(ITS#6078) Slapd.exe crashes when writing entries while reading many entries. - openldap-bugs

27 Apr 2009


      Full_Name: Frank Offermanns
Version: 2.4.16
OS: Windows XP
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (217.6.189.242)
Hello,
i am using OpenLDAP 2.4.16 with Windows XP, compiled with msys/mingw with
Windows XP and the berkeley DB 4.7.25 NC
The database runs without any replication (though with replication it also
crashes)
I use the "hdb" Backend.
I started the datatbase with "slapd -F slapd.d".
Now i add about 21000 Entries under the DN: o=test6,ou=Users,o=caesar.
When this is finished i start adding another 21000 Entries under the DN:
o=test7,ou=Users,o=caesar.
While adding these seconds entries i use the LDAPAdmin 1.1.0.0 to view the
subtree at "o=test6,ou=Users,o=caesar". LDAPAdmin starts to read the Users and
fails at about the 4000 Entry. At this time my slapd.exe crashed. 
I also did the test with "slapd -d 1 -F slapd.d" and found the following entries
in the logfile:
ber_flush2: 77 bytes to sd 2268
ber_get_next: tag 0x30 len 93 contents:
<= send_search_entry: conn 0 exit.
ber_get_next
conn=2 op=4017 do_modify
entry_decode: ""
ber_scanf fmt ({m) ber:
ber_scanf fmt ({e{m[W]}}) ber:
<= entry_decode()
...
...
...
dnPrettyNormal: <cn=10250,o=test7,ou=users,o=caesar>
<<< dnPrettyNormal: <cn=10250,o=test7,ou=users,o=caesar>,
<cn=10250,o=test7,ou=users,o=caesar>
bdb_dn2entry("cn=10250,o=test7,ou=users,o=caesar")
=> send_search_entry: conn 0 dn="cn=12034,o=test6,ou=Users,o=caesar"
=> hdb_dn2id("cn=10250,o=test7,ou=users,o=caesar")
ber_flush2: 77 bytes to sd 2268
<= hdb_dn2id: got id=0x53da
<= send_search_entry: conn 0 exit.
entry_decode: ""
entry_decode: ""
<= entry_decode()
<= entry_decode()
bdb_dn2entry("cn=10250,o=test7,ou=users,o=caesar")
=> send_search_entry: conn 0 dn="cn=12035,o=test6,ou=Users,o=caesar"
ber_flush2: 77 bytes to sd 2268
entry_decode: ""
<= entry_decode()
<= send_search_entry: conn 0 exit.
bdb_modify_internal: 0x000053da: c²0250
entry_decode: ""
<= entry_decode()
entry failed schema check: unrecongized attribute type(s) in RDN
hdb_modify: modify failed (34)
send_ldap_result: conn=2 op=4017 p=3
send_ldap_response: msgid=1026718 tag=103 err=34
=> send_search_entry: conn 0 dn="cn=12036,o=test6,ou=Users,o=caesar"
ber_flush2: 53 bytes to sd 2336
ber_flush2: 77 bytes to sd 2268
<= send_search_entry: conn 0 exit.
entry_decode: ""
<= entry_decode()
connection_get(4): got connid=2
connection_read(4): checking for input on id=2
ber_get_next
ber_get_next: tag 0x30 len 87 contents:
ber_get_next
conn=2 op=4018 do_modify
ber_scanf fmt ({m) ber:
ber_scanf fmt ({e{m[W]}}) ber:
...
...
...
dnPrettyNormal: <cn=10250,o=test7,ou=users,o=caesar>
<<< dnPrettyNormal: <cn=10250,o=test7,ou=users,o=caesar>,
<cn=10250,o=test7,ou=users,o=caesar>
bdb_dn2entry("cn=10250,o=test7,ou=users,o=caesar")
=> hdb_dn2id("cn=10250,o=test7,ou=users,o=caesar")
<= hdb_dn2id: got id=0x53da
entry_decode: ""
<= entry_decode()
The pointer at "bdb_modify_internal" seems to be dangling.
When i didn't read the entries while writing, there seems to be no problem.
The Version 2.4.13 also does not seem to have this problem. (I couldn't get the
2.4.13 Version crashing and the 2.4.16 crashes every time after a short time)
Additional Info:
I am not sure, if this could be a reason for this error, so i am writing this
problem also.
Starting at Version 2.4.14 the configure command creates makefiles which are not
linkable at mingw/msys at windows.
I manually added "-lws2_32" in the makefiles at the lines:
AC_LIBS = -lregex  -lws2_32
I apologize me for my bad english and hope you can understand me.
Regards,
Frank Offermanns