linux-mips-fnet
[Top] [All Lists]

Re: Olivetti tftp booting

To: linux-mips@fnet.fr
Subject: Re: Olivetti tftp booting
From: Systemkennung Linux <linux@mailhost.uni-koblenz.de>
Date: Tue, 16 Jan 1996 17:04:21 +0100 (MET)
In-reply-to: <199601161339.OAA22842@bandsept.univ-evry.fr> from "Stoned Elipot" at Jan 16, 96 02:38:55 pm
Hi all,

> last night I tried to boot my Oily by tftp,... no luck I can't found
> the hardware path to tell the ARC bios to use the network interface,
> even after browsing ARC.5 man page of Milo and seeking any kind of
> strings in the BIOS. BTW, on oily you have "superuser" mode for the
> 'system services' BIOS shell, just type 'superuser on', and then you
> have function like create a TLB entry for accessing any physical RAM
> (by 16M page) do dump and fill on memory adresses do read write on
> port adresses, and may be other stuff I can recall just now... 
> I didn't find any reference to this in any docs I got: is this kind of
> stuff in ARC BIOS standard ?

No; the ARC standard is any more than incomplete anyway.  Just one
example: It provides file services similar to open(2), close(2), read(2),
write(2), lseek(2) (extremly buggy) but no way to remove files.
(C) Microsoft ...

The network booting facilities described in the ARC standard are not
implemented in any ARC BIOS that I've yet used.  So network booting in
this case means booting the kernel from harddisk/floppy/CDROM.  Then
the kernel then continues booting with RARP/NFS.

                               Oh yes! there is provision for hooking an
> 'external' debugger: unfortunately I don't know what they call an
> external debugger (firmware upload like the bi-endian one or 'serial
> debugger'), any idea ???

Not in the ARC standard.  Everything that the BIOS provides is any useless
after the kernel has been started.  With the machine getting reinitialized
and all BIOS variables overwritten BIOS stops working.  On the other side
keeping the BIOS alive is not an option since it uses a large part of the
memory for ... doing nothing with it ...

Someone from SNI once wrote that they had written their own BIOS
implementation from scratch on and so I hope that the RM200 the SNI promised
to send will have the first machine with an complete and bugfree
implementation of an ARC BIOS.

                           (I was trying to thrown in 1.3.48 kernel the
> gdb support Andy wrote to track down a keyboard interupt problems that
> is "pretty" persistent if you see what I mean Ralf:)

Oh ...  Well, for me 1.3.48 was pretty reliable.  Yesterday I felt like
testing 1.3.57 and so I rebooted the machine after an uptime of five days
and one hour.

   Ralf

<Prev in Thread] Current Thread [Next in Thread>