Re: [PATCH v3 1/3] dt-bindings: pwm: amlogic: Add core clock for S4 like

From
Chuan Liu <hidden>
Date
2026-09-30 07:13:45
Also in
linux-amlogic, linux-devicetree, linux-pwm, lkml
Reply to
Conor Dooley, 2026-09-28
Hi Ronald, Cc Conor,

On 9/29/2026 12:46 AM, Conor Dooley wrote:
On Sun, Sep 27, 2026 at 04:28:15PM +0200, Ronald Claveau via B4 Relay wrote:
quoted
From: Ronald Claveau <redacted>

The T7, A1, C3, and S4 PWM controllers have a system clock in
addition to the two input clocks of channels A and B.

The channel clocks are acquired by index in the pwm driver,
where the core clock is fetched by name, with optional get
which is silent when not provided.

Require clock-names and minimum 3 items for S4, and devices
which fallback to S4, where the PWM devices do not work
without all 3 clocks.

Add an example showing T7 usage with its fallback to
amlogic,meson-s4-pwm.

Signed-off-by: Ronald Claveau <redacted>
---
  .../devicetree/bindings/pwm/pwm-amlogic.yaml        | 21 ++++++++++++++++++---
  1 file changed, 18 insertions(+), 3 deletions(-)
diff --git a/Documentation/devicetree/bindings/pwm/pwm-amlogic.yaml b/Documentation/devicetree/bindings/pwm/pwm-amlogic.yaml
index c337d85da40f1..e9cd5f1fdf264 100644
--- a/Documentation/devicetree/bindings/pwm/pwm-amlogic.yaml
+++ b/Documentation/devicetree/bindings/pwm/pwm-amlogic.yaml
@@ -62,7 +62,7 @@ properties:
  
    clock-names:
      minItems: 1
-    maxItems: 2
+    maxItems: 3
  
    power-domains:
      maxItems: 1
@@ -143,9 +143,15 @@ allOf:
            items:
              - description: input clock of PWM channel A
              - description: input clock of PWM channel B
-        clock-names: false
+            - description: gate clock for the PWM controller
This is the APB bus clock gate for the PWM controller. If this clock is disabled, the CPU can no longer access the PWM controller's registers.

This clock serves the same purpose as the "bus" clock in the thermal sensor patch you submitted previously [1].

By the way, in the Amlogic clock driver, many sys_clk/clk81 clocks are used to provide the clock for the APB bus of individual controllers. Some of them also provide the functional clock for the controller, such as I2C.

Therefore, would it be reasonable to keep the description here consistent with the description in the thermal dt-binding mentioned above?

[1] https://lore.kernel.org/all/20260923-fix-t7-thermal-clock-v2-1-bff7a09eb487@aliel.fr/ (local)
quoted
+        clock-names:
+          items:
+            - const: channel-a
+            - const: channel-b
+            - const: core
Similarly, would it be more appropriate to name this clock "bus" as well?
quoted
        required:
          - clocks
+        - clock-names
  
    - if:
        properties:
@@ -179,6 +185,15 @@ examples:
      pwm@1000 {
        compatible = "amlogic,meson-s4-pwm";
        reg = <0x1000 0x10>;
-      clocks = <&pwm_src_a>, <&pwm_src_b>;
+      clocks = <&pwm_src_a>, <&pwm_src_b>, <&pwm_src_core>;
+      clock-names = "channel-a", "channel-b", "core";
+      #pwm-cells = <3>;
+    };
+  - |
+    pwm@60000 {
If you respin, drop the second example please, since it doesn't actually
do anything different anymore.

Acked-by: Conor Dooley <conor.dooley@microchip.com>
pw-bot: not-applicable
quoted
+      compatible = "amlogic,t7-pwm", "amlogic,meson-s4-pwm";
+      reg = <0x60000 0x24>;
+      clocks = <&channel_a>, <&channel_b>, <&core>;
+      clock-names = "channel-a", "channel-b", "core";
        #pwm-cells = <3>;
      };

-- 
2.49.0
-- 
Best regards,
Chuan

Thread (7 messages) 7 messages, 3 authors, last reply 2026-10-01
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help