[PATCH v2 0/2] i2c: ls2x: Add clock- related properties and parsing

[PATCH v2 0/2] i2c: ls2x: Add clock- related properties and parsing

[Hongliang Wang]

From: wanghongliang <wanghongliang@loongson.cn>

Hi all,

On 3A/7A/2K platform, the input clock of i2c controller and the divisor of
input clock maybe different, so define the clock-input and clock-div properties
in i2c controller node description of dts or acpi. clock-input describes the
input clock of i2c controller and clock-div describes the clock divisor of the
input clock.

Compared to v1, v2 modified the following content:
1. Add dt-bindings
2. In i2c-ls2x driver, the definition of pclk and div variables is placed on
one line
3. list the different value of clock-input and clock-div on different platforms

Thanks.

wanghongliang (2):
  dt-bindings: i2c: ls2x-i2c: Add clock- related properties
  i2c: ls2x: Add clock-input and clock-div properties parsing

 .../devicetree/bindings/i2c/loongson,ls2x-i2c.yaml  | 13 +++++++++++++
 drivers/i2c/busses/i2c-ls2x.c                       |  7 ++++++-
 2 files changed, 19 insertions(+), 1 deletion(-)

-- 
2.47.2

[PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

From: wanghongliang <wanghongliang@loongson.cn>

Add clock-input and clock-div properties.

clock-input describes the input clock of i2c controller.
clock-div describes the clock divisor of the input clock
of i2c controller.

Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1

Signed-off-by: wanghongliang <wanghongliang@loongson.cn>
---
 .../devicetree/bindings/i2c/loongson,ls2x-i2c.yaml  | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml b/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
index 67882ec6e06a..4659d7012bce 100644
--- a/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
+++ b/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
@@ -24,6 +24,16 @@ properties:
   interrupts:
     maxItems: 1
 
+  clock-input:
+    $ref: /schemas/types.yaml#/definitions/uint32
+    description:
+      Describes the input clock of i2c controller in HZ.
+
+  clock-div:
+    $ref: /schemas/types.yaml#/definitions/uint32
+    description:
+      Describes the divisor of the input clock of i2c controller.
+
 required:
   - compatible
   - reg
@@ -40,6 +50,9 @@ examples:
         reg = <0x1fe21000 0x8>;
         interrupt-parent = <&extioiic>;
         interrupts = <22 IRQ_TYPE_LEVEL_LOW>;
+        clock-frequency = <100000>;
+        clock-input = <100000000>;
+        clock-div = <40>;
         #address-cells = <1>;
         #size-cells = <0>;
 
-- 
2.47.2

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On Wed, Mar 25, 2026 at 09:18:51AM +0800, Hongliang Wang wrote:
> From: wanghongliang <wanghongliang@loongson.cn>
> 
> Add clock-input and clock-div properties.
> 
> clock-input describes the input clock of i2c controller.

No, it does not.


> clock-div describes the clock divisor of the input clock
> of i2c controller.

No, it does not.

Write proper explanations, instead of justyfing your code with
tautology.

> 
> Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1
> 
> Signed-off-by: wanghongliang <wanghongliang@loongson.cn>
> ---
>  .../devicetree/bindings/i2c/loongson,ls2x-i2c.yaml  | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml b/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
> index 67882ec6e06a..4659d7012bce 100644
> --- a/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
> +++ b/Documentation/devicetree/bindings/i2c/loongson,ls2x-i2c.yaml
> @@ -24,6 +24,16 @@ properties:
>    interrupts:
>      maxItems: 1
>  
> +  clock-input:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    description:
> +      Describes the input clock of i2c controller in HZ.

"clocks" property describes the clock input. And it is not in Hz (not
HZ! It is coming from a person's name) but phandle already defined.

> +
> +  clock-div:
> +    $ref: /schemas/types.yaml#/definitions/uint32
> +    description:
> +      Describes the divisor of the input clock of i2c controller.

Use clock framework.

> +
>  required:
>    - compatible
>    - reg
> @@ -40,6 +50,9 @@ examples:
>          reg = <0x1fe21000 0x8>;
>          interrupt-parent = <&extioiic>;
>          interrupts = <22 IRQ_TYPE_LEVEL_LOW>;
> +        clock-frequency = <100000>;
> +        clock-input = <100000000>;
> +        clock-div = <40>;
>          #address-cells = <1>;
>          #size-cells = <0>;
>  
> -- 
> 2.47.2
> 

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On Wed, Mar 25, 2026 at 09:18:51AM +0800, Hongliang Wang wrote:
> From: wanghongliang <wanghongliang@loongson.cn>
> 
> Add clock-input and clock-div properties.
> 
> clock-input describes the input clock of i2c controller.
> clock-div describes the clock divisor of the input clock
> of i2c controller.
> 
> Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1
> 
> Signed-off-by: wanghongliang <wanghongliang@loongson.cn>

Also, why are you using inconsistent identities?

