Re: 2.6.5 Sensors & USB problems

[Jean Delvare <khali@linux-fr.org>]

> Seems that the w83781d driver no longer detects whatever the sensor
> chip on the P4PE is, but I see there is now a new driver called asb100
> which does work. An old conflicting lm_sensors install was breaking
> the sensors-detect script, but once resolved it nicely detected the
> asb100.

The hardware monitoring chip on the P4PE is an Asus ASB100 "Bach". It
used to be somewhat supported by the w83781d driver (see as an
"as99127f" kind) but then a new, dedicated driver was developed by Mark
M. Hoffman, which works better. For this reason, support was dropped
from the w83781d, which explains why it suddenly stopped working.

I agree that this should have been advertised a little more, and I am
adding information about this on our 2.6 kernel dedicated page at the
moment.

> Can anyone explain, however, why my i2c bus showed up as number 0
> under linux <= 2.6.4, and now always as number 1 under linux 2.6.5?
> The is no number 0 any more.

The bus number allocation scheme is such that once a number has been
used once (since the machine last booted) it will not be used again.
This is admittedly not ideal and should be fixed. I suspect that the fix
isn't trivial because the current structures would make the new scheme
have a poor algorithmic complexity (O(2) maybe), but I haven't checked
yet. Greg, can you confirm?

This isn't critical (unless you cycle your i2c adapter drivers a great
number of times, but nodoby does this) which is why no attempt has been
made to fix it yet. Anyone with a nice fix is welcome however ;)

Thanks.

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