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                 next
 539                         if exists $watchme{$_};
 540                 if($progs->{temps}{$_})
 541                 {
 542                         # shared temp
 543                         $watchme{$_} = WATCHME_T | WATCHME_X
 544                 }
 545                 else
 546                 {
 547                         # unique temp
 548                         $watchme{$_} = WATCHME_X
 549                 }
 550         }
 551
 552         $watchme{$_} |= WATCHME_R
 553                 for keys %{$func->{globals_read}};
 554         $watchme{$_} |= WATCHME_W
 555                 for keys %{$func->{globals_written}};
 556
 557         my %write_places = ();
 558         for my $ofs(keys %{$func->{globals_written}})
 559         {
 560                 next
 561                         unless exists $watchme{$ofs} and $watchme{$ofs} & WATCHME_X;
 562                 for my $ip(keys %{$func->{globals_written}{$ofs}})
 563                 {
 564                         for my $op(keys %{$func->{globals_written}{$ofs}{$ip}})
 565                         {
 566                                 push @{$write_places{$ip}{$op}}, $ofs;
 567                         }
 568                 }
 569         }
 570
 571         for(keys %watchme)
 572         {
 573                 delete $watchme{$_}
 574                         if ($watchme{$_} & (WATCHME_R | WATCHME_W | WATCHME_X)) != (WATCHME_R | WATCHME_W | WATCHME_X);
 575         }
 576
 577         return
 578                 if not keys %watchme;
 579
 580         for(keys %watchme)
 581         {
 582                 $watchme{$_} = {
 583                         flags => $watchme{$_},
 584                         valid => [0, undef, undef]
 585                 };
 586         }
 587
 588         # mark parameters as initialized
 589         for($func->{parm_start} .. ($p-1))
 590         {
 591                 $watchme{$_}{valid} = [1, undef, undef]
 592                         if defined $watchme{$_};
 593         }
 594         # an initial run of STORE instruction is for receiving extra parameters
 595         # (beyond 8). Only possible if the function is declared as having 8 params.
 596         # Extra parameters behave otherwise like temps, but are initialized at
 597         # startup.
 598         for($func->{first_statement} .. (@{$progs->{statements}}-1))
 599         {
 600                 my $s = $progs->{statements}[$_];
 601                 if($s->{op} eq 'STORE_V')
 602                 {
 603                         $watchme{$s->{a}}{valid} = [1, undef, undef]
 604                                 if defined $watchme{$s->{a}};
 605                         $watchme{$s->{a}+1}{valid} = [1, undef, undef]
 606                                 if defined $watchme{$s->{a}+1};
 607                         $watchme{$s->{a}+2}{valid} = [1, undef, undef]
 608                                 if defined $watchme{$s->{a}+2};
 609                 }
 610                 elsif($s->{op} =~ /^STORE_/)
 611                 {
 612                         $watchme{$s->{a}}{valid} = [1, undef, undef]
 613                                 if defined $watchme{$s->{a}};
 614                 }
 615                 else
 616                 {
 617                         last;
 618                 }
 619         }
 620
 621         my %warned = ();
 622         my %ip_seen = ();
 623         run_nfa $progs, $func->{first_statement}, \%watchme,
 624                 sub {
 625                         my ($h) = @_;
 626                         return { map { $_ => { %{$h->{$_}} } } keys %$h };
 627                 },
 628                 sub {
 629                         my ($ip, $state) = @_;
 630
 631                         my $s = $ip_seen{$ip};
 632                         if($s)
 633                         {
 634                                 # if $state is stronger or equal to $s, return 1
 635
 636                                 for(keys %$state)
 637                                 {
 638                                         if($state->{$_}{valid}[0] < $s->{$_})
 639                                         {
 640                                                 # The current state is LESS valid than the previously run one. We NEED to run this.
 641                                                 # The saved state can safely become the intersection [citation needed].
 642                                                 for(keys %$state)
 643                                                 {
 644                                                         $s->{$_} = $state->{$_}{valid}[0]
 645                                                                 if $state->{$_}{valid}[0] < $s->{$_};
 646                                                 }
 647                                                 return 0;
 648                                         }
 649                                 }
 650                                 # if we get here, $state is stronger or equal. No need to try it.
 651                                 return 1;
 652                         }
 653                         else
 654                         {
 655                                 # Never seen this IP yet.
