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

…t/dledford/rdma

Pull rdma fixes from Doug Ledford:
 "This has been a slow -rc cycle for the RDMA subsystem. We really
  haven't had a lot of rc fixes come in. This pull request is the first
  of this entire rc cycle and it has all of the suitable fixes so far
  and it's still only about 20 patches. The fix for the minor breakage
  cause by the dma mapping patchset is in here, as well as a couple
  other potential oops fixes, but the rest is more minor.

  Summary:

   - fix for dma_ops change in this kernel, resolving the s390, powerpc,
     and IOMMU operation

   - a few other oops fixes

   - the rest are all minor fixes"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma:
  IB/qib: fix false-postive maybe-uninitialized warning
  RDMA/iser: Fix possible mr leak on device removal event
  IB/device: Convert ib-comp-wq to be CPU-bound
  IB/cq: Don't process more than the given budget
  IB/rxe: increment msn only when completing a request
  uapi: fix rdma/mlx5-abi.h userspace compilation errors
  IB/core: Restore I/O MMU, s390 and powerpc support
  IB/rxe: Update documentation link
  RDMA/ocrdma: fix a type issue in ocrdma_put_pd_num()
  IB/rxe: double free on error
  RDMA/vmw_pvrdma: Activate device on ethernet link up
  RDMA/vmw_pvrdma: Dont hardcode QP header page
  RDMA/vmw_pvrdma: Cleanup unused variables
  infiniband: Fix alignment of mmap cookies to support VIPT caching
  IB/core: Protect against self-requeue of a cq work item
  i40iw: Receive netdev events post INET_NOTIFIER state

drivers/infiniband/core/cq.c

@@ -29,7 +29,13 @@ static int __ib_process_cq(struct ib_cq *cq, int budget)
 {
 	int i, n, completed = 0;
 
-	while ((n = ib_poll_cq(cq, IB_POLL_BATCH, cq->wc)) > 0) {
+	/*
+	 * budget might be (-1) if the caller does not
+	 * want to bound this call, thus we need unsigned
+	 * minimum here.
+	 */
+	while ((n = ib_poll_cq(cq, min_t(u32, IB_POLL_BATCH,
+			budget - completed), cq->wc)) > 0) {
 		for (i = 0; i < n; i++) {
 			struct ib_wc *wc = &cq->wc[i];
 
@@ -196,7 +202,7 @@ void ib_free_cq(struct ib_cq *cq)
 		irq_poll_disable(&cq->iop);
 		break;
 	case IB_POLL_WORKQUEUE:
-		flush_work(&cq->work);
+		cancel_work_sync(&cq->work);
 		break;
 	default:
 		WARN_ON_ONCE(1);

drivers/infiniband/core/device.c

@@ -336,12 +336,26 @@ int ib_register_device(struct ib_device *device,
 	struct device *parent = device->dev.parent;
 
 	WARN_ON_ONCE(!parent);
-	if (!device->dev.dma_ops)
-		device->dev.dma_ops = parent->dma_ops;
-	if (!device->dev.dma_mask)
-		device->dev.dma_mask = parent->dma_mask;
-	if (!device->dev.coherent_dma_mask)
-		device->dev.coherent_dma_mask = parent->coherent_dma_mask;
+	WARN_ON_ONCE(device->dma_device);
+	if (device->dev.dma_ops) {
+		/*
+		 * The caller provided custom DMA operations. Copy the
+		 * DMA-related fields that are used by e.g. dma_alloc_coherent()
+		 * into device->dev.
+		 */
+		device->dma_device = &device->dev;
+		if (!device->dev.dma_mask)
+			device->dev.dma_mask = parent->dma_mask;
+		if (!device->dev.coherent_dma_mask)
+			device->dev.coherent_dma_mask =
+				parent->coherent_dma_mask;
+	} else {
+		/*
+		 * The caller did not provide custom DMA operations. Use the
+		 * DMA mapping operations of the parent device.
+		 */
+		device->dma_device = parent;
+	}
 
 	mutex_lock(&device_mutex);
 
