On Fri, Aug 18, 2023 at 10:31:03AM +0200, Rafael J. Wysocki wrote:
> On Fri, Aug 18, 2023 at 7:15 AM Mario Limonciello
> <mario.limonciello@xxxxxxx> wrote:
...
> > +int acpi_get_lps0_constraint(struct device *dev)
>
> I think that some overhead would be reduced below if this were taking
> a struct acpi_device pointer as the argument.
Hmm... Either you need a pointer to handle, which involves pointer arithmetics
or something else. I would believe if you tell that ACPI handle should be passed,
but current suggestion is not obvious to me how it may help.
> > +{
> > + struct lpi_constraints *entry;
> > +
> > + for_each_lpi_constraint(entry) {
> > + if (!device_match_acpi_handle(dev, entry->handle))
Here we retrieve handle...
> > + continue;
> > + acpi_handle_debug(entry->handle,
> > + "ACPI device constraint: %d\n", entry->min_dstate);
> > + return entry->min_dstate;
> > + }
> > + dev_dbg(dev, "No ACPI device constraint specified\n");
...and here we are using dev directly (otherwise acpi_handle_dbg() should be used).
> > + return -ENODEV;
>
> ACPI_STATE_UNKNOWN?
>
> > +}
--
With Best Regards,
Andy Shevchenko
