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 #else
 104   const char *symprefix = ctx->mode != BUILD_elfasm ? "_" : "";
 105 #endif
 106   sprintf(name, "%s%s%s", symprefix, prefix, suffix);
 107   p = strchr(name, '@');
 108   if (p) {
 109     if (!LJ_64 && (ctx->mode == BUILD_coffasm || ctx->mode == BUILD_peobj))
 110       name[0] = '@';
 111     else
 112       *p = '\0';
 113   }
 114   p = (char *)malloc(strlen(name)+1);  /* MSVC doesn't like strdup. */
 115   strcpy(p, name);
 116   return p;
 117 }
 118
 119 #define NRELOCSYM       (sizeof(extnames)/sizeof(extnames[0])-1)
 120
 121 static int relocmap[NRELOCSYM];
 122
 123 /* Collect external relocations. */
 124 static int collect_reloc(BuildCtx *ctx, uint8_t *addr, int idx, int type)
 125 {
 126   if (ctx->nreloc >= BUILD_MAX_RELOC) {
 127     fprintf(stderr, "Error: too many relocations, increase BUILD_MAX_RELOC.\n");
 128     exit(1);
 129   }
 130   if (relocmap[idx] < 0) {
 131     relocmap[idx] = ctx->nrelocsym;
 132     ctx->relocsym[ctx->nrelocsym] = sym_decorate(ctx, "", extnames[idx]);
 133     ctx->nrelocsym++;
 134   }
 135   ctx->reloc[ctx->nreloc].ofs = (int32_t)(addr - ctx->code);
 136   ctx->reloc[ctx->nreloc].sym = relocmap[idx];
 137   ctx->reloc[ctx->nreloc].type = type;
 138   ctx->nreloc++;
 139   return 0;  /* Encode symbol offset of 0. */
 140 }
 141
 142 /* Naive insertion sort. Performance doesn't matter here. */
 143 static void sym_insert(BuildCtx *ctx, int32_t ofs,
 144                        const char *prefix, const char *suffix)
 145 {
 146   ptrdiff_t i = ctx->nsym++;
 147   while (i > 0) {
 148     if (ctx->sym[i-1].ofs <= ofs)
 149       break;
 150     ctx->sym[i] = ctx->sym[i-1];
 151     i--;
 152   }
 153   ctx->sym[i].ofs = ofs;
 154   ctx->sym[i].name = sym_decorate(ctx, prefix, suffix);
 155 }
 156
 157 /* Build the machine code. */
 158 static int build_code(BuildCtx *ctx)
 159 {
 160   int status;
 161   int i;
 162
 163   /* Initialize DynASM structures. */
 164   ctx->nglob = GLOB__MAX;
 165   ctx->glob = (void **)malloc(ctx->nglob*sizeof(void *));
 166   memset(ctx->glob, 0, ctx->nglob*sizeof(void *));
 167   ctx->nreloc = 0;
 168
 169   ctx->globnames = globnames;
 170   ctx->relocsym = (const char **)malloc(NRELOCSYM*sizeof(const char *));
 171   ctx->nrelocsym = 0;
 172   for (i = 0; i < (int)NRELOCSYM; i++) relocmap[i] = -1;
 173
 174   ctx->dasm_ident = DASM_IDENT;
 175   ctx->dasm_arch = DASM_ARCH;
 176
 177   dasm_init(Dst, DASM_MAXSECTION);
 178   dasm_setupglobal(Dst, ctx->glob, ctx->nglob);
 179   dasm_setup(Dst, build_actionlist);
 180
 181   /* Call arch-specific backend to emit the code. */
 182   ctx->npc = build_backend(ctx);
 183
 184   /* Finalize the code. */
 185   (void)dasm_checkstep(Dst, -1);
 186   if ((status = dasm_link(Dst, &ctx->codesz))) return status;
 187   ctx->code = (uint8_t *)malloc(ctx->codesz);
 188   if ((status = dasm_encode(Dst, (void *)ctx->code))) return status;
 189
