Home | History | Annotate | Line # | Download | only in netinet6
ip6_flow.c revision 1.12 
      1 /*	$NetBSD: ip6_flow.c,v 1.12 2008/01/04 23:26:44 dyoung Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 2007 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by the 3am Software Foundry ("3am").  It was developed by Liam J. Foy
      9  * <liamjfoy (at) netbsd.org> and Matt Thomas <matt (at) netbsd.org>.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  * 3. All advertising materials mentioning features or use of this software
     20  *    must display the following acknowledgement:
     21  *	This product includes software developed by the NetBSD
     22  *	Foundation, Inc. and its contributors.
     23  * 4. Neither the name of The NetBSD Foundation nor the names of its
     24  *    contributors may be used to endorse or promote products derived
     25  *    from this software without specific prior written permission.
     26  *
     27  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     28  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     37  * POSSIBILITY OF SUCH DAMAGE.
     38  *
     39  * IPv6 version was developed by Liam J. Foy. Original source existed in IPv4
     40  * format developed by Matt Thomas. Thanks to Joerg Sonnenberger, Matt
     41  * Thomas and Christos Zoulas.
     42  *
     43  * Thanks to Liverpool John Moores University, especially Dr. David Llewellyn-Jones
     44  * for providing resources (to test) and Professor Madjid Merabti.
     45  */
     46 
     47 #include <sys/cdefs.h>
     48 __KERNEL_RCSID(0, "$NetBSD: ip6_flow.c,v 1.12 2008/01/04 23:26:44 dyoung Exp $");
     49 
     50 #include <sys/param.h>
     51 #include <sys/systm.h>
     52 #include <sys/malloc.h>
     53 #include <sys/mbuf.h>
     54 #include <sys/domain.h>
     55 #include <sys/protosw.h>
     56 #include <sys/socket.h>
     57 #include <sys/socketvar.h>
     58 #include <sys/time.h>
     59 #include <sys/kernel.h>
     60 #include <sys/pool.h>
     61 #include <sys/sysctl.h>
     62 
     63 #include <net/if.h>
     64 #include <net/if_dl.h>
     65 #include <net/route.h>
     66 #include <net/pfil.h>
     67 
     68 #include <netinet/in.h>
     69 #include <netinet6/in6_var.h>
     70 #include <netinet/in_systm.h>
     71 #include <netinet/ip6.h>
     72 #include <netinet6/ip6_var.h>
     73 
     74 /*
     75  * IPv6 Fast Forward caches/hashes flows from one source to destination.
     76  *
     77  * Upon a successful forward IPv6FF caches and hashes details such as the
     78  * route, source and destination. Once another packet is received matching
     79  * the source and destination the packet is forwarded straight onto if_output
     80  * using the cached details.
     81  *
     82  * Example:
     83  * ether/fddi_input -> ip6flow_fastfoward -> if_output
     84  */
     85 
     86 POOL_INIT(ip6flow_pool, sizeof(struct ip6flow), 0, 0, 0, "ip6flowpl", NULL,
     87     IPL_NET);
     88 
     89 LIST_HEAD(ip6flowhead, ip6flow);
     90 
     91 /*
     92  * We could use IPv4 defines (IPFLOW_HASHBITS) but we'll
     93  * use our own (possibly for future expansion).
     94  */
     95 #define	IP6FLOW_TIMER		(5 * PR_SLOWHZ)
     96 #define	IP6FLOW_DEFAULT_HASHSIZE	(1 << IP6FLOW_HASHBITS)
     97 
     98 static struct ip6flowhead *ip6flowtable = NULL;
     99 static struct ip6flowhead ip6flowlist;
    100 static int ip6flow_inuse;
    101 
    102 /*
    103  * Insert an ip6flow into the list.
    104  */
    105 #define	IP6FLOW_INSERT(bucket, ip6f) \
    106 do { \
    107 	LIST_INSERT_HEAD((bucket), (ip6f), ip6f_hash); \
    108 	LIST_INSERT_HEAD(&ip6flowlist, (ip6f), ip6f_list); \
    109 } while (/*CONSTCOND*/ 0)
    110 
    111 /*
    112  * Remove an ip6flow from the list.
