xref: /freebsd/cddl/contrib/opensolaris/tools/ctf/cvt/dwarf.c (revision d11ff7f93386f36a385ea01498fd9ff9ff3862ce)
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  * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 /*
27  * DWARF to tdata conversion
28  *
29  * For the most part, conversion is straightforward, proceeding in two passes.
30  * On the first pass, we iterate through every die, creating new type nodes as
31  * necessary.  Referenced tdesc_t's are created in an uninitialized state, thus
32  * allowing type reference pointers to be filled in.  If the tdesc_t
33  * corresponding to a given die can be completely filled out (sizes and offsets
34  * calculated, and so forth) without using any referenced types, the tdesc_t is
35  * marked as resolved.  Consider an array type.  If the type corresponding to
36  * the array contents has not yet been processed, we will create a blank tdesc
37  * for the contents type (only the type ID will be filled in, relying upon the
38  * later portion of the first pass to encounter and complete the referenced
39  * type).  We will then attempt to determine the size of the array.  If the
40  * array has a byte size attribute, we will have completely characterized the
41  * array type, and will be able to mark it as resolved.  The lack of a byte
42  * size attribute, on the other hand, will prevent us from fully resolving the
43  * type, as the size will only be calculable with reference to the contents
44  * type, which has not, as yet, been encountered.  The array type will thus be
45  * left without the resolved flag, and the first pass will continue.
46  *
47  * When we begin the second pass, we will have created tdesc_t nodes for every
48  * type in the section.  We will traverse the tree, from the iidescs down,
49  * processing each unresolved node.  As the referenced nodes will have been
50  * populated, the array type used in our example above will be able to use the
51  * size of the referenced types (if available) to determine its own type.  The
52  * traversal will be repeated until all types have been resolved or we have
53  * failed to make progress.  When all tdescs have been resolved, the conversion
54  * is complete.
55  *
56  * There are, as always, a few special cases that are handled during the first
57  * and second passes:
58  *
59  *  1. Empty enums - GCC will occasionally emit an enum without any members.
60  *     Later on in the file, it will emit the same enum type, though this time
61  *     with the full complement of members.  All references to the memberless
62  *     enum need to be redirected to the full definition.  During the first
63  *     pass, each enum is entered in dm_enumhash, along with a pointer to its
64  *     corresponding tdesc_t.  If, during the second pass, we encounter a
65  *     memberless enum, we use the hash to locate the full definition.  All
66  *     tdescs referencing the empty enum are then redirected.
67  *
68  *  2. Forward declarations - If the compiler sees a forward declaration for
69  *     a structure, followed by the definition of that structure, it will emit
70  *     DWARF data for both the forward declaration and the definition.  We need
71  *     to resolve the forward declarations when possible, by redirecting
72  *     forward-referencing tdescs to the actual struct/union definitions.  This
73  *     redirection is done completely within the first pass.  We begin by
74  *     recording all forward declarations in dw_fwdhash.  When we define a
75  *     structure, we check to see if there have been any corresponding forward
76  *     declarations.  If so, we redirect the tdescs which referenced the forward
77  *     declarations to the structure or union definition.
78  *
79  * XXX see if a post traverser will allow the elimination of repeated pass 2
80  * traversals.
81  */
82 
83 #include <stdio.h>
84 #include <stdlib.h>
85 #include <string.h>
86 #include <strings.h>
87 #include <errno.h>
88 #include <libelf.h>
89 #include <libdwarf.h>
90 #include <libgen.h>
91 #include <dwarf.h>
92 
93 #include "ctf_headers.h"
94 #include "ctftools.h"
95 #include "memory.h"
96 #include "list.h"
97 #include "traverse.h"
98 
99 /*
100  * We need to define a couple of our own intrinsics, to smooth out some of the
101  * differences between the GCC and DevPro DWARF emitters.  See the referenced
102  * routines and the special cases in the file comment for more details.
103  *
104  * Type IDs are 32 bits wide.  We're going to use the top of that field to
105  * indicate types that we've created ourselves.
106  */
107 #define	TID_FILEMAX		0x3fffffff	/* highest tid from file */
108 #define	TID_VOID		0x40000001	/* see die_void() */
109 #define	TID_LONG		0x40000002	/* see die_array() */
110 
111 #define	TID_MFGTID_BASE		0x40000003	/* first mfg'd tid */
112 
113 /*
114  * To reduce the staggering amount of error-handling code that would otherwise
115  * be required, the attribute-retrieval routines handle most of their own
116  * errors.  If the following flag is supplied as the value of the `req'
117  * argument, they will also handle the absence of a requested attribute by
118  * terminating the program.
119  */
120 #define	DW_ATTR_REQ	1
121 
122 #define	TDESC_HASH_BUCKETS	511
123 
124 typedef struct dwarf {
125 	Dwarf_Debug dw_dw;		/* for libdwarf */
126 	Dwarf_Error dw_err;		/* for libdwarf */
127 	Dwarf_Off dw_maxoff;		/* highest legal offset in this cu */
128 	tdata_t *dw_td;			/* root of the tdesc/iidesc tree */
129 	hash_t *dw_tidhash;		/* hash of tdescs by t_id */
130 	hash_t *dw_fwdhash;		/* hash of fwd decls by name */
131 	hash_t *dw_enumhash;		/* hash of memberless enums by name */
132 	tdesc_t *dw_void;		/* manufactured void type */
133 	tdesc_t *dw_long;		/* manufactured long type for arrays */
134 	size_t dw_ptrsz;		/* size of a pointer in this file */
135 	tid_t dw_mfgtid_last;		/* last mfg'd type ID used */
136 	uint_t dw_nunres;		/* count of unresolved types */
137 	char *dw_cuname;		/* name of compilation unit */
138 } dwarf_t;
139 
140 static void die_create_one(dwarf_t *, Dwarf_Die);
141 static void die_create(dwarf_t *, Dwarf_Die);
142 
143 static tid_t
mfgtid_next(dwarf_t * dw)144 mfgtid_next(dwarf_t *dw)
145 {
146 	return (++dw->dw_mfgtid_last);
147 }
148 
149 static void
tdesc_add(dwarf_t * dw,tdesc_t * tdp)150 tdesc_add(dwarf_t *dw, tdesc_t *tdp)
151 {
152 	hash_add(dw->dw_tidhash, tdp);
153 }
154 
155 static tdesc_t *
tdesc_lookup(dwarf_t * dw,int tid)156 tdesc_lookup(dwarf_t *dw, int tid)
157 {
158 	tdesc_t tmpl;
159 	void *tdp;
160 
161 	tmpl.t_id = tid;
162 
163 	if (hash_find(dw->dw_tidhash, &tmpl, &tdp))
164 		return (tdp);
165 	else
166 		return (NULL);
167 }
168 
169 /*
170  * Resolve a tdesc down to a node which should have a size.  Returns the size,
171  * zero if the size hasn't yet been determined.
172  */
173 static size_t
tdesc_size(tdesc_t * tdp)174 tdesc_size(tdesc_t *tdp)
175 {
176 	for (;;) {
177 		switch (tdp->t_type) {
178 		case INTRINSIC:
179 		case POINTER:
180 		case ARRAY:
181 		case FUNCTION:
182 		case STRUCT:
183 		case UNION:
184 		case ENUM:
185 			return (tdp->t_size);
186 
187 		case FORWARD:
188 			return (0);
189 
190 		case TYPEDEF:
191 		case VOLATILE:
192 		case CONST:
193 		case RESTRICT:
194 			tdp = tdp->t_tdesc;
195 			continue;
196 
197 		case 0: /* not yet defined */
198 			return (0);
199 
200 		default:
201 			terminate("tdp %u: tdesc_size on unknown type %d\n",
202 			    tdp->t_id, tdp->t_type);
203 		}
204 	}
205 }
206 
207 static size_t
tdesc_bitsize(tdesc_t * tdp)208 tdesc_bitsize(tdesc_t *tdp)
209 {
210 	for (;;) {
211 		switch (tdp->t_type) {
212 		case INTRINSIC:
213 			return (tdp->t_intr->intr_nbits);
214 
215 		case ARRAY:
216 		case FUNCTION:
217 		case STRUCT:
218 		case UNION:
219 		case ENUM:
220 		case POINTER:
221 			return (tdp->t_size * NBBY);
222 
223 		case FORWARD:
224 			return (0);
225 
226 		case TYPEDEF:
227 		case VOLATILE:
228 		case RESTRICT:
229 		case CONST:
230 			tdp = tdp->t_tdesc;
231 			continue;
232 
233 		case 0: /* not yet defined */
234 			return (0);
235 
236 		default:
237 			terminate("tdp %u: tdesc_bitsize on unknown type %d\n",
238 			    tdp->t_id, tdp->t_type);
239 		}
240 	}
241 }
242 
243 static tdesc_t *
tdesc_basetype(tdesc_t * tdp)244 tdesc_basetype(tdesc_t *tdp)
245 {
246 	for (;;) {
247 		switch (tdp->t_type) {
248 		case TYPEDEF:
249 		case VOLATILE:
250 		case RESTRICT:
251 		case CONST:
252 			tdp = tdp->t_tdesc;
253 			break;
254 		case 0: /* not yet defined */
255 			return (NULL);
256 		default:
257 			return (tdp);
258 		}
259 	}
260 }
261 
262 static Dwarf_Off
die_off(dwarf_t * dw,Dwarf_Die die)263 die_off(dwarf_t *dw, Dwarf_Die die)
264 {
265 	Dwarf_Off off;
266 
