module integration issue: keyboard driver _still_ not working after port from 2.4.27
hi,
after some thought, i decided on friday to port all of the 2.4.27 code over to 2.6.11.7.
bearing in mind that reading from /dev/input/evdev0 worked fine on 2.4.27 - and now also works fine under 2.6.11.7 - the exact same problem occurs on 2.6.11.7 as occurred under 2.4.27 - hanging of the linux kernel whilst providing absolutely zero crash/debugging information responses whatsoever.
this device does _not_ have and cannot have a standard PC keyboard attached to it.
in order to be able to debug what is going on, i have enabled a dummy/virtual serial console, all is well so far.
in order to test the screen, i have written, enabled, tested, confirmed as reasonably working, a framebuffer driver (which i would like to make the console framebuffer - eventually - when the serial console is disabled and no longer needed - so i am enabling Framebuffer Console support AS WELL as serial console support)
now i load the keyboard event module... splat - absolutely no response:
complete lock-up.
okay, let's try that again.
reboot.
install the keyboard event module.
install evdev module.
run a program (multilog) which is a bit like tail -f except it outputs to a logging file rather than to stdout on /dev/input/evdev0 and press a few buttons.
do a hexdump -C on the output from multilog - GREAT!
oh...keeyyy...
so, individually, the components work fine.
put them together, and something goes badly wrong.
has _anyone_ else tried running two console drivers at the same time?
should it be expected to work?
any assistance / guidance greatly appreciated.
l.
----- Forwarded message from Luke Kenneth Casson Leighton <[email protected]> -----
From: Luke Kenneth Casson Leighton <[email protected]>
To: [email protected]
Subject: [2.4.27 arm skyminder] writing new keyboard handler - help!
hi, please respond cc to me because i am not on the lkml.
for a new arm linux embedded device called a skyminder, i'm responsible for getting all the drivers working. ha ha.
the success of this device presents the linux community with an opportunity to own a linux-based mobile phone (even if it's not a very small phone - 8cm x 10cm x 1cm) - and it has a GPS module in it, as well.
i'm endeavouring to adapt various bits of code to create a keyboard driver. they've adopted 2.4.27 and are too far down the line to move to 2.6 - yet.
i particularly want to avoid - if i can - compiling this keyboard driver under development into the kernel (even though it's the primary keyboard) because downloading 600k over a serial link into flash ram isn't a) funny b) a good idea c) slows development time down.
with that in mind, so far, i have:
- cut/paste pc_keyb.c just like everyone else has (in celps_keyb.c, c711x_keyb.c, dummy_keyb.c etc.) to create k_translate, k_unexpected_up, k_setkeycode and k_getkeycode routines.
- cut/paste usbkbd.c and adapted it to successfully call input_report_key on a key press and key release.
then, on installation of module input, keybdev and sky_buttons, i happily get debug messages indicating key presses (keycode 31 indicating 's') ... but no actual key events appear down my serial console.
so, my question is: does anyone know off the top of their heads what i may have missed out that causes the keybdev event handler to _not_ actually stuff keys out?
am i... like... missing something really obvious, given that the console has been set to "serial"?
where should i look to, to find the keys being outputted, if they're going anywhere?
help, help, gloop.
cheers,
l.
Signature · 10 lines
-- <a href="http://lkcl.net">http://lkcl.net</a> -- ----- End forwarded message ----- -- -- <a href="http://lkcl.net">http://lkcl.net</a> --
okay - i found this:
http://seclists.org/lists/linux-kernel/2000/Mar/0093.html
it talks about "we must force people to choose between vgacon and fbcon".
when that forcing decision was taken, was dual fbcon-and-serial-console usage taken into account?
booting with ..... console=ttyCL1 fbcon=vc:1 and then later on doing modprobe fbcon has exactly the same consequences:fbcon_takeover is called, the fbcon=vc:1 parameters appear to be ignored.... greeeeaat.
how the _heck_ am i gonna over-ride the fb_console "setup" paramaters on a module load, to stop this happening???
i know - HACK TIME!!!
muhahahahah.
l.
Quoted 12 lines · Luke Kenneth Casson Leighton
in order to be able to debug what is going on, i have enabled a dummy/virtual serial console, all is well so far.
in order to test the screen, i have written, enabled, tested, confirmed as reasonably working, a framebuffer driver (which i would like to make the console framebuffer - eventually - when the serial console is disabled and no longer needed - so i am enabling Framebuffer Console support AS WELL as serial console support)
now i load the keyboard event module... splat - absolutely no response:
complete lock-up.