forked from ~ljy/RK356X_SDK_RELEASE
blame | history | patch | commit | commitdiff
hc
2023-12-11 072de836f53be56a70cecf70b43ae43b7ce17376 
 kernel/net/packet/af_packet.c
....@@ -1,3 +1,4 @@
1
+// SPDX-License-Identifier: GPL-2.0-or-later
12 /*
23  * INET		An implementation of the TCP/IP protocol suite for the LINUX
34  *		operating system.  INET is implemented using the  BSD Socket
....@@ -43,13 +44,6 @@
4344  *		Chetan Loke	:	Implemented TPACKET_V3 block abstraction
4445  *					layer.
4546  *					Copyright (C) 2011, <lokec@ccs.neu.edu>
46
- *
47
- *
48
- *		This program is free software; you can redistribute it and/or
49
- *		modify it under the terms of the GNU General Public License
50
- *		as published by the Free Software Foundation; either version
51
- *		2 of the License, or (at your option) any later version.
52
- *
5347  */
5448 
5549 #include <linux/types.h>
....@@ -99,52 +93,56 @@
9993 
10094 /*
10195    Assumptions:
102
-   - if device has no dev->hard_header routine, it adds and removes ll header
103
-     inside itself. In this case ll header is invisible outside of device,
104
-     but higher levels still should reserve dev->hard_header_len.
105
-     Some devices are enough clever to reallocate skb, when header
106
-     will not fit to reserved space (tunnel), another ones are silly
107
-     (PPP).
96
+   - If the device has no dev->header_ops->create, there is no LL header
97
+     visible above the device. In this case, its hard_header_len should be 0.
98
+     The device may prepend its own header internally. In this case, its
99
+     needed_headroom should be set to the space needed for it to add its
100
+     internal header.
101
+     For example, a WiFi driver pretending to be an Ethernet driver should
102
+     set its hard_header_len to be the Ethernet header length, and set its
103
+     needed_headroom to be (the real WiFi header length - the fake Ethernet
104
+     header length).
108105    - packet socket receives packets with pulled ll header,
109106      so that SOCK_RAW should push it back.
110107 
111108 On receive:
112109 -----------
113110 
114
-Incoming, dev->hard_header!=NULL
111
+Incoming, dev_has_header(dev) == true
115112    mac_header -> ll header
116113    data       -> data
117114 
118
-Outgoing, dev->hard_header!=NULL
115
+Outgoing, dev_has_header(dev) == true
119116    mac_header -> ll header
120117    data       -> ll header
121118 
122
-Incoming, dev->hard_header==NULL
123
-   mac_header -> UNKNOWN position. It is very likely, that it points to ll
124
-		 header.  PPP makes it, that is wrong, because introduce
125
-		 assymetry between rx and tx paths.
119
+Incoming, dev_has_header(dev) == false
120
+   mac_header -> data
121
+     However drivers often make it point to the ll header.
122
+     This is incorrect because the ll header should be invisible to us.
126123    data       -> data
127124 
128
-Outgoing, dev->hard_header==NULL
129
-   mac_header -> data. ll header is still not built!
125
+Outgoing, dev_has_header(dev) == false
126
+   mac_header -> data. ll header is invisible to us.
130127    data       -> data
131128 
132129 Resume
133
-  If dev->hard_header==NULL we are unlikely to restore sensible ll header.
130
+  If dev_has_header(dev) == false we are unable to restore the ll header,
131
+    because it is invisible to us.
134132 
135133 
136134 On transmit:
137135 ------------
138136 
139
-dev->hard_header != NULL
137
+dev->header_ops != NULL
140138    mac_header -> ll header
141139    data       -> ll header
142140 
143
-dev->hard_header == NULL (ll header is added by device, we cannot control it)
141
+dev->header_ops == NULL (ll header is invisible to us)
144142    mac_header -> data
145143    data       -> data
146144 
147
-   We should set nh.raw on output to correct posistion,
145
+   We should set network_header on output to the correct position,
148146    packet classifier depends on it.
149147  */
150148 
....@@ -183,7 +181,6 @@
183181 #define BLOCK_LEN(x)		((x)->hdr.bh1.blk_len)
184182 #define BLOCK_SNUM(x)		((x)->hdr.bh1.seq_num)
185183 #define BLOCK_O2PRIV(x)	((x)->offset_to_priv)
186
-#define BLOCK_PRIV(x)		((void *)((char *)(x) + BLOCK_O2PRIV(x)))
187184 
188185 struct packet_sock;
189186 static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev,
....@@ -275,24 +272,22 @@
275272 	return po->xmit == packet_direct_xmit;
276273 }
277274 
278
-static u16 __packet_pick_tx_queue(struct net_device *dev, struct sk_buff *skb,
279
-				  struct net_device *sb_dev)
280
-{
281
-	return dev_pick_tx_cpu_id(dev, skb, sb_dev, NULL);
282
-}
283
-
284275 static u16 packet_pick_tx_queue(struct sk_buff *skb)
285276 {
286277 	struct net_device *dev = skb->dev;
287278 	const struct net_device_ops *ops = dev->netdev_ops;
279
+	int cpu = raw_smp_processor_id();
288280 	u16 queue_index;
289281 
282
+#ifdef CONFIG_XPS
283
+	skb->sender_cpu = cpu + 1;
284
+#endif
285
+	skb_record_rx_queue(skb, cpu % dev->real_num_tx_queues);
290286 	if (ops->ndo_select_queue) {
291
-		queue_index = ops->ndo_select_queue(dev, skb, NULL,
292
-						    __packet_pick_tx_queue);
287
+		queue_index = ops->ndo_select_queue(dev, skb, NULL);
293288 		queue_index = netdev_cap_txqueue(dev, queue_index);
294289 	} else {
295
-		queue_index = __packet_pick_tx_queue(dev, skb, NULL);
290
+		queue_index = netdev_pick_tx(dev, skb, NULL);
296291 	}
297292 
298293 	return queue_index;
....@@ -392,7 +387,7 @@
392387 	smp_wmb();
393388 }
394389 
395
-static int __packet_get_status(struct packet_sock *po, void *frame)
390
+static int __packet_get_status(const struct packet_sock *po, void *frame)
396391 {
397392 	union tpacket_uhdr h;
398393 
....@@ -416,17 +411,18 @@
416411 	}
417412 }
418413 
419
-static __u32 tpacket_get_timestamp(struct sk_buff *skb, struct timespec *ts,
414
+static __u32 tpacket_get_timestamp(struct sk_buff *skb, struct timespec64 *ts,
420415 				   unsigned int flags)
421416 {
422417 	struct skb_shared_hwtstamps *shhwtstamps = skb_hwtstamps(skb);
423418 
424419 	if (shhwtstamps &&
425420 	    (flags & SOF_TIMESTAMPING_RAW_HARDWARE) &&
426
-	    ktime_to_timespec_cond(shhwtstamps->hwtstamp, ts))
421
+	    ktime_to_timespec64_cond(shhwtstamps->hwtstamp, ts))
427422 		return TP_STATUS_TS_RAW_HARDWARE;
428423 
429
-	if (ktime_to_timespec_cond(skb->tstamp, ts))
424
+	if ((flags & SOF_TIMESTAMPING_SOFTWARE) &&
425
+	    ktime_to_timespec64_cond(skb->tstamp, ts))
430426 		return TP_STATUS_TS_SOFTWARE;
431427 
432428 	return 0;
....@@ -436,13 +432,20 @@
436432 				    struct sk_buff *skb)
437433 {
438434 	union tpacket_uhdr h;
439
-	struct timespec ts;
435
+	struct timespec64 ts;
440436 	__u32 ts_status;
441437 
442438 	if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp)))
443439 		return 0;
444440 
445441 	h.raw = frame;
442
+	/*
443
+	 * versions 1 through 3 overflow the timestamps in y2106, since they
444
+	 * all store the seconds in a 32-bit unsigned integer.