267 	if (dwarf_dieoffset(die, &off, &dw->dw_err) == DW_DLV_OK)
268 		return (off);
269 
270 	terminate("failed to get offset for die: %s\n",
271 	    dwarf_errmsg(dw->dw_err));
272 	/*NOTREACHED*/
273 	return (0);
274 }
275 
276 static Dwarf_Die
die_sibling(dwarf_t * dw,Dwarf_Die die)277 die_sibling(dwarf_t *dw, Dwarf_Die die)
278 {
279 	Dwarf_Die sib;
280 	int rc;
281 
282 	if ((rc = dwarf_siblingof(dw->dw_dw, die, &sib, &dw->dw_err)) ==
283 	    DW_DLV_OK)
284 		return (sib);
285 	else if (rc == DW_DLV_NO_ENTRY)
286 		return (NULL);
287 
288 	terminate("die %llu: failed to find type sibling: %s\n",
289 	    die_off(dw, die), dwarf_errmsg(dw->dw_err));
290 	/*NOTREACHED*/
291 	return (NULL);
292 }
293 
294 static Dwarf_Die
die_child(dwarf_t * dw,Dwarf_Die die)295 die_child(dwarf_t *dw, Dwarf_Die die)
296 {
297 	Dwarf_Die child;
298 	int rc;
299 
300 	if ((rc = dwarf_child(die, &child, &dw->dw_err)) == DW_DLV_OK)
301 		return (child);
302 	else if (rc == DW_DLV_NO_ENTRY)
303 		return (NULL);
304 
305 	terminate("die %llu: failed to find type child: %s\n",
306 	    die_off(dw, die), dwarf_errmsg(dw->dw_err));
307 	/*NOTREACHED*/
308 	return (NULL);
309 }
310 
311 static Dwarf_Half
die_tag(dwarf_t * dw,Dwarf_Die die)312 die_tag(dwarf_t *dw, Dwarf_Die die)
313 {
314 	Dwarf_Half tag;
315 
316 	if (dwarf_tag(die, &tag, &dw->dw_err) == DW_DLV_OK)
317 		return (tag);
318 
319 	terminate("die %llu: failed to get tag for type: %s\n",
320 	    die_off(dw, die), dwarf_errmsg(dw->dw_err));
321 	/*NOTREACHED*/
322 	return (0);
323 }
324 
325 static Dwarf_Attribute
die_attr(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,int req)326 die_attr(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name, int req)
327 {
328 	Dwarf_Attribute attr;
329 	int rc;
330 
331 	if ((rc = dwarf_attr(die, name, &attr, &dw->dw_err)) == DW_DLV_OK) {
332 		return (attr);
333 	} else if (rc == DW_DLV_NO_ENTRY) {
334 		if (req) {
335 			terminate("die %llu: no attr 0x%x\n", die_off(dw, die),
336 			    name);
337 		} else {
338 			return (NULL);
339 		}
340 	}
341 
342 	terminate("die %llu: failed to get attribute for type: %s\n",
343 	    die_off(dw, die), dwarf_errmsg(dw->dw_err));
344 	/*NOTREACHED*/
345 	return (NULL);
346 }
347 
348 static int
die_signed(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,Dwarf_Signed * valp,int req)349 die_signed(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name, Dwarf_Signed *valp,
350     int req)
351 {
352 	*valp = 0;
353 	if (dwarf_attrval_signed(die, name, valp, &dw->dw_err) != DW_DLV_OK) {
354 		if (req)
355 			terminate("die %llu: failed to get signed: %s\n",
356 			    die_off(dw, die), dwarf_errmsg(dw->dw_err));
357 		return (0);
358 	}
359 
360 	return (1);
361 }
362 
363 static int
die_unsigned(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,Dwarf_Unsigned * valp,int req)364 die_unsigned(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name, Dwarf_Unsigned *valp,
365     int req)
366 {
367 	*valp = 0;
368 	if (dwarf_attrval_unsigned(die, name, valp, &dw->dw_err) != DW_DLV_OK) {
369 		if (req)
370 			terminate("die %llu: failed to get unsigned: %s\n",
371 			    die_off(dw, die), dwarf_errmsg(dw->dw_err));
372 		return (0);
373 	}
374 
375 	return (1);
376 }
377 
378 static int
die_bool(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,Dwarf_Bool * valp,int req)379 die_bool(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name, Dwarf_Bool *valp, int req)
380 {
381 	*valp = 0;
382 
383 	if (dwarf_attrval_flag(die, name, valp, &dw->dw_err) != DW_DLV_OK) {
384 		if (req)
385 			terminate("die %llu: failed to get flag: %s\n",
386 			    die_off(dw, die), dwarf_errmsg(dw->dw_err));
387 		return (0);
388 	}
389 
390 	return (1);
391 }
392 
393 static int
die_string(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,char ** strp,int req)394 die_string(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name, char **strp, int req)
395 {
396 	const char *str = NULL;
397 
398 	if (dwarf_attrval_string(die, name, &str, &dw->dw_err) != DW_DLV_OK ||
399 	    str == NULL) {
400 		if (req)
401 			terminate("die %llu: failed to get string: %s\n",
402 			    die_off(dw, die), dwarf_errmsg(dw->dw_err));
403 		else
404 			*strp = NULL;
405 		return (0);
406 	} else
407 		*strp = xstrdup(str);
408 
409 	return (1);
410 }
411 
412 static Dwarf_Off
die_attr_ref(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name)413 die_attr_ref(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name)
414 {
415 	Dwarf_Off off;
416 
417 	if (dwarf_attrval_unsigned(die, name, &off, &dw->dw_err) != DW_DLV_OK) {
418 		terminate("die %llu: failed to get ref: %s\n",
419 		    die_off(dw, die), dwarf_errmsg(dw->dw_err));
420 	}
421 
422 	return (off);
423 }
424 
425 static char *
die_name(dwarf_t * dw,Dwarf_Die die)426 die_name(dwarf_t *dw, Dwarf_Die die)
427 {
428 	char *str = NULL;
429 
430 	(void) die_string(dw, die, DW_AT_name, &str, 0);
431 	if (str == NULL)
432 		str = xstrdup("");
433 
434 	return (str);
435 }
436 
437 static int
die_isdecl(dwarf_t * dw,Dwarf_Die die)438 die_isdecl(dwarf_t *dw, Dwarf_Die die)
439 {
440 	Dwarf_Bool val;
441 
442 	return (die_bool(dw, die, DW_AT_declaration, &val, 0) && val);
443 }
444 
445 static int
die_isglobal(dwarf_t * dw,Dwarf_Die die)446 die_isglobal(dwarf_t *dw, Dwarf_Die die)
447 {
448 	Dwarf_Signed vis;
449 	Dwarf_Bool ext;
450 
451 	/*
452 	 * Some compilers (gcc) use DW_AT_external to indicate function
453 	 * visibility.  Others (Sun) use DW_AT_visibility.
454 	 */
455 	if (die_signed(dw, die, DW_AT_visibility, &vis, 0))
456 		return (vis == DW_VIS_exported);
457 	else
458 		return (die_bool(dw, die, DW_AT_external, &ext, 0) && ext);
459 }
460 
461 static tdesc_t *
die_add(dwarf_t * dw,Dwarf_Off off)462 die_add(dwarf_t *dw, Dwarf_Off off)
463 {
464 	tdesc_t *tdp = xcalloc(sizeof (tdesc_t));
465 
466 	tdp->t_id = off;
467 
468 	tdesc_add(dw, tdp);
469 
470 	return (tdp);
471 }
472 
473 static tdesc_t *
die_lookup_pass1(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name)474 die_lookup_pass1(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name)
475 {
476 	Dwarf_Off ref = die_attr_ref(dw, die, name);
477 	tdesc_t *tdp;
478 
479 	if ((tdp = tdesc_lookup(dw, ref)) != NULL)
480 		return (tdp);
481 
482 	return (die_add(dw, ref));
483 }
484 
485 static int
die_mem_offset(dwarf_t * dw,Dwarf_Die die,Dwarf_Half name,Dwarf_Unsigned * valp,int req __unused)486 die_mem_offset(dwarf_t *dw, Dwarf_Die die, Dwarf_Half name,
487     Dwarf_Unsigned *valp, int req __unused)
488 {
489 	Dwarf_Locdesc *loc = NULL;
490 	Dwarf_Signed locnum = 0;
491 	Dwarf_Attribute at;
492 	Dwarf_Half form;
493 
494 	if (name != DW_AT_data_member_location)
495 		terminate("die %llu: can only process attribute "
496 		    "DW_AT_data_member_location\n", die_off(dw, die));
497 
498 	if ((at = die_attr(dw, die, name, 0)) == NULL)
499 		return (0);
500 
501 	if (dwarf_whatform(at, &form, &dw->dw_err) != DW_DLV_OK)
502 		return (0);
503 
504 	switch (form) {
505 	case DW_FORM_sec_offset:
506 	case DW_FORM_block:
507 	case DW_FORM_block1:
508 	case DW_FORM_block2:
509 	case DW_FORM_block4:
510 		/*
511 		 * GCC in base and Clang (3.3 or below) generates
512 		 * DW_AT_data_member_location attribute with DW_FORM_block*
513 		 * form. The attribute contains one DW_OP_plus_uconst
514 		 * operator. The member offset stores in the operand.
515 		 */
516 		if (dwarf_loclist(at, &loc, &locnum, &dw->dw_err) != DW_DLV_OK)
517 			return (0);
518 		if (locnum != 1 || loc->ld_s->lr_atom != DW_OP_plus_uconst) {
519 			terminate("die %llu: cannot parse member offset with "
520 			    "operator other than DW_OP_plus_uconst\n",
521 			    die_off(dw, die));
522 		}
523 		*valp = loc->ld_s->lr_number;
524 		if (loc != NULL) {
525 			dwarf_dealloc(dw->dw_dw, loc->ld_s, DW_DLA_LOC_BLOCK);
526 			dwarf_dealloc(dw->dw_dw, loc, DW_DLA_LOCDESC);
527 		}
528 		break;
529 
530 	case DW_FORM_data1:
531 	case DW_FORM_data2:
532 	case DW_FORM_data4:
533 	case DW_FORM_data8:
534 	case DW_FORM_udata:
535 		/*
536 		 * Clang 3.4 generates DW_AT_data_member_location attribute
537 		 * with DW_FORM_data* form (constant class). The attribute
538 		 * stores a contant value which is the member offset.
539 		 *
540 		 * However, note that DW_FORM_data[48] in DWARF version 2 or 3
541 		 * could be used as a section offset (offset into .debug_loc in
542 		 * this case). Here we assume the attribute always stores a
543 		 * constant because we know Clang 3.4 does this and GCC in
544 		 * base won't emit DW_FORM_data[48] for this attribute. This
545 		 * code will remain correct if future vesrions of Clang and
546 		 * GCC conform to DWARF4 standard and only use the form
547 		 * DW_FORM_sec_offset for section offset.
