misc/tools/progs-analyzer.pl

   1 use strict;
   2 use warnings;
   3 use Digest::SHA;
   4
   5 sub id()
   6 {
   7         return sub { $_[0]; };
   8 }
   9
  10 sub signed($)
  11 {
  12         my ($bits) = @_;
  13         return sub { $_[0] >= (2**($bits-1)) ? $_[0]-(2**$bits) : $_[0]; };
  14 }
  15
  16 use constant OPCODE_E => [qw[
  17         DONE
  18         MUL_F MUL_V MUL_FV MUL_VF
  19         DIV_F
  20         ADD_F ADD_V
  21         SUB_F SUB_V
  22         EQ_F EQ_V EQ_S EQ_E EQ_FNC
  23         NE_F NE_V NE_S NE_E NE_FNC
  24         LE GE LT GT
  25         LOAD_F LOAD_V LOAD_S LOAD_ENT LOAD_FLD LOAD_FNC
  26         ADDRESS
  27         STORE_F STORE_V STORE_S STORE_ENT STORE_FLD STORE_FNC
  28         STOREP_F STOREP_V STOREP_S STOREP_ENT STOREP_FLD STOREP_FNC
  29         RETURN
  30         NOT_F NOT_V NOT_S NOT_ENT NOT_FNC
  31         IF IFNOT
  32         CALL0 CALL1 CALL2 CALL3 CALL4 CALL5 CALL6 CALL7 CALL8
  33         STATE
  34         GOTO
  35         AND OR
  36         BITAND BITOR
  37 ]];
  38 use constant ETYPE_E => [qw[
  39         void
  40         string
  41         float
  42         vector
  43         entity
  44         field
  45         function
  46         pointer
  47 ]];
  48 use constant DEF_SAVEGLOBAL => 32768;
  49 sub typesize($)
  50 {
  51         my ($type) = @_;
  52         return 3 if $type eq 'vector';
  53         return 1;
  54 }
  55
  56 sub checkop($)
  57 {
  58         my ($op) = @_;
  59         if($op =~ /^IF.*_V$/)
  60         {
  61                 return { a => 'inglobalvec', b => 'ipoffset', isjump => 'b', isconditional => 1 };
  62         }
  63         if($op =~ /^IF/)
  64         {
  65                 return { a => 'inglobal', b => 'ipoffset', isjump => 'b', isconditional => 1 };
  66         }
  67         if($op eq 'GOTO')
  68         {
  69                 return { a => 'ipoffset', isjump => 'a', isconditional => 0 };
  70         }
  71         if($op =~ /^ADD_V$|^SUB_V$/)
  72         {
  73                 return { a => 'inglobalvec', b => 'inglobalvec', c => 'outglobalvec' };
  74         }
  75         if($op =~ /^MUL_V$|^EQ_V$|^NE_V$/)
  76         {
  77                 return { a => 'inglobalvec', b => 'inglobalvec', c => 'outglobal' };
  78         }
  79         if($op eq 'MUL_FV')
  80         {
  81                 return { a => 'inglobal', b => 'inglobalvec', c => 'outglobalvec' };
  82         }
  83         if($op eq 'MUL_VF')
  84         {
  85                 return { a => 'inglobalvec', b => 'inglobal', c => 'outglobalvec' };
  86         }
  87         if($op eq 'LOAD_V')
  88         {
  89                 return { a => 'inglobal', b => 'inglobal', c => 'outglobalvec' };
  90         }
  91         if($op =~ /^NOT_V/)
  92         {
  93                 return { a => 'inglobalvec', c => 'outglobal' };
  94         }
  95         if($op =~ /^NOT_/)
  96         {
  97                 return { a => 'inglobal', c => 'outglobal' };
  98         }
  99         if($op eq 'STOREP_V')
 100         {
 101                 return { a => 'inglobalvec', b => 'inglobal' };
 102         }
 103         if($op eq 'STORE_V')
 104         {
 105                 return { a => 'inglobalvec', b => 'outglobalvec' };
 106         }
 107         if($op =~ /^STOREP_/)
 108         {
 109                 return { a => 'inglobal', b => 'inglobal' };
 110         }
 111         if($op =~ /^STORE_/)
 112         {
 113                 return { a => 'inglobal', b => 'outglobal' };
 114         }
 115         if($op =~ /^CALL/)
 116         {
 117                 return { a => 'inglobalfunc', iscall => 1 };
 118         }
 119         if($op =~ /^DONE$|^RETURN$/)
 120         {
 121                 return { a => 'inglobalvec', isreturn => 1 };
 122         }
 123         return { a => 'inglobal', b => 'inglobal', c => 'outglobal' };
 124 }
 125
 126 use constant TYPES => {
 127         int => ['V', 4, signed 32],
 128         ushort => ['v', 2, id],
 129         short => ['v', 2, signed 16],
 130         opcode => ['v', 2, sub { OPCODE_E->[$_[0]] or do { warn "Invalid opcode: $_[0]"; "INVALID#$_[0]"; }; }],
 131         float => ['f', 4, id],
 132         uchar8 => ['a8', 8, sub { [unpack 'C8', $_[0]] }],
 133         global => ['i', 4, sub { { int => $_[0], float => unpack "f", pack "L", $_[0] }; }],
 134         deftype => ['v', 2, sub { { type => ETYPE_E->[$_[0] & ~DEF_SAVEGLOBAL], save => !!($_[0] & DEF_SAVEGLOBAL) }; }],
 135 };
 136
 137 use constant DPROGRAMS_T => [
 138         [int => 'version'],
 139         [int => 'crc'],
 140         [int => 'ofs_statements'],
 141         [int => 'numstatements'],
 142         [int => 'ofs_globaldefs'],
 143         [int => 'numglobaldefs'],
 144         [int => 'ofs_fielddefs'],
 145         [int => 'numfielddefs'],
 146         [int => 'ofs_functions'],
 147         [int => 'numfunctions'],
 148         [int => 'ofs_strings'],
 149         [int => 'numstrings'],
 150         [int => 'ofs_globals'],
 151         [int => 'numglobals'],
 152         [int => 'entityfields']
 153 ];
 154
 155 use constant DSTATEMENT_T => [
 156         [opcode => 'op'],
 157         [short => 'a'],
 158         [short => 'b'],
 159         [short => 'c']
 160 ];
 161
 162 use constant DDEF_T => [
 163         [deftype => 'type'],
 164         [ushort => 'ofs'],
 165         [int => 's_name']
 166 ];
 167
 168 use constant DGLOBAL_T => [
 169         [global => 'v'],
 170 ];
 171
 172 use constant DFUNCTION_T => [
 173         [int => 'first_statement'],
 174         [int => 'parm_start'],
 175         [int => 'locals'],
 176         [int => 'profile'],
 177         [int => 's_name'],
 178         [int => 's_file'],
 179         [int => 'numparms'],
 180         [uchar8 => 'parm_size'],
 181 ];
 182
 183 sub get_section($$$)
 184 {
 185         my ($fh, $start, $len) = @_;
 186         seek $fh, $start, 0
 187                 or die "seek: $!";
 188         $len == read $fh, my $buf, $len
 189                 or die "short read";
 190         return $buf;
 191 }
 192
 193 sub parse_section($$$$$)
 194 {
 195         my ($fh, $struct, $start, $len, $cnt) = @_;
 196
 197         my $itemlen = 0;
 198         $itemlen += TYPES->{$_->[0]}->[1]
 199                 for @$struct;
 200         my $packspec = join '', map { TYPES->{$_->[0]}->[0]; } @$struct;
 201         my @packnames = map { $_->[1]; } @$struct;
 202
 203         $len = $cnt * $itemlen
 204                 if not defined $len and defined $cnt;
 205         $cnt = int($len / $itemlen)
 206                 if not defined $cnt and defined $len;
 207         die "Invalid length specification"
 208                 unless defined $len and defined $cnt and $len == $cnt * $itemlen;
 209         die "Invalid length specification in scalar context"
 210                 unless wantarray or $cnt == 1;
 211
 212         seek $fh, $start, 0
 213                 or die "seek: $!";
 214         my @out = map
 215         {
 216                 $itemlen == read $fh, my $buf, $itemlen
 217                         or die "short read";
 218                 my %h = ();
 219                 @h{@packnames} = unpack $packspec, $buf;
 220                 $h{$_->[1]} = TYPES->{$_->[0]}->[2]->($h{$_->[1]})
 221                         for @$struct;
 222                 \%h;
 223         }
 224         0..($cnt-1);
 225         return @out
 226                 if wantarray;
 227         return $out[0];
 228 }
 229
 230 sub nfa_default_state_checker()
 231 {
 232         my %seen;
 233         return sub
 234         {
 235                 my ($ip, $state) = @_;
 236                 return $seen{"$ip $state"}++;
 237         };
 238 }
 239
 240 sub run_nfa($$$$$$)
 241 {
 242         my ($progs, $ip, $state, $copy_handler, $state_checker, $instruction_handler) = @_;
 243
 244         my $statements = $progs->{statements};
 245
 246         my $nfa;
 247         $nfa = sub
 248         {
 249                 no warnings 'recursion';
 250
 251                 my ($ip, $state) = @_;
 252                 my $ret = 0;
 253
 254                 for(;;)
 255                 {
 256                         return $ret
 257                                 if $state_checker->($ip, $state);
 258
 259                         my $s = $statements->[$ip];
 260                         my $c = checkop $s->{op};
 261
 262                         if(($ret = $instruction_handler->($ip, $state, $s, $c)))
 263                         {
 264                                 # abort execution
 265                                 last;
 266                         }
 267
 268                         if($c->{isreturn})
 269                         {
 270                                 last;
 271                         }
 272                         elsif($c->{iscall})
 273                         {
 274                                 my $func = $s->{a};
 275                                 my $funcid = $progs->{globals}[$func]{v}{int};
 276                                 last
 277                                         if $progs->{error_func}{$funcid};
 278                                 $ip += 1;
 279                         }
 280                         elsif($c->{isjump})
 281                         {
 282                                 if($c->{isconditional})
 283                                 {
 284                                         if(rand 2)
 285                                         {
 286                                                 if(($ret = $nfa->($ip+$s->{$c->{isjump}}, $copy_handler->($state))) < 0)
 287                                                 {
 288                                                         last;
 289                                                 }
 290                                                 $ip += 1;
 291                                         }
 292                                         else
 293                                         {
 294                                                 $nfa->($ip+1, $copy_handler->($state));
 295                                                 $ip += $s->{$c->{isjump}};
 296                                         }
 297                                 }
 298                                 else
 299                                 {
 300                                         $ip += $s->{$c->{isjump}};
 301                                 }
 302                         }
 303                         else
 304                         {
 305                                 $ip += 1;
 306                         }
 307                 }
 308
 309                 return $ret;
 310         };
 311
 312         $nfa->($ip, $copy_handler->($state));
 313 }
 314
 315 sub get_constant($$)
 316 {
 317         my ($progs, $g) = @_;
 318         if($g->{int} == 0)
 319         {
 320                 return 0;
 321         }
 322         elsif($g->{int} > 0 && $g->{int} < 8388608)
 323         {
 324                 if($g->{int} < length $progs->{strings} && $g->{int} > 0)
 325                 {
 326                         return str($progs->{getstring}->($g->{int}));
 327                 }
 328                 else
 329                 {
 330                         return $g->{int} . "i";
 331                 }
 332         }
 333         else
 334         {
 335                 return $g->{float};
 336         }
 337 }
 338
 339 use constant PRE_MARK_STATEMENT => "";
 340 use constant POST_MARK_STATEMENT => "";
 341 use constant PRE_MARK_OPERAND => "*** ";
 342 use constant POST_MARK_OPERAND => " ***";
 343
 344 use constant INSTRUCTION_FORMAT => "%8s %3s | %-12s ";
 345 use constant OPERAND_FORMAT => "%s";
 346 use constant OPERAND_SEPARATOR => ", ";
 347 use constant INSTRUCTION_SEPARATOR => "\n";
 348
 349 sub str($)
 350 {
 351         my ($str) = @_;
 352         $str =~ s/[\000-\037\\\"\177-\377]/sprintf "\\%03o", ord $&/ge;
 353         return "\"$str\"";
 354 }
 355
 356 sub disassemble_function($$;$)
 357 {
 358         my ($progs, $func, $highlight) = @_;
 359
 360         print "$func->{debugname}:\n";
 361
 362         my $initializer = sub
 363         {
 364                 my ($ofs) = @_;
 365                 my $g = get_constant($progs, $progs->{globals}[$ofs]{v});
 366                 print " = $g"
 367                         if defined $g;
 368         };
 369
 370         printf INSTRUCTION_FORMAT, '', '', '.PARM_START';
 371         printf OPERAND_FORMAT, "$func->{parm_start}";
 372         print INSTRUCTION_SEPARATOR;
 373
 374         printf INSTRUCTION_FORMAT, '', '', '.LOCALS';
 375         printf OPERAND_FORMAT, "$func->{locals}";
 376         print INSTRUCTION_SEPARATOR;
 377
 378         my %override_locals = ();
 379         my $p = $func->{parm_start};
 380         for(0..($func->{numparms}-1))
 381         {
 382                 $override_locals{$p} //= "argv_$_";
 383                 for my $comp(0..($func->{parm_size}[$_]-1))
 384                 {
 385                         $override_locals{$p} //= "argv_$_\[$comp]";
 386                         ++$p;
 387                 }
 388                 printf INSTRUCTION_FORMAT, '', '', '.ARG';
 389                 printf OPERAND_FORMAT, "argv_$_";
 390                 print OPERAND_SEPARATOR;
 391                 printf OPERAND_FORMAT, $func->{parm_size}[$_];
 392                 print INSTRUCTION_SEPARATOR;
 393         }
 394         for($func->{parm_start}..($func->{parm_start} + $func->{locals} - 1))
 395         {
 396                 next
 397                         if exists $override_locals{$_};
 398                 $override_locals{$_} = "local_$_";
 399
 400                 printf INSTRUCTION_FORMAT, '', '', '.LOCAL';
 401                 printf OPERAND_FORMAT, "local_$_";
 402                 $initializer->($_);
 403                 print INSTRUCTION_SEPARATOR;
 404         }
 405
 406         my $getname = sub
 407         {
 408                 my ($ofs) = @_;
 409                 return $override_locals{$ofs}
 410                         if exists $override_locals{$ofs};
 411                 my $def = $progs->{globaldef_byoffset}->($ofs);
 412                 return $def->{debugname};
 413         };
 414
 415         my $operand = sub
 416         {
 417                 my ($ip, $type, $operand) = @_;
 418                 if($type eq 'inglobal')
 419                 {
 420                         my $name = $getname->($operand);
 421                         printf OPERAND_FORMAT, "$name";
 422                 }
 423                 elsif($type eq 'outglobal')
 424                 {
 425                         my $name = $getname->($operand);
 426                         printf OPERAND_FORMAT, "&$name";
 427                 }
 428                 elsif($type eq 'inglobalvec')
 429                 {
 430                         my $name = $getname->($operand);
 431                         printf OPERAND_FORMAT, "$name\[\]";
 432                 }
 433                 elsif($type eq 'outglobalvec')
 434                 {
 435                         my $name = $getname->($operand);
 436                         printf OPERAND_FORMAT, "&$name\[\]";
 437                 }
 438                 elsif($type eq 'inglobalfunc')
 439                 {
 440                         my $name = $getname->($operand);
 441                         printf OPERAND_FORMAT, "$name()";
 442                 }
 443                 elsif($type eq 'ipoffset')
 444                 {
 445                         printf OPERAND_FORMAT, "@{[$ip + $operand]}" . sprintf ' ($%+d)', $operand;
 446                 }
 447                 else
 448                 {
 449                         die "unknown type: $type";
 450                 }
 451         };
 452
 453         my $statements = $func->{statements};
 454         my $come_from = $func->{come_from};
 455
 456         my $ipprev = undef;
 457         for my $ip(sort { $a <=> $b } keys %$statements)
 458         {
 459                 if($ip == $func->{first_statement})
 460                 {
 461                         printf INSTRUCTION_FORMAT, $ip, '', '.ENTRY';
 462                         print INSTRUCTION_SEPARATOR;
 463                 }
 464                 if(defined $ipprev && $ip != $ipprev + 1)
 465                 {
 466                         printf INSTRUCTION_FORMAT, $ip, '', '.SKIP';
 467                         printf OPERAND_FORMAT, $ip - $ipprev - 1;
 468                         print INSTRUCTION_SEPARATOR;
 469                 }
 470                 if(my $cf = $come_from->{$ip})
 471                 {
 472                         printf INSTRUCTION_FORMAT, $ip, '', '.XREF';
 473                         my $cnt = 0;
 474                         for(sort { $a <=> $b } keys %$cf)
 475                         {
 476                                 print OPERAND_SEPARATOR
 477                                         if $cnt++;
 478                                 printf OPERAND_FORMAT, ($cf->{$_} ? 'c' : 'j') . $_ . sprintf ' ($%+d)', $_ - $ip;
 479                         }
 480                         print INSTRUCTION_SEPARATOR;
 481                 }
 482
 483                 my $op = $progs->{statements}[$ip]{op};
 484                 my $ipt = $progs->{statements}[$ip];
 485                 my $opprop = checkop $op;
 486
 487                 print PRE_MARK_STATEMENT
 488                         if $highlight and $highlight->{$ip};
 489
 490                 my $showip = $opprop->{isjump};
 491                 printf INSTRUCTION_FORMAT, $showip ? $ip : '', $highlight->{$ip} ? "<!>" : "", $op;
 492
 493                 my $cnt = 0;
 494                 for my $o(qw(a b c))
 495                 {
 496                         next
 497                                 if not defined $opprop->{$o};
 498                         print OPERAND_SEPARATOR
 499                                 if $cnt++;
 500                         print PRE_MARK_OPERAND
 501                                 if $highlight and $highlight->{$ip} and $highlight->{$ip}{$o};
 502                         $operand->($ip, $opprop->{$o}, $ipt->{$o});
 503                         print POST_MARK_OPERAND
 504                                 if $highlight and $highlight->{$ip} and $highlight->{$ip}{$o};
 505                 }
 506
 507                 print POST_MARK_STATEMENT
 508                         if $highlight and $highlight->{$ip};
 509
 510                 print INSTRUCTION_SEPARATOR;
 511         }
 512 }
 513
 514 sub find_uninitialized_locals($$)
 515 {
 516         my ($progs, $func) = @_;
 517
 518
 519         return
 520                 if $func->{first_statement} < 0; # builtin
 521
 522         print STDERR "Checking $func->{debugname}...\n";
 523
 524         my $p = $func->{parm_start};
 525         for(0..($func->{numparms}-1))
 526         {
 527                 $p += $func->{parm_size}[$_];
 528         }
 529
 530         use constant WATCHME_R => 1;
 531         use constant WATCHME_W => 2;
 532         use constant WATCHME_X => 4;
 533         use constant WATCHME_T => 8;
 534         my %watchme = map { $_ => WATCHME_X } ($func->{parm_start} .. ($func->{parm_start} + $func->{locals} - 1));
 535
 536         for(keys %{$progs->{temps}})
 537         {
 538                 $watchme{$_} = WATCHME_T | WATCHME_X
 539                         if not exists $watchme{$_};
 540         }
 541
 542         $watchme{$_} |= WATCHME_R
 543                 for keys %{$func->{globals_read}};
 544         $watchme{$_} |= WATCHME_W
 545                 for keys %{$func->{globals_written}};
 546
 547         my %write_places = ();
 548         for my $ofs(keys %{$func->{globals_written}})
 549         {
 550                 next
 551                         unless exists $watchme{$ofs} and $watchme{$ofs} & WATCHME_X;
 552                 for my $ip(keys %{$func->{globals_written}{$ofs}})
 553                 {
 554                         for my $op(keys %{$func->{globals_written}{$ofs}{$ip}})
 555                         {
 556                                 push @{$write_places{$ip}{$op}}, $ofs;
 557                         }
 558                 }
 559         }
 560
 561         for(keys %watchme)
 562         {
 563                 delete $watchme{$_}
 564                         if ($watchme{$_} & (WATCHME_R | WATCHME_W | WATCHME_X)) != (WATCHME_R | WATCHME_W | WATCHME_X);
 565         }
 566
 567         return
 568                 if not keys %watchme;
 569
 570         for(keys %watchme)
 571         {
 572                 $watchme{$_} = {
 573                         flags => $watchme{$_},
 574                         valid => [0, undef, undef]
 575                 };
 576         }
 577
 578         # mark parameters as initialized
 579         for($func->{parm_start} .. ($p-1))
 580         {
 581                 $watchme{$_}{valid} = [1, undef, undef]
 582                         if defined $watchme{$_};
 583         }
 584         # an initial run of STORE instruction is for receiving extra parameters
 585         # (beyond 8). Only possible if the function is declared as having 8 params.
 586         # Extra parameters behave otherwise like temps, but are initialized at
 587         # startup.
 588         for($func->{first_statement} .. (@{$progs->{statements}}-1))
 589         {
 590                 my $s = $progs->{statements}[$_];
 591                 if($s->{op} eq 'STORE_V')
 592                 {
 593                         $watchme{$s->{a}}{valid} = [1, undef, undef]
 594                                 if defined $watchme{$s->{a}};
 595                         $watchme{$s->{a}+1}{valid} = [1, undef, undef]
 596                                 if defined $watchme{$s->{a}+1};
 597                         $watchme{$s->{a}+2}{valid} = [1, undef, undef]
 598                                 if defined $watchme{$s->{a}+2};
 599                 }
 600                 elsif($s->{op} =~ /^STORE_/)
 601                 {
 602                         $watchme{$s->{a}}{valid} = [1, undef, undef]
 603                                 if defined $watchme{$s->{a}};
 604                 }
 605                 else
 606                 {
 607                         last;
 608                 }
 609         }
 610
 611         my %warned = ();
 612         my %ip_seen = ();
 613         run_nfa $progs, $func->{first_statement}, \%watchme,
 614                 sub {
 615                         my ($h) = @_;
 616                         return { map { $_ => { %{$h->{$_}} } } keys %$h };
 617                 },
 618                 sub {
 619                         my ($ip, $state) = @_;
 620
 621                         my $s = $ip_seen{$ip};
 622                         if($s)
 623                         {
 624                                 # if $state is stronger or equal to $s, return 1
 625
 626                                 for(keys %$state)
 627                                 {
 628                                         if($state->{$_}{valid}[0] < $s->{$_})
 629                                         {
 630                                                 # The current state is LESS valid than the previously run one. We NEED to run this.
 631                                                 # The saved state can safely become the intersection [citation needed].
 632                                                 for(keys %$state)
 633                                                 {
 634                                                         $s->{$_} = $state->{$_}{valid}[0]
 635                                                                 if $state->{$_}{valid}[0] < $s->{$_};
 636                                                 }
 637                                                 return 0;
 638                                         }
 639                                 }
 640                                 # if we get here, $state is stronger or equal. No need to try it.
 641                                 return 1;
 642                         }
 643                         else
 644                         {
 645                                 # Never seen this IP yet.
 646                                 $ip_seen{$ip} = { map { ($_ => $state->{$_}{valid}[0]); } keys %$state };
 647                                 return 0;
 648                         }
 649                 },
 650                 sub {
 651                         my ($ip, $state, $s, $c) = @_;
 652                         my $op = $s->{op};
 653
 654                         my $return_hack = $c->{isreturn} // 0;
 655
 656                         for(qw(a b c))
 657                         {
 658                                 my $type = $c->{$_};
 659                                 next
 660                                         unless defined $type;
 661
 662                                 my $ofs = $s->{$_};
 663
 664                                 my $read = sub
 665                                 {
 666                                         my ($ofs) = @_;
 667                                         ++$return_hack
 668                                                 if $return_hack;
 669                                         return
 670                                                 if not exists $state->{$ofs};
 671                                         my $valid = $state->{$ofs}{valid};
 672                                         if($valid->[0] == 0)
 673                                         {
 674                                                 if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
 675                                                 {
 676                                                         print "; Use of uninitialized value $ofs in $func->{debugname} at $ip.$_\n";
 677                                                         ++$warned{$ip}{$_};
 678                                                 }
 679                                         }
 680                                         elsif($valid->[0] < 0)
 681                                         {
 682                                                 if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
 683                                                 {
 684                                                         print "; Use of temporary $ofs across CALL in $func->{debugname} at $ip.$_\n";
 685                                                         ++$warned{$ip}{$_};
 686                                                 }
 687                                         }
 688                                         else
 689                                         {
 690                                                 # it's VALID
 691                                                 if(defined $valid->[1])
 692                                                 {
 693                                                         delete $write_places{$valid->[1]}{$valid->[2]};
 694                                                 }
 695                                         }
 696                                 };
 697                                 my $write = sub
 698                                 {
 699                                         my ($ofs) = @_;
 700                                         $state->{$ofs}{valid} = [1, $ip, $_]
 701                                                 if exists $state->{$ofs};
 702                                 };
 703
 704                                 if($type eq 'inglobal' || $type eq 'inglobalfunc')
 705                                 {
 706                                         $read->($ofs);
 707                                 }
 708                                 elsif($type eq 'inglobalvec')
 709                                 {
 710                                         $read->($ofs);
 711                                         $read->($ofs+1);
 712                                         $read->($ofs+2);
 713                                 }
 714                                 elsif($type eq 'outglobal')
 715                                 {
 716                                         $write->($ofs);
 717                                 }
 718                                 elsif($type eq 'outglobalvec')
 719                                 {
 720                                         $write->($ofs);
 721                                         $write->($ofs+1);
 722                                         $write->($ofs+2);
 723                                 }
 724                         }
 725                         if($c->{iscall})
 726                         {
 727                                 # builtin calls may clobber stuff
 728                                 my $func = $s->{a};
 729                                 my $funcid = $progs->{globals}[$func]{v}{int};
 730                                 my $funcobj = $progs->{functions}[$funcid];
 731                                 if(!$funcobj || $funcobj->{first_statement} >= 0)
 732                                 {
 733                                         # invalidate temps
 734                                         for(values %$state)
 735                                         {
 736                                                 if($_->{flags} & WATCHME_T)
 737                                                 {
 738                                                         $_->{valid} = [-1, undef, undef];
 739                                                 }
 740                                         }
 741                                 }
 742                         }
 743
 744                         return 0;
 745                 };
 746
 747         for my $ip(keys %write_places)
 748         {
 749                 for my $operand(keys %{$write_places{$ip}})
 750                 {
 751                         # TODO verify it
 752                         my %left = map { $_ => 1 } @{$write_places{$ip}{$operand}};