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

Hi Krzysztof,

Consider the clock framework relies on the device tree, and can only be 
used on
Loongson 2K platform with dts parameter mechanism, It cannot be used on 
Loongson
3A+7A platform with the acpi parameter mechanism.
The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A 
platform, parse
the same parameters regardless of dts or acpi parameter passing, So 
clock-input
and clock-div attributes are defined to describe input clock of i2c 
controller and
divisor of input clock. It can be used on both 2K and 3A+7A platform.

Thanks

Hongliang Wang

On 2026/3/25 下午7:43, Krzysztof Kozlowski wrote:
> On Wed, Mar 25, 2026 at 09:18:51AM +0800, Hongliang Wang wrote:
>> From: wanghongliang <wanghongliang@loongson.cn>
>>
>> Add clock-input and clock-div properties.
>>
>> clock-input describes the input clock of i2c controller.
>> clock-div describes the clock divisor of the input clock
>> of i2c controller.
>>
>> Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1
>>
>> Signed-off-by: wanghongliang <wanghongliang@loongson.cn>
> Also, why are you using inconsistent identities?
>
> Best regards,
> Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 26/03/2026 03:12, Hongliang Wang wrote:
> Hi Krzysztof,
> 
> Consider the clock framework relies on the device tree, and can only be 
> used on
> Loongson 2K platform with dts parameter mechanism, It cannot be used on 
> Loongson

Don't top post or request read receipts.

> 3A+7A platform with the acpi parameter mechanism.

And this patch is for ACPI? Then we finish discussion here, because
dt-bindings is not for ACPI.

> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A 
> platform, parse
> the same parameters regardless of dts or acpi parameter passing, So 
> clock-input
> and clock-div attributes are defined to describe input clock of i2c 
> controller and
> divisor of input clock. It can be used on both 2K and 3A+7A platform.

And you cannot use them in DTS.

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 26/03/2026 08:02, Krzysztof Kozlowski wrote:
> On 26/03/2026 03:12, Hongliang Wang wrote:
>> Hi Krzysztof,
>>
>> Consider the clock framework relies on the device tree, and can only be 
>> used on
>> Loongson 2K platform with dts parameter mechanism, It cannot be used on 
>> Loongson
> 
> Don't top post or request read receipts.
> 
>> 3A+7A platform with the acpi parameter mechanism.
> 
> And this patch is for ACPI? Then we finish discussion here, because
> dt-bindings is not for ACPI.
> 
>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A 
>> platform, parse
>> the same parameters regardless of dts or acpi parameter passing, So 
>> clock-input
>> and clock-div attributes are defined to describe input clock of i2c 
>> controller and
>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
> 
> And you cannot use them in DTS.

I need to keep guessing what you want to achieve, because neither your
message nor commit text was explicit - if you need properties for ACPI
and you want to be sure that DTS does not have them, then you could
define them as "foo:false" with a comment why (you always explain WHY
you are doing things). We don't have such convention so far, but I think
it will be useful when Rob finishes the ABI checker.

*Otherwise* minimum would be a comment in the driver that these are not
allowed in DTS.

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

Hi Krzysztof and Andi,

Sorry, I didn't express myself clearly. The purpose of defining clock-input and
clock-div properties is to use them in both acpi and dts. If these properties
cannot be used in dts, can I first use them in acpi? and second, use clock
framework in dts.
if these properties can be used in acpi, use clocks property describes the
clock input may be appropriate.

the [PATCH v2] i2c: ls2x: Add clocks and clock-div properties parsing for ACPI
as follows:

 From 40bdf5588594d2b51f50682fb841c6cbe87c4297 Mon Sep 17 00:00:00 2001
From: wanghongliang <wanghongliang@loongson.cn>
Date: Sun, 12 Apr 2026 03:13:09 +0800
Subject: [PATCH v2] i2c: ls2x: Add clocks and clock-div properties parsing for
  ACPI

On 3A/7A/2K3000 platform, the input clock of i2c controller
and the divisor of input clock maybe different, so define
the clocks and clock-div properties in Device (I2Cx) in acpi
parameter passing. clocks describes the input clock of i2c
controller and clock-div describes the clock divisor of the
input clock.

  Device (I2C1)
  {
    Name (_HID, "LOON0004" /* AT Real-Time Clock */)  // _HID: Hardware ID
    Name (_UID, 0x1)  // _UID: Unique ID
    Name (_DSD, Package (0x02)  // _DSD: Device-Specific Data
    {
      ToUUID ("daffd814-6eba-4d8c-8a91-bc9bbf4aa301")
      Package (0x03)
      {
        Package (0x02)
        {
          "clock-frequency",
          100000
        }
        Package (0x02)
        {
          "clocks",
          100000000
        }
        Package (0x02)
        {
          "clock-div",
          40
        }
      }
    })

The formulas is
Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1

The value of clocks and clock-div on different platforms are
listed below:

1.7A1000/7A2000:
clock-input = 80MHZ
clock-div = 50

2.3A5000/3C5000/3A6000/3C6000:
clock-input = 100MHZ
clock-div = 40

3.2K3000:
clock-input = 100MHZ
clock-div = 55

Signed-off-by: wanghongliang <wanghongliang@loongson.cn>
---
  drivers/i2c/busses/i2c-ls2x.c | 13 ++++++++++++-
  1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/drivers/i2c/busses/i2c-ls2x.c b/drivers/i2c/busses/i2c-ls2x.c
index b475dd27b7af..4d1d461d065a 100644
--- a/drivers/i2c/busses/i2c-ls2x.c
+++ b/drivers/i2c/busses/i2c-ls2x.c
@@ -96,6 +96,7 @@ static irqreturn_t ls2x_i2c_isr(int this_irq, void *dev_id)
  static void ls2x_i2c_adjust_bus_speed(struct ls2x_i2c_priv *priv)
  {
  	u16 val;
+	u32 pclk, div;
  	struct i2c_timings *t = &priv->i2c_t;
  	struct device *dev = priv->adapter.dev.parent;
  	u32 acpi_speed = i2c_acpi_find_bus_speed(dev);
@@ -107,12 +108,22 @@ static void ls2x_i2c_adjust_bus_speed(struct ls2x_i2c_priv *priv)
  	else
  		t->bus_freq_hz = LS2X_I2C_FREQ_STD;
  
+	if (has_acpi_companion(dev)) {
+		if (!device_property_read_u32(dev, "clock-input", &pclk) &&
+		    !device_property_read_u32(dev, "clock-div", &div) &&
+		    div != 0)
+			val = (pclk * 10) / (div * t->bus_freq_hz) - 1;
+		else
+			val = LS2X_I2C_PCLK_FREQ / (5 * t->bus_freq_hz) - 1;
+	} else {
+		val = LS2X_I2C_PCLK_FREQ / (5 * t->bus_freq_hz) - 1;
+	}
+
  	/*
  	 * According to the chip manual, we can only access the registers as bytes,
  	 * otherwise the high bits will be truncated.
  	 * So set the I2C frequency with a sequential writeb() instead of writew().
  	 */
-	val = LS2X_I2C_PCLK_FREQ / (5 * t->bus_freq_hz) - 1;
  	writeb(FIELD_GET(GENMASK(7, 0), val), priv->base + I2C_LS2X_PRER_LO);
  	writeb(FIELD_GET(GENMASK(15, 8), val), priv->base + I2C_LS2X_PRER_HI);
  }
-- 
2.47.2

Best regards,
Hongliang Wang

On 2026/3/26 下午3:49, Krzysztof Kozlowski wrote:
> On 26/03/2026 08:02, Krzysztof Kozlowski wrote:
>> On 26/03/2026 03:12, Hongliang Wang wrote:
>>> Hi Krzysztof,
>>>
>>> Consider the clock framework relies on the device tree, and can only be
>>> used on
>>> Loongson 2K platform with dts parameter mechanism, It cannot be used on
>>> Loongson
>> Don't top post or request read receipts.
>>
>>> 3A+7A platform with the acpi parameter mechanism.
>> And this patch is for ACPI? Then we finish discussion here, because
>> dt-bindings is not for ACPI.
>>
>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>> platform, parse
>>> the same parameters regardless of dts or acpi parameter passing, So
>>> clock-input
>>> and clock-div attributes are defined to describe input clock of i2c
>>> controller and
>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>> And you cannot use them in DTS.
> I need to keep guessing what you want to achieve, because neither your
> message nor commit text was explicit - if you need properties for ACPI
> and you want to be sure that DTS does not have them, then you could
> define them as "foo:false" with a comment why (you always explain WHY
> you are doing things). We don't have such convention so far, but I think
> it will be useful when Rob finishes the ABI checker.
>
> *Otherwise* minimum would be a comment in the driver that these are not
> allowed in DTS.
>
> Best regards,
> Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 26/03/2026 12:48, Hongliang Wang wrote:
> Hi Krzysztof and Andi,
> 
> Sorry, I didn't express myself clearly. The purpose of defining clock-input and
> clock-div properties is to use them in both acpi and dts. If these properties
> cannot be used in dts, can I first use them in acpi? and second, use clock
> framework in dts.

I already asked you:
1. Stop requesting read receipts on your email
2. Don't top post.

I already told you also why these cannot be in DTS. Respond to technical
arguments, not top-post of what you want.

