On Sun, Jun 13, 2010 at 4:29 PM, David Härdeman <david@xxxxxxxxxxx> wrote: > On Wed, Jun 09, 2010 at 09:25:44PM -0400, Jarod Wilson wrote: >> On Wed, Jun 9, 2010 at 2:15 PM, Jarod Wilson <jarod@xxxxxxxxxx> wrote: >> > On Wed, Jun 09, 2010 at 07:56:21PM +0200, David Härdeman wrote: >> >> On Wed, Jun 09, 2010 at 09:29:08AM -0400, Jarod Wilson wrote: >> ... >> >> > So this definitely negatively impacts my ir-core-to-lirc_dev >> >> > (ir-lirc-codec.c) bridge driver, as it was doing the lirc_dev device >> >> > registration work in its register function. However, if (after your >> >> > patchset) we add a new pair of callbacks replacing raw_register and >> >> > raw_unregister, which are optional, that work could be done there instead, >> >> > so I don't think this is an insurmountable obstacle for the lirc bits. >> >> >> >> While I'm not sure exactly what callbacks you're suggesting, >> > >> > Essentially: >> > >> > .setup_other_crap >> > .tear_down_other_crap >> > >> > ...which in the ir-lirc-codec case, register ir-lirc-codec for a specific >> > hardware receiver as an lirc_dev client, and conversely, tear it down. >> > >> >> it still >> >> sounds like the callbacks would have the exact same problems that the >> >> current code has (i.e. the decoder will be blissfully unaware of >> >> hardware which exists before the decoder is loaded). Right? >> > >> > In my head, this was going to work out, but you're correct, I still have >> > the exact same problem -- its not in ir_raw_handler_list yet when >> > ir_raw_event_register runs, and thus the callback never fires, so lirc_dev >> > never actually gets wired up to ir-lirc-codec. It now knows about the lirc >> > decoder, but its completely useless. Narf. >> >> And now I have it working atop your patches. Its a bit of a nasty-ish >> hack, at least for the lirc case, but its working, even in the case >> where the decoder drivers aren't actually loaded until after the >> device driver. I've added one extra param to each protocol-specific >> struct in ir-core-priv.h (bool initialized) and hooked into the >> protocol-specific decode functions to both determine whether a >> protocol should be enabled or disabled by default, and to run any >> additionally required initialization (such as in the ir-lirc-codec >> case). >> >> So initially, mceusb comes up with all decoders enabled. Then when ir >> comes in, every protocol-specific decoder fires. Each of them check >> for whether or not they've been fully initialized, and if not, we >> check the loaded keymap, and if it doesn't match, we disable that >> decoder (bringing back the "disable protocol handlers we don't need" >> functionality that disappeared w/this patchset). In the lirc case, we >> actually do all the work needed to wire up the connection over to >> lirc_dev. >> >> This works perfectly fine for all the in-kernel decoders, but has one >> minor shortcoming for ir-lirc-codec, in that /dev/lirc0 won't actually >> exist until the first incoming ir signal is seen. lircd can handle >> this case just fine, it'll wait for /dev/lirc0 to show up, but it >> doesn't come up fast enough to catch and decode the very first >> incoming ir signal. Subsequent ones work perfectly fine though. This >> need to initialize the link via incoming ir is a bit problematic if >> you're using a device for transmit-only (e.g., and mceusb device >> hooked to a mythtv backend in the closet or something), as there would >> be a strong possibility of /dev/lirc0 never getting hooked up. I can >> think of a few workarounds, but none are particularly clean and/or >> automagic. >> >> Not sure how palatable it is, but here it is: > > I think it sounds pretty awful :) So my suspicion wrt palatability was correct. :) > I have another suggestion, let's keep the client register/unregister > callbacks for decoders (but add a comment that they're only used for > lirc). Then teach drivers/media/IR/ir-raw-event.c to keep track of the > raw clients so that it can pass all pre-existing clients to newly added > decoders. > > I'll post two patches (compile tested only) in a few seconds to show > what I mean. Consider them now runtime tested as well. They appear to do the trick, the lirc bridge comes up just fine, even when ir-lirc-codec isn't loaded until after mceusb. *Much* better implementation than my ugly trick. I'll ack your patches and submit a series on top of them for lirc support, hopefully this evening (in addition to a few other fixes that aren't dependent on any of them). -- Jarod Wilson jarod@xxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
