Experimental Physics and
Industrial Control System

"Mark Rivers" <[email protected]> · Mon, 29 Nov 2010 20:29:57 -0600

Hi David,

A very simple asyn driver based on asynPortDriver would do the trick.

The constructor would set the asynFloat64Mask bit in the interface mask.  You would implement the readFloat64 function, which would open and read from /proc/acpi/thermal_zone/THRM, parse the temperature and return that as the value.  That would work with periodically scanned records.

The whole thing could probably be about 30 lines of code.

If you want to get more sophisticated you could set a polling frequency in the configure command, which you pass to the constructor.  It would then have a polling thread that reads, parses, calls setDoubleParam() and callParamCallbacks().  That way the record only processes when the temperature changes.  

You could also support other parameters (power, fan, etc.) by using the drvInfo field to specify what parameter is to be returned.

Mark

________________________________

From: Eric Norum [mailto:[email protected]]
Sent: Mon 11/29/2010 7:30 PM
To: David Morris
Cc: Mark Rivers; Steve Shoaf
Subject: Re: Monitoring CPU temperature for small IOC

Sounds like a fairly simple 'devEPICS/ASYN' driver would do the trick.
I think that Steve Shoaf at the APS may have something similar for controlling/reading digital I/O points and load average on an embedded linux system.

On Nov 29, 2010, at 5:20 PM, David Morris wrote:

> Hello all
>  I have a Linux based IOC running on a dedicated small hardware platform (Atom CPU in Mini-ITX form factor) We would like this to have no moving parts, which means monitoring the CPU state becomes critical on hot summer days. Is there support for monitoring CPU temperatures using the /proc/acpi service? It seems like this would be something that devIocStats could provide, but perhaps it is too system dependent.
>
>  Is there an easy way to collect information from the IOC host system and place it into PVs? Regular reading of /proc/acpi/thermal_zone/THRM will give an updated CPU core temperature, but it is not clear to me how to place this into a PV.
>
>  Thanks in advance.
>
> David
>
> --
>
> David Morris                              TRIUMF
> Systems Designer                          4004 Wesbrook Mall
> Ph  604-222-7450                          Vancouver, B.C.
> Fax 604-222-7307                          V6T 2A3, Canada
>
>
>

--
Eric Norum
[email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System