-
Notifications
You must be signed in to change notification settings - Fork 2
/
601-add-kernel-imq-support.patch
1706 lines (1678 loc) · 47.2 KB
/
601-add-kernel-imq-support.patch
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
diff -Naupr linux-5.4.10_orig/drivers/net/Kconfig linux-5.4.10/drivers/net/Kconfig
--- linux-5.4.10_orig/drivers/net/Kconfig 2020-01-09 11:25:53.000000000 +0200
+++ linux-5.4.10/drivers/net/Kconfig 2020-01-10 15:16:47.222143243 +0200
@@ -339,6 +339,125 @@ config RIONET_RX_SIZE
depends on RIONET
default "128"
+config IMQ
+ tristate "IMQ (intermediate queueing device) support"
+ depends on NETDEVICES && NETFILTER
+ help
+ The IMQ device(s) is used as placeholder for QoS queueing
+ disciplines. Every packet entering/leaving the IP stack can be
+ directed through the IMQ device where it's enqueued/dequeued to the
+ attached qdisc. This allows you to treat network devices as classes
+ and distribute bandwidth among them. Iptables is used to specify
+ through which IMQ device, if any, packets travel.
+
+ More information at: https://github.com/imq/linuximq
+
+ To compile this driver as a module, choose M here: the module
+ will be called imq. If unsure, say N.
+
+choice
+ prompt "IMQ behavior (PRE/POSTROUTING)"
+ depends on IMQ
+ default IMQ_BEHAVIOR_AB
+ help
+ This setting defines how IMQ behaves in respect to its
+ hooking in PREROUTING and POSTROUTING.
+
+ IMQ can work in any of the following ways:
+
+ PREROUTING | POSTROUTING
+ -----------------|-------------------
+ #1 After NAT | After NAT
+ #2 After NAT | Before NAT
+ #3 Before NAT | After NAT
+ #4 Before NAT | Before NAT
+
+ The default behavior is to hook before NAT on PREROUTING
+ and after NAT on POSTROUTING (#3).
+
+ This settings are specially usefull when trying to use IMQ
+ to shape NATed clients.
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
+config IMQ_BEHAVIOR_AA
+ bool "IMQ AA"
+ help
+ This setting defines how IMQ behaves in respect to its
+ hooking in PREROUTING and POSTROUTING.
+
+ Choosing this option will make IMQ hook like this:
+
+ PREROUTING: After NAT
+ POSTROUTING: After NAT
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
+config IMQ_BEHAVIOR_AB
+ bool "IMQ AB"
+ help
+ This setting defines how IMQ behaves in respect to its
+ hooking in PREROUTING and POSTROUTING.
+
+ Choosing this option will make IMQ hook like this:
+
+ PREROUTING: After NAT
+ POSTROUTING: Before NAT
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
+config IMQ_BEHAVIOR_BA
+ bool "IMQ BA"
+ help
+ This setting defines how IMQ behaves in respect to its
+ hooking in PREROUTING and POSTROUTING.
+
+ Choosing this option will make IMQ hook like this:
+
+ PREROUTING: Before NAT
+ POSTROUTING: After NAT
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
+config IMQ_BEHAVIOR_BB
+ bool "IMQ BB"
+ help
+ This setting defines how IMQ behaves in respect to its
+ hooking in PREROUTING and POSTROUTING.
+
+ Choosing this option will make IMQ hook like this:
+
+ PREROUTING: Before NAT
+ POSTROUTING: Before NAT
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
+endchoice
+
+config IMQ_NUM_DEVS
+ int "Number of IMQ devices"
+ range 2 16
+ depends on IMQ
+ default "16"
+ help
+ This setting defines how many IMQ devices will be created.
+
+ The default value is 16.
+
+ More information can be found at: https://github.com/imq/linuximq
+
+ If not sure leave the default settings alone.
+
config TUN
tristate "Universal TUN/TAP device driver support"
depends on INET
diff -Naupr linux-5.4.10_orig/drivers/net/Makefile linux-5.4.10/drivers/net/Makefile
--- linux-5.4.10_orig/drivers/net/Makefile 2020-01-09 11:25:53.000000000 +0200
+++ linux-5.4.10/drivers/net/Makefile 2020-01-10 15:16:47.222143243 +0200
@@ -14,6 +14,7 @@ obj-$(CONFIG_DUMMY) += dummy.o
obj-$(CONFIG_EQUALIZER) += eql.o
obj-$(CONFIG_IFB) += ifb.o
obj-$(CONFIG_MACSEC) += macsec.o
+obj-$(CONFIG_IMQ) += imq.o
obj-$(CONFIG_MACVLAN) += macvlan.o
obj-$(CONFIG_MACVTAP) += macvtap.o
obj-$(CONFIG_MII) += mii.o
diff -Naupr linux-5.4.10_orig/drivers/net/imq.c linux-5.4.10/drivers/net/imq.c
--- linux-5.4.10_orig/drivers/net/imq.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-5.4.10/drivers/net/imq.c 2020-01-10 15:21:17.815149199 +0200
@@ -0,0 +1,963 @@
+/*
+ * Pseudo-driver for the intermediate queue device.
+ *
+ * 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.
+ *
+ * Authors: Patrick McHardy, <[email protected]>
+ *
+ * The first version was written by Martin Devera, <[email protected]>
+ *
+ * See Credits.txt
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/moduleparam.h>
+#include <linux/list.h>
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/rtnetlink.h>
+#include <linux/if_arp.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter_ipv4.h>
+#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
+#include <linux/netfilter_ipv6.h>
+#endif
+#include <linux/imq.h>
+#include <net/pkt_sched.h>
+#include <net/netfilter/nf_queue.h>
+#include <net/sock.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/if_vlan.h>
+#include <linux/if_pppox.h>
+#include <net/ip.h>
+#include <net/ipv6.h>
+
+static int imq_nf_queue(struct nf_queue_entry *entry, unsigned queue_num);
+
+static nf_hookfn imq_nf_hook;
+
+static struct nf_hook_ops imq_ops[] = {
+ {
+ /* imq_ingress_ipv4 */
+ .hook = imq_nf_hook,
+ .pf = PF_INET,
+ .hooknum = NF_INET_PRE_ROUTING,
+#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB)
+ .priority = NF_IP_PRI_MANGLE + 1,
+#else
+ .priority = NF_IP_PRI_NAT_DST + 1,
+#endif
+ },
+ {
+ /* imq_egress_ipv4 */
+ .hook = imq_nf_hook,
+ .pf = PF_INET,
+ .hooknum = NF_INET_POST_ROUTING,
+#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA)
+ .priority = NF_IP_PRI_LAST,
+#else
+ .priority = NF_IP_PRI_NAT_SRC - 1,
+#endif
+ },
+#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
+ {
+ /* imq_ingress_ipv6 */
+ .hook = imq_nf_hook,
+ .pf = PF_INET6,
+ .hooknum = NF_INET_PRE_ROUTING,
+#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB)
+ .priority = NF_IP6_PRI_MANGLE + 1,
+#else
+ .priority = NF_IP6_PRI_NAT_DST + 1,
+#endif
+ },
+ {
+ /* imq_egress_ipv6 */
+ .hook = imq_nf_hook,
+ .pf = PF_INET6,
+ .hooknum = NF_INET_POST_ROUTING,
+#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA)
+ .priority = NF_IP6_PRI_LAST,
+#else
+ .priority = NF_IP6_PRI_NAT_SRC - 1,
+#endif
+ },
+#endif
+};
+
