On Wed, 2005-03-09 at 01:34 -0500, Dmitry Torokhov wrote: > On Wednesday 09 March 2005 00:59, Adam Belay wrote: > > > > I agree that individual drivers should track their own usage. I was > > just wondering if we could provide some utility functions that would > > help with this. > > > > Did we remove pm_access() and friends just a tad early? > Perhaps so... I'm not sure if pm_access() was correct. For example, what if the power management policy decided to turn off a device in the middle of device access? Just because pm_access() is called, it doesn't mean the device state is locked to active. Wouldn't this be a race condition? In general, I think locking for this sort of thing could get very tricky. Also, in our new design, I think something like pm_access() should be called primarily at the class level. Comments? Thanks, Adam
