[net-next 00/12][pull request] 40GbE Intel Wired LAN Driver Updates 2018-02-14

[net-next 00/12][pull request] 40GbE Intel Wired LAN Driver Updates 2018-02-14

[Jeff Kirsher]

This patch series enables the new mqprio hardware offload mechanism
creating traffic classes on VFs for XL710 devices. The parameters
needed to configure these traffic classes/queue channels are provides
by the user via the tc tool. A maximum of four traffic classes can be
created on each VF. This patch series also enables application of cloud
filters to each of these traffic classes. The cloud filters are applied
using the tc-flower classifier.

Example:
    1. tc qdisc add dev vf0 root mqprio num_tc 4 map 0 0 0 0 1 2 2 3\
        queues 2@0 2@2 1@4 1@5 hw 1 mode channel
    2. tc qdisc add dev vf0 ingress
    3. ethtool -K vf0 hw-tc-offload on
    4. ip link set eth0 vf 0 spoofchk off
    5. tc filter add dev vf0 protocol ip parent ffff: prio 1 flower dst_ip\
        192.168.3.5/32 ip_proto udp dst_port 25 skip_sw hw_tc 2

The following are changes since commit 1d631583ae5e40681dc9a41d7cfa845222c3c7ff:
  Merge branch 'net-dev-Make-protocol-ptr-dependent-on-CONFIG'
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queue 40GbE

Avinash Dayanand (6):
  i40evf: Fix link up issue when queues are disabled
  i40e: Enable ADq and create queue channel/s on VF
  i40evf: Alloc queues for ADq on VF
  i40e: Delete queue channel for ADq on VF
  i40e: Service request to configure bandwidth for ADq on a VF
  i40e: Add and delete cloud filter

Harshitha Ramamurthy (6):
  virtchnl: Add virtchl structures to support queue channels
  i40evf: add ndo_setup_tc callback to i40evf
  i40evf: Add support to configure bw via tc tool
  virtchnl: Add a macro to check the size of a union
  virtchnl: Add filter data structures
  i40evf: Add support to apply cloud filters

 drivers/net/ethernet/intel/i40e/i40e.h             |    6 +
 drivers/net/ethernet/intel/i40e/i40e_main.c        |   16 +-
 drivers/net/ethernet/intel/i40e/i40e_type.h        |    2 +-
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 1000 ++++++++++++++++++--
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h |   20 +
 drivers/net/ethernet/intel/i40evf/i40evf.h         |   72 ++
 drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c |    6 +
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    |  803 +++++++++++++++-
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    |  300 +++++-
 include/linux/avf/virtchnl.h                       |  107 ++-
 10 files changed, 2237 insertions(+), 95 deletions(-)

-- 
2.14.3

[net-next 01/12] i40evf: Fix link up issue when queues are disabled

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

One of the previous patch fixes the link up issue by ignoring it if
i40evf is not in __I40EVF_RUNNING state. However this doesn't fix the
race condition when queues are disabled esp for ADq on VF. Hence check
if all queues are enabled before starting all queues.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40evf/i40evf.h         |  1 +
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    |  1 +
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    | 28 +++++++++++++++-------
 3 files changed, 22 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40evf/i40evf.h b/drivers/net/ethernet/intel/i40evf/i40evf.h
index b6991e8014d8..89ce2f9a0fbe 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf.h
+++ b/drivers/net/ethernet/intel/i40evf/i40evf.h
@@ -240,6 +240,7 @@ struct i40evf_adapter {
 #define I40EVF_FLAG_ALLMULTI_ON			BIT(14)
 #define I40EVF_FLAG_LEGACY_RX			BIT(15)
 #define I40EVF_FLAG_REINIT_ITR_NEEDED		BIT(16)
+#define I40EVF_FLAG_QUEUES_DISABLED		BIT(17)
 /* duplicates for common code */
 #define I40E_FLAG_DCB_ENABLED			0
 #define I40E_FLAG_RX_CSUM_ENABLED		I40EVF_FLAG_RX_CSUM_ENABLED
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 34fd6c553879..0776b07477a2 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -1913,6 +1913,7 @@ static void i40evf_reset_task(struct work_struct *work)
 	i40evf_free_all_rx_resources(adapter);
 	i40evf_free_all_tx_resources(adapter);
 
+	adapter->flags |= I40EVF_FLAG_QUEUES_DISABLED;
 	/* kill and reinit the admin queue */
 	i40evf_shutdown_adminq(hw);
 	adapter->current_op = VIRTCHNL_OP_UNKNOWN;
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index 0700f0afe2d3..e8dcc31fcbf2 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -1017,14 +1017,25 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 			if (adapter->link_up == link_up)
 				break;
 
-			/* If we get link up message and start queues before
-			 * our queues are configured it will trigger a TX hang.
-			 * In that case, just ignore the link status message,
-			 * we'll get another one after we enable queues and
-			 * actually prepared to send traffic.
-			 */
-			if (link_up && adapter->state != __I40EVF_RUNNING)
-				break;
+			if (link_up) {
+				/* If we get link up message and start queues
+				 * before our queues are configured it will
+				 * trigger a TX hang. In that case, just ignore
+				 * the link status message,we'll get another one
+				 * after we enable queues and actually prepared
+				 * to send traffic.
+				 */
+				if (adapter->state != __I40EVF_RUNNING)
+					break;
+
+				/* For ADq enabled VF, we reconfigure VSIs and
+				 * re-allocate queues. Hence wait till all
+				 * queues are enabled.
+				 */
+				if (adapter->flags &
+				    I40EVF_FLAG_QUEUES_DISABLED)
+					break;
+			}
 
 			adapter->link_up = link_up;
 			if (link_up) {
@@ -1108,6 +1119,7 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 	case VIRTCHNL_OP_ENABLE_QUEUES:
 		/* enable transmits */
 		i40evf_irq_enable(adapter, true);
+		adapter->flags &= ~I40EVF_FLAG_QUEUES_DISABLED;
 		break;
 	case VIRTCHNL_OP_DISABLE_QUEUES:
 		i40evf_free_all_tx_resources(adapter);
-- 
2.14.3

[net-next 02/12] virtchnl: Add virtchl structures to support queue channels

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch defines new structs in support of the virtchannel message
that the VF sends to the PF to create a queue channel specified by the
user via tc tool.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 include/linux/avf/virtchnl.h | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h
index 3ce61342fa31..1f652ceecf35 100644
--- a/include/linux/avf/virtchnl.h
+++ b/include/linux/avf/virtchnl.h
@@ -136,6 +136,8 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_ENABLE_VLAN_STRIPPING = 27,
 	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING = 28,
 	VIRTCHNL_OP_REQUEST_QUEUES = 29,
+	VIRTCHNL_OP_ENABLE_CHANNELS = 30,
+	VIRTCHNL_OP_DISABLE_CHANNELS = 31,
 };
 
 /* This macro is used to generate a compilation error if a structure
@@ -244,6 +246,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_ENCAP		0X00100000
 #define VIRTCHNL_VF_OFFLOAD_ENCAP_CSUM		0X00200000
 #define VIRTCHNL_VF_OFFLOAD_RX_ENCAP_CSUM	0X00400000
+#define VIRTCHNL_VF_OFFLOAD_ADQ			0X00800000
 
 #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
 			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -496,6 +499,30 @@ struct virtchnl_rss_hena {
 
 VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_rss_hena);
 
+/* VIRTCHNL_OP_ENABLE_CHANNELS
+ * VIRTCHNL_OP_DISABLE_CHANNELS
+ * VF sends these messages to enable or disable channels based on
+ * the user specified queue count and queue offset for each traffic class.
+ * This struct encompasses all the information that the PF needs from
+ * VF to create a channel.
+ */
+struct virtchnl_channel_info {
+	u16 count; /* number of queues in a channel */
+	u16 offset; /* queues in a channel start from 'offset' */
+	u32 pad;
+	u64 max_tx_rate;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_channel_info);
+
+struct virtchnl_tc_info {
+	u32	num_tc;
+	u32	pad;
+	struct	virtchnl_channel_info list[1];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(24, virtchnl_tc_info);
+
 /* VIRTCHNL_OP_EVENT
  * PF sends this message to inform the VF driver of events that may affect it.
  * No direct response is expected from the VF, though it may generate other
@@ -711,6 +738,19 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_REQUEST_QUEUES:
 		valid_len = sizeof(struct virtchnl_vf_res_request);
 		break;
+	case VIRTCHNL_OP_ENABLE_CHANNELS:
+		valid_len = sizeof(struct virtchnl_tc_info);
+		if (msglen >= valid_len) {
+			struct virtchnl_tc_info *vti =
+				(struct virtchnl_tc_info *)msg;
+			valid_len += vti->num_tc *
+				sizeof(struct virtchnl_channel_info);
+			if (vti->num_tc == 0)
+				err_msg_format = true;
+		}
+		break;
+	case VIRTCHNL_OP_DISABLE_CHANNELS:
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
-- 
2.14.3

[net-next 03/12] i40evf: add ndo_setup_tc callback to i40evf

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch introduces the callback to the ndo_setup_tc function
in the VF driver. We add a wrapper function to make room for the
upcoming cloud filter patches which add calls to different functions
from setup_tc.

First, we add support for capability exchange for ADQ between the
PF and VF. Next, we add support to take in the mqprio configuration
and configure queues as per the traffic classes, rate limit and the
priorities specified by the user. This is done by passing the channel
config to the PF driver through a virtchannel message.

The flags and bits added, track if ADq is enabled, set max number of
traffic classes to 4 and provide ability to negotiate capability with
the PF.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40evf/i40evf.h         |  22 +++
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 154 +++++++++++++++++++++
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    |  82 ++++++++++-
 3 files changed, 257 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/i40evf/i40evf.h b/drivers/net/ethernet/intel/i40evf/i40evf.h
index 89ce2f9a0fbe..c8d68c8d2c33 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf.h
+++ b/drivers/net/ethernet/intel/i40evf/i40evf.h
@@ -52,6 +52,7 @@
 #include <linux/socket.h>
 #include <linux/jiffies.h>
 #include <net/ip6_checksum.h>
+#include <net/pkt_cls.h>
 #include <net/udp.h>
 
 #include "i40e_type.h"
@@ -168,6 +169,20 @@ struct i40evf_vlan_filter {
 	bool add;		/* filter needs to be added */
 };
 
+#define I40EVF_MAX_TRAFFIC_CLASS	4
+/* State of traffic class creation */
+enum i40evf_tc_state_t {
+	__I40EVF_TC_INVALID, /* no traffic class, default state */
+	__I40EVF_TC_RUNNING, /* traffic classes have been created */
+};
+
+/* channel info */
+struct i40evf_channel_config {
+	struct virtchnl_channel_info ch_info[I40EVF_MAX_TRAFFIC_CLASS];
+	enum i40evf_tc_state_t state;
+	u8 total_qps;
+};
+
 /* Driver state. The order of these is important! */
 enum i40evf_state_t {
 	__I40EVF_STARTUP,		/* driver loaded, probe complete */
@@ -269,6 +284,8 @@ struct i40evf_adapter {
 #define I40EVF_FLAG_AQ_RELEASE_ALLMULTI		BIT(18)
 #define I40EVF_FLAG_AQ_ENABLE_VLAN_STRIPPING	BIT(19)
 #define I40EVF_FLAG_AQ_DISABLE_VLAN_STRIPPING	BIT(20)
+#define I40EVF_FLAG_AQ_ENABLE_CHANNELS		BIT(21)
+#define I40EVF_FLAG_AQ_DISABLE_CHANNELS		BIT(22)
 
 	/* OS defined structs */
 	struct net_device *netdev;
@@ -314,6 +331,9 @@ struct i40evf_adapter {
 	u16 rss_lut_size;
 	u8 *rss_key;
 	u8 *rss_lut;
+	/* ADQ related members */
+	struct i40evf_channel_config ch_config;
+	u8 num_tc;
 };
 
 
@@ -380,4 +400,6 @@ void i40evf_notify_client_message(struct i40e_vsi *vsi, u8 *msg, u16 len);
 void i40evf_notify_client_l2_params(struct i40e_vsi *vsi);
 void i40evf_notify_client_open(struct i40e_vsi *vsi);
 void i40evf_notify_client_close(struct i40e_vsi *vsi, bool reset);
+void i40evf_enable_channels(struct i40evf_adapter *adapter);
+void i40evf_disable_channels(struct i40evf_adapter *adapter);
 #endif /* _I40EVF_H_ */
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 0776b07477a2..099d4f59e445 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -1732,6 +1732,17 @@ static void i40evf_watchdog_task(struct work_struct *work)
 		i40evf_set_promiscuous(adapter, 0);
 		goto watchdog_done;
 	}
+
+	if (adapter->aq_required & I40EVF_FLAG_AQ_ENABLE_CHANNELS) {
+		i40evf_enable_channels(adapter);
+		goto watchdog_done;
+	}
+
+	if (adapter->aq_required & I40EVF_FLAG_AQ_DISABLE_CHANNELS) {
+		i40evf_disable_channels(adapter);
+		goto watchdog_done;
+	}
+
 	schedule_delayed_work(&adapter->client_task, msecs_to_jiffies(5));
 
 	if (adapter->state == __I40EVF_RUNNING)
@@ -2211,6 +2222,148 @@ void i40evf_free_all_rx_resources(struct i40evf_adapter *adapter)
 			i40evf_free_rx_resources(&adapter->rx_rings[i]);
 }
 
+/**
+ * i40evf_validate_channel_config - validate queue mapping info
+ * @adapter: board private structure
+ * @mqprio_qopt: queue parameters
+ *
+ * This function validates if the config provided by the user to
+ * configure queue channels is valid or not. Returns 0 on a valid
+ * config.
+ **/
+static int i40evf_validate_ch_config(struct i40evf_adapter *adapter,
+				     struct tc_mqprio_qopt_offload *mqprio_qopt)
+{
+	int i, num_qps = 0;
+
+	if (mqprio_qopt->qopt.num_tc > I40EVF_MAX_TRAFFIC_CLASS ||
+	    mqprio_qopt->qopt.num_tc < 1)
+		return -EINVAL;
+
+	for (i = 0; i <= mqprio_qopt->qopt.num_tc - 1; i++) {
+		if (!mqprio_qopt->qopt.count[i] ||
+		    mqprio_qopt->min_rate[i] ||
+		    mqprio_qopt->max_rate[i] ||
+		    mqprio_qopt->qopt.offset[i] != num_qps)
+			return -EINVAL;
+		num_qps += mqprio_qopt->qopt.count[i];
+	}
+	if (num_qps > MAX_QUEUES)
+		return -EINVAL;
+
+	return 0;
+}
+
+/**
+ * __i40evf_setup_tc - configure multiple traffic classes
+ * @netdev: network interface device structure
+ * @type_date: tc offload data
+ *
+ * This function processes the config information provided by the
+ * user to configure traffic classes/queue channels and packages the
+ * information to request the PF to setup traffic classes.
+ *
+ * Returns 0 on success.
+ **/
+static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
+{
+	struct tc_mqprio_qopt_offload *mqprio_qopt = type_data;
+	struct i40evf_adapter *adapter = netdev_priv(netdev);
+	struct virtchnl_vf_resource *vfres = adapter->vf_res;
+	u8 num_tc = 0, total_qps = 0;
+	int ret = 0, netdev_tc = 0;
+	u16 mode;
+	int i;
+
+	num_tc = mqprio_qopt->qopt.num_tc;
+	mode = mqprio_qopt->mode;
+
+	/* delete queue_channel */
+	if (!mqprio_qopt->qopt.hw) {
+		if (adapter->ch_config.state == __I40EVF_TC_RUNNING) {
+			/* reset the tc configuration */
+			netdev_reset_tc(netdev);
+			adapter->num_tc = 0;
+			netif_tx_stop_all_queues(netdev);
+			netif_tx_disable(netdev);
+			adapter->aq_required = I40EVF_FLAG_AQ_DISABLE_CHANNELS;
+			goto exit;
+		} else {
+			return -EINVAL;
+		}
+	}
+
+	/* add queue channel */
+	if (mode == TC_MQPRIO_MODE_CHANNEL) {
+		if (!(vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ)) {
+			dev_err(&adapter->pdev->dev, "ADq not supported\n");
+			return -EOPNOTSUPP;
+		}
+		if (adapter->ch_config.state != __I40EVF_TC_INVALID) {
+			dev_err(&adapter->pdev->dev, "TC configuration already exists\n");
+			return -EINVAL;
+		}
+
+		ret = i40evf_validate_ch_config(adapter, mqprio_qopt);
+		if (ret)
+			return ret;
+		/* Return if same TC config is requested */
+		if (adapter->num_tc == num_tc)
+			return 0;
+		adapter->num_tc = num_tc;
+
+		for (i = 0; i < I40EVF_MAX_TRAFFIC_CLASS; i++) {
+			if (i < num_tc) {
+				adapter->ch_config.ch_info[i].count =
+					mqprio_qopt->qopt.count[i];
+				adapter->ch_config.ch_info[i].offset =
+					mqprio_qopt->qopt.offset[i];
+				total_qps += mqprio_qopt->qopt.count[i];
+			} else {
+				adapter->ch_config.ch_info[i].count = 1;
+				adapter->ch_config.ch_info[i].offset = 0;
+			}
+		}
+		adapter->ch_config.total_qps = total_qps;
+		netif_tx_stop_all_queues(netdev);
+		netif_tx_disable(netdev);
+		adapter->aq_required |= I40EVF_FLAG_AQ_ENABLE_CHANNELS;
+		netdev_reset_tc(netdev);
+		/* Report the tc mapping up the stack */
+		netdev_set_num_tc(adapter->netdev, num_tc);
+		for (i = 0; i < I40EVF_MAX_TRAFFIC_CLASS; i++) {
+			u16 qcount = mqprio_qopt->qopt.count[i];
+			u16 qoffset = mqprio_qopt->qopt.offset[i];
+
+			if (i < num_tc)
+				netdev_set_tc_queue(netdev, netdev_tc++, qcount,
+						    qoffset);
+		}
+	}
+exit:
+	return ret;
+}
+
+/**
+ * i40evf_setup_tc - configure multiple traffic classes
+ * @netdev: network interface device structure
+ * @type: type of offload
+ * @type_date: tc offload data
+ *
+ * This function is the callback to ndo_setup_tc in the
+ * netdev_ops.
+ *
+ * Returns 0 on success
+ **/
+static int i40evf_setup_tc(struct net_device *netdev, enum tc_setup_type type,
+			   void *type_data)
+{
+	if (type != TC_SETUP_QDISC_MQPRIO)
+		return -EOPNOTSUPP;
+
+	return __i40evf_setup_tc(netdev, type_data);
+}
+
 /**
  * i40evf_open - Called when a network interface is made active
  * @netdev: network interface device structure
@@ -2478,6 +2631,7 @@ static const struct net_device_ops i40evf_netdev_ops = {
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	.ndo_poll_controller	= i40evf_netpoll,
 #endif
+	.ndo_setup_tc		= i40evf_setup_tc,
 };
 
