Merge 3.12-rc6 into driver-core-next

We want these fixes here too.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

4 files changed, 76 insertions, 94 deletions

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 3f831de9a4d8..b1a6cb3a2809 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -164,6 +164,7 @@ int mlx5_vector2eqn(struct mlx5_ib_dev *dev, int vector, int *eqn, int *irqn)
 static int alloc_comp_eqs(struct mlx5_ib_dev *dev)
 {
 	struct mlx5_eq_table *table = &dev->mdev.priv.eq_table;
+	char name[MLX5_MAX_EQ_NAME];
 	struct mlx5_eq *eq, *n;
 	int ncomp_vec;
 	int nent;
@@ -180,11 +181,10 @@ static int alloc_comp_eqs(struct mlx5_ib_dev *dev)
 			goto clean;
 		}
 
-		snprintf(eq->name, MLX5_MAX_EQ_NAME, "mlx5_comp%d", i);
+		snprintf(name, MLX5_MAX_EQ_NAME, "mlx5_comp%d", i);
 		err = mlx5_create_map_eq(&dev->mdev, eq,
 					 i + MLX5_EQ_VEC_COMP_BASE, nent, 0,
-					 eq->name,
-					 &dev->mdev.priv.uuari.uars[0]);
+					 name, &dev->mdev.priv.uuari.uars[0]);
 		if (err) {
 			kfree(eq);
 			goto clean;
@@ -301,9 +301,8 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	props->max_srq_sge	   = max_rq_sg - 1;
 	props->max_fast_reg_page_list_len = (unsigned int)-1;
 	props->local_ca_ack_delay  = dev->mdev.caps.local_ca_ack_delay;
-	props->atomic_cap	   = dev->mdev.caps.flags & MLX5_DEV_CAP_FLAG_ATOMIC ?
-		IB_ATOMIC_HCA : IB_ATOMIC_NONE;
-	props->masked_atomic_cap   = IB_ATOMIC_HCA;
+	props->atomic_cap	   = IB_ATOMIC_NONE;
+	props->masked_atomic_cap   = IB_ATOMIC_NONE;
 	props->max_pkeys	   = be16_to_cpup((__be16 *)(out_mad->data + 28));
 	props->max_mcast_grp	   = 1 << dev->mdev.caps.log_max_mcg;
 	props->max_mcast_qp_attach = dev->mdev.caps.max_qp_mcg;
@@ -1006,6 +1005,11 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event,
 	ibev.device	      = &ibdev->ib_dev;
 	ibev.element.port_num = port;
 
+	if (port < 1 || port > ibdev->num_ports) {
+		mlx5_ib_warn(ibdev, "warning: event on port %d\n", port);
+		return;
+	}
+
 	if (ibdev->ib_active)
 		ib_dispatch_event(&ibev);
 }
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index bd41df95b6f0..3453580b1eb2 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -42,6 +42,10 @@ enum {
 	DEF_CACHE_SIZE	= 10,
 };
 
+enum {
+	MLX5_UMR_ALIGN	= 2048
+};
+
 static __be64 *mr_align(__be64 *ptr, int align)
 {
 	unsigned long mask = align - 1;
@@ -61,13 +65,11 @@ static int order2idx(struct mlx5_ib_dev *dev, int order)
 
 static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
 {
-	struct device *ddev = dev->ib_dev.dma_device;
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent = &cache->ent[c];
 	struct mlx5_create_mkey_mbox_in *in;
 	struct mlx5_ib_mr *mr;
 	int npages = 1 << ent->order;
-	int size = sizeof(u64) * npages;
 	int err = 0;
 	int i;
 
@@ -83,21 +85,6 @@ static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
 		}
 		mr->order = ent->order;
 		mr->umred = 1;
-		mr->pas = kmalloc(size + 0x3f, GFP_KERNEL);
-		if (!mr->pas) {
-			kfree(mr);
-			err = -ENOMEM;
-			goto out;
-		}
-		mr->dma = dma_map_single(ddev, mr_align(mr->pas, 0x40), size,
-					 DMA_TO_DEVICE);
-		if (dma_mapping_error(ddev, mr->dma)) {
-			kfree(mr->pas);
-			kfree(mr);
-			err = -ENOMEM;
-			goto out;
-		}
-
 		in->seg.status = 1 << 6;
 		in->seg.xlt_oct_size = cpu_to_be32((npages + 1) / 2);
 		in->seg.qpn_mkey7_0 = cpu_to_be32(0xffffff << 8);
@@ -108,8 +95,6 @@ static int add_keys(struct mlx5_ib_dev *dev, int c, int num)
 					    sizeof(*in));
 		if (err) {
 			mlx5_ib_warn(dev, "create mkey failed %d\n", err);
-			dma_unmap_single(ddev, mr->dma, size, DMA_TO_DEVICE);
-			kfree(mr->pas);
 			kfree(mr);
 			goto out;
 		}
@@ -129,11 +114,9 @@ out:
 
 static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 {
-	struct device *ddev = dev->ib_dev.dma_device;
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent = &cache->ent[c];
 	struct mlx5_ib_mr *mr;
-	int size;
 	int err;
 	int i;
 
