RDMA/rxe: Use kzmalloc/kfree for mca

Remove rxe_mca (was rxe_mc_elem) from rxe pools and use kzmalloc and kfree
to allocate and free in rxe_mcast.c. Call kzalloc outside of spinlocks to
avoid having to use GFP_ATOMIC.

Link: https://lore.kernel.org/r/20220208211644.123457-3-rpearsonhpe@gmail.com
Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

drivers/infiniband/sw/rxe/rxe.c

@@ -29,7 +29,6 @@ void rxe_dealloc(struct ib_device *ib_dev)
 	rxe_pool_cleanup(&rxe->mr_pool);
 	rxe_pool_cleanup(&rxe->mw_pool);
 	rxe_pool_cleanup(&rxe->mc_grp_pool);
-	rxe_pool_cleanup(&rxe->mc_elem_pool);
 
 	if (rxe->tfm)
 		crypto_free_shash(rxe->tfm);
@@ -163,15 +162,8 @@ static int rxe_init_pools(struct rxe_dev *rxe)
 	if (err)
 		goto err9;
 
-	err = rxe_pool_init(rxe, &rxe->mc_elem_pool, RXE_TYPE_MC_ELEM,
-			    rxe->attr.max_total_mcast_qp_attach);
-	if (err)
-		goto err10;
-
 	return 0;
 
-err10:
-	rxe_pool_cleanup(&rxe->mc_grp_pool);
 err9:
 	rxe_pool_cleanup(&rxe->mw_pool);
 err8:

drivers/infiniband/sw/rxe/rxe_mcast.c

@@ -26,96 +26,104 @@ static int rxe_mcast_delete(struct rxe_dev *rxe, union ib_gid *mgid)
 }
 
 /* caller should hold rxe->mcg_lock */
-static struct rxe_mcg *create_grp(struct rxe_dev *rxe,
-				     struct rxe_pool *pool,
-				     union ib_gid *mgid)
+static struct rxe_mcg *__rxe_create_grp(struct rxe_dev *rxe,
+					struct rxe_pool *pool,
+					union ib_gid *mgid)
 {
-	int err;
 	struct rxe_mcg *grp;
+	int err;
 
-	grp = rxe_alloc_locked(&rxe->mc_grp_pool);
+	grp = rxe_alloc_locked(pool);
 	if (!grp)
 		return ERR_PTR(-ENOMEM);
 
-	INIT_LIST_HEAD(&grp->qp_list);
-	grp->rxe = rxe;
-	rxe_add_key_locked(grp, mgid);
-
 	err = rxe_mcast_add(rxe, mgid);
 	if (unlikely(err)) {
-		rxe_drop_key_locked(grp);
 		rxe_drop_ref(grp);
 		return ERR_PTR(err);
 	}
 
+	INIT_LIST_HEAD(&grp->qp_list);
+	grp->rxe = rxe;
+
+	/* rxe_alloc_locked takes a ref on grp but that will be
+	 * dropped when grp goes out of scope. We need to take a ref
+	 * on the pointer that will be saved in the red-black tree
+	 * by rxe_add_key and used to lookup grp from mgid later.
+	 * Adding key makes object visible to outside so this should
+	 * be done last after the object is ready.
+	 */
+	rxe_add_ref(grp);
+	rxe_add_key_locked(grp, mgid);
+
 	return grp;
 }
 
-static int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid,
-			     struct rxe_mcg **grp_p)
+static struct rxe_mcg *rxe_mcast_get_grp(struct rxe_dev *rxe,
+					 union ib_gid *mgid)
 {
-	int err;
 	struct rxe_mcg *grp;
 	struct rxe_pool *pool = &rxe->mc_grp_pool;
 	unsigned long flags;
 
 	if (rxe->attr.max_mcast_qp_attach == 0)
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 
 	spin_lock_irqsave(&rxe->mcg_lock, flags);
-
 	grp = rxe_pool_get_key_locked(pool, mgid);
-	if (grp)
-		goto done;
-
-	grp = create_grp(rxe, pool, mgid);
-	if (IS_ERR(grp)) {
-		spin_unlock_irqrestore(&rxe->mcg_lock, flags);
-		err = PTR_ERR(grp);
-		return err;
-	}
-
-done:
+	if (!grp)
+		grp = __rxe_create_grp(rxe, pool, mgid);
 	spin_unlock_irqrestore(&rxe->mcg_lock, flags);
-	*grp_p = grp;
-	return 0;
+
+	return grp;
 }
 
 static int rxe_mcast_add_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
-			   struct rxe_mcg *grp)
+				  struct rxe_mcg *grp)
 {
-	int err;
-	struct rxe_mca *elem;
+	struct rxe_mca *mca, *tmp;
 	unsigned long flags;
+	int err;
 
-	/* check to see of the qp is already a member of the group */
+	/* check to see if the qp is already a member of the group */
 	spin_lock_irqsave(&rxe->mcg_lock, flags);
-	list_for_each_entry(elem, &grp->qp_list, qp_list) {
-		if (elem->qp == qp) {
+	list_for_each_entry(mca, &grp->qp_list, qp_list) {
+		if (mca->qp == qp) {
+			spin_unlock_irqrestore(&rxe->mcg_lock, flags);
+			return 0;
+		}
+	}
+	spin_unlock_irqrestore(&rxe->mcg_lock, flags);
+
+	/* speculative alloc new mca without using GFP_ATOMIC */
+	mca = kzalloc(sizeof(*mca), GFP_KERNEL);
+	if (!mca)
+		return -ENOMEM;
+
+	spin_lock_irqsave(&rxe->mcg_lock, flags);
+	/* re-check to see if someone else just attached qp */
+	list_for_each_entry(tmp, &grp->qp_list, qp_list) {
+		if (tmp->qp == qp) {
+			kfree(mca);
 			err = 0;
 			goto out;
 		}
 	}
 
+	/* check limits after checking if already attached */
 	if (grp->num_qp >= rxe->attr.max_mcast_qp_attach) {
+		kfree(mca);
 		err = -ENOMEM;
 		goto out;
 	}
 
-	elem = rxe_alloc_locked(&rxe->mc_elem_pool);
-	if (!elem) {
-		err = -ENOMEM;
-		goto out;
-	}
+	/* protect pointer to qp in mca */
+	rxe_add_ref(qp);
+	mca->qp = qp;
 
-	/* each qp holds a ref on the grp */
-	rxe_add_ref(grp);
-
-	grp->num_qp++;
-	elem->qp = qp;
 	atomic_inc(&qp->mcg_num);
-
-	list_add(&elem->qp_list, &grp->qp_list);
+	grp->num_qp++;
+	list_add(&mca->qp_list, &grp->qp_list);
 
 	err = 0;
 out:
@@ -123,46 +131,80 @@ out:
 	return err;
 }
 
-static int rxe_mcast_drop_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
-				   union ib_gid *mgid)
+/* caller should be holding rxe->mcg_lock */
+static void __rxe_destroy_grp(struct rxe_mcg *grp)
 {
-	struct rxe_mcg *grp;
-	struct rxe_mca *elem, *tmp;
+	/* first remove grp from red-black tree then drop ref */
+	rxe_drop_key_locked(grp);
+	rxe_drop_ref(grp);
+
+	rxe_mcast_delete(grp->rxe, &grp->mgid);
+}
+
+static void rxe_destroy_grp(struct rxe_mcg *grp)
+{
+	struct rxe_dev *rxe = grp->rxe;
 	unsigned long flags;
 
-	grp = rxe_pool_get_key(&rxe->mc_grp_pool, mgid);
-	if (!grp)
-		goto err1;
-
 	spin_lock_irqsave(&rxe->mcg_lock, flags);
-
-	list_for_each_entry_safe(elem, tmp, &grp->qp_list, qp_list) {
-		if (elem->qp == qp) {
-			list_del(&elem->qp_list);
-			grp->num_qp--;
-			atomic_dec(&qp->mcg_num);
-
-			spin_unlock_irqrestore(&rxe->mcg_lock, flags);
-			rxe_drop_ref(elem);
-			rxe_drop_ref(grp);	/* ref held by QP */
-			rxe_drop_ref(grp);	/* ref from get_key */
-			return 0;
-		}
-	}
-
+	__rxe_destroy_grp(grp);
 	spin_unlock_irqrestore(&rxe->mcg_lock, flags);
-	rxe_drop_ref(grp);			/* ref from get_key */
-err1:
-	return -EINVAL;
 }
 
 void rxe_mc_cleanup(struct rxe_pool_elem *elem)
 {
-	struct rxe_mcg *grp = container_of(elem, typeof(*grp), elem);
-	struct rxe_dev *rxe = grp->rxe;
+	/* nothing left to do for now */
+}
 
-	rxe_drop_key(grp);
-	rxe_mcast_delete(rxe, &grp->mgid);
+static int rxe_mcast_drop_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
+				   union ib_gid *mgid)
+{
+	struct rxe_mcg *grp;
+	struct rxe_mca *mca, *tmp;
+	unsigned long flags;
+	int err;
+
+	spin_lock_irqsave(&rxe->mcg_lock, flags);
+	grp = rxe_pool_get_key_locked(&rxe->mc_grp_pool, mgid);
+	if (!grp) {
+		/* we didn't find the mcast group for mgid */
+		err = -EINVAL;
+		goto out_unlock;
+	}
+
+	list_for_each_entry_safe(mca, tmp, &grp->qp_list, qp_list) {
+		if (mca->qp == qp) {
+			list_del(&mca->qp_list);
+
+			/* if the number of qp's attached to the
+			 * mcast group falls to zero go ahead and
+			 * tear it down. This will not free the
+			 * object since we are still holding a ref
+			 * from the get key above.
+			 */
+			grp->num_qp--;
+			if (grp->num_qp <= 0)
+				__rxe_destroy_grp(grp);
+
+			atomic_dec(&qp->mcg_num);
+
+			/* drop the ref from get key. This will free the
+			 * object if num_qp is zero.
+			 */
+			rxe_drop_ref(grp);
+			kfree(mca);
+			err = 0;
+			goto out_unlock;
+		}
+	}
+
+	/* we didn't find the qp on the list */
+	rxe_drop_ref(grp);
+	err = -EINVAL;
+
+out_unlock:
+	spin_unlock_irqrestore(&rxe->mcg_lock, flags);
+	return err;
 }
 
 int rxe_attach_mcast(struct ib_qp *ibqp, union ib_gid *mgid, u16 mlid)
@@ -173,12 +215,16 @@ int rxe_attach_mcast(struct ib_qp *ibqp, union ib_gid *mgid, u16 mlid)
 	struct rxe_mcg *grp;
 
 	/* takes a ref on grp if successful */
-	err = rxe_mcast_get_grp(rxe, mgid, &grp);
-	if (err)
-		return err;
+	grp = rxe_mcast_get_grp(rxe, mgid);
+	if (IS_ERR(grp))
+		return PTR_ERR(grp);
 
 	err = rxe_mcast_add_grp_elem(rxe, qp, grp);
 
+	/* if we failed to attach the first qp to grp tear it down */
+	if (grp->num_qp == 0)
+		rxe_destroy_grp(grp);
+
 	rxe_drop_ref(grp);
 	return err;
 }

drivers/infiniband/sw/rxe/rxe_pool.c

@@ -90,11 +90,6 @@ static const struct rxe_type_info {
 		.key_offset	= offsetof(struct rxe_mcg, mgid),
 		.key_size	= sizeof(union ib_gid),
 	},
-	[RXE_TYPE_MC_ELEM] = {
-		.name		= "rxe-mc_elem",
-		.size		= sizeof(struct rxe_mca),
-		.elem_offset	= offsetof(struct rxe_mca, elem),
-	},
 };
 
 static int rxe_pool_init_index(struct rxe_pool *pool, u32 max, u32 min)

drivers/infiniband/sw/rxe/rxe_pool.h

@@ -23,7 +23,6 @@ enum rxe_elem_type {
 	RXE_TYPE_MR,
 	RXE_TYPE_MW,
 	RXE_TYPE_MC_GRP,
-	RXE_TYPE_MC_ELEM,
 	RXE_NUM_TYPES,		/* keep me last */
 };
 
@@ -156,4 +155,6 @@ void rxe_elem_release(struct kref *kref);
 /* drop a reference on an object */
 #define rxe_drop_ref(obj) kref_put(&(obj)->elem.ref_cnt, rxe_elem_release)
 
+#define rxe_read_ref(obj) kref_read(&(obj)->elem.ref_cnt)
+
 #endif /* RXE_POOL_H */

drivers/infiniband/sw/rxe/rxe_verbs.h

@@ -362,7 +362,6 @@ struct rxe_mcg {
 };
 
 struct rxe_mca {
-	struct rxe_pool_elem	elem;
 	struct list_head	qp_list;
 	struct rxe_qp		*qp;
 };
@@ -396,7 +395,6 @@ struct rxe_dev {
 	struct rxe_pool		mr_pool;
 	struct rxe_pool		mw_pool;
 	struct rxe_pool		mc_grp_pool;
-	struct rxe_pool		mc_elem_pool;
 
 	spinlock_t		mcg_lock;