+#if defined(CONFIG_IMQ_NUM_DEVS)
+static int numdevs = CONFIG_IMQ_NUM_DEVS;
+#else
+static int numdevs = IMQ_MAX_DEVS;
+#endif
+
+static struct net_device *imq_devs_cache[IMQ_MAX_DEVS];
+
+#define IMQ_MAX_QUEUES 32
+static int numqueues = 1;
+static u32 imq_hashrnd;
+static int imq_dev_accurate_stats = 1;
+
+static inline __be16 pppoe_proto(const struct sk_buff *skb)
+{
+ return *((__be16 *)(skb_mac_header(skb) + ETH_HLEN +
+ sizeof(struct pppoe_hdr)));
+}
+
+static u16 imq_hash(struct net_device *dev, struct sk_buff *skb)
+{
+ unsigned int pull_len;
+ u16 protocol = skb->protocol;
+ u32 addr1, addr2;
+ u32 hash, ihl = 0;
+ union {
+ u16 in16[2];
+ u32 in32;
+ } ports;
+ u8 ip_proto;
+
+ pull_len = 0;
+
+recheck:
+ switch (protocol) {
+ case htons(ETH_P_8021Q): {
+ if (unlikely(skb_pull(skb, VLAN_HLEN) == NULL))
+ goto other;
+
+ pull_len += VLAN_HLEN;
+ skb->network_header += VLAN_HLEN;
+
+ protocol = vlan_eth_hdr(skb)->h_vlan_encapsulated_proto;
+ goto recheck;
+ }
+
+ case htons(ETH_P_PPP_SES): {
+ if (unlikely(skb_pull(skb, PPPOE_SES_HLEN) == NULL))
+ goto other;
+
+ pull_len += PPPOE_SES_HLEN;
+ skb->network_header += PPPOE_SES_HLEN;
+
+ protocol = pppoe_proto(skb);
+ goto recheck;
+ }
+
+ case htons(ETH_P_IP): {
+ const struct iphdr *iph = ip_hdr(skb);
+
+ if (unlikely(!pskb_may_pull(skb, sizeof(struct iphdr))))
+ goto other;
+
+ addr1 = iph->daddr;
+ addr2 = iph->saddr;
+
+ ip_proto = !(ip_hdr(skb)->frag_off & htons(IP_MF | IP_OFFSET)) ?
+ iph->protocol : 0;
+ ihl = ip_hdrlen(skb);
+
+ break;
+ }
+#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
+ case htons(ETH_P_IPV6): {
+ const struct ipv6hdr *iph = ipv6_hdr(skb);
+ __be16 fo = 0;
+
+ if (unlikely(!pskb_may_pull(skb, sizeof(struct ipv6hdr))))
+ goto other;
+
+ addr1 = iph->daddr.s6_addr32[3];
+ addr2 = iph->saddr.s6_addr32[3];
+ ihl = ipv6_skip_exthdr(skb, sizeof(struct ipv6hdr), &ip_proto,
+ &fo);
+ if (unlikely(ihl < 0))
+ goto other;
+
+ break;
+ }
+#endif
+ default:
+other:
+ if (pull_len != 0) {
+ skb_push(skb, pull_len);
+ skb->network_header -= pull_len;
+ }
+
+ return (u16)(ntohs(protocol) % dev->real_num_tx_queues);
+ }
+
+ if (addr1 > addr2)
+ swap(addr1, addr2);
+
+ switch (ip_proto) {
+ case IPPROTO_TCP:
+ case IPPROTO_UDP:
+ case IPPROTO_DCCP:
+ case IPPROTO_ESP:
+ case IPPROTO_AH:
+ case IPPROTO_SCTP:
+ case IPPROTO_UDPLITE: {
+ if (likely(skb_copy_bits(skb, ihl, &ports.in32, 4) >= 0)) {
+ if (ports.in16[0] > ports.in16[1])
+ swap(ports.in16[0], ports.in16[1]);
+ break;
+ }
+ /* fall-through */
+ }
+ default:
+ ports.in32 = 0;
+ break;
+ }
+
+ if (pull_len != 0) {
+ skb_push(skb, pull_len);
+ skb->network_header -= pull_len;
+ }
+
+ hash = jhash_3words(addr1, addr2, ports.in32, imq_hashrnd ^ ip_proto);
+
+ return (u16)(((u64)hash * dev->real_num_tx_queues) >> 32);
+}
+
+static inline bool sk_tx_queue_recorded(struct sock *sk)
+{
+ return (sk_tx_queue_get(sk) >= 0);
+}
+
+static struct netdev_queue *imq_select_queue(struct net_device *dev,
+ struct sk_buff *skb)
+{
+ u16 queue_index = 0;
+ u32 hash;
+
+ if (likely(dev->real_num_tx_queues == 1))
+ goto out;
+
+ /* IMQ can be receiving ingress or engress packets. */
+
+ /* Check first for if rx_queue is set */
+ if (skb_rx_queue_recorded(skb)) {
+ queue_index = skb_get_rx_queue(skb);
+ goto out;
+ }
+
+ /* Check if socket has tx_queue set */
+ if (sk_tx_queue_recorded(skb->sk)) {
+ queue_index = sk_tx_queue_get(skb->sk);
+ goto out;
+ }
+
+ /* Try use socket hash */
+ if (skb->sk && skb->sk->sk_hash) {
+ hash = skb->sk->sk_hash;
+ queue_index =
+ (u16)(((u64)hash * dev->real_num_tx_queues) >> 32);
+ goto out;
+ }
+
+ /* Generate hash from packet data */
+ queue_index = imq_hash(dev, skb);
+
+out:
+ if (unlikely(queue_index >= dev->real_num_tx_queues))
+ queue_index = (u16)((u32)queue_index % dev->real_num_tx_queues);
+
+ skb_set_queue_mapping(skb, queue_index);
+ return netdev_get_tx_queue(dev, queue_index);
+}
+
+static struct net_device_stats *imq_get_stats(struct net_device *dev)
+{
+ return &dev->stats;
+}
+
+/* called for packets kfree'd in qdiscs at places other than enqueue */
+static void imq_skb_destructor(struct sk_buff *skb)
+{
+ struct nf_queue_entry *entry = skb->nf_queue_entry;
+
+ skb->nf_queue_entry = NULL;
+
+ if (entry) {
+ nf_queue_entry_free(entry);
+ kfree(entry);
+ }
+
+ skb_restore_cb(skb); /* kfree backup */
+}
+
+static void imq_done_check_queue_mapping(struct sk_buff *skb,
+ struct net_device *dev)
+{
+ unsigned int queue_index;
+
+ /* Don't let queue_mapping be left too large after exiting IMQ */
+ if (likely(skb->dev != dev && skb->dev != NULL)) {
+ queue_index = skb_get_queue_mapping(skb);
+ if (unlikely(queue_index >= skb->dev->real_num_tx_queues)) {
+ queue_index = (u16)((u32)queue_index %
+ skb->dev->real_num_tx_queues);
+ skb_set_queue_mapping(skb, queue_index);
+ }
+ } else {
+ /* skb->dev was IMQ device itself or NULL, be on safe side and
+ * just clear queue mapping.
+ */
+ skb_set_queue_mapping(skb, 0);
+ }
+}
+
+static netdev_tx_t imq_dev_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+ struct nf_queue_entry *entry = skb->nf_queue_entry;
+
+ rcu_read_lock();
+
+ skb->nf_queue_entry = NULL;
+ netif_trans_update(dev);
+
+ dev->stats.tx_bytes += skb->len;
+ dev->stats.tx_packets++;
+
+ if (unlikely(entry == NULL)) {
+ /* We don't know what is going on here.. packet is queued for
+ * imq device, but (probably) not by us.
+ *
+ * If this packet was not send here by imq_nf_queue(), then
+ * skb_save_cb() was not used and skb_free() should not show:
+ * WARNING: IMQ: kfree_skb: skb->cb_next:..
+ * and/or
+ * WARNING: IMQ: kfree_skb: skb->nf_queue_entry...
+ *
+ * However if this message is shown, then IMQ is somehow broken
+ * and you should report this to linuximq.net.
+ */
+
+ /* imq_dev_xmit is black hole that eats all packets, report that
+ * we eat this packet happily and increase dropped counters.
