- openldap-bugs - openldap.org

Re: (ITS#7918) error building back-ldap
by hyc＠symas.com 13 Aug '14

13 Aug '14

dieter(a)dkluenter.de wrote: > Full_Name: Dieter Klünter > Version: openldap-OPENLDAP_REL_ENG_2_4-b046124 > OS: openSuse-13.1 > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (93.220.56.191) > > > cd back-ldap; make -w all > make[3]: Entering directory > `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd/back-ldap' > rm -f version.c > ../../../build/mkversion -v "2.4.X" back_ldap > version.c > /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 > -m64 -I../../../include -I../../../include -I.. -I./.. -c init.c Don't define LDAP_DEVEL unless you know what you're doing. Closing this ITS. > cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c > init.c -o init.o > /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 > -m64 -I../../../include -I../../../include -I.. -I./.. -c config.c > cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c > config.c -o config.o > /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 > -m64 -I../../../include -I../../../include -I.. -I./.. -c search.c > cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c > search.c -o search.o > /bin/sh %2/../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 > -m64 -I../../../include -I../../../include -I.. -I./.. -c bind.c > cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c > bind.c -o bind.o > bind.c: In function 'ldapack_k_proxy_authz_bind': > bind.c:2325:32: error: 'LDAP_BACK_AUTH_DN_AUTHZID' undeclared (first use in this > function) > if ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_AUTHZID ) { > ^ > bind.c:2325:32: note: each undeclared identifier is reportednlnly once for each > function it appears in > bind.c:2328:21: error: 'LDAP_CONTROL_AUTHZID_REQUEST' undeclared (first use in > this function) > ctrl.ldctl_oid = LDAP_CONTROL_AUTHZID_REQUEST; > ^ > bind.c:2371:32: error: 'LDAP_CONTROL_AUTHZID_RESPONSE' undeclared (first use in > this function) > ctrl = ldap_control_find( LDAP_CONTROL_AUTHZID_RESPONSE, > ^ > bind.c:2389:40: error: 'LDAP_BACK_AUTH_DN_WHOAMI' undeclared (first use in this > function) > } else if ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_WHOAMI ) { > ^ > bind.c:2407:35: error: 'LDAP_BACK_AUTH_DN_MASK' undeclared (first use in this > function) > if ( ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_MASK ) && > ^ > make[3]: *** [bind.lo] Fehler 1 > make[3]: Leaving directory > `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd/back-ldap' > make[2]: *** [.backend] Fehler 1 > make[2]: Leaving directory > `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd' > make[1]: *** [all-common] Fehler 1 > make[1]: Leaving directory > `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers' > make: *** [all-common] Fehler 1 > > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7919) slapd would not start with back-ldap database
by hyc＠symas.com 13 Aug '14

13 Aug '14

dieter(a)dkluenter.de wrote: > Full_Name: Dieter > Version: 2.4.39 > OS: openSuse-13.1 > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (62.226.250.246) > > > When slapd configured with database ldap only, an error occurs, > > gdb) bt > #0 0x00007ffff7bb8038 in get_token (sp=sp@entry=0x7fffffff9590, > token_val=token_val@entry=0x7fffffff9598) > at schema.c:1018 > #1 0x00007ffff7bbc6f2 in ldap_str2objectclass (s=s@entry=0x7 <Address 0x7 out > of bounds>, > code=code@entry=0x7fffffff95f4, errp=errp@entry=0x7fffffff95f8, > flags=flags@entry=63) at schema.c:2461 > #2 0x00005555555e7c1c in register_oc (def=0x7 <Address 0x7 out of bounds>, > soc=0x7ffff30a59d0 <pcocs+112>, > dupok=dupok@entry=1) at oc.c:917 > #3 0x00005555555935ee in init_config_ocs (ocs=ocs@entry=0x7ffff30a5960 <pcocs>) > at config.c:556 > #4 0x000055555558bf18 in config_register_schema (ct=0x7ffff30a54c0 <pccfg>, > ocs=0x7ffff30a5960 <pcocs>) > > The trace indicates something initializing the proxycache overlay, which contradicts your statement that only back-ldap was configured. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

(ITS#7919) slapd would not start with back-ldap database
by dieter＠dkluenter.de 13 Aug '14

13 Aug '14

Full_Name: Dieter Version: 2.4.39 OS: openSuse-13.1 URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (62.226.250.246) When slapd configured with database ldap only, an error occurs, gdb) bt #0 0x00007ffff7bb8038 in get_token (sp=sp@entry=0x7fffffff9590, token_val=token_val@entry=0x7fffffff9598) at schema.c:1018 #1 0x00007ffff7bbc6f2 in ldap_str2objectclass (s=s@entry=0x7 <Address 0x7 out of bounds>, code=code@entry=0x7fffffff95f4, errp=errp@entry=0x7fffffff95f8, flags=flags@entry=63) at schema.c:2461 #2 0x00005555555e7c1c in register_oc (def=0x7 <Address 0x7 out of bounds>, soc=0x7ffff30a59d0 <pcocs+112>, dupok=dupok@entry=1) at oc.c:917 #3 0x00005555555935ee in init_config_ocs (ocs=ocs@entry=0x7ffff30a5960 <pcocs>) at config.c:556 #4 0x000055555558bf18 in config_register_schema (ct=0x7ffff30a54c0 <pccfg>, ocs=0x7ffff30a5960 <pcocs>)

1 0

(ITS#7918) error building back-ldap
by dieter＠dkluenter.de 12 Aug '14

12 Aug '14

Full_Name: Dieter Klünter Version: openldap-OPENLDAP_REL_ENG_2_4-b046124 OS: openSuse-13.1 URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (93.220.56.191) cd back-ldap; make -w all make[3]: Entering directory `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd/back-ldap' rm -f version.c ../../../build/mkversion -v "2.4.X" back_ldap > version.c /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c init.c cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c init.c -o init.o /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c config.c cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c config.c -o config.o /bin/sh ../../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c search.c cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c search.c -o search.o /bin/sh %2/../../libtool --tag=disable-shared --mode=compile cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c bind.c cc -DLDAP_DEVEL -g3 -m64 -I../../../include -I../../../include -I.. -I./.. -c bind.c -o bind.o bind.c: In function 'ldapack_k_proxy_authz_bind': bind.c:2325:32: error: 'LDAP_BACK_AUTH_DN_AUTHZID' undeclared (first use in this function) if ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_AUTHZID ) { ^ bind.c:2325:32: note: each undeclared identifier is reportednlnly once for each function it appears in bind.c:2328:21: error: 'LDAP_CONTROL_AUTHZID_REQUEST' undeclared (first use in this function) ctrl.ldctl_oid = LDAP_CONTROL_AUTHZID_REQUEST; ^ bind.c:2371:32: error: 'LDAP_CONTROL_AUTHZID_RESPONSE' undeclared (first use in this function) ctrl = ldap_control_find( LDAP_CONTROL_AUTHZID_RESPONSE, ^ bind.c:2389:40: error: 'LDAP_BACK_AUTH_DN_WHOAMI' undeclared (first use in this function) } else if ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_WHOAMI ) { ^ bind.c:2407:35: error: 'LDAP_BACK_AUTH_DN_MASK' undeclared (first use in this function) if ( ( li->li_idassert_flags & LDAP_BACK_AUTH_DN_MASK ) && ^ make[3]: *** [bind.lo] Fehler 1 make[3]: Leaving directory `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd/back-ldap' make[2]: *** [.backend] Fehler 1 make[2]: Leaving directory `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers/slapd' make[1]: *** [all-common] Fehler 1 make[1]: Leaving directory `/home/dieter/build/openldap-OPENLDAP_REL_ENG_2_4-b046124/servers' make: *** [all-common] Fehler 1

1 0

(ITS#7917) LMDB: bug in mdb_dbi_open creating subDB
by hyc＠openldap.org 11 Aug '14

11 Aug '14

Full_Name: Howard Chu Version: 2.4 OS: Solaris/Sparc URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (78.155.233.73) Submitted by: hyc mdb_dbi_open may use "dummy" after it has gone out of scope, when creating a new named subDB. This never caused any problem under gcc, but breaks using the Sun C compiler when debugging is enabled and optimization is disabled. A fix is coming shortly.

1 0

Re: (ITS#7916) ppolicy doesn't set pwdAccountLockedTime
by quanah＠zimbra.com 06 Aug '14

06 Aug '14

--On Wednesday, August 06, 2014 11:26 AM +0000 anshman.osc(a)gmail.com wrote: > Full_Name: Anshuman > Version: 2.4.23 2.4.23 is 4+ years old. In addition, the ITS system is for filing bugs, not for asking usage questions. You need to (a) upgrade to a current release or (b) contact the vendor of your massively out of date packages. For (a), I suggest packages from either Symas or the LTB project if you are not able to build OpenLDAP yourself. This ITS will be closed. --Quanah -- Quanah Gibson-Mount Server Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration

1 0

(ITS#7916) ppolicy doesn't set pwdAccountLockedTime
by anshman.osc＠gmail.com 06 Aug '14

06 Aug '14

Full_Name: Anshuman Version: 2.4.23 OS: RHEL 6.4 URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (198.241.211.15) Hello, I am trying to get the ppolicy to lock account after N unsuccessful attempts. To accomplish this, I defined the overlay policy in slapd.conf, and also attached the pwdPolicySubentry to the user object. It is able to detect the password policy, because the number of times "pwdFailureTime" appears is always 1 less than the value I set for "pwdMaxFailure" in the password policy. So, if I set pwdMaxFailure=4, the count pwdFailureTime stops growing after 3. However, the pwdAccountLockedTime is never set. Up until release 2.3.x adding a rootdn entry to the slapd.conf solved this issue. But today we are trying to upgrade to 2.4.23, and this "fix" no longer works. Could someone please let me know what needs to be done to make this work? -- slapd.conf--- # Load dynamic backend modules: modulepath /usr/lib64/openldap moduleload ppolicy.la moduleload auditlog.la overlay ppolicy ppolicy_default "cn=Standard,ou=Policies,dc=mycompany,dc=com" ppolicy_use_lockout

1 0

Re: (ITS#7906) segv when changing olcDbDirectory with ppolicy running
by ebackes＠symas.com 06 Aug '14

06 Aug '14

> While testing out the new mdb_copy compaction, I tried switching a live instance > to use a different db path via MOD to change olcDbDirectory. > This results in repeatable segv in ppolicy.c:ppolicy_connection_destroy, where > conn refers to the correction making the change that is likely in the process of > disconnecting, and pwcons is NULL. There are more pwcons hits than just ppolicy_connection_destroy, looking for the others but at a minimum ppolicy_restrict. -- Emily Backes Symas Corporation ebackes(a)symas.com

1 0

Re: (ITS#7915) pcache and translucent crashes OL
by hyc＠symas.com 04 Aug '14

04 Aug '14

nvoutsin(a)gmail.com wrote: > Full_Name: Nikos Voutsinas > Version: 2.4.39 > OS: Debian > URL: http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(bdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(bdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(bdb)-pcache%mdmd… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(bdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(mdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(mdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(mdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(mdb)-pcache(bdb)… > Submission from: (NULL) (5.54.11.83) Thanks for the report, fixed in git master. > The combination of translucent and pcache overlays crashes OL or makes it > unresponsive. Any attempt to modify an object on the replica by using the > following ldif: > > dn: uid=joe,dc=foo,dc=com > changetype: modify > add: description > description: first > - > > dn: uid=joe,dc=foo,dc=com > changetype: modify > replace: description > description: second > - > > cause OL to fail either during the modification or on the subsequent operation. > Specifically, OL behaves differently depending on the type of database backends > that are used for translucent and pcache DBs, as it is shown below: > > translucent db,pcache db: comment > bdb,bdb: freezes on the second modify op > bdb,mdb: crashes on the secend modify op (invalid pointer) > mdb,mdb: completes the modify ops. crashes on the first search on object > mdb,bdb: completes the modify ops. crashes, if not on the first, then on > subsequent op > > Attached files: a)the cn=config ldif for each of the above cases and b)the > corresponding debug output during the modification of the user object or if that > completes during the subsequent search on it. > > > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7915) pcache and translucent crashes OL
by quanah＠zimbra.com 04 Aug '14

04 Aug '14

--On Monday, August 04, 2014 10:16 PM +0300 Nikos Voutsinas=20 <nvoutsin(a)gmail.com> wrote: > I am aware of #7587, however based on the release dates shouldn't > this=C2=A0fix be=C2=A0included in 2.4.39; I just told you specifically that the fix was made after 2.4.39 was=20 released. Again, please test current RE24. Thanks! --Quanah -- Quanah Gibson-Mount Server Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration

1 0

Re: (ITS#7915) pcache and translucent crashes OL
by nvoutsin＠gmail.com 04 Aug '14

04 Aug '14

--001a11364318a34ed604ffd1be1c Content-Type: text/plain; charset=UTF-8 I am aware of #7587, however based on the release dates shouldn't this fix be included in 2.4.39; On Monday, August 4, 2014, Quanah Gibson-Mount <quanah(a)zimbra.com> wrote: > --On Monday, August 04, 2014 1:04 AM +0000 nvoutsin(a)gmail.com wrote: > > Full_Name: Nikos Voutsinas >> Version: 2.4.39 >> OS: Debian >> > > This looks like a duplicate of ITS#7587. Can you please re-test with > current code from RE24 where this is fixed? > > Thanks! > > --Quanah > > > > -- > > Quanah Gibson-Mount > Server Architect > Zimbra, Inc. > -------------------- > Zimbra :: the leader in open source messaging and collaboration > --001a11364318a34ed604ffd1be1c Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable I am aware of #7587, however based on the release dates shouldn't this= =C2=A0fix be=C2=A0<span></span>included in 2.4.39;<br><br>On Monday, August= 4, 2014, Quanah Gibson-Mount <<a href=3D"mailto:quanah@zimbra.com">quan= ah(a)zimbra.com</a>> wrote:<br> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex">--On Monday, August 04, 2014 1:04 AM +0000 <= a>nvoutsin(a)gmail.com</a> wrote:<br> <br> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex"> Full_Name: Nikos Voutsinas<br> Version: 2.4.39<br> OS: Debian<br> </blockquote> <br> This looks like a duplicate of ITS#7587. =C2=A0Can you please re-test with = current code from RE24 where this is fixed?<br> <br> Thanks!<br> <br> --Quanah<br> <br> <br> <br> --<br> <br> Quanah Gibson-Mount<br> Server Architect<br> Zimbra, Inc.<br> --------------------<br> Zimbra :: =C2=A0the leader in open source messaging and collaboration<br> </blockquote> --001a11364318a34ed604ffd1be1c--

