Second RDMA 5.7 rc pull request

A few minor bug fixes for user visible defects, and one regression:
 
 - Various bugs from static checkers and syzkaller
 
 - Add missing error checking in mlx4
 
 - Prevent RTNL lock recursion in i40iw
 
 - Fix segfault in cxgb4 in peer abort cases
 
 - Fix a regression added in 5.7 where the IB_EVENT_DEVICE_FATAL could be
   lost, and wasn't delivered to all the FDs
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEEfB7FMLh+8QxL+6i3OG33FX4gmxoFAl6+5V8ACgkQOG33FX4g
 mxo/Zw//QlrhDSM4DG4hZsFhFX3i/vT7xbxkNTp/U1+Vue9cKcUlotNOzQ79lQur
 /12QEiDffLZNsy2jCD1nxf/rEo9NDc8iSHPcryQ3aGa2K3XJ3QQ2ZhHBzfFU9GVP
 GQYlGt9uB4t6LEcU4P/ZCdf1nmkfYFvNDfmveVWrPLasABK7WqxCBXaqbLed+0ca
 lGhFBGLdGNJqyK2BkPCtXr9XYTjzZhW0lJqMuex0YD7cIAFfn+qbzvLQheBy/mhH
 o9n28GQbIgpNXvYz2HvUkTfiwDLylFaNVBYVctnm10cbNtLHv2J0bBQQb2ZaXUAM
 xt54AQ2QAMHKC2h6sUqyDAmWKfPEOnxZ9LycYDa0ZaIvm/uK/jiEOXE9XbQeiOKF
 eHiyuDg0EB4AVcYlNSm7roHdbh3rAluHerGe4Kv3efF/b1Zt2mtTW7q/XuROfSeT
 WBnALyl7RurnSG0HY9hyUMak65JwolgqPdBDpzRjqIF5jeKW0nn8MsbGTPGWlcyN
 zLu5IQn/vv+sfISav1cVMRoijboe85+3SOVtMBQOJ2m7EkdzRyRSQh3oG1556n2y
 oqseNSQFKNFA1p4hEeTf8BVEqVX3gj50ykj+2g1HyxH3sB+iE3+EctnHNhfd3IKI
 n6e/67Nm0ub/WM1OXeuXBmW0NvsTt0aPBfsp4s+oq3eICD4YbFE=
 =R5BJ
 -----END PGP SIGNATURE-----

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

Pull rdma fixes from Jason Gunthorpe:
 "A few minor bug fixes for user visible defects, and one regression:

   - Various bugs from static checkers and syzkaller

   - Add missing error checking in mlx4

   - Prevent RTNL lock recursion in i40iw

   - Fix segfault in cxgb4 in peer abort cases

   - Fix a regression added in 5.7 where the IB_EVENT_DEVICE_FATAL could
     be lost, and wasn't delivered to all the FDs"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/uverbs: Move IB_EVENT_DEVICE_FATAL to destroy_uobj
  RDMA/uverbs: Do not discard the IB_EVENT_DEVICE_FATAL event
  RDMA/iw_cxgb4: Fix incorrect function parameters
  RDMA/core: Fix double put of resource
  IB/core: Fix potential NULL pointer dereference in pkey cache
  IB/hfi1: Fix another case where pq is left on waitlist
  IB/i40iw: Remove bogus call to netdev_master_upper_dev_get()
  IB/mlx4: Test return value of calls to ib_get_cached_pkey
  RDMA/rxe: Always return ERR_PTR from rxe_create_mmap_info()
  i40iw: Fix error handling in i40iw_manage_arp_cache()

drivers/infiniband/core/cache.c

@@ -1553,8 +1553,11 @@ int ib_cache_setup_one(struct ib_device *device)
 	if (err)
 		return err;
 
-	rdma_for_each_port (device, p)
-		ib_cache_update(device, p, true);
+	rdma_for_each_port (device, p) {
+		err = ib_cache_update(device, p, true);
+		if (err)
+			return err;
+	}
 
 	return 0;
 }

drivers/infiniband/core/nldev.c

@@ -1292,11 +1292,10 @@ static int res_get_common_doit(struct sk_buff *skb, struct nlmsghdr *nlh,
 	has_cap_net_admin = netlink_capable(skb, CAP_NET_ADMIN);
 
 	ret = fill_func(msg, has_cap_net_admin, res, port);
-
-	rdma_restrack_put(res);
 	if (ret)
 		goto err_free;
 
+	rdma_restrack_put(res);
 	nlmsg_end(msg, nlh);
 	ib_device_put(device);
 	return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid);

drivers/infiniband/core/rdma_core.c

@@ -459,7 +459,8 @@ alloc_begin_fd_uobject(const struct uverbs_api_object *obj,
 	struct ib_uobject *uobj;
 	struct file *filp;
 
-	if (WARN_ON(fd_type->fops->release != &uverbs_uobject_fd_release))
+	if (WARN_ON(fd_type->fops->release != &uverbs_uobject_fd_release &&
+		    fd_type->fops->release != &uverbs_async_event_release))
 		return ERR_PTR(-EINVAL);
 
 	new_fd = get_unused_fd_flags(O_CLOEXEC);

drivers/infiniband/core/uverbs.h

@@ -219,6 +219,7 @@ void ib_uverbs_init_event_queue(struct ib_uverbs_event_queue *ev_queue);
 void ib_uverbs_init_async_event_file(struct ib_uverbs_async_event_file *ev_file);
 void ib_uverbs_free_event_queue(struct ib_uverbs_event_queue *event_queue);
 void ib_uverbs_flow_resources_free(struct ib_uflow_resources *uflow_res);
+int uverbs_async_event_release(struct inode *inode, struct file *filp);
 
 int ib_alloc_ucontext(struct uverbs_attr_bundle *attrs);
 int ib_init_ucontext(struct uverbs_attr_bundle *attrs);
@@ -227,6 +228,9 @@ void ib_uverbs_release_ucq(struct ib_uverbs_completion_event_file *ev_file,
 			   struct ib_ucq_object *uobj);
 void ib_uverbs_release_uevent(struct ib_uevent_object *uobj);
 void ib_uverbs_release_file(struct kref *ref);
+void ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
+			     __u64 element, __u64 event,
+			     struct list_head *obj_list, u32 *counter);
 
 void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context);
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr);

drivers/infiniband/core/uverbs_main.c

@@ -346,7 +346,7 @@ const struct file_operations uverbs_async_event_fops = {
 	.owner	 = THIS_MODULE,
 	.read	 = ib_uverbs_async_event_read,
 	.poll    = ib_uverbs_async_event_poll,
-	.release = uverbs_uobject_fd_release,
+	.release = uverbs_async_event_release,
 	.fasync  = ib_uverbs_async_event_fasync,
 	.llseek	 = no_llseek,
 };
@@ -386,10 +386,9 @@ void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context)
 	kill_fasync(&ev_queue->async_queue, SIGIO, POLL_IN);
 }
 
-static void
-ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
-			__u64 element, __u64 event, struct list_head *obj_list,
-			u32 *counter)
+void ib_uverbs_async_handler(struct ib_uverbs_async_event_file *async_file,
+			     __u64 element, __u64 event,
+			     struct list_head *obj_list, u32 *counter)
 {
 	struct ib_uverbs_event *entry;
 	unsigned long flags;
@@ -1187,9 +1186,6 @@ static void ib_uverbs_free_hw_resources(struct ib_uverbs_device *uverbs_dev,
 		 */
 		mutex_unlock(&uverbs_dev->lists_mutex);
 
-		ib_uverbs_async_handler(READ_ONCE(file->async_file), 0,
-					IB_EVENT_DEVICE_FATAL, NULL, NULL);
-
 		uverbs_destroy_ufile_hw(file, RDMA_REMOVE_DRIVER_REMOVE);
 		kref_put(&file->ref, ib_uverbs_release_file);
 

drivers/infiniband/core/uverbs_std_types_async_fd.c

@@ -26,10 +26,38 @@ static int uverbs_async_event_destroy_uobj(struct ib_uobject *uobj,
 		container_of(uobj, struct ib_uverbs_async_event_file, uobj);
 
 	ib_unregister_event_handler(&event_file->event_handler);
-	ib_uverbs_free_event_queue(&event_file->ev_queue);
+
+	if (why == RDMA_REMOVE_DRIVER_REMOVE)
+		ib_uverbs_async_handler(event_file, 0, IB_EVENT_DEVICE_FATAL,
+					NULL, NULL);
 	return 0;
 }
 
