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
|
/*-
* Copyright (c) 2012 Chelsio Communications, Inc.
* All rights reserved.
* Written by: Navdeep Parhar <np@FreeBSD.org>
*
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/types.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/module.h>
#include <sys/protosw.h>
#include <sys/domain.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/taskqueue.h>
#include <net/if.h>
#include <net/if_var.h>
#include <netinet/in.h>
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/ip6.h>
#include <netinet/tcp_var.h>
#include <netinet6/scope6_var.h>
#define TCPSTATES
#include <netinet/tcp_fsm.h>
#include <netinet/toecore.h>
#ifdef TCP_OFFLOAD
#include "common/common.h"
#include "common/t4_msg.h"
#include "common/t4_regs.h"
#include "common/t4_regs_values.h"
#include "common/t4_tcb.h"
#include "tom/t4_tom_l2t.h"
#include "tom/t4_tom.h"
static struct protosw ddp_protosw;
static struct pr_usrreqs ddp_usrreqs;
static struct protosw ddp6_protosw;
static struct pr_usrreqs ddp6_usrreqs;
/* Module ops */
static int t4_tom_mod_load(void);
static int t4_tom_mod_unload(void);
static int t4_tom_modevent(module_t, int, void *);
/* ULD ops and helpers */
static int t4_tom_activate(struct adapter *);
static int t4_tom_deactivate(struct adapter *);
static struct uld_info tom_uld_info = {
.uld_id = ULD_TOM,
.activate = t4_tom_activate,
.deactivate = t4_tom_deactivate,
};
static void queue_tid_release(struct adapter *, int);
static void release_offload_resources(struct toepcb *);
static int alloc_tid_tabs(struct tid_info *);
static void free_tid_tabs(struct tid_info *);
static int add_lip(struct adapter *, struct in6_addr *);
static int delete_lip(struct adapter *, struct in6_addr *);
static struct clip_entry *search_lip(struct tom_data *, struct in6_addr *);
static void init_clip_table(struct adapter *, struct tom_data *);
static void update_clip(struct adapter *, void *);
static void t4_clip_task(void *, int);
static void update_clip_table(struct adapter *, struct tom_data *);
static void destroy_clip_table(struct adapter *, struct tom_data *);
static void free_tom_data(struct adapter *, struct tom_data *);
static int in6_ifaddr_gen;
static eventhandler_tag ifaddr_evhandler;
static struct timeout_task clip_task;
struct toepcb *
alloc_toepcb(struct port_info *pi, int txqid, int rxqid, int flags)
{
struct adapter *sc = pi->adapter;
struct toepcb *toep;
int tx_credits, txsd_total, len;
/*
* The firmware counts tx work request credits in units of 16 bytes
* each. Reserve room for an ABORT_REQ so the driver never has to worry
* about tx credits if it wants to abort a connection.
*/
tx_credits = sc->params.ofldq_wr_cred;
tx_credits -= howmany(sizeof(struct cpl_abort_req), 16);
/*
* Shortest possible tx work request is a fw_ofld_tx_data_wr + 1 byte
* immediate payload, and firmware counts tx work request credits in
* units of 16 byte. Calculate the maximum work requests possible.
*/
txsd_total = tx_credits /
howmany((sizeof(struct fw_ofld_tx_data_wr) + 1), 16);
if (txqid < 0)
txqid = (arc4random() % pi->nofldtxq) + pi->first_ofld_txq;
KASSERT(txqid >= pi->first_ofld_txq &&
txqid < pi->first_ofld_txq + pi->nofldtxq,
("%s: txqid %d for port %p (first %d, n %d)", __func__, txqid, pi,
pi->first_ofld_txq, pi->nofldtxq));
if (rxqid < 0)
rxqid = (arc4random() % pi->nofldrxq) + pi->first_ofld_rxq;
KASSERT(rxqid >= pi->first_ofld_rxq &&
rxqid < pi->first_ofld_rxq + pi->nofldrxq,
("%s: rxqid %d for port %p (first %d, n %d)", __func__, rxqid, pi,
pi->first_ofld_rxq, pi->nofldrxq));
len = offsetof(struct toepcb, txsd) +
txsd_total * sizeof(struct ofld_tx_sdesc);
toep = malloc(len, M_CXGBE, M_ZERO | flags);
if (toep == NULL)
return (NULL);
toep->td = sc->tom_softc;
toep->port = pi;
toep->tx_total = tx_credits;
toep->tx_credits = tx_credits;
toep->ofld_txq = &sc->sge.ofld_txq[txqid];
toep->ofld_rxq = &sc->sge.ofld_rxq[rxqid];
toep->ctrlq = &sc->sge.ctrlq[pi->port_id];
toep->txsd_total = txsd_total;
toep->txsd_avail = txsd_total;
toep->txsd_pidx = 0;
toep->txsd_cidx = 0;
return (toep);
}
void
free_toepcb(struct toepcb *toep)
{
KASSERT(!(toep->flags & TPF_ATTACHED),
("%s: attached to an inpcb", __func__));
KASSERT(!(toep->flags & TPF_CPL_PENDING),
("%s: CPL pending", __func__));
free(toep, M_CXGBE);
}
/*
* Set up the socket for TCP offload.
*/
void
offload_socket(struct socket *so, struct toepcb *toep)
{
struct tom_data *td = toep->td;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp = intotcpcb(inp);
struct sockbuf *sb;
INP_WLOCK_ASSERT(inp);
/* Update socket */
sb = &so->so_snd;
SOCKBUF_LOCK(sb);
sb->sb_flags |= SB_NOCOALESCE;
SOCKBUF_UNLOCK(sb);
sb = &so->so_rcv;
SOCKBUF_LOCK(sb);
sb->sb_flags |= SB_NOCOALESCE;
if (toep->ulp_mode == ULP_MODE_TCPDDP) {
if (inp->inp_vflag & INP_IPV6)
so->so_proto = &ddp6_protosw;
else
so->so_proto = &ddp_protosw;
}
SOCKBUF_UNLOCK(sb);
/* Update TCP PCB */
tp->tod = &td->tod;
tp->t_toe = toep;
tp->t_flags |= TF_TOE;
/* Install an extra hold on inp */
toep->inp = inp;
toep->flags |= TPF_ATTACHED;
in_pcbref(inp);
/* Add the TOE PCB to the active list */
mtx_lock(&td->toep_list_lock);
TAILQ_INSERT_HEAD(&td->toep_list, toep, link);
mtx_unlock(&td->toep_list_lock);
}
/* This is _not_ the normal way to "unoffload" a socket. */
void
undo_offload_socket(struct socket *so)
{
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp = intotcpcb(inp);
struct toepcb *toep = tp->t_toe;
struct tom_data *td = toep->td;
struct sockbuf *sb;
INP_WLOCK_ASSERT(inp);
sb = &so->so_snd;
SOCKBUF_LOCK(sb);
sb->sb_flags &= ~SB_NOCOALESCE;
SOCKBUF_UNLOCK(sb);
sb = &so->so_rcv;
SOCKBUF_LOCK(sb);
sb->sb_flags &= ~SB_NOCOALESCE;
SOCKBUF_UNLOCK(sb);
tp->tod = NULL;
tp->t_toe = NULL;
tp->t_flags &= ~TF_TOE;
toep->inp = NULL;
toep->flags &= ~TPF_ATTACHED;
if (in_pcbrele_wlocked(inp))
panic("%s: inp freed.", __func__);
mtx_lock(&td->toep_list_lock);
TAILQ_REMOVE(&td->toep_list, toep, link);
mtx_unlock(&td->toep_list_lock);
}
static void
release_offload_resources(struct toepcb *toep)
{
struct tom_data *td = toep->td;
struct adapter *sc = td_adapter(td);
int tid = toep->tid;
KASSERT(!(toep->flags & TPF_CPL_PENDING),
("%s: %p has CPL pending.", __func__, toep));
KASSERT(!(toep->flags & TPF_ATTACHED),
("%s: %p is still attached.", __func__, toep));
CTR5(KTR_CXGBE, "%s: toep %p (tid %d, l2te %p, ce %p)",
__func__, toep, tid, toep->l2te, toep->ce);
if (toep->ulp_mode == ULP_MODE_TCPDDP)
release_ddp_resources(toep);
if (toep->l2te)
t4_l2t_release(toep->l2te);
if (tid >= 0) {
remove_tid(sc, tid);
release_tid(sc, tid, toep->ctrlq);
}
if (toep->ce)
release_lip(td, toep->ce);
mtx_lock(&td->toep_list_lock);
TAILQ_REMOVE(&td->toep_list, toep, link);
mtx_unlock(&td->toep_list_lock);
free_toepcb(toep);
}
/*
* The kernel is done with the TCP PCB and this is our opportunity to unhook the
* toepcb hanging off of it. If the TOE driver is also done with the toepcb (no
* pending CPL) then it is time to release all resources tied to the toepcb.
*
* Also gets called when an offloaded active open fails and the TOM wants the
* kernel to take the TCP PCB back.
*/
static void
t4_pcb_detach(struct toedev *tod __unused, struct tcpcb *tp)
{
#if defined(KTR) || defined(INVARIANTS)
struct inpcb *inp = tp->t_inpcb;
#endif
struct toepcb *toep = tp->t_toe;
INP_WLOCK_ASSERT(inp);
KASSERT(toep != NULL, ("%s: toep is NULL", __func__));
KASSERT(toep->flags & TPF_ATTACHED,
("%s: not attached", __func__));
#ifdef KTR
if (tp->t_state == TCPS_SYN_SENT) {
CTR6(KTR_CXGBE, "%s: atid %d, toep %p (0x%x), inp %p (0x%x)",
__func__, toep->tid, toep, toep->flags, inp,
inp->inp_flags);
} else {
CTR6(KTR_CXGBE,
"t4_pcb_detach: tid %d (%s), toep %p (0x%x), inp %p (0x%x)",
toep->tid, tcpstates[tp->t_state], toep, toep->flags, inp,
inp->inp_flags);
}
#endif
tp->t_toe = NULL;
tp->t_flags &= ~TF_TOE;
toep->flags &= ~TPF_ATTACHED;
if (!(toep->flags & TPF_CPL_PENDING))
release_offload_resources(toep);
}
/*
* setsockopt handler.
*/
static void
t4_ctloutput(struct toedev *tod, struct tcpcb *tp, int dir, int name)
{
struct adapter *sc = tod->tod_softc;
struct toepcb *toep = tp->t_toe;
if (dir == SOPT_GET)
return;
CTR4(KTR_CXGBE, "%s: tp %p, dir %u, name %u", __func__, tp, dir, name);
switch (name) {
case TCP_NODELAY:
t4_set_tcb_field(sc, toep, 1, W_TCB_T_FLAGS, V_TF_NAGLE(1),
V_TF_NAGLE(tp->t_flags & TF_NODELAY ? 0 : 1));
break;
default:
break;
}
}
/*
* The TOE driver will not receive any more CPLs for the tid associated with the
* toepcb; release the hold on the inpcb.
*/
void
final_cpl_received(struct toepcb *toep)
{
struct inpcb *inp = toep->inp;
KASSERT(inp != NULL, ("%s: inp is NULL", __func__));
INP_WLOCK_ASSERT(inp);
KASSERT(toep->flags & TPF_CPL_PENDING,
("%s: CPL not pending already?", __func__));
CTR6(KTR_CXGBE, "%s: tid %d, toep %p (0x%x), inp %p (0x%x)",
__func__, toep->tid, toep, toep->flags, inp, inp->inp_flags);
toep->inp = NULL;
toep->flags &= ~TPF_CPL_PENDING;
if (!(toep->flags & TPF_ATTACHED))
release_offload_resources(toep);
if (!in_pcbrele_wlocked(inp))
INP_WUNLOCK(inp);
}
void
insert_tid(struct adapter *sc, int tid, void *ctx)
{
struct tid_info *t = &sc->tids;
t->tid_tab[tid] = ctx;
atomic_add_int(&t->tids_in_use, 1);
}
void *
lookup_tid(struct adapter *sc, int tid)
{
struct tid_info *t = &sc->tids;
return (t->tid_tab[tid]);
}
void
update_tid(struct adapter *sc, int tid, void *ctx)
{
struct tid_info *t = &sc->tids;
t->tid_tab[tid] = ctx;
}
void
remove_tid(struct adapter *sc, int tid)
{
struct tid_info *t = &sc->tids;
t->tid_tab[tid] = NULL;
atomic_subtract_int(&t->tids_in_use, 1);
}
void
release_tid(struct adapter *sc, int tid, struct sge_wrq *ctrlq)
{
struct wrqe *wr;
struct cpl_tid_release *req;
wr = alloc_wrqe(sizeof(*req), ctrlq);
if (wr == NULL) {
queue_tid_release(sc, tid); /* defer */
return;
}
req = wrtod(wr);
INIT_TP_WR_MIT_CPL(req, CPL_TID_RELEASE, tid);
t4_wrq_tx(sc, wr);
}
static void
queue_tid_release(struct adapter *sc, int tid)
{
CXGBE_UNIMPLEMENTED("deferred tid release");
}
/*
* What mtu_idx to use, given a 4-tuple and/or an MSS cap
*/
int
find_best_mtu_idx(struct adapter *sc, struct in_conninfo *inc, int pmss)
{
unsigned short *mtus = &sc->params.mtus[0];
int i, mss, n;
KASSERT(inc != NULL || pmss > 0,
("%s: at least one of inc/pmss must be specified", __func__));
mss = inc ? tcp_mssopt(inc) : pmss;
if (pmss > 0 && mss > pmss)
mss = pmss;
if (inc->inc_flags & INC_ISIPV6)
n = sizeof(struct ip6_hdr) + sizeof(struct tcphdr);
else
n = sizeof(struct ip) + sizeof(struct tcphdr);
for (i = 0; i < NMTUS - 1 && mtus[i + 1] <= mss + n; i++)
continue;
return (i);
}
/*
* Determine the receive window size for a socket.
*/
u_long
select_rcv_wnd(struct socket *so)
{
unsigned long wnd;
SOCKBUF_LOCK_ASSERT(&so->so_rcv);
wnd = sbspace(&so->so_rcv);
if (wnd < MIN_RCV_WND)
wnd = MIN_RCV_WND;
return min(wnd, MAX_RCV_WND);
}
int
select_rcv_wscale(void)
{
int wscale = 0;
unsigned long space = sb_max;
if (space > MAX_RCV_WND)
space = MAX_RCV_WND;
while (wscale < TCP_MAX_WINSHIFT && (TCP_MAXWIN << wscale) < space)
wscale++;
return (wscale);
}
extern int always_keepalive;
#define VIID_SMACIDX(v) (((unsigned int)(v) & 0x7f) << 1)
/*
* socket so could be a listening socket too.
*/
uint64_t
calc_opt0(struct socket *so, struct port_info *pi, struct l2t_entry *e,
int mtu_idx, int rscale, int rx_credits, int ulp_mode)
{
uint64_t opt0;
KASSERT(rx_credits <= M_RCV_BUFSIZ,
("%s: rcv_bufsiz too high", __func__));
opt0 = F_TCAM_BYPASS | V_WND_SCALE(rscale) | V_MSS_IDX(mtu_idx) |
V_ULP_MODE(ulp_mode) | V_RCV_BUFSIZ(rx_credits);
if (so != NULL) {
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp = intotcpcb(inp);
int keepalive = always_keepalive ||
so_options_get(so) & SO_KEEPALIVE;
opt0 |= V_NAGLE((tp->t_flags & TF_NODELAY) == 0);
opt0 |= V_KEEP_ALIVE(keepalive != 0);
}
if (e != NULL)
opt0 |= V_L2T_IDX(e->idx);
if (pi != NULL) {
opt0 |= V_SMAC_SEL(VIID_SMACIDX(pi->viid));
opt0 |= V_TX_CHAN(pi->tx_chan);
}
return htobe64(opt0);
}
uint64_t
select_ntuple(struct port_info *pi, struct l2t_entry *e)
{
struct adapter *sc = pi->adapter;
struct tp_params *tp = &sc->params.tp;
uint16_t viid = pi->viid;
uint64_t ntuple = 0;
/*
* Initialize each of the fields which we care about which are present
* in the Compressed Filter Tuple.
*/
if (tp->vlan_shift >= 0 && e->vlan != CPL_L2T_VLAN_NONE)
ntuple |= (uint64_t)(F_FT_VLAN_VLD | e->vlan) << tp->vlan_shift;
if (tp->port_shift >= 0)
ntuple |= (uint64_t)e->lport << tp->port_shift;
if (tp->protocol_shift >= 0)
ntuple |= (uint64_t)IPPROTO_TCP << tp->protocol_shift;
if (tp->vnic_shift >= 0) {
uint32_t vf = G_FW_VIID_VIN(viid);
uint32_t pf = G_FW_VIID_PFN(viid);
uint32_t vld = G_FW_VIID_VIVLD(viid);
ntuple |= (uint64_t)(V_FT_VNID_ID_VF(vf) | V_FT_VNID_ID_PF(pf) |
V_FT_VNID_ID_VLD(vld)) << tp->vnic_shift;
}
if (is_t4(sc))
return (htobe32((uint32_t)ntuple));
else
return (htobe64(V_FILTER_TUPLE(ntuple)));
}
void
set_tcpddp_ulp_mode(struct toepcb *toep)
{
toep->ulp_mode = ULP_MODE_TCPDDP;
toep->ddp_flags = DDP_OK;
toep->ddp_score = DDP_LOW_SCORE;
}
int
negative_advice(int status)
{
return (status == CPL_ERR_RTX_NEG_ADVICE ||
status == CPL_ERR_PERSIST_NEG_ADVICE ||
status == CPL_ERR_KEEPALV_NEG_ADVICE);
}
static int
alloc_tid_tabs(struct tid_info *t)
{
size_t size;
unsigned int i;
size = t->ntids * sizeof(*t->tid_tab) +
t->natids * sizeof(*t->atid_tab) +
t->nstids * sizeof(*t->stid_tab);
t->tid_tab = malloc(size, M_CXGBE, M_ZERO | M_NOWAIT);
if (t->tid_tab == NULL)
return (ENOMEM);
mtx_init(&t->atid_lock, "atid lock", NULL, MTX_DEF);
t->atid_tab = (union aopen_entry *)&t->tid_tab[t->ntids];
t->afree = t->atid_tab;
t->atids_in_use = 0;
for (i = 1; i < t->natids; i++)
t->atid_tab[i - 1].next = &t->atid_tab[i];
t->atid_tab[t->natids - 1].next = NULL;
mtx_init(&t->stid_lock, "stid lock", NULL, MTX_DEF);
t->stid_tab = (struct listen_ctx **)&t->atid_tab[t->natids];
t->stids_in_use = 0;
TAILQ_INIT(&t->stids);
t->nstids_free_head = t->nstids;
atomic_store_rel_int(&t->tids_in_use, 0);
return (0);
}
static void
free_tid_tabs(struct tid_info *t)
{
KASSERT(t->tids_in_use == 0,
("%s: %d tids still in use.", __func__, t->tids_in_use));
KASSERT(t->atids_in_use == 0,
("%s: %d atids still in use.", __func__, t->atids_in_use));
KASSERT(t->stids_in_use == 0,
("%s: %d tids still in use.", __func__, t->stids_in_use));
free(t->tid_tab, M_CXGBE);
t->tid_tab = NULL;
if (mtx_initialized(&t->atid_lock))
mtx_destroy(&t->atid_lock);
if (mtx_initialized(&t->stid_lock))
mtx_destroy(&t->stid_lock);
}
static int
add_lip(struct adapter *sc, struct in6_addr *lip)
{
struct fw_clip_cmd c;
ASSERT_SYNCHRONIZED_OP(sc);
/* mtx_assert(&td->clip_table_lock, MA_OWNED); */
memset(&c, 0, sizeof(c));
c.op_to_write = htonl(V_FW_CMD_OP(FW_CLIP_CMD) | F_FW_CMD_REQUEST |
F_FW_CMD_WRITE);
c.alloc_to_len16 = htonl(F_FW_CLIP_CMD_ALLOC | FW_LEN16(c));
c.ip_hi = *(uint64_t *)&lip->s6_addr[0];
c.ip_lo = *(uint64_t *)&lip->s6_addr[8];
return (-t4_wr_mbox_ns(sc, sc->mbox, &c, sizeof(c), &c));
}
static int
delete_lip(struct adapter *sc, struct in6_addr *lip)
{
struct fw_clip_cmd c;
ASSERT_SYNCHRONIZED_OP(sc);
/* mtx_assert(&td->clip_table_lock, MA_OWNED); */
memset(&c, 0, sizeof(c));
c.op_to_write = htonl(V_FW_CMD_OP(FW_CLIP_CMD) | F_FW_CMD_REQUEST |
F_FW_CMD_READ);
c.alloc_to_len16 = htonl(F_FW_CLIP_CMD_FREE | FW_LEN16(c));
c.ip_hi = *(uint64_t *)&lip->s6_addr[0];
c.ip_lo = *(uint64_t *)&lip->s6_addr[8];
return (-t4_wr_mbox_ns(sc, sc->mbox, &c, sizeof(c), &c));
}
static struct clip_entry *
search_lip(struct tom_data *td, struct in6_addr *lip)
{
struct clip_entry *ce;
mtx_assert(&td->clip_table_lock, MA_OWNED);
TAILQ_FOREACH(ce, &td->clip_table, link) {
if (IN6_ARE_ADDR_EQUAL(&ce->lip, lip))
return (ce);
}
return (NULL);
}
struct clip_entry *
hold_lip(struct tom_data *td, struct in6_addr *lip)
{
struct clip_entry *ce;
mtx_lock(&td->clip_table_lock);
ce = search_lip(td, lip);
if (ce != NULL)
ce->refcount++;
mtx_unlock(&td->clip_table_lock);
return (ce);
}
void
release_lip(struct tom_data *td, struct clip_entry *ce)
{
mtx_lock(&td->clip_table_lock);
KASSERT(search_lip(td, &ce->lip) == ce,
("%s: CLIP entry %p p not in CLIP table.", __func__, ce));
KASSERT(ce->refcount > 0,
("%s: CLIP entry %p has refcount 0", __func__, ce));
--ce->refcount;
mtx_unlock(&td->clip_table_lock);
}
static void
init_clip_table(struct adapter *sc, struct tom_data *td)
{
ASSERT_SYNCHRONIZED_OP(sc);
mtx_init(&td->clip_table_lock, "CLIP table lock", NULL, MTX_DEF);
TAILQ_INIT(&td->clip_table);
td->clip_gen = -1;
update_clip_table(sc, td);
}
static void
update_clip(struct adapter *sc, void *arg __unused)
{
if (begin_synchronized_op(sc, NULL, HOLD_LOCK, "t4tomuc"))
return;
if (sc->flags & TOM_INIT_DONE)
update_clip_table(sc, sc->tom_softc);
end_synchronized_op(sc, LOCK_HELD);
}
static void
t4_clip_task(void *arg, int count)
{
t4_iterate(update_clip, NULL);
}
static void
update_clip_table(struct adapter *sc, struct tom_data *td)
{
struct in6_ifaddr *ia;
struct in6_addr *lip, tlip;
struct clip_head stale;
struct clip_entry *ce, *ce_temp;
int rc, gen = atomic_load_acq_int(&in6_ifaddr_gen);
ASSERT_SYNCHRONIZED_OP(sc);
IN6_IFADDR_RLOCK();
mtx_lock(&td->clip_table_lock);
if (gen == td->clip_gen)
goto done;
TAILQ_INIT(&stale);
TAILQ_CONCAT(&stale, &td->clip_table, link);
TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
lip = &ia->ia_addr.sin6_addr;
KASSERT(!IN6_IS_ADDR_MULTICAST(lip),
("%s: mcast address in in6_ifaddr list", __func__));
if (IN6_IS_ADDR_LOOPBACK(lip))
continue;
if (IN6_IS_SCOPE_EMBED(lip)) {
/* Remove the embedded scope */
tlip = *lip;
lip = &tlip;
in6_clearscope(lip);
}
/*
* XXX: how to weed out the link local address for the loopback
* interface? It's fe80::1 usually (always?).
*/
/*
* If it's in the main list then we already know it's not stale.
*/
TAILQ_FOREACH(ce, &td->clip_table, link) {
if (IN6_ARE_ADDR_EQUAL(&ce->lip, lip))
goto next;
}
/*
* If it's in the stale list we should move it to the main list.
*/
TAILQ_FOREACH(ce, &stale, link) {
if (IN6_ARE_ADDR_EQUAL(&ce->lip, lip)) {
TAILQ_REMOVE(&stale, ce, link);
TAILQ_INSERT_TAIL(&td->clip_table, ce, link);
goto next;
}
}
/* A new IP6 address; add it to the CLIP table */
ce = malloc(sizeof(*ce), M_CXGBE, M_NOWAIT);
memcpy(&ce->lip, lip, sizeof(ce->lip));
ce->refcount = 0;
rc = add_lip(sc, lip);
if (rc == 0)
TAILQ_INSERT_TAIL(&td->clip_table, ce, link);
else {
char ip[INET6_ADDRSTRLEN];
inet_ntop(AF_INET6, &ce->lip, &ip[0], sizeof(ip));
log(LOG_ERR, "%s: could not add %s (%d)\n",
__func__, ip, rc);
free(ce, M_CXGBE);
}
next:
continue;
}
/*
* Remove stale addresses (those no longer in V_in6_ifaddrhead) that are
* no longer referenced by the driver.
