xref: /netbsd-src/external/cddl/osnet/dist/tools/ctf/cvt/ctftools.h (revision 413d532bcc3f62d122e56d92e13ac64825a40baf)
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 2006 Sun Microsystems, Inc.  All rights reserved.
23  * Use is subject to license terms.
24  */
25 
26 #ifndef _CTFTOOLS_H
27 #define	_CTFTOOLS_H
28 
29 #pragma ident	"%Z%%M%	%I%	%E% SMI"
30 
31 /*
32  * Functions and data structures used in the manipulation of stabs and CTF data
33  */
34 
35 #include <stdio.h>
36 #include <stdlib.h>
37 #include <stdarg.h>
38 #include <libelf.h>
39 #include <gelf.h>
40 #include <pthread.h>
41 
42 #ifdef __cplusplus
43 extern "C" {
44 #endif
45 
46 #include "list.h"
47 #include "hash.h"
48 
49 #ifndef DEBUG_LEVEL
50 #define	DEBUG_LEVEL 0
51 #endif
52 #ifndef DEBUG_PARSE
53 #define	DEBUG_PARSE 0
54 #endif
55 
56 #ifndef DEBUG_STREAM
57 #define	DEBUG_STREAM stderr
58 #endif
59 
60 #ifndef MAX
61 #define	MAX(a, b) 		((a) < (b) ? (b) : (a))
62 #endif
63 
64 #ifndef MIN
65 #define	MIN(a, b) 		((a) > (b) ? (b) : (a))
66 #endif
67 
68 #ifndef TRUE
69 #define	TRUE	1
70 #endif
71 #ifndef FALSE
72 #define	FALSE	0
73 #endif
74 
75 #define	CTF_ELF_SCN_NAME	".SUNW_ctf"
76 
77 #define	CTF_LABEL_LASTIDX	-1
78 
79 #define	CTF_DEFAULT_LABEL	"*** No Label Provided ***"
80 
81 /*
82  * Default hash sizes
83  */
84 #define	TDATA_LAYOUT_HASH_SIZE	8191	/* A tdesc hash based on layout */
85 #define	TDATA_ID_HASH_SIZE	997	/* A tdesc hash based on type id */
86 #define	IIDESC_HASH_SIZE	8191	/* Hash of iidesc's */
87 
88 /*
89  * The default function argument array size.  We'll realloc the array larger
90  * if we need to, but we want a default value that will allow us to avoid
91  * reallocation in the common case.
92  */
93 #define	FUNCARG_DEF	5
94 
95 extern const char *progname;
96 extern int debug_level;
97 extern int debug_parse;
98 extern char *curhdr;
99 
100 /*
101  * This is a partial copy of the stab.h that DevPro includes with their
102  * compiler.
103  */
104 typedef struct stab {
105 	uint32_t	n_strx;
106 	uint8_t		n_type;
107 	int8_t		n_other;
108 	int16_t		n_desc;
109 	uint32_t	n_value;
110 } stab_t;
111 
112 #define	N_GSYM	0x20	/* global symbol: name,,0,type,0 */
113 #define	N_FUN	0x24	/* procedure: name,,0,linenumber,0 */
114 #define	N_STSYM	0x26	/* static symbol: name,,0,type,0 or section relative */
115 #define	N_LCSYM	0x28	/* .lcomm symbol: name,,0,type,0 or section relative */
116 #define	N_ROSYM	0x2c	/* ro_data: name,,0,type,0 or section relative */
117 #define	N_OPT	0x3c	/* compiler options */
118 #define	N_RSYM	0x40	/* register sym: name,,0,type,register */
119 #define	N_SO	0x64	/* source file name: name,,0,0,0 */
120 #define	N_LSYM	0x80	/* local sym: name,,0,type,offset */
121 #define	N_SOL	0x84	/* #included file name: name,,0,0,0 */
122 #define	N_PSYM	0xa0	/* parameter: name,,0,type,offset */
123 #define	N_LBRAC	0xc0	/* left bracket: 0,,0,nesting level,function relative */
124 #define	N_RBRAC	0xe0	/* right bracket: 0,,0,nesting level,func relative */
125 #define	N_BINCL 0x82	/* header file: name,,0,0,0 */
126 #define	N_EINCL 0xa2	/* end of include file */
127 
128 /*
129  * Nodes in the type tree
130  *
131  * Each node consists of a single tdesc_t, with one of several auxiliary
132  * structures linked in via the `data' union.
133  */
134 
135 /* The type of tdesc_t node */
136 typedef enum stabtype {
137 	STABTYPE_FIRST, /* do not use */
138 	INTRINSIC,
139 	POINTER,
140 	ARRAY,
141 	FUNCTION,
142 	STRUCT,
143 	UNION,
144 	ENUM,
145 	FORWARD,
146 	TYPEDEF,
147 	TYPEDEF_UNRES,
148 	VOLATILE,
149 	CONST,
150 	RESTRICT,
151 	STABTYPE_LAST /* do not use */
152 } stabtype_t;
153 
154 typedef struct tdesc tdesc_t;
155 
156 /* Auxiliary structure for array tdesc_t */
157 typedef struct ardef {
158 	tdesc_t	*ad_contents;
159 	tdesc_t *ad_idxtype;
160 	uint_t	ad_nelems;
161 } ardef_t;
162 
163 /* Auxiliary structure for structure/union tdesc_t */
164 typedef struct mlist {
165 	int	ml_offset;	/* Offset from start of structure (in bits) */
166 	uint_t	ml_size;	/* Member size (in bits) */
167 	char	*ml_name;	/* Member name */
168 	struct	tdesc *ml_type;	/* Member type */
169 	struct	mlist *ml_next;	/* Next member */
170 } mlist_t;
171 
172 /* Auxiliary structure for enum tdesc_t */
173 typedef struct elist {
174 	char	*el_name;
175 	int	el_number;
176 	struct elist *el_next;
177 } elist_t;
178 
179 /* Auxiliary structure for intrinsics (integers and reals) */
180 typedef enum {
181 	INTR_INT,
182 	INTR_REAL
183 } intrtype_t;
184 
185 typedef struct intr {
186 	intrtype_t	intr_type;
187 	int		intr_signed;
188 	union {
189 			char _iformat;
190 			int _fformat;
191 	} _u;
192 	int		intr_offset;
193 	int		intr_nbits;
194 } intr_t;
195 
196 #define	intr_iformat _u._iformat
197 #define	intr_fformat _u._fformat
198 
199 typedef struct fnarg {
200 	char *fna_name;
201 	struct tdesc *fna_type;
202 } fnarg_t;
203 
204 #define	FN_F_GLOBAL	0x1
205 #define	FN_F_VARARGS	0x2
206 
207 typedef struct fndef {
208 	struct tdesc *fn_ret;
209 	uint_t fn_nargs;
210 	tdesc_t **fn_args;
211 	uint_t fn_vargs;
212 } fndef_t;
213 
214 typedef int32_t tid_t;
215 
216 /*
217  * The tdesc_t (Type DESCription) is the basic node type used in the stabs data
218  * structure.  Each data node gets a tdesc structure.  Each node is linked into
219  * a directed graph (think of it as a tree with multiple roots and multiple
220  * leaves), with the root nodes at the top, and intrinsics at the bottom.  The
221  * root nodes, which are pointed to by iidesc nodes, correspond to the types,
222  * globals, and statics defined by the stabs.
223  */
224 struct tdesc {
225 	char	*t_name;
226 	tdesc_t *t_next;	/* Name hash next pointer */
227 
228 	tid_t t_id;
229 	tdesc_t *t_hash;	/* ID hash next pointer */
230 
231 	stabtype_t t_type;
232 	int	t_size;	/* Size in bytes of object represented by this node */
233 
234 	union {
235 		intr_t	*intr;		/* int, real */
236 		tdesc_t *tdesc;		/* ptr, typedef, vol, const, restr */
237 		ardef_t *ardef;		/* array */
238 		mlist_t *members;	/* struct, union */
239 		elist_t *emem;		/* enum */
240 		fndef_t *fndef;		/* function - first is return type */
241 	} t_data;
242 
243 	int t_flags;
244 	int t_vgen;	/* Visitation generation (see traverse.c) */
245 	int t_emark;	/* Equality mark (see equiv_cb() in merge.c) */
246 };
247 
248 #define	t_intr		t_data.intr
249 #define	t_tdesc		t_data.tdesc
250 #define	t_ardef		t_data.ardef
251 #define	t_members	t_data.members
252 #define	t_emem		t_data.emem
253 #define	t_fndef		t_data.fndef
254 
255 #define	TDESC_F_ISROOT		0x1	/* Has an iidesc_t (see below) */
256 #define	TDESC_F_GLOBAL		0x2
257 #define	TDESC_F_RESOLVED	0x4
258 
259 /*
260  * iidesc_t (Interesting Item DESCription) nodes point to tdesc_t nodes that
261  * correspond to "interesting" stabs.  A stab is interesting if it defines a
262  * global or static variable, a global or static function, or a data type.
