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

Pull InfiniBand/RDMA fixes from Roland Dreier:
 - A couple more IPoIB fixes for regressions introduced by path database
   conversion
 - Minor other fixes to low-level drivers (cxgb4, mlx4, qib, ocrdma)

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/qib: Fix failure of compliance test C14-024#06_LocalPortNum
  RDMA/ocrdma: Fix CQE expansion of unsignaled WQE
  mlx4_core: Fix integer overflows so 8TBs of memory registration works
  IPoIB: Fix AB-BA deadlock when deleting neighbours
  IPoIB: Fix memory leak in the neigh table deletion flow
  RDMA/cxgb4: Move dereference below NULL test

drivers/infiniband/hw/cxgb4/cm.c

@@ -1361,11 +1361,11 @@ static int abort_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
 	struct tid_info *t = dev->rdev.lldi.tids;
 
 	ep = lookup_tid(t, tid);
-	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
 	if (!ep) {
 		printk(KERN_WARNING MOD "Abort rpl to freed endpoint\n");
 		return 0;
 	}
+	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
 	mutex_lock(&ep->com.mutex);
 	switch (ep->com.state) {
 	case ABORTING:

drivers/infiniband/hw/ocrdma/ocrdma_verbs.c

@@ -2219,7 +2219,6 @@ static bool ocrdma_poll_success_scqe(struct ocrdma_qp *qp,
 	u32 wqe_idx;
 
 	if (!qp->wqe_wr_id_tbl[tail].signaled) {
-		expand = true;	/* CQE cannot be consumed yet */
 		*polled = false;    /* WC cannot be consumed yet */
 	} else {
 		ibwc->status = IB_WC_SUCCESS;
@@ -2227,10 +2226,11 @@ static bool ocrdma_poll_success_scqe(struct ocrdma_qp *qp,
 		ibwc->qp = &qp->ibqp;
 		ocrdma_update_wc(qp, ibwc, tail);
 		*polled = true;
-		wqe_idx = le32_to_cpu(cqe->wq.wqeidx) &	OCRDMA_CQE_WQEIDX_MASK;
-		if (tail != wqe_idx)
-			expand = true; /* Coalesced CQE can't be consumed yet */
 	}
+	wqe_idx = le32_to_cpu(cqe->wq.wqeidx) &	OCRDMA_CQE_WQEIDX_MASK;
+	if (tail != wqe_idx)
+		expand = true; /* Coalesced CQE can't be consumed yet */
+
 	ocrdma_hwq_inc_tail(&qp->sq);
 	return expand;
 }

drivers/infiniband/hw/qib/qib_mad.c

@@ -471,9 +471,10 @@ static int subn_get_portinfo(struct ib_smp *smp, struct ib_device *ibdev,
 		if (port_num != port) {
 			ibp = to_iport(ibdev, port_num);
 			ret = check_mkey(ibp, smp, 0);
-			if (ret)
+			if (ret) {
 				ret = IB_MAD_RESULT_FAILURE;
 				goto bail;
+			}
 		}
 	}
 

drivers/infiniband/ulp/ipoib/ipoib.h

@@ -262,7 +262,10 @@ struct ipoib_ethtool_st {
 	u16     max_coalesced_frames;
 };
 
+struct ipoib_neigh_table;
+
 struct ipoib_neigh_hash {
+	struct ipoib_neigh_table       *ntbl;
 	struct ipoib_neigh __rcu      **buckets;
 	struct rcu_head			rcu;
 	u32				mask;
@@ -271,9 +274,9 @@ struct ipoib_neigh_hash {
 
 struct ipoib_neigh_table {
 	struct ipoib_neigh_hash __rcu  *htbl;
-	rwlock_t			rwlock;
 	atomic_t			entries;
 	struct completion		flushed;
+	struct completion		deleted;
 };
 
 /*

drivers/infiniband/ulp/ipoib/ipoib_main.c

@@ -546,15 +546,15 @@ static void neigh_add_path(struct sk_buff *skb, u8 *daddr,
 	struct ipoib_neigh *neigh;
 	unsigned long flags;
 
+	spin_lock_irqsave(&priv->lock, flags);
 	neigh = ipoib_neigh_alloc(daddr, dev);
 	if (!neigh) {
+		spin_unlock_irqrestore(&priv->lock, flags);
 		++dev->stats.tx_dropped;
 		dev_kfree_skb_any(skb);
 		return;
 	}
 
-	spin_lock_irqsave(&priv->lock, flags);
-
 	path = __path_find(dev, daddr + 4);
 	if (!path) {
 		path = path_rec_create(dev, daddr + 4);
@@ -863,10 +863,10 @@ static void __ipoib_reap_neigh(struct ipoib_dev_priv *priv)
 	if (test_bit(IPOIB_STOP_NEIGH_GC, &priv->flags))
 		return;
 
-	write_lock_bh(&ntbl->rwlock);
+	spin_lock_irqsave(&priv->lock, flags);
 
 	htbl = rcu_dereference_protected(ntbl->htbl,
-					 lockdep_is_held(&ntbl->rwlock));
+					 lockdep_is_held(&priv->lock));
 
 	if (!htbl)
 		goto out_unlock;
@@ -883,16 +883,14 @@ static void __ipoib_reap_neigh(struct ipoib_dev_priv *priv)
 		struct ipoib_neigh __rcu **np = &htbl->buckets[i];
 
 		while ((neigh = rcu_dereference_protected(*np,
-							  lockdep_is_held(&ntbl->rwlock))) != NULL) {
+							  lockdep_is_held(&priv->lock))) != NULL) {
 			/* was the neigh idle for two GC periods */
 			if (time_after(neigh_obsolete, neigh->alive)) {
 				rcu_assign_pointer(*np,
 						   rcu_dereference_protected(neigh->hnext,
-									     lockdep_is_held(&ntbl->rwlock)));
+									     lockdep_is_held(&priv->lock)));
 				/* remove from path/mc list */
-				spin_lock_irqsave(&priv->lock, flags);
 				list_del(&neigh->list);
-				spin_unlock_irqrestore(&priv->lock, flags);
 				call_rcu(&neigh->rcu, ipoib_neigh_reclaim);
 			} else {
 				np = &neigh->hnext;
@@ -902,7 +900,7 @@ static void __ipoib_reap_neigh(struct ipoib_dev_priv *priv)
 	}
 
 out_unlock:
-	write_unlock_bh(&ntbl->rwlock);
+	spin_unlock_irqrestore(&priv->lock, flags);
 }
 
 static void ipoib_reap_neigh(struct work_struct *work)
@@ -947,10 +945,8 @@ struct ipoib_neigh *ipoib_neigh_alloc(u8 *daddr,
 	struct ipoib_neigh *neigh;
 	u32 hash_val;
 
-	write_lock_bh(&ntbl->rwlock);
-
 	htbl = rcu_dereference_protected(ntbl->htbl,
-					 lockdep_is_held(&ntbl->rwlock));
+					 lockdep_is_held(&priv->lock));
 	if (!htbl) {
 		neigh = NULL;
 		goto out_unlock;
@@ -961,10 +957,10 @@ struct ipoib_neigh *ipoib_neigh_alloc(u8 *daddr,
 	 */
 	hash_val = ipoib_addr_hash(htbl, daddr);
 	for (neigh = rcu_dereference_protected(htbl->buckets[hash_val],
-					       lockdep_is_held(&ntbl->rwlock));
+					       lockdep_is_held(&priv->lock));
 	     neigh != NULL;
 	     neigh = rcu_dereference_protected(neigh->hnext,
-					       lockdep_is_held(&ntbl->rwlock))) {
+					       lockdep_is_held(&priv->lock))) {
 		if (memcmp(daddr, neigh->daddr, INFINIBAND_ALEN) == 0) {
 			/* found, take one ref on behalf of the caller */
 			if (!atomic_inc_not_zero(&neigh->refcnt)) {
@@ -987,12 +983,11 @@ struct ipoib_neigh *ipoib_neigh_alloc(u8 *daddr,
 	/* put in hash */
 	rcu_assign_pointer(neigh->hnext,
 			   rcu_dereference_protected(htbl->buckets[hash_val],
-						     lockdep_is_held(&ntbl->rwlock)));
+						     lockdep_is_held(&priv->lock)));
 	rcu_assign_pointer(htbl->buckets[hash_val], neigh);
 	atomic_inc(&ntbl->entries);
 
 out_unlock:
-	write_unlock_bh(&ntbl->rwlock);
 
 	return neigh;
 }
@@ -1040,35 +1035,29 @@ void ipoib_neigh_free(struct ipoib_neigh *neigh)
 	struct ipoib_neigh *n;
 	u32 hash_val;
 
-	write_lock_bh(&ntbl->rwlock);
-
 	htbl = rcu_dereference_protected(ntbl->htbl,
-					lockdep_is_held(&ntbl->rwlock));
+					lockdep_is_held(&priv->lock));
 	if (!htbl)
-		goto out_unlock;
+		return;
 
 	hash_val = ipoib_addr_hash(htbl, neigh->daddr);
 	np = &htbl->buckets[hash_val];
 	for (n = rcu_dereference_protected(*np,
-					    lockdep_is_held(&ntbl->rwlock));
+					    lockdep_is_held(&priv->lock));
 	     n != NULL;
 	     n = rcu_dereference_protected(*np,
-					lockdep_is_held(&ntbl->rwlock))) {
+					lockdep_is_held(&priv->lock))) {
 		if (n == neigh) {
 			/* found */
 			rcu_assign_pointer(*np,
 					   rcu_dereference_protected(neigh->hnext,
-								     lockdep_is_held(&ntbl->rwlock)));
+								     lockdep_is_held(&priv->lock)));
 			call_rcu(&neigh->rcu, ipoib_neigh_reclaim);
-			goto out_unlock;
+			return;
 		} else {
 			np = &n->hnext;
 		}
 	}
-
-out_unlock:
-	write_unlock_bh(&ntbl->rwlock);
-
 }
 
 static int ipoib_neigh_hash_init(struct ipoib_dev_priv *priv)
@@ -1080,7 +1069,6 @@ static int ipoib_neigh_hash_init(struct ipoib_dev_priv *priv)
 
 	clear_bit(IPOIB_NEIGH_TBL_FLUSH, &priv->flags);
 	ntbl->htbl = NULL;
-	rwlock_init(&ntbl->rwlock);
 	htbl = kzalloc(sizeof(*htbl), GFP_KERNEL);
 	if (!htbl)
 		return -ENOMEM;
@@ -1095,6 +1083,7 @@ static int ipoib_neigh_hash_init(struct ipoib_dev_priv *priv)
 	htbl->mask = (size - 1);
 	htbl->buckets = buckets;
 	ntbl->htbl = htbl;
+	htbl->ntbl = ntbl;
 	atomic_set(&ntbl->entries, 0);
 
 	/* start garbage collection */
@@ -1111,9 +1100,11 @@ static void neigh_hash_free_rcu(struct rcu_head *head)
 						    struct ipoib_neigh_hash,
 						    rcu);
 	struct ipoib_neigh __rcu **buckets = htbl->buckets;
+	struct ipoib_neigh_table *ntbl = htbl->ntbl;
 
 	kfree(buckets);
 	kfree(htbl);
+	complete(&ntbl->deleted);
 }
 
 void ipoib_del_neighs_by_gid(struct net_device *dev, u8 *gid)
@@ -1125,10 +1116,10 @@ void ipoib_del_neighs_by_gid(struct net_device *dev, u8 *gid)
 	int i;
 
 	/* remove all neigh connected to a given path or mcast */
-	write_lock_bh(&ntbl->rwlock);
+	spin_lock_irqsave(&priv->lock, flags);
 
 	htbl = rcu_dereference_protected(ntbl->htbl,
-					 lockdep_is_held(&ntbl->rwlock));
+					 lockdep_is_held(&priv->lock));
 
 	if (!htbl)
 		goto out_unlock;
@@ -1138,16 +1129,14 @@ void ipoib_del_neighs_by_gid(struct net_device *dev, u8 *gid)
 		struct ipoib_neigh __rcu **np = &htbl->buckets[i];
 
 		while ((neigh = rcu_dereference_protected(*np,
-							  lockdep_is_held(&ntbl->rwlock))) != NULL) {
+							  lockdep_is_held(&priv->lock))) != NULL) {
 			/* delete neighs belong to this parent */
 			if (!memcmp(gid, neigh->daddr + 4, sizeof (union ib_gid))) {
 				rcu_assign_pointer(*np,
 						   rcu_dereference_protected(neigh->hnext,
-									     lockdep_is_held(&ntbl->rwlock)));
+									     lockdep_is_held(&priv->lock)));
 				/* remove from parent list */
-				spin_lock_irqsave(&priv->lock, flags);
 				list_del(&neigh->list);
-				spin_unlock_irqrestore(&priv->lock, flags);
 				call_rcu(&neigh->rcu, ipoib_neigh_reclaim);
 			} else {
 				np = &neigh->hnext;
@@ -1156,7 +1145,7 @@ void ipoib_del_neighs_by_gid(struct net_device *dev, u8 *gid)
 		}
 	}
 out_unlock:
-	write_unlock_bh(&ntbl->rwlock);
+	spin_unlock_irqrestore(&priv->lock, flags);
 }
 
 static void ipoib_flush_neighs(struct ipoib_dev_priv *priv)
@@ -1164,37 +1153,44 @@ static void ipoib_flush_neighs(struct ipoib_dev_priv *priv)
 	struct ipoib_neigh_table *ntbl = &priv->ntbl;
 	struct ipoib_neigh_hash *htbl;
 	unsigned long flags;
-	int i;
+	int i, wait_flushed = 0;
 
-	write_lock_bh(&ntbl->rwlock);
+	init_completion(&priv->ntbl.flushed);
+
+	spin_lock_irqsave(&priv->lock, flags);
 
 	htbl = rcu_dereference_protected(ntbl->htbl,
-					lockdep_is_held(&ntbl->rwlock));
+					lockdep_is_held(&priv->lock));
 	if (!htbl)
 		goto out_unlock;
 
+	wait_flushed = atomic_read(&priv->ntbl.entries);
+	if (!wait_flushed)
+		goto free_htbl;
+
 	for (i = 0; i < htbl->size; i++) {
 		struct ipoib_neigh *neigh;
 		struct ipoib_neigh __rcu **np = &htbl->buckets[i];
 
 		while ((neigh = rcu_dereference_protected(*np,
-							  lockdep_is_held(&ntbl->rwlock))) != NULL) {
+				       lockdep_is_held(&priv->lock))) != NULL) {
 			rcu_assign_pointer(*np,
 					   rcu_dereference_protected(neigh->hnext,
-								     lockdep_is_held(&ntbl->rwlock)));
+								     lockdep_is_held(&priv->lock)));
 			/* remove from path/mc list */
-			spin_lock_irqsave(&priv->lock, flags);
 			list_del(&neigh->list);
-			spin_unlock_irqrestore(&priv->lock, flags);
 			call_rcu(&neigh->rcu, ipoib_neigh_reclaim);
 		}
 	}
 
+free_htbl:
 	rcu_assign_pointer(ntbl->htbl, NULL);
 	call_rcu(&htbl->rcu, neigh_hash_free_rcu);
 
 out_unlock:
-	write_unlock_bh(&ntbl->rwlock);
+	spin_unlock_irqrestore(&priv->lock, flags);
+	if (wait_flushed)
+		wait_for_completion(&priv->ntbl.flushed);
 }
 
 static void ipoib_neigh_hash_uninit(struct net_device *dev)
@@ -1203,7 +1199,7 @@ static void ipoib_neigh_hash_uninit(struct net_device *dev)
 	int stopped;
 
 	ipoib_dbg(priv, "ipoib_neigh_hash_uninit\n");
-	init_completion(&priv->ntbl.flushed);
+	init_completion(&priv->ntbl.deleted);
 	set_bit(IPOIB_NEIGH_TBL_FLUSH, &priv->flags);
 
 	/* Stop GC if called at init fail need to cancel work */
@@ -1211,10 +1207,9 @@ static void ipoib_neigh_hash_uninit(struct net_device *dev)
 	if (!stopped)
 		cancel_delayed_work(&priv->neigh_reap_task);
 
-	if (atomic_read(&priv->ntbl.entries)) {
-		ipoib_flush_neighs(priv);
-		wait_for_completion(&priv->ntbl.flushed);
-	}
+	ipoib_flush_neighs(priv);
+
+	wait_for_completion(&priv->ntbl.deleted);
 }
 
 

drivers/infiniband/ulp/ipoib/ipoib_multicast.c

@@ -707,9 +707,7 @@ out:
 		neigh = ipoib_neigh_get(dev, daddr);
 		spin_lock_irqsave(&priv->lock, flags);
 		if (!neigh) {
-			spin_unlock_irqrestore(&priv->lock, flags);
 			neigh = ipoib_neigh_alloc(daddr, dev);
-			spin_lock_irqsave(&priv->lock, flags);
 			if (neigh) {
 				kref_get(&mcast->ah->ref);
 				neigh->ah	= mcast->ah;

drivers/net/ethernet/mellanox/mlx4/icm.c

@@ -227,9 +227,10 @@ int mlx4_UNMAP_ICM_AUX(struct mlx4_dev *dev)
 			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
 }
 
