[PATCH 1/2] dt-bindings: i2c: cadence: Document reset property

[PATCH 1/2] dt-bindings: i2c: cadence: Document reset property

[Lars-Peter Clausen]

The Cadence I2C controller has an external reset that needs to be
de-asserted before the I2C controller can be accessed.

Document the `resets` devicetree property that can be used to describe how
the reset signal is connected.

While the reset signal will always be present in hardware the devicetree
property is kept optional for backwards compatibility with existing systems
that do not specify the reset property and where the reset signal might not
be controlled by operating system.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
---
 Documentation/devicetree/bindings/i2c/cdns,i2c-r1p10.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/Documentation/devicetree/bindings/i2c/cdns,i2c-r1p10.yaml b/Documentation/devicetree/bindings/i2c/cdns,i2c-r1p10.yaml
index 9187015d9702..76e2b9a10de8 100644
--- a/Documentation/devicetree/bindings/i2c/cdns,i2c-r1p10.yaml
+++ b/Documentation/devicetree/bindings/i2c/cdns,i2c-r1p10.yaml
@@ -24,6 +24,9 @@ properties:
   clocks:
     minItems: 1
 
+  resets:
+    maxItems: 1
+
   interrupts:
     maxItems: 1
 
-- 
2.30.2

[PATCH 2/2] i2c: cadence: Add reset controller support

[Lars-Peter Clausen]

The Cadence I2C controller has an external reset signal that needs to be
de-asserted before the I2C controller can be used.

Add support to the driver to be able to take the peripheral out of reset
using the reset controller API. The reset is optional in the driver for
compatibility to systems where the reset managed by the bootloader.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
---
 drivers/i2c/busses/i2c-cadence.c | 26 ++++++++++++++++++++++----
 1 file changed, 22 insertions(+), 4 deletions(-)

diff --git a/drivers/i2c/busses/i2c-cadence.c b/drivers/i2c/busses/i2c-cadence.c
index 8f61a633c42c..102774ab2497 100644
--- a/drivers/i2c/busses/i2c-cadence.c
+++ b/drivers/i2c/busses/i2c-cadence.c
@@ -16,6 +16,7 @@
 #include <linux/of.h>
 #include <linux/pm_runtime.h>
 #include <linux/pinctrl/consumer.h>
+#include <linux/reset.h>
 
 /* Register offsets for the I2C device. */
 #define CDNS_I2C_CR_OFFSET		0x00 /* Control Register, RW */
@@ -178,6 +179,7 @@ enum cdns_i2c_slave_state {
  * @bus_hold_flag:	Flag used in repeated start for clearing HOLD bit
  * @clk:		Pointer to struct clk
  * @clk_rate_change_nb:	Notifier block for clock rate changes
+ * @reset:		Reset control for the device
  * @quirks:		flag for broken hold bit usage in r1p10
  * @ctrl_reg:		Cached value of the control register.
  * @ctrl_reg_diva_divb: value of fields DIV_A and DIV_B from CR register
@@ -204,6 +206,7 @@ struct cdns_i2c {
 	unsigned int bus_hold_flag;
 	struct clk *clk;
 	struct notifier_block clk_rate_change_nb;
+	struct reset_control *reset;
 	u32 quirks;
 	u32 ctrl_reg;
 	struct i2c_bus_recovery_info rinfo;
@@ -1325,10 +1328,22 @@ static int cdns_i2c_probe(struct platform_device *pdev)
 		return dev_err_probe(&pdev->dev, PTR_ERR(id->clk),
 				     "input clock not found.\n");
 
+	id->reset = devm_reset_control_get_optional_shared(&pdev->dev, NULL);
+	if (IS_ERR(id->reset))
+		return dev_err_probe(&pdev->dev, PTR_ERR(id->reset),
+							 "Failed to request reset.\n");
+
 	ret = clk_prepare_enable(id->clk);
 	if (ret)
 		dev_err(&pdev->dev, "Unable to enable clock.\n");
 
+	ret = reset_control_deassert(id->reset);
+	if (ret) {
+		dev_err_probe(&pdev->dev, ret,
+					  "Failed to de-assert reset.\n");
+		goto err_clk_dis;
+	}
+
 	pm_runtime_set_autosuspend_delay(id->dev, CNDS_I2C_PM_TIMEOUT);
 	pm_runtime_use_autosuspend(id->dev);
 	pm_runtime_set_active(id->dev);
@@ -1360,28 +1375,30 @@ static int cdns_i2c_probe(struct platform_device *pdev)
 	if (ret) {
 		dev_err(&pdev->dev, "invalid SCL clock: %u Hz\n", id->i2c_clk);
 		ret = -EINVAL;
-		goto err_clk_dis;
+		goto err_clk_notifier_unregister;
 	}
 
 	ret = devm_request_irq(&pdev->dev, irq, cdns_i2c_isr, 0,
 				 DRIVER_NAME, id);
 	if (ret) {
 		dev_err(&pdev->dev, "cannot get irq %d\n", irq);
-		goto err_clk_dis;
+		goto err_clk_notifier_unregister;
 	}
 	cdns_i2c_init(id);
 
 	ret = i2c_add_adapter(&id->adap);
 	if (ret < 0)
-		goto err_clk_dis;
+		goto err_clk_notifier_unregister;
 
 	dev_info(&pdev->dev, "%u kHz mmio %08lx irq %d\n",
 		 id->i2c_clk / 1000, (unsigned long)r_mem->start, irq);
 
 	return 0;
 