263  */
264 typedef enum iitype {
265 	II_NOT = 0,
266 	II_GFUN,	/* Global function */
267 	II_SFUN,	/* Static function */
268 	II_GVAR,	/* Global variable */
269 	II_SVAR,	/* Static variable */
270 	II_PSYM,	/* Function argument */
271 	II_SOU,		/* Struct or union */
272 	II_TYPE		/* Type (typedef) */
273 } iitype_t;
274 
275 typedef struct iidesc {
276 	iitype_t	ii_type;
277 	char		*ii_name;
278 	tdesc_t 	*ii_dtype;
279 	char		*ii_owner;	/* File that defined this node */
280 	int		ii_flags;
281 
282 	/* Function arguments (if any) */
283 	int		ii_nargs;
284 	tdesc_t 	**ii_args;
285 	int		ii_vargs;	/* Function uses varargs */
286 } iidesc_t;
287 
288 #define	IIDESC_F_USED	0x1	/* Write this iidesc out */
289 
290 /*
291  * labelent_t nodes identify labels and corresponding type ranges associated
292  * with them.  The label in a given labelent_t is associated with types with
293  * ids <= le_idx.
294  */
295 typedef struct labelent {
296 	char *le_name;
297 	int le_idx;
298 } labelent_t;
299 
300 /*
301  * The tdata_t (Type DATA) structure contains or references all type data for
302  * a given file or, during merging, several files.
303  */
304 typedef struct tdata {
305 	int	td_curemark;	/* Equality mark (see merge.c) */
306 	int	td_curvgen;	/* Visitation generation (see traverse.c) */
307 	int	td_nextid;	/* The ID for the next tdesc_t created */
308 	hash_t	*td_iihash;	/* The iidesc_t nodes for this file */
309 
310 	hash_t	*td_layouthash;	/* The tdesc nodes, hashed by structure */
311 	hash_t	*td_idhash;	/* The tdesc nodes, hashed by type id */
312 	list_t	*td_fwdlist;	/* All forward declaration tdesc nodes */
313 
314 	char	*td_parlabel;	/* Top label uniq'd against in parent */
315 	char	*td_parname;	/* Basename of parent */
316 	list_t	*td_labels;	/* Labels and their type ranges */
317 
318 	pthread_mutex_t td_mergelock;
319 
320 	int	td_ref;
321 } tdata_t;
322 
323 /*
324  * By design, the iidesc hash is heterogeneous.  The CTF emitter, on the
325  * other hand, needs to be able to access the elements of the list by type,
326  * and in a specific sorted order.  An iiburst holds these elements in that
327  * order.  (A burster is a machine that separates carbon-copy forms)
328  */
329 typedef struct iiburst {
330 	int iib_nfuncs;
331 	int iib_curfunc;
332 	iidesc_t **iib_funcs;
333 
334 	int iib_nobjts;
335 	int iib_curobjt;
336 	iidesc_t **iib_objts;
337 
338 	list_t *iib_types;
339 	int iib_maxtypeid;
340 
341 	tdata_t *iib_td;
342 	struct tdtrav_data *iib_tdtd; /* tdtrav_data_t */
343 } iiburst_t;
344 
345 typedef struct ctf_buf ctf_buf_t;
346 
347 typedef struct symit_data symit_data_t;
348 
349 /* fixup_tdescs.c */
350 void cvt_fixstabs(tdata_t *);
351 void cvt_fixups(tdata_t *, size_t);
352 
353 /* ctf.c */
354 caddr_t ctf_gen(iiburst_t *, size_t *, int);
355 tdata_t *ctf_load(char *, caddr_t, size_t, symit_data_t *, char *);
356 
357 /* iidesc.c */
358 iidesc_t *iidesc_new(char *);
359 int iidesc_hash(int, void *);
360 void iter_iidescs_by_name(tdata_t *, const char *,
361     int (*)(void *, void *), void *);
362 iidesc_t *iidesc_dup(iidesc_t *);
363 iidesc_t *iidesc_dup_rename(iidesc_t *, char const *, char const *);
