| | 1 | /* |
| | 2 | * (C) GraÅžvydas "notaz" Ignotas, 2009-2010 |
| | 3 | * |
| | 4 | * This work is licensed under the terms of any of these licenses |
| | 5 | * (at your option): |
| | 6 | * - GNU GPL, version 2 or later. |
| | 7 | * - GNU LGPL, version 2.1 or later. |
| | 8 | * - MAME license. |
| | 9 | * See the COPYING file in the top-level directory. |
| | 10 | */ |
| | 11 | |
| | 12 | #include <stdio.h> |
| | 13 | #include <stdlib.h> |
| | 14 | #include <string.h> |
| | 15 | #include <stdarg.h> |
| | 16 | #include <bfd.h> |
| | 17 | #include <dis-asm.h> |
| | 18 | |
| | 19 | #include "host_dasm.h" |
| | 20 | |
| | 21 | extern char **g_argv; |
| | 22 | |
| | 23 | static struct disassemble_info di; |
| | 24 | static disassembler_ftype print_insn_func; |
| | 25 | |
| | 26 | #if defined __arm__ |
| | 27 | //#define print_insn_func print_insn_little_arm |
| | 28 | #define BFD_ARCH bfd_arch_arm |
| | 29 | #define BFD_MACH bfd_mach_arm_unknown |
| | 30 | #define DASM_OPTS "reg-names-std" |
| | 31 | #elif defined __aarch64__ |
| | 32 | //#define print_insn_func print_insn_aarch64 |
| | 33 | #define BFD_ARCH bfd_arch_aarch64 |
| | 34 | #define BFD_MACH bfd_mach_aarch64 |
| | 35 | #define DASM_OPTS NULL |
| | 36 | #elif defined __mips__ |
| | 37 | //#define print_insn_func print_insn_little_mips |
| | 38 | #define BFD_ARCH bfd_arch_mips |
| | 39 | #define BFD_MACH bfd_mach_mipsisa64r2 |
| | 40 | #define DASM_OPTS NULL |
| | 41 | #elif defined __riscv |
| | 42 | //#define print_insn_func print_insn_riscv |
| | 43 | #define BFD_ARCH bfd_arch_riscv |
| | 44 | #define BFD_MACH bfd_mach_riscv64 |
| | 45 | #define DASM_OPTS NULL |
| | 46 | #elif defined __powerpc__ |
| | 47 | //#define print_insn_func print_insn_little_powerpc |
| | 48 | #define BFD_ARCH bfd_arch_powerpc |
| | 49 | #define BFD_MACH bfd_mach_ppc64 |
| | 50 | #define DASM_OPTS NULL |
| | 51 | #elif defined(__x86_64__) || defined(__i386__) |
| | 52 | //#define print_insn_func print_insn_i386_intel |
| | 53 | #define BFD_ARCH bfd_arch_i386 |
| | 54 | #ifdef __x86_64__ |
| | 55 | #define BFD_MACH bfd_mach_x86_64_intel_syntax |
| | 56 | #else |
| | 57 | #define BFD_MACH bfd_mach_i386_i386_intel_syntax |
| | 58 | #endif |
| | 59 | #define DASM_OPTS NULL |
| | 60 | #else |
| | 61 | #error "missing arch support" |
| | 62 | #endif |
| | 63 | |
| | 64 | /* symbols */ |
| | 65 | static asymbol **symbols; |
| | 66 | static long symcount, symstorage; |
| | 67 | static int init_done; |
| | 68 | |
| | 69 | /* Filter out (in place) symbols that are useless for disassembly. |
| | 70 | COUNT is the number of elements in SYMBOLS. |
| | 71 | Return the number of useful symbols. */ |
| | 72 | static long |
| | 73 | remove_useless_symbols (asymbol **symbols, long count) |
| | 74 | { |
| | 75 | asymbol **in_ptr = symbols, **out_ptr = symbols; |
| | 76 | |
| | 77 | while (--count >= 0) |
| | 78 | { |
| | 79 | asymbol *sym = *in_ptr++; |
| | 80 | |
| | 81 | if (sym->name == NULL || sym->name[0] == '\0' || sym->name[0] == '$') |
| | 82 | continue; |
| | 83 | if (sym->flags & (BSF_DEBUGGING | BSF_SECTION_SYM)) |
| | 84 | continue; |
| | 85 | if (bfd_is_und_section (sym->section) |
| | 86 | || bfd_is_com_section (sym->section)) |
| | 87 | continue; |
| | 88 | if (sym->value + sym->section->vma == 0) |
| | 89 | continue; |
