dt_module.c revision 297077
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 (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
24 */
25/*
26 * Copyright (c) 2013, Joyent, Inc.  All rights reserved.
27 */
28
29#include <sys/types.h>
30#ifdef illumos
31#include <sys/modctl.h>
32#include <sys/kobj.h>
33#include <sys/kobj_impl.h>
34#include <sys/sysmacros.h>
35#include <sys/elf.h>
36#include <sys/task.h>
37#else
38#include <sys/param.h>
39#include <sys/linker.h>
40#include <sys/stat.h>
41#endif
42
43#include <unistd.h>
44#ifdef illumos
45#include <project.h>
46#endif
47#include <strings.h>
48#include <stdlib.h>
49#include <libelf.h>
50#include <limits.h>
51#include <assert.h>
52#include <errno.h>
53#include <dirent.h>
54#ifndef illumos
55#include <fcntl.h>
56#include <libproc_compat.h>
57#endif
58
59#include <dt_strtab.h>
60#include <dt_module.h>
61#include <dt_impl.h>
62
63static const char *dt_module_strtab; /* active strtab for qsort callbacks */
64
65static void
66dt_module_symhash_insert(dt_module_t *dmp, const char *name, uint_t id)
67{
68	dt_sym_t *dsp = &dmp->dm_symchains[dmp->dm_symfree];
69	uint_t h;
70
71	assert(dmp->dm_symfree < dmp->dm_nsymelems + 1);
72
73	dsp->ds_symid = id;
74	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
75	dsp->ds_next = dmp->dm_symbuckets[h];
76	dmp->dm_symbuckets[h] = dmp->dm_symfree++;
77}
78
79static uint_t
80dt_module_syminit32(dt_module_t *dmp)
81{
82#if STT_NUM != (STT_TLS + 1)
83#error "STT_NUM has grown. update dt_module_syminit32()"
84#endif
85
86	Elf32_Sym *sym = dmp->dm_symtab.cts_data;
87	const char *base = dmp->dm_strtab.cts_data;
88	size_t ss_size = dmp->dm_strtab.cts_size;
89	uint_t i, n = dmp->dm_nsymelems;
90	uint_t asrsv = 0;
91
92#if defined(__FreeBSD__)
93	GElf_Ehdr ehdr;
94	int is_elf_obj;
95
96	gelf_getehdr(dmp->dm_elf, &ehdr);
97	is_elf_obj = (ehdr.e_type == ET_REL);
98#endif
99
100	for (i = 0; i < n; i++, sym++) {
101		const char *name = base + sym->st_name;
102		uchar_t type = ELF32_ST_TYPE(sym->st_info);
103
104		if (type >= STT_NUM || type == STT_SECTION)
105			continue; /* skip sections and unknown types */
106
107		if (sym->st_name == 0 || sym->st_name >= ss_size)
108			continue; /* skip null or invalid names */
109
110		if (sym->st_value != 0 &&
111		    (ELF32_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size)) {
112			asrsv++; /* reserve space in the address map */
113
114#if defined(__FreeBSD__)
115			sym->st_value += (Elf_Addr) dmp->dm_reloc_offset;
116			if (is_elf_obj && sym->st_shndx != SHN_UNDEF &&
117			    sym->st_shndx < ehdr.e_shnum)
118				sym->st_value +=
119				    dmp->dm_sec_offsets[sym->st_shndx];
120#endif
121		}
122
123		dt_module_symhash_insert(dmp, name, i);
124	}
125
126	return (asrsv);
127}
128
129static uint_t
130dt_module_syminit64(dt_module_t *dmp)
131{
132#if STT_NUM != (STT_TLS + 1)
133#error "STT_NUM has grown. update dt_module_syminit64()"
134#endif
135
136	Elf64_Sym *sym = dmp->dm_symtab.cts_data;
137	const char *base = dmp->dm_strtab.cts_data;
138	size_t ss_size = dmp->dm_strtab.cts_size;
139	uint_t i, n = dmp->dm_nsymelems;
140	uint_t asrsv = 0;
141
142#if defined(__FreeBSD__)
143	GElf_Ehdr ehdr;
144	int is_elf_obj;
145
146	gelf_getehdr(dmp->dm_elf, &ehdr);
147	is_elf_obj = (ehdr.e_type == ET_REL);
148#endif
149
150	for (i = 0; i < n; i++, sym++) {
151		const char *name = base + sym->st_name;
152		uchar_t type = ELF64_ST_TYPE(sym->st_info);
153
154		if (type >= STT_NUM || type == STT_SECTION)
155			continue; /* skip sections and unknown types */
156
157		if (sym->st_name == 0 || sym->st_name >= ss_size)
158			continue; /* skip null or invalid names */
159
160		if (sym->st_value != 0 &&
161		    (ELF64_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size)) {
162			asrsv++; /* reserve space in the address map */
163#if defined(__FreeBSD__)
164			sym->st_value += (Elf_Addr) dmp->dm_reloc_offset;
165			if (is_elf_obj && sym->st_shndx != SHN_UNDEF &&
166			    sym->st_shndx < ehdr.e_shnum)
167				sym->st_value +=
168				    dmp->dm_sec_offsets[sym->st_shndx];
169#endif
170		}
171
172		dt_module_symhash_insert(dmp, name, i);
173	}
174
175	return (asrsv);
176}
177
178/*
179 * Sort comparison function for 32-bit symbol address-to-name lookups.  We sort
180 * symbols by value.  If values are equal, we prefer the symbol that is
181 * non-zero sized, typed, not weak, or lexically first, in that order.
182 */
183static int
184dt_module_symcomp32(const void *lp, const void *rp)
185{
186	Elf32_Sym *lhs = *((Elf32_Sym **)lp);
187	Elf32_Sym *rhs = *((Elf32_Sym **)rp);
188
189	if (lhs->st_value != rhs->st_value)
190		return (lhs->st_value > rhs->st_value ? 1 : -1);
191
192	if ((lhs->st_size == 0) != (rhs->st_size == 0))
193		return (lhs->st_size == 0 ? 1 : -1);
194
195	if ((ELF32_ST_TYPE(lhs->st_info) == STT_NOTYPE) !=
196	    (ELF32_ST_TYPE(rhs->st_info) == STT_NOTYPE))
197		return (ELF32_ST_TYPE(lhs->st_info) == STT_NOTYPE ? 1 : -1);
198
199	if ((ELF32_ST_BIND(lhs->st_info) == STB_WEAK) !=
200	    (ELF32_ST_BIND(rhs->st_info) == STB_WEAK))
201		return (ELF32_ST_BIND(lhs->st_info) == STB_WEAK ? 1 : -1);
202
203	return (strcmp(dt_module_strtab + lhs->st_name,
204	    dt_module_strtab + rhs->st_name));
205}
206
207/*
208 * Sort comparison function for 64-bit symbol address-to-name lookups.  We sort
209 * symbols by value.  If values are equal, we prefer the symbol that is
210 * non-zero sized, typed, not weak, or lexically first, in that order.
211 */
212static int
213dt_module_symcomp64(const void *lp, const void *rp)
214{
215	Elf64_Sym *lhs = *((Elf64_Sym **)lp);
216	Elf64_Sym *rhs = *((Elf64_Sym **)rp);
217
218	if (lhs->st_value != rhs->st_value)
219		return (lhs->st_value > rhs->st_value ? 1 : -1);
220
221	if ((lhs->st_size == 0) != (rhs->st_size == 0))
222		return (lhs->st_size == 0 ? 1 : -1);
223
224	if ((ELF64_ST_TYPE(lhs->st_info) == STT_NOTYPE) !=
225	    (ELF64_ST_TYPE(rhs->st_info) == STT_NOTYPE))
226		return (ELF64_ST_TYPE(lhs->st_info) == STT_NOTYPE ? 1 : -1);
227
228	if ((ELF64_ST_BIND(lhs->st_info) == STB_WEAK) !=
229	    (ELF64_ST_BIND(rhs->st_info) == STB_WEAK))
230		return (ELF64_ST_BIND(lhs->st_info) == STB_WEAK ? 1 : -1);
231
232	return (strcmp(dt_module_strtab + lhs->st_name,
233	    dt_module_strtab + rhs->st_name));
234}
235
236static void
237dt_module_symsort32(dt_module_t *dmp)
238{
239	Elf32_Sym *symtab = (Elf32_Sym *)dmp->dm_symtab.cts_data;
240	Elf32_Sym **sympp = (Elf32_Sym **)dmp->dm_asmap;
241	const dt_sym_t *dsp = dmp->dm_symchains + 1;
242	uint_t i, n = dmp->dm_symfree;
243
244	for (i = 1; i < n; i++, dsp++) {
245		Elf32_Sym *sym = symtab + dsp->ds_symid;
246		if (sym->st_value != 0 &&
247		    (ELF32_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size))
248			*sympp++ = sym;
249	}
250
251	dmp->dm_aslen = (uint_t)(sympp - (Elf32_Sym **)dmp->dm_asmap);
252	assert(dmp->dm_aslen <= dmp->dm_asrsv);
253
254	dt_module_strtab = dmp->dm_strtab.cts_data;
255	qsort(dmp->dm_asmap, dmp->dm_aslen,
256	    sizeof (Elf32_Sym *), dt_module_symcomp32);
257	dt_module_strtab = NULL;
258}
259
260static void
261dt_module_symsort64(dt_module_t *dmp)
262{
263	Elf64_Sym *symtab = (Elf64_Sym *)dmp->dm_symtab.cts_data;
264	Elf64_Sym **sympp = (Elf64_Sym **)dmp->dm_asmap;
265	const dt_sym_t *dsp = dmp->dm_symchains + 1;
266	uint_t i, n = dmp->dm_symfree;
267
268	for (i = 1; i < n; i++, dsp++) {
269		Elf64_Sym *sym = symtab + dsp->ds_symid;
270		if (sym->st_value != 0 &&
271		    (ELF64_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size))
272			*sympp++ = sym;
273	}
274
275	dmp->dm_aslen = (uint_t)(sympp - (Elf64_Sym **)dmp->dm_asmap);
276	assert(dmp->dm_aslen <= dmp->dm_asrsv);
277
278	dt_module_strtab = dmp->dm_strtab.cts_data;
279	qsort(dmp->dm_asmap, dmp->dm_aslen,
280	    sizeof (Elf64_Sym *), dt_module_symcomp64);
281	dt_module_strtab = NULL;
282}
283
284static GElf_Sym *
285dt_module_symgelf32(const Elf32_Sym *src, GElf_Sym *dst)
286{
287	if (dst != NULL) {
288		dst->st_name = src->st_name;
289		dst->st_info = src->st_info;
290		dst->st_other = src->st_other;
291		dst->st_shndx = src->st_shndx;
292		dst->st_value = src->st_value;
293		dst->st_size = src->st_size;
294	}
295
296	return (dst);
297}
298
299static GElf_Sym *
300dt_module_symgelf64(const Elf64_Sym *src, GElf_Sym *dst)
301{
302	if (dst != NULL)
303		bcopy(src, dst, sizeof (GElf_Sym));
304
305	return (dst);
306}
307
308static GElf_Sym *
309dt_module_symname32(dt_module_t *dmp, const char *name,
310    GElf_Sym *symp, uint_t *idp)
311{
312	const Elf32_Sym *symtab = dmp->dm_symtab.cts_data;
313	const char *strtab = dmp->dm_strtab.cts_data;
314
315	const Elf32_Sym *sym;
316	const dt_sym_t *dsp;
317	uint_t i, h;
318
319	if (dmp->dm_nsymelems == 0)
320		return (NULL);
321
322	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
323
324	for (i = dmp->dm_symbuckets[h]; i != 0; i = dsp->ds_next) {
325		dsp = &dmp->dm_symchains[i];
326		sym = symtab + dsp->ds_symid;
327
328		if (strcmp(name, strtab + sym->st_name) == 0) {
329			if (idp != NULL)
330				*idp = dsp->ds_symid;
331			return (dt_module_symgelf32(sym, symp));
332		}
333	}
334
335	return (NULL);
336}
337
338static GElf_Sym *
339dt_module_symname64(dt_module_t *dmp, const char *name,
340    GElf_Sym *symp, uint_t *idp)
341{
342	const Elf64_Sym *symtab = dmp->dm_symtab.cts_data;
343	const char *strtab = dmp->dm_strtab.cts_data;
344
345	const Elf64_Sym *sym;
346	const dt_sym_t *dsp;
347	uint_t i, h;
348
349	if (dmp->dm_nsymelems == 0)
350		return (NULL);
351
352	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
353
354	for (i = dmp->dm_symbuckets[h]; i != 0; i = dsp->ds_next) {
355		dsp = &dmp->dm_symchains[i];
356		sym = symtab + dsp->ds_symid;
357
358		if (strcmp(name, strtab + sym->st_name) == 0) {
359			if (idp != NULL)
360				*idp = dsp->ds_symid;
361			return (dt_module_symgelf64(sym, symp));
362		}
363	}
364
365	return (NULL);
366}
367
368static GElf_Sym *
369dt_module_symaddr32(dt_module_t *dmp, GElf_Addr addr,
370    GElf_Sym *symp, uint_t *idp)
371{
372	const Elf32_Sym **asmap = (const Elf32_Sym **)dmp->dm_asmap;
373	const Elf32_Sym *symtab = dmp->dm_symtab.cts_data;
374	const Elf32_Sym *sym;
375
376	uint_t i, mid, lo = 0, hi = dmp->dm_aslen - 1;
377	Elf32_Addr v;
378
379	if (dmp->dm_aslen == 0)
380		return (NULL);
381
382	while (hi - lo > 1) {
383		mid = (lo + hi) / 2;
384		if (addr >= asmap[mid]->st_value)
385			lo = mid;
386		else
387			hi = mid;
388	}
389
390	i = addr < asmap[hi]->st_value ? lo : hi;
391	sym = asmap[i];
392	v = sym->st_value;
393
394	/*
395	 * If the previous entry has the same value, improve our choice.  The
396	 * order of equal-valued symbols is determined by the comparison func.
397	 */
398	while (i-- != 0 && asmap[i]->st_value == v)
399		sym = asmap[i];
400
401	if (addr - sym->st_value < MAX(sym->st_size, 1)) {
402		if (idp != NULL)
403			*idp = (uint_t)(sym - symtab);
404		return (dt_module_symgelf32(sym, symp));
405	}
406
407	return (NULL);
408}
409
410static GElf_Sym *
411dt_module_symaddr64(dt_module_t *dmp, GElf_Addr addr,
412    GElf_Sym *symp, uint_t *idp)
413{
414	const Elf64_Sym **asmap = (const Elf64_Sym **)dmp->dm_asmap;
415	const Elf64_Sym *symtab = dmp->dm_symtab.cts_data;
416	const Elf64_Sym *sym;
417
418	uint_t i, mid, lo = 0, hi = dmp->dm_aslen - 1;
419	Elf64_Addr v;
420
421	if (dmp->dm_aslen == 0)
422		return (NULL);
423
424	while (hi - lo > 1) {
425		mid = (lo + hi) / 2;
426		if (addr >= asmap[mid]->st_value)
427			lo = mid;
428		else
429			hi = mid;
430	}
431
432	i = addr < asmap[hi]->st_value ? lo : hi;
433	sym = asmap[i];
434	v = sym->st_value;
435
436	/*
437	 * If the previous entry has the same value, improve our choice.  The
438	 * order of equal-valued symbols is determined by the comparison func.
439	 */
440	while (i-- != 0 && asmap[i]->st_value == v)
441		sym = asmap[i];
442
443	if (addr - sym->st_value < MAX(sym->st_size, 1)) {
444		if (idp != NULL)
445			*idp = (uint_t)(sym - symtab);
446		return (dt_module_symgelf64(sym, symp));
447	}
448
449	return (NULL);
450}
451
452static const dt_modops_t dt_modops_32 = {
453	dt_module_syminit32,
454	dt_module_symsort32,
455	dt_module_symname32,
456	dt_module_symaddr32
457};
458
459static const dt_modops_t dt_modops_64 = {
460	dt_module_syminit64,
461	dt_module_symsort64,
462	dt_module_symname64,
463	dt_module_symaddr64
464};
465
466dt_module_t *
467dt_module_create(dtrace_hdl_t *dtp, const char *name)
468{
469	long pid;
470	char *eptr;
471	dt_ident_t *idp;
472	uint_t h = dt_strtab_hash(name, NULL) % dtp->dt_modbuckets;
473	dt_module_t *dmp;
474
475	for (dmp = dtp->dt_mods[h]; dmp != NULL; dmp = dmp->dm_next) {
476		if (strcmp(dmp->dm_name, name) == 0)
477			return (dmp);
478	}
479
480	if ((dmp = malloc(sizeof (dt_module_t))) == NULL)
481		return (NULL); /* caller must handle allocation failure */
482
483	bzero(dmp, sizeof (dt_module_t));
484	(void) strlcpy(dmp->dm_name, name, sizeof (dmp->dm_name));
485	dt_list_append(&dtp->dt_modlist, dmp);
486	dmp->dm_next = dtp->dt_mods[h];
487	dtp->dt_mods[h] = dmp;
488	dtp->dt_nmods++;
489
490	if (dtp->dt_conf.dtc_ctfmodel == CTF_MODEL_LP64)
491		dmp->dm_ops = &dt_modops_64;
492	else
493		dmp->dm_ops = &dt_modops_32;
494
495	/*
496	 * Modules for userland processes are special. They always refer to a
497	 * specific process and have a copy of their CTF data from a specific
498	 * instant in time. Any dt_module_t that begins with 'pid' is a module
499	 * for a specific process, much like how any probe description that
500	 * begins with 'pid' is special. pid123 refers to process 123. A module
501	 * that is just 'pid' refers specifically to pid$target. This is
502	 * generally done as D does not currently allow for macros to be
503	 * evaluated when working with types.
