> On Mon, 16 Aug 2004 18:01:02 -0700 > "David S. Miller" <davem@redhat.com> wrote: > > > On Sun, 15 Aug 2004 23:31:50 -0700 > > "David S. Miller" <davem@redhat.com> wrote: > > > > > I can't even figure out why this change makes a difference, but > > > I will investigate it some more. Thanks a lot. > > > > Tony, please revert your patch and try this one > > instead. Does it make your MTU changing test > > case work? > > Actually, scatch that, I see what the real but is, > we fail to re-enable interrupts on code paths > that use tg3_init_hw(), so try this patch instead > against a vanilla tg3.c driver Sorry, neither patch made a difference for my test case. The problem does seem to be interrupt-related though. cat /proc/interrupts shows approx. 10 interrupts on eth0 every second when the interface is working (with an idle network) but 0 interrupts per second when the interface is not working. When the interface is not working and the interrupt count is remaining constant, doing "ifconfig eth0 down; ifconfig eth0 up" brings it back to 10 interrupts per second and the interface works again. If it helps, here is some dmesg output from the driver: tg3.c:v3.8 (July 14, 2004) eth0: Tigon3 [partno(A6825-60101) rev 0105 PHY(5701)] (PCI:66MHz:64-bit) 10/100/1000BaseT Ethernet 00:30:6e:49:46:6c eth0: HostTXDS[1] RXcsums[1] LinkChgREG[0] MIirq[0] ASF[0] Split[0] WireSpeed[1] TSOcap[0] tg3: eth0: Link is up at 1000 Mbps, full duplex. tg3: eth0: Flow control is on for TX and on for RX. Anthony J. Battersy Cybernetics - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
