|
1 /* |
|
2 * Copyright (c) 2010 The WebM project authors. All Rights Reserved. |
|
3 * |
|
4 * Use of this source code is governed by a BSD-style license |
|
5 * that can be found in the LICENSE file in the root of the source |
|
6 * tree. An additional intellectual property rights grant can be found |
|
7 * in the file PATENTS. All contributing project authors may |
|
8 * be found in the AUTHORS file in the root of the source tree. |
|
9 */ |
|
10 |
|
11 |
|
12 #include <stdarg.h> |
|
13 #include <stdio.h> |
|
14 #include <stdlib.h> |
|
15 #include <string.h> |
|
16 |
|
17 #include "vpx_config.h" |
|
18 #include "vpx/vpx_integer.h" |
|
19 |
|
20 typedef enum { |
|
21 OUTPUT_FMT_PLAIN, |
|
22 OUTPUT_FMT_RVDS, |
|
23 OUTPUT_FMT_GAS, |
|
24 } output_fmt_t; |
|
25 |
|
26 int log_msg(const char *fmt, ...) { |
|
27 int res; |
|
28 va_list ap; |
|
29 va_start(ap, fmt); |
|
30 res = vfprintf(stderr, fmt, ap); |
|
31 va_end(ap); |
|
32 return res; |
|
33 } |
|
34 |
|
35 #if defined(__GNUC__) && __GNUC__ |
|
36 #if defined(__MACH__) |
|
37 |
|
38 #include <mach-o/loader.h> |
|
39 #include <mach-o/nlist.h> |
|
40 |
|
41 int print_macho_equ(output_fmt_t mode, uint8_t* name, int val) { |
|
42 switch (mode) { |
|
43 case OUTPUT_FMT_RVDS: |
|
44 printf("%-40s EQU %5d\n", name, val); |
|
45 return 0; |
|
46 case OUTPUT_FMT_GAS: |
|
47 printf(".set %-40s, %5d\n", name, val); |
|
48 return 0; |
|
49 default: |
|
50 log_msg("Unsupported mode: %d", mode); |
|
51 return 1; |
|
52 } |
|
53 } |
|
54 |
|
55 int parse_macho(uint8_t *base_buf, size_t sz, output_fmt_t mode) { |
|
56 int i, j; |
|
57 struct mach_header header; |
|
58 uint8_t *buf = base_buf; |
|
59 int base_data_section = 0; |
|
60 int bits = 0; |
|
61 |
|
62 /* We can read in mach_header for 32 and 64 bit architectures |
|
63 * because it's identical to mach_header_64 except for the last |
|
64 * element (uint32_t reserved), which we don't use. Then, when |
|
65 * we know which architecture we're looking at, increment buf |
|
66 * appropriately. |
|
67 */ |
|
68 memcpy(&header, buf, sizeof(struct mach_header)); |
|
69 |
|
70 if (header.magic == MH_MAGIC) { |
|
71 if (header.cputype == CPU_TYPE_ARM |
|
72 || header.cputype == CPU_TYPE_X86) { |
|
73 bits = 32; |
|
74 buf += sizeof(struct mach_header); |
|
75 } else { |
|
76 log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_[ARM|X86].\n"); |
|
77 goto bail; |
|
78 } |
|
79 } else if (header.magic == MH_MAGIC_64) { |
|
80 if (header.cputype == CPU_TYPE_X86_64) { |
|
81 bits = 64; |
|
82 buf += sizeof(struct mach_header_64); |
|
83 } else { |
|
84 log_msg("Bad cputype for object file. Currently only tested for CPU_TYPE_X86_64.\n"); |
|
85 goto bail; |
|
86 } |
|
87 } else { |
|
88 log_msg("Bad magic number for object file. 0x%x or 0x%x expected, 0x%x found.\n", |
|
89 MH_MAGIC, MH_MAGIC_64, header.magic); |
|
90 goto bail; |
|
91 } |
|
92 |
|
93 if (header.filetype != MH_OBJECT) { |
|
94 log_msg("Bad filetype for object file. Currently only tested for MH_OBJECT.\n"); |
|
95 goto bail; |
|
96 } |
|
97 |
|
98 for (i = 0; i < header.ncmds; i++) { |
|
99 struct load_command lc; |
|
100 |
|
101 memcpy(&lc, buf, sizeof(struct load_command)); |
|
102 |
|
103 if (lc.cmd == LC_SEGMENT) { |
|
104 uint8_t *seg_buf = buf; |
|
105 struct section s; |
|
106 struct segment_command seg_c; |
|
107 |
|
108 memcpy(&seg_c, seg_buf, sizeof(struct segment_command)); |
|
109 seg_buf += sizeof(struct segment_command); |
