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

Pull rdma fixes from Jason Gunthorpe:
"A small set of late-rc patches, mostly fixes for various crashers,
some syzkaller fixes and a mlx5 HW limitation:

- Several MAINTAINERS updates

- Memory leak regression in ODP

- Several fixes for syzkaller related crashes. Google recently taught
syzkaller to create the software RDMA devices

- Crash fixes for HFI1

- Several fixes for mlx5 crashes

- Prevent unprivileged access to an unsafe mlx5 HW resource"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
RDMA/mlx5: Block delay drop to unprivileged users
RDMA/mlx5: Fix access to wrong pointer while performing flush due to error
RDMA/core: Ensure security pkey modify is not lost
MAINTAINERS: Clean RXE section and add Zhu as RXE maintainer
IB/hfi1: Ensure pq is not left on waitlist
IB/rdmavt: Free kernel completion queue when done
RDMA/mad: Do not crash if the rdma device does not have a umad interface
RDMA/core: Fix missing error check on dev_set_name()
RDMA/nl: Do not permit empty devices names during RDMA_NLDEV_CMD_NEWLINK/SET
RDMA/mlx5: Fix the number of hwcounters of a dynamic counter
MAINTAINERS: Update maintainers for HISILICON ROCE DRIVER
RDMA/odp: Fix leaking the tgid for implicit ODP

MAINTAINERS

···
 
 HISILICON ROCE DRIVER
 M:	Lijun Ou <[e-mail]>
-M:	Wei Hu(Xavier) <[e-mail]>
+M:	Wei Hu(Xavier) <[e-mail]>
+M:	Weihang Li <[e-mail]>
 L:	[e-mail]
 S:	Maintained
 F:	drivers/infiniband/hw/hns/
···
 F:	include/uapi/rdma/siw-abi.h
 
 SOFT-ROCE DRIVER (rxe)
-M:	Moni Shoua <[e-mail]>
+M:	Zhu Yanjun <[e-mail]>
 L:	[e-mail]
 S:	Supported
-W:	https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home
-Q:	http://patchwork.kernel.org/project/linux-rdma/list/
 F:	drivers/infiniband/sw/rxe/
 F:	include/uapi/rdma/rdma_user_rxe.h
 

drivers/infiniband/core/device.c

···
 	cdev->dev.parent = device->dev.parent;
 	rdma_init_coredev(cdev, device, read_pnet(&rnet->net));
 	cdev->dev.release = compatdev_release;
-	dev_set_name(&cdev->dev, "%s", dev_name(&device->dev));
+	ret = dev_set_name(&cdev->dev, "%s", dev_name(&device->dev));
+	if (ret)
+		goto add_err;
 
 	ret = device_add(&cdev->dev);
 	if (ret)

drivers/infiniband/core/nldev.c

···
 
 		nla_strlcpy(name, tb[RDMA_NLDEV_ATTR_DEV_NAME],
 			    IB_DEVICE_NAME_MAX);
+		if (strlen(name) == 0) {
+			err = -EINVAL;
+			goto done;
+		}
 		err = ib_device_rename(device, name);
 		goto done;
 	}
···
 
 	nla_strlcpy(ibdev_name, tb[RDMA_NLDEV_ATTR_DEV_NAME],
 		    sizeof(ibdev_name));
-	if (strchr(ibdev_name, '%'))
+	if (strchr(ibdev_name, '%') || strlen(ibdev_name) == 0)
 		return -EINVAL;
 
 	nla_strlcpy(type, tb[RDMA_NLDEV_ATTR_LINK_TYPE], sizeof(type));

drivers/infiniband/core/security.c

···
 	else if (qp_pps)
 		new_pps->main.pkey_index = qp_pps->main.pkey_index;
 
-	if ((qp_attr_mask & IB_QP_PKEY_INDEX) && (qp_attr_mask & IB_QP_PORT))
+	if (((qp_attr_mask & IB_QP_PKEY_INDEX) &&
+	     (qp_attr_mask & IB_QP_PORT)) ||
+	    (qp_pps && qp_pps->main.state != IB_PORT_PKEY_NOT_VALID))
 		new_pps->main.state = IB_PORT_PKEY_VALID;
-
-	if (!(qp_attr_mask & (IB_QP_PKEY_INDEX | IB_QP_PORT)) && qp_pps) {
-		new_pps->main.port_num = qp_pps->main.port_num;
-		new_pps->main.pkey_index = qp_pps->main.pkey_index;
-		if (qp_pps->main.state != IB_PORT_PKEY_NOT_VALID)
-			new_pps->main.state = IB_PORT_PKEY_VALID;
-	}
 
 	if (qp_attr_mask & IB_QP_ALT_PATH) {
 		new_pps->alt.port_num = qp_attr->alt_port_num;

drivers/infiniband/core/umem_odp.c

···
 		mmu_interval_notifier_remove(&umem_odp->notifier);
 		kvfree(umem_odp->dma_list);
 		kvfree(umem_odp->page_list);
-		put_pid(umem_odp->tgid);
 	}
+	put_pid(umem_odp->tgid);
 	kfree(umem_odp);
 }
 EXPORT_SYMBOL(ib_umem_odp_release);

drivers/infiniband/core/user_mad.c

···
 	.llseek	 = no_llseek,
 };
 
+static struct ib_umad_port *get_port(struct ib_device *ibdev,
+				     struct ib_umad_device *umad_dev,
+				     unsigned int port)
+{
+	if (!umad_dev)
+		return ERR_PTR(-EOPNOTSUPP);
+	if (!rdma_is_port_valid(ibdev, port))
+		return ERR_PTR(-EINVAL);
+	if (!rdma_cap_ib_mad(ibdev, port))
+		return ERR_PTR(-EOPNOTSUPP);
+
+	return &umad_dev->ports[port - rdma_start_port(ibdev)];
+}
+
 static int ib_umad_get_nl_info(struct ib_device *ibdev, void *client_data,
 			       struct ib_client_nl_info *res)
 {
-	struct ib_umad_device *umad_dev = client_data;
+	struct ib_umad_port *port = get_port(ibdev, client_data, res->port);
 
-	if (!rdma_is_port_valid(ibdev, res->port))
-		return -EINVAL;
+	if (IS_ERR(port))
+		return PTR_ERR(port);
 
 	res->abi = IB_USER_MAD_ABI_VERSION;
-	res->cdev = &umad_dev->ports[res->port - rdma_start_port(ibdev)].dev;
-
+	res->cdev = &port->dev;
 	return 0;
 }
 
···
 static int ib_issm_get_nl_info(struct ib_device *ibdev, void *client_data,
 			       struct ib_client_nl_info *res)
 {
-	struct ib_umad_device *umad_dev =
-		ib_get_client_data(ibdev, &umad_client);
+	struct ib_umad_port *port = get_port(ibdev, client_data, res->port);
 
-	if (!rdma_is_port_valid(ibdev, res->port))
-		return -EINVAL;
+	if (IS_ERR(port))
+		return PTR_ERR(port);
 
 	res->abi = IB_USER_MAD_ABI_VERSION;
-	res->cdev = &umad_dev->ports[res->port - rdma_start_port(ibdev)].sm_dev;
-
+	res->cdev = &port->sm_dev;
 	return 0;
 }
 

drivers/infiniband/hw/hfi1/user_sdma.c

···
 	 */
 	xchg(&pq->state, SDMA_PKT_Q_DEFERRED);
 	if (list_empty(&pq->busy.list)) {
+		pq->busy.lock = &sde->waitlock;
 		iowait_get_priority(&pq->busy);
 		iowait_queue(pkts_sent, &pq->busy, &sde->dmawait);
 	}
···
 {
 	struct hfi1_user_sdma_pkt_q *pq =
 		container_of(wait, struct hfi1_user_sdma_pkt_q, busy);
+	pq->busy.lock = NULL;
 	xchg(&pq->state, SDMA_PKT_Q_ACTIVE);
 	wake_up(&wait->wait_dma);
 };
···
 	return ret;
 }
 
+static void flush_pq_iowait(struct hfi1_user_sdma_pkt_q *pq)
+{
+	unsigned long flags;
+	seqlock_t *lock = pq->busy.lock;
+
+	if (!lock)
+		return;
+	write_seqlock_irqsave(lock, flags);
+	if (!list_empty(&pq->busy.list)) {
+		list_del_init(&pq->busy.list);
+		pq->busy.lock = NULL;
+	}
+	write_sequnlock_irqrestore(lock, flags);
+}
+
 int hfi1_user_sdma_free_queues(struct hfi1_filedata *fd,
 			       struct hfi1_ctxtdata *uctxt)
 {
···
 		kfree(pq->reqs);
 		kfree(pq->req_in_use);
 		kmem_cache_destroy(pq->txreq_cache);
+		flush_pq_iowait(pq);
 		kfree(pq);
 	} else {
 		spin_unlock(&fd->pq_rcu_lock);
···
 		if (ret < 0) {
 			if (ret != -EBUSY)
 				goto free_req;
-			wait_event_interruptible_timeout(
+			if (wait_event_interruptible_timeout(
 				pq->busy.wait_dma,
-				(pq->state == SDMA_PKT_Q_ACTIVE),
+				pq->state == SDMA_PKT_Q_ACTIVE,
 				msecs_to_jiffies(
-					SDMA_IOWAIT_TIMEOUT));
+					SDMA_IOWAIT_TIMEOUT)) <= 0)
+				flush_pq_iowait(pq);
 		}
 	}
 	*count += idx;

