dt_module.c revision 296816
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 * Copyright (c) 2016, Pedro Giffuni.  All rights reserved.
28 */
29
30#include <sys/types.h>
31#ifdef illumos
32#include <sys/modctl.h>
33#include <sys/kobj.h>
34#include <sys/kobj_impl.h>
35#include <sys/sysmacros.h>
36#include <sys/elf.h>
37#include <sys/task.h>
38#else
39#include <sys/param.h>
40#include <sys/linker.h>
41#include <sys/module.h>
42#include <sys/stat.h>
43#endif
44
45#include <unistd.h>
46#ifdef illumos
47#include <project.h>
48#endif
49#include <strings.h>
50#include <stdlib.h>
51#include <libelf.h>
52#include <limits.h>
53#include <assert.h>
54#include <errno.h>
55#include <dirent.h>
56#ifndef illumos
57#include <fcntl.h>
58#include <libproc_compat.h>
59#endif
60
61#include <dt_strtab.h>
62#include <dt_module.h>
63#include <dt_impl.h>
64
65static const char *dt_module_strtab; /* active strtab for qsort callbacks */
66
67static void
68dt_module_symhash_insert(dt_module_t *dmp, const char *name, uint_t id)
69{
70	dt_sym_t *dsp = &dmp->dm_symchains[dmp->dm_symfree];
71	uint_t h;
72
73	assert(dmp->dm_symfree < dmp->dm_nsymelems + 1);
74
75	dsp->ds_symid = id;
76	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
77	dsp->ds_next = dmp->dm_symbuckets[h];
78	dmp->dm_symbuckets[h] = dmp->dm_symfree++;
79}
80
81static uint_t
82dt_module_syminit32(dt_module_t *dmp)
83{
84#if STT_NUM != (STT_TLS + 1)
85#error "STT_NUM has grown. update dt_module_syminit32()"
86#endif
87
88	Elf32_Sym *sym = dmp->dm_symtab.cts_data;
89	const char *base = dmp->dm_strtab.cts_data;
90	size_t ss_size = dmp->dm_strtab.cts_size;
91	uint_t i, n = dmp->dm_nsymelems;
92	uint_t asrsv = 0;
93
94#if defined(__FreeBSD__)
95	GElf_Ehdr ehdr;
96	int is_elf_obj;
97
98	gelf_getehdr(dmp->dm_elf, &ehdr);
99	is_elf_obj = (ehdr.e_type == ET_REL);
100#endif
101
102	for (i = 0; i < n; i++, sym++) {
103		const char *name = base + sym->st_name;
104		uchar_t type = ELF32_ST_TYPE(sym->st_info);
105
106		if (type >= STT_NUM || type == STT_SECTION)
107			continue; /* skip sections and unknown types */
108
109		if (sym->st_name == 0 || sym->st_name >= ss_size)
110			continue; /* skip null or invalid names */
111
112		if (sym->st_value != 0 &&
113		    (ELF32_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size)) {
114			asrsv++; /* reserve space in the address map */
115
116#if defined(__FreeBSD__)
117			sym->st_value += (Elf_Addr) dmp->dm_reloc_offset;
118			if (is_elf_obj && sym->st_shndx != SHN_UNDEF &&
119			    sym->st_shndx < ehdr.e_shnum)
120				sym->st_value +=
121				    dmp->dm_sec_offsets[sym->st_shndx];
122#endif
123		}
124
125		dt_module_symhash_insert(dmp, name, i);
126	}
127
128	return (asrsv);
129}
130
131static uint_t
132dt_module_syminit64(dt_module_t *dmp)
133{
134#if STT_NUM != (STT_TLS + 1)
135#error "STT_NUM has grown. update dt_module_syminit64()"
136#endif
137
138	Elf64_Sym *sym = dmp->dm_symtab.cts_data;
139	const char *base = dmp->dm_strtab.cts_data;
140	size_t ss_size = dmp->dm_strtab.cts_size;
141	uint_t i, n = dmp->dm_nsymelems;
142	uint_t asrsv = 0;
143
144#if defined(__FreeBSD__)
145	GElf_Ehdr ehdr;
146	int is_elf_obj;
147
148	gelf_getehdr(dmp->dm_elf, &ehdr);
149	is_elf_obj = (ehdr.e_type == ET_REL);
150#endif
151
152	for (i = 0; i < n; i++, sym++) {
153		const char *name = base + sym->st_name;
154		uchar_t type = ELF64_ST_TYPE(sym->st_info);
155
156		if (type >= STT_NUM || type == STT_SECTION)
157			continue; /* skip sections and unknown types */
158
159		if (sym->st_name == 0 || sym->st_name >= ss_size)
160			continue; /* skip null or invalid names */
161
162		if (sym->st_value != 0 &&
163		    (ELF64_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size)) {
164			asrsv++; /* reserve space in the address map */
165#if defined(__FreeBSD__)
166			sym->st_value += (Elf_Addr) dmp->dm_reloc_offset;
167			if (is_elf_obj && sym->st_shndx != SHN_UNDEF &&
168			    sym->st_shndx < ehdr.e_shnum)
169				sym->st_value +=
170				    dmp->dm_sec_offsets[sym->st_shndx];
171#endif
172		}
173
174		dt_module_symhash_insert(dmp, name, i);
175	}
176
177	return (asrsv);
178}
179
180/*
181 * Sort comparison function for 32-bit symbol address-to-name lookups.  We sort
182 * symbols by value.  If values are equal, we prefer the symbol that is
183 * non-zero sized, typed, not weak, or lexically first, in that order.
184 */
185static int
186dt_module_symcomp32(const void *lp, const void *rp)
187{
188	Elf32_Sym *lhs = *((Elf32_Sym **)lp);
189	Elf32_Sym *rhs = *((Elf32_Sym **)rp);
190
191	if (lhs->st_value != rhs->st_value)
192		return (lhs->st_value > rhs->st_value ? 1 : -1);
193
194	if ((lhs->st_size == 0) != (rhs->st_size == 0))
195		return (lhs->st_size == 0 ? 1 : -1);
196
197	if ((ELF32_ST_TYPE(lhs->st_info) == STT_NOTYPE) !=
198	    (ELF32_ST_TYPE(rhs->st_info) == STT_NOTYPE))
199		return (ELF32_ST_TYPE(lhs->st_info) == STT_NOTYPE ? 1 : -1);
200
201	if ((ELF32_ST_BIND(lhs->st_info) == STB_WEAK) !=
202	    (ELF32_ST_BIND(rhs->st_info) == STB_WEAK))
203		return (ELF32_ST_BIND(lhs->st_info) == STB_WEAK ? 1 : -1);
204
205	return (strcmp(dt_module_strtab + lhs->st_name,
206	    dt_module_strtab + rhs->st_name));
207}
208
209/*
210 * Sort comparison function for 64-bit symbol address-to-name lookups.  We sort
211 * symbols by value.  If values are equal, we prefer the symbol that is
212 * non-zero sized, typed, not weak, or lexically first, in that order.
