linux-kernel - Re: [PATCH 1/4] hwmon (max6639): Use regmap

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <afb5d43b-14ff-4859-afce-000dbd4bb62e@roeck-us.net>
Date: Mon, 29 Apr 2024 06:49:21 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Naresh Solanki <naresh.solanki@...ements.com>
Cc: krzysztof.kozlowski+dt@...aro.org, u.kleine-koenig@...gutronix.de,
 Jean Delvare <jdelvare@...e.com>, linux-hwmon@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] hwmon (max6639): Use regmap

On 4/29/24 01:19, Naresh Solanki wrote:
> Hi Guenter,
> 
> 
> On Sun, 28 Apr 2024 at 22:48, Guenter Roeck <linux@...ck-us.net> wrote:
>>
>> On 4/25/24 02:50, Naresh Solanki wrote:
>> ...
>>> This driver has 27 regmap accesses. Except volatile registers, others are
>>> cached by regmap.
>>> Some function which only access volatile registers will not be able to take
>>> advantage of caching. This is also the case in various other drivers for similar
>>> devices.
>>> Also regmap offers bit handling which makes the code much cleaner.
>>>
>>
>> Maybe I need to make it explicit in documentation. I will not accept regmap
>> conversions unless local caching is dropped. Yes, that means that volatile
>> registers will not be cached. I consider that a positive.
> I agree with you. Regmap conversion wouldn't make sense if local caching
> is present.
> Correct me if I'm wrong, but in this context, local caching points to the
> various variables in max6639_data ?
> i.e.,
>      bool valid;     /* true if following fields are valid */
>      unsigned long last_updated; /* In jiffies */
> 
>      /* Register values sampled regularly */
>      u16 temp[2];        /* Temperature, in 1/8 C, 0..255 C */
>      bool temp_fault[2]; /* Detected temperature diode failure */
>      u8 fan[2];      /* Register value: TACH count for fans >=30 */
>      u8 status;      /* Detected channel alarms and fan failures */
> 
>      /* Register values only written to */
>      u8 pwm[2];      /* Register value: Duty cycle 0..120 */
>      u8 temp_therm[2];   /* THERM Temperature, 0..255 C (->_max) */
>      u8 temp_alert[2];   /* ALERT Temperature, 0..255 C (->_crit) */
>      u8 temp_ot[2];      /* OT Temperature, 0..255 C (->_emergency) */
> 
>      /* Register values initialized only once */
>      u8 ppr;         /* Pulses per rotation 0..3 for 1..4 ppr */
>      u8 rpm_range;       /* Index in above rpm_ranges table */
> 
> Are you asking for removal of all these variables & each read sysfs
> attribute read should access regmap cache directly ?
> 

Mostly yes. Note that "ppr" is only used in max6639_init_client(),
and it is unnecessary to keep it in max6639_data to start with.
rpm_range is ok to keep because it is a calculated initialization value.
The fixed initialization of temp_therm, temp_alert, and temp_ot
is questionable to start with because it overrides bios/rommon
initialization values.

Guenter