Talk:How to make use of IrDA
The "wrong chip version ff" is a real issue, and having it mentioned makes sure that google searches turn it up with a solution. Tonko 01:41, 8 Oct 2005 (CEST)
The present text confused me (as you now have noticed). Maybe it should better reflect that the error is incorrect ("wrong chip version ff" instead of "FIR mode not enabled" or whatever). Would that be acceptable?
Paul Bolle 02:05, 8 Oct 2005 (CEST)
If you think you can explain it better, go right ahead. The IrDA document could use some major cleaning.
As long as we still point out that the only error in syslog is "Wrong chip version ff" when trying to load the FIR module without first activating the PnP device. It would be nice if the ISA-PNP patch to the nsc-ircc driver got accepted upstream, so we dont have to do these hacks anymore. Tonko 03:58, 8 Oct 2005 (CEST)
Is there a reason why you deleted the TODO entry for linux 2.6 kernel config for SIR? I would like to readd it, but like to hear your reason to remove it first.
Paul Bolle 10:23, 14 Oct 2005 (CEST)
T23 debianized kernel with probs
Found another crazy thing here. T23 2648-2GG, with 2.6.13 and 2.6.14 (Debianized Source). Clear init of nsc-ircc and dongle, but the dongle stays offline (dark and blind). Booted with Knoppix or Bart-PE Windows give correct function of the whole device, viewed with a digicam shows some blinks at init. But not with plain 2.6.13 an 2.6.14 from Debian. Eric
since the nsc-ircc patch isn't in 2.6.15, I tried to go the setpnp route, but I don't even get to see /proc/bus/pnp even though I enabled the support for it in the kernel config (once I found out that I need to enable ISA support to even get the option). Has anyone successfully used setpnp to enable the IRDA port or is that just some sort of urban legend? ;)
--spiney 19:12, 4 Jan 2006 (CET)
Make sure you enable pnp-bios support specifically when compiling the kernel, just enabling pnp support is not enough. I must however admit that I have not done this, since my ancient TP 770 running a 2.4 kernel, which was at least 5 years ago.
I just looked at a 2.6 menuconfig myself, and could not find the option, but looking at my .config file I can see the CONFIG_PNPBIOS option, so you might just want to edit the config file directly, enable the PNPBIOS option, and run make oldconfig.
--Tonko 19:38, 4 Jan 2006 (CET)
I do have the pnp-bios support enabled (the option is depending on ISA among others, checked the Kconfig file in the source tree to find it), and also the proc-Interface option, but still no go... maybe it's not compatible with CONFIG_PNPACPI? Oh well, nevermind, going back to the patch, that works with less hassles, I just hope it'll end up in the vanilla kernel eventually. Still thanks for the help.
--spiney 20:22, 4 Jan 2006 (CET)
Ok, on further investigation (not giving up that easily ;) it really seems that CONFIG_PNPACPI is disabling PnPBIOS, at least dmesg showed something along the lines of "PnPBIOS: disabled by PnPACPI". But, after recompiling without PnPACPI, I still can't get it to work. I can use
lspnp just fine, but the nsc-ircc module won't load:
t43p:~# lspnp 12 12 IBM0071 IBM Thinkpad infrared port t43p:~# setpnp 12 io 0x2f8 irq 3 t43p:~# lspnp -v 12 12 IBM0071 IBM Thinkpad infrared port dma 3 io 0x02f8-0x02ff irq 3 t43p:~# modprobe nsc-ircc io=0x2f8 irq=3 dongle_id=0x09 FATAL: Error inserting nsc_ircc (/lib/modules/2.6.15/kernel/drivers/net/irda/nsc-ircc.ko): No such device
Hmm, I'm getting somewhat annoyed... any ideas?
--spiney 20:58, 4 Jan 2006 (CET)
have you tried just doing
setpnp 12 on
--Tonko 21:05, 4 Jan 2006 (CET)
Yes, that was the first thing I tried, didn't help. BTW, I just fiddled around with lirc a bit (never tried it before), and the lirc_sir module works just fine:
lirc_sir: I/O port 0x02f8, IRQ 3. lirc_sir: Installed.
and I was able to get it to control xmms.
setserial doesn't claim the device either:
t43p:~# setserial -ag /dev/ttyS0 /dev/ttyS0, Line 0, UART: unknown, Port: 0x0000, IRQ: 0 Baud_base: 921600, close_delay: 50, divisor: 0 closing_wait: 3000 Flags: spd_normal skip_test
I'm really out of ideas.
--spiney 21:20, 4 Jan 2006 (CET)
Hmm, it should not be ttyS0, but ttyS1 which might be stepping on the resources ttyS0 is IO 0x3f8 and IRQ 4, while ttyS1 is IO 0x2f8 and IRQ 3.
ttyS0 is your integrated serial port, which even ThinkPads without physical serial port have.
--Tonko 21:29, 4 Jan 2006 (CET)
Sorry, I cut'n'paste the wrong device since I checked them both, the output for ttyS1 is identical (apart from "/dev/ttyS1, Line 1" of course).
--spiney 21:36, 4 Jan 2006 (CET)
No idea what might be wrong, perhaps it does not work on more recent machines, or kernels.
As I said, it worked for me a long time ago on my 770X with something like this:
/bin/setserial /dev/ttyS1 uart none port 0 irq 0 setpnp 12 io 0x2f8 irq 3 setpnp 12 on modprobe nsc-ircc dongle_id=0x09 io=0x2f8 irq=3
What is the error you get from nsc-ircc in syslog? the "Wrong chip version ff" error?
--Tonko 22:22, 4 Jan 2006 (CET)
No errors are recorded at all (dmesg, syslog), just the message above on stderr from
--spiney 09:27, 5 Jan 2006 (CET)
BTW, is it just me or does the nsc-ircc patch only work if CONFIG_PNPACPI is set in the kernel config? Otherwise the module loads fine, (but without stating in the kernel log that it found the dongle) and using IRDA does not work, irattach returns "irattach: ioctl(SIOCGIFFLAGS): No such device".
--spiney 14:37, 12 Jan 2006 (CET)