Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Include fixes for netrom and dsa (Fabian Frederick and Florian
    Fainelli)

 2) Fix FIXED_PHY support in stmmac, from Giuseppe CAVALLARO.

 3) Several SKB use after free fixes (vxlan, openvswitch, vxlan,
    ip_tunnel, fou), from Li ROngQing.

 4) fec driver PTP support fixes from Luwei Zhou and Nimrod Andy.

 5) Use after free in virtio_net, from Michael S Tsirkin.

 6) Fix flow mask handling for megaflows in openvswitch, from Pravin B
    Shelar.

 7) ISDN gigaset and capi bug fixes from Tilman Schmidt.

 8) Fix route leak in ip_send_unicast_reply(), from Vasily Averin.

 9) Fix two eBPF JIT bugs on x86, from Alexei Starovoitov.

10) TCP_SKB_CB() reorganization caused a few regressions, fixed by Cong
    Wang and Eric Dumazet.

11) Don't overwrite end of SKB when parsing malformed sctp ASCONF
    chunks, from Daniel Borkmann.

12) Don't call sock_kfree_s() with NULL pointers, this function also has
    the side effect of adjusting the socket memory usage.  From Cong Wang.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (90 commits)
  bna: fix skb->truesize underestimation
  net: dsa: add includes for ethtool and phy_fixed definitions
  openvswitch: Set flow-key members.
  netrom: use linux/uaccess.h
  dsa: Fix conversion from host device to mii bus
  tipc: fix bug in bundled buffer reception
  ipv6: introduce tcp_v6_iif()
  sfc: add support for skb->xmit_more
  r8152: return -EBUSY for runtime suspend
  ipv4: fix a potential use after free in fou.c
  ipv4: fix a potential use after free in ip_tunnel_core.c
  hyperv: Add handling of IP header with option field in netvsc_set_hash()
  openvswitch: Create right mask with disabled megaflows
  vxlan: fix a free after use
  openvswitch: fix a use after free
  ipv4: dst_entry leak in ip_send_unicast_reply()
  ipv4: clean up cookie_v4_check()
  ipv4: share tcp_v4_save_options() with cookie_v4_check()
  ipv4: call __ip_options_echo() in cookie_v4_check()
  atm: simplify lanai.c by using module_pci_driver
  ...

net/ipv4/fib_semantics.c

@@ -537,7 +537,7 @@ int fib_nh_match(struct fib_config *cfg, struct fib_info *fi)
 			return 1;
 
 		attrlen = rtnh_attrlen(rtnh);
-		if (attrlen < 0) {
+		if (attrlen > 0) {
 			struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
 
 			nla = nla_find(attrs, attrlen, RTA_GATEWAY);

net/ipv4/fou.c

@@ -87,6 +87,9 @@ static int gue_udp_recv(struct sock *sk, struct sk_buff *skb)
 	if (!pskb_may_pull(skb, len))
 		goto drop;
 
+	uh = udp_hdr(skb);
+	guehdr = (struct guehdr *)&uh[1];
+
 	if (guehdr->version != 0)
 		goto drop;
 

net/ipv4/ip_output.c

@@ -1535,6 +1535,7 @@ void ip_send_unicast_reply(struct net *net, struct sk_buff *skb,
 	struct sk_buff *nskb;
 	struct sock *sk;
 	struct inet_sock *inet;
+	int err;
 
 	if (__ip_options_echo(&replyopts.opt.opt, skb, sopt))
 		return;
@@ -1574,8 +1575,13 @@ void ip_send_unicast_reply(struct net *net, struct sk_buff *skb,
 	sock_net_set(sk, net);
 	__skb_queue_head_init(&sk->sk_write_queue);
 	sk->sk_sndbuf = sysctl_wmem_default;
-	ip_append_data(sk, &fl4, ip_reply_glue_bits, arg->iov->iov_base, len, 0,
-		       &ipc, &rt, MSG_DONTWAIT);
+	err = ip_append_data(sk, &fl4, ip_reply_glue_bits, arg->iov->iov_base,
+			     len, 0, &ipc, &rt, MSG_DONTWAIT);
+	if (unlikely(err)) {
+		ip_flush_pending_frames(sk);
+		goto out;
+	}
+
 	nskb = skb_peek(&sk->sk_write_queue);
 	if (nskb) {
 		if (arg->csumoffset >= 0)
@@ -1587,7 +1593,7 @@ void ip_send_unicast_reply(struct net *net, struct sk_buff *skb,
 		skb_set_queue_mapping(nskb, skb_get_queue_mapping(skb));
 		ip_push_pending_frames(sk, &fl4);
 	}
-
+out:
 	put_cpu_var(unicast_sock);
 