548 		 */
549 		if (dwarf_attrval_unsigned(die, name, valp, &dw->dw_err) !=
550 		    DW_DLV_OK)
551 			return (0);
552 		break;
553 
554 	default:
555 		terminate("die %llu: cannot parse member offset with form "
556 		    "%u\n", die_off(dw, die), form);
557 	}
558 
559 	return (1);
560 }
561 
562 static tdesc_t *
tdesc_intr_common(dwarf_t * dw,int tid,const char * name,size_t sz)563 tdesc_intr_common(dwarf_t *dw, int tid, const char *name, size_t sz)
564 {
565 	tdesc_t *tdp;
566 	intr_t *intr;
567 
568 	intr = xcalloc(sizeof (intr_t));
569 	intr->intr_type = INTR_INT;
570 	intr->intr_signed = 1;
571 	intr->intr_nbits = sz * NBBY;
572 
573 	tdp = xcalloc(sizeof (tdesc_t));
574 	tdp->t_name = xstrdup(name);
575 	tdp->t_size = sz;
576 	tdp->t_id = tid;
577 	tdp->t_type = INTRINSIC;
578 	tdp->t_intr = intr;
579 	tdp->t_flags = TDESC_F_RESOLVED;
580 
581 	tdesc_add(dw, tdp);
582 
583 	return (tdp);
584 }
585 
586 /*
587  * Manufacture a void type.  Used for gcc-emitted stabs, where the lack of a
588  * type reference implies a reference to a void type.  A void *, for example
589  * will be represented by a pointer die without a DW_AT_type.  CTF requires
590  * that pointer nodes point to something, so we'll create a void for use as
591  * the target.  Note that the DWARF data may already create a void type.  Ours
592  * would then be a duplicate, but it'll be removed in the self-uniquification
593  * merge performed at the completion of DWARF->tdesc conversion.
594  */
595 static tdesc_t *
tdesc_intr_void(dwarf_t * dw)596 tdesc_intr_void(dwarf_t *dw)
597 {
598 	if (dw->dw_void == NULL)
599 		dw->dw_void = tdesc_intr_common(dw, TID_VOID, "void", 0);
600 
601 	return (dw->dw_void);
602 }
603 
604 static tdesc_t *
tdesc_intr_long(dwarf_t * dw)605 tdesc_intr_long(dwarf_t *dw)
606 {
607 	if (dw->dw_long == NULL) {
608 		dw->dw_long = tdesc_intr_common(dw, TID_LONG, "long",
609 		    dw->dw_ptrsz);
610 	}
611 
612 	return (dw->dw_long);
613 }
614 
615 /*
616  * Used for creating bitfield types.  We create a copy of an existing intrinsic,
617  * adjusting the size of the copy to match what the caller requested.  The
618  * caller can then use the copy as the type for a bitfield structure member.
619  */
620 static tdesc_t *
tdesc_intr_clone(dwarf_t * dw,tdesc_t * old,size_t bitsz,const char * suffix)621 tdesc_intr_clone(dwarf_t *dw, tdesc_t *old, size_t bitsz, const char *suffix)
622 {
623 	tdesc_t *new = xcalloc(sizeof (tdesc_t));
624 
625 	if (!(old->t_flags & TDESC_F_RESOLVED)) {
626 		terminate("tdp %u: attempt to make a bit field from an "
627 		    "unresolved type\n", old->t_id);
628 	}
629 
630 	xasprintf(&new->t_name, "%s %s", old->t_name, suffix);
631 	new->t_size = old->t_size;
632 	new->t_id = mfgtid_next(dw);
633 	new->t_type = INTRINSIC;
634 	new->t_flags = TDESC_F_RESOLVED;
635 
636 	new->t_intr = xcalloc(sizeof (intr_t));
637 	bcopy(old->t_intr, new->t_intr, sizeof (intr_t));
638 	new->t_intr->intr_nbits = bitsz;
639 
640 	tdesc_add(dw, new);
641 
642 	return (new);
643 }
644 
645 static void
tdesc_array_create(dwarf_t * dw,Dwarf_Die dim,tdesc_t * arrtdp,tdesc_t * dimtdp)646 tdesc_array_create(dwarf_t *dw, Dwarf_Die dim, tdesc_t *arrtdp,
647     tdesc_t *dimtdp)
648 {
649 	Dwarf_Unsigned uval;
650 	Dwarf_Signed sval;
651 	tdesc_t *ctdp = NULL;
652 	Dwarf_Die dim2;
653 	ardef_t *ar;
654 
655 	if ((dim2 = die_sibling(dw, dim)) == NULL) {
656 		ctdp = arrtdp;
657 	} else if (die_tag(dw, dim2) == DW_TAG_subrange_type) {
658 		ctdp = xcalloc(sizeof (tdesc_t));
659 		ctdp->t_id = mfgtid_next(dw);
660 		debug(3, "die %llu: creating new type %u for sub-dimension\n",
661 		    die_off(dw, dim2), ctdp->t_id);
662 		tdesc_array_create(dw, dim2, arrtdp, ctdp);
663 	} else {
664 		terminate("die %llu: unexpected non-subrange node in array\n",
665 		    die_off(dw, dim2));
666 	}
667 
668 	dimtdp->t_type = ARRAY;
669 	dimtdp->t_ardef = ar = xcalloc(sizeof (ardef_t));
670 
671 	/*
672 	 * Array bounds can be signed or unsigned, but there are several kinds
673 	 * of signless forms (data1, data2, etc) that take their sign from the
674 	 * routine that is trying to interpret them.  That is, data1 can be
675 	 * either signed or unsigned, depending on whether you use the signed or
676 	 * unsigned accessor function.  GCC will use the signless forms to store
677 	 * unsigned values which have their high bit set, so we need to try to
678 	 * read them first as unsigned to get positive values.  We could also
679 	 * try signed first, falling back to unsigned if we got a negative
680 	 * value.
681 	 */
682 	if (die_unsigned(dw, dim, DW_AT_upper_bound, &uval, 0))
683 		ar->ad_nelems = uval + 1;
684 	else if (die_signed(dw, dim, DW_AT_upper_bound, &sval, 0))
685 		ar->ad_nelems = sval + 1;
686 	else if (die_unsigned(dw, dim, DW_AT_count, &uval, 0))
687 		ar->ad_nelems = uval;
688 	else if (die_signed(dw, dim, DW_AT_count, &sval, 0))
689 		ar->ad_nelems = sval;
690 	else
691 		ar->ad_nelems = 0;
692 
693 	/*
694 	 * Different compilers use different index types.  Force the type to be
695 	 * a common, known value (long).
696 	 */
697 	ar->ad_idxtype = tdesc_intr_long(dw);
698 	ar->ad_contents = ctdp;
699 
700 	if (ar->ad_contents->t_size != 0) {
701 		dimtdp->t_size = ar->ad_contents->t_size * ar->ad_nelems;
702 		dimtdp->t_flags |= TDESC_F_RESOLVED;
703 	}
704 }
705 
706 /*
707  * Create a tdesc from an array node.  Some arrays will come with byte size
708  * attributes, and thus can be resolved immediately.  Others don't, and will
709  * need to wait until the second pass for resolution.
710  */
711 static void
die_array_create(dwarf_t * dw,Dwarf_Die arr,Dwarf_Off off,tdesc_t * tdp)712 die_array_create(dwarf_t *dw, Dwarf_Die arr, Dwarf_Off off, tdesc_t *tdp)
713 {
714 	tdesc_t *arrtdp = die_lookup_pass1(dw, arr, DW_AT_type);
715 	Dwarf_Unsigned uval;
716 	Dwarf_Die dim;
717 
718 	debug(3, "die %llu <%llx>: creating array\n", off, off);
719 
720 	if ((dim = die_child(dw, arr)) == NULL ||
721 	    die_tag(dw, dim) != DW_TAG_subrange_type)
722 		terminate("die %llu: failed to retrieve array bounds\n", off);
723 
724 	tdesc_array_create(dw, dim, arrtdp, tdp);
725 
726 	if (die_unsigned(dw, arr, DW_AT_byte_size, &uval, 0)) {
727 		tdesc_t *dimtdp;
728 		int flags;
729 
730 		tdp->t_size = uval;
731 
732 		/*
733 		 * Ensure that sub-dimensions have sizes too before marking
734 		 * as resolved.
735 		 */
736 		flags = TDESC_F_RESOLVED;
737 		for (dimtdp = tdp->t_ardef->ad_contents;
738 		    dimtdp->t_type == ARRAY;
739 		    dimtdp = dimtdp->t_ardef->ad_contents) {
740 			if (!(dimtdp->t_flags & TDESC_F_RESOLVED)) {
741 				flags = 0;
742 				break;
743 			}
744 		}
745 
746 		tdp->t_flags |= flags;
747 	}
748 
749 	debug(3, "die %llu <%llx>: array nelems %u size %u\n", off, off,
750 	    tdp->t_ardef->ad_nelems, tdp->t_size);
751 }
752 
753 /*ARGSUSED1*/
754 static int
die_array_resolve(tdesc_t * tdp,tdesc_t ** tdpp __unused,void * private)755 die_array_resolve(tdesc_t *tdp, tdesc_t **tdpp __unused, void *private)
756 {
757 	dwarf_t *dw = private;
758 	size_t sz;
759 
760 	if (tdp->t_flags & TDESC_F_RESOLVED)
761 		return (1);
762 
763 	debug(3, "trying to resolve array %d (cont %d)\n", tdp->t_id,
764 	    tdp->t_ardef->ad_contents->t_id);
765 
766 	if ((sz = tdesc_size(tdp->t_ardef->ad_contents)) == 0 &&
767 	    (tdp->t_ardef->ad_contents->t_flags & TDESC_F_RESOLVED) == 0) {
768 		debug(3, "unable to resolve array %s (%d) contents %d\n",
769 		    tdesc_name(tdp), tdp->t_id,
770 		    tdp->t_ardef->ad_contents->t_id);
771 
772 		dw->dw_nunres++;
773 		return (1);
774 	}
775 
776 	tdp->t_size = sz * tdp->t_ardef->ad_nelems;
777 	tdp->t_flags |= TDESC_F_RESOLVED;
778 
779 	debug(3, "resolved array %d: %u bytes\n", tdp->t_id, tdp->t_size);
780 
781 	return (1);
782 }
783 
784 /*ARGSUSED1*/
785 static int
die_array_failed(tdesc_t * tdp,tdesc_t ** tdpp __unused,void * private __unused)786 die_array_failed(tdesc_t *tdp, tdesc_t **tdpp __unused, void *private __unused)
787 {
788 	tdesc_t *cont = tdp->t_ardef->ad_contents;
789 
790 	if (tdp->t_flags & TDESC_F_RESOLVED)
791 		return (1);
792 
793 	fprintf(stderr, "Array %d: failed to size contents type %s (%d)\n",
794 	    tdp->t_id, tdesc_name(cont), cont->t_id);
795 
796 	return (1);
797 }
798 
799 /*
800  * Most enums (those with members) will be resolved during this first pass.
