dyngen.c

   1 /*
   2  *  Generic Dynamic compiler generator
   3  *
   4  *  Copyright (c) 2003 Fabrice Bellard
   5  *
   6  *  This program is free software; you can redistribute it and/or modify
   7  *  it under the terms of the GNU General Public License as published by
   8  *  the Free Software Foundation; either version 2 of the License, or
   9  *  (at your option) any later version.
  10  *
  11  *  This program is distributed in the hope that it will be useful,
  12  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  *  GNU General Public License for more details.
  15  *
  16  *  You should have received a copy of the GNU General Public License
  17  *  along with this program; if not, write to the Free Software
  18  *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  19  */
  20 #include <stdlib.h>
  21 #include <stdio.h>
  22 #include <string.h>
  23 #include <stdarg.h>
  24 #include <inttypes.h>
  25 #include <unistd.h>
  26 #include <fcntl.h>
  27
  28 #include "config.h"
  29
  30 /* elf format definitions. We use these macros to test the CPU to
  31    allow cross compilation (this tool must be ran on the build
  32    platform) */
  33 #if defined(HOST_I386)
  34
  35 #define ELF_CLASS       ELFCLASS32
  36 #define ELF_ARCH        EM_386
  37 #define elf_check_arch(x) ( ((x) == EM_386) || ((x) == EM_486) )
  38 #undef ELF_USES_RELOCA
  39
  40 #elif defined(HOST_PPC)
  41
  42 #define ELF_CLASS       ELFCLASS32
  43 #define ELF_ARCH        EM_PPC
  44 #define elf_check_arch(x) ((x) == EM_PPC)
  45 #define ELF_USES_RELOCA
  46
  47 #elif defined(HOST_S390)
  48
  49 #define ELF_CLASS       ELFCLASS32
  50 #define ELF_ARCH        EM_S390
  51 #define elf_check_arch(x) ((x) == EM_S390)
  52 #define ELF_USES_RELOCA
  53
  54 #elif defined(HOST_ALPHA)
  55
  56 #define ELF_CLASS       ELFCLASS64
  57 #define ELF_ARCH        EM_ALPHA
  58 #define elf_check_arch(x) ((x) == EM_ALPHA)
  59 #define ELF_USES_RELOCA
  60
  61 #elif defined(HOST_IA64)
  62
  63 #define ELF_CLASS       ELFCLASS64
  64 #define ELF_ARCH        EM_IA_64
  65 #define elf_check_arch(x) ((x) == EM_IA_64)
  66 #define ELF_USES_RELOCA
  67
  68 #elif defined(HOST_SPARC)
  69
  70 #define ELF_CLASS       ELFCLASS32
  71 #define ELF_ARCH        EM_SPARC
  72 #define elf_check_arch(x) ((x) == EM_SPARC || (x) == EM_SPARC32PLUS)
  73 #define ELF_USES_RELOCA
  74
  75 #elif defined(HOST_SPARC64)
  76
  77 #define ELF_CLASS       ELFCLASS64
  78 #define ELF_ARCH        EM_SPARCV9
  79 #define elf_check_arch(x) ((x) == EM_SPARCV9)
  80 #define ELF_USES_RELOCA
  81
  82 #else
  83 #error unsupported CPU - please update the code
  84 #endif
  85
  86 #include "elf.h"
  87
  88 #if ELF_CLASS == ELFCLASS32
  89 typedef int32_t host_long;
  90 typedef uint32_t host_ulong;
  91 #define swabls(x) swab32s(x)
  92 #else
  93 typedef int64_t host_long;
  94 typedef uint64_t host_ulong;
  95 #define swabls(x) swab64s(x)
  96 #endif
  97
  98 #include "thunk.h"
  99
 100 /* all dynamically generated functions begin with this code */
 101 #define OP_PREFIX "op_"
 102
 103 int elf_must_swap(struct elfhdr *h)
 104 {
 105   union {
 106       uint32_t i;
 107       uint8_t b[4];
 108   } swaptest;
 109
 110   swaptest.i = 1;
 111   return (h->e_ident[EI_DATA] == ELFDATA2MSB) !=
 112       (swaptest.b[0] == 0);
 113 }
 114
 115 void swab16s(uint16_t *p)
 116 {
 117     *p = bswap16(*p);
 118 }
 119
 120 void swab32s(uint32_t *p)
 121 {
 122     *p = bswap32(*p);
 123 }
 124
 125 void swab64s(uint64_t *p)
 126 {
 127     *p = bswap64(*p);
 128 }
 129
 130 void elf_swap_ehdr(struct elfhdr *h)
 131 {
 132     swab16s(&h->e_type);                        /* Object file type */
 133     swab16s(&h->        e_machine);             /* Architecture */
 134     swab32s(&h->        e_version);             /* Object file version */
 135     swabls(&h-> e_entry);               /* Entry point virtual address */
 136     swabls(&h-> e_phoff);               /* Program header table file offset */
 137     swabls(&h-> e_shoff);               /* Section header table file offset */
 138     swab32s(&h->        e_flags);               /* Processor-specific flags */
 139     swab16s(&h->        e_ehsize);              /* ELF header size in bytes */
 140     swab16s(&h->        e_phentsize);           /* Program header table entry size */
 141     swab16s(&h->        e_phnum);               /* Program header table entry count */
 142     swab16s(&h->        e_shentsize);           /* Section header table entry size */
 143     swab16s(&h->        e_shnum);               /* Section header table entry count */
