Hop Limit = IPv4 TTL */
+       ih6->hop_limit = ih4->ttl;
+
+       /* Translate source destination addresses,
+          for IPv6 host it's IPv4-translated IPv6 address,
+          for IPv4 host it's IPv4-mapped IPv6 address
+
+          !!WARNING!! Instead IPv4-mapped IPv6 addresses we use addreesses
+          with unused prefix ::ffff:ffff:0:0/96, because KAME implementation
+          doesn't support IPv4-mapped addresses in IPv6 packets and discard them.
+
+       */
+
+       if (include_flag) {
+               /*
+                  It's ICMP included IP packet and there is a diffirence
+                  in src/dst addresses then src/dst in normal direction
+                */
+
+               /*
+                  Source address
+                  is IPv4-translated IPv6 address because packet traveled
+                  from IPv6 to IPv4 area
+               */
+               ih6->saddr.in6_u.u6_addr32[0] = 0;
+               ih6->saddr.in6_u.u6_addr32[1] = 0;
+               ih6->saddr.in6_u.u6_addr32[2] = htonl(TRANSLATED_PREFIX); /* to network order bytes */
+               ih6->saddr.in6_u.u6_addr32[3] = ih4->saddr;
+
+               /*
+                  Destination address
+                  is IPv4-mapped address (but it's not IPv4- mapped, we use
+                  prefix ::ffff:ffff:0:0/96
+                */
+               ih6->daddr.in6_u.u6_addr32[0] = 0;
+               ih6->daddr.in6_u.u6_addr32[1] = 0;
+               ih6->daddr.in6_u.u6_addr32[2] = htonl(MAPPED_PREFIX); /* to network order bytes */
+               ih6->daddr.in6_u.u6_addr32[3] = ih4->daddr;
+       }
+       else {
+
+               /*
+                  This is normal case (packet isn't included IP packet)
+
+                  Source address
+                  is IPv4-mapped address (but it's not IPv4- mapped, we use
+                  prefix ::ffff:ffff:0:0/96)
+               */
+               ih6->saddr.in6_u.u6_addr32[0] = 0;
+               ih6->saddr.in6_u.u6_addr32[1] = 0;
+               ih6->saddr.in6_u.u6_addr32[2] = htonl(MAPPED_PREFIX); /* to network order bytes */
+               ih6->saddr.in6_u.u6_addr32[3] = ih4->saddr;
+
+               /* Destination address
+                  is is IPv4-translated IPv6 address
+                */
+               ih6->daddr.in6_u.u6_addr32[0] = 0;
+               ih6->daddr.in6_u.u6_addr32[1] = 0;
+               ih6->daddr.in6_u.u6_addr32[2] = htonl(TRANSLATED_PREFIX); /* to network order bytes */
+               ih6->daddr.in6_u.u6_addr32[3] = ih4->daddr;
+       }
+
+       /* Payload Length */
+       plen = new_tot_len - hdr_len; /* Payload length = IPv4 total len - IPv4 header len */
+       ih6->payload_len = htons(plen);
+
+       /* Next Header */
+       ih6->nexthdr = new_nexthdr; /* Next Header */
+
+       /* Process ICMP protocols data */
+
+       switch (ih4->protocol) {
+       case IPPROTO_ICMP:
+               if ( (ntohs(ih4->frag_off) & IP_OFFSET) != 0 || (ntohs(ih4->frag_off) & IP_MF) != 0 ) {
+                       PDEBUG("ip4_ip6(): don't translate ICMPv4 fragments - packet dropped.\n");
+                       return -1;
+               }
+
+               icmp_hdr = (struct icmphdr *) (src+hdr_len); /* point to ICMPv4 header */
+               csum = 0;
+               icmp_len =  ntohs(ih4->tot_len) - hdr_len; /* ICMPv4 packet length */
+               icmp6_hdr = (struct icmp6hdr *)(dst+sizeof(struct ipv6hdr)
+                                                                           +fr_flag*sizeof(struct frag_hdr)); /* point to ICMPv6 header */
+
+               if (include_flag) {
+                       /* ICMPv4 packet cannot be included in ICMPv4 Error message */
+                       /* !!! May be it's WRONG !!! ICMPv4 QUERY packet can be included
+                          in ICMPv4 Error message */
+                       PDEBUG("ip4_ip6(): It's included ICMPv4 in ICMPv4 Error message - packet dropped.\n");
+                       return -1;
+               }
+
+               /* Check ICMPv4 Type field */
+               switch (icmp_hdr->type) {
+               /* ICMP Error messages */
+               /* Destination Unreachable (Type 3) */
+               case ICMP_DEST_UNREACH:
+                       icmp6_hdr->icmp6_type = ICMPV6_DEST_UNREACH; /* to Type 1 */
+                       icmp6_hdr->icmp6_unused = 0;
+                       switch (icmp_hdr->code)
+                       {
+                       case ICMP_NET_UNREACH: /* Code 0 */
+                       case ICMP_HOST_UNREACH: /* Code 1 */
+                       case ICMP_SR_FAILED: /* Code 5 */
+                       case ICMP_NET_UNKNOWN: /* Code 6 */
+                       case ICMP_HOST_UNKNOWN: /* Code 7 */
+                       case ICMP_HOST_ISOLATED: /* Code 8 */
+                       case ICMP_NET_UNR_TOS: /* Code 11 */
+                       case ICMP_HOST_UNR_TOS: /* Code 12 */
+                               icmp6_hdr->icmp6_code = ICMPV6_NOROUTE; /* to Code 0 */
+                               break;
+                       case ICMP_PROT_UNREACH: /* Code 2 */
+                               icmp6_hdr->icmp6_type = ICMPV6_PARAMPROB; /* to Type 4 */
+                               icmp6_hdr->icmp6_code = ICMPV6_UNK_NEXTHDR; /* to Code 1 */
+                               /* Set pointer filed to 6, it's octet offset IPv6 Next Header field */
+                               icmp6_hdr->icmp6_pointer = htonl(6);
+                               break;
+                       case ICMP_PORT_UNREACH: /* Code 3 */
+                               icmp6_hdr->icmp6_code = ICMPV6_PORT_UNREACH; /* to Code 4 */
+                               break;
+                       case ICMP_FRAG_NEEDED: /* Code 4 */
+                               icmp6_hdr->icmp6_type = ICMPV6_PKT_TOOBIG; /* to Type 2 */
+                               icmp6_hdr->icmp6_code = 0;
+                               /* Correct MTU  */
+                               if (icmp_hdr->un.frag.mtu == 0)
+                                       /* we use minimum MTU for IPv4 PMTUv4 RFC1191, section 5;
+                                          IPv6 implementation wouldn't accept Path MTU < 1280,
+                                          but it records info correctly to always include
+                                          a fragment header */
+                                       icmp6_hdr->icmp6_mtu = htonl(576);
+                               else
+                                       /* needs to adjusted for difference between IPv4/IPv6 headers
+                                        * SIIT RFC2765, section 3.3,
+                                        * we assume that difference is 20 bytes */
+                                       icmp6_hdr->icmp6_mtu = htonl(ntohs(icmp_hdr->un.frag.mtu)+IP4_IP6_HDR_DIFF);
+
+                               break;
+                       case ICMP_NET_ANO: /* Code 9 */
+                       case ICMP_HOST_ANO: /* Code 10 */
+                               icmp6_hdr->icmp6_code = ICMPV6_ADM_PROHIBITED; /* to Code 1 */
+                               break;
+                       default: /* discard any other Code */
+                               PDEBUG("ip4_ip6(): Unknown ICMPv4 Type %d Code %d - packet dropped.\n",
+                                          ICMP_DEST_UNREACH, icmp_hdr->code);
+                               return -1;
+                       }
+                       break;
+                       /* Time Exceeded (Type 11) */
+               case ICMP_TIME_EXCEEDED:
+                       icmp6_hdr->icmp6_type = ICMPV6_TIME_EXCEED;
+                       icmp6_hdr->icmp6_code = icmp_hdr->code;
+                       break;
+                       /* Parameter Problem (Type 12) */
+               case ICMP_PARAMETERPROB:
+                       icmp6_hdr->icmp6_type = ICMPV6_PARAMPROB;
+                       icmp6_hdr->icmp6_code = icmp_hdr->code;
+
+                       icmp_ptr = ntohs(icmp_hdr->un.echo.id) >> 8;
+                       switch (icmp_ptr) {
+                       case 0:
+                               icmp6_hdr->icmp6_pointer = 0; /* IPv4 Version -> IPv6 Version */
+                               break;
+                       case 2:
+                               icmp6_hdr->icmp6_pointer = __constant_htonl(4); /* IPv4 length -> IPv6 Payload Length */
+                               break;
+                       case 8:
+                               icmp6_hdr->icmp6_pointer = __constant_htonl(7); /* IPv4 TTL -> IPv6 Hop Limit */
+                               break;
+                       case 9:
+                               icmp6_hdr->icmp6_pointer = __constant_htonl(6); /* IPv4 Protocol -> IPv6 Next Header */
+                               break;
+                       case 12:
+                               icmp6_hdr->icmp6_pointer = __constant_htonl(8); /* IPv4 Src Addr -> IPv6 Src Addr */
+                               break;
+                       case 16:
+                               icmp6_hdr->icmp6_pointer = __constant_htonl(24); /* IPv4 Dst Addr -> IPv6 Dst Addr */
+                               break;
+                       default:
+                               icmp6_hdr->icmp6_pointer = 0xffffffff; /* set to all ones in any other cases */
+                               break;
+                       }
+                       break;
+               case ICMP_ECHO:
+                       icmperr = 0;
+                       icmp6_hdr->icmp6_type = ICMPV6_ECHO_REQUEST;
+                       icmp6_hdr->icmp6_code = 0;
+                       /* Copy rest ICMP data to new IPv6 packet without changing */
+                       memcpy(((char *)icmp6_hdr)+4, ((char *)icmp_hdr)+4, len - hdr_len - 4);
+                       break;
+
+               case ICMP_ECHOREPLY:
+                       icmperr = 0;
+                       icmp6_hdr->icmp6_type = ICMPV6_ECHO_REPLY;
+                       icmp6_hdr->icmp6_code = 0;
+                       /* Copy rest ICMP data to new IPv6 packet without changing */
+                       memcpy(((char *)icmp6_hdr)+4, ((char *)icmp_hdr)+4, len - hdr_len - 4);
+                       break;
+
+                       /* Discard any other ICMP messages */
+               default:
+                       PDEBUG("ip4_ip6(): Unknown ICMPv4 packet Type %x - packet dropped.\n", icmp_hdr->type);
+                       return -1;
+               }
+
+               /* Now if it's ICMPv4 Error message we must translate included IP packet */
+
+               if (icmperr) {
+                       /* Call our ip4_ip6() to translate included IP packet */
+                       if (ip4_ip6(src+hdr_len+sizeof(struct icmphdr), len - hdr_len - sizeof(struct icmphdr),
+                                               dst+sizeof(struct ipv6hdr)+fr_flag*sizeof(struct frag_hdr)
+                                               +sizeof(struct icmp6hdr), 1) == -1) {
+                               PDEBUG("ip4_ip6(): Uncorrect translation of ICMPv4 Error message - packet dropped.\n");
+                               return -1;
+                       }
+                       /* correct ICMPv6 packet length for diffirence between IPv4 and IPv6 headers
+                          in included IP packet
+                          */
+                       icmp_len += 20;
+                       /* and correct Payload length for diffirence between IPv4 and IPv6 headers */
+                       plen += 20;
+                       ih6->payload_len = htons(plen);
+               }
+
+               /* Calculate ICMPv6 checksum */
+
+               icmp6_hdr->icmp6_cksum = 0;
+               csum = 0;
+
+               csum = csum_partial((u_char *)icmp6_hdr, icmp_len, csum);
+               icmp6_hdr->icmp6_cksum = csum_ipv6_magic(&ih6->saddr, &ih6->daddr, icmp_len,
+                                                                                IPPROTO_ICMPV6, csum);
+               break;
+
+       /* Process TCP protocols data */
+       case IPPROTO_TCP:
+               /* Copy TCP data to new IPv6 packet without changing */
+               memcpy(dst+sizeof(struct ipv6hdr)+fr_flag*sizeof(struct frag_hdr),
+                                  src+hdr_len, len - hdr_len);
+               break;
+
+       /* Process UDP protocols data */
+       case IPPROTO_UDP:
+               udp_hdr = (struct udphdr *)(src+hdr_len);
+               if ((ntohs(ih4->frag_off) & IP_OFFSET) == 0) {
+                       if ((ntohs(ih4->frag_off) & IP_MF) != 0) {
+                               /* It's a first fragment */
+                               if (udp_hdr->check == 0) {
+                                       /* System management event */
+                                       printk("siit: First fragment of UDP with zero checksum - packet droped\n");
+                                       printk("siit: addr: %x src port: %d dst port: %d\n",
+                                                  htonl(ih4->saddr), htons(udp_hdr->source), htons(udp_hdr->dest));
+                                       return -1;
+                               }
+                       }
+                       else if (udp_hdr->check == 0)
+                               fl_csum = 1;
+               }
+
+               /* Copy UDP data to new IPv6 packet */
+               udp_hdr = (struct udphdr *)(dst+sizeof(struct ipv6hdr)
+                                                                       + fr_flag*sizeof(struct frag_hdr));
+               memcpy((char *)udp_hdr, src+hdr_len, len - hdr_len);
+
+               /* Calculate UDP checksum if UDP checksum in IPv4 packet was ZERO
+                  and if it isn't included IP packet
+                */
+               if (fl_csum && (!include_flag)) {
+                       udp_hdr->check = 0;
+                       csum = 0;
+                       csum = csum_partial((unsigned char *)udp_hdr, plen - fr_flag*sizeof(struct frag_hdr), csum);
+                       udp_hdr->check = csum_ipv6_magic(&ih6->saddr, &ih6->daddr, plen -
+                                                                                    fr_flag*sizeof(struct frag_hdr), IPPROTO_UDP, csum);
+               }
+               break;
+
+       /* Discard packets with any other protocol */
+       default:
+               PDEBUG("ip4_ip6(): Unknown upper protocol - packet dropped.\n");
+               return -1;
+       }
+
+#ifdef SIIT_DEBUG
+       siit_print_dump(dst, sizeof(struct ipv6hdr), "siit: ip4_ip6(): (out) ipv6 header dump");
+#endif
+
+       return 0;
+}
+
+/*
+ * Translation IPv6 to IPv4 stuff
+ *
+ * ip6_ip4(src, len, dst, include_flag)
+ *
+ * where
+ * src - buffer with original IPv6 packet,
+ * len - size of original packet,
+ * dst - new buffer for IPv4 packet,
+ * include_flag - if = 1, dst point to IPv6 packet that is ICMP error
+ *                included IP packet, else = 0
+ *
+ */
+
+static int ip6_ip4(char *src, int len, char *dst, int include_flag)
+{
+       struct ipv6hdr *ip6_hdr;    /* point to current IPv6 header struct */
+       struct iphdr *ip_hdr;       /* point to current IPv4 header struct */
+       int opts_len = 0;           /* to sum Option Headers length */
+       int icmperr = 1;            /* if = 1, indicate that packet is ICMP Error message, else = 0 */
+       int ntot_len = 0;           /* to calculate IPv6 Total Length field */
+       int real_len;
+       int len_delta;
+       int ip6_payload_len;
+       int inc_opts_len = 0;       /* to sum Option Headers length in ICMP included IP packet */
+       __u8 next_hdr;              /* Next Header */
+
+#ifdef SIIT_DEBUG
+       siit_print_dump(src, sizeof(struct ipv6hdr), "siit: ip6_ip4(): (in) ipv6 header dump");
+#endif
+
+       if ( (len_delta = len - sizeof(struct ipv6hdr)) >= 0)
+       {
+               ip6_hdr = (struct ipv6hdr *)src;
+               ip_hdr = (struct iphdr *)dst;
+
+               real_len = sizeof(struct iphdr);
+
+               /* Check validation of Saddr & Daddr? is a packet to fall under our translation? */
+               if (include_flag) { /* It's ICMP included IP packet,
+                                                          about process include_flag see comment in ip4_ip6() */
+                       if (ip6_hdr->saddr.s6_addr32[2] != htonl(MAPPED_PREFIX)) {
+                               PDEBUG("ip6_ip4(): Included IP packet Src addr isn't mapped addr: %x%x%x%x, packet dropped.\n",
+                                          ip6_hdr->saddr.s6_addr32[0], ip6_hdr->saddr.s6_addr32[1],
+                                          ip6_hdr->saddr.s6_addr32[2], ip6_hdr->saddr.s6_addr32[3]);
+                               return -1;
+                       }
+                       if ( ip6_hdr->daddr.s6_addr32[2] != htonl(TRANSLATED_PREFIX)) {
+                               PDEBUG("ip6_ip4(): Included IP packet Dst addr isn't translated addr: %x%x%x%x, packet dropped.\n",
+                                          ip6_hdr->daddr.s6_addr32[0], ip6_hdr->daddr.s6_addr32[1],
+                                          ip6_hdr->daddr.s6_addr32[2], ip6_hdr->daddr.s6_addr32[3]);
+                               return -1;
+                       }
+               }
+               else { /* It's normal IP packet (not included in ICMP) */
+                       if (ip6_hdr->saddr.s6_addr32[2] != htonl(TRANSLATED_PREFIX)) {
+                               PDEBUG("ip6_ip4(): Src addr isn't translated addr: %x%x%x%x, packet dropped.\n",
+                                          ip6_hdr->saddr.s6_addr32[0], ip6_hdr->saddr.s6_addr32[1],
+                                          ip6_hdr->saddr.s6_addr32[2], ip6_hdr->saddr.s6_addr32[3]);
+                               return -1;
+                       }
+                       if ( ip6_hdr->daddr.s6_addr32[2] != htonl(MAPPED_PREFIX)) {
+                               PDEBUG("ip6_ip4(): Dst addr isn't mapped addr: %x%x%x%x, packet dropped.\n",
+                                          ip6_hdr->daddr.s6_addr32[0], ip6_hdr->daddr.s6_addr32[1],
+                                          ip6_hdr->daddr.s6_addr32[2], ip6_hdr->daddr.s6_addr32[3]);
+                               return -1;
+                       }
+               }
+
+               /* Set IPv4 Fragment Offset and ID to 0
+                  before process any Option Headers */
+               ip_hdr->frag_off = 0;
+               ip_hdr->id = 0;
+
+               /*
+                * We process only Fragment Header. Any other options headers
+                * are ignored, i.e. there is no attempt to translate them.
+                * However, the Total Length field and the Protocol field would
+                * have to be adjusted to "skip" these extension headers.
+                */
+
+               next_hdr = ip6_hdr->nexthdr;
+
+               /* Hop_by_Hop options header (ip6_hdr->nexthdr = 0). It must
+                * appear only in IPv6 header's Next Header field.
+                */
+               if (next_hdr == NEXTHDR_HOP) {
+                       if ( (len_delta - sizeof(struct ipv6_opt_hdr)) >= 0)
+                       {
+                               struct ipv6_opt_hdr *ip6h =
+                                       (struct ipv6_opt_hdr *)(src+sizeof(struct ipv6hdr) + opts_len);
+                               if ( (len_delta -= ip6h->hdrlen*8 + 8) >= 0)
+                               {
+                                       opts_len += ip6h->hdrlen*8 + 8;  /* See  RFC 2460 page 11:
+                                                       Hdr Ext Len  8-bit unsigned integer.  Length of the Hop-by-
+                                                                                Hop Options header in 8-octet units, not
+                                                                                including the first 8 octets.
+                                                       */
+                                       next_hdr = ip6h->nexthdr;
+                               }
+                               else
+                               {
+                                       PDEBUG("ip6_ip4(): hop_by_hop header error, packet droped");
+                                       /* Generate ICMP Parameter Problem */
+                                       return -1;
+                               }
+                       }
+               }
+
+               if (len_delta > 0)
+               {
+                       while(next_hdr != NEXTHDR_ICMP && next_hdr != NEXTHDR_TCP
+                                 && next_hdr != NEXTHDR_UDP)
+                       {
+                               /* Destination options header */
+                               if (next_hdr == NEXTHDR_DEST)
+                               {
+                                       if ( (len_delta - sizeof(struct ipv6_opt_hdr)) >= 0)
+                                       {
+                                               struct ipv6_opt_hdr *ip6d =
+                                                       (struct ipv6_opt_hdr *)(src + sizeof(struct ipv6hdr) + opts_len);
+                                               if ( (len_delta -= ip6d->hdrlen*8 + 8) >= 0)
+                                               {
+                                                       opts_len += ip6d->hdrlen*8 + 8;
+                                                       next_hdr = ip6d->nexthdr;
+                                               }
+                                       }
+                                       else
+                                       {
+                                               PDEBUG("ip6_ip4(): destination header error, packet droped");
+                                               /* Generate ICMP Parameter Problem */
+                                               return -1;
+                                       }
+                               }
+                               /* Routing options header */
+                               else if (next_hdr == NEXTHDR_ROUTING)
+                               {
+                                       if ( (len_delta - sizeof(struct ipv6_rt_hdr)) >= 0)
+                                       {
+                                               struct ipv6_rt_hdr *ip6rt =
+                                                       (struct ipv6_rt_hdr *)(src+sizeof(struct ipv6hdr) + opts_len);
+                                               /* RFC 2765 SIIT, 4.1:
+                                                  If a routing header with a non-zero Segments Left field is present
+                                                  then the packet MUST NOT be translated, and an ICMPv6 "parameter
+                                                  problem/ erroneous header field encountered" (Type 4/Code 0) error
+                                                  message, with the Pointer field indicating the first byte of the
+                                                  Segments Left field, SHOULD be returned to the sender.
+                                                  */
+                                               if (ip6rt->segments_left != 0) {
+                                                       /* Build ICMPv6 "Parameter Problem/Erroneous Header
+                                                          Field Encountered" & drop the packet */
+                                                       /* !!! We don't send ICMPv6 "Parameter Problem" !!! */
+                                                       PDEBUG("ip6_ip4(): routing header type != 0\n");
+                                                       return -1;
+                                               }
+                                               if ( (len_delta -= ip6rt->hdrlen*8 + 8) >= 0)
+                                               {
+                                                       opts_len += ip6rt->hdrlen*8 + 8;
+                                                       next_hdr = ip6rt->nexthdr;
+                                               }
+                                               else
+                                               {
+                                                       PDEBUG("ip6_ip4(): routing header error, packet droped");
+                                                       /* Generate ICMP Parameter Problem */
+                                                       return -1;
+                                               }
+                                       }
+                               }
+                               /* Fragment options header */
+                               else if (next_hdr == NEXTHDR_FRAGMENT)
+                               {
+                                       if ( (len_delta -= sizeof(struct frag_hdr)) >= 0)
+                                       {
+                                               struct frag_hdr *ip6f =
+                                                       (struct frag_hdr *)(src+sizeof(struct ipv6hdr)+opts_len);
+
+                                               opts_len += sizeof(struct frag_hdr);      /* Frag Header Length = 8 */
+                                               ip_hdr->id = htons(ntohl(ip6f->identification)); /* ID field */
+                                               ip_hdr->frag_off = htons((ntohs(ip6f->frag_off) & IP6F_OFF_MASK) >> 3);
+                                                                                                                  /* fragment offset */
+                                               ip_hdr->frag_off = htons(ntohs(ip_hdr->frag_off) |
+                                                                                        ((ntohs(ip6f->frag_off) & IP6F_MORE_FRAG) << 13));
+                                                                                                                 /* more fragments flag */
+                                               next_hdr = ip6f->nexthdr;
+                                       }
+                                       else
+                                       {
+                                               PDEBUG("ip6_ip4(): fragment header error, packet droped");
+                                               /* Generate ICMP Parameter Problem */
+                                               return -1;
+                                       }
+                               }
+                               /* No Next Header */
+                               else if (next_hdr == NEXTHDR_NONE)
+                               {
+                                       /* RFC 2460 IPv6 Specification, 4.7
+                                          4.7 No Next Header
+
+                                          The value 59 in the Next Header field of an IPv6 header or any
+                                          extension header indicates that there is nothing following that
+                                          header.  If the Payload Length field of the IPv6 header indicates the
+                                          presence of octets past the end of a header whose Next Header field
+                                          contains 59, those octets must be ignored, and passed on unchanged if
+                                          the packet is forwarded.
+                                          */
+                                       break;
+                               }
+                               else if (next_hdr == NEXTHDR_ESP || next_hdr == NEXTHDR_AUTH)
+                               {
+                                       PDEBUG("ip6_ip4(): cannot translate AUTH or ESP extention header, packet dropped\n");
+                                       return -1;
+                               }
+                               else if (next_hdr == NEXTHDR_IPV6)
+                               {
+                                       PDEBUG("ip6_ip4(): cannot translate IPv6-IPv6 packet, packet dropped\n");
+                                       return -1;
+                               }
+                               else if (next_hdr == 0)
+                               {
+                                       /* As say RFC 2460 (IPv6 Spec) we should discard the packet and send an
+                                          ICMP Parameter Problem message to the source of the packet, with an
+                                          ICMP Code value of 1 ("unrecognized Next Header type encountered")
+                                          and the ICMP Pointer field containing the offset of the unrecognized
+                                          value within the original packet
+                                          */
+                                       /* NOT IMPLEMENTED */
+                                       PDEBUG("ip6_ip4(): NEXTHDR in extention header = 0, packet dropped\n");
+                                       return -1;
+                               }
+                               else
+                               {
+                                       PDEBUG("ip6_ip4(): cannot translate extention header = %d, packet dropped\n", next_hdr);
+                                       return -1;
+                               }
+                       }
+               }
+       }
+       else
+       {
+          PDEBUG("ip6_ip4(): error packet len, packet dropped.\n");
+          return -1;
+       }
+
+       /* Building ipv4 packet */
+
+       ip_hdr->version = IPVERSION;
+       ip_hdr->ihl = 5;
+
+       /* TOS see comment about TOS in ip4_ip6() */
+       if (tos_ignore_flag)
+               ip_hdr->tos = 0;
+       else {
+               ip_hdr->tos = ip6_hdr->priority << 4;
+               ip_hdr->tos = ip_hdr->tos | (ip6_hdr->flow_lbl[0] >> 4);
+       }
+
+       /* IPv4 Total Len = IPv6 Payload Len +
+          IPv4 Header Len (without options) - Options Headers Len */
+       ip6_payload_len = ntohs(ip6_hdr->payload_len);
+
+       if (ip6_payload_len == 0)
+               ntot_len = 0;
+       else
+               ntot_len = ip6_payload_len + IP4_IP6_HDR_DIFF - opts_len;
+
+       ip_hdr->tot_len = htons(ntot_len);
+
+       /* IPv4 TTL = IPv6 Hop Limit */
+       ip_hdr->ttl = ip6_hdr->hop_limit;
+
+       /* IPv4 Protocol = Next Header that will point to upper layer protocol */
+       ip_hdr->protocol = next_hdr;
+
+       /* IPv4 Src addr = last 4 bytes from IPv6 Src addr */
+       ip_hdr->saddr = ip6_hdr->saddr.s6_addr32[3];
+       /* IPv4 Dst addr = last 4 bytes from IPv6 Dst addr */
+       ip_hdr->daddr = ip6_hdr->daddr.s6_addr32[3];
+
+       /* Calculate IPv4 header checksum */
+       ip_hdr->check = 0;
+       ip_hdr->check = ip_fast_csum((unsigned char *)ip_hdr, ip_hdr->ihl);
+
+       if (len_delta > 0)
+       {
+               /* PROCESS ICMP */
+
+               if (next_hdr == NEXTHDR_ICMP)
+               {
+                       struct icmp6hdr *icmp6_hdr;
+                       struct icmphdr *icmp_hdr;
+
+                       if ((len_delta -= sizeof(struct icmp6hdr)) >= 0)
+                       {
+                               icmp6_hdr = (struct icmp6hdr *)(src + sizeof(struct ipv6hdr) + opts_len);
+                               icmp_hdr = (struct icmphdr *)(dst + sizeof(struct iphdr));
+
+                               real_len += len_delta + sizeof(struct icmphdr);
+
+                               /* There is diffirent between ICMPv4/ICMPv6 protocol codes
+                                  IPPROTO_ICMP = 1
+                                  IPPROTO_ICMPV6 = 58        */
+                               ip_hdr->protocol = IPPROTO_ICMP;
+
+                               if (include_flag) {
+                                       /* !!! Warnig !!! We discard ICMP packets with any ICMP as included
+                                          in ICMP Error. But ICMP Error messages can include ICMP Query message
+                                          */
+                                       if (icmp6_hdr->icmp6_type != ICMPV6_ECHO_REQUEST)
+                                       {
+                                               PDEBUG("ip6_ip4(): included ICMPv6 in ICMPv6 Error message, packet dropped\n");
+                                               return -1;
+                                       }
+                               }
+
+                       /* Translate ICMPv6 to ICMPv4 */
+                               switch (icmp6_hdr->icmp6_type)
+                               {
+/* ICMP Error messages */
+                       /* Destination Unreachable (Type 1) */
+                               case ICMPV6_DEST_UNREACH: /* Type 1 */
+                                       icmp_hdr->type = ICMP_DEST_UNREACH; /* to Type 3 */
+                                       icmp_hdr->un.echo.id = 0;
+                                       icmp_hdr->un.echo.sequence = 0;
+                                       switch (icmp6_hdr->icmp6_code)
+                                       {
+                                       case ICMPV6_NOROUTE: /* Code 0 */
+                                       case ICMPV6_NOT_NEIGHBOUR: /* Code 2 */
+                                       case ICMPV6_ADDR_UNREACH: /* Code 3  */
+                                               icmp_hdr->code = ICMP_HOST_UNREACH; /* To Code 1 */
+                                               break;
+                                       case ICMPV6_ADM_PROHIBITED: /* Code 1 */
+                                               icmp_hdr->code = ICMP_HOST_ANO; /* To Code 10 */
+                                               break;
+                                       case ICMPV6_PORT_UNREACH: /* Code 4 */
+                                               icmp_hdr->code = ICMP_PORT_UNREACH; /* To Code 3 */
+
+                                               break;
+                                       default:            /* discard any other codes */
+                                               PDEBUG("ip6_ip4(): Unknown ICMPv6 Type %d Code %d - packet dropped.\n",
+                                                          ICMPV6_DEST_UNREACH, icmp6_hdr->icmp6_code);
+                                               return -1;
+                                       }
+                                       break;
+                       /* Packet Too Big (Type 2) */
+                               case ICMPV6_PKT_TOOBIG: /* Type 2 */
+                                       icmp_hdr->type = ICMP_DEST_UNREACH; /* to Type 3  */
+                                       icmp_hdr->code = ICMP_FRAG_NEEDED; /*  to Code 4 */
+                                       /* Change MTU, RFC 2765 (SIIT), 4.2:
+                                          The MTU field needs to be adjusted for the difference between
+                                          the IPv4 and IPv6 header sizes taking into account whether or
+                                          not the packet in error includes a Fragment header.