    113  */
    114 #define	IP6FLOW_REMOVE(ip6f) \
    115 do { \
    116 	LIST_REMOVE((ip6f), ip6f_hash); \
    117 	LIST_REMOVE((ip6f), ip6f_list); \
    118 } while (/*CONSTCOND*/ 0)
    119 
    120 #ifndef IP6FLOW_DEFAULT
    121 #define	IP6FLOW_DEFAULT		256
    122 #endif
    123 
    124 int ip6_maxflows = IP6FLOW_DEFAULT;
    125 int ip6_hashsize = IP6FLOW_DEFAULT_HASHSIZE;
    126 
    127 /*
    128  * Calculate hash table position.
    129  */
    130 static size_t
    131 ip6flow_hash(struct ip6_hdr *ip6)
    132 {
    133 	size_t hash;
    134 	uint32_t dst_sum, src_sum;
    135 	size_t idx;
    136 
    137 	src_sum = ip6->ip6_src.s6_addr32[0] + ip6->ip6_src.s6_addr32[1]
    138 	    + ip6->ip6_src.s6_addr32[2] + ip6->ip6_src.s6_addr32[3];
    139 	dst_sum = ip6->ip6_dst.s6_addr32[0] + ip6->ip6_dst.s6_addr32[1]
    140 	    + ip6->ip6_dst.s6_addr32[2] + ip6->ip6_dst.s6_addr32[3];
    141 
    142 	hash = ip6->ip6_flow;
    143 
    144 	for (idx = 0; idx < 32; idx += IP6FLOW_HASHBITS)
    145 		hash += (dst_sum >> (32 - idx)) + (src_sum >> idx);
    146 
    147 	return hash & (ip6_hashsize-1);
    148 }
    149 
    150 /*
    151  * Check to see if a flow already exists - if so return it.
    152  */
    153 static struct ip6flow *
    154 ip6flow_lookup(struct ip6_hdr *ip6)
    155 {
    156 	size_t hash;
    157 	struct ip6flow *ip6f;
    158 
    159 	hash = ip6flow_hash(ip6);
    160 
    161 	LIST_FOREACH(ip6f, &ip6flowtable[hash], ip6f_hash) {
    162 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &ip6f->ip6f_dst)
    163 		    && IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, &ip6f->ip6f_src)
    164 		    && ip6f->ip6f_flow == ip6->ip6_flow) {
    165 		    	/* A cached flow has been found. */
    166 			return ip6f;
    167 		}
    168 	}
    169 
    170 	return NULL;
    171 }
    172 
    173 /*
    174  * Allocate memory and initialise lists. This function is called
    175  * from ip6_init and called there after to resize the hash table.
    176  * If a newly sized table cannot be malloc'ed we just continue
    177  * to use the old one.
    178  */
    179 int
    180 ip6flow_init(int table_size)
    181 {
    182 	struct ip6flowhead *new_table;
    183 	size_t i;
    184 
    185 	new_table = (struct ip6flowhead *)malloc(sizeof(struct ip6flowhead) *
    186 	    table_size, M_RTABLE, M_NOWAIT);
    187 
    188 	if (new_table == NULL)
    189 		return 1;
    190 
    191 	if (ip6flowtable != NULL)
    192 		free(ip6flowtable, M_RTABLE);
    193 
    194 	ip6flowtable = new_table;
    195 	ip6_hashsize = table_size;
    196 
    197 	LIST_INIT(&ip6flowlist);
    198 	for (i = 0; i < ip6_hashsize; i++)
    199 		LIST_INIT(&ip6flowtable[i]);
    200 
    201 	return 0;
    202 }
    203 
    204 /*
    205  * IPv6 Fast Forward routine. Attempt to forward the packet -
    206  * if any problems are found return to the main IPv6 input
    207  * routine to deal with.
