Jarod Wilson wrote: > On Monday 21 February 2005 05:05, Thomas B�rkel wrote: > >>Jarod Wilson wrote: >> >>>I just did a clean install today. There are a few problems related to >>>udev that haven't yet been properly solved, but my remote is functioning >>>just fine, using the ATrpms lirc packages and the ATrpms Myth packages. >> >>What are the udev problems? > > > There are some race conditions. With prior non-udev systems, drivers were > loaded when daemons started up and tried to access their associated device > nodes. > > For example, when lircd starts up, it tries to access /dev/lirc. On a > non-udev > system (i.e., FC2), /dev/lirc was a symlink to /dev/lirc0. You have a > modprobe.conf entry that says 'alias char-major-61 lirc_i2c', which since its > the first and only line, is analogous to char-major-61-0, character device > with major number 61, minor number 0, aka /dev/lirc0 (look at 'll /dev/lirc0' > output). The alias line told the system that when accessing lirc0, we > actually need lirc_i2c loaded, as lirc0 is an i2c-based IR receiver. > > All this falls down if the device node isn't already there. Since there's no > device node to be accessed, the system doesn't recognize the correspondence > between lirc0 and lirc_i2c, and the module doesn't get loaded, and lircd > tanks. Similar thing goes for the ivtv driver and it not loading/mythbackend > dying when loaded at boot time.
I use Debian rather than RH/Fedora, so perhaps this isn't possible for you. But why mess with all this aliasing stuff to load modules on demand? I thought that was for seldom-used drivers to save some memory, etc. That obviously isn't the case for a lirc driver on a mythtv box. On Debian, I just add lirc_i2c to /etc/modules. It is then unconditionally loaded early in the boot sequence. I would be surprised if RH/Fedora did not have an equivalent mechanism. _______________________________________________ mythtv-users mailing list [email protected] http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
