[Qemu-devel] Open Hack'Ware test image.

[Qemu-devel] Open Hack'Ware test image.

[J. Mayer]

A binary and svn snapshot of qemu PPC BIOS (version 0.4-pre1d) is now
available here:
<http://perso.magic.fr/l_indien/OpenHackWare/>

The BIOS code have been reorganized. It also include new features
including a console with ADB/PC keyboard input.
It includes bugfixes, including a major HFS rework, from Thayne Harbaugh
& Matthew S. Wood.

With this release, most PowerMac boot CDROM seem recognized. All Linux
2.4 kernel based distributions should boot. A few 2.6 ones should also
boot OK. Please note that early 2.6 kernels frame-buffer drivers are
broken, so some distributions may boot without any output on the
console.
Debug is on the way to find other known issues with 2.6 kernels
(including IDE & ADB problems).
Most other OSes will start booting and then crash sooner or later
because of:
- lack of executable file format support (OpenBSD)
- buggy OF tree (NetBSD)
- other unknown/not clearly identified problems (Darwin, MacOS X 10.2 &
10.3, MacOS 9).
Debug is in progress to fix ISO9660 El-torito boot as well as some known
bugs in executable files loaders.

Note that PREP support has not been checked with this snapshot (but will
be before it'll become a release).

Please report any other bug, problem or success.

-- 
J. Mayer <l_indien@magic.fr>
Never organized

[Qemu-devel] Re: Open Hack'Ware test image.

[Ronald]

Le Tue, 22 Feb 2005 12:45:20 +0100, J. Mayer a écrit :

> A binary and svn snapshot of qemu PPC BIOS (version 0.4-pre1d) is now
> available here:
> <http://perso.magic.fr/l_indien/OpenHackWare/>
> 
[...]
> 
> Please report any other bug, problem or success.

I am not able to boot anything with it, I have tried 2.4.29 prep, pmac and
chrp. I only get:
Booting from device m
...

ERROR: Unable to get memory bloc device

I use latest cvs.
  

[Qemu-devel] Re: Open Hack'Ware test image.

[Thayne Harbaugh]

On Tue, 2005-02-22 at 12:45 +0100, J. Mayer wrote:

> Note that PREP support has not been checked with this snapshot (but will
> be before it'll become a release).

This won't work because the PCI host bridge *still* hasn't been added to
qemu.  I've sent the patch at least twice.  Tha patch hasn't been
applied and there hasn't been an explanation as to why it was ignored.

Right now qemu/hw/pci.c:pci_prep_init() only adds a pci-to-pci bridge
which won't get very far if it doesn't have a pci host bridge to attach
to.

Re: [Qemu-devel] Re: Open Hack'Ware test image.

[Fabrice Bellard]

Thayne Harbaugh wrote:
> On Tue, 2005-02-22 at 12:45 +0100, J. Mayer wrote:
> 
> 
>>Note that PREP support has not been checked with this snapshot (but will
>>be before it'll become a release).
> 
> 
> This won't work because the PCI host bridge *still* hasn't been added to
> qemu.  I've sent the patch at least twice.  Tha patch hasn't been
> applied and there hasn't been an explanation as to why it was ignored.

Hi,

Sorry for the lack of explanation: I was not sure whether QEMU could 
work with your patch without an updated PPC BIOS.

Fabrice.

Re: [Qemu-devel] Re: Open Hack'Ware test image.

[Thayne Harbaugh]

On Tue, 2005-02-22 at 19:58 +0100, Fabrice Bellard wrote:
> Thayne Harbaugh wrote:
> > On Tue, 2005-02-22 at 12:45 +0100, J. Mayer wrote:
> > 
> > 
> >>Note that PREP support has not been checked with this snapshot (but will
> >>be before it'll become a release).
> > 
> > 
> > This won't work because the PCI host bridge *still* hasn't been added to
> > qemu.  I've sent the patch at least twice.  Tha patch hasn't been
> > applied and there hasn't been an explanation as to why it was ignored.
> 
> Hi,
> 
> Sorry for the lack of explanation: I was not sure whether QEMU could 
> work with your patch without an updated PPC BIOS.

Yes, *both* Qemu and OHW need to be patched.  It shouldn't matter which
one supports the host bridge first.  You shouldn't even worry about
making PReP unstable by adding support for the PCI host bridge to either
qemu or OHW because PReP isn't stable now - it doesn't even boot without
the host bridge.

[Qemu-devel] Re: Re: Open Hack'Ware test image.

[Ronald]

Le Tue, 22 Feb 2005 14:02:35 -0700, Thayne Harbaugh a écrit :

> You shouldn't even worry about making
> PReP unstable by adding support for the PCI host bridge to either qemu or
> OHW because PReP isn't stable now - it doesn't even boot without the host
> bridge.

Sorry but prep works fine for me, that's even the only ppc emulation mode
I can use.

