On Fri, Dec 27, 2019 at 2:51 PM Roger Lu <roger.lu@xxxxxxxxxxxx> wrote:
>
> Dear Rob,
>
> Sorry for the late reply.
>
> On Mon, 2019-09-30 at 08:35 -0500, Rob Herring wrote:
> > On Fri, Sep 06, 2019 at 06:05:13PM +0800, Roger Lu wrote:
> > > Document the binding for enabling mtk svs on MediaTek SoC.
> > >
> > > Signed-off-by: Roger Lu <roger.lu@xxxxxxxxxxxx>
> > > ---
> > > .../devicetree/bindings/power/mtk-svs.txt | 88 +++++++++++++++++++
> > > 1 file changed, 88 insertions(+)
> > > create mode 100644 Documentation/devicetree/bindings/power/mtk-svs.txt
> > >
> > > diff --git a/Documentation/devicetree/bindings/power/mtk-svs.txt b/Documentation/devicetree/bindings/power/mtk-svs.txt
> > > new file mode 100644
> > > index 000000000000..6a71992ef162
> > > --- /dev/null
> > > +++ b/Documentation/devicetree/bindings/power/mtk-svs.txt
> > > @@ -0,0 +1,88 @@
> > > +* Mediatek Smart Voltage Scaling (MTK SVS)
> > > +
> > > +This describes the device tree binding for the MTK SVS controller (bank)
> > > +which helps provide the optimized CPU/GPU/CCI voltages. This device also
> > > +needs thermal data to calculate thermal slope for accurately compensate
> > > +the voltages when temperature change.
> > > +
> > > +Required properties:
> > > +- compatible:
> > > + - "mediatek,mt8183-svs" : For MT8183 family of SoCs
> > > +- reg: Address range of the MTK SVS controller.
> > > +- interrupts: IRQ for the MTK SVS controller.
> > > +- clocks, clock-names: Clocks needed for the svs controller. required
> > > + clocks are:
> > > + "main_clk": Main clock needed for register access
> >
> > '_clk' is redundant.
>
> Oh Okay. I'll remove _clk. Thanks.
>
> >
> > > +- nvmem-cells: Phandle to the calibration data provided by a nvmem device.
> > > +- nvmem-cell-names: Should be "svs-calibration-data" and "calibration-data"
> > > +
> > > +Subnodes:
> > > +- svs_cpu_little: SVS bank device node of little CPU
> > > + compatible: "mediatek,mt8183-svs-cpu-little"
> > > + operating-points-v2: OPP table hooked by SVS little CPU bank.
> > > + SVS will optimze this OPP table voltage part.
> > > + vcpu-little-supply: PMIC buck of little CPU
> > > +- svs_cpu_big: SVS bank device node of big CPU
> > > + compatible: "mediatek,mt8183-svs-cpu-big"
> > > + operating-points-v2: OPP table hooked by SVS big CPU bank.
> > > + SVS will optimze this OPP table voltage part.
> > > + vcpu-big-supply: PMIC buck of big CPU
> > > +- svs_cci: SVS bank device node of CCI
> > > + compatible: "mediatek,mt8183-svs-cci"
> > > + operating-points-v2: OPP table hooked by SVS CCI bank.
> > > + SVS will optimze this OPP table voltage part.
> > > + vcci-supply: PMIC buck of CCI
> > > +- svs_gpu: SVS bank device node of GPU
> > > + compatible: "mediatek,mt8183-svs-gpu"
> > > + operating-points-v2: OPP table hooked by SVS GPU bank.
> > > + SVS will optimze this OPP table voltage part.
> > > + vgpu-spply: PMIC buck of GPU
> > > +
> > > +Example:
> > > +
> > > + svs: svs@1100b000 {
> > > + compatible = "mediatek,mt8183-svs";
> > > + reg = <0 0x1100b000 0 0x1000>;
> > > + interrupts = <GIC_SPI 127 IRQ_TYPE_LEVEL_LOW 0>;
> >
> > GIC interrupts are 3 cells, you have 4.
>
> Oops, I'll remove the fourth parameter. Thanks a lot.
>
> >
> > > + clocks = <&infracfg CLK_INFRA_THERM>;
> > > + clock-names = "main_clk";
> > > + nvmem-cells = <&svs_calibration>, <&thermal_calibration>;
> > > + nvmem-cell-names = "svs-calibration-data", "calibration-data";
> > > +
> > > + svs_cpu_little: svs_cpu_little {
> >
> > Don't use '_' in node names.
>
> Okay. I'll replace it with '-'. Thanks.
>
> >
> > > + compatible = "mediatek,mt8183-svs-cpu-little";
> > > + operating-points-v2 = <&cluster0_opp>;
> > > + };
> > > +
> > > + svs_cpu_big: svs_cpu_big {
> > > + compatible = "mediatek,mt8183-svs-cpu-big";
> > > + operating-points-v2 = <&cluster1_opp>;
> > > + };
> > > +
> > > + svs_cci: svs_cci {
> > > + compatible = "mediatek,mt8183-svs-cci";
> > > + operating-points-v2 = <&cci_opp>;
> > > + };
> > > +
> > > + svs_gpu: svs_gpu {
> > > + compatible = "mediatek,mt8183-svs-gpu";
> > > + power-domains = <&scpsys MT8183_POWER_DOMAIN_MFG_2D>;
> > > + operating-points-v2 = <&gpu_opp_table>;
> > > + };
> > > + };
> > > +
> > > + &svs_cpu_little {
> > > + vcpu-little-supply = <&mt6358_vproc12_reg>;
> >
> > It's already defined to have OPP and supply in the cpu nodes. Parse them
> > to get this information rather than duplicating it here.
> >
> > The same should apply to the CCI and GPU.
>
> Please let me explain the reason why I add SVS sub-nodes. I ever try to
> parse other nodes to get desired power-domains/OPP table. However, it
> makes SVS driver harder to develop and maintain.
>
> 1. When a SVS-controller-init wants GPU_CORE0's OPP table in one node
> but it needs power-domains(GPU_MFG_2D) in another node, it becomes
> complicated and confusing when SVS sub-node tries to parse many nodes.
> Therefore, we want SVS sub-node to focus on what SVS bank requires by
> how we do in this patch.
>
> 2. In hardware point of view, SVS controller depends on other hardware's
> power only. All the SVS controller registers are in SVS hardware. So, we
> think It's good that SVS sub-node describes what SVS controller requires
> instead of linking other subsys nodes and parse the property that SVS
> controller needs.
>
> 3. We want SVS driver to have a generic way to attain subsys device for
> using "pm_runtime and OPP framework" API. If SVS driver tries to parse
> CPU(little/big core) and other subsys device node(e.g cci/gpu), it means
> SVS driver has to maintain different methodologies(cpu-specific?
> devfreq? others?) in order to get CPU(little/big core) and other subsys
> device(e.g cci/gpu) for using "pm_runtime and OPP framework" API.
(Didn't see this more complete reply before replying to v6, I can see
your argument, but if we wanted to push further to have the sub-device
node in the DT)
>From what I see, the SVS driver for node x (cpu/cci/gpu) requires only 3 things:
- An OPP table => that should always be "operating-points-v2"
property of the node x.
- A power domain => that should always be power-domains property of node x.
- A regulator. That one is a bit tricky as the cpu/cci uses "proc",
but gpu uses "mali" (at least on 8183). But maybe you can add a
child-regulator-name property or something to the DT so that the SVS
driver can find the correct regulator?
Seems like the solution could be quite generic?
> >
> > Rob
>
> Sincerely,
> Roger Lu.