213 */
214static int
215dt_module_symcomp64(const void *lp, const void *rp)
216{
217	Elf64_Sym *lhs = *((Elf64_Sym **)lp);
218	Elf64_Sym *rhs = *((Elf64_Sym **)rp);
219
220	if (lhs->st_value != rhs->st_value)
221		return (lhs->st_value > rhs->st_value ? 1 : -1);
222
223	if ((lhs->st_size == 0) != (rhs->st_size == 0))
224		return (lhs->st_size == 0 ? 1 : -1);
225
226	if ((ELF64_ST_TYPE(lhs->st_info) == STT_NOTYPE) !=
227	    (ELF64_ST_TYPE(rhs->st_info) == STT_NOTYPE))
228		return (ELF64_ST_TYPE(lhs->st_info) == STT_NOTYPE ? 1 : -1);
229
230	if ((ELF64_ST_BIND(lhs->st_info) == STB_WEAK) !=
231	    (ELF64_ST_BIND(rhs->st_info) == STB_WEAK))
232		return (ELF64_ST_BIND(lhs->st_info) == STB_WEAK ? 1 : -1);
233
234	return (strcmp(dt_module_strtab + lhs->st_name,
235	    dt_module_strtab + rhs->st_name));
236}
237
238static void
239dt_module_symsort32(dt_module_t *dmp)
240{
241	Elf32_Sym *symtab = (Elf32_Sym *)dmp->dm_symtab.cts_data;
242	Elf32_Sym **sympp = (Elf32_Sym **)dmp->dm_asmap;
243	const dt_sym_t *dsp = dmp->dm_symchains + 1;
244	uint_t i, n = dmp->dm_symfree;
245
246	for (i = 1; i < n; i++, dsp++) {
247		Elf32_Sym *sym = symtab + dsp->ds_symid;
248		if (sym->st_value != 0 &&
249		    (ELF32_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size))
250			*sympp++ = sym;
251	}
252
253	dmp->dm_aslen = (uint_t)(sympp - (Elf32_Sym **)dmp->dm_asmap);
254	assert(dmp->dm_aslen <= dmp->dm_asrsv);
255
256	dt_module_strtab = dmp->dm_strtab.cts_data;
257	qsort(dmp->dm_asmap, dmp->dm_aslen,
258	    sizeof (Elf32_Sym *), dt_module_symcomp32);
259	dt_module_strtab = NULL;
260}
261
262static void
263dt_module_symsort64(dt_module_t *dmp)
264{
265	Elf64_Sym *symtab = (Elf64_Sym *)dmp->dm_symtab.cts_data;
266	Elf64_Sym **sympp = (Elf64_Sym **)dmp->dm_asmap;
267	const dt_sym_t *dsp = dmp->dm_symchains + 1;
268	uint_t i, n = dmp->dm_symfree;
269
270	for (i = 1; i < n; i++, dsp++) {
271		Elf64_Sym *sym = symtab + dsp->ds_symid;
272		if (sym->st_value != 0 &&
273		    (ELF64_ST_BIND(sym->st_info) != STB_LOCAL || sym->st_size))
274			*sympp++ = sym;
275	}
276
277	dmp->dm_aslen = (uint_t)(sympp - (Elf64_Sym **)dmp->dm_asmap);
278	assert(dmp->dm_aslen <= dmp->dm_asrsv);
279
280	dt_module_strtab = dmp->dm_strtab.cts_data;
281	qsort(dmp->dm_asmap, dmp->dm_aslen,
282	    sizeof (Elf64_Sym *), dt_module_symcomp64);
283	dt_module_strtab = NULL;
284}
285
286static GElf_Sym *
287dt_module_symgelf32(const Elf32_Sym *src, GElf_Sym *dst)
288{
289	if (dst != NULL) {
290		dst->st_name = src->st_name;
291		dst->st_info = src->st_info;
292		dst->st_other = src->st_other;
293		dst->st_shndx = src->st_shndx;
294		dst->st_value = src->st_value;
295		dst->st_size = src->st_size;
296	}
297
298	return (dst);
299}
300
301static GElf_Sym *
302dt_module_symgelf64(const Elf64_Sym *src, GElf_Sym *dst)
303{
304	if (dst != NULL)
305		bcopy(src, dst, sizeof (GElf_Sym));
306
307	return (dst);
308}
309
310static GElf_Sym *
311dt_module_symname32(dt_module_t *dmp, const char *name,
312    GElf_Sym *symp, uint_t *idp)
313{
314	const Elf32_Sym *symtab = dmp->dm_symtab.cts_data;
315	const char *strtab = dmp->dm_strtab.cts_data;
316
317	const Elf32_Sym *sym;
318	const dt_sym_t *dsp;
319	uint_t i, h;
320
321	if (dmp->dm_nsymelems == 0)
322		return (NULL);
323
324	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
325
326	for (i = dmp->dm_symbuckets[h]; i != 0; i = dsp->ds_next) {
327		dsp = &dmp->dm_symchains[i];
328		sym = symtab + dsp->ds_symid;
329
330		if (strcmp(name, strtab + sym->st_name) == 0) {
331			if (idp != NULL)
332				*idp = dsp->ds_symid;
333			return (dt_module_symgelf32(sym, symp));
334		}
335	}
336
337	return (NULL);
338}
339
340static GElf_Sym *
341dt_module_symname64(dt_module_t *dmp, const char *name,
342    GElf_Sym *symp, uint_t *idp)
343{
344	const Elf64_Sym *symtab = dmp->dm_symtab.cts_data;
345	const char *strtab = dmp->dm_strtab.cts_data;
346
347	const Elf64_Sym *sym;
348	const dt_sym_t *dsp;
349	uint_t i, h;
350
351	if (dmp->dm_nsymelems == 0)
352		return (NULL);
353
354	h = dt_strtab_hash(name, NULL) % dmp->dm_nsymbuckets;
355
356	for (i = dmp->dm_symbuckets[h]; i != 0; i = dsp->ds_next) {
357		dsp = &dmp->dm_symchains[i];
358		sym = symtab + dsp->ds_symid;
359
360		if (strcmp(name, strtab + sym->st_name) == 0) {
361			if (idp != NULL)
362				*idp = dsp->ds_symid;
363			return (dt_module_symgelf64(sym, symp));
364		}
365	}
366
367	return (NULL);
368}
369
370static GElf_Sym *
371dt_module_symaddr32(dt_module_t *dmp, GElf_Addr addr,
372    GElf_Sym *symp, uint_t *idp)
373{
374	const Elf32_Sym **asmap = (const Elf32_Sym **)dmp->dm_asmap;
375	const Elf32_Sym *symtab = dmp->dm_symtab.cts_data;
376	const Elf32_Sym *sym;
377
378	uint_t i, mid, lo = 0, hi = dmp->dm_aslen - 1;
379	Elf32_Addr v;
380
381	if (dmp->dm_aslen == 0)
382		return (NULL);
383
384	while (hi - lo > 1) {
385		mid = (lo + hi) / 2;
386		if (addr >= asmap[mid]->st_value)
387			lo = mid;
388		else
389			hi = mid;
390	}
391
392	i = addr < asmap[hi]->st_value ? lo : hi;
393	sym = asmap[i];
394	v = sym->st_value;
395
396	/*
397	 * If the previous entry has the same value, improve our choice.  The
398	 * order of equal-valued symbols is determined by the comparison func.
399	 */
400	while (i-- != 0 && asmap[i]->st_value == v)
401		sym = asmap[i];
402
403	if (addr - sym->st_value < MAX(sym->st_size, 1)) {
404		if (idp != NULL)
405			*idp = (uint_t)(sym - symtab);
406		return (dt_module_symgelf32(sym, symp));
407	}
408
409	return (NULL);
410}
411
412static GElf_Sym *
413dt_module_symaddr64(dt_module_t *dmp, GElf_Addr addr,
414    GElf_Sym *symp, uint_t *idp)
415{
416	const Elf64_Sym **asmap = (const Elf64_Sym **)dmp->dm_asmap;
417	const Elf64_Sym *symtab = dmp->dm_symtab.cts_data;
418	const Elf64_Sym *sym;
419
420	uint_t i, mid, lo = 0, hi = dmp->dm_aslen - 1;
421	Elf64_Addr v;
422
423	if (dmp->dm_aslen == 0)
424		return (NULL);
425
426	while (hi - lo > 1) {
427		mid = (lo + hi) / 2;
428		if (addr >= asmap[mid]->st_value)
429			lo = mid;
430		else
431			hi = mid;
432	}
433
434	i = addr < asmap[hi]->st_value ? lo : hi;
435	sym = asmap[i];
436	v = sym->st_value;
437
438	/*
439	 * If the previous entry has the same value, improve our choice.  The
440	 * order of equal-valued symbols is determined by the comparison func.
441	 */
442	while (i-- != 0 && asmap[i]->st_value == v)
443		sym = asmap[i];
444
445	if (addr - sym->st_value < MAX(sym->st_size, 1)) {
446		if (idp != NULL)
447			*idp = (uint_t)(sym - symtab);
448		return (dt_module_symgelf64(sym, symp));
449	}
450
451	return (NULL);
452}
453
454static const dt_modops_t dt_modops_32 = {
455	dt_module_syminit32,
456	dt_module_symsort32,
457	dt_module_symname32,
458	dt_module_symaddr32
459};
460
461static const dt_modops_t dt_modops_64 = {
462	dt_module_syminit64,
463	dt_module_symsort64,
464	dt_module_symname64,
465	dt_module_symaddr64
466};
467
468dt_module_t *
469dt_module_create(dtrace_hdl_t *dtp, const char *name)
470{
471	long pid;
472	char *eptr;
473	dt_ident_t *idp;
474	uint_t h = dt_strtab_hash(name, NULL) % dtp->dt_modbuckets;
475	dt_module_t *dmp;
476
477	for (dmp = dtp->dt_mods[h]; dmp != NULL; dmp = dmp->dm_next) {
478		if (strcmp(dmp->dm_name, name) == 0)
479			return (dmp);
480	}
481
482	if ((dmp = malloc(sizeof (dt_module_t))) == NULL)
483		return (NULL); /* caller must handle allocation failure */
484
485	bzero(dmp, sizeof (dt_module_t));
486	(void) strlcpy(dmp->dm_name, name, sizeof (dmp->dm_name));
487	dt_list_append(&dtp->dt_modlist, dmp);
488	dmp->dm_next = dtp->dt_mods[h];
489	dtp->dt_mods[h] = dmp;
490	dtp->dt_nmods++;
491
492	if (dtp->dt_conf.dtc_ctfmodel == CTF_MODEL_LP64)
493		dmp->dm_ops = &dt_modops_64;
494	else
495		dmp->dm_ops = &dt_modops_32;
496
497	/*
498	 * Modules for userland processes are special. They always refer to a
499	 * specific process and have a copy of their CTF data from a specific
500	 * instant in time. Any dt_module_t that begins with 'pid' is a module
501	 * for a specific process, much like how any probe description that
502	 * begins with 'pid' is special. pid123 refers to process 123. A module
503	 * that is just 'pid' refers specifically to pid$target. This is
504	 * generally done as D does not currently allow for macros to be
505	 * evaluated when working with types.
