summaryrefslogtreecommitdiff
path: root/lwip/src/core/ipv6/nd6.c
blob: db0c132e48d56d2278e8228167923b7ad92d4770 (plain)
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
/**
 * @file
 *
 * Neighbor discovery and stateless address autoconfiguration for IPv6.
 * Aims to be compliant with RFC 4861 (Neighbor discovery) and RFC 4862
 * (Address autoconfiguration).
 */

/*
 * Copyright (c) 2010 Inico Technologies Ltd.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification,
 * are permitted provided that the following conditions are met:
 *
 * 1. Redistributions of source code must retain the above copyright notice,
 *    this list of conditions and the following disclaimer.
 * 2. 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.
 * 3. The name of the author may not be used to endorse or promote products
 *    derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
 *
 * This file is part of the lwIP TCP/IP stack.
 *
 * Author: Ivan Delamer <delamer@inicotech.com>
 *
 *
 * Please coordinate changes and requests with Ivan Delamer
 * <delamer@inicotech.com>
 */

#include "lwip/opt.h"

#if LWIP_IPV6  /* don't build if not configured for use in lwipopts.h */

#include "lwip/nd6.h"
#include "lwip/priv/nd6_priv.h"
#include "lwip/prot/nd6.h"
#include "lwip/prot/icmp6.h"
#include "lwip/pbuf.h"
#include "lwip/mem.h"
#include "lwip/memp.h"
#include "lwip/ip6.h"
#include "lwip/ip6_addr.h"
#include "lwip/inet_chksum.h"
#include "lwip/netif.h"
#include "lwip/icmp6.h"
#include "lwip/mld6.h"
#include "lwip/dhcp6.h"
#include "lwip/ip.h"
#include "lwip/stats.h"
#include "lwip/dns.h"

#include <string.h>

#ifdef LWIP_HOOK_FILENAME
#include LWIP_HOOK_FILENAME
#endif

#if LWIP_IPV6_DUP_DETECT_ATTEMPTS > IP6_ADDR_TENTATIVE_COUNT_MASK
#error LWIP_IPV6_DUP_DETECT_ATTEMPTS > IP6_ADDR_TENTATIVE_COUNT_MASK
#endif

/* Router tables. */
struct nd6_neighbor_cache_entry neighbor_cache[LWIP_ND6_NUM_NEIGHBORS];
struct nd6_destination_cache_entry destination_cache[LWIP_ND6_NUM_DESTINATIONS];
struct nd6_prefix_list_entry prefix_list[LWIP_ND6_NUM_PREFIXES];
struct nd6_router_list_entry default_router_list[LWIP_ND6_NUM_ROUTERS];

/* Default values, can be updated by a RA message. */
u32_t reachable_time = LWIP_ND6_REACHABLE_TIME;
u32_t retrans_timer = LWIP_ND6_RETRANS_TIMER; /* @todo implement this value in timer */

/* Index for cache entries. */
static u8_t nd6_cached_neighbor_index;
static netif_addr_idx_t nd6_cached_destination_index;

/* Multicast address holder. */
static ip6_addr_t multicast_address;

static u8_t nd6_tmr_rs_reduction;

/* Static buffer to parse RA packet options */
union ra_options {
  struct lladdr_option  lladdr;
  struct mtu_option     mtu;
  struct prefix_option  prefix;
#if LWIP_ND6_RDNSS_MAX_DNS_SERVERS
  struct rdnss_option   rdnss;
#endif
};
static union ra_options nd6_ra_buffer;

/* Forward declarations. */
static s8_t nd6_find_neighbor_cache_entry(const ip6_addr_t *ip6addr);
static s8_t nd6_new_neighbor_cache_entry(void);
static void nd6_free_neighbor_cache_entry(s8_t i);
static s16_t nd6_find_destination_cache_entry(const ip6_addr_t *ip6addr);
static s16_t nd6_new_destination_cache_entry(void);
static int nd6_is_prefix_in_netif(const ip6_addr_t *ip6addr, struct netif *netif);
static s8_t nd6_select_router(const ip6_addr_t *ip6addr, struct netif *netif);
static s8_t nd6_get_router(const ip6_addr_t *router_addr, struct netif *netif);
static s8_t nd6_new_router(const ip6_addr_t *router_addr, struct netif *netif);
static s8_t nd6_get_onlink_prefix(const ip6_addr_t *prefix, struct netif *netif);
static s8_t nd6_new_onlink_prefix(const ip6_addr_t *prefix, struct netif *netif);
static s8_t nd6_get_next_hop_entry(const ip6_addr_t *ip6addr, struct netif *netif);
static err_t nd6_queue_packet(s8_t neighbor_index, struct pbuf *q);

#define ND6_SEND_FLAG_MULTICAST_DEST 0x01
#define ND6_SEND_FLAG_ALLNODES_DEST 0x02
#define ND6_SEND_FLAG_ANY_SRC 0x04
static void nd6_send_ns(struct netif *netif, const ip6_addr_t *target_addr, u8_t flags);
static void nd6_send_na(struct netif *netif, const ip6_addr_t *target_addr, u8_t flags);
static void nd6_send_neighbor_cache_probe(struct nd6_neighbor_cache_entry *entry, u8_t flags);
#if LWIP_IPV6_SEND_ROUTER_SOLICIT
static err_t nd6_send_rs(struct netif *netif);
#endif /* LWIP_IPV6_SEND_ROUTER_SOLICIT */

#if LWIP_ND6_QUEUEING
static void nd6_free_q(struct nd6_q_entry *q);
#else /* LWIP_ND6_QUEUEING */
#define nd6_free_q(q) pbuf_free(q)
#endif /* LWIP_ND6_QUEUEING */
static void nd6_send_q(s8_t i);


/**
 * A local address has been determined to be a duplicate. Take the appropriate
 * action(s) on the address and the interface as a whole.
 *
 * @param netif the netif that owns the address
 * @param addr_idx the index of the address detected to be a duplicate
 */
static void
nd6_duplicate_addr_detected(struct netif *netif, s8_t addr_idx)
{

  /* Mark the address as duplicate, but leave its lifetimes alone. If this was
   * a manually assigned address, it will remain in existence as duplicate, and
   * as such be unusable for any practical purposes until manual intervention.
   * If this was an autogenerated address, the address will follow normal
   * expiration rules, and thus disappear once its valid lifetime expires. */
  netif_ip6_addr_set_state(netif, addr_idx, IP6_ADDR_DUPLICATED);

#if LWIP_IPV6_AUTOCONFIG
  /* If the affected address was the link-local address that we use to generate
   * all other addresses, then we should not continue to use those derived
   * addresses either, so mark them as duplicate as well. For autoconfig-only
   * setups, this will make the interface effectively unusable, approaching the
   * intention of RFC 4862 Sec. 5.4.5. @todo implement the full requirements */
  if (addr_idx == 0) {
    s8_t i;
    for (i = 1; i < LWIP_IPV6_NUM_ADDRESSES; i++) {
      if (!ip6_addr_isinvalid(netif_ip6_addr_state(netif, i)) &&
          !netif_ip6_addr_isstatic(netif, i)) {
        netif_ip6_addr_set_state(netif, i, IP6_ADDR_DUPLICATED);
      }
    }
  }
#endif /* LWIP_IPV6_AUTOCONFIG */
}

#if LWIP_IPV6_AUTOCONFIG
/**
 * We received a router advertisement that contains a prefix with the
 * autoconfiguration flag set. Add or update an associated autogenerated
 * address.
 *
 * @param netif the netif on which the router advertisement arrived
 * @param prefix_opt a pointer to the prefix option data
 * @param prefix_addr an aligned copy of the prefix address
 */
static void
nd6_process_autoconfig_prefix(struct netif *netif,
  struct prefix_option *prefix_opt, const ip6_addr_t *prefix_addr)
{
  ip6_addr_t ip6addr;
  u32_t valid_life, pref_life;
  u8_t addr_state;
  s8_t i, free_idx;

  /* The caller already checks RFC 4862 Sec. 5.5.3 points (a) and (b). We do
   * the rest, starting with checks for (c) and (d) here. */
  valid_life = lwip_htonl(prefix_opt->valid_lifetime);
  pref_life = lwip_htonl(prefix_opt->preferred_lifetime);
  if (pref_life > valid_life || prefix_opt->prefix_length != 64) {
    return; /* silently ignore this prefix for autoconfiguration purposes */
  }

  /* If an autogenerated address already exists for this prefix, update its
   * lifetimes. An address is considered autogenerated if 1) it is not static
   * (i.e., manually assigned), and 2) there is an advertised autoconfiguration
   * prefix for it (the one we are processing here). This does not necessarily
   * exclude the possibility that the address was actually assigned by, say,
   * DHCPv6. If that distinction becomes important in the future, more state
   * must be kept. As explained elsewhere we also update lifetimes of tentative
   * and duplicate addresses. Skip address slot 0 (the link-local address). */
  for (i = 1; i < LWIP_IPV6_NUM_ADDRESSES; i++) {
    addr_state = netif_ip6_addr_state(netif, i);
    if (!ip6_addr_isinvalid(addr_state) && !netif_ip6_addr_isstatic(netif, i) &&
        ip6_addr_netcmp(prefix_addr, netif_ip6_addr(netif, i))) {
      /* Update the valid lifetime, as per RFC 4862 Sec. 5.5.3 point (e).
       * The valid lifetime will never drop to zero as a result of this. */
      u32_t remaining_life = netif_ip6_addr_valid_life(netif, i);
      if (valid_life > ND6_2HRS || valid_life > remaining_life) {
        netif_ip6_addr_set_valid_life(netif, i, valid_life);
      } else if (remaining_life > ND6_2HRS) {
        netif_ip6_addr_set_valid_life(netif, i, ND6_2HRS);
      }
      LWIP_ASSERT("bad valid lifetime", !netif_ip6_addr_isstatic(netif, i));
      /* Update the preferred lifetime. No bounds checks are needed here. In
       * rare cases the advertisement may un-deprecate the address, though.
       * Deprecation is left to the timer code where it is handled anyway. */
      if (pref_life > 0 && addr_state == IP6_ADDR_DEPRECATED) {
        netif_ip6_addr_set_state(netif, i, IP6_ADDR_PREFERRED);
      }
      netif_ip6_addr_set_pref_life(netif, i, pref_life);
      return; /* there should be at most one matching address */
    }
  }

  /* No autogenerated address exists for this prefix yet. See if we can add a
   * new one. However, if IPv6 autoconfiguration is administratively disabled,
   * do not generate new addresses, but do keep updating lifetimes for existing
   * addresses. Also, when adding new addresses, we must protect explicitly
   * against a valid lifetime of zero, because again, we use that as a special
   * value. The generated address would otherwise expire immediately anyway.
   * Finally, the original link-local address must be usable at all. We start
   * creating addresses even if the link-local address is still in tentative
   * state though, and deal with the fallout of that upon DAD collision. */
  addr_state = netif_ip6_addr_state(netif, 0);
  if (!netif->ip6_autoconfig_enabled || valid_life == IP6_ADDR_LIFE_STATIC ||
      ip6_addr_isinvalid(addr_state) || ip6_addr_isduplicated(addr_state)) {
    return;
  }

  /* Construct the new address that we intend to use, and then see if that
   * address really does not exist. It might have been added manually, after
   * all. As a side effect, find a free slot. Note that we cannot use
   * netif_add_ip6_address() here, as it would return ERR_OK if the address
   * already did exist, resulting in that address being given lifetimes. */
  IP6_ADDR(&ip6addr, prefix_addr->addr[0], prefix_addr->addr[1],
    netif_ip6_addr(netif, 0)->addr[2], netif_ip6_addr(netif, 0)->addr[3]);
  ip6_addr_assign_zone(&ip6addr, IP6_UNICAST, netif);

  free_idx = 0;
  for (i = 1; i < LWIP_IPV6_NUM_ADDRESSES; i++) {
    if (!ip6_addr_isinvalid(netif_ip6_addr_state(netif, i))) {
      if (ip6_addr_cmp(&ip6addr, netif_ip6_addr(netif, i))) {
        return; /* formed address already exists */
      }
    } else if (free_idx == 0) {
      free_idx = i;
    }
  }
  if (free_idx == 0) {
    return; /* no address slots available, try again on next advertisement */
  }

  /* Assign the new address to the interface. */
  ip_addr_copy_from_ip6(netif->ip6_addr[free_idx], ip6addr);
  netif_ip6_addr_set_valid_life(netif, free_idx, valid_life);
  netif_ip6_addr_set_pref_life(netif, free_idx, pref_life);
  netif_ip6_addr_set_state(netif, free_idx, IP6_ADDR_TENTATIVE);
}
#endif /* LWIP_IPV6_AUTOCONFIG */

/**
 * Process an incoming neighbor discovery message
 *
 * @param p the nd packet, p->payload pointing to the icmpv6 header
 * @param inp the netif on which this packet was received
 */
void
nd6_input(struct pbuf *p, struct netif *inp)
{
  u8_t msg_type;
  s8_t i;
  s16_t dest_idx;

  ND6_STATS_INC(nd6.recv);

  msg_type = *((u8_t *)p->payload);
  switch (msg_type) {
  case ICMP6_TYPE_NA: /* Neighbor Advertisement. */
  {
    struct na_header *na_hdr;
    struct lladdr_option *lladdr_opt;
    ip6_addr_t target_address;

    /* Check that na header fits in packet. */
    if (p->len < (sizeof(struct na_header))) {
      /* @todo debug message */
      pbuf_free(p);
      ND6_STATS_INC(nd6.lenerr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    na_hdr = (struct na_header *)p->payload;

    /* Create an aligned, zoned copy of the target address. */
    ip6_addr_copy_from_packed(target_address, na_hdr->target_address);
    ip6_addr_assign_zone(&target_address, IP6_UNICAST, inp);

    /* Check a subset of the other RFC 4861 Sec. 7.1.2 requirements. */
    if (IP6H_HOPLIM(ip6_current_header()) != ND6_HOPLIM || na_hdr->code != 0 ||
        ip6_addr_ismulticast(&target_address)) {
      pbuf_free(p);
      ND6_STATS_INC(nd6.proterr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    /* @todo RFC MUST: if IP destination is multicast, Solicited flag is zero */
    /* @todo RFC MUST: all included options have a length greater than zero */

    /* Unsolicited NA?*/
    if (ip6_addr_ismulticast(ip6_current_dest_addr())) {
      /* This is an unsolicited NA.
       * link-layer changed?
       * part of DAD mechanism? */

#if LWIP_IPV6_DUP_DETECT_ATTEMPTS
      /* If the target address matches this netif, it is a DAD response. */
      for (i = 0; i < LWIP_IPV6_NUM_ADDRESSES; i++) {
        if (!ip6_addr_isinvalid(netif_ip6_addr_state(inp, i)) &&
            !ip6_addr_isduplicated(netif_ip6_addr_state(inp, i)) &&
            ip6_addr_cmp(&target_address, netif_ip6_addr(inp, i))) {
          /* We are using a duplicate address. */
          nd6_duplicate_addr_detected(inp, i);

          pbuf_free(p);
          return;
        }
      }
#endif /* LWIP_IPV6_DUP_DETECT_ATTEMPTS */

      /* Check that link-layer address option also fits in packet. */
      if (p->len < (sizeof(struct na_header) + 2)) {
        /* @todo debug message */
        pbuf_free(p);
        ND6_STATS_INC(nd6.lenerr);
        ND6_STATS_INC(nd6.drop);
        return;
      }

      lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct na_header));

      if (p->len < (sizeof(struct na_header) + (lladdr_opt->length << 3))) {
        /* @todo debug message */
        pbuf_free(p);
        ND6_STATS_INC(nd6.lenerr);
        ND6_STATS_INC(nd6.drop);
        return;
      }

      /* This is an unsolicited NA, most likely there was a LLADDR change. */
      i = nd6_find_neighbor_cache_entry(&target_address);
      if (i >= 0) {
        if (na_hdr->flags & ND6_FLAG_OVERRIDE) {
          MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);
        }
      }
    } else {
      /* This is a solicited NA.
       * neighbor address resolution response?
       * neighbor unreachability detection response? */

      /* Find the cache entry corresponding to this na. */
      i = nd6_find_neighbor_cache_entry(&target_address);
      if (i < 0) {
        /* We no longer care about this target address. drop it. */
        pbuf_free(p);
        return;
      }

      /* Update cache entry. */
      if ((na_hdr->flags & ND6_FLAG_OVERRIDE) ||
          (neighbor_cache[i].state == ND6_INCOMPLETE)) {
        /* Check that link-layer address option also fits in packet. */
        if (p->len < (sizeof(struct na_header) + 2)) {
          /* @todo debug message */
          pbuf_free(p);
          ND6_STATS_INC(nd6.lenerr);
          ND6_STATS_INC(nd6.drop);
          return;
        }

        lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct na_header));

        if (p->len < (sizeof(struct na_header) + (lladdr_opt->length << 3))) {
          /* @todo debug message */
          pbuf_free(p);
          ND6_STATS_INC(nd6.lenerr);
          ND6_STATS_INC(nd6.drop);
          return;
        }

        MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);
      }

      neighbor_cache[i].netif = inp;
      neighbor_cache[i].state = ND6_REACHABLE;
      neighbor_cache[i].counter.reachable_time = reachable_time;

      /* Send queued packets, if any. */
      if (neighbor_cache[i].q != NULL) {
        nd6_send_q(i);
      }
    }

    break; /* ICMP6_TYPE_NA */
  }
  case ICMP6_TYPE_NS: /* Neighbor solicitation. */
  {
    struct ns_header *ns_hdr;
    struct lladdr_option *lladdr_opt;
    ip6_addr_t target_address;
    u8_t accepted;

    /* Check that ns header fits in packet. */
    if (p->len < sizeof(struct ns_header)) {
      /* @todo debug message */
      pbuf_free(p);
      ND6_STATS_INC(nd6.lenerr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    ns_hdr = (struct ns_header *)p->payload;

    /* Create an aligned, zoned copy of the target address. */
    ip6_addr_copy_from_packed(target_address, ns_hdr->target_address);
    ip6_addr_assign_zone(&target_address, IP6_UNICAST, inp);

    /* Check a subset of the other RFC 4861 Sec. 7.1.1 requirements. */
    if (IP6H_HOPLIM(ip6_current_header()) != ND6_HOPLIM || ns_hdr->code != 0 ||
       ip6_addr_ismulticast(&target_address)) {
      pbuf_free(p);
      ND6_STATS_INC(nd6.proterr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    /* @todo RFC MUST: all included options have a length greater than zero */
    /* @todo RFC MUST: if IP source is 'any', destination is solicited-node multicast address */
    /* @todo RFC MUST: if IP source is 'any', there is no source LL address option */

    /* Check if there is a link-layer address provided. Only point to it if in this buffer. */
    if (p->len >= (sizeof(struct ns_header) + 2)) {
      lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct ns_header));
      if (p->len < (sizeof(struct ns_header) + (lladdr_opt->length << 3))) {
        lladdr_opt = NULL;
      }
    } else {
      lladdr_opt = NULL;
    }

    /* Check if the target address is configured on the receiving netif. */
    accepted = 0;
    for (i = 0; i < LWIP_IPV6_NUM_ADDRESSES; ++i) {
      if ((ip6_addr_isvalid(netif_ip6_addr_state(inp, i)) ||
           (ip6_addr_istentative(netif_ip6_addr_state(inp, i)) &&
            ip6_addr_isany(ip6_current_src_addr()))) &&
          ip6_addr_cmp(&target_address, netif_ip6_addr(inp, i))) {
        accepted = 1;
        break;
      }
    }

    /* NS not for us? */
    if (!accepted) {
      pbuf_free(p);
      return;
    }

    /* Check for ANY address in src (DAD algorithm). */
    if (ip6_addr_isany(ip6_current_src_addr())) {
      /* Sender is validating this address. */
      for (i = 0; i < LWIP_IPV6_NUM_ADDRESSES; ++i) {
        if (!ip6_addr_isinvalid(netif_ip6_addr_state(inp, i)) &&
            ip6_addr_cmp(&target_address, netif_ip6_addr(inp, i))) {
          /* Send a NA back so that the sender does not use this address. */
          nd6_send_na(inp, netif_ip6_addr(inp, i), ND6_FLAG_OVERRIDE | ND6_SEND_FLAG_ALLNODES_DEST);
          if (ip6_addr_istentative(netif_ip6_addr_state(inp, i))) {
            /* We shouldn't use this address either. */
            nd6_duplicate_addr_detected(inp, i);
          }
        }
      }
    } else {
      /* Sender is trying to resolve our address. */
      /* Verify that they included their own link-layer address. */
      if (lladdr_opt == NULL) {
        /* Not a valid message. */
        pbuf_free(p);
        ND6_STATS_INC(nd6.proterr);
        ND6_STATS_INC(nd6.drop);
        return;
      }

      i = nd6_find_neighbor_cache_entry(ip6_current_src_addr());
      if (i>= 0) {
        /* We already have a record for the solicitor. */
        if (neighbor_cache[i].state == ND6_INCOMPLETE) {
          neighbor_cache[i].netif = inp;
          MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);

          /* Delay probe in case we get confirmation of reachability from upper layer (TCP). */
          neighbor_cache[i].state = ND6_DELAY;
          neighbor_cache[i].counter.delay_time = LWIP_ND6_DELAY_FIRST_PROBE_TIME / ND6_TMR_INTERVAL;
        }
      } else {
        /* Add their IPv6 address and link-layer address to neighbor cache.
         * We will need it at least to send a unicast NA message, but most
         * likely we will also be communicating with this node soon. */
        i = nd6_new_neighbor_cache_entry();
        if (i < 0) {
          /* We couldn't assign a cache entry for this neighbor.
           * we won't be able to reply. drop it. */
          pbuf_free(p);
          ND6_STATS_INC(nd6.memerr);
          return;
        }
        neighbor_cache[i].netif = inp;
        MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);
        ip6_addr_set(&(neighbor_cache[i].next_hop_address), ip6_current_src_addr());

        /* Receiving a message does not prove reachability: only in one direction.
         * Delay probe in case we get confirmation of reachability from upper layer (TCP). */
        neighbor_cache[i].state = ND6_DELAY;
        neighbor_cache[i].counter.delay_time = LWIP_ND6_DELAY_FIRST_PROBE_TIME / ND6_TMR_INTERVAL;
      }

      /* Send back a NA for us. Allocate the reply pbuf. */
      nd6_send_na(inp, &target_address, ND6_FLAG_SOLICITED | ND6_FLAG_OVERRIDE);
    }

    break; /* ICMP6_TYPE_NS */
  }
  case ICMP6_TYPE_RA: /* Router Advertisement. */
  {
    struct ra_header *ra_hdr;
    u8_t *buffer; /* Used to copy options. */
    u16_t offset;
#if LWIP_ND6_RDNSS_MAX_DNS_SERVERS
    /* There can be multiple RDNSS options per RA */
    u8_t rdnss_server_idx = 0;
#endif /* LWIP_ND6_RDNSS_MAX_DNS_SERVERS */

    /* Check that RA header fits in packet. */
    if (p->len < sizeof(struct ra_header)) {
      /* @todo debug message */
      pbuf_free(p);
      ND6_STATS_INC(nd6.lenerr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    ra_hdr = (struct ra_header *)p->payload;

    /* Check a subset of the other RFC 4861 Sec. 6.1.2 requirements. */
    if (!ip6_addr_islinklocal(ip6_current_src_addr()) ||
        IP6H_HOPLIM(ip6_current_header()) != ND6_HOPLIM || ra_hdr->code != 0) {
      pbuf_free(p);
      ND6_STATS_INC(nd6.proterr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    /* @todo RFC MUST: all included options have a length greater than zero */

    /* If we are sending RS messages, stop. */
#if LWIP_IPV6_SEND_ROUTER_SOLICIT
    /* ensure at least one solicitation is sent (see RFC 4861, ch. 6.3.7) */
    if ((inp->rs_count < LWIP_ND6_MAX_MULTICAST_SOLICIT) ||
        (nd6_send_rs(inp) == ERR_OK)) {
      inp->rs_count = 0;
    } else {
      inp->rs_count = 1;
    }
#endif /* LWIP_IPV6_SEND_ROUTER_SOLICIT */

    /* Get the matching default router entry. */
    i = nd6_get_router(ip6_current_src_addr(), inp);
    if (i < 0) {
      /* Create a new router entry. */
      i = nd6_new_router(ip6_current_src_addr(), inp);
    }

    if (i < 0) {
      /* Could not create a new router entry. */
      pbuf_free(p);
      ND6_STATS_INC(nd6.memerr);
      return;
    }

    /* Re-set invalidation timer. */
    default_router_list[i].invalidation_timer = lwip_htons(ra_hdr->router_lifetime);

    /* Re-set default timer values. */
#if LWIP_ND6_ALLOW_RA_UPDATES
    if (ra_hdr->retrans_timer > 0) {
      retrans_timer = lwip_htonl(ra_hdr->retrans_timer);
    }
    if (ra_hdr->reachable_time > 0) {
      reachable_time = lwip_htonl(ra_hdr->reachable_time);
    }
#endif /* LWIP_ND6_ALLOW_RA_UPDATES */

    /* @todo set default hop limit... */
    /* ra_hdr->current_hop_limit;*/

    /* Update flags in local entry (incl. preference). */
    default_router_list[i].flags = ra_hdr->flags;

#if LWIP_IPV6_DHCP6
    /* Trigger DHCPv6 if enabled */
    dhcp6_nd6_ra_trigger(inp, ra_hdr->flags & ND6_RA_FLAG_MANAGED_ADDR_CONFIG,
      ra_hdr->flags & ND6_RA_FLAG_OTHER_CONFIG);
#endif

    /* Offset to options. */
    offset = sizeof(struct ra_header);

    /* Process each option. */
    while ((p->tot_len - offset) >= 2) {
      u8_t option_type;
      u16_t option_len;
      int option_len8 = pbuf_try_get_at(p, offset + 1);
      if (option_len8 <= 0) {
        /* read beyond end or zero length */
        goto lenerr_drop_free_return;
      }
      option_len = ((u8_t)option_len8) << 3;
      if (option_len > p->tot_len - offset) {
        /* short packet (option does not fit in) */
        goto lenerr_drop_free_return;
      }
      if (p->len == p->tot_len) {
        /* no need to copy from contiguous pbuf */
        buffer = &((u8_t*)p->payload)[offset];
      } else {
        /* check if this option fits into our buffer */
        if (option_len > sizeof(nd6_ra_buffer)) {
          option_type = pbuf_get_at(p, offset);
          /* invalid option length */
          if (option_type != ND6_OPTION_TYPE_RDNSS) {
            goto lenerr_drop_free_return;
          }
          /* we allow RDNSS option to be longer - we'll just drop some servers */
          option_len = sizeof(nd6_ra_buffer);
        }
        buffer = (u8_t*)&nd6_ra_buffer;
        option_len = pbuf_copy_partial(p, &nd6_ra_buffer, option_len, offset);
      }
      option_type = buffer[0];
      switch (option_type) {
      case ND6_OPTION_TYPE_SOURCE_LLADDR:
      {
        struct lladdr_option *lladdr_opt;
        if (option_len < sizeof(struct lladdr_option)) {
          goto lenerr_drop_free_return;
        }
        lladdr_opt = (struct lladdr_option *)buffer;
        if ((default_router_list[i].neighbor_entry != NULL) &&
            (default_router_list[i].neighbor_entry->state == ND6_INCOMPLETE)) {
          SMEMCPY(default_router_list[i].neighbor_entry->lladdr, lladdr_opt->addr, inp->hwaddr_len);
          default_router_list[i].neighbor_entry->state = ND6_REACHABLE;
          default_router_list[i].neighbor_entry->counter.reachable_time = reachable_time;
        }
        break;
      }
      case ND6_OPTION_TYPE_MTU:
      {
        struct mtu_option *mtu_opt;
        u32_t mtu32;
        if (option_len < sizeof(struct mtu_option)) {
          goto lenerr_drop_free_return;
        }
        mtu_opt = (struct mtu_option *)buffer;
        mtu32 = lwip_htonl(mtu_opt->mtu);
        if ((mtu32 >= 1280) && (mtu32 <= 0xffff)) {
#if LWIP_ND6_ALLOW_RA_UPDATES
          if (inp->mtu) {
            /* don't set the mtu for IPv6 higher than the netif driver supports */
            inp->mtu6 = LWIP_MIN(inp->mtu, (u16_t)mtu32);
          } else {
            inp->mtu6 = (u16_t)mtu32;
          }
#endif /* LWIP_ND6_ALLOW_RA_UPDATES */
        }
        break;
      }
      case ND6_OPTION_TYPE_PREFIX_INFO:
      {
        struct prefix_option *prefix_opt;
        ip6_addr_t prefix_addr;
        if (option_len < sizeof(struct prefix_option)) {
          goto lenerr_drop_free_return;
        }

        prefix_opt = (struct prefix_option *)buffer;

        /* Get a memory-aligned copy of the prefix. */
        ip6_addr_copy_from_packed(prefix_addr, prefix_opt->prefix);
        ip6_addr_assign_zone(&prefix_addr, IP6_UNICAST, inp);

        if (!ip6_addr_islinklocal(&prefix_addr)) {
          if ((prefix_opt->flags & ND6_PREFIX_FLAG_ON_LINK) &&
              (prefix_opt->prefix_length == 64)) {
            /* Add to on-link prefix list. */
            u32_t valid_life;
            s8_t prefix;

            valid_life = lwip_htonl(prefix_opt->valid_lifetime);

            /* find cache entry for this prefix. */
            prefix = nd6_get_onlink_prefix(&prefix_addr, inp);
            if (prefix < 0 && valid_life > 0) {
              /* Create a new cache entry. */
              prefix = nd6_new_onlink_prefix(&prefix_addr, inp);
            }
            if (prefix >= 0) {
              prefix_list[prefix].invalidation_timer = valid_life;
            }
          }
#if LWIP_IPV6_AUTOCONFIG
          if (prefix_opt->flags & ND6_PREFIX_FLAG_AUTONOMOUS) {
            /* Perform processing for autoconfiguration. */
            nd6_process_autoconfig_prefix(inp, prefix_opt, &prefix_addr);
          }
#endif /* LWIP_IPV6_AUTOCONFIG */
        }

        break;
      }
      case ND6_OPTION_TYPE_ROUTE_INFO:
        /* @todo implement preferred routes.
        struct route_option * route_opt;
        route_opt = (struct route_option *)buffer;*/

        break;
#if LWIP_ND6_RDNSS_MAX_DNS_SERVERS
      case ND6_OPTION_TYPE_RDNSS:
      {
        u8_t num, n;
        u16_t copy_offset = offset + SIZEOF_RDNSS_OPTION_BASE;
        struct rdnss_option * rdnss_opt;
        if (option_len < SIZEOF_RDNSS_OPTION_BASE) {
          goto lenerr_drop_free_return;
        }

        rdnss_opt = (struct rdnss_option *)buffer;
        num = (rdnss_opt->length - 1) / 2;
        for (n = 0; (rdnss_server_idx < DNS_MAX_SERVERS) && (n < num); n++) {
          ip_addr_t rdnss_address;

          /* Copy directly from pbuf to get an aligned, zoned copy of the prefix. */
          if (pbuf_copy_partial(p, &rdnss_address, sizeof(ip6_addr_p_t), copy_offset) == sizeof(ip6_addr_p_t)) {
            IP_SET_TYPE_VAL(rdnss_address, IPADDR_TYPE_V6);
            ip6_addr_assign_zone(ip_2_ip6(&rdnss_address), IP6_UNKNOWN, inp);

            if (htonl(rdnss_opt->lifetime) > 0) {
              /* TODO implement Lifetime > 0 */
              dns_setserver(rdnss_server_idx++, &rdnss_address);
            } else {
              /* TODO implement DNS removal in dns.c */
              u8_t s;
              for (s = 0; s < DNS_MAX_SERVERS; s++) {
                const ip_addr_t *addr = dns_getserver(s);
                if(ip_addr_cmp(addr, &rdnss_address)) {
                  dns_setserver(s, NULL);
                }
              }
            }
          }
        }
        break;
      }
#endif /* LWIP_ND6_RDNSS_MAX_DNS_SERVERS */
      default:
        /* Unrecognized option, abort. */
        ND6_STATS_INC(nd6.proterr);
        break;
      }
      /* option length is checked earlier to be non-zero to make sure loop ends */
      offset += 8 * (u8_t)option_len8;
    }

    break; /* ICMP6_TYPE_RA */
  }
  case ICMP6_TYPE_RD: /* Redirect */
  {
    struct redirect_header *redir_hdr;
    struct lladdr_option *lladdr_opt;
    ip6_addr_t destination_address, target_address;

    /* Check that Redir header fits in packet. */
    if (p->len < sizeof(struct redirect_header)) {
      /* @todo debug message */
      pbuf_free(p);
      ND6_STATS_INC(nd6.lenerr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    redir_hdr = (struct redirect_header *)p->payload;

    /* Create an aligned, zoned copy of the destination address. */
    ip6_addr_copy_from_packed(destination_address, redir_hdr->destination_address);
    ip6_addr_assign_zone(&destination_address, IP6_UNICAST, inp);

    /* Check a subset of the other RFC 4861 Sec. 8.1 requirements. */
    if (!ip6_addr_islinklocal(ip6_current_src_addr()) ||
        IP6H_HOPLIM(ip6_current_header()) != ND6_HOPLIM ||
        redir_hdr->code != 0 || ip6_addr_ismulticast(&destination_address)) {
      pbuf_free(p);
      ND6_STATS_INC(nd6.proterr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    /* @todo RFC MUST: IP source address equals first-hop router for destination_address */
    /* @todo RFC MUST: ICMP target address is either link-local address or same as destination_address */
    /* @todo RFC MUST: all included options have a length greater than zero */

    if (p->len >= (sizeof(struct redirect_header) + 2)) {
      lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct redirect_header));
      if (p->len < (sizeof(struct redirect_header) + (lladdr_opt->length << 3))) {
        lladdr_opt = NULL;
      }
    } else {
      lladdr_opt = NULL;
    }

    /* Find dest address in cache */
    dest_idx = nd6_find_destination_cache_entry(&destination_address);
    if (dest_idx < 0) {
      /* Destination not in cache, drop packet. */
      pbuf_free(p);
      return;
    }

    /* Create an aligned, zoned copy of the target address. */
    ip6_addr_copy_from_packed(target_address, redir_hdr->target_address);
    ip6_addr_assign_zone(&target_address, IP6_UNICAST, inp);

    /* Set the new target address. */
    ip6_addr_copy(destination_cache[dest_idx].next_hop_addr, target_address);

    /* If Link-layer address of other router is given, try to add to neighbor cache. */
    if (lladdr_opt != NULL) {
      if (lladdr_opt->type == ND6_OPTION_TYPE_TARGET_LLADDR) {
        i = nd6_find_neighbor_cache_entry(&target_address);
        if (i < 0) {
          i = nd6_new_neighbor_cache_entry();
          if (i >= 0) {
            neighbor_cache[i].netif = inp;
            MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);
            ip6_addr_copy(neighbor_cache[i].next_hop_address, target_address);

