VIA KT133 data corruption update

VIA KT133 data corruption update

[Bryan O'Sullivan]

[-- Attachment #1: Type: text/plain, Size: 1720 bytes --]

After several months of begrudgingly putting up with my ASUS A7V
motherboard corrupting roughly 1 byte per 100 million read during
moderate to heavy PCI bus activity, I flashed VIA's 1009 BIOS this
evening.

I have not been able to reproduce any corruption since then (it was
ridiculously easy before the new BIOS), and my machine seems otherwise
as stable as I would hope.  This marks the first time since 2.4.6 that
I've been able to run a Linus kernel without cowering.

I also discovered, of necessity, a halfway manageable process for
creating a DOS boot floppy using Windows ME, which Microsoft would
apparently prefer was not possible.  I'll reproduce the steps here,
since otherwise flashing a new BIOS is likely to be nightmarish for
people stuck dual booting into WinME.

Most of these steps occur under Linux, and I'll assume that your Windows
Me "C:" drive is mounted as /dos/c.

- Format a floppy:
  fdformat /dev/fd0H1440

- Create a FAT filesystem:
  mkdosfs /dev/fd0

- Mount the floppy:
  mount /dev/fd0 /mnt

- Copy across a few files:
  cp /dos/c/command.com /mnt
  cp /dos/c/io.sys /mnt
  cp /dos/c/msdos.sys /mnt

- Edit /mnt/msdos.sys, and change values as follows:
  [Paths]
  WinDir=a:\
  WinBootDir=a:\
  HostWinBootDrv=a

  [Options]
  BootMulti=0
  BootGUI=0
  AutoScan=0

- Copy across your BIOS flash utility (probably aflash.exe) and BIOS
  image.  Unmount the floppy (important; don't just reboot):
  umount /mnt/floppy

- When you reboot to the floppy, it will desperately try to boot into
  Windows.  When it prompts you for the path to some Windows VXD, just
  type "a:\command.com", and lo, you've got a DOS prompt.

Cheers,

	<b

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: VIA KT133 data corruption update

[Hans-Peter Jansen]

On Saturday, 27. October 2001 12:48, Bryan O'Sullivan wrote:
> After several months of begrudgingly putting up with my ASUS A7V
> motherboard corrupting roughly 1 byte per 100 million read during
> moderate to heavy PCI bus activity, I flashed VIA's 1009 BIOS this
> evening.
>
> I have not been able to reproduce any corruption since then (it was
> ridiculously easy before the new BIOS), and my machine seems otherwise
> as stable as I would hope.  This marks the first time since 2.4.6 that
> I've been able to run a Linus kernel without cowering.

Good news. Congrats.

> I also discovered, of necessity, a halfway manageable process for
> creating a DOS boot floppy using Windows ME, which Microsoft would
> apparently prefer was not possible.  I'll reproduce the steps here,
> since otherwise flashing a new BIOS is likely to be nightmarish for
> people stuck dual booting into WinME.
>
> Most of these steps occur under Linux, and I'll assume that your Windows
> Me "C:" drive is mounted as /dos/c.
>
> - Format a floppy:
>   fdformat /dev/fd0H1440
>
> - Create a FAT filesystem:
>   mkdosfs /dev/fd0
>
> - Mount the floppy:
>   mount /dev/fd0 /mnt
>
> - Copy across a few files:
>   cp /dos/c/command.com /mnt
>   cp /dos/c/io.sys /mnt
>   cp /dos/c/msdos.sys /mnt
>
> - Edit /mnt/msdos.sys, and change values as follows:
>   [Paths]
>   WinDir=a:\
>   WinBootDir=a:\
>   HostWinBootDrv=a
>
>   [Options]
>   BootMulti=0
>   BootGUI=0
>   AutoScan=0
>
> - Copy across your BIOS flash utility (probably aflash.exe) and BIOS
>   image.  Unmount the floppy (important; don't just reboot):
>   umount /mnt/floppy
>
> - When you reboot to the floppy, it will desperately try to boot into
>   Windows.  When it prompts you for the path to some Windows VXD, just
>   type "a:\command.com", and lo, you've got a DOS prompt.

Since your floppy doesn't contain a valid bootsector, it won't start
from floppy here. At least my mkdosfs created boot sector says:
This is not a bootable disk.  Please insert a bootable floppy and
press any key to try again ...

The simplest solution is choosing win(9*|me) from boot manager then, press
F8 immediately, select boot to command prompt. (Hope this works in me, too) 
Issue
c:\> sys a:
and
c:\> edit msdos.sys
then, as above. If booting from floppy still fails, check BIOS boot order.

Note: if you don't trust your winloose inst., check the created bootsector:
1) Label at offset 3 should read MSWIN4.1 or the like
2) some readable error messages at around 0x180 
3) IO      SYSMSDOS  SYS at ~0x1d8

BTW[FAR OT]: Somebody remember IHC here?

> Cheers,
>
> 	<b

Have-a-nice-stay-where-ever-you-are-ly yours,
FrisPete

Re: VIA KT133 data corruption update

[Jan Niehusmann]

On Sat, Oct 27, 2001 at 03:48:56AM -0700, Bryan O'Sullivan wrote:
> After several months of begrudgingly putting up with my ASUS A7V
> motherboard corrupting roughly 1 byte per 100 million read during
> moderate to heavy PCI bus activity, I flashed VIA's 1009 BIOS this
> evening.