 190   /* Allocate symbol table and bytecode offsets. */
 191   ctx->beginsym = sym_decorate(ctx, "", LABEL_PREFIX "vm_asm_begin");
 192   ctx->sym = (BuildSym *)malloc((ctx->npc+ctx->nglob+1)*sizeof(BuildSym));
 193   ctx->nsym = 0;
 194   ctx->bc_ofs = (int32_t *)malloc(ctx->npc*sizeof(int32_t));
 195
 196   /* Collect the opcodes (PC labels). */
 197   for (i = 0; i < ctx->npc; i++) {
 198     int32_t ofs = dasm_getpclabel(Dst, i);
 199     if (ofs < 0) return 0x22000000|i;
 200     ctx->bc_ofs[i] = ofs;
 201     if ((LJ_HASJIT ||
 202          !(i == BC_JFORI || i == BC_JFORL || i == BC_JITERL || i == BC_JLOOP ||
 203            i == BC_IFORL || i == BC_IITERL || i == BC_ILOOP)) &&
 204         (LJ_HASFFI || i != BC_KCDATA))
 205       sym_insert(ctx, ofs, LABEL_PREFIX_BC, bc_names[i]);
 206   }
 207
 208   /* Collect the globals (named labels). */
 209   for (i = 0; i < ctx->nglob; i++) {
 210     const char *gl = globnames[i];
 211     int len = (int)strlen(gl);
 212     if (!ctx->glob[i]) {
 213       fprintf(stderr, "Error: undefined global %s\n", gl);
 214       exit(2);
 215     }
 216     /* Skip the _Z symbols. */
 217     if (!(len >= 2 && gl[len-2] == '_' && gl[len-1] == 'Z'))
 218       sym_insert(ctx, (int32_t)((uint8_t *)(ctx->glob[i]) - ctx->code),
 219                  LABEL_PREFIX, globnames[i]);
 220   }
 221
 222   /* Close the address range. */
 223   sym_insert(ctx, (int32_t)ctx->codesz, "", "");
 224   ctx->nsym--;
 225
 226   dasm_free(Dst);
 227
 228   return 0;
 229 }
 230
 231 /* -- Generate VM enums --------------------------------------------------- */
 232
 233 const char *const bc_names[] = {
 234 #define BCNAME(name, ma, mb, mc, mt)       #name,
 235 BCDEF(BCNAME)
 236 #undef BCNAME
 237   NULL
 238 };
 239
 240 const char *const ir_names[] = {
 241 #define IRNAME(name, m, m1, m2) #name,
 242 IRDEF(IRNAME)
 243 #undef IRNAME
 244   NULL
 245 };
 246
 247 const char *const irt_names[] = {
 248 #define IRTNAME(name, size)     #name,
 249 IRTDEF(IRTNAME)
 250 #undef IRTNAME
 251   NULL
 252 };
 253
 254 const char *const irfpm_names[] = {
 255 #define FPMNAME(name)           #name,
 256 IRFPMDEF(FPMNAME)
 257 #undef FPMNAME
 258   NULL
 259 };
 260
 261 const char *const irfield_names[] = {
 262 #define FLNAME(name, ofs)       #name,
 263 IRFLDEF(FLNAME)
 264 #undef FLNAME
 265   NULL
 266 };
 267
 268 const char *const ircall_names[] = {
 269 #define IRCALLNAME(cond, name, nargs, kind, type, flags)        #name,
 270 IRCALLDEF(IRCALLNAME)
 271 #undef IRCALLNAME
 272   NULL
 273 };
 274
 275 static const char *const trace_errors[] = {
 276 #define TREDEF(name, msg)       msg,
 277 #include "lj_traceerr.h"
 278   NULL
 279 };
 280
 281 static const char *lower(char *buf, const char *s)
 282 {
 283   char *p = buf;
 284   while (*s) {
 285     *p++ = (*s >= 'A' && *s <= 'Z') ? *s+0x20 : *s;
 286     s++;
 287   }
 288   *p = '\0';
 289   return buf;
 290 }
 291
 292 /* Emit C source code for bytecode-related definitions. */
