From e4a8065910cd6b2e7e0448cc6431ca2906322389 Mon Sep 17 00:00:00 2001 From: Sebastian Huber Date: Thu, 9 Aug 2018 13:04:41 +0200 Subject: Update to FreeBSD head 2017-10-01 Git mirror commit b2f0376b45428f13151d229c5ae9d4d8f74acbd1. Update #3472. --- freebsd/sys/netinet/if_ether.c | 2 +- freebsd/sys/netinet/in_pcb.c | 32 +++++++++++- freebsd/sys/netinet/in_pcb.h | 1 + freebsd/sys/netinet/ip_icmp.c | 65 +++++++++++++++---------- freebsd/sys/netinet/sctp.h | 1 - freebsd/sys/netinet/sctp_asconf.c | 33 ------------- freebsd/sys/netinet/sctp_asconf.h | 3 -- freebsd/sys/netinet/sctp_constants.h | 9 +--- freebsd/sys/netinet/sctp_indata.c | 26 +++++----- freebsd/sys/netinet/sctp_input.c | 43 +++++++++++------ freebsd/sys/netinet/sctp_os_bsd.h | 5 -- freebsd/sys/netinet/sctp_output.c | 36 +++++++------- freebsd/sys/netinet/sctp_output.h | 2 +- freebsd/sys/netinet/sctp_pcb.c | 12 ++--- freebsd/sys/netinet/sctp_pcb.h | 4 -- freebsd/sys/netinet/sctp_sysctl.c | 8 +++ freebsd/sys/netinet/sctp_uio.h | 4 +- freebsd/sys/netinet/sctp_usrreq.c | 21 ++++++-- freebsd/sys/netinet/sctputil.c | 94 ++++++++++++------------------------ freebsd/sys/netinet/tcp_input.c | 38 +++++++-------- freebsd/sys/netinet/tcp_syncache.c | 10 ++-- freebsd/sys/netinet/tcp_timer.c | 49 ++++++------------- freebsd/sys/netinet/tcp_usrreq.c | 2 + freebsd/sys/netinet/tcp_var.h | 7 ++- 24 files changed, 243 insertions(+), 264 deletions(-) (limited to 'freebsd/sys/netinet') diff --git a/freebsd/sys/netinet/if_ether.c b/freebsd/sys/netinet/if_ether.c index 802fedbb..bdb3b054 100644 --- a/freebsd/sys/netinet/if_ether.c +++ b/freebsd/sys/netinet/if_ether.c @@ -1281,7 +1281,7 @@ arp_mark_lle_reachable(struct llentry *la) } /* - * Add pernament link-layer record for given interface address. + * Add permanent link-layer record for given interface address. */ static __noinline void arp_add_ifa_lle(struct ifnet *ifp, const struct sockaddr *dst) diff --git a/freebsd/sys/netinet/in_pcb.c b/freebsd/sys/netinet/in_pcb.c index 3d43ed92..44890e3f 100644 --- a/freebsd/sys/netinet/in_pcb.c +++ b/freebsd/sys/netinet/in_pcb.c @@ -2808,6 +2808,35 @@ in_pcbquery_txrtlmt(struct inpcb *inp, uint32_t *p_max_pacing_rate) return (error); } +/* + * Query existing TX queue level based on the existing + * "inp->inp_snd_tag", if any. + */ +int +in_pcbquery_txrlevel(struct inpcb *inp, uint32_t *p_txqueue_level) +{ + union if_snd_tag_query_params params = { }; + struct m_snd_tag *mst; + struct ifnet *ifp; + int error; + + mst = inp->inp_snd_tag; + if (mst == NULL) + return (EINVAL); + + ifp = mst->ifp; + if (ifp == NULL) + return (EINVAL); + + if (ifp->if_snd_tag_query == NULL) + return (EOPNOTSUPP); + + error = ifp->if_snd_tag_query(mst, ¶ms); + if (error == 0 && p_txqueue_level != NULL) + *p_txqueue_level = params.rate_limit.queue_level; + return (error); +} + /* * Allocate a new TX rate limit send tag from the network interface * given by the "ifp" argument and save it in "inp->inp_snd_tag": @@ -2817,7 +2846,8 @@ in_pcbattach_txrtlmt(struct inpcb *inp, struct ifnet *ifp, uint32_t flowtype, uint32_t flowid, uint32_t max_pacing_rate) { union if_snd_tag_alloc_params params = { - .rate_limit.hdr.type = IF_SND_TAG_TYPE_RATE_LIMIT, + .rate_limit.hdr.type = (max_pacing_rate == -1U) ? + IF_SND_TAG_TYPE_UNLIMITED : IF_SND_TAG_TYPE_RATE_LIMIT, .rate_limit.hdr.flowid = flowid, .rate_limit.hdr.flowtype = flowtype, .rate_limit.max_rate = max_pacing_rate, diff --git a/freebsd/sys/netinet/in_pcb.h b/freebsd/sys/netinet/in_pcb.h index 42fd23d0..059ea90c 100644 --- a/freebsd/sys/netinet/in_pcb.h +++ b/freebsd/sys/netinet/in_pcb.h @@ -761,6 +761,7 @@ int in_pcbattach_txrtlmt(struct inpcb *, struct ifnet *, uint32_t, uint32_t, uin void in_pcbdetach_txrtlmt(struct inpcb *); int in_pcbmodify_txrtlmt(struct inpcb *, uint32_t); int in_pcbquery_txrtlmt(struct inpcb *, uint32_t *); +int in_pcbquery_txrlevel(struct inpcb *, uint32_t *); void in_pcboutput_txrtlmt(struct inpcb *, struct ifnet *, struct mbuf *); void in_pcboutput_eagain(struct inpcb *); #endif diff --git a/freebsd/sys/netinet/ip_icmp.c b/freebsd/sys/netinet/ip_icmp.c index acc2e6b6..4843efcd 100644 --- a/freebsd/sys/netinet/ip_icmp.c +++ b/freebsd/sys/netinet/ip_icmp.c @@ -187,17 +187,14 @@ kmod_icmpstat_inc(int statnum) void icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu) { - struct ip *oip = mtod(n, struct ip *), *nip; - unsigned oiphlen = oip->ip_hl << 2; + struct ip *oip, *nip; struct icmp *icp; struct mbuf *m; - unsigned icmplen, icmpelen, nlen; + unsigned icmplen, icmpelen, nlen, oiphlen; + + KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", + __func__)); - KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__)); -#ifdef ICMPPRINTFS - if (icmpprintfs) - printf("icmp_error(%p, %x, %d)\n", oip, type, code); -#endif if (type != ICMP_REDIRECT) ICMPSTAT_INC(icps_error); /* @@ -209,19 +206,28 @@ icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu) */ if (n->m_flags & M_DECRYPTED) goto freeit; - if (oip->ip_off & htons(~(IP_MF|IP_DF))) - goto freeit; if (n->m_flags & (M_BCAST|M_MCAST)) goto freeit; + + /* Drop if IP header plus 8 bytes is not contiguous in first mbuf. */ + if (n->m_len < sizeof(struct ip) + ICMP_MINLEN) + goto freeit; + oip = mtod(n, struct ip *); + oiphlen = oip->ip_hl << 2; + if (n->m_len < oiphlen + ICMP_MINLEN) + goto freeit; +#ifdef ICMPPRINTFS + if (icmpprintfs) + printf("icmp_error(%p, %x, %d)\n", oip, type, code); +#endif + if (oip->ip_off & htons(~(IP_MF|IP_DF))) + goto freeit; if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && - n->m_len >= oiphlen + ICMP_MINLEN && - !