 /**
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index e8dcc31fcbf2..f6c56141b4fe 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -161,7 +161,8 @@ int i40evf_send_vf_config_msg(struct i40evf_adapter *adapter)
 	       VIRTCHNL_VF_OFFLOAD_RSS_PCTYPE_V2 |
 	       VIRTCHNL_VF_OFFLOAD_ENCAP |
 	       VIRTCHNL_VF_OFFLOAD_ENCAP_CSUM |
-	       VIRTCHNL_VF_OFFLOAD_REQ_QUEUES;
+	       VIRTCHNL_VF_OFFLOAD_REQ_QUEUES |
+	       VIRTCHNL_VF_OFFLOAD_ADQ;
 
 	adapter->current_op = VIRTCHNL_OP_GET_VF_RESOURCES;
 	adapter->aq_required &= ~I40EVF_FLAG_AQ_GET_CONFIG;
@@ -972,6 +973,70 @@ static void i40evf_print_link_message(struct i40evf_adapter *adapter)
 	netdev_info(netdev, "NIC Link is Up %sbps Full Duplex\n", speed);
 }
 
+/**
+ * i40evf_enable_channel
+ * @adapter: adapter structure
+ *
+ * Request that the PF enable channels as specified by
+ * the user via tc tool.
+ **/
+void i40evf_enable_channels(struct i40evf_adapter *adapter)
+{
+	struct virtchnl_tc_info *vti = NULL;
+	u16 len;
+	int i;
+
+	if (adapter->current_op != VIRTCHNL_OP_UNKNOWN) {
+		/* bail because we already have a command pending */
+		dev_err(&adapter->pdev->dev, "Cannot configure mqprio, command %d pending\n",
+			adapter->current_op);
+		return;
+	}
+
+	len = (adapter->num_tc * sizeof(struct virtchnl_channel_info)) +
+	       sizeof(struct virtchnl_tc_info);
+
+	vti = kzalloc(len, GFP_KERNEL);
+	if (!vti)
+		return;
+	vti->num_tc = adapter->num_tc;
+	for (i = 0; i < vti->num_tc; i++) {
+		vti->list[i].count = adapter->ch_config.ch_info[i].count;
+		vti->list[i].offset = adapter->ch_config.ch_info[i].offset;
+	}
+
+	adapter->ch_config.state = __I40EVF_TC_RUNNING;
+	adapter->flags |= I40EVF_FLAG_REINIT_ITR_NEEDED;
+	adapter->current_op = VIRTCHNL_OP_ENABLE_CHANNELS;
+	adapter->aq_required &= ~I40EVF_FLAG_AQ_ENABLE_CHANNELS;
+	i40evf_send_pf_msg(adapter, VIRTCHNL_OP_ENABLE_CHANNELS,
+			   (u8 *)vti, len);
+	kfree(vti);
+}
+
+/**
+ * i40evf_disable_channel
+ * @adapter: adapter structure
+ *
+ * Request that the PF disable channels that are configured
+ **/
+void i40evf_disable_channels(struct i40evf_adapter *adapter)
+{
+	if (adapter->current_op != VIRTCHNL_OP_UNKNOWN) {
+		/* bail because we already have a command pending */
+		dev_err(&adapter->pdev->dev, "Cannot configure mqprio, command %d pending\n",
+			adapter->current_op);
+		return;
+	}
+
+	adapter->ch_config.state = __I40EVF_TC_INVALID;
+	adapter->flags |= I40EVF_FLAG_REINIT_ITR_NEEDED;
+	adapter->current_op = VIRTCHNL_OP_DISABLE_CHANNELS;
+	adapter->aq_required &= ~I40EVF_FLAG_AQ_DISABLE_CHANNELS;
+	i40evf_send_pf_msg(adapter, VIRTCHNL_OP_DISABLE_CHANNELS,
+			   NULL, 0);
+}
+
 /**
  * i40evf_request_reset
  * @adapter: adapter structure
@@ -1080,6 +1145,21 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 			dev_err(&adapter->pdev->dev, "Failed to delete MAC filter, error %s\n",
 				i40evf_stat_str(&adapter->hw, v_retval));
 			break;
+		case VIRTCHNL_OP_ENABLE_CHANNELS:
+			dev_err(&adapter->pdev->dev, "Failed to configure queue channels, error %s\n",
+				i40evf_stat_str(&adapter->hw, v_retval));
+			adapter->flags &= ~I40EVF_FLAG_REINIT_ITR_NEEDED;
+			adapter->ch_config.state = __I40EVF_TC_INVALID;
+			netdev_reset_tc(netdev);
+			netif_tx_start_all_queues(netdev);
+			break;
+		case VIRTCHNL_OP_DISABLE_CHANNELS:
+			dev_err(&adapter->pdev->dev, "Failed to disable queue channels, error %s\n",
+				i40evf_stat_str(&adapter->hw, v_retval));
+			adapter->flags &= ~I40EVF_FLAG_REINIT_ITR_NEEDED;
+			adapter->ch_config.state = __I40EVF_TC_RUNNING;
+			netif_tx_start_all_queues(netdev);
+			break;
 		default:
 			dev_err(&adapter->pdev->dev, "PF returned error %d (%s) to our request %d\n",
 				v_retval,
-- 
2.14.3

[net-next 04/12] i40e: Enable ADq and create queue channel/s on VF

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

This patch enables ADq and creates queue channels on a VF. An ADq
enabled VF can have up to 4 VSIs and each one of them represents
a traffic class and this is termed as a queue channel. Each of these
VSIs can have up to 4 queues. This patch services the request for
enabling ADq and adds queue channel based on the TC mqprio info
provided by the user in the VF.

Initially a check is made to see if spoof check is OFF, if not ADq
will not be enabled. PF notifies VF for a reset in order to complete
the creation of ADq resources i.e. creation of additional VSIs and
allocation of queues as per TC information, all in the reset path.

Steps:
======
1. Turn off the spoof check
2. Enable ADq using tc mqprio command with or without rate limit.
3. Pass traffic.

Example:
========
% ip link set dev eth0 vf 0 spoofchk off
% tc qdisc add dev $iface root mqprio num_tc 4 map\
	0 0 0 0 1 1 1 1 2 2 2 2 3 3 3 3 queues\
	4@0 4@4 4@8 4@8 hw 1 mode channel

Expected results:
=================
1. Total number of queues for the VF should be sum of queues of all TCs.
2. Traffic flow should be normal without errors.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_type.h        |   2 +-
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 429 +++++++++++++++++----
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h |  17 +
 3 files changed, 379 insertions(+), 69 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_type.h b/drivers/net/ethernet/intel/i40e/i40e_type.h
index cd294e6a8587..b0eed8c0b2f2 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_type.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_type.h
@@ -39,7 +39,7 @@
 #define I40E_MASK(mask, shift) ((u32)(mask) << (shift))
 
 #define I40E_MAX_VSI_QP			16
-#define I40E_MAX_VF_VSI			3
+#define I40E_MAX_VF_VSI			4
 #define I40E_MAX_CHAINED_RX_BUFFERS	5
 #define I40E_MAX_PF_UDP_OFFLOAD_PORTS	16
 
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index e9309fb9084b..7452d4387818 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -257,6 +257,38 @@ static u16 i40e_vc_get_pf_queue_id(struct i40e_vf *vf, u16 vsi_id,
 	return pf_queue_id;
 }
 
+/**
+ * i40e_get_real_pf_qid
+ * @vf: pointer to the VF info
+ * @vsi_id: vsi id
+ * @queue_id: queue number
+ *
+ * wrapper function to get pf_queue_id handling ADq code as well
+ **/
+static u16 i40e_get_real_pf_qid(struct i40e_vf *vf, u16 vsi_id, u16 queue_id)
+{
+	int i;
+
+	if (vf->adq_enabled) {
+		/* Although VF considers all the queues(can be 1 to 16) as its
+		 * own but they may actually belong to different VSIs(up to 4).
+		 * We need to find which queues belongs to which VSI.
+		 */
+		for (i = 0; i < vf->num_tc; i++) {
+			if (queue_id < vf->ch[i].num_qps) {
+				vsi_id = vf->ch[i].vsi_id;
+				break;
+			}
+			/* find right queue id which is relative to a
+			 * given VSI.
+			 */
+			queue_id -= vf->ch[i].num_qps;
+			}
+		}
+
+	return i40e_vc_get_pf_queue_id(vf, vsi_id, queue_id);
+}
+
 /**
  * i40e_config_irq_link_list
  * @vf: pointer to the VF info
@@ -310,7 +342,7 @@ static void i40e_config_irq_link_list(struct i40e_vf *vf, u16 vsi_id,
 
 	vsi_queue_id = next_q / I40E_VIRTCHNL_SUPPORTED_QTYPES;
 	qtype = next_q % I40E_VIRTCHNL_SUPPORTED_QTYPES;
-	pf_queue_id = i40e_vc_get_pf_queue_id(vf, vsi_id, vsi_queue_id);
+	pf_queue_id = i40e_get_real_pf_qid(vf, vsi_id, vsi_queue_id);
 	reg = ((qtype << I40E_VPINT_LNKLSTN_FIRSTQ_TYPE_SHIFT) | pf_queue_id);
 
 	wr32(hw, reg_idx, reg);
@@ -333,8 +365,9 @@ static void i40e_config_irq_link_list(struct i40e_vf *vf, u16 vsi_id,
 		if (next_q < size) {
 			vsi_queue_id = next_q / I40E_VIRTCHNL_SUPPORTED_QTYPES;
 			qtype = next_q % I40E_VIRTCHNL_SUPPORTED_QTYPES;
-			pf_queue_id = i40e_vc_get_pf_queue_id(vf, vsi_id,
-							      vsi_queue_id);
+			pf_queue_id = i40e_get_real_pf_qid(vf,
+							   vsi_id,
+							   vsi_queue_id);
 		} else {
 			pf_queue_id = I40E_QUEUE_END_OF_LIST;
 			qtype = 0;
@@ -669,18 +702,19 @@ static int i40e_config_vsi_rx_queue(struct i40e_vf *vf, u16 vsi_id,
 /**
  * i40e_alloc_vsi_res
  * @vf: pointer to the VF info
- * @type: type of VSI to allocate
+ * @idx: VSI index, applies only for ADq mode, zero otherwise
  *
  * alloc VF vsi context & resources
  **/
-static int i40e_alloc_vsi_res(struct i40e_vf *vf, enum i40e_vsi_type type)
+static int i40e_alloc_vsi_res(struct i40e_vf *vf, u8 idx)
 {
 	struct i40e_mac_filter *f = NULL;
 	struct i40e_pf *pf = vf->pf;
 	struct i40e_vsi *vsi;
 	int ret = 0;
 
-	vsi = i40e_vsi_setup(pf, type, pf->vsi[pf->lan_vsi]->seid, vf->vf_id);
+	vsi = i40e_vsi_setup(pf, I40E_VSI_SRIOV, pf->vsi[pf->lan_vsi]->seid,
+			     vf->vf_id);
 
 	if (!vsi) {
 		dev_err(&pf->pdev->dev,
@@ -689,7 +723,8 @@ static int i40e_alloc_vsi_res(struct i40e_vf *vf, enum i40e_vsi_type type)
 		ret = -ENOENT;
 		goto error_alloc_vsi_res;
 	}
-	if (type == I40E_VSI_SRIOV) {
+
+	if (!idx) {
 		u64 hena = i40e_pf_get_default_rss_hena(pf);
 		u8 broadcast[ETH_ALEN];
 
@@ -721,12 +756,17 @@ static int i40e_alloc_vsi_res(struct i40e_vf *vf, enum i40e_vsi_type type)
 		spin_unlock_bh(&vsi->mac_filter_hash_lock);
 		wr32(&pf->hw, I40E_VFQF_HENA1(0, vf->vf_id), (u32)hena);
 		wr32(&pf->hw, I40E_VFQF_HENA1(1, vf->vf_id), (u32)(hena >> 32));
+		/* program mac filter only for VF VSI */
+		ret = i40e_sync_vsi_filters(vsi);
+		if (ret)
+			dev_err(&pf->pdev->dev, "Unable to program ucast filters\n");
 	}
 
-	/* program mac filter */
-	ret = i40e_sync_vsi_filters(vsi);
-	if (ret)
-		dev_err(&pf->pdev->dev, "Unable to program ucast filters\n");
+	/* storing VSI index and id for ADq and don't apply the mac filter */
+	if (vf->adq_enabled) {
+		vf->ch[idx].vsi_idx = vsi->idx;
+		vf->ch[idx].vsi_id = vsi->id;
+	}
 
 	/* Set VF bandwidth if specified */
 	if (vf->tx_rate) {
@@ -741,6 +781,92 @@ static int i40e_alloc_vsi_res(struct i40e_vf *vf, enum i40e_vsi_type type)
 	return ret;
 }
 