 	ip_rt_put(rt);

net/ipv4/ip_tunnel_core.c

@@ -91,11 +91,12 @@ int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto)
 	skb_pull_rcsum(skb, hdr_len);
 
 	if (inner_proto == htons(ETH_P_TEB)) {
-		struct ethhdr *eh = (struct ethhdr *)skb->data;
+		struct ethhdr *eh;
 
 		if (unlikely(!pskb_may_pull(skb, ETH_HLEN)))
 			return -ENOMEM;
 
+		eh = (struct ethhdr *)skb->data;
 		if (likely(ntohs(eh->h_proto) >= ETH_P_802_3_MIN))
 			skb->protocol = eh->h_proto;
 		else

net/ipv4/syncookies.c

@@ -255,9 +255,9 @@ bool cookie_check_timestamp(struct tcp_options_received *tcp_opt,
 }
 EXPORT_SYMBOL(cookie_check_timestamp);
 
-struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
-			     struct ip_options *opt)
+struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 {
+	struct ip_options *opt = &TCP_SKB_CB(skb)->header.h4.opt;
 	struct tcp_options_received tcp_opt;
 	struct inet_request_sock *ireq;
 	struct tcp_request_sock *treq;
@@ -317,15 +317,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 	/* We throwed the options of the initial SYN away, so we hope
 	 * the ACK carries the same options again (see RFC1122 4.2.3.8)
 	 */
-	if (opt && opt->optlen) {
-		int opt_size = sizeof(struct ip_options_rcu) + opt->optlen;
-
-		ireq->opt = kmalloc(opt_size, GFP_ATOMIC);
-		if (ireq->opt != NULL && ip_options_echo(&ireq->opt->opt, skb)) {
-			kfree(ireq->opt);
-			ireq->opt = NULL;
-		}
-	}
+	ireq->opt = tcp_v4_save_options(skb);
 
 	if (security_inet_conn_request(sk, skb, req)) {
 		reqsk_free(req);
@@ -344,7 +336,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 	flowi4_init_output(&fl4, sk->sk_bound_dev_if, ireq->ir_mark,
 			   RT_CONN_FLAGS(sk), RT_SCOPE_UNIVERSE, IPPROTO_TCP,
 			   inet_sk_flowi_flags(sk),
-			   (opt && opt->srr) ? opt->faddr : ireq->ir_rmt_addr,
+			   opt->srr ? opt->faddr : ireq->ir_rmt_addr,
 			   ireq->ir_loc_addr, th->source, th->dest);
 	security_req_classify_flow(req, flowi4_to_flowi(&fl4));
 	rt = ip_route_output_key(sock_net(sk), &fl4);

net/ipv4/tcp_input.c

@@ -68,6 +68,7 @@
 #include <linux/module.h>
 #include <linux/sysctl.h>
 #include <linux/kernel.h>
+#include <linux/prefetch.h>
 #include <net/dst.h>
 #include <net/tcp.h>
 #include <net/inet_common.h>
@@ -3029,6 +3030,21 @@ static u32 tcp_tso_acked(struct sock *sk, struct sk_buff *skb)
 	return packets_acked;
 }
 
