[Soekris] BIOS setting stored in NatSemi EEPROM?

[Rob Flickenger]

On Aug 17, 2004, at 10:17 AM, Dev Kumar wrote:

> It will try to PXE boot every time, even when there is a valid CF in 
> the slot, and it will eventually timeout and boot the CF if there is 
> no server present.
> If you change the value in this 17th word back to all 0xFFFF, then it 
> boots as expected: from the CF if present, from PXE if not.

Interesting.  I have a bunch of net4526s running BIOS v1.24, and 
apparently the values are different:

 > dw 16
0000:0016  F000 0457 F000 0457 F000 2EB9 F000 0457  .*W..*W..*+..*W.
0000:0026  F000 0457 F000 0457 F000 0457 F000 0457  .*W..*W..*W..*W.

It looks like 0000:0016 is F000, not FFFF.  Perhaps there are 
differences between the 4526 and the 4511?  Or am I misinterpreting 
this command?  Boy, I'd love a thorough reference to the BIOS on these 
boards.

> I think you can make it PXE boot by default normally just with a BIOS 
> setting.

I thought so too, but apparently not.  Setting the FLASH parameter to 
Secondary just makes it attempt PXE and then die if a server is not 
present (not fail over to CF, which would be really handy in many 
circumstances...)  Typing "boot f0" on the console is not an option 
when the box is on a tower in a waterproof case, but cycling the power 
and bringing up a PXE server remotely would solve it nicely.

> The bottom line is that we are trying to store some simple ID info on 
> the unused parts of the EEPROM, at least according to the NatSemi 
> datasheet, and then the board behavior changes. Any ideas on what 
> parts of the EEPROM are truly reserved and used, and which are really 
> free for use, is appreciated.

Wish I could help.  In the process of solving your problem, you've 
found a way to make the board do something that would solve a real 
problem for us...  If only I could replicate it!

Regards,

--Rob