    208  */
    209 int
    210 ip6flow_fastforward(struct mbuf *m)
    211 {
    212 	struct ip6flow *ip6f;
    213 	struct ip6_hdr *ip6;
    214 	struct rtentry *rt;
    215 	const struct sockaddr *dst;
    216 	int error;
    217 
    218 	/*
    219 	 * Are we forwarding packets and have flows?
    220 	 */
    221 	if (!ip6_forwarding || ip6flow_inuse == 0)
    222 		return 0;
    223 
    224 	/*
    225 	 * At least size of IPv6 Header?
    226 	 */
    227 	if (m->m_len < sizeof(struct ip6_hdr))
    228 		return 0;
    229 	/*
    230 	 * Was packet received as a link-level multicast or broadcast?
    231 	 * If so, don't try to fast forward.
    232 	 */
    233 	if ((m->m_flags & (M_BCAST|M_MCAST)) != 0)
    234 		return 0;
    235 
    236 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    237 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    238 				(max_linkhdr + 3) & ~3)) == NULL) {
    239 			return 0;
    240 		}
    241 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    242 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    243 			return 0;
    244 		}
    245 	}
    246 
    247 	ip6 = mtod(m, struct ip6_hdr *);
    248 
    249 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    250 		/* Bad version. */
    251 		return 0;
    252 	}
    253 
    254 	/*
    255 	 * If we have a hop-by-hop extension we must process it.
    256 	 * We just leave this up to ip6_input to deal with.
    257 	 */
    258 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS)
    259 		return 0;
    260 
    261 	/*
    262 	 * Attempt to find a flow.
    263 	 */
    264 	if ((ip6f = ip6flow_lookup(ip6)) == NULL) {
    265 		/* No flow found. */
    266 		return 0;
    267 	}
    268 
    269 	/*
    270 	 * Route and interface still up?
    271 	 */
    272 	if ((rt = rtcache_validate(&ip6f->ip6f_ro)) == NULL ||
    273 	    (rt->rt_ifp->if_flags & IFF_UP) == 0) {
    274 	    	/* Route or interface is down */
    275 		return 0;
    276 	}
    277 
    278 	/*
    279 	 * Packet size greater than MTU?
    280 	 */
    281 	if (m->m_pkthdr.len > rt->rt_ifp->if_mtu) {
    282 		/* Return to main IPv6 input function. */
    283 		return 0;
    284 	}
    285 
    286 	if (ip6->ip6_hlim <= IPV6_HLIMDEC)
    287 		return 0;
    288 
    289 	/* Decrement hop limit (same as TTL) */
    290 	ip6->ip6_hlim -= IPV6_HLIMDEC;
    291 
    292 	if (rt->rt_flags & RTF_GATEWAY)
    293 		dst = rt->rt_gateway;
    294 	else
    295 		dst = rtcache_getdst(&ip6f->ip6f_ro);
    296 
    297 	PRT_SLOW_ARM(ip6f->ip6f_timer, IP6FLOW_TIMER);
    298 
    299 	ip6f->ip6f_uses++;
    300 
    301 	/* Send on its way - straight to the interface output routine. */
    302 	if ((error = (*rt->rt_ifp->if_output)(rt->rt_ifp, m, dst, rt)) != 0) {
    303 		ip6f->ip6f_dropped++;
    304 	} else {
    305 		ip6f->ip6f_forwarded++;
    306 	}
    307 
    308 	return 1;
    309 }
    310 
    311 /*
    312  * Add the IPv6 flow statistics to the main IPv6 statistics.
    313  */
    314 static void
    315 ip6flow_addstats(struct ip6flow *ip6f)
    316 {
    317 	struct rtentry *rt;
    318 
    319 	if ((rt = rtcache_validate(&ip6f->ip6f_ro)) != NULL)
    320 		rt->rt_use += ip6f->ip6f_uses;
    321 	ip6stat.ip6s_fastforwardflows = ip6flow_inuse;
    322 	ip6stat.ip6s_cantforward += ip6f->ip6f_dropped;
    323 	ip6stat.ip6s_odropped += ip6f->ip6f_dropped;
    324 	ip6stat.ip6s_total += ip6f->ip6f_uses;
    325 	ip6stat.ip6s_forward += ip6f->ip6f_forwarded;
    326 	ip6stat.ip6s_fastforward += ip6f->ip6f_forwarded;
    327 }
    328 
    329 /*
    330  * Add statistics and free the flow.
