https://bugs.openldap.org/show_bug.cgi?id=10029
Issue ID: 10029
Summary: slapd crashes when run with unlimited open files
Product: OpenLDAP
Version: 2.6.4
Hardware: All
OS: Mac OS
Status: UNCONFIRMED
Keywords: needs_review
Severity: normal
Priority: ---
Component: slapd
Assignee: [email protected]
Reporter: [email protected]
Target Milestone: ---
To reproduce:
% ulimit -n
unlimited
% $T/openldap-2.6.4/libexec/slapd -d-1
641ee8bc.32f05820 0x1dc760140 @(#) $OpenLDAP: slapd 2.6.4 (Mar 25 2023
12:25:49) $
openldap
641ee8bc.32f39ff8 0x1dc760140 daemon_init: <null>
641ee8bc.32f40588 0x1dc760140 daemon: SLAP_SOCK_INIT: dtblsize=-1
641ee8bc.32f43080 0x1dc760140 ch_calloc of 1 elems of 18446744073709551615
bytes failed
Assertion failed: (0), function ch_calloc, file ch_malloc.c, line 107.
zsh: abort $T/openldap-2.6.4/libexec/slapd -d-1
This is because `daemon.c` (line 1867) uses the maximum number of open files to
set `dtblsize`, which is subsequently used to size an array:
1867 #ifdef HAVE_SYSCONF
1868 dtblsize = sysconf( _SC_OPEN_MAX );
1869 #elif defined(HAVE_GETDTABLESIZE)
1870 dtblsize = getdtablesize();
1871 #else /* ! HAVE_SYSCONF && ! HAVE_GETDTABLESIZE */
1872 dtblsize = FD_SETSIZE;
1873 #endif /* ! HAVE_SYSCONF && ! HAVE_GETDTABLESIZE */
If the maximum number of FDs is unlimited, then sysconf(_SC_OPEN_MAX) returns
-1, and the program crashes when it tries to malloc that much memory.
I've marked this as OS=macOS because that's what I'm illustrating this on, but
the same thing would happen on any OS where the sysconf call returns a negative
number for the 'unlimited' case.
--
You are receiving this mail because:
You are on the CC list for the issue.
