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
|
/*
* MPTCP implementation - Sending side
*
* Initial Design & Implementation:
* Sébastien Barré <sebastien.barre@uclouvain.be>
*
* Current Maintainer & Author:
* Christoph Paasch <christoph.paasch@uclouvain.be>
*
* Additional authors:
* Jaakko Korkeaniemi <jaakko.korkeaniemi@aalto.fi>
* Gregory Detal <gregory.detal@uclouvain.be>
* Fabien Duchêne <fabien.duchene@uclouvain.be>
* Andreas Seelinger <Andreas.Seelinger@rwth-aachen.de>
* Lavkesh Lahngir <lavkesh51@gmail.com>
* Andreas Ripke <ripke@neclab.eu>
* Vlad Dogaru <vlad.dogaru@intel.com>
* Octavian Purdila <octavian.purdila@intel.com>
* John Ronan <jronan@tssg.org>
* Catalin Nicutar <catalin.nicutar@gmail.com>
* Brandon Heller <brandonh@stanford.edu>
*
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <asm/unaligned.h>
#include <net/mptcp.h>
#include <net/mptcp_v4.h>
#include <net/mptcp_v6.h>
#include <linux/kconfig.h>
/* is seq1 < seq2 ? */
static inline bool before64(const u64 seq1, const u64 seq2)
{
return (s64)(seq1 - seq2) < 0;
}
/* is seq1 > seq2 ? */
#define after64(seq1, seq2) before64(seq2, seq1)
static inline void mptcp_become_fully_estab(struct sock *sk)
{
tcp_sk(sk)->mptcp->fully_established = 1;
if (is_master_tp(tcp_sk(sk)) &&
tcp_sk(sk)->mpcb->pm_ops->fully_established)
tcp_sk(sk)->mpcb->pm_ops->fully_established(mptcp_meta_sk(sk));
}
/* Similar to tcp_tso_acked without any memory accounting */
static inline int mptcp_tso_acked_reinject(struct sock *meta_sk, struct sk_buff *skb)
{
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
u32 packets_acked, len;
BUG_ON(!after(TCP_SKB_CB(skb)->end_seq, meta_tp->snd_una));
packets_acked = tcp_skb_pcount(skb);
if (skb_unclone(skb, GFP_ATOMIC))
return 0;
len = meta_tp->snd_una - TCP_SKB_CB(skb)->seq;
__pskb_trim_head(skb, len);
TCP_SKB_CB(skb)->seq += len;
skb->ip_summed = CHECKSUM_PARTIAL;
skb->truesize -= len;
/* Any change of skb->len requires recalculation of tso factor. */
if (tcp_skb_pcount(skb) > 1)
tcp_set_skb_tso_segs(meta_sk, skb, tcp_skb_mss(skb));
packets_acked -= tcp_skb_pcount(skb);
if (packets_acked) {
BUG_ON(tcp_skb_pcount(skb) == 0);
BUG_ON(!before(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq));
}
return packets_acked;
}
/**
* Cleans the meta-socket retransmission queue and the reinject-queue.
* @sk must be the metasocket.
*/
static void mptcp_clean_rtx_queue(struct sock *meta_sk, u32 prior_snd_una)
{
struct sk_buff *skb, *tmp;
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
struct mptcp_cb *mpcb = meta_tp->mpcb;
bool acked = false;
u32 acked_pcount;
while ((skb = tcp_write_queue_head(meta_sk)) &&
skb != tcp_send_head(meta_sk)) {
bool fully_acked = true;
if (before(meta_tp->snd_una, TCP_SKB_CB(skb)->end_seq)) {
if (tcp_skb_pcount(skb) == 1 ||
!after(meta_tp->snd_una, TCP_SKB_CB(skb)->seq))
break;
acked_pcount = tcp_tso_acked(meta_sk, skb);
if (!acked_pcount)
break;
fully_acked = false;
} else {
acked_pcount = tcp_skb_pcount(skb);
}
acked = true;
meta_tp->packets_out -= acked_pcount;
meta_tp->retrans_stamp = 0;
if (!fully_acked)
break;
tcp_unlink_write_queue(skb, meta_sk);
if (mptcp_is_data_fin(skb)) {
struct sock *sk_it;
/* DATA_FIN has been acknowledged - now we can close
* the subflows
*/
mptcp_for_each_sk(mpcb, sk_it) {
unsigned long delay = 0;
/* If we are the passive closer, don't trigger
* subflow-fin until the subflow has been finned
* by the peer - thus we add a delay.
*/
if (mpcb->passive_close &&
sk_it->sk_state == TCP_ESTABLISHED)
delay = inet_csk(sk_it)->icsk_rto << 3;
mptcp_sub_close(sk_it, delay);
}
}
sk_wmem_free_skb(meta_sk, skb);
}
/* Remove acknowledged data from the reinject queue */
skb_queue_walk_safe(&mpcb->reinject_queue, skb, tmp) {
if (before(meta_tp->snd_una, TCP_SKB_CB(skb)->end_seq)) {
if (tcp_skb_pcount(skb) == 1 ||
!after(meta_tp->snd_una, TCP_SKB_CB(skb)->seq))
break;
mptcp_tso_acked_reinject(meta_sk, skb);
break;
}
__skb_unlink(skb, &mpcb->reinject_queue);
__kfree_skb(skb);
}
if (likely(between(meta_tp->snd_up, prior_snd_una, meta_tp->snd_una)))
meta_tp->snd_up = meta_tp->snd_una;
if (acked) {
tcp_rearm_rto(meta_sk);
/* Normally this is done in tcp_try_undo_loss - but MPTCP
* does not call this function.
*/
inet_csk(meta_sk)->icsk_retransmits = 0;
}
}
/* Inspired by tcp_rcv_state_process */
static int mptcp_rcv_state_process(struct sock *meta_sk, struct sock *sk,
const struct sk_buff *skb, u32 data_seq,
u16 data_len)
{
struct tcp_sock *meta_tp = tcp_sk(meta_sk), *tp = tcp_sk(sk);
struct tcphdr *th = tcp_hdr(skb);
/* State-machine handling if FIN has been enqueued and he has
* been acked (snd_una == write_seq) - it's important that this
* here is after sk_wmem_free_skb because otherwise
* sk_forward_alloc is wrong upon inet_csk_destroy_sock()
*/
switch (meta_sk->sk_state) {
case TCP_FIN_WAIT1: {
struct dst_entry *dst;
int tmo;
if (meta_tp->snd_una != meta_tp->write_seq)
break;
tcp_set_state(meta_sk, TCP_FIN_WAIT2);
meta_sk->sk_shutdown |= SEND_SHUTDOWN;
dst = __sk_dst_get(sk);
if (dst)
dst_confirm(dst);
if (!sock_flag(meta_sk, SOCK_DEAD)) {
/* Wake up lingering close() */
meta_sk->sk_state_change(meta_sk);
break;
}
if (meta_tp->linger2 < 0 ||
(data_len &&
after(data_seq + data_len - (mptcp_is_data_fin2(skb, tp) ? 1 : 0),
meta_tp->rcv_nxt))) {
mptcp_send_active_reset(meta_sk, GFP_ATOMIC);
tcp_done(meta_sk);
NET_INC_STATS_BH(sock_net(meta_sk), LINUX_MIB_TCPABORTONDATA);
return 1;
}
tmo = tcp_fin_time(meta_sk);
if (tmo > TCP_TIMEWAIT_LEN) {
inet_csk_reset_keepalive_timer(meta_sk, tmo - TCP_TIMEWAIT_LEN);
} else if (mptcp_is_data_fin2(skb, tp) || sock_owned_by_user(meta_sk)) {
/* Bad case. We could lose such FIN otherwise.
* It is not a big problem, but it looks confusing
* and not so rare event. We still can lose it now,
* if it spins in bh_lock_sock(), but it is really
* marginal case.
*/
inet_csk_reset_keepalive_timer(meta_sk, tmo);
} else {
meta_tp->time_wait(meta_sk, TCP_FIN_WAIT2, tmo);
}
break;
}
case TCP_CLOSING:
case TCP_LAST_ACK:
if (meta_tp->snd_una == meta_tp->write_seq) {
tcp_done(meta_sk);
return 1;
}
break;
}
/* step 7: process the segment text */
switch (meta_sk->sk_state) {
case TCP_FIN_WAIT1:
case TCP_FIN_WAIT2:
/* RFC 793 says to queue data in these states,
* RFC 1122 says we MUST send a reset.
* BSD 4.4 also does reset.
*/
if (meta_sk->sk_shutdown & RCV_SHUTDOWN) {
if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq &&
after(TCP_SKB_CB(skb)->end_seq - th->fin, tp->rcv_nxt) &&
!mptcp_is_data_fin2(skb, tp)) {
NET_INC_STATS_BH(sock_net(meta_sk), LINUX_MIB_TCPABORTONDATA);
mptcp_send_active_reset(meta_sk, GFP_ATOMIC);
tcp_reset(meta_sk);
return 1;
}
}
break;
}
return 0;
}
/**
* @return:
* i) 1: Everything's fine.
* ii) -1: A reset has been sent on the subflow - csum-failure
* iii) 0: csum-failure but no reset sent, because it's the last subflow.
* Last packet should not be destroyed by the caller because it has
* been done here.
*/
static int mptcp_verif_dss_csum(struct sock *sk)
{
struct tcp_sock *tp = tcp_sk(sk);
struct sk_buff *tmp, *tmp1, *last = NULL;
__wsum csum_tcp = 0; /* cumulative checksum of pld + mptcp-header */
int ans = 1, overflowed = 0, offset = 0, dss_csum_added = 0;
int iter = 0;
skb_queue_walk_safe(&sk->sk_receive_queue, tmp, tmp1) {
unsigned int csum_len;
if (before(tp->mptcp->map_subseq + tp->mptcp->map_data_len, TCP_SKB_CB(tmp)->end_seq))
/* Mapping ends in the middle of the packet -
* csum only these bytes
*/
csum_len = tp->mptcp->map_subseq + tp->mptcp->map_data_len - TCP_SKB_CB(tmp)->seq;
else
csum_len = tmp->len;
offset = 0;
if (overflowed) {
char first_word[4];
first_word[0] = 0;
first_word[1] = 0;
first_word[2] = 0;
first_word[3] = *(tmp->data);
csum_tcp = csum_partial(first_word, 4, csum_tcp);
offset = 1;
csum_len--;
overflowed = 0;
}
csum_tcp = skb_checksum(tmp, offset, csum_len, csum_tcp);
/* Was it on an odd-length? Then we have to merge the next byte
* correctly (see above)
*/
if (csum_len != (csum_len & (~1)))
overflowed = 1;
if (mptcp_is_data_seq(tmp) && !dss_csum_added) {
__be32 data_seq = htonl((u32)(tp->mptcp->map_data_seq >> 32));
/* If a 64-bit dss is present, we increase the offset
* by 4 bytes, as the high-order 64-bits will be added
* in the final csum_partial-call.
*/
u32 offset = skb_transport_offset(tmp) +
TCP_SKB_CB(tmp)->dss_off;
if (TCP_SKB_CB(tmp)->mptcp_flags & MPTCPHDR_SEQ64_SET)
offset += 4;
csum_tcp = skb_checksum(tmp, offset,
MPTCP_SUB_LEN_SEQ_CSUM,
csum_tcp);
csum_tcp = csum_partial(&data_seq,
sizeof(data_seq), csum_tcp);
dss_csum_added = 1; /* Just do it once */
}
last = tmp;
iter++;
if (!skb_queue_is_last(&sk->sk_receive_queue, tmp) &&
!before(TCP_SKB_CB(tmp1)->seq,
tp->mptcp->map_subseq + tp->mptcp->map_data_len))
break;
}
/* Now, checksum must be 0 */
if (unlikely(csum_fold(csum_tcp))) {
pr_err("%s csum is wrong: %#x data_seq %u dss_csum_added %d overflowed %d iterations %d\n",
__func__, csum_fold(csum_tcp), TCP_SKB_CB(last)->seq,
dss_csum_added, overflowed, iter);
tp->mptcp->send_mp_fail = 1;
/* map_data_seq is the data-seq number of the
* mapping we are currently checking
*/
tp->mpcb->csum_cutoff_seq = tp->mptcp->map_data_seq;
if (tp->mpcb->cnt_subflows > 1) {
mptcp_send_reset(sk);
ans = -1;
} else {
tp->mpcb->send_infinite_mapping = 1;
/* Need to purge the rcv-queue as it's no more valid */
while ((tmp = __skb_dequeue(&sk->sk_receive_queue)) != NULL) {
tp->copied_seq = TCP_SKB_CB(tmp)->end_seq;
kfree_skb(tmp);
}
ans = 0;
}
}
return ans;
}
static inline void mptcp_prepare_skb(struct sk_buff *skb, struct sk_buff *next,
struct sock *sk)
{
struct tcp_sock *tp = tcp_sk(sk);
struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
u32 inc = 0;
/* If skb is the end of this mapping (end is always at mapping-boundary
* thanks to the splitting/trimming), then we need to increase
* data-end-seq by 1 if this here is a data-fin.
*
* We need to do -1 because end_seq includes the subflow-FIN.
*/
if (tp->mptcp->map_data_fin &&
(tcb->end_seq - (tcp_hdr(skb)->fin ? 1 : 0)) ==
(tp->mptcp->map_subseq + tp->mptcp->map_data_len)) {
inc = 1;
/* We manually set the fin-flag if it is a data-fin. For easy
* processing in tcp_recvmsg.
*/
tcp_hdr(skb)->fin = 1;
} else {
/* We may have a subflow-fin with data but without data-fin */
tcp_hdr(skb)->fin = 0;
}
/* Adapt data-seq's to the packet itself. We kinda transform the
* dss-mapping to a per-packet granularity. This is necessary to
* correctly handle overlapping mappings coming from different
* subflows. Otherwise it would be a complete mess.
*/
tcb->seq = ((u32)tp->mptcp->map_data_seq) + tcb->seq - tp->mptcp->map_subseq;
tcb->end_seq = tcb->seq + skb->len + inc;
}
/**
* @return: 1 if the segment has been eaten and can be suppressed,
* otherwise 0.
*/
static inline int mptcp_direct_copy(struct sk_buff *skb, struct sock *meta_sk)
{
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
int chunk = min_t(unsigned int, skb->len, meta_tp->ucopy.len);
int eaten = 0;
__set_current_state(TASK_RUNNING);
local_bh_enable();
if (!skb_copy_datagram_iovec(skb, 0, meta_tp->ucopy.iov, chunk)) {
meta_tp->ucopy.len -= chunk;
meta_tp->copied_seq += chunk;
eaten = (chunk == skb->len);
tcp_rcv_space_adjust(meta_sk);
}
local_bh_disable();
return eaten;
}
static inline void mptcp_reset_mapping(struct tcp_sock *tp)
{
tp->mptcp->map_data_len = 0;
tp->mptcp->map_data_seq = 0;
tp->mptcp->map_subseq = 0;
tp->mptcp->map_data_fin = 0;
tp->mptcp->mapping_present = 0;
}
/* The DSS-mapping received on the sk only covers the second half of the skb
* (cut at seq). We trim the head from the skb.
* Data will be freed upon kfree().
*
* Inspired by tcp_trim_head().
*/
static void mptcp_skb_trim_head(struct sk_buff *skb, struct sock *sk, u32 seq)
{
int len = seq - TCP_SKB_CB(skb)->seq;
u32 new_seq = TCP_SKB_CB(skb)->seq + len;
if (len < skb_headlen(skb))
__skb_pull(skb, len);
else
__pskb_trim_head(skb, len - skb_headlen(skb));
TCP_SKB_CB(skb)->seq = new_seq;
skb->truesize -= len;
atomic_sub(len, &sk->sk_rmem_alloc);
sk_mem_uncharge(sk, len);
}
/* The DSS-mapping received on the sk only covers the first half of the skb
* (cut at seq). We create a second skb (@return), and queue it in the rcv-queue
* as further packets may resolve the mapping of the second half of data.
*
* Inspired by tcp_fragment().
*/
static int mptcp_skb_split_tail(struct sk_buff *skb, struct sock *sk, u32 seq)
{
struct sk_buff *buff;
int nsize;
int nlen, len;
len = seq - TCP_SKB_CB(skb)->seq;
nsize = skb_headlen(skb) - len + tcp_sk(sk)->tcp_header_len;
if (nsize < 0)
nsize = 0;
/* Get a new skb... force flag on. */
buff = alloc_skb(nsize, GFP_ATOMIC);
if (buff == NULL)
return -ENOMEM;
skb_reserve(buff, tcp_sk(sk)->tcp_header_len);
skb_reset_transport_header(buff);
tcp_hdr(buff)->fin = tcp_hdr(skb)->fin;
tcp_hdr(skb)->fin = 0;
/* We absolutly need to call skb_set_owner_r before refreshing the
* truesize of buff, otherwise the moved data will account twice.
*/
skb_set_owner_r(buff, sk);
nlen = skb->len - len - nsize;
buff->truesize += nlen;
skb->truesize -= nlen;
/* Correct the sequence numbers. */
TCP_SKB_CB(buff)->seq = TCP_SKB_CB(skb)->seq + len;
TCP_SKB_CB(buff)->end_seq = TCP_SKB_CB(skb)->end_seq;
TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(buff)->seq;
skb_split(skb, buff, len);
__skb_queue_after(&sk->sk_receive_queue, skb, buff);
return 0;
}
/* @return: 0 everything is fine. Just continue processing
* 1 subflow is broken stop everything
* -1 this packet was broken - continue with the next one.
*/
static int mptcp_prevalidate_skb(struct sock *sk, struct sk_buff *skb)
{
struct tcp_sock *tp = tcp_sk(sk);
/* If we are in infinite mode, the subflow-fin is in fact a data-fin. */
if (!skb->len && tcp_hdr(skb)->fin && !mptcp_is_data_fin(skb) &&
!tp->mpcb->infinite_mapping_rcv) {
/* Remove a pure subflow-fin from the queue and increase
* copied_seq.
*/
tp->copied_seq = TCP_SKB_CB(skb)->end_seq;
__skb_unlink(skb, &sk->sk_receive_queue);
__kfree_skb(skb);
return -1;
}
/* If we are not yet fully established and do not know the mapping for
* this segment, this path has to fallback to infinite or be torn down.
*/
if (!tp->mptcp->fully_established && !mptcp_is_data_seq(skb) &&
!tp->mptcp->mapping_present && !tp->mpcb->infinite_mapping_rcv) {
pr_err("%s %#x will fallback - pi %d from %pS, seq %u\n",
__func__, tp->mpcb->mptcp_loc_token,
tp->mptcp->path_index, __builtin_return_address(0),
TCP_SKB_CB(skb)->seq);
if (!is_master_tp(tp)) {
mptcp_send_reset(sk);
return 1;
}
tp->mpcb->infinite_mapping_snd = 1;
tp->mpcb->infinite_mapping_rcv = 1;
/* We do a seamless fallback and should not send a inf.mapping. */
tp->mpcb->send_infinite_mapping = 0;
tp->mptcp->fully_established = 1;
}
/* Receiver-side becomes fully established when a whole rcv-window has
* been received without the need to fallback due to the previous
* condition.
*/
if (!tp->mptcp->fully_established) {
tp->mptcp->init_rcv_wnd -= skb->len;
if (tp->mptcp->init_rcv_wnd < 0)
mptcp_become_fully_estab(sk);
}
return 0;
}
/* @return: 0 everything is fine. Just continue processing
* 1 subflow is broken stop everything
* -1 this packet was broken - continue with the next one.
*/
static int mptcp_detect_mapping(struct sock *sk, struct sk_buff *skb)
{
struct tcp_sock *tp = tcp_sk(sk), *meta_tp = mptcp_meta_tp(tp);
struct mptcp_cb *mpcb = tp->mpcb;
struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
u32 *ptr;
u32 data_seq, sub_seq, data_len, tcp_end_seq;
/* If we are in infinite-mapping-mode, the subflow is guaranteed to be
* in-order at the data-level. Thus data-seq-numbers can be inferred
* from what is expected at the data-level.
