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 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 $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, $nd) = &format(@_);
  25   if ($formatted) {
  26     $insns++;
  27     $aname = "aa_$_[0]";
  28     push @$aname, $formatted;
  29   }
  30   if ( $_[0] =~ /cc$/ ) {
  31       $k_opcodes_cc{$_[0]}++;
  32   } else {
  33       $k_opcodes{$_[0]}++;
  34   }
  35   $done{$_[0]} = 1 if !$done{$_[0]};
  36   if ($formatted && !$nd) {
  37     push @big, $formatted;
  38     foreach $i (&startbyte($_[2])) {
  39       $aname = sprintf "dd_%02X",$i;
  40       push @$aname, $#big;
  41     }
  42   }
  43 }
  44
  45 close F;
  46
  47 @opcodes    = sort keys(%k_opcodes);    # Unconditional instructions
  48 @opcodes_cc = sort keys(%k_opcodes_cc); # Conditional instructions
  49
  50 print STDERR "Writing insnsa.c...\n";
  51
  52 open A, ">insnsa.c";
  53
  54 print A "/* This file auto-generated from insns.dat by insns.pl" .
  55         " - don't edit it */\n\n";
  56 print A "#include <stdio.h>\n";
  57 print A "#include \"nasm.h\"\n";
  58 print A "#include \"insns.h\"\n";
  59 print A "\n";
  60
  61 foreach $i (@opcodes, @opcodes_cc) {
  62   print A "static struct itemplate instrux_${i}[] = {\n";
  63   $aname = "aa_$i";
  64   foreach $j (@$aname) {
  65     print A "    $j\n";
  66   }
  67   print A "    {-1}\n};\n\n";
  68 }
  69 print A "struct itemplate *nasm_instructions[] = {\n";
  70 foreach $i (@opcodes, @opcodes_cc) {
  71   print A "    instrux_${i},\n";
  72 }
  73 print A "};\n";
  74
  75 close A;
  76
  77 print STDERR "Writing insnsd.c...\n";
  78
  79 open D, ">insnsd.c";
  80
  81 print D "/* This file auto-generated from insns.dat by insns.pl" .
  82         " - don't edit it */\n\n";
  83 print D "#include <stdio.h>\n";
  84 print D "#include \"nasm.h\"\n";
  85 print D "#include \"insns.h\"\n";
  86 print D "\n";
  87
  88 print D "static struct itemplate instrux[] = {\n";
  89 foreach $j (@big) {
  90   print D "    $j\n";
  91 }
  92 print D "    {-1}\n};\n\n";
  93
  94 for ($c=0; $c<256; $c++) {
  95   $h = sprintf "%02X", $c;
  96   print D "static struct itemplate *itable_${h}[] = {\n";
  97   $aname = "dd_$h";
  98   foreach $j (@$aname) {
  99     print D "    instrux + $j,\n";
 100   }
 101   print D "    NULL\n};\n\n";
 102 }
 103
 104 print D "struct itemplate **itable[] = {\n";
 105 for ($c=0; $c<256; $c++) {
 106   printf D "    itable_%02X,\n", $c;
 107 }
 108 print D "};\n";
 109
 110 close D;
 111
 112 print STDERR "Writing insnsi.h...\n";
 113
 114 open I, ">insnsi.h";
 115
 116 print I "/* This file is auto-generated from insns.dat by insns.pl" .
 117         " - don't edit it */\n\n";
 118 print I "/* This file in included by nasm.h */\n\n";
 119
 120 print I "/* Instruction names */\n";
 121 print I "enum {";
 122 $first  = 1;
 123 $maxlen = 0;
 124 foreach $i (@opcodes, @opcodes_cc) {
 125     print I "," if ( !$first );
 126     $first = 0;
 127     print I "\n\tI_${i}";
 128     $len = length($i);
 129     $len++ if ( $i =~ /cc$/ );  # Condition codes can be 3 characters long
 130     $maxlen = $len if ( $len > $maxlen );
 131 }
 132 print I "\n};\n\n";
 133 print I "#define MAX_INSLEN ", $maxlen, "\n";
 134
 135 close I;
 136
 137 print STDERR "Writing insnsn.c...\n";
 138
 139 open N, ">insnsn.c";
 140
 141 print N "/* This file is auto-generated from insns.dat by insns.pl" .
