1 /* Advance to next CU header.
2 Copyright (C) 2002-2010, 2016, 2017 Red Hat, Inc.
3 This file is part of elfutils.
4 Written by Ulrich Drepper <drepper@redhat.com>, 2002.
5
6 This file is free software; you can redistribute it and/or modify
7 it under the terms of either
8
9 * the GNU Lesser General Public License as published by the Free
10 Software Foundation; either version 3 of the License, or (at
11 your option) any later version
12
13 or
14
15 * the GNU General Public License as published by the Free
16 Software Foundation; either version 2 of the License, or (at
17 your option) any later version
18
19 or both in parallel, as here.
20
21 elfutils is distributed in the hope that it will be useful, but
22 WITHOUT ANY WARRANTY; without even the implied warranty of
23 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
24 General Public License for more details.
25
26 You should have received copies of the GNU General Public License and
27 the GNU Lesser General Public License along with this program. If
28 not, see <http://www.gnu.org/licenses/>. */
29
30 #ifdef HAVE_CONFIG_H
31 # include <config.h>
32 #endif
33
34 #include <libdwP.h>
35 #include <dwarf.h>
36
37
38 int
dwarf_next_unit(Dwarf * dwarf,Dwarf_Off off,Dwarf_Off * next_off,size_t * header_sizep,Dwarf_Half * versionp,Dwarf_Off * abbrev_offsetp,uint8_t * address_sizep,uint8_t * offset_sizep,uint64_t * v4_type_signaturep,Dwarf_Off * v4_type_offsetp)39 dwarf_next_unit (Dwarf *dwarf, Dwarf_Off off, Dwarf_Off *next_off,
40 size_t *header_sizep, Dwarf_Half *versionp,
41 Dwarf_Off *abbrev_offsetp, uint8_t *address_sizep,
42 uint8_t *offset_sizep, uint64_t *v4_type_signaturep,
43 Dwarf_Off *v4_type_offsetp)
44 {
45 const bool v4_debug_types = v4_type_signaturep != NULL;
46 return __libdw_next_unit (dwarf, v4_debug_types, off, next_off,
47 header_sizep, versionp, NULL,
48 abbrev_offsetp, address_sizep, offset_sizep,
49 v4_type_signaturep, v4_type_offsetp);
50 }
INTDEF(dwarf_next_unit)51 INTDEF(dwarf_next_unit)
52
53 int
54 internal_function
55 __libdw_next_unit (Dwarf *dwarf, bool v4_debug_types, Dwarf_Off off,
56 Dwarf_Off *next_off, size_t *header_sizep,
57 Dwarf_Half *versionp, uint8_t *unit_typep,
58 Dwarf_Off *abbrev_offsetp, uint8_t *address_sizep,
59 uint8_t *offset_sizep, uint64_t *unit_id8p,
60 Dwarf_Off *subdie_offsetp)
61 {
62 /* Note that debug_type units come from .debug_types in DWARF < 5 and
63 from .debug_info in DWARF >= 5. If the user requested the
64 v4_type_signature we return from .debug_types always. If no signature
65 is requested we return units (any type) from .debug_info. */
66 const size_t sec_idx = v4_debug_types ? IDX_debug_types : IDX_debug_info;
67
68 /* Maybe there has been an error before. */
69 if (dwarf == NULL)
70 return -1;
71
72 /* If we reached the end before don't do anything. */
73 if (off == (Dwarf_Off) -1l
74 || unlikely (dwarf->sectiondata[sec_idx] == NULL)
75 /* Make sure there is enough space in the .debug_info section
76 for at least the initial word. We cannot test the rest since
77 we don't know yet whether this is a 64-bit object or not. */
78 || unlikely (off + 4 >= dwarf->sectiondata[sec_idx]->d_size))
79 {
80 *next_off = (Dwarf_Off) -1l;
81 return 1;
82 }
83
84 /* This points into the .debug_info or .debug_types section to the
85 beginning of the CU entry. */
86 const unsigned char *data = dwarf->sectiondata[sec_idx]->d_buf;
87 const unsigned char *bytes = data + off;
88 const unsigned char *bytes_end = data + dwarf->sectiondata[sec_idx]->d_size;
89
90 /* The format of the CU header is described in dwarf2p1 7.5.1 and
91 changed in DWARFv5 (to include unit type, switch location of some
92 fields and add some optional fields).