*/
if (mpcb->infinite_mapping_rcv) {
tp->mptcp->map_data_seq = mptcp_get_rcv_nxt_64(meta_tp);
tp->mptcp->map_subseq = tcb->seq;
tp->mptcp->map_data_len = skb->len;
tp->mptcp->map_data_fin = tcp_hdr(skb)->fin;
tp->mptcp->mapping_present = 1;
return 0;
}
/* No mapping here? Exit - it is either already set or still on its way */
if (!mptcp_is_data_seq(skb)) {
/* Too many packets without a mapping - this subflow is broken */
if (!tp->mptcp->mapping_present &&
tp->rcv_nxt - tp->copied_seq > 65536) {
mptcp_send_reset(sk);
return 1;
}
return 0;
}
ptr = mptcp_skb_set_data_seq(skb, &data_seq, mpcb);
ptr++;
sub_seq = get_unaligned_be32(ptr) + tp->mptcp->rcv_isn;
ptr++;
data_len = get_unaligned_be16(ptr);
/* If it's an empty skb with DATA_FIN, sub_seq must get fixed.
* The draft sets it to 0, but we really would like to have the
* real value, to have an easy handling afterwards here in this
* function.
*/
if (mptcp_is_data_fin(skb) && skb->len == 0)
sub_seq = TCP_SKB_CB(skb)->seq;
/* If there is already a mapping - we check if it maps with the current
* one. If not - we reset.
*/
if (tp->mptcp->mapping_present &&
(data_seq != (u32)tp->mptcp->map_data_seq ||
sub_seq != tp->mptcp->map_subseq ||
data_len != tp->mptcp->map_data_len + tp->mptcp->map_data_fin ||
mptcp_is_data_fin(skb) != tp->mptcp->map_data_fin)) {
/* Mapping in packet is different from what we want */
pr_err("%s Mappings do not match!\n", __func__);
pr_err("%s dseq %u mdseq %u, sseq %u msseq %u dlen %u mdlen %u dfin %d mdfin %d\n",
__func__, data_seq, (u32)tp->mptcp->map_data_seq,
sub_seq, tp->mptcp->map_subseq, data_len,
tp->mptcp->map_data_len, mptcp_is_data_fin(skb),
tp->mptcp->map_data_fin);
mptcp_send_reset(sk);
return 1;
}
/* If the previous check was good, the current mapping is valid and we exit. */
if (tp->mptcp->mapping_present)
return 0;
/* Mapping not yet set on this subflow - we set it here! */
if (!data_len) {
mpcb->infinite_mapping_rcv = 1;
tp->mptcp->fully_established = 1;
/* We need to repeat mp_fail's until the sender felt
* back to infinite-mapping - here we stop repeating it.
*/
tp->mptcp->send_mp_fail = 0;
/* We have to fixup data_len - it must be the same as skb->len */
data_len = skb->len + (mptcp_is_data_fin(skb) ? 1 : 0);
sub_seq = tcb->seq;
/* TODO kill all other subflows than this one */
/* data_seq and so on are set correctly */
/* At this point, the meta-ofo-queue has to be emptied,
* as the following data is guaranteed to be in-order at
* the data and subflow-level
*/
mptcp_purge_ofo_queue(meta_tp);
}
/* We are sending mp-fail's and thus are in fallback mode.
* Ignore packets which do not announce the fallback and still
* want to provide a mapping.
*/
if (tp->mptcp->send_mp_fail) {
tp->copied_seq = TCP_SKB_CB(skb)->end_seq;
__skb_unlink(skb, &sk->sk_receive_queue);
__kfree_skb(skb);
return -1;
}
/* FIN increased the mapping-length by 1 */
if (mptcp_is_data_fin(skb))
data_len--;
/* Subflow-sequences of packet must be
* (at least partially) be part of the DSS-mapping's
* subflow-sequence-space.
*
* Basically the mapping is not valid, if either of the
* following conditions is true:
*
* 1. It's not a data_fin and
* MPTCP-sub_seq >= TCP-end_seq
*
* 2. It's a data_fin and TCP-end_seq > TCP-seq and
* MPTCP-sub_seq >= TCP-end_seq
*
* The previous two can be merged into:
* TCP-end_seq > TCP-seq and MPTCP-sub_seq >= TCP-end_seq
* Because if it's not a data-fin, TCP-end_seq > TCP-seq
*
* 3. It's a data_fin and skb->len == 0 and
* MPTCP-sub_seq > TCP-end_seq
*
* 4. It's not a data_fin and TCP-end_seq > TCP-seq and
* MPTCP-sub_seq + MPTCP-data_len <= TCP-seq
*
* 5. MPTCP-sub_seq is prior to what we already copied (copied_seq)
*/
/* subflow-fin is not part of the mapping - ignore it here ! */
tcp_end_seq = tcb->end_seq - tcp_hdr(skb)->fin;
if ((!before(sub_seq, tcb->end_seq) && after(tcp_end_seq, tcb->seq)) ||
(mptcp_is_data_fin(skb) && skb->len == 0 && after(sub_seq, tcb->end_seq)) ||
(!after(sub_seq + data_len, tcb->seq) && after(tcp_end_seq, tcb->seq)) ||
before(sub_seq, tp->copied_seq)) {
/* Subflow-sequences of packet is different from what is in the
* packet's dss-mapping. The peer is misbehaving - reset
*/
pr_err("%s Packet's mapping does not map to the DSS sub_seq %u "
"end_seq %u, tcp_end_seq %u seq %u dfin %u len %u data_len %u"
"copied_seq %u\n", __func__, sub_seq, tcb->end_seq, tcp_end_seq, tcb->seq, mptcp_is_data_fin(skb),
skb->len, data_len, tp->copied_seq);
mptcp_send_reset(sk);
return 1;
}
/* Does the DSS had 64-bit seqnum's ? */
if (!(tcb->mptcp_flags & MPTCPHDR_SEQ64_SET)) {
/* Wrapped around? */
if (unlikely(after(data_seq, meta_tp->rcv_nxt) && data_seq < meta_tp->rcv_nxt)) {
tp->mptcp->map_data_seq = mptcp_get_data_seq_64(mpcb, !mpcb->rcv_hiseq_index, data_seq);
} else {
/* Else, access the default high-order bits */
tp->mptcp->map_data_seq = mptcp_get_data_seq_64(mpcb, mpcb->rcv_hiseq_index, data_seq);
}
} else {
tp->mptcp->map_data_seq = mptcp_get_data_seq_64(mpcb, (tcb->mptcp_flags & MPTCPHDR_SEQ64_INDEX) ? 1 : 0, data_seq);
if (unlikely(tcb->mptcp_flags & MPTCPHDR_SEQ64_OFO)) {
/* We make sure that the data_seq is invalid.
* It will be dropped later.
*/
tp->mptcp->map_data_seq += 0xFFFFFFFF;
tp->mptcp->map_data_seq += 0xFFFFFFFF;
}
}
tp->mptcp->map_data_len = data_len;
tp->mptcp->map_subseq = sub_seq;
tp->mptcp->map_data_fin = mptcp_is_data_fin(skb) ? 1 : 0;
tp->mptcp->mapping_present = 1;
return 0;
}
/* Similar to tcp_sequence(...) */
static inline bool mptcp_sequence(const struct tcp_sock *meta_tp,
u64 data_seq, u64 end_data_seq)
{
struct mptcp_cb *mpcb = meta_tp->mpcb;
u64 rcv_wup64;
/* Wrap-around? */
if (meta_tp->rcv_wup > meta_tp->rcv_nxt) {
rcv_wup64 = ((u64)(mpcb->rcv_high_order[mpcb->rcv_hiseq_index] - 1) << 32) |
meta_tp->rcv_wup;
} else {
rcv_wup64 = mptcp_get_data_seq_64(mpcb, mpcb->rcv_hiseq_index,
meta_tp->rcv_wup);
}
return !before64(end_data_seq, rcv_wup64) &&
!after64(data_seq, mptcp_get_rcv_nxt_64(meta_tp) + tcp_receive_window(meta_tp));
}
/* @return: 0 everything is fine. Just continue processing
* -1 this packet was broken - continue with the next one.
*/
static int mptcp_validate_mapping(struct sock *sk, struct sk_buff *skb)
{
struct tcp_sock *tp = tcp_sk(sk);
struct sk_buff *tmp, *tmp1;
u32 tcp_end_seq;
if (!tp->mptcp->mapping_present)
return 0;
/* either, the new skb gave us the mapping and the first segment
* in the sub-rcv-queue has to be trimmed ...
*/
tmp = skb_peek(&sk->sk_receive_queue);
if (before(TCP_SKB_CB(tmp)->seq, tp->mptcp->map_subseq) &&
after(TCP_SKB_CB(tmp)->end_seq, tp->mptcp->map_subseq))
mptcp_skb_trim_head(tmp, sk, tp->mptcp->map_subseq);
/* ... or the new skb (tail) has to be split at the end. */
tcp_end_seq = TCP_SKB_CB(skb)->end_seq - (tcp_hdr(skb)->fin ? 1 : 0);
if (after(tcp_end_seq, tp->mptcp->map_subseq + tp->mptcp->map_data_len)) {
u32 seq = tp->mptcp->map_subseq + tp->mptcp->map_data_len;
if (mptcp_skb_split_tail(skb, sk, seq)) { /* Allocation failed */
/* TODO : maybe handle this here better.
* We now just force meta-retransmission.
*/
tp->copied_seq = TCP_SKB_CB(skb)->end_seq;
__skb_unlink(skb, &sk->sk_receive_queue);
__kfree_skb(skb);
return -1;
}
}
/* Now, remove old sk_buff's from the receive-queue.
* This may happen if the mapping has been lost for these segments and
* the next mapping has already been received.