801  * Others - those without members (see the file comment) - won't be, and will
802  * need to wait until the second pass when they can be matched with their full
803  * definitions.
804  */
805 static void
die_enum_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)806 die_enum_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
807 {
808 	Dwarf_Die mem;
809 	Dwarf_Unsigned uval;
810 	Dwarf_Signed sval;
811 
812 	if (die_isdecl(dw, die)) {
813 		tdp->t_type = FORWARD;
814 		return;
815 	}
816 
817 	debug(3, "die %llu: creating enum\n", off);
818 
819 	tdp->t_type = ENUM;
820 
821 	(void) die_unsigned(dw, die, DW_AT_byte_size, &uval, DW_ATTR_REQ);
822 	tdp->t_size = uval;
823 
824 	if ((mem = die_child(dw, die)) != NULL) {
825 		elist_t **elastp = &tdp->t_emem;
826 
827 		do {
828 			elist_t *el;
829 
830 			if (die_tag(dw, mem) != DW_TAG_enumerator) {
831 				/* Nested type declaration */
832 				die_create_one(dw, mem);
833 				continue;
834 			}
835 
836 			el = xcalloc(sizeof (elist_t));
837 			el->el_name = die_name(dw, mem);
838 
839 			if (die_signed(dw, mem, DW_AT_const_value, &sval, 0)) {
840 				el->el_number = sval;
841 			} else if (die_unsigned(dw, mem, DW_AT_const_value,
842 			    &uval, 0)) {
843 				el->el_number = uval;
844 			} else {
845 				terminate("die %llu: enum %llu: member without "
846 				    "value\n", off, die_off(dw, mem));
847 			}
848 
849 			debug(3, "die %llu: enum %llu: created %s = %d\n", off,
850 			    die_off(dw, mem), el->el_name, el->el_number);
851 
852 			*elastp = el;
853 			elastp = &el->el_next;
854 
855 		} while ((mem = die_sibling(dw, mem)) != NULL);
856 
857 		hash_add(dw->dw_enumhash, tdp);
858 
859 		tdp->t_flags |= TDESC_F_RESOLVED;
860 
861 		if (tdp->t_name != NULL) {
862 			iidesc_t *ii = xcalloc(sizeof (iidesc_t));
863 			ii->ii_type = II_SOU;
864 			ii->ii_name = xstrdup(tdp->t_name);
865 			ii->ii_dtype = tdp;
866 
867 			iidesc_add(dw->dw_td->td_iihash, ii);
868 		}
869 	}
870 }
871 
872 static int
die_enum_match(void * arg1,void * arg2)873 die_enum_match(void *arg1, void *arg2)
874 {
875 	tdesc_t *tdp = arg1, **fullp = arg2;
876 
877 	if (tdp->t_emem != NULL) {
878 		*fullp = tdp;
879 		return (-1); /* stop the iteration */
880 	}
881 
882 	return (0);
883 }
884 
885 /*ARGSUSED1*/
886 static int
die_enum_resolve(tdesc_t * tdp,tdesc_t ** tdpp __unused,void * private)887 die_enum_resolve(tdesc_t *tdp, tdesc_t **tdpp __unused, void *private)
888 {
889 	dwarf_t *dw = private;
890 	tdesc_t *full = NULL;
891 
892 	if (tdp->t_flags & TDESC_F_RESOLVED)
893 		return (1);
894 
895 	(void) hash_find_iter(dw->dw_enumhash, tdp, die_enum_match, &full);
896 
897 	/*
898 	 * The answer to this one won't change from iteration to iteration,
899 	 * so don't even try.
900 	 */
901 	if (full == NULL) {
902 		terminate("tdp %u: enum %s has no members\n", tdp->t_id,
903 		    tdesc_name(tdp));
904 	}
905 
906 	debug(3, "tdp %u: enum %s redirected to %u\n", tdp->t_id,
907 	    tdesc_name(tdp), full->t_id);
908 
909 	tdp->t_flags |= TDESC_F_RESOLVED;
910 
911 	return (1);
912 }
913 
914 static int
die_fwd_map(void * arg1,void * arg2)915 die_fwd_map(void *arg1, void *arg2)
916 {
917 	tdesc_t *fwd = arg1, *sou = arg2;
918 
919 	debug(3, "tdp %u: mapped forward %s to sou %u\n", fwd->t_id,
920 	    tdesc_name(fwd), sou->t_id);
921 	fwd->t_tdesc = sou;
922 
923 	return (0);
924 }
925 
926 /*
927  * Structures and unions will never be resolved during the first pass, as we
928  * won't be able to fully determine the member sizes.  The second pass, which
929  * have access to sizing information, will be able to complete the resolution.
930  */
931 static void
die_sou_create(dwarf_t * dw,Dwarf_Die str,Dwarf_Off off,tdesc_t * tdp,int type,const char * typename)932 die_sou_create(dwarf_t *dw, Dwarf_Die str, Dwarf_Off off, tdesc_t *tdp,
933     int type, const char *typename)
934 {
935 	Dwarf_Unsigned sz, bitsz, bitoff;
936 #if BYTE_ORDER == _LITTLE_ENDIAN
937 	Dwarf_Unsigned bysz;
938 #endif
939 	Dwarf_Die mem;
940 	mlist_t *ml, **mlastp;
941 	iidesc_t *ii;
942 
943 	tdp->t_type = (die_isdecl(dw, str) ? FORWARD : type);
944 
945 	debug(3, "die %llu: creating %s %s\n", off,
946 	    (tdp->t_type == FORWARD ? "forward decl" : typename),
947 	    tdesc_name(tdp));
948 
949 	if (tdp->t_type == FORWARD) {
950 		hash_add(dw->dw_fwdhash, tdp);
951 		return;
952 	}
953 
954 	(void) hash_find_iter(dw->dw_fwdhash, tdp, die_fwd_map, tdp);
955 
956 	(void) die_unsigned(dw, str, DW_AT_byte_size, &sz, DW_ATTR_REQ);
957 	tdp->t_size = sz;
958 
959 	/*
960 	 * GCC allows empty SOUs as an extension.
961 	 */
962 	if ((mem = die_child(dw, str)) == NULL) {
963 		goto out;
964 	}
965 
966 	mlastp = &tdp->t_members;
967 
968 	do {
969 		Dwarf_Off memoff = die_off(dw, mem);
970 		Dwarf_Half tag = die_tag(dw, mem);
971 		Dwarf_Unsigned mloff;
972 
973 		if (tag != DW_TAG_member) {
974 			/* Nested type declaration */
975 			die_create_one(dw, mem);
976 			continue;
977 		}
978 
979 		debug(3, "die %llu: mem %llu: creating member\n", off, memoff);
980 
981 		ml = xcalloc(sizeof (mlist_t));
982 
983 		/*
984 		 * This could be a GCC anon struct/union member, so we'll allow
985 		 * an empty name, even though nothing can really handle them
986 		 * properly.  Note that some versions of GCC miss out debug
987 		 * info for anon structs, though recent versions are fixed (gcc
988 		 * bug 11816).
989 		 */
990 		if ((ml->ml_name = die_name(dw, mem)) == NULL)
991 			ml->ml_name = NULL;
992 
993 		ml->ml_type = die_lookup_pass1(dw, mem, DW_AT_type);
994 
995 		if (die_mem_offset(dw, mem, DW_AT_data_member_location,
996 		    &mloff, 0)) {
997 			debug(3, "die %llu: got mloff %llx\n", off,
998 			    (u_longlong_t)mloff);
999 			ml->ml_offset = mloff * 8;
1000 		}
1001 
1002 		if (die_unsigned(dw, mem, DW_AT_bit_size, &bitsz, 0))
1003 			ml->ml_size = bitsz;
1004 		else
1005 			ml->ml_size = tdesc_bitsize(ml->ml_type);
1006 
1007 		if (die_unsigned(dw, mem, DW_AT_data_bit_offset, &bitoff, 0)) {
1008 			ml->ml_offset += bitoff;
1009 		} else if (die_unsigned(dw, mem, DW_AT_bit_offset, &bitoff, 0)) {
1010 #if BYTE_ORDER == _BIG_ENDIAN
1011 			ml->ml_offset += bitoff;
1012 #else
1013 			/*
1014 			 * Note that Clang 3.4 will sometimes generate
1015 			 * member DIE before generating the DIE for the
1016 			 * member's type. The code can not handle this
1017 			 * properly so that tdesc_bitsize(ml->ml_type) will
1018 			 * return 0 because ml->ml_type is unknown. As a
1019 			 * result, a wrong member offset will be calculated.
1020 			 * To workaround this, we can instead try to
1021 			 * retrieve the value of DW_AT_byte_size attribute
1022 			 * which stores the byte size of the space occupied
1023 			 * by the type. If this attribute exists, its value
1024 			 * should equal to tdesc_bitsize(ml->ml_type)/NBBY.
1025 			 */
1026 			if (die_unsigned(dw, mem, DW_AT_byte_size, &bysz, 0) &&
1027 			    bysz > 0)
1028 				ml->ml_offset += bysz * NBBY - bitoff -
1029 				    ml->ml_size;
1030 			else
1031 				ml->ml_offset += tdesc_bitsize(ml->ml_type) -
1032 				    bitoff - ml->ml_size;
1033 #endif
1034 		}
1035 
1036 		debug(3, "die %llu: mem %llu: created \"%s\" (off %u sz %u)\n",
1037 		    off, memoff, ml->ml_name, ml->ml_offset, ml->ml_size);
1038 
1039 		*mlastp = ml;
1040 		mlastp = &ml->ml_next;
1041 	} while ((mem = die_sibling(dw, mem)) != NULL);
1042 
1043 	/*
1044 	 * GCC will attempt to eliminate unused types, thus decreasing the
1045 	 * size of the emitted dwarf.  That is, if you declare a foo_t in your
1046 	 * header, include said header in your source file, and neglect to
1047 	 * actually use (directly or indirectly) the foo_t in the source file,
1048 	 * the foo_t won't make it into the emitted DWARF.  So, at least, goes
1049 	 * the theory.
