xref: /illumos-gate/usr/src/cmd/sgs/libld/common/util.c (revision 581cede61ac9c14d8d4ea452562a567189eead78)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 
22 /*
23  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 /*
28  *	Copyright (c) 1988 AT&T
29  *	  All Rights Reserved
30  */
31 
32 /*
33  * Utility functions
34  */
35 #include <unistd.h>
36 #include <stdio.h>
37 #include <stdarg.h>
38 #include <string.h>
39 #include <fcntl.h>
40 #include <sys/types.h>
41 #include <sys/mman.h>
42 #include <errno.h>
43 #include <sgs.h>
44 #include <libintl.h>
45 #include <debug.h>
46 #include "msg.h"
47 #include "_libld.h"
48 
49 /*
50  * libld_malloc() and dz_map() are used for both performance and for ease of
51  * programming:
52  *
53  * Performance:
54  *	The link-edit is a short lived process which doesn't really free much
55  *	of the dynamic memory that it requests.  Because of this, it is more
56  *	important to optimize for quick memory allocations than the
57  *	re-usability of the memory.
58  *
59  *	By also mmaping blocks of pages in from /dev/zero we don't need to
60  *	waste the overhead of zeroing out these pages for calloc() requests.
61  *
62  * Memory Management:
63  *	By doing all libld memory management through the ld_malloc routine
64  *	it's much easier to free up all memory at the end by simply unmaping
65  *	all of the blocks that were mapped in through dz_map().  This is much
66  *	simpler then trying to track all of the libld structures that were
67  *	dynamically allocate and are actually pointers into the ELF files.
68  *
69  *	It's important that we can free up all of our dynamic memory because
70  *	libld is used by ld.so.1 when it performs dlopen()'s of relocatable
71  *	objects.
72  *
73  * Format:
74  *	The memory blocks for each allocation store the size of the allocation
75  *	in the first 8 bytes of the block.  The pointer that is returned by
76  *	libld_malloc() is actually the address of (block + 8):
77  *
78  *		(addr - 8)	block_size
79  *		(addr)		<allocated block>
80  *
81  *	The size is retained in order to implement realloc(), and to perform
82  *	the required memcpy().  8 bytes are uses, as the memory area returned
83  *	by libld_malloc() must be 8 byte-aligned.  Even in a 32-bit environment,
84  *	u_longlog_t pointers are employed.
85  *
86  * Map anonymous memory via MAP_ANON (added in Solaris 8).
87  */
88 static void *
89 dz_map(size_t size)
90 {
91 	void	*addr;
92 
93 	if ((addr = mmap(0, size, (PROT_READ | PROT_WRITE | PROT_EXEC),
94 	    (MAP_PRIVATE | MAP_ANON), -1, 0)) == MAP_FAILED) {
95 		int	err = errno;
96 		eprintf(0, ERR_FATAL, MSG_INTL(MSG_SYS_MMAPANON),
97 		    strerror(err));
98 		return (MAP_FAILED);
99 	}
100 	return (addr);
101 }
102 
103 void *
104 libld_malloc(size_t size)
105 {
106 	Ld_heap		*chp = ld_heap;
107 	void		*vptr;
108 	size_t		asize = size + HEAPALIGN;
109 
110 	/*
111 	 * If this is the first allocation, or the allocation request is greater
112 	 * than the current free space available, allocate a new heap.
113 	 */
114 	if ((chp == NULL) ||
115 	    (((size_t)chp->lh_end - (size_t)chp->lh_free) <= asize)) {
116 		Ld_heap	*nhp;
117 		size_t	hsize = (size_t)S_ROUND(sizeof (Ld_heap), HEAPALIGN);
118 		size_t	tsize = (size_t)S_ROUND((asize + hsize), HEAPALIGN);
119 
120 		/*
121 		 * Allocate a block that is at minimum 'HEAPBLOCK' size
122 		 */
123 		if (tsize < HEAPBLOCK)
124 			tsize = HEAPBLOCK;
125 
126 		if ((nhp = dz_map(tsize)) == MAP_FAILED)
127 			return (NULL);
128 
129 		nhp->lh_next = chp;
130 		nhp->lh_free = (void *)((size_t)nhp + hsize);
131 		nhp->lh_end = (void *)((size_t)nhp + tsize);
132 
133 		ld_heap = chp = nhp;
134 	}
135 	vptr = chp->lh_free;
136 
137 	/*
138 	 * Assign size to head of allocated block (used by realloc), and
139 	 * memory arena as then next 8-byte aligned offset.