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/3/26 下午8:23, Krzysztof Kozlowski wrote:
> On 26/03/2026 12:48, Hongliang Wang wrote:
>> Hi Krzysztof and Andi,
>>
>> Sorry, I didn't express myself clearly. The purpose of defining clock-input and
>> clock-div properties is to use them in both acpi and dts. If these properties
>> cannot be used in dts, can I first use them in acpi? and second, use clock
>> framework in dts.
> I already asked you:
> 1. Stop requesting read receipts on your email
OK, I have set my email and removed the read receipts.
> 2. Don't top post.
OK
>
> I already told you also why these cannot be in DTS. Respond to technical
> arguments, not top-post of what you want.
OK
>
> Best regards,
> Krzysztof
Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/3/26 下午3:49, Krzysztof Kozlowski wrote:
> On 26/03/2026 08:02, Krzysztof Kozlowski wrote:
>> On 26/03/2026 03:12, Hongliang Wang wrote:
>>> Hi Krzysztof,
>>>
>>> Consider the clock framework relies on the device tree, and can only be
>>> used on
>>> Loongson 2K platform with dts parameter mechanism, It cannot be used on
>>> Loongson
>> Don't top post or request read receipts.
OK
>>> 3A+7A platform with the acpi parameter mechanism.
>> And this patch is for ACPI? Then we finish discussion here, because
>> dt-bindings is not for ACPI.
>>
The initial idea was that this patch could be used for both ACPI and DTS.
>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>> platform, parse
>>> the same parameters regardless of dts or acpi parameter passing, So
>>> clock-input
>>> and clock-div attributes are defined to describe input clock of i2c
>>> controller and
>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>> And you cannot use them in DTS.
OK
> I need to keep guessing what you want to achieve, because neither your
> message nor commit text was explicit
What I want to achieve is to describe the input clock and divisor of I2C 
controller
through parameters passing, and the parameters can be used in both ACPI 
and DTS.
because clock framework cannot be used for ACPI, So I defined two new 
properties.
>   - if you need properties for ACPI
> and you want to be sure that DTS does not have them, then you could
> define them as "foo:false" with a comment why (you always explain WHY
> you are doing things). We don't have such convention so far, but I think
> it will be useful when Rob finishes the ABI checker.
>
> *Otherwise* minimum would be a comment in the driver that these are not
> allowed in DTS.
>
> Best regards,
> Krzysztof

Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 27/03/2026 04:09, Hongliang Wang wrote:
> The initial idea was that this patch could be used for both ACPI and DTS.
>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>> platform, parse
>>>> the same parameters regardless of dts or acpi parameter passing, So
>>>> clock-input
>>>> and clock-div attributes are defined to describe input clock of i2c
>>>> controller and
>>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>>> And you cannot use them in DTS.
> OK
>> I need to keep guessing what you want to achieve, because neither your
>> message nor commit text was explicit
> What I want to achieve is to describe the input clock and divisor of I2C 
> controller

Input clocks are defined as clock inputs obviously in DT, not as
integers. Bindings need to describe the hardware, so start with that.


Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
> On 27/03/2026 04:09, Hongliang Wang wrote:
>> The initial idea was that this patch could be used for both ACPI and DTS.
>>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>>> platform, parse
>>>>> the same parameters regardless of dts or acpi parameter passing, So
>>>>> clock-input
>>>>> and clock-div attributes are defined to describe input clock of i2c
>>>>> controller and
>>>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>>>> And you cannot use them in DTS.
>> OK
>>> I need to keep guessing what you want to achieve, because neither your
>>> message nor commit text was explicit
>> What I want to achieve is to describe the input clock and divisor of I2C
>> controller
> Input clocks are defined as clock inputs obviously in DT, not as
> integers. Bindings need to describe the hardware, so start with that.
I can describe the hardware in loongson,ls2x-i2c.yaml, and I would like to
confirm with you what final implementation plan you agree to? clock 
framework
or custom clock-input an clock-div attributes? if clock framework, how 
can it
also be used for ACPI?
>
> Best regards,
> Krzysztof

Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 30/03/2026 09:18, Hongliang Wang wrote:
> 
> On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
>> On 27/03/2026 04:09, Hongliang Wang wrote:
>>> The initial idea was that this patch could be used for both ACPI and DTS.
>>>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>>>> platform, parse
>>>>>> the same parameters regardless of dts or acpi parameter passing, So
>>>>>> clock-input
>>>>>> and clock-div attributes are defined to describe input clock of i2c
>>>>>> controller and
>>>>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>>>>> And you cannot use them in DTS.
>>> OK
>>>> I need to keep guessing what you want to achieve, because neither your
>>>> message nor commit text was explicit
>>> What I want to achieve is to describe the input clock and divisor of I2C
>>> controller
>> Input clocks are defined as clock inputs obviously in DT, not as
>> integers. Bindings need to describe the hardware, so start with that.
> I can describe the hardware in loongson,ls2x-i2c.yaml, and I would like to
> confirm with you what final implementation plan you agree to? clock 
> framework
> or custom clock-input an clock-div attributes? if clock framework, how 
> can it
> also be used for ACPI?

And you ask DT maintainer for that? It's not relevant. You sent DT
bindings patch, so this patch must be correct and we discuss this patch
here.

