insns.pl

   1 #!/usr/bin/perl
   2 #
   3 # insns.pl   produce insnsa.c, insnsd.c, insnsi.h, insnsn.c from insns.dat
   4 #
   5 # The Netwide Assembler is copyright (C) 1996 Simon Tatham and
   6 # Julian Hall. All rights reserved. The software is
   7 # redistributable under the license given in the file "LICENSE"
   8 # distributed in the NASM archive.
   9
  10 # Opcode prefixes which need their own opcode tables
  11 # LONGER PREFIXES FIRST!
  12 @disasm_prefixes = qw(0F24 0F25 0F38 0F3A 0F7A 0FA6 0FA7 0F);
  13
  14 # This should match MAX_OPERANDS from nasm.h
  15 $MAX_OPERANDS = 5;
  16
  17 print STDERR "Reading insns.dat...\n";
  18
  19 @args   = ();
  20 undef $output;
  21 foreach $arg ( @ARGV ) {
  22     if ( $arg =~ /^\-/ ) {
  23         if  ( $arg =~ /^\-([abdin])$/ ) {
  24             $output = $1;
  25         } else {
  26             die "$0: Unknown option: ${arg}\n";
  27         }
  28     } else {
  29         push (@args, $arg);
  30     }
  31 }
  32
  33 $fname = "insns.dat" unless $fname = $args[0];
  34 open (F, $fname) || die "unable to open $fname";
  35
  36 %dinstables = ();
  37 @bytecode_list = ();
  38
  39 $line = 0;
  40 $insns = 0;
  41 while (<F>) {
  42   $line++;
  43   chomp;
  44   next if ( /^\s*(\;.*|)$/ );   # comments or blank lines
  45
  46   unless (/^\s*(\S+)\s+(\S+)\s+(\S+|\[.*\])\s+(\S+)\s*$/) {
  47       warn "line $line does not contain four fields\n";
  48       next;
  49   }
  50   @fields = ($1, $2, $3, $4);
  51   ($formatted, $nd) = format_insn(@fields);
  52   if ($formatted) {
  53     $insns++;
  54     $aname = "aa_$fields[0]";
  55     push @$aname, $formatted;
  56   }
  57   if ( $fields[0] =~ /cc$/ ) {
  58       # Conditional instruction
  59       $k_opcodes_cc{$fields[0]}++;
  60   } else {
  61       # Unconditional instruction
  62       $k_opcodes{$fields[0]}++;
  63   }
  64   if ($formatted && !$nd) {
  65     push @big, $formatted;
  66     my @sseq = startseq($fields[2]);
  67     foreach $i (@sseq) {
  68         if (!defined($dinstables{$i})) {
  69             $dinstables{$i} = [];
  70         }
  71         push(@{$dinstables{$i}}, $#big);
  72     }
  73   }
  74 }
  75
  76 close F;
  77
  78 #
  79 # Generate the bytecode array.  At this point, @bytecode_list contains
  80 # the full set of bytecodes.
  81 #
  82
  83 # Sort by descending length
  84 @bytecode_list = sort { scalar(@$b) <=> scalar(@$a) } @bytecode_list;
  85 @bytecode_array = ();
  86 %bytecode_pos = ();
  87 $bytecode_next = 0;
  88 foreach $bl (@bytecode_list) {
  89     my $h = hexstr(@$bl);
  90     next if (defined($bytecode_pos{$h}));
  91
  92     push(@bytecode_array, $bl);
  93     while ($h ne '') {
  94         $bytecode_pos{$h} = $bytecode_next;
  95         $h = substr($h, 2);
  96         $bytecode_next++;
  97     }
  98 }
  99 undef @bytecode_list;
 100
 101 @opcodes    = sort keys(%k_opcodes);
 102 @opcodes_cc = sort keys(%k_opcodes_cc);
 103
 104 if ( !defined($output) || $output eq 'b') {
 105     print STDERR "Writing insnsb.c...\n";
 106
 107     open B, ">insnsb.c";
 108
 109     print B "/* This file auto-generated from insns.dat by insns.pl" .
