Subject: Problems with wm(4) with ip/tcp/udp checksum
To: None <email@example.com>
From: A.R.S. <firstname.lastname@example.org>
Date: 05/01/2005 18:44:14
I have the folowing setup: a ring of 6 netbsd-2-0-2 routers all having
some fxp(4) and the folowing wm(4) GigE cards:
wm0 at pci2 dev 1 function 0: Intel i82547EI 1000BASE-T Ethernet, rev. 0
wm0: interrupting at irq 10
wm0: Communication Streaming Architecture
wm0: using 82547 Tx FIFO stall work-around
wm0: 65536 word (16 address bits) SPI EEPROM
wm0: Ethernet address 00:11:11:7e:24:f4
igphy0 at wm0 phy 1: Intel IGP01E1000 Gigabit PHY, rev. 0
igphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseT,
wm1 at pci3 dev 0 function 0: Intel i82541GI 1000BASE-T Ethernet, rev. 0
wm1: interrupting at irq 3
wm1: 32-bit 33MHz PCI bus
wm1: 64 word (6 address bits) MicroWire EEPROM
wm1: Ethernet address 00:0e:0c:64:5c:1a
igphy1 at wm1 phy 1: Intel IGP01E1000 Gigabit PHY, rev. 0
igphy1: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseT,
They are plugged in a manageable switch and have ips from a /27 network
assigned and they route via bgp/ospf .
Having a high load of traffic the ip/tcp/udp checksuming was needed on
those wm(4). But once enabled there was a drastic upload rate reduction,
from 300Mbps to 100Mbps.
By monitoring the switch FDB i noticed new mac-addresses appearing:
1 default 00-11-11-7E-24-F4 2 Dynamic
1 default 00-11-11-7E-24-F8 2 Dynamic
the first mac-address belongs to the first wm0 from up.
What is odd is that the second mac appearing on the same port where wm0
is plugged in_ is from an fxp on the same server which doesn't have any
link with the switch. So this mac-address shouldn't be advertised. The
tcpdump -e on a server plugged in the switch too but not routing
anything shows those odd packets originating with mac-addresses from
behind the switch network/routers which are of course blackholed.
Any other information needed (dmesgs, ifconfigs) is available upon request.