From owner-freebsd-net Sun May 3 16:44:25 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA24865 for freebsd-net-outgoing; Sun, 3 May 1998 16:44:25 -0700 (PDT) (envelope-from owner-freebsd-net@FreeBSD.ORG) Received: from frmug.org (frmug-gw.frmug.org [193.56.58.252]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id QAA24854 for ; Sun, 3 May 1998 16:44:09 -0700 (PDT) (envelope-from pb@fasterix.frmug.org) Received: (from uucp@localhost) by frmug.org (8.9.0.Beta7/frmug-2.3/nospam) with UUCP id BAA23707 for freebsd-net@freebsd.org; Mon, 4 May 1998 01:44:02 +0200 (CEST) (envelope-from pb@fasterix.frmug.org) Received: (from pb@localhost) by fasterix.frmug.org (8.8.8/8.8.5/pb-19970302) id BAA27510 for freebsd-net@freebsd.org; Mon, 4 May 1998 01:43:11 +0200 (CEST) Message-ID: <19980504014310.A22037@fasterix.frmug.fr.net> Date: Mon, 4 May 1998 01:43:10 +0200 From: Pierre Beyssac To: freebsd-net@FreeBSD.ORG Subject: patches for fast forwarding (with kernel option & sysctl) Mime-Version: 1.0 Content-Type: multipart/mixed; boundary=VS++wcV0S1rZb1Fb X-Mailer: Mutt 0.92.2 Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org --VS++wcV0S1rZb1Fb Content-Type: text/plain; charset=us-ascii Here are new patches to use NetBSD's IPFLOW stuff on -current. To use, store ipflow.c in /sys/netinet, then add IPFLOW to your kernel config. After booting, activate with: sysctl -w net.inet.ip.fastforwarding=1 Two attachments follow: the patches and the additional file. -- Pierre Beyssac pb@fasterix.frmug.org pb@fasterix.freenix.org {Free,Net,Open}BSD, Linux : il y a moins bien, mais c'est plus cher Free domains: http://www.eu.org/ or mail dns-manager@EU.org --VS++wcV0S1rZb1Fb Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename=patch2 Index: conf/files =================================================================== RCS file: /home/ncvs/src/sys/conf/files,v retrieving revision 1.136 diff -u -r1.136 files --- files 1998/04/22 18:12:29 1.136 +++ files 1998/05/03 23:30:21 @@ -268,6 +268,7 @@ netinet/ip_auth.c optional ipfilter inet netinet/ip_divert.c optional ipdivert netinet/ip_fil.c optional ipfilter inet +netinet/ip_flow.c optional ipflow inet netinet/ip_frag.c optional ipfilter inet netinet/ip_fw.c optional ipfirewall netinet/ip_icmp.c optional inet Index: conf/options =================================================================== RCS file: /home/ncvs/src/sys/conf/options,v retrieving revision 1.72 diff -u -r1.72 options --- options 1998/04/20 04:30:41 1.72 +++ options 1998/05/03 23:30:24 @@ -172,6 +172,7 @@ IPFIREWALL_VERBOSE opt_ipfw.h IPFIREWALL_VERBOSE_LIMIT opt_ipfw.h IPFIREWALL_DEFAULT_TO_ACCEPT opt_ipfw.h +IPFLOW opt_ipflow.h IPX opt_ipx.h IPXIP opt_ipx.h IPTUNNEL opt_ipx.h Index: i386/conf/LINT =================================================================== RCS file: /home/ncvs/src/sys/i386/conf/LINT,v retrieving revision 1.429 diff -u -r1.429 LINT --- LINT 1998/04/29 17:09:41 1.429 +++ LINT 1998/05/03 23:30:53 @@ -419,6 +419,9 @@ # IPFILTER_LOG enables ipfilter's logging. # IPFILTER_LKM enables LKM support for an ipfilter module (untested). # +# IPFLOW enables Matt Thomas' fast IP forwarding code +# (activate using "sysctl -w net.inet.ip.fastforwarding=1") +# # TCPDEBUG is undocumented. # options "TCP_COMPAT_42" #emulate 4.2BSD TCP bugs @@ -432,6 +435,7 @@ options IPFILTER #kernel ipfilter support options IPFILTER_LOG #ipfilter logging #options IPFILTER_LKM #kernel support for ip_fil.o LKM +options IPFLOW options TCPDEBUG Index: net/if_ethersubr.c =================================================================== RCS file: /home/ncvs/src/sys/net/if_ethersubr.c,v retrieving revision 1.47 diff -u -r1.47 if_ethersubr.c --- if_ethersubr.c 1998/03/30 09:51:39 1.47 +++ if_ethersubr.c 1998/05/03 23:31:06 @@ -36,6 +36,7 @@ #include "opt_atalk.h" #include "opt_inet.h" +#include "opt_ipflow.h" #include "opt_ipx.h" #include @@ -501,6 +502,10 @@ switch (ether_type) { #ifdef INET case ETHERTYPE_IP: +#ifdef IPFLOW + if (ipflow_fastforward(m)) + return; +#endif schednetisr(NETISR_IP); inq = &ipintrq; break; Index: net/if_fddisubr.c =================================================================== RCS file: /home/ncvs/src/sys/net/if_fddisubr.c,v retrieving revision 1.27 diff -u -r1.27 if_fddisubr.c --- if_fddisubr.c 1998/03/30 09:51:44 1.27 +++ if_fddisubr.c 1998/05/03 23:31:18 @@ -38,6 +38,7 @@ #include "opt_atalk.h" #include "opt_inet.h" +#include "opt_ipflow.h" #include "opt_ipx.h" #include @@ -533,6 +534,10 @@ switch (type) { #ifdef INET case ETHERTYPE_IP: +#ifdef IPFLOW + if (ipflow_fastforward(m)) + return; +#endif schednetisr(NETISR_IP); inq = &ipintrq; break; Index: net/if_ppp.c =================================================================== RCS file: /home/ncvs/src/sys/net/if_ppp.c,v retrieving revision 1.56 diff -u -r1.56 if_ppp.c --- if_ppp.c 1998/04/06 11:43:10 1.56 +++ if_ppp.c 1998/05/03 23:31:50 @@ -77,6 +77,7 @@ #if NPPP > 0 #include "opt_inet.h" +#include "opt_ipflow.h" #include "opt_ipx.h" #include "opt_ppp.h" @@ -1488,6 +1489,12 @@ m->m_pkthdr.len -= PPP_HDRLEN; m->m_data += PPP_HDRLEN; m->m_len -= PPP_HDRLEN; +#ifdef IPFLOW + if (ipflow_fastforward(m)) { + sc->sc_last_recv = time_second; + return; + } +#endif schednetisr(NETISR_IP); inq = &ipintrq; sc->sc_last_recv = time_second; /* update time of last pkt rcvd */ Index: netinet/in.h =================================================================== RCS file: /home/ncvs/src/sys/netinet/in.h,v retrieving revision 1.31 diff -u -r1.31 in.h --- in.h 1998/04/19 17:22:27 1.31 +++ in.h 1998/05/03 23:32:00 @@ -398,7 +398,8 @@ #define IPCTL_INTRQDROPS 11 /* number of netisr q drops */ #define IPCTL_STATS 12 /* ipstat structure */ #define IPCTL_ACCEPTSOURCEROUTE 13 /* may accept source routed packets */ -#define IPCTL_MAXID 14 +#define IPCTL_FASTFORWARDING 14 /* use fast IP forwarding code */ +#define IPCTL_MAXID 15 #define IPCTL_NAMES { \ { 0, 0 }, \ @@ -415,6 +416,7 @@ { "intr-queue-drops", CTLTYPE_INT }, \ { "stats", CTLTYPE_STRUCT }, \ { "accept_sourceroute", CTLTYPE_INT }, \ + { "fastforwarding", CTLTYPE_INT }, \ } Index: netinet/in_var.h =================================================================== RCS file: /home/ncvs/src/sys/netinet/in_var.h,v retrieving revision 1.27 diff -u -r1.27 in_var.h --- in_var.h 1997/09/07 05:26:43 1.27 +++ in_var.h 1998/05/03 23:32:02 @@ -211,6 +211,7 @@ IN_NEXT_MULTI((step), (inm)); \ } while(0) +struct route; struct in_multi *in_addmulti __P((struct in_addr *, struct ifnet *)); void in_delmulti __P((struct in_multi *)); int in_control __P((struct socket *, int, caddr_t, struct ifnet *, @@ -219,6 +220,9 @@ void ip_input __P((struct