 144     swab16s(&h->        e_shstrndx);            /* Section header string table index */
 145 }
 146
 147 void elf_swap_shdr(struct elf_shdr *h)
 148 {
 149   swab32s(&h->  sh_name);               /* Section name (string tbl index) */
 150   swab32s(&h->  sh_type);               /* Section type */
 151   swabls(&h->   sh_flags);              /* Section flags */
 152   swabls(&h->   sh_addr);               /* Section virtual addr at execution */
 153   swabls(&h->   sh_offset);             /* Section file offset */
 154   swabls(&h->   sh_size);               /* Section size in bytes */
 155   swab32s(&h->  sh_link);               /* Link to another section */
 156   swab32s(&h->  sh_info);               /* Additional section information */
 157   swabls(&h->   sh_addralign);          /* Section alignment */
 158   swabls(&h->   sh_entsize);            /* Entry size if section holds table */
 159 }
 160
 161 void elf_swap_phdr(struct elf_phdr *h)
 162 {
 163     swab32s(&h->p_type);                        /* Segment type */
 164     swabls(&h->p_offset);               /* Segment file offset */
 165     swabls(&h->p_vaddr);                /* Segment virtual address */
 166     swabls(&h->p_paddr);                /* Segment physical address */
 167     swabls(&h->p_filesz);               /* Segment size in file */
 168     swabls(&h->p_memsz);                /* Segment size in memory */
 169     swab32s(&h->p_flags);               /* Segment flags */
 170     swabls(&h->p_align);                /* Segment alignment */
 171 }
 172
 173 /* ELF file info */
 174 int do_swap;
 175 struct elf_shdr *shdr;
 176 struct elfhdr ehdr;
 177 ElfW(Sym) *symtab;
 178 int nb_syms;
 179 char *strtab;
 180 /* data section */
 181 uint8_t *data_data, *sdata_data;
 182 int data_shndx, sdata_shndx;
 183
 184 uint16_t get16(uint16_t *p)
 185 {
 186     uint16_t val;
 187     val = *p;
 188     if (do_swap)
 189         val = bswap16(val);
 190     return val;
 191 }
 192
 193 uint32_t get32(uint32_t *p)
 194 {
 195     uint32_t val;
 196     val = *p;
 197     if (do_swap)
 198         val = bswap32(val);
 199     return val;
 200 }
 201
 202 void put16(uint16_t *p, uint16_t val)
 203 {
 204     if (do_swap)
 205         val = bswap16(val);
 206     *p = val;
 207 }
 208
 209 void put32(uint32_t *p, uint32_t val)
 210 {
 211     if (do_swap)
 212         val = bswap32(val);
 213     *p = val;
 214 }
 215
 216 void __attribute__((noreturn)) __attribute__((format (printf, 1, 2))) error(const char *fmt, ...)
 217 {
 218     va_list ap;
 219     va_start(ap, fmt);
 220     fprintf(stderr, "dyngen: ");
 221     vfprintf(stderr, fmt, ap);
 222     fprintf(stderr, "\n");
 223     va_end(ap);
 224     exit(1);
 225 }
 226
 227
 228 struct elf_shdr *find_elf_section(struct elf_shdr *shdr, int shnum, const char *shstr,
 229                                   const char *name)
 230 {
 231     int i;
 232     const char *shname;
 233     struct elf_shdr *sec;
 234
 235     for(i = 0; i < shnum; i++) {
 236         sec = &shdr[i];
 237         if (!sec->sh_name)
 238             continue;
 239         shname = shstr + sec->sh_name;
 240         if (!strcmp(shname, name))
 241             return sec;
 242     }
 243     return NULL;
 244 }
 245
 246 void *load_data(int fd, long offset, unsigned int size)
 247 {
 248     char *data;
 249
 250     data = malloc(size);
 251     if (!data)
 252         return NULL;
 253     lseek(fd, offset, SEEK_SET);
 254     if (read(fd, data, size) != size) {
 255         free(data);
 256         return NULL;
 257     }
 258     return data;
 259 }
 260
 261 int strstart(const char *str, const char *val, const char **ptr)
 262 {
 263     const char *p, *q;
 264     p = str;
 265     q = val;
 266     while (*q != '\0') {
 267         if (*p != *q)
 268             return 0;
 269         p++;
 270         q++;
 271     }
 272     if (ptr)
 273         *ptr = p;
 274     return 1;
 275 }
 276
 277 #define MAX_ARGS 3
 278
 279 /* generate op code */
 280 void gen_code(const char *name, host_ulong offset, host_ulong size,
 281               FILE *outfile, uint8_t *text, ELF_RELOC *relocs, int nb_relocs, int reloc_sh_type,
 282               int gen_switch)
 283 {
 284     int copy_size = 0;
 285     uint8_t *p_start, *p_end;
 286     host_ulong start_offset;
 287     int nb_args, i, n;
 288     uint8_t args_present[MAX_ARGS];
 289     const char *sym_name, *p;
 290     ELF_RELOC *rel;
 291
 292     /* Compute exact size excluding prologue and epilogue instructions.
 293      * Increment start_offset to skip epilogue instructions, then compute
 294      * copy_size the indicate the size of the remaining instructions (in
 295      * bytes).
