Playing with the DB provided in ITS#10041 on master, interrupting the ldapsearch sometimes leads to a slapd crash. It's not 100% repeatable and the debugger shows dynlist_search2resp touching memory freed by dynlist_search_cleanup already, which doesn't make sense. Might be something else is happening at the same time.
likely fixed by bug#10135, marking as duplicate unless it can be reproduced again *** This issue has been marked as a duplicate of issue 10135 ***
Created attachment 1012 [details] slapd stats logs + backtrace
I can still replicate this on current master with the set up in ITS#10041, just run slapd and run this ldapsearch twice, Ctrl+C it before it finishes: ldapsearch -x -H 'ldap://localhost:9011/' -D 'cn=admin,dc=my-organization,dc=com' -w pw -b 'ou=groups,dc=my-organization,dc=com' '(&(cn=*group*)(cn=*1)(member=uid=user1,ou=people,dc=my-organization,dc=com))' cn stats logs + backtrace attached just now
https://git.openldap.org/openldap/openldap/-/merge_requests/689
head: • ce524021 by Howard Chu at 2024-02-22T19:54:30+00:00 ITS#10044 dynlist: check for abandon in search2resp RE26: • 97474d96 by Howard Chu at 2024-03-26T16:42:34+00:00 ITS#10044 dynlist: check for abandon in search2resp RE25: • abd8706e by Howard Chu at 2024-03-26T16:43:13+00:00 ITS#10044 dynlist: check for abandon in search2resp