(ITS#7429) MDB crashing with slapo-valsort

[marco.pizzoli@gmail.com]

Full_Name: Marco Pizzoli
Version: 2.4.33
OS: Linux 64bit
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (194.11.209.11)


Hi,
I'm aware I'm not providing a complete bug report, but as I wrote on the mailing
list I'm in a situation where I simply can't do it. Feel free to close if not
enough information will be provided or try to ask me other infos if it could be
worth the doing.

I slapadd-ed a pre-existent OpenLDAP database (2.4.16) on a new test system
(2.4.33).
The new setup uses back-mdb as database. 

I created a second specular system and successfully completed a sync-replication
from the first of the servers. Repeated the procedure (drop/populate the db) 5
or 6 times, without any problem.
I decided to enable the valsort overlay, instanciated on the mdb db.
My configuration was simply:
overlay valsort
valsort-attr member dc=company,dc=XX alpha-ascend

Now I have the slave always crashing at almost 70% of progress.

I produced a "bt full" (which I can't copy/paste here) of the core and I can see
in the "mdb_search <cut> search.c:352" section a lot of rows similar to this
one:
isc{
bv_val = 0x8 <Address 0x8 out of bounds>}, {bv_len = 140172391141264, bv_val =
0x7f7c6e041000 <Address 0x7f7c6e041000 out of bounds>},[cut]

I would be happy to provide other infos but I can't copy paste the entire "bt
full" output. I'm sorry about that.

Thanks
Marco Pizzoli