[LinuxBIOS] [PATCH] Tyan S1846: Device tree fixes

[Uwe Hermann]

The static device tree of the Tyan S1846 was, um... totally bogus.

I'm using the ide0_enble/ide1_enable variables now to enable IDE.
Hm, what happens if those variables are not set at all in a Confib.lb
file of some mainboard? Do they default to zero?

Uwe.

[roger]

On Tue, 2007-05-29 at 06:37 +0200, Uwe Hermann wrote:
> The static device tree of the Tyan S1846 was, um... totally bogus.
>
> I'm using the ide0_enble/ide1_enable variables now to enable IDE.
> Hm, what happens if those variables are not set at all in a Confib.lb
> file of some mainboard? Do they default to zero?

I've applied both, this patch, and the 82371 patch.

1) I get a random error when in filo/grub menu.lst after pressing "p"
for password to manually edit my options while in grub.
A quick typed, not exact copy of the error --> "Attempt to free
non-allocated memory: 00000006".

2) /dev/sda2 seems to be easily recognized now.  (I'm guessing this is
the device tree fix?)

3) I now get a long hang at "Loading kernel".  I then get a very very,
if not sometimes permanent hang after "Jumping to Entry Point" is
printed to tty.  Sometimes the kernel loads at this point, sometimes it
never seems to. (82371 patch?)

(If I can get a minute or two more tonight, I'll try reversing the
patches one at a time, my guess is, the 82371 might be causing the
hang.)

Posted session logs at:
http://www.eskimo.com/~roger/programming/linuxbios/index.html
http://www.eskimo.com/~roger/files/linuxbios/440BX.S1832DL/session-20070528b...

Mon May 28 23:59:30 PDT 2007

[Corey Osgood]

roger wrote:
> On Tue, 2007-05-29 at 06:37 +0200, Uwe Hermann wrote:
> > The static device tree of the Tyan S1846 was, um... totally bogus.
> >
> > I'm using the ide0_enble/ide1_enable variables now to enable IDE.
> > Hm, what happens if those variables are not set at all in a Confib.lb
> > file of some mainboard? Do they default to zero?
>
> I've applied both, this patch, and the 82371 patch.
> I get a random error when in filo/grub menu.lst after pressing "p"
>
> for password to manually edit my options while in grub.
> A quick typed, not exact copy of the error --> "Attempt to free
> non-allocated memory: 00000006".

Would that be the 640k-768k memory range? I honestly have no clue, but
it seems to make sense...
> /dev/sda2 seems to be easily recognized now.  (I'm guessing this is
>
> the device tree fix?)

Are you complaining? ;) Yeah, this is probably due to the device tree
fixes. If I were you though, I'd grab a spare IDE drive and try to get
that working first, before trying with add-on cards.
> I now get a long hang at "Loading kernel".  I then get a very very,
>
> if not sometimes permanent hang after "Jumping to Entry Point" is
> printed to tty.  Sometimes the kernel loads at this point, sometimes it
> never seems to. (82371 patch?)

I'm getting the same thing on the i810 (but it's progress for me!). I
figured it was just because filo was only doing byte reads, or is filo
not in charge any more at this point? Anyways, need to dig up Peter's
patch...

-Corey

[Uwe Hermann]

On Tue, May 29, 2007 at 03:41:01AM -0400, Corey Osgood wrote:
> > I get a random error when in filo/grub menu.lst after pressing "p"
> >
> > for password to manually edit my options while in grub.
> > A quick typed, not exact copy of the error --> "Attempt to free
> > non-allocated memory: 00000006".

My guess is that the RAM timings are incorrect for your board, thus you
get random hangs(?)
> > /dev/sda2 seems to be easily recognized now.  (I'm guessing this is
> >
> > the device tree fix?)
>
> Are you complaining? ;) Yeah, this is probably due to the device tree
> fixes.

That, and/or the IDE init code in the southbridge.
> > I now get a long hang at "Loading kernel".  I then get a very very,
> >
> > if not sometimes permanent hang after "Jumping to Entry Point" is
> > printed to tty.  Sometimes the kernel loads at this point, sometimes it
> > never seems to. (82371 patch?)
>
> I'm getting the same thing on the i810 (but it's progress for me!). I
> figured it was just because filo was only doing byte reads, or is filo
> not in charge any more at this point? Anyways, need to dig up Peter's
> patch...

