[tulip-bug] Recurring Lite-on PNIC loss of traffic after unplug/replug from switch

[John Connett]

Here are some edited highlights using debug=2 (with some duplicated
messages removed):

tulip.c:v0.92 4/17/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth1: Lite-On 82c168 PNIC rev 32 at 0xf0838f00, 00:C0:F0:3E:BA:78, IRQ
9.
eth1:  MII transceiver #1 config 3100 status 7829 advertising 01e1.
eth1: tulip_open() irq 9.
eth1: MII link partner 0080, negotiated 0080.
# successfully connected at 100baseTx-HD
# disconnected
eth1: Transmit error, Tx status 7fff8c04.
eth1: MII link partner 0020, negotiated 0020.
# successfully reconnected at 10baseT-HD
eth1: MII link partner 0020, negotiated 0020.
# disconnected
eth1: MII link partner 0000, negotiated 0000.
eth1: No link beat on the MII interface, status 7809.
eth1: MII link partner 0080, negotiated 0080.
# successfully reconnected at 100baseTx-HD
eth1: MII link partner 0080, negotiated 0080.
# disconnected
eth1: MII link partner 41e1, negotiated 01e1.
eth1: Setting full-duplex based on MII #1 link partner capability of
41e1.
eth1: PNIC link changed state 0001f978, CSR5 02678016.
# successfully reconnected at 100baseTx-FD
# disconnected
eth1: PNIC link changed state 0001f978, CSR5 02669000.
eth1: MII link partner 0000, negotiated 0000.
eth1: No link beat on the MII interface, status 7809.
eth1: MII link partner 0080, negotiated 0080.
eth1: Setting half-duplex based on MII #1 link partner capability of
0080.
eth1: MII link partner 0080, negotiated 0080.
eth1: Transmit timeout using MII device.
# FAILED TO RECONNECT AT 100baseTx-HD <<<<<<<<<<
eth1: Transmit timeout using MII device.
eth1: Transmit timeout using MII device.
eth1: MII link partner 0080, negotiated 0080.
eth1: Transmit timeout using MII device.
eth1: Transmit timeout using MII device.
# ifdown eth1
eth1: Shutting down ethercard, status was 2261000.
# ifup eth1
eth1: tulip_open() irq 9.
eth1: MII link partner 0080, negotiated 0080.
# successfully reconnected at 100baseTx-FD

In this case the problem occurs when changing from 100baseTx-FD to
100baseTx-HD.  Shutting down and restarting eth1 clears the problem.

I'm willing to spent some more time investigating this.  Unfortunately,
the only documentation on the KNE110TX I have is a preliminary data
sheet for the Myson Technology MTD972 100BaseTX PCS/PMA.  Can someone
with greater knowledge in this area venture an opinion as to whether
this is fixable in the driver on not?  

Thanks in anticipation
--
John Connett (jrc@art-render.com)