David Malone wrote (2001/06/11):
> On Sun, Jun 10, 2001 at 05:20:50PM -0700, Peter Wemm wrote:
> > I agree totally. This should have been done ages ago, I've been burned on
> > it a few times, but never badly enough to go fix it.
>
> I've committed this - I'll let Matt do the MFC when he feels ready.
Thanks for this commit, but it works exactly in the way, which
I wanted to avoid. It is something like "partially submitted and
not approved by me". Please, could you look once again on it?
Please please please... Thanks. ;-)
- Hunk #1 from commit adds PS_NOCLDWAIT when SIG_IGN is used for
SIGCHLD, so if I use
sigemptyset(&sa.sa_mask);
sa.sa_flags = 0;
sa.sa_handler = SIG_IGN;
sigaction(SIGCHLD, &sa, NULL);
sigaction(SIGCHLD, NULL, &so);
I get additional flag SA_NOCLDWAIT in so.sa_flags. None of tested
systems (Solaris, Unixware, Irix, Linux, Aix) works in this way.
- If hunk #1 is applied, I see hunk #2 (resetting SIG_IGN to SIG_DFL
for SIGCHLD) as almost unnecessary, because it is just cosmetic
change now and behavior in exec() is very different among various
systems.
- (And I have found another bug in my original patch. Three code lines
and two bad bugs... :-( :-)
Here are patches for current -current (... which I had to do in the
first posting). These should change the behavior as is in Solaris,
where flags are not added and SIGCHLD is resetted to SIG_DFL after an
exec(). I'm testing these patches on my -stable production backup
server running amanda and up to now without any problem.
# Back out hunk #1 from applied commit
--- sys/kern/kern_sig.c.orig Mon Jun 11 21:47:29 2001
+++ sys/kern/kern_sig.c Tue Jun 12 07:42:42 2001
@@ -293,8 +293,7 @@
p->p_procsig->ps_flag |= PS_NOCLDSTOP;
else
p->p_procsig->ps_flag &= ~PS_NOCLDSTOP;
- if ((act->sa_flags & SA_NOCLDWAIT) ||
- ps->ps_sigact[_SIG_IDX(SIGCHLD)] == SIG_IGN) {
+ if (act->sa_flags & SA_NOCLDWAIT) {
/*
* Paranoia: since SA_NOCLDWAIT is implemented
* by reparenting the dying child to PID 1 (and
# Add test for SIG_IGN to exit1()
--- sys/kern/kern_exit.c.orig Mon Jun 11 21:47:35 2001
+++ sys/kern/kern_exit.c Tue Jun 12 07:41:22 2001
@@ -341,7 +341,8 @@
* flag set, notify process 1 instead (and hope it will handle
* this situation).
*/
- if (p->p_pptr->p_procsig->ps_flag & PS_NOCLDWAIT) {
+ if ((p->p_pptr->p_procsig->ps_flag & PS_NOCLDWAIT)
+ || p->p_pptr->p_sigacts->ps_sigact[_SIG_IDX(SIGCHLD)] == SIG_IGN) {
struct proc *pp = p->p_pptr;
proc_reparent(p, initproc);
/*
# And this is patch for question from my first posting: ... Why? I know
# that sa_handler and sa_sigaction are the same pointer in an union,
# but I think that "act->sa_handler" should be exchanged with
# "(__sighandler_t *)act->sa_sigaction" for clarification purposes.
--- sys/kern/kern_sig.c Tue Jun 12 07:42:42 2001
+++ sys/kern/kern_sig.c.new Tue Jun 12 08:16:12 2001
@@ -259,11 +259,11 @@
ps->ps_catchmask[_SIG_IDX(sig)] = act->sa_mask;
SIG_CANTMASK(ps->ps_catchmask[_SIG_IDX(sig)]);
if (act->sa_flags & SA_SIGINFO) {
- ps->ps_sigact[_SIG_IDX(sig)] = act->sa_handler;
- SIGADDSET(ps->ps_siginfo, sig);
- } else {
ps->ps_sigact[_SIG_IDX(sig)] =
(__sighandler_t *)act->sa_sigaction;
+ SIGADDSET(ps->ps_siginfo, sig);
+ } else {
+ ps->ps_sigact[_SIG_IDX(sig)] = act->sa_handler;
SIGDELSET(ps->ps_siginfo, sig);
}
if (!(act->sa_flags & SA_RESTART))
Thanks.
--
Rudolf Cejka ([EMAIL PROTECTED]; http://www.fee.vutbr.cz/~cejkar)
Brno University of Technology, Faculty of El. Engineering and Comp. Science
Bozetechova 2, 612 66 Brno, Czech Republic
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