|
110 |
|
111 /* Although each section is given it's own offset, nlist.n_value |
|
112 * references the offset of the first section. This isn't |
|
113 * apparent without debug information because the offset of the |
|
114 * data section is the same as the first section. However, with |
|
115 * debug sections mixed in, the offset of the debug section |
|
116 * increases but n_value still references the first section. |
|
117 */ |
|
118 if (seg_c.nsects < 1) { |
|
119 log_msg("Not enough sections\n"); |
|
120 goto bail; |
|
121 } |
|
122 |
|
123 memcpy(&s, seg_buf, sizeof(struct section)); |
|
124 base_data_section = s.offset; |
|
125 } else if (lc.cmd == LC_SEGMENT_64) { |
|
126 uint8_t *seg_buf = buf; |
|
127 struct section_64 s; |
|
128 struct segment_command_64 seg_c; |
|
129 |
|
130 memcpy(&seg_c, seg_buf, sizeof(struct segment_command_64)); |
|
131 seg_buf += sizeof(struct segment_command_64); |
|
132 |
|
133 /* Explanation in LG_SEGMENT */ |
|
134 if (seg_c.nsects < 1) { |
|
135 log_msg("Not enough sections\n"); |
|
136 goto bail; |
|
137 } |
|
138 |
|
139 memcpy(&s, seg_buf, sizeof(struct section_64)); |
|
140 base_data_section = s.offset; |
|
141 } else if (lc.cmd == LC_SYMTAB) { |
|
142 if (base_data_section != 0) { |
|
143 struct symtab_command sc; |
|
144 uint8_t *sym_buf = base_buf; |
|
145 uint8_t *str_buf = base_buf; |
|
146 |
|
147 memcpy(&sc, buf, sizeof(struct symtab_command)); |
|
148 |
|
149 if (sc.cmdsize != sizeof(struct symtab_command)) { |
|
150 log_msg("Can't find symbol table!\n"); |
|
151 goto bail; |
|
152 } |
|
153 |
|
154 sym_buf += sc.symoff; |
|
155 str_buf += sc.stroff; |
|
156 |
|
157 for (j = 0; j < sc.nsyms; j++) { |
|
158 /* Location of string is cacluated each time from the |
|
159 * start of the string buffer. On darwin the symbols |
|
160 * are prefixed by "_", so we bump the pointer by 1. |
|
161 * The target value is defined as an int in *_asm_*_offsets.c, |
|
162 * which is 4 bytes on all targets we currently use. |
|
163 */ |
|
164 if (bits == 32) { |
|
165 struct nlist nl; |
|
166 int val; |
|
167 |
|
168 memcpy(&nl, sym_buf, sizeof(struct nlist)); |
|
169 sym_buf += sizeof(struct nlist); |
|
170 |
|
171 memcpy(&val, base_buf + base_data_section + nl.n_value, |
|
172 sizeof(val)); |
|
173 print_macho_equ(mode, str_buf + nl.n_un.n_strx + 1, val); |
|
174 } else { /* if (bits == 64) */ |
|
175 struct nlist_64 nl; |
|
176 int val; |
|
177 |
|
178 memcpy(&nl, sym_buf, sizeof(struct nlist_64)); |
|
179 sym_buf += sizeof(struct nlist_64); |
|
180 |
|
181 memcpy(&val, base_buf + base_data_section + nl.n_value, |
|
182 sizeof(val)); |
|
183 print_macho_equ(mode, str_buf + nl.n_un.n_strx + 1, val); |
|
184 } |
|
185 } |
|
186 } |
|
187 } |
|
188 |
|
189 buf += lc.cmdsize; |
|
190 } |
|
191 |
|
192 return 0; |
|
193 bail: |
|
194 return 1; |
|
195 |
|
196 } |
|
197 |
|
198 #elif defined(__ELF__) |
|
199 #include "elf.h" |
|
200 |
|
201 #define COPY_STRUCT(dst, buf, ofst, sz) do {\ |
|
202 if(ofst + sizeof((*(dst))) > sz) goto bail;\ |
|
203 memcpy(dst, buf+ofst, sizeof((*(dst))));\ |
|
204 } while(0) |
|
205 |
|
206 #define ENDIAN_ASSIGN(val, memb) do {\ |
|
207 if(!elf->le_data) {log_msg("Big Endian data not supported yet!\n");goto bail;}\ |
|
208 (val) = (memb);\ |
|
209 } while(0) |
|
210 |
|
211 #define ENDIAN_ASSIGN_IN_PLACE(memb) do {\ |
|
212 ENDIAN_ASSIGN(memb, memb);\ |
|
213 } while(0) |
|
214 |
|
215 typedef struct { |
