src/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 and backend. */
  58 #if LJ_TARGET_X86
  59 #include "../dynasm/dasm_x86.h"
  60 #include "buildvm_x86.h"
  61 #elif LJ_TARGET_X64
  62 #include "../dynasm/dasm_x86.h"
  63 #if LJ_ABI_WIN
  64 #include "buildvm_x64win.h"
  65 #else
  66 #include "buildvm_x64.h"
  67 #endif
  68 #elif LJ_TARGET_ARM
  69 #include "../dynasm/dasm_arm.h"
  70 #include "buildvm_arm.h"
  71 #elif LJ_TARGET_PPC
  72 #include "../dynasm/dasm_ppc.h"
  73 #include "buildvm_ppc.h"
  74 #elif LJ_TARGET_PPCSPE
  75 #include "../dynasm/dasm_ppc.h"
  76 #include "buildvm_ppcspe.h"
  77 #elif LJ_TARGET_MIPS
  78 #include "../dynasm/dasm_mips.h"
  79 #include "buildvm_mips.h"
  80 #else
  81 #error "No support for this architecture (yet)"
  82 #endif
  83
  84 /* ------------------------------------------------------------------------ */
  85
  86 void owrite(BuildCtx *ctx, const void *ptr, size_t sz)
  87 {
  88   if (fwrite(ptr, 1, sz, ctx->fp) != sz) {
  89     fprintf(stderr, "Error: cannot write to output file: %s\n",
  90             strerror(errno));
  91     exit(1);
  92   }
  93 }
  94
  95 /* ------------------------------------------------------------------------ */
  96
  97 /* Emit code as raw bytes. Only used for DynASM debugging. */
  98 static void emit_raw(BuildCtx *ctx)
  99 {
 100   owrite(ctx, ctx->code, ctx->codesz);
 101 }
 102
 103 /* -- Build machine code -------------------------------------------------- */
 104
 105 static const char *sym_decorate(BuildCtx *ctx,
 106                                 const char *prefix, const char *suffix)
 107 {
 108   char name[256];
 109   char *p;
 110 #if LJ_64
 111   const char *symprefix = ctx->mode == BUILD_machasm ? "_" : "";
 112 #else
 113   const char *symprefix = ctx->mode != BUILD_elfasm ? "_" : "";
 114 #endif
 115   sprintf(name, "%s%s%s", symprefix, prefix, suffix);
 116   p = strchr(name, '@');
 117   if (p) {
 118     if (!LJ_64 && (ctx->mode == BUILD_coffasm || ctx->mode == BUILD_peobj))
 119       name[0] = '@';
 120     else
 121       *p = '\0';
 122   }
 123   p = (char *)malloc(strlen(name)+1);  /* MSVC doesn't like strdup. */
 124   strcpy(p, name);
 125   return p;
 126 }
 127
 128 #define NRELOCSYM       (sizeof(extnames)/sizeof(extnames[0])-1)
 129
 130 static int relocmap[NRELOCSYM];
 131
 132 /* Collect external relocations. */
 133 static int collect_reloc(BuildCtx *ctx, uint8_t *addr, int idx, int type)
 134 {
 135   if (ctx->nreloc >= BUILD_MAX_RELOC) {
 136     fprintf(stderr, "Error: too many relocations, increase BUILD_MAX_RELOC.\n");
 137     exit(1);
 138   }
 139   if (relocmap[idx] < 0) {
 140     relocmap[idx] = ctx->nrelocsym;
 141     ctx->relocsym[ctx->nrelocsym] = sym_decorate(ctx, "", extnames[idx]);
 142     ctx->nrelocsym++;
 143   }
 144   ctx->reloc[ctx->nreloc].ofs = (int32_t)(addr - ctx->code);
 145   ctx->reloc[ctx->nreloc].sym = relocmap[idx];
 146   ctx->reloc[ctx->nreloc].type = type;
 147   ctx->nreloc++;
 148   return 0;  /* Encode symbol offset of 0. */
 149 }
 150
 151 /* Naive insertion sort. Performance doesn't matter here. */
 152 static void sym_insert(BuildCtx *ctx, int32_t ofs,
 153                        const char *prefix, const char *suffix)
 154 {
 155   ptrdiff_t i = ctx->nsym++;
 156   while (i > 0) {
 157     if (ctx->sym[i-1].ofs <= ofs)
 158       break;
 159     ctx->sym[i] = ctx->sym[i-1];
 160     i--;
 161   }
 162   ctx->sym[i].ofs = ofs;
 163   ctx->sym[i].name = sym_decorate(ctx, prefix, suffix);
 164 }
 165
 166 /* Build the machine code. */
 167 static int build_code(BuildCtx *ctx)
 168 {
 169   int status;
 170   int i;
 171
 172   /* Initialize DynASM structures. */
 173   ctx->nglob = GLOB__MAX;
 174   ctx->glob = (void **)malloc(ctx->nglob*sizeof(void *));
 175   memset(ctx->glob, 0, ctx->nglob*sizeof(void *));
 176   ctx->nreloc = 0;
 177
 178   ctx->globnames = globnames;
 179   ctx->relocsym = (const char **)malloc(NRELOCSYM*sizeof(const char *));
 180   ctx->nrelocsym = 0;
 181   for (i = 0; i < (int)NRELOCSYM; i++) relocmap[i] = -1;
 182
 183   ctx->dasm_ident = DASM_IDENT;
 184   ctx->dasm_arch = DASM_ARCH;
 185
 186   dasm_init(Dst, DASM_MAXSECTION);
 187   dasm_setupglobal(Dst, ctx->glob, ctx->nglob);
 188   dasm_setup(Dst, build_actionlist);
 189
 190   /* Call arch-specific backend to emit the code. */
 191   ctx->npc = build_backend(ctx);
 192
 193   /* Finalize the code. */
 194   (void)dasm_checkstep(Dst, -1);
 195   if ((status = dasm_link(Dst, &ctx->codesz))) return status;
 196   ctx->code = (uint8_t *)malloc(ctx->codesz);
 197   if ((status = dasm_encode(Dst, (void *)ctx->code))) return status;
 198
 199   /* Allocate symbol table and bytecode offsets. */
 200   ctx->beginsym = sym_decorate(ctx, "", LABEL_PREFIX "vm_asm_begin");
 201   ctx->sym = (BuildSym *)malloc((ctx->npc+ctx->nglob+1)*sizeof(BuildSym));
 202   ctx->nsym = 0;
 203   ctx->bc_ofs = (int32_t *)malloc(ctx->npc*sizeof(int32_t));
 204
 205   /* Collect the opcodes (PC labels). */
 206   for (i = 0; i < ctx->npc; i++) {
 207     int32_t ofs = dasm_getpclabel(Dst, i);
 208     if (ofs < 0) return 0x22000000|i;
 209     ctx->bc_ofs[i] = ofs;
 210     if ((LJ_HASJIT ||
 211          !(i == BC_JFORI || i == BC_JFORL || i == BC_JITERL || i == BC_JLOOP ||
 212            i == BC_IFORL || i == BC_IITERL || i == BC_ILOOP)) &&
 213         (LJ_HASFFI || i != BC_KCDATA))
 214       sym_insert(ctx, ofs, LABEL_PREFIX_BC, bc_names[i]);
 215   }
 216
 217   /* Collect the globals (named labels). */
 218   for (i = 0; i < ctx->nglob; i++) {
 219     const char *gl = globnames[i];
 220     int len = (int)strlen(gl);
 221     if (!ctx->glob[i]) {
 222       fprintf(stderr, "Error: undefined global %s\n", gl);
 223       exit(2);
 224     }
 225     /* Skip the _Z symbols. */
 226     if (!(len >= 2 && gl[len-2] == '_' && gl[len-1] == 'Z'))
 227       sym_insert(ctx, (int32_t)((uint8_t *)(ctx->glob[i]) - ctx->code),
 228                  LABEL_PREFIX, globnames[i]);
 229   }
 230
 231   /* Close the address range. */
 232   sym_insert(ctx, (int32_t)ctx->codesz, "", "");
 233   ctx->nsym--;
 234
 235   dasm_free(Dst);
 236
 237   return 0;
 238 }
 239
 240 /* -- Generate VM enums --------------------------------------------------- */
 241
 242 const char *const bc_names[] = {
 243 #define BCNAME(name, ma, mb, mc, mt)       #name,
 244 BCDEF(BCNAME)
 245 #undef BCNAME
 246   NULL
 247 };
 248
 249 const char *const ir_names[] = {
 250 #define IRNAME(name, m, m1, m2) #name,
 251 IRDEF(IRNAME)
 252 #undef IRNAME
 253   NULL
 254 };
 255
 256 const char *const irt_names[] = {
 257 #define IRTNAME(name)   #name,
 258 IRTDEF(IRTNAME)
 259 #undef IRTNAME
 260   NULL
 261 };
 262
 263 const char *const irfpm_names[] = {
 264 #define FPMNAME(name)           #name,
