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 $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 print STDERR "Writing insnsi.h...\n";
 106
 107 open I, ">insnsi.h";
 108
 109 print I "/* This file is auto-generated from insns.dat by insns.pl" .
 110         " - don't exit it */\n\n";
 111 print I "/* This file in included by nasm.h */\n\n";
 112
 113 print I "/* Instruction names */\n";
 114 print I "enum {";
 115 $first  = 1;
 116 $maxlen = 0;
 117 foreach $i (@opcodes) {
 118     print I "," if ( !$first );
 119     $first = 0;
 120     print I "\n\tI_${i}";
 121     $len = length($i);
 122     $len++ if ( $i =~ /cc$/ );  # Condition codes can be 3 characters long
 123     $maxlen = $len if ( $len > $maxlen );
 124 }
 125 print I "\n};\n\n";
 126 print I "#define MAX_INSLEN ", $maxlen, "\n";
 127
 128 close I;
 129
 130 print STDERR "Writing insnsn.c...\n";
 131
 132 open N, ">insnsn.c";
 133
 134 print N "/* This file is auto-generated from insns.dat by insns.pl" .
 135         " - don't exit it */\n\n";
 136 print N "/* This file in included by names.c */\n\n";
 137
 138 print N "static char *insn_names[] = {";
 139 $first = 1;
 140 foreach $i (@opcodes) {
 141     # Don't include conditional instructions
 142     if ( $i !~ /cc$/ ) {
 143         print N "," if ( !$first );
 144         $first = 0;
 145         $ilower = $i;
 146         $ilower =~ tr/A-Z/a-z/;
 147         print N "\n\t\"${ilower}\"";
 148     }
 149 }
 150 print N "\n};\n\n";
 151 print N "/* Conditional instructions */\n";
 152 print N "static char *icn[] = {";
 153 $first = 1;
 154 foreach $i (@opcodes) {
 155     # Only conditional instructions
 156     if ( $i =~ /cc$/ ) {
 157         $ins = $`;              # Skip cc suffix
 158         print N "," if ( !$first );
 159         $first = 0;
 160         $ilower = $i;
 161         $ilower =~ tr/A-Z/a-z/;
 162         print N "\n\t\"${ilower}\"";
 163     }
 164 }
 165
 166 print N "\n};\n\n";
 167 print N "/* and the corresponding opcodes */\n";
 168 print N "static int ico[] = {";
 169 $first = 1;
 170 foreach $i (@opcodes) {
 171     # Only conditional instructions
 172     if ( $i =~ /cc$/ ) {
 173         $ins = $`;              # Skip cc suffix
 174         print N "," if ( !$first );
 175         $first = 0;
 176         print N "\n\tI_$i";
 177     }
 178 }
 179
 180 print N "\n};\n";
 181
 182 close N;
 183
 184 printf STDERR "Done: %d instructions\n", $insns;
 185
 186 sub format {
 187   local ($opcode, $operands, $codes, $flags) = @_;
 188   local $num, $nd = 0;
 189
 190   return (undef, undef) if $operands eq "ignore";
 191
 192   # format the operands
 193   $operands =~ s/:/|colon,/g;
 194   $operands =~ s/mem(\d+)/mem|bits$1/g;
 195   $operands =~ s/mem/memory/g;
 196   $operands =~ s/memory_offs/mem_offs/g;
 197   $operands =~ s/imm(\d+)/imm|bits$1/g;
 198   $operands =~ s/imm/immediate/g;
 199   $operands =~ s/rm(\d+)/regmem|bits$1/g;
 200   $num = 3;
 201   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 202   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 203   $operands =~ tr/a-z/A-Z/;
 204
 205   # format the flags
 206   $flags =~ s/,/|IF_/g;
 207   $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 208   $flags = "IF_" . $flags;
 209
 210   ("{I_$opcode, $num, {$operands}, \"$codes\", $flags},", $nd);
 211 }
 212
 213 # Here we determine the range of possible starting bytes for a given
 214 # instruction. We need only consider the codes:
 215 # \1 \2 \3     mean literal bytes, of course
 216 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 217 # \10 \11 \12  mean byte plus register value
 218 # \17          means byte zero
 219 # \330         means byte plus condition code
 220 # \0 or \340   mean give up and return empty set
 221 sub startbyte {
 222   local ($codes) = @_;
 223   local $word, @range;
 224
 225   while (1) {
 226     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 227     $word = $1, $codes = $2;
 228     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 229     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 230     return (0xA1, 0xA9) if $word eq "\\5";
 231     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 232     return (0xA0, 0xA8) if $word eq "\\7";
 233     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 234     return (0) if $word eq "\\17";
 235     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 236     return () if $word eq "\\0" || $word eq "\\340";
 237   }
 238   @range = ();
 239   push @range, $start++ while ($r-- > 0);
 240   @range;
 241 }