From 03f0d916aa0317592dda11bd17c7357858719b6c Mon Sep 17 00:00:00 2001 From: Andy Zhou Date: Wed, 7 Aug 2013 20:01:00 -0700 Subject: openvswitch: Mega flow implementation Add wildcarded flow support in kernel datapath. Wildcarded flow can improve OVS flow set up performance by avoid sending matching new flows to the user space program. The exact performance boost will largely dependent on wildcarded flow hit rate. In case all new flows hits wildcard flows, the flow set up rate is within 5% of that of linux bridge module. Pravin has made significant contributions to this patch. Including API clean ups and bug fixes. Signed-off-by: Pravin B Shelar Signed-off-by: Andy Zhou Signed-off-by: Jesse Gross --- include/uapi/linux/openvswitch.h | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'include/uapi') diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h index 52490b0e62b5..de1fa5d3780f 100644 --- a/include/uapi/linux/openvswitch.h +++ b/include/uapi/linux/openvswitch.h @@ -1,6 +1,6 @@ /* - * Copyright (c) 2007-2011 Nicira Networks. + * Copyright (c) 2007-2013 Nicira, Inc. * * This program is free software; you can redistribute it and/or * modify it under the terms of version 2 of the GNU General Public @@ -379,6 +379,12 @@ struct ovs_key_nd { * @OVS_FLOW_ATTR_CLEAR: If present in a %OVS_FLOW_CMD_SET request, clears the * last-used time, accumulated TCP flags, and statistics for this flow. * Otherwise ignored in requests. Never present in notifications. + * @OVS_FLOW_ATTR_MASK: Nested %OVS_KEY_ATTR_* attributes specifying the + * mask bits for wildcarded flow match. Mask bit value '1' specifies exact + * match with corresponding flow key bit, while mask bit value '0' specifies + * a wildcarded match. Omitting attribute is treated as wildcarding all + * corresponding fields. Optional for all requests. If not present, + * all flow key bits are exact match bits. * * These attributes follow the &struct ovs_header within the Generic Netlink * payload for %OVS_FLOW_* commands. @@ -391,6 +397,7 @@ enum ovs_flow_attr { OVS_FLOW_ATTR_TCP_FLAGS, /* 8-bit OR'd TCP flags. */ OVS_FLOW_ATTR_USED, /* u64 msecs last used in monotonic time. */ OVS_FLOW_ATTR_CLEAR, /* Flag to clear stats, tcp_flags, used. */ + OVS_FLOW_ATTR_MASK, /* Sequence of OVS_KEY_ATTR_* attributes. */ __OVS_FLOW_ATTR_MAX }; -- cgit v1.2.3 From a175a723301a8a4a9fedf9ce5b8ca586e7a97b40 Mon Sep 17 00:00:00 2001 From: Joe Stringer Date: Thu, 22 Aug 2013 12:30:48 -0700 Subject: openvswitch: Add SCTP support This patch adds support for rewriting SCTP src,dst ports similar to the functionality already available for TCP/UDP. Rewriting SCTP ports is expensive due to double-recalculation of the SCTP checksums; this is performed to ensure that packets traversing OVS with invalid checksums will continue to the destination with any checksum corruption intact. Reviewed-by: Simon Horman Signed-off-by: Joe Stringer Signed-off-by: Ben Pfaff Signed-off-by: Jesse Gross --- include/uapi/linux/openvswitch.h | 6 ++++ net/openvswitch/Kconfig | 1 + net/openvswitch/actions.c | 39 ++++++++++++++++++++++++ net/openvswitch/datapath.c | 6 ++++ net/openvswitch/flow.c | 65 ++++++++++++++++++++++++++++++++++++++++ net/openvswitch/flow.h | 8 ++--- 6 files changed, 121 insertions(+), 4 deletions(-) (limited to 'include/uapi') diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h index de1fa5d3780f..a74d375b439b 100644 --- a/include/uapi/linux/openvswitch.h +++ b/include/uapi/linux/openvswitch.h @@ -259,6 +259,7 @@ enum ovs_key_attr { OVS_KEY_ATTR_ND, /* struct ovs_key_nd */ OVS_KEY_ATTR_SKB_MARK, /* u32 skb mark */ OVS_KEY_ATTR_TUNNEL, /* Nested set of ovs_tunnel attributes */ + OVS_KEY_ATTR_SCTP, /* struct ovs_key_sctp */ #ifdef __KERNEL__ OVS_KEY_ATTR_IPV4_TUNNEL, /* struct ovs_key_ipv4_tunnel */ @@ -333,6 +334,11 @@ struct ovs_key_udp { __be16 udp_dst; }; +struct ovs_key_sctp { + __be16 