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       # Conditional instruction
  32       $k_opcodes_cc{$_[0]}++;
  33   } else {
  34       # Unconditional instruction
  35       $k_opcodes{$_[0]}++;
  36   }
  37   if ($formatted && !$nd) {
  38     push @big, $formatted;
  39     foreach $i (&startbyte($_[2])) {
  40       $aname = sprintf "dd_%02X",$i;
  41       push @$aname, $#big;
  42     }
  43   }
  44 }
  45
  46 close F;
  47
  48 @opcodes    = sort keys(%k_opcodes);
  49 @opcodes_cc = sort keys(%k_opcodes_cc);
  50
  51 print STDERR "Writing insnsa.c...\n";
  52
  53 open A, ">insnsa.c";
  54
  55 print A "/* This file auto-generated from insns.dat by insns.pl" .
  56         " - don't edit it */\n\n";
  57 print A "#include <stdio.h>\n";
  58 print A "#include \"nasm.h\"\n";
  59 print A "#include \"insns.h\"\n";
  60 print A "\n";
  61
  62 foreach $i (@opcodes, @opcodes_cc) {
  63   print A "static struct itemplate instrux_${i}[] = {\n";
  64   $aname = "aa_$i";
  65   foreach $j (@$aname) {
  66     print A "    $j\n";
  67   }
  68   print A "    {-1}\n};\n\n";
  69 }
  70 print A "struct itemplate *nasm_instructions[] = {\n";
  71 foreach $i (@opcodes, @opcodes_cc) {
  72   print A "    instrux_${i},\n";
  73 }
  74 print A "};\n";
  75
  76 close A;
  77
  78 print STDERR "Writing insnsd.c...\n";
  79
  80 open D, ">insnsd.c";
  81
  82 print D "/* This file auto-generated from insns.dat by insns.pl" .
  83         " - don't edit it */\n\n";
  84 print D "#include <stdio.h>\n";
  85 print D "#include \"nasm.h\"\n";
  86 print D "#include \"insns.h\"\n";
  87 print D "\n";
  88
  89 print D "static struct itemplate instrux[] = {\n";
  90 foreach $j (@big) {
  91   print D "    $j\n";
  92 }
  93 print D "    {-1}\n};\n\n";
  94
  95 for ($c=0; $c<256; $c++) {
  96   $h = sprintf "%02X", $c;
  97   print D "static struct itemplate *itable_${h}[] = {\n";
  98   $aname = "dd_$h";
  99   foreach $j (@$aname) {
 100     print D "    instrux + $j,\n";
 101   }
 102   print D "    NULL\n};\n\n";
 103 }
 104
 105 print D "struct itemplate **itable[] = {\n";
 106 for ($c=0; $c<256; $c++) {
 107   printf D "    itable_%02X,\n", $c;
 108 }
 109 print D "};\n";
 110
 111 close D;
 112
 113 print STDERR "Writing insnsi.h...\n";
 114
 115 open I, ">insnsi.h";
 116
 117 print I "/* This file is auto-generated from insns.dat by insns.pl" .
 118         " - don't edit it */\n\n";
 119 print I "/* This file in included by nasm.h */\n\n";
 120
 121 print I "/* Instruction names */\n";
 122 print I "enum {";
 123 $first  = 1;
 124 $maxlen = 0;
 125 foreach $i (@opcodes, @opcodes_cc) {
 126     print I "," if ( !$first );
 127     $first = 0;
 128     print I "\n\tI_${i}";
 129     $len = length($i);
 130     $len++ if ( $i =~ /cc$/ );  # Condition codes can be 3 characters long
 131     $maxlen = $len if ( $len > $maxlen );
 132 }
 133 print I "\n};\n\n";
 134 print I "#define MAX_INSLEN ", $maxlen, "\n";
 135
 136 close I;
 137
 138 print STDERR "Writing insnsn.c...\n";
 139
 140 open N, ">insnsn.c";
 141
 142 print N "/* This file is auto-generated from insns.dat by insns.pl" .
 143         " - don't edit it */\n\n";
 144 print N "/* This file in included by names.c */\n\n";
 145
 146 print N "static char *insn_names[] = {";
 147 $first = 1;
 148 foreach $i (@opcodes) {
 149     print N "," if ( !$first );
 150     $first = 0;
 151     $ilower = $i;
 152     $ilower =~ tr/A-Z/a-z/;     # Change to lower case (Perl 4 compatible)
 153     print N "\n\t\"${ilower}\"";
 154 }
 155 print N "\n};\n\n";
 156 print N "/* Conditional instructions */\n";
 157 print N "static char *icn[] = {";
 158 $first = 1;
 159 foreach $i (@opcodes_cc) {
 160     print N "," if ( !$first );
 161     $first = 0;
 162     $ilower = $i;
 163     $ilower =~ s/cc$//;         # Skip cc suffix
 164     $ilower =~ tr/A-Z/a-z/;     # Change to lower case (Perl 4 compatible)
 165     print N "\n\t\"${ilower}\"";
 166 }
 167
 168 print N "\n};\n\n";
 169 print N "/* and the corresponding opcodes */\n";
 170 print N "static int ico[] = {";
 171 $first = 1;
 172 foreach $i (@opcodes_cc) {
 173     print N "," if ( !$first );
 174     $first = 0;
 175     print N "\n\tI_$i";
 176 }
 177
 178 print N "\n};\n";
 179
 180 close N;
 181
 182 printf STDERR "Done: %d instructions\n", $insns;
 183
 184 sub format {
 185   local ($opcode, $operands, $codes, $flags) = @_;
 186   local $num, $nd = 0;
 187
 188   return (undef, undef) if $operands eq "ignore";
 189
 190   # format the operands
 191   $operands =~ s/:/|colon,/g;
 192   $operands =~ s/mem(\d+)/mem|bits$1/g;
 193   $operands =~ s/mem/memory/g;
 194   $operands =~ s/memory_offs/mem_offs/g;
 195   $operands =~ s/imm(\d+)/imm|bits$1/g;
 196   $operands =~ s/imm/immediate/g;
 197   $operands =~ s/rm(\d+)/regmem|bits$1/g;
 198   $num = 3;
 199   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 200   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 201   $operands =~ tr/a-z/A-Z/;
 202
 203   # format the flags
 204   $flags =~ s/,/|IF_/g;
 205   $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 206   $flags = "IF_" . $flags;
 207
 208   ("{I_$opcode, $num, {$operands}, \"$codes\", $flags},", $nd);
 209 }
 210
 211 # Here we determine the range of possible starting bytes for a given
 212 # instruction. We need only consider the codes:
 213 # \1 \2 \3     mean literal bytes, of course
 214 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 215 # \10 \11 \12  mean byte plus register value
 216 # \17          means byte zero
 217 # \330         means byte plus condition code
 218 # \0 or \340   mean give up and return empty set
 219 sub startbyte {
 220   local ($codes) = @_;
 221   local $word, @range;
 222
 223   while (1) {
 224     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 225     $word = $1, $codes = $2;
 226     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 227     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 228     return (0xA1, 0xA9) if $word eq "\\5";
 229     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 230     return (0xA0, 0xA8) if $word eq "\\7";
 231     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 232     return (0) if $word eq "\\17";
 233     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 234     return () if $word eq "\\0" || $word eq "\\340";
 235   }
 236   @range = ();
 237   push @range, $start++ while ($r-- > 0);
 238   @range;
 239 }