+                                          */
+                                       /* !!! Don't implement !!! */
+                                       icmp_hdr->un.frag.mtu = (__u16) icmp6_hdr->icmp6_mtu;
+                                       break;
+                       /* Time Exceeded (Type 3) */
+                               case ICMPV6_TIME_EXCEED:
+                                       icmp_hdr->type = ICMP_TIME_EXCEEDED; /* to Type 11 */
+                                       icmp_hdr->code = icmp6_hdr->icmp6_code; /* Code unchanged */
+                                       break;
+                       /* Parameter Problem (Type 4) */
+                               case ICMPV6_PARAMPROB:
+                                       switch (icmp6_hdr->icmp6_code) {
+                                       case ICMPV6_UNK_NEXTHDR: /* Code 1 */
+                                               icmp_hdr->type = ICMP_DEST_UNREACH; /* to Type 3 */
+                                               icmp_hdr->code = ICMP_PROT_UNREACH; /* to Code 2 */
+                                               break;
+                                       default: /* if Code != 1 */
+                                               icmp_hdr->type = ICMP_PARAMETERPROB; /* to Type 12 */
+                                               icmp_hdr->code = 0; /* to Code 0 */
+                                               /* Update Pointer field
+                                                  RFC 2765 (SIIT), 4.2:
+                                                  The Pointer needs to be updated to point to the corresponding
+                                                  field in the translated include IP header.
+                                                  */
+                                               switch (ntohl(icmp6_hdr->icmp6_pointer))
+                                               {
+                                               case 0: /* IPv6 Version -> IPv4 Version */
+                                                       icmp_hdr->un.echo.id = 0;
+                                                       break;
+                                               case 4: /* IPv6 PayloadLength -> IPv4 Total Length */
+                                                       icmp_hdr->un.echo.id = 0x0002; /* 2 */
+                                                       break;
+                                               case 6: /* IPv6 Next Header-> IPv4 Protocol */
+                                                       icmp_hdr->un.echo.id = 0x0009; /* 9 */
+                                                       break;
+                                               case 7: /* IPv6 Hop Limit -> IPv4 TTL */
+                                                       icmp_hdr->un.echo.id = 0x0008; /* 8 */
+                                                       break;
+                                               case 8: /* IPv6 Src addr -> IPv4 Src addr */
+                                                       icmp_hdr->un.echo.id = 0x000c; /* 12 */
+                                                       break;
+                                               case 24: /* IPv6 Dst addr -> IPv4 Dst addr*/
+                                                       icmp_hdr->un.echo.id = 0x0010; /* 16 */
+                                                       break;
+                                               default: /* set all ones in other cases */
+                                                       icmp_hdr->un.echo.id = 0xff;
+                                                       break;
+                                               }
+                                               break;
+                                       }
+                                       break;
+
+/* End of ICMP Error messages */
+
+                       /* Echo Request and Echo Reply (Type 128 and 129)  */
+                               case ICMPV6_ECHO_REQUEST:
+                                       icmperr = 0;        /* not error ICMP message */
+                                       icmp_hdr->type = ICMP_ECHO; /* to Type 8 */
+                                       icmp_hdr->code = 0; /* to Code 0 */
+                                       icmp_hdr->un.echo.id = icmp6_hdr->icmp6_identifier;
+                                       icmp_hdr->un.echo.sequence = icmp6_hdr->icmp6_sequence;
+                                       /* copy rest of ICMP data to result packet */
+                                       if (len_delta > 0)
+                                               memcpy(((char *)icmp_hdr) + sizeof(struct icmphdr),
+                                                          ((char *)icmp6_hdr) + sizeof(struct icmp6hdr), len_delta);
+                                       break;
+                               case ICMPV6_ECHO_REPLY:
+                                       icmperr = 0;        /* not error ICMP message */
+                                       icmp_hdr->type = ICMP_ECHOREPLY; /* to Type 0 */
+                                       icmp_hdr->code = 0; /* to Code 0 */
+                                       icmp_hdr->un.echo.id = icmp6_hdr->icmp6_identifier;
+                                       icmp_hdr->un.echo.sequence = icmp6_hdr->icmp6_sequence;
+                                       /* copy rest of ICMP data */
+                                       if (len_delta > 0)
+                                               memcpy(((char *)icmp_hdr) + sizeof(struct icmphdr),
+                                                          ((char *)icmp6_hdr) + sizeof(struct icmp6hdr), len_delta);
+                                       break;
+                               default:
+                                       /* Unknown error messages. Silently drop. */
+                                       PDEBUG("ip6_ip4(): unknown ICMPv6 Type %d, packet dropped.\n", icmp6_hdr->icmp6_type);
+                                       return -1;
+                               }
+
+                               if (icmperr)
+                               {
+                                       /* If ICMP Error message, we translate IP included packet*/
+                                       if (len_delta >= sizeof(struct ipv6hdr))
+                                       {
+                                               if((inc_opts_len = ip6_ip4((char *)icmp6_hdr + sizeof(struct icmp6hdr), len_delta,
+                                                                                          (char *)icmp_hdr + sizeof(struct icmphdr), 1)) == -1) {
+                                                       PDEBUG("ip6_ip4(): incorrect translation of ICMPv6 Error message, packet dropped\n");
+                                                       return -1;
+                                               }
+                                               /* correct IPv4 Total Len that = old Total Len
+                                                  - Options Headers Len in included IP packet
+                                                  - diffirence between IPv6 Header Len and IPv4 Header Len
+                                                  */
+                                               if (ntot_len != 0)
+                                                       ip_hdr->tot_len = htons(ntot_len - inc_opts_len - IP4_IP6_HDR_DIFF);
+                                               real_len = real_len - inc_opts_len - IP4_IP6_HDR_DIFF;
+                                       }
+                                       else if (len_delta > 0)
+                                       {
+                                               /* May be it need set 0x0 to rest area in result IPv4 packet,
+                                                * but we copy rest data unchanged
+                                                */
+                                               memcpy(((char *)icmp_hdr) + sizeof(struct icmphdr),
+                                                          ((char *)icmp6_hdr) + sizeof(struct icmp6hdr), len_delta);
+                                       }
+                               }
+
+                               /* Calculate IPv4 Header checksum */
+                               ip_hdr->check = 0;
+                               ip_hdr->check = ip_fast_csum((unsigned char *)ip_hdr, ip_hdr->ihl);
+
+                               /* Calculate ICMPv4 checksum */
+                               if (ntot_len != 0)
