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

Pull rdma updates from Jason Gunthorpe:
 "This has been a smaller cycle with many of the commits being smallish
  code fixes and improvements across the drivers.

   - Driver updates for bnxt_re, cxgb4, hfi1, hns, mlx5, nes, qedr, and
     rxe

   - Memory window support in hns

   - mlx5 user API 'flow mutate/steering' allows accessing the full
     packet mangling and matching machinery from user space

   - Support inter-working with verbs API calls in the 'devx' mlx5 user
     API, and provide options to use devx with less privilege

   - Modernize the use of syfs and the device interface to use attribute
     groups and cdev properly for uverbs, and clean up some of the core
     code's device list management

   - More progress on net namespaces for RDMA devices

   - Consolidate driver BAR mmapping support into core code helpers and
     rework how RDMA holds poitners to mm_struct for get_user_pages
     cases

   - First pass to use 'dev_name' instead of ib_device->name

   - Device renaming for RDMA devices"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (242 commits)
  IB/mlx5: Add support for extended atomic operations
  RDMA/core: Fix comment for hw stats init for port == 0
  RDMA/core: Refactor ib_register_device() function
  RDMA/core: Fix unwinding flow in case of error to register device
  ib_srp: Remove WARN_ON in srp_terminate_io()
  IB/mlx5: Allow scatter to CQE without global signaled WRs
  IB/mlx5: Verify that driver supports user flags
  IB/mlx5: Support scatter to CQE for DC transport type
  RDMA/drivers: Use core provided API for registering device attributes
  RDMA/core: Allow existing drivers to set one sysfs group per device
  IB/rxe: Remove unnecessary enum values
  RDMA/umad: Use kernel API to allocate umad indexes
  RDMA/uverbs: Use kernel API to allocate uverbs indexes
  RDMA/core: Increase total number of RDMA ports across all devices
  IB/mlx4: Add port and TID to MAD debug print
  IB/mlx4: Enable debug print of SMPs
  RDMA/core: Rename ports_parent to ports_kobj
  RDMA/core: Do not expose unsupported counters
  IB/mlx4: Refer to the device kobject instead of ports_parent
  RDMA/nldev: Allow IB device rename through RDMA netlink
  ...

10 files changed, 94 insertions, 479 deletions

diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h
index 3461df002f81..83d2349188db 100644
--- a/drivers/infiniband/hw/qib/qib.h
+++ b/drivers/infiniband/hw/qib/qib.h
@@ -1390,13 +1390,13 @@ static inline u32 qib_get_hdrqtail(const struct qib_ctxtdata *rcd)
  */
 
 extern const char ib_qib_version[];
+extern const struct attribute_group qib_attr_group;
 
 int qib_device_create(struct qib_devdata *);
 void qib_device_remove(struct qib_devdata *);
 
 int qib_create_port_files(struct ib_device *ibdev, u8 port_num,
 			  struct kobject *kobj);
-int qib_verbs_register_sysfs(struct qib_devdata *);
 void qib_verbs_unregister_sysfs(struct qib_devdata *);
 /* Hook for sysfs read of QSFP */
 extern int qib_qsfp_dump(struct qib_pportdata *ppd, char *buf, int len);
diff --git a/drivers/infiniband/hw/qib/qib_qp.c b/drivers/infiniband/hw/qib/qib_qp.c
index 344e401915f7..a81905df2d0f 100644
--- a/drivers/infiniband/hw/qib/qib_qp.c
+++ b/drivers/infiniband/hw/qib/qib_qp.c
@@ -378,25 +378,22 @@ void qib_flush_qp_waiters(struct rvt_qp *qp)
  * qib_check_send_wqe - validate wr/wqe
  * @qp - The qp
  * @wqe - The built wqe
+ * @call_send - Determine if the send should be posted or scheduled
  *
- * validate wr/wqe.  This is called
- * prior to inserting the wqe into
- * the ring but after the wqe has been
- * setup.
- *
- * Returns 1 to force direct progress, 0 otherwise, -EINVAL on failure
+ * Returns 0 on success, -EINVAL on failure
  */
 int qib_check_send_wqe(struct rvt_qp *qp,
-		       struct rvt_swqe *wqe)
+		       struct rvt_swqe *wqe, bool *call_send)
 {
 	struct rvt_ah *ah;
-	int ret = 0;
 
 	switch (qp->ibqp.qp_type) {
 	case IB_QPT_RC:
 	case IB_QPT_UC:
 		if (wqe->length > 0x80000000U)
 			return -EINVAL;
+		if (wqe->length > qp->pmtu)
+			*call_send = false;
 		break;
 	case IB_QPT_SMI:
 	case IB_QPT_GSI:
@@ -405,12 +402,12 @@ int qib_check_send_wqe(struct rvt_qp *qp,
 		if (wqe->length > (1 << ah->log_pmtu))
 			return -EINVAL;
 		/* progress hint */
-		ret = 1;
+		*call_send = true;
 		break;
 	default:
 		break;
 	}
-	return ret;
+	return 0;
 }
 
 #ifdef CONFIG_DEBUG_FS
diff --git a/drivers/infiniband/hw/qib/qib_rc.c b/drivers/infiniband/hw/qib/qib_rc.c
index f35fdeb14347..6fa002940451 100644
--- a/drivers/infiniband/hw/qib/qib_rc.c
+++ b/drivers/infiniband/hw/qib/qib_rc.c
@@ -254,7 +254,7 @@ int qib_make_rc_req(struct rvt_qp *qp, unsigned long *flags)
 			goto bail;
 		}
 		wqe = rvt_get_swqe_ptr(qp, qp->s_last);
-		qib_send_complete(qp, wqe, qp->s_last != qp->s_acked ?
+		rvt_send_complete(qp, wqe, qp->s_last != qp->s_acked ?
 			IB_WC_SUCCESS : IB_WC_WR_FLUSH_ERR);
 		/* will get called again */
 		goto done;
@@ -838,7 +838,7 @@ void qib_restart_rc(struct rvt_qp *qp, u32 psn, int wait)
 			qib_migrate_qp(qp);
 			qp->s_retry = qp->s_retry_cnt;
 		} else if (qp->s_last == qp->s_acked) {
-			qib_send_complete(qp, wqe, IB_WC_RETRY_EXC_ERR);
+			rvt_send_complete(qp, wqe, IB_WC_RETRY_EXC_ERR);
 			rvt_error_qp(qp, IB_WC_WR_FLUSH_ERR);
 			return;
 		} else /* XXX need to handle delayed completion */
@@ -1221,7 +1221,7 @@ static int do_rc_ack(struct rvt_qp *qp, u32 aeth, u32 psn, int opcode,
 			ibp->rvp.n_other_naks++;
 class_b:
 			if (qp->s_last == qp->s_acked) {
-				qib_send_complete(qp, wqe, status);
+				rvt_send_complete(qp, wqe, status);
 				rvt_error_qp(qp, IB_WC_WR_FLUSH_ERR);
 			}
 			break;
@@ -1425,7 +1425,8 @@ read_middle:
 		qp->s_rdma_read_len -= pmtu;
 		update_last_psn(qp, psn);
 		spin_unlock_irqrestore(&qp->s_lock, flags);
-		qib_copy_sge(&qp->s_rdma_read_sge, data, pmtu, 0);
+		rvt_copy_sge(qp, &qp->s_rdma_read_sge,
+			     data, pmtu, false, false);
 		goto bail;
 
 	case OP(RDMA_READ_RESPONSE_ONLY):
@@ -1471,7 +1472,8 @@ read_last:
 		if (unlikely(tlen != qp->s_rdma_read_len))
 			goto ack_len_err;
 		aeth = be32_to_cpu(ohdr->u.aeth);
-		qib_copy_sge(&qp->s_rdma_read_sge, data, tlen, 0);
+		rvt_copy_sge(qp, &qp->s_rdma_read_sge,
+			     data, tlen, false, false);
 		WARN_ON(qp->s_rdma_read_sge.num_sge);
 		(void) do_rc_ack(qp, aeth, psn,
 				 OP(RDMA_READ_RESPONSE_LAST), 0, rcd);
@@ -1490,7 +1492,7 @@ ack_len_err:
 	status = IB_WC_LOC_LEN_ERR;
 ack_err:
 	if (qp->s_last == qp->s_acked) {
-		qib_send_complete(qp, wqe, status);
+		rvt_send_complete(qp, wqe, status);
 		rvt_error_qp(qp, IB_WC_WR_FLUSH_ERR);
 	}
 ack_done:
@@ -1844,7 +1846,7 @@ send_middle:
 		qp->r_rcv_len += pmtu;
 		if (unlikely(qp->r_rcv_len > qp->r_len))
 			goto nack_inv;
-		qib_copy_sge(&qp->r_sge, data, pmtu, 1);
+		rvt_copy_sge(qp, &qp->r_sge, data, pmtu, true, false);
 		break;
 
 	case OP(RDMA_WRITE_LAST_WITH_IMMEDIATE):
@@ -1890,7 +1892,7 @@ send_last:
 		wc.byte_len = tlen + qp->r_rcv_len;
 		if (unlikely(wc.byte_len > qp->r_len))
 			goto nack_inv;
-		qib_copy_sge(&qp->r_sge, data, tlen, 1);
+		rvt_copy_sge(qp, &qp->r_sge, data, tlen, true, false);
 		rvt_put_ss(&qp->r_sge);
 		qp->r_msn++;
 		if (!test_and_clear_bit(RVT_R_WRID_VALID, &qp->r_aflags))
diff --git a/drivers/infiniband/hw/qib/qib_ruc.c b/drivers/infiniband/hw/qib/qib_ruc.c
index f8a7de795beb..1fa21938f310 100644
--- a/drivers/infiniband/hw/qib/qib_ruc.c
+++ b/drivers/infiniband/hw/qib/qib_ruc.c
@@ -171,307 +171,6 @@ err:
 }
 
