Hi Alain, >> From a high level, this looks good for me although I agree, this is an >> order of magnitude bigger in terms of scope. Can you suggest perhaps >> an interactive way to deliver this over a period of time, perhaps >> prioritizing the BT_DEBUG kernel messages first? :) > > I am always in favor of increasing the ability to debug things, but we need to do this in a clean fashion and not some short term hacks (since they will come back and haunt us). I like to get some review on my idea first. > > What we could do is work on the BT_DBG etc infrastructure to allow switching when dynamic_debug is not available. Then you would use some debugfs toggle in /sys/kernel/debug/bluetooth since that is no stable API for us (and of course the clear understanding that this toggle is temporary). I just posted a patch to allow a short term solution. It is similar to the patch you pointed me to with the difference that it is a debugfs entry and limited to the case when dynamic debug is disabled. Is this something that would work until we get a more complete solution worked out and tested? Regards Marcel