sctp_src; + __be16 sctp_dst; +}; + struct ovs_key_icmp { __u8 icmp_type; __u8 icmp_code; diff --git a/net/openvswitch/Kconfig b/net/openvswitch/Kconfig index bed30e69baa7..6ecf491ad509 100644 --- a/net/openvswitch/Kconfig +++ b/net/openvswitch/Kconfig @@ -4,6 +4,7 @@ config OPENVSWITCH tristate "Open vSwitch" + select LIBCRC32C ---help--- Open vSwitch is a multilayer Ethernet switch targeted at virtualized environments. In addition to supporting a variety of features diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c index 1f680222f4f7..65cfaa816075 100644 --- a/net/openvswitch/actions.c +++ b/net/openvswitch/actions.c @@ -22,6 +22,7 @@ #include #include #include +#include #include #include #include @@ -31,6 +32,7 @@ #include #include #include +#include #include "datapath.h" #include "vport.h" @@ -352,6 +354,39 @@ static int set_tcp(struct sk_buff *skb, const struct ovs_key_tcp *tcp_port_key) return 0; } +static int set_sctp(struct sk_buff *skb, + const struct ovs_key_sctp *sctp_port_key) +{ + struct sctphdr *sh; + int err; + unsigned int sctphoff = skb_transport_offset(skb); + + err = make_writable(skb, sctphoff + sizeof(struct sctphdr)); + if (unlikely(err)) + return err; + + sh = sctp_hdr(skb); + if (sctp_port_key->sctp_src != sh->source || + sctp_port_key->sctp_dst != sh->dest) { + __le32 old_correct_csum, new_csum, old_csum; + + old_csum = sh->checksum; + old_correct_csum = sctp_compute_cksum(skb, sctphoff); + + sh->source = sctp_port_key->sctp_src; + sh->dest = sctp_port_key->sctp_dst; + + new_csum = sctp_compute_cksum(skb, sctphoff); + + /* Carry any checksum errors through. */ + sh->checksum = old_csum ^ old_correct_csum ^ new_csum; + + skb->rxhash = 0; + } + + return 0; +} + static int do_output(struct datapath *dp, struct sk_buff *skb, int out_port) { struct vport *vport; @@ -461,6 +496,10 @@ static int execute_set_action(struct sk_buff *skb, case OVS_KEY_ATTR_UDP: err = set_udp(skb, nla_data(nested_attr)); break; + + case OVS_KEY_ATTR_SCTP: + err = set_sctp(skb, nla_data(nested_attr)); + break; } return err; diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c index d29cd9aa4a67..2aa13bd7f2b2 100644 --- a/net/openvswitch/datapath.c +++ b/net/openvswitch/datapath.c @@ -712,6 +712,12 @@ static int validate_set(const struct nlattr *a, return validate_tp_port(flow_key); + case OVS_KEY_ATTR_SCTP: + if (flow_key->ip.proto != IPPROTO_SCTP) + return -EINVAL; + + return validate_tp_port(flow_key); + default: return -EINVAL; } diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c index 1fceb9653598..2b4785590b56 100644 --- a/net/openvswitch/flow.c +++ b/net/openvswitch/flow.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include @@ -129,6 +130,7 @@ static bool ovs_match_validate(const struct sw_flow_match *match, | (1 << OVS_KEY_ATTR_IPV6) | (1 << OVS_KEY_ATTR_TCP) | (1 << OVS_KEY_ATTR_UDP) + | (1 << OVS_KEY_ATTR_SCTP) | (1 << OVS_KEY_ATTR_ICMP) | (1 << OVS_KEY_ATTR_ICMPV6) | (1 << OVS_KEY_ATTR_ARP) @@ -159,6 +161,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match, mask_allowed |= 1 << OVS_KEY_ATTR_UDP; } + if (match->key->ip.proto == IPPROTO_SCTP) { + key_expected |= 1 << OVS_KEY_ATTR_SCTP; + if (match->mask && (match->mask->key.ip.proto == 0xff)) + mask_allowed |= 1 << OVS_KEY_ATTR_SCTP; + } + if (match->key->ip.proto == IPPROTO_TCP) { key_expected |= 1 << OVS_KEY_ATTR_TCP; if (match->mask && (match->mask->key.ip.proto == 0xff)) @@ -185,6 +193,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match, mask_allowed |= 1 << OVS_KEY_ATTR_UDP; } + if (match->key->ip.proto == IPPROTO_SCTP) { + key_expected |= 1 << OVS_KEY_ATTR_SCTP; + if (match->mask && (match->mask->key.ip.proto == 0xff)) + mask_allowed |= 1 << OVS_KEY_ATTR_SCTP; + } + if (match->key->ip.proto == IPPROTO_TCP) { key_expected |= 1 << OVS_KEY_ATTR_TCP; if (match->mask && (match->mask->key.ip.proto == 0xff)) @@ -280,6 +294,12 @@ static bool udphdr_ok(struct sk_buff *skb) sizeof(struct udphdr)); } +static bool sctphdr_ok(struct sk_buff *skb) +{ + return pskb_may_pull(skb, skb_transport_offset(skb) + + sizeof(struct sctphdr)); +} + static bool icmphdr_ok(struct sk_buff *skb) { return pskb_may_pull(skb, skb_transport_offset(skb) + @@ -891,6 +911,12 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key) key->ipv4.tp.src = udp->source; key->ipv4.tp.dst = udp->dest; } + } else if (key->ip.proto == IPPROTO_SCTP) { + if (sctphdr_ok(skb)) { + struct sctphdr *sctp = sctp_hdr(skb); + key->ipv4.tp.src = sctp->source; + key->ipv4.tp.dst = sctp->dest; + } } else if (key->ip.proto == IPPROTO_ICMP) { if (icmphdr_ok(skb)) { struct icmphdr *icmp = icmp_hdr(skb); @@ -953,6 +979,12 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key) key->ipv6.tp.src = udp->source; key->ipv6.tp.dst = udp->dest; } + } else if (key->ip.proto == NEXTHDR_SCTP) { + if (sctphdr_ok(skb)) { + struct sctphdr *sctp = sctp_hdr(skb); + key->ipv6.tp.src = sctp->source; + key->ipv6.tp.dst = sctp->dest; + } } else if (key->ip.proto == NEXTHDR_ICMP) { if (icmp6hdr_ok(skb)) { error = parse_icmpv6(skb, key, nh_len); @@ -1087,6 +1119,7 @@ const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = { [OVS_KEY_ATTR_IPV6] = sizeof(struct ovs_key_ipv6), [OVS_KEY_ATTR_TCP] = sizeof(struct ovs_key_tcp), [OVS_KEY_ATTR_UDP] = sizeof(struct ovs_key_udp), + [OVS_KEY_ATTR_SCTP] = sizeof(struct ovs_key_sctp), [OVS_KEY_ATTR_ICMP] = sizeof(struct ovs_key_icmp), [OVS_KEY_ATTR_ICMPV6] = sizeof(struct ovs_key_icmpv6), [OVS_KEY_ATTR_ARP] = sizeof(struct ovs_key_arp), @@ -1500,6 +1533,24 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match, u64 attrs, attrs &= ~(1 << OVS_KEY_ATTR_UDP); } + if (attrs & (1 << OVS_KEY_ATTR_SCTP)) { + const struct ovs_key_sctp *sctp_key; + + sctp_key = nla_data(a[OVS_KEY_ATTR_SCTP]); + if (orig_attrs & (1 << OVS_KEY_ATTR_IPV4)) { + SW_FLOW_KEY_PUT(match, ipv4.tp.src, + sctp_key->sctp_src, is_mask); + SW_FLOW_KEY_PUT(match, ipv4.tp.dst, + sctp_key->sctp_dst, is_mask); + } else { + SW_FLOW_KEY_PUT(match, ipv6.tp.src, + sctp_key->sctp_src, is_mask); + SW_FLOW_KEY_PUT(match, ipv6.tp.dst, + sctp_key->sctp_dst, is_mask); + } + attrs &= ~(1 << OVS_KEY_ATTR_SCTP); + } + if (attrs & (1 << OVS_KEY_ATTR_ICMP)) { const struct ovs_key_icmp *icmp_key; @@ -1843,6 +1894,20 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, udp_key->udp_src = output->ipv6.tp.src; udp_key->udp_dst = output->ipv6.tp.dst; } + } else if (swkey->ip.proto == IPPROTO_SCTP) { + struct ovs_key_sctp *sctp_key; + + nla = nla_reserve(skb, OVS_KEY_ATTR_SCTP, sizeof(*sctp_key)); + if (!nla) + goto nla_put_failure; + sctp_key = nla_data(nla); + if (swkey->eth.type == htons(ETH_P_IP)) { + sctp_key->sctp_src = swkey->ipv4.tp.src; + sctp_key->sctp_dst = swkey->ipv4.tp.dst; + } else if (swkey->eth.type == htons(ETH_P_IPV6)) { + sctp_key->sctp_src = swkey->ipv6.tp.src; + sctp_key->sctp_dst = swkey->ipv6.tp.dst; + } } else if (swkey->eth.type == htons(ETH_P_IP) && swkey->ip.proto == IPPROTO_ICMP) { struct ovs_key_icmp *icmp_key; diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h index 9674e45f6969..d08dcf78dbf3 100644 --- a/net/openvswitch/flow.h +++ b/net/openvswitch/flow.h @@ -99,8 +99,8 @@ struct sw_flow_key { } addr; union { struct { - __be16 src; /* TCP/UDP source port. */ - __be16 dst; /* TCP/UDP destination port. */ + __be16 src; /* TCP/UDP/SCTP source port. */ + __be16 dst; /* TCP/UDP/SCTP destination port. */ } tp; struct { u8 sha[ETH_ALEN]; /* ARP source hardware address. */ @@ -115,8 +115,8 @@ struct sw_flow_key { } addr; __be32 label; /* IPv6 flow label. */ struct { - __be16 src; /* TCP/UDP source port. */ - __be16 dst; /* TCP/UDP destination port. */ + __be16 src; /* TCP/UDP/SCTP source port. */ + __be16 dst; /* TCP/UDP/SCTP destination port. */ } tp; struct { struct in6_addr target; /* ND target address. */ -- cgit v1.2.3