 293 static void emit_bcdef(BuildCtx *ctx)
 294 {
 295   int i;
 296   fprintf(ctx->fp, "/* This is a generated file. DO NOT EDIT! */\n\n");
 297   fprintf(ctx->fp, "LJ_DATADEF const uint16_t lj_bc_ofs[] = {\n");
 298   for (i = 0; i < ctx->npc; i++) {
 299     if (i != 0)
 300       fprintf(ctx->fp, ",\n");
 301     fprintf(ctx->fp, "%d", ctx->bc_ofs[i]);
 302   }
 303 }
 304
 305 /* Emit VM definitions as Lua code for debug modules. */
 306 static void emit_vmdef(BuildCtx *ctx)
 307 {
 308   char buf[80];
 309   int i;
 310   fprintf(ctx->fp, "-- This is a generated file. DO NOT EDIT!\n\n");
 311   fprintf(ctx->fp, "module(...)\n\n");
 312
 313   fprintf(ctx->fp, "bcnames = \"");
 314   for (i = 0; bc_names[i]; i++) fprintf(ctx->fp, "%-6s", bc_names[i]);
 315   fprintf(ctx->fp, "\"\n\n");
 316
 317   fprintf(ctx->fp, "irnames = \"");
 318   for (i = 0; ir_names[i]; i++) fprintf(ctx->fp, "%-6s", ir_names[i]);
 319   fprintf(ctx->fp, "\"\n\n");
 320
 321   fprintf(ctx->fp, "irfpm = { [0]=");
 322   for (i = 0; irfpm_names[i]; i++)
 323     fprintf(ctx->fp, "\"%s\", ", lower(buf, irfpm_names[i]));
 324   fprintf(ctx->fp, "}\n\n");
 325
 326   fprintf(ctx->fp, "irfield = { [0]=");
 327   for (i = 0; irfield_names[i]; i++) {
 328     char *p;
 329     lower(buf, irfield_names[i]);
 330     p = strchr(buf, '_');
 331     if (p) *p = '.';
 332     fprintf(ctx->fp, "\"%s\", ", buf);
 333   }
 334   fprintf(ctx->fp, "}\n\n");
 335
 336   fprintf(ctx->fp, "ircall = {\n[0]=");
 337   for (i = 0; ircall_names[i]; i++)
 338     fprintf(ctx->fp, "\"%s\",\n", ircall_names[i]);
 339   fprintf(ctx->fp, "}\n\n");
 340
 341   fprintf(ctx->fp, "traceerr = {\n[0]=");
 342   for (i = 0; trace_errors[i]; i++)
 343     fprintf(ctx->fp, "\"%s\",\n", trace_errors[i]);
 344   fprintf(ctx->fp, "}\n\n");
 345 }
 346
 347 /* -- Argument parsing ---------------------------------------------------- */
 348
 349 /* Build mode names. */
 350 static const char *const modenames[] = {
 351 #define BUILDNAME(name)         #name,
 352 BUILDDEF(BUILDNAME)
 353 #undef BUILDNAME
 354   NULL
 355 };
 356
 357 /* Print usage information and exit. */
 358 static void usage(void)
 359 {
 360   int i;
 361   fprintf(stderr, LUAJIT_VERSION " VM builder.\n");
 362   fprintf(stderr, LUAJIT_COPYRIGHT ", " LUAJIT_URL "\n");
 363   fprintf(stderr, "Target architecture: " LJ_ARCH_NAME "\n\n");
 364   fprintf(stderr, "Usage: buildvm -m mode [-o outfile] [infiles...]\n\n");
 365   fprintf(stderr, "Available modes:\n");
 366   for (i = 0; i < BUILD__MAX; i++)
 367     fprintf(stderr, "  %s\n", modenames[i]);
 368   exit(1);
 369 }
 370
 371 /* Parse the output mode name. */
 372 static BuildMode parsemode(const char *mode)
 373 {
 374   int i;
 375   for (i = 0; modenames[i]; i++)
 376     if (!strcmp(mode, modenames[i]))
 377       return (BuildMode)i;
 378   usage();
 379   return (BuildMode)-1;
 380 }
 381
 382 /* Parse arguments. */