445
+	 * If we create a version 4, that should have a 64-bit timestamp,
446
+	 * either 64-bit seconds + 32-bit nanoseconds, or just 64-bit
447
+	 * nanoseconds.
448
+	 */
446449 	switch (po->tp_version) {
447450 	case TPACKET_V1:
448451 		h.h1->tp_sec = ts.tv_sec;
....@@ -468,10 +471,10 @@
468471 	return ts_status;
469472 }
470473 
471
-static void *packet_lookup_frame(struct packet_sock *po,
472
-		struct packet_ring_buffer *rb,
473
-		unsigned int position,
474
-		int status)
474
+static void *packet_lookup_frame(const struct packet_sock *po,
475
+				 const struct packet_ring_buffer *rb,
476
+				 unsigned int position,
477
+				 int status)
475478 {
476479 	unsigned int pg_vec_pos, frame_offset;
477480 	union tpacket_uhdr h;
....@@ -528,7 +531,7 @@
528531 				int blk_size_in_bytes)
529532 {
530533 	struct net_device *dev;
531
-	unsigned int mbits = 0, msec = 0, div = 0, tmo = 0;
534
+	unsigned int mbits, div;
532535 	struct ethtool_link_ksettings ecmd;
533536 	int err;
534537 
....@@ -540,31 +543,25 @@
540543 	}
541544 	err = __ethtool_get_link_ksettings(dev, &ecmd);
542545 	rtnl_unlock();
543
-	if (!err) {
544
-		/*
545
-		 * If the link speed is so slow you don't really
546
-		 * need to worry about perf anyways
547
-		 */
548
-		if (ecmd.base.speed < SPEED_1000 ||
549
-		    ecmd.base.speed == SPEED_UNKNOWN) {
550
-			return DEFAULT_PRB_RETIRE_TOV;
551
-		} else {
552
-			msec = 1;
553
-			div = ecmd.base.speed / 1000;
554
-		}
555
-	} else
546
+	if (err)
556547 		return DEFAULT_PRB_RETIRE_TOV;
557548 
549
+	/* If the link speed is so slow you don't really
550
+	 * need to worry about perf anyways
551
+	 */
552
+	if (ecmd.base.speed < SPEED_1000 ||
553
+	    ecmd.base.speed == SPEED_UNKNOWN)
554
+		return DEFAULT_PRB_RETIRE_TOV;
555
+
556
+	div = ecmd.base.speed / 1000;
558557 	mbits = (blk_size_in_bytes * 8) / (1024 * 1024);
559558 
560559 	if (div)
561560 		mbits /= div;
562561 
563
-	tmo = mbits * msec;
564
-
565562 	if (div)
566
-		return tmo+1;
567
-	return tmo;
563
+		return mbits + 1;
564
+	return mbits;
568565 }
569566 
570567 static void prb_init_ft_ops(struct tpacket_kbdq_core *p1,
....@@ -600,6 +597,7 @@
600597 						req_u->req3.tp_block_size);
601598 	p1->tov_in_jiffies = msecs_to_jiffies(p1->retire_blk_tov);
602599 	p1->blk_sizeof_priv = req_u->req3.tp_sizeof_priv;
600
+	rwlock_init(&p1->blk_fill_in_prog_lock);
603601 
604602 	p1->max_frame_len = p1->kblk_size - BLK_PLUS_PRIV(p1->blk_sizeof_priv);
605603 	prb_init_ft_ops(p1, req_u);
....@@ -666,10 +664,9 @@
666664 	 *
667665 	 */
668666 	if (BLOCK_NUM_PKTS(pbd)) {
669
-		while (atomic_read(&pkc->blk_fill_in_prog)) {
670
-			/* Waiting for skb_copy_bits to finish... */
671
-			cpu_relax();
672
-		}
667
+		/* Waiting for skb_copy_bits to finish... */
668
+		write_lock(&pkc->blk_fill_in_prog_lock);
669
+		write_unlock(&pkc->blk_fill_in_prog_lock);
673670 	}
674671 
675672 	if (pkc->last_kactive_blk_num == pkc->kactive_blk_num) {
....@@ -767,7 +764,7 @@
767764 	struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1;
768765 	struct sock *sk = &po->sk;
769766 
770
-	if (po->stats.stats3.tp_drops)
767
+	if (atomic_read(&po->tp_drops))
771768 		status |= TP_STATUS_LOSING;
772769 
773770 	last_pkt = (struct tpacket3_hdr *)pkc1->prev;
....@@ -783,8 +780,8 @@
783780 		 * It shouldn't really happen as we don't close empty
784781 		 * blocks. See prb_retire_rx_blk_timer_expired().
785782 		 */
786
-		struct timespec ts;
787
-		getnstimeofday(&ts);
783
+		struct timespec64 ts;
784
+		ktime_get_real_ts64(&ts);
788785 		h1->ts_last_pkt.ts_sec = ts.tv_sec;
789786 		h1->ts_last_pkt.ts_nsec	= ts.tv_nsec;
790787 	}
....@@ -814,7 +811,7 @@
814811 static void prb_open_block(struct tpacket_kbdq_core *pkc1,
815812 	struct tpacket_block_desc *pbd1)
816813 {
817
-	struct timespec ts;
814
+	struct timespec64 ts;
818815 	struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1;
819816 
820817 	smp_rmb();
....@@ -827,7 +824,7 @@
827824 	BLOCK_NUM_PKTS(pbd1) = 0;
828825 	BLOCK_LEN(pbd1) = BLK_PLUS_PRIV(pkc1->blk_sizeof_priv);
829826 
830
-	getnstimeofday(&ts);
827
+	ktime_get_real_ts64(&ts);
831828 
832829 	h1->ts_first_pkt.ts_sec = ts.tv_sec;
833830 	h1->ts_first_pkt.ts_nsec = ts.tv_nsec;
....@@ -928,10 +925,9 @@
928925 		 * the timer-handler already handled this case.
929926 		 */
930927 		if (!(status & TP_STATUS_BLK_TMO)) {
931
-			while (atomic_read(&pkc->blk_fill_in_prog)) {
932
-				/* Waiting for skb_copy_bits to finish... */
933
-				cpu_relax();
934
-			}
928
+			/* Waiting for skb_copy_bits to finish... */
929
+			write_lock(&pkc->blk_fill_in_prog_lock);
930
+			write_unlock(&pkc->blk_fill_in_prog_lock);
935931 		}
936932 		prb_close_block(pkc, pbd, po, status);
937933 		return;
....@@ -952,7 +948,8 @@
952948 	__releases(&pkc->blk_fill_in_prog_lock)
953949 {
954950 	struct tpacket_kbdq_core *pkc  = GET_PBDQC_FROM_RB(rb);
955
-	atomic_dec(&pkc->blk_fill_in_prog);
951
+
952
+	read_unlock(&pkc->blk_fill_in_prog_lock);
956953 }
957954 
958955 static void prb_fill_rxhash(struct tpacket_kbdq_core *pkc,
....@@ -1007,14 +1004,13 @@
10071004 	pkc->nxt_offset += TOTAL_PKT_LEN_INCL_ALIGN(len);
10081005 	BLOCK_LEN(pbd) += TOTAL_PKT_LEN_INCL_ALIGN(len);
10091006 	BLOCK_NUM_PKTS(pbd) += 1;
1010
-	atomic_inc(&pkc->blk_fill_in_prog);
1007
+	read_lock(&pkc->blk_fill_in_prog_lock);
10111008 	prb_run_all_ft_ops(pkc, ppd);
10121009 }
10131010 
10141011 /* Assumes caller has the sk->rx_queue.lock */
10151012 static void *__packet_lookup_frame_in_block(struct packet_sock *po,
10161013 					    struct sk_buff *skb,
1017
-						int status,
10181014 					    unsigned int len
10191015 					    )
10201016 {
....@@ -1086,7 +1082,7 @@
10861082 					po->rx_ring.head, status);
10871083 		return curr;
10881084 	case TPACKET_V3:
1089
-		return __packet_lookup_frame_in_block(po, skb, status, len);
1085
+		return __packet_lookup_frame_in_block(po, skb, len);
10901086 	default:
10911087 		WARN(1, "TPACKET version not supported\n");
10921088 		BUG();
....@@ -1094,10 +1090,10 @@
10941090 	}
10951091 }
10961092 
1097
-static void *prb_lookup_block(struct packet_sock *po,
1098
-				     struct packet_ring_buffer *rb,
1099
-				     unsigned int idx,
1100
-				     int status)
1093
+static void *prb_lookup_block(const struct packet_sock *po,
1094
+			      const struct packet_ring_buffer *rb,
1095
+			      unsigned int idx,
1096
+			      int status)
11011097 {
11021098 	struct tpacket_kbdq_core *pkc  = GET_PBDQC_FROM_RB(rb);
11031099 	struct tpacket_block_desc *pbd = GET_PBLOCK_DESC(pkc, idx);
....@@ -1210,12 +1206,12 @@
12101206 #define ROOM_LOW	0x1
12111207 #define ROOM_NORMAL	0x2
12121208 
1213
-static bool __tpacket_has_room(struct packet_sock *po, int pow_off)
1209
+static bool __tpacket_has_room(const struct packet_sock *po, int pow_off)
12141210 {
12151211 	int idx, len;
12161212 
1217
-	len = po->rx_ring.frame_max + 1;
1218
-	idx = po->rx_ring.head;
1213
+	len = READ_ONCE(po->rx_ring.frame_max) + 1;
1214
+	idx = READ_ONCE(po->rx_ring.head);
12191215 	if (pow_off)
12201216 		idx += len >> pow_off;
12211217 	if (idx >= len)
....@@ -1223,12 +1219,12 @@
12231219 	return packet_lookup_frame(po, &po->rx_ring, idx, TP_STATUS_KERNEL);
12241220 }
12251221 
1226
-static bool __tpacket_v3_has_room(struct packet_sock *po, int pow_off)
1222
+static bool __tpacket_v3_has_room(const struct packet_sock *po, int pow_off)
12271223 {
12281224 	int idx, len;
12291225 
1230
-	len = po->rx_ring.prb_bdqc.knum_blocks;
1231
-	idx = po->rx_ring.prb_bdqc.kactive_blk_num;
1226
+	len = READ_ONCE(po->rx_ring.prb_bdqc.knum_blocks);
1227
+	idx = READ_ONCE(po->rx_ring.prb_bdqc.kactive_blk_num);
12321228 	if (pow_off)
12331229 		idx += len >> pow_off;
12341230 	if (idx >= len)
....@@ -1236,15 +1232,18 @@
12361232 	return prb_lookup_block(po, &po->rx_ring, idx, TP_STATUS_KERNEL);
12371233 }
12381234 
1239
-static int __packet_rcv_has_room(struct packet_sock *po, struct sk_buff *skb)
1235
+static int __packet_rcv_has_room(const struct packet_sock *po,
1236
+				 const struct sk_buff *skb)
12401237 {
1241
-	struct sock *sk = &po->sk;
1238
+	const struct sock *sk = &po->sk;
12421239 	int ret = ROOM_NONE;
12431240 
12441241 	if (po->prot_hook.func != tpacket_rcv) {
1245
-		int avail = sk->sk_rcvbuf - atomic_read(&sk->sk_rmem_alloc)
1246
-					  - (skb ? skb->truesize : 0);
1247
-		if (avail > (sk->sk_rcvbuf >> ROOM_POW_OFF))
1242
+		int rcvbuf = READ_ONCE(sk->sk_rcvbuf);
1243
+		int avail = rcvbuf - atomic_read(&sk->sk_rmem_alloc)
1244
+				   - (skb ? skb->truesize : 0);
1245
+
1246
+		if (avail > (rcvbuf >> ROOM_POW_OFF))
12481247 			return ROOM_NORMAL;
12491248 		else if (avail > 0)
12501249 			return ROOM_LOW;
....@@ -1269,17 +1268,22 @@
12691268 
12701269 static int packet_rcv_has_room(struct packet_sock *po, struct sk_buff *skb)
12711270 {
1272
-	int ret;
1273
-	bool has_room;
1271
+	int pressure, ret;
12741272 
1275
-	spin_lock_bh(&po->sk.sk_receive_queue.lock);
12761273 	ret = __packet_rcv_has_room(po, skb);
1277
-	has_room = ret == ROOM_NORMAL;
1278
-	if (po->pressure == has_room)
1279
-		po->pressure = !has_room;
1280
-	spin_unlock_bh(&po->sk.sk_receive_queue.lock);
1274
+	pressure = ret != ROOM_NORMAL;
1275
+
1276
+	if (READ_ONCE(po->pressure) != pressure)
1277
+		WRITE_ONCE(po->pressure, pressure);
12811278 
12821279 	return ret;
1280
+}
1281
+
1282
+static void packet_rcv_try_clear_pressure(struct packet_sock *po)
1283
+{
1284
+	if (READ_ONCE(po->pressure) &&
1285
+	    __packet_rcv_has_room(po, NULL) == ROOM_NORMAL)
1286
+		WRITE_ONCE(po->pressure,  0);
12831287 }
12841288 
12851289 static void packet_sock_destruct(struct sock *sk)
....@@ -1355,7 +1359,7 @@
13551359 	struct packet_sock *po, *po_next, *po_skip = NULL;
13561360 	unsigned int i, j, room = ROOM_NONE;
13571361 
1358
-	po = pkt_sk(f->arr[idx]);
1362
+	po = pkt_sk(rcu_dereference(f->arr[idx]));
13591363 
13601364 	if (try_self) {
13611365 		room = packet_rcv_has_room(po, skb);
....@@ -1367,8 +1371,8 @@
13671371 
13681372 	i = j = min_t(int, po->rollover->sock, num - 1);
13691373 	do {
1370
-		po_next = pkt_sk(f->arr[i]);
1371
-		if (po_next != po_skip && !po_next->pressure &&
1374
+		po_next = pkt_sk(rcu_dereference(f->arr[i]));
1375
+		if (po_next != po_skip && !READ_ONCE(po_next->pressure) &&
13721376 		    packet_rcv_has_room(po_next, skb) == ROOM_NORMAL) {
13731377 			if (i != j)
13741378 				po->rollover->sock = i;
....@@ -1462,7 +1466,7 @@
14621466 	if (fanout_has_flag(f, PACKET_FANOUT_FLAG_ROLLOVER))
14631467 		idx = fanout_demux_rollover(f, skb, idx, true, num);
14641468 
1465
-	po = pkt_sk(f->arr[idx]);
1469
+	po = pkt_sk(rcu_dereference(f->arr[idx]));
14661470 	return po->prot_hook.func(skb, dev, &po->prot_hook, orig_dev);
14671471 }
14681472 
....@@ -1476,7 +1480,7 @@
14761480 	struct packet_fanout *f = po->fanout;
14771481 
14781482 	spin_lock(&f->lock);
1479
-	f->arr[f->num_members] = sk;
1483
+	rcu_assign_pointer(f->arr[f->num_members], sk);
14801484 	smp_wmb();
14811485 	f->num_members++;
14821486 	if (f->num_members == 1)
....@@ -1491,11 +1495,14 @@
14911495 
14921496 	spin_lock(&f->lock);
14931497 	for (i = 0; i < f->num_members; i++) {
1494
-		if (f->arr[i] == sk)
1498
+		if (rcu_dereference_protected(f->arr[i],
1499
+					      lockdep_is_held(&f->lock)) == sk)
14951500 			break;
14961501 	}
14971502 	BUG_ON(i >= f->num_members);
1498
-	f->arr[i] = f->arr[f->num_members - 1];
1503
+	rcu_assign_pointer(f->arr[i],
1504
+			   rcu_dereference_protected(f->arr[f->num_members - 1],
1505
+						     lockdep_is_held(&f->lock)));
14991506 	f->num_members--;
15001507 	if (f->num_members == 0)
15011508 		__dev_remove_pack(&f->prot_hook);
....@@ -1538,7 +1545,7 @@
15381545 	}
15391546 }
15401547 
1541
-static int fanout_set_data_cbpf(struct packet_sock *po, char __user *data,
1548
+static int fanout_set_data_cbpf(struct packet_sock *po, sockptr_t data,
15421549 				unsigned int len)
15431550 {
15441551 	struct bpf_prog *new;
....@@ -1547,10 +1554,10 @@
15471554 
15481555 	if (sock_flag(&po->sk, SOCK_FILTER_LOCKED))
15491556 		return -EPERM;
1550
-	if (len != sizeof(fprog))
1551
-		return -EINVAL;
1552
-	if (copy_from_user(&fprog, data, len))
1553
-		return -EFAULT;
1557
+
1558
+	ret = copy_bpf_fprog_from_user(&fprog, data, len);
1559
+	if (ret)
1560
+		return ret;
15541561 
15551562 	ret = bpf_prog_create_from_user(&new, &fprog, NULL, false);
15561563 	if (ret)
....@@ -1560,7 +1567,7 @@
15601567 	return 0;
15611568 }
15621569 
1563
-static int fanout_set_data_ebpf(struct packet_sock *po, char __user *data,
1570
+static int fanout_set_data_ebpf(struct packet_sock *po, sockptr_t data,
15641571 				unsigned int len)
15651572 {
15661573 	struct bpf_prog *new;
....@@ -1570,7 +1577,7 @@
15701577 		return -EPERM;
15711578 	if (len != sizeof(fd))
15721579 		return -EINVAL;
1573
-	if (copy_from_user(&fd, data, len))
1580
+	if (copy_from_sockptr(&fd, data, len))
15741581 		return -EFAULT;
15751582 
15761583 	new = bpf_prog_get_type(fd, BPF_PROG_TYPE_SOCKET_FILTER);
....@@ -1581,7 +1588,7 @@
15811588 	return 0;
15821589 }
15831590 
1584
-static int fanout_set_data(struct packet_sock *po, char __user *data,
1591
+static int fanout_set_data(struct packet_sock *po, sockptr_t data,
15851592 			   unsigned int len)
15861593 {
15871594 	switch (po->fanout->type) {
....@@ -1633,13 +1640,15 @@
16331640 	return false;
16341641 }
16351642 
1636
-static int fanout_add(struct sock *sk, u16 id, u16 type_flags)
1643
+static int fanout_add(struct sock *sk, struct fanout_args *args)
16371644 {
16381645 	struct packet_rollover *rollover = NULL;
16391646 	struct packet_sock *po = pkt_sk(sk);
1647
+	u16 type_flags = args->type_flags;
16401648 	struct packet_fanout *f, *match;
16411649 	u8 type = type_flags & 0xff;
16421650 	u8 flags = type_flags >> 8;
1651
+	u16 id = args->id;
16431652 	int err;
16441653 
16451654 	switch (type) {
....@@ -1697,11 +1706,21 @@
16971706 		}
16981707 	}
16991708 	err = -EINVAL;
1700
-	if (match && match->flags != flags)
1701
-		goto out;
1702
-	if (!match) {
1709
+	if (match) {
1710
+		if (match->flags != flags)
1711
+			goto out;
1712
+		if (args->max_num_members &&
1713
+		    args->max_num_members != match->max_num_members)
1714
+			goto out;
1715
+	} else {
1716
+		if (args->max_num_members > PACKET_FANOUT_MAX)
1717
+			goto out;
1718
+		if (!args->max_num_members)
1719
+			/* legacy PACKET_FANOUT_MAX */
1720
+			args->max_num_members = 256;
17031721 		err = -ENOMEM;
1704
-		match = kzalloc(sizeof(*match), GFP_KERNEL);
1722
+		match = kvzalloc(struct_size(match, arr, args->max_num_members),
1723
+				 GFP_KERNEL);
17051724 		if (!match)
17061725 			goto out;
17071726 		write_pnet(&match->net, sock_net(sk));
....@@ -1718,6 +1737,7 @@
17181737 		match->prot_hook.af_packet_priv = match;
