>From: Dhanotia, Abhishek [mailto:abhishek.dhano...@intel.com] > >Hi Timo, > >We are working on wakeup reduction for accelerometer during >idle. We found an issue in the sensorfw that on screen blank, >sensor framework only disables reading the driver >/sys/class/input/eventX. The accelerometer adaptor does not >have a mechanism to actually set polling to "0" in the sysfs >on standby, hence there are still wakeups during idle. > >We are working on a patch to disable polling in driver by >implementing standby() and resume() functions in the >accelerometer adaptor which would set the polling to 0 on >standby and resume it to the last polling value using the >evaluateIntervalRequest() function in the accelerometer >adaptor. Do you think this can be integrated into the >accelerometer adaptor?
Hi Abhishek, The expectation has been that the accelerometer chip should power down once the driver handle is released. Seems like we can't trust this to be the behavior everywhere, so I think resetting the poll parameters would be a good idea (and generally good behavior anyway). Perhaps we can make it more generic than just accelerometerAdaptor: standby()/stop() in DeviceAdaptor could set interval to defaultInterval, assuming default is always the most power consumption friendly value available. Then resume()/start() would set the correct value by running the evaluateIntervalRequest() again. This way every adaptor would follow the same rules. How does that sound? // Timo _______________________________________________ MeeGo-dev mailing list MeeGo-dev@meego.com http://lists.meego.com/listinfo/meego-dev