mirror of
https://github.com/netwide-assembler/nasm.git
synced 2024-11-27 08:10:07 +08:00
5255fd1f36
Combining arithmetric (add) and bitwise (xor) mixing seems to give better result than either. With the new prehash function, we find a valid hash much quicker.
210 lines
5.5 KiB
Perl
Executable File
210 lines
5.5 KiB
Perl
Executable File
#!/usr/bin/perl
|
|
#
|
|
# Produce pptok.c and pptok.h from pptok.dat
|
|
#
|
|
|
|
require 'phash.ph';
|
|
|
|
my($what, $in, $out) = @ARGV;
|
|
|
|
#
|
|
# Read pptok.dat
|
|
#
|
|
open(IN, "< $in") or die "$0: cannot open: $in\n";
|
|
while (defined($line = <IN>)) {
|
|
chomp $line;
|
|
$line =~ s/^\s+//; # Remove leading whitespace
|
|
$line =~ s/\s*\#.*$//; # Remove comments and trailing whitespace
|
|
next if ($line eq '');
|
|
|
|
if ($line =~ /^\%(.*)\*$/) {
|
|
push(@cctok, $1);
|
|
} elsif ($line =~ /^\%(.*)$/) {
|
|
push(@pptok, $1);
|
|
} elsif ($line =~ /^\*(.*)$/) {
|
|
push(@cond, $1);
|
|
}
|
|
}
|
|
close(IN);
|
|
|
|
@cctok = sort @cctok;
|
|
@cond = sort @cond;
|
|
@pptok = sort @pptok;
|
|
|
|
# Generate the expanded list including conditionals. The conditionals
|
|
# are at the beginning, padded to a power of 2, with the inverses
|
|
# interspersed; this allows a simple mask to pick out the condition.
|
|
|
|
while ((scalar @cond) & (scalar @cond)-1) {
|
|
push(@cond, undef);
|
|
}
|
|
|
|
@cptok = ();
|
|
foreach $ct (@cctok) {
|
|
foreach $cc (@cond) {
|
|
if (defined($cc)) {
|
|
push(@cptok, $ct.$cc);
|
|
push(@cptok, $ct.'n'.$cc);
|
|
} else {
|
|
push(@cptok, undef, undef);
|
|
}
|
|
}
|
|
}
|
|
$first_uncond = $pptok[0];
|
|
@pptok = (@cptok, @pptok);
|
|
|
|
open(OUT, "> $out") or die "$0: cannot open: $out\n";
|
|
print OUT "/* Automatically generated from $in by $0 */\n";
|
|
print OUT "/* Do not edit */\n";
|
|
print OUT "\n";
|
|
|
|
#
|
|
# Output pptok.h
|
|
#
|
|
if ($what eq 'h') {
|
|
print OUT "enum preproc_token {\n";
|
|
$n = 0;
|
|
foreach $pt (@pptok) {
|
|
if (defined($pt)) {
|
|
printf OUT " %-16s = %3d,\n", "PP_\U$pt\E", $n;
|
|
}
|
|
$n++;
|
|
}
|
|
printf OUT " %-16s = %3d\n", 'PP_INVALID', -1;
|
|
print OUT "};\n";
|
|
print OUT "\n";
|
|
|
|
print OUT "enum pp_conditional {\n";
|
|
$n = 0;
|
|
foreach $cc (@cond) {
|
|
if (defined($cc)) {
|
|
printf OUT " %-16s = %3d,\n", "PPC_IF\U$cc\E", $n;
|
|
}
|
|
$n += 2;
|
|
}
|
|
print OUT "};\n\n";
|
|
|
|
printf OUT "#define PP_COND(x) ((enum pp_conditional)((x) & 0x%x))\n",
|
|
(scalar(@cond)-1) << 1;
|
|
print OUT "#define PP_IS_COND(x) ((unsigned int)(x) < PP_\U$first_uncond\E)\n";
|
|
print OUT "#define PP_NEGATIVE(x) ((x) & 1)\n";
|
|
print OUT "\n";
|
|
|
|
foreach $ct (@cctok) {
|
|
print OUT "#define CASE_PP_\U$ct\E";
|
|
$pref = " \\\n";
|
|
foreach $cc (@cond) {
|
|
if (defined($cc)) {
|
|
print OUT "$pref\tcase PP_\U${ct}${cc}\E: \\\n";
|
|
print OUT "\tcase PP_\U${ct}N${cc}\E";
|
|
$pref = ":\\\n";
|
|
}
|
|
}
|
|
print OUT "\n"; # No colon or newline on the last one
|
|
}
|
|
}
|
|
|
|
#
|
|
# Output pptok.c
|
|
#
|
|
if ($what eq 'c') {
|
|
my %tokens = ();
|
|
my @tokendata = ();
|
|
|
|
my $n = 0;
|
|
foreach $pt (@pptok) {
|
|
if (defined($pt)) {
|
|
$tokens{'%'.$pt} = $n;
|
|
if ($pt =~ /[\@\[\]\\_]/) {
|
|
# Fail on characters which look like upper-case letters
|
|
# to the quick-and-dirty downcasing in the prehash
|
|
# (see below)
|
|
die "$in: invalid character in token: $pt";
|
|
}
|
|
}
|
|
$n++;
|
|
}
|
|
|
|
my @hashinfo = gen_perfect_hash(\%tokens);
|
|
if (!defined(@hashinfo)) {
|
|
die "$0: no hash found\n";