Please note that there have been broken versions of the 1009 BIOS
around. I know one person, and I read from serveral ones, who flashed
1009 to their A7V and were unable to start the computer afterwards.
(Hangs before/during POST).

Apparently ASUS has replaced the broken version with a working one
without updating the version number. But that's just a guess based on
recent success stories about 1009.

> I also discovered, of necessity, a halfway manageable process for
> creating a DOS boot floppy using Windows ME, which Microsoft would
> apparently prefer was not possible.  I'll reproduce the steps here,
> since otherwise flashing a new BIOS is likely to be nightmarish for
> people stuck dual booting into WinME.

As I don't use Windows at all, FreeDOS has proven very useful for 
flashing the bios. (www.freedos.org)
But, of course, no guarantees.

Jan

Re: VIA KT133 data corruption update

[Tobias Ringstrom]

On 27 Oct 2001, Bryan O'Sullivan wrote:

> I also discovered, of necessity, a halfway manageable process for
> creating a DOS boot floppy using Windows ME, which Microsoft would
> apparently prefer was not possible.  I'll reproduce the steps here,
> since otherwise flashing a new BIOS is likely to be nightmarish for
> people stuck dual booting into WinME.

Slightly off topic, but you can also use the rescue disk (made using
add/remove programs in the control panel) to get a DOS prompt, and 
use that to flash your new BIOS.

The Control or F8 trick to go to a DOS prompt does not work with ME.

Using FreeDOS is an even nicer approach, of course.

/Tobias

Re: VIA KT133 data corruption update

[Bryan O'Sullivan]

[-- Attachment #1: Type: text/plain, Size: 591 bytes --]

On Sat, 2001-10-27 at 06:17, Jan Niehusmann wrote:

> Please note that there have been broken versions of the 1009 BIOS
> around. I know one person, and I read from serveral ones, who flashed
> 1009 to their A7V and were unable to start the computer afterwards.
> (Hangs before/during POST).

Ouch.  As it happens, my system now beeps repeatedly on POST, but
hitting the reset button convinces it to behave.

> As I don't use Windows at all, FreeDOS has proven very useful for 
> flashing the bios. (www.freedos.org)

I should have thought of that, had it not been 3am.

	<b

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: VIA KT133 data corruption update

[David Grant]

Bryan,

I have an A7V-133 motherboard, and the VIA IDE just started to work finally,
after upgrading my BIOS to 1007, the latest one, which came out on the same
day as your 1009.  I never experienced "data corruption" as you described,
just irq timeouts which crashed fatally.  It seems like everything is normal
now, in other words, I could install Mandrake 8.1 with no problems whereas
it failed dozens of times before.  There was one other I did different this
time, and that was that I created the partitions in Partition Magic 7.0 for
Windows instead of using DiskDrake.  I can't see this would have any bearing
on things, so I can only attribute the recent successful install to ASUS's
new BIOS.

David Grant

On Saturday October 27, Bryan O'Sullivan said:
>
>After several months of begrudgingly putting up with my ASUS A7V
>motherboard corrupting roughly 1 byte per 100 million read during
>moderate to heavy PCI bus activity, I flashed VIA's 1009 BIOS this
>evening.
>
>I have not been able to reproduce any corruption since then (it was
>ridiculously easy before the new BIOS), and my machine seems otherwise
>as stable as I would hope. This marks the first time since 2.4.6 that
>I've been able to run a Linus kernel without cowering.
>

Re: VIA KT133 data corruption update

[Sam Vilain]

On 27 Oct 2001 03:48:56 -0700
"Bryan O'Sullivan" <bos@serpentine.com> wrote:

> I also discovered, of necessity, a halfway manageable process for
> creating a DOS boot floppy using Windows ME, which Microsoft would
> apparently prefer was not possible.  I'll reproduce the steps here,
> since otherwise flashing a new BIOS is likely to be nightmarish for
> people stuck dual booting into WinME.

I notice that Dell use FreeDOS on their utility disks that do this sort of
thing.  Dell++.  Perhaps you might have more luck using that over
WinBlowsME.

http://www.freedos.org/

Sam.

Re: VIA KT133 data corruption update

[Martin Eriksson]

----- Original Message -----
From: "Sam Vilain" <sam@vilain.net>
To: "Bryan O'Sullivan" <bos@serpentine.com>
Cc: <linux-kernel@vger.kernel.org>
Sent: Tuesday, October 30, 2001 2:26 PM
Subject: Re: VIA KT133 data corruption update


<snip>
>
> I notice that Dell use FreeDOS on their utility disks that do this sort of
> thing.  Dell++.  Perhaps you might have more luck using that over
> WinBlowsME.
>
> http://www.freedos.org/
>

Even better: www.bootdisk.org

Re: VIA KT133 data corruption update

[Sam Vilain]

"Martin Eriksson" <nitrax@giron.wox.org> wrote:

> > I notice that Dell use FreeDOS on their utility disks that do this
> > sort of thing.  Dell++.  Perhaps you might have more luck using that
> > over WinBlowsME.
> > http://www.freedos.org/
> Even better: www.bootdisk.org

www.whydidntithinkofthat.com

Don't know about you, but I can't get to bootdisk.org's nameservers.  The
site isn't indexed by Google, either.

I tried bootdisk.com, but they have a distinctly w4r3z feel to them.  That
tends to turn me off.  Strictly speaking, you're not supposed to
distribute those images to people without a license...

Sam.