Re: [PATCH v4 2/3] dt-bindings: rng: add bindings for MediaTek ARMv8 SoCs

[Neal Liu]

On Wed, 2019-10-02 at 21:47 +0800, Neal Liu wrote:
> Dear Rob,
> 
> Just a gentle ping.
> 
> Neal
> 
> On Wed, 2019-09-18 at 09:44 +0800, Neal Liu wrote:
> > On Tue, 2019-07-23 at 22:35 +0800, Rob Herring wrote:
> > > On Mon, Jul 22, 2019 at 8:21 PM Neal Liu <neal.liu@xxxxxxxxxxxx> wrote:
> > > >
> > > 
> > > Please don't top post to lists.
> > > 
> > > > Dear Rob,
> > > >         You can check my driver for detail:
> > > >         http://patchwork.kernel.org/patch/11012475/ or patchset 3/3
> > > 
> > > I could, or you could just answer my question.
> > > 
> > > >
> > > >         This driver is registered as hardware random number generator, and
> > > > combines with rng-core.
> > > >         We want to add one rng hw based on the dts. Is this proper or do you
> > > > have other suggestion to meet this requirement?
> > > 
> > > It depends. There doesn't appear to be any resource configuration, so
> > > why does it need to be in DT. DT is not the only way instantiate
> > > drivers.
> > > 
> > > Rob
> > > 
> > 
> > We would like to consult more about this patch.
> > We cannot figure out what method should be used instead of DT.
> > The interface to access firmware is "smc" and firmware function only
> > exists on certain platforms.
> > Some DT has similar way, like:
> > http://github.com/torvalds/linux/blob/master/arch/arm64/boot/dts/hisilicon/hi6220-hikey.dts#L470
> > 
> > 	firmware {
> > 		optee {
> > 			compatible = "linaro,optee-tz";
> > 			method = "smc";
> > 		};
> > 	};
> > 
> > Is there any way to instantiate driver on certain platforms without DT?
> > Could you give us some examples?
> > Thanks
> > 
> > > >
> > > >         Thanks
> > > >
> > > >
> > > > On Tue, 2019-07-23 at 01:13 +0800, Rob Herring wrote:
> > > > > On Mon, Jun 24, 2019 at 03:24:11PM +0800, Neal Liu wrote:
> > > > > > Document the binding used by the MediaTek ARMv8 SoCs random
> > > > > > number generator with TrustZone enabled.
> > > > > >
> > > > > > Signed-off-by: Neal Liu <neal.liu@xxxxxxxxxxxx>
> > > > > > ---
> > > > > >  .../devicetree/bindings/rng/mtk-sec-rng.txt        |   10 ++++++++++
> > > > > >  1 file changed, 10 insertions(+)
> > > > > >  create mode 100644 Documentation/devicetree/bindings/rng/mtk-sec-rng.txt
> > > > > >
> > > > > > diff --git a/Documentation/devicetree/bindings/rng/mtk-sec-rng.txt b/Documentation/devicetree/bindings/rng/mtk-sec-rng.txt
> > > > > > new file mode 100644
> > > > > > index 0000000..c04ce15
> > > > > > --- /dev/null
> > > > > > +++ b/Documentation/devicetree/bindings/rng/mtk-sec-rng.txt
> > > > > > @@ -0,0 +1,10 @@
> > > > > > +MediaTek random number generator with TrustZone enabled
> > > > > > +
> > > > > > +Required properties:
> > > > > > +- compatible : Should be "mediatek,mtk-sec-rng"
> > > > >
> > > > > What's the interface to access this?
> > > > >
> > > > > A node with a 'compatible' and nothing else is a sign of something that
> > > > > a parent device should instantiate and doesn't need to be in DT. IOW,
> > > > > what do complete bindings for firmware functions look like?
> > > > >

We would like to revise our DT node as below:

firmware {
	hwrng {
		compatible = "mediatek,mtk-sec-rng";
		method = "smc";
	};
};

And dt-bindings path would be changed because it's base on ARM TrustZone
Firmware.
>From "Documentation/devicetree/bindings/rng/mtk-sec-rng.txt"
To
"Documentation/devicetree/bindings/arm/firmware/mediatek,mtk-sec-rng.txt"

We found some similar examples which also provide an interface to trap
to Secure State through SMC instruction.
Example 1: Documentation/devicetree/bindings/arm/firmware/xxx.txt
Example 2: Documentation/devicetree/bindings/arm/psci.txt

Is that okay for you?

Neal

> > > > > > +
> > > > > > +Example:
> > > > > > +
> > > > > > +hwrng: hwrng {
> > > > > > +   compatible = "mediatek,mtk-sec-rng";
> > > > > > +}
> > > > > > --
> > > > > > 1.7.9.5
> > > > > >
> > > > >
> > > > > _______________________________________________
> > > > > Linux-mediatek mailing list
> > > > > Linux-mediatek@xxxxxxxxxxxxxxxxxxx
> > > > > http://lists.infradead.org/mailman/listinfo/linux-mediatek
> > > >
> > > >
> > 
>