src/lj_bcread.c

   1 /*
   2 ** Bytecode reader.
   3 ** Copyright (C) 2005-2023 Mike Pall. See Copyright Notice in luajit.h
   4 */
   5
   6 #define lj_bcread_c
   7 #define LUA_CORE
   8
   9 #include "lj_obj.h"
  10 #include "lj_gc.h"
  11 #include "lj_err.h"
  12 #include "lj_buf.h"
  13 #include "lj_str.h"
  14 #include "lj_tab.h"
  15 #include "lj_bc.h"
  16 #if LJ_HASFFI
  17 #include "lj_ctype.h"
  18 #include "lj_cdata.h"
  19 #include "lualib.h"
  20 #endif
  21 #include "lj_lex.h"
  22 #include "lj_bcdump.h"
  23 #include "lj_state.h"
  24 #include "lj_strfmt.h"
  25
  26 /* Reuse some lexer fields for our own purposes. */
  27 #define bcread_flags(ls)        ls->level
  28 #define bcread_swap(ls) \
  29   ((bcread_flags(ls) & BCDUMP_F_BE) != LJ_BE*BCDUMP_F_BE)
  30 #define bcread_oldtop(L, ls)    restorestack(L, ls->lastline)
  31 #define bcread_savetop(L, ls, top) \
  32   ls->lastline = (BCLine)savestack(L, (top))
  33
  34 /* -- Input buffer handling ----------------------------------------------- */
  35
  36 /* Throw reader error. */
  37 static LJ_NOINLINE void bcread_error(LexState *ls, ErrMsg em)
  38 {
  39   lua_State *L = ls->L;
  40   const char *name = ls->chunkarg;
  41   if (*name == BCDUMP_HEAD1) name = "(binary)";
  42   else if (*name == '@' || *name == '=') name++;
  43   lj_strfmt_pushf(L, "%s: %s", name, err2msg(em));
  44   lj_err_throw(L, LUA_ERRSYNTAX);
  45 }
  46
  47 /* Refill buffer. */
  48 static LJ_NOINLINE void bcread_fill(LexState *ls, MSize len, int need)
  49 {
  50   lj_assertLS(len != 0, "empty refill");
  51   if (len > LJ_MAX_BUF || ls->c < 0)
  52     bcread_error(ls, LJ_ERR_BCBAD);
  53   do {
  54     const char *buf;
  55     size_t sz;
  56     char *p = ls->sb.b;
  57     MSize n = (MSize)(ls->pe - ls->p);
  58     if (n) {  /* Copy remainder to buffer. */
  59       if (sbuflen(&ls->sb)) {  /* Move down in buffer. */
  60         lj_assertLS(ls->pe == ls->sb.w, "bad buffer pointer");
  61         if (ls->p != p) memmove(p, ls->p, n);
  62       } else {  /* Copy from buffer provided by reader. */
  63         p = lj_buf_need(&ls->sb, len);
  64         memcpy(p, ls->p, n);
  65       }
  66       ls->p = p;
  67       ls->pe = p + n;
  68     }
  69     ls->sb.w = p + n;
  70     buf = ls->rfunc(ls->L, ls->rdata, &sz);  /* Get more data from reader. */
  71     if (buf == NULL || sz == 0) {  /* EOF? */
  72       if (need) bcread_error(ls, LJ_ERR_BCBAD);
  73       ls->c = -1;  /* Only bad if we get called again. */
  74       break;
  75     }
  76     if (sz >= LJ_MAX_BUF - n) lj_err_mem(ls->L);
  77     if (n) {  /* Append to buffer. */
  78       n += (MSize)sz;
  79       p = lj_buf_need(&ls->sb, n < len ? len : n);
  80       memcpy(ls->sb.w, buf, sz);
  81       ls->sb.w = p + n;
  82       ls->p = p;
  83       ls->pe = p + n;
  84     } else {  /* Return buffer provided by reader. */
  85       ls->p = buf;
  86       ls->pe = buf + sz;
  87     }
  88   } while ((MSize)(ls->pe - ls->p) < len);
  89 }
  90
  91 /* Need a certain number of bytes. */
  92 static LJ_AINLINE void bcread_need(LexState *ls, MSize len)
  93 {
  94   if (LJ_UNLIKELY((MSize)(ls->pe - ls->p) < len))
  95     bcread_fill(ls, len, 1);
  96 }
  97
  98 /* Want to read up to a certain number of bytes, but may need less. */
  99 static LJ_AINLINE void bcread_want(LexState *ls, MSize len)
 100 {
 101   if (LJ_UNLIKELY((MSize)(ls->pe - ls->p) < len))
 102     bcread_fill(ls, len, 0);
 103 }
 104
 105 /* Return memory block from buffer. */
 106 static LJ_AINLINE uint8_t *bcread_mem(LexState *ls, MSize len)
 107 {
 108   uint8_t *p = (uint8_t *)ls->p;
 109   ls->p += len;
 110   lj_assertLS(ls->p <= ls->pe, "buffer read overflow");
 111   return p;
 112 }
 113
 114 /* Copy memory block from buffer. */
 115 static void bcread_block(LexState *ls, void *q, MSize len)
 116 {
 117   memcpy(q, bcread_mem(ls, len), len);
 118 }
 119
 120 /* Read byte from buffer. */
 121 static LJ_AINLINE uint32_t bcread_byte(LexState *ls)
 122 {
 123   lj_assertLS(ls->p < ls->pe, "buffer read overflow");
 124   return (uint32_t)(uint8_t)*ls->p++;
 125 }
 126
 127 /* Read ULEB128 value from buffer. */
 128 static LJ_AINLINE uint32_t bcread_uleb128(LexState *ls)
 129 {
 130   uint32_t v = lj_buf_ruleb128(&ls->p);
 131   lj_assertLS(ls->p <= ls->pe, "buffer read overflow");
 132   return v;
 133 }
 134
 135 /* Read top 32 bits of 33 bit ULEB128 value from buffer. */
 136 static uint32_t bcread_uleb128_33(LexState *ls)
 137 {
 138   const uint8_t *p = (const uint8_t *)ls->p;
 139   uint32_t v = (*p++ >> 1);
 140   if (LJ_UNLIKELY(v >= 0x40)) {
 141     int sh = -1;
 142     v &= 0x3f;
 143     do {
 144      v |= ((*p & 0x7f) << (sh += 7));
 145    } while (*p++ >= 0x80);
 146   }
 147   ls->p = (char *)p;
 148   lj_assertLS(ls->p <= ls->pe, "buffer read overflow");
 149   return v;
 150 }
 151
 152 /* -- Bytecode reader ----------------------------------------------------- */
 153
 154 /* Read debug info of a prototype. */
 155 static void bcread_dbg(LexState *ls, GCproto *pt, MSize sizedbg)
 156 {
 157   void *lineinfo = (void *)proto_lineinfo(pt);
 158   bcread_block(ls, lineinfo, sizedbg);
 159   /* Swap lineinfo if the endianess differs. */
 160   if (bcread_swap(ls) && pt->numline >= 256) {
 161     MSize i, n = pt->sizebc-1;
 162     if (pt->numline < 65536) {
 163       uint16_t *p = (uint16_t *)lineinfo;
 164       for (i = 0; i < n; i++) p[i] = (uint16_t)((p[i] >> 8)|(p[i] << 8));
 165     } else {
 166       uint32_t *p = (uint32_t *)lineinfo;
 167       for (i = 0; i < n; i++) p[i] = lj_bswap(p[i]);
 168     }
 169   }
 170 }
 171
 172 /* Find pointer to varinfo. */
 173 static const void *bcread_varinfo(GCproto *pt)
 174 {
 175   const uint8_t *p = proto_uvinfo(pt);
 176   MSize n = pt->sizeuv;
 177   if (n) while (*p++ || --n) ;
 178   return p;
 179 }
 180
 181 /* Read a single constant key/value of a template table. */
 182 static void bcread_ktabk(LexState *ls, TValue *o)
 183 {
 184   MSize tp = bcread_uleb128(ls);
 185   if (tp >= BCDUMP_KTAB_STR) {
 186     MSize len = tp - BCDUMP_KTAB_STR;
 187     const char *p = (const char *)bcread_mem(ls, len);
 188     setstrV(ls->L, o, lj_str_new(ls->L, p, len));