364 void iidesc_add(hash_t *, iidesc_t *);
365 void iidesc_free(void *, void *);
366 int iidesc_count_type(void *, void *);
367 void iidesc_stats(hash_t *);
368 int iidesc_dump(iidesc_t *);
369 
370 /* input.c */
371 typedef enum source_types {
372 	SOURCE_NONE 	= 0,
373 	SOURCE_UNKNOWN	= 1,
374 	SOURCE_C	= 2,
375 	SOURCE_S	= 4
376 } source_types_t;
377 
378 source_types_t built_source_types(Elf *, const char *);
379 int count_files(char **, int);
380 int read_ctf(char **, int, char *, int (*)(tdata_t *, char *, void *),
381     void *, int);
382 int read_ctf_save_cb(tdata_t *, char *, void *);
383 symit_data_t *symit_new(Elf *, const char *);
384 void symit_reset(symit_data_t *);
385 char *symit_curfile(symit_data_t *);
386 GElf_Sym *symit_next(symit_data_t *, int);
387 char *symit_name(symit_data_t *);
388 void symit_free(symit_data_t *);
389 
390 /* merge.c */
391 void merge_into_master(tdata_t *, tdata_t *, tdata_t *, int);
392 
393 /* output.c */
394 #define	CTF_FUZZY_MATCH	0x1 /* match local symbols to global CTF */
395 #define	CTF_USE_DYNSYM	0x2 /* use .dynsym not .symtab */
396 #define	CTF_COMPRESS	0x4 /* compress CTF output */
397 #define	CTF_KEEP_STABS	0x8 /* keep .stabs sections */
398 #define	CTF_SWAP_BYTES	0x10 /* target byte order is different from host */
399 
400 void write_ctf(tdata_t *, const char *, const char *, int);
401 
402 /* parse.c */
403 void parse_init(tdata_t *);
404 void parse_finish(tdata_t *);
405 int parse_stab(stab_t *, char *, iidesc_t **);
406 tdesc_t *lookup(int);
407 tdesc_t *lookupname(const char *);
408 void check_hash(void);
409 void resolve_typed_bitfields(void);
410 
411 /* stabs.c */
412 int stabs_read(tdata_t *, Elf *, char *);
413 
414 /* dwarf.c */
415 int dw_read(tdata_t *, Elf *, char *);
416 const char *dw_tag2str(uint_t);
417 
418 /* tdata.c */
419 tdata_t *tdata_new(void);
420 void tdata_free(tdata_t *);
421 void tdata_build_hashes(tdata_t *td);
422 const char *tdesc_name(tdesc_t *);
423 int tdesc_idhash(int, void *);
424 int tdesc_idcmp(void *, void *);
425 int tdesc_namehash(int, void *);
426 int tdesc_namecmp(void *, void *);
427 int tdesc_layouthash(int, void *);
428 int tdesc_layoutcmp(void *, void *);
429 void tdesc_free(tdesc_t *);
430 void tdata_label_add(tdata_t *, const char *, int);
431 labelent_t *tdata_label_top(tdata_t *);
432 int tdata_label_find(tdata_t *, char *);
433 void tdata_label_free(tdata_t *);
434 void tdata_merge(tdata_t *, tdata_t *);
435 void tdata_label_newmax(tdata_t *, int);
436 
437 /* util.c */
438 int streq(const char *, const char *);
439 int findelfsecidx(Elf *, const char *, const char *);
440 size_t elf_ptrsz(Elf *);
441 char *mktmpname(const char *, const char *);
442 void terminate(const char *, ...);
443 void aborterr(const char *, ...);
444 void set_terminate_cleanup(void (*)(void));
445 void elfterminate(const char *, const char *, ...);
446 void warning(const char *, ...);
447 void vadebug(int, const char *, va_list);
448 void debug(int, const char *, ...);
449 
450 
451 void watch_dump(int);
452 void watch_set(void *, int);
453 
454 #ifdef __cplusplus
455 }
456 #endif
457 
458 #endif /* _CTFTOOLS_H */
459