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) { + !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + + oiphlen))->icmp_type)) { ICMPSTAT_INC(icps_oldicmp); goto freeit; } - /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */ - if (oiphlen + 8 > n->m_len) - goto freeit; /* * Calculate length to quote from original packet and * prevent the ICMP mbuf from overflowing. @@ -237,9 +243,10 @@ icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu) n->m_next == NULL) goto stdreply; if (n->m_len < oiphlen + sizeof(struct tcphdr) && - ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL)) + (n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL) goto freeit; - th = (struct tcphdr *)((caddr_t)oip + oiphlen); + oip = mtod(n, struct ip *); + th = mtodo(n, oiphlen); tcphlen = th->th_off << 2; if (tcphlen < sizeof(struct tcphdr)) goto freeit; @@ -247,8 +254,8 @@ icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu) goto freeit; if (oiphlen + tcphlen > n->m_len && n->m_next == NULL) goto stdreply; - if (n->m_len < oiphlen + tcphlen && - ((n = m_pullup(n, oiphlen + tcphlen)) == NULL)) + if (n->m_len < oiphlen + tcphlen && + (n = m_pullup(n, oiphlen + tcphlen)) == NULL) goto freeit; icmpelen = max(tcphlen, min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen)); @@ -264,24 +271,31 @@ icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu) if (n->m_len < oiphlen + sizeof(struct sctphdr) && (n = m_pullup(n, oiphlen + sizeof(struct sctphdr))) == NULL) goto freeit; + oip = mtod(n, struct ip *); icmpelen = max(sizeof(struct sctphdr), min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen)); - sh = (struct sctphdr *)((caddr_t)oip + oiphlen); + sh = mtodo(n, oiphlen); if (ntohl(sh->v_tag) == 0 && - ntohs(oip->ip_len) >= oiphlen + sizeof(struct sctphdr) + 8 && + ntohs(oip->ip_len) >= oiphlen + + sizeof(struct sctphdr) + 8 && (n->m_len >= oiphlen + sizeof(struct sctphdr) + 8 || n->m_next != NULL)) { if (n->m_len < oiphlen + sizeof(struct sctphdr) + 8 && - (n = m_pullup(n, oiphlen + sizeof(struct sctphdr) + 8)) == NULL) + (n = m_pullup(n, oiphlen + + sizeof(struct sctphdr) + 8)) == NULL) goto freeit; + oip = mtod(n, struct ip *); + sh = mtodo(n, oiphlen); ch = (struct sctp_chunkhdr *)(sh + 1); if (ch->chunk_type == SCTP_INITIATION) { icmpelen = max(sizeof(struct sctphdr) + 8, - min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen)); + min(V_icmp_quotelen, ntohs(oip->ip_len) - + oiphlen)); } } } else -stdreply: icmpelen = max(8, min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen)); +stdreply: icmpelen = max(8, min(V_icmp_quotelen, ntohs(oip->ip_len) - + oiphlen)); icmplen = min(oiphlen + icmpelen, nlen); if (icmplen < sizeof(struct ip)) @@ -296,7 +310,8 @@ stdreply: icmpelen = max(8, min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen)); #ifdef MAC mac_netinet_icmp_reply(n, m); #endif - icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN); + icmplen = min(icmplen, M_TRAILINGSPACE(m) - + sizeof(struct ip) - ICMP_MINLEN); m_align(m, ICMP_MINLEN + icmplen); m->m_len = ICMP_MINLEN + icmplen; diff --git a/freebsd/sys/netinet/sctp.h b/freebsd/sys/netinet/sctp.h index ec42cffa..578be4f6 100644 --- a/freebsd/sys/netinet/sctp.h +++ b/freebsd/sys/netinet/sctp.h @@ -545,7 +545,6 @@ struct sctp_error_auth_invalid_hmac { #define SCTP_PCB_FLAGS_INTERLEAVE_STRMS 0x0000000000000010 #define SCTP_PCB_FLAGS_DO_ASCONF 0x0000000000000020 #define SCTP_PCB_FLAGS_AUTO_ASCONF 0x0000000000000040 -#define SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE 0x0000000000000080 /* socket options */ #define SCTP_PCB_FLAGS_NODELAY 0x0000000000000100 #define SCTP_PCB_FLAGS_AUTOCLOSE 0x0000000000000200 diff --git a/freebsd/sys/netinet/sctp_asconf.c b/freebsd/sys/netinet/sctp_asconf.c index aa31e488..5f1c4e8f 100644 --- a/freebsd/sys/netinet/sctp_asconf.c +++ b/freebsd/sys/netinet/sctp_asconf.c @@ -2307,39 +2307,6 @@ sctp_set_primary_ip_address_sa(struct sctp_tcb *stcb, struct sockaddr *sa) return (0); } -void -sctp_set_primary_ip_address(struct sctp_ifa *ifa) -{ - struct sctp_inpcb *inp; - - /* go through all our PCB's */ - LIST_FOREACH(inp, &SCTP_BASE_INFO(listhead), sctp_list) { - struct sctp_tcb *stcb; - - /* process for all associations for this endpoint */ - LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { - /* queue an ASCONF:SET_PRIM_ADDR to be sent */ - if (!sctp_asconf_queue_add(stcb, ifa, - SCTP_SET_PRIM_ADDR)) { - /* set primary queuing succeeded */ - SCTPDBG(SCTP_DEBUG_ASCONF1, "set_primary_ip_address: queued on stcb=%p, ", - (void *)stcb); - SCTPDBG_ADDR(SCTP_DEBUG_ASCONF1, &ifa->address.sa); - if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || - (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { -#ifdef SCTP_TIMER_BASED_ASCONF - sctp_timer_start(SCTP_TIMER_TYPE_ASCONF, - stcb->sctp_ep, stcb, - stcb->asoc.primary_destination); -#else - sctp_send_asconf(stcb, NULL, SCTP_ADDR_NOT_LOCKED); -#endif - } - } - } /* for each stcb */ - } /* for each inp */ -} - int sctp_is_addr_pending(struct sctp_tcb *stcb, struct sctp_ifa *sctp_ifa) { diff --git a/freebsd/sys/netinet/sctp_asconf.h b/freebsd/sys/netinet/sctp_asconf.h index c8d670e5..ed9160e1 100644 --- a/freebsd/sys/netinet/sctp_asconf.h +++ b/freebsd/sys/netinet/sctp_asconf.h @@ -72,9 +72,6 @@ extern int32_t sctp_set_primary_ip_address_sa(struct sctp_tcb *, struct sockaddr *); -extern void - sctp_set_primary_ip_address(struct sctp_ifa *ifa); - extern void sctp_check_address_list(struct sctp_tcb *, struct mbuf *, int, int, struct sockaddr *, uint16_t, uint16_t, uint16_t, uint16_t); diff --git a/freebsd/sys/netinet/sctp_constants.h b/freebsd/sys/netinet/sctp_constants.h index dca34cc5..56772b34 100644 --- a/freebsd/sys/netinet/sctp_constants.h +++ b/freebsd/sys/netinet/sctp_constants.h @@ -555,11 +555,9 @@ __FBSDID("$FreeBSD$"); #define SCTP_TIMER_TYPE_INPKILL 15 #define SCTP_TIMER_TYPE_ASOCKILL 16 #define SCTP_TIMER_TYPE_ADDR_WQ 17 -#define SCTP_TIMER_TYPE_ZERO_COPY 18 -#define SCTP_TIMER_TYPE_ZCOPY_SENDQ 19 -#define SCTP_TIMER_TYPE_PRIM_DELETED 20 +#define SCTP_TIMER_TYPE_PRIM_DELETED 18 /* add new timers here - and increment LAST */ -#define SCTP_TIMER_TYPE_LAST 21 +#define SCTP_TIMER_TYPE_LAST 19 #define SCTP_IS_TIMER_TYPE_VALID(t) (((t) > SCTP_TIMER_TYPE_NONE) && \ ((t) < SCTP_TIMER_TYPE_LAST)) @@ -971,9 +969,6 @@ __FBSDID("$FreeBSD$"); #define SCTP_SO_NOT_LOCKED 0 -#define SCTP_HOLDS_LOCK 1 -#define SCTP_NOT_LOCKED 0 - /*- * For address locks, do we hold the lock? */ diff --git a/freebsd/sys/netinet/sctp_indata.c b/freebsd/sys/netinet/sctp_indata.c index 1924aeab..c27dd485 100644 --- a/freebsd/sys/netinet/sctp_indata.c +++ b/freebsd/sys/netinet/sctp_indata.c @@ -1381,10 +1381,10 @@ sctp_queue_data_for_reasm(struct sctp_tcb *stcb, struct sctp_association *asoc, } if (sctp_place_control_in_stream(strm, asoc, control)) { /* Duplicate SSN? */ - sctp_clean_up_control(stcb, control); sctp_abort_in_reasm(stcb, control, chk, abort_flag, SCTP_FROM_SCTP_INDATA + SCTP_LOC_6); + sctp_clean_up_control(stcb, control); return; } if ((tsn == (asoc->cumulative_tsn + 1) && (asoc->idata_supported == 0))) { @@ -1596,9 +1596,17 @@ sctp_queue_data_for_reasm(struct sctp_tcb *stcb, struct sctp_association *asoc, next_fsn, control->fsn_included); TAILQ_REMOVE(&control->reasm, at, sctp_next); lenadded = sctp_add_chk_to_control(control, strm, stcb, asoc, at, SCTP_READ_LOCK_NOT_HELD); - asoc->size_on_all_streams += lenadded; if (control->on_read_q) { do_wakeup = 1; + } else { + /* + * We only add to the + * size-on-all-streams if its not on + * the read q. The read q flag will + * cause a sballoc so its accounted + * for there. + */ + asoc->size_on_all_streams += lenadded; } next_fsn++; if (control->end_added && control->pdapi_started) { @@ -1611,7 +1619,6 @@ sctp_queue_data_for_reasm(struct sctp_tcb *stcb, struct sctp_association *asoc, control, &stcb->sctp_socket->so_rcv, control->end_added, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); - do_wakeup = 1; } break; } @@ -4207,8 +4214,6 @@ sctp_express_handle_sack(struct sctp_tcb *stcb, uint32_t cumack, again: j = 0; TAILQ_FOREACH(net, &asoc->nets, sctp_next) { - int to_ticks; - if (win_probe_recovery && (net->window_probe)) { win_probe_recovered = 1; /* @@ -4224,15 +4229,9 @@ again: } } } - if (net->RTO == 0) { - to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); - } else { - to_ticks = MSEC_TO_TICKS(net->RTO); - } if (net->flight_size) { j++; - (void)SCTP_OS_TIMER_START(&net->rxt_timer.timer, to_ticks, - sctp_timeout_handler, &net->rxt_timer); + sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net); if (net->window_probe) { net->window_probe = 0; } @@ -4244,8 +4243,7 @@ again: */ net->window_probe = 0; if (!SCTP_OS_TIMER_PENDING(&net->rxt_timer.timer)) { - SCTP_OS_TIMER_START(&net->rxt_timer.timer, to_ticks, - sctp_timeout_handler, &net->rxt_timer); + sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net); } } else if (SCTP_OS_TIMER_PENDING(&net->rxt_timer.timer)) { sctp_timer_stop(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, diff --git a/freebsd/sys/netinet/sctp_input.c b/freebsd/sys/netinet/sctp_input.c index 2e844fb5..0871bdb8 100644 --- a/freebsd/sys/netinet/sctp_input.c +++ b/freebsd/sys/netinet/sctp_input.c @@ -199,8 +199,7 @@ sctp_handle_init(struct mbuf *m, int iphlen, int offset, sctp_send_initiate_ack(inp, stcb, net, m, iphlen, offset, src, dst, sh, cp, mflowtype, mflowid, - vrf_id, port, - ((stcb == NULL) ? SCTP_HOLDS_LOCK : SCTP_NOT_LOCKED)); + vrf_id, port); } outnow: if (stcb == NULL) { @@ -702,6 +701,14 @@ sctp_handle_nat_colliding_state(struct sctp_tcb *stcb) */ struct sctpasochead *head; + if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_WAIT) || + (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_ECHOED)) { + atomic_add_int(&stcb->asoc.refcnt, 1); + SCTP_TCB_UNLOCK(stcb); + SCTP_INP_INFO_WLOCK(); + SCTP_TCB_LOCK(stcb); + atomic_subtract_int(&stcb->asoc.refcnt, 1); + } if (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_WAIT) { /* generate a new vtag and send init */ LIST_REMOVE(stcb, sctp_asocs); @@ -713,6 +720,7 @@ sctp_handle_nat_colliding_state(struct sctp_tcb *stcb) */ LIST_INSERT_HEAD(head, stcb, sctp_asocs); sctp_send_initiate(stcb->sctp_ep, stcb, SCTP_SO_NOT_LOCKED); + SCTP_INP_INFO_WUNLOCK(); return (1); } if (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_ECHOED) { @@ -734,6 +742,7 @@ sctp_handle_nat_colliding_state(struct sctp_tcb *stcb) */ LIST_INSERT_HEAD(head, stcb, sctp_asocs); sctp_send_initiate(stcb->sctp_ep, stcb, SCTP_SO_NOT_LOCKED); + SCTP_INP_INFO_WUNLOCK(); return (1); } return (0); @@ -2555,15 +2564,12 @@ sctp_handle_cookie_echo(struct mbuf *m, int iphlen, int offset, /* Expire time is in Ticks, so we convert to seconds */ time_expires.tv_sec = cookie->time_entered.tv_sec + TICKS_TO_SEC(cookie->cookie_life); time_expires.tv_usec = cookie->time_entered.tv_usec; - /* - * TODO sctp_constants.h needs alternative time macros when _KERNEL - * is undefined. - */ if (timevalcmp(&now, &time_expires, >)) { /* cookie is stale! */ struct mbuf *op_err; struct sctp_error_stale_cookie *cause; - uint32_t tim; + struct timeval diff; + uint32_t staleness; op_err = sctp_get_mbuf_for_msg(sizeof(struct sctp_error_stale_cookie), 0, M_NOWAIT, 1, MT_DATA); @@ -2577,12 +2583,19 @@ sctp_handle_cookie_echo(struct mbuf *m, int iphlen, int offset, cause->cause.code = htons(SCTP_CAUSE_STALE_COOKIE); cause->cause.length = htons((sizeof(struct sctp_paramhdr) + (sizeof(uint32_t)))); - /* seconds to usec */ - tim = (now.tv_sec - time_expires.tv_sec) * 1000000; - /* add in usec */ - if (tim == 0) - tim = now.tv_usec - cookie->time_entered.tv_usec; - cause->stale_time = htonl(tim); + diff = now; + timevalsub(&diff, &time_expires); + if (diff.tv_sec > UINT32_MAX / 1000000) { + staleness = UINT32_MAX; + } else { + staleness = diff.tv_sec * 1000000; + } + if (UINT32_MAX - staleness >= (uint32_t)diff.tv_usec) { + staleness += diff.tv_usec; + } else { + staleness = UINT32_MAX; + } + cause->stale_time = htonl(staleness); sctp_send_operr_to(src, dst, sh, cookie->peers_vtag, op_err, mflowtype, mflowid, l_inp->fibnum, vrf_id, port); @@ -3135,14 +3148,14 @@ sctp_handle_ecn_cwr(struct sctp_cwr_chunk *cp, struct sctp_tcb *stcb, struct sct * make sure that we have a covered ECNE in the control chunk part. * If so remove it. */ - struct sctp_tmit_chunk *chk; + struct sctp_tmit_chunk *chk, *nchk; struct sctp_ecne_chunk *ecne; int override; uint32_t cwr_tsn; cwr_tsn = ntohl(cp->tsn); override = cp->ch.chunk_flags & SCTP_CWR_REDUCE_OVERRIDE; - TAILQ_FOREACH(chk, &stcb->asoc.control_send_queue, sctp_next) { + TAILQ_FOREACH_SAFE(chk, &stcb->asoc.control_send_queue, sctp_next, nchk) { if (chk->rec.chunk_id.id != SCTP_ECN_ECHO) { continue; } diff --git a/freebsd/sys/netinet/sctp_os_bsd.h b/freebsd/sys/netinet/sctp_os_bsd.h index 194a03ed..045ccc38 100644 --- a/freebsd/sys/netinet/sctp_os_bsd.h +++ b/freebsd/sys/netinet/sctp_os_bsd.h @@ -404,11 +404,6 @@ typedef struct rtentry sctp_rtentry_t; #define SCTP_RTALLOC(ro, vrf_id, fibnum) \ rtalloc_ign_fib((struct route *)ro, 0UL, fibnum) -/* Future zero copy wakeup/send function */ -#define SCTP_ZERO_COPY_EVENT(inp, so) -/* This is re-pulse ourselves for sendbuf */ -#define SCTP_ZERO_COPY_SENDQ_EVENT(inp, so) - /* * SCTP protocol specific mbuf flags. */ diff --git a/freebsd/sys/netinet/sctp_output.c b/freebsd/sys/netinet/sctp_output.c index 5e2fbbc2..2540b5b9 100644 --- a/freebsd/sys/netinet/sctp_output.c +++ b/freebsd/sys/netinet/sctp_output.c @@ -4279,11 +4279,13 @@ sctp_lowlevel_chunk_output(struct sctp_inpcb *inp, uint32_t mtu; mtu = SCTP_GATHER_MTU_FROM_ROUTE(net->ro._s_addr, &net->ro._l_addr.sa, ro->ro_rt); - if (net->port) { - mtu -= sizeof(struct udphdr); - } - if (mtu && (stcb->asoc.smallest_mtu > mtu)) { - sctp_mtu_size_reset(inp, &stcb->asoc, mtu); + if (mtu > 0) { + if (net->port) { + mtu -= sizeof(struct udphdr); + } + if ((stcb != NULL) && (stcb->asoc.smallest_mtu > mtu)) { + sctp_mtu_size_reset(inp, &stcb->asoc, mtu); + } net->mtu = mtu; } } else if (ro->ro_rt == NULL) { @@ -4635,13 +4637,14 @@ sctp_lowlevel_chunk_output(struct sctp_inpcb *inp, uint32_t mtu; mtu = SCTP_GATHER_MTU_FROM_ROUTE(net->ro._s_addr, &net->ro._l_addr.sa, ro->ro_rt); - if (mtu && - (stcb->asoc.smallest_mtu > mtu)) { - sctp_mtu_size_reset(inp, &stcb->asoc, mtu); - net->mtu = mtu; + if (mtu > 0) { if (net->port) { - net->mtu -= sizeof(struct udphdr); + mtu -= sizeof(struct udphdr); + } + if ((stcb != NULL) && (stcb->asoc.smallest_mtu > mtu)) { + sctp_mtu_size_reset(inp, &stcb->asoc, mtu); } + net->mtu = mtu; } } else if (ifp) { if (ND_IFINFO(ifp)->linkmtu && @@ -5490,7 +5493,7 @@ sctp_send_initiate_ack(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sockaddr *src, struct sockaddr *dst, struct sctphdr *sh, struct sctp_init_chunk *init_chk, uint8_t mflowtype, uint32_t mflowid, - uint32_t vrf_id, uint16_t port, int hold_inp_lock) + uint32_t vrf_id, uint16_t port) { struct sctp_association *asoc; struct mbuf *m, *m_tmp, *m_last, *m_cookie, *op_err; @@ -5838,10 +5841,6 @@ do_a_abort: } else { uint32_t vtag, itsn; - if (hold_inp_lock) { - SCTP_INP_INCR_REF(inp); - SCTP_INP_RUNLOCK(inp); - } if (asoc) { atomic_add_int(&asoc->refcnt, 1); SCTP_TCB_UNLOCK(stcb); @@ -5861,12 +5860,12 @@ do_a_abort: SCTP_TCB_LOCK(stcb); atomic_add_int(&asoc->refcnt, -1); } else { + SCTP_INP_INCR_REF(inp); + SCTP_INP_RUNLOCK(inp); vtag = sctp_select_a_tag(inp, inp->sctp_lport, sh->src_port, 1); initack->init.initiate_tag = htonl(vtag); /* get a TSN to use too */ initack->init.initial_tsn = htonl(sctp_select_initial_TSN(&inp->sctp_ep)); - } - if (hold_inp_lock) { SCTP_INP_RLOCK(inp); SCTP_INP_DECR_REF(inp); } @@ -12423,7 +12422,6 @@ sctp_copy_it_in(struct sctp_tcb *stcb, resv_in_first = SCTP_DATA_CHUNK_OVERHEAD(stcb); sp->data = sp->tail_mbuf = NULL; if (sp->length == 0) { - *error = 0; goto skip_copy; } if (srcv->sinfo_keynumber_valid) { @@ -13156,7 +13154,7 @@ skip_preblock: if (strm->last_msg_incomplete == 0) { do_a_copy_in: sp = sctp_copy_it_in(stcb, asoc, srcv, uio, net, max_len, user_marks_eor, &error); - if ((sp == NULL) || (error)) { + if (error) { goto out; } SCTP_TCB_SEND_LOCK(stcb); diff --git a/freebsd/sys/netinet/sctp_output.h b/freebsd/sys/netinet/sctp_output.h index 32f968c1..0a5cf10b 100644 --- a/freebsd/sys/netinet/sctp_output.h +++ b/freebsd/sys/netinet/sctp_output.h @@ -86,7 +86,7 @@ sctp_send_initiate_ack(struct sctp_inpcb *, struct sctp_tcb *, struct sockaddr *, struct sockaddr *, struct sctphdr *, struct sctp_init_chunk *, uint8_t, uint32_t, - uint32_t, uint16_t, int); + uint32_t, uint16_t); struct mbuf * sctp_arethere_unrecognized_parameters(struct mbuf *, int, int *, diff --git a/freebsd/sys/netinet/sctp_pcb.c b/freebsd/sys/netinet/sctp_pcb.c index 1907991e..46a91110 100644 --- a/freebsd/sys/netinet/sctp_pcb.c +++ b/freebsd/sys/netinet/sctp_pcb.c @@ -711,12 +711,11 @@ sctp_add_addr_to_vrf(uint32_t vrf_id, void *ifn, uint32_t ifn_index, SCTP_WQ_ADDR_LOCK(); LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); - SCTP_WQ_ADDR_UNLOCK(); - sctp_timer_start(SCTP_TIMER_TYPE_ADDR_WQ, (struct sctp_inpcb *)NULL, (struct sctp_tcb *)NULL, (struct sctp_nets *)NULL); + SCTP_WQ_ADDR_UNLOCK(); } else { /* it's ready for use */ sctp_ifap->localifa_flags &= ~SCTP_ADDR_DEFER_USE; @@ -823,12 +822,11 @@ out_now: * the newest first :-0 */ LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); - SCTP_WQ_ADDR_UNLOCK(); - sctp_timer_start(SCTP_TIMER_TYPE_ADDR_WQ, (struct sctp_inpcb *)NULL, (struct sctp_tcb *)NULL, (struct sctp_nets *)NULL); + SCTP_WQ_ADDR_UNLOCK(); } return; } @@ -2488,7 +2486,9 @@ sctp_inpcb_alloc(struct socket *so, uint32_t vrf_id) inp->sctp_flags = (SCTP_PCB_FLAGS_TCPTYPE | SCTP_PCB_FLAGS_UNBOUND); /* Be sure we have blocking IO by default */ + SOCK_LOCK(so); SCTP_CLEAR_SO_NBIO(so); + SOCK_UNLOCK(so); } else { /* * unsupported socket type (RAW, etc)- in case we missed it @@ -2921,9 +2921,9 @@ sctp_inpcb_bind(struct socket *so, struct sockaddr *addr, */ /* got to be root to get at low ports */ if (ntohs(lport) < IPPORT_RESERVED) { - if (p && (error = + if ((p != NULL) && ((error = priv_check(p, PRIV_NETINET_RESERVEDPORT) - )) { + ) != 0)) { SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); SCTP_INP_INFO_WUNLOCK(); diff --git a/freebsd/sys/netinet/sctp_pcb.h b/freebsd/sys/netinet/sctp_pcb.h index 86fae1f4..acc68311 100644 --- a/freebsd/sys/netinet/sctp_pcb.h +++ b/freebsd/sys/netinet/sctp_pcb.h @@ -314,10 +314,6 @@ struct sctp_pcb { */ struct sctp_timer signature_change; - /* Zero copy full buffer timer */ - struct sctp_timer zero_copy_timer; - /* Zero copy app to transport (sendq) read repulse timer */ - struct sctp_timer zero_copy_sendq_timer; uint32_t def_cookie_life; /* defaults to 0 */ int auto_close_time; diff --git a/freebsd/sys/netinet/sctp_sysctl.c b/freebsd/sys/netinet/sctp_sysctl.c index a2364431..52361602 100644 --- a/freebsd/sys/netinet/sctp_sysctl.c +++ b/freebsd/sys/netinet/sctp_sysctl.c @@ -508,6 +508,14 @@ sctp_sysctl_handle_assoclist(SYSCTL_HANDLER_ARGS) xraddr.rtt = net->rtt / 1000; xraddr.heartbeat_interval = net->heart_beat_delay; xraddr.ssthresh = net->ssthresh; + xraddr.encaps_port = net->port; + if (net->dest_state & SCTP_ADDR_UNCONFIRMED) { + xraddr.state = SCTP_UNCONFIRMED; + } else if (net->dest_state & SCTP_ADDR_REACHABLE) { + xraddr.state = SCTP_ACTIVE; + } else { + xraddr.state = SCTP_INACTIVE; + } xraddr.start_time.tv_sec = (uint32_t)net->start_time.tv_sec; xraddr.start_time.tv_usec = (uint32_t)net->start_time.tv_usec; SCTP_INP_RUNLOCK(inp); diff --git a/freebsd/sys/netinet/sctp_uio.h b/freebsd/sys/netinet/sctp_uio.h index d6055163..daa6847d 100644 --- a/freebsd/sys/netinet/sctp_uio.h +++ b/freebsd/sys/netinet/sctp_uio.h @@ -1239,7 +1239,9 @@ struct xsctp_raddr { uint32_t rtt; uint32_t heartbeat_interval; uint32_t ssthresh; - uint32_t extra_padding[30]; /* future */ + uint16_t encaps_port; + uint16_t state; + uint32_t extra_padding[29]; /* future */ }; #define SCTP_MAX_LOGGING_SIZE 30000 diff --git a/freebsd/sys/netinet/sctp_usrreq.c b/freebsd/sys/netinet/sctp_usrreq.c index fa136d4e..6ab04456 100644 --- a/freebsd/sys/netinet/sctp_usrreq.c +++ b/freebsd/sys/netinet/sctp_usrreq.c @@ -1961,6 +1961,7 @@ flags_out: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { id->assoc_value = stcb->asoc.vrf_id; + SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_assoc_value); } break; @@ -3065,7 +3066,6 @@ flags_out: if (event_type > 0) { if (stcb) { event->se_on = sctp_stcb_is_feature_on(inp, stcb, event_type); - SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || @@ -3079,6 +3079,9 @@ flags_out: } } } + if (stcb != NULL) { + SCTP_TCB_UNLOCK(stcb); + } if (error == 0) { *optsize = sizeof(struct sctp_event); } @@ -3649,12 +3652,16 @@ flags_out: sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[0]; #endif - SCTP_TCB_UNLOCK(stcb); - *optsize = sizeof(struct sctp_prstatus); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } + if (stcb != NULL) { + SCTP_TCB_UNLOCK(stcb); + } + if (error == 0) { + *optsize = sizeof(struct sctp_prstatus); + } break; } case SCTP_PR_ASSOC_STATUS: @@ -3677,12 +3684,16 @@ flags_out: sprstat->sprstat_abandoned_unsent = stcb->asoc.abandoned_unsent[policy]; sprstat->sprstat_abandoned_sent = stcb->asoc.abandoned_sent[policy]; } - SCTP_TCB_UNLOCK(stcb); - *optsize = sizeof(struct sctp_prstatus); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } + if (stcb != NULL) { + SCTP_TCB_UNLOCK(stcb); + } + if (error == 0) { + *optsize = sizeof(struct sctp_prstatus); + } break; } case SCTP_MAX_CWND: diff --git a/freebsd/sys/netinet/sctputil.c b/freebsd/sys/netinet/sctputil.c index 75c989d3..f8fd23f1 100644 --- a/freebsd/sys/netinet/sctputil.c +++ b/freebsd/sys/netinet/sctputil.c @@ -1464,13 +1464,11 @@ sctp_handle_addr_wq(void) LIST_INIT(&asc->list_of_work); asc->cnt = 0; - SCTP_WQ_ADDR_LOCK(); LIST_FOREACH_SAFE(wi, &SCTP_BASE_INFO(addr_wq), sctp_nxt_addr, nwi) { LIST_REMOVE(wi, sctp_nxt_addr); LIST_INSERT_HEAD(&asc->list_of_work, wi, sctp_nxt_addr); asc->cnt++; } - SCTP_WQ_ADDR_UNLOCK(); if (asc->cnt == 0) { SCTP_FREE(asc, SCTP_M_ASC_IT); @@ -1494,11 +1492,9 @@ sctp_handle_addr_wq(void) if (SCTP_BASE_VAR(sctp_pcb_initialized) == 0) { sctp_asconf_iterator_end(asc, 0); } else { - SCTP_WQ_ADDR_LOCK(); LIST_FOREACH(wi, &asc->list_of_work, sctp_nxt_addr) { LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); } - SCTP_WQ_ADDR_UNLOCK(); SCTP_FREE(asc, SCTP_M_ASC_IT); } } @@ -1567,8 +1563,7 @@ sctp_timeout_handler(void *t) (tmr->type != SCTP_TIMER_TYPE_SHUTDOWN) && (tmr->type != SCTP_TIMER_TYPE_SHUTDOWNACK) && (tmr->type != SCTP_TIMER_TYPE_SHUTDOWNGUARD) && - (tmr->type != SCTP_TIMER_TYPE_ASOCKILL)) - ) { + (tmr->type != SCTP_TIMER_TYPE_ASOCKILL))) { SCTP_INP_DECR_REF(inp); CURVNET_RESTORE(); return; @@ -1614,6 +1609,12 @@ sctp_timeout_handler(void *t) CURVNET_RESTORE(); return; } + } else if (inp != NULL) { + if (type != SCTP_TIMER_TYPE_INPKILL) { + SCTP_INP_WLOCK(inp); + } + } else { + SCTP_WQ_ADDR_LOCK(); } /* record in stopped what t-o occurred */ tmr->stopped_from = type; @@ -1635,22 +1636,6 @@ sctp_timeout_handler(void *t) /* call the handler for the appropriate timer type */ switch (type) { - case SCTP_TIMER_TYPE_ZERO_COPY: - if (inp == NULL) { - break; - } - if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { - SCTP_ZERO_COPY_EVENT(inp, inp->sctp_socket); - } - break; - case SCTP_TIMER_TYPE_ZCOPY_SENDQ: - if (inp == NULL) { - break; - } - if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { - SCTP_ZERO_COPY_SENDQ_EVENT(inp, inp->sctp_socket); - } - break; case SCTP_TIMER_TYPE_ADDR_WQ: sctp_handle_addr_wq(); break; @@ -1777,7 +1762,6 @@ sctp_timeout_handler(void *t) } SCTP_STAT_INCR(sctps_timosecret); (void)SCTP_GETTIME_TIMEVAL(&tv); - SCTP_INP_WLOCK(inp); inp->sctp_ep.time_of_secret_change = tv.tv_sec; inp->sctp_ep.last_secret_number = inp->sctp_ep.current_secret_number; @@ -1791,7 +1775,6 @@ sctp_timeout_handler(void *t) inp->sctp_ep.secret_key[secret][i] = sctp_select_initial_TSN(&inp->sctp_ep); } - SCTP_INP_WUNLOCK(inp); sctp_timer_start(SCTP_TIMER_TYPE_NEWCOOKIE, inp, stcb, net); } did_output = 0; @@ -1939,7 +1922,12 @@ sctp_timeout_handler(void *t) get_out: if (stcb) { SCTP_TCB_UNLOCK(stcb); + } else if (inp != NULL) { + SCTP_INP_WUNLOCK(inp); + } else { + SCTP_WQ_ADDR_UNLOCK(); } + out_decr: if (inp) { SCTP_INP_DECR_REF(inp); @@ -1964,14 +1952,6 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, SCTP_TCB_LOCK_ASSERT(stcb); } switch (t_type) { - case SCTP_TIMER_TYPE_ZERO_COPY: - tmr = &inp->sctp_ep.zero_copy_timer; - to_ticks = SCTP_ZERO_COPY_TICK_DELAY; - break; - case SCTP_TIMER_TYPE_ZCOPY_SENDQ: - tmr = &inp->sctp_ep.zero_copy_sendq_timer; - to_ticks = SCTP_ZERO_COPY_SENDQ_TICK_DELAY; - break; case SCTP_TIMER_TYPE_ADDR_WQ: /* Only 1 tick away :-) */ tmr = &SCTP_BASE_INFO(addr_wq_timer); @@ -2253,12 +2233,6 @@ sctp_timer_stop(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, SCTP_TCB_LOCK_ASSERT(stcb); } switch (t_type) { - case SCTP_TIMER_TYPE_ZERO_COPY: - tmr = &inp->sctp_ep.zero_copy_timer; - break; - case SCTP_TIMER_TYPE_ZCOPY_SENDQ: - tmr = &inp->sctp_ep.zero_copy_sendq_timer; - break; case SCTP_TIMER_TYPE_ADDR_WQ: tmr = &SCTP_BASE_INFO(addr_wq_timer); break; @@ -4472,36 +4446,32 @@ sctp_wakeup_the_read_socket(struct sctp_inpcb *inp, ) { if ((inp != NULL) && (inp->sctp_socket != NULL)) { - if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { - SCTP_ZERO_COPY_EVENT(inp, inp->sctp_socket); - } else { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) - struct socket *so; + struct socket *so; - so = SCTP_INP_SO(inp); - if (!so_locked) { - if (stcb) { - atomic_add_int(&stcb->asoc.refcnt, 1); - SCTP_TCB_UNLOCK(stcb); - } - SCTP_SOCKET_LOCK(so, 1); - if (stcb) { - SCTP_TCB_LOCK(stcb); - atomic_subtract_int(&stcb->asoc.refcnt, 1); - } - if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { - SCTP_SOCKET_UNLOCK(so, 1); - return; - } + so = SCTP_INP_SO(inp); + if (!so_locked) { + if (stcb) { + atomic_add_int(&stcb->asoc.refcnt, 1); + SCTP_TCB_UNLOCK(stcb); } -#endif - sctp_sorwakeup(inp, inp->sctp_socket); -#if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) - if (!so_locked) { + SCTP_SOCKET_LOCK(so, 1); + if (stcb) { + SCTP_TCB_LOCK(stcb); + atomic_subtract_int(&stcb->asoc.refcnt, 1); + } + if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { SCTP_SOCKET_UNLOCK(so, 1); + return; } + } #endif + sctp_sorwakeup(inp, inp->sctp_socket); +#if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) + if (!so_locked) { + SCTP_SOCKET_UNLOCK(so, 1); } +#endif } } @@ -6182,11 +6152,11 @@ sctp_dynamic_set_primary(struct sockaddr *sa, uint32_t vrf_id) * newest first :-0 */ LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); - SCTP_WQ_ADDR_UNLOCK(); sctp_timer_start(SCTP_TIMER_TYPE_ADDR_WQ, (struct sctp_inpcb *)NULL, (struct sctp_tcb *)NULL, (struct sctp_nets *)NULL); + SCTP_WQ_ADDR_UNLOCK(); return (0); } diff --git a/freebsd/sys/netinet/tcp_input.c b/freebsd/sys/netinet/tcp_input.c index 2195a778..663b8bc2 100644 --- a/freebsd/sys/netinet/tcp_input.c +++ b/freebsd/sys/netinet/tcp_input.c @@ -1695,25 +1695,6 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so, if (TSTMP_GT(to.to_tsecr, tcp_ts_getticks())) to.to_tsecr = 0; } - /* - * If timestamps were negotiated during SYN/ACK they should - * appear on every segment during this session and vice versa. - */ - if ((tp->t_flags & TF_RCVD_TSTMP) && !(to.to_flags & TOF_TS)) { - if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { - log(LOG_DEBUG, "%s; %s: Timestamp missing, " - "no action\n", s, __func__); - free(s, M_TCPLOG); - } - } - if (!(tp->t_flags & TF_RCVD_TSTMP) && (to.to_flags & TOF_TS)) { - if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { - log(LOG_DEBUG, "%s; %s: Timestamp not expected, " - "no action\n", s, __func__); - free(s, M_TCPLOG); - } - } - /* * Process options only when we get SYN/ACK back. The SYN case * for incoming connections is handled in tcp_syncache. @@ -1744,6 +1725,25 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so, tp->t_flags &= ~TF_SACK_PERMIT; } + /* + * If timestamps were negotiated during SYN/ACK they should + * appear on every segment during this session and vice versa. + */ + if ((tp->t_flags & TF_RCVD_TSTMP) && !(to.to_flags & TOF_TS)) { + if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { + log(LOG_DEBUG, "%s; %s: Timestamp missing, " + "no action\n", s, __func__); + free(s, M_TCPLOG); + } + } + if (!(tp->t_flags & TF_RCVD_TSTMP) && (to.to_flags & TOF_TS)) { + if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { + log(LOG_DEBUG, "%s; %s: Timestamp not expected, " + "no action\n", s, __func__); + free(s, M_TCPLOG); + } + } + /* * Header prediction: check for the two common cases * of a uni-directional data xfer. If the packet has diff --git a/freebsd/sys/netinet/tcp_syncache.c b/freebsd/sys/netinet/tcp_syncache.c index cf1f0ad1..a30def41 100644 --- a/freebsd/sys/netinet/tcp_syncache.c +++ b/freebsd/sys/netinet/tcp_syncache.c @@ -1644,9 +1644,7 @@ syncache_respond(struct syncache *sc, struct syncache_head *sch, int locked, tlen = hlen + sizeof(struct tcphdr); /* Determine MSS we advertize to other end of connection. */ - mssopt = tcp_mssopt(&sc->sc_inc); - if (sc->sc_peer_mss) - mssopt = max( min(sc->sc_peer_mss, mssopt), V_tcp_minmss); + mssopt = max(tcp_mssopt(&sc->sc_inc), V_tcp_minmss); /* XXX: Assume that the entire packet will fit in a header mbuf. */ KASSERT(max_linkhdr + tlen + TCP_MAXOLEN <= MHLEN, @@ -1995,7 +1993,7 @@ syncookie_mac(struct in_conninfo *inc, tcp_seq irs, uint8_t flags, static tcp_seq syncookie_generate(struct syncache_head *sch, struct syncache *sc) { - u_int i, mss, secbit, wscale; + u_int i, secbit, wscale; uint32_t iss, hash; uint8_t *secbits; union syncookie cookie; @@ -2005,8 +2003,8 @@ syncookie_generate(struct syncache_head *sch, struct syncache *sc) cookie.cookie = 0; /* Map our computed MSS into the 3-bit index. */ - mss = min(tcp_mssopt(&sc->sc_inc), max(sc->sc_peer_mss, V_tcp_minmss)); - for (i = nitems(tcp_sc_msstab) - 1; tcp_sc_msstab[i] > mss && i > 0; + for (i = nitems(tcp_sc_msstab) - 1; + tcp_sc_msstab[i] > sc->sc_peer_mss && i > 0; i--) ; cookie.flags.mss_idx = i; diff --git a/freebsd/sys/netinet/tcp_timer.c b/freebsd/sys/netinet/tcp_timer.c index ebd9e852..b4528801 100644 --- a/freebsd/sys/netinet/tcp_timer.c +++ b/freebsd/sys/netinet/tcp_timer.c @@ -150,29 +150,6 @@ SYSCTL_INT(_net_inet_tcp, OID_AUTO, pmtud_blackhole_detection, &VNET_NAME(tcp_pmtud_blackhole_detect), 