From e51171019bb0e1f9fb57c25bd2e38ce652eaea27 Mon Sep 17 00:00:00 2001 From: YOSHIFUJI Hideaki Date: Thu, 29 May 2008 19:55:05 +0900 Subject: [SCTP]: Fix NULL dereference of asoc. Commit 7cbca67c073263c179f605bdbbdc565ab29d801d ("[IPV6]: Support Source Address Selection API (RFC5014)") introduced NULL dereference of asoc to sctp_v6_get_saddr in net/sctp/ipv6.c. Pointed out by Johann Felix Soden . Signed-off-by: YOSHIFUJI Hideaki --- include/net/sctp/structs.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/net/sctp') diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h index 0ce0443c5b7..917d425f054 100644 --- a/include/net/sctp/structs.h +++ b/include/net/sctp/structs.h @@ -548,7 +548,8 @@ struct sctp_af { struct dst_entry *(*get_dst) (struct sctp_association *asoc, union sctp_addr *daddr, union sctp_addr *saddr); - void (*get_saddr) (struct sctp_association *asoc, + void (*get_saddr) (struct sctp_sock *sk, + struct sctp_association *asoc, struct dst_entry *dst, union sctp_addr *daddr, union sctp_addr *saddr); -- cgit v1.2.3 From a6465234814efda9ed1dccdba852953f7508e827 Mon Sep 17 00:00:00 2001 From: Vlad Yasevich Date: Wed, 4 Jun 2008 12:38:43 -0700 Subject: sctp: Correctly implement Fast Recovery cwnd manipulations. Correctly keep track of Fast Recovery state and do not reduce congestion window multiple times during sucht state. Signed-off-by: Vlad Yasevich Tested-by: Wei Yongjun Signed-off-by: David S. Miller --- include/net/sctp/structs.h | 8 +++++++- net/sctp/transport.c | 44 ++++++++++++++++++++++++++++++++------------ 2 files changed, 39 insertions(+), 13 deletions(-) (limited to 'include/net/sctp') diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h index 917d425f054..67592072a32 100644 --- a/include/net/sctp/structs.h +++ b/include/net/sctp/structs.h @@ -902,7 +902,10 @@ struct sctp_transport { * calculation completes (i.e. the DATA chunk * is SACK'd) clear this flag. */ - int rto_pending; + __u8 rto_pending; + + /* Flag to track the current fast recovery state */ + __u8 fast_recovery; /* * These are the congestion stats. @@ -921,6 +924,9 @@ struct sctp_transport { /* Data that has been sent, but not acknowledged. */ __u32 flight_size; + /* TSN marking the fast recovery exit point */ + __u32 fast_recovery_exit; + /* Destination */ struct dst_entry *dst; /* Source address. */ diff --git a/net/sctp/transport.c b/net/sctp/transport.c index 62082e7b797..9647fb27722 100644 --- a/net/sctp/transport.c +++ b/net/sctp/transport.c @@ -79,6 +79,7 @@ static struct sctp_transport *sctp_transport_init(struct sctp_transport *peer, peer->rttvar = 0; peer->srtt = 0; peer->rto_pending = 0; + peer->fast_recovery = 0; peer->last_time_heard = jiffies; peer->last_time_used = jiffies; @@ -403,11 +404,16 @@ void sctp_transport_raise_cwnd(struct sctp_transport *transport, cwnd = transport->cwnd; flight_size = transport->flight_size; + /* See if we need to exit Fast Recovery first */ + if (transport->fast_recovery && + TSN_lte(transport->fast_recovery_exit, sack_ctsn)) + transport->fast_recovery = 0; + /* The appropriate cwnd increase algorithm is performed if, and only - * if the cumulative TSN has advanced and the congestion window is + * if the cumulative TSN whould advanced and the congestion window is * being fully utilized. */ - if ((transport->asoc->ctsn_ack_point >= sack_ctsn) || + if (TSN_lte(sack_ctsn, transport->asoc->ctsn_ack_point) || (flight_size < cwnd)) return; @@ -416,17 +422,23 @@ void sctp_transport_raise_cwnd(struct sctp_transport *transport, pmtu = transport->asoc->pathmtu; if (cwnd <= ssthresh) { - /* RFC 2960 7.2.1, sctpimpguide-05 2.14.2 When cwnd is less - * than or equal to ssthresh an SCTP endpoint MUST use the - * slow start algorithm to increase cwnd only if the current - * congestion window is being fully utilized and an incoming - * SACK advances the Cumulative TSN Ack Point. Only when these - * two conditions are met can the cwnd be increased otherwise - * the cwnd MUST not be increased. If these conditions are met - * then cwnd MUST be increased by at most the lesser of - * 1) the total size of the previously outstanding DATA - * chunk(s) acknowledged, and 2) the destination's path MTU. + /* RFC 4960 7.2.1 + * o When cwnd is less than or equal to ssthresh, an SCTP + * endpoint MUST use the slow-start algorithm to increase + * cwnd only if the current congestion window is being fully + * utilized, an incoming SACK advances the Cumulative TSN + * Ack Point, and the data sender is not in Fast Recovery. + * Only when these three conditions are met can the cwnd be + * increased; otherwise, the cwnd MUST not be increased. + * If these conditions are met, then cwnd MUST be increased + * by, at most, the lesser of 1) the total size of the + * previously outstanding DATA chunk(s) acknowledged, and + * 2) the destination's path MTU. This upper bound protects + * against the ACK-Splitting attack outlined in [SAVAGE99]. */ + if (transport->fast_recovery) + return; + if (bytes_acked > pmtu) cwnd += pmtu; else @@ -502,6 +514,13 @@ void sctp_transport_lower_cwnd(struct sctp_transport *transport, * cwnd = ssthresh * partial_bytes_acked = 0 */ + if (transport->fast_recovery) + return; + + /* Mark Fast recovery */ + transport->fast_recovery = 1; + transport->fast_recovery_exit = transport->asoc->next_tsn - 1; + transport->ssthresh = max(transport->cwnd/2, 4*transport->asoc->pathmtu); transport->cwnd = transport->ssthresh; @@ -586,6 +605,7 @@ void sctp_transport_reset(struct sctp_transport *t) t->flight_size = 0; t->error_count = 0; t->rto_pending = 0; + t->fast_recovery = 0; /* Initialize the state information for SFR-CACC */ t->cacc.changeover_active = 0; -- cgit v1.2.3 From 62aeaff5ccd96462b7077046357a6d7886175a57 Mon Sep 17 00:00:00 2001 From: Vlad Yasevich Date: Wed, 4 Jun 2008 12:39:11 -0700 Subject: sctp: Start T3-RTX timer when fast retransmitting lowest TSN When we are trying to fast retransmit the lowest outstanding TSN, we need to restart the T3-RTX timer, so that subsequent timeouts will correctly tag all the packets necessary for retransmissions. Signed-off-by: Vlad Yasevich Tested-by: Wei Yongjun Signed-off-by: David S. Miller --- include/net/sctp/structs.h | 5 ++++- net/sctp/outqueue.c | 42 +++++++++++++++++++++++++++++++----------- net/sctp/transport.c | 4 ++-- 3 files changed, 37 insertions(+), 14 deletions(-) (limited to 'include/net/sctp') diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h index 67592072a32..714dc43c034 100644 --- a/include/net/sctp/structs.h +++ b/include/net/sctp/structs.h @@ -1051,7 +1051,7 @@ void sctp_transport_route(struct sctp_transport *, union sctp_addr *, struct sctp_sock *); void sctp_transport_pmtu(struct sctp_transport *); void sctp_transport_free(struct sctp_transport *); -void sctp_transport_reset_timers(struct sctp_transport *); +void sctp_transport_reset_timers(struct sctp_transport *, int); void sctp_transport_hold(struct sctp_transport *); void sctp_transport_put(struct sctp_transport *); void sctp_transport_update_rto(struct sctp_transport *, __u32); @@ -1141,6 +1141,9 @@ struct sctp_outq { /* How many unackd bytes do we have in-flight? */ __u32 outstanding_bytes; + /* Are we doing fast-rtx on this queue */ + char fast_rtx; + /* Corked? */ char cork; diff --git a/net/sctp/outqueue.c b/net/sctp/outqueue.c index 59edfd25a19..5d3c441e84d 100644 --- a/net/sctp/outqueue.c +++ b/net/sctp/outqueue.c @@ -208,6 +208,7 @@ void sctp_outq_init(struct sctp_association *asoc, struct sctp_outq *q) INIT_LIST_HEAD(&q->sacked); INIT_LIST_HEAD(&q->abandoned); + q->fast_rtx = 0; q->outstanding_bytes = 0; q->empty = 