I saw something like the problem in bugs #566501 and #570338 since about
February and just spent some time with a more knowledgeable collegue to
investigate.
I have two PCs that I run with the 'testing' distribution.
On one of these PCs, from February thru July, I was able to change the
line in /etc/default/console-setup from:
VERBOSE_OUTPUT="no"
to
VERBOSE_OUTPUT="yes"
to fix the problem on one of my systems.
When I did a upgrade in August, the problem re-occurred, and the final
VT could be any of tty[1-6] after booting.
(My other PC never exhibited this problem in any way.)
We found this thread: http://lkml.org/lkml/2009/11/15/72, talking about
a problem in /usr/bin/consolechars (package: console-tools) that seems
to explain this problem as being in the kernel. We couldn't find any
patch that fixed it.
'/usr/bin/consolechars' is used by '/bin/setupcon' which is used in
'/etc/init.d/console-setup' (both in package: console-setup) which is
where my system seemed to have the problem. I hacked '/bin/setupcon' to
save the vt number at entry (from '/bin/fgconsole') and restored it at
exit (with '/usr/bin/chvt'). This fixed the problem on my PC.
'/usr/bin/consolechars' is also used by /etc/init.d/console-screen.sh
(package: console-tools) which might explain the behavior in bug
#566501.
I found it is also used by '/usr/bin/charset' and
'/usr/bin/unicode_start' (also package: console-tools).
Hope this helps
Paul Higgins
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
