1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
|
// Copyright (c) 2010 Google Inc. All Rights Reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
// * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
// CFI reader author: Jim Blandy <jimb@mozilla.com> <jimb@red-bean.com>
// Implementation of dwarf2reader::LineInfo, dwarf2reader::CompilationUnit,
// and dwarf2reader::CallFrameInfo. See dwarf2reader.h for details.
#include "common/dwarf/dwarf2reader.h"
#include <assert.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <map>
#include <memory>
#include <stack>
#include <string>
#include <utility>
#include <sys/stat.h>
#include "common/dwarf/bytereader-inl.h"
#include "common/dwarf/bytereader.h"
#include "common/dwarf/line_state_machine.h"
#include "common/using_std_string.h"
namespace dwarf2reader {
CompilationUnit::CompilationUnit(const string& path,
const SectionMap& sections, uint64 offset,
ByteReader* reader, Dwarf2Handler* handler)
: path_(path), offset_from_section_start_(offset), reader_(reader),
sections_(sections), handler_(handler), abbrevs_(),
string_buffer_(NULL), string_buffer_length_(0),
str_offsets_buffer_(NULL), str_offsets_buffer_length_(0),
addr_buffer_(NULL), addr_buffer_length_(0),
is_split_dwarf_(false), dwo_id_(0), dwo_name_(),
skeleton_dwo_id_(0), ranges_base_(0), addr_base_(0),
have_checked_for_dwp_(false), dwp_path_(),
dwp_byte_reader_(), dwp_reader_() {}
// Initialize a compilation unit from a .dwo or .dwp file.
// In this case, we need the .debug_addr section from the
// executable file that contains the corresponding skeleton
// compilation unit. We also inherit the Dwarf2Handler from
// the executable file, and call it as if we were still
// processing the original compilation unit.
void CompilationUnit::SetSplitDwarf(const uint8_t* addr_buffer,
uint64 addr_buffer_length,
uint64 addr_base,
uint64 ranges_base,
uint64 dwo_id) {
is_split_dwarf_ = true;
addr_buffer_ = addr_buffer;
addr_buffer_length_ = addr_buffer_length;
addr_base_ = addr_base;
ranges_base_ = ranges_base;
skeleton_dwo_id_ = dwo_id;
}
// Read a DWARF2/3 abbreviation section.
// Each abbrev consists of a abbreviation number, a tag, a byte
// specifying whether the tag has children, and a list of
// attribute/form pairs.
// The list of forms is terminated by a 0 for the attribute, and a
// zero for the form. The entire abbreviation section is terminated
// by a zero for the code.
void CompilationUnit::ReadAbbrevs() {
if (abbrevs_)
return;
// First get the debug_abbrev section. ".debug_abbrev" is the name
// recommended in the DWARF spec, and used on Linux;
// "__debug_abbrev" is the name used in Mac OS X Mach-O files.
SectionMap::const_iterator iter = sections_.find(".debug_abbrev");
if (iter == sections_.end())
iter = sections_.find("__debug_abbrev");
assert(iter != sections_.end());
abbrevs_ = new std::vector<Abbrev>;
abbrevs_->resize(1);
// The only way to check whether we are reading over the end of the
// buffer would be to first compute the size of the leb128 data by
// reading it, then go back and read it again.
const uint8_t *abbrev_start = iter->second.first +
header_.abbrev_offset;
const uint8_t *abbrevptr = abbrev_start;
#ifndef NDEBUG
const uint64 abbrev_length = iter->second.second - header_.abbrev_offset;
#endif
while (1) {
CompilationUnit::Abbrev abbrev;
size_t len;
const uint64 number = reader_->ReadUnsignedLEB128(abbrevptr, &len);
if (number == 0)
break;
abbrev.number = number;
abbrevptr += len;
assert(abbrevptr < abbrev_start + abbrev_length);
const uint64 tag = reader_->ReadUnsignedLEB128(abbrevptr, &len);
abbrevptr += len;
abbrev.tag = static_cast<enum DwarfTag>(tag);
assert(abbrevptr < abbrev_start + abbrev_length);
abbrev.has_children = reader_->ReadOneByte(abbrevptr);
abbrevptr += 1;
assert(abbrevptr < abbrev_start + abbrev_length);
while (1) {
const uint64 nametemp = reader_->ReadUnsignedLEB128(abbrevptr, &len);
abbrevptr += len;
assert(abbrevptr < abbrev_start + abbrev_length);
const uint64 formtemp = reader_->ReadUnsignedLEB128(abbrevptr, &len);
abbrevptr += len;
if (nametemp == 0 && formtemp == 0)
break;
const enum DwarfAttribute name =
static_cast<enum DwarfAttribute>(nametemp);
const enum DwarfForm form = static_cast<enum DwarfForm>(formtemp);
abbrev.attributes.push_back(std::make_pair(name, form));
}
assert(abbrev.number == abbrevs_->size());
abbrevs_->push_back(abbrev);
}
}
// Skips a single DIE's attributes.
const uint8_t *CompilationUnit::SkipDIE(const uint8_t* start,
const Abbrev& abbrev) {
for (AttributeList::const_iterator i = abbrev.attributes.begin();
i != abbrev.attributes.end();
i++) {
start = SkipAttribute(start, i->second);
}
return start;
}
// Skips a single attribute form's data.
const uint8_t *CompilationUnit::SkipAttribute(const uint8_t *start,
enum DwarfForm form) {
size_t len;
switch (form) {
case DW_FORM_indirect:
form = static_cast<enum DwarfForm>(reader_->ReadUnsignedLEB128(start,
&len));
start += len;
return SkipAttribute(start, form);
case DW_FORM_flag_present:
return start;
case DW_FORM_data1:
case DW_FORM_flag:
case DW_FORM_ref1:
return start + 1;
case DW_FORM_ref2:
case DW_FORM_data2:
return start + 2;
case DW_FORM_ref4:
case DW_FORM_data4:
return start + 4;
case DW_FORM_ref8:
case DW_FORM_data8:
case DW_FORM_ref_sig8:
return start + 8;
case DW_FORM_string:
return start + strlen(reinterpret_cast<const char *>(start)) + 1;
case DW_FORM_udata:
case DW_FORM_ref_udata:
case DW_FORM_GNU_str_index:
case DW_FORM_GNU_addr_index:
reader_->ReadUnsignedLEB128(start, &len);
return start + len;
case DW_FORM_sdata:
reader_->ReadSignedLEB128(start, &len);
return start + len;
case DW_FORM_addr:
return start + reader_->AddressSize();
case DW_FORM_ref_addr:
// DWARF2 and 3/4 differ on whether ref_addr is address size or
// offset size.
assert(header_.version >= 2);
if (header_.version == 2) {
return start + reader_->AddressSize();
} else if (header_.version >= 3) {
return start + reader_->OffsetSize();
}
break;
case DW_FORM_block1:
return start + 1 + reader_->ReadOneByte(start);
case DW_FORM_block2:
return start + 2 + reader_->ReadTwoBytes(start);
case DW_FORM_block4:
return start + 4 + reader_->ReadFourBytes(start);
case DW_FORM_block:
case DW_FORM_exprloc: {
uint64 size = reader_->ReadUnsignedLEB128(start, &len);
return start + size + len;
}
case DW_FORM_strp:
case DW_FORM_sec_offset:
return start + reader_->OffsetSize();
}
fprintf(stderr,"Unhandled form type");
return NULL;
}
// Read a DWARF2/3 header.
// The header is variable length in DWARF3 (and DWARF2 as extended by
// most compilers), and consists of an length field, a version number,
// the offset in the .debug_abbrev section for our abbrevs, and an
// address size.
void CompilationUnit::ReadHeader() {
const uint8_t *headerptr = buffer_;
size_t initial_length_size;
assert(headerptr + 4 < buffer_ + buffer_length_);
const uint64 initial_length
= reader_->ReadInitialLength(headerptr, &initial_length_size);
headerptr += initial_length_size;
header_.length = initial_length;
assert(headerptr + 2 < buffer_ + buffer_length_);
header_.version = reader_->ReadTwoBytes(headerptr);
headerptr += 2;
assert(headerptr + reader_->OffsetSize() < buffer_ + buffer_length_);
header_.abbrev_offset = reader_->ReadOffset(headerptr);
headerptr += reader_->OffsetSize();
// Compare against less than or equal because this may be the last
// section in the file.
assert(headerptr + 1 <= buffer_ + buffer_length_);
header_.address_size = reader_->ReadOneByte(headerptr);
reader_->SetAddressSize(header_.address_size);
headerptr += 1;
after_header_ = headerptr;
// This check ensures that we don't have to do checking during the
// reading of DIEs. header_.length does not include the size of the
// initial length.
assert(buffer_ + initial_length_size + header_.length <=
buffer_ + buffer_length_);
}
uint64 CompilationUnit::Start() {
// First get the debug_info section. ".debug_info" is the name
// recommended in the DWARF spec, and used on Linux; "__debug_info"
// is the name used in Mac OS X Mach-O files.
SectionMap::const_iterator iter = sections_.find(".debug_info");
if (iter == sections_.end())
iter = sections_.find("__debug_info");
assert(iter != sections_.end());
// Set up our buffer
buffer_ = iter->second.first + offset_from_section_start_;
buffer_length_ = iter->second.second - offset_from_section_start_;
// Read the header
ReadHeader();
// Figure out the real length from the end of the initial length to
// the end of the compilation unit, since that is the value we
// return.
uint64 ourlength = header_.length;
if (reader_->OffsetSize() == 8)
ourlength += 12;
else
ourlength += 4;
// See if the user wants this compilation unit, and if not, just return.
if (!handler_->StartCompilationUnit(offset_from_section_start_,
reader_->AddressSize(),
reader_->OffsetSize(),
header_.length,
header_.version))
return ourlength;
// Otherwise, continue by reading our abbreviation entries.
ReadAbbrevs();
// Set the string section if we have one. ".debug_str" is the name
// recommended in the DWARF spec, and used on Linux; "__debug_str"
// is the name used in Mac OS X Mach-O files.
iter = sections_.find(".debug_str");
if (iter == sections_.end())
iter = sections_.find("__debug_str");
if (iter != sections_.end()) {
string_buffer_ = iter->second.first;
string_buffer_length_ = iter->second.second;
}
// Set the string offsets section if we have one.
iter = sections_.find(".debug_str_offsets");
if (iter != sections_.end()) {
str_offsets_buffer_ = iter->second.first;
str_offsets_buffer_length_ = iter->second.second;
}
// Set the address section if we have one.
iter = sections_.find(".debug_addr");
if (iter != sections_.end()) {
addr_buffer_ = iter->second.first;
addr_buffer_length_ = iter->second.second;
}
// Now that we have our abbreviations, start processing DIE's.
ProcessDIEs();
// If this is a skeleton compilation unit generated with split DWARF,
// and the client needs the full debug info, we need to find the full
// compilation unit in a .dwo or .dwp file.
if (!is_split_dwarf_
&& dwo_name_ != NULL
&& handler_->NeedSplitDebugInfo())
ProcessSplitDwarf();
return ourlength;
}
// If one really wanted, you could merge SkipAttribute and
// ProcessAttribute
// This is all boring data manipulation and calling of the handler.
const uint8_t *CompilationUnit::ProcessAttribute(
uint64 dieoffset, const uint8_t *start, enum DwarfAttribute attr,
enum DwarfForm form) {
size_t len;
switch (form) {
// DW_FORM_indirect is never used because it is such a space
// waster.
case DW_FORM_indirect:
form = static_cast<enum DwarfForm>(reader_->ReadUnsignedLEB128(start,
&len));
start += len;
return ProcessAttribute(dieoffset, start, attr, form);
case DW_FORM_flag_present:
ProcessAttributeUnsigned(dieoffset, attr, form, 1);
return start;
case DW_FORM_data1:
case DW_FORM_flag:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadOneByte(start));
return start + 1;
case DW_FORM_data2:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadTwoBytes(start));
return start + 2;
case DW_FORM_data4:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadFourBytes(start));
return start + 4;
case DW_FORM_data8:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadEightBytes(start));
return start + 8;
case DW_FORM_string: {
const char *str = reinterpret_cast<const char *>(start);
ProcessAttributeString(dieoffset, attr, form, str);
return start + strlen(str) + 1;
}
case DW_FORM_udata:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadUnsignedLEB128(start, &len));
return start + len;
case DW_FORM_sdata:
ProcessAttributeSigned(dieoffset, attr, form,
reader_->ReadSignedLEB128(start, &len));
return start + len;
case DW_FORM_addr:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadAddress(start));
return start + reader_->AddressSize();
case DW_FORM_sec_offset:
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadOffset(start));
return start + reader_->OffsetSize();
case DW_FORM_ref1:
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadOneByte(start)
+ offset_from_section_start_);
return start + 1;
case DW_FORM_ref2:
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadTwoBytes(start)
+ offset_from_section_start_);
return start + 2;
case DW_FORM_ref4:
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadFourBytes(start)
+ offset_from_section_start_);
return start + 4;
case DW_FORM_ref8:
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadEightBytes(start)
+ offset_from_section_start_);
return start + 8;
case DW_FORM_ref_udata:
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadUnsignedLEB128(start,
&len)
+ offset_from_section_start_);
return start + len;
case DW_FORM_ref_addr:
// DWARF2 and 3/4 differ on whether ref_addr is address size or
// offset size.
assert(header_.version >= 2);
if (header_.version == 2) {
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadAddress(start));
return start + reader_->AddressSize();
} else if (header_.version >= 3) {
handler_->ProcessAttributeReference(dieoffset, attr, form,
reader_->ReadOffset(start));
return start + reader_->OffsetSize();
}
break;
case DW_FORM_ref_sig8:
handler_->ProcessAttributeSignature(dieoffset, attr, form,
reader_->ReadEightBytes(start));
return start + 8;
case DW_FORM_block1: {
uint64 datalen = reader_->ReadOneByte(start);
handler_->ProcessAttributeBuffer(dieoffset, attr, form, start + 1,
datalen);
return start + 1 + datalen;
}
case DW_FORM_block2: {
uint64 datalen = reader_->ReadTwoBytes(start);
handler_->ProcessAttributeBuffer(dieoffset, attr, form, start + 2,
datalen);
return start + 2 + datalen;
}
case DW_FORM_block4: {
uint64 datalen = reader_->ReadFourBytes(start);
handler_->ProcessAttributeBuffer(dieoffset, attr, form, start + 4,
datalen);
return start + 4 + datalen;
}
case DW_FORM_block:
case DW_FORM_exprloc: {
uint64 datalen = reader_->ReadUnsignedLEB128(start, &len);
handler_->ProcessAttributeBuffer(dieoffset, attr, form, start + len,
datalen);
return start + datalen + len;
}
case DW_FORM_strp: {
assert(string_buffer_ != NULL);
const uint64 offset = reader_->ReadOffset(start);
assert(string_buffer_ + offset < string_buffer_ + string_buffer_length_);
const char *str = reinterpret_cast<const char *>(string_buffer_ + offset);
ProcessAttributeString(dieoffset, attr, form, str);
return start + reader_->OffsetSize();
}
case DW_FORM_GNU_str_index: {
uint64 str_index = reader_->ReadUnsignedLEB128(start, &len);
const uint8_t* offset_ptr =
str_offsets_buffer_ + str_index * reader_->OffsetSize();
const uint64 offset = reader_->ReadOffset(offset_ptr);
if (offset >= string_buffer_length_) {
return NULL;
}
const char* str = reinterpret_cast<const char *>(string_buffer_) + offset;
ProcessAttributeString(dieoffset, attr, form, str);
return start + len;
break;
}
case DW_FORM_GNU_addr_index: {
uint64 addr_index = reader_->ReadUnsignedLEB128(start, &len);
const uint8_t* addr_ptr =
addr_buffer_ + addr_base_ + addr_index * reader_->AddressSize();
ProcessAttributeUnsigned(dieoffset, attr, form,
reader_->ReadAddress(addr_ptr));
return start + len;
}
}
fprintf(stderr, "Unhandled form type\n");
return NULL;
}
const uint8_t *CompilationUnit::ProcessDIE(uint64 dieoffset,
const uint8_t *start,
const Abbrev& abbrev) {
for (AttributeList::const_iterator i = abbrev.attributes.begin();
i != abbrev.attributes.end();
i++) {
start = ProcessAttribute(dieoffset, start, i->first, i->second);
}
// If this is a compilation unit in a split DWARF object, verify that
// the dwo_id matches. If it does not match, we will ignore this
// compilation unit.
if (abbrev.tag == DW_TAG_compile_unit
&& is_split_dwarf_
&& dwo_id_ != skeleton_dwo_id_) {
return NULL;
}
return start;
}
void CompilationUnit::ProcessDIEs() {
const uint8_t *dieptr = after_header_;
size_t len;
// lengthstart is the place the length field is based on.
// It is the point in the header after the initial length field
const uint8_t *lengthstart = buffer_;
// In 64 bit dwarf, the initial length is 12 bytes, because of the
// 0xffffffff at the start.
if (reader_->OffsetSize() == 8)
lengthstart += 12;
else
lengthstart += 4;
std::stack<uint64> die_stack;
while (dieptr < (lengthstart + header_.length)) {
// We give the user the absolute offset from the beginning of
// debug_info, since they need it to deal with ref_addr forms.
uint64 absolute_offset = (dieptr - buffer_) + offset_from_section_start_;
uint64 abbrev_num = reader_->ReadUnsignedLEB128(dieptr, &len);
dieptr += len;
// Abbrev == 0 represents the end of a list of children, or padding
// at the end of the compilation unit.
if (abbrev_num == 0) {
if (die_stack.size() == 0)
// If it is padding, then we are done with the compilation unit's DIEs.
return;
const uint64 offset = die_stack.top();
die_stack.pop();
handler_->EndDIE(offset);
continue;
}
const Abbrev& abbrev = abbrevs_->at(static_cast<size_t>(abbrev_num));
const enum DwarfTag tag = abbrev.tag;
if (!handler_->StartDIE(absolute_offset, tag)) {
dieptr = SkipDIE(dieptr, abbrev);
} else {
dieptr = ProcessDIE(absolute_offset, dieptr, abbrev);
}
if (abbrev.has_children) {
die_stack.push(absolute_offset);
} else {
handler_->EndDIE(absolute_offset);
}
}
}
// Check for a valid ELF file and return the Address size.
// Returns 0 if not a valid ELF file.
inline int GetElfWidth(const ElfReader& elf) {
if (elf.IsElf32File())
return 4;
if (elf.IsElf64File())
return 8;
return 0;
}
void CompilationUnit::ProcessSplitDwarf() {
struct stat statbuf;
if (!have_checked_for_dwp_) {
// Look for a .dwp file in the same directory as the executable.
have_checked_for_dwp_ = true;
string dwp_suffix(".dwp");
dwp_path_ = path_ + dwp_suffix;
if (stat(dwp_path_.c_str(), &statbuf) != 0) {
// Fall back to a split .debug file in the same directory.
string debug_suffix(".debug");
dwp_path_ = path_;
size_t found = path_.rfind(debug_suffix);
if (found + debug_suffix.length() == path_.length())
dwp_path_ = dwp_path_.replace(found, debug_suffix.length(), dwp_suffix);
}
if (stat(dwp_path_.c_str(), &statbuf) == 0) {
ElfReader* elf = new ElfReader(dwp_path_);
int width = GetElfWidth(*elf);
if (width != 0) {
dwp_byte_reader_.reset(new ByteReader(reader_->GetEndianness()));
dwp_byte_reader_->SetAddressSize(width);
dwp_reader_.reset(new DwpReader(*dwp_byte_reader_, elf));
dwp_reader_->Initialize();
} else {
delete elf;
}
}
}
bool found_in_dwp = false;
if (dwp_reader_) {
// If we have a .dwp file, read the debug sections for the requested CU.
SectionMap sections;
dwp_reader_->ReadDebugSectionsForCU(dwo_id_, §ions);
if (!sections.empty()) {
found_in_dwp = true;
CompilationUnit dwp_comp_unit(dwp_path_, sections, 0,
dwp_byte_reader_.get(), handler_);
dwp_comp_unit.SetSplitDwarf(addr_buffer_, addr_buffer_length_, addr_base_,
ranges_base_, dwo_id_);
dwp_comp_unit.Start();
}
}
if (!found_in_dwp) {
// If no .dwp file, try to open the .dwo file.
if (stat(dwo_name_, &statbuf) == 0) {
ElfReader elf(dwo_name_);
int width = GetElfWidth(elf);
if (width != 0) {
ByteReader reader(ENDIANNESS_LITTLE);
reader.SetAddressSize(width);
SectionMap sections;
ReadDebugSectionsFromDwo(&elf, §ions);
CompilationUnit dwo_comp_unit(dwo_name_, sections, 0, &reader,
handler_);
dwo_comp_unit.SetSplitDwarf(addr_buffer_, addr_buffer_length_,
addr_base_, ranges_base_, dwo_id_);
dwo_comp_unit.Start();
}
}
}
}
void CompilationUnit::ReadDebugSectionsFromDwo(ElfReader* elf_reader,
SectionMap* sections) {
static const char* const section_names[] = {
".debug_abbrev",
".debug_info",
".debug_str_offsets",
".debug_str"
};
for (unsigned int i = 0u;
i < sizeof(section_names)/sizeof(*(section_names)); ++i) {
string base_name = section_names[i];
string dwo_name = base_name + ".dwo";
size_t section_size;
const char* section_data = elf_reader->GetSectionByName(dwo_name,
§ion_size);
if (section_data != NULL)
sections->insert(std::make_pair(
base_name, std::make_pair(
reinterpret_cast<const uint8_t *>(section_data),
section_size)));
}
}
DwpReader::DwpReader(const ByteReader& byte_reader, ElfReader* elf_reader)
: elf_reader_(elf_reader), byte_reader_(byte_reader),
cu_index_(NULL), cu_index_size_(0), string_buffer_(NULL),
string_buffer_size_(0), version_(0), ncolumns_(0), nunits_(0),
nslots_(0), phash_(NULL), pindex_(NULL), shndx_pool_(NULL),
offset_table_(NULL), size_table_(NULL), abbrev_data_(NULL),
abbrev_size_(0), info_data_(NULL), info_size_(0),
str_offsets_data_(NULL), str_offsets_size_(0) {}
DwpReader::~DwpReader() {
if (elf_reader_) delete elf_reader_;
}
void DwpReader::Initialize() {
cu_index_ = elf_reader_->GetSectionByName(".debug_cu_index",
&cu_index_size_);
if (cu_index_ == NULL) {
return;
}
// The .debug_str.dwo section is shared by all CUs in the file.
string_buffer_ = elf_reader_->GetSectionByName(".debug_str.dwo",
&string_buffer_size_);
version_ = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(cu_index_));
if (version_ == 1) {
nslots_ = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(cu_index_)
+ 3 * sizeof(uint32));
phash_ = cu_index_ + 4 * sizeof(uint32);
pindex_ = phash_ + nslots_ * sizeof(uint64);
shndx_pool_ = pindex_ + nslots_ * sizeof(uint32);
if (shndx_pool_ >= cu_index_ + cu_index_size_) {
version_ = 0;
}
} else if (version_ == 2) {
ncolumns_ = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(cu_index_) + sizeof(uint32));
nunits_ = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(cu_index_) + 2 * sizeof(uint32));
nslots_ = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(cu_index_) + 3 * sizeof(uint32));
phash_ = cu_index_ + 4 * sizeof(uint32);
pindex_ = phash_ + nslots_ * sizeof(uint64);
offset_table_ = pindex_ + nslots_ * sizeof(uint32);
size_table_ = offset_table_ + ncolumns_ * (nunits_ + 1) * sizeof(uint32);
abbrev_data_ = elf_reader_->GetSectionByName(".debug_abbrev.dwo",
&abbrev_size_);
info_data_ = elf_reader_->GetSectionByName(".debug_info.dwo", &info_size_);
str_offsets_data_ = elf_reader_->GetSectionByName(".debug_str_offsets.dwo",
&str_offsets_size_);
if (size_table_ >= cu_index_ + cu_index_size_) {
version_ = 0;
}
}
}
void DwpReader::ReadDebugSectionsForCU(uint64 dwo_id,
SectionMap* sections) {
if (version_ == 1) {
int slot = LookupCU(dwo_id);
if (slot == -1) {
return;
}
// The index table points to the section index pool, where we
// can read a list of section indexes for the debug sections
// for the CU whose dwo_id we are looking for.
int index = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(pindex_)
+ slot * sizeof(uint32));
const char* shndx_list = shndx_pool_ + index * sizeof(uint32);
for (;;) {
if (shndx_list >= cu_index_ + cu_index_size_) {
version_ = 0;
return;
}
unsigned int shndx = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(shndx_list));
shndx_list += sizeof(uint32);
if (shndx == 0)
break;
const char* section_name = elf_reader_->GetSectionName(shndx);
size_t section_size;
const char* section_data;
// We're only interested in these four debug sections.
// The section names in the .dwo file end with ".dwo", but we
// add them to the sections table with their normal names.
if (!strncmp(section_name, ".debug_abbrev", strlen(".debug_abbrev"))) {
section_data = elf_reader_->GetSectionByIndex(shndx, §ion_size);
sections->insert(std::make_pair(
".debug_abbrev",
std::make_pair(reinterpret_cast<const uint8_t *> (section_data),
section_size)));
} else if (!strncmp(section_name, ".debug_info", strlen(".debug_info"))) {
section_data = elf_reader_->GetSectionByIndex(shndx, §ion_size);
sections->insert(std::make_pair(
".debug_info",
std::make_pair(reinterpret_cast<const uint8_t *> (section_data),
section_size)));
} else if (!strncmp(section_name, ".debug_str_offsets",
strlen(".debug_str_offsets"))) {
section_data = elf_reader_->GetSectionByIndex(shndx, §ion_size);
sections->insert(std::make_pair(
".debug_str_offsets",
std::make_pair(reinterpret_cast<const uint8_t *> (section_data),
section_size)));
}
}
sections->insert(std::make_pair(
".debug_str",
std::make_pair(reinterpret_cast<const uint8_t *> (string_buffer_),
string_buffer_size_)));
} else if (version_ == 2) {
uint32 index = LookupCUv2(dwo_id);
if (index == 0) {
return;
}
// The index points to a row in each of the section offsets table
// and the section size table, where we can read the offsets and sizes
// of the contributions to each debug section from the CU whose dwo_id
// we are looking for. Row 0 of the section offsets table has the
// section ids for each column of the table. The size table begins
// with row 1.
const char* id_row = offset_table_;
const char* offset_row = offset_table_
+ index * ncolumns_ * sizeof(uint32);
const char* size_row =
size_table_ + (index - 1) * ncolumns_ * sizeof(uint32);
if (size_row + ncolumns_ * sizeof(uint32) > cu_index_ + cu_index_size_) {
version_ = 0;
return;
}
for (unsigned int col = 0u; col < ncolumns_; ++col) {
uint32 section_id =
byte_reader_.ReadFourBytes(reinterpret_cast<const uint8_t *>(id_row)
+ col * sizeof(uint32));
uint32 offset = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(offset_row)
+ col * sizeof(uint32));
uint32 size = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(size_row) + col * sizeof(uint32));
if (section_id == DW_SECT_ABBREV) {
sections->insert(std::make_pair(
".debug_abbrev",
std::make_pair(reinterpret_cast<const uint8_t *> (abbrev_data_)
+ offset, size)));
} else if (section_id == DW_SECT_INFO) {
sections->insert(std::make_pair(
".debug_info",
std::make_pair(reinterpret_cast<const uint8_t *> (info_data_)
+ offset, size)));
} else if (section_id == DW_SECT_STR_OFFSETS) {
sections->insert(std::make_pair(
".debug_str_offsets",
std::make_pair(reinterpret_cast<const uint8_t *> (str_offsets_data_)
+ offset, size)));
}
}
sections->insert(std::make_pair(
".debug_str",
std::make_pair(reinterpret_cast<const uint8_t *> (string_buffer_),
string_buffer_size_)));
}
}
int DwpReader::LookupCU(uint64 dwo_id) {
uint32 slot = static_cast<uint32>(dwo_id) & (nslots_ - 1);
uint64 probe = byte_reader_.ReadEightBytes(
reinterpret_cast<const uint8_t *>(phash_) + slot * sizeof(uint64));
if (probe != 0 && probe != dwo_id) {
uint32 secondary_hash =
(static_cast<uint32>(dwo_id >> 32) & (nslots_ - 1)) | 1;
do {
slot = (slot + secondary_hash) & (nslots_ - 1);
probe = byte_reader_.ReadEightBytes(
reinterpret_cast<const uint8_t *>(phash_) + slot * sizeof(uint64));
} while (probe != 0 && probe != dwo_id);
}
if (probe == 0)
return -1;
return slot;
}
uint32 DwpReader::LookupCUv2(uint64 dwo_id) {
uint32 slot = static_cast<uint32>(dwo_id) & (nslots_ - 1);
uint64 probe = byte_reader_.ReadEightBytes(
reinterpret_cast<const uint8_t *>(phash_) + slot * sizeof(uint64));
uint32 index = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(pindex_) + slot * sizeof(uint32));
if (index != 0 && probe != dwo_id) {
uint32 secondary_hash =
(static_cast<uint32>(dwo_id >> 32) & (nslots_ - 1)) | 1;
do {
slot = (slot + secondary_hash) & (nslots_ - 1);
probe = byte_reader_.ReadEightBytes(
reinterpret_cast<const uint8_t *>(phash_) + slot * sizeof(uint64));
index = byte_reader_.ReadFourBytes(
reinterpret_cast<const uint8_t *>(pindex_) + slot * sizeof(uint32));
} while (index != 0 && probe != dwo_id);
}
return index;
}
LineInfo::LineInfo(const uint8_t *buffer, uint64 buffer_length,
ByteReader* reader, LineInfoHandler* handler):
handler_(handler), reader_(reader), buffer_(buffer) {
#ifndef NDEBUG
buffer_length_ = buffer_length;
#endif
header_.std_opcode_lengths = NULL;
}
uint64 LineInfo::Start() {
ReadHeader();
ReadLines();
return after_header_ - buffer_;
}
// The header for a debug_line section is mildly complicated, because
// the line info is very tightly encoded.
void LineInfo::ReadHeader() {
const uint8_t *lineptr = buffer_;
size_t initial_length_size;
const uint64 initial_length
= reader_->ReadInitialLength(lineptr, &initial_length_size);
lineptr += initial_length_size;
header_.total_length = initial_length;
assert(buffer_ + initial_length_size + header_.total_length <=
buffer_ + buffer_length_);
// Address size *must* be set by CU ahead of time.
assert(reader_->AddressSize() != 0);
header_.version = reader_->ReadTwoBytes(lineptr);
lineptr += 2;
header_.prologue_length = reader_->ReadOffset(lineptr);
lineptr += reader_->OffsetSize();
header_.min_insn_length = reader_->ReadOneByte(lineptr);
lineptr += 1;
if (header_.version >= 4) {
__attribute__((unused)) uint8 max_ops_per_insn =
reader_->ReadOneByte(lineptr);
++lineptr;
assert(max_ops_per_insn == 1);
}
header_.default_is_stmt = reader_->ReadOneByte(lineptr);
lineptr += 1;
header_.line_base = *reinterpret_cast<const int8*>(lineptr);
lineptr += 1;
header_.line_range = reader_->ReadOneByte(lineptr);
lineptr += 1;
header_.opcode_base = reader_->ReadOneByte(lineptr);
lineptr += 1;
header_.std_opcode_lengths = new std::vector<unsigned char>;
header_.std_opcode_lengths->resize(header_.opcode_base + 1);
(*header_.std_opcode_lengths)[0] = 0;
for (int i = 1; i < header_.opcode_base; i++) {
(*header_.std_opcode_lengths)[i] = reader_->ReadOneByte(lineptr);
lineptr += 1;
}
// It is legal for the directory entry table to be empty.
if (*lineptr) {
uint32 dirindex = 1;
while (*lineptr) {
const char *dirname = reinterpret_cast<const char *>(lineptr);
handler_->DefineDir(dirname, dirindex);
lineptr += strlen(dirname) + 1;
dirindex++;
}
}
lineptr++;
// It is also legal for the file entry table to be empty.
if (*lineptr) {
uint32 fileindex = 1;
size_t len;
while (*lineptr) {
const char *filename = reinterpret_cast<const char *>(lineptr);
lineptr += strlen(filename) + 1;
uint64 dirindex = reader_->ReadUnsignedLEB128(lineptr, &len);
lineptr += len;
uint64 mod_time = reader_->ReadUnsignedLEB128(lineptr, &len);
lineptr += len;
uint64 filelength = reader_->ReadUnsignedLEB128(lineptr, &len);
lineptr += len;
handler_->DefineFile(filename, fileindex, static_cast<uint32>(dirindex),
mod_time, filelength);
fileindex++;
}
}
lineptr++;
after_header_ = lineptr;
}
/* static */
bool LineInfo::ProcessOneOpcode(ByteReader* reader,
LineInfoHandler* handler,
const struct LineInfoHeader &header,
const uint8_t *start,
struct LineStateMachine* lsm,
size_t* len,
uintptr pc,
bool *lsm_passes_pc) {
size_t oplen = 0;
size_t templen;
uint8 opcode = reader->ReadOneByte(start);
oplen++;
start++;
// If the opcode is great than the opcode_base, it is a special
// opcode. Most line programs consist mainly of special opcodes.