|
|
}
|
|
|
|
# Paranoia...
|
|
verify_hash_table(\%tokens, \@hashinfo);
|
|
|
|
($n, $sv, $g) = @hashinfo;
|
|
$sv2 = $sv+2;
|
|
|
|
die if ($n & ($n-1));
|
|
|
|
print OUT "#include <inttypes.h>\n";
|
|
print OUT "#include <ctype.h>\n";
|
|
print OUT "#include \"nasmlib.h\"\n";
|
|
print OUT "#include \"preproc.h\"\n";
|
|
print OUT "\n";
|
|
|
|
print OUT "#define rot(x,y) (((uint32_t)(x) << (y))+((uint32_t)(x) >> (32-(y))))\n";
|
|
print OUT "\n";
|
|
|
|
# Note that this is global.
|
|
printf OUT "const char * const pp_directives[%d] = {\n", scalar(@pptok);
|
|
foreach $d (@pptok) {
|
|
if (defined($d)) {
|
|
print OUT " \"%$d\",\n";
|
|
} else {
|
|
print OUT " NULL,\n";
|
|
}
|
|
}
|
|
print OUT "};\n";
|
|
|
|
print OUT "enum preproc_token pp_token_hash(const char *token)\n";
|
|
print OUT "{\n";
|
|
|
|
# Put a large value in unused slots. This makes it extremely unlikely
|
|
# that any combination that involves unused slot will pass the range test.
|
|
# This speeds up rejection of unrecognized tokens, i.e. identifiers.
|
|
print OUT "#define UNUSED 16383\n";
|
|
|
|
print OUT " static const int16_t hash1[$n] = {\n";
|
|
for ($i = 0; $i < $n; $i++) {
|
|
my $h = ${$g}[$i*2+0];
|
|
print OUT " ", defined($h) ? $h : 'UNUSED', ",\n";
|
|
}
|
|
print OUT " };\n";
|
|
|
|
print OUT " static const int16_t hash2[$n] = {\n";
|
|
for ($i = 0; $i < $n; $i++) {
|
|
my $h = ${$g}[$i*2+1];
|
|
print OUT " ", defined($h) ? $h : 'UNUSED', ",\n";
|
|
}
|
|
print OUT " };\n";
|
|
|
|
print OUT " uint32_t k1 = 0, k2 = 0;\n";
|
|
print OUT " uint8_t c;\n";
|
|
# For correct overflow behavior, "ix" should be unsigned of the same
|
|
# width as the hash arrays.
|
|
print OUT " uint16_t ix;\n";
|
|
print OUT " const char *p = token;\n";
|
|
print OUT "\n";
|
|
|
|
print OUT " while ((c = *p++) != 0) {\n";
|
|
print OUT " uint32_t kn1, kn2;\n";
|
|
print OUT " c |= 0x20; /* convert to lower case */\n";
|
|
printf OUT " kn1 = rot(k1,%2d)^(rot(k2,%2d) + c);\n", ${$sv}[0], ${$sv}[1];
|
|
printf OUT " kn2 = rot(k2,%2d)^(rot(k1,%2d) + c);\n", ${$sv}[2], ${$sv}[3];
|
|
print OUT " k1 = kn1; k2 = kn2;\n";
|
|
print OUT " }\n";
|
|
print OUT "\n";
|
|
printf OUT " ix = hash1[k1 & 0x%x] + hash2[k2 & 0x%x];\n", $n-1, $n-1;
|
|
printf OUT " if (ix >= %d)\n", scalar(@pptok);
|
|
print OUT " return PP_INVALID;\n";
|
|
print OUT "\n";
|
|
|
|
print OUT " if (!pp_directives[ix] || nasm_stricmp(pp_directives[ix], token))\n";
|
|
print OUT " return PP_INVALID;\n";
|
|
print OUT "\n";
|
|
print OUT " return ix;\n";
|
|
print OUT "}\n";
|
|
}
|