@@ -149,14 +132,10 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 		ent->size--;
 		spin_unlock(&ent->lock);
 		err = mlx5_core_destroy_mkey(&dev->mdev, &mr->mmr);
-		if (err) {
+		if (err)
 			mlx5_ib_warn(dev, "failed destroy mkey\n");
-		} else {
-			size = ALIGN(sizeof(u64) * (1 << mr->order), 0x40);
-			dma_unmap_single(ddev, mr->dma, size, DMA_TO_DEVICE);
-			kfree(mr->pas);
+		else
 			kfree(mr);
-		}
 	}
 }
 
@@ -408,13 +387,12 @@ static void free_cached_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 
 static void clean_keys(struct mlx5_ib_dev *dev, int c)
 {
-	struct device *ddev = dev->ib_dev.dma_device;
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent = &cache->ent[c];
 	struct mlx5_ib_mr *mr;
-	int size;
 	int err;
 
+	cancel_delayed_work(&ent->dwork);
 	while (1) {
 		spin_lock(&ent->lock);
 		if (list_empty(&ent->head)) {
@@ -427,14 +405,10 @@ static void clean_keys(struct mlx5_ib_dev *dev, int c)
 		ent->size--;
 		spin_unlock(&ent->lock);
 		err = mlx5_core_destroy_mkey(&dev->mdev, &mr->mmr);
-		if (err) {
+		if (err)
 			mlx5_ib_warn(dev, "failed destroy mkey\n");
-		} else {
-			size = ALIGN(sizeof(u64) * (1 << mr->order), 0x40);
-			dma_unmap_single(ddev, mr->dma, size, DMA_TO_DEVICE);
-			kfree(mr->pas);
+		else
 			kfree(mr);
-		}
 	}
 }
 
@@ -540,13 +514,15 @@ int mlx5_mr_cache_cleanup(struct mlx5_ib_dev *dev)
 	int i;
 
 	dev->cache.stopped = 1;
-	destroy_workqueue(dev->cache.wq);
+	flush_workqueue(dev->cache.wq);
 
 	mlx5_mr_cache_debugfs_cleanup(dev);
 
 	for (i = 0; i < MAX_MR_CACHE_ENTRIES; i++)
 		clean_keys(dev, i);
 
+	destroy_workqueue(dev->cache.wq);
+
 	return 0;
 }
 
@@ -675,10 +651,12 @@ static struct mlx5_ib_mr *reg_umr(struct ib_pd *pd, struct ib_umem *umem,
 				  int page_shift, int order, int access_flags)
 {
 	struct mlx5_ib_dev *dev = to_mdev(pd->device);
+	struct device *ddev = dev->ib_dev.dma_device;
 	struct umr_common *umrc = &dev->umrc;
 	struct ib_send_wr wr, *bad;
 	struct mlx5_ib_mr *mr;
 	struct ib_sge sg;
+	int size = sizeof(u64) * npages;
 	int err;
 	int i;
 
@@ -697,7 +675,22 @@ static struct mlx5_ib_mr *reg_umr(struct ib_pd *pd, struct ib_umem *umem,
 	if (!mr)
 		return ERR_PTR(-EAGAIN);
 
-	mlx5_ib_populate_pas(dev, umem, page_shift, mr_align(mr->pas, 0x40), 1);
+	mr->pas = kmalloc(size + MLX5_UMR_ALIGN - 1, GFP_KERNEL);
+	if (!mr->pas) {
+		err = -ENOMEM;
+		goto error;
+	}
+
+	mlx5_ib_populate_pas(dev, umem, page_shift,
+			     mr_align(mr->pas, MLX5_UMR_ALIGN), 1);
+
+	mr->dma = dma_map_single(ddev, mr_align(mr->pas, MLX5_UMR_ALIGN), size,
+				 DMA_TO_DEVICE);
+	if (dma_mapping_error(ddev, mr->dma)) {
+		kfree(mr->pas);
+		err = -ENOMEM;
+		goto error;
+	}
 
 	memset(&wr, 0, sizeof(wr));
 	wr.wr_id = (u64)(unsigned long)mr;
@@ -718,6 +711,9 @@ static struct mlx5_ib_mr *reg_umr(struct ib_pd *pd, struct ib_umem *umem,
 	wait_for_completion(&mr->done);
 	up(&umrc->sem);
 
