icmp.c

linux 内核源代码

/*
 *	NET3:	Implementation of the ICMP protocol layer.
 *
 *		Alan Cox, <[e-mail]>
 *
 *	Version: $Id: icmp.c,v 1.85 2002/02/01 22:01:03 davem Exp $
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 *
 *	Some of the function names and the icmp unreach table for this
 *	module were derived from [icmp.c 1.0.11 06/02/93] by
 *	Ross Biro, Fred N. van Kempen, Mark Evans, Alan Cox, Gerhard Koerting.
 *	Other than that this module is a complete rewrite.
 *
 *	Fixes:
 *	Clemens Fruhwirth	:	introduce global icmp rate limiting
 *					with icmp type masking ability instead
 *					of broken per type icmp timeouts.
 *		Mike Shaver	:	RFC1122 checks.
 *		Alan Cox	:	Multicast ping reply as self.
 *		Alan Cox	:	Fix atomicity lockup in ip_build_xmit
 *					call.
 *		Alan Cox	:	Added 216,128 byte paths to the MTU
 *					code.
 *		Martin Mares	:	RFC1812 checks.
 *		Martin Mares	:	Can be configured to follow redirects
 *					if acting as a router _without_ a
 *					routing protocol (RFC 1812).
 *		Martin Mares	:	Echo requests may be configured to
 *					be ignored (RFC 1812).
 *		Martin Mares	:	Limitation of ICMP error message
 *					transmit rate (RFC 1812).
 *		Martin Mares	:	TOS and Precedence set correctly
 *					(RFC 1812).
 *		Martin Mares	:	Now copying as much data from the
 *					original packet as we can without
 *					exceeding 576 bytes (RFC 1812).
 *	Willy Konynenberg	:	Transparent proxying support.
 *		Keith Owens	:	RFC1191 correction for 4.2BSD based
 *					path MTU bug.
 *		Thomas Quinot	:	ICMP Dest Unreach codes up to 15 are
 *					valid (RFC 1812).
 *		Andi Kleen	:	Check all packet lengths properly
 *					and moved all kfree_skb() up to
 *					icmp_rcv.
 *		Andi Kleen	:	Move the rate limit bookkeeping
 *					into the dest entry and use a token
 *					bucket filter (thanks to ANK). Make
 *					the rates sysctl configurable.
 *		Yu Tianli	:	Fixed two ugly bugs in icmp_send
 *					- IP option length was accounted wrongly
 *					- ICMP header length was not accounted
 *					  at all.
 *              Tristan Greaves :       Added sysctl option to ignore bogus
 *              			broadcast responses from broken routers.
 *
 * To Fix:
 *
 *	- Should use skb_pull() instead of all the manual checking.
 *	  This would also greatly simply some upper layer error handlers. --AK
 *
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/jiffies.h>
#include <linux/kernel.h>
#include <linux/fcntl.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/inetdevice.h>
#include <linux/netdevice.h>
#include <linux/string.h>
#include <linux/netfilter_ipv4.h>
#include <net/snmp.h>
#include <net/ip.h>
#include <net/route.h>
#include <net/protocol.h>
#include <net/icmp.h>
#include <net/tcp.h>
#include <net/udp.h>
#include <net/raw.h>
#include <linux/skbuff.h>
#include <net/sock.h>
#include <linux/errno.h>
#include <linux/timer.h>
#include <linux/init.h>
#include <asm/system.h>
#include <asm/uaccess.h>
#include <net/checksum.h>

/*
 *	Build xmit assembly blocks
 */

struct icmp_bxm {
	struct sk_buff *skb;
	int offset;
	int data_len;

	struct {
		struct icmphdr icmph;
		__be32	       times[3];
	} data;
	int head_len;
	struct ip_options replyopts;
	unsigned char  optbuf[40];
};

/*
 *	Statistics
 */
DEFINE_SNMP_STAT(struct icmp_mib, icmp_statistics) __read_mostly;
DEFINE_SNMP_STAT(struct icmpmsg_mib, icmpmsg_statistics) __read_mostly;