1; q->cork = 0; @@ -500,6 +501,7 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport, case SCTP_RTXR_FAST_RTX: SCTP_INC_STATS(SCTP_MIB_FAST_RETRANSMITS); sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_FAST_RTX); + q->fast_rtx = 1; break; case SCTP_RTXR_PMTUD: SCTP_INC_STATS(SCTP_MIB_PMTUD_RETRANSMITS); @@ -543,10 +545,13 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, sctp_xmit_t status; struct sctp_chunk *chunk, *chunk1; struct sctp_association *asoc; + int fast_rtx; int error = 0; + int timer = 0; asoc = q->asoc; lqueue = &q->retransmit; + fast_rtx = q->fast_rtx; /* RFC 2960 6.3.3 Handle T3-rtx Expiration * @@ -587,13 +592,12 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, switch (status) { case SCTP_XMIT_PMTU_FULL: /* Send this packet. */ - if ((error = sctp_packet_transmit(pkt)) == 0) - *start_timer = 1; + error = sctp_packet_transmit(pkt); /* If we are retransmitting, we should only * send a single packet. */ - if (rtx_timeout) { + if (rtx_timeout || fast_rtx) { list_add(lchunk, lqueue); lchunk = NULL; } @@ -603,8 +607,7 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, case SCTP_XMIT_RWND_FULL: /* Send this packet. */ - if ((error = sctp_packet_transmit(pkt)) == 0) - *start_timer = 1; + error = sctp_packet_transmit(pkt); /* Stop sending DATA as there is no more room * at the receiver. @@ -615,8 +618,7 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, case SCTP_XMIT_NAGLE_DELAY: /* Send this packet. */ - if ((error = sctp_packet_transmit(pkt)) == 0) - *start_timer = 1; + error = sctp_packet_transmit(pkt); /* Stop sending DATA because of nagle delay. */ list_add(lchunk, lqueue); @@ -635,7 +637,14 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, if (chunk->fast_retransmit > 0) chunk->fast_retransmit = -1; - *start_timer = 1; + /* Force start T3-rtx timer when fast retransmitting + * the earliest outstanding TSN + */ + if (!timer && fast_rtx && + ntohl(chunk->subh.data_hdr->tsn) == + asoc->ctsn_ack_point + 1) + timer = 2; + q->empty = 0; /* Retrieve a new chunk to bundle. */ @@ -643,12 +652,16 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, break; } + /* Set the timer if there were no errors */ + if (!error && !timer) + timer = 1; + /* If we are here due to a retransmit timeout or a fast * retransmit and if there are any chunks left in the retransmit * queue that could not fit in the PMTU sized packet, they need * to be marked as ineligible for a subsequent fast retransmit. */ - if (rtx_timeout && !lchunk) { + if (rtx_timeout && fast_rtx) { list_for_each_entry(chunk1, lqueue, transmitted_list) { if (chunk1->fast_retransmit > 0) chunk1->fast_retransmit = -1; @@ -656,6 +669,12 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt, } } + *start_timer = timer; + + /* Clear fast retransmit hint */ + if (fast_rtx) + q->fast_rtx = 0; + return error; } @@ -862,7 +881,8 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout) rtx_timeout, &start_timer); if (start_timer) - sctp_transport_reset_timers(transport); + sctp_transport_reset_timers(transport, + start_timer-1); /* This can happen on COOKIE-ECHO resend. Only * one chunk can get bundled with a COOKIE-ECHO. @@ -977,7 +997,7 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout) list_add_tail(&chunk->transmitted_list, &transport->transmitted); - sctp_transport_reset_timers(transport); + sctp_transport_reset_timers(transport, start_timer-1); q->empty = 0; diff --git a/net/sctp/transport.c b/net/sctp/transport.c index 9647fb27722..3f34f61221e 100644 --- a/net/sctp/transport.c +++ b/net/sctp/transport.c @@ -191,7 +191,7 @@ static void sctp_transport_destroy(struct sctp_transport *transport) /* Start T3_rtx timer if it is not already running and update the heartbeat * timer. This routine is called every time a DATA chunk is sent. */ -void