 656                                 $ip_seen{$ip} = { map { ($_ => $state->{$_}{valid}[0]); } keys %$state };
 657                                 return 0;
 658                         }
 659                 },
 660                 sub {
 661                         my ($ip, $state, $s, $c) = @_;
 662                         my $op = $s->{op};
 663
 664                         my $return_hack = $c->{isreturn} // 0;
 665
 666                         for(qw(a b c))
 667                         {
 668                                 my $type = $c->{$_};
 669                                 next
 670                                         unless defined $type;
 671
 672                                 my $ofs = $s->{$_};
 673
 674                                 my $read = sub
 675                                 {
 676                                         my ($ofs) = @_;
 677                                         ++$return_hack
 678                                                 if $return_hack;
 679                                         return
 680                                                 if not exists $state->{$ofs};
 681                                         my $valid = $state->{$ofs}{valid};
 682                                         if($valid->[0] == 0)
 683                                         {
 684                                                 if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
 685                                                 {
 686                                                         print "; Use of uninitialized value $ofs in $func->{debugname} at $ip.$_\n";
 687                                                         ++$warned{$ip}{$_};
 688                                                 }
 689                                         }
 690                                         elsif($valid->[0] < 0)
 691                                         {
 692                                                 if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
 693                                                 {
 694                                                         print "; Use of temporary $ofs across CALL in $func->{debugname} at $ip.$_\n";
 695                                                         ++$warned{$ip}{$_};
 696                                                 }
 697                                         }
 698                                         else
 699                                         {
 700                                                 # it's VALID
 701                                                 if(defined $valid->[1])
 702                                                 {
 703                                                         delete $write_places{$valid->[1]}{$valid->[2]};
 704                                                 }
 705                                         }
 706                                 };
 707                                 my $write = sub
 708                                 {
 709                                         my ($ofs) = @_;
 710                                         $state->{$ofs}{valid} = [1, $ip, $_]
 711                                                 if exists $state->{$ofs};
 712                                 };
 713
 714                                 if($type eq 'inglobal' || $type eq 'inglobalfunc')
 715                                 {
 716                                         $read->($ofs);
 717                                 }
 718                                 elsif($type eq 'inglobalvec')
 719                                 {
 720                                         $read->($ofs);
 721                                         $read->($ofs+1);
 722                                         $read->($ofs+2);
 723                                 }
 724                                 elsif($type eq 'outglobal')
 725                                 {
 726                                         $write->($ofs);
 727                                 }
 728                                 elsif($type eq 'outglobalvec')
 729                                 {
 730                                         $write->($ofs);
 731                                         $write->($ofs+1);
 732                                         $write->($ofs+2);
 733                                 }
 734                         }
 735                         if($c->{iscall})
 736                         {
 737                                 # builtin calls may clobber stuff
 738                                 my $func = $s->{a};
 739                                 my $funcid = $progs->{globals}[$func]{v}{int};
 740                                 my $funcobj = $progs->{functions}[$funcid];
 741                                 if(!$funcobj || $funcobj->{first_statement} >= 0)
 742                                 {
 743                                         # invalidate temps
 744                                         for(values %$state)
 745                                         {
 746                                                 if($_->{flags} & WATCHME_T)
 747                                                 {
 748                                                         $_->{valid} = [-1, undef, undef];
 749                                                 }
 750                                         }
 751                                 }
 752                         }
 753
 754                         return 0;
 755                 };
 756
 757         for my $ip(keys %write_places)
 758         {
 759                 for my $operand(keys %{$write_places{$ip}})
 760                 {
 761                         # TODO verify it
 762                         my %left = map { $_ => 1 } @{$write_places{$ip}{$operand}};
 763                         my $isread = 0;
 764
 765                         my %writeplace_seen = ();
 766                         run_nfa $progs, $ip+1, \%left,
 767                                 sub
 768                                 {
 769                                         return { %{$_[0]} };
 770                                 },
 771                                 sub
 772                                 {
 773                                         my ($ip, $state) = @_;
 774                                         return $writeplace_seen{"$ip " . join " ", sort keys %$state}++;
 775                                 },
 776                                 sub
 777                                 {
 778                                         my ($ip, $state, $s, $c) = @_;
 779                                         for(qw(a b c))
 780                                         {
 781                                                 my $type = $c->{$_};
 782                                                 next
 783                                                         unless defined $type;
 784
 785                                                 my $ofs = $s->{$_};
 786                                                 if($type eq 'inglobal' || $type eq 'inglobalfunc')
 787                                                 {
 788                                                         if($state->{$ofs})
 789                                                         {
 790                                                                 $isread = 1;
 791                                                                 return -1; # exit TOTALLY
 792                                                         }
 793                                                 }
 794                                                 elsif($type eq 'inglobalvec')
 795                                                 {
 796                                                         if($state->{$ofs} || $state->{$ofs+1} || $state->{$ofs+2})
 797                                                         {
 798                                                                 $isread = 1;
 799                                                                 return -1; # exit TOTALLY
 800                                                         }
 801                                                 }
 802                                                 elsif($type eq 'outglobal')
 803                                                 {
 804                                                         delete $state->{$ofs};
 805                                                         return 1
 806                                                                 if !%$state;
 807                                                 }
 808                                                 elsif($type eq 'outglobalvec')
 809                                                 {
 810                                                         delete $state->{$ofs};
 811                                                         delete $state->{$ofs+1};
 812                                                         delete $state->{$ofs+2};
 813                                                         return 1
 814                                                                 if !%$state;
 815                                                 }
 816                                         }
 817                                         return 0;
 818                                 };
 819
 820                         if(!$isread)
 821                         {
 822                                 print "; Value is never used in $func->{debugname} at $ip.$operand\n";
 823                                 ++$warned{$ip}{$operand};
 824                         }
 825                 }
 826         }
 827
 828         disassemble_function($progs, $func, \%warned)
 829                 if keys %warned;
 830 }
 831
 832 use constant DEFAULTGLOBALS => [
 833         "OFS_NULL",
 834         "OFS_RETURN",
 835         "OFS_RETURN[1]",
 836         "OFS_RETURN[2]",
 837         "OFS_PARM0",
 838         "OFS_PARM0[1]",
 839         "OFS_PARM0[2]",
 840         "OFS_PARM1",
 841         "OFS_PARM1[1]",
 842         "OFS_PARM1[2]",
 843         "OFS_PARM2",
 844         "OFS_PARM2[1]",
 845         "OFS_PARM2[2]",
 846         "OFS_PARM3",
 847         "OFS_PARM3[1]",
 848         "OFS_PARM3[2]",
 849         "OFS_PARM4",
 850         "OFS_PARM4[1]",
 851         "OFS_PARM4[2]",
 852         "OFS_PARM5",
 853         "OFS_PARM5[1]",
 854         "OFS_PARM5[2]",
 855         "OFS_PARM6",
 856         "OFS_PARM6[1]",
 857         "OFS_PARM6[2]",
 858         "OFS_PARM7",
 859         "OFS_PARM7[1]",
 860         "OFS_PARM7[2]"
 861 ];
 862
 863 sub defaultglobal($)
 864 {
 865         my ($ofs) = @_;
 866         if($ofs < @{(DEFAULTGLOBALS)})
 867         {
 868                 return { ofs => $ofs, s_name => undef, debugname => DEFAULTGLOBALS->[$ofs], type => undef };
 869         }
 870         return { ofs => $ofs, s_name => undef, debugname => "<undefined>\@$ofs", type => undef };
 871 }
 872
 873 sub detect_constants($)
 874 {
 875         my ($progs) = @_;
 876         use constant GLOBALFLAG_R => 1; # read
 877         use constant GLOBALFLAG_W => 2; # written
 878         use constant GLOBALFLAG_S => 4; # saved
 879         use constant GLOBALFLAG_I => 8; # initialized
 880         use constant GLOBALFLAG_N => 16; # named
 881         use constant GLOBALFLAG_Q => 32; # unique to function
 882         use constant GLOBALFLAG_U => 64; # unused
 883         my @globalflags = (GLOBALFLAG_Q | GLOBALFLAG_U) x @{$progs->{globals}};
 884
 885         for(@{$progs->{functions}})
 886         {
 887                 for(keys %{$_->{globals_used}})
 888                 {
 889                         if($globalflags[$_] & GLOBALFLAG_U)
 890                         {
 891                                 $globalflags[$_] &= ~GLOBALFLAG_U;
 892                         }
 893                         elsif($globalflags[$_] & GLOBALFLAG_Q)
 894                         {
 895                                 $globalflags[$_] &= ~GLOBALFLAG_Q;
 896                         }
 897                 }
 898                 $globalflags[$_] |= GLOBALFLAG_R
 899                         for keys %{$_->{globals_read}};
 900                 $globalflags[$_] |= GLOBALFLAG_W
 901                         for keys %{$_->{globals_written}};
 902         }
 903
 904         my %offsets_saved = ();
 905         for(@{$progs->{globaldefs}})
 906         {
 907                 my $type = $_->{type};
 908                 my $name = $progs->{getstring}->($_->{s_name});
 909                 if($type->{save})
 910                 {
 911                         for my $i(0..(typesize($_->{type}{type})-1))
 912                         {
 913                                 $globalflags[$_->{ofs}] |= GLOBALFLAG_S;
 914                         }
 915                 }
 916                 if($name ne "")
 917                 {
 918                         for my $i(0..(typesize($_->{type}{type})-1))
 919                         {
 920                                 $globalflags[$_->{ofs}] |= GLOBALFLAG_N;
 921                         }
 922                 }
 923         }
 924         my %offsets_initialized = ();
 925         for(0..(@{$progs->{globals}}-1))
 926         {
 927                 if($progs->{globals}[$_]{v}{int})
 928                 {
 929                         $globalflags[$_] |= GLOBALFLAG_I;
 930                 }
 931         }
 932
 933         my @globaltypes = (undef) x @{$progs->{globals}};
 934
 935         my %istemp = ();
 936         for(0..(@{$progs->{globals}}-1))
 937         {
 938                 next
 939                         if $_ < @{(DEFAULTGLOBALS)};
 940                 if(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == 0)
 941                 {
 942                         $globaltypes[$_] = "unused";
 943                 }
 944                 elsif(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == GLOBALFLAG_R)
 945                 {
 946                         # so it is ro
 947                         if(($globalflags[$_] & GLOBALFLAG_N) == GLOBALFLAG_N)
 948                         {
 949                                 $globaltypes[$_] = "read_only";
 950                         }
 951                         elsif(($globalflags[$_] & GLOBALFLAG_S) == 0)
 952                         {
 953                                 $globaltypes[$_] = "const";
 954                         }
 955                         else
 956                         {
 957                                 $globaltypes[$_] = "read_only";
 958                         }
 959                 }
 960                 elsif(($globalflags[$_] & (GLOBALFLAG_R | GLOBALFLAG_W)) == GLOBALFLAG_W)
 961                 {
 962                         $globaltypes[$_] = "write_only";
 963                 }
 964                 else
 965                 {
 966                         # now we know it is rw
 967                         if(($globalflags[$_] & GLOBALFLAG_N) == GLOBALFLAG_N)
 968                         {
 969                                 $globaltypes[$_] = "global";
 970                         }
 971                         elsif(($globalflags[$_] & (GLOBALFLAG_S | GLOBALFLAG_I | GLOBALFLAG_Q)) == GLOBALFLAG_Q)
 972                         {
 973                                 $globaltypes[$_] = "uniquetemp";
 974                                 $istemp{$_} = 0;
 975                         }
 976                         elsif(($globalflags[$_] & (GLOBALFLAG_S | GLOBALFLAG_I | GLOBALFLAG_Q)) == 0)
 977                         {
 978                                 $globaltypes[$_] = "temp";
 979                                 $istemp{$_} = 1;
 980                         }
 981                         elsif(($globalflags[$_] & (GLOBALFLAG_S | GLOBALFLAG_I)) == GLOBALFLAG_I)
 982                         {
 983                                 $globaltypes[$_] = "not_saved";
 984                         }
 985                         else
 986                         {
 987                                 $globaltypes[$_] = "global";
 988                         }
 989                 }
 990         }
 991         $progs->{temps} = \%istemp;
 992
 993         # globaldefs
 994         my @globaldefs = (undef) x @{$progs->{globaldefs}};
 995         for(@{$progs->{globaldefs}})
 996         {
 997                 my $s = $progs->{getstring}->($_->{s_name});
 998                 $_->{debugname} //= "\$" . "$s"
 999                         if length $s;
1000         }
1001         for(@{$progs->{globaldefs}})
1002         {
1003                 $globaldefs[$_->{ofs}] //= $_
1004                         if defined $_->{debugname};
1005         }
1006         for(@{$progs->{globaldefs}})
1007         {
1008                 $globaldefs[$_->{ofs}] //= $_;
1009         }
1010         for(0..(@{$progs->{globals}}-1))
1011         {
1012                 $globaldefs[$_] //= {
1013                         ofs => $_,
1014                         s_name => undef,
1015                         debugname => undef
1016                 };
1017         }
1018         for(0..(@{(DEFAULTGLOBALS)}-1))
1019         {
1020                 $globaldefs[$_] = { ofs => $_, s_name => undef, debugname => DEFAULTGLOBALS->[$_], type => undef };
1021                 $globaltypes[$_] = 'defglobal';
1022         }
1023         my %globaldefs_namecount = ();
1024         for(@globaldefs)
1025         {
1026                 $_->{globaltype} = $globaltypes[$_->{ofs}];
1027                 if(defined $_->{debugname})
1028                 {
1029                         # already has debugname
1030                 }
1031                 elsif($_->{globaltype} eq 'const')
1032                 {
1033                         $_->{debugname} = get_constant($progs, $progs->{globals}[$_->{ofs}]{v});
1034                 }
1035                 else
1036                 {
1037                         $_->{debugname} = "$_->{globaltype}_$_->{ofs}";
1038                 }
1039                 ++$globaldefs_namecount{$_->{debugname}};
1040         }
1041         for(@globaldefs)
1042         {
1043                 next
1044                         if $globaldefs_namecount{$_->{debugname}} <= 1;
1045                 #print "Not unique: $_->{debugname} at $_->{ofs}\n";
1046                 $_->{debugname} .= "\@$_->{ofs}";
1047         }
1048         $progs->{globaldef_byoffset} = sub
1049         {
1050                 my ($ofs) = @_;
1051                 my $def = $globaldefs[$ofs];
1052                 return $def;
1053         };
1054 }
1055
1056 sub parse_progs($)
1057 {
1058         my ($fh) = @_;
1059
1060         my %p = ();
1061
1062         print STDERR "Parsing header...\n";
1063         $p{header} = parse_section $fh, DPROGRAMS_T, 0, undef, 1;
1064
1065         print STDERR "Parsing strings...\n";
1066         $p{strings} = get_section $fh, $p{header}{ofs_strings}, $p{header}{numstrings};
1067         $p{getstring} = sub
1068         {
1069                 my ($startpos) = @_;
1070                 my $endpos = index $p{strings}, "\0", $startpos;
1071                 return substr $p{strings}, $startpos, $endpos - $startpos;
1072         };
1073
1074         print STDERR "Parsing statements...\n";
1075         $p{statements} = [parse_section $fh, DSTATEMENT_T, $p{header}{ofs_statements}, undef, $p{header}{numstatements}];
1076
1077         print STDERR "Fixing statements...\n";
1078         for my $s(@{$p{statements}})
1079         {
1080                 my $c = checkop $s->{op};
1081
1082                 for(qw(a b c))
1083                 {
1084                         my $type = $c->{$_};
1085                         next
1086                                 unless defined $type;
1087
1088                         if($type eq 'inglobal' || $type eq 'inglobalfunc')
1089                         {
1090                                 $s->{$_} &= 0xFFFF;
1091                         }
1092                         elsif($type eq 'inglobalvec')
1093                         {
1094                                 $s->{$_} &= 0xFFFF;
1095                         }
1096                         elsif($type eq 'outglobal')
1097                         {
1098                                 $s->{$_} &= 0xFFFF;
1099                         }
1100                         elsif($type eq 'outglobalvec')
1101                         {
1102                                 $s->{$_} &= 0xFFFF;
1103                         }
1104                 }
1105
1106         }
1107
1108         print STDERR "Parsing globaldefs...\n";
1109         $p{globaldefs} = [parse_section $fh, DDEF_T, $p{header}{ofs_globaldefs}, undef, $p{header}{numglobaldefs}];
1110
1111         print STDERR "Parsing fielddefs...\n";
1112         $p{fielddefs} = [parse_section $fh, DDEF_T, $p{header}{ofs_fielddefs}, undef, $p{header}{numfielddefs}];
1113
1114         print STDERR "Parsing globals...\n";
1115         $p{globals} = [parse_section $fh, DGLOBAL_T, $p{header}{ofs_globals}, undef, $p{header}{numglobals}];
1116
1117         print STDERR "Parsing functions...\n";
1118         $p{functions} = [parse_section $fh, DFUNCTION_T, $p{header}{ofs_functions}, undef, $p{header}{numfunctions}];
1119
1120         print STDERR "Looking for error()...\n";
1121         $p{error_func} = {};