0, "Path MTU Discovery Black Hole Detection Enabled"); -static VNET_DEFINE(int, tcp_pmtud_blackhole_activated); -#define V_tcp_pmtud_blackhole_activated \ - VNET(tcp_pmtud_blackhole_activated) -SYSCTL_INT(_net_inet_tcp, OID_AUTO, pmtud_blackhole_activated, - CTLFLAG_RD|CTLFLAG_VNET, - &VNET_NAME(tcp_pmtud_blackhole_activated), 0, - "Path MTU Discovery Black Hole Detection, Activation Count"); - -static VNET_DEFINE(int, tcp_pmtud_blackhole_activated_min_mss); -#define V_tcp_pmtud_blackhole_activated_min_mss \ - VNET(tcp_pmtud_blackhole_activated_min_mss) -SYSCTL_INT(_net_inet_tcp, OID_AUTO, pmtud_blackhole_activated_min_mss, - CTLFLAG_RD|CTLFLAG_VNET, - &VNET_NAME(tcp_pmtud_blackhole_activated_min_mss), 0, - "Path MTU Discovery Black Hole Detection, Activation Count at min MSS"); - -static VNET_DEFINE(int, tcp_pmtud_blackhole_failed); -#define V_tcp_pmtud_blackhole_failed VNET(tcp_pmtud_blackhole_failed) -SYSCTL_INT(_net_inet_tcp, OID_AUTO, pmtud_blackhole_failed, - CTLFLAG_RD|CTLFLAG_VNET, - &VNET_NAME(tcp_pmtud_blackhole_failed), 0, - "Path MTU Discovery Black Hole Detection, Failure Count"); - #ifdef INET static VNET_DEFINE(int, tcp_pmtud_blackhole_mss) = 1200; #define V_tcp_pmtud_blackhole_mss VNET(tcp_pmtud_blackhole_mss) @@ -751,18 +728,20 @@ tcp_timer_rexmt(void * xtp) */ if (((tp->t_flags2 & (TF2_PLPMTU_PMTUD|TF2_PLPMTU_MAXSEGSNT)) == (TF2_PLPMTU_PMTUD|TF2_PLPMTU_MAXSEGSNT)) && - (tp->t_rxtshift >= 2 && tp->t_rxtshift % 2 == 0)) { + (tp->t_rxtshift >= 2 && tp->t_rxtshift < 6 && + tp->t_rxtshift % 2 == 0)) { /* * Enter Path MTU Black-hole Detection mechanism: * - Disable Path MTU Discovery (IP "DF" bit). * - Reduce MTU to lower value than what we * negotiated with peer. */ - /* Record that we may have found a black hole. */ - tp->t_flags2 |= TF2_PLPMTU_BLACKHOLE; - - /* Keep track of previous MSS. */ - tp->t_pmtud_saved_maxseg = tp->t_maxseg; + if ((tp->t_flags2 & TF2_PLPMTU_BLACKHOLE) == 0) { + /* Record that we may have found a black hole. */ + tp->t_flags2 |= TF2_PLPMTU_BLACKHOLE; + /* Keep track of previous MSS. */ + tp->t_pmtud_saved_maxseg = tp->t_maxseg; + } /* * Reduce the MSS to blackhole value or to the default @@ -774,7 +753,7 @@ tcp_timer_rexmt(void * xtp) tp->t_maxseg > V_tcp_v6pmtud_blackhole_mss) { /* Use the sysctl tuneable blackhole MSS. */ tp->t_maxseg = V_tcp_v6pmtud_blackhole_mss; - V_tcp_pmtud_blackhole_activated++; + TCPSTAT_INC(tcps_pmtud_blackhole_activated); } else if (isipv6) { /* Use the default MSS. */ tp->t_maxseg = V_tcp_v6mssdflt; @@ -783,7 +762,7 @@ tcp_timer_rexmt(void * xtp) * minmss. */ tp->t_flags2 &= ~TF2_PLPMTU_PMTUD; - V_tcp_pmtud_blackhole_activated_min_mss++; + TCPSTAT_INC(tcps_pmtud_blackhole_activated_min_mss); } #endif #if defined(INET6) && defined(INET) @@ -793,7 +772,7 @@ tcp_timer_rexmt(void * xtp) if (tp->t_maxseg > V_tcp_pmtud_blackhole_mss) { /* Use the sysctl tuneable blackhole MSS. */ tp->t_maxseg = V_tcp_pmtud_blackhole_mss; - V_tcp_pmtud_blackhole_activated++; + TCPSTAT_INC(tcps_pmtud_blackhole_activated); } else { /* Use the default MSS. */ tp->t_maxseg = V_tcp_mssdflt; @@ -802,7 +781,7 @@ tcp_timer_rexmt(void * xtp) * minmss. */ tp->t_flags2 &= ~TF2_PLPMTU_PMTUD; - V_tcp_pmtud_blackhole_activated_min_mss++; + TCPSTAT_INC(tcps_pmtud_blackhole_activated_min_mss); } #endif /* @@ -821,11 +800,11 @@ tcp_timer_rexmt(void * xtp) * stage (1448, 1188, 524) 2 chances to recover. */ if ((tp->t_flags2 & TF2_PLPMTU_BLACKHOLE) && - (tp->t_rxtshift > 6)) { + (tp->t_rxtshift >= 6)) { tp->t_flags2 |= TF2_PLPMTU_PMTUD; tp->t_flags2 &= ~TF2_PLPMTU_BLACKHOLE; tp->t_maxseg = tp->t_pmtud_saved_maxseg; - V_tcp_pmtud_blackhole_failed++; + TCPSTAT_INC(tcps_pmtud_blackhole_failed); /* * Reset the slow-start flight size as it * may depend on the new MSS. diff --git a/freebsd/sys/netinet/tcp_usrreq.c b/freebsd/sys/netinet/tcp_usrreq.c index 198291f2..88cc3501 100644 --- a/freebsd/sys/netinet/tcp_usrreq.c +++ b/freebsd/sys/netinet/tcp_usrreq.c @@ -1920,6 +1920,8 @@ tcp_attach(struct socket *so) #ifdef INET6 if (inp->inp_vflag & INP_IPV6PROTO) { inp->inp_vflag |= INP_IPV6; + if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) + inp->inp_vflag |= INP_IPV4; inp->in6p_hops = -1; /* use kernel default */ } else diff --git a/freebsd/sys/netinet/tcp_var.h b/freebsd/sys/netinet/tcp_var.h index 2e2f88ac..967a5139 100644 --- a/freebsd/sys/netinet/tcp_var.h +++ b/freebsd/sys/netinet/tcp_var.h @@ -580,6 +580,11 @@ struct tcpstat { uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ + /* Path MTU Discovery Black Hole Detection related stats */ + uint64_t tcps_pmtud_blackhole_activated; /* Black Hole Count */ + uint64_t tcps_pmtud_blackhole_activated_min_mss; /* BH at min MSS Count */ + uint64_t tcps_pmtud_blackhole_failed; /* Black Hole Failure Count */ + uint64_t _pad[12]; /* 6 UTO, 6 TBD */ }; @@ -650,7 +655,7 @@ struct tcp_hhook_data { struct xtcpcb { size_t xt_len; /* length of this structure */ struct xinpcb xt_inp; - char xt_stack[TCP_FUNCTION_NAME_LEN_MAX]; /* (n) */ + char xt_stack[TCP_FUNCTION_NAME_LEN_MAX]; /* (s) */ int64_t spare64[8]; int32_t t_state; /* (s,p) */ uint32_t t_flags; /* (s,p) */ -- cgit v1.2.3