+/**
+ * i40e_map_pf_queues_to_vsi
+ * @vf: pointer to the VF info
+ *
+ * PF maps LQPs to a VF by programming VSILAN_QTABLE & VPLAN_QTABLE. This
+ * function takes care of first part VSILAN_QTABLE, mapping pf queues to VSI.
+ **/
+static void i40e_map_pf_queues_to_vsi(struct i40e_vf *vf)
+{
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_hw *hw = &pf->hw;
+	u32 reg, num_tc = 1; /* VF has at least one traffic class */
+	u16 vsi_id, qps;
+	int i, j;
+
+	if (vf->adq_enabled)
+		num_tc = vf->num_tc;
+
+	for (i = 0; i < num_tc; i++) {
+		if (vf->adq_enabled) {
+			qps = vf->ch[i].num_qps;
+			vsi_id =  vf->ch[i].vsi_id;
+		} else {
+			qps = pf->vsi[vf->lan_vsi_idx]->alloc_queue_pairs;
+			vsi_id = vf->lan_vsi_id;
+		}
+
+		for (j = 0; j < 7; j++) {
+			if (j * 2 >= qps) {
+				/* end of list */
+				reg = 0x07FF07FF;
+			} else {
+				u16 qid = i40e_vc_get_pf_queue_id(vf,
+								  vsi_id,
+								  j * 2);
+				reg = qid;
+				qid = i40e_vc_get_pf_queue_id(vf, vsi_id,
+							      (j * 2) + 1);
+				reg |= qid << 16;
+			}
+			i40e_write_rx_ctl(hw,
+					  I40E_VSILAN_QTABLE(j, vsi_id),
+					  reg);
+		}
+	}
+}
+
+/**
+ * i40e_map_pf_to_vf_queues
+ * @vf: pointer to the VF info
+ *
+ * PF maps LQPs to a VF by programming VSILAN_QTABLE & VPLAN_QTABLE. This
+ * function takes care of the second part VPLAN_QTABLE & completes VF mappings.
+ **/
+static void i40e_map_pf_to_vf_queues(struct i40e_vf *vf)
+{
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_hw *hw = &pf->hw;
+	u32 reg, total_qps = 0;
+	u32 qps, num_tc = 1; /* VF has at least one traffic class */
+	u16 vsi_id, qid;
+	int i, j;
+
+	if (vf->adq_enabled)
+		num_tc = vf->num_tc;
+
+	for (i = 0; i < num_tc; i++) {
+		if (vf->adq_enabled) {
+			qps = vf->ch[i].num_qps;
+			vsi_id =  vf->ch[i].vsi_id;
+		} else {
+			qps = pf->vsi[vf->lan_vsi_idx]->alloc_queue_pairs;
+			vsi_id = vf->lan_vsi_id;
+		}
+
+		for (j = 0; j < qps; j++) {
+			qid = i40e_vc_get_pf_queue_id(vf, vsi_id, j);
+
+			reg = (qid & I40E_VPLAN_QTABLE_QINDEX_MASK);
+			wr32(hw, I40E_VPLAN_QTABLE(total_qps, vf->vf_id),
+			     reg);
+			total_qps++;
+		}
+	}
+}
+
 /**
  * i40e_enable_vf_mappings
  * @vf: pointer to the VF info
@@ -751,8 +877,7 @@ static void i40e_enable_vf_mappings(struct i40e_vf *vf)
 {
 	struct i40e_pf *pf = vf->pf;
 	struct i40e_hw *hw = &pf->hw;
-	u32 reg, total_queue_pairs = 0;
-	int j;
+	u32 reg;
 
 	/* Tell the hardware we're using noncontiguous mapping. HW requires
 	 * that VF queues be mapped using this method, even when they are
@@ -765,30 +890,8 @@ static void i40e_enable_vf_mappings(struct i40e_vf *vf)
 	reg = I40E_VPLAN_MAPENA_TXRX_ENA_MASK;
 	wr32(hw, I40E_VPLAN_MAPENA(vf->vf_id), reg);
 
-	/* map PF queues to VF queues */
-	for (j = 0; j < pf->vsi[vf->lan_vsi_idx]->alloc_queue_pairs; j++) {
-		u16 qid = i40e_vc_get_pf_queue_id(vf, vf->lan_vsi_id, j);
-
-		reg = (qid & I40E_VPLAN_QTABLE_QINDEX_MASK);
-		wr32(hw, I40E_VPLAN_QTABLE(total_queue_pairs, vf->vf_id), reg);
-		total_queue_pairs++;
-	}
-
-	/* map PF queues to VSI */
-	for (j = 0; j < 7; j++) {
-		if (j * 2 >= pf->vsi[vf->lan_vsi_idx]->alloc_queue_pairs) {
-			reg = 0x07FF07FF;	/* unused */
-		} else {
-			u16 qid = i40e_vc_get_pf_queue_id(vf, vf->lan_vsi_id,
-							  j * 2);
-			reg = qid;
-			qid = i40e_vc_get_pf_queue_id(vf, vf->lan_vsi_id,
-						      (j * 2) + 1);
-			reg |= qid << 16;
-		}
-		i40e_write_rx_ctl(hw, I40E_VSILAN_QTABLE(j, vf->lan_vsi_id),
-				  reg);
-	}
+	i40e_map_pf_to_vf_queues(vf);
+	i40e_map_pf_queues_to_vsi(vf);
 
 	i40e_flush(hw);
 }
@@ -824,7 +927,7 @@ static void i40e_free_vf_res(struct i40e_vf *vf)
 	struct i40e_pf *pf = vf->pf;
 	struct i40e_hw *hw = &pf->hw;
 	u32 reg_idx, reg;
-	int i, msix_vf;
+	int i, j, msix_vf;
 
 	/* Start by disabling VF's configuration API to prevent the OS from
 	 * accessing the VF's VSI after it's freed / invalidated.
@@ -846,6 +949,20 @@ static void i40e_free_vf_res(struct i40e_vf *vf)
 		vf->lan_vsi_id = 0;
 		vf->num_mac = 0;
 	}
+
+	/* do the accounting and remove additional ADq VSI's */
+	if (vf->adq_enabled && vf->ch[0].vsi_idx) {
+		for (j = 0; j < vf->num_tc; j++) {
+			/* At this point VSI0 is already released so don't
+			 * release it again and only clear their values in
+			 * structure variables
+			 */
+			if (j)
+				i40e_vsi_release(pf->vsi[vf->ch[j].vsi_idx]);
+			vf->ch[j].vsi_idx = 0;
+			vf->ch[j].vsi_id = 0;
+		}
+	}
 	msix_vf = pf->hw.func_caps.num_msix_vectors_vf;
 
 	/* disable interrupts so the VF starts in a known state */
@@ -891,7 +1008,7 @@ static int i40e_alloc_vf_res(struct i40e_vf *vf)
 {
 	struct i40e_pf *pf = vf->pf;
 	int total_queue_pairs = 0;
-	int ret;
+	int ret, idx;
 
 	if (vf->num_req_queues &&
 	    vf->num_req_queues <= pf->queues_left + I40E_DEFAULT_QUEUES_PER_VF)
@@ -900,11 +1017,30 @@ static int i40e_alloc_vf_res(struct i40e_vf *vf)
 		pf->num_vf_qps = I40E_DEFAULT_QUEUES_PER_VF;
 
 	/* allocate hw vsi context & associated resources */
-	ret = i40e_alloc_vsi_res(vf, I40E_VSI_SRIOV);
+	ret = i40e_alloc_vsi_res(vf, 0);
 	if (ret)
 		goto error_alloc;
 	total_queue_pairs += pf->vsi[vf->lan_vsi_idx]->alloc_queue_pairs;
 
+	/* allocate additional VSIs based on tc information for ADq */
+	if (vf->adq_enabled) {
+		if (pf->queues_left >=
+		    (I40E_MAX_VF_QUEUES - I40E_DEFAULT_QUEUES_PER_VF)) {
+			/* TC 0 always belongs to VF VSI */
+			for (idx = 1; idx < vf->num_tc; idx++) {
+				ret = i40e_alloc_vsi_res(vf, idx);
+				if (ret)
+					goto error_alloc;
+			}
+			/* send correct number of queues */
+			total_queue_pairs = I40E_MAX_VF_QUEUES;
+		} else {
+			dev_info(&pf->pdev->dev, "VF %d: Not enough queues to allocate, disabling ADq\n",
+				 vf->vf_id);
+			vf->adq_enabled = false;
+		}
+	}
+
 	/* We account for each VF to get a default number of queue pairs.  If
 	 * the VF has now requested more, we need to account for that to make
 	 * certain we never request more queues than we actually have left in
@@ -1631,6 +1767,9 @@ static int i40e_vc_get_vf_resources_msg(struct i40e_vf *vf, u8 *msg)
 	if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_REQ_QUEUES)
 		vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_REQ_QUEUES;
 
+	if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_ADQ)
+		vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_ADQ;
+
 	vfres->num_vsis = num_vsis;
 	vfres->num_queue_pairs = vf->num_queue_pairs;
 	vfres->max_vectors = pf->hw.func_caps.num_msix_vectors_vf;
@@ -1855,27 +1994,37 @@ static int i40e_vc_config_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 	    (struct virtchnl_vsi_queue_config_info *)msg;
 	struct virtchnl_queue_pair_info *qpi;
 	struct i40e_pf *pf = vf->pf;
-	u16 vsi_id, vsi_queue_id;
+	u16 vsi_id, vsi_queue_id = 0;
 	i40e_status aq_ret = 0;
-	int i;
+	int i, j = 0, idx = 0;
+
+	vsi_id = qci->vsi_id;
 
 	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
 		aq_ret = I40E_ERR_PARAM;
 		goto error_param;
 	}
 
-	vsi_id = qci->vsi_id;
 	if (!i40e_vc_isvalid_vsi_id(vf, vsi_id)) {
 		aq_ret = I40E_ERR_PARAM;
 		goto error_param;
 	}
+
 	for (i = 0; i < qci->num_queue_pairs; i++) {
 		qpi = &qci->qpair[i];
-		vsi_queue_id = qpi->txq.queue_id;
-		if ((qpi->txq.vsi_id != vsi_id) ||
-		    (qpi->rxq.vsi_id != vsi_id) ||
-		    (qpi->rxq.queue_id != vsi_queue_id) ||
-		    !i40e_vc_isvalid_queue_id(vf, vsi_id, vsi_queue_id)) {
+
+		if (!vf->adq_enabled) {
+			vsi_queue_id = qpi->txq.queue_id;
+
+			if (qpi->txq.vsi_id != qci->vsi_id ||
+			    qpi->rxq.vsi_id != qci->vsi_id ||
+			    qpi->rxq.queue_id != vsi_queue_id) {
+				aq_ret = I40E_ERR_PARAM;
+				goto error_param;
+			}
+		}
+
+		if (!i40e_vc_isvalid_queue_id(vf, vsi_id, vsi_queue_id)) {
 			aq_ret = I40E_ERR_PARAM;
 			goto error_param;
 		}
@@ -1887,9 +2036,33 @@ static int i40e_vc_config_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 			aq_ret = I40E_ERR_PARAM;
 			goto error_param;
 		}
+
+		/* For ADq there can be up to 4 VSIs with max 4 queues each.
+		 * VF does not know about these additional VSIs and all
+		 * it cares is about its own queues. PF configures these queues
+		 * to its appropriate VSIs based on TC mapping
+		 **/
+		if (vf->adq_enabled) {
+			if (j == (vf->ch[idx].num_qps - 1)) {
+				idx++;
+				j = 0; /* resetting the queue count */
+				vsi_queue_id = 0;
+			} else {
+				j++;
+				vsi_queue_id++;
+			}
+			vsi_id = vf->ch[idx].vsi_id;
+		}
 	}
 	/* set vsi num_queue_pairs in use to num configured by VF */
-	pf->vsi[vf->lan_vsi_idx]->num_queue_pairs = qci->num_queue_pairs;
+	if (!vf->adq_enabled) {
+		pf->vsi[vf->lan_vsi_idx]->num_queue_pairs =
+			qci->num_queue_pairs;
+	} else {
+		for (i = 0; i < vf->num_tc; i++)
+			pf->vsi[vf->ch[i].vsi_idx]->num_queue_pairs =
+			       vf->ch[i].num_qps;
+	}
 
 error_param:
 	/* send the response to the VF */
@@ -1897,6 +2070,33 @@ static int i40e_vc_config_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 				       aq_ret);
 }
 
+/**
+ * i40e_validate_queue_map
+ * @vsi_id: vsi id
+ * @queuemap: Tx or Rx queue map
+ *
+ * check if Tx or Rx queue map is valid
+ **/
+static int i40e_validate_queue_map(struct i40e_vf *vf, u16 vsi_id,
+				   unsigned long queuemap)
+{
+	u16 vsi_queue_id, queue_id;
+
+	for_each_set_bit(vsi_queue_id, &queuemap, I40E_MAX_VSI_QP) {
+		if (vf->adq_enabled) {
+			vsi_id = vf->ch[vsi_queue_id / I40E_MAX_VF_VSI].vsi_id;
+			queue_id = (vsi_queue_id % I40E_DEFAULT_QUEUES_PER_VF);
+		} else {
+			queue_id = vsi_queue_id;
+		}
+
+		if (!i40e_vc_isvalid_queue_id(vf, vsi_id, queue_id))
+			return -EINVAL;
+	}
+
+	return 0;
+}
+
 /**
  * i40e_vc_config_irq_map_msg
  * @vf: pointer to the VF info
@@ -1911,9 +2111,8 @@ static int i40e_vc_config_irq_map_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 	struct virtchnl_irq_map_info *irqmap_info =
 	    (struct virtchnl_irq_map_info *)msg;
 	struct virtchnl_vector_map *map;
-	u16 vsi_id, vsi_queue_id, vector_id;
+	u16 vsi_id, vector_id;
 	i40e_status aq_ret = 0;
-	unsigned long tempmap;
 	int i;
 
 	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
@@ -1923,7 +2122,6 @@ static int i40e_vc_config_irq_map_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 
 	for (i = 0; i < irqmap_info->num_vectors; i++) {
 		map = &irqmap_info->vecmap[i];
-
 		vector_id = map->vector_id;
 		vsi_id = map->vsi_id;
 		/* validate msg params */
@@ -1933,23 +2131,14 @@ static int i40e_vc_config_irq_map_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 			goto error_param;
 		}
 
-		/* lookout for the invalid queue index */
-		tempmap = map->rxq_map;
-		for_each_set_bit(vsi_queue_id, &tempmap, I40E_MAX_VSI_QP) {
-			if (!i40e_vc_isvalid_queue_id(vf, vsi_id,
-						      vsi_queue_id)) {
-				aq_ret = I40E_ERR_PARAM;
-				goto error_param;
-			}
+		if (i40e_validate_queue_map(vf, vsi_id, map->rxq_map)) {
+			aq_ret = I40E_ERR_PARAM;
+			goto error_param;
 		}
 
-		tempmap = map->txq_map;
-		for_each_set_bit(vsi_queue_id, &tempmap, I40E_MAX_VSI_QP) {
-			if (!i40e_vc_isvalid_queue_id(vf, vsi_id,
-						      vsi_queue_id)) {
-				aq_ret = I40E_ERR_PARAM;
-				goto error_param;
-			}
+		if (i40e_validate_queue_map(vf, vsi_id, map->txq_map)) {
+			aq_ret = I40E_ERR_PARAM;
+			goto error_param;
 		}
 
 		i40e_config_irq_link_list(vf, vsi_id, map);
@@ -1975,6 +2164,7 @@ static int i40e_vc_enable_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 	struct i40e_pf *pf = vf->pf;
 	u16 vsi_id = vqs->vsi_id;
 	i40e_status aq_ret = 0;
+	int i;
 
 	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
 		aq_ret = I40E_ERR_PARAM;
@@ -1993,6 +2183,16 @@ static int i40e_vc_enable_queues_msg(struct i40e_vf *vf, u8 *msg, u16 msglen)
 
 	if (i40e_vsi_start_rings(pf->vsi[vf->lan_vsi_idx]))
 		aq_ret = I40E_ERR_TIMEOUT;
+
+	/* need to start the rings for additional ADq VSI's as well */
+	if (vf->adq_enabled) {
+		/* zero belongs to LAN VSI */
+		for (i = 1; i < vf->num_tc; i++) {
+			if (i40e_vsi_start_rings(pf->vsi[vf->ch[i].vsi_idx]))
+				aq_ret = I40E_ERR_TIMEOUT;
+		}
+	}
+
 error_param:
 	/* send the response to the VF */
 	return i40e_vc_send_resp_to_vf(vf, VIRTCHNL_OP_ENABLE_QUEUES,
@@ -2687,6 +2887,97 @@ static int i40e_vc_disable_vlan_stripping(struct i40e_vf *vf, u8 *msg,
 				       aq_ret);
 }
 