            /* Receiving a message does not prove reachability: only in one direction.
             * Delay probe in case we get confirmation of reachability from upper layer (TCP). */
            neighbor_cache[i].state = ND6_DELAY;
            neighbor_cache[i].counter.delay_time = LWIP_ND6_DELAY_FIRST_PROBE_TIME / ND6_TMR_INTERVAL;
          }
        }
        if (i >= 0) {
          if (neighbor_cache[i].state == ND6_INCOMPLETE) {
            MEMCPY(neighbor_cache[i].lladdr, lladdr_opt->addr, inp->hwaddr_len);
            /* Receiving a message does not prove reachability: only in one direction.
             * Delay probe in case we get confirmation of reachability from upper layer (TCP). */
            neighbor_cache[i].state = ND6_DELAY;
            neighbor_cache[i].counter.delay_time = LWIP_ND6_DELAY_FIRST_PROBE_TIME / ND6_TMR_INTERVAL;
          }
        }
      }
    }
    break; /* ICMP6_TYPE_RD */
  }
  case ICMP6_TYPE_PTB: /* Packet too big */
  {
    struct icmp6_hdr *icmp6hdr; /* Packet too big message */
    struct ip6_hdr *ip6hdr; /* IPv6 header of the packet which caused the error */
    u32_t pmtu;
    ip6_addr_t destination_address;

    /* Check that ICMPv6 header + IPv6 header fit in payload */
    if (p->len < (sizeof(struct icmp6_hdr) + IP6_HLEN)) {
      /* drop short packets */
      pbuf_free(p);
      ND6_STATS_INC(nd6.lenerr);
      ND6_STATS_INC(nd6.drop);
      return;
    }

    icmp6hdr = (struct icmp6_hdr *)p->payload;
    ip6hdr = (struct ip6_hdr *)((u8_t*)p->payload + sizeof(struct icmp6_hdr));

    /* Create an aligned, zoned copy of the destination address. */
    ip6_addr_copy_from_packed(destination_address, ip6hdr->dest);
    ip6_addr_assign_zone(&destination_address, IP6_UNKNOWN, inp);

    /* Look for entry in destination cache. */
    dest_idx = nd6_find_destination_cache_entry(&destination_address);
    if (dest_idx < 0) {
      /* Destination not in cache, drop packet. */
      pbuf_free(p);
      return;
    }

    /* Change the Path MTU. */
    pmtu = lwip_htonl(icmp6hdr->data);
    destination_cache[dest_idx].pmtu = (u16_t)LWIP_MIN(pmtu, 0xFFFF);

    break; /* ICMP6_TYPE_PTB */
  }

  default:
    ND6_STATS_INC(nd6.proterr);
    ND6_STATS_INC(nd6.drop);
    break; /* default */
  }

  pbuf_free(p);
  return;
lenerr_drop_free_return:
  ND6_STATS_INC(nd6.lenerr);
  ND6_STATS_INC(nd6.drop);
  pbuf_free(p);
}


/**
 * Periodic timer for Neighbor discovery functions:
 *
 * - Update neighbor reachability states
 * - Update destination cache entries age
 * - Update invalidation timers of default routers and on-link prefixes
 * - Update lifetimes of our addresses
 * - Perform duplicate address detection (DAD) for our addresses
 * - Send router solicitations
 */
void
nd6_tmr(void)
{
  s8_t i;
  struct netif *netif;

  /* Process neighbor entries. */
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    switch (neighbor_cache[i].state) {
    case ND6_INCOMPLETE:
      if ((neighbor_cache[i].counter.probes_sent >= LWIP_ND6_MAX_MULTICAST_SOLICIT) &&
          (!neighbor_cache[i].isrouter)) {
        /* Retries exceeded. */
        nd6_free_neighbor_cache_entry(i);
      } else {
        /* Send a NS for this entry. */
        neighbor_cache[i].counter.probes_sent++;
        nd6_send_neighbor_cache_probe(&neighbor_cache[i], ND6_SEND_FLAG_MULTICAST_DEST);
      }
      break;
    case ND6_REACHABLE:
      /* Send queued packets, if any are left. Should have been sent already. */
      if (neighbor_cache[i].q != NULL) {
        nd6_send_q(i);
      }
      if (neighbor_cache[i].counter.reachable_time <= ND6_TMR_INTERVAL) {
        /* Change to stale state. */
        neighbor_cache[i].state = ND6_STALE;
        neighbor_cache[i].counter.stale_time = 0;
      } else {
        neighbor_cache[i].counter.reachable_time -= ND6_TMR_INTERVAL;
      }
      break;
    case ND6_STALE:
      neighbor_cache[i].counter.stale_time++;
      break;
    case ND6_DELAY:
      if (neighbor_cache[i].counter.delay_time <= 1) {
        /* Change to PROBE state. */
        neighbor_cache[i].state = ND6_PROBE;
        neighbor_cache[i].counter.probes_sent = 0;
      } else {
        neighbor_cache[i].counter.delay_time--;
      }
      break;
    case ND6_PROBE:
      if ((neighbor_cache[i].counter.probes_sent >= LWIP_ND6_MAX_MULTICAST_SOLICIT) &&
          (!neighbor_cache[i].isrouter)) {
        /* Retries exceeded. */
        nd6_free_neighbor_cache_entry(i);
      } else {
        /* Send a NS for this entry. */
        neighbor_cache[i].counter.probes_sent++;
        nd6_send_neighbor_cache_probe(&neighbor_cache[i], 0);
      }
      break;
    case ND6_NO_ENTRY:
    default:
      /* Do nothing. */
      break;
    }
  }

  /* Process destination entries. */
  for (i = 0; i < LWIP_ND6_NUM_DESTINATIONS; i++) {
    destination_cache[i].age++;
  }

  /* Process router entries. */
  for (i = 0; i < LWIP_ND6_NUM_ROUTERS; i++) {
    if (default_router_list[i].neighbor_entry != NULL) {
      /* Active entry. */
      if (default_router_list[i].invalidation_timer <= ND6_TMR_INTERVAL / 1000) {
        /* No more than 1 second remaining. Clear this entry. Also clear any of
         * its destination cache entries, as per RFC 4861 Sec. 5.3 and 6.3.5. */
        s8_t j;
        for (j = 0; j < LWIP_ND6_NUM_DESTINATIONS; j++) {
          if (ip6_addr_cmp(&destination_cache[j].next_hop_addr,
               &default_router_list[i].neighbor_entry->next_hop_address)) {
             ip6_addr_set_any(&destination_cache[j].destination_addr);
          }
        }
        default_router_list[i].neighbor_entry->isrouter = 0;
        default_router_list[i].neighbor_entry = NULL;
        default_router_list[i].invalidation_timer = 0;
        default_router_list[i].flags = 0;
      } else {
        default_router_list[i].invalidation_timer -= ND6_TMR_INTERVAL / 1000;
      }
    }
  }

  /* Process prefix entries. */
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; i++) {
    if (prefix_list[i].netif != NULL) {
      if (prefix_list[i].invalidation_timer <= ND6_TMR_INTERVAL / 1000) {
        /* Entry timed out, remove it */
        prefix_list[i].invalidation_timer = 0;
        prefix_list[i].netif = NULL;
      } else {
        prefix_list[i].invalidation_timer -= ND6_TMR_INTERVAL / 1000;
      }
    }
  }

  /* Process our own addresses, updating address lifetimes and/or DAD state. */
  NETIF_FOREACH(netif) {
    for (i = 0; i < LWIP_IPV6_NUM_ADDRESSES; ++i) {
      u8_t addr_state;
#if LWIP_IPV6_ADDRESS_LIFETIMES
      /* Step 1: update address lifetimes (valid and preferred). */
      addr_state = netif_ip6_addr_state(netif, i);
      /* RFC 4862 is not entirely clear as to whether address lifetimes affect
       * tentative addresses, and is even less clear as to what should happen
       * with duplicate addresses. We choose to track and update lifetimes for
       * both those types, although for different reasons:
       * - for tentative addresses, the line of thought of Sec. 5.7 combined
       *   with the potentially long period that an address may be in tentative
       *   state (due to the interface being down) suggests that lifetimes
       *   should be independent of external factors which would include DAD;
       * - for duplicate addresses, retiring them early could result in a new
       *   but unwanted attempt at marking them as valid, while retiring them
       *   late/never could clog up address slots on the netif.
       * As a result, we may end up expiring addresses of either type here.
       */
      if (!ip6_addr_isinvalid(addr_state) &&
          !netif_ip6_addr_isstatic(netif, i)) {
        u32_t life = netif_ip6_addr_valid_life(netif, i);
        if (life <= ND6_TMR_INTERVAL / 1000) {
          /* The address has expired. */
          netif_ip6_addr_set_valid_life(netif, i, 0);
          netif_ip6_addr_set_pref_life(netif, i, 0);
          netif_ip6_addr_set_state(netif, i, IP6_ADDR_INVALID);
        } else {
          if (!ip6_addr_life_isinfinite(life)) {
            life -= ND6_TMR_INTERVAL / 1000;
            LWIP_ASSERT("bad valid lifetime", life != IP6_ADDR_LIFE_STATIC);
            netif_ip6_addr_set_valid_life(netif, i, life);
          }
          /* The address is still here. Update the preferred lifetime too. */
          life = netif_ip6_addr_pref_life(netif, i);
          if (life <= ND6_TMR_INTERVAL / 1000) {
            /* This case must also trigger if 'life' was already zero, so as to
             * deal correctly with advertised preferred-lifetime reductions. */
            netif_ip6_addr_set_pref_life(netif, i, 0);
            if (addr_state == IP6_ADDR_PREFERRED)
              netif_ip6_addr_set_state(netif, i, IP6_ADDR_DEPRECATED);
          } else if (!ip6_addr_life_isinfinite(life)) {
            life -= ND6_TMR_INTERVAL / 1000;
            netif_ip6_addr_set_pref_life(netif, i, life);
          }
        }
      }
      /* The address state may now have changed, so reobtain it next. */
#endif /* LWIP_IPV6_ADDRESS_LIFETIMES */
      /* Step 2: update DAD state. */
      addr_state = netif_ip6_addr_state(netif, i);
      if (ip6_addr_istentative(addr_state)) {
        if ((addr_state & IP6_ADDR_TENTATIVE_COUNT_MASK) >= LWIP_IPV6_DUP_DETECT_ATTEMPTS) {
          /* No NA received in response. Mark address as valid. For dynamic
           * addresses with an expired preferred lifetime, the state is set to
           * deprecated right away. That should almost never happen, though. */
          addr_state = IP6_ADDR_PREFERRED;
#if LWIP_IPV6_ADDRESS_LIFETIMES
          if (!netif_ip6_addr_isstatic(netif, i) &&
              netif_ip6_addr_pref_life(netif, i) == 0) {
            addr_state = IP6_ADDR_DEPRECATED;
          }
#endif /* LWIP_IPV6_ADDRESS_LIFETIMES */
          netif_ip6_addr_set_state(netif, i, addr_state);
        } else if (netif_is_up(netif) && netif_is_link_up(netif)) {
          /* tentative: set next state by increasing by one */
          netif_ip6_addr_set_state(netif, i, addr_state + 1);
          /* Send a NS for this address. Use the unspecified address as source
           * address in all cases (RFC 4862 Sec. 5.4.2), not in the least
           * because as it is, we only consider multicast replies for DAD. */
          nd6_send_ns(netif, netif_ip6_addr(netif, i),
            ND6_SEND_FLAG_MULTICAST_DEST | ND6_SEND_FLAG_ANY_SRC);
        }
      }
    }
  }

#if LWIP_IPV6_SEND_ROUTER_SOLICIT
  /* Send router solicitation messages, if necessary. */
  if (!nd6_tmr_rs_reduction) {
    nd6_tmr_rs_reduction = (ND6_RTR_SOLICITATION_INTERVAL / ND6_TMR_INTERVAL) - 1;
    NETIF_FOREACH(netif) {
      if ((netif->rs_count > 0) && netif_is_up(netif) &&
          netif_is_link_up(netif) &&
          !ip6_addr_isinvalid(netif_ip6_addr_state(netif, 0)) &&
          !ip6_addr_isduplicated(netif_ip6_addr_state(netif, 0))) {
        if (nd6_send_rs(netif) == ERR_OK) {
          netif->rs_count--;
        }
      }
    }
  } else {
    nd6_tmr_rs_reduction--;
  }
#endif /* LWIP_IPV6_SEND_ROUTER_SOLICIT */

}

/** Send a neighbor solicitation message for a specific neighbor cache entry
 *
 * @param entry the neightbor cache entry for wich to send the message
 * @param flags one of ND6_SEND_FLAG_*
 */
static void
nd6_send_neighbor_cache_probe(struct nd6_neighbor_cache_entry *entry, u8_t flags)
{
  nd6_send_ns(entry->netif, &entry->next_hop_address, flags);
}

/**
 * Send a neighbor solicitation message
 *
 * @param netif the netif on which to send the message
 * @param target_addr the IPv6 target address for the ND message
 * @param flags one of ND6_SEND_FLAG_*
 */
static void
nd6_send_ns(struct netif *netif, const ip6_addr_t *target_addr, u8_t flags)
{
  struct ns_header *ns_hdr;
  struct pbuf *p;
  const ip6_addr_t *src_addr;
  u16_t lladdr_opt_len;

  LWIP_ASSERT("target address is required", target_addr != NULL);

  if (!(flags & ND6_SEND_FLAG_ANY_SRC) &&
      ip6_addr_isvalid(netif_ip6_addr_state(netif,0))) {
    /* Use link-local address as source address. */
    src_addr = netif_ip6_addr(netif, 0);
    /* calculate option length (in 8-byte-blocks) */
    lladdr_opt_len = ((netif->hwaddr_len + 2) + 7) >> 3;
  } else {
    src_addr = IP6_ADDR_ANY6;
    /* Option "MUST NOT be included when the source IP address is the unspecified address." */
    lladdr_opt_len = 0;
  }

  /* Allocate a packet. */
  p = pbuf_alloc(PBUF_IP, sizeof(struct ns_header) + (lladdr_opt_len << 3), PBUF_RAM);
  if (p == NULL) {
    ND6_STATS_INC(nd6.memerr);
    return;
  }

  /* Set fields. */
  ns_hdr = (struct ns_header *)p->payload;

  ns_hdr->type = ICMP6_TYPE_NS;
  ns_hdr->code = 0;
  ns_hdr->chksum = 0;
  ns_hdr->reserved = 0;
  ip6_addr_copy_to_packed(ns_hdr->target_address, *target_addr);

  if (lladdr_opt_len != 0) {
    struct lladdr_option *lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct ns_header));
    lladdr_opt->type = ND6_OPTION_TYPE_SOURCE_LLADDR;
    lladdr_opt->length = (u8_t)lladdr_opt_len;
    SMEMCPY(lladdr_opt->addr, netif->hwaddr, netif->hwaddr_len);
  }

  /* Generate the solicited node address for the target address. */
  if (flags & ND6_SEND_FLAG_MULTICAST_DEST) {
    ip6_addr_set_solicitednode(&multicast_address, target_addr->addr[3]);
    ip6_addr_assign_zone(&multicast_address, IP6_MULTICAST, netif);
    target_addr = &multicast_address;
  }

#if CHECKSUM_GEN_ICMP6
  IF__NETIF_CHECKSUM_ENABLED(netif, NETIF_CHECKSUM_GEN_ICMP6) {
    ns_hdr->chksum = ip6_chksum_pseudo(p, IP6_NEXTH_ICMP6, p->len, src_addr,
      target_addr);
  }
#endif /* CHECKSUM_GEN_ICMP6 */

  /* Send the packet out. */
  ND6_STATS_INC(nd6.xmit);
  ip6_output_if(p, (src_addr == IP6_ADDR_ANY6) ? NULL : src_addr, target_addr,
      ND6_HOPLIM, 0, IP6_NEXTH_ICMP6, netif);
  pbuf_free(p);
}

/**
 * Send a neighbor advertisement message
 *
 * @param netif the netif on which to send the message
 * @param target_addr the IPv6 target address for the ND message
 * @param flags one of ND6_SEND_FLAG_*
 */
static void
nd6_send_na(struct netif *netif, const ip6_addr_t *target_addr, u8_t flags)
{
  struct na_header *na_hdr;
  struct lladdr_option *lladdr_opt;
  struct pbuf *p;
  const ip6_addr_t *src_addr;
  const ip6_addr_t *dest_addr;
  u16_t lladdr_opt_len;

  LWIP_ASSERT("target address is required", target_addr != NULL);

  /* Use link-local address as source address. */
  /* src_addr = netif_ip6_addr(netif, 0); */
  /* Use target address as source address. */
  src_addr = target_addr;

  /* Allocate a packet. */
  lladdr_opt_len = ((netif->hwaddr_len + 2) >> 3) + (((netif->hwaddr_len + 2) & 0x07) ? 1 : 0);
  p = pbuf_alloc(PBUF_IP, sizeof(struct na_header) + (lladdr_opt_len << 3), PBUF_RAM);
  if (p == NULL) {
    ND6_STATS_INC(nd6.memerr);
    return;
  }

  /* Set fields. */
  na_hdr = (struct na_header *)p->payload;
  lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct na_header));

  na_hdr->type = ICMP6_TYPE_NA;
  na_hdr->code = 0;
  na_hdr->chksum = 0;
  na_hdr->flags = flags & 0xf0;
  na_hdr->reserved[0] = 0;
  na_hdr->reserved[1] = 0;
  na_hdr->reserved[2] = 0;
  ip6_addr_copy_to_packed(na_hdr->target_address, *target_addr);

  lladdr_opt->type = ND6_OPTION_TYPE_TARGET_LLADDR;
  lladdr_opt->length = (u8_t)lladdr_opt_len;
  SMEMCPY(lladdr_opt->addr, netif->hwaddr, netif->hwaddr_len);

  /* Generate the solicited node address for the target address. */
  if (flags & ND6_SEND_FLAG_MULTICAST_DEST) {
    ip6_addr_set_solicitednode(&multicast_address, target_addr->addr[3]);
    ip6_addr_assign_zone(&multicast_address, IP6_MULTICAST, netif);
    dest_addr = &multicast_address;
  } else if (flags & ND6_SEND_FLAG_ALLNODES_DEST) {
    ip6_addr_set_allnodes_linklocal(&multicast_address);
    ip6_addr_assign_zone(&multicast_address, IP6_MULTICAST, netif);
    dest_addr = &multicast_address;
  } else {
    dest_addr = ip6_current_src_addr();
  }

#if CHECKSUM_GEN_ICMP6
  IF__NETIF_CHECKSUM_ENABLED(netif, NETIF_CHECKSUM_GEN_ICMP6) {
    na_hdr->chksum = ip6_chksum_pseudo(p, IP6_NEXTH_ICMP6, p->len, src_addr,
      dest_addr);
  }
#endif /* CHECKSUM_GEN_ICMP6 */

  /* Send the packet out. */
  ND6_STATS_INC(nd6.xmit);
  ip6_output_if(p, src_addr, dest_addr,
      ND6_HOPLIM, 0, IP6_NEXTH_ICMP6, netif);
  pbuf_free(p);
}

#if LWIP_IPV6_SEND_ROUTER_SOLICIT
/**
 * Send a router solicitation message
 *
 * @param netif the netif on which to send the message
 */
static err_t
nd6_send_rs(struct netif *netif)
{
  struct rs_header *rs_hdr;
  struct lladdr_option *lladdr_opt;
  struct pbuf *p;
  const ip6_addr_t *src_addr;
  err_t err;
  u16_t lladdr_opt_len = 0;

  /* Link-local source address, or unspecified address? */
  if (ip6_addr_isvalid(netif_ip6_addr_state(netif, 0))) {
    src_addr = netif_ip6_addr(netif, 0);
  } else {
    src_addr = IP6_ADDR_ANY6;
  }

  /* Generate the all routers target address. */
  ip6_addr_set_allrouters_linklocal(&multicast_address);
  ip6_addr_assign_zone(&multicast_address, IP6_MULTICAST, netif);

  /* Allocate a packet. */
  if (src_addr != IP6_ADDR_ANY6) {
    lladdr_opt_len = ((netif->hwaddr_len + 2) >> 3) + (((netif->hwaddr_len + 2) & 0x07) ? 1 : 0);
  }
  p = pbuf_alloc(PBUF_IP, sizeof(struct rs_header) + (lladdr_opt_len << 3), PBUF_RAM);
  if (p == NULL) {
    ND6_STATS_INC(nd6.memerr);
    return ERR_BUF;
  }

  /* Set fields. */
  rs_hdr = (struct rs_header *)p->payload;

  rs_hdr->type = ICMP6_TYPE_RS;
  rs_hdr->code = 0;
  rs_hdr->chksum = 0;
  rs_hdr->reserved = 0;

  if (src_addr != IP6_ADDR_ANY6) {
    /* Include our hw address. */
    lladdr_opt = (struct lladdr_option *)((u8_t*)p->payload + sizeof(struct rs_header));
    lladdr_opt->type = ND6_OPTION_TYPE_SOURCE_LLADDR;
    lladdr_opt->length = (u8_t)lladdr_opt_len;
    SMEMCPY(lladdr_opt->addr, netif->hwaddr, netif->hwaddr_len);
  }

#if CHECKSUM_GEN_ICMP6
  IF__NETIF_CHECKSUM_ENABLED(netif, NETIF_CHECKSUM_GEN_ICMP6) {
    rs_hdr->chksum = ip6_chksum_pseudo(p, IP6_NEXTH_ICMP6, p->len, src_addr,
      &multicast_address);
  }
#endif /* CHECKSUM_GEN_ICMP6 */

  /* Send the packet out. */
  ND6_STATS_INC(nd6.xmit);

  err = ip6_output_if(p, (src_addr == IP6_ADDR_ANY6) ? NULL : src_addr, &multicast_address,
      ND6_HOPLIM, 0, IP6_NEXTH_ICMP6, netif);
  pbuf_free(p);

  return err;
}
#endif /* LWIP_IPV6_SEND_ROUTER_SOLICIT */

/**
 * Search for a neighbor cache entry
 *
 * @param ip6addr the IPv6 address of the neighbor
 * @return The neighbor cache entry index that matched, -1 if no
 * entry is found
 */
static s8_t
nd6_find_neighbor_cache_entry(const ip6_addr_t *ip6addr)
{
  s8_t i;
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if (ip6_addr_cmp(ip6addr, &(neighbor_cache[i].next_hop_address))) {
      return i;
    }
  }
  return -1;
}

/**
 * Create a new neighbor cache entry.
 *
 * If no unused entry is found, will try to recycle an old entry
 * according to ad-hoc "age" heuristic.
 *
 * @return The neighbor cache entry index that was created, -1 if no
 * entry could be created
 */
static s8_t
nd6_new_neighbor_cache_entry(void)
{
  s8_t i;
  s8_t j;
  u32_t time;


  /* First, try to find an empty entry. */
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if (neighbor_cache[i].state == ND6_NO_ENTRY) {
      return i;
    }
  }

  /* We need to recycle an entry. in general, do not recycle if it is a router. */

  /* Next, try to find a Stale entry. */
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if ((neighbor_cache[i].state == ND6_STALE) &&
        (!neighbor_cache[i].isrouter)) {
      nd6_free_neighbor_cache_entry(i);
      return i;
    }
  }

  /* Next, try to find a Probe entry. */
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if ((neighbor_cache[i].state == ND6_PROBE) &&
        (!neighbor_cache[i].isrouter)) {
      nd6_free_neighbor_cache_entry(i);
      return i;
    }
  }

  /* Next, try to find a Delayed entry. */
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if ((neighbor_cache[i].state == ND6_DELAY) &&
        (!neighbor_cache[i].isrouter)) {
      nd6_free_neighbor_cache_entry(i);
      return i;
    }
  }

  /* Next, try to find the oldest reachable entry. */
  time = 0xfffffffful;
  j = -1;
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if ((neighbor_cache[i].state == ND6_REACHABLE) &&
        (!neighbor_cache[i].isrouter)) {
      if (neighbor_cache[i].counter.reachable_time < time) {
        j = i;
        time = neighbor_cache[i].counter.reachable_time;
      }
    }
  }
  if (j >= 0) {
    nd6_free_neighbor_cache_entry(j);
    return j;
  }

  /* Next, find oldest incomplete entry without queued packets. */
  time = 0;
  j = -1;
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if (
        (neighbor_cache[i].q == NULL) &&
        (neighbor_cache[i].state == ND6_INCOMPLETE) &&
        (!neighbor_cache[i].isrouter)) {
      if (neighbor_cache[i].counter.probes_sent >= time) {
        j = i;
        time = neighbor_cache[i].counter.probes_sent;
      }
    }
  }
  if (j >= 0) {
    nd6_free_neighbor_cache_entry(j);
    return j;
  }

  /* Next, find oldest incomplete entry with queued packets. */
  time = 0;
  j = -1;
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if ((neighbor_cache[i].state == ND6_INCOMPLETE) &&
        (!neighbor_cache[i].isrouter)) {
      if (neighbor_cache[i].counter.probes_sent >= time) {
        j = i;
        time = neighbor_cache[i].counter.probes_sent;
      }
    }
  }
  if (j >= 0) {
    nd6_free_neighbor_cache_entry(j);
    return j;
  }

  /* No more entries to try. */
  return -1;
}

/**
 * Will free any resources associated with a neighbor cache
 * entry, and will mark it as unused.
 *
 * @param i the neighbor cache entry index to free
 */
static void
nd6_free_neighbor_cache_entry(s8_t i)
{
  if ((i < 0) || (i >= LWIP_ND6_NUM_NEIGHBORS)) {
    return;
  }
  if (neighbor_cache[i].isrouter) {
    /* isrouter needs to be cleared before deleting a neighbor cache entry */
    return;
  }

  /* Free any queued packets. */
  if (neighbor_cache[i].q != NULL) {
    nd6_free_q(neighbor_cache[i].q);
    neighbor_cache[i].q = NULL;
  }

  neighbor_cache[i].state = ND6_NO_ENTRY;
  neighbor_cache[i].isrouter = 0;
  neighbor_cache[i].netif = NULL;
  neighbor_cache[i].counter.reachable_time = 0;
  ip6_addr_set_zero(&(neighbor_cache[i].next_hop_address));
}

/**
 * Search for a destination cache entry
 *
 * @param ip6addr the IPv6 address of the destination
 * @return The destination cache entry index that matched, -1 if no
 * entry is found
 */
static s16_t
nd6_find_destination_cache_entry(const ip6_addr_t *ip6addr)
{
  s16_t i;

  IP6_ADDR_ZONECHECK(ip6addr);

  for (i = 0; i < LWIP_ND6_NUM_DESTINATIONS; i++) {
    if (ip6_addr_cmp(ip6addr, &(destination_cache[i].destination_addr))) {
      return i;
    }
  }
  return -1;
}

/**
 * Create a new destination cache entry. If no unused entry is found,
 * will recycle oldest entry.
 *
 * @return The destination cache entry index that was created, -1 if no
 * entry was created
 */
static s16_t
nd6_new_destination_cache_entry(void)
{
  s16_t i, j;
  u32_t age;

  /* Find an empty entry. */
  for (i = 0; i < LWIP_ND6_NUM_DESTINATIONS; i++) {
    if (ip6_addr_isany(&(destination_cache[i].destination_addr))) {
      return i;
    }
  }

  /* Find oldest entry. */
  age = 0;
  j = LWIP_ND6_NUM_DESTINATIONS - 1;
  for (i = 0; i < LWIP_ND6_NUM_DESTINATIONS; i++) {
    if (destination_cache[i].age > age) {
      j = i;
    }
  }

  return j;
}

/**
 * Clear the destination cache.
 *
 * This operation may be necessary for consistency in the light of changing
 * local addresses and/or use of the gateway hook.
 */
void
nd6_clear_destination_cache(void)
{
  int i;

  for (i = 0; i < LWIP_ND6_NUM_DESTINATIONS; i++) {
    ip6_addr_set_any(&destination_cache[i].destination_addr);
  }
}

/**
 * Determine whether an address matches an on-link prefix or the subnet of a
 * statically assigned address.
 *
 * @param ip6addr the IPv6 address to match
 * @return 1 if the address is on-link, 0 otherwise
 */
static int
nd6_is_prefix_in_netif(const ip6_addr_t *ip6addr, struct netif *netif)
{
  s8_t i;

  /* Check to see if the address matches an on-link prefix. */
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; i++) {
    if ((prefix_list[i].netif == netif) &&
        (prefix_list[i].invalidation_timer > 0) &&
        ip6_addr_netcmp(ip6addr, &(prefix_list[i].prefix))) {
      return 1;
    }
  }
  /* Check to see if address prefix matches a manually configured (= static)
   * address. Static addresses have an implied /64 subnet assignment. Dynamic
   * addresses (from autoconfiguration) have no implied subnet assignment, and
   * are thus effectively /128 assignments. See RFC 5942 for more on this. */
  for (i = 0; i < LWIP_IPV6_NUM_ADDRESSES; i++) {
    if (ip6_addr_isvalid(netif_ip6_addr_state(netif, i)) &&
        netif_ip6_addr_isstatic(netif, i) &&
        ip6_addr_netcmp(ip6addr, netif_ip6_addr(netif, i))) {
      return 1;
    }
  }
  return 0;
}

/**
 * Select a default router for a destination.
 *
 * This function is used both for routing and for finding a next-hop target for
 * a packet. In the former case, the given netif is NULL, and the returned
 * router entry must be for a netif suitable for sending packets (up, link up).
 * In the latter case, the given netif is not NULL and restricts router choice.
 *
 * @param ip6addr the destination address
 * @param netif the netif for the outgoing packet, if known
 * @return the default router entry index, or -1 if no suitable
 *         router is found
 */
static s8_t
nd6_select_router(const ip6_addr_t *ip6addr, struct netif *netif)
{
  struct netif *router_netif;
  s8_t i, j, valid_router;
  static s8_t last_router;

  LWIP_UNUSED_ARG(ip6addr); /* @todo match preferred routes!! (must implement ND6_OPTION_TYPE_ROUTE_INFO) */

  /* @todo: implement default router preference */

  /* Look for valid routers. A reachable router is preferred. */
  valid_router = -1;
  for (i = 0; i < LWIP_ND6_NUM_ROUTERS; i++) {
    /* Is the router netif both set and apppropriate? */
    if (default_router_list[i].neighbor_entry != NULL) {
      router_netif = default_router_list[i].neighbor_entry->netif;
      if ((router_netif != NULL) && (netif != NULL ? netif == router_netif :
          (netif_is_up(router_netif) && netif_is_link_up(router_netif)))) {
        /* Is the router valid, i.e., reachable or probably reachable as per
         * RFC 4861 Sec. 6.3.6? Note that we will never return a router that
         * has no neighbor cache entry, due to the netif association tests. */
        if (default_router_list[i].neighbor_entry->state != ND6_INCOMPLETE) {
          /* Is the router known to be reachable? */
          if (default_router_list[i].neighbor_entry->state == ND6_REACHABLE) {
            return i; /* valid and reachable - done! */
          } else if (valid_router < 0) {
            valid_router = i; /* valid but not known to be reachable */
          }
        }
      }
    }
  }
  if (valid_router >= 0) {
    return valid_router;
  }

  /* Look for any router for which we have any information at all. */
  /* last_router is used for round-robin selection of incomplete routers, as
   * recommended in RFC 4861 Sec. 6.3.6 point (2). Advance only when picking a
   * route, to select the same router as next-hop target in the common case. */
  if ((netif == NULL) && (++last_router >= LWIP_ND6_NUM_ROUTERS)) {
    last_router = 0;
  }
  i = last_router;
  for (j = 0; j < LWIP_ND6_NUM_ROUTERS; j++) {
    if (default_router_list[i].neighbor_entry != NULL) {
      router_netif = default_router_list[i].neighbor_entry->netif;
      if ((router_netif != NULL) && (netif != NULL ? netif == router_netif :
          (netif_is_up(router_netif) && netif_is_link_up(router_netif)))) {
        return i;
      }
    }
    if (++i >= LWIP_ND6_NUM_ROUTERS) {
      i = 0;
    }
  }

  /* no suitable router found. */
  return -1;
}

/**
 * Find a router-announced route to the given destination. This route may be
 * based on an on-link prefix or a default router.
 *
 * If a suitable route is found, the returned netif is guaranteed to be in a
 * suitable state (up, link up) to be used for packet transmission.
 *
 * @param ip6addr the destination IPv6 address
 * @return the netif to use for the destination, or NULL if none found
 */
struct netif *
nd6_find_route(const ip6_addr_t *ip6addr)
{
  struct netif *netif;
  s8_t i;

  /* @todo decide if it makes sense to check the destination cache first */

  /* Check if there is a matching on-link prefix. There may be multiple
   * matches. Pick the first one that is associated with a suitable netif. */
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; ++i) {
    netif = prefix_list[i].netif;
    if ((netif != NULL) && ip6_addr_netcmp(&prefix_list[i].prefix, ip6addr) &&
        netif_is_up(netif) && netif_is_link_up(netif)) {
      return netif;
    }
  }

  /* No on-link prefix match. Find a router that can forward the packet. */
  i = nd6_select_router(ip6addr, NULL);
  if (i >= 0) {
    LWIP_ASSERT("selected router must have a neighbor entry",
      default_router_list[i].neighbor_entry != NULL);
    return default_router_list[i].neighbor_entry->netif;
  }

  return NULL;
}

/**
 * Find an entry for a default router.
 *
 * @param router_addr the IPv6 address of the router
 * @param netif the netif on which the router is found, if known
 * @return the index of the router entry, or -1 if not found
 */
static s8_t
nd6_get_router(const ip6_addr_t *router_addr, struct netif *netif)
{
  s8_t i;

  IP6_ADDR_ZONECHECK_NETIF(router_addr, netif);

  /* Look for router. */
  for (i = 0; i < LWIP_ND6_NUM_ROUTERS; i++) {
    if ((default_router_list[i].neighbor_entry != NULL) &&
        ((netif != NULL) ? netif == default_router_list[i].neighbor_entry->netif : 1) &&
        ip6_addr_cmp(router_addr, &(default_router_list[i].neighbor_entry->next_hop_address))) {
      return i;
    }
  }

  /* router not found. */
  return -1;
}

/**
 * Create a new entry for a default router.
 *
 * @param router_addr the IPv6 address of the router
 * @param netif the netif on which the router is connected, if known
 * @return the index on the router table, or -1 if could not be created
 */
static s8_t
nd6_new_router(const ip6_addr_t *router_addr, struct netif *netif)
{
  s8_t router_index;
  s8_t free_router_index;
  s8_t neighbor_index;

  IP6_ADDR_ZONECHECK_NETIF(router_addr, netif);

  /* Do we have a neighbor entry for this router? */
  neighbor_index = nd6_find_neighbor_cache_entry(router_addr);
  if (neighbor_index < 0) {
    /* Create a neighbor entry for this router. */
    neighbor_index = nd6_new_neighbor_cache_entry();
    if (neighbor_index < 0) {
      /* Could not create neighbor entry for this router. */
      return -1;
    }
    ip6_addr_set(&(neighbor_cache[neighbor_index].next_hop_address), router_addr);
    neighbor_cache[neighbor_index].netif = netif;
    neighbor_cache[neighbor_index].q = NULL;
    neighbor_cache[neighbor_index].state = ND6_INCOMPLETE;
    neighbor_cache[neighbor_index].counter.probes_sent = 1;
    nd6_send_neighbor_cache_probe(&neighbor_cache[neighbor_index], ND6_SEND_FLAG_MULTICAST_DEST);
  }

  /* Mark neighbor as router. */
  neighbor_cache[neighbor_index].isrouter = 1;

  /* Look for empty entry. */
  free_router_index = LWIP_ND6_NUM_ROUTERS;
  for (router_index = LWIP_ND6_NUM_ROUTERS - 1; router_index >= 0; router_index--) {
    /* check if router already exists (this is a special case for 2 netifs on the same subnet
       - e.g. wifi and cable) */
    if(default_router_list[router_index].neighbor_entry == &(neighbor_cache[neighbor_index])){ 
      return router_index; 
    } 
    if (default_router_list[router_index].neighbor_entry == NULL) {
      /* remember lowest free index to create a new entry */
      free_router_index = router_index;
    }
  }
  if (free_router_index < LWIP_ND6_NUM_ROUTERS) {
    default_router_list[free_router_index].neighbor_entry = &(neighbor_cache[neighbor_index]);
    return free_router_index;
  }

  /* Could not create a router entry. */

  /* Mark neighbor entry as not-router. Entry might be useful as neighbor still. */
  neighbor_cache[neighbor_index].isrouter = 0;

  /* router not found. */
  return -1;
}

/**
 * Find the cached entry for an on-link prefix.
 *
 * @param prefix the IPv6 prefix that is on-link
 * @param netif the netif on which the prefix is on-link
 * @return the index on the prefix table, or -1 if not found
 */
static s8_t
nd6_get_onlink_prefix(const ip6_addr_t *prefix, struct netif *netif)
{
  s8_t i;

  /* Look for prefix in list. */
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; ++i) {
    if ((ip6_addr_netcmp(&(prefix_list[i].prefix), prefix)) &&
        (prefix_list[i].netif == netif)) {
      return i;
    }
  }

  /* Entry not available. */
  return -1;
}

/**
 * Creates a new entry for an on-link prefix.
 *
 * @param prefix the IPv6 prefix that is on-link
 * @param netif the netif on which the prefix is on-link
 * @return the index on the prefix table, or -1 if not created
 */
static s8_t
nd6_new_onlink_prefix(const ip6_addr_t *prefix, struct netif *netif)
{
  s8_t i;

  /* Create new entry. */
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; ++i) {
    if ((prefix_list[i].netif == NULL) ||
        (prefix_list[i].invalidation_timer == 0)) {
      /* Found empty prefix entry. */
      prefix_list[i].netif = netif;
      ip6_addr_set(&(prefix_list[i].prefix), prefix);
      return i;
    }
  }

  /* Entry not available. */
  return -1;
}

/**
 * Determine the next hop for a destination. Will determine if the
 * destination is on-link, else a suitable on-link router is selected.
 *
 * The last entry index is cached for fast entry search.
 *
 * @param ip6addr the destination address
 * @param netif the netif on which the packet will be sent
 * @return the neighbor cache entry for the next hop, ERR_RTE if no
 *         suitable next hop was found, ERR_MEM if no cache entry
 *         could be created
 */
static s8_t
nd6_get_next_hop_entry(const ip6_addr_t *ip6addr, struct netif *netif)
{
#ifdef LWIP_HOOK_ND6_GET_GW
  const ip6_addr_t *next_hop_addr;
#endif /* LWIP_HOOK_ND6_GET_GW */
  s8_t i;
  s16_t dst_idx;

  IP6_ADDR_ZONECHECK_NETIF(ip6addr, netif);

#if LWIP_NETIF_HWADDRHINT
  if (netif->hints != NULL) {
    /* per-pcb cached entry was given */
    netif_addr_idx_t addr_hint = netif->hints->addr_hint;
    if (addr_hint < LWIP_ND6_NUM_DESTINATIONS) {
      nd6_cached_destination_index = addr_hint;
    }
  }
#endif /* LWIP_NETIF_HWADDRHINT */

  /* Look for ip6addr in destination cache. */
  if (ip6_addr_cmp(ip6addr, &(destination_cache[nd6_cached_destination_index].destination_addr))) {
    /* the cached entry index is the right one! */
    /* do nothing. */
    ND6_STATS_INC(nd6.cachehit);
  } else {
    /* Search destination cache. */
    dst_idx = nd6_find_destination_cache_entry(ip6addr);
    if (dst_idx >= 0) {
      /* found destination entry. make it our new cached index. */
      LWIP_ASSERT("type overflow", (size_t)dst_idx < NETIF_ADDR_IDX_MAX);
      nd6_cached_destination_index = (netif_addr_idx_t)dst_idx;
    } else {
      /* Not found. Create a new destination entry. */
      dst_idx = nd6_new_destination_cache_entry();
      if (dst_idx >= 0) {
        /* got new destination entry. make it our new cached index. */
        LWIP_ASSERT("type overflow", (size_t)dst_idx < NETIF_ADDR_IDX_MAX);
        nd6_cached_destination_index = (netif_addr_idx_t)dst_idx;
      } else {
        /* Could not create a destination cache entry. */
        return ERR_MEM;
      }

      /* Copy dest address to destination cache. */
      ip6_addr_set(&(destination_cache[nd6_cached_destination_index].destination_addr), ip6addr);

      /* Now find the next hop. is it a neighbor? */
      if (ip6_addr_islinklocal(ip6addr) ||
          nd6_is_prefix_in_netif(ip6addr, netif)) {
        /* Destination in local link. */
        destination_cache[nd6_cached_destination_index].pmtu = netif_mtu6(netif);
        ip6_addr_copy(destination_cache[nd6_cached_destination_index].next_hop_addr, destination_cache[nd6_cached_destination_index].destination_addr);
#ifdef LWIP_HOOK_ND6_GET_GW
      } else if ((next_hop_addr = LWIP_HOOK_ND6_GET_GW(netif, ip6addr)) != NULL) {
        /* Next hop for destination provided by hook function. */
        destination_cache[nd6_cached_destination_index].pmtu = netif->mtu;
        ip6_addr_set(&destination_cache[nd6_cached_destination_index].next_hop_addr, next_hop_addr);
#endif /* LWIP_HOOK_ND6_GET_GW */
      } else {
        /* We need to select a router. */
        i = nd6_select_router(ip6addr, netif);
        if (i < 0) {
          /* No router found. */
          ip6_addr_set_any(&(destination_cache[nd6_cached_destination_index].destination_addr));
          return ERR_RTE;
        }
        destination_cache[nd6_cached_destination_index].pmtu = netif_mtu6(netif); /* Start with netif mtu, correct through ICMPv6 if necessary */
        ip6_addr_copy(destination_cache[nd6_cached_destination_index].next_hop_addr, default_router_list[i].neighbor_entry->next_hop_address);
      }
    }
  }

#if LWIP_NETIF_HWADDRHINT
  if (netif->hints != NULL) {
    /* per-pcb cached entry was given */
    netif->hints->addr_hint = nd6_cached_destination_index;
  }
#endif /* LWIP_NETIF_HWADDRHINT */

  /* Look in neighbor cache for the next-hop address. */
  if (ip6_addr_cmp(&(destination_cache[nd6_cached_destination_index].next_hop_addr),
                   &(neighbor_cache[nd6_cached_neighbor_index].next_hop_address))) {
    /* Cache hit. */
    /* Do nothing. */
    ND6_STATS_INC(nd6.cachehit);
  } else {
    i = nd6_find_neighbor_cache_entry(&(destination_cache[nd6_cached_destination_index].next_hop_addr));
    if (i >= 0) {
      /* Found a matching record, make it new cached entry. */
      nd6_cached_neighbor_index = i;
    } else {
      /* Neighbor not in cache. Make a new entry. */
      i = nd6_new_neighbor_cache_entry();
      if (i >= 0) {
        /* got new neighbor entry. make it our new cached index. */
        nd6_cached_neighbor_index = i;
      } else {
        /* Could not create a neighbor cache entry. */
        return ERR_MEM;
      }

      /* Initialize fields. */
      ip6_addr_copy(neighbor_cache[i].next_hop_address,
                   destination_cache[nd6_cached_destination_index].next_hop_addr);
      neighbor_cache[i].isrouter = 0;
      neighbor_cache[i].netif = netif;
      neighbor_cache[i].state = ND6_INCOMPLETE;
      neighbor_cache[i].counter.probes_sent = 1;
      nd6_send_neighbor_cache_probe(&neighbor_cache[i], ND6_SEND_FLAG_MULTICAST_DEST);
    }
  }

  /* Reset this destination's age. */
  destination_cache[nd6_cached_destination_index].age = 0;

  return nd6_cached_neighbor_index;
}

/**
 * Queue a packet for a neighbor.
 *
 * @param neighbor_index the index in the neighbor cache table
 * @param q packet to be queued
 * @return ERR_OK if succeeded, ERR_MEM if out of memory
 */
static err_t
nd6_queue_packet(s8_t neighbor_index, struct pbuf *q)
{
  err_t result = ERR_MEM;
  struct pbuf *p;
  int copy_needed = 0;
#if LWIP_ND6_QUEUEING
  struct nd6_q_entry *new_entry, *r;
#endif /* LWIP_ND6_QUEUEING */

  if ((neighbor_index < 0) || (neighbor_index >= LWIP_ND6_NUM_NEIGHBORS)) {
    return ERR_ARG;
  }

  /* IF q includes a pbuf that must be copied, we have to copy the whole chain
   * into a new PBUF_RAM. See the definition of PBUF_NEEDS_COPY for details. */
  p = q;
  while (p) {
    if (PBUF_NEEDS_COPY(p)) {
      copy_needed = 1;
      break;
    }
    p = p->next;
  }
  if (copy_needed) {
    /* copy the whole packet into new pbufs */
    p = pbuf_clone(PBUF_LINK, PBUF_RAM, q);
    while ((p == NULL) && (neighbor_cache[neighbor_index].q != NULL)) {
      /* Free oldest packet (as per RFC recommendation) */
#if LWIP_ND6_QUEUEING
      r = neighbor_cache[neighbor_index].q;
      neighbor_cache[neighbor_index].q = r->next;
      r->next = NULL;
      nd6_free_q(r);
#else /* LWIP_ND6_QUEUEING */
      pbuf_free(neighbor_cache[neighbor_index].q);
      neighbor_cache[neighbor_index].q = NULL;
#endif /* LWIP_ND6_QUEUEING */
      p = pbuf_clone(PBUF_LINK, PBUF_RAM, q);
    }
  } else {
    /* referencing the old pbuf is enough */
    p = q;
    pbuf_ref(p);
  }
  /* packet was copied/ref'd? */
  if (p != NULL) {
    /* queue packet ... */
#if LWIP_ND6_QUEUEING
    /* allocate a new nd6 queue entry */
    new_entry = (struct nd6_q_entry *)memp_malloc(MEMP_ND6_QUEUE);
    if ((new_entry == NULL) && (neighbor_cache[neighbor_index].q != NULL)) {
      /* Free oldest packet (as per RFC recommendation) */
      r = neighbor_cache[neighbor_index].q;
      neighbor_cache[neighbor_index].q = r->next;
      r->next = NULL;
      nd6_free_q(r);
      new_entry = (struct nd6_q_entry *)memp_malloc(MEMP_ND6_QUEUE);
    }
    if (new_entry != NULL) {
      new_entry->next = NULL;
      new_entry->p = p;
      if (neighbor_cache[neighbor_index].q != NULL) {
        /* queue was already existent, append the new entry to the end */
        r = neighbor_cache[neighbor_index].q;
        while (r->next != NULL) {
          r = r->next;
        }
        r->next = new_entry;
      } else {
        /* queue did not exist, first item in queue */
        neighbor_cache[neighbor_index].q = new_entry;
      }
      LWIP_DEBUGF(LWIP_DBG_TRACE, ("ipv6: queued packet %p on neighbor entry %"S16_F"\n", (void *)p, (s16_t)neighbor_index));
      result = ERR_OK;
    } else {
      /* the pool MEMP_ND6_QUEUE is empty */
      pbuf_free(p);
      LWIP_DEBUGF(LWIP_DBG_TRACE, ("ipv6: could not queue a copy of packet %p (out of memory)\n", (void *)p));
      /* { result == ERR_MEM } through initialization */
    }
#else /* LWIP_ND6_QUEUEING */
    /* Queue a single packet. If an older packet is already queued, free it as per RFC. */
    if (neighbor_cache[neighbor_index].q != NULL) {
      pbuf_free(neighbor_cache[neighbor_index].q);
    }
    neighbor_cache[neighbor_index].q = p;
    LWIP_DEBUGF(LWIP_DBG_TRACE, ("ipv6: queued packet %p on neighbor entry %"S16_F"\n", (void *)p, (s16_t)neighbor_index));
    result = ERR_OK;
#endif /* LWIP_ND6_QUEUEING */
  } else {
    LWIP_DEBUGF(LWIP_DBG_TRACE, ("ipv6: could not queue a copy of packet %p (out of memory)\n", (void *)q));
    /* { result == ERR_MEM } through initialization */
  }

  return result;
}

#if LWIP_ND6_QUEUEING
/**
 * Free a complete queue of nd6 q entries
 *
 * @param q a queue of nd6_q_entry to free
 */
static void
nd6_free_q(struct nd6_q_entry *q)
{
  struct nd6_q_entry *r;
  LWIP_ASSERT("q != NULL", q != NULL);
  LWIP_ASSERT("q->p != NULL", q->p != NULL);
  while (q) {
    r = q;
    q = q->next;
    LWIP_ASSERT("r->p != NULL", (r->p != NULL));
    pbuf_free(r->p);
    memp_free(MEMP_ND6_QUEUE, r);
  }
}
#endif /* LWIP_ND6_QUEUEING */

/**
 * Send queued packets for a neighbor
 *
 * @param i the neighbor to send packets to
 */
static void
nd6_send_q(s8_t i)
{
  struct ip6_hdr *ip6hdr;
  ip6_addr_t dest;
#if LWIP_ND6_QUEUEING
  struct nd6_q_entry *q;
#endif /* LWIP_ND6_QUEUEING */

  if ((i < 0) || (i >= LWIP_ND6_NUM_NEIGHBORS)) {
    return;
  }

#if LWIP_ND6_QUEUEING
  while (neighbor_cache[i].q != NULL) {
    /* remember first in queue */
    q = neighbor_cache[i].q;
    /* pop first item off the queue */
    neighbor_cache[i].q = q->next;
    /* Get ipv6 header. */
    ip6hdr = (struct ip6_hdr *)(q->p->payload);
    /* Create an aligned copy. */
    ip6_addr_copy_from_packed(dest, ip6hdr->dest);
    /* Restore the zone, if applicable. */
    ip6_addr_assign_zone(&dest, IP6_UNKNOWN, neighbor_cache[i].netif);
    /* send the queued IPv6 packet */
    (neighbor_cache[i].netif)->output_ip6(neighbor_cache[i].netif, q->p, &dest);
    /* free the queued IP packet */
    pbuf_free(q->p);
    /* now queue entry can be freed */
    memp_free(MEMP_ND6_QUEUE, q);
  }
#else /* LWIP_ND6_QUEUEING */
  if (neighbor_cache[i].q != NULL) {
    /* Get ipv6 header. */
    ip6hdr = (struct ip6_hdr *)(neighbor_cache[i].q->payload);
    /* Create an aligned copy. */
    ip6_addr_copy_from_packed(dest, ip6hdr->dest);
    /* Restore the zone, if applicable. */
    ip6_addr_assign_zone(&dest, IP6_UNKNOWN, neighbor_cache[i].netif);
    /* send the queued IPv6 packet */
    (neighbor_cache[i].netif)->output_ip6(neighbor_cache[i].netif, neighbor_cache[i].q, &dest);
    /* free the queued IP packet */
    pbuf_free(neighbor_cache[i].q);
    neighbor_cache[i].q = NULL;
  }
#endif /* LWIP_ND6_QUEUEING */
}

/**
 * A packet is to be transmitted to a specific IPv6 destination on a specific
 * interface. Check if we can find the hardware address of the next hop to use
 * for the packet. If so, give the hardware address to the caller, which should
 * use it to send the packet right away. Otherwise, enqueue the packet for
 * later transmission while looking up the hardware address, if possible.
 *
 * As such, this function returns one of three different possible results:
 *
 * - ERR_OK with a non-NULL 'hwaddrp': the caller should send the packet now.
 * - ERR_OK with a NULL 'hwaddrp': the packet has been enqueued for later.
 * - not ERR_OK: something went wrong; forward the error upward in the stack.
 *
 * @param netif The lwIP network interface on which the IP packet will be sent.
 * @param q The pbuf(s) containing the IP packet to be sent.
 * @param ip6addr The destination IPv6 address of the packet.
 * @param hwaddrp On success, filled with a pointer to a HW address or NULL (meaning
 *        the packet has been queued).
 * @return
 * - ERR_OK on success, ERR_RTE if no route was found for the packet,
 * or ERR_MEM if low memory conditions prohibit sending the packet at all.
 */
err_t
nd6_get_next_hop_addr_or_queue(struct netif *netif, struct pbuf *q, const ip6_addr_t *ip6addr, const u8_t **hwaddrp)
{
  s8_t i;

  /* Get next hop record. */
  i = nd6_get_next_hop_entry(ip6addr, netif);
  if (i < 0) {
    /* failed to get a next hop neighbor record. */
    return i;
  }

  /* Now that we have a destination record, send or queue the packet. */
  if (neighbor_cache[i].state == ND6_STALE) {
    /* Switch to delay state. */
    neighbor_cache[i].state = ND6_DELAY;
    neighbor_cache[i].counter.delay_time = LWIP_ND6_DELAY_FIRST_PROBE_TIME / ND6_TMR_INTERVAL;
  }
  /* @todo should we send or queue if PROBE? send for now, to let unicast NS pass. */
  if ((neighbor_cache[i].state == ND6_REACHABLE) ||
      (neighbor_cache[i].state == ND6_DELAY) ||
      (neighbor_cache[i].state == ND6_PROBE)) {

    /* Tell the caller to send out the packet now. */
    *hwaddrp = neighbor_cache[i].lladdr;
    return ERR_OK;
  }

  /* We should queue packet on this interface. */
  *hwaddrp = NULL;
  return nd6_queue_packet(i, q);
}


/**
 * Get the Path MTU for a destination.
 *
 * @param ip6addr the destination address
 * @param netif the netif on which the packet will be sent
 * @return the Path MTU, if known, or the netif default MTU
 */
u16_t
nd6_get_destination_mtu(const ip6_addr_t *ip6addr, struct netif *netif)
{
  s16_t i;

  i = nd6_find_destination_cache_entry(ip6addr);
  if (i >= 0) {
    if (destination_cache[i].pmtu > 0) {
      return destination_cache[i].pmtu;
    }
  }

  if (netif != NULL) {
    return netif_mtu6(netif);
  }

  return 1280; /* Minimum MTU */
}


#if LWIP_ND6_TCP_REACHABILITY_HINTS
/**
 * Provide the Neighbor discovery process with a hint that a
 * destination is reachable. Called by tcp_receive when ACKs are
 * received or sent (as per RFC). This is useful to avoid sending
 * NS messages every 30 seconds.
 *
 * @param ip6addr the destination address which is know to be reachable
 *                by an upper layer protocol (TCP)
 */
void
nd6_reachability_hint(const ip6_addr_t *ip6addr)
{
  s8_t i;
  s16_t dst_idx;

  /* Find destination in cache. */
  if (ip6_addr_cmp(ip6addr, &(destination_cache[nd6_cached_destination_index].destination_addr))) {
    dst_idx = nd6_cached_destination_index;
    ND6_STATS_INC(nd6.cachehit);
  } else {
    dst_idx = nd6_find_destination_cache_entry(ip6addr);
  }
  if (dst_idx < 0) {
    return;
  }

  /* Find next hop neighbor in cache. */
  if (ip6_addr_cmp(&(destination_cache[dst_idx].next_hop_addr), &(neighbor_cache[nd6_cached_neighbor_index].next_hop_address))) {
    i = nd6_cached_neighbor_index;
    ND6_STATS_INC(nd6.cachehit);
  } else {
    i = nd6_find_neighbor_cache_entry(&(destination_cache[dst_idx].next_hop_addr));
  }
  if (i < 0) {
    return;
  }

  /* For safety: don't set as reachable if we don't have a LL address yet. Misuse protection. */
  if (neighbor_cache[i].state == ND6_INCOMPLETE || neighbor_cache[i].state == ND6_NO_ENTRY) {
    return;
  }

  /* Set reachability state. */
  neighbor_cache[i].state = ND6_REACHABLE;
  neighbor_cache[i].counter.reachable_time = reachable_time;
}
#endif /* LWIP_ND6_TCP_REACHABILITY_HINTS */

/**
 * Remove all prefix, neighbor_cache and router entries of the specified netif.
 *
 * @param netif points to a network interface
 */
void
nd6_cleanup_netif(struct netif *netif)
{
  u8_t i;
  s8_t router_index;
  for (i = 0; i < LWIP_ND6_NUM_PREFIXES; i++) {
    if (prefix_list[i].netif == netif) {
      prefix_list[i].netif = NULL;
    }
  }
  for (i = 0; i < LWIP_ND6_NUM_NEIGHBORS; i++) {
    if (neighbor_cache[i].netif == netif) {
      for (router_index = 0; router_index < LWIP_ND6_NUM_ROUTERS; router_index++) {
        if (default_router_list[router_index].neighbor_entry == &neighbor_cache[i]) {
          default_router_list[router_index].neighbor_entry = NULL;
          default_router_list[router_index].flags = 0;
        }
      }
      neighbor_cache[i].isrouter = 0;
      nd6_free_neighbor_cache_entry(i);
    }
  }
  /* Clear the destination cache, since many entries may now have become
   * invalid for one of several reasons. As destination cache entries have no
   * netif association, use a sledgehammer approach (this can be improved). */
  nd6_clear_destination_cache();
}

#if LWIP_IPV6_MLD
/**
 * The state of a local IPv6 address entry is about to change. If needed, join
 * or leave the solicited-node multicast group for the address.
 *
 * @param netif The netif that owns the address.
 * @param addr_idx The index of the address.
 * @param new_state The new (IP6_ADDR_) state for the address.
 */
void
nd6_adjust_mld_membership(struct netif *netif, s8_t addr_idx, u8_t new_state)
{
  u8_t old_state, old_member, new_member;

  old_state = netif_ip6_addr_state(netif, addr_idx);

  /* Determine whether we were, and should be, a member of the solicited-node
   * multicast group for this address. For tentative addresses, the group is
   * not joined until the address enters the TENTATIVE_1 (or VALID) state. */
  old_member = (old_state != IP6_ADDR_INVALID && old_state != IP6_ADDR_DUPLICATED && old_state != IP6_ADDR_TENTATIVE);
  new_member = (new_state != IP6_ADDR_INVALID && new_state != IP6_ADDR_DUPLICATED && new_state != IP6_ADDR_TENTATIVE);

  if (old_member != new_member) {
    ip6_addr_set_solicitednode(&multicast_address, netif_ip6_addr(netif, addr_idx)->addr[3]);
    ip6_addr_assign_zone(&multicast_address, IP6_MULTICAST, netif);

    if (new_member) {
      mld6_joingroup_netif(netif, &multicast_address);
    } else {
      mld6_leavegroup_netif(netif, &multicast_address);
    }
  }
}
#endif /* LWIP_IPV6_MLD */

/** Netif was added, set up, or reconnected (link up) */
void
nd6_restart_netif(struct netif *netif)
{
#if LWIP_IPV6_SEND_ROUTER_SOLICIT
  /* Send Router Solicitation messages (see RFC 4861, ch. 6.3.7). */
  netif->rs_count = LWIP_ND6_MAX_MULTICAST_SOLICIT;
#endif /* LWIP_IPV6_SEND_ROUTER_SOLICIT */
}

#endif /* LWIP_IPV6 */