|
216 uint8_t *buf; /* Buffer containing ELF data */ |
|
217 size_t sz; /* Buffer size */ |
|
218 int le_data; /* Data is little-endian */ |
|
219 unsigned char e_ident[EI_NIDENT]; /* Magic number and other info */ |
|
220 int bits; /* 32 or 64 */ |
|
221 Elf32_Ehdr hdr32; |
|
222 Elf64_Ehdr hdr64; |
|
223 } elf_obj_t; |
|
224 |
|
225 int parse_elf_header(elf_obj_t *elf) { |
|
226 int res; |
|
227 /* Verify ELF Magic numbers */ |
|
228 COPY_STRUCT(&elf->e_ident, elf->buf, 0, elf->sz); |
|
229 res = elf->e_ident[EI_MAG0] == ELFMAG0; |
|
230 res &= elf->e_ident[EI_MAG1] == ELFMAG1; |
|
231 res &= elf->e_ident[EI_MAG2] == ELFMAG2; |
|
232 res &= elf->e_ident[EI_MAG3] == ELFMAG3; |
|
233 res &= elf->e_ident[EI_CLASS] == ELFCLASS32 |
|
234 || elf->e_ident[EI_CLASS] == ELFCLASS64; |
|
235 res &= elf->e_ident[EI_DATA] == ELFDATA2LSB; |
|
236 |
|
237 if (!res) goto bail; |
|
238 |
|
239 elf->le_data = elf->e_ident[EI_DATA] == ELFDATA2LSB; |
|
240 |
|
241 /* Read in relevant values */ |
|
242 if (elf->e_ident[EI_CLASS] == ELFCLASS32) { |
|
243 elf->bits = 32; |
|
244 COPY_STRUCT(&elf->hdr32, elf->buf, 0, elf->sz); |
|
245 |
|
246 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_type); |
|
247 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_machine); |
|
248 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_version); |
|
249 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_entry); |
|
250 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phoff); |
|
251 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shoff); |
|
252 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_flags); |
|
253 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_ehsize); |
|
254 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phentsize); |
|
255 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_phnum); |
|
256 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shentsize); |
|
257 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shnum); |
|
258 ENDIAN_ASSIGN_IN_PLACE(elf->hdr32.e_shstrndx); |
|
259 } else { /* if (elf->e_ident[EI_CLASS] == ELFCLASS64) */ |
|
260 elf->bits = 64; |
|
261 COPY_STRUCT(&elf->hdr64, elf->buf, 0, elf->sz); |
|
262 |
|
263 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_type); |
|
264 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_machine); |
|
265 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_version); |
|
266 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_entry); |
|
267 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phoff); |
|
268 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shoff); |
|
269 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_flags); |
|
270 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_ehsize); |
|
271 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phentsize); |
|
272 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_phnum); |
|
273 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shentsize); |
|
274 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shnum); |
|
275 ENDIAN_ASSIGN_IN_PLACE(elf->hdr64.e_shstrndx); |
|
276 } |
|
277 |
|
278 return 0; |
|
279 bail: |
|
280 log_msg("Failed to parse ELF file header"); |
|
281 return 1; |
|
282 } |
|
283 |
|
284 int parse_elf_section(elf_obj_t *elf, int idx, Elf32_Shdr *hdr32, Elf64_Shdr *hdr64) { |
|
285 if (hdr32) { |
|
286 if (idx >= elf->hdr32.e_shnum) |
|
287 goto bail; |
|
288 |
|
289 COPY_STRUCT(hdr32, elf->buf, elf->hdr32.e_shoff + idx * elf->hdr32.e_shentsize, |
|
290 elf->sz); |
|
291 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_name); |
|
292 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_type); |
|
293 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_flags); |
|
294 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addr); |
|
295 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_offset); |
|
296 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_size); |
|
297 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_link); |
|
298 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_info); |
|
299 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_addralign); |
|
300 ENDIAN_ASSIGN_IN_PLACE(hdr32->sh_entsize); |
|
301 } else { /* if (hdr64) */ |
|
302 if (idx >= elf->hdr64.e_shnum) |
|
303 goto bail; |
|
304 |
|
305 COPY_STRUCT(hdr64, elf->buf, elf->hdr64.e_shoff + idx * elf->hdr64.e_shentsize, |
|
306 elf->sz); |
|
307 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_name); |
|
308 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_type); |
|
309 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_flags); |
|
310 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addr); |
|
311 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_offset); |
|
312 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_size); |
|
313 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_link); |
|
314 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_info); |
|
315 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_addralign); |
|
316 ENDIAN_ASSIGN_IN_PLACE(hdr64->sh_entsize); |
|
317 } |
|
318 |
|
319 return 0; |
|
320 bail: |
|
321 return 1; |
|
322 } |
|
323 |
|
324 char *parse_elf_string_table(elf_obj_t *elf, int s_idx, int idx) { |
|
325 if (elf->bits == 32) { |
|
326 Elf32_Shdr shdr; |
|
327 |
|
328 if (parse_elf_section(elf, s_idx, &shdr, NULL)) { |
|
329 log_msg("Failed to parse ELF string table: section %d, index %d\n", |
|
330 s_idx, idx); |
|
331 return ""; |
|
332 } |
|
333 |
|
334 return (char *)(elf->buf + shdr.sh_offset + idx); |
|
335 } else { /* if (elf->bits == 64) */ |
|
336 Elf64_Shdr shdr; |
|
337 |
|
338 if (parse_elf_section(elf, s_idx, NULL, &shdr)) { |
|
339 log_msg("Failed to parse ELF string table: section %d, index %d\n", |
|
340 s_idx, idx); |
|
341 return ""; |
|
342 } |
|
343 |
|
344 return (char *)(elf->buf + shdr.sh_offset + idx); |
|
345 } |
|
346 } |
|
347 |
|
348 int parse_elf_symbol(elf_obj_t *elf, unsigned int ofst, Elf32_Sym *sym32, Elf64_Sym *sym64) { |
|
349 if (sym32) { |
|
350 COPY_STRUCT(sym32, elf->buf, ofst, elf->sz); |
|
351 ENDIAN_ASSIGN_IN_PLACE(sym32->st_name); |
|
352 ENDIAN_ASSIGN_IN_PLACE(sym32->st_value); |
|
353 ENDIAN_ASSIGN_IN_PLACE(sym32->st_size); |
|
354 ENDIAN_ASSIGN_IN_PLACE(sym32->st_info); |
|
355 ENDIAN_ASSIGN_IN_PLACE(sym32->st_other); |
|
356 ENDIAN_ASSIGN_IN_PLACE(sym32->st_shndx); |
|
357 } else { /* if (sym64) */ |
|
358 COPY_STRUCT(sym64, elf->buf, ofst, elf->sz); |
|
359 ENDIAN_ASSIGN_IN_PLACE(sym64->st_name); |
|
360 ENDIAN_ASSIGN_IN_PLACE(sym64->st_value); |
|
361 ENDIAN_ASSIGN_IN_PLACE(sym64->st_size); |
|
362 ENDIAN_ASSIGN_IN_PLACE(sym64->st_info); |
|
363 ENDIAN_ASSIGN_IN_PLACE(sym64->st_other); |
|
364 ENDIAN_ASSIGN_IN_PLACE(sym64->st_shndx); |
|
365 } |
|
366 return 0; |
|
367 bail: |
|
368 return 1; |
|
369 } |
|
370 |
|
