v3 ck804 support

[Myles Watson]

This patch adds southbridge/nvidia/ck804 from v2 minimally updated for v3.

Signed-off-by: Myles Watson

Thanks,
Myles

[Peter Stuge]

Myles Watson wrote:
> This patch adds southbridge/nvidia/ck804 from v2 minimally updated
> for v3.

Mostly fine, but..
> /* If that is invalid we will read that from romstrap. */
> if (!eeprom_valid) {
> unsigned long mac_pos;
> mac_pos = 0xffffffd0; /* See romstrap.inc and romstrap.lds. */
> mac_l = readl(mac_pos) + nic_index;
> mac_h = readl(mac_pos + 4);
> }

..this won't work so well in v3 I think. I'd like a LAR file for the
MAC.
> +#if CONFIG_PCI_ROM_RUN == 1
> pci_dev_init(dev);	/* It will init Option ROM. */
>
> +#endif

Should this be in the NIC code?
> +// 0x7a or e3
> +#define PREVIOUS_POWER_STATE 0x7A
>
> +#define MAINBOARD_POWER_OFF 0
> +#define MAINBOARD_POWER_ON 1
> +#define SLOW_CPU_OFF 0
> +#define SLOW_CPU__ON 1
>
> +#ifndef MAINBOARD_POWER_ON_AFTER_POWER_FAIL
> +#define MAINBOARD_POWER_ON_AFTER_POWER_FAIL MAINBOARD_POWER_ON
> +#endif

We should make this a CONFIG_ variable.

Anyway:
Acked-by: Peter Stuge

[Myles Watson]

On Mon, Jan 5, 2009 at 3:36 PM, Peter Stuge  wrote:
> Myles Watson wrote:
> > This patch adds southbridge/nvidia/ck804 from v2 minimally updated
> > for v3.
>
> Mostly fine, but..
> > /* If that is invalid we will read that from romstrap. */
> > if (!eeprom_valid) {
> >         unsigned long mac_pos;
> >         mac_pos = 0xffffffd0; /* See romstrap.inc and romstrap.lds. */
> >         mac_l = readl(mac_pos) + nic_index;
> >         mac_h = readl(mac_pos + 4);
> > }
>
> ..this won't work so well in v3 I think. I'd like a LAR file for the
> MAC.
> > +#if CONFIG_PCI_ROM_RUN == 1
> > pci_dev_init(dev);      /* It will init Option ROM. */
> >
> > +#endif
>
> Should this be in the NIC code?
> > +// 0x7a or e3
> > +#define PREVIOUS_POWER_STATE 0x7A
> >
> > +#define MAINBOARD_POWER_OFF 0
> > +#define MAINBOARD_POWER_ON 1
> > +#define SLOW_CPU_OFF 0
> > +#define SLOW_CPU__ON 1
> >
> > +#ifndef MAINBOARD_POWER_ON_AFTER_POWER_FAIL
> > +#define MAINBOARD_POWER_ON_AFTER_POWER_FAIL MAINBOARD_POWER_ON
> > +#endif
>
> We should make this a CONFIG_ variable.
>
> Anyway:
> Acked-by: Peter Stuge

Rev 1100.

I added your concerns to the commit message so they won't be forgotten.

Thanks,
Myles

[Peter Stuge]

Myles Watson wrote:
> I added your concerns to the commit message so they won't be
> forgotten.

Didn't see them, but anyway maybe TODOs in the source would be
better.

//Peter

[Myles Watson]

On Mon, Jan 5, 2009 at 4:12 PM, Peter Stuge  wrote:
> Myles Watson wrote:
> > I added your concerns to the commit message so they won't be
> > forgotten.
>
> Didn't see them, but anyway maybe TODOs in the source would be
> better.

Sure.  I'll put them in.

Thanks,
Myles

[ron minnich]

On Mon, Jan 5, 2009 at 2:36 PM, Peter Stuge  wrote:
> Myles Watson wrote:
> > This patch adds southbridge/nvidia/ck804 from v2 minimally updated
> > for v3.
>
> Mostly fine, but..
> > /* If that is invalid we will read that from romstrap. */
> > if (!eeprom_valid) {
> >         unsigned long mac_pos;
> >         mac_pos = 0xffffffd0; /* See romstrap.inc and romstrap.lds. */
> >         mac_l = readl(mac_pos) + nic_index;
> >         mac_h = readl(mac_pos + 4);
> > }
>
> ..this won't work so well in v3 I think. I'd like a LAR file for the
> MAC.

