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 @args   = ();
  13 undef $output;
  14 foreach $arg ( @ARGV ) {
  15     if ( $arg =~ /^\-/ ) {
  16         if  ( $arg =~ /^\-([adin])$/ ) {
  17             $output = $1;
  18         } else {
  19             die "$0: Unknown option: ${arg}\n";
  20         }
  21     } else {
  22         push (@args, $arg);
  23     }
  24 }
  25
  26 $fname = "insns.dat" unless $fname = $args[0];
  27 open (F, $fname) || die "unable to open $fname";
  28
  29 $line = 0;
  30 $insns = 0;
  31 while (<F>) {
  32   $line++;
  33   next if /^\s*;/;   # comments
  34   chomp;
  35   split;
  36   next if $#_ == -1; # blank lines
  37   (warn "line $line does not contain four fields\n"), next if $#_ != 3;
  38   ($formatted, $nd) = &format(@_);
  39   if ($formatted) {
  40     $insns++;
  41     $aname = "aa_$_[0]";
  42     push @$aname, $formatted;
  43   }
  44   if ( $_[0] =~ /cc$/ ) {
  45       # Conditional instruction
  46       $k_opcodes_cc{$_[0]}++;
  47   } else {
  48       # Unconditional instruction
  49       $k_opcodes{$_[0]}++;
  50   }
  51   if ($formatted && !$nd) {
  52     push @big, $formatted;
  53     foreach $i (&startbyte($_[2])) {
  54       $aname = sprintf "dd_%02X",$i;
  55       push @$aname, $#big;
  56     }
  57   }
  58 }
  59
  60 close F;
  61
  62 @opcodes    = sort keys(%k_opcodes);
  63 @opcodes_cc = sort keys(%k_opcodes_cc);
  64
  65 if ( !defined($output) || $output eq 'a' ) {
  66     print STDERR "Writing insnsa.c...\n";
  67
  68     open A, ">insnsa.c";
  69
  70     print A "/* This file auto-generated from insns.dat by insns.pl" .
  71         " - don't edit it */\n\n";
  72     print A "#include \"nasm.h\"\n";
  73     print A "#include \"insns.h\"\n";
  74     print A "\n";
  75
  76     foreach $i (@opcodes, @opcodes_cc) {
  77         print A "static const struct itemplate instrux_${i}[] = {\n";
  78         $aname = "aa_$i";
  79         foreach $j (@$aname) {
  80             print A "    $j\n";
  81         }
  82         print A "    ITEMPLATE_END\n};\n\n";
  83     }
  84     print A "const struct itemplate * const nasm_instructions[] = {\n";
  85     foreach $i (@opcodes, @opcodes_cc) {
  86         print A "    instrux_${i},\n";
  87     }
  88     print A "};\n";
  89
  90     close A;
  91 }
  92
  93 if ( !defined($output) || $output eq 'd' ) {
  94     print STDERR "Writing insnsd.c...\n";
  95
  96     open D, ">insnsd.c";
  97
  98     print D "/* This file auto-generated from insns.dat by insns.pl" .
  99         " - don't edit it */\n\n";
 100     print D "#include \"nasm.h\"\n";
 101     print D "#include \"insns.h\"\n";
 102     print D "\n";
 103
 104     print D "static const struct itemplate instrux[] = {\n";
 105     foreach $j (@big) {
 106         print D "    $j\n";
 107     }
 108         print D "    ITEMPLATE_END\n};\n\n";
 109
 110     for ($c=0; $c<256; $c++) {
 111         $h = sprintf "%02X", $c;
 112         print D "static const struct itemplate * const itable_${h}[] = {\n";
 113         $aname = "dd_$h";
 114         foreach $j (@$aname) {
 115             print D "    instrux + $j,\n";
 116         }
 117         print D "    NULL\n};\n\n";
 118     }
 119
 120     print D "const struct itemplate * const * const itable[] = {\n";
 121     for ($c=0; $c<256; $c++) {
 122         printf D "    itable_%02X,\n", $c;
 123     }
 124     print D "};\n";
 125
 126     close D;
 127 }
 128
 129 if ( !defined($output) || $output eq 'i' ) {
 130     print STDERR "Writing insnsi.h...\n";
 131
 132     open I, ">insnsi.h";
 133
 134     print I "/* This file is auto-generated from insns.dat by insns.pl" .
 135         " - don't edit it */\n\n";
 136     print I "/* This file in included by nasm.h */\n\n";
 137
 138     print I "/* Instruction names */\n\n";
 139     print I "#ifndef NASM_INSNSI_H\n";
 140     print I "#define NASM_INSNSI_H 1\n\n";
 141     print I "enum opcode {\n";
 142     $maxlen = 0;
 143     foreach $i (@opcodes, @opcodes_cc) {
 144         print I "\tI_${i},\n";
 145         $len = length($i);
 146         $len++ if ( $i =~ /cc$/ );      # Condition codes can be 3 characters long
 147         $maxlen = $len if ( $len > $maxlen );
 148     }
 149     print I "\tI_none = -1\n";
 150     print I "\n};\n\n";
 151     print I "#define MAX_INSLEN ", $maxlen, "\n\n";
 152     print I "#endif /* NASM_INSNSI_H */\n";
 153
 154     close I;
 155 }
 156
 157 if ( !defined($output) || $output eq 'n' ) {
 158     print STDERR "Writing insnsn.c...\n";
 159
 160     open N, ">insnsn.c";
 161
 162     print N "/* This file is auto-generated from insns.dat by insns.pl" .
 163         " - don't edit it */\n\n";
 164     print N "/* This file in included by names.c */\n\n";
 165
 166     print N "static const char * const insn_names[] = {";
 167     $first = 1;
 168     foreach $i (@opcodes) {
 169         print N "," if ( !$first );
 170         $first = 0;
 171         $ilower = $i;
 172         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 173         print N "\n\t\"${ilower}\"";
 174     }
 175     print N "\n};\n\n";
 176     print N "/* Conditional instructions */\n";
 177     print N "static const char *icn[] = {";
 178     $first = 1;
 179     foreach $i (@opcodes_cc) {
 180         print N "," if ( !$first );
 181         $first = 0;
 182         $ilower = $i;
 183         $ilower =~ s/cc$//;             # Skip cc suffix
 184         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 185         print N "\n\t\"${ilower}\"";
 186     }
 187
 188     print N "\n};\n\n";
 189     print N "/* and the corresponding opcodes */\n";
 190     print N "static const enum opcode ico[] = {";
 191     $first = 1;
 192     foreach $i (@opcodes_cc) {
 193         print N "," if ( !$first );
 194         $first = 0;
 195         print N "\n\tI_$i";
 196     }
 197
 198     print N "\n};\n";
 199
 200     close N;
 201 }
 202
 203 printf STDERR "Done: %d instructions\n", $insns;
 204
 205 sub format {
 206   local ($opcode, $operands, $codes, $flags) = @_;
 207   local $num, $nd = 0;
 208
 209   return (undef, undef) if $operands eq "ignore";
 210
 211   # format the operands
 212   $operands =~ s/:/|colon,/g;
 213   $operands =~ s/mem(\d+)/mem|bits$1/g;
 214   $operands =~ s/mem/memory/g;
 215   $operands =~ s/memory_offs/mem_offs/g;
 216   $operands =~ s/imm(\d+)/imm|bits$1/g;
 217   $operands =~ s/imm/immediate/g;
 218   $operands =~ s/rm(\d+)/rm_gpr|bits$1/g;
 219   $operands =~ s/mmxrm/rm_mmx/g;
 220   $operands =~ s/xmmrm/rm_xmm/g;
 221   $num = 3;
 222   $operands = '0,0,0', $num = 0 if $operands eq 'void';
 223   $operands .= ',0', $num-- while $operands !~ /,.*,/;
 224   $operands =~ tr/a-z/A-Z/;
 225
 226   # format the flags
 227   $flags =~ s/,/|IF_/g;
 228   $flags =~ s/(\|IF_ND|IF_ND\|)//, $nd = 1 if $flags =~ /IF_ND/;
 229   $flags = "IF_" . $flags;
 230
 231   ("{I_$opcode, $num, {$operands}, \"$codes\", $flags},", $nd);
 232 }
 233
 234 # Here we determine the range of possible starting bytes for a given
 235 # instruction. We need only consider the codes:
 236 # \1 \2 \3     mean literal bytes, of course
 237 # \4 \5 \6 \7  mean PUSH/POP of segment registers: special case
 238 # \10 \11 \12  mean byte plus register value
 239 # \17          means byte zero
 240 # \330         means byte plus condition code
 241 # \0 or \340   mean give up and return empty set
 242 sub startbyte {
 243   local ($codes) = @_;
 244   local $word, @range;
 245
 246   while (1) {
 247     die "couldn't get code in '$codes'" if $codes !~ /^(\\[^\\]+)(\\.*)?$/;
 248     $word = $1, $codes = $2;
 249     return (hex $1) if $word =~ /^\\[123]$/ && $codes =~ /^\\x(..)/;
 250     return (0x07, 0x17, 0x1F) if $word eq "\\4";
 251     return (0xA1, 0xA9) if $word eq "\\5";
 252     return (0x06, 0x0E, 0x16, 0x1E) if $word eq "\\6";
 253     return (0xA0, 0xA8) if $word eq "\\7";
 254     $start=hex $1, $r=8, last if $word =~ /^\\1[012]$/ && $codes =~/^\\x(..)/;
 255     return (0) if $word eq "\\17";
 256     $start=hex $1, $r=16, last if $word =~ /^\\330$/ && $codes =~ /^\\x(..)/;
 257     return () if $word eq "\\0" || $word eq "\\340";
 258   }
 259   @range = ();
 260   push @range, $start++ while ($r-- > 0);
 261   @range;
 262 }