140 	 */
141 	*((size_t *)vptr) = size;
142 	vptr = (void *)((size_t)vptr + HEAPALIGN);
143 
144 	/*
145 	 * Increment free to point to next available block
146 	 */
147 	chp->lh_free = (void *)S_ROUND((size_t)chp->lh_free + asize,
148 	    HEAPALIGN);
149 
150 	return (vptr);
151 }
152 
153 void *
154 libld_realloc(void *ptr, size_t size)
155 {
156 	size_t	psize;
157 	void	*vptr;
158 
159 	if (ptr == NULL)
160 		return (libld_malloc(size));
161 
162 	/*
163 	 * Size of the allocated blocks is stored *just* before the blocks
164 	 * address.
165 	 */
166 	psize = *((size_t *)((size_t)ptr - HEAPALIGN));
167 
168 	/*
169 	 * If the block actually fits then just return.
170 	 */
171 	if (size <= psize)
172 		return (ptr);
173 
174 	if ((vptr = libld_malloc(size)) != NULL)
175 		(void) memcpy(vptr, ptr, psize);
176 
177 	return (vptr);
178 }
179 
180 void
181 /* ARGSUSED 0 */
182 libld_free(void *ptr)
183 {
184 }
185 
186 /*
187  * Determine if a shared object definition structure already exists and if
188  * not create one.  These definitions provide for recording information
189  * regarding shared objects that are still to be processed.  Once processed
190  * shared objects are maintained on the ofl_sos list.  The information
191  * recorded in this structure includes:
192  *
193  *  o	DT_USED requirements.  In these cases definitions are added during
194  *	mapfile processing of `-' entries (see map_dash()).
195  *
196  *  o	implicit NEEDED entries.  As shared objects are processed from the
197  *	command line so any of their dependencies are recorded in these
198  *	structures for later processing (see process_dynamic()).
199  *
200  *  o	version requirements.  Any explicit shared objects that have version
201  *	dependencies on other objects have their version requirements recorded.
202  *	In these cases definitions are added during mapfile processing of `-'
203  *	entries (see map_dash()).  Also, shared objects may have versioning
204  *	requirements on their NEEDED entries.  These cases are added during
205  *	their version processing (see vers_need_process()).
206  *
207  *	Note: Both process_dynamic() and vers_need_process() may generate the
208  *	initial version definition structure because you can't rely on what
209  *	section (.dynamic or .SUNW_version) may be processed first from	any
210  *	input file.
211  */
212 Sdf_desc *
213 sdf_find(const char *name, APlist *alp)
214 {
215 	Aliste		idx;
216 	Sdf_desc	*sdf;
217 
218 	for (APLIST_TRAVERSE(alp, idx, sdf))
219 		if (strcmp(name, sdf->sdf_name) == 0)
220 			return (sdf);
221 
222 	return (NULL);
223 }
224 
225 Sdf_desc *
226 sdf_add(const char *name, APlist **alpp)
227 {
228 	Sdf_desc	*sdf;
229 
230 	if ((sdf = libld_calloc(sizeof (Sdf_desc), 1)) == NULL)
231 		return ((Sdf_desc *)S_ERROR);
232 
233 	sdf->sdf_name = name;
234 
235 	if (aplist_append(alpp, sdf, AL_CNT_OFL_LIBS) == NULL)
236 		return ((Sdf_desc *)S_ERROR);
237 
238 	return (sdf);
239 }
240 
241 /*
242  * Add a string, separated by a colon, to an existing string.  Typically used
243  * to maintain filter, rpath and audit names, of which there is normally only
244  * one string supplied anyway.
245  */
246 char *
247 add_string(char *old, char *str)
248 {
249 	char	*new;
250 
251 	if (old) {
252 		char	*_str;
253 		size_t	len;
254 
255 		/*
256 		 * If an original string exists, make sure this new string
257 		 * doesn't get duplicated.
258 		 */
259 		if ((_str = strstr(old, str)) != NULL) {
260 			if (((_str == old) ||
261 			    (*(_str - 1) == *(MSG_ORIG(MSG_STR_COLON)))) &&
262 			    (_str += strlen(str)) &&
263 			    ((*_str == '\0') ||
264 			    (*_str == *(MSG_ORIG(MSG_STR_COLON)))))
265 				return (old);
266 		}
267 
268 		len = strlen(old) + strlen(str) + 2;
269 		if ((new = libld_calloc(1, len)) == NULL)
270 			return ((char *)S_ERROR);
271 		(void) snprintf(new, len, MSG_ORIG(MSG_FMT_COLPATH), old, str);
272 	} else {
273 		if ((new = libld_malloc(strlen(str) + 1)) == NULL)
274 			return ((char *)S_ERROR);
275 		(void) strcpy(new, str);
276 	}
277 
278 	return (new);
279 }
280 
281 /*
282  * Determine whether this string, possibly with an associated option, should be
283  * translated to an option character.  If so, update the optind and optarg
284  * as described for short options in getopt(3c).
285  */
286 static int
287 str2chr(Lm_list *lml, int ndx, int argc, char **argv, char *arg, int c,
288     const char *opt, size_t optsz)
289 {
290 	if (optsz == 0) {
291 		/*
292 		 * Compare a single option (ie. there's no associated option
293 		 * argument).
294 		 */
295 		if (strcmp(arg, opt) == 0) {
296 			DBG_CALL(Dbg_args_str2chr(lml, ndx, opt, c));
297 			optind += 1;
298 			return (c);
299 		}
300 
301 	} else if (strncmp(arg, opt, optsz) == 0) {
302 		/*
303 		 * Otherwise, compare the option name, which may be
304 		 * concatenated with the option argument.
305 		 */
306 		DBG_CALL(Dbg_args_str2chr(lml, ndx, opt, c));
307 
308 		if (arg[optsz] == '\0') {
309 			/*
310 			 * Optarg is the next argument (white space separated).
311 			 * Make sure an optarg is available, and if not return
312 			 * a failure to prevent any fall-through to the generic
313 			 * getopt() processing.
314 			 */
315 			if ((++optind + 1) > argc) {
316 				return ('?');
317 			}
318 			optarg = argv[optind];
319 			optind++;
320 		} else {
321 			/*
322 			 * Optarg concatenated to option (no white space).
323 			 * GNU option/option argument pairs can be represented
324 			 * with a "=" separator.  If this is the case, remove
325 			 * the separator.
326 			 */
327 			optarg = &arg[optsz];
328 			optind++;
329 			if (*optarg == '=') {
330 				if (*(++optarg) == '\0')
331 					return ('?');
332 			}
333 		}
334 		return (c);
335 	}
336 	return (0);
337 }
338 
339 /*
340  * Parse an individual option.  The intent of this function is to determine if
341  * any known, non-Solaris options have been passed to ld(1).  This condition
342  * can occur as a result of build configuration tools, because of users
343  * familiarity with other systems, or simply the users preferences.  If a known
344  * non-Solaris option can be determined, translate that option into the Solaris
345  * counterpart.
346  *
347  * This function will probably never be a complete solution, as new, non-Solaris
348  * options are discovered, their translation will have to be added.  Other
349  * non-Solaris options are incompatible with the Solaris link-editor, and will
350  * never be recognized.  We support what we can.
351  */
352 int
353 ld_getopt(Lm_list *lml, int ndx, int argc, char **argv)
354 {
355 	int	c;
356 
357 	if ((optind < argc) && argv[optind] && (argv[optind][0] == '-')) {
358 		char	*arg = &argv[optind][1];
359 
360 		switch (*arg) {
361 		case 'r':
362 			/* Translate -rpath <optarg> to -R <optarg> */
363 			if ((c = str2chr(lml, ndx, argc, argv, arg, 'R',
364 			    MSG_ORIG(MSG_ARG_T_RPATH),
365 			    MSG_ARG_T_RPATH_SIZE)) != 0) {
366 				return (c);
367 			}
368 			break;
369 		case 's':
370 			/* Translate -shared to -G */
371 			if ((c = str2chr(lml, ndx, argc, argv, arg, 'G',
372 			    MSG_ORIG(MSG_ARG_T_SHARED), 0)) != 0) {
373 				return (c);
374 
375 			/* Translate -soname <optarg> to -h <optarg> */
376 			} else if ((c = str2chr(lml, ndx, argc, argv, arg, 'h',
377 			    MSG_ORIG(MSG_ARG_T_SONAME),
378 			    MSG_ARG_T_SONAME_SIZE)) != 0) {
379 				return (c);
380 			}
381 			break;
382 		case '(':
383 			/*
384 			 * Translate -( to -z rescan-start
385 			 */
386 			if ((c = str2chr(lml, ndx, argc, argv,
387 			    arg, 'z', MSG_ORIG(MSG_ARG_T_OPAR), 0)) != 0) {
388 				optarg = (char *)MSG_ORIG(MSG_ARG_RESCAN_START);
389 				return (c);
390 			}
391 			break;
392 		case ')':
393 			/*
394 			 * Translate -) to -z rescan-end
395 			 */
396 			if ((c = str2chr(lml, ndx, argc, argv,
397 			    arg, 'z', MSG_ORIG(MSG_ARG_T_CPAR), 0)) != 0) {
398 				optarg = (char *)MSG_ORIG(MSG_ARG_RESCAN_END);
399 				return (c);
400 			}
401 			break;
402 		case '-':
403 			switch (*(arg + 1)) {
404 			case 'a':
405 				/*
406 				 * Translate --allow-multiple-definition to
407 				 * -zmuldefs
408 				 */
409 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'z',
410 				    MSG_ORIG(MSG_ARG_T_MULDEFS), 0)) != 0) {
411 					optarg =
412 					    (char *)MSG_ORIG(MSG_ARG_MULDEFS);
413 					return (c);
414 
415 				/*
416 				 * Translate --auxiliary <optarg> to
417 				 * -f <optarg>
418 				 */
419 				} else if ((c = str2chr(lml, argc, ndx, argv,
420 				    arg, 'f', MSG_ORIG(MSG_ARG_T_AUXFLTR),
421 				    MSG_ARG_T_AUXFLTR_SIZE)) != 0) {
422 					return (c);
423 				}
424 				break;
425 			case 'd':
426 				/*
427 				 * Translate --dynamic-linker <optarg> to
428 				 * -I <optarg>
429 				 */
430 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'I',
431 				    MSG_ORIG(MSG_ARG_T_INTERP),
432 				    MSG_ARG_T_INTERP_SIZE)) != 0) {
433 					return (c);
434 				}
435 				break;
436 			case 'e':
437 				/* Translate --entry <optarg> to -e <optarg> */
438 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'e',
439 				    MSG_ORIG(MSG_ARG_T_ENTRY),
440 				    MSG_ARG_T_ENTRY_SIZE)) != 0) {
441 					return (c);
442 				}
443 				/*
444 				 * Translate --end-group to -z rescan-end
445 				 */
446 				if ((c = str2chr(lml, ndx, argc, argv,
447 				    arg, 'z',
448 				    MSG_ORIG(MSG_ARG_T_ENDGROUP), 0)) != 0) {
449 					optarg = (char *)
450 					    MSG_ORIG(MSG_ARG_RESCAN_END);
451 					return (c);
452 				}
453 				break;
454 			case 'f':
455 				/* Translate --filter <optarg> to -F <optarg> */
456 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'F',
457 				    MSG_ORIG(MSG_ARG_T_STDFLTR),
458 				    MSG_ARG_T_STDFLTR_SIZE)) != 0) {
459 					return (c);
460 				}
461 				break;
462 			case 'h':
463 				/* Translate --help to -zhelp */
464 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'z',
465 				    MSG_ORIG(MSG_ARG_T_HELP), 0)) != 0) {
466 					optarg = (char *)MSG_ORIG(MSG_ARG_HELP);
467 					return (c);
468 				}
469 				break;
470 			case 'l':
471 				/*
472 				 * Translate --library <optarg> to -l <optarg>
473 				 */
474 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'l',
475 				    MSG_ORIG(MSG_ARG_T_LIBRARY),
476 				    MSG_ARG_T_LIBRARY_SIZE)) != 0) {
477 					return (c);
478 
479 				/*
480 				 * Translate --library-path <optarg> to
481 				 * -L <optarg>
482 				 */
483 				} else if ((c = str2chr(lml, ndx, argc, argv,
484 				    arg, 'L', MSG_ORIG(MSG_ARG_T_LIBPATH),
485 				    MSG_ARG_T_LIBPATH_SIZE)) != 0) {
486 					return (c);
487 				}
488 				break;
489 			case 'n':
490 				/* Translate --no-undefined to -zdefs */
491 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'z',
492 				    MSG_ORIG(MSG_ARG_T_NOUNDEF), 0)) != 0) {
493 					optarg = (char *)MSG_ORIG(MSG_ARG_DEFS);
494 					return (c);
495 
496 				/*
497 				 * Translate --no-whole-archive to
498 				 * -z defaultextract
499 				 */
500 				} else if ((c = str2chr(lml, ndx, argc, argv,
501 				    arg, 'z',
502 				    MSG_ORIG(MSG_ARG_T_NOWHOLEARC), 0)) != 0) {
503 					optarg =
504 					    (char *)MSG_ORIG(MSG_ARG_DFLEXTRT);
505 					return (c);
506 				}
507 				break;
508 			case 'o':
509 				/* Translate --output <optarg> to -o <optarg> */
510 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'o',
