linux-kernel.diff

基于linux内核2.6.9的mpls补丁

+ */
+
+void spec_nh_remove(struct spec_nh *spec)
+{
+	struct list_head *head;
+	struct spec_nh *spec1;
+
+	spin_lock_bh(&spec_nh_lock);
+	head = &spec_nh_base[ntohs(spec->type) & 15];
+
+	list_for_each_entry(spec1, head, list) {
+		if (spec == spec1) {
+			list_del_rcu(&spec->list);
+			goto out;
+		}
+	}
+	printk(KERN_WARNING "spec_nh_remove: %p not found.\n", spec);
+out:
+	spin_unlock_bh(&spec_nh_lock);
+
+	synchronize_net();
+
+}
+
+/**
+ *	spec_nh_find - find a special nexthop handler by it's protocol type
+ *	@proto: protocol type declaration
+ *
+ *	Search the kernels list of special nexthops handlers looking for
+ *	a handler for this specific protocol.
+ */
+struct spec_nh *spec_nh_find(unsigned short proto)
+{
+	struct list_head *head;
+	struct spec_nh *spec;
+
+	spin_lock_bh(&spec_nh_lock);
+	head = &spec_nh_base[ntohs(proto) & 15];
+
+	list_for_each_entry(spec, head, list) {
+		if (proto == spec->type) {
+			goto out;
+		}
+	}
+	spec = NULL;
+out:
+	spin_unlock_bh(&spec_nh_lock);
+
+	return spec;
+}
+
+/*
+ * Proc filesystem directory entries.
+ */
+										
+/*
+ *      /proc/net/spec_nh
+ */
+										
+static struct proc_dir_entry *proc_spec_nh_dir;
+										
+/*
+ *      /proc/net/spec_nh/config
+ */
+										
+static struct proc_dir_entry *proc_spec_nh_conf;
+
+/*
+ *      Names of the proc directory entries
+ */
+										
+static const char name_root[]    = "spec_nh";
+static const char name_conf[]    = "config";
+
+/*
+ * The following few functions build the content of /proc/net/spec_nh/config
+ */
+
+/* starting at spec, find the next registered protocol */
+struct spec_nh *spec_nh_skip(struct spec_nh *spec)
+{
+	struct list_head *head;
+	struct spec_nh *spec1;
+	int next = 0;
+	int slot = 0;
+
+	if (spec)
+		slot = ntohs(spec->type) & 15;
+	else
+		next = 1;
+		
+	for (;slot < 16;slot++) {
+		head = &spec_nh_base[slot];
+		list_for_each_entry(spec1, head, list) {
+			if (next)
+				return spec1;
+
+			if (spec1 == spec)
+				next = 1;
+		}
+	}
+
+	return NULL;
+}
+
+										
+/* start read of /proc/net/spec_nh/config */
+static void *spec_nh_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct spec_nh *spec;
+	loff_t i = 1;
+
+	spin_lock_bh(&spec_nh_lock);
+
+	if (*pos == 0)
+		return SEQ_START_TOKEN;
+
+	for (spec = spec_nh_skip(NULL); spec && i < *pos;
+		spec = spec_nh_skip(spec), ++i);
+										
+	return (i == *pos) ? spec : NULL;
+}
+
+static void *spec_nh_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	++*pos;
+										
+	return spec_nh_skip((v == SEQ_START_TOKEN)
+			    ? NULL
+			    : (struct spec_nh *)v);
+}
+										
+static void spec_nh_seq_stop(struct seq_file *seq, void *v)
+{
+	spin_unlock_bh(&spec_nh_lock);
+}
+
+static int spec_nh_seq_show(struct seq_file *seq, void *v)
+{
+	struct spec_nh* spec = (struct spec_nh*)v;
+	if (v != SEQ_START_TOKEN)
+		seq_printf(seq, "0x%04x\t%s\n",
+		    ntohs(spec->type), spec->name ? spec->name : "(none)");
+	return 0;
+}
+
+/*
+ *      Generic /proc/net/spec_nh/<file> file and inode operations
+ */
+										
+static struct seq_operations spec_nh_seq_ops = {
+	.start = spec_nh_seq_start,
+	.next = spec_nh_seq_next,
+	.stop = spec_nh_seq_stop,
+	.show = spec_nh_seq_show,
+};
+										
+static int spec_nh_seq_open(struct inode *inode, struct file *file)
+{
+	return seq_open(file, &spec_nh_seq_ops);
+}
+										
+static struct file_operations spec_nh_fops = {
+	.owner   = THIS_MODULE,
+	.open    = spec_nh_seq_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release,
+};
+
+/*
+ *      Clean up /proc/net/spec_nh entries
+ */
+										
+void spec_nh_proc_cleanup(void)
+{
+	if (proc_spec_nh_conf)
+		remove_proc_entry(name_conf, proc_spec_nh_dir);
+										
+	if (proc_spec_nh_dir)
+		proc_net_remove(name_root);
+}
+
+int __init spec_nh_proc_init(void)
+{
+	proc_spec_nh_dir = proc_mkdir(name_root, proc_net);
+	if (proc_spec_nh_dir) {
+		proc_spec_nh_conf = create_proc_entry(name_conf,
+						   S_IFREG|S_IRUSR|S_IWUSR,
+						   proc_spec_nh_dir);
+		if (proc_spec_nh_conf) {
+			proc_spec_nh_conf->proc_fops = &spec_nh_fops;
+			return 0;
+		}
+	}
+	spec_nh_proc_cleanup();
+	return -ENOBUFS;
+}
+
+void __init spec_nh_init(void)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < 16; i++)
+		INIT_LIST_HEAD(&spec_nh_base[i]);
+
+	/* proc file system initialization */
+	err = spec_nh_proc_init();
+	if (err < 0) {
+		printk(KERN_ERR
+		       "%s: can't create entry in proc filesystem!\n",
+		       __FUNCTION__);
+	}
+}
+
+static void __exit spec_nh_cleanup(void)
+{
+	spec_nh_proc_cleanup();
+	synchronize_net();
+}
+
+EXPORT_SYMBOL(spec_nh_add);
+EXPORT_SYMBOL(spec_nh_remove);
+EXPORT_SYMBOL(spec_nh_find);
diff -uNr linux-kernel/net/ipv4/fib_semantics.c mpls-kernel-1/net/ipv4/fib_semantics.c
--- linux-kernel/net/ipv4/fib_semantics.c	2004-10-22 13:40:45.000000000 -0500
+++ mpls-kernel-1/net/ipv4/fib_semantics.c	2005-01-05 01:10:11.000000000 -0600
@@ -185,6 +185,8 @@
 #ifdef CONFIG_NET_CLS_ROUTE
 		    nh->nh_tclassid != onh->nh_tclassid ||
 #endif
+		    nh->nh_spec_proto != onh->nh_spec_proto ||
+		    nh->nh_spec_data != onh->nh_spec_data ||
 		    ((nh->nh_flags^onh->nh_flags)&~RTNH_F_DEAD))
 			return -1;
 		onh++;
@@ -309,6 +311,8 @@
 		nh->nh_flags = (r->rtm_flags&~0xFF) | nhp->rtnh_flags;
 		nh->nh_oif = nhp->rtnh_ifindex;
 		nh->nh_weight = nhp->rtnh_hops + 1;
+		nh->nh_spec_proto = nhp->rtnh_spec_proto;
+		nh->nh_spec_data = nhp->rtnh_spec_data;
 		if (attrlen) {
 			nh->nh_gw = fib_get_attr32(RTNH_DATA(nhp), attrlen, RTA_GATEWAY);
 #ifdef CONFIG_NET_CLS_ROUTE
@@ -336,6 +340,8 @@
 
 	if (rta->rta_oif || rta->rta_gw) {
 		if ((!rta->rta_oif || *rta->rta_oif == fi->fib_nh->nh_oif) &&
+		    (!rta->rta_spec_proto || *rta->rta_spec_proto == fi->fib_nh->nh_spec_proto) &&
+		    (!rta->rta_spec_data || memcmp(rta->rta_spec_data, &fi->fib_nh->nh_spec_data, 4) == 0) &&
 		    (!rta->rta_gw  || memcmp(rta->rta_gw, &fi->fib_nh->nh_gw, 4) == 0))
 			return 0;
 		return 1;
@@ -355,6 +361,12 @@
 			return -EINVAL;
 		if (nhp->rtnh_ifindex && nhp->rtnh_ifindex != nh->nh_oif)
 			return 1;
+		if (nhp->rtnh_spec_data &&
+		    nhp->rtnh_spec_data != nh->nh_spec_data)
+			return 1;
+		if (nhp->rtnh_spec_proto &&
+		    nhp->rtnh_spec_proto != nh->nh_spec_proto)
+			return 1;
 		if (attrlen) {
 			gw = fib_get_attr32(RTNH_DATA(nhp), attrlen, RTA_GATEWAY);
 			if (gw && gw != nh->nh_gw)
@@ -659,6 +671,10 @@
 			goto err_inval;
 		if (rta->rta_gw && memcmp(&fi->fib_nh->nh_gw, rta->rta_gw, 4))
 			goto err_inval;
+		if (rta->rta_spec_proto && 
+		    (fi->fib_nh->nh_spec_proto != *rta->rta_spec_proto ||
+		    fi->fib_nh->nh_spec_data != *rta->rta_spec_data))
+			goto err_inval;
 #ifdef CONFIG_NET_CLS_ROUTE
 		if (rta->rta_flow && memcmp(&fi->fib_nh->nh_tclassid, rta->rta_flow, 4))
 			goto err_inval;
@@ -680,6 +696,13 @@
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
 		nh->nh_weight = 1;
 #endif
+		if (rta->rta_spec_proto) {
+			nh->nh_spec_proto = *rta->rta_spec_proto;
+			if (rta->rta_spec_data)
+				nh->nh_spec_data = *rta->rta_spec_data;
+			else
+				nh->nh_spec_data = 0;
+		}
 	}
 
