On Thu, Aug 31, 2017 at 03:49:44PM +0100, Howard Chu wrote:

Christopher Wood wrote:

...

Good morning, long time no list.

For backstory, I downloaded the 2.4.45 source tarball from https://www.openldap.org/software/download/, compiled, ran "make test" (which was fine as near as I can tell), but doing a slapadd of cn=config caused a segfault.

I used the points on http://www.openldap.org/faq/data/cache/59.html to create the following backtrace:

https://gist.github.com/christopherwood/701fbc70ef352a3bf2d2893a9ae0a65e

And this is the ldif (temporary password is fakepass333):

https://gist.github.com/christopherwood/2cf46f5a3384f9edc89e7eabbefc465e

Does anything jump out at you as obviously wrong or a good place to look? I've chopped the ldif down from the original with confidential data and am still reducing, but it's still segfaulting.

Yes. Look at your LDIF line 62 more carefully. Compare it to line 65 if it's still not obvious to you.

Granted it shouldn't SEGV but the backtrace shows it was shutting itself down already because it (correctly) rejected your input.

Oh ouch, that is obvious now. Nothing like typo'ing in front of the internet first thing in the morning. Thank you, ahem.

(For posterity: The accesslog overlay configuration at line 62 specifies the syncprov overlay in the DN, which won't work. When I specify olcOverlay={1}accesslog in the DN line, it works without a segfault.)

-- -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/