From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gdb-return-21683-listarch-gdb=sources.redhat.com@sources.redhat.com>
Received: (qmail 6001 invoked by alias); 26 May 2005 19:40:47 -0000
Mailing-List: contact gdb-help@sources.redhat.com; run by ezmlm
Precedence: bulk
List-Subscribe: <mailto:gdb-subscribe@sources.redhat.com>
List-Archive: <http://sources.redhat.com/ml/gdb/>
List-Post: <mailto:gdb@sources.redhat.com>
List-Help: <mailto:gdb-help@sources.redhat.com>, <http://sources.redhat.com/ml/#faqs>
Sender: gdb-owner@sources.redhat.com
Received: (qmail 5922 invoked by uid 22791); 26 May 2005 19:40:41 -0000
Received: from 64-6-182-228.t1.lax.megapath.net (HELO abbott.domain.trailerparc.com) (64.6.182.228)
    by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Thu, 26 May 2005 19:40:41 +0000
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Subject: gdb 6.3  misses breakpoint on Linux when inferior does clone()
Date: Thu, 26 May 2005 19:40:00 -0000
Message-ID: <2DC041637C94C64C8039281448B323B2870FBA@abbott.domain.trailerparc.com>
From: "Satish Mohan" <Satish.Mohan@Accenia.com>
To: <gdb@sources.redhat.com>
Cc: "Satish Mohan" <Satish.Mohan@accenia.com>
X-SW-Source: 2005-05/txt/msg00338.txt.bz2

Hi
=A0If the inferior does a clone(...,CLONE_VM,...) on Linux, then gdb delete=
s breakpoints in the child which indirectly affects the parent because of C=
LONE_VM. The following sample code illustrates the problem:

int mythread (void * unused)=20
{
=A0=A0 printf("my thread running\n");
=A0=A0 return 0;
}

int main (int argc, char **argv)=20
{
=A0=A0 int status, pid;
=A0=A0 char * stack =3D (char *)malloc(4096);
=A0=A0 if ((pid =3D clone (mythread, (char *)(stack + 4096 - 4),=20
                     CLONE_VM | SIGCHLD, 0)) =3D=3D -1)
=A0=A0 {
=A0=A0=A0=A0=A0=A0 printf("clone failed, errno=3D%d\n", errno);=20
=A0=A0=A0=A0=A0=A0 return -1;
=A0=A0 }

=A0=A0 if (waitpid(pid, &status, 0) =3D=3D -1)=20
   {
=A0=A0=A0=A0=A0=A0 perror("waitpid failed\n");
=A0=A0=A0=A0=A0=A0 return -1;
=A0=A0 }
}

If a breakpoint is set on the waitpid() call above, then gdb misses it. Thi=
s seems to be because gdb deletes breakpoints in the child after the clone(=
) but doesn't seem to know that CLONE_VM is set, and that the breakpoints w=
ill also get deleted from the parent.=20

Commenting the detach_breakpoints(child_pid) in child_follow_fork() in linu=
x-nat.c fixes the problem as long as parent and child don't execute the sam=
e code after clone.=20

Is this a gdb bug ? Could anyone let me know the right fix for this ?

Thanks,
Satish Mohan.