Re: Since chvt works, the problem must be the keymap

On Sun, Apr 29, 2007 at 02:51:13PM EDT, Florian Kulzer wrote:
On Sun, Apr 29, 2007 at 14:10:14 -0400, cga2000 wrote:
On Sun, Apr 29, 2007 at 12:10:22PM EDT, Florian Kulzer wrote:

[...]

It would be interesting to see which keycodes and keysyms are reported
if you run "xev", press (and hold) both CTRL and ALT, and then press F1,
F2, etc. Does xev really display the keycodes for the Fn keys and the
keysyms "XF86_Switch_VT_n"? Are the hexadecimal keysym values the same
as the ones that you get with "grep VT /usr/share/X11/XKeysymDB"?

Another way of looking at it is that _unless I am totally misunderstand
how these things work_ .. the CTRL+Alt.. sequence invokes "X server"
code and the "disconnection" from the VT as Andrew nicely put it, is
eventually processed (in part at least) by your card's driver .. whereas
with "chvt" it is code that lives in the kernel that is invoked.

Hence a buggy video card driver could cause CTRL+Alt.. to fail on a
system where chvt works.. Regardless of keyboard mapping ..

The "logic" :-) behind this embarrassing guesswork of mine is that I
have experienced this kind of problem with embedded chips (together with
a slew of other issues) in cases where the driver was clearly described
as immature in the (then) xfree86 doc.

That's why I suggested .. possibly in another recent thread on a similar
subject

It probably was the same thread. One of the morons from debianhelp.org
has gratuitously changed the subject line in the meantime, which does
not really contribute to making things any clearer.

.. was it an nvidia card then? .. switching to the VESA driver
just to see if it makes any difference.

Naturally, checking whether you are running the latest version of the
driver and taking a peek at it's change log is matter or course.

Let me know if this makes sense.

I think we certainly cannot rule out anything at the moment. I seem to
recall, though, that some people who reported the VT switching problem
in the past did try the vesa driver and found that it did not help.

Not even sure it did in my case .. unfortunately this was on machines I
no longer own so I can't just run a quick test myself. My very vague
hunch is that this may turn out to be something more involved than a
problem with keyboard mapping. And then again it might be. It's
difficult to tell without actually having the machine to fiddle with.

My feeling is that a problem with the graphics driver would either
lead to a lock-up of the system, a crash of X, or at least to error
messages in the Xorg log. I don't remember if we ever suggested that
people check the log.

It is certainly annoying that this is a persistent problem for a small
percentage of users and that we cannot figure out what causes it.

Maybe someone closer to X development might shed some light on how this
works. I don't remember going to a specialized list or newsgroup when I
ran into a similar problem myself. To make things more complicated, the
machines shared a kvm with a couple of others and since I mostly ssh'd
to them, I have a feeling I just decided this wasn't worth the trouble.

Thanks,
cga


--
To UNSUBSCRIBE, email to debian-user-REQUEST@xxxxxxxxxxxxxxxx
with a subject of "unsubscribe". Trouble? Contact listmaster@xxxxxxxxxxxxxxxx

Relevant Pages

  • Re: Since chvt works, the problem must be the keymap
    ... etc. Does xev really display the keycodes for the Fn keys and the ... with "chvt" it is code that lives in the kernel that is invoked. ... Hence a buggy video card driver could cause CTRL+Alt.. ... To UNSUBSCRIBE, email to debian-user-REQUEST@xxxxxxxxxxxxxxxx ...
    (Debian-User)
  • Re: Since chvt works, the problem must be the keymap
    ... etc. Does xev really display the keycodes for the Fn keys and the ... with "chvt" it is code that lives in the kernel that is invoked. ... Hence a buggy video card driver could cause CTRL+Alt.. ...
    (Debian-User)
  • Re: xev can only see 2 of 20 btn mouse [Re: map a mouse button to a keyboard key?]
    ... > i'm able to get one of the 18 extra buttons to be seen by xev with the ... > moused per driver. ... mouse protocols are bits to be variously interpreted in three to five ... byte packets. ...
    (comp.unix.bsd.freebsd.misc)
  • Re: mfi freebsd7
    ... Failed to open driver node /dev/megadev0 ... you might have more luck with sysutils/linux-megacli since that explicitly mentions LSI MegaRAID SAS, I wish they would make the util open source though. ... A RAID1 config shoudl suit a busy webserver well. ... To unsubscribe, ...
    (freebsd-questions)
  • missing NICs in FC2
    ... With most machines that I load FC2 on, ... I'll say that this is a custom install with only eth0 being ... I'm trying to understand if the kernel is not loading the driver until ... divert: allocating divert_blk for eth1 ...
    (Fedora)