+/**
+ * i40e_vc_add_qch_msg: Add queue channel and enable ADq
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer
+ **/
+static int i40e_vc_add_qch_msg(struct i40e_vf *vf, u8 *msg)
+{
+	struct virtchnl_tc_info *tci =
+		(struct virtchnl_tc_info *)msg;
+	struct i40e_pf *pf = vf->pf;
+	int i, adq_request_qps = 0;
+	i40e_status aq_ret = 0;
+
+	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	/* ADq cannot be applied if spoof check is ON */
+	if (vf->spoofchk) {
+		dev_err(&pf->pdev->dev,
+			"Spoof check is ON, turn it OFF to enable ADq\n");
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (!(vf->driver_caps & VIRTCHNL_VF_OFFLOAD_ADQ)) {
+		dev_err(&pf->pdev->dev,
+			"VF %d attempting to enable ADq, but hasn't properly negotiated that capability\n",
+			vf->vf_id);
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	/* max number of traffic classes for VF currently capped at 4 */
+	if (!tci->num_tc || tci->num_tc > I40E_MAX_VF_VSI) {
+		dev_err(&pf->pdev->dev,
+			"VF %d trying to set %u TCs, valid range 1-4 TCs per VF\n",
+			vf->vf_id, tci->num_tc);
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	/* validate queues for each TC */
+	for (i = 0; i < tci->num_tc; i++)
+		if (!tci->list[i].count ||
+		    tci->list[i].count > I40E_DEFAULT_QUEUES_PER_VF) {
+			dev_err(&pf->pdev->dev,
+				"VF %d: TC %d trying to set %u queues, valid range 1-4 queues per TC\n",
+				vf->vf_id, i, tci->list[i].count);
+			aq_ret = I40E_ERR_PARAM;
+			goto err;
+		}
+
+	/* need Max VF queues but already have default number of queues */
+	adq_request_qps = I40E_MAX_VF_QUEUES - I40E_DEFAULT_QUEUES_PER_VF;
+
+	if (pf->queues_left < adq_request_qps) {
+		dev_err(&pf->pdev->dev,
+			"No queues left to allocate to VF %d\n",
+			vf->vf_id);
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	} else {
+		/* we need to allocate max VF queues to enable ADq so as to
+		 * make sure ADq enabled VF always gets back queues when it
+		 * goes through a reset.
+		 */
+		vf->num_queue_pairs = I40E_MAX_VF_QUEUES;
+	}
+
+	/* parse data from the queue channel info */
+	vf->num_tc = tci->num_tc;
+	for (i = 0; i < vf->num_tc; i++)
+		vf->ch[i].num_qps = tci->list[i].count;
+
+	/* set this flag only after making sure all inputs are sane */
+	vf->adq_enabled = true;
+
+	/* reset the VF in order to allocate resources */
+	i40e_vc_notify_vf_reset(vf);
+	i40e_reset_vf(vf, false);
+
+	return I40E_SUCCESS;
+
+	/* send the response to the VF */
+err:
+	return i40e_vc_send_resp_to_vf(vf, VIRTCHNL_OP_ENABLE_CHANNELS,
+				       aq_ret);
+}
+
 /**
  * i40e_vc_process_vf_msg
  * @pf: pointer to the PF structure
@@ -2816,7 +3107,9 @@ int i40e_vc_process_vf_msg(struct i40e_pf *pf, s16 vf_id, u32 v_opcode,
 	case VIRTCHNL_OP_REQUEST_QUEUES:
 		ret = i40e_vc_request_queues_msg(vf, msg, msglen);
 		break;
-
+	case VIRTCHNL_OP_ENABLE_CHANNELS:
+		ret = i40e_vc_add_qch_msg(vf, msg);
+		break;
 	case VIRTCHNL_OP_UNKNOWN:
 	default:
 		dev_err(&pf->pdev->dev, "Unsupported opcode %d from VF %d\n",
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
index 5efc4f92bb37..67543dbefcd8 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
@@ -69,6 +69,18 @@ enum i40e_vf_capabilities {
 	I40E_VIRTCHNL_VF_CAP_IWARP,
 };
 
+/* In ADq, max 4 VSI's can be allocated per VF including primary VF VSI.
+ * These variables are used to store indices, id's and number of queues
+ * for each VSI including that of primary VF VSI. Each Traffic class is
+ * termed as channel and each channel can in-turn have 4 queues which
+ * means max 16 queues overall per VF.
+ */
+struct i40evf_channel {
+	u16 vsi_idx; /* index in PF struct for all channel VSIs */
+	u16 vsi_id; /* VSI ID used by firmware */
+	u16 num_qps; /* number of queue pairs requested by user */
+};
+
 /* VF information structure */
 struct i40e_vf {
 	struct i40e_pf *pf;
@@ -111,6 +123,11 @@ struct i40e_vf {
 	u16 num_mac;
 	u16 num_vlan;
 
+	/* ADq related variables */
+	bool adq_enabled; /* flag to enable adq */
+	u8 num_tc;
+	struct i40evf_channel ch[I40E_MAX_VF_VSI];
+
 	/* RDMA Client */
 	struct virtchnl_iwarp_qvlist_info *qvlist_info;
 };
-- 
2.14.3

[net-next 05/12] i40evf: Alloc queues for ADq on VF

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

This patch allocates number of queues requested by the user as a part
of TC command when ADq is enabled on a VF.

In order to be consistent in design with PF implementation of ADq,
don't allow to set channels via ethtool from VF when ADq is already
enabled. This means the users will not be able to change the number of
queues/channels via ethtool for a VF when ADq is ON. In order to be
able to use set channels, users will be required to disable ADq first
and then try setting the channels again.

When ADq is enabled on VF, it goes through a reset during which VSIs
and queues are re-configured. Meanwhile if we receive link status
message from PF even before the queues are re-configured, just ignore
this link up message.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c |  6 ++++++
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 14 ++++++++++++++
 2 files changed, 20 insertions(+)

diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c b/drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c
index aded3ad7763e..e6793255de0b 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_ethtool.c
@@ -695,6 +695,12 @@ static int i40evf_set_channels(struct net_device *netdev,
 		return -EINVAL;
 	}
 
+	if ((adapter->vf_res->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) &&
+	    adapter->num_tc) {
+		dev_info(&adapter->pdev->dev, "Cannot set channels since ADq is enabled.\n");
+		return -EINVAL;
+	}
+
 	/* All of these should have already been checked by ethtool before this
 	 * even gets to us, but just to be sure.
 	 */
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 099d4f59e445..8a26393459ef 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -1164,6 +1164,9 @@ static int i40evf_alloc_queues(struct i40evf_adapter *adapter)
 	 */
 	if (adapter->num_req_queues)
 		num_active_queues = adapter->num_req_queues;
+	else if ((adapter->vf_res->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) &&
+		 adapter->num_tc)
+		num_active_queues = adapter->ch_config.total_qps;
 	else
 		num_active_queues = min_t(int,
 					  adapter->vsi_res->num_queue_pairs,
@@ -1491,6 +1494,16 @@ int i40evf_init_interrupt_scheme(struct i40evf_adapter *adapter)
 		goto err_alloc_q_vectors;
 	}
 
+	/* If we've made it so far while ADq flag being ON, then we haven't
+	 * bailed out anywhere in middle. And ADq isn't just enabled but actual
+	 * resources have been allocated in the reset path.
+	 * Now we can truly claim that ADq is enabled.
+	 */
+	if ((adapter->vf_res->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) &&
+	    adapter->num_tc)
+		dev_info(&adapter->pdev->dev, "ADq Enabled, %u TCs created",
+			 adapter->num_tc);
+
 	dev_info(&adapter->pdev->dev, "Multiqueue %s: Queue pair count = %u",
 		 (adapter->num_active_queues > 1) ? "Enabled" : "Disabled",
 		 adapter->num_active_queues);
@@ -3263,6 +3276,7 @@ static void i40evf_remove(struct pci_dev *pdev)
 	/* Shut down all the garbage mashers on the detention level */
 	adapter->state = __I40EVF_REMOVE;
 	adapter->aq_required = 0;
+	adapter->flags &= ~I40EVF_FLAG_REINIT_ITR_NEEDED;
 	i40evf_request_reset(adapter);
 	msleep(50);
 	/* If the FW isn't responding, kick it once, but only once. */
-- 
2.14.3

[net-next 06/12] i40e: Delete queue channel for ADq on VF

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

This patch takes care of freeing up all the VSIs, queues and
other ADq related software and hardware resources, when a user
requests for deletion of ADq on VF.

Example command:
tc qdisc del dev eth0 root

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 63 ++++++++++++++++++++++
 1 file changed, 63 insertions(+)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index 7452d4387818..a5565d07f569 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -1672,6 +1672,27 @@ static int i40e_vc_get_version_msg(struct i40e_vf *vf, u8 *msg)
 				      sizeof(struct virtchnl_version_info));
 }
 
+/**
+ * i40e_del_qch - delete all the additional VSIs created as a part of ADq
+ * @vf: pointer to VF structure
+ **/
+static void i40e_del_qch(struct i40e_vf *vf)
+{
+	struct i40e_pf *pf = vf->pf;
+	int i;
+
+	/* first element in the array belongs to primary VF VSI and we shouldn't
+	 * delete it. We should however delete the rest of the VSIs created
+	 */
+	for (i = 1; i < vf->num_tc; i++) {
+		if (vf->ch[i].vsi_idx) {
+			i40e_vsi_release(pf->vsi[vf->ch[i].vsi_idx]);
+			vf->ch[i].vsi_idx = 0;
+			vf->ch[i].vsi_id = 0;
+		}
+	}
+}
+
 /**
  * i40e_vc_get_vf_resources_msg
  * @vf: pointer to the VF info
@@ -2978,6 +2999,45 @@ static int i40e_vc_add_qch_msg(struct i40e_vf *vf, u8 *msg)
 				       aq_ret);
 }
 
+/**
+ * i40e_vc_del_qch_msg
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer
+ **/
+static int i40e_vc_del_qch_msg(struct i40e_vf *vf, u8 *msg)
+{
+	struct i40e_pf *pf = vf->pf;
+	i40e_status aq_ret = 0;
+
+	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (vf->adq_enabled) {
+		i40e_del_qch(vf);
+		vf->adq_enabled = false;
+		vf->num_tc = 0;
+		dev_info(&pf->pdev->dev,
+			 "Deleting Queue Channels for ADq on VF %d\n",
+			 vf->vf_id);
+	} else {
+		dev_info(&pf->pdev->dev, "VF %d trying to delete queue channels but ADq isn't enabled\n",
+			 vf->vf_id);
+		aq_ret = I40E_ERR_PARAM;
+	}
+
+	/* reset the VF in order to allocate resources */
+	i40e_vc_notify_vf_reset(vf);
+	i40e_reset_vf(vf, false);
+
+	return I40E_SUCCESS;
+
+err:
+	return i40e_vc_send_resp_to_vf(vf, VIRTCHNL_OP_DISABLE_CHANNELS,
+				       aq_ret);
+}
+
 /**
  * i40e_vc_process_vf_msg
  * @pf: pointer to the PF structure
@@ -3110,6 +3170,9 @@ int i40e_vc_process_vf_msg(struct i40e_pf *pf, s16 vf_id, u32 v_opcode,
 	case VIRTCHNL_OP_ENABLE_CHANNELS:
 		ret = i40e_vc_add_qch_msg(vf, msg);
 		break;
+	case VIRTCHNL_OP_DISABLE_CHANNELS:
+		ret = i40e_vc_del_qch_msg(vf, msg);
+		break;
 	case VIRTCHNL_OP_UNKNOWN:
 	default:
 		dev_err(&pf->pdev->dev, "Unsupported opcode %d from VF %d\n",
-- 
2.14.3

[net-next 07/12] i40evf: Add support to configure bw via tc tool

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch adds support to configure bandwidth for the traffic
classes via tc tool. The required information is passed to the PF
which is used in the process of setting up the traffic classes.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40evf/i40evf.h         |  1 +
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 66 +++++++++++++++++++++-
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    |  3 +
 3 files changed, 67 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40evf/i40evf.h b/drivers/net/ethernet/intel/i40evf/i40evf.h
index c8d68c8d2c33..150d676ab87c 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf.h
+++ b/drivers/net/ethernet/intel/i40evf/i40evf.h
@@ -107,6 +107,7 @@ struct i40e_vsi {
 
 #define I40EVF_HKEY_ARRAY_SIZE ((I40E_VFQF_HKEY_MAX_INDEX + 1) * 4)
 #define I40EVF_HLUT_ARRAY_SIZE ((I40E_VFQF_HLUT_MAX_INDEX + 1) * 4)
+#define I40EVF_MBPS_DIVISOR	125000 /* divisor to convert to Mbps */
 
 /* MAX_MSIX_Q_VECTORS of these are allocated,
  * but we only use one per queue-specific vector.
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 8a26393459ef..2f677a2a4b5c 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -2235,6 +2235,48 @@ void i40evf_free_all_rx_resources(struct i40evf_adapter *adapter)
 			i40evf_free_rx_resources(&adapter->rx_rings[i]);
 }
 
+/**
+ * i40evf_validate_tx_bandwidth - validate the max Tx bandwidth
+ * @adapter: board private structure
+ * @max_tx_rate: max Tx bw for a tc
+ **/
+static int i40evf_validate_tx_bandwidth(struct i40evf_adapter *adapter,
+					u64 max_tx_rate)
+{
+	int speed = 0, ret = 0;
+
+	switch (adapter->link_speed) {
+	case I40E_LINK_SPEED_40GB:
+		speed = 40000;
+		break;
+	case I40E_LINK_SPEED_25GB:
+		speed = 25000;
+		break;
+	case I40E_LINK_SPEED_20GB:
+		speed = 20000;
+		break;
+	case I40E_LINK_SPEED_10GB:
+		speed = 10000;
+		break;
+	case I40E_LINK_SPEED_1GB:
+		speed = 1000;
+		break;
+	case I40E_LINK_SPEED_100MB:
+		speed = 100;
+		break;
+	default:
+		break;
+	}
+
+	if (max_tx_rate > speed) {
+		dev_err(&adapter->pdev->dev,
+			"Invalid tx rate specified\n");
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
 /**
  * i40evf_validate_channel_config - validate queue mapping info
  * @adapter: board private structure
@@ -2247,7 +2289,10 @@ void i40evf_free_all_rx_resources(struct i40evf_adapter *adapter)
 static int i40evf_validate_ch_config(struct i40evf_adapter *adapter,
 				     struct tc_mqprio_qopt_offload *mqprio_qopt)
 {
+	u64 total_max_rate = 0;
 	int i, num_qps = 0;
+	u64 tx_rate = 0;
+	int ret = 0;
 
 	if (mqprio_qopt->qopt.num_tc > I40EVF_MAX_TRAFFIC_CLASS ||
 	    mqprio_qopt->qopt.num_tc < 1)
@@ -2255,16 +2300,24 @@ static int i40evf_validate_ch_config(struct i40evf_adapter *adapter,
 
 	for (i = 0; i <= mqprio_qopt->qopt.num_tc - 1; i++) {
 		if (!mqprio_qopt->qopt.count[i] ||
-		    mqprio_qopt->min_rate[i] ||
-		    mqprio_qopt->max_rate[i] ||
 		    mqprio_qopt->qopt.offset[i] != num_qps)
 			return -EINVAL;
+		if (mqprio_qopt->min_rate[i]) {
+			dev_err(&adapter->pdev->dev,
+				"Invalid min tx rate (greater than 0) specified\n");
+			return -EINVAL;
+		}
+		/*convert to Mbps */
+		tx_rate = div_u64(mqprio_qopt->max_rate[i],
+				  I40EVF_MBPS_DIVISOR);
+		total_max_rate += tx_rate;
 		num_qps += mqprio_qopt->qopt.count[i];
 	}
 	if (num_qps > MAX_QUEUES)
 		return -EINVAL;
 
-	return 0;
+	ret = i40evf_validate_tx_bandwidth(adapter, total_max_rate);
+	return ret;
 }
 
 /**
@@ -2285,6 +2338,7 @@ static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
 	struct virtchnl_vf_resource *vfres = adapter->vf_res;
 	u8 num_tc = 0, total_qps = 0;
 	int ret = 0, netdev_tc = 0;
+	u64 max_tx_rate;
 	u16 mode;
 	int i;
 
@@ -2332,6 +2386,12 @@ static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
 				adapter->ch_config.ch_info[i].offset =
 					mqprio_qopt->qopt.offset[i];
 				total_qps += mqprio_qopt->qopt.count[i];
+				max_tx_rate = mqprio_qopt->max_rate[i];
+				/* convert to Mbps */
+				max_tx_rate = div_u64(max_tx_rate,
+						      I40EVF_MBPS_DIVISOR);
+				adapter->ch_config.ch_info[i].max_tx_rate =
+					max_tx_rate;
 			} else {
 				adapter->ch_config.ch_info[i].count = 1;
 				adapter->ch_config.ch_info[i].offset = 0;
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index f6c56141b4fe..647ea104d435 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -1003,6 +1003,9 @@ void i40evf_enable_channels(struct i40evf_adapter *adapter)
 	for (i = 0; i < vti->num_tc; i++) {
 		vti->list[i].count = adapter->ch_config.ch_info[i].count;
 		vti->list[i].offset = adapter->ch_config.ch_info[i].offset;
+		vti->list[i].pad = 0;
+		vti->list[i].max_tx_rate =
+				adapter->ch_config.ch_info[i].max_tx_rate;
 	}
 
 	adapter->ch_config.state = __I40EVF_TC_RUNNING;
-- 
2.14.3

[net-next 08/12] i40e: Service request to configure bandwidth for ADq on a VF

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

This patch handles the request from ADq enabled VF to allocate
bandwidth to each traffic class which means for each VSI.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 56 ++++++++++++++++++++--
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h |  1 +
 2 files changed, 54 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index a5565d07f569..b4bad70c82cd 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -711,6 +711,7 @@ static int i40e_alloc_vsi_res(struct i40e_vf *vf, u8 idx)
 	struct i40e_mac_filter *f = NULL;
 	struct i40e_pf *pf = vf->pf;
 	struct i40e_vsi *vsi;
+	u64 max_tx_rate = 0;
 	int ret = 0;
 
 	vsi = i40e_vsi_setup(pf, I40E_VSI_SRIOV, pf->vsi[pf->lan_vsi]->seid,
@@ -770,8 +771,15 @@ static int i40e_alloc_vsi_res(struct i40e_vf *vf, u8 idx)
 
 	/* Set VF bandwidth if specified */
 	if (vf->tx_rate) {
+		max_tx_rate = vf->tx_rate;
+	} else if (vf->ch[idx].max_tx_rate) {
+		max_tx_rate = vf->ch[idx].max_tx_rate;
+	}
+
+	if (max_tx_rate) {
+		max_tx_rate = div_u64(max_tx_rate, I40E_BW_CREDIT_DIVISOR);
 		ret = i40e_aq_config_vsi_bw_limit(&pf->hw, vsi->seid,
-						  vf->tx_rate / 50, 0, NULL);
+						  max_tx_rate, 0, NULL);
 		if (ret)
 			dev_err(&pf->pdev->dev, "Unable to set tx rate, VF %d, error code %d.\n",
 				vf->vf_id, ret);
@@ -2918,7 +2926,8 @@ static int i40e_vc_add_qch_msg(struct i40e_vf *vf, u8 *msg)
 	struct virtchnl_tc_info *tci =
 		(struct virtchnl_tc_info *)msg;
 	struct i40e_pf *pf = vf->pf;
-	int i, adq_request_qps = 0;
+	struct i40e_link_status *ls = &pf->hw.phy.link_info;
+	int i, adq_request_qps = 0, speed = 0;
 	i40e_status aq_ret = 0;
 
 	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
@@ -2979,10 +2988,51 @@ static int i40e_vc_add_qch_msg(struct i40e_vf *vf, u8 *msg)
 		vf->num_queue_pairs = I40E_MAX_VF_QUEUES;
 	}
 
+	/* get link speed in MB to validate rate limit */
+	switch (ls->link_speed) {
+	case VIRTCHNL_LINK_SPEED_100MB:
+		speed = SPEED_100;
+		break;
+	case VIRTCHNL_LINK_SPEED_1GB:
+		speed = SPEED_1000;
+		break;
+	case VIRTCHNL_LINK_SPEED_10GB:
+		speed = SPEED_10000;
+		break;
+	case VIRTCHNL_LINK_SPEED_20GB:
+		speed = SPEED_20000;
+		break;
+	case VIRTCHNL_LINK_SPEED_25GB:
+		speed = SPEED_25000;
+		break;
+	case VIRTCHNL_LINK_SPEED_40GB:
+		speed = SPEED_40000;
+		break;
+	default:
+		dev_err(&pf->pdev->dev,
+			"Cannot detect link speed\n");
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
 	/* parse data from the queue channel info */
 	vf->num_tc = tci->num_tc;
-	for (i = 0; i < vf->num_tc; i++)
+	for (i = 0; i < vf->num_tc; i++) {
+		if (tci->list[i].max_tx_rate) {
+			if (tci->list[i].max_tx_rate > speed) {
+				dev_err(&pf->pdev->dev,
+					"Invalid max tx rate %llu specified for VF %d.",
+					tci->list[i].max_tx_rate,
+					vf->vf_id);
+				aq_ret = I40E_ERR_PARAM;
+				goto err;
+			} else {
+				vf->ch[i].max_tx_rate =
+					tci->list[i].max_tx_rate;
+			}
+		}
 		vf->ch[i].num_qps = tci->list[i].count;
+	}
 
 	/* set this flag only after making sure all inputs are sane */
 	vf->adq_enabled = true;
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
index 67543dbefcd8..6e794c8a1ef3 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
@@ -79,6 +79,7 @@ struct i40evf_channel {
 	u16 vsi_idx; /* index in PF struct for all channel VSIs */
 	u16 vsi_id; /* VSI ID used by firmware */
 	u16 num_qps; /* number of queue pairs requested by user */
+	u64 max_tx_rate; /* bandwidth rate allocation for VSIs */
 };
 
 /* VF information structure */
-- 
2.14.3

[net-next 09/12] virtchnl: Add a macro to check the size of a union

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch adds a macro to check if the size of a union is correct.
It throws a divide by zero error if the union is not of the correct
size.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 include/linux/avf/virtchnl.h | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h
index 1f652ceecf35..6fe630ebbf23 100644
--- a/include/linux/avf/virtchnl.h
+++ b/include/linux/avf/virtchnl.h
@@ -140,13 +140,15 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_DISABLE_CHANNELS = 31,
 };
 
-/* This macro is used to generate a compilation error if a structure
+/* These macros are used to generate compilation errors if a structure/union
  * is not exactly the correct length. It gives a divide by zero error if the
- * structure is not of the correct size, otherwise it creates an enum that is
- * never used.
+ * structure/union is not of the correct size, otherwise it creates an enum
+ * that is never used.
  */
 #define VIRTCHNL_CHECK_STRUCT_LEN(n, X) enum virtchnl_static_assert_enum_##X \
 	{ virtchnl_static_assert_##X = (n)/((sizeof(struct X) == (n)) ? 1 : 0) }
+#define VIRTCHNL_CHECK_UNION_LEN(n, X) enum virtchnl_static_asset_enum_##X \
+	{ virtchnl_static_assert_##X = (n)/((sizeof(union X) == (n)) ? 1 : 0) }
 
 /* Virtual channel message descriptor. This overlays the admin queue
  * descriptor. All other data is passed in external buffers.
-- 
2.14.3

[net-next 10/12] virtchnl: Add filter data structures

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch adds infrastructure to send virtchnl messages to the
PF to configure filters on the VF. The patch adds a struct
called virtchnl_filter which contains information about the fields
in the user-specified tc filter.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 include/linux/avf/virtchnl.h | 59 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 59 insertions(+)

diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h
index 6fe630ebbf23..b0a7f315bfbe 100644
--- a/include/linux/avf/virtchnl.h
+++ b/include/linux/avf/virtchnl.h
@@ -138,6 +138,8 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_REQUEST_QUEUES = 29,
 	VIRTCHNL_OP_ENABLE_CHANNELS = 30,
 	VIRTCHNL_OP_DISABLE_CHANNELS = 31,
+	VIRTCHNL_OP_ADD_CLOUD_FILTER = 32,
+	VIRTCHNL_OP_DEL_CLOUD_FILTER = 33,
 };
 
 /* These macros are used to generate compilation errors if a structure/union
@@ -525,6 +527,57 @@ struct virtchnl_tc_info {
 
 VIRTCHNL_CHECK_STRUCT_LEN(24, virtchnl_tc_info);
 
+/* VIRTCHNL_ADD_CLOUD_FILTER
+ * VIRTCHNL_DEL_CLOUD_FILTER
+ * VF sends these messages to add or delete a cloud filter based on the
+ * user specified match and action filters. These structures encompass
+ * all the information that the PF needs from the VF to add/delete a
+ * cloud filter.
+ */
+
+struct virtchnl_l4_spec {
+	u8	src_mac[ETH_ALEN];
+	u8	dst_mac[ETH_ALEN];
+	__be16	vlan_id;
+	__be16	pad; /* reserved for future use */
+	__be32	src_ip[4];
+	__be32	dst_ip[4];
+	__be16	src_port;
+	__be16	dst_port;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(52, virtchnl_l4_spec);
+
+union virtchnl_flow_spec {
+	struct	virtchnl_l4_spec tcp_spec;
+	u8	buffer[128]; /* reserved for future use */
+};
+
+VIRTCHNL_CHECK_UNION_LEN(128, virtchnl_flow_spec);
+
+enum virtchnl_action {
+	/* action types */
+	VIRTCHNL_ACTION_DROP = 0,
+	VIRTCHNL_ACTION_TC_REDIRECT,
+};
+
+enum virtchnl_flow_type {
+	/* flow types */
+	VIRTCHNL_TCP_V4_FLOW = 0,
+	VIRTCHNL_TCP_V6_FLOW,
+};
+
+struct virtchnl_filter {
+	union	virtchnl_flow_spec data;
+	union	virtchnl_flow_spec mask;
+	enum	virtchnl_flow_type flow_type;
+	enum	virtchnl_action action;
+	u32	action_meta;
+	__u8	field_flags;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(272, virtchnl_filter);
+
 /* VIRTCHNL_OP_EVENT
  * PF sends this message to inform the VF driver of events that may affect it.
  * No direct response is expected from the VF, though it may generate other
@@ -753,6 +806,12 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 		break;
 	case VIRTCHNL_OP_DISABLE_CHANNELS:
 		break;
+	case VIRTCHNL_OP_ADD_CLOUD_FILTER:
+		valid_len = sizeof(struct virtchnl_filter);
+		break;
+	case VIRTCHNL_OP_DEL_CLOUD_FILTER:
+		valid_len = sizeof(struct virtchnl_filter);
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
-- 
2.14.3

[net-next 11/12] i40evf: Add support to apply cloud filters

[Jeff Kirsher]

From: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>

This patch enables a tc filter to be applied as a cloud
filter for the VF. This patch adds functions which parse the
tc filter, extract the necessary fields needed to configure the
filter and package them in a virtchnl message to be sent to the
PF to apply them.

Signed-off-by: Harshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40evf/i40evf.h         |  48 ++
 drivers/net/ethernet/intel/i40evf/i40evf_main.c    | 580 ++++++++++++++++++++-
 .../net/ethernet/intel/i40evf/i40evf_virtchnl.c    | 187 +++++++
 3 files changed, 810 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40evf/i40evf.h b/drivers/net/ethernet/intel/i40evf/i40evf.h
index 150d676ab87c..e46555ad7122 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf.h
+++ b/drivers/net/ethernet/intel/i40evf/i40evf.h
@@ -54,6 +54,8 @@
 #include <net/ip6_checksum.h>
 #include <net/pkt_cls.h>
 #include <net/udp.h>
+#include <net/tc_act/tc_gact.h>
+#include <net/tc_act/tc_mirred.h>
 
 #include "i40e_type.h"
 #include <linux/avf/virtchnl.h>
@@ -184,6 +186,14 @@ struct i40evf_channel_config {
 	u8 total_qps;
 };
 
+/* State of cloud filter */
+enum i40evf_cloud_filter_state_t {
+	__I40EVF_CF_INVALID,	 /* cloud filter not added */
+	__I40EVF_CF_ADD_PENDING, /* cloud filter pending add by the PF */
+	__I40EVF_CF_DEL_PENDING, /* cloud filter pending del by the PF */
+	__I40EVF_CF_ACTIVE,	 /* cloud filter is active */
+};
+
 /* Driver state. The order of these is important! */
 enum i40evf_state_t {
 	__I40EVF_STARTUP,		/* driver loaded, probe complete */
@@ -205,6 +215,36 @@ enum i40evf_critical_section_t {
 	__I40EVF_IN_REMOVE_TASK,	/* device being removed */
 };
 
+#define I40EVF_CLOUD_FIELD_OMAC		0x01
+#define I40EVF_CLOUD_FIELD_IMAC		0x02
+#define I40EVF_CLOUD_FIELD_IVLAN	0x04
+#define I40EVF_CLOUD_FIELD_TEN_ID	0x08
+#define I40EVF_CLOUD_FIELD_IIP		0x10
+
+#define I40EVF_CF_FLAGS_OMAC	I40EVF_CLOUD_FIELD_OMAC
+#define I40EVF_CF_FLAGS_IMAC	I40EVF_CLOUD_FIELD_IMAC
+#define I40EVF_CF_FLAGS_IMAC_IVLAN	(I40EVF_CLOUD_FIELD_IMAC |\
+					 I40EVF_CLOUD_FIELD_IVLAN)
+#define I40EVF_CF_FLAGS_IMAC_TEN_ID	(I40EVF_CLOUD_FIELD_IMAC |\
+					 I40EVF_CLOUD_FIELD_TEN_ID)
+#define I40EVF_CF_FLAGS_OMAC_TEN_ID_IMAC	(I40EVF_CLOUD_FIELD_OMAC |\
+						 I40EVF_CLOUD_FIELD_IMAC |\
+						 I40EVF_CLOUD_FIELD_TEN_ID)
+#define I40EVF_CF_FLAGS_IMAC_IVLAN_TEN_ID	(I40EVF_CLOUD_FIELD_IMAC |\
+						 I40EVF_CLOUD_FIELD_IVLAN |\
+						 I40EVF_CLOUD_FIELD_TEN_ID)
+#define I40EVF_CF_FLAGS_IIP	I40E_CLOUD_FIELD_IIP
+
+/* bookkeeping of cloud filters */
+struct i40evf_cloud_filter {
+	enum i40evf_cloud_filter_state_t state;
+	struct list_head list;
+	struct virtchnl_filter f;
+	unsigned long cookie;
+	bool del;		/* filter needs to be deleted */
+	bool add;		/* filter needs to be added */
+};
+
 /* board specific private data structure */
 struct i40evf_adapter {
 	struct timer_list watchdog_timer;
@@ -287,6 +327,8 @@ struct i40evf_adapter {
 #define I40EVF_FLAG_AQ_DISABLE_VLAN_STRIPPING	BIT(20)
 #define I40EVF_FLAG_AQ_ENABLE_CHANNELS		BIT(21)
 #define I40EVF_FLAG_AQ_DISABLE_CHANNELS		BIT(22)
+#define I40EVF_FLAG_AQ_ADD_CLOUD_FILTER		BIT(23)
+#define I40EVF_FLAG_AQ_DEL_CLOUD_FILTER		BIT(24)
 
 	/* OS defined structs */
 	struct net_device *netdev;
@@ -335,6 +377,10 @@ struct i40evf_adapter {
 	/* ADQ related members */
 	struct i40evf_channel_config ch_config;
 	u8 num_tc;
+	struct list_head cloud_filter_list;
+	/* lock to protest access to the cloud filter list */
+	spinlock_t cloud_filter_list_lock;
+	u16 num_cloud_filters;
 };
 
 
@@ -403,4 +449,6 @@ void i40evf_notify_client_open(struct i40e_vsi *vsi);
 void i40evf_notify_client_close(struct i40e_vsi *vsi, bool reset);
 void i40evf_enable_channels(struct i40evf_adapter *adapter);
 void i40evf_disable_channels(struct i40evf_adapter *adapter);
+void i40evf_add_cloud_filter(struct i40evf_adapter *adapter);
+void i40evf_del_cloud_filter(struct i40evf_adapter *adapter);
 #endif /* _I40EVF_H_ */
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_main.c b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
index 2f677a2a4b5c..4955ce3ab6a2 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_main.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_main.c
@@ -1041,6 +1041,7 @@ void i40evf_down(struct i40evf_adapter *adapter)
 	struct net_device *netdev = adapter->netdev;
 	struct i40evf_vlan_filter *vlf;
 	struct i40evf_mac_filter *f;
+	struct i40evf_cloud_filter *cf;
 
 	if (adapter->state <= __I40EVF_DOWN_PENDING)
 		return;
@@ -1064,11 +1065,18 @@ void i40evf_down(struct i40evf_adapter *adapter)
 
 	/* remove all VLAN filters */
 	list_for_each_entry(vlf, &adapter->vlan_filter_list, list) {
-		f->remove = true;
+		vlf->remove = true;
 	}
 
 	spin_unlock_bh(&adapter->mac_vlan_list_lock);
 
+	/* remove all cloud filters */
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+		cf->del = true;
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+
 	if (!(adapter->flags & I40EVF_FLAG_PF_COMMS_FAILED) &&
 	    adapter->state != __I40EVF_RESETTING) {
 		/* cancel any current operation */
@@ -1079,6 +1087,7 @@ void i40evf_down(struct i40evf_adapter *adapter)
 		 */
 		adapter->aq_required = I40EVF_FLAG_AQ_DEL_MAC_FILTER;
 		adapter->aq_required |= I40EVF_FLAG_AQ_DEL_VLAN_FILTER;
+		adapter->aq_required |= I40EVF_FLAG_AQ_DEL_CLOUD_FILTER;
 		adapter->aq_required |= I40EVF_FLAG_AQ_DISABLE_QUEUES;
 	}
 
@@ -1756,6 +1765,16 @@ static void i40evf_watchdog_task(struct work_struct *work)
 		goto watchdog_done;
 	}
 
+	if (adapter->aq_required & I40EVF_FLAG_AQ_ADD_CLOUD_FILTER) {
+		i40evf_add_cloud_filter(adapter);
+		goto watchdog_done;
+	}
+
+	if (adapter->aq_required & I40EVF_FLAG_AQ_DEL_CLOUD_FILTER) {
+		i40evf_del_cloud_filter(adapter);
+		goto watchdog_done;
+	}
+
 	schedule_delayed_work(&adapter->client_task, msecs_to_jiffies(5));
 
 	if (adapter->state == __I40EVF_RUNNING)
@@ -1779,6 +1798,7 @@ static void i40evf_disable_vf(struct i40evf_adapter *adapter)
 {
 	struct i40evf_mac_filter *f, *ftmp;
 	struct i40evf_vlan_filter *fv, *fvtmp;
+	struct i40evf_cloud_filter *cf, *cftmp;
 
 	adapter->flags |= I40EVF_FLAG_PF_COMMS_FAILED;
 
@@ -1800,7 +1820,7 @@ static void i40evf_disable_vf(struct i40evf_adapter *adapter)
 
 	spin_lock_bh(&adapter->mac_vlan_list_lock);
 
-	/* Delete all of the filters, both MAC and VLAN. */
+	/* Delete all of the filters */
 	list_for_each_entry_safe(f, ftmp, &adapter->mac_filter_list, list) {
 		list_del(&f->list);
 		kfree(f);
@@ -1813,6 +1833,14 @@ static void i40evf_disable_vf(struct i40evf_adapter *adapter)
 
 	spin_unlock_bh(&adapter->mac_vlan_list_lock);
 
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	list_for_each_entry_safe(cf, cftmp, &adapter->cloud_filter_list, list) {
+		list_del(&cf->list);
+		kfree(cf);
+		adapter->num_cloud_filters--;
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+
 	i40evf_free_misc_irq(adapter);
 	i40evf_reset_interrupt_capability(adapter);
 	i40evf_free_queues(adapter);
@@ -1842,9 +1870,11 @@ static void i40evf_reset_task(struct work_struct *work)
 	struct i40evf_adapter *adapter = container_of(work,
 						      struct i40evf_adapter,
 						      reset_task);
+	struct virtchnl_vf_resource *vfres = adapter->vf_res;
 	struct net_device *netdev = adapter->netdev;
 	struct i40e_hw *hw = &adapter->hw;
 	struct i40evf_vlan_filter *vlf;
+	struct i40evf_cloud_filter *cf;
 	struct i40evf_mac_filter *f;
 	u32 reg_val;
 	int i = 0, err;
@@ -1969,8 +1999,19 @@ static void i40evf_reset_task(struct work_struct *work)
 
 	spin_unlock_bh(&adapter->mac_vlan_list_lock);
 
+	/* check if TCs are running and re-add all cloud filters */
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	if ((vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) &&
+	    adapter->num_tc) {
+		list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+			cf->add = true;
+		}
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+
 	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_MAC_FILTER;
 	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_VLAN_FILTER;
+	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_CLOUD_FILTER;
 	i40evf_misc_irq_enable(adapter);
 
 	mod_timer(&adapter->watchdog_timer, jiffies + 2);
@@ -2320,6 +2361,24 @@ static int i40evf_validate_ch_config(struct i40evf_adapter *adapter,
 	return ret;
 }
 
+/**
+ * i40evf_del_all_cloud_filters - delete all cloud filters
+ * on the traffic classes
+ **/
+static void i40evf_del_all_cloud_filters(struct i40evf_adapter *adapter)
+{
+	struct i40evf_cloud_filter *cf, *cftmp;
+
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	list_for_each_entry_safe(cf, cftmp, &adapter->cloud_filter_list,
+				 list) {
+		list_del(&cf->list);
+		kfree(cf);
+		adapter->num_cloud_filters--;
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+}
+
 /**
  * __i40evf_setup_tc - configure multiple traffic classes
  * @netdev: network interface device structure
@@ -2353,6 +2412,7 @@ static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
 			adapter->num_tc = 0;
 			netif_tx_stop_all_queues(netdev);
 			netif_tx_disable(netdev);
+			i40evf_del_all_cloud_filters(adapter);
 			adapter->aq_required = I40EVF_FLAG_AQ_DISABLE_CHANNELS;
 			goto exit;
 		} else {
@@ -2417,6 +2477,499 @@ static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
 	return ret;
 }
 
+/**
+ * i40evf_parse_cls_flower - Parse tc flower filters provided by kernel
+ * @adapter: board private structure
+ * @cls_flower: pointer to struct tc_cls_flower_offload
+ * @filter: pointer to cloud filter structure
+ */
+static int i40evf_parse_cls_flower(struct i40evf_adapter *adapter,
+				   struct tc_cls_flower_offload *f,
+				   struct i40evf_cloud_filter *filter)
+{
+	u16 n_proto_mask = 0;
+	u16 n_proto_key = 0;
+	u8 field_flags = 0;
+	u16 addr_type = 0;
+	u16 n_proto = 0;
+	int i = 0;
+
+	if (f->dissector->used_keys &
+	    ~(BIT(FLOW_DISSECTOR_KEY_CONTROL) |
+	      BIT(FLOW_DISSECTOR_KEY_BASIC) |
+	      BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_VLAN) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |
+	      BIT(FLOW_DISSECTOR_KEY_PORTS) |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_KEYID))) {
+		dev_err(&adapter->pdev->dev, "Unsupported key used: 0x%x\n",
+			f->dissector->used_keys);
+		return -EOPNOTSUPP;
+	}
+
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_ENC_KEYID)) {
+		struct flow_dissector_key_keyid *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_ENC_KEYID,
+						  f->mask);
+
+		if (mask->keyid != 0)
+			field_flags |= I40EVF_CLOUD_FIELD_TEN_ID;
+	}
+
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_BASIC)) {
+		struct flow_dissector_key_basic *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_BASIC,
+						  f->key);
+
+		struct flow_dissector_key_basic *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_BASIC,
+						  f->mask);
+		n_proto_key = ntohs(key->n_proto);
+		n_proto_mask = ntohs(mask->n_proto);
+
+		if (n_proto_key == ETH_P_ALL) {
+			n_proto_key = 0;
+			n_proto_mask = 0;
+		}
+		n_proto = n_proto_key & n_proto_mask;
+		if (n_proto != ETH_P_IP && n_proto != ETH_P_IPV6)
+			return -EINVAL;
+		if (n_proto == ETH_P_IPV6) {
+			/* specify flow type as TCP IPv6 */
+			filter->f.flow_type = VIRTCHNL_TCP_V6_FLOW;
+		}
+
+		if (key->ip_proto != IPPROTO_TCP) {
+			dev_info(&adapter->pdev->dev, "Only TCP transport is supported\n");
+			return -EINVAL;
+		}
+	}
+
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
+		struct flow_dissector_key_eth_addrs *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_ETH_ADDRS,
+						  f->key);
+
+		struct flow_dissector_key_eth_addrs *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_ETH_ADDRS,
+						  f->mask);
+		/* use is_broadcast and is_zero to check for all 0xf or 0 */
+		if (!is_zero_ether_addr(mask->dst)) {
+			if (is_broadcast_ether_addr(mask->dst)) {
+				field_flags |= I40EVF_CLOUD_FIELD_OMAC;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad ether dest mask %pM\n",
+					mask->dst);
+				return I40E_ERR_CONFIG;
+			}
+		}
+
+		if (!is_zero_ether_addr(mask->src)) {
+			if (is_broadcast_ether_addr(mask->src)) {
+				field_flags |= I40EVF_CLOUD_FIELD_IMAC;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad ether src mask %pM\n",
+					mask->src);
+				return I40E_ERR_CONFIG;
+			}
+		}
+
+		if (!is_zero_ether_addr(key->dst))
+			if (is_valid_ether_addr(key->dst) ||
+			    is_multicast_ether_addr(key->dst)) {
+				/* set the mask if a valid dst_mac address */
+				for (i = 0; i < ETH_ALEN; i++)
+					filter->f.mask.tcp_spec.dst_mac[i] |=
+									0xff;
+				ether_addr_copy(filter->f.data.tcp_spec.dst_mac,
+						key->dst);
+			}
+
+		if (!is_zero_ether_addr(key->src))
+			if (is_valid_ether_addr(key->src) ||
+			    is_multicast_ether_addr(key->src)) {
+				/* set the mask if a valid dst_mac address */
+				for (i = 0; i < ETH_ALEN; i++)
+					filter->f.mask.tcp_spec.src_mac[i] |=
+									0xff;
+				ether_addr_copy(filter->f.data.tcp_spec.src_mac,
+						key->src);
+		}
+	}
+
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_VLAN)) {
+		struct flow_dissector_key_vlan *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_VLAN,
+						  f->key);
+		struct flow_dissector_key_vlan *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_VLAN,
+						  f->mask);
+
+		if (mask->vlan_id) {
+			if (mask->vlan_id == VLAN_VID_MASK) {
+				field_flags |= I40EVF_CLOUD_FIELD_IVLAN;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad vlan mask %u\n",
+					mask->vlan_id);
+				return I40E_ERR_CONFIG;
+			}
+		}
+		filter->f.mask.tcp_spec.vlan_id |= cpu_to_be16(0xffff);
+		filter->f.data.tcp_spec.vlan_id = cpu_to_be16(key->vlan_id);
+	}
+
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_dissector_key_control *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_CONTROL,
+						  f->key);
+
+		addr_type = key->addr_type;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		struct flow_dissector_key_ipv4_addrs *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_IPV4_ADDRS,
+						  f->key);
+		struct flow_dissector_key_ipv4_addrs *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_IPV4_ADDRS,
+						  f->mask);
+
+		if (mask->dst) {
+			if (mask->dst == cpu_to_be32(0xffffffff)) {
+				field_flags |= I40EVF_CLOUD_FIELD_IIP;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad ip dst mask 0x%08x\n",
+					be32_to_cpu(mask->dst));
+				return I40E_ERR_CONFIG;
+			}
+		}
+
+		if (mask->src) {
+			if (mask->src == cpu_to_be32(0xffffffff)) {
+				field_flags |= I40EVF_CLOUD_FIELD_IIP;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad ip src mask 0x%08x\n",
+					be32_to_cpu(mask->dst));
+				return I40E_ERR_CONFIG;
+			}
+		}
+
+		if (field_flags & I40EVF_CLOUD_FIELD_TEN_ID) {
+			dev_info(&adapter->pdev->dev, "Tenant id not allowed for ip filter\n");
+			return I40E_ERR_CONFIG;
+		}
+		if (key->dst) {
+			filter->f.mask.tcp_spec.dst_ip[0] |=
+							cpu_to_be32(0xffffffff);
+			filter->f.data.tcp_spec.dst_ip[0] = key->dst;
+		}
+		if (key->src) {
+			filter->f.mask.tcp_spec.src_ip[0] |=
+							cpu_to_be32(0xffffffff);
+			filter->f.data.tcp_spec.src_ip[0] = key->src;
+		}
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct flow_dissector_key_ipv6_addrs *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_IPV6_ADDRS,
+						  f->key);
+		struct flow_dissector_key_ipv6_addrs *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_IPV6_ADDRS,
+						  f->mask);
+
+		/* validate mask, make sure it is not IPV6_ADDR_ANY */
+		if (ipv6_addr_any(&mask->dst)) {
+			dev_err(&adapter->pdev->dev, "Bad ipv6 dst mask 0x%02x\n",
+				IPV6_ADDR_ANY);
+			return I40E_ERR_CONFIG;
+		}
+
+		/* src and dest IPv6 address should not be LOOPBACK
+		 * (0:0:0:0:0:0:0:1) which can be represented as ::1
+		 */
+		if (ipv6_addr_loopback(&key->dst) ||
+		    ipv6_addr_loopback(&key->src)) {
+			dev_err(&adapter->pdev->dev,
+				"ipv6 addr should not be loopback\n");
+			return I40E_ERR_CONFIG;
+		}
+		if (!ipv6_addr_any(&mask->dst) || !ipv6_addr_any(&mask->src))
+			field_flags |= I40EVF_CLOUD_FIELD_IIP;
+
+		if (key->dst.s6_addr) {
+			for (i = 0; i < 4; i++)
+				filter->f.mask.tcp_spec.dst_ip[i] |=
+							cpu_to_be32(0xffffffff);
+			memcpy(&filter->f.data.tcp_spec.dst_ip,
+			       &key->dst.s6_addr32,
+			       sizeof(filter->f.data.tcp_spec.dst_ip));
+		}
+		if (key->src.s6_addr) {
+			for (i = 0; i < 4; i++)
+				filter->f.mask.tcp_spec.src_ip[i] |=
+							cpu_to_be32(0xffffffff);
+			memcpy(&filter->f.data.tcp_spec.src_ip,
+			       &key->src.s6_addr32,
+			       sizeof(filter->f.data.tcp_spec.src_ip));
+		}
+	}
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_PORTS)) {
+		struct flow_dissector_key_ports *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_PORTS,
+						  f->key);
+		struct flow_dissector_key_ports *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_PORTS,
+						  f->mask);
+
+		if (mask->src) {
+			if (mask->src == cpu_to_be16(0xffff)) {
+				field_flags |= I40EVF_CLOUD_FIELD_IIP;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad src port mask %u\n",
+					be16_to_cpu(mask->src));
+				return I40E_ERR_CONFIG;
+			}
+		}
+
+		if (mask->dst) {
+			if (mask->dst == cpu_to_be16(0xffff)) {
+				field_flags |= I40EVF_CLOUD_FIELD_IIP;
+			} else {
+				dev_err(&adapter->pdev->dev, "Bad dst port mask %u\n",
+					be16_to_cpu(mask->dst));
+				return I40E_ERR_CONFIG;
+			}
+		}
+		if (key->dst) {
+			filter->f.mask.tcp_spec.dst_port |= cpu_to_be16(0xffff);
+			filter->f.data.tcp_spec.dst_port = key->dst;
+		}
+
+		if (key->src) {
+			filter->f.mask.tcp_spec.src_port |= cpu_to_be16(0xffff);
+			filter->f.data.tcp_spec.src_port = key->dst;
+		}
+	}
+	filter->f.field_flags = field_flags;
+
+	return 0;
+}
+
+/**
+ * i40evf_handle_tclass - Forward to a traffic class on the device
+ * @adapter: board private structure
+ * @tc: traffic class index on the device
+ * @filter: pointer to cloud filter structure
+ */
+static int i40evf_handle_tclass(struct i40evf_adapter *adapter, u32 tc,
+				struct i40evf_cloud_filter *filter)
+{
+	if (tc == 0)
+		return 0;
+	if (tc < adapter->num_tc) {
+		if (!filter->f.data.tcp_spec.dst_port) {
+			dev_err(&adapter->pdev->dev,
+				"Specify destination port to redirect to traffic class other than TC0\n");
+			return -EINVAL;
+		}
+	}
+	/* redirect to a traffic class on the same device */
+	filter->f.action = VIRTCHNL_ACTION_TC_REDIRECT;
+	filter->f.action_meta = tc;
+	return 0;
+}
+
+/**
+ * i40evf_configure_clsflower - Add tc flower filters
+ * @adapter: board private structure
+ * @cls_flower: Pointer to struct tc_cls_flower_offload
+ */
+static int i40evf_configure_clsflower(struct i40evf_adapter *adapter,
+				      struct tc_cls_flower_offload *cls_flower)
+{
+	int tc = tc_classid_to_hwtc(adapter->netdev, cls_flower->classid);
+	struct i40evf_cloud_filter *filter = NULL;
+	int err = 0, count = 50;
+
+	while (test_and_set_bit(__I40EVF_IN_CRITICAL_TASK,
+				&adapter->crit_section)) {
+		udelay(1);
+		if (--count == 0)
+			return -EINVAL;
+	}
+
+	if (tc < 0) {
+		dev_err(&adapter->pdev->dev, "Invalid traffic class\n");
+		return -EINVAL;
+	}
+
+	filter = kzalloc(sizeof(*filter), GFP_KERNEL);
+	if (!filter) {
+		err = -ENOMEM;
+		goto clearout;
+	}
+	filter->cookie = cls_flower->cookie;
+
+	/* set the mask to all zeroes to begin with */
+	memset(&filter->f.mask.tcp_spec, 0, sizeof(struct virtchnl_l4_spec));
+	/* start out with flow type and eth type IPv4 to begin with */
+	filter->f.flow_type = VIRTCHNL_TCP_V4_FLOW;
+	err = i40evf_parse_cls_flower(adapter, cls_flower, filter);
+	if (err < 0)
+		goto err;
+
+	err = i40evf_handle_tclass(adapter, tc, filter);
+	if (err < 0)
+		goto err;
+
+	/* add filter to the list */
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	list_add_tail(&filter->list, &adapter->cloud_filter_list);
+	adapter->num_cloud_filters++;
+	filter->add = true;
+	adapter->aq_required |= I40EVF_FLAG_AQ_ADD_CLOUD_FILTER;
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+err:
+	if (err)
+		kfree(filter);
+clearout:
+	clear_bit(__I40EVF_IN_CRITICAL_TASK, &adapter->crit_section);
+	return err;
+}
+
+/* i40evf_find_cf - Find the cloud filter in the list
+ * @adapter: Board private structure
+ * @cookie: filter specific cookie
+ *
+ * Returns ptr to the filter object or NULL. Must be called while holding the
+ * cloud_filter_list_lock.
+ */
+static struct i40evf_cloud_filter *i40evf_find_cf(struct i40evf_adapter *adapter,
+						  unsigned long *cookie)
+{
+	struct i40evf_cloud_filter *filter = NULL;
+
+	if (!cookie)
+		return NULL;
+
+	list_for_each_entry(filter, &adapter->cloud_filter_list, list) {
+		if (!memcmp(cookie, &filter->cookie, sizeof(filter->cookie)))
+			return filter;
+	}
+	return NULL;
+}
+
+/**
+ * i40evf_delete_clsflower - Remove tc flower filters
+ * @adapter: board private structure
+ * @cls_flower: Pointer to struct tc_cls_flower_offload
+ */
+static int i40evf_delete_clsflower(struct i40evf_adapter *adapter,
+				   struct tc_cls_flower_offload *cls_flower)
+{
+	struct i40evf_cloud_filter *filter = NULL;
+	int err = 0;
+
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	filter = i40evf_find_cf(adapter, &cls_flower->cookie);
+	if (filter) {
+		filter->del = true;
+		adapter->aq_required |= I40EVF_FLAG_AQ_DEL_CLOUD_FILTER;
+	} else {
+		err = -EINVAL;
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+
+	return err;
+}
+
+/**
+ * i40evf_setup_tc_cls_flower - flower classifier offloads
+ * @netdev: net device to configure
+ * @type_data: offload data
+ */
+static int i40evf_setup_tc_cls_flower(struct i40evf_adapter *adapter,
+				      struct tc_cls_flower_offload *cls_flower)
+{
+	if (cls_flower->common.chain_index)
+		return -EOPNOTSUPP;
+
+	switch (cls_flower->command) {
+	case TC_CLSFLOWER_REPLACE:
+		return i40evf_configure_clsflower(adapter, cls_flower);
+	case TC_CLSFLOWER_DESTROY:
+		return i40evf_delete_clsflower(adapter, cls_flower);
+	case TC_CLSFLOWER_STATS:
+		return -EOPNOTSUPP;
+	default:
+		return -EINVAL;
+	}
+}
+
+/**
+ * i40evf_setup_tc_block_cb - block callback for tc
+ * @type: type of offload
+ * @type_data: offload data
+ * @cb_priv:
+ *
+ * This function is the block callback for traffic classes
+ **/
+static int i40evf_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
+				    void *cb_priv)
+{
+	switch (type) {
+	case TC_SETUP_CLSFLOWER:
+		return i40evf_setup_tc_cls_flower(cb_priv, type_data);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+/**
+ * i40evf_setup_tc_block - register callbacks for tc
+ * @netdev: network interface device structure
+ * @f: tc offload data
+ *
+ * This function registers block callbacks for tc
+ * offloads
+ **/
+static int i40evf_setup_tc_block(struct net_device *dev,
+				 struct tc_block_offload *f)
+{
+	struct i40evf_adapter *adapter = netdev_priv(dev);
+
+	if (f->binder_type != TCF_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
+		return -EOPNOTSUPP;
+
+	switch (f->command) {
+	case TC_BLOCK_BIND:
+		return tcf_block_cb_register(f->block, i40evf_setup_tc_block_cb,
+					     adapter, adapter);
+	case TC_BLOCK_UNBIND:
+		tcf_block_cb_unregister(f->block, i40evf_setup_tc_block_cb,
+					adapter);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
 /**
  * i40evf_setup_tc - configure multiple traffic classes
  * @netdev: network interface device structure
@@ -2431,10 +2984,14 @@ static int __i40evf_setup_tc(struct net_device *netdev, void *type_data)
 static int i40evf_setup_tc(struct net_device *netdev, enum tc_setup_type type,
 			   void *type_data)
 {
-	if (type != TC_SETUP_QDISC_MQPRIO)
+	switch (type) {
+	case TC_SETUP_QDISC_MQPRIO:
+		return __i40evf_setup_tc(netdev, type_data);
+	case TC_SETUP_BLOCK:
+		return i40evf_setup_tc_block(netdev, type_data);
+	default:
 		return -EOPNOTSUPP;
-
-	return __i40evf_setup_tc(netdev, type_data);
+	}
 }
 
 /**
@@ -2819,6 +3376,9 @@ int i40evf_process_config(struct i40evf_adapter *adapter)
 	if (vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_VLAN)
 		hw_features |= (NETIF_F_HW_VLAN_CTAG_TX |
 				NETIF_F_HW_VLAN_CTAG_RX);
+	/* Enable cloud filter if ADQ is supported */
+	if (vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ)
+		hw_features |= NETIF_F_HW_TC;
 
 	netdev->hw_features |= hw_features;
 
@@ -3186,9 +3746,11 @@ static int i40evf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	mutex_init(&hw->aq.arq_mutex);
 
 	spin_lock_init(&adapter->mac_vlan_list_lock);
+	spin_lock_init(&adapter->cloud_filter_list_lock);
 
 	INIT_LIST_HEAD(&adapter->mac_filter_list);
 	INIT_LIST_HEAD(&adapter->vlan_filter_list);
+	INIT_LIST_HEAD(&adapter->cloud_filter_list);
 
 	INIT_WORK(&adapter->reset_task, i40evf_reset_task);
 	INIT_WORK(&adapter->adminq_task, i40evf_adminq_task);
@@ -3315,6 +3877,7 @@ static void i40evf_remove(struct pci_dev *pdev)
 	struct i40evf_adapter *adapter = netdev_priv(netdev);
 	struct i40evf_vlan_filter *vlf, *vlftmp;
 	struct i40evf_mac_filter *f, *ftmp;
+	struct i40evf_cloud_filter *cf, *cftmp;
 	struct i40e_hw *hw = &adapter->hw;
 	int err;
 	/* Indicate we are in remove and not to run reset_task */
@@ -3385,6 +3948,13 @@ static void i40evf_remove(struct pci_dev *pdev)
 
 	spin_unlock_bh(&adapter->mac_vlan_list_lock);
 
+	spin_lock_bh(&adapter->cloud_filter_list_lock);
+	list_for_each_entry_safe(cf, cftmp, &adapter->cloud_filter_list, list) {
+		list_del(&cf->list);
+		kfree(cf);
+	}
+	spin_unlock_bh(&adapter->cloud_filter_list_lock);
+
 	free_netdev(netdev);
 
 	pci_disable_pcie_error_reporting(pdev);
diff --git a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
index 647ea104d435..6134b61e0938 100644
--- a/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
+++ b/drivers/net/ethernet/intel/i40evf/i40evf_virtchnl.c
@@ -1040,6 +1040,134 @@ void i40evf_disable_channels(struct i40evf_adapter *adapter)
 			   NULL, 0);
 }
 
+/**
+ * i40evf_print_cloud_filter
+ * @adapter: adapter structure
+ * @f: cloud filter to print
+ *
+ * Print the cloud filter
+ **/
+static void i40evf_print_cloud_filter(struct i40evf_adapter *adapter,
+				      struct virtchnl_filter f)
+{
+	switch (f.flow_type) {
+	case VIRTCHNL_TCP_V4_FLOW:
+		dev_info(&adapter->pdev->dev, "dst_mac: %pM src_mac: %pM vlan_id: %hu dst_ip: %pI4 src_ip %pI4 dst_port %hu src_port %hu\n",
+			 &f.data.tcp_spec.dst_mac, &f.data.tcp_spec.src_mac,
+			 ntohs(f.data.tcp_spec.vlan_id),
+			 &f.data.tcp_spec.dst_ip[0], &f.data.tcp_spec.src_ip[0],
+			 ntohs(f.data.tcp_spec.dst_port),
+			 ntohs(f.data.tcp_spec.src_port));
+		break;
+	case VIRTCHNL_TCP_V6_FLOW:
+		dev_info(&adapter->pdev->dev, "dst_mac: %pM src_mac: %pM vlan_id: %hu dst_ip: %pI6 src_ip %pI6 dst_port %hu src_port %hu\n",
+			 &f.data.tcp_spec.dst_mac, &f.data.tcp_spec.src_mac,
+			 ntohs(f.data.tcp_spec.vlan_id),
+			 &f.data.tcp_spec.dst_ip, &f.data.tcp_spec.src_ip,
+			 ntohs(f.data.tcp_spec.dst_port),
+			 ntohs(f.data.tcp_spec.src_port));
+		break;
+	}
+}
+
+/**
+ * i40evf_add_cloud_filter
+ * @adapter: adapter structure
+ *
+ * Request that the PF add cloud filters as specified
+ * by the user via tc tool.
+ **/
+void i40evf_add_cloud_filter(struct i40evf_adapter *adapter)
+{
+	struct i40evf_cloud_filter *cf;
+	struct virtchnl_filter *f;
+	int len = 0, count = 0;
+
+	if (adapter->current_op != VIRTCHNL_OP_UNKNOWN) {
+		/* bail because we already have a command pending */
+		dev_err(&adapter->pdev->dev, "Cannot add cloud filter, command %d pending\n",
+			adapter->current_op);
+		return;
+	}
+	list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+		if (cf->add) {
+			count++;
+			break;
+		}
+	}
+	if (!count) {
+		adapter->aq_required &= ~I40EVF_FLAG_AQ_ADD_CLOUD_FILTER;
+		return;
+	}
+	adapter->current_op = VIRTCHNL_OP_ADD_CLOUD_FILTER;
+
+	len = sizeof(struct virtchnl_filter);
+	f = kzalloc(len, GFP_KERNEL);
+	if (!f)
+		return;
+
+	list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+		if (cf->add) {
+			memcpy(f, &cf->f, sizeof(struct virtchnl_filter));
+			cf->add = false;
+			cf->state = __I40EVF_CF_ADD_PENDING;
+			i40evf_send_pf_msg(adapter,
+					   VIRTCHNL_OP_ADD_CLOUD_FILTER,
+					   (u8 *)f, len);
+		}
+	}
+	kfree(f);
+}
+
+/**
+ * i40evf_del_cloud_filter
+ * @adapter: adapter structure
+ *
+ * Request that the PF delete cloud filters as specified
+ * by the user via tc tool.
+ **/
+void i40evf_del_cloud_filter(struct i40evf_adapter *adapter)
+{
+	struct i40evf_cloud_filter *cf, *cftmp;
+	struct virtchnl_filter *f;
+	int len = 0, count = 0;
+
+	if (adapter->current_op != VIRTCHNL_OP_UNKNOWN) {
+		/* bail because we already have a command pending */
+		dev_err(&adapter->pdev->dev, "Cannot remove cloud filter, command %d pending\n",
+			adapter->current_op);
+		return;
+	}
+	list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+		if (cf->del) {
+			count++;
+			break;
+		}
+	}
+	if (!count) {
+		adapter->aq_required &= ~I40EVF_FLAG_AQ_DEL_CLOUD_FILTER;
+		return;
+	}
+	adapter->current_op = VIRTCHNL_OP_DEL_CLOUD_FILTER;
+
+	len = sizeof(struct virtchnl_filter);
+	f = kzalloc(len, GFP_KERNEL);
+	if (!f)
+		return;
+
+	list_for_each_entry_safe(cf, cftmp, &adapter->cloud_filter_list, list) {
+		if (cf->del) {
+			memcpy(f, &cf->f, sizeof(struct virtchnl_filter));
+			cf->del = false;
+			cf->state = __I40EVF_CF_DEL_PENDING;
+			i40evf_send_pf_msg(adapter,
+					   VIRTCHNL_OP_DEL_CLOUD_FILTER,
+					   (u8 *)f, len);
+		}
+	}
+	kfree(f);
+}
+
 /**
  * i40evf_request_reset
  * @adapter: adapter structure
@@ -1163,6 +1291,42 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 			adapter->ch_config.state = __I40EVF_TC_RUNNING;
 			netif_tx_start_all_queues(netdev);
 			break;
+		case VIRTCHNL_OP_ADD_CLOUD_FILTER: {
+			struct i40evf_cloud_filter *cf, *cftmp;
+
+			list_for_each_entry_safe(cf, cftmp,
+						 &adapter->cloud_filter_list,
+						 list) {
+				if (cf->state == __I40EVF_CF_ADD_PENDING) {
+					cf->state = __I40EVF_CF_INVALID;
+					dev_info(&adapter->pdev->dev, "Failed to add cloud filter, error %s\n",
+						 i40evf_stat_str(&adapter->hw,
+								 v_retval));
+					i40evf_print_cloud_filter(adapter,
+								  cf->f);
+					list_del(&cf->list);
+					kfree(cf);
+					adapter->num_cloud_filters--;
+				}
+			}
+			}
+			break;
+		case VIRTCHNL_OP_DEL_CLOUD_FILTER: {
+			struct i40evf_cloud_filter *cf;
+
+			list_for_each_entry(cf, &adapter->cloud_filter_list,
+					    list) {
+				if (cf->state == __I40EVF_CF_DEL_PENDING) {
+					cf->state = __I40EVF_CF_ACTIVE;
+					dev_info(&adapter->pdev->dev, "Failed to del cloud filter, error %s\n",
+						 i40evf_stat_str(&adapter->hw,
+								 v_retval));
+					i40evf_print_cloud_filter(adapter,
+								  cf->f);
+				}
+			}
+			}
+			break;
 		default:
 			dev_err(&adapter->pdev->dev, "PF returned error %d (%s) to our request %d\n",
 				v_retval,
@@ -1257,6 +1421,29 @@ void i40evf_virtchnl_completion(struct i40evf_adapter *adapter,
 		}
 		}
 		break;
+	case VIRTCHNL_OP_ADD_CLOUD_FILTER: {
+		struct i40evf_cloud_filter *cf;
+
+		list_for_each_entry(cf, &adapter->cloud_filter_list, list) {
+			if (cf->state == __I40EVF_CF_ADD_PENDING)
+				cf->state = __I40EVF_CF_ACTIVE;
+		}
+		}
+		break;
+	case VIRTCHNL_OP_DEL_CLOUD_FILTER: {
+		struct i40evf_cloud_filter *cf, *cftmp;
+
+		list_for_each_entry_safe(cf, cftmp, &adapter->cloud_filter_list,
+					 list) {
+			if (cf->state == __I40EVF_CF_DEL_PENDING) {
+				cf->state = __I40EVF_CF_INVALID;
+				list_del(&cf->list);
+				kfree(cf);
+				adapter->num_cloud_filters--;
+			}
+		}
+		}
+		break;
 	default:
 		if (adapter->current_op && (v_opcode != adapter->current_op))
 			dev_warn(&adapter->pdev->dev, "Expected response %d from PF, received %d\n",
-- 
2.14.3

[net-next 12/12] i40e: Add and delete cloud filter

[Jeff Kirsher]

From: Avinash Dayanand <avinash.dayanand@intel.com>

This patch provides support to add or delete cloud filter for queue
channels created for ADq on VF.
We are using the HW's cloud filter feature and programming it to act
as a TC filter applied to a group of queues.

There are two possible modes for a VF when applying a cloud filter
1. Basic Mode:	Intended to apply filters that don't need a VF to be
		Trusted. This would include the following
		  Dest MAC + L4 port
		  Dest MAC + VLAN + L4 port
2. Advanced Mode: This mode is only for filters with combination that
		  requires VF to be Trusted.
		  Dest IP + L4 port

When cloud filters are applied on a trusted VF and for some reason
the same VF is later made as untrusted then all cloud filters
will be deleted. All cloud filters has to be re-applied in
such a case.
Cloud filters are also deleted when queue channel is deleted.