 110         " - don't edit it */\n\n";
 111
 112     print B "#include \"nasm.h\"\n";
 113     print B "#include \"insns.h\"\n\n";
 114
 115     print B "const uint8_t nasm_bytecodes[$bytecode_next] = {\n";
 116
 117     $p = 0;
 118     foreach $bl (@bytecode_array) {
 119         printf B "    /* %5d */ ", $p;
 120         foreach $d (@$bl) {
 121             printf B "%#o,", $d;
 122             $p++;
 123         }
 124         printf B "\n";
 125     }
 126     print B "};\n";
 127
 128     close B;
 129 }
 130
 131 if ( !defined($output) || $output eq 'a' ) {
 132     print STDERR "Writing insnsa.c...\n";
 133
 134     open A, ">insnsa.c";
 135
 136     print A "/* This file auto-generated from insns.dat by insns.pl" .
 137         " - don't edit it */\n\n";
 138
 139     print A "#include \"nasm.h\"\n";
 140     print A "#include \"insns.h\"\n\n";
 141
 142     foreach $i (@opcodes, @opcodes_cc) {
 143         print A "static const struct itemplate instrux_${i}[] = {\n";
 144         $aname = "aa_$i";
 145         foreach $j (@$aname) {
 146             print A "    ", codesubst($j), "\n";
 147         }
 148         print A "    ITEMPLATE_END\n};\n\n";
 149     }
 150     print A "const struct itemplate * const nasm_instructions[] = {\n";
 151     foreach $i (@opcodes, @opcodes_cc) {
 152         print A "    instrux_${i},\n";
 153     }
 154     print A "};\n";
 155
 156     close A;
 157 }
 158
 159 if ( !defined($output) || $output eq 'd' ) {
 160     print STDERR "Writing insnsd.c...\n";
 161
 162     open D, ">insnsd.c";
 163
 164     print D "/* This file auto-generated from insns.dat by insns.pl" .
 165         " - don't edit it */\n\n";
 166
 167     print D "#include \"nasm.h\"\n";
 168     print D "#include \"insns.h\"\n\n";
 169
 170     print D "static const struct itemplate instrux[] = {\n";
 171     $n = 0;
 172     foreach $j (@big) {
 173         printf D "    /* %4d */ %s\n", $n++, codesubst($j);
 174     }
 175     print D "};\n";
 176
 177     foreach $h (sort(keys(%dinstables))) {
 178         print D "\nstatic const struct itemplate * const itable_${h}[] = {\n";
 179         foreach $j (@{$dinstables{$h}}) {
 180             print D "    instrux + $j,\n";
 181         }
 182         print D "};\n";
 183     }
 184
 185     foreach $h (@disasm_prefixes, '') {
 186         $is_prefix{$h} = 1;
 187         print D "\n";
 188         print D "static " unless ($h eq '');
 189         print D "const struct disasm_index ";
 190         print D ($h eq '') ? 'itable' : "itable_$h";
 191         print D "[256] = {\n";
 192         for ($c = 0; $c < 256; $c++) {
 193             $nn = sprintf("%s%02X", $h, $c);
 194             if ($is_prefix{$nn}) {
 195                 die "$0: ambiguous decoding of $nn\n"
 196                     if (defined($dinstables{$nn}));
 197                 printf D "    { itable_%s, -1 },\n", $nn;
 198             } elsif (defined($dinstables{$nn})) {
 199                 printf D "    { itable_%s, %u },\n",
 200                 $nn, scalar(@{$dinstables{$nn}});
 201             } else {
 202                 printf D "    { NULL, 0 },\n";
 203             }
 204         }
 205     print D "};\n";
 206     }
 207
 208     close D;
 209 }
 210
 211 if ( !defined($output) || $output eq 'i' ) {
 212     print STDERR "Writing insnsi.h...\n";
 213
 214     open I, ">insnsi.h";
 215
 216     print I "/* This file is auto-generated from insns.dat by insns.pl" .