 	if (fib_props[r->rtm_type].error) {
@@ -873,6 +896,12 @@
 			RTA_PUT(skb, RTA_GATEWAY, 4, &fi->fib_nh->nh_gw);
 		if (fi->fib_nh->nh_oif)
 			RTA_PUT(skb, RTA_OIF, sizeof(int), &fi->fib_nh->nh_oif);
+		if (fi->fib_nh->nh_spec_proto) {
+			RTA_PUT(skb, RTA_SPEC_PROTO, sizeof(unsigned short),
+				&fi->fib_nh->nh_spec_proto);
+			RTA_PUT(skb, RTA_SPEC_DATA, sizeof(u32),
+				&fi->fib_nh->nh_spec_data);
+		}
 	}
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
 	if (fi->fib_nhs > 1) {
@@ -889,6 +918,8 @@
 			nhp->rtnh_flags = nh->nh_flags & 0xFF;
 			nhp->rtnh_hops = nh->nh_weight-1;
 			nhp->rtnh_ifindex = nh->nh_oif;
+			nhp->rtnh_spec_proto = nh->nh_spec_proto;
+			nhp->rtnh_spec_data = nh->nh_spec_data;
 			if (nh->nh_gw)
 				RTA_PUT(skb, RTA_GATEWAY, 4, &nh->nh_gw);
 			nhp->rtnh_len = skb->tail - (unsigned char*)nhp;
diff -uNr linux-kernel/net/ipv4/ip_output.c mpls-kernel-1/net/ipv4/ip_output.c
--- linux-kernel/net/ipv4/ip_output.c	2004-10-22 13:40:45.000000000 -0500
+++ mpls-kernel-1/net/ipv4/ip_output.c	2005-01-05 01:10:11.000000000 -0600
@@ -176,6 +176,11 @@
 	struct net_device *dev = dst->dev;
 	int hh_len = LL_RESERVED_SPACE(dev);
 
+	if (dst->child) {
+		skb->dst = dst_pop(skb->dst);
+		return dst_output(skb);
+	}
+
 	/* Be paranoid, rather than too clever. */
 	if (unlikely(skb_headroom(skb) < hh_len && dev->hard_header)) {
 		struct sk_buff *skb2;
diff -uNr linux-kernel/net/ipv4/Kconfig mpls-kernel-1/net/ipv4/Kconfig
--- linux-kernel/net/ipv4/Kconfig	2004-10-22 13:40:45.000000000 -0500
+++ mpls-kernel-1/net/ipv4/Kconfig	2005-01-05 01:10:09.000000000 -0600
@@ -158,6 +158,15 @@
 	  operating on your network. Read <file:Documentation/nfsroot.txt> for
 	  details.
 
