Re: [OpenBIOS] OpenBIOS: reality check

[David Woodhouse]

[e-mail] said:
> On Sun, 1 Mar 1998, Benjamin Scott wrote:
> > > > > Do  Suns,  Apollos,  HPs,  IBMs,  NeXTs,  etc.  have  rudamentary
> > > > > filesystem code in their BIOS?
> > > >
> > > > I know for an absolute fact that is has been done on the Alpha.
> > >
> > > Which filesystem does it use?
> >
> > Anything Linux does.  The MILO loader, which is designed to be blown
> > into firmware, actually incorporates regular Linux device drivers.  So
> > take your pick.  :-)
>
> Not exactly true.  The SRM firmware found on most Alpha systems knows
> nothing about filesystems or disk partitions.  Instead, SRM relies on the
> ability to load a secondary bootstap loader, the location of which is
> described by values stored on the first 512 byte sector of the physical
> disk.

Not so. At least on the SX164 we have, it knows about the filesystem. Anyway,
the phrase "the location of which is described by values stored on the first
512 byte sector" sounds suspiciously like understanding disk partitions to me.

The SRM firmware understands at least FAT, and possibly also NTFS partitions.
Certainly it's capable of loading second-stage bootloaders like "LINLOAD.EXE"
from a FAT floppy disk. I believe that even the emergency firmware update is
managed from a FAT floppy.

When you configure SRM to load Linux using LINLOAD, you have to give a dummy 
argument for the "OPERATING SYSTEM DIRECTORY", and the SRM console complains 
if that directory is not there. It definitely understands the filesystem.

Looking at the ext2 drivers for Win95, I see that the VXD files gzip down to 
15596 bytes. I'm sure we can find that amount of space in the flash for those 
who choose to configure ext2 support into their BIOS. I for one, would like to 
do so.

Making new filesystems just so that the BIOS can access them easily is not 
practicable. Neither is the idea of dedicating a partition to BIOS extensions.
Simple read-only filesystem support can be kept fairly small, and is going to 
be required by UNIX people to load either the second-stage bootloader or the 
kernel itself.

----                              ----                              ----
David Woodhouse, Robinson College, CB3 9AN, England.   ([phone]
    [e-mail]        http://dwmw2.robinson.cam.ac.uk
        finger [e-mail] for PGP key.

---
OpenBIOS -- http://www.linkscape.net/openbios/
[e-mail]   Body: un/subscribe
Problems?  [e-mail]

OpenBIOS: reality check

[Aaron M. Gowatch]

On Sun, 1 Mar 1998, David Woodhouse wrote:
> Not so. At least on the SX164 we have, it knows about the filesystem. Anyway,
> the phrase "the location of which is described by values stored on the first
> 512 byte sector" sounds suspiciously like understanding disk partitions to me.

Ahh... the Windows NT AlphaBIOS that runs on the PC164.  Its a lot like
the Windows NT ARC firmware, but it is not SRM.

What does reading raw sectors off of a disk have to do with partitions? 
Before any disk has partitions or even filesystems on it, it has sectors. 
Do you think the MBR is a partition?  You store these values in the first
512 byte sector of raw disk and you read them.
> The SRM firmware understands at least FAT, and possibly also NTFS partitions.
> Certainly it's capable of loading second-stage bootloaders like "LINLOAD.EXE"
> from a FAT floppy disk. I believe that even the emergency firmware update is
> managed from a FAT floppy.

Again, the Windows NT AlphaBIOS, tailored specifically for running Windows
NT, has support for these filesystems.  It is not SRM.  I dont see the
point in developing such OS specific BIOS extensions.
> Making new filesystems just so that the BIOS can access them easily is not 
> practicable. Neither is the idea of dedicating a partition to BIOS extensions.
> Simple read-only filesystem support can be kept fairly small, and is going to 
> be required by UNIX people to load either the second-stage bootloader or the 
> kernel itself.

Agreed, creating new filesystems is probably not practical.  But saying
that a filesystem is required in order to boot UNIX is entirely false. 
The concept of reading sectors directly from disk independent of
partitions or filesystem is not a new one.  I'd suggest that you read the
SRM Firmware HOWTO:

http://www.azstarnet.com/~axplinux/srm.html

Aa.

---
OpenBIOS -- http://www.linkscape.net/openbios/
[e-mail]   Body: un/subscribe
Problems?  [e-mail]