[PATCH 0/6] staging: bcm2835-firmware-gpio: Initial staging commit

[Michael Zoran]

On Fri, 2017-03-17 at 18:24 +0100, Stefan Wahren wrote:
> Hi Michael,
> 
> > Michael Zoran <mzoran at crowfest.net> hat am 17. März 2017 um 17:48
> > geschrieben:
> > 
> > 
> > On Fri, 2017-03-17 at 17:31 +0100, Stefan Wahren wrote:
> > > Hi Michael,
> > > 
> > > 
> > > Am 17.03.2017 um 16:22 schrieb Michael Zoran:
> > > > The firmware now has a mailbox API for performing generalized
> > > > gpio
> > > > through
> > > > the firmware.  This driver builds upon a driver written by Dave
> > > > Stevenson
> > > > that was written specifically for the expander on the RPI 3,
> > > > but I
> > > > have
> > > > generalized for generic GPIO through the firmware.
> > > > 
> > > > With this change I was able to test VC4 on a RPI 3 running in
> > > > arm64
> > > > with
> > > > a pure mainline tree from the staging branch.
> > > > 
> > > > The firmware gpio numbers are currently documented at:
> > > > https://github.com/raspberrypi/firmware/blob/master/extra/dt-bl
> > > > ob.d
> > > > ts
> > > > 
> > > > 
> > > 
> > > could you please explain why this driver should go to staging?
> > 
> > I can think of a few reasons:
> > 
> > 1. The mailbox API is still very new and a work in progress. I'm
> > sure
> > it's probably going to change some.
> > 
> > 2. It fits in with the VideoCore subsystem idea in that it doesn't
> > talk
> > directly to hardware. Submitting it to staging would allow us to
> > get
> > VC4 working sooner to unblock proper testing on a RPI 3.
> > 
> > 3. I have some ideas for this driver for future work that I'm still
> > thinking about but I'm not ready yet to submit any changes.
> > 
> 
> before running into the wrong direction i suggest to send at least 1
> RFC series to the GPIO and the DT mailing list.
> 
> If you interested i can send you my comments to this series.
> 
> Stefan

Sure I'm interested in comments, but keep in mind I generalized the
driver really, really fast.  We probably need to change the name of the
driver and the comment block at the top.  I also noticed when the
driver loads it's complaining about no GPIO parent so it certainly
needs work.  I was hoping people would be happy that we have some way
to get some VC4 support on mainline.

I still think this belongs with VideoCore rather then the general
kernel tree.  And for now VideoCore is in staging.