Re: [linux-sunxi] Re: [PATCH v3 02/12] arm64: allwinner: a64: add NMI controller on A64

From: icenowy@aosc.io
To: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: devicetree@vger.kernel.org, Mark Brown <broonie@kernel.org>,
	linux-sunxi@googlegroups.com, Liam Girdwood <lgirdwood@gmail.com>,
	linux-kernel@vger.kernel.org, Chen-Yu Tsai <wens@csie.org>,
	Rob Herring <robh+dt@kernel.org>,
	Lee Jones <lee.jones@linaro.org>,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [linux-sunxi] Re: [PATCH v3 02/12] arm64: allwinner: a64: add NMI controller on A64
Date: Mon, 24 Apr 2017 18:25:51 +0800	[thread overview]
Message-ID: <2970665c4666a321f14a8d2ff13bc57c@aosc.io> (raw)
In-Reply-To: <20170424071746.u2lrk43kmvvd7m25@lukather>

在 2017-04-24 15:17，Maxime Ripard 写道：
> On Thu, Apr 20, 2017 at 03:03:38PM +0800, icenowy@aosc.io wrote:
>> 在 2017-04-20 13:58，Maxime Ripard 写道：
>> > On Tue, Apr 18, 2017 at 06:56:43PM +0800, Icenowy Zheng wrote:
>> > >
>> > >
>> > > 于 2017年4月18日 GMT+08:00 下午3:00:16, Maxime Ripard
>> > > <maxime.ripard@free-electrons.com> 写到:
>> > > >On Mon, Apr 17, 2017 at 07:57:37PM +0800, Icenowy Zheng wrote:
>> > > >> Allwinner A64 SoC features a NMI controller, which is usually
>> > > >connected
>> > > >> to the AXP PMIC.
>> > > >>
>> > > >> Add support for it.
>> > > >>
>> > > >> Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
>> > > >> Acked-by: Chen-Yu Tsai <wens@csie.org>
>> > > >> ---
>> > > >> Changes in v2:
>> > > >> - Added Chen-Yu's ACK.
>> > > >>
>> > > >>  arch/arm64/boot/dts/allwinner/sun50i-a64.dtsi | 8 ++++++++
>> > > >>  1 file changed, 8 insertions(+)
>> > > >>
>> > > >> diff --git a/arch/arm64/boot/dts/allwinner/sun50i-a64.dtsi
>> > > >b/arch/arm64/boot/dts/allwinner/sun50i-a64.dtsi
>> > > >> index 05ec9fc5e81f..53c18ca372ea 100644
>> > > >> --- a/arch/arm64/boot/dts/allwinner/sun50i-a64.dtsi
>> > > >> +++ b/arch/arm64/boot/dts/allwinner/sun50i-a64.dtsi
>> > > >> @@ -403,6 +403,14 @@
>> > > >>  				     <GIC_SPI 41 IRQ_TYPE_LEVEL_HIGH>;
>> > > >>  		};
>> > > >>
>> > > >> +		nmi_intc: interrupt-controller@01f00c0c {
>> > > >> +			compatible = "allwinner,sun6i-a31-sc-nmi";
>> > > >> +			interrupt-controller;
>> > > >> +			#interrupt-cells = <2>;
>> > > >> +			reg = <0x01f00c0c 0x38>;
>> > > >
>> > > >The base address is not correct, and there's uncertainty on whether
>> > > >this is this particular controller or not. Did you even test this?
>> > >
>> > > Tested by axp20x-pek.
>> >
>> > Still, the base address is wrong, which is yet another hint that this
>> > is not the same interrupt controller, and just works by accident.
>> 
>> No, it's the same as other post-sun6i device trees.
>> See other post-sun6i device trees: (or maybe they're all wrong, but
>> as we have no document for it, we should temporarily keep them)
>> 
>> sun6i-a31.dtsi
>> ```
>> 		nmi_intc: interrupt-controller@01f00c0c {
>> 			compatible = "allwinner,sun6i-a31-sc-nmi";
>> 			interrupt-controller;
>> 			#interrupt-cells = <2>;
>> 			reg = <0x01f00c0c 0x38>;
>> 			interrupts = <GIC_SPI 32 IRQ_TYPE_LEVEL_HIGH>;
>> 		};
>> ```
>> 
>> sun8i-a23-a33.dtsi
>> ```
>> 		nmi_intc: interrupt-controller@01f00c0c {
>> 			compatible = "allwinner,sun6i-a31-sc-nmi";
>> 			interrupt-controller;
>> 			#interrupt-cells = <2>;
>> 			reg = <0x01f00c0c 0x38>;
>> 			interrupts = <GIC_SPI 32 IRQ_TYPE_LEVEL_HIGH>;
>> 		};
>> ```
>> 
>> But according to the BSP device tree, the base address should be
>> 0x01f00c00. Should I send some patch to fix all of them? (but it will
>> break device tree compatibility)
> 
> I'm really not a big fan of "if we see something that is broken, just
> let it rot" to be honest.
> 
> We have no idea how this controller works exactly, just like we have
> no idea if it is exactly the same controller or not.
> 
> The only thing we have today is the memory map, and it tells us that
> it has more registers than what you express here.
> 
> Because of the DT backward compatibility, you have to think of it the
> other way around: what will happen if it turns out we need to setup
> any register outside of that region you described in the DT, in
> something like a year or so?
> 
> We can't, really. While if you have the full memory region from the
> beginning, then you just have to add a single writel in your driver.

So things are now already broken, and we may need to fix also A31 and
A23/33.

How should we do this?

> 
> Maxime
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel