[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <ce3866a2-d264-4da9-9411-8427fc1c8f46@roeck-us.net>
Date: Fri, 28 Feb 2025 06:47:42 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: "lihuisong (C)" <lihuisong@...wei.com>
Cc: linux-hwmon@...r.kernel.org, linux-kernel@...r.kernel.org,
zhanjie9@...ilicon.com, zhenglifeng1@...wei.com, liuyonglong@...wei.com,
jdelvare@...e.com
Subject: Re: [RFC] hwmon: (acpi_power_meter) Replace hwmon_device_register
On 2/28/25 00:26, lihuisong (C) wrote:
...
>>> BTW, I have another problem as commit log described:
>>> -->
>>> The path of these sysfs interfaces are modified accordingly if use hwmon_device_register_with_info():
>>> Old: all sysfs interfaces are under acpi device, namely, path is "/sys/class/hwmon/hwmon1/device/" ('device' in the path is a soft link of acpi device)
>>> Now: all sysfs interfaces are under hwmon device, namely, path is "/sys/class/hwmon/hwmon1/"
>>> What do you think about this?
>>>
>>
>> That is as intended. The ABI states that the attributes are under
>> /sys/class/hwmon/hwmonX _or_ /sys/class/hwmon/hwmonX/device/.
>> The ABI does not guarantee that the underlying path remains the same.
> Yes
>> libsensors handles this automatically, as should any userspace programs
>> accessing the attributes directly.
>>
>>
> So we don't need to care or handle this change of syfs path, right?
>
Correct. Each driver converted to the new API experiences the same change.
If you insist keeping the path the same, you'll have to drop this effort
and live with the deprecated message.
Guenter
Powered by blists - more mailing lists