--On Monday, April 02, 2007 9:09 PM +0200 Pierangelo Masarati
<ando(a)sys-net.it> wrote:
>> It looks like there's soemthing incorrect in the way it is allocating
>> memory on startup.
>
> Apparently, there's nothing directly related to dynlist; can you post
> the entire slapd.conf? I suspect something like ITS#4848, which is NOT
> fixed in re23 (and likely won't).
I've looked at the ITS, and I don't think it is related. As far as I know,
everything in it is specifically located where it should be (global bits in
the global area, database specific bits in the database areas):
My slapd.conf is:
include /usr/local/etc/openldap/schema/core.schema
include /usr/local/etc/openldap/schema/cosine.schema
include /usr/local/etc/openldap/schema/dyngroup.schema
include /usr/local/etc/openldap/schema/krb5-kdc.schema
include /usr/local/etc/openldap/schema/inetorgperson.schema
include /usr/local/etc/openldap/schema/misc.schema
include /usr/local/etc/openldap/schema/nis.schema
include /usr/local/etc/openldap/schema/eduperson.schema
include /usr/local/etc/openldap/schema/stanford-oids.schema
include /usr/local/etc/openldap/schema/suacct.schema
include /usr/local/etc/openldap/schema/superson.schema
include /usr/local/etc/openldap/schema/suapplication.schema
include /usr/local/etc/openldap/schema/suorg.schema
include /usr/local/etc/openldap/schema/eduorg.schema
include /usr/local/etc/openldap/schema/suworkgroup.schema
allow bind_v2
TLSCertificateFile /usr/local/etc/openldap/stardomain.crt
TLSCertificateKeyFile /usr/local/etc/openldap/stardomain.key
TLSCACertificateFile /usr/local/etc/openldap/comodo.pem
include /usr/local/etc/openldap/slapd.acl
pidfile /var/run/slapd.pid
argsfile /var/run/slapd.args
defaultsearchbase "dc=stanford,dc=edu"
gentlehup off
loglevel 256
threads 8
tool-threads 2
sasl-realm stanford.edu
sasl-authz-policy both
sasl-regexp uid=(.*)/cgi,cn=stanford.edu,cn=gssapi,cn=auth
ldap:///cn=cgi,cn=applications,dc=stanford,dc=edu??sub?krb5PrincipalName=$1/cgi@stanford.edu
sasl-regexp uid=service/(.*),cn=stanford.edu,cn=gssapi,cn=auth
ldap:///cn=Service,cn=Applications,dc=stanford,dc=edu??sub?krb5PrincipalName=service/$1@stanford.edu
sasl-regexp uid=webauth/(.*),cn=stanford.edu,cn=gssapi,cn=auth
ldap:///cn=Webauth,cn=Applications,dc=stanford,dc=edu??sub?krb5PrincipalName=webauth/$1@stanford.edu
sasl-regexp uid=(.*),cn=stanford.edu,cn=gssapi,cn=auth
ldap:///uid=$1,cn=Accounts,dc=stanford,dc=edu??sub?suSeasStatus=active
reverse-lookup on
modulepath /usr/local/lib/openldap
moduleload back_hdb.la
moduleload back_monitor.la
moduleload valsort.la
moduleload dynlist.la
database hdb
suffix "dc=stanford,dc=edu"
rootdn "cn=manager,dc=stanford,dc=edu"
overlay valsort
valsort-attr suOrgContactStanford cn=orgs,dc=stanford,dc=edu weighted
valsort-attr suOrgContactWorld cn=orgs,dc=stanford,dc=edu weighted
overlay dynlist
dynlist-attrset groupOfURLS memberURL member
limits group="cn=ldapadmin,cn=applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=unlimited
size.hard=unlimited
limits
dn.exact="cn=athletics,cn=service,cn=applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=unlimited
size.hard=unlimited
limits
dn.exact="cn=workgroup-audit,cn=service,cn=applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=unlimited
size.hard=unlimited
limits
dn.exact="cn=RegistryDataAuditor,cn=Service,cn=Applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=unlimited
size.hard=unlimited
limits dn.exact="cn=ispace,cn=Service,cn=Applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=5000 size.hard=5000
limits
dn.exact="cn=gsb-person,cn=service,cn=applications,dc=stanford,dc=edu"
time.soft=unlimited time.hard=unlimited size.soft=unlimited
size.hard=unlimited
lastmod on
syncrepl rid=0
provider=ldap://ldap-devmaster.stanford.edu:389
bindmethod=sasl