 142         " - don't edit it */\n\n";
 143 print N "/* This file in included by names.c */\n\n";
 144
 145 print N "static char *insn_names[] = {";
 146 $first = 1;
 147 foreach $i (@opcodes) {
 148     print N "," if ( !$first );
 149     $first = 0;
 150     $ilower = $i;
 151     $ilower =~ tr/A-Z/a-z/;     # Change to lower case (Perl 4 compatible)
 152     print N "\n\t\"${ilower}\"";
 153 }
 154 print N "\n};\n\n";
 155 print N "/* Conditional instructions */\n";
 156 print N "static char *icn[] = {";
 157 $first = 1;
 158 foreach $i (@opcodes_cc) {
 159     print N "," if ( !$first );
 160     $first = 0;
 161     $ilower = $i;
 162     $ilower =~ s/cc$//;         # Skip cc suffix
 163     $ilower =~ tr/A-Z/a-z/;     # Change to lower case (Perl 4 compatible)
 164     print N "\n\t\"${ilower}\"";
 165 }
 166
 167 print N "\n};\n\n";
 168 print N "/* and the corresponding opcodes */\n";
 169 print N "static int ico[] = {";
 170 $first = 1;
 171 foreach $i (@opcodes_cc) {
 172     print N "," if ( !$first );
 173     $first = 0;
 174     print N "\n\tI_$i";
 175 }
 176
 177 print N "\n};\n";
 178
 179 close N;
 180
 181 printf STDERR "Done: %d instructions\n", $insns;
 182
 183 sub format {
 184   local ($opcode, $operands, $codes, $flags) = @_;
 185   local $num, $nd = 0;
 186
 187   return (undef, undef) if $operands eq "ignore";
 188
 189   # format the operands
 190   $operands =~ s/:/|colon,/g;
 191   $operands =~ s/mem(\d+)/mem|bits$1/g;
 192   $operands =~ s/mem/memory/g;
 193   $operands =~ s/memory_offs/mem_offs/g;
 194   $operands =~ s/imm(\d+)/imm|bits$1/g;
 195   $operands =~ s/imm/immediate/g;
 196   $operands =~ s/rm(\d+)/regmem|bits$1/g;
 197   $num = 3;
 198   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 199   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 200   $operands =~ tr/a-z/A-Z/;
 201
 202   # format the flags
 203   $flags =~ s/,/|IF_/g;
 204   $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 205   $flags = "IF_" . $flags;
 206
 207   ("{I_$opcode, $num, {$operands}, \"$codes\", $flags},", $nd);
 208 }
 209
 210 # Here we determine the range of possible starting bytes for a given
 211 # instruction. We need only consider the codes:
 212 # \1 \2 \3     mean literal bytes, of course
 213 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 214 # \10 \11 \12  mean byte plus register value
 215 # \17          means byte zero
 216 # \330         means byte plus condition code
 217 # \0 or \340   mean give up and return empty set
 218 sub startbyte {
 219   local ($codes) = @_;
 220   local $word, @range;
 221
 222   while (1) {
 223     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 224     $word = $1, $codes = $2;
 225     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 226     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 227     return (0xA1, 0xA9) if $word eq "\\5";
 228     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 229     return (0xA0, 0xA8) if $word eq "\\7";
 230     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 231     return (0) if $word eq "\\17";
 232     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 233     return () if $word eq "\\0" || $word eq "\\340";
 234   }
 235   @range = ();
 236   push @range, $start++ while ($r-- > 0);
 237   @range;
 238 }