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 /* Sanity check for cross-build bootstrap tools */
69 #if !defined(BYTE_ORDER)
70 #error "Missing BYTE_ORDER defines"
71 #elif !defined(LITTLE_ENDIAN)
72 #error "Missing LITTLE_ENDIAN defines"
73 #elif !defined(BIG_ENDIAN)
74 #error "Missing BIG_ENDIAN defines"
75 #endif
76 
77 #ifndef TRUE
78 #define   TRUE      1
79 #endif
80 #ifndef FALSE
81 #define   FALSE     0
82 #endif
83 
84 #define   CTF_ELF_SCN_NAME    ".SUNW_ctf"
85 
86 #define   CTF_LABEL_LASTIDX   -1
87 
88 #define   CTF_DEFAULT_LABEL   "*** No Label Provided ***"
89 
90 /*
91  * Default hash sizes
92  */
93 #define   TDATA_LAYOUT_HASH_SIZE        8191      /* A tdesc hash based on layout */
94 #define   TDATA_ID_HASH_SIZE  997       /* A tdesc hash based on type id */
95 #define   IIDESC_HASH_SIZE    8191      /* Hash of iidesc's */
96 
97 /*
98  * The default function argument array size.  We'll realloc the array larger
99  * if we need to, but we want a default value that will allow us to avoid
100  * reallocation in the common case.
101  */
102 #define   FUNCARG_DEF         5
103 
104 extern const char *progname;
105 extern int debug_level;
106 extern int debug_parse;
107 extern char *curhdr;
108 
109 /*
110  * This is a partial copy of the stab.h that DevPro includes with their
111  * compiler.
112  */
113 typedef struct stab {
114           uint32_t  n_strx;
115           uint8_t             n_type;
116           int8_t              n_other;
117           int16_t             n_desc;
118           uint32_t  n_value;
119 } stab_t;
120 
121 #define   N_GSYM    0x20      /* global symbol: name,,0,type,0 */
122 #define   N_FUN     0x24      /* procedure: name,,0,linenumber,0 */
123 #define   N_STSYM   0x26      /* static symbol: name,,0,type,0 or section relative */
124 #define   N_LCSYM   0x28      /* .lcomm symbol: name,,0,type,0 or section relative */
125 #define   N_ROSYM   0x2c      /* ro_data: name,,0,type,0 or section relative */
126 #define   N_OPT     0x3c      /* compiler options */
127 #define   N_RSYM    0x40      /* register sym: name,,0,type,register */
128 #define   N_SO      0x64      /* source file name: name,,0,0,0 */
129 #define   N_LSYM    0x80      /* local sym: name,,0,type,offset */
130 #define   N_SOL     0x84      /* #included file name: name,,0,0,0 */
131 #define   N_PSYM    0xa0      /* parameter: name,,0,type,offset */
132 #define   N_LBRAC   0xc0      /* left bracket: 0,,0,nesting level,function relative */
133 #define   N_RBRAC   0xe0      /* right bracket: 0,,0,nesting level,func relative */
134 #define   N_BINCL 0x82        /* header file: name,,0,0,0 */
135 #define   N_EINCL 0xa2        /* end of include file */
136 
137 /*
138  * Nodes in the type tree
139  *
140  * Each node consists of a single tdesc_t, with one of several auxiliary
141  * structures linked in via the `data' union.
