Re: [PATCH] arm64: dts: qcom: sc8280xp-huawei-gaokun3: Add dsi panel DT node

[Pengyu Luo]

On Wed, Jul 1, 2026 at 7:45 PM Konrad Dybcio
<konrad.dybcio@xxxxxxxxxxxxxxxx> wrote:
>
> On 6/30/26 12:41 PM, Pengyu Luo wrote:
> > On Tue, Jun 30, 2026 at 6:34 PM Konrad Dybcio
> > <konrad.dybcio@xxxxxxxxxxxxxxxx> wrote:
> >>
> >> On 6/30/26 12:30 PM, Pengyu Luo wrote:
> >>> On Tue, Jun 30, 2026 at 6:29 PM Konrad Dybcio
> >>> <konrad.dybcio@xxxxxxxxxxxxxxxx> wrote:
> >>>>
> >>>> On 6/30/26 11:28 AM, Pengyu Luo wrote:
> >>>>> On Tue, Jun 30, 2026 at 5:18 PM Konrad Dybcio
> >>>>> <konrad.dybcio@xxxxxxxxxxxxxxxx> wrote:
> >>>>>>
> >>>>>> On 6/30/26 6:22 AM, Pengyu Luo wrote:
> >>>>>>> DSI panel driver have been added for a while, so add the DT node to
> >>>>>>> enable it. vdd{a,s}-supply for dsi, dsi_phy are blank since the DSDT
> >>>>>>> describes the wrong, no impact after manual disabling these wrong
> >>>>>>> supplies.
> >>>>>>>
> >>>>>>> Signed-off-by: Pengyu Luo <mitltlatltl@xxxxxxxxx>
> >>>>>>> ---
> >>>>>>
> >>>>>> [...]
> >>>>>>
> >>>>>>> +&mdss0_dsi0 {
> >>>>>>> +     /* real vdda is unknown */
> >>>>>>> +
> >>>>>>
> >>>>>> L3B: DSI core 1.2 V (both?)
> >>>>>> L6B: DSI0/1 PLL and core 0.9 V
> >>>>>>
> >>>>>
> >>>>> Yes, that is what I saw in the DSDT, however, I can disable them
> >>>>> safely (hack the regulator driver to add a sysfs to handle these
> >>>>> regulators), the panel works well.
> >>>>
> >>>> L3B additionally powers CSIPHY2/3, EDP0/1, REFGEN (probably why it
> >>>> ends up being on), UFSPHY and USB1_PHY
> >>>>
> >>>> L6B additionally powers the same components, minus refgen and USB
> >>>>
> >>>
> >>> No, I confirmed they are totally disabled, since there is a way to check it now.
> >>> https://lore.kernel.org/linux-arm-msm/20260420-read-rpmh-v3-v4-0-70c152e6c958@xxxxxxxxxxxxxxxx/
> >>
> >> This presents you with the status of the votes casted by HLOS, not
> >> the actual hardware state
> >>
> >
> > If so, non-HLOS will vote on some critical regulators to keep them on,
> > IIRC, once I disabled L9D, system reseted immediately.
>
> NHLOS will generally ensure that the hardware that it needs is
> powered. This does not extend to hardware that *you* may want to
> access in Linux (or which was left online by UEFI), which I suspect
> was the case here
>
> >> To get the latter, you'd have to dump the SPMI registers
> >>
> > I will appreciate it if you provide more information.
>
> Take a look at drivers/regulator/qcom_spmi-regulator.c, you can get
> the actual live value of the enable register and the voltage selector
> for each regulator
>

Hi, Konrad. The real case seems to be as you said. After disabling l3b
and l6b via rpmh,

# cat /sys/kernel/debug/regmap/0-01/registers | grep -E 'c[1-9a]46:'
c146: 80
c246: 00
c346: 80 # l3b en
c446: 80
c546: 00
c646: 80 # l6b en
c746: 00
c846: 80
c946: 00
ca46: 80

Could you please have a check if this node is available for
sc8280xp?(I want to write 0 to 0xc346 and 0x0646 to ensure the ldo
disabling will  turn off the display.) This is copied from sm8350, if
the fuse register is 0x7820a8 for sc8280xp as well, then it is fused
as disabled.

        spmi_debug_bus: qcom,spmi-debug@6b12000 {
                compatible = "qcom,spmi-pmic-arb-debug";
                reg = <0 0x06b12000 0 0x60>, <0 0x7820a8 0 0x4>;
                reg-names = "core", "fuse";
                clocks = <&aoss_qmp>;
                clock-names = "core_clk";
                qcom,fuse-disable-bit = <24>;
                #address-cells = <2>;
                #size-cells = <0>;
                status = "okay";
        }

Best wishes,
Pengyu


> Konrad