1 0

Re: (ITS#7915) pcache and translucent crashes OL
by quanah＠zimbra.com 04 Aug '14

04 Aug '14

--On Monday, August 04, 2014 1:04 AM +0000 nvoutsin(a)gmail.com wrote: > Full_Name: Nikos Voutsinas > Version: 2.4.39 > OS: Debian This looks like a duplicate of ITS#7587. Can you please re-test with current code from RE24 where this is fixed? Thanks! --Quanah -- Quanah Gibson-Mount Server Architect Zimbra, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration

1 0

(ITS#7915) pcache and translucent crashes OL
by nvoutsin＠gmail.com 04 Aug '14

04 Aug '14

Full_Name: Nikos Voutsinas Version: 2.4.39 OS: Debian URL: http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(bdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(bdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(bdb)-pcache%mdmd… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(bdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(mdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(mdb)-pcache(mdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-trasnlucent(mdb)-pcache(bdb)… http://users.uoa.gr/~nvoutsin/OpenLdap_ITS/add-translucent(mdb)-pcache(bdb)… Submission from: (NULL) (5.54.11.83) The combination of translucent and pcache overlays crashes OL or makes it unresponsive. Any attempt to modify an object on the replica by using the following ldif: dn: uid=joe,dc=foo,dc=com changetype: modify add: description description: first - dn: uid=joe,dc=foo,dc=com changetype: modify replace: description description: second - cause OL to fail either during the modification or on the subsequent operation. Specifically, OL behaves differently depending on the type of database backends that are used for translucent and pcache DBs, as it is shown below: translucent db,pcache db: comment bdb,bdb: freezes on the second modify op bdb,mdb: crashes on the secend modify op (invalid pointer) mdb,mdb: completes the modify ops. crashes on the first search on object mdb,bdb: completes the modify ops. crashes, if not on the first, then on subsequent op Attached files: a)the cn=config ldif for each of the above cases and b)the corresponding debug output during the modification of the user object or if that completes during the subsequent search on it.

1 0

Re: (ITS#7914) LMDB: MDB_DUPSORT and mdb_cursor_put(.., MDB_CURRENT) can put DB in illegal state
by h.b.furuseth＠usit.uio.no 03 Aug '14

03 Aug '14

On 03/08/14 20:13, hyc(a)symas.com wrote: > The recent doc changes are a bit over-zealous here. The only requirement is > that the new value would naturally sort into the same position. So yes, you > can store a value different than the old value. No, MDB_CURRENT does not update branch keys. That was the reason for those latest doc changes.

1 0

Re: (ITS#7914) LMDB: MDB_DUPSORT and mdb_cursor_put(.., MDB_CURRENT) can put DB in illegal state
by hyc＠symas.com 03 Aug '14

03 Aug '14

Max Bolingbroke wrote: > On 2 August 2014 17:43, Howard Chu <hyc(a)symas.com <mailto:hyc@symas.com>> wrote: > > Max Bolingbroke wrote: > > OK, that does make sense. So MDB_CURRENT is totally useless with > MDB_DUPSORT? > Since you are either pointing at the (k, v) pair already (so the put > will do > nothing) or you aren't (so the put will break the DB). > > > The feature would not exist if it were useless. > > > An alternative explanation for the existence of MDB_CURRENT could be that it > is useful without MDB_DUPSORT (which is certainly is!) but is useless with it > on. I take it from this remark that you do see some use for the combination of > the two options but I guess I must be too stupid to see what that is! > > I suppose that MDB_CURRENT might be useful with MDB_DUPSORT if you could use > it to replace the (key, value) pair being pointed to by the cursor with a > different (key, value) pair, so long as the LMDB user knew that the > replacement would not violate the LMDB invariant that duplicates must be > sorted and distinct. Though this use case "seems" to work with the current > implementation the docs certainly do not promise that this is supported. What defines a "match" depends on the comparison function. If you use a comparator that only uses a portion of the value then of course you can store different values that still match. Prefix comparison is the most obvious example. The recent doc changes are a bit over-zealous here. The only requirement is that the new value would naturally sort into the same position. So yes, you can store a value different than the old value. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7894) LMDB: MDB_DUPSORT and mdb_cursor_put(.., MDB_CURRENT) can put DB in illegal state
by batterseapower＠hotmail.com 03 Aug '14

