Re: obsd 4.6 on Netra X1

> Date: Sun, 3 Jan 2010 13:12:04 -0800 (PST)
> From: abs <[email protected]>
> 
> Hello all,
> I am running OpenBSD 4.6 on a Netra X1 as a layer2 firewall.  The problem I
> seem to be experiencing is that every time the NIC gets saturated (100Mb) i
> get the following error on the console twice followed by a kernel panic and
> core dump.
> 
> dc0: watchdog timeout
> 
> As long as I don't max out the network port (i.e 11MBps) everything works as
> expected. Is there a bug or am I doing something wrong?  thank you in
> advance.

Can you try the following diff?  It probably won't fix the watchdog
timeouts, but it should prevent the kernel panic.

Index: dc.c
===================================================================
RCS file: /cvs/src/sys/dev/ic/dc.c,v
retrieving revision 1.113
diff -u -p -r1.113 dc.c
--- dc.c	15 Oct 2009 17:54:54 -0000	1.113
+++ dc.c	12 Dec 2009 00:15:42 -0000
@@ -3062,6 +3062,7 @@ void
 dc_stop(struct dc_softc *sc, int softonly)
 {
 	struct ifnet *ifp;
+	u_int32_t isr;
 	int i;
 
 	ifp = &sc->sc_arpcom.ac_if;
@@ -3073,6 +3074,28 @@ dc_stop(struct dc_softc *sc, int softonl
 
 	if (!softonly) {
 		DC_CLRBIT(sc, DC_NETCFG, (DC_NETCFG_RX_ON|DC_NETCFG_TX_ON));
+
+		for (i = 0; i < DC_TIMEOUT; i++) {
+			isr = CSR_READ_4(sc, DC_ISR);
+			if ((isr & DC_ISR_TX_IDLE ||
+			    (isr & DC_ISR_TX_STATE) == DC_TXSTATE_RESET) &&
+			    (isr & DC_ISR_RX_STATE) == DC_RXSTATE_STOPPED)
+				break;
+			DELAY(10);
+		}
+
+		if (i == DC_TIMEOUT) {
+			if (!((isr & DC_ISR_TX_IDLE) ||
+			    (isr & DC_ISR_TX_STATE) == DC_TXSTATE_RESET) &&
+			    !DC_IS_ASIX(sc) && !DC_IS_DAVICOM(sc))
+				printf("%s: failed to force tx to idle state\n",
+				    sc->sc_dev.dv_xname);
+			if (!((isr & DC_ISR_RX_STATE) == DC_RXSTATE_STOPPED) &&
+			    !DC_HAS_BROKEN_RXSTATE(sc))
+				printf("%s: failed to force rx to idle state\n",
+				    sc->sc_dev.dv_xname);
+		}
+
 		CSR_WRITE_4(sc, DC_IMR, 0x00000000);
 		CSR_WRITE_4(sc, DC_TXADDR, 0x00000000);
 		CSR_WRITE_4(sc, DC_RXADDR, 0x00000000);