On Mon, 11 Oct 2004 08:43:44 -0700, "David A. Case" <case.scripps.edu> wrote:
> (I wrote)
>
>> FC1 compiles fine with Intel's v8 compiler, but running tests freezes
>> the machine:
>>
>> cd tip5p; ./Run.tip5p
>> forrtl: info: Fortran error message number is 174.
>> forrtl: warning: Could not open message catalog: ifcore_msg.cat.
>> forrtl: info: Check environment variable NLSPATH and protection of /usr/lib/ifcore_msg.cat.
>
> What happens when you follow the suggestions above?
The warning means "I can't find the error message file so here is the
number of the error message instead". Of course it would be nice to
know what it means. Googling for "intel fortran 174" found the following:
<URL:
http://softwareforums.intel.com/ids/board/message?board.id=11&message.id=1345&jump=true>
It means "forrtl: severe (174): SIGSEGV, segmentation fault occurred".
Setting NLSPATH to the lib directory of the Intel compiler installation
did not help, but making a symlink from the file to /usr/lib did, so I
got this:
cd tip5p; ./Run.tip5p
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
sander 4013956E Unknown Unknown Unknown
sander 4013839C Unknown Unknown Unknown
sander 0820846C Unknown Unknown Unknown
sander 082084CF Unknown Unknown Unknown
sander 081952D9 Unknown Unknown Unknown
sander 00000000 Unknown Unknown Unknown
(It froze, so I pressed ^C here)
forrtl: error (69): process interrupted (SIGINT)
(Still, nothing happened. I went to another window and killed it)
forrtl: error (78): process killed (SIGTERM)
(Still, nothing happened. I killed it again with -HUP)
Hangup
make: *** [test.sander.no_lmod] Interrupt
("make test" ended there)
> Sander operates completely in the user space, and no such program
> should be able to freeze the machine, even if it has a bug.
Sorry, I was not explaining myself clearly enough. The machine itself
was fine and sander could be killed from another login session. Sander
wasn't doing anything and wasn't taking up any CPU time while it was
frozen. It just wasn't getting anywhere, and I can't complete the
tests. Obviously, I can't make available a binary that segfaults and
freezes. We do also have the Absoft compiler, so if the Intel 8.0
compiler is a dead end, I could try that instead. We do have the
earlier Intel compiler but no RH7.x to try it on anymore.
>> On Solaris 8, I get the following. (Compiler = Sun Workshop 6 update 2)
>
> Look at the definition of _Xconst in /usr/lib/X11/Xfuncproto.h.
Such a file does not exist. I assume you mean /usr/include/X11/.
#ifndef _Xconst
#if defined(__STDC__) || defined(__cplusplus) || defined(c_plusplus) || (FUNCPROTO&4)
#define _Xconst const
#else
#define _Xconst
#endif
#endif /* _Xconst */
> _Xconst should get expanded to "const", but it sounds like it is not.
Let's look at the compilation command line again:
cc -c -I/usr/openwin/include -DSYSV -o Atoms.o Atoms.c
Xfuncproto.h is in /usr/openwin/include/X11, and Atoms.h includes
<X11/Xfuncproto.h>, so it should be found correctly.
> You may need to add -D__STDC__ to your $(CC) variable in config.h.
> (I am just guessing here, based on files in SunOS 5.8, which works fine.)
SunOS 5.8 is Solaris 8 (those damn marketoons!).
Looking at the definition in Xfuncproto.h, adding __STDC__ definitely
seems like the key.
> What is the output of "uname -a"? From "cc -V"?
SunOS mymachine 5.8 Generic_108528-29 sun4u sparc SUNW,Sun-Fire-V240
cc: Sun WorkShop 6 update 2 C 5.3 2001/05/15
On Mon, 11 Oct 2004 10:17:50 -0700, "Ross Walker" <ross.rosswalker.co.uk> wrote:
> You can set an alias to alias make to gmake. This should work even if there
> is a 'make' command in your path. E.g.
>
> alias make=gmake
Yep.
In the meantime, I fixed the ar problem by adding spaces to the command
lines in the Makefiles. Why leave room for such a problem in the first
place?
--
Atro Tossavainen (Mr.) / The Institute of Biotechnology at
Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
+358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
< URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue Oct 12 2004 - 08:53:00 PDT
