[ovs-dev] [PATCH v4 2/4] datapath: Add SCTP support
Joe Stringer
joe at wand.net.nz
Wed Jul 31 00:31:56 UTC 2013
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 <horms at verge.net.au>
Signed-off-by: Joe Stringer <joe at wand.net.nz>
---
This patch introduces sparse warnings in some cases when calling
sctp_end_cksum(), due to a bug that was fixed in v3.10:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=eee1d5a14780b9391ec51f3feaf4cffb521ddbb1
v4: Rebase against megaflow
Move compute_sctp_checksum to compat/
Move openvswitch.h header changes in from other patch
Don't set CHECKSUM_UNNECESSARY on port-rewrite
Remove checksumming from Xen compat code
v3: Rebase
Refactor sctp checksum computation
Handle skb fragments for checksum calculation
Fix sparse errors
v2: Remove SCTP checksum recalculation when changing IP address
Calculate checksums as delta from incoming checksum
---
datapath/actions.c | 39 +++++++++++++
datapath/datapath.c | 6 ++
datapath/flow.c | 65 +++++++++++++++++++++
datapath/flow.h | 8 +--
datapath/linux/Modules.mk | 4 +-
datapath/linux/compat/include/linux/sctp.h | 13 +++++
datapath/linux/compat/include/net/ipv6.h | 4 ++
datapath/linux/compat/include/net/sctp/checksum.h | 28 +++++++++
include/linux/openvswitch.h | 6 ++
9 files changed, 168 insertions(+), 5 deletions(-)
create mode 100644 datapath/linux/compat/include/linux/sctp.h
create mode 100644 datapath/linux/compat/include/net/sctp/checksum.h
diff --git a/datapath/actions.c b/datapath/actions.c
index 0a2def6..37c1a64 100644
--- a/datapath/actions.c
+++ b/datapath/actions.c
@@ -22,6 +22,7 @@
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/openvswitch.h>
+#include <linux/sctp.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/in6.h>
@@ -31,6 +32,7 @@
#include <net/ipv6.h>
#include <net/checksum.h>
#include <net/dsfield.h>
+#include <net/sctp/checksum.h>
#include "checksum.h"
#include "datapath.h"
@@ -360,6 +362,39 @@ 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_clear_rxhash(skb);
+ }
+
+ return 0;
+}
+
static int do_output(struct datapath *dp, struct sk_buff *skb, int out_port)
{
struct vport *vport;
@@ -469,6 +504,10 @@ 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/datapath/datapath.c b/datapath/datapath.c
index 4330ce3..ca2a0fb 100644
--- a/datapath/datapath.c
+++ b/datapath/datapath.c
@@ -726,6 +726,12 @@ 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/datapath/flow.c b/datapath/flow.c
index 3acc743..92a592a 100644
--- a/datapath/flow.c
+++ b/datapath/flow.c
@@ -34,6 +34,7 @@
#include <linux/if_arp.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
+#include <linux/sctp.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/icmp.h>
@@ -130,6 +131,7 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
| (1ULL << OVS_KEY_ATTR_IPV6)
| (1ULL << OVS_KEY_ATTR_TCP)
| (1ULL << OVS_KEY_ATTR_UDP)
+ | (1ULL << OVS_KEY_ATTR_SCTP)
| (1ULL << OVS_KEY_ATTR_ICMP)
| (1ULL << OVS_KEY_ATTR_ICMPV6)
| (1ULL << OVS_KEY_ATTR_ARP)
@@ -163,6 +165,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
mask_allowed |= 1ULL << OVS_KEY_ATTR_UDP;
}
+ if (match->key->ip.proto == IPPROTO_SCTP) {
+ key_expected |= 1ULL << OVS_KEY_ATTR_SCTP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1ULL << OVS_KEY_ATTR_SCTP;
+ }
+
if (match->key->ip.proto == IPPROTO_TCP) {
key_expected |= 1ULL << OVS_KEY_ATTR_TCP;
if (match->mask && (match->mask->key.ip.proto == 0xff))
@@ -189,6 +197,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
mask_allowed |= 1ULL << OVS_KEY_ATTR_UDP;
}
+ if (match->key->ip.proto == IPPROTO_SCTP) {
+ key_expected |= 1ULL << OVS_KEY_ATTR_SCTP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1ULL << OVS_KEY_ATTR_SCTP;
+ }
+
if (match->key->ip.proto == IPPROTO_TCP) {
key_expected |= 1ULL << OVS_KEY_ATTR_TCP;
if (match->mask && (match->mask->key.ip.proto == 0xff))
@@ -284,6 +298,12 @@ 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) +
@@ -903,6 +923,12 @@ 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);
@@ -965,6 +991,12 @@ 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);
@@ -1099,6 +1131,7 @@ 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),
@@ -1510,6 +1543,24 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match, u64 attrs,
attrs &= ~(1ULL << OVS_KEY_ATTR_UDP);
}
+ if (attrs & (1ULL << OVS_KEY_ATTR_SCTP)) {
+ const struct ovs_key_sctp *sctp_key;
+
+ sctp_key = nla_data(a[OVS_KEY_ATTR_SCTP]);
+ if (orig_attrs & (1ULL << 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 &= ~(1ULL << OVS_KEY_ATTR_SCTP);
+ }
+
if (attrs & (1ULL << OVS_KEY_ATTR_ICMP)) {
const struct ovs_key_icmp *icmp_key;
@@ -1833,6 +1884,20 @@ 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/datapath/flow.h b/datapath/flow.h
index 59c7f6e..b8cac71 100644
--- a/datapath/flow.h
+++ b/datapath/flow.h
@@ -85,8 +85,8 @@ 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. */
@@ -101,8 +101,8 @@ 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. */
diff --git a/datapath/linux/Modules.mk b/datapath/linux/Modules.mk
index edaeabb..2162372 100644
--- a/datapath/linux/Modules.mk
+++ b/datapath/linux/Modules.mk
@@ -56,6 +56,7 @@ openvswitch_headers += \
linux/compat/include/linux/rcupdate.h \
linux/compat/include/linux/reciprocal_div.h \
linux/compat/include/linux/rtnetlink.h \
+ linux/compat/include/linux/sctp.h \
linux/compat/include/linux/skbuff.h \
linux/compat/include/linux/slab.h \
linux/compat/include/linux/stddef.h \
@@ -80,4 +81,5 @@ openvswitch_headers += \
linux/compat/include/net/route.h \
linux/compat/include/net/sock.h \
linux/compat/include/net/netns/generic.h \
- linux/compat/include/net/vxlan.h
+ linux/compat/include/net/vxlan.h \
+ linux/compat/include/net/sctp/checksum.h
diff --git a/datapath/linux/compat/include/linux/sctp.h b/datapath/linux/compat/include/linux/sctp.h
new file mode 100644
index 0000000..b91c9c6
--- /dev/null
+++ b/datapath/linux/compat/include/linux/sctp.h
@@ -0,0 +1,13 @@
+#ifndef __LINUX_SCTP_WRAPPER_H
+#define __LINUX_SCTP_WRAPPER_H 1
+
+#include_next <linux/sctp.h>
+
+#ifndef HAVE_SKBUFF_HEADER_HELPERS
+static inline struct sctphdr *sctp_hdr(const struct sk_buff *skb)
+{
+ return (struct sctphdr *)skb_transport_header(skb);
+}
+#endif /* HAVE_SKBUFF_HEADER_HELPERS */
+
+#endif
diff --git a/datapath/linux/compat/include/net/ipv6.h b/datapath/linux/compat/include/net/ipv6.h
index 7ab234a..71f4708 100644
--- a/datapath/linux/compat/include/net/ipv6.h
+++ b/datapath/linux/compat/include/net/ipv6.h
@@ -5,6 +5,10 @@
#include_next <net/ipv6.h>
+#ifndef NEXTHDR_SCTP
+#define NEXTHDR_SCTP 132 /* Stream Control Transport Protocol */
+#endif
+
#if LINUX_VERSION_CODE < KERNEL_VERSION(3,3,0)
#define ipv6_skip_exthdr rpl_ipv6_skip_exthdr
extern int ipv6_skip_exthdr(const struct sk_buff *skb, int start,
diff --git a/datapath/linux/compat/include/net/sctp/checksum.h b/datapath/linux/compat/include/net/sctp/checksum.h
new file mode 100644
index 0000000..f1c2819
--- /dev/null
+++ b/datapath/linux/compat/include/net/sctp/checksum.h
@@ -0,0 +1,28 @@
+#ifndef __SCTP_CHECKSUM_WRAPPER_H
+#define __SCTP_CHECKSUM_WRAPPER_H 1
+
+#include <linux/version.h>
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25)
+#include <net/sctp/sctp.h>
+#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) */
+#include_next <net/sctp/checksum.h>
+#endif
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(3,12,0)
+static inline __le32 sctp_compute_cksum(const struct sk_buff *skb,
+ unsigned int offset)
+{
+ const struct sk_buff *iter;
+
+ __u32 crc32 = sctp_start_cksum(skb->data + offset,
+ skb_headlen(skb) - offset);
+ skb_walk_frags(skb, iter)
+ crc32 = sctp_update_cksum((__u8 *) iter->data,
+ skb_headlen(iter), crc32);
+
+ return sctp_end_cksum(crc32);
+}
+#endif
+
+#endif
diff --git a/include/linux/openvswitch.h b/include/linux/openvswitch.h
index a119b14..09c26b5 100644
--- a/include/linux/openvswitch.h
+++ b/include/linux/openvswitch.h
@@ -282,6 +282,7 @@ 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 */
@@ -364,6 +365,11 @@ 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;
--
1.7.10.4
More information about the dev
mailing list