[mvapich-discuss] viacheck.c error

[Karl W. Schulz]

Hello,

For one more data point, we went to 2.6.12.6 to try and avoid the fork()
problem based on the release notes referenced below, but I can confirm
that we still cannot call system within non-threaded MPI applications.
If anyone has had success with fork() and 2.6.12 I'd be interested to
hear about it.

Thanks,

Karl



> -----Original Message-----
> From: mvapich-discuss-bounces at cse.ohio-state.edu
[mailto:mvapich-discuss-bounces at cse.ohio-state.edu]
> On Behalf Of Christian Guggenberger
> Sent: Friday, April 20, 2007 6:58 AM
> To: Sayantan Sur
> Cc: Acero Fernandez Alicia; mvapich-discuss at cse.ohio-state.edu
> Subject: Re: [mvapich-discuss] viacheck.c error
> 
> On Thu, Apr 19, 2007 at 02:58:00PM -0400, Sayantan Sur wrote:
> > Hello Alicia,
> >
> > Thanks for reporting the error. As you said, the error appears after
> > calling "system" from within an MPI program. The "system" call
involves
> > an internal call to fork(). As of OFED-1.1, fork() is *not*
supported by
> > the underlying InfiniBand libraries. I believe OFED-1.2 will support
fork().
> >
> FYI, from the OFED-1.1 release notes:
> 
> "2. Fork support from kernel 2.6.12 and above is available provided
> that applications do not use threads. The fork() is supported as long
> as parent process does not run before child exits or calls exec().
> The former can be achieved by calling wait(childpid) the later
> can be achieved by application specific means.  Posix system() call
> is supported."
> 
> cheers.
>  - Christian