From 4f90e73538f1faf101733fcd95392bb77ba9467c Mon Sep 17 00:00:00 2001

From: William Brown <firstyear@redhat.com>

Date: Wed, 22 Mar 2017 14:10:11 +1000

Subject: [PATCH] Ticket 49174 - nunc-stans can not use negative timeout

Bug Description:  FreeIPA regularly sets up service accounts with

an nsIdleTimeout of -1. As a result of an issue with NS and libevent

this would cause an instant timeout and disconnect of the service

account.

Fix Description:  Correctly check that jobs are registered to NS.

Add validation to NS for negative timeouts. During the job registration,

we force the timeout to be a valid value.

https://pagure.io/389-ds-base/issue/49174

Author: wibrown

Review by: mreynolds(Thanks!!!)

Signed-off-by: Mark Reynolds <mreynolds@redhat.com>

---

 ldap/servers/slapd/daemon.c           | 39 ++++++++++++++++++++++++++++-------

 src/nunc-stans/ns/ns_event_fw_event.c |  8 -------

 src/nunc-stans/ns/ns_thrpool.c        | 16 ++++++++++++++

 src/nunc-stans/test/test_nuncstans.c  | 20 ++++++++++++++++++

 4 files changed, 68 insertions(+), 15 deletions(-)

diff --git a/ldap/servers/slapd/daemon.c b/ldap/servers/slapd/daemon.c

index e17a858..a4ea4c0 100644

--- a/ldap/servers/slapd/daemon.c

+++ b/ldap/servers/slapd/daemon.c

@@ -1891,15 +1891,32 @@ ns_connection_post_io_or_closing(Connection *conn)

 			tv.tv_usec = slapd_wakeup_timer * 1000;

 			conn->c_ns_close_jobs++; /* now 1 active closure job */

 			connection_acquire_nolock_ext(conn, 1 /* allow acquire even when closing */); /* event framework now has a reference */

-			ns_add_timeout_job(conn->c_tp, &tv, NS_JOB_TIMER,

+			PRStatus job_result = ns_add_timeout_job(conn->c_tp, &tv, NS_JOB_TIMER,

 					   ns_handle_closure, conn, NULL);

-			slapi_log_err(SLAPI_LOG_CONNS, "ns_connection_post_io_or_closing", "post closure job "

-				"for conn %" NSPRIu64 " for fd=%d\n", conn->c_connid, conn->c_sd);

+#ifdef DEBUG

+			PR_ASSERT(job_result == PR_SUCCESS);

+#endif

+			if (job_result != PR_SUCCESS) {

+				slapi_log_err(SLAPI_LOG_WARNING, "ns_connection_post_io_or_closing", "post closure job "

+					"for conn %" NSPRIu64 " for fd=%d failed to be added to event queue\n", conn->c_connid, conn->c_sd);

+			} else {

+				slapi_log_err(SLAPI_LOG_CONNS, "ns_connection_post_io_or_closing", "post closure job "

+					"for conn %" NSPRIu64 " for fd=%d\n", conn->c_connid, conn->c_sd);

+			}

 		}

 	} else {

 		/* process event normally - wait for I/O until idletimeout */

-		tv.tv_sec = conn->c_idletimeout;

+		/* With nunc-stans there is a quirk. When we have idleTimeout of -1

+		 * which is set on some IPA bind dns for infinite, this causes libevent 

+		 * to *instantly* timeout. So if we detect < 0, we set 0 to this timeout, to

+		 * catch all possible times that an admin could set.

+		 */

+		if (conn->c_idletimeout < 0) {

+			tv.tv_sec = 0;

+		} else {

+			tv.tv_sec = conn->c_idletimeout;

+		}

 		tv.tv_usec = 0;

 #ifdef DEBUG

 		PR_ASSERT(0 == connection_acquire_nolock(conn));

@@ -1913,11 +1930,19 @@ ns_connection_post_io_or_closing(Connection *conn)

 			return;

 		}

 #endif

-		ns_add_io_timeout_job(conn->c_tp, conn->c_prfd, &tv,

+		PRStatus job_result = ns_add_io_timeout_job(conn->c_tp, conn->c_prfd, &tv,

 				      NS_JOB_READ|NS_JOB_PRESERVE_FD,

 				      ns_handle_pr_read_ready, conn, NULL);

-		slapi_log_err(SLAPI_LOG_CONNS, "ns_connection_post_io_or_closing", "post I/O job for "

-			"conn %" NSPRIu64 " for fd=%d\n", conn->c_connid, conn->c_sd);

+#ifdef DEBUG

+		PR_ASSERT(job_result == PR_SUCCESS);

+#endif

+		if (job_result != PR_SUCCESS) {

+			slapi_log_err(SLAPI_LOG_WARNING, "ns_connection_post_io_or_closing", "post I/O job for "

