1. Patchew
  2. linux
  3. [PATCH 0/6] misc: add amd side-band functionality
  4. View patch

[PATCH 2/6] misc: sbrmi: Add platform device add to create platform device

Akshay Gupta posted 6 patches 1 year, 7 months ago
[PATCH 2/6] misc: sbrmi: Add platform device add to create platform device
Posted by Akshay Gupta 1 year, 7 months ago 
- AMD provides socket power information from out of band
  which can be read by sensors.
- platform driver will probe drivers/hwmon/sbrmi as a platform device
  and share the sbrmi device data.

Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
Reviewed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
---
 drivers/misc/amd-sb/sbrmi-i2c.c | 25 ++++++++++++++++++++++++-
 include/misc/amd-sb.h           |  2 ++
 2 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/drivers/misc/amd-sb/sbrmi-i2c.c b/drivers/misc/amd-sb/sbrmi-i2c.c
index c4903d9e9f0f..b593bbdd78e0 100644
--- a/drivers/misc/amd-sb/sbrmi-i2c.c
+++ b/drivers/misc/amd-sb/sbrmi-i2c.c
@@ -72,7 +72,29 @@ static int sbrmi_i2c_probe(struct i2c_client *client)
 		return ret;
 
 	/* Cache maximum power limit */
-	return sbrmi_get_max_pwr_limit(data);
+	ret = sbrmi_get_max_pwr_limit(data);
+	if (ret < 0)
+		return ret;
+
+	dev_set_drvdata(dev, (void *)data);
+	data->pdev = platform_device_register_data(dev, "sbrmi-hwmon",
+						   PLATFORM_DEVID_NONE,
+						   data,
+						   sizeof(struct sbrmi_data));
+	if (IS_ERR(data->pdev)) {
+		pr_err("unable to register platform device for sbrmi-hwmon\n");
+		return PTR_ERR(data->pdev);
+	}
+	return ret;
+}
+
+static void sbrmi_i2c_remove(struct i2c_client *client)
+{
+	struct sbrmi_data *data = dev_get_drvdata(&client->dev);
+
+	if (!data)
+		return;
+	platform_device_unregister(data->pdev);
 }
 
 static const struct i2c_device_id sbrmi_id[] = {
@@ -95,6 +117,7 @@ static struct i2c_driver sbrmi_driver = {
 		.of_match_table = of_match_ptr(sbrmi_of_match),
 	},
 	.probe = sbrmi_i2c_probe,
+	.remove = sbrmi_i2c_remove,
 	.id_table = sbrmi_id,
 };
 
diff --git a/include/misc/amd-sb.h b/include/misc/amd-sb.h
index e1a012fcdff9..79b76dd6068a 100644
--- a/include/misc/amd-sb.h
+++ b/include/misc/amd-sb.h
@@ -8,6 +8,7 @@
 
 #include <linux/mutex.h>
 #include <linux/i2c.h>
+#include <linux/platform_device.h>
 /*
  * SB-RMI supports soft mailbox service request to MP1 (power management
  * firmware) through SBRMI inbound/outbound message registers.
@@ -24,6 +25,7 @@ enum sbrmi_msg_id {
 struct sbrmi_data {
 	struct i2c_client *client;
 	struct mutex lock;
+	struct platform_device *pdev;
 	u32 pwr_limit_max;
 };
 
-- 
2.25.1
Re: [PATCH 2/6] misc: sbrmi: Add platform device add to create platform device
Posted by Greg KH 1 year, 7 months ago 
On Thu, Jul 04, 2024 at 11:16:20AM +0000, Akshay Gupta wrote:
> - AMD provides socket power information from out of band
>   which can be read by sensors.
> - platform driver will probe drivers/hwmon/sbrmi as a platform device
>   and share the sbrmi device data.

So you are "splitting" a real device into different ones using a
platform device?  THat's not ok, and an abuse of the platform api.
Please use the correct one for that instead.


> 
> Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
> Reviewed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
> ---
>  drivers/misc/amd-sb/sbrmi-i2c.c | 25 ++++++++++++++++++++++++-
>  include/misc/amd-sb.h           |  2 ++
>  2 files changed, 26 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/misc/amd-sb/sbrmi-i2c.c b/drivers/misc/amd-sb/sbrmi-i2c.c
> index c4903d9e9f0f..b593bbdd78e0 100644
> --- a/drivers/misc/amd-sb/sbrmi-i2c.c
> +++ b/drivers/misc/amd-sb/sbrmi-i2c.c
> @@ -72,7 +72,29 @@ static int sbrmi_i2c_probe(struct i2c_client *client)
>  		return ret;
>  
>  	/* Cache maximum power limit */
> -	return sbrmi_get_max_pwr_limit(data);
> +	ret = sbrmi_get_max_pwr_limit(data);
> +	if (ret < 0)
> +		return ret;
> +
> +	dev_set_drvdata(dev, (void *)data);

No need to cast, right?

> +	data->pdev = platform_device_register_data(dev, "sbrmi-hwmon",
> +						   PLATFORM_DEVID_NONE,

Yeah, that's not ok.  Please do this correctly, as this is NOT a
platform device, but rather a made-up one that you just created out of
no where.  Instead use the correct apis for that.

> +						   data,
> +						   sizeof(struct sbrmi_data));
> +	if (IS_ERR(data->pdev)) {
> +		pr_err("unable to register platform device for sbrmi-hwmon\n");
> +		return PTR_ERR(data->pdev);

You don't need to unwind anything else here?



> +	}
> +	return ret;
> +}
> +
> +static void sbrmi_i2c_remove(struct i2c_client *client)
> +{
> +	struct sbrmi_data *data = dev_get_drvdata(&client->dev);
> +
> +	if (!data)
> +		return;

How can that happen?

thanks,

greg k-h
Re: [PATCH 2/6] misc: sbrmi: Add platform device add to create platform device
Posted by Gupta, Akshay 1 year, 7 months ago 
On 7/4/2024 5:20 PM, Greg KH wrote:
> Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
>
>
> On Thu, Jul 04, 2024 at 11:16:20AM +0000, Akshay Gupta wrote:
>> - AMD provides socket power information from out of band
>>    which can be read by sensors.
>> - platform driver will probe drivers/hwmon/sbrmi as a platform device
>>    and share the sbrmi device data.
> So you are "splitting" a real device into different ones using a
> platform device?  THat's not ok, and an abuse of the platform api.
> Please use the correct one for that instead.

is it ok to call API, devm_hwmon_device_register_with_info() from misc 
driver

which will require to defining the hwmon ops in here.

>
>> Signed-off-by: Akshay Gupta <akshay.gupta@amd.com>
>> Reviewed-by: Naveen Krishna Chatradhi <naveenkrishna.chatradhi@amd.com>
>> ---
>>   drivers/misc/amd-sb/sbrmi-i2c.c | 25 ++++++++++++++++++++++++-
>>   include/misc/amd-sb.h           |  2 ++
>>   2 files changed, 26 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/misc/amd-sb/sbrmi-i2c.c b/drivers/misc/amd-sb/sbrmi-i2c.c
>> index c4903d9e9f0f..b593bbdd78e0 100644
>> --- a/drivers/misc/amd-sb/sbrmi-i2c.c
>> +++ b/drivers/misc/amd-sb/sbrmi-i2c.c
>> @@ -72,7 +72,29 @@ static int sbrmi_i2c_probe(struct i2c_client *client)
>>                return ret;
>>
>>        /* Cache maximum power limit */
>> -     return sbrmi_get_max_pwr_limit(data);
>> +     ret = sbrmi_get_max_pwr_limit(data);
>> +     if (ret < 0)
>> +             return ret;
>> +
>> +     dev_set_drvdata(dev, (void *)data);
> No need to cast, right?
Yes, will update.
>
>> +     data->pdev = platform_device_register_data(dev, "sbrmi-hwmon",
>> +                                                PLATFORM_DEVID_NONE,
> Yeah, that's not ok.  Please do this correctly, as this is NOT a
> platform device, but rather a made-up one that you just created out of
> no where.  Instead use the correct apis for that.
>
>> +                                                data,
>> +                                                sizeof(struct sbrmi_data));
>> +     if (IS_ERR(data->pdev)) {
>> +             pr_err("unable to register platform device for sbrmi-hwmon\n");
>> +             return PTR_ERR(data->pdev);
> You don't need to unwind anything else here?
Yes, not required.
>
>
>
>> +     }
>> +     return ret;
>> +}
>> +
>> +static void sbrmi_i2c_remove(struct i2c_client *client)
>> +{
>> +     struct sbrmi_data *data = dev_get_drvdata(&client->dev);
>> +
>> +     if (!data)
>> +             return;
> How can that happen?
Its just a safe check, once we remove platform driver this will not be 
required.
> greg k-h

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.