saslmech=gssapi
realm=stanford.edu
searchbase="dc=stanford,dc=edu"
logbase="cn=accesslog"
logfilter="(&(objectClass=auditWriteObject)(reqResult=0))"
schemachecking=on
type=refreshAndPersist
retry="60 +"
syncdata=accesslog
# Replica Directives
updateref ldap://ldap-devmaster.stanford.edu
directory /var/lib/ldap
dbconfig set_cachesize 3 536870912 1
dbconfig set_lg_regionmax 262144
dbconfig set_lg_bsize 2097152
dbconfig set_lg_dir /var/log/bdb
dbconfig set_lk_max_locks 3000
dbconfig set_lk_max_objects 1500
dbconfig set_lk_max_lockers 1500
dbconfig set_flags DB_LOG_AUTOREMOVE
dbconfig set_tas_spins 1
checkpoint 1024 5
cachesize 50000
idlcachesize 50000
cachefree 1000
index_substr_any_len 3
index default eq
(Long list of indices)
database monitor
--
Quanah Gibson-Mount
Senior Systems Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html
Full_Name: Howard Chu
Version: 2.3/HEAD
OS:
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (76.168.84.21)
Submitted by: hyc
In syncrepl_message_to_op the modlist is copied to op->orm_modlist and then
freed after op->o_bd->be_modify. But the modify can alter the modlist (e.g.,
ppolicy overlay) causing a redundant free later on.
quanah(a)stanford.edu wrote:
> Full_Name: Quanah Gibson-Mount
> Version: RE_23
> OS: Linux 2.6 (64-bit)
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (171.64.19.81)
>
>
> In testing the RE_23 tree, I put it under hoard after everything passed using
> glibc as the memory allocator. I found that if I enable dynlist:
>
> moduleload dynlist.la
>
> then in my database section add:
>
> # Dynlist Overlay
> overlay dynlist
> dynlist-attrset groupOfURLS memberURL member
>
>
> slapd will core dump on startup with:
>
> Loaded symbols for /usr/local/lib/openldap/dynlist-2.3-releng.so.0
> #0 0x00002ab7651490c9 in malloc () from /usr/local/lib/libhoard.so
> (gdb) bt
> #0 0x00002ab7651490c9 in malloc () from /usr/local/lib/libhoard.so
> #1 0x00002ab765149e38 in realloc () from /usr/local/lib/libhoard.so
> #2 0x00002ab76554f890 in ber_memrealloc_x (p=Variable "p" is not available.
> ) at memory.c:354
> #3 0x00002ab76542bf88 in ldap_charray_add (a=0x7fff45a7eec0, s=0x2ab7667c2518
> "olcConnMaxPending") at charray.c:49
> #4 0x000000000045e9f9 in matching_rule_use_init () at mr.c:364
> #5 0x0000000000459f62 in slap_schema_check () at schema_prep.c:1368
> #6 0x0000000000415f1b in main (argc=3, argv=0x7fff45a7f0b8) at main.c:711
>
>
> It looks like there's soemthing incorrect in the way it is allocating memory on
> startup.
Apparently, there's nothing directly related to dynlist; can you post
the entire slapd.conf? I suspect something like ITS#4848, which is NOT
fixed in re23 (and likely won't).
p.
Ing. Pierangelo Masarati
OpenLDAP Core Team
SysNet s.n.c.
Via Dossi, 8 - 27100 Pavia - ITALIA
http://www.sys-net.it
------------------------------------------
Office: +39.02.23998309
Mobile: +39.333.4963172
Email: pierangelo.masarati(a)sys-net.it
------------------------------------------
Full_Name: Quanah Gibson-Mount
Version: RE_23
OS: Linux 2.6 (64-bit)
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (171.64.19.81)
In testing the RE_23 tree, I put it under hoard after everything passed using
glibc as the memory allocator. I found that if I enable dynlist:
moduleload dynlist.la
then in my database section add:
# Dynlist Overlay
overlay dynlist
dynlist-attrset groupOfURLS memberURL member
slapd will core dump on startup with:
Loaded symbols for /usr/local/lib/openldap/dynlist-2.3-releng.so.0
#0 0x00002ab7651490c9 in malloc () from /usr/local/lib/libhoard.so
(gdb) bt
#0 0x00002ab7651490c9 in malloc () from /usr/local/lib/libhoard.so
#1 0x00002ab765149e38 in realloc () from /usr/local/lib/libhoard.so
#2 0x00002ab76554f890 in ber_memrealloc_x (p=Variable "p" is not available.