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/3/30 下午3:23, Krzysztof Kozlowski wrote:
> On 30/03/2026 09:18, Hongliang Wang wrote:
>> On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
>>> On 27/03/2026 04:09, Hongliang Wang wrote:
>>>> The initial idea was that this patch could be used for both ACPI and DTS.
>>>>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>>>>> platform, parse
>>>>>>> the same parameters regardless of dts or acpi parameter passing, So
>>>>>>> clock-input
>>>>>>> and clock-div attributes are defined to describe input clock of i2c
>>>>>>> controller and
>>>>>>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
>>>>>> And you cannot use them in DTS.
>>>> OK
>>>>> I need to keep guessing what you want to achieve, because neither your
>>>>> message nor commit text was explicit
>>>> What I want to achieve is to describe the input clock and divisor of I2C
>>>> controller
>>> Input clocks are defined as clock inputs obviously in DT, not as
>>> integers. Bindings need to describe the hardware, so start with that.
>> I can describe the hardware in loongson,ls2x-i2c.yaml, and I would like to
>> confirm with you what final implementation plan you agree to? clock
>> framework
>> or custom clock-input an clock-div attributes? if clock framework, how
>> can it
>> also be used for ACPI?
> And you ask DT maintainer for that? It's not relevant. You sent DT
> bindings patch, so this patch must be correct and we discuss this patch
> here.
I don't. My idea is that if the clock input attribute can't be used for both
dts and acpi, then clock framework will be used for dts and new define 
attribute
will be used for acpi. I will first implement the hardware description 
and clock
framework in Bindings.
> Best regards,
> Krzysztof

Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

Hi, Krzysztof

On 2026/3/31 下午3:11, Hongliang Wang wrote:
>
> On 2026/3/30 下午3:23, Krzysztof Kozlowski wrote:
>> On 30/03/2026 09:18, Hongliang Wang wrote:
>>> On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
>>>> On 27/03/2026 04:09, Hongliang Wang wrote:
>>>>> The initial idea was that this patch could be used for both ACPI 
>>>>> and DTS.
>>>>>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>>>>>> platform, parse
>>>>>>>> the same parameters regardless of dts or acpi parameter 
>>>>>>>> passing, So
>>>>>>>> clock-input
>>>>>>>> and clock-div attributes are defined to describe input clock of 
>>>>>>>> i2c
>>>>>>>> controller and
>>>>>>>> divisor of input clock. It can be used on both 2K and 3A+7A 
>>>>>>>> platform.
>>>>>>> And you cannot use them in DTS.
>>>>> OK
>>>>>> I need to keep guessing what you want to achieve, because neither 
>>>>>> your
>>>>>> message nor commit text was explicit
>>>>> What I want to achieve is to describe the input clock and divisor 
>>>>> of I2C
>>>>> controller
>>>> Input clocks are defined as clock inputs obviously in DT, not as
>>>> integers. Bindings need to describe the hardware, so start with that.
>>> I can describe the hardware in loongson,ls2x-i2c.yaml, and I would 
>>> like to
>>> confirm with you what final implementation plan you agree to? clock
>>> framework
>>> or custom clock-input an clock-div attributes? if clock framework, how
>>> can it
>>> also be used for ACPI?
>> And you ask DT maintainer for that? It's not relevant. You sent DT
>> bindings patch, so this patch must be correct and we discuss this patch
>> here.
> I don't. My idea is that if the clock input attribute can't be used 
> for both
> dts and acpi, then clock framework will be used for dts and new define 
> attribute
> will be used for acpi. I will first implement the hardware description 
> and clock
> framework in Bindings.
>> Best regards,
>> Krzysztof
>
> Best regards,
> Hongliang Wang
>

I have a question, the input clock of i2c controller can be described by 
"clocks",
but there is no existing attribute can describe the divisor of the input 
clock,
Can I define a new attribute named "clock-div" to describe it in DT 
bindings?
or do you have any standard solutions for the divisor problem? Thank you.

Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 09/04/2026 14:03, Hongliang Wang wrote:
> 
> I have a question, the input clock of i2c controller can be described by 
> "clocks",
> but there is no existing attribute can describe the divisor of the input 
> clock,
> Can I define a new attribute named "clock-div" to describe it in DT 
> bindings?
> or do you have any standard solutions for the divisor problem? Thank you.
>

You should determine/calculate the divisor in the driver code, depending
on clocks and bus frequencies. You don't need a property for that, usually.

Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

Hi, Krzysztof

On 2026/4/9 下午8:11, Krzysztof Kozlowski wrote:
> On 09/04/2026 14:03, Hongliang Wang wrote:
>> I have a question, the input clock of i2c controller can be described by
>> "clocks",
>> but there is no existing attribute can describe the divisor of the input
>> clock,
>> Can I define a new attribute named "clock-div" to describe it in DT
>> bindings?
>> or do you have any standard solutions for the divisor problem? Thank you.
>>
> You should determine/calculate the divisor in the driver code, depending
> on clocks and bus frequencies. You don't need a property for that, usually.

Not only clocks and bus frequencies, but also a third property is required.

The frequency divison calculation formula of i2c is
Prcescale = clock_a/(clock_div*clock_s)-1

There is three parameters in this formula:
clock_a represents the input clock, which is described by "clocks",
clock_s represents the i2c bus frequency, which is described by 
"clock-frequency",
but there is no existing property to describe clock_div, which has 
different value
on different platform (for example, it is 5 on 7a1000/7a2000, 4 on 
2K1000/2K2000,
5.5 on 2K3000.), So I need a property to describe clock_div in this formula.