03 Aug '14

--001a11c2f26695c81a04ffbb03f9 Content-Type: text/plain; charset=UTF-8 On 2 August 2014 17:43, Howard Chu <hyc(a)symas.com> wrote: > Max Bolingbroke wrote: > >> OK, that does make sense. So MDB_CURRENT is totally useless with >> MDB_DUPSORT? >> Since you are either pointing at the (k, v) pair already (so the put will >> do >> nothing) or you aren't (so the put will break the DB). >> > > The feature would not exist if it were useless. An alternative explanation for the existence of MDB_CURRENT could be that it is useful without MDB_DUPSORT (which is certainly is!) but is useless with it on. I take it from this remark that you do see some use for the combination of the two options but I guess I must be too stupid to see what that is! I suppose that MDB_CURRENT might be useful with MDB_DUPSORT if you could use it to replace the (key, value) pair being pointed to by the cursor with a different (key, value) pair, so long as the LMDB user knew that the replacement would not violate the LMDB invariant that duplicates must be sorted and distinct. Though this use case "seems" to work with the current implementation the docs certainly do not promise that this is supported. > Do you think it would be better to error out when the supplied key/data >> doesn't match the data pointed to by the key? That would protect the DB >> from >> this kind of corruption in the presence of bugs in the application. >> > > We don't attempt to protect from user errors; we expect competent > programmers. Is that really true? For example your docs from MDB_WRITEMAP say: """Use a writeable memory map unless MDB_RDONLY is set. This is faster and uses fewer mallocs, but loses protection from application bugs like wild pointer writes and other bad updates into the database.""" So LMDB's default behaviour is actually to take a perf hit in the name of protecting programmers from themselves. My proposal is consistent with that philosophy. > p.s. you may want to add a note to the docs that MDB_RESERVE is not >> allowed >> with MDB_DUPSORT. The combination doesn't seem to make a lot of sense >> > > Indeed it makes no sense at all. And no, we don't attempt to protect from > nonsensical usage. Except that in e.g. mdb_put you explicitly check for nonsensical flags making it through to mdb_cursor_put: if ((flags & (MDB_NOOVERWRITE|MDB_NODUPDATA|MDB_RESERVE|MDB_APPEND|MDB_APPENDDUP)) != flags) return EINVAL; A check for the nonsensical combination of MDB_RESERVE and MDB_DUPSORT would be in that spirit. --001a11c2f26695c81a04ffbb03f9 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On 2= August 2014 17:43, Howard Chu <span dir=3D"ltr"><<a href=3D"mailto:hyc@= symas.com" target=3D"_blank">hyc(a)symas.com</a>></span> wrote:<br><blockq= uote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-wi= dth:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-= left:1ex"> <div class=3D"">Max Bolingbroke wrote:<br> <blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-= left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p= adding-left:1ex"> OK, that does make sense. So MDB_CURRENT is totally useless with MDB_DUPSOR= T?<br> Since you are either pointing at the (k, v) pair already (so the put will d= o<br> nothing) or you aren't (so the put will break the DB).<br> </blockquote> <br></div> The feature would not exist if it were useless.</blockquote><div><br></div>= <div>An alternative explanation for the existence of MDB_CURRENT could be t= hat it is useful without MDB_DUPSORT (which is certainly is!) but is useles= s with it on. I take it from this remark that you do see some use for the c= ombination of the two options but I guess I must be too stupid to see what = that is!</div> <div><br>I suppose that MDB_CURRENT might be useful with MDB_DUPSORT if you= could use it to replace the (key, value) pair being pointed to by the curs= or with a different (key, value) pair, so long as the LMDB user knew that t= he replacement would not violate the LMDB invariant that duplicates must be= sorted and distinct. Though this use case "seems" to work with t= he current implementation the docs certainly do not promise that this is su= pported.</div> <div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px = 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l= eft-style:solid;padding-left:1ex"><div class=3D""> <blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-= left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p= adding-left:1ex"> Do you think it would be better to error out when the supplied key/data<br> doesn't match the data pointed to by the key? That would protect the DB= from<br> this kind of corruption in the presence of bugs in the application.<br> </blockquote> <br></div> We don't attempt to protect from user errors; we expect competent progr= ammers.</blockquote><div><br></div><div>Is that really true? For example yo= ur docs from MDB_WRITEMAP say:</div><div><br></div><div><div>""&q= uot;Use a writeable memory map unless MDB_RDONLY is set. This is faster</di= v> <div>and uses fewer mallocs, but loses protection from application bugs</di= v><div>like wild pointer writes and other bad updates into the database.&qu= ot;""</div></div><div><br></div><div>So LMDB's default behavi= our is actually to take a perf hit in the name of protecting programmers fr= om themselves. My proposal is consistent with that philosophy.</div> <div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px = 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l= eft-style:solid;padding-left:1ex"><div class=3D""> <blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-= left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p= adding-left:1ex"> p.s. you may want to add a note to the docs that MDB_RESERVE is not allowed= <br> with MDB_DUPSORT. The combination doesn't seem to make a lot of sense<b= r> </blockquote> <br></div> Indeed it makes no sense at all. And no, we don't attempt to protect fr= om nonsensical usage.</blockquote><div><br></div><div>Except that in e.g. m= db_put you explicitly check for nonsensical flags making it through to mdb_= cursor_put:</div> <div><br></div><div><div>=C2=A0 =C2=A0 if ((flags & (MDB_NOOVERWRITE|MD= B_NODUPDATA|MDB_RESERVE|MDB_APPEND|MDB_APPENDDUP)) !=3D flags)</div><div><s= pan class=3D"" style=3D"white-space:pre"> </span>return EINVAL;</div></div= ><div><br></div> <div>A check for the nonsensical combination of MDB_RESERVE and MDB_DUPSORT= would be in that spirit.</div></div></div></div> --001a11c2f26695c81a04ffbb03f9--

