dyngen.c

   1 #include <stdlib.h>
   2 #include <stdio.h>
   3 #include <stdarg.h>
   4 #include <inttypes.h>
   5 #include <elf.h>
   6 #include <unistd.h>
   7 #include <fcntl.h>
   8
   9 #include "thunk.h"
  10
  11 /* all dynamically generated functions begin with this code */
  12 #define OP_PREFIX "op"
  13
  14 int elf_must_swap(Elf32_Ehdr *h)
  15 {
  16   union {
  17       uint32_t i;
  18       uint8_t b[4];
  19   } swaptest;
  20
  21   swaptest.i = 1;
  22   return (h->e_ident[EI_DATA] == ELFDATA2MSB) !=
  23       (swaptest.b[0] == 0);
  24 }
  25
  26 void swab16s(uint16_t *p)
  27 {
  28     *p = bswap16(*p);
  29 }
  30
  31 void swab32s(uint32_t *p)
  32 {
  33     *p = bswap32(*p);
  34 }
  35
  36 void swab64s(uint32_t *p)
  37 {
  38     *p = bswap64(*p);
  39 }
  40
  41 void elf_swap_ehdr(Elf32_Ehdr *h)
  42 {
  43     swab16s(&h->e_type);                        /* Object file type */
  44     swab16s(&h->        e_machine);             /* Architecture */
  45     swab32s(&h->        e_version);             /* Object file version */
  46     swab32s(&h->        e_entry);               /* Entry point virtual address */
  47     swab32s(&h->        e_phoff);               /* Program header table file offset */
  48     swab32s(&h->        e_shoff);               /* Section header table file offset */
  49     swab32s(&h->        e_flags);               /* Processor-specific flags */
  50     swab16s(&h->        e_ehsize);              /* ELF header size in bytes */
  51     swab16s(&h->        e_phentsize);           /* Program header table entry size */
  52     swab16s(&h->        e_phnum);               /* Program header table entry count */
  53     swab16s(&h->        e_shentsize);           /* Section header table entry size */
  54     swab16s(&h->        e_shnum);               /* Section header table entry count */
  55     swab16s(&h->        e_shstrndx);            /* Section header string table index */
  56 }
  57
  58 void elf_swap_shdr(Elf32_Shdr *h)
  59 {
  60   swab32s(&h->  sh_name);               /* Section name (string tbl index) */
  61   swab32s(&h->  sh_type);               /* Section type */
  62   swab32s(&h->  sh_flags);              /* Section flags */
  63   swab32s(&h->  sh_addr);               /* Section virtual addr at execution */
  64   swab32s(&h->  sh_offset);             /* Section file offset */
  65   swab32s(&h->  sh_size);               /* Section size in bytes */
  66   swab32s(&h->  sh_link);               /* Link to another section */
  67   swab32s(&h->  sh_info);               /* Additional section information */
  68   swab32s(&h->  sh_addralign);          /* Section alignment */
  69   swab32s(&h->  sh_entsize);            /* Entry size if section holds table */
  70 }
  71
  72 void elf_swap_phdr(Elf32_Phdr *h)
  73 {
  74     swab32s(&h->p_type);                        /* Segment type */
  75     swab32s(&h->p_offset);              /* Segment file offset */
  76     swab32s(&h->p_vaddr);               /* Segment virtual address */
  77     swab32s(&h->p_paddr);               /* Segment physical address */
  78     swab32s(&h->p_filesz);              /* Segment size in file */
  79     swab32s(&h->p_memsz);               /* Segment size in memory */
  80     swab32s(&h->p_flags);               /* Segment flags */
  81     swab32s(&h->p_align);               /* Segment alignment */
  82 }
  83
  84 int do_swap;
  85 int e_machine;
  86
  87 uint16_t get16(uint16_t *p)
  88 {
  89     uint16_t val;
  90     val = *p;
  91     if (do_swap)
  92         val = bswap16(val);
  93     return val;
  94 }
  95
  96 uint32_t get32(uint32_t *p)
  97 {
  98     uint32_t val;
  99     val = *p;
 100     if (do_swap)
 101         val = bswap32(val);
 102     return val;
 103 }
 104
 105 void put16(uint16_t *p, uint16_t val)
 106 {
 107     if (do_swap)
 108         val = bswap16(val);
 109     *p = val;
 110 }
 111
 112 void put32(uint32_t *p, uint32_t val)
 113 {
 114     if (do_swap)
 115         val = bswap32(val);
 116     *p = val;
 117 }
 118
 119 void __attribute__((noreturn)) error(const char *fmt, ...)