17191738 		match->prot_hook.af_packet_net = read_pnet(&match->net);
17201739 		match->prot_hook.id_match = match_fanout_group;
1740
+		match->max_num_members = args->max_num_members;
17211741 		list_add(&match->list, &fanout_list);
17221742 	}
17231743 	err = -EINVAL;
....@@ -1728,7 +1748,7 @@
17281748 	    match->prot_hook.type == po->prot_hook.type &&
17291749 	    match->prot_hook.dev == po->prot_hook.dev) {
17301750 		err = -ENOSPC;
1731
-		if (refcount_read(&match->sk_ref) < PACKET_FANOUT_MAX) {
1751
+		if (refcount_read(&match->sk_ref) < match->max_num_members) {
17321752 			__dev_remove_pack(&po->prot_hook);
17331753 
17341754 			/* Paired with packet_setsockopt(PACKET_FANOUT_DATA) */
....@@ -1745,7 +1765,7 @@
17451765 
17461766 	if (err && !refcount_read(&match->sk_ref)) {
17471767 		list_del(&match->list);
1748
-		kfree(match);
1768
+		kvfree(match);
17491769 	}
17501770 
17511771 out:
....@@ -1835,7 +1855,7 @@
18351855 	skb_dst_drop(skb);
18361856 
18371857 	/* drop conntrack reference */
1838
-	nf_reset(skb);
1858
+	nf_reset_ct(skb);
18391859 
18401860 	spkt = &PACKET_SKB_CB(skb)->sa.pkt;
18411861 
....@@ -1863,6 +1883,16 @@
18631883 	return 0;
18641884 }
18651885 
1886
+static void packet_parse_headers(struct sk_buff *skb, struct socket *sock)
1887
+{
1888
+	if ((!skb->protocol || skb->protocol == htons(ETH_P_ALL)) &&
1889
+	    sock->type == SOCK_RAW) {
1890
+		skb_reset_mac_header(skb);
1891
+		skb->protocol = dev_parse_header_protocol(skb);
1892
+	}
1893
+
1894
+	skb_probe_transport_header(skb);
1895
+}
18661896 
18671897 /*
18681898  *	Output a raw packet to a device layer. This bypasses all the other
....@@ -1978,12 +2008,12 @@
19782008 	skb->mark = sk->sk_mark;
19792009 	skb->tstamp = sockc.transmit_time;
19802010 
1981
-	sock_tx_timestamp(sk, sockc.tsflags, &skb_shinfo(skb)->tx_flags);
2011
+	skb_setup_tx_timestamp(skb, sockc.tsflags);
19822012 
19832013 	if (unlikely(extra_len == 4))
19842014 		skb->no_fcs = 1;
19852015 
1986
-	skb_probe_transport_header(skb, 0);
2016
+	packet_parse_headers(skb, sock);
19872017 
19882018 	dev_queue_xmit(skb);
19892019 	rcu_read_unlock();
....@@ -2060,7 +2090,7 @@
20602090 
20612091 	skb->dev = dev;
20622092 
2063
-	if (dev->header_ops) {
2093
+	if (dev_has_header(dev)) {
20642094 		/* The device has an explicit notion of ll header,
20652095 		 * exported to higher levels.
20662096 		 *
....@@ -2125,7 +2155,7 @@
21252155 	skb_dst_drop(skb);
21262156 
21272157 	/* drop conntrack reference */
2128
-	nf_reset(skb);
2158
+	nf_reset_ct(skb);
21292159 
21302160 	spin_lock(&sk->sk_receive_queue.lock);
21312161 	po->stats.stats1.tp_packets++;
....@@ -2137,10 +2167,8 @@
21372167 
21382168 drop_n_acct:
21392169 	is_drop_n_account = true;
2140
-	spin_lock(&sk->sk_receive_queue.lock);
2141
-	po->stats.stats1.tp_drops++;
2170
+	atomic_inc(&po->tp_drops);
21422171 	atomic_inc(&sk->sk_drops);
2143
-	spin_unlock(&sk->sk_receive_queue.lock);
21442172 
21452173 drop_n_restore:
21462174 	if (skb_head != skb->data && skb_shared(skb)) {
....@@ -2169,7 +2197,7 @@
21692197 	unsigned short macoff, hdrlen;
21702198 	unsigned int netoff;
21712199 	struct sk_buff *copy_skb = NULL;
2172
-	struct timespec ts;
2200
+	struct timespec64 ts;
21732201 	__u32 ts_status;
21742202 	bool is_drop_n_account = false;
21752203 	unsigned int slot_id = 0;
....@@ -2191,7 +2219,7 @@
21912219 	if (!net_eq(dev_net(dev), sock_net(sk)))
21922220 		goto drop;
21932221 
2194
-	if (dev->header_ops) {
2222
+	if (dev_has_header(dev)) {
21952223 		if (sk->sk_type != SOCK_DGRAM)
21962224 			skb_push(skb, skb->data - skb_mac_header(skb));
21972225 		else if (skb->pkt_type == PACKET_OUTGOING) {
....@@ -2206,11 +2234,16 @@
22062234 	if (!res)
22072235 		goto drop_n_restore;
22082236 
2237
+	/* If we are flooded, just give up */
2238
+	if (__packet_rcv_has_room(po, skb) == ROOM_NONE) {
2239
+		atomic_inc(&po->tp_drops);
2240
+		goto drop_n_restore;
2241
+	}
2242
+
22092243 	if (skb->ip_summed == CHECKSUM_PARTIAL)
22102244 		status |= TP_STATUS_CSUMNOTREADY;
22112245 	else if (skb->pkt_type != PACKET_OUTGOING &&
2212
-		 (skb->ip_summed == CHECKSUM_COMPLETE ||
2213
-		  skb_csum_unnecessary(skb)))
2246
+		 skb_csum_unnecessary(skb))
22142247 		status |= TP_STATUS_CSUM_VALID;
22152248 
22162249 	if (snaplen > res)
....@@ -2231,9 +2264,7 @@
22312264 		macoff = netoff - maclen;
22322265 	}
22332266 	if (netoff > USHRT_MAX) {
2234
-		spin_lock(&sk->sk_receive_queue.lock);
2235
-		po->stats.stats1.tp_drops++;
2236
-		spin_unlock(&sk->sk_receive_queue.lock);
2267
+		atomic_inc(&po->tp_drops);
22372268 		goto drop_n_restore;
22382269 	}
22392270 	if (po->tp_version <= TPACKET_V2) {
....@@ -2246,8 +2277,11 @@
22462277 					copy_skb = skb_get(skb);
22472278 					skb_head = skb->data;
22482279 				}
2249
-				if (copy_skb)
2280
+				if (copy_skb) {
2281
+					memset(&PACKET_SKB_CB(copy_skb)->sa.ll, 0,
2282
+					       sizeof(PACKET_SKB_CB(copy_skb)->sa.ll));
22502283 					skb_set_owner_r(copy_skb, sk);
2284
+				}
22512285 			}
22522286 			snaplen = po->rx_ring.frame_size - macoff;
22532287 			if ((int)snaplen < 0) {
....@@ -2299,7 +2333,7 @@
22992333 	 * Anyways, moving it for V1/V2 only as V3 doesn't need this
23002334 	 * at packet level.
