src/host/buildvm.c

   1 /*
   2 ** LuaJIT VM builder.
   3 ** Copyright (C) 2005-2012 Mike Pall. See Copyright Notice in luajit.h
   4 **
   5 ** This is a tool to build the hand-tuned assembler code required for
   6 ** LuaJIT's bytecode interpreter. It supports a variety of output formats
   7 ** to feed different toolchains (see usage() below).
   8 **
   9 ** This tool is not particularly optimized because it's only used while
  10 ** _building_ LuaJIT. There's no point in distributing or installing it.
  11 ** Only the object code generated by this tool is linked into LuaJIT.
  12 **
  13 ** Caveat: some memory is not free'd, error handling is lazy.
  14 ** It's a one-shot tool -- any effort fixing this would be wasted.
  15 */
  16
  17 #include "buildvm.h"
  18 #include "lj_obj.h"
  19 #include "lj_gc.h"
  20 #include "lj_bc.h"
  21 #include "lj_ir.h"
  22 #include "lj_ircall.h"
  23 #include "lj_frame.h"
  24 #include "lj_dispatch.h"
  25 #if LJ_HASFFI
  26 #include "lj_ctype.h"
  27 #include "lj_ccall.h"
  28 #endif
  29 #include "luajit.h"
  30
  31 #if defined(_WIN32)
  32 #include <fcntl.h>
  33 #include <io.h>
  34 #endif
  35
  36 /* ------------------------------------------------------------------------ */
  37
  38 /* DynASM glue definitions. */
  39 #define Dst             ctx
  40 #define Dst_DECL        BuildCtx *ctx
  41 #define Dst_REF         (ctx->D)
  42 #define DASM_CHECKS     1
  43
  44 #include "../dynasm/dasm_proto.h"
  45
  46 /* Glue macros for DynASM. */
  47 static int collect_reloc(BuildCtx *ctx, uint8_t *addr, int idx, int type);
  48
  49 #define DASM_EXTERN(ctx, addr, idx, type) \
  50   collect_reloc(ctx, addr, idx, type)
  51
  52 /* ------------------------------------------------------------------------ */
  53
  54 /* Avoid trouble if cross-compiling for an x86 target. Speed doesn't matter. */
  55 #define DASM_ALIGNED_WRITES     1
  56
  57 /* Embed architecture-specific DynASM encoder. */
  58 #if LJ_TARGET_X86ORX64
  59 #include "../dynasm/dasm_x86.h"
  60 #elif LJ_TARGET_ARM
  61 #include "../dynasm/dasm_arm.h"
  62 #elif LJ_TARGET_PPC
  63 #include "../dynasm/dasm_ppc.h"
  64 #elif LJ_TARGET_PPCSPE
  65 #include "../dynasm/dasm_ppc.h"
  66 #elif LJ_TARGET_MIPS
  67 #include "../dynasm/dasm_mips.h"
  68 #else
  69 #error "No support for this architecture (yet)"
  70 #endif
  71
  72 /* Embed generated architecture-specific backend. */
  73 #include "buildvm_arch.h"
  74
  75 /* ------------------------------------------------------------------------ */
  76
  77 void owrite(BuildCtx *ctx, const void *ptr, size_t sz)
  78 {
  79   if (fwrite(ptr, 1, sz, ctx->fp) != sz) {
  80     fprintf(stderr, "Error: cannot write to output file: %s\n",
  81             strerror(errno));
  82     exit(1);
  83   }
  84 }
  85
  86 /* ------------------------------------------------------------------------ */
  87
  88 /* Emit code as raw bytes. Only used for DynASM debugging. */
  89 static void emit_raw(BuildCtx *ctx)
  90 {
  91   owrite(ctx, ctx->code, ctx->codesz);
  92 }
  93