@@ -1015,8 +1029,7 @@ static int __init ib_core_init(void)
 		return -ENOMEM;
 
 	ib_comp_wq = alloc_workqueue("ib-comp-wq",
-			WQ_UNBOUND | WQ_HIGHPRI | WQ_MEM_RECLAIM,
-			WQ_UNBOUND_MAX_ACTIVE);
+			WQ_HIGHPRI | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
 	if (!ib_comp_wq) {
 		ret = -ENOMEM;
 		goto err;

drivers/infiniband/hw/i40iw/i40iw_utils.c

@@ -160,6 +160,9 @@ int i40iw_inetaddr_event(struct notifier_block *notifier,
 		return NOTIFY_DONE;
 
 	iwdev = &hdl->device;
+	if (iwdev->init_state < INET_NOTIFIER)
+		return NOTIFY_DONE;
+
 	netdev = iwdev->ldev->netdev;
 	upper_dev = netdev_master_upper_dev_get(netdev);
 	if (netdev != event_netdev)
@@ -214,6 +217,9 @@ int i40iw_inet6addr_event(struct notifier_block *notifier,
 		return NOTIFY_DONE;
 
 	iwdev = &hdl->device;
+	if (iwdev->init_state < INET_NOTIFIER)
+		return NOTIFY_DONE;
+
 	netdev = iwdev->ldev->netdev;
 	if (netdev != event_netdev)
 		return NOTIFY_DONE;
@@ -260,6 +266,8 @@ int i40iw_net_event(struct notifier_block *notifier, unsigned long event, void *
 		if (!iwhdl)
 			return NOTIFY_DONE;
 		iwdev = &iwhdl->device;
+		if (iwdev->init_state < INET_NOTIFIER)
+			return NOTIFY_DONE;
 		p = (__be32 *)neigh->primary_key;
 		i40iw_copy_ip_ntohl(local_ipaddr, p);
 		if (neigh->nud_state & NUD_VALID) {

drivers/infiniband/hw/ocrdma/ocrdma_verbs.c

@@ -372,7 +372,7 @@ static int _ocrdma_pd_mgr_put_bitmap(struct ocrdma_dev *dev, u16 pd_id,
 	return 0;
 }
 
-static u8 ocrdma_put_pd_num(struct ocrdma_dev *dev, u16 pd_id,
+static int ocrdma_put_pd_num(struct ocrdma_dev *dev, u16 pd_id,
 				   bool dpp_pool)
 {
 	int status;

drivers/infiniband/hw/qib/qib_iba7322.c

@@ -7068,7 +7068,7 @@ static void qib_7322_txchk_change(struct qib_devdata *dd, u32 start,
 	unsigned long flags;
 
 	while (wait) {
-		unsigned long shadow;
+		unsigned long shadow = 0;
 		int cstart, previ = -1;
 
 		/*

drivers/infiniband/hw/vmw_pvrdma/pvrdma.h

@@ -69,6 +69,9 @@
  */
 #define PCI_DEVICE_ID_VMWARE_PVRDMA	0x0820
 
+#define PVRDMA_NUM_RING_PAGES		4
+#define PVRDMA_QP_NUM_HEADER_PAGES	1
+
 struct pvrdma_dev;
 
 struct pvrdma_page_dir {

drivers/infiniband/hw/vmw_pvrdma/pvrdma_dev_api.h

@@ -132,7 +132,7 @@ enum pvrdma_pci_resource {
 
 enum pvrdma_device_ctl {
 	PVRDMA_DEVICE_CTL_ACTIVATE,	/* Activate device. */
-	PVRDMA_DEVICE_CTL_QUIESCE,	/* Quiesce device. */
+	PVRDMA_DEVICE_CTL_UNQUIESCE,	/* Unquiesce device. */
 	PVRDMA_DEVICE_CTL_RESET,	/* Reset device. */
 };
 

drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c

@@ -56,7 +56,7 @@
 #include "pvrdma.h"
 