Re: [Qemu-devel] Re: Re: Open Hack'Ware test image.

[Thayne Harbaugh]

On Tue, 2005-02-22 at 22:53 +0100, Ronald wrote:
> Le Tue, 22 Feb 2005 14:02:35 -0700, Thayne Harbaugh a écrit :
> 
> > You shouldn't even worry about making
> > PReP unstable by adding support for the PCI host bridge to either qemu or
> > OHW because PReP isn't stable now - it doesn't even boot without the host
> > bridge.
> 
> Sorry but prep works fine for me, that's even the only ppc emulation mode
> I can use.

You've applied the "Raven" host bridge patch and it breaks your PReP?

Until someone can tell me that having a correct PCI host bridge actually
breaks something I'll continue to ask that the patch be applied.  I
believe fixes some problems and that even in the working situations it
makes things more correct.  Real PReP hardware has a PCI host bridge -
whether it's the Motorola Raven, Eagle, Grackle or whatever.

If someone can show that having a correct PCI host bridge breaks
something then I'll either rework the patch or drop it.  I know that I
have created some broken patches before (HFS comes to mind) and I don't
have a problem rethinking.

Re: [Qemu-devel] Re: Re: Open Hack'Ware test image.

[J. Mayer]

On Wed, 2005-02-23 at 00:14, Thayne Harbaugh wrote:
> On Tue, 2005-02-22 at 22:53 +0100, Ronald wrote:
> > Le Tue, 22 Feb 2005 14:02:35 -0700, Thayne Harbaugh a écrit :
> > 
> > > You shouldn't even worry about making
> > > PReP unstable by adding support for the PCI host bridge to either qemu or
> > > OHW because PReP isn't stable now - it doesn't even boot without the host
> > > bridge.
> > 
> > Sorry but prep works fine for me, that's even the only ppc emulation mode
> > I can use.
> 
> You've applied the "Raven" host bridge patch and it breaks your PReP?
> 
> Until someone can tell me that having a correct PCI host bridge actually
> breaks something I'll continue to ask that the patch be applied.  I
> believe fixes some problems and that even in the working situations it
> makes things more correct.  Real PReP hardware has a PCI host bridge -
> whether it's the Motorola Raven, Eagle, Grackle or whatever.
> 
> If someone can show that having a correct PCI host bridge breaks
> something then I'll either rework the patch or drop it.  I know that I
> have created some broken patches before (HFS comes to mind) and I don't
> have a problem rethinking.

You're absolutely right, having no PCI host bridge is not correct but
should not prevent OSes to run then is not a problem for Linux.

There is absolutely no reason that your patch may break anything. It may
also help some other OSes to run.

The only problem here is that qemu & OpenHackWare have to stay
synchronised on this point. As the OHW used in qemu CVS is 0.3 and as it
does not recognize Raven bridge, then qemu has to stay with its current
configuration or the PCI bus won't get initialized by the BIOS, and this
may break the target (even if it's not sure).
When OHW 0.4 will be stable and will be committed in qemu CVS, then qemu
will have to apply the patch so PREP will still not be broken, for the
same reason.
In the meantime, people who want to test PREP target with OHW 0.4
preview need to apply the Raven patch to qemu.

-- 
J. Mayer <l_indien@magic.fr>
Never organized

Re: [Qemu-devel] Re: Open Hack'Ware test image.

[J. Mayer]

On Tue, 2005-02-22 at 22:02, Thayne Harbaugh wrote:
> On Tue, 2005-02-22 at 19:58 +0100, Fabrice Bellard wrote:
> > Thayne Harbaugh wrote:
> > > On Tue, 2005-02-22 at 12:45 +0100, J. Mayer wrote:
> > > 
> > > 
> > >>Note that PREP support has not been checked with this snapshot (but will
> > >>be before it'll become a release).
> > > 
> > > 
> > > This won't work because the PCI host bridge *still* hasn't been added to
> > > qemu.  I've sent the patch at least twice.  Tha patch hasn't been
> > > applied and there hasn't been an explanation as to why it was ignored.
> > 
> > Hi,
> > 
> > Sorry for the lack of explanation: I was not sure whether QEMU could 
> > work with your patch without an updated PPC BIOS.
> 
> Yes, *both* Qemu and OHW need to be patched.  It shouldn't matter which
> one supports the host bridge first.  You shouldn't even worry about
> making PReP unstable by adding support for the PCI host bridge to either
> qemu or OHW because PReP isn't stable now - it doesn't even boot without
> the host bridge.

What you say is trange:
it used to boot with the current qemu configuration. As PREP is not
supposed to use the open-firmware interface, it does not care about
seing or not the PCI host bridge in the tree.
So, imho, it would be a good choice not to make a change into qemu until
we get a new stable OHW BIOS.

-- 
J. Mayer <l_indien@magic.fr>
Never organized