From: mikeb <mikeb@openbsd.org>
Date: Fri, 29 Jul 2016 22:01:57 +0000 (+0000)
Subject: Disable receive ring slot accounting
X-Git-Url: http://artulab.com/gitweb/?a=commitdiff_plain;h=2e8a8535ca9fd69692b134ba27ecf3105a6e0e23;p=openbsd

Disable receive ring slot accounting

In the Netfront case it appears that by having scheduler decide when
to run the workload we don't need to involve additional mechanisms
to artificially limit resource availability to achieve better
performance under heavy load.  On the contrary, by performing a
single pass through Rx and Tx completion rings and having scheduler
decide when to run the next attempt, we limit the amount of time
spent in the packet processing and achieve system responsiveness.
---

diff --git a/sys/dev/pv/if_xnf.c b/sys/dev/pv/if_xnf.c
index 5cdbac6522e..15fa81e5499 100644
--- a/sys/dev/pv/if_xnf.c
+++ b/sys/dev/pv/if_xnf.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: if_xnf.c,v 1.27 2016/07/29 18:33:12 mikeb Exp $	*/
+/*	$OpenBSD: if_xnf.c,v 1.28 2016/07/29 22:01:57 mikeb Exp $	*/
 
 /*
  * Copyright (c) 2015, 2016 Mike Belopuhov
@@ -176,7 +176,6 @@ struct xnf_softc {
 	struct mbuf		*sc_rx_buf[XNF_RX_DESC];
 	bus_dmamap_t		 sc_rx_dmap[XNF_RX_DESC]; /* maps for packets */
 	struct mbuf		*sc_rx_cbuf[2];	  	  /* chain handling */
-	struct if_rxring	 sc_rx_slots;
 	struct timeout		 sc_rx_fill;
 
 	/* Tx ring */
@@ -380,10 +379,6 @@ xnf_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
 	case SIOCSIFMEDIA:
 		error = ifmedia_ioctl(ifp, ifr, &sc->sc_media, command);
 		break;
-	case SIOCGIFRXR:
-		error = if_rxr_ioctl((struct if_rxrinfo *)ifr->ifr_data,
-		    NULL, XNF_MCLEN, &sc->sc_rx_slots);
-		break;
 	default:
 		error = ether_ioctl(ifp, &sc->sc_ac, command, data);
 		break;
@@ -714,8 +709,6 @@ xnf_rxeof(struct xnf_softc *sc)
 		KASSERT(m != NULL);
 		sc->sc_rx_buf[i] = NULL;
 
-		if_rxr_put(&sc->sc_rx_slots, 1);
-
 		if (flags & XNF_RXF_MGMT) {
 			printf("%s: management data present\n",
 			    ifp->if_xname);
@@ -780,24 +773,14 @@ xnf_rx_ring_fill(void *arg)
 	bus_dmamap_t dmap;
 	struct mbuf *m;
 	uint32_t cons, prod, oprod;
-	int i, flags, n, s;
+	int i, flags, s;
 
 	s = splnet();
 
 	cons = rxr->rxr_cons;
 	prod = oprod = rxr->rxr_prod;
 
-	n = if_rxr_get(&sc->sc_rx_slots, XNF_RX_DESC);
-
-	/* Less than XNF_RX_MIN slots available? */
-	if (n == 0 && prod - cons < XNF_RX_MIN) {
-		splx(s);
-		if (ifp->if_flags & IFF_RUNNING)
-			timeout_add(&sc->sc_rx_fill, 10);
-		return;
-	}
-
-	for (; n > 0; prod++, n--) {
+	while (prod - cons < XNF_RX_DESC) {
 		i = prod & (XNF_RX_DESC - 1);
 		if (sc->sc_rx_buf[i])
 			break;
@@ -814,22 +797,15 @@ xnf_rx_ring_fill(void *arg)
 		sc->sc_rx_buf[i] = m;
 		rxr->rxr_desc[i].rxd_req.rxq_ref = dmap->dm_segs[0].ds_addr;
 		bus_dmamap_sync(sc->sc_dmat, dmap, 0, 0, BUS_DMASYNC_PREWRITE);
+		prod++;
 	}
 
-	if (n > 0)
-		if_rxr_put(&sc->sc_rx_slots, n);
-
 	rxr->rxr_prod = prod;
 	bus_dmamap_sync(sc->sc_dmat, sc->sc_rx_rmap, 0, 0,
 	    BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
 
-	if (prod - cons < XNF_RX_MIN) {
-		splx(s);
-		if (ifp->if_flags & IFF_RUNNING)
-			timeout_add(&sc->sc_rx_fill, 10);
-		return;
-	}
-
+	if ((prod - cons < XNF_RX_DESC) && (ifp->if_flags & IFF_RUNNING))
+		timeout_add(&sc->sc_rx_fill, 10);
 	if (prod - rxr->rxr_prod_event < prod - oprod)
 		xen_intr_signal(sc->sc_xih);
 
@@ -884,7 +860,6 @@ xnf_rx_ring_create(struct xnf_softc *sc)
 		sc->sc_rx_ring->rxr_desc[i].rxd_req.rxq_id = i;
 	}
 
-	if_rxr_init(&sc->sc_rx_slots, XNF_RX_MIN, XNF_RX_DESC);
 	xnf_rx_ring_fill(sc);
 
 	return (0);
@@ -919,8 +894,6 @@ xnf_rx_ring_destroy(struct xnf_softc *sc)
 		slots++;
 	}
 
-	if_rxr_put(&sc->sc_rx_slots, slots);
-
 	for (i = 0; i < XNF_RX_DESC; i++) {
 		if (sc->sc_rx_dmap[i] == NULL)
 			continue;