bytecode.pl

   1 BEGIN {
   2   push @INC, './lib';
   3   require 'regen_lib.pl';
   4 }
   5 use strict;
   6 my %alias_to = (
   7     U32 => [qw(PADOFFSET STRLEN line_t)],
   8     I32 => [qw(SSize_t long)],
   9     U16 => [qw(OPCODE short)],
  10     U8  => [qw(char)],
  11 );
  12
  13 my @optype= qw(OP UNOP BINOP LOGOP LISTOP PMOP SVOP PADOP PVOP LOOP COP);
  14
  15 # Nullsv *must* come first in the following so that the condition
  16 # ($$sv == 0) can continue to be used to test (sv == Nullsv).
  17 my @specialsv = qw(Nullsv &PL_sv_undef &PL_sv_yes &PL_sv_no pWARN_ALL pWARN_NONE);
  18
  19 my (%alias_from, $from, $tos);
  20 while (($from, $tos) = each %alias_to) {
  21     map { $alias_from{$_} = $from } @$tos;
  22 }
  23
  24 my $c_header = <<'EOT';
  25 /*
  26  *      Copyright (c) 1996-1999 Malcolm Beattie
  27  *
  28  *      You may distribute under the terms of either the GNU General Public
  29  *      License or the Artistic License, as specified in the README file.
  30  *
  31  */
  32 /*
  33  * This file is autogenerated from bytecode.pl. Changes made here will be lost.
  34  */
  35 EOT
  36
  37 my $perl_header;
  38 ($perl_header = $c_header) =~ s{[/ ]?\*/?}{#}g;
  39
  40 safer_unlink "ext/ByteLoader/byterun.c", "ext/ByteLoader/byterun.h", "ext/B/B/Asmdata.pm";
  41
  42 #
  43 # Start with boilerplate for Asmdata.pm
  44 #
  45 open(ASMDATA_PM, ">ext/B/B/Asmdata.pm") or die "ext/B/B/Asmdata.pm: $!";
  46 print ASMDATA_PM $perl_header, <<'EOT';
  47 package B::Asmdata;
  48
  49 our $VERSION = '1.00';
  50
  51 use Exporter;
  52 @ISA = qw(Exporter);
  53 @EXPORT_OK = qw(%insn_data @insn_name @optype @specialsv_name);
  54 our(%insn_data, @insn_name, @optype, @specialsv_name);
  55
  56 EOT
  57 print ASMDATA_PM <<"EOT";
  58 \@optype = qw(@optype);
  59 \@specialsv_name = qw(@specialsv);
  60
  61 # XXX insn_data is initialised this way because with a large
  62 # %insn_data = (foo => [...], bar => [...], ...) initialiser
  63 # I get a hard-to-track-down stack underflow and segfault.
  64 EOT
  65
  66 #
  67 # Boilerplate for byterun.c
  68 #
  69 open(BYTERUN_C, ">ext/ByteLoader/byterun.c") or die "ext/ByteLoader/byterun.c: $!";
  70 print BYTERUN_C $c_header, <<'EOT';
  71
  72 #define PERL_NO_GET_CONTEXT
  73 #include "EXTERN.h"
  74 #include "perl.h"
  75 #define NO_XSLOCKS
  76 #include "XSUB.h"
  77
  78 #include "byterun.h"
  79 #include "bytecode.h"
  80
  81
  82 static const int optype_size[] = {
  83 EOT
  84 my $i = 0;
  85 for ($i = 0; $i < @optype - 1; $i++) {
  86     printf BYTERUN_C "    sizeof(%s),\n", $optype[$i], $i;
  87 }
  88 printf BYTERUN_C "    sizeof(%s)\n", $optype[$i], $i;
  89 print BYTERUN_C <<'EOT';
  90 };
  91
  92 void *
  93 bset_obj_store(pTHX_ struct byteloader_state *bstate, void *obj, I32 ix)
  94 {
  95     if (ix > bstate->bs_obj_list_fill) {
  96         Renew(bstate->bs_obj_list, ix + 32, void*);
  97         bstate->bs_obj_list_fill = ix + 31;
  98     }
  99     bstate->bs_obj_list[ix] = obj;
 100     return obj;
 101 }
 102
 103 int
 104 byterun(pTHX_ register struct byteloader_state *bstate)
 105 {
 106     register int insn;
 107     U32 ix;
 108     SV *specialsv_list[6];
 109
 110     BYTECODE_HEADER_CHECK;      /* croak if incorrect platform */
 111     New(666, bstate->bs_obj_list, 32, void*); /* set op objlist */
 112     bstate->bs_obj_list_fill = 31;
 113     bstate->bs_obj_list[0] = NULL; /* first is always Null */
 114     bstate->bs_ix = 1;
 115
 116 EOT
 117
 118 for my $i ( 0 .. $#specialsv ) {
 119     print BYTERUN_C "    specialsv_list[$i] = $specialsv[$i];\n";
 120 }
 121
 122 print BYTERUN_C <<'EOT';
 123
 124     while ((insn = BGET_FGETC()) != EOF) {
 125         switch (insn) {
 126 EOT
 127
 128
 129 my (@insn_name, $insn_num, $insn, $lvalue, $argtype, $flags, $fundtype);
 130
 131 while (<DATA>) {
 132     if (/^\s*#/) {
 133         print BYTERUN_C if /^\s*#\s*(?:if|endif|el)/;
 134         next;
 135     }
 136     chop;
 137     next unless length;
 138     if (/^%number\s+(.*)/) {
 139         $insn_num = $1;
 140         next;
 141     } elsif (/%enum\s+(.*?)\s+(.*)/) {
 142         create_enum($1, $2);    # must come before instructions
 143         next;
 144     }
 145     ($insn, $lvalue, $argtype, $flags) = split;
 146     $insn_name[$insn_num] = $insn;
 147     $fundtype = $alias_from{$argtype} || $argtype;
 148
 149     #
 150     # Add the case statement and code for the bytecode interpreter in byterun.c
 151     #
 152     printf BYTERUN_C "\t  case INSN_%s:\t\t/* %d */\n\t    {\n",
 153         uc($insn), $insn_num;
 154     my $optarg = $argtype eq "none" ? "" : ", arg";
 155     if ($optarg) {
 156         printf BYTERUN_C "\t\t$argtype arg;\n\t\tBGET_%s(arg);\n", $fundtype;
 157     }
 158     if ($flags =~ /x/) {
 159         print BYTERUN_C "\t\tBSET_$insn($lvalue$optarg);\n";
 160     } elsif ($flags =~ /s/) {
 161         # Store instructions store to bytecode_obj_list[arg]. "lvalue" field is rvalue.
 162         print BYTERUN_C "\t\tBSET_OBJ_STORE($lvalue$optarg);\n";
 163     }
 164     elsif ($optarg && $lvalue ne "none") {
 165         print BYTERUN_C "\t\t$lvalue = arg;\n";
 166     }
 167     print BYTERUN_C "\t\tbreak;\n\t    }\n";
 168
 169     #
 170     # Add the initialiser line for %insn_data in Asmdata.pm
 171     #
 172     print ASMDATA_PM <<"EOT";
 173 \$insn_data{$insn} = [$insn_num, \\&PUT_$fundtype, "GET_$fundtype"];
 174 EOT
 175
 176     # Find the next unused instruction number
 177     do { $insn_num++ } while $insn_name[$insn_num];
 178 }
 179
 180 #
 181 # Finish off byterun.c
 182 #
 183 print BYTERUN_C <<'EOT';
 184           default:
 185             Perl_croak(aTHX_ "Illegal bytecode instruction %d\n", insn);
 186             /* NOTREACHED */
 187         }
 188     }
 189     return 0;
 190 }
 191 EOT
 192
 193 #
 194 # Write the instruction and optype enum constants into byterun.h
 195 #
 196 open(BYTERUN_H, ">ext/ByteLoader/byterun.h") or die "ext/ByteLoader/byterun.h: $!";
 197 print BYTERUN_H $c_header, <<'EOT';
 198 struct byteloader_fdata {
 199     SV  *datasv;
 200     int next_out;
 201     int idx;
 202 };
 203
 204 struct byteloader_state {
 205     struct byteloader_fdata     *bs_fdata;
 206     SV                          *bs_sv;
 207     void                        **bs_obj_list;
 208     int                         bs_obj_list_fill;