mbuf *)); int in_ifadown __P((struct ifaddr *ifa)); void in_ifscrub __P((struct ifnet *, struct in_ifaddr *)); +int ipflow_fastforward __P((struct mbuf *)); +void ipflow_create __P((const struct route *, struct mbuf *)); +void ipflow_slowtimo __P((void)); #endif /* KERNEL */ Index: netinet/ip_fw.c =================================================================== RCS file: /home/ncvs/src/sys/netinet/ip_fw.c,v retrieving revision 1.82 diff -u -r1.82 ip_fw.c --- ip_fw.c 1998/04/21 18:54:53 1.82 +++ ip_fw.c 1998/05/03 23:32:13 @@ -36,6 +36,7 @@ #include #include #include +#include #include #include #include Index: netinet/ip_input.c =================================================================== RCS file: /home/ncvs/src/sys/netinet/ip_input.c,v retrieving revision 1.82 diff -u -r1.82 ip_input.c --- ip_input.c 1998/04/13 17:27:08 1.82 +++ ip_input.c 1998/05/03 23:32:27 @@ -41,6 +41,7 @@ #include "opt_ipfw.h" #include "opt_ipdivert.h" #include "opt_ipfilter.h" +#include "opt_ipflow.h" #include @@ -80,7 +81,7 @@ static int ip_rsvp_on; struct socket *ip_rsvpd; -static int ipforwarding = 0; +int ipforwarding = 0; SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW, &ipforwarding, 0, ""); @@ -878,6 +879,9 @@ } } } +#ifdef IPFLOW + ipflow_slowtimo(); +#endif splx(s); } @@ -1381,8 +1385,12 @@ if (type) ipstat.ips_redirectsent++; else { - if (mcopy) + if (mcopy) { +#ifdef IPFLOW + ipflow_create(&ipforward_rt, mcopy); +#endif m_freem(mcopy); + } return; } } Index: netinet/ip_var.h =================================================================== RCS file: /home/ncvs/src/sys/netinet/ip_var.h,v retrieving revision 1.34 diff -u -r1.34 ip_var.h --- ip_var.h 1997/09/07 05:26:46 1.34 +++ ip_var.h 1998/05/03 23:32:29 @@ -132,6 +132,7 @@ u_long ips_fragdropped; /* frags dropped (dups, out of space) */ u_long ips_fragtimeout; /* fragments timed out */ u_long ips_forward; /* packets forwarded */ + u_long ips_fastforward; /* packets fast forwarded */ u_long ips_cantforward; /* packets rcvd for unreachable dest */ u_long ips_redirectsent; /* packets forwarded on same net */ u_long ips_noproto; /* unknown or unsupported protocol */ @@ -150,6 +151,21 @@ u_long ips_notmember; /* multicasts for unregistered grps */ }; +#define IPFLOW_HASHBITS 6 /* should not be a multiple of 8 */ +struct ipflow { + LIST_ENTRY(ipflow) ipf_next; /* next ipflow in bucket */ + struct in_addr ipf_dst; /* destination address */ + struct in_addr ipf_src; /* source address */ + u_int8_t ipf_tos; /* type-of-service */ + struct route ipf_ro; /* associated route entry */ + u_long ipf_uses; /* number of uses in this period */ + u_long ipf_last_uses; /* number of uses in last period */ + u_long ipf_dropped; /* ENOBUFS returned by if_output */ + u_long ipf_errors; /* other errors returned by if_output */ + int ipf_timer; /* remaining lifetime of this entry */ + time_t ipf_start; /* creation time */ +}; + #ifdef KERNEL /* flags passed to ip_output as last parameter */ #define IP_FORWARDING 0x1 /* most of ip header exists */ @@ -163,6 +179,7 @@ extern struct ipstat ipstat; extern u_short ip_id; /* ip packet ctr, for ids */ extern int ip_defttl; /* default IP ttl */ +extern int ipforwarding; /* ip forwarding */ extern u_char ip_protox[]; extern struct socket *ip_rsvpd; /* reservation protocol daemon */ extern struct socket *ip_mrouter; /* multicast routing daemon */ --VS++wcV0S1rZb1Fb Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename="ip_flow.c" /* $NetBSD: ip_flow.c,v 1.1 1998/04/29 21:37:55 matt Exp $ */ /*- * Copyright (c) 1998 The NetBSD Foundation, Inc. * All rights reserved. * * This code is derived from software contributed to The NetBSD Foundation * by the 3am Software Foundry ("3am"). It was developed by Matt Thomas. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the NetBSD * Foundation, Inc. and its contributors. * 4. Neither the name of The NetBSD Foundation nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define IPFLOW_TIMER (5 * PR_SLOWHZ) #define IPFLOW_HASHSIZE (1 << IPFLOW_HASHBITS) static LIST_HEAD(ipflowhead, ipflow) ipflows[IPFLOW_HASHSIZE]; static int ipflow_inuse; #define IPFLOW_MAX 256 static int ipflow_active = 0; SYSCTL_INT(_net_inet_ip, IPCTL_FASTFORWARDING, fastforwarding, CTLFLAG_RW, &ipflow_active, 0, ""); MALLOC_DEFINE(M_IPFLOW, "ip_flow", "IP flow"); static unsigned ipflow_hash( struct in_addr dst, struct in_addr src, unsigned tos) { unsigned hash = tos; int idx; for (idx = 0; idx < 32; idx += IPFLOW_HASHBITS) hash += (dst.s_addr >> (32 - idx)) + (src.s_addr >> idx); return hash & (IPFLOW_HASHSIZE-1); } static struct ipflow * ipflow_lookup( const struct ip *ip) { unsigned hash; struct ipflow *ipf; hash = ipflow_hash(ip->ip_dst, ip->ip_src, ip->ip_tos); ipf = LIST_FIRST(&ipflows[hash]); while (ipf != NULL) { if (ip->ip_dst.s_addr == ipf->ipf_dst.s_addr && ip->ip_src.s_addr == ipf->ipf_src.s_addr && ip->ip_tos == ipf->ipf_tos) break; ipf = LIST_NEXT(ipf, ipf_next); } return ipf; } int ipflow_fastforward( struct mbuf *m) { struct ip *ip; struct ipflow *ipf; struct rtentry *rt; u_int32_t sum; int error; /* * Are we forwarding packets? Big enough for an IP packet? */ if (!ipforwarding || !ipflow_active || m->m_len < sizeof(struct ip)) return 0; /* * IP header with no option and valid version and length */ ip = mtod(m, struct ip *); if (ip->ip_v != IPVERSION || ip->ip_hl != (sizeof(struct ip) >> 2) || ntohs(ip->ip_len) > m->m_pkthdr.len) return 0; /* * Find a flow. */ if ((ipf = ipflow_lookup(ip)) == NULL) return 0; /* * Route and interface still up? */ rt = ipf->ipf_ro.ro_rt; if ((rt->rt_flags & RTF_UP) == 0 || (rt->rt_ifp->if_flags & IFF_UP) == 0) return 0; /* * Packet size OK? TTL? */ if (m->m_pkthdr.len > rt->rt_ifp->if_mtu || ip->ip_ttl <= IPTTLDEC) return 0; /* * Everything checks out and so we can forward this packet. * Modify the TTL and incrementally change the checksum. * On little endian machine, the TTL is in LSB position * (so we can simply add) while on big-endian it's in the * MSB position (so we have to do two calculation; the first * is the add and second is to wrap the results into 17 bits, * 16 bits and a carry). */ ip->ip_ttl -= IPTTLDEC; #if BYTE_ORDER == LITTLE_ENDIAN sum = ip->ip_sum + IPTTLDEC; #endif #if BYTE_ORDER == BIG_ENDIAN sum = ip->ip_sum + (IPTTLDEC << 8); sum = (sum & 0xFFFF) + (sum >> 16); #endif if (sum > 0x10000) /* add in carry if needed */ sum++; ip->ip_sum = sum; /* bit 16 is dropped */ /* * Send the packet on its way. All we can get back is ENOBUFS */ ipf->ipf_uses++; ipf->ipf_timer = IPFLOW_TIMER; if ((error = (*rt->rt_ifp->if_output)(rt->rt_ifp, m, &ipf->ipf_ro.ro_dst, rt)) != 0) { if (error == ENOBUFS) ipf->ipf_dropped++; else ipf->ipf_errors++; } return 1; } static void ipflow_addstats( struct ipflow *ipf) { ipf->ipf_ro.ro_rt->rt_use += ipf->ipf_uses; ipstat.ips_cantforward += ipf->ipf_errors + ipf->ipf_dropped; ipstat.ips_forward += ipf->ipf_uses; ipstat.ips_fastforward += ipf->ipf_uses; } static void ipflow_free( struct ipflow *ipf) { int s; /* * Remove the flow from the hash table (at elevated IPL). * Once it's off the list, we can deal with it at normal * network IPL. */ s = splimp(); LIST_REMOVE(ipf, ipf_next); splx(s); ipflow_addstats(ipf); RTFREE(ipf->ipf_ro.ro_rt); ipflow_inuse--; FREE(ipf, M_IPFLOW); } static struct ipflow * ipflow_reap( void) { struct ipflow *ipf, *maybe_ipf = NULL; int idx; int s; for (idx = 0; idx < IPFLOW_HASHSIZE; idx++) { ipf = LIST_FIRST(&ipflows[idx]); while (ipf != NULL) { /* * If this no longer points to a valid route * reclaim it. */ if ((ipf->ipf_ro.ro_rt->rt_flags & RTF_UP) == 0) goto done; /* * choose the one that's been least recently used * or has had the least uses in the last 1.5 * intervals. */ if (ipf == NULL || ipf->ipf_timer < maybe_ipf->ipf_timer || (ipf->ipf_timer == maybe_ipf->ipf_timer && ipf->ipf_last_uses + ipf->ipf_uses < maybe_ipf->ipf_last_uses + maybe_ipf->ipf_uses)) maybe_ipf = ipf; ipf = LIST_NEXT(ipf, ipf_next); } } ipf = maybe_ipf; done: /* * Remove the entry from the flow table. */ s = splimp(); LIST_REMOVE(ipf, ipf_next); splx(s); ipflow_addstats(ipf); RTFREE(ipf->ipf_ro.ro_rt); return ipf; } void ipflow_slowtimo( void) { struct ipflow *ipf; int idx; for (idx = 0; idx < IPFLOW_HASHSIZE; idx++) { ipf = LIST_FIRST(&ipflows[idx]); while (ipf != NULL) { struct ipflow *next_ipf = LIST_NEXT(ipf, ipf_next); if (--ipf->ipf_timer == 0) { ipflow_free(ipf); } else { ipf->ipf_last_uses = ipf->ipf_uses; ipf->ipf_ro.ro_rt->rt_use += ipf->ipf_uses; ipstat.ips_forward += ipf->ipf_uses; ipstat.ips_fastforward += ipf->ipf_uses; ipf->ipf_uses = 0; } ipf = next_ipf; } } } void ipflow_create( const struct route *ro, struct mbuf *m) { const struct ip *const ip = mtod(m, struct ip *); struct ipflow *ipf; unsigned hash; int s; /* * Don't create cache entries for ICMP messages. */ if (!ipflow_active || ip->ip_p == IPPROTO_ICMP) return; /* * See if an existing flow struct exists. If so remove it from it's * list and free the old route. If not, try to malloc a new one * (if we aren't at our limit). */ ipf = ipflow_lookup(ip); if (ipf == NULL) { if (ipflow_inuse == IPFLOW_MAX) { ipf = ipflow_reap(); } else { ipf = (struct ipflow *) malloc(sizeof(*ipf), M_IPFLOW, M_NOWAIT); if (ipf == NULL) return; ipflow_inuse++; } bzero((caddr_t) ipf, sizeof(*ipf)); } else { s = splimp(); LIST_REMOVE(ipf, ipf_next); splx(s); ipflow_addstats(ipf); RTFREE(ipf->ipf_ro.ro_rt); ipf->ipf_uses = ipf->ipf_last_uses = 0; ipf->ipf_errors = ipf->ipf_dropped = 0; } /* * Fill in the updated information. */ ipf->ipf_ro = *ro; ro->ro_rt->rt_refcnt++; ipf->ipf_dst = ip->ip_dst; ipf->ipf_src = ip->ip_src; ipf->ipf_tos = ip->ip_tos; ipf->ipf_timer = IPFLOW_TIMER; ipf->ipf_start = time_second; /* * Insert into the approriate bucket of the flow table. */ hash = ipflow_hash(ip->ip_dst, ip->ip_src, ip->ip_tos); s = splimp(); LIST_INSERT_HEAD(&ipflows[hash], ipf, ipf_next); splx(s); } --VS++wcV0S1rZb1Fb-- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message