+config IP_MPLS
+	tristate "IP: MPLS support"
+	depends on INET && MPLS
+	---help---
+	  If you say Y here, the kernel will support being an ingress and
+	  egress LER for IPv4 packets
+
+	  If unsure, say N.
+
 # not yet ready..
 #   bool '    IP: ARP support' CONFIG_IP_PNP_ARP		
 config NET_IPIP
diff -uNr linux-kernel/net/ipv4/Makefile mpls-kernel-1/net/ipv4/Makefile
--- linux-kernel/net/ipv4/Makefile	2004-10-22 13:40:45.000000000 -0500
+++ mpls-kernel-1/net/ipv4/Makefile	2005-01-05 01:10:09.000000000 -0600
@@ -20,6 +20,7 @@
 obj-$(CONFIG_INET_IPCOMP) += ipcomp.o
 obj-$(CONFIG_INET_TUNNEL) += xfrm4_tunnel.o 
 obj-$(CONFIG_IP_PNP) += ipconfig.o
+obj-$(CONFIG_IP_MPLS) += mpls4.o
 obj-$(CONFIG_NETFILTER)	+= netfilter/
 obj-$(CONFIG_IP_VS) += ipvs/
 
diff -uNr linux-kernel/net/ipv4/mpls4.c mpls-kernel-1/net/ipv4/mpls4.c
--- linux-kernel/net/ipv4/mpls4.c	1969-12-31 18:00:00.000000000 -0600
+++ mpls-kernel-1/net/ipv4/mpls4.c	2005-01-07 08:09:51.000000000 -0600
@@ -0,0 +1,390 @@
+/* mpls4.c: IPv4 MPLS protocol driver.
+ *
+ * Copyright (C) 2003 David S. Miller ([e-mail])
+ *
+ * Changes:
+ *	JLEU: 	Add ICMP handling
+ *		Add nexthop printing
+ *		Change nexthop resolve signature
+ *	JLEU:	Added mpls4_cache_flush()
+ *	JLEU:	un/register reserved labels in fini/init
+ *	JLEU:	removed sysfs print routin
+ */
+
+#include <linux/module.h>
+#include <linux/socket.h>
+#include <linux/skbuff.h>
+#include <linux/in.h>
+#include <linux/init.h>
+#include <net/dsfield.h>
+#include <net/neighbour.h>
+#include <net/route.h>
+#include <net/ip.h>
+#include <net/mpls.h>
+#include <net/icmp.h>
+#include <net/checksum.h>
+
+static void mpls4_cache_flush(void)
+{
+	rt_cache_flush(0);
+}
+
+static void mpls4_set_ttl(struct sk_buff *skb, int ttl)
+{
+	skb->nh.iph->ttl = ttl;
+	ip_send_check(skb->nh.iph);
+}
+
+static int mpls4_get_ttl(struct sk_buff *skb)
+{
+	return skb->nh.iph->ttl;
+}
+
+static void mpls4_change_dsfield(struct sk_buff *skb, int ds)
+{
+	ipv4_change_dsfield(skb->nh.iph, 0x3, ds << 2);
+}
+
+static int mpls4_get_dsfield(struct sk_buff *skb)
+{
+	return ipv4_get_dsfield(skb->nh.iph) >> 2;
+}
+
+struct mpls_icmp_common {
+#if defined(__LITTLE_ENDIAN_BITFIELD)
+	__u8    res1:4,
+		version:4;
+#elif defined (__BIG_ENDIAN_BITFIELD)
+	__u8    version:4,
+		res1:4;
+#else
+#error  "Please fix <asm/byteorder.h>"
+#endif
+	__u8	res2;
+	__u16	check;
+};
+
+struct mpls_icmp_object {
+	__u16	length;
+	__u8	class;
+	__u8	type;
+};
+
+/* we can probably used a modified ip_append_data to build this */
+static struct sk_buff*
+mpls4_build_icmp(struct sk_buff *skb, int type, unsigned int icmp_data,
+	int mpls)
+{
+	unsigned char buf[576];
+