> Best regards,
> Krzysztof
Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Krzysztof Kozlowski]

On 10/04/2026 05:04, Hongliang Wang wrote:
> Hi, Krzysztof
> 
> On 2026/4/9 下午8:11, Krzysztof Kozlowski wrote:
>> On 09/04/2026 14:03, Hongliang Wang wrote:
>>> I have a question, the input clock of i2c controller can be described by
>>> "clocks",
>>> but there is no existing attribute can describe the divisor of the input
>>> clock,
>>> Can I define a new attribute named "clock-div" to describe it in DT
>>> bindings?
>>> or do you have any standard solutions for the divisor problem? Thank you.
>>>
>> You should determine/calculate the divisor in the driver code, depending
>> on clocks and bus frequencies. You don't need a property for that, usually.
> 
> Not only clocks and bus frequencies, but also a third property is required.
> 
> The frequency divison calculation formula of i2c is
> Prcescale = clock_a/(clock_div*clock_s)-1
> 
> There is three parameters in this formula:
> clock_a represents the input clock, which is described by "clocks",
> clock_s represents the i2c bus frequency, which is described by 
> "clock-frequency",
> but there is no existing property to describe clock_div, which has 
> different value
> on different platform (for example, it is 5 on 7a1000/7a2000, 4 on 
> 2K1000/2K2000,
> 5.5 on 2K3000.), So I need a property to describe clock_div in this formula.

So it is fixed per compatible? Then you do not need.


Best regards,
Krzysztof

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/4/10 下午2:42, Krzysztof Kozlowski wrote:
> On 10/04/2026 05:04, Hongliang Wang wrote:
>> Hi, Krzysztof
>>
>> On 2026/4/9 下午8:11, Krzysztof Kozlowski wrote:
>>> On 09/04/2026 14:03, Hongliang Wang wrote:
>>>> I have a question, the input clock of i2c controller can be described by
>>>> "clocks",
>>>> but there is no existing attribute can describe the divisor of the input
>>>> clock,
>>>> Can I define a new attribute named "clock-div" to describe it in DT
>>>> bindings?
>>>> or do you have any standard solutions for the divisor problem? Thank you.
>>>>
>>> You should determine/calculate the divisor in the driver code, depending
>>> on clocks and bus frequencies. You don't need a property for that, usually.
>> Not only clocks and bus frequencies, but also a third property is required.
>>
>> The frequency divison calculation formula of i2c is
>> Prcescale = clock_a/(clock_div*clock_s)-1
>>
>> There is three parameters in this formula:
>> clock_a represents the input clock, which is described by "clocks",
>> clock_s represents the i2c bus frequency, which is described by
>> "clock-frequency",
>> but there is no existing property to describe clock_div, which has
>> different value
>> on different platform (for example, it is 5 on 7a1000/7a2000, 4 on
>> 2K1000/2K2000,
>> 5.5 on 2K3000.), So I need a property to describe clock_div in this formula.
> So it is fixed per compatible? Then you do not need.
>
OK, I will fix the clock_div based on per compatible. thank you.
> Best regards,
> Krzysztof
Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Yao Zi]

On Thu, Apr 09, 2026 at 08:03:47PM +0800, Hongliang Wang wrote:
> Hi, Krzysztof
> 
> On 2026/3/31 下午3:11, Hongliang Wang wrote:
> > 
> > On 2026/3/30 下午3:23, Krzysztof Kozlowski wrote:
> > > On 30/03/2026 09:18, Hongliang Wang wrote:
> > > > On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
> > > > > On 27/03/2026 04:09, Hongliang Wang wrote:
> > > > > > The initial idea was that this patch could be used for
> > > > > > both ACPI and DTS.
> > > > > > > > > The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
> > > > > > > > > platform, parse
> > > > > > > > > the same parameters regardless of dts or
> > > > > > > > > acpi parameter passing, So
> > > > > > > > > clock-input
> > > > > > > > > and clock-div attributes are defined to
> > > > > > > > > describe input clock of i2c
> > > > > > > > > controller and
> > > > > > > > > divisor of input clock. It can be used on
> > > > > > > > > both 2K and 3A+7A platform.
> > > > > > > > And you cannot use them in DTS.
> > > > > > OK
> > > > > > > I need to keep guessing what you want to achieve,
> > > > > > > because neither your
> > > > > > > message nor commit text was explicit
> > > > > > What I want to achieve is to describe the input clock
> > > > > > and divisor of I2C
> > > > > > controller
> > > > > Input clocks are defined as clock inputs obviously in DT, not as
> > > > > integers. Bindings need to describe the hardware, so start with that.
> > > > I can describe the hardware in loongson,ls2x-i2c.yaml, and I
> > > > would like to
> > > > confirm with you what final implementation plan you agree to? clock
> > > > framework
> > > > or custom clock-input an clock-div attributes? if clock framework, how
> > > > can it
> > > > also be used for ACPI?
> > > And you ask DT maintainer for that? It's not relevant. You sent DT
> > > bindings patch, so this patch must be correct and we discuss this patch
> > > here.
> > I don't. My idea is that if the clock input attribute can't be used for
> > both
> > dts and acpi, then clock framework will be used for dts and new define
> > attribute
> > will be used for acpi. I will first implement the hardware description
> > and clock
> > framework in Bindings.
> > > Best regards,
> > > Krzysztof
> > 
> > Best regards,
> > Hongliang Wang
> > 
> 
> I have a question, the input clock of i2c controller can be described by
> "clocks",
> but there is no existing attribute can describe the divisor of the input
> clock,

From the description of 7A1000's user manual (section 2.3
"时钟功能描述"), it seems the divider isn't part of the I2C controller,
but instead is an on-chip divider with fixed 1/2 factor, feeding both
"MISC" block (including I2C) and SPI.

