Hi, Regarding the multiple instances of .so files getting dynamically-linked to various processes, There might be some doc somewhere which says that. My conclusions were from studying the "framework" code and experimenting and observing the results on a device with my "custom" HAL in place.
Here is a simple experiment you can try on your own: 1. Declare a Global variable in sensors.c 2. Update the global variable in *control_activate*. 3. Log the value of the global variable in *data_poll*. You can see that the update made in control_activate is NOT reflected in * data_poll*. __________________________ Regarding the necessity for any locking mechanism (mutex?) I do not foresee the need for any locks. Though multiple instances of *sensors.so* exist and each may try to read the same input file, Each *sensors.so* does so with it own handle to the file. This means the kernel already is managing multiple access using multiple handles & you need not worry about it. __________________________ ...aaand yes. All this changes with gingerbread. :) regards CVS -- unsubscribe: [email protected] website: http://groups.google.com/group/android-porting