1122         for(@{$p{globaldefs}})
1123         {
1124                 next
1125                         if $p{getstring}($_->{s_name}) ne 'error';
1126                 my $v = $p{globals}[$_->{ofs}]{v}{int};
1127                 next
1128                         if $v <= 0 || $v >= @{$p{functions}};
1129                 my $first = $p{functions}[$v]{first_statement};
1130                 next
1131                         if $first >= 0;
1132                 print STDERR "Detected error() at offset $_->{ofs} (builtin #@{[-$first]})\n";
1133                 $p{error_func}{$_->{ofs}} = 1;
1134         }
1135
1136         print STDERR "Scanning functions...\n";
1137         for(@{$p{functions}})
1138         {
1139                 my $file = $p{getstring}->($_->{s_file});
1140                 my $name = $p{getstring}->($_->{s_name});
1141                 $name = "$file:$name"
1142                         if length $file;
1143                 $_->{debugname} = $name;
1144
1145                 next
1146                         if $_->{first_statement} < 0;
1147
1148                 my %statements = ();
1149                 my %come_from = ();
1150                 my %go_to = ();
1151                 my %globals_read = ();
1152                 my %globals_written = ();
1153                 my %globals_used = ();
1154
1155                 run_nfa \%p, $_->{first_statement}, "", id, nfa_default_state_checker,
1156                         sub
1157                         {
1158                                 my ($ip, $state, $s, $c) = @_;
1159                                 ++$statements{$ip};
1160
1161                                 if(my $j = $c->{isjump})
1162                                 {
1163                                         my $t = $ip + $s->{$j};
1164                                         $come_from{$t}{$ip} = $c->{isconditional};
1165                                         $go_to{$ip}{$t} = $c->{isconditional};
1166                                 }
1167
1168                                 for my $o(qw(a b c))
1169                                 {
1170                                         my $type = $c->{$o}
1171                                                 or next;
1172                                         my $ofs = $s->{$o};
1173
1174                                         my $read = sub
1175                                         {
1176                                                 my ($ofs) = @_;
1177                                                 $globals_read{$ofs}{$ip}{$o} = 1;
1178                                                 $globals_used{$ofs} = 1;
1179                                         };
1180                                         my $write = sub
1181                                         {
1182                                                 my ($ofs) = @_;
1183                                                 $globals_written{$ofs}{$ip}{$o} = 1;
1184                                                 $globals_used{$ofs} = 1;
1185                                         };
1186
1187                                         if($type eq 'inglobal' || $type eq 'inglobalfunc')
1188                                         {
1189                                                 $read->($ofs);
1190                                         }
1191                                         elsif($type eq 'inglobalvec')
1192                                         {
1193                                                 $read->($ofs);
1194                                                 $read->($ofs+1);
1195                                                 $read->($ofs+2);
1196                                         }
1197                                         elsif($type eq 'outglobal')
1198                                         {
1199                                                 $write->($ofs);
1200                                         }
1201                                         elsif($type eq 'outglobalvec')
1202                                         {
1203                                                 $write->($ofs);
1204                                                 $write->($ofs+1);
1205                                                 $write->($ofs+2);
1206                                         }
1207                                 }
1208
1209                                 return 0;
1210                         };
1211
1212                 $_->{statements} = \%statements;
1213                 $_->{come_from} = \%come_from;
1214                 $_->{go_to} = \%go_to;
1215                 $_->{globals_read} = \%globals_read;
1216                 $_->{globals_written} = \%globals_written;
1217                 $_->{globals_used} = \%globals_used;
1218
1219                 # using this info, we could now identify basic blocks
1220         }
1221
1222         print STDERR "Detecting constants and temps, and naming...\n";
1223         detect_constants \%p;
1224
1225         # what do we want to do?
1226         my $checkfunc = \&find_uninitialized_locals;
1227         #my $checkfunc = \&disassemble_function;
1228         for(sort { $a->{debugname} cmp $b->{debugname} } @{$p{functions}})
1229         {
1230                 $checkfunc->(\%p, $_);
1231         }
1232 }
1233
1234 open my $fh, '<', $ARGV[0];
1235 parse_progs $fh;