371 int parse_elf(uint8_t *buf, size_t sz, output_fmt_t mode) { |
|
372 elf_obj_t elf; |
|
373 unsigned int ofst; |
|
374 int i; |
|
375 Elf32_Off strtab_off32; |
|
376 Elf64_Off strtab_off64; /* save String Table offset for later use */ |
|
377 |
|
378 memset(&elf, 0, sizeof(elf)); |
|
379 elf.buf = buf; |
|
380 elf.sz = sz; |
|
381 |
|
382 /* Parse Header */ |
|
383 if (parse_elf_header(&elf)) |
|
384 goto bail; |
|
385 |
|
386 if (elf.bits == 32) { |
|
387 Elf32_Shdr shdr; |
|
388 for (i = 0; i < elf.hdr32.e_shnum; i++) { |
|
389 parse_elf_section(&elf, i, &shdr, NULL); |
|
390 |
|
391 if (shdr.sh_type == SHT_STRTAB) { |
|
392 char strtsb_name[128]; |
|
393 |
|
394 strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name)); |
|
395 |
|
396 if (!(strcmp(strtsb_name, ".shstrtab"))) { |
|
397 /* log_msg("found section: %s\n", strtsb_name); */ |
|
398 strtab_off32 = shdr.sh_offset; |
|
399 break; |
|
400 } |
|
401 } |
|
402 } |
|
403 } else { /* if (elf.bits == 64) */ |
|
404 Elf64_Shdr shdr; |
|
405 for (i = 0; i < elf.hdr64.e_shnum; i++) { |
|
406 parse_elf_section(&elf, i, NULL, &shdr); |
|
407 |
|
408 if (shdr.sh_type == SHT_STRTAB) { |
|
409 char strtsb_name[128]; |
|
410 |
|
411 strcpy(strtsb_name, (char *)(elf.buf + shdr.sh_offset + shdr.sh_name)); |
|
412 |
|
413 if (!(strcmp(strtsb_name, ".shstrtab"))) { |
|
414 /* log_msg("found section: %s\n", strtsb_name); */ |
|
415 strtab_off64 = shdr.sh_offset; |
|
416 break; |
|
417 } |
|
418 } |
|
419 } |
|
420 } |
|
421 |
|
422 /* Parse all Symbol Tables */ |
|
423 if (elf.bits == 32) { |
|
424 Elf32_Shdr shdr; |
|
425 for (i = 0; i < elf.hdr32.e_shnum; i++) { |
|
426 parse_elf_section(&elf, i, &shdr, NULL); |
|
427 |
|
428 if (shdr.sh_type == SHT_SYMTAB) { |
|
429 for (ofst = shdr.sh_offset; |
|
430 ofst < shdr.sh_offset + shdr.sh_size; |
|
431 ofst += shdr.sh_entsize) { |
|
432 Elf32_Sym sym; |
|
433 |
|
434 parse_elf_symbol(&elf, ofst, &sym, NULL); |
|
435 |
|
436 /* For all OBJECTS (data objects), extract the value from the |
|
437 * proper data segment. |
|
438 */ |
|
439 /* if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name) |
|
440 log_msg("found data object %s\n", |
|
441 parse_elf_string_table(&elf, |
|
442 shdr.sh_link, |
|
443 sym.st_name)); |
|
444 */ |
|
445 |
|
446 if (ELF32_ST_TYPE(sym.st_info) == STT_OBJECT |
|
447 && sym.st_size == 4) { |
|
448 Elf32_Shdr dhdr; |
|
449 int val = 0; |
|
450 char section_name[128]; |
|
451 |
|
452 parse_elf_section(&elf, sym.st_shndx, &dhdr, NULL); |
|
453 |
|
454 /* For explanition - refer to _MSC_VER version of code */ |
|
455 strcpy(section_name, (char *)(elf.buf + strtab_off32 + dhdr.sh_name)); |
|
456 /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */ |
|
457 |
|
458 if (strcmp(section_name, ".bss")) { |
|
459 if (sizeof(val) != sym.st_size) { |
|
460 /* The target value is declared as an int in |
|
461 * *_asm_*_offsets.c, which is 4 bytes on all |
|
462 * targets we currently use. Complain loudly if |
|
463 * this is not true. |
|
464 */ |
|
465 log_msg("Symbol size is wrong\n"); |
|
466 goto bail; |
|
467 } |
|
468 |
|
469 memcpy(&val, |
|
470 elf.buf + dhdr.sh_offset + sym.st_value, |
|
471 sym.st_size); |
|
472 } |
|
473 |
|
474 if (!elf.le_data) { |
|
475 log_msg("Big Endian data not supported yet!\n"); |
|
476 goto bail; |
|
477 } |
|
478 |
|
479 switch (mode) { |
|
480 case OUTPUT_FMT_RVDS: |
|
481 printf("%-40s EQU %5d\n", |
|
482 parse_elf_string_table(&elf, |
|
483 shdr.sh_link, |
