Promise PDC20268 FastTrack 100 TX2 (PDC20268)

Promise PDC20268 FastTrack 100 TX2 (PDC20268)

[lkml]

Hello all,

I have a Promise FastTrack 100 TX2 (PDC20268) IDE-controller
(BIOS v2.00.0.24) used in a linux MD-RAID.  Aside from various
other annoying Promise-problems, I am not able to perform a
remote boot because the brain-dead Promise-BIOS "complains" that
no array is defined, and requires one to press ESC to continue
booting.  I would very much appreciate any tips as to how I can
circumvent this "feature".

Best regards,

Robert

PS. Please CC me, as I am not in the list.

Re: Promise PDC20268 FastTrack 100 TX2 (PDC20268)

[John Bradford]

> I have a Promise FastTrack 100 TX2 (PDC20268) IDE-controller
> (BIOS v2.00.0.24) used in a linux MD-RAID.  Aside from various
> other annoying Promise-problems, I am not able to perform a
> remote boot because the brain-dead Promise-BIOS "complains" that
> no array is defined, and requires one to press ESC to continue
> booting.  I would very much appreciate any tips as to how I can
> circumvent this "feature".

Well, if you don't usually need a keyboard on that machine, in theory,
you might be able to connect the keyboard input to the PS/2 mouse port
of another machine, and write a program to send the correct bytes for
that keypress to the other machine.  Then, you could reboot the
machine with the Promise card in it, then log in to the other machine,
and run the program to send the keypress.

Not sure how practical this solution would be though...  You'd
probably have to simulate the keyboard initialisation responses as
well, which would make it a bit complicated

John.

Re: Promise PDC20268 FastTrack 100 TX2 (PDC20268)

[Vojtech Pavlik]

On Mon, Jan 20, 2003 at 07:12:53PM +0000, John Bradford wrote:
> > I have a Promise FastTrack 100 TX2 (PDC20268) IDE-controller
> > (BIOS v2.00.0.24) used in a linux MD-RAID.  Aside from various
> > other annoying Promise-problems, I am not able to perform a
> > remote boot because the brain-dead Promise-BIOS "complains" that
> > no array is defined, and requires one to press ESC to continue
> > booting.  I would very much appreciate any tips as to how I can
> > circumvent this "feature".
> 
> Well, if you don't usually need a keyboard on that machine, in theory,
> you might be able to connect the keyboard input to the PS/2 mouse port
> of another machine, and write a program to send the correct bytes for
> that keypress to the other machine. Then, you could reboot the
> machine with the Promise card in it, then log in to the other machine,
> and run the program to send the keypress.
> 
> Not sure how practical this solution would be though...  You'd
> probably have to simulate the keyboard initialisation responses as
> well, which would make it a bit complicated

That won't work, of course - while the PS/2 and keyboard ports are the
same, they are master/slave - the computer is master and the device is
slave - you cannot connect master to master.

It'd work if you used bitbanging on the parallel port, though.

Or a keyboard with something heavy on the 'esc' key ;)

Another possibility is just to remove the BIOS from the card or tell the
on-board BIOS not to initalize it.

-- 
Vojtech Pavlik
SuSE Labs

Re: Promise PDC20268 FastTrack 100 TX2 (PDC20268)

[Milan Roubal]

make a single drive strip array - promise will see array defined and the
reboot will pass correctly.
Then you can use sw raid.
    Milan Roubal
    roubm9am@barbora.ms.mff.cuni.cz

----- Original Message -----
From: <lkml@scienceworks.com>
To: <linux-kernel@vger.kernel.org>
Sent: Monday, January 20, 2003 7:34 PM
Subject: Promise PDC20268 FastTrack 100 TX2 (PDC20268)


> Hello all,
>
> I have a Promise FastTrack 100 TX2 (PDC20268) IDE-controller
> (BIOS v2.00.0.24) used in a linux MD-RAID.  Aside from various
> other annoying Promise-problems, I am not able to perform a
> remote boot because the brain-dead Promise-BIOS "complains" that
> no array is defined, and requires one to press ESC to continue
> booting.  I would very much appreciate any tips as to how I can
> circumvent this "feature".
>
> Best regards,
>
> Robert
>
> PS. Please CC me, as I am not in the list.