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_unlock(dtp, p);
715		dt_proc_release(dtp, p);
716		return (dt_set_errno(dtp, EDT_CANTLOAD));
717	}
718
719	if (arg.dpa_count == 0) {
720		dt_dprintf("no ctf data present\n");
721		dt_proc_unlock(dtp, p);
722		dt_proc_release(dtp, p);
723		return (dt_set_errno(dtp, EDT_CANTLOAD));
724	}
725
726	dmp->dm_libctfp = calloc(arg.dpa_count, sizeof (ctf_file_t *));
727	if (dmp->dm_libctfp == NULL) {
728		dt_proc_unlock(dtp, p);
729		dt_proc_release(dtp, p);
730		return (dt_set_errno(dtp, EDT_NOMEM));
731	}
732
733	dmp->dm_libctfn = calloc(arg.dpa_count, sizeof (char *));
734	if (dmp->dm_libctfn == NULL) {
735		free(dmp->dm_libctfp);
736		dt_proc_unlock(dtp, p);
737		dt_proc_release(dtp, p);
738		return (dt_set_errno(dtp, EDT_NOMEM));
739	}
740
741	dmp->dm_nctflibs = arg.dpa_count;
742
743	arg.dpa_count = 0;
744	if (Pobject_iter_resolved(p, dt_module_load_proc_build, &arg) != 0) {
745		dt_proc_unlock(dtp, p);
746		dt_module_unload(dtp, dmp);
747		dt_proc_release(dtp, p);
748		return (dt_set_errno(dtp, EDT_CANTLOAD));
749	}
750	assert(arg.dpa_count == dmp->dm_nctflibs);
751	dt_dprintf("loaded %d ctf modules for pid %d\n", arg.dpa_count,
752	    (int)dmp->dm_pid);
753
754	dt_proc_unlock(dtp, p);
755	dt_proc_release(dtp, p);
756	dmp->dm_flags |= DT_DM_LOADED;
757
758	return (0);
759}
760
761int
762dt_module_load(dtrace_hdl_t *dtp, dt_module_t *dmp)
763{
764	if (dmp->dm_flags & DT_DM_LOADED)
765		return (0); /* module is already loaded */
766
767	if (dmp->dm_pid != 0)
768		return (dt_module_load_proc(dtp, dmp));
769
770	dmp->dm_ctdata.cts_name = ".SUNW_ctf";
771	dmp->dm_ctdata.cts_type = SHT_PROGBITS;
772	dmp->dm_ctdata.cts_flags = 0;
773	dmp->dm_ctdata.cts_data = NULL;
774	dmp->dm_ctdata.cts_size = 0;
775	dmp->dm_ctdata.cts_entsize = 0;
776	dmp->dm_ctdata.cts_offset = 0;
777
778	dmp->dm_symtab.cts_name = ".symtab";
779	dmp->dm_symtab.cts_type = SHT_SYMTAB;
780	dmp->dm_symtab.cts_flags = 0;
781	dmp->dm_symtab.cts_data = NULL;
782	dmp->dm_symtab.cts_size = 0;
783	dmp->dm_symtab.cts_entsize = dmp->dm_ops == &dt_modops_64 ?
784	    sizeof (Elf64_Sym) : sizeof (Elf32_Sym);
785	dmp->dm_symtab.cts_offset = 0;
786
787	dmp->dm_strtab.cts_name = ".strtab";
788	dmp->dm_strtab.cts_type = SHT_STRTAB;
789	dmp->dm_strtab.cts_flags = 0;
790	dmp->dm_strtab.cts_data = NULL;
791	dmp->dm_strtab.cts_size = 0;
792	dmp->dm_strtab.cts_entsize = 0;
793	dmp->dm_strtab.cts_offset = 0;
794
795	/*
796	 * Attempt to load the module's CTF section, symbol table section, and
797	 * string table section.  Note that modules may not contain CTF data:
798	 * this will result in a successful load_sect but data of size zero.
799	 * We will then fail if dt_module_getctf() is called, as shown below.
800	 */
801	if (dt_module_load_sect(dtp, dmp, &dmp->dm_ctdata) == -1 ||
802	    dt_module_load_sect(dtp, dmp, &dmp->dm_symtab) == -1 ||
803	    dt_module_load_sect(dtp, dmp, &dmp->dm_strtab) == -1) {
804		dt_module_unload(dtp, dmp);
805		return (-1); /* dt_errno is set for us */
806	}
807
808	/*
809	 * Allocate the hash chains and hash buckets for symbol name lookup.
810	 * This is relatively simple since the symbol table is of fixed size
811	 * and is known in advance.  We allocate one extra element since we
812	 * use element indices instead of pointers and zero is our sentinel.
813	 */
814	dmp->dm_nsymelems =
815	    dmp->dm_symtab.cts_size / dmp->dm_symtab.cts_entsize;
816
817	dmp->dm_nsymbuckets = _dtrace_strbuckets;
818	dmp->dm_symfree = 1;		/* first free element is index 1 */
819
820	dmp->dm_symbuckets = calloc(dmp->dm_nsymbuckets, sizeof (uint_t));
821	dmp->dm_symchains = calloc(dmp->dm_nsymelems + 1, sizeof (dt_sym_t));
822
823	if (dmp->dm_symbuckets == NULL || dmp->dm_symchains == NULL) {
824		dt_module_unload(dtp, dmp);
825		return (dt_set_errno(dtp, EDT_NOMEM));
826	}
827
828	/*
829	 * Iterate over the symbol table data buffer and insert each symbol
830	 * name into the name hash if the name and type are valid.  Then
831	 * allocate the address map, fill it in, and sort it.
832	 */
833	dmp->dm_asrsv = dmp->dm_ops->do_syminit(dmp);
834
835	dt_dprintf("hashed %s [%s] (%u symbols)\n",
836	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_symfree - 1);
837
838	if ((dmp->dm_asmap = malloc(sizeof (void *) * dmp->dm_asrsv)) == NULL) {
839		dt_module_unload(dtp, dmp);
840		return (dt_set_errno(dtp, EDT_NOMEM));
841	}
842
843	dmp->dm_ops->do_symsort(dmp);
844
845	dt_dprintf("sorted %s [%s] (%u symbols)\n",
846	    dmp->dm_name, dmp->dm_symtab.cts_name, dmp->dm_aslen);
847
848	dmp->dm_flags |= DT_DM_LOADED;
849	return (0);
850}
851
852int
853dt_module_hasctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
854{
855	if (dmp->dm_pid != 0 && dmp->dm_nctflibs > 0)
856		return (1);
857	return (dt_module_getctf(dtp, dmp) != NULL);
858}
859
860ctf_file_t *
861dt_module_getctf(dtrace_hdl_t *dtp, dt_module_t *dmp)
862{
863	const char *parent;
864	dt_module_t *pmp;
865	ctf_file_t *pfp;
866	int model;
867
868	if (dmp->dm_ctfp != NULL || dt_module_load(dtp, dmp) != 0)
869		return (dmp->dm_ctfp);
870
871	if (dmp->dm_ops == &dt_modops_64)
872		model = CTF_MODEL_LP64;
873	else
874		model = CTF_MODEL_ILP32;
875
876	/*
877	 * If the data model of the module does not match our program data
878	 * model, then do not permit CTF from this module to be opened and
879	 * returned to the compiler.  If we support mixed data models in the
880	 * future for combined kernel/user tracing, this can be removed.
881	 */
882	if (dtp->dt_conf.dtc_ctfmodel != model) {
883		(void) dt_set_errno(dtp, EDT_DATAMODEL);
884		return (NULL);
885	}
886
887	if (dmp->dm_ctdata.cts_size == 0) {
888		(void) dt_set_errno(dtp, EDT_NOCTF);
889		return (NULL);
890	}
891
892	dmp->dm_ctfp = ctf_bufopen(&dmp->dm_ctdata,
893	    &dmp->dm_symtab, &dmp->dm_strtab, &dtp->dt_ctferr);
894
895	if (dmp->dm_ctfp == NULL) {
896		(void) dt_set_errno(dtp, EDT_CTF);
897		return (NULL);
898	}
899
900	(void) ctf_setmodel(dmp->dm_ctfp, model);
901	ctf_setspecific(dmp->dm_ctfp, dmp);
902
903	if ((parent = ctf_parent_name(dmp->dm_ctfp)) != NULL) {
904		if ((pmp = dt_module_create(dtp, parent)) == NULL ||
905		    (pfp = dt_module_getctf(dtp, pmp)) == NULL) {
906			if (pmp == NULL)
907				(void) dt_set_errno(dtp, EDT_NOMEM);
908			goto err;
909		}
910
911		if (ctf_import(dmp->dm_ctfp, pfp) == CTF_ERR) {
912			dtp->dt_ctferr = ctf_errno(dmp->dm_ctfp);
913			(void) dt_set_errno(dtp, EDT_CTF);
914			goto err;
915		}
916	}
917
918	dt_dprintf("loaded CTF container for %s (%p)\n",
919	    dmp->dm_name, (void *)dmp->dm_ctfp);
920
921	return (dmp->dm_ctfp);
922
923err:
924	ctf_close(dmp->dm_ctfp);
925	dmp->dm_ctfp = NULL;
926	return (NULL);
927}
928
929/*ARGSUSED*/
930void
931dt_module_unload(dtrace_hdl_t *dtp, dt_module_t *dmp)
932{
933	int i;
934
935	ctf_close(dmp->dm_ctfp);
936	dmp->dm_ctfp = NULL;
937
938#ifndef illumos
939	if (dmp->dm_ctdata.cts_data != NULL) {
940		free(dmp->dm_ctdata.cts_data);
941	}
942	if (dmp->dm_symtab.cts_data != NULL) {
943		free(dmp->dm_symtab.cts_data);
944	}
945	if (dmp->dm_strtab.cts_data != NULL) {
946		free(dmp->dm_strtab.cts_data);
947	}
948#endif
949
950	if (dmp->dm_libctfp != NULL) {
951		for (i = 0; i < dmp->dm_nctflibs; i++) {
952			ctf_close(dmp->dm_libctfp[i]);
953			free(dmp->dm_libctfn[i]);
954		}
955		free(dmp->dm_libctfp);
956		free(dmp->dm_libctfn);
957		dmp->dm_libctfp = NULL;
958		dmp->dm_nctflibs = 0;
959	}
960
961	bzero(&dmp->dm_ctdata, sizeof (ctf_sect_t));
962	bzero(&dmp->dm_symtab, sizeof (ctf_sect_t));
963	bzero(&dmp->dm_strtab, sizeof (ctf_sect_t));
964
965	if (dmp->dm_symbuckets != NULL) {
966		free(dmp->dm_symbuckets);
967		dmp->dm_symbuckets = NULL;
968	}
969
970	if (dmp->dm_symchains != NULL) {
971		free(dmp->dm_symchains);
972		dmp->dm_symchains = NULL;
973	}
974
975	if (dmp->dm_asmap != NULL) {
976		free(dmp->dm_asmap);
977		dmp->dm_asmap = NULL;
978	}
979#if defined(__FreeBSD__)
980	if (dmp->dm_sec_offsets != NULL) {
981		free(dmp->dm_sec_offsets);
982		dmp->dm_sec_offsets = NULL;
983	}
984#endif
985	dmp->dm_symfree = 0;
986	dmp->dm_nsymbuckets = 0;
987	dmp->dm_nsymelems = 0;
988	dmp->dm_asrsv = 0;
989	dmp->dm_aslen = 0;
990
991	dmp->dm_text_va = 0;
992	dmp->dm_text_size = 0;
993	dmp->dm_data_va = 0;
994	dmp->dm_data_size = 0;
995	dmp->dm_bss_va = 0;
996	dmp->dm_bss_size = 0;
997
998	if (dmp->dm_extern != NULL) {
999		dt_idhash_destroy(dmp->dm_extern);
1000		dmp->dm_extern = NULL;
1001	}
1002
1003	(void) elf_end(dmp->dm_elf);
1004	dmp->dm_elf = NULL;
1005
1006	dmp->dm_pid = 0;
1007
1008	dmp->dm_flags &= ~DT_DM_LOADED;
1009}
1010
1011void
1012dt_module_destroy(dtrace_hdl_t *dtp, dt_module_t *dmp)
1013{
1014	uint_t h = dt_strtab_hash(dmp->dm_name, NULL) % dtp->dt_modbuckets;
1015	dt_module_t **dmpp = &dtp->dt_mods[h];
1016
1017	dt_list_delete(&dtp->dt_modlist, dmp);
1018	assert(dtp->dt_nmods != 0);
1019	dtp->dt_nmods--;
1020
1021	/*
1022	 * Now remove this module from its hash chain.  We expect to always
1023	 * find the module on its hash chain, so in this loop we assert that
1024	 * we don't run off the end of the list.
1025	 */
1026	while (*dmpp != dmp) {
1027		dmpp = &((*dmpp)->dm_next);
1028		assert(*dmpp != NULL);
1029	}
1030
1031	*dmpp = dmp->dm_next;
1032
1033	dt_module_unload(dtp, dmp);
1034	free(dmp);
1035}
1036
1037/*
1038 * Insert a new external symbol reference into the specified module.  The new
1039 * symbol will be marked as undefined and is assigned a symbol index beyond
1040 * any existing cached symbols from this module.  We use the ident's di_data
1041 * field to store a pointer to a copy of the dtrace_syminfo_t for this symbol.
1042 */
1043dt_ident_t *
1044dt_module_extern(dtrace_hdl_t *dtp, dt_module_t *dmp,
1045    const char *name, const dtrace_typeinfo_t *tip)
1046{
1047	dtrace_syminfo_t *sip;
1048	dt_ident_t *idp;
1049	uint_t id;
1050
1051	if (dmp->dm_extern == NULL && (dmp->dm_extern = dt_idhash_create(
1052	    "extern", NULL, dmp->dm_nsymelems, UINT_MAX)) == NULL) {
1053		(void) dt_set_errno(dtp, EDT_NOMEM);
1054		return (NULL);
1055	}
1056
1057	if (dt_idhash_nextid(dmp->dm_extern, &id) == -1) {
1058		(void) dt_set_errno(dtp, EDT_SYMOFLOW);
1059		return (NULL);
1060	}
1061
1062	if ((sip = malloc(sizeof (dtrace_syminfo_t))) == NULL) {
1063		(void) dt_set_errno(dtp, EDT_NOMEM);
1064		return (NULL);
1065	}
1066
1067	idp = dt_idhash_insert(dmp->dm_extern, name, DT_IDENT_SYMBOL, 0, id,
1068	    _dtrace_symattr, 0, &dt_idops_thaw, NULL, dtp->dt_gen);
1069
1070	if (idp == NULL) {
1071		(void) dt_set_errno(dtp, EDT_NOMEM);
1072		free(sip);
1073		return (NULL);
1074	}
1075
1076	sip->dts_object = dmp->dm_name;
1077	sip->dts_name = idp->di_name;
1078	sip->dts_id = idp->di_id;
1079
1080	idp->di_data = sip;
1081	idp->di_ctfp = tip->dtt_ctfp;
1082	idp->di_type = tip->dtt_type;
1083
1084	return (idp);
1085}
1086
1087const char *
1088dt_module_modelname(dt_module_t *dmp)
1089{
1090	if (dmp->dm_ops == &dt_modops_64)
1091		return ("64-bit");
1092	else
1093		return ("32-bit");
1094}
1095
1096/* ARGSUSED */
1097int
1098dt_module_getlibid(dtrace_hdl_t *dtp, dt_module_t *dmp, const ctf_file_t *fp)
1099{
1100	int i;
1101
1102	for (i = 0; i < dmp->dm_nctflibs; i++) {
1103		if (dmp->dm_libctfp[i] == fp)
1104			return (i);
1105	}
1106
1107	return (-1);
1108}
1109
1110/* ARGSUSED */
1111ctf_file_t *
1112dt_module_getctflib(dtrace_hdl_t *dtp, dt_module_t *dmp, const char *name)
1113{
1114	int i;
1115
1116	for (i = 0; i < dmp->dm_nctflibs; i++) {
1117		if (strcmp(dmp->dm_libctfn[i], name) == 0)
1118			return (dmp->dm_libctfp[i]);
1119	}
1120
1121	return (NULL);
1122}
1123
1124/*
1125 * Update our module cache by adding an entry for the specified module 'name'.
1126 * We create the dt_module_t and populate it using /system/object/<name>/.
1127 *
1128 * On FreeBSD, the module name is passed as the full module file name,
1129 * including the path.
1130 */
1131static void
1132dt_module_update(dtrace_hdl_t *dtp, struct kld_file_stat *k_stat)
1133{
1134	char fname[MAXPATHLEN];
1135	struct stat64 st;
1136	int fd, err, bits;
1137	struct module_stat ms;
1138	dt_kmodule_t *dkmp;
1139	uint_t h;
1140	int modid;
1141	dt_module_t *dmp;
1142	const char *s;
1143	size_t shstrs;
1144	GElf_Shdr sh;
1145	Elf_Data *dp;
1146	Elf_Scn *sp;
1147	GElf_Ehdr ehdr;
1148	GElf_Phdr ph;
1149	char name[MAXPATHLEN];
1150	uintptr_t mapbase, alignmask;
1151	int i = 0;
1152	int is_elf_obj;
1153
1154	(void) strlcpy(name, k_stat->name, sizeof(name));
1155	(void) strlcpy(fname, k_stat->pathname, sizeof(fname));
1156
1157	if ((fd = open(fname, O_RDONLY)) == -1 || fstat64(fd, &st) == -1 ||
1158	    (dmp = dt_module_create(dtp, name)) == NULL) {
1159		dt_dprintf("failed to open %s: %s\n", fname, strerror(errno));
1160		(void) close(fd);
1161		return;
1162	}
1163
1164	(void) strlcpy(dmp->dm_file, fname, sizeof(dmp->dm_file));
1165	dmp->dm_modid = k_stat->id;
1166
1167	/*
1168	 * Since the module can unload out from under us (and /system/object
1169	 * will return ENOENT), tell libelf to cook the entire file now and
1170	 * then close the underlying file descriptor immediately.  If this
1171	 * succeeds, we know that we can continue safely using dmp->dm_elf.
1172	 */
1173	dmp->dm_elf = elf_begin(fd, ELF_C_READ, NULL);
1174	err = elf_cntl(dmp->dm_elf, ELF_C_FDREAD);
1175	(void) close(fd);
1176
1177	if (dmp->dm_elf == NULL || err == -1 ||
1178	    elf_getshdrstrndx(dmp->dm_elf, &shstrs) == -1) {
1179		dt_dprintf("failed to load %s: %s\n",
1180		    fname, elf_errmsg(elf_errno()));
1181		dt_module_destroy(dtp, dmp);
1182		return;
1183	}
1184
1185	switch (gelf_getclass(dmp->dm_elf)) {
1186	case ELFCLASS32:
1187		dmp->dm_ops = &dt_modops_32;
1188		bits = 32;
1189		break;
1190	case ELFCLASS64:
1191		dmp->dm_ops = &dt_modops_64;
1192		bits = 64;
1193		break;
1194	default:
1195		dt_dprintf("failed to load %s: unknown ELF class\n", fname);
1196		dt_module_destroy(dtp, dmp);
1197		return;
1198	}
1199	mapbase = (uintptr_t)k_stat->address;
1200	gelf_getehdr(dmp->dm_elf, &ehdr);
1201	is_elf_obj = (ehdr.e_type == ET_REL);
1202	if (is_elf_obj) {
1203		dmp->dm_sec_offsets =
1204		    malloc(ehdr.e_shnum * sizeof(*dmp->dm_sec_offsets));
1205		if (dmp->dm_sec_offsets == NULL) {
1206			dt_dprintf("failed to allocate memory\n");
1207			dt_module_destroy(dtp, dmp);
1208			return;
1209		}
1210	}
1211	/*
1212	 * Iterate over the section headers locating various sections of
1213	 * interest and use their attributes to flesh out the dt_module_t.
1214	 */
1215	for (sp = NULL; (sp = elf_nextscn(dmp->dm_elf, sp)) != NULL; ) {
1216		if (gelf_getshdr(sp, &sh) == NULL || sh.sh_type == SHT_NULL ||
1217		    (s = elf_strptr(dmp->dm_elf, shstrs, sh.sh_name)) == NULL)
1218			continue; /* skip any malformed sections */
1219		if (sh.sh_size == 0)
1220			continue;
1221		if (sh.sh_type == SHT_PROGBITS || sh.sh_type == SHT_NOBITS) {
1222			alignmask = sh.sh_addralign - 1;
1223			mapbase += alignmask;
1224			mapbase &= ~alignmask;
1225			sh.sh_addr = mapbase;
1226			if (is_elf_obj)
1227				dmp->dm_sec_offsets[elf_ndxscn(sp)] = sh.sh_addr;
1228			mapbase += sh.sh_size;
1229		}
1230		if (strcmp(s, ".text") == 0) {
1231			dmp->dm_text_size = sh.sh_size;
1232			dmp->dm_text_va = sh.sh_addr;
1233		} else if (strcmp(s, ".data") == 0) {
1234			dmp->dm_data_size = sh.sh_size;
1235			dmp->dm_data_va = sh.sh_addr;
1236		} else if (strcmp(s, ".bss") == 0) {
1237			dmp->dm_bss_size = sh.sh_size;
1238			dmp->dm_bss_va = sh.sh_addr;
1239		} else if (strcmp(s, ".info") == 0 &&
1240		    (dp = elf_getdata(sp, NULL)) != NULL) {
1241			bcopy(dp->d_buf, &dmp->dm_info,
1242			    MIN(sh.sh_size, sizeof (dmp->dm_info)));
1243		}
1244	}
1245
1246	dmp->dm_flags |= DT_DM_KERNEL;
1247	/*
1248	 * Include .rodata and special sections into .text.
