Subject: xi0 dying
To: NetBSD TechNet <tech-net@netbsd.org>
From: Jonathan Perkin <sketch@rd.bbc.co.uk>
List: tech-net
Date: 10/11/2001 14:36:42
[Not sure if this is better list than tech-kern for this query]

Another quickie, I simply cannot transfer a large file onto my NetBSD
laptop with xi0 ethernet as it always cuts off half way and subsequent
attempts to send any data over the network fails.

It's an 8Mb cisco image, and cuts off at random points within the file.
Odd though because I can spend days transferring small amounts of data
around (ie cvsup), but doing a large and fast one-off transfer just
kills it.

I understand there have been problems with xi0 driver in the past, no
doubt due to the quality of the actual card rather than the driver,
but I'm not sure that I've had this problem before.

maxusers is set to 64.
NetBSD 1.5Y (CARDBUS-$Revision: 1.39 $) #0: Wed Oct 10 16:07:31 BST 2001

-% netstat -m
5 mbufs in use:
        1 mbufs allocated to data
        4 mbufs allocated to packet headers
28/36 mapped pages in use
116 Kbytes allocated to network (79% in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines

xi0: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        address: 00:03:47:18:42:4c
        media: Ethernet 10baseT
        status: active
        inet 132.185.130.238 netmask 0xfffffc00 broadcast 132.185.131.255
        inet6 fe80::203:47ff:fe18:424c%xi0 prefixlen 64 scopeid 0x5

PING 132.185.128.94 (132.185.128.94): 56 data bytes
ping: sendto: No buffer space available
ping: sendto: No buffer space available
[..]

Basically, anyone else seen similar problems, and/or give me information
as to where to look in the driver?  I don't mind having a poke around
and providing a patch, but I suck at knowing where to start to debug :)

Jon