drivers/infiniband/hw/mlx5/cq.c

···
 		dump_cqe(dev, cqe);
 }
 
+static void handle_atomics(struct mlx5_ib_qp *qp, struct mlx5_cqe64 *cqe64,
+			   u16 tail, u16 head)
+{
+	u16 idx;
+
+	do {
+		idx = tail & (qp->sq.wqe_cnt - 1);
+		if (idx == head)
+			break;
+
+		tail = qp->sq.w_list[idx].next;
+	} while (1);
+	tail = qp->sq.w_list[idx].next;
+	qp->sq.last_poll = tail;
+}
+
 static void free_cq_buf(struct mlx5_ib_dev *dev, struct mlx5_ib_cq_buf *buf)
 {
 	mlx5_frag_buf_free(dev->mdev, &buf->frag_buf);
···
 }
 
 static void sw_comp(struct mlx5_ib_qp *qp, int num_entries, struct ib_wc *wc,
-		    int *npolled, int is_send)
+		    int *npolled, bool is_send)
 {
 	struct mlx5_ib_wq *wq;
 	unsigned int cur;
···
 		return;
 
 	for (i = 0;  i < cur && np < num_entries; i++) {
-		wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
+		unsigned int idx;
+
+		idx = (is_send) ? wq->last_poll : wq->tail;
+		idx &= (wq->wqe_cnt - 1);
+		wc->wr_id = wq->wrid[idx];
 		wc->status = IB_WC_WR_FLUSH_ERR;
 		wc->vendor_err = MLX5_CQE_SYNDROME_WR_FLUSH_ERR;
 		wq->tail++;
+		if (is_send)
+			wq->last_poll = wq->w_list[idx].next;
 		np++;
 		wc->qp = &qp->ibqp;
 		wc++;
···
 		wqe_ctr = be16_to_cpu(cqe64->wqe_counter);
 		idx = wqe_ctr & (wq->wqe_cnt - 1);
 		handle_good_req(wc, cqe64, wq, idx);
+		handle_atomics(*cur_qp, cqe64, wq->last_poll, idx);
 		wc->wr_id = wq->wrid[idx];
 		wq->tail = wq->wqe_head[idx] + 1;
 		wc->status = IB_WC_SUCCESS;

drivers/infiniband/hw/mlx5/main.c

···
 	const struct mlx5_ib_counters *cnts =
 		get_counters(dev, counter->port - 1);
 
-	/* Q counters are in the beginning of all counters */
 	return rdma_alloc_hw_stats_struct(cnts->names,
-					  cnts->num_q_counters,
+					  cnts->num_q_counters +
+					  cnts->num_cong_counters +
+					  cnts->num_ext_ppcnt_counters,
 					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
 }
 

drivers/infiniband/hw/mlx5/mlx5_ib.h

···
 	unsigned		head;
 	unsigned		tail;
 	u16			cur_post;
+	u16			last_poll;
 	void			*cur_edge;
 };
 

drivers/infiniband/hw/mlx5/qp.c

···
 		qp->sq.cur_post = 0;
 		if (qp->sq.wqe_cnt)
 			qp->sq.cur_edge = get_sq_edge(&qp->sq, 0);
+		qp->sq.last_poll = 0;
 		qp->db.db[MLX5_RCV_DBR] = 0;
 		qp->db.db[MLX5_SND_DBR] = 0;
 	}
···
 	min_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved);
 	if (udata->outlen && udata->outlen < min_resp_len)
 		return ERR_PTR(-EINVAL);
+
+	if (!capable(CAP_SYS_RAWIO) &&
+	    init_attr->create_flags & IB_WQ_FLAGS_DELAY_DROP)
+		return ERR_PTR(-EPERM);
 
 	dev = to_mdev(pd->device);
 	switch (init_attr->wq_type) {

drivers/infiniband/sw/rdmavt/cq.c

···
 	if (cq->ip)
 		kref_put(&cq->ip->ref, rvt_release_mmap_info);
 	else
-		vfree(cq->queue);
+		vfree(cq->kqueue);
 }
 
 /**