Re: Solving suspend-level confusion

[David Brownell <david-b@pacbell.net>]

On Saturday 31 July 2004 10:01, Oliver Neukum wrote:
> 
> Maybe a better approach would be to describe the required features to
> the drivers rather than encoding them in a single integer. Rather
> like passing a request that states "lowest power level with device state
> retained, must not do DMA, enable remote wake up"

A pointer to some sort of struct could be generic and typesafe;
better than an integer or enum.

	// <linux/device.h>
	struct device {
		... unchanged; but fix or remove this:
		struct dev_pm_state *power_state;
		...
	}; 

	// <linux/pm.>
	struct dev_pm_state {
		char *name;
		unsigned number;
		// maybe more
	};
	struct dev_pm_info {
		struct dev_pm_state	*power_state;
		struct dev_pm_state	**supported;
		unsigned could_wakeup:1;
		unsigned should_wakeup:1;
		... plus the rest
	};

There would be bus-specific rules about how those get used,
for example what device power states exist. 

	// <linux/pci.h>
	extern const struct dev_pm_state PCI_D0, PCI_D1, PCI_D2,
				PCI_D3cold, PCI_D3hot;

	... eventually signatures change:
	struct pci_driver {
		int  (*suspend) (struct pci_dev *dev, struct dev_pm_state *state);
		int  (*enable_wake) (struct pci_dev *dev, struct dev_pm *state,
					int enable);
	};
	int pci_set_power_state(struct pci_dev *dev, struct dev_pm *state);
	int pci_enable_wake(struct pci_dev *dev, struct dev_pm *state, int enable);

	// <asm/arch/bus.h>
	... support for whatever non-PCI device states work here
	... maybe even special states for the platform's busses:
	struct mybus_dev_pm_state {
		struct dev_pm_state public;
		// private state:  clock hooks, power switches,
		// linkage to devices sharing clock or power, etc
	};

So for example maybe dev->power.power_state == PCI_D0, and
dev->power->supported == { PCI_D0, PCI_D3hot, PCI_D3cold, NULL }
on some controller, and sysfs would say "D3hot" not "3" (or "2" or
whatever).  Other busses would report different states..

- Dave