504	 */
505	if (strncmp(dmp->dm_name, "pid", 3) == 0) {
506		errno = 0;
507		if (dmp->dm_name[3] == '\0') {
508			idp = dt_idhash_lookup(dtp->dt_macros, "target");
509			if (idp != NULL && idp->di_id != 0)
510				dmp->dm_pid = idp->di_id;
511		} else {
512			pid = strtol(dmp->dm_name + 3, &eptr, 10);
513			if (errno == 0 && *eptr == '\0')
514				dmp->dm_pid = (pid_t)pid;
515			else
516				dt_dprintf("encountered malformed pid "
517				    "module: %s\n", dmp->dm_name);
518		}
519	}
520
521	return (dmp);
522}
523
524dt_module_t *
525dt_module_lookup_by_name(dtrace_hdl_t *dtp, const char *name)
526{
527	uint_t h = dt_strtab_hash(name, NULL) % dtp->dt_modbuckets;
528	dt_module_t *dmp;
529
530	for (dmp = dtp->dt_mods[h]; dmp != NULL; dmp = dmp->dm_next) {
531		if (strcmp(dmp->dm_name, name) == 0)
532			return (dmp);
533	}
534
535	return (NULL);
536}
537
538/*ARGSUSED*/
539dt_module_t *
540dt_module_lookup_by_ctf(dtrace_hdl_t *dtp, ctf_file_t *ctfp)
541{
542	return (ctfp ? ctf_getspecific(ctfp) : NULL);
543}
544
545static int
546dt_module_load_sect(dtrace_hdl_t *dtp, dt_module_t *dmp, ctf_sect_t *ctsp)
547{
548	const char *s;
549	size_t shstrs;
550	GElf_Shdr sh;
551	Elf_Data *dp;
552	Elf_Scn *sp;
553
554	if (elf_getshdrstrndx(dmp->dm_elf, &shstrs) == -1)
555		return (dt_set_errno(dtp, EDT_NOTLOADED));
556
557	for (sp = NULL; (sp = elf_nextscn(dmp->dm_elf, sp)) != NULL; ) {
558		if (gelf_getshdr(sp, &sh) == NULL || sh.sh_type == SHT_NULL ||
559		    (s = elf_strptr(dmp->dm_elf, shstrs, sh.sh_name)) == NULL)
560			continue; /* skip any malformed sections */
561
562		if (sh.sh_type == ctsp->cts_type &&
563		    sh.sh_entsize == ctsp->cts_entsize &&
564		    strcmp(s, ctsp->cts_name) == 0)
565			break; /* section matches specification */
566	}
567
568	/*
569	 * If the section isn't found, return success but leave cts_data set
570	 * to NULL and cts_size set to zero for our caller.
571	 */
572	if (sp == NULL || (dp = elf_getdata(sp, NULL)) == NULL)
573		return (0);
574
575#ifdef illumos
576	ctsp->cts_data = dp->d_buf;
577#else
578	if ((ctsp->cts_data = malloc(dp->d_size)) == NULL)
579		return (0);
580	memcpy(ctsp->cts_data, dp->d_buf, dp->d_size);
581#endif
582	ctsp->cts_size = dp->d_size;
583
584	dt_dprintf("loaded %s [%s] (%lu bytes)\n",
585	    dmp->dm_name, ctsp->cts_name, (ulong_t)ctsp->cts_size);
586
587	return (0);
588}
589
590typedef struct dt_module_cb_arg {
591	struct ps_prochandle *dpa_proc;
592	dtrace_hdl_t *dpa_dtp;
593	dt_module_t *dpa_dmp;
594	uint_t dpa_count;
595} dt_module_cb_arg_t;
596
597/* ARGSUSED */
598static int
599dt_module_load_proc_count(void *arg, const prmap_t *prmap, const char *obj)
600{
601	ctf_file_t *fp;
602	dt_module_cb_arg_t *dcp = arg;
603
604	/* Try to grab a ctf container if it exists */
605	fp = Pname_to_ctf(dcp->dpa_proc, obj);
606	if (fp != NULL)
607		dcp->dpa_count++;
608	return (0);
609}
610
611/* ARGSUSED */
612static int
613dt_module_load_proc_build(void *arg, const prmap_t *prmap, const char *obj)
614{
615	ctf_file_t *fp;
616	char buf[MAXPATHLEN], *p;
617	dt_module_cb_arg_t *dcp = arg;
618	int count = dcp->dpa_count;
619	Lmid_t lmid;
620
621	fp = Pname_to_ctf(dcp->dpa_proc, obj);
622	if (fp == NULL)
623		return (0);
624	fp = ctf_dup(fp);
625	if (fp == NULL)
626		return (0);
627	dcp->dpa_dmp->dm_libctfp[count] = fp;
628	/*
629	 * While it'd be nice to simply use objname here, because of our prior
630	 * actions we'll always get a resolved object name to its on disk file.
631	 * Like the pid provider, we need to tell a bit of a lie here. The type
632	 * that the user thinks of is in terms of the libraries they requested,
633	 * eg. libc.so.1, they don't care about the fact that it's
634	 * libc_hwcap.so.1.
635	 */
636	(void) Pobjname(dcp->dpa_proc, prmap->pr_vaddr, buf, sizeof (buf));
637	if ((p = strrchr(buf, '/')) == NULL)
638		p = buf;
639	else
640		p++;
641
642	/*
643	 * If for some reason we can't find a link map id for this module, which
644	 * would be really quite weird. We instead just say the link map id is
645	 * zero.
646	 */
647	if (Plmid(dcp->dpa_proc, prmap->pr_vaddr, &lmid) != 0)
648		lmid = 0;
649
650	if (lmid == 0)
651		dcp->dpa_dmp->dm_libctfn[count] = strdup(p);
652	else
653		(void) asprintf(&dcp->dpa_dmp->dm_libctfn[count],
654		    "LM%x`%s", lmid, p);
655	if (dcp->dpa_dmp->dm_libctfn[count] == NULL)
656		return (1);
657	ctf_setspecific(fp, dcp->dpa_dmp);
658	dcp->dpa_count++;
659	return (0);
660}
661
662/*
663 * We've been asked to load data that belongs to another process. As such we're
664 * going to pgrab it at this instant, load everything that we might ever care
665 * about, and then drive on. The reason for this is that the process that we're
666 * interested in might be changing. As long as we have grabbed it, then this
667 * can't be a problem for us.
668 *
669 * For now, we're actually going to punt on most things and just try to get CTF
670 * data, nothing else. Basically this is only useful as a source of type
671 * information, we can't go and do the stacktrace lookups, etc.
672 */
673static int
674dt_module_load_proc(dtrace_hdl_t *dtp, dt_module_t *dmp)
675{
676	struct ps_prochandle *p;
677	dt_module_cb_arg_t arg;
678
679	/*
680	 * Note that on success we do not release this hold. We must hold this
681	 * for our life time.
682	 */
683	p = dt_proc_grab(dtp, dmp->dm_pid, 0, PGRAB_RDONLY | PGRAB_FORCE);
684	if (p == NULL) {
685		dt_dprintf("failed to grab pid: %d\n", (int)dmp->dm_pid);
686		return (dt_set_errno(dtp, EDT_CANTLOAD));
687	}
688	dt_proc_lock(dtp, p);
689
690	arg.dpa_proc = p;
691	arg.dpa_dtp = dtp;
692	arg.dpa_dmp = dmp;
693	arg.dpa_count = 0;
694	if (Pobject_iter_resolved(p, dt_module_load_proc_count, &arg) != 0) {
695		dt_dprintf("failed to iterate objects\n");
696		dt_proc_release(dtp, p);
697		return (dt_set_errno(dtp, EDT_CANTLOAD));
698	}
699
700	if (arg.dpa_count == 0) {
701		dt_dprintf("no ctf data present\n");
702		dt_proc_unlock(dtp, p);
703		dt_proc_release(dtp, p);
704		return (dt_set_errno(dtp, EDT_CANTLOAD));
705	}
706
707	dmp->dm_libctfp = malloc(sizeof (ctf_file_t *) * arg.dpa_count);
708	if (dmp->dm_libctfp == NULL) {
709		dt_proc_unlock(dtp, p);
710		dt_proc_release(dtp, p);
711		return (dt_set_errno(dtp, EDT_NOMEM));
712	}
713	bzero(dmp->dm_libctfp, sizeof (ctf_file_t *) * arg.dpa_count);
714
715	dmp->dm_libctfn = malloc(sizeof (char *) * arg.dpa_count);
716	if (dmp->dm_libctfn == NULL) {
717		free(dmp->dm_libctfp);
718		dt_proc_unlock(dtp, p);
719		dt_proc_release(dtp, p);
720		return (dt_set_errno(dtp, EDT_NOMEM));
721	}
722	bzero(dmp->dm_libctfn, sizeof (char *) * arg.dpa_count);
723
724	dmp->dm_nctflibs = arg.dpa_count;
725
726	arg.dpa_count = 0;
727	if (Pobject_iter_resolved(p, dt_module_load_proc_build, &arg) != 0) {
728		dt_proc_unlock(dtp, p);
729		dt_module_unload(dtp, dmp);
730		dt_proc_release(dtp, p);
731		return (dt_set_errno(dtp, EDT_CANTLOAD));
732	}
733	assert(arg.dpa_count == dmp->dm_nctflibs);
734	dt_dprintf("loaded %d ctf modules for pid %d\n", arg.dpa_count,
735	    (int)dmp->dm_pid);
736
737	dt_proc_unlock(dtp, p);
738	dt_proc_release(dtp, p);
739	dmp->dm_flags |= DT_DM_LOADED;
740
741	return (0);
742}
743
744int
745dt_module_load(dtrace_hdl_t *dtp, dt_module_t *dmp)
746{
747	if (dmp->dm_flags & DT_DM_LOADED)
748		return (0); /* module is already loaded */
749
750	if (dmp->dm_pid != 0)
751		return (dt_module_load_proc(dtp, dmp));
752
753	dmp->dm_ctdata.cts_name = ".SUNW_ctf";
754	dmp->dm_ctdata.cts_type = SHT_PROGBITS;
755	dmp->dm_ctdata.cts_flags = 0;
756	dmp->dm_ctdata.cts_data = NULL;
757	dmp->dm_ctdata.cts_size = 0;
758	dmp->dm_ctdata.cts_entsize = 0;
759	dmp->dm_ctdata.cts_offset = 0;
760
761	dmp->dm_symtab.cts_name = ".symtab";
762	dmp->dm_symtab.cts_type = SHT_SYMTAB;
763	dmp->dm_symtab.cts_flags = 0;
764	dmp->dm_symtab.cts_data = NULL;
765	dmp->dm_symtab.cts_size = 0;
766	dmp->dm_symtab.cts_entsize = dmp->dm_ops == &dt_modops_64 ?
767	    sizeof (Elf64_Sym) : sizeof (Elf32_Sym);
768	dmp->dm_symtab.cts_offset = 0;
769
770	dmp->dm_strtab.cts_name = ".strtab";
771	dmp->dm_strtab.cts_type = SHT_STRTAB;
772	dmp->dm_strtab.cts_flags = 0;
773	dmp->dm_strtab.cts_data = NULL;
774	dmp->dm_strtab.cts_size = 0;
775	dmp->dm_strtab.cts_entsize = 0;
776	dmp->dm_strtab.cts_offset = 0;
777
778	/*
779	 * Attempt to load the module's CTF section, symbol table section, and
780	 * string table section.  Note that modules may not contain CTF data:
781	 * this will result in a successful load_sect but data of size zero.
782	 * We will then fail if dt_module_getctf() is called, as shown below.
783	 */
784	if (dt_module_load_sect(dtp, dmp, &dmp->dm_ctdata) == -1 ||
785	    dt_module_load_sect(dtp, dmp, &dmp->dm_symtab) == -1 ||
786	    dt_module_load_sect(dtp, dmp, &dmp->dm_strtab) == -1) {
787		dt_module_unload(dtp, dmp);
788		return (-1); /* dt_errno is set for us */
789	}
790
791	/*
792	 * Allocate the hash chains and hash buckets for symbol name lookup.
793	 * This is relatively simple since the symbol table is of fixed size
794	 * and is known in advance.  We allocate one extra element since we
795	 * use element indices instead of pointers and zero is our sentinel.
796	 */
797	dmp->dm_nsymelems =
798	    dmp->dm_symtab.cts_size / dmp->dm_symtab.cts_entsize;
799
800	dmp->dm_nsymbuckets = _dtrace_strbuckets;
801	dmp->dm_symfree = 1;		/* first free element is index 1 */
802
803	dmp->dm_symbuckets = malloc(sizeof (uint_t) * dmp->dm_nsymbuckets);
804	dmp->dm_symchains = malloc(sizeof (dt_sym_t) * dmp->dm_nsymelems + 1);
805
806	if (dmp->dm_symbuckets == NULL || dmp->dm_symchains == NULL) {
807		dt_module_unload(dtp, dmp);
808		return (dt_set_errno(dtp, EDT_NOMEM));
809	}
810
811	bzero(dmp->dm_symbuckets, sizeof (uint_t) * dmp->dm_nsymbuckets);
812	bzero(dmp->dm_symchains, sizeof (dt_sym_t) * dmp->dm_nsymelems + 1);
813
814	/*
815	 * Iterate over the symbol table data buffer and insert each symbol
816	 * name into the name hash if the name and type are valid.  Then
817	 * allocate the address map, fill it in, and sort it.
818	 */
819	dmp->dm_asrsv = dmp->dm_ops->do_syminit(dmp);
820
821	dt_dprintf("hashed %s [%s] (%u symbols)\n",
822	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_symfree - 1);
823
824	if ((dmp->dm_asmap = malloc(sizeof (void *) * dmp->dm_asrsv)) == NULL) {
825		dt_module_unload(dtp, dmp);
826		return (dt_set_errno(dtp, EDT_NOMEM));
827	}
828
829	dmp->dm_ops->do_symsort(dmp);
830
831	dt_dprintf("sorted %s [%s] (%u symbols)\n",
832	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_aslen);
833
834	dmp->dm_flags |= DT_DM_LOADED;
835	return (0);
836}
837
838int
839dt_module_hasctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
840{
841	if (dmp->dm_pid != 0 && dmp->dm_nctflibs > 0)
842		return (1);
843	return (dt_module_getctf(dtp, dmp) != NULL);
844}
845
846ctf_file_t *
847dt_module_getctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
848{
849	const char *parent;
850	dt_module_t *pmp;
851	ctf_file_t *pfp;
852	int model;
853
854	if (dmp->dm_ctfp != NULL || dt_module_load(dtp, dmp) != 0)
855		return (dmp->dm_ctfp);
856
857	if (dmp->dm_ops == &dt_modops_64)
858		model = CTF_MODEL_LP64;
859	else
860		model = CTF_MODEL_ILP32;
861
862	/*
863	 * If the data model of the module does not match our program data
864	 * model, then do not permit CTF from this module to be opened and
865	 * returned to the compiler.  If we support mixed data models in the
866	 * future for combined kernel/user tracing, this can be removed.
