Hi Mike,
On 18.09.2009 18:55, Spangler, Mike T wrote:
I've corrected my email and am now sending to the group.
Thanks.
Of course I'd prefer it if you tried to flash coreboot (new name of LinuxBIOS) to these boards, but that's your choice. Anyway, you said you have the factory BIOS around somewhere. Is it the one at http://flextronics.com/arima/Server/Download/Download/HDAMA189.zip ?
I can't use flash coreboot (LinuxBIOS) because our cluster management tools require pxeboot. We've transitioned away from it.
That's unfortunate. Did you know that coreboot+SeaBIOS works fine with gPXE? We coreboot/flashrom developers are always interested in helping people update coreboot to the latest and greatest. So if you need PXE booting, there's no need to move back to the factory BIOS, a simple coreboot update should suffice.
So, the bios zip file above is correct, but I'm not using it. It will not work for flashing a chip with LinuxBios. The way I do this is to do a "flashrom -r" on a system with factory bios and then take that newly created flashrom file and do a "flashrom -w" with it on a board with LinuxBios.
That's one option. The other option would be to extract the BIOS from the .ROM file. That looks reasonably easy. No guarantees, though. dd if=something.ROM of=bios.bin bs=1k count=512
This is the method we used with the old bios_copy software from Linux Networx, but that was extremely unreliable.
flashrom strives to be the most reliable flashing tool out there. The current version has builtin verification for every step it does.
Hmm ok. I think we have enough docs to iterate through all possible board enables. How many of these boards do you have (in case we trash the ROM of one of them)?
The ROM is removeable, so we only potentially trash a PLCC chip. I have a half dozen or so spares.
Great. And with hotplugging/hotflashing it should be possible to reflash a trashed chip in another board.
Can you supply the output of "lspci -nnvvvxxx" as root? That will allow me to dig further. I'd like to limit the search to a reasonable subset of options.
Regards, Carl-Daniel