Problems with Devel::DProf

[HuckinFappy]

Good morning your holinesses

I have had a lot of success using Devel::DProf to help me identify the slower parts of my programs, and speed them up.

However, I have several programs that does not work for. These are larger, longer running programs. For example, the one I'm focusing on now generates a 242 MB tmon.out file when run through Devel::DProf

What happens is when I run dprofpp I'll get a message similar to:

Bad profile: - 26f at /tools/xgs/perl/5.8.5/bin/dprofpp line 743, <fh>
+ line 20464.

When I look, line 20464 contains 26941 ^@ characters

I anticipate a variety of responses telling me to break it down to a smaller problem, profile pieces of it, etc. I've considered that and have made some attempts. And perhaps I'll get the profiling information I need via some other route. But I'd like to understand what is happening here, and if there's anything I can do to get Devel::DProf to work when I do experience this.

Thanks,
~Jeff

[shmem]

Namaste, fellow monk.

Just ideas..

That just looks like if /dev/zero were leaking... ;-) is it the only corrupt line in the tmon.out file? Maybe there is a pattern. Flushing buffers to the wrong place... hm.

I'd check whether the tmon.out file is otherwise useable and if so, copied dprofpp to $HOME/trashbin and hacked it not to die and ignore these lines, to get at my profiling data (which I wouldn't trust too much, given the circumstances :-). I'd check also the surroundings of the strange noise lines and looked whether there's something strange happening in the called subs. *shrug*

Maybe it's not even perl-related, but an OS/FS issue. Where is the tmon.out file written to? /tmp ? tmpfs mounted? Which OS do you run and what perl version? It would be handy if you posted that information, so others with more wisdom than I had it at hand whilst thinking, and must not ask.

Do the programs which don't work have anything in common? well, just ideas..

greets,
--shmem

[HuckinFappy]

Thank you for the ideas. And you're correct, I need to provide more information:
• perl 5.8.5
• Devel::DProf::VERSION 20030813.00
• Red Hat machines, with kernels from 2.4.21-2.6.16
• Writing either to local ext3 filesystems or over NFS mounts
I can't seem to make the file usable...removing the bad data leads to other messages about 'garbled data', and further efforts just make things worse.

[convenientstore]

I seem to have some issue also running this.
I do have tmout.out but get below


[root@myserver]# dprofpp
$Monfile is tmon.out
Exporter::Heavy::heavy_export
Garbled profile, unexpected exit time stamp at /usr/bin/dprofpp line 6
+67, <fh> line 195.

line where issue is below
sub exitstamp {
        my($stack, $tstack, $t, $times, $ctimes, $name, $in, $tab, $de
+ep) = @_;
        my( $x, $c, $z );

        $x = pop( @$stack );
        if( ! defined $x ){
                die "Garbled profile, missing an enter time stamp";
        }
        if( $x->[0] ne $name ){
          if ($x->[0] =~ /::AUTOLOAD$/) {
            if ($opt_A) {
              $name = $x->[0];
            }
          } elsif ( $opt_F ) {
            warn( "Garbled profile, faking exit timestamp:\n\t$name =>
+ $x->[0].\n");
            $name = $x->[0];
          } else {
            foreach $z (@stack, $x) {
              print $z->[0],"\n";
            }
            die "Garbled profile, unexpected exit time stamp";
          }

[root@myserver]# uname -a
Linux myserver 2.6.18-1.2798.fc6 #1 SMP Mon Oct 16 14:54:20 EDT 2006 i
+686 i686 i386 GNU/Linux
[root@myserver]# cat /proc/cpuinfo 
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 8
model name      : Pentium III (Coppermine)
stepping        : 3
cpu MHz         : 647.233
cache size      : 256 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 mtrr pge mca cmov
+ pat pse36 mmx fxsr sse up
bogomips        : 12

<UPDATE> for now I just commenented out the DIE part and program ran.. not sure if that's a solution but I get to see some results