 189   } else if (tp == BCDUMP_KTAB_INT) {
 190     setintV(o, (int32_t)bcread_uleb128(ls));
 191   } else if (tp == BCDUMP_KTAB_NUM) {
 192     o->u32.lo = bcread_uleb128(ls);
 193     o->u32.hi = bcread_uleb128(ls);
 194   } else {
 195     lj_assertLS(tp <= BCDUMP_KTAB_TRUE, "bad constant type %d", tp);
 196     setpriV(o, ~tp);
 197   }
 198 }
 199
 200 /* Read a template table. */
 201 static GCtab *bcread_ktab(LexState *ls)
 202 {
 203   MSize narray = bcread_uleb128(ls);
 204   MSize nhash = bcread_uleb128(ls);
 205   GCtab *t = lj_tab_new(ls->L, narray, hsize2hbits(nhash));
 206   if (narray) {  /* Read array entries. */
 207     MSize i;
 208     TValue *o = tvref(t->array);
 209     for (i = 0; i < narray; i++, o++)
 210       bcread_ktabk(ls, o);
 211   }
 212   if (nhash) {  /* Read hash entries. */
 213     MSize i;
 214     for (i = 0; i < nhash; i++) {
 215       TValue key;
 216       bcread_ktabk(ls, &key);
 217       lj_assertLS(!tvisnil(&key), "nil key");
 218       bcread_ktabk(ls, lj_tab_set(ls->L, t, &key));
 219     }
 220   }
 221   return t;
 222 }
 223
 224 /* Read GC constants of a prototype. */
 225 static void bcread_kgc(LexState *ls, GCproto *pt, MSize sizekgc)
 226 {
 227   MSize i;
 228   GCRef *kr = mref(pt->k, GCRef) - (ptrdiff_t)sizekgc;
 229   for (i = 0; i < sizekgc; i++, kr++) {
 230     MSize tp = bcread_uleb128(ls);
 231     if (tp >= BCDUMP_KGC_STR) {
 232       MSize len = tp - BCDUMP_KGC_STR;
 233       const char *p = (const char *)bcread_mem(ls, len);
 234       setgcref(*kr, obj2gco(lj_str_new(ls->L, p, len)));
 235     } else if (tp == BCDUMP_KGC_TAB) {
 236       setgcref(*kr, obj2gco(bcread_ktab(ls)));
 237 #if LJ_HASFFI
 238     } else if (tp != BCDUMP_KGC_CHILD) {
 239       CTypeID id = tp == BCDUMP_KGC_COMPLEX ? CTID_COMPLEX_DOUBLE :
 240                    tp == BCDUMP_KGC_I64 ? CTID_INT64 : CTID_UINT64;
 241       CTSize sz = tp == BCDUMP_KGC_COMPLEX ? 16 : 8;
 242       GCcdata *cd = lj_cdata_new_(ls->L, id, sz);
 243       TValue *p = (TValue *)cdataptr(cd);
 244       setgcref(*kr, obj2gco(cd));
 245       p[0].u32.lo = bcread_uleb128(ls);
 246       p[0].u32.hi = bcread_uleb128(ls);
 247       if (tp == BCDUMP_KGC_COMPLEX) {
 248         p[1].u32.lo = bcread_uleb128(ls);
 249         p[1].u32.hi = bcread_uleb128(ls);
 250       }
 251 #endif
 252     } else {
 253       lua_State *L = ls->L;
 254       lj_assertLS(tp == BCDUMP_KGC_CHILD, "bad constant type %d", tp);
 255       if (L->top <= bcread_oldtop(L, ls))  /* Stack underflow? */
 256         bcread_error(ls, LJ_ERR_BCBAD);
 257       L->top--;
 258       setgcref(*kr, obj2gco(protoV(L->top)));
 259     }
 260   }
 261 }
 262
 263 /* Read number constants of a prototype. */
 264 static void bcread_knum(LexState *ls, GCproto *pt, MSize sizekn)
 265 {
 266   MSize i;
 267   TValue *o = mref(pt->k, TValue);
 268   for (i = 0; i < sizekn; i++, o++) {
 269     int isnum = (ls->p[0] & 1);
 270     uint32_t lo = bcread_uleb128_33(ls);
 271     if (isnum) {
 272       o->u32.lo = lo;
 273       o->u32.hi = bcread_uleb128(ls);
 274     } else {
