use strict;
use warnings;

my %msgs;
while (<DATA>) {
    s/^\s+//;
    chomp;
    $msgs{$_}++;
}

# Sort by number of occurrences and only show top 8:
my $i = 0;
for my $m (sort {$msgs{$b} <=> $msgs{$a}} keys %msgs) {
    print "$msgs{$m} $m\n";
    $i++;
    last if $i == 8;
}

__DATA__
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol unlock_
+page
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol generic
+_file_read
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol generic
+_file_write
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol generic
+_file_mmap
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol generic
+_file_sendfile
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: disagrees about versio
+n of symbol zone_table
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol zone_ta
+ble
     Mar  9 08:15:05  gen-vcs11 kernel: kjslahdisagrees about version 
+of symbol unlock_page
Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol unlock_page
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol filemap
+_fdatawrite
     Mar  9 08:15:05  gen-vcs11 kernel: kjslah: Unknown symbol find_or
+_create_page
[download]

See also:

perlintro

perldoc -q sort

You are including the time stamps in the key, so identical events a second apart increment separate counts. or maybe you were showing the overall concept, and leaving the trimming as an exercise for the student?

It looks like gen-vcs11 kernel is a standard component of every line, so I would ignore it. Using split to extract the second and third components, and using those as a key:

my ( $code, $msg ) = ( split, ':', $_)[2,3];
$msgs{$code}{$msg}++;
[download]

It becomes even simplar if you only want to preserve the msg component.

As Occam said: Entia non sunt multiplicanda praeter necessitatem.

or maybe you were showing the overall concept, and leaving the trimming as an exercise for the student?

Yes.

This is what perl is great for.

Probably the most direct approach would be to parse each line with a regex to get the parts you care about (IE not the timestamp). Create a hash where the key is the relevant part of the line and the value is a number that you increment every time you find the same message. Notice you're not keeping the whole file in memory, just one instance of each line and a number.

--Pileofrogs