insns.pl

   1 #!/usr/bin/perl
   2 #
   3 # insns.pl   produce insnsa.c and insnsd.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 licence given in the file "Licence"
   8 # distributed in the NASM archive.
   9
  10 print STDERR "Reading insns.dat...\n";
  11
  12 open (F, "insns.dat") || die "unable to open insns.dat";
  13
  14 $line = 0;
  15 $opcodes = 0;
  16 $insns = 0;
  17 while (<F>) {
  18   $line++;
  19   next if /^\s*;/;   # comments
  20   chomp;
  21   split;
  22   next if $#_ == -1; # blank lines
  23   (warn "line $line does not contain four fields\n"), next if $#_ != 3;
  24   $formatted = &format(@_);
  25   if ($formatted) {
  26     $insns++;
  27     $aname = "aa_$_[0]";
  28     push @$aname, $formatted;
  29   }
  30   $opcodes[$opcodes++] = $_[0], $done{$_[0]} = 1 if !$done{$_[0]};
  31   if ($formatted && $formatted !~ /IF_ND/) {
  32     push @big, $formatted;
  33     foreach $i (&startbyte($_[2])) {
  34       $aname = sprintf "dd_%02X",$i;
  35       push @$aname, $#big;
  36     }
  37   }
  38 }
  39
  40 close F;
  41
  42 print STDERR "Writing insnsa.c...\n";
  43
  44 open A, ">insnsa.c";
  45
  46 print A "/* This file auto-generated from insns.dat by insns.pl" .
  47         " - don't edit it */\n\n";
  48 print A "#include <stdio.h>\n";
  49 print A "#include \"nasm.h\"\n";
  50 print A "#include \"insns.h\"\n";
  51 print A "\n";
  52
  53 foreach $i (@opcodes) {
  54   print A "static struct itemplate instrux_${i}[] = {\n";
  55   $aname = "aa_$i";
  56   foreach $j (@$aname) {
  57     print A "    $j\n";
  58   }
  59   print A "    {-1}\n};\n\n";
  60 }
  61 print A "struct itemplate *nasm_instructions[] = {\n";
  62 foreach $i (@opcodes) {
  63   print A "    instrux_${i},\n";
  64 }
  65 print A "};\n";
  66
  67 close A;
  68
  69 print STDERR "Writing insnsd.c...\n";
  70
  71 open D, ">insnsd.c";
  72
  73 print D "/* This file auto-generated from insns.dat by insns.pl" .
  74         " - don't edit it */\n\n";
  75 print D "#include <stdio.h>\n";
  76 print D "#include \"nasm.h\"\n";
  77 print D "#include \"insns.h\"\n";
  78 print D "\n";
  79
  80 print D "static struct itemplate instrux[] = {\n";
  81 foreach $j (@big) {
  82   print D "    $j\n";
  83 }
  84 print D "    {-1}\n};\n\n";
  85
  86 for ($c=0; $c<256; $c++) {
  87   $h = sprintf "%02X", $c;
  88   print D "static struct itemplate *itable_${h}[] = {\n";
  89   $aname = "dd_$h";
  90   foreach $j (@$aname) {
  91     print D "    instrux + $j,\n";
  92   }
  93   print D "    NULL\n};\n\n";
  94 }
  95
  96 print D "struct itemplate **itable[] = {\n";
  97 for ($c=0; $c<256; $c++) {
  98   printf D "    itable_%02X,\n", $c;
  99 }
 100 print D "};\n";
 101
 102 close D;
 103
 104 printf STDERR "Done: %d instructions\n", $insns;
 105
 106 sub format {
 107   local ($opcode, $operands, $codes, $flags) = @_;
 108   local $num;
 109
 110   return undef if $operands eq "ignore";
 111
 112   # format the operands
 113   $operands =~ s/:/|colon,/g;
 114   $operands =~ s/mem(\d+)/mem|bits$1/g;
 115   $operands =~ s/mem/memory/g;
 116   $operands =~ s/memory_offs/mem_offs/g;
 117   $operands =~ s/imm(\d+)/imm|bits$1/g;
 118   $operands =~ s/imm/immediate/g;
 119   $operands =~ s/rm(\d+)/regmem|bits$1/g;
 120   $num = 3;
 121   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 122   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 123   $operands =~ tr/a-z/A-Z/;
 124
 125   # format the flags
 126   $flags =~ s/,/|IF_/g;
 127   $flags = "IF_" . $flags;
 128
 129   "{I_$opcode, $num, {$operands}, \"$codes\", $flags},";
 130 }
 131
 132 # Here we determine the range of possible starting bytes for a given
 133 # instruction. We need only consider the codes:
 134 # \1 \2 \3     mean literal bytes, of course
 135 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 136 # \10 \11 \12  mean byte plus register value
 137 # \17          means byte zero
 138 # \330         means byte plus condition code
 139 # \0 or \340   mean give up and return empty set
 140 sub startbyte {  # FIXME we cheat, for now :-)
 141   local ($codes) = @_;
 142   local $word, @range;
 143
 144   while (1) {
 145     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 146     $word = $1, $codes = $2;
 147     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 148     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 149     return (0xA1, 0xA9) if $word eq "\\5";
 150     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 151     return (0xA0, 0xA8) if $word eq "\\7";
 152     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 153     return (0) if $word eq "\\17";
 154     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 155     return () if $word eq "\\0" || $word eq "\\340";
 156   }
 157   @range = ();
 158   push @range, $start++ while ($r-- > 0);
 159   @range;
 160 }