sctp_transport_reset_timers(struct sctp_transport *transport) +void sctp_transport_reset_timers(struct sctp_transport *transport, int force) { /* RFC 2960 6.3.2 Retransmission Timer Rules * @@ -201,7 +201,7 @@ void sctp_transport_reset_timers(struct sctp_transport *transport) * address. */ - if (!timer_pending(&transport->T3_rtx_timer)) + if (force || !timer_pending(&transport->T3_rtx_timer)) if (!mod_timer(&transport->T3_rtx_timer, jiffies + transport->rto)) sctp_transport_hold(transport); -- cgit v1.2.3 From b9031d9d87b24e24cd32ea15b5f4220a1e8da909 Mon Sep 17 00:00:00 2001 From: Vlad Yasevich Date: Wed, 4 Jun 2008 12:40:15 -0700 Subject: sctp: Fix ECN markings for IPv6 Commit e9df2e8fd8fbc95c57dbd1d33dada66c4627b44c ("[IPV6]: Use appropriate sock tclass setting for routing lookup.") also changed the way that ECN capable transports mark this capability in IPv6. As a result, SCTP was not marking ECN capablity because the traffic class was never set. This patch brings back the markings for IPv6 traffic. Signed-off-by: Vlad Yasevich Signed-off-by: David S. Miller --- include/net/sctp/structs.h | 1 + net/sctp/ipv6.c | 6 ++++++ net/sctp/output.c | 2 +- net/sctp/protocol.c | 6 ++++++ 4 files changed, 14 insertions(+), 1 deletion(-) (limited to 'include/net/sctp') diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h index 714dc43c034..7f25195f985 100644 --- a/include/net/sctp/structs.h +++ b/include/net/sctp/structs.h @@ -588,6 +588,7 @@ struct sctp_af { int (*is_ce) (const struct sk_buff *sk); void (*seq_dump_addr)(struct seq_file *seq, union sctp_addr *addr); + void (*ecn_capable)(struct sock *sk); __u16 net_header_len; int sockaddr_len; sa_family_t sa_family; diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c index e4aac3266fc..a2f4d4d5159 100644 --- a/net/sctp/ipv6.c +++ b/net/sctp/ipv6.c @@ -727,6 +727,11 @@ static void sctp_v6_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr) seq_printf(seq, NIP6_FMT " ", NIP6(addr->v6.sin6_addr)); } +static void sctp_v6_ecn_capable(struct sock *sk) +{ + inet6_sk(sk)->tclass |= INET_ECN_ECT_0; +} + /* Initialize a PF_INET6 socket msg_name. */ static void sctp_inet6_msgname(char *msgname, int *addr_len) { @@ -997,6 +1002,7 @@ static struct sctp_af sctp_af_inet6 = { .skb_iif = sctp_v6_skb_iif, .is_ce = sctp_v6_is_ce, .seq_dump_addr = sctp_v6_seq_dump_addr, + .ecn_capable = sctp_v6_ecn_capable, .net_header_len = sizeof(struct ipv6hdr), .sockaddr_len = sizeof(struct sockaddr_in6), #ifdef CONFIG_COMPAT diff --git a/net/sctp/output.c b/net/sctp/output.c index cf4f9fb6819..6d45bae93b4 100644 --- a/net/sctp/output.c +++ b/net/sctp/output.c @@ -548,7 +548,7 @@ int sctp_packet_transmit(struct sctp_packet *packet) * Note: The works for IPv6 layer checks this bit too later * in transmission. See IP6_ECN_flow_xmit(). */ - INET_ECN_xmit(nskb->sk); + (*tp->af_specific->ecn_capable)(nskb->sk); /* Set up the IP options. */ /* BUG: not implemented diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c index 56bdaf7fc42..b435a193c5d 100644 --- a/net/sctp/protocol.c +++ b/net/sctp/protocol.c @@ -617,6 +617,11 @@ static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr) seq_printf(seq, "%d.%d.%d.%d ", NIPQUAD(addr->v4.sin_addr)); } +static void sctp_v4_ecn_capable(struct sock *sk) +{ + INET_ECN_xmit(sk); +} + /* Event handler for inet address addition/deletion events. * The sctp_local_addr_list needs to be protocted by a spin lock since * multiple notifiers (say IPv4 and IPv6) may be running at the same @@ -935,6 +940,7 @@ static struct sctp_af sctp_af_inet = { .skb_iif = sctp_v4_skb_iif, .is_ce = sctp_v4_is_ce, .seq_dump_addr = sctp_v4_seq_dump_addr, + .ecn_capable = sctp_v4_ecn_capable, .net_header_len = sizeof(struct iphdr), .sockaddr_len = sizeof(struct sockaddr_in), #ifdef CONFIG_COMPAT -- cgit v1.2.3