1 0

Re: (ITS#7914) LMDB: MDB_DUPSORT and mdb_cursor_put(.., MDB_CURRENT) can put DB in illegal state
by hyc＠symas.com 02 Aug '14

02 Aug '14

Due to an error in the transition from old to new OpenLDAP.org server, your ITS was erroneously assigned #7894 which was already allocated. Just following up here to let you know the correct ITS#. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

(ITS#7894) LMDB: MDB_DUPSORT and mdb_cursor_put(.., MDB_CURRENT) can put DB in illegal state
by batterseapower＠hotmail.com 01 Aug '14

01 Aug '14

Full_Name: Max Bolingbroke Version: LMDB HEAD OS: OS X URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (149.254.58.67) It is possible to insert a (key, value) pair into an MDB_DUPSORTed LMDB database twice using mdb_cursor_put(..., MDB_CURRENT). If you attempt to insert the same (key, value) pair twice with mdb_put then the second insert has no effect. This seems like a bug in mdb_cursor_put. Furthermore, you can use mdb_cursor_put(..., MDB_CURRENT) to insert duplicates in the opposite of their correctly sorted order. This is certainly a bug. The following C program demonstrates both issues: #define _XOPEN_SOURCE 500 #include <stdio.h> #include <stdlib.h> #include <unistd.h> #include <stdarg.h> #include "lmdb.h" void panic(const char * restrict format, ...) { va_list argp; va_start(argp, format); vprintf%foformat, argp); va_end(argp); exit(1); } void mdb_check(char *context, int rc) { if (rc != 0) { panic("%s: %s", context, mdb_strerror(rc)); } } void mdb_check_notfound(char *context, int rc) { if (rc != MDB_NOTFOUND) { panic("%s: %s", context, mdb_strerror(rc)); } } int BE(int x) { return (x & 0x000000FF) << 24 | (x & 0x0000FF00) << 8 | (x & 0x00FF0000) >> 8 | (x & 0xFF000000) >> 24; } int main(int argc, char **argv) { srand(1337); MDB_e * *env; mdb_check("mdb_env_create", mdb_env_create(&env)); mdb_check("mdb_env_set_maxdbs", mdb_env_set_maxdbs(env, 1)); mdb_check("mdb_env_set_mapsize", mdb_env_set_mapsize(env, 1024 * 1024)); mdb_check("mdb_env_open", mdb_env_open(env, "testdb", MDB_NOSUBDIR, 0644)); MDB_txn *txn; mdb_check("mdb_txn_begin", mdb_txn_begin(env, NULL, 0, &txn)); MDB_dbi dbi; mdb_check("mdb_dbi_open", mdb_dbi_open(txn, "root", MDB_CREATE | MDB_DUPSORT, &dbi)); int key; int value; MDB_val keyBuffer; keyBuffer.mv_size = sizeof(int); keyBuffer.mv_data = &key; MDB_val valueBuffer; valueBuffer.mv_size = sizeof(int); valueBuffer.mv_data = &value; { key = BE(2); value = BE(200); mdb_check("mdb_put", mdb_put(txn, dbi, &keyBuffer, &valueBuffer, 0)); key = BE(2); value = BE(300); mdb_check("mdb_put", mdb_put(txn, dbi, &keyBuffer, &valueBuffer, 0)); // Inserting the same (key, value) pair twice: second insert is ignored key = BE(2); value = BE(300); mdb_check("mdb_put", mdb_put(txn, dbi, &keyBuffer, &valueBuffer, 0)); } { MDB_cursor *cursor; mdb_check("mdb_cursor_open", mdb_cursor_open(txn, dbi, &cursor)); size_t count; mdb_check("mdb_cursor_get", mdb_cursor_get(cursor, &keyBuffer, &valueBuffer, MDB_FIRST))0D0D mdb_check("mdb_cursor_count", mdb_cursor_count(cursor, &count)); if (count != 2) panic("count != 2"); if (0) { // Inserting the same (key, value) pair twice: second insert is not ignored. // While this is different to mdb_put, I can't tell from the docs whether this is a bug or not. keyBuffer.mv_size = sizeof(int); valueBuffer.mv_size = sizeof(int); keyBuffer.mv_data = &key; valueBuffer.mv_data = &value; key = BE(2); value = BE(300); mdb_check("mdb_cursor_put2,2C mdb_cursor_put(cursor, &keyBuffer, &valueBuffer, MDB_CURRENT)); mdb_check("mdb_cursor_count", mdb_cursor_count(cursor, &count)); if (count != 1) panic("count != 1"); } else { // This on the other hand is definitely a bug: after doing this the duplicates are not sorted. keyBuffer.mv_size = sizeof(int); valueBuffer.mv_size = sizeof(int); keyBuffer.mv_data = &key; valueBuffer.mv_data = &value; key = BE(2); value = BE(400); mdb_check("mdb_cursor_put", mdb_cursor_p(c8cursor, &keyBuffer, &valueBuffer, MDB_CURRENT)); mdb_check("mdb_cursor_count", mdb_cursor_count(cursor, &count)); if (count != 2) panic("count != 2"); mdb_check("mdb_cursor_get", mdb_cursor_get(cursor, &keyBuffer, &valueBuffer, MDB_FIRST)); printf("%d\n", BE(*((int*)valueBuffer.mv_data))); mdb_check("mdb_cursor_get", mdb_cursor_get(cursor, &keyBuffer, &valueBuffer, MDB_NEXT)); printf("%d\n", BE(*((int*)valueBuffer.mv_data))); mdb_check_notfound("mdb_cursor_get", mdb_cursor_get(cursor, &keyBuffer, &valueBuffer, MDB_NEXT)); } mdb_cursor_close(cursor); } mdb_check("mdb_txn_commit", mdb_txn_commit(txn)); return 0; }

