Merge tag 'v5.3-rc8' into rdma.git for-next

To resolve dependencies in following patches

mlx5_ib.h conflict resolved by keeing both hunks

Linux 5.3-rc8

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

drivers/infiniband/hw/bnxt_re/qplib_rcfw.c

@@ -136,6 +136,13 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
 		spin_unlock_irqrestore(&cmdq->lock, flags);
 		return -EBUSY;
 	}
+
+	size = req->cmd_size;
+	/* change the cmd_size to the number of 16byte cmdq unit.
+	 * req->cmd_size is modified here
+	 */
+	bnxt_qplib_set_cmd_slots(req);
+
 	memset(resp, 0, sizeof(*resp));
 	crsqe->resp = (struct creq_qp_event *)resp;
 	crsqe->resp->cookie = req->cookie;
@@ -150,7 +157,6 @@ static int __send_message(struct bnxt_qplib_rcfw *rcfw, struct cmdq_base *req,
 
 	cmdq_ptr = (struct bnxt_qplib_cmdqe **)cmdq->pbl_ptr;
 	preq = (u8 *)req;
-	size = req->cmd_size * BNXT_QPLIB_CMDQE_UNITS;
 	do {
 		/* Locate the next cmdq slot */
 		sw_prod = HWQ_CMP(cmdq->prod, cmdq);

drivers/infiniband/hw/bnxt_re/qplib_rcfw.h

@@ -55,9 +55,7 @@
 	do {								\
 		memset(&(req), 0, sizeof((req)));			\
 		(req).opcode = CMDQ_BASE_OPCODE_##CMD;			\
-		(req).cmd_size = (sizeof((req)) +			\
-				BNXT_QPLIB_CMDQE_UNITS - 1) /		\
-				BNXT_QPLIB_CMDQE_UNITS;			\
+		(req).cmd_size = sizeof((req));				\
 		(req).flags = cpu_to_le16(cmd_flags);			\
 	} while (0)
 
@@ -95,6 +93,13 @@ static inline u32 bnxt_qplib_cmdqe_cnt_per_pg(u32 depth)
 		 BNXT_QPLIB_CMDQE_UNITS);
 }
 
+/* Set the cmd_size to a factor of CMDQE unit */
+static inline void bnxt_qplib_set_cmd_slots(struct cmdq_base *req)
+{
+	req->cmd_size = (req->cmd_size + BNXT_QPLIB_CMDQE_UNITS - 1) /
+			 BNXT_QPLIB_CMDQE_UNITS;
+}
+
 #define MAX_CMDQ_IDX(depth)		((depth) - 1)
 
 static inline u32 bnxt_qplib_max_cmdq_idx_per_pg(u32 depth)

drivers/infiniband/hw/hfi1/fault.c

@@ -141,12 +141,14 @@ static ssize_t fault_opcodes_write(struct file *file, const char __user *buf,
 	if (!data)
 		return -ENOMEM;
 	copy = min(len, datalen - 1);
-	if (copy_from_user(data, buf, copy))
-		return -EFAULT;
+	if (copy_from_user(data, buf, copy)) {
+		ret = -EFAULT;
+		goto free_data;
+	}
 
 	ret = debugfs_file_get(file->f_path.dentry);
 	if (unlikely(ret))
-		return ret;
+		goto free_data;
 	ptr = data;
 	token = ptr;
 	for (ptr = data; *ptr; ptr = end + 1, token = ptr) {
@@ -195,6 +197,7 @@ static ssize_t fault_opcodes_write(struct file *file, const char __user *buf,
 	ret = len;
 
 	debugfs_file_put(file->f_path.dentry);
+free_data:
 	kfree(data);
 	return ret;
 }
@@ -214,7 +217,7 @@ static ssize_t fault_opcodes_read(struct file *file, char __user *buf,
 		return -ENOMEM;
 	ret = debugfs_file_get(file->f_path.dentry);
 	if (unlikely(ret))
-		return ret;
+		goto free_data;
 	bit = find_first_bit(fault->opcodes, bitsize);
 	while (bit < bitsize) {
 		zero = find_next_zero_bit(fault->opcodes, bitsize, bit);
@@ -232,6 +235,7 @@ static ssize_t fault_opcodes_read(struct file *file, char __user *buf,
 	data[size - 1] = '\n';
 	data[size] = '\0';
 	ret = simple_read_from_buffer(buf, len, pos, data, size);
+free_data:
 	kfree(data);
 	return ret;
 }

drivers/infiniband/hw/hfi1/tid_rdma.c

@@ -2574,18 +2574,9 @@ void hfi1_kern_read_tid_flow_free(struct rvt_qp *qp)
 	hfi1_kern_clear_hw_flow(priv->rcd, qp);
 }
 
