Subject: Sun-3/80 "le0: memory error"
To: None <port-sun3@NetBSD.ORG>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: port-sun3
Date: 08/07/1997 11:41:13
I'm running NetBSD/sun3x in a Sun-3/80.  I have no particular reason to
think the hardware is bad, except as described below; it was running
SunOS just fine until recently.

The world is built from the same July 8th source tree I use everywhere
else.  Everything works fine...EXCEPT...once it's connected to the
network, I start getting "le0: memory error".  They seem to be related
to traffic; if the machine is idle (ie, receiving only link-layer
broadcasts), they show up at the rate of four or five an hour.
Blasting data at it from another machine connected via Ethernet
produces such messages at a rate more like one every few seconds.  But
talking to my home machines, which are on the other end of a serial
line netlink, does not visibly increase the rate of the messages.

I have also seen one "le0: dropping chained buffer" message.

If I unplug the drop cable (it's got AUI Ethernet), the messages stop
completely - I've never seen even one occur with the drop cable off.

These messages have been showing up from the very first NetBSD kernel I
ever booted on the machine.  Does this indicate flaky hardware that
either SunOS didn't diagnose or went bad since we brought SunOS down?
Is it normal and I should ignore it(!)?  Known bug?  Something else?

The main reason I'm concerned is that I did one largish data transfer
that provoked fifteen or so of those messages, and a couple of bytes in
the middle got corrupted, despite the Ethernet CRC and TCP's
checksumming...which makes me suspect it's a real problem that the
checksums are stopping, most of the time.

					der Mouse

		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B