 753                         my $isread = 0;
 754
 755                         my %writeplace_seen = ();
 756                         run_nfa $progs, $ip+1, \%left,
 757                                 sub
 758                                 {
 759                                         return { %{$_[0]} };
 760                                 },
 761                                 sub
 762                                 {
 763                                         my ($ip, $state) = @_;
 764                                         return $writeplace_seen{"$ip " . join " ", sort keys %$state}++;
 765                                 },
 766                                 sub
 767                                 {
 768                                         my ($ip, $state, $s, $c) = @_;
 769                                         for(qw(a b c))
 770                                         {
 771                                                 my $type = $c->{$_};
 772                                                 next
 773                                                         unless defined $type;
 774
 775                                                 my $ofs = $s->{$_};
 776                                                 if($type eq 'inglobal' || $type eq 'inglobalfunc')
 777                                                 {
 778                                                         if($state->{$ofs})
 779                                                         {
 780                                                                 $isread = 1;
 781                                                                 return -1; # exit TOTALLY
 782                                                         }
 783                                                 }
 784                                                 elsif($type eq 'inglobalvec')
 785                                                 {
 786                                                         if($state->{$ofs} || $state->{$ofs+1} || $state->{$ofs+2})
 787                                                         {
 788                                                                 $isread = 1;
 789                                                                 return -1; # exit TOTALLY
 790                                                         }
 791                                                 }
 792                                                 elsif($type eq 'outglobal')
 793                                                 {
 794                                                         delete $state->{$ofs};
 795                                                         return 1
 796                                                                 if !%$state;
 797                                                 }
 798                                                 elsif($type eq 'outglobalvec')
 799                                                 {
 800                                                         delete $state->{$ofs};
 801                                                         delete $state->{$ofs+1};
 802                                                         delete $state->{$ofs+2};
 803                                                         return 1
 804                                                                 if !%$state;
 805                                                 }
 806                                         }
 807                                         return 0;
 808                                 };
 809
 810                         if(!$isread)
 811                         {
 812                                 print "; Value is never used in $func->{debugname} at $ip.$operand\n";
 813                                 ++$warned{$ip}{$operand};
 814                         }
 815                 }
 816         }
 817
 818         disassemble_function($progs, $func, \%warned)
 819                 if keys %warned;
 820 }
 821
 822 use constant DEFAULTGLOBALS => [
 823         "OFS_NULL",
 824         "OFS_RETURN",
 825         "OFS_RETURN[1]",
 826         "OFS_RETURN[2]",
 827         "OFS_PARM0",
 828         "OFS_PARM0[1]",
 829         "OFS_PARM0[2]",
 830         "OFS_PARM1",
 831         "OFS_PARM1[1]",
 832         "OFS_PARM1[2]",
 833         "OFS_PARM2",
 834         "OFS_PARM2[1]",
 835         "OFS_PARM2[2]",
 836         "OFS_PARM3",
 837         "OFS_PARM3[1]",
 838         "OFS_PARM3[2]",
 839         "OFS_PARM4",
 840         "OFS_PARM4[1]",
 841         "OFS_PARM4[2]",
 842         "OFS_PARM5",
 843         "OFS_PARM5[1]",
 844         "OFS_PARM5[2]",
 845         "OFS_PARM6",
 846         "OFS_PARM6[1]",
 847         "OFS_PARM6[2]",
 848         "OFS_PARM7",
 849         "OFS_PARM7[1]",
 850         "OFS_PARM7[2]"
 851 ];
 852
 853 sub defaultglobal($)
 854 {
 855         my ($ofs) = @_;
 856         if($ofs < @{(DEFAULTGLOBALS)})
 857         {
 858                 return { ofs => $ofs, s_name => undef, debugname => DEFAULTGLOBALS->[$ofs], type => undef };
 859         }
 860         return { ofs => $ofs, s_name => undef, debugname => "<undefined>\@$ofs", type => undef };
 861 }
 862
 863 sub detect_constants($)
 864 {
 865         my ($progs) = @_;
 866         use constant GLOBALFLAG_R => 1; # read
 867         use constant GLOBALFLAG_W => 2; # written
 868         use constant GLOBALFLAG_S => 4; # saved
 869         use constant GLOBALFLAG_I => 8; # initialized
 870         use constant GLOBALFLAG_N => 16; # named
 871         use constant GLOBALFLAG_Q => 32; # unique to function
 872         use constant GLOBALFLAG_U => 64; # unused
 873         my @globalflags = (GLOBALFLAG_Q | GLOBALFLAG_U) x @{$progs->{globals}};
 874
 875         for(@{$progs->{functions}})
 876         {
 877                 for(keys %{$_->{globals_used}})
 878                 {
 879                         if($globalflags[$_] & GLOBALFLAG_Q)
 880                         {
 881                                 $globalflags[$_] &= ~GLOBALFLAG_Q;
 882                         }
 883                         else
 884                         {
 885                                 $globalflags[$_] &= ~GLOBALFLAG_U;
 886                         }
 887                 }
 888                 $globalflags[$_] |= GLOBALFLAG_R
 889                         for keys %{$_->{globals_read}};
 890                 $globalflags[$_] |= GLOBALFLAG_W
 891                         for keys %{$_->{globals_written}};
 892         }
 893
 894         my %offsets_saved = ();
 895         for(@{$progs->{globaldefs}})
 896         {
 897                 my $type = $_->{type};
 898                 my $name = $progs->{getstring}->($_->{s_name});
 899                 if($type->{save})
 900                 {
 901                         for my $i(0..(typesize($_->{type}{type})-1))
 902                         {
 903                                 $globalflags[$_->{ofs}] |= GLOBALFLAG_S;
 904                         }
 905                 }
 906                 if($name ne "")
 907                 {
 908                         for my $i(0..(typesize($_->{type}{type})-1))
 909                         {
 910                                 $globalflags[$_->{ofs}] |= GLOBALFLAG_N;
 911                         }
 912                 }
 913         }
 914         my %offsets_initialized = ();
 915         for(0..(@{$progs->{globals}}-1))
 916         {
 917                 if($progs->{globals}[$_]{v}{int})
 918                 {
 919                         $globalflags[$_] |= GLOBALFLAG_I;
 920                 }
 921         }
 922
 923         my @globaltypes = (undef) x @{$progs->{globals}};
 924
 925         my %istemp = ();
 926         for(0..(@{$progs->{globals}}-1))
 927         {
 928                 next
 929                         if $_ < @{(DEFAULTGLOBALS)};
 930                 if(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == 0)
 931                 {
 932                         $globaltypes[$_] = "unused";
 933                 }
 934                 elsif(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == GLOBALFLAG_R)
 935                 {
 936                         # so it is ro
 937                         if(($globalflags[$_] & GLOBALFLAG_N) == GLOBALFLAG_N)
 938                         {
 939                                 $globaltypes[$_] = "read_only";
 940                         }
 941                         elsif(($globalflags[$_] & GLOBALFLAG_S) == 0)
 942                         {
 943                                 $globaltypes[$_] = "const";
 944                         }
 945                         else
 946                         {
 947                                 $globaltypes[$_] = "read_only";
 948                         }
 949                 }
 950                 elsif(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == GLOBALFLAG_W)
 951                 {
 952                         $globaltypes[$_] = "write_only";
 953                 }
 954                 else
 955                 {
 956                         # now we know it is rw
 957                         if(($globalflags[$_] & GLOBALFLAG_N) == GLOBALFLAG_N)
 958                         {
 959                                 $globaltypes[$_] = "global";
 960                         }
 961                         elsif(($globalflags[$_] & (GLOBALFLAG_S | GLOBALFLAG_I)) == 0)
 962                         {
 963                                 $globaltypes[$_] = "temp";
 964                                 ++$istemp{$_};
 965                         }
 966                         elsif(($globalflags[$_] & (GLOBALFLAG_S | GLOBALFLAG_I)) == GLOBALFLAG_I)
 967                         {
 968                                 $globaltypes[$_] = "not_saved";
 969                         }
 970                         else
 971                         {
 972                                 $globaltypes[$_] = "global";
 973                         }
 974                 }
 975         }
 976         $progs->{temps} = \%istemp;
 977
 978         # globaldefs
 979         my @globaldefs = (undef) x @{$progs->{globaldefs}};
 980         for(@{$progs->{globaldefs}})
 981         {
 982                 my $s = $progs->{getstring}->($_->{s_name});
 983                 $_->{debugname} //= "\$" . "$s"
 984                         if length $s;
 985         }
 986         for(@{$progs->{globaldefs}})
 987         {
 988                 $globaldefs[$_->{ofs}] //= $_
 989                         if defined $_->{debugname};
 990         }
 991         for(@{$progs->{globaldefs}})
 992         {
 993                 $globaldefs[$_->{ofs}] //= $_;
 994         }
 995         for(0..(@{$progs->{globals}}-1))
 996         {
 997                 $globaldefs[$_] //= {
 998                         ofs => $_,
 999                         s_name => undef,
1000                         debugname => undef
1001                 };
1002         }
1003         for(0..(@{(DEFAULTGLOBALS)}-1))
1004         {
1005                 $globaldefs[$_] = { ofs => $_, s_name => undef, debugname => DEFAULTGLOBALS->[$_], type => undef };
1006                 $globaltypes[$_] = 'defglobal';
1007         }
1008         my %globaldefs_namecount = ();
1009         for(@globaldefs)
1010         {
1011                 $_->{globaltype} = $globaltypes[$_->{ofs}];
1012                 if(defined $_->{debugname})
1013                 {
1014                         # already has debugname
1015                 }
1016                 elsif($_->{globaltype} eq 'const')
1017                 {
1018                         $_->{debugname} = get_constant($progs, $progs->{globals}[$_->{ofs}]{v});
1019                 }
1020                 else
1021                 {
1022                         $_->{debugname} = "$_->{globaltype}_$_->{ofs}";
1023                 }
1024                 ++$globaldefs_namecount{$_->{debugname}};
1025         }
1026         for(@globaldefs)
1027         {
1028                 next
1029                         if $globaldefs_namecount{$_->{debugname}} <= 1;
1030                 #print "Not unique: $_->{debugname} at $_->{ofs}\n";
1031                 $_->{debugname} .= "\@$_->{ofs}";
1032         }
1033         $progs->{globaldef_byoffset} = sub
1034         {
1035                 my ($ofs) = @_;
1036                 my $def = $globaldefs[$ofs];
1037                 return $def;
1038         };
1039 }
1040
1041 sub parse_progs($)
1042 {
1043         my ($fh) = @_;
1044
1045         my %p = ();
1046
1047         print STDERR "Parsing header...\n";
1048         $p{header} = parse_section $fh, DPROGRAMS_T, 0, undef, 1;
1049
1050         print STDERR "Parsing strings...\n";
1051         $p{strings} = get_section $fh, $p{header}{ofs_strings}, $p{header}{numstrings};
1052         $p{getstring} = sub
1053         {
1054                 my ($startpos) = @_;
1055                 my $endpos = index $p{strings}, "\0", $startpos;
1056                 return substr $p{strings}, $startpos, $endpos - $startpos;
1057         };
1058
1059         print STDERR "Parsing statements...\n";
1060         $p{statements} = [parse_section $fh, DSTATEMENT_T, $p{header}{ofs_statements}, undef, $p{header}{numstatements}];
1061
1062         print STDERR "Fixing statements...\n";
1063         for my $s(@{$p{statements}})
1064         {
1065                 my $c = checkop $s->{op};
1066
1067                 for(qw(a b c))
1068                 {
1069                         my $type = $c->{$_};
1070                         next
1071                                 unless defined $type;
1072
1073                         if($type eq 'inglobal' || $type eq 'inglobalfunc')