23012335 	 */
2302
-		if (po->stats.stats1.tp_drops)
2336
+		if (atomic_read(&po->tp_drops))
23032337 			status |= TP_STATUS_LOSING;
23042338 	}
23052339 
....@@ -2312,8 +2346,13 @@
23122346 
23132347 	skb_copy_bits(skb, 0, h.raw + macoff, snaplen);
23142348 
2315
-	if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp)))
2316
-		getnstimeofday(&ts);
2349
+	/* Always timestamp; prefer an existing software timestamp taken
2350
+	 * closer to the time of capture.
2351
+	 */
2352
+	ts_status = tpacket_get_timestamp(skb, &ts,
2353
+					  po->tp_tstamp | SOF_TIMESTAMPING_SOFTWARE);
2354
+	if (!ts_status)
2355
+		ktime_get_real_ts64(&ts);
23172356 
23182357 	status |= ts_status;
23192358 
....@@ -2412,9 +2451,9 @@
24122451 	return 0;
24132452 
24142453 drop_n_account:
2415
-	is_drop_n_account = true;
2416
-	po->stats.stats1.tp_drops++;
24172454 	spin_unlock(&sk->sk_receive_queue.lock);
2455
+	atomic_inc(&po->tp_drops);
2456
+	is_drop_n_account = true;
24182457 
24192458 	sk->sk_data_ready(sk);
24202459 	kfree_skb(copy_skb);
....@@ -2440,15 +2479,6 @@
24402479 	}
24412480 
24422481 	sock_wfree(skb);
2443
-}
2444
-
2445
-static void tpacket_set_protocol(const struct net_device *dev,
2446
-				 struct sk_buff *skb)
2447
-{
2448
-	if (dev->type == ARPHRD_ETHER) {
2449
-		skb_reset_mac_header(skb);
2450
-		skb->protocol = eth_hdr(skb)->h_proto;
2451
-	}
24522482 }
24532483 
24542484 static int __packet_snd_vnet_parse(struct virtio_net_hdr *vnet_hdr, size_t len)
....@@ -2498,7 +2528,7 @@
24982528 	skb->priority = po->sk.sk_priority;
24992529 	skb->mark = po->sk.sk_mark;
25002530 	skb->tstamp = sockc->transmit_time;
2501
-	sock_tx_timestamp(&po->sk, sockc->tsflags, &skb_shinfo(skb)->tx_flags);
2531
+	skb_setup_tx_timestamp(skb, sockc->tsflags);
25022532 	skb_zcopy_set_nouarg(skb, ph.raw);
25032533 
25042534 	skb_reserve(skb, hlen);
....@@ -2521,8 +2551,6 @@
25212551 			return err;
25222552 		if (!dev_validate_header(dev, skb->data, hdrlen))
25232553 			return -EINVAL;
2524
-		if (!skb->protocol)
2525
-			tpacket_set_protocol(dev, skb);
25262554 
25272555 		data += hdrlen;
25282556 		to_write -= hdrlen;
....@@ -2557,7 +2585,7 @@
25572585 		len = ((to_write > len_max) ? len_max : to_write);
25582586 	}
25592587 
2560
-	skb_probe_transport_header(skb, 0);
2588
+	packet_parse_headers(skb, sock);
25612589 
25622590 	return tp_len;
25632591 }
....@@ -2788,8 +2816,9 @@
27882816 
27892817 		status = TP_STATUS_SEND_REQUEST;
27902818 		err = po->xmit(skb);
2791
-		if (unlikely(err > 0)) {
2792
-			err = net_xmit_errno(err);
2819
+		if (unlikely(err != 0)) {
2820
+			if (err > 0)
2821
+				err = net_xmit_errno(err);
27932822 			if (err && __packet_get_status(po, ph) ==
27942823 				   TP_STATUS_AVAILABLE) {
27952824 				/* skb was destructed already */
....@@ -2956,13 +2985,13 @@
29562985 	if (err)
29572986 		goto out_free;
29582987 
2959
-	if (sock->type == SOCK_RAW &&
2960
-	    !dev_validate_header(dev, skb->data, len)) {
2988
+	if ((sock->type == SOCK_RAW &&
2989
+	     !dev_validate_header(dev, skb->data, len)) || !skb->len) {
29612990 		err = -EINVAL;
29622991 		goto out_free;
29632992 	}
29642993 
2965
-	sock_tx_timestamp(sk, sockc.tsflags, &skb_shinfo(skb)->tx_flags);
2994
+	skb_setup_tx_timestamp(skb, sockc.tsflags);
29662995 
29672996 	if (!vnet_hdr.gso_type && (len > dev->mtu + reserve + extra_len) &&
29682997 	    !packet_extra_vlan_len_allowed(dev, skb)) {
....@@ -2984,14 +3013,18 @@
29843013 		virtio_net_hdr_set_proto(skb, &vnet_hdr);
29853014 	}
29863015 
2987
-	skb_probe_transport_header(skb, reserve);
3016
+	packet_parse_headers(skb, sock);
29883017 
29893018 	if (unlikely(extra_len == 4))
29903019 		skb->no_fcs = 1;
29913020 
29923021 	err = po->xmit(skb);
2993
-	if (err > 0 && (err = net_xmit_errno(err)) != 0)
2994
-		goto out_unlock;
3022
+	if (unlikely(err != 0)) {
3023
+		if (err > 0)
3024
+			err = net_xmit_errno(err);
3025
+		if (err)
3026
+			goto out_unlock;
3027
+	}
29953028 
29963029 	dev_put(dev);
29973030 
....@@ -3011,10 +3044,13 @@
30113044 	struct sock *sk = sock->sk;
30123045 	struct packet_sock *po = pkt_sk(sk);
30133046 
3014
-	if (po->tx_ring.pg_vec)
3047
+	/* Reading tx_ring.pg_vec without holding pg_vec_lock is racy.
3048
+	 * tpacket_snd() will redo the check safely.
3049
+	 */
3050
+	if (data_race(po->tx_ring.pg_vec))
30153051 		return tpacket_snd(po, msg);
3016
-	else
3017
-		return packet_snd(sock, msg, len);
3052
+
3053
+	return packet_snd(sock, msg, len);
30183054 }
30193055 
30203056 /*
....@@ -3075,7 +3111,7 @@
30753111 	kfree(po->rollover);
30763112 	if (f) {
30773113 		fanout_release_data(f);
3078
-		kfree(f);
3114
+		kvfree(f);
30793115 	}
30803116 	/*
30813117 	 *	Now the socket is dead. No more input will appear.