    331  */
    332 static void
    333 ip6flow_free(struct ip6flow *ip6f)
    334 {
    335 	int s;
    336 
    337 	/*
    338 	 * Remove the flow from the hash table (at elevated IPL).
    339 	 * Once it's off the list, we can deal with it at normal
    340 	 * network IPL.
    341 	 */
    342 	s = splnet();
    343 	IP6FLOW_REMOVE(ip6f);
    344 	splx(s);
    345 	ip6flow_inuse--;
    346 	ip6flow_addstats(ip6f);
    347 	rtcache_free(&ip6f->ip6f_ro);
    348 	pool_put(&ip6flow_pool, ip6f);
    349 }
    350 
    351 /*
    352  * Reap one or more flows - ip6flow_reap may remove
    353  * multiple flows if net.inet6.ip6.maxflows is reduced.
    354  */
    355 struct ip6flow *
    356 ip6flow_reap(int just_one)
    357 {
    358 	while (just_one || ip6flow_inuse > ip6_maxflows) {
    359 		struct ip6flow *ip6f, *maybe_ip6f = NULL;
    360 		int s;
    361 
    362 		ip6f = LIST_FIRST(&ip6flowlist);
    363 		while (ip6f != NULL) {
    364 			/*
    365 			 * If this no longer points to a valid route -
    366 			 * reclaim it.
    367 			 */
    368 			if (rtcache_validate(&ip6f->ip6f_ro) == NULL)
    369 				goto done;
    370 			/*
    371 			 * choose the one that's been least recently
    372 			 * used or has had the least uses in the
    373 			 * last 1.5 intervals.
    374 			 */
    375 			if (maybe_ip6f == NULL ||
    376 			    ip6f->ip6f_timer < maybe_ip6f->ip6f_timer ||
    377 			    (ip6f->ip6f_timer == maybe_ip6f->ip6f_timer &&
    378 			     ip6f->ip6f_last_uses + ip6f->ip6f_uses <
    379 			         maybe_ip6f->ip6f_last_uses +
    380 			         maybe_ip6f->ip6f_uses))
    381 				maybe_ip6f = ip6f;
    382 			ip6f = LIST_NEXT(ip6f, ip6f_list);
    383 		}
    384 		ip6f = maybe_ip6f;
    385 	    done:
    386 		/*
    387 		 * Remove the entry from the flow table
    388 		 */
    389 		s = splnet();
    390 		IP6FLOW_REMOVE(ip6f);
    391 		splx(s);
    392 		rtcache_free(&ip6f->ip6f_ro);
    393 		if (just_one) {
    394 			ip6flow_addstats(ip6f);
    395 			return ip6f;
    396 		}
    397 		ip6flow_inuse--;
    398 		ip6flow_addstats(ip6f);
    399 		pool_put(&ip6flow_pool, ip6f);
    400 	}
    401 	return NULL;
    402 }
    403 
    404 void
    405 ip6flow_slowtimo(void)
    406 {
    407 	struct ip6flow *ip6f, *next_ip6f;
    408 
    409 	for (ip6f = LIST_FIRST(&ip6flowlist); ip6f != NULL; ip6f = next_ip6f) {
    410 		next_ip6f = LIST_NEXT(ip6f, ip6f_list);
    411 		if (PRT_SLOW_ISEXPIRED(ip6f->ip6f_timer) ||
    412 		    rtcache_validate(&ip6f->ip6f_ro) == NULL) {
    413 			ip6flow_free(ip6f);
    414 		} else {
    415 			ip6f->ip6f_last_uses = ip6f->ip6f_uses;
    416 			ip6flow_addstats(ip6f);
    417 			ip6f->ip6f_uses = 0;
    418 			ip6f->ip6f_dropped = 0;
    419 			ip6f->ip6f_forwarded = 0;
    420 		}
    421 	}
    422 }
    423 
    424 /*
    425  * We have successfully forwarded a packet using the normal
    426  * IPv6 stack. Now create/update a flow.