-static bool tid_rdma_tid_err(struct hfi1_ctxtdata *rcd,
-			     struct hfi1_packet *packet, u8 rcv_type,
-			     u8 opcode)
+static bool tid_rdma_tid_err(struct hfi1_packet *packet, u8 rcv_type)
 {
 	struct rvt_qp *qp = packet->qp;
-	struct hfi1_qp_priv *qpriv = qp->priv;
-	u32 ipsn;
-	struct ib_other_headers *ohdr = packet->ohdr;
-	struct rvt_ack_entry *e;
-	struct tid_rdma_request *req;
-	struct rvt_dev_info *rdi = ib_to_rvt(qp->ibqp.device);
-	u32 i;
 
 	if (rcv_type >= RHF_RCV_TYPE_IB)
 		goto done;
@@ -2602,41 +2593,9 @@ static bool tid_rdma_tid_err(struct hfi1_ctxtdata *rcd,
 	if (rcv_type == RHF_RCV_TYPE_EAGER) {
 		hfi1_restart_rc(qp, qp->s_last_psn + 1, 1);
 		hfi1_schedule_send(qp);
-		goto done_unlock;
 	}
 
-	/*
-	 * For TID READ response, error out QP after freeing the tid
-	 * resources.
-	 */
-	if (opcode == TID_OP(READ_RESP)) {
-		ipsn = mask_psn(be32_to_cpu(ohdr->u.tid_rdma.r_rsp.verbs_psn));
-		if (cmp_psn(ipsn, qp->s_last_psn) > 0 &&
-		    cmp_psn(ipsn, qp->s_psn) < 0) {
-			hfi1_kern_read_tid_flow_free(qp);
-			spin_unlock(&qp->s_lock);
-			rvt_rc_error(qp, IB_WC_LOC_QP_OP_ERR);
-			goto done;
-		}
-		goto done_unlock;
-	}
-
-	/*
-	 * Error out the qp for TID RDMA WRITE
-	 */
-	hfi1_kern_clear_hw_flow(qpriv->rcd, qp);
-	for (i = 0; i < rvt_max_atomic(rdi); i++) {
-		e = &qp->s_ack_queue[i];
-		if (e->opcode == TID_OP(WRITE_REQ)) {
-			req = ack_to_tid_req(e);
-			hfi1_kern_exp_rcv_clear_all(req);
-		}
-	}
-	spin_unlock(&qp->s_lock);
-	rvt_rc_error(qp, IB_WC_LOC_LEN_ERR);
-	goto done;
-
-done_unlock:
+	/* Since no payload is delivered, just drop the packet */
 	spin_unlock(&qp->s_lock);
 done:
 	return true;
@@ -2687,12 +2646,12 @@ static bool handle_read_kdeth_eflags(struct hfi1_ctxtdata *rcd,
 	u32 fpsn;
 
 	lockdep_assert_held(&qp->r_lock);
+	spin_lock(&qp->s_lock);
 	/* If the psn is out of valid range, drop the packet */
 	if (cmp_psn(ibpsn, qp->s_last_psn) < 0 ||
 	    cmp_psn(ibpsn, qp->s_psn) > 0)
-		return ret;
+		goto s_unlock;
 
-	spin_lock(&qp->s_lock);
 	/*
 	 * Note that NAKs implicitly ACK outstanding SEND and RDMA write
 	 * requests and implicitly NAK RDMA read and atomic requests issued
@@ -2740,9 +2699,12 @@ static bool handle_read_kdeth_eflags(struct hfi1_ctxtdata *rcd,
 
 		wqe = do_rc_completion(qp, wqe, ibp);
 		if (qp->s_acked == qp->s_tail)
-			break;
+			goto s_unlock;
 	}
 
+	if (qp->s_acked == qp->s_tail)
+		goto s_unlock;
+
 	/* Handle the eflags for the request */
 	if (wqe->wr.opcode != IB_WR_TID_RDMA_READ)
 		goto s_unlock;
@@ -2922,7 +2884,7 @@ bool hfi1_handle_kdeth_eflags(struct hfi1_ctxtdata *rcd,
 		if (lnh == HFI1_LRH_GRH)
 			goto r_unlock;
 
-		if (tid_rdma_tid_err(rcd, packet, rcv_type, opcode))
+		if (tid_rdma_tid_err(packet, rcv_type))
 			goto r_unlock;
 	}
 
@@ -2942,8 +2904,15 @@ bool hfi1_handle_kdeth_eflags(struct hfi1_ctxtdata *rcd,
 	 */
 	spin_lock(&qp->s_lock);
 	qpriv = qp->priv;
+	if (qpriv->r_tid_tail == HFI1_QP_WQE_INVALID ||
+	    qpriv->r_tid_tail == qpriv->r_tid_head)
+		goto unlock;
 	e = &qp->s_ack_queue[qpriv->r_tid_tail];
+	if (e->opcode != TID_OP(WRITE_REQ))
+		goto unlock;
 	req = ack_to_tid_req(e);
+	if (req->comp_seg == req->cur_seg)
+		goto unlock;
 	flow = &req->flows[req->clear_tail];
 	trace_hfi1_eflags_err_write(qp, rcv_type, rte, psn);
 	trace_hfi1_rsp_handle_kdeth_eflags(qp, psn);
@@ -4509,7 +4478,7 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 	struct rvt_swqe *wqe;
 	struct tid_rdma_request *req;
 	struct tid_rdma_flow *flow;
-	u32 aeth, psn, req_psn, ack_psn, resync_psn, ack_kpsn;
+	u32 aeth, psn, req_psn, ack_psn, flpsn, resync_psn, ack_kpsn;
 	unsigned long flags;
 	u16 fidx;
 
@@ -4538,6 +4507,9 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 		ack_kpsn--;
 	}
 
+	if (unlikely(qp->s_acked == qp->s_tail))
+		goto ack_op_err;
+
 	wqe = rvt_get_swqe_ptr(qp, qp->s_acked);
 
 	if (wqe->wr.opcode != IB_WR_TID_RDMA_WRITE)
@@ -4550,7 +4522,8 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 	trace_hfi1_tid_flow_rcv_tid_ack(qp, req->acked_tail, flow);
 
 	/* Drop stale ACK/NAK */
-	if (cmp_psn(psn, full_flow_psn(flow, flow->flow_state.spsn)) < 0)
+	if (cmp_psn(psn, full_flow_psn(flow, flow->flow_state.spsn)) < 0 ||
+	    cmp_psn(req_psn, flow->flow_state.resp_ib_psn) < 0)
 		goto ack_op_err;
 
 	while (cmp_psn(ack_kpsn,
@@ -4712,7 +4685,12 @@ done:
 		switch ((aeth >> IB_AETH_CREDIT_SHIFT) &
 			IB_AETH_CREDIT_MASK) {
 		case 0: /* PSN sequence error */
+			if (!req->flows)
+				break;
 			flow = &req->flows[req->acked_tail];
+			flpsn = full_flow_psn(flow, flow->flow_state.lpsn);
+			if (cmp_psn(psn, flpsn) > 0)
+				break;
 			trace_hfi1_tid_flow_rcv_tid_ack(qp, req->acked_tail,
 							flow);
 			req->r_ack_psn = mask_psn(be32_to_cpu(ohdr->bth[2]));

drivers/infiniband/hw/mlx4/mad.c

@@ -1677,8 +1677,6 @@ tx_err:
 				    tx_buf_size, DMA_TO_DEVICE);
 		kfree(tun_qp->tx_ring[i].buf.addr);
 	}
-	kfree(tun_qp->tx_ring);
-	tun_qp->tx_ring = NULL;
 	i = MLX4_NUM_TUNNEL_BUFS;
 err:
 	while (i > 0) {
@@ -1687,6 +1685,8 @@ err:
 				    rx_buf_size, DMA_FROM_DEVICE);
 		kfree(tun_qp->ring[i].addr);
 	}
+	kfree(tun_qp->tx_ring);
+	tun_qp->tx_ring = NULL;
 	kfree(tun_qp->ring);
 	tun_qp->ring = NULL;
 	return -ENOMEM;

drivers/infiniband/hw/mlx5/main.c

@@ -1023,7 +1023,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL;
 
