1 /* input_scrub.c - Break up input buffers into whole numbers of lines. 2 Copyright (C) 1987-2016 Free Software Foundation, Inc. 3 4 This file is part of GAS, the GNU Assembler. 5 6 GAS 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 3, or (at your option) 9 any later version. 10 11 GAS 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 GAS; see the file COPYING. If not, write to the Free 18 Software Foundation, 51 Franklin Street - Fifth Floor, Boston, MA 19 02110-1301, USA. */ 20 21 #include "as.h" 22 #include "filenames.h" 23 #include "input-file.h" 24 #include "sb.h" 25 #include "listing.h" 26 27 /* 28 * O/S independent module to supply buffers of sanitised source code 29 * to rest of assembler. We get sanitised input data of arbitrary length. 30 * We break these buffers on line boundaries, recombine pieces that 31 * were broken across buffers, and return a buffer of full lines to 32 * the caller. 33 * The last partial line begins the next buffer we build and return to caller. 34 * The buffer returned to caller is preceded by BEFORE_STRING and followed 35 * by AFTER_STRING, as sentinels. The last character before AFTER_STRING 36 * is a newline. 37 * Also looks after line numbers, for e.g. error messages. 38 */ 39 40 /* 41 * We don't care how filthy our buffers are, but our callers assume 42 * that the following sanitation has already been done. 43 * 44 * No comments, reduce a comment to a space. 45 * Reduce a tab to a space unless it is 1st char of line. 46 * All multiple tabs and spaces collapsed into 1 char. Tab only 47 * legal if 1st char of line. 48 * # line file statements converted to .line x;.file y; statements. 49 * Escaped newlines at end of line: remove them but add as many newlines 50 * to end of statement as you removed in the middle, to synch line numbers. 51 */ 52 53 #define BEFORE_STRING ("\n") 55 #define AFTER_STRING ("\0") /* memcpy of 0 chars might choke. */ 56 #define BEFORE_SIZE (1) 57 #define AFTER_SIZE (1) 58 59 #ifndef TC_EOL_IN_INSN 60 #define TC_EOL_IN_INSN(P) 0 61 #endif 62 63 static char *buffer_start; /*->1st char of full buffer area. */ 64 static char *partial_where; /*->after last full line in buffer. */ 65 static int partial_size; /* >=0. Number of chars in partial line in buffer. */ 66 67 /* Because we need AFTER_STRING just after last full line, it clobbers 68 1st part of partial line. So we preserve 1st part of partial line 69 here. */ 70 static char save_source[AFTER_SIZE]; 71 72 /* What is the largest size buffer that input_file_give_next_buffer() 73 could return to us? */ 74 static unsigned int buffer_length; 75 76 /* The index into an sb structure we are reading from. -1 if none. */ 77 static size_t sb_index = -1; 78 79 /* If we are reading from an sb structure, this is it. */ 80 static sb from_sb; 81 82 /* Should we do a conditional check on from_sb? */ 83 static int from_sb_is_expansion = 1; 84 85 /* The number of nested sb structures we have included. */ 86 int macro_nest; 87 88 /* We can have more than one source file open at once, though the info for all 89 but the latest one are saved off in a struct input_save. These files remain 90 open, so we are limited by the number of open files allowed by the 91 underlying OS. We may also sequentially read more than one source file in an 92 assembly. */ 93 94 /* We must track the physical file and line number for error messages. We also 95 track a "logical" file and line number corresponding to (C?) compiler 96 source line numbers. Whenever we open a file we must fill in 97 physical_input_file. So if it is NULL we have not opened any files yet. */ 98 99 static const char *physical_input_file; 100 static const char *logical_input_file; 101 102 /* 1-origin line number in a source file. */ 103 /* A line ends in '\n' or eof. */ 104 static unsigned int physical_input_line; 105 static int logical_input_line; 106 107 /* Struct used to save the state of the input handler during include files */ 108 struct input_save { 109 char * buffer_start; 110 char * partial_where; 111 int partial_size; 112 char save_source[AFTER_SIZE]; 113 size_t buffer_length; 114 const char * physical_input_file; 115 const char * logical_input_file; 116 unsigned int physical_input_line; 117 int logical_input_line; 118 size_t sb_index; 119 sb from_sb; 120 int from_sb_is_expansion; /* Should we do a conditional check? */ 121 struct input_save * next_saved_file; /* Chain of input_saves. */ 122 char * input_file_save; /* Saved state of input routines. */ 123 char * saved_position; /* Caller's saved position in buf. */ 124 }; 125 126 static struct input_save *input_scrub_push (char *saved_position); 127 static char *input_scrub_pop (struct input_save *arg); 128 129 /* Saved information about the file that .include'd this one. When we hit EOF, 130 we automatically pop to that file. */ 131 132 static struct input_save *next_saved_file; 133 134 /* Push the state of input reading and scrubbing so that we can #include. 135 The return value is a 'void *' (fudged for old compilers) to a save 136 area, which can be restored by passing it to input_scrub_pop(). */ 137 138 static struct input_save * 139 input_scrub_push (char *saved_position) 140 { 141 struct input_save *saved; 142 143 saved = XNEW (struct input_save); 144 145 saved->saved_position = saved_position; 146 saved->buffer_start = buffer_start; 147 saved->partial_where = partial_where; 148 saved->partial_size = partial_size; 149 saved->buffer_length = buffer_length; 150 saved->physical_input_file = physical_input_file; 151 saved->logical_input_file = logical_input_file; 152 saved->physical_input_line = physical_input_line; 153 saved->logical_input_line = logical_input_line; 154 saved->sb_index = sb_index; 155 saved->from_sb = from_sb; 156 saved->from_sb_is_expansion = from_sb_is_expansion; 157 memcpy (saved->save_source, save_source, sizeof (save_source)); 158 saved->next_saved_file = next_saved_file; 159 saved->input_file_save = input_file_push (); 160 161 input_file_begin (); /* Reinitialize! */ 162 logical_input_line = -1; 163 logical_input_file = NULL; 164 buffer_length = input_file_buffer_size (); 165 sb_index = -1; 166 167 buffer_start = XNEWVEC (char, (BEFORE_SIZE + buffer_length 168 + buffer_length + AFTER_SIZE + 1)); 169 memcpy (buffer_start, BEFORE_STRING, (int) BEFORE_SIZE); 170 171 return saved; 172 } 173 174 static char * 175 input_scrub_pop (struct input_save *saved) 176 { 177 char *saved_position; 178 179 input_scrub_end (); /* Finish off old buffer */ 180 181 input_file_pop (saved->input_file_save); 182 saved_position = saved->saved_position; 183 buffer_start = saved->buffer_start; 184 buffer_length = saved->buffer_length; 185 physical_input_file = saved->physical_input_file; 186 logical_input_file = saved->logical_input_file; 187 physical_input_line = saved->physical_input_line; 188 logical_input_line = saved->logical_input_line; 189 sb_index = saved->sb_index; 190 from_sb = saved->from_sb; 191 from_sb_is_expansion = saved->from_sb_is_expansion; 192 partial_where = saved->partial_where; 193 partial_size = saved->partial_size; 194 next_saved_file = saved->next_saved_file; 195 memcpy (save_source, saved->save_source, sizeof (save_source)); 196 197 free (saved); 198 return saved_position; 199 } 200 201 void 203 input_scrub_begin (void) 204 { 205 know (strlen (BEFORE_STRING) == BEFORE_SIZE); 206 know (strlen (AFTER_STRING) == AFTER_SIZE 207 || (AFTER_STRING[0] == '\0' && AFTER_SIZE == 1)); 208 209 input_file_begin (); 210 211 buffer_length = input_file_buffer_size (); 212 213 buffer_start = XNEWVEC (char, (BEFORE_SIZE + buffer_length 214 + buffer_length + AFTER_SIZE + 1)); 215 memcpy (buffer_start, BEFORE_STRING, (int) BEFORE_SIZE); 216 217 /* Line number things. */ 218 logical_input_line = -1; 219 logical_input_file = NULL; 220 physical_input_file = NULL; /* No file read yet. */ 221 next_saved_file = NULL; /* At EOF, don't pop to any other file */ 222 do_scrub_begin (flag_m68k_mri); 223 } 224 225 void 226 input_scrub_end (void) 227 { 228 if (buffer_start) 229 { 230 free (buffer_start); 231 buffer_start = 0; 232 input_file_end (); 233 } 234 } 235 236 /* Start reading input from a new file. 237 Return start of caller's part of buffer. */ 238 239 char * 240 input_scrub_new_file (const char *filename) 241 { 242 input_file_open (filename, !flag_no_comments); 243 physical_input_file = filename[0] ? filename : _("{standard input}"); 244 physical_input_line = 0; 245 246 partial_size = 0; 247 return (buffer_start + BEFORE_SIZE); 248 } 249 250 /* Include a file from the current file. Save our state, cause it to 251 be restored on EOF, and begin handling a new file. Same result as 252 input_scrub_new_file. */ 253 254 char * 255 input_scrub_include_file (const char *filename, char *position) 256 { 257 next_saved_file = input_scrub_push (position); 258 return input_scrub_new_file (filename); 259 } 260 261 /* Start getting input from an sb structure. This is used when 262 expanding a macro. */ 263 264 void 265 input_scrub_include_sb (sb *from, char *position, int is_expansion) 266 { 267 int newline; 268 269 if (macro_nest > max_macro_nest) 270 as_fatal (_("macros nested too deeply")); 271 ++macro_nest; 272 273 #ifdef md_macro_start 274 if (is_expansion) 275 { 276 md_macro_start (); 277 } 278 #endif 279 280 next_saved_file = input_scrub_push (position); 281 282 /* Allocate sufficient space: from->len + optional newline. */ 283 newline = from->len >= 1 && from->ptr[0] != '\n'; 284 sb_build (&from_sb, from->len + newline); 285 from_sb_is_expansion = is_expansion; 286 if (newline) 287 { 288 /* Add the sentinel required by read.c. */ 289 sb_add_char (&from_sb, '\n'); 290 } 291 sb_scrub_and_add_sb (&from_sb, from); 292 293 /* Make sure the parser looks at defined contents when it scans for 294 e.g. end-of-line at the end of a macro. */ 295 sb_terminate (&from_sb); 296 297 sb_index = 1; 298 299 /* These variables are reset by input_scrub_push. Restore them 300 since we are, after all, still at the same point in the file. */ 301 logical_input_line = next_saved_file->logical_input_line; 302 logical_input_file = next_saved_file->logical_input_file; 303 } 304 305 void 306 input_scrub_close (void) 307 { 308 input_file_close (); 309 physical_input_line = 0; 310 logical_input_line = -1; 311 } 312 313 char * 314 input_scrub_next_buffer (char **bufp) 315 { 316 char *limit; /*->just after last char of buffer. */ 317 318 if (sb_index != (size_t) -1) 319 { 320 if (sb_index >= from_sb.len) 321 { 322 sb_kill (&from_sb); 323 if (from_sb_is_expansion) 324 { 325 cond_finish_check (macro_nest); 326 #ifdef md_macro_end 327 /* Allow the target to clean up per-macro expansion 328 data. */ 329 md_macro_end (); 330 #endif 331 } 332 --macro_nest; 333 partial_where = NULL; 334 partial_size = 0; 335 if (next_saved_file != NULL) 336 *bufp = input_scrub_pop (next_saved_file); 337 return partial_where; 338 } 339 340 partial_where = from_sb.ptr + from_sb.len; 341 partial_size = 0; 342 *bufp = from_sb.ptr + sb_index; 343 sb_index = from_sb.len; 344 return partial_where; 345 } 346 347 if (partial_size) 348 { 349 memmove (buffer_start + BEFORE_SIZE, partial_where, 350 (unsigned int) partial_size); 351 memcpy (buffer_start + BEFORE_SIZE, save_source, AFTER_SIZE); 352 } 353 354 while (1) 355 { 356 char *p; 357 358 *bufp = buffer_start + BEFORE_SIZE; 359 limit = input_file_give_next_buffer (buffer_start 360 + BEFORE_SIZE 361 + partial_size); 362 if (!limit) 363 { 364 if (partial_size == 0) 365 break; 366 367 as_warn (_("end of file not at end of a line; newline inserted")); 368 p = buffer_start + BEFORE_SIZE + partial_size; 369 *p++ = '\n'; 370 limit = p; 371 } 372 else 373 { 374 /* Terminate the buffer to avoid confusing TC_EOL_IN_INSN. */ 375 *limit = '\0'; 376 377 /* Find last newline. */ 378 for (p = limit - 1; *p != '\n' || TC_EOL_IN_INSN (p); --p) 379 ; 380 ++p; 381 } 382 383 if (p != buffer_start + BEFORE_SIZE) 384 { 385 partial_where = p; 386 partial_size = limit - p; 387 memcpy (save_source, partial_where, (int) AFTER_SIZE); 388 memcpy (partial_where, AFTER_STRING, (int) AFTER_SIZE); 389 return partial_where; 390 } 391 392 partial_size = limit - (buffer_start + BEFORE_SIZE); 393 buffer_length += input_file_buffer_size (); 394 buffer_start = XRESIZEVEC (char, buffer_start, 395 (BEFORE_SIZE 396 + 2 * buffer_length 397 + AFTER_SIZE + 1)); 398 } 399 400 /* Tell the listing we've finished the file. */ 401 LISTING_EOF (); 402 403 /* If we should pop to another file at EOF, do it. */ 404 partial_where = NULL; 405 if (next_saved_file) 406 *bufp = input_scrub_pop (next_saved_file); 407 408 return partial_where; 409 } 410 411 /* The remaining part of this file deals with line numbers, error 413 messages and so on. Return TRUE if we opened any file. */ 414 415 int 416 seen_at_least_1_file (void) 417 { 418 return (physical_input_file != NULL); 419 } 420 421 void 422 bump_line_counters (void) 423 { 424 if (sb_index == (size_t) -1) 425 { 426 ++physical_input_line; 427 if (logical_input_line >= 0) 428 ++logical_input_line; 429 } 430 } 431 432 /* Tells us what the new logical line number and file are. 434 If the line_number is -1, we don't change the current logical line 435 number. If it is -2, we decrement the logical line number (this is 436 to support the .appfile pseudo-op inserted into the stream by 437 do_scrub_chars). 438 If the fname is NULL, we don't change the current logical file name. 439 Returns nonzero if the filename actually changes. */ 440 441 int 442 new_logical_line_flags (const char *fname, /* DON'T destroy it! We point to it! */ 443 int line_number, 444 int flags) 445 { 446 switch (flags) 447 { 448 case 0: 449 break; 450 case 1: 451 if (line_number != -1) 452 abort (); 453 break; 454 case 1 << 1: 455 case 1 << 2: 456 /* FIXME: we could check that include nesting is correct. */ 457 break; 458 default: 459 abort (); 460 } 461 462 if (line_number >= 0) 463 logical_input_line = line_number; 464 else if (line_number == -1 && fname && !*fname && (flags & (1 << 2))) 465 { 466 logical_input_file = physical_input_file; 467 logical_input_line = physical_input_line; 468 fname = NULL; 469 } 470 471 if (fname 472 && (logical_input_file == NULL 473 || filename_cmp (logical_input_file, fname))) 474 { 475 logical_input_file = fname; 476 return 1; 477 } 478 else 479 return 0; 480 } 481 482 int 483 new_logical_line (const char *fname, int line_number) 484 { 485 return new_logical_line_flags (fname, line_number, 0); 486 } 487 488 489 /* Return the current file name and line number. */ 491 492 const char * 493 as_where (unsigned int *linep) 494 { 495 if (logical_input_file != NULL 496 && (linep == NULL || logical_input_line >= 0)) 497 { 498 if (linep != NULL) 499 *linep = logical_input_line; 500 return logical_input_file; 501 } 502 else if (physical_input_file != NULL) 503 { 504 if (linep != NULL) 505 *linep = physical_input_line; 506 return physical_input_file; 507 } 508 else 509 { 510 if (linep != NULL) 511 *linep = 0; 512 return NULL; 513 } 514 } 515