Merge branch 'mlx5-next'

Amir Vadai says:

====================
net/mlx5: ConnectX-4 100G Ethernet driver

This patchset extends the mlx5_core driver to support Ethernet
functionality. The Ethernet functionality in the mlx5 driver is
integrated into the core driver and not as separated driver. The
IB functionality remains in the mlx5_ib driver as before.

This functionality will enable the Ethernet capability of Mellanox's new
famility of cards - ConnectX-4. Due to the fact that backword
compatability is being kept, existing Connect-IB cards that are using
this driver are fully working with the modified driver, and no issues
with current deployments should be seen.

Like the ConnectX-3 cards, ConnectX-4 is a VPI (Virtual Port Interface -
every port can be configured as Infiniband or Ethernet) card.
Unlike previous generations, the ConnectX-4 has a separate PCI function
per port.

The current code has a limitation that Infiniband and Ethernet port types
are mutually exclusive. When the driver is compiled with Ethernet
support, the Infiniband functionality is disabled and vice versa. To
control that we added the CONFIG_MLX5_CORE_EN config directive
which is 'n' by default, but can be changed by the user.

This limitation is short-lived and would be addressed soon.

As part of this patchset, mlx5_ifc.h was heavily modified [1]. This file
is now generated automatically from the device specification document.
Since this patch is too big for the mail server, it might be missing in
the mailing list, but could be pulled from an external git repository [2].

irq name selection is done at driver initialization and doesn't contain the
interface name as part of the irq name.
irq_balancer will still work thanks to an improvement introduced by Neil Horman
[3] to use sysfs instead of /proc/interrupts.

Patchset was applied on top of commit ed2dfd90 ("tcp/dccp: warn user for
preferred ip_local_port_range")

[1] - Patch 4/11 ("net/mlx5_core: HW data structs/types definitions preparation for mlx5 ehternet driver")
[2] - http://git.openfabrics.org/?p=~amirv/linux.git;a=shortlog;h=refs/heads/mlx5e_v1
[3] - kernel: da8d1c8b PCI/sysfs: add per pci device msi[x] irq listing (v5)
      irq_balancer: 32a7757 Complete rework of how we detect and classify irqs

Thanks to Achiad, Saeed, Yevheny, Or and the whole team for making this happen,
Amir

Changes from V4:
- Removed Patch 3/12: net/mlx5_core: Add EQ renaming mechanism
- Patch 12/12: net/mlx5: Extend mlx5_core to support ConnectX-4 Ethernet functionality
  - irq name is created on driver initialization, therefore it won't contain
    the network interface name in it. This won't effect irq_balancer thanks to
    patches introduced by Neil Horman to use sysfs instead of /proc/interrupts.

Changes from V3:
- PATCH 8/11: net/mlx5_core: Set/Query port MTU commands
  - Return value directly - no need for err.

Changes from V2:
- Improved changelogs and cover-letter
- Added CONFIG_MLX5_EN to disable/enable the Ethernet functionality
- Moved en.h and wq.[ch] into the patch with data-path related code

Changes from V1:
- Added patch 1/12 ("net/mlx5_core,mlx5_ib: Do not use vmap() on coherent
  memory")

Changes from V0:
- Removed V0 Patch 1/11 ("net/mlx5_core: Virtually extend work/completion queue
  buffers by one page") due to misuse of DMA API. Thanks Dave.
- Patch 1/11 ("net/mlx5_core: Set irq affinity hints"):
  - Use kcalloc instead of kzalloc
  - Fix build error when CONFIG_CPUMASK_OFFSTACK=n. Driver loading will fail
    now if cpumask allocation is failing.
  - Using dev_to_node helper. Thanks, Ido.
- Patch 3/11 ("net/mlx5_core: HW data structs/types definitions preparation for
  mlx5 ehternet driver")
  - Removed Mellanox internal comment at the head of the file. Thanks Joe
- Patch 6/11 ("net/mlx5_core: Implement get/set port status")
  - Use direct return of function's result. Thanks Sergei.
- Added Patch 8/11 ("net/mlx5_core: Set/Query port MTU commands")
- Patch 9/11 ("net/mlx5: Ethernet Datapath files")
  - Use rq->wqe_sz instead of skb_end_offset. Thanks Ido.
  - Use dma_wmb() when possible instead of wmb(). Thanks Alex.
  - Fix checkpatch issues
- Patch 10/11 ("net/mlx5: ethernet resources handling")
  - checkpatch issues
  - Added missing include
- Patch 11/11 ("net/mlx5: Ethernet driver")
  - checkpatch issues
  - fixed typo
  - Modified use of affinity hint
  - Using dev_to_node helper. Thanks, Ido.
  - Use new hardware commands from Patch 8/11 ("net/mlx5_core: Set/Query port
    MTU commands") to get/set port MTU in hardware.
  - Removed NETIF_F_SG since hardware ring wraparound is not supported
  - Use dma_wmb() when possible instead of wmb(). Thanks Alex.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/infiniband/hw/mlx5/Kconfig

 config MLX5_INFINIBAND
 	tristate "Mellanox Connect-IB HCA support"
-	depends on NETDEVICES && ETHERNET && PCI
-	select NET_VENDOR_MELLANOX
-	select MLX5_CORE
+	depends on NETDEVICES && ETHERNET && PCI && MLX5_CORE
 	---help---
 	  This driver provides low-level InfiniBand support for
 	  Mellanox Connect-IB PCI Express host channel adapters (HCAs).

drivers/infiniband/hw/mlx5/cq.c

@@ -590,8 +590,7 @@ static int alloc_cq_buf(struct mlx5_ib_dev *dev, struct mlx5_ib_cq_buf *buf,
 {
 	int err;
 
-	err = mlx5_buf_alloc(dev->mdev, nent * cqe_size,
-			     PAGE_SIZE * 2, &buf->buf);
+	err = mlx5_buf_alloc(dev->mdev, nent * cqe_size, &buf->buf);
 	if (err)
 		return err;
 
@@ -754,7 +753,7 @@ struct ib_cq *mlx5_ib_create_cq(struct ib_device *ibdev, int entries,
 		return ERR_PTR(-EINVAL);
 
 	entries = roundup_pow_of_two(entries + 1);
-	if (entries > dev->mdev->caps.gen.max_cqes)
+	if (entries > (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)))
 		return ERR_PTR(-EINVAL);
 
 	cq = kzalloc(sizeof(*cq), GFP_KERNEL);
@@ -921,7 +920,7 @@ int mlx5_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
 	int err;
 	u32 fsel;
 
-	if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_CQ_MODER))
+	if (!MLX5_CAP_GEN(dev->mdev, cq_moderation))
 		return -ENOSYS;
 
 	in = kzalloc(sizeof(*in), GFP_KERNEL);
@@ -1076,7 +1075,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
 	int uninitialized_var(cqe_size);
 	unsigned long flags;
 
-	if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_RESIZE_CQ)) {
+	if (!MLX5_CAP_GEN(dev->mdev, cq_resize)) {
 		pr_info("Firmware does not support resize CQ\n");
 		return -ENOSYS;
 	}
@@ -1085,7 +1084,7 @@ int mlx5_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
 		return -EINVAL;
 
 	entries = roundup_pow_of_two(entries + 1);
-	if (entries > dev->mdev->caps.gen.max_cqes + 1)
+	if (entries >  (1 << MLX5_CAP_GEN(dev->mdev, log_max_cq_sz)) + 1)
 		return -EINVAL;
 
 	if (entries == ibcq->cqe + 1)

drivers/infiniband/hw/mlx5/mad.c

@@ -129,7 +129,7 @@ int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port)
 
 	packet_error = be16_to_cpu(out_mad->status);
 
-	dev->mdev->caps.gen.ext_port_cap[port - 1] = (!err && !packet_error) ?
+	dev->mdev->port_caps[port - 1].ext_port_cap = (!err && !packet_error) ?
 		MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO : 0;
 
 out:

drivers/infiniband/hw/mlx5/main.c

@@ -66,15 +66,13 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 				struct ib_device_attr *props)
 {
 	struct mlx5_ib_dev *dev = to_mdev(ibdev);
+	struct mlx5_core_dev *mdev = dev->mdev;
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
-	struct mlx5_general_caps *gen;
 	int err = -ENOMEM;
 	int max_rq_sg;
 	int max_sq_sg;
-	u64 flags;
 
-	gen = &dev->mdev->caps.gen;
 	in_mad  = kzalloc(sizeof(*in_mad), GFP_KERNEL);
 	out_mad = kmalloc(sizeof(*out_mad), GFP_KERNEL);
 	if (!in_mad || !out_mad)
@@ -96,18 +94,18 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 		IB_DEVICE_PORT_ACTIVE_EVENT		|
 		IB_DEVICE_SYS_IMAGE_GUID		|
 		IB_DEVICE_RC_RNR_NAK_GEN;
-	flags = gen->flags;
-	if (flags & MLX5_DEV_CAP_FLAG_BAD_PKEY_CNTR)
+
+	if (MLX5_CAP_GEN(mdev, pkv))
 		props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR;
-	if (flags & MLX5_DEV_CAP_FLAG_BAD_QKEY_CNTR)
+	if (MLX5_CAP_GEN(mdev, qkv))
 		props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR;
-	if (flags & MLX5_DEV_CAP_FLAG_APM)
+	if (MLX5_CAP_GEN(mdev, apm))
 		props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG;
 	props->device_cap_flags |= IB_DEVICE_LOCAL_DMA_LKEY;
-	if (flags & MLX5_DEV_CAP_FLAG_XRC)
+	if (MLX5_CAP_GEN(mdev, xrc))
 		props->device_cap_flags |= IB_DEVICE_XRC;
 	props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS;
-	if (flags & MLX5_DEV_CAP_FLAG_SIG_HAND_OVER) {
+	if (MLX5_CAP_GEN(mdev, sho)) {
 		props->device_cap_flags |= IB_DEVICE_SIGNATURE_HANDOVER;
 		/* At this stage no support for signature handover */
 		props->sig_prot_cap = IB_PROT_T10DIF_TYPE_1 |
@@ -116,7 +114,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 		props->sig_guard_cap = IB_GUARD_T10DIF_CRC |
 				       IB_GUARD_T10DIF_CSUM;
 	}
-	if (flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST)
+	if (MLX5_CAP_GEN(mdev, block_lb_mc))
 		props->device_cap_flags |= IB_DEVICE_BLOCK_MULTICAST_LOOPBACK;
 
 	props->vendor_id	   = be32_to_cpup((__be32 *)(out_mad->data + 36)) &
@@ -126,37 +124,38 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	memcpy(&props->sys_image_guid, out_mad->data +	4, 8);
 
 	props->max_mr_size	   = ~0ull;
-	props->page_size_cap	   = gen->min_page_sz;
-	props->max_qp		   = 1 << gen->log_max_qp;
-	props->max_qp_wr	   = gen->max_wqes;
-	max_rq_sg = gen->max_rq_desc_sz / sizeof(struct mlx5_wqe_data_seg);
-	max_sq_sg = (gen->max_sq_desc_sz - sizeof(struct mlx5_wqe_ctrl_seg)) /
-		sizeof(struct mlx5_wqe_data_seg);
+	props->page_size_cap	   = 1ull << MLX5_CAP_GEN(mdev, log_pg_sz);
+	props->max_qp		   = 1 << MLX5_CAP_GEN(mdev, log_max_qp);
+	props->max_qp_wr	   = 1 << MLX5_CAP_GEN(mdev, log_max_qp_sz);
+	max_rq_sg =  MLX5_CAP_GEN(mdev, max_wqe_sz_rq) /
+		     sizeof(struct mlx5_wqe_data_seg);
+	max_sq_sg = (MLX5_CAP_GEN(mdev, max_wqe_sz_sq) -
+		     sizeof(struct mlx5_wqe_ctrl_seg)) /
+		     sizeof(struct mlx5_wqe_data_seg);
 	props->max_sge = min(max_rq_sg, max_sq_sg);
-	props->max_cq		   = 1 << gen->log_max_cq;
-	props->max_cqe		   = gen->max_cqes - 1;
-	props->max_mr		   = 1 << gen->log_max_mkey;
-	props->max_pd		   = 1 << gen->log_max_pd;
-	props->max_qp_rd_atom	   = 1 << gen->log_max_ra_req_qp;
-	props->max_qp_init_rd_atom = 1 << gen->log_max_ra_res_qp;
-	props->max_srq		   = 1 << gen->log_max_srq;
-	props->max_srq_wr	   = gen->max_srq_wqes - 1;
-	props->local_ca_ack_delay  = gen->local_ca_ack_delay;
+	props->max_cq		   = 1 << MLX5_CAP_GEN(mdev, log_max_cq);
+	props->max_cqe = (1 << MLX5_CAP_GEN(mdev, log_max_eq_sz)) - 1;
+	props->max_mr		   = 1 << MLX5_CAP_GEN(mdev, log_max_mkey);
+	props->max_pd		   = 1 << MLX5_CAP_GEN(mdev, log_max_pd);
+	props->max_qp_rd_atom	   = 1 << MLX5_CAP_GEN(mdev, log_max_ra_req_qp);
+	props->max_qp_init_rd_atom = 1 << MLX5_CAP_GEN(mdev, log_max_ra_res_qp);
+	props->max_srq		   = 1 << MLX5_CAP_GEN(mdev, log_max_srq);
+	props->max_srq_wr = (1 << MLX5_CAP_GEN(mdev, log_max_srq_sz)) - 1;
+	props->local_ca_ack_delay  = MLX5_CAP_GEN(mdev, local_ca_ack_delay);
 	props->max_res_rd_atom	   = props->max_qp_rd_atom * props->max_qp;
 	props->max_srq_sge	   = max_rq_sg - 1;
 	props->max_fast_reg_page_list_len = (unsigned int)-1;
-	props->local_ca_ack_delay  = gen->local_ca_ack_delay;
 	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 << gen->log_max_mcg;
-	props->max_mcast_qp_attach = gen->max_qp_mcg;
+	props->max_mcast_grp	   = 1 << MLX5_CAP_GEN(mdev, log_max_mcg);
+	props->max_mcast_qp_attach = MLX5_CAP_GEN(mdev, max_qp_mcg);
 	props->max_total_mcast_qp_attach = props->max_mcast_qp_attach *
 					   props->max_mcast_grp;
 	props->max_map_per_fmr = INT_MAX; /* no limit in ConnectIB */
 
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG)
+	if (MLX5_CAP_GEN(mdev, pg))
 		props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
 	props->odp_caps = dev->odp_caps;
 #endif
@@ -172,14 +171,13 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port,
 		       struct ib_port_attr *props)
 {
 	struct mlx5_ib_dev *dev = to_mdev(ibdev);
+	struct mlx5_core_dev *mdev = dev->mdev;
 	struct ib_smp *in_mad  = NULL;
 	struct ib_smp *out_mad = NULL;
-	struct mlx5_general_caps *gen;
 	int ext_active_speed;
 	int err = -ENOMEM;
 
-	gen = &dev->mdev->caps.gen;
-	if (port < 1 || port > gen->num_ports) {
+	if (port < 1 || port > MLX5_CAP_GEN(mdev, num_ports)) {
 		mlx5_ib_warn(dev, "invalid port number %d\n", port);
 		return -EINVAL;
 	}
@@ -210,8 +208,8 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port,
 	props->phys_state	= out_mad->data[33] >> 4;
 	props->port_cap_flags	= be32_to_cpup((__be32 *)(out_mad->data + 20));
 	props->gid_tbl_len	= out_mad->data[50];
-	props->max_msg_sz	= 1 << gen->log_max_msg;
-	props->pkey_tbl_len	= gen->port[port - 1].pkey_table_len;
+	props->max_msg_sz	= 1 << MLX5_CAP_GEN(mdev, log_max_msg);
+	props->pkey_tbl_len	= mdev->port_caps[port - 1].pkey_table_len;
 	props->bad_pkey_cntr	= be16_to_cpup((__be16 *)(out_mad->data + 46));
 	props->qkey_viol_cntr	= be16_to_cpup((__be16 *)(out_mad->data + 48));
 	props->active_width	= out_mad->data[31] & 0xf;
@@ -238,7 +236,7 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port,
 
 	/* If reported active speed is QDR, check if is FDR-10 */
 	if (props->active_speed == 4) {
-		if (gen->ext_port_cap[port - 1] &
+		if (mdev->port_caps[port - 1].ext_port_cap &
 		    MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO) {
 			init_query_mad(in_mad);
 			in_mad->attr_id = MLX5_ATTR_EXTENDED_PORT_INFO;
@@ -392,7 +390,6 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	struct mlx5_ib_alloc_ucontext_req_v2 req;
 	struct mlx5_ib_alloc_ucontext_resp resp;
 	struct mlx5_ib_ucontext *context;
-	struct mlx5_general_caps *gen;
 	struct mlx5_uuar_info *uuari;
 	struct mlx5_uar *uars;
 	int gross_uuars;
@@ -403,7 +400,6 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	int i;
 	size_t reqlen;
 
-	gen = &dev->mdev->caps.gen;
 	if (!dev->ib_active)
 		return ERR_PTR(-EAGAIN);
 
@@ -436,14 +432,14 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 
 	num_uars = req.total_num_uuars / MLX5_NON_FP_BF_REGS_PER_PAGE;
 	gross_uuars = num_uars * MLX5_BF_REGS_PER_PAGE;
-	resp.qp_tab_size      = 1 << gen->log_max_qp;
-	resp.bf_reg_size      = gen->bf_reg_size;
-	resp.cache_line_size  = L1_CACHE_BYTES;
-	resp.max_sq_desc_sz = gen->max_sq_desc_sz;
-	resp.max_rq_desc_sz = gen->max_rq_desc_sz;
-	resp.max_send_wqebb = gen->max_wqes;
-	resp.max_recv_wr = gen->max_wqes;
-	resp.max_srq_recv_wr = gen->max_srq_wqes;
+	resp.qp_tab_size = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp);
+	resp.bf_reg_size = 1 << MLX5_CAP_GEN(dev->mdev, log_bf_reg_size);
+	resp.cache_line_size = L1_CACHE_BYTES;
+	resp.max_sq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq);
+	resp.max_rq_desc_sz = MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq);
+	resp.max_send_wqebb = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz);
+	resp.max_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz);
+	resp.max_srq_recv_wr = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz);
 
 	context = kzalloc(sizeof(*context), GFP_KERNEL);
 	if (!context)
@@ -493,7 +489,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	mutex_init(&context->db_page_mutex);
 
 	resp.tot_uuars = req.total_num_uuars;
-	resp.num_ports = gen->num_ports;
+	resp.num_ports = MLX5_CAP_GEN(dev->mdev, num_ports);
 	err = ib_copy_to_udata(udata, &resp,
 			       sizeof(resp) - sizeof(resp.reserved));
 	if (err)
@@ -895,11 +891,9 @@ static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context,
 
 static void get_ext_port_caps(struct mlx5_ib_dev *dev)
 {
-	struct mlx5_general_caps *gen;
 	int port;
 
-	gen = &dev->mdev->caps.gen;
-	for (port = 1; port <= gen->num_ports; port++)
+	for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++)
 		mlx5_query_ext_port_caps(dev, port);
 }
 
@@ -907,11 +901,9 @@ static int get_port_caps(struct mlx5_ib_dev *dev)
 {
 	struct ib_device_attr *dprops = NULL;
 	struct ib_port_attr *pprops = NULL;
-	struct mlx5_general_caps *gen;
 	int err = -ENOMEM;
 	int port;
 
-	gen = &dev->mdev->caps.gen;
 	pprops = kmalloc(sizeof(*pprops), GFP_KERNEL);
 	if (!pprops)
 		goto out;
@@ -926,14 +918,17 @@ static int get_port_caps(struct mlx5_ib_dev *dev)
 		goto out;
 	}
 
-	for (port = 1; port <= gen->num_ports; port++) {
+	for (port = 1; port <= MLX5_CAP_GEN(dev->mdev, num_ports); port++) {
 		err = mlx5_ib_query_port(&dev->ib_dev, port, pprops);
 		if (err) {
-			mlx5_ib_warn(dev, "query_port %d failed %d\n", port, err);
+			mlx5_ib_warn(dev, "query_port %d failed %d\n",
+				     port, err);
 			break;
 		}
-		gen->port[port - 1].pkey_table_len = dprops->max_pkeys;
-		gen->port[port - 1].gid_table_len = pprops->gid_tbl_len;
+		dev->mdev->port_caps[port - 1].pkey_table_len =
+						dprops->max_pkeys;
+		dev->mdev->port_caps[port - 1].gid_table_len =
+						pprops->gid_tbl_len;
 		mlx5_ib_dbg(dev, "pkey_table_len %d, gid_table_len %d\n",
 			    dprops->max_pkeys, pprops->gid_tbl_len);
 	}
@@ -1207,8 +1202,8 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	strlcpy(dev->ib_dev.name, "mlx5_%d", IB_DEVICE_NAME_MAX);
 	dev->ib_dev.owner		= THIS_MODULE;
 	dev->ib_dev.node_type		= RDMA_NODE_IB_CA;
-	dev->ib_dev.local_dma_lkey	= mdev->caps.gen.reserved_lkey;
-	dev->num_ports		= mdev->caps.gen.num_ports;
+	dev->ib_dev.local_dma_lkey	= 0 /* not supported for now */;
+	dev->num_ports		= MLX5_CAP_GEN(mdev, num_ports);
 	dev->ib_dev.phys_port_cnt     = dev->num_ports;
 	dev->ib_dev.num_comp_vectors    =
 		dev->mdev->priv.eq_table.num_comp_vectors;
@@ -1286,9 +1281,9 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev)
 	dev->ib_dev.free_fast_reg_page_list  = mlx5_ib_free_fast_reg_page_list;
 	dev->ib_dev.check_mr_status	= mlx5_ib_check_mr_status;
 
-	mlx5_ib_internal_query_odp_caps(dev);
+	mlx5_ib_internal_fill_odp_caps(dev);
 
-	if (mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_XRC) {
+	if (MLX5_CAP_GEN(mdev, xrc)) {
 		dev->ib_dev.alloc_xrcd = mlx5_ib_alloc_xrcd;
 		dev->ib_dev.dealloc_xrcd = mlx5_ib_dealloc_xrcd;
 		dev->ib_dev.uverbs_cmd_mask |=

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -617,7 +617,7 @@ int mlx5_ib_check_mr_status(struct ib_mr *ibmr, u32 check_mask,
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 extern struct workqueue_struct *mlx5_ib_page_fault_wq;
 
-int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev);
+void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev);
 void mlx5_ib_mr_pfault_handler(struct mlx5_ib_qp *qp,
 			       struct mlx5_ib_pfault *pfault);
 void mlx5_ib_odp_create_qp(struct mlx5_ib_qp *qp);
@@ -631,9 +631,9 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 			      unsigned long end);
 
 #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
-static inline int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev)
+static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 {
-	return 0;
+	return;
 }
 
 static inline void mlx5_ib_odp_create_qp(struct mlx5_ib_qp *qp)		{}

drivers/infiniband/hw/mlx5/mr.c

@@ -975,8 +975,7 @@ static struct mlx5_ib_mr *reg_create(struct ib_pd *pd, u64 virt_addr,
 	struct mlx5_ib_mr *mr;
 	int inlen;
 	int err;
-	bool pg_cap = !!(dev->mdev->caps.gen.flags &
-			 MLX5_DEV_CAP_FLAG_ON_DMND_PG);
+	bool pg_cap = !!(MLX5_CAP_GEN(dev->mdev, pg));
 
 	mr = kzalloc(sizeof(*mr), GFP_KERNEL);
 	if (!mr)

drivers/infiniband/hw/mlx5/odp.c

@@ -109,40 +109,33 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 	ib_umem_odp_unmap_dma_pages(umem, start, end);
 }
 
-#define COPY_ODP_BIT_MLX_TO_IB(reg, ib_caps, field_name, bit_name) do {	\
-	if (be32_to_cpu(reg.field_name) & MLX5_ODP_SUPPORT_##bit_name)	\
-		ib_caps->field_name |= IB_ODP_SUPPORT_##bit_name;	\
-} while (0)
-
-int mlx5_ib_internal_query_odp_caps(struct mlx5_ib_dev *dev)
+void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev)
 {
-	int err;
-	struct mlx5_odp_caps hw_caps;
 	struct ib_odp_caps *caps = &dev->odp_caps;
 
 	memset(caps, 0, sizeof(*caps));
 
-	if (!(dev->mdev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG))
-		return 0;
-
-	err = mlx5_query_odp_caps(dev->mdev, &hw_caps);
-	if (err)
-		goto out;
+	if (!MLX5_CAP_GEN(dev->mdev, pg))
+		return;
 
 	caps->general_caps = IB_ODP_SUPPORT;
-	COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.ud_odp_caps,
-			       SEND);
-	COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps,
-			       SEND);
-	COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps,
-			       RECV);
-	COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps,
-			       WRITE);
-	COPY_ODP_BIT_MLX_TO_IB(hw_caps, caps, per_transport_caps.rc_odp_caps,
-			       READ);
-
-out:
-	return err;
+
+	if (MLX5_CAP_ODP(dev->mdev, ud_odp_caps.send))
+		caps->per_transport_caps.ud_odp_caps |= IB_ODP_SUPPORT_SEND;
+
+	if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.send))
+		caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_SEND;
+
+	if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.receive))
+		caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_RECV;
+
+	if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.write))
+		caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_WRITE;
+
+	if (MLX5_CAP_ODP(dev->mdev, rc_odp_caps.read))
+		caps->per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_READ;
+
+	return;
 }
 
 static struct mlx5_ib_mr *mlx5_ib_odp_find_mr_lkey(struct mlx5_ib_dev *dev,

drivers/infiniband/hw/mlx5/qp.c

@@ -220,13 +220,11 @@ static void mlx5_ib_qp_event(struct mlx5_core_qp *qp, int type)
 static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap,
 		       int has_rq, struct mlx5_ib_qp *qp, struct mlx5_ib_create_qp *ucmd)
 {
-	struct mlx5_general_caps *gen;
 	int wqe_size;
 	int wq_size;
 
-	gen = &dev->mdev->caps.gen;
 	/* Sanity check RQ size before proceeding */
-	if (cap->max_recv_wr  > gen->max_wqes)
+	if (cap->max_recv_wr > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz)))
 		return -EINVAL;
 
 	if (!has_rq) {
@@ -246,10 +244,11 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap,
 			wq_size = roundup_pow_of_two(cap->max_recv_wr) * wqe_size;
 			wq_size = max_t(int, wq_size, MLX5_SEND_WQE_BB);
 			qp->rq.wqe_cnt = wq_size / wqe_size;
-			if (wqe_size > gen->max_rq_desc_sz) {
+			if (wqe_size > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_rq)) {
 				mlx5_ib_dbg(dev, "wqe_size %d, max %d\n",
 					    wqe_size,
-					    gen->max_rq_desc_sz);
+					    MLX5_CAP_GEN(dev->mdev,
+							 max_wqe_sz_rq));
 				return -EINVAL;
 			}
 			qp->rq.wqe_shift = ilog2(wqe_size);
@@ -330,11 +329,9 @@ static int calc_send_wqe(struct ib_qp_init_attr *attr)
 static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr,
 			struct mlx5_ib_qp *qp)
 {
-	struct mlx5_general_caps *gen;
 	int wqe_size;
 	int wq_size;
 
-	gen = &dev->mdev->caps.gen;
 	if (!attr->cap.max_send_wr)
 		return 0;
 
@@ -343,9 +340,9 @@ static int calc_sq_size(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr,
 	if (wqe_size < 0)
 		return wqe_size;
 
-	if (wqe_size > gen->max_sq_desc_sz) {
+	if (wqe_size > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)) {
 		mlx5_ib_dbg(dev, "wqe_size(%d) > max_sq_desc_sz(%d)\n",
-			    wqe_size, gen->max_sq_desc_sz);
+			    wqe_size, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq));
 		return -EINVAL;
 	}
 
@@ -358,9 +355,10 @@ 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 > gen->max_wqes) {
+	if (qp->sq.wqe_cnt > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) {
 		mlx5_ib_dbg(dev, "wqe count(%d) exceeds limits(%d)\n",
-			    qp->sq.wqe_cnt, gen->max_wqes);
+			    qp->sq.wqe_cnt,
+			    1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz));
 		return -ENOMEM;
 	}
 	qp->sq.wqe_shift = ilog2(MLX5_SEND_WQE_BB);
@@ -375,13 +373,11 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev,
 			    struct mlx5_ib_qp *qp,
 			    struct mlx5_ib_create_qp *ucmd)
 {
-	struct mlx5_general_caps *gen;
 	int desc_sz = 1 << qp->sq.wqe_shift;
 
-	gen = &dev->mdev->caps.gen;
-	if (desc_sz > gen->max_sq_desc_sz) {
+	if (desc_sz > MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq)) {
 		mlx5_ib_warn(dev, "desc_sz %d, max_sq_desc_sz %d\n",
-			     desc_sz, gen->max_sq_desc_sz);
+			     desc_sz, MLX5_CAP_GEN(dev->mdev, max_wqe_sz_sq));
 		return -EINVAL;
 	}
 
@@ -393,9 +389,10 @@ static int set_user_buf_size(struct mlx5_ib_dev *dev,
 
 	qp->sq.wqe_cnt = ucmd->sq_wqe_count;
 
-	if (qp->sq.wqe_cnt > gen->max_wqes) {
+	if (qp->sq.wqe_cnt > (1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz))) {
 		mlx5_ib_warn(dev, "wqe_cnt %d, max_wqes %d\n",
-			     qp->sq.wqe_cnt, gen->max_wqes);
+			     qp->sq.wqe_cnt,
+			     1 << MLX5_CAP_GEN(dev->mdev, log_max_qp_sz));
 		return -EINVAL;
 	}
 
@@ -768,7 +765,7 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev,
 	qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift;
 	qp->buf_size = err + (qp->rq.wqe_cnt << qp->rq.wqe_shift);
 
-	err = mlx5_buf_alloc(dev->mdev, qp->buf_size, PAGE_SIZE * 2, &qp->buf);
+	err = mlx5_buf_alloc(dev->mdev, qp->buf_size, &qp->buf);
 	if (err) {
 		mlx5_ib_dbg(dev, "err %d\n", err);
 		goto err_uuar;
@@ -866,22 +863,21 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 			    struct ib_udata *udata, struct mlx5_ib_qp *qp)
 {
 	struct mlx5_ib_resources *devr = &dev->devr;
+	struct mlx5_core_dev *mdev = dev->mdev;
 	struct mlx5_ib_create_qp_resp resp;
 	struct mlx5_create_qp_mbox_in *in;
-	struct mlx5_general_caps *gen;
 	struct mlx5_ib_create_qp ucmd;
 	int inlen = sizeof(*in);
 	int err;
 
 	mlx5_ib_odp_create_qp(qp);
 
-	gen = &dev->mdev->caps.gen;
 	mutex_init(&qp->mutex);
 	spin_lock_init(&qp->sq.lock);
 	spin_lock_init(&qp->rq.lock);
 
 	if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK) {
-		if (!(gen->flags & MLX5_DEV_CAP_FLAG_BLOCK_MCAST)) {
+		if (!MLX5_CAP_GEN(mdev, block_lb_mc)) {
 			mlx5_ib_dbg(dev, "block multicast loopback isn't supported\n");
 			return -EINVAL;
 		} else {
@@ -914,15 +910,17 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 
 	if (pd) {
 		if (pd->uobject) {
+			__u32 max_wqes =
+				1 << MLX5_CAP_GEN(mdev, log_max_qp_sz);
 			mlx5_ib_dbg(dev, "requested sq_wqe_count (%d)\n", ucmd.sq_wqe_count);
 			if (ucmd.rq_wqe_shift != qp->rq.wqe_shift ||
 			    ucmd.rq_wqe_count != qp->rq.wqe_cnt) {
 				mlx5_ib_dbg(dev, "invalid rq params\n");
 				return -EINVAL;
 			}
-			if (ucmd.sq_wqe_count > gen->max_wqes) {
+			if (ucmd.sq_wqe_count > max_wqes) {
 				mlx5_ib_dbg(dev, "requested sq_wqe_count (%d) > max allowed (%d)\n",
-					    ucmd.sq_wqe_count, gen->max_wqes);
+					    ucmd.sq_wqe_count, max_wqes);
 				return -EINVAL;
 			}
 			err = create_user_qp(dev, pd, qp, udata, &in, &resp, &inlen);
@@ -1226,7 +1224,6 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd,
 				struct ib_qp_init_attr *init_attr,
 				struct ib_udata *udata)
 {
-	struct mlx5_general_caps *gen;
 	struct mlx5_ib_dev *dev;
 	struct mlx5_ib_qp *qp;
 	u16 xrcdn = 0;
@@ -1244,12 +1241,11 @@ struct ib_qp *mlx5_ib_create_qp(struct ib_pd *pd,
 		}
 		dev = to_mdev(to_mxrcd(init_attr->xrcd)->ibxrcd.device);
 	}
-	gen = &dev->mdev->caps.gen;
 
 	switch (init_attr->qp_type) {
 	case IB_QPT_XRC_TGT:
 	case IB_QPT_XRC_INI:
-		if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC)) {
+		if (!MLX5_CAP_GEN(dev->mdev, xrc)) {
 			mlx5_ib_dbg(dev, "XRC not supported\n");
 			return ERR_PTR(-ENOSYS);
 		}
@@ -1356,9 +1352,6 @@ enum {
 
 static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate)
 {
-	struct mlx5_general_caps *gen;
-
-	gen = &dev->mdev->caps.gen;
 	if (rate == IB_RATE_PORT_CURRENT) {
 		return 0;
 	} else if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_300_GBPS) {
@@ -1366,7 +1359,7 @@ static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate)
 	} else {
 		while (rate != IB_RATE_2_5_GBPS &&
 		       !(1 << (rate + MLX5_STAT_RATE_OFFSET) &
-			 gen->stat_rate_support))
+			 MLX5_CAP_GEN(dev->mdev, stat_rate_support)))
 			--rate;
 	}
 
@@ -1377,10 +1370,8 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah,
 			 struct mlx5_qp_path *path, u8 port, int attr_mask,
 			 u32 path_flags, const struct ib_qp_attr *attr)
 {
-	struct mlx5_general_caps *gen;
 	int err;
 
-	gen = &dev->mdev->caps.gen;
 	path->fl = (path_flags & MLX5_PATH_FLAG_FL) ? 0x80 : 0;
 	path->free_ar = (path_flags & MLX5_PATH_FLAG_FREE_AR) ? 0x80 : 0;
 
@@ -1391,9 +1382,11 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, const struct ib_ah_attr *ah,
 	path->rlid	= cpu_to_be16(ah->dlid);
 
 	if (ah->ah_flags & IB_AH_GRH) {
-		if (ah->grh.sgid_index >= gen->port[port - 1].gid_table_len) {
+		if (ah->grh.sgid_index >=
+		    dev->mdev->port_caps[port - 1].gid_table_len) {
 			pr_err("sgid_index (%u) too large. max is %d\n",
-			       ah->grh.sgid_index, gen->port[port - 1].gid_table_len);
+			       ah->grh.sgid_index,
+			       dev->mdev->port_caps[port - 1].gid_table_len);
 			return -EINVAL;
 		}
 		path->grh_mlid |= 1 << 7;
@@ -1570,7 +1563,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
 	struct mlx5_ib_qp *qp = to_mqp(ibqp);
 	struct mlx5_ib_cq *send_cq, *recv_cq;
 	struct mlx5_qp_context *context;
-	struct mlx5_general_caps *gen;
 	struct mlx5_modify_qp_mbox_in *in;
 	struct mlx5_ib_pd *pd;
 	enum mlx5_qp_state mlx5_cur, mlx5_new;
@@ -1579,7 +1571,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
 	int mlx5_st;
 	int err;
 
-	gen = &dev->mdev->caps.gen;
 	in = kzalloc(sizeof(*in), GFP_KERNEL);
 	if (!in)
 		return -ENOMEM;
@@ -1619,7 +1610,8 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp,
 			err = -EINVAL;
 			goto out;
 		}
-		context->mtu_msgmax = (attr->path_mtu << 5) | gen->log_max_msg;
+		context->mtu_msgmax = (attr->path_mtu << 5) |
+				      (u8)MLX5_CAP_GEN(dev->mdev, log_max_msg);
 	}
 
 	if (attr_mask & IB_QP_DEST_QPN)
@@ -1777,11 +1769,9 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
 	struct mlx5_ib_qp *qp = to_mqp(ibqp);
 	enum ib_qp_state cur_state, new_state;
-	struct mlx5_general_caps *gen;
 	int err = -EINVAL;
 	int port;
 
-	gen = &dev->mdev->caps.gen;
 	mutex_lock(&qp->mutex);
 
 	cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state;
@@ -1793,21 +1783,25 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 		goto out;
 
 	if ((attr_mask & IB_QP_PORT) &&
-	    (attr->port_num == 0 || attr->port_num > gen->num_ports))
+	    (attr->port_num == 0 ||
+	     attr->port_num > MLX5_CAP_GEN(dev->mdev, num_ports)))
 		goto out;
 
 	if (attr_mask & IB_QP_PKEY_INDEX) {
 		port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port;
-		if (attr->pkey_index >= gen->port[port - 1].pkey_table_len)
+		if (attr->pkey_index >=
+		    dev->mdev->port_caps[port - 1].pkey_table_len)
 			goto out;
 	}
 
 	if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC &&
-	    attr->max_rd_atomic > (1 << gen->log_max_ra_res_qp))
+	    attr->max_rd_atomic >
+	    (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_res_qp)))
 		goto out;
 
 	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC &&
-	    attr->max_dest_rd_atomic > (1 << gen->log_max_ra_req_qp))
+	    attr->max_dest_rd_atomic >
+	    (1 << MLX5_CAP_GEN(dev->mdev, log_max_ra_req_qp)))
 		goto out;
 
 	if (cur_state == new_state && cur_state == IB_QPS_RESET) {
@@ -3009,7 +3003,7 @@ static void to_ib_ah_attr(struct mlx5_ib_dev *ibdev, struct ib_ah_attr *ib_ah_at
 	ib_ah_attr->port_num	  = path->port;
 
 	if (ib_ah_attr->port_num == 0 ||
-	    ib_ah_attr->port_num > dev->caps.gen.num_ports)
+	    ib_ah_attr->port_num > MLX5_CAP_GEN(dev, num_ports))
 		return;
 
 	ib_ah_attr->sl = path->sl & 0xf;
@@ -3135,12 +3129,10 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev,
 					  struct ib_udata *udata)
 {
 	struct mlx5_ib_dev *dev = to_mdev(ibdev);
-	struct mlx5_general_caps *gen;
 	struct mlx5_ib_xrcd *xrcd;
 	int err;
 
-	gen = &dev->mdev->caps.gen;
-	if (!(gen->flags & MLX5_DEV_CAP_FLAG_XRC))
+	if (!MLX5_CAP_GEN(dev->mdev, xrc))
 		return ERR_PTR(-ENOSYS);
 
 	xrcd = kmalloc(sizeof(*xrcd), GFP_KERNEL);

drivers/infiniband/hw/mlx5/srq.c

@@ -165,7 +165,7 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq,
 		return err;
 	}
 
-	if (mlx5_buf_alloc(dev->mdev, buf_size, PAGE_SIZE * 2, &srq->buf)) {
+	if (mlx5_buf_alloc(dev->mdev, buf_size, &srq->buf)) {
 		mlx5_ib_dbg(dev, "buf alloc failed\n");
 		err = -ENOMEM;
 		goto err_db;
@@ -236,7 +236,6 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd,
 				  struct ib_udata *udata)
 {
 	struct mlx5_ib_dev *dev = to_mdev(pd->device);
-	struct mlx5_general_caps *gen;
 	struct mlx5_ib_srq *srq;
 	int desc_size;
 	int buf_size;
@@ -245,13 +244,13 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd,
 	int uninitialized_var(inlen);
 	int is_xrc;
 	u32 flgs, xrcdn;
+	__u32 max_srq_wqes = 1 << MLX5_CAP_GEN(dev->mdev, log_max_srq_sz);
 
-	gen = &dev->mdev->caps.gen;
 	/* Sanity check SRQ size before proceeding */
-	if (init_attr->attr.max_wr >= gen->max_srq_wqes) {
+	if (init_attr->attr.max_wr >= max_srq_wqes) {
 		mlx5_ib_dbg(dev, "max_wr %d, cap %d\n",
 			    init_attr->attr.max_wr,
-			    gen->max_srq_wqes);
+			    max_srq_wqes);
 		return ERR_PTR(-EINVAL);
 	}
 

drivers/net/ethernet/mellanox/mlx5/core/Kconfig

@@ -3,6 +3,18 @@
 #
 
 config MLX5_CORE
-	tristate
+	tristate "Mellanox Technologies ConnectX-4 and Connect-IB core driver"
 	depends on PCI
 	default n
+	---help---
+	  Core driver for low level functionality of the ConnectX-4 and
+	  Connect-IB cards by Mellanox Technologies.
+
+config MLX5_CORE_EN
+	bool "Mellanox Technologies ConnectX-4 Ethernet support"
+	depends on MLX5_INFINIBAND=n && NETDEVICES && ETHERNET && PCI && MLX5_CORE
+	default n
+	---help---
+	  Ethernet support in Mellanox Technologies ConnectX-4 NIC.
+	  Ethernet and Infiniband support in ConnectX-4 are currently mutually
+	  exclusive.

drivers/net/ethernet/mellanox/mlx5/core/Makefile

@@ -3,3 +3,6 @@ obj-$(CONFIG_MLX5_CORE)		+= mlx5_core.o
 mlx5_core-y :=	main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
 		health.o mcg.o cq.o srq.o alloc.o qp.o port.o mr.o pd.o   \
 		mad.o
+mlx5_core-$(CONFIG_MLX5_CORE_EN) += wq.o flow_table.o vport.o transobj.o \
+		en_main.o en_flow_table.o en_ethtool.o en_tx.o en_rx.o \
+		en_txrx.o

drivers/net/ethernet/mellanox/mlx5/core/alloc.c

@@ -42,95 +42,36 @@
 #include "mlx5_core.h"
 
 /* Handling for queue buffers -- we allocate a bunch of memory and
- * register it in a memory region at HCA virtual address 0.  If the
- * requested size is > max_direct, we split the allocation into
- * multiple pages, so we don't require too much contiguous memory.
+ * register it in a memory region at HCA virtual address 0.
  */
 
-int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, int max_direct,
-		   struct mlx5_buf *buf)
+int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, struct mlx5_buf *buf)
 {
 	dma_addr_t t;
 
 	buf->size = size;
-	if (size <= max_direct) {
-		buf->nbufs        = 1;
-		buf->npages       = 1;
-		buf->page_shift   = (u8)get_order(size) + PAGE_SHIFT;
-		buf->direct.buf   = dma_zalloc_coherent(&dev->pdev->dev,
-							size, &t, GFP_KERNEL);
-		if (!buf->direct.buf)
-			return -ENOMEM;
-
-		buf->direct.map = t;
-
-		while (t & ((1 << buf->page_shift) - 1)) {
-			--buf->page_shift;
-			buf->npages *= 2;
-		}
-	} else {
-		int i;
-
-		buf->direct.buf  = NULL;
-		buf->nbufs       = (size + PAGE_SIZE - 1) / PAGE_SIZE;
-		buf->npages      = buf->nbufs;
-		buf->page_shift  = PAGE_SHIFT;
-		buf->page_list   = kcalloc(buf->nbufs, sizeof(*buf->page_list),
-					   GFP_KERNEL);
-		if (!buf->page_list)
-			return -ENOMEM;
-
-		for (i = 0; i < buf->nbufs; i++) {
-			buf->page_list[i].buf =
-				dma_zalloc_coherent(&dev->pdev->dev, PAGE_SIZE,
-						    &t, GFP_KERNEL);
-			if (!buf->page_list[i].buf)
-				goto err_free;
-
-			buf->page_list[i].map = t;
-		}
-
-		if (BITS_PER_LONG == 64) {
-			struct page **pages;
-			pages = kmalloc(sizeof(*pages) * buf->nbufs, GFP_KERNEL);
-			if (!pages)
-				goto err_free;
-			for (i = 0; i < buf->nbufs; i++)
-				pages[i] = virt_to_page(buf->page_list[i].buf);
-			buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP, PAGE_KERNEL);
-			kfree(pages);
-			if (!buf->direct.buf)
-				goto err_free;
-		}
-	}
+	buf->npages       = 1;
+	buf->page_shift   = (u8)get_order(size) + PAGE_SHIFT;
+	buf->direct.buf   = dma_zalloc_coherent(&dev->pdev->dev,
+						size, &t, GFP_KERNEL);
+	if (!buf->direct.buf)
+		return -ENOMEM;
 
-	return 0;
+	buf->direct.map = t;
 
-err_free:
-	mlx5_buf_free(dev, buf);
+	while (t & ((1 << buf->page_shift) - 1)) {
+		--buf->page_shift;
+		buf->npages *= 2;
+	}
 
-	return -ENOMEM;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(mlx5_buf_alloc);
 
 void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_buf *buf)
 {
-	int i;
-
-	if (buf->nbufs == 1)
-		dma_free_coherent(&dev->pdev->dev, buf->size, buf->direct.buf,
-				  buf->direct.map);
-	else {
-		if (BITS_PER_LONG == 64)
-			vunmap(buf->direct.buf);
-
-		for (i = 0; i < buf->nbufs; i++)
-			if (buf->page_list[i].buf)
-				dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
-						  buf->page_list[i].buf,
-						  buf->page_list[i].map);
-		kfree(buf->page_list);
-	}
+	dma_free_coherent(&dev->pdev->dev, buf->size, buf->direct.buf,
+			  buf->direct.map);
 }
 EXPORT_SYMBOL_GPL(mlx5_buf_free);
 
@@ -230,10 +171,7 @@ void mlx5_fill_page_array(struct mlx5_buf *buf, __be64 *pas)
 	int i;
 
 	for (i = 0; i < buf->npages; i++) {
-		if (buf->nbufs == 1)
-			addr = buf->direct.map + (i << buf->page_shift);
-		else
-			addr = buf->page_list[i].map;
+		addr = buf->direct.map + (i << buf->page_shift);
 
 		pas[i] = cpu_to_be64(addr);
 	}

drivers/net/ethernet/mellanox/mlx5/core/cmd.c

@@ -75,25 +75,6 @@ enum {
 	MLX5_CMD_DELIVERY_STAT_CMD_DESCR_ERR		= 0x10,
 };
 
-enum {
-	MLX5_CMD_STAT_OK			= 0x0,
-	MLX5_CMD_STAT_INT_ERR			= 0x1,
-	MLX5_CMD_STAT_BAD_OP_ERR		= 0x2,
-	MLX5_CMD_STAT_BAD_PARAM_ERR		= 0x3,
-	MLX5_CMD_STAT_BAD_SYS_STATE_ERR		= 0x4,
-	MLX5_CMD_STAT_BAD_RES_ERR		= 0x5,
-	MLX5_CMD_STAT_RES_BUSY			= 0x6,
-	MLX5_CMD_STAT_LIM_ERR			= 0x8,
-	MLX5_CMD_STAT_BAD_RES_STATE_ERR		= 0x9,
-	MLX5_CMD_STAT_IX_ERR			= 0xa,
-	MLX5_CMD_STAT_NO_RES_ERR		= 0xf,
-	MLX5_CMD_STAT_BAD_INP_LEN_ERR		= 0x50,
-	MLX5_CMD_STAT_BAD_OUTP_LEN_ERR		= 0x51,
-	MLX5_CMD_STAT_BAD_QP_STATE_ERR		= 0x10,
-	MLX5_CMD_STAT_BAD_PKT_ERR		= 0x30,
-	MLX5_CMD_STAT_BAD_SIZE_OUTS_CQES_ERR	= 0x40,
-};
-
 static struct mlx5_cmd_work_ent *alloc_cmd(struct mlx5_cmd *cmd,
 					   struct mlx5_cmd_msg *in,
 					   struct mlx5_cmd_msg *out,
@@ -390,8 +371,17 @@ const char *mlx5_command_str(int command)
 	case MLX5_CMD_OP_ARM_RQ:
 		return "ARM_RQ";
 
-	case MLX5_CMD_OP_RESIZE_SRQ:
-		return "RESIZE_SRQ";
+	case MLX5_CMD_OP_CREATE_XRC_SRQ:
+		return "CREATE_XRC_SRQ";
+
+	case MLX5_CMD_OP_DESTROY_XRC_SRQ:
+		return "DESTROY_XRC_SRQ";
+
+	case MLX5_CMD_OP_QUERY_XRC_SRQ:
+		return "QUERY_XRC_SRQ";
+
+	case MLX5_CMD_OP_ARM_XRC_SRQ:
+		return "ARM_XRC_SRQ";
 
 	case MLX5_CMD_OP_ALLOC_PD:
 		return "ALLOC_PD";
@@ -408,8 +398,8 @@ const char *mlx5_command_str(int command)
 	case MLX5_CMD_OP_ATTACH_TO_MCG:
 		return "ATTACH_TO_MCG";
 
-	case MLX5_CMD_OP_DETACH_FROM_MCG:
-		return "DETACH_FROM_MCG";
+	case MLX5_CMD_OP_DETTACH_FROM_MCG:
+		return "DETTACH_FROM_MCG";
 
 	case MLX5_CMD_OP_ALLOC_XRCD:
 		return "ALLOC_XRCD";

drivers/net/ethernet/mellanox/mlx5/core/cq.c

@@ -219,6 +219,24 @@ int mlx5_core_modify_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
 }
 EXPORT_SYMBOL(mlx5_core_modify_cq);
 
+int mlx5_core_modify_cq_moderation(struct mlx5_core_dev *dev,
+				   struct mlx5_core_cq *cq,
+				   u16 cq_period,
+				   u16 cq_max_count)
+{
+	struct mlx5_modify_cq_mbox_in in;
+
+	memset(&in, 0, sizeof(in));
+
+	in.cqn              = cpu_to_be32(cq->cqn);
+	in.ctx.cq_period    = cpu_to_be16(cq_period);
+	in.ctx.cq_max_count = cpu_to_be16(cq_max_count);
+	in.field_select     = cpu_to_be32(MLX5_CQ_MODIFY_PERIOD |
+					  MLX5_CQ_MODIFY_COUNT);
+
+	return mlx5_core_modify_cq(dev, cq, &in, sizeof(in));
+}
+
 int mlx5_init_cq_table(struct mlx5_core_dev *dev)
 {
 	struct mlx5_cq_table *table = &dev->priv.cq_table;

drivers/net/ethernet/mellanox/mlx5/core/en.h

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/if_vlan.h>
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/qp.h>
+#include <linux/mlx5/cq.h>
+#include "vport.h"
+#include "wq.h"
+#include "transobj.h"
+#include "mlx5_core.h"
+
+#define MLX5E_MAX_NUM_TC	8
+
+#define MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE                0x7
+#define MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE                0xa
+#define MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE                0xd
+
+#define MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE                0x7
+#define MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE                0xa
+#define MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE                0xd
+
+#define MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ                 (16 * 1024)
+#define MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC      0x10
+#define MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_PKTS      0x20
+#define MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC      0x10
+#define MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_PKTS      0x20
+#define MLX5E_PARAMS_DEFAULT_MIN_RX_WQES                0x80
+#define MLX5E_PARAMS_DEFAULT_RX_HASH_LOG_TBL_SZ         0x7
+#define MLX5E_PARAMS_MIN_MTU                            46
+
+#define MLX5E_TX_CQ_POLL_BUDGET        128
+#define MLX5E_UPDATE_STATS_INTERVAL    200 /* msecs */
+
+static const char vport_strings[][ETH_GSTRING_LEN] = {
+	/* vport statistics */
+	"rx_packets",
+	"rx_bytes",
+	"tx_packets",
+	"tx_bytes",
+	"rx_error_packets",
+	"rx_error_bytes",
+	"tx_error_packets",
+	"tx_error_bytes",
+	"rx_unicast_packets",
+	"rx_unicast_bytes",
+	"tx_unicast_packets",
+	"tx_unicast_bytes",
+	"rx_multicast_packets",
+	"rx_multicast_bytes",
+	"tx_multicast_packets",
+	"tx_multicast_bytes",
+	"rx_broadcast_packets",
+	"rx_broadcast_bytes",
+	"tx_broadcast_packets",
+	"tx_broadcast_bytes",
+
+	/* SW counters */
+	"tso_packets",
+	"tso_bytes",
+	"lro_packets",
+	"lro_bytes",
+	"rx_csum_good",
+	"rx_csum_none",
+	"tx_csum_offload",
+	"tx_queue_stopped",
+	"tx_queue_wake",
+	"tx_queue_dropped",
+	"rx_wqe_err",
+};
+
+struct mlx5e_vport_stats {
+	/* HW counters */
+	u64 rx_packets;
+	u64 rx_bytes;
+	u64 tx_packets;
+	u64 tx_bytes;
+	u64 rx_error_packets;
+	u64 rx_error_bytes;
+	u64 tx_error_packets;
+	u64 tx_error_bytes;
+	u64 rx_unicast_packets;
+	u64 rx_unicast_bytes;
+	u64 tx_unicast_packets;
+	u64 tx_unicast_bytes;
+	u64 rx_multicast_packets;
+	u64 rx_multicast_bytes;
+	u64 tx_multicast_packets;
+	u64 tx_multicast_bytes;
+	u64 rx_broadcast_packets;
+	u64 rx_broadcast_bytes;
+	u64 tx_broadcast_packets;
+	u64 tx_broadcast_bytes;
+
+	/* SW counters */
+	u64 tso_packets;
+	u64 tso_bytes;
+	u64 lro_packets;
+	u64 lro_bytes;
+	u64 rx_csum_good;
+	u64 rx_csum_none;
+	u64 tx_csum_offload;
+	u64 tx_queue_stopped;
+	u64 tx_queue_wake;
+	u64 tx_queue_dropped;
+	u64 rx_wqe_err;
+
+#define NUM_VPORT_COUNTERS     31
+};
+
+static const char rq_stats_strings[][ETH_GSTRING_LEN] = {
+	"packets",
+	"csum_none",
+	"lro_packets",
+	"lro_bytes",
+	"wqe_err"
+};
+
+struct mlx5e_rq_stats {
+	u64 packets;
+	u64 csum_none;
+	u64 lro_packets;
+	u64 lro_bytes;
+	u64 wqe_err;
+#define NUM_RQ_STATS 5
+};
+
+static const char sq_stats_strings[][ETH_GSTRING_LEN] = {
+	"packets",
+	"tso_packets",
+	"tso_bytes",
+	"csum_offload_none",
+	"stopped",
+	"wake",
+	"dropped",
+	"nop"
+};
+
+struct mlx5e_sq_stats {
+	u64 packets;
+	u64 tso_packets;
+	u64 tso_bytes;
+	u64 csum_offload_none;
+	u64 stopped;
+	u64 wake;
+	u64 dropped;
+	u64 nop;
+#define NUM_SQ_STATS 8
+};
+
+struct mlx5e_stats {
+	struct mlx5e_vport_stats   vport;
+};
+
+struct mlx5e_params {
+	u8  log_sq_size;
+	u8  log_rq_size;
+	u16 num_channels;
+	u8  default_vlan_prio;
+	u8  num_tc;
+	u16 rx_cq_moderation_usec;
+	u16 rx_cq_moderation_pkts;
+	u16 tx_cq_moderation_usec;
+	u16 tx_cq_moderation_pkts;
+	u16 min_rx_wqes;
+	u16 rx_hash_log_tbl_sz;
+	bool lro_en;
+	u32 lro_wqe_sz;
+};
+
+enum {
+	MLX5E_RQ_STATE_POST_WQES_ENABLE,
+};
+
+enum cq_flags {
+	MLX5E_CQ_HAS_CQES = 1,
+};
+
+struct mlx5e_cq {
+	/* data path - accessed per cqe */
+	struct mlx5_cqwq           wq;
+	void                      *sqrq;
+	unsigned long              flags;
+
+	/* data path - accessed per napi poll */
+	struct napi_struct        *napi;
+	struct mlx5_core_cq        mcq;
+	struct mlx5e_channel      *channel;
+
+	/* control */
+	struct mlx5_wq_ctrl        wq_ctrl;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_rq {
+	/* data path */
+	struct mlx5_wq_ll      wq;
+	u32                    wqe_sz;
+	struct sk_buff       **skb;
+
+	struct device         *pdev;
+	struct net_device     *netdev;
+	struct mlx5e_rq_stats  stats;
+	struct mlx5e_cq        cq;
+
+	unsigned long          state;
+	int                    ix;
+
+	/* control */
+	struct mlx5_wq_ctrl    wq_ctrl;
+	u32                    rqn;
+	struct mlx5e_channel  *channel;
+} ____cacheline_aligned_in_smp;
+
+struct mlx5e_tx_skb_cb {
+	u32 num_bytes;
+	u8  num_wqebbs;
+	u8  num_dma;
+};
+
+#define MLX5E_TX_SKB_CB(__skb) ((struct mlx5e_tx_skb_cb *)__skb->cb)
+
+struct mlx5e_sq_dma {
+	dma_addr_t addr;
+	u32        size;
+};
+
+enum {
+	MLX5E_SQ_STATE_WAKE_TXQ_ENABLE,
+};
+
+struct mlx5e_sq {
+	/* data path */
+
+	/* dirtied @completion */
+	u16                        cc;
+	u32                        dma_fifo_cc;
+
+	/* dirtied @xmit */
+	u16                        pc ____cacheline_aligned_in_smp;
+	u32                        dma_fifo_pc;
+	u32                        bf_offset;
+	struct mlx5e_sq_stats      stats;
+
+	struct mlx5e_cq            cq;
+
+	/* pointers to per packet info: write@xmit, read@completion */
+	struct sk_buff           **skb;
+	struct mlx5e_sq_dma       *dma_fifo;
+
+	/* read only */
+	struct mlx5_wq_cyc         wq;
+	u32                        dma_fifo_mask;
+	void __iomem              *uar_map;
+	struct netdev_queue       *txq;
+	u32                        sqn;
+	u32                        bf_buf_size;
+	struct device             *pdev;
+	__be32                     mkey_be;
+	unsigned long              state;
+
+	/* control path */
+	struct mlx5_wq_ctrl        wq_ctrl;
+	struct mlx5_uar            uar;
+	struct mlx5e_channel      *channel;
+	int                        tc;
+} ____cacheline_aligned_in_smp;
+
+static inline bool mlx5e_sq_has_room_for(struct mlx5e_sq *sq, u16 n)
+{
+	return (((sq->wq.sz_m1 & (sq->cc - sq->pc)) >= n) ||
+		(sq->cc  == sq->pc));
+}
+
+enum channel_flags {
+	MLX5E_CHANNEL_NAPI_SCHED = 1,
+};
+
+struct mlx5e_channel {
+	/* data path */
+	struct mlx5e_rq            rq;
+	struct mlx5e_sq            sq[MLX5E_MAX_NUM_TC];
+	struct napi_struct         napi;
+	struct device             *pdev;
+	struct net_device         *netdev;
+	__be32                     mkey_be;
+	u8                         num_tc;
+	unsigned long              flags;
+
+	/* control */
+	struct mlx5e_priv         *priv;
+	int                        ix;
+	int                        cpu;
+};
+
+enum mlx5e_traffic_types {
+	MLX5E_TT_IPV4_TCP = 0,
+	MLX5E_TT_IPV6_TCP = 1,
+	MLX5E_TT_IPV4_UDP = 2,
+	MLX5E_TT_IPV6_UDP = 3,
+	MLX5E_TT_IPV4     = 4,
+	MLX5E_TT_IPV6     = 5,
+	MLX5E_TT_ANY      = 6,
+	MLX5E_NUM_TT      = 7,
+};
+
+enum {
+	MLX5E_RQT_SPREADING  = 0,
+	MLX5E_RQT_DEFAULT_RQ = 1,
+	MLX5E_NUM_RQT        = 2,
+};
+
+struct mlx5e_eth_addr_info {
+	u8  addr[ETH_ALEN + 2];
+	u32 tt_vec;
+	u32 ft_ix[MLX5E_NUM_TT]; /* flow table index per traffic type */
+};
+
+#define MLX5E_ETH_ADDR_HASH_SIZE (1 << BITS_PER_BYTE)
+
+struct mlx5e_eth_addr_db {
+	struct hlist_head          netdev_uc[MLX5E_ETH_ADDR_HASH_SIZE];
+	struct hlist_head          netdev_mc[MLX5E_ETH_ADDR_HASH_SIZE];
+	struct mlx5e_eth_addr_info broadcast;
+	struct mlx5e_eth_addr_info allmulti;
+	struct mlx5e_eth_addr_info promisc;
+	bool                       broadcast_enabled;
+	bool                       allmulti_enabled;
+	bool                       promisc_enabled;
+};
+
+enum {
+	MLX5E_STATE_ASYNC_EVENTS_ENABLE,
+	MLX5E_STATE_OPENED,
+};
+
+struct mlx5e_vlan_db {
+	unsigned long active_vlans[BITS_TO_LONGS(VLAN_N_VID)];
+	u32           active_vlans_ft_ix[VLAN_N_VID];
+	u32           untagged_rule_ft_ix;
+	u32           any_vlan_rule_ft_ix;
+	bool          filter_disabled;
+};
+
+struct mlx5e_flow_table {
+	void *vlan;
+	void *main;
+};
+
+struct mlx5e_priv {
+	/* priv data path fields - start */
+	int                        order_base_2_num_channels;
+	int                        queue_mapping_channel_mask;
+	int                        num_tc;
+	int                        default_vlan_prio;
+	/* priv data path fields - end */
+
+	unsigned long              state;
+	struct mutex               state_lock; /* Protects Interface state */
+	struct mlx5_uar            cq_uar;
+	u32                        pdn;
+	struct mlx5_core_mr        mr;
+
+	struct mlx5e_channel     **channel;
+	u32                        tisn[MLX5E_MAX_NUM_TC];
+	u32                        rqtn;
+	u32                        tirn[MLX5E_NUM_TT];
+
+	struct mlx5e_flow_table    ft;
+	struct mlx5e_eth_addr_db   eth_addr;
+	struct mlx5e_vlan_db       vlan;
+
+	struct mlx5e_params        params;
+	spinlock_t                 async_events_spinlock; /* sync hw events */
+	struct work_struct         update_carrier_work;
+	struct work_struct         set_rx_mode_work;
+	struct delayed_work        update_stats_work;
+
+	struct mlx5_core_dev      *mdev;
+	struct net_device         *netdev;
+	struct mlx5e_stats         stats;
+};
+
+#define MLX5E_NET_IP_ALIGN 2
+
+struct mlx5e_tx_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_eth_seg  eth;
+};
+
+struct mlx5e_rx_wqe {
+	struct mlx5_wqe_srq_next_seg  next;
+	struct mlx5_wqe_data_seg      data;
+};
+
+enum mlx5e_link_mode {
+	MLX5E_1000BASE_CX_SGMII	 = 0,
+	MLX5E_1000BASE_KX	 = 1,
+	MLX5E_10GBASE_CX4	 = 2,
+	MLX5E_10GBASE_KX4	 = 3,
+	MLX5E_10GBASE_KR	 = 4,
+	MLX5E_20GBASE_KR2	 = 5,
+	MLX5E_40GBASE_CR4	 = 6,
+	MLX5E_40GBASE_KR4	 = 7,
+	MLX5E_56GBASE_R4	 = 8,
+	MLX5E_10GBASE_CR	 = 12,
+	MLX5E_10GBASE_SR	 = 13,
+	MLX5E_10GBASE_ER	 = 14,
+	MLX5E_40GBASE_SR4	 = 15,
+	MLX5E_40GBASE_LR4	 = 16,
+	MLX5E_100GBASE_CR4	 = 20,
+	MLX5E_100GBASE_SR4	 = 21,
+	MLX5E_100GBASE_KR4	 = 22,
+	MLX5E_100GBASE_LR4	 = 23,
+	MLX5E_100BASE_TX	 = 24,
+	MLX5E_100BASE_T		 = 25,
+	MLX5E_10GBASE_T		 = 26,
+	MLX5E_25GBASE_CR	 = 27,
+	MLX5E_25GBASE_KR	 = 28,
+	MLX5E_25GBASE_SR	 = 29,
+	MLX5E_50GBASE_CR2	 = 30,
+	MLX5E_50GBASE_KR2	 = 31,
+	MLX5E_LINK_MODES_NUMBER,
+};
+
+#define MLX5E_PROT_MASK(link_mode) (1 << link_mode)
+
+u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
+		       void *accel_priv, select_queue_fallback_t fallback);
+netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev);
+netdev_tx_t mlx5e_xmit_multi_tc(struct sk_buff *skb, struct net_device *dev);
+
+void mlx5e_completion_event(struct mlx5_core_cq *mcq);
+void mlx5e_cq_error_event(struct mlx5_core_cq *mcq, enum mlx5_event event);
+int mlx5e_napi_poll(struct napi_struct *napi, int budget);
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq);
+bool mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget);
+bool mlx5e_post_rx_wqes(struct mlx5e_rq *rq);
+struct mlx5_cqe64 *mlx5e_get_cqe(struct mlx5e_cq *cq);
+
+void mlx5e_update_stats(struct mlx5e_priv *priv);
+
+int mlx5e_open_flow_table(struct mlx5e_priv *priv);
+void mlx5e_close_flow_table(struct mlx5e_priv *priv);
+void mlx5e_init_eth_addr(struct mlx5e_priv *priv);
+void mlx5e_set_rx_mode_core(struct mlx5e_priv *priv);
+void mlx5e_set_rx_mode_work(struct work_struct *work);
+
+int mlx5e_vlan_rx_add_vid(struct net_device *dev, __always_unused __be16 proto,
+			  u16 vid);
+int mlx5e_vlan_rx_kill_vid(struct net_device *dev, __always_unused __be16 proto,
+			   u16 vid);
+void mlx5e_enable_vlan_filter(struct mlx5e_priv *priv);
+void mlx5e_disable_vlan_filter(struct mlx5e_priv *priv);
+int mlx5e_add_all_vlan_rules(struct mlx5e_priv *priv);
+void mlx5e_del_all_vlan_rules(struct mlx5e_priv *priv);
+
+int mlx5e_open_locked(struct net_device *netdev);
+int mlx5e_close_locked(struct net_device *netdev);
+int mlx5e_update_priv_params(struct mlx5e_priv *priv,
+			     struct mlx5e_params *new_params);
+
+static inline void mlx5e_tx_notify_hw(struct mlx5e_sq *sq,
+				      struct mlx5e_tx_wqe *wqe)
+{
+	/* ensure wqe is visible to device before updating doorbell record */
+	dma_wmb();
+
+	*sq->wq.db = cpu_to_be32(sq->pc);
+
+	/* ensure doorbell record is visible to device before ringing the
+	 * doorbell
+	 */
+	wmb();
+
+	mlx5_write64((__be32 *)&wqe->ctrl,
+		     sq->uar_map + MLX5_BF_OFFSET + sq->bf_offset,
+		     NULL);
+
+	sq->bf_offset ^= sq->bf_buf_size;
+}
+
+static inline void mlx5e_cq_arm(struct mlx5e_cq *cq)
+{
+	struct mlx5_core_cq *mcq;
+
+	mcq = &cq->mcq;
+	mlx5_cq_arm(mcq, MLX5_CQ_DB_REQ_NOT, mcq->uar->map, NULL, cq->wq.cc);
+}
+
+extern const struct ethtool_ops mlx5e_ethtool_ops;

drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+
+static void mlx5e_get_drvinfo(struct net_device *dev,
+			      struct ethtool_drvinfo *drvinfo)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	strlcpy(drvinfo->driver, DRIVER_NAME, sizeof(drvinfo->driver));
+	strlcpy(drvinfo->version, DRIVER_VERSION " (" DRIVER_RELDATE ")",
+		sizeof(drvinfo->version));
+	snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+		 "%d.%d.%d",
+		 fw_rev_maj(mdev), fw_rev_min(mdev), fw_rev_sub(mdev));
+	strlcpy(drvinfo->bus_info, pci_name(mdev->pdev),
+		sizeof(drvinfo->bus_info));
+}
+
+static const struct {
+	u32 supported;
+	u32 advertised;
+	u32 speed;
+} ptys2ethtool_table[MLX5E_LINK_MODES_NUMBER] = {
+	[MLX5E_1000BASE_CX_SGMII] = {
+		.supported  = SUPPORTED_1000baseKX_Full,
+		.advertised = ADVERTISED_1000baseKX_Full,
+		.speed      = 1000,
+	},
+	[MLX5E_1000BASE_KX] = {
+		.supported  = SUPPORTED_1000baseKX_Full,
+		.advertised = ADVERTISED_1000baseKX_Full,
+		.speed      = 1000,
+	},
+	[MLX5E_10GBASE_CX4] = {
+		.supported  = SUPPORTED_10000baseKX4_Full,
+		.advertised = ADVERTISED_10000baseKX4_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_10GBASE_KX4] = {
+		.supported  = SUPPORTED_10000baseKX4_Full,
+		.advertised = ADVERTISED_10000baseKX4_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_10GBASE_KR] = {
+		.supported  = SUPPORTED_10000baseKR_Full,
+		.advertised = ADVERTISED_10000baseKR_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_20GBASE_KR2] = {
+		.supported  = SUPPORTED_20000baseKR2_Full,
+		.advertised = ADVERTISED_20000baseKR2_Full,
+		.speed      = 20000,
+	},
+	[MLX5E_40GBASE_CR4] = {
+		.supported  = SUPPORTED_40000baseCR4_Full,
+		.advertised = ADVERTISED_40000baseCR4_Full,
+		.speed      = 40000,
+	},
+	[MLX5E_40GBASE_KR4] = {
+		.supported  = SUPPORTED_40000baseKR4_Full,
+		.advertised = ADVERTISED_40000baseKR4_Full,
+		.speed      = 40000,
+	},
+	[MLX5E_56GBASE_R4] = {
+		.supported  = SUPPORTED_56000baseKR4_Full,
+		.advertised = ADVERTISED_56000baseKR4_Full,
+		.speed      = 56000,
+	},
+	[MLX5E_10GBASE_CR] = {
+		.supported  = SUPPORTED_10000baseKR_Full,
+		.advertised = ADVERTISED_10000baseKR_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_10GBASE_SR] = {
+		.supported  = SUPPORTED_10000baseKR_Full,
+		.advertised = ADVERTISED_10000baseKR_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_10GBASE_ER] = {
+		.supported  = SUPPORTED_10000baseKR_Full,
+		.advertised = ADVERTISED_10000baseKR_Full,
+		.speed      = 10000,
+	},
+	[MLX5E_40GBASE_SR4] = {
+		.supported  = SUPPORTED_40000baseSR4_Full,
+		.advertised = ADVERTISED_40000baseSR4_Full,
+		.speed      = 40000,
+	},
+	[MLX5E_40GBASE_LR4] = {
+		.supported  = SUPPORTED_40000baseLR4_Full,
+		.advertised = ADVERTISED_40000baseLR4_Full,
+		.speed      = 40000,
+	},
+	[MLX5E_100GBASE_CR4] = {
+		.speed      = 100000,
+	},
+	[MLX5E_100GBASE_SR4] = {
+		.speed      = 100000,
+	},
+	[MLX5E_100GBASE_KR4] = {
+		.speed      = 100000,
+	},
+	[MLX5E_100GBASE_LR4] = {
+		.speed      = 100000,
+	},
+	[MLX5E_100BASE_TX]   = {
+		.speed      = 100,
+	},
+	[MLX5E_100BASE_T]    = {
+		.supported  = SUPPORTED_100baseT_Full,
+		.advertised = ADVERTISED_100baseT_Full,
+		.speed      = 100,
+	},
+	[MLX5E_10GBASE_T]    = {
+		.supported  = SUPPORTED_10000baseT_Full,
+		.advertised = ADVERTISED_10000baseT_Full,
+		.speed      = 1000,
+	},
+	[MLX5E_25GBASE_CR]   = {
+		.speed      = 25000,
+	},
+	[MLX5E_25GBASE_KR]   = {
+		.speed      = 25000,
+	},
+	[MLX5E_25GBASE_SR]   = {
+		.speed      = 25000,
+	},
+	[MLX5E_50GBASE_CR2]  = {
+		.speed      = 50000,
+	},
+	[MLX5E_50GBASE_KR2]  = {
+		.speed      = 50000,
+	},
+};
+
+static int mlx5e_get_sset_count(struct net_device *dev, int sset)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (sset) {
+	case ETH_SS_STATS:
+		return NUM_VPORT_COUNTERS +
+		       priv->params.num_channels * NUM_RQ_STATS +
+		       priv->params.num_channels * priv->num_tc *
+						   NUM_SQ_STATS;
+	/* fallthrough */
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void mlx5e_get_strings(struct net_device *dev,
+			      uint32_t stringset, uint8_t *data)
+{
+	int i, j, tc, idx = 0;
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (stringset) {
+	case ETH_SS_PRIV_FLAGS:
+		break;
+
+	case ETH_SS_TEST:
+		break;
+
+	case ETH_SS_STATS:
+		/* VPORT counters */
+		for (i = 0; i < NUM_VPORT_COUNTERS; i++)
+			strcpy(data + (idx++) * ETH_GSTRING_LEN,
+			       vport_strings[i]);
+
+		/* per channel counters */
+		for (i = 0; i < priv->params.num_channels; i++)
+			for (j = 0; j < NUM_RQ_STATS; j++)
+				sprintf(data + (idx++) * ETH_GSTRING_LEN,
+					"rx%d_%s", i, rq_stats_strings[j]);
+
+		for (i = 0; i < priv->params.num_channels; i++)
+			for (tc = 0; tc < priv->num_tc; tc++)
+				for (j = 0; j < NUM_SQ_STATS; j++)
+					sprintf(data +
+						(idx++) * ETH_GSTRING_LEN,
+						"tx%d_%d_%s", i, tc,
+						sq_stats_strings[j]);
+		break;
+	}
+}
+
+static void mlx5e_get_ethtool_stats(struct net_device *dev,
+				    struct ethtool_stats *stats, u64 *data)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int i, j, tc, idx = 0;
+
+	if (!data)
+		return;
+
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		mlx5e_update_stats(priv);
+	mutex_unlock(&priv->state_lock);
+
+	for (i = 0; i < NUM_VPORT_COUNTERS; i++)
+		data[idx++] = ((u64 *)&priv->stats.vport)[i];
+
+	/* per channel counters */
+	for (i = 0; i < priv->params.num_channels; i++)
+		for (j = 0; j < NUM_RQ_STATS; j++)
+			data[idx++] = !test_bit(MLX5E_STATE_OPENED,
+						&priv->state) ? 0 :
+				       ((u64 *)&priv->channel[i]->rq.stats)[j];
+
+	for (i = 0; i < priv->params.num_channels; i++)
+		for (tc = 0; tc < priv->num_tc; tc++)
+			for (j = 0; j < NUM_SQ_STATS; j++)
+				data[idx++] = !test_bit(MLX5E_STATE_OPENED,
+							&priv->state) ? 0 :
+				((u64 *)&priv->channel[i]->sq[tc].stats)[j];
+}
+
+static void mlx5e_get_ringparam(struct net_device *dev,
+				struct ethtool_ringparam *param)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	param->rx_max_pending = 1 << MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE;
+	param->tx_max_pending = 1 << MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE;
+	param->rx_pending     = 1 << priv->params.log_rq_size;
+	param->tx_pending     = 1 << priv->params.log_sq_size;
+}
+
+static int mlx5e_set_ringparam(struct net_device *dev,
+			       struct ethtool_ringparam *param)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_params new_params;
+	u16 min_rx_wqes;
+	u8 log_rq_size;
+	u8 log_sq_size;
+	int err = 0;
+
+	if (param->rx_jumbo_pending) {
+		netdev_info(dev, "%s: rx_jumbo_pending not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+	if (param->rx_mini_pending) {
+		netdev_info(dev, "%s: rx_mini_pending not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+	if (param->rx_pending < (1 << MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE)) {
+		netdev_info(dev, "%s: rx_pending (%d) < min (%d)\n",
+			    __func__, param->rx_pending,
+			    1 << MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE);
+		return -EINVAL;
+	}
+	if (param->rx_pending > (1 << MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE)) {
+		netdev_info(dev, "%s: rx_pending (%d) > max (%d)\n",
+			    __func__, param->rx_pending,
+			    1 << MLX5E_PARAMS_MAXIMUM_LOG_RQ_SIZE);
+		return -EINVAL;
+	}
+	if (param->tx_pending < (1 << MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE)) {
+		netdev_info(dev, "%s: tx_pending (%d) < min (%d)\n",
+			    __func__, param->tx_pending,
+			    1 << MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE);
+		return -EINVAL;
+	}
+	if (param->tx_pending > (1 << MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE)) {
+		netdev_info(dev, "%s: tx_pending (%d) > max (%d)\n",
+			    __func__, param->tx_pending,
+			    1 << MLX5E_PARAMS_MAXIMUM_LOG_SQ_SIZE);
+		return -EINVAL;
+	}
+
+	log_rq_size = order_base_2(param->rx_pending);
+	log_sq_size = order_base_2(param->tx_pending);
+	min_rx_wqes = min_t(u16, param->rx_pending - 1,
+			    MLX5E_PARAMS_DEFAULT_MIN_RX_WQES);
+
+	if (log_rq_size == priv->params.log_rq_size &&
+	    log_sq_size == priv->params.log_sq_size &&
+	    min_rx_wqes == priv->params.min_rx_wqes)
+		return 0;
+
+	mutex_lock(&priv->state_lock);
+	new_params = priv->params;
+	new_params.log_rq_size = log_rq_size;
+	new_params.log_sq_size = log_sq_size;
+	new_params.min_rx_wqes = min_rx_wqes;
+	err = mlx5e_update_priv_params(priv, &new_params);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static void mlx5e_get_channels(struct net_device *dev,
+			       struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int ncv = priv->mdev->priv.eq_table.num_comp_vectors;
+
+	ch->max_combined   = ncv;
+	ch->combined_count = priv->params.num_channels;
+}
+
+static int mlx5e_set_channels(struct net_device *dev,
+			      struct ethtool_channels *ch)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int ncv = priv->mdev->priv.eq_table.num_comp_vectors;
+	unsigned int count = ch->combined_count;
+	struct mlx5e_params new_params;
+	int err = 0;
+
+	if (!count) {
+		netdev_info(dev, "%s: combined_count=0 not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+	if (ch->rx_count || ch->tx_count) {
+		netdev_info(dev, "%s: separate rx/tx count not supported\n",
+			    __func__);
+		return -EINVAL;
+	}
+	if (count > ncv) {
+		netdev_info(dev, "%s: count (%d) > max (%d)\n",
+			    __func__, count, ncv);
+		return -EINVAL;
+	}
+
+	if (priv->params.num_channels == count)
+		return 0;
+
+	mutex_lock(&priv->state_lock);
+	new_params = priv->params;
+	new_params.num_channels = count;
+	err = mlx5e_update_priv_params(priv, &new_params);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static int mlx5e_get_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	coal->rx_coalesce_usecs       = priv->params.rx_cq_moderation_usec;
+	coal->rx_max_coalesced_frames = priv->params.rx_cq_moderation_pkts;
+	coal->tx_coalesce_usecs       = priv->params.tx_cq_moderation_usec;
+	coal->tx_max_coalesced_frames = priv->params.tx_cq_moderation_pkts;
+
+	return 0;
+}
+
+static int mlx5e_set_coalesce(struct net_device *netdev,
+			      struct ethtool_coalesce *coal)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_channel *c;
+	int tc;
+	int i;
+
+	priv->params.tx_cq_moderation_usec = coal->tx_coalesce_usecs;
+	priv->params.tx_cq_moderation_pkts = coal->tx_max_coalesced_frames;
+	priv->params.rx_cq_moderation_usec = coal->rx_coalesce_usecs;
+	priv->params.rx_cq_moderation_pkts = coal->rx_max_coalesced_frames;
+
+	for (i = 0; i < priv->params.num_channels; ++i) {
+		c = priv->channel[i];
+
+		for (tc = 0; tc < c->num_tc; tc++) {
+			mlx5_core_modify_cq_moderation(mdev,
+						&c->sq[tc].cq.mcq,
+						coal->tx_coalesce_usecs,
+						coal->tx_max_coalesced_frames);
+		}
+
+		mlx5_core_modify_cq_moderation(mdev, &c->rq.cq.mcq,
+					       coal->rx_coalesce_usecs,
+					       coal->rx_max_coalesced_frames);
+	}
+
+	return 0;
+}
+
+static u32 ptys2ethtool_supported_link(u32 eth_proto_cap)
+{
+	int i;
+	u32 supported_modes = 0;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (eth_proto_cap & MLX5E_PROT_MASK(i))
+			supported_modes |= ptys2ethtool_table[i].supported;
+	}
+	return supported_modes;
+}
+
+static u32 ptys2ethtool_adver_link(u32 eth_proto_cap)
+{
+	int i;
+	u32 advertising_modes = 0;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (eth_proto_cap & MLX5E_PROT_MASK(i))
+			advertising_modes |= ptys2ethtool_table[i].advertised;
+	}
+	return advertising_modes;
+}
+
+static u32 ptys2ethtool_supported_port(u32 eth_proto_cap)
+{
+	if (eth_proto_cap & (MLX5E_PROT_MASK(MLX5E_10GBASE_CR)
+			   | MLX5E_PROT_MASK(MLX5E_10GBASE_SR)
+			   | MLX5E_PROT_MASK(MLX5E_40GBASE_CR4)
+			   | MLX5E_PROT_MASK(MLX5E_40GBASE_SR4)
+			   | MLX5E_PROT_MASK(MLX5E_100GBASE_SR4)
+			   | MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII))) {
+		return SUPPORTED_FIBRE;
+	}
+
+	if (eth_proto_cap & (MLX5E_PROT_MASK(MLX5E_100GBASE_KR4)
+			   | MLX5E_PROT_MASK(MLX5E_40GBASE_KR4)
+			   | MLX5E_PROT_MASK(MLX5E_10GBASE_KR)
+			   | MLX5E_PROT_MASK(MLX5E_10GBASE_KX4)
+			   | MLX5E_PROT_MASK(MLX5E_1000BASE_KX))) {
+		return SUPPORTED_Backplane;
+	}
+	return 0;
+}
+
+static void get_speed_duplex(struct net_device *netdev,
+			     u32 eth_proto_oper,
+			     struct ethtool_cmd *cmd)
+{
+	int i;
+	u32 speed = SPEED_UNKNOWN;
+	u8 duplex = DUPLEX_UNKNOWN;
+
+	if (!netif_carrier_ok(netdev))
+		goto out;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (eth_proto_oper & MLX5E_PROT_MASK(i)) {
+			speed = ptys2ethtool_table[i].speed;
+			duplex = DUPLEX_FULL;
+			break;
+		}
+	}
+out:
+	ethtool_cmd_speed_set(cmd, speed);
+	cmd->duplex = duplex;
+}
+
+static void get_supported(u32 eth_proto_cap, u32 *supported)
+{
+	*supported |= ptys2ethtool_supported_port(eth_proto_cap);
+	*supported |= ptys2ethtool_supported_link(eth_proto_cap);
+	*supported |= SUPPORTED_Pause | SUPPORTED_Asym_Pause;
+}
+
+static void get_advertising(u32 eth_proto_cap, u8 tx_pause,
+			    u8 rx_pause, u32 *advertising)
+{
+	*advertising |= ptys2ethtool_adver_link(eth_proto_cap);
+	*advertising |= tx_pause ? ADVERTISED_Pause : 0;
+	*advertising |= (tx_pause ^ rx_pause) ? ADVERTISED_Asym_Pause : 0;
+}
+
+static u8 get_connector_port(u32 eth_proto)
+{
+	if (eth_proto & (MLX5E_PROT_MASK(MLX5E_10GBASE_SR)
+			 | MLX5E_PROT_MASK(MLX5E_40GBASE_SR4)
+			 | MLX5E_PROT_MASK(MLX5E_100GBASE_SR4)
+			 | MLX5E_PROT_MASK(MLX5E_1000BASE_CX_SGMII))) {
+			return PORT_FIBRE;
+	}
+
+	if (eth_proto & (MLX5E_PROT_MASK(MLX5E_40GBASE_CR4)
+			 | MLX5E_PROT_MASK(MLX5E_10GBASE_CR)
+			 | MLX5E_PROT_MASK(MLX5E_100GBASE_CR4))) {
+			return PORT_DA;
+	}
+
+	if (eth_proto & (MLX5E_PROT_MASK(MLX5E_10GBASE_KX4)
+			 | MLX5E_PROT_MASK(MLX5E_10GBASE_KR)
+			 | MLX5E_PROT_MASK(MLX5E_40GBASE_KR4)
+			 | MLX5E_PROT_MASK(MLX5E_100GBASE_KR4))) {
+			return PORT_NONE;
+	}
+
+	return PORT_OTHER;
+}
+
+static void get_lp_advertising(u32 eth_proto_lp, u32 *lp_advertising)
+{
+	*lp_advertising = ptys2ethtool_adver_link(eth_proto_lp);
+}
+
+static int mlx5e_get_settings(struct net_device *netdev,
+			      struct ethtool_cmd *cmd)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	u32 eth_proto_cap;
+	u32 eth_proto_admin;
+	u32 eth_proto_lp;
+	u32 eth_proto_oper;
+	int err;
+
+	err = mlx5_query_port_ptys(mdev, out, sizeof(out), MLX5_PTYS_EN);
+
+	if (err) {
+		netdev_err(netdev, "%s: query port ptys failed: %d\n",
+			   __func__, err);
+		goto err_query_ptys;
+	}
+
+	eth_proto_cap   = MLX5_GET(ptys_reg, out, eth_proto_capability);
+	eth_proto_admin = MLX5_GET(ptys_reg, out, eth_proto_admin);
+	eth_proto_oper  = MLX5_GET(ptys_reg, out, eth_proto_oper);
+	eth_proto_lp    = MLX5_GET(ptys_reg, out, eth_proto_lp_advertise);
+
+	cmd->supported   = 0;
+	cmd->advertising = 0;
+
+	get_supported(eth_proto_cap, &cmd->supported);
+	get_advertising(eth_proto_admin, 0, 0, &cmd->advertising);
+	get_speed_duplex(netdev, eth_proto_oper, cmd);
+
+	eth_proto_oper = eth_proto_oper ? eth_proto_oper : eth_proto_cap;
+
+	cmd->port = get_connector_port(eth_proto_oper);
+	get_lp_advertising(eth_proto_lp, &cmd->lp_advertising);
+
+	cmd->transceiver = XCVR_INTERNAL;
+
+err_query_ptys:
+	return err;
+}
+
+static u32 mlx5e_ethtool2ptys_adver_link(u32 link_modes)
+{
+	u32 i, ptys_modes = 0;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (ptys2ethtool_table[i].advertised & link_modes)
+			ptys_modes |= MLX5E_PROT_MASK(i);
+	}
+
+	return ptys_modes;
+}
+
+static u32 mlx5e_ethtool2ptys_speed_link(u32 speed)
+{
+	u32 i, speed_links = 0;
+
+	for (i = 0; i < MLX5E_LINK_MODES_NUMBER; ++i) {
+		if (ptys2ethtool_table[i].speed == speed)
+			speed_links |= MLX5E_PROT_MASK(i);
+	}
+
+	return speed_links;
+}
+
+static int mlx5e_set_settings(struct net_device *netdev,
+			      struct ethtool_cmd *cmd)
+{
+	struct mlx5e_priv *priv    = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 link_modes;
+	u32 speed;
+	u32 eth_proto_cap, eth_proto_admin;
+	u8 port_status;
+	int err;
+
+	speed = ethtool_cmd_speed(cmd);
+
+	link_modes = cmd->autoneg == AUTONEG_ENABLE ?
+		mlx5e_ethtool2ptys_adver_link(cmd->advertising) :
+		mlx5e_ethtool2ptys_speed_link(speed);
+
+	err = mlx5_query_port_proto_cap(mdev, &eth_proto_cap, MLX5_PTYS_EN);
+	if (err) {
+		netdev_err(netdev, "%s: query port eth proto cap failed: %d\n",
+			   __func__, err);
+		goto out;
+	}
+
+	link_modes = link_modes & eth_proto_cap;
+	if (!link_modes) {
+		netdev_err(netdev, "%s: Not supported link mode(s) requested",
+			   __func__);
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = mlx5_query_port_proto_admin(mdev, &eth_proto_admin, MLX5_PTYS_EN);
+	if (err) {
+		netdev_err(netdev, "%s: query port eth proto admin failed: %d\n",
+			   __func__, err);
+		goto out;
+	}
+
+	if (link_modes == eth_proto_admin)
+		goto out;
+
+	err = mlx5_set_port_proto(mdev, link_modes, MLX5_PTYS_EN);
+	if (err) {
+		netdev_err(netdev, "%s: set port eth proto admin failed: %d\n",
+			   __func__, err);
+		goto out;
+	}
+
+	err = mlx5_query_port_status(mdev, &port_status);
+	if (err)
+		goto out;
+
+	if (port_status == MLX5_PORT_DOWN)
+		return 0;
+
+	err = mlx5_set_port_status(mdev, MLX5_PORT_DOWN);
+	if (err)
+		goto out;
+	err = mlx5_set_port_status(mdev, MLX5_PORT_UP);
+out:
+	return err;
+}
+
+const struct ethtool_ops mlx5e_ethtool_ops = {
+	.get_drvinfo       = mlx5e_get_drvinfo,
+	.get_link          = ethtool_op_get_link,
+	.get_strings       = mlx5e_get_strings,
+	.get_sset_count    = mlx5e_get_sset_count,
+	.get_ethtool_stats = mlx5e_get_ethtool_stats,
+	.get_ringparam     = mlx5e_get_ringparam,
+	.set_ringparam     = mlx5e_set_ringparam,
+	.get_channels      = mlx5e_get_channels,
+	.set_channels      = mlx5e_set_channels,
+	.get_coalesce      = mlx5e_get_coalesce,
+	.set_coalesce      = mlx5e_set_coalesce,
+	.get_settings      = mlx5e_get_settings,
+	.set_settings      = mlx5e_set_settings,
+};

drivers/net/ethernet/mellanox/mlx5/core/en_flow_table.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/list.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/mlx5/flow_table.h>
+#include "en.h"
+
+enum {
+	MLX5E_FULLMATCH = 0,
+	MLX5E_ALLMULTI  = 1,
+	MLX5E_PROMISC   = 2,
+};
+
+enum {
+	MLX5E_UC        = 0,
+	MLX5E_MC_IPV4   = 1,
+	MLX5E_MC_IPV6   = 2,
+	MLX5E_MC_OTHER  = 3,
+};
+
+enum {
+	MLX5E_ACTION_NONE = 0,
+	MLX5E_ACTION_ADD  = 1,
+	MLX5E_ACTION_DEL  = 2,
+};
+
+struct mlx5e_eth_addr_hash_node {
+	struct hlist_node          hlist;
+	u8                         action;
+	struct mlx5e_eth_addr_info ai;
+};
+
+static inline int mlx5e_hash_eth_addr(u8 *addr)
+{
+	return addr[5];
+}
+
+static void mlx5e_add_eth_addr_to_hash(struct hlist_head *hash, u8 *addr)
+{
+	struct mlx5e_eth_addr_hash_node *hn;
+	int ix = mlx5e_hash_eth_addr(addr);
+	int found = 0;
+
+	hlist_for_each_entry(hn, &hash[ix], hlist)
+		if (ether_addr_equal_64bits(hn->ai.addr, addr)) {
+			found = 1;
+			break;
+		}
+
+	if (found) {
+		hn->action = MLX5E_ACTION_NONE;
+		return;
+	}
+
+	hn = kzalloc(sizeof(*hn), GFP_ATOMIC);
+	if (!hn)
+		return;
+
+	ether_addr_copy(hn->ai.addr, addr);
+	hn->action = MLX5E_ACTION_ADD;
+
+	hlist_add_head(&hn->hlist, &hash[ix]);
+}
+
+static void mlx5e_del_eth_addr_from_hash(struct mlx5e_eth_addr_hash_node *hn)
+{
+	hlist_del(&hn->hlist);
+	kfree(hn);
+}
+
+static void mlx5e_del_eth_addr_from_flow_table(struct mlx5e_priv *priv,
+					       struct mlx5e_eth_addr_info *ai)
+{
+	void *ft = priv->ft.main;
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV6_TCP))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV6_TCP]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV4_TCP))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV4_TCP]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV6_UDP))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV6_UDP]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV4_UDP))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV4_UDP]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV6))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV6]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_IPV4))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_IPV4]);
+
+	if (ai->tt_vec & (1 << MLX5E_TT_ANY))
+		mlx5_del_flow_table_entry(ft, ai->ft_ix[MLX5E_TT_ANY]);
+}
+
+static int mlx5e_get_eth_addr_type(u8 *addr)
+{
+	if (is_unicast_ether_addr(addr))
+		return MLX5E_UC;
+
+	if ((addr[0] == 0x01) &&
+	    (addr[1] == 0x00) &&
+	    (addr[2] == 0x5e) &&
+	   !(addr[3] &  0x80))
+		return MLX5E_MC_IPV4;
+
+	if ((addr[0] == 0x33) &&
+	    (addr[1] == 0x33))
+		return MLX5E_MC_IPV6;
+
+	return MLX5E_MC_OTHER;
+}
+
+static u32 mlx5e_get_tt_vec(struct mlx5e_eth_addr_info *ai, int type)
+{
+	int eth_addr_type;
+	u32 ret;
+
+	switch (type) {
+	case MLX5E_FULLMATCH:
+		eth_addr_type = mlx5e_get_eth_addr_type(ai->addr);
+		switch (eth_addr_type) {
+		case MLX5E_UC:
+			ret =
+				(1 << MLX5E_TT_IPV4_TCP) |
+				(1 << MLX5E_TT_IPV6_TCP) |
+				(1 << MLX5E_TT_IPV4_UDP) |
+				(1 << MLX5E_TT_IPV6_UDP) |
+				(1 << MLX5E_TT_IPV4)     |
+				(1 << MLX5E_TT_IPV6)     |
+				(1 << MLX5E_TT_ANY)      |
+				0;
+			break;
+
+		case MLX5E_MC_IPV4:
+			ret =
+				(1 << MLX5E_TT_IPV4_UDP) |
+				(1 << MLX5E_TT_IPV4)     |
+				0;
+			break;
+
+		case MLX5E_MC_IPV6:
+			ret =
+				(1 << MLX5E_TT_IPV6_UDP) |
+				(1 << MLX5E_TT_IPV6)     |
+				0;
+			break;
+
+		case MLX5E_MC_OTHER:
+			ret =
+				(1 << MLX5E_TT_ANY)      |
+				0;
+			break;
+		}
+
+		break;
+
+	case MLX5E_ALLMULTI:
+		ret =
+			(1 << MLX5E_TT_IPV4_UDP) |
+			(1 << MLX5E_TT_IPV6_UDP) |
+			(1 << MLX5E_TT_IPV4)     |
+			(1 << MLX5E_TT_IPV6)     |
+			(1 << MLX5E_TT_ANY)      |
+			0;
+		break;
+
+	default: /* MLX5E_PROMISC */
+		ret =
+			(1 << MLX5E_TT_IPV4_TCP) |
+			(1 << MLX5E_TT_IPV6_TCP) |
+			(1 << MLX5E_TT_IPV4_UDP) |
+			(1 << MLX5E_TT_IPV6_UDP) |
+			(1 << MLX5E_TT_IPV4)     |
+			(1 << MLX5E_TT_IPV6)     |
+			(1 << MLX5E_TT_ANY)      |
+			0;
+		break;
+	}
+
+	return ret;
+}
+
+static int __mlx5e_add_eth_addr_rule(struct mlx5e_priv *priv,
+				     struct mlx5e_eth_addr_info *ai, int type,
+				     void *flow_context, void *match_criteria)
+{
+	u8 match_criteria_enable = 0;
+	void *match_value;
+	void *dest;
+	u8   *dmac;
+	u8   *match_criteria_dmac;
+	void *ft   = priv->ft.main;
+	u32  *tirn = priv->tirn;
+	u32  tt_vec;
+	int  err;
+
+	match_value = MLX5_ADDR_OF(flow_context, flow_context, match_value);
+	dmac = MLX5_ADDR_OF(fte_match_param, match_value,
+			    outer_headers.dmac_47_16);
+	match_criteria_dmac = MLX5_ADDR_OF(fte_match_param, match_criteria,
+					   outer_headers.dmac_47_16);
+	dest = MLX5_ADDR_OF(flow_context, flow_context, destination);
+
+	MLX5_SET(flow_context, flow_context, action,
+		 MLX5_FLOW_CONTEXT_ACTION_FWD_DEST);
+	MLX5_SET(flow_context, flow_context, destination_list_size, 1);
+	MLX5_SET(dest_format_struct, dest, destination_type,
+		 MLX5_FLOW_CONTEXT_DEST_TYPE_TIR);
+
+	switch (type) {
+	case MLX5E_FULLMATCH:
+		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		memset(match_criteria_dmac, 0xff, ETH_ALEN);
+		ether_addr_copy(dmac, ai->addr);
+		break;
+
+	case MLX5E_ALLMULTI:
+		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		match_criteria_dmac[0] = 0x01;
+		dmac[0] = 0x01;
+		break;
+
+	case MLX5E_PROMISC:
+		break;
+	}
+
+	tt_vec = mlx5e_get_tt_vec(ai, type);
+
+	if (tt_vec & (1 << MLX5E_TT_ANY)) {
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_ANY]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_ANY]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_ANY);
+	}
+
+	match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 outer_headers.ethertype);
+
+	if (tt_vec & (1 << MLX5E_TT_IPV4)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IP);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV4]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV4]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV4);
+	}
+
+	if (tt_vec & (1 << MLX5E_TT_IPV6)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IPV6);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV6]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV6]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV6);
+	}
+
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 outer_headers.ip_protocol);
+	MLX5_SET(fte_match_param, match_value, outer_headers.ip_protocol,
+		 IPPROTO_UDP);
+
+	if (tt_vec & (1 << MLX5E_TT_IPV4_UDP)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IP);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV4_UDP]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV4_UDP]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV4_UDP);
+	}
+
+	if (tt_vec & (1 << MLX5E_TT_IPV6_UDP)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IPV6);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV6_UDP]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV6_UDP]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV6_UDP);
+	}
+
+	MLX5_SET(fte_match_param, match_value, outer_headers.ip_protocol,
+		 IPPROTO_TCP);
+
+	if (tt_vec & (1 << MLX5E_TT_IPV4_TCP)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IP);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV4_TCP]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV4_TCP]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV4_TCP);
+	}
+
+	if (tt_vec & (1 << MLX5E_TT_IPV6_TCP)) {
+		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+			 ETH_P_IPV6);
+		MLX5_SET(dest_format_struct, dest, destination_id,
+			 tirn[MLX5E_TT_IPV6_TCP]);
+		err = mlx5_add_flow_table_entry(ft, match_criteria_enable,
+						match_criteria, flow_context,
+						&ai->ft_ix[MLX5E_TT_IPV6_TCP]);
+		if (err) {
+			mlx5e_del_eth_addr_from_flow_table(priv, ai);
+			return err;
+		}
+		ai->tt_vec |= (1 << MLX5E_TT_IPV6_TCP);
+	}
+
+	return 0;
+}
+
+static int mlx5e_add_eth_addr_rule(struct mlx5e_priv *priv,
+				   struct mlx5e_eth_addr_info *ai, int type)
+{
+	u32 *flow_context;
+	u32 *match_criteria;
+	int err;
+
+	flow_context   = mlx5_vzalloc(MLX5_ST_SZ_BYTES(flow_context) +
+				      MLX5_ST_SZ_BYTES(dest_format_struct));
+	match_criteria = mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
+	if (!flow_context || !match_criteria) {
+		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
+		err = -ENOMEM;
+		goto add_eth_addr_rule_out;
+	}
+
+	err = __mlx5e_add_eth_addr_rule(priv, ai, type, flow_context,
+					match_criteria);
+	if (err)
+		netdev_err(priv->netdev, "%s: failed\n", __func__);
+
+add_eth_addr_rule_out:
+	kvfree(match_criteria);
+	kvfree(flow_context);
+	return err;
+}
+
+enum mlx5e_vlan_rule_type {
+	MLX5E_VLAN_RULE_TYPE_UNTAGGED,
+	MLX5E_VLAN_RULE_TYPE_ANY_VID,
+	MLX5E_VLAN_RULE_TYPE_MATCH_VID,
+};
+
+static int mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
+			       enum mlx5e_vlan_rule_type rule_type, u16 vid)
+{
+	u8 match_criteria_enable = 0;
+	u32 *flow_context;
+	void *match_value;
+	void *dest;
+	u32 *match_criteria;
+	u32 *ft_ix;
+	int err;
+
+	flow_context   = mlx5_vzalloc(MLX5_ST_SZ_BYTES(flow_context) +
+				      MLX5_ST_SZ_BYTES(dest_format_struct));
+	match_criteria = mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
+	if (!flow_context || !match_criteria) {
+		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
+		err = -ENOMEM;
+		goto add_vlan_rule_out;
+	}
+	match_value = MLX5_ADDR_OF(flow_context, flow_context, match_value);
+	dest = MLX5_ADDR_OF(flow_context, flow_context, destination);
+
+	MLX5_SET(flow_context, flow_context, action,
+		 MLX5_FLOW_CONTEXT_ACTION_FWD_DEST);
+	MLX5_SET(flow_context, flow_context, destination_list_size, 1);
+	MLX5_SET(dest_format_struct, dest, destination_type,
+		 MLX5_FLOW_CONTEXT_DEST_TYPE_FLOW_TABLE);
+	MLX5_SET(dest_format_struct, dest, destination_id,
+		 mlx5_get_flow_table_id(priv->ft.main));
+
+	match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+			 outer_headers.vlan_tag);
+
+	switch (rule_type) {
+	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
+		ft_ix = &priv->vlan.untagged_rule_ft_ix;
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_VID:
+		ft_ix = &priv->vlan.any_vlan_rule_ft_ix;
+		MLX5_SET(fte_match_param, match_value, outer_headers.vlan_tag,
+			 1);
+		break;
+	default: /* MLX5E_VLAN_RULE_TYPE_MATCH_VID */
+		ft_ix = &priv->vlan.active_vlans_ft_ix[vid];
+		MLX5_SET(fte_match_param, match_value, outer_headers.vlan_tag,
+			 1);
+		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, match_value, outer_headers.first_vid,
+			 vid);
+		break;
+	}
+
+	err = mlx5_add_flow_table_entry(priv->ft.vlan, match_criteria_enable,
+					match_criteria, flow_context, ft_ix);
+	if (err)
+		netdev_err(priv->netdev, "%s: failed\n", __func__);
+
+add_vlan_rule_out:
+	kvfree(match_criteria);
+	kvfree(flow_context);
+	return err;
+}
+
+static void mlx5e_del_vlan_rule(struct mlx5e_priv *priv,
+				enum mlx5e_vlan_rule_type rule_type, u16 vid)
+{
+	switch (rule_type) {
+	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
+		mlx5_del_flow_table_entry(priv->ft.vlan,
+					  priv->vlan.untagged_rule_ft_ix);
+		break;
+	case MLX5E_VLAN_RULE_TYPE_ANY_VID:
+		mlx5_del_flow_table_entry(priv->ft.vlan,
+					  priv->vlan.any_vlan_rule_ft_ix);
+		break;
+	case MLX5E_VLAN_RULE_TYPE_MATCH_VID:
+		mlx5_del_flow_table_entry(priv->ft.vlan,
+					  priv->vlan.active_vlans_ft_ix[vid]);
+		break;
+	}
+}
+
+void mlx5e_enable_vlan_filter(struct mlx5e_priv *priv)
+{
+	WARN_ON(!mutex_is_locked(&priv->state_lock));
+
+	if (priv->vlan.filter_disabled) {
+		priv->vlan.filter_disabled = false;
+		if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+			mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_VID,
+					    0);
+	}
+}
+
+void mlx5e_disable_vlan_filter(struct mlx5e_priv *priv)
+{
+	WARN_ON(!mutex_is_locked(&priv->state_lock));
+
+	if (!priv->vlan.filter_disabled) {
+		priv->vlan.filter_disabled = true;
+		if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+			mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_VID,
+					    0);
+	}
+}
+
+int mlx5e_vlan_rx_add_vid(struct net_device *dev, __always_unused __be16 proto,
+			  u16 vid)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int err = 0;
+
+	mutex_lock(&priv->state_lock);
+
+	set_bit(vid, priv->vlan.active_vlans);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_VID,
+					  vid);
+
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_vlan_rx_kill_vid(struct net_device *dev, __always_unused __be16 proto,
+			   u16 vid)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	mutex_lock(&priv->state_lock);
+
+	clear_bit(vid, priv->vlan.active_vlans);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_VID, vid);
+
+	mutex_unlock(&priv->state_lock);
+
+	return 0;
+}
+
+int mlx5e_add_all_vlan_rules(struct mlx5e_priv *priv)
+{
+	u16 vid;
+	int err;
+
+	for_each_set_bit(vid, priv->vlan.active_vlans, VLAN_N_VID) {
+		err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_VID,
+					  vid);
+		if (err)
+			return err;
+	}
+
+	err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
+	if (err)
+		return err;
+
+	if (priv->vlan.filter_disabled) {
+		err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_VID,
+					  0);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+void mlx5e_del_all_vlan_rules(struct mlx5e_priv *priv)
+{
+	u16 vid;
+
+	if (priv->vlan.filter_disabled)
+		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_VID, 0);
+
+	mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
+
+	for_each_set_bit(vid, priv->vlan.active_vlans, VLAN_N_VID)
+		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_VID, vid);
+}
+
+#define mlx5e_for_each_hash_node(hn, tmp, hash, i) \
+	for (i = 0; i < MLX5E_ETH_ADDR_HASH_SIZE; i++) \
+		hlist_for_each_entry_safe(hn, tmp, &hash[i], hlist)
+
+static void mlx5e_execute_action(struct mlx5e_priv *priv,
+				 struct mlx5e_eth_addr_hash_node *hn)
+{
+	switch (hn->action) {
+	case MLX5E_ACTION_ADD:
+		mlx5e_add_eth_addr_rule(priv, &hn->ai, MLX5E_FULLMATCH);
+		hn->action = MLX5E_ACTION_NONE;
+		break;
+
+	case MLX5E_ACTION_DEL:
+		mlx5e_del_eth_addr_from_flow_table(priv, &hn->ai);
+		mlx5e_del_eth_addr_from_hash(hn);
+		break;
+	}
+}
+
+static void mlx5e_sync_netdev_addr(struct mlx5e_priv *priv)
+{
+	struct net_device *netdev = priv->netdev;
+	struct netdev_hw_addr *ha;
+
+	netif_addr_lock_bh(netdev);
+
+	mlx5e_add_eth_addr_to_hash(priv->eth_addr.netdev_uc,
+				   priv->netdev->dev_addr);
+
+	netdev_for_each_uc_addr(ha, netdev)
+		mlx5e_add_eth_addr_to_hash(priv->eth_addr.netdev_uc, ha->addr);
+
+	netdev_for_each_mc_addr(ha, netdev)
+		mlx5e_add_eth_addr_to_hash(priv->eth_addr.netdev_mc, ha->addr);
+
+	netif_addr_unlock_bh(netdev);
+}
+
+static void mlx5e_apply_netdev_addr(struct mlx5e_priv *priv)
+{
+	struct mlx5e_eth_addr_hash_node *hn;
+	struct hlist_node *tmp;
+	int i;
+
+	mlx5e_for_each_hash_node(hn, tmp, priv->eth_addr.netdev_uc, i)
+		mlx5e_execute_action(priv, hn);
+
+	mlx5e_for_each_hash_node(hn, tmp, priv->eth_addr.netdev_mc, i)
+		mlx5e_execute_action(priv, hn);
+}
+
+static void mlx5e_handle_netdev_addr(struct mlx5e_priv *priv)
+{
+	struct mlx5e_eth_addr_hash_node *hn;
+	struct hlist_node *tmp;
+	int i;
+
+	mlx5e_for_each_hash_node(hn, tmp, priv->eth_addr.netdev_uc, i)
+		hn->action = MLX5E_ACTION_DEL;
+	mlx5e_for_each_hash_node(hn, tmp, priv->eth_addr.netdev_mc, i)
+		hn->action = MLX5E_ACTION_DEL;
+
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		mlx5e_sync_netdev_addr(priv);
+
+	mlx5e_apply_netdev_addr(priv);
+}
+
+void mlx5e_set_rx_mode_core(struct mlx5e_priv *priv)
+{
+	struct mlx5e_eth_addr_db *ea = &priv->eth_addr;
+	struct net_device *ndev = priv->netdev;
+
+	bool rx_mode_enable   = test_bit(MLX5E_STATE_OPENED, &priv->state);
+	bool promisc_enabled   = rx_mode_enable && (ndev->flags & IFF_PROMISC);
+	bool allmulti_enabled  = rx_mode_enable && (ndev->flags & IFF_ALLMULTI);
+	bool broadcast_enabled = rx_mode_enable;
+
+	bool enable_promisc    = !ea->promisc_enabled   &&  promisc_enabled;
+	bool disable_promisc   =  ea->promisc_enabled   && !promisc_enabled;
+	bool enable_allmulti   = !ea->allmulti_enabled  &&  allmulti_enabled;
+	bool disable_allmulti  =  ea->allmulti_enabled  && !allmulti_enabled;
+	bool enable_broadcast  = !ea->broadcast_enabled &&  broadcast_enabled;
+	bool disable_broadcast =  ea->broadcast_enabled && !broadcast_enabled;
+
+	if (enable_promisc)
+		mlx5e_add_eth_addr_rule(priv, &ea->promisc, MLX5E_PROMISC);
+	if (enable_allmulti)
+		mlx5e_add_eth_addr_rule(priv, &ea->allmulti, MLX5E_ALLMULTI);
+	if (enable_broadcast)
+		mlx5e_add_eth_addr_rule(priv, &ea->broadcast, MLX5E_FULLMATCH);
+
+	mlx5e_handle_netdev_addr(priv);
+
+	if (disable_broadcast)
+		mlx5e_del_eth_addr_from_flow_table(priv, &ea->broadcast);
+	if (disable_allmulti)
+		mlx5e_del_eth_addr_from_flow_table(priv, &ea->allmulti);
+	if (disable_promisc)
+		mlx5e_del_eth_addr_from_flow_table(priv, &ea->promisc);
+
+	ea->promisc_enabled   = promisc_enabled;
+	ea->allmulti_enabled  = allmulti_enabled;
+	ea->broadcast_enabled = broadcast_enabled;
+}
+
+void mlx5e_set_rx_mode_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       set_rx_mode_work);
+
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		mlx5e_set_rx_mode_core(priv);
+	mutex_unlock(&priv->state_lock);
+}
+
+void mlx5e_init_eth_addr(struct mlx5e_priv *priv)
+{
+	ether_addr_copy(priv->eth_addr.broadcast.addr, priv->netdev->broadcast);
+}
+
+static int mlx5e_create_main_flow_table(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_table_group *g;
+	u8 *dmac;
+
+	g = kcalloc(9, sizeof(*g), GFP_KERNEL);
+
+	g[0].log_sz = 2;
+	g[0].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, g[0].match_criteria,
+			 outer_headers.ethertype);
+	MLX5_SET_TO_ONES(fte_match_param, g[0].match_criteria,
+			 outer_headers.ip_protocol);
+
+	g[1].log_sz = 1;
+	g[1].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, g[1].match_criteria,
+			 outer_headers.ethertype);
+
+	g[2].log_sz = 0;
+
+	g[3].log_sz = 14;
+	g[3].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[3].match_criteria,
+			    outer_headers.dmac_47_16);
+	memset(dmac, 0xff, ETH_ALEN);
+	MLX5_SET_TO_ONES(fte_match_param, g[3].match_criteria,
+			 outer_headers.ethertype);
+	MLX5_SET_TO_ONES(fte_match_param, g[3].match_criteria,
+			 outer_headers.ip_protocol);
+
+	g[4].log_sz = 13;
+	g[4].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[4].match_criteria,
+			    outer_headers.dmac_47_16);
+	memset(dmac, 0xff, ETH_ALEN);
+	MLX5_SET_TO_ONES(fte_match_param, g[4].match_criteria,
+			 outer_headers.ethertype);
+
+	g[5].log_sz = 11;
+	g[5].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[5].match_criteria,
+			    outer_headers.dmac_47_16);
+	memset(dmac, 0xff, ETH_ALEN);
+
+	g[6].log_sz = 2;
+	g[6].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[6].match_criteria,
+			    outer_headers.dmac_47_16);
+	dmac[0] = 0x01;
+	MLX5_SET_TO_ONES(fte_match_param, g[6].match_criteria,
+			 outer_headers.ethertype);
+	MLX5_SET_TO_ONES(fte_match_param, g[6].match_criteria,
+			 outer_headers.ip_protocol);
+
+	g[7].log_sz = 1;
+	g[7].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[7].match_criteria,
+			    outer_headers.dmac_47_16);
+	dmac[0] = 0x01;
+	MLX5_SET_TO_ONES(fte_match_param, g[7].match_criteria,
+			 outer_headers.ethertype);
+
+	g[8].log_sz = 0;
+	g[8].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	dmac = MLX5_ADDR_OF(fte_match_param, g[8].match_criteria,
+			    outer_headers.dmac_47_16);
+	dmac[0] = 0x01;
+	priv->ft.main = mlx5_create_flow_table(priv->mdev, 1,
+					       MLX5_FLOW_TABLE_TYPE_NIC_RCV,
+					       9, g);
+	kfree(g);
+
+	return priv->ft.main ? 0 : -ENOMEM;
+}
+
+static void mlx5e_destroy_main_flow_table(struct mlx5e_priv *priv)
+{
+	mlx5_destroy_flow_table(priv->ft.main);
+}
+
+static int mlx5e_create_vlan_flow_table(struct mlx5e_priv *priv)
+{
+	struct mlx5_flow_table_group *g;
+
+	g = kcalloc(2, sizeof(*g), GFP_KERNEL);
+	if (!g)
+		return -ENOMEM;
+
+	g[0].log_sz = 12;
+	g[0].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, g[0].match_criteria,
+			 outer_headers.vlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, g[0].match_criteria,
+			 outer_headers.first_vid);
+
+	/* untagged + any vlan id */
+	g[1].log_sz = 1;
+	g[1].match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, g[1].match_criteria,
+			 outer_headers.vlan_tag);
+
+	priv->ft.vlan = mlx5_create_flow_table(priv->mdev, 0,
+					       MLX5_FLOW_TABLE_TYPE_NIC_RCV,
+					       2, g);
+
+	kfree(g);
+	return priv->ft.vlan ? 0 : -ENOMEM;
+}
+
+static void mlx5e_destroy_vlan_flow_table(struct mlx5e_priv *priv)
+{
+	mlx5_destroy_flow_table(priv->ft.vlan);
+}
+
+int mlx5e_open_flow_table(struct mlx5e_priv *priv)
+{
+	int err;
+
+	err = mlx5e_create_main_flow_table(priv);
+	if (err)
+		return err;
+
+	err = mlx5e_create_vlan_flow_table(priv);
+	if (err)
+		goto err_destroy_main_flow_table;
+
+	return 0;
+
+err_destroy_main_flow_table:
+	mlx5e_destroy_main_flow_table(priv);
+
+	return err;
+}
+
+void mlx5e_close_flow_table(struct mlx5e_priv *priv)
+{
+	mlx5e_destroy_vlan_flow_table(priv);
+	mlx5e_destroy_main_flow_table(priv);
+}