1 0

Re: (ITS#7713) Segmentation fault if the pagesize of the Operating system is not equal to 4096.
by h.b.furuseth＠usit.uio.no 09 Jul '14

09 Jul '14

On 09/07/14 01:41, hyc(a)symas.com wrote: > The fix is now available in mdb.master; currently you must compile with > -DMDB_DEVEL to enable it. Since this is an incompatible DB format change, you > should backup any existing DBs before using this option. Well, there is a fix in LDMB 0.9.10: It uses pages <= 32k, doing less efficient write operations if that's less than the OS page size. -- Hallvard

1 0

Re: (ITS#7893) memberof: bad a_numvals count
by hyc＠symas.com 09 Jul '14

09 Jul '14

jsoula(a)univ-lille2.fr wrote: > Full_Name: julien soula > Version: 2.4.39 > OS: > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (194.254.117.28) > > > Hi, > > this is related to ITS#6508. I don't think that the supplied patch is enough to > fix the bug. > > When the last "member" entry is dangling then "( j - i == 1 )" is true and the > loop is broken without reducing a_numvals. Looks like you're right. Committed to master, thanks. > > So I propose this patch : > > --- servers/slapd/overlays/memberof.c.orig 2014-07-09 21:17:14.487296363 > +0200 > +++ servers/slapd/overlays/memberof.c 2014-07-09 21:31:44.071565223 +0200 > @@ -596,6 +596,7 @@ > ber_memfree( a->a_nvals[ i > ].bv_val ); > BER_BVZERO( &a->a_nvals[ i ] ); > } > + a->a_numvals--; > if ( j - i == 1 ) { > break; > } > @@ -607,7 +608,6 @@ > sizeof( struct berval ) > * ( j - i ) ); > } > i--; > - a->a_numvals--; > } > } > > > > Best regards, > Julien > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

(ITS#7893) memberof: bad a_numvals count
by jsoula＠univ-lille2.fr 09 Jul '14

09 Jul '14

Full_Name: julien soula Version: 2.4.39 OS: URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (194.254.117.28) Hi, this is related to ITS#6508. I don't think that the supplied patch is enough to fix the bug. When the last "member" entry is dangling then "( j - i == 1 )" is true and the loop is broken without reducing a_numvals. So I propose this patch : --- servers/slapd/overlays/memberof.c.orig 2014-07-09 21:17:14.487296363 +0200 +++ servers/slapd/overlays/memberof.c 2014-07-09 21:31:44.071565223 +0200 @@ -596,6 +596,7 @@ ber_memfree( a->a_nvals[ i ].bv_val ); BER_BVZERO( &a->a_nvals[ i ] ); } + a->a_numvals--; if ( j - i == 1 ) { break; } @@ -607,7 +608,6 @@ sizeof( struct berval ) * ( j - i ) ); } i--; - a->a_numvals--; } } Best regards, Julien

1 0

Re: (ITS#7351) can not start slapd when enable accesslog relay
by hyc＠symas.com 09 Jul '14

