Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma updates from Jason Gunthorpe:
 "A smaller cycle this time. Notably we see another new driver, 'Soft
  iWarp', and the deletion of an ancient unused driver for nes.

   - Revise and simplify the signature offload RDMA MR APIs

   - More progress on hoisting object allocation boiler plate code out
     of the drivers

   - Driver bug fixes and revisions for hns, hfi1, efa, cxgb4, qib,
     i40iw

   - Tree wide cleanups: struct_size, put_user_page, xarray, rst doc
     conversion

   - Removal of obsolete ib_ucm chardev and nes driver

   - netlink based discovery of chardevs and autoloading of the modules
     providing them

   - Move more of the rdamvt/hfi1 uapi to include/uapi/rdma

   - New driver 'siw' for software based iWarp running on top of netdev,
     much like rxe's software RoCE.

   - mlx5 feature to report events in their raw devx format to userspace

   - Expose per-object counters through rdma tool

   - Adaptive interrupt moderation for RDMA (DIM), sharing the DIM core
     from netdev"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (194 commits)
  RMDA/siw: Require a 64 bit arch
  RDMA/siw: Mark expected switch fall-throughs
  RDMA/core: Fix -Wunused-const-variable warnings
  rdma/siw: Remove set but not used variable 's'
  rdma/siw: Add missing dependencies on LIBCRC32C and DMA_VIRT_OPS
  RDMA/siw: Add missing rtnl_lock around access to ifa
  rdma/siw: Use proper enumerated type in map_cqe_status
  RDMA/siw: Remove unnecessary kthread create/destroy printouts
  IB/rdmavt: Fix variable shadowing issue in rvt_create_cq
  RDMA/core: Fix race when resolving IP address
  RDMA/core: Make rdma_counter.h compile stand alone
  IB/core: Work on the caller socket net namespace in nldev_newlink()
  RDMA/rxe: Fill in wc byte_len with IB_WC_RECV_RDMA_WITH_IMM
  RDMA/mlx5: Set RDMA DIM to be enabled by default
  RDMA/nldev: Added configuration of RDMA dynamic interrupt moderation to netlink
  RDMA/core: Provide RDMA DIM support for ULPs
  linux/dim: Implement RDMA adaptive moderation (DIM)
  IB/mlx5: Report correctly tag matching rendezvous capability
  docs: infiniband: add it to the driver-api bookset
  IB/mlx5: Implement VHCA tunnel mechanism in DEVX
  ...

8 files changed, 99 insertions, 125 deletions

diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
index 09fcfc9e052d..e87fc0408470 100644
--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -953,7 +953,7 @@ static int send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb,
 	mpalen = sizeof(*mpa) + ep->plen;
 	if (mpa_rev_to_use == 2)
 		mpalen += sizeof(struct mpa_v2_conn_params);
-	wrlen = roundup(mpalen + sizeof *req, 16);
+	wrlen = roundup(mpalen + sizeof(*req), 16);
 	skb = get_skb(skb, wrlen, GFP_KERNEL);
 	if (!skb) {
 		connect_reply_upcall(ep, -ENOMEM);
@@ -997,8 +997,9 @@ static int send_mpa_req(struct c4iw_ep *ep, struct sk_buff *skb,
 	}
 
 	if (mpa_rev_to_use == 2) {
-		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
-					       sizeof (struct mpa_v2_conn_params));
+		mpa->private_data_size =
+			htons(ntohs(mpa->private_data_size) +
+			      sizeof(struct mpa_v2_conn_params));
 		pr_debug("initiator ird %u ord %u\n", ep->ird,
 			 ep->ord);
 		mpa_v2_params.ird = htons((u16)ep->ird);
@@ -1057,7 +1058,7 @@ static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen)
 	mpalen = sizeof(*mpa) + plen;
 	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn)
 		mpalen += sizeof(struct mpa_v2_conn_params);
-	wrlen = roundup(mpalen + sizeof *req, 16);
+	wrlen = roundup(mpalen + sizeof(*req), 16);
 
 	skb = get_skb(NULL, wrlen, GFP_KERNEL);
 	if (!skb) {
@@ -1088,8 +1089,9 @@ static int send_mpa_reject(struct c4iw_ep *ep, const void *pdata, u8 plen)
 
 	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) {
 		mpa->flags |= MPA_ENHANCED_RDMA_CONN;
-		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
-					       sizeof (struct mpa_v2_conn_params));
+		mpa->private_data_size =
+			htons(ntohs(mpa->private_data_size) +
+			      sizeof(struct mpa_v2_conn_params));
 		mpa_v2_params.ird = htons(((u16)ep->ird) |
 					  (peer2peer ? MPA_V2_PEER2PEER_MODEL :
 					   0));
@@ -1136,7 +1138,7 @@ static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen)
 	mpalen = sizeof(*mpa) + plen;
 	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn)
 		mpalen += sizeof(struct mpa_v2_conn_params);
-	wrlen = roundup(mpalen + sizeof *req, 16);
+	wrlen = roundup(mpalen + sizeof(*req), 16);
 
 	skb = get_skb(NULL, wrlen, GFP_KERNEL);
 	if (!skb) {
@@ -1171,8 +1173,9 @@ static int send_mpa_reply(struct c4iw_ep *ep, const void *pdata, u8 plen)
 
 	if (ep->mpa_attr.version == 2 && ep->mpa_attr.enhanced_rdma_conn) {
 		mpa->flags |= MPA_ENHANCED_RDMA_CONN;
-		mpa->private_data_size = htons(ntohs(mpa->private_data_size) +
-					       sizeof (struct mpa_v2_conn_params));
+		mpa->private_data_size =
+			htons(ntohs(mpa->private_data_size) +
+			      sizeof(struct mpa_v2_conn_params));
 		mpa_v2_params.ird = htons((u16)ep->ird);
 		mpa_v2_params.ord = htons((u16)ep->ord);
 		if (peer2peer && (ep->mpa_attr.p2p_type !=
diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c
index 52ce586621c6..b1bb61c65f4f 100644
--- a/drivers/infiniband/hw/cxgb4/cq.c
+++ b/drivers/infiniband/hw/cxgb4/cq.c
@@ -34,16 +34,15 @@
 
 #include "iw_cxgb4.h"
 
-static int destroy_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
-		      struct c4iw_dev_ucontext *uctx, struct sk_buff *skb,
-		      struct c4iw_wr_wait *wr_waitp)
+static void destroy_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
+		       struct c4iw_dev_ucontext *uctx, struct sk_buff *skb,
+		       struct c4iw_wr_wait *wr_waitp)
 {
 	struct fw_ri_res_wr *res_wr;
 	struct fw_ri_res *res;
 	int wr_len;
-	int ret;
 
-	wr_len = sizeof *res_wr + sizeof *res;
+	wr_len = sizeof(*res_wr) + sizeof(*res);
 	set_wr_txq(skb, CPL_PRIORITY_CONTROL, 0);
 
 	res_wr = __skb_put_zero(skb, wr_len);
@@ -59,14 +58,13 @@ static int destroy_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
 	res->u.cq.iqid = cpu_to_be32(cq->cqid);
 
 	c4iw_init_wr_wait(wr_waitp);
-	ret = c4iw_ref_send_wait(rdev, skb, wr_waitp, 0, 0, __func__);
+	c4iw_ref_send_wait(rdev, skb, wr_waitp, 0, 0, __func__);
 
 	kfree(cq->sw_queue);
 	dma_free_coherent(&(rdev->lldi.pdev->dev),
 			  cq->memsize, cq->queue,
 			  dma_unmap_addr(cq, mapping));
 	c4iw_put_cqid(rdev, cq->cqid, uctx);
-	return ret;
 }
 
 static int create_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
@@ -104,7 +102,6 @@ static int create_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
 		goto err3;
 	}
 	dma_unmap_addr_set(cq, mapping, cq->dma_addr);
-	memset(cq->queue, 0, cq->memsize);
 
 	if (user && ucontext->is_32b_cqe) {
 		cq->qp_errp = &((struct t4_status_page *)
@@ -117,7 +114,7 @@ static int create_cq(struct c4iw_rdev *rdev, struct t4_cq *cq,
 	}
 
 	/* build fw_ri_res_wr */
-	wr_len = sizeof *res_wr + sizeof *res;
+	wr_len = sizeof(*res_wr) + sizeof(*res);
 
 	skb = alloc_skb(wr_len, GFP_KERNEL);
 	if (!skb) {
@@ -970,7 +967,7 @@ int c4iw_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
 	return !err || err == -ENODATA ? npolled : err;
 }
 
-int c4iw_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata)
+void c4iw_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata)
 {
 	struct c4iw_cq *chp;
 	struct c4iw_ucontext *ucontext;
@@ -988,18 +985,16 @@ int c4iw_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata)
 		   ucontext ? &ucontext->uctx : &chp->cq.rdev->uctx,
 		   chp->destroy_skb, chp->wr_waitp);
 	c4iw_put_wr_wait(chp->wr_waitp);
-	kfree(chp);
-	return 0;
 }
 
-struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
-			     const struct ib_cq_init_attr *attr,
-			     struct ib_udata *udata)
+int c4iw_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
+		   struct ib_udata *udata)
 {
+	struct ib_device *ibdev = ibcq->device;
 	int entries = attr->cqe;
 	int vector = attr->comp_vector;
-	struct c4iw_dev *rhp;
-	struct c4iw_cq *chp;
+	struct c4iw_dev *rhp = to_c4iw_dev(ibcq->device);
+	struct c4iw_cq *chp = to_c4iw_cq(ibcq);
 	struct c4iw_create_cq ucmd;
 	struct c4iw_create_cq_resp uresp;
 	int ret, wr_len;
@@ -1010,22 +1005,16 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
 
 	pr_debug("ib_dev %p entries %d\n", ibdev, entries);
 	if (attr->flags)
-		return ERR_PTR(-EINVAL);
-
-	rhp = to_c4iw_dev(ibdev);
+		return -EINVAL;
 
 	if (vector >= rhp->rdev.lldi.nciq)
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	if (udata) {
 		if (udata->inlen < sizeof(ucmd))
 			ucontext->is_32b_cqe = 1;
 	}
 
-	chp = kzalloc(sizeof(*chp), GFP_KERNEL);
-	if (!chp)
-		return ERR_PTR(-ENOMEM);
-
 	chp->wr_waitp = c4iw_alloc_wr_wait(GFP_KERNEL);
 	if (!chp->wr_waitp) {
 		ret = -ENOMEM;
@@ -1095,10 +1084,10 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
 
 	if (ucontext) {
 		ret = -ENOMEM;
-		mm = kmalloc(sizeof *mm, GFP_KERNEL);
+		mm = kmalloc(sizeof(*mm), GFP_KERNEL);
 		if (!mm)
 			goto err_remove_handle;
-		mm2 = kmalloc(sizeof *mm2, GFP_KERNEL);
+		mm2 = kmalloc(sizeof(*mm2), GFP_KERNEL);
 		if (!mm2)
 			goto err_free_mm;
 
@@ -1135,10 +1124,11 @@ struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
 		mm2->len = PAGE_SIZE;
 		insert_mmap(ucontext, mm2);
 	}
-	pr_debug("cqid 0x%0x chp %p size %u memsize %zu, dma_addr 0x%0llx\n",
-		 chp->cq.cqid, chp, chp->cq.size,
-		 chp->cq.memsize, (unsigned long long)chp->cq.dma_addr);
-	return &chp->ibcq;
+
+	pr_debug("cqid 0x%0x chp %p size %u memsize %zu, dma_addr %pad\n",
+		 chp->cq.cqid, chp, chp->cq.size, chp->cq.memsize,
+		 &chp->cq.dma_addr);
+	return 0;
 err_free_mm2:
 	kfree(mm2);
 err_free_mm:
@@ -1154,8 +1144,7 @@ err_free_skb:
 err_free_wr_wait:
 	c4iw_put_wr_wait(chp->wr_waitp);
 err_free_chp:
-	kfree(chp);
-	return ERR_PTR(ret);
+	return ret;
 }
 
 int c4iw_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags)
diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c
index 4c0d925c5ff5..a8b9548bd1a2 100644
--- a/drivers/infiniband/hw/cxgb4/device.c
+++ b/drivers/infiniband/hw/cxgb4/device.c
@@ -327,7 +327,7 @@ static int qp_open(struct inode *inode, struct file *file)
 	unsigned long index;
 	int count = 1;
 
-	qpd = kmalloc(sizeof *qpd, GFP_KERNEL);
+	qpd = kmalloc(sizeof(*qpd), GFP_KERNEL);
 	if (!qpd)
 		return -ENOMEM;
 
@@ -421,7 +421,7 @@ static int stag_open(struct inode *inode, struct file *file)
 	int ret = 0;
 	int count = 1;
 
-	stagd = kmalloc(sizeof *stagd, GFP_KERNEL);
+	stagd = kmalloc(sizeof(*stagd), GFP_KERNEL);
 	if (!stagd) {
 		ret = -ENOMEM;
 		goto out;
@@ -1075,7 +1075,7 @@ static void *c4iw_uld_add(const struct cxgb4_lld_info *infop)
 		pr_info("Chelsio T4/T5 RDMA Driver - version %s\n",
 			DRV_VERSION);
 
-	ctx = kzalloc(sizeof *ctx, GFP_KERNEL);
+	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
 	if (!ctx) {
 		ctx = ERR_PTR(-ENOMEM);
 		goto out;
@@ -1243,10 +1243,9 @@ static int c4iw_uld_state_change(void *handle, enum cxgb4_state new_state)
 	case CXGB4_STATE_START_RECOVERY:
 		pr_info("%s: Fatal Error\n", pci_name(ctx->lldi.pdev));
 		if (ctx->dev) {
-			struct ib_event event;
+			struct ib_event event = {};
 