1074                         {
1075                                 $s->{$_} &= 0xFFFF;
1076                         }
1077                         elsif($type eq 'inglobalvec')
1078                         {
1079                                 $s->{$_} &= 0xFFFF;
1080                         }
1081                         elsif($type eq 'outglobal')
1082                         {
1083                                 $s->{$_} &= 0xFFFF;
1084                         }
1085                         elsif($type eq 'outglobalvec')
1086                         {
1087                                 $s->{$_} &= 0xFFFF;
1088                         }
1089                 }
1090
1091         }
1092
1093         print STDERR "Parsing globaldefs...\n";
1094         $p{globaldefs} = [parse_section $fh, DDEF_T, $p{header}{ofs_globaldefs}, undef, $p{header}{numglobaldefs}];
1095
1096         print STDERR "Parsing fielddefs...\n";
1097         $p{fielddefs} = [parse_section $fh, DDEF_T, $p{header}{ofs_fielddefs}, undef, $p{header}{numfielddefs}];
1098
1099         print STDERR "Parsing globals...\n";
1100         $p{globals} = [parse_section $fh, DGLOBAL_T, $p{header}{ofs_globals}, undef, $p{header}{numglobals}];
1101
1102         print STDERR "Parsing functions...\n";
1103         $p{functions} = [parse_section $fh, DFUNCTION_T, $p{header}{ofs_functions}, undef, $p{header}{numfunctions}];
1104
1105         print STDERR "Looking for error()...\n";
1106         $p{error_func} = {};
1107         for(@{$p{globaldefs}})
1108         {
1109                 next
1110                         if $p{getstring}($_->{s_name}) ne 'error';
1111                 my $v = $p{globals}[$_->{ofs}]{v}{int};
1112                 next
1113                         if $v <= 0 || $v >= @{$p{functions}};
1114                 my $first = $p{functions}[$v]{first_statement};
1115                 next
1116                         if $first >= 0;
1117                 print STDERR "Detected error() at offset $_->{ofs} (builtin #@{[-$first]})\n";
1118                 $p{error_func}{$_->{ofs}} = 1;
1119         }
1120
1121         print STDERR "Scanning functions...\n";
1122         for(@{$p{functions}})
1123         {
1124                 my $file = $p{getstring}->($_->{s_file});
1125                 my $name = $p{getstring}->($_->{s_name});
1126                 $name = "$file:$name"
1127                         if length $file;
1128                 $_->{debugname} = $name;
1129
1130                 next
1131                         if $_->{first_statement} < 0;
1132
1133                 my %statements = ();
1134                 my %come_from = ();
1135                 my %go_to = ();
1136                 my %globals_read = ();
1137                 my %globals_written = ();
1138                 my %globals_used = ();
1139
1140                 run_nfa \%p, $_->{first_statement}, "", id, nfa_default_state_checker,
1141                         sub
1142                         {
1143                                 my ($ip, $state, $s, $c) = @_;
1144                                 ++$statements{$ip};
1145
1146                                 if(my $j = $c->{isjump})
1147                                 {
1148                                         my $t = $ip + $s->{$j};
1149                                         $come_from{$t}{$ip} = $c->{isconditional};
1150                                         $go_to{$ip}{$t} = $c->{isconditional};
1151                                 }
1152
1153                                 for my $o(qw(a b c))
1154                                 {
1155                                         my $type = $c->{$o}
1156                                                 or next;
1157                                         my $ofs = $s->{$o};
1158
1159                                         my $read = sub
1160                                         {
1161                                                 my ($ofs) = @_;
1162                                                 $globals_read{$ofs}{$ip}{$o} = 1;
1163                                                 $globals_used{$ofs} = 1;
1164                                         };
1165                                         my $write = sub
1166                                         {
1167                                                 my ($ofs) = @_;
1168                                                 $globals_written{$ofs}{$ip}{$o} = 1;
1169                                                 $globals_used{$ofs} = 1;
1170                                         };
1171
1172                                         if($type eq 'inglobal' || $type eq 'inglobalfunc')
1173                                         {
1174                                                 $read->($ofs);
1175                                         }
1176                                         elsif($type eq 'inglobalvec')
1177                                         {
1178                                                 $read->($ofs);
1179                                                 $read->($ofs+1);
1180                                                 $read->($ofs+2);
1181                                         }
1182                                         elsif($type eq 'outglobal')
1183                                         {
1184                                                 $write->($ofs);
1185                                         }
1186                                         elsif($type eq 'outglobalvec')
1187                                         {
1188                                                 $write->($ofs);
1189                                                 $write->($ofs+1);
1190                                                 $write->($ofs+2);
1191                                         }
1192                                 }
1193
1194                                 return 0;
1195                         };
1196
1197                 $_->{statements} = \%statements;
1198                 $_->{come_from} = \%come_from;
1199                 $_->{go_to} = \%go_to;
1200                 $_->{globals_read} = \%globals_read;
1201                 $_->{globals_written} = \%globals_written;
1202                 $_->{globals_used} = \%globals_used;
1203
1204                 # using this info, we could now identify basic blocks
1205         }
1206
1207         print STDERR "Detecting constants and temps, and naming...\n";
1208         detect_constants \%p;
1209
1210         # what do we want to do?
1211         my $checkfunc = \&find_uninitialized_locals;
1212         #my $checkfunc = \&disassemble_function;
1213         for(sort { $a->{debugname} cmp $b->{debugname} } @{$p{functions}})
1214         {
1215                 $checkfunc->(\%p, $_);
1216         }
1217 }
1218
1219 open my $fh, '<', $ARGV[0];
1220 parse_progs $fh;