Re: [REGRESSION] pinctrl, of, unable to find hogs

[Tony Lindgren <tony@atomide.com>]

* Gary Bisson <gary.bisson@boundarydevices.com> [170227 08:42]:
> On Mon, Feb 27, 2017 at 05:27:47PM +0100, Gary Bisson wrote:
> > Mika, Tony, All,
> > 
> > On Mon, Feb 27, 2017 at 07:53:53AM -0800, Tony Lindgren wrote:
> > > * Mika Penttilä <mika.penttila@nextfour.com> [170226 21:46]:
> > > > 
> > > > With current linus git (pre 4.11), unable to find the pinctrl hogs :
> > > > 
> > > > 
> > > >  imx6q-pinctrl 20e0000.iomuxc: unable to find group for node hoggrp
> > > > 
> > > > 
> > > > Device is i.MX6 based.
> > > 
> > > Sorry to hear about that, maybe imx_pinctrl_probe_dt() should be
> > > called before devm_pinctrl_register_and_init()?
> > > 
> > > Things got moved around a bit with e566fc11ea76 ("pinctrl: imx: use
> > > generic pinctrl helpers for managing groups") it seems. But maybe that
> > > was done because we did not have commit 950b0d91dc10 ("pinctrl: core:
> > > Fix regression caused by delayed work for hogs") when the imx_pinctrl
> > > changes got merged.
> > 
> > Indeed the i.MX changes were made before your the rework.
> 
> s/the/hog/
> 
> > The reason imx_pinctrl_probe_dt got moved around is because
> > devm_pinctrl_register is the one that initializes the radix trees that
> > are needed when probing the dt.

OK

> > > Gary, are you able to reproduce this? Seems it should happen with
> > > any imx with hogs configured in the dts.
> > 
> > Yes I can reproduce the issue.

OK good to hear.

> > Not sure how to fix it though since we can't move the dt probing before
> > radix tree init.

Yup looks like we still have an issue with pinctrl driver functions
getting called before driver probe has completed.

How about we introduce something like:

int pinctrl_claim_hogs(struct pinctrl_dev *pctldev);

Then the drivers can call that at the end of the probe after
the pins have been parsed?

This should be safe as no other driver can claim the pins either
before the pins have been parsed :)

Regards,

Tony