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 \<inttypes.h\>\n\n";
  73     print A "#include \"nasm.h\"\n";
  74     print A "#include \"insns.h\"\n";
  75     print A "\n";
  76
  77     foreach $i (@opcodes, @opcodes_cc) {
  78         print A "static struct itemplate instrux_${i}[] = {\n";
  79         $aname = "aa_$i";
  80         foreach $j (@$aname) {
  81             print A "    $j\n";
  82         }
  83         print A "    ITEMPLATE_END\n};\n\n";
  84     }
  85     print A "struct itemplate *nasm_instructions[] = {\n";
  86     foreach $i (@opcodes, @opcodes_cc) {
  87         print A "    instrux_${i},\n";
  88     }
  89     print A "};\n";
  90
  91     close A;
  92 }
  93
  94 if ( !defined($output) || $output eq 'd' ) {
  95     print STDERR "Writing insnsd.c...\n";
  96
  97     open D, ">insnsd.c";
  98
  99     print D "/* This file auto-generated from insns.dat by insns.pl" .
 100         " - don't edit it */\n\n";
 101     print D "#include \<inttypes.h\>\n\n";
 102     print D "#include \"nasm.h\"\n";
 103     print D "#include \"insns.h\"\n";
 104     print D "\n";
 105
 106     print D "static struct itemplate instrux[] = {\n";
 107     foreach $j (@big) {
 108         print D "    $j\n";
 109     }
 110         print D "    ITEMPLATE_END\n};\n\n";
 111
 112     for ($c=0; $c<256; $c++) {
 113         $h = sprintf "%02X", $c;
 114         print D "static struct itemplate *itable_${h}[] = {\n";
 115         $aname = "dd_$h";
 116         foreach $j (@$aname) {
 117             print D "    instrux + $j,\n";
 118         }
 119         print D "    NULL\n};\n\n";
 120     }
 121
 122     print D "struct itemplate **itable[] = {\n";
 123     for ($c=0; $c<256; $c++) {
 124         printf D "    itable_%02X,\n", $c;
 125     }
 126     print D "};\n";
 127
 128     close D;
 129 }
 130
 131 if ( !defined($output) || $output eq 'i' ) {
 132     print STDERR "Writing insnsi.h...\n";
 133
 134     open I, ">insnsi.h";
 135
 136     print I "/* This file is auto-generated from insns.dat by insns.pl" .
 137         " - don't edit it */\n\n";
 138     print I "/* This file in included by nasm.h */\n\n";
 139
 140     print I "/* Instruction names */\n";
 141     print I "enum {";
 142     $first  = 1;
 143     $maxlen = 0;
 144     foreach $i (@opcodes, @opcodes_cc) {
 145         print I "," if ( !$first );
 146         $first = 0;
 147         print I "\n\tI_${i}";
 148         $len = length($i);
 149         $len++ if ( $i =~ /cc$/ );      # Condition codes can be 3 characters long
 150         $maxlen = $len if ( $len > $maxlen );
 151     }
 152     print I "\n};\n\n";
 153     print I "#define MAX_INSLEN ", $maxlen, "\n";
 154
 155     close I;
 156 }
 157
 158 if ( !defined($output) || $output eq 'n' ) {
 159     print STDERR "Writing insnsn.c...\n";
 160
 161     open N, ">insnsn.c";
 162
 163     print N "/* This file is auto-generated from insns.dat by insns.pl" .
 164         " - don't edit it */\n\n";
 165     print N "/* This file in included by names.c */\n\n";
 166     print N "#include \<inttypes.h\>\n\n";
 167
 168     print N "static const int8_t *insn_names[] = {";
 169     $first = 1;
 170     foreach $i (@opcodes) {
 171         print N "," if ( !$first );
 172         $first = 0;
 173         $ilower = $i;
 174         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 175         print N "\n\t\"${ilower}\"";
 176     }
 177     print N "\n};\n\n";
 178     print N "/* Conditional instructions */\n";
 179     print N "static const int8_t *icn[] = {";
 180     $first = 1;
 181     foreach $i (@opcodes_cc) {
 182         print N "," if ( !$first );
 183         $first = 0;
 184         $ilower = $i;
 185         $ilower =~ s/cc$//;             # Skip cc suffix
 186         $ilower =~ tr/A-Z/a-z/; # Change to lower case (Perl 4 compatible)
 187         print N "\n\t\"${ilower}\"";
 188     }
 189
 190     print N "\n};\n\n";
 191     print N "/* and the corresponding opcodes */\n";
 192     print N "static int ico[] = {";
 193     $first = 1;
 194     foreach $i (@opcodes_cc) {
 195         print N "," if ( !$first );
 196         $first = 0;
 197         print N "\n\tI_$i";
 198     }
 199
 200     print N "\n};\n";
 201
 202     close N;
 203 }
 204
 205 printf STDERR "Done: %d instructions\n", $insns;
 206
 207 sub format {
 208   local ($opcode, $operands, $codes, $flags) = @_;
 209   local $num, $nd = 0;
 210
 211   return (undef, undef) if $operands eq "ignore";
 212
 213   # format the operands
 214   $operands =~ s/:/|colon,/g;
 215   $operands =~ s/mem(\d+)/mem|bits$1/g;
 216   $operands =~ s/mem/memory/g;
 217   $operands =~ s/memory_offs/mem_offs/g;
 218   $operands =~ s/imm(\d+)/imm|bits$1/g;
 219   $operands =~ s/imm/immediate/g;
 220   $operands =~ s/rm(\d+)/regmem|bits$1/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 }