> -----Original Message----- > From: Joe Stringer [mailto:j...@ovn.org] > Sent: Tuesday, May 24, 2016 7:26 PM > To: Fischetti, Antonio <antonio.fische...@intel.com> > Cc: Daniele Di Proietto <diproiet...@vmware.com>; dev@openvswitch.org > Subject: Re: [ovs-dev] [PATCH v3 04/16] conntrack: New userspace > connection tracker. > > On 24 May 2016 at 07:19, Fischetti, Antonio > <antonio.fische...@intel.com> wrote: > > Hi Daniele, just a comment below. > > Apart from that, it looks good to me, thanks. > > > > Acked-by: Antonio Fischetti <antonio.fische...@intel.com> > > > >> -----Original Message----- > >> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of > Daniele > >> Di Proietto > >> Sent: Tuesday, May 17, 2016 1:56 AM > >> To: dev@openvswitch.org > >> Subject: [ovs-dev] [PATCH v3 04/16] conntrack: New userspace > >> connection tracker. > >> > >> This commit adds the conntrack module. > >> > >> It is a connection tracker that resides entirely in userspace. > Its > >> primary user will be the dpif-netdev datapath. > >> > >> The module main goal is to provide conntrack_execute(), which > offers > >> a > >> convenient interface to implement the datapath ct() action. > >> > >> The conntrack module uses two submodules to deal with the l4 > protocol > >> details (conntrack-other for UDP and ICMP, conntrack-tcp for TCP). > >> > >> The conntrack-tcp submodule implementation is adapted from > FreeBSD's > >> pf > >> subsystem, therefore it's BSD licensed. It has been slightly > altered > >> to > >> match the OVS coding style and to allow the pickup of already > >> established connections. > >> > >> Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> > >> --- > >> COPYING | 1 + > >> debian/copyright.in | 4 + > >> include/openvswitch/types.h | 4 + > >> lib/automake.mk | 5 + > >> lib/conntrack-other.c | 85 +++++ > >> lib/conntrack-private.h | 88 +++++ > >> lib/conntrack-tcp.c | 463 +++++++++++++++++++++++ > >> lib/conntrack.c | 883 > >> ++++++++++++++++++++++++++++++++++++++++++++ > >> lib/conntrack.h | 151 ++++++++ > >> lib/util.h | 9 + > >> 10 files changed, 1693 insertions(+) > >> create mode 100644 lib/conntrack-other.c > >> create mode 100644 lib/conntrack-private.h > >> create mode 100644 lib/conntrack-tcp.c > >> create mode 100644 lib/conntrack.c > >> create mode 100644 lib/conntrack.h > >> > >> diff --git a/COPYING b/COPYING > >> index 308e3ea..afb98b9 100644 > >> --- a/COPYING > >> +++ b/COPYING > >> @@ -25,6 +25,7 @@ License, version 2. > >> The following files are licensed under the 2-clause BSD license. > >> include/windows/getopt.h > >> lib/getopt_long.c > >> + lib/conntrack-tcp.c > >> > >> The following files are licensed under the 3-clause BSD-license > >> include/windows/netinet/icmp6.h > >> diff --git a/debian/copyright.in b/debian/copyright.in > >> index 57d007a..a15f4dd 100644 > >> --- a/debian/copyright.in > >> +++ b/debian/copyright.in > >> @@ -21,6 +21,9 @@ Upstream Copyright Holders: > >> Copyright (c) 2014 Michael Chapman > >> Copyright (c) 2014 WindRiver, Inc. > >> Copyright (c) 2014 Avaya, Inc. > >> + Copyright (c) 2001 Daniel Hartmeier > >> + Copyright (c) 2002 - 2008 Henning Brauer > >> + Copyright (c) 2012 Gleb Smirnoff <gleb...@freebsd.org> > >> > >> License: > >> > >> @@ -90,6 +93,7 @@ License: > >> lib/getopt_long.c > >> include/windows/getopt.h > >> datapath-windows/ovsext/Conntrack-tcp.c > >> + lib/conntrack-tcp.c > >> > >> * The following files are licensed under the 3-clause BSD-license > >> > >> diff --git a/include/openvswitch/types.h > >> b/include/openvswitch/types.h > >> index 5f3347d..d7e94a6 100644 > >> --- a/include/openvswitch/types.h > >> +++ b/include/openvswitch/types.h > >> @@ -107,6 +107,10 @@ static const ovs_u128 OVS_U128_MAX = { { > >> UINT32_MAX, UINT32_MAX, > >> UINT32_MAX, UINT32_MAX } > }; > >> static const ovs_be128 OVS_BE128_MAX OVS_UNUSED = { { > OVS_BE32_MAX, > >> OVS_BE32_MAX, > >> OVS_BE32_MAX, > >> OVS_BE32_MAX } }; > >> +static const ovs_u128 OVS_U128_MIN OVS_UNUSED = { {0, 0, 0, 0} }; > >> +static const ovs_u128 OVS_BE128_MIN OVS_UNUSED = { {0, 0, 0, 0} > }; > >> + > >> +#define OVS_U128_ZERO OVS_U128_MIN > >> > >> /* A 64-bit value, in network byte order, that is only