[3180] in Release_7.7_team
Etherboot floppy trashes GX150 EEPROM!
daemon@ATHENA.MIT.EDU (Bill Cattey)
Fri Mar 22 20:41:34 2002
Message-ID: <Awaxp=xz00011VUZos@mit.edu>
Date: Sat, 23 Mar 2002 01:41:31 +0000 ()
From: Bill Cattey <wdc@MIT.EDU>
To: owls@MIT.EDU, release-team@MIT.EDU
CC: kcr@MIT.EDU, tb@MIT.EDU, linux-dev@MIT.EDU
Today, Danilo and I discovered that the Athena Etherboot install floppy,
even if you just run it as far as the first IP address prompt and stop,
renders a GX150 incapable of doing a PXE boot.
What this means is that WE, not Dell, have been setting a toxic set of
values in the EEPROMs.
I read the code and discovered it carefully sets
0x0160 in location 0x000d in the EEPROM
to cover the case of one buggy chipset.
This value all by itself is enough to cause problems.
We need to study this some more to be fully sure of what we're up
against, and to understand just what is being set, and whether this
represents an easy or difficult case. I am pursuing further discussions
with the Ethernet intelligencia.
Current Implications:
If you use the floppy to install a dual boot Windows Athena system,
it will trash the PXE boot ability.
The EEPROM fix we sent out to the clusters will quit working as soon
as a system is reinstalled.
Possible Remedies:
Fix the Etherboot driver, and give everyone new floppies.
See if the slightly different timing with the new experimental,
diverse hardware floppy is acceptable.
Thing to learn from this:
With things like Ethernet device drivers, stick to drivers used by LOTS
of people instead of hairy experimental things, unless you are prepared
to audit code!
-wdc