+int uverbs_async_event_release(struct inode *inode, struct file *filp)
+{
+	struct ib_uverbs_async_event_file *event_file;
+	struct ib_uobject *uobj = filp->private_data;
+	int ret;
+
+	if (!uobj)
+		return uverbs_uobject_fd_release(inode, filp);
+
+	event_file =
+		container_of(uobj, struct ib_uverbs_async_event_file, uobj);
+
+	/*
+	 * The async event FD has to deliver IB_EVENT_DEVICE_FATAL even after
+	 * disassociation, so cleaning the event list must only happen after
+	 * release. The user knows it has reached the end of the event stream
+	 * when it sees IB_EVENT_DEVICE_FATAL.
+	 */
+	uverbs_uobject_get(uobj);
+	ret = uverbs_uobject_fd_release(inode, filp);
+	ib_uverbs_free_event_queue(&event_file->ev_queue);
+	uverbs_uobject_put(uobj);
+	return ret;
+}
+
 DECLARE_UVERBS_NAMED_METHOD(
 	UVERBS_METHOD_ASYNC_EVENT_ALLOC,
 	UVERBS_ATTR_FD(UVERBS_ATTR_ASYNC_EVENT_ALLOC_FD_HANDLE,

drivers/infiniband/hw/cxgb4/cm.c

@@ -2891,8 +2891,7 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb)
 			srqidx = ABORT_RSS_SRQIDX_G(
 					be32_to_cpu(req->srqidx_status));
 			if (srqidx) {
-				complete_cached_srq_buffers(ep,
-							    req->srqidx_status);
+				complete_cached_srq_buffers(ep, srqidx);
 			} else {
 				/* Hold ep ref until finish_peer_abort() */
 				c4iw_get_ep(&ep->com);
@@ -3878,8 +3877,8 @@ static int read_tcb_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
 		return 0;
 	}
 
-	ep->srqe_idx = t4_tcb_get_field32(tcb, TCB_RQ_START_W, TCB_RQ_START_W,
-			TCB_RQ_START_S);
+	ep->srqe_idx = t4_tcb_get_field32(tcb, TCB_RQ_START_W, TCB_RQ_START_M,
+					  TCB_RQ_START_S);
 cleanup:
 	pr_debug("ep %p tid %u %016x\n", ep, ep->hwtid, ep->srqe_idx);
 

drivers/infiniband/hw/hfi1/user_sdma.c

@@ -589,10 +589,6 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
 
 	set_comp_state(pq, cq, info.comp_idx, QUEUED, 0);
 	pq->state = SDMA_PKT_Q_ACTIVE;
-	/* Send the first N packets in the request to buy us some time */
-	ret = user_sdma_send_pkts(req, pcount);
-	if (unlikely(ret < 0 && ret != -EBUSY))
-		goto free_req;
 
 	/*
 	 * This is a somewhat blocking send implementation.

drivers/infiniband/hw/i40iw/i40iw_cm.c

@@ -1987,7 +1987,6 @@ static int i40iw_addr_resolve_neigh(struct i40iw_device *iwdev,
 	struct rtable *rt;
 	struct neighbour *neigh;
 	int rc = arpindex;
-	struct net_device *netdev = iwdev->netdev;
 	__be32 dst_ipaddr = htonl(dst_ip);
 	__be32 src_ipaddr = htonl(src_ip);
 
@@ -1997,9 +1996,6 @@ static int i40iw_addr_resolve_neigh(struct i40iw_device *iwdev,
 		return rc;
 	}
 
-	if (netif_is_bond_slave(netdev))
-		netdev = netdev_master_upper_dev_get(netdev);
-
 	neigh = dst_neigh_lookup(&rt->dst, &dst_ipaddr);
 
 	rcu_read_lock();
@@ -2065,7 +2061,6 @@ static int i40iw_addr_resolve_neigh_ipv6(struct i40iw_device *iwdev,
 {
 	struct neighbour *neigh;
 	int rc = arpindex;
-	struct net_device *netdev = iwdev->netdev;
 	struct dst_entry *dst;
 	struct sockaddr_in6 dst_addr;
 	struct sockaddr_in6 src_addr;
@@ -2086,9 +2081,6 @@ static int i40iw_addr_resolve_neigh_ipv6(struct i40iw_device *iwdev,
 		return rc;
 	}
 