 #define DRV_NAME	"vmw_pvrdma"
-#define DRV_VERSION	"1.0.0.0-k"
+#define DRV_VERSION	"1.0.1.0-k"
 
 static DEFINE_MUTEX(pvrdma_device_list_lock);
 static LIST_HEAD(pvrdma_device_list);
@@ -660,7 +660,16 @@ static void pvrdma_netdevice_event_handle(struct pvrdma_dev *dev,
 		pvrdma_dispatch_event(dev, 1, IB_EVENT_PORT_ERR);
 		break;
 	case NETDEV_UP:
-		pvrdma_dispatch_event(dev, 1, IB_EVENT_PORT_ACTIVE);
+		pvrdma_write_reg(dev, PVRDMA_REG_CTL,
+				 PVRDMA_DEVICE_CTL_UNQUIESCE);
+
+		mb();
+
+		if (pvrdma_read_reg(dev, PVRDMA_REG_ERR))
+			dev_err(&dev->pdev->dev,
+				"failed to activate device during link up\n");
+		else
+			pvrdma_dispatch_event(dev, 1, IB_EVENT_PORT_ACTIVE);
 		break;
 	default:
 		dev_dbg(&dev->pdev->dev, "ignore netdevice event %ld on %s\n",
@@ -858,7 +867,7 @@ static int pvrdma_pci_probe(struct pci_dev *pdev,
 	dev->dsr->resp_slot_dma = (u64)slot_dma;
 
 	/* Async event ring */
-	dev->dsr->async_ring_pages.num_pages = 4;
+	dev->dsr->async_ring_pages.num_pages = PVRDMA_NUM_RING_PAGES;
 	ret = pvrdma_page_dir_init(dev, &dev->async_pdir,
 				   dev->dsr->async_ring_pages.num_pages, true);
 	if (ret)
@@ -867,7 +876,7 @@ static int pvrdma_pci_probe(struct pci_dev *pdev,
 	dev->dsr->async_ring_pages.pdir_dma = dev->async_pdir.dir_dma;
 
 	/* CQ notification ring */
-	dev->dsr->cq_ring_pages.num_pages = 4;
+	dev->dsr->cq_ring_pages.num_pages = PVRDMA_NUM_RING_PAGES;
 	ret = pvrdma_page_dir_init(dev, &dev->cq_pdir,
 				   dev->dsr->cq_ring_pages.num_pages, true);
 	if (ret)

drivers/infiniband/hw/vmw_pvrdma/pvrdma_qp.c

@@ -170,8 +170,9 @@ static int pvrdma_set_sq_size(struct pvrdma_dev *dev, struct ib_qp_cap *req_cap,
 					     sizeof(struct pvrdma_sge) *
 					     qp->sq.max_sg);
 	/* Note: one extra page for the header. */
-	qp->npages_send = 1 + (qp->sq.wqe_cnt * qp->sq.wqe_size +
-			       PAGE_SIZE - 1) / PAGE_SIZE;
+	qp->npages_send = PVRDMA_QP_NUM_HEADER_PAGES +
+			  (qp->sq.wqe_cnt * qp->sq.wqe_size + PAGE_SIZE - 1) /
+								PAGE_SIZE;
 
 	return 0;
 }
@@ -288,7 +289,7 @@ struct ib_qp *pvrdma_create_qp(struct ib_pd *pd,
 			qp->npages = qp->npages_send + qp->npages_recv;
 
 			/* Skip header page. */
-			qp->sq.offset = PAGE_SIZE;
+			qp->sq.offset = PVRDMA_QP_NUM_HEADER_PAGES * PAGE_SIZE;
 
 			/* Recv queue pages are after send pages. */
 			qp->rq.offset = qp->npages_send * PAGE_SIZE;
@@ -341,7 +342,7 @@ struct ib_qp *pvrdma_create_qp(struct ib_pd *pd,
 	cmd->qp_type = ib_qp_type_to_pvrdma(init_attr->qp_type);
 	cmd->access_flags = IB_ACCESS_LOCAL_WRITE;
 	cmd->total_chunks = qp->npages;
-	cmd->send_chunks = qp->npages_send - 1;
+	cmd->send_chunks = qp->npages_send - PVRDMA_QP_NUM_HEADER_PAGES;
 	cmd->pdir_dma = qp->pdir.dir_dma;
 
 	dev_dbg(&dev->pdev->dev, "create queuepair with %d, %d, %d, %d\n",
@@ -554,13 +555,13 @@ int pvrdma_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 	return ret;
 }
 