*/
if (before(TCP_SKB_CB(skb_peek(&sk->sk_receive_queue))->seq, tp->mptcp->map_subseq)) {
skb_queue_walk_safe(&sk->sk_receive_queue, tmp1, tmp) {
if (!before(TCP_SKB_CB(tmp1)->seq, tp->mptcp->map_subseq))
break;
tp->copied_seq = TCP_SKB_CB(tmp1)->end_seq;
__skb_unlink(tmp1, &sk->sk_receive_queue);
/* Impossible that we could free skb here, because his
* mapping is known to be valid from previous checks
*/
__kfree_skb(tmp1);
}
}
return 0;
}
/* @return: 0 everything is fine. Just continue processing
* 1 subflow is broken stop everything
* -1 this mapping has been put in the meta-receive-queue
* -2 this mapping has been eaten by the application
*/
static int mptcp_queue_skb(struct sock *sk)
{
struct tcp_sock *tp = tcp_sk(sk), *meta_tp = mptcp_meta_tp(tp);
struct sock *meta_sk = mptcp_meta_sk(sk);
struct mptcp_cb *mpcb = tp->mpcb;
struct sk_buff *tmp, *tmp1;
u64 rcv_nxt64 = mptcp_get_rcv_nxt_64(meta_tp);
bool data_queued = false;
/* Have we not yet received the full mapping? */
if (!tp->mptcp->mapping_present ||
before(tp->rcv_nxt, tp->mptcp->map_subseq + tp->mptcp->map_data_len))
return 0;
/* Is this an overlapping mapping? rcv_nxt >= end_data_seq
* OR
* This mapping is out of window
*/
if (!before64(rcv_nxt64, tp->mptcp->map_data_seq + tp->mptcp->map_data_len + tp->mptcp->map_data_fin) ||
!mptcp_sequence(meta_tp, tp->mptcp->map_data_seq,
tp->mptcp->map_data_seq + tp->mptcp->map_data_len + tp->mptcp->map_data_fin)) {
skb_queue_walk_safe(&sk->sk_receive_queue, tmp1, tmp) {
__skb_unlink(tmp1, &sk->sk_receive_queue);
tp->copied_seq = TCP_SKB_CB(tmp1)->end_seq;
__kfree_skb(tmp1);
if (!skb_queue_empty(&sk->sk_receive_queue) &&
!before(TCP_SKB_CB(tmp)->seq,
tp->mptcp->map_subseq + tp->mptcp->map_data_len))
break;
}
mptcp_reset_mapping(tp);
return -1;
}
/* Record it, because we want to send our data_fin on the same path */
if (tp->mptcp->map_data_fin) {
mpcb->dfin_path_index = tp->mptcp->path_index;
mpcb->dfin_combined = !!(sk->sk_shutdown & RCV_SHUTDOWN);
}
/* Verify the checksum */
if (mpcb->dss_csum && !mpcb->infinite_mapping_rcv) {
int ret = mptcp_verif_dss_csum(sk);
if (ret <= 0) {
mptcp_reset_mapping(tp);
return 1;
}
}
if (before64(rcv_nxt64, tp->mptcp->map_data_seq)) {
/* Seg's have to go to the meta-ofo-queue */
skb_queue_walk_safe(&sk->sk_receive_queue, tmp1, tmp) {
tp->copied_seq = TCP_SKB_CB(tmp1)->end_seq;
mptcp_prepare_skb(tmp1, tmp, sk);
__skb_unlink(tmp1, &sk->sk_receive_queue);
/* MUST be done here, because fragstolen may be true later.
* Then, kfree_skb_partial will not account the memory.
*/
skb_orphan(tmp1);
if (!mpcb->in_time_wait) /* In time-wait, do not receive data */
mptcp_add_meta_ofo_queue(meta_sk, tmp1, sk);
else
__kfree_skb(tmp1);
if (!skb_queue_empty(&sk->sk_receive_queue) &&
!before(TCP_SKB_CB(tmp)->seq,
tp->mptcp->map_subseq + tp->mptcp->map_data_len))
break;
}
tcp_enter_quickack_mode(sk);
} else {
/* Ready for the meta-rcv-queue */
skb_queue_walk_safe(&sk->sk_receive_queue, tmp1, tmp) {
int eaten = 0;
const bool copied_early = false;
bool fragstolen = false;
u32 old_rcv_nxt = meta_tp->rcv_nxt;
tp->copied_seq = TCP_SKB_CB(tmp1)->end_seq;
mptcp_prepare_skb(tmp1, tmp, sk);
__skb_unlink(tmp1, &sk->sk_receive_queue);
/* MUST be done here, because fragstolen may be true.
* Then, kfree_skb_partial will not account the memory.
*/
skb_orphan(tmp1);
/* This segment has already been received */
if (!after(TCP_SKB_CB(tmp1)->end_seq, meta_tp->rcv_nxt)) {
__kfree_skb(tmp1);
goto next;
}
#ifdef CONFIG_NET_DMA
if (TCP_SKB_CB(tmp1)->seq == meta_tp->rcv_nxt &&
meta_tp->ucopy.task == current &&
meta_tp->copied_seq == meta_tp->rcv_nxt &&
tmp1->len <= meta_tp->ucopy.len &&
sock_owned_by_user(meta_sk) &&
tcp_dma_try_early_copy(meta_sk, tmp1, 0)) {
copied_early = true;
eaten = 1;
}
#endif
/* Is direct copy possible ? */
if (TCP_SKB_CB(tmp1)->seq == meta_tp->rcv_nxt &&
meta_tp->ucopy.task == current &&
meta_tp->copied_seq == meta_tp->rcv_nxt &&
meta_tp->ucopy.len && sock_owned_by_user(meta_sk) &&
!copied_early)
eaten = mptcp_direct_copy(tmp1, meta_sk);
if (mpcb->in_time_wait) /* In time-wait, do not receive data */
eaten = 1;
if (!eaten)
eaten = tcp_queue_rcv(meta_sk, tmp1, 0, &fragstolen);
meta_tp->rcv_nxt = TCP_SKB_CB(tmp1)->end_seq;
mptcp_check_rcvseq_wrap(meta_tp, old_rcv_nxt);
//#ifdef CONFIG_NET_DMA
if (copied_early)
meta_tp->cleanup_rbuf(meta_sk, tmp1->len);
//#endif
if (tcp_hdr(tmp1)->fin && !mpcb->in_time_wait)
mptcp_fin(meta_sk);
/* Check if this fills a gap in the ofo queue */
if (!skb_queue_empty(&meta_tp->out_of_order_queue))
mptcp_ofo_queue(meta_sk);
#ifdef CONFIG_NET_DMA
if (copied_early)
__skb_queue_tail(&meta_sk->sk_async_wait_queue,
tmp1);
else
#endif
if (eaten)
kfree_skb_partial(tmp1, fragstolen);
data_queued = true;
next:
if (!skb_queue_empty(&sk->sk_receive_queue) &&
!before(TCP_SKB_CB(tmp)->seq,
tp->mptcp->map_subseq + tp->mptcp->map_data_len))
break;
}
}
inet_csk(meta_sk)->icsk_ack.lrcvtime = tcp_time_stamp;
mptcp_reset_mapping(tp);
return data_queued ? -1 : -2;
}
void mptcp_data_ready(struct sock *sk, int bytes)
{
struct sock *meta_sk = mptcp_meta_sk(sk);
struct sk_buff *skb, *tmp;
int queued = 0;
/* restart before the check, because mptcp_fin might have changed the
* state.
*/
restart:
/* If the meta cannot receive data, there is no point in pushing data.
* If we are in time-wait, we may still be waiting for the final FIN.
* So, we should proceed with the processing.
*/
if (!mptcp_sk_can_recv(meta_sk) && !tcp_sk(sk)->mpcb->in_time_wait) {
skb_queue_purge(&sk->sk_receive_queue);
tcp_sk(sk)->copied_seq = tcp_sk(sk)->rcv_nxt;
goto exit;
}
/* Iterate over all segments, detect their mapping (if we don't have
* one yet), validate them and push everything one level higher.
*/
skb_queue_walk_safe(&sk->sk_receive_queue, skb, tmp) {
int ret;
/* Pre-validation - e.g., early fallback */
ret = mptcp_prevalidate_skb(sk, skb);
if (ret < 0)
goto restart;
else if (ret > 0)
break;
/* Set the current mapping */
ret = mptcp_detect_mapping(sk, skb);
if (ret < 0)
goto restart;
else if (ret > 0)
break;
/* Validation */
if (mptcp_validate_mapping(sk, skb) < 0)
goto restart;
/* Push a level higher */
ret = mptcp_queue_skb(sk);
if (ret < 0) {
if (ret == -1)
queued = ret;
goto restart;
} else if (ret == 0) {
continue;
} else { /* ret == 1 */
break;
}
}
exit:
if (tcp_sk(sk)->close_it) {
tcp_send_ack(sk);
tcp_sk(sk)->time_wait(sk, TCP_TIME_WAIT, 0);
}
if (queued == -1 && !sock_flag(meta_sk, SOCK_DEAD))
meta_sk->sk_data_ready(meta_sk, 0);
}
int mptcp_check_req(struct sk_buff *skb, struct net *net)
{
struct tcphdr *th = tcp_hdr(skb);
struct sock *meta_sk = NULL;
/* MPTCP structures not initialized */
if (mptcp_init_failed)
return 0;
if (skb->protocol == htons(ETH_P_IP))
meta_sk = mptcp_v4_search_req(th->source, ip_hdr(skb)->saddr,
ip_hdr(skb)->daddr, net);
#if IS_ENABLED(CONFIG_IPV6)
else /* IPv6 */
meta_sk = mptcp_v6_search_req(th->source, &ipv6_hdr(skb)->saddr,
&ipv6_hdr(skb)->daddr, net);
#endif /* CONFIG_IPV6 */
if (!meta_sk)
return 0;
TCP_SKB_CB(skb)->mptcp_flags |= MPTCPHDR_JOIN;
bh_lock_sock_nested(meta_sk);
if (sock_owned_by_user(meta_sk)) {
skb->sk = meta_sk;
if (unlikely(sk_add_backlog(meta_sk, skb,
meta_sk->sk_rcvbuf + meta_sk->sk_sndbuf))) {
bh_unlock_sock(meta_sk);
NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
sock_put(meta_sk); /* Taken by mptcp_search_req */
kfree_skb(skb);
return 1;
}
} else if (skb->protocol == htons(ETH_P_IP)) {
tcp_v4_do_rcv(meta_sk, skb);
#if IS_ENABLED(CONFIG_IPV6)
} else { /* IPv6 */
tcp_v6_do_rcv(meta_sk, skb);
#endif /* CONFIG_IPV6 */
}
bh_unlock_sock(meta_sk);
sock_put(meta_sk); /* Taken by mptcp_vX_search_req */
return 1;
}
struct mp_join *mptcp_find_join(struct sk_buff *skb)
{
struct tcphdr *th = tcp_hdr(skb);
unsigned char *ptr;
int length = (th->doff * 4) - sizeof(struct tcphdr);
/* Jump through the options to check whether JOIN is there */
ptr = (unsigned char *)(th + 1);
while (length > 0) {
int opcode = *ptr++;
int opsize;
switch (opcode) {
case TCPOPT_EOL:
return NULL;
case TCPOPT_NOP: /* Ref: RFC 793 section 3.1 */
length--;
continue;
default:
opsize = *ptr++;
if (opsize < 2) /* "silly options" */
return NULL;
if (opsize > length)
return NULL; /* don't parse partial options */
if (opcode == TCPOPT_MPTCP &&
((struct mptcp_option *)(ptr - 2))->sub == MPTCP_SUB_JOIN) {
return (struct mp_join *)(ptr - 2);
}
ptr += opsize - 2;
length -= opsize;
}
}
return NULL;
}
int mptcp_lookup_join(struct sk_buff *skb, struct inet_timewait_sock *tw)
{
struct mptcp_cb *mpcb;
struct sock *meta_sk;
u32 token;
struct mp_join *join_opt = mptcp_find_join(skb);
if (!join_opt)
return 0;
/* MPTCP structures were not initialized, so return error */
if (mptcp_init_failed)
return -1;
token = join_opt->u.syn.token;
meta_sk = mptcp_hash_find(dev_net(skb_dst(skb)->dev), token);
if (!meta_sk) {
mptcp_debug("%s:mpcb not found:%x\n", __func__, token);
return -1;
}
mpcb = tcp_sk(meta_sk)->mpcb;
if (mpcb->infinite_mapping_rcv || mpcb->send_infinite_mapping) {
/* We are in fallback-mode on the reception-side -
* no new subflows!
