[Pkg-openldap-devel] Bug#866122: slapd-mtread crash on ppc64{, el} in stretch/sid

[Ryan Tandy]

I spent some more time on this issue and I believe I isolated the 
problem to the following line:

int     r = ((double)nvalues)*rand()/(RAND_MAX + 1.0);

http://sources.debian.net/src/openldap/2.4.44%2Bdfsg-7/tests/progs/slapd-mtread.c/#L682

Specifically, during the multi-threaded tests, casting 'nvalues' from 
int to double occasionally comes up with the wrong result.

e.g.:
(double)19 -> 61.000000

Note that 'nvalues' is a local variable in the same function. There 
should be zero chance of it being shared with another thread or anything 
like that.

The bug reproduces consistently when building with stretch's toolchain 
and running stretch's kernel, but I have not seen it at all with 
jessie's kernel *nor* in a jessie chroot on stretch's kernel. I have 
also reproduced it with clang as well as gcc.