Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!yale!eagle.wesleyan.edu!awilcox From: awilcox@eagle.wesleyan.edu Newsgroups: comp.lang.perl Subject: Short core dump file produced Message-ID: <1990Dec2.145832.36355@eagle.wesleyan.edu> Date: 2 Dec 90 19:58:31 GMT Organization: Wesleyan University Computing Center Lines: 88 I have a Perl program that I leave running in a window, which wakes up every 30 seconds and reports on how much disk space I have left and who is on the system. Once in a blue moon Perl has a segmentation fault and core dumps. The strange thing about this is that the core file is short (and gdb complains about not being able to read things from the file): $ ls -l core -rw-r--r-- 1 andrew astro 8192 Dec 2 10:22 core I checked, and the core file really does come from Perl: $ strings core ... $Header: perly.c,v 3.0.1.8 90/10/16 10:14:20 lwall Locked $ Patch level: 37 ... I wonder what could cause only part of the core file to be written? I have lots of free disk space (16meg) on the local file system the core file is written out to. This happened twice, and the other core file was only 8K too. This is a Sparcstation 1, and I'm using Perl @ PL37 compiled with gcc version 1.36. Here is the Perl program, "watch": #!/data1/andrew/bin/perl while (1) { system( "clear" ); %users = (); $_ = `users`; for $user (split) { $users{$user} = ''; } open( USERS, "ps -aux|" ); while () { ($user) = split; $users{$user} = '' if $user ne 'USER'; } close( USERS ); for $user (sort(keys(%users))) { print "$user "; } print "\n"; system( "mail -e" ); if (system("mail -e") >> 8) { print "\n"; } else { print "*** Mail ***\n"; } print "\n"; open(F,"df|"); while () { print $_ if m./dev/sd0g.; print $_ if m./data1.; } close(F); system( "pstat -s" ); sleep(30); } I run it from an xterm window: xterm -title Watch -g 65x7+115+0 -fn 6x10 -e watch & I thought maybe when xterm terminated, that would somehow stop the rest of the core file from being written. But I tried this with a Perl program that killed itself, and a normal core file was produced. Any thoughts? Andrew Wilcox Department of Astronomy & Astrophysics, Pennsylvania State University (andrew@astro.psu.edu) Brought to you by Super Global Mega Corp .com