867	 */
868	if (dtp->dt_conf.dtc_ctfmodel != model) {
869		(void) dt_set_errno(dtp, EDT_DATAMODEL);
870		return (NULL);
871	}
872
873	if (dmp->dm_ctdata.cts_size == 0) {
874		(void) dt_set_errno(dtp, EDT_NOCTF);
875		return (NULL);
876	}
877
878	dmp->dm_ctfp = ctf_bufopen(&dmp->dm_ctdata,
879	    &dmp->dm_symtab, &dmp->dm_strtab, &dtp->dt_ctferr);
880
881	if (dmp->dm_ctfp == NULL) {
882		(void) dt_set_errno(dtp, EDT_CTF);
883		return (NULL);
884	}
885
886	(void) ctf_setmodel(dmp->dm_ctfp, model);
887	ctf_setspecific(dmp->dm_ctfp, dmp);
888
889	if ((parent = ctf_parent_name(dmp->dm_ctfp)) != NULL) {
890		if ((pmp = dt_module_create(dtp, parent)) == NULL ||
891		    (pfp = dt_module_getctf(dtp, pmp)) == NULL) {
892			if (pmp == NULL)
893				(void) dt_set_errno(dtp, EDT_NOMEM);
894			goto err;
895		}
896
897		if (ctf_import(dmp->dm_ctfp, pfp) == CTF_ERR) {
898			dtp->dt_ctferr = ctf_errno(dmp->dm_ctfp);
899			(void) dt_set_errno(dtp, EDT_CTF);
900			goto err;
901		}
902	}
903
904	dt_dprintf("loaded CTF container for %s (%p)\n",
905	    dmp->dm_name, (void *)dmp->dm_ctfp);
906
907	return (dmp->dm_ctfp);
908
909err:
910	ctf_close(dmp->dm_ctfp);
911	dmp->dm_ctfp = NULL;
912	return (NULL);
913}
914
915/*ARGSUSED*/
916void
917dt_module_unload(dtrace_hdl_t *dtp, dt_module_t *dmp)
918{
919	int i;
920
921	ctf_close(dmp->dm_ctfp);
922	dmp->dm_ctfp = NULL;
923
924#ifndef illumos
925	if (dmp->dm_ctdata.cts_data != NULL) {
926		free(dmp->dm_ctdata.cts_data);
927	}
928	if (dmp->dm_symtab.cts_data != NULL) {
929		free(dmp->dm_symtab.cts_data);
930	}
931	if (dmp->dm_strtab.cts_data != NULL) {
932		free(dmp->dm_strtab.cts_data);
933	}
934#endif
935
936	if (dmp->dm_libctfp != NULL) {
937		for (i = 0; i < dmp->dm_nctflibs; i++) {
938			ctf_close(dmp->dm_libctfp[i]);
939			free(dmp->dm_libctfn[i]);
940		}
941		free(dmp->dm_libctfp);
942		free(dmp->dm_libctfn);
943		dmp->dm_libctfp = NULL;
944		dmp->dm_nctflibs = 0;
945	}
946
947	bzero(&dmp->dm_ctdata, sizeof (ctf_sect_t));
948	bzero(&dmp->dm_symtab, sizeof (ctf_sect_t));
949	bzero(&dmp->dm_strtab, sizeof (ctf_sect_t));
950
951	if (dmp->dm_symbuckets != NULL) {
952		free(dmp->dm_symbuckets);
953		dmp->dm_symbuckets = NULL;
954	}
955
956	if (dmp->dm_symchains != NULL) {
957		free(dmp->dm_symchains);
958		dmp->dm_symchains = NULL;
959	}
960
961	if (dmp->dm_asmap != NULL) {
962		free(dmp->dm_asmap);
963		dmp->dm_asmap = NULL;
964	}
965#if defined(__FreeBSD__)
966	if (dmp->dm_sec_offsets != NULL) {
967		free(dmp->dm_sec_offsets);
968		dmp->dm_sec_offsets = NULL;
969	}
970#endif
971	dmp->dm_symfree = 0;
972	dmp->dm_nsymbuckets = 0;
973	dmp->dm_nsymelems = 0;
974	dmp->dm_asrsv = 0;
975	dmp->dm_aslen = 0;
976
977	dmp->dm_text_va = 0;
978	dmp->dm_text_size = 0;
979	dmp->dm_data_va = 0;
980	dmp->dm_data_size = 0;
981	dmp->dm_bss_va = 0;
982	dmp->dm_bss_size = 0;
983
984	if (dmp->dm_extern != NULL) {
985		dt_idhash_destroy(dmp->dm_extern);
986		dmp->dm_extern = NULL;
987	}
988
989	(void) elf_end(dmp->dm_elf);
990	dmp->dm_elf = NULL;
991
992	dmp->dm_pid = 0;
993
994	dmp->dm_flags &= ~DT_DM_LOADED;
995}
996
997void
998dt_module_destroy(dtrace_hdl_t *dtp, dt_module_t *dmp)
999{
1000	uint_t h = dt_strtab_hash(dmp->dm_name, NULL) % dtp->dt_modbuckets;
1001	dt_module_t **dmpp = &dtp->dt_mods[h];
1002
1003	dt_list_delete(&dtp->dt_modlist, dmp);
1004	assert(dtp->dt_nmods != 0);
1005	dtp->dt_nmods--;
1006
1007	/*
1008	 * Now remove this module from its hash chain.  We expect to always
1009	 * find the module on its hash chain, so in this loop we assert that
1010	 * we don't run off the end of the list.
1011	 */
1012	while (*dmpp != dmp) {
1013		dmpp = &((*dmpp)->dm_next);
1014		assert(*dmpp != NULL);
1015	}
1016
1017	*dmpp = dmp->dm_next;
1018
1019	dt_module_unload(dtp, dmp);
1020	free(dmp);
1021}
1022
1023/*
1024 * Insert a new external symbol reference into the specified module.  The new
1025 * symbol will be marked as undefined and is assigned a symbol index beyond
1026 * any existing cached symbols from this module.  We use the ident's di_data
1027 * field to store a pointer to a copy of the dtrace_syminfo_t for this symbol.
1028 */
1029dt_ident_t *
1030dt_module_extern(dtrace_hdl_t *dtp, dt_module_t *dmp,
1031    const char *name, const dtrace_typeinfo_t *tip)
1032{
1033	dtrace_syminfo_t *sip;
1034	dt_ident_t *idp;
1035	uint_t id;
1036
1037	if (dmp->dm_extern == NULL && (dmp->dm_extern = dt_idhash_create(
1038	    "extern", NULL, dmp->dm_nsymelems, UINT_MAX)) == NULL) {
1039		(void) dt_set_errno(dtp, EDT_NOMEM);
1040		return (NULL);
1041	}
1042
1043	if (dt_idhash_nextid(dmp->dm_extern, &id) == -1) {
1044		(void) dt_set_errno(dtp, EDT_SYMOFLOW);
1045		return (NULL);
1046	}
1047
1048	if ((sip = malloc(sizeof (dtrace_syminfo_t))) == NULL) {
1049		(void) dt_set_errno(dtp, EDT_NOMEM);
1050		return (NULL);
1051	}
1052
1053	idp = dt_idhash_insert(dmp->dm_extern, name, DT_IDENT_SYMBOL, 0, id,
1054	    _dtrace_symattr, 0, &dt_idops_thaw, NULL, dtp->dt_gen);
1055
1056	if (idp == NULL) {
1057		(void) dt_set_errno(dtp, EDT_NOMEM);
1058		free(sip);
1059		return (NULL);
1060	}
1061
1062	sip->dts_object = dmp->dm_name;
1063	sip->dts_name = idp->di_name;
1064	sip->dts_id = idp->di_id;
1065
1066	idp->di_data = sip;
1067	idp->di_ctfp = tip->dtt_ctfp;
1068	idp->di_type = tip->dtt_type;
1069
1070	return (idp);
1071}
1072
1073const char *
1074dt_module_modelname(dt_module_t *dmp)
1075{
1076	if (dmp->dm_ops == &dt_modops_64)
1077		return ("64-bit");
1078	else
1079		return ("32-bit");
1080}
1081
1082/* ARGSUSED */
1083int
1084dt_module_getlibid(dtrace_hdl_t *dtp, dt_module_t *dmp, const ctf_file_t *fp)
1085{
1086	int i;
1087
1088	for (i = 0; i < dmp->dm_nctflibs; i++) {
1089		if (dmp->dm_libctfp[i] == fp)
1090			return (i);
1091	}
1092
1093	return (-1);
1094}
1095
1096/* ARGSUSED */
1097ctf_file_t *
1098dt_module_getctflib(dtrace_hdl_t *dtp, dt_module_t *dmp, const char *name)
1099{
1100	int i;
1101
1102	for (i = 0; i < dmp->dm_nctflibs; i++) {
1103		if (strcmp(dmp->dm_libctfn[i], name) == 0)
1104			return (dmp->dm_libctfp[i]);
1105	}
1106
1107	return (NULL);
1108}
1109
1110/*
1111 * Update our module cache by adding an entry for the specified module 'name'.
1112 * We create the dt_module_t and populate it using /system/object/<name>/.
1113 *
1114 * On FreeBSD, the module name is passed as the full module file name,
1115 * including the path.
1116 */
1117static void
1118#ifdef illumos
1119dt_module_update(dtrace_hdl_t *dtp, const char *name)
1120#else
1121dt_module_update(dtrace_hdl_t *dtp, struct kld_file_stat *k_stat)
1122#endif
1123{
1124	char fname[MAXPATHLEN];
1125	struct stat64 st;
1126	int fd, err, bits;
1127
1128	dt_module_t *dmp;
1129	const char *s;
1130	size_t shstrs;
1131	GElf_Shdr sh;
1132	Elf_Data *dp;
1133	Elf_Scn *sp;
1134
1135#ifdef illumos
1136	(void) snprintf(fname, sizeof (fname),
1137	    "%s/%s/object", OBJFS_ROOT, name);
1138#else
1139	GElf_Ehdr ehdr;
1140	GElf_Phdr ph;
1141	char name[MAXPATHLEN];
1142	uintptr_t mapbase, alignmask;
1143	int i = 0;
1144	int is_elf_obj;
1145
1146	(void) strlcpy(name, k_stat->name, sizeof(name));
1147	(void) strlcpy(fname, k_stat->pathname, sizeof(fname));
1148#endif
1149
1150	if ((fd = open(fname, O_RDONLY)) == -1 || fstat64(fd, &st) == -1 ||
1151	    (dmp = dt_module_create(dtp, name)) == NULL) {
1152		dt_dprintf("failed to open %s: %s\n", fname, strerror(errno));
1153		(void) close(fd);
1154		return;
1155	}
1156
1157	/*
1158	 * Since the module can unload out from under us (and /system/object
1159	 * will return ENOENT), tell libelf to cook the entire file now and
1160	 * then close the underlying file descriptor immediately.  If this
1161	 * succeeds, we know that we can continue safely using dmp->dm_elf.
1162	 */
1163	dmp->dm_elf = elf_begin(fd, ELF_C_READ, NULL);
1164	err = elf_cntl(dmp->dm_elf, ELF_C_FDREAD);
1165	(void) close(fd);
1166
1167	if (dmp->dm_elf == NULL || err == -1 ||
1168	    elf_getshdrstrndx(dmp->dm_elf, &shstrs) == -1) {
1169		dt_dprintf("failed to load %s: %s\n",
1170		    fname, elf_errmsg(elf_errno()));
1171		dt_module_destroy(dtp, dmp);
1172		return;
1173	}
1174
1175	switch (gelf_getclass(dmp->dm_elf)) {
1176	case ELFCLASS32:
1177		dmp->dm_ops = &dt_modops_32;
1178		bits = 32;
1179		break;
1180	case ELFCLASS64:
1181		dmp->dm_ops = &dt_modops_64;
1182		bits = 64;
1183		break;
1184	default:
1185		dt_dprintf("failed to load %s: unknown ELF class\n", fname);
1186		dt_module_destroy(dtp, dmp);
1187		return;
1188	}
1189#if defined(__FreeBSD__)
1190	mapbase = (uintptr_t)k_stat->address;
1191	gelf_getehdr(dmp->dm_elf, &ehdr);
1192	is_elf_obj = (ehdr.e_type == ET_REL);
1193	if (is_elf_obj) {
1194		dmp->dm_sec_offsets =
1195		    malloc(ehdr.e_shnum * sizeof(*dmp->dm_sec_offsets));
1196		if (dmp->dm_sec_offsets == NULL) {
1197			dt_dprintf("failed to allocate memory\n");
1198			dt_module_destroy(dtp, dmp);
1199			return;
1200		}
1201	}
1202#endif
1203	/*
1204	 * Iterate over the section headers locating various sections of
1205	 * interest and use their attributes to flesh out the dt_module_t.
1206	 */
1207	for (sp = NULL; (sp = elf_nextscn(dmp->dm_elf, sp)) != NULL; ) {
1208		if (gelf_getshdr(sp, &sh) == NULL || sh.sh_type == SHT_NULL ||
1209		    (s = elf_strptr(dmp->dm_elf, shstrs, sh.sh_name)) == NULL)
1210			continue; /* skip any malformed sections */
1211#if defined(__FreeBSD__)
1212		if (sh.sh_size == 0)
1213			continue;
1214		if (sh.sh_type == SHT_PROGBITS || sh.sh_type == SHT_NOBITS) {
1215			alignmask = sh.sh_addralign - 1;
1216			mapbase += alignmask;
1217			mapbase &= ~alignmask;
1218			sh.sh_addr = mapbase;
1219			if (is_elf_obj)
1220				dmp->dm_sec_offsets[elf_ndxscn(sp)] = sh.sh_addr;
1221			mapbase += sh.sh_size;
1222		}
1223#endif
1224		if (strcmp(s, ".text") == 0) {
1225			dmp->dm_text_size = sh.sh_size;
1226			dmp->dm_text_va = sh.sh_addr;
1227		} else if (strcmp(s, ".data") == 0) {
1228			dmp->dm_data_size = sh.sh_size;
1229			dmp->dm_data_va = sh.sh_addr;
1230		} else if (strcmp(s, ".bss") == 0) {
1231			dmp->dm_bss_size = sh.sh_size;
1232			dmp->dm_bss_va = sh.sh_addr;
1233		} else if (strcmp(s, ".info") == 0 &&
1234		    (dp = elf_getdata(sp, NULL)) != NULL) {
1235			bcopy(dp->d_buf, &dmp->dm_info,
1236			    MIN(sh.sh_size, sizeof (dmp->dm_info)));
1237		} else if (strcmp(s, ".filename") == 0 &&
1238		    (dp = elf_getdata(sp, NULL)) != NULL) {
1239			(void) strlcpy(dmp->dm_file,
1240			    dp->d_buf, sizeof (dmp->dm_file));
1241		}
1242	}
1243
1244	dmp->dm_flags |= DT_DM_KERNEL;
1245#ifdef illumos
1246	dmp->dm_modid = (int)OBJFS_MODID(st.st_ino);
1247#else
1248	/*
1249	 * Include .rodata and special sections into .text.
1250	 * This depends on default section layout produced by GNU ld
1251	 * for ELF objects and libraries:
1252	 * [Text][R/O data][R/W data][Dynamic][BSS][Non loadable]
1253	 */
1254	dmp->dm_text_size = dmp->dm_data_va - dmp->dm_text_va;
1255#if defined(__i386__)
1256	/*
1257	 * Find the first load section and figure out the relocation
1258	 * offset for the symbols. The kernel module will not need
1259	 * relocation, but the kernel linker modules will.
1260	 */
1261	for (i = 0; gelf_getphdr(dmp->dm_elf, i, &ph) != NULL; i++) {
1262		if (ph.p_type == PT_LOAD) {
1263			dmp->dm_reloc_offset = k_stat->address - ph.p_vaddr;
1264			break;
1265		}
1266	}
1267#endif
1268#endif /* illumos */
1269
1270	if (dmp->dm_info.objfs_info_primary)
1271		dmp->dm_flags |= DT_DM_PRIMARY;
1272
1273	dt_dprintf("opened %d-bit module %s (%s) [%d]\n",
1274	    bits, dmp->dm_name, dmp->dm_file, dmp->dm_modid);
1275}
1276
1277/*
1278 * Unload all the loaded modules and then refresh the module cache with the
1279 * latest list of loaded modules and their address ranges.
1280 */
1281void
1282dtrace_update(dtrace_hdl_t *dtp)
1283{
1284	dt_module_t *dmp;
1285	DIR *dirp;
1286#if defined(__FreeBSD__)
1287	int fileid;
1288#endif
1289
1290	for (dmp = dt_list_next(&dtp->dt_modlist);
1291	    dmp != NULL; dmp = dt_list_next(dmp))
1292		dt_module_unload(dtp, dmp);
1293
1294#ifdef illumos
1295	/*
1296	 * Open /system/object and attempt to create a libdtrace module for
1297	 * each kernel module that is loaded on the current system.
1298	 */
1299	if (!(dtp->dt_oflags & DTRACE_O_NOSYS) &&
1300	    (dirp = opendir(OBJFS_ROOT)) != NULL) {
1301		struct dirent *dp;
1302
1303		while ((dp = readdir(dirp)) != NULL) {
1304			if (dp->d_name[0] != '.')
1305				dt_module_update(dtp, dp->d_name);
1306		}
1307
1308		(void) closedir(dirp);
1309	}
1310#elif defined(__FreeBSD__)
1311	/*
1312	 * Use FreeBSD's kernel loader interface to discover what kernel
1313	 * modules are loaded and create a libdtrace module for each one.