    427  */
    428 void
    429 ip6flow_create(const struct route *ro, struct mbuf *m)
    430 {
    431 	struct ip6_hdr *ip6;
    432 	struct ip6flow *ip6f;
    433 	size_t hash;
    434 	int s;
    435 
    436 	ip6 = mtod(m, struct ip6_hdr *);
    437 
    438 	/*
    439 	 * If IPv6 Fast Forward is disabled, don't create a flow.
    440 	 * It can be disabled by setting net.inet6.ip6.maxflows to 0.
    441 	 *
    442 	 * Don't create a flow for ICMPv6 messages.
    443 	 */
    444 	if (ip6_maxflows == 0 || ip6->ip6_nxt == IPPROTO_IPV6_ICMP)
    445 		return;
    446 
    447 	/*
    448 	 * See if an existing flow exists.  If so:
    449 	 *	- Remove the flow
    450 	 *	- Add flow statistics
    451 	 *	- Free the route
    452 	 *	- Reset statistics
    453 	 *
    454 	 * If a flow doesn't exist allocate a new one if
    455 	 * ip6_maxflows hasn't reached its limit. If it has
    456 	 * been reached, reap some flows.
    457 	 */
    458 	ip6f = ip6flow_lookup(ip6);
    459 	if (ip6f == NULL) {
    460 		if (ip6flow_inuse >= ip6_maxflows) {
    461 			ip6f = ip6flow_reap(1);
    462 		} else {
    463 			ip6f = pool_get(&ip6flow_pool, PR_NOWAIT);
    464 			if (ip6f == NULL)
    465 				return;
    466 			ip6flow_inuse++;
    467 		}
    468 		memset(ip6f, 0, sizeof(*ip6f));
    469 	} else {
    470 		s = splnet();
    471 		IP6FLOW_REMOVE(ip6f);
    472 		splx(s);
    473 		ip6flow_addstats(ip6f);
    474 		rtcache_free(&ip6f->ip6f_ro);
    475 		ip6f->ip6f_uses = 0;
    476 		ip6f->ip6f_last_uses = 0;
    477 		ip6f->ip6f_dropped = 0;
    478 		ip6f->ip6f_forwarded = 0;
    479 	}
    480 
    481 	/*
    482 	 * Fill in the updated/new details.
    483 	 */
    484 	rtcache_copy(&ip6f->ip6f_ro, ro);
    485 	ip6f->ip6f_dst = ip6->ip6_dst;
    486 	ip6f->ip6f_src = ip6->ip6_src;
    487 	ip6f->ip6f_flow = ip6->ip6_flow;
    488 	PRT_SLOW_ARM(ip6f->ip6f_timer, IP6FLOW_TIMER);
    489 	ip6f->ip6f_start = time_uptime;
    490 
    491 	/*
    492 	 * Insert into the approriate bucket of the flow table.
    493 	 */
    494 	hash = ip6flow_hash(ip6);
    495 	s = splnet();
    496 	IP6FLOW_INSERT(&ip6flowtable[hash], ip6f);
    497 	splx(s);
    498 }
    499 
    500 /*
    501  * Invalidate/remove all flows - if new_size is positive we
    502  * resize the hash table.
    503  */
    504 int
    505 ip6flow_invalidate_all(int new_size)
    506 {
    507 	struct ip6flow *ip6f, *next_ip6f;
    508 	int s, error;
    509 
    510 	error = 0;
    511 	s = splnet();
    512 	for (ip6f = LIST_FIRST(&ip6flowlist); ip6f != NULL; ip6f = next_ip6f) {
    513 		next_ip6f = LIST_NEXT(ip6f, ip6f_list);
    514 		ip6flow_free(ip6f);
    515 	}
    516 
    517 	if (new_size)
    518 		error = ip6flow_init(new_size);
    519 	splx(s);
    520 
    521 	return error;
    522 }
    523