506	 */
507	if (strncmp(dmp->dm_name, "pid", 3) == 0) {
508		errno = 0;
509		if (dmp->dm_name[3] == '\0') {
510			idp = dt_idhash_lookup(dtp->dt_macros, "target");
511			if (idp != NULL && idp->di_id != 0)
512				dmp->dm_pid = idp->di_id;
513		} else {
514			pid = strtol(dmp->dm_name + 3, &eptr, 10);
515			if (errno == 0 && *eptr == '\0')
516				dmp->dm_pid = (pid_t)pid;
517			else
518				dt_dprintf("encountered malformed pid "
519				    "module: %s\n", dmp->dm_name);
520		}
521	}
522
523	return (dmp);
524}
525
526dt_module_t *
527dt_module_lookup_by_name(dtrace_hdl_t *dtp, const char *name)
528{
529	uint_t h = dt_strtab_hash(name, NULL) % dtp->dt_modbuckets;
530	dt_module_t *dmp;
531
532	for (dmp = dtp->dt_mods[h]; dmp != NULL; dmp = dmp->dm_next) {
533		if (strcmp(dmp->dm_name, name) == 0)
534			return (dmp);
535	}
536
537	return (NULL);
538}
539
540/*ARGSUSED*/
541dt_module_t *
542dt_module_lookup_by_ctf(dtrace_hdl_t *dtp, ctf_file_t *ctfp)
543{
544	return (ctfp ? ctf_getspecific(ctfp) : NULL);
545}
546
547#ifdef __FreeBSD__
548dt_kmodule_t *
549dt_kmodule_lookup(dtrace_hdl_t *dtp, const char *name)
550{
551	uint_t h = dt_strtab_hash(name, NULL) % dtp->dt_modbuckets;
552	dt_kmodule_t *dkmp;
553
554	for (dkmp = dtp->dt_kmods[h]; dkmp != NULL; dkmp = dkmp->dkm_next) {
555		if (strcmp(dkmp->dkm_name, name) == 0)
556			return (dkmp);
557	}
558
559	return (NULL);
560}
561#endif
562
563static int
564dt_module_load_sect(dtrace_hdl_t *dtp, dt_module_t *dmp, ctf_sect_t *ctsp)
565{
566	const char *s;
567	size_t shstrs;
568	GElf_Shdr sh;
569	Elf_Data *dp;
570	Elf_Scn *sp;
571
572	if (elf_getshdrstrndx(dmp->dm_elf, &shstrs) == -1)
573		return (dt_set_errno(dtp, EDT_NOTLOADED));
574
575	for (sp = NULL; (sp = elf_nextscn(dmp->dm_elf, sp)) != NULL; ) {
576		if (gelf_getshdr(sp, &sh) == NULL || sh.sh_type == SHT_NULL ||
577		    (s = elf_strptr(dmp->dm_elf, shstrs, sh.sh_name)) == NULL)
578			continue; /* skip any malformed sections */
579
580		if (sh.sh_type == ctsp->cts_type &&
581		    sh.sh_entsize == ctsp->cts_entsize &&
582		    strcmp(s, ctsp->cts_name) == 0)
583			break; /* section matches specification */
584	}
585
586	/*
587	 * If the section isn't found, return success but leave cts_data set
588	 * to NULL and cts_size set to zero for our caller.
589	 */
590	if (sp == NULL || (dp = elf_getdata(sp, NULL)) == NULL)
591		return (0);
592
593#ifdef illumos
594	ctsp->cts_data = dp->d_buf;
595#else
596	if ((ctsp->cts_data = malloc(dp->d_size)) == NULL)
597		return (0);
598	memcpy(ctsp->cts_data, dp->d_buf, dp->d_size);
599#endif
600	ctsp->cts_size = dp->d_size;
601
602	dt_dprintf("loaded %s [%s] (%lu bytes)\n",
603	    dmp->dm_name, ctsp->cts_name, (ulong_t)ctsp->cts_size);
604
605	return (0);
606}
607
608typedef struct dt_module_cb_arg {
609	struct ps_prochandle *dpa_proc;
610	dtrace_hdl_t *dpa_dtp;
611	dt_module_t *dpa_dmp;
612	uint_t dpa_count;
613} dt_module_cb_arg_t;
614
615/* ARGSUSED */
616static int
617dt_module_load_proc_count(void *arg, const prmap_t *prmap, const char *obj)
618{
619	ctf_file_t *fp;
620	dt_module_cb_arg_t *dcp = arg;
621
622	/* Try to grab a ctf container if it exists */
623	fp = Pname_to_ctf(dcp->dpa_proc, obj);
624	if (fp != NULL)
625		dcp->dpa_count++;
626	return (0);
627}
628
629/* ARGSUSED */
630static int
631dt_module_load_proc_build(void *arg, const prmap_t *prmap, const char *obj)
632{
633	ctf_file_t *fp;
634	char buf[MAXPATHLEN], *p;
635	dt_module_cb_arg_t *dcp = arg;
636	int count = dcp->dpa_count;
637	Lmid_t lmid;
638
639	fp = Pname_to_ctf(dcp->dpa_proc, obj);
640	if (fp == NULL)
641		return (0);
642	fp = ctf_dup(fp);
643	if (fp == NULL)
644		return (0);
645	dcp->dpa_dmp->dm_libctfp[count] = fp;
646	/*
647	 * While it'd be nice to simply use objname here, because of our prior
648	 * actions we'll always get a resolved object name to its on disk file.
649	 * Like the pid provider, we need to tell a bit of a lie here. The type
650	 * that the user thinks of is in terms of the libraries they requested,
651	 * eg. libc.so.1, they don't care about the fact that it's
652	 * libc_hwcap.so.1.
653	 */
654	(void) Pobjname(dcp->dpa_proc, prmap->pr_vaddr, buf, sizeof (buf));
655	if ((p = strrchr(buf, '/')) == NULL)
656		p = buf;
657	else
658		p++;
659
660	/*
661	 * If for some reason we can't find a link map id for this module, which
662	 * would be really quite weird. We instead just say the link map id is
663	 * zero.
664	 */
665	if (Plmid(dcp->dpa_proc, prmap->pr_vaddr, &lmid) != 0)
666		lmid = 0;
667
668	if (lmid == 0)
669		dcp->dpa_dmp->dm_libctfn[count] = strdup(p);
670	else
671		(void) asprintf(&dcp->dpa_dmp->dm_libctfn[count],
672		    "LM%x`%s", lmid, p);
673	if (dcp->dpa_dmp->dm_libctfn[count] == NULL)
674		return (1);
675	ctf_setspecific(fp, dcp->dpa_dmp);
676	dcp->dpa_count++;
677	return (0);
678}
679
680/*
681 * We've been asked to load data that belongs to another process. As such we're
682 * going to pgrab it at this instant, load everything that we might ever care
683 * about, and then drive on. The reason for this is that the process that we're
684 * interested in might be changing. As long as we have grabbed it, then this
685 * can't be a problem for us.
686 *
687 * For now, we're actually going to punt on most things and just try to get CTF
688 * data, nothing else. Basically this is only useful as a source of type
689 * information, we can't go and do the stacktrace lookups, etc.
690 */
691static int
692dt_module_load_proc(dtrace_hdl_t *dtp, dt_module_t *dmp)
693{
694	struct ps_prochandle *p;
695	dt_module_cb_arg_t arg;
696
697	/*
698	 * Note that on success we do not release this hold. We must hold this
699	 * for our life time.