if (opcode >= header.opcode_base) {
opcode -= header.opcode_base;
const int64 advance_address = (opcode / header.line_range)
* header.min_insn_length;
const int32 advance_line = (opcode % header.line_range)
+ header.line_base;
// Check if the lsm passes "pc". If so, mark it as passed.
if (lsm_passes_pc &&
lsm->address <= pc && pc < lsm->address + advance_address) {
*lsm_passes_pc = true;
}
lsm->address += advance_address;
lsm->line_num += advance_line;
lsm->basic_block = true;
*len = oplen;
return true;
}
// Otherwise, we have the regular opcodes
switch (opcode) {
case DW_LNS_copy: {
lsm->basic_block = false;
*len = oplen;
return true;
}
case DW_LNS_advance_pc: {
uint64 advance_address = reader->ReadUnsignedLEB128(start, &templen);
oplen += templen;
// Check if the lsm passes "pc". If so, mark it as passed.
if (lsm_passes_pc && lsm->address <= pc &&
pc < lsm->address + header.min_insn_length * advance_address) {
*lsm_passes_pc = true;
}
lsm->address += header.min_insn_length * advance_address;
}
break;
case DW_LNS_advance_line: {
const int64 advance_line = reader->ReadSignedLEB128(start, &templen);
oplen += templen;
lsm->line_num += static_cast<int32>(advance_line);
// With gcc 4.2.1, we can get the line_no here for the first time
// since DW_LNS_advance_line is called after DW_LNE_set_address is
// called. So we check if the lsm passes "pc" here, not in
// DW_LNE_set_address.
if (lsm_passes_pc && lsm->address == pc) {
*lsm_passes_pc = true;
}
}
break;
case DW_LNS_set_file: {
const uint64 fileno = reader->ReadUnsignedLEB128(start, &templen);
oplen += templen;
lsm->file_num = static_cast<uint32>(fileno);
}
break;
case DW_LNS_set_column: {
const uint64 colno = reader->ReadUnsignedLEB128(start, &templen);
oplen += templen;
lsm->column_num = static_cast<uint32>(colno);
}
break;
case DW_LNS_negate_stmt: {
lsm->is_stmt = !lsm->is_stmt;
}
break;
case DW_LNS_set_basic_block: {
lsm->basic_block = true;
}
break;
case DW_LNS_fixed_advance_pc: {
const uint16 advance_address = reader->ReadTwoBytes(start);
oplen += 2;
// Check if the lsm passes "pc". If so, mark it as passed.
if (lsm_passes_pc &&
lsm->address <= pc && pc < lsm->address + advance_address) {
*lsm_passes_pc = true;
}
lsm->address += advance_address;
}
break;
case DW_LNS_const_add_pc: {
const int64 advance_address = header.min_insn_length
* ((255 - header.opcode_base)
/ header.line_range);
// Check if the lsm passes "pc". If so, mark it as passed.
if (lsm_passes_pc &&
lsm->address <= pc && pc < lsm->address + advance_address) {
*lsm_passes_pc = true;
}
lsm->address += advance_address;
}
break;
case DW_LNS_extended_op: {
const uint64 extended_op_len = reader->ReadUnsignedLEB128(start,
&templen);
start += templen;
oplen += templen + extended_op_len;
const uint64 extended_op = reader->ReadOneByte(start);
start++;
switch (extended_op) {
case DW_LNE_end_sequence: {
lsm->end_sequence = true;
*len = oplen;
return true;
}
break;
case DW_LNE_set_address: {
// With gcc 4.2.1, we cannot tell the line_no here since
// DW_LNE_set_address is called before DW_LNS_advance_line is
// called. So we do not check if the lsm passes "pc" here. See
// also the comment in DW_LNS_advance_line.
uint64 address = reader->ReadAddress(start);
lsm->address = address;
}
break;
case DW_LNE_define_file: {
const char *filename = reinterpret_cast<const char *>(start);
templen = strlen(filename) + 1;
start += templen;
uint64 dirindex = reader->ReadUnsignedLEB128(start, &templen);
oplen += templen;
const uint64 mod_time = reader->ReadUnsignedLEB128(start,
&templen);
oplen += templen;
const uint64 filelength = reader->ReadUnsignedLEB128(start,
&templen);
oplen += templen;
if (handler) {
handler->DefineFile(filename, -1, static_cast<uint32>(dirindex),
mod_time, filelength);
}
}
break;
}
}
break;
default: {
// Ignore unknown opcode silently
if (header.std_opcode_lengths) {
for (int i = 0; i < (*header.std_opcode_lengths)[opcode]; i++) {
reader->ReadUnsignedLEB128(start, &templen);
start += templen;
oplen += templen;
}
}
}
break;
}
*len = oplen;
return false;
}
void LineInfo::ReadLines() {
struct LineStateMachine lsm;
// lengthstart is the place the length field is based on.
// It is the point in the header after the initial length field
const uint8_t *lengthstart = buffer_;
// In 64 bit dwarf, the initial length is 12 bytes, because of the
// 0xffffffff at the start.
if (reader_->OffsetSize() == 8)
lengthstart += 12;
else
lengthstart += 4;
const uint8_t *lineptr = after_header_;
lsm.Reset(header_.default_is_stmt);
// The LineInfoHandler interface expects each line's length along
// with its address, but DWARF only provides addresses (sans
// length), and an end-of-sequence address; one infers the length
// from the next address. So we report a line only when we get the
// next line's address, or the end-of-sequence address.
bool have_pending_line = false;
uint64 pending_address = 0;
uint32 pending_file_num = 0, pending_line_num = 0, pending_column_num = 0;
while (lineptr < lengthstart + header_.total_length) {
size_t oplength;
bool add_row = ProcessOneOpcode(reader_, handler_, header_,
lineptr, &lsm, &oplength, (uintptr)-1,
NULL);
if (add_row) {
if (have_pending_line)
handler_->AddLine(pending_address, lsm.address - pending_address,
pending_file_num, pending_line_num,
pending_column_num);
if (lsm.end_sequence) {
lsm.Reset(header_.default_is_stmt);
have_pending_line = false;
} else {
pending_address = lsm.address;
pending_file_num = lsm.file_num;
pending_line_num = lsm.line_num;
pending_column_num = lsm.column_num;
have_pending_line = true;
}
}
lineptr += oplength;
}
after_header_ = lengthstart + header_.total_length;
}
// A DWARF rule for recovering the address or value of a register, or
// computing the canonical frame address. There is one subclass of this for
// each '*Rule' member function in CallFrameInfo::Handler.
//
// It's annoying that we have to handle Rules using pointers (because
// the concrete instances can have an arbitrary size). They're small,
// so it would be much nicer if we could just handle them by value
// instead of fretting about ownership and destruction.
//
// It seems like all these could simply be instances of std::tr1::bind,
// except that we need instances to be EqualityComparable, too.
//
// This could logically be nested within State, but then the qualified names
// get horrendous.
class CallFrameInfo::Rule {
public:
virtual ~Rule() { }
// Tell HANDLER that, at ADDRESS in the program, REGISTER can be
// recovered using this rule. If REGISTER is kCFARegister, then this rule
// describes how to compute the canonical frame address. Return what the
// HANDLER member function returned.
virtual bool Handle(Handler *handler,
uint64 address, int register) const = 0;
// Equality on rules. We use these to decide which rules we need
// to report after a DW_CFA_restore_state instruction.
virtual bool operator==(const Rule &rhs) const = 0;
bool operator!=(const Rule &rhs) const { return ! (*this == rhs); }
// Return a pointer to a copy of this rule.
virtual Rule *Copy() const = 0;
// If this is a base+offset rule, change its base register to REG.
// Otherwise, do nothing. (Ugly, but required for DW_CFA_def_cfa_register.)
virtual void SetBaseRegister(unsigned reg) { }
// If this is a base+offset rule, change its offset to OFFSET. Otherwise,
// do nothing. (Ugly, but required for DW_CFA_def_cfa_offset.)
virtual void SetOffset(long long offset) { }
};
// Rule: the value the register had in the caller cannot be recovered.
class CallFrameInfo::UndefinedRule: public CallFrameInfo::Rule {
public:
UndefinedRule() { }
~UndefinedRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->UndefinedRule(address, reg);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const UndefinedRule *our_rhs = dynamic_cast<const UndefinedRule *>(&rhs);
return (our_rhs != NULL);
}
Rule *Copy() const { return new UndefinedRule(*this); }
};
// Rule: the register's value is the same as that it had in the caller.
class CallFrameInfo::SameValueRule: public CallFrameInfo::Rule {
public:
SameValueRule() { }
~SameValueRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->SameValueRule(address, reg);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const SameValueRule *our_rhs = dynamic_cast<const SameValueRule *>(&rhs);
return (our_rhs != NULL);
}
Rule *Copy() const { return new SameValueRule(*this); }
};
// Rule: the register is saved at OFFSET from BASE_REGISTER. BASE_REGISTER
// may be CallFrameInfo::Handler::kCFARegister.
class CallFrameInfo::OffsetRule: public CallFrameInfo::Rule {
public:
OffsetRule(int base_register, long offset)
: base_register_(base_register), offset_(offset) { }
~OffsetRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->OffsetRule(address, reg, base_register_, offset_);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const OffsetRule *our_rhs = dynamic_cast<const OffsetRule *>(&rhs);
return (our_rhs &&
base_register_ == our_rhs->base_register_ &&
offset_ == our_rhs->offset_);
}
Rule *Copy() const { return new OffsetRule(*this); }
// We don't actually need SetBaseRegister or SetOffset here, since they
// are only ever applied to CFA rules, for DW_CFA_def_cfa_offset, and it
// doesn't make sense to use OffsetRule for computing the CFA: it
// computes the address at which a register is saved, not a value.
private:
int base_register_;
long offset_;
};
// Rule: the value the register had in the caller is the value of
// BASE_REGISTER plus offset. BASE_REGISTER may be
// CallFrameInfo::Handler::kCFARegister.
class CallFrameInfo::ValOffsetRule: public CallFrameInfo::Rule {
public:
ValOffsetRule(int base_register, long offset)
: base_register_(base_register), offset_(offset) { }
~ValOffsetRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->ValOffsetRule(address, reg, base_register_, offset_);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const ValOffsetRule *our_rhs = dynamic_cast<const ValOffsetRule *>(&rhs);
return (our_rhs &&
base_register_ == our_rhs->base_register_ &&
offset_ == our_rhs->offset_);
}
Rule *Copy() const { return new ValOffsetRule(*this); }
void SetBaseRegister(unsigned reg) { base_register_ = reg; }
void SetOffset(long long offset) { offset_ = offset; }
private:
int base_register_;
long offset_;
};
// Rule: the register has been saved in another register REGISTER_NUMBER_.
class CallFrameInfo::RegisterRule: public CallFrameInfo::Rule {
public:
explicit RegisterRule(int register_number)
: register_number_(register_number) { }
~RegisterRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->RegisterRule(address, reg, register_number_);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const RegisterRule *our_rhs = dynamic_cast<const RegisterRule *>(&rhs);
return (our_rhs && register_number_ == our_rhs->register_number_);
}
Rule *Copy() const { return new RegisterRule(*this); }
private:
int register_number_;
};
// Rule: EXPRESSION evaluates to the address at which the register is saved.
class CallFrameInfo::ExpressionRule: public CallFrameInfo::Rule {
public:
explicit ExpressionRule(const string &expression)
: expression_(expression) { }
~ExpressionRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->ExpressionRule(address, reg, expression_);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const ExpressionRule *our_rhs = dynamic_cast<const ExpressionRule *>(&rhs);
return (our_rhs && expression_ == our_rhs->expression_);
}
Rule *Copy() const { return new ExpressionRule(*this); }
private:
string expression_;
};
// Rule: EXPRESSION evaluates to the address at which the register is saved.
class CallFrameInfo::ValExpressionRule: public CallFrameInfo::Rule {
public:
explicit ValExpressionRule(const string &expression)
: expression_(expression) { }
~ValExpressionRule() { }
bool Handle(Handler *handler, uint64 address, int reg) const {
return handler->ValExpressionRule(address, reg, expression_);
}
bool operator==(const Rule &rhs) const {
// dynamic_cast is allowed by the Google C++ Style Guide, if the use has
// been carefully considered; cheap RTTI-like workarounds are forbidden.
const ValExpressionRule *our_rhs =
dynamic_cast<const ValExpressionRule *>(&rhs);
return (our_rhs && expression_ == our_rhs->expression_);
}
Rule *Copy() const { return new ValExpressionRule(*this); }
private:
string expression_;
};
// A map from register numbers to rules.
class CallFrameInfo::RuleMap {
public:
RuleMap() : cfa_rule_(NULL) { }
RuleMap(const RuleMap &rhs) : cfa_rule_(NULL) { *this = rhs; }
~RuleMap() { Clear(); }
RuleMap &operator=(const RuleMap &rhs);
// Set the rule for computing the CFA to RULE. Take ownership of RULE.
void SetCFARule(Rule *rule) { delete cfa_rule_; cfa_rule_ = rule; }
// Return the current CFA rule. Unlike RegisterRule, this RuleMap retains
// ownership of the rule. We use this for DW_CFA_def_cfa_offset and
// DW_CFA_def_cfa_register, and for detecting references to the CFA before
// a rule for it has been established.
Rule *CFARule() const { return cfa_rule_; }
// Return the rule for REG, or NULL if there is none. The caller takes
// ownership of the result.
Rule *RegisterRule(int reg) const;
// Set the rule for computing REG to RULE. Take ownership of RULE.
void SetRegisterRule(int reg, Rule *rule);
// Make all the appropriate calls to HANDLER as if we were changing from
// this RuleMap to NEW_RULES at ADDRESS. We use this to implement
// DW_CFA_restore_state, where lots of rules can change simultaneously.
// Return true if all handlers returned true; otherwise, return false.
bool HandleTransitionTo(Handler *handler, uint64 address,
const RuleMap &new_rules) const;
private:
// A map from register numbers to Rules.
typedef std::map<int, Rule *> RuleByNumber;
// Remove all register rules and clear cfa_rule_.
void Clear();
// The rule for computing the canonical frame address. This RuleMap owns
// this rule.
Rule *cfa_rule_;
// A map from register numbers to postfix expressions to recover
// their values. This RuleMap owns the Rules the map refers to.
RuleByNumber registers_;
};
CallFrameInfo::RuleMap &CallFrameInfo::RuleMap::operator=(const RuleMap &rhs) {
Clear();
// Since each map owns the rules it refers to, assignment must copy them.
if (rhs.cfa_rule_) cfa_rule_ = rhs.cfa_rule_->Copy();
for (RuleByNumber::const_iterator it = rhs.registers_.begin();
it != rhs.registers_.end(); it++)
registers_[it->first] = it->second->Copy();
return *this;
}
CallFrameInfo::Rule *CallFrameInfo::RuleMap::RegisterRule(int reg) const {
assert(reg != Handler::kCFARegister);
RuleByNumber::const_iterator it = registers_.find(reg);
if (it != registers_.end())
return it->second->Copy();
else
return NULL;
}
void CallFrameInfo::RuleMap::SetRegisterRule(int reg, Rule *rule) {
assert(reg != Handler::kCFARegister);
assert(rule);
Rule **slot = ®isters_[reg];
delete *slot;
*slot = rule;
}
bool CallFrameInfo::RuleMap::HandleTransitionTo(
Handler *handler,
uint64 address,
const RuleMap &new_rules) const {
// Transition from cfa_rule_ to new_rules.cfa_rule_.
if (cfa_rule_ && new_rules.cfa_rule_) {
if (*cfa_rule_ != *new_rules.cfa_rule_ &&
!new_rules.cfa_rule_->Handle(handler, address,
Handler::kCFARegister))
return false;
} else if (cfa_rule_) {
// this RuleMap has a CFA rule but new_rules doesn't.
// CallFrameInfo::Handler has no way to handle this --- and shouldn't;
// it's garbage input. The instruction interpreter should have
// detected this and warned, so take no action here.
} else if (new_rules.cfa_rule_) {
// This shouldn't be possible: NEW_RULES is some prior state, and
// there's no way to remove entries.
assert(0);
} else {
// Both CFA rules are empty. No action needed.
}
// Traverse the two maps in order by register number, and report
// whatever differences we find.
RuleByNumber::const_iterator old_it = registers_.begin();
RuleByNumber::const_iterator new_it = new_rules.registers_.begin();
while (old_it != registers_.end() && new_it != new_rules.registers_.end()) {
if (old_it->first < new_it->first) {
// This RuleMap has an entry for old_it->first, but NEW_RULES
// doesn't.
//
// This isn't really the right thing to do, but since CFI generally
// only mentions callee-saves registers, and GCC's convention for
// callee-saves registers is that they are unchanged, it's a good
// approximation.
if (!handler->SameValueRule(address, old_it->first))
return false;
old_it++;
} else if (old_it->first > new_it->first) {
// NEW_RULES has entry for new_it->first, but this RuleMap
// doesn't. This shouldn't be possible: NEW_RULES is some prior
// state, and there's no way to remove entries.
assert(0);
} else {
// Both maps have an entry for this register. Report the new
// rule if it is different.
if (*old_it->second != *new_it->second &&
!new_it->second->Handle(handler, address, new_it->first))
return false;
new_it++, old_it++;
}
}
// Finish off entries from this RuleMap with no counterparts in new_rules.
while (old_it != registers_.end()) {
if (!handler->SameValueRule(address, old_it->first))
return false;
old_it++;
}
// Since we only make transitions from a rule set to some previously
// saved rule set, and we can only add rules to the map, NEW_RULES
// must have fewer rules than *this.
assert(new_it == new_rules.registers_.end());
return true;
}
// Remove all register rules and clear cfa_rule_.
void CallFrameInfo::RuleMap::Clear() {
delete cfa_rule_;
cfa_rule_ = NULL;
for (RuleByNumber::iterator it = registers_.begin();
it != registers_.end(); it++)
delete it->second;
registers_.clear();
}
// The state of the call frame information interpreter as it processes
// instructions from a CIE and FDE.
class CallFrameInfo::State {
public:
// Create a call frame information interpreter state with the given
// reporter, reader, handler, and initial call frame info address.
State(ByteReader *reader, Handler *handler, Reporter *reporter,
uint64 address)
: reader_(reader), handler_(handler), reporter_(reporter),
address_(address), entry_(NULL), cursor_(NULL) { }
// Interpret instructions from CIE, save the resulting rule set for
// DW_CFA_restore instructions, and return true. On error, report
// the problem to reporter_ and return false.
bool InterpretCIE(const CIE &cie);
// Interpret instructions from FDE, and return true. On error,
// report the problem to reporter_ and return false.
bool InterpretFDE(const FDE &fde);
private:
// The operands of a CFI instruction, for ParseOperands.
struct Operands {
unsigned register_number; // A register number.
uint64 offset; // An offset or address.
long signed_offset; // A signed offset.
string expression; // A DWARF expression.
};
// Parse CFI instruction operands from STATE's instruction stream as
// described by FORMAT. On success, populate OPERANDS with the
// results, and return true. On failure, report the problem and
// return false.
//
// Each character of FORMAT should be one of the following:
//
// 'r' unsigned LEB128 register number (OPERANDS->register_number)
// 'o' unsigned LEB128 offset (OPERANDS->offset)
// 's' signed LEB128 offset (OPERANDS->signed_offset)
// 'a' machine-size address (OPERANDS->offset)
// (If the CIE has a 'z' augmentation string, 'a' uses the
// encoding specified by the 'R' argument.)
// '1' a one-byte offset (OPERANDS->offset)
// '2' a two-byte offset (OPERANDS->offset)
// '4' a four-byte offset (OPERANDS->offset)
// '8' an eight-byte offset (OPERANDS->offset)
// 'e' a DW_FORM_block holding a (OPERANDS->expression)
// DWARF expression
bool ParseOperands(const char *format, Operands *operands);
// Interpret one CFI instruction from STATE's instruction stream, update
// STATE, report any rule changes to handler_, and return true. On
// failure, report the problem and return false.
bool DoInstruction();
// The following Do* member functions are subroutines of DoInstruction,
// factoring out the actual work of operations that have several
// different encodings.
// Set the CFA rule to be the value of BASE_REGISTER plus OFFSET, and
// return true. On failure, report and return false. (Used for
// DW_CFA_def_cfa and DW_CFA_def_cfa_sf.)
bool DoDefCFA(unsigned base_register, long offset);
// Change the offset of the CFA rule to OFFSET, and return true. On
// failure, report and return false. (Subroutine for
// DW_CFA_def_cfa_offset and DW_CFA_def_cfa_offset_sf.)
bool DoDefCFAOffset(long offset);
// Specify that REG can be recovered using RULE, and return true. On
// failure, report and return false.
bool DoRule(unsigned reg, Rule *rule);
// Specify that REG can be found at OFFSET from the CFA, and return true.
// On failure, report and return false. (Subroutine for DW_CFA_offset,
// DW_CFA_offset_extended, and DW_CFA_offset_extended_sf.)
bool DoOffset(unsigned reg, long offset);
// Specify that the caller's value for REG is the CFA plus OFFSET,
// and return true. On failure, report and return false. (Subroutine
// for DW_CFA_val_offset and DW_CFA_val_offset_sf.)
bool DoValOffset(unsigned reg, long offset);
// Restore REG to the rule established in the CIE, and return true. On
// failure, report and return false. (Subroutine for DW_CFA_restore and
// DW_CFA_restore_extended.)
bool DoRestore(unsigned reg);
// Return the section offset of the instruction at cursor. For use
// in error messages.
uint64 CursorOffset() { return entry_->offset + (cursor_ - entry_->start); }
// Report that entry_ is incomplete, and return false. For brevity.
bool ReportIncomplete() {
reporter_->Incomplete(entry_->offset, entry_->kind);
return false;
}
// For reading multi-byte values with the appropriate endianness.
ByteReader *reader_;
// The handler to which we should report the data we find.
Handler *handler_;
// For reporting problems in the info we're parsing.
Reporter *reporter_;
// The code address to which the next instruction in the stream applies.
uint64 address_;
// The entry whose instructions we are currently processing. This is
// first a CIE, and then an FDE.
const Entry *entry_;
// The next instruction to process.
const uint8_t *cursor_;
// The current set of rules.
RuleMap rules_;
// The set of rules established by the CIE, used by DW_CFA_restore
// and DW_CFA_restore_extended. We set this after interpreting the
// CIE's instructions.
RuleMap cie_rules_;
// A stack of saved states, for DW_CFA_remember_state and
// DW_CFA_restore_state.
std::stack<RuleMap> saved_rules_;
};
bool CallFrameInfo::State::InterpretCIE(const CIE &cie) {
entry_ = &cie;
cursor_ = entry_->instructions;
while (cursor_ < entry_->end)
if (!DoInstruction())
return false;
// Note the rules established by the CIE, for use by DW_CFA_restore
// and DW_CFA_restore_extended.
cie_rules_ = rules_;
return true;
}
bool CallFrameInfo::State::InterpretFDE(const FDE &fde) {
entry_ = &fde;
cursor_ = entry_->instructions;
while (cursor_ < entry_->end)
if (!DoInstruction())
return false;
return true;
}
bool CallFrameInfo::State::ParseOperands(const char *format,
Operands *operands) {
size_t len;
const char *operand;
for (operand = format; *operand; operand++) {
size_t bytes_left = entry_->end - cursor_;
switch (*operand) {
case 'r':
operands->register_number = reader_->ReadUnsignedLEB128(cursor_, &len);
if (len > bytes_left) return ReportIncomplete();
cursor_ += len;
break;
case 'o':
operands->offset = reader_->ReadUnsignedLEB128(cursor_, &len);
if (len > bytes_left) return ReportIncomplete();
cursor_ += len;
break;
case 's':
operands->signed_offset = reader_->ReadSignedLEB128(cursor_, &len);
if (len > bytes_left) return ReportIncomplete();
cursor_ += len;
break;
case 'a':
operands->offset =
reader_->ReadEncodedPointer(cursor_, entry_->cie->pointer_encoding,
&len);
if (len > bytes_left) return ReportIncomplete();
cursor_ += len;
break;
case '1':
if (1 > bytes_left) return ReportIncomplete();
operands->offset = static_cast<unsigned char>(*cursor_++);
break;
case '2':
if (2 > bytes_left) return ReportIncomplete();
operands->offset = reader_->ReadTwoBytes(cursor_);
cursor_ += 2;
break;
case '4':
if (4 > bytes_left) return ReportIncomplete();
operands->offset = reader_->ReadFourBytes(cursor_);
cursor_ += 4;
break;
case '8':
if (8 > bytes_left) return ReportIncomplete();
operands->offset = reader_->ReadEightBytes(cursor_);
cursor_ += 8;
break;
case 'e': {
size_t expression_length = reader_->ReadUnsignedLEB128(cursor_, &len);
if (len > bytes_left || expression_length > bytes_left - len)
return ReportIncomplete();
cursor_ += len;
operands->expression = string(reinterpret_cast<const char *>(cursor_),
expression_length);
cursor_ += expression_length;
break;
}
default:
assert(0);
}
}
return true;
}
bool CallFrameInfo::State::DoInstruction() {
CIE *cie = entry_->cie;
Operands ops;
// Our entry's kind should have been set by now.
assert(entry_->kind != kUnknown);
// We shouldn't have been invoked unless there were more
// instructions to parse.
assert(cursor_ < entry_->end);
unsigned opcode = *cursor_++;
if ((opcode & 0xc0) != 0) {
switch (opcode & 0xc0) {
// Advance the address.
case DW_CFA_advance_loc: {
size_t code_offset = opcode & 0x3f;
address_ += code_offset * cie->code_alignment_factor;
break;
}
// Find a register at an offset from the CFA.
case DW_CFA_offset:
if (!ParseOperands("o", &ops) ||
!DoOffset(opcode & 0x3f, ops.offset * cie->data_alignment_factor))
return false;
break;
// Restore the rule established for a register by the CIE.
case DW_CFA_restore:
if (!DoRestore(opcode & 0x3f)) return false;
break;
// The 'if' above should have excluded this possibility.
default:
assert(0);
}
// Return here, so the big switch below won't be indented.
return true;
}
switch (opcode) {
// Set the address.
case DW_CFA_set_loc:
if (!ParseOperands("a", &ops)) return false;
address_ = ops.offset;
break;
// Advance the address.
case DW_CFA_advance_loc1:
if (!ParseOperands("1", &ops)) return false;
address_ += ops.offset * cie->code_alignment_factor;
break;
// Advance the address.
case DW_CFA_advance_loc2:
if (!ParseOperands("2", &ops)) return false;
address_ += ops.offset * cie->code_alignment_factor;
break;
// Advance the address.
case DW_CFA_advance_loc4:
if (!ParseOperands("4", &ops)) return false;
address_ += ops.offset * cie->code_alignment_factor;
break;
// Advance the address.
case DW_CFA_MIPS_advance_loc8:
if (!ParseOperands("8", &ops)) return false;
address_ += ops.offset * cie->code_alignment_factor;
break;
// Compute the CFA by adding an offset to a register.
case DW_CFA_def_cfa:
if (!ParseOperands("ro", &ops) ||
!DoDefCFA(ops.register_number, ops.offset))
return false;
break;
// Compute the CFA by adding an offset to a register.
case DW_CFA_def_cfa_sf:
if (!ParseOperands("rs", &ops) ||
!DoDefCFA(ops.register_number,
ops.signed_offset * cie->data_alignment_factor))
return false;
break;
// Change the base register used to compute the CFA.
case DW_CFA_def_cfa_register: {
if (!ParseOperands("r", &ops)) return false;
Rule *cfa_rule = rules_.CFARule();
if (!cfa_rule) {
if (!DoDefCFA(ops.register_number, ops.offset)) {
reporter_->NoCFARule(entry_->offset, entry_->kind, CursorOffset());
return false;
}
} else {
cfa_rule->SetBaseRegister(ops.register_number);
if (!cfa_rule->Handle(handler_, address_,
Handler::kCFARegister))
return false;
}
break;
}
// Change the offset used to compute the CFA.
case DW_CFA_def_cfa_offset:
if (!ParseOperands("o", &ops) ||
!DoDefCFAOffset(ops.offset))
return false;
break;
// Change the offset used to compute the CFA.
case DW_CFA_def_cfa_offset_sf:
if (!ParseOperands("s", &ops) ||
!DoDefCFAOffset(ops.signed_offset * cie->data_alignment_factor))
return false;
break;
// Specify an expression whose value is the CFA.
case DW_CFA_def_cfa_expression: {
if (!ParseOperands("e", &ops))
return false;
Rule *rule = new ValExpressionRule(ops.expression);
rules_.SetCFARule(rule);
if (!rule->Handle(handler_, address_,
Handler::kCFARegister))
return false;
break;
}
// The register's value cannot be recovered.
case DW_CFA_undefined: {
if (!ParseOperands("r", &ops) ||
!DoRule(ops.register_number, new UndefinedRule()))
return false;
break;
}
// The register's value is unchanged from its value in the caller.
case DW_CFA_same_value: {
if (!ParseOperands("r", &ops) ||
!DoRule(ops.register_number, new SameValueRule()))
return false;
break;
}
// Find a register at an offset from the CFA.
case DW_CFA_offset_extended:
if (!ParseOperands("ro", &ops) ||
!DoOffset(ops.register_number,
ops.offset * cie->data_alignment_factor))
return false;
break;
// The register is saved at an offset from the CFA.
case DW_CFA_offset_extended_sf:
if (!ParseOperands("rs", &ops) ||
!DoOffset(ops.register_number,
ops.signed_offset * cie->data_alignment_factor))
return false;
break;
// The register is saved at an offset from the CFA.
case DW_CFA_GNU_negative_offset_extended:
if (!ParseOperands("ro", &ops) ||
!DoOffset(ops.register_number,
-ops.offset * cie->data_alignment_factor))
return false;
break;
// The register's value is the sum of the CFA plus an offset.
case DW_CFA_val_offset:
if (!ParseOperands("ro", &ops) ||
!DoValOffset(ops.register_number,
ops.offset * cie->data_alignment_factor))
return false;
break;
// The register's value is the sum of the CFA plus an offset.
case DW_CFA_val_offset_sf:
if (!ParseOperands("rs", &ops) ||
!DoValOffset(ops.register_number,
ops.signed_offset * cie->data_alignment_factor))
return false;
break;
// The register has been saved in another register.
case DW_CFA_register: {
if (!ParseOperands("ro", &ops) ||
!DoRule(ops.register_number, new RegisterRule(ops.offset)))
return false;
break;
}
// An expression yields the address at which the register is saved.
case DW_CFA_expression: {
if (!ParseOperands("re", &ops) ||
!DoRule(ops.register_number, new ExpressionRule(ops.expression)))
return false;
break;
}
// An expression yields the caller's value for the register.
case DW_CFA_val_expression: {
if (!ParseOperands("re", &ops) ||
!DoRule(ops.register_number, new ValExpressionRule(ops.expression)))
return false;
break;
}
// Restore the rule established for a register by the CIE.
case DW_CFA_restore_extended:
if (!ParseOperands("r", &ops) ||
!DoRestore( ops.register_number))
return false;
break;
// Save the current set of rules on a stack.
case DW_CFA_remember_state:
saved_rules_.push(rules_);
break;
// Pop the current set of rules off the stack.
case DW_CFA_restore_state: {
if (saved_rules_.empty()) {
reporter_->EmptyStateStack(entry_->offset, entry_->kind,
CursorOffset());
return false;
}
const RuleMap &new_rules = saved_rules_.top();
if (rules_.CFARule() && !new_rules.CFARule()) {
reporter_->ClearingCFARule(entry_->offset, entry_->kind,
CursorOffset());
return false;
}
rules_.HandleTransitionTo(handler_, address_, new_rules);
rules_ = new_rules;
saved_rules_.pop();
break;
}
// No operation. (Padding instruction.)
case DW_CFA_nop:
break;
// A SPARC register window save: Registers 8 through 15 (%o0-%o7)
// are saved in registers 24 through 31 (%i0-%i7), and registers
// 16 through 31 (%l0-%l7 and %i0-%i7) are saved at CFA offsets
// (0-15 * the register size). The register numbers must be
// hard-coded. A GNU extension, and not a pretty one.
case DW_CFA_GNU_window_save: {
// Save %o0-%o7 in %i0-%i7.
for (int i = 8; i < 16; i++)
if (!DoRule(i, new RegisterRule(i + 16)))
return false;
// Save %l0-%l7 and %i0-%i7 at the CFA.
for (int i = 16; i < 32; i++)
// Assume that the byte reader's address size is the same as
// the architecture's register size. !@#%*^ hilarious.
if (!DoRule(i, new OffsetRule(Handler::kCFARegister,
(i - 16) * reader_->AddressSize())))
return false;
break;
}
// I'm not sure what this is. GDB doesn't use it for unwinding.
case DW_CFA_GNU_args_size:
if (!ParseOperands("o", &ops)) return false;
break;
// An opcode we don't recognize.
default: {
reporter_->BadInstruction(entry_->offset, entry_->kind, CursorOffset());
return false;
}
}
return true;
}
bool CallFrameInfo::State::DoDefCFA(unsigned base_register, long offset) {
Rule *rule = new ValOffsetRule(base_register, offset);
rules_.SetCFARule(rule);
return rule->Handle(handler_, address_,
Handler::kCFARegister);
}
bool CallFrameInfo::State::DoDefCFAOffset(long offset) {
Rule *cfa_rule = rules_.CFARule();
if (!cfa_rule) {
reporter_->NoCFARule(entry_->offset, entry_->kind, CursorOffset());
return false;
}
cfa_rule->SetOffset(offset);
return cfa_rule->Handle(handler_, address_,
Handler::kCFARegister);
}
bool CallFrameInfo::State::DoRule(unsigned reg, Rule *rule) {
rules_.SetRegisterRule(reg, rule);
return rule->Handle(handler_, address_, reg);
}
bool CallFrameInfo::State::DoOffset(unsigned reg, long offset) {
if (!rules_.CFARule()) {
reporter_->NoCFARule(entry_->offset, entry_->kind, CursorOffset());
return false;
}
return DoRule(reg,
new OffsetRule(Handler::kCFARegister, offset));
}
bool CallFrameInfo::State::DoValOffset(unsigned reg, long offset) {
if (!rules_.CFARule()) {
reporter_->NoCFARule(entry_->offset, entry_->kind, CursorOffset());
return false;
}
return DoRule(reg,
new ValOffsetRule(Handler::kCFARegister, offset));
}
bool CallFrameInfo::State::DoRestore(unsigned reg) {
// DW_CFA_restore and DW_CFA_restore_extended don't make sense in a CIE.
if (entry_->kind == kCIE) {
reporter_->RestoreInCIE(entry_->offset, CursorOffset());
return false;
}
Rule *rule = cie_rules_.RegisterRule(reg);
if (!rule) {
// This isn't really the right thing to do, but since CFI generally
// only mentions callee-saves registers, and GCC's convention for
// callee-saves registers is that they are unchanged, it's a good
// approximation.
rule = new SameValueRule();
}
return DoRule(reg, rule);
}
bool CallFrameInfo::ReadEntryPrologue(const uint8_t *cursor, Entry *entry) {
const uint8_t *buffer_end = buffer_ + buffer_length_;
// Initialize enough of ENTRY for use in error reporting.
entry->offset = cursor - buffer_;
entry->start = cursor;
entry->kind = kUnknown;
entry->end = NULL;
// Read the initial length. This sets reader_'s offset size.
size_t length_size;
uint64 length = reader_->ReadInitialLength(cursor, &length_size);
if (length_size > size_t(buffer_end - cursor))
return ReportIncomplete(entry);
cursor += length_size;
// In a .eh_frame section, a length of zero marks the end of the series
// of entries.
if (length == 0 && eh_frame_) {
entry->kind = kTerminator;
entry->end = cursor;
return true;
}
// Validate the length.
if (length > size_t(buffer_end - cursor))
return ReportIncomplete(entry);
// The length is the number of bytes after the initial length field;
// we have that position handy at this point, so compute the end
// now. (If we're parsing 64-bit-offset DWARF on a 32-bit machine,
// and the length didn't fit in a size_t, we would have rejected it
// above.)
entry->end = cursor + length;
// Parse the next field: either the offset of a CIE or a CIE id.
size_t offset_size = reader_->OffsetSize();
if (offset_size > size_t(entry->end - cursor)) return ReportIncomplete(entry);
entry->id = reader_->ReadOffset(cursor);
// Don't advance cursor past id field yet; in .eh_frame data we need
// the id's position to compute the section offset of an FDE's CIE.
// Now we can decide what kind of entry this is.
if (eh_frame_) {
// In .eh_frame data, an ID of zero marks the entry as a CIE, and
// anything else is an offset from the id field of the FDE to the start
// of the CIE.
if (entry->id == 0) {
entry->kind = kCIE;
} else {
entry->kind = kFDE;
// Turn the offset from the id into an offset from the buffer's start.
entry->id = (cursor - buffer_) - entry->id;
}
} else {
// In DWARF CFI data, an ID of ~0 (of the appropriate width, given the
// offset size for the entry) marks the entry as a CIE, and anything
// else is the offset of the CIE from the beginning of the section.
if (offset_size == 4)
entry->kind = (entry->id == 0xffffffff) ? kCIE : kFDE;
else {
assert(offset_size == 8);
entry->kind = (entry->id == 0xffffffffffffffffULL) ? kCIE : kFDE;
}
}
// Now advance cursor past the id.
cursor += offset_size;
// The fields specific to this kind of entry start here.
entry->fields = cursor;
entry->cie = NULL;
return true;
}
bool CallFrameInfo::ReadCIEFields(CIE *cie) {
const uint8_t *cursor = cie->fields;
size_t len;
assert(cie->kind == kCIE);
// Prepare for early exit.
cie->version = 0;
cie->augmentation.clear();
cie->code_alignment_factor = 0;
cie->data_alignment_factor = 0;
cie->return_address_register = 0;
cie->has_z_augmentation = false;
cie->pointer_encoding = DW_EH_PE_absptr;
cie->instructions = 0;
// Parse the version number.
if (cie->end - cursor < 1)
return ReportIncomplete(cie);
cie->version = reader_->ReadOneByte(cursor);
cursor++;
// If we don't recognize the version, we can't parse any more fields of the
// CIE. For DWARF CFI, we handle versions 1 through 4 (there was never a
// version 2 of CFI data). For .eh_frame, we handle versions 1 and 4 as well;
// the difference between those versions seems to be the same as for
// .debug_frame.
if (cie->version < 1 || cie->version > 4) {
reporter_->UnrecognizedVersion(cie->offset, cie->version);
return false;
}
const uint8_t *augmentation_start = cursor;
const uint8_t *augmentation_end =
reinterpret_cast<const uint8_t *>(memchr(augmentation_start, '\0',
cie->end - augmentation_start));
if (! augmentation_end) return ReportIncomplete(cie);
cursor = augmentation_end;
cie->augmentation = string(reinterpret_cast<const char *>(augmentation_start),
cursor - augmentation_start);
// Skip the terminating '\0'.
cursor++;
// Is this CFI augmented?
if (!cie->augmentation.empty()) {
// Is it an augmentation we recognize?
if (cie->augmentation[0] == DW_Z_augmentation_start) {
// Linux C++ ABI 'z' augmentation, used for exception handling data.
cie->has_z_augmentation = true;
} else {
// Not an augmentation we recognize. Augmentations can have arbitrary
// effects on the form of rest of the content, so we have to give up.
reporter_->UnrecognizedAugmentation(cie->offset, cie->augmentation);
return false;
}
}
if (cie->version >= 4) {
uint8_t address_size = *cursor++;
if (address_size != 8) {
// TODO(scottmg): Only supporting x64 for now.
reporter_->UnexpectedAddressSize(cie->offset, address_size);
return false;
}
uint8_t segment_size = *cursor++;
if (segment_size != 0) {
// TODO(scottmg): Only supporting x64 for now.
// I would have perhaps expected 4 here, but LLVM emits a 0, near
// http://llvm.org/docs/doxygen/html/MCDwarf_8cpp_source.html#l00606. As
// we are not using the value, only succeed for now if it's the expected
// 0.
reporter_->UnexpectedSegmentSize(cie->offset, segment_size);
return false;
}
}
// Parse the code alignment factor.
cie->code_alignment_factor = reader_->ReadUnsignedLEB128(cursor, &len);
if (size_t(cie->end - cursor) < len) return ReportIncomplete(cie);
cursor += len;
// Parse the data alignment factor.
cie->data_alignment_factor = reader_->ReadSignedLEB128(cursor, &len);
if (size_t(cie->end - cursor) < len) return ReportIncomplete(cie);
cursor += len;
// Parse the return address register. This is a ubyte in version 1, and
// a ULEB128 in version 3.
if (cie->version == 1) {
if (cursor >= cie->end) return ReportIncomplete(cie);
cie->return_address_register = uint8(*cursor++);
} else {
cie->return_address_register = reader_->ReadUnsignedLEB128(cursor, &len);
if (size_t(cie->end - cursor) < len) return ReportIncomplete(cie);
cursor += len;
}
// If we have a 'z' augmentation string, find the augmentation data and
// use the augmentation string to parse it.
if (cie->has_z_augmentation) {
uint64_t data_size = reader_->ReadUnsignedLEB128(cursor, &len);
if (size_t(cie->end - cursor) < len + data_size)
return ReportIncomplete(cie);
cursor += len;
const uint8_t *data = cursor;
cursor += data_size;
const uint8_t *data_end = cursor;
cie->has_z_lsda = false;
cie->has_z_personality = false;
cie->has_z_signal_frame = false;
// Walk the augmentation string, and extract values from the
// augmentation data as the string directs.
for (size_t i = 1; i < cie->augmentation.size(); i++) {
switch (cie->augmentation[i]) {
case DW_Z_has_LSDA:
// The CIE's augmentation data holds the language-specific data
// area pointer's encoding, and the FDE's augmentation data holds
// the pointer itself.
cie->has_z_lsda = true;
// Fetch the LSDA encoding from the augmentation data.
if (data >= data_end) return ReportIncomplete(cie);
cie->lsda_encoding = DwarfPointerEncoding(*data++);
if (!reader_->ValidEncoding(cie->lsda_encoding)) {
reporter_->InvalidPointerEncoding(cie->offset, cie->lsda_encoding);
return false;
}
// Don't check if the encoding is usable here --- we haven't
// read the FDE's fields yet, so we're not prepared for
// DW_EH_PE_funcrel, although that's a fine encoding for the
// LSDA to use, since it appears in the FDE.
break;
case DW_Z_has_personality_routine:
// The CIE's augmentation data holds the personality routine
// pointer's encoding, followed by the pointer itself.
cie->has_z_personality = true;
// Fetch the personality routine pointer's encoding from the
// augmentation data.
if (data >= data_end) return ReportIncomplete(cie);
cie->personality_encoding = DwarfPointerEncoding(*data++);
if (!reader_->ValidEncoding(cie->personality_encoding)) {
reporter_->InvalidPointerEncoding(cie->offset,
cie->personality_encoding);
return false;
}
if (!reader_->UsableEncoding(cie->personality_encoding)) {
reporter_->UnusablePointerEncoding(cie->offset,
cie->personality_encoding);
return false;
}
// Fetch the personality routine's pointer itself from the data.
cie->personality_address =
reader_->ReadEncodedPointer(data, cie->personality_encoding,
&len);
if (len > size_t(data_end - data))
return ReportIncomplete(cie);
data += len;
break;
case DW_Z_has_FDE_address_encoding:
// The CIE's augmentation data holds the pointer encoding to use
// for addresses in the FDE.
if (data >= data_end) return ReportIncomplete(cie);
cie->pointer_encoding = DwarfPointerEncoding(*data++);
if (!reader_->ValidEncoding(cie->pointer_encoding)) {
reporter_->InvalidPointerEncoding(cie->offset,
cie->pointer_encoding);
return false;
}
if (!reader_->UsableEncoding(cie->pointer_encoding)) {
reporter_->UnusablePointerEncoding(cie->offset,
cie->pointer_encoding);
return false;
}
break;
case DW_Z_is_signal_trampoline:
// Frames using this CIE are signal delivery frames.
cie->has_z_signal_frame = true;
break;
default:
// An augmentation we don't recognize.
reporter_->UnrecognizedAugmentation(cie->offset, cie->augmentation);
return false;
}
}
}
// The CIE's instructions start here.
cie->instructions = cursor;
return true;
}
bool CallFrameInfo::ReadFDEFields(FDE *fde) {
const uint8_t *cursor = fde->fields;
size_t size;
fde->address = reader_->ReadEncodedPointer(cursor, fde->cie->pointer_encoding,
&size);
if (size > size_t(fde->end - cursor))
return ReportIncomplete(fde);
cursor += size;
reader_->SetFunctionBase(fde->address);
// For the length, we strip off the upper nybble of the encoding used for
// the starting address.