aligned on > a > >> 32-bit > >> * boundary. */ > >> diff --git a/lib/automake.mk b/lib/automake.mk > >> index affbb5c..df8b07d 100644 > >> --- a/lib/automake.mk > >> +++ b/lib/automake.mk > >> @@ -47,6 +47,11 @@ lib_libopenvswitch_la_SOURCES = \ > >> lib/compiler.h \ > >> lib/connectivity.c \ > >> lib/connectivity.h \ > >> + lib/conntrack-private.h \ > >> + lib/conntrack-tcp.c \ > >> + lib/conntrack-other.c \ > >> + lib/conntrack.c \ > >> + lib/conntrack.h \ > >> lib/coverage.c \ > >> lib/coverage.h \ > >> lib/crc32c.c \ > >> diff --git a/lib/conntrack-other.c b/lib/conntrack-other.c > >> new file mode 100644 > >> index 0000000..295cb2c > >> --- /dev/null > >> +++ b/lib/conntrack-other.c > >> @@ -0,0 +1,85 @@ > >> +/* > >> + * Copyright (c) 2015, 2016 Nicira, Inc. > >> + * > >> + * Licensed under the Apache License, Version 2.0 (the > "License"); > >> + * you may not use this file except in compliance with the > License. > >> + * You may obtain a copy of the License at: > >> + * > >> + * http://www.apache.org/licenses/LICENSE-2.0 > >> + * > >> + * Unless required by applicable law or agreed to in writing, > >> software > >> + * distributed under the License is distributed on an "AS IS" > BASIS, > >> + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express > or > >> implied. > >> + * See the License for the specific language governing > permissions > >> and > >> + * limitations under the License. > >> + */ > >> + > >> +#include <config.h> > >> + > >> +#include "conntrack-private.h" > >> +#include "dp-packet.h" > >> + > >> +enum other_state { > >> + OTHERS_FIRST, > >> + OTHERS_MULTIPLE, > >> + OTHERS_BIDIR, > >> +}; > >> + > >> +struct conn_other { > >> + struct conn up; > >> + enum other_state state; > >> +}; > >> + > >> +static const enum ct_timeout other_timeouts[] = { > >> + [OTHERS_FIRST] = CT_TM_OTHER_FIRST, > >> + [OTHERS_MULTIPLE] = CT_TM_OTHER_MULTIPLE, > >> + [OTHERS_BIDIR] = CT_TM_OTHER_BIDIR, > >> +}; > >> + > >> +static struct conn_other * > >> +conn_other_cast(const struct conn *conn) > >> +{ > >> + return CONTAINER_OF(conn, struct conn_other, up); > >> +} > >> + > >> +static enum ct_update_res > >> +other_conn_update(struct conn *conn_, struct dp_packet *pkt > >> OVS_UNUSED, > >> + bool reply, long long now) > >> +{ > >> + struct conn_other *conn = conn_other_cast(conn_); > >> + > >> + if (reply && conn->state != OTHERS_BIDIR) { > >> + conn->state = OTHERS_BIDIR; > >> + } else if (conn->state == OTHERS_FIRST) { > >> + conn->state = OTHERS_MULTIPLE; > >> + } > >> + > >> + update_expiration(conn_, other_timeouts[conn->state], now); > >> + > >> + return CT_UPDATE_VALID; > >> +} > >> + > >> +static bool > >> +other_valid_new(struct dp_packet *pkt OVS_UNUSED) > >> +{ > >> + return true; > >> +} > >> + > >> +static struct conn * > >> +other_new_conn(struct dp_packet *pkt OVS_UNUSED, long long now) > >> +{ > >> + struct conn_other *conn; > >> + > >> + conn = xzalloc(sizeof *conn); > >> + conn->state = OTHERS_FIRST; > >> + > >> + update_expiration(&conn->up, other_timeouts[conn->state], > now); > >> + > >> + return &conn->up; > >> +} > >> + > >> +struct ct_l4_proto ct_proto_other = { > >> + .new_conn = other_new_conn, > >> + .valid_new = other_valid_new, > >> + .conn_update = other_conn_update, > >> +}; > >> diff --git a/lib/conntrack-private.h b/lib/conntrack-private.h > >> new file mode 100644 > >> index 0000000..d3e0099 > >> --- /dev/null > >> +++ b/lib/conntrack-private.h > >> @@ -0,0 +1,88 @@ > >> +/* > >> + * Copyright (c) 2015, 2016 Nicira, Inc. > >> + * > >> + * Licensed under the Apache License, Version 2.0 (the > "License"); > >> + * you may not use this file except in compliance with the > License. > >> + * You may obtain a copy of the License at: > >> + * > >> + * http://www.apache.org/licenses/LICENSE-2.0 > >> + * > >> + * Unless required by applicable law or agreed to in writing, > >> software > >> + * distributed under the License is distributed on an "AS IS" > BASIS, > >> + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express > or > >> implied. > >> + * See the License for the specific language governing > permissions > >> and > >> + * limitations under the License. > >> + */ > >> + > >> +#ifndef