 /**
- * qib_ruc_loopback - handle UC and RC lookback requests
- * @sqp: the sending QP
- *
- * This is called from qib_do_send() to
- * forward a WQE addressed to the same HCA.
- * Note that although we are single threaded due to the tasklet, we still
- * have to protect against post_send().  We don't have to worry about
- * receive interrupts since this is a connected protocol and all packets
- * will pass through here.
- */
-static void qib_ruc_loopback(struct rvt_qp *sqp)
-{
-	struct qib_ibport *ibp = to_iport(sqp->ibqp.device, sqp->port_num);
-	struct qib_pportdata *ppd = ppd_from_ibp(ibp);
-	struct qib_devdata *dd = ppd->dd;
-	struct rvt_dev_info *rdi = &dd->verbs_dev.rdi;
-	struct rvt_qp *qp;
-	struct rvt_swqe *wqe;
-	struct rvt_sge *sge;
-	unsigned long flags;
-	struct ib_wc wc;
-	u64 sdata;
-	atomic64_t *maddr;
-	enum ib_wc_status send_status;
-	int release;
-	int ret;
-
-	rcu_read_lock();
-	/*
-	 * Note that we check the responder QP state after
-	 * checking the requester's state.
-	 */
-	qp = rvt_lookup_qpn(rdi, &ibp->rvp, sqp->remote_qpn);
-	if (!qp)
-		goto done;
-
-	spin_lock_irqsave(&sqp->s_lock, flags);
-
-	/* Return if we are already busy processing a work request. */
-	if ((sqp->s_flags & (RVT_S_BUSY | RVT_S_ANY_WAIT)) ||
-	    !(ib_rvt_state_ops[sqp->state] & RVT_PROCESS_OR_FLUSH_SEND))
-		goto unlock;
-
-	sqp->s_flags |= RVT_S_BUSY;
-
-again:
-	if (sqp->s_last == READ_ONCE(sqp->s_head))
-		goto clr_busy;
-	wqe = rvt_get_swqe_ptr(sqp, sqp->s_last);
-
-	/* Return if it is not OK to start a new work reqeust. */
-	if (!(ib_rvt_state_ops[sqp->state] & RVT_PROCESS_NEXT_SEND_OK)) {
-		if (!(ib_rvt_state_ops[sqp->state] & RVT_FLUSH_SEND))
-			goto clr_busy;
-		/* We are in the error state, flush the work request. */
-		send_status = IB_WC_WR_FLUSH_ERR;
-		goto flush_send;
-	}
-
-	/*
-	 * We can rely on the entry not changing without the s_lock
-	 * being held until we update s_last.
-	 * We increment s_cur to indicate s_last is in progress.
-	 */
-	if (sqp->s_last == sqp->s_cur) {
-		if (++sqp->s_cur >= sqp->s_size)
-			sqp->s_cur = 0;
-	}
-	spin_unlock_irqrestore(&sqp->s_lock, flags);
-
-	if (!qp || !(ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) ||
-	    qp->ibqp.qp_type != sqp->ibqp.qp_type) {
-		ibp->rvp.n_pkt_drops++;
-		/*
-		 * For RC, the requester would timeout and retry so
-		 * shortcut the timeouts and just signal too many retries.
-		 */
-		if (sqp->ibqp.qp_type == IB_QPT_RC)
-			send_status = IB_WC_RETRY_EXC_ERR;
-		else
-			send_status = IB_WC_SUCCESS;
-		goto serr;
-	}
-
-	memset(&wc, 0, sizeof(wc));
-	send_status = IB_WC_SUCCESS;
-
-	release = 1;
-	sqp->s_sge.sge = wqe->sg_list[0];
-	sqp->s_sge.sg_list = wqe->sg_list + 1;
-	sqp->s_sge.num_sge = wqe->wr.num_sge;
-	sqp->s_len = wqe->length;
-	switch (wqe->wr.opcode) {
-	case IB_WR_SEND_WITH_IMM:
-		wc.wc_flags = IB_WC_WITH_IMM;
-		wc.ex.imm_data = wqe->wr.ex.imm_data;
-		/* FALLTHROUGH */
-	case IB_WR_SEND:
-		ret = rvt_get_rwqe(qp, false);
-		if (ret < 0)
-			goto op_err;
-		if (!ret)
-			goto rnr_nak;
-		break;
-
-	case IB_WR_RDMA_WRITE_WITH_IMM:
-		if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
-			goto inv_err;
-		wc.wc_flags = IB_WC_WITH_IMM;
-		wc.ex.imm_data = wqe->wr.ex.imm_data;
-		ret = rvt_get_rwqe(qp, true);
-		if (ret < 0)
-			goto op_err;
-		if (!ret)
-			goto rnr_nak;
-		/* FALLTHROUGH */
-	case IB_WR_RDMA_WRITE:
-		if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_WRITE)))
-			goto inv_err;
-		if (wqe->length == 0)
-			break;
-		if (unlikely(!rvt_rkey_ok(qp, &qp->r_sge.sge, wqe->length,
-					  wqe->rdma_wr.remote_addr,
-					  wqe->rdma_wr.rkey,
-					  IB_ACCESS_REMOTE_WRITE)))
-			goto acc_err;
-		qp->r_sge.sg_list = NULL;
-		qp->r_sge.num_sge = 1;
-		qp->r_sge.total_len = wqe->length;
-		break;
-
-	case IB_WR_RDMA_READ:
-		if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_READ)))
-			goto inv_err;
-		if (unlikely(!rvt_rkey_ok(qp, &sqp->s_sge.sge, wqe->length,
-					  wqe->rdma_wr.remote_addr,
-					  wqe->rdma_wr.rkey,
-					  IB_ACCESS_REMOTE_READ)))
-			goto acc_err;
-		release = 0;
-		sqp->s_sge.sg_list = NULL;
-		sqp->s_sge.num_sge = 1;
-		qp->r_sge.sge = wqe->sg_list[0];
-		qp->r_sge.sg_list = wqe->sg_list + 1;
-		qp->r_sge.num_sge = wqe->wr.num_sge;
-		qp->r_sge.total_len = wqe->length;
-		break;
-
-	case IB_WR_ATOMIC_CMP_AND_SWP:
-	case IB_WR_ATOMIC_FETCH_AND_ADD:
-		if (unlikely(!(qp->qp_access_flags & IB_ACCESS_REMOTE_ATOMIC)))
-			goto inv_err;
-		if (unlikely(!rvt_rkey_ok(qp, &qp->r_sge.sge, sizeof(u64),
-					  wqe->atomic_wr.remote_addr,
-					  wqe->atomic_wr.rkey,
-					  IB_ACCESS_REMOTE_ATOMIC)))
-			goto acc_err;
-		/* Perform atomic OP and save result. */
-		maddr = (atomic64_t *) qp->r_sge.sge.vaddr;
-		sdata = wqe->atomic_wr.compare_add;
-		*(u64 *) sqp->s_sge.sge.vaddr =
-			(wqe->atomic_wr.wr.opcode == IB_WR_ATOMIC_FETCH_AND_ADD) ?
-			(u64) atomic64_add_return(sdata, maddr) - sdata :
-			(u64) cmpxchg((u64 *) qp->r_sge.sge.vaddr,
-				      sdata, wqe->atomic_wr.swap);
-		rvt_put_mr(qp->r_sge.sge.mr);
-		qp->r_sge.num_sge = 0;
-		goto send_comp;
-
-	default:
-		send_status = IB_WC_LOC_QP_OP_ERR;
-		goto serr;
-	}
-
-	sge = &sqp->s_sge.sge;
-	while (sqp->s_len) {
-		u32 len = sqp->s_len;
-
-		if (len > sge->length)
-			len = sge->length;
-		if (len > sge->sge_length)
-			len = sge->sge_length;
-		BUG_ON(len == 0);
-		qib_copy_sge(&qp->r_sge, sge->vaddr, len, release);
-		sge->vaddr += len;
-		sge->length -= len;
-		sge->sge_length -= len;
-		if (sge->sge_length == 0) {
-			if (!release)
-				rvt_put_mr(sge->mr);
-			if (--sqp->s_sge.num_sge)
-				*sge = *sqp->s_sge.sg_list++;
-		} else if (sge->length == 0 && sge->mr->lkey) {
-			if (++sge->n >= RVT_SEGSZ) {
-				if (++sge->m >= sge->mr->mapsz)
-					break;
-				sge->n = 0;
-			}
-			sge->vaddr =
-				sge->mr->map[sge->m]->segs[sge->n].vaddr;
-			sge->length =
-				sge->mr->map[sge->m]->segs[sge->n].length;
-		}
-		sqp->s_len -= len;
-	}
-	if (release)
-		rvt_put_ss(&qp->r_sge);
-
-	if (!test_and_clear_bit(RVT_R_WRID_VALID, &qp->r_aflags))
-		goto send_comp;
-
-	if (wqe->wr.opcode == IB_WR_RDMA_WRITE_WITH_IMM)
-		wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
-	else
-		wc.opcode = IB_WC_RECV;
-	wc.wr_id = qp->r_wr_id;
-	wc.status = IB_WC_SUCCESS;
-	wc.byte_len = wqe->length;
-	wc.qp = &qp->ibqp;
-	wc.src_qp = qp->remote_qpn;
-	wc.slid = rdma_ah_get_dlid(&qp->remote_ah_attr);
-	wc.sl = rdma_ah_get_sl(&qp->remote_ah_attr);
-	wc.port_num = 1;
-	/* Signal completion event if the solicited bit is set. */
-	rvt_cq_enter(ibcq_to_rvtcq(qp->ibqp.recv_cq), &wc,
-		     wqe->wr.send_flags & IB_SEND_SOLICITED);
-
-send_comp:
-	spin_lock_irqsave(&sqp->s_lock, flags);
-	ibp->rvp.n_loop_pkts++;
-flush_send:
-	sqp->s_rnr_retry = sqp->s_rnr_retry_cnt;
-	qib_send_complete(sqp, wqe, send_status);
-	goto again;
-
-rnr_nak:
-	/* Handle RNR NAK */
-	if (qp->ibqp.qp_type == IB_QPT_UC)
-		goto send_comp;
-	ibp->rvp.n_rnr_naks++;
-	/*
-	 * Note: we don't need the s_lock held since the BUSY flag
-	 * makes this single threaded.
-	 */
-	if (sqp->s_rnr_retry == 0) {
-		send_status = IB_WC_RNR_RETRY_EXC_ERR;
-		goto serr;
-	}
-	if (sqp->s_rnr_retry_cnt < 7)
-		sqp->s_rnr_retry--;
-	spin_lock_irqsave(&sqp->s_lock, flags);
-	if (!(ib_rvt_state_ops[sqp->state] & RVT_PROCESS_RECV_OK))
-		goto clr_busy;
-	rvt_add_rnr_timer(sqp, qp->r_min_rnr_timer <<
-				IB_AETH_CREDIT_SHIFT);
-	goto clr_busy;
-
-op_err:
-	send_status = IB_WC_REM_OP_ERR;
-	wc.status = IB_WC_LOC_QP_OP_ERR;
-	goto err;
-
-inv_err:
-	send_status = IB_WC_REM_INV_REQ_ERR;
-	wc.status = IB_WC_LOC_QP_OP_ERR;
-	goto err;
-
-acc_err:
-	send_status = IB_WC_REM_ACCESS_ERR;
-	wc.status = IB_WC_LOC_PROT_ERR;
-err:
-	/* responder goes to error state */
-	rvt_rc_error(qp, wc.status);
-
-serr:
-	spin_lock_irqsave(&sqp->s_lock, flags);
-	qib_send_complete(sqp, wqe, send_status);
-	if (sqp->ibqp.qp_type == IB_QPT_RC) {
-		int lastwqe = rvt_error_qp(sqp, IB_WC_WR_FLUSH_ERR);
-
-		sqp->s_flags &= ~RVT_S_BUSY;
-		spin_unlock_irqrestore(&sqp->s_lock, flags);
-		if (lastwqe) {
-			struct ib_event ev;
-
-			ev.device = sqp->ibqp.device;
-			ev.element.qp = &sqp->ibqp;
-			ev.event = IB_EVENT_QP_LAST_WQE_REACHED;
-			sqp->ibqp.event_handler(&ev, sqp->ibqp.qp_context);
-		}
-		goto done;
-	}
-clr_busy:
-	sqp->s_flags &= ~RVT_S_BUSY;
-unlock:
-	spin_unlock_irqrestore(&sqp->s_lock, flags);
-done:
-	rcu_read_unlock();
-}
-
-/**
  * qib_make_grh - construct a GRH header
  * @ibp: a pointer to the IB port
  * @hdr: a pointer to the GRH header being constructed
@@ -573,7 +272,7 @@ void qib_do_send(struct rvt_qp *qp)
 	     qp->ibqp.qp_type == IB_QPT_UC) &&
 	    (rdma_ah_get_dlid(&qp->remote_ah_attr) &
 	     ~((1 << ppd->lmc) - 1)) == ppd->lid) {
-		qib_ruc_loopback(qp);
+		rvt_ruc_loopback(qp);
 		return;
 	}
 
