On Thu, 13 Jun 2013, Tomasz Figa wrote:

> > +   rc = ohci_suspend(hcd, do_wakeup);
> > +   if (rc == 0 && do_wakeup && HCD_WAKEUP_PENDING(hcd)) {
> > +           ohci_resume(hcd, false);
> > +           rc = -EBUSY;
> > +   }
>
> I'm not into USB host subsystem, so I might just ask a stupid question.
>
> Can't we make ohci_suspend check this for us, so the drivers would just
> check for error code? It seems like in all your patches this part of code
> is duplicated, looking as a good candidate to be generic.

Argh!  You're right, of course.

I didn't see it, because the only existing place where this check is
made is in the PCI glue layer.  Pushing it into the HCDs themselves is
obviously the right thing to do.

Manjanuth, let's do this.  You can write a preliminary patch that puts
this check at the end of the ohci_suspend() routine, and then resubmit
your series.

Alan Stern