[PATCH] fix object/driver confusion of all northbridge code

[Stefan Reinauer]

See patch

[PATCH] fix object/driver confusion of all northbridge code

[Joseph Smith]

On Tue, 24 Mar 2009 12:41:15 +0100, Stefan Reinauer 
wrote:
> See patch

This may be trivial, but in case not
Acked-by: Joseph Smith

[PATCH] fix object/driver confusion of all northbridge code

[Peter Stuge]

Stefan Reinauer wrote:
> Every object file with a struct pci_driver ... __pci_driver needs to
> be marked as "driver" instead of "object" in order to get the init
> code actually executed.
>
> This patch fixes up all northbridges that did not do this before.
>
> Signed-off-by: Stefan Reinauer

Acked-by: Peter Stuge

How come we have not noticed this before? Do these NB drivers not do
anything critical?

//Peter

[PATCH] fix object/driver confusion of all northbridge code

[Corey Osgood]

On Tue, Mar 24, 2009 at 12:06 PM, Peter Stuge  wrote:
> Stefan Reinauer wrote:
> > Every object file with a struct pci_driver ... __pci_driver needs to
> > be marked as "driver" instead of "object" in order to get the init
> > code actually executed.
> >
> > This patch fixes up all northbridges that did not do this before.
> >
> > Signed-off-by: Stefan Reinauer
>
> Acked-by: Peter Stuge
>
> How come we have not noticed this before? Do these NB drivers not do
> anything critical?

Please hold off until tomorrow on committing this. IIRC I tried this with
cn700 and something broke, but I don't remember what or why.

Thanks,
Corey

[PATCH] fix object/driver confusion of all northbridge code

[Peter Stuge]

Corey Osgood wrote:
> Please hold off until tomorrow on committing this. IIRC I tried
> this with cn700 and something broke, but I don't remember what or
> why.

It's already in now, but can be reverted of course. It's an
interesting bug..

//Peter

[PATCH] fix object/driver confusion of all northbridge code

[Joseph Smith]

On Tue, 24 Mar 2009 17:17:50 +0100, Peter Stuge  wrote:
> Corey Osgood wrote:
> > Please hold off until tomorrow on committing this. IIRC I tried
> > this with cn700 and something broke, but I don't remember what or
> > why.
>
> It's already in now, but can be reverted of course. It's an
> interesting bug..

I don't think it is a good idea to revert if cn700 is the only one broken.
I have tested the i82830 and it works fine. For the fist time I can see
"Northbridge init", not that it does anything but it is nice to know the
code is working now, just in case I need to do something with it in the
future. Maybe another patch switching cn700 back if confirmed it is broken?

[PATCH] fix object/driver confusion of all northbridge code

[Corey Osgood]

On Tue, Mar 24, 2009 at 3:24 PM, Joseph Smith  wrote:
> On Tue, 24 Mar 2009 17:17:50 +0100, Peter Stuge  wrote:
> > Corey Osgood wrote:
> > > Please hold off until tomorrow on committing this. IIRC I tried
> > > this with cn700 and something broke, but I don't remember what or
> > > why.
> >
> > It's already in now, but can be reverted of course. It's an
> > interesting bug..
>
> I don't think it is a good idea to revert if cn700 is the only one broken.
> I have tested the i82830 and it works fine. For the fist time I can see
> "Northbridge init", not that it does anything but it is nice to know the
> code is working now, just in case I need to do something with it in the
> future. Maybe another patch switching cn700 back if confirmed it is broken?

It doesn't break abuild. If something does break, it's probably a problem
with the code and not the config, I'll fix it this weekend if that's the
case.

-Corey

[PATCH] fix object/driver confusion ofall northbridge code

[Myles Watson]

> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]]
> On Behalf Of Peter Stuge
> Sent: Tuesday, March 24, 2009 10:06 AM
> To: [e-mail]
> Subject: Re: [coreboot] [PATCH] fix object/driver confusion ofall
> northbridge code
>
> Stefan Reinauer wrote:
> > Every object file with a struct pci_driver ... __pci_driver needs to
> > be marked as "driver" instead of "object" in order to get the init
> > code actually executed.
> >
> > This patch fixes up all northbridges that did not do this before.
> >
> > Signed-off-by: Stefan Reinauer
>
> Acked-by: Peter Stuge
>
> How come we have not noticed this before? Do these NB drivers not do
> anything critical?

It is hard to believe that the amdk8 northbridge code never got executed.
Is this only true if it's not contained in a "chip", or if you don't specify
its position in the device tree?

Thanks,
Myles

[PATCH] fix object/driver confusion ofall northbridge code

[Corey Osgood]

On Tue, Mar 24, 2009 at 12:32 PM, Myles Watson  wrote:
> > -----Original Message-----
> > From: [e-mail] [mailto:
>
> [e-mail]]
> > On Behalf Of Peter Stuge
> > Sent: Tuesday, March 24, 2009 10:06 AM
> > To: [e-mail]
> > Subject: Re: [coreboot] [PATCH] fix object/driver confusion ofall
> > northbridge code
> >
> > Stefan Reinauer wrote:
> > > Every object file with a struct pci_driver ... __pci_driver needs to
> > > be marked as "driver" instead of "object" in order to get the init
> > > code actually executed.
> > >
> > > This patch fixes up all northbridges that did not do this before.
> > >
> > > Signed-off-by: Stefan Reinauer
> >
> > Acked-by: Peter Stuge
> >
> > How come we have not noticed this before? Do these NB drivers not do
> > anything critical?
>
> It is hard to believe that the amdk8 northbridge code never got executed.
> Is this only true if it's not contained in a "chip", or if you don't
> specify
> its position in the device tree?

I'm grasping at straws, but I think it had something to do with multiple
drivers in northbridge.c, if it's marked as a driver only the first driver
would get run, or it might have been something to do with the special pci
domain and cpu bus drivers. Then again, I could be entirely wrong, it's been
a while.

-Corey

[PATCH] fix object/driver confusion ofall northbridge code

[Joseph Smith]

On Tue, 24 Mar 2009 12:53:53 -0400, Corey Osgood 
wrote:
> On Tue, Mar 24, 2009 at 12:32 PM, Myles Watson  wrote:
> > > -----Original Message-----
> > > From: [e-mail] [mailto:
> >
> > [e-mail]]
> > > On Behalf Of Peter Stuge
> > > Sent: Tuesday, March 24, 2009 10:06 AM
> > > To: [e-mail]
> > > Subject: Re: [coreboot] [PATCH] fix object/driver confusion ofall
> > > northbridge code
> > >
> > > Stefan Reinauer wrote:
> > > > Every object file with a struct pci_driver ... __pci_driver needs to
> > > > be marked as "driver" instead of "object" in order to get the init
> > > > code actually executed.
> > > >
> > > > This patch fixes up all northbridges that did not do this before.
> > > >
> > > > Signed-off-by: Stefan Reinauer
> > >
> > > Acked-by: Peter Stuge
> > >
> > > How come we have not noticed this before? Do these NB drivers not do
> > > anything critical?
> >
> > It is hard to believe that the amdk8 northbridge code never got
>
> executed.
> > Is this only true if it's not contained in a "chip", or if you don't
> > specify
> > its position in the device tree?
>
> I'm grasping at straws, but I think it had something to do with multiple
> drivers in northbridge.c, if it's marked as a driver only the first

driver
> would get run, or it might have been something to do with the special pci
> domain and cpu bus drivers. Then again, I could be entirely wrong, it's
> been
> a while.

I don't have access right now, can someone do a abuild?