/* An array of errno for error messages from dest unreach. */
/* RFC 1122: 3.2.2.1 States that NET_UNREACH, HOST_UNREACH and SR_FAILED MUST be considered 'transient errs'. */

struct icmp_err icmp_err_convert[] = {
	{
		.errno = ENETUNREACH,	/* ICMP_NET_UNREACH */
		.fatal = 0,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_HOST_UNREACH */
		.fatal = 0,
	},
	{
		.errno = ENOPROTOOPT	/* ICMP_PROT_UNREACH */,
		.fatal = 1,
	},
	{
		.errno = ECONNREFUSED,	/* ICMP_PORT_UNREACH */
		.fatal = 1,
	},
	{
		.errno = EMSGSIZE,	/* ICMP_FRAG_NEEDED */
		.fatal = 0,
	},
	{
		.errno = EOPNOTSUPP,	/* ICMP_SR_FAILED */
		.fatal = 0,
	},
	{
		.errno = ENETUNREACH,	/* ICMP_NET_UNKNOWN */
		.fatal = 1,
	},
	{
		.errno = EHOSTDOWN,	/* ICMP_HOST_UNKNOWN */
		.fatal = 1,
	},
	{
		.errno = ENONET,	/* ICMP_HOST_ISOLATED */
		.fatal = 1,
	},
	{
		.errno = ENETUNREACH,	/* ICMP_NET_ANO	*/
		.fatal = 1,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_HOST_ANO */
		.fatal = 1,
	},
	{
		.errno = ENETUNREACH,	/* ICMP_NET_UNR_TOS */
		.fatal = 0,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_HOST_UNR_TOS */
		.fatal = 0,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_PKT_FILTERED */
		.fatal = 1,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_PREC_VIOLATION */
		.fatal = 1,
	},
	{
		.errno = EHOSTUNREACH,	/* ICMP_PREC_CUTOFF */
		.fatal = 1,
	},
};

/* Control parameters for ECHO replies. */
int sysctl_icmp_echo_ignore_all __read_mostly;
int sysctl_icmp_echo_ignore_broadcasts __read_mostly = 1;

/* Control parameter - ignore bogus broadcast responses? */
int sysctl_icmp_ignore_bogus_error_responses __read_mostly = 1;

/*
 * 	Configurable global rate limit.
 *
 *	ratelimit defines tokens/packet consumed for dst->rate_token bucket
 *	ratemask defines which icmp types are ratelimited by setting
 * 	it's bit position.
 *
 *	default:
 *	dest unreachable (3), source quench (4),
 *	time exceeded (11), parameter problem (12)
 */

int sysctl_icmp_ratelimit __read_mostly = 1 * HZ;
int sysctl_icmp_ratemask __read_mostly = 0x1818;
int sysctl_icmp_errors_use_inbound_ifaddr __read_mostly;

/*
 *	ICMP control array. This specifies what to do with each ICMP.
 */

struct icmp_control {
	void (*handler)(struct sk_buff *skb);
	short   error;		/* This ICMP is classed as an error message */
};

static const struct icmp_control icmp_pointers[NR_ICMP_TYPES+1];

/*
 *	The ICMP socket(s). This is the most convenient way to flow control
 *	our ICMP output as well as maintain a clean interface throughout
 *	all layers. All Socketless IP sends will soon be gone.
 *
 *	On SMP we have one ICMP socket per-cpu.
 */
static DEFINE_PER_CPU(struct socket *, __icmp_socket) = NULL;
#define icmp_socket	__get_cpu_var(__icmp_socket)

static __inline__ int icmp_xmit_lock(void)
{
	local_bh_disable();

	if (unlikely(!spin_trylock(&icmp_socket->sk->sk_lock.slock))) {
		/* This can happen if the output path signals a
		 * dst_link_failure() for an outgoing ICMP packet.
		 */
		local_bh_enable();
		return 1;
	}
	return 0;
}

static void icmp_xmit_unlock(void)
{
	spin_unlock_bh(&icmp_socket->sk->sk_lock.slock);
}

/*
 *	Send an ICMP frame.
 */

/*
 *	Check transmit rate limitation for given message.
 *	The rate information is held in the destination cache now.
 *	This function is generic and could be used for other purposes
 *	too. It uses a Token bucket filter as suggested by Alexey Kuznetsov.
 *
 *	Note that the same dst_entry fields are modified by functions in
 *	route.c too, but these work for packet destinations while xrlim_allow
 *	works for icmp destinations. This means the rate limiting information
 *	for one "ip object" is shared - and these ICMPs are twice limited:
 *	by source and by destination.
 *
 *	RFC 1812: 4.3.2.8 SHOULD be able to limit error message rate
 *			  SHOULD allow setting of rate limits
 *
 * 	Shared between ICMPv4 and ICMPv6.
 */
#define XRLIM_BURST_FACTOR 6
int xrlim_allow(struct dst_entry *dst, int timeout)
{
	unsigned long now;
	int rc = 0;