-static inline void *get_sq_wqe(struct pvrdma_qp *qp, int n)
+static inline void *get_sq_wqe(struct pvrdma_qp *qp, unsigned int n)
 {
 	return pvrdma_page_dir_get_ptr(&qp->pdir,
 				       qp->sq.offset + n * qp->sq.wqe_size);
 }
 
-static inline void *get_rq_wqe(struct pvrdma_qp *qp, int n)
+static inline void *get_rq_wqe(struct pvrdma_qp *qp, unsigned int n)
 {
 	return pvrdma_page_dir_get_ptr(&qp->pdir,
 				       qp->rq.offset + n * qp->rq.wqe_size);
@@ -598,9 +599,7 @@ int pvrdma_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 	unsigned long flags;
 	struct pvrdma_sq_wqe_hdr *wqe_hdr;
 	struct pvrdma_sge *sge;
-	int i, index;
-	int nreq;
-	int ret;
+	int i, ret;
 
 	/*
 	 * In states lower than RTS, we can fail immediately. In other states,
@@ -613,9 +612,8 @@ int pvrdma_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 
 	spin_lock_irqsave(&qp->sq.lock, flags);
 
-	index = pvrdma_idx(&qp->sq.ring->prod_tail, qp->sq.wqe_cnt);
-	for (nreq = 0; wr; nreq++, wr = wr->next) {
-		unsigned int tail;
+	while (wr) {
+		unsigned int tail = 0;
 
 		if (unlikely(!pvrdma_idx_ring_has_space(
 				qp->sq.ring, qp->sq.wqe_cnt, &tail))) {
@@ -680,7 +678,7 @@ int pvrdma_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			}
 		}
 
-		wqe_hdr = (struct pvrdma_sq_wqe_hdr *)get_sq_wqe(qp, index);
+		wqe_hdr = (struct pvrdma_sq_wqe_hdr *)get_sq_wqe(qp, tail);
 		memset(wqe_hdr, 0, sizeof(*wqe_hdr));
 		wqe_hdr->wr_id = wr->wr_id;
 		wqe_hdr->num_sge = wr->num_sge;
@@ -771,12 +769,11 @@ int pvrdma_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 		/* Make sure wqe is written before index update */
 		smp_wmb();
 
-		index++;
-		if (unlikely(index >= qp->sq.wqe_cnt))
-			index = 0;
 		/* Update shared sq ring */
 		pvrdma_idx_ring_inc(&qp->sq.ring->prod_tail,
 				    qp->sq.wqe_cnt);
+
+		wr = wr->next;
 	}
 
 	ret = 0;
@@ -806,7 +803,6 @@ int pvrdma_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 	struct pvrdma_qp *qp = to_vqp(ibqp);
 	struct pvrdma_rq_wqe_hdr *wqe_hdr;
 	struct pvrdma_sge *sge;
-	int index, nreq;
 	int ret = 0;
 	int i;
 
@@ -821,9 +817,8 @@ int pvrdma_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 
 	spin_lock_irqsave(&qp->rq.lock, flags);
 
-	index = pvrdma_idx(&qp->rq.ring->prod_tail, qp->rq.wqe_cnt);
-	for (nreq = 0; wr; nreq++, wr = wr->next) {
-		unsigned int tail;
+	while (wr) {
+		unsigned int tail = 0;
 
 		if (unlikely(wr->num_sge > qp->rq.max_sg ||
 			     wr->num_sge < 0)) {
@@ -843,7 +838,7 @@ int pvrdma_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 			goto out;
 		}
 
-		wqe_hdr = (struct pvrdma_rq_wqe_hdr *)get_rq_wqe(qp, index);
+		wqe_hdr = (struct pvrdma_rq_wqe_hdr *)get_rq_wqe(qp, tail);
 		wqe_hdr->wr_id = wr->wr_id;
 		wqe_hdr->num_sge = wr->num_sge;
 		wqe_hdr->total_len = 0;
@@ -859,12 +854,11 @@ int pvrdma_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 		/* Make sure wqe is written before index update */
 		smp_wmb();
 