) at memory.c:354
#3 0x00002ab76542bf88 in ldap_charray_add (a=0x7fff45a7eec0, s=0x2ab7667c2518
"olcConnMaxPending") at charray.c:49
#4 0x000000000045e9f9 in matching_rule_use_init () at mr.c:364
#5 0x0000000000459f62 in slap_schema_check () at schema_prep.c:1368
#6 0x0000000000415f1b in main (argc=3, argv=0x7fff45a7f0b8) at main.c:711
It looks like there's soemthing incorrect in the way it is allocating memory on
startup.
--Quanah
quanah(a)stanford.edu wrote:
>> I don't know if the people who are in charge to build a RPM are paying
>> attention to what they are doing or not.
>>
>> You can do a man ldap_init, and then look at /usr/include/ldap.h for the
>> function definition.
>>
>> I am not a LDAP expert or anything close to that. But it seems to me that
>> function called ldap_init has already been replaced by the function called
>> ldap_initialize, and they take even different set of parameters.
>>
>> And yet the manpage not only didn't reflect that change, but enforce the
>> wrong idea of using ldap_init function by having ldap_init in manpages.
>>
>> An extremely poor job done by the package maintainer.
>
> Hi,
>
> I suggest you complain to RedHat if you don't like the way they've packaged
> something. The OpenLDAP foundation is not responsible for the way in which
> RedHat packages its releases. In any case, if you are taking the time to
> read the header file, then pay close attention to the following:
>
> #if LDAP_DEPRECATED
> LDAP_F( LDAP * )
> ldap_init LDAP_P(( /* deprecated, use ldap_create or ldap_initialize */
> LDAP_CONST char *host,
> int port ));
>
>
> I.e., ldap_init is a deprecated function, and ldap_initialize should be
> used instead. However, there are many programs out there that still use
> the old API, so the old functions are maintained for backwards
> compatibility purposes. Hence the man page.
I'd add that OpenLDAP is a voluntary based project, and volunteers
devote their efforts to what they consider mostly important - typically
writing good useful code that runs (fast). Man pages typically lag
behind, but eventually they get updated/fixed (as they are in currently
released code). If you cared so much about up-to-date code and man
pages, you would be using the latest release, not a 15 months old one.
And if you were able to infer the really useful API from ldap.h, you
could have more proficiently spent your time by writing and submitting a
useful man page, instead of just complaining quite unpolitely.
p.
Ing. Pierangelo Masarati
OpenLDAP Core Team
SysNet s.n.c.
Via Dossi, 8 - 27100 Pavia - ITALIA
http://www.sys-net.it
------------------------------------------
Office: +39.02.23998309
Mobile: +39.333.4963172
Email: pierangelo.masarati(a)sys-net.it
------------------------------------------
--On Monday, April 02, 2007 5:23 PM +0000 wweng(a)kencast.com wrote:
> Full_Name: Wei Weng
> Version: 2.3.19
> OS: Linux(Fedora 5)
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (69.120.101.108)
>
>
> I don't know if the people who are in charge to build a RPM are paying
> attention to what they are doing or not.
>
> You can do a man ldap_init, and then look at /usr/include/ldap.h for the
> function definition.
>
> I am not a LDAP expert or anything close to that. But it seems to me that
> function called ldap_init has already been replaced by the function called
> ldap_initialize, and they take even different set of parameters.
>
> And yet the manpage not only didn't reflect that change, but enforce the
> wrong idea of using ldap_init function by having ldap_init in manpages.
>
> An extremely poor job done by the package maintainer.