700	 */
701	p = dt_proc_grab(dtp, dmp->dm_pid, 0, PGRAB_RDONLY | PGRAB_FORCE);
702	if (p == NULL) {
703		dt_dprintf("failed to grab pid: %d\n", (int)dmp->dm_pid);
704		return (dt_set_errno(dtp, EDT_CANTLOAD));
705	}
706	dt_proc_lock(dtp, p);
707
708	arg.dpa_proc = p;
709	arg.dpa_dtp = dtp;
710	arg.dpa_dmp = dmp;
711	arg.dpa_count = 0;
712	if (Pobject_iter_resolved(p, dt_module_load_proc_count, &arg) != 0) {
713		dt_dprintf("failed to iterate objects\n");
714		dt_proc_release(dtp, p);
715		return (dt_set_errno(dtp, EDT_CANTLOAD));
716	}
717
718	if (arg.dpa_count == 0) {
719		dt_dprintf("no ctf data present\n");
720		dt_proc_unlock(dtp, p);
721		dt_proc_release(dtp, p);
722		return (dt_set_errno(dtp, EDT_CANTLOAD));
723	}
724
725	dmp->dm_libctfp = calloc(arg.dpa_count, sizeof (ctf_file_t *));
726	if (dmp->dm_libctfp == NULL) {
727		dt_proc_unlock(dtp, p);
728		dt_proc_release(dtp, p);
729		return (dt_set_errno(dtp, EDT_NOMEM));
730	}
731
732	dmp->dm_libctfn = calloc(arg.dpa_count, sizeof (char *));
733	if (dmp->dm_libctfn == NULL) {
734		free(dmp->dm_libctfp);
735		dt_proc_unlock(dtp, p);
736		dt_proc_release(dtp, p);
737		return (dt_set_errno(dtp, EDT_NOMEM));
738	}
739
740	dmp->dm_nctflibs = arg.dpa_count;
741
742	arg.dpa_count = 0;
743	if (Pobject_iter_resolved(p, dt_module_load_proc_build, &arg) != 0) {
744		dt_proc_unlock(dtp, p);
745		dt_module_unload(dtp, dmp);
746		dt_proc_release(dtp, p);
747		return (dt_set_errno(dtp, EDT_CANTLOAD));
748	}
749	assert(arg.dpa_count == dmp->dm_nctflibs);
750	dt_dprintf("loaded %d ctf modules for pid %d\n", arg.dpa_count,
751	    (int)dmp->dm_pid);
752
753	dt_proc_unlock(dtp, p);
754	dt_proc_release(dtp, p);
755	dmp->dm_flags |= DT_DM_LOADED;
756
757	return (0);
758}
759
760int
761dt_module_load(dtrace_hdl_t *dtp, dt_module_t *dmp)
762{
763	if (dmp->dm_flags & DT_DM_LOADED)
764		return (0); /* module is already loaded */
765
766	if (dmp->dm_pid != 0)
767		return (dt_module_load_proc(dtp, dmp));
768
769	dmp->dm_ctdata.cts_name = ".SUNW_ctf";
770	dmp->dm_ctdata.cts_type = SHT_PROGBITS;
771	dmp->dm_ctdata.cts_flags = 0;
772	dmp->dm_ctdata.cts_data = NULL;
773	dmp->dm_ctdata.cts_size = 0;
774	dmp->dm_ctdata.cts_entsize = 0;
775	dmp->dm_ctdata.cts_offset = 0;
776
777	dmp->dm_symtab.cts_name = ".symtab";
778	dmp->dm_symtab.cts_type = SHT_SYMTAB;
779	dmp->dm_symtab.cts_flags = 0;
780	dmp->dm_symtab.cts_data = NULL;
781	dmp->dm_symtab.cts_size = 0;
782	dmp->dm_symtab.cts_entsize = dmp->dm_ops == &dt_modops_64 ?
783	    sizeof (Elf64_Sym) : sizeof (Elf32_Sym);
784	dmp->dm_symtab.cts_offset = 0;
785
786	dmp->dm_strtab.cts_name = ".strtab";
787	dmp->dm_strtab.cts_type = SHT_STRTAB;
788	dmp->dm_strtab.cts_flags = 0;
789	dmp->dm_strtab.cts_data = NULL;
790	dmp->dm_strtab.cts_size = 0;
791	dmp->dm_strtab.cts_entsize = 0;
792	dmp->dm_strtab.cts_offset = 0;
793
794	/*
795	 * Attempt to load the module's CTF section, symbol table section, and
796	 * string table section.  Note that modules may not contain CTF data:
797	 * this will result in a successful load_sect but data of size zero.
798	 * We will then fail if dt_module_getctf() is called, as shown below.
799	 */
800	if (dt_module_load_sect(dtp, dmp, &dmp->dm_ctdata) == -1 ||
801	    dt_module_load_sect(dtp, dmp, &dmp->dm_symtab) == -1 ||
802	    dt_module_load_sect(dtp, dmp, &dmp->dm_strtab) == -1) {
803		dt_module_unload(dtp, dmp);
804		return (-1); /* dt_errno is set for us */
805	}
806
807	/*
808	 * Allocate the hash chains and hash buckets for symbol name lookup.
809	 * This is relatively simple since the symbol table is of fixed size
810	 * and is known in advance.  We allocate one extra element since we
811	 * use element indices instead of pointers and zero is our sentinel.
812	 */
813	dmp->dm_nsymelems =
814	    dmp->dm_symtab.cts_size / dmp->dm_symtab.cts_entsize;
815
816	dmp->dm_nsymbuckets = _dtrace_strbuckets;
817	dmp->dm_symfree = 1;		/* first free element is index 1 */
818
819	dmp->dm_symbuckets = calloc(dmp->dm_nsymbuckets, sizeof (uint_t));
820	dmp->dm_symchains = calloc(dmp->dm_nsymelems + 1, sizeof (dt_sym_t));
821
822	if (dmp->dm_symbuckets == NULL || dmp->dm_symchains == NULL) {
823		dt_module_unload(dtp, dmp);
824		return (dt_set_errno(dtp, EDT_NOMEM));
825	}
826
827	/*
828	 * Iterate over the symbol table data buffer and insert each symbol
829	 * name into the name hash if the name and type are valid.  Then
830	 * allocate the address map, fill it in, and sort it.
831	 */
832	dmp->dm_asrsv = dmp->dm_ops->do_syminit(dmp);
833
834	dt_dprintf("hashed %s [%s] (%u symbols)\n",
835	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_symfree - 1);
836
837	if ((dmp->dm_asmap = malloc(sizeof (void *) * dmp->dm_asrsv)) == NULL) {
838		dt_module_unload(dtp, dmp);
839		return (dt_set_errno(dtp, EDT_NOMEM));
840	}
841
842	dmp->dm_ops->do_symsort(dmp);
843
844	dt_dprintf("sorted %s [%s] (%u symbols)\n",
845	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_aslen);
846
847	dmp->dm_flags |= DT_DM_LOADED;
848	return (0);
849}
850
851int
852dt_module_hasctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
853{
854	if (dmp->dm_pid != 0 && dmp->dm_nctflibs > 0)
855		return (1);
856	return (dt_module_getctf(dtp, dmp) != NULL);
857}
858
859ctf_file_t *
860dt_module_getctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
861{
862	const char *parent;
863	dt_module_t *pmp;
864	ctf_file_t *pfp;
865	int model;
866
867	if (dmp->dm_ctfp != NULL || dt_module_load(dtp, dmp) != 0)
868		return (dmp->dm_ctfp);
869
870	if (dmp->dm_ops == &dt_modops_64)
871		model = CTF_MODEL_LP64;
872	else
873		model = CTF_MODEL_ILP32;
874
875	/*
876	 * If the data model of the module does not match our program data
877	 * model, then do not permit CTF from this module to be opened and
878	 * returned to the compiler.  If we support mixed data models in the
879	 * future for combined kernel/user tracing, this can be removed.