-		index++;
-		if (unlikely(index >= qp->rq.wqe_cnt))
-			index = 0;
 		/* Update shared rq ring */
 		pvrdma_idx_ring_inc(&qp->rq.ring->prod_tail,
 				    qp->rq.wqe_cnt);
+
+		wr = wr->next;
 	}
 
 	spin_unlock_irqrestore(&qp->rq.lock, flags);

drivers/infiniband/sw/rdmavt/mmap.c

@@ -170,9 +170,9 @@ struct rvt_mmap_info *rvt_create_mmap_info(struct rvt_dev_info *rdi,
 
 	spin_lock_irq(&rdi->mmap_offset_lock);
 	if (rdi->mmap_offset == 0)
-		rdi->mmap_offset = PAGE_SIZE;
+		rdi->mmap_offset = ALIGN(PAGE_SIZE, SHMLBA);
 	ip->offset = rdi->mmap_offset;
-	rdi->mmap_offset += size;
+	rdi->mmap_offset += ALIGN(size, SHMLBA);
 	spin_unlock_irq(&rdi->mmap_offset_lock);
 
 	INIT_LIST_HEAD(&ip->pending_mmaps);

drivers/infiniband/sw/rxe/Kconfig

@@ -22,4 +22,4 @@ config RDMA_RXE
 	To configure and work with soft-RoCE driver please use the
 	following wiki page under "configure Soft-RoCE (RXE)" section:
 
-	https://github.com/SoftRoCE/rxe-dev/wiki/rxe-dev:-Home
+	https://github.com/linux-rdma/rdma-core/blob/master/Documentation/rxe.md

drivers/infiniband/sw/rxe/rxe_mmap.c

@@ -156,10 +156,10 @@ struct rxe_mmap_info *rxe_create_mmap_info(struct rxe_dev *rxe,
 	spin_lock_bh(&rxe->mmap_offset_lock);
 
 	if (rxe->mmap_offset == 0)
-		rxe->mmap_offset = PAGE_SIZE;
+		rxe->mmap_offset = ALIGN(PAGE_SIZE, SHMLBA);
 
 	ip->info.offset = rxe->mmap_offset;
-	rxe->mmap_offset += size;
+	rxe->mmap_offset += ALIGN(size, SHMLBA);
 
 	spin_unlock_bh(&rxe->mmap_offset_lock);
 

drivers/infiniband/sw/rxe/rxe_req.c

@@ -729,11 +729,11 @@ int rxe_requester(void *arg)
 	ret = rxe_xmit_packet(to_rdev(qp->ibqp.device), qp, &pkt, skb);
 	if (ret) {
 		qp->need_req_skb = 1;
-		kfree_skb(skb);
 
 		rollback_state(wqe, qp, &rollback_wqe, rollback_psn);
 
 		if (ret == -EAGAIN) {
+			kfree_skb(skb);
 			rxe_run_task(&qp->req.task, 1);
 			goto exit;
 		}

drivers/infiniband/sw/rxe/rxe_resp.c

@@ -813,18 +813,17 @@ static enum resp_states execute(struct rxe_qp *qp, struct rxe_pkt_info *pkt)
 		WARN_ON_ONCE(1);
 	}
 
-	/* We successfully processed this new request. */
-	qp->resp.msn++;
-
 	/* next expected psn, read handles this separately */
 	qp->resp.psn = (pkt->psn + 1) & BTH_PSN_MASK;
 
 	qp->resp.opcode = pkt->opcode;
 	qp->resp.status = IB_WC_SUCCESS;
 
-	if (pkt->mask & RXE_COMP_MASK)
+	if (pkt->mask & RXE_COMP_MASK) {
+		/* We successfully processed this new request. */
+		qp->resp.msn++;
 		return RESPST_COMPLETE;
-	else if (qp_type(qp) == IB_QPT_RC)
+	} else if (qp_type(qp) == IB_QPT_RC)
 		return RESPST_ACKNOWLEDGE;
 	else
 		return RESPST_CLEANUP;

drivers/infiniband/ulp/iser/iscsi_iser.h

@@ -430,6 +430,7 @@ struct iser_fr_desc {
 	struct list_head		  list;
 	struct iser_reg_resources	  rsc;
 	struct iser_pi_context		 *pi_ctx;
+	struct list_head                  all_list;
 };
 
 /**
@@ -443,6 +444,7 @@ struct iser_fr_pool {
 	struct list_head        list;
 	spinlock_t              lock;
 	int                     size;
+	struct list_head        all_list;
 };
 
 /**