xref: /freebsd/cddl/contrib/opensolaris/tools/ctf/cvt/ctftools.h (revision 0b3105a37d7adcadcb720112fed4dc4e8040be99)
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 /*
30  * Functions and data structures used in the manipulation of stabs and CTF data
31  */
32 
33 #include <stdio.h>
34 #include <stdlib.h>
35 #include <stdarg.h>
36 #include <libelf.h>
37 #include <gelf.h>
38 #include <pthread.h>
39 
40 #include <sys/ccompile.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 #define	TRUE	1
69 #define	FALSE	0
70 
71 #define	CTF_ELF_SCN_NAME	".SUNW_ctf"
72 
73 #define	CTF_LABEL_LASTIDX	-1
74 
75 #define	CTF_DEFAULT_LABEL	"*** No Label Provided ***"
76 
77 /*
78  * Default hash sizes
79  */
80 #define	TDATA_LAYOUT_HASH_SIZE	8191	/* A tdesc hash based on layout */
81 #define	TDATA_ID_HASH_SIZE	997	/* A tdesc hash based on type id */
82 #define	IIDESC_HASH_SIZE	8191	/* Hash of iidesc's */
83 
84 /*
85  * The default function argument array size.  We'll realloc the array larger
86  * if we need to, but we want a default value that will allow us to avoid
87  * reallocation in the common case.
88  */
89 #define	FUNCARG_DEF	5
90 
91 extern const char *progname;
92 extern int debug_level;
93 extern int debug_parse;
94 extern char *curhdr;
95 
96 /*
97  * This is a partial copy of the stab.h that DevPro includes with their
98  * compiler.
99  */
100 typedef struct stab {
101 	uint32_t	n_strx;
102 	uint8_t		n_type;
103 	int8_t		n_other;
104 	int16_t		n_desc;
105 	uint32_t	n_value;
106 } stab_t;
107 
108 #define	N_GSYM	0x20	/* global symbol: name,,0,type,0 */
109 #define	N_FUN	0x24	/* procedure: name,,0,linenumber,0 */
110 #define	N_STSYM	0x26	/* static symbol: name,,0,type,0 or section relative */
111 #define	N_LCSYM	0x28	/* .lcomm symbol: name,,0,type,0 or section relative */
112 #define	N_ROSYM	0x2c	/* ro_data: name,,0,type,0 or section relative */
113 #define	N_OPT	0x3c	/* compiler options */
114 #define	N_RSYM	0x40	/* register sym: name,,0,type,register */
115 #define	N_SO	0x64	/* source file name: name,,0,0,0 */
116 #define	N_LSYM	0x80	/* local sym: name,,0,type,offset */
117 #define	N_SOL	0x84	/* #included file name: name,,0,0,0 */
118 #define	N_PSYM	0xa0	/* parameter: name,,0,type,offset */
119 #define	N_LBRAC	0xc0	/* left bracket: 0,,0,nesting level,function relative */
120 #define	N_RBRAC	0xe0	/* right bracket: 0,,0,nesting level,func relative */
121 #define	N_BINCL 0x82	/* header file: name,,0,0,0 */
122 #define	N_EINCL 0xa2	/* end of include file */
123 
124 /*
125  * Nodes in the type tree
126  *
127  * Each node consists of a single tdesc_t, with one of several auxiliary
128  * structures linked in via the `data' union.
129  */
130 
131 /* The type of tdesc_t node */
132 typedef enum stabtype {
133 	STABTYPE_FIRST, /* do not use */
134 	INTRINSIC,
135 	POINTER,
136 	ARRAY,
137 	FUNCTION,
138 	STRUCT,
139 	UNION,
140 	ENUM,
141 	FORWARD,
142 	TYPEDEF,
143 	TYPEDEF_UNRES,
144 	VOLATILE,
145 	CONST,
146 	RESTRICT,
147 	STABTYPE_LAST /* do not use */
148 } stabtype_t;
149 
150 typedef struct tdesc tdesc_t;
151 
152 /* Auxiliary structure for array tdesc_t */
153 typedef struct ardef {
154 	tdesc_t	*ad_contents;
155 	tdesc_t *ad_idxtype;
156 	uint_t	ad_nelems;
157 } ardef_t;
158 
159 /* Auxiliary structure for structure/union tdesc_t */
160 typedef struct mlist {
161 	int	ml_offset;	/* Offset from start of structure (in bits) */
162 	uint_t	ml_size;	/* Member size (in bits) */
163 	char	*ml_name;	/* Member name */
164 	struct	tdesc *ml_type;	/* Member type */
165 	struct	mlist *ml_next;	/* Next member */
166 } mlist_t;
167 
168 /* Auxiliary structure for enum tdesc_t */
169 typedef struct elist {
170 	char	*el_name;
171 	int	el_number;
172 	struct elist *el_next;
173 } elist_t;
174 
175 /* Auxiliary structure for intrinsics (integers and reals) */
176 typedef enum {
177 	INTR_INT,
178 	INTR_REAL
179 } intrtype_t;
180 
181 typedef struct intr {
182 	intrtype_t	intr_type;
183 	int		intr_signed;
184 	union {
185 			char _iformat;
186 			int _fformat;
187 	} _u;
188 	int		intr_offset;
189 	int		intr_nbits;
190 } intr_t;
191 
192 #define	intr_iformat _u._iformat
193 #define	intr_fformat _u._fformat
194 
195 typedef struct fnarg {
196 	char *fna_name;
197 	struct tdesc *fna_type;
198 } fnarg_t;
199 
200 #define	FN_F_GLOBAL	0x1
201 #define	FN_F_VARARGS	0x2
202 
203 typedef struct fndef {
204 	struct tdesc *fn_ret;
205 	uint_t fn_nargs;
206 	tdesc_t **fn_args;
207 	uint_t fn_vargs;
208 } fndef_t;
209 
210 typedef int32_t tid_t;
211 
212 /*
213  * The tdesc_t (Type DESCription) is the basic node type used in the stabs data
214  * structure.  Each data node gets a tdesc structure.  Each node is linked into
215  * a directed graph (think of it as a tree with multiple roots and multiple
216  * leaves), with the root nodes at the top, and intrinsics at the bottom.  The
217  * root nodes, which are pointed to by iidesc nodes, correspond to the types,
218  * globals, and statics defined by the stabs.
219  */
220 struct tdesc {
221 	char	*t_name;
222 	tdesc_t *t_next;	/* Name hash next pointer */
223 
224 	tid_t t_id;
225 	tdesc_t *t_hash;	/* ID hash next pointer */
226 
227 	stabtype_t t_type;
228 	int	t_size;	/* Size in bytes of object represented by this node */
229 
230 	union {
231 		intr_t	*intr;		/* int, real */
232 		tdesc_t *tdesc;		/* ptr, typedef, vol, const, restr */
233 		ardef_t *ardef;		/* array */
234 		mlist_t *members;	/* struct, union */
235 		elist_t *emem;		/* enum */
236 		fndef_t *fndef;		/* function - first is return type */
237 	} t_data;
238 
239 	int t_flags;
240 	int t_vgen;	/* Visitation generation (see traverse.c) */
241 	int t_emark;	/* Equality mark (see equiv_cb() in merge.c) */
242 };
243 
244 #define	t_intr		t_data.intr
245 #define	t_tdesc		t_data.tdesc
246 #define	t_ardef		t_data.ardef
247 #define	t_members	t_data.members
248 #define	t_emem		t_data.emem
249 #define	t_fndef		t_data.fndef
250 
251 #define	TDESC_F_ISROOT		0x1	/* Has an iidesc_t (see below) */
252 #define	TDESC_F_GLOBAL		0x2
253 #define	TDESC_F_RESOLVED	0x4
254 
255 /*
256  * iidesc_t (Interesting Item DESCription) nodes point to tdesc_t nodes that
257  * correspond to "interesting" stabs.  A stab is interesting if it defines a
258  * global or static variable, a global or static function, or a data type.
259  */
260 typedef enum iitype {
261 	II_NOT = 0,
262 	II_GFUN,	/* Global function */
263 	II_SFUN,	/* Static function */
264 	II_GVAR,	/* Global variable */
265 	II_SVAR,	/* Static variable */
266 	II_PSYM,	/* Function argument */
267 	II_SOU,		/* Struct or union */
268 	II_TYPE		/* Type (typedef) */
269 } iitype_t;
270 
271 typedef struct iidesc {
272 	iitype_t	ii_type;
273 	char		*ii_name;
274 	tdesc_t 	*ii_dtype;
275 	char		*ii_owner;	/* File that defined this node */
276 	int		ii_flags;
277 
278 	/* Function arguments (if any) */
279 	int		ii_nargs;
280 	tdesc_t 	**ii_args;
281 	int		ii_vargs;	/* Function uses varargs */
282 } iidesc_t;
283 
284 #define	IIDESC_F_USED	0x1	/* Write this iidesc out */
285 
286 /*
287  * labelent_t nodes identify labels and corresponding type ranges associated
288  * with them.  The label in a given labelent_t is associated with types with
289  * ids <= le_idx.
290  */
291 typedef struct labelent {
292 	char *le_name;
293 	int le_idx;
294 } labelent_t;
295 
296 /*
297  * The tdata_t (Type DATA) structure contains or references all type data for
298  * a given file or, during merging, several files.
299  */
300 typedef struct tdata {
301 	int	td_curemark;	/* Equality mark (see merge.c) */
302 	int	td_curvgen;	/* Visitation generation (see traverse.c) */
303 	int	td_nextid;	/* The ID for the next tdesc_t created */
304 	hash_t	*td_iihash;	/* The iidesc_t nodes for this file */
305 
306 	hash_t	*td_layouthash;	/* The tdesc nodes, hashed by structure */
307 	hash_t	*td_idhash;	/* The tdesc nodes, hashed by type id */
308 	list_t	*td_fwdlist;	/* All forward declaration tdesc nodes */
309 
310 	char	*td_parlabel;	/* Top label uniq'd against in parent */
311 	char	*td_parname;	/* Basename of parent */
312 	list_t	*td_labels;	/* Labels and their type ranges */
313 
314 	pthread_mutex_t td_mergelock;
315 
316 	int	td_ref;
317 } tdata_t;
318 
319 /*
320  * By design, the iidesc hash is heterogeneous.  The CTF emitter, on the
321  * other hand, needs to be able to access the elements of the list by type,
322  * and in a specific sorted order.  An iiburst holds these elements in that
323  * order.  (A burster is a machine that separates carbon-copy forms)
324  */
325 typedef struct iiburst {
326 	int iib_nfuncs;
327 	int iib_curfunc;
328 	iidesc_t **iib_funcs;
329 
330 	int iib_nobjts;
331 	int iib_curobjt;
332 	iidesc_t **iib_objts;
333 
334 	list_t *iib_types;
335 	int iib_maxtypeid;
336 
337 	tdata_t *iib_td;
338 	struct tdtrav_data *iib_tdtd; /* tdtrav_data_t */
339 } iiburst_t;
340 
341 typedef struct ctf_buf ctf_buf_t;
342 
343 typedef struct symit_data symit_data_t;
344 
345 /* fixup_tdescs.c */
346 void cvt_fixstabs(tdata_t *);
347 void cvt_fixups(tdata_t *, size_t);
348 
349 /* ctf.c */
350 caddr_t ctf_gen(iiburst_t *, size_t *, int);
351 tdata_t *ctf_load(char *, caddr_t, size_t, symit_data_t *, char *);
352 
353 /* iidesc.c */
354 iidesc_t *iidesc_new(char *);
355 int iidesc_hash(int, void *);
356 void iter_iidescs_by_name(tdata_t *, const char *,
357     int (*)(void *, void *), void *);
358 iidesc_t *iidesc_dup(iidesc_t *);
359 iidesc_t *iidesc_dup_rename(iidesc_t *, char const *, char const *);
360 void iidesc_add(hash_t *, iidesc_t *);
361 void iidesc_free(void *, void *);
362 int iidesc_count_type(void *, void *);
363 void iidesc_stats(hash_t *);
364 int iidesc_dump(iidesc_t *);
365 
366 /* input.c */
367 typedef enum source_types {
368 	SOURCE_NONE 	= 0,
369 	SOURCE_UNKNOWN	= 1,
370 	SOURCE_C	= 2,
371 	SOURCE_S	= 4
372 } source_types_t;
373 
374 source_types_t built_source_types(Elf *, const char *);
375 int count_files(char **, int);
376 int read_ctf(char **, int, char *, int (*)(tdata_t *, char *, void *),
377     void *, int);
378 int read_ctf_save_cb(tdata_t *, char *, void *);
379 symit_data_t *symit_new(Elf *, const char *);
380 void symit_reset(symit_data_t *);
381 char *symit_curfile(symit_data_t *);
382 GElf_Sym *symit_next(symit_data_t *, int);
383 char *symit_name(symit_data_t *);
384 void symit_free(symit_data_t *);
385 
386 /* merge.c */
387 void merge_into_master(tdata_t *, tdata_t *, tdata_t *, int);
388 
389 /* output.c */
390 #define	CTF_FUZZY_MATCH	0x1 /* match local symbols to global CTF */
391 #define	CTF_USE_DYNSYM	0x2 /* use .dynsym not .symtab */
392 #define	CTF_COMPRESS	0x4 /* compress CTF output */
393 #define	CTF_KEEP_STABS	0x8 /* keep .stabs sections */
394 #define	CTF_SWAP_BYTES	0x10 /* target byte order is different from host */
395 
396 void write_ctf(tdata_t *, const char *, const char *, int);
397 
398 /* parse.c */
399 void parse_init(tdata_t *);
400 void parse_finish(tdata_t *);
401 int parse_stab(stab_t *, char *, iidesc_t **);
402 tdesc_t *lookup(int);
403 tdesc_t *lookupname(const char *);
404 void check_hash(void);
405 void resolve_typed_bitfields(void);
406 
407 /* stabs.c */
408 int stabs_read(tdata_t *, Elf *, char *);
409 
410 /* dwarf.c */
411 int dw_read(tdata_t *, Elf *, char *);
412 const char *dw_tag2str(uint_t);
413 
414 /* tdata.c */
415 tdata_t *tdata_new(void);
416 void tdata_free(tdata_t *);
417 void tdata_build_hashes(tdata_t *td);
418 const char *tdesc_name(tdesc_t *);
419 int tdesc_idhash(int, void *);
420 int tdesc_idcmp(void *, void *);
421 int tdesc_namehash(int, void *);
422 int tdesc_namecmp(void *, void *);
423 int tdesc_layouthash(int, void *);
424 int tdesc_layoutcmp(void *, void *);
425 void tdesc_free(tdesc_t *);
426 void tdata_label_add(tdata_t *, const char *, int);
427 labelent_t *tdata_label_top(tdata_t *);
428 int tdata_label_find(tdata_t *, char *);
429 void tdata_label_free(tdata_t *);
430 void tdata_merge(tdata_t *, tdata_t *);
431 void tdata_label_newmax(tdata_t *, int);
432 
433 /* util.c */
434 int streq(const char *, const char *);
435 int findelfsecidx(Elf *, const char *, const char *);
436 size_t elf_ptrsz(Elf *);
437 char *mktmpname(const char *, const char *);
438 void terminate(const char *, ...) __NORETURN;
439 void aborterr(const char *, ...) __NORETURN;
440 void set_terminate_cleanup(void (*)(void));
441 void elfterminate(const char *, const char *, ...);
442 void warning(const char *, ...);
443 void vadebug(int, const char *, va_list);
444 void debug(int, const char *, ...);
445 
446 
447 void watch_dump(int);
448 void watch_set(void *, int);
449 
450 #ifdef __cplusplus
451 }
452 #endif
453 
454 #endif /* _CTFTOOLS_H */
455