1050 	 *
1051 	 * Occasionally, it'll emit the DW_TAG_structure_type for the foo_t,
1052 	 * and then neglect to emit the members.  Strangely, the loner struct
1053 	 * tag will always be followed by a proper nested declaration of
1054 	 * something else.  This is clearly a bug, but we're not going to have
1055 	 * time to get it fixed before this goo goes back, so we'll have to work
1056 	 * around it.  If we see a no-membered struct with a nested declaration
1057 	 * (i.e. die_child of the struct tag won't be null), we'll ignore it.
1058 	 * Being paranoid, we won't simply remove it from the hash.  Instead,
1059 	 * we'll decline to create an iidesc for it, thus ensuring that this
1060 	 * type won't make it into the output file.  To be safe, we'll also
1061 	 * change the name.
1062 	 */
1063 	if (tdp->t_members == NULL) {
1064 		const char *old = tdesc_name(tdp);
1065 		size_t newsz = 7 + strlen(old) + 1;
1066 		char *new = xmalloc(newsz);
1067 		(void) snprintf(new, newsz, "orphan %s", old);
1068 
1069 		debug(3, "die %llu: worked around %s %s\n", off, typename, old);
1070 
1071 		if (tdp->t_name != NULL)
1072 			free(tdp->t_name);
1073 		tdp->t_name = new;
1074 		return;
1075 	}
1076 
1077 out:
1078 	if (tdp->t_name != NULL) {
1079 		ii = xcalloc(sizeof (iidesc_t));
1080 		ii->ii_type = II_SOU;
1081 		ii->ii_name = xstrdup(tdp->t_name);
1082 		ii->ii_dtype = tdp;
1083 
1084 		iidesc_add(dw->dw_td->td_iihash, ii);
1085 	}
1086 }
1087 
1088 static void
die_struct_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1089 die_struct_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1090 {
1091 	die_sou_create(dw, die, off, tdp, STRUCT, "struct");
1092 }
1093 
1094 static void
die_union_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1095 die_union_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1096 {
1097 	die_sou_create(dw, die, off, tdp, UNION, "union");
1098 }
1099 
1100 /*ARGSUSED1*/
1101 static int
die_sou_resolve(tdesc_t * tdp,tdesc_t ** tdpp __unused,void * private)1102 die_sou_resolve(tdesc_t *tdp, tdesc_t **tdpp __unused, void *private)
1103 {
1104 	dwarf_t *dw = private;
1105 	mlist_t *ml;
1106 	tdesc_t *mt;
1107 
1108 	if (tdp->t_flags & TDESC_F_RESOLVED)
1109 		return (1);
1110 
1111 	debug(3, "resolving sou %s\n", tdesc_name(tdp));
1112 
1113 	for (ml = tdp->t_members; ml != NULL; ml = ml->ml_next) {
1114 		if (ml->ml_size == 0) {
1115 			mt = tdesc_basetype(ml->ml_type);
1116 
1117 			if ((ml->ml_size = tdesc_bitsize(mt)) != 0)
1118 				continue;
1119 
1120 			/*
1121 			 * For empty members, or GCC/C99 flexible array
1122 			 * members, a size of 0 is correct. Structs and unions
1123 			 * consisting of flexible array members will also have
1124 			 * size 0.
1125 			 */
1126 			if (mt->t_members == NULL)
1127 				continue;
1128 			if (mt->t_type == ARRAY) {
1129 				if (mt->t_ardef->ad_nelems == 0)
1130 					continue;
1131 				mt = tdesc_basetype(mt->t_ardef->ad_contents);
1132 				if ((mt->t_flags & TDESC_F_RESOLVED) != 0 &&
1133 				    (mt->t_type == STRUCT ||
1134 				    mt->t_type == UNION) &&
1135 				    mt->t_members == NULL)
1136 					continue;
1137 			}
1138 			if ((mt->t_flags & TDESC_F_RESOLVED) != 0 &&
1139 			    (mt->t_type == STRUCT || mt->t_type == UNION))
1140 				continue;
1141 
1142 			dw->dw_nunres++;
1143 			return (1);
1144 		}
1145 
1146 		if ((mt = tdesc_basetype(ml->ml_type)) == NULL) {
1147 			dw->dw_nunres++;
1148 			return (1);
1149 		}
1150 
1151 		if (ml->ml_size != 0 && mt->t_type == INTRINSIC &&
1152 		    mt->t_intr->intr_nbits != ml->ml_size) {
1153 			/*
1154 			 * This member is a bitfield, and needs to reference
1155 			 * an intrinsic type with the same width.  If the
1156 			 * currently-referenced type isn't of the same width,
1157 			 * we'll copy it, adjusting the width of the copy to
1158 			 * the size we'd like.
1159 			 */
1160 			debug(3, "tdp %u: creating bitfield for %d bits\n",
1161 			    tdp->t_id, ml->ml_size);
1162 
1163 			ml->ml_type = tdesc_intr_clone(dw, mt, ml->ml_size,
1164 			    "bitfield");
1165 		}
1166 	}
1167 
1168 	tdp->t_flags |= TDESC_F_RESOLVED;
1169 
1170 	return (1);
1171 }
1172 
1173 /*ARGSUSED1*/
1174 static int
die_sou_failed(tdesc_t * tdp,tdesc_t ** tdpp __unused,void * private __unused)1175 die_sou_failed(tdesc_t *tdp, tdesc_t **tdpp __unused, void *private __unused)
1176 {
1177 	const char *typename = (tdp->t_type == STRUCT ? "struct" : "union");
1178 	mlist_t *ml;
1179 
1180 	if (tdp->t_flags & TDESC_F_RESOLVED)
1181 		return (1);
1182 
1183 	for (ml = tdp->t_members; ml != NULL; ml = ml->ml_next) {
1184 		if (ml->ml_size == 0) {
1185 			fprintf(stderr, "%s %d <%x>: failed to size member \"%s\" "
1186 			    "of type %s (%d <%x>)\n", typename, tdp->t_id,
1187 			    tdp->t_id,
1188 			    ml->ml_name, tdesc_name(ml->ml_type),
1189 			    ml->ml_type->t_id, ml->ml_type->t_id);
1190 		}
1191 	}
1192 
1193 	return (1);
1194 }
1195 
1196 static void
die_funcptr_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1197 die_funcptr_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1198 {
1199 	Dwarf_Attribute attr;
1200 	Dwarf_Half tag;
1201 	Dwarf_Die arg;
1202 	fndef_t *fn;
1203 	int i;
1204 
1205 	debug(3, "die %llu <%llx>: creating function pointer\n", off, off);
1206 
1207 	/*
1208 	 * We'll begin by processing any type definition nodes that may be
1209 	 * lurking underneath this one.
1210 	 */
1211 	for (arg = die_child(dw, die); arg != NULL;
1212 	    arg = die_sibling(dw, arg)) {
1213 		if ((tag = die_tag(dw, arg)) != DW_TAG_formal_parameter &&
1214 		    tag != DW_TAG_unspecified_parameters) {
1215 			/* Nested type declaration */
1216 			die_create_one(dw, arg);
1217 		}
1218 	}
1219 
1220 	if (die_isdecl(dw, die)) {
1221 		/*
1222 		 * This is a prototype.  We don't add prototypes to the
1223 		 * tree, so we're going to drop the tdesc.  Unfortunately,
1224 		 * it has already been added to the tree.  Nobody will reference
1225 		 * it, though, and it will be leaked.
1226 		 */
1227 		return;
1228 	}
1229 
1230 	fn = xcalloc(sizeof (fndef_t));
1231 
1232 	tdp->t_type = FUNCTION;
1233 
1234 	if ((attr = die_attr(dw, die, DW_AT_type, 0)) != NULL) {
1235 		fn->fn_ret = die_lookup_pass1(dw, die, DW_AT_type);
1236 	} else {
1237 		fn->fn_ret = tdesc_intr_void(dw);
1238 	}
1239 
1240 	/*
1241 	 * Count the arguments to the function, then read them in.
1242 	 */
1243 	for (fn->fn_nargs = 0, arg = die_child(dw, die); arg != NULL;
1244 	    arg = die_sibling(dw, arg)) {
1245 		if ((tag = die_tag(dw, arg)) == DW_TAG_formal_parameter)
1246 			fn->fn_nargs++;
1247 		else if (tag == DW_TAG_unspecified_parameters &&
1248 		    fn->fn_nargs > 0)
1249 			fn->fn_vargs = 1;
1250 	}
1251 
1252 	if (fn->fn_nargs != 0) {
1253 		debug(3, "die %llu: adding %d argument%s\n", off, fn->fn_nargs,
1254 		    (fn->fn_nargs > 1 ? "s" : ""));
1255 
1256 		fn->fn_args = xcalloc(sizeof (tdesc_t *) * fn->fn_nargs);
1257 		for (i = 0, arg = die_child(dw, die);
1258 		    arg != NULL && i < (int) fn->fn_nargs;
1259 		    arg = die_sibling(dw, arg)) {
1260 			if (die_tag(dw, arg) != DW_TAG_formal_parameter)
1261 				continue;
1262 
1263 			fn->fn_args[i++] = die_lookup_pass1(dw, arg,
1264 			    DW_AT_type);
1265 		}
1266 	}
1267 
1268 	tdp->t_fndef = fn;
1269 	tdp->t_flags |= TDESC_F_RESOLVED;
1270 }
1271 
1272 /*
1273  * GCC and DevPro use different names for the base types.  While the terms are
1274  * the same, they are arranged in a different order.  Some terms, such as int,
1275  * are implied in one, and explicitly named in the other.  Given a base type
1276  * as input, this routine will return a common name, along with an intr_t
1277  * that reflects said name.
