Several years after initial development, merge prototype support for

linking NIC Receive Side Scaling (RSS) to the network stack's
connection-group implementation.  This prototype (and derived patches)
are in use at Juniper and several other FreeBSD-using companies, so
despite some reservations about its maturity, merge the patch to the
base tree so that it can be iteratively refined in collaboration rather
than maintained as a set of gradually diverging patch sets.

(1) Merge a software implementation of the Toeplitz hash specified in
    RSS implemented by David Malone.  This is used to allow suitable
    pcbgroup placement of connections before the first packet is
    received from the NIC.  Software hashing is generally avoided,
    however, due to high cost of the hash on general-purpose CPUs.

(2) In in_rss.c, maintain authoritative versions of RSS state intended
    to be pushed to each NIC, including keying material, hash
    algorithm/ configuration, and buckets.  Provide software-facing
    interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both
    the RSS standardised Toeplitz and a 'naive' variation with a hash
    efficient in software but with poor distribution properties.
    Implement rss_m2cpuid()to be used by netisr and other load
    balancing code to look up the CPU on which an mbuf should be
    processed.

(3) In the Ethernet link layer, allow netisr distribution using RSS as
    a source of policy as an alternative to source ordering; continue
    to default to direct dispatch (i.e., don't try and requeue packets
    for processing on the 'right' CPU if they arrive in a directly
    dispatchable context).

(4) Allow RSS to control tuning of connection groups in order to align
    groups with RSS buckets.  If a packet arrives on a protocol using
    connection groups, and contains a suitable hardware-generated
    hash, use that hash value to select the connection group for pcb
    lookup for both IPv4 and IPv6.  If no hardware-generated Toeplitz
    hash is available, we fall back on regular PCB lookup risking
    contention rather than pay the cost of Toeplitz in software --
    this is a less scalable but, at my last measurement, faster
    approach.  As core counts go up, we may want to revise this
    strategy despite CPU overhead.

Where device drivers suitably configure NICs, and connection groups /
RSS are enabled, this should avoid both lock and line contention during
connection lookup for TCP.  This commit does not modify any device
drivers to tune device RSS configuration to the global RSS
configuration; patches are in circulation to do this for at least
Chelsio T3 and Intel 1G/10G drivers.  Currently, the KPI for device
drivers is not particularly robust, nor aware of more advanced features
such as runtime reconfiguration/rebalancing.  This will hopefully prove
a useful starting point for refinement.

No MFC is scheduled as we will first want to nail down a more mature
and maintainable KPI/KBI for device drivers.

Sponsored by:   Juniper Networks (original work)
Sponsored by:   EMC/Isilon (patch update and merge)

1 files changed, 24 insertions, 1 deletions

diff --git a/sys/net/if_ethersubr.c b/sys/net/if_ethersubr.c
index 860971c..4c2bd4e 100644
--- a/sys/net/if_ethersubr.c
+++ b/sys/net/if_ethersubr.c
@@ -34,6 +34,7 @@
 #include "opt_inet6.h"
 #include "opt_netgraph.h"
 #include "opt_mbuf_profiling.h"
+#include "opt_rss.h"
 
 #include <sys/param.h>
 #include <sys/systm.h>
@@ -70,6 +71,7 @@
 #include <netinet/in.h>
 #include <netinet/in_var.h>
 #include <netinet/if_ether.h>
+#include <netinet/in_rss.h>
 #include <netinet/ip_carp.h>
 #include <netinet/ip_var.h>
 #endif
@@ -583,7 +585,22 @@ ether_input_internal(struct ifnet *ifp, struct mbuf *m)
 
 /*
  * Ethernet input dispatch; by default, direct dispatch here regardless of
- * global configuration.
+ * global configuration.  However, if RSS is enabled, hook up RSS affinity
+ * so that when deferred or hybrid dispatch is enabled, we can redistribute
+ * load based on RSS.
+ *
+ * XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
+ * not it had already done work distribution via multi-queue.  Then we could
+ * direct dispatch in the event load balancing was already complete and
+ * handle the case of interfaces with different capabilities better.
+ *
+ * XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
+ * at multiple layers?
+ *
+ * XXXRW: For now, enable all this only if RSS is compiled in, although it
+ * works fine without RSS.  Need to characterise the performance overhead
+ * of the detour through the netisr code in the event the result is always
+ * direct dispatch.
  */
 static void
 ether_nh_input(struct mbuf *m)
@@ -596,8 +613,14 @@ static struct netisr_handler	ether_nh = {
 	.nh_name = "ether",
 	.nh_handler = ether_nh_input,
 	.nh_proto = NETISR_ETHER,
+#ifdef RSS
+	.nh_policy = NETISR_POLICY_CPU,
+	.nh_dispatch = NETISR_DISPATCH_DIRECT,
+	.nh_m2cpuid = rss_m2cpuid,
+#else
 	.nh_policy = NETISR_POLICY_SOURCE,
 	.nh_dispatch = NETISR_DISPATCH_DIRECT,
+#endif
 };
 
 static void