LM83 support

LM83 support

[Jean Delvare]

Hi everyone,

I will add support to LM83 to lm_sensors, as per request from ZOleg
Matrix <zoleg@matrixc.ru>. This is perfectly what I need to start with
real driver coding.

My question now. The LM83 could be supported by the ADM1021 module,
which already supports ADM1021, ADM1023, MAX1617, MAX1617A, NE1617,
NE1617A, TMC10, MC1066 and LM84. The LM84 can easily be seen as a light
version of the LM83. However, I think I remember we decided to avoid
adding support for new devices to existing drivers. Or is is just true
for the w83781d?

I see three possibilities:

1* A new driver for the LM83.

2* Support for LM83 added to adm1021

3* A new driver for the LM83 and move LM84 support to this new driver
(or any other support combination that would make as much sense as
possible.

I did not take a look at the code yet (just read LM83 and LM84
datasheets (very great BTW) and adm1021 and such detection routine in
sensors-detect) so I can't say what is technically the best choice.
However, I suppose that any solution will be easy, so the choice is
mainly a philosophical one, so to say.

So, what should we do?

-- 
Jean Delvare
http://www.ensicaen.ismra.fr/~delvare/

LM83 support

[Jean Delvare]

> I would recommend extending the adm1021 driver.  It's a relatively 
> simple driver and not nearly as complex as the w83781d.  And the 
> supported chips have fewer differences.

After taking a look at the code, I had come to the opposite conclusion.
The adm1021 driver handles only 2 temperature sensors while the LM83
specificity is that it has 4 temperature sensors. If we are to limit the
LM83 to 2 temperature readings, then it becomes a LM84 that is already
supported. And changing the adm1021 driver code to support 4 temperature
sensors probably makes no sense. So I was planing to write a new driver
for the LM83.

Wether or not existing support should then be moved from adm1021 to lm83
could be discussed later (although my arguments right above make it
quite obvious that older chipsets should be kept supported by adm1021).

> You should probably look at modularizing the extended features
> somewhat. 
>   You can look at my lm85.c driver for an example.

I'll take a look at the code and see if it can change my mind. As said
in a previous post, I never write a driver myself so my opinion may not
be the wisest.

One thing that should be taken into account however is the fact that the
adm1021 driver already supports 8 devices, which is our current max.
Adding a ninth chipset to this driver would introduce (again)
incompatibilities.

-- 
Jean Delvare
http://www.ensicaen.ismra.fr/~delvare/

LM83 support

[Philip Pokorny]

I would recommend extending the adm1021 driver.  It's a relatively 
simple driver and not nearly as complex as the w83781d.  And the 
supported chips have fewer differences.

You should probably look at modularizing the extended features somewhat. 
  You can look at my lm85.c driver for an example.

Also, could you change the adm1023 support a bit?  It puts the 
temperature offset value in with the min, max and current values.  But 
this breaks the print routines in sensors.  The temp offset(s) should be 
in their own sysctl.

:v)

Jean Delvare wrote:
> Hi everyone,
> 
> I will add support to LM83 to lm_sensors, as per request from ZOleg
> Matrix <zoleg@matrixc.ru>. This is perfectly what I need to start with
> real driver coding.
> 
> My question now. The LM83 could be supported by the ADM1021 module,
> which already supports ADM1021, ADM1023, MAX1617, MAX1617A, NE1617,
> NE1617A, TMC10, MC1066 and LM84. The LM84 can easily be seen as a light
> version of the LM83. However, I think I remember we decided to avoid
> adding support for new devices to existing drivers. Or is is just true
> for the w83781d?
> 
> I see three possibilities:
> 
> 1* A new driver for the LM83.
> 
> 2* Support for LM83 added to adm1021
> 
> 3* A new driver for the LM83 and move LM84 support to this new driver
> (or any other support combination that would make as much sense as
> possible.
> 
> I did not take a look at the code yet (just read LM83 and LM84
> datasheets (very great BTW) and adm1021 and such detection routine in
> sensors-detect) so I can't say what is technically the best choice.
> However, I suppose that any solution will be easy, so the choice is
> mainly a philosophical one, so to say.
> 
> So, what should we do?
> 



-- 
Philip Pokorny, Director of Engineering
Tel: 415-358-2635   Fax: 415-358-2646   Toll Free: 888-PENGUIN
PENGUIN COMPUTING, INC.
www.penguincomputing.com

LM83 support

[Mark D. Studebaker]

agreed that a separate driver for lm83 is best, and that
eventually adm1023 support should be moved out of adm1021.

BTW Nat. Semi has always had the best datasheets in the industry.

Also, if you ever have a chance to see Bob Pease on his lecture tour 
don't miss it...

http://www.national.com/rap/

Jean Delvare wrote:
>>I would recommend extending the adm1021 driver.  It's a relatively 
>>simple driver and not nearly as complex as the w83781d.  And the 
>>supported chips have fewer differences.
> 
> 
> After taking a look at the code, I had come to the opposite conclusion.
> The adm1021 driver handles only 2 temperature sensors while the LM83
> specificity is that it has 4 temperature sensors. If we are to limit the
> LM83 to 2 temperature readings, then it becomes a LM84 that is already
> supported. And changing the adm1021 driver code to support 4 temperature
> sensors probably makes no sense. So I was planing to write a new driver
> for the LM83.
> 
> Wether or not existing support should then be moved from adm1021 to lm83
> could be discussed later (although my arguments right above make it
> quite obvious that older chipsets should be kept supported by adm1021).
> 
> 
>>You should probably look at modularizing the extended features
>>somewhat. 
>>  You can look at my lm85.c driver for an example.
> 
> 
> I'll take a look at the code and see if it can change my mind. As said
> in a previous post, I never write a driver myself so my opinion may not
> be the wisest.
> 
> One thing that should be taken into account however is the fact that the
> adm1021 driver already supports 8 devices, which is our current max.
> Adding a ninth chipset to this driver would introduce (again)
> incompatibilities.
>