1278  */
1279 static intr_t *
die_base_name_parse(const char * name,char ** newp)1280 die_base_name_parse(const char *name, char **newp)
1281 {
1282 	char buf[256];
1283 	char const *base;
1284 	char *c;
1285 	int nlong = 0, nshort = 0, nchar = 0, nint = 0;
1286 	int sign = 1;
1287 	char fmt = '\0';
1288 	intr_t *intr;
1289 
1290 	if (strlen(name) > sizeof (buf) - 1)
1291 		terminate("base type name \"%s\" is too long\n", name);
1292 
1293 	strncpy(buf, name, sizeof (buf));
1294 
1295 	for (c = strtok(buf, " "); c != NULL; c = strtok(NULL, " ")) {
1296 		if (strcmp(c, "signed") == 0)
1297 			sign = 1;
1298 		else if (strcmp(c, "unsigned") == 0)
1299 			sign = 0;
1300 		else if (strcmp(c, "long") == 0)
1301 			nlong++;
1302 		else if (strcmp(c, "char") == 0) {
1303 			nchar++;
1304 			fmt = 'c';
1305 		} else if (strcmp(c, "short") == 0)
1306 			nshort++;
1307 		else if (strcmp(c, "int") == 0)
1308 			nint++;
1309 		else {
1310 			/*
1311 			 * If we don't recognize any of the tokens, we'll tell
1312 			 * the caller to fall back to the dwarf-provided
1313 			 * encoding information.
1314 			 */
1315 			return (NULL);
1316 		}
1317 	}
1318 
1319 	if (nchar > 1 || nshort > 1 || nint > 1 || nlong > 2)
1320 		return (NULL);
1321 
1322 	if (nchar > 0) {
1323 		if (nlong > 0 || nshort > 0 || nint > 0)
1324 			return (NULL);
1325 
1326 		base = "char";
1327 
1328 	} else if (nshort > 0) {
1329 		if (nlong > 0)
1330 			return (NULL);
1331 
1332 		base = "short";
1333 
1334 	} else if (nlong > 0) {
1335 		base = "long";
1336 
1337 	} else {
1338 		base = "int";
1339 	}
1340 
1341 	intr = xcalloc(sizeof (intr_t));
1342 	intr->intr_type = INTR_INT;
1343 	intr->intr_signed = sign;
1344 	intr->intr_iformat = fmt;
1345 
1346 	snprintf(buf, sizeof (buf), "%s%s%s",
1347 	    (sign ? "" : "unsigned "),
1348 	    (nlong > 1 ? "long " : ""),
1349 	    base);
1350 
1351 	*newp = xstrdup(buf);
1352 	return (intr);
1353 }
1354 
1355 typedef struct fp_size_map {
1356 	size_t fsm_typesz[2];	/* size of {32,64} type */
1357 	uint_t fsm_enc[3];	/* CTF_FP_* for {bare,cplx,imagry} type */
1358 } fp_size_map_t;
1359 
1360 static const fp_size_map_t fp_encodings[] = {
1361 	{ { 4, 4 }, { CTF_FP_SINGLE, CTF_FP_CPLX, CTF_FP_IMAGRY } },
1362 	{ { 8, 8 }, { CTF_FP_DOUBLE, CTF_FP_DCPLX, CTF_FP_DIMAGRY } },
1363 	{ { 12, 16 }, { CTF_FP_LDOUBLE, CTF_FP_LDCPLX, CTF_FP_LDIMAGRY } },
1364 	{ { 0, 0 }, { 0, 0, 0 } }
1365 };
1366 
1367 static uint_t
die_base_type2enc(dwarf_t * dw,Dwarf_Off off,Dwarf_Unsigned enc,size_t sz)1368 die_base_type2enc(dwarf_t *dw, Dwarf_Off off, Dwarf_Unsigned enc, size_t sz)
1369 {
1370 	const fp_size_map_t *map = fp_encodings;
1371 	uint_t szidx = dw->dw_ptrsz == sizeof (uint64_t);
1372 	uint_t mult = 1, col = 0;
1373 
1374 	if (enc == DW_ATE_complex_float) {
1375 		mult = 2;
1376 		col = 1;
1377 	} else if (enc == DW_ATE_imaginary_float
1378 #ifdef illumos
1379 	    || enc == DW_ATE_SUN_imaginary_float
1380 #endif
1381 	    )
1382 		col = 2;
1383 
1384 	while (map->fsm_typesz[szidx] != 0) {
1385 		if (map->fsm_typesz[szidx] * mult == sz)
1386 			return (map->fsm_enc[col]);
1387 		map++;
1388 	}
1389 
1390 	terminate("die %llu: unrecognized real type size %u\n", off, sz);
1391 	/*NOTREACHED*/
1392 	return (0);
1393 }
1394 
1395 static intr_t *
die_base_from_dwarf(dwarf_t * dw,Dwarf_Die base,Dwarf_Off off,size_t sz)1396 die_base_from_dwarf(dwarf_t *dw, Dwarf_Die base, Dwarf_Off off, size_t sz)
1397 {
1398 	intr_t *intr = xcalloc(sizeof (intr_t));
1399 	Dwarf_Unsigned enc;
1400 
1401 	(void) die_unsigned(dw, base, DW_AT_encoding, &enc, DW_ATTR_REQ);
1402 
1403 	switch (enc) {
1404 	case DW_ATE_unsigned:
1405 	case DW_ATE_address:
1406 		intr->intr_type = INTR_INT;
1407 		break;
1408 	case DW_ATE_unsigned_char:
1409 		intr->intr_type = INTR_INT;
1410 		intr->intr_iformat = 'c';
1411 		break;
1412 	case DW_ATE_signed:
1413 		intr->intr_type = INTR_INT;
1414 		intr->intr_signed = 1;
1415 		break;
1416 	case DW_ATE_signed_char:
1417 		intr->intr_type = INTR_INT;
1418 		intr->intr_signed = 1;
1419 		intr->intr_iformat = 'c';
1420 		break;
1421 	case DW_ATE_boolean:
1422 		intr->intr_type = INTR_INT;
1423 		intr->intr_signed = 1;
1424 		intr->intr_iformat = 'b';
1425 		break;
1426 	case DW_ATE_float:
1427 	case DW_ATE_complex_float:
1428 	case DW_ATE_imaginary_float:
1429 #ifdef illumos
1430 	case DW_ATE_SUN_imaginary_float:
1431 	case DW_ATE_SUN_interval_float:
1432 #endif
1433 		intr->intr_type = INTR_REAL;
1434 		intr->intr_signed = 1;
1435 		intr->intr_fformat = die_base_type2enc(dw, off, enc, sz);
1436 		break;
1437 	default:
1438 		terminate("die %llu: unknown base type encoding 0x%llx\n",
1439 		    off, enc);
1440 	}
1441 
1442 	return (intr);
1443 }
1444 
1445 static void
die_base_create(dwarf_t * dw,Dwarf_Die base,Dwarf_Off off,tdesc_t * tdp)1446 die_base_create(dwarf_t *dw, Dwarf_Die base, Dwarf_Off off, tdesc_t *tdp)
1447 {
1448 	Dwarf_Unsigned sz;
1449 	intr_t *intr;
1450 	char *new;
1451 
1452 	debug(3, "die %llu: creating base type\n", off);
1453 
1454 	/*
1455 	 * The compilers have their own clever (internally inconsistent) ideas
1456 	 * as to what base types should look like.  Some times gcc will, for
1457 	 * example, use DW_ATE_signed_char for char.  Other times, however, it
1458 	 * will use DW_ATE_signed.  Needless to say, this causes some problems
1459 	 * down the road, particularly with merging.  We do, however, use the
1460 	 * DWARF idea of type sizes, as this allows us to avoid caring about
1461 	 * the data model.
1462 	 */
1463 	(void) die_unsigned(dw, base, DW_AT_byte_size, &sz, DW_ATTR_REQ);
1464 
1465 	if (tdp->t_name == NULL)
1466 		terminate("die %llu: base type without name\n", off);
1467 
1468 	/* XXX make a name parser for float too */
1469 	if ((intr = die_base_name_parse(tdp->t_name, &new)) != NULL) {
1470 		/* Found it.  We'll use the parsed version */
1471 		debug(3, "die %llu: name \"%s\" remapped to \"%s\"\n", off,
1472 		    tdesc_name(tdp), new);
1473 
1474 		free(tdp->t_name);
1475 		tdp->t_name = new;
1476 	} else {
1477 		/*
1478 		 * We didn't recognize the type, so we'll create an intr_t
1479 		 * based on the DWARF data.
1480 		 */
1481 		debug(3, "die %llu: using dwarf data for base \"%s\"\n", off,
1482 		    tdesc_name(tdp));
1483 
1484 		intr = die_base_from_dwarf(dw, base, off, sz);
1485 	}
1486 
1487 	intr->intr_nbits = sz * 8;
1488 
1489 	tdp->t_type = INTRINSIC;
1490 	tdp->t_intr = intr;
1491 	tdp->t_size = sz;
1492 
1493 	tdp->t_flags |= TDESC_F_RESOLVED;
1494 }
1495 
1496 static void
die_through_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp,int type,const char * typename)1497 die_through_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp,
1498     int type, const char *typename)
1499 {
1500 	Dwarf_Attribute attr;
1501 
1502 	debug(3, "die %llu <%llx>: creating %s type %d\n", off, off, typename, type);
1503 
1504 	tdp->t_type = type;
1505 
1506 	if ((attr = die_attr(dw, die, DW_AT_type, 0)) != NULL) {
1507 		tdp->t_tdesc = die_lookup_pass1(dw, die, DW_AT_type);
1508 	} else {
1509 		tdp->t_tdesc = tdesc_intr_void(dw);
1510 	}
1511 
1512 	if (type == POINTER)
1513 		tdp->t_size = dw->dw_ptrsz;
1514 
1515 	tdp->t_flags |= TDESC_F_RESOLVED;
1516 
1517 	if (type == TYPEDEF) {
1518 		iidesc_t *ii = xcalloc(sizeof (iidesc_t));
1519 		ii->ii_type = II_TYPE;
1520 		ii->ii_name = xstrdup(tdp->t_name);
1521 		ii->ii_dtype = tdp;
1522 
1523 		iidesc_add(dw->dw_td->td_iihash, ii);
1524 	}
1525 }
1526 
1527 static void
die_typedef_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1528 die_typedef_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1529 {
1530 	die_through_create(dw, die, off, tdp, TYPEDEF, "typedef");
1531 }
1532 
1533 static void
die_const_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1534 die_const_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1535 {
1536 	die_through_create(dw, die, off, tdp, CONST, "const");
1537 }
1538 
1539 static void
die_pointer_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1540 die_pointer_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1541 {
1542 	die_through_create(dw, die, off, tdp, POINTER, "pointer");
1543 }
1544 
1545 static void
die_restrict_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1546 die_restrict_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1547 {
1548 	die_through_create(dw, die, off, tdp, RESTRICT, "restrict");
1549 }
1550 
1551 static void
die_volatile_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp)1552 die_volatile_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp)
1553 {
1554 	die_through_create(dw, die, off, tdp, VOLATILE, "volatile");
1555 }
1556 
1557 /*ARGSUSED3*/
1558 static void
die_function_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp __unused)1559 die_function_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp __unused)
1560 {
1561 	Dwarf_Die arg;
1562 	Dwarf_Half tag;
1563 	iidesc_t *ii;
1564 	char *name;
1565 
1566 	debug(3, "die %llu <%llx>: creating function definition\n", off, off);
1567 
1568 	/*
1569 	 * We'll begin by processing any type definition nodes that may be
1570 	 * lurking underneath this one.
