cdebconf and DFB signal handling issues
Hi
recently we discovered that DFB's signal handling mechanism may be
occasionally broken and could lead to a crash of the active DFB
application, cdebconf in the case of the g-i (not clear why and when,
exactly, still to discover ).
DFB internally handles SIGUSR1 and SIGUSR2: SIGUSR1 suspends the DFB
core, while SIGUSR2 resumes it.
After the internal handler has been executed, DFB should run the
previously stored handlers for SIGUSR1 and SIGUSR2 that, in the case of
cdebconf, are used to save the question database and/or quitting.
Now, is this signal-handling mechanism of cdebconf really used or is
that a never-used functionality? is it really worth fixing the bug in
DFB's signal handling system?
ciao
Attilio
Reply to: