Message-ID: <457CAF41.7010207@scytek.de>
Date: Sun, 10 Dec 2006 20:07:13 -0500
From: Volker Quetschke <quetschke AT scytek DOT de>
User-Agent: Thunderbird 1.5.0.8 (Windows/20061025)
MIME-Version: 1.0
To: cygwin AT cygwin DOT com
Subject: CTRL-c kills all children of current parent process.
Content-Type: multipart/signed; micalg=pgp-sha1;  protocol="application/pgp-signature";  boundary="------------enig30DACB1CEAAF8296E1D62797"
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

--------------enig30DACB1CEAAF8296E1D62797
Content-Type: multipart/mixed;
 boundary="------------090203030608010001090309"

This is a multi-part message in MIME format.
--------------090203030608010001090309
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

I encountered the problem that a CTRL-c in a shell (tested with bash in
rxvt and in cmd.exe with/without CYGWIN=3Dtty (Yes, before I started the
bash)) kills all children started by the running process even though
it has its own signal handler in place.

Take the attached source and compile it.
$ gcc breaktest.c -o breaktest.exe

Start it and press CTRL-c:
$ ./breaktest.exe
Started child pid: 372 from parent 3504
you have pressed ctrl-c
wait returned: 372
pid: 372 wid: 372
Done

Hmm, the CTRL-c kills the child and the parent returns because
the child finished.

Now start it again and send the SIGINT via `kill -INT <parentid>` from
another shell (3x in this example):

$ ./breaktest.exe
Started child pid: 2140 from parent 3608
you have pressed ctrl-c
you have pressed ctrl-c
you have pressed ctrl-c


That works as expected. So the question is, can killing the children
of a running process by CTRL-c be considered a bug?


    Volker


--=20
PGP/GPG key  (ID: 0x9F8A785D)  available  from  wwwkeys.de.pgp.net
key-fingerprint 550D F17E B082 A3E9 F913  9E53 3D35 C9BA 9F8A 785D

--------------090203030608010001090309
Content-Type: text/plain;
 name="breaktest.c"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline;
 filename="breaktest.c"

#include <stdio.h>
#include <process.h>
#include <errno.h>
#include <signal.h>
#include <sys/wait.h>

void sigproc(int);
void quitproc(int);

int
main(void)
{
  pid_t pid1, wid;
  int status;

  char *cmnd[] =3D {"sleep", "2000", (char *)0 };

  signal(SIGINT, sigproc);
  signal(SIGQUIT, quitproc);=20

  pid1 =3D spawnvp(_P_NOWAIT, cmnd[0], (const char**) cmnd);
  if(pid1 =3D=3D -1) {
    fprintf(stderr,"%s : %s\n", cmnd[0], strerror(errno));
  }
  fprintf(stderr, "Started child pid: %d from parent %d\n", pid1, getpid());

  do {
     wid =3D wait(&status);
     fprintf(stderr, "wait returned: %d\n", wid);
  } while( wid !=3D pid1 );

  fprintf(stderr, "pid: %d wid: %d\nDone\n", pid1, wid);
}

void sigproc(int a) {
   printf("you have pressed ctrl-c \n");
}
=20
void quitproc(int a) {
   printf("ctrl-\\ pressed to quit\n");
   exit(0); /* normal exit status */
}

--------------090203030608010001090309--

--------------enig30DACB1CEAAF8296E1D62797
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (MinGW)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFfK9EPTXJup+KeF0RAmJcAKCryCFCac2EONfiFWtKfQafwxfRUACePT3d
anutCAvOCwi9qGG3pDdk2Wo=
=VafK
-----END PGP SIGNATURE-----

--------------enig30DACB1CEAAF8296E1D62797--