1314	 */
1315	for (fileid = kldnext(0); fileid > 0; fileid = kldnext(fileid)) {
1316		struct kld_file_stat k_stat;
1317		k_stat.version = sizeof(k_stat);
1318		if (kldstat(fileid, &k_stat) == 0)
1319			dt_module_update(dtp, &k_stat);
1320	}
1321#endif
1322
1323	/*
1324	 * Look up all the macro identifiers and set di_id to the latest value.
1325	 * This code collaborates with dt_lex.l on the use of di_id.  We will
1326	 * need to implement something fancier if we need to support non-ints.
1327	 */
1328	dt_idhash_lookup(dtp->dt_macros, "egid")->di_id = getegid();
1329	dt_idhash_lookup(dtp->dt_macros, "euid")->di_id = geteuid();
1330	dt_idhash_lookup(dtp->dt_macros, "gid")->di_id = getgid();
1331	dt_idhash_lookup(dtp->dt_macros, "pid")->di_id = getpid();
1332	dt_idhash_lookup(dtp->dt_macros, "pgid")->di_id = getpgid(0);
1333	dt_idhash_lookup(dtp->dt_macros, "ppid")->di_id = getppid();
1334#ifdef illumos
1335	dt_idhash_lookup(dtp->dt_macros, "projid")->di_id = getprojid();
1336#endif
1337	dt_idhash_lookup(dtp->dt_macros, "sid")->di_id = getsid(0);
1338#ifdef illumos
1339	dt_idhash_lookup(dtp->dt_macros, "taskid")->di_id = gettaskid();
1340#endif
1341	dt_idhash_lookup(dtp->dt_macros, "uid")->di_id = getuid();
1342
1343	/*
1344	 * Cache the pointers to the modules representing the base executable
1345	 * and the run-time linker in the dtrace client handle. Note that on
1346	 * x86 krtld is folded into unix, so if we don't find it, use unix
1347	 * instead.
1348	 */
1349	dtp->dt_exec = dt_module_lookup_by_name(dtp, "genunix");
1350	dtp->dt_rtld = dt_module_lookup_by_name(dtp, "krtld");
1351	if (dtp->dt_rtld == NULL)
1352		dtp->dt_rtld = dt_module_lookup_by_name(dtp, "unix");
1353
1354	/*
1355	 * If this is the first time we are initializing the module list,
1356	 * remove the module for genunix from the module list and then move it
1357	 * to the front of the module list.  We do this so that type and symbol
1358	 * queries encounter genunix and thereby optimize for the common case
1359	 * in dtrace_lookup_by_name() and dtrace_lookup_by_type(), below.
1360	 */
1361	if (dtp->dt_exec != NULL &&
1362	    dtp->dt_cdefs == NULL && dtp->dt_ddefs == NULL) {
1363		dt_list_delete(&dtp->dt_modlist, dtp->dt_exec);
1364		dt_list_prepend(&dtp->dt_modlist, dtp->dt_exec);
1365	}
1366}
1367
1368static dt_module_t *
1369dt_module_from_object(dtrace_hdl_t *dtp, const char *object)
1370{
1371	int err = EDT_NOMOD;
1372	dt_module_t *dmp;
1373
1374	switch ((uintptr_t)object) {
1375	case (uintptr_t)DTRACE_OBJ_EXEC:
1376		dmp = dtp->dt_exec;
1377		break;
1378	case (uintptr_t)DTRACE_OBJ_RTLD:
1379		dmp = dtp->dt_rtld;
1380		break;
1381	case (uintptr_t)DTRACE_OBJ_CDEFS:
1382		dmp = dtp->dt_cdefs;
1383		break;
1384	case (uintptr_t)DTRACE_OBJ_DDEFS:
1385		dmp = dtp->dt_ddefs;
1386		break;
1387	default:
1388		dmp = dt_module_create(dtp, object);
1389		err = EDT_NOMEM;
1390	}
1391
1392	if (dmp == NULL)
1393		(void) dt_set_errno(dtp, err);
1394
1395	return (dmp);
1396}
1397
1398/*
1399 * Exported interface to look up a symbol by name.  We return the GElf_Sym and
1400 * complete symbol information for the matching symbol.
1401 */
1402int
1403dtrace_lookup_by_name(dtrace_hdl_t *dtp, const char *object, const char *name,
1404    GElf_Sym *symp, dtrace_syminfo_t *sip)
1405{
1406	dt_module_t *dmp;
1407	dt_ident_t *idp;
1408	uint_t n, id;
1409	GElf_Sym sym;
1410
1411	uint_t mask = 0; /* mask of dt_module flags to match */
1412	uint_t bits = 0; /* flag bits that must be present */
1413
1414	if (object != DTRACE_OBJ_EVERY &&
1415	    object != DTRACE_OBJ_KMODS &&
1416	    object != DTRACE_OBJ_UMODS) {
1417		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1418			return (-1); /* dt_errno is set for us */
1419
1420		if (dt_module_load(dtp, dmp) == -1)
1421			return (-1); /* dt_errno is set for us */
1422		n = 1;
1423
1424	} else {
1425		if (object == DTRACE_OBJ_KMODS)
1426			mask = bits = DT_DM_KERNEL;
1427		else if (object == DTRACE_OBJ_UMODS)
1428			mask = DT_DM_KERNEL;
1429
1430		dmp = dt_list_next(&dtp->dt_modlist);
1431		n = dtp->dt_nmods;
1432	}
1433
1434	if (symp == NULL)
1435		symp = &sym;
1436
1437	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1438		if ((dmp->dm_flags & mask) != bits)
1439			continue; /* failed to match required attributes */
1440
1441		if (dt_module_load(dtp, dmp) == -1)
1442			continue; /* failed to load symbol table */
1443
1444		if (dmp->dm_ops->do_symname(dmp, name, symp, &id) != NULL) {
1445			if (sip != NULL) {
1446				sip->dts_object = dmp->dm_name;
1447				sip->dts_name = (const char *)
1448				    dmp->dm_strtab.cts_data + symp->st_name;
1449				sip->dts_id = id;
1450			}
1451			return (0);
1452		}
1453
1454		if (dmp->dm_extern != NULL &&
1455		    (idp = dt_idhash_lookup(dmp->dm_extern, name)) != NULL) {
1456			if (symp != &sym) {
1457				symp->st_name = (uintptr_t)idp->di_name;
1458				symp->st_info =
1459				    GELF_ST_INFO(STB_GLOBAL, STT_NOTYPE);
1460				symp->st_other = 0;
1461				symp->st_shndx = SHN_UNDEF;
1462				symp->st_value = 0;
1463				symp->st_size =
1464				    ctf_type_size(idp->di_ctfp, idp->di_type);
1465			}
1466
1467			if (sip != NULL) {
1468				sip->dts_object = dmp->dm_name;
1469				sip->dts_name = idp->di_name;
1470				sip->dts_id = idp->di_id;
1471			}
1472
1473			return (0);
1474		}
1475	}
1476
1477	return (dt_set_errno(dtp, EDT_NOSYM));
1478}
1479
1480/*
1481 * Exported interface to look up a symbol by address.  We return the GElf_Sym
1482 * and complete symbol information for the matching symbol.
1483 */
1484int
1485dtrace_lookup_by_addr(dtrace_hdl_t *dtp, GElf_Addr addr,
1486    GElf_Sym *symp, dtrace_syminfo_t *sip)
1487{
1488	dt_module_t *dmp;
1489	uint_t id;
1490	const dtrace_vector_t *v = dtp->dt_vector;
1491
1492	if (v != NULL)
1493		return (v->dtv_lookup_by_addr(dtp->dt_varg, addr, symp, sip));
1494
1495	for (dmp = dt_list_next(&dtp->dt_modlist); dmp != NULL;
1496	    dmp = dt_list_next(dmp)) {
1497		if (addr - dmp->dm_text_va < dmp->dm_text_size ||
1498		    addr - dmp->dm_data_va < dmp->dm_data_size ||
1499		    addr - dmp->dm_bss_va < dmp->dm_bss_size)
1500			break;
1501	}
1502
1503	if (dmp == NULL)
1504		return (dt_set_errno(dtp, EDT_NOSYMADDR));
1505
1506	if (dt_module_load(dtp, dmp) == -1)
1507		return (-1); /* dt_errno is set for us */
1508
1509	if (symp != NULL) {
1510		if (dmp->dm_ops->do_symaddr(dmp, addr, symp, &id) == NULL)
1511			return (dt_set_errno(dtp, EDT_NOSYMADDR));
1512	}
1513
1514	if (sip != NULL) {
1515		sip->dts_object = dmp->dm_name;
1516
1517		if (symp != NULL) {
1518			sip->dts_name = (const char *)
1519			    dmp->dm_strtab.cts_data + symp->st_name;
1520			sip->dts_id = id;
1521		} else {
1522			sip->dts_name = NULL;
1523			sip->dts_id = 0;
1524		}
1525	}
1526
1527	return (0);
1528}
1529
1530int
1531dtrace_lookup_by_type(dtrace_hdl_t *dtp, const char *object, const char *name,
1532    dtrace_typeinfo_t *tip)
1533{
1534	dtrace_typeinfo_t ti;
1535	dt_module_t *dmp;
1536	int found = 0;
1537	ctf_id_t id;
1538	uint_t n, i;
1539	int justone;
1540	ctf_file_t *fp;
1541	char *buf, *p, *q;
1542
1543	uint_t mask = 0; /* mask of dt_module flags to match */
1544	uint_t bits = 0; /* flag bits that must be present */
1545
1546	if (object != DTRACE_OBJ_EVERY &&
1547	    object != DTRACE_OBJ_KMODS &&
1548	    object != DTRACE_OBJ_UMODS) {
1549		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1550			return (-1); /* dt_errno is set for us */
1551
1552		if (dt_module_load(dtp, dmp) == -1)
1553			return (-1); /* dt_errno is set for us */
1554		n = 1;
1555		justone = 1;
1556	} else {
1557		if (object == DTRACE_OBJ_KMODS)
1558			mask = bits = DT_DM_KERNEL;
1559		else if (object == DTRACE_OBJ_UMODS)
1560			mask = DT_DM_KERNEL;
1561
1562		dmp = dt_list_next(&dtp->dt_modlist);
1563		n = dtp->dt_nmods;
1564		justone = 0;
1565	}
1566
1567	if (tip == NULL)
1568		tip = &ti;
1569
1570	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1571		if ((dmp->dm_flags & mask) != bits)
1572			continue; /* failed to match required attributes */
1573
1574		/*
1575		 * If we can't load the CTF container, continue on to the next
1576		 * module.  If our search was scoped to only one module then
1577		 * return immediately leaving dt_errno unmodified.
1578		 */
1579		if (dt_module_hasctf(dtp, dmp) == 0) {
1580			if (justone)
1581				return (-1);
1582			continue;
1583		}
1584
1585		/*
1586		 * Look up the type in the module's CTF container.  If our
1587		 * match is a forward declaration tag, save this choice in
1588		 * 'tip' and keep going in the hope that we will locate the
1589		 * underlying structure definition.  Otherwise just return.
1590		 */
1591		if (dmp->dm_pid == 0) {
1592			id = ctf_lookup_by_name(dmp->dm_ctfp, name);
1593			fp = dmp->dm_ctfp;
1594		} else {
1595			if ((p = strchr(name, '`')) != NULL) {
1596				buf = strdup(name);
1597				if (buf == NULL)
1598					return (dt_set_errno(dtp, EDT_NOMEM));
1599				p = strchr(buf, '`');
1600				if ((q = strchr(p + 1, '`')) != NULL)
1601					p = q;
1602				*p = '\0';
1603				fp = dt_module_getctflib(dtp, dmp, buf);
1604				if (fp == NULL || (id = ctf_lookup_by_name(fp,
1605				    p + 1)) == CTF_ERR)
1606					id = CTF_ERR;
1607				free(buf);
1608			} else {
1609				for (i = 0; i < dmp->dm_nctflibs; i++) {
1610					fp = dmp->dm_libctfp[i];
1611					id = ctf_lookup_by_name(fp, name);
1612					if (id != CTF_ERR)
1613						break;
1614				}
1615			}
1616		}
1617		if (id != CTF_ERR) {
1618			tip->dtt_object = dmp->dm_name;
1619			tip->dtt_ctfp = fp;
1620			tip->dtt_type = id;
1621			if (ctf_type_kind(fp, ctf_type_resolve(fp, id)) !=
1622			    CTF_K_FORWARD)
1623				return (0);
1624
1625			found++;
1626		}
1627	}
1628
1629	if (found == 0)
1630		return (dt_set_errno(dtp, EDT_NOTYPE));
1631
1632	return (0);
1633}
1634
1635int
1636dtrace_symbol_type(dtrace_hdl_t *dtp, const GElf_Sym *symp,
1637    const dtrace_syminfo_t *sip, dtrace_typeinfo_t *tip)
1638{
1639	dt_module_t *dmp;
1640
1641	tip->dtt_object = NULL;
1642	tip->dtt_ctfp = NULL;
1643	tip->dtt_type = CTF_ERR;
1644	tip->dtt_flags = 0;
1645
1646	if ((dmp = dt_module_lookup_by_name(dtp, sip->dts_object)) == NULL)
1647		return (dt_set_errno(dtp, EDT_NOMOD));
1648
1649	if (symp->st_shndx == SHN_UNDEF && dmp->dm_extern != NULL) {
1650		dt_ident_t *idp =
1651		    dt_idhash_lookup(dmp->dm_extern, sip->dts_name);
1652
1653		if (idp == NULL)
1654			return (dt_set_errno(dtp, EDT_NOSYM));
1655
1656		tip->dtt_ctfp = idp->di_ctfp;
1657		tip->dtt_type = idp->di_type;
1658
1659	} else if (GELF_ST_TYPE(symp->st_info) != STT_FUNC) {
1660		if (dt_module_getctf(dtp, dmp) == NULL)
1661			return (-1); /* errno is set for us */
1662
1663		tip->dtt_ctfp = dmp->dm_ctfp;
1664		tip->dtt_type = ctf_lookup_by_symbol(dmp->dm_ctfp, sip->dts_id);
1665
1666		if (tip->dtt_type == CTF_ERR) {
1667			dtp->dt_ctferr = ctf_errno(tip->dtt_ctfp);
1668			return (dt_set_errno(dtp, EDT_CTF));
1669		}
1670
1671	} else {
1672		tip->dtt_ctfp = DT_FPTR_CTFP(dtp);
1673		tip->dtt_type = DT_FPTR_TYPE(dtp);
1674	}
1675
1676	tip->dtt_object = dmp->dm_name;
1677	return (0);
1678}
1679
1680static dtrace_objinfo_t *
1681dt_module_info(const dt_module_t *dmp, dtrace_objinfo_t *dto)
1682{
1683	dto->dto_name = dmp->dm_name;
1684	dto->dto_file = dmp->dm_file;
1685	dto->dto_id = dmp->dm_modid;
1686	dto->dto_flags = 0;
1687
1688	if (dmp->dm_flags & DT_DM_KERNEL)
1689		dto->dto_flags |= DTRACE_OBJ_F_KERNEL;
1690	if (dmp->dm_flags & DT_DM_PRIMARY)
1691		dto->dto_flags |= DTRACE_OBJ_F_PRIMARY;
1692
1693	dto->dto_text_va = dmp->dm_text_va;
1694	dto->dto_text_size = dmp->dm_text_size;
1695	dto->dto_data_va = dmp->dm_data_va;
1696	dto->dto_data_size = dmp->dm_data_size;
1697	dto->dto_bss_va = dmp->dm_bss_va;
1698	dto->dto_bss_size = dmp->dm_bss_size;
1699
1700	return (dto);
1701}
1702
1703int
1704dtrace_object_iter(dtrace_hdl_t *dtp, dtrace_obj_f *func, void *data)
1705{
1706	const dt_module_t *dmp = dt_list_next(&dtp->dt_modlist);
1707	dtrace_objinfo_t dto;
1708	int rv;
1709
1710	for (; dmp != NULL; dmp = dt_list_next(dmp)) {
1711		if ((rv = (*func)(dtp, dt_module_info(dmp, &dto), data)) != 0)
1712			return (rv);
1713	}
1714
1715	return (0);
1716}
1717
1718int
1719dtrace_object_info(dtrace_hdl_t *dtp, const char *object, dtrace_objinfo_t *dto)
1720{
1721	dt_module_t *dmp;
1722
1723	if (object == DTRACE_OBJ_EVERY || object == DTRACE_OBJ_KMODS ||
1724	    object == DTRACE_OBJ_UMODS || dto == NULL)
1725		return (dt_set_errno(dtp, EINVAL));
1726
1727	if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1728		return (-1); /* dt_errno is set for us */
1729
1730	if (dt_module_load(dtp, dmp) == -1)
1731		return (-1); /* dt_errno is set for us */
1732
1733	(void) dt_module_info(dmp, dto);
1734	return (0);
1735}
1736