Hello,
I'm using a cubox-i, which has a gpio remote receiver built in. The remote
I'm using is a universal remote, using an RC-6 protocol. This is using
kernel 3.14.14 (from geexbox)
What happens is that if you press key1, key1, key2 with spaces between the
presses that are larger than IR_KEYPRESS_TIMEOUT but shorter than the
repeat delay, then the driver often delivers four down/up pairs of events:
key1, key1, key1, key2.
Similarly, key1, key2, key2 gives key1, key2, key2, key2
and key1, key2, key1 gives key1, key2, key2, key1
Strangely, key1, key1, key1 gives only the three expected events.
If I set the repeat delay to be equal to the IR_KEYPRESS_TIMEOUT
(using ir-keytable --delay) then the problem goes away (or is hidden).
I've been looking though the relevant kernel code, and while I don't think
I fully understand all the subtleties, one thing I noticed that seems
odd is that in ir_raw_event_store_edge in ir-raw.c, the repeat delay is
used to determine if an edge is part of a new event or not. It seems to me
that this might be way too long. Shouldn't it be something more like
IR_KEYPRESS_TIMEOUT ? Or even shorter - like just longer than the longest
possible gap between edges in a code?
I tried changing the "delay" value in ir_raw_event_store_edge to match the
IR_KEYPRESS_TIMEOUT, and it does seem to resolve the problem for me.
Carl
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
