On Thu, Feb 02, 2023 at 08:10:01PM +0000, Thinh Nguyen wrote:
> On Thu, Feb 02, 2023, Linyu Yuan wrote:
> >
> > On 2/2/2023 3:05 AM, Thinh Nguyen wrote:
> > > On Wed, Feb 01, 2023, Linyu Yuan wrote:
> > > > Consider there is interrpt sequences as suspend (U3) -> wakeup (U0) ->
> > > interrupt?
> >
> >
> > thanks, will change next version.
> >
> >
> > >
> > > > suspend (U3), as there is no update to link state in wakeup interrupt,
> > > Instead of "no update", can you note in the commit that the link state
> > > change event is not enabled for most devices, so the driver doesn't
> > > update its link_state.
> >
> >
> > thanks, will change next version.
> >
> >
> > >
> > > > the second suspend interrupt will not report to upper layer.
> > > >
> > > > Fix it by update link state in wakeup interrupt handler.
> > > >
> > > > Cc: stable@xxxxxxxxxxxxxxx
> > > Can you add fix tag?
> >
> >
> > seem this change can apply to all current stable kernel.
>
> Did we have handling of suspend/resume since the beginning? If we did,
> please add a fix tag to the commit when the driver first added.
>
> That helps to know that this is a fix patch.
Suspend was added with my change: d1d90dd27254 ("usb: dwc3: gadget: Enable suspend
events"), so arguably the link_state mismatch of $SUBJECT wouldn't have occurred
prior to that if suspend_interrupt() never got called right? ;)
But strictly speaking, wakeup_interrupt() was introduced all the way back
in the first commit 72246da40f37 ("usb: Introduce DesignWare USB3 DRD Driver")
and had not been changed since.
Any preference which tag to choose for Fixes?
Jack
