change system file

hc

2024-10-09 244b2c5ca8b14627e4a17755e5922221e121c771

 kernel/net/ipv4/ip_tunnel.c
..
..
@@ -1,19 +1,6 @@
1
+// SPDX-License-Identifier: GPL-2.0-only
1
2
 /*
2
3
  * Copyright (c) 2013 Nicira, Inc.
3
- *
4
- * This program is free software; you can redistribute it and/or
5
- * modify it under the terms of version 2 of the GNU General Public
6
- * License as published by the Free Software Foundation.
7
- *
8
- * This program is distributed in the hope that it will be useful, but
9
- * WITHOUT ANY WARRANTY; without even the implied warranty of
10
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11
- * General Public License for more details.
12
- *
13
- * You should have received a copy of the GNU General Public License
14
- * along with this program; if not, write to the Free Software
15
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
16
- * 02110-1301, USA
17
4
  */
18
5
 
19
6
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
..
..
@@ -308,7 +295,7 @@
308
295
 		ip_tunnel_init_flow(&fl4, iph->protocol, iph->daddr,
309
296
 				    iph->saddr, tunnel->parms.o_key,
310
297
 				    RT_TOS(iph->tos), tunnel->parms.link,
311
-				    tunnel->fwmark);
298
+				    tunnel->fwmark, 0);
312
299
 		rt = ip_route_output_key(tunnel->net, &fl4);
313
300
 
314
301
 		if (!IS_ERR(rt)) {
..
..
@@ -376,7 +363,6 @@
376
363
 		  const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst,
377
364
 		  bool log_ecn_error)
378
365
 {
379
-	struct pcpu_sw_netstats *tstats;
380
366
 	const struct iphdr *iph = ip_hdr(skb);
381
367
 	int err;
382
368
 
..
..
@@ -404,7 +390,7 @@
404
390
 		tunnel->i_seqno = ntohl(tpi->seq) + 1;
405
391
 	}
406
392
 
407
-	skb_reset_network_header(skb);
393
+	skb_set_network_header(skb, (tunnel->dev->type == ARPHRD_ETHER) ? ETH_HLEN : 0);
408
394
 
409
395
 	err = IP_ECN_decapsulate(iph, skb);
410
396
 	if (unlikely(err)) {
..
..
@@ -418,12 +404,7 @@
418
404
 		}
419
405
 	}
420
406
 
421
-	tstats = this_cpu_ptr(tunnel->dev->tstats);
422
-	u64_stats_update_begin(&tstats->syncp);
423
-	tstats->rx_packets++;
424
-	tstats->rx_bytes += skb->len;
425
-	u64_stats_update_end(&tstats->syncp);
426
-
407
+	dev_sw_netstats_rx_add(tunnel->dev, skb->len);
427
408
 	skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(tunnel->dev)));
428
409
 
429
410
 	if (tunnel->dev->type == ARPHRD_ETHER) {
..
..
@@ -502,41 +483,47 @@
502
483
 
503
484
 static int tnl_update_pmtu(struct net_device *dev, struct sk_buff *skb,
504
485
 			    struct rtable *rt, __be16 df,
505
-			    const struct iphdr *inner_iph)
486
+			    const struct iphdr *inner_iph,
487
+			    int tunnel_hlen, __be32 dst, bool md)
506
488
 {
507
489
 	struct ip_tunnel *tunnel = netdev_priv(dev);
508
490
 	int pkt_size;
509
491
 	int mtu;
510
492
 
511
-	pkt_size = skb->len - tunnel->hlen;
493
+	tunnel_hlen = md ? tunnel_hlen : tunnel->hlen;
494
+	pkt_size = skb->len - tunnel_hlen;
512
495
 	pkt_size -= dev->type == ARPHRD_ETHER ? dev->hard_header_len : 0;
513
496
 
514
497
 	if (df) {
515
-		mtu = dst_mtu(&rt->dst) - (sizeof(struct iphdr) + tunnel->hlen);
498
+		mtu = dst_mtu(&rt->dst) - (sizeof(struct iphdr) + tunnel_hlen);
516
499
 		mtu -= dev->type == ARPHRD_ETHER ? dev->hard_header_len : 0;
517
500
 	} else {
518
-		mtu = skb_dst(skb) ? dst_mtu(skb_dst(skb)) : dev->mtu;
501
+		mtu = skb_valid_dst(skb) ? dst_mtu(skb_dst(skb)) : dev->mtu;
519
502
 	}
520
503
 