DwarfPointerEncoding length_encoding =
DwarfPointerEncoding(fde->cie->pointer_encoding & 0x0f);
fde->size = reader_->ReadEncodedPointer(cursor, length_encoding, &size);
if (size > size_t(fde->end - cursor))
return ReportIncomplete(fde);
cursor += size;
// If the CIE has a 'z' augmentation string, then augmentation data
// appears here.
if (fde->cie->has_z_augmentation) {
uint64_t data_size = reader_->ReadUnsignedLEB128(cursor, &size);
if (size_t(fde->end - cursor) < size + data_size)
return ReportIncomplete(fde);
cursor += size;
// In the abstract, we should walk the augmentation string, and extract
// items from the FDE's augmentation data as we encounter augmentation
// string characters that specify their presence: the ordering of items
// in the augmentation string determines the arrangement of values in
// the augmentation data.
//
// In practice, there's only ever one value in FDE augmentation data
// that we support --- the LSDA pointer --- and we have to bail if we
// see any unrecognized augmentation string characters. So if there is
// anything here at all, we know what it is, and where it starts.
if (fde->cie->has_z_lsda) {
// Check whether the LSDA's pointer encoding is usable now: only once
// we've parsed the FDE's starting address do we call reader_->
// SetFunctionBase, so that the DW_EH_PE_funcrel encoding becomes
// usable.
if (!reader_->UsableEncoding(fde->cie->lsda_encoding)) {
reporter_->UnusablePointerEncoding(fde->cie->offset,
fde->cie->lsda_encoding);
return false;
}
fde->lsda_address =
reader_->ReadEncodedPointer(cursor, fde->cie->lsda_encoding, &size);
if (size > data_size)
return ReportIncomplete(fde);
// Ideally, we would also complain here if there were unconsumed
// augmentation data.
}
cursor += data_size;
}
// The FDE's instructions start after those.
fde->instructions = cursor;
return true;
}
bool CallFrameInfo::Start() {
const uint8_t *buffer_end = buffer_ + buffer_length_;
const uint8_t *cursor;
bool all_ok = true;
const uint8_t *entry_end;
bool ok;
// Traverse all the entries in buffer_, skipping CIEs and offering
// FDEs to the handler.
for (cursor = buffer_; cursor < buffer_end;
cursor = entry_end, all_ok = all_ok && ok) {
FDE fde;
// Make it easy to skip this entry with 'continue': assume that
// things are not okay until we've checked all the data, and
// prepare the address of the next entry.
ok = false;
// Read the entry's prologue.
if (!ReadEntryPrologue(cursor, &fde)) {
if (!fde.end) {
// If we couldn't even figure out this entry's extent, then we
// must stop processing entries altogether.
all_ok = false;
break;
}
entry_end = fde.end;
continue;
}
// The next iteration picks up after this entry.
entry_end = fde.end;
// Did we see an .eh_frame terminating mark?
if (fde.kind == kTerminator) {
// If there appears to be more data left in the section after the
// terminating mark, warn the user. But this is just a warning;
// we leave all_ok true.
if (fde.end < buffer_end) reporter_->EarlyEHTerminator(fde.offset);
break;
}
// In this loop, we skip CIEs. We only parse them fully when we
// parse an FDE that refers to them. This limits our memory
// consumption (beyond the buffer itself) to that needed to
// process the largest single entry.
if (fde.kind != kFDE) {
ok = true;
continue;
}
// Validate the CIE pointer.
if (fde.id > buffer_length_) {
reporter_->CIEPointerOutOfRange(fde.offset, fde.id);
continue;
}
CIE cie;
// Parse this FDE's CIE header.
if (!ReadEntryPrologue(buffer_ + fde.id, &cie))
continue;
// This had better be an actual CIE.
if (cie.kind != kCIE) {
reporter_->BadCIEId(fde.offset, fde.id);
continue;
}
if (!ReadCIEFields(&cie))
continue;
// We now have the values that govern both the CIE and the FDE.
cie.cie = &cie;
fde.cie = &cie;
// Parse the FDE's header.
if (!ReadFDEFields(&fde))
continue;
// Call Entry to ask the consumer if they're interested.
if (!handler_->Entry(fde.offset, fde.address, fde.size,
cie.version, cie.augmentation,
cie.return_address_register)) {
// The handler isn't interested in this entry. That's not an error.
ok = true;
continue;
}
if (cie.has_z_augmentation) {
// Report the personality routine address, if we have one.
if (cie.has_z_personality) {
if (!handler_
->PersonalityRoutine(cie.personality_address,
IsIndirectEncoding(cie.personality_encoding)))
continue;
}
// Report the language-specific data area address, if we have one.
if (cie.has_z_lsda) {
if (!handler_
->LanguageSpecificDataArea(fde.lsda_address,
IsIndirectEncoding(cie.lsda_encoding)))
continue;
}
// If this is a signal-handling frame, report that.
if (cie.has_z_signal_frame) {
if (!handler_->SignalHandler())
continue;
}
}
// Interpret the CIE's instructions, and then the FDE's instructions.
State state(reader_, handler_, reporter_, fde.address);
ok = state.InterpretCIE(cie) && state.InterpretFDE(fde);
// Tell the ByteReader that the function start address from the
// FDE header is no longer valid.
reader_->ClearFunctionBase();
// Report the end of the entry.
handler_->End();
}
return all_ok;
}
const char *CallFrameInfo::KindName(EntryKind kind) {
if (kind == CallFrameInfo::kUnknown)
return "entry";
else if (kind == CallFrameInfo::kCIE)
return "common information entry";
else if (kind == CallFrameInfo::kFDE)
return "frame description entry";
else {
assert (kind == CallFrameInfo::kTerminator);
return ".eh_frame sequence terminator";
}
}
bool CallFrameInfo::ReportIncomplete(Entry *entry) {
reporter_->Incomplete(entry->offset, entry->kind);
return false;
}
void CallFrameInfo::Reporter::Incomplete(uint64 offset,
CallFrameInfo::EntryKind kind) {
fprintf(stderr,
"%s: CFI %s at offset 0x%llx in '%s': entry ends early\n",
filename_.c_str(), CallFrameInfo::KindName(kind), offset,
section_.c_str());
}
void CallFrameInfo::Reporter::EarlyEHTerminator(uint64 offset) {
fprintf(stderr,
"%s: CFI at offset 0x%llx in '%s': saw end-of-data marker"
" before end of section contents\n",
filename_.c_str(), offset, section_.c_str());
}
void CallFrameInfo::Reporter::CIEPointerOutOfRange(uint64 offset,
uint64 cie_offset) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE pointer is out of range: 0x%llx\n",
filename_.c_str(), offset, section_.c_str(), cie_offset);
}
void CallFrameInfo::Reporter::BadCIEId(uint64 offset, uint64 cie_offset) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE pointer does not point to a CIE: 0x%llx\n",
filename_.c_str(), offset, section_.c_str(), cie_offset);
}
void CallFrameInfo::Reporter::UnexpectedAddressSize(uint64 offset,
uint8_t address_size) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE specifies unexpected address size: %d\n",
filename_.c_str(), offset, section_.c_str(), address_size);
}
void CallFrameInfo::Reporter::UnexpectedSegmentSize(uint64 offset,
uint8_t segment_size) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE specifies unexpected segment size: %d\n",
filename_.c_str(), offset, section_.c_str(), segment_size);
}
void CallFrameInfo::Reporter::UnrecognizedVersion(uint64 offset, int version) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE specifies unrecognized version: %d\n",
filename_.c_str(), offset, section_.c_str(), version);
}
void CallFrameInfo::Reporter::UnrecognizedAugmentation(uint64 offset,
const string &aug) {
fprintf(stderr,
"%s: CFI frame description entry at offset 0x%llx in '%s':"
" CIE specifies unrecognized augmentation: '%s'\n",
filename_.c_str(), offset, section_.c_str(), aug.c_str());
}
void CallFrameInfo::Reporter::InvalidPointerEncoding(uint64 offset,
uint8 encoding) {
fprintf(stderr,
"%s: CFI common information entry at offset 0x%llx in '%s':"
" 'z' augmentation specifies invalid pointer encoding: 0x%02x\n",
filename_.c_str(), offset, section_.c_str(), encoding);
}
void CallFrameInfo::Reporter::UnusablePointerEncoding(uint64 offset,
uint8 encoding) {
fprintf(stderr,
"%s: CFI common information entry at offset 0x%llx in '%s':"
" 'z' augmentation specifies a pointer encoding for which"
" we have no base address: 0x%02x\n",
filename_.c_str(), offset, section_.c_str(), encoding);
}
void CallFrameInfo::Reporter::RestoreInCIE(uint64 offset, uint64 insn_offset) {
fprintf(stderr,
"%s: CFI common information entry at offset 0x%llx in '%s':"
" the DW_CFA_restore instruction at offset 0x%llx"
" cannot be used in a common information entry\n",
filename_.c_str(), offset, section_.c_str(), insn_offset);
}
void CallFrameInfo::Reporter::BadInstruction(uint64 offset,
CallFrameInfo::EntryKind kind,
uint64 insn_offset) {
fprintf(stderr,
"%s: CFI %s at offset 0x%llx in section '%s':"
" the instruction at offset 0x%llx is unrecognized\n",
filename_.c_str(), CallFrameInfo::KindName(kind),
offset, section_.c_str(), insn_offset);
}
void CallFrameInfo::Reporter::NoCFARule(uint64 offset,
CallFrameInfo::EntryKind kind,
uint64 insn_offset) {
fprintf(stderr,
"%s: CFI %s at offset 0x%llx in section '%s':"
" the instruction at offset 0x%llx assumes that a CFA rule has"
" been set, but none has been set\n",
filename_.c_str(), CallFrameInfo::KindName(kind), offset,
section_.c_str(), insn_offset);
}
void CallFrameInfo::Reporter::EmptyStateStack(uint64 offset,
CallFrameInfo::EntryKind kind,
uint64 insn_offset) {
fprintf(stderr,
"%s: CFI %s at offset 0x%llx in section '%s':"
" the DW_CFA_restore_state instruction at offset 0x%llx"
" should pop a saved state from the stack, but the stack is empty\n",
filename_.c_str(), CallFrameInfo::KindName(kind), offset,
section_.c_str(), insn_offset);
}
void CallFrameInfo::Reporter::ClearingCFARule(uint64 offset,
CallFrameInfo::EntryKind kind,
uint64 insn_offset) {
fprintf(stderr,
"%s: CFI %s at offset 0x%llx in section '%s':"
" the DW_CFA_restore_state instruction at offset 0x%llx"
" would clear the CFA rule in effect\n",
filename_.c_str(), CallFrameInfo::KindName(kind), offset,
section_.c_str(), insn_offset);
}
} // namespace dwarf2reader
|