s3:winbindd: fix problems with SIGCHLD handling (bug #7317)

The main problem is that we call CatchChild() within the
parent winbindd, which overwrites the signal handler
that was registered by winbindd_setup_sig_chld_handler().

That means winbindd_sig_chld_handler() and winbind_child_died()
are never triggered when a winbindd domain child dies.
As a result will get "broken pipe" for all requests to that domain.

To reduce the risk of similar bugs in future we call
CatchChild() in winbindd_reinit_after_fork() now.

We also use a full winbindd_reinit_after_fork() in the
cache validation child now instead instead of just resetting
the SIGCHLD handler by hand. This will also fix possible
tdb problems on systems without pread/pwrite and disabled mmap
as we now correctly reopen the tdb handle for the child.

metze
(cherry picked from commit 73577205cf81644e7fe853eaf3e6459f7f443096)
(cherry picked from commit e0ece652956292cc67383535a0fa174b5015d91e)
(cherry picked from commit 26bdc249310b71dc45e087347e456c9f5b0f4f9b)

3 files changed, 6 insertions, 17 deletions

diff --git a/source3/winbindd/winbindd.c b/source3/winbindd/winbindd.c
index e98b7994adb..eb5429c3efb 100644
--- a/source3/winbindd/winbindd.c
+++ b/source3/winbindd/winbindd.c
@@ -378,7 +378,6 @@ static void winbind_msg_validate_cache(struct messaging_context *msg_ctx,
 	 * so we don't block the main winbindd and the validation
 	 * code can safely use fork/waitpid...
 	 */
-	CatchChild();
 	child_pid = sys_fork();
 
 	if (child_pid == -1) {
@@ -396,16 +395,9 @@ static void winbind_msg_validate_cache(struct messaging_context *msg_ctx,
 
 	/* child */
 
-	/* install default SIGCHLD handler: validation code uses fork/waitpid */
-	ZERO_STRUCT(act);
-	act.sa_handler = SIG_DFL;
-#ifdef SA_RESTART
-	/* We *want* SIGALRM to interrupt a system call. */
-	act.sa_flags = SA_RESTART;
-#endif
-	sigemptyset(&act.sa_mask);
-	sigaddset(&act.sa_mask,SIGCHLD);
-	sigaction(SIGCHLD,&act,&oldact);
+	if (!winbindd_reinit_after_fork(NULL)) {
+		_exit(0);
+	}
 
 	ret = (uint8)winbindd_validate_cache_nobackup();
 	DEBUG(10, ("winbindd_msg_validata_cache: got return value %d\n", ret));
diff --git a/source3/winbindd/winbindd_cm.c b/source3/winbindd/winbindd_cm.c
index d78742f5041..d8bde4690f6 100644
--- a/source3/winbindd/winbindd_cm.c
+++ b/source3/winbindd/winbindd_cm.c
@@ -174,9 +174,6 @@ static bool fork_child_dc_connect(struct winbindd_domain *domain)
 	pid_t parent_pid = sys_getpid();
 	char *lfile = NULL;
 
-	/* Stop zombies */
-	CatchChild();
-
 	if (domain->dc_probe_pid != (pid_t)-1) {
 		/*
 		 * We might already have a DC probe
diff --git a/source3/winbindd/winbindd_dual.c b/source3/winbindd/winbindd_dual.c
index b6287dd3ae8..be6b1371b06 100644
--- a/source3/winbindd/winbindd_dual.c
+++ b/source3/winbindd/winbindd_dual.c
@@ -1167,6 +1167,9 @@ bool winbindd_reinit_after_fork(const char *logfilename)
 					    logfilename))
 		return false;
 
+	/* Stop zombies in children */
+	CatchChild();
+
 	/* Don't handle the same messages as our parent. */
 	messaging_deregister(winbind_messaging_context(),
 			     MSG_SMB_CONF_UPDATED, NULL);
@@ -1282,9 +1285,6 @@ static bool fork_domain_child(struct winbindd_child *child)
 
 	DEBUG(10, ("Child process %d\n", (int)sys_getpid()));
 
-	/* Stop zombies in children */
-	CatchChild();
-
 	state.sock = fdpair[0];
 	close(fdpair[1]);