Re: Cleanup PCI power states

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > > Care to resend this, I seem to have lost them :(
> > 
> > Could this go to "after 2.6.10 tree", too? It is a helper that
> > converts system state into PCI state. We really do not want to have
> > this copied into every driver, because it will need to change when
> > system state gets type-checked / expanded to struct.
> 
> So this is how you want to switch stuff over?  Can you give me an
> example of how this will be used?

Yes. Some way to convert between system-state and pci-state is needed;
for now it only silences warning, but I guess it will get more complex
in future.

Example is here:

--- clean/drivers/net/e100.c	2004-10-01 00:30:15.000000000 +0200
+++ linux/drivers/net/e100.c	2004-11-14 23:36:46.000000000 +0100
@@ -2313,7 +2313,7 @@
 	pci_save_state(pdev, nic->pm_state);
 	pci_enable_wake(pdev, state, nic->flags & (wol_magic | e100_asf(nic)));
 	pci_disable_device(pdev);
-	pci_set_power_state(pdev, state);
+	pci_set_power_state(pdev, pci_choose_state(pdev, state));
 
 	return 0;
 }
@@ -2323,7 +2323,7 @@
 	struct net_device *netdev = pci_get_drvdata(pdev);
 	struct nic *nic = netdev_priv(netdev);
 
-	pci_set_power_state(pdev, 0);
+	pci_set_power_state(pdev, PCI_D0);
 	pci_restore_state(pdev, nic->pm_state);
 	e100_hw_init(nic);
 

> > --- clean/drivers/pci/pci.c	2004-10-01 00:30:16.000000000 +0200
> > +++ linux/drivers/pci/pci.c	2004-11-14 23:36:46.000000000 +0100
> > @@ -300,6 +300,30 @@
> >  }
> >  
> >  /**
> > + * pci_choose_state - Choose the power state of a PCI device
> > + * @dev: PCI device to be suspended
> > + * @state: target sleep state for the whole system
> > + *
> > + * Returns PCI power state suitable for given device and given system
> > + * message.
> > + */
> > +
> > +pci_power_t pci_choose_state(struct pci_dev *dev, u32 state)
> > +{
> > +	if (!pci_find_capability(dev, PCI_CAP_ID_PM))
> > +		return PCI_D0;
> > +
> > +	switch (state) {
> > +	case 0:	return PCI_D0;
> > +	case 2: return PCI_D2;
> > +	case 3: return PCI_D3hot;
> > +	default: BUG();
> > +	}
> > +}
> > +
> > +EXPORT_SYMBOL(pci_choose_state);
> 
> EXPORT_SYMBOL_GPL() perhaps?

Ugh, I do not have any strong feelings here, but I want nvidia etc to
use this one so that they implement driver model properly.

								Pavel
-- 
People were complaining that M$ turns users into beta-testers...
...jr ghea gurz vagb qrirybcref, naq gurl frrz gb yvxr vg gung jnl!