On Fri, 18 Aug 2006, Philippe Gerum wrote:
On Wed, 2006-08-16 at 16:14 +0200, Klaas Gadeyne wrote:
I just wanted to drop a note that I did some refactoring on the
drivers directory (16550A->serial) and the config menus. If anything is
broken, shoot me.
I don't think this is related, but 2 students of mine got stuck on a linker
error this morning trying to compile xenomai-trunk on a i386 target
(2.4.33 kernel with the adeos-ipipe-2.4.32-i386-1.2-06.patch (applied
with a few minor hunks), gcc 3.3.6 from Debian).
The "quick-fix" solution was to disable the magic_sysrq_key option
from the kernel. I first thought the problem was due to a missing
dependency in the the kernel and not at all xenomai-related (and since
the quick hack worked, they didn't search any further), but now that
I've read your mail I started to doubt if it might be related.
There seems to be a few "generic" 2.4 kernel issues in the same area.
For instance, building the kernel without CONFIG_VT causes
handle_scancode to be undefined, since driver/char/Makefile does not
build keyboard.c in such a case, albeit it's used by the input
subsystem. In any case, this is not related to Adeos/Xenomai.
Indeed, that's what they've concluded too, the vanilla kernel suffers
from the same problem. They should have posted this themselves a
couple of day ago, but they seem to have some trouble getting started
with the ML :-)
Xenomai-core mailing list