1571 	 */
1572 	for (arg = die_child(dw, die); arg != NULL;
1573 	    arg = die_sibling(dw, arg)) {
1574 		if ((tag = die_tag(dw, arg)) != DW_TAG_formal_parameter &&
1575 		    tag != DW_TAG_variable) {
1576 			/* Nested type declaration */
1577 			die_create_one(dw, arg);
1578 		}
1579 	}
1580 
1581 	if (die_isdecl(dw, die) || (name = die_name(dw, die)) == NULL) {
1582 		/*
1583 		 * We process neither prototypes nor subprograms without
1584 		 * names.
1585 		 */
1586 		return;
1587 	}
1588 
1589 	ii = xcalloc(sizeof (iidesc_t));
1590 	ii->ii_type = die_isglobal(dw, die) ? II_GFUN : II_SFUN;
1591 	ii->ii_name = name;
1592 	if (ii->ii_type == II_SFUN)
1593 		ii->ii_owner = xstrdup(dw->dw_cuname);
1594 
1595 	debug(3, "die %llu: function %s is %s\n", off, ii->ii_name,
1596 	    (ii->ii_type == II_GFUN ? "global" : "static"));
1597 
1598 	if (die_attr(dw, die, DW_AT_type, 0) != NULL)
1599 		ii->ii_dtype = die_lookup_pass1(dw, die, DW_AT_type);
1600 	else
1601 		ii->ii_dtype = tdesc_intr_void(dw);
1602 
1603 	for (arg = die_child(dw, die); arg != NULL;
1604 	    arg = die_sibling(dw, arg)) {
1605 		char *name1;
1606 
1607 		debug(3, "die %llu: looking at sub member at %llu\n",
1608 		    off, die_off(dw, die));
1609 
1610 		if (die_tag(dw, arg) != DW_TAG_formal_parameter)
1611 			continue;
1612 
1613 		if ((name1 = die_name(dw, arg)) == NULL) {
1614 			terminate("die %llu: func arg %d has no name\n",
1615 			    off, ii->ii_nargs + 1);
1616 		}
1617 
1618 		if (strcmp(name1, "...") == 0) {
1619 			free(name1);
1620 			ii->ii_vargs = 1;
1621 			continue;
1622 		}
1623 		free(name1);
1624 
1625 		ii->ii_nargs++;
1626 	}
1627 
1628 	if (ii->ii_nargs > 0) {
1629 		int i;
1630 
1631 		debug(3, "die %llu: function has %d argument%s\n", off,
1632 		    ii->ii_nargs, (ii->ii_nargs == 1 ? "" : "s"));
1633 
1634 		ii->ii_args = xcalloc(sizeof (tdesc_t) * ii->ii_nargs);
1635 
1636 		for (arg = die_child(dw, die), i = 0;
1637 		    arg != NULL && i < ii->ii_nargs;
1638 		    arg = die_sibling(dw, arg)) {
1639 			if (die_tag(dw, arg) != DW_TAG_formal_parameter)
1640 				continue;
1641 
1642 			ii->ii_args[i++] = die_lookup_pass1(dw, arg,
1643 			    DW_AT_type);
1644 		}
1645 	}
1646 
1647 	iidesc_add(dw->dw_td->td_iihash, ii);
1648 }
1649 
1650 /*ARGSUSED3*/
1651 static void
die_variable_create(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off,tdesc_t * tdp __unused)1652 die_variable_create(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off, tdesc_t *tdp __unused)
1653 {
1654 	iidesc_t *ii;
1655 	char *name;
1656 
1657 	debug(3, "die %llu: creating object definition\n", off);
1658 
1659 	if (die_isdecl(dw, die) || (name = die_name(dw, die)) == NULL)
1660 		return; /* skip prototypes and nameless objects */
1661 
1662 	ii = xcalloc(sizeof (iidesc_t));
1663 	ii->ii_type = die_isglobal(dw, die) ? II_GVAR : II_SVAR;
1664 	ii->ii_name = name;
1665 	ii->ii_dtype = die_lookup_pass1(dw, die, DW_AT_type);
1666 	if (ii->ii_type == II_SVAR)
1667 		ii->ii_owner = xstrdup(dw->dw_cuname);
1668 
1669 	iidesc_add(dw->dw_td->td_iihash, ii);
1670 }
1671 
1672 /*ARGSUSED2*/
1673 static int
die_fwd_resolve(tdesc_t * fwd,tdesc_t ** fwdp,void * private __unused)1674 die_fwd_resolve(tdesc_t *fwd, tdesc_t **fwdp, void *private __unused)
1675 {
1676 	if (fwd->t_flags & TDESC_F_RESOLVED)
1677 		return (1);
1678 
1679 	if (fwd->t_tdesc != NULL) {
1680 		debug(3, "tdp %u: unforwarded %s\n", fwd->t_id,
1681 		    tdesc_name(fwd));
1682 		*fwdp = fwd->t_tdesc;
1683 	}
1684 
1685 	fwd->t_flags |= TDESC_F_RESOLVED;
1686 
1687 	return (1);
1688 }
1689 
1690 /*ARGSUSED*/
1691 static void
die_lexblk_descend(dwarf_t * dw,Dwarf_Die die,Dwarf_Off off __unused,tdesc_t * tdp __unused)1692 die_lexblk_descend(dwarf_t *dw, Dwarf_Die die, Dwarf_Off off __unused, tdesc_t *tdp __unused)
1693 {
1694 	Dwarf_Die child = die_child(dw, die);
1695 
1696 	if (child != NULL)
1697 		die_create(dw, child);
1698 }
1699 
1700 /*
1701  * Used to map the die to a routine which can parse it, using the tag to do the
1702  * mapping.  While the processing of most tags entails the creation of a tdesc,
1703  * there are a few which don't - primarily those which result in the creation of
1704  * iidescs which refer to existing tdescs.
1705  */
1706 
1707 #define	DW_F_NOTDP	0x1	/* Don't create a tdesc for the creator */
1708 
1709 typedef struct die_creator {
1710 	Dwarf_Half dc_tag;
1711 	uint16_t dc_flags;
1712 	void (*dc_create)(dwarf_t *, Dwarf_Die, Dwarf_Off, tdesc_t *);
1713 } die_creator_t;
1714 
1715 static const die_creator_t die_creators[] = {
1716 	{ DW_TAG_array_type,		0,		die_array_create },
1717 	{ DW_TAG_enumeration_type,	0,		die_enum_create },
1718 	{ DW_TAG_lexical_block,		DW_F_NOTDP,	die_lexblk_descend },
1719 	{ DW_TAG_pointer_type,		0,		die_pointer_create },
1720 	{ DW_TAG_structure_type,	0,		die_struct_create },
1721 	{ DW_TAG_subroutine_type,	0,		die_funcptr_create },
1722 	{ DW_TAG_typedef,		0,		die_typedef_create },
1723 	{ DW_TAG_union_type,		0,		die_union_create },
1724 	{ DW_TAG_base_type,		0,		die_base_create },
1725 	{ DW_TAG_const_type,		0,		die_const_create },
1726 	{ DW_TAG_subprogram,		DW_F_NOTDP,	die_function_create },
1727 	{ DW_TAG_variable,		DW_F_NOTDP,	die_variable_create },
1728 	{ DW_TAG_volatile_type,		0,		die_volatile_create },
1729 	{ DW_TAG_restrict_type,		0,		die_restrict_create },
1730 	{ 0, 0, NULL }
1731 };
1732 
1733 static const die_creator_t *
die_tag2ctor(Dwarf_Half tag)1734 die_tag2ctor(Dwarf_Half tag)
1735 {
1736 	const die_creator_t *dc;
1737 
1738 	for (dc = die_creators; dc->dc_create != NULL; dc++) {
1739 		if (dc->dc_tag == tag)
1740 			return (dc);
1741 	}
1742 
1743 	return (NULL);
1744 }
1745 
1746 static void
die_create_one(dwarf_t * dw,Dwarf_Die die)1747 die_create_one(dwarf_t *dw, Dwarf_Die die)
1748 {
1749 	Dwarf_Off off = die_off(dw, die);
1750 	const die_creator_t *dc;
1751 	Dwarf_Half tag;
1752 	tdesc_t *tdp;
1753 
1754 	debug(3, "die %llu <%llx>: create_one\n", off, off);
1755 
1756 	if (off > dw->dw_maxoff) {
1757 		terminate("illegal die offset %llu (max %llu)\n", off,
1758 		    dw->dw_maxoff);
1759 	}
1760 
1761 	tag = die_tag(dw, die);
1762 
1763 	if ((dc = die_tag2ctor(tag)) == NULL) {
1764 		debug(2, "die %llu: ignoring tag type %x\n", off, tag);
1765 		return;
1766 	}
1767 
1768 	if ((tdp = tdesc_lookup(dw, off)) == NULL &&
1769 	    !(dc->dc_flags & DW_F_NOTDP)) {
1770 		tdp = xcalloc(sizeof (tdesc_t));
1771 		tdp->t_id = off;
1772 		tdesc_add(dw, tdp);
1773 	}
1774 
1775 	if (tdp != NULL)
1776 		tdp->t_name = die_name(dw, die);
1777 
1778 	dc->dc_create(dw, die, off, tdp);
1779 }
1780 
1781 static void
die_create(dwarf_t * dw,Dwarf_Die die)1782 die_create(dwarf_t *dw, Dwarf_Die die)
1783 {
1784 	do {
1785 		die_create_one(dw, die);
1786 	} while ((die = die_sibling(dw, die)) != NULL);
1787 }
1788 
1789 static tdtrav_cb_f die_resolvers[] = {
1790 	NULL,
1791 	NULL,			/* intrinsic */
1792 	NULL,			/* pointer */
1793 	die_array_resolve,	/* array */
1794 	NULL,			/* function */
1795 	die_sou_resolve,	/* struct */
1796 	die_sou_resolve,	/* union */
1797 	die_enum_resolve,	/* enum */
1798 	die_fwd_resolve,	/* forward */
1799 	NULL,			/* typedef */
1800 	NULL,			/* typedef unres */
1801 	NULL,			/* volatile */
1802 	NULL,			/* const */
1803 	NULL,			/* restrict */
1804 };
1805 
1806 static tdtrav_cb_f die_fail_reporters[] = {
1807 	NULL,
1808 	NULL,			/* intrinsic */
1809 	NULL,			/* pointer */
1810 	die_array_failed,	/* array */
1811 	NULL,			/* function */
1812 	die_sou_failed,		/* struct */
1813 	die_sou_failed,		/* union */
1814 	NULL,			/* enum */
1815 	NULL,			/* forward */
1816 	NULL,			/* typedef */
1817 	NULL,			/* typedef unres */
1818 	NULL,			/* volatile */
1819 	NULL,			/* const */
1820 	NULL,			/* restrict */
1821 };
1822 
1823 static void
die_resolve(dwarf_t * dw)1824 die_resolve(dwarf_t *dw)
1825 {
1826 	int last = -1;
1827 	int pass = 0;
1828 
1829 	do {
1830 		pass++;
1831 		dw->dw_nunres = 0;
1832 
1833 		(void) iitraverse_hash(dw->dw_td->td_iihash,
1834 		    &dw->dw_td->td_curvgen, NULL, NULL, die_resolvers, dw);
1835 
1836 		debug(3, "resolve: pass %d, %u left\n", pass, dw->dw_nunres);
1837 
1838 		if ((int) dw->dw_nunres == last) {
1839 			fprintf(stderr, "%s: failed to resolve the following "
1840 			    "types:\n", progname);
1841 
1842 			(void) iitraverse_hash(dw->dw_td->td_iihash,
1843 			    &dw->dw_td->td_curvgen, NULL, NULL,
1844 			    die_fail_reporters, dw);
1845 
1846 			terminate("failed to resolve types\n");
1847 		}
1848 
1849 		last = dw->dw_nunres;
1850 
1851 	} while (dw->dw_nunres != 0);
1852 }
1853 
1854 /*
1855  * Any object containing a function or object symbol at any scope should also
1856  * contain DWARF data.