drivers/net/ethernet/mellanox/mlx5/core/en_main.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/flow_table.h>
+#include "en.h"
+
+struct mlx5e_rq_param {
+	u32                        rqc[MLX5_ST_SZ_DW(rqc)];
+	struct mlx5_wq_param       wq;
+};
+
+struct mlx5e_sq_param {
+	u32                        sqc[MLX5_ST_SZ_DW(sqc)];
+	struct mlx5_wq_param       wq;
+};
+
+struct mlx5e_cq_param {
+	u32                        cqc[MLX5_ST_SZ_DW(cqc)];
+	struct mlx5_wq_param       wq;
+	u16                        eq_ix;
+};
+
+struct mlx5e_channel_param {
+	struct mlx5e_rq_param      rq;
+	struct mlx5e_sq_param      sq;
+	struct mlx5e_cq_param      rx_cq;
+	struct mlx5e_cq_param      tx_cq;
+};
+
+static void mlx5e_update_carrier(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u8 port_state;
+
+	port_state = mlx5_query_vport_state(mdev,
+		MLX5_QUERY_VPORT_STATE_IN_OP_MOD_VNIC_VPORT);
+
+	if (port_state == VPORT_STATE_UP)
+		netif_carrier_on(priv->netdev);
+	else
+		netif_carrier_off(priv->netdev);
+}
+
+static void mlx5e_update_carrier_work(struct work_struct *work)
+{
+	struct mlx5e_priv *priv = container_of(work, struct mlx5e_priv,
+					       update_carrier_work);
+
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state))
+		mlx5e_update_carrier(priv);
+	mutex_unlock(&priv->state_lock);
+}
+
+void mlx5e_update_stats(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5e_vport_stats *s = &priv->stats.vport;
+	struct mlx5e_rq_stats *rq_stats;
+	struct mlx5e_sq_stats *sq_stats;
+	u32 in[MLX5_ST_SZ_DW(query_vport_counter_in)];
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_vport_counter_out);
+	u64 tx_offload_none;
+	int i, j;
+
+	out = mlx5_vzalloc(outlen);
+	if (!out)
+		return;
+
+	/* Collect firts the SW counters and then HW for consistency */
+	s->tso_packets		= 0;
+	s->tso_bytes		= 0;
+	s->tx_queue_stopped	= 0;
+	s->tx_queue_wake	= 0;
+	s->tx_queue_dropped	= 0;
+	tx_offload_none		= 0;
+	s->lro_packets		= 0;
+	s->lro_bytes		= 0;
+	s->rx_csum_none		= 0;
+	s->rx_wqe_err		= 0;
+	for (i = 0; i < priv->params.num_channels; i++) {
+		rq_stats = &priv->channel[i]->rq.stats;
+
+		s->lro_packets	+= rq_stats->lro_packets;
+		s->lro_bytes	+= rq_stats->lro_bytes;
+		s->rx_csum_none	+= rq_stats->csum_none;
+		s->rx_wqe_err   += rq_stats->wqe_err;
+
+		for (j = 0; j < priv->num_tc; j++) {
+			sq_stats = &priv->channel[i]->sq[j].stats;
+
+			s->tso_packets		+= sq_stats->tso_packets;
+			s->tso_bytes		+= sq_stats->tso_bytes;
+			s->tx_queue_stopped	+= sq_stats->stopped;
+			s->tx_queue_wake	+= sq_stats->wake;
+			s->tx_queue_dropped	+= sq_stats->dropped;
+			tx_offload_none		+= sq_stats->csum_offload_none;
+		}
+	}
+
+	/* HW counters */
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(query_vport_counter_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VPORT_COUNTER);
+	MLX5_SET(query_vport_counter_in, in, op_mod, 0);
+	MLX5_SET(query_vport_counter_in, in, other_vport, 0);
+
+	memset(out, 0, outlen);
+
+	if (mlx5_cmd_exec(mdev, in, sizeof(in), out, outlen))
+		goto free_out;
+
+#define MLX5_GET_CTR(p, x) \
+	MLX5_GET64(query_vport_counter_out, p, x)
+
+	s->rx_error_packets     =
+		MLX5_GET_CTR(out, received_errors.packets);
+	s->rx_error_bytes       =
+		MLX5_GET_CTR(out, received_errors.octets);
+	s->tx_error_packets     =
+		MLX5_GET_CTR(out, transmit_errors.packets);
+	s->tx_error_bytes       =
+		MLX5_GET_CTR(out, transmit_errors.octets);
+
+	s->rx_unicast_packets   =
+		MLX5_GET_CTR(out, received_eth_unicast.packets);
+	s->rx_unicast_bytes     =
+		MLX5_GET_CTR(out, received_eth_unicast.octets);
+	s->tx_unicast_packets   =
+		MLX5_GET_CTR(out, transmitted_eth_unicast.packets);
+	s->tx_unicast_bytes     =
+		MLX5_GET_CTR(out, transmitted_eth_unicast.octets);
+
+	s->rx_multicast_packets =
+		MLX5_GET_CTR(out, received_eth_multicast.packets);
+	s->rx_multicast_bytes   =
+		MLX5_GET_CTR(out, received_eth_multicast.octets);
+	s->tx_multicast_packets =
+		MLX5_GET_CTR(out, transmitted_eth_multicast.packets);
+	s->tx_multicast_bytes   =
+		MLX5_GET_CTR(out, transmitted_eth_multicast.octets);
+
+	s->rx_broadcast_packets =
+		MLX5_GET_CTR(out, received_eth_broadcast.packets);
+	s->rx_broadcast_bytes   =
+		MLX5_GET_CTR(out, received_eth_broadcast.octets);
+	s->tx_broadcast_packets =
+		MLX5_GET_CTR(out, transmitted_eth_broadcast.packets);
+	s->tx_broadcast_bytes   =
+		MLX5_GET_CTR(out, transmitted_eth_broadcast.octets);
+
+	s->rx_packets =
+		s->rx_unicast_packets +
+		s->rx_multicast_packets +
+		s->rx_broadcast_packets;
+	s->rx_bytes =
+		s->rx_unicast_bytes +
+		s->rx_multicast_bytes +
+		s->rx_broadcast_bytes;
+	s->tx_packets =
+		s->tx_unicast_packets +
+		s->tx_multicast_packets +
+		s->tx_broadcast_packets;
+	s->tx_bytes =
+		s->tx_unicast_bytes +
+		s->tx_multicast_bytes +
+		s->tx_broadcast_bytes;
+
+	/* Update calculated offload counters */
+	s->tx_csum_offload = s->tx_packets - tx_offload_none;
+	s->rx_csum_good    = s->rx_packets - s->rx_csum_none;
+
+free_out:
+	kvfree(out);
+}
+
+static void mlx5e_update_stats_work(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct mlx5e_priv *priv = container_of(dwork, struct mlx5e_priv,
+					       update_stats_work);
+	mutex_lock(&priv->state_lock);
+	if (test_bit(MLX5E_STATE_OPENED, &priv->state)) {
+		mlx5e_update_stats(priv);
+		schedule_delayed_work(dwork,
+				      msecs_to_jiffies(
+					      MLX5E_UPDATE_STATS_INTERVAL));
+	}
+	mutex_unlock(&priv->state_lock);
+}
+
+static void __mlx5e_async_event(struct mlx5e_priv *priv,
+				enum mlx5_dev_event event)
+{
+	switch (event) {
+	case MLX5_DEV_EVENT_PORT_UP:
+	case MLX5_DEV_EVENT_PORT_DOWN:
+		schedule_work(&priv->update_carrier_work);
+		break;
+
+	default:
+		break;
+	}
+}
+
+static void mlx5e_async_event(struct mlx5_core_dev *mdev, void *vpriv,
+			      enum mlx5_dev_event event, unsigned long param)
+{
+	struct mlx5e_priv *priv = vpriv;
+
+	spin_lock(&priv->async_events_spinlock);
+	if (test_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLE, &priv->state))
+		__mlx5e_async_event(priv, event);
+	spin_unlock(&priv->async_events_spinlock);
+}
+
+static void mlx5e_enable_async_events(struct mlx5e_priv *priv)
+{
+	set_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLE, &priv->state);
+}
+
+static void mlx5e_disable_async_events(struct mlx5e_priv *priv)
+{
+	spin_lock_irq(&priv->async_events_spinlock);
+	clear_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLE, &priv->state);
+	spin_unlock_irq(&priv->async_events_spinlock);
+}
+
+static void mlx5e_send_nop(struct mlx5e_sq *sq)
+{
+	struct mlx5_wq_cyc                *wq  = &sq->wq;
+
+	u16 pi = sq->pc & wq->sz_m1;
+	struct mlx5e_tx_wqe              *wqe  = mlx5_wq_cyc_get_wqe(wq, pi);
+
+	struct mlx5_wqe_ctrl_seg         *cseg = &wqe->ctrl;
+
+	memset(cseg, 0, sizeof(*cseg));
+
+	cseg->opmod_idx_opcode = cpu_to_be32((sq->pc << 8) | MLX5_OPCODE_NOP);
+	cseg->qpn_ds           = cpu_to_be32((sq->sqn << 8) | 0x01);
+	cseg->fm_ce_se         = MLX5_WQE_CTRL_CQ_UPDATE;
+
+	sq->skb[pi] = NULL;
+	sq->pc++;
+	mlx5e_tx_notify_hw(sq, wqe);
+}
+
+static int mlx5e_create_rq(struct mlx5e_channel *c,
+			   struct mlx5e_rq_param *param,
+			   struct mlx5e_rq *rq)
+{
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	void *rqc = param->rqc;
+	void *rqc_wq = MLX5_ADDR_OF(rqc, rqc, wq);
+	int wq_sz;
+	int err;
+	int i;
+
+	err = mlx5_wq_ll_create(mdev, &param->wq, rqc_wq, &rq->wq,
+				&rq->wq_ctrl);
+	if (err)
+		return err;
+
+	rq->wq.db = &rq->wq.db[MLX5_RCV_DBR];
+
+	wq_sz = mlx5_wq_ll_get_size(&rq->wq);
+	rq->skb = kzalloc_node(wq_sz * sizeof(*rq->skb), GFP_KERNEL,
+			       cpu_to_node(c->cpu));
+	if (!rq->skb) {
+		err = -ENOMEM;
+		goto err_rq_wq_destroy;
+	}
+
+	rq->wqe_sz = (priv->params.lro_en) ? priv->params.lro_wqe_sz :
+				priv->netdev->mtu + ETH_HLEN + VLAN_HLEN;
+
+	for (i = 0; i < wq_sz; i++) {
+		struct mlx5e_rx_wqe *wqe = mlx5_wq_ll_get_wqe(&rq->wq, i);
+
+		wqe->data.lkey       = c->mkey_be;
+		wqe->data.byte_count = cpu_to_be32(rq->wqe_sz);
+	}
+
+	rq->pdev    = c->pdev;
+	rq->netdev  = c->netdev;
+	rq->channel = c;
+	rq->ix      = c->ix;
+
+	return 0;
+
+err_rq_wq_destroy:
+	mlx5_wq_destroy(&rq->wq_ctrl);
+
+	return err;
+}
+
+static void mlx5e_destroy_rq(struct mlx5e_rq *rq)
+{
+	kfree(rq->skb);
+	mlx5_wq_destroy(&rq->wq_ctrl);
+}
+
+static int mlx5e_enable_rq(struct mlx5e_rq *rq, struct mlx5e_rq_param *param)
+{
+	struct mlx5e_channel *c = rq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	void *in;
+	void *rqc;
+	void *wq;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_rq_in) +
+		sizeof(u64) * rq->wq_ctrl.buf.npages;
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	rqc = MLX5_ADDR_OF(create_rq_in, in, ctx);
+	wq  = MLX5_ADDR_OF(rqc, rqc, wq);
+
+	memcpy(rqc, param->rqc, sizeof(param->rqc));
+
+	MLX5_SET(rqc,  rqc, cqn,		c->rq.cq.mcq.cqn);
+	MLX5_SET(rqc,  rqc, state,		MLX5_RQC_STATE_RST);
+	MLX5_SET(rqc,  rqc, flush_in_error_en,	1);
+	MLX5_SET(wq,   wq,  wq_type,		MLX5_WQ_TYPE_LINKED_LIST);
+	MLX5_SET(wq,   wq,  log_wq_pg_sz,	rq->wq_ctrl.buf.page_shift -
+						PAGE_SHIFT);
+	MLX5_SET64(wq, wq,  dbr_addr,		rq->wq_ctrl.db.dma);
+
+	mlx5_fill_page_array(&rq->wq_ctrl.buf,
+			     (__be64 *)MLX5_ADDR_OF(wq, wq, pas));
+
+	err = mlx5_create_rq(mdev, in, inlen, &rq->rqn);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5e_modify_rq(struct mlx5e_rq *rq, int curr_state, int next_state)
+{
+	struct mlx5e_channel *c = rq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	void *in;
+	void *rqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_rq_in);
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	rqc = MLX5_ADDR_OF(modify_rq_in, in, ctx);
+
+	MLX5_SET(modify_rq_in, in, rq_state, curr_state);
+	MLX5_SET(rqc, rqc, state, next_state);
+
+	err = mlx5_modify_rq(mdev, rq->rqn, in, inlen);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5e_disable_rq(struct mlx5e_rq *rq)
+{
+	struct mlx5e_channel *c = rq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	mlx5_destroy_rq(mdev, rq->rqn);
+}
+
+static int mlx5e_wait_for_min_rx_wqes(struct mlx5e_rq *rq)
+{
+	struct mlx5e_channel *c = rq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_wq_ll *wq = &rq->wq;
+	int i;
+
+	for (i = 0; i < 1000; i++) {
+		if (wq->cur_sz >= priv->params.min_rx_wqes)
+			return 0;
+
+		msleep(20);
+	}
+
+	return -ETIMEDOUT;
+}
+
+static int mlx5e_open_rq(struct mlx5e_channel *c,
+			 struct mlx5e_rq_param *param,
+			 struct mlx5e_rq *rq)
+{
+	int err;
+
+	err = mlx5e_create_rq(c, param, rq);
+	if (err)
+		return err;
+
+	err = mlx5e_enable_rq(rq, param);
+	if (err)
+		goto err_destroy_rq;
+
+	err = mlx5e_modify_rq(rq, MLX5_RQC_STATE_RST, MLX5_RQC_STATE_RDY);
+	if (err)
+		goto err_disable_rq;
+
+	set_bit(MLX5E_RQ_STATE_POST_WQES_ENABLE, &rq->state);
+	mlx5e_send_nop(&c->sq[0]); /* trigger mlx5e_post_rx_wqes() */
+
+	return 0;
+
+err_disable_rq:
+	mlx5e_disable_rq(rq);
+err_destroy_rq:
+	mlx5e_destroy_rq(rq);
+
+	return err;
+}
+
+static void mlx5e_close_rq(struct mlx5e_rq *rq)
+{
+	clear_bit(MLX5E_RQ_STATE_POST_WQES_ENABLE, &rq->state);
+	napi_synchronize(&rq->channel->napi); /* prevent mlx5e_post_rx_wqes */
+
+	mlx5e_modify_rq(rq, MLX5_RQC_STATE_RDY, MLX5_RQC_STATE_ERR);
+	while (!mlx5_wq_ll_is_empty(&rq->wq))
+		msleep(20);
+
+	/* avoid destroying rq before mlx5e_poll_rx_cq() is done with it */
+	napi_synchronize(&rq->channel->napi);
+
+	mlx5e_disable_rq(rq);
+	mlx5e_destroy_rq(rq);
+}
+
+static void mlx5e_free_sq_db(struct mlx5e_sq *sq)
+{
+	kfree(sq->dma_fifo);
+	kfree(sq->skb);
+}
+
+static int mlx5e_alloc_sq_db(struct mlx5e_sq *sq, int numa)
+{
+	int wq_sz = mlx5_wq_cyc_get_size(&sq->wq);
+	int df_sz = wq_sz * MLX5_SEND_WQEBB_NUM_DS;
+
+	sq->skb = kzalloc_node(wq_sz * sizeof(*sq->skb), GFP_KERNEL, numa);
+	sq->dma_fifo = kzalloc_node(df_sz * sizeof(*sq->dma_fifo), GFP_KERNEL,
+				    numa);
+
+	if (!sq->skb || !sq->dma_fifo) {
+		mlx5e_free_sq_db(sq);
+		return -ENOMEM;
+	}
+
+	sq->dma_fifo_mask = df_sz - 1;
+
+	return 0;
+}
+
+static int mlx5e_create_sq(struct mlx5e_channel *c,
+			   int tc,
+			   struct mlx5e_sq_param *param,
+			   struct mlx5e_sq *sq)
+{
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	void *sqc = param->sqc;
+	void *sqc_wq = MLX5_ADDR_OF(sqc, sqc, wq);
+	int err;
+
+	err = mlx5_alloc_map_uar(mdev, &sq->uar);
+	if (err)
+		return err;
+
+	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, &sq->wq,
+				 &sq->wq_ctrl);
+	if (err)
+		goto err_unmap_free_uar;
+
+	sq->wq.db       = &sq->wq.db[MLX5_SND_DBR];
+	sq->uar_map     = sq->uar.map;
+	sq->bf_buf_size = (1 << MLX5_CAP_GEN(mdev, log_bf_reg_size)) / 2;
+
+	if (mlx5e_alloc_sq_db(sq, cpu_to_node(c->cpu)))
+		goto err_sq_wq_destroy;
+
+	sq->txq = netdev_get_tx_queue(priv->netdev,
+				      c->ix + tc * priv->params.num_channels);
+
+	sq->pdev    = c->pdev;
+	sq->mkey_be = c->mkey_be;
+	sq->channel = c;
+	sq->tc      = tc;
+
+	return 0;
+
+err_sq_wq_destroy:
+	mlx5_wq_destroy(&sq->wq_ctrl);
+
+err_unmap_free_uar:
+	mlx5_unmap_free_uar(mdev, &sq->uar);
+
+	return err;
+}
+
+static void mlx5e_destroy_sq(struct mlx5e_sq *sq)
+{
+	struct mlx5e_channel *c = sq->channel;
+	struct mlx5e_priv *priv = c->priv;
+
+	mlx5e_free_sq_db(sq);
+	mlx5_wq_destroy(&sq->wq_ctrl);
+	mlx5_unmap_free_uar(priv->mdev, &sq->uar);
+}
+
+static int mlx5e_enable_sq(struct mlx5e_sq *sq, struct mlx5e_sq_param *param)
+{
+	struct mlx5e_channel *c = sq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	void *in;
+	void *sqc;
+	void *wq;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_sq_in) +
+		sizeof(u64) * sq->wq_ctrl.buf.npages;
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	sqc = MLX5_ADDR_OF(create_sq_in, in, ctx);
+	wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	memcpy(sqc, param->sqc, sizeof(param->sqc));
+
+	MLX5_SET(sqc,  sqc, user_index,		sq->tc);
+	MLX5_SET(sqc,  sqc, tis_num_0,		priv->tisn[sq->tc]);
+	MLX5_SET(sqc,  sqc, cqn,		c->sq[sq->tc].cq.mcq.cqn);
+	MLX5_SET(sqc,  sqc, state,		MLX5_SQC_STATE_RST);
+	MLX5_SET(sqc,  sqc, tis_lst_sz,		1);
+	MLX5_SET(sqc,  sqc, flush_in_error_en,	1);
+
+	MLX5_SET(wq,   wq, wq_type,       MLX5_WQ_TYPE_CYCLIC);
+	MLX5_SET(wq,   wq, uar_page,      sq->uar.index);
+	MLX5_SET(wq,   wq, log_wq_pg_sz,  sq->wq_ctrl.buf.page_shift -
+					  PAGE_SHIFT);
+	MLX5_SET64(wq, wq, dbr_addr,      sq->wq_ctrl.db.dma);
+
+	mlx5_fill_page_array(&sq->wq_ctrl.buf,
+			     (__be64 *)MLX5_ADDR_OF(wq, wq, pas));
+
+	err = mlx5_create_sq(mdev, in, inlen, &sq->sqn);
+
+	kvfree(in);
+
+	return err;
+}
+
+static int mlx5e_modify_sq(struct mlx5e_sq *sq, int curr_state, int next_state)
+{
+	struct mlx5e_channel *c = sq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	void *in;
+	void *sqc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(modify_sq_in);
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	sqc = MLX5_ADDR_OF(modify_sq_in, in, ctx);
+
+	MLX5_SET(modify_sq_in, in, sq_state, curr_state);
+	MLX5_SET(sqc, sqc, state, next_state);
+
+	err = mlx5_modify_sq(mdev, sq->sqn, in, inlen);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5e_disable_sq(struct mlx5e_sq *sq)
+{
+	struct mlx5e_channel *c = sq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	mlx5_destroy_sq(mdev, sq->sqn);
+}
+
+static int mlx5e_open_sq(struct mlx5e_channel *c,
+			 int tc,
+			 struct mlx5e_sq_param *param,
+			 struct mlx5e_sq *sq)
+{
+	int err;
+
+	err = mlx5e_create_sq(c, tc, param, sq);
+	if (err)
+		return err;
+
+	err = mlx5e_enable_sq(sq, param);
+	if (err)
+		goto err_destroy_sq;
+
+	err = mlx5e_modify_sq(sq, MLX5_SQC_STATE_RST, MLX5_SQC_STATE_RDY);
+	if (err)
+		goto err_disable_sq;
+
+	set_bit(MLX5E_SQ_STATE_WAKE_TXQ_ENABLE, &sq->state);
+	netdev_tx_reset_queue(sq->txq);
+	netif_tx_start_queue(sq->txq);
+
+	return 0;
+
+err_disable_sq:
+	mlx5e_disable_sq(sq);
+err_destroy_sq:
+	mlx5e_destroy_sq(sq);
+
+	return err;
+}
+
+static inline void netif_tx_disable_queue(struct netdev_queue *txq)
+{
+	__netif_tx_lock_bh(txq);
+	netif_tx_stop_queue(txq);
+	__netif_tx_unlock_bh(txq);
+}
+
+static void mlx5e_close_sq(struct mlx5e_sq *sq)
+{
+	clear_bit(MLX5E_SQ_STATE_WAKE_TXQ_ENABLE, &sq->state);
+	napi_synchronize(&sq->channel->napi); /* prevent netif_tx_wake_queue */
+	netif_tx_disable_queue(sq->txq);
+
+	/* ensure hw is notified of all pending wqes */
+	if (mlx5e_sq_has_room_for(sq, 1))
+		mlx5e_send_nop(sq);
+
+	mlx5e_modify_sq(sq, MLX5_SQC_STATE_RDY, MLX5_SQC_STATE_ERR);
+	while (sq->cc != sq->pc) /* wait till sq is empty */
+		msleep(20);
+
+	/* avoid destroying sq before mlx5e_poll_tx_cq() is done with it */
+	napi_synchronize(&sq->channel->napi);
+
+	mlx5e_disable_sq(sq);
+	mlx5e_destroy_sq(sq);
+}
+
+static int mlx5e_create_cq(struct mlx5e_channel *c,
+			   struct mlx5e_cq_param *param,
+			   struct mlx5e_cq *cq)
+{
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_core_cq *mcq = &cq->mcq;
+	int eqn_not_used;
+	int irqn;
+	int err;
+	u32 i;
+
+	param->wq.numa = cpu_to_node(c->cpu);
+	param->eq_ix   = c->ix;
+
+	err = mlx5_cqwq_create(mdev, &param->wq, param->cqc, &cq->wq,
+			       &cq->wq_ctrl);
+	if (err)
+		return err;
+
+	mlx5_vector2eqn(mdev, param->eq_ix, &eqn_not_used, &irqn);
+
+	cq->napi        = &c->napi;
+
+	mcq->cqe_sz     = 64;
+	mcq->set_ci_db  = cq->wq_ctrl.db.db;
+	mcq->arm_db     = cq->wq_ctrl.db.db + 1;
+	*mcq->set_ci_db = 0;
+	*mcq->arm_db    = 0;
+	mcq->vector     = param->eq_ix;
+	mcq->comp       = mlx5e_completion_event;
+	mcq->event      = mlx5e_cq_error_event;
+	mcq->irqn       = irqn;
+	mcq->uar        = &priv->cq_uar;
+
+	for (i = 0; i < mlx5_cqwq_get_size(&cq->wq); i++) {
+		struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(&cq->wq, i);
+
+		cqe->op_own = 0xf1;
+	}
+
+	cq->channel = c;
+
+	return 0;
+}
+
+static void mlx5e_destroy_cq(struct mlx5e_cq *cq)
+{
+	mlx5_wq_destroy(&cq->wq_ctrl);
+}
+
+static int mlx5e_enable_cq(struct mlx5e_cq *cq, struct mlx5e_cq_param *param)
+{
+	struct mlx5e_channel *c = cq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_core_cq *mcq = &cq->mcq;
+
+	void *in;
+	void *cqc;
+	int inlen;
+	int irqn_not_used;
+	int eqn;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_cq_in) +
+		sizeof(u64) * cq->wq_ctrl.buf.npages;
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
+
+	memcpy(cqc, param->cqc, sizeof(param->cqc));
+
+	mlx5_fill_page_array(&cq->wq_ctrl.buf,
+			     (__be64 *)MLX5_ADDR_OF(create_cq_in, in, pas));
+
+	mlx5_vector2eqn(mdev, param->eq_ix, &eqn, &irqn_not_used);
+
+	MLX5_SET(cqc,   cqc, c_eqn,         eqn);
+	MLX5_SET(cqc,   cqc, uar_page,      mcq->uar->index);
+	MLX5_SET(cqc,   cqc, log_page_size, cq->wq_ctrl.buf.page_shift -
+					    PAGE_SHIFT);
+	MLX5_SET64(cqc, cqc, dbr_addr,      cq->wq_ctrl.db.dma);
+
+	err = mlx5_core_create_cq(mdev, mcq, in, inlen);
+
+	kvfree(in);
+
+	if (err)
+		return err;
+
+	mlx5e_cq_arm(cq);
+
+	return 0;
+}
+
+static void mlx5e_disable_cq(struct mlx5e_cq *cq)
+{
+	struct mlx5e_channel *c = cq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	mlx5_core_destroy_cq(mdev, &cq->mcq);
+}
+
+static int mlx5e_open_cq(struct mlx5e_channel *c,
+			 struct mlx5e_cq_param *param,
+			 struct mlx5e_cq *cq,
+			 u16 moderation_usecs,
+			 u16 moderation_frames)
+{
+	int err;
+	struct mlx5e_priv *priv = c->priv;
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	err = mlx5e_create_cq(c, param, cq);
+	if (err)
+		return err;
+
+	err = mlx5e_enable_cq(cq, param);
+	if (err)
+		goto err_destroy_cq;
+
+	err = mlx5_core_modify_cq_moderation(mdev, &cq->mcq,
+					     moderation_usecs,
+					     moderation_frames);
+	if (err)
+		goto err_destroy_cq;
+
+	return 0;
+
+err_destroy_cq:
+	mlx5e_destroy_cq(cq);
+
+	return err;
+}
+
+static void mlx5e_close_cq(struct mlx5e_cq *cq)
+{
+	mlx5e_disable_cq(cq);
+	mlx5e_destroy_cq(cq);
+}
+
+static int mlx5e_get_cpu(struct mlx5e_priv *priv, int ix)
+{
+	return cpumask_first(priv->mdev->priv.irq_info[ix].mask);
+}
+
+static int mlx5e_open_tx_cqs(struct mlx5e_channel *c,
+			     struct mlx5e_channel_param *cparam)
+{
+	struct mlx5e_priv *priv = c->priv;
+	int err;
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++) {
+		err = mlx5e_open_cq(c, &cparam->tx_cq, &c->sq[tc].cq,
+				    priv->params.tx_cq_moderation_usec,
+				    priv->params.tx_cq_moderation_pkts);
+		if (err)
+			goto err_close_tx_cqs;
+
+		c->sq[tc].cq.sqrq = &c->sq[tc];
+	}
+
+	return 0;
+
+err_close_tx_cqs:
+	for (tc--; tc >= 0; tc--)
+		mlx5e_close_cq(&c->sq[tc].cq);
+
+	return err;
+}
+
+static void mlx5e_close_tx_cqs(struct mlx5e_channel *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_cq(&c->sq[tc].cq);
+}
+
+static int mlx5e_open_sqs(struct mlx5e_channel *c,
+			  struct mlx5e_channel_param *cparam)
+{
+	int err;
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++) {
+		err = mlx5e_open_sq(c, tc, &cparam->sq, &c->sq[tc]);
+		if (err)
+			goto err_close_sqs;
+	}
+
+	return 0;
+
+err_close_sqs:
+	for (tc--; tc >= 0; tc--)
+		mlx5e_close_sq(&c->sq[tc]);
+
+	return err;
+}
+
+static void mlx5e_close_sqs(struct mlx5e_channel *c)
+{
+	int tc;
+
+	for (tc = 0; tc < c->num_tc; tc++)
+		mlx5e_close_sq(&c->sq[tc]);
+}
+
+static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
+			      struct mlx5e_channel_param *cparam,
+			      struct mlx5e_channel **cp)
+{
+	struct net_device *netdev = priv->netdev;
+	int cpu = mlx5e_get_cpu(priv, ix);
+	struct mlx5e_channel *c;
+	int err;
+
+	c = kzalloc_node(sizeof(*c), GFP_KERNEL, cpu_to_node(cpu));
+	if (!c)
+		return -ENOMEM;
+
+	c->priv     = priv;
+	c->ix       = ix;
+	c->cpu      = cpu;
+	c->pdev     = &priv->mdev->pdev->dev;
+	c->netdev   = priv->netdev;
+	c->mkey_be  = cpu_to_be32(priv->mr.key);
+	c->num_tc   = priv->num_tc;
+
+	netif_napi_add(netdev, &c->napi, mlx5e_napi_poll, 64);
+
+	err = mlx5e_open_tx_cqs(c, cparam);
+	if (err)
+		goto err_napi_del;
+
+	err = mlx5e_open_cq(c, &cparam->rx_cq, &c->rq.cq,
+			    priv->params.rx_cq_moderation_usec,
+			    priv->params.rx_cq_moderation_pkts);
+	if (err)
+		goto err_close_tx_cqs;
+	c->rq.cq.sqrq = &c->rq;
+
+	napi_enable(&c->napi);
+
+	err = mlx5e_open_sqs(c, cparam);
+	if (err)
+		goto err_disable_napi;
+
+	err = mlx5e_open_rq(c, &cparam->rq, &c->rq);
+	if (err)
+		goto err_close_sqs;
+
+	netif_set_xps_queue(netdev, get_cpu_mask(c->cpu), ix);
+	*cp = c;
+
+	return 0;
+
+err_close_sqs:
+	mlx5e_close_sqs(c);
+
+err_disable_napi:
+	napi_disable(&c->napi);
+	mlx5e_close_cq(&c->rq.cq);
+
+err_close_tx_cqs:
+	mlx5e_close_tx_cqs(c);
+
+err_napi_del:
+	netif_napi_del(&c->napi);
+	kfree(c);
+
+	return err;
+}
+
+static void mlx5e_close_channel(struct mlx5e_channel *c)
+{
+	mlx5e_close_rq(&c->rq);
+	mlx5e_close_sqs(c);
+	napi_disable(&c->napi);
+	mlx5e_close_cq(&c->rq.cq);
+	mlx5e_close_tx_cqs(c);
+	netif_napi_del(&c->napi);
+	kfree(c);
+}
+
+static void mlx5e_build_rq_param(struct mlx5e_priv *priv,
+				 struct mlx5e_rq_param *param)
+{
+	void *rqc = param->rqc;
+	void *wq = MLX5_ADDR_OF(rqc, rqc, wq);
+
+	MLX5_SET(wq, wq, wq_type,          MLX5_WQ_TYPE_LINKED_LIST);
+	MLX5_SET(wq, wq, end_padding_mode, MLX5_WQ_END_PAD_MODE_ALIGN);
+	MLX5_SET(wq, wq, log_wq_stride,    ilog2(sizeof(struct mlx5e_rx_wqe)));
+	MLX5_SET(wq, wq, log_wq_sz,        priv->params.log_rq_size);
+	MLX5_SET(wq, wq, pd,               priv->pdn);
+
+	param->wq.numa   = dev_to_node(&priv->mdev->pdev->dev);
+	param->wq.linear = 1;
+}
+
+static void mlx5e_build_sq_param(struct mlx5e_priv *priv,
+				 struct mlx5e_sq_param *param)
+{
+	void *sqc = param->sqc;
+	void *wq = MLX5_ADDR_OF(sqc, sqc, wq);
+
+	MLX5_SET(wq, wq, log_wq_sz,     priv->params.log_sq_size);
+	MLX5_SET(wq, wq, log_wq_stride, ilog2(MLX5_SEND_WQE_BB));
+	MLX5_SET(wq, wq, pd,            priv->pdn);
+
+	param->wq.numa = dev_to_node(&priv->mdev->pdev->dev);
+}
+
+static void mlx5e_build_common_cq_param(struct mlx5e_priv *priv,
+					struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, uar_page, priv->cq_uar.index);
+}
+
+static void mlx5e_build_rx_cq_param(struct mlx5e_priv *priv,
+				    struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, log_cq_size,  priv->params.log_rq_size);
+
+	mlx5e_build_common_cq_param(priv, param);
+}
+
+static void mlx5e_build_tx_cq_param(struct mlx5e_priv *priv,
+				    struct mlx5e_cq_param *param)
+{
+	void *cqc = param->cqc;
+
+	MLX5_SET(cqc, cqc, log_cq_size,  priv->params.log_sq_size);
+
+	mlx5e_build_common_cq_param(priv, param);
+}
+
+static void mlx5e_build_channel_param(struct mlx5e_priv *priv,
+				      struct mlx5e_channel_param *cparam)
+{
+	memset(cparam, 0, sizeof(*cparam));
+
+	mlx5e_build_rq_param(priv, &cparam->rq);
+	mlx5e_build_sq_param(priv, &cparam->sq);
+	mlx5e_build_rx_cq_param(priv, &cparam->rx_cq);
+	mlx5e_build_tx_cq_param(priv, &cparam->tx_cq);
+}
+
+static int mlx5e_open_channels(struct mlx5e_priv *priv)
+{
+	struct mlx5e_channel_param cparam;
+	int err;
+	int i;
+	int j;
+
+	priv->channel = kcalloc(priv->params.num_channels,
+				sizeof(struct mlx5e_channel *), GFP_KERNEL);
+	if (!priv->channel)
+		return -ENOMEM;
+
+	mlx5e_build_channel_param(priv, &cparam);
+	for (i = 0; i < priv->params.num_channels; i++) {
+		err = mlx5e_open_channel(priv, i, &cparam, &priv->channel[i]);
+		if (err)
+			goto err_close_channels;
+	}
+
+	for (j = 0; j < priv->params.num_channels; j++) {
+		err = mlx5e_wait_for_min_rx_wqes(&priv->channel[j]->rq);
+		if (err)
+			goto err_close_channels;
+	}
+
+	return 0;
+
+err_close_channels:
+	for (i--; i >= 0; i--)
+		mlx5e_close_channel(priv->channel[i]);
+
+	kfree(priv->channel);
+
+	return err;
+}
+
+static void mlx5e_close_channels(struct mlx5e_priv *priv)
+{
+	int i;
+
+	for (i = 0; i < priv->params.num_channels; i++)
+		mlx5e_close_channel(priv->channel[i]);
+
+	kfree(priv->channel);
+}
+
+static int mlx5e_open_tis(struct mlx5e_priv *priv, int tc)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 in[MLX5_ST_SZ_DW(create_tis_in)];
+	void *tisc = MLX5_ADDR_OF(create_tis_in, in, ctx);
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(tisc, tisc, prio,  tc);
+
+	return mlx5_create_tis(mdev, in, sizeof(in), &priv->tisn[tc]);
+}
+
+static void mlx5e_close_tis(struct mlx5e_priv *priv, int tc)
+{
+	mlx5_destroy_tis(priv->mdev, priv->tisn[tc]);
+}
+
+static int mlx5e_open_tises(struct mlx5e_priv *priv)
+{
+	int num_tc = priv->num_tc;
+	int err;
+	int tc;
+
+	for (tc = 0; tc < num_tc; tc++) {
+		err = mlx5e_open_tis(priv, tc);
+		if (err)
+			goto err_close_tises;
+	}
+
+	return 0;
+
+err_close_tises:
+	for (tc--; tc >= 0; tc--)
+		mlx5e_close_tis(priv, tc);
+
+	return err;
+}
+
+static void mlx5e_close_tises(struct mlx5e_priv *priv)
+{
+	int num_tc = priv->num_tc;
+	int tc;
+
+	for (tc = 0; tc < num_tc; tc++)
+		mlx5e_close_tis(priv, tc);
+}
+
+static int mlx5e_open_rqt(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 *in;
+	u32 out[MLX5_ST_SZ_DW(create_rqt_out)];
+	void *rqtc;
+	int inlen;
+	int err;
+	int sz;
+	int i;
+
+	sz = 1 << priv->params.rx_hash_log_tbl_sz;
+
+	inlen = MLX5_ST_SZ_BYTES(create_rqt_in) + sizeof(u32) * sz;
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	rqtc = MLX5_ADDR_OF(create_rqt_in, in, rqt_context);
+
+	MLX5_SET(rqtc, rqtc, rqt_actual_size, sz);
+	MLX5_SET(rqtc, rqtc, rqt_max_size, sz);
+
+	for (i = 0; i < sz; i++) {
+		int ix = i % priv->params.num_channels;
+
+		MLX5_SET(rqtc, rqtc, rq_num[i], priv->channel[ix]->rq.rqn);
+	}
+
+	MLX5_SET(create_rqt_in, in, opcode, MLX5_CMD_OP_CREATE_RQT);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(mdev, in, inlen, out, sizeof(out));
+	if (!err)
+		priv->rqtn = MLX5_GET(create_rqt_out, out, rqtn);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5e_close_rqt(struct mlx5e_priv *priv)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_rqt_in)];
+	u32 out[MLX5_ST_SZ_DW(destroy_rqt_out)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(destroy_rqt_in, in, opcode, MLX5_CMD_OP_DESTROY_RQT);
+	MLX5_SET(destroy_rqt_in, in, rqtn, priv->rqtn);
+
+	mlx5_cmd_exec_check_status(priv->mdev, in, sizeof(in), out,
+				   sizeof(out));
+}
+
+static void mlx5e_build_tir_ctx(struct mlx5e_priv *priv, u32 *tirc, int tt)
+{
+	void *hfso = MLX5_ADDR_OF(tirc, tirc, rx_hash_field_selector_outer);
+
+#define ROUGH_MAX_L2_L3_HDR_SZ 256
+
+#define MLX5_HASH_IP     (MLX5_HASH_FIELD_SEL_SRC_IP   |\
+			  MLX5_HASH_FIELD_SEL_DST_IP)
+
+#define MLX5_HASH_ALL    (MLX5_HASH_FIELD_SEL_SRC_IP   |\
+			  MLX5_HASH_FIELD_SEL_DST_IP   |\
+			  MLX5_HASH_FIELD_SEL_L4_SPORT |\
+			  MLX5_HASH_FIELD_SEL_L4_DPORT)
+
+	if (priv->params.lro_en) {
+		MLX5_SET(tirc, tirc, lro_enable_mask,
+			 MLX5_TIRC_LRO_ENABLE_MASK_IPV4_LRO |
+			 MLX5_TIRC_LRO_ENABLE_MASK_IPV6_LRO);
+		MLX5_SET(tirc, tirc, lro_max_ip_payload_size,
+			 (priv->params.lro_wqe_sz -
+			  ROUGH_MAX_L2_L3_HDR_SZ) >> 8);
+		MLX5_SET(tirc, tirc, lro_timeout_period_usecs,
+			 MLX5_CAP_ETH(priv->mdev,
+				      lro_timer_supported_periods[3]));
+	}
+
+	switch (tt) {
+	case MLX5E_TT_ANY:
+		MLX5_SET(tirc, tirc, disp_type,
+			 MLX5_TIRC_DISP_TYPE_DIRECT);
+		MLX5_SET(tirc, tirc, inline_rqn,
+			 priv->channel[0]->rq.rqn);
+		break;
+	default:
+		MLX5_SET(tirc, tirc, disp_type,
+			 MLX5_TIRC_DISP_TYPE_INDIRECT);
+		MLX5_SET(tirc, tirc, indirect_table,
+			 priv->rqtn);
+		MLX5_SET(tirc, tirc, rx_hash_fn,
+			 MLX5_TIRC_RX_HASH_FN_HASH_TOEPLITZ);
+		MLX5_SET(tirc, tirc, rx_hash_symmetric, 1);
+		netdev_rss_key_fill(MLX5_ADDR_OF(tirc, tirc,
+						 rx_hash_toeplitz_key),
+				    MLX5_FLD_SZ_BYTES(tirc,
+						      rx_hash_toeplitz_key));
+		break;
+	}
+
+	switch (tt) {
+	case MLX5E_TT_IPV4_TCP:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV4);
+		MLX5_SET(rx_hash_field_select, hfso, l4_prot_type,
+			 MLX5_L4_PROT_TYPE_TCP);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_ALL);
+		break;
+
+	case MLX5E_TT_IPV6_TCP:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV6);
+		MLX5_SET(rx_hash_field_select, hfso, l4_prot_type,
+			 MLX5_L4_PROT_TYPE_TCP);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_ALL);
+		break;
+
+	case MLX5E_TT_IPV4_UDP:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV4);
+		MLX5_SET(rx_hash_field_select, hfso, l4_prot_type,
+			 MLX5_L4_PROT_TYPE_UDP);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_ALL);
+		break;
+
+	case MLX5E_TT_IPV6_UDP:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV6);
+		MLX5_SET(rx_hash_field_select, hfso, l4_prot_type,
+			 MLX5_L4_PROT_TYPE_UDP);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_ALL);
+		break;
+
+	case MLX5E_TT_IPV4:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV4);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_IP);
+		break;
+
+	case MLX5E_TT_IPV6:
+		MLX5_SET(rx_hash_field_select, hfso, l3_prot_type,
+			 MLX5_L3_PROT_TYPE_IPV6);
+		MLX5_SET(rx_hash_field_select, hfso, selected_fields,
+			 MLX5_HASH_IP);
+		break;
+	}
+}
+
+static int mlx5e_open_tir(struct mlx5e_priv *priv, int tt)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 *in;
+	void *tirc;
+	int inlen;
+	int err;
+
+	inlen = MLX5_ST_SZ_BYTES(create_tir_in);
+	in = mlx5_vzalloc(inlen);
+	if (!in)
+		return -ENOMEM;
+
+	tirc = MLX5_ADDR_OF(create_tir_in, in, ctx);
+
+	mlx5e_build_tir_ctx(priv, tirc, tt);
+
+	err = mlx5_create_tir(mdev, in, inlen, &priv->tirn[tt]);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5e_close_tir(struct mlx5e_priv *priv, int tt)
+{
+	mlx5_destroy_tir(priv->mdev, priv->tirn[tt]);
+}
+
+static int mlx5e_open_tirs(struct mlx5e_priv *priv)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < MLX5E_NUM_TT; i++) {
+		err = mlx5e_open_tir(priv, i);
+		if (err)
+			goto err_close_tirs;
+	}
+
+	return 0;
+
+err_close_tirs:
+	for (i--; i >= 0; i--)
+		mlx5e_close_tir(priv, i);
+
+	return err;
+}
+
+static void mlx5e_close_tirs(struct mlx5e_priv *priv)
+{
+	int i;
+
+	for (i = 0; i < MLX5E_NUM_TT; i++)
+		mlx5e_close_tir(priv, i);
+}
+
+int mlx5e_open_locked(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int actual_mtu;
+	int num_txqs;
+	int err;
+
+	num_txqs = roundup_pow_of_two(priv->params.num_channels) *
+		   priv->params.num_tc;
+	netif_set_real_num_tx_queues(netdev, num_txqs);
+	netif_set_real_num_rx_queues(netdev, priv->params.num_channels);
+
+	err = mlx5_set_port_mtu(mdev, netdev->mtu);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5_set_port_mtu failed %d\n",
+			   __func__, err);
+		return err;
+	}
+
+	err = mlx5_query_port_oper_mtu(mdev, &actual_mtu);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5_query_port_oper_mtu failed %d\n",
+			   __func__, err);
+		return err;
+	}
+
+	if (actual_mtu != netdev->mtu)
+		netdev_warn(netdev, "%s: Failed to set MTU to %d\n",
+			    __func__, netdev->mtu);
+
+	netdev->mtu = actual_mtu;
+
+	err = mlx5e_open_tises(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_open_tises failed, %d\n",
+			   __func__, err);
+		return err;
+	}
+
+	err = mlx5e_open_channels(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_open_channels failed, %d\n",
+			   __func__, err);
+		goto err_close_tises;
+	}
+
+	err = mlx5e_open_rqt(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_open_rqt failed, %d\n",
+			   __func__, err);
+		goto err_close_channels;
+	}
+
+	err = mlx5e_open_tirs(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_open_tir failed, %d\n",
+			   __func__, err);
+		goto err_close_rqls;
+	}
+
+	err = mlx5e_open_flow_table(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_open_flow_table failed, %d\n",
+			   __func__, err);
+		goto err_close_tirs;
+	}
+
+	err = mlx5e_add_all_vlan_rules(priv);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_add_all_vlan_rules failed, %d\n",
+			   __func__, err);
+		goto err_close_flow_table;
+	}
+
+	mlx5e_init_eth_addr(priv);
+
+	set_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	mlx5e_update_carrier(priv);
+	mlx5e_set_rx_mode_core(priv);
+
+	schedule_delayed_work(&priv->update_stats_work, 0);
+	return 0;
+
+err_close_flow_table:
+	mlx5e_close_flow_table(priv);
+
+err_close_tirs:
+	mlx5e_close_tirs(priv);
+
+err_close_rqls:
+	mlx5e_close_rqt(priv);
+
+err_close_channels:
+	mlx5e_close_channels(priv);
+
+err_close_tises:
+	mlx5e_close_tises(priv);
+
+	return err;
+}
+
+static int mlx5e_open(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_open_locked(netdev);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_close_locked(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	clear_bit(MLX5E_STATE_OPENED, &priv->state);
+
+	mlx5e_set_rx_mode_core(priv);
+	mlx5e_del_all_vlan_rules(priv);
+	netif_carrier_off(priv->netdev);
+	mlx5e_close_flow_table(priv);
+	mlx5e_close_tirs(priv);
+	mlx5e_close_rqt(priv);
+	mlx5e_close_channels(priv);
+	mlx5e_close_tises(priv);
+
+	return 0;
+}
+
+static int mlx5e_close(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	int err;
+
+	mutex_lock(&priv->state_lock);
+	err = mlx5e_close_locked(netdev);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+int mlx5e_update_priv_params(struct mlx5e_priv *priv,
+			     struct mlx5e_params *new_params)
+{
+	int err = 0;
+	int was_opened;
+
+	WARN_ON(!mutex_is_locked(&priv->state_lock));
+
+	was_opened = test_bit(MLX5E_STATE_OPENED, &priv->state);
+	if (was_opened)
+		mlx5e_close_locked(priv->netdev);
+
+	priv->params = *new_params;
+
+	if (was_opened)
+		err = mlx5e_open_locked(priv->netdev);
+
+	return err;
+}
+
+static struct rtnl_link_stats64 *
+mlx5e_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	struct mlx5e_vport_stats *vstats = &priv->stats.vport;
+
+	stats->rx_packets = vstats->rx_packets;
+	stats->rx_bytes   = vstats->rx_bytes;
+	stats->tx_packets = vstats->tx_packets;
+	stats->tx_bytes   = vstats->tx_bytes;
+	stats->multicast  = vstats->rx_multicast_packets +
+			    vstats->tx_multicast_packets;
+	stats->tx_errors  = vstats->tx_error_packets;
+	stats->rx_errors  = vstats->rx_error_packets;
+	stats->tx_dropped = vstats->tx_queue_dropped;
+	stats->rx_crc_errors = 0;
+	stats->rx_length_errors = 0;
+
+	return stats;
+}
+
+static void mlx5e_set_rx_mode(struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	schedule_work(&priv->set_rx_mode_work);
+}
+
+static int mlx5e_set_mac(struct net_device *netdev, void *addr)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct sockaddr *saddr = addr;
+
+	if (!is_valid_ether_addr(saddr->sa_data))
+		return -EADDRNOTAVAIL;
+
+	netif_addr_lock_bh(netdev);
+	ether_addr_copy(netdev->dev_addr, saddr->sa_data);
+	netif_addr_unlock_bh(netdev);
+
+	schedule_work(&priv->set_rx_mode_work);
+
+	return 0;
+}
+
+static int mlx5e_set_features(struct net_device *netdev,
+			      netdev_features_t features)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	netdev_features_t changes = features ^ netdev->features;
+	struct mlx5e_params new_params;
+	bool update_params = false;
+
+	mutex_lock(&priv->state_lock);
+	new_params = priv->params;
+
+	if (changes & NETIF_F_LRO) {
+		new_params.lro_en = !!(features & NETIF_F_LRO);
+		update_params = true;
+	}
+
+	if (update_params)
+		mlx5e_update_priv_params(priv, &new_params);
+
+	if (changes & NETIF_F_HW_VLAN_CTAG_FILTER) {
+		if (features & NETIF_F_HW_VLAN_CTAG_FILTER)
+			mlx5e_enable_vlan_filter(priv);
+		else
+			mlx5e_disable_vlan_filter(priv);
+	}
+
+	mutex_unlock(&priv->state_lock);
+
+	return 0;
+}
+
+static int mlx5e_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+	int max_mtu;
+	int err = 0;
+
+	err = mlx5_query_port_max_mtu(mdev, &max_mtu);
+	if (err)
+		return err;
+
+	if (new_mtu > max_mtu || new_mtu < MLX5E_PARAMS_MIN_MTU) {
+		netdev_err(netdev, "%s: Bad MTU size, mtu must be [%d-%d]\n",
+			   __func__, MLX5E_PARAMS_MIN_MTU, max_mtu);
+		return -EINVAL;
+	}
+
+	mutex_lock(&priv->state_lock);
+	netdev->mtu = new_mtu;
+	err = mlx5e_update_priv_params(priv, &priv->params);
+	mutex_unlock(&priv->state_lock);
+
+	return err;
+}
+
+static struct net_device_ops mlx5e_netdev_ops = {
+	.ndo_open                = mlx5e_open,
+	.ndo_stop                = mlx5e_close,
+	.ndo_start_xmit          = mlx5e_xmit,
+	.ndo_get_stats64         = mlx5e_get_stats,
+	.ndo_set_rx_mode         = mlx5e_set_rx_mode,
+	.ndo_set_mac_address     = mlx5e_set_mac,
+	.ndo_vlan_rx_add_vid	 = mlx5e_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid	 = mlx5e_vlan_rx_kill_vid,
+	.ndo_set_features        = mlx5e_set_features,
+	.ndo_change_mtu		 = mlx5e_change_mtu,
+};
+
+static int mlx5e_check_required_hca_cap(struct mlx5_core_dev *mdev)
+{
+	if (MLX5_CAP_GEN(mdev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
+		return -ENOTSUPP;
+	if (!MLX5_CAP_GEN(mdev, eth_net_offloads) ||
+	    !MLX5_CAP_GEN(mdev, nic_flow_table) ||
+	    !MLX5_CAP_ETH(mdev, csum_cap) ||
+	    !MLX5_CAP_ETH(mdev, max_lso_cap) ||
+	    !MLX5_CAP_ETH(mdev, vlan_cap) ||
+	    !MLX5_CAP_ETH(mdev, rss_ind_tbl_cap)) {
+		mlx5_core_warn(mdev,
+			       "Not creating net device, some required device capabilities are missing\n");
+		return -ENOTSUPP;
+	}
+	return 0;
+}
+
+static void mlx5e_build_netdev_priv(struct mlx5_core_dev *mdev,
+				    struct net_device *netdev,
+				    int num_comp_vectors)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	priv->params.log_sq_size           =
+		MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
+	priv->params.log_rq_size           =
+		MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE;
+	priv->params.rx_cq_moderation_usec =
+		MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC;
+	priv->params.rx_cq_moderation_pkts =
+		MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_PKTS;
+	priv->params.tx_cq_moderation_usec =
+		MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_USEC;
+	priv->params.tx_cq_moderation_pkts =
+		MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_PKTS;
+	priv->params.min_rx_wqes           =
+		MLX5E_PARAMS_DEFAULT_MIN_RX_WQES;
+	priv->params.rx_hash_log_tbl_sz    =
+		(order_base_2(num_comp_vectors) >
+		 MLX5E_PARAMS_DEFAULT_RX_HASH_LOG_TBL_SZ) ?
+		order_base_2(num_comp_vectors)           :
+		MLX5E_PARAMS_DEFAULT_RX_HASH_LOG_TBL_SZ;
+	priv->params.num_tc                = 1;
+	priv->params.default_vlan_prio     = 0;
+
+	priv->params.lro_en = false && !!MLX5_CAP_ETH(priv->mdev, lro_cap);
+	priv->params.lro_wqe_sz            =
+		MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ;
+
+	priv->mdev                         = mdev;
+	priv->netdev                       = netdev;
+	priv->params.num_channels          = num_comp_vectors;
+	priv->order_base_2_num_channels    = order_base_2(num_comp_vectors);
+	priv->queue_mapping_channel_mask   =
+		roundup_pow_of_two(num_comp_vectors) - 1;
+	priv->num_tc                       = priv->params.num_tc;
+	priv->default_vlan_prio            = priv->params.default_vlan_prio;
+
+	spin_lock_init(&priv->async_events_spinlock);
+	mutex_init(&priv->state_lock);
+
+	INIT_WORK(&priv->update_carrier_work, mlx5e_update_carrier_work);
+	INIT_WORK(&priv->set_rx_mode_work, mlx5e_set_rx_mode_work);
+	INIT_DELAYED_WORK(&priv->update_stats_work, mlx5e_update_stats_work);
+}
+
+static void mlx5e_set_netdev_dev_addr(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+
+	mlx5_query_vport_mac_address(priv->mdev, netdev->dev_addr);
+}
+
+static void mlx5e_build_netdev(struct net_device *netdev)
+{
+	struct mlx5e_priv *priv = netdev_priv(netdev);
+	struct mlx5_core_dev *mdev = priv->mdev;
+
+	SET_NETDEV_DEV(netdev, &mdev->pdev->dev);
+
+	if (priv->num_tc > 1) {
+		mlx5e_netdev_ops.ndo_select_queue = mlx5e_select_queue;
+		mlx5e_netdev_ops.ndo_start_xmit   = mlx5e_xmit_multi_tc;
+	}
+
+	netdev->netdev_ops        = &mlx5e_netdev_ops;
+	netdev->watchdog_timeo    = 15 * HZ;
+
+	netdev->ethtool_ops	  = &mlx5e_ethtool_ops;
+
+	netdev->vlan_features    |= NETIF_F_IP_CSUM;
+	netdev->vlan_features    |= NETIF_F_IPV6_CSUM;
+	netdev->vlan_features    |= NETIF_F_GRO;
+	netdev->vlan_features    |= NETIF_F_TSO;
+	netdev->vlan_features    |= NETIF_F_TSO6;
+	netdev->vlan_features    |= NETIF_F_RXCSUM;
+	netdev->vlan_features    |= NETIF_F_RXHASH;
+
+	if (!!MLX5_CAP_ETH(mdev, lro_cap))
+		netdev->vlan_features    |= NETIF_F_LRO;
+
+	netdev->hw_features       = netdev->vlan_features;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_TX;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_RX;
+	netdev->hw_features      |= NETIF_F_HW_VLAN_CTAG_FILTER;
+
+	netdev->features          = netdev->hw_features;
+	if (!priv->params.lro_en)
+		netdev->features  &= ~NETIF_F_LRO;
+
+	netdev->features         |= NETIF_F_HIGHDMA;
+
+	netdev->priv_flags       |= IFF_UNICAST_FLT;
+
+	mlx5e_set_netdev_dev_addr(netdev);
+}
+
+static int mlx5e_create_mkey(struct mlx5e_priv *priv, u32 pdn,
+			     struct mlx5_core_mr *mr)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	struct mlx5_create_mkey_mbox_in *in;
+	int err;
+
+	in = mlx5_vzalloc(sizeof(*in));
+	if (!in)
+		return -ENOMEM;
+
+	in->seg.flags = MLX5_PERM_LOCAL_WRITE |
+			MLX5_PERM_LOCAL_READ  |
+			MLX5_ACCESS_MODE_PA;
+	in->seg.flags_pd = cpu_to_be32(pdn | MLX5_MKEY_LEN64);
+	in->seg.qpn_mkey7_0 = cpu_to_be32(0xffffff << 8);
+
+	err = mlx5_core_create_mkey(mdev, mr, in, sizeof(*in), NULL, NULL,
+				    NULL);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void *mlx5e_create_netdev(struct mlx5_core_dev *mdev)
+{
+	struct net_device *netdev;
+	struct mlx5e_priv *priv;
+	int ncv = mdev->priv.eq_table.num_comp_vectors;
+	int err;
+
+	if (mlx5e_check_required_hca_cap(mdev))
+		return NULL;
+
+	netdev = alloc_etherdev_mqs(sizeof(struct mlx5e_priv),
+				    roundup_pow_of_two(ncv) * MLX5E_MAX_NUM_TC,
+				    ncv);
+	if (!netdev) {
+		mlx5_core_err(mdev, "alloc_etherdev_mqs() failed\n");
+		return NULL;
+	}
+
+	mlx5e_build_netdev_priv(mdev, netdev, ncv);
+	mlx5e_build_netdev(netdev);
+
+	netif_carrier_off(netdev);
+
+	priv = netdev_priv(netdev);
+
+	err = mlx5_alloc_map_uar(mdev, &priv->cq_uar);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5_alloc_map_uar failed, %d\n",
+			   __func__, err);
+		goto err_free_netdev;
+	}
+
+	err = mlx5_core_alloc_pd(mdev, &priv->pdn);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5_core_alloc_pd failed, %d\n",
+			   __func__, err);
+		goto err_unmap_free_uar;
+	}
+
+	err = mlx5e_create_mkey(priv, priv->pdn, &priv->mr);
+	if (err) {
+		netdev_err(netdev, "%s: mlx5e_create_mkey failed, %d\n",
+			   __func__, err);
+		goto err_dealloc_pd;
+	}
+
+	err = register_netdev(netdev);
+	if (err) {
+		netdev_err(netdev, "%s: register_netdev failed, %d\n",
+			   __func__, err);
+		goto err_destroy_mkey;
+	}
+
+	mlx5e_enable_async_events(priv);
+
+	return priv;
+
+err_destroy_mkey:
+	mlx5_core_destroy_mkey(mdev, &priv->mr);
+
+err_dealloc_pd:
+	mlx5_core_dealloc_pd(mdev, priv->pdn);
+
+err_unmap_free_uar:
+	mlx5_unmap_free_uar(mdev, &priv->cq_uar);
+
+err_free_netdev:
+	free_netdev(netdev);
+
+	return NULL;
+}
+
+static void mlx5e_destroy_netdev(struct mlx5_core_dev *mdev, void *vpriv)
+{
+	struct mlx5e_priv *priv = vpriv;
+	struct net_device *netdev = priv->netdev;
+
+	unregister_netdev(netdev);
+	mlx5_core_destroy_mkey(priv->mdev, &priv->mr);
+	mlx5_core_dealloc_pd(priv->mdev, priv->pdn);
+	mlx5_unmap_free_uar(priv->mdev, &priv->cq_uar);
+	mlx5e_disable_async_events(priv);
+	flush_scheduled_work();
+	free_netdev(netdev);
+}
+
+static void *mlx5e_get_netdev(void *vpriv)
+{
+	struct mlx5e_priv *priv = vpriv;
+
+	return priv->netdev;
+}
+
+static struct mlx5_interface mlx5e_interface = {
+	.add       = mlx5e_create_netdev,
+	.remove    = mlx5e_destroy_netdev,
+	.event     = mlx5e_async_event,
+	.protocol  = MLX5_INTERFACE_PROTOCOL_ETH,
+	.get_dev   = mlx5e_get_netdev,
+};
+
+void mlx5e_init(void)
+{
+	mlx5_register_interface(&mlx5e_interface);
+}
+
+void mlx5e_cleanup(void)
+{
+	mlx5_unregister_interface(&mlx5e_interface);
+}