  94 /* -- Build machine code -------------------------------------------------- */
  95
  96 static const char *sym_decorate(BuildCtx *ctx,
  97                                 const char *prefix, const char *suffix)
  98 {
  99   char name[256];
 100   char *p;
 101 #if LJ_64
 102   const char *symprefix = ctx->mode == BUILD_machasm ? "_" : "";
 103 #elif LJ_TARGET_XBOX360
 104   const char *symprefix = "";
 105 #else
 106   const char *symprefix = ctx->mode != BUILD_elfasm ? "_" : "";
 107 #endif
 108   sprintf(name, "%s%s%s", symprefix, prefix, suffix);
 109   p = strchr(name, '@');
 110   if (p) {
 111     if (!LJ_64 && (ctx->mode == BUILD_coffasm || ctx->mode == BUILD_peobj))
 112       name[0] = '@';
 113     else
 114       *p = '\0';
 115   }
 116   p = (char *)malloc(strlen(name)+1);  /* MSVC doesn't like strdup. */
 117   strcpy(p, name);
 118   return p;
 119 }
 120
 121 #define NRELOCSYM       (sizeof(extnames)/sizeof(extnames[0])-1)
 122
 123 static int relocmap[NRELOCSYM];
 124
 125 /* Collect external relocations. */
 126 static int collect_reloc(BuildCtx *ctx, uint8_t *addr, int idx, int type)
 127 {
 128   if (ctx->nreloc >= BUILD_MAX_RELOC) {
 129     fprintf(stderr, "Error: too many relocations, increase BUILD_MAX_RELOC.\n");
 130     exit(1);
 131   }
 132   if (relocmap[idx] < 0) {
 133     relocmap[idx] = ctx->nrelocsym;
 134     ctx->relocsym[ctx->nrelocsym] = sym_decorate(ctx, "", extnames[idx]);
 135     ctx->nrelocsym++;
 136   }
 137   ctx->reloc[ctx->nreloc].ofs = (int32_t)(addr - ctx->code);
 138   ctx->reloc[ctx->nreloc].sym = relocmap[idx];
 139   ctx->reloc[ctx->nreloc].type = type;
 140   ctx->nreloc++;
 141 #if LJ_TARGET_XBOX360
 142   return (int)(ctx->code - addr) + 4;  /* Encode symbol offset of .text. */
 143 #else
 144   return 0;  /* Encode symbol offset of 0. */
 145 #endif
 146 }
 147
 148 /* Naive insertion sort. Performance doesn't matter here. */
 149 static void sym_insert(BuildCtx *ctx, int32_t ofs,
 150                        const char *prefix, const char *suffix)
 151 {
 152   ptrdiff_t i = ctx->nsym++;
 153   while (i > 0) {
 154     if (ctx->sym[i-1].ofs <= ofs)
 155       break;
 156     ctx->sym[i] = ctx->sym[i-1];
 157     i--;
 158   }
 159   ctx->sym[i].ofs = ofs;
 160   ctx->sym[i].name = sym_decorate(ctx, prefix, suffix);
 161 }
 162
 163 /* Build the machine code. */
 164 static int build_code(BuildCtx *ctx)
 165 {
 166   int status;
 167   int i;
 168
 169   /* Initialize DynASM structures. */
 170   ctx->nglob = GLOB__MAX;
 171   ctx->glob = (void **)malloc(ctx->nglob*sizeof(void *));
 172   memset(ctx->glob, 0, ctx->nglob*sizeof(void *));
 173   ctx->nreloc = 0;
 174
 175   ctx->globnames = globnames;
 176   ctx->relocsym = (const char **)malloc(NRELOCSYM*sizeof(const char *));
 177   ctx->nrelocsym = 0;
 178   for (i = 0; i < (int)NRELOCSYM; i++) relocmap[i] = -1;
 179
 180   ctx->dasm_ident = DASM_IDENT;
 181   ctx->dasm_arch = DASM_ARCH;
 182
 183   dasm_init(Dst, DASM_MAXSECTION);
 184   dasm_setupglobal(Dst, ctx->glob, ctx->nglob);
 185   dasm_setup(Dst, build_actionlist);
 186
 187   /* Call arch-specific backend to emit the code. */
 188   ctx->npc = build_backend(ctx);
 189
 190   /* Finalize the code. */
 191   (void)dasm_checkstep(Dst, -1);
 192   if ((status = dasm_link(Dst, &ctx->codesz))) return status;
 193   ctx->code = (uint8_t *)malloc(ctx->codesz);
 194   if ((status = dasm_encode(Dst, (void *)ctx->code))) return status;
 195
 196   /* Allocate symbol table and bytecode offsets. */
 197   ctx->beginsym = sym_decorate(ctx, "", LABEL_PREFIX "vm_asm_begin");
 198   ctx->sym = (BuildSym *)malloc((ctx->npc+ctx->nglob+1)*sizeof(BuildSym));
 199   ctx->nsym = 0;
 200   ctx->bc_ofs = (int32_t *)malloc(ctx->npc*sizeof(int32_t));
 201
 202   /* Collect the opcodes (PC labels). */
 203   for (i = 0; i < ctx->npc; i++) {
 204     int32_t ofs = dasm_getpclabel(Dst, i);
 205     if (ofs < 0) return 0x22000000|i;
 206     ctx->bc_ofs[i] = ofs;
 207     if ((LJ_HASJIT ||
 208          !(i == BC_JFORI || i == BC_JFORL || i == BC_JITERL || i == BC_JLOOP ||
 209            i == BC_IFORL || i == BC_IITERL || i == BC_ILOOP)) &&
 210         (LJ_HASFFI || i != BC_KCDATA))
 211       sym_insert(ctx, ofs, LABEL_PREFIX_BC, bc_names[i]);
 212   }
 213
 214   /* Collect the globals (named labels). */
 215   for (i = 0; i < ctx->nglob; i++) {
 216     const char *gl = globnames[i];
 217     int len = (int)strlen(gl);
 218     if (!ctx->glob[i]) {
 219       fprintf(stderr, "Error: undefined global %s\n", gl);
 220       exit(2);
 221     }
 222     /* Skip the _Z symbols. */
 223     if (!(len >= 2 && gl[len-2] == '_' && gl[len-1] == 'Z'))
 224       sym_insert(ctx, (int32_t)((uint8_t *)(ctx->glob[i]) - ctx->code),
 225                  LABEL_PREFIX, globnames[i]);
 226   }
 227
 228   /* Close the address range. */
 229   sym_insert(ctx, (int32_t)ctx->codesz, "", "");
 230   ctx->nsym--;
 231
 232   dasm_free(Dst);
 233
 234   return 0;
 235 }
 236
 237 /* -- Generate VM enums --------------------------------------------------- */
 238
 239 const char *const bc_names[] = {
 240 #define BCNAME(name, ma, mb, mc, mt)       #name,
 241 BCDEF(BCNAME)
 242 #undef BCNAME
 243   NULL
 244 };
 245
 246 const char *const ir_names[] = {
 247 #define IRNAME(name, m, m1, m2) #name,
 248 IRDEF(IRNAME)
 249 #undef IRNAME
 250   NULL
 251 };
 252
 253 const char *const irt_names[] = {
 254 #define IRTNAME(name, size)     #name,
 255 IRTDEF(IRTNAME)
 256 #undef IRTNAME
 257   NULL
 258 };
 259
 260 const char *const irfpm_names[] = {
 261 #define FPMNAME(name)           #name,
 262 IRFPMDEF(FPMNAME)
 263 #undef FPMNAME
 264   NULL
 265 };
 266
 267 const char *const irfield_names[] = {
 268 #define FLNAME(name, ofs)       #name,
 269 IRFLDEF(FLNAME)
 270 #undef FLNAME
 271   NULL
 272 };
 273
 274 const char *const ircall_names[] = {
 275 #define IRCALLNAME(cond, name, nargs, kind, type, flags)        #name,
 276 IRCALLDEF(IRCALLNAME)
 277 #undef IRCALLNAME
 278   NULL
 279 };
 280
 281 static const char *const trace_errors[] = {
 282 #define TREDEF(name, msg)       msg,
 283 #include "lj_traceerr.h"
 284   NULL
 285 };
 286
 287 static const char *lower(char *buf, const char *s)
 288 {
 289   char *p = buf;
 290   while (*s) {
 291     *p++ = (*s >= 'A' && *s <= 'Z') ? *s+0x20 : *s;
 292     s++;
 293   }
 294   *p = '\0';
 295   return buf;
 296 }
 297
 298 /* Emit C source code for bytecode-related definitions. */
 299 static void emit_bcdef(BuildCtx *ctx)
 300 {
 301   int i;
 302   fprintf(ctx->fp, "/* This is a generated file. DO NOT EDIT! */\n\n");
 303   fprintf(ctx->fp, "LJ_DATADEF const uint16_t lj_bc_ofs[] = {\n");
 304   for (i = 0; i < ctx->npc; i++) {
 305     if (i != 0)
 306       fprintf(ctx->fp, ",\n");
 307     fprintf(ctx->fp, "%d", ctx->bc_ofs[i]);
 308   }
 309 }
 310
 311 /* Emit VM definitions as Lua code for debug modules. */
 312 static void emit_vmdef(BuildCtx *ctx)
 313 {
 314   char buf[80];
 315   int i;
 316   fprintf(ctx->fp, "-- This is a generated file. DO NOT EDIT!\n\n");
 317   fprintf(ctx->fp, "module(...)\n\n");
 318
 319   fprintf(ctx->fp, "bcnames = \"");
 320   for (i = 0; bc_names[i]; i++) fprintf(ctx->fp, "%-6s", bc_names[i]);
 321   fprintf(ctx->fp, "\"\n\n");
 322
 323   fprintf(ctx->fp, "irnames = \"");
 324   for (i = 0; ir_names[i]; i++) fprintf(ctx->fp, "%-6s", ir_names[i]);
 325   fprintf(ctx->fp, "\"\n\n");
 326
 327   fprintf(ctx->fp, "irfpm = { [0]=");
 328   for (i = 0; irfpm_names[i]; i++)
 329     fprintf(ctx->fp, "\"%s\", ", lower(buf, irfpm_names[i]));
 330   fprintf(ctx->fp, "}\n\n");
 331
 332   fprintf(ctx->fp, "irfield = { [0]=");
 333   for (i = 0; irfield_names[i]; i++) {
 334     char *p;
 335     lower(buf, irfield_names[i]);
 336     p = strchr(buf, '_');
 337     if (p) *p = '.';
 338     fprintf(ctx->fp, "\"%s\", ", buf);
 339   }
 340   fprintf(ctx->fp, "}\n\n");
 341
 342   fprintf(ctx->fp, "ircall = {\n[0]=");
 343   for (i = 0; ircall_names[i]; i++)
 344     fprintf(ctx->fp, "\"%s\",\n", ircall_names[i]);
 345   fprintf(ctx->fp, "}\n\n");
 346
 347   fprintf(ctx->fp, "traceerr = {\n[0]=");
 348   for (i = 0; trace_errors[i]; i++)
 349     fprintf(ctx->fp, "\"%s\",\n", trace_errors[i]);
 350   fprintf(ctx->fp, "}\n\n");
 351 }
 352
 353 /* -- Argument parsing ---------------------------------------------------- */
 354
 355 /* Build mode names. */
 356 static const char *const modenames[] = {
 357 #define BUILDNAME(name)         #name,
 358 BUILDDEF(BUILDNAME)
 359 #undef BUILDNAME
 360   NULL
 361 };
 362
 363 /* Print usage information and exit. */
 364 static void usage(void)
 365 {
 366   int i;
 367   fprintf(stderr, LUAJIT_VERSION " VM builder.\n");
 368   fprintf(stderr, LUAJIT_COPYRIGHT ", " LUAJIT_URL "\n");
 369   fprintf(stderr, "Target architecture: " LJ_ARCH_NAME "\n\n");
 370   fprintf(stderr, "Usage: buildvm -m mode [-o outfile] [infiles...]\n\n");
 371   fprintf(stderr, "Available modes:\n");
 372   for (i = 0; i < BUILD__MAX; i++)
 373     fprintf(stderr, "  %s\n", modenames[i]);
 374   exit(1);
 375 }
 376
 377 /* Parse the output mode name. */
 378 static BuildMode parsemode(const char *mode)
 379 {
 380   int i;
 381   for (i = 0; modenames[i]; i++)
 382     if (!strcmp(mode, modenames[i]))
 383       return (BuildMode)i;