 217         " - don't edit it */\n\n";
 218     print I "/* This file in included by nasm.h */\n\n";
 219
 220     print I "/* Instruction names */\n\n";
 221     print I "#ifndef NASM_INSNSI_H\n";
 222     print I "#define NASM_INSNSI_H 1\n\n";
 223     print I "enum opcode {\n";
 224     $maxlen = 0;
 225     foreach $i (@opcodes, @opcodes_cc) {
 226         print I "\tI_${i},\n";
 227         $len = length($i);
 228         $len++ if ( $i =~ /cc$/ );      # Condition codes can be 3 characters long
 229         $maxlen = $len if ( $len > $maxlen );
 230     }
 231     print I "\tI_none = -1\n";
 232     print I "\n};\n\n";
 233     print I "#define MAX_INSLEN ", $maxlen, "\n";
 234     print I "#define FIRST_COND_OPCODE I_", $opcodes_cc[0], "\n\n";
 235     print I "#endif /* NASM_INSNSI_H */\n";
 236
 237     close I;
 238 }
 239
 240 if ( !defined($output) || $output eq 'n' ) {
 241     print STDERR "Writing insnsn.c...\n";
 242
 243     open N, ">insnsn.c";
 244
 245     print N "/* This file is auto-generated from insns.dat by insns.pl" .
 246         " - don't edit it */\n\n";
 247     print N "#include \"tables.h\"\n\n";
 248
 249     print N "const char * const nasm_insn_names[] = {";
 250     $first = 1;
 251     foreach $i (@opcodes) {
 252         print N "," if ( !$first );
 253         $first = 0;
 254         $ilower = $i;
 255         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 256         print N "\n\t\"${ilower}\"";
 257     }
 258     print N "\n};\n\n";
 259     print N "/* Conditional instructions */\n";
 260     print N "const char * const nasm_cond_insn_names[] = {";
 261     $first = 1;
 262     foreach $i (@opcodes_cc) {
 263         print N "," if ( !$first );
 264         $first = 0;
 265         $ilower = $i;
 266         $ilower =~ s/cc$//;             # Skip cc suffix
 267         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 268         print N "\n\t\"${ilower}\"";
 269     }
 270
 271     print N "\n};\n\n";
 272     print N "/* and the corresponding opcodes */\n";
 273     print N "const enum opcode nasm_cond_insn_opcodes[] = {";
 274     $first = 1;
 275     foreach $i (@opcodes_cc) {
 276         print N "," if ( !$first );
 277         $first = 0;
 278         print N "\n\tI_$i";
 279     }
 280
 281     print N "\n};\n";
 282     close N;
 283 }
 284
 285 printf STDERR "Done: %d instructions\n", $insns;
 286
 287 sub format_insn(@) {
 288     my ($opcode, $operands, $codes, $flags) = @_;
 289     my $num, $nd = 0;
 290     my @bytecode;
 291
 292     return (undef, undef) if $operands eq "ignore";
 293
 294     # format the operands
 295     $operands =~ s/:/|colon,/g;
 296     $operands =~ s/mem(\d+)/mem|bits$1/g;
 297     $operands =~ s/mem/memory/g;
 298     $operands =~ s/memory_offs/mem_offs/g;
 299     $operands =~ s/imm(\d+)/imm|bits$1/g;
 300     $operands =~ s/imm/immediate/g;
 301     $operands =~ s/rm(\d+)/rm_gpr|bits$1/g;
 302     $operands =~ s/(mmx|xmm|ymm)rm/rm_$1/g;
 303     $operands =~ s/\=([0-9]+)/same_as|$1/g;
 304     if ($operands eq 'void') {
 305         @ops = ();
 306     } else {
 307         @ops = split(/\,/, $operands);
 308     }
 309     $num = scalar(@ops);
 310     while (scalar(@ops) < $MAX_OPERANDS) {
 311         push(@ops, '0');
 312     }
 313     $operands = join(',', @ops);
 314     $operands =~ tr/a-z/A-Z/;
 315
 316     # format the flags
 317     $flags =~ s/,/|IF_/g;
 318     $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 319     $flags = "IF_" . $flags;
 320
 321     @bytecode = (decodify($codes), 0);
 322     push(@bytecode_list, [@bytecode]);
 323     $codes = hexstr(@bytecode);
 324
 325     ("{I_$opcode, $num, {$operands}, \@\@CODES-$codes\@\@, $flags},", $nd);
 326 }
 327
 328 #
 329 # Look for @@CODES-xxx@@ sequences and replace them with the appropriate
 330 # offset into nasm_bytecodes
 331 #
 332 sub codesubst($) {
 333     my($s) = @_;
 334     my $n;
 335
 336     while ($s =~ /\@\@CODES-([0-9A-F]+)\@\@/) {
 337         my $pos = $bytecode_pos{$1};
 338         if (!defined($pos)) {
 339             die "$0: no position assigned to byte code $1\n";
 340         }
 341         $s = $` . "nasm_bytecodes+${pos}" . "$'";
 342     }
 343     return $s;
 344 }
 345
 346 sub addprefix ($@) {
 347     my ($prefix, @list) = @_;
 348     my $x;
 349     my @l = ();
 350
 351     foreach $x (@list) {
 352         push(@l, sprintf("%s%02X", $prefix, $x));
 353     }
 354
 355     return @l;
 356 }
 357
 358 #
 359 # Turn a code string into a sequence of bytes
 360 #
 361 sub decodify($) {
 362   # Although these are C-syntax strings, by convention they should have
 363   # only octal escapes (for directives) and hexadecimal escapes
 364   # (for verbatim bytes)
 365     my($codestr) = @_;
 366
 367     if ($codestr =~ /^\s*\[([^\]]*)\]\s*$/) {
 368         return byte_code_compile($1);
 369     }
 370
 371     my $c = $codestr;
 372     my @codes = ();
 373
 374     while ($c ne '') {
 375         if ($c =~ /^\\x([0-9a-f]+)(.*)$/i) {
 376             push(@codes, hex $1);
 377             $c = $2;
 378             next;
 379         } elsif ($c =~ /^\\([0-7]{1,3})(.*)$/) {
 380             push(@codes, oct $1);
 381             $c = $2;
 382             next;
 383         } else {
 384             die "$0: unknown code format in \"$codestr\"\n";
 385         }
 386     }
 387
 388     return @codes;
 389 }
 390
 391 # Turn a numeric list into a hex string
 392 sub hexstr(@) {
 393     my $s = '';
 394     my $c;
 395
 396     foreach $c (@_) {
 397         $s .= sprintf("%02X", $c);
 398     }
 399     return $s;
 400 }
 401
 402 # Here we determine the range of possible starting bytes for a given
 403 # instruction. We need only consider the codes:
 404 # \1 \2 \3     mean literal bytes, of course
 405 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 406 # \1[0123]     mean byte plus register value
 407 # \330         means byte plus condition code
 408 # \0 or \340   mean give up and return empty set
 409 # \17[234]     skip is4 control byte
 410 # \26x \270    skip VEX control bytes
 411 sub startseq($) {
 412   my ($codestr) = @_;
 413   my $word, @range;
 414   my @codes = ();
 415   my $c = $codestr;
 416   my $c0, $c1, $i;
 417   my $prefix = '';
 418
 419   @codes = decodify($codestr);
 420
 421   while ($c0 = shift(@codes)) {
 422       $c1 = $codes[0];
 423       if ($c0 == 01 || $c0 == 02 || $c0 == 03) {
 424           # Fixed byte string
 425           my $fbs = $prefix;
 426           while (1) {
 427               if ($c0 == 01 || $c0 == 02 || $c0 == 03) {
 428                   while ($c0--) {
 429                       $fbs .= sprintf("%02X", shift(@codes));
 430                   }
 431               } else {
 432                   last;
 433               }
 434               $c0 = shift(@codes);
 435           }
 436
 437           foreach $pfx (@disasm_prefixes) {
 438               if (substr($fbs, 0, length($pfx)) eq $pfx) {
 439                   $prefix = $pfx;
 440                   $fbs = substr($fbs, length($pfx));
 441                   last;
 442               }
 443           }
 444
 445           if ($fbs ne '') {
 446               return ($prefix.substr($fbs,0,2));
 447           }
 448
 449           unshift(@codes, $c0);
 450       } elsif ($c0 == 04) {
 451           return addprefix($prefix, 0x07, 0x17, 0x1F);
 452       } elsif ($c0 == 05) {
 453           return addprefix($prefix, 0xA1, 0xA9);
 454       } elsif ($c0 == 06) {
 455           return addprefix($prefix, 0x06, 0x0E, 0x16, 0x1E);
 456       } elsif ($c0 == 07) {
 457           return addprefix($prefix, 0xA0, 0xA8);
 458       } elsif ($c0 >= 010 && $c0 <= 013) {
 459           return addprefix($prefix, $c1..($c1+7));
 460       } elsif (($c0 & ~013) == 0144) {
 461           return addprefix($prefix, $c1, $c1|2);
 462       } elsif ($c0 == 0330) {
 463           return addprefix($prefix, $c1..($c1+15));
 464       } elsif ($c0 == 0 || $c0 == 0340) {
 465           return $prefix;
 466       } elsif (($c0 & ~3) == 0260 || $c0 == 0270) {
 467           shift(@codes);        # Skip VEX control bytes
 468           shift(@codes);
 469       } elsif ($c0 >= 0172 && $c0 <= 174) {
 470           shift(@codes);        # Skip is4 control byte
 471       } else {
 472           # We really need to be able to distinguish "forbidden"
 473           # and "ignorable" codes here
 474       }
 475   }
 476   return $prefix;
 477 }
 478
 479 #
 480 # This function takes a series of byte codes in a format which is more
 481 # typical of the Intel documentation, and encode it.
