Messages in this thread | | | Date | Sun, 11 Sep 2011 16:44:31 -0400 (EDT) | From | Alan Stern <> | Subject | Re: [PATCH 2/2] PM/runtime: handle ->runtime_suspend failure correctly |
| |
On Sat, 10 Sep 2011 tom.leiming@gmail.com wrote:
> From: Ming Lei <tom.leiming@gmail.com> > > If ->runtime_suspend returns -EAGAIN or -EBUSY, the device should > still be in ACTIVE state, so it is not needed to handle defered > resume and idle notification to its parent; if ->runtime_suspend > returns other fatal failure, it doesn't make sense to process defered > resume and send idle notification to its parent.
> @@ -422,6 +425,9 @@ static int rpm_suspend(struct device *dev, int rpmflags) > } > wake_up_all(&dev->power.wait_queue); > > + if (retval) > + goto out; > + > if (dev->power.deferred_resume) { > rpm_resume(dev, 0); > retval = -EAGAIN;
If there's a suspend failure, the deferred_resume flag gets turned off anyway. But skipping this test won't hurt, and skipping the parent notification is a good idea.
In fact, it might be even better to put a copy of the wake_up_all() in the "if (retval)" branch after the suspend callback and then go directly to out. The "else" branch could then become part of the straight-through code, not indented.
Alan Stern
| |