Rework netisr policy mechanism so that per-protocol dispatch policies can

be represented:

- A single policy namespace is defined, consisting of four possible
  policies: "default" to use the global default, "deferred" to force
  deferred dispatch, "direct" to employ direct dispatch where possible, and
  "hybrid" which makes a dynamic decision based on CPU affinity, ordering,
  etc.  Routines are implemented to convert between strings and an integer
  namespace.

- A new global variable, netisr_dispatch_policy, subsumes existing global
  variables for direct dispatch, forced direct dispatch, etc, and is used
  for explicit policy interpretation and composition.  Old variables remain
  so that they can be exported by legacy sysctls for use by old netstat(1)
  binaries.  A new sysctl and tunable, netisr.dispatch.policy, accepts the
  above strings for specifying a global policy default.

- The protocol registration structure, netisr_handler, grows an nh_dispatch
  field, which accepts a per-policy policy override.  The default value is
  '0', which corresponds to "default", meaning that protocols will accept
  the global default policy unless otherwise specified.

- Policies are now interpreted and composed explicitly at various points in
  packet dispatch; protocol policies override global policies.

- Protocols grow the ability to express a non-opinion about affinity even
  when implenting m2cpuid by returning NETISR_CPUID_NONE.  In that case, the
  framework falls back on source ordering, rather than simply using the
  current CPU.

These changes are in support of allowing link layer re-dispatch based on
RSS or similar hashes provided by NICs, especially in the case where the
number of hardware receive queues matches hardware core count, rather than
hardware thread count, requiring further software redistributeon.  (i.e.,
on RMI XLR).

MFC after:      3 weeks
Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.

1 files changed, 16 insertions, 3 deletions

diff --git a/sys/net/netisr.h b/sys/net/netisr.h
index cd692f6d601d..83bf9ce5d4d1 100644
--- a/sys/net/netisr.h
+++ b/sys/net/netisr.h
@@ -1,6 +1,6 @@
 /*-
  * Copyright (c) 2007-2009 Robert N. M. Watson
- * Copyright (c) 2010 Juniper Networks, Inc.
+ * Copyright (c) 2010-2011 Juniper Networks, Inc.
  * All rights reserved.
  *
  * This software was developed by Robert N. M. Watson under contract
@@ -71,6 +71,15 @@
 #define	NETISR_POLICY_CPU	3	/* Protocol determines CPU placement. */
 
 /*
+ * Protocol dispatch policy constants; selects whether and when direct
+ * dispatch is permitted.
+ */
+#define	NETISR_DISPATCH_DEFAULT		0	/* Use global default. */
+#define	NETISR_DISPATCH_DEFERRED	1	/* Always defer dispatch. */
+#define	NETISR_DISPATCH_HYBRID		2	/* Allow hybrid dispatch. */
+#define	NETISR_DISPATCH_DIRECT		3	/* Always direct dispatch. */
+
+/*
  * Monitoring data structures, exported by sysctl(2).
  *
  * Three sysctls are defined.  First, a per-protocol structure exported by
@@ -84,7 +93,8 @@ struct sysctl_netisr_proto {
 	u_int	snp_qlimit;			/* nh_qlimit */
 	u_int	snp_policy;			/* nh_policy */
 	u_int	snp_flags;			/* Various flags. */
-	u_int	_snp_ispare[7];
+	u_int	snp_dispatch;			/* Dispatch policy. */
+	u_int	_snp_ispare[6];
 };
 
 /*
@@ -173,6 +183,8 @@ typedef struct mbuf	*netisr_m2cpuid_t(struct mbuf *m, uintptr_t source,
 typedef	struct mbuf	*netisr_m2flow_t(struct mbuf *m, uintptr_t source);
 typedef void		 netisr_drainedcpu_t(u_int cpuid);
 
+#define	NETISR_CPUID_NONE	((u_int)-1)	/* No affinity returned. */
+
 /*
  * Data structure describing a protocol handler.
  */
@@ -185,7 +197,8 @@ struct netisr_handler {
 	u_int		 nh_proto;	/* Integer protocol ID. */
 	u_int		 nh_qlimit;	/* Maximum per-CPU queue depth. */
 	u_int		 nh_policy;	/* Work placement policy. */
-	u_int		 nh_ispare[5];	/* For future use. */
+	u_int		 nh_dispatch;	/* Dispatch policy. */
+	u_int		 nh_ispare[4];	/* For future use. */
 	void		*nh_pspare[4];	/* For future use. */
 };