 482 #
 483 # The format looks like:
 484 #
 485 # [operands: opcodes]
 486 #
 487 # The operands word lists the order of the operands:
 488 #
 489 # r = register field in the modr/m
 490 # m = modr/m
 491 # v = VEX "v" field
 492 # d = DREX "dst" field
 493 # i = immediate
 494 # s = register field of is4/imz2 field
 495 # - = implicit (unencoded) operand
 496 #
 497 # For an operand that should be filled into more than one field,
 498 # enter it as e.g. "r+v".
 499 #
 500 sub byte_code_compile($) {
 501     my($str) = @_;
 502     my $opr;
 503     my $opc;
 504     my @codes = ();
 505     my $litix = undef;
 506     my %oppos = ();
 507     my $i;
 508     my $op, $oq;
 509
 510     if ($str =~ /^(\S*)\:\s*(.*\S)\s*$/) {
 511         $opr = "\L$1";
 512         $opc = "\L$2";
 513     } else {
 514         $opr = '';
 515         $opc = "\L$str";
 516     }
 517
 518     my $op = 0;
 519     for ($i = 0; $i < length($opr); $i++) {
 520         my $c = substr($opr,$i,1);
 521         if ($c eq '+') {
 522             $op--;
 523         } else {
 524             $oppos{$c} = $op++;
 525         }
 526     }
 527
 528     $prefix_ok = 1;
 529     foreach $op (split(/\s*(?:\s|(?=[\/\\]))/, $opc)) {
 530         if ($op eq 'o16') {
 531             push(@codes, 0320);
 532         } elsif ($op eq 'o32') {
 533             push(@codes, 0321);
 534         } elsif ($op eq 'o64') {  # 64-bit operand size requiring REX.W
 535             push(@codes, 0324);
 536         } elsif ($op eq 'o64i') { # Implied 64-bit operand size (no REX.W)
 537             push(@codes, 0323);
 538         } elsif ($op eq 'a16') {
 539             push(@codes, 0310);
 540         } elsif ($op eq 'a32') {
 541             push(@codes, 0311);
 542         } elsif ($op eq 'a64') {
 543             push(@codes, 0313);
 544         } elsif ($op eq '!osp') {
 545             push(@codes, 0364);
 546         } elsif ($op eq '!asp') {
 547             push(@codes, 0365);
 548         } elsif ($op eq 'rex.l') {
 549             push(@codes, 0334);
 550         } elsif ($op eq 'repe') {
 551             push(@codes, 0335);
 552         } elsif ($prefix_ok && $op =~ /^(66|f2|f3|np)$/) {
 553             # 66/F2/F3 prefix used as an opcode extension, or np = no prefix
 554             if ($op eq '66') {
 555                 push(@codes, 0361);
 556             } elsif ($op eq 'f2') {
 557                 push(@codes, 0362);
 558             } elsif ($op eq 'f3') {
 559                 push(@codes, 0363);
 560             } else {
 561                 push(@codes, 0360);
 562             }
 563         } elsif ($op =~ /^[0-9a-f]{2}$/) {
 564             if (defined($litix) && $litix+$codes[$litix]+1 == scalar @codes) {
 565                 $codes[$litix]++;
 566                 push(@codes, hex $op);
 567             } else {
 568                 $litix = scalar(@codes);
 569                 push(@codes, 01, hex $op);
 570             }
 571             $prefix_ok = 0;
 572         } elsif ($op eq '/r') {
 573             if (!defined($oppos{'r'}) || !defined($oppos{'m'})) {
 574                 die "$0: $line: $op requires r and m operands\n";
 575             }
 576             push(@codes, 0100 + ($oppos{'m'} << 3) + $oppos{'r'});
 577             $prefix_ok = 0;
 578         } elsif ($op =~ m:^/([0-7])$:) {
 579             if (!defined($oppos{'m'})) {
 580                 die "$0: $line: $op requires m operand\n";
