Summary: | sys-auth/nss_ldap-265 and dev-libs/openssl-1.0.0a-r1 produce segfaults in libcrypto.so.1.0.0 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Torsten Kurbad <torsten> |
Component: | [OLD] Core system | Assignee: | Gentoo's Team for Core System packages <base-system> |
Status: | RESOLVED DUPLICATE | ||
Severity: | normal | CC: | ldap-bugs, robbat2, ssuominen |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Torsten Kurbad
2010-08-18 12:22:51 UTC
Contents of /etc/ldap.conf: uri ldap://ldapsrv.iwm-kmrc.de base o=IWM,dc=iwm-kmrc,dc=de rootbinddn cn=ldapadmin,o=IWM,dc=iwm-kmrc,dc=de scope one nss_base_passwd ou=People,o=IWM,dc=iwm-kmrc,dc=de?one nss_base_passwd ou=Machines,o=IWM,dc=iwm-kmrc,dc=de?one nss_base_shadow ou=People,o=IWM,dc=iwm-kmrc,dc=de?one nss_base_group ou=Groups,o=IWM,dc=iwm-kmrc,dc=de?one ssl start_tls tls_cacertfile /etc/iwm/ssl/certs/iwm-CA_cert.pem tls_ciphers HIGH:MEDIUM:+SSLv2 Contents of /etc/openldap/slapd.conf (openldap version 2.4.21): pidfile /var/run/openldap/slapd.pid argsfile /var/run/openldap/slapd.args modulepath /usr/lib/openldap/openldap moduleload back_bdb.la back_hdb.la loglevel 0 include /etc/openldap/schema/core.schema include /etc/openldap/schema/cosine.schema include /etc/openldap/schema/inetorgperson.schema include /etc/openldap/schema/nis.schema include /etc/openldap/schema/krb5-kdc.schema include /etc/openldap/schema/samba.schema include /etc/openldap/schema/quota.schema sasl-realm IWM-KMRC.DE sasl-host ldapsrv.iwm-kmrc.de sasl-regexp uid=ldapadmin,cn=iwm-kmrc.de,cn=gssapi,cn=auth cn=ldapadmin,o=IWM,dc=iwm-kmrc,dc=de sasl-regexp uid=.*/admin,cn=iwm-kmrc.de,cn=gssapi,cn=auth cn=ldapadmin,o=IWM,dc=iwm-kmrc,dc=de sasl-regexp uid=(.*),cn=iwm-kmrc.de,cn=gssapi,cn=auth uid=$1,ou=People,o=IWM,dc=iwm-kmrc,dc=de sasl-secprops maxssf=0 TLSCipherSuite HIGH:MEDIUM:+SSLv2 TLSCACertificateFile /etc/iwm/ssl/certs/iwm-CA_cert.pem TLSCertificateFile /etc/iwm/ssl/certs/iwm-ldap_cert.pem TLSCertificateKeyFile /etc/iwm/ssl/private/iwm-ldap_key.pem database hdb suffix "o=IWM,dc=iwm-kmrc,dc=de" directory /var/lib/openldap-data/iwm lastmod on index objectClass,uid,uidNumber,gidNumber,memberUid eq index sambaSID eq index sambaPrimaryGroupSID eq index sambaDomainName eq index cn,mail,surname,givenname eq,subinitial sizelimit unlimited password-hash {SASL} rootdn "cn=ldapadmin,o=IWM,dc=iwm-kmrc,dc=de" rootpw {SSHA}... ... ACL definitions Need a backtrace[1] of the segmentation fault. [1] http://www.gentoo.org/proj/en/qa/backtraces.xml (In reply to comment #2) > Need a backtrace[1] of the segmentation fault. > > [1] http://www.gentoo.org/proj/en/qa/backtraces.xml > After building all required packages with FEATURES, CFLAGS and so on needed for debugging, no further segfaults occured. So could this problem be related to my CFLAGS? (In reply to comment #3) > debugging, no further segfaults occured. So could this problem be related to my > CFLAGS? It's possible, for example -O3 with GCC 4.4.x is known to produce bad code on x86 wrt bug 270120. > It's possible, for example -O3 with GCC 4.4.x is known to produce bad code on > x86 wrt bug 270120. Further investigating this, I found that the problem lies with openssl. Apparently to prevent the segfaults from occuring, it's necessary to compile dev-libs/openssl with -O2, while using GCC 4.4.x. These flags seem work for me now: CFLAGS="-march=pentium4 -O2 -msse2 -mfpmath=sse -mmmx" CXXFLAGS="${CFLAGS}" If I understand bug #270120 correctly, these problems aren't to be expected on amd64 machines, right? Would this one warrant some flag-o-matic for the combination of x86 and GCC 4.4.x to a dev-libs/openssl-1.0.0a-r2? (In reply to comment #5) > > It's possible, for example -O3 with GCC 4.4.x is known to produce bad code on > > x86 wrt bug 270120. > > Further investigating this, I found that the problem lies with openssl. > > Apparently to prevent the segfaults from occuring, it's necessary to compile > dev-libs/openssl with -O2, while using GCC 4.4.x. > These flags seem work for me now: > > CFLAGS="-march=pentium4 -O2 -msse2 -mfpmath=sse -mmmx" CXXFLAGS="${CFLAGS}" > > If I understand bug #270120 correctly, these problems aren't to be expected on > amd64 machines, right? > > Would this one warrant some flag-o-matic for the combination of x86 and GCC > 4.4.x to a dev-libs/openssl-1.0.0a-r2? > Try combination of "-O3 -fno-tree-vectorize", if that works, this is a duplicate of bug 270120 for sure And nope, I don't think any flag-o-matic hackery is required, -O3 is expected to break... > Try combination of "-O3 -fno-tree-vectorize", if that works, this is a > duplicate of bug 270120 for sure It is, definitely. I'll go rebuild my world with -fno-tree-vectorize... ;-) Thanks, Torsten *** This bug has been marked as a duplicate of bug 270120 *** |