*/
TAILQ_FOREACH_SAFE(ce, &stale, link, ce_temp) {
if (ce->refcount == 0) {
rc = delete_lip(sc, &ce->lip);
if (rc == 0) {
TAILQ_REMOVE(&stale, ce, link);
free(ce, M_CXGBE);
} else {
char ip[INET6_ADDRSTRLEN];
inet_ntop(AF_INET6, &ce->lip, &ip[0],
sizeof(ip));
log(LOG_ERR, "%s: could not delete %s (%d)\n",
__func__, ip, rc);
}
}
}
/* The ones that are still referenced need to stay in the CLIP table */
TAILQ_CONCAT(&td->clip_table, &stale, link);
td->clip_gen = gen;
done:
mtx_unlock(&td->clip_table_lock);
IN6_IFADDR_RUNLOCK();
}
static void
destroy_clip_table(struct adapter *sc, struct tom_data *td)
{
struct clip_entry *ce, *ce_temp;
if (mtx_initialized(&td->clip_table_lock)) {
mtx_lock(&td->clip_table_lock);
TAILQ_FOREACH_SAFE(ce, &td->clip_table, link, ce_temp) {
KASSERT(ce->refcount == 0,
("%s: CLIP entry %p still in use (%d)", __func__,
ce, ce->refcount));
TAILQ_REMOVE(&td->clip_table, ce, link);
delete_lip(sc, &ce->lip);
free(ce, M_CXGBE);
}
mtx_unlock(&td->clip_table_lock);
mtx_destroy(&td->clip_table_lock);
}
}
static void
free_tom_data(struct adapter *sc, struct tom_data *td)
{
ASSERT_SYNCHRONIZED_OP(sc);
KASSERT(TAILQ_EMPTY(&td->toep_list),
("%s: TOE PCB list is not empty.", __func__));
KASSERT(td->lctx_count == 0,
("%s: lctx hash table is not empty.", __func__));
t4_uninit_l2t_cpl_handlers(sc);
t4_uninit_cpl_io_handlers(sc);
t4_uninit_ddp(sc, td);
destroy_clip_table(sc, td);
if (td->listen_mask != 0)
hashdestroy(td->listen_hash, M_CXGBE, td->listen_mask);
if (mtx_initialized(&td->lctx_hash_lock))
mtx_destroy(&td->lctx_hash_lock);
if (mtx_initialized(&td->toep_list_lock))
mtx_destroy(&td->toep_list_lock);
free_tid_tabs(&sc->tids);
free(td, M_CXGBE);
}
/*
* Ground control to Major TOM
* Commencing countdown, engines on
*/
static int
t4_tom_activate(struct adapter *sc)
{
struct tom_data *td;
struct toedev *tod;
int i, rc;
ASSERT_SYNCHRONIZED_OP(sc);
/* per-adapter softc for TOM */
td = malloc(sizeof(*td), M_CXGBE, M_ZERO | M_NOWAIT);
if (td == NULL)
return (ENOMEM);
/* List of TOE PCBs and associated lock */
mtx_init(&td->toep_list_lock, "PCB list lock", NULL, MTX_DEF);
TAILQ_INIT(&td->toep_list);
/* Listen context */
mtx_init(&td->lctx_hash_lock, "lctx hash lock", NULL, MTX_DEF);
td->listen_hash = hashinit_flags(LISTEN_HASH_SIZE, M_CXGBE,
&td->listen_mask, HASH_NOWAIT);
/* TID tables */
rc = alloc_tid_tabs(&sc->tids);
if (rc != 0)
goto done;
/* DDP page pods and CPL handlers */
t4_init_ddp(sc, td);
/* CLIP table for IPv6 offload */
init_clip_table(sc, td);
/* CPL handlers */
t4_init_connect_cpl_handlers(sc);
t4_init_l2t_cpl_handlers(sc);
t4_init_listen_cpl_handlers(sc);
t4_init_cpl_io_handlers(sc);
/* toedev ops */
tod = &td->tod;
init_toedev(tod);
tod->tod_softc = sc;
tod->tod_connect = t4_connect;
tod->tod_listen_start = t4_listen_start;
tod->tod_listen_stop = t4_listen_stop;
tod->tod_rcvd = t4_rcvd;
tod->tod_output = t4_tod_output;
tod->tod_send_rst = t4_send_rst;
tod->tod_send_fin = t4_send_fin;
tod->tod_pcb_detach = t4_pcb_detach;
tod->tod_l2_update = t4_l2_update;
tod->tod_syncache_added = t4_syncache_added;
tod->tod_syncache_removed = t4_syncache_removed;
tod->tod_syncache_respond = t4_syncache_respond;
tod->tod_offload_socket = t4_offload_socket;