*/
sock_put(meta_sk); /* Taken by mptcp_hash_find */
return -1;
}
/* Coming from time-wait-sock processing in tcp_v4_rcv.
* We have to deschedule it before continuing, because otherwise
* mptcp_v4_do_rcv will hit again on it inside tcp_v4_hnd_req.
*/
if (tw) {
inet_twsk_deschedule(tw, &tcp_death_row);
inet_twsk_put(tw);
}
TCP_SKB_CB(skb)->mptcp_flags |= MPTCPHDR_JOIN;
/* OK, this is a new syn/join, let's create a new open request and
* send syn+ack
*/
bh_lock_sock_nested(meta_sk);
if (sock_owned_by_user(meta_sk)) {
skb->sk = meta_sk;
if (unlikely(sk_add_backlog(meta_sk, skb,
meta_sk->sk_rcvbuf + meta_sk->sk_sndbuf))) {
bh_unlock_sock(meta_sk);
NET_INC_STATS_BH(sock_net(meta_sk),
LINUX_MIB_TCPBACKLOGDROP);
sock_put(meta_sk); /* Taken by mptcp_hash_find */
kfree_skb(skb);
return 1;
}
} else if (skb->protocol == htons(ETH_P_IP)) {
tcp_v4_do_rcv(meta_sk, skb);
#if IS_ENABLED(CONFIG_IPV6)
} else {
tcp_v6_do_rcv(meta_sk, skb);
#endif /* CONFIG_IPV6 */
}
bh_unlock_sock(meta_sk);
sock_put(meta_sk); /* Taken by mptcp_hash_find */
return 1;
}
int mptcp_do_join_short(struct sk_buff *skb, struct mptcp_options_received *mopt,
struct net *net)
{
struct sock *meta_sk;
u32 token;
token = mopt->mptcp_rem_token;
meta_sk = mptcp_hash_find(net, token);
if (!meta_sk) {
mptcp_debug("%s:mpcb not found:%x\n", __func__, token);
return -1;
}
TCP_SKB_CB(skb)->mptcp_flags |= MPTCPHDR_JOIN;
/* OK, this is a new syn/join, let's create a new open request and
* send syn+ack
*/
bh_lock_sock(meta_sk);
/* This check is also done in mptcp_vX_do_rcv. But, there we cannot
* call tcp_vX_send_reset, because we hold already two socket-locks.
* (the listener and the meta from above)
*
* And the send-reset will try to take yet another one (ip_send_reply).
* Thus, we propagate the reset up to tcp_rcv_state_process.
*/
if (tcp_sk(meta_sk)->mpcb->infinite_mapping_rcv ||
tcp_sk(meta_sk)->mpcb->send_infinite_mapping ||
meta_sk->sk_state == TCP_CLOSE || !tcp_sk(meta_sk)->inside_tk_table) {
bh_unlock_sock(meta_sk);
sock_put(meta_sk); /* Taken by mptcp_hash_find */
return -1;
}
if (sock_owned_by_user(meta_sk)) {
skb->sk = meta_sk;
if (unlikely(sk_add_backlog(meta_sk, skb,
meta_sk->sk_rcvbuf + meta_sk->sk_sndbuf)))
NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
else
/* Must make sure that upper layers won't free the
* skb if it is added to the backlog-queue.
*/
skb_get(skb);
} else {
/* mptcp_v4_do_rcv tries to free the skb - we prevent this, as
* the skb will finally be freed by tcp_v4_do_rcv (where we are
* coming from)
*/
skb_get(skb);
if (skb->protocol == htons(ETH_P_IP)) {
tcp_v4_do_rcv(meta_sk, skb);
#if IS_ENABLED(CONFIG_IPV6)
} else { /* IPv6 */
tcp_v6_do_rcv(meta_sk, skb);
#endif /* CONFIG_IPV6 */
}
}
bh_unlock_sock(meta_sk);
sock_put(meta_sk); /* Taken by mptcp_hash_find */
return 0;
}
/**
* Equivalent of tcp_fin() for MPTCP
* Can be called only when the FIN is validly part
* of the data seqnum space. Not before when we get holes.
*/
void mptcp_fin(struct sock *meta_sk)
{
struct sock *sk = NULL, *sk_it;
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
struct mptcp_cb *mpcb = meta_tp->mpcb;
mptcp_for_each_sk(mpcb, sk_it) {
if (tcp_sk(sk_it)->mptcp->path_index == mpcb->dfin_path_index) {
sk = sk_it;
break;
}
}
if (!sk || sk->sk_state == TCP_CLOSE)
sk = mptcp_select_ack_sock(meta_sk);
inet_csk_schedule_ack(sk);
meta_sk->sk_shutdown |= RCV_SHUTDOWN;
sock_set_flag(meta_sk, SOCK_DONE);
switch (meta_sk->sk_state) {
case TCP_SYN_RECV:
case TCP_ESTABLISHED:
/* Move to CLOSE_WAIT */
tcp_set_state(meta_sk, TCP_CLOSE_WAIT);
inet_csk(sk)->icsk_ack.pingpong = 1;
break;
case TCP_CLOSE_WAIT:
case TCP_CLOSING:
/* Received a retransmission of the FIN, do
* nothing.
*/
break;
case TCP_LAST_ACK:
/* RFC793: Remain in the LAST-ACK state. */
break;
case TCP_FIN_WAIT1:
/* This case occurs when a simultaneous close
* happens, we must ack the received FIN and
* enter the CLOSING state.
*/
tcp_send_ack(sk);
tcp_set_state(meta_sk, TCP_CLOSING);
break;
case TCP_FIN_WAIT2:
/* Received a FIN -- send ACK and enter TIME_WAIT. */
tcp_send_ack(sk);
meta_tp->time_wait(meta_sk, TCP_TIME_WAIT, 0);
break;
default:
/* Only TCP_LISTEN and TCP_CLOSE are left, in these
* cases we should never reach this piece of code.
*/
pr_err("%s: Impossible, meta_sk->sk_state=%d\n", __func__,
meta_sk->sk_state);
break;
}
/* It _is_ possible, that we have something out-of-order _after_ FIN.
* Probably, we should reset in this case. For now drop them.
*/
mptcp_purge_ofo_queue(meta_tp);
sk_mem_reclaim(meta_sk);
if (!sock_flag(meta_sk, SOCK_DEAD)) {
meta_sk->sk_state_change(meta_sk);
/* Do not send POLL_HUP for half duplex close. */
if (meta_sk->sk_shutdown == SHUTDOWN_MASK ||
meta_sk->sk_state == TCP_CLOSE)
sk_wake_async(meta_sk, SOCK_WAKE_WAITD, POLL_HUP);
else
sk_wake_async(meta_sk, SOCK_WAKE_WAITD, POLL_IN);
}
return;
}
static void mptcp_xmit_retransmit_queue(struct sock *meta_sk)
{
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
struct sk_buff *skb;
if (!meta_tp->packets_out)
return;
tcp_for_write_queue(skb, meta_sk) {
if (skb == tcp_send_head(meta_sk))
break;
if (mptcp_retransmit_skb(meta_sk, skb))
return;
if (skb == tcp_write_queue_head(meta_sk))
inet_csk_reset_xmit_timer(meta_sk, ICSK_TIME_RETRANS,
inet_csk(meta_sk)->icsk_rto,
TCP_RTO_MAX);
}
}
/* Handle the DATA_ACK */
static void mptcp_data_ack(struct sock *sk, const struct sk_buff *skb)
{
struct sock *meta_sk = mptcp_meta_sk(sk);
struct tcp_sock *meta_tp = tcp_sk(meta_sk), *tp = tcp_sk(sk);
struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
u32 prior_snd_una = meta_tp->snd_una;
int prior_packets;
u32 nwin, data_ack, data_seq;
u16 data_len = 0;
/* A valid packet came in - subflow is operational again */
tp->pf = 0;
/* Even if there is no data-ack, we stop retransmitting.