+static void tcp_ack_tstamp(struct sock *sk, struct sk_buff *skb,
+			   u32 prior_snd_una)
+{
+	const struct skb_shared_info *shinfo;
+
+	/* Avoid cache line misses to get skb_shinfo() and shinfo->tx_flags */
+	if (likely(!(sk->sk_tsflags & SOF_TIMESTAMPING_TX_ACK)))
+		return;
+
+	shinfo = skb_shinfo(skb);
+	if ((shinfo->tx_flags & SKBTX_ACK_TSTAMP) &&
+	    between(shinfo->tskey, prior_snd_una, tcp_sk(sk)->snd_una - 1))
+		__skb_tstamp_tx(skb, NULL, sk, SCM_TSTAMP_ACK);
+}
+
 /* Remove acknowledged frames from the retransmission queue. If our packet
  * is before the ack sequence we can discard it as it's confirmed to have
  * arrived at the other end.
@@ -3052,14 +3068,11 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	first_ackt.v64 = 0;
 
 	while ((skb = tcp_write_queue_head(sk)) && skb != tcp_send_head(sk)) {
-		struct skb_shared_info *shinfo = skb_shinfo(skb);
 		struct tcp_skb_cb *scb = TCP_SKB_CB(skb);
 		u8 sacked = scb->sacked;
 		u32 acked_pcount;
 
-		if (unlikely(shinfo->tx_flags & SKBTX_ACK_TSTAMP) &&
-		    between(shinfo->tskey, prior_snd_una, tp->snd_una - 1))
-			__skb_tstamp_tx(skb, NULL, sk, SCM_TSTAMP_ACK);
+		tcp_ack_tstamp(sk, skb, prior_snd_una);
 
 		/* Determine how many packets and what bytes were acked, tso and else */
 		if (after(scb->end_seq, tp->snd_una)) {
@@ -3073,10 +3086,12 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 
 			fully_acked = false;
 		} else {
+			/* Speedup tcp_unlink_write_queue() and next loop */
+			prefetchw(skb->next);
 			acked_pcount = tcp_skb_pcount(skb);
 		}
 
-		if (sacked & TCPCB_RETRANS) {
+		if (unlikely(sacked & TCPCB_RETRANS)) {
 			if (sacked & TCPCB_SACKED_RETRANS)
 				tp->retrans_out -= acked_pcount;
 			flag |= FLAG_RETRANS_DATA_ACKED;
@@ -3107,7 +3122,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 		 * connection startup slow start one packet too
 		 * quickly.  This is severely frowned upon behavior.
 		 */
-		if (!(scb->tcp_flags & TCPHDR_SYN)) {
+		if (likely(!(scb->tcp_flags & TCPHDR_SYN))) {
 			flag |= FLAG_DATA_ACKED;
 		} else {
 			flag |= FLAG_SYN_ACKED;
@@ -3119,9 +3134,9 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 
 		tcp_unlink_write_queue(skb, sk);
 		sk_wmem_free_skb(sk, skb);
-		if (skb == tp->retransmit_skb_hint)
+		if (unlikely(skb == tp->retransmit_skb_hint))
 			tp->retransmit_skb_hint = NULL;
-		if (skb == tp->lost_skb_hint)
+		if (unlikely(skb == tp->lost_skb_hint))
 			tp->lost_skb_hint = NULL;
 	}
 
@@ -3132,7 +3147,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 		flag |= FLAG_SACK_RENEGING;
 
 	skb_mstamp_get(&now);
-	if (first_ackt.v64) {
+	if (likely(first_ackt.v64)) {
 		seq_rtt_us = skb_mstamp_us_delta(&now, &first_ackt);
 		ca_seq_rtt_us = skb_mstamp_us_delta(&now, &last_ackt);
 	}
@@ -3394,6 +3409,9 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	int acked = 0; /* Number of packets newly acked */
 	long sack_rtt_us = -1L;
 
+	/* We very likely will need to access write queue head. */
+	prefetchw(sk->sk_write_queue.next);
+
 	/* If the ack is older than previous acks
 	 * then we can probably ignore it.
 	 */

net/ipv4/tcp_ipv4.c

@@ -880,26 +880,6 @@ bool tcp_syn_flood_action(struct sock *sk,
 }
 EXPORT_SYMBOL(tcp_syn_flood_action);
 
