[mythtv-users] lirc_serial move to serial_ir nightmare in kernel 4.12

[Tom Dexter]

On Tue, Aug 22, 2017 at 10:47 AM, Stephen Worthington
<stephen_agent at jsw.gen.nz> wrote:
> On Tue, 22 Aug 2017 09:43:53 -0400, you wrote:
>>I'm using a simple hardware serial port receiver as well. Not only
>>that, none of the potential issues with the serial port uart are an
>>issue for me as I have the SERIAL_8250 disabled in my kernel
>>altogether. All the lirc stuff is compiled in as well...no modules
>>involved at all. What's ever more perplexing is the fact that mode2
>>works with the default driver and /dev/lirc0 that I'm using, which
>>rules about everything except something breaking the decoding. That
>>would seem to be an issue with the ir-lirc-codec. That's used by the
>>serial_ir but wasn't needed for lirc_serial.
>>
>>There were even several bugs around just that in ir-lirc-codec.c, all
>>of which I've patched into the kernel, and still no luck. Just
>>horrible.
>>
>>Tom
>
> So maybe you need to compile serial_ir as a module, so that you can
> reload it after disabling the uart module from grabbing the port.

Again: I have SERIAL_8250, that is the standard serial port interface
stuff that even *would* grab the port completely disabled in the
kernel, so that's a non issue. If that were a problem I'd never be
getting as close as I am. Again, I'm getting data from mode2 using the
/dev/lirc0 device and the default driver as configured in
/etc/lirc/lirc_options.conf. That would never happen with any uart
issue.

It actually proves that everything up to, but excluding, the actual
decoding of the IR data, is working. I'm going to do some tests with
irrecord to see what it's actually getting. I have a feeling that it's
just not decoding properly from some issue with ir-lirc-codec.

Tom