+	dma_unmap_single(ddev, mr->dma, size, DMA_TO_DEVICE);
+	kfree(mr->pas);
+
 	if (mr->status != IB_WC_SUCCESS) {
 		mlx5_ib_warn(dev, "reg umr failed\n");
 		err = -EFAULT;
diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index 045f8cdbd303..5659ea880741 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -203,7 +203,7 @@ static int sq_overhead(enum ib_qp_type qp_type)
 
 	switch (qp_type) {
 	case IB_QPT_XRC_INI:
-		size = sizeof(struct mlx5_wqe_xrc_seg);
+		size += sizeof(struct mlx5_wqe_xrc_seg);
 		/* fall through */
 	case IB_QPT_RC:
 		size += sizeof(struct mlx5_wqe_ctrl_seg) +
@@ -211,20 +211,23 @@ static int sq_overhead(enum ib_qp_type qp_type)
 			sizeof(struct mlx5_wqe_raddr_seg);
 		break;
 
+	case IB_QPT_XRC_TGT:
+		return 0;
+
 	case IB_QPT_UC:
-		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+		size += sizeof(struct mlx5_wqe_ctrl_seg) +
 			sizeof(struct mlx5_wqe_raddr_seg);
 		break;
 
 	case IB_QPT_UD:
 	case IB_QPT_SMI:
 	case IB_QPT_GSI:
-		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+		size += sizeof(struct mlx5_wqe_ctrl_seg) +
 			sizeof(struct mlx5_wqe_datagram_seg);
 		break;
 
 	case MLX5_IB_QPT_REG_UMR:
-		size = sizeof(struct mlx5_wqe_ctrl_seg) +
+		size += sizeof(struct mlx5_wqe_ctrl_seg) +
 			sizeof(struct mlx5_wqe_umr_ctrl_seg) +
 			sizeof(struct mlx5_mkey_seg);
 		break;
@@ -270,7 +273,8 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr,
 		return wqe_size;
 
 	if (wqe_size > dev->mdev.caps.max_sq_desc_sz) {
-		mlx5_ib_dbg(dev, "\n");
+		mlx5_ib_dbg(dev, "wqe_size(%d) > max_sq_desc_sz(%d)\n",
+			    wqe_size, dev->mdev.caps.max_sq_desc_sz);
 		return -EINVAL;
 	}
 
@@ -280,9 +284,15 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr,
 
 	wq_size = roundup_pow_of_two(attr->cap.max_send_wr * wqe_size);
 	qp->sq.wqe_cnt = wq_size / MLX5_SEND_WQE_BB;
+	if (qp->sq.wqe_cnt > dev->mdev.caps.max_wqes) {
+		mlx5_ib_dbg(dev, "wqe count(%d) exceeds limits(%d)\n",
+			    qp->sq.wqe_cnt, dev->mdev.caps.max_wqes);
+		return -ENOMEM;
+	}
 	qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB);
 	qp->sq.max_gs = attr->cap.max_send_sge;
-	qp->sq.max_post = 1 << ilog2(wq_size / wqe_size);
+	qp->sq.max_post = wq_size / wqe_size;
+	attr->cap.max_send_wr = qp->sq.max_post;
 
 	return wq_size;
 }
@@ -1280,6 +1290,11 @@ static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_Q
 					  MLX5_QP_OPTPAR_Q_KEY,
 			[MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_PKEY_INDEX	|
 					   MLX5_QP_OPTPAR_Q_KEY,
+			[MLX5_QP_ST_XRC] = MLX5_QP_OPTPAR_ALT_ADDR_PATH |
+					  MLX5_QP_OPTPAR_RRE            |
+					  MLX5_QP_OPTPAR_RAE            |
+					  MLX5_QP_OPTPAR_RWE            |
+					  MLX5_QP_OPTPAR_PKEY_INDEX,
 		},
 	},
 	[MLX5_QP_STATE_RTR] = {
@@ -1314,6 +1329,11 @@ static enum mlx5_qp_optpar opt_mask[MLX5_QP_NUM_STATE][MLX5_QP_NUM_STATE][MLX5_Q
 		[MLX5_QP_STATE_RTS] = {
 			[MLX5_QP_ST_UD]	 = MLX5_QP_OPTPAR_Q_KEY,
 			[MLX5_QP_ST_MLX] = MLX5_QP_OPTPAR_Q_KEY,
+			[MLX5_QP_ST_UC]	 = MLX5_QP_OPTPAR_RWE,
+			[MLX5_QP_ST_RC]	 = MLX5_QP_OPTPAR_RNR_TIMEOUT	|
+					   MLX5_QP_OPTPAR_RWE		|
+					   MLX5_QP_OPTPAR_RAE		|
+					   MLX5_QP_OPTPAR_RRE,
 		},
 	},
 };
@@ -1651,29 +1671,6 @@ static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg,
 	rseg->reserved = 0;
 }
 