drivers/net/ethernet/mellanox/mlx5/core/en_rx.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include "en.h"
+
+static inline int mlx5e_alloc_rx_wqe(struct mlx5e_rq *rq,
+				     struct mlx5e_rx_wqe *wqe, u16 ix)
+{
+	struct sk_buff *skb;
+	dma_addr_t dma_addr;
+
+	skb = netdev_alloc_skb(rq->netdev, rq->wqe_sz);
+	if (unlikely(!skb))
+		return -ENOMEM;
+
+	skb_reserve(skb, MLX5E_NET_IP_ALIGN);
+
+	dma_addr = dma_map_single(rq->pdev,
+				  /* hw start padding */
+				  skb->data - MLX5E_NET_IP_ALIGN,
+				  /* hw   end padding */
+				  rq->wqe_sz,
+				  DMA_FROM_DEVICE);
+
+	if (unlikely(dma_mapping_error(rq->pdev, dma_addr)))
+		goto err_free_skb;
+
+	*((dma_addr_t *)skb->cb) = dma_addr;
+	wqe->data.addr = cpu_to_be64(dma_addr + MLX5E_NET_IP_ALIGN);
+
+	rq->skb[ix] = skb;
+
+	return 0;
+
+err_free_skb:
+	dev_kfree_skb(skb);
+
+	return -ENOMEM;
+}
+
+bool mlx5e_post_rx_wqes(struct mlx5e_rq *rq)
+{
+	struct mlx5_wq_ll *wq = &rq->wq;
+
+	if (unlikely(!test_bit(MLX5E_RQ_STATE_POST_WQES_ENABLE, &rq->state)))
+		return false;
+
+	while (!mlx5_wq_ll_is_full(wq)) {
+		struct mlx5e_rx_wqe *wqe = mlx5_wq_ll_get_wqe(wq, wq->head);
+
+		if (unlikely(mlx5e_alloc_rx_wqe(rq, wqe, wq->head)))
+			break;
+
+		mlx5_wq_ll_push(wq, be16_to_cpu(wqe->next.next_wqe_index));
+	}
+
+	/* ensure wqes are visible to device before updating doorbell record */
+	dma_wmb();
+
+	mlx5_wq_ll_update_db_record(wq);
+
+	return !mlx5_wq_ll_is_full(wq);
+}
+
+static void mlx5e_lro_update_hdr(struct sk_buff *skb, struct mlx5_cqe64 *cqe)
+{
+	struct ethhdr	*eth	= (struct ethhdr *)(skb->data);
+	struct iphdr	*ipv4	= (struct iphdr *)(skb->data + ETH_HLEN);
+	struct ipv6hdr	*ipv6	= (struct ipv6hdr *)(skb->data + ETH_HLEN);
+	struct tcphdr	*tcp;
+
+	u8 l4_hdr_type = get_cqe_l4_hdr_type(cqe);
+	int tcp_ack = ((CQE_L4_HDR_TYPE_TCP_ACK_NO_DATA  == l4_hdr_type) ||
+		       (CQE_L4_HDR_TYPE_TCP_ACK_AND_DATA == l4_hdr_type));
+
+	u16 tot_len = be32_to_cpu(cqe->byte_cnt) - ETH_HLEN;
+
+	if (eth->h_proto == htons(ETH_P_IP)) {
+		tcp = (struct tcphdr *)(skb->data + ETH_HLEN +
+					sizeof(struct iphdr));
+		ipv6 = NULL;
+	} else {
+		tcp = (struct tcphdr *)(skb->data + ETH_HLEN +
+					sizeof(struct ipv6hdr));
+		ipv4 = NULL;
+	}
+
+	if (get_cqe_lro_tcppsh(cqe))
+		tcp->psh                = 1;
+
+	if (tcp_ack) {
+		tcp->ack                = 1;
+		tcp->ack_seq            = cqe->lro_ack_seq_num;
+		tcp->window             = cqe->lro_tcp_win;
+	}
+
+	if (ipv4) {
+		ipv4->ttl               = cqe->lro_min_ttl;
+		ipv4->tot_len           = cpu_to_be16(tot_len);
+		ipv4->check             = 0;
+		ipv4->check             = ip_fast_csum((unsigned char *)ipv4,
+						       ipv4->ihl);
+	} else {
+		ipv6->hop_limit         = cqe->lro_min_ttl;
+		ipv6->payload_len       = cpu_to_be16(tot_len -
+						      sizeof(struct ipv6hdr));
+	}
+}
+
+static inline void mlx5e_skb_set_hash(struct mlx5_cqe64 *cqe,
+				      struct sk_buff *skb)
+{
+	u8 cht = cqe->rss_hash_type;
+	int ht = (cht & CQE_RSS_HTYPE_L4) ? PKT_HASH_TYPE_L4 :
+		 (cht & CQE_RSS_HTYPE_IP) ? PKT_HASH_TYPE_L3 :
+					    PKT_HASH_TYPE_NONE;
+	skb_set_hash(skb, be32_to_cpu(cqe->rss_hash_result), ht);
+}
+
+static inline void mlx5e_build_rx_skb(struct mlx5_cqe64 *cqe,
+				      struct mlx5e_rq *rq,
+				      struct sk_buff *skb)
+{
+	struct net_device *netdev = rq->netdev;
+	u32 cqe_bcnt = be32_to_cpu(cqe->byte_cnt);
+	int lro_num_seg;
+
+	skb_put(skb, cqe_bcnt);
+
+	lro_num_seg = be32_to_cpu(cqe->srqn) >> 24;
+	if (lro_num_seg > 1) {
+		mlx5e_lro_update_hdr(skb, cqe);
+		skb_shinfo(skb)->gso_size = MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ;
+		rq->stats.lro_packets++;
+		rq->stats.lro_bytes += cqe_bcnt;
+	}
+
+	if (likely(netdev->features & NETIF_F_RXCSUM) &&
+	    (cqe->hds_ip_ext & CQE_L2_OK) &&
+	    (cqe->hds_ip_ext & CQE_L3_OK) &&
+	    (cqe->hds_ip_ext & CQE_L4_OK)) {
+		skb->ip_summed = CHECKSUM_UNNECESSARY;
+	} else {
+		skb->ip_summed = CHECKSUM_NONE;
+		rq->stats.csum_none++;
+	}
+
+	skb->protocol = eth_type_trans(skb, netdev);
+
+	skb_record_rx_queue(skb, rq->ix);
+
+	if (likely(netdev->features & NETIF_F_RXHASH))
+		mlx5e_skb_set_hash(cqe, skb);
+
+	if (cqe_has_vlan(cqe))
+		__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
+				       be16_to_cpu(cqe->vlan_info));
+}
+
+bool mlx5e_poll_rx_cq(struct mlx5e_cq *cq, int budget)
+{
+	struct mlx5e_rq *rq = cq->sqrq;
+	int i;
+
+	/* avoid accessing cq (dma coherent memory) if not needed */
+	if (!test_and_clear_bit(MLX5E_CQ_HAS_CQES, &cq->flags))
+		return false;
+
+	for (i = 0; i < budget; i++) {
+		struct mlx5e_rx_wqe *wqe;
+		struct mlx5_cqe64 *cqe;
+		struct sk_buff *skb;
+		__be16 wqe_counter_be;
+		u16 wqe_counter;
+
+		cqe = mlx5e_get_cqe(cq);
+		if (!cqe)
+			break;
+
+		wqe_counter_be = cqe->wqe_counter;
+		wqe_counter    = be16_to_cpu(wqe_counter_be);
+		wqe            = mlx5_wq_ll_get_wqe(&rq->wq, wqe_counter);
+		skb            = rq->skb[wqe_counter];
+		rq->skb[wqe_counter] = NULL;
+
+		dma_unmap_single(rq->pdev,
+				 *((dma_addr_t *)skb->cb),
+				 skb_end_offset(skb),
+				 DMA_FROM_DEVICE);
+
+		if (unlikely((cqe->op_own >> 4) != MLX5_CQE_RESP_SEND)) {
+			rq->stats.wqe_err++;
+			dev_kfree_skb(skb);
+			goto wq_ll_pop;
+		}
+
+		mlx5e_build_rx_skb(cqe, rq, skb);
+		rq->stats.packets++;
+		napi_gro_receive(cq->napi, skb);
+
+wq_ll_pop:
+		mlx5_wq_ll_pop(&rq->wq, wqe_counter_be,
+			       &wqe->next.next_wqe_index);
+	}
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	if (i == budget) {
+		set_bit(MLX5E_CQ_HAS_CQES, &cq->flags);
+		return true;
+	}
+
+	return false;
+}

drivers/net/ethernet/mellanox/mlx5/core/en_tx.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/tcp.h>
+#include <linux/if_vlan.h>
+#include "en.h"
+
+static void mlx5e_dma_pop_last_pushed(struct mlx5e_sq *sq, dma_addr_t *addr,
+				      u32 *size)
+{
+	sq->dma_fifo_pc--;
+	*addr = sq->dma_fifo[sq->dma_fifo_pc & sq->dma_fifo_mask].addr;
+	*size = sq->dma_fifo[sq->dma_fifo_pc & sq->dma_fifo_mask].size;
+}
+
+static void mlx5e_dma_unmap_wqe_err(struct mlx5e_sq *sq, struct sk_buff *skb)
+{
+	dma_addr_t addr;
+	u32 size;
+	int i;
+
+	for (i = 0; i < MLX5E_TX_SKB_CB(skb)->num_dma; i++) {
+		mlx5e_dma_pop_last_pushed(sq, &addr, &size);
+		dma_unmap_single(sq->pdev, addr, size, DMA_TO_DEVICE);
+	}
+}
+
+static inline void mlx5e_dma_push(struct mlx5e_sq *sq, dma_addr_t addr,
+				  u32 size)
+{
+	sq->dma_fifo[sq->dma_fifo_pc & sq->dma_fifo_mask].addr = addr;
+	sq->dma_fifo[sq->dma_fifo_pc & sq->dma_fifo_mask].size = size;
+	sq->dma_fifo_pc++;
+}
+
+static inline void mlx5e_dma_get(struct mlx5e_sq *sq, u32 i, dma_addr_t *addr,
+				 u32 *size)
+{
+	*addr = sq->dma_fifo[i & sq->dma_fifo_mask].addr;
+	*size = sq->dma_fifo[i & sq->dma_fifo_mask].size;
+}
+
+u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
+		       void *accel_priv, select_queue_fallback_t fallback)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int channel_ix = fallback(dev, skb);
+	int up = skb_vlan_tag_present(skb)        ?
+		 skb->vlan_tci >> VLAN_PRIO_SHIFT :
+		 priv->default_vlan_prio;
+	int tc = netdev_get_prio_tc_map(dev, up);
+
+	return (tc << priv->order_base_2_num_channels) | channel_ix;
+}
+
+static inline u16 mlx5e_get_inline_hdr_size(struct mlx5e_sq *sq,
+					    struct sk_buff *skb)
+{
+#define MLX5E_MIN_INLINE 16 /* eth header with vlan (w/o next ethertype) */
+	return MLX5E_MIN_INLINE;
+}
+
+static inline void mlx5e_insert_vlan(void *start, struct sk_buff *skb, u16 ihs)
+{
+	struct vlan_ethhdr *vhdr = (struct vlan_ethhdr *)start;
+	int cpy1_sz = 2 * ETH_ALEN;
+	int cpy2_sz = ihs - cpy1_sz - VLAN_HLEN;
+
+	skb_copy_from_linear_data(skb, vhdr, cpy1_sz);
+	skb_pull_inline(skb, cpy1_sz);
+	vhdr->h_vlan_proto = skb->vlan_proto;
+	vhdr->h_vlan_TCI = cpu_to_be16(skb_vlan_tag_get(skb));
+	skb_copy_from_linear_data(skb, &vhdr->h_vlan_encapsulated_proto,
+				  cpy2_sz);
+	skb_pull_inline(skb, cpy2_sz);
+}
+
+static netdev_tx_t mlx5e_sq_xmit(struct mlx5e_sq *sq, struct sk_buff *skb)
+{
+	struct mlx5_wq_cyc       *wq   = &sq->wq;
+
+	u16 pi = sq->pc & wq->sz_m1;
+	struct mlx5e_tx_wqe      *wqe  = mlx5_wq_cyc_get_wqe(wq, pi);
+
+	struct mlx5_wqe_ctrl_seg *cseg = &wqe->ctrl;
+	struct mlx5_wqe_eth_seg  *eseg = &wqe->eth;
+	struct mlx5_wqe_data_seg *dseg;
+
+	u8  opcode = MLX5_OPCODE_SEND;
+	dma_addr_t dma_addr = 0;
+	u16 headlen;
+	u16 ds_cnt;
+	u16 ihs;
+	int i;
+
+	memset(wqe, 0, sizeof(*wqe));
+
+	if (likely(skb->ip_summed == CHECKSUM_PARTIAL))
+		eseg->cs_flags	= MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
+	else
+		sq->stats.csum_offload_none++;
+
+	if (skb_is_gso(skb)) {
+		u32 payload_len;
+		int num_pkts;
+
+		eseg->mss    = cpu_to_be16(skb_shinfo(skb)->gso_size);
+		opcode       = MLX5_OPCODE_LSO;
+		ihs          = skb_transport_offset(skb) + tcp_hdrlen(skb);
+		payload_len  = skb->len - ihs;
+		num_pkts     =    (payload_len / skb_shinfo(skb)->gso_size) +
+				!!(payload_len % skb_shinfo(skb)->gso_size);
+		MLX5E_TX_SKB_CB(skb)->num_bytes = skb->len +
+						  (num_pkts - 1) * ihs;
+		sq->stats.tso_packets++;
+		sq->stats.tso_bytes += payload_len;
+	} else {
+		ihs             = mlx5e_get_inline_hdr_size(sq, skb);
+		MLX5E_TX_SKB_CB(skb)->num_bytes = max_t(unsigned int, skb->len,
+							ETH_ZLEN);
+	}
+
+	if (skb_vlan_tag_present(skb)) {
+		mlx5e_insert_vlan(eseg->inline_hdr_start, skb, ihs);
+	} else {
+		skb_copy_from_linear_data(skb, eseg->inline_hdr_start, ihs);
+		skb_pull_inline(skb, ihs);
+	}
+
+	eseg->inline_hdr_sz	= cpu_to_be16(ihs);
+
+	ds_cnt  = sizeof(*wqe) / MLX5_SEND_WQE_DS;
+	ds_cnt += DIV_ROUND_UP(ihs - sizeof(eseg->inline_hdr_start),
+			       MLX5_SEND_WQE_DS);
+	dseg    = (struct mlx5_wqe_data_seg *)cseg + ds_cnt;
+
+	MLX5E_TX_SKB_CB(skb)->num_dma = 0;
+
+	headlen = skb_headlen(skb);
+	if (headlen) {
+		dma_addr = dma_map_single(sq->pdev, skb->data, headlen,
+					  DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(sq->pdev, dma_addr)))
+			goto dma_unmap_wqe_err;
+
+		dseg->addr       = cpu_to_be64(dma_addr);
+		dseg->lkey       = sq->mkey_be;
+		dseg->byte_count = cpu_to_be32(headlen);
+
+		mlx5e_dma_push(sq, dma_addr, headlen);
+		MLX5E_TX_SKB_CB(skb)->num_dma++;
+
+		dseg++;
+	}
+
+	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
+		struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i];
+		int fsz = skb_frag_size(frag);
+
+		dma_addr = skb_frag_dma_map(sq->pdev, frag, 0, fsz,
+					    DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(sq->pdev, dma_addr)))
+			goto dma_unmap_wqe_err;
+
+		dseg->addr       = cpu_to_be64(dma_addr);
+		dseg->lkey       = sq->mkey_be;
+		dseg->byte_count = cpu_to_be32(fsz);
+
+		mlx5e_dma_push(sq, dma_addr, fsz);
+		MLX5E_TX_SKB_CB(skb)->num_dma++;
+
+		dseg++;
+	}
+
+	ds_cnt += MLX5E_TX_SKB_CB(skb)->num_dma;
+
+	cseg->opmod_idx_opcode	= cpu_to_be32((sq->pc << 8) | opcode);
+	cseg->qpn_ds		= cpu_to_be32((sq->sqn << 8) | ds_cnt);
+	cseg->fm_ce_se		= MLX5_WQE_CTRL_CQ_UPDATE;
+
+	sq->skb[pi] = skb;
+
+	MLX5E_TX_SKB_CB(skb)->num_wqebbs = DIV_ROUND_UP(ds_cnt,
+							MLX5_SEND_WQEBB_NUM_DS);
+	sq->pc += MLX5E_TX_SKB_CB(skb)->num_wqebbs;
+
+	netdev_tx_sent_queue(sq->txq, MLX5E_TX_SKB_CB(skb)->num_bytes);
+
+	if (unlikely(!mlx5e_sq_has_room_for(sq, MLX5_SEND_WQE_MAX_WQEBBS))) {
+		netif_tx_stop_queue(sq->txq);
+		sq->stats.stopped++;
+	}
+
+	if (!skb->xmit_more || netif_xmit_stopped(sq->txq))
+		mlx5e_tx_notify_hw(sq, wqe);
+
+	sq->stats.packets++;
+	return NETDEV_TX_OK;
+
+dma_unmap_wqe_err:
+	sq->stats.dropped++;
+	mlx5e_dma_unmap_wqe_err(sq, skb);
+
+	dev_kfree_skb_any(skb);
+
+	return NETDEV_TX_OK;
+}
+
+netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int ix = skb->queue_mapping;
+	int tc = 0;
+	struct mlx5e_channel *c = priv->channel[ix];
+	struct mlx5e_sq *sq = &c->sq[tc];
+
+	return mlx5e_sq_xmit(sq, skb);
+}
+
+netdev_tx_t mlx5e_xmit_multi_tc(struct sk_buff *skb, struct net_device *dev)
+{
+	struct mlx5e_priv *priv = netdev_priv(dev);
+	int ix = skb->queue_mapping & priv->queue_mapping_channel_mask;
+	int tc = skb->queue_mapping >> priv->order_base_2_num_channels;
+	struct mlx5e_channel *c = priv->channel[ix];
+	struct mlx5e_sq *sq = &c->sq[tc];
+
+	return mlx5e_sq_xmit(sq, skb);
+}
+
+bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq)
+{
+	struct mlx5e_sq *sq;
+	u32 dma_fifo_cc;
+	u32 nbytes;
+	u16 npkts;
+	u16 sqcc;
+	int i;
+
+	/* avoid accessing cq (dma coherent memory) if not needed */
+	if (!test_and_clear_bit(MLX5E_CQ_HAS_CQES, &cq->flags))
+		return false;
+
+	sq = cq->sqrq;
+
+	npkts = 0;
+	nbytes = 0;
+
+	/* sq->cc must be updated only after mlx5_cqwq_update_db_record(),
+	 * otherwise a cq overrun may occur
+	 */
+	sqcc = sq->cc;
+
+	/* avoid dirtying sq cache line every cqe */
+	dma_fifo_cc = sq->dma_fifo_cc;
+
+	for (i = 0; i < MLX5E_TX_CQ_POLL_BUDGET; i++) {
+		struct mlx5_cqe64 *cqe;
+		struct sk_buff *skb;
+		u16 ci;
+		int j;
+
+		cqe = mlx5e_get_cqe(cq);
+		if (!cqe)
+			break;
+
+		ci = sqcc & sq->wq.sz_m1;
+		skb = sq->skb[ci];
+
+		if (unlikely(!skb)) { /* nop */
+			sq->stats.nop++;
+			sqcc++;
+			goto free_skb;
+		}
+
+		for (j = 0; j < MLX5E_TX_SKB_CB(skb)->num_dma; j++) {
+			dma_addr_t addr;
+			u32 size;
+
+			mlx5e_dma_get(sq, dma_fifo_cc, &addr, &size);
+			dma_fifo_cc++;
+			dma_unmap_single(sq->pdev, addr, size, DMA_TO_DEVICE);
+		}
+
+		npkts++;
+		nbytes += MLX5E_TX_SKB_CB(skb)->num_bytes;
+		sqcc += MLX5E_TX_SKB_CB(skb)->num_wqebbs;
+
+free_skb:
+		dev_kfree_skb(skb);
+	}
+
+	mlx5_cqwq_update_db_record(&cq->wq);
+
+	/* ensure cq space is freed before enabling more cqes */
+	wmb();
+
+	sq->dma_fifo_cc = dma_fifo_cc;
+	sq->cc = sqcc;
+
+	netdev_tx_completed_queue(sq->txq, npkts, nbytes);
+
+	if (netif_tx_queue_stopped(sq->txq) &&
+	    mlx5e_sq_has_room_for(sq, MLX5_SEND_WQE_MAX_WQEBBS) &&
+	    likely(test_bit(MLX5E_SQ_STATE_WAKE_TXQ_ENABLE, &sq->state))) {
+				netif_tx_wake_queue(sq->txq);
+				sq->stats.wake++;
+	}
+	if (i == MLX5E_TX_CQ_POLL_BUDGET) {
+		set_bit(MLX5E_CQ_HAS_CQES, &cq->flags);
+		return true;
+	}
+
+	return false;
+}

drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c

+/*
+ * Copyright (c) 2015, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "en.h"
+
+struct mlx5_cqe64 *mlx5e_get_cqe(struct mlx5e_cq *cq)
+{
+	struct mlx5_cqwq *wq = &cq->wq;
+	u32 ci = mlx5_cqwq_get_ci(wq);
+	struct mlx5_cqe64 *cqe = mlx5_cqwq_get_wqe(wq, ci);
+	int cqe_ownership_bit = cqe->op_own & MLX5_CQE_OWNER_MASK;
+	int sw_ownership_val = mlx5_cqwq_get_wrap_cnt(wq) & 1;
+
+	if (cqe_ownership_bit != sw_ownership_val)
+		return NULL;
+
+	mlx5_cqwq_pop(wq);
+
+	/* ensure cqe content is read after cqe ownership bit */
+	rmb();
+
+	return cqe;
+}
+
+int mlx5e_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct mlx5e_channel *c = container_of(napi, struct mlx5e_channel,
+					       napi);
+	bool busy = false;
+	int i;
+
+	clear_bit(MLX5E_CHANNEL_NAPI_SCHED, &c->flags);
+
+	for (i = 0; i < c->num_tc; i++)
+		busy |= mlx5e_poll_tx_cq(&c->sq[i].cq);
+
+	busy |= mlx5e_poll_rx_cq(&c->rq.cq, budget);
+
+	busy |= mlx5e_post_rx_wqes(c->rq.cq.sqrq);
+
+	if (busy)
+		return budget;
+
+	napi_complete(napi);
+
+	/* avoid losing completion event during/after polling cqs */
+	if (test_bit(MLX5E_CHANNEL_NAPI_SCHED, &c->flags)) {
+		napi_schedule(napi);
+		return 0;
+	}
+
+	for (i = 0; i < c->num_tc; i++)
+		mlx5e_cq_arm(&c->sq[i].cq);
+	mlx5e_cq_arm(&c->rq.cq);
+
+	return 0;
+}
+
+void mlx5e_completion_event(struct mlx5_core_cq *mcq)
+{
+	struct mlx5e_cq *cq = container_of(mcq, struct mlx5e_cq, mcq);
+
+	set_bit(MLX5E_CQ_HAS_CQES, &cq->flags);
+	set_bit(MLX5E_CHANNEL_NAPI_SCHED, &cq->channel->flags);
+	barrier();
+	napi_schedule(cq->napi);
+}
+
+void mlx5e_cq_error_event(struct mlx5_core_cq *mcq, enum mlx5_event event)
+{
+	struct mlx5e_cq *cq = container_of(mcq, struct mlx5e_cq, mcq);
+	struct mlx5e_channel *c = cq->channel;
+	struct mlx5e_priv *priv = c->priv;
+	struct net_device *netdev = priv->netdev;
+
+	netdev_err(netdev, "%s: cqn=0x%.6x event=0x%.2x\n",
+		   __func__, mcq->cqn, event);
+}

drivers/net/ethernet/mellanox/mlx5/core/eq.c

@@ -339,15 +339,14 @@ static void init_eq_buf(struct mlx5_eq *eq)
 int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx,
 		       int nent, u64 mask, const char *name, struct mlx5_uar *uar)
 {
-	struct mlx5_eq_table *table = &dev->priv.eq_table;
+	struct mlx5_priv *priv = &dev->priv;
 	struct mlx5_create_eq_mbox_in *in;
 	struct mlx5_create_eq_mbox_out out;
 	int err;
 	int inlen;
 
 	eq->nent = roundup_pow_of_two(nent + MLX5_NUM_SPARE_EQE);
-	err = mlx5_buf_alloc(dev, eq->nent * MLX5_EQE_SIZE, 2 * PAGE_SIZE,
-			     &eq->buf);
+	err = mlx5_buf_alloc(dev, eq->nent * MLX5_EQE_SIZE, &eq->buf);
 	if (err)
 		return err;
 
@@ -378,14 +377,15 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx,
 		goto err_in;
 	}
 
-	snprintf(eq->name, MLX5_MAX_EQ_NAME, "%s@pci:%s",
+	snprintf(priv->irq_info[vecidx].name, MLX5_MAX_IRQ_NAME, "%s@pci:%s",
 		 name, pci_name(dev->pdev));
+
 	eq->eqn = out.eq_number;
 	eq->irqn = vecidx;
 	eq->dev = dev;
 	eq->doorbell = uar->map + MLX5_EQ_DOORBEL_OFFSET;
-	err = request_irq(table->msix_arr[vecidx].vector, mlx5_msix_handler, 0,
-			  eq->name, eq);
+	err = request_irq(priv->msix_arr[vecidx].vector, mlx5_msix_handler, 0,
+			  priv->irq_info[vecidx].name, eq);
 	if (err)
 		goto err_eq;
 
@@ -401,7 +401,7 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx,
 	return 0;
 
 err_irq:
-	free_irq(table->msix_arr[vecidx].vector, eq);
+	free_irq(priv->msix_arr[vecidx].vector, eq);
 
 err_eq:
 	mlx5_cmd_destroy_eq(dev, eq->eqn);
@@ -417,16 +417,15 @@ EXPORT_SYMBOL_GPL(mlx5_create_map_eq);
 
 int mlx5_destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
 {
-	struct mlx5_eq_table *table = &dev->priv.eq_table;
 	int err;
 
 	mlx5_debug_eq_remove(dev, eq);
-	free_irq(table->msix_arr[eq->irqn].vector, eq);
+	free_irq(dev->priv.msix_arr[eq->irqn].vector, eq);
 	err = mlx5_cmd_destroy_eq(dev, eq->eqn);
 	if (err)
 		mlx5_core_warn(dev, "failed to destroy a previously created eq: eqn %d\n",
 			       eq->eqn);
-	synchronize_irq(table->msix_arr[eq->irqn].vector);
+	synchronize_irq(dev->priv.msix_arr[eq->irqn].vector);
 	mlx5_buf_free(dev, &eq->buf);
 
 	return err;
@@ -456,7 +455,7 @@ int mlx5_start_eqs(struct mlx5_core_dev *dev)
 	u32 async_event_mask = MLX5_ASYNC_EVENT_MASK;
 	int err;
 
-	if (dev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG)
+	if (MLX5_CAP_GEN(dev, pg))
 		async_event_mask |= (1ull << MLX5_EVENT_TYPE_PAGE_FAULT);
 
 	err = mlx5_create_map_eq(dev, &table->cmd_eq, MLX5_EQ_VEC_CMD,
@@ -479,7 +478,7 @@ int mlx5_start_eqs(struct mlx5_core_dev *dev)
 
 	err = mlx5_create_map_eq(dev, &table->pages_eq,
 				 MLX5_EQ_VEC_PAGES,
-				 dev->caps.gen.max_vf + 1,
+				 /* TODO: sriov max_vf + */ 1,
 				 1 << MLX5_EVENT_TYPE_PAGE_REQUEST, "mlx5_pages_eq",
 				 &dev->priv.uuari.uars[0]);
 	if (err) {

drivers/net/ethernet/mellanox/mlx5/core/flow_table.c

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/export.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/flow_table.h>
+#include "mlx5_core.h"
+
+struct mlx5_ftg {
+	struct mlx5_flow_table_group    g;
+	u32				id;
+	u32				start_ix;
+};
+
+struct mlx5_flow_table {
+	struct mlx5_core_dev	*dev;
+	u8			level;
+	u8			type;
+	u32			id;
+	struct mutex		mutex; /* sync bitmap alloc */
+	u16			num_groups;
+	struct mlx5_ftg		*group;
+	unsigned long		*bitmap;
+	u32			size;
+};
+
+static int mlx5_set_flow_entry_cmd(struct mlx5_flow_table *ft, u32 group_ix,
+				   u32 flow_index, void *flow_context)
+{
+	u32 out[MLX5_ST_SZ_DW(set_fte_out)];
+	u32 *in;
+	void *in_flow_context;
+	int fcdls =
+		MLX5_GET(flow_context, flow_context, destination_list_size) *
+		MLX5_ST_SZ_BYTES(dest_format_struct);
+	int inlen = MLX5_ST_SZ_BYTES(set_fte_in) + fcdls;
+	int err;
+
+	in = mlx5_vzalloc(inlen);
+	if (!in) {
+		mlx5_core_warn(ft->dev, "failed to allocate inbox\n");
+		return -ENOMEM;
+	}
+
+	MLX5_SET(set_fte_in, in, table_type, ft->type);
+	MLX5_SET(set_fte_in, in, table_id,   ft->id);
+	MLX5_SET(set_fte_in, in, flow_index, flow_index);
+	MLX5_SET(set_fte_in, in, opcode, MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY);
+
+	in_flow_context = MLX5_ADDR_OF(set_fte_in, in, flow_context);
+	memcpy(in_flow_context, flow_context,
+	       MLX5_ST_SZ_BYTES(flow_context) + fcdls);
+
+	MLX5_SET(flow_context, in_flow_context, group_id,
+		 ft->group[group_ix].id);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(ft->dev, in, inlen, out,
+					 sizeof(out));
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5_del_flow_entry_cmd(struct mlx5_flow_table *ft, u32 flow_index)
+{
+	u32 in[MLX5_ST_SZ_DW(delete_fte_in)];
+	u32 out[MLX5_ST_SZ_DW(delete_fte_out)];
+
+	memset(in, 0, sizeof(in));
+	memset(out, 0, sizeof(out));
+
+#define MLX5_SET_DFTEI(p, x, v) MLX5_SET(delete_fte_in, p, x, v)
+	MLX5_SET_DFTEI(in, table_type, ft->type);
+	MLX5_SET_DFTEI(in, table_id,   ft->id);
+	MLX5_SET_DFTEI(in, flow_index, flow_index);
+	MLX5_SET_DFTEI(in, opcode,     MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY);
+
+	mlx5_cmd_exec_check_status(ft->dev, in, sizeof(in), out, sizeof(out));
+}
+
+static void mlx5_destroy_flow_group_cmd(struct mlx5_flow_table *ft, int i)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_group_in)];
+	u32 out[MLX5_ST_SZ_DW(destroy_flow_group_out)];
+
+	memset(in, 0, sizeof(in));
+	memset(out, 0, sizeof(out));
+
+#define MLX5_SET_DFGI(p, x, v) MLX5_SET(destroy_flow_group_in, p, x, v)
+	MLX5_SET_DFGI(in, table_type, ft->type);
+	MLX5_SET_DFGI(in, table_id,   ft->id);
+	MLX5_SET_DFGI(in, opcode, MLX5_CMD_OP_DESTROY_FLOW_GROUP);
+	MLX5_SET_DFGI(in, group_id, ft->group[i].id);
+	mlx5_cmd_exec_check_status(ft->dev, in, sizeof(in), out, sizeof(out));
+}
+
+static int mlx5_create_flow_group_cmd(struct mlx5_flow_table *ft, int i)
+{
+	u32 out[MLX5_ST_SZ_DW(create_flow_group_out)];
+	u32 *in;
+	void *in_match_criteria;
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table_group *g = &ft->group[i].g;
+	u32 start_ix = ft->group[i].start_ix;
+	u32 end_ix = start_ix + (1 << g->log_sz) - 1;
+	int err;
+
+	in = mlx5_vzalloc(inlen);
+	if (!in) {
+		mlx5_core_warn(ft->dev, "failed to allocate inbox\n");
+		return -ENOMEM;
+	}
+	in_match_criteria = MLX5_ADDR_OF(create_flow_group_in, in,
+					 match_criteria);
+
+	memset(out, 0, sizeof(out));
+
+#define MLX5_SET_CFGI(p, x, v) MLX5_SET(create_flow_group_in, p, x, v)
+	MLX5_SET_CFGI(in, table_type,            ft->type);
+	MLX5_SET_CFGI(in, table_id,              ft->id);
+	MLX5_SET_CFGI(in, opcode,                MLX5_CMD_OP_CREATE_FLOW_GROUP);
+	MLX5_SET_CFGI(in, start_flow_index,      start_ix);
+	MLX5_SET_CFGI(in, end_flow_index,        end_ix);
+	MLX5_SET_CFGI(in, match_criteria_enable, g->match_criteria_enable);
+
+	memcpy(in_match_criteria, g->match_criteria,
+	       MLX5_ST_SZ_BYTES(fte_match_param));
+
+	err = mlx5_cmd_exec_check_status(ft->dev, in, inlen, out,
+					 sizeof(out));
+	if (!err)
+		ft->group[i].id = MLX5_GET(create_flow_group_out, out,
+					   group_id);
+
+	kvfree(in);
+
+	return err;
+}
+
+static void mlx5_destroy_flow_table_groups(struct mlx5_flow_table *ft)
+{
+	int i;
+
+	for (i = 0; i < ft->num_groups; i++)
+		mlx5_destroy_flow_group_cmd(ft, i);
+}
+
+static int mlx5_create_flow_table_groups(struct mlx5_flow_table *ft)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < ft->num_groups; i++) {
+		err = mlx5_create_flow_group_cmd(ft, i);
+		if (err)
+			goto err_destroy_flow_table_groups;
+	}
+
+	return 0;
+
+err_destroy_flow_table_groups:
+	for (i--; i >= 0; i--)
+		mlx5_destroy_flow_group_cmd(ft, i);
+
+	return err;
+}
+
+static int mlx5_create_flow_table_cmd(struct mlx5_flow_table *ft)
+{
+	u32 in[MLX5_ST_SZ_DW(create_flow_table_in)];
+	u32 out[MLX5_ST_SZ_DW(create_flow_table_out)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(create_flow_table_in, in, table_type, ft->type);
+	MLX5_SET(create_flow_table_in, in, level,      ft->level);
+	MLX5_SET(create_flow_table_in, in, log_size,   order_base_2(ft->size));
+
+	MLX5_SET(create_flow_table_in, in, opcode,
+		 MLX5_CMD_OP_CREATE_FLOW_TABLE);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(ft->dev, in, sizeof(in), out,
+					 sizeof(out));
+	if (err)
+		return err;
+
+	ft->id = MLX5_GET(create_flow_table_out, out, table_id);
+
+	return 0;
+}
+
+static void mlx5_destroy_flow_table_cmd(struct mlx5_flow_table *ft)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_flow_table_in)];
+	u32 out[MLX5_ST_SZ_DW(destroy_flow_table_out)];
+
+	memset(in, 0, sizeof(in));
+	memset(out, 0, sizeof(out));
+
+#define MLX5_SET_DFTI(p, x, v) MLX5_SET(destroy_flow_table_in, p, x, v)
+	MLX5_SET_DFTI(in, table_type, ft->type);
+	MLX5_SET_DFTI(in, table_id,   ft->id);
+	MLX5_SET_DFTI(in, opcode, MLX5_CMD_OP_DESTROY_FLOW_TABLE);
+
+	mlx5_cmd_exec_check_status(ft->dev, in, sizeof(in), out, sizeof(out));
+}
+
+static int mlx5_find_group(struct mlx5_flow_table *ft, u8 match_criteria_enable,
+			   u32 *match_criteria, int *group_ix)
+{
+	void *mc_outer = MLX5_ADDR_OF(fte_match_param, match_criteria,
+				      outer_headers);
+	void *mc_misc  = MLX5_ADDR_OF(fte_match_param, match_criteria,
+				      misc_parameters);
+	void *mc_inner = MLX5_ADDR_OF(fte_match_param, match_criteria,
+				      inner_headers);
+	int mc_outer_sz = MLX5_ST_SZ_BYTES(fte_match_set_lyr_2_4);
+	int mc_misc_sz  = MLX5_ST_SZ_BYTES(fte_match_set_misc);
+	int mc_inner_sz = MLX5_ST_SZ_BYTES(fte_match_set_lyr_2_4);
+	int i;
+
+	for (i = 0; i < ft->num_groups; i++) {
+		struct mlx5_flow_table_group *g = &ft->group[i].g;
+		void *gmc_outer = MLX5_ADDR_OF(fte_match_param,
+					       g->match_criteria,
+					       outer_headers);
+		void *gmc_misc  = MLX5_ADDR_OF(fte_match_param,
+					       g->match_criteria,
+					       misc_parameters);
+		void *gmc_inner = MLX5_ADDR_OF(fte_match_param,
+					       g->match_criteria,
+					       inner_headers);
+
+		if (g->match_criteria_enable != match_criteria_enable)
+			continue;
+
+		if (match_criteria_enable & MLX5_MATCH_OUTER_HEADERS)
+			if (memcmp(mc_outer, gmc_outer, mc_outer_sz))
+				continue;
+
+		if (match_criteria_enable & MLX5_MATCH_MISC_PARAMETERS)
+			if (memcmp(mc_misc, gmc_misc, mc_misc_sz))
+				continue;
+
+		if (match_criteria_enable & MLX5_MATCH_INNER_HEADERS)
+			if (memcmp(mc_inner, gmc_inner, mc_inner_sz))
+				continue;
+
+		*group_ix = i;
+		return 0;
+	}
+
+	return -EINVAL;
+}
+
+static int alloc_flow_index(struct mlx5_flow_table *ft, int group_ix, u32 *ix)
+{
+	struct mlx5_ftg *g = &ft->group[group_ix];
+	int err = 0;
+
+	mutex_lock(&ft->mutex);
+
+	*ix = find_next_zero_bit(ft->bitmap, ft->size, g->start_ix);
+	if (*ix >= (g->start_ix + (1 << g->g.log_sz)))
+		err = -ENOSPC;
+	else
+		__set_bit(*ix, ft->bitmap);
+
+	mutex_unlock(&ft->mutex);
+
+	return err;
+}
+
+static void mlx5_free_flow_index(struct mlx5_flow_table *ft, u32 ix)
+{
+	__clear_bit(ix, ft->bitmap);
+}
+
+int mlx5_add_flow_table_entry(void *flow_table, u8 match_criteria_enable,
+			      void *match_criteria, void *flow_context,
+			      u32 *flow_index)
+{
+	struct mlx5_flow_table *ft = flow_table;
+	int group_ix;
+	int err;
+
+	err = mlx5_find_group(ft, match_criteria_enable, match_criteria,
+			      &group_ix);
+	if (err) {
+		mlx5_core_warn(ft->dev, "mlx5_find_group failed\n");
+		return err;
+	}
+
+	err = alloc_flow_index(ft, group_ix, flow_index);
+	if (err) {
+		mlx5_core_warn(ft->dev, "alloc_flow_index failed\n");
+		return err;
+	}
+
+	return mlx5_set_flow_entry_cmd(ft, group_ix, *flow_index, flow_context);
+}
+EXPORT_SYMBOL(mlx5_add_flow_table_entry);
+
+void mlx5_del_flow_table_entry(void *flow_table, u32 flow_index)
+{
+	struct mlx5_flow_table *ft = flow_table;
+
+	mlx5_del_flow_entry_cmd(ft, flow_index);
+	mlx5_free_flow_index(ft, flow_index);
+}
+EXPORT_SYMBOL(mlx5_del_flow_table_entry);
+
+void *mlx5_create_flow_table(struct mlx5_core_dev *dev, u8 level, u8 table_type,
+			     u16 num_groups,
+			     struct mlx5_flow_table_group *group)
+{
+	struct mlx5_flow_table *ft;
+	u32 start_ix = 0;
+	u32 ft_size = 0;
+	void *gr;
+	void *bm;
+	int err;
+	int i;
+
+	for (i = 0; i < num_groups; i++)
+		ft_size += (1 << group[i].log_sz);
+
+	ft = kzalloc(sizeof(*ft), GFP_KERNEL);
+	gr = kcalloc(num_groups, sizeof(struct mlx5_ftg), GFP_KERNEL);
+	bm = kcalloc(BITS_TO_LONGS(ft_size), sizeof(uintptr_t), GFP_KERNEL);
+	if (!ft || !gr || !bm)
+		goto err_free_ft;
+
+	ft->group	= gr;
+	ft->bitmap	= bm;
+	ft->num_groups	= num_groups;
+	ft->level	= level;
+	ft->type	= table_type;
+	ft->size	= ft_size;
+	ft->dev		= dev;
+	mutex_init(&ft->mutex);
+
+	for (i = 0; i < ft->num_groups; i++) {
+		memcpy(&ft->group[i].g, &group[i], sizeof(*group));
+		ft->group[i].start_ix = start_ix;
+		start_ix += 1 << group[i].log_sz;
+	}
+
+	err = mlx5_create_flow_table_cmd(ft);
+	if (err)
+		goto err_free_ft;
+
+	err = mlx5_create_flow_table_groups(ft);
+	if (err)
+		goto err_destroy_flow_table_cmd;
+
+	return ft;
+
+err_destroy_flow_table_cmd:
+	mlx5_destroy_flow_table_cmd(ft);
+
+err_free_ft:
+	mlx5_core_warn(dev, "failed to alloc flow table\n");
+	kfree(bm);
+	kfree(gr);
+	kfree(ft);
+
+	return NULL;
+}
+EXPORT_SYMBOL(mlx5_create_flow_table);
+
+void mlx5_destroy_flow_table(void *flow_table)
+{
+	struct mlx5_flow_table *ft = flow_table;
+
+	mlx5_destroy_flow_table_groups(ft);
+	mlx5_destroy_flow_table_cmd(ft);
+	kfree(ft->bitmap);
+	kfree(ft->group);
+	kfree(ft);
+}
+EXPORT_SYMBOL(mlx5_destroy_flow_table);
+
+u32 mlx5_get_flow_table_id(void *flow_table)
+{
+	struct mlx5_flow_table *ft = flow_table;
+
+	return ft->id;
+}
+EXPORT_SYMBOL(mlx5_get_flow_table_id);

drivers/net/ethernet/mellanox/mlx5/core/fw.c

@@ -64,50 +64,74 @@ int mlx5_cmd_query_adapter(struct mlx5_core_dev *dev)
 	return err;
 }
 
-int mlx5_cmd_query_hca_cap(struct mlx5_core_dev *dev, struct mlx5_caps *caps)
+int mlx5_query_hca_caps(struct mlx5_core_dev *dev)
 {
-	return mlx5_core_get_caps(dev, caps, HCA_CAP_OPMOD_GET_CUR);
-}
-
-int mlx5_query_odp_caps(struct mlx5_core_dev *dev, struct mlx5_odp_caps *caps)
-{
-	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)];
-	int out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
-	void *out;
 	int err;
 
-	if (!(dev->caps.gen.flags & MLX5_DEV_CAP_FLAG_ON_DMND_PG))
-		return -ENOTSUPP;
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL, HCA_CAP_OPMOD_GET_CUR);
+	if (err)
+		return err;
 
-	memset(in, 0, sizeof(in));
-	out = kzalloc(out_sz, GFP_KERNEL);
-	if (!out)
-		return -ENOMEM;
-	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
-	MLX5_SET(query_hca_cap_in, in, op_mod, HCA_CAP_OPMOD_GET_ODP_CUR);
-	err = mlx5_cmd_exec(dev, in, sizeof(in), out, out_sz);
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL, HCA_CAP_OPMOD_GET_MAX);
 	if (err)
-		goto out;
+		return err;
 
-	err = mlx5_cmd_status_to_err_v2(out);
-	if (err) {
-		mlx5_core_warn(dev, "query cur hca ODP caps failed, %d\n", err);
-		goto out;
+	if (MLX5_CAP_GEN(dev, eth_net_offloads)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ETHERNET_OFFLOADS,
+					 HCA_CAP_OPMOD_GET_CUR);
+		if (err)
+			return err;
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ETHERNET_OFFLOADS,
+					 HCA_CAP_OPMOD_GET_MAX);
+		if (err)
+			return err;
 	}
 
-	memcpy(caps, MLX5_ADDR_OF(query_hca_cap_out, out, capability_struct),
-	       sizeof(*caps));
+	if (MLX5_CAP_GEN(dev, pg)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ODP,
+					 HCA_CAP_OPMOD_GET_CUR);
+		if (err)
+			return err;
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ODP,
+					 HCA_CAP_OPMOD_GET_MAX);
+		if (err)
+			return err;
+	}
 
-	mlx5_core_dbg(dev, "on-demand paging capabilities:\nrc: %08x\nuc: %08x\nud: %08x\n",
-		be32_to_cpu(caps->per_transport_caps.rc_odp_caps),
-		be32_to_cpu(caps->per_transport_caps.uc_odp_caps),
-		be32_to_cpu(caps->per_transport_caps.ud_odp_caps));
+	if (MLX5_CAP_GEN(dev, atomic)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ATOMIC,
+					 HCA_CAP_OPMOD_GET_CUR);
+		if (err)
+			return err;
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ATOMIC,
+					 HCA_CAP_OPMOD_GET_MAX);
+		if (err)
+			return err;
+	}
 
-out:
-	kfree(out);
-	return err;
+	if (MLX5_CAP_GEN(dev, roce)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ROCE,
+					 HCA_CAP_OPMOD_GET_CUR);
+		if (err)
+			return err;
+		err = mlx5_core_get_caps(dev, MLX5_CAP_ROCE,
+					 HCA_CAP_OPMOD_GET_MAX);
+		if (err)
+			return err;
+	}
+
+	if (MLX5_CAP_GEN(dev, nic_flow_table)) {
+		err = mlx5_core_get_caps(dev, MLX5_CAP_FLOW_TABLE,
+					 HCA_CAP_OPMOD_GET_CUR);
+		if (err)
+			return err;
+		err = mlx5_core_get_caps(dev, MLX5_CAP_FLOW_TABLE,
+					 HCA_CAP_OPMOD_GET_MAX);
+		if (err)
+			return err;
+	}
+	return 0;
 }
-EXPORT_SYMBOL(mlx5_query_odp_caps);
 
 int mlx5_cmd_init_hca(struct mlx5_core_dev *dev)
 {

drivers/net/ethernet/mellanox/mlx5/core/main.c

@@ -38,6 +38,7 @@
 #include <linux/dma-mapping.h>
 #include <linux/slab.h>
 #include <linux/io-mapping.h>
+#include <linux/interrupt.h>
 #include <linux/mlx5/driver.h>
 #include <linux/mlx5/cq.h>
 #include <linux/mlx5/qp.h>
@@ -47,10 +48,6 @@
 #include <linux/mlx5/mlx5_ifc.h>
 #include "mlx5_core.h"
 