| | 90 | /* |
| | 91 | printf("sym: %08lx %04x %08x v %08x \"%s\"\n", |
| | 92 | (unsigned int)sym->value, (unsigned int)sym->flags, (unsigned int)sym->udata.i, |
| | 93 | (unsigned int)sym->section->vma, sym->name); |
| | 94 | */ |
| | 95 | *out_ptr++ = sym; |
| | 96 | } |
| | 97 | |
| | 98 | return out_ptr - symbols; |
| | 99 | } |
| | 100 | |
| | 101 | static void slurp_symtab(const char *filename) |
| | 102 | { |
| | 103 | bfd *abfd; |
| | 104 | |
| | 105 | symcount = 0; |
| | 106 | |
| | 107 | abfd = bfd_openr(filename, NULL); |
| | 108 | if (abfd == NULL) { |
| | 109 | fprintf(stderr, "failed to open: %s\n", filename); |
| | 110 | goto no_symbols; |
| | 111 | } |
| | 112 | |
| | 113 | if (!bfd_check_format(abfd, bfd_object)) |
| | 114 | goto no_symbols; |
| | 115 | |
| | 116 | if (!(bfd_get_file_flags(abfd) & HAS_SYMS)) |
| | 117 | goto no_symbols; |
| | 118 | |
| | 119 | symstorage = bfd_get_symtab_upper_bound(abfd); |
| | 120 | if (symstorage <= 0) |
| | 121 | goto no_symbols; |
| | 122 | |
| | 123 | symbols = malloc(symstorage); |
| | 124 | if (symbols == NULL) |
| | 125 | goto no_symbols; |
| | 126 | |
| | 127 | symcount = bfd_canonicalize_symtab(abfd, symbols); |
| | 128 | if (symcount < 0) |
| | 129 | goto no_symbols; |
| | 130 | |
| | 131 | symcount = remove_useless_symbols(symbols, symcount); |
| | 132 | // bfd_close(abfd); |
| | 133 | return; |
| | 134 | |
| | 135 | no_symbols: |
| | 136 | fprintf(stderr, "no symbols in %s\n", bfd_get_filename(abfd)); |
| | 137 | if (symbols != NULL) |
| | 138 | free(symbols); |
| | 139 | symbols = NULL; |
| | 140 | if (abfd != NULL) |
| | 141 | bfd_close(abfd); |
| | 142 | } |
| | 143 | |
| | 144 | static const char *lookup_name(bfd_vma addr) |
| | 145 | { |
| | 146 | asymbol **sptr = symbols; |
| | 147 | int i; |
| | 148 | |
| | 149 | for (i = 0; i < symcount; i++) { |
| | 150 | asymbol *sym = *sptr++; |
| | 151 | |
| | 152 | if (addr == sym->value + sym->section->vma) |
| | 153 | return sym->name; |
| | 154 | } |
| | 155 | |
| | 156 | return NULL; |
| | 157 | } |
| | 158 | |
| | 159 | /* Like target_read_memory, but slightly different parameters. */ |
| | 160 | static int |
| | 161 | dis_asm_read_memory(bfd_vma memaddr, bfd_byte *myaddr, unsigned int len, |
| | 162 | struct disassemble_info *info) |
| | 163 | { |
| | 164 | memcpy(myaddr, (void *)memaddr, len); |
| | 165 | return 0; |
| | 166 | } |
| | 167 | |
| | 168 | static void |
| | 169 | dis_asm_memory_error(int status, bfd_vma memaddr, |
| | 170 | struct disassemble_info *info) |
| | 171 | { |
| | 172 | fprintf(stderr, "memory_error %p\n", (void *)(long)memaddr); |
| | 173 | } |
| | 174 | |
| | 175 | static void |
| | 176 | dis_asm_print_address(bfd_vma addr, struct disassemble_info *info) |
| | 177 | { |
| | 178 | const char *name; |
| | 179 | |
| | 180 | printf("%08x", (int)addr); |
| | 181 | |
| | 182 | name = lookup_name(addr); |
| | 183 | if (name != NULL) |
| | 184 | printf(" <%s>", name); |
| | 185 | } |
| | 186 | |
| | 187 | static int insn_printf(void *f, const char *format, ...) |
| | 188 | { |
| | 189 | va_list args; |
| | 190 | size_t n; |
| | 191 | |
| | 192 | va_start(args, format); |
| | 193 | n = vprintf(format, args); |
| | 194 | va_end(args); |
| | 195 | |
| | 196 | return n; |
| | 197 | } |
| | 198 | |