-static void set_atomic_seg(struct mlx5_wqe_atomic_seg *aseg, struct ib_send_wr *wr)
-{
-	if (wr->opcode == IB_WR_ATOMIC_CMP_AND_SWP) {
-		aseg->swap_add = cpu_to_be64(wr->wr.atomic.swap);
-		aseg->compare  = cpu_to_be64(wr->wr.atomic.compare_add);
-	} else if (wr->opcode == IB_WR_MASKED_ATOMIC_FETCH_AND_ADD) {
-		aseg->swap_add = cpu_to_be64(wr->wr.atomic.compare_add);
-		aseg->compare  = cpu_to_be64(wr->wr.atomic.compare_add_mask);
-	} else {
-		aseg->swap_add = cpu_to_be64(wr->wr.atomic.compare_add);
-		aseg->compare  = 0;
-	}
-}
-
-static void set_masked_atomic_seg(struct mlx5_wqe_masked_atomic_seg *aseg,
-				  struct ib_send_wr *wr)
-{
-	aseg->swap_add		= cpu_to_be64(wr->wr.atomic.swap);
-	aseg->swap_add_mask	= cpu_to_be64(wr->wr.atomic.swap_mask);
-	aseg->compare		= cpu_to_be64(wr->wr.atomic.compare_add);
-	aseg->compare_mask	= cpu_to_be64(wr->wr.atomic.compare_add_mask);
-}
-
 static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg,
 			     struct ib_send_wr *wr)
 {
@@ -2063,28 +2060,11 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 
 			case IB_WR_ATOMIC_CMP_AND_SWP:
 			case IB_WR_ATOMIC_FETCH_AND_ADD:
-				set_raddr_seg(seg, wr->wr.atomic.remote_addr,
-					      wr->wr.atomic.rkey);
-				seg  += sizeof(struct mlx5_wqe_raddr_seg);
-
-				set_atomic_seg(seg, wr);
-				seg  += sizeof(struct mlx5_wqe_atomic_seg);
-
-				size += (sizeof(struct mlx5_wqe_raddr_seg) +
-					 sizeof(struct mlx5_wqe_atomic_seg)) / 16;
-				break;
-
 			case IB_WR_MASKED_ATOMIC_CMP_AND_SWP:
-				set_raddr_seg(seg, wr->wr.atomic.remote_addr,
-					      wr->wr.atomic.rkey);
-				seg  += sizeof(struct mlx5_wqe_raddr_seg);
-
-				set_masked_atomic_seg(seg, wr);
-				seg  += sizeof(struct mlx5_wqe_masked_atomic_seg);
-
-				size += (sizeof(struct mlx5_wqe_raddr_seg) +
-					 sizeof(struct mlx5_wqe_masked_atomic_seg)) / 16;
-				break;
+				mlx5_ib_warn(dev, "Atomic operations are not supported yet\n");
+				err = -ENOSYS;
+				*bad_wr = wr;
+				goto out;
 
 			case IB_WR_LOCAL_INV:
 				next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL;
diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c
index 84d297afd6a9..0aa478bc291a 100644
--- a/drivers/infiniband/hw/mlx5/srq.c
+++ b/drivers/infiniband/hw/mlx5/srq.c
@@ -295,7 +295,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd,
 	mlx5_vfree(in);
 	if (err) {
 		mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err);
-		goto err_srq;
+		goto err_usr_kern_srq;
 	}
 
 	mlx5_ib_dbg(dev, "create SRQ with srqn 0x%x\n", srq->msrq.srqn);
@@ -316,6 +316,8 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd,
 
 err_core:
 	mlx5_core_destroy_srq(&dev->mdev, &srq->msrq);
+
+err_usr_kern_srq:
 	if (pd->uobject)
 		destroy_srq_user(pd, srq);
 	else