 120 {
 121     va_list ap;
 122     va_start(ap, fmt);
 123     fprintf(stderr, "dyngen: ");
 124     vfprintf(stderr, fmt, ap);
 125     fprintf(stderr, "\n");
 126     va_end(ap);
 127     exit(1);
 128 }
 129
 130
 131 Elf32_Shdr *find_elf_section(Elf32_Shdr *shdr, int shnum, const char *shstr,
 132                              const char *name)
 133 {
 134     int i;
 135     const char *shname;
 136     Elf32_Shdr *sec;
 137
 138     for(i = 0; i < shnum; i++) {
 139         sec = &shdr[i];
 140         if (!sec->sh_name)
 141             continue;
 142         shname = shstr + sec->sh_name;
 143         if (!strcmp(shname, name))
 144             return sec;
 145     }
 146     return NULL;
 147 }
 148
 149 void *load_data(int fd, long offset, unsigned int size)
 150 {
 151     char *data;
 152
 153     data = malloc(size);
 154     if (!data)
 155         return NULL;
 156     lseek(fd, offset, SEEK_SET);
 157     if (read(fd, data, size) != size) {
 158         free(data);
 159         return NULL;
 160     }
 161     return data;
 162 }
 163
 164 int strstart(const char *str, const char *val, const char **ptr)
 165 {
 166     const char *p, *q;
 167     p = str;
 168     q = val;
 169     while (*q != '\0') {
 170         if (*p != *q)
 171             return 0;
 172         p++;
 173         q++;
 174     }
 175     if (ptr)
 176         *ptr = p;
 177     return 1;
 178 }
 179
 180 #define MAX_ARGS 3
 181
 182 /* generate op code */
 183 void gen_code(const char *name, unsigned long offset, unsigned long size,
 184               FILE *outfile, uint8_t *text, void *relocs, int nb_relocs, int reloc_sh_type,
 185               Elf32_Sym *symtab, char *strtab)
 186 {
 187     int copy_size = 0;
 188     uint8_t *p_start, *p_end;
 189     int nb_args, i;
 190     uint8_t args_present[MAX_ARGS];
 191     const char *sym_name, *p;
 192
 193     /* compute exact size excluding return instruction */
 194     p_start = text + offset;
 195     p_end = p_start + size;
 196     switch(e_machine) {
 197     case EM_386:
 198         {
 199             uint8_t *p;
 200             p = p_end - 1;
 201             /* find ret */
 202             while (p > p_start && *p != 0xc3)
 203                 p--;
 204             /* skip double ret */
 205             if (p > p_start && p[-1] == 0xc3)
 206                 p--;
 207             if (p == p_start)
 208                 error("empty code for %s", name);
 209             copy_size = p - p_start;
 210         }
 211         break;
 212     case EM_PPC:
 213         {
 214             uint8_t *p;
 215             p = (void *)(p_end - 4);
 216             /* find ret */
 217             while (p > p_start && get32((uint32_t *)p) != 0x4e800020)
 218                 p -= 4;
 219             /* skip double ret */
 220             if (p > p_start && get32((uint32_t *)(p - 4)) == 0x4e800020)
 221                 p -= 4;
 222             if (p == p_start)
 223                 error("empty code for %s", name);
 224             copy_size = p - p_start;
 225         }
 226         break;
 227     default:
 228         error("unsupported CPU (%d)", e_machine);
 229     }
 230
 231     /* compute the number of arguments by looking at the relocations */
 232     for(i = 0;i < MAX_ARGS; i++)
 233         args_present[i] = 0;
 234
 235     if (reloc_sh_type == SHT_REL) {
 236         Elf32_Rel *rel;
 237         int n;
 238         for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 239             if (rel->r_offset >= offset && rel->r_offset < offset + copy_size) {
 240                 sym_name = strtab + symtab[ELF32_R_SYM(rel->r_info)].st_name;
 241                 if (strstart(sym_name, "__op_param", &p)) {
 242                     n = strtoul(p, NULL, 10);
 243                     if (n >= MAX_ARGS)
 244                         error("too many arguments in %s", name);
 245                     args_present[n - 1] = 1;
 246                 }
 247             }
 248         }
 249     } else {
 250         Elf32_Rela *rel;
 251         int n;
 252         for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 253             if (rel->r_offset >= offset && rel->r_offset < offset + copy_size) {
 254                 sym_name = strtab + symtab[ELF32_R_SYM(rel->r_info)].st_name;
 255                 if (strstart(sym_name, "__op_param", &p)) {
 256                     n = strtoul(p, NULL, 10);
 257                     if (n >= MAX_ARGS)
 258                         error("too many arguments in %s", name);
 259                     args_present[n - 1] = 1;
 260                 }
 261             }
 262         }
 263     }
 264
 265     nb_args = 0;
 266     while (nb_args < MAX_ARGS && args_present[nb_args])
 267         nb_args++;
 268     for(i = nb_args; i < MAX_ARGS; i++) {
 269         if (args_present[i])
 270             error("inconsistent argument numbering in %s", name);
 271     }
 272
 273     /* output C code */
 274     fprintf(outfile, "extern void %s();\n", name);
 275     fprintf(outfile, "static inline void gen_%s(", name);
 276     if (nb_args == 0) {
 277         fprintf(outfile, "void");
 278     } else {
 279         for(i = 0; i < nb_args; i++) {
 280             if (i != 0)
 281                 fprintf(outfile, ", ");
 282             fprintf(outfile, "long param%d", i + 1);
 283         }
 284     }
 285     fprintf(outfile, ")\n");
 286     fprintf(outfile, "{\n");
 287     fprintf(outfile, "    memcpy(gen_code_ptr, &%s, %d);\n", name, copy_size);
 288
 289     /* patch relocations */
 290     switch(e_machine) {
 291     case EM_386:
 292         {
 293             Elf32_Rel *rel;
 294             char name[256];
 295             int type;
 296             long addend;
 297             for(i = 0, rel = relocs;i < nb_relocs; i++, rel++) {
 298                 if (rel->r_offset >= offset && rel->r_offset < offset + copy_size) {
 299                     sym_name = strtab + symtab[ELF32_R_SYM(rel->r_info)].st_name;
 300                     if (strstart(sym_name, "__op_param", &p)) {
 301                         snprintf(name, sizeof(name), "param%s", p);
 302                     } else {
 303                         snprintf(name, sizeof(name), "(long)(&%s)", sym_name);
 304                     }
 305                     type = ELF32_R_TYPE(rel->r_info);
 306                     addend = get32((uint32_t *)(text + rel->r_offset));
 307                     switch(type) {
 308                     case R_386_32:
 309                         fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %ld) = %s + %ld;\n",
 310                                 rel->r_offset - offset, name, addend);
 311                         break;
 312                     case R_386_PC32:
 313                         fprintf(outfile, "    *(uint32_t *)(gen_code_ptr + %ld) = %s - (long)(gen_code_ptr + %ld) + %ld;\n",
 314                                 rel->r_offset - offset, name, rel->r_offset - offset, addend);
 315                         break;
 316                     default:
 317                         error("unsupported i386 relocation (%d)", type);
 318                     }
 319                 }
 320             }
 321         }
 322         break;
 323     default:
 324         error("unsupported CPU for relocations (%d)", e_machine);
 325     }
 326
 327
 328     fprintf(outfile, "    gen_code_ptr += %d;\n", copy_size);
 329     fprintf(outfile, "}\n\n");
 330 }
 331
 332 /* load an elf object file */
 333 int load_elf(const char *filename, FILE *outfile)
 334 {
 335     int fd;
 336     Elf32_Ehdr ehdr;
 337     Elf32_Shdr *sec, *shdr, *symtab_sec, *strtab_sec, *text_sec;
 338     int i, j, nb_syms;
 339     Elf32_Sym *symtab, *sym;
 340     const char *cpu_name;
 341     char *shstr, *strtab;
 342     uint8_t *text;
 343     void *relocs;
 344     int nb_relocs, reloc_sh_type;
 345
 346     fd = open(filename, O_RDONLY);
 347     if (fd < 0)
 348         error("can't open file '%s'", filename);
 349
 350     /* Read ELF header.  */
 351     if (read(fd, &ehdr, sizeof (ehdr)) != sizeof (ehdr))
 352         error("unable to read file header");
 353
 354     /* Check ELF identification.  */
 355     if (ehdr.e_ident[EI_MAG0] != ELFMAG0
 356      || ehdr.e_ident[EI_MAG1] != ELFMAG1
 357      || ehdr.e_ident[EI_MAG2] != ELFMAG2
 358      || ehdr.e_ident[EI_MAG3] != ELFMAG3
 359      || ehdr.e_ident[EI_CLASS] != ELFCLASS32
 360      || ehdr.e_ident[EI_VERSION] != EV_CURRENT) {
 361         error("bad ELF header");
 362     }
 363
 364     do_swap = elf_must_swap(&ehdr);
 365     if (do_swap)
 366         elf_swap_ehdr(&ehdr);
 367     if (ehdr.e_type != ET_REL)
 368         error("ELF object file expected");
 369     if (ehdr.e_version != EV_CURRENT)
 370         error("Invalid ELF version");
 371     e_machine = ehdr.e_machine;
 372
 373     /* read section headers */
 374     shdr = load_data(fd, ehdr.e_shoff, ehdr.e_shnum * sizeof(Elf32_Shdr));
 375     if (do_swap) {
 376         for(i = 0; i < ehdr.e_shnum; i++) {
 377             elf_swap_shdr(&shdr[i]);
 378         }
 379     }
 380
 381     sec = &shdr[ehdr.e_shstrndx];
 382     shstr = load_data(fd, sec->sh_offset, sec->sh_size);
 383
 384     /* text section */
 385
 386     text_sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".text");
 387     if (!text_sec)
 388         error("could not find .text section");
 389     text = load_data(fd, text_sec->sh_offset, text_sec->sh_size);
 390
 391     /* find text relocations, if any */
 392     nb_relocs = 0;
 393     relocs = NULL;
 394     reloc_sh_type = 0;
 395     for(i = 0; i < ehdr.e_shnum; i++) {
 396         sec = &shdr[i];
 397         if ((sec->sh_type == SHT_REL || sec->sh_type == SHT_RELA) &&
 398             sec->sh_info == (text_sec - shdr)) {
 399             reloc_sh_type = sec->sh_type;
 400             relocs = load_data(fd, sec->sh_offset, sec->sh_size);
 401             nb_relocs = sec->sh_size / sec->sh_entsize;
 402             if (do_swap) {
 403                 if (sec->sh_type == SHT_REL) {
 404                     Elf32_Rel *rel = relocs;
 405                     for(j = 0, rel = relocs; j < nb_relocs; j++, rel++) {
 406                         swab32s(&rel->r_offset);
 407                         swab32s(&rel->r_info);
 408                     }
 409                 } else {
 410                     Elf32_Rela *rel = relocs;
 411                     for(j = 0, rel = relocs; j < nb_relocs; j++, rel++) {
 412                         swab32s(&rel->r_offset);
 413                         swab32s(&rel->r_info);
 414                         swab32s(&rel->r_addend);
 415                     }
 416                 }
 417             }
 418             break;
 419         }
 420     }
 421
 422     symtab_sec = find_elf_section(shdr, ehdr.e_shnum, shstr, ".symtab");
 423     if (!symtab_sec)
 424         error("could not find .symtab section");
 425     strtab_sec = &shdr[symtab_sec->sh_link];
 426
 427     symtab = load_data(fd, symtab_sec->sh_offset, symtab_sec->sh_size);
 428     strtab = load_data(fd, strtab_sec->sh_offset, strtab_sec->sh_size);
 429
 430     nb_syms = symtab_sec->sh_size / sizeof(Elf32_Sym);
 431     if (do_swap) {
 432         for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
 433             swab32s(&sym->st_name);
 434             swab32s(&sym->st_value);
 435             swab32s(&sym->st_size);
 436             swab16s(&sym->st_shndx);
 437         }
 438     }
 439
 440     switch(e_machine) {
 441     case EM_386:
 442         cpu_name = "i386";
 443         break;
 444     case EM_PPC:
 445         cpu_name = "ppc";
 446         break;
 447     case EM_MIPS:
 448         cpu_name = "mips";
 449         break;
 450     case EM_ARM:
 451         cpu_name = "arm";
 452         break;
 453     case EM_SPARC:
 454         cpu_name = "sparc";
 455         break;
 456     default:
 457         error("unsupported CPU (e_machine=%d)", e_machine);
 458     }
 459
 460     fprintf(outfile, "#include \"gen-%s.h\"\n\n", cpu_name);
 461
 462     for(i = 0, sym = symtab; i < nb_syms; i++, sym++) {
 463         const char *name;
 464         name = strtab + sym->st_name;
 465         if (strstart(name, "op_", NULL) ||
 466             strstart(name, "op1_", NULL) ||
 467             strstart(name, "op2_", NULL) ||
 468             strstart(name, "op3_", NULL)) {
 469 #if 0
 470             printf("%4d: %s pos=0x%08x len=%d\n",
 471                    i, name, sym->st_value, sym->st_size);
 472 #endif
 473             if (sym->st_shndx != (text_sec - shdr))
 474                 error("invalid section for opcode (0x%x)", sym->st_shndx);
 475             gen_code(name, sym->st_value, sym->st_size, outfile,
 476                      text, relocs, nb_relocs, reloc_sh_type, symtab, strtab);
 477         }
 478     }
 479
 480     close(fd);
 481     return 0;
 482 }
 483
 484 void usage(void)
 485 {
 486     printf("dyngen (c) 2003 Fabrice Bellard\n"
 487            "usage: dyngen [-o outfile] objfile\n"
 488            "Generate a dynamic code generator from an object file\n");
 489     exit(1);
 490 }
 491
 492 int main(int argc, char **argv)
 493 {
 494     int c;
 495     const char *filename, *outfilename;
 496     FILE *outfile;
 497
 498     outfilename = "out.c";
 499     for(;;) {
 500         c = getopt(argc, argv, "ho:");
 501         if (c == -1)
 502             break;
 503         switch(c) {
 504         case 'h':
 505             usage();
 506             break;
 507         case 'o':
 508             outfilename = optarg;
 509             break;
 510         }
 511     }
 512     if (optind >= argc)
 513         usage();
 514     filename = argv[optind];
 515     outfile = fopen(outfilename, "w");
 516     if (!outfile)
 517         error("could not open '%s'", outfilename);
 518     load_elf(filename, outfile);
 519     fclose(outfile);
 520     return 0;
 521 }