	now = jiffies;
	dst->rate_tokens += now - dst->rate_last;
	dst->rate_last = now;
	if (dst->rate_tokens > XRLIM_BURST_FACTOR * timeout)
		dst->rate_tokens = XRLIM_BURST_FACTOR * timeout;
	if (dst->rate_tokens >= timeout) {
		dst->rate_tokens -= timeout;
		rc = 1;
	}
	return rc;
}

static inline int icmpv4_xrlim_allow(struct rtable *rt, int type, int code)
{
	struct dst_entry *dst = &rt->u.dst;
	int rc = 1;

	if (type > NR_ICMP_TYPES)
		goto out;

	/* Don't limit PMTU discovery. */
	if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
		goto out;

	/* No rate limit on loopback */
	if (dst->dev && (dst->dev->flags&IFF_LOOPBACK))
		goto out;

	/* Limit if icmp type is enabled in ratemask. */
	if ((1 << type) & sysctl_icmp_ratemask)
		rc = xrlim_allow(dst, sysctl_icmp_ratelimit);
out:
	return rc;
}

/*
 *	Maintain the counters used in the SNMP statistics for outgoing ICMP
 */
void icmp_out_count(unsigned char type)
{
	ICMPMSGOUT_INC_STATS(type);
	ICMP_INC_STATS(ICMP_MIB_OUTMSGS);
}

/*
 *	Checksum each fragment, and on the first include the headers and final
 *	checksum.
 */
static int icmp_glue_bits(void *from, char *to, int offset, int len, int odd,
			  struct sk_buff *skb)
{
	struct icmp_bxm *icmp_param = (struct icmp_bxm *)from;
	__wsum csum;

	csum = skb_copy_and_csum_bits(icmp_param->skb,
				      icmp_param->offset + offset,
				      to, len, 0);

	skb->csum = csum_block_add(skb->csum, csum, odd);
	if (icmp_pointers[icmp_param->data.icmph.type].error)
		nf_ct_attach(skb, icmp_param->skb);
	return 0;
}

static void icmp_push_reply(struct icmp_bxm *icmp_param,
			    struct ipcm_cookie *ipc, struct rtable *rt)
{
	struct sk_buff *skb;

	if (ip_append_data(icmp_socket->sk, icmp_glue_bits, icmp_param,
			   icmp_param->data_len+icmp_param->head_len,
			   icmp_param->head_len,
			   ipc, rt, MSG_DONTWAIT) < 0)
		ip_flush_pending_frames(icmp_socket->sk);
	else if ((skb = skb_peek(&icmp_socket->sk->sk_write_queue)) != NULL) {
		struct icmphdr *icmph = icmp_hdr(skb);
		__wsum csum = 0;
		struct sk_buff *skb1;

		skb_queue_walk(&icmp_socket->sk->sk_write_queue, skb1) {
			csum = csum_add(csum, skb1->csum);
		}
		csum = csum_partial_copy_nocheck((void *)&icmp_param->data,
						 (char *)icmph,
						 icmp_param->head_len, csum);
		icmph->checksum = csum_fold(csum);
		skb->ip_summed = CHECKSUM_NONE;
		ip_push_pending_frames(icmp_socket->sk);
	}
}

/*
 *	Driving logic for building and sending ICMP messages.
 */

static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
{
	struct sock *sk = icmp_socket->sk;
	struct inet_sock *inet = inet_sk(sk);
	struct ipcm_cookie ipc;
	struct rtable *rt = (struct rtable *)skb->dst;
	__be32 daddr;

	if (ip_options_echo(&icmp_param->replyopts, skb))
		return;

	if (icmp_xmit_lock())
		return;

	icmp_param->data.icmph.checksum = 0;

	inet->tos = ip_hdr(skb)->tos;
	daddr = ipc.addr = rt->rt_src;
	ipc.opt = NULL;
	if (icmp_param->replyopts.optlen) {
		ipc.opt = &icmp_param->replyopts;
		if (ipc.opt->srr)
			daddr = icmp_param->replyopts.faddr;
	}
	{
		struct flowi fl = { .nl_u = { .ip4_u =
					      { .daddr = daddr,
						.saddr = rt->rt_spec_dst,
						.tos = RT_TOS(ip_hdr(skb)->tos) } },
				    .proto = IPPROTO_ICMP };
		security_skb_classify_flow(skb, &fl);
		if (ip_route_output_key(&rt, &fl))
			goto out_unlock;
	}
	if (icmpv4_xrlim_allow(rt, icmp_param->data.icmph.type,
			       icmp_param->data.icmph.code))
		icmp_push_reply(icmp_param, &ipc, rt);
	ip_rt_put(rt);
out_unlock:
	icmp_xmit_unlock();
}