> Can I define a new attribute named "clock-div" to describe it in DT
> bindings?
> or do you have any standard solutions for the divisor problem? Thank you.

If these devicetree-based Loongson platforms follow a similar pattern as
the bridge chip, then the divisor shouldn't be described in the I2C
controller node. You may want to include a "fixed-factor-clock" node to
match the hardware.

> Best regards,
> Hongliang Wang
> 
> 
> 

Regards,
Yao Zi

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

Hi, Yao Zi

On 2026/4/9 下午11:29, Yao Zi wrote:
> On Thu, Apr 09, 2026 at 08:03:47PM +0800, Hongliang Wang wrote:
>> Hi, Krzysztof
>>
>> On 2026/3/31 下午3:11, Hongliang Wang wrote:
>>> On 2026/3/30 下午3:23, Krzysztof Kozlowski wrote:
>>>> On 30/03/2026 09:18, Hongliang Wang wrote:
>>>>> On 2026/3/27 下午2:39, Krzysztof Kozlowski wrote:
>>>>>> On 27/03/2026 04:09, Hongliang Wang wrote:
>>>>>>> The initial idea was that this patch could be used for
>>>>>>> both ACPI and DTS.
>>>>>>>>>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A
>>>>>>>>>> platform, parse
>>>>>>>>>> the same parameters regardless of dts or
>>>>>>>>>> acpi parameter passing, So
>>>>>>>>>> clock-input
>>>>>>>>>> and clock-div attributes are defined to
>>>>>>>>>> describe input clock of i2c
>>>>>>>>>> controller and
>>>>>>>>>> divisor of input clock. It can be used on
>>>>>>>>>> both 2K and 3A+7A platform.
>>>>>>>>> And you cannot use them in DTS.
>>>>>>> OK
>>>>>>>> I need to keep guessing what you want to achieve,
>>>>>>>> because neither your
>>>>>>>> message nor commit text was explicit
>>>>>>> What I want to achieve is to describe the input clock
>>>>>>> and divisor of I2C
>>>>>>> controller
>>>>>> Input clocks are defined as clock inputs obviously in DT, not as
>>>>>> integers. Bindings need to describe the hardware, so start with that.
>>>>> I can describe the hardware in loongson,ls2x-i2c.yaml, and I
>>>>> would like to
>>>>> confirm with you what final implementation plan you agree to? clock
>>>>> framework
>>>>> or custom clock-input an clock-div attributes? if clock framework, how
>>>>> can it
>>>>> also be used for ACPI?
>>>> And you ask DT maintainer for that? It's not relevant. You sent DT
>>>> bindings patch, so this patch must be correct and we discuss this patch
>>>> here.
>>> I don't. My idea is that if the clock input attribute can't be used for
>>> both
>>> dts and acpi, then clock framework will be used for dts and new define
>>> attribute
>>> will be used for acpi. I will first implement the hardware description
>>> and clock
>>> framework in Bindings.
>>>> Best regards,
>>>> Krzysztof
>>> Best regards,
>>> Hongliang Wang
>>>
>> I have a question, the input clock of i2c controller can be described by
>> "clocks",
>> but there is no existing attribute can describe the divisor of the input
>> clock,
>  From the description of 7A1000's user manual (section 2.3
> "时钟功能描述"), it seems the divider isn't part of the I2C controller,
> but instead is an on-chip divider with fixed 1/2 factor, feeding both
> "MISC" block (including I2C) and SPI.
>
>> Can I define a new attribute named "clock-div" to describe it in DT
>> bindings?
>> or do you have any standard solutions for the divisor problem? Thank you.
> If these devicetree-based Loongson platforms follow a similar pattern as
> the bridge chip, then the divisor shouldn't be described in the I2C
> controller node. You may want to include a "fixed-factor-clock" node to
> match the hardware.
>
Sorry, I didn't describe it clearly, The divisor I described doesn't 
refer to
the fixed 1/2 factor feeding MISC block(including I2C). What I have 
described
is included in the formula in section 10.2 of the 7a1000 user manual.

