Subject: tlp regression on 1.6 branch
To: None <tech-kern@netbsd.org>
From: Emmanuel Dreyfus <manu@netbsd.org>
List: tech-kern
Date: 10/15/2003 12:27:44
Hi

I was recently burnt by a regression on the 1.6 branch. Some ethernet
board handled by the tlp driver stopped working.

Here is the dmesg output for the board on a functionnal system:
(as of 6th november 2002)
tlp0 at pci0 dev 7 function 0: Lite-On 82C169 Ethernet, pass 2.0
tlp0: interrupting at irq 11
tlp0: Ethernet address 00:02:e3:07:fb:3c
bmtphy0 at tlp0 phy 1: BCM5201 10/100 media interface, rev. 2
bmtphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto

On a recent system, as of 22th septeber 2003, dmesg does not report the phy:
tlp0 at pci0 dev 7 function 0: Lite-On 82C169 Ethernet, pass 2.0
tlp0: interrupting at irq 11
tlp0: Ethernet address 00:02:e3:07:fb:3

ifconfig tlp0 up will cause recurent "tlp: filter setup and transmit timeout"
messages, one per second. But I suspect it is related to the missing phy.

What is strange is that neither tlphy.c, nor if_tlp_pci.c nor mbtphy.c were
modified in the meantime.

Anyone has an idea of what went wrong? 

-- 
Emmanuel Dreyfus
manu@netbsd.org