1857  */
1858 static boolean_t
should_have_dwarf(Elf * elf)1859 should_have_dwarf(Elf *elf)
1860 {
1861 	Elf_Scn *scn = NULL;
1862 	Elf_Data *data = NULL;
1863 	GElf_Shdr shdr;
1864 	GElf_Sym sym;
1865 	uint32_t symdx = 0;
1866 	size_t nsyms = 0;
1867 	boolean_t found = B_FALSE;
1868 
1869 	while ((scn = elf_nextscn(elf, scn)) != NULL) {
1870 		gelf_getshdr(scn, &shdr);
1871 
1872 		if (shdr.sh_type == SHT_SYMTAB) {
1873 			found = B_TRUE;
1874 			break;
1875 		}
1876 	}
1877 
1878 	if (!found)
1879 		terminate("cannot convert stripped objects\n");
1880 
1881 	data = elf_getdata(scn, NULL);
1882 	nsyms = shdr.sh_size / shdr.sh_entsize;
1883 
1884 	for (symdx = 0; symdx < nsyms; symdx++) {
1885 		gelf_getsym(data, symdx, &sym);
1886 
1887 		if ((GELF_ST_TYPE(sym.st_info) == STT_FUNC) ||
1888 		    (GELF_ST_TYPE(sym.st_info) == STT_TLS) ||
1889 		    (GELF_ST_TYPE(sym.st_info) == STT_OBJECT)) {
1890 			char *name;
1891 
1892 			name = elf_strptr(elf, shdr.sh_link, sym.st_name);
1893 
1894 			/* Studio emits these local symbols regardless */
1895 			if ((strcmp(name, "Bbss.bss") != 0) &&
1896 			    (strcmp(name, "Ttbss.bss") != 0) &&
1897 			    (strcmp(name, "Ddata.data") != 0) &&
1898 			    (strcmp(name, "Ttdata.data") != 0) &&
1899 			    (strcmp(name, "Drodata.rodata") != 0))
1900 				return (B_TRUE);
1901 		}
1902 	}
1903 
1904 	return (B_FALSE);
1905 }
1906 
1907 /*ARGSUSED*/
1908 int
dw_read(tdata_t * td,Elf * elf,char * filename __unused)1909 dw_read(tdata_t *td, Elf *elf, char *filename __unused)
1910 {
1911 	Dwarf_Unsigned abboff, hdrlen, lang, nxthdr;
1912 	Dwarf_Half vers, addrsz, offsz;
1913 	Dwarf_Die cu = 0;
1914 	Dwarf_Die child = 0;
1915 	dwarf_t dw;
1916 	char *prod = NULL;
1917 	int rc;
1918 
1919 	bzero(&dw, sizeof (dwarf_t));
1920 	dw.dw_td = td;
1921 	dw.dw_ptrsz = elf_ptrsz(elf);
1922 	dw.dw_mfgtid_last = TID_MFGTID_BASE;
1923 	dw.dw_tidhash = hash_new(TDESC_HASH_BUCKETS, tdesc_idhash, tdesc_idcmp);
1924 	dw.dw_fwdhash = hash_new(TDESC_HASH_BUCKETS, tdesc_namehash,
1925 	    tdesc_namecmp);
1926 	dw.dw_enumhash = hash_new(TDESC_HASH_BUCKETS, tdesc_namehash,
1927 	    tdesc_namecmp);
1928 
1929 	if ((rc = dwarf_elf_init(elf, DW_DLC_READ, NULL, NULL, &dw.dw_dw,
1930 	    &dw.dw_err)) == DW_DLV_NO_ENTRY) {
1931 		if (should_have_dwarf(elf)) {
1932 			errno = ENOENT;
1933 			return (-1);
1934 		} else {
1935 			return (0);
1936 		}
1937 	} else if (rc != DW_DLV_OK) {
1938 		if (dwarf_errno(dw.dw_err) == DW_DLE_DEBUG_INFO_NULL) {
1939 			/*
1940 			 * There's no type data in the DWARF section, but
1941 			 * libdwarf is too clever to handle that properly.
1942 			 */
1943 			return (0);
1944 		}
1945 
1946 		terminate("failed to initialize DWARF: %s\n",
1947 		    dwarf_errmsg(dw.dw_err));
1948 	}
1949 
1950 	if ((rc = dwarf_next_cu_header_b(dw.dw_dw, &hdrlen, &vers, &abboff,
1951 	    &addrsz, &offsz, NULL, &nxthdr, &dw.dw_err)) != DW_DLV_OK) {
1952 		if (dw.dw_err.err_error == DW_DLE_NO_ENTRY)
1953 			exit(0);
1954 		else
1955 			terminate("rc = %d %s\n", rc, dwarf_errmsg(dw.dw_err));
1956 	}
1957 	if ((cu = die_sibling(&dw, NULL)) == NULL ||
1958 	    (((child = die_child(&dw, cu)) == NULL) &&
1959 	    should_have_dwarf(elf))) {
1960 		terminate("file does not contain dwarf type data "
1961 		    "(try compiling with -g)\n");
1962 	} else if (child == NULL) {
1963 		return (0);
1964 	}
1965 
1966 	dw.dw_maxoff = nxthdr - 1;
1967 
1968 	if (dw.dw_maxoff > TID_FILEMAX)
1969 		terminate("file contains too many types\n");
1970 
1971 	debug(1, "DWARF version: %d\n", vers);
1972 	if (vers < 2 || vers > 4) {
1973 		terminate("file contains incompatible version %d DWARF code "
1974 		    "(version 2, 3 or 4 required)\n", vers);
1975 	}
1976 
1977 	if (die_string(&dw, cu, DW_AT_producer, &prod, 0)) {
1978 		debug(1, "DWARF emitter: %s\n", prod);
1979 		free(prod);
1980 	}
1981 
1982 	if (dwarf_attrval_unsigned(cu, DW_AT_language, &lang, &dw.dw_err) == 0)
1983 		switch (lang) {
1984 		case DW_LANG_C:
1985 		case DW_LANG_C89:
1986 		case DW_LANG_C99:
1987 		case DW_LANG_C11:
1988 		case DW_LANG_C_plus_plus:
1989 		case DW_LANG_C_plus_plus_03:
1990 		case DW_LANG_C_plus_plus_11:
1991 		case DW_LANG_C_plus_plus_14:
1992 		case DW_LANG_Mips_Assembler:
1993 			break;
1994 		default:
1995 			terminate("file contains DWARF for unsupported "
1996 			    "language %#x", lang);
1997 		}
1998 	else
1999 		warning("die %llu: failed to get language attribute: %s\n",
2000 		    die_off(&dw, cu), dwarf_errmsg(dw.dw_err));
2001 
2002 	if ((dw.dw_cuname = die_name(&dw, cu)) != NULL) {
2003 		char *base = xstrdup(basename(dw.dw_cuname));
2004 		free(dw.dw_cuname);
2005 		dw.dw_cuname = base;
2006 
2007 		debug(1, "CU name: %s\n", dw.dw_cuname);
2008 	}
2009 
2010 	if ((child = die_child(&dw, cu)) != NULL)
2011 		die_create(&dw, child);
2012 
2013 	if ((rc = dwarf_next_cu_header_b(dw.dw_dw, &hdrlen, &vers, &abboff,
2014 	    &addrsz, &offsz, NULL, &nxthdr, &dw.dw_err)) != DW_DLV_NO_ENTRY)
2015 		terminate("multiple compilation units not supported\n");
2016 
2017 	(void) dwarf_finish(dw.dw_dw, &dw.dw_err);
2018 
2019 	die_resolve(&dw);
2020 
2021 	cvt_fixups(td, dw.dw_ptrsz);
2022 
2023 	/* leak the dwarf_t */
2024 
2025 	return (0);
2026 }
2027