[plug] Firewire causes system to freeze - solution - not Firewire, but eepro100

[Onno Benschop]

On Wed, 2003-01-15 at 16:54, Onno Benschop wrote:
> I'm running a Debian kernel 2.4.20-686 on an Intel 815 series
> motherboard. I have an external firewire case with an 80Gb drive. The
> case uses an Oxford Semiconducter Ltd / Wise Advanced Co., Ltd. chipset
> and drives an 80Gb MAXTOR - L080J4.
> 
> When I transfer data to or from the drive, for no apparent reason at
> all, my workstation will hang. There are no kernel messages, no log
> entries, it just stops.

While I don't generally try to make a habit out of talking to myself,
for archival purposes I'm adding what the problem really was and what
the solution was.

The machine was connected via its on-board eepro100 to a 10Mb hub, which
in turn was connected to a Macintosh PCMCIA ethernet adapter. The
problem - given that I continued to test and loose hair - became
apparent when I ran in single user mode in the console and actually got
a hint that the eepro100 wasn't a happy camper. (wait_for_cmd_done
timeout!)

After searching for a solution, on a hunch I decided to remove the hub
and replace it with a 10/100 switch, and all my problems just vanished.
Overnight, the machine has transferred 16Gb to the Macintosh without
missing a beat.

Some postings on the net refer to strange lock-up problems, but AFAIK,
it seems to be related to high-traffic/high collision environments -
which makes sense, since intermittently I've noticed that on a good
morning /. gives me many interesting links to click on and after about
six sites are opening, the machine also dies.

So, while the finger was pointing at fire-wire - since thats where all
my interesting data resides, the mayhem was happening on eth0.

I hope this helps the next unwary traveler in the future.
 
-- 
()/)/)()        ..ASCII for Onno.. 
|>>?            ..EBCDIC for Onno.. 
--- -. -. ---   ..Morse for Onno.. 

ITmaze   -   ABN: 56 178 057 063   -  ph: 04 1219 8888   -  
onno at itmaze.com.au