93
94 1. A 4-byte or 12-byte unsigned integer representing the length
95 of the .debug_info contribution for that compilation unit, not
96 including the length field itself. In the 32-bit DWARF format,
97 this is a 4-byte unsigned integer (which must be less than
98 0xfffffff0); in the 64-bit DWARF format, this consists of the
99 4-byte value 0xffffffff followed by an 8-byte unsigned integer
100 that gives the actual length (see Section 7.2.2). This field
101 indicates whether this unit is 32-bit of 64-bit DWARF, which
102 affects all other offset fields in this header.
103
104 2. A 2-byte unsigned integer representing the version of the
105 DWARF information for that compilation unit. For DWARF Version
106 2.1, the value in this field is 2 (3 for v3, 4 for v4, 5 for v5).
107 This fields determines the order of the next fields and whether
108 there are any optional fields in this header.
109
110 3. For DWARF 2, 3 and 4 (including v4 type units):
111 A 4-byte or 8-byte unsigned offset into the .debug_abbrev
112 section. This offset associates the compilation unit with a
113 particular set of debugging information entry abbreviations. In
114 the 32-bit DWARF format, this is a 4-byte unsigned length; in
115 the 64-bit DWARF format, this is an 8-byte unsigned length (see
116 Section 7.4).
117
118 For DWARF 5:
119 A 1-byte unsigned integer representing the unit (header) type.
120 This field determines what the optional fields in the header
121 represent. If this is an unknown unit type then we cannot
122 assume anything about the rest of the unit (header).
123
124 4. For all DWARF versions (including v4 type units):
125 A 1-byte unsigned integer representing the size in bytes of
126 an address on the target architecture. If the system uses
127 segmented addressing, this value represents the size of the
128 offset portion of an address. This is the last field in the header
129 for DWARF versions 2, 3 and 4 (except for v4 type units).
130
131 5. For DWARF 5 only (this is field 3 for DWARF 2, 3, 4 and v4 types):
132 A 4-byte or 8-byte unsigned offset into the .debug_abbrev
133 section. This offset associates the compilation unit with a
134 particular set of debugging information entry abbreviations. In
135 the 32-bit DWARF format, this is a 4-byte unsigned length; in
136 the 64-bit DWARF format, this is an 8-byte unsigned length.
137
138 6. For v4 type units (this is really field 5 for v4 types) and
139 DWARF 5 optional (skeleton, split_compile, type and
140 split_type): An 8 byte (opaque) integer constant value. For
141 v4 and v5 type units this is the type signature. For skeleton
142 and split compile units this is the compilation ID.
143
144 7. For v4 type units (this is really field 6 for v4 types) and
145 DWARF 5 optional (type and split_type) and v4 type units:
146 A 4-byte or 8-byte unsigned offset. In the 32-bit DWARF format,
147 this is a 4-byte unsigned length; in the 64-bit DWARF format,
148 this is an 8-byte unsigned length. This is the type DIE offset
149 (which is not necessarily the first DIE in the unit).
150 */
151
152 uint64_t length = read_4ubyte_unaligned_inc (dwarf, bytes);
153 size_t offset_size = 4;
154 /* Lengths of 0xfffffff0 - 0xffffffff are escape codes. Oxffffffff is
155 used to indicate that 64-bit dwarf information is being used, the
156 other values are currently reserved. */
157 if (length == DWARF3_LENGTH_64_BIT)
158 offset_size = 8;
159 else if (unlikely (length >= DWARF3_LENGTH_MIN_ESCAPE_CODE
160 && length <= DWARF3_LENGTH_MAX_ESCAPE_CODE))
161 {
162 invalid:
163 __libdw_seterrno (DWARF_E_INVALID_DWARF);
164 return -1;
165 }
166
167 if (length == DWARF3_LENGTH_64_BIT)
168 {
169 /* This is a 64-bit DWARF format. */
170 if (bytes_end - bytes < 8)
171 goto invalid;
172 length = read_8ubyte_unaligned_inc (dwarf, bytes);
173 }
174
175 /* Read the version stamp. Always a 16-bit value. */
176 if (bytes_end - bytes < 2)
177 goto invalid;
178 uint_fast16_t version = read_2ubyte_unaligned_inc (dwarf, bytes);
179
180 /* We keep unit_type at zero for older DWARF since we cannot
181 easily guess whether it is a compile or partial unit. */
182 uint8_t unit_type = 0;
183 if (version >= 5)
184 {
185 if (bytes_end - bytes < 1)
186 goto invalid;
187 unit_type = *bytes++;
188 }
189
190 /* All these are optional. */
191 Dwarf_Off subdie_off = 0;
192 uint64_t sig_id = 0;
193 Dwarf_Off abbrev_offset = 0;
194 uint8_t address_size = 0;
195
196 if (version < 2 || version > 5
197 || (version == 5 && ! (unit_type == DW_UT_compile
198 || unit_type == DW_UT_partial
199 || unit_type == DW_UT_skeleton
200 || unit_type == DW_UT_split_compile
201 || unit_type == DW_UT_type
202 || unit_type == DW_UT_split_type)))
203 {
204 /* We cannot really know more about the header. Just report
205 the length of the unit, version and unit type. */
206 goto done;
207 }
208
209 /* We have to guess the unit_type. But we don't have a real CUDIE. */
210 if (version < 5)
211 unit_type = v4_debug_types ? DW_UT_type : DW_UT_compile;
212
213 /* Now we know how large the header is (should be). */
214 if (unlikely (__libdw_first_die_from_cu_start (off, offset_size, version,
215 unit_type)
216 >= dwarf->sectiondata[sec_idx]->d_size))
217 {
218 *next_off = -1;
219 return 1;
220 }
221
222 /* The address size. Always an 8-bit value.
