Jason Thorpe <thorpej@wasabisystems.com> writes:

> > However, the SMBus 2.0 protocol does, via a somewhat isapnp-like
> > mechanism. Is this implementation readily extensible or layerable to a
> > SMBus implementaion with direct configuration?
> 
> Yes.  One of the reasons the back-end provides the "bus lock" is to
> allow the back-end to potentially attach an "smbus" as well as an
> "iic".

Okay, good.

> [Note, the SMBus devices that I've encountered while doing this work
> do not have an PnP ID mechanism that I can tell.. but these are mostly
> things like fan controllers and whatnot.]

The mechanism I'm referring to is the one described on page 34 of
http://www.smbus.org/specs/smbus20.pdf. I can't speak to how many
SMBus 2.0 devices are really out there, although modern PC
motherboards (such as my AMD 760MPX+-based box) have their sensors
enumerable this way; see pksrc/sysutils/mbmon for some code that does
the enumeration by prodding the SMBus registers from userspace.

        - Nathan