Hi,
I suggest you complain to RedHat if you don't like the way they've packaged
something. The OpenLDAP foundation is not responsible for the way in which
RedHat packages its releases. In any case, if you are taking the time to
read the header file, then pay close attention to the following:
#if LDAP_DEPRECATED
LDAP_F( LDAP * )
ldap_init LDAP_P(( /* deprecated, use ldap_create or ldap_initialize */
LDAP_CONST char *host,
int port ));
I.e., ldap_init is a deprecated function, and ldap_initialize should be
used instead. However, there are many programs out there that still use
the old API, so the old functions are maintained for backwards
compatibility purposes. Hence the man page.
This ITS will be closed.
Regards,
Quanah
--
Quanah Gibson-Mount
Senior Systems Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html
Full_Name: Wei Weng
Version: 2.3.19
OS: Linux(Fedora 5)
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (69.120.101.108)
I don't know if the people who are in charge to build a RPM are paying attention
to what they are doing or not.
You can do a man ldap_init, and then look at /usr/include/ldap.h for the
function definition.
I am not a LDAP expert or anything close to that. But it seems to me that
function called ldap_init has already been replaced by the function called
ldap_initialize, and they take even different set of parameters.
And yet the manpage not only didn't reflect that change, but enforce the wrong
idea of using ldap_init function by having ldap_init in manpages.
An extremely poor job done by the package maintainer.
--On Saturday, March 31, 2007 12:54 AM +0000 quanah(a)stanford.edu wrote:
> --On Friday, March 30, 2007 2:25 PM +0000 chris.ridd(a)isode.com wrote:
>
>> Full_Name: Chris Ridd
>> Version: 2.3.34
>> OS: Windows
>> URL: ftp://ftp.openldap.org/incoming/
>> Submission from: (NULL) (62.3.217.250)
>>
>>
>> On Windows, it is important that mallocs done in one DLL are balanced by
>> frees done from the same DLL. Failure to do this consistently leads to
>> heap corruption.
>>
>> This was observed to occur in several places in ldapadd - several buffers
>> allocated by ber_memfree() were freed using free().
>>
>> Our customer has successfully used a version of ldapadd on Windows
>> containing this patch (and the one in ITS 4900.)
>
> Thanks, testing.
--- openldap-2.3.34/clients/tools/common.c.orig Fri Mar 30 15:17:32 2007
+++ openldap-2.3.34/clients/tools/common.c Fri Mar 30 15:21:42 2007
@@ -1244,7 +1244,7 @@
crit ? "critical " : "" );
}
- free( ctrls );
+ ldap_controls_free( ctrls );
if ( crit ) {
exit( EXIT_FAILURE );
}
Just to note, this part of the patch is bad. ctrls is a pointer to an
array of controls, not an array of controls itself. Therefore, the use of
"free" here is appropriate. Using ldap_controls_free results in segfaults.
--Quanah
--
Quanah Gibson-Mount
Senior Systems Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html
--On Friday, March 30, 2007 2:25 PM +0000 chris.ridd(a)isode.com wrote:
> Full_Name: Chris Ridd
> Version: 2.3.34
> OS: Windows
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (62.3.217.250)
>
>
> On Windows, it is important that mallocs done in one DLL are balanced by
> frees done from the same DLL. Failure to do this consistently leads to
> heap corruption.
>
> This was observed to occur in several places in ldapadd - several buffers
> allocated by ber_memfree() were freed using free().
>
> Our customer has successfully used a version of ldapadd on Windows
> containing this patch (and the one in ITS 4900.)
Thanks, testing.
--Quanah
--
Quanah Gibson-Mount
Senior Systems Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html
--On Friday, March 30, 2007 2:09 PM +0000 chris.ridd(a)isode.com wrote:
> Full_Name: Chris Ridd
> Version: 2.3.34
> OS: Windows
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (62.3.217.250)
>
>
> On Windows, the ldapadd tool often crashes while parsing the command-line
> arguments.
Hi Chris,
We formally dropped support for MSVC a long time ago. That being said,
working build of OpenLDAP was accomplished with MSVCC a few months ago
without hitting this issue. Which version of MSVC is being used, and
please doublecheck the conditions spelled out in the comments in
<ldap_cdefs.h>, where it starts with "Support for Windows DLLs".
Also, instead of declaring things as extern, does changing LDAP_LIBC_V to
LDAP_LUTIL_V resolve your issue?
--Quanah
--
Quanah Gibson-Mount
Senior Systems Software Developer
ITS/Shared Application Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html