Hi, * Rafael J. Wysocki <r...@rjwysocki.net> [150305 17:38]: > Please always CC linux-pm on CC patches.
Sure will do for the next rev, sorry forgot to add that. > On Thursday, March 05, 2015 04:34:06 PM Tony Lindgren wrote: > > +/** > > + * handle_dedicated_wakeirq - Handler for device wake-up interrupts > > + * @wakeirq: Separate wake-up interrupt for a device different > > + * @_wirq: Wake-up interrupt data > > + * > > + * Some devices have a separate wake-up interrupt in addition to the > > + * regular device interrupt. The wake-up interrupts signal that the > > + * device should be woken up from a deeper idle state. This handler > > + * uses device specific pm_runtime functions to wake-up the device > > + * and then it's up to the device to do whatever it needs to. Note > > + * as the device may need to restore context and start up regulators, > > + * this is not a fast path. > > + * > > + * Note that we are not resending the lost device interrupts. We assume > > + * that the wake-up interrupt just needs to wake-up the device, and > > + * the device pm_runtime_resume() can deal with the situation. > > + */ > > +static irqreturn_t handle_dedicated_wakeirq(int wakeirq, void *_wirq) > > +{ > > + struct wakeirq_source *wirq = _wirq; > > + irqreturn_t ret = IRQ_NONE; > > + > > + /* We don't want RPM_ASYNC or RPM_NOWAIT here */ > > + if (pm_runtime_suspended(wirq->dev)) { > > What if the device is resumed on a different CPU right here? Good point, sounds like we need to do this in some pm_runtime function directly for the locking. > > + pm_runtime_mark_last_busy(wirq->dev); > > + pm_runtime_resume(wirq->dev); > > Calling this with disabled interrupts is a bad idea in general. > Is the device guaranteed to have power.irq_safe set? Well right now it's using threaded irq, and I'd like to get rid of the pm_runtime calls in the regular driver interrupts completely. We need to ensure the device runtime_resume is completed before returning IRQ_HANDLED here. > I guess what you want to call here is pm_request_resume() and > I wouldn't say that calling pm_runtime_mark_last_busy() on a > suspended device was valid. I'll verify again, but I believe the issue was that without doing mark_last_busy here the device falls back asleep right away. That probably should be fixed in pm_runtime in general if that's the case. Considering the above, should we add a new function something like pm_resume_complete() that does not need irq_safe set but does not return until the device has completed resume? I think that would be pretty much probably just pm_request_resume + pm_runtime_barrier. > > +/** > > + * dev_pm_wakeirq_arm_for_suspend - Configure device wake-up > > + * @wirq: Device wake-up interrupt > > + * > > + * Called from the bus code or the device driver for > > + * device suspend(). Just sets up the wake-up event > > + * conditionally based on the device_may_wake(). The > > + * rest is handled automatically by the generic suspend() > > + * code and runtime_suspend(). > > + */ > > +void dev_pm_wakeirq_arm_for_suspend(struct wakeirq_source *wirq) > > +{ > > + if (is_invalid_wakeirq(wirq)) > > + return; > > + > > + irq_set_irq_wake(wirq->wakeirq, > > + device_may_wakeup(wirq->dev)); > > You want to do > > if (device_may_wakeup(wirq->dev)) > enable_irq_wake(wirq->wakeirq); > > here or strange things may happen if two devices share a wakeup IRQ. OK sure. > Also instead of doing it this way, I'd prefer to hook system wakeup > interrupts into the wakeup source objects pointed to by the power.wakeup > fields in struct device. > > Then we could just walk the list of wakeup sources and do enable_irq_wake() > automatically for the wakeup interrupts hooked up to them at the > suspend_device_irqs() time without the need to do anything from drivers > at suspend time. OK that's a good idea. Then we can drop dev_pm_wakeirq_arm_for_suspend() and make that part automatic. Then for runtime_pm, we could make the toggling of the wakeirq handling automatic too. Or do you see a problem with that? > > +struct wakeirq_source { > > + struct device *dev; > > + int wakeirq; > > + bool initialized; > > + bool enabled; > > + irq_handler_t handler; > > + void *data; > > +}; > > Well, so now we have struct wakeup_source already and here we get struct > wakeirq_source > and they mean different things ... Well I was trying to keep it out of the way for most drivers not needing to use wakeirqs. I'll take a look at making it a pointer in the struct wakeup_source. Regards, Tony -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/