Il 24/10/25 18:29, Conor Dooley ha scritto:

On Fri, Oct 24, 2025 at 10:32:13AM +0200, AngeloGioacchino Del Regno wrote:

quoted

Add bindings for the regulators found in the MediaTek MT6316 PMIC,
usually found in board designs using the MT6991 Dimensity 9400 and
on MT8196 Kompanio SoC for Chromebooks.

This chip is fully controlled by SPMI and has multiple variants
providing different phase configurations.

Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
---
  .../regulator/mediatek,mt6316b-regulator.yaml | 78 +++++++++++++++++++
  .../regulator/mediatek,mt6316c-regulator.yaml | 78 +++++++++++++++++++
  .../regulator/mediatek,mt6316d-regulator.yaml | 77 ++++++++++++++++++
  3 files changed, 233 insertions(+)
  create mode 100644 Documentation/devicetree/bindings/regulator/mediatek,mt6316b-regulator.yaml
  create mode 100644 Documentation/devicetree/bindings/regulator/mediatek,mt6316c-regulator.yaml
  create mode 100644 Documentation/devicetree/bindings/regulator/mediatek,mt6316d-regulator.yaml

diff --git a/Documentation/devicetree/bindings/regulator/mediatek,mt6316b-regulator.yaml b/Documentation/devicetree/bindings/regulator/mediatek,mt6316b-regulator.yaml
new file mode 100644
index 000000000000..65b70dd90728
--- /dev/null
+++ b/Documentation/devicetree/bindings/regulator/mediatek,mt6316b-regulator.yaml

@@ -0,0 +1,78 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/regulator/mediatek,mt6316b-regulator.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: MediaTek MT6316 BP/VP SPMI PMIC Regulators
+
+maintainers:
+  - AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
+
+description:
+  The MediaTek MT6316BP/VP PMICs are fully controlled by SPMI interface, both
+  feature four step-down DC/DC (buck) converters, and provides 2+2 Phases,
+  joining Buck 1+2 for the first phase, and Buck 3+4 for the second phase.
+
+properties:
+  compatible:
+    const: mediatek,mt6316b-regulator
+
+  reg:
+    maxItems: 1
+
+patternProperties:
+  "^vbuck(12|34)$":
+    type: object
+    $ref: regulator.yaml#
+    unevaluatedProperties: false
+    properties:
+      regulator-allowed-modes:
+        description: |
+          Allowed Buck regulator operating modes allowed. Valid values below.
+            0 - Normal mode with automatic power saving, reducing the switching
+                frequency when light load conditions are detected
+            1 - Forced Continuous Conduction mode (FCCM) for improved voltage
+                regulation accuracy with constant switching frequency but lower
+                regulator efficiency
+            2 - Forced Low Power mode for improved regulator efficiency, used
+                when no heavy load is expected, will shut down unnecessary IP
+                blocks and secondary phases to reduce quiescent current.
+                This mode does not limit the maximum output current but unless
+                only a light load is applied, there will be regulation accuracy
+                and efficiency losses.
+        minItems: 1
+        maxItems: 3
+        items:
+          enum: [ 0, 1, 2 ]

This property has no default, and the property is not required. Is one
of these modes the default, or is there another mode beyond what's here
that is used if the property is absent? Or are all modes allowed with no
property?

The default is what the bootloader sets before jumping to the kernel... this may
vary from one to the other.

Even though "in theory" the default should be 0, I can't guarantee that this will
really be the default for when Linux boots.... and "resetting" is not possible
(either as a real reset or forcing a default) because those are CPU regulators
and if anything goes wrong the CPUs may freeze execution.

It should be kinda safe to force the default to 0 at boot, but I'm not sure that
I should really take that assumption for granted - I prefer leaving this untouched
and without any "forced" default unless DT specifies some.

So, well, that's why there's no default, and that's why this is not a required
property at the end of the day.

quoted

+
+required:
+  - compatible
+  - reg
+  - '#address-cells'

Why is address-cells required here? Your bucks don't have addresses.

Ugh. I forgot it there. No, that's not intentional.

The #address-cells makes no sense here.

If it is actually required, Rob's bot has pointed out that the property
isn't defined for the device anyway.

pw-bot: changes-requested

quoted

+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/spmi/spmi.h>
+
+    spmi {
+      #address-cells = <2>;
+      #size-cells = <0>;
+
+      pmic@8 {
+        compatible = "mediatek,mt6316b-regulator";
+        reg = <0x8 SPMI_USID>;
+        #address-cells = <0>;
+
+        vbuck12 {
+          regulator-name = "dvdd_core";
+          regulator-min-microvolt = <450000>;
+          regulator-max-microvolt = <965000>;
+          regulator-allowed-modes = <0 1 2>;
+          regulator-enable-ramp-delay = <256>;
+        };
+      };
+    };
+...