511 				    MSG_ORIG(MSG_ARG_T_OUTPUT),
512 				    MSG_ARG_T_OUTPUT_SIZE)) != 0) {
513 					return (c);
514 				}
515 				break;
516 			case 'r':
517 				/* Translate --relocatable to -r */
518 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'r',
519 				    MSG_ORIG(MSG_ARG_T_RELOCATABLE), 0)) != 0) {
520 					return (c);
521 				}
522 				break;
523 			case 's':
524 				/* Translate --strip-all to -s */
525 				if ((c = str2chr(lml, ndx, argc, argv, arg, 's',
526 				    MSG_ORIG(MSG_ARG_T_STRIP), 0)) != 0) {
527 					return (c);
528 				}
529 				/*
530 				 * Translate --start-group to -z rescan-start
531 				 */
532 				if ((c = str2chr(lml, ndx, argc, argv,
533 				    arg, 'z',
534 				    MSG_ORIG(MSG_ARG_T_STARTGROUP), 0)) != 0) {
535 					optarg = (char *)
536 					    MSG_ORIG(MSG_ARG_RESCAN_START);
537 					return (c);
538 				}
539 				break;
540 			case 'u':
541 				/*
542 				 * Translate --undefined <optarg> to
543 				 * -u <optarg>
544 				 */
545 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'u',
546 				    MSG_ORIG(MSG_ARG_T_UNDEF),
547 				    MSG_ARG_T_UNDEF_SIZE)) != 0) {
548 					return (c);
549 				}
550 				break;
551 			case 'v':
552 				/* Translate --version to -V */
553 				if ((c = str2chr(lml, ndx, argc, argv, arg, 'V',
554 				    MSG_ORIG(MSG_ARG_T_VERSION), 0)) != 0) {
555 					return (c);
556 				}
557 				break;
558 			case 'w':
559 				/*
560 				 * Translate --whole-archive to -z alltextract
561 				 */
562 				if ((c = str2chr(lml, ndx, argc, argv,
563 				    arg, 'z',
564 				    MSG_ORIG(MSG_ARG_T_WHOLEARC), 0)) != 0) {
565 					optarg =
566 					    (char *)MSG_ORIG(MSG_ARG_ALLEXTRT);
567 					return (c);
568 				}
569 				break;
570 			}
571 			break;
572 		}
573 	}
574 	if ((c = getopt(argc, argv, MSG_ORIG(MSG_STR_OPTIONS))) != -1) {
575 		/*
576 		 * It is possible that a "-Wl," argument has been used to
577 		 * specify an option.  This isn't advertized ld(1) syntax, but
578 		 * compiler drivers and configuration tools, have been known to
579 		 * pass this compiler option to ld(1).  Strip off the "-Wl,"
580 		 * prefix and pass the option through.
581 		 */
582 		if ((c == 'W') && (strncmp(optarg,
583 		    MSG_ORIG(MSG_ARG_T_WL), MSG_ARG_T_WL_SIZE) == 0)) {
584 			DBG_CALL(Dbg_args_Wldel(lml, ndx, optarg));
585 			c = optarg[MSG_ARG_T_WL_SIZE];
586 			optarg += MSG_ARG_T_WL_SIZE + 1;
587 		}
588 	}
589 
590 	return (c);
591 }
592 
593 /*
594  * A compare routine for Isd_node AVLT trees.
595  */
596 int
597 isdavl_compare(const void *n1, const void *n2)
598 {
599 	uint_t		hash1, hash2;
600 	const char	*st1, *st2;
601 	int		rc;
602 
603 	hash1 = ((Isd_node *)n1)->isd_hash;
604 	hash2 = ((Isd_node *)n2)->isd_hash;
605 
606 	if (hash1 > hash2)
607 		return (1);
608 	if (hash1 < hash2)
609 		return (-1);
610 
611 	st1 = ((Isd_node *)n1)->isd_isp->is_name;
612 	st2 = ((Isd_node *)n2)->isd_isp->is_name;
613 
614 	rc = strcmp(st1, st2);
615 	if (rc > 0)
616 		return (1);
617 	if (rc < 0)
618 		return (-1);
619 	return (0);
620 }
621 
622 /*
623  * Messaging support - funnel everything through dgettext().
624  */
625 const char *
626 _libld_msg(Msg mid)
627 {
628 	return (dgettext(MSG_ORIG(MSG_SUNW_OST_SGS), MSG_ORIG(mid)));
629 }
630 
631 /*
632  * Determine whether a symbol name should be demangled.
633  */
634 const char *
635 demangle(const char *name)
636 {
637 	if (demangle_flag)
638 		return (Elf_demangle_name(name));
639 	else
640 		return (name);
641 }
642