> > I would think that if you are copying all of the BIOS settings, then the
> > checksum should be right.
> 
> Why? As I see it, the checksum is presumably created and stored
> somewhere when you select save in the BIOS setting program. Since this
> has been bypassed by using /dev/nvram the stored checksum will not
> match the new BIOS settings. It might have worked if the checksum
> value was stored in the BIOS settings and the checksum test was only
> for the rest of the settings but this would appear not to be the case.

It sounds like the question is: if you copy from another known good BIOS,
perhaps the (known good) checksum is being copied as well (in-band as opposed
to out-of-band), and therefore the second BIOS should come up fine.  However,
perhaps the assumption that the checksum is in-band is not correct?

Maybe there's a second NVRAM just for the checksum?

Maybe the checksum is not accessible to the kernel in order to be exposed via
/dev/nvram?

Maybe the checksum includes a parameter from the hardware itself, e.g. MAC
address or motherboard serial number or what have you, which precludes
transfering a known good BIOS image from one machine to another without going
through the manufacturer's checksum routine each time?

*sigh*

Good luck,
Dan W.