On Wed, 26 Mar 2014 08:16:23 +0100 Johan Hake <[email protected]> wrote:
> If none else object, I think we should just push this fix into > instant, as it has been a returning headache for quite a long time. > > Could you make a pull request for your fork? But as os.system does > not work on windows it would be nice to stick with Popen there. Also > see the commented out code in the same module. It seems that os.system works on windows http://docs.python.org/2/library/os.html#os.system Moreover, it is logical choice as it is implemented by std C function system() which probably guarantees OFED-fork safety. Only drawback seems to be the system-dependency of return value. Jan > > Johan > > > On Tue, Mar 25, 2014 at 2:57 PM, Jan Blechta > <[email protected]>wrote: > > > After some digging, it seems that segfaults observed on > > OFED/InfiniBand clusters (see [1], [2], [3]) are caused by > > implementation of > > > > subprocess.Popen > > > > Check your local subprocess.py:_execute_child. Parent does not seem > > to keep hands of memory between fork() and exec() (as required [4], > > [5]), especially as it is fiddling with garbage collector. > > > > I tried switching to os.system (posix implementation in [6]) > > instead of subprocess.Popen and it seems promising. Check out [7] > > or enclosed patch. > > > > Jan > > > > [1] https://answers.launchpad.net/dolfin/+question/219270 > > [2] https://answers.launchpad.net/dolfin/+question/225946 > > [3] http://fenicsproject.org/pipermail/fenics/2013-June/000398.html > > [4] https://www.open-mpi.org/faq/?category=openfabrics#ofa-fork > > [5] > > http://www.openfabrics.org/downloads/OFED/release_notes/OFED_3.12_rc1_release_notes#3.03 > > [6] > > http://svn.python.org/projects/python/trunk/Modules/posixmodule.c > > [7] https://bitbucket.org/blechta/instant/branch/blechta/ofed-fork > > _______________________________________________ fenics mailing list > > [email protected] > > http://fenicsproject.org/mailman/listinfo/fenics > > > > _______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