Yep. Put a #warning here so we can fix it.
> > +#if CONFIG_PCI_ROM_RUN == 1
> > pci_dev_init(dev);      /* It will init Option ROM. */
> >
> > +#endif
>
> Should this be in the NIC code?

Seems that way to me.
> > +// 0x7a or e3
> > +#define PREVIOUS_POWER_STATE 0x7A
> >
> > +#define MAINBOARD_POWER_OFF 0
> > +#define MAINBOARD_POWER_ON 1
> > +#define SLOW_CPU_OFF 0
> > +#define SLOW_CPU__ON 1
> >
> > +#ifndef MAINBOARD_POWER_ON_AFTER_POWER_FAIL
> > +#define MAINBOARD_POWER_ON_AFTER_POWER_FAIL MAINBOARD_POWER_ON
> > +#endif
>
> We should make this a CONFIG_ variable.

Actually it should be a cmos option, as in any real mainboard. Code
for this should not be #ifdef'ed.

A #warning here would be useful too.

thanks

ron

[Myles Watson]

On Wed, Jan 7, 2009 at 10:29 AM, ron minnich  wrote:
> On Mon, Jan 5, 2009 at 2:36 PM, Peter Stuge  wrote:
> > Myles Watson wrote:
> > > This patch adds southbridge/nvidia/ck804 from v2 minimally updated
> > > for v3.
> >
> > Mostly fine, but..
> > > /* If that is invalid we will read that from romstrap. */
> > > if (!eeprom_valid) {
> > >         unsigned long mac_pos;
> > >         mac_pos = 0xffffffd0; /* See romstrap.inc and romstrap.lds. */
> > >         mac_l = readl(mac_pos) + nic_index;
> > >         mac_h = readl(mac_pos + 4);
> > > }
> >
> > ..this won't work so well in v3 I think. I'd like a LAR file for the
> > MAC.
>
> Yep. Put a #warning here so we can fix it.

Added.
> > > +#if CONFIG_PCI_ROM_RUN == 1
> > > pci_dev_init(dev);      /* It will init Option ROM. */
> > >
> > > +#endif
> >
> > Should this be in the NIC code?

There could be an option ROM for the NIC.
> > > +// 0x7a or e3
> > > +#define PREVIOUS_POWER_STATE 0x7A
> > >
> > > +#define MAINBOARD_POWER_OFF 0
> > > +#define MAINBOARD_POWER_ON 1
> > > +#define SLOW_CPU_OFF 0
> > > +#define SLOW_CPU__ON 1
> > >
> > > +#ifndef MAINBOARD_POWER_ON_AFTER_POWER_FAIL
> > > +#define MAINBOARD_POWER_ON_AFTER_POWER_FAIL MAINBOARD_POWER_ON
> > > +#endif
> >
> > We should make this a CONFIG_ variable.
>
> Actually it should be a cmos option, as in any real mainboard. Code
> for this should not be #ifdef'ed.
>
> A #warning here would be useful too.

Added.

Rev 1104.  Thanks for the reminder.

Thanks,
Myles

[Peter Stuge]

Myles Watson wrote:
> > > > +#if CONFIG_PCI_ROM_RUN == 1
> > > > pci_dev_init(dev);      /* It will init Option ROM. */
> > > >
> > > > +#endif
> > >
> > > Should this be in the NIC code?
>
> There could be an option ROM for the NIC.

The point was that this wasn't in the NIC code.

Thanks for adding the warnings.

//Peter

[Myles Watson]

On Wed, Jan 7, 2009 at 1:12 PM, Peter Stuge  wrote:
> Myles Watson wrote:
> > > > > +#if CONFIG_PCI_ROM_RUN == 1
> > > > > pci_dev_init(dev);      /* It will init Option ROM. */
> > > > >
> > > > > +#endif
> > > >
> > > > Should this be in the NIC code?
> >
> > There could be an option ROM for the NIC.
>
> The point was that this wasn't in the NIC code.

Sorry I misunderstood.  It's in both places, ide and nic.  I just took
it from v2.
> Thanks for adding the warnings.

Sure.

Thanks,
Myles