|
484 sym.st_name), |
|
485 val); |
|
486 break; |
|
487 case OUTPUT_FMT_GAS: |
|
488 printf(".equ %-40s, %5d\n", |
|
489 parse_elf_string_table(&elf, |
|
490 shdr.sh_link, |
|
491 sym.st_name), |
|
492 val); |
|
493 break; |
|
494 default: |
|
495 printf("%s = %d\n", |
|
496 parse_elf_string_table(&elf, |
|
497 shdr.sh_link, |
|
498 sym.st_name), |
|
499 val); |
|
500 } |
|
501 } |
|
502 } |
|
503 } |
|
504 } |
|
505 } else { /* if (elf.bits == 64) */ |
|
506 Elf64_Shdr shdr; |
|
507 for (i = 0; i < elf.hdr64.e_shnum; i++) { |
|
508 parse_elf_section(&elf, i, NULL, &shdr); |
|
509 |
|
510 if (shdr.sh_type == SHT_SYMTAB) { |
|
511 for (ofst = shdr.sh_offset; |
|
512 ofst < shdr.sh_offset + shdr.sh_size; |
|
513 ofst += shdr.sh_entsize) { |
|
514 Elf64_Sym sym; |
|
515 |
|
516 parse_elf_symbol(&elf, ofst, NULL, &sym); |
|
517 |
|
518 /* For all OBJECTS (data objects), extract the value from the |
|
519 * proper data segment. |
|
520 */ |
|
521 /* if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT && sym.st_name) |
|
522 log_msg("found data object %s\n", |
|
523 parse_elf_string_table(&elf, |
|
524 shdr.sh_link, |
|
525 sym.st_name)); |
|
526 */ |
|
527 |
|
528 if (ELF64_ST_TYPE(sym.st_info) == STT_OBJECT |
|
529 && sym.st_size == 4) { |
|
530 Elf64_Shdr dhdr; |
|
531 int val = 0; |
|
532 char section_name[128]; |
|
533 |
|
534 parse_elf_section(&elf, sym.st_shndx, NULL, &dhdr); |
|
535 |
|
536 /* For explanition - refer to _MSC_VER version of code */ |
|
537 strcpy(section_name, (char *)(elf.buf + strtab_off64 + dhdr.sh_name)); |
|
538 /* log_msg("Section_name: %s, Section_type: %d\n", section_name, dhdr.sh_type); */ |
|
539 |
|
540 if ((strcmp(section_name, ".bss"))) { |
|
541 if (sizeof(val) != sym.st_size) { |
|
542 /* The target value is declared as an int in |
|
543 * *_asm_*_offsets.c, which is 4 bytes on all |
|
544 * targets we currently use. Complain loudly if |
|
545 * this is not true. |
|
546 */ |
|
547 log_msg("Symbol size is wrong\n"); |
|
548 goto bail; |
|
549 } |
|
550 |
|
551 memcpy(&val, |
|
552 elf.buf + dhdr.sh_offset + sym.st_value, |
|
553 sym.st_size); |
|
554 } |
|
555 |
|
556 if (!elf.le_data) { |
|
557 log_msg("Big Endian data not supported yet!\n"); |
|
558 goto bail; |
|
559 } |
|
560 |
|
561 switch (mode) { |
|
562 case OUTPUT_FMT_RVDS: |
|
563 printf("%-40s EQU %5d\n", |
|
564 parse_elf_string_table(&elf, |
|
565 shdr.sh_link, |
|
566 sym.st_name), |
|
567 val); |
|
568 break; |
|
569 case OUTPUT_FMT_GAS: |
|
570 printf(".equ %-40s, %5d\n", |
|
571 parse_elf_string_table(&elf, |
|
572 shdr.sh_link, |
|
573 sym.st_name), |
|
574 val); |
|
575 break; |
|
576 default: |
|
577 printf("%s = %d\n", |
|
578 parse_elf_string_table(&elf, |
|
579 shdr.sh_link, |
|
580 sym.st_name), |
|
581 val); |
|
582 } |
|
583 } |
|
584 } |
|
585 } |
|
586 } |
|
587 } |
|
588 |
|
589 if (mode == OUTPUT_FMT_RVDS) |
|
590 printf(" END\n"); |
|
591 |
|
592 return 0; |
|
593 bail: |
|
594 log_msg("Parse error: File does not appear to be valid ELF32 or ELF64\n"); |
|
595 return 1; |
|
596 } |
|
597 |
|
598 #endif |
|
599 #endif /* defined(__GNUC__) && __GNUC__ */ |
|
600 |
|
601 |
|
602 #if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__) |
|
603 /* See "Microsoft Portable Executable and Common Object File Format Specification" |
|
604 for reference. |
|
605 */ |
|
606 #define get_le32(x) ((*(x)) | (*(x+1)) << 8 |(*(x+2)) << 16 | (*(x+3)) << 24 ) |
|