+				"conn %" NSPRIu64 " for fd=%d failed to be added to event queue\n", conn->c_connid, conn->c_sd);

+		} else {

+			slapi_log_err(SLAPI_LOG_CONNS, "ns_connection_post_io_or_closing", "post I/O job for "

+				"conn %" NSPRIu64 " for fd=%d\n", conn->c_connid, conn->c_sd);

+		}

 	}

 #endif

 }

diff --git a/src/nunc-stans/ns/ns_event_fw_event.c b/src/nunc-stans/ns/ns_event_fw_event.c

index 3acbaf7..76936de 100644

--- a/src/nunc-stans/ns/ns_event_fw_event.c

+++ b/src/nunc-stans/ns/ns_event_fw_event.c

@@ -48,7 +48,6 @@ typedef struct event ns_event_fw_sig_t;

 #include "ns_event_fw.h"

 #include <syslog.h>

-

 static void

 event_logger_cb(int severity, const char *msg)

 {

@@ -248,13 +247,6 @@ ns_event_fw_mod_io(

     }

     if (events) {

         job->ns_event_fw_fd->ev_events = events;

-

-#ifdef DEBUG_FSM

-        /* REALLY make sure that we aren't being re-added */

-        if (event_pending(job->ns_event_fw_fd, events, tv)) {

-            abort();

-        }

-#endif

         event_add(job->ns_event_fw_fd, tv);

     } else {

         /* setting the job_type to remove IO events will remove it from the event system */

diff --git a/src/nunc-stans/ns/ns_thrpool.c b/src/nunc-stans/ns/ns_thrpool.c

index a867b39..9d87384 100644

--- a/src/nunc-stans/ns/ns_thrpool.c

+++ b/src/nunc-stans/ns/ns_thrpool.c

@@ -180,6 +180,14 @@ ns_thrpool_is_event_shutdown(struct ns_thrpool_t *tp)

     return result;

 }

+static int32_t

+validate_event_timeout(struct timeval *tv) {

+    if (tv->tv_sec < 0 || tv->tv_usec < 0) {

+        /* If we get here, you have done something WRONG */

+        return 1;

+    }

+    return 0;

+}

 static void

 job_queue_cleanup(void *arg) {

@@ -864,6 +872,10 @@ ns_add_timeout_job(ns_thrpool_t *tp, struct timeval *tv, ns_job_type_t job_type,

         return PR_FAILURE;

     }

+    if (validate_event_timeout(tv)) {

+        return PR_FAILURE;

+    }

+

     /* get an event context for a timer job */

     _job = alloc_timeout_context(tp, tv, job_type, func, data);

     if (!_job) {

@@ -900,6 +912,10 @@ ns_add_io_timeout_job(ns_thrpool_t *tp, PRFileDesc *fd, struct timeval *tv,

         return PR_FAILURE;

     }

+    if (validate_event_timeout(tv)) {

+        return PR_FAILURE;

+    }

+

     /* Don't allow an accept job to be run outside of the event thread.

      * We do this so a listener job won't shut down while still processing

      * current connections in other threads.

diff --git a/src/nunc-stans/test/test_nuncstans.c b/src/nunc-stans/test/test_nuncstans.c

index 8eef9e6..2795302 100644

--- a/src/nunc-stans/test/test_nuncstans.c

+++ b/src/nunc-stans/test/test_nuncstans.c

@@ -385,6 +385,23 @@ ns_job_signal_cb_test(void **state)

     assert_int_equal(ns_job_done(job), 0);

 }

+/*

+ * Test that given a timeout of -1, we fail to create a job.

+ */

+

+static void

+ns_job_neg_timeout_test(void **state)

+{

+    struct ns_thrpool_t *tp = *state;

+

+    struct timeval tv = { -1, 0 };

+

+    PR_ASSERT(PR_FAILURE == ns_add_io_timeout_job(tp, 0, &tv, NS_JOB_THREAD, ns_init_do_nothing_cb, NULL, NULL));

+

+    PR_ASSERT(PR_FAILURE == ns_add_timeout_job(tp, &tv, NS_JOB_THREAD, ns_init_do_nothing_cb, NULL, NULL));

+

+}

+

 int

 main(void)

 {

@@ -410,6 +427,9 @@ main(void)

         cmocka_unit_test_setup_teardown(ns_job_signal_cb_test,

                                         ns_test_setup,

                                         ns_test_teardown),

+        cmocka_unit_test_setup_teardown(ns_job_neg_timeout_test,

+                                        ns_test_setup,

+                                        ns_test_teardown),

     };

     return cmocka_run_group_tests(tests, NULL, NULL);

 }

-- 

2.9.3