Hi Stefan, Thanks for your feedback. > i understand the motivation, but could you please explain more in detail > why you decided to add vchiq_instance instead of device reference? I > think vchiq_instance is a more internal structure which should be > avoided in kernel consumers like bcm2835-audio or mmal. I used the vchiq_instance instead of the device reference because in order to get rid of the vchiq_states array (patch 3/3 [1]), I needed another way to access the vchiq_state in the 'handle_to_service' function. So I passed the vchiq_instance to it (I could also have passed the state directly instead of the instance), and this propagated in the caller chain all the way up to 'vchiq_bulk_transmit' and friends which are used in the bcm2835-audio consumer. Please let me know if you see a better way of doing this :) Thanks, Adrien [1] https://lore.kernel.org/all/20220502183045.206519-4-athierry@xxxxxxxxxx/