/*
 *	Send an ICMP message in response to a situation
 *
 *	RFC 1122: 3.2.2	MUST send at least the IP header and 8 bytes of header.
 *		  MAY send more (we do).
 *			MUST NOT change this header information.
 *			MUST NOT reply to a multicast/broadcast IP address.
 *			MUST NOT reply to a multicast/broadcast MAC address.
 *			MUST reply to only the first fragment.
 */

void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
{
	struct iphdr *iph;
	int room;
	struct icmp_bxm icmp_param;
	struct rtable *rt = (struct rtable *)skb_in->dst;
	struct ipcm_cookie ipc;
	__be32 saddr;
	u8  tos;

	if (!rt)
		goto out;

	/*
	 *	Find the original header. It is expected to be valid, of course.
	 *	Check this, icmp_send is called from the most obscure devices
	 *	sometimes.
	 */
	iph = ip_hdr(skb_in);

	if ((u8 *)iph < skb_in->head ||
	    (skb_in->network_header + sizeof(*iph)) > skb_in->tail)
		goto out;

	/*
	 *	No replies to physical multicast/broadcast
	 */
	if (skb_in->pkt_type != PACKET_HOST)
		goto out;

	/*
	 *	Now check at the protocol level
	 */
	if (rt->rt_flags & (RTCF_BROADCAST | RTCF_MULTICAST))
		goto out;

	/*
	 *	Only reply to fragment 0. We byte re-order the constant
	 *	mask for efficiency.
	 */
	if (iph->frag_off & htons(IP_OFFSET))
		goto out;

	/*
	 *	If we send an ICMP error to an ICMP error a mess would result..
	 */
	if (icmp_pointers[type].error) {
		/*
		 *	We are an error, check if we are replying to an
		 *	ICMP error
		 */
		if (iph->protocol == IPPROTO_ICMP) {
			u8 _inner_type, *itp;

			itp = skb_header_pointer(skb_in,
						 skb_network_header(skb_in) +
						 (iph->ihl << 2) +
						 offsetof(struct icmphdr,
							  type) -
						 skb_in->data,
						 sizeof(_inner_type),
						 &_inner_type);
			if (itp == NULL)
				goto out;

			/*
			 *	Assume any unknown ICMP type is an error. This
			 *	isn't specified by the RFC, but think about it..
			 */
			if (*itp > NR_ICMP_TYPES ||
			    icmp_pointers[*itp].error)
				goto out;
		}
	}

	if (icmp_xmit_lock())
		return;

	/*
	 *	Construct source address and options.
	 */

	saddr = iph->daddr;
	if (!(rt->rt_flags & RTCF_LOCAL)) {
		struct net_device *dev = NULL;

		if (rt->fl.iif && sysctl_icmp_errors_use_inbound_ifaddr)
			dev = dev_get_by_index(&init_net, rt->fl.iif);

		if (dev) {
			saddr = inet_select_addr(dev, 0, RT_SCOPE_LINK);
			dev_put(dev);
		} else
			saddr = 0;
	}

	tos = icmp_pointers[type].error ? ((iph->tos & IPTOS_TOS_MASK) |
					   IPTOS_PREC_INTERNETCONTROL) :
					  iph->tos;

	if (ip_options_echo(&icmp_param.replyopts, skb_in))
		goto out_unlock;


	/*
	 *	Prepare data for ICMP header.
	 */

	icmp_param.data.icmph.type	 = type;
	icmp_param.data.icmph.code	 = code;
	icmp_param.data.icmph.un.gateway = info;
	icmp_param.data.icmph.checksum	 = 0;
	icmp_param.skb	  = skb_in;
	icmp_param.offset = skb_network_offset(skb_in);
	inet_sk(icmp_socket->sk)->tos = tos;
	ipc.addr = iph->saddr;
	ipc.opt = &icmp_param.replyopts;

	{
		struct flowi fl = {
			.nl_u = {
				.ip4_u = {
					.daddr = icmp_param.replyopts.srr ?
						icmp_param.replyopts.faddr :
						iph->saddr,
					.saddr = saddr,