|  support |  documentation |  report a bug |  advanced search |  search howto |  statistics |  random bug |  login
Bug #37931 PHP in Apache crash after Database-Restart while using persistent connections
Submitted: 2006-06-27 15:34 UTC Modified: 2006-06-28 15:04 UTC
From: Olaf dot Imig at bifab dot de Assigned:
Status: Closed Package: OCI8 related
PHP Version: 5.1.4 OS: Linux(x86)/Solaris(sparc)
Private report: No CVE-ID: None
Welcome back! If you're the original bug submitter, here's where you can edit the bug or add additional notes.
If this is not your bug, you can add a comment by following this link.
If this is your bug, but you forgot your password, you can retrieve your password here.
Bug Type:
From: Olaf dot Imig at bifab dot de
New email:
PHP Version: OS:


 [2006-06-27 15:34 UTC] Olaf dot Imig at bifab dot de
The following behavior ist reproduceable:
-Call the script
-shutdown/startup the Database
-Call the script
=> Apache Instance crash

The bug is forced, when you call the script many times. For the tests I used the OCI8 stuff from Oracle8i on Solaris and from Oracle8i and Oracle10g on Linux. The result was the same. Without persistent connection all is perfect.

Perhaps I can deliver a Solution. I made two changes in the code and the bug vanished.
After restart of Database php_oci_connection_ping found invalid connections.
The first problem was here, that the function php_oci_connection_close was called twice, first time triggered through  zend_hash_del(&EG(persistent_list)...
and then through a direct call. The Oracle-C-functions don't like this. Furthermore some pointers was freed twice.
The second problem is the call of OCISessionEnd, when the Connection is invalid. Setting of connection->is_close=0 before calling php_oci_connection_close was helpful in this special case.
On demand I can send my patch.

Reproduce code:
    print "Connection failed\n";


Add a Patch

Pull Requests

Add a Pull Request


AllCommentsChangesGit/SVN commitsRelated reports
 [2006-06-27 15:48 UTC]
Please try using this CVS snapshot:
For Windows:

 [2006-06-27 16:22 UTC] Olaf dot Imig at bifab dot de
Ok, I made the test with php5.2-20060627143 on Solaris. I found the same result.
 [2006-06-27 16:57 UTC]
Thank you for this bug report. To properly diagnose the problem, we
need a backtrace to see what is happening behind the scenes. To
find out how to generate a backtrace, please read for *NIX and for Win32

Once you have generated a backtrace, please submit it to this bug
report and change the status back to "Open". Thank you for helping
us make PHP better.

Also, I should note that ocilogoff($db) does not do anything with persistent connections, it just ignores them, so this piece of code doesn't make much sense.
 [2006-06-28 08:56 UTC] Olaf dot Imig at bifab dot de
Here my backtrace. When I start with httpd -X the server crash immediately after database restart