@@ -613,42 +312,3 @@ void qib_do_send(struct rvt_qp *qp)
 
 	spin_unlock_irqrestore(&qp->s_lock, flags);
 }
-
-/*
- * This should be called with s_lock held.
- */
-void qib_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
-		       enum ib_wc_status status)
-{
-	u32 old_last, last;
-
-	if (!(ib_rvt_state_ops[qp->state] & RVT_PROCESS_OR_FLUSH_SEND))
-		return;
-
-	last = qp->s_last;
-	old_last = last;
-	if (++last >= qp->s_size)
-		last = 0;
-	qp->s_last = last;
-	/* See post_send() */
-	barrier();
-	rvt_put_swqe(wqe);
-	if (qp->ibqp.qp_type == IB_QPT_UD ||
-	    qp->ibqp.qp_type == IB_QPT_SMI ||
-	    qp->ibqp.qp_type == IB_QPT_GSI)
-		atomic_dec(&ibah_to_rvtah(wqe->ud_wr.ah)->refcount);
-
-	rvt_qp_swqe_complete(qp,
-			     wqe,
-			     ib_qib_wc_opcode[wqe->wr.opcode],
-			     status);
-
-	if (qp->s_acked == old_last)
-		qp->s_acked = last;
-	if (qp->s_cur == old_last)
-		qp->s_cur = last;
-	if (qp->s_tail == old_last)
-		qp->s_tail = last;
-	if (qp->state == IB_QPS_SQD && last == qp->s_cur)
-		qp->s_draining = 0;
-}
diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c
index d0723d4aef5c..757d4c9d713d 100644
--- a/drivers/infiniband/hw/qib/qib_sdma.c
+++ b/drivers/infiniband/hw/qib/qib_sdma.c
@@ -651,7 +651,7 @@ unmap:
 		if (ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK)
 			rvt_error_qp(qp, IB_WC_GENERAL_ERR);
 	} else if (qp->s_wqe)
-		qib_send_complete(qp, qp->s_wqe, IB_WC_GENERAL_ERR);
+		rvt_send_complete(qp, qp->s_wqe, IB_WC_GENERAL_ERR);
 	spin_unlock(&qp->s_lock);
 	spin_unlock(&qp->r_lock);
 	/* return zero to process the next send work request */
diff --git a/drivers/infiniband/hw/qib/qib_sysfs.c b/drivers/infiniband/hw/qib/qib_sysfs.c
index ca2638d8f35e..1cf4ca3f23e3 100644
--- a/drivers/infiniband/hw/qib/qib_sysfs.c
+++ b/drivers/infiniband/hw/qib/qib_sysfs.c
@@ -551,17 +551,18 @@ static struct kobj_type qib_diagc_ktype = {
  * Start of per-unit (or driver, in some cases, but replicated
  * per unit) functions (these get a device *)
  */
-static ssize_t show_rev(struct device *device, struct device_attribute *attr,
-			char *buf)
+static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr,
+			   char *buf)
 {
 	struct qib_ibdev *dev =
 		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
 
 	return sprintf(buf, "%x\n", dd_from_dev(dev)->minrev);
 }
+static DEVICE_ATTR_RO(hw_rev);
 
-static ssize_t show_hca(struct device *device, struct device_attribute *attr,
-			char *buf)
+static ssize_t hca_type_show(struct device *device,
+			     struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
 		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
@@ -574,15 +575,18 @@ static ssize_t show_hca(struct device *device, struct device_attribute *attr,
 		ret = scnprintf(buf, PAGE_SIZE, "%s\n", dd->boardname);
 	return ret;
 }
+static DEVICE_ATTR_RO(hca_type);
+static DEVICE_ATTR(board_id, 0444, hca_type_show, NULL);
 
-static ssize_t show_version(struct device *device,
+static ssize_t version_show(struct device *device,
 			    struct device_attribute *attr, char *buf)
 {
 	/* The string printed here is already newline-terminated. */
 	return scnprintf(buf, PAGE_SIZE, "%s", (char *)ib_qib_version);
 }
+static DEVICE_ATTR_RO(version);
 
-static ssize_t show_boardversion(struct device *device,
+static ssize_t boardversion_show(struct device *device,
 				 struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
@@ -592,9 +596,9 @@ static ssize_t show_boardversion(struct device *device,
 	/* The string printed here is already newline-terminated. */
 	return scnprintf(buf, PAGE_SIZE, "%s", dd->boardversion);
 }
+static DEVICE_ATTR_RO(boardversion);
 
-
-static ssize_t show_localbus_info(struct device *device,
+static ssize_t localbus_info_show(struct device *device,
 				  struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
@@ -604,9 +608,9 @@ static ssize_t show_localbus_info(struct device *device,
 	/* The string printed here is already newline-terminated. */
 	return scnprintf(buf, PAGE_SIZE, "%s", dd->lbus_info);
 }
+static DEVICE_ATTR_RO(localbus_info);
 
-
-static ssize_t show_nctxts(struct device *device,
+static ssize_t nctxts_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
@@ -620,9 +624,10 @@ static ssize_t show_nctxts(struct device *device,
 			(dd->first_user_ctxt > dd->cfgctxts) ? 0 :
 			(dd->cfgctxts - dd->first_user_ctxt));
 }
+static DEVICE_ATTR_RO(nctxts);
 
-static ssize_t show_nfreectxts(struct device *device,
-			   struct device_attribute *attr, char *buf)
+static ssize_t nfreectxts_show(struct device *device,
+			       struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
 		container_of(device, struct qib_ibdev, rdi.ibdev.dev);
@@ -631,8 +636,9 @@ static ssize_t show_nfreectxts(struct device *device,
 	/* Return the number of free user ports (contexts) available. */
 	return scnprintf(buf, PAGE_SIZE, "%u\n", dd->freectxts);
 }
+static DEVICE_ATTR_RO(nfreectxts);
 
-static ssize_t show_serial(struct device *device,
+static ssize_t serial_show(struct device *device,
 			   struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
@@ -644,8 +650,9 @@ static ssize_t show_serial(struct device *device,
 	strcat(buf, "\n");
 	return strlen(buf);
 }
+static DEVICE_ATTR_RO(serial);
 
-static ssize_t store_chip_reset(struct device *device,
+static ssize_t chip_reset_store(struct device *device,
 				struct device_attribute *attr, const char *buf,
 				size_t count)
 {
@@ -663,11 +670,12 @@ static ssize_t store_chip_reset(struct device *device,
 bail:
 	return ret < 0 ? ret : count;
 }
+static DEVICE_ATTR_WO(chip_reset);
 
 /*
  * Dump tempsense regs. in decimal, to ease shell-scripts.
  */
-static ssize_t show_tempsense(struct device *device,
+static ssize_t tempsense_show(struct device *device,
 			      struct device_attribute *attr, char *buf)
 {
 	struct qib_ibdev *dev =
@@ -695,6 +703,7 @@ static ssize_t show_tempsense(struct device *device,
 				*(signed char *)(regvals + 7));
 	return ret;
 }
+static DEVICE_ATTR_RO(tempsense);
 
 /*
  * end of per-unit (or driver, in some cases, but replicated
@@ -702,30 +711,23 @@ static ssize_t show_tempsense(struct device *device,
  */
 
 /* start of per-unit file structures and support code */
-static DEVICE_ATTR(hw_rev, S_IRUGO, show_rev, NULL);
-static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL);
-static DEVICE_ATTR(board_id, S_IRUGO, show_hca, NULL);
-static DEVICE_ATTR(version, S_IRUGO, show_version, NULL);
-static DEVICE_ATTR(nctxts, S_IRUGO, show_nctxts, NULL);
-static DEVICE_ATTR(nfreectxts, S_IRUGO, show_nfreectxts, NULL);
-static DEVICE_ATTR(serial, S_IRUGO, show_serial, NULL);
-static DEVICE_ATTR(boardversion, S_IRUGO, show_boardversion, NULL);
-static DEVICE_ATTR(tempsense, S_IRUGO, show_tempsense, NULL);
-static DEVICE_ATTR(localbus_info, S_IRUGO, show_localbus_info, NULL);
-static DEVICE_ATTR(chip_reset, S_IWUSR, NULL, store_chip_reset);
-
-static struct device_attribute *qib_attributes[] = {
-	&dev_attr_hw_rev,
-	&dev_attr_hca_type,
-	&dev_attr_board_id,
-	&dev_attr_version,
-	&dev_attr_nctxts,
-	&dev_attr_nfreectxts,
-	&dev_attr_serial,
-	&dev_attr_boardversion,
-	&dev_attr_tempsense,
-	&dev_attr_localbus_info,
-	&dev_attr_chip_reset,
+static struct attribute *qib_attributes[] = {
+	&dev_attr_hw_rev.attr,
+	&dev_attr_hca_type.attr,
+	&dev_attr_board_id.attr,
+	&dev_attr_version.attr,
+	&dev_attr_nctxts.attr,
+	&dev_attr_nfreectxts.attr,
+	&dev_attr_serial.attr,
+	&dev_attr_boardversion.attr,
+	&dev_attr_tempsense.attr,
+	&dev_attr_localbus_info.attr,
+	&dev_attr_chip_reset.attr,
+	NULL,
+};
+
+const struct attribute_group qib_attr_group = {
+	.attrs = qib_attributes,
 };
 
 int qib_create_port_files(struct ib_device *ibdev, u8 port_num,
@@ -827,27 +829,6 @@ bail:
 }
 
 /*
- * Register and create our files in /sys/class/infiniband.
- */
-int qib_verbs_register_sysfs(struct qib_devdata *dd)
-{
-	struct ib_device *dev = &dd->verbs_dev.rdi.ibdev;
-	int i, ret;
-
-	for (i = 0; i < ARRAY_SIZE(qib_attributes); ++i) {
-		ret = device_create_file(&dev->dev, qib_attributes[i]);
-		if (ret)
-			goto bail;
-	}
-
-	return 0;
-bail:
-	for (i = 0; i < ARRAY_SIZE(qib_attributes); ++i)
-		device_remove_file(&dev->dev, qib_attributes[i]);
-	return ret;
-}
-
-/*
  * Unregister and remove our files in /sys/class/infiniband.
  */
 void qib_verbs_unregister_sysfs(struct qib_devdata *dd)
diff --git a/drivers/infiniband/hw/qib/qib_uc.c b/drivers/infiniband/hw/qib/qib_uc.c
index 3e54bc11e0ae..30c70ad0f4bf 100644
--- a/drivers/infiniband/hw/qib/qib_uc.c
+++ b/drivers/infiniband/hw/qib/qib_uc.c
@@ -68,7 +68,7 @@ int qib_make_uc_req(struct rvt_qp *qp, unsigned long *flags)
 			goto bail;
 		}
 		wqe = rvt_get_swqe_ptr(qp, qp->s_last);
-		qib_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR);
+		rvt_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR);
 		goto done;
 	}
 
@@ -359,7 +359,7 @@ send_first:
 		qp->r_rcv_len += pmtu;
 		if (unlikely(qp->r_rcv_len > qp->r_len))
 			goto rewind;
-		qib_copy_sge(&qp->r_sge, data, pmtu, 0);
+		rvt_copy_sge(qp, &qp->r_sge, data, pmtu, false, false);
 		break;
 
 	case OP(SEND_LAST_WITH_IMMEDIATE):
@@ -385,7 +385,7 @@ send_last:
 		if (unlikely(wc.byte_len > qp->r_len))
 			goto rewind;
 		wc.opcode = IB_WC_RECV;
-		qib_copy_sge(&qp->r_sge, data, tlen, 0);
+		rvt_copy_sge(qp, &qp->r_sge, data, tlen, false, false);
 		rvt_put_ss(&qp->s_rdma_read_sge);
 last_imm:
 		wc.wr_id = qp->r_wr_id;
@@ -449,7 +449,7 @@ rdma_first:
 		qp->r_rcv_len += pmtu;
 		if (unlikely(qp->r_rcv_len > qp->r_len))
 			goto drop;
-		qib_copy_sge(&qp->r_sge, data, pmtu, 1);
+		rvt_copy_sge(qp, &qp->r_sge, data, pmtu, true, false);
 		break;
 
 	case OP(RDMA_WRITE_LAST_WITH_IMMEDIATE):
@@ -479,7 +479,7 @@ rdma_last_imm:
 		}
 		wc.byte_len = qp->r_len;
 		wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
-		qib_copy_sge(&qp->r_sge, data, tlen, 1);
+		rvt_copy_sge(qp, &qp->r_sge, data, tlen, true, false);
 		rvt_put_ss(&qp->r_sge);
 		goto last_imm;
 
@@ -495,7 +495,7 @@ rdma_last:
 		tlen -= (hdrsize + pad + 4);
 		if (unlikely(tlen + qp->r_rcv_len != qp->r_len))
 			goto drop;
-		qib_copy_sge(&qp->r_sge, data, tlen, 1);
+		rvt_copy_sge(qp, &qp->r_sge, data, tlen, true, false);
 		rvt_put_ss(&qp->r_sge);
 		break;
 
diff --git a/drivers/infiniband/hw/qib/qib_ud.c b/drivers/infiniband/hw/qib/qib_ud.c
index f8d029a2390f..4d4c31ea4e2d 100644
--- a/drivers/infiniband/hw/qib/qib_ud.c
+++ b/drivers/infiniband/hw/qib/qib_ud.c
@@ -162,8 +162,8 @@ static void qib_ud_loopback(struct rvt_qp *sqp, struct rvt_swqe *swqe)
 		const struct ib_global_route *grd = rdma_ah_read_grh(ah_attr);
 
 		qib_make_grh(ibp, &grh, grd, 0, 0);
-		qib_copy_sge(&qp->r_sge, &grh,
-			     sizeof(grh), 1);
+		rvt_copy_sge(qp, &qp->r_sge, &grh,
+			     sizeof(grh), true, false);
 		wc.wc_flags |= IB_WC_GRH;
 	} else
 		rvt_skip_sge(&qp->r_sge, sizeof(struct ib_grh), true);
@@ -179,7 +179,7 @@ static void qib_ud_loopback(struct rvt_qp *sqp, struct rvt_swqe *swqe)
 		if (len > sge->sge_length)
 			len = sge->sge_length;
 		BUG_ON(len == 0);
-		qib_copy_sge(&qp->r_sge, sge->vaddr, len, 1);
+		rvt_copy_sge(qp, &qp->r_sge, sge->vaddr, len, true, false);
 		sge->vaddr += len;
 		sge->length -= len;
 		sge->sge_length -= len;
@@ -260,7 +260,7 @@ int qib_make_ud_req(struct rvt_qp *qp, unsigned long *flags)
 			goto bail;
 		}
 		wqe = rvt_get_swqe_ptr(qp, qp->s_last);
-		qib_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR);
+		rvt_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR);
 		goto done;
 	}
 
@@ -304,7 +304,7 @@ int qib_make_ud_req(struct rvt_qp *qp, unsigned long *flags)
 			qib_ud_loopback(qp, wqe);
 			spin_lock_irqsave(&qp->s_lock, tflags);
 			*flags = tflags;
-			qib_send_complete(qp, wqe, IB_WC_SUCCESS);
+			rvt_send_complete(qp, wqe, IB_WC_SUCCESS);
 			goto done;
 		}
 	}
@@ -551,12 +551,13 @@ void qib_ud_rcv(struct qib_ibport *ibp, struct ib_header *hdr,
 		goto drop;
 	}
 	if (has_grh) {
-		qib_copy_sge(&qp->r_sge, &hdr->u.l.grh,
-			     sizeof(struct ib_grh), 1);
+		rvt_copy_sge(qp, &qp->r_sge, &hdr->u.l.grh,
+			     sizeof(struct ib_grh), true, false);
 		wc.wc_flags |= IB_WC_GRH;
 	} else
 		rvt_skip_sge(&qp->r_sge, sizeof(struct ib_grh), true);
-	qib_copy_sge(&qp->r_sge, data, wc.byte_len - sizeof(struct ib_grh), 1);
+	rvt_copy_sge(qp, &qp->r_sge, data, wc.byte_len - sizeof(struct ib_grh),
+		     true, false);
 	rvt_put_ss(&qp->r_sge);
 	if (!test_and_clear_bit(RVT_R_WRID_VALID, &qp->r_aflags))
 		return;
diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c
index 41babbc0db58..4b0f5761a646 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.c
+++ b/drivers/infiniband/hw/qib/qib_verbs.c
@@ -131,27 +131,6 @@ const enum ib_wc_opcode ib_qib_wc_opcode[] = {
  */
 __be64 ib_qib_sys_image_guid;
 
-/**
- * qib_copy_sge - copy data to SGE memory
- * @ss: the SGE state
- * @data: the data to copy
- * @length: the length of the data
- */
-void qib_copy_sge(struct rvt_sge_state *ss, void *data, u32 length, int release)
-{
-	struct rvt_sge *sge = &ss->sge;
-
-	while (length) {
-		u32 len = rvt_get_sge_length(sge, length);
-
-		WARN_ON_ONCE(len == 0);
-		memcpy(sge->vaddr, data, len);
-		rvt_update_sge(ss, len, release);
-		data += len;
-		length -= len;
-	}
-}
-
 /*
  * Count the number of DMA descriptors needed to send length bytes of data.
  * Don't modify the qib_sge_state to get the count.
@@ -752,7 +731,7 @@ static void sdma_complete(struct qib_sdma_txreq *cookie, int status)
 
 	spin_lock(&qp->s_lock);
 	if (tx->wqe)
-		qib_send_complete(qp, tx->wqe, IB_WC_SUCCESS);
+		rvt_send_complete(qp, tx->wqe, IB_WC_SUCCESS);
 	else if (qp->ibqp.qp_type == IB_QPT_RC) {
 		struct ib_header *hdr;
 
@@ -1025,7 +1004,7 @@ done:
 	}
 	if (qp->s_wqe) {
 		spin_lock_irqsave(&qp->s_lock, flags);
-		qib_send_complete(qp, qp->s_wqe, IB_WC_SUCCESS);
+		rvt_send_complete(qp, qp->s_wqe, IB_WC_SUCCESS);
 		spin_unlock_irqrestore(&qp->s_lock, flags);
 	} else if (qp->ibqp.qp_type == IB_QPT_RC) {
 		spin_lock_irqsave(&qp->s_lock, flags);
@@ -1512,6 +1491,9 @@ static void qib_fill_device_attr(struct qib_devdata *dd)
 					rdi->dparms.props.max_mcast_grp;
 	/* post send table */
 	dd->verbs_dev.rdi.post_parms = qib_post_parms;
+
+	/* opcode translation table */
+	dd->verbs_dev.rdi.wc_opcode = ib_qib_wc_opcode;
 }
 
 /**
@@ -1588,7 +1570,7 @@ int qib_register_ib_device(struct qib_devdata *dd)
 	dd->verbs_dev.rdi.driver_f.port_callback = qib_create_port_files;
 	dd->verbs_dev.rdi.driver_f.get_pci_dev = qib_get_pci_dev;
 	dd->verbs_dev.rdi.driver_f.check_ah = qib_check_ah;
-	dd->verbs_dev.rdi.driver_f.check_send_wqe = qib_check_send_wqe;
+	dd->verbs_dev.rdi.driver_f.setup_wqe = qib_check_send_wqe;
 	dd->verbs_dev.rdi.driver_f.notify_new_ah = qib_notify_new_ah;
 	dd->verbs_dev.rdi.driver_f.alloc_qpn = qib_alloc_qpn;
 	dd->verbs_dev.rdi.driver_f.qp_priv_alloc = qib_qp_priv_alloc;
@@ -1631,6 +1613,7 @@ int qib_register_ib_device(struct qib_devdata *dd)
 	dd->verbs_dev.rdi.dparms.node = dd->assigned_node_id;
 	dd->verbs_dev.rdi.dparms.core_cap_flags = RDMA_CORE_PORT_IBA_IB;
 	dd->verbs_dev.rdi.dparms.max_mad_size = IB_MGMT_MAD_SIZE;
+	dd->verbs_dev.rdi.dparms.sge_copy_mode = RVT_SGE_COPY_MEMCPY;
 
 	qib_fill_device_attr(dd);
 
@@ -1642,19 +1625,14 @@ int qib_register_ib_device(struct qib_devdata *dd)
 			      i,
 			      dd->rcd[ctxt]->pkeys);
 	}
+	rdma_set_device_sysfs_group(&dd->verbs_dev.rdi.ibdev, &qib_attr_group);
 
 	ret = rvt_register_device(&dd->verbs_dev.rdi, RDMA_DRIVER_QIB);
 	if (ret)
 		goto err_tx;
 
-	ret = qib_verbs_register_sysfs(dd);
-	if (ret)
-		goto err_class;
-
 	return ret;
 
-err_class:
-	rvt_unregister_device(&dd->verbs_dev.rdi);
 err_tx:
 	while (!list_empty(&dev->txreq_free)) {
 		struct list_head *l = dev->txreq_free.next;
@@ -1716,14 +1694,14 @@ void qib_unregister_ib_device(struct qib_devdata *dd)
  * It is only used in post send, which doesn't hold
  * the s_lock.
  */
-void _qib_schedule_send(struct rvt_qp *qp)
+bool _qib_schedule_send(struct rvt_qp *qp)
 {
 	struct qib_ibport *ibp =
 		to_iport(qp->ibqp.device, qp->port_num);
 	struct qib_pportdata *ppd = ppd_from_ibp(ibp);
 	struct qib_qp_priv *priv = qp->priv;
 
-	queue_work(ppd->qib_wq, &priv->s_work);
+	return queue_work(ppd->qib_wq, &priv->s_work);
 }
 
 /**
@@ -1733,8 +1711,9 @@ void _qib_schedule_send(struct rvt_qp *qp)
  * This schedules qp progress.  The s_lock
  * should be held.
  */
-void qib_schedule_send(struct rvt_qp *qp)
+bool qib_schedule_send(struct rvt_qp *qp)
 {
 	if (qib_send_ok(qp))
-		_qib_schedule_send(qp);
+		return _qib_schedule_send(qp);
+	return false;
 }
diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h
index 666613eef88f..a4426c24b0d1 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.h
+++ b/drivers/infiniband/hw/qib/qib_verbs.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012 - 2017 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2012 - 2018 Intel Corporation.  All rights reserved.
  * Copyright (c) 2006 - 2012 QLogic Corporation. All rights reserved.
  * Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
  *
@@ -223,8 +223,8 @@ static inline int qib_send_ok(struct rvt_qp *qp)
 		 !(qp->s_flags & RVT_S_ANY_WAIT_SEND));
 }
 
-void _qib_schedule_send(struct rvt_qp *qp);
-void qib_schedule_send(struct rvt_qp *qp);
+bool _qib_schedule_send(struct rvt_qp *qp);
+bool qib_schedule_send(struct rvt_qp *qp);
 
 static inline int qib_pkey_ok(u16 pkey1, u16 pkey2)
 {
@@ -292,9 +292,6 @@ void qib_put_txreq(struct qib_verbs_txreq *tx);
 int qib_verbs_send(struct rvt_qp *qp, struct ib_header *hdr,
 		   u32 hdrwords, struct rvt_sge_state *ss, u32 len);
 
-void qib_copy_sge(struct rvt_sge_state *ss, void *data, u32 length,
-		  int release);
-
 void qib_uc_rcv(struct qib_ibport *ibp, struct ib_header *hdr,
 		int has_grh, void *data, u32 tlen, struct rvt_qp *qp);
 
@@ -303,7 +300,8 @@ void qib_rc_rcv(struct qib_ctxtdata *rcd, struct ib_header *hdr,
 
 int qib_check_ah(struct ib_device *ibdev, struct rdma_ah_attr *ah_attr);
 
-int qib_check_send_wqe(struct rvt_qp *qp, struct rvt_swqe *wqe);
+int qib_check_send_wqe(struct rvt_qp *qp, struct rvt_swqe *wqe,
+		       bool *call_send);
 
 struct ib_ah *qib_create_qp0_ah(struct qib_ibport *ibp, u16 dlid);
 
@@ -333,9 +331,6 @@ void _qib_do_send(struct work_struct *work);
 
 void qib_do_send(struct rvt_qp *qp);
 
-void qib_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
-		       enum ib_wc_status status);
-
 void qib_send_rc_ack(struct rvt_qp *qp);
 
 int qib_make_rc_req(struct rvt_qp *qp, unsigned long *flags);