 275       setintV(o, lo);
 276     }
 277   }
 278 }
 279
 280 /* Read bytecode instructions. */
 281 static void bcread_bytecode(LexState *ls, GCproto *pt, MSize sizebc)
 282 {
 283   BCIns *bc = proto_bc(pt);
 284   BCIns op;
 285   if (ls->fr2 != LJ_FR2) op = BC_NOT;  /* Mark non-native prototype. */
 286   else if ((pt->flags & PROTO_VARARG)) op = BC_FUNCV;
 287   else op = BC_FUNCF;
 288   bc[0] = BCINS_AD(op, pt->framesize, 0);
 289   bcread_block(ls, bc+1, (sizebc-1)*(MSize)sizeof(BCIns));
 290   /* Swap bytecode instructions if the endianess differs. */
 291   if (bcread_swap(ls)) {
 292     MSize i;
 293     for (i = 1; i < sizebc; i++) bc[i] = lj_bswap(bc[i]);
 294   }
 295 }
 296
 297 /* Read upvalue refs. */
 298 static void bcread_uv(LexState *ls, GCproto *pt, MSize sizeuv)
 299 {
 300   if (sizeuv) {
 301     uint16_t *uv = proto_uv(pt);
 302     bcread_block(ls, uv, sizeuv*2);
 303     /* Swap upvalue refs if the endianess differs. */
 304     if (bcread_swap(ls)) {
 305       MSize i;
 306       for (i = 0; i < sizeuv; i++)
 307         uv[i] = (uint16_t)((uv[i] >> 8)|(uv[i] << 8));
 308     }
 309   }
 310 }
 311
 312 /* Read a prototype. */
 313 GCproto *lj_bcread_proto(LexState *ls)
 314 {
 315   GCproto *pt;
 316   MSize framesize, numparams, flags, sizeuv, sizekgc, sizekn, sizebc, sizept;
 317   MSize ofsk, ofsuv, ofsdbg;
 318   MSize sizedbg = 0;
 319   BCLine firstline = 0, numline = 0;
 320
 321   /* Read prototype header. */
 322   flags = bcread_byte(ls);
 323   numparams = bcread_byte(ls);
 324   framesize = bcread_byte(ls);
 325   sizeuv = bcread_byte(ls);
 326   sizekgc = bcread_uleb128(ls);
 327   sizekn = bcread_uleb128(ls);
 328   sizebc = bcread_uleb128(ls) + 1;
 329   if (!(bcread_flags(ls) & BCDUMP_F_STRIP)) {
 330     sizedbg = bcread_uleb128(ls);
 331     if (sizedbg) {
 332       firstline = bcread_uleb128(ls);
 333       numline = bcread_uleb128(ls);
 334     }
 335   }
 336
 337   /* Calculate total size of prototype including all colocated arrays. */
 338   sizept = (MSize)sizeof(GCproto) +
 339            sizebc*(MSize)sizeof(BCIns) +
 340            sizekgc*(MSize)sizeof(GCRef);
 341   sizept = (sizept + (MSize)sizeof(TValue)-1) & ~((MSize)sizeof(TValue)-1);
 342   ofsk = sizept; sizept += sizekn*(MSize)sizeof(TValue);
 343   ofsuv = sizept; sizept += ((sizeuv+1)&~1)*2;
 344   ofsdbg = sizept; sizept += sizedbg;
 345
 346   /* Allocate prototype object and initialize its fields. */
 347   pt = (GCproto *)lj_mem_newgco(ls->L, (MSize)sizept);
 348   pt->gct = ~LJ_TPROTO;
 349   pt->numparams = (uint8_t)numparams;
 350   pt->framesize = (uint8_t)framesize;
 351   pt->sizebc = sizebc;
 352   setmref(pt->k, (char *)pt + ofsk);
 353   setmref(pt->uv, (char *)pt + ofsuv);
 354   pt->sizekgc = 0;  /* Set to zero until fully initialized. */
 355   pt->sizekn = sizekn;
 356   pt->sizept = sizept;
 357   pt->sizeuv = (uint8_t)sizeuv;
 358   pt->flags = (uint8_t)flags;
 359   pt->trace = 0;
 360   setgcref(pt->chunkname, obj2gco(ls->chunkname));
 361
 362   /* Close potentially uninitialized gap between bc and kgc. */