1249	 * This depends on default section layout produced by GNU ld
1250	 * for ELF objects and libraries:
1251	 * [Text][R/O data][R/W data][Dynamic][BSS][Non loadable]
1252	 */
1253	dmp->dm_text_size = dmp->dm_data_va - dmp->dm_text_va;
1254#if defined(__i386__)
1255	/*
1256	 * Find the first load section and figure out the relocation
1257	 * offset for the symbols. The kernel module will not need
1258	 * relocation, but the kernel linker modules will.
1259	 */
1260	for (i = 0; gelf_getphdr(dmp->dm_elf, i, &ph) != NULL; i++) {
1261		if (ph.p_type == PT_LOAD) {
1262			dmp->dm_reloc_offset = k_stat->address - ph.p_vaddr;
1263			break;
1264		}
1265	}
1266#endif
1267
1268	if (dmp->dm_info.objfs_info_primary)
1269		dmp->dm_flags |= DT_DM_PRIMARY;
1270
1271	ms.version = sizeof(ms);
1272	for (modid = kldfirstmod(k_stat->id); modid > 0;
1273	    modid = modnext(modid)) {
1274		if (modstat(modid, &ms) != 0) {
1275			dt_dprintf("modstat failed for id %d in %s: %s\n",
1276			    modid, k_stat->name, strerror(errno));
1277			continue;
1278		}
1279		if (dt_kmodule_lookup(dtp, ms.name) != NULL)
1280			continue;
1281
1282		dkmp = malloc(sizeof (*dkmp));
1283		if (dkmp == NULL) {
1284			dt_dprintf("failed to allocate memory\n");
1285			dt_module_destroy(dtp, dmp);
1286			return;
1287		}
1288
1289		h = dt_strtab_hash(ms.name, NULL) % dtp->dt_modbuckets;
1290		dkmp->dkm_next = dtp->dt_kmods[h];
1291		dkmp->dkm_name = strdup(ms.name);
1292		dkmp->dkm_module = dmp;
1293		dtp->dt_kmods[h] = dkmp;
1294	}
1295
1296	dt_dprintf("opened %d-bit module %s (%s) [%d]\n",
1297	    bits, dmp->dm_name, dmp->dm_file, dmp->dm_modid);
1298}
1299
1300/*
1301 * Unload all the loaded modules and then refresh the module cache with the
1302 * latest list of loaded modules and their address ranges.
1303 */
1304void
1305dtrace_update(dtrace_hdl_t *dtp)
1306{
1307	dt_module_t *dmp;
1308	DIR *dirp;
1309#if defined(__FreeBSD__)
1310	int fileid;
1311#endif
1312
1313	for (dmp = dt_list_next(&dtp->dt_modlist);
1314	    dmp != NULL; dmp = dt_list_next(dmp))
1315		dt_module_unload(dtp, dmp);
1316
1317#ifdef illumos
1318	/*
1319	 * Open /system/object and attempt to create a libdtrace module for
1320	 * each kernel module that is loaded on the current system.
1321	 */
1322	if (!(dtp->dt_oflags & DTRACE_O_NOSYS) &&
1323	    (dirp = opendir(OBJFS_ROOT)) != NULL) {
1324		struct dirent *dp;
1325
1326		while ((dp = readdir(dirp)) != NULL) {
1327			if (dp->d_name[0] != '.')
1328				dt_module_update(dtp, dp->d_name);
1329		}
1330
1331		(void) closedir(dirp);
1332	}
1333#elif defined(__FreeBSD__)
1334	/*
1335	 * Use FreeBSD's kernel loader interface to discover what kernel
1336	 * modules are loaded and create a libdtrace module for each one.
1337	 */
1338	for (fileid = kldnext(0); fileid > 0; fileid = kldnext(fileid)) {
1339		struct kld_file_stat k_stat;
1340		k_stat.version = sizeof(k_stat);
1341		if (kldstat(fileid, &k_stat) == 0)
1342			dt_module_update(dtp, &k_stat);
1343	}
1344#endif
1345
1346	/*
1347	 * Look up all the macro identifiers and set di_id to the latest value.
1348	 * This code collaborates with dt_lex.l on the use of di_id.  We will
1349	 * need to implement something fancier if we need to support non-ints.
1350	 */
1351	dt_idhash_lookup(dtp->dt_macros, "egid")->di_id = getegid();
1352	dt_idhash_lookup(dtp->dt_macros, "euid")->di_id = geteuid();
1353	dt_idhash_lookup(dtp->dt_macros, "gid")->di_id = getgid();
1354	dt_idhash_lookup(dtp->dt_macros, "pid")->di_id = getpid();
1355	dt_idhash_lookup(dtp->dt_macros, "pgid")->di_id = getpgid(0);
1356	dt_idhash_lookup(dtp->dt_macros, "ppid")->di_id = getppid();
1357#ifdef illumos
1358	dt_idhash_lookup(dtp->dt_macros, "projid")->di_id = getprojid();
1359#endif
1360	dt_idhash_lookup(dtp->dt_macros, "sid")->di_id = getsid(0);
1361#ifdef illumos
1362	dt_idhash_lookup(dtp->dt_macros, "taskid")->di_id = gettaskid();
1363#endif
1364	dt_idhash_lookup(dtp->dt_macros, "uid")->di_id = getuid();
1365
1366	/*
1367	 * Cache the pointers to the modules representing the base executable
1368	 * and the run-time linker in the dtrace client handle. Note that on
1369	 * x86 krtld is folded into unix, so if we don't find it, use unix
1370	 * instead.
1371	 */
1372	dtp->dt_exec = dt_module_lookup_by_name(dtp, "genunix");
1373	dtp->dt_rtld = dt_module_lookup_by_name(dtp, "krtld");
1374	if (dtp->dt_rtld == NULL)
1375		dtp->dt_rtld = dt_module_lookup_by_name(dtp, "unix");
1376
1377	/*
1378	 * If this is the first time we are initializing the module list,
1379	 * remove the module for genunix from the module list and then move it
1380	 * to the front of the module list.  We do this so that type and symbol
1381	 * queries encounter genunix and thereby optimize for the common case
1382	 * in dtrace_lookup_by_name() and dtrace_lookup_by_type(), below.
1383	 */
1384	if (dtp->dt_exec != NULL &&
1385	    dtp->dt_cdefs == NULL && dtp->dt_ddefs == NULL) {
1386		dt_list_delete(&dtp->dt_modlist, dtp->dt_exec);
1387		dt_list_prepend(&dtp->dt_modlist, dtp->dt_exec);
1388	}
1389}
1390
1391static dt_module_t *
1392dt_module_from_object(dtrace_hdl_t *dtp, const char *object)
1393{
1394	int err = EDT_NOMOD;
1395	dt_module_t *dmp;
1396
1397	switch ((uintptr_t)object) {
1398	case (uintptr_t)DTRACE_OBJ_EXEC:
1399		dmp = dtp->dt_exec;
1400		break;
1401	case (uintptr_t)DTRACE_OBJ_RTLD:
1402		dmp = dtp->dt_rtld;
1403		break;
1404	case (uintptr_t)DTRACE_OBJ_CDEFS:
1405		dmp = dtp->dt_cdefs;
1406		break;
1407	case (uintptr_t)DTRACE_OBJ_DDEFS:
1408		dmp = dtp->dt_ddefs;
1409		break;
1410	default:
1411		dmp = dt_module_create(dtp, object);
1412		err = EDT_NOMEM;
1413	}
1414
1415	if (dmp == NULL)
1416		(void) dt_set_errno(dtp, err);
1417
1418	return (dmp);
1419}
1420
1421/*
1422 * Exported interface to look up a symbol by name.  We return the GElf_Sym and
1423 * complete symbol information for the matching symbol.
1424 */
1425int
1426dtrace_lookup_by_name(dtrace_hdl_t *dtp, const char *object, const char *name,
1427    GElf_Sym *symp, dtrace_syminfo_t *sip)
1428{
1429	dt_module_t *dmp;
1430	dt_ident_t *idp;
1431	uint_t n, id;
1432	GElf_Sym sym;
1433
1434	uint_t mask = 0; /* mask of dt_module flags to match */
1435	uint_t bits = 0; /* flag bits that must be present */
1436
1437	if (object != DTRACE_OBJ_EVERY &&
1438	    object != DTRACE_OBJ_KMODS &&
1439	    object != DTRACE_OBJ_UMODS) {
1440		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1441			return (-1); /* dt_errno is set for us */
1442
1443		if (dt_module_load(dtp, dmp) == -1)
1444			return (-1); /* dt_errno is set for us */
1445		n = 1;
1446
1447	} else {
1448		if (object == DTRACE_OBJ_KMODS)
1449			mask = bits = DT_DM_KERNEL;
1450		else if (object == DTRACE_OBJ_UMODS)
1451			mask = DT_DM_KERNEL;
1452
1453		dmp = dt_list_next(&dtp->dt_modlist);
1454		n = dtp->dt_nmods;
1455	}
1456
1457	if (symp == NULL)
1458		symp = &sym;
1459
1460	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1461		if ((dmp->dm_flags & mask) != bits)
1462			continue; /* failed to match required attributes */
1463
1464		if (dt_module_load(dtp, dmp) == -1)
1465			continue; /* failed to load symbol table */
1466
1467		if (dmp->dm_ops->do_symname(dmp, name, symp, &id) != NULL) {
1468			if (sip != NULL) {
1469				sip->dts_object = dmp->dm_name;
1470				sip->dts_name = (const char *)
1471				    dmp->dm_strtab.cts_data + symp->st_name;
1472				sip->dts_id = id;
1473			}
1474			return (0);
1475		}
1476
1477		if (dmp->dm_extern != NULL &&
1478		    (idp = dt_idhash_lookup(dmp->dm_extern, name)) != NULL) {
1479			if (symp != &sym) {
1480				symp->st_name = (uintptr_t)idp->di_name;
1481				symp->st_info =
1482				    GELF_ST_INFO(STB_GLOBAL, STT_NOTYPE);
1483				symp->st_other = 0;
1484				symp->st_shndx = SHN_UNDEF;
1485				symp->st_value = 0;
1486				symp->st_size =
1487				    ctf_type_size(idp->di_ctfp, idp->di_type);
1488			}
1489
1490			if (sip != NULL) {
1491				sip->dts_object = dmp->dm_name;
1492				sip->dts_name = idp->di_name;
1493				sip->dts_id = idp->di_id;
1494			}
1495
1496			return (0);
1497		}
1498	}
1499
1500	return (dt_set_errno(dtp, EDT_NOSYM));
1501}
1502
1503/*
1504 * Exported interface to look up a symbol by address.  We return the GElf_Sym
1505 * and complete symbol information for the matching symbol.