-/*
- * Save and compile IPv4 options into the request_sock if needed.
- */
-static struct ip_options_rcu *tcp_v4_save_options(struct sk_buff *skb)
-{
-	const struct ip_options *opt = &TCP_SKB_CB(skb)->header.h4.opt;
-	struct ip_options_rcu *dopt = NULL;
-
-	if (opt && opt->optlen) {
-		int opt_size = sizeof(*dopt) + opt->optlen;
-
-		dopt = kmalloc(opt_size, GFP_ATOMIC);
-		if (dopt && __ip_options_echo(&dopt->opt, skb, opt)) {
-			kfree(dopt);
-			dopt = NULL;
-		}
-	}
-	return dopt;
-}
-
 #ifdef CONFIG_TCP_MD5SIG
 /*
  * RFC2385 MD5 checksumming requires a mapping of
@@ -1428,7 +1408,7 @@ static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
 
 #ifdef CONFIG_SYN_COOKIES
 	if (!th->syn)
-		sk = cookie_v4_check(sk, skb, &TCP_SKB_CB(skb)->header.h4.opt);
+		sk = cookie_v4_check(sk, skb);
 #endif
 	return sk;
 }

net/ipv4/tcp_output.c

@@ -839,26 +839,38 @@ void tcp_wfree(struct sk_buff *skb)
 {
 	struct sock *sk = skb->sk;
 	struct tcp_sock *tp = tcp_sk(sk);
+	int wmem;
+
+	/* Keep one reference on sk_wmem_alloc.
+	 * Will be released by sk_free() from here or tcp_tasklet_func()
+	 */
+	wmem = atomic_sub_return(skb->truesize - 1, &sk->sk_wmem_alloc);
+
+	/* If this softirq is serviced by ksoftirqd, we are likely under stress.
+	 * Wait until our queues (qdisc + devices) are drained.
+	 * This gives :
+	 * - less callbacks to tcp_write_xmit(), reducing stress (batches)
+	 * - chance for incoming ACK (processed by another cpu maybe)
+	 *   to migrate this flow (skb->ooo_okay will be eventually set)
+	 */
+	if (wmem >= SKB_TRUESIZE(1) && this_cpu_ksoftirqd() == current)
+		goto out;
 
 	if (test_and_clear_bit(TSQ_THROTTLED, &tp->tsq_flags) &&
 	    !test_and_set_bit(TSQ_QUEUED, &tp->tsq_flags)) {
 		unsigned long flags;
 		struct tsq_tasklet *tsq;
 
-		/* Keep a ref on socket.
-		 * This last ref will be released in tcp_tasklet_func()
-		 */
-		atomic_sub(skb->truesize - 1, &sk->sk_wmem_alloc);
-
 		/* queue this socket to tasklet queue */
 		local_irq_save(flags);
 		tsq = this_cpu_ptr(&tsq_tasklet);
 		list_add(&tp->tsq_node, &tsq->head);
 		tasklet_schedule(&tsq->tasklet);
 		local_irq_restore(flags);
-	} else {
-		sock_wfree(skb);
+		return;
 	}
+out:
+	sk_free(sk);
 }
 
 /* This routine actually transmits TCP packets queued in by
@@ -914,9 +926,13 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
 		tcp_ca_event(sk, CA_EVENT_TX_START);
 
 	/* if no packet is in qdisc/device queue, then allow XPS to select
-	 * another queue.
+	 * another queue. We can be called from tcp_tsq_handler()
+	 * which holds one reference to sk_wmem_alloc.
+	 *
+	 * TODO: Ideally, in-flight pure ACK packets should not matter here.
+	 * One way to get this would be to set skb->truesize = 2 on them.
 	 */
-	skb->ooo_okay = sk_wmem_alloc_get(sk) == 0;
+	skb->ooo_okay = sk_wmem_alloc_get(sk) < SKB_TRUESIZE(1);
 
 	skb_push(skb, tcp_header_size);
 	skb_reset_transport_header(skb);