....@@ -3370,8 +3406,7 @@
33703406 	if (skb == NULL)
33713407 		goto out;
33723408 
3373
-	if (pkt_sk(sk)->pressure)
3374
-		packet_rcv_has_room(pkt_sk(sk), NULL);
3409
+	packet_rcv_try_clear_pressure(pkt_sk(sk));
33753410 
33763411 	if (pkt_sk(sk)->has_vnet_hdr) {
33773412 		err = packet_rcv_vnet(msg, skb, &len);
....@@ -3406,6 +3441,8 @@
34063441 	sock_recv_ts_and_drops(msg, sk, skb);
34073442 
34083443 	if (msg->msg_name) {
3444
+		const size_t max_len = min(sizeof(skb->cb),
3445
+					   sizeof(struct sockaddr_storage));
34093446 		int copy_len;
34103447 
34113448 		/* If the address length field is there to be filled
....@@ -3428,6 +3465,10 @@
34283465 				msg->msg_namelen = sizeof(struct sockaddr_ll);
34293466 			}
34303467 		}
3468
+		if (WARN_ON_ONCE(copy_len > max_len)) {
3469
+			copy_len = max_len;
3470
+			msg->msg_namelen = copy_len;
3471
+		}
34313472 		memcpy(msg->msg_name, &PACKET_SKB_CB(skb)->sa, copy_len);
34323473 	}
34333474 
....@@ -3438,8 +3479,7 @@
34383479 		if (skb->ip_summed == CHECKSUM_PARTIAL)
34393480 			aux.tp_status |= TP_STATUS_CSUMNOTREADY;
34403481 		else if (skb->pkt_type != PACKET_OUTGOING &&
3441
-			 (skb->ip_summed == CHECKSUM_COMPLETE ||
3442
-			  skb_csum_unnecessary(skb)))
3482
+			 skb_csum_unnecessary(skb))
34433483 			aux.tp_status |= TP_STATUS_CSUM_VALID;
34443484 
34453485 		aux.tp_len = origlen;
....@@ -3669,7 +3709,8 @@
36693709 }
36703710 
36713711 static int
3672
-packet_setsockopt(struct socket *sock, int level, int optname, char __user *optval, unsigned int optlen)
3712
+packet_setsockopt(struct socket *sock, int level, int optname, sockptr_t optval,
3713
+		  unsigned int optlen)
36733714 {
36743715 	struct sock *sk = sock->sk;
36753716 	struct packet_sock *po = pkt_sk(sk);
....@@ -3689,7 +3730,7 @@
36893730 			return -EINVAL;
36903731 		if (len > sizeof(mreq))
36913732 			len = sizeof(mreq);
3692
-		if (copy_from_user(&mreq, optval, len))
3733
+		if (copy_from_sockptr(&mreq, optval, len))
36933734 			return -EFAULT;
36943735 		if (len < (mreq.mr_alen + offsetof(struct packet_mreq, mr_address)))
36953736 			return -EINVAL;
....@@ -3720,7 +3761,7 @@
37203761 		if (optlen < len) {
37213762 			ret = -EINVAL;
37223763 		} else {
3723
-			if (copy_from_user(&req_u.req, optval, len))
3764
+			if (copy_from_sockptr(&req_u.req, optval, len))
37243765 				ret = -EFAULT;
37253766 			else
37263767 				ret = packet_set_ring(sk, &req_u, 0,
....@@ -3735,7 +3776,7 @@
37353776 
37363777 		if (optlen != sizeof(val))
37373778 			return -EINVAL;
3738
-		if (copy_from_user(&val, optval, sizeof(val)))
3779
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
37393780 			return -EFAULT;
37403781 
37413782 		pkt_sk(sk)->copy_thresh = val;
....@@ -3747,7 +3788,7 @@
37473788 
37483789 		if (optlen != sizeof(val))
37493790 			return -EINVAL;
3750
-		if (copy_from_user(&val, optval, sizeof(val)))
3791
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
37513792 			return -EFAULT;
37523793 		switch (val) {
37533794 		case TPACKET_V1:
....@@ -3773,7 +3814,7 @@
37733814 
37743815 		if (optlen != sizeof(val))
37753816 			return -EINVAL;
3776
-		if (copy_from_user(&val, optval, sizeof(val)))
3817
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
37773818 			return -EFAULT;
37783819 		if (val > INT_MAX)
37793820 			return -EINVAL;
....@@ -3793,7 +3834,7 @@
37933834 
37943835 		if (optlen != sizeof(val))
37953836 			return -EINVAL;
3796
-		if (copy_from_user(&val, optval, sizeof(val)))
3837
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
37973838 			return -EFAULT;
37983839 
37993840 		lock_sock(sk);
....@@ -3812,7 +3853,7 @@
38123853 
38133854 		if (optlen < sizeof(val))
38143855 			return -EINVAL;
3815
-		if (copy_from_user(&val, optval, sizeof(val)))
3856
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
38163857 			return -EFAULT;
38173858 
38183859 		lock_sock(sk);
....@@ -3826,7 +3867,7 @@
38263867 
38273868 		if (optlen < sizeof(val))
38283869 			return -EINVAL;
3829
-		if (copy_from_user(&val, optval, sizeof(val)))
3870
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
38303871 			return -EFAULT;
38313872 
38323873 		lock_sock(sk);
....@@ -3842,7 +3883,7 @@
38423883 			return -EINVAL;
38433884 		if (optlen < sizeof(val))
38443885 			return -EINVAL;
3845
-		if (copy_from_user(&val, optval, sizeof(val)))
3886
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
38463887 			return -EFAULT;
38473888 
38483889 		lock_sock(sk);
....@@ -3861,7 +3902,7 @@
38613902 
38623903 		if (optlen != sizeof(val))
38633904 			return -EINVAL;
3864
-		if (copy_from_user(&val, optval, sizeof(val)))
3905
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
38653906 			return -EFAULT;
38663907 
38673908 		po->tp_tstamp = val;
....@@ -3869,14 +3910,14 @@
38693910 	}
38703911 	case PACKET_FANOUT:
38713912 	{
3872
-		int val;
3913
+		struct fanout_args args = { 0 };
38733914 
3874
-		if (optlen != sizeof(val))
3915
+		if (optlen != sizeof(int) && optlen != sizeof(args))
38753916 			return -EINVAL;
3876
-		if (copy_from_user(&val, optval, sizeof(val)))
3917
+		if (copy_from_sockptr(&args, optval, optlen))
38773918 			return -EFAULT;
38783919 
3879
-		return fanout_add(sk, val & 0xffff, val >> 16);
3920
+		return fanout_add(sk, &args);
38803921 	}
38813922 	case PACKET_FANOUT_DATA:
38823923 	{
....@@ -3886,13 +3927,27 @@
38863927 
38873928 		return fanout_set_data(po, optval, optlen);
38883929 	}
3930
+	case PACKET_IGNORE_OUTGOING:
3931
+	{
3932
+		int val;
3933
+
3934
+		if (optlen != sizeof(val))
3935
+			return -EINVAL;
3936
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
3937
+			return -EFAULT;
3938
+		if (val < 0 || val > 1)
