On Thu, Sep 01, 2011 at 12:09:14AM -0400, Paul Walmsley wrote: > On Wed, 31 Aug 2011, Guenter Roeck wrote: > > > On Wed, Aug 31, 2011 at 08:36:43PM -0400, Paul Walmsley wrote: > > > Hi > > > > > > Some comments. > > > > > > On Wed, 31 Aug 2011, Keerthy wrote: > > > > > [ ... ] > > > > > > > +} > > > > + > > > > +/* Sysfs hook functions */ > > > > > > These should be conditionally compiled out if sysfs isn't compiled in. > > > > > The whole point of the hwmon subsystem is to expose hardware monitoring information > > to userland using sysfs. hwmon without sysfs doesn't make sense. > > > > So, if anything, it might make sense to disable the entire hwmon tree if sysfs is disabled. > > But please no conditionals in the code. > > Hmm. This IP block is more than just a sensor. It also can interrupt the > CPU and/or trigger a GPIO line (to shut down the chip) if the chip > temperature crosses some thresholds. On some OMAPs, the thresholds are > fixed; on others, they are software-programmable. That functionality > shouldn't require sysfs; it's almost closer to an x86 MCE. > > So based on your comments, it sounds like we should move that part of the > code to a different driver, and just leave the basic software thermal > monitoring here? > Good point. This definitely requires some thought. hwmon is meant to be hw monitoring, as the name says, not thermal management. Maybe this entire driver should be a thermal driver instead ? Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