880	 */
881	if (dtp->dt_conf.dtc_ctfmodel != model) {
882		(void) dt_set_errno(dtp, EDT_DATAMODEL);
883		return (NULL);
884	}
885
886	if (dmp->dm_ctdata.cts_size == 0) {
887		(void) dt_set_errno(dtp, EDT_NOCTF);
888		return (NULL);
889	}
890
891	dmp->dm_ctfp = ctf_bufopen(&dmp->dm_ctdata,
892	    &dmp->dm_symtab, &dmp->dm_strtab, &dtp->dt_ctferr);
893
894	if (dmp->dm_ctfp == NULL) {
895		(void) dt_set_errno(dtp, EDT_CTF);
896		return (NULL);
897	}
898
899	(void) ctf_setmodel(dmp->dm_ctfp, model);
900	ctf_setspecific(dmp->dm_ctfp, dmp);
901
902	if ((parent = ctf_parent_name(dmp->dm_ctfp)) != NULL) {
903		if ((pmp = dt_module_create(dtp, parent)) == NULL ||
904		    (pfp = dt_module_getctf(dtp, pmp)) == NULL) {
905			if (pmp == NULL)
906				(void) dt_set_errno(dtp, EDT_NOMEM);
907			goto err;
908		}
909
910		if (ctf_import(dmp->dm_ctfp, pfp) == CTF_ERR) {
911			dtp->dt_ctferr = ctf_errno(dmp->dm_ctfp);
912			(void) dt_set_errno(dtp, EDT_CTF);
913			goto err;
914		}
915	}
916
917	dt_dprintf("loaded CTF container for %s (%p)\n",
918	    dmp->dm_name, (void *)dmp->dm_ctfp);
919
920	return (dmp->dm_ctfp);
921
922err:
923	ctf_close(dmp->dm_ctfp);
924	dmp->dm_ctfp = NULL;
925	return (NULL);
926}
927
928/*ARGSUSED*/
929void
930dt_module_unload(dtrace_hdl_t *dtp, dt_module_t *dmp)
931{
932	int i;
933
934	ctf_close(dmp->dm_ctfp);
935	dmp->dm_ctfp = NULL;
936
937#ifndef illumos
938	if (dmp->dm_ctdata.cts_data != NULL) {
939		free(dmp->dm_ctdata.cts_data);
940	}
941	if (dmp->dm_symtab.cts_data != NULL) {
942		free(dmp->dm_symtab.cts_data);
943	}
944	if (dmp->dm_strtab.cts_data != NULL) {
945		free(dmp->dm_strtab.cts_data);
946	}
947#endif
948
949	if (dmp->dm_libctfp != NULL) {
950		for (i = 0; i < dmp->dm_nctflibs; i++) {
951			ctf_close(dmp->dm_libctfp[i]);
952			free(dmp->dm_libctfn[i]);
953		}
954		free(dmp->dm_libctfp);
955		free(dmp->dm_libctfn);
956		dmp->dm_libctfp = NULL;
957		dmp->dm_nctflibs = 0;
958	}
959
960	bzero(&dmp->dm_ctdata, sizeof (ctf_sect_t));
961	bzero(&dmp->dm_symtab, sizeof (ctf_sect_t));
962	bzero(&dmp->dm_strtab, sizeof (ctf_sect_t));
963
964	if (dmp->dm_symbuckets != NULL) {
965		free(dmp->dm_symbuckets);
966		dmp->dm_symbuckets = NULL;
967	}
968
969	if (dmp->dm_symchains != NULL) {
970		free(dmp->dm_symchains);
971		dmp->dm_symchains = NULL;
972	}
973
974	if (dmp->dm_asmap != NULL) {
975		free(dmp->dm_asmap);
976		dmp->dm_asmap = NULL;
977	}
978#if defined(__FreeBSD__)
979	if (dmp->dm_sec_offsets != NULL) {
980		free(dmp->dm_sec_offsets);
981		dmp->dm_sec_offsets = NULL;
982	}
983#endif
984	dmp->dm_symfree = 0;
985	dmp->dm_nsymbuckets = 0;
986	dmp->dm_nsymelems = 0;
987	dmp->dm_asrsv = 0;
988	dmp->dm_aslen = 0;
989
990	dmp->dm_text_va = 0;
991	dmp->dm_text_size = 0;
992	dmp->dm_data_va = 0;
993	dmp->dm_data_size = 0;
994	dmp->dm_bss_va = 0;
995	dmp->dm_bss_size = 0;
996
997	if (dmp->dm_extern != NULL) {
998		dt_idhash_destroy(dmp->dm_extern);
999		dmp->dm_extern = NULL;
1000	}
1001
1002	(void) elf_end(dmp->dm_elf);
1003	dmp->dm_elf = NULL;
1004
1005	dmp->dm_pid = 0;
1006
1007	dmp->dm_flags &= ~DT_DM_LOADED;
1008}
1009
1010void
1011dt_module_destroy(dtrace_hdl_t *dtp, dt_module_t *dmp)
1012{
1013	uint_t h = dt_strtab_hash(dmp->dm_name, NULL) % dtp->dt_modbuckets;
1014	dt_module_t **dmpp = &dtp->dt_mods[h];
1015
1016	dt_list_delete(&dtp->dt_modlist, dmp);
1017	assert(dtp->dt_nmods != 0);
1018	dtp->dt_nmods--;
1019
1020	/*
1021	 * Now remove this module from its hash chain.  We expect to always
1022	 * find the module on its hash chain, so in this loop we assert that
1023	 * we don't run off the end of the list.
1024	 */
1025	while (*dmpp != dmp) {
1026		dmpp = &((*dmpp)->dm_next);
1027		assert(*dmpp != NULL);
1028	}
1029
1030	*dmpp = dmp->dm_next;
1031
1032	dt_module_unload(dtp, dmp);
1033	free(dmp);
1034}
1035
1036/*
1037 * Insert a new external symbol reference into the specified module.  The new
1038 * symbol will be marked as undefined and is assigned a symbol index beyond
1039 * any existing cached symbols from this module.  We use the ident's di_data
1040 * field to store a pointer to a copy of the dtrace_syminfo_t for this symbol.
1041 */
1042dt_ident_t *
1043dt_module_extern(dtrace_hdl_t *dtp, dt_module_t *dmp,
1044    const char *name, const dtrace_typeinfo_t *tip)
1045{
1046	dtrace_syminfo_t *sip;
1047	dt_ident_t *idp;
1048	uint_t id;
1049
1050	if (dmp->dm_extern == NULL && (dmp->dm_extern = dt_idhash_create(
1051	    "extern", NULL, dmp->dm_nsymelems, UINT_MAX)) == NULL) {
1052		(void) dt_set_errno(dtp, EDT_NOMEM);
1053		return (NULL);
1054	}
1055
1056	if (dt_idhash_nextid(dmp->dm_extern, &id) == -1) {
1057		(void) dt_set_errno(dtp, EDT_SYMOFLOW);
1058		return (NULL);
1059	}
1060
1061	if ((sip = malloc(sizeof (dtrace_syminfo_t))) == NULL) {
1062		(void) dt_set_errno(dtp, EDT_NOMEM);
1063		return (NULL);
1064	}
1065
1066	idp = dt_idhash_insert(dmp->dm_extern, name, DT_IDENT_SYMBOL, 0, id,
1067	    _dtrace_symattr, 0, &dt_idops_thaw, NULL, dtp->dt_gen);
1068
1069	if (idp == NULL) {
1070		(void) dt_set_errno(dtp, EDT_NOMEM);
1071		free(sip);
1072		return (NULL);
1073	}
1074
1075	sip->dts_object = dmp->dm_name;
1076	sip->dts_name = idp->di_name;
1077	sip->dts_id = idp->di_id;
1078
1079	idp->di_data = sip;
1080	idp->di_ctfp = tip->dtt_ctfp;
1081	idp->di_type = tip->dtt_type;
1082
1083	return (idp);
1084}
1085
1086const char *
1087dt_module_modelname(dt_module_t *dmp)
1088{
1089	if (dmp->dm_ops == &dt_modops_64)
1090		return ("64-bit");
1091	else
1092		return ("32-bit");
1093}
1094
1095/* ARGSUSED */
1096int
1097dt_module_getlibid(dtrace_hdl_t *dtp, dt_module_t *dmp, const ctf_file_t *fp)
1098{
1099	int i;
1100
1101	for (i = 0; i < dmp->dm_nctflibs; i++) {
1102		if (dmp->dm_libctfp[i] == fp)
1103			return (i);
1104	}
1105
1106	return (-1);
1107}
1108
1109/* ARGSUSED */
1110ctf_file_t *
1111dt_module_getctflib(dtrace_hdl_t *dtp, dt_module_t *dmp, const char *name)
1112{
1113	int i;
1114
1115	for (i = 0; i < dmp->dm_nctflibs; i++) {
1116		if (strcmp(dmp->dm_libctfn[i], name) == 0)
1117			return (dmp->dm_libctfp[i]);
1118	}
1119
1120	return (NULL);
1121}
1122
1123/*
1124 * Update our module cache by adding an entry for the specified module 'name'.
1125 * We create the dt_module_t and populate it using /system/object/<name>/.
1126 *
1127 * On FreeBSD, the module name is passed as the full module file name,
1128 * including the path.
1129 */
1130static void
1131#ifdef illumos
1132dt_module_update(dtrace_hdl_t *dtp, const char *name)
1133#else
1134dt_module_update(dtrace_hdl_t *dtp, struct kld_file_stat *k_stat)
1135#endif
1136{
1137	char fname[MAXPATHLEN];
1138	struct stat64 st;
1139	int fd, err, bits;
1140#ifdef __FreeBSD__
1141	struct module_stat ms;
1142	dt_kmodule_t *dkmp;
1143	uint_t h;
1144	int modid;
1145#endif
1146
1147	dt_module_t *dmp;
1148	const char *s;
1149	size_t shstrs;
1150	GElf_Shdr sh;
1151	Elf_Data *dp;
1152	Elf_Scn *sp;
1153
1154#ifdef illumos
1155	(void) snprintf(fname, sizeof (fname),
1156	    "%s/%s/object", OBJFS_ROOT, name);
1157#else
1158	GElf_Ehdr ehdr;
1159	GElf_Phdr ph;
1160	char name[MAXPATHLEN];
1161	uintptr_t mapbase, alignmask;
1162	int i = 0;
1163	int is_elf_obj;
1164
1165	(void) strlcpy(name, k_stat->name, sizeof(name));
1166	(void) strlcpy(fname, k_stat->pathname, sizeof(fname));
1167#endif
1168
1169	if ((fd = open(fname, O_RDONLY)) == -1 || fstat64(fd, &st) == -1 ||
1170	    (dmp = dt_module_create(dtp, name)) == NULL) {
1171		dt_dprintf("failed to open %s: %s\n", fname, strerror(errno));
1172		(void) close(fd);
1173		return;
1174	}
1175
1176	/*
1177	 * Since the module can unload out from under us (and /system/object
1178	 * will return ENOENT), tell libelf to cook the entire file now and
1179	 * then close the underlying file descriptor immediately.  If this
1180	 * succeeds, we know that we can continue safely using dmp->dm_elf.
1181	 */
1182	dmp->dm_elf = elf_begin(fd, ELF_C_READ, NULL);
1183	err = elf_cntl(dmp->dm_elf, ELF_C_FDREAD);
1184	(void) close(fd);
1185
1186	if (dmp->dm_elf == NULL || err == -1 ||
1187	    elf_getshdrstrndx(dmp->dm_elf, &shstrs) == -1) {
1188		dt_dprintf("failed to load %s: %s\n",
1189		    fname, elf_errmsg(elf_errno()));
1190		dt_module_destroy(dtp, dmp);
1191		return;
1192	}
1193
1194	switch (gelf_getclass(dmp->dm_elf)) {
1195	case ELFCLASS32:
1196		dmp->dm_ops = &dt_modops_32;
1197		bits = 32;
1198		break;
1199	case ELFCLASS64:
1200		dmp->dm_ops = &dt_modops_64;
1201		bits = 64;
1202		break;
1203	default:
1204		dt_dprintf("failed to load %s: unknown ELF class\n", fname);
1205		dt_module_destroy(dtp, dmp);
1206		return;
1207	}
1208#if defined(__FreeBSD__)
1209	mapbase = (uintptr_t)k_stat->address;
1210	gelf_getehdr(dmp->dm_elf, &ehdr);
1211	is_elf_obj = (ehdr.e_type == ET_REL);
1212	if (is_elf_obj) {
1213		dmp->dm_sec_offsets =
1214		    malloc(ehdr.e_shnum * sizeof(*dmp->dm_sec_offsets));
1215		if (dmp->dm_sec_offsets == NULL) {
1216			dt_dprintf("failed to allocate memory\n");
1217			dt_module_destroy(dtp, dmp);
1218			return;
1219		}
1220	}
1221#endif
1222	/*
1223	 * Iterate over the section headers locating various sections of
1224	 * interest and use their attributes to flesh out the dt_module_t.
1225	 */
1226	for (sp = NULL; (sp = elf_nextscn(dmp->dm_elf, sp)) != NULL; ) {
1227		if (gelf_getshdr(sp, &sh) == NULL || sh.sh_type == SHT_NULL ||
1228		    (s = elf_strptr(dmp->dm_elf, shstrs, sh.sh_name)) == NULL)
1229			continue; /* skip any malformed sections */
1230#if defined(__FreeBSD__)
1231		if (sh.sh_size == 0)
1232			continue;
1233		if (sh.sh_type == SHT_PROGBITS || sh.sh_type == SHT_NOBITS) {
1234			alignmask = sh.sh_addralign - 1;
1235			mapbase += alignmask;
1236			mapbase &= ~alignmask;
1237			sh.sh_addr = mapbase;
1238			if (is_elf_obj)
1239				dmp->dm_sec_offsets[elf_ndxscn(sp)] = sh.sh_addr;
1240			mapbase += sh.sh_size;
1241		}
1242#endif
1243		if (strcmp(s, ".text") == 0) {
1244			dmp->dm_text_size = sh.sh_size;
1245			dmp->dm_text_va = sh.sh_addr;
1246		} else if (strcmp(s, ".data") == 0) {
1247			dmp->dm_data_size = sh.sh_size;
1248			dmp->dm_data_va = sh.sh_addr;
1249		} else if (strcmp(s, ".bss") == 0) {
1250			dmp->dm_bss_size = sh.sh_size;
1251			dmp->dm_bss_va = sh.sh_addr;
1252		} else if (strcmp(s, ".info") == 0 &&
1253		    (dp = elf_getdata(sp, NULL)) != NULL) {
1254			bcopy(dp->d_buf, &dmp->dm_info,
1255			    MIN(sh.sh_size, sizeof (dmp->dm_info)));
1256		} else if (strcmp(s, ".filename") == 0 &&
1257		    (dp = elf_getdata(sp, NULL)) != NULL) {
1258			(void) strlcpy(dmp->dm_file,
1259			    dp->d_buf, sizeof (dmp->dm_file));
1260		}
1261	}
1262
1263	dmp->dm_flags |= DT_DM_KERNEL;
1264#ifdef illumos
1265	dmp->dm_modid = (int)OBJFS_MODID(st.st_ino);
1266#else
1267	/*
1268	 * Include .rodata and special sections into .text.
1269	 * This depends on default section layout produced by GNU ld
1270	 * for ELF objects and libraries:
1271	 * [Text][R/O data][R/W data][Dynamic][BSS][Non loadable]
1272	 */
1273	dmp->dm_text_size = dmp->dm_data_va - dmp->dm_text_va;
1274#if defined(__i386__)
1275	/*
1276	 * Find the first load section and figure out the relocation
1277	 * offset for the symbols. The kernel module will not need
1278	 * relocation, but the kernel linker modules will.
1279	 */
1280	for (i = 0; gelf_getphdr(dmp->dm_elf, i, &ph) != NULL; i++) {
1281		if (ph.p_type == PT_LOAD) {
1282			dmp->dm_reloc_offset = k_stat->address - ph.p_vaddr;
1283			break;
1284		}
1285	}
1286#endif
1287#endif /* illumos */
1288
1289	if (dmp->dm_info.objfs_info_primary)
1290		dmp->dm_flags |= DT_DM_PRIMARY;
1291
1292#ifdef __FreeBSD__
1293	ms.version = sizeof(ms);
1294	for (modid = kldfirstmod(k_stat->id); modid > 0;
1295	    modid = modnext(modid)) {
1296		if (modstat(modid, &ms) != 0) {
1297			dt_dprintf("modstat failed for id %d in %s: %s\n",
1298			    modid, k_stat->name, strerror(errno));
1299			continue;
1300		}
1301		if (dt_kmodule_lookup(dtp, ms.name) != NULL)
1302			continue;
1303
1304		dkmp = malloc(sizeof (*dkmp));
1305		if (dkmp == NULL) {
1306			dt_dprintf("failed to allocate memory\n");
1307			dt_module_destroy(dtp, dmp);
1308			return;
1309		}
1310
1311		h = dt_strtab_hash(ms.name, NULL) % dtp->dt_modbuckets;
1312		dkmp->dkm_next = dtp->dt_kmods[h];
1313		dkmp->dkm_name = strdup(ms.name);
1314		dkmp->dkm_module = dmp;
1315		dtp->dt_kmods[h] = dkmp;
1316	}
1317#endif
1318
1319	dt_dprintf("opened %d-bit module %s (%s) [%d]\n",
1320	    bits, dmp->dm_name, dmp->dm_file, dmp->dm_modid);
1321}
1322
1323/*
1324 * Unload all the loaded modules and then refresh the module cache with the
1325 * latest list of loaded modules and their address ranges.
1326 */
1327void
1328dtrace_update(dtrace_hdl_t *dtp)
1329{
1330	dt_module_t *dmp;
1331	DIR *dirp;
1332#if defined(__FreeBSD__)
1333	int fileid;
1334#endif
1335
1336	for (dmp = dt_list_next(&dtp->dt_modlist);
1337	    dmp != NULL; dmp = dt_list_next(dmp))
1338		dt_module_unload(dtp, dmp);
1339
1340#ifdef illumos
1341	/*
1342	 * Open /system/object and attempt to create a libdtrace module for
1343	 * each kernel module that is loaded on the current system.
1344	 */
1345	if (!(dtp->dt_oflags & DTRACE_O_NOSYS) &&
1346	    (dirp = opendir(OBJFS_ROOT)) != NULL) {
1347		struct dirent *dp;
1348
1349		while ((dp = readdir(dirp)) != NULL) {
1350			if (dp->d_name[0] != '.')