 265 IRFPMDEF(FPMNAME)
 266 #undef FPMNAME
 267   NULL
 268 };
 269
 270 const char *const irfield_names[] = {
 271 #define FLNAME(name, ofs)       #name,
 272 IRFLDEF(FLNAME)
 273 #undef FLNAME
 274   NULL
 275 };
 276
 277 const char *const ircall_names[] = {
 278 #define IRCALLNAME(cond, name, nargs, kind, type, flags)        #name,
 279 IRCALLDEF(IRCALLNAME)
 280 #undef IRCALLNAME
 281   NULL
 282 };
 283
 284 static const char *const trace_errors[] = {
 285 #define TREDEF(name, msg)       msg,
 286 #include "lj_traceerr.h"
 287   NULL
 288 };
 289
 290 static const char *lower(char *buf, const char *s)
 291 {
 292   char *p = buf;
 293   while (*s) {
 294     *p++ = (*s >= 'A' && *s <= 'Z') ? *s+0x20 : *s;
 295     s++;
 296   }
 297   *p = '\0';
 298   return buf;
 299 }
 300
 301 /* Emit C source code for bytecode-related definitions. */
 302 static void emit_bcdef(BuildCtx *ctx)
 303 {
 304   int i;
 305   fprintf(ctx->fp, "/* This is a generated file. DO NOT EDIT! */\n\n");
 306   fprintf(ctx->fp, "LJ_DATADEF const uint16_t lj_bc_ofs[] = {\n");
 307   for (i = 0; i < ctx->npc; i++) {
 308     if (i != 0)
 309       fprintf(ctx->fp, ",\n");
 310     fprintf(ctx->fp, "%d", ctx->bc_ofs[i]);
 311   }
 312 }
 313
 314 /* Emit VM definitions as Lua code for debug modules. */
 315 static void emit_vmdef(BuildCtx *ctx)
 316 {
 317   char buf[80];
 318   int i;
 319   fprintf(ctx->fp, "-- This is a generated file. DO NOT EDIT!\n\n");
 320   fprintf(ctx->fp, "module(...)\n\n");
 321
 322   fprintf(ctx->fp, "bcnames = \"");
 323   for (i = 0; bc_names[i]; i++) fprintf(ctx->fp, "%-6s", bc_names[i]);
 324   fprintf(ctx->fp, "\"\n\n");
 325
 326   fprintf(ctx->fp, "irnames = \"");
 327   for (i = 0; ir_names[i]; i++) fprintf(ctx->fp, "%-6s", ir_names[i]);
 328   fprintf(ctx->fp, "\"\n\n");
 329
 330   fprintf(ctx->fp, "irfpm = { [0]=");
 331   for (i = 0; irfpm_names[i]; i++)
 332     fprintf(ctx->fp, "\"%s\", ", lower(buf, irfpm_names[i]));
 333   fprintf(ctx->fp, "}\n\n");
 334
 335   fprintf(ctx->fp, "irfield = { [0]=");
 336   for (i = 0; irfield_names[i]; i++) {
 337     char *p;
 338     lower(buf, irfield_names[i]);
 339     p = strchr(buf, '_');
 340     if (p) *p = '.';
 341     fprintf(ctx->fp, "\"%s\", ", buf);
 342   }
 343   fprintf(ctx->fp, "}\n\n");
 344
 345   fprintf(ctx->fp, "ircall = {\n[0]=");
 346   for (i = 0; ircall_names[i]; i++)
 347     fprintf(ctx->fp, "\"%s\",\n", ircall_names[i]);
 348   fprintf(ctx->fp, "}\n\n");
 349
 350   fprintf(ctx->fp, "traceerr = {\n[0]=");
 351   for (i = 0; trace_errors[i]; i++)
 352     fprintf(ctx->fp, "\"%s\",\n", trace_errors[i]);
 353   fprintf(ctx->fp, "}\n\n");
 354 }
 355
 356 /* -- Argument parsing ---------------------------------------------------- */
 357
 358 /* Build mode names. */
 359 static const char *const modenames[] = {
 360 #define BUILDNAME(name)         #name,
 361 BUILDDEF(BUILDNAME)
 362 #undef BUILDNAME
 363   NULL
 364 };
 365
 366 /* Print usage information and exit. */
 367 static void usage(void)
 368 {
 369   int i;
 370   fprintf(stderr, LUAJIT_VERSION " VM builder.\n");
 371   fprintf(stderr, LUAJIT_COPYRIGHT ", " LUAJIT_URL "\n");
 372   fprintf(stderr, "Target architecture: " LJ_ARCH_NAME "\n\n");
 373   fprintf(stderr, "Usage: buildvm -m mode [-o outfile] [infiles...]\n\n");
 374   fprintf(stderr, "Available modes:\n");
 375   for (i = 0; i < BUILD__MAX; i++)