 			ctx->dev->rdev.flags |= T4_FATAL_ERROR;
-			memset(&event, 0, sizeof event);
 			event.event  = IB_EVENT_DEVICE_FATAL;
 			event.device = &ctx->dev->ibdev;
 			ib_dispatch_event(&event);
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index 916ef982172e..7d06b0f8d49a 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -490,13 +490,13 @@ struct c4iw_qp {
 	struct t4_wq wq;
 	spinlock_t lock;
 	struct mutex mutex;
-	struct kref kref;
 	wait_queue_head_t wait;
 	int sq_sig_all;
 	struct c4iw_srq *srq;
-	struct work_struct free_work;
 	struct c4iw_ucontext *ucontext;
 	struct c4iw_wr_wait *wr_waitp;
+	struct completion qp_rel_comp;
+	refcount_t qp_refcnt;
 };
 
 static inline struct c4iw_qp *to_c4iw_qp(struct ib_qp *ibqp)
@@ -992,10 +992,9 @@ struct ib_mr *c4iw_reg_user_mr(struct ib_pd *pd, u64 start,
 					   struct ib_udata *udata);
 struct ib_mr *c4iw_get_dma_mr(struct ib_pd *pd, int acc);
 int c4iw_dereg_mr(struct ib_mr *ib_mr, struct ib_udata *udata);
-int c4iw_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata);
-struct ib_cq *c4iw_create_cq(struct ib_device *ibdev,
-			     const struct ib_cq_init_attr *attr,
-			     struct ib_udata *udata);
+void c4iw_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata);
+int c4iw_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
+		   struct ib_udata *udata);
 int c4iw_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags);
 int c4iw_modify_srq(struct ib_srq *ib_srq, struct ib_srq_attr *attr,
 		    enum ib_srq_attr_mask srq_attr_mask,
diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c
index 811c0c8c5b16..aa772ee0706f 100644
--- a/drivers/infiniband/hw/cxgb4/mem.c
+++ b/drivers/infiniband/hw/cxgb4/mem.c
@@ -130,8 +130,9 @@ static int _c4iw_write_mem_inline(struct c4iw_rdev *rdev, u32 addr, u32 len,
 
 		copy_len = len > C4IW_MAX_INLINE_SIZE ? C4IW_MAX_INLINE_SIZE :
 			   len;
-		wr_len = roundup(sizeof *req + sizeof *sc +
-				 roundup(copy_len, T4_ULPTX_MIN_IO), 16);
+		wr_len = roundup(sizeof(*req) + sizeof(*sc) +
+					 roundup(copy_len, T4_ULPTX_MIN_IO),
+				 16);
 
 		if (!skb) {
 			skb = alloc_skb(wr_len, GFP_KERNEL | __GFP_NOFAIL);
@@ -807,8 +808,7 @@ int c4iw_dereg_mr(struct ib_mr *ib_mr, struct ib_udata *udata)
 				  mhp->attr.pbl_size << 3);
 	if (mhp->kva)
 		kfree((void *) (unsigned long) mhp->kva);
-	if (mhp->umem)
-		ib_umem_release(mhp->umem);
+	ib_umem_release(mhp->umem);
 	pr_debug("mmid 0x%x ptr %p\n", mmid, mhp);
 	c4iw_put_wr_wait(mhp->wr_waitp);
 	kfree(mhp);
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index 74b795642fca..5e59c5708729 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -271,7 +271,6 @@ static int c4iw_query_device(struct ib_device *ibdev, struct ib_device_attr *pro
 		return -EINVAL;
 
 	dev = to_c4iw_dev(ibdev);
-	memset(props, 0, sizeof *props);
 	memcpy(&props->sys_image_guid, dev->rdev.lldi.ports[0]->dev_addr, 6);
 	props->hw_ver = CHELSIO_CHIP_RELEASE(dev->rdev.lldi.adapter_type);
 	props->fw_ver = dev->rdev.lldi.fw_vers;
@@ -490,6 +489,10 @@ static int fill_res_entry(struct sk_buff *msg, struct rdma_restrack_entry *res)
 }
 
 static const struct ib_device_ops c4iw_dev_ops = {
+	.owner = THIS_MODULE,
+	.driver_id = RDMA_DRIVER_CXGB4,
+	.uverbs_abi_ver = C4IW_UVERBS_ABI_VERSION,
+
 	.alloc_hw_stats = c4iw_alloc_stats,
 	.alloc_mr = c4iw_alloc_mr,
 	.alloc_mw = c4iw_alloc_mw,
@@ -534,6 +537,7 @@ static const struct ib_device_ops c4iw_dev_ops = {
 	.reg_user_mr = c4iw_reg_user_mr,
 	.req_notify_cq = c4iw_arm_cq,
 	INIT_RDMA_OBJ_SIZE(ib_pd, c4iw_pd, ibpd),
+	INIT_RDMA_OBJ_SIZE(ib_cq, c4iw_cq, ibcq),
 	INIT_RDMA_OBJ_SIZE(ib_srq, c4iw_srq, ibsrq),
 	INIT_RDMA_OBJ_SIZE(ib_ucontext, c4iw_ucontext, ibucontext),
 };
@@ -561,7 +565,6 @@ void c4iw_register_device(struct work_struct *work)
 	pr_debug("c4iw_dev %p\n", dev);
 	memset(&dev->ibdev.node_guid, 0, sizeof(dev->ibdev.node_guid));
 	memcpy(&dev->ibdev.node_guid, dev->rdev.lldi.ports[0]->dev_addr, 6);
-	dev->ibdev.owner = THIS_MODULE;
 	dev->device_cap_flags = IB_DEVICE_LOCAL_DMA_LKEY | IB_DEVICE_MEM_WINDOW;
 	if (fastreg_support)
 		dev->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS;
@@ -594,13 +597,11 @@ void c4iw_register_device(struct work_struct *work)
 	dev->ibdev.phys_port_cnt = dev->rdev.lldi.nports;
 	dev->ibdev.num_comp_vectors =  dev->rdev.lldi.nciq;
 	dev->ibdev.dev.parent = &dev->rdev.lldi.pdev->dev;
-	dev->ibdev.uverbs_abi_ver = C4IW_UVERBS_ABI_VERSION;
 
 	memcpy(dev->ibdev.iw_ifname, dev->rdev.lldi.ports[0]->name,
 	       sizeof(dev->ibdev.iw_ifname));
 
 	rdma_set_device_sysfs_group(&dev->ibdev, &c4iw_attr_group);
-	dev->ibdev.driver_id = RDMA_DRIVER_CXGB4;
 	ib_set_device_ops(&dev->ibdev, &c4iw_dev_ops);
 	ret = set_netdevs(&dev->ibdev, &dev->rdev);
 	if (ret)
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index e92b9544357a..eb9368be28c1 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -274,7 +274,6 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 			 (unsigned long long)virt_to_phys(wq->sq.queue),
 			 wq->rq.queue,
 			 (unsigned long long)virt_to_phys(wq->rq.queue));
-		memset(wq->rq.queue, 0, wq->rq.memsize);
 		dma_unmap_addr_set(&wq->rq, mapping, wq->rq.dma_addr);
 	}
 
@@ -303,7 +302,7 @@ static int create_qp(struct c4iw_rdev *rdev, struct t4_wq *wq,
 	wq->rq.msn = 1;
 
 	/* build fw_ri_res_wr */
-	wr_len = sizeof *res_wr + 2 * sizeof *res;
+	wr_len = sizeof(*res_wr) + 2 * sizeof(*res);
 	if (need_rq)
 		wr_len += sizeof(*res);
 	skb = alloc_skb(wr_len, GFP_KERNEL);
@@ -439,7 +438,7 @@ static int build_immd(struct t4_sq *sq, struct fw_ri_immd *immdp,
 			rem -= len;
 		}
 	}
-	len = roundup(plen + sizeof *immdp, 16) - (plen + sizeof *immdp);
+	len = roundup(plen + sizeof(*immdp), 16) - (plen + sizeof(*immdp));
 	if (len)
 		memset(dstp, 0, len);
 	immdp->op = FW_RI_DATA_IMMD;
@@ -528,7 +527,7 @@ static int build_rdma_send(struct t4_sq *sq, union t4_wr *wqe,
 					 T4_MAX_SEND_INLINE, &plen);
 			if (ret)
 				return ret;
-			size = sizeof wqe->send + sizeof(struct fw_ri_immd) +
+			size = sizeof(wqe->send) + sizeof(struct fw_ri_immd) +
 			       plen;
 		} else {
 			ret = build_isgl((__be64 *)sq->queue,
@@ -537,7 +536,7 @@ static int build_rdma_send(struct t4_sq *sq, union t4_wr *wqe,
 					 wr->sg_list, wr->num_sge, &plen);
 			if (ret)
 				return ret;
-			size = sizeof wqe->send + sizeof(struct fw_ri_isgl) +
+			size = sizeof(wqe->send) + sizeof(struct fw_ri_isgl) +
 			       wr->num_sge * sizeof(struct fw_ri_sge);
 		}
 	} else {
@@ -545,7 +544,7 @@ static int build_rdma_send(struct t4_sq *sq, union t4_wr *wqe,
 		wqe->send.u.immd_src[0].r1 = 0;
 		wqe->send.u.immd_src[0].r2 = 0;
 		wqe->send.u.immd_src[0].immdlen = 0;
-		size = sizeof wqe->send + sizeof(struct fw_ri_immd);
+		size = sizeof(wqe->send) + sizeof(struct fw_ri_immd);
 		plen = 0;
 	}
 	*len16 = DIV_ROUND_UP(size, 16);
@@ -579,7 +578,7 @@ static int build_rdma_write(struct t4_sq *sq, union t4_wr *wqe,
 					 T4_MAX_WRITE_INLINE, &plen);
 			if (ret)
 				return ret;
-			size = sizeof wqe->write + sizeof(struct fw_ri_immd) +
+			size = sizeof(wqe->write) + sizeof(struct fw_ri_immd) +
 			       plen;
 		} else {
 			ret = build_isgl((__be64 *)sq->queue,
@@ -588,7 +587,7 @@ static int build_rdma_write(struct t4_sq *sq, union t4_wr *wqe,
 					 wr->sg_list, wr->num_sge, &plen);
 			if (ret)
 				return ret;
-			size = sizeof wqe->write + sizeof(struct fw_ri_isgl) +
+			size = sizeof(wqe->write) + sizeof(struct fw_ri_isgl) +
 			       wr->num_sge * sizeof(struct fw_ri_sge);
 		}
 	} else {
@@ -596,7 +595,7 @@ static int build_rdma_write(struct t4_sq *sq, union t4_wr *wqe,
 		wqe->write.u.immd_src[0].r1 = 0;
 		wqe->write.u.immd_src[0].r2 = 0;
 		wqe->write.u.immd_src[0].immdlen = 0;
-		size = sizeof wqe->write + sizeof(struct fw_ri_immd);
+		size = sizeof(wqe->write) + sizeof(struct fw_ri_immd);
 		plen = 0;
 	}
 	*len16 = DIV_ROUND_UP(size, 16);
@@ -683,7 +682,7 @@ static int build_rdma_read(union t4_wr *wqe, const struct ib_send_wr *wr,
 	}
 	wqe->read.r2 = 0;
 	wqe->read.r5 = 0;
-	*len16 = DIV_ROUND_UP(sizeof wqe->read, 16);
+	*len16 = DIV_ROUND_UP(sizeof(wqe->read), 16);
 	return 0;
 }
 
@@ -766,8 +765,8 @@ static int build_rdma_recv(struct c4iw_qp *qhp, union t4_recv_wr *wqe,
 			 &wqe->recv.isgl, wr->sg_list, wr->num_sge, NULL);
 	if (ret)
 		return ret;
-	*len16 = DIV_ROUND_UP(sizeof wqe->recv +
-			      wr->num_sge * sizeof(struct fw_ri_sge), 16);
+	*len16 = DIV_ROUND_UP(
+		sizeof(wqe->recv) + wr->num_sge * sizeof(struct fw_ri_sge), 16);
 	return 0;
 }
 
@@ -886,47 +885,21 @@ static int build_inv_stag(union t4_wr *wqe, const struct ib_send_wr *wr,
 {
 	wqe->inv.stag_inv = cpu_to_be32(wr->ex.invalidate_rkey);
 	wqe->inv.r2 = 0;
-	*len16 = DIV_ROUND_UP(sizeof wqe->inv, 16);
+	*len16 = DIV_ROUND_UP(sizeof(wqe->inv), 16);
 	return 0;
 }
 
-static void free_qp_work(struct work_struct *work)
-{
-	struct c4iw_ucontext *ucontext;
-	struct c4iw_qp *qhp;
-	struct c4iw_dev *rhp;
-
-	qhp = container_of(work, struct c4iw_qp, free_work);
-	ucontext = qhp->ucontext;
-	rhp = qhp->rhp;
-
-	pr_debug("qhp %p ucontext %p\n", qhp, ucontext);
-	destroy_qp(&rhp->rdev, &qhp->wq,
-		   ucontext ? &ucontext->uctx : &rhp->rdev.uctx, !qhp->srq);
-
-	c4iw_put_wr_wait(qhp->wr_waitp);
-	kfree(qhp);
-}
-
-static void queue_qp_free(struct kref *kref)
-{
-	struct c4iw_qp *qhp;
-
-	qhp = container_of(kref, struct c4iw_qp, kref);
-	pr_debug("qhp %p\n", qhp);
-	queue_work(qhp->rhp->rdev.free_workq, &qhp->free_work);
-}
-
 void c4iw_qp_add_ref(struct ib_qp *qp)
 {
 	pr_debug("ib_qp %p\n", qp);
-	kref_get(&to_c4iw_qp(qp)->kref);
+	refcount_inc(&to_c4iw_qp(qp)->qp_refcnt);
 }
 
 void c4iw_qp_rem_ref(struct ib_qp *qp)
 {
 	pr_debug("ib_qp %p\n", qp);
-	kref_put(&to_c4iw_qp(qp)->kref, queue_qp_free);
+	if (refcount_dec_and_test(&to_c4iw_qp(qp)->qp_refcnt))
+		complete(&to_c4iw_qp(qp)->qp_rel_comp);
 }
 
 static void add_to_fc_list(struct list_head *head, struct list_head *entry)
@@ -1606,7 +1579,7 @@ static void post_terminate(struct c4iw_qp *qhp, struct t4_cqe *err_cqe,
 		FW_WR_LEN16_V(DIV_ROUND_UP(sizeof(*wqe), 16)));
 
 	wqe->u.terminate.type = FW_RI_TYPE_TERMINATE;
-	wqe->u.terminate.immdlen = cpu_to_be32(sizeof *term);
+	wqe->u.terminate.immdlen = cpu_to_be32(sizeof(*term));
 	term = (struct terminate_message *)wqe->u.terminate.termmsg;
 	if (qhp->attr.layer_etype == (LAYER_MPA|DDP_LLP)) {
 		term->layer_etype = qhp->attr.layer_etype;
@@ -1751,16 +1724,15 @@ static int rdma_fini(struct c4iw_dev *rhp, struct c4iw_qp *qhp,
 static void build_rtr_msg(u8 p2p_type, struct fw_ri_init *init)
 {
 	pr_debug("p2p_type = %d\n", p2p_type);
-	memset(&init->u, 0, sizeof init->u);
+	memset(&init->u, 0, sizeof(init->u));
 	switch (p2p_type) {
 	case FW_RI_INIT_P2PTYPE_RDMA_WRITE:
 		init->u.write.opcode = FW_RI_RDMA_WRITE_WR;
 		init->u.write.stag_sink = cpu_to_be32(1);
 		init->u.write.to_sink = cpu_to_be64(1);
 		init->u.write.u.immd_src[0].op = FW_RI_DATA_IMMD;
-		init->u.write.len16 = DIV_ROUND_UP(sizeof init->u.write +
-						   sizeof(struct fw_ri_immd),
-						   16);
+		init->u.write.len16 = DIV_ROUND_UP(
+			sizeof(init->u.write) + sizeof(struct fw_ri_immd), 16);
 		break;
 	case FW_RI_INIT_P2PTYPE_READ_REQ:
 		init->u.write.opcode = FW_RI_RDMA_READ_WR;
@@ -1768,7 +1740,7 @@ static void build_rtr_msg(u8 p2p_type, struct fw_ri_init *init)
 		init->u.read.to_src_lo = cpu_to_be32(1);
 		init->u.read.stag_sink = cpu_to_be32(1);
 		init->u.read.to_sink_lo = cpu_to_be32(1);
-		init->u.read.len16 = DIV_ROUND_UP(sizeof init->u.read, 16);
+		init->u.read.len16 = DIV_ROUND_UP(sizeof(init->u.read), 16);
 		break;
 	}
 }
@@ -1782,7 +1754,7 @@ static int rdma_init(struct c4iw_dev *rhp, struct c4iw_qp *qhp)
 	pr_debug("qhp %p qid 0x%x tid %u ird %u ord %u\n", qhp,
 		 qhp->wq.sq.qid, qhp->ep->hwtid, qhp->ep->ird, qhp->ep->ord);
 
-	skb = alloc_skb(sizeof *wqe, GFP_KERNEL);
+	skb = alloc_skb(sizeof(*wqe), GFP_KERNEL);
 	if (!skb) {
 		ret = -ENOMEM;
 		goto out;
@@ -2099,10 +2071,12 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp, struct ib_udata *udata)
 {
 	struct c4iw_dev *rhp;
 	struct c4iw_qp *qhp;
+	struct c4iw_ucontext *ucontext;
 	struct c4iw_qp_attributes attrs;
 
 	qhp = to_c4iw_qp(ib_qp);
 	rhp = qhp->rhp;
+	ucontext = qhp->ucontext;
 
 	attrs.next_state = C4IW_QP_STATE_ERROR;
 	if (qhp->attr.state == C4IW_QP_STATE_TERMINATE)
@@ -2120,7 +2094,17 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp, struct ib_udata *udata)
 
 	c4iw_qp_rem_ref(ib_qp);
 
+	wait_for_completion(&qhp->qp_rel_comp);
+
 	pr_debug("ib_qp %p qpid 0x%0x\n", ib_qp, qhp->wq.sq.qid);
+	pr_debug("qhp %p ucontext %p\n", qhp, ucontext);
+
+	destroy_qp(&rhp->rdev, &qhp->wq,
+		   ucontext ? &ucontext->uctx : &rhp->rdev.uctx, !qhp->srq);
+
+	c4iw_put_wr_wait(qhp->wr_waitp);
+
+	kfree(qhp);
 	return 0;
 }
 