+                               {
+                                       icmp_hdr->checksum = 0;
+                                       icmp_hdr->checksum = ip_compute_csum((unsigned char *)icmp_hdr, ntohs(ip_hdr->tot_len)
+                                                                                            - sizeof(struct iphdr));
+                               }
+                       }
+                       else
+                       {
+                               PDEBUG("ip6_ip4(): error length ICMP packet, packet dropped.\n");
+                               return -1;
+                       }
+
+               }
+               /* PROCESS TCP and UDP (and rest data) */
+
+               else {
+                       real_len += len_delta;
+                       /* we copy rest data to IPv4 packet without changing */
+                       memcpy(dst+sizeof(struct iphdr), src + sizeof(struct ipv6hdr) + opts_len, len_delta);
+               }
+       }
+
+       if (include_flag)           /* if it's included IP packet */
+               return opts_len;        /* return options headers length */
+       else
+               return real_len; /* result packet len */
+}
+
+/*
+ * ip4_fragment(skb, len, hdr_len, dev, eth_h)
+ * to fragment original IPv4 packet if result IPv6 packet will be > 1280
+ */
+
+static int ip4_fragment(struct sk_buff *skb, int len, int hdr_len, struct net_device *dev, struct ethhdr *eth_h)
+{
+       struct sk_buff *skb2 = NULL;       /* pointer to new struct sk_buff for transleded packet */
+       char buff[FRAG_BUFF_SIZE+hdr_len]; /* buffer to form new fragment packet */
+       char *cur_ptr = skb->data+hdr_len; /* pointter to current packet data with len = frag_len */
+       struct iphdr *ih4 = (struct iphdr *) skb->data;
+       struct iphdr *new_ih4 = (struct iphdr *) buff; /* point to new IPv4 hdr */
+       struct ethhdr *new_eth_h;   /* point to ether hdr, need to set hard header data in fragment */
+       int data_len = len - hdr_len; /* origin packet data len */
+       int rest_len = data_len;    /* rest data to fragment */
+       int frag_len = 0;           /* current fragment len */
+       int last_frag = 0;          /* last fragment flag, if = 1, it's last fragment */
+       int flag_last_mf = 0;
+       __u16 new_id = 0;           /* to generate identification field */
+       __u16 frag_offset = 0;      /* fragment offset */
+       unsigned int csum;
+       unsigned short udp_len;
+
+#ifdef SIIT_DEBUG
+       printk("siit: it's DF == 0 and result IPv6 packet will be > 1280\n");
+       siit_print_dump(skb->data, hdr_len, "siit: (orig) ipv4_hdr dump");
+#endif
+
+       if ((ntohs(ih4->frag_off) & IP_MF) == 0 )
+               /* it's a case we'll clear MF flag in our last packet */
+               flag_last_mf = 1;
+
+       if (ih4->protocol == IPPROTO_UDP) {
+               if ( (ntohs(ih4->frag_off) & IP_OFFSET) == 0) {
+                       struct udphdr *udp_hdr = (struct udphdr *)((char *)ih4 + hdr_len);
+                       if (!flag_last_mf) {
+                               if (udp_hdr->check == 0) {
+                                       /* it's a first fragment with ZERO checksum and we drop packet */
+                                       printk("siit: First fragment of UDP with zero checksum - packet droped\n");
+                                       printk("siit: addr: %x src port: %d dst port: %d\n",
+                                                  htonl(ih4->saddr), htons(udp_hdr->source), htons(udp_hdr->dest));
+                                       return -1;
+                               }
+                       }
+                       else if (udp_hdr->check == 0) {
+                               /* Calculate UDP checksum only if it's not fragment */
+                               udp_len = ntohs(udp_hdr->len);
+                               csum = 0;
+                               csum = csum_partial((unsigned char *)udp_hdr, udp_len, csum);
+                               udp_hdr->check = csum_tcpudp_magic(ih4->saddr, ih4->daddr, udp_len, IPPROTO_UDP, csum);
+                       }
+               }
+       }
+
+       frag_offset = ntohs(ih4->frag_off) & IP_OFFSET;
+
+       new_id = ih4->id;
+
+       while(1) {
+               if (rest_len <= FRAG_BUFF_SIZE) {
+                       /* it's last fragmen */
+                       frag_len = rest_len; /* rest data */
+                       last_frag = 1;
+               }
+               else
+                       frag_len = FRAG_BUFF_SIZE;
+
+               /* copy IP header to buffer */
+               memcpy(buff, skb->data, hdr_len);
+               /* copy data to buffer with len = frag_len */
+               memcpy(buff + hdr_len, cur_ptr, frag_len);
+
+               /* set id field in new IPv4 header*/
+               new_ih4->id = new_id;
+
+               /* is it last fragmet */
+               if(last_frag && flag_last_mf)
+                       /* clear MF flag */
+                       new_ih4->frag_off = htons(frag_offset & (~IP_MF));
+               else
+                       /* set MF flag */
+                       new_ih4->frag_off = htons(frag_offset | IP_MF);
+
+               /* change packet total length */
+               new_ih4->tot_len = htons(frag_len+hdr_len);
+
+               /* rebuild the header checksum (IP needs it) */
+               new_ih4->check = 0;
+               new_ih4->check = ip_fast_csum((unsigned char *)new_ih4,new_ih4->ihl);
+
+               /* Allocate new sk_buff to compose translated packet */
+               skb2 = dev_alloc_skb(frag_len+hdr_len+dev->hard_header_len+IP4_IP6_HDR_DIFF+IP6_FRAGMENT_SIZE);
+               if (!skb2) {
+                       printk(KERN_DEBUG "%s: alloc_skb failure - packet dropped.\n", dev->name);
+                       dev_kfree_skb(skb2);
+                       return -1;
+               }
+               /* allocate skb->data portion for IP header len, fragment data len and ether header len
+                * and copy to head ether header from origin skb
+                */
+               memcpy(skb_put(skb2, frag_len+hdr_len+dev->hard_header_len+IP4_IP6_HDR_DIFF+IP6_FRAGMENT_SIZE), (char *) eth_h,
+                          dev->hard_header_len);
+               /* correct ether header data, ether protocol field to ETH_P_IPV6 */
+               new_eth_h = (struct ethhdr *)skb2->data;
+               new_eth_h->h_proto = htons(ETH_P_IPV6);
+
+               /* reset the mac header */
+               skb_reset_mac_header(skb2);
+
+               /* pull ether header from new skb->data */
+               skb_pull(skb2, dev->hard_header_len);
+               /* set skb protocol to IPV6 */
+               skb2->protocol = htons(ETH_P_IPV6);
+
+               /* call translation function */
+               if ( ip4_ip6(buff, frag_len+hdr_len, skb2->data, 0) == -1) {
+                       dev_kfree_skb(skb2);
+                       return -1;
+               }
+
+               /*
+                * Set needed fields in new sk_buff
+                */
+               skb2->dev = dev;
+               skb2->ip_summed = CHECKSUM_UNNECESSARY;
+               skb2->pkt_type = PACKET_HOST;
+
+               /* Add transmit statistic */
+               siit_stats(dev)->tx_packets++;
+               siit_stats(dev)->tx_bytes += skb2->len;
+
+               /* send packet to upper layer */
+               netif_rx(skb2);
+
+               /* exit if it was last fragment */
+               if (last_frag)
+                       break;
+
+               /* correct current data pointer */
+               cur_ptr += frag_len;
+               /* rest data len */
+               rest_len -= frag_len;
+               /* current fragment offset */
+               frag_offset = (frag_offset*8 + frag_len)/8;
+       }
+
+       return 0;
+}
+/*
+ * Transmit a packet (called by the kernel)
+ *
+ * siit_xmit(skb, dev)
+ *
+ * where
+ * skb - pointer to struct sk_buff with incomed packet
+ * dev - pointer to struct device on which packet revieved
+ *
+ * Statistic:
+ * for all incoming packes:
+ *            stats.rx_bytes+=skb->len
+ *            stats.rx_packets++
+ * for packets we can't transle:
+ *            stats.tx_errors++
+ * device busy:
+ *            stats.tx_errors++
+ * for packets we can't allocate sk_buff:
+ *            stats.tx_dropped++
+ * for outgoing packes:
+ *            stats.tx_packets++
+ *            stats.tx_bytes+=skb2->len !!! But we don't set skb2->len !!!
+ */
+
+static int siit_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+       struct sk_buff *skb2 = NULL;/* pointer to new struct sk_buff for transleded packet */
+       struct ethhdr *eth_h;       /* pointer to incoming Ether header */
+       int len;                    /* original packets length */
+       int new_packet_len;
+       int skb_delta = 0;          /* delta size for allocate new skb */
+       char new_packet_buff[2048];
+
+       /* Check pointer to sk_buff and device structs */
+       if (skb == NULL || dev == NULL)
+               return -EINVAL;
+
+       /* Add receive statistic */
+       siit_stats(dev)->rx_bytes += skb->len;
+       siit_stats(dev)->rx_packets++;
+
+       dev->trans_start = jiffies;
+
+       /* Upper layer (IP) protocol forms sk_buff for outgoing packet
+        * and sets IP header + Ether header too. IP layer sets outgoing
+        * device in sk_buff->dev.
+        * In function (from linux/net/core/dev.c) ther is a call to
+        * device transmit function (dev->hard_start_xmit):
+        *
+        *    dev_queue_xmit(struct sk_buff *skb)
+        *    {
+        *    ...
+        *          device *dev = skb->dev;
+        *    ...
+        *          dev->hard_start_xmit(skb, dev);
+        *    ...
+        *    }
+        * We save pointer to ether header in eth_h and skb_pull ether header
+        * from data field of skb_buff
+        */
+
+       eth_h = (struct ethhdr *)skb->data; /* point to incoming packet Ether Header */
+
+#ifdef SIIT_DEBUG
+       siit_print_dump(skb->data, ETH_HLEN, "siit: eth_hdr dump");
+#endif
+
+       /* Remove hardware header from origin sk_buff */
+       skb_pull(skb,dev->hard_header_len);
+
+       /*
+        * Process IPv4 paket
+        */
+       if (ntohs(skb->protocol) == ETH_P_IP) {
+               int hdr_len;            /* IPv4 header length */
+               int data_len;           /* IPv4 data length */
+               struct iphdr *ih4;      /* pointer to IPv4 header */
+               struct icmphdr *icmp_hdr;   /* point to current ICMPv4 header struct */
+
+               ih4 = (struct iphdr *)skb->data; /* point to incoming packet's IPv4 header */
+
+               /* Check IPv4 Total Length */
+               if (skb->len != ntohs(ih4->tot_len)) {
+                       PDEBUG("siit_xmit(): Different skb_len %x and ip4 tot_len %x - packet dropped.\n",
+                                  skb->len, ih4->tot_len);
+                       siit_stats(dev)->tx_errors++;
+                       dev_kfree_skb(skb);
+                       return 0;
+               }
+
+               len = skb->len;     /* packet's total len */
+               hdr_len = (int)(ih4->ihl * 4); /* packet's header len */
+               data_len = len - hdr_len; /* packet's data len */
+
+               /* If DF == 0 */
+               if ( (ntohs(ih4->frag_off) & IP_DF) == 0 ) {
+                       /* If result IPv6 packet will be > 1280
+                          we need to fragment original IPv4 packet
+                       */
+                       if ( data_len > FRAG_BUFF_SIZE ) {
+                               /* call function that fragment packet and translate to IPv6 each fragment
+                                * and send to upper layer
+                                */
+                               if ( ip4_fragment(skb, len, hdr_len, dev, eth_h) == -1) {
+                                       siit_stats(dev)->tx_errors++;
+                               }
+                               /* Free incoming skb */
+                               dev_kfree_skb(skb);
+                               /* Device can accept a new packet */
+
+                               return 0;
+
+                       }
+               }
+               /* If DF == 1 && MF == 0 && Fragment Offset == 0
+                * we don't include fragment header
+                */
+               if ( ntohs(ih4->frag_off) == IP_DF )
+                       skb_delta = IP4_IP6_HDR_DIFF; /* delta is +20 */
+               else
+                       skb_delta = IP4_IP6_HDR_DIFF + IP6_FRAGMENT_SIZE; /* delta is +20 and +8 */
+
+               /* If it's ICMP, check is it included IP packet in it */
+               if ( ih4->protocol == IPPROTO_ICMP) {
+                       icmp_hdr = (struct icmphdr *) (skb->data+hdr_len); /* point to ICMPv4 header */
+                       if ( icmp_hdr->type != ICMP_ECHO && icmp_hdr->type != ICMP_ECHOREPLY) {
+                               /*
+                                * It's ICMP Error that has included IP packet
+                                * we'll add only +20 because we don't include Fragment Header
+                                * into translated included IP packet
+                                */
+                               skb_delta += IP4_IP6_HDR_DIFF;
+                       }
+               }
+
+               /* Allocate new sk_buff to compose translated packet */
+               skb2 = dev_alloc_skb(len+dev->hard_header_len+skb_delta);
+               if (!skb2) {
+                       printk(KERN_DEBUG "%s: alloc_skb failure - packet dropped.\n", dev->name);
+                       dev_kfree_skb(skb);
+                       siit_stats(dev)->rx_dropped++;
+
+                       return 0;
+               }
+               /* allocate skb->data portion = IPv4 packet len + ether header len
+                * + skb_delta (max = two times (diffirence between IPv4 header and
+                * IPv6 header + Frag Header), second for included packet,
+                * and copy to head of skb->data ether header from origin skb
+                */
+               memcpy(skb_put(skb2, len+dev->hard_header_len+skb_delta), (char *)eth_h, dev->hard_header_len);
+               /* correct ether header data, ether protocol field to ETH_P_IPV6 */
+               eth_h = (struct ethhdr *)skb2->data;
+               eth_h->h_proto = htons(ETH_P_IPV6);
+               skb_reset_mac_header(skb2);
+               /* remove ether header from new skb->data,
+                * NOTE! data will rest, pointer to data and data len will change
+                */
+               skb_pull(skb2,dev->hard_header_len);
+               /* set skb protocol to IPV6 */
+               skb2->protocol = htons(ETH_P_IPV6);
+
+               /* call translation function */
+               if (ip4_ip6(skb->data, len, skb2->data, 0) == -1 ) {
+                       dev_kfree_skb(skb);
+                       dev_kfree_skb(skb2);
+                       siit_stats(dev)->rx_errors++;
+
+                       return 0;
+               }
+       }
+       /*
+        * IPv6 paket
+        */
+       else if (ntohs(skb->protocol) == ETH_P_IPV6) {
+
+#ifdef SIIT_DEBUG
+               siit_print_dump(skb->data, sizeof(struct ipv6hdr), "siit: (in) ip6_hdr dump");
+#endif
+               /* packet len = skb->data len*/
+               len = skb->len;
+
+               /* call translation function */
+               if ((new_packet_len = ip6_ip4(skb->data, len, new_packet_buff, 0)) == -1 )
+               {
+                       PDEBUG("siit_xmit(): error translation ipv6->ipv4, packet dropped.\n");
+                       siit_stats(dev)->rx_dropped++;
+                       goto end;
+               }
+
+               /* Allocate new sk_buff to compose translated packet */
+               skb2 = dev_alloc_skb(new_packet_len + dev->hard_header_len);
+               if (!skb2) {
+                       printk(KERN_DEBUG "%s: alloc_skb failure, packet dropped.\n", dev->name);
+                       siit_stats(dev)->rx_dropped++;
+                       goto end;
+               }
+               memcpy(skb_put(skb2, new_packet_len + dev->hard_header_len), (char *)eth_h, dev->hard_header_len);
+               eth_h = (struct ethhdr *)skb2->data;
+               eth_h->h_proto = htons(ETH_P_IP);
+               skb_reset_mac_header(skb2);
+               skb_pull(skb2, dev->hard_header_len);
+               memcpy(skb2->data, new_packet_buff, new_packet_len);
+               skb2->protocol = htons(ETH_P_IP);
+       }
+       else {
+               PDEBUG("siit_xmit(): unsupported protocol family %x, packet dropped.\n", skb->protocol);
+               goto end;
+       }
+
+       /*
+        * Set needed fields in new sk_buff
+        */
+       skb2->pkt_type = PACKET_HOST;
+       skb2->dev = dev;
+       skb2->ip_summed = CHECKSUM_UNNECESSARY;
+
+       /* Add transmit statistic */
+       siit_stats(dev)->tx_packets++;
+       siit_stats(dev)->tx_bytes += skb2->len;
+
+       /* Send packet to upper layer protocol */
+       netif_rx(skb2);
+
+end:
+       dev_kfree_skb(skb);
+
+       return 0;
+}
+
+#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,0)
+static bool header_ops_init = false;
+static struct header_ops siit_header_ops ____cacheline_aligned;
+#endif
+
+/*
+ * The init function initialize of the SIIT device..
+ * It is invoked by register_netdev()
+ */
+static void
+siit_init(struct net_device *dev)
+{
+       ether_setup(dev);    /* assign some of the fields */
+       random_ether_addr(dev->dev_addr);
+
+       /*
+        * Assign device function.
+        */
+       dev->open            = siit_open;
+       dev->stop            = siit_release;
+       dev->hard_start_xmit = siit_xmit;
+       dev->flags           |= IFF_NOARP;     /* ARP not used */
+       dev->tx_queue_len = 10;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+       dev->hard_header_cache = NULL;        /* Disable caching */
+       memset(netdev_priv(dev), 0, sizeof(struct net_device_stats));
+       dev->get_stats = siit_get_stats;
+#else
+       if (!header_ops_init) {
+               memcpy(&siit_header_ops, dev->header_ops, sizeof(struct header_ops));
+               siit_header_ops.cache = NULL;
+       }
+       dev->header_ops = &siit_header_ops;
+#endif
+}
+
+/*
+ * Finally, the module stuff
+ */
+static struct net_device *siit_dev = NULL;
+
+int init_module(void)
+{
+       int res = -ENOMEM;
+       int priv_size;
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)
+       priv_size = sizeof(struct net_device_stats);
+#else
+       priv_size = sizeof(struct header_ops);
+#endif
+       siit_dev = alloc_netdev(priv_size, "siit%d", siit_init);
+       if (!siit_dev)
+               goto err_alloc;
+
+       res = register_netdev(siit_dev);
+       if (res)
+               goto err_register;
+
+       return 0;
+
+err_register:
+       free_netdev(siit_dev);
+err_alloc:
+       printk(KERN_ERR "Error creating siit device: %d\n", res);
+       return res;
+}
+
+void cleanup_module(void)
+{
+       unregister_netdev(siit_dev);
+       free_netdev(siit_dev);
+}
+
+
diff --git a/package/siit/src/siit.h b/package/siit/src/siit.h
new file mode 100644 (file)
index 0000000..47cf776
--- /dev/null
@@ -0,0 +1,61 @@
+/*
+ * siit.h -- definitions for the SIIT module
+ *
+ *
+ */
+
+/*
+ * Constants
+ */
+
+/* SIIT_ETH control the name of SIIT interface:
+ * 0 - interface name is siit0,
+ * 1 - interface name is ethX.
+ */
+#define SIIT_ETH 0
+
+#define BUFF_SIZE 4096
+#define FRAG_BUFF_SIZE 1232     /* IPv6 max fragment size without IPv6 header 
+                                 * to fragmanet IPv4 if result IPv6 packet will be > 1280
+                                 */
+
+#define TRANSLATED_PREFIX 0x0000ffff /* third byte in IPv4-translated addr prefix */
+#define MAPPED_PREFIX 0x0000ffff     /* third byte in IPv4-mapped addr prefix */
+
+#define IP4_IP6_HDR_DIFF 20     /* diffirence between IPv4 and IPv6 headers */
+#define IP6_FRAGMENT_SIZE 8     /* size of Fragment Header */
+
+/* IPv6 header fields masks */
+#define IP6F_OFF_MASK       0xfff8  /* mask out offset from frag_off */
+#define IP6F_RESERVED_MASK  0x0006  /* reserved bits in frag_off */
+#define IP6F_MORE_FRAG      0x0001  /* more-fragments flag */
+
+
+
+/*
+ * Macros to help debugging
+ */
+
+#undef PDEBUG             /* undef it, just in case */
+#ifdef SIIT_DEBUG
+#  ifdef __KERNEL__
+     /* This one if debugging is on, and kernel space */
+#    define PDEBUG(fmt, args...) printk(KERN_DEBUG "siit: " fmt, ## args)
+#  else
+     /* This one for user space */
+#    define PDEBUG(fmt, args...) fprintf(stderr, fmt, ## args)
+#  endif
+#else
+#  define PDEBUG(fmt, args...) /* not debugging: nothing */
+#endif
+
+#undef PDEBUGG
+#define PDEBUGG(fmt, args...)
+
+
+
+
+
+
+
+
This page took 0.068849 seconds and 4 git commands to generate.