3939
+			return -EINVAL;
3940
+
3941
+		po->prot_hook.ignore_outgoing = !!val;
3942
+		return 0;
3943
+	}
38893944 	case PACKET_TX_HAS_OFF:
38903945 	{
38913946 		unsigned int val;
38923947 
38933948 		if (optlen != sizeof(val))
38943949 			return -EINVAL;
3895
-		if (copy_from_user(&val, optval, sizeof(val)))
3950
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
38963951 			return -EFAULT;
38973952 
38983953 		lock_sock(sk);
....@@ -3911,7 +3966,7 @@
39113966 
39123967 		if (optlen != sizeof(val))
39133968 			return -EINVAL;
3914
-		if (copy_from_user(&val, optval, sizeof(val)))
3969
+		if (copy_from_sockptr(&val, optval, sizeof(val)))
39153970 			return -EFAULT;
39163971 
39173972 		po->xmit = val ? packet_direct_xmit : dev_queue_xmit;
....@@ -3932,6 +3987,7 @@
39323987 	void *data = &val;
39333988 	union tpacket_stats_u st;
39343989 	struct tpacket_rollover_stats rstats;
3990
+	int drops;
39353991 
39363992 	if (level != SOL_PACKET)
39373993 		return -ENOPROTOOPT;
....@@ -3948,14 +4004,17 @@
39484004 		memcpy(&st, &po->stats, sizeof(st));
39494005 		memset(&po->stats, 0, sizeof(po->stats));
39504006 		spin_unlock_bh(&sk->sk_receive_queue.lock);
4007
+		drops = atomic_xchg(&po->tp_drops, 0);
39514008 
39524009 		if (po->tp_version == TPACKET_V3) {
39534010 			lv = sizeof(struct tpacket_stats_v3);
3954
-			st.stats3.tp_packets += st.stats3.tp_drops;
4011
+			st.stats3.tp_drops = drops;
4012
+			st.stats3.tp_packets += drops;
39554013 			data = &st.stats3;
39564014 		} else {
39574015 			lv = sizeof(struct tpacket_stats);
3958
-			st.stats1.tp_packets += st.stats1.tp_drops;
4016
+			st.stats1.tp_drops = drops;
4017
+			st.stats1.tp_packets += drops;
39594018 			data = &st.stats1;
39604019 		}
39614020 
....@@ -4009,6 +4068,9 @@
40094068 			((u32)po->fanout->flags << 24)) :
40104069 		       0);
40114070 		break;
4071
+	case PACKET_IGNORE_OUTGOING:
4072
+		val = po->prot_hook.ignore_outgoing;
4073
+		break;
40124074 	case PACKET_ROLLOVER_STATS:
40134075 		if (!po->rollover)
40144076 			return -EINVAL;
....@@ -4037,28 +4099,6 @@
40374099 	return 0;
40384100 }
40394101 
4040
-
4041
-#ifdef CONFIG_COMPAT
4042
-static int compat_packet_setsockopt(struct socket *sock, int level, int optname,
4043
-				    char __user *optval, unsigned int optlen)
4044
-{
4045
-	struct packet_sock *po = pkt_sk(sock->sk);
4046
-
4047
-	if (level != SOL_PACKET)
4048
-		return -ENOPROTOOPT;
4049
-
4050
-	if (optname == PACKET_FANOUT_DATA &&
4051
-	    po->fanout && po->fanout->type == PACKET_FANOUT_CBPF) {
4052
-		optval = (char __user *)get_compat_bpf_fprog(optval);
4053
-		if (!optval)
4054
-			return -EFAULT;
4055
-		optlen = sizeof(struct sock_fprog);
4056
-	}
4057
-
4058
-	return packet_setsockopt(sock, level, optname, optval, optlen);
4059
-}
4060
-#endif
4061
-
40624102 static int packet_notifier(struct notifier_block *this,
40634103 			   unsigned long msg, void *ptr)
40644104 {
....@@ -4074,7 +4114,7 @@
40744114 		case NETDEV_UNREGISTER:
40754115 			if (po->mclist)
40764116 				packet_dev_mclist_delete(dev, &po->mclist);
4077
-			/* fallthrough */
4117
+			fallthrough;
40784118 
40794119 		case NETDEV_DOWN:
40804120 			if (dev->ifindex == po->ifindex) {
....@@ -4134,11 +4174,6 @@
41344174 		spin_unlock_bh(&sk->sk_receive_queue.lock);
41354175 		return put_user(amount, (int __user *)arg);
41364176 	}
4137
-	case SIOCGSTAMP:
4138
-		return sock_get_timestamp(sk, (struct timeval __user *)arg);
4139
-	case SIOCGSTAMPNS:
4140
-		return sock_get_timestampns(sk, (struct timespec __user *)arg);
4141
-
41424177 #ifdef CONFIG_INET
41434178 	case SIOCADDRT:
41444179 	case SIOCDELRT:
....@@ -4176,8 +4211,7 @@
41764211 			TP_STATUS_KERNEL))
41774212 			mask |= EPOLLIN | EPOLLRDNORM;
41784213 	}
4179
-	if (po->pressure && __packet_rcv_has_room(po, NULL) == ROOM_NORMAL)
4180
-		po->pressure = 0;
4214
+	packet_rcv_try_clear_pressure(po);
41814215 	spin_unlock_bh(&sk->sk_receive_queue.lock);
41824216 	spin_lock_bh(&sk->sk_write_queue.lock);
41834217 	if (po->tx_ring.pg_vec) {
....@@ -4296,7 +4330,7 @@
42964330 	struct packet_ring_buffer *rb;
42974331 	struct sk_buff_head *rb_queue;
42984332 	__be16 num;
4299
-	int err = -EINVAL;
4333
+	int err;
43004334 	/* Added to avoid minimal code churn */
43014335 	struct tpacket_req *req = &req_u->req;
43024336 
....@@ -4526,10 +4560,9 @@
45264560 	.getname =	packet_getname_spkt,
45274561 	.poll =		datagram_poll,
45284562 	.ioctl =	packet_ioctl,
4563
+	.gettstamp =	sock_gettstamp,
45294564 	.listen =	sock_no_listen,
45304565 	.shutdown =	sock_no_shutdown,
4531
-	.setsockopt =	sock_no_setsockopt,
4532
-	.getsockopt =	sock_no_getsockopt,
45334566 	.sendmsg =	packet_sendmsg_spkt,
45344567 	.recvmsg =	packet_recvmsg,
45354568 	.mmap =		sock_no_mmap,
....@@ -4547,13 +4580,11 @@
45474580 	.getname =	packet_getname,
45484581 	.poll =		packet_poll,
45494582 	.ioctl =	packet_ioctl,
4583
+	.gettstamp =	sock_gettstamp,
45504584 	.listen =	sock_no_listen,
45514585 	.shutdown =	sock_no_shutdown,
45524586 	.setsockopt =	packet_setsockopt,
45534587 	.getsockopt =	packet_getsockopt,
4554
-#ifdef CONFIG_COMPAT
4555
-	.compat_setsockopt = compat_packet_setsockopt,
4556
-#endif
45574588 	.sendmsg =	packet_sendmsg,
45584589 	.recvmsg =	packet_recvmsg,
45594590 	.mmap =		packet_mmap,
....@@ -4630,9 +4661,11 @@
46304661 	mutex_init(&net->packet.sklist_lock);
46314662 	INIT_HLIST_HEAD(&net->packet.sklist);
46324663 
4664
+#ifdef CONFIG_PROC_FS
46334665 	if (!proc_create_net("packet", 0, net->proc_net, &packet_seq_ops,
46344666 			sizeof(struct seq_net_private)))
46354667 		return -ENOMEM;
4668
+#endif /* CONFIG_PROC_FS */
46364669 
46374670 	return 0;
46384671 }