-err_clk_dis:
+err_clk_notifier_unregister:
 	clk_notifier_unregister(id->clk, &id->clk_rate_change_nb);
+	reset_control_assert(id->reset);
+err_clk_dis:
 	clk_disable_unprepare(id->clk);
 	pm_runtime_disable(&pdev->dev);
 	pm_runtime_set_suspended(&pdev->dev);
@@ -1406,6 +1423,7 @@ static int cdns_i2c_remove(struct platform_device *pdev)
 
 	i2c_del_adapter(&id->adap);
 	clk_notifier_unregister(id->clk, &id->clk_rate_change_nb);
+	reset_control_assert(id->reset);
 	clk_disable_unprepare(id->clk);
 
 	return 0;
-- 
2.30.2

Re: [PATCH 2/2] i2c: cadence: Add reset controller support

[Michal Simek]

On 3/30/23 20:04, Lars-Peter Clausen wrote:
> The Cadence I2C controller has an external reset signal that needs to be
> de-asserted before the I2C controller can be used.
> 
> Add support to the driver to be able to take the peripheral out of reset
> using the reset controller API. The reset is optional in the driver for
> compatibility to systems where the reset managed by the bootloader.
> 
> Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
> ---
>   drivers/i2c/busses/i2c-cadence.c | 26 ++++++++++++++++++++++----
>   1 file changed, 22 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-cadence.c b/drivers/i2c/busses/i2c-cadence.c
> index 8f61a633c42c..102774ab2497 100644
> --- a/drivers/i2c/busses/i2c-cadence.c
> +++ b/drivers/i2c/busses/i2c-cadence.c
> @@ -16,6 +16,7 @@
>   #include <linux/of.h>
>   #include <linux/pm_runtime.h>
>   #include <linux/pinctrl/consumer.h>
> +#include <linux/reset.h>
>   
>   /* Register offsets for the I2C device. */
>   #define CDNS_I2C_CR_OFFSET		0x00 /* Control Register, RW */
> @@ -178,6 +179,7 @@ enum cdns_i2c_slave_state {
>    * @bus_hold_flag:	Flag used in repeated start for clearing HOLD bit
>    * @clk:		Pointer to struct clk
>    * @clk_rate_change_nb:	Notifier block for clock rate changes
> + * @reset:		Reset control for the device
>    * @quirks:		flag for broken hold bit usage in r1p10
>    * @ctrl_reg:		Cached value of the control register.
>    * @ctrl_reg_diva_divb: value of fields DIV_A and DIV_B from CR register
> @@ -204,6 +206,7 @@ struct cdns_i2c {
>   	unsigned int bus_hold_flag;
>   	struct clk *clk;
>   	struct notifier_block clk_rate_change_nb;
> +	struct reset_control *reset;
>   	u32 quirks;
>   	u32 ctrl_reg;
>   	struct i2c_bus_recovery_info rinfo;
> @@ -1325,10 +1328,22 @@ static int cdns_i2c_probe(struct platform_device *pdev)
>   		return dev_err_probe(&pdev->dev, PTR_ERR(id->clk),
>   				     "input clock not found.\n");
>   
> +	id->reset = devm_reset_control_get_optional_shared(&pdev->dev, NULL);
> +	if (IS_ERR(id->reset))
> +		return dev_err_probe(&pdev->dev, PTR_ERR(id->reset),
> +							 "Failed to request reset.\n");

incorrect alignment.

> +
>   	ret = clk_prepare_enable(id->clk);
>   	if (ret)
>   		dev_err(&pdev->dev, "Unable to enable clock.\n");
>   
> +	ret = reset_control_deassert(id->reset);
> +	if (ret) {
> +		dev_err_probe(&pdev->dev, ret,
> +					  "Failed to de-assert reset.\n");

incorrect alignment

The rest looks good to me.

M

Re: [PATCH 1/2] dt-bindings: i2c: cadence: Document reset property

[Krzysztof Kozlowski]

On 30/03/2023 20:04, Lars-Peter Clausen wrote:
> The Cadence I2C controller has an external reset that needs to be
> de-asserted before the I2C controller can be accessed.
> 
> Document the `resets` devicetree property that can be used to describe how
> the reset signal is connected.