09 Jul '14

bashift(a)gmail.com wrote: > Full_Name: ChenShuJun > Version: 2.4.32 > OS: RHEL5.5(2.6.18-194.el5) > URL: ftp://ftp.openldap.org/incoming/ > Submission from: (NULL) (218.246.127.3) > Due to lack of cooperation from Oracle, back-ndb is no longer supported. Some info here http://www.openldap.org/lists/openldap-technical/201203/msg00216.html http://www.openldap.org/lists/openldap-devel/201307/msg00018.html This ITS will be closed. > > ####################################################################### > # BDB database definitions > ####################################################################### > > database ndb > suffix "dc=bgctv,dc=net,dc=cn" > rootdn "cn=admin,dc=bgctv,dc=net,dc=cn" > rootpw ghwl > directory /usr/local/openldap/var/openldap-data > dbconnect 172.30.5.20:1186 > dbhost 172.30.5.20 > dbname bgctv > dbuser bgctv > dbpass ghwl_h0 > dbconnections 1 > overlay accesslog > logdb "cn=auditlog" > logops writes > logold (objectclass=*) > > #dbconnect 172.30.5.20:1186 > #dbhost 172.30.5.20 > #dbname auditlog > #dbuser bgctv > #dbpass ghwl_h0 > #dbconnections 1 > > access to attrs=userPassword,shadowLastChange > by anonymous auth > by self write > by * none > access to dn.sub="ou=cm,dc=bgctv,dc=net,dc=cn" > by dn="userid=cmadmin,dc=bgctv,dc=net,dc=cn" write > by anonymous auth > by * none > access to * > by dn="userid=cmadmin,dc=bgctv,dc=net,dc=cn" read > by * none > > database bdb > suffix "cn=auditlog" > index reqStart eq > > -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7825) LMDB: misleading error message
by hyc＠symas.com 09 Jul '14

09 Jul '14

h.b.furuseth(a)usit.uio.no wrote: > 5) Add field MDB_dbx.md_dbiseq = DBI usage sequence number, > incremented when dbi_open creates (not reuses) a DBI and in > dbi_close. Copy it to a new malloced array txn->mt_dbiseqs[] > in write txns, verify it in at least commit and drop. > This means close()-open(same DB) will also be caught, which > seems a good thing since with (4) it will work unreliably: > Only if open() reuses the same DBI for the same DB. This is now fixed in mdb.master, using basically this approach. There is a master array of DBI sequence numbers in the environment and another array in each write transaction. It is verified anywhere the dbx->md_name would get used. The sequence number is incremented in dbi_open and when a handle is closed. Note - it is only checked in write txns, and only a few places (drop, commit, cursor_touch, page_search) ever check. There is a new error code MDB_BAD_DBI as well. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7789) Unreliable mdb_env_set_mapsize()
by hyc＠symas.com 09 Jul '14

09 Jul '14

Hallvard Breien Furuseth wrote: > Should have enclosed a test program. Here: > http://folk.uio.no/hbf/testsize.c > Output: > process #0: open env > process #1: open env > process #0: write txn, mapsize 204800 > process #1: write txn, mapsize 40960 > process #2: open env, write txn, mapsize 40960 > process #3: open env, write txn, mapsize 204800 > > > Summarizing a reply which got sent privately instead of to ITS: > > On 2014-01-24 00:07, Howard Chu wrote: >> The doc says the caller of set_mapsize is required to make sure there >> are no active transactions when it is called. As such, X failed this >> requirement, and this sequence of events is explicitly unsupported. > > No, I'm talking about X changing the mapsize when there is no txn, > then afterwards doing a write txn so the mapsize gets written. > >> If Y doesn't start its write txn until after X finishes, then Y will >> see the new size. > > It doesn't, that's the point. Only txn_commit pays attention, > and it checks the size which existed in the wrong metapage. > > Come to think of it, set_mapsize(); env_open(); txn begin/end > should do the same. Y will either way not pay attention to > the new mapsize. > Fixed now in mdb.master -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

Re: (ITS#7713) Segmentation fault if the pagesize of the Operating system is not equal to 4096.
by hyc＠symas.com 09 Jul '14

09 Jul '14

hyc(a)symas.com wrote: > sumanth wrote: >> As per ur suggestions,tried changing >> >> 1. MDB_PAGESIZE to 65536 >> 2. changing pb_lower and pb_upper to uint32_t. >> As of now it is running fine for the existing testcases in openldap-2.4.36. >> However there were problems in 2.4.35. Also saw the comments in >> openldap-devel. So whether this will make to openldap 2.4.36? > > The current plan is to fix this in OpenLDAP 2.5, and we are now halting new > development in 2.4. > The fix is now available in mdb.master; currently you must compile with -DMDB_DEVEL to enable it. Since this is an incompatible DB format change, you should backup any existing DBs before using this option. -- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/

1 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

openldap-bugs