-#define DRIVER_NAME "mlx5_core"
-#define DRIVER_VERSION "3.0"
-#define DRIVER_RELDATE  "January 2015"
-
 MODULE_AUTHOR("Eli Cohen <eli@mellanox.com>");
 MODULE_DESCRIPTION("Mellanox Connect-IB, ConnectX-4 core driver");
 MODULE_LICENSE("Dual BSD/GPL");
@@ -208,24 +205,28 @@ static void release_bar(struct pci_dev *pdev)
 
 static int mlx5_enable_msix(struct mlx5_core_dev *dev)
 {
-	struct mlx5_eq_table *table = &dev->priv.eq_table;
-	int num_eqs = 1 << dev->caps.gen.log_max_eq;
+	struct mlx5_priv *priv = &dev->priv;
+	struct mlx5_eq_table *table = &priv->eq_table;
+	int num_eqs = 1 << MLX5_CAP_GEN(dev, log_max_eq);
 	int nvec;
 	int i;
 
-	nvec = dev->caps.gen.num_ports * num_online_cpus() + MLX5_EQ_VEC_COMP_BASE;
+	nvec = MLX5_CAP_GEN(dev, num_ports) * num_online_cpus() +
+	       MLX5_EQ_VEC_COMP_BASE;
 	nvec = min_t(int, nvec, num_eqs);
 	if (nvec <= MLX5_EQ_VEC_COMP_BASE)
 		return -ENOMEM;
 
-	table->msix_arr = kzalloc(nvec * sizeof(*table->msix_arr), GFP_KERNEL);
-	if (!table->msix_arr)
-		return -ENOMEM;
+	priv->msix_arr = kcalloc(nvec, sizeof(*priv->msix_arr), GFP_KERNEL);
+
+	priv->irq_info = kcalloc(nvec, sizeof(*priv->irq_info), GFP_KERNEL);
+	if (!priv->msix_arr || !priv->irq_info)
+		goto err_free_msix;
 
 	for (i = 0; i < nvec; i++)
-		table->msix_arr[i].entry = i;
+		priv->msix_arr[i].entry = i;
 
-	nvec = pci_enable_msix_range(dev->pdev, table->msix_arr,
+	nvec = pci_enable_msix_range(dev->pdev, priv->msix_arr,
 				     MLX5_EQ_VEC_COMP_BASE + 1, nvec);
 	if (nvec < 0)
 		return nvec;
@@ -233,14 +234,20 @@ static int mlx5_enable_msix(struct mlx5_core_dev *dev)
 	table->num_comp_vectors = nvec - MLX5_EQ_VEC_COMP_BASE;
 
 	return 0;
+
+err_free_msix:
+	kfree(priv->irq_info);
+	kfree(priv->msix_arr);
+	return -ENOMEM;
 }
 
 static void mlx5_disable_msix(struct mlx5_core_dev *dev)
 {
-	struct mlx5_eq_table *table = &dev->priv.eq_table;
+	struct mlx5_priv *priv = &dev->priv;
 
 	pci_disable_msix(dev->pdev);
-	kfree(table->msix_arr);
+	kfree(priv->irq_info);
+	kfree(priv->msix_arr);
 }
 
 struct mlx5_reg_host_endianess {
@@ -277,98 +284,28 @@ static u16 to_fw_pkey_sz(u32 size)
 	}
 }
 
-/* selectively copy writable fields clearing any reserved area
- */
-static void copy_rw_fields(void *to, struct mlx5_caps *from)
+static u16 to_sw_pkey_sz(int pkey_sz)
 {
-	__be64 *flags_off = (__be64 *)MLX5_ADDR_OF(cmd_hca_cap, to, reserved_22);
-	u64 v64;
-
-	MLX5_SET(cmd_hca_cap, to, log_max_qp, from->gen.log_max_qp);
-	MLX5_SET(cmd_hca_cap, to, log_max_ra_req_qp, from->gen.log_max_ra_req_qp);
-	MLX5_SET(cmd_hca_cap, to, log_max_ra_res_qp, from->gen.log_max_ra_res_qp);
-	MLX5_SET(cmd_hca_cap, to, pkey_table_size, from->gen.pkey_table_size);
-	MLX5_SET(cmd_hca_cap, to, pkey_table_size, to_fw_pkey_sz(from->gen.pkey_table_size));
-	MLX5_SET(cmd_hca_cap, to, log_uar_page_sz, PAGE_SHIFT - 12);
-	v64 = from->gen.flags & MLX5_CAP_BITS_RW_MASK;
-	*flags_off = cpu_to_be64(v64);
-}
-
-static u16 get_pkey_table_size(int pkey)
-{
-	if (pkey > MLX5_MAX_LOG_PKEY_TABLE)
+	if (pkey_sz > MLX5_MAX_LOG_PKEY_TABLE)
 		return 0;
 
-	return MLX5_MIN_PKEY_TABLE_SIZE << pkey;
+	return MLX5_MIN_PKEY_TABLE_SIZE << pkey_sz;
 }
 
-static void fw2drv_caps(struct mlx5_caps *caps, void *out)
-{
-	struct mlx5_general_caps *gen = &caps->gen;
-
-	gen->max_srq_wqes = 1 << MLX5_GET_PR(cmd_hca_cap, out, log_max_srq_sz);
-	gen->max_wqes = 1 << MLX5_GET_PR(cmd_hca_cap, out, log_max_qp_sz);
-	gen->log_max_qp = MLX5_GET_PR(cmd_hca_cap, out, log_max_qp);
-	gen->log_max_strq = MLX5_GET_PR(cmd_hca_cap, out, log_max_strq_sz);
-	gen->log_max_srq = MLX5_GET_PR(cmd_hca_cap, out, log_max_srqs);
-	gen->max_cqes = 1 << MLX5_GET_PR(cmd_hca_cap, out, log_max_cq_sz);
-	gen->log_max_cq = MLX5_GET_PR(cmd_hca_cap, out, log_max_cq);
-	gen->max_eqes = 1 << MLX5_GET_PR(cmd_hca_cap, out, log_max_eq_sz);
-	gen->log_max_mkey = MLX5_GET_PR(cmd_hca_cap, out, log_max_mkey);
-	gen->log_max_eq = MLX5_GET_PR(cmd_hca_cap, out, log_max_eq);
-	gen->max_indirection = MLX5_GET_PR(cmd_hca_cap, out, max_indirection);
-	gen->log_max_mrw_sz = MLX5_GET_PR(cmd_hca_cap, out, log_max_mrw_sz);
-	gen->log_max_bsf_list_size = MLX5_GET_PR(cmd_hca_cap, out, log_max_bsf_list_size);
-	gen->log_max_klm_list_size = MLX5_GET_PR(cmd_hca_cap, out, log_max_klm_list_size);
-	gen->log_max_ra_req_dc = MLX5_GET_PR(cmd_hca_cap, out, log_max_ra_req_dc);
-	gen->log_max_ra_res_dc = MLX5_GET_PR(cmd_hca_cap, out, log_max_ra_res_dc);
-	gen->log_max_ra_req_qp = MLX5_GET_PR(cmd_hca_cap, out, log_max_ra_req_qp);
-	gen->log_max_ra_res_qp = MLX5_GET_PR(cmd_hca_cap, out, log_max_ra_res_qp);
-	gen->max_qp_counters = MLX5_GET_PR(cmd_hca_cap, out, max_qp_cnt);
-	gen->pkey_table_size = get_pkey_table_size(MLX5_GET_PR(cmd_hca_cap, out, pkey_table_size));
-	gen->local_ca_ack_delay = MLX5_GET_PR(cmd_hca_cap, out, local_ca_ack_delay);
-	gen->num_ports = MLX5_GET_PR(cmd_hca_cap, out, num_ports);
-	gen->log_max_msg = MLX5_GET_PR(cmd_hca_cap, out, log_max_msg);
-	gen->stat_rate_support = MLX5_GET_PR(cmd_hca_cap, out, stat_rate_support);
-	gen->flags = be64_to_cpu(*(__be64 *)MLX5_ADDR_OF(cmd_hca_cap, out, reserved_22));
-	pr_debug("flags = 0x%llx\n", gen->flags);
-	gen->uar_sz = MLX5_GET_PR(cmd_hca_cap, out, uar_sz);
-	gen->min_log_pg_sz = MLX5_GET_PR(cmd_hca_cap, out, log_pg_sz);
-	gen->bf_reg_size = MLX5_GET_PR(cmd_hca_cap, out, bf);
-	gen->bf_reg_size = 1 << MLX5_GET_PR(cmd_hca_cap, out, log_bf_reg_size);
-	gen->max_sq_desc_sz = MLX5_GET_PR(cmd_hca_cap, out, max_wqe_sz_sq);
-	gen->max_rq_desc_sz = MLX5_GET_PR(cmd_hca_cap, out, max_wqe_sz_rq);
-	gen->max_dc_sq_desc_sz = MLX5_GET_PR(cmd_hca_cap, out, max_wqe_sz_sq_dc);
-	gen->max_qp_mcg = MLX5_GET_PR(cmd_hca_cap, out, max_qp_mcg);
-	gen->log_max_pd = MLX5_GET_PR(cmd_hca_cap, out, log_max_pd);
-	gen->log_max_xrcd = MLX5_GET_PR(cmd_hca_cap, out, log_max_xrcd);
-	gen->log_uar_page_sz = MLX5_GET_PR(cmd_hca_cap, out, log_uar_page_sz);
-}
-
-static const char *caps_opmod_str(u16 opmod)
-{
-	switch (opmod) {
-	case HCA_CAP_OPMOD_GET_MAX:
-		return "GET_MAX";
-	case HCA_CAP_OPMOD_GET_CUR:
-		return "GET_CUR";
-	default:
-		return "Invalid";
-	}
-}
-
-int mlx5_core_get_caps(struct mlx5_core_dev *dev, struct mlx5_caps *caps,
-		       u16 opmod)
+int mlx5_core_get_caps(struct mlx5_core_dev *dev, enum mlx5_cap_type cap_type,
+		       enum mlx5_cap_mode cap_mode)
 {
 	u8 in[MLX5_ST_SZ_BYTES(query_hca_cap_in)];
 	int out_sz = MLX5_ST_SZ_BYTES(query_hca_cap_out);
-	void *out;
+	void *out, *hca_caps;
+	u16 opmod = (cap_type << 1) | (cap_mode & 0x01);
 	int err;
 
 	memset(in, 0, sizeof(in));
 	out = kzalloc(out_sz, GFP_KERNEL);
 	if (!out)
 		return -ENOMEM;
+
 	MLX5_SET(query_hca_cap_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_CAP);
 	MLX5_SET(query_hca_cap_in, in, op_mod, opmod);
 	err = mlx5_cmd_exec(dev, in, sizeof(in), out, out_sz);
@@ -377,12 +314,30 @@ int mlx5_core_get_caps(struct mlx5_core_dev *dev, struct mlx5_caps *caps,
 
 	err = mlx5_cmd_status_to_err_v2(out);
 	if (err) {
-		mlx5_core_warn(dev, "query max hca cap failed, %d\n", err);
+		mlx5_core_warn(dev,
+			       "QUERY_HCA_CAP : type(%x) opmode(%x) Failed(%d)\n",
+			       cap_type, cap_mode, err);
 		goto query_ex;
 	}
-	mlx5_core_dbg(dev, "%s\n", caps_opmod_str(opmod));
-	fw2drv_caps(caps, MLX5_ADDR_OF(query_hca_cap_out, out, capability_struct));
 
+	hca_caps =  MLX5_ADDR_OF(query_hca_cap_out, out, capability);
+
+	switch (cap_mode) {
+	case HCA_CAP_OPMOD_GET_MAX:
+		memcpy(dev->hca_caps_max[cap_type], hca_caps,
+		       MLX5_UN_SZ_BYTES(hca_cap_union));
+		break;
+	case HCA_CAP_OPMOD_GET_CUR:
+		memcpy(dev->hca_caps_cur[cap_type], hca_caps,
+		       MLX5_UN_SZ_BYTES(hca_cap_union));
+		break;
+	default:
+		mlx5_core_warn(dev,
+			       "Tried to query dev cap type(%x) with wrong opmode(%x)\n",
+			       cap_type, cap_mode);
+		err = -EINVAL;
+		break;
+	}
 query_ex:
 	kfree(out);
 	return err;
@@ -409,49 +364,45 @@ static int handle_hca_cap(struct mlx5_core_dev *dev)
 {
 	void *set_ctx = NULL;
 	struct mlx5_profile *prof = dev->profile;
-	struct mlx5_caps *cur_caps = NULL;
-	struct mlx5_caps *max_caps = NULL;
 	int err = -ENOMEM;
 	int set_sz = MLX5_ST_SZ_BYTES(set_hca_cap_in);
+	void *set_hca_cap;
 
 	set_ctx = kzalloc(set_sz, GFP_KERNEL);
 	if (!set_ctx)
 		goto query_ex;
 
-	max_caps = kzalloc(sizeof(*max_caps), GFP_KERNEL);
-	if (!max_caps)
-		goto query_ex;
-
-	cur_caps = kzalloc(sizeof(*cur_caps), GFP_KERNEL);
-	if (!cur_caps)
-		goto query_ex;
-
-	err = mlx5_core_get_caps(dev, max_caps, HCA_CAP_OPMOD_GET_MAX);
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL, HCA_CAP_OPMOD_GET_MAX);
 	if (err)
 		goto query_ex;
 
-	err = mlx5_core_get_caps(dev, cur_caps, HCA_CAP_OPMOD_GET_CUR);
+	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL, HCA_CAP_OPMOD_GET_CUR);
 	if (err)
 		goto query_ex;
 
+	set_hca_cap = MLX5_ADDR_OF(set_hca_cap_in, set_ctx,
+				   capability);
+	memcpy(set_hca_cap, dev->hca_caps_cur[MLX5_CAP_GENERAL],
+	       MLX5_ST_SZ_BYTES(cmd_hca_cap));
+
+	mlx5_core_dbg(dev, "Current Pkey table size %d Setting new size %d\n",
+		      to_sw_pkey_sz(MLX5_CAP_GEN(dev, pkey_table_size)),
+		      128);
 	/* we limit the size of the pkey table to 128 entries for now */
-	cur_caps->gen.pkey_table_size = 128;
+	MLX5_SET(cmd_hca_cap, set_hca_cap, pkey_table_size,
+		 to_fw_pkey_sz(128));
 
 	if (prof->mask & MLX5_PROF_MASK_QP_SIZE)
-		cur_caps->gen.log_max_qp = prof->log_max_qp;
+		MLX5_SET(cmd_hca_cap, set_hca_cap, log_max_qp,
+			 prof->log_max_qp);
 
-	/* disable checksum */
-	cur_caps->gen.flags &= ~MLX5_DEV_CAP_FLAG_CMDIF_CSUM;
+	/* disable cmdif checksum */
+	MLX5_SET(cmd_hca_cap, set_hca_cap, cmdif_checksum, 0);
 
-	copy_rw_fields(MLX5_ADDR_OF(set_hca_cap_in, set_ctx, hca_capability_struct),
-		       cur_caps);
 	err = set_caps(dev, set_ctx, set_sz);
 
 query_ex:
-	kfree(cur_caps);
-	kfree(max_caps);
 	kfree(set_ctx);
-
 	return err;
 }
 
@@ -507,6 +458,77 @@ static int mlx5_core_disable_hca(struct mlx5_core_dev *dev)
 	return 0;
 }
 
+static int mlx5_irq_set_affinity_hint(struct mlx5_core_dev *mdev, int i)
+{
+	struct mlx5_priv *priv  = &mdev->priv;
+	struct msix_entry *msix = priv->msix_arr;
+	int irq                 = msix[i + MLX5_EQ_VEC_COMP_BASE].vector;
+	int numa_node           = dev_to_node(&mdev->pdev->dev);
+	int err;
+
+	if (!zalloc_cpumask_var(&priv->irq_info[i].mask, GFP_KERNEL)) {
+		mlx5_core_warn(mdev, "zalloc_cpumask_var failed");
+		return -ENOMEM;
+	}
+
+	err = cpumask_set_cpu_local_first(i, numa_node, priv->irq_info[i].mask);
+	if (err) {
+		mlx5_core_warn(mdev, "cpumask_set_cpu_local_first failed");
+		goto err_clear_mask;
+	}
+
+	err = irq_set_affinity_hint(irq, priv->irq_info[i].mask);
+	if (err) {
+		mlx5_core_warn(mdev, "irq_set_affinity_hint failed,irq 0x%.4x",
+			       irq);
+		goto err_clear_mask;
+	}
+
+	return 0;
+
+err_clear_mask:
+	free_cpumask_var(priv->irq_info[i].mask);
+	return err;
+}
+
+static void mlx5_irq_clear_affinity_hint(struct mlx5_core_dev *mdev, int i)
+{
+	struct mlx5_priv *priv  = &mdev->priv;
+	struct msix_entry *msix = priv->msix_arr;
+	int irq                 = msix[i + MLX5_EQ_VEC_COMP_BASE].vector;
+
+	irq_set_affinity_hint(irq, NULL);
+	free_cpumask_var(priv->irq_info[i].mask);
+}
+
+static int mlx5_irq_set_affinity_hints(struct mlx5_core_dev *mdev)
+{
+	int err;
+	int i;
+
+	for (i = 0; i < mdev->priv.eq_table.num_comp_vectors; i++) {
+		err = mlx5_irq_set_affinity_hint(mdev, i);
+		if (err)
+			goto err_out;
+	}
+
+	return 0;
+
+err_out:
+	for (i--; i >= 0; i--)
+		mlx5_irq_clear_affinity_hint(mdev, i);
+
+	return err;
+}
+
+static void mlx5_irq_clear_affinity_hints(struct mlx5_core_dev *mdev)
+{
+	int i;
+
+	for (i = 0; i < mdev->priv.eq_table.num_comp_vectors; i++)
+		mlx5_irq_clear_affinity_hint(mdev, i);
+}
+
 int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn, int *irqn)
 {
 	struct mlx5_eq_table *table = &dev->priv.eq_table;
@@ -549,7 +571,7 @@ static void free_comp_eqs(struct mlx5_core_dev *dev)
 static int alloc_comp_eqs(struct mlx5_core_dev *dev)
 {
 	struct mlx5_eq_table *table = &dev->priv.eq_table;
-	char name[MLX5_MAX_EQ_NAME];
+	char name[MLX5_MAX_IRQ_NAME];
 	struct mlx5_eq *eq;
 	int ncomp_vec;
 	int nent;
@@ -566,7 +588,7 @@ static int alloc_comp_eqs(struct mlx5_core_dev *dev)
 			goto clean;
 		}
 
-		snprintf(name, MLX5_MAX_EQ_NAME, "mlx5_comp%d", i);
+		snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_comp%d", i);
 		err = mlx5_create_map_eq(dev, eq,
 					 i + MLX5_EQ_VEC_COMP_BASE, nent, 0,
 					 name, &dev->priv.uuari.uars[0]);
@@ -588,6 +610,61 @@ static int alloc_comp_eqs(struct mlx5_core_dev *dev)
 	return err;
 }
 
+#ifdef CONFIG_MLX5_CORE_EN
+static int mlx5_core_set_issi(struct mlx5_core_dev *dev)
+{
+	u32 query_in[MLX5_ST_SZ_DW(query_issi_in)];
+	u32 query_out[MLX5_ST_SZ_DW(query_issi_out)];
+	u32 set_in[MLX5_ST_SZ_DW(set_issi_in)];
+	u32 set_out[MLX5_ST_SZ_DW(set_issi_out)];
+	int err;
+	u32 sup_issi;
+
+	memset(query_in, 0, sizeof(query_in));
+	memset(query_out, 0, sizeof(query_out));
+
+	MLX5_SET(query_issi_in, query_in, opcode, MLX5_CMD_OP_QUERY_ISSI);
+
+	err = mlx5_cmd_exec_check_status(dev, query_in, sizeof(query_in),
+					 query_out, sizeof(query_out));
+	if (err) {
+		if (((struct mlx5_outbox_hdr *)query_out)->status ==
+		    MLX5_CMD_STAT_BAD_OP_ERR) {
+			pr_debug("Only ISSI 0 is supported\n");
+			return 0;
+		}
+
+		pr_err("failed to query ISSI\n");
+		return err;
+	}
+
+	sup_issi = MLX5_GET(query_issi_out, query_out, supported_issi_dw0);
+
+	if (sup_issi & (1 << 1)) {
+		memset(set_in, 0, sizeof(set_in));
+		memset(set_out, 0, sizeof(set_out));
+
+		MLX5_SET(set_issi_in, set_in, opcode, MLX5_CMD_OP_SET_ISSI);
+		MLX5_SET(set_issi_in, set_in, current_issi, 1);
+
+		err = mlx5_cmd_exec_check_status(dev, set_in, sizeof(set_in),
+						 set_out, sizeof(set_out));
+		if (err) {
+			pr_err("failed to set ISSI=1\n");
+			return err;
+		}
+
+		dev->issi = 1;
+
+		return 0;
+	} else if (sup_issi & (1 << 0)) {
+		return 0;
+	}
+
+	return -ENOTSUPP;
+}
+#endif
+
 static int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev)
 {
 	struct mlx5_priv *priv = &dev->priv;
@@ -650,6 +727,14 @@ static int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev)
 		goto err_pagealloc_cleanup;
 	}
 
+#ifdef CONFIG_MLX5_CORE_EN
+	err = mlx5_core_set_issi(dev);
+	if (err) {
+		dev_err(&pdev->dev, "failed to set issi\n");
+		goto err_disable_hca;
+	}
+#endif
+
 	err = mlx5_satisfy_startup_pages(dev, 1);
 	if (err) {
 		dev_err(&pdev->dev, "failed to allocate boot pages\n");
@@ -688,7 +773,7 @@ static int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev)
 
 	mlx5_start_health_poll(dev);
 
-	err = mlx5_cmd_query_hca_cap(dev, &dev->caps);
+	err = mlx5_query_hca_caps(dev);
 	if (err) {
 		dev_err(&pdev->dev, "query hca failed\n");
 		goto err_stop_poll;
@@ -730,6 +815,12 @@ static int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev)
 		goto err_stop_eqs;
 	}
 
+	err = mlx5_irq_set_affinity_hints(dev);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to alloc affinity hint cpumask\n");
+		goto err_free_comp_eqs;
+	}
+
 	MLX5_INIT_DOORBELL_LOCK(&priv->cq_uar_lock);
 
 	mlx5_init_cq_table(dev);
@@ -739,6 +830,9 @@ static int mlx5_dev_init(struct mlx5_core_dev *dev, struct pci_dev *pdev)
 
 	return 0;
 
+err_free_comp_eqs:
+	free_comp_eqs(dev);
+
 err_stop_eqs:
 	mlx5_stop_eqs(dev);
 
@@ -793,6 +887,7 @@ static void mlx5_dev_cleanup(struct mlx5_core_dev *dev)
 	mlx5_cleanup_srq_table(dev);
 	mlx5_cleanup_qp_table(dev);
 	mlx5_cleanup_cq_table(dev);
+	mlx5_irq_clear_affinity_hints(dev);
 	free_comp_eqs(dev);
 	mlx5_stop_eqs(dev);
 	mlx5_free_uuars(dev, &priv->uuari);
@@ -1048,6 +1143,10 @@ static int __init init(void)
 	if (err)
 		goto err_health;
 
+#ifdef CONFIG_MLX5_CORE_EN
+	mlx5e_init();
+#endif
+
 	return 0;
 
 err_health:
@@ -1060,6 +1159,9 @@ static int __init init(void)
 
 static void __exit cleanup(void)
 {
+#ifdef CONFIG_MLX5_CORE_EN
+	mlx5e_cleanup();
+#endif
 	pci_unregister_driver(&mlx5_core_driver);
 	mlx5_health_cleanup();
 	destroy_workqueue(mlx5_core_wq);

drivers/net/ethernet/mellanox/mlx5/core/mcg.c

@@ -91,7 +91,7 @@ int mlx5_core_detach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn)
 
 	memset(&in, 0, sizeof(in));
 	memset(&out, 0, sizeof(out));
-	in.hdr.opcode = cpu_to_be16(MLX5_CMD_OP_DETACH_FROM_MCG);
+	in.hdr.opcode = cpu_to_be16(MLX5_CMD_OP_DETTACH_FROM_MCG);
 	memcpy(in.gid, mgid, sizeof(*mgid));
 	in.qpn = cpu_to_be32(qpn);
 	err = mlx5_cmd_exec(dev, &in, sizeof(in), &out, sizeof(out));

drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h

 /*
- * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -37,6 +37,10 @@
 #include <linux/kernel.h>
 #include <linux/sched.h>
 
+#define DRIVER_NAME "mlx5_core"
+#define DRIVER_VERSION "3.0-1"
+#define DRIVER_RELDATE  "January 2015"
+
 extern int mlx5_core_debug_mask;
 
 #define mlx5_core_dbg(dev, format, ...)					\
@@ -65,11 +69,20 @@ enum {
 	MLX5_CMD_TIME, /* print command execution time */
 };
 
+static inline int mlx5_cmd_exec_check_status(struct mlx5_core_dev *dev, u32 *in,
+					     int in_size, u32 *out,
+					     int out_size)
+{
+	mlx5_cmd_exec(dev, in, in_size, out, out_size);
+	return mlx5_cmd_status_to_err((struct mlx5_outbox_hdr *)out);
+}
 
-int mlx5_cmd_query_hca_cap(struct mlx5_core_dev *dev,
-			   struct mlx5_caps *caps);
+int mlx5_query_hca_caps(struct mlx5_core_dev *dev);
 int mlx5_cmd_query_adapter(struct mlx5_core_dev *dev);
 int mlx5_cmd_init_hca(struct mlx5_core_dev *dev);
 int mlx5_cmd_teardown_hca(struct mlx5_core_dev *dev);
 
+void mlx5e_init(void);
+void mlx5e_cleanup(void);
+
 #endif /* __MLX5_CORE_H__ */

drivers/net/ethernet/mellanox/mlx5/core/port.c

@@ -102,3 +102,165 @@ int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps)
 	return err;
 }
 EXPORT_SYMBOL_GPL(mlx5_set_port_caps);
+
+int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys,
+			 int ptys_size, int proto_mask)
+{
+	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+	MLX5_SET(ptys_reg, in, local_port, 1);
+	MLX5_SET(ptys_reg, in, proto_mask, proto_mask);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), ptys,
+				   ptys_size, MLX5_REG_PTYS, 0, 0);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_ptys);
+
+int mlx5_query_port_proto_cap(struct mlx5_core_dev *dev,
+			      u32 *proto_cap, int proto_mask)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	err = mlx5_query_port_ptys(dev, out, sizeof(out), proto_mask);
+	if (err)
+		return err;
+
+	if (proto_mask == MLX5_PTYS_EN)
+		*proto_cap = MLX5_GET(ptys_reg, out, eth_proto_capability);
+	else
+		*proto_cap = MLX5_GET(ptys_reg, out, ib_proto_capability);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_proto_cap);
+
+int mlx5_query_port_proto_admin(struct mlx5_core_dev *dev,
+				u32 *proto_admin, int proto_mask)
+{
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	err = mlx5_query_port_ptys(dev, out, sizeof(out), proto_mask);
+	if (err)
+		return err;
+
+	if (proto_mask == MLX5_PTYS_EN)
+		*proto_admin = MLX5_GET(ptys_reg, out, eth_proto_admin);
+	else
+		*proto_admin = MLX5_GET(ptys_reg, out, ib_proto_admin);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_proto_admin);
+
+int mlx5_set_port_proto(struct mlx5_core_dev *dev, u32 proto_admin,
+			int proto_mask)
+{
+	u32 in[MLX5_ST_SZ_DW(ptys_reg)];
+	u32 out[MLX5_ST_SZ_DW(ptys_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(ptys_reg, in, local_port, 1);
+	MLX5_SET(ptys_reg, in, proto_mask, proto_mask);
+	if (proto_mask == MLX5_PTYS_EN)
+		MLX5_SET(ptys_reg, in, eth_proto_admin, proto_admin);
+	else
+		MLX5_SET(ptys_reg, in, ib_proto_admin, proto_admin);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PTYS, 0, 1);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_proto);
+
+int mlx5_set_port_status(struct mlx5_core_dev *dev,
+			 enum mlx5_port_status status)
+{
+	u32 in[MLX5_ST_SZ_DW(paos_reg)];
+	u32 out[MLX5_ST_SZ_DW(paos_reg)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(paos_reg, in, admin_status, status);
+	MLX5_SET(paos_reg, in, ase, 1);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out,
+				    sizeof(out), MLX5_REG_PAOS, 0, 1);
+}
+
+int mlx5_query_port_status(struct mlx5_core_dev *dev, u8 *status)
+{
+	u32 in[MLX5_ST_SZ_DW(paos_reg)];
+	u32 out[MLX5_ST_SZ_DW(paos_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PAOS, 0, 0);
+	if (err)
+		return err;
+
+	*status = MLX5_GET(paos_reg, out, oper_status);
+	return err;
+}
+
+static int mlx5_query_port_mtu(struct mlx5_core_dev *dev,
+			       int *admin_mtu, int *max_mtu, int *oper_mtu)
+{
+	u32 in[MLX5_ST_SZ_DW(pmtu_reg)];
+	u32 out[MLX5_ST_SZ_DW(pmtu_reg)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(pmtu_reg, in, local_port, 1);
+
+	err = mlx5_core_access_reg(dev, in, sizeof(in), out,
+				   sizeof(out), MLX5_REG_PMTU, 0, 0);
+	if (err)
+		return err;
+
+	if (max_mtu)
+		*max_mtu  = MLX5_GET(pmtu_reg, out, max_mtu);
+	if (oper_mtu)
+		*oper_mtu = MLX5_GET(pmtu_reg, out, oper_mtu);
+	if (admin_mtu)
+		*admin_mtu = MLX5_GET(pmtu_reg, out, admin_mtu);
+
+	return 0;
+}
+
+int mlx5_set_port_mtu(struct mlx5_core_dev *dev, int mtu)
+{
+	u32 in[MLX5_ST_SZ_DW(pmtu_reg)];
+	u32 out[MLX5_ST_SZ_DW(pmtu_reg)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(pmtu_reg, in, admin_mtu, mtu);
+	MLX5_SET(pmtu_reg, in, local_port, 1);
+
+	return mlx5_core_access_reg(dev, in, sizeof(in), out, sizeof(out),
+				    MLX5_REG_PMTU, 0, 1);
+}
+EXPORT_SYMBOL_GPL(mlx5_set_port_mtu);
+
+int mlx5_query_port_max_mtu(struct mlx5_core_dev *dev, int *max_mtu)
+{
+	return mlx5_query_port_mtu(dev, NULL, max_mtu, NULL);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_max_mtu);
+
+int mlx5_query_port_oper_mtu(struct mlx5_core_dev *dev, int *oper_mtu)
+{
+	return mlx5_query_port_mtu(dev, NULL, NULL, oper_mtu);
+}
+EXPORT_SYMBOL_GPL(mlx5_query_port_oper_mtu);

drivers/net/ethernet/mellanox/mlx5/core/transobj.c

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+#include "transobj.h"
+
+int mlx5_create_rq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *rqn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_rq_out)];
+	int err;
+
+	MLX5_SET(create_rq_in, in, opcode, MLX5_CMD_OP_CREATE_RQ);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*rqn = MLX5_GET(create_rq_out, out, rqn);
+
+	return err;
+}
+
+int mlx5_modify_rq(struct mlx5_core_dev *dev, u32 rqn, u32 *in, int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_rq_out)];
+
+	MLX5_SET(modify_rq_in, in, rqn, rqn);
+	MLX5_SET(modify_rq_in, in, opcode, MLX5_CMD_OP_MODIFY_RQ);
+
+	memset(out, 0, sizeof(out));
+	return mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+}
+
+void mlx5_destroy_rq(struct mlx5_core_dev *dev, u32 rqn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_rq_in)];
+	u32 out[MLX5_ST_SZ_DW(destroy_rq_out)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(destroy_rq_in, in, opcode, MLX5_CMD_OP_DESTROY_RQ);
+	MLX5_SET(destroy_rq_in, in, rqn, rqn);
+
+	mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_create_sq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *sqn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_sq_out)];
+	int err;
+
+	MLX5_SET(create_sq_in, in, opcode, MLX5_CMD_OP_CREATE_SQ);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*sqn = MLX5_GET(create_sq_out, out, sqn);
+
+	return err;
+}
+
+int mlx5_modify_sq(struct mlx5_core_dev *dev, u32 sqn, u32 *in, int inlen)
+{
+	u32 out[MLX5_ST_SZ_DW(modify_sq_out)];
+
+	MLX5_SET(modify_sq_in, in, sqn, sqn);
+	MLX5_SET(modify_sq_in, in, opcode, MLX5_CMD_OP_MODIFY_SQ);
+
+	memset(out, 0, sizeof(out));
+	return mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+}
+
+void mlx5_destroy_sq(struct mlx5_core_dev *dev, u32 sqn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_sq_in)];
+	u32 out[MLX5_ST_SZ_DW(destroy_sq_out)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(destroy_sq_in, in, opcode, MLX5_CMD_OP_DESTROY_SQ);
+	MLX5_SET(destroy_sq_in, in, sqn, sqn);
+
+	mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_create_tir(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *tirn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_tir_out)];
+	int err;
+
+	MLX5_SET(create_tir_in, in, opcode, MLX5_CMD_OP_CREATE_TIR);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*tirn = MLX5_GET(create_tir_out, out, tirn);
+
+	return err;
+}
+
+void mlx5_destroy_tir(struct mlx5_core_dev *dev, u32 tirn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_tir_out)];
+	u32 out[MLX5_ST_SZ_DW(destroy_tir_out)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(destroy_tir_in, in, opcode, MLX5_CMD_OP_DESTROY_TIR);
+	MLX5_SET(destroy_tir_in, in, tirn, tirn);
+
+	mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, sizeof(out));
+}
+
+int mlx5_create_tis(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *tisn)
+{
+	u32 out[MLX5_ST_SZ_DW(create_tis_out)];
+	int err;
+
+	MLX5_SET(create_tis_in, in, opcode, MLX5_CMD_OP_CREATE_TIS);
+
+	memset(out, 0, sizeof(out));
+	err = mlx5_cmd_exec_check_status(dev, in, inlen, out, sizeof(out));
+	if (!err)
+		*tisn = MLX5_GET(create_tis_out, out, tisn);
+
+	return err;
+}
+
+void mlx5_destroy_tis(struct mlx5_core_dev *dev, u32 tisn)
+{
+	u32 in[MLX5_ST_SZ_DW(destroy_tis_out)];
+	u32 out[MLX5_ST_SZ_DW(destroy_tis_out)];
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(destroy_tis_in, in, opcode, MLX5_CMD_OP_DESTROY_TIS);
+	MLX5_SET(destroy_tis_in, in, tisn, tisn);
+
+	mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, sizeof(out));
+}

drivers/net/ethernet/mellanox/mlx5/core/transobj.h

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __TRANSOBJ_H__
+#define __TRANSOBJ_H__
+
+int mlx5_create_rq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *rqn);
+int mlx5_modify_rq(struct mlx5_core_dev *dev, u32 rqn, u32 *in, int inlen);
+void mlx5_destroy_rq(struct mlx5_core_dev *dev, u32 rqn);
+int mlx5_create_sq(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *sqn);
+int mlx5_modify_sq(struct mlx5_core_dev *dev, u32 sqn, u32 *in, int inlen);
+void mlx5_destroy_sq(struct mlx5_core_dev *dev, u32 sqn);
+int mlx5_create_tir(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *tirn);
+void mlx5_destroy_tir(struct mlx5_core_dev *dev, u32 tirn);
+int mlx5_create_tis(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *tisn);
+void mlx5_destroy_tis(struct mlx5_core_dev *dev, u32 tisn);
+
+#endif /* __TRANSOBJ_H__ */

drivers/net/ethernet/mellanox/mlx5/core/uar.c

@@ -175,12 +175,13 @@ int mlx5_alloc_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari)
 	for (i = 0; i < tot_uuars; i++) {
 		bf = &uuari->bfs[i];
 
-		bf->buf_size = dev->caps.gen.bf_reg_size / 2;
+		bf->buf_size = (1 << MLX5_CAP_GEN(dev, log_bf_reg_size)) / 2;
 		bf->uar = &uuari->uars[i / MLX5_BF_REGS_PER_PAGE];
 		bf->regreg = uuari->uars[i / MLX5_BF_REGS_PER_PAGE].map;
 		bf->reg = NULL; /* Add WC support */
-		bf->offset = (i % MLX5_BF_REGS_PER_PAGE) * dev->caps.gen.bf_reg_size +
-			MLX5_BF_OFFSET;
+		bf->offset = (i % MLX5_BF_REGS_PER_PAGE) *
+			     (1 << MLX5_CAP_GEN(dev, log_bf_reg_size)) +
+			     MLX5_BF_OFFSET;
 		bf->need_lock = need_uuar_lock(i);
 		spin_lock_init(&bf->lock);
 		spin_lock_init(&bf->lock32);
@@ -223,3 +224,40 @@ int mlx5_free_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari)
 
 	return 0;
 }
+
+int mlx5_alloc_map_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar)
+{
+	phys_addr_t pfn;
+	phys_addr_t uar_bar_start;
+	int err;
+
+	err = mlx5_cmd_alloc_uar(mdev, &uar->index);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_cmd_alloc_uar() failed, %d\n", err);
+		return err;
+	}
+
+	uar_bar_start = pci_resource_start(mdev->pdev, 0);
+	pfn           = (uar_bar_start >> PAGE_SHIFT) + uar->index;
+	uar->map      = ioremap(pfn << PAGE_SHIFT, PAGE_SIZE);
+	if (!uar->map) {
+		mlx5_core_warn(mdev, "ioremap() failed, %d\n", err);
+		err = -ENOMEM;
+		goto err_free_uar;
+	}
+
+	return 0;
+
+err_free_uar:
+	mlx5_cmd_free_uar(mdev, uar->index);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_alloc_map_uar);
+
+void mlx5_unmap_free_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar)
+{
+	iounmap(uar->map);
+	mlx5_cmd_free_uar(mdev, uar->index);
+}
+EXPORT_SYMBOL(mlx5_unmap_free_uar);

drivers/net/ethernet/mellanox/mlx5/core/vport.c

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/export.h>
+#include <linux/etherdevice.h>
+#include <linux/mlx5/driver.h>
+#include "vport.h"
+#include "mlx5_core.h"
+
+u8 mlx5_query_vport_state(struct mlx5_core_dev *mdev, u8 opmod)
+{
+	u32 in[MLX5_ST_SZ_DW(query_vport_state_in)];
+	u32 out[MLX5_ST_SZ_DW(query_vport_state_out)];
+	int err;
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(query_vport_state_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_VPORT_STATE);
+	MLX5_SET(query_vport_state_in, in, op_mod, opmod);
+
+	err = mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out,
+					 sizeof(out));
+	if (err)
+		mlx5_core_warn(mdev, "MLX5_CMD_OP_QUERY_VPORT_STATE failed\n");
+
+	return MLX5_GET(query_vport_state_out, out, state);
+}
+
+void mlx5_query_vport_mac_address(struct mlx5_core_dev *mdev, u8 *addr)
+{
+	u32  in[MLX5_ST_SZ_DW(query_nic_vport_context_in)];
+	u32 *out;
+	int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out);
+	u8 *out_addr;
+
+	out = mlx5_vzalloc(outlen);
+	if (!out)
+		return;
+
+	out_addr = MLX5_ADDR_OF(query_nic_vport_context_out, out,
+				nic_vport_context.permanent_address);
+
+	memset(in, 0, sizeof(in));
+
+	MLX5_SET(query_nic_vport_context_in, in, opcode,
+		 MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT);
+
+	memset(out, 0, outlen);
+	mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, outlen);
+
+	ether_addr_copy(addr, &out_addr[2]);
+
+	kvfree(out);
+}

drivers/net/ethernet/mellanox/mlx5/core/vport.h

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_VPORT_H__
+#define __MLX5_VPORT_H__
+
+#include <linux/mlx5/driver.h>
+
+u8 mlx5_query_vport_state(struct mlx5_core_dev *mdev, u8 opmod);
+void mlx5_query_vport_mac_address(struct mlx5_core_dev *mdev, u8 *addr);
+
+#endif /* __MLX5_VPORT_H__ */

drivers/net/ethernet/mellanox/mlx5/core/wq.c

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/driver.h>
+#include "wq.h"
+#include "mlx5_core.h"
+
+u32 mlx5_wq_cyc_get_size(struct mlx5_wq_cyc *wq)
+{
+	return (u32)wq->sz_m1 + 1;
+}
+
+u32 mlx5_cqwq_get_size(struct mlx5_cqwq *wq)
+{
+	return wq->sz_m1 + 1;
+}
+
+u32 mlx5_wq_ll_get_size(struct mlx5_wq_ll *wq)
+{
+	return (u32)wq->sz_m1 + 1;
+}
+
+static u32 mlx5_wq_cyc_get_byte_size(struct mlx5_wq_cyc *wq)
+{
+	return mlx5_wq_cyc_get_size(wq) << wq->log_stride;
+}
+
+static u32 mlx5_cqwq_get_byte_size(struct mlx5_cqwq *wq)
+{
+	return mlx5_cqwq_get_size(wq) << wq->log_stride;
+}
+
+static u32 mlx5_wq_ll_get_byte_size(struct mlx5_wq_ll *wq)
+{
+	return mlx5_wq_ll_get_size(wq) << wq->log_stride;
+}
+
+int mlx5_wq_cyc_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		       void *wqc, struct mlx5_wq_cyc *wq,
+		       struct mlx5_wq_ctrl *wq_ctrl)
+{
+	int err;
+
+	wq->log_stride = MLX5_GET(wq, wqc, log_wq_stride);
+	wq->sz_m1 = (1 << MLX5_GET(wq, wqc, log_wq_sz)) - 1;
+
+	err = mlx5_db_alloc(mdev, &wq_ctrl->db);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc() failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5_buf_alloc(mdev, mlx5_wq_cyc_get_byte_size(wq), &wq_ctrl->buf);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_buf_alloc() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	wq->buf = wq_ctrl->buf.direct.buf;
+	wq->db  = wq_ctrl->db.db;
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+int mlx5_cqwq_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		     void *cqc, struct mlx5_cqwq *wq,
+		     struct mlx5_wq_ctrl *wq_ctrl)
+{
+	int err;
+
+	wq->log_stride = 6 + MLX5_GET(cqc, cqc, cqe_sz);
+	wq->log_sz = MLX5_GET(cqc, cqc, log_cq_size);
+	wq->sz_m1 = (1 << wq->log_sz) - 1;
+
+	err = mlx5_db_alloc(mdev, &wq_ctrl->db);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc() failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5_buf_alloc(mdev, mlx5_cqwq_get_byte_size(wq), &wq_ctrl->buf);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_buf_alloc() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	wq->buf = wq_ctrl->buf.direct.buf;
+	wq->db  = wq_ctrl->db.db;
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+int mlx5_wq_ll_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *wqc, struct mlx5_wq_ll *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl)
+{
+	struct mlx5_wqe_srq_next_seg *next_seg;
+	int err;
+	int i;
+
+	wq->log_stride = MLX5_GET(wq, wqc, log_wq_stride);
+	wq->sz_m1 = (1 << MLX5_GET(wq, wqc, log_wq_sz)) - 1;
+
+	err = mlx5_db_alloc(mdev, &wq_ctrl->db);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_db_alloc() failed, %d\n", err);
+		return err;
+	}
+
+	err = mlx5_buf_alloc(mdev, mlx5_wq_ll_get_byte_size(wq), &wq_ctrl->buf);
+	if (err) {
+		mlx5_core_warn(mdev, "mlx5_buf_alloc() failed, %d\n", err);
+		goto err_db_free;
+	}
+
+	wq->buf = wq_ctrl->buf.direct.buf;
+	wq->db  = wq_ctrl->db.db;
+
+	for (i = 0; i < wq->sz_m1; i++) {
+		next_seg = mlx5_wq_ll_get_wqe(wq, i);
+		next_seg->next_wqe_index = cpu_to_be16(i + 1);
+	}
+	next_seg = mlx5_wq_ll_get_wqe(wq, i);
+	wq->tail_next = &next_seg->next_wqe_index;
+
+	wq_ctrl->mdev = mdev;
+
+	return 0;
+
+err_db_free:
+	mlx5_db_free(mdev, &wq_ctrl->db);
+
+	return err;
+}
+
+void mlx5_wq_destroy(struct mlx5_wq_ctrl *wq_ctrl)
+{
+	mlx5_buf_free(wq_ctrl->mdev, &wq_ctrl->buf);
+	mlx5_db_free(wq_ctrl->mdev, &wq_ctrl->db);
+}

drivers/net/ethernet/mellanox/mlx5/core/wq.h

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __MLX5_WQ_H__
+#define __MLX5_WQ_H__
+
+#include <linux/mlx5/mlx5_ifc.h>
+
+struct mlx5_wq_param {
+	int		linear;
+	int		numa;
+};
+
+struct mlx5_wq_ctrl {
+	struct mlx5_core_dev	*mdev;
+	struct mlx5_buf		buf;
+	struct mlx5_db		db;
+};
+
+struct mlx5_wq_cyc {
+	void			*buf;
+	__be32			*db;
+	u16			sz_m1;
+	u8			log_stride;
+};
+
+struct mlx5_cqwq {
+	void			*buf;
+	__be32			*db;
+	u32			sz_m1;
+	u32			cc; /* consumer counter */
+	u8			log_sz;
+	u8			log_stride;
+};
+
+struct mlx5_wq_ll {
+	void			*buf;
+	__be32			*db;
+	__be16			*tail_next;
+	u16			sz_m1;
+	u16			head;
+	u16			wqe_ctr;
+	u16			cur_sz;
+	u8			log_stride;
+};
+
+int mlx5_wq_cyc_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		       void *wqc, struct mlx5_wq_cyc *wq,
+		       struct mlx5_wq_ctrl *wq_ctrl);
+u32 mlx5_wq_cyc_get_size(struct mlx5_wq_cyc *wq);
+
+int mlx5_cqwq_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		     void *cqc, struct mlx5_cqwq *wq,
+		     struct mlx5_wq_ctrl *wq_ctrl);
+u32 mlx5_cqwq_get_size(struct mlx5_cqwq *wq);
+
+int mlx5_wq_ll_create(struct mlx5_core_dev *mdev, struct mlx5_wq_param *param,
+		      void *wqc, struct mlx5_wq_ll *wq,
+		      struct mlx5_wq_ctrl *wq_ctrl);
+u32 mlx5_wq_ll_get_size(struct mlx5_wq_ll *wq);
+
+void mlx5_wq_destroy(struct mlx5_wq_ctrl *wq_ctrl);
+
+static inline u16 mlx5_wq_cyc_ctr2ix(struct mlx5_wq_cyc *wq, u16 ctr)
+{
+	return ctr & wq->sz_m1;
+}
+
+static inline void *mlx5_wq_cyc_get_wqe(struct mlx5_wq_cyc *wq, u16 ix)
+{
+	return wq->buf + (ix << wq->log_stride);
+}
+
+static inline int mlx5_wq_cyc_cc_bigger(u16 cc1, u16 cc2)
+{
+	int equal   = (cc1 == cc2);
+	int smaller = 0x8000 & (cc1 - cc2);
+
+	return !equal && !smaller;
+}
+
+static inline u32 mlx5_cqwq_get_ci(struct mlx5_cqwq *wq)
+{
+	return wq->cc & wq->sz_m1;
+}
+
+static inline void *mlx5_cqwq_get_wqe(struct mlx5_cqwq *wq, u32 ix)
+{
+	return wq->buf + (ix << wq->log_stride);
+}
+
+static inline u32 mlx5_cqwq_get_wrap_cnt(struct mlx5_cqwq *wq)
+{
+	return wq->cc >> wq->log_sz;
+}
+
+static inline void mlx5_cqwq_pop(struct mlx5_cqwq *wq)
+{
+	wq->cc++;
+}
+
+static inline void mlx5_cqwq_update_db_record(struct mlx5_cqwq *wq)
+{
+	*wq->db = cpu_to_be32(wq->cc & 0xffffff);
+}
+
+static inline int mlx5_wq_ll_is_full(struct mlx5_wq_ll *wq)
+{
+	return wq->cur_sz == wq->sz_m1;
+}
+
+static inline int mlx5_wq_ll_is_empty(struct mlx5_wq_ll *wq)
+{
+	return !wq->cur_sz;
+}
+
+static inline void *mlx5_wq_ll_get_wqe(struct mlx5_wq_ll *wq, u16 ix)
+{
+	return wq->buf + (ix << wq->log_stride);
+}
+
+static inline void mlx5_wq_ll_push(struct mlx5_wq_ll *wq, u16 head_next)
+{
+	wq->head = head_next;
+	wq->wqe_ctr++;
+	wq->cur_sz++;
+}
+
+static inline void mlx5_wq_ll_pop(struct mlx5_wq_ll *wq, __be16 ix,
+				  __be16 *next_tail_next)
+{
+	*wq->tail_next = ix;
+	wq->tail_next = next_tail_next;
+	wq->cur_sz--;
+}
+
+static inline void mlx5_wq_ll_update_db_record(struct mlx5_wq_ll *wq)
+{
+	*wq->db = cpu_to_be32(wq->wqe_ctr);
+}
+
+#endif /* __MLX5_WQ_H__ */

include/linux/mlx5/cq.h

@@ -169,6 +169,9 @@ int mlx5_core_query_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
 		       struct mlx5_query_cq_mbox_out *out);
 int mlx5_core_modify_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
 			struct mlx5_modify_cq_mbox_in *in, int in_sz);
+int mlx5_core_modify_cq_moderation(struct mlx5_core_dev *dev,
+				   struct mlx5_core_cq *cq, u16 cq_period,
+				   u16 cq_max_count);
 int mlx5_debug_cq_add(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq);
 void mlx5_debug_cq_remove(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq);
 

include/linux/mlx5/device.h

@@ -35,6 +35,7 @@
 
 #include <linux/types.h>
 #include <rdma/ib_verbs.h>
+#include <linux/mlx5/mlx5_ifc.h>
 
 #if defined(__LITTLE_ENDIAN)
 #define MLX5_SET_HOST_ENDIANNESS	0
@@ -58,6 +59,8 @@
 #define MLX5_FLD_SZ_BYTES(typ, fld) (__mlx5_bit_sz(typ, fld) / 8)
 #define MLX5_ST_SZ_BYTES(typ) (sizeof(struct mlx5_ifc_##typ##_bits) / 8)
 #define MLX5_ST_SZ_DW(typ) (sizeof(struct mlx5_ifc_##typ##_bits) / 32)
+#define MLX5_UN_SZ_BYTES(typ) (sizeof(union mlx5_ifc_##typ##_bits) / 8)
+#define MLX5_UN_SZ_DW(typ) (sizeof(union mlx5_ifc_##typ##_bits) / 32)
 #define MLX5_BYTE_OFF(typ, fld) (__mlx5_bit_off(typ, fld) / 8)
 #define MLX5_ADDR_OF(typ, p, fld) ((char *)(p) + MLX5_BYTE_OFF(typ, fld))
 
@@ -70,6 +73,14 @@
 		     << __mlx5_dw_bit_off(typ, fld))); \
 } while (0)
 
+#define MLX5_SET_TO_ONES(typ, p, fld) do { \
+	BUILD_BUG_ON(__mlx5_st_sz_bits(typ) % 32);             \
+	*((__be32 *)(p) + __mlx5_dw_off(typ, fld)) = \
+	cpu_to_be32((be32_to_cpu(*((__be32 *)(p) + __mlx5_dw_off(typ, fld))) & \
+		     (~__mlx5_dw_mask(typ, fld))) | ((__mlx5_mask(typ, fld)) \
+		     << __mlx5_dw_bit_off(typ, fld))); \
+} while (0)
+
 #define MLX5_GET(typ, p, fld) ((be32_to_cpu(*((__be32 *)(p) +\
 __mlx5_dw_off(typ, fld))) >> __mlx5_dw_bit_off(typ, fld)) & \
 __mlx5_mask(typ, fld))
@@ -264,6 +275,7 @@ enum {
 	MLX5_OPCODE_RDMA_WRITE_IMM	= 0x09,
 	MLX5_OPCODE_SEND		= 0x0a,
 	MLX5_OPCODE_SEND_IMM		= 0x0b,
+	MLX5_OPCODE_LSO			= 0x0e,
 	MLX5_OPCODE_RDMA_READ		= 0x10,
 	MLX5_OPCODE_ATOMIC_CS		= 0x11,
 	MLX5_OPCODE_ATOMIC_FA		= 0x12,
@@ -312,13 +324,6 @@ enum {
 	MLX5_CAP_OFF_CMDIF_CSUM		= 46,
 };
 
-enum {
-	HCA_CAP_OPMOD_GET_MAX	= 0,
-	HCA_CAP_OPMOD_GET_CUR	= 1,
-	HCA_CAP_OPMOD_GET_ODP_MAX = 4,
-	HCA_CAP_OPMOD_GET_ODP_CUR = 5
-};
-
 struct mlx5_inbox_hdr {
 	__be16		opcode;
 	u8		rsvd[4];
@@ -541,6 +546,10 @@ struct mlx5_cmd_prot_block {
 	u8		sig;
 };
 
+enum {
+	MLX5_CQE_SYND_FLUSHED_IN_ERROR = 5,
+};
+
 struct mlx5_err_cqe {
 	u8	rsvd0[32];
 	__be32	srqn;
@@ -554,13 +563,22 @@ struct mlx5_err_cqe {
 };
 
 struct mlx5_cqe64 {
-	u8		rsvd0[17];
+	u8		rsvd0[4];
+	u8		lro_tcppsh_abort_dupack;
+	u8		lro_min_ttl;
+	__be16		lro_tcp_win;
+	__be32		lro_ack_seq_num;
+	__be32		rss_hash_result;
+	u8		rss_hash_type;
 	u8		ml_path;
-	u8		rsvd20[4];
+	u8		rsvd20[2];
+	__be16		check_sum;
 	__be16		slid;
 	__be32		flags_rqpn;
-	u8		rsvd28[4];
-	__be32		srqn;
+	u8		hds_ip_ext;
+	u8		l4_hdr_type_etc;
+	__be16		vlan_info;
+	__be32		srqn; /* [31:24]: lro_num_seg, [23:0]: srqn */
 	__be32		imm_inval_pkey;
 	u8		rsvd40[4];
 	__be32		byte_cnt;
@@ -571,6 +589,40 @@ struct mlx5_cqe64 {
 	u8		op_own;
 };
 
+static inline int get_cqe_lro_tcppsh(struct mlx5_cqe64 *cqe)
+{
+	return (cqe->lro_tcppsh_abort_dupack >> 6) & 1;
+}
+
+static inline u8 get_cqe_l4_hdr_type(struct mlx5_cqe64 *cqe)
+{
+	return (cqe->l4_hdr_type_etc >> 4) & 0x7;
+}
+
+static inline int cqe_has_vlan(struct mlx5_cqe64 *cqe)
+{
+	return !!(cqe->l4_hdr_type_etc & 0x1);
+}
+
+enum {
+	CQE_L4_HDR_TYPE_NONE			= 0x0,
+	CQE_L4_HDR_TYPE_TCP_NO_ACK		= 0x1,
+	CQE_L4_HDR_TYPE_UDP			= 0x2,
+	CQE_L4_HDR_TYPE_TCP_ACK_NO_DATA		= 0x3,
+	CQE_L4_HDR_TYPE_TCP_ACK_AND_DATA	= 0x4,
+};
+
+enum {
+	CQE_RSS_HTYPE_IP	= 0x3 << 6,
+	CQE_RSS_HTYPE_L4	= 0x3 << 2,
+};
+
+enum {
+	CQE_L2_OK	= 1 << 0,
+	CQE_L3_OK	= 1 << 1,
+	CQE_L4_OK	= 1 << 2,
+};
+
 struct mlx5_sig_err_cqe {
 	u8		rsvd0[16];
 	__be32		expected_trans_sig;
@@ -996,4 +1048,128 @@ struct mlx5_destroy_psv_out {
 	u8                      rsvd[8];
 };
 
+#define MLX5_CMD_OP_MAX 0x920
+
+enum {
+	VPORT_STATE_DOWN		= 0x0,
+	VPORT_STATE_UP			= 0x1,
+};
+
+enum {
+	MLX5_L3_PROT_TYPE_IPV4		= 0,
+	MLX5_L3_PROT_TYPE_IPV6		= 1,
+};
+
+enum {
+	MLX5_L4_PROT_TYPE_TCP		= 0,
+	MLX5_L4_PROT_TYPE_UDP		= 1,
+};
+
+enum {
+	MLX5_HASH_FIELD_SEL_SRC_IP	= 1 << 0,
+	MLX5_HASH_FIELD_SEL_DST_IP	= 1 << 1,
+	MLX5_HASH_FIELD_SEL_L4_SPORT	= 1 << 2,
+	MLX5_HASH_FIELD_SEL_L4_DPORT	= 1 << 3,
+	MLX5_HASH_FIELD_SEL_IPSEC_SPI	= 1 << 4,
+};
+
+enum {
+	MLX5_MATCH_OUTER_HEADERS	= 1 << 0,
+	MLX5_MATCH_MISC_PARAMETERS	= 1 << 1,
+	MLX5_MATCH_INNER_HEADERS	= 1 << 2,
+
+};
+
+enum {
+	MLX5_FLOW_TABLE_TYPE_NIC_RCV	= 0,
+	MLX5_FLOW_TABLE_TYPE_ESWITCH	= 4,
+};
+
+enum {
+	MLX5_FLOW_CONTEXT_DEST_TYPE_VPORT	= 0,
+	MLX5_FLOW_CONTEXT_DEST_TYPE_FLOW_TABLE	= 1,
+	MLX5_FLOW_CONTEXT_DEST_TYPE_TIR		= 2,
+};
+
+enum {
+	MLX5_RQC_RQ_TYPE_MEMORY_RQ_INLINE = 0x0,
+	MLX5_RQC_RQ_TYPE_MEMORY_RQ_RPM    = 0x1,
+};
+
+/* MLX5 DEV CAPs */
+
+/* TODO: EAT.ME */
+enum mlx5_cap_mode {
+	HCA_CAP_OPMOD_GET_MAX	= 0,
+	HCA_CAP_OPMOD_GET_CUR	= 1,
+};
+
+enum mlx5_cap_type {
+	MLX5_CAP_GENERAL = 0,
+	MLX5_CAP_ETHERNET_OFFLOADS,
+	MLX5_CAP_ODP,
+	MLX5_CAP_ATOMIC,
+	MLX5_CAP_ROCE,
+	MLX5_CAP_IPOIB_OFFLOADS,
+	MLX5_CAP_EOIB_OFFLOADS,
+	MLX5_CAP_FLOW_TABLE,
+	/* NUM OF CAP Types */
+	MLX5_CAP_NUM
+};
+
+/* GET Dev Caps macros */
+#define MLX5_CAP_GEN(mdev, cap) \
+	MLX5_GET(cmd_hca_cap, mdev->hca_caps_cur[MLX5_CAP_GENERAL], cap)
+
+#define MLX5_CAP_GEN_MAX(mdev, cap) \
+	MLX5_GET(cmd_hca_cap, mdev->hca_caps_max[MLX5_CAP_GENERAL], cap)
+
+#define MLX5_CAP_ETH(mdev, cap) \
+	MLX5_GET(per_protocol_networking_offload_caps,\
+		 mdev->hca_caps_cur[MLX5_CAP_ETHERNET_OFFLOADS], cap)
+
+#define MLX5_CAP_ETH_MAX(mdev, cap) \
+	MLX5_GET(per_protocol_networking_offload_caps,\
+		 mdev->hca_caps_max[MLX5_CAP_ETHERNET_OFFLOADS], cap)
+
+#define MLX5_CAP_ROCE(mdev, cap) \
+	MLX5_GET(roce_cap, mdev->hca_caps_cur[MLX5_CAP_ROCE], cap)
+
+#define MLX5_CAP_ROCE_MAX(mdev, cap) \
+	MLX5_GET(roce_cap, mdev->hca_caps_max[MLX5_CAP_ROCE], cap)
+
+#define MLX5_CAP_ATOMIC(mdev, cap) \
+	MLX5_GET(atomic_caps, mdev->hca_caps_cur[MLX5_CAP_ATOMIC], cap)
+
+#define MLX5_CAP_ATOMIC_MAX(mdev, cap) \
+	MLX5_GET(atomic_caps, mdev->hca_caps_max[MLX5_CAP_ATOMIC], cap)
+
+#define MLX5_CAP_FLOWTABLE(mdev, cap) \
+	MLX5_GET(flow_table_nic_cap, mdev->hca_caps_cur[MLX5_CAP_FLOW_TABLE], cap)
+
+#define MLX5_CAP_FLOWTABLE_MAX(mdev, cap) \
+	MLX5_GET(flow_table_nic_cap, mdev->hca_caps_max[MLX5_CAP_FLOW_TABLE], cap)
+
+#define MLX5_CAP_ODP(mdev, cap)\
+	MLX5_GET(odp_cap, mdev->hca_caps_cur[MLX5_CAP_ODP], cap)
+
+enum {
+	MLX5_CMD_STAT_OK			= 0x0,
+	MLX5_CMD_STAT_INT_ERR			= 0x1,
+	MLX5_CMD_STAT_BAD_OP_ERR		= 0x2,
+	MLX5_CMD_STAT_BAD_PARAM_ERR		= 0x3,
+	MLX5_CMD_STAT_BAD_SYS_STATE_ERR		= 0x4,
+	MLX5_CMD_STAT_BAD_RES_ERR		= 0x5,
+	MLX5_CMD_STAT_RES_BUSY			= 0x6,
+	MLX5_CMD_STAT_LIM_ERR			= 0x8,
+	MLX5_CMD_STAT_BAD_RES_STATE_ERR		= 0x9,
+	MLX5_CMD_STAT_IX_ERR			= 0xa,
+	MLX5_CMD_STAT_NO_RES_ERR		= 0xf,
+	MLX5_CMD_STAT_BAD_INP_LEN_ERR		= 0x50,
+	MLX5_CMD_STAT_BAD_OUTP_LEN_ERR		= 0x51,
+	MLX5_CMD_STAT_BAD_QP_STATE_ERR		= 0x10,
+	MLX5_CMD_STAT_BAD_PKT_ERR		= 0x30,
+	MLX5_CMD_STAT_BAD_SIZE_OUTS_CQES_ERR	= 0x40,
+};
+
 #endif /* MLX5_DEVICE_H */

include/linux/mlx5/driver.h

@@ -44,7 +44,6 @@
 
 #include <linux/mlx5/device.h>
 #include <linux/mlx5/doorbell.h>
-#include <linux/mlx5/mlx5_ifc.h>
 
 enum {
 	MLX5_BOARD_ID_LEN = 64,
@@ -85,7 +84,7 @@ enum {
 };
 
 enum {
-	MLX5_MAX_EQ_NAME	= 32
+	MLX5_MAX_IRQ_NAME	= 32
 };
 
 enum {
@@ -150,6 +149,11 @@ enum mlx5_dev_event {
 	MLX5_DEV_EVENT_CLIENT_REREG,
 };
 
+enum mlx5_port_status {
+	MLX5_PORT_UP        = 1 << 1,
+	MLX5_PORT_DOWN      = 1 << 2,
+};
+
 struct mlx5_uuar_info {
 	struct mlx5_uar	       *uars;
 	int			num_uars;
@@ -269,56 +273,7 @@ struct mlx5_cmd {
 struct mlx5_port_caps {
 	int	gid_table_len;
 	int	pkey_table_len;
-};
-
-struct mlx5_general_caps {
-	u8	log_max_eq;
-	u8	log_max_cq;
-	u8	log_max_qp;
-	u8	log_max_mkey;
-	u8	log_max_pd;
-	u8	log_max_srq;
-	u8	log_max_strq;
-	u8	log_max_mrw_sz;
-	u8	log_max_bsf_list_size;
-	u8	log_max_klm_list_size;
-	u32	max_cqes;
-	int	max_wqes;
-	u32	max_eqes;
-	u32	max_indirection;
-	int	max_sq_desc_sz;
-	int	max_rq_desc_sz;
-	int	max_dc_sq_desc_sz;
-	u64	flags;
-	u16	stat_rate_support;
-	int	log_max_msg;
-	int	num_ports;
-	u8	log_max_ra_res_qp;
-	u8	log_max_ra_req_qp;
-	int	max_srq_wqes;
-	int	bf_reg_size;
-	int	bf_regs_per_page;
-	struct mlx5_port_caps	port[MLX5_MAX_PORTS];
-	u8			ext_port_cap[MLX5_MAX_PORTS];
-	int	max_vf;
-	u32	reserved_lkey;
-	u8	local_ca_ack_delay;
-	u8	log_max_mcg;
-	u32	max_qp_mcg;
-	int	min_page_sz;
-	int	pd_cap;
-	u32	max_qp_counters;
-	u32	pkey_table_size;
-	u8	log_max_ra_req_dc;
-	u8	log_max_ra_res_dc;
-	u32	uar_sz;
-	u8	min_log_pg_sz;
-	u8	log_max_xrcd;
-	u16	log_uar_page_sz;
-};
-
-struct mlx5_caps {
-	struct mlx5_general_caps gen;
+	u8	ext_port_cap;
 };
 
 struct mlx5_cmd_mailbox {
@@ -334,8 +289,6 @@ struct mlx5_buf_list {
 
 struct mlx5_buf {
 	struct mlx5_buf_list	direct;
-	struct mlx5_buf_list   *page_list;
-	int			nbufs;
 	int			npages;
 	int			size;
 	u8			page_shift;
@@ -351,7 +304,6 @@ struct mlx5_eq {
 	u8			eqn;
 	int			nent;
 	u64			mask;
-	char			name[MLX5_MAX_EQ_NAME];
 	struct list_head	list;
 	int			index;
 	struct mlx5_rsc_debug	*dbg;
@@ -414,7 +366,6 @@ struct mlx5_eq_table {
 	struct mlx5_eq		pages_eq;
 	struct mlx5_eq		async_eq;
 	struct mlx5_eq		cmd_eq;
-	struct msix_entry	*msix_arr;
 	int			num_comp_vectors;
 	/* protect EQs list
 	 */
@@ -467,9 +418,16 @@ struct mlx5_mr_table {
 	struct radix_tree_root	tree;
 };
 
+struct mlx5_irq_info {
+	cpumask_var_t mask;
+	char name[MLX5_MAX_IRQ_NAME];
+};
+
 struct mlx5_priv {
 	char			name[MLX5_MAX_NAME_LEN];
 	struct mlx5_eq_table	eq_table;
+	struct msix_entry	*msix_arr;
+	struct mlx5_irq_info	*irq_info;
 	struct mlx5_uuar_info	uuari;
 	MLX5_DECLARE_DOORBELL_LOCK(cq_uar_lock);
 
@@ -520,7 +478,9 @@ struct mlx5_core_dev {
 	u8			rev_id;
 	char			board_id[MLX5_BOARD_ID_LEN];
 	struct mlx5_cmd		cmd;
-	struct mlx5_caps	caps;
+	struct mlx5_port_caps	port_caps[MLX5_MAX_PORTS];
+	u32 hca_caps_cur[MLX5_CAP_NUM][MLX5_UN_SZ_DW(hca_cap_union)];
+	u32 hca_caps_max[MLX5_CAP_NUM][MLX5_UN_SZ_DW(hca_cap_union)];
 	phys_addr_t		iseg_base;
 	struct mlx5_init_seg __iomem *iseg;
 	void			(*event) (struct mlx5_core_dev *dev,
@@ -529,6 +489,7 @@ struct mlx5_core_dev {
 	struct mlx5_priv	priv;
 	struct mlx5_profile	*profile;
 	atomic_t		num_qps;
+	u32			issi;
 };
 
 struct mlx5_db {
@@ -549,6 +510,11 @@ enum {
 	MLX5_COMP_EQ_SIZE = 1024,
 };
 
+enum {
+	MLX5_PTYS_IB = 1 << 0,
+	MLX5_PTYS_EN = 1 << 2,
+};
+
 struct mlx5_db_pgdir {
 	struct list_head	list;
 	DECLARE_BITMAP(bitmap, MLX5_DB_PER_PAGE);
@@ -586,11 +552,7 @@ struct mlx5_pas {
 
 static inline void *mlx5_buf_offset(struct mlx5_buf *buf, int offset)
 {
-	if (likely(BITS_PER_LONG == 64 || buf->nbufs == 1))
 		return buf->direct.buf + offset;
-	else
-		return buf->page_list[offset >> PAGE_SHIFT].buf +
-			(offset & (PAGE_SIZE - 1));
 }
 
 extern struct workqueue_struct *mlx5_core_wq;
@@ -654,8 +616,8 @@ void mlx5_cmd_use_events(struct mlx5_core_dev *dev);
 void mlx5_cmd_use_polling(struct mlx5_core_dev *dev);
 int mlx5_cmd_status_to_err(struct mlx5_outbox_hdr *hdr);
 int mlx5_cmd_status_to_err_v2(void *ptr);
-int mlx5_core_get_caps(struct mlx5_core_dev *dev, struct mlx5_caps *caps,
-		       u16 opmod);
+int mlx5_core_get_caps(struct mlx5_core_dev *dev, enum mlx5_cap_type cap_type,
+		       enum mlx5_cap_mode cap_mode);
 int mlx5_cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out,
 		  int out_size);
 int mlx5_cmd_exec_cb(struct mlx5_core_dev *dev, void *in, int in_size,
@@ -665,12 +627,13 @@ int mlx5_cmd_alloc_uar(struct mlx5_core_dev *dev, u32 *uarn);
 int mlx5_cmd_free_uar(struct mlx5_core_dev *dev, u32 uarn);
 int mlx5_alloc_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari);
 int mlx5_free_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari);
+int mlx5_alloc_map_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar);
+void mlx5_unmap_free_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar);
 void mlx5_health_cleanup(void);
 void  __init mlx5_health_init(void);
 void mlx5_start_health_poll(struct mlx5_core_dev *dev);
 void mlx5_stop_health_poll(struct mlx5_core_dev *dev);
-int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, int max_direct,
-		   struct mlx5_buf *buf);
+int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, struct mlx5_buf *buf);
 void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_buf *buf);
 struct mlx5_cmd_mailbox *mlx5_alloc_cmd_mailbox_chain(struct mlx5_core_dev *dev,
 						      gfp_t flags, int npages);
@@ -734,7 +697,23 @@ void mlx5_qp_debugfs_cleanup(struct mlx5_core_dev *dev);
 int mlx5_core_access_reg(struct mlx5_core_dev *dev, void *data_in,
 			 int size_in, void *data_out, int size_out,
 			 u16 reg_num, int arg, int write);
+
 int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps);
+int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys,
+			 int ptys_size, int proto_mask);
+int mlx5_query_port_proto_cap(struct mlx5_core_dev *dev,
+			      u32 *proto_cap, int proto_mask);
+int mlx5_query_port_proto_admin(struct mlx5_core_dev *dev,
+				u32 *proto_admin, int proto_mask);
+int mlx5_set_port_proto(struct mlx5_core_dev *dev, u32 proto_admin,
+			int proto_mask);
+int mlx5_set_port_status(struct mlx5_core_dev *dev,
+			 enum mlx5_port_status status);
+int mlx5_query_port_status(struct mlx5_core_dev *dev, u8 *status);
+
+int mlx5_set_port_mtu(struct mlx5_core_dev *dev, int mtu);
+int mlx5_query_port_max_mtu(struct mlx5_core_dev *dev, int *max_mtu);
+int mlx5_query_port_oper_mtu(struct mlx5_core_dev *dev, int *oper_mtu);
 
 int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq);
 void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq);

include/linux/mlx5/flow_table.h

+/*
+ * Copyright (c) 2013-2015, Mellanox Technologies, Ltd.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef MLX5_FLOW_TABLE_H
+#define MLX5_FLOW_TABLE_H
+
+#include <linux/mlx5/driver.h>
+
+struct mlx5_flow_table_group {
+	u8	log_sz;
+	u8	match_criteria_enable;
+	u32	match_criteria[MLX5_ST_SZ_DW(fte_match_param)];
+};
+
+void *mlx5_create_flow_table(struct mlx5_core_dev *dev, u8 level, u8 table_type,
+			     u16 num_groups,
+			     struct mlx5_flow_table_group *group);
+void mlx5_destroy_flow_table(void *flow_table);
+int mlx5_add_flow_table_entry(void *flow_table, u8 match_criteria_enable,
+			      void *match_criteria, void *flow_context,
+			      u32 *flow_index);
+void mlx5_del_flow_table_entry(void *flow_table, u32 flow_index);
+u32 mlx5_get_flow_table_id(void *flow_table);
+
+#endif /* MLX5_FLOW_TABLE_H */

include/linux/mlx5/qp.h

@@ -134,13 +134,21 @@ enum {
 
 enum {
 	MLX5_WQE_CTRL_CQ_UPDATE		= 2 << 2,
+	MLX5_WQE_CTRL_CQ_UPDATE_AND_EQE	= 3 << 2,
 	MLX5_WQE_CTRL_SOLICITED		= 1 << 1,
 };
 
 enum {
+	MLX5_SEND_WQE_DS	= 16,
 	MLX5_SEND_WQE_BB	= 64,
 };
 
+#define MLX5_SEND_WQEBB_NUM_DS	(MLX5_SEND_WQE_BB / MLX5_SEND_WQE_DS)
+
+enum {
+	MLX5_SEND_WQE_MAX_WQEBBS	= 16,
+};
+
 enum {
 	MLX5_WQE_FMR_PERM_LOCAL_READ	= 1 << 27,
 	MLX5_WQE_FMR_PERM_LOCAL_WRITE	= 1 << 28,
@@ -200,6 +208,23 @@ struct mlx5_wqe_ctrl_seg {
 #define MLX5_WQE_CTRL_WQE_INDEX_MASK 0x00ffff00
 #define MLX5_WQE_CTRL_WQE_INDEX_SHIFT 8
 
+enum {
+	MLX5_ETH_WQE_L3_INNER_CSUM      = 1 << 4,
+	MLX5_ETH_WQE_L4_INNER_CSUM      = 1 << 5,
+	MLX5_ETH_WQE_L3_CSUM            = 1 << 6,
+	MLX5_ETH_WQE_L4_CSUM            = 1 << 7,
+};
+
+struct mlx5_wqe_eth_seg {
+	u8              rsvd0[4];
+	u8              cs_flags;
+	u8              rsvd1;
+	__be16          mss;
+	__be32          rsvd2;
+	__be16          inline_hdr_sz;
+	u8              inline_hdr_start[2];
+};
+
 struct mlx5_wqe_xrc_seg {
 	__be32			xrc_srqn;
 	u8			rsvd[12];