tod->tod_ctloutput = t4_ctloutput;
for_each_port(sc, i)
TOEDEV(sc->port[i]->ifp) = &td->tod;
sc->tom_softc = td;
sc->flags |= TOM_INIT_DONE;
register_toedev(sc->tom_softc);
done:
if (rc != 0)
free_tom_data(sc, td);
return (rc);
}
static int
t4_tom_deactivate(struct adapter *sc)
{
int rc = 0;
struct tom_data *td = sc->tom_softc;
ASSERT_SYNCHRONIZED_OP(sc);
if (td == NULL)
return (0); /* XXX. KASSERT? */
if (sc->offload_map != 0)
return (EBUSY); /* at least one port has IFCAP_TOE enabled */
mtx_lock(&td->toep_list_lock);
if (!TAILQ_EMPTY(&td->toep_list))
rc = EBUSY;
mtx_unlock(&td->toep_list_lock);
mtx_lock(&td->lctx_hash_lock);
if (td->lctx_count > 0)
rc = EBUSY;
mtx_unlock(&td->lctx_hash_lock);
if (rc == 0) {
unregister_toedev(sc->tom_softc);
free_tom_data(sc, td);
sc->tom_softc = NULL;
sc->flags &= ~TOM_INIT_DONE;
}
return (rc);
}
static void
t4_tom_ifaddr_event(void *arg __unused, struct ifnet *ifp)
{
atomic_add_rel_int(&in6_ifaddr_gen, 1);
taskqueue_enqueue_timeout(taskqueue_thread, &clip_task, -hz / 4);
}
static int
t4_tom_mod_load(void)
{
int rc;
struct protosw *tcp_protosw, *tcp6_protosw;
tcp_protosw = pffindproto(PF_INET, IPPROTO_TCP, SOCK_STREAM);
if (tcp_protosw == NULL)
return (ENOPROTOOPT);
bcopy(tcp_protosw, &ddp_protosw, sizeof(ddp_protosw));
bcopy(tcp_protosw->pr_usrreqs, &ddp_usrreqs, sizeof(ddp_usrreqs));
ddp_usrreqs.pru_soreceive = t4_soreceive_ddp;
ddp_protosw.pr_usrreqs = &ddp_usrreqs;
tcp6_protosw = pffindproto(PF_INET6, IPPROTO_TCP, SOCK_STREAM);
if (tcp6_protosw == NULL)
return (ENOPROTOOPT);
bcopy(tcp6_protosw, &ddp6_protosw, sizeof(ddp6_protosw));
bcopy(tcp6_protosw->pr_usrreqs, &ddp6_usrreqs, sizeof(ddp6_usrreqs));
ddp6_usrreqs.pru_soreceive = t4_soreceive_ddp;
ddp6_protosw.pr_usrreqs = &ddp6_usrreqs;
TIMEOUT_TASK_INIT(taskqueue_thread, &clip_task, 0, t4_clip_task, NULL);
ifaddr_evhandler = EVENTHANDLER_REGISTER(ifaddr_event,
t4_tom_ifaddr_event, NULL, EVENTHANDLER_PRI_ANY);
rc = t4_register_uld(&tom_uld_info);
if (rc != 0)
t4_tom_mod_unload();
return (rc);
}
static void
tom_uninit(struct adapter *sc, void *arg __unused)
{
if (begin_synchronized_op(sc, NULL, SLEEP_OK | INTR_OK, "t4tomun"))
return;
/* Try to free resources (works only if no port has IFCAP_TOE) */
if (sc->flags & TOM_INIT_DONE)
t4_deactivate_uld(sc, ULD_TOM);
end_synchronized_op(sc, 0);
}
static int
t4_tom_mod_unload(void)
{
t4_iterate(tom_uninit, NULL);
if (t4_unregister_uld(&tom_uld_info) == EBUSY)
return (EBUSY);
if (ifaddr_evhandler) {
EVENTHANDLER_DEREGISTER(ifaddr_event, ifaddr_evhandler);
taskqueue_cancel_timeout(taskqueue_thread, &clip_task, NULL);
}
return (0);
}
#endif /* TCP_OFFLOAD */
static int
t4_tom_modevent(module_t mod, int cmd, void *arg)
{
int rc = 0;
#ifdef TCP_OFFLOAD
switch (cmd) {
case MOD_LOAD:
rc = t4_tom_mod_load();
break;
case MOD_UNLOAD:
rc = t4_tom_mod_unload();
break;
default:
rc = EINVAL;
}
#else
printf("t4_tom: compiled without TCP_OFFLOAD support.\n");
rc = EOPNOTSUPP;
#endif
return (rc);
}
static moduledata_t t4_tom_moddata= {
"t4_tom",
t4_tom_modevent,
0
};
MODULE_VERSION(t4_tom, 1);
MODULE_DEPEND(t4_tom, toecore, 1, 1, 1);
MODULE_DEPEND(t4_tom, t4nex, 1, 1, 1);
DECLARE_MODULE(t4_tom, t4_tom_moddata, SI_SUB_EXEC, SI_ORDER_ANY);
|