 296      */
 297     p_start = text + offset;
 298     p_end = p_start + size;
 299     start_offset = offset;
 300     switch(ELF_ARCH) {
 301     case EM_386:
 302         {
 303             int len;
 304             len = p_end - p_start;
 305             if (len == 0)
 306                 error("empty code for %s", name);
 307             if (p_end[-1] == 0xc3) {
 308                 len--;
 309             } else {
 310                 error("ret or jmp expected at the end of %s", name);
 311             }
 312             copy_size = len;
 313         }
 314         break;
 315     case EM_PPC:
 316         {
 317             uint8_t *p;
 318             p = (void *)(p_end - 4);
 319             if (p == p_start)
 320                 error("empty code for %s", name);
 321             if (get32((uint32_t *)p) != 0x4e800020)
 322                 error("blr expected at the end of %s", name);
 323             copy_size = p - p_start;
 324         }
 325         break;
 326     case EM_S390:
 327         {
 328             uint8_t *p;
 329             p = (void *)(p_end - 2);
 330             if (p == p_start)
 331                 error("empty code for %s", name);
 332             if (get16((uint16_t *)p) != 0x07fe && get16((uint16_t *)p) != 0x07f4)
 333                 error("br %%r14 expected at the end of %s", name);
 334             copy_size = p - p_start;
 335         }
 336         break;
 337     case EM_ALPHA:
 338         {
 339             uint8_t *p;
 340             p = p_end - 4;
 341             if (p == p_start)
 342                 error("empty code for %s", name);
 343             if (get32((uint32_t *)p) != 0x6bfa8001)
 344                 error("ret expected at the end of %s", name);
 345             copy_size = p - p_start;
 346         }
 347         break;
 348     case EM_IA_64:
 349         {
 350             uint8_t *p;
 351             p = (void *)(p_end - 4);
 352             if (p == p_start)
 353                 error("empty code for %s", name);
 354             /* br.ret.sptk.many b0;; */
 355             /* 08 00 84 00 */
 356             if (get32((uint32_t *)p) != 0x00840008)
 357                 error("br.ret.sptk.many b0;; expected at the end of %s", name);
 358             copy_size = p - p_start;
 359         }
 360         break;
 361     case EM_SPARC:
 362     case EM_SPARC32PLUS:
 363         {
 364             uint32_t start_insn, end_insn1, end_insn2, skip_insn;
 365             uint8_t *p;
 366             p = (void *)(p_end - 8);
 367             if (p <= p_start)
 368                 error("empty code for %s", name);
 369             start_insn = get32((uint32_t *)(p_start + 0x0));
 370             end_insn1 = get32((uint32_t *)(p + 0x0));
 371             end_insn2 = get32((uint32_t *)(p + 0x4));
 372             if ((start_insn & ~0x1fff) == 0x9de3a000) {
 373                 p_start += 0x4;
 374                 start_offset += 0x4;
 375                 if ((int)(start_insn | ~0x1fff) < -128)
 376                     error("Found bogus save at the start of %s", name);
 377                 if (end_insn1 != 0x81c7e008 || end_insn2 != 0x81e80000)
 378                     error("ret; restore; not found at end of %s", name);
 379             } else {
 380                 error("No save at the beginning of %s", name);
 381             }
 382
 383             /* Skip a preceeding nop, if present.  */
 384             if (p > p_start) {
 385                 skip_insn = get32((uint32_t *)(p - 0x4));
 386                 if (skip_insn == 0x01000000)
 387                     p -= 4;
 388             }
 389
 390             copy_size = p - p_start;
 391         }
 392         break;
 393     case EM_SPARCV9:
 394         {
 395             uint32_t start_insn, end_insn1, end_insn2, skip_insn;
 396             uint8_t *p;
 397             p = (void *)(p_end - 8);
 398             if (p <= p_start)
 399                 error("empty code for %s", name);
 400             start_insn = get32((uint32_t *)(p_start + 0x0));
 401             end_insn1 = get32((uint32_t *)(p + 0x0));
 402             end_insn2 = get32((uint32_t *)(p + 0x4));
 403             if ((start_insn & ~0x1fff) == 0x9de3a000) {
 404                 p_start += 0x4;
 405                 start_offset += 0x4;
 406                 if ((int)(start_insn | ~0x1fff) < -256)
 407                     error("Found bogus save at the start of %s", name);
 408                 if (end_insn1 != 0x81c7e008 || end_insn2 != 0x81e80000)
 409                     error("ret; restore; not found at end of %s", name);
 410             } else {
 411                 error("No save at the beginning of %s", name);
 412             }
 413
 414             /* Skip a preceeding nop, if present.  */
 415             if (p > p_start) {
 416                 skip_insn = get32((uint32_t *)(p - 0x4));
 417                 if (skip_insn == 0x01000000)
 418                     p -= 4;
 419             }
 420
 421             copy_size = p - p_start;
 422         }
 423         break;
 424     default:
 425         error("unknown ELF architecture");
 426     }
 427
 428     /* compute the number of arguments by looking at the relocations */
 429     for(i = 0;i < MAX_ARGS; i++)
 430         args_present[i] = 0;
 431
 432     for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 433         if (rel->r_offset >= start_offset &&
 434             rel->r_offset < start_offset + copy_size) {
 435             sym_name = strtab + symtab[ELFW(R_SYM)(rel->r_info)].st_name;
 436             if (strstart(sym_name, "__op_param", &p)) {
 437                 n = strtoul(p, NULL, 10);
 438                 if (n > MAX_ARGS)
 439                     error("too many arguments in %s", name);
 440                 args_present[n - 1] = 1;
 441             }
 442         }
 443     }
 444
 445     nb_args = 0;
 446     while (nb_args < MAX_ARGS && args_present[nb_args])
 447         nb_args++;
 448     for(i = nb_args; i < MAX_ARGS; i++) {
 449         if (args_present[i])
 450             error("inconsistent argument numbering in %s", name);
 451     }
 452
 453     if (gen_switch == 2) {
 454         fprintf(outfile, "DEF(%s, %d, %d)\n", name + 3, nb_args, copy_size);
 455     } else if (gen_switch == 1) {
 456
 457         /* output C code */
 458         fprintf(outfile, "case INDEX_%s: {\n", name);
 459         if (nb_args > 0) {
 460             fprintf(outfile, "    long ");
 461             for(i = 0; i < nb_args; i++) {
 462                 if (i != 0)
 463                     fprintf(outfile, ", ");
 464                 fprintf(outfile, "param%d", i + 1);
 465             }
 466             fprintf(outfile, ";\n");
 467         }
 468         fprintf(outfile, "    extern void %s();\n", name);
 469
 470         for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 471             if (rel->r_offset >= start_offset &&
 472                 rel->r_offset < start_offset + copy_size) {
 473                 sym_name = strtab + symtab[ELFW(R_SYM)(rel->r_info)].st_name;
 474                 if (*sym_name &&
 475                     !strstart(sym_name, "__op_param", NULL) &&
 476                     !strstart(sym_name, "__op_jmp", NULL)) {
 477 #if defined(HOST_SPARC)
 478                     if (sym_name[0] == '.') {
 479                         fprintf(outfile,
 480                                 "extern char __dot_%s __asm__(\"%s\");\n",
