Newsgroups: comp.parallel.pvm
From: "Paul A. Gray" <pgray@mth.msu.edu>
Subject: (no subject)
Organization: Michigan State University Mathematics Department
Date: 7 Jul 1995 11:03:16 GMT
Mime-Version: 1.0
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Message-ID: <3tj49k$1c5m@msunews.cl.msu.edu>

From pgray Wed Jul  5 10:53:22 1995
Newsgroups: comp.parallel.pvm
From: pgray@math.msu.edu (Paul Gray)
Path: pgray
Distribution: 
Followup-To: 
Organization: Michigan State University
Subject: Problem spawning FORTRAN programs in LINUX
Keywords: 

Greetings all.

I would appreciate any advice on the following dilemma...

I am having problems spawning FORTRAN processes on a Linux machine.  For
instance, the example programs master1.f and slave1.f (when compiled via 
"aimk fmaster fslave") make fine.  However, upon running the master program, I am
able to enter the number of processes, nproc, but then I encounter the error: 

[tid000something] pvm_spawn: Bad argument
Segmentation fault

Note that this is _only_ with FORTRAN programs...the "c" versions of master and
slave work fine.

In trying to find the "bad argument" to pvm_spawn, I find that by replacing
"nproc" in master1.f with an integer, say 8,  (i.e., I hardcoded "8" into the
pvmfspawn call in place of "nproc") I just obtain:

Segmentation fault

Usually, when I obtain "segmentation faults", it means that I've overshot the
bounds on an array...so one other bit of information that might be relevant
is that one of the FORTRAN compilers that I use does not recognize integer*2...
Is there some translation to integer*2 when calling the c-function pvm_spawn?

The FORTRAN compilers that I'm using are f77 -- which is actually a perl scripted
front end to f2c -- and g77 (the versions escape me at the moment).

Also, this is only a problem with the Linux arch's...The SUN4, SUN4SOL2, and
SUNMP executions of fpvmspawn that I've tried work fine. 

Certainly there are others out there using FORTRAN pvm programs on a LINUX
architecture...Any suggestions?

ADVthanksANCE

-pgray@mth.msu.edu

From pgray Wed Jul  5 10:55:21 1995
Newsgroups: comp.parallel.pvm
From: pgray@math.msu.edu (Paul Gray)
Path: pgray
Distribution: 
Followup-To: 
Organization: Michigan State University
Subject: From pgray Wed Jul  5 10:53:22 1995
Path: pgray
Distribution: 
Followup-To: 
Organization: Michigan State University
Subject: Problem spawning FORTRAN programs in LINUX
Keywords: 

Greetings all.

I would appreciate any advice on the following dilemma...

I am having problems spawning FORTRAN processes on a Linux machine.  For
instance, the example programs master1.f and slave1.f (when compiled via 
"aimk fmaster fslave") make fine.  However, upon running the master program, I am
able to enter the number of processes, nproc, but then I encounter the error: 

[tid000something] pvm_spawn: Bad argument
Segmentation fault

Note that this is _only_ with FORTRAN programs...the "c" versions of master and
slave work fine.

In trying to find the "bad argument" to pvm_spawn, I find that by replacing
"nproc" in master1.f with an integer, say 8,  (i.e., I hardcoded "8" into the
pvmfspawn call in place of "nproc") I just obtain:

Segmentation fault

Usually, when I obtain "segmentation faults", it means that I've overshot the
bounds on an array...so one other bit of information that might be relevant
is that one of the FORTRAN compilers that I use does not recognize integer*2...
Is there some translation to integer*2 when calling the c-function pvm_spawn?

The FORTRAN compilers that I'm using are f77 -- which is actually a perl scripted
front end to f2c -- and g77 (the versions escape me at the moment).

Also, this is only a problem with the Linux arch's...The SUN4, SUN4SOL2, and
SUNMP executions of fpvmspawn that I've tried work fine. 

Certainly there are others out there using FORTRAN pvm programs on a LINUX
architecture...Any suggestions?

ADVthanksANCE

-pgray@mth.msu.edu


Keywords: 





