I'm adding Runtime PM support to the SDIO subsystem, and was planning
to use the generic runtime pm handlers.
Those handlers return -EINVAL if the driver didn't explicitly define
the runtime pm handlers:
That may result in some drivers defining nop handlers, just to return
0 (in case there's nothing else they need to do).
Do we want that ?
Alternatively, we may want to allow drivers to enable Runtime PM (by
taking the appropriate action for their subsystem, e.g. calling
put_noidle in probe and get_noresume in remove), but still not define
any runtime pm handlers (implicitly always returning a success), with
something like:
diff --git a/drivers/base/power/generic_ops.c b/drivers/base/power/generic_ops.
index 304c831..531762a 100644
--- a/drivers/base/power/generic_ops.c
+++ b/drivers/base/power/generic_ops.c
@@ -49,7 +49,7 @@ int pm_generic_runtime_suspend(struct device *dev)
const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
int ret;
- ret = pm && pm->runtime_suspend ? pm->runtime_suspend(dev) : -EINVAL;
+ ret = pm && pm->runtime_suspend ? pm->runtime_suspend(dev) : 0;
return ret;
}
@@ -68,7 +68,7 @@ int pm_generic_runtime_resume(struct device *dev)
const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
int ret;
- ret = pm && pm->runtime_resume ? pm->runtime_resume(dev) : -EINVAL;
+ ret = pm && pm->runtime_resume ? pm->runtime_resume(dev) : 0;
return ret;
}
What do you think ?
Please note that the generic pm_generic_runtime_idle handler does
allow drivers not to explicitly define a runtime_idle handler.
Thanks,
Ohad.
_______________________________________________
linux-pm mailing list
linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/linux-pm