Prcescale = clock_a/(clock_div*clock_s)-1

clock_a represents the input clock(it is the fixed 1/2 factor feeding 
MISC block
(including I2C), 50M on 7a1000), which is described by "clocks",
clock_s represents the i2c bus frequency, which is described by 
"clock-frequency",
The divisor I described is clock_div in formula. which has different 
value on
different platform. for example, it is 5 on 7a1000/7a2000, 4 on 
2K1000/2K2000,
5.5 on 2K3000. I need a property to describe clock_div in this formula.
>> Best regards,
>> Hongliang Wang
>>
>>
>>
> Regards,
> Yao Zi
Best regards,
Hongliang Wang

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Andi Shyti]

Hi Hongliang,

> Consider the clock framework relies on the device tree, and can only be used
> on
> Loongson 2K platform with dts parameter mechanism, It cannot be used on
> Loongson
> 3A+7A platform with the acpi parameter mechanism.
> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A platform,
> parse
> the same parameters regardless of dts or acpi parameter passing, So
> clock-input
> and clock-div attributes are defined to describe input clock of i2c
> controller and
> divisor of input clock. It can be used on both 2K and 3A+7A platform.

Thanks for the explanation, but please do consider Krzysztof's
review. Without him agreeing on the change I can't take your
patch.

If you have questions, please ask.

Andi

Re: [PATCH v1 1/2] dt-bindings: i2c: ls2x-i2c: Add clock- related properties

[Hongliang Wang]

On 2026/3/26 下午3:21, Andi Shyti wrote:
> Hi Hongliang,
>
>> Consider the clock framework relies on the device tree, and can only be used
>> on
>> Loongson 2K platform with dts parameter mechanism, It cannot be used on
>> Loongson
>> 3A+7A platform with the acpi parameter mechanism.
>> The i2c-ls2x driver is compatible with both Loongson 2K and 3A+7A platform,
>> parse
>> the same parameters regardless of dts or acpi parameter passing, So
>> clock-input
>> and clock-div attributes are defined to describe input clock of i2c
>> controller and
>> divisor of input clock. It can be used on both 2K and 3A+7A platform.
> Thanks for the explanation, but please do consider Krzysztof's
> review. Without him agreeing on the change I can't take your
> patch.
>
> If you have questions, please ask.
OK, thank you
> Andi

[PATCH v1 2/2] i2c: ls2x: Add clock-input and clock-div properties parsing

[Hongliang Wang]

From: wanghongliang <wanghongliang@loongson.cn>

On 3A/7A/2K platform, the input clock of i2c controller and
the divisor of input clock maybe different, so define the
clock-input and clock-div properties in i2c controller node
description of dts or acpi. clock-input describes the input
clock of i2c controller and clock-div describes the clock
divisor of the input clock.

The formulas is
Prcescale = (clock-input * 10) / (clock-div * t->bus_freq_hz) - 1

The value of clock-input and clock-div on different platforms are
listed below:

1.7A1000/7A2000:
clock-input = 80MHZ
clock-div = 50

2.3A5000/3C5000/3A6000/3C6000:
clock-input = 100MHZ
clock-div = 40

3.2K0500/2K1000LA/2K2000:
clock-input = 100MHZ
clock-div = 40

4.2K3000:
clock-input = 100MHZ
clock-div = 55

Signed-off-by: wanghongliang <wanghongliang@loongson.cn>
---
 drivers/i2c/busses/i2c-ls2x.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/i2c/busses/i2c-ls2x.c b/drivers/i2c/busses/i2c-ls2x.c
index b475dd27b7af..08c242e5fd75 100644
--- a/drivers/i2c/busses/i2c-ls2x.c
+++ b/drivers/i2c/busses/i2c-ls2x.c
@@ -96,6 +96,7 @@ static irqreturn_t ls2x_i2c_isr(int this_irq, void *dev_id)
 static void ls2x_i2c_adjust_bus_speed(struct ls2x_i2c_priv *priv)
 {
 	u16 val;
+	u32 pclk, div;
 	struct i2c_timings *t = &priv->i2c_t;
 	struct device *dev = priv->adapter.dev.parent;
 	u32 acpi_speed = i2c_acpi_find_bus_speed(dev);
@@ -107,12 +108,16 @@ static void ls2x_i2c_adjust_bus_speed(struct ls2x_i2c_priv *priv)
 	else
 		t->bus_freq_hz = LS2X_I2C_FREQ_STD;
 
+	if (!device_property_read_u32(dev, "clock-input", &pclk) &&
+	    !device_property_read_u32(dev, "clock-div", &div) && div != 0)
+		val = (pclk * 10) / (div * t->bus_freq_hz) - 1;
+	else
+		val = LS2X_I2C_PCLK_FREQ / (5 * t->bus_freq_hz) - 1;
 	/*
 	 * According to the chip manual, we can only access the registers as bytes,
 	 * otherwise the high bits will be truncated.
 	 * So set the I2C frequency with a sequential writeb() instead of writew().
 	 */
-	val = LS2X_I2C_PCLK_FREQ / (5 * t->bus_freq_hz) - 1;
 	writeb(FIELD_GET(GENMASK(7, 0), val), priv->base + I2C_LS2X_PRER_LO);
 	writeb(FIELD_GET(GENMASK(15, 8), val), priv->base + I2C_LS2X_PRER_HI);
 }
-- 
2.47.2