| | 199 | static int print_insn_hex(bfd_vma addr, struct disassemble_info *info) |
| | 200 | { |
| | 201 | unsigned op; |
| | 202 | |
| | 203 | dis_asm_read_memory(addr, (bfd_byte *)&op, 4, info); |
| | 204 | printf("%p %08lx",(void *)addr, (long)op); |
| | 205 | return 4; |
| | 206 | } |
| | 207 | |
| | 208 | #ifdef BFD_COMPRESS_ZSTD |
| | 209 | static int insn_printf_styled(void *f, enum disassembler_style style, const char* format, ...) |
| | 210 | { |
| | 211 | va_list args; |
| | 212 | size_t n; |
| | 213 | |
| | 214 | va_start(args, format); |
| | 215 | n = vprintf(format, args); |
| | 216 | va_end(args); |
| | 217 | |
| | 218 | return n; |
| | 219 | } |
| | 220 | #endif |
| | 221 | |
| | 222 | static void host_dasm_init(void) |
| | 223 | { |
| | 224 | bfd_init(); |
| | 225 | if (g_argv && g_argv[0]) |
| | 226 | slurp_symtab(g_argv[0]); |
| | 227 | |
| | 228 | #ifdef BFD_COMPRESS_ZSTD |
| | 229 | init_disassemble_info(&di, NULL, insn_printf, insn_printf_styled); |
| | 230 | #else |
| | 231 | init_disassemble_info(&di, NULL, insn_printf); |
| | 232 | #endif |
| | 233 | di.flavour = bfd_target_unknown_flavour; |
| | 234 | di.memory_error_func = dis_asm_memory_error; |
| | 235 | di.print_address_func = dis_asm_print_address; |
| | 236 | // di.symbol_at_address_func = dis_asm_symbol_at_address; |
| | 237 | di.read_memory_func = dis_asm_read_memory; |
| | 238 | di.arch = BFD_ARCH; |
| | 239 | di.mach = BFD_MACH; |
| | 240 | di.endian = BFD_ENDIAN_LITTLE; |
| | 241 | di.disassembler_options = DASM_OPTS; |
| | 242 | disassemble_init_for_target(&di); |
| | 243 | #ifndef print_insn_func |
| | 244 | print_insn_func = disassembler(BFD_ARCH, 0, BFD_MACH, NULL); |
| | 245 | if (!print_insn_func) print_insn_func = print_insn_hex; |
| | 246 | #endif |
| | 247 | init_done = 1; |
| | 248 | } |
| | 249 | |
| | 250 | void host_dasm(void *addr, int len) |
| | 251 | { |
| | 252 | bfd_vma vma_end, vma = (bfd_vma)addr; |
| | 253 | const char *name; |
| | 254 | |
| | 255 | if (!init_done) |
| | 256 | host_dasm_init(); |
| | 257 | |
| | 258 | vma_end = vma + len; |
| | 259 | while (vma < vma_end) { |
| | 260 | name = lookup_name(vma); |
| | 261 | if (name != NULL) |
| | 262 | printf("%s:\n", name); |
| | 263 | |
| | 264 | printf(" %08lx ", (long)vma); |
| | 265 | vma += print_insn_func(vma, &di); |
| | 266 | printf("\n"); |
| | 267 | } |
| | 268 | } |
| | 269 | |
| | 270 | void host_dasm_new_symbol_(void *addr, const char *name) |
| | 271 | { |
| | 272 | bfd_vma vma = (bfd_vma)(long)addr; |
| | 273 | asymbol *sym, **tmp; |
| | 274 | |
| | 275 | if (!init_done) |
| | 276 | host_dasm_init(); |
| | 277 | if (symbols == NULL) |
| | 278 | return; |
| | 279 | if (symstorage <= symcount * sizeof(symbols[0])) { |
| | 280 | tmp = realloc(symbols, symstorage * 2); |
| | 281 | if (tmp == NULL) |
| | 282 | return; |
| | 283 | symstorage *= 2; |
| | 284 | symbols = tmp; |
| | 285 | } |
| | 286 | |
| | 287 | symbols[symcount] = calloc(sizeof(*symbols[0]), 1); |
| | 288 | if (symbols[symcount] == NULL) |
| | 289 | return; |
| | 290 | |
| | 291 | // a HACK (should use correct section), but ohwell |
| | 292 | sym = symbols[symcount]; |
| | 293 | sym->section = symbols[0]->section; |
| | 294 | sym->value = vma - sym->section->vma; |
| | 295 | sym->name = name; |
| | 296 | symcount++; |
| | 297 | } |
| | 298 | |