1506 */
1507int
1508dtrace_lookup_by_addr(dtrace_hdl_t *dtp, GElf_Addr addr,
1509    GElf_Sym *symp, dtrace_syminfo_t *sip)
1510{
1511	dt_module_t *dmp;
1512	uint_t id;
1513	const dtrace_vector_t *v = dtp->dt_vector;
1514
1515	if (v != NULL)
1516		return (v->dtv_lookup_by_addr(dtp->dt_varg, addr, symp, sip));
1517
1518	for (dmp = dt_list_next(&dtp->dt_modlist); dmp != NULL;
1519	    dmp = dt_list_next(dmp)) {
1520		if (addr - dmp->dm_text_va < dmp->dm_text_size ||
1521		    addr - dmp->dm_data_va < dmp->dm_data_size ||
1522		    addr - dmp->dm_bss_va < dmp->dm_bss_size)
1523			break;
1524	}
1525
1526	if (dmp == NULL)
1527		return (dt_set_errno(dtp, EDT_NOSYMADDR));
1528
1529	if (dt_module_load(dtp, dmp) == -1)
1530		return (-1); /* dt_errno is set for us */
1531
1532	if (symp != NULL) {
1533		if (dmp->dm_ops->do_symaddr(dmp, addr, symp, &id) == NULL)
1534			return (dt_set_errno(dtp, EDT_NOSYMADDR));
1535	}
1536
1537	if (sip != NULL) {
1538		sip->dts_object = dmp->dm_name;
1539
1540		if (symp != NULL) {
1541			sip->dts_name = (const char *)
1542			    dmp->dm_strtab.cts_data + symp->st_name;
1543			sip->dts_id = id;
1544		} else {
1545			sip->dts_name = NULL;
1546			sip->dts_id = 0;
1547		}
1548	}
1549
1550	return (0);
1551}
1552
1553int
1554dtrace_lookup_by_type(dtrace_hdl_t *dtp, const char *object, const char *name,
1555    dtrace_typeinfo_t *tip)
1556{
1557	dtrace_typeinfo_t ti;
1558	dt_module_t *dmp;
1559	int found = 0;
1560	ctf_id_t id;
1561	uint_t n, i;
1562	int justone;
1563	ctf_file_t *fp;
1564	char *buf, *p, *q;
1565
1566	uint_t mask = 0; /* mask of dt_module flags to match */
1567	uint_t bits = 0; /* flag bits that must be present */
1568
1569	if (object != DTRACE_OBJ_EVERY &&
1570	    object != DTRACE_OBJ_KMODS &&
1571	    object != DTRACE_OBJ_UMODS) {
1572		if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1573			return (-1); /* dt_errno is set for us */
1574
1575		if (dt_module_load(dtp, dmp) == -1)
1576			return (-1); /* dt_errno is set for us */
1577		n = 1;
1578		justone = 1;
1579	} else {
1580		if (object == DTRACE_OBJ_KMODS)
1581			mask = bits = DT_DM_KERNEL;
1582		else if (object == DTRACE_OBJ_UMODS)
1583			mask = DT_DM_KERNEL;
1584
1585		dmp = dt_list_next(&dtp->dt_modlist);
1586		n = dtp->dt_nmods;
1587		justone = 0;
1588	}
1589
1590	if (tip == NULL)
1591		tip = &ti;
1592
1593	for (; n > 0; n--, dmp = dt_list_next(dmp)) {
1594		if ((dmp->dm_flags & mask) != bits)
1595			continue; /* failed to match required attributes */
1596
1597		/*
1598		 * If we can't load the CTF container, continue on to the next
1599		 * module.  If our search was scoped to only one module then
1600		 * return immediately leaving dt_errno unmodified.
1601		 */
1602		if (dt_module_hasctf(dtp, dmp) == 0) {
1603			if (justone)
1604				return (-1);
1605			continue;
1606		}
1607
1608		/*
1609		 * Look up the type in the module's CTF container.  If our
1610		 * match is a forward declaration tag, save this choice in
1611		 * 'tip' and keep going in the hope that we will locate the
1612		 * underlying structure definition.  Otherwise just return.
1613		 */
1614		if (dmp->dm_pid == 0) {
1615			id = ctf_lookup_by_name(dmp->dm_ctfp, name);
1616			fp = dmp->dm_ctfp;
1617		} else {
1618			if ((p = strchr(name, '`')) != NULL) {
1619				buf = strdup(name);
1620				if (buf == NULL)
1621					return (dt_set_errno(dtp, EDT_NOMEM));
1622				p = strchr(buf, '`');
1623				if ((q = strchr(p + 1, '`')) != NULL)
1624					p = q;
1625				*p = '\0';
1626				fp = dt_module_getctflib(dtp, dmp, buf);
1627				if (fp == NULL || (id = ctf_lookup_by_name(fp,
1628				    p + 1)) == CTF_ERR)
1629					id = CTF_ERR;
1630				free(buf);
1631			} else {
1632				for (i = 0; i < dmp->dm_nctflibs; i++) {
1633					fp = dmp->dm_libctfp[i];
1634					id = ctf_lookup_by_name(fp, name);
1635					if (id != CTF_ERR)
1636						break;
1637				}
1638			}
1639		}
1640		if (id != CTF_ERR) {
1641			tip->dtt_object = dmp->dm_name;
1642			tip->dtt_ctfp = fp;
1643			tip->dtt_type = id;
1644			if (ctf_type_kind(fp, ctf_type_resolve(fp, id)) !=
1645			    CTF_K_FORWARD)
1646				return (0);
1647
1648			found++;
1649		}
1650	}
1651
1652	if (found == 0)
1653		return (dt_set_errno(dtp, EDT_NOTYPE));
1654
1655	return (0);
1656}
1657
1658int
1659dtrace_symbol_type(dtrace_hdl_t *dtp, const GElf_Sym *symp,
1660    const dtrace_syminfo_t *sip, dtrace_typeinfo_t *tip)
1661{
1662	dt_module_t *dmp;
1663
1664	tip->dtt_object = NULL;
1665	tip->dtt_ctfp = NULL;
1666	tip->dtt_type = CTF_ERR;
1667	tip->dtt_flags = 0;
1668
1669	if ((dmp = dt_module_lookup_by_name(dtp, sip->dts_object)) == NULL)
1670		return (dt_set_errno(dtp, EDT_NOMOD));
1671
1672	if (symp->st_shndx == SHN_UNDEF && dmp->dm_extern != NULL) {
1673		dt_ident_t *idp =
1674		    dt_idhash_lookup(dmp->dm_extern, sip->dts_name);
1675
1676		if (idp == NULL)
1677			return (dt_set_errno(dtp, EDT_NOSYM));
1678
1679		tip->dtt_ctfp = idp->di_ctfp;
1680		tip->dtt_type = idp->di_type;
1681
1682	} else if (GELF_ST_TYPE(symp->st_info) != STT_FUNC) {
1683		if (dt_module_getctf(dtp, dmp) == NULL)
1684			return (-1); /* errno is set for us */
1685
1686		tip->dtt_ctfp = dmp->dm_ctfp;
1687		tip->dtt_type = ctf_lookup_by_symbol(dmp->dm_ctfp, sip->dts_id);
1688
1689		if (tip->dtt_type == CTF_ERR) {
1690			dtp->dt_ctferr = ctf_errno(tip->dtt_ctfp);
1691			return (dt_set_errno(dtp, EDT_CTF));
1692		}
1693
1694	} else {
1695		tip->dtt_ctfp = DT_FPTR_CTFP(dtp);
1696		tip->dtt_type = DT_FPTR_TYPE(dtp);
1697	}
1698
1699	tip->dtt_object = dmp->dm_name;
1700	return (0);
1701}
1702
1703static dtrace_objinfo_t *
1704dt_module_info(const dt_module_t *dmp, dtrace_objinfo_t *dto)
1705{
1706	dto->dto_name = dmp->dm_name;
1707	dto->dto_file = dmp->dm_file;
1708	dto->dto_id = dmp->dm_modid;
1709	dto->dto_flags = 0;
1710
1711	if (dmp->dm_flags & DT_DM_KERNEL)
1712		dto->dto_flags |= DTRACE_OBJ_F_KERNEL;
1713	if (dmp->dm_flags & DT_DM_PRIMARY)
1714		dto->dto_flags |= DTRACE_OBJ_F_PRIMARY;
1715
1716	dto->dto_text_va = dmp->dm_text_va;
1717	dto->dto_text_size = dmp->dm_text_size;
1718	dto->dto_data_va = dmp->dm_data_va;
1719	dto->dto_data_size = dmp->dm_data_size;
1720	dto->dto_bss_va = dmp->dm_bss_va;
1721	dto->dto_bss_size = dmp->dm_bss_size;
1722
1723	return (dto);
1724}
1725
1726int
1727dtrace_object_iter(dtrace_hdl_t *dtp, dtrace_obj_f *func, void *data)
1728{
1729	const dt_module_t *dmp = dt_list_next(&dtp->dt_modlist);
1730	dtrace_objinfo_t dto;
1731	int rv;
1732
1733	for (; dmp != NULL; dmp = dt_list_next(dmp)) {
1734		if ((rv = (*func)(dtp, dt_module_info(dmp, &dto), data)) != 0)
1735			return (rv);
1736	}
1737
1738	return (0);
1739}
1740
1741int
1742dtrace_object_info(dtrace_hdl_t *dtp, const char *object, dtrace_objinfo_t *dto)
1743{
1744	dt_module_t *dmp;
1745
1746	if (object == DTRACE_OBJ_EVERY || object == DTRACE_OBJ_KMODS ||
1747	    object == DTRACE_OBJ_UMODS || dto == NULL)
1748		return (dt_set_errno(dtp, EINVAL));
1749
1750	if ((dmp = dt_module_from_object(dtp, object)) == NULL)
1751		return (-1); /* dt_errno is set for us */
1752
1753	if (dt_module_load(dtp, dmp) == -1)
1754		return (-1); /* dt_errno is set for us */
1755
1756	(void) dt_module_info(dmp, dto);
1757	return (0);
1758}
1759