 481                                 sym_name+1, sym_name);
 482                         continue;
 483                     }
 484 #endif
 485                     fprintf(outfile, "extern char %s;\n", sym_name);
 486                 }
 487             }
 488         }
 489
 490         fprintf(outfile, "    memcpy(gen_code_ptr, (void *)((char *)&%s+%d), %d);\n", name, start_offset - offset, copy_size);
 491
 492         /* emit code offset information */
 493         {
 494             ElfW(Sym) *sym;
 495             const char *sym_name, *p;
 496             target_ulong val;
 497             int n;
 498
 499             for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
 500                 sym_name = strtab + sym->st_name;
 501                 if (strstart(sym_name, "__op_label", &p)) {
 502                     uint8_t *ptr;
 503
 504                     /* test if the variable refers to a label inside
 505                        the code we are generating */
 506                     if (sym->st_shndx == data_shndx)
 507                         ptr = data_data;
 508                     else if (sym->st_shndx == sdata_shndx)
 509                         ptr = sdata_data;
 510                     else
 511                         error("__op_labelN symbols must be in .data or .sdata section");
 512                     val = *(target_ulong *)(ptr + sym->st_value);
 513                     if (val >= start_offset && val < start_offset + copy_size) {
 514                         n = strtol(p, NULL, 10);
 515                         fprintf(outfile, "    label_offsets[%d] = %d + (gen_code_ptr - gen_code_buf);\n", n, val - start_offset);
 516                     }
 517                 }
 518             }
 519         }
 520
 521         /* load parameres in variables */
 522         for(i = 0; i < nb_args; i++) {
 523             fprintf(outfile, "    param%d = *opparam_ptr++;\n", i + 1);
 524         }
 525
 526         /* patch relocations */
 527 #if defined(HOST_I386)
 528             {
 529                 char name[256];
 530                 int type;
 531                 int addend;
 532                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 533                 if (rel->r_offset >= start_offset &&
 534                     rel->r_offset < start_offset + copy_size) {
 535                     sym_name = strtab + symtab[ELFW(R_SYM)(rel->r_info)].st_name;
 536                     if (strstart(sym_name, "__op_param", &p)) {
 537                         snprintf(name, sizeof(name), "param%s", p);
 538                     } else {
 539                         snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 540                     }
 541                     type = ELF32_R_TYPE(rel->r_info);
 542                     addend = get32((uint32_t *)(text + rel->r_offset));
 543                     switch(type) {
 544                     case R_386_32:
 545                         fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s + %d;\n",
 546                                 rel->r_offset - start_offset, name, addend);
 547                         break;
 548                     case R_386_PC32:
 549                         fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s - (long)(gen_code_ptr + %d) + %d;\n",
 550                                 rel->r_offset - start_offset, name, rel->r_offset - start_offset, addend);
 551                         break;
 552                     default:
 553                         error("unsupported i386 relocation (%d)", type);
 554                     }
 555                 }
 556                 }
 557             }
 558 #elif defined(HOST_PPC)
 559             {
 560                 char name[256];
 561                 int type;
 562                 int addend;
 563                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 564                     if (rel->r_offset >= start_offset &&
 565                         rel->r_offset < start_offset + copy_size) {
 566                         sym_name = strtab + symtab[ELFW(R_SYM)(rel->r_info)].st_name;
 567                         if (strstart(sym_name, "__op_jmp", &p)) {
 568                             int n;
 569                             n = strtol(p, NULL, 10);
 570                             /* __op_jmp relocations are done at
 571                                runtime to do translated block
 572                                chaining: the offset of the instruction
 573                                needs to be stored */
 574                             fprintf(outfile, "    jmp_offsets[%d] = %d + (gen_code_ptr - gen_code_buf);\n",
 575                                     n, rel->r_offset - start_offset);
 576                             continue;
 577                         }
 578
 579                         if (strstart(sym_name, "__op_param", &p)) {
 580                             snprintf(name, sizeof(name), "param%s", p);
 581                         } else {
 582                             snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 583                         }
 584                         type = ELF32_R_TYPE(rel->r_info);
 585                         addend = rel->r_addend;
 586                         switch(type) {
 587                         case R_PPC_ADDR32:
 588                             fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s + %d;\n",
 589                                     rel->r_offset - start_offset, name, addend);
 590                             break;
 591                         case R_PPC_ADDR16_LO:
 592                             fprintf(outfile, "    *(uint16_t *)(gen_code_ptr + %d) = (%s + %d);\n",
 593                                     rel->r_offset - start_offset, name, addend);
 594                             break;
 595                         case R_PPC_ADDR16_HI:
 596                             fprintf(outfile, "    *(uint16_t *)(gen_code_ptr + %d) = (%s + %d) >> 16;\n",
 597                                     rel->r_offset - start_offset, name, addend);
 598                             break;
 599                         case R_PPC_ADDR16_HA:
 600                             fprintf(outfile, "    *(uint16_t *)(gen_code_ptr + %d) = (%s + %d + 0x8000) >> 16;\n",
 601                                     rel->r_offset - start_offset, name, addend);
 602                             break;
 603                         case R_PPC_REL24:
 604                             /* warning: must be at 32 MB distancy */
 605                             fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = (*(uint32_t *)(gen_code_ptr + %d) & ~0x03fffffc) | ((%s - (long)(gen_code_ptr + %d) + %d) & 0x03fffffc);\n",
 606                                     rel->r_offset - start_offset, rel->r_offset - start_offset, name, rel->r_offset - start_offset, addend);
 607                             break;
 608                         default:
 609                             error("unsupported powerpc relocation (%d)", type);
 610                         }
 611                     }
 612                 }
 613             }
 614 #elif defined(HOST_S390)
 615             {
 616                 char name[256];
 617                 int type;
 618                 int addend;
 619                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 620                     if (rel->r_offset >= start_offset &&
