Re: [PATCH v4] gpio: lib-sysfs: Add 'wakeup' attribute

[Johan Hovold <johan@kernel.org>]

On Fri, Jan 16, 2015 at 08:49:17AM -0800, Sören Brinkmann wrote:
> On Fri, 2015-01-16 at 12:11PM +0100, Johan Hovold wrote:
> > On Thu, Jan 15, 2015 at 11:49:49AM -0800, Soren Brinkmann wrote:
> > > Add an attribute 'wakeup' to the GPIO sysfs interface which allows
> > > marking/unmarking a GPIO as wake IRQ.
> > > The file 'wakeup' is created in each exported GPIOs directory, if an IRQ
> > > is associated with that GPIO and the irqchip implements set_wake().
> > > Writing 'enabled' to that file will enable wake for that GPIO, while
> > > writing 'disabled' will disable wake.
> > > Reading that file will return either 'disabled' or 'enabled' depening on
> > > the currently set flag for the GPIO's IRQ.
> > > 
> > > Signed-off-by: Soren Brinkmann <soren.brinkmann@xilinx.com>
> > > Reviewed-by: Alexandre Courbot <acourbot@nvidia.com>
> > > ---
> > > Hi Linus, Johan,
> > > 
> > > I rebased my patch. And things look good.
> > 
> > I took at closer look at this patch now and I really don't think it
> > should be merged at all.
> > 
> > We have a mechanism for handling wake-up sources (documented in
> > Documentation/power/devices.txt) as well as an ABI to enable/disable
> > them using the power/wakeup device attribute from userspace.
> 
> Doesn't work for GPIOs AFAIK.

Not today no, that's why I said it would take some work.

> > Implementing proper wakeup support for unclaimed GPIOs would take some
> > work (if at all desired), but that is not a reason to be adding custom
> > implementations that violates the kernel's power policies and new ABIs
> > that would need to be maintained forever.
> 
> These are claimed, by the sysfs interface.

Unclaimed by a proper device and driver in the driver model.

> > [ And we really shouldn't be adding anything to the broken gpio sysfs
> > interface until it's been redesigned. ]
> > 
> > Meanwhile you can (should) use gpio-keys if you need to wake your system
> > on gpio events.
> 
> We had that discussion and I don't think GPIO keys is the right solution
> for every use-case.

I can see that, but this still needs to be implemented properly and not
just as a quick hack on top of the already fragile gpio sysfs-interface.

Since pretty much everyone agrees that the current interface needs to be
replaced, we really shouldn't be adding more stuff to the broken
interface before that happens.

> > > But the 'is_visible' things does not behave the way I expected it to.
> > > It seems to be only triggered on an export but not when attributes
> > > change. Hence, in my case, everything was visiible since the inital
> > > state matches that, but even when changing the direction or things
> > > like that, attributes don't disappear. Is that something still worked
> > > on? Expected
> > 
> > That's expected. We generally don't want attributes to appear or
> > disappear after the device has been registered (although there is a
> > mechanism for cases were it makes sense). This is no different from
> > how your v3 patch worked either.
> 
> Sure, but the is_visible thing is effectively broken for GPIO. I think a
> GPIO is in a defined state when exported and the checks all work on that
> state during export. But then this state can be changed through the
> sysfs interface. So, if the initial state hides something it becomes
> unavailable for all times and, vice versa, if the initial state makes
> something visible, it will stay even when it is no longer a valid
> property to change.

Again, this is exactly how the interface has always worked, and that's
exactly how your v3, which added the attributes manually, also worked.

The group-visibility mechanism is not broken. What's broken is interface
designs based on attributes magically disappearing and reappearing after
the device has been created.

Johan