| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1388935422-7331-3-git-send-email-ogerlitz@mellanox.com>
Date: Sun, 5 Jan 2014 17:23:42 +0200
From: Or Gerlitz <ogerlitz@...lanox.com>
To: hkchu@...gle.com, edumazet@...gle.com, herbert@...dor.apana.org.au
Cc: netdev@...r.kernel.org, davem@...emloft.net, yanb@...lanox.com,
shlomop@...lanox.com, Or Gerlitz <ogerlitz@...lanox.com>
Subject: [PATCH V1 net-next 2/2] net: Add UDP GRO support for vxlan traffic
Add GRO handlers for UDP, with the intent of being able to coalesce vxlan
packets which encapsulate packets belonging to the same TCP session.
The UDP GRO handler will only attempt to coalesce packets whose
destination port is used as vxlan listening port. There are two
issues here for which I will be happy to get feedback
1. what is the most efficient way to determine if a udp port is
owned by the vxlan driver, e.g maybe through a bit map exported by
the vxlan driver?
2. how to prevent the host GRO stack from coalescing guest UDP
packets which went through vxlan encapsulation and happen to carry
the same udp destination port as used by the host vxlan driver..
On my setup, which is net-next (now with the mlx4 vxlan offloads patches) -- for
single TCP session that goes through vxlan tunneling I got nice improvement
from 6.8Gbs to 11.7Gbs
--> UDP/VXLAN GRO disabled
$ netperf -H 192.168.52.147 -c -C
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.52.147 () port 0 AF_INET
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 65536 65536 10.00 6821.21 14.99 29.14 0.720 1.400
--> UDP/VXLAN GRO enabled
$ netperf -H 192.168.52.147 -c -C
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.52.147 () port 0 AF_INET
Recv Send Send Utilization Service Demand
Socket Socket Message Elapsed Send Recv Send Recv
Size Size Size Time Throughput local remote local remote
bytes bytes bytes secs. 10^6bits/s % S % S us/KB us/KB
87380 65536 65536 10.00 11694.65 24.87 20.71 0.697 0.580
Signed-off-by: Or Gerlitz <ogerlitz@...lanox.com>
---
drivers/net/vxlan.c | 3 +
include/net/protocol.h | 3 +
net/ipv4/udp_offload.c | 127 ++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 133 insertions(+), 0 deletions(-)
diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index e3d8183..c1d6647 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -40,6 +40,7 @@
#include <net/net_namespace.h>
#include <net/netns/generic.h>
#include <net/vxlan.h>
+#include <net/protocol.h>
#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6.h>
#include <net/addrconf.h>
@@ -562,6 +563,7 @@ static void vxlan_notify_add_rx_port(struct sock *sk)
dev->netdev_ops->ndo_add_vxlan_port(dev, sa_family,
port);
}
+ udp_gro_add_vxlan_port(port);
rcu_read_unlock();
}
@@ -579,6 +581,7 @@ static void vxlan_notify_del_rx_port(struct sock *sk)
dev->netdev_ops->ndo_del_vxlan_port(dev, sa_family,
port);
}
+ udp_gro_del_vxlan_port(port);
rcu_read_unlock();
}
diff --git a/include/net/protocol.h b/include/net/protocol.h
index fbf7676..ca42395 100644
--- a/include/net/protocol.h
+++ b/include/net/protocol.h
@@ -103,6 +103,9 @@ int inet_del_offload(const struct net_offload *prot, unsigned char num);
void inet_register_protosw(struct inet_protosw *p);
void inet_unregister_protosw(struct inet_protosw *p);
+void udp_gro_add_vxlan_port(__be16 port);
+void udp_gro_del_vxlan_port(__be16 port);
+
#if IS_ENABLED(CONFIG_IPV6)
int inet6_add_protocol(const struct inet6_protocol *prot, unsigned char num);
int inet6_del_protocol(const struct inet6_protocol *prot, unsigned char num);
diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
index bd09f65..2462469 100644
--- a/net/ipv4/udp_offload.c
+++ b/net/ipv4/udp_offload.c
@@ -13,6 +13,7 @@
#include <linux/skbuff.h>
#include <net/udp.h>
#include <net/protocol.h>
+#include <net/vxlan.h>
static int udp4_ufo_send_check(struct sk_buff *skb)
{
@@ -90,10 +91,136 @@ out:
return segs;
}
+
+static u8 udp_gro_vxlan_ports[1 << 16];
+
+void udp_gro_add_vxlan_port(__be16 port)
+{
+ udp_gro_vxlan_ports[port] = 1;
+}
+EXPORT_SYMBOL(udp_gro_add_vxlan_port);
+
+void udp_gro_del_vxlan_port(__be16 port)
+{
+ udp_gro_vxlan_ports[port] = 0;
+}
+EXPORT_SYMBOL(udp_gro_del_vxlan_port);
+
+static struct sk_buff **udp_gro_receive(struct sk_buff **head, struct sk_buff *skb)
+{
+ struct sk_buff *p, **pp = NULL;
+ struct udphdr *uh, *uh2;
+ struct vxlanhdr *vh, *vh2;
+ struct ethhdr *eh;
+ unsigned int hlen, off_udp, off_vxlan, off_eth;
+ struct packet_offload *ptype;
+ __be16 type;
+ int flush = 1;
+
+ off_udp = skb_gro_offset(skb);
+ hlen = off_udp + sizeof(*uh);
+ uh = skb_gro_header_fast(skb, off_udp);
+ if (skb_gro_header_hard(skb, hlen)) {
+ uh = skb_gro_header_slow(skb, hlen, off_udp);
+ if (unlikely(!uh))
+ goto out;
+ }
+
+ if (!udp_gro_vxlan_ports[uh->dest] || !skb->encapsulation)
+ goto out;
+
+ skb_gro_pull(skb, sizeof(struct udphdr)); /* pull encapsulating udp header */
+
+ off_vxlan = skb_gro_offset(skb);
+ hlen = off_vxlan + sizeof(*vh);
+ vh = skb_gro_header_fast(skb, off_vxlan);
+ if (skb_gro_header_hard(skb, hlen)) {
+ vh = skb_gro_header_slow(skb, hlen, off_vxlan);
+ if (unlikely(!vh))
+ goto out;
+ }
+ skb_gro_pull(skb, sizeof(struct vxlanhdr)); /* pull vxlan header */
+ flush = 0;
+
+
+ for (p = *head; p; p = p->next) {
+ if (!NAPI_GRO_CB(p)->same_flow)
+ continue;
+
+ uh2 = (struct udphdr *)(p->data + off_udp);
+ vh2 = (struct vxlanhdr *)(p->data + off_vxlan);
+
+ if ((*(u32 *)&uh->source ^ *(u32 *)&uh2->source) |
+ (vh->vx_vni ^ vh2->vx_vni)) {
+ NAPI_GRO_CB(p)->same_flow = 0;
+ continue;
+ }
+
+ goto found;
+ }
+
+found:
+
+ off_eth = skb_gro_offset(skb);
+ hlen = off_eth + sizeof(*eh);
+ eh = skb_gro_header_fast(skb, off_eth);
+ if (skb_gro_header_hard(skb, hlen)) {
+ eh = skb_gro_header_slow(skb, hlen, off_eth);
+ if (unlikely(!eh))
+ goto out;
+ }
+ type = eh->h_proto;
+ skb_gro_pull(skb, sizeof(*eh)); /* pull inner eth header */
+
+ rcu_read_lock();
+ ptype = gro_find_receive_by_type(type);
+ if (ptype == NULL) {
+ flush = 1;
+ goto out_unlock;
+ }
+
+ pp = ptype->callbacks.gro_receive(head, skb);
+
+out_unlock:
+ rcu_read_unlock();
+out:
+ NAPI_GRO_CB(skb)->flush |= flush;
+
+ return pp;
+}
+
+static int udp_gro_complete(struct sk_buff *skb, int nhoff)
+{
+ __be16 newlen = htons(skb->len - nhoff);
+ struct udphdr *uh = (struct udphdr *)(skb->data + nhoff);
+ struct ethhdr *eh;
+ struct packet_offload *ptype;
+ __be16 type;
+ /* 22 = 8 bytes for the vlxan header + 14 bytes for the inner eth header */
+ int vxlan_len = 22;
+ int err = -ENOENT;
+
+ uh->len = newlen;
+
+ /* 16 = 8 bytes for the udp header + 8 bytes for the vxlan header */
+ eh = (struct ethhdr *)(skb->data + nhoff + 16);
+ type = eh->h_proto;
+
+ rcu_read_lock();
+ ptype = gro_find_complete_by_type(type);
+ if (ptype != NULL)
+ err = ptype->callbacks.gro_complete(skb, nhoff + sizeof(struct udphdr) + vxlan_len);
+
+ rcu_read_unlock();
+ return err;
+}
+
static const struct net_offload udpv4_offload = {
.callbacks = {
.gso_send_check = udp4_ufo_send_check,
.gso_segment = udp4_ufo_fragment,
+ .gro_receive = udp_gro_receive,
+ .gro_complete = udp_gro_complete,
},
};
--
1.7.1
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists