Package: socat
Version: 1.4.3.1-1
Severity: normal
Tags: patch
I don't currently know that I'm experiencing any symptom of this, other than a
warning generated by what I'm told is an obsolete warning flag to Perl.
mart...@whitewater:~$ perl -we 0
mart...@whitewater:~$ socat - EXEC:'perl -we 0'
Can't ignore signal CHLD, forcing to default.
mart...@whitewater:~$
Here's a demonstration with a more recent version of socat, unpatched then
patched:
mart...@whitewater:/tmp/socat-1.7.1.0$ /tmp/socat-1.7.1.0/socat - EXEC:'perl
-we 0'
Can't ignore signal CHLD, forcing to default.
mart...@whitewater:/tmp/socat-1.7.1.0$ patch -p0 < xio-progcall.c.patch
patching file xio-progcall.c
mart...@whitewater:/tmp/socat-1.7.1.0$ make
gcc -O -D_GNU_SOURCE -Wall -Wno-parentheses -DHAVE_CONFIG_H -I. -I. -c -o
xio-progcall.o xio-progcall.c
ar r libxio.a <snipped>
ranlib libxio.a
gcc -O -D_GNU_SOURCE -Wall -Wno-parentheses -DHAVE_CONFIG_H -I. -o socat
socat.o libxio.a -lutil -lssl
mart...@whitewater:/tmp/socat-1.7.1.0$ /tmp/socat-1.7.1.0/socat - EXEC:'perl
-we 0'
mart...@whitewater:/tmp/socat-1.7.1.0$
I fear there might be other symptoms, which is why I'm raising the bug.
My fears are based partly on:
In http://www.pasc.org/interps/unofficial/db/p1003.1/pasc-1003.1-132.html,
a "David Korn" (perhaps *the* Dave Korn) is quoted as having written:
A conforming application would not set SIGCHLD to SIG_IGN
since the standard leaves this behavior unspecified. An application
that does set SIGCHLD to SIG_IGN should set it back to SIG_DFL
before the call to exec.
In working on terminal emulators, I've seen a number of programs revert SIGCHLD
to SIG_DFL before calling exec(), generally without convincingly specifying
the reason. Perhaps the unspecified heritability of SIG_IGN is the reason.
I wonder why socat ignores SIGCHLD in the child. Perhaps it's for the case
where the child will then call system(). Then the SIG_IGN will affect socat
code as well as any child that inherits it. But system() waits for its child.
Perhaps this SIG_IGN is a relic of earlier efforts to prevent zombies.
More likely, I'm missing something.
-- System Information:
Debian Release: 4.0
APT prefers oldstable
APT policy: (500, 'oldstable'), (50, 'unstable')
Architecture: amd64 (x86_64)
Shell: /bin/sh linked to /bin/bash
Kernel: Linux 2.6.18-4-amd64
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL
set to en_US.UTF-8)
Versions of packages socat depends on:
ii libc6 2.3.6.ds1-13etch9 GNU C Library: Shared libraries
ii libreadline5 5.2-2 GNU readline and history libraries
ii libssl0.9.8 0.9.8c-4etch5 SSL shared libraries
ii libwrap0 7.6.dbs-13 Wietse Venema's TCP wrappers libra
socat recommends no packages.
-- no debconf information
--- xio-progcall.c.orig 2009-05-29 12:20:53.000000000 -0700
+++ xio-progcall.c 2009-05-29 12:24:43.000000000 -0700
@@ -414,8 +414,10 @@
gid_t group;
if (withfork) {
- if (Signal(SIGCHLD, SIG_IGN) == SIG_ERR) {
- Warn1("signal(SIGCHLD, SIG_IGN): %s", strerror(errno));
+ /* The child should have default handling for SIGCHLD. */
+ /* In particular, it's not defined whether ignoring SIGCHLD is inheritable. */
+ if (Signal(SIGCHLD, SIG_DFL) == SIG_ERR) {
+ Warn1("signal(SIGCHLD, SIG_DFL): %s", strerror(errno));
}
#if HAVE_PTY