-int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj)
 {
-	int i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
+	u32 i = (obj & (table->num_obj - 1)) /
+			(MLX4_TABLE_CHUNK_SIZE / table->obj_size);
 	int ret = 0;
 
 	mutex_lock(&table->mutex);
@@ -262,16 +263,18 @@ out:
 	return ret;
 }
 
-void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj)
 {
-	int i;
+	u32 i;
+	u64 offset;
 
 	i = (obj & (table->num_obj - 1)) / (MLX4_TABLE_CHUNK_SIZE / table->obj_size);
 
 	mutex_lock(&table->mutex);
 
 	if (--table->icm[i]->refcount == 0) {
-		mlx4_UNMAP_ICM(dev, table->virt + i * MLX4_TABLE_CHUNK_SIZE,
+		offset = (u64) i * MLX4_TABLE_CHUNK_SIZE;
+		mlx4_UNMAP_ICM(dev, table->virt + offset,
 			       MLX4_TABLE_CHUNK_SIZE / MLX4_ICM_PAGE_SIZE);
 		mlx4_free_icm(dev, table->icm[i], table->coherent);
 		table->icm[i] = NULL;
@@ -280,9 +283,11 @@ void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj)
 	mutex_unlock(&table->mutex);
 }
 
-void *mlx4_table_find(struct mlx4_icm_table *table, int obj, dma_addr_t *dma_handle)
+void *mlx4_table_find(struct mlx4_icm_table *table, u32 obj,
+			dma_addr_t *dma_handle)
 {
-	int idx, offset, dma_offset, i;
+	int offset, dma_offset, i;
+	u64 idx;
 	struct mlx4_icm_chunk *chunk;
 	struct mlx4_icm *icm;
 	struct page *page = NULL;
@@ -292,7 +297,7 @@ void *mlx4_table_find(struct mlx4_icm_table *table, int obj, dma_addr_t *dma_han
 
 	mutex_lock(&table->mutex);
 
-	idx = (obj & (table->num_obj - 1)) * table->obj_size;
+	idx = (u64) (obj & (table->num_obj - 1)) * table->obj_size;
 	icm = table->icm[idx / MLX4_TABLE_CHUNK_SIZE];
 	dma_offset = offset = idx % MLX4_TABLE_CHUNK_SIZE;
 
@@ -326,10 +331,11 @@ out:
 }
 
 int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
-			 int start, int end)
+			 u32 start, u32 end)
 {
 	int inc = MLX4_TABLE_CHUNK_SIZE / table->obj_size;
-	int i, err;
+	int err;
+	u32 i;
 
 	for (i = start; i <= end; i += inc) {
 		err = mlx4_table_get(dev, table, i);
@@ -349,9 +355,9 @@ fail:
 }
 
 void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
-			  int start, int end)
+			  u32 start, u32 end)
 {
-	int i;
+	u32 i;
 
 	for (i = start; i <= end; i += MLX4_TABLE_CHUNK_SIZE / table->obj_size)
 		mlx4_table_put(dev, table, i);

drivers/net/ethernet/mellanox/mlx4/icm.h

@@ -71,17 +71,17 @@ struct mlx4_icm *mlx4_alloc_icm(struct mlx4_dev *dev, int npages,
 				gfp_t gfp_mask, int coherent);
 void mlx4_free_icm(struct mlx4_dev *dev, struct mlx4_icm *icm, int coherent);
 
-int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
-void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, int obj);
+int mlx4_table_get(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj);
+void mlx4_table_put(struct mlx4_dev *dev, struct mlx4_icm_table *table, u32 obj);
 int mlx4_table_get_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
-			 int start, int end);
+			 u32 start, u32 end);
 void mlx4_table_put_range(struct mlx4_dev *dev, struct mlx4_icm_table *table,
-			  int start, int end);
+			  u32 start, u32 end);
 int mlx4_init_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table,
 			u64 virt, int obj_size,	u32 nobj, int reserved,
 			int use_lowmem, int use_coherent);
 void mlx4_cleanup_icm_table(struct mlx4_dev *dev, struct mlx4_icm_table *table);
-void *mlx4_table_find(struct mlx4_icm_table *table, int obj, dma_addr_t *dma_handle);
+void *mlx4_table_find(struct mlx4_icm_table *table, u32 obj, dma_addr_t *dma_handle);
 
 static inline void mlx4_icm_first(struct mlx4_icm *icm,
 				  struct mlx4_icm_iter *iter)