607 #define get_le16(x) ((*(x)) | (*(x+1)) << 8) |
|
608 |
|
609 int parse_coff(uint8_t *buf, size_t sz) { |
|
610 unsigned int nsections, symtab_ptr, symtab_sz, strtab_ptr; |
|
611 unsigned int sectionrawdata_ptr; |
|
612 unsigned int i; |
|
613 uint8_t *ptr; |
|
614 uint32_t symoffset; |
|
615 |
|
616 char **sectionlist; // this array holds all section names in their correct order. |
|
617 // it is used to check if the symbol is in .bss or .rdata section. |
|
618 |
|
619 nsections = get_le16(buf + 2); |
|
620 symtab_ptr = get_le32(buf + 8); |
|
621 symtab_sz = get_le32(buf + 12); |
|
622 strtab_ptr = symtab_ptr + symtab_sz * 18; |
|
623 |
|
624 if (nsections > 96) { |
|
625 log_msg("Too many sections\n"); |
|
626 return 1; |
|
627 } |
|
628 |
|
629 sectionlist = malloc(nsections * sizeof(sectionlist)); |
|
630 |
|
631 if (sectionlist == NULL) { |
|
632 log_msg("Allocating first level of section list failed\n"); |
|
633 return 1; |
|
634 } |
|
635 |
|
636 // log_msg("COFF: Found %u symbols in %u sections.\n", symtab_sz, nsections); |
|
637 |
|
638 /* |
|
639 The size of optional header is always zero for an obj file. So, the section header |
|
640 follows the file header immediately. |
|
641 */ |
|
642 |
|
643 ptr = buf + 20; // section header |
|
644 |
|
645 for (i = 0; i < nsections; i++) { |
|
646 char sectionname[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0}; |
|
647 strncpy(sectionname, ptr, 8); |
|
648 // log_msg("COFF: Parsing section %s\n",sectionname); |
|
649 |
|
650 sectionlist[i] = malloc(strlen(sectionname) + 1); |
|
651 |
|
652 if (sectionlist[i] == NULL) { |
|
653 log_msg("Allocating storage for %s failed\n", sectionname); |
|
654 goto bail; |
|
655 } |
|
656 strcpy(sectionlist[i], sectionname); |
|
657 |
|
658 if (!strcmp(sectionname, ".rdata")) sectionrawdata_ptr = get_le32(ptr + 20); |
|
659 |
|
660 ptr += 40; |
|
661 } |
|
662 |
|
663 // log_msg("COFF: Symbol table at offset %u\n", symtab_ptr); |
|
664 // log_msg("COFF: raw data pointer ofset for section .rdata is %u\n", sectionrawdata_ptr); |
|
665 |
|
666 /* The compiler puts the data with non-zero offset in .rdata section, but puts the data with |
|
667 zero offset in .bss section. So, if the data in in .bss section, set offset=0. |
|
668 Note from Wiki: In an object module compiled from C, the bss section contains |
|
669 the local variables (but not functions) that were declared with the static keyword, |
|
670 except for those with non-zero initial values. (In C, static variables are initialized |
|
671 to zero by default.) It also contains the non-local (both extern and static) variables |
|
672 that are also initialized to zero (either explicitly or by default). |
|
673 */ |
|
674 // move to symbol table |
|
675 /* COFF symbol table: |
|
676 offset field |
|
677 0 Name(*) |
|
678 8 Value |
|
679 12 SectionNumber |
|
680 14 Type |
|
681 16 StorageClass |
|
682 17 NumberOfAuxSymbols |
|
683 */ |
|
684 ptr = buf + symtab_ptr; |
|
685 |
|
686 for (i = 0; i < symtab_sz; i++) { |
|
687 int16_t section = get_le16(ptr + 12); // section number |
|
688 |
|
689 if (section > 0 && ptr[16] == 2) { |
|
690 // if(section > 0 && ptr[16] == 3 && get_le32(ptr+8)) { |
|
691 |
|
692 if (get_le32(ptr)) { |
|
693 char name[9] = {0, 0, 0, 0, 0, 0, 0, 0, 0}; |
|
694 strncpy(name, ptr, 8); |
|
695 // log_msg("COFF: Parsing symbol %s\n",name); |
|
696 /* The 64bit Windows compiler doesn't prefix with an _. |
|
