DFB and cdebconf signal handling
This bug is currently assigned to package directfb, but no one yet
proved this is really DFB rlated.
In an old bugreport about cdebconf [1], cjw warned about wrong/dangerous
SIGRUSR1/2 handling on PPC/AMD64 by some d-i components.
What do this has to do with this br? when a VT switch is requested by
the user, the DFB's input thread sends a SIGUSR1 to the VT thread, and
VT switching is performed.
But SIGUSR1 signal is also handled by cdebconf, which catches it to save
the questions database, IIRC.
We ran into some kind of nasty clash in signal handling between cdebconf
and DFB already before [2].
If this bug should be proved to be unreproducible outside the d-i, i
think looking at bug 316484 too would be a good idea.
cheers
Attilio
[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=316484
[2] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=339379
Reply to: