Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <f6e4a78d-0199-3135-f85d-800457a731b0@linux.intel.com>
Date:   Thu, 20 Apr 2023 09:46:20 -0500
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.0
Subject: Re: [PATCH 3/5] hwmon: (socfpga) Add hardware monitoring support on
 SoCFPGA platforms
To:     Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc:     Guenter Roeck <linux@roeck-us.net>, linux-hwmon@vger.kernel.org,
        dinguyen@kernel.org, devicetree@vger.kernel.org,
        robh+dt@kernel.org, krzysztof.kozlowski+dt@linaro.org,
        linux-kernel@vger.kernel.org, jdelvare@suse.com
References: <20230410153314.27127-1-dinh.nguyen@linux.intel.com>
 <20230410153314.27127-3-dinh.nguyen@linux.intel.com>
 <09730359-8731-e21e-3335-bf60ba7f1280@roeck-us.net>
 <a3e966f8-8e9d-7081-1665-9d2e87acb310@linux.intel.com>
 <8d158880-1e6a-5fdd-dae7-a7647794eb60@roeck-us.net>
 <a1a34c32-dbd4-7a77-ab7e-5e34af85900f@linux.intel.com>
 <ZD/UsuzhZmK3AFJn@smile.fi.intel.com>
Content-Language: en-US
From:   Dinh Nguyen <dinh.nguyen@linux.intel.com>
In-Reply-To: <ZD/UsuzhZmK3AFJn@smile.fi.intel.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


On 4/19/2023 6:46 AM, Andy Shevchenko wrote:
> On Tue, Apr 18, 2023 at 12:29:40PM -0500, Dinh Nguyen wrote:
>> On 4/17/2023 4:51 PM, Guenter Roeck wrote:
>>> On 4/17/23 13:55, Dinh Nguyen wrote:
> ...
>
>>> ... and this contradict each other. If bit 31 indicates an error,
>>> this can not be a signed 32-bit value.
>>>
>> You're right! I've re-read the spec and should have the the code look for
>> the specific error values:
>>
>> 0x80000000 - inactive
>> 0x80000001 - old value
>> 0x80000002 - invalid channel
>> 0x80000003 -  corrupted.
> No, they are not hex. Probably you need to define an error space with it, but
> at least just use signed _decimal_ values.
>
> Instead of BIT(31) this should go as
>
> #define ..._ERR_BASE   INT_MIN // or equivalent if the type is not int
> #define ..._ERR_MAX ... // or whatever name is better
>
> Then in your code
>
> 	if (value >= _ERR_MAX)
> 		return 0;
>
> 	err = _ERR_MAX - value;
> 	switch (err) {
> 		...
> 	}
>
> P.S. I asked during internal review if the values are bit fielded when errors.
> AFAIU that time they are, now it seems different.

Can I ask what's wrong with this simple implementation?

static int socfpga_hwmon_err_to_errno(struct socfpga_hwmon_priv *priv)
{
         int value = priv->temperature.value;

         switch (value) {
         case ETEMP_NOT_PRESENT:
                 return -ENOENT;
         case ETEMP_CORRUPT:
         case ETEMP_NOT_INITIALIZED:
                 return -ENODATA;
         case ETEMP_BUSY:
                 return -EBUSY;
         case ETEMP_INACTIVE:
         case ETEMP_TIMEOUT:
         case ETEMP_TOO_OLD:
                 return -EAGAIN;
         default:
                 /* No error */
                 return 0;
         }
}

Dinh