 383 static void parseargs(BuildCtx *ctx, char **argv)
 384 {
 385   const char *a;
 386   int i;
 387   ctx->mode = (BuildMode)-1;
 388   ctx->outname = "-";
 389   for (i = 1; (a = argv[i]) != NULL; i++) {
 390     if (a[0] != '-')
 391       break;
 392     switch (a[1]) {
 393     case '-':
 394       if (a[2]) goto err;
 395       i++;
 396       goto ok;
 397     case '\0':
 398       goto ok;
 399     case 'm':
 400       i++;
 401       if (a[2] || argv[i] == NULL) goto err;
 402       ctx->mode = parsemode(argv[i]);
 403       break;
 404     case 'o':
 405       i++;
 406       if (a[2] || argv[i] == NULL) goto err;
 407       ctx->outname = argv[i];
 408       break;
 409     default: err:
 410       usage();
 411       break;
 412     }
 413   }
 414 ok:
 415   ctx->args = argv+i;
 416   if (ctx->mode == (BuildMode)-1) goto err;
 417 }
 418
 419 int main(int argc, char **argv)
 420 {
 421   BuildCtx ctx_;
 422   BuildCtx *ctx = &ctx_;
 423   int status, binmode;
 424
 425   if (sizeof(void *) != 4*LJ_32+8*LJ_64) {
 426     fprintf(stderr,"Error: pointer size mismatch in cross-build.\n");
 427     fprintf(stderr,"Try: make HOST_CC=\"gcc -m32\" CROSS=...\n\n");
 428     return 1;
 429   }
 430
 431   UNUSED(argc);
 432   parseargs(ctx, argv);
 433
 434   if ((status = build_code(ctx))) {
 435     fprintf(stderr,"Error: DASM error %08x\n", status);
 436     return 1;
 437   }
 438
 439   switch (ctx->mode) {
 440   case BUILD_peobj:
 441   case BUILD_raw:
 442     binmode = 1;
 443     break;
 444   default:
 445     binmode = 0;
 446     break;
 447   }
 448
 449   if (ctx->outname[0] == '-' && ctx->outname[1] == '\0') {
 450     ctx->fp = stdout;
 451 #if defined(_WIN32)
 452     if (binmode)
 453       _setmode(_fileno(stdout), _O_BINARY);  /* Yuck. */
 454 #endif
 455   } else if (!(ctx->fp = fopen(ctx->outname, binmode ? "wb" : "w"))) {
 456     fprintf(stderr, "Error: cannot open output file '%s': %s\n",
 457             ctx->outname, strerror(errno));
 458     exit(1);
 459   }
 460
 461   switch (ctx->mode) {
 462   case BUILD_elfasm:
 463   case BUILD_coffasm:
 464   case BUILD_machasm:
 465     emit_asm(ctx);
 466     emit_asm_debug(ctx);
 467     break;
 468   case BUILD_peobj:
 469     emit_peobj(ctx);
 470     break;
 471   case BUILD_raw:
 472     emit_raw(ctx);
 473     break;
 474   case BUILD_bcdef:
 475     emit_bcdef(ctx);
 476     emit_lib(ctx);
 477     break;
 478   case BUILD_vmdef:
 479     emit_vmdef(ctx);
 480     emit_lib(ctx);
 481     break;
 482   case BUILD_ffdef:
 483   case BUILD_libdef:
 484   case BUILD_recdef:
 485     emit_lib(ctx);
 486     break;
 487   case BUILD_folddef:
 488     emit_fold(ctx);
 489     break;
 490   default:
 491     break;
 492   }
 493
 494   fflush(ctx->fp);
 495   if (ferror(ctx->fp)) {
 496     fprintf(stderr, "Error: cannot write to output file: %s\n",
 497             strerror(errno));
 498     exit(1);
 499   }
 500   fclose(ctx->fp);
 501
 502   return 0;
 503 }
 504