 209     int                         bs_ix;
 210     XPV                         bs_pv;
 211     int                         bs_iv_overflows;
 212 };
 213
 214 int bl_getc(struct byteloader_fdata *);
 215 int bl_read(struct byteloader_fdata *, char *, size_t, size_t);
 216 extern int byterun(pTHX_ struct byteloader_state *);
 217
 218 enum {
 219 EOT
 220
 221 my $add_enum_value = 0;
 222 my $max_insn;
 223 for $i ( 0 .. $#insn_name ) {
 224     $insn = uc($insn_name[$i]);
 225     if (defined($insn)) {
 226         $max_insn = $i;
 227         if ($add_enum_value) {
 228             print BYTERUN_H "    INSN_$insn = $i,\t\t\t/* $i */\n";
 229             $add_enum_value = 0;
 230         } else {
 231             print BYTERUN_H "    INSN_$insn,\t\t\t/* $i */\n";
 232         }
 233     } else {
 234         $add_enum_value = 1;
 235     }
 236 }
 237
 238 print BYTERUN_H "    MAX_INSN = $max_insn\n};\n";
 239
 240 print BYTERUN_H "\nenum {\n";
 241 for ($i = 0; $i < @optype - 1; $i++) {
 242     printf BYTERUN_H "    OPt_%s,\t\t/* %d */\n", $optype[$i], $i;
 243 }
 244 printf BYTERUN_H "    OPt_%s\t\t/* %d */\n};\n\n", $optype[$i], $i;
 245
 246 #
 247 # Finish off insn_data and create array initialisers in Asmdata.pm
 248 #
 249 print ASMDATA_PM <<'EOT';
 250
 251 my ($insn_name, $insn_data);
 252 while (($insn_name, $insn_data) = each %insn_data) {
 253     $insn_name[$insn_data->[0]] = $insn_name;
 254 }
 255 # Fill in any gaps
 256 @insn_name = map($_ || "unused", @insn_name);
 257
 258 1;
 259
 260 __END__
 261
 262 =head1 NAME
 263
 264 B::Asmdata - Autogenerated data about Perl ops, used to generate bytecode
 265
 266 =head1 SYNOPSIS
 267
 268         use B::Asmdata qw(%insn_data @insn_name @optype @specialsv_name);
 269
 270 =head1 DESCRIPTION
 271
 272 Provides information about Perl ops in order to generate bytecode via
 273 a bunch of exported variables.  Its mostly used by B::Assembler and
 274 B::Disassembler.
 275
 276 =over 4
 277
 278 =item %insn_data
 279
 280   my($bytecode_num, $put_sub, $get_meth) = @$insn_data{$op_name};
 281
 282 For a given $op_name (for example, 'cop_label', 'sv_flags', etc...)
 283 you get an array ref containing the bytecode number of the op, a
 284 reference to the subroutine used to 'PUT', and the name of the method
 285 used to 'GET'.
 286
 287 =for _private
 288 Add more detail about what $put_sub and $get_meth are and how to use them.
 289
 290 =item @insn_name
 291
 292   my $op_name = $insn_name[$bytecode_num];
 293
 294 A simple mapping of the bytecode number to the name of the op.
 295 Suitable for using with %insn_data like so:
 296
 297   my $op_info = $insn_data{$insn_name[$bytecode_num]};
 298
 299 =item @optype
 300
 301   my $op_type = $optype[$op_type_num];
 302
 303 A simple mapping of the op type number to its type (like 'COP' or 'BINOP').
 304
 305 =item @specialsv_name
 306
 307   my $sv_name = $specialsv_name[$sv_index];
 308
 309 Certain SV types are considered 'special'.  They're represented by
 310 B::SPECIAL and are refered to by a number from the specialsv_list.
 311 This array maps that number back to the name of the SV (like 'Nullsv'
 312 or '&PL_sv_undef').