142  */
143 
144 /* The type of tdesc_t node */
145 typedef enum stabtype {
146           STABTYPE_FIRST, /* do not use */
147           INTRINSIC,
148           POINTER,
149           REFERENCE,
150           ARRAY,
151           FUNCTION,
152           STRUCT,
153           UNION,
154           CLASS,
155           ENUM,
156           FORWARD,
157           TYPEDEF,
158           TYPEDEF_UNRES,
159           VOLATILE,
160           CONST,
161           RESTRICT,
162           STABTYPE_LAST /* do not use */
163 } stabtype_t;
164 
165 typedef struct tdesc tdesc_t;
166 
167 /* Auxiliary structure for array tdesc_t */
168 typedef struct ardef {
169           tdesc_t   *ad_contents;
170           tdesc_t *ad_idxtype;
171           uint_t    ad_nelems;
172 } ardef_t;
173 
174 /* Auxiliary structure for structure/union tdesc_t */
175 typedef struct mlist {
176           int       ml_offset;          /* Offset from start of structure (in bits) */
177           uint_t    ml_size;  /* Member size (in bits) */
178           char      *ml_name; /* Member name */
179           struct    tdesc *ml_type;     /* Member type */
180           struct    mlist *ml_next;     /* Next member */
181 } mlist_t;
182 
183 /* Auxiliary structure for enum tdesc_t */
184 typedef struct elist {
185           char      *el_name;
186           int       el_number;
187           struct elist *el_next;
188 } elist_t;
189 
190 /* Auxiliary structure for intrinsics (integers and reals) */
191 typedef enum {
192           INTR_INT,
193           INTR_REAL
194 } intrtype_t;
195 
196 typedef struct intr {
197           intrtype_t          intr_type;
198           int                 intr_signed;
199           union {
200                               char _iformat;
201                               int _fformat;
202           } _u;
203           int                 intr_offset;
204           int                 intr_nbits;
205 } intr_t;
206 
207 #define   intr_iformat _u._iformat
208 #define   intr_fformat _u._fformat
209 
210 typedef struct fnarg {
211           char *fna_name;
212           struct tdesc *fna_type;
213 } fnarg_t;
214 
215 #define   FN_F_GLOBAL         0x1
216 #define   FN_F_VARARGS        0x2
217 
218 typedef struct fndef {
219           struct tdesc *fn_ret;
220           uint_t fn_nargs;
221           tdesc_t **fn_args;
222           uint_t fn_vargs;
223 } fndef_t;
224 
225 typedef int32_t tid_t;
226 
227 /*
228  * The tdesc_t (Type DESCription) is the basic node type used in the stabs data
229  * structure.  Each data node gets a tdesc structure.  Each node is linked into
230  * a directed graph (think of it as a tree with multiple roots and multiple
231  * leaves), with the root nodes at the top, and intrinsics at the bottom.  The
232  * root nodes, which are pointed to by iidesc nodes, correspond to the types,
233  * globals, and statics defined by the stabs.
234  */
235 struct tdesc {
236           char      *t_name;
237           tdesc_t *t_next;    /* Name hash next pointer */
238 
239           tid_t t_id;
240           tdesc_t *t_hash;    /* ID hash next pointer */
241 
242           stabtype_t t_type;
243           int       t_size;   /* Size in bytes of object represented by this node */
244 
245           union {
246                     intr_t    *intr;              /* int, real */
247                     tdesc_t *tdesc;               /* ptr, typedef, vol, const, restr */
248                     ardef_t *ardef;               /* array */
249                     mlist_t *members;   /* struct, union */
250                     elist_t *emem;                /* enum */
251                     fndef_t *fndef;               /* function - first is return type */
252           } t_data;
253 
254           int t_flags;
255           int t_vgen;         /* Visitation generation (see traverse.c) */
256           int t_emark;        /* Equality mark (see equiv_cb() in merge.c) */
257 };
258 
259 #define   t_intr              t_data.intr
260 #define   t_tdesc             t_data.tdesc
261 #define   t_ardef             t_data.ardef
262 #define   t_members t_data.members
263 #define   t_emem              t_data.emem
264 #define   t_fndef             t_data.fndef
265 
266 #define   TDESC_F_ISROOT                0x1       /* Has an iidesc_t (see below) */
267 #define   TDESC_F_GLOBAL                0x2
268 #define   TDESC_F_RESOLVED    0x4
269 
270 /*
271  * iidesc_t (Interesting Item DESCription) nodes point to tdesc_t nodes that
272  * correspond to "interesting" stabs.  A stab is interesting if it defines a
273  * global or static variable, a global or static function, or a data type.
274  */
275 typedef enum iitype {
276           II_NOT = 0,
277           II_GFUN,  /* Global function */
278           II_SFUN,  /* Static function */
279           II_GVAR,  /* Global variable */
280           II_SVAR,  /* Static variable */
281           II_PSYM,  /* Function argument */
282           II_SOU,             /* Struct or union */
283           II_TYPE             /* Type (typedef) */
284 } iitype_t;
285 
286 typedef struct iidesc {
287           iitype_t  ii_type;
288           char                *ii_name;
289           tdesc_t   *ii_dtype;
290           char                *ii_owner;          /* File that defined this node */
291           int                 ii_flags;
292 
293           /* Function arguments (if any) */
294           int                 ii_nargs;
295           tdesc_t   **ii_args;
296           int                 ii_vargs; /* Function uses varargs */
297 } iidesc_t;
298 
299 #define   IIDESC_F_USED       0x1       /* Write this iidesc out */
300 
301 /*
302  * labelent_t nodes identify labels and corresponding type ranges associated
303  * with them.  The label in a given labelent_t is associated with types with
304  * ids <= le_idx.
305  */
306 typedef struct labelent {
307           char *le_name;
308           int le_idx;
309 } labelent_t;
310 
311 /*
312  * The tdata_t (Type DATA) structure contains or references all type data for
313  * a given file or, during merging, several files.
314  */
315 typedef struct tdata {
316           int       td_curemark;        /* Equality mark (see merge.c) */
317           int       td_curvgen;         /* Visitation generation (see traverse.c) */
318           int       td_nextid;          /* The ID for the next tdesc_t created */
319           hash_t    *td_iihash;         /* The iidesc_t nodes for this file */
320 
321           hash_t    *td_layouthash;     /* The tdesc nodes, hashed by structure */
322           hash_t    *td_idhash;         /* The tdesc nodes, hashed by type id */
323           list_t    *td_fwdlist;        /* All forward declaration tdesc nodes */
324 
325           char      *td_parlabel;       /* Top label uniq'd against in parent */
326           char      *td_parname;        /* Basename of parent */
327           list_t    *td_labels;         /* Labels and their type ranges */
328 
329           pthread_mutex_t td_mergelock;
330 
331           int       td_ref;
332 } tdata_t;
333 
334 /*
335  * By design, the iidesc hash is heterogeneous.  The CTF emitter, on the
336  * other hand, needs to be able to access the elements of the list by type,
337  * and in a specific sorted order.  An iiburst holds these elements in that
338  * order.  (A burster is a machine that separates carbon-copy forms)
339  */
340 typedef struct iiburst {
341           int iib_nfuncs;
342           int iib_curfunc;
343           iidesc_t **iib_funcs;
344 
345           int iib_nobjts;
346           int iib_curobjt;
347           iidesc_t **iib_objts;
348 
349           list_t *iib_types;
350           int iib_maxtypeid;
351 
352           tdata_t *iib_td;
353           struct tdtrav_data *iib_tdtd; /* tdtrav_data_t */
354 } iiburst_t;
355 
356 typedef struct ctf_buf ctf_buf_t;
357 
358 typedef struct symit_data symit_data_t;
359 
360 /* fixup_tdescs.c */
361 void cvt_fixstabs(tdata_t *);
362 void cvt_fixups(tdata_t *, size_t);
363 
364 /* ctf.c */
365 caddr_t ctf_gen(iiburst_t *, size_t *, int);
366 tdata_t *ctf_load(char *, caddr_t, size_t, symit_data_t *, char *);
367 
368 /* iidesc.c */
369 iidesc_t *iidesc_new(char *);
370 int iidesc_hash(int, void *);
371 void iter_iidescs_by_name(tdata_t *, const char *,
372     int (*)(void *, void *), void *);
373 iidesc_t *iidesc_dup(iidesc_t *);
374 iidesc_t *iidesc_dup_rename(iidesc_t *, char const *, char const *);
375 void iidesc_add(hash_t *, iidesc_t *);
376 void iidesc_free(void *, void *);
377 int iidesc_count_type(void *, void *);
378 void iidesc_stats(hash_t *);
379 int iidesc_dump(iidesc_t *);
380 
381 /* input.c */
382 typedef enum source_types {
383           SOURCE_NONE         = 0,
384           SOURCE_UNKNOWN      = 1,
385           SOURCE_C  = 2,
386           SOURCE_S  = 4
387 } source_types_t;
388 
389 source_types_t built_source_types(Elf *, const char *);
390 int count_files(char **, int);
391 int read_ctf(char **, int, char *, int (*)(tdata_t *, char *, void *),
392     void *, int);
393 int read_ctf_save_cb(tdata_t *, char *, void *);
394 symit_data_t *symit_new(Elf *, const char *);
395 void symit_reset(symit_data_t *);
396 char *symit_curfile(symit_data_t *);
397 GElf_Sym *symit_next(symit_data_t *, int);
398 char *symit_name(symit_data_t *);
399 void symit_free(symit_data_t *);
400 
401 /* merge.c */
402 void merge_into_master(tdata_t *, tdata_t *, tdata_t *, int);
403 
404 /* output.c */
405 #define   CTF_FUZZY_MATCH     0x1 /* match local symbols to global CTF */
406 #define   CTF_USE_DYNSYM      0x2 /* use .dynsym not .symtab */
407 #define   CTF_COMPRESS        0x4 /* compress CTF output */
408 #define   CTF_KEEP_STABS      0x8 /* keep .stabs sections */
409 #define   CTF_SWAP_BYTES      0x10 /* target byte order is different from host */
410 
411 void write_ctf(tdata_t *, const char *, const char *, int);
412 
413 /* parse.c */
414 void parse_init(tdata_t *);
415 void parse_finish(tdata_t *);
416 int parse_stab(stab_t *, char *, iidesc_t **);
417 tdesc_t *lookup(int);
418 tdesc_t *lookupname(const char *);
419 void check_hash(void);
420 void resolve_typed_bitfields(void);
421 
422 /* stabs.c */
423 int stabs_read(tdata_t *, Elf *, char *);
424 
425 /* dwarf.c */
426 int dw_read(tdata_t *, Elf *, char *);
427 const char *dw_tag2str(uint_t);
428 
429 /* tdata.c */
430 tdata_t *tdata_new(void);
431 void tdata_free(tdata_t *);
432 void tdata_build_hashes(tdata_t *td);
433 const char *tdesc_name(tdesc_t *);
434 int tdesc_idhash(int, void *);
435 int tdesc_idcmp(void *, void *);
436 int tdesc_namehash(int, void *);
437 int tdesc_namecmp(void *, void *);
438 int tdesc_layouthash(int, void *);
439 int tdesc_layoutcmp(void *, void *);
440 void tdesc_free(tdesc_t *);
441 void tdata_label_add(tdata_t *, const char *, int);
442 labelent_t *tdata_label_top(tdata_t *);
443 int tdata_label_find(tdata_t *, char *);
444 void tdata_label_free(tdata_t *);
445 void tdata_merge(tdata_t *, tdata_t *);
446 void tdata_label_newmax(tdata_t *, int);
447 
448 /* util.c */
449 int streq(const char *, const char *);
450 int findelfsecidx(Elf *, const char *, const char *);
451 size_t elf_ptrsz(Elf *);
452 char *mktmpname(const char *, const char *);
453 void terminate(const char *, ...) __printflike(1, 2) __dead;
454 void aborterr(const char *, ...) __printflike(1, 2) __dead;
455 void set_terminate_cleanup(void (*)(void));
456 void elfterminate(const char *, const char *, ...) __printflike(2, 3) __dead;
457 void warning(const char *, ...) __printflike(1, 2);
458 void vadebug(int, const char *, va_list) __printflike(2, 0);
459 void debug(int, const char *, ...) __printflike(2, 3);
460 
461 
462 void watch_dump(int);
463 void watch_set(void *, int);
464 
465 #ifdef __cplusplus
466 }
467 #endif
468 
469 #endif /* _CTFTOOLS_H */
470