#!/usr/bin/perl -w

## sort-addresses.pl
##  by Mike Melanson (mike at multimedia.cx)
## This Perl script takes an file with a list of addresses and any number of
## function files indicating function boundaries, and sorts the addresses 
## into the function boundaries.
##
## Usage:
##   sort-addresses.pl <address-file> <function-file-1> [function-file-2 ...
##     ... function-file-n]

use strict;

my @original_address_bins;
my @address_bins;
my @sorted_address_bins;
my @addresses;
my @sorted_addresses;
my $i;
my $j;

my $address_filename = shift;
my $function_filename;
my @function_filenames;

while ($function_filename = shift) {
  $function_filenames[++$#function_filenames] = $function_filename;
}

sub sort_bins {

  $b->{'count'} <=> $a->{'count'};

}

if (!defined $address_filename || !defined $function_filenames[0]) {
  print <<EOM;
Usage: sort-addresses.pl <address file> <function-file> [function-file-2 ... function-file-n]
  This script requires an address file followed by at least one function file.
  Any number of additional function files may be specified adterwards.
EOM
  exit;
}

print "loading known function information...\n";

for ($i = 0; $i <= $#function_filenames; $i++) {
  open (FUNCTIONFILE, $function_filenames[$i])
    or die "could not open $function_filenames[$i]\n";

  while (($_ = <FUNCTIONFILE>)) {
    if (/^function: ([0-9A-Fa-f]{8}) -> ([0-9A-Fa-f]{8}): (.+)/) {
# small hole right now: disregard the function whose last address is FFFFFFFF
if ($2 ne "FFFFFFFF") {
      $#original_address_bins++;
      $original_address_bins[$#original_address_bins]->{'name'} = $3;
      $original_address_bins[$#original_address_bins]->{'first_address'} = $1;
      $original_address_bins[$#original_address_bins]->{'last_address'} = $2;
      $original_address_bins[$#original_address_bins]->{'count'} = 0;
}
    }
  }
}

# let the last bin count everything above the last bin
$original_address_bins[$#original_address_bins]->{'last_address'} = "FFFFFFFF";

close (FUNCTIONFILE);

print "loading addresses...\n";

open (ADDRESSFILE, $address_filename)
  or die "could not open $address_filename\n";

while ($_ = <ADDRESSFILE>) {
  if (/START PROFILE/) {
    $_ =~ s/^START PROFILE: (.*)$//;
    print <<EOM;

*************************************************************************
Profile: $1
EOM

    # reset everything and start logging
    undef @addresses;
    undef @sorted_addresses;
    undef @address_bins;
    undef @sorted_address_bins;
    for ($i = 0; $i <= $#original_address_bins; $i++) {
      $address_bins[$i]->{'name'} = 
        $original_address_bins[$i]->{'name'};
      $address_bins[$i]->{'first_address'} = 
        $original_address_bins[$i]->{'first_address'};
      $address_bins[$i]->{'last_address'} = 
        $original_address_bins[$i]->{'last_address'};
      $address_bins[$i]->{'count'} = 0;
    }
    my $count = 0xFFFFFFFF;
#    my $count = 5000;   # good for debugging since it takes a lot less time
    while (($_ = <ADDRESSFILE>) && ($_ !~ /STOP PROFILE/) && ($count-- > 0)) {
      if (/^:/) {
        $_ =~ s/^:(.*)$//;
        $addresses[++$#addresses] = $1;
      }
    }

    print <<EOM;
total addresses executed: $#addresses
*************************************************************************
EOM
    @sorted_addresses = sort @addresses;

    # $i is the current bin number, $j is the current address
    $i = 0;
    $j = 0;
    while (($i <= $#address_bins) && ($j <= $#sorted_addresses)) {
      if ($sorted_addresses[$j] gt $address_bins[$i]->{'last_address'}) {
        $i++;  # next bin; loop again
      } else {
        # add the address to this bin's count and move to the next address
        $address_bins[$i]->{'count'}++;
        $j++;
      }
    }

    @sorted_address_bins = sort sort_bins @address_bins;
    for ($i = 0; $i <= $#sorted_address_bins; $i++) {
      if ($sorted_address_bins[$i]->{'count'} > 0) {
        print "$sorted_address_bins[$i]->{'name'}: $sorted_address_bins[$i]->{'first_address'} -> $sorted_address_bins[$i]->{'last_address'}, count = $sorted_address_bins[$i]->{'count'} (" . $sorted_address_bins[$i]->{'count'} / $#sorted_addresses * 100 . "%)\n";
      }
    }
  }
}

close (ADDRESSFILE);