 313
 314 =back
 315
 316 =head1 AUTHOR
 317
 318 Malcolm Beattie, C<mbeattie@sable.ox.ac.uk>
 319
 320 =cut
 321 EOT
 322
 323
 324 close ASMDATA_PM or die "Error closing ASMDATA_PM: $!";
 325 close BYTERUN_H or die "Error closing BYTERUN_H: $!";
 326 close BYTERUN_C or die "Error closing BYTERUN_C: $!";
 327
 328 __END__
 329 # First set instruction ord("#") to read comment to end-of-line (sneaky)
 330 %number 35
 331 comment         arg                     comment_t
 332 # Then make ord("\n") into a no-op
 333 %number 10
 334 nop             none                    none
 335
 336 # Now for the rest of the ordinary ones, beginning with \0 which is
 337 # ret so that \0-terminated strings can be read properly as bytecode.
 338 %number 0
 339 #
 340 #opcode         lvalue                                  argtype         flags
 341 #
 342 ret             none                                    none            x
 343 ldsv            bstate->bs_sv                           svindex
 344 ldop            PL_op                                   opindex
 345 stsv            bstate->bs_sv                           U32             s
 346 stop            PL_op                                   U32             s
 347 stpv            bstate->bs_pv.xpv_pv                    U32             x
 348 ldspecsv        bstate->bs_sv                           U8              x
 349 ldspecsvx       bstate->bs_sv                           U8              x
 350 newsv           bstate->bs_sv                           U8              x
 351 newsvx          bstate->bs_sv                           U32             x
 352 newop           PL_op                                   U8              x
 353 newopx          PL_op                                   U16             x
 354 newopn          PL_op                                   U8              x
 355 newpv           none                                    PV
 356 pv_cur          bstate->bs_pv.xpv_cur                   STRLEN
 357 pv_free         bstate->bs_pv                           none            x
 358 sv_upgrade      bstate->bs_sv                           U8              x
 359 sv_refcnt       SvREFCNT(bstate->bs_sv)                 U32
 360 sv_refcnt_add   SvREFCNT(bstate->bs_sv)                 I32             x
 361 sv_flags        SvFLAGS(bstate->bs_sv)                  U32
 362 xrv             SvRV(bstate->bs_sv)                     svindex
 363 xpv             bstate->bs_sv                           none            x
 364 xpv_cur         SvCUR(bstate->bs_sv)                    STRLEN
 365 xpv_len         SvLEN(bstate->bs_sv)                    STRLEN
 366 xiv             SvIVX(bstate->bs_sv)                    IV
 367 xnv             SvNVX(bstate->bs_sv)                    NV
 368 xlv_targoff     LvTARGOFF(bstate->bs_sv)                STRLEN
 369 xlv_targlen     LvTARGLEN(bstate->bs_sv)                STRLEN
 370 xlv_targ        LvTARG(bstate->bs_sv)                   svindex
 371 xlv_type        LvTYPE(bstate->bs_sv)                   char
 372 xbm_useful      BmUSEFUL(bstate->bs_sv)                 I32
 373 xbm_previous    BmPREVIOUS(bstate->bs_sv)               U16
 374 xbm_rare        BmRARE(bstate->bs_sv)                   U8
 375 xfm_lines       FmLINES(bstate->bs_sv)                  IV
 376 xio_lines       IoLINES(bstate->bs_sv)                  IV
 377 xio_page        IoPAGE(bstate->bs_sv)                   IV
 378 xio_page_len    IoPAGE_LEN(bstate->bs_sv)               IV
 379 xio_lines_left  IoLINES_LEFT(bstate->bs_sv)             IV
 380 xio_top_name    IoTOP_NAME(bstate->bs_sv)               pvindex
 381 xio_top_gv      *(SV**)&IoTOP_GV(bstate->bs_sv)         svindex
 382 xio_fmt_name    IoFMT_NAME(bstate->bs_sv)               pvindex
 383 xio_fmt_gv      *(SV**)&IoFMT_GV(bstate->bs_sv)         svindex
 384 xio_bottom_name IoBOTTOM_NAME(bstate->bs_sv)            pvindex
 385 xio_bottom_gv   *(SV**)&IoBOTTOM_GV(bstate->bs_sv)      svindex
 386 xio_subprocess  IoSUBPROCESS(bstate->bs_sv)             short
 387 xio_type        IoTYPE(bstate->bs_sv)                   char
 388 xio_flags       IoFLAGS(bstate->bs_sv)                  char
 389 xcv_xsubany     *(SV**)&CvXSUBANY(bstate->bs_sv).any_ptr        svindex
 390 xcv_stash       *(SV**)&CvSTASH(bstate->bs_sv)          svindex
 391 xcv_start       CvSTART(bstate->bs_sv)                  opindex
 392 xcv_root        CvROOT(bstate->bs_sv)                   opindex
 393 xcv_gv          *(SV**)&CvGV(bstate->bs_sv)             svindex
 394 xcv_file        CvFILE(bstate->bs_sv)                   pvindex
 395 xcv_depth       CvDEPTH(bstate->bs_sv)                  long
 396 xcv_padlist     *(SV**)&CvPADLIST(bstate->bs_sv)        svindex
 397 xcv_outside     *(SV**)&CvOUTSIDE(bstate->bs_sv)        svindex
 398 xcv_outside_seq CvOUTSIDE_SEQ(bstate->bs_sv)            U32
 399 xcv_flags       CvFLAGS(bstate->bs_sv)                  U16
 400 av_extend       bstate->bs_sv                           SSize_t         x
 401 av_pushx        bstate->bs_sv                           svindex         x
 402 av_push         bstate->bs_sv                           svindex         x
 403 xav_fill        AvFILLp(bstate->bs_sv)                  SSize_t
 404 xav_max         AvMAX(bstate->bs_sv)                    SSize_t
 405 xav_flags       AvFLAGS(bstate->bs_sv)                  U8
 406 xhv_riter       HvRITER(bstate->bs_sv)                  I32
 407 xhv_name        HvNAME(bstate->bs_sv)                   pvindex
 408 xhv_pmroot      *(OP**)&HvPMROOT(bstate->bs_sv)         opindex
 409 hv_store        bstate->bs_sv                           svindex         x
 410 sv_magic        bstate->bs_sv                           char            x
 411 mg_obj          SvMAGIC(bstate->bs_sv)->mg_obj          svindex
 412 mg_private      SvMAGIC(bstate->bs_sv)->mg_private      U16
 413 mg_flags        SvMAGIC(bstate->bs_sv)->mg_flags        U8
 414 mg_name         SvMAGIC(bstate->bs_sv)                  pvcontents      x
 415 mg_namex        SvMAGIC(bstate->bs_sv)                  svindex         x
 416 xmg_stash       *(SV**)&SvSTASH(bstate->bs_sv)          svindex
 417 gv_fetchpv      bstate->bs_sv                           strconst        x
 418 gv_fetchpvx     bstate->bs_sv                           strconst        x
 419 gv_stashpv      bstate->bs_sv                           strconst        x
 420 gv_stashpvx     bstate->bs_sv                           strconst        x
 421 gp_sv           GvSV(bstate->bs_sv)                     svindex
 422 gp_refcnt       GvREFCNT(bstate->bs_sv)                 U32
 423 gp_refcnt_add   GvREFCNT(bstate->bs_sv)                 I32             x
 424 gp_av           *(SV**)&GvAV(bstate->bs_sv)             svindex
 425 gp_hv           *(SV**)&GvHV(bstate->bs_sv)             svindex
 426 gp_cv           *(SV**)&GvCV(bstate->bs_sv)             svindex
 427 gp_file         GvFILE(bstate->bs_sv)                   pvindex
 428 gp_io           *(SV**)&GvIOp(bstate->bs_sv)            svindex
 429 gp_form         *(SV**)&GvFORM(bstate->bs_sv)           svindex
 430 gp_cvgen        GvCVGEN(bstate->bs_sv)                  U32
 431 gp_line         GvLINE(bstate->bs_sv)                   line_t
 432 gp_share        bstate->bs_sv                           svindex         x
 433 xgv_flags       GvFLAGS(bstate->bs_sv)                  U8
 434 op_next         PL_op->op_next                          opindex
 435 op_sibling      PL_op->op_sibling                       opindex
 436 op_ppaddr       PL_op->op_ppaddr                        strconst        x
 437 op_targ         PL_op->op_targ                          PADOFFSET
 438 op_type         PL_op                                   OPCODE          x
 439 op_seq          PL_op->op_seq                           U16
 440 op_flags        PL_op->op_flags                         U8
 441 op_private      PL_op->op_private                       U8
 442 op_first        cUNOP->op_first                         opindex
 443 op_last         cBINOP->op_last                         opindex
 444 op_other        cLOGOP->op_other                        opindex
 445 op_pmreplroot   cPMOP->op_pmreplroot                    opindex
 446 op_pmreplstart  cPMOP->op_pmreplstart                   opindex
 447 op_pmnext       *(OP**)&cPMOP->op_pmnext                opindex
 448 #ifdef USE_ITHREADS
 449 op_pmstashpv    cPMOP->op_pmstashpv                     pvindex
 450 op_pmreplrootpo (PADOFFSET)cPMOP->op_pmreplroot         PADOFFSET
 451 #else
 452 op_pmstash      *(SV**)&cPMOP->op_pmstash               svindex
 453 op_pmreplrootgv *(SV**)&cPMOP->op_pmreplroot            svindex
 454 #endif
 455 pregcomp        PL_op                                   pvcontents      x
 456 op_pmflags      cPMOP->op_pmflags                       U16
 457 op_pmpermflags  cPMOP->op_pmpermflags                   U16
 458 op_pmdynflags   cPMOP->op_pmdynflags                    U8
 459 op_sv           cSVOP->op_sv                            svindex
 460 op_padix        cPADOP->op_padix                        PADOFFSET
 461 op_pv           cPVOP->op_pv                            pvcontents
 462 op_pv_tr        cPVOP->op_pv                            op_tr_array
 463 op_redoop       cLOOP->op_redoop                        opindex
 464 op_nextop       cLOOP->op_nextop                        opindex
 465 op_lastop       cLOOP->op_lastop                        opindex
 466 cop_label       cCOP->cop_label                         pvindex
 467 #ifdef USE_ITHREADS
 468 cop_stashpv     cCOP                                    pvindex         x
 469 cop_file        cCOP                                    pvindex         x
 470 #else
 471 cop_stash       cCOP                                    svindex         x
 472 cop_filegv      cCOP                                    svindex         x
 473 #endif
 474 cop_seq         cCOP->cop_seq                           U32
 475 cop_arybase     cCOP->cop_arybase                       I32
 476 cop_line        cCOP->cop_line                          line_t
 477 cop_io          cCOP->cop_io                            svindex
 478 cop_warnings    cCOP->cop_warnings                      svindex
 479 main_start      PL_main_start                           opindex
 480 main_root       PL_main_root                            opindex
 481 main_cv         *(SV**)&PL_main_cv                      svindex
 482 curpad          PL_curpad                               svindex         x
 483 push_begin      PL_beginav                              svindex         x
 484 push_init       PL_initav                               svindex         x
 485 push_end        PL_endav                                svindex         x
 486 curstash        *(SV**)&PL_curstash                     svindex
 487 defstash        *(SV**)&PL_defstash                     svindex
 488 data            none                                    U8              x
 489 incav           *(SV**)&PL_incgv                        svindex
 490 load_glob       none                                    svindex         x
 491 #ifdef USE_ITHREADS
 492 regex_padav     *(SV**)&PL_regex_padav                  svindex
 493 #endif
 494 dowarn          PL_dowarn                               U8
 495 comppad_name    *(SV**)&PL_comppad_name                 svindex
 496 xgv_stash       *(SV**)&GvSTASH(bstate->bs_sv)          svindex
 497 signal          bstate->bs_sv                           strconst        x
 498 # to be removed
 499 formfeed        PL_formfeed                             svindex