521
-	skb_dst_update_pmtu_no_confirm(skb, mtu);
504
+	if (skb_valid_dst(skb))
505
+		skb_dst_update_pmtu_no_confirm(skb, mtu);
522
506
 
523
507
 	if (skb->protocol == htons(ETH_P_IP)) {
524
508
 		if (!skb_is_gso(skb) &&
525
509
 		    (inner_iph->frag_off & htons(IP_DF)) &&
526
510
 		    mtu < pkt_size) {
527
-			memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
528
-			icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
511
+			icmp_ndo_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
529
512
 			return -E2BIG;
530
513
 		}
531
514
 	}
532
515
 #if IS_ENABLED(CONFIG_IPV6)
533
516
 	else if (skb->protocol == htons(ETH_P_IPV6)) {
534
-		struct rt6_info *rt6 = (struct rt6_info *)skb_dst(skb);
517
+		struct rt6_info *rt6;
518
+		__be32 daddr;
519
+
520
+		rt6 = skb_valid_dst(skb) ? (struct rt6_info *)skb_dst(skb) :
521
+					   NULL;
522
+		daddr = md ? dst : tunnel->parms.iph.daddr;
535
523
 
536
524
 		if (rt6 && mtu < dst_mtu(skb_dst(skb)) &&
537
525
 			   mtu >= IPV6_MIN_MTU) {
538
-			if ((tunnel->parms.iph.daddr &&
539
-			    !ipv4_is_multicast(tunnel->parms.iph.daddr)) ||
526
+			if ((daddr && !ipv4_is_multicast(daddr)) ||
540
527
 			    rt6->rt6i_dst.plen == 128) {
541
528
 				rt6->rt6i_flags |= RTF_MODIFIED;
542
529
 				dst_metric_set(skb_dst(skb), RTAX_MTU, mtu);
..
..
@@ -545,7 +532,7 @@
545
532
 
546
533
 		if (!skb_is_gso(skb) && mtu >= IPV6_MIN_MTU &&
547
534
 					mtu < pkt_size) {
548
-			icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
535
+			icmpv6_ndo_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
549
536
 			return -E2BIG;
550
537
 		}
551
538
 	}
..
..
@@ -553,17 +540,19 @@
553
540
 	return 0;
554
541
 }
555
542
 
556
-void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev, u8 proto)
543
+void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
544
+		       u8 proto, int tunnel_hlen)
557
545
 {
558
546
 	struct ip_tunnel *tunnel = netdev_priv(dev);
559
547
 	u32 headroom = sizeof(struct iphdr);
560
548
 	struct ip_tunnel_info *tun_info;
561
549
 	const struct ip_tunnel_key *key;
562
550
 	const struct iphdr *inner_iph;
563
-	struct rtable *rt;
551
+	struct rtable *rt = NULL;
564
552
 	struct flowi4 fl4;
565
553
 	__be16 df = 0;
566
554
 	u8 tos, ttl;
555
+	bool use_cache;
567
556
 
