Re: MILO vs BIOS on Linux/Intel.

Werner Almesberger (almesber@lrc.di.epfl.ch)
Thu, 12 Feb 1998 18:22:34 +0100 (MET)


David Woodhouse wrote:
> SCSI controllers generally provide their own BIOS. The BIOS would have to
> provide direct drivers for on-board IDE, but not SCSI - it could just use INT
> 13h for that.

... and get back many of the restrictions which make people unhappy with
their current BIOS (e.g. that 1024 cylinder limit or some variation of
it).

> Recently, we've seen examples of features that have been unusable or at least
> difficult to use in Linux because the proprietary BIOS doesn't enable them or
> support them correctly. (UDMA, IO-APIC, ThinkPad APM...)

I'm not familiar with the details of those cases, but I'd assume you could
just replace whatever the BIOS currently does if you really want to.

> In addition to this improvement, we could offer extra functionality, like
> password-protection upon booting,

There are several ways to accomplish this already, e.g.
- some BIOS' already support password protection
- restrict boot options in BIOS and require password in boot loader
- if all else fails, move untrusted hardware somewhere where the BIOS
doesn't find it and require password in boot loader

> an interface to a Linux device which allows you to edit BIOS settings
> (imagine an Xbios config program :).

You can already access the configuration RAM and, with some effort, you
could probably also find where the BIOS stores that data and in what
format.

> We could incorporate serial console support,

Besides accessing the BIOS setup screens, this already works.

> MILO-style boot loading,

What exactly do you mean ?

> With hardware watchdogs, we could implement complete core dumps to the
> network on the way back up...

... or to a disk. Yes, this, and a network boot without hardware changes
would probably be the most interesting features.

> Calls to re-implement ROM BASIC might be shot down in flames though.

Just switch to 40x25 text mode and display "*** NO ROM BASIC ***" ;-)

> There are already people offering BIOS upgrades that haven't been tailored to
> particular boards. Take a look at http://www.mrbios.com/

They have a very long list of supported boards along with the order
numbers for the corresponding BIOS, so I guess what they put on the
chips isn't exactly generic ...

> The basic functionality required from a BIOS implementation is clearly
> documented in a number of places, and should be fairly easy to conform to.

Hmm, I wonder what the DOSEMU folks think about that statement.

> Providing a GPL'd BIOS implementation is a feasible project. It could be
> extremely useful.

I'm not so sure about the "extremely". It seems that most added
functionality such a BIOS could provide is already available in a
perhaps less elegant but acceptable way.

But if you're determined and have a few spare years, you can certainly
try :-)

- Werner

-- 
  _________________________________________________________________________
 / Werner Almesberger, DI-ICA,EPFL,CH   werner.almesberger@lrc.di.epfl.ch /
/_IN_R_131__Tel_+41_21_693_6621__Fax_+41_21_693_6610_____________________/

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu