2018-05-07, 12:39
(2018-05-07, 02:45)mikeb8591 Wrote: Ok, the results of all the tests are here. It should be noted that I also see occasional, seemingly random key activity several seconds after I've stopped my keypress. Some of these are captured in the .txt file, inside this .zip. Each test captures one "normal" press of the OK button.Thanks a lot, these logs are very helpful! I can clearly see that the IR receiver is behaving very odd with the short timeout. It's not sending the messages we'd expect and also not at the times where we'd expect them - which messes up everything.
I also observed that the red led on the IR receiver is basically on solid for the duration of my keypress in the fixed_timeout=1 case (and with 0417 and earlier) but flashes rapidly in all the other test scenarios.
http://www.awen.com/~mburgett/kodi/ir-de..._tests.zip
This remote has performed flawlessly up til now, and indeed still works flawlessly with 0417 and earlier builds, so I'm not inclined to agree that I have worn out buttons.
I'll discuss this with the IR kernel maintainer, it could be that we need to disable timeout configuration for this specific IR receiver.
As you mentioned that running "ir-ctl -t 100000" worked on newer builds there's one more interesting thing to find out - if only this timeout works or if some other timeouts work, too and we just happen to set some timeout value it doesn't like.
So, could you play a bit with ir-ctl -t values on any build newer than 0418? The value set by the kernel is 25650, so checking with 30000, 40000, ... 100000 would be a good start. If some of these work and some others don't it would also be interesting to test nearby values (eg if 30000 works and 40000 doesn't try 35000, 38000, 39000, 41000, 42000, 45000). Just run ir-ctl -t while kodi is running and watch out for the typical symptoms like "double-clicks" and spurious button-presses. The ir-ctl -t changes will take effect immediately but the very first button press after the change could still be messed up due to remains of the previous configuration.
Edit: testing with 20000 and 25000 would also be interesting.
BTW: the changes in the LED behaviour on the receiver are about to be expected, so this is nothing to worry about.
so long,
Hias