 363   *(uint32_t *)((char *)pt + ofsk - sizeof(GCRef)*(sizekgc+1)) = 0;
 364
 365   /* Read bytecode instructions and upvalue refs. */
 366   bcread_bytecode(ls, pt, sizebc);
 367   bcread_uv(ls, pt, sizeuv);
 368
 369   /* Read constants. */
 370   bcread_kgc(ls, pt, sizekgc);
 371   pt->sizekgc = sizekgc;
 372   bcread_knum(ls, pt, sizekn);
 373
 374   /* Read and initialize debug info. */
 375   pt->firstline = firstline;
 376   pt->numline = numline;
 377   if (sizedbg) {
 378     MSize sizeli = (sizebc-1) << (numline < 256 ? 0 : numline < 65536 ? 1 : 2);
 379     setmref(pt->lineinfo, (char *)pt + ofsdbg);
 380     setmref(pt->uvinfo, (char *)pt + ofsdbg + sizeli);
 381     bcread_dbg(ls, pt, sizedbg);
 382     setmref(pt->varinfo, bcread_varinfo(pt));
 383   } else {
 384     setmref(pt->lineinfo, NULL);
 385     setmref(pt->uvinfo, NULL);
 386     setmref(pt->varinfo, NULL);
 387   }
 388   return pt;
 389 }
 390
 391 /* Read and check header of bytecode dump. */
 392 static int bcread_header(LexState *ls)
 393 {
 394   uint32_t flags;
 395   bcread_want(ls, 3+5+5);
 396   if (bcread_byte(ls) != BCDUMP_HEAD2 ||
 397       bcread_byte(ls) != BCDUMP_HEAD3 ||
 398       bcread_byte(ls) != BCDUMP_VERSION) return 0;
 399   bcread_flags(ls) = flags = bcread_uleb128(ls);
 400   if ((flags & ~(BCDUMP_F_KNOWN)) != 0) return 0;
 401   if ((flags & BCDUMP_F_FR2) != (uint32_t)ls->fr2*BCDUMP_F_FR2) return 0;
 402   if ((flags & BCDUMP_F_FFI)) {
 403 #if LJ_HASFFI
 404     lua_State *L = ls->L;
 405     ctype_loadffi(L);
 406 #else
 407     return 0;
 408 #endif
 409   }
 410   if ((flags & BCDUMP_F_STRIP)) {
 411     ls->chunkname = lj_str_newz(ls->L, ls->chunkarg);
 412   } else {
 413     MSize len = bcread_uleb128(ls);
 414     bcread_need(ls, len);
 415     ls->chunkname = lj_str_new(ls->L, (const char *)bcread_mem(ls, len), len);
 416   }
 417   return 1;  /* Ok. */
 418 }
 419
 420 /* Read a bytecode dump. */
 421 GCproto *lj_bcread(LexState *ls)
 422 {
 423   lua_State *L = ls->L;
 424   lj_assertLS(ls->c == BCDUMP_HEAD1, "bad bytecode header");
 425   bcread_savetop(L, ls, L->top);
 426   lj_buf_reset(&ls->sb);
 427   /* Check for a valid bytecode dump header. */
 428   if (!bcread_header(ls))
 429     bcread_error(ls, LJ_ERR_BCFMT);
 430   for (;;) {  /* Process all prototypes in the bytecode dump. */
 431     GCproto *pt;
 432     MSize len;
 433     const char *startp;
 434     /* Read length. */
 435     if (ls->p < ls->pe && ls->p[0] == 0) {  /* Shortcut EOF. */
 436       ls->p++;
 437       break;
 438     }
 439     bcread_want(ls, 5);
 440     len = bcread_uleb128(ls);
 441     if (!len) break;  /* EOF */
 442     bcread_need(ls, len);
 443     startp = ls->p;
 444     pt = lj_bcread_proto(ls);
 445     if (ls->p != startp + len)
 446       bcread_error(ls, LJ_ERR_BCBAD);
 447     setprotoV(L, L->top, pt);
 448     incr_top(L);
 449   }
 450   if ((ls->pe != ls->p && !ls->endmark) || L->top-1 != bcread_oldtop(L, ls))
 451     bcread_error(ls, LJ_ERR_BCBAD);
 452   /* Pop off last prototype. */
 453   L->top--;
 454   return protoV(L->top);
 455 }
 456