[LinuxBIOS] ADLO on a real machine

[Myles Watson]

With some fiddling I booted into the Windows install CD and installed
Windows XP on a hard drive.  It took fiddling because the BIOS is so old
that it doesn't conform to the ATA-3 specifications.

I haven't been able to boot using a hard drive from ADLO yet because it
doesn't recognize my hard drives (same age related problems of the Bochs
BIOS.)

I've looked at porting the FILO drivers into the Bochs BIOS, but it doesn't
appear trivial.

Any questions, suggestions, or comments besides "Why would you want to do
this?"

Thanks,
Myles

[Stefan Reinauer]

Myles Watson wrote:
> With some fiddling I booted into the Windows install CD and installed
> Windows XP on a hard drive.  It took fiddling because the BIOS is so old
> that it doesn't conform to the ATA-3 specifications.
>
> I haven't been able to boot using a hard drive from ADLO yet because it
> doesn't recognize my hard drives (same age related problems of the Bochs
> BIOS.)
>
> I've looked at porting the FILO drivers into the Bochs BIOS, but it doesn't
> appear trivial.
>
> Any questions, suggestions, or comments besides "Why would you want to do
> this?"

Hmm. First off, this is pretty cool. Which version of Windows? :-)

Does your hard disk controller have an option rom? P-ATA? S-ATA? SCSI?

It might make sense to have the controller's option rom executed by ADLO
so it can install an int13 hook.

Have you been able to use FILO?

Stefan

[Augusto Pedroza]

This is great,
  I'd like to help with ADLO, could give me some more details when you try
to boot using ADLO?

Thanks,
  Augusto Pedroza

On 9/14/07, Stefan Reinauer  wrote:
> Myles Watson wrote:
> > With some fiddling I booted into the Windows install CD and installed
> > Windows XP on a hard drive.  It took fiddling because the BIOS is so old
> > that it doesn't conform to the ATA-3 specifications.
> >
> > I haven't been able to boot using a hard drive from ADLO yet because it
> > doesn't recognize my hard drives (same age related problems of the Bochs
> > BIOS.)
> >
> > I've looked at porting the FILO drivers into the Bochs BIOS, but it
>
> doesn't
> > appear trivial.
> >
> > Any questions, suggestions, or comments besides "Why would you want to
>
> do
> > this?"
>
> Hmm. First off, this is pretty cool. Which version of Windows? :-)
>
> Does your hard disk controller have an option rom? P-ATA? S-ATA? SCSI?
>
> It might make sense to have the controller's option rom executed by ADLO
> so it can install an int13 hook.
>
> Have you been able to use FILO?
>
> Stefan

[Myles Watson]

> I'd like to help with ADLO, could give me some more details when you try

to boot using ADLO?

It's pretty unspectacular.  The error message is:

ata0 master: Unknown device

It won't even try to boot.

Myles

[Myles Watson]

> > I'd like to help with ADLO, could give me some more details when you
>
> try
> to boot using ADLO?
>
> It's pretty unspectacular.  The error message is:
>
> ata0 master: Unknown device
>
> It won't even try to boot.

There are two ways it seemed like I could do this
1. Add support for new drives to Bochs' BIOS
2. Backport Filo's driver

I started #2.  It starts and fails quickly.  I'll work on it more on Monday,
but if anyone is interested in looking it over in the mean time, that would
be great.

To build it, replace your rombios.c with my rombios.c, filo_ide.h, and
filo_ide.c files, then cross your fingers :)

Right now it still has the ATA support from Bochs, it just calls the FILO
probe routines and then tries to continue.

Once it detects the disks correctly, I'll have to make the BIOS callbacks
function correctly with the new disk data.

Myles

[Peter Stuge]

On Fri, Sep 14, 2007 at 04:53:17PM -0600, Myles Watson wrote:
> Backport Filo's driver
>
> I started #2.

Please don't spend time on the FILO IDE driver. It's a mess and not
worth the effort.

Instead I'd suggest the OpenBIOS IDE driver, which is much nicer.
> To build it, replace your rombios.c with my rombios.c, filo_ide.h,
> and filo_ide.c files, then cross your fingers :)

I was curious to have a look. Please send patches instead of
tarballs.
> Right now it still has the ATA support from Bochs, it just calls
> the FILO probe routines and then tries to continue.

How about LBA32/LBA48 support in the BIOS calls?

//Peter

[Myles Watson]

> > Backport Filo's driver
> >
> > I started #2.
>
> Please don't spend time on the FILO IDE driver. It's a mess and not
> worth the effort.
>
> Instead I'd suggest the OpenBIOS IDE driver, which is much nicer.

I'm confused why FILO hasn't started using the OpenBIOS IDE driver if FILO's
driver is a mess.
> > To build it, replace your rombios.c with my rombios.c, filo_ide.h,
> > and filo_ide.c files, then cross your fingers :)
>
> I was curious to have a look. Please send patches instead of
> tarballs.

I thought patches would be too hard to look through, since there are changes
on nearly every line to support segments.  All of the data structures had to
be moved into the EBDA (Extended BIOS Data Area) segment, and that meant
passing different parameters to every function.  Generating the patches is
easy, though, so I'll send you one on Monday.
> > Right now it still has the ATA support from Bochs, it just calls
> > the FILO probe routines and then tries to continue.
>
> How about LBA32/LBA48 support in the BIOS calls?

Yep, that would be the next thing to do.

Is there another free BIOS somewhere that already has that support?  Maybe I
shouldn't be starting with ADLO at all.

Myles

[Myles Watson]

Peter,

Here are the patches I promised.  I recombined filo_ide.h and filo_ide.c so
that the diff would make more sense with the original file (ide_x.c).

I'm still open for suggestions if there is something better than ADLO that I
should be using for the BIOS call backs after initializing the machine with
LinuxBIOS.

Myles