* Except if this is a SYN/ACK. Then it is just a retransmission
*/
if (tp->mptcp->pre_established && !tcp_hdr(skb)->syn) {
tp->mptcp->pre_established = 0;
sk_stop_timer(sk, &tp->mptcp->mptcp_ack_timer);
}
/* If we are in infinite mapping mode, rx_opt.data_ack has been
* set by mptcp_clean_rtx_infinite.
*/
if (!(tcb->mptcp_flags & MPTCPHDR_ACK) && !tp->mpcb->infinite_mapping_snd)
goto exit;
data_ack = tp->mptcp->rx_opt.data_ack;
if (unlikely(!tp->mptcp->fully_established) &&
tp->mptcp->snt_isn + 1 != TCP_SKB_CB(skb)->ack_seq)
/* As soon as a subflow-data-ack (not acking syn, thus snt_isn + 1)
* includes a data-ack, we are fully established
*/
mptcp_become_fully_estab(sk);
/* Get the data_seq */
if (mptcp_is_data_seq(skb)) {
data_seq = tp->mptcp->rx_opt.data_seq;
data_len = tp->mptcp->rx_opt.data_len;
} else {
data_seq = meta_tp->snd_wl1;
}
/* If the ack is older than previous acks
* then we can probably ignore it.
*/
if (before(data_ack, prior_snd_una))
goto exit;
/* If the ack includes data we haven't sent yet, discard
* this segment (RFC793 Section 3.9).
*/
if (after(data_ack, meta_tp->snd_nxt))
goto exit;
/*** Now, update the window - inspired by tcp_ack_update_window ***/
nwin = ntohs(tcp_hdr(skb)->window);
if (likely(!tcp_hdr(skb)->syn))
nwin <<= tp->rx_opt.snd_wscale;
if (tcp_may_update_window(meta_tp, data_ack, data_seq, nwin)) {
tcp_update_wl(meta_tp, data_seq);
/* Draft v09, Section 3.3.5:
* [...] It should only update its local receive window values
* when the largest sequence number allowed (i.e. DATA_ACK +
* receive window) increases. [...]
*/
if (meta_tp->snd_wnd != nwin &&
!before(data_ack + nwin, tcp_wnd_end(meta_tp))) {
meta_tp->snd_wnd = nwin;
if (nwin > meta_tp->max_window)
meta_tp->max_window = nwin;
}
}
/*** Done, update the window ***/
/* We passed data and got it acked, remove any soft error
* log. Something worked...
*/
sk->sk_err_soft = 0;
inet_csk(meta_sk)->icsk_probes_out = 0;
meta_tp->rcv_tstamp = tcp_time_stamp;
prior_packets = meta_tp->packets_out;
if (!prior_packets)
goto no_queue;
meta_tp->snd_una = data_ack;
mptcp_clean_rtx_queue(meta_sk, prior_snd_una);
/* We are in loss-state, and something got acked, retransmit the whole
* queue now!
*/
if (inet_csk(meta_sk)->icsk_ca_state == TCP_CA_Loss &&
after(data_ack, prior_snd_una)) {
mptcp_xmit_retransmit_queue(meta_sk);
inet_csk(meta_sk)->icsk_ca_state = TCP_CA_Open;
}
/* Simplified version of tcp_new_space, because the snd-buffer
* is handled by all the subflows.
*/
if (sock_flag(meta_sk, SOCK_QUEUE_SHRUNK)) {
sock_reset_flag(meta_sk, SOCK_QUEUE_SHRUNK);
if (meta_sk->sk_socket &&
test_bit(SOCK_NOSPACE, &meta_sk->sk_socket->flags))
meta_sk->sk_write_space(meta_sk);
}
if (meta_sk->sk_state != TCP_ESTABLISHED &&
mptcp_rcv_state_process(meta_sk, sk, skb, data_seq, data_len))
return;
exit:
mptcp_push_pending_frames(meta_sk);
return;
no_queue:
if (tcp_send_head(meta_sk))
tcp_ack_probe(meta_sk);
mptcp_push_pending_frames(meta_sk);
return;
}
void mptcp_clean_rtx_infinite(struct sk_buff *skb, struct sock *sk)
{
struct tcp_sock *tp = tcp_sk(sk), *meta_tp = tcp_sk(mptcp_meta_sk(sk));
if (!tp->mpcb->infinite_mapping_snd)
return;
/* The difference between both write_seq's represents the offset between
* data-sequence and subflow-sequence. As we are infinite, this must
* match.
*
* Thus, from this difference we can infer the meta snd_una.
*/
tp->mptcp->rx_opt.data_ack = meta_tp->snd_nxt - tp->snd_nxt +
tp->snd_una;
mptcp_data_ack(sk, skb);
}
/**** static functions used by mptcp_parse_options */
static void mptcp_send_reset_rem_id(const struct mptcp_cb *mpcb, u8 rem_id)
{
struct sock *sk_it, *tmpsk;
mptcp_for_each_sk_safe(mpcb, sk_it, tmpsk) {
if (tcp_sk(sk_it)->mptcp->rem_id == rem_id) {
mptcp_reinject_data(sk_it, 0);
sk_it->sk_err = ECONNRESET;
if (tcp_need_reset(sk_it->sk_state))
tcp_sk(sk_it)->send_active_reset(sk_it,
GFP_ATOMIC);
mptcp_sub_force_close(sk_it);
}
}
}
void mptcp_parse_options(const uint8_t *ptr, int opsize,
struct mptcp_options_received *mopt,
const struct sk_buff *skb)
{
struct mptcp_option *mp_opt = (struct mptcp_option *)ptr;
/* If the socket is mp-capable we would have a mopt. */
if (!mopt)
return;
switch (mp_opt->sub) {
case MPTCP_SUB_CAPABLE:
{
struct mp_capable *mpcapable = (struct mp_capable *)ptr;
if (opsize != MPTCP_SUB_LEN_CAPABLE_SYN &&
opsize != MPTCP_SUB_LEN_CAPABLE_ACK) {
mptcp_debug("%s: mp_capable: bad option size %d\n",
__func__, opsize);
break;
}
/* We only support MPTCP version 0 */
if (mpcapable->ver != 0)
break;
/* MPTCP-RFC 6824:
* "If receiving a message with the 'B' flag set to 1, and this
* is not understood, then this SYN MUST be silently ignored;
*/
if (mpcapable->b) {
mopt->drop_me = 1;
break;
}
/* MPTCP-RFC 6824:
* "An implementation that only supports this method MUST set
* bit "H" to 1, and bits "C" through "G" to 0."
*/
if (!mpcapable->h)
break;
mopt->saw_mpc = 1;
mopt->dss_csum = sysctl_mptcp_checksum || mpcapable->a;
if (opsize >= MPTCP_SUB_LEN_CAPABLE_SYN)
mopt->mptcp_key = mpcapable->sender_key;
break;
}
case MPTCP_SUB_JOIN:
{
struct mp_join *mpjoin = (struct mp_join *)ptr;
if (opsize != MPTCP_SUB_LEN_JOIN_SYN &&
opsize != MPTCP_SUB_LEN_JOIN_SYNACK &&
opsize != MPTCP_SUB_LEN_JOIN_ACK) {
mptcp_debug("%s: mp_join: bad option size %d\n",
__func__, opsize);
break;
}
/* saw_mpc must be set, because in tcp_check_req we assume that
* it is set to support falling back to reg. TCP if a rexmitted
* SYN has no MP_CAPABLE or MP_JOIN
*/
switch (opsize) {
case MPTCP_SUB_LEN_JOIN_SYN:
mopt->is_mp_join = 1;
mopt->saw_mpc = 1;
mopt->low_prio = mpjoin->b;
mopt->rem_id = mpjoin->addr_id;
mopt->mptcp_rem_token = mpjoin->u.syn.token;
mopt->mptcp_recv_nonce = mpjoin->u.syn.nonce;
break;
case MPTCP_SUB_LEN_JOIN_SYNACK:
mopt->saw_mpc = 1;
mopt->low_prio = mpjoin->b;
mopt->rem_id = mpjoin->addr_id;
mopt->mptcp_recv_tmac = mpjoin->u.synack.mac;
mopt->mptcp_recv_nonce = mpjoin->u.synack.nonce;
break;
case MPTCP_SUB_LEN_JOIN_ACK:
mopt->saw_mpc = 1;
mopt->join_ack = 1;
memcpy(mopt->mptcp_recv_mac, mpjoin->u.ack.mac, 20);
break;
}
break;
}
case MPTCP_SUB_DSS:
{
struct mp_dss *mdss = (struct mp_dss *)ptr;
struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
/* We check opsize for the csum and non-csum case. We do this,
* because the draft says that the csum SHOULD be ignored if
* it has not been negotiated in the MP_CAPABLE but still is
* present in the data.
*
* It will get ignored later in mptcp_queue_skb.