You could add it also to the example to have complete picture (and
validate your change).


Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>

Best regards,
Krzysztof

Re: [PATCH 2/2] i2c: cadence: Add reset controller support

[Lars-Peter Clausen]

On 3/30/23 23:49, Michal Simek wrote:
>
>
> On 3/30/23 20:04, Lars-Peter Clausen wrote:
>> The Cadence I2C controller has an external reset signal that needs to be
>> de-asserted before the I2C controller can be used.
>>
>> Add support to the driver to be able to take the peripheral out of reset
>> using the reset controller API. The reset is optional in the driver for
>> compatibility to systems where the reset managed by the bootloader.
>>
>> Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
>> ---
>>   drivers/i2c/busses/i2c-cadence.c | 26 ++++++++++++++++++++++----
>>   1 file changed, 22 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/i2c/busses/i2c-cadence.c 
>> b/drivers/i2c/busses/i2c-cadence.c
>> index 8f61a633c42c..102774ab2497 100644
>> --- a/drivers/i2c/busses/i2c-cadence.c
>> +++ b/drivers/i2c/busses/i2c-cadence.c
>> @@ -16,6 +16,7 @@
>>   #include <linux/of.h>
>>   #include <linux/pm_runtime.h>
>>   #include <linux/pinctrl/consumer.h>
>> +#include <linux/reset.h>
>>     /* Register offsets for the I2C device. */
>>   #define CDNS_I2C_CR_OFFSET        0x00 /* Control Register, RW */
>> @@ -178,6 +179,7 @@ enum cdns_i2c_slave_state {
>>    * @bus_hold_flag:    Flag used in repeated start for clearing HOLD 
>> bit
>>    * @clk:        Pointer to struct clk
>>    * @clk_rate_change_nb:    Notifier block for clock rate changes
>> + * @reset:        Reset control for the device
>>    * @quirks:        flag for broken hold bit usage in r1p10
>>    * @ctrl_reg:        Cached value of the control register.
>>    * @ctrl_reg_diva_divb: value of fields DIV_A and DIV_B from CR 
>> register
>> @@ -204,6 +206,7 @@ struct cdns_i2c {
>>       unsigned int bus_hold_flag;
>>       struct clk *clk;
>>       struct notifier_block clk_rate_change_nb;
>> +    struct reset_control *reset;
>>       u32 quirks;
>>       u32 ctrl_reg;
>>       struct i2c_bus_recovery_info rinfo;
>> @@ -1325,10 +1328,22 @@ static int cdns_i2c_probe(struct 
>> platform_device *pdev)
>>           return dev_err_probe(&pdev->dev, PTR_ERR(id->clk),
>>                        "input clock not found.\n");
>>   +    id->reset = devm_reset_control_get_optional_shared(&pdev->dev, 
>> NULL);
>> +    if (IS_ERR(id->reset))
>> +        return dev_err_probe(&pdev->dev, PTR_ERR(id->reset),
>> +                             "Failed to request reset.\n");
>
> incorrect alignment.
>
>> +
>>       ret = clk_prepare_enable(id->clk);
>>       if (ret)
>>           dev_err(&pdev->dev, "Unable to enable clock.\n");
>>   +    ret = reset_control_deassert(id->reset);
>> +    if (ret) {
>> +        dev_err_probe(&pdev->dev, ret,
>> +                      "Failed to de-assert reset.\n");
>
> incorrect alignment
Wrong tabstop, that's embarrassing. Will fix.

Re: [PATCH 1/2] dt-bindings: i2c: cadence: Document reset property

[Lars-Peter Clausen]

On 3/31/23 02:42, Krzysztof Kozlowski wrote:
> On 30/03/2023 20:04, Lars-Peter Clausen wrote:
>> The Cadence I2C controller has an external reset that needs to be
>> de-asserted before the I2C controller can be accessed.
>>
>> Document the `resets` devicetree property that can be used to describe how
>> the reset signal is connected.
> You could add it also to the example to have complete picture (and
> validate your change).
>
>
> Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
>
When I resend with it added to the example, do you want me to keep or 
drop the acked-by?

Re: [PATCH 1/2] dt-bindings: i2c: cadence: Document reset property

[Krzysztof Kozlowski]

On 31/03/2023 19:11, Lars-Peter Clausen wrote:
> On 3/31/23 02:42, Krzysztof Kozlowski wrote:
>> On 30/03/2023 20:04, Lars-Peter Clausen wrote:
>>> The Cadence I2C controller has an external reset that needs to be
>>> de-asserted before the I2C controller can be accessed.
>>>
>>> Document the `resets` devicetree property that can be used to describe how
>>> the reset signal is connected.
>> You could add it also to the example to have complete picture (and
>> validate your change).
>>
>>
>> Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
>>
> When I resend with it added to the example, do you want me to keep or 
> drop the acked-by?

Keep the ack, please.

Best regards,
Krzysztof