697 * Check what's there, and bump if necessary |
|
698 */ |
|
699 if (name[0] == '_') |
|
700 printf("%-40s EQU ", name + 1); |
|
701 else |
|
702 printf("%-40s EQU ", name); |
|
703 } else { |
|
704 // log_msg("COFF: Parsing symbol %s\n", |
|
705 // buf + strtab_ptr + get_le32(ptr+4)); |
|
706 if ((buf + strtab_ptr + get_le32(ptr + 4))[0] == '_') |
|
707 printf("%-40s EQU ", |
|
708 buf + strtab_ptr + get_le32(ptr + 4) + 1); |
|
709 else |
|
710 printf("%-40s EQU ", buf + strtab_ptr + get_le32(ptr + 4)); |
|
711 } |
|
712 |
|
713 if (!(strcmp(sectionlist[section - 1], ".bss"))) { |
|
714 symoffset = 0; |
|
715 } else { |
|
716 symoffset = get_le32(buf + sectionrawdata_ptr + get_le32(ptr + 8)); |
|
717 } |
|
718 |
|
719 // log_msg(" Section: %d\n",section); |
|
720 // log_msg(" Class: %d\n",ptr[16]); |
|
721 // log_msg(" Address: %u\n",get_le32(ptr+8)); |
|
722 // log_msg(" Offset: %u\n", symoffset); |
|
723 |
|
724 printf("%5d\n", symoffset); |
|
725 } |
|
726 |
|
727 ptr += 18; |
|
728 } |
|
729 |
|
730 printf(" END\n"); |
|
731 |
|
732 for (i = 0; i < nsections; i++) { |
|
733 free(sectionlist[i]); |
|
734 } |
|
735 |
|
736 free(sectionlist); |
|
737 |
|
738 return 0; |
|
739 bail: |
|
740 |
|
741 for (i = 0; i < nsections; i++) { |
|
742 free(sectionlist[i]); |
|
743 } |
|
744 |
|
745 free(sectionlist); |
|
746 |
|
747 return 1; |
|
748 } |
|
749 #endif /* defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__) */ |
|
750 |
|
751 int main(int argc, char **argv) { |
|
752 output_fmt_t mode = OUTPUT_FMT_PLAIN; |
|
753 const char *f; |
|
754 uint8_t *file_buf; |
|
755 int res; |
|
756 FILE *fp; |
|
757 long int file_size; |
|
758 |
|
759 if (argc < 2 || argc > 3) { |
|
760 fprintf(stderr, "Usage: %s [output format] <obj file>\n\n", argv[0]); |
|
761 fprintf(stderr, " <obj file>\tobject file to parse\n"); |
|
762 fprintf(stderr, "Output Formats:\n"); |
|
763 fprintf(stderr, " gas - compatible with GNU assembler\n"); |
|
764 fprintf(stderr, " rvds - compatible with armasm\n"); |
|
765 goto bail; |
|
766 } |
|
767 |
|
768 f = argv[2]; |
|
769 |
|
770 if (!strcmp(argv[1], "rvds")) |
|
771 mode = OUTPUT_FMT_RVDS; |
|
772 else if (!strcmp(argv[1], "gas")) |
|
773 mode = OUTPUT_FMT_GAS; |
|
774 else |
|
775 f = argv[1]; |
|
776 |
|
777 fp = fopen(f, "rb"); |
|
778 |
|
779 if (!fp) { |
|
780 perror("Unable to open file"); |
|
781 goto bail; |
|
782 } |
|
783 |
|
784 if (fseek(fp, 0, SEEK_END)) { |
|
785 perror("stat"); |
|
786 goto bail; |
|
787 } |
|
788 |
|
789 file_size = ftell(fp); |
|
790 file_buf = malloc(file_size); |
|
791 |
|
792 if (!file_buf) { |
|
793 perror("malloc"); |
|
794 goto bail; |
|
795 } |
|
796 |
|
797 rewind(fp); |
|
798 |
|
799 if (fread(file_buf, sizeof(char), file_size, fp) != file_size) { |
|
800 perror("read"); |
|
801 goto bail; |
|
802 } |
|
803 |
|
804 if (fclose(fp)) { |
|
805 perror("close"); |
|
806 goto bail; |
|
807 } |
|
808 |
|
809 #if defined(__GNUC__) && __GNUC__ |
|
810 #if defined(__MACH__) |
|
811 res = parse_macho(file_buf, file_size, mode); |
|
812 #elif defined(__ELF__) |
|
813 res = parse_elf(file_buf, file_size, mode); |
|
814 #endif |
|
815 #endif |
|
816 #if defined(_MSC_VER) || defined(__MINGW32__) || defined(__CYGWIN__) |
|
817 res = parse_coff(file_buf, file_size); |
|
818 #endif |
|
819 |
|
820 free(file_buf); |
|
821 |
|
822 if (!res) |
|
823 return EXIT_SUCCESS; |
|
824 |
|
825 bail: |
|
826 return EXIT_FAILURE; |
|
827 } |