 581             }
 582             push(@codes, 0200 + ($oppos{'m'} << 3) + $1);
 583             $prefix_ok = 0;
 584         } elsif ($op =~ /^vex(|\..*)$/) {
 585             my ($m,$w,$l,$p) = (undef,2,undef,0);
 586             foreach $oq (split(/\./, $op)) {
 587                 if ($oq eq 'vex') {
 588                     # prefix
 589                 } elsif ($oq eq '128' || $oq eq 'l0') {
 590                     $l = 0;
 591                 } elsif ($oq eq '256' || $oq eq 'l1') {
 592                     $l = 1;
 593                 } elsif ($oq eq 'w0') {
 594                     $w = 0;
 595                 } elsif ($oq eq 'w1') {
 596                     $w = 1;
 597                 } elsif ($oq eq '66') {
 598                     $p = 1;
 599                 } elsif ($oq eq 'f3') {
 600                     $p = 2;
 601                 } elsif ($oq eq 'f2') {
 602                     $p = 3;
 603                 } elsif ($oq eq '0f') {
 604                     $m = 1;
 605                 } elsif ($oq eq '0f38') {
 606                     $m = 2;
 607                 } elsif ($oq eq '0f3a') {
 608                     $m = 3;
 609                 } elsif ($oq =~ /^m([0-9]+)$/) {
 610                     $m = $1+0;
 611                 } elsif ($oq eq 'nds' || $oq eq 'ndd') {
 612                     if (!defined($oppos{'v'})) {
 613                         die "$0: $line: vex.$oq without 'v' operand\n";
 614                     }
 615                 } else {
 616                     die "$0: $line: undefined VEX subcode: $oq\n";
 617                 }
 618             }
 619             if (!defined($m) || !defined($w) || !defined($l) || !defined($p)) {
 620                 die "$0: $line: missing fields in VEX specification\n";
 621             }
 622             push(@codes, defined($oppos{'v'}) ? 0260+$oppos{'v'} : 0270,
 623                  $m, ($w << 3)+($l << 2)+$p);
 624             $prefix_ok = 0;
 625         } elsif ($op =~ /^drex(|..*)$/) {
 626             my ($oc0) = (0);
 627             foreach $oq (split(/\./, $op)) {
 628                 if ($oq eq 'drex') {
 629                     #prefix
 630                 } elsif ($oq eq 'oc0') {
 631                     $oc0 = 1;
 632                 } else {
 633                     die "$0: $line: undefined DREX subcode: $oq\n";
 634                 }
 635             }
 636             if (!defined($oppos{'d'})) {
 637                 die "$0: $line: DREX without a 'd' operand\n";
 638             }
 639             push(@codes, 0160+$oppos{'d'}+($oc0 ? 4 : 0));
 640         } elsif ($op =~ /^(ib\,s|ib|ib\,w|iw|iwd|id|iwdq|rel|rel8|rel16|rel32|iq|seg|ibw|ibd|ibd,s)$/) {
 641             if (!defined($oppos{'i'})) {
 642                 die "$0: $op without 'i' operand\n";
 643             }
 644             if ($op eq 'ib,s') { # Signed imm8
 645                 push(@codes, 014+$oppos{'i'});
 646             } elsif ($op eq 'ib') { # imm8
 647                 push(@codes, 020+$oppos{'i'});
 648             } elsif ($op eq 'ib,u') { # Unsigned imm8
 649                 push(@codes, 024+$oppos{'i'});
 650             } elsif ($op eq 'iw') { # imm16
 651                 push(@codes, 030+$oppos{'i'});
 652             } elsif ($op eq 'iwd') { # imm16 or imm32, depending on opsize
 653                 push(@codes, 034+$oppos{'i'});
 654             } elsif ($op eq 'id') { # imm32
 655                 push(@codes, 040+$oppos{'i'});
 656             } elsif ($op eq 'iwdq') { # imm16/32/64, depending on opsize
 657                 push(@codes, 044+$oppos{'i'});
 658             } elsif ($op eq 'rel8') {
 659                 push(@codes, 050+$oppos{'i'});
 660             } elsif ($op eq 'iq') {
 661                 push(@codes, 054+$oppos{'i'});
 662             } elsif ($op eq 'rel16') {
 663                 push(@codes, 060+$oppos{'i'});
 664             } elsif ($op eq 'rel') { # 16 or 32 bit relative operand
 665                 push(@codes, 064+$oppos{'i'});
 666             } elsif ($op eq 'rel32') {
 667                 push(@codes, 070+$oppos{'i'});
 668             } elsif ($op eq 'seg') {
 669                 push(@codes, 074+$oppos{'i'});
 670             } elsif ($op eq 'ibw') { # imm16 that can be bytified
 671                 if (!defined($s_pos)) {
 672                     die "$0: $line: $op without a +s byte\n";
 673                 }
 674                 $codes[$s_pos] += 0144;
 675                 push(@codes, 0140+$oppos{'i'});
 676             } elsif ($op eq 'ibd') { # imm32 that can be bytified
 677                 if (!defined($s_pos)) {
 678                     die "$0: $line: $op without a +s byte\n";
 679                 }
 680                 $codes[$s_pos] += 0154;
 681                 push(@codes, 0150+$oppos{'i'});
 682             } elsif ($op eq 'ibd,s') {
 683                 # imm32 that can be bytified, sign extended to 64 bits
 684                 if (!defined($s_pos)) {
 685                     die "$0: $line: $op without a +s byte\n";
 686                 }
 687                 $codes[$s_pos] += 0154;
 688                 push(@codes, 0250+$oppos{'i'});
 689             }
 690             $prefix_ok = 0;
 691         } elsif ($op eq '/is4') {
 692             if (!defined($oppos{'s'})) {
 693                 die "$0: $line: $op without 's' operand\n";
 694             }
 695             if (defined($oppos{'i'})) {
 696                 push(@codes, 0172, ($oppos{'s'} << 3)+$oppos{'i'});
 697             } else {
 698                 push(@codes, 0174, $oppos{'s'});
 699             }
 700             $prefix_ok = 0;
 701         } elsif ($op =~ /^\/is4\=([0-9]+)$/) {
 702             my $imm = $1;
 703             if (!defined($oppos{'s'})) {
 704                 die "$0: $line: $op without 's' operand\n";
 705             }
 706             if ($imm < 0 || $imm > 15) {
 707                 die "$0: $line: invalid imm4 value for $op: $imm\n";
 708             }
 709             push(@codes, 0173, ($oppos{'s'} << 4) + $imm);
 710             $prefix_ok = 0;
 711         } elsif ($op =~ /^([0-9a-f]{2})\+s$/) {
 712             if (!defined($oppos{'i'})) {
 713                 die "$0: $op without 'i' operand\n";
 714             }
 715             $s_pos = scalar @codes;
 716             push(@codes, $oppos{'i'}, hex $1);
 717             $prefix_ok = 0;
 718         } elsif ($op =~ /^([0-9a-f]{2})\+c$/) {
 719             push(@codes, 0330, hex $1);
 720             $prefix_ok = 0;
 721         } elsif ($op =~ /^\\([0-7]+|x[0-9a-f]{2})$/) {
 722             # Escape to enter literal bytecodes
 723             push(@codes, oct $1);
 724         } else {
 725             die "$0: unknown operation: $op\n";
 726         }
 727     }
 728
 729     return @codes;
 730 }