snappyHexMesh under valgrind reports lots of uninitialised memory

(unfinished) With 2a4ddaa9c6b0190e46493280d1f96218d3f242dd: processor0.log

(unfinished) Under 1712: processor0.log

Same openmpi? I run my 96478baa5d44f5 through with SYSTEMOPENMPI (1.10.6rc1) and valgrind on incompressible/simpleFoam/motorBike started with this:

mpirunDebug -method=5 -local -yes -np 6 snappyHexMesh -decomposeParDict system/decomposeParDict.6 -parallel &

This sets valgrind running with valgrind --leak-check=full --show-reachable=yes - do I need more flags than this?

I've gotten to Morph iteration 6, and still nothing untoward happening.

my mpirun:

ThirdParty-plus.develop/platforms/linux64Gcc/openmpi-2.1.1/bin/mpirun

From the traceback it looks like it has been configured with memory check support:

==10213==    at 0xDE1A891: valgrind_module_isdefined (memchecker_valgrind_module.c:111)
==10213==    by 0xDE1A2CB: opal_memchecker_base_isdefined (memchecker_base_wrappers.c:33)
==10213==    by 0xD371482: memchecker_call (memchecker.h:110)
==10213==    by 0xD3715FC: PMPI_Send (psend.c:49)

That could be the reason.

mentioned in commit 8a74970efb9a923d2e4355fa9bcd6fb9f6a7db73

assuming resolved considering 8a74970e. please reopen the ticket if the issue persists.

closed