Re: [PATCHv2 5/5] arm64: allwinner: a64: Add support for TERES-I laptop

From: Harald Geyer <harald@ccbib.org>
To: Icenowy Zheng <icenowy@aosc.io>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>,
	Chen-Yu Tsai <wens@csie.org>, Kalle Valo <kvalo@codeaurora.org>,
	Rob Herring <robh+dt@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will.deacon@arm.com>,
	linux-arm-kernel@lists.infradead.org, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Andre Przywara <andre.przywara@arm.com>,
	info@olimex.com, linux-wireless@vger.kernel.org
Subject: Re: [PATCHv2 5/5] arm64: allwinner: a64: Add support for TERES-I laptop
Date: Sun, 24 Jun 2018 18:34:40 +0200	[thread overview]
Message-ID: <E1fX7yK-0000OC-HA@stardust.g4.wien.funkfeuer.at> (raw)
In-Reply-To: <a7011fe78d171658742406bf01d50b7e3d7eaf1d.camel@aosc.io>

Icenowy Zheng writes:
> 在 2018-03-15四的 16:25 +0000，Harald Geyer写道：
> > +&mmc1 {
> > +	pinctrl-names = "default";
> > +	pinctrl-0 = <&mmc1_pins>;
> > +	vmmc-supply = <&reg_aldo2>;
> > +	vqmmc-supply = <&reg_dldo4>;
> > +	mmc-pwrseq = <&wifi_pwrseq>;
> > +	bus-width = <4>;
> > +	non-removable;
> > +	status = "okay";
> > +
> > +	rtl8723bs: wifi@1 {
> > +		reg = <1>;
> > +		interrupt-parent = <&r_pio>;
> > +		interrupts = <0 3 IRQ_TYPE_LEVEL_LOW>; /* PL3 */
> > +		interrupt-names = "host-wake";
> > +	};
> 
> I think this node has some problem:

Thanks for the heads up! Admittedly, I simply copied this node from
sun50i-a64-olinuxino.dts and since it worked, didn't look into it in
too much detail.

> - This device node has no binding. The "host-wake" interrupt is part of
>  Broadcom SDIO Wi-Fi binding, rather than a generic one.

I think the general mmc and interrupts bindings apply. And the mmc binding
clearly states that for sub-nodes a compatible string is optional.

However I just realized that the 'interrupt-names' property is not part
of the general interrupts binding, so I guess at least this property should
be removed.

> - Without the interrupt this device node isn't needed, as RTL8723BS has
> MAC eFUSE and doesn't need a MAC in device tree.

Indeed. I wasn't aware of this, but I just tested and the device probes
fine without the subnode present. I think the devicetree is mainly for
information which cannot be probed, so maybe the subnode should just get
removed.

> In order to solve the problems. I suggest either drop this device node
> or make a generic "sdio-wifi" device tree binding. Personally I prefer
> the latter, as it's more accurate device representation.
> 
> If such a device tree binding is added, I think it should contain the
> "host-wake" interrupt and a "local-mac-address" property. Both can be
> ignored by the driver. (This interrupt can be needed if a more card-
> ventor-neutral name is found.)

I don't feel qualified to comment on this. If you want to propose such
a patch and fix above node accordingly, I won't object. Otherwise I'll
just send a patch to remove the subnode.

Thanks,
Harald