 621                         rel->r_offset < start_offset + copy_size) {
 622                         sym_name = strtab + symtab[ELFW(R_SYM)(rel->r_info)].st_name;
 623                         if (strstart(sym_name, "__op_param", &p)) {
 624                             snprintf(name, sizeof(name), "param%s", p);
 625                         } else {
 626                             snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 627                         }
 628                         type = ELF32_R_TYPE(rel->r_info);
 629                         addend = rel->r_addend;
 630                         switch(type) {
 631                         case R_390_32:
 632                             fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s + %d;\n",
 633                                     rel->r_offset - start_offset, name, addend);
 634                             break;
 635                         case R_390_16:
 636                             fprintf(outfile, "    *(uint16_t *)(gen_code_ptr + %d) = %s + %d;\n",
 637                                     rel->r_offset - start_offset, name, addend);
 638                             break;
 639                         case R_390_8:
 640                             fprintf(outfile, "    *(uint8_t *)(gen_code_ptr + %d) = %s + %d;\n",
 641                                     rel->r_offset - start_offset, name, addend);
 642                             break;
 643                         default:
 644                             error("unsupported s390 relocation (%d)", type);
 645                         }
 646                     }
 647                 }
 648             }
 649 #elif defined(HOST_ALPHA)
 650             {
 651                 for (i = 0, rel = relocs; i < nb_relocs; i++, rel++) {
 652                     if (rel->r_offset >= start_offset && rel->r_offset < start_offset + copy_size) {
 653                         int type;
 654
 655                         type = ELF64_R_TYPE(rel->r_info);
 656                         sym_name = strtab + symtab[ELF64_R_SYM(rel->r_info)].st_name;
 657                         switch (type) {
 658                         case R_ALPHA_GPDISP:
 659                             /* The gp is just 32 bit, and never changes, so it's easiest to emit it
 660                                as an immediate instead of constructing it from the pv or ra.  */
 661                             fprintf(outfile, "    immediate_ldah(gen_code_ptr + %ld, gp);\n",
 662                                     rel->r_offset - start_offset);
 663                             fprintf(outfile, "    immediate_lda(gen_code_ptr + %ld, gp);\n",
 664                                     rel->r_offset - start_offset + rel->r_addend);
 665                             break;
 666                         case R_ALPHA_LITUSE:
 667                             /* jsr to literal hint. Could be used to optimize to bsr. Ignore for
 668                                now, since some called functions (libc) need pv to be set up.  */
 669                             break;
 670                         case R_ALPHA_HINT:
 671                             /* Branch target prediction hint. Ignore for now.  Should be already
 672                                correct for in-function jumps.  */
 673                             break;
 674                         case R_ALPHA_LITERAL:
 675                             /* Load a literal from the GOT relative to the gp.  Since there's only a
 676                                single gp, nothing is to be done.  */
 677                             break;
 678                         case R_ALPHA_GPRELHIGH:
 679                             /* Handle fake relocations against __op_param symbol.  Need to emit the
 680                                high part of the immediate value instead.  Other symbols need no
 681                                special treatment.  */
 682                             if (strstart(sym_name, "__op_param", &p))
 683                                 fprintf(outfile, "    immediate_ldah(gen_code_ptr + %ld, param%s);\n",
 684                                         rel->r_offset - start_offset, p);
 685                             break;
 686                         case R_ALPHA_GPRELLOW:
 687                             if (strstart(sym_name, "__op_param", &p))
 688                                 fprintf(outfile, "    immediate_lda(gen_code_ptr + %ld, param%s);\n",
 689                                         rel->r_offset - start_offset, p);
 690                             break;
 691                         case R_ALPHA_BRSGP:
 692                             /* PC-relative jump. Tweak offset to skip the two instructions that try to
 693                                set up the gp from the pv.  */
 694                             fprintf(outfile, "    fix_bsr(gen_code_ptr + %ld, (uint8_t *) &%s - (gen_code_ptr + %ld + 4) + 8);\n",
 695                                     rel->r_offset - start_offset, sym_name, rel->r_offset - start_offset);
 696                             break;
 697                         default:
 698                             error("unsupported Alpha relocation (%d)", type);
 699                         }
 700                     }
 701                 }
 702             }
 703 #elif defined(HOST_IA64)
 704             {
 705                 char name[256];
 706                 int type;
 707                 int addend;
 708                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 709                     if (rel->r_offset >= start_offset && rel->r_offset < start_offset + copy_size) {
 710                         sym_name = strtab + symtab[ELF64_R_SYM(rel->r_info)].st_name;
 711                         if (strstart(sym_name, "__op_param", &p)) {
 712                             snprintf(name, sizeof(name), "param%s", p);
 713                         } else {
 714                             snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 715                         }
 716                         type = ELF64_R_TYPE(rel->r_info);
 717                         addend = rel->r_addend;
 718                         switch(type) {
 719                         case R_IA64_LTOFF22:
 720                             error("must implemnt R_IA64_LTOFF22 relocation");
 721                         case R_IA64_PCREL21B:
 722                             error("must implemnt R_IA64_PCREL21B relocation");
 723                         default:
 724                             error("unsupported ia64 relocation (%d)", type);
 725                         }
 726                     }
 727                 }
 728             }
 729 #elif defined(HOST_SPARC)
 730             {
 731                 char name[256];
 732                 int type;
 733                 int addend;
 734                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 735                     if (rel->r_offset >= start_offset &&
 736                         rel->r_offset < start_offset + copy_size) {
 737                         sym_name = strtab + symtab[ELF32_R_SYM(rel->r_info)].st_name;
 738                         if (strstart(sym_name, "__op_param", &p)) {
 739                             snprintf(name, sizeof(name), "param%s", p);
 740                         } else {
 741                                 if (sym_name[0] == '.')