568
557
 	tun_info = skb_tunnel_info(skb);
569
558
 	if (unlikely(!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX) ||
..
..
@@ -581,19 +570,37 @@
581
570
 	}
582
571
 	ip_tunnel_init_flow(&fl4, proto, key->u.ipv4.dst, key->u.ipv4.src,
583
572
 			    tunnel_id_to_key32(key->tun_id), RT_TOS(tos),
584
-			    0, skb->mark);
573
+			    0, skb->mark, skb_get_hash(skb));
585
574
 	if (tunnel->encap.type != TUNNEL_ENCAP_NONE)
586
575
 		goto tx_error;
587
-	rt = ip_route_output_key(tunnel->net, &fl4);
588
-	if (IS_ERR(rt)) {
589
-		dev->stats.tx_carrier_errors++;
590
-		goto tx_error;
576
+
577
+	use_cache = ip_tunnel_dst_cache_usable(skb, tun_info);
578
+	if (use_cache)
579
+		rt = dst_cache_get_ip4(&tun_info->dst_cache, &fl4.saddr);
580
+	if (!rt) {
581
+		rt = ip_route_output_key(tunnel->net, &fl4);
582
+		if (IS_ERR(rt)) {
583
+			dev->stats.tx_carrier_errors++;
584
+			goto tx_error;
585
+		}
586
+		if (use_cache)
587
+			dst_cache_set_ip4(&tun_info->dst_cache, &rt->dst,
588
+					  fl4.saddr);
591
589
 	}
592
590
 	if (rt->dst.dev == dev) {
593
591
 		ip_rt_put(rt);
594
592
 		dev->stats.collisions++;
595
593
 		goto tx_error;
596
594
 	}
595
+
596
+	if (key->tun_flags & TUNNEL_DONT_FRAGMENT)
597
+		df = htons(IP_DF);
598
+	if (tnl_update_pmtu(dev, skb, rt, df, inner_iph, tunnel_hlen,
599
+			    key->u.ipv4.dst, true)) {
600
+		ip_rt_put(rt);
601
+		goto tx_error;
602
+	}
603
+
597
604
 	tos = ip_tunnel_ecn_encap(tos, inner_iph, skb);
598
605
 	ttl = key->ttl;
599
606
 	if (ttl == 0) {
..
..
@@ -604,15 +611,12 @@
604
611
 		else
605
612
 			ttl = ip4_dst_hoplimit(&rt->dst);
606
613
 	}
607
-	if (key->tun_flags & TUNNEL_DONT_FRAGMENT)
608
-		df = htons(IP_DF);
609
-	else if (skb->protocol == htons(ETH_P_IP))
610
-		df = inner_iph->frag_off & htons(IP_DF);
611
-	headroom += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
612
-	if (headroom > dev->needed_headroom)
613
-		dev->needed_headroom = headroom;
614
614
 
615
-	if (skb_cow_head(skb, dev->needed_headroom)) {
615
+	headroom += LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len;
616
+	if (headroom > READ_ONCE(dev->needed_headroom))
617
+		WRITE_ONCE(dev->needed_headroom, headroom);
618
+
619
+	if (skb_cow_head(skb, READ_ONCE(dev->needed_headroom))) {
616
620
 		ip_rt_put(rt);
617
621
 		goto tx_dropped;
618
622
 	}
..
..
@@ -633,14 +637,17 @@
633
637
 		    const struct iphdr *tnl_params, u8 protocol)
634
638
 {
635
639
 	struct ip_tunnel *tunnel = netdev_priv(dev);
640
+	struct ip_tunnel_info *tun_info = NULL;
636
641
 	const struct iphdr *inner_iph;
637
-	struct flowi4 fl4;
638
-	u8     tos, ttl;
639
-	__be16 df;
640
-	struct rtable *rt;		/* Route to the other host */
641
642
 	unsigned int max_headroom;	/* The extra header space needed */
642
-	__be32 dst;
643
+	struct rtable *rt = NULL;		/* Route to the other host */
644
+	bool use_cache = false;
645
+	struct flowi4 fl4;
646
+	bool md = false;
643
647
 	bool connected;
648
+	u8 tos, ttl;
649
+	__be32 dst;
650
+	__be16 df;
644
651
 
645
652
 	inner_iph = (const struct iphdr *)skb_inner_network_header(skb);
646
653
 	connected = (tunnel->parms.iph.daddr != 0);
..
..
@@ -650,7 +657,6 @@
650
657
 	dst = tnl_params->daddr;
651
658
 	if (dst == 0) {
652
659
 		/* NBMA tunnel */
653
-		struct ip_tunnel_info *tun_info;
654
660
 
655
661
 		if (!skb_dst(skb)) {
656
662
 			dev->stats.tx_fifo_errors++;
..
..
@@ -660,8 +666,11 @@
660
666
 		tun_info = skb_tunnel_info(skb);
661
667
 		if (tun_info && (tun_info->mode & IP_TUNNEL_INFO_TX) &&
662
668
 		    ip_tunnel_info_af(tun_info) == AF_INET &&
663
-		    tun_info->key.u.ipv4.dst)
669
+		    tun_info->key.u.ipv4.dst) {
664
670
 			dst = tun_info->key.u.ipv4.dst;
671
+			md = true;
672
+			connected = true;
673
+		}
665
674
 		else if (skb->protocol == htons(ETH_P_IP)) {
666
675
 			rt = skb_rtable(skb);
667
676
 			dst = rt_nexthop(rt, inner_iph->daddr);
..
..
@@ -700,7 +709,8 @@
700
709
 		else
701
710
 			goto tx_error;
702
711
 
703
-		connected = false;
712
+		if (!md)
713
+			connected = false;
704
714
 	}
705
715
 
706
716
 	tos = tnl_params->tos;
..
..
@@ -717,13 +727,20 @@
717
727
 
718
728
 	ip_tunnel_init_flow(&fl4, protocol, dst, tnl_params->saddr,
719
729
 			    tunnel->parms.o_key, RT_TOS(tos), tunnel->parms.link,
720
-			    tunnel->fwmark);
730
+			    tunnel->fwmark, skb_get_hash(skb));
721
731
 
722
732
 	if (ip_tunnel_encap(skb, tunnel, &protocol, &fl4) < 0)
723
733
 		goto tx_error;
724
734
 
725
-	rt = connected ? dst_cache_get_ip4(&tunnel->dst_cache, &fl4.saddr) :
726
-			 NULL;
735
+	if (connected && md) {
736
+		use_cache = ip_tunnel_dst_cache_usable(skb, tun_info);
737
+		if (use_cache)
738
+			rt = dst_cache_get_ip4(&tun_info->dst_cache,
739
+					       &fl4.saddr);
740
+	} else {
741
+		rt = connected ? dst_cache_get_ip4(&tunnel->dst_cache,
742
+						&fl4.saddr) : NULL;
743
+	}
727
744
 
728
745
 	if (!rt) {
729
746
 		rt = ip_route_output_key(tunnel->net, &fl4);
..
..
@@ -732,7 +749,10 @@
732
749
 			dev->stats.tx_carrier_errors++;
733
750
 			goto tx_error;
734
751
 		}
735
-		if (connected)
752
+		if (use_cache)
753
+			dst_cache_set_ip4(&tun_info->dst_cache, &rt->dst,
754
+					  fl4.saddr);
755
+		else if (!md && connected)
736
756
 			dst_cache_set_ip4(&tunnel->dst_cache, &rt->dst,
737
757
 					  fl4.saddr);
738
758
 	}
..
..
@@ -747,7 +767,7 @@
747
767
 	if (skb->protocol == htons(ETH_P_IP) && !tunnel->ignore_df)
748
768
 		df |= (inner_iph->frag_off & htons(IP_DF));
749
769
 
750
-	if (tnl_update_pmtu(dev, skb, rt, df, inner_iph)) {
770
+	if (tnl_update_pmtu(dev, skb, rt, df, inner_iph, 0, 0, false)) {
751
771
 		ip_rt_put(rt);
752
772
 		goto tx_error;
753
773
 	}
..
..
@@ -777,10 +797,10 @@
777
797
 
778
798
 	max_headroom = LL_RESERVED_SPACE(rt->dst.dev) + sizeof(struct iphdr)
779
799
 			+ rt->dst.header_len + ip_encap_hlen(&tunnel->encap);
780
-	if (max_headroom > dev->needed_headroom)
781
-		dev->needed_headroom = max_headroom;
800
+	if (max_headroom > READ_ONCE(dev->needed_headroom))
801
+		WRITE_ONCE(dev->needed_headroom, max_headroom);
782
802
 
783
-	if (skb_cow_head(skb, dev->needed_headroom)) {
803
+	if (skb_cow_head(skb, READ_ONCE(dev->needed_headroom))) {
784
804
 		ip_rt_put(rt);
785
805
 		dev->stats.tx_dropped++;
786
806
 		kfree_skb(skb);
..
..
@@ -836,7 +856,7 @@
836
856
 	netdev_state_change(dev);
837
857
 }
838
858
 
839
-int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd)
859
+int ip_tunnel_ctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd)
840
860
 {
841
861
 	int err = 0;
842
862
 	struct ip_tunnel *t = netdev_priv(dev);
..
..
@@ -936,6 +956,20 @@
936
956
 done:
937
957
 	return err;
938
958
 }
959
+EXPORT_SYMBOL_GPL(ip_tunnel_ctl);
960
+
961
+int ip_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
962
+{
963
+	struct ip_tunnel_parm p;
964
+	int err;
965
+
966
+	if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
967
+		return -EFAULT;
968
+	err = dev->netdev_ops->ndo_tunnel_ctl(dev, &p, cmd);
969
+	if (!err && copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
970
+		return -EFAULT;
971
+	return err;
972
+}
939
973
 EXPORT_SYMBOL_GPL(ip_tunnel_ioctl);
940
974
 
941
975
 int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict)