1 /* 2 * Copyright (C) 1984-2012 Mark Nudelman 3 * Modified for use with illumos by Garrett D'Amore. 4 * Copyright 2014 Garrett D'Amore <garrett@damore.org> 5 * 6 * You may distribute under the terms of either the GNU General Public 7 * License or the Less License, as specified in the README file. 8 * 9 * For more information, see the README file. 10 */ 11 12 /* 13 * Low level character input from the input file. 14 * We use these special purpose routines which optimize moving 15 * both forward and backward from the current read pointer. 16 */ 17 18 #include <sys/stat.h> 19 20 #include "less.h" 21 22 extern dev_t curr_dev; 23 extern ino_t curr_ino; 24 extern int less_is_more; 25 26 typedef off_t BLOCKNUM; 27 28 int ignore_eoi; 29 30 /* 31 * Pool of buffers holding the most recently used blocks of the input file. 32 * The buffer pool is kept as a doubly-linked circular list, 33 * in order from most- to least-recently used. 34 * The circular list is anchored by the file state "thisfile". 35 */ 36 struct bufnode { 37 struct bufnode *next, *prev; 38 struct bufnode *hnext, *hprev; 39 }; 40 41 #define LBUFSIZE 8192 42 struct buf { 43 struct bufnode node; 44 BLOCKNUM block; 45 unsigned int datasize; 46 unsigned char data[LBUFSIZE]; 47 }; 48 #define bufnode_buf(bn) ((struct buf *)bn) 49 50 /* 51 * The file state is maintained in a filestate structure. 52 * A pointer to the filestate is kept in the ifile structure. 53 */ 54 #define BUFHASH_SIZE 64 55 struct filestate { 56 struct bufnode buflist; 57 struct bufnode hashtbl[BUFHASH_SIZE]; 58 int file; 59 int flags; 60 off_t fpos; 61 int nbufs; 62 BLOCKNUM block; 63 unsigned int offset; 64 off_t fsize; 65 }; 66 67 #define ch_bufhead thisfile->buflist.next 68 #define ch_buftail thisfile->buflist.prev 69 #define ch_nbufs thisfile->nbufs 70 #define ch_block thisfile->block 71 #define ch_offset thisfile->offset 72 #define ch_fpos thisfile->fpos 73 #define ch_fsize thisfile->fsize 74 #define ch_flags thisfile->flags 75 #define ch_file thisfile->file 76 77 #define END_OF_CHAIN (&thisfile->buflist) 78 #define END_OF_HCHAIN(h) (&thisfile->hashtbl[h]) 79 #define BUFHASH(blk) ((blk) & (BUFHASH_SIZE-1)) 80 81 /* 82 * Macros to manipulate the list of buffers in thisfile->buflist. 83 */ 84 #define FOR_BUFS(bn) \ 85 for ((bn) = ch_bufhead; (bn) != END_OF_CHAIN; (bn) = (bn)->next) 86 87 #define BUF_RM(bn) \ 88 (bn)->next->prev = (bn)->prev; \ 89 (bn)->prev->next = (bn)->next; 90 91 #define BUF_INS_HEAD(bn) \ 92 (bn)->next = ch_bufhead; \ 93 (bn)->prev = END_OF_CHAIN; \ 94 ch_bufhead->prev = (bn); \ 95 ch_bufhead = (bn); 96 97 #define BUF_INS_TAIL(bn) \ 98 (bn)->next = END_OF_CHAIN; \ 99 (bn)->prev = ch_buftail; \ 100 ch_buftail->next = (bn); \ 101 ch_buftail = (bn); 102 103 /* 104 * Macros to manipulate the list of buffers in thisfile->hashtbl[n]. 105 */ 106 #define FOR_BUFS_IN_CHAIN(h, bn) \ 107 for ((bn) = thisfile->hashtbl[h].hnext; \ 108 (bn) != END_OF_HCHAIN(h); (bn) = (bn)->hnext) 109 110 #define BUF_HASH_RM(bn) \ 111 (bn)->hnext->hprev = (bn)->hprev; \ 112 (bn)->hprev->hnext = (bn)->hnext; 113 114 #define BUF_HASH_INS(bn, h) \ 115 (bn)->hnext = thisfile->hashtbl[h].hnext; \ 116 (bn)->hprev = END_OF_HCHAIN(h); \ 117 thisfile->hashtbl[h].hnext->hprev = (bn); \ 118 thisfile->hashtbl[h].hnext = (bn); 119 120 static struct filestate *thisfile; 121 static int ch_ungotchar = -1; 122 static int maxbufs = -1; 123 124 extern int autobuf; 125 extern volatile sig_atomic_t sigs; 126 extern int secure; 127 extern int screen_trashed; 128 extern int follow_mode; 129 extern IFILE curr_ifile; 130 extern int logfile; 131 extern char *namelogfile; 132 133 static int ch_addbuf(void); 134 135 136 /* 137 * Get the character pointed to by the read pointer. 138 */ 139 int 140 ch_get(void) 141 { 142 struct buf *bp; 143 struct bufnode *bn; 144 int n; 145 int slept; 146 int h; 147 off_t pos; 148 off_t len; 149 150 if (thisfile == NULL) 151 return (EOI); 152 153 /* 154 * Quick check for the common case where 155 * the desired char is in the head buffer. 156 */ 157 if (ch_bufhead != END_OF_CHAIN) { 158 bp = bufnode_buf(ch_bufhead); 159 if (ch_block == bp->block && ch_offset < bp->datasize) 160 return (bp->data[ch_offset]); 161 } 162 163 slept = FALSE; 164 165 /* 166 * Look for a buffer holding the desired block. 167 */ 168 h = BUFHASH(ch_block); 169 FOR_BUFS_IN_CHAIN(h, bn) { 170 bp = bufnode_buf(bn); 171 if (bp->block == ch_block) { 172 if (ch_offset >= bp->datasize) 173 /* 174 * Need more data in this buffer. 175 */ 176 break; 177 goto found; 178 } 179 } 180 if (bn == END_OF_HCHAIN(h)) { 181 /* 182 * Block is not in a buffer. 183 * Take the least recently used buffer 184 * and read the desired block into it. 185 * If the LRU buffer has data in it, 186 * then maybe allocate a new buffer. 187 */ 188 if (ch_buftail == END_OF_CHAIN || 189 bufnode_buf(ch_buftail)->block != -1) { 190 /* 191 * There is no empty buffer to use. 192 * Allocate a new buffer if: 193 * 1. We can't seek on this file and -b is not in 194 * effect; or 195 * 2. We haven't allocated the max buffers for this 196 * file yet. 197 */ 198 if ((autobuf && !(ch_flags & CH_CANSEEK)) || 199 (maxbufs < 0 || ch_nbufs < maxbufs)) 200 if (ch_addbuf()) 201 /* 202 * Allocation failed: turn off autobuf. 203 */ 204 autobuf = OPT_OFF; 205 } 206 bn = ch_buftail; 207 bp = bufnode_buf(bn); 208 BUF_HASH_RM(bn); /* Remove from old hash chain. */ 209 bp->block = ch_block; 210 bp->datasize = 0; 211 BUF_HASH_INS(bn, h); /* Insert into new hash chain. */ 212 } 213 214 read_more: 215 pos = (ch_block * LBUFSIZE) + bp->datasize; 216 if ((len = ch_length()) != -1 && pos >= len) 217 /* 218 * At end of file. 219 */ 220 return (EOI); 221 222 if (pos != ch_fpos) { 223 /* 224 * Not at the correct position: must seek. 225 * If input is a pipe, we're in trouble (can't seek on a pipe). 226 * Some data has been lost: just return "?". 227 */ 228 if (!(ch_flags & CH_CANSEEK)) 229 return ('?'); 230 if (lseek(ch_file, (off_t)pos, SEEK_SET) == (off_t)-1) { 231 error("seek error", NULL); 232 clear_eol(); 233 return (EOI); 234 } 235 ch_fpos = pos; 236 } 237 238 /* 239 * Read the block. 240 * If we read less than a full block, that's ok. 241 * We use partial block and pick up the rest next time. 242 */ 243 if (ch_ungotchar != -1) { 244 bp->data[bp->datasize] = (unsigned char)ch_ungotchar; 245 n = 1; 246 ch_ungotchar = -1; 247 } else { 248 n = iread(ch_file, &bp->data[bp->datasize], 249 (unsigned int)(LBUFSIZE - bp->datasize)); 250 } 251 252 if (n == READ_INTR) 253 return (EOI); 254 if (n < 0) { 255 error("read error", NULL); 256 clear_eol(); 257 n = 0; 258 } 259 260 /* 261 * If we have a log file, write the new data to it. 262 */ 263 if (!secure && logfile >= 0 && n > 0) 264 (void) write(logfile, (char *)&bp->data[bp->datasize], n); 265 266 ch_fpos += n; 267 bp->datasize += n; 268 269 /* 270 * If we have read to end of file, set ch_fsize to indicate 271 * the position of the end of file. 272 */ 273 if (n == 0) { 274 ch_fsize = pos; 275 if (ignore_eoi) { 276 /* 277 * We are ignoring EOF. 278 * Wait a while, then try again. 279 */ 280 if (!slept) { 281 PARG parg; 282 parg.p_string = wait_message(); 283 ierror("%s", &parg); 284 } 285 sleep(1); 286 slept = TRUE; 287 288 if (follow_mode == FOLLOW_NAME) { 289 /* 290 * See whether the file's i-number has changed. 291 * If so, force the file to be closed and 292 * reopened. 293 */ 294 struct stat st; 295 int r = stat(get_filename(curr_ifile), &st); 296 if (r == 0 && (st.st_ino != curr_ino || 297 st.st_dev != curr_dev)) { 298 /* 299 * screen_trashed=2 causes 300 * make_display to reopen the file. 301 */ 302 screen_trashed = 2; 303 return (EOI); 304 } 305 } 306 } 307 if (sigs) 308 return (EOI); 309 } 310 311 found: 312 if (ch_bufhead != bn) { 313 /* 314 * Move the buffer to the head of the buffer chain. 315 * This orders the buffer chain, most- to least-recently used. 316 */ 317 BUF_RM(bn); 318 BUF_INS_HEAD(bn); 319 320 /* 321 * Move to head of hash chain too. 322 */ 323 BUF_HASH_RM(bn); 324 BUF_HASH_INS(bn, h); 325 } 326 327 if (ch_offset >= bp->datasize) 328 /* 329 * After all that, we still don't have enough data. 330 * Go back and try again. 331 */ 332 goto read_more; 333 334 return (bp->data[ch_offset]); 335 } 336 337 /* 338 * ch_ungetchar is a rather kludgy and limited way to push 339 * a single char onto an input file descriptor. 340 */ 341 void 342 ch_ungetchar(int c) 343 { 344 if (c != -1 && ch_ungotchar != -1) 345 error("ch_ungetchar overrun", NULL); 346 ch_ungotchar = c; 347 } 348 349 /* 350 * Close the logfile. 351 * If we haven't read all of standard input into it, do that now. 352 */ 353 void 354 end_logfile(void) 355 { 356 static int tried = FALSE; 357 358 if (logfile < 0) 359 return; 360 if (!tried && ch_fsize == -1) { 361 tried = TRUE; 362 ierror("Finishing logfile", NULL); 363 while (ch_forw_get() != EOI) 364 if (ABORT_SIGS()) 365 break; 366 } 367 close(logfile); 368 logfile = -1; 369 free(namelogfile); 370 namelogfile = NULL; 371 } 372 373 /* 374 * Start a log file AFTER less has already been running. 375 * Invoked from the - command; see toggle_option(). 376 * Write all the existing buffered data to the log file. 377 */ 378 void 379 sync_logfile(void) 380 { 381 struct buf *bp; 382 struct bufnode *bn; 383 int warned = FALSE; 384 BLOCKNUM block; 385 BLOCKNUM nblocks; 386 387 nblocks = (ch_fpos + LBUFSIZE - 1) / LBUFSIZE; 388 for (block = 0; block < nblocks; block++) { 389 int wrote = FALSE; 390 FOR_BUFS(bn) { 391 bp = bufnode_buf(bn); 392 if (bp->block == block) { 393 (void) write(logfile, (char *)bp->data, 394 bp->datasize); 395 wrote = TRUE; 396 break; 397 } 398 } 399 if (!wrote && !warned) { 400 error("Warning: log file is incomplete", NULL); 401 warned = TRUE; 402 } 403 } 404 } 405 406 /* 407 * Determine if a specific block is currently in one of the buffers. 408 */ 409 static int 410 buffered(BLOCKNUM block) 411 { 412 struct buf *bp; 413 struct bufnode *bn; 414 int h; 415 416 h = BUFHASH(block); 417 FOR_BUFS_IN_CHAIN(h, bn) { 418 bp = bufnode_buf(bn); 419 if (bp->block == block) 420 return (TRUE); 421 } 422 return (FALSE); 423 } 424 425 /* 426 * Seek to a specified position in the file. 427 * Return 0 if successful, non-zero if can't seek there. 428 */ 429 int 430 ch_seek(off_t pos) 431 { 432 BLOCKNUM new_block; 433 off_t len; 434 435 if (thisfile == NULL) 436 return (0); 437 438 len = ch_length(); 439 if (pos < ch_zero() || (len != -1 && pos > len)) 440 return (1); 441 442 new_block = pos / LBUFSIZE; 443 if (!(ch_flags & CH_CANSEEK) && pos != ch_fpos && 444 !buffered(new_block)) { 445 if (ch_fpos > pos) 446 return (1); 447 while (ch_fpos < pos) { 448 if (ch_forw_get() == EOI) 449 return (1); 450 if (ABORT_SIGS()) 451 return (1); 452 } 453 return (0); 454 } 455 /* 456 * Set read pointer. 457 */ 458 ch_block = new_block; 459 ch_offset = pos % LBUFSIZE; 460 return (0); 461 } 462 463 /* 464 * Seek to the end of the file. 465 */ 466 int 467 ch_end_seek(void) 468 { 469 off_t len; 470 471 if (thisfile == NULL) 472 return (0); 473 474 if (ch_flags & CH_CANSEEK) 475 ch_fsize = filesize(ch_file); 476 477 len = ch_length(); 478 if (len != -1) 479 return (ch_seek(len)); 480 481 /* 482 * Do it the slow way: read till end of data. 483 */ 484 while (ch_forw_get() != EOI) 485 if (ABORT_SIGS()) 486 return (1); 487 return (0); 488 } 489 490 /* 491 * Seek to the beginning of the file, or as close to it as we can get. 492 * We may not be able to seek there if input is a pipe and the 493 * beginning of the pipe is no longer buffered. 494 */ 495 int 496 ch_beg_seek(void) 497 { 498 struct bufnode *bn; 499 struct bufnode *firstbn; 500 501 /* 502 * Try a plain ch_seek first. 503 */ 504 if (ch_seek(ch_zero()) == 0) 505 return (0); 506 507 /* 508 * Can't get to position 0. 509 * Look thru the buffers for the one closest to position 0. 510 */ 511 firstbn = ch_bufhead; 512 if (firstbn == END_OF_CHAIN) 513 return (1); 514 FOR_BUFS(bn) { 515 if (bufnode_buf(bn)->block < bufnode_buf(firstbn)->block) 516 firstbn = bn; 517 } 518 ch_block = bufnode_buf(firstbn)->block; 519 ch_offset = 0; 520 return (0); 521 } 522 523 /* 524 * Return the length of the file, if known. 525 */ 526 off_t 527 ch_length(void) 528 { 529 if (thisfile == NULL) 530 return (-1); 531 if (ignore_eoi) 532 return (-1); 533 if (ch_flags & CH_NODATA) 534 return (0); 535 return (ch_fsize); 536 } 537 538 /* 539 * Return the current position in the file. 540 */ 541 off_t 542 ch_tell(void) 543 { 544 if (thisfile == NULL) 545 return (-1); 546 return ((ch_block * LBUFSIZE) + ch_offset); 547 } 548 549 /* 550 * Get the current char and post-increment the read pointer. 551 */ 552 int 553 ch_forw_get(void) 554 { 555 int c; 556 557 if (thisfile == NULL) 558 return (EOI); 559 c = ch_get(); 560 if (c == EOI) 561 return (EOI); 562 if (ch_offset < LBUFSIZE-1) { 563 ch_offset++; 564 } else { 565 ch_block ++; 566 ch_offset = 0; 567 } 568 return (c); 569 } 570 571 /* 572 * Pre-decrement the read pointer and get the new current char. 573 */ 574 int 575 ch_back_get(void) 576 { 577 if (thisfile == NULL) 578 return (EOI); 579 if (ch_offset > 0) { 580 ch_offset --; 581 } else { 582 if (ch_block <= 0) 583 return (EOI); 584 if (!(ch_flags & CH_CANSEEK) && !buffered(ch_block-1)) 585 return (EOI); 586 ch_block--; 587 ch_offset = LBUFSIZE-1; 588 } 589 return (ch_get()); 590 } 591 592 /* 593 * Set max amount of buffer space. 594 * bufspace is in units of 1024 bytes. -1 mean no limit. 595 */ 596 void 597 ch_setbufspace(int bufspace) 598 { 599 if (bufspace < 0) { 600 maxbufs = -1; 601 } else { 602 maxbufs = ((bufspace * 1024) + LBUFSIZE-1) / LBUFSIZE; 603 if (maxbufs < 1) 604 maxbufs = 1; 605 } 606 } 607 608 /* 609 * Flush (discard) any saved file state, including buffer contents. 610 */ 611 void 612 ch_flush(void) 613 { 614 struct bufnode *bn; 615 616 if (thisfile == NULL) 617 return; 618 619 if (!(ch_flags & CH_CANSEEK)) { 620 /* 621 * If input is a pipe, we don't flush buffer contents, 622 * since the contents can't be recovered. 623 */ 624 ch_fsize = -1; 625 return; 626 } 627 628 /* 629 * Initialize all the buffers. 630 */ 631 FOR_BUFS(bn) { 632 bufnode_buf(bn)->block = -1; 633 } 634 635 /* 636 * Figure out the size of the file, if we can. 637 */ 638 ch_fsize = filesize(ch_file); 639 640 /* 641 * Seek to a known position: the beginning of the file. 642 */ 643 ch_fpos = 0; 644 ch_block = 0; /* ch_fpos / LBUFSIZE; */ 645 ch_offset = 0; /* ch_fpos % LBUFSIZE; */ 646 647 #if 1 648 /* 649 * This is a kludge to workaround a Linux kernel bug: files in 650 * /proc have a size of 0 according to fstat() but have readable 651 * data. They are sometimes, but not always, seekable. 652 * Force them to be non-seekable here. 653 */ 654 if (ch_fsize == 0) { 655 ch_fsize = -1; 656 ch_flags &= ~CH_CANSEEK; 657 } 658 #endif 659 660 if (lseek(ch_file, (off_t)0, SEEK_SET) == (off_t)-1) { 661 /* 662 * Warning only; even if the seek fails for some reason, 663 * there's a good chance we're at the beginning anyway. 664 * {{ I think this is bogus reasoning. }} 665 */ 666 error("seek error to 0", NULL); 667 } 668 } 669 670 /* 671 * Allocate a new buffer. 672 * The buffer is added to the tail of the buffer chain. 673 */ 674 static int 675 ch_addbuf(void) 676 { 677 struct buf *bp; 678 struct bufnode *bn; 679 680 /* 681 * Allocate and initialize a new buffer and link it 682 * onto the tail of the buffer list. 683 */ 684 bp = calloc(1, sizeof (struct buf)); 685 if (bp == NULL) 686 return (1); 687 ch_nbufs++; 688 bp->block = -1; 689 bn = &bp->node; 690 691 BUF_INS_TAIL(bn); 692 BUF_HASH_INS(bn, 0); 693 return (0); 694 } 695 696 /* 697 * 698 */ 699 static void 700 init_hashtbl(void) 701 { 702 int h; 703 704 for (h = 0; h < BUFHASH_SIZE; h++) { 705 thisfile->hashtbl[h].hnext = END_OF_HCHAIN(h); 706 thisfile->hashtbl[h].hprev = END_OF_HCHAIN(h); 707 } 708 } 709 710 /* 711 * Delete all buffers for this file. 712 */ 713 static void 714 ch_delbufs(void) 715 { 716 struct bufnode *bn; 717 718 while (ch_bufhead != END_OF_CHAIN) { 719 bn = ch_bufhead; 720 BUF_RM(bn); 721 free(bufnode_buf(bn)); 722 } 723 ch_nbufs = 0; 724 init_hashtbl(); 725 } 726 727 /* 728 * Is it possible to seek on a file descriptor? 729 */ 730 int 731 seekable(int f) 732 { 733 return (lseek(f, (off_t)1, SEEK_SET) != (off_t)-1); 734 } 735 736 /* 737 * Force EOF to be at the current read position. 738 * This is used after an ignore_eof read, during which the EOF may change. 739 */ 740 void 741 ch_set_eof(void) 742 { 743 ch_fsize = ch_fpos; 744 } 745 746 747 /* 748 * Initialize file state for a new file. 749 */ 750 void 751 ch_init(int f, int flags) 752 { 753 /* 754 * See if we already have a filestate for this file. 755 */ 756 thisfile = get_filestate(curr_ifile); 757 if (thisfile == NULL) { 758 /* 759 * Allocate and initialize a new filestate. 760 */ 761 thisfile = calloc(1, sizeof (struct filestate)); 762 thisfile->buflist.next = thisfile->buflist.prev = END_OF_CHAIN; 763 thisfile->nbufs = 0; 764 thisfile->flags = 0; 765 thisfile->fpos = 0; 766 thisfile->block = 0; 767 thisfile->offset = 0; 768 thisfile->file = -1; 769 thisfile->fsize = -1; 770 ch_flags = flags; 771 init_hashtbl(); 772 /* 773 * Try to seek; set CH_CANSEEK if it works. 774 */ 775 if ((flags & CH_CANSEEK) && !seekable(f)) 776 ch_flags &= ~CH_CANSEEK; 777 set_filestate(curr_ifile, (void *) thisfile); 778 } 779 if (thisfile->file == -1) 780 thisfile->file = f; 781 ch_flush(); 782 } 783 784 /* 785 * Close a filestate. 786 */ 787 void 788 ch_close(void) 789 { 790 int keepstate = FALSE; 791 792 if (thisfile == NULL) 793 return; 794 795 if (ch_flags & (CH_CANSEEK|CH_POPENED|CH_HELPFILE)) { 796 /* 797 * We can seek or re-open, so we don't need to keep buffers. 798 */ 799 ch_delbufs(); 800 } else { 801 keepstate = TRUE; 802 } 803 if (!(ch_flags & CH_KEEPOPEN)) { 804 /* 805 * We don't need to keep the file descriptor open 806 * (because we can re-open it.) 807 * But don't really close it if it was opened via popen(), 808 * because pclose() wants to close it. 809 */ 810 if (!(ch_flags & CH_POPENED)) 811 close(ch_file); 812 ch_file = -1; 813 } else { 814 keepstate = TRUE; 815 } 816 if (!keepstate) { 817 /* 818 * We don't even need to keep the filestate structure. 819 */ 820 free(thisfile); 821 thisfile = NULL; 822 set_filestate(curr_ifile, NULL); 823 } 824 } 825 826 /* 827 * Return ch_flags for the current file. 828 */ 829 int 830 ch_getflags(void) 831 { 832 if (thisfile == NULL) 833 return (0); 834 return (ch_flags); 835 } 836