 742                                         snprintf(name, sizeof(name),
 743                                                  "(long)(&__dot_%s)",
 744                                                  sym_name + 1);
 745                                 else
 746                                         snprintf(name, sizeof(name),
 747                                                  "(long)(&%s)", sym_name);
 748                         }
 749                         type = ELF32_R_TYPE(rel->r_info);
 750                         addend = rel->r_addend;
 751                         switch(type) {
 752                         case R_SPARC_32:
 753                             fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s + %d;\n",
 754                                     rel->r_offset - start_offset, name, addend);
 755                             break;
 756                         case R_SPARC_HI22:
 757                             fprintf(outfile,
 758                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 759                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 760                                     " & ~0x3fffff) "
 761                                     " | (((%s + %d) >> 10) & 0x3fffff);\n",
 762                                     rel->r_offset - start_offset,
 763                                     rel->r_offset - start_offset,
 764                                     name, addend);
 765                             break;
 766                         case R_SPARC_LO10:
 767                             fprintf(outfile,
 768                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 769                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 770                                     " & ~0x3ff) "
 771                                     " | ((%s + %d) & 0x3ff);\n",
 772                                     rel->r_offset - start_offset,
 773                                     rel->r_offset - start_offset,
 774                                     name, addend);
 775                             break;
 776                         case R_SPARC_WDISP30:
 777                             fprintf(outfile,
 778                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 779                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 780                                     " & ~0x3fffffff) "
 781                                     " | ((((%s + %d) - (long)(gen_code_ptr + %d))>>2) "
 782                                     "    & 0x3fffffff);\n",
 783                                     rel->r_offset - start_offset,
 784                                     rel->r_offset - start_offset,
 785                                     name, addend,
 786                                     rel->r_offset - start_offset);
 787                             break;
 788                         default:
 789                             error("unsupported sparc relocation (%d)", type);
 790                         }
 791                     }
 792                 }
 793             }
 794 #elif defined(HOST_SPARC64)
 795             {
 796                 char name[256];
 797                 int type;
 798                 int addend;
 799                 for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 800                     if (rel->r_offset >= start_offset &&
 801                         rel->r_offset < start_offset + copy_size) {
 802                         sym_name = strtab + symtab[ELF64_R_SYM(rel->r_info)].st_name;
 803                         if (strstart(sym_name, "__op_param", &p)) {
 804                             snprintf(name, sizeof(name), "param%s", p);
 805                         } else {
 806                             snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 807                         }
 808                         type = ELF64_R_TYPE(rel->r_info);
 809                         addend = rel->r_addend;
 810                         switch(type) {
 811                         case R_SPARC_32:
 812                             fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %d) = %s + %d;\n",
 813                                     rel->r_offset - start_offset, name, addend);
 814                             break;
 815                         case R_SPARC_HI22:
 816                             fprintf(outfile,
 817                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 818                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 819                                     " & ~0x3fffff) "
 820                                     " | (((%s + %d) >> 10) & 0x3fffff);\n",
 821                                     rel->r_offset - start_offset,
 822                                     rel->r_offset - start_offset,
 823                                     name, addend);
 824                             break;
 825                         case R_SPARC_LO10:
 826                             fprintf(outfile,
 827                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 828                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 829                                     " & ~0x3ff) "
 830                                     " | ((%s + %d) & 0x3ff);\n",
 831                                     rel->r_offset - start_offset,
 832                                     rel->r_offset - start_offset,
 833                                     name, addend);
 834                             break;
 835                         case R_SPARC_WDISP30:
 836                             fprintf(outfile,
 837                                     "    *(uint32_t *)(gen_code_ptr + %d) = "
 838                                     "((*(uint32_t *)(gen_code_ptr + %d)) "
 839                                     " & ~0x3fffffff) "
 840                                     " | ((((%s + %d) - (long)(gen_code_ptr + %d))>>2) "
 841                                     "    & 0x3fffffff);\n",
 842                                     rel->r_offset - start_offset,
 843                                     rel->r_offset - start_offset,
 844                                     name, addend,
 845                                     rel->r_offset - start_offset);
 846                             break;
