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 $fname = "insns.dat" unless $fname = $ARGV[0];
  13 open (F, $fname) || die "unable to open $fname";
  14
  15 $line = 0;
  16 $opcodes = 0;
  17 $insns = 0;
  18 while (<F>) {
  19   $line++;
  20   next if /^\s*;/;   # comments
  21   chomp;
  22   split;
  23   next if $#_ == -1; # blank lines
  24   (warn "line $line does not contain four fields\n"), next if $#_ != 3;
  25   ($formatted, $nd) = &format(@_);
  26   if ($formatted) {
  27     $insns++;
  28     $aname = "aa_$_[0]";
  29     push @$aname, $formatted;
  30   }
  31   $opcodes[$opcodes++] = $_[0], $done{$_[0]} = 1 if !$done{$_[0]};
  32   if ($formatted && !$nd) {
  33     push @big, $formatted;
  34     foreach $i (&startbyte($_[2])) {
  35       $aname = sprintf "dd_%02X",$i;
  36       push @$aname, $#big;
  37     }
  38   }
  39 }
  40
  41 close F;
  42
  43 print STDERR "Writing insnsa.c...\n";
  44
  45 open A, ">insnsa.c";
  46
  47 print A "/* This file auto-generated from insns.dat by insns.pl" .
  48         " - don't edit it */\n\n";
  49 print A "#include <stdio.h>\n";
  50 print A "#include \"nasm.h\"\n";
  51 print A "#include \"insns.h\"\n";
  52 print A "\n";
  53
  54 foreach $i (@opcodes) {
  55   print A "static struct itemplate instrux_${i}[] = {\n";
  56   $aname = "aa_$i";
  57   foreach $j (@$aname) {
  58     print A "    $j\n";
  59   }
  60   print A "    {-1}\n};\n\n";
  61 }
  62 print A "struct itemplate *nasm_instructions[] = {\n";
  63 foreach $i (@opcodes) {
  64   print A "    instrux_${i},\n";
  65 }
  66 print A "};\n";
  67
  68 close A;
  69
  70 print STDERR "Writing insnsd.c...\n";
  71
  72 open D, ">insnsd.c";
  73
  74 print D "/* This file auto-generated from insns.dat by insns.pl" .
  75         " - don't edit it */\n\n";
  76 print D "#include <stdio.h>\n";
  77 print D "#include \"nasm.h\"\n";
  78 print D "#include \"insns.h\"\n";
  79 print D "\n";
  80
  81 print D "static struct itemplate instrux[] = {\n";
  82 foreach $j (@big) {
  83   print D "    $j\n";
  84 }
  85 print D "    {-1}\n};\n\n";
  86
  87 for ($c=0; $c<256; $c++) {
  88   $h = sprintf "%02X", $c;
  89   print D "static struct itemplate *itable_${h}[] = {\n";
  90   $aname = "dd_$h";
  91   foreach $j (@$aname) {
  92     print D "    instrux + $j,\n";
  93   }
  94   print D "    NULL\n};\n\n";
  95 }
  96
  97 print D "struct itemplate **itable[] = {\n";
  98 for ($c=0; $c<256; $c++) {
  99   printf D "    itable_%02X,\n", $c;
 100 }
 101 print D "};\n";
 102
 103 close D;
 104
 105 printf STDERR "Done: %d instructions\n", $insns;
 106
 107 sub format {
 108   local ($opcode, $operands, $codes, $flags) = @_;
 109   local $num, $nd = 0;
 110
 111   return (undef, undef) if $operands eq "ignore";
 112
 113   # format the operands
 114   $operands =~ s/:/|colon,/g;
 115   $operands =~ s/mem(\d+)/mem|bits$1/g;
 116   $operands =~ s/mem/memory/g;
 117   $operands =~ s/memory_offs/mem_offs/g;
 118   $operands =~ s/imm(\d+)/imm|bits$1/g;
 119   $operands =~ s/imm/immediate/g;
 120   $operands =~ s/rm(\d+)/regmem|bits$1/g;
 121   $num = 3;
 122   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 123   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 124   $operands =~ tr/a-z/A-Z/;
 125
 126   # format the flags
 127   $flags =~ s/,/|IF_/g;
 128   $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 129   $flags = "IF_" . $flags;
 130
 131   ("{I_$opcode, $num, {$operands}, \"$codes\", $flags},", $nd);
 132 }
 133
 134 # Here we determine the range of possible starting bytes for a given
 135 # instruction. We need only consider the codes:
 136 # \1 \2 \3     mean literal bytes, of course
 137 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 138 # \10 \11 \12  mean byte plus register value
 139 # \17          means byte zero
 140 # \330         means byte plus condition code
 141 # \0 or \340   mean give up and return empty set
 142 sub startbyte {
 143   local ($codes) = @_;
 144   local $word, @range;
 145
 146   while (1) {
 147     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 148     $word = $1, $codes = $2;
 149     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 150     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 151     return (0xA1, 0xA9) if $word eq "\\5";
 152     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 153     return (0xA0, 0xA8) if $word eq "\\7";
 154     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 155     return (0) if $word eq "\\17";
 156     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 157     return () if $word eq "\\0" || $word eq "\\340";
 158   }
 159   @range = ();
 160   push @range, $start++ while ($r-- > 0);
 161   @range;
 162 }