*/
if (opsize != mptcp_sub_len_dss(mdss, 0) &&
opsize != mptcp_sub_len_dss(mdss, 1)) {
mptcp_debug("%s: mp_dss: bad option size %d\n",
__func__, opsize);
break;
}
ptr += 4;
if (mdss->A) {
tcb->mptcp_flags |= MPTCPHDR_ACK;
if (mdss->a) {
mopt->data_ack = (u32) get_unaligned_be64(ptr);
ptr += MPTCP_SUB_LEN_ACK_64;
} else {
mopt->data_ack = get_unaligned_be32(ptr);
ptr += MPTCP_SUB_LEN_ACK;
}
}
tcb->dss_off = (ptr - skb_transport_header(skb));
if (mdss->M) {
if (mdss->m) {
u64 data_seq64 = get_unaligned_be64(ptr);
tcb->mptcp_flags |= MPTCPHDR_SEQ64_SET;
mopt->data_seq = (u32) data_seq64;
ptr += 12; /* 64-bit dseq + subseq */
} else {
mopt->data_seq = get_unaligned_be32(ptr);
ptr += 8; /* 32-bit dseq + subseq */
}
mopt->data_len = get_unaligned_be16(ptr);
tcb->mptcp_flags |= MPTCPHDR_SEQ;
/* Is a check-sum present? */
if (opsize == mptcp_sub_len_dss(mdss, 1))
tcb->mptcp_flags |= MPTCPHDR_DSS_CSUM;
/* DATA_FIN only possible with DSS-mapping */
if (mdss->F)
tcb->mptcp_flags |= MPTCPHDR_FIN;
}
break;
}
case MPTCP_SUB_ADD_ADDR:
{
#if IS_ENABLED(CONFIG_IPV6)
struct mp_add_addr *mpadd = (struct mp_add_addr *)ptr;
if ((mpadd->ipver == 4 && opsize != MPTCP_SUB_LEN_ADD_ADDR4 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR4 + 2) ||
(mpadd->ipver == 6 && opsize != MPTCP_SUB_LEN_ADD_ADDR6 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR6 + 2)) {
#else
if (opsize != MPTCP_SUB_LEN_ADD_ADDR4 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR4 + 2) {
#endif /* CONFIG_IPV6 */
mptcp_debug("%s: mp_add_addr: bad option size %d\n",
__func__, opsize);
break;
}
/* We have to manually parse the options if we got two of them. */
if (mopt->saw_add_addr) {
mopt->more_add_addr = 1;
break;
}
mopt->saw_add_addr = 1;
mopt->add_addr_ptr = ptr;
break;
}
case MPTCP_SUB_REMOVE_ADDR:
if ((opsize - MPTCP_SUB_LEN_REMOVE_ADDR) < 0) {
mptcp_debug("%s: mp_remove_addr: bad option size %d\n",
__func__, opsize);
break;
}
if (mopt->saw_rem_addr) {
mopt->more_rem_addr = 1;
break;
}
mopt->saw_rem_addr = 1;
mopt->rem_addr_ptr = ptr;
break;
case MPTCP_SUB_PRIO:
{
struct mp_prio *mpprio = (struct mp_prio *)ptr;
if (opsize != MPTCP_SUB_LEN_PRIO &&
opsize != MPTCP_SUB_LEN_PRIO_ADDR) {
mptcp_debug("%s: mp_prio: bad option size %d\n",
__func__, opsize);
break;
}
mopt->saw_low_prio = 1;
mopt->low_prio = mpprio->b;
if (opsize == MPTCP_SUB_LEN_PRIO_ADDR) {
mopt->saw_low_prio = 2;
mopt->prio_addr_id = mpprio->addr_id;
}
break;
}
case MPTCP_SUB_FAIL:
if (opsize != MPTCP_SUB_LEN_FAIL) {
mptcp_debug("%s: mp_fail: bad option size %d\n",
__func__, opsize);
break;
}
mopt->mp_fail = 1;
break;
case MPTCP_SUB_FCLOSE:
if (opsize != MPTCP_SUB_LEN_FCLOSE) {
mptcp_debug("%s: mp_fclose: bad option size %d\n",
__func__, opsize);
break;
}
mopt->mp_fclose = 1;
mopt->mptcp_key = ((struct mp_fclose *)ptr)->key;
break;
default:
mptcp_debug("%s: Received unkown subtype: %d\n",
__func__, mp_opt->sub);
break;
}
}
/** Parse only MPTCP options */
void tcp_parse_mptcp_options(const struct sk_buff *skb,
struct mptcp_options_received *mopt)
{
const struct tcphdr *th = tcp_hdr(skb);
int length = (th->doff * 4) - sizeof(struct tcphdr);
const unsigned char *ptr = (const unsigned char *)(th + 1);
while (length > 0) {
int opcode = *ptr++;
int opsize;
switch (opcode) {
case TCPOPT_EOL:
return;
case TCPOPT_NOP: /* Ref: RFC 793 section 3.1 */
length--;
continue;
default:
opsize = *ptr++;
if (opsize < 2) /* "silly options" */
return;
if (opsize > length)
return; /* don't parse partial options */
if (opcode == TCPOPT_MPTCP)
mptcp_parse_options(ptr - 2, opsize, mopt, skb);
}
ptr += opsize - 2;
length -= opsize;
}
}
int mptcp_check_rtt(const struct tcp_sock *tp, int time)
{
struct mptcp_cb *mpcb = tp->mpcb;
struct sock *sk;
u32 rtt_max = 0;
/* In MPTCP, we take the max delay across all flows,
* in order to take into account meta-reordering buffers.
*/
mptcp_for_each_sk(mpcb, sk) {
if (!mptcp_sk_can_recv(sk))
continue;
if (rtt_max < tcp_sk(sk)->rcv_rtt_est.rtt)
rtt_max = tcp_sk(sk)->rcv_rtt_est.rtt;
}
if (time < (rtt_max >> 3) || !rtt_max)
return 1;
return 0;
}
static void mptcp_handle_add_addr(const unsigned char *ptr, struct sock *sk)
{
struct mp_add_addr *mpadd = (struct mp_add_addr *)ptr;
struct mptcp_cb *mpcb = tcp_sk(sk)->mpcb;
__be16 port = 0;
union inet_addr addr;
sa_family_t family;
if (mpadd->ipver == 4) {
if (mpadd->len == MPTCP_SUB_LEN_ADD_ADDR4 + 2)
port = mpadd->u.v4.port;
family = AF_INET;
addr.in = mpadd->u.v4.addr;
#if IS_ENABLED(CONFIG_IPV6)
} else if (mpadd->ipver == 6) {
if (mpadd->len == MPTCP_SUB_LEN_ADD_ADDR6 + 2)
port = mpadd->u.v6.port;
family = AF_INET6;
addr.in6 = mpadd->u.v6.addr;
#endif /* CONFIG_IPV6 */
} else {
return;
}
if (mpcb->pm_ops->add_raddr)
mpcb->pm_ops->add_raddr(mpcb, &addr, family, port, mpadd->addr_id);
}
static void mptcp_handle_rem_addr(const unsigned char *ptr, struct sock *sk)
{
struct mp_remove_addr *mprem = (struct mp_remove_addr *)ptr;
int i;
u8 rem_id;
struct mptcp_cb *mpcb = tcp_sk(sk)->mpcb;
for (i = 0; i <= mprem->len - MPTCP_SUB_LEN_REMOVE_ADDR; i++) {
rem_id = (&mprem->addrs_id)[i];
if (mpcb->pm_ops->rem_raddr)
mpcb->pm_ops->rem_raddr(mpcb, rem_id);
mptcp_send_reset_rem_id(mpcb, rem_id);
}
}
static void mptcp_parse_addropt(const struct sk_buff *skb, struct sock *sk)
{
struct tcphdr *th = tcp_hdr(skb);
unsigned char *ptr;
int length = (th->doff * 4) - sizeof(struct tcphdr);
/* Jump through the options to check whether ADD_ADDR is there */
ptr = (unsigned char *)(th + 1);
while (length > 0) {
int opcode = *ptr++;
int opsize;
switch (opcode) {
case TCPOPT_EOL:
return;
case TCPOPT_NOP:
length--;
continue;
default:
opsize = *ptr++;
if (opsize < 2)
return;
if (opsize > length)
return; /* don't parse partial options */
if (opcode == TCPOPT_MPTCP &&
((struct mptcp_option *)ptr)->sub == MPTCP_SUB_ADD_ADDR) {
#if IS_ENABLED(CONFIG_IPV6)
struct mp_add_addr *mpadd = (struct mp_add_addr *)ptr;
if ((mpadd->ipver == 4 && opsize != MPTCP_SUB_LEN_ADD_ADDR4 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR4 + 2) ||
(mpadd->ipver == 6 && opsize != MPTCP_SUB_LEN_ADD_ADDR6 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR6 + 2))
#else
if (opsize != MPTCP_SUB_LEN_ADD_ADDR4 &&
opsize != MPTCP_SUB_LEN_ADD_ADDR4 + 2)
#endif /* CONFIG_IPV6 */
goto cont;
mptcp_handle_add_addr(ptr, sk);
}
if (opcode == TCPOPT_MPTCP &&
((struct mptcp_option *)ptr)->sub == MPTCP_SUB_REMOVE_ADDR) {
if ((opsize - MPTCP_SUB_LEN_REMOVE_ADDR) < 0)
goto cont;
mptcp_handle_rem_addr(ptr, sk);
}
cont:
ptr += opsize - 2;
length -= opsize;
}
}
return;
}
static inline int mptcp_mp_fail_rcvd(struct sock *sk, const struct tcphdr *th)
{
struct mptcp_tcp_sock *mptcp = tcp_sk(sk)->mptcp;
struct sock *meta_sk = mptcp_meta_sk(sk);
struct mptcp_cb *mpcb = tcp_sk(sk)->mpcb;
if (unlikely(mptcp->rx_opt.mp_fail)) {
mptcp->rx_opt.mp_fail = 0;
if (!th->rst && !mpcb->infinite_mapping_snd) {
struct sock *sk_it;
mpcb->send_infinite_mapping = 1;
/* We resend everything that has not been acknowledged */
meta_sk->sk_send_head = tcp_write_queue_head(meta_sk);
/* We artificially restart the whole send-queue. Thus,
* it is as if no packets are in flight
*/
tcp_sk(meta_sk)->packets_out = 0;
/* If the snd_nxt already wrapped around, we have to
* undo the wrapping, as we are restarting from snd_una
* on.
*/
if (tcp_sk(meta_sk)->snd_nxt < tcp_sk(meta_sk)->snd_una) {
mpcb->snd_high_order[mpcb->snd_hiseq_index] -= 2;
mpcb->snd_hiseq_index = mpcb->snd_hiseq_index ? 0 : 1;
}
tcp_sk(meta_sk)->snd_nxt = tcp_sk(meta_sk)->snd_una;
/* Trigger a sending on the meta. */
mptcp_push_pending_frames(meta_sk);
mptcp_for_each_sk(mpcb, sk_it) {
if (sk != sk_it)
mptcp_sub_force_close(sk_it);
}
}
return 0;
}
if (unlikely(mptcp->rx_opt.mp_fclose)) {
struct sock *sk_it, *tmpsk;
mptcp->rx_opt.mp_fclose = 0;
if (mptcp->rx_opt.mptcp_key != mpcb->mptcp_loc_key)
return 0;
if (tcp_need_reset(sk->sk_state))
tcp_sk(sk)->send_active_reset(sk, GFP_ATOMIC);
mptcp_for_each_sk_safe(mpcb, sk_it, tmpsk)
mptcp_sub_force_close(sk_it);
tcp_reset(meta_sk);
return 1;
}
return 0;
}
static inline void mptcp_path_array_check(struct sock *meta_sk)
{
struct mptcp_cb *mpcb = tcp_sk(meta_sk)->mpcb;
if (unlikely(mpcb->list_rcvd)) {
mpcb->list_rcvd = 0;
if (mpcb->pm_ops->new_remote_address)
mpcb->pm_ops->new_remote_address(meta_sk);
}
}
int mptcp_handle_options(struct sock *sk, const struct tcphdr *th, struct sk_buff *skb)
{
struct tcp_sock *tp = tcp_sk(sk);
struct mptcp_options_received *mopt = &tp->mptcp->rx_opt;
if (tp->mpcb->infinite_mapping_rcv || tp->mpcb->infinite_mapping_snd)
return 0;
if (mptcp_mp_fail_rcvd(sk, th))
return 1;
/* RFC 6824, Section 3.3:
* If a checksum is not present when its use has been negotiated, the
* receiver MUST close the subflow with a RST as it is considered broken.
*/
if (mptcp_is_data_seq(skb) && tp->mpcb->dss_csum &&
!(TCP_SKB_CB(skb)->mptcp_flags & MPTCPHDR_DSS_CSUM)) {
if (tcp_need_reset(sk->sk_state))
tp->send_active_reset(sk, GFP_ATOMIC);
mptcp_sub_force_close(sk);
return 1;
}
/* We have to acknowledge retransmissions of the third
* ack.
*/
if (mopt->join_ack) {
tcp_send_delayed_ack(sk);
mopt->join_ack = 0;
}
if (mopt->saw_add_addr || mopt->saw_rem_addr) {
if (mopt->more_add_addr || mopt->more_rem_addr) {
mptcp_parse_addropt(skb, sk);
} else {
if (mopt->saw_add_addr)
mptcp_handle_add_addr(mopt->add_addr_ptr, sk);
if (mopt->saw_rem_addr)
mptcp_handle_rem_addr(mopt->rem_addr_ptr, sk);
}
mopt->more_add_addr = 0;
mopt->saw_add_addr = 0;
mopt->more_rem_addr = 0;
mopt->saw_rem_addr = 0;
}
if (mopt->saw_low_prio) {
if (mopt->saw_low_prio == 1) {
tp->mptcp->rcv_low_prio = mopt->low_prio;
} else {
struct sock *sk_it;
mptcp_for_each_sk(tp->mpcb, sk_it) {
struct mptcp_tcp_sock *mptcp = tcp_sk(sk_it)->mptcp;
if (mptcp->rem_id == mopt->prio_addr_id)
mptcp->rcv_low_prio = mopt->low_prio;
}
}
mopt->saw_low_prio = 0;
}
mptcp_data_ack(sk, skb);
mptcp_path_array_check(mptcp_meta_sk(sk));
/* Socket may have been mp_killed by a REMOVE_ADDR */
if (tp->mp_killed)
return 1;
return 0;
}
/* The skptr is needed, because if we become MPTCP-capable, we have to switch
* from meta-socket to master-socket.
*
* @return: 1 - we want to reset this connection
* 2 - we want to discard the received syn/ack
* 0 - everything is fine - continue
*/
int mptcp_rcv_synsent_state_process(struct sock *sk, struct sock **skptr,
struct sk_buff *skb,
struct mptcp_options_received *mopt)
{
struct tcp_sock *tp = tcp_sk(sk);
if (mptcp(tp)) {
u8 hash_mac_check[20];
struct mptcp_cb *mpcb = tp->mpcb;
mptcp_hmac_sha1((u8 *)&mpcb->mptcp_rem_key,
(u8 *)&mpcb->mptcp_loc_key,
(u8 *)&tp->mptcp->rx_opt.mptcp_recv_nonce,
(u8 *)&tp->mptcp->mptcp_loc_nonce,
(u32 *)hash_mac_check);
if (memcmp(hash_mac_check,
(char *)&tp->mptcp->rx_opt.mptcp_recv_tmac, 8)) {
mptcp_sub_force_close(sk);
return 1;
}
/* Set this flag in order to postpone data sending
* until the 4th ack arrives.
*/
tp->mptcp->pre_established = 1;
tp->mptcp->rcv_low_prio = tp->mptcp->rx_opt.low_prio;
mptcp_hmac_sha1((u8 *)&mpcb->mptcp_loc_key,
(u8 *)&mpcb->mptcp_rem_key,
(u8 *)&tp->mptcp->mptcp_loc_nonce,
(u8 *)&tp->mptcp->rx_opt.mptcp_recv_nonce,
(u32 *)&tp->mptcp->sender_mac[0]);
} else if (mopt->saw_mpc) {
if (mptcp_create_master_sk(sk, mopt->mptcp_key,
ntohs(tcp_hdr(skb)->window)))
return 2;
sk = tcp_sk(sk)->mpcb->master_sk;
*skptr = sk;
tp = tcp_sk(sk);
/* snd_nxt - 1, because it has been incremented
* by tcp_connect for the SYN
*/
tp->mptcp->snt_isn = tp->snd_nxt - 1;
tp->mpcb->dss_csum = mopt->dss_csum;
tp->mptcp->include_mpc = 1;
/* Ensure that fastopen is handled at the meta-level. */
tp->fastopen_req = NULL;
sk_set_socket(sk, mptcp_meta_sk(sk)->sk_socket);
sk->sk_wq = mptcp_meta_sk(sk)->sk_wq;
/* hold in mptcp_inherit_sk due to initialization to 2 */
sock_put(sk);
} else {
tp->request_mptcp = 0;
if (tp->inside_tk_table)
mptcp_hash_remove(tp);
}
if (mptcp(tp))
tp->mptcp->rcv_isn = TCP_SKB_CB(skb)->seq;
return 0;
}
bool mptcp_should_expand_sndbuf(const struct sock *sk)
{
struct sock *sk_it;
struct sock *meta_sk = mptcp_meta_sk(sk);
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
int cnt_backups = 0;
int backup_available = 0;
/* We circumvent this check in tcp_check_space, because we want to
* always call sk_write_space. So, we reproduce the check here.
*/
if (!meta_sk->sk_socket ||
!test_bit(SOCK_NOSPACE, &meta_sk->sk_socket->flags))
return false;
/* If the user specified a specific send buffer setting, do
* not modify it.
*/
if (meta_sk->sk_userlocks & SOCK_SNDBUF_LOCK)
return false;
/* If we are under global TCP memory pressure, do not expand. */
if (sk_under_memory_pressure(meta_sk))
return false;
/* If we are under soft global TCP memory pressure, do not expand. */
if (sk_memory_allocated(meta_sk) >= sk_prot_mem_limits(meta_sk, 0))
return false;
/* For MPTCP we look for a subsocket that could send data.
* If we found one, then we update the send-buffer.
*/
mptcp_for_each_sk(meta_tp->mpcb, sk_it) {
struct tcp_sock *tp_it = tcp_sk(sk_it);
if (!mptcp_sk_can_send(sk_it))
continue;
/* Backup-flows have to be counted - if there is no other
* subflow we take the backup-flow into account.
*/
if (tp_it->mptcp->rcv_low_prio || tp_it->mptcp->low_prio)
cnt_backups++;
if (tp_it->packets_out < tp_it->snd_cwnd) {
if (tp_it->mptcp->rcv_low_prio || tp_it->mptcp->low_prio) {
backup_available = 1;
continue;
}
return true;
}
}
/* Backup-flow is available for sending - update send-buffer */
if (meta_tp->mpcb->cnt_established == cnt_backups && backup_available)
return true;
return false;
}
void mptcp_init_buffer_space(struct sock *sk)
{
struct tcp_sock *tp = tcp_sk(sk);
struct sock *meta_sk = mptcp_meta_sk(sk);
struct tcp_sock *meta_tp = tcp_sk(meta_sk);
int space;
tcp_init_buffer_space(sk);
if (is_master_tp(tp)) {
meta_tp->rcvq_space.space = meta_tp->rcv_wnd;
meta_tp->rcvq_space.time = tcp_time_stamp;
meta_tp->rcvq_space.seq = meta_tp->copied_seq;
/* If there is only one subflow, we just use regular TCP
* autotuning. User-locks are handled already by
* tcp_init_buffer_space
*/
meta_tp->window_clamp = tp->window_clamp;
meta_tp->rcv_ssthresh = tp->rcv_ssthresh;
meta_sk->sk_rcvbuf = sk->sk_rcvbuf;
meta_sk->sk_sndbuf = sk->sk_sndbuf;
return;
}
if (meta_sk->sk_userlocks & SOCK_RCVBUF_LOCK)
goto snd_buf;
/* Adding a new subflow to the rcv-buffer space. We make a simple
* addition, to give some space to allow traffic on the new subflow.
* Autotuning will increase it further later on.
*/
space = min(meta_sk->sk_rcvbuf + sk->sk_rcvbuf, sysctl_tcp_rmem[2]);
if (space > meta_sk->sk_rcvbuf) {
meta_tp->window_clamp += tp->window_clamp;
meta_tp->rcv_ssthresh += tp->rcv_ssthresh;
meta_sk->sk_rcvbuf = space;
}
snd_buf:
if (meta_sk->sk_userlocks & SOCK_SNDBUF_LOCK)
return;
/* Adding a new subflow to the send-buffer space. We make a simple
* addition, to give some space to allow traffic on the new subflow.
* Autotuning will increase it further later on.
*/
space = min(meta_sk->sk_sndbuf + sk->sk_sndbuf, sysctl_tcp_wmem[2]);
if (space > meta_sk->sk_sndbuf) {
meta_sk->sk_sndbuf = space;
meta_sk->sk_write_space(meta_sk);
}
}
void mptcp_tcp_set_rto(struct sock *sk)
{
tcp_set_rto(sk);
mptcp_set_rto(sk);
}
|