1351				dt_module_update(dtp, dp->d_name);
1352		}
1353
1354		(void) closedir(dirp);
1355	}
1356#elif defined(__FreeBSD__)
1357	/*
1358	 * Use FreeBSD's kernel loader interface to discover what kernel
1359	 * modules are loaded and create a libdtrace module for each one.
1360	 */
1361	for (fileid = kldnext(0); fileid > 0; fileid = kldnext(fileid)) {
1362		struct kld_file_stat k_stat;
1363		k_stat.version = sizeof(k_stat);
1364		if (kldstat(fileid, &k_stat) == 0)
1365			dt_module_update(dtp, &k_stat);
1366	}
1367#endif
1368
1369	/*
1370	 * Look up all the macro identifiers and set di_id to the latest value.
1371	 * This code collaborates with dt_lex.l on the use of di_id.  We will
1372	 * need to implement something fancier if we need to support non-ints.
1373	 */
1374	dt_idhash_lookup(dtp->dt_macros, "egid")->di_id = getegid();
1375	dt_idhash_lookup(dtp->dt_macros, "euid")->di_id = geteuid();
1376	dt_idhash_lookup(dtp->dt_macros, "gid")->di_id = getgid();
1377	dt_idhash_lookup(dtp->dt_macros, "pid")->di_id = getpid();
1378	dt_idhash_lookup(dtp->dt_macros, "pgid")->di_id = getpgid(0);
1379	dt_idhash_lookup(dtp->dt_macros, "ppid")->di_id = getppid();
1380#ifdef illumos
1381	dt_idhash_lookup(dtp->dt_macros, "projid")->di_id = getprojid();
1382#endif
1383	dt_idhash_lookup(dtp->dt_macros, "sid")->di_id = getsid(0);
1384#ifdef illumos
1385	dt_idhash_lookup(dtp->dt_macros, "taskid")->di_id = gettaskid();
1386#endif
1387	dt_idhash_lookup(dtp->dt_macros, "uid")->di_id = getuid();
1388
1389	/*
1390	 * Cache the pointers to the modules representing the base executable
1391	 * and the run-time linker in the dtrace client handle. Note that on
1392	 * x86 krtld is folded into unix, so if we don't find it, use unix
1393	 * instead.
1394	 */
1395	dtp->dt_exec = dt_module_lookup_by_name(dtp, "genunix");
1396	dtp->dt_rtld = dt_module_lookup_by_name(dtp, "krtld");
1397	if (dtp->dt_rtld == NULL)
1398		dtp->dt_rtld = dt_module_lookup_by_name(dtp, "unix");
1399
1400	/*
1401	 * If this is the first time we are initializing the module list,
1402	 * remove the module for genunix from the module list and then move it
1403	 * to the front of the module list.  We do this so that type and symbol
1404	 * queries encounter genunix and thereby optimize for the common case
1405	 * in dtrace_lookup_by_name() and dtrace_lookup_by_type(), below.
1406	 */
1407	if (dtp->dt_exec != NULL &&
1408	    dtp->dt_cdefs == NULL && dtp->dt_ddefs == NULL) {
1409		dt_list_delete(&dtp->dt_modlist, dtp->dt_exec);
1410		dt_list_prepend(&dtp->dt_modlist, dtp->dt_exec);
1411	}
1412}
1413
1414static dt_module_t *
1415dt_module_from_object(dtrace_hdl_t *dtp, const char *object)
1416{
1417	int err = EDT_NOMOD;
1418	dt_module_t *dmp;
1419
1420	switch ((uintptr_t)object) {
1421	case (uintptr_t)DTRACE_OBJ_EXEC:
1422		dmp = dtp->dt_exec;
1423		break;
1424	case (uintptr_t)DTRACE_OBJ_RTLD:
1425		dmp = dtp->dt_rtld;
1426		break;
1427	case (uintptr_t)DTRACE_OBJ_CDEFS:
1428		dmp = dtp->dt_cdefs;
1429		break;
1430	case (uintptr_t)DTRACE_OBJ_DDEFS:
1431		dmp = dtp->dt_ddefs;
1432		break;
1433	default:
1434		dmp = dt_module_create(dtp, object);
1435		err = EDT_NOMEM;
1436	}
1437
1438	if (dmp == NULL)
1439		(void) dt_set_errno(dtp, err);
1440
1441	return (dmp);
1442}
1443
1444/*
1445 * Exported interface to look up a symbol by name.  We return the GElf_Sym and
1446 * complete symbol information for the matching symbol.
1447 */
1448int
1449dtrace_lookup_by_name(dtrace_hdl_t *dtp, const char *object, const char *name,
1450    GElf_Sym *symp, dtrace_syminfo_t *sip)
1451{
1452	dt_module_t *dmp;
1453	dt_ident_t *idp;
1454	uint_t n, id;
1455	GElf_Sym sym;
1456
1457	uint_t mask = 0; /* mask of dt_module flags to match */
1458	uint_t bits = 0; /* flag bits that must be present */
1459
1460	if (object != DTRACE_OBJ_EVERY &&
1461	    object != DTRACE_OBJ_KMODS &&
1462	    object != DTRACE_OBJ_UMODS) {
1463		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1464			return (-1); /* dt_errno is set for us */
1465
1466		if (dt_module_load(dtp, dmp) == -1)
1467			return (-1); /* dt_errno is set for us */
1468		n = 1;
1469
1470	} else {
1471		if (object == DTRACE_OBJ_KMODS)
1472			mask = bits = DT_DM_KERNEL;
1473		else if (object == DTRACE_OBJ_UMODS)
1474			mask = DT_DM_KERNEL;
1475
1476		dmp = dt_list_next(&dtp->dt_modlist);
1477		n = dtp->dt_nmods;
1478	}
1479
1480	if (symp == NULL)
1481		symp = &sym;
1482
1483	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1484		if ((dmp->dm_flags & mask) != bits)
1485			continue; /* failed to match required attributes */
1486
1487		if (dt_module_load(dtp, dmp) == -1)
1488			continue; /* failed to load symbol table */
1489
1490		if (dmp->dm_ops->do_symname(dmp, name, symp, &id) != NULL) {
1491			if (sip != NULL) {
1492				sip->dts_object = dmp->dm_name;
1493				sip->dts_name = (const char *)
1494				    dmp->dm_strtab.cts_data + symp->st_name;
1495				sip->dts_id = id;
1496			}
1497			return (0);
1498		}
1499
1500		if (dmp->dm_extern != NULL &&
1501		    (idp = dt_idhash_lookup(dmp->dm_extern, name)) != NULL) {
1502			if (symp != &sym) {
1503				symp->st_name = (uintptr_t)idp->di_name;
1504				symp->st_info =
1505				    GELF_ST_INFO(STB_GLOBAL, STT_NOTYPE);
1506				symp->st_other = 0;
1507				symp->st_shndx = SHN_UNDEF;
1508				symp->st_value = 0;
1509				symp->st_size =
1510				    ctf_type_size(idp->di_ctfp, idp->di_type);
1511			}
1512
1513			if (sip != NULL) {
1514				sip->dts_object = dmp->dm_name;
1515				sip->dts_name = idp->di_name;
1516				sip->dts_id = idp->di_id;
1517			}
1518
1519			return (0);
1520		}
1521	}
1522
1523	return (dt_set_errno(dtp, EDT_NOSYM));
1524}
1525
1526/*
1527 * Exported interface to look up a symbol by address.  We return the GElf_Sym
1528 * and complete symbol information for the matching symbol.