 	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
-		if (MLX5_CAP_GEN(mdev, pg))
+		if (dev->odp_caps.general_caps & IB_ODP_SUPPORT)
 			props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
 		props->odp_caps = dev->odp_caps;
 	}
@@ -6179,6 +6179,8 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
 		dev->port[i].roce.last_port_state = IB_PORT_DOWN;
 	}
 
+	mlx5_ib_internal_fill_odp_caps(dev);
+
 	err = mlx5_ib_init_multiport_master(dev);
 	if (err)
 		return err;
@@ -6603,8 +6605,6 @@ static void mlx5_ib_stage_dev_res_cleanup(struct mlx5_ib_dev *dev)
 
 static int mlx5_ib_stage_odp_init(struct mlx5_ib_dev *dev)
 {
-	mlx5_ib_internal_fill_odp_caps(dev);
-
 	return mlx5_ib_odp_init_one(dev);
 }
 

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -1477,4 +1477,18 @@ int bfregn_to_uar_index(struct mlx5_ib_dev *dev,
 
 int mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter);
 u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num);
+
+static inline bool mlx5_ib_can_use_umr(struct mlx5_ib_dev *dev,
+				       bool do_modify_atomic)
+{
+	if (MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled))
+		return false;
+
+	if (do_modify_atomic &&
+	    MLX5_CAP_GEN(dev->mdev, atomic) &&
+	    MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled))
+		return false;
+
+	return true;
+}
 #endif /* MLX5_IB_H */

drivers/infiniband/hw/mlx5/mr.c

@@ -1311,9 +1311,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	if (err < 0)
 		return ERR_PTR(err);
 
-	use_umr = !MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled) &&
-		  (!MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled) ||
-		   !MLX5_CAP_GEN(dev->mdev, atomic));
+	use_umr = mlx5_ib_can_use_umr(dev, true);
 
 	if (order <= mr_cache_max_order(dev) && use_umr) {
 		mr = alloc_mr_from_cache(pd, umem, virt_addr, length, ncont,
@@ -1466,7 +1464,8 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 			goto err;
 	}
 
-	if (flags & IB_MR_REREG_TRANS && !use_umr_mtt_update(mr, addr, len)) {
+	if (!mlx5_ib_can_use_umr(dev, true) ||
+	    (flags & IB_MR_REREG_TRANS && !use_umr_mtt_update(mr, addr, len))) {
 		/*
 		 * UMR can't be used - MKey needs to be replaced.
 		 */

drivers/infiniband/hw/mlx5/odp.c

@@ -301,7 +301,8 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 
 	memset(caps, 0, sizeof(*caps));
 
-	if (!MLX5_CAP_GEN(dev->mdev, pg))
+	if (!MLX5_CAP_GEN(dev->mdev, pg) ||
+	    !mlx5_ib_can_use_umr(dev, true))
 		return;
 
 	caps->general_caps = IB_ODP_SUPPORT;
@@ -355,7 +356,8 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 
 	if (MLX5_CAP_GEN(dev->mdev, fixed_buffer_size) &&
 	    MLX5_CAP_GEN(dev->mdev, null_mkey) &&
-	    MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset))
+	    MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset) &&
+	    !MLX5_CAP_GEN(dev->mdev, umr_indirect_mkey_disabled))
 		caps->general_caps |= IB_ODP_SUPPORT_IMPLICIT;
 
 	return;
@@ -1549,8 +1551,10 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev)
 {
 	int ret = 0;
 
-	if (dev->odp_caps.general_caps & IB_ODP_SUPPORT)
-		ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_odp_ops);
+	if (!(dev->odp_caps.general_caps & IB_ODP_SUPPORT))
+		return ret;
+
+	ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_odp_ops);
 
 	if (dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT) {
 		ret = mlx5_cmd_null_mkey(dev->mdev, &dev->null_mkey);
@@ -1560,9 +1564,6 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev)
 		}
 	}
 
-	if (!MLX5_CAP_GEN(dev->mdev, pg))
-		return ret;
-
 	ret = mlx5_ib_create_pf_eq(dev, &dev->odp_pf_eq);
 
 	return ret;
@@ -1570,7 +1571,7 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev)
 
 void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *dev)
 {
-	if (!MLX5_CAP_GEN(dev->mdev, pg))
+	if (!(dev->odp_caps.general_caps & IB_ODP_SUPPORT))
 		return;
 
 	mlx5_ib_destroy_pf_eq(dev, &dev->odp_pf_eq);

drivers/infiniband/hw/mlx5/qp.c

@@ -4161,7 +4161,7 @@ static u64 get_xlt_octo(u64 bytes)
 	       MLX5_IB_UMR_OCTOWORD;
 }
 
-static __be64 frwr_mkey_mask(void)
+static __be64 frwr_mkey_mask(bool atomic)
 {
 	u64 result;
 
@@ -4174,10 +4174,12 @@ static __be64 frwr_mkey_mask(void)
 		MLX5_MKEY_MASK_LW		|
 		MLX5_MKEY_MASK_RR		|
 		MLX5_MKEY_MASK_RW		|
-		MLX5_MKEY_MASK_A		|
 		MLX5_MKEY_MASK_SMALL_FENCE	|
 		MLX5_MKEY_MASK_FREE;
 
+	if (atomic)
+		result |= MLX5_MKEY_MASK_A;
+
 	return cpu_to_be64(result);
 }
 
@@ -4203,7 +4205,7 @@ static __be64 sig_mkey_mask(void)
 }
 
 static void set_reg_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr,
-			    struct mlx5_ib_mr *mr, u8 flags)
+			    struct mlx5_ib_mr *mr, u8 flags, bool atomic)
 {
 	int size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size;
 
@@ -4211,7 +4213,7 @@ static void set_reg_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr,
 
 	umr->flags = flags;
 	umr->xlt_octowords = cpu_to_be16(get_xlt_octo(size));
-	umr->mkey_mask = frwr_mkey_mask();
+	umr->mkey_mask = frwr_mkey_mask(atomic);
 }
 
 static void set_linv_umr_seg(struct mlx5_wqe_umr_ctrl_seg *umr)
@@ -4810,10 +4812,22 @@ static int set_reg_wr(struct mlx5_ib_qp *qp,
 {
 	struct mlx5_ib_mr *mr = to_mmr(wr->mr);
 	struct mlx5_ib_pd *pd = to_mpd(qp->ibqp.pd);
+	struct mlx5_ib_dev *dev = to_mdev(pd->ibpd.device);
 	int mr_list_size = (mr->ndescs + mr->meta_ndescs) * mr->desc_size;
 	bool umr_inline = mr_list_size <= MLX5_IB_SQ_UMR_INLINE_THRESHOLD;
+	bool atomic = wr->access & IB_ACCESS_REMOTE_ATOMIC;
 	u8 flags = 0;
 
+	if (!mlx5_ib_can_use_umr(dev, atomic)) {
+		mlx5_ib_warn(to_mdev(qp->ibqp.device),
+			     "Fast update of %s for MR is disabled\n",
+			     (MLX5_CAP_GEN(dev->mdev,
+					   umr_modify_entity_size_disabled)) ?
+				     "entity size" :
+				     "atomic access");
+		return -EINVAL;
+	}
+
 	if (unlikely(wr->wr.send_flags & IB_SEND_INLINE)) {
 		mlx5_ib_warn(to_mdev(qp->ibqp.device),
 			     "Invalid IB_SEND_INLINE send flag\n");
@@ -4825,7 +4839,7 @@ static int set_reg_wr(struct mlx5_ib_qp *qp,
 	if (umr_inline)
 		flags |= MLX5_UMR_INLINE;
 
-	set_reg_umr_seg(*seg, mr, flags);
+	set_reg_umr_seg(*seg, mr, flags, atomic);
 	*seg += sizeof(struct mlx5_wqe_umr_ctrl_seg);
 	*size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16;
 	handle_post_send_edge(&qp->sq, seg, *size, cur_edge);