223 Comes after abbrev_offset for version < 5, otherwise unit type
224 and address size (if a known unit type) comes before abbrev_offset. */
225 if (version >= 5)
226 address_size = *bytes++;
227
228 /* Get offset in .debug_abbrev. Note that the size of the entry
229 depends on whether this is a 32-bit or 64-bit DWARF definition. */
230 if (__libdw_read_offset_inc (dwarf, sec_idx, &bytes, offset_size,
231 &abbrev_offset, IDX_debug_abbrev, 0))
232 return -1;
233
234 if (version < 5)
235 address_size = *bytes++;
236
237 /* Extra fields, signature/id and type offset/padding. */
238 if (v4_debug_types
239 || (version >= 5
240 && (unit_type == DW_UT_skeleton || unit_type == DW_UT_split_compile
241 || unit_type == DW_UT_type || unit_type == DW_UT_split_type)))
242 {
243 sig_id = read_8ubyte_unaligned_inc (dwarf, bytes);
244
245 if ((v4_debug_types
246 || unit_type == DW_UT_type || unit_type == DW_UT_split_type))
247 {
248 if (__libdw_read_offset_inc (dwarf, sec_idx, &bytes, offset_size,
249 &subdie_off, sec_idx, 0))
250 return -1;
251
252 /* Validate that the TYPE_OFFSET points past the header. */
253 if (unlikely (subdie_off < (size_t) (bytes - (data + off))))
254 goto invalid;
255 }
256 }
257
258 done:
259 if (unit_id8p != NULL)
260 *unit_id8p = sig_id;
261
262 if (subdie_offsetp != NULL)
263 *subdie_offsetp = subdie_off;
264
265 /* Store the header length. This is really how much we have read
266 from the header. If we didn't recognize the unit type the
267 header might actually be bigger. */
268 if (header_sizep != NULL)
269 *header_sizep = bytes - (data + off);
270
271 if (versionp != NULL)
272 *versionp = version;
273
274 if (unit_typep != NULL)
275 *unit_typep = unit_type;
276
277 if (abbrev_offsetp != NULL)
278 *abbrev_offsetp = abbrev_offset;
279
280 if (address_sizep != NULL)
281 *address_sizep = address_size;
282
283 /* Store the offset size. */
284 if (offset_sizep != NULL)
285 *offset_sizep = offset_size;
286
287 /* The length of the unit doesn't include the length field itself.
288 The length field is either, with offset == 4: 2 * 4 - 4 == 4,
289 or with offset == 8: 2 * 8 - 4 == 12. */
290 *next_off = off + 2 * offset_size - 4 + length;
291
292 /* This means that the length field is bogus, but return the CU anyway.
293 We just won't return anything after this. */
294 if (*next_off <= off)
295 *next_off = (Dwarf_Off) -1;
296
297 return 0;
298 }
299
300 int
dwarf_nextcu(Dwarf * dwarf,Dwarf_Off off,Dwarf_Off * next_off,size_t * header_sizep,Dwarf_Off * abbrev_offsetp,uint8_t * address_sizep,uint8_t * offset_sizep)301 dwarf_nextcu (Dwarf *dwarf, Dwarf_Off off, Dwarf_Off *next_off,
302 size_t *header_sizep, Dwarf_Off *abbrev_offsetp,
303 uint8_t *address_sizep, uint8_t *offset_sizep)
304 {
305 return INTUSE(dwarf_next_unit) (dwarf, off, next_off, header_sizep, NULL,
306 abbrev_offsetp, address_sizep, offset_sizep,
307 NULL, NULL);
308 }
309 INTDEF(dwarf_nextcu)
310