+ */
+
+ dev->stats.tx_dropped++;
+ dev_kfree_skb(skb);
+
+ rcu_read_unlock();
+ return NETDEV_TX_OK;
+ }
+
+ skb_restore_cb(skb); /* restore skb->cb */
+
+ skb->imq_flags = 0;
+ skb->destructor = NULL;
+
+ imq_done_check_queue_mapping(skb, dev);
+
+ nf_reinject(entry, NF_ACCEPT);
+
+ rcu_read_unlock();
+ return NETDEV_TX_OK;
+}
+
+static struct net_device *get_imq_device_by_index(int index)
+{
+ struct net_device *dev = NULL;
+ struct net *net;
+ char buf[8];
+
+ /* get device by name and cache result */
+ snprintf(buf, sizeof(buf), "imq%d", index);
+
+ /* Search device from all namespaces. */
+ for_each_net(net) {
+ dev = dev_get_by_name(net, buf);
+ if (dev)
+ break;
+ }
+
+ if (WARN_ON_ONCE(dev == NULL)) {
+ /* IMQ device not found. Exotic config? */
+ return ERR_PTR(-ENODEV);
+ }
+
+ imq_devs_cache[index] = dev;
+ dev_put(dev);
+
+ return dev;
+}
+
+static struct nf_queue_entry *nf_queue_entry_dup(struct nf_queue_entry *e)
+{
+ struct nf_queue_entry *entry = kmemdup(e, e->size, GFP_ATOMIC);
+ if (entry) {
+ nf_queue_entry_get_refs(entry);
+ return entry;
+ }
+ return NULL;
+}
+
+#ifdef CONFIG_BRIDGE_NETFILTER
+/* When called from bridge netfilter, skb->data must point to MAC header
+ * before calling skb_gso_segment(). Else, original MAC header is lost
+ * and segmented skbs will be sent to wrong destination.
+ */
+static void nf_bridge_adjust_skb_data(struct sk_buff *skb)
+{
+ if (skb->nf_bridge)
+ __skb_push(skb, skb->network_header - skb->mac_header);
+}
+
+static void nf_bridge_adjust_segmented_data(struct sk_buff *skb)
+{
+ if (skb->nf_bridge)
+ __skb_pull(skb, skb->network_header - skb->mac_header);
+}
+#else
+#define nf_bridge_adjust_skb_data(s) do {} while (0)
+#define nf_bridge_adjust_segmented_data(s) do {} while (0)
+#endif
+
+static void free_entry(struct nf_queue_entry *entry)
+{
+ nf_queue_entry_free(entry);
+ kfree(entry);
+}
+
+static int __imq_nf_queue(struct nf_queue_entry *entry, struct net_device *dev);
+
+static int __imq_nf_queue_gso(struct nf_queue_entry *entry,
+ struct net_device *dev, struct sk_buff *skb)
+{
+ int ret = -ENOMEM;
+ struct nf_queue_entry *entry_seg;
+
+ nf_bridge_adjust_segmented_data(skb);
+
+ if (skb->next == NULL) { /* last packet, no need to copy entry */
+ struct sk_buff *gso_skb = entry->skb;
+ entry->skb = skb;
+ ret = __imq_nf_queue(entry, dev);
+ if (ret)
+ entry->skb = gso_skb;
+ return ret;
+ }
+
+ skb->next = NULL;
+
+ entry_seg = nf_queue_entry_dup(entry);
+ if (entry_seg) {
+ entry_seg->skb = skb;
+ ret = __imq_nf_queue(entry_seg, dev);
+ if (ret)
+ nf_queue_entry_free(entry_seg);
+ }
+ return ret;
+}
+
+static int imq_nf_queue(struct nf_queue_entry *entry, unsigned queue_num)
+{
+ struct sk_buff *skb, *segs;
+ struct net_device *dev;
+ unsigned int queued;
+ int index, retval, err;
+
+ index = entry->skb->imq_flags & IMQ_F_IFMASK;
+ if (unlikely(index > numdevs - 1)) {
+ if (net_ratelimit())
+ pr_warn("IMQ: invalid device specified, highest is %u\n",
+ numdevs - 1);
+ retval = -EINVAL;
+ goto out_no_dev;
+ }
+
+ /* check for imq device by index from cache */
+ dev = imq_devs_cache[index];
+ if (unlikely(!dev)) {
+ dev = get_imq_device_by_index(index);
+ if (IS_ERR(dev)) {
+ retval = PTR_ERR(dev);
+ goto out_no_dev;
+ }
+ }
+
+ if (unlikely(!(dev->flags & IFF_UP))) {
+ entry->skb->imq_flags = 0;
+ retval = -ECANCELED;
+ goto out_no_dev;
+ }
+
+ /* Since 3.10.x, GSO handling moved here as result of upstream commit
+ * a5fedd43d5f6c94c71053a66e4c3d2e35f1731a2 (netfilter: move
+ * skb_gso_segment into nfnetlink_queue module).
+ *
+ * Following code replicates the gso handling from
+ * 'net/netfilter/nfnetlink_queue_core.c':nfqnl_enqueue_packet().
+ */
+
+ skb = entry->skb;
+
+ switch (entry->state.pf) {
+ case NFPROTO_IPV4:
+ skb->protocol = htons(ETH_P_IP);
+ break;
+ case NFPROTO_IPV6:
+ skb->protocol = htons(ETH_P_IPV6);
+ break;
+ }
+
+ if (!skb_is_gso(entry->skb))
+ return __imq_nf_queue(entry, dev);
+
+ nf_bridge_adjust_skb_data(skb);
+ segs = skb_gso_segment(skb, 0);
+ /* Does not use PTR_ERR to limit the number of error codes that can be
+ * returned by nf_queue. For instance, callers rely on -ECANCELED to
+ * mean 'ignore this hook'.
+ */
+ err = -ENOBUFS;
+ if (IS_ERR(segs))
+ goto out_err;
+ queued = 0;
+ err = 0;
+ do {
+ struct sk_buff *nskb = segs->next;
+ if (nskb && nskb->next)
+ nskb->cb_next = NULL;
+ if (err == 0)
+ err = __imq_nf_queue_gso(entry, dev, segs);
+ if (err == 0)
+ queued++;
+ else
+ kfree_skb(segs);
+ segs = nskb;
+ } while (segs);
+
+ if (queued) {
+ if (err) /* some segments are already queued */
+ nf_queue_entry_free(entry);
+ kfree_skb(skb);
+ return 0;
+ }
+
+out_err:
+ nf_bridge_adjust_segmented_data(skb);
+ retval = err;
+out_no_dev:
+ return retval;
+}
+
+static int __imq_nf_queue(struct nf_queue_entry *entry, struct net_device *dev)
+{
+ struct sk_buff *skb_orig, *skb, *skb_shared, *skb_popd;
+ struct Qdisc *q;
+ struct sk_buff *to_free = NULL;
+ struct netdev_queue *txq;
+ spinlock_t *root_lock;
+ int users;
+ int retval = -EINVAL;
+ unsigned int orig_queue_index;
+ bool again = false;
+
+ dev->last_rx = jiffies;
+
+ skb = entry->skb;
+ skb_orig = NULL;
+
+ /* skb has owner? => make clone */
+ if (unlikely(skb->destructor)) {
+ skb_orig = skb;
+ skb = skb_clone(skb, GFP_ATOMIC);
+ if (unlikely(!skb)) {
+ retval = -ENOMEM;
+ goto out;
+ }
+ skb->cb_next = NULL;
+ entry->skb = skb;
+ }
+
+ dev->stats.rx_bytes += skb->len;
+ dev->stats.rx_packets++;
+
+ if (!skb->dev) {
+ /* skb->dev == NULL causes problems, try the find cause. */
+ if (net_ratelimit()) {
+ dev_warn(&dev->dev,
+ "received packet with skb->dev == NULL\n");
+ dump_stack();
+ }
+
+ skb->dev = dev;
+ }
+
+ /* Disables softirqs for lock below */
+ rcu_read_lock_bh();
+
+ /* Multi-queue selection */
+ orig_queue_index = skb_get_queue_mapping(skb);
+ txq = imq_select_queue(dev, skb);
+
+ q = rcu_dereference_bh(txq->qdisc);
+ if (unlikely(!q->enqueue))
+ goto packet_not_eaten_by_imq_dev;
+
+ skb->nf_queue_entry = entry;
+ root_lock = qdisc_lock(q);
+ spin_lock(root_lock);
+
+ users = refcount_read(&skb->users);
+
+ skb_shared = skb_get(skb); /* increase reference count by one */
+
+ /* backup skb->cb, as qdisc layer will overwrite it */
+ skb_save_cb(skb_shared);
+ qdisc_enqueue_root(skb_shared, q, &to_free); /* might kfree_skb */
+ if (likely(refcount_read(&skb_shared->users) == users + 1)) {
+ bool validate;
+
+ kfree_skb(skb_shared); /* decrease reference count by one */
+
+ skb->destructor = &imq_skb_destructor;
+
+ skb_popd = qdisc_dequeue_skb(q, &validate);
+
+ /* cloned? */
+ if (unlikely(skb_orig))
+ kfree_skb(skb_orig); /* free original */
+
+ spin_unlock(root_lock);
+
+#if 0
+ /* schedule qdisc dequeue */
+ __netif_schedule(q);
+#else
+ if (likely(skb_popd)) {
+ /* Note that we validate skb (GSO, checksum, ...) outside of locks */
+ if (validate)
+ skb_popd = validate_xmit_skb_list(skb_popd, dev, &again);
+
+ if (skb_popd) {
+ int dummy_ret;
+ int cpu = smp_processor_id(); /* ok because BHs are off */
+
+ txq = skb_get_tx_queue(dev, skb_popd);
+ /*
+ IMQ device will not be frozen or stoped, and it always be successful.
+ So we need not check its status and return value to accelerate.
+ */
+ if (imq_dev_accurate_stats && txq->xmit_lock_owner != cpu) {
+ HARD_TX_LOCK(dev, txq, cpu);
+ if (!netif_xmit_frozen_or_stopped(txq)) {
+ dev_hard_start_xmit(skb_popd, dev, txq, &dummy_ret);
+ }
+ HARD_TX_UNLOCK(dev, txq);
+ } else {
+ if (!netif_xmit_frozen_or_stopped(txq)) {
+ dev_hard_start_xmit(skb_popd, dev, txq, &dummy_ret);
+ }
+ }
+ }
+ } else {
+ /* No ready skb, then schedule it */
+ __netif_schedule(q);
+ }
+#endif
+ rcu_read_unlock_bh();
+ retval = 0;
+ goto out;
+ } else {
+ skb_restore_cb(skb_shared); /* restore skb->cb */
+ skb->nf_queue_entry = NULL;
+ /*
+ * qdisc dropped packet and decreased skb reference count of
+ * skb, so we don't really want to and try refree as that would
+ * actually destroy the skb.
+ */
+ spin_unlock(root_lock);
+ goto packet_not_eaten_by_imq_dev;
+ }
+
+packet_not_eaten_by_imq_dev:
+ skb_set_queue_mapping(skb, orig_queue_index);
+ rcu_read_unlock_bh();
+
+ /* cloned? restore original */
+ if (unlikely(skb_orig)) {
+ kfree_skb(skb);
+ entry->skb = skb_orig;
+ }
+ retval = -1;
+out:
+ if (unlikely(to_free)) {
+ kfree_skb_list(to_free);
+ }
+ return retval;
+}
+static unsigned int imq_nf_hook(void *priv,
+ struct sk_buff *skb,
+ const struct nf_hook_state *state)
+{
+ return (skb->imq_flags & IMQ_F_ENQUEUE) ? NF_IMQ_QUEUE : NF_ACCEPT;
+}
+
+static int imq_close(struct net_device *dev)
+{
+ netif_stop_queue(dev);
+ return 0;
+}
+
+static int imq_open(struct net_device *dev)
+{
+ netif_start_queue(dev);
+ return 0;
+}
+
+static struct device_type imq_device_type = {
+ .name = "imq",
+};
+
+static const struct net_device_ops imq_netdev_ops = {
+ .ndo_open = imq_open,
+ .ndo_stop = imq_close,
+ .ndo_start_xmit = imq_dev_xmit,
+ .ndo_get_stats = imq_get_stats,
+};
+
+static void imq_setup(struct net_device *dev)
+{
+ dev->netdev_ops = &imq_netdev_ops;
+ dev->type = ARPHRD_VOID;
+ dev->mtu = 16000; /* too small? */
+ dev->tx_queue_len = 11000; /* too big? */
