[PATCH][2.5] Support for W83627THF sensor chip (Was Re: [2.5] Help with ICH5 SMBus and W83627THF)

Mark D. Studebaker mds at paradyne.com
Tue Jun 17 03:50:09 CEST 2003

If your patch is all that's required and it tests out well (have you tested it?)
then sure, why not. It assumes that a w83627thf is exactly the same
as a w83627hf. It's when the additional patches come in
(if kind == w83627thf.....) then it's making a bad situation worse in
my opinion.

So my preference is to limit the additions to w83781d.

There's no plans to split the existing w83781d driver.
We don't rename existing drivers because of CVS limitations and
wanting to minimize changes for our users.

So I'd feel better about your patch after seeing some test results.
And as I said before, you will get the best results by using
the new 627hf driver, either after porting it yourself or
hoping somebody else does.


Martin Schlemmer wrote:
> On Tue, 2003-06-17 at 03:06, Mark D. Studebaker wrote:
>>About a week ago I worked with
>>  Matthias Hentges <matthias at hentges.net>
>>to determine the device ID for this chip, and he supplied a patch which
>>I worked into CVS, into the w83627hf driver and into sensors-detect.
>>It was checked in last week.
>>This driver is designed for Super I/O chips and includes detection
>>and activation. It uses ISA accesses.
> Ok, checked out CVS.
>>I would rather not keep adding to the w83781d cruft, especially
>>for Super I/O chips. Not only is the
>>driver quite unwieldy already, but people have had lots of trouble
>>with the Winbond Super I/O chips because they often aren't
>>initialzed by the bios so the w83781d driver can't find them.
>>ISA accesses are also much more reliable.
>>Please test the w83627hf driver in CVS and give us some feedback.
> Well, it is not yet ported to 2.5, and the state this box is
> in (NPTL, etc), I _cannot_ use a 2.4 kernel.  I also do not know
> if I currently have the time to port it to 2.5.
> Finally, what is the plan .. split the w83781d into smaller
> drivers for each class of chip ?  If so, don't you still
> need i2c support for some w836* chips?  Won't then also
> be better to call the drivers w837xx.c and w836xx.c ?
> Regards,

More information about the lm-sensors mailing list