On Thu, Dec 07, 2006 at 11:27:33AM -0500, Woody Suwalski wrote:
> However, much better would be to create a symlink from /dev/temp to 
> /dev/temperature.
> 
> Reason: set_therm and fan_ctrl programs use /dev/temperature for setting the 
> thermostat configuration.
> fand is taking advantage of 0.5 degree temperature resolution - using 
> propriatory ioctl call.
> All of them need a non-standard /dev/temp interface. As such it should not be 
> attempted if the correct Netwinder driver is not installed.
> Hence since it is hardware-specific, it may be better to keep it as a 
> specific /dev/ interface... (rename to nwtemp????)
> 
> So the quick fix will be to add /dev/temp to the man page. The proper - to 
> add /dev/temperature symlink to udev rules...

It sounds like you're saying that /dev/temp might, sometimes, not be
what I want to use. On a default etch install it has the same
major/minor as recommended in the manpage. So, by default it *is* what
I want to use.

Can you explain in more detail what the conditions are that would
cause /dev/temp to not be what I want? The only possibility I can see
is if there's another driver that I could load that would also use the
same major/minor but would provide an incompatible interface. In that
case I can understand why you might be using the device name to
differentiate, but that would only work if you're relying on a udev
environment that only exposes /dev entries if the right driver is
there. In other words, if /dev/temperature is there and the wrong
driver gets loaded, /dev/temperature will also be providing the wrong
interface.

If that's the case, it seems like the *right* answer is to make udev
create something like your suggested /dev/nwtemp with a unique
major/minor.

I don't see how a symlink is of any value - by doing a symlink we're
assuming /dev/temp is correct. But if its correct, why not use it
directly?

-- 
dann frazier | HP Open Source and Linux Organization


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to