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

Pull networking fixes from David Miller:
 "This pull request is dedicated to the upcoming snowpocalypse parts 2
  and 3 in the Pacific Northwest:

   1) Drop profiles are broken because some drivers use dev_kfree_skb*
      instead of dev_consume_skb*, from Yang Wei.

   2) Fix IWLWIFI kconfig deps, from Luca Coelho.

   3) Fix percpu maps updating in bpftool, from Paolo Abeni.

   4) Missing station release in batman-adv, from Felix Fietkau.

   5) Fix some networking compat ioctl bugs, from Johannes Berg.

   6) ucc_geth must reset the BQL queue state when stopping the device,
      from Mathias Thore.

   7) Several XDP bug fixes in virtio_net from Toshiaki Makita.

   8) TSO packets must be sent always on queue 0 in stmmac, from Jose
      Abreu.

   9) Fix socket refcounting bug in RDS, from Eric Dumazet.

  10) Handle sparse cpu allocations in bpf selftests, from Martynas
      Pumputis.

  11) Make sure mgmt frames have enough tailroom in mac80211, from Felix
      Feitkau.

  12) Use safe list walking in sctp_sendmsg() asoc list traversal, from
      Greg Kroah-Hartman.

  13) Make DCCP's ccid_hc_[rt]x_parse_options always check for NULL
      ccid, from Eric Dumazet.

  14) Need to reload WoL password into bcmsysport device after deep
      sleeps, from Florian Fainelli.

  15) Remove filter from mask before freeing in cls_flower, from Petr
      Machata.

  16) Missing release and use after free in error paths of s390 qeth
      code, from Julian Wiedmann.

  17) Fix lockdep false positive in dsa code, from Marc Zyngier.

  18) Fix counting of ATU violations in mv88e6xxx, from Andrew Lunn.

  19) Fix EQ firmware assert in qed driver, from Manish Chopra.

  20) Don't default Caivum PTP to Y in kconfig, from Bjorn Helgaas"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (116 commits)
  net: dsa: b53: Fix for failure when irq is not defined in dt
  sit: check if IPv6 enabled before calling ip6_err_gen_icmpv6_unreach()
  geneve: should not call rt6_lookup() when ipv6 was disabled
  net: Don't default Cavium PTP driver to 'y'
  net: broadcom: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: via-velocity: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: tehuti: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: sun: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: fsl_ucc_hdlc: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: fec_mpc52xx: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: smsc: epic100: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: dscc4: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: tulip: de2104x: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net: defxx: replace dev_kfree_skb_irq by dev_consume_skb_irq for drop profiles
  net/mlx5e: Don't overwrite pedit action when multiple pedit used
  net/mlx5e: Update hw flows when encap source mac changed
  qed*: Advance drivers version to 8.37.0.20
  qed: Change verbosity for coalescing message.
  qede: Fix system crash on configuring channels.
  qed: Consider TX tcs while deriving the max num_queues for PF.
  ...

1 files changed, 31 insertions, 33 deletions

diff --git a/net/smc/smc_tx.c b/net/smc/smc_tx.c
index d8366ed51757..f93f3580c100 100644
--- a/net/smc/smc_tx.c
+++ b/net/smc/smc_tx.c
@@ -165,12 +165,11 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
 			conn->local_tx_ctrl.prod_flags.urg_data_pending = 1;
 
 		if (!atomic_read(&conn->sndbuf_space) || conn->urg_tx_pend) {
+			if (send_done)
+				return send_done;
 			rc = smc_tx_wait(smc, msg->msg_flags);
-			if (rc) {
-				if (send_done)
-					return send_done;
+			if (rc)
 				goto out_err;
-			}
 			continue;
 		}
 
@@ -267,27 +266,23 @@ int smcd_tx_ism_write(struct smc_connection *conn, void *data, size_t len,
 
 /* sndbuf consumer: actual data transfer of one target chunk with RDMA write */
 static int smc_tx_rdma_write(struct smc_connection *conn, int peer_rmbe_offset,
-			     int num_sges, struct ib_sge sges[])
+			     int num_sges, struct ib_rdma_wr *rdma_wr)
 {
 	struct smc_link_group *lgr = conn->lgr;
-	struct ib_rdma_wr rdma_wr;
 	struct smc_link *link;
 	int rc;
 
-	memset(&rdma_wr, 0, sizeof(rdma_wr));
 	link = &lgr->lnk[SMC_SINGLE_LINK];
-	rdma_wr.wr.wr_id = smc_wr_tx_get_next_wr_id(link);
-	rdma_wr.wr.sg_list = sges;
-	rdma_wr.wr.num_sge = num_sges;
-	rdma_wr.wr.opcode = IB_WR_RDMA_WRITE;
-	rdma_wr.remote_addr =
+	rdma_wr->wr.wr_id = smc_wr_tx_get_next_wr_id(link);
+	rdma_wr->wr.num_sge = num_sges;
+	rdma_wr->remote_addr =
 		lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].dma_addr +
 		/* RMBE within RMB */
 		conn->tx_off +
 		/* offset within RMBE */
 		peer_rmbe_offset;
-	rdma_wr.rkey = lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey;
-	rc = ib_post_send(link->roce_qp, &rdma_wr.wr, NULL);
+	rdma_wr->rkey = lgr->rtokens[conn->rtoken_idx][SMC_SINGLE_LINK].rkey;
+	rc = ib_post_send(link->roce_qp, &rdma_wr->wr, NULL);
 	if (rc) {
 		conn->local_tx_ctrl.conn_state_flags.peer_conn_abort = 1;
 		smc_lgr_terminate(lgr);
@@ -314,24 +309,25 @@ static inline void smc_tx_advance_cursors(struct smc_connection *conn,
 /* SMC-R helper for smc_tx_rdma_writes() */
 static int smcr_tx_rdma_writes(struct smc_connection *conn, size_t len,
 			       size_t src_off, size_t src_len,
-			       size_t dst_off, size_t dst_len)
+			       size_t dst_off, size_t dst_len,
+			       struct smc_rdma_wr *wr_rdma_buf)
 {
 	dma_addr_t dma_addr =
 		sg_dma_address(conn->sndbuf_desc->sgt[SMC_SINGLE_LINK].sgl);
-	struct smc_link *link = &conn->lgr->lnk[SMC_SINGLE_LINK];
 	int src_len_sum = src_len, dst_len_sum = dst_len;
-	struct ib_sge sges[SMC_IB_MAX_SEND_SGE];
 	int sent_count = src_off;
 	int srcchunk, dstchunk;
 	int num_sges;
 	int rc;
 
 	for (dstchunk = 0; dstchunk < 2; dstchunk++) {
+		struct ib_sge *sge =
+			wr_rdma_buf->wr_tx_rdma[dstchunk].wr.sg_list;
+
 		num_sges = 0;
 		for (srcchunk = 0; srcchunk < 2; srcchunk++) {
-			sges[srcchunk].addr = dma_addr + src_off;
-			sges[srcchunk].length = src_len;
-			sges[srcchunk].lkey = link->roce_pd->local_dma_lkey;
+			sge[srcchunk].addr = dma_addr + src_off;
+			sge[srcchunk].length = src_len;
 			num_sges++;
 
 			src_off += src_len;
@@ -344,7 +340,8 @@ static int smcr_tx_rdma_writes(struct smc_connection *conn, size_t len,
 			src_len = dst_len - src_len; /* remainder */
 			src_len_sum += src_len;
 		}
-		rc = smc_tx_rdma_write(conn, dst_off, num_sges, sges);
+		rc = smc_tx_rdma_write(conn, dst_off, num_sges,
+				       &wr_rdma_buf->wr_tx_rdma[dstchunk]);
 		if (rc)
 			return rc;
 		if (dst_len_sum == len)
@@ -403,7 +400,8 @@ static int smcd_tx_rdma_writes(struct smc_connection *conn, size_t len,
 /* sndbuf consumer: prepare all necessary (src&dst) chunks of data transmit;
  * usable snd_wnd as max transmit
  */
-static int smc_tx_rdma_writes(struct smc_connection *conn)
+static int smc_tx_rdma_writes(struct smc_connection *conn,
+			      struct smc_rdma_wr *wr_rdma_buf)
 {
 	size_t len, src_len, dst_off, dst_len; /* current chunk values */
 	union smc_host_cursor sent, prep, prod, cons;
@@ -464,7 +462,7 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 					 dst_off, dst_len);
 	else
 		rc = smcr_tx_rdma_writes(conn, len, sent.count, src_len,
-					 dst_off, dst_len);
+					 dst_off, dst_len, wr_rdma_buf);
 	if (rc)
 		return rc;
 
@@ -485,31 +483,30 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
 static int smcr_tx_sndbuf_nonempty(struct smc_connection *conn)
 {
 	struct smc_cdc_producer_flags *pflags;
+	struct smc_rdma_wr *wr_rdma_buf;
 	struct smc_cdc_tx_pend *pend;
 	struct smc_wr_buf *wr_buf;
 	int rc;
 
-	spin_lock_bh(&conn->send_lock);
-	rc = smc_cdc_get_free_slot(conn, &wr_buf, &pend);
+	rc = smc_cdc_get_free_slot(conn, &wr_buf, &wr_rdma_buf, &pend);
 	if (rc < 0) {
 		if (rc == -EBUSY) {
 			struct smc_sock *smc =
 				container_of(conn, struct smc_sock, conn);
 
-			if (smc->sk.sk_err == ECONNABORTED) {
-				rc = sock_error(&smc->sk);
-				goto out_unlock;
-			}
+			if (smc->sk.sk_err == ECONNABORTED)
+				return sock_error(&smc->sk);
 			rc = 0;
 			if (conn->alert_token_local) /* connection healthy */
 				mod_delayed_work(system_wq, &conn->tx_work,
 						 SMC_TX_WORK_DELAY);
 		}
-		goto out_unlock;
+		return rc;
 	}
 
+	spin_lock_bh(&conn->send_lock);
 	if (!conn->local_tx_ctrl.prod_flags.urg_data_present) {
-		rc = smc_tx_rdma_writes(conn);
+		rc = smc_tx_rdma_writes(conn, wr_rdma_buf);
 		if (rc) {
 			smc_wr_tx_put_slot(&conn->lgr->lnk[SMC_SINGLE_LINK],
 					   (struct smc_wr_tx_pend_priv *)pend);
@@ -536,7 +533,7 @@ static int smcd_tx_sndbuf_nonempty(struct smc_connection *conn)
 
 	spin_lock_bh(&conn->send_lock);
 	if (!pflags->urg_data_present)
-		rc = smc_tx_rdma_writes(conn);
+		rc = smc_tx_rdma_writes(conn, NULL);
 	if (!rc)
 		rc = smcd_cdc_msg_send(conn);
 
@@ -598,7 +595,8 @@ void smc_tx_consumer_update(struct smc_connection *conn, bool force)
 	if (to_confirm > conn->rmbe_update_limit) {
 		smc_curs_copy(&prod, &conn->local_rx_ctrl.prod, conn);
 		sender_free = conn->rmb_desc->len -
-			      smc_curs_diff(conn->rmb_desc->len, &prod, &cfed);
+			      smc_curs_diff_large(conn->rmb_desc->len,
+						  &cfed, &prod);
 	}
 
 	if (conn->local_rx_ctrl.prod_flags.cons_curs_upd_req ||