1529 */
1530int
1531dtrace_lookup_by_addr(dtrace_hdl_t *dtp, GElf_Addr addr,
1532    GElf_Sym *symp, dtrace_syminfo_t *sip)
1533{
1534	dt_module_t *dmp;
1535	uint_t id;
1536	const dtrace_vector_t *v = dtp->dt_vector;
1537
1538	if (v != NULL)
1539		return (v->dtv_lookup_by_addr(dtp->dt_varg, addr, symp, sip));
1540
1541	for (dmp = dt_list_next(&dtp->dt_modlist); dmp != NULL;
1542	    dmp = dt_list_next(dmp)) {
1543		if (addr - dmp->dm_text_va < dmp->dm_text_size ||
1544		    addr - dmp->dm_data_va < dmp->dm_data_size ||
1545		    addr - dmp->dm_bss_va < dmp->dm_bss_size)
1546			break;
1547	}
1548
1549	if (dmp == NULL)
1550		return (dt_set_errno(dtp, EDT_NOSYMADDR));
1551
1552	if (dt_module_load(dtp, dmp) == -1)
1553		return (-1); /* dt_errno is set for us */
1554
1555	if (symp != NULL) {
1556		if (dmp->dm_ops->do_symaddr(dmp, addr, symp, &id) == NULL)
1557			return (dt_set_errno(dtp, EDT_NOSYMADDR));
1558	}
1559
1560	if (sip != NULL) {
1561		sip->dts_object = dmp->dm_name;
1562
1563		if (symp != NULL) {
1564			sip->dts_name = (const char *)
1565			    dmp->dm_strtab.cts_data + symp->st_name;
1566			sip->dts_id = id;
1567		} else {
1568			sip->dts_name = NULL;
1569			sip->dts_id = 0;
1570		}
1571	}
1572
1573	return (0);
1574}
1575
1576int
1577dtrace_lookup_by_type(dtrace_hdl_t *dtp, const char *object, const char *name,
1578    dtrace_typeinfo_t *tip)
1579{
1580	dtrace_typeinfo_t ti;
1581	dt_module_t *dmp;
1582	int found = 0;
1583	ctf_id_t id;
1584	uint_t n, i;
1585	int justone;
1586	ctf_file_t *fp;
1587	char *buf, *p, *q;
1588
1589	uint_t mask = 0; /* mask of dt_module flags to match */
1590	uint_t bits = 0; /* flag bits that must be present */
1591
1592	if (object != DTRACE_OBJ_EVERY &&
1593	    object != DTRACE_OBJ_KMODS &&
1594	    object != DTRACE_OBJ_UMODS) {
1595		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1596			return (-1); /* dt_errno is set for us */
1597
1598		if (dt_module_load(dtp, dmp) == -1)
1599			return (-1); /* dt_errno is set for us */
1600		n = 1;
1601		justone = 1;
1602	} else {
1603		if (object == DTRACE_OBJ_KMODS)
1604			mask = bits = DT_DM_KERNEL;
1605		else if (object == DTRACE_OBJ_UMODS)
1606			mask = DT_DM_KERNEL;
1607
1608		dmp = dt_list_next(&dtp->dt_modlist);
1609		n = dtp->dt_nmods;
1610		justone = 0;
1611	}
1612
1613	if (tip == NULL)
1614		tip = &ti;
1615
1616	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1617		if ((dmp->dm_flags & mask) != bits)
1618			continue; /* failed to match required attributes */
1619
1620		/*
1621		 * If we can't load the CTF container, continue on to the next
1622		 * module.  If our search was scoped to only one module then
1623		 * return immediately leaving dt_errno unmodified.
1624		 */
1625		if (dt_module_hasctf(dtp, dmp) == 0) {
1626			if (justone)
1627				return (-1);
1628			continue;
1629		}
1630
1631		/*
1632		 * Look up the type in the module's CTF container.  If our
1633		 * match is a forward declaration tag, save this choice in
1634		 * 'tip' and keep going in the hope that we will locate the
1635		 * underlying structure definition.  Otherwise just return.
1636		 */
1637		if (dmp->dm_pid == 0) {
1638			id = ctf_lookup_by_name(dmp->dm_ctfp, name);
1639			fp = dmp->dm_ctfp;
1640		} else {
1641			if ((p = strchr(name, '`')) != NULL) {
1642				buf = strdup(name);
1643				if (buf == NULL)
1644					return (dt_set_errno(dtp, EDT_NOMEM));
1645				p = strchr(buf, '`');
1646				if ((q = strchr(p + 1, '`')) != NULL)
1647					p = q;
1648				*p = '\0';
1649				fp = dt_module_getctflib(dtp, dmp, buf);
1650				if (fp == NULL || (id = ctf_lookup_by_name(fp,
1651				    p + 1)) == CTF_ERR)
1652					id = CTF_ERR;
1653				free(buf);
1654			} else {
1655				for (i = 0; i < dmp->dm_nctflibs; i++) {
1656					fp = dmp->dm_libctfp[i];
1657					id = ctf_lookup_by_name(fp, name);
1658					if (id != CTF_ERR)
1659						break;
1660				}
1661			}
1662		}
1663		if (id != CTF_ERR) {
1664			tip->dtt_object = dmp->dm_name;
1665			tip->dtt_ctfp = fp;
1666			tip->dtt_type = id;
1667			if (ctf_type_kind(fp, ctf_type_resolve(fp, id)) !=
1668			    CTF_K_FORWARD)
1669				return (0);
1670
1671			found++;
1672		}
1673	}
1674
1675	if (found == 0)
1676		return (dt_set_errno(dtp, EDT_NOTYPE));
1677
1678	return (0);
1679}
1680
1681int
1682dtrace_symbol_type(dtrace_hdl_t *dtp, const GElf_Sym *symp,
1683    const dtrace_syminfo_t *sip, dtrace_typeinfo_t *tip)
1684{
1685	dt_module_t *dmp;
1686
1687	tip->dtt_object = NULL;
1688	tip->dtt_ctfp = NULL;
1689	tip->dtt_type = CTF_ERR;
1690	tip->dtt_flags = 0;
1691
1692	if ((dmp = dt_module_lookup_by_name(dtp, sip->dts_object)) == NULL)
1693		return (dt_set_errno(dtp, EDT_NOMOD));
1694
1695	if (symp->st_shndx == SHN_UNDEF && dmp->dm_extern != NULL) {
1696		dt_ident_t *idp =
1697		    dt_idhash_lookup(dmp->dm_extern, sip->dts_name);
1698
1699		if (idp == NULL)
1700			return (dt_set_errno(dtp, EDT_NOSYM));
1701
1702		tip->dtt_ctfp = idp->di_ctfp;
1703		tip->dtt_type = idp->di_type;
1704
1705	} else if (GELF_ST_TYPE(symp->st_info) != STT_FUNC) {
1706		if (dt_module_getctf(dtp, dmp) == NULL)
1707			return (-1); /* errno is set for us */
1708
1709		tip->dtt_ctfp = dmp->dm_ctfp;
1710		tip->dtt_type = ctf_lookup_by_symbol(dmp->dm_ctfp, sip->dts_id);
1711
1712		if (tip->dtt_type == CTF_ERR) {
1713			dtp->dt_ctferr = ctf_errno(tip->dtt_ctfp);
1714			return (dt_set_errno(dtp, EDT_CTF));
1715		}
1716
1717	} else {
1718		tip->dtt_ctfp = DT_FPTR_CTFP(dtp);
1719		tip->dtt_type = DT_FPTR_TYPE(dtp);
1720	}
1721
1722	tip->dtt_object = dmp->dm_name;
1723	return (0);
1724}
1725
1726static dtrace_objinfo_t *
1727dt_module_info(const dt_module_t *dmp, dtrace_objinfo_t *dto)
1728{
1729	dto->dto_name = dmp->dm_name;
1730	dto->dto_file = dmp->dm_file;
1731	dto->dto_id = dmp->dm_modid;
1732	dto->dto_flags = 0;
1733
1734	if (dmp->dm_flags & DT_DM_KERNEL)
1735		dto->dto_flags |= DTRACE_OBJ_F_KERNEL;
1736	if (dmp->dm_flags & DT_DM_PRIMARY)
1737		dto->dto_flags |= DTRACE_OBJ_F_PRIMARY;
1738
1739	dto->dto_text_va = dmp->dm_text_va;
1740	dto->dto_text_size = dmp->dm_text_size;
1741	dto->dto_data_va = dmp->dm_data_va;
1742	dto->dto_data_size = dmp->dm_data_size;
1743	dto->dto_bss_va = dmp->dm_bss_va;
1744	dto->dto_bss_size = dmp->dm_bss_size;
1745
1746	return (dto);
1747}
1748
1749int
1750dtrace_object_iter(dtrace_hdl_t *dtp, dtrace_obj_f *func, void *data)
1751{
1752	const dt_module_t *dmp = dt_list_next(&dtp->dt_modlist);
1753	dtrace_objinfo_t dto;
1754	int rv;
1755
1756	for (; dmp != NULL; dmp = dt_list_next(dmp)) {
1757		if ((rv = (*func)(dtp, dt_module_info(dmp, &dto), data)) != 0)
1758			return (rv);
1759	}
1760
1761	return (0);
1762}
1763
1764int
1765dtrace_object_info(dtrace_hdl_t *dtp, const char *object, dtrace_objinfo_t *dto)
1766{
1767	dt_module_t *dmp;
1768
1769	if (object == DTRACE_OBJ_EVERY || object == DTRACE_OBJ_KMODS ||
1770	    object == DTRACE_OBJ_UMODS || dto == NULL)
1771		return (dt_set_errno(dtp, EINVAL));
1772
1773	if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1774		return (-1); /* dt_errno is set for us */
1775
1776	if (dt_module_load(dtp, dmp) == -1)
1777		return (-1); /* dt_errno is set for us */
1778
1779	(void) dt_module_info(dmp, dto);
1780	return (0);
1781}
1782