Subject: Re: CVS commit: src/bin/sh (trap.c:1.32)
To: David Young <firstname.lastname@example.org>
From: Greywolf <email@example.com>
Date: 07/14/2005 18:00:09
[Thus spake David Young ("DY: ") 6:42pm...]
DY: Christos, maybe that should be "Don't hard ignore signals that were
DY: ignored by our environment ..." after a trap is set?
As I understand it, subprocesses are supposed to inherit the signal
mask if possible. You obviously can't inherit signals with handlers;
they get reset to the default. Blocked signals, however, are inherited.
Realistically, I've had to hack around this with
( trap 1; command... & )
( trap 1; command... ) &
(depending on desired semantics) in order to get this to work.
Now, maybe something real can be applied to the SIGQUIT problem we have
with fsck immediately terminating instead of finishing the check and
returning to single-user mode as its output message promises...