This bug also affects my system. Everything was working fine with 10.10;
fails with upgrade to 11.04. I'm using a Philips (or NXP) eHome Infrared
Receiver with a Hauppauge RC5 remote.

The short story is that I had to use the devinput driver,
/dev/input/event# device and a new lircd.conf file. Everything is
working 100% with this configuration. There seems to be a bug when using
/dev/lirc0 with the default driver (REMOTE_DRIVER="").

Details of my findings:

Following the various comments that I've read, I modified the
hardware.conf file to remove the driver in REMOTE_DRIVER, and to change
the LOAD_MODULES to false. Note that I've always been using /dev/lirc0
and still using it for this setup. My remote still failed to operate
under this configuration. So, I tried using irrecord. The first line of
the process asks for random button presses, only about 20% of the
buttons on my remote would produce a dot. The second line of the process
then asks for more button presses, now every button would generate a
dot. Finally, it asks for a button name, followed by that button press;
this failed everytime with the message "something went wrong, try
again".

I then tried using the "raw" mode of irrecord "irrecord -f". In this
way, I was able to generate a new lircd.conf file that mostly worked.
I'd say that button presses were about 75% successful. Interestingly, a
button press might fail once but work on a repeat press. It also seemed
to me that button sequences could have been a factor, i.e. "left" button
fails if it came after "9" but it would work if it came after some other
button press. (example only, I didn't dwell on this or document what I
was seeing)

Finally, I read about using the devinput driver. Using this method, I
was able to generate a fully working lircd.conf file, see attached. The
only thing that I don't understand is Jorod's comment above: "The
current problem with trying to use devinput mode is that the lirc
initscript is disabling all the in-kernel decoders, upon which devinput
mode relies to get raw IR scancodes translated into Linux input layer
event keycodes." Do I need to do something different or change
something?

The hex codes in the devinput lircd.conf file are basically the same as
my old working /dev/lirc0 RC5 conf file. The difference is that the
address/header portion is treated as common. (add 0x1780 to all of the
devinput values and they match the RC5 codes). It seems that the raw
codes are a mess; I don't understand the exact timing structure here, I
would have expected the the first half or so of all the buttons to be
the same (address, etc) but that is clearly not the case - I'm not
surprized that this method wasn't 100% successful.

** Attachment added: "Working lircd.conf file using /dev/input/event#"
   
https://bugs.launchpad.net/ubuntu/+source/lirc/+bug/778296/+attachment/2223858/+files/WORKING_dev-input-event%23_lircd.conf

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/778296

Title:
  lirc not working correctly on 2.6.38 kernel

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/lirc/+bug/778296/+subscriptions

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to