As long as NumLock or CapsLock are not touched, everything works fine (even
console switching). As soon as one of these is touched (maybe others, I have
not experimented over-long) the keyboard will lock up for some seconds, and
from then on, console switching will work, with the same time lag before
accepting keypresses. After having had 4-5 of these lockups, the keyboard
driver becomes very confused (I was in vi at the time, not the very best place
for blind keypressing :)
Luca
P.S. I could not experiment with pre-2.1.37-2, it wouldn't compile as
non-SMP. I expect the problem would be present, as the changes in vt.c and
tty_io.c were already there (not that I'm implying I'm sure the problem
is in those changes :)
P.P.S If i compile as SMP, it barfs in calibrate_APIC_clock on my Cyrix 6x86
P166+ (if necessary, I will repeat the experiment and hand transcribe
the Oops). Yes I know that Cyrix 6x86s are not exactly smp-capable, just
wondering if a more graceful fallback is possible or desirable.