[389-ds-base] branch master updated (bdcbf5b -> 1b2a2d6)

[git repository hosting]

This is an automated email from the git hooks/post-receive script.

tbordaz pushed a change to branch master
in repository 389-ds-base.

from  bdcbf5b   Ticket 48118 - fix compiler warning for incorrect return type
       new  1b2a2d6   Ticket 49410 - opened connection can remain no longer poll, like hanging

The 1 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails.  The revisions
listed as "adds" were already present in the repository and have only
been added to this reference.

Summary of changes:
 ldap/servers/slapd/connection.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

[389-ds-base] 01/01: Ticket 49410 - opened connection can remain no longer poll, like hanging

[git repository hosting]

This is an automated email from the git hooks/post-receive script.

tbordaz pushed a commit to branch master
in repository 389-ds-base.

commit 1b2a2d6c08c78637ac1b83a7ef55878eced266ff
Author: Thierry Bordaz 
Date:   Tue Nov 14 16:29:03 2017 +0100

Ticket 49410 - opened connection can remain no longer poll, like hanging

Bug Description:
    	Some opened connection are no longer poll.
    	Those connections has 'gettingber' toggle set although there is
    	no more worker thread reading it.
    	The reason they have gettingber set is that the last
    	operation had 'persistent search' flag. With such flag
    	gettingber is not reset.
    	persistent flag is set even when no persistent search/sync_repl
    	was received on the connection.
    	The problem is that the flag is tested on the wrong operation.
    	The tested operation can be
    		- the first operation when the connection entered in turbo mode
    		- the previous operation if several ops PDUs were read on the network
    		- accessing random memory

In theory testing the flag can lead to sigsev even
    	if it never crash

Fix Description:
    	The fix is to use the operation that is in the pblock
    	In such case pb_op is no longer used, so we can get rid of it.
    	In addition make pb_conn a local variable where it is used

https://pagure.io/389-ds-base/issue/49410

Reviewed by: Ludwig Krispenz, Mark Reynolds

Platforms tested: F26

Flag Day: no

Doc impact: no
---
 ldap/servers/slapd/connection.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/ldap/servers/slapd/connection.c b/ldap/servers/slapd/connection.c
index 0128986..8ef1156 100644
--- a/ldap/servers/slapd/connection.c
+++ b/ldap/servers/slapd/connection.c
@@ -1498,8 +1498,6 @@ connection_threadmain()
     int maxthreads = 0;
     int enable_nunc_stans = 0;
     long bypasspollcnt = 0;
-    Connection *pb_conn = NULL;
-    Operation *pb_op = NULL;

enable_nunc_stans = config_get_enable_nunc_stans();
 #if defined(hpux)
@@ -1520,6 +1518,8 @@ connection_threadmain()
         }

if (!thread_turbo_flag && !more_data) {
+	    Connection *pb_conn = NULL;
+
             /* If more data is left from the previous connection_read_operation,
                we should finish the op now.  Client might be thinking it's
                done sending the request and wait for the response forever.
@@ -1530,7 +1530,6 @@ connection_threadmain()
              * Connection wait for new work provides the conn and op for us.
              */
             slapi_pblock_get(pb, SLAPI_CONNECTION, &pb_conn);
-            slapi_pblock_get(pb, SLAPI_OPERATION, &pb_op);

switch (ret) {
             case CONN_NOWORK:
@@ -1786,7 +1785,7 @@ connection_threadmain()
         /* total number of ops for the server */
         slapi_counter_increment(ops_completed);
         /* If this op isn't a persistent search, remove it */
-        if (pb_op->o_flags & OP_FLAG_PS) {
+        if (op->o_flags & OP_FLAG_PS) {
             PR_EnterMonitor(conn->c_mutex);
             connection_release_nolock(conn); /* psearch acquires ref to conn - release this one now */
             PR_ExitMonitor(conn->c_mutex);