 847                         default:
 848                             error("unsupported sparc64 relocation (%d)", type);
 849                         }
 850                     }
 851                 }
 852             }
 853 #else
 854 #error unsupported CPU
 855 #endif
 856         fprintf(outfile, "    gen_code_ptr += %d;\n", copy_size);
 857         fprintf(outfile, "}\n");
 858         fprintf(outfile, "break;\n\n");
 859     } else {
 860         fprintf(outfile, "static inline void gen_%s(", name);
 861         if (nb_args == 0) {
 862             fprintf(outfile, "void");
 863         } else {
 864             for(i = 0; i < nb_args; i++) {
 865                 if (i != 0)
 866                     fprintf(outfile, ", ");
 867                 fprintf(outfile, "long param%d", i + 1);
 868             }
 869         }
 870         fprintf(outfile, ")\n");
 871         fprintf(outfile, "{\n");
 872         for(i = 0; i < nb_args; i++) {
 873             fprintf(outfile, "    *gen_opparam_ptr++ = param%d;\n", i + 1);
 874         }
 875         fprintf(outfile, "    *gen_opc_ptr++ = INDEX_%s;\n", name);
 876         fprintf(outfile, "}\n\n");
 877     }
 878 }
 879
 880 /* load an elf object file */
 881 int load_elf(const char *filename, FILE *outfile, int do_print_enum)
 882 {
 883     int fd;
 884     struct elf_shdr *sec, *symtab_sec, *strtab_sec, *text_sec;
 885     int i, j;
 886     ElfW(Sym) *sym;
 887     char *shstr;
 888     uint8_t *text;
 889     void *relocs;
 890     int nb_relocs, reloc_sh_type;
 891
 892     fd = open(filename, O_RDONLY);
 893     if (fd < 0)
 894         error("can't open file '%s'", filename);
 895
 896     /* Read ELF header.  */
 897     if (read(fd, &ehdr, sizeof (ehdr)) != sizeof (ehdr))
 898         error("unable to read file header");
 899
 900     /* Check ELF identification.  */
 901     if (ehdr.e_ident[EI_MAG0] != ELFMAG0
 902      || ehdr.e_ident[EI_MAG1] != ELFMAG1
 903      || ehdr.e_ident[EI_MAG2] != ELFMAG2
 904      || ehdr.e_ident[EI_MAG3] != ELFMAG3
 905      || ehdr.e_ident[EI_VERSION] != EV_CURRENT) {
 906         error("bad ELF header");
 907     }
 908
 909     do_swap = elf_must_swap(&ehdr);
 910     if (do_swap)
 911         elf_swap_ehdr(&ehdr);
 912     if (ehdr.e_ident[EI_CLASS] != ELF_CLASS)
 913         error("Unsupported ELF class");
 914     if (ehdr.e_type != ET_REL)
 915         error("ELF object file expected");
 916     if (ehdr.e_version != EV_CURRENT)
 917         error("Invalid ELF version");
 918     if (!elf_check_arch(ehdr.e_machine))
 919         error("Unsupported CPU (e_machine=%d)", ehdr.e_machine);
 920
 921     /* read section headers */
 922     shdr = load_data(fd, ehdr.e_shoff, ehdr.e_shnum * sizeof(struct elf_shdr));
 923     if (do_swap) {
 924         for(i = 0; i < ehdr.e_shnum; i++) {
 925             elf_swap_shdr(&shdr[i]);
 926         }
 927     }
 928
 929     sec = &shdr[ehdr.e_shstrndx];
 930     shstr = load_data(fd, sec->sh_offset, sec->sh_size);
 931
 932     /* text section */
 933
 934     text_sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".text");
 935     if (!text_sec)
 936         error("could not find .text section");
 937     text = load_data(fd, text_sec->sh_offset, text_sec->sh_size);
 938
 939     data_shndx = -1;
 940     sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".data");
 941     if (sec) {
 942         data_shndx = sec - shdr;
 943         data_data = load_data(fd, sec->sh_offset, sec->sh_size);
 944     }
 945     sdata_shndx = -1;
 946     sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".sdata");
 947     if (sec) {
 948         sdata_shndx = sec - shdr;
 949         sdata_data = load_data(fd, sec->sh_offset, sec->sh_size);
 950     }
 951
 952     /* find text relocations, if any */
 953     nb_relocs = 0;
 954     relocs = NULL;
 955     reloc_sh_type = 0;
 956     for(i = 0; i < ehdr.e_shnum; i++) {
 957         sec = &shdr[i];
 958         if ((sec->sh_type == SHT_REL || sec->sh_type == SHT_RELA) &&
 959             sec->sh_info == (text_sec - shdr)) {
 960             reloc_sh_type = sec->sh_type;
 961             relocs = load_data(fd, sec->sh_offset, sec->sh_size);
 962             nb_relocs = sec->sh_size / sec->sh_entsize;
 963             if (do_swap) {
 964                 if (sec->sh_type == SHT_REL) {
 965                     ElfW(Rel) *rel = relocs;
 966                     for(j = 0, rel = relocs; j < nb_relocs; j++, rel++) {
 967                         swabls(&rel->r_offset);
 968                         swabls(&rel->r_info);
 969                     }
 970                 } else {
 971                     ElfW(Rela) *rel = relocs;
 972                     for(j = 0, rel = relocs; j < nb_relocs; j++, rel++) {
 973                         swabls(&rel->r_offset);
 974                         swabls(&rel->r_info);
 975                         swabls(&rel->r_addend);
 976                     }
 977                 }
 978             }
 979             break;
 980         }
 981     }
 982
 983     symtab_sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".symtab");
 984     if (!symtab_sec)
 985         error("could not find .symtab section");
 986     strtab_sec = &shdr[symtab_sec->sh_link];
 987
 988     symtab = load_data(fd, symtab_sec->sh_offset, symtab_sec->sh_size);
 989     strtab = load_data(fd, strtab_sec->sh_offset, strtab_sec->sh_size);
 990
 991     nb_syms = symtab_sec->sh_size / sizeof(ElfW(Sym));
 992     if (do_swap) {
 993         for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
 994             swab32s(&sym->st_name);
 995             swabls(&sym->st_value);
 996             swabls(&sym->st_size);
 997             swab16s(&sym->st_shndx);
 998         }
 999     }
1000
1001     if (do_print_enum) {
1002         fprintf(outfile, "DEF(end, 0, 0)\n");
1003         for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
1004             const char *name, *p;
1005             name = strtab + sym->st_name;
1006             if (strstart(name, OP_PREFIX, &p)) {
1007                 gen_code(name, sym->st_value, sym->st_size, outfile,
1008                          text, relocs, nb_relocs, reloc_sh_type, 2);
1009             }
1010         }
1011     } else {
1012         /* generate big code generation switch */
1013 #ifdef HOST_ALPHA
1014 fprintf(outfile,
1015 "register int gp asm(\"$29\");\n"
1016 "static inline void immediate_ldah(void *p, int val) {\n"
1017 "    uint32_t *dest = p;\n"
1018 "    long high = ((val >> 16) + ((val >> 15) & 1)) & 0xffff;\n"
1019 "\n"
1020 "    *dest &= ~0xffff;\n"
1021 "    *dest |= high;\n"
1022 "    *dest |= 31 << 16;\n"
1023 "}\n"
1024 "static inline void immediate_lda(void *dest, int val) {\n"
1025 "    *(uint16_t *) dest = val;\n"
1026 "}\n"
1027 "void fix_bsr(void *p, int offset) {\n"
1028 "    uint32_t *dest = p;\n"
1029 "    *dest &= ~((1 << 21) - 1);\n"
1030 "    *dest |= (offset >> 2) & ((1 << 21) - 1);\n"
1031 "}\n");
1032 #endif
1033 fprintf(outfile,
1034 "int dyngen_code(uint8_t *gen_code_buf,\n"
1035 "                uint16_t *label_offsets, uint16_t *jmp_offsets,\n"
1036 "                const uint16_t *opc_buf, const uint32_t *opparam_buf)\n"
1037 "{\n"
1038 "    uint8_t *gen_code_ptr;\n"
1039 "    const uint16_t *opc_ptr;\n"
1040 "    const uint32_t *opparam_ptr;\n"
1041 "    gen_code_ptr = gen_code_buf;\n"
1042 "    opc_ptr = opc_buf;\n"
1043 "    opparam_ptr = opparam_buf;\n");
1044
1045         /* Generate prologue, if needed. */
1046         switch(ELF_ARCH) {
1047         case EM_SPARC:
1048                 fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x9c23a080; /* sub %%sp, 128, %%sp */\n");
1049                 fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0xbc27a080; /* sub %%fp, 128, %%fp */\n");
1050                 break;
1051
1052         case EM_SPARCV9:
1053                 fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x9c23a100; /* sub %%sp, 256, %%sp */\n");
1054                 fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0xbc27a100; /* sub %%fp, 256, %%fp */\n");
1055                 break;
1056         };
1057
1058 fprintf(outfile,
1059 "    for(;;) {\n"
1060 "        switch(*opc_ptr++) {\n"
1061 );
1062
1063         for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
1064             const char *name;
1065             name = strtab + sym->st_name;
1066             if (strstart(name, OP_PREFIX, NULL)) {
1067 #if 0
1068                 printf("%4d: %s pos=0x%08x len=%d\n",
1069                        i, name, sym->st_value, sym->st_size);
1070 #endif
1071                 if (sym->st_shndx != (text_sec - shdr))
1072                     error("invalid section for opcode (0x%x)", sym->st_shndx);
1073                 gen_code(name, sym->st_value, sym->st_size, outfile,
1074                          text, relocs, nb_relocs, reloc_sh_type, 1);
1075             }
1076         }
1077
1078 fprintf(outfile,
1079 "        default:\n"
1080 "            goto the_end;\n"
1081 "        }\n"
1082 "    }\n"
1083 " the_end:\n"
1084 );
1085
1086 /* generate epilogue */
1087     switch(ELF_ARCH) {
1088     case EM_386:
1089         fprintf(outfile, "*gen_code_ptr++ = 0xc3; /* ret */\n");
1090         break;
1091     case EM_PPC:
1092         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x4e800020; /* blr */\n");
1093         break;
1094     case EM_S390:
1095         fprintf(outfile, "*((uint16_t *)gen_code_ptr)++ = 0x07fe; /* br %%r14 */\n");
1096         break;
1097     case EM_ALPHA:
1098         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x6bfa8001; /* ret */\n");
1099         break;
1100     case EM_IA_64:
1101         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x00840008; /* br.ret.sptk.many b0;; */\n");
1102         break;
1103     case EM_SPARC:
1104     case EM_SPARC32PLUS:
1105         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0xbc07a080; /* add %%fp, 256, %%fp */\n");
1106         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x81c62008; /* jmpl %%i0 + 8, %%g0 */\n");
1107         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x9c03a080; /* add %%sp, 256, %%sp */\n");
1108         break;
1109     case EM_SPARCV9:
1110         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x81c7e008; /* ret */\n");
1111         fprintf(outfile, "*((uint32_t *)gen_code_ptr)++ = 0x81e80000; /* restore */\n");
1112         break;
1113     default:
1114         error("unknown ELF architecture");
1115     }
1116
1117     fprintf(outfile, "return gen_code_ptr -  gen_code_buf;\n");
1118     fprintf(outfile, "}\n\n");
1119
1120 /* generate gen_xxx functions */
1121 /* XXX: suppress the use of these functions to simplify code */
1122         for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
1123             const char *name;
1124             name = strtab + sym->st_name;
1125             if (strstart(name, OP_PREFIX, NULL)) {
1126                 if (sym->st_shndx != (text_sec - shdr))
1127                     error("invalid section for opcode (0x%x)", sym->st_shndx);
1128                 gen_code(name, sym->st_value, sym->st_size, outfile,
1129                          text, relocs, nb_relocs, reloc_sh_type, 0);
1130             }
1131         }
1132     }
1133
1134     close(fd);
1135     return 0;
1136 }
1137
1138 void usage(void)
1139 {
1140     printf("dyngen (c) 2003 Fabrice Bellard\n"
1141            "usage: dyngen [-o outfile] [-c] objfile\n"
1142            "Generate a dynamic code generator from an object file\n"
1143            "-c     output enum of operations\n"
1144            );
1145     exit(1);
1146 }
1147
1148 int main(int argc, char **argv)
1149 {
1150     int c, do_print_enum;
1151     const char *filename, *outfilename;
1152     FILE *outfile;
1153
1154     outfilename = "out.c";
1155     do_print_enum = 0;
1156     for(;;) {
1157         c = getopt(argc, argv, "ho:c");
1158         if (c == -1)
1159             break;
1160         switch(c) {
1161         case 'h':
1162             usage();
1163             break;
1164         case 'o':
1165             outfilename = optarg;
1166             break;
1167         case 'c':
1168             do_print_enum = 1;
1169             break;
1170         }
1171     }
1172     if (optind >= argc)
1173         usage();
1174     filename = argv[optind];
1175     outfile = fopen(outfilename, "w");
1176     if (!outfile)
1177         error("could not open '%s'", outfilename);
1178     load_elf(filename, outfile, do_print_enum);
1179     fclose(outfile);
1180     return 0;
1181 }