-	if (netif_is_bond_slave(netdev))
-		netdev = netdev_master_upper_dev_get(netdev);
-
 	neigh = dst_neigh_lookup(dst, dst_addr.sin6_addr.in6_u.u6_addr32);
 
 	rcu_read_lock();

drivers/infiniband/hw/i40iw/i40iw_hw.c

@@ -534,7 +534,7 @@ void i40iw_manage_arp_cache(struct i40iw_device *iwdev,
 	int arp_index;
 
 	arp_index = i40iw_arp_table(iwdev, ip_addr, ipv4, mac_addr, action);
-	if (arp_index == -1)
+	if (arp_index < 0)
 		return;
 	cqp_request = i40iw_get_cqp_request(&iwdev->cqp, false);
 	if (!cqp_request)

drivers/infiniband/hw/mlx4/qp.c

@@ -2891,6 +2891,7 @@ static int build_sriov_qp0_header(struct mlx4_ib_sqp *sqp,
 	int send_size;
 	int header_size;
 	int spc;
+	int err;
 	int i;
 
 	if (wr->wr.opcode != IB_WR_SEND)
@@ -2925,7 +2926,9 @@ static int build_sriov_qp0_header(struct mlx4_ib_sqp *sqp,
 
 	sqp->ud_header.lrh.virtual_lane    = 0;
 	sqp->ud_header.bth.solicited_event = !!(wr->wr.send_flags & IB_SEND_SOLICITED);
-	ib_get_cached_pkey(ib_dev, sqp->qp.port, 0, &pkey);
+	err = ib_get_cached_pkey(ib_dev, sqp->qp.port, 0, &pkey);
+	if (err)
+		return err;
 	sqp->ud_header.bth.pkey = cpu_to_be16(pkey);
 	if (sqp->qp.mlx4_ib_qp_type == MLX4_IB_QPT_TUN_SMI_OWNER)
 		sqp->ud_header.bth.destination_qpn = cpu_to_be32(wr->remote_qpn);
@@ -3212,9 +3215,14 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, const struct ib_ud_wr *wr,
 	}
 	sqp->ud_header.bth.solicited_event = !!(wr->wr.send_flags & IB_SEND_SOLICITED);
 	if (!sqp->qp.ibqp.qp_num)
-		ib_get_cached_pkey(ib_dev, sqp->qp.port, sqp->pkey_index, &pkey);
+		err = ib_get_cached_pkey(ib_dev, sqp->qp.port, sqp->pkey_index,
+					 &pkey);
 	else
-		ib_get_cached_pkey(ib_dev, sqp->qp.port, wr->pkey_index, &pkey);
+		err = ib_get_cached_pkey(ib_dev, sqp->qp.port, wr->pkey_index,
+					 &pkey);
+	if (err)
+		return err;
+
 	sqp->ud_header.bth.pkey = cpu_to_be16(pkey);
 	sqp->ud_header.bth.destination_qpn = cpu_to_be32(wr->remote_qpn);
 	sqp->ud_header.bth.psn = cpu_to_be32((sqp->send_psn++) & ((1 << 24) - 1));

drivers/infiniband/sw/rxe/rxe_mmap.c

@@ -151,7 +151,7 @@ struct rxe_mmap_info *rxe_create_mmap_info(struct rxe_dev *rxe, u32 size,
 
 	ip = kmalloc(sizeof(*ip), GFP_KERNEL);
 	if (!ip)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
 	size = PAGE_ALIGN(size);
 

drivers/infiniband/sw/rxe/rxe_queue.c

@@ -45,12 +45,15 @@ int do_mmap_info(struct rxe_dev *rxe, struct mminfo __user *outbuf,
 
 	if (outbuf) {
 		ip = rxe_create_mmap_info(rxe, buf_size, udata, buf);
-		if (!ip)
+		if (IS_ERR(ip)) {
+			err = PTR_ERR(ip);
 			goto err1;
+		}
 
-		err = copy_to_user(outbuf, &ip->info, sizeof(ip->info));
-		if (err)
+		if (copy_to_user(outbuf, &ip->info, sizeof(ip->info))) {
+			err = -EFAULT;
 			goto err2;
+		}
 
 		spin_lock_bh(&rxe->pending_lock);
 		list_add(&ip->pending_mmaps, &rxe->pending_mmaps);
@@ -64,7 +67,7 @@ int do_mmap_info(struct rxe_dev *rxe, struct mminfo __user *outbuf,
 err2:
 	kfree(ip);
 err1:
-	return -EINVAL;
+	return err;
 }
 
 inline void rxe_queue_reset(struct rxe_queue *q)