+ dev->flags = IFF_NOARP;
+ dev->features = NETIF_F_SG | NETIF_F_FRAGLIST |
+ NETIF_F_GSO | NETIF_F_HW_CSUM |
+ NETIF_F_HIGHDMA;
+ dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE |
+ IFF_TX_SKB_SHARING);
+}
+
+static int imq_validate(struct nlattr *tb[], struct nlattr *data[],
+ struct netlink_ext_ack *extack)
+{
+ int ret = 0;
+
+ if (tb[IFLA_ADDRESS]) {
+ if (nla_len(tb[IFLA_ADDRESS]) != ETH_ALEN) {
+ ret = -EINVAL;
+ goto end;
+ }
+ if (!is_valid_ether_addr(nla_data(tb[IFLA_ADDRESS]))) {
+ ret = -EADDRNOTAVAIL;
+ goto end;
+ }
+ }
+ return 0;
+end:
+ pr_warn("IMQ: imq_validate failed (%d)\n", ret);
+ return ret;
+}
+
+static struct rtnl_link_ops imq_link_ops __read_mostly = {
+ .kind = "imq",
+ .priv_size = 0,
+ .setup = imq_setup,
+ .validate = imq_validate,
+};
+
+static const struct nf_queue_handler imq_nfqh = {
+ .outfn = imq_nf_queue,
+};
+
+static int __net_init imq_nf_register(struct net *net)
+{
+ return nf_register_net_hooks(net, imq_ops,
+ ARRAY_SIZE(imq_ops));
+};
+
+static void __net_exit imq_nf_unregister(struct net *net)
+{
+ nf_unregister_net_hooks(net, imq_ops,
+ ARRAY_SIZE(imq_ops));
+};
+
+static struct pernet_operations imq_net_ops = {
+ .init = imq_nf_register,
+ .exit = imq_nf_unregister,
+};
+
+static int __net_init imq_init_hooks(void)
+{
+ int ret;
+ nf_register_queue_imq_handler(&imq_nfqh);
+
+ ret = register_pernet_subsys(&imq_net_ops);
+ if (ret < 0)
+ nf_unregister_queue_imq_handler();
+
+ return ret;
+}
+
+#ifdef CONFIG_LOCKDEP
+ static struct lock_class_key imq_netdev_addr_lock_key;
+
+ static void __init imq_dev_set_lockdep_one(struct net_device *dev,
+ struct netdev_queue *txq, void *arg)
+ {
+ /*
+ * the IMQ transmit locks can be taken recursively,
+ * for example with one IMQ rule for input- and one for
+ * output network devices in iptables!
+ * until we find a better solution ignore them.
+ */
+ lockdep_set_novalidate_class(&txq->_xmit_lock);
+ }
+
+ static void imq_dev_set_lockdep_class(struct net_device *dev)
+ {
+ lockdep_set_class_and_name(&dev->addr_list_lock,
+ &imq_netdev_addr_lock_key, "_xmit_addr_IMQ");
+ netdev_for_each_tx_queue(dev, imq_dev_set_lockdep_one, NULL);
+}
+#else
+ static inline void imq_dev_set_lockdep_class(struct net_device *dev)
+ {
+ }
+#endif
+
+static int __init imq_init_one(int index)
+{
+ struct net_device *dev;
+ int ret;
+
+ dev = alloc_netdev_mq(0, "imq%d", NET_NAME_UNKNOWN, imq_setup, numqueues);
+ if (!dev)
+ return -ENOMEM;
+
+ ret = dev_alloc_name(dev, dev->name);
+ if (ret < 0)
+ goto fail;
+
+ dev->rtnl_link_ops = &imq_link_ops;
+ SET_NETDEV_DEVTYPE(dev, &imq_device_type);
+ ret = register_netdevice(dev);
+ if (ret < 0)
+ goto fail;
+
+ imq_dev_set_lockdep_class(dev);
+
+ return 0;
+fail:
+ free_netdev(dev);
+ return ret;
+}
+