CONNTRACK_PRIVATE_H > >> +#define CONNTRACK_PRIVATE_H 1 > >> + > >> +#include <sys/types.h> > >> +#include <netinet/in.h> > >> +#include <netinet/ip6.h> > >> + > >> +#include "conntrack.h" > >> +#include "hmap.h" > >> +#include "openvswitch/list.h" > >> +#include "openvswitch/types.h" > >> +#include "packets.h" > >> +#include "unaligned.h" > >> + > >> +struct ct_addr { > >> + union { > >> + ovs_16aligned_be32 ipv4; > >> + union ovs_16aligned_in6_addr ipv6; > >> + ovs_be32 ipv4_aligned; > >> + struct in6_addr ipv6_aligned; > >> + }; > >> +}; > >> + > >> +struct ct_endpoint { > >> + struct ct_addr addr; > >> + ovs_be16 port; > >> +}; > >> + > >> +struct conn_key { > >> + struct ct_endpoint src; > >> + struct ct_endpoint dst; > >> + > >> + ovs_be16 dl_type; > >> + uint8_t nw_proto; > >> + uint16_t zone; > >> +}; > >> + > >> +struct conn { > >> + struct conn_key key; > >> + struct conn_key rev_key; > >> + long long expiration; > >> + struct ovs_list exp_node; > >> + struct hmap_node node; > >> + uint32_t mark; > >> + ovs_u128 label; > >> +}; > >> + > >> +enum ct_update_res { > >> + CT_UPDATE_INVALID, > >> + CT_UPDATE_VALID, > >> + CT_UPDATE_NEW, > >> +}; > >> + > >> +struct ct_l4_proto { > >> + struct conn *(*new_conn)(struct dp_packet *pkt, long long > now); > >> + bool (*valid_new)(struct dp_packet *pkt); > >> + enum ct_update_res (*conn_update)(struct conn *conn, struct > >> dp_packet *pkt, > >> + bool reply, long long now); > >> +}; > >> + > >> +extern struct ct_l4_proto ct_proto_tcp; > >> +extern struct ct_l4_proto ct_proto_other; > >> + > >> +extern long long ct_timeout_val[]; > >> + > >> +static inline void > >> +update_expiration(struct conn *conn, enum ct_timeout tm, long > long > >> now) > >> +{ > >> + conn->expiration = now + ct_timeout_val[tm]; > >> +} > >> + > >> +#endif /* conntrack-private.h */ > >> diff --git a/lib/conntrack-tcp.c b/lib/conntrack-tcp.c > >> new file mode 100644 > >> index 0000000..4d3d9c3 > >> --- /dev/null > >> +++ b/lib/conntrack-tcp.c > >> @@ -0,0 +1,463 @@ > >> +/*- > >> + * Copyright (c) 2001 Daniel Hartmeier > >> + * Copyright (c) 2002 - 2008 Henning Brauer > >> + * Copyright (c) 2012 Gleb Smirnoff <gleb...@freebsd.org> > >> + * Copyright (c) 2015, 2016 Nicira, Inc. > >> + * All rights reserved. > >> + * > >> + * Redistribution and use in source and binary forms, with or > >> without > >> + * modification, are permitted provided that the following > >> conditions > >> + * are met: > >> + * > >> + * - Redistributions of source code must retain the above > >> copyright > >> + * notice, this list of conditions and the following > >> disclaimer. > >> + * - 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. > >> + * > >> + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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 > >> + * COPYRIGHT HOLDERS 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. > >> + * > >> + * Effort sponsored in part by the Defense Advanced Research > >> Projects > >> + * Agency (DARPA) and Air Force Research Laboratory, Air Force > >> + * Materiel Command, USAF, under agreement number F30602-01-2- > 0537. > >> + * > >> + * $OpenBSD: pf.c,v 1.634 2009/02/27 12:37:45 henning Exp $ > >> + */ > >> + > >> +#include <config.h> > >> + > >> +#include "conntrack-private.h" > >> +#include "ct-dpif.h" > >> +#include "dp-packet.h" > >> +#include "util.h" > >> + > >> +struct tcp_peer { > >> + enum ct_dpif_tcp_state state; > >> + uint32_t seqlo; /* Max sequence number > >> sent */ > >> + uint32_t seqhi; /* Max the other end > ACKd > >> + win */ > >> + uint16_t max_win; /* largest window (pre > >> scaling) */ > >> + uint8_t wscale; /* window scaling > factor > >> */ > >> +}; > >> + > >> +struct conn_tcp { > >> + struct conn up; > >> + struct tcp_peer peer[2]; > >> +}; > >> + > >> +enum { > >> + TCPOPT_EOL, > >> + TCPOPT_NOP, > >> + TCPOPT_WINDOW = 3, > >> +}; > >> + > >> +/* TCP sequence numbers are 32 bit integers operated > >> + * on with modular arithmetic. These macros can be > >> + * used to compare such integers. */ > >> +#define SEQ_LT(a,b) INT_MOD_LT(a, b) > >> +#define SEQ_LEQ(a,b) INT_MOD_LEQ(a, b) > >> +#define SEQ_GT(a,b) INT_MOD_GT(a, b) > >> +#define