Update netisr handling; Each SWI now registers its queue, and all queue

drain routines are done by swi_net, which allows for better queue control
at some future point.  Packets may also be directly dispatched to a netisr
instead of queued, this may be of interest at some installations, but
currently defaults to off.

Reviewed by: hsu, silby, jayanth, sam
Sponsored by: DARPA, NAI Labs

1 files changed, 46 insertions, 49 deletions

diff --git a/sys/net/if_ef.c b/sys/net/if_ef.c
index e136a6a..86f73c1 100644
--- a/sys/net/if_ef.c
+++ b/sys/net/if_ef.c
@@ -240,70 +240,74 @@ ef_start(struct ifnet *ifp)
  * parameter passing but simplify the code
  */
 static int __inline
-ef_inputEII(struct mbuf *m, struct ether_header *eh,
-	u_short ether_type, struct ifqueue **inq)
+ef_inputEII(struct mbuf *m, struct ether_header *eh, u_short ether_type)
 {
+	int isr;
+
 	switch(ether_type) {
 #ifdef IPX
-	    case ETHERTYPE_IPX:
-		schednetisr(NETISR_IPX);
-		*inq = &ipxintrq;
+	case ETHERTYPE_IPX:
+		isr = NETISR_IPX;
 		break;
 #endif
 #ifdef INET
-	    case ETHERTYPE_IP:
+	case ETHERTYPE_IP:
 		if (ipflow_fastforward(m))
-			return 1;
-		schednetisr(NETISR_IP);
-		*inq = &ipintrq;
+			return (0);
+		isr = NETISR_IP;
 		break;
 
-	    case ETHERTYPE_ARP:
-		schednetisr(NETISR_ARP);
-		*inq = &arpintrq;
+	case ETHERTYPE_ARP:
+		isr = NETISR_ARP;
 		break;
 #endif
-	    default:
-		return EPROTONOSUPPORT;
+	default:
+		return (EPROTONOSUPPORT);
 	}
-	return 0;
+	netisr_dispatch(isr, m);
+	return (0);
 }
 
 static int __inline
 ef_inputSNAP(struct mbuf *m, struct ether_header *eh, struct llc* l,
-	u_short ether_type, struct ifqueue **inq)
+	u_short ether_type)
 {
+	int isr;
+
 	switch(ether_type) {
 #ifdef IPX
-	    case ETHERTYPE_IPX:
+	case ETHERTYPE_IPX:
 		m_adj(m, 8);
-		schednetisr(NETISR_IPX);
-		*inq = &ipxintrq;
+		isr = NETISR_IPX;
 		break;
 #endif
-	    default:
-		return EPROTONOSUPPORT;
+	default:
+		return (EPROTONOSUPPORT);
 	}
-	return 0;
+	netisr_dispatch(isr, m);
+	return (0);
 }
 
 static int __inline
 ef_input8022(struct mbuf *m, struct ether_header *eh, struct llc* l,
-	u_short ether_type, struct ifqueue **inq)
+	u_short ether_type)
 {
+	int isr;
+
 	switch(ether_type) {
 #ifdef IPX
-	    case 0xe0:
+	case 0xe0:
 		m_adj(m, 3);
-		schednetisr(NETISR_IPX);
-		*inq = &ipxintrq;
+		isr = NETISR_IPX;
 		break;
 #endif
-	    default:
-		return EPROTONOSUPPORT;
+	default:
+		return (EPROTONOSUPPORT);
 	}
-	return 0;
+	netisr_dispatch(isr, m);
+	return (0);
 }
+
 /*
  * Called from ether_input()
  */
@@ -312,11 +316,11 @@ ef_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
 {
 	u_short ether_type;
 	int ft = -1;
-	struct ifqueue *inq;
 	struct efnet *efp;
 	struct ifnet *eifp;
 	struct llc *l;
 	struct ef_link *efl;
+	int isr;
 
 	ether_type = ntohs(eh->ether_type);
 	if (ether_type < ETHERMTU) {
@@ -377,35 +381,28 @@ ef_input(struct ifnet *ifp, struct ether_header *eh, struct mbuf *m)
 	/*
 	 * Now we ready to adjust mbufs and pass them to protocol intr's
 	 */
-	inq = NULL;
 	switch(ft) {
-	    case ETHER_FT_EII:
-		if (ef_inputEII(m, eh, ether_type, &inq) != 0)
-			return EPROTONOSUPPORT;
+	case ETHER_FT_EII:
+		return (ef_inputEII(m, eh, ether_type));
 		break;
 #ifdef IPX
-	    case ETHER_FT_8023:		/* only IPX can be here */
-		schednetisr(NETISR_IPX);
-		inq = &ipxintrq;
+	case ETHER_FT_8023:		/* only IPX can be here */
+		isr = NETISR_IPX;
 		break;
 #endif
-	    case ETHER_FT_SNAP:
-		if (ef_inputSNAP(m, eh, l, ether_type, &inq) != 0)
-			return EPROTONOSUPPORT;
+	case ETHER_FT_SNAP:
+		return (ef_inputSNAP(m, eh, l, ether_type));
 		break;
-	    case ETHER_FT_8022:
-		if (ef_input8022(m, eh, l, ether_type, &inq) != 0)
-			return EPROTONOSUPPORT;
+	case ETHER_FT_8022:
+		return (ef_input8022(m, eh, l, ether_type));
 		break;
-	}
-
-	if (inq == NULL) {
+	default:
 		EFDEBUG("No support for frame %d and proto %04x\n",
 			ft, ether_type);
-		return EPROTONOSUPPORT;
+		return (EPROTONOSUPPORT);
 	}
-	(void) IF_HANDOFF(inq, m, NULL);
-	return 0;
+	netisr_dispatch(isr, m);
+	return (0);
 }
 
 static int