H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 1 | /* ndisasm.c the Netwide Disassembler main module |
| 2 | * |
| 3 | * The Netwide Assembler is copyright (C) 1996 Simon Tatham and |
| 4 | * Julian Hall. All rights reserved. The software is |
| 5 | * redistributable under the licence given in the file "Licence" |
| 6 | * distributed in the NASM archive. |
| 7 | */ |
| 8 | |
| 9 | #include <stdio.h> |
H. Peter Anvin | d7ed89e | 2002-04-30 20:52:08 +0000 | [diff] [blame] | 10 | #include <stdlib.h> |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 11 | #include <string.h> |
| 12 | #include <ctype.h> |
H. Peter Anvin | 76690a1 | 2002-04-30 20:52:49 +0000 | [diff] [blame] | 13 | #include <errno.h> |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 14 | |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 15 | #include "insns.h" |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 16 | #include "nasm.h" |
| 17 | #include "nasmlib.h" |
| 18 | #include "sync.h" |
| 19 | #include "disasm.h" |
| 20 | |
| 21 | #define BPL 8 /* bytes per line of hex dump */ |
| 22 | |
| 23 | static const char *help = |
| 24 | "usage: ndisasm [-a] [-i] [-h] [-r] [-u] [-b bits] [-o origin] [-s sync...]\n" |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 25 | " [-e bytes] [-k start,bytes] [-p vendor] file\n" |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 26 | " -a or -i activates auto (intelligent) sync\n" |
| 27 | " -u sets USE32 (32-bit mode)\n" |
| 28 | " -b 16 or -b 32 sets number of bits too\n" |
| 29 | " -h displays this text\n" |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 30 | " -r or -v displays the version number\n" |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 31 | " -e skips <bytes> bytes of header\n" |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 32 | " -k avoids disassembling <bytes> bytes from position <start>\n" |
H. Peter Anvin | ef7468f | 2002-04-30 20:57:59 +0000 | [diff] [blame] | 33 | " -p selects the preferred vendor instruction set (intel, amd, cyrix, idt)\n"; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 34 | |
| 35 | static void output_ins (unsigned long, unsigned char *, int, char *); |
| 36 | static void skip (unsigned long dist, FILE *fp); |
| 37 | |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 38 | int main(int argc, char **argv) |
| 39 | { |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 40 | unsigned char buffer[INSN_MAX * 2], *p, *q; |
| 41 | char outbuf[256]; |
| 42 | char *pname = *argv; |
| 43 | char *filename = NULL; |
| 44 | unsigned long nextsync, synclen, initskip = 0L; |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 45 | int lenread; |
| 46 | long lendis; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 47 | int autosync = FALSE; |
| 48 | int bits = 16; |
H. Peter Anvin | 87bc619 | 2002-04-30 20:53:16 +0000 | [diff] [blame] | 49 | int eof = FALSE; |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 50 | unsigned long prefer = 0; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 51 | int rn_error; |
| 52 | long offset; |
| 53 | FILE *fp; |
| 54 | |
| 55 | offset = 0; |
| 56 | init_sync(); |
| 57 | |
| 58 | while (--argc) { |
| 59 | char *v, *vv, *p = *++argv; |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 60 | if (*p == '-' && p[1]) { |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 61 | p++; |
| 62 | while (*p) switch (tolower(*p)) { |
| 63 | case 'a': /* auto or intelligent sync */ |
| 64 | case 'i': |
| 65 | autosync = TRUE; |
| 66 | p++; |
| 67 | break; |
| 68 | case 'h': |
| 69 | fprintf(stderr, help); |
| 70 | return 0; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 71 | case 'r': |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 72 | case 'v': |
| 73 | fprintf(stderr, "NDISASM version %s compiled " __DATE__ "\n", NASM_VER); |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 74 | return 0; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 75 | case 'u': /* USE32 */ |
| 76 | bits = 32; |
| 77 | p++; |
| 78 | break; |
| 79 | case 'b': /* bits */ |
| 80 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 81 | if (!v) { |
| 82 | fprintf(stderr, "%s: `-b' requires an argument\n", pname); |
| 83 | return 1; |
| 84 | } |
| 85 | if (!strcmp(v, "16")) |
| 86 | bits = 16; |
| 87 | else if (!strcmp(v, "32")) |
| 88 | bits = 32; |
| 89 | else { |
| 90 | fprintf(stderr, "%s: argument to `-b' should" |
| 91 | " be `16' or `32'\n", pname); |
| 92 | } |
| 93 | p = ""; /* force to next argument */ |
| 94 | break; |
| 95 | case 'o': /* origin */ |
| 96 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 97 | if (!v) { |
| 98 | fprintf(stderr, "%s: `-o' requires an argument\n", pname); |
| 99 | return 1; |
| 100 | } |
| 101 | offset = readnum (v, &rn_error); |
| 102 | if (rn_error) { |
| 103 | fprintf(stderr, "%s: `-o' requires a numeric argument\n", |
| 104 | pname); |
| 105 | return 1; |
| 106 | } |
| 107 | p = ""; /* force to next argument */ |
| 108 | break; |
| 109 | case 's': /* sync point */ |
| 110 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 111 | if (!v) { |
| 112 | fprintf(stderr, "%s: `-s' requires an argument\n", pname); |
| 113 | return 1; |
| 114 | } |
| 115 | add_sync (readnum (v, &rn_error), 0L); |
| 116 | if (rn_error) { |
| 117 | fprintf(stderr, "%s: `-s' requires a numeric argument\n", |
| 118 | pname); |
| 119 | return 1; |
| 120 | } |
| 121 | p = ""; /* force to next argument */ |
| 122 | break; |
| 123 | case 'e': /* skip a header */ |
| 124 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 125 | if (!v) { |
| 126 | fprintf(stderr, "%s: `-e' requires an argument\n", pname); |
| 127 | return 1; |
| 128 | } |
| 129 | initskip = readnum (v, &rn_error); |
| 130 | if (rn_error) { |
| 131 | fprintf(stderr, "%s: `-e' requires a numeric argument\n", |
| 132 | pname); |
| 133 | return 1; |
| 134 | } |
| 135 | p = ""; /* force to next argument */ |
| 136 | break; |
| 137 | case 'k': /* skip a region */ |
| 138 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 139 | if (!v) { |
| 140 | fprintf(stderr, "%s: `-k' requires an argument\n", pname); |
| 141 | return 1; |
| 142 | } |
| 143 | vv = strchr(v, ','); |
| 144 | if (!vv) { |
| 145 | fprintf(stderr, "%s: `-k' requires two numbers separated" |
| 146 | " by a comma\n", pname); |
| 147 | return 1; |
| 148 | } |
| 149 | *vv++ = '\0'; |
| 150 | nextsync = readnum (v, &rn_error); |
| 151 | if (rn_error) { |
| 152 | fprintf(stderr, "%s: `-k' requires numeric arguments\n", |
| 153 | pname); |
| 154 | return 1; |
| 155 | } |
| 156 | synclen = readnum (vv, &rn_error); |
| 157 | if (rn_error) { |
| 158 | fprintf(stderr, "%s: `-k' requires numeric arguments\n", |
| 159 | pname); |
| 160 | return 1; |
| 161 | } |
| 162 | add_sync (nextsync, synclen); |
| 163 | p = ""; /* force to next argument */ |
| 164 | break; |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 165 | case 'p': /* preferred vendor */ |
| 166 | v = p[1] ? p+1 : --argc ? *++argv : NULL; |
| 167 | if (!v) { |
| 168 | fprintf(stderr, "%s: `-p' requires an argument\n", pname); |
| 169 | return 1; |
| 170 | } |
| 171 | if ( !strcmp(v, "intel") ) { |
| 172 | prefer = 0; /* Default */ |
| 173 | } else if ( !strcmp(v, "amd") ) { |
| 174 | prefer = IF_AMD|IF_3DNOW; |
| 175 | } else if ( !strcmp(v, "cyrix") ) { |
| 176 | prefer = IF_CYRIX|IF_3DNOW; |
H. Peter Anvin | 620515a | 2002-04-30 20:57:38 +0000 | [diff] [blame] | 177 | } else if ( !strcmp(v, "idt") || !strcmp(v, "centaur") || |
| 178 | !strcmp(v, "winchip") ) { |
| 179 | prefer = IF_3DNOW; |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 180 | } else { |
| 181 | fprintf(stderr, "%s: unknown vendor `%s' specified with `-p'\n", pname, v); |
| 182 | return 1; |
| 183 | } |
| 184 | p = ""; /* force to next argument */ |
| 185 | break; |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 186 | default: /*bf*/ |
| 187 | fprintf(stderr, "%s: unrecognised option `-%c'\n", |
| 188 | pname, *p); |
| 189 | return 1; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 190 | } |
| 191 | } else if (!filename) { |
| 192 | filename = p; |
| 193 | } else { |
| 194 | fprintf(stderr, "%s: more than one filename specified\n", pname); |
| 195 | return 1; |
| 196 | } |
| 197 | } |
| 198 | |
| 199 | if (!filename) { |
| 200 | fprintf(stderr, help, pname); |
| 201 | return 0; |
| 202 | } |
| 203 | |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 204 | if (strcmp(filename, "-")) { |
| 205 | fp = fopen(filename, "rb"); |
| 206 | if (!fp) { |
| 207 | fprintf(stderr, "%s: unable to open `%s': %s\n", |
| 208 | pname, filename, strerror(errno)); |
| 209 | return 1; |
| 210 | } |
| 211 | } else |
| 212 | fp = stdin; |
| 213 | |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 214 | if (initskip > 0) |
| 215 | skip (initskip, fp); |
| 216 | |
| 217 | /* |
| 218 | * This main loop is really horrible, and wants rewriting with |
| 219 | * an axe. It'll stay the way it is for a while though, until I |
| 220 | * find the energy... |
| 221 | */ |
| 222 | |
| 223 | p = q = buffer; |
| 224 | nextsync = next_sync (offset, &synclen); |
| 225 | do { |
| 226 | unsigned long to_read = buffer+sizeof(buffer)-p; |
| 227 | if (to_read > nextsync-offset-(p-q)) |
| 228 | to_read = nextsync-offset-(p-q); |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 229 | if (to_read) { |
| 230 | lenread = fread (p, 1, to_read, fp); |
| 231 | if (lenread == 0) |
| 232 | eof = TRUE; /* help along systems with bad feof */ |
| 233 | } else |
| 234 | lenread = 0; |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 235 | p += lenread; |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 236 | if ((unsigned long)offset == nextsync) { |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 237 | if (synclen) { |
| 238 | printf("%08lX skipping 0x%lX bytes\n", offset, synclen); |
| 239 | offset += synclen; |
| 240 | skip (synclen, fp); |
| 241 | } |
| 242 | p = q = buffer; |
| 243 | nextsync = next_sync (offset, &synclen); |
| 244 | } |
| 245 | while (p > q && (p - q >= INSN_MAX || lenread == 0)) { |
H. Peter Anvin | 4836e33 | 2002-04-30 20:56:43 +0000 | [diff] [blame] | 246 | lendis = disasm (q, outbuf, bits, offset, autosync, prefer); |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 247 | if (!lendis || lendis > (p - q) || |
H. Peter Anvin | 4cf1748 | 2002-04-30 21:01:38 +0000 | [diff] [blame^] | 248 | (unsigned long)lendis > nextsync-offset) |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 249 | lendis = eatbyte (q, outbuf); |
| 250 | output_ins (offset, q, lendis, outbuf); |
| 251 | q += lendis; |
| 252 | offset += lendis; |
| 253 | } |
| 254 | if (q >= buffer+INSN_MAX) { |
| 255 | unsigned char *r = buffer, *s = q; |
| 256 | int count = p - q; |
| 257 | while (count--) |
| 258 | *r++ = *s++; |
| 259 | p -= (q - buffer); |
| 260 | q = buffer; |
| 261 | } |
H. Peter Anvin | 87bc619 | 2002-04-30 20:53:16 +0000 | [diff] [blame] | 262 | } while (lenread > 0 || !(eof || feof(fp))); |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 263 | |
| 264 | if (fp != stdin) |
| 265 | fclose (fp); |
| 266 | |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 267 | return 0; |
| 268 | } |
| 269 | |
| 270 | static void output_ins (unsigned long offset, unsigned char *data, |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 271 | int datalen, char *insn) |
| 272 | { |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 273 | int bytes; |
| 274 | printf("%08lX ", offset); |
| 275 | |
| 276 | bytes = 0; |
| 277 | while (datalen > 0 && bytes < BPL) { |
| 278 | printf("%02X", *data++); |
| 279 | bytes++; |
| 280 | datalen--; |
| 281 | } |
| 282 | |
| 283 | printf("%*s%s\n", (BPL+1-bytes)*2, "", insn); |
| 284 | |
| 285 | while (datalen > 0) { |
| 286 | printf(" -"); |
| 287 | bytes = 0; |
| 288 | while (datalen > 0 && bytes < BPL) { |
| 289 | printf("%02X", *data++); |
| 290 | bytes++; |
| 291 | datalen--; |
| 292 | } |
| 293 | printf("\n"); |
| 294 | } |
| 295 | } |
| 296 | |
| 297 | /* |
| 298 | * Skip a certain amount of data in a file, either by seeking if |
| 299 | * possible, or if that fails then by reading and discarding. |
| 300 | */ |
H. Peter Anvin | eba20a7 | 2002-04-30 20:53:55 +0000 | [diff] [blame] | 301 | static void skip (unsigned long dist, FILE *fp) |
| 302 | { |
H. Peter Anvin | ea6e34d | 2002-04-30 20:51:32 +0000 | [diff] [blame] | 303 | char buffer[256]; /* should fit on most stacks :-) */ |
| 304 | |
| 305 | /* |
| 306 | * Got to be careful with fseek: at least one fseek I've tried |
| 307 | * doesn't approve of SEEK_CUR. So I'll use SEEK_SET and |
| 308 | * ftell... horrible but apparently necessary. |
| 309 | */ |
| 310 | if (fseek (fp, dist+ftell(fp), SEEK_SET)) { |
| 311 | while (dist > 0) { |
| 312 | unsigned long len = (dist < sizeof(buffer) ? |
| 313 | dist : sizeof(buffer)); |
| 314 | if (fread (buffer, 1, len, fp) < len) { |
| 315 | perror("fread"); |
| 316 | exit(1); |
| 317 | } |
| 318 | dist -= len; |
| 319 | } |
| 320 | } |
| 321 | } |