Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next

Pull networking updates from David Miller:
 "Another merge window, another pull full of stuff:

   1) Support alternative names for network devices, from Jiri Pirko.

   2) Introduce per-netns netdev notifiers, also from Jiri Pirko.

   3) Support MSG_PEEK in vsock/virtio, from Matias Ezequiel Vara
      Larsen.

   4) Allow compiling out the TLS TOE code, from Jakub Kicinski.

   5) Add several new tracepoints to the kTLS code, also from Jakub.

   6) Support set channels ethtool callback in ena driver, from Sameeh
      Jubran.

   7) New SCTP events SCTP_ADDR_ADDED, SCTP_ADDR_REMOVED,
      SCTP_ADDR_MADE_PRIM, and SCTP_SEND_FAILED_EVENT. From Xin Long.

   8) Add XDP support to mvneta driver, from Lorenzo Bianconi.

   9) Lots of netfilter hw offload fixes, cleanups and enhancements,
      from Pablo Neira Ayuso.

  10) PTP support for aquantia chips, from Egor Pomozov.

  11) Add UDP segmentation offload support to igb, ixgbe, and i40e. From
      Josh Hunt.

  12) Add smart nagle to tipc, from Jon Maloy.

  13) Support L2 field rewrite by TC offloads in bnxt_en, from Venkat
      Duvvuru.

  14) Add a flow mask cache to OVS, from Tonghao Zhang.

  15) Add XDP support to ice driver, from Maciej Fijalkowski.

  16) Add AF_XDP support to ice driver, from Krzysztof Kazimierczak.

  17) Support UDP GSO offload in atlantic driver, from Igor Russkikh.

  18) Support it in stmmac driver too, from Jose Abreu.

  19) Support TIPC encryption and auth, from Tuong Lien.

  20) Introduce BPF trampolines, from Alexei Starovoitov.

  21) Make page_pool API more numa friendly, from Saeed Mahameed.

  22) Introduce route hints to ipv4 and ipv6, from Paolo Abeni.

  23) Add UDP segmentation offload to cxgb4, Rahul Lakkireddy"

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1857 commits)
  libbpf: Fix usage of u32 in userspace code
  mm: Implement no-MMU variant of vmalloc_user_node_flags
  slip: Fix use-after-free Read in slip_open
  net: dsa: sja1105: fix sja1105_parse_rgmii_delays()
  macvlan: schedule bc_work even if error
  enetc: add support Credit Based Shaper(CBS) for hardware offload
  net: phy: add helpers phy_(un)lock_mdio_bus
  mdio_bus: don't use managed reset-controller
  ax88179_178a: add ethtool_op_get_ts_info()
  mlxsw: spectrum_router: Fix use of uninitialized adjacency index
  mlxsw: spectrum_router: After underlay moves, demote conflicting tunnels
  bpf: Simplify __bpf_arch_text_poke poke type handling
  bpf: Introduce BPF_TRACE_x helper for the tracing tests
  bpf: Add bpf_jit_blinding_enabled for !CONFIG_BPF_JIT
  bpf, testing: Add various tail call test cases
  bpf, x86: Emit patchable direct jump as tail call
  bpf: Constant map key tracking for prog array pokes
  bpf: Add poke dependency tracking for prog array maps
  bpf: Add initial poke descriptor table for jit images
  bpf: Move owner type, jited info into array auxiliary data
  ...

1 files changed, 122 insertions, 45 deletions

diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c
index 18c6fac6ead9..c71f4328d138 100644
--- a/net/rds/ib_cm.c
+++ b/net/rds/ib_cm.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006, 2018 Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2006, 2019 Oracle and/or its affiliates. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -30,6 +30,7 @@
  * SOFTWARE.
  *
  */
+#include <linux/dmapool.h>
 #include <linux/kernel.h>
 #include <linux/in.h>
 #include <linux/slab.h>
@@ -439,6 +440,68 @@ static inline void ibdev_put_vector(struct rds_ib_device *rds_ibdev, int index)
 	rds_ibdev->vector_load[index]--;
 }
 
+/* Allocate DMA coherent memory to be used to store struct rds_header for
+ * sending/receiving packets.  The pointers to the DMA memory and the
+ * associated DMA addresses are stored in two arrays.
+ *
+ * @ibdev: the IB device
+ * @pool: the DMA memory pool
+ * @dma_addrs: pointer to the array for storing DMA addresses
+ * @num_hdrs: number of headers to allocate
+ *
+ * It returns the pointer to the array storing the DMA memory pointers.  On
+ * error, NULL pointer is returned.
+ */
+struct rds_header **rds_dma_hdrs_alloc(struct ib_device *ibdev,
+				       struct dma_pool *pool,
+				       dma_addr_t **dma_addrs, u32 num_hdrs)
+{
+	struct rds_header **hdrs;
+	dma_addr_t *hdr_daddrs;
+	u32 i;
+
+	hdrs = kvmalloc_node(sizeof(*hdrs) * num_hdrs, GFP_KERNEL,
+			     ibdev_to_node(ibdev));
+	if (!hdrs)
+		return NULL;
+
+	hdr_daddrs = kvmalloc_node(sizeof(*hdr_daddrs) * num_hdrs, GFP_KERNEL,
+				   ibdev_to_node(ibdev));
+	if (!hdr_daddrs) {
+		kvfree(hdrs);
+		return NULL;
+	}
+
+	for (i = 0; i < num_hdrs; i++) {
+		hdrs[i] = dma_pool_zalloc(pool, GFP_KERNEL, &hdr_daddrs[i]);
+		if (!hdrs[i]) {
+			rds_dma_hdrs_free(pool, hdrs, hdr_daddrs, i);
+			return NULL;
+		}
+	}
+
+	*dma_addrs = hdr_daddrs;
+	return hdrs;
+}
+
+/* Free the DMA memory used to store struct rds_header.
+ *
+ * @pool: the DMA memory pool
+ * @hdrs: pointer to the array storing DMA memory pointers
+ * @dma_addrs: pointer to the array storing DMA addresses
+ * @num_hdars: number of headers to free.
+ */
+void rds_dma_hdrs_free(struct dma_pool *pool, struct rds_header **hdrs,
+		       dma_addr_t *dma_addrs, u32 num_hdrs)
+{
+	u32 i;
+
+	for (i = 0; i < num_hdrs; i++)
+		dma_pool_free(pool, hdrs[i], dma_addrs[i]);
+	kvfree(hdrs);
+	kvfree(dma_addrs);
+}
+
 /*
  * This needs to be very careful to not leave IS_ERR pointers around for
  * cleanup to trip over.
@@ -452,6 +515,7 @@ static int rds_ib_setup_qp(struct rds_connection *conn)
 	struct rds_ib_device *rds_ibdev;
 	unsigned long max_wrs;
 	int ret, fr_queue_space;
+	struct dma_pool *pool;
 
 	/*
 	 * It's normal to see a null device if an incoming connection races
@@ -547,31 +611,28 @@ static int rds_ib_setup_qp(struct rds_connection *conn)
 		goto recv_cq_out;
 	}
 
-	ic->i_send_hdrs = ib_dma_alloc_coherent(dev,
-					   ic->i_send_ring.w_nr *
-						sizeof(struct rds_header),
-					   &ic->i_send_hdrs_dma, GFP_KERNEL);
+	pool = rds_ibdev->rid_hdrs_pool;
+	ic->i_send_hdrs = rds_dma_hdrs_alloc(dev, pool, &ic->i_send_hdrs_dma,
+					     ic->i_send_ring.w_nr);
 	if (!ic->i_send_hdrs) {
 		ret = -ENOMEM;
-		rdsdebug("ib_dma_alloc_coherent send failed\n");
+		rdsdebug("DMA send hdrs alloc failed\n");
 		goto qp_out;
 	}
 
-	ic->i_recv_hdrs = ib_dma_alloc_coherent(dev,
-					   ic->i_recv_ring.w_nr *
-						sizeof(struct rds_header),
-					   &ic->i_recv_hdrs_dma, GFP_KERNEL);
+	ic->i_recv_hdrs = rds_dma_hdrs_alloc(dev, pool, &ic->i_recv_hdrs_dma,
+					     ic->i_recv_ring.w_nr);
 	if (!ic->i_recv_hdrs) {
 		ret = -ENOMEM;
-		rdsdebug("ib_dma_alloc_coherent recv failed\n");
+		rdsdebug("DMA recv hdrs alloc failed\n");
 		goto send_hdrs_dma_out;
 	}
 
-	ic->i_ack = ib_dma_alloc_coherent(dev, sizeof(struct rds_header),
-				       &ic->i_ack_dma, GFP_KERNEL);
+	ic->i_ack = dma_pool_zalloc(pool, GFP_KERNEL,
+				    &ic->i_ack_dma);
 	if (!ic->i_ack) {
 		ret = -ENOMEM;
-		rdsdebug("ib_dma_alloc_coherent ack failed\n");
+		rdsdebug("DMA ack header alloc failed\n");
 		goto recv_hdrs_dma_out;
 	}
 
@@ -602,17 +663,23 @@ static int rds_ib_setup_qp(struct rds_connection *conn)
 
 sends_out:
 	vfree(ic->i_sends);
+
 ack_dma_out:
-	ib_dma_free_coherent(dev, sizeof(struct rds_header),
-			     ic->i_ack, ic->i_ack_dma);
+	dma_pool_free(pool, ic->i_ack, ic->i_ack_dma);
+	ic->i_ack = NULL;
+
 recv_hdrs_dma_out:
-	ib_dma_free_coherent(dev, ic->i_recv_ring.w_nr *
-					sizeof(struct rds_header),
-					ic->i_recv_hdrs, ic->i_recv_hdrs_dma);
+	rds_dma_hdrs_free(pool, ic->i_recv_hdrs, ic->i_recv_hdrs_dma,
+			  ic->i_recv_ring.w_nr);
+	ic->i_recv_hdrs = NULL;
+	ic->i_recv_hdrs_dma = NULL;
+
 send_hdrs_dma_out:
-	ib_dma_free_coherent(dev, ic->i_send_ring.w_nr *
-					sizeof(struct rds_header),
-					ic->i_send_hdrs, ic->i_send_hdrs_dma);
+	rds_dma_hdrs_free(pool, ic->i_send_hdrs, ic->i_send_hdrs_dma,
+			  ic->i_send_ring.w_nr);
+	ic->i_send_hdrs = NULL;
+	ic->i_send_hdrs_dma = NULL;
+
 qp_out:
 	rdma_destroy_qp(ic->i_cm_id);
 recv_cq_out:
@@ -990,8 +1057,6 @@ void rds_ib_conn_path_shutdown(struct rds_conn_path *cp)
 		 ic->i_cm_id ? ic->i_cm_id->qp : NULL);
 
 	if (ic->i_cm_id) {
-		struct ib_device *dev = ic->i_cm_id->device;
-
 		rdsdebug("disconnecting cm %p\n", ic->i_cm_id);
 		err = rdma_disconnect(ic->i_cm_id);
 		if (err) {
@@ -1041,24 +1106,39 @@ void rds_ib_conn_path_shutdown(struct rds_conn_path *cp)
 			ib_destroy_cq(ic->i_recv_cq);
 		}
 
-		/* then free the resources that ib callbacks use */
-		if (ic->i_send_hdrs)
-			ib_dma_free_coherent(dev,
-					   ic->i_send_ring.w_nr *
-						sizeof(struct rds_header),
-					   ic->i_send_hdrs,
-					   ic->i_send_hdrs_dma);
-
-		if (ic->i_recv_hdrs)
-			ib_dma_free_coherent(dev,
-					   ic->i_recv_ring.w_nr *
-						sizeof(struct rds_header),
-					   ic->i_recv_hdrs,
-					   ic->i_recv_hdrs_dma);
-
-		if (ic->i_ack)
-			ib_dma_free_coherent(dev, sizeof(struct rds_header),
-					     ic->i_ack, ic->i_ack_dma);
+		if (ic->rds_ibdev) {
+			struct dma_pool *pool;
+
+			pool = ic->rds_ibdev->rid_hdrs_pool;
+
+			/* then free the resources that ib callbacks use */
+			if (ic->i_send_hdrs) {
+				rds_dma_hdrs_free(pool, ic->i_send_hdrs,
+						  ic->i_send_hdrs_dma,
+						  ic->i_send_ring.w_nr);
+				ic->i_send_hdrs = NULL;
+				ic->i_send_hdrs_dma = NULL;
+			}
+
+			if (ic->i_recv_hdrs) {
+				rds_dma_hdrs_free(pool, ic->i_recv_hdrs,
+						  ic->i_recv_hdrs_dma,
+						  ic->i_recv_ring.w_nr);
+				ic->i_recv_hdrs = NULL;
+				ic->i_recv_hdrs_dma = NULL;
+			}
+
+			if (ic->i_ack) {
+				dma_pool_free(pool, ic->i_ack, ic->i_ack_dma);
+				ic->i_ack = NULL;
+			}
+		} else {
+			WARN_ON(ic->i_send_hdrs);
+			WARN_ON(ic->i_send_hdrs_dma);
+			WARN_ON(ic->i_recv_hdrs);
+			WARN_ON(ic->i_recv_hdrs_dma);
+			WARN_ON(ic->i_ack);
+		}
 
 		if (ic->i_sends)
 			rds_ib_send_clear_ring(ic);
@@ -1077,9 +1157,6 @@ void rds_ib_conn_path_shutdown(struct rds_conn_path *cp)
 		ic->i_pd = NULL;
 		ic->i_send_cq = NULL;
 		ic->i_recv_cq = NULL;
-		ic->i_send_hdrs = NULL;
-		ic->i_recv_hdrs = NULL;
-		ic->i_ack = NULL;
 	}
 	BUG_ON(ic->rds_ibdev);