SEQ_GEQ(a,b) INT_MOD_GEQ(a, b) > >> + > >> +#define SEQ_MIN(a, b) INT_MOD_MIN(a, b) > >> +#define SEQ_MAX(a, b) INT_MOD_MAX(a, b) > >> + > >> +static struct conn_tcp* > >> +conn_tcp_cast(const struct conn* conn) > >> +{ > >> + return CONTAINER_OF(conn, struct conn_tcp, up); > >> +} > >> + > >> +/* pf does this in in pf_normalize_tcp(), and it is called only > if > >> scrub > >> + * is enabled. We're not scrubbing, but this check seems > >> reasonable. */ > >> +static bool > >> +tcp_invalid_flags(uint16_t flags) > >> +{ > >> + > >> + if (flags & TCP_SYN) { > >> + if (flags & TCP_RST || flags & TCP_FIN) { > >> + return true; > >> + } > >> + } else { > >> + /* Illegal packet */ > >> + if (!(flags & (TCP_ACK|TCP_RST))) { > >> + return true; > >> + } > >> + } > >> + > >> + if (!(flags & TCP_ACK)) { > >> + /* These flags are only valid if ACK is set */ > >> + if ((flags & TCP_FIN) || (flags & TCP_PSH) || (flags & > >> TCP_URG)) { > >> + return true; > >> + } > >> + } > >> + > >> + return false; > >> +} > >> + > >> +#define TCP_MAX_WSCALE 14 > >> +#define CT_WSCALE_FLAG 0x80 > >> +#define CT_WSCALE_UNKNOWN 0x40 > >> +#define CT_WSCALE_MASK 0xf > >> + > >> +static uint8_t > >> +tcp_get_wscale(const struct tcp_header *tcp) > >> +{ > >> + int len = TCP_OFFSET(tcp->tcp_ctl) * 4 - sizeof *tcp; > >> + const uint8_t *opt = (const uint8_t *)(tcp + 1); > >> + uint8_t wscale = 0; > >> + uint8_t optlen; > >> + > >> + while (len >= 3) { > >> + switch (*opt) { > >> + case TCPOPT_EOL: > >> + return wscale; > >> + case TCPOPT_NOP: > >> + opt++; > >> + len--; > >> + break; > >> + case TCPOPT_WINDOW: > >> + wscale = MIN(opt[2], TCP_MAX_WSCALE); > >> + wscale |= CT_WSCALE_FLAG; > >> + /* fall through */ > >> + default: > >> + optlen = opt[1]; > >> + if (optlen < 2) { > >> + optlen = 2; > >> + } > >> + len -= optlen; > >> + opt += optlen; > >> + } > >> + } > >> + > >> + return wscale; > >> +} > >> + > >> +static uint32_t > >> +tcp_payload_length(struct dp_packet *pkt) > >> +{ > >> + return (char *) dp_packet_tail(pkt) - > dp_packet_l2_pad_size(pkt) > >> + - (char *) dp_packet_get_tcp_payload(pkt); > >> +} > >> + > >> +static enum ct_update_res > >> +tcp_conn_update(struct conn* conn_, struct dp_packet *pkt, bool > >> reply, > >> + long long now) > >> +{ > >> + struct conn_tcp *conn = conn_tcp_cast(conn_); > >> + struct tcp_header *tcp = dp_packet_l4(pkt); > >> + /* The peer that sent 'pkt' */ > >> + struct tcp_peer *src = &conn->peer[reply ? 1 : 0]; > >> + /* The peer that should receive 'pkt' */ > >> + struct tcp_peer *dst = &conn->peer[reply ? 0 : 1]; > >> + uint8_t sws = 0, dws = 0; > >> + uint16_t tcp_flags = TCP_FLAGS(tcp->tcp_ctl); > >> + > >> + uint16_t win = ntohs(tcp->tcp_winsz); > >> + uint32_t ack, end, seq, orig_seq; > >> + uint32_t p_len = tcp_payload_length(pkt); > >> + int ackskew; > >> + > >> + if (tcp_invalid_flags(tcp_flags)) { > >> + return CT_UPDATE_INVALID; > >> + } > >> + > >> + if (((tcp_flags & (TCP_SYN|TCP_ACK)) == TCP_SYN) && > >> + dst->state >= CT_DPIF_TCPS_FIN_WAIT_2 && > >> + src->state >= CT_DPIF_TCPS_FIN_WAIT_2) { > >> + src->state = dst->state = CT_DPIF_TCPS_CLOSED; > >> + return CT_UPDATE_NEW; > >> + } > >> + > >> + if (src->wscale & CT_WSCALE_FLAG > >> + && dst->wscale & CT_WSCALE_FLAG > >> + && !(tcp_flags & TCP_SYN)) { > >> + > >> + sws = src->wscale & CT_WSCALE_MASK; > >> + dws = dst->wscale & CT_WSCALE_MASK; > >> + > >> + } else if (src->wscale & CT_WSCALE_UNKNOWN > >> + && dst->wscale & CT_WSCALE_UNKNOWN > >> + && !(tcp_flags & TCP_SYN)) { > >> + > >> + sws = TCP_MAX_WSCALE; > >> + dws = TCP_MAX_WSCALE; > >> + } > >> + > >> + /* > >> + * Sequence tracking algorithm from Guido van Rooij's paper: > >> + * http://www.madison- > gurkha.com/publications/tcp_filtering/ > >> + * tcp_filtering.ps > >> + */ > >> + > >> + orig_seq = seq = ntohl(get_16aligned_be32(&tcp->tcp_seq)); > >> + if (src->state < CT_DPIF_TCPS_SYN_SENT) { > >> + /* First packet from this end. Set its state */ > >> + > >> + ack = ntohl(get_16aligned_be32(&tcp->tcp_ack)); > >> + > >> + end = seq + p_len; > >> + if (tcp_flags & TCP_SYN) { > >> + end++; > >> + if (dst->wscale & CT_WSCALE_FLAG) { > >> + src->wscale = tcp_get_wscale(tcp); > >> + if (src->wscale & CT_WSCALE_FLAG) { > >> + /* Remove scale factor from initial window */ > >> + sws = src->wscale & CT_WSCALE_MASK; > >> + win = DIV_ROUND_UP((uint32_t) win, 1 << sws); > >> + dws = dst->wscale & CT_WSCALE_MASK; > >> + } else { > >> + /* fixup other window */ > >> + dst->max_win <<= dst->wscale & > >> + CT_WSCALE_MASK; > >> + /* in case of a retrans SYN|ACK */ > >> + dst->wscale = 0; > >> + } > >> + } > >> + } > >> + if (tcp_flags & TCP_FIN) { > >> + end++; > >> + } > >> + > >> + src->seqlo = seq; > >> + src->state = CT_DPIF_TCPS_SYN_SENT; > >> + /* > >> + * May need to slide the window (seqhi may have been set > by > >> + * the crappy stack check or if we picked up the > connection > >> + * after establishment) > >> + */ > >> + if (src->seqhi == 1 || > >> + SEQ_GEQ(end + MAX(1, dst->max_win << dws), src- > >> >seqhi)) { > >> + src->seqhi = end + MAX(1, dst->max_win << dws); > >> + } > >> + if (win > src->max_win) { > >> + src->max_win = win; > >> + } > >> + > >> + } else { > >> + ack = ntohl(get_16aligned_be32(&tcp->tcp_ack)); > >> + end = seq + p_len; > >> + if (tcp_flags & TCP_SYN) { > >> + end++; > >> + } > >> + if (tcp_flags & TCP_FIN) { > >> + end++; > >> + } > >> + } > >> + > >> + if ((tcp_flags & TCP_ACK) == 0) { > >> + /* Let it pass through the ack skew check */ > >> + ack = dst->seqlo; > >> + } else if ((ack == 0 > >> + && (tcp_flags & (TCP_ACK|TCP_RST)) == > >> (TCP_ACK|TCP_RST)) > >> + /* broken tcp stacks do not set ack */) { > >> + /* Many stacks (ours included) will set the ACK number in > an > >> + * FIN|ACK if the SYN times out -- no sequence to ACK. */ > >> + ack = dst->seqlo; > >> + } > >> + > >> + if (seq == end) { > >> + /* Ease sequencing restrictions on no data packets */ > >> + seq = src->seqlo; > >> + end = seq; > >> + } > >> + > >> + ackskew = dst->seqlo - ack; > >> +#define MAXACKWINDOW (0xffff + 1500) /* 1500 is an arbitrary > >> fudge factor */ > >> + if (SEQ_GEQ(src->seqhi, end) > >> + /* Last octet inside other's window space */ > >> + && SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws)) > >> + /* Retrans: not more than one window back */ > >> + && (ackskew >= -MAXACKWINDOW) > >> + /* Acking not more than one reassembled fragment > backwards > >> */ > >> + && (ackskew <= (MAXACKWINDOW << sws)) > >> + /* Acking not more than one window forward */ > >> + && ((tcp_flags & TCP_RST) == 0 || orig_seq == src->seqlo > >> + || (orig_seq == src->seqlo + 1) || (orig_seq + 1 == > src- > >> >seqlo))) { > >> + /* Require an exact/+1 sequence match on resets when > >> possible */ > >> + > >> + /* update max window */ > >> + if (src->max_win < win) { > >> + src->max_win = win; > >> + } > >> + /* synchronize sequencing */ > >> + if (SEQ_GT(end, src->seqlo)) { > >> + src->seqlo = end; > >> + } > >> + /* slide the window of what the other end can send */ > >> + if (SEQ_GEQ(ack + (win << sws), dst->seqhi)) { > >> + dst->seqhi = ack + MAX((win << sws), 1); > >> + } > >> + > >> + /* update states */ > >> + if (tcp_flags & TCP_SYN && src->state < > >> CT_DPIF_TCPS_SYN_SENT) { > >> + src->state = CT_DPIF_TCPS_SYN_SENT; > >> + } > >> + if (tcp_flags & TCP_FIN && src->state < > >> CT_DPIF_TCPS_CLOSING) { > >> + src->state = CT_DPIF_TCPS_CLOSING; > >> + } > >> + if (tcp_flags & TCP_ACK) { > >> + if (dst->state == CT_DPIF_TCPS_SYN_SENT) { > >> + dst->state = CT_DPIF_TCPS_ESTABLISHED; > >> + } else if (dst->state == CT_DPIF_TCPS_CLOSING) { > >> + dst->state = CT_DPIF_TCPS_FIN_WAIT_2; > >> + } > >> + } > >> + if (tcp_flags & TCP_RST) { > >> + src->state = dst->state = CT_DPIF_TCPS_TIME_WAIT; > >> + } > >> + > >> + if (src->state >= CT_DPIF_TCPS_FIN_WAIT_2 > >> + && dst->state >= CT_DPIF_TCPS_FIN_WAIT_2) { > >> + update_expiration(conn_, CT_TM_TCP_CLOSED, now); > >> + } else if (src->state >= CT_DPIF_TCPS_CLOSING > >> + && dst->state >= CT_DPIF_TCPS_CLOSING) { > >> + update_expiration(conn_, CT_TM_TCP_FIN_WAIT, now); > >> + } else if (src->state < CT_DPIF_TCPS_ESTABLISHED > >> + || dst->state < CT_DPIF_TCPS_ESTABLISHED) { > >> + update_expiration(conn_, now, CT_TM_TCP_OPENING); > >> + } else if (src->state >= CT_DPIF_TCPS_CLOSING > >> + || dst->state >= CT_DPIF_TCPS_CLOSING) { > >> + update_expiration(conn_, now, CT_TM_TCP_CLOSING); > >> + } else { > >> + update_expiration(conn_, now, CT_TM_TCP_ESTABLISHED); > >> + } > >> + } else if ((dst->state < CT_DPIF_TCPS_SYN_SENT > >> + || dst->state >= CT_DPIF_TCPS_FIN_WAIT_2 > >> + || src->state >= CT_DPIF_TCPS_FIN_WAIT_2) > >> + && SEQ_GEQ(src->seqhi + MAXACKWINDOW, end) > >> + /* Within a window forward of the originating > packet > >> */ > >> + && SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW)) { > >> + /* Within a window backward of the originating > packet > >> */ > >> + > >> + /* > >> + * This currently handles three situations: > >> + * 1) Stupid stacks will shotgun SYNs before their peer > >> + * replies. > >> + * 2) When PF catches an already established stream (the > >> + * firewall rebooted, the state table was flushed, > >> routes > >> + * changed...) > >> + * 3) Packets get funky immediately after the connection > >> + * closes (this should catch Solaris spurious > ACK|FINs > >> + * that web servers like to spew after a close) > >> + * > >> + * This must be a little more careful than the above code > >> + * since packet floods will also be caught here. We don't > >> + * update the TTL here to mitigate the damage of a packet > >> + * flood and so the same code can handle awkward > >> establishment > >> + * and a loosened connection close. > >> + * In the establishment case, a correct peer response > will > >> + * validate the connection, go through the normal state > code > >> + * and keep updating the state TTL. > >> + */ > >> + > >> + /* update max window */ > >> + if (src->max_win < win) { > >> + src->max_win = win; > >> + } > >> + /* synchronize sequencing */ > >> + if (SEQ_GT(end, src->seqlo)) { > >> + src->seqlo = end; > >> + } > >> + /* slide the window of what the other end can send */ > >> + if (SEQ_GEQ(ack + (win << sws), dst->seqhi)) { > >> + dst->seqhi = ack + MAX((win << sws), 1); > >> + } > >> + > >> + /* > >> + * Cannot set dst->seqhi here since this could be a > >> shotgunned > >> + * SYN and not an already established connection. > >> + */ > >> + > >> + if (tcp_flags & TCP_FIN && src->state < > >> CT_DPIF_TCPS_CLOSING) { > >> + src->state = CT_DPIF_TCPS_CLOSING; > >> + } > >> + > >> + if (tcp_flags & TCP_RST) { > >> + src->state = dst->state = CT_DPIF_TCPS_TIME_WAIT; > >> + } > >> + } else { > >> + return CT_UPDATE_INVALID; > >> + } > >> + > >> + return CT_UPDATE_VALID; > >> +} > >> + > >> +static bool > >> +tcp_valid_new(struct dp_packet *pkt) > >> +{ > >> + struct tcp_header *tcp = dp_packet_l4(pkt); > >> + uint16_t tcp_flags = TCP_FLAGS(tcp->tcp_ctl); > >> + > >> + if (tcp_invalid_flags(tcp_flags)) { > >> + return false; > >> + } > >> + > >> + /* A syn+ack is not allowed to create a connection. We want > to > >> allow > >> + * totally new connections (syn) or already established, not > >> partially > >> + * open (syn+ack). */ > >> + if ((tcp_flags & TCP_SYN) && (tcp_flags & TCP_ACK)) { > >> + return false; > >> + } > >> + > >> + return true; > >> +} > >> + > >> +static struct conn * > >> +tcp_new_conn(struct dp_packet *pkt, long long now) > >> +{ > >> + struct conn_tcp* newconn = NULL; > >> + struct tcp_header *tcp = dp_packet_l4(pkt); > >> + struct tcp_peer *src, *dst; > >> + uint16_t tcp_flags = TCP_FLAGS(tcp->tcp_ctl); > >> + > >> + newconn = xzalloc(sizeof *newconn); > >> + > >> + src = &newconn->peer[0]; > >> + dst = &newconn->peer[1]; > >> + > >> + src->seqlo = ntohl(get_16aligned_be32(&tcp->tcp_seq)); > >> + src->seqhi = src->seqlo + tcp_payload_length(pkt) + 1; > >> + > >> + if (tcp_flags & TCP_SYN) { > >> + src->seqhi++; > >> + src->wscale = tcp_get_wscale(tcp); > >> + } else { > >> + src->wscale = CT_WSCALE_UNKNOWN; > >> + dst->wscale = CT_WSCALE_UNKNOWN; > >> + } > >> + src->max_win = MAX(ntohs(tcp->tcp_winsz), 1); > >> + if (src->wscale & CT_WSCALE_MASK) { > >> + /* Remove scale factor from initial window */ > >> + uint8_t sws = src->wscale & CT_WSCALE_MASK; > >> + src->max_win = DIV_ROUND_UP((uint32_t) src->max_win, 1 << > >> sws); > >> + } > >> + if (tcp_flags & TCP_FIN) { > >> + src->seqhi++; > >> + } > >> + dst->seqhi = 1; > >> + dst->max_win = 1; > >> + src->state = CT_DPIF_TCPS_SYN_SENT; > >> + dst->state = CT_DPIF_TCPS_CLOSED; > >> + > >> + update_expiration(&newconn->up, now, CT_TM_TCP_FIRST_PACKET); > >> + > >> + return &newconn->up; > >> +} > >> + > >> +struct ct_l4_proto ct_proto_tcp = { > >> + .new_conn = tcp_new_conn, > >> + .valid_new = tcp_valid_new, > >> + .conn_update = tcp_conn_update, > >> +}; > >> diff --git a/lib/conntrack.c b/lib/conntrack.c > >> new file mode 100644 > >> index 0000000..e282485 > >> --- /dev/null > >> +++ b/lib/conntrack.c > >> @@ -0,0 +1,883 @@ > >> +/* > >> + * Copyright (c) 2015, 2016 Nicira, Inc. > >> + * > >> + * Licensed under the Apache License, Version 2.0 (the > "License"); > >> + * you may not use this file except in compliance with the > License. > >> + * You may obtain a copy of the License at: > >> + * > >> + * http://www.apache.org/licenses/LICENSE-2.0 > >> + * > >> + * Unless required by applicable law or agreed to in writing, > >> software > >> + * distributed under the License is distributed on an "AS IS" > BASIS, > >> + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express > or > >> implied. > >> + * See the License for the specific language governing > permissions > >> and > >> + * limitations under the License. > >> + */ > >> + > >> +#include <config.h> > >> +#include "conntrack.h" > >> + > >> +#include <errno.h> > >> +#include <sys/types.h> > >> +#include <netinet/in.h> > >> +#include <netinet/icmp6.h> > >> + > >> +#include "bitmap.h" > >> +#include "conntrack-private.h" > >> +#include "coverage.h" > >> +#include "csum.h" > >> +#include "dp-packet.h" > >> +#include "flow.h" > >> +#include "hmap.h" > >> +#include "netdev.h" > >> +#include "odp-netlink.h" > >> +#include "openvswitch/vlog.h" > >> +#include "ovs-rcu.h" > >> +#include "random.h" > >> +#include "timeval.h" > >> + > >> +VLOG_DEFINE_THIS_MODULE(conntrack); > >> + > >> +COVERAGE_DEFINE(conntrack_new_full); > >> + > >> +struct conn_lookup_ctx { > >> + struct conn_key key; > >> + struct conn *conn; > >> + uint32_t hash; > >> + bool reply; > >> + bool related; > >> +}; > >> + > >> +static bool conn_key_extract(struct conntrack *, struct dp_packet > *, > >> + struct conn_lookup_ctx *, uint16_t > >> zone); > >> +static uint32_t conn_key_hash(const struct conn_key *, uint32_t > >> basis); > >> +static void conn_key_reverse(struct conn_key *); > >> +static void conn_key_lookup(struct conntrack_bucket *ctb, > >> + struct conn_lookup_ctx *ctx, > >> + long long now); > >> +static bool valid_new(struct dp_packet *pkt, struct conn_key *); > >> +static struct conn *new_conn(struct dp_packet *pkt, struct > conn_key > >> *, > >> + long long now); > >> +static void delete_conn(struct conn *); > >> +static enum ct_update_res conn_update(struct conn *, struct > >> dp_packet*, > >> + bool reply, long long now); > >> +static bool conn_expired(struct conn *, long long now); > >> +static void set_mark(struct dp_packet *, struct conn *, > >> + uint32_t val, uint32_t mask); > >> +static void set_label(struct dp_packet *, struct conn *, > >> + const struct ovs_key_ct_labels *val, > >> + const struct ovs_key_ct_labels *mask); > >> + > >> +static struct ct_l4_proto *l4_protos[] = { > >> + [IPPROTO_TCP] = &ct_proto_tcp, > >> + [IPPROTO_UDP] = &ct_proto_other, > >> + [IPPROTO_ICMP] = &ct_proto_other, > >> + [IPPROTO_ICMPV6] = &ct_proto_other, > >> +}; > >> + > >> +long long ct_timeout_val[] = { > >> +#define CT_TIMEOUT(NAME, VAL) [CT_TM_##NAME] = VAL, > >> + CT_TIMEOUTS > >> +#undef CT_TIMEOUT > >> +}; > >> + > >> +/* If the total number of connections goes above this value, no > new > >> connections > >> + * are accepted */ > >> +#define DEFAULT_N_CONN_LIMIT 3000000 > >> + > >> +/* Initializes the connection tracker 'ct'. The caller is > >> responbile for > >> + * calling 'conntrack_destroy()', when the instance is not needed > >> anymore */ > >> +void > >> +conntrack_init(struct conntrack *ct) > >> +{ > >> + unsigned i; > >> + > >> + for (i = 0; i < CONNTRACK_BUCKETS; i++) { > >> + struct conntrack_bucket *ctb = &ct->buckets[i]; > >> + > >> + ct_lock_init(&ctb->lock); > >> + ct_lock_lock(&ctb->lock); > >> + hmap_init(&ctb->connections); > >> + ct_lock_unlock(&ctb->lock); > >> + } > >> + ct->hash_basis = random_uint32(); > >> + atomic_count_init(&ct->n_conn, 0); > >> + atomic_init(&ct->n_conn_limit, DEFAULT_N_CONN_LIMIT); > >> +} > >> + > >> +/* Destroys the connection tracker 'ct' and frees all the > allocated > >> memory. */ > >> +void > >> +conntrack_destroy(struct conntrack *ct) > >> +{ > >> + unsigned i; > >> + > >> + for (i = 0; i < CONNTRACK_BUCKETS; i++) { > >> + struct conntrack_bucket *ctb = &ct->buckets[i]; > >> + struct conn *conn; > >> + > >> + ct_lock_lock(&ctb->lock); > >> + HMAP_FOR_EACH_POP(conn, node, &ctb->connections) { > >> + atomic_count_dec(&ct->n_conn); > >> + delete_conn(conn); > >> + } > >> + hmap_destroy(&ctb->connections); > >> + ct_lock_unlock(&ctb->lock); > >> + ct_lock_destroy(&ctb->lock); > >> + } > >> +} > >> + > > > >> +static unsigned hash_to_bucket(uint32_t hash) > >> +{ > >> + /* Extracts the most significant bits in hash. The least > >> significant bits > >> + * are already used internally by the hmap implementation. */ > >> + BUILD_ASSERT(CONNTRACK_BUCKETS_SHIFT < 32 && > >> CONNTRACK_BUCKETS_SHIFT >= 1); > >> + > >> + return (hash >> (32 - CONNTRACK_BUCKETS_SHIFT)) % > >> CONNTRACK_BUCKETS; > >> +} > >> + > >> +static void > >> +write_ct_md(struct dp_packet *pkt, uint16_t state, uint16_t zone, > >> + uint32_t mark, ovs_u128 label) > >> +{ > >> + pkt->md.ct_state = state | CS_TRACKED; > >> + pkt->md.ct_zone = zone; > >> + pkt->md.ct_mark = mark; > >> + pkt->md.ct_label = label; > >> +} > >> + > >> +static struct conn * > >> +conn_not_found(struct conntrack *ct, struct dp_packet *pkt, > >> + struct conn_lookup_ctx *ctx, uint16_t *state, bool > >> commit, > >> + long long now) > >> +{ > >> + unsigned bucket = hash_to_bucket(ctx->hash); > >> + struct conn *nc = NULL; > >> + > >> + if (!valid_new(pkt, &ctx->key)) { > >> + *state |= CS_INVALID; > >> + return nc; > >> + } > >> + > >> + *state |= CS_NEW; > >> + > >> + if (commit) { > >> + unsigned int n_conn_limit; > >> + > >> + atomic_read_relaxed(&ct->n_conn_limit, &n_conn_limit); > >> + > >> + if (atomic_count_get(&ct->n_conn) >= n_conn_limit) { > >> + COVERAGE_INC(conntrack_new_full); > >> + return nc; > >> + } > >> + > >> + nc = new_conn(pkt, &ctx->key, now); > >> + > >> + memcpy(&nc->rev_key, &ctx->key, sizeof nc->rev_key); > >> + > >> + conn_key_reverse(&nc->rev_key); > >> + hmap_insert(&ct->buckets[bucket].connections, &nc->node, > >> ctx->hash); > >> + atomic_count_inc(&ct->n_conn); > >> + } > >> + > >> + return nc; > >> +} > >> + > >> +static struct conn * > >> +process_one(struct conntrack *ct, struct dp_packet *pkt, > >> + struct conn_lookup_ctx *ctx, uint16_t zone, > >> + bool commit, long long now) > >> +{ > >> + unsigned bucket = hash_to_bucket(ctx->hash); > >> + struct conn *conn = ctx->conn; > >> + uint16_t state = 0; > >> + > >> + if (conn) { > >> + if (ctx->related) { > >> + state |= CS_RELATED; > >> + if (ctx->reply) { > >> + state |= CS_REPLY_DIR; > >> + } > >> + } else { > >> + enum ct_update_res res; > >> + > >> + res = conn_update(conn, pkt, ctx->reply, now); > >> + > >> + switch (res) { > >> + case CT_UPDATE_VALID: > >> + state |= CS_ESTABLISHED; > >> + if (ctx->reply) { > >> + state |= CS_REPLY_DIR; > >> + } > >> + break; > >> + case CT_UPDATE_INVALID: > >> + state |= CS_INVALID; > >> + break; > >> + case CT_UPDATE_NEW: > >> + hmap_remove(&ct->buckets[bucket].connections, > &conn- > >> >node); > >> + atomic_count_dec(&ct->n_conn); > >> + delete_conn(conn); > >> + conn = conn_not_found(ct, pkt, ctx, &state, > commit, > >> now); > >> + break; > >> + } > > > > [Antonio F] Sorry to repeat, but I'd prefer to add the 'default' > > case, here. I mean something like > > > > default: > > state |= CS_INVALID; > > break; > > > > I know if we add new items to enum ct_update_res we can get a > > warning from the compiler, but I wouldn't rely on that. > > If we're going to rely upon a default case that we don't expect to > hit, we should consider calling OVS_NOT_REACHED() rather than > treating the > traffic as invalid; this would be easier to track down than quietly > marking some traffic as invalid.
[Antonio F] Both options looks fine to me, I just wanted to point out that it should be better to add a default case to cover any possible value. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