 376     fprintf(stderr, "  %s\n", modenames[i]);
 377   exit(1);
 378 }
 379
 380 /* Parse the output mode name. */
 381 static BuildMode parsemode(const char *mode)
 382 {
 383   int i;
 384   for (i = 0; modenames[i]; i++)
 385     if (!strcmp(mode, modenames[i]))
 386       return (BuildMode)i;
 387   usage();
 388   return (BuildMode)-1;
 389 }
 390
 391 /* Parse arguments. */
 392 static void parseargs(BuildCtx *ctx, char **argv)
 393 {
 394   const char *a;
 395   int i;
 396   ctx->mode = (BuildMode)-1;
 397   ctx->outname = "-";
 398   for (i = 1; (a = argv[i]) != NULL; i++) {
 399     if (a[0] != '-')
 400       break;
 401     switch (a[1]) {
 402     case '-':
 403       if (a[2]) goto err;
 404       i++;
 405       goto ok;
 406     case '\0':
 407       goto ok;
 408     case 'm':
 409       i++;
 410       if (a[2] || argv[i] == NULL) goto err;
 411       ctx->mode = parsemode(argv[i]);
 412       break;
 413     case 'o':
 414       i++;
 415       if (a[2] || argv[i] == NULL) goto err;
 416       ctx->outname = argv[i];
 417       break;
 418     default: err:
 419       usage();
 420       break;
 421     }
 422   }
 423 ok:
 424   ctx->args = argv+i;
 425   if (ctx->mode == (BuildMode)-1) goto err;
 426 }
 427
 428 int main(int argc, char **argv)
 429 {
 430   BuildCtx ctx_;
 431   BuildCtx *ctx = &ctx_;
 432   int status, binmode;
 433
 434   if (sizeof(void *) != 4*LJ_32+8*LJ_64) {
 435     fprintf(stderr,"Error: pointer size mismatch in cross-build.\n");
 436     fprintf(stderr,"Try: make HOST_CC=\"gcc -m32\" CROSS=... TARGET=...\n\n");
 437     return 1;
 438   }
 439
 440   UNUSED(argc);
 441   parseargs(ctx, argv);
 442
 443   if ((status = build_code(ctx))) {
 444     fprintf(stderr,"Error: DASM error %08x\n", status);
 445     return 1;
 446   }
 447
 448   switch (ctx->mode) {
 449   case BUILD_peobj:
 450   case BUILD_raw:
 451     binmode = 1;
 452     break;
 453   default:
 454     binmode = 0;
 455     break;
 456   }
 457
 458   if (ctx->outname[0] == '-' && ctx->outname[1] == '\0') {
 459     ctx->fp = stdout;
 460 #if defined(_WIN32)
 461     if (binmode)
 462       _setmode(_fileno(stdout), _O_BINARY);  /* Yuck. */
 463 #endif
 464   } else if (!(ctx->fp = fopen(ctx->outname, binmode ? "wb" : "w"))) {
 465     fprintf(stderr, "Error: cannot open output file '%s': %s\n",
 466             ctx->outname, strerror(errno));
 467     exit(1);
 468   }
 469
 470   switch (ctx->mode) {
 471   case BUILD_elfasm:
 472   case BUILD_coffasm:
 473   case BUILD_machasm:
 474     emit_asm(ctx);
 475     emit_asm_debug(ctx);
 476     break;
 477   case BUILD_peobj:
 478     emit_peobj(ctx);
 479     break;
 480   case BUILD_raw:
 481     emit_raw(ctx);
 482     break;
 483   case BUILD_bcdef:
 484     emit_bcdef(ctx);
 485     emit_lib(ctx);
 486     break;
 487   case BUILD_vmdef:
 488     emit_vmdef(ctx);
 489     emit_lib(ctx);
 490     break;
 491   case BUILD_ffdef:
 492   case BUILD_libdef:
 493   case BUILD_recdef:
 494     emit_lib(ctx);
 495     break;
 496   case BUILD_folddef:
 497     emit_fold(ctx);
 498     break;
 499   default:
 500     break;
 501   }
 502
 503   fflush(ctx->fp);
 504   if (ferror(ctx->fp)) {
 505     fprintf(stderr, "Error: cannot write to output file: %s\n",
 506             strerror(errno));
 507     exit(1);
 508   }
 509   fclose(ctx->fp);
 510
 511   return 0;
 512 }
 513