 #0  0xa70c228c in kputac () from /opt/oracle/oracle8i/lib/
#1  0xa710ccb1 in OCISessionEnd ()
   from /opt/oracle/oracle8i/lib/
#2  0xa79cbf43 in php_oci_connection_close (connection=0x824b368)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/ext/oci8/oci8.c:1457
#3  0xa79c99ba in php_oci_pconnection_list_dtor (entry=0x826b4d8)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/ext/oci8/oci8.c:707
#4  0xa7b23bd9 in plist_entry_destructor (ptr=0x826b4d8)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/Zend/zend_list.c:210
#5  0xa7b21590 in zend_hash_del_key_or_index (ht=0xa7c64bd4, 
    arKey=0x823a5b4 "oci8___opasuser__topasl__0", nKeyLength=27, h=513927878, 
    at /export/home/users/minstall/medea/software/php5.2-200606271430/Zend/zend_hash.c:492
#6  0xa79cad30 in php_oci_do_connect_ex (username=0x823edb4 "OPASUSER", 
    username_len=8, password=0x823a634 "OPASPASS", password_len=8, 
    new_password=0x0, new_password_len=0, dbname=0x823d18c "TOPASL", 
    dbname_len=6, charset=0x0, session_mode=0, persistent=1, exclusive=0)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/ext/oci8/oci8.c:1077
#7  0xa79ca187 in php_oci_do_connect (ht=3, return_value=0x823a6b4, 
    return_value_ptr=0x0, this_ptr=0x0, return_value_used=1, persistent=1, 
    at /export/home/users/minstall/medea/software/php5.2-200606271430/ext/oci8/oci8.c:947
#8  0xa79d9cc7 in zif_oci_pconnect (ht=3, return_value=0x823a6b4, 
    return_value_ptr=0x0, this_ptr=0x0, return_value_used=1)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/ext/oci8/oci8_interface.c:1536
#9  0xa7b3844c in zend_do_fcall_common_helper_SPEC (execute_data=0xafb2e6f0)
    at zend_vm_execute.h:200
#10 0xa7b3b7ec in ZEND_DO_FCALL_SPEC_CONST_HANDLER (execute_data=0xafb2e6f0)
    at zend_vm_execute.h:1642
#11 0xa7b38003 in execute (op_array=0x823ecf4) at zend_vm_execute.h:92
#12 0xa7b16625 in zend_execute_scripts (type=8, retval=0x0, file_count=3)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/Zend/zend.c:1110
#13 0xa7acb122 in php_execute_script (primary_file=0xafb30a80)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/main/main.c:1748
#14 0xa7b85710 in php_handler (r=0x8277d98)
    at /export/home/users/minstall/medea/software/php5.2-200606271430/sapi/apache2handler/sapi_apache2.c:593
#15 0x08067ce9 in ap_invoke_handler ()
#16 0x080656b0 in ap_process_request ()
#17 0x08060bcb in ap_process_http_connection ()
#18 0x08070cca in ap_process_connection ()
#19 0x08066b13 in child_main ()
#20 0x08066bfc in make_child ()
#21 0x08066ffe in ap_mpm_run ()
#22 0x0806cfad in main ()
 [2006-06-28 09:32 UTC]
I don't know how did you manage to shutdown Oracle instance when there are active connections and I definitely cannot reproduce it.
Please provide more information.
 [2006-06-28 10:54 UTC] Olaf dot Imig at bifab dot de
Here my procedure. I work on a Linux(x86) system. Webserver and Database are on different computers (communication over TCP!). First I start the webserver with httpd  -X. Then I run my script. netstat show the connection:
tcp       20      0 medea-dev.l.bifab:59019   ESTABLISHED
Then I stop the database:
lsnrctl stop listener
dbshut (with shutdown immediate)
Then I start the database:
lsnrctl start listener
After a short time (2-3 min) I call the my script once more.
=>Segmentation fault
 [2006-06-28 14:03 UTC]
Does this patch help you?
 [2006-06-28 14:47 UTC] Olaf dot Imig at bifab dot de
No, this patch isn't the full solution. Setting connection->is_open to 0 in php_oci_connection_ping is not enough. The server will crash, when php_oci_connection_status
fail. Because of the twice call of  php_oci_connection_close  
after the ping-test the server crash. Its enough to call zend_hash_del(&EG(persistent_list). The destructor of the persistent_list then call php_oci_connection_close. 
Here is my patch (work on php5.2-200606271430)
--- php-5.1.4/ext/oci8/oci8.c.ori	2006-06-26 15:59:03.000000000 +0200
+++ php-5.1.4/ext/oci8/oci8.c	2006-06-27 16:02:28.000000000 +0200
@@ -1065,8 +1065,11 @@
 					/* server died */
+					connection->is_open=0;
 					zend_hash_del(&EG(persistent_list), hashed_details.c, hashed_details.len+1);
+					/* one call too much 
 					php_oci_connection_close(connection TSRMLS_CC);
+					*/
 					connection = NULL;
 					goto open;
 [2006-06-28 15:04 UTC]
This bug has been fixed in CVS.

Snapshots of the sources are packaged every three hours; this change
will be in the next snapshot. You can grab the snapshot at
Thank you for the report, and for helping us make PHP better.

PHP Copyright © 2001-2020 The PHP Group
All rights reserved.
Last updated: Tue Jul 07 16:01:26 2020 UTC