Re: [PATCH 2/2] drivercore: Fix unregistration path of platform devices

From: Tony Lindgren <tony@atomide.com>
To: Rob Herring <robherring2@gmail.com>
Cc: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>,
	Kevin Hilman <khilman@kernel.org>,
	Grant Likely <grant.likely@linaro.org>,
	Bjorn Helgaas <bhelgaas@google.com>,
	"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Pantelis Antoniou <pantelis.antoniou@konsulko.com>,
	Wolfram Sang <wsa@the-dreams.de>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Tyler Baker <tyler.baker@linaro.org>,
	Olof Johansson <olof@lixom.net>
Subject: Re: [PATCH 2/2] drivercore: Fix unregistration path of platform devices
Date: Tue, 16 Jun 2015 00:58:10 -0700	[thread overview]
Message-ID: <20150616075809.GD2740@atomide.com> (raw)
In-Reply-To: <CAL_Jsq+RyuiYszWSGF-XSL_fqKNWpR0VjFkJMdnW2QuaKNT-AA@mail.gmail.com>

* Rob Herring <robherring2@gmail.com> [150610 07:06]:
> 
> I've looked at some of the failures. Armada 370 looks normal AFAICT,
> but the network device evidently does not probe. i.MX6 has no log, but
> IIRC it is what failed previously on Grant's last attempt. For OMAP4,
> I found the overlapping region here:
> 
>                         omap4_padconf_core: scm@100000 {
>                                 compatible = "ti,omap4-scm-padconf-core",
>                                              "simple-bus";
>                                 #address-cells = <1>;
>                                 #size-cells = <1>;
>                                 ranges = <0 0x100000 0x1000>;
> 
>                                 omap4_pmx_core: pinmux@40 {
>                                         compatible = "ti,omap4-padconf",
>                                                      "pinctrl-single";
>                                         reg = <0x40 0x0196>;
>                                         #address-cells = <1>;
>                                         #size-cells = <0>;
>                                         #interrupt-cells = <1>;
>                                         interrupt-controller;
>                                         pinctrl-single,register-width = <16>;
>                                         pinctrl-single,function-mask = <0x7fff>;
>                                 };
> 
>                                 omap4_padconf_global: omap4_padconf_global@5a0 {
>                                         compatible = "syscon";
>                                         reg = <0x5a0 0x170>;
>                                         #address-cells = <1>;
>                                         #size-cells = <1>;
> 
>                                         pbias_regulator: pbias_regulator {
>                                                 compatible = "ti,pbias-omap";
>                                                 reg = <0x60 0x4>;
> 
> 0x60 is within the pinmux range of 0x40-0x1d6.
> 
> But why is the regulator a sub node here instead of omap4_pmx_core?

I don't think the reg entry is in use here as the pbias_regulator uses
syscon_regmap_lookup_by_phandle via syscon.

>                                                 syscon =
> <&omap4_padconf_global>;
> 
> This seems to indicate that 0x60 is supposed to be an offset from
> 0x5a0. That would require a ranges property in the parent. Is this an
> error?

Yeah we should add ranges to padconf_global so drivers not using syscon
can just do of_ioremap for a dedicated range of registers within the
padconf_global. That area has things like PHYs, regulators and clocks.

Regards,

Tony