On 21.07.2015 23:41, Mark Brown wrote: > On Tue, Jul 21, 2015 at 04:35:24PM +0200, Uwe Kleine-König wrote: >> On Tue, Jul 21, 2015 at 10:09:32PM +0900, Krzysztof Kozlowski wrote: > >>> The function looks empty so it can be removed entirely. > >> I assumed that not having a remove function makes the device not >> detachable. Not sure about that. > > No, of course not - the remove function is completely optional. > >> Looking at the code I found that not having a remove function can yield >> surprises, though. If your driver has a probe but no remove function the >> platform bus glue calls > >> dev_pm_domain_attach(_dev, true); > >> at probe time, but not > >> dev_pm_domain_detach(_dev, true); > >> at remove. I admit I don't know about that dev_pm_domain stuff, but it >> looks wrong to only have one but not the other. Greg? > > That looks like a bug, yes. Cc: linux-pm, Kevin, Rafael, Ulf I agree, device should be detached from domain regardless of presence of remove callback. Documentation (like Documentation/driver-model/driver.txt) does not mention that remove callback is necessary for unbinding devices. There is no sense in storing empty removal callbacks. Best regards, Krzysztof -- To unsubscribe from this list: send the line "unsubscribe linux-next" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