How long are you waiting at the "Loading kernel"? It _can_ take quite a
while, don't reboot too fast here...

Uwe.

[roger]

On Tue, 2007-05-29 at 13:46 +0200, Uwe Hermann wrote:
> On Tue, May 29, 2007 at 03:41:01AM -0400, Corey Osgood wrote:
> > > I get a random error when in filo/grub menu.lst after pressing "p"
> > >
> > > for password to manually edit my options while in grub.
> > > A quick typed, not exact copy of the error --> "Attempt to free
> > > non-allocated memory: 00000006".
>
> My guess is that the RAM timings are incorrect for your board, thus you
> get random hangs(?)

Could be.  Just remember, this is the S1832 board.
> > > /dev/sda2 seems to be easily recognized now.  (I'm guessing this is
> > >
> > > the device tree fix?)
> >
> > Are you complaining? ;) Yeah, this is probably due to the device tree
> > fixes.
>
> That, and/or the IDE init code in the southbridge.
> > > I now get a long hang at "Loading kernel".  I then get a very very,
> > >
> > > if not sometimes permanent hang after "Jumping to Entry Point" is
> > > printed to tty.  Sometimes the kernel loads at this point, sometimes it
> > > never seems to. (82371 patch?)
> >
> > I'm getting the same thing on the i810 (but it's progress for me!). I
> > figured it was just because filo was only doing byte reads, or is filo
> > not in charge any more at this point? Anyways, need to dig up Peter's
> > patch...
>
> How long are you waiting at the "Loading kernel"? It _can_ take quite a
> while, don't reboot too fast here...

5-10 minutes waiting, but it could take 6-11 minutes.  You're probably
right.  One or two times, it just continued right before the 82371 or
device patch.

Really, I need this all calls for more debugging, but I just wanted to
get some logs up before I split for the week.

Tue May 29 11:02:26 PDT 2007

[Corey Osgood]

Uwe Hermann wrote:
>  device pci 7.1 on end		# IDE
>  device pci 7.2 on end		# USB
>  device pci 7.3 on end		# ACPI
>  register "ide0_enable" = "1"
>  register "ide1_enable" = "1"

Shouldn't this be

device pci 7.1 on 		# IDE
     register "ide0_enable" = "1"
     register "ide1_enable" = "1"
end
device pci 7.2 on end		# USB
device pci 7.3 on end		# ACPI

?

-corey

[Uwe Hermann]

On Tue, May 29, 2007 at 03:47:14AM -0400, Corey Osgood wrote:
> Uwe Hermann wrote:
> >  device pci 7.1 on end		# IDE
> >  device pci 7.2 on end		# USB
> >  device pci 7.3 on end		# ACPI
> >  register "ide0_enable" = "1"
> >  register "ide1_enable" = "1"
>
> Shouldn't this be
>
> device pci 7.1 on 		# IDE
>      register "ide0_enable" = "1"
>      register "ide1_enable" = "1"
> end
> device pci 7.2 on end		# USB
> device pci 7.3 on end		# ACPI

Nope, trying that I get:

===> ERROR: Only chips can have register values

I tested the version in my patch and I can confirm that it works, I
tried disabling IDE0 and FILO didn't find the disk anymore...

Uwe.

[Stefan Reinauer]

* Uwe Hermann  [070529 06:37]:
> The static device tree of the Tyan S1846 was, um... totally bogus.
>
> I'm using the ide0_enble/ide1_enable variables now to enable IDE.
> Hm, what happens if those variables are not set at all in a Confib.lb
> file of some mainboard? Do they default to zero?
> Uwe.
>
> http://www.hermann-uwe.de  | http://www.holsham-traders.de
> http://www.crazy-hacks.org | http://www.unmaintained-free-software.org
> Fix the static device tree of the Tyan S1846. Especially the
> Super I/O part was incorrect.
>
> Also, add ide0_enable/ide1_enable variables, and enable both the
> primary and secondary IDE interface per default.
> Signed-off-by: Uwe Hermann

I dont have the board, but I think it makes sense. Please someone else
intervent otherwise. I say:

Acked-by: Stefan Reinauer

[Uwe Hermann]

On Sun, Jun 03, 2007 at 09:31:15PM +0200, Stefan Reinauer wrote:
> Acked-by: Stefan Reinauer

Thanks, r2708.

Uwe.