 384   usage();
 385   return (BuildMode)-1;
 386 }
 387
 388 /* Parse arguments. */
 389 static void parseargs(BuildCtx *ctx, char **argv)
 390 {
 391   const char *a;
 392   int i;
 393   ctx->mode = (BuildMode)-1;
 394   ctx->outname = "-";
 395   for (i = 1; (a = argv[i]) != NULL; i++) {
 396     if (a[0] != '-')
 397       break;
 398     switch (a[1]) {
 399     case '-':
 400       if (a[2]) goto err;
 401       i++;
 402       goto ok;
 403     case '\0':
 404       goto ok;
 405     case 'm':
 406       i++;
 407       if (a[2] || argv[i] == NULL) goto err;
 408       ctx->mode = parsemode(argv[i]);
 409       break;
 410     case 'o':
 411       i++;
 412       if (a[2] || argv[i] == NULL) goto err;
 413       ctx->outname = argv[i];
 414       break;
 415     default: err:
 416       usage();
 417       break;
 418     }
 419   }
 420 ok:
 421   ctx->args = argv+i;
 422   if (ctx->mode == (BuildMode)-1) goto err;
 423 }
 424
 425 int main(int argc, char **argv)
 426 {
 427   BuildCtx ctx_;
 428   BuildCtx *ctx = &ctx_;
 429   int status, binmode;
 430
 431   if (sizeof(void *) != 4*LJ_32+8*LJ_64) {
 432     fprintf(stderr,"Error: pointer size mismatch in cross-build.\n");
 433     fprintf(stderr,"Try: make HOST_CC=\"gcc -m32\" CROSS=...\n\n");
 434     return 1;
 435   }
 436
 437   UNUSED(argc);
 438   parseargs(ctx, argv);
 439
 440   if ((status = build_code(ctx))) {
 441     fprintf(stderr,"Error: DASM error %08x\n", status);
 442     return 1;
 443   }
 444
 445   switch (ctx->mode) {
 446   case BUILD_peobj:
 447   case BUILD_raw:
 448     binmode = 1;
 449     break;
 450   default:
 451     binmode = 0;
 452     break;
 453   }
 454
 455   if (ctx->outname[0] == '-' && ctx->outname[1] == '\0') {
 456     ctx->fp = stdout;
 457 #if defined(_WIN32)
 458     if (binmode)
 459       _setmode(_fileno(stdout), _O_BINARY);  /* Yuck. */
 460 #endif
 461   } else if (!(ctx->fp = fopen(ctx->outname, binmode ? "wb" : "w"))) {
 462     fprintf(stderr, "Error: cannot open output file '%s': %s\n",
 463             ctx->outname, strerror(errno));
 464     exit(1);
 465   }
 466
 467   switch (ctx->mode) {
 468   case BUILD_elfasm:
 469   case BUILD_coffasm:
 470   case BUILD_machasm:
 471     emit_asm(ctx);
 472     emit_asm_debug(ctx);
 473     break;
 474   case BUILD_peobj:
 475     emit_peobj(ctx);
 476     break;
 477   case BUILD_raw:
 478     emit_raw(ctx);
 479     break;
 480   case BUILD_bcdef:
 481     emit_bcdef(ctx);
 482     emit_lib(ctx);
 483     break;
 484   case BUILD_vmdef:
 485     emit_vmdef(ctx);
 486     emit_lib(ctx);
 487     break;
 488   case BUILD_ffdef:
 489   case BUILD_libdef:
 490   case BUILD_recdef:
 491     emit_lib(ctx);
 492     break;
 493   case BUILD_folddef:
 494     emit_fold(ctx);
 495     break;
 496   default:
 497     break;
 498   }
 499
 500   fflush(ctx->fp);
 501   if (ferror(ctx->fp)) {
 502     fprintf(stderr, "Error: cannot write to output file: %s\n",
 503             strerror(errno));
 504     exit(1);
 505   }
 506   fclose(ctx->fp);
 507
 508   return 0;
 509 }
 510