@@ -2230,8 +2214,8 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 	spin_lock_init(&qhp->lock);
 	mutex_init(&qhp->mutex);
 	init_waitqueue_head(&qhp->wait);
-	kref_init(&qhp->kref);
-	INIT_WORK(&qhp->free_work, free_qp_work);
+	init_completion(&qhp->qp_rel_comp);
+	refcount_set(&qhp->qp_refcnt, 1);
 
 	ret = xa_insert_irq(&rhp->qps, qhp->wq.sq.qid, qhp, GFP_KERNEL);
 	if (ret)
@@ -2302,7 +2286,7 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 			ucontext->key += PAGE_SIZE;
 		}
 		spin_unlock(&ucontext->mmap_lock);
-		ret = ib_copy_to_udata(udata, &uresp, sizeof uresp);
+		ret = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 		if (ret)
 			goto err_free_ma_sync_key;
 		sq_key_mm->key = uresp.sq_key;
@@ -2386,7 +2370,7 @@ int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 	struct c4iw_dev *rhp;
 	struct c4iw_qp *qhp;
 	enum c4iw_qp_attr_mask mask = 0;
-	struct c4iw_qp_attributes attrs;
+	struct c4iw_qp_attributes attrs = {};
 
 	pr_debug("ib_qp %p\n", ibqp);
 
@@ -2398,7 +2382,6 @@ int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 	if (!attr_mask)
 		return 0;
 
-	memset(&attrs, 0, sizeof attrs);
 	qhp = to_c4iw_qp(ibqp);
 	rhp = qhp->rhp;
 
@@ -2482,8 +2465,8 @@ int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 {
 	struct c4iw_qp *qhp = to_c4iw_qp(ibqp);
 
-	memset(attr, 0, sizeof *attr);
-	memset(init_attr, 0, sizeof *init_attr);
+	memset(attr, 0, sizeof(*attr));
+	memset(init_attr, 0, sizeof(*init_attr));
 	attr->qp_state = to_ib_qp_state(qhp->attr.state);
 	init_attr->cap.max_send_wr = qhp->attr.sq_num_entries;
 	init_attr->cap.max_recv_wr = qhp->attr.rq_num_entries;
diff --git a/drivers/infiniband/hw/cxgb4/resource.c b/drivers/infiniband/hw/cxgb4/resource.c
index 57ed26b3cc21..5c95c789f302 100644
--- a/drivers/infiniband/hw/cxgb4/resource.c
+++ b/drivers/infiniband/hw/cxgb4/resource.c
@@ -126,7 +126,7 @@ u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		rdev->stats.qid.cur += rdev->qpmask + 1;
 		mutex_unlock(&rdev->stats.lock);
 		for (i = qid+1; i & rdev->qpmask; i++) {
-			entry = kmalloc(sizeof *entry, GFP_KERNEL);
+			entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 			if (!entry)
 				goto out;
 			entry->qid = i;
@@ -137,13 +137,13 @@ u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		 * now put the same ids on the qp list since they all
 		 * map to the same db/gts page.
 		 */
-		entry = kmalloc(sizeof *entry, GFP_KERNEL);
+		entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 		if (!entry)
 			goto out;
 		entry->qid = qid;
 		list_add_tail(&entry->entry, &uctx->qpids);
 		for (i = qid+1; i & rdev->qpmask; i++) {
-			entry = kmalloc(sizeof *entry, GFP_KERNEL);
+			entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 			if (!entry)
 				goto out;
 			entry->qid = i;
@@ -165,7 +165,7 @@ void c4iw_put_cqid(struct c4iw_rdev *rdev, u32 qid,
 {
 	struct c4iw_qid_list *entry;
 
-	entry = kmalloc(sizeof *entry, GFP_KERNEL);
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 	if (!entry)
 		return;
 	pr_debug("qid 0x%x\n", qid);
@@ -200,7 +200,7 @@ u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		rdev->stats.qid.cur += rdev->qpmask + 1;
 		mutex_unlock(&rdev->stats.lock);
 		for (i = qid+1; i & rdev->qpmask; i++) {
-			entry = kmalloc(sizeof *entry, GFP_KERNEL);
+			entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 			if (!entry)
 				goto out;
 			entry->qid = i;
@@ -211,13 +211,13 @@ u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		 * now put the same ids on the cq list since they all
 		 * map to the same db/gts page.
 		 */
-		entry = kmalloc(sizeof *entry, GFP_KERNEL);
+		entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 		if (!entry)
 			goto out;
 		entry->qid = qid;
 		list_add_tail(&entry->entry, &uctx->cqids);
 		for (i = qid; i & rdev->qpmask; i++) {
-			entry = kmalloc(sizeof *entry, GFP_KERNEL);
+			entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 			if (!entry)
 				goto out;
 			entry->qid = i;
@@ -239,7 +239,7 @@ void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qid,
 {
 	struct c4iw_qid_list *entry;
 
-	entry = kmalloc(sizeof *entry, GFP_KERNEL);
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
 	if (!entry)
 		return;
 	pr_debug("qid 0x%x\n", qid);