Re: [PATCH] wl12xx: Allow wl12xx interrupts to wake up the host

[Juuso Oikarinen <juuso.oikarinen@nokia.com>]

On Mon, 2011-02-07 at 14:38 +0200, ext Ohad Ben-Cohen wrote:
> On Mon, Feb 7, 2011 at 2:29 PM, Juuso Oikarinen
> <juuso.oikarinen@nokia.com> wrote:
> > On Mon, 2011-02-07 at 14:09 +0200, ext Ohad Ben-Cohen wrote:
> >> On Mon, Feb 7, 2011 at 1:30 PM, <juuso.oikarinen@nokia.com> wrote:
> >> >
> >> > From: Juuso Oikarinen <juuso.oikarinen@nokia.com>
> >> >
> >> > Currently the wl12xx interrupts are not configured to wake up the host, which
> >> > leads to reduced performance.
> >>
> >> Can you elaborate what do you mean ?
> >>
> >> Today the chip is powered off at system suspend, so it won't trigger
> >> any interrupts anyway.
> >>
> >> When additional suspend modes will be introduced (e.g. WoWL) this will
> >> be needed of course.
> >
> > These are not just for suspend, AFAIK, but apply also to power saving
> > states.
> 
> What power states do you mean ? do you mean the cpuidle path ? I don't
> think it's related.
> 
> Can you please describe what was the reduced performance that this
> solved for you ? performance is a hot topic for us now :)

Actually this I don't see any performance impact for better or worse of
this patch on brief testing. I was instructed by some of our other
kernel-dudes that this is actually needed for the cpuidle path too, or
there might be some implications.

I'm currently trying to find out if this is really needed or not.

-Juuso