Testing-Hints:
=============
1. Adding Basic Mode filter should be possible on a VF in
   Non-Trusted mode.
2. In Advanced mode all filters should be able to be created.

Steps:
======
1. Enable ADq and create TCs using TC mqprio command
2. Apply cloud filter.
3. Turn-off the spoof check.
4. Pass traffic.

Example:
========
1. tc qdisc add dev enp4s2 root mqprio num_tc 4 map 0 0 0 0 1 2 2 3\
	queues 2@0 2@2 1@4 1@5 hw 1 mode channel
2. tc qdisc add dev enp4s2 ingress
3. ethtool -K enp4s2 hw-tc-offload on
4. ip link set ens261f0 vf 0 spoofchk off
5. tc filter add dev enp4s2 protocol ip parent ffff: prio 1 flower\
	dst_ip 192.168.3.5/32 ip_proto udp dst_port 25 skip_sw hw_tc 2

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
 drivers/net/ethernet/intel/i40e/i40e.h             |   6 +
 drivers/net/ethernet/intel/i40e/i40e_main.c        |  16 +-
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c | 458 ++++++++++++++++++++-
 drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h |   2 +
 4 files changed, 470 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
index 4cf99292fc2f..36d9401a6258 100644
--- a/drivers/net/ethernet/intel/i40e/i40e.h
+++ b/drivers/net/ethernet/intel/i40e/i40e.h
@@ -1109,4 +1109,10 @@ static inline bool i40e_enabled_xdp_vsi(struct i40e_vsi *vsi)
 
 int i40e_create_queue_channel(struct i40e_vsi *vsi, struct i40e_channel *ch);
 int i40e_set_bw_limit(struct i40e_vsi *vsi, u16 seid, u64 max_tx_rate);
+int i40e_add_del_cloud_filter(struct i40e_vsi *vsi,
+			      struct i40e_cloud_filter *filter,
+			      bool add);
+int i40e_add_del_cloud_filter_big_buf(struct i40e_vsi *vsi,
+				      struct i40e_cloud_filter *filter,
+				      bool add);
 #endif /* _I40E_H_ */
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index 101702af099f..f6d37456f3b7 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -69,12 +69,6 @@ static int i40e_reset(struct i40e_pf *pf);
 static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired);
 static void i40e_fdir_sb_setup(struct i40e_pf *pf);
 static int i40e_veb_get_bw_info(struct i40e_veb *veb);
-static int i40e_add_del_cloud_filter(struct i40e_vsi *vsi,
-				     struct i40e_cloud_filter *filter,
-				     bool add);
-static int i40e_add_del_cloud_filter_big_buf(struct i40e_vsi *vsi,
-					     struct i40e_cloud_filter *filter,
-					     bool add);
 static int i40e_get_capabilities(struct i40e_pf *pf,
 				 enum i40e_admin_queue_opc list_type);
 
@@ -6841,8 +6835,8 @@ i40e_set_cld_element(struct i40e_cloud_filter *filter,
  * Add or delete a cloud filter for a specific flow spec.
  * Returns 0 if the filter were successfully added.
  **/
-static int i40e_add_del_cloud_filter(struct i40e_vsi *vsi,
-				     struct i40e_cloud_filter *filter, bool add)
+int i40e_add_del_cloud_filter(struct i40e_vsi *vsi,
+			      struct i40e_cloud_filter *filter, bool add)
 {
 	struct i40e_aqc_cloud_filters_element_data cld_filter;
 	struct i40e_pf *pf = vsi->back;
@@ -6908,9 +6902,9 @@ static int i40e_add_del_cloud_filter(struct i40e_vsi *vsi,
  * Add or delete a cloud filter for a specific flow spec using big buffer.
  * Returns 0 if the filter were successfully added.
  **/
-static int i40e_add_del_cloud_filter_big_buf(struct i40e_vsi *vsi,
-					     struct i40e_cloud_filter *filter,
-					     bool add)
+int i40e_add_del_cloud_filter_big_buf(struct i40e_vsi *vsi,
+				      struct i40e_cloud_filter *filter,
+				      bool add)
 {
 	struct i40e_aqc_cloud_filters_element_bb cld_filter;
 	struct i40e_pf *pf = vsi->back;
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
index b4bad70c82cd..5cca083da93c 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.c
@@ -2916,6 +2916,440 @@ static int i40e_vc_disable_vlan_stripping(struct i40e_vf *vf, u8 *msg,
 				       aq_ret);
 }
 
+/**
+ * i40e_validate_cloud_filter
+ * @mask: mask for TC filter
+ * @data: data for TC filter
+ *
+ * This function validates cloud filter programmed as TC filter for ADq
+ **/
+static int i40e_validate_cloud_filter(struct i40e_vf *vf,
+				      struct virtchnl_filter *tc_filter)
+{
+	struct virtchnl_l4_spec mask = tc_filter->mask.tcp_spec;
+	struct virtchnl_l4_spec data = tc_filter->data.tcp_spec;
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_vsi *vsi = NULL;
+	struct i40e_mac_filter *f;
+	struct hlist_node *h;
+	bool found = false;
+	int bkt;
+
+	if (!tc_filter->action) {
+		dev_info(&pf->pdev->dev,
+			 "VF %d: Currently ADq doesn't support Drop Action\n",
+			 vf->vf_id);
+		goto err;
+	}
+
+	/* action_meta is TC number here to which the filter is applied */
+	if (!tc_filter->action_meta ||
+	    tc_filter->action_meta > I40E_MAX_VF_VSI) {
+		dev_info(&pf->pdev->dev, "VF %d: Invalid TC number %u\n",
+			 vf->vf_id, tc_filter->action_meta);
+		goto err;
+	}
+
+	/* Check filter if it's programmed for advanced mode or basic mode.
+	 * There are two ADq modes (for VF only),
+	 * 1. Basic mode: intended to allow as many filter options as possible
+	 *		  to be added to a VF in Non-trusted mode. Main goal is
+	 *		  to add filters to its own MAC and VLAN id.
+	 * 2. Advanced mode: is for allowing filters to be applied other than
+	 *		  its own MAC or VLAN. This mode requires the VF to be
+	 *		  Trusted.
+	 */
+	if (mask.dst_mac[0] && !mask.dst_ip[0]) {
+		vsi = pf->vsi[vf->lan_vsi_idx];
+		f = i40e_find_mac(vsi, data.dst_mac);
+
+		if (!f) {
+			dev_info(&pf->pdev->dev,
+				 "Destination MAC %pM doesn't belong to VF %d\n",
+				 data.dst_mac, vf->vf_id);
+			goto err;
+		}
+
+		if (mask.vlan_id) {
+			hash_for_each_safe(vsi->mac_filter_hash, bkt, h, f,
+					   hlist) {
+				if (f->vlan == ntohs(data.vlan_id)) {
+					found = true;
+					break;
+				}
+			}
+			if (!found) {
+				dev_info(&pf->pdev->dev,
+					 "VF %d doesn't have any VLAN id %u\n",
+					 vf->vf_id, ntohs(data.vlan_id));
+				goto err;
+			}
+		}
+	} else {
+		/* Check if VF is trusted */
+		if (!test_bit(I40E_VIRTCHNL_VF_CAP_PRIVILEGE, &vf->vf_caps)) {
+			dev_err(&pf->pdev->dev,
+				"VF %d not trusted, make VF trusted to add advanced mode ADq cloud filters\n",
+				vf->vf_id);
+			return I40E_ERR_CONFIG;
+		}
+	}
+
+	if (mask.dst_mac[0] & data.dst_mac[0]) {
+		if (is_broadcast_ether_addr(data.dst_mac) ||
+		    is_zero_ether_addr(data.dst_mac)) {
+			dev_info(&pf->pdev->dev, "VF %d: Invalid Dest MAC addr %pM\n",
+				 vf->vf_id, data.dst_mac);
+			goto err;
+		}
+	}
+
+	if (mask.src_mac[0] & data.src_mac[0]) {
+		if (is_broadcast_ether_addr(data.src_mac) ||
+		    is_zero_ether_addr(data.src_mac)) {
+			dev_info(&pf->pdev->dev, "VF %d: Invalid Source MAC addr %pM\n",
+				 vf->vf_id, data.src_mac);
+			goto err;
+		}
+	}
+
+	if (mask.dst_port & data.dst_port) {
+		if (!data.dst_port || be16_to_cpu(data.dst_port) > 0xFFFF) {
+			dev_info(&pf->pdev->dev, "VF %d: Invalid Dest port\n",
+				 vf->vf_id);
+			goto err;
+		}
+	}
+
+	if (mask.src_port & data.src_port) {
+		if (!data.src_port || be16_to_cpu(data.src_port) > 0xFFFF) {
+			dev_info(&pf->pdev->dev, "VF %d: Invalid Source port\n",
+				 vf->vf_id);
+			goto err;
+		}
+	}
+
+	if (tc_filter->flow_type != VIRTCHNL_TCP_V6_FLOW &&
+	    tc_filter->flow_type != VIRTCHNL_TCP_V4_FLOW) {
+		dev_info(&pf->pdev->dev, "VF %d: Invalid Flow type\n",
+			 vf->vf_id);
+		goto err;
+	}
+
+	if (mask.vlan_id & data.vlan_id) {
+		if (ntohs(data.vlan_id) > I40E_MAX_VLANID) {
+			dev_info(&pf->pdev->dev, "VF %d: invalid VLAN ID\n",
+				 vf->vf_id);
+			goto err;
+		}
+	}
+
+	return I40E_SUCCESS;
+err:
+	return I40E_ERR_CONFIG;
+}
+
+/**
+ * i40e_find_vsi_from_seid - searches for the vsi with the given seid
+ * @vf: pointer to the VF info
+ * @seid - seid of the vsi it is searching for
+ **/
+static struct i40e_vsi *i40e_find_vsi_from_seid(struct i40e_vf *vf, u16 seid)
+{
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_vsi *vsi = NULL;
+	int i;
+
+	for (i = 0; i < vf->num_tc ; i++) {
+		vsi = i40e_find_vsi_from_id(pf, vf->ch[i].vsi_id);
+		if (vsi->seid == seid)
+			return vsi;
+	}
+	return NULL;
+}
+
+/**
+ * i40e_del_all_cloud_filters
+ * @vf: pointer to the VF info
+ *
+ * This function deletes all cloud filters
+ **/
+static void i40e_del_all_cloud_filters(struct i40e_vf *vf)
+{
+	struct i40e_cloud_filter *cfilter = NULL;
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_vsi *vsi = NULL;
+	struct hlist_node *node;
+	int ret;
+
+	hlist_for_each_entry_safe(cfilter, node,
+				  &vf->cloud_filter_list, cloud_node) {
+		vsi = i40e_find_vsi_from_seid(vf, cfilter->seid);
+
+		if (!vsi) {
+			dev_err(&pf->pdev->dev, "VF %d: no VSI found for matching %u seid, can't delete cloud filter\n",
+				vf->vf_id, cfilter->seid);
+			continue;
+		}
+
+		if (cfilter->dst_port)
+			ret = i40e_add_del_cloud_filter_big_buf(vsi, cfilter,
+								false);
+		else
+			ret = i40e_add_del_cloud_filter(vsi, cfilter, false);
+		if (ret)
+			dev_err(&pf->pdev->dev,
+				"VF %d: Failed to delete cloud filter, err %s aq_err %s\n",
+				vf->vf_id, i40e_stat_str(&pf->hw, ret),
+				i40e_aq_str(&pf->hw,
+					    pf->hw.aq.asq_last_status));
+
+		hlist_del(&cfilter->cloud_node);
+		kfree(cfilter);
+		vf->num_cloud_filters--;
+	}
+}
+
+/**
+ * i40e_vc_del_cloud_filter
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer
+ *
+ * This function deletes a cloud filter programmed as TC filter for ADq
+ **/
+static int i40e_vc_del_cloud_filter(struct i40e_vf *vf, u8 *msg)
+{
+	struct virtchnl_filter *vcf = (struct virtchnl_filter *)msg;
+	struct virtchnl_l4_spec mask = vcf->mask.tcp_spec;
+	struct virtchnl_l4_spec tcf = vcf->data.tcp_spec;
+	struct i40e_cloud_filter cfilter, *cf = NULL;
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_vsi *vsi = NULL;
+	struct hlist_node *node;
+	i40e_status aq_ret = 0;
+	int i, ret;
+
+	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (!vf->adq_enabled) {
+		dev_info(&pf->pdev->dev,
+			 "VF %d: ADq not enabled, can't apply cloud filter\n",
+			 vf->vf_id);
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (i40e_validate_cloud_filter(vf, vcf)) {
+		dev_info(&pf->pdev->dev,
+			 "VF %d: Invalid input, can't apply cloud filter\n",
+			 vf->vf_id);
+			aq_ret = I40E_ERR_PARAM;
+			goto err;
+	}
+
+	memset(&cfilter, 0, sizeof(cfilter));
+	/* parse destination mac address */
+	for (i = 0; i < ETH_ALEN; i++)
+		cfilter.dst_mac[i] = mask.dst_mac[i] & tcf.dst_mac[i];
+
+	/* parse source mac address */
+	for (i = 0; i < ETH_ALEN; i++)
+		cfilter.src_mac[i] = mask.src_mac[i] & tcf.src_mac[i];
+
+	cfilter.vlan_id = mask.vlan_id & tcf.vlan_id;
+	cfilter.dst_port = mask.dst_port & tcf.dst_port;
+	cfilter.src_port = mask.src_port & tcf.src_port;
+
+	switch (vcf->flow_type) {
+	case VIRTCHNL_TCP_V4_FLOW:
+		cfilter.n_proto = ETH_P_IP;
+		if (mask.dst_ip[0] & tcf.dst_ip[0])
+			memcpy(&cfilter.ip.v4.dst_ip, tcf.dst_ip,
+			       ARRAY_SIZE(tcf.dst_ip));
+		else if (mask.src_ip[0] & tcf.dst_ip[0])
+			memcpy(&cfilter.ip.v4.src_ip, tcf.src_ip,
+			       ARRAY_SIZE(tcf.dst_ip));
+		break;
+	case VIRTCHNL_TCP_V6_FLOW:
+		cfilter.n_proto = ETH_P_IPV6;
+		if (mask.dst_ip[3] & tcf.dst_ip[3])
+			memcpy(&cfilter.ip.v6.dst_ip6, tcf.dst_ip,
+			       sizeof(cfilter.ip.v6.dst_ip6));
+		if (mask.src_ip[3] & tcf.src_ip[3])
+			memcpy(&cfilter.ip.v6.src_ip6, tcf.src_ip,
+			       sizeof(cfilter.ip.v6.src_ip6));
+		break;
+	default:
+		/* TC filter can be configured based on different combinations
+		 * and in this case IP is not a part of filter config
+		 */
+		dev_info(&pf->pdev->dev, "VF %d: Flow type not configured\n",
+			 vf->vf_id);
+	}
+
+	/* get the vsi to which the tc belongs to */
+	vsi = pf->vsi[vf->ch[vcf->action_meta].vsi_idx];
+	cfilter.seid = vsi->seid;
+	cfilter.flags = vcf->field_flags;
+
+	/* Deleting TC filter */
+	if (tcf.dst_port)
+		ret = i40e_add_del_cloud_filter_big_buf(vsi, &cfilter, false);
+	else
+		ret = i40e_add_del_cloud_filter(vsi, &cfilter, false);
+	if (ret) {
+		dev_err(&pf->pdev->dev,
+			"VF %d: Failed to delete cloud filter, err %s aq_err %s\n",
+			vf->vf_id, i40e_stat_str(&pf->hw, ret),
+			i40e_aq_str(&pf->hw, pf->hw.aq.asq_last_status));
+		goto err;
+	}
+
+	hlist_for_each_entry_safe(cf, node,
+				  &vf->cloud_filter_list, cloud_node) {
+		if (cf->seid != cfilter.seid)
+			continue;
+		if (mask.dst_port)
+			if (cfilter.dst_port != cf->dst_port)
+				continue;
+		if (mask.dst_mac[0])
+			if (!ether_addr_equal(cf->src_mac, cfilter.src_mac))
+				continue;
+		/* for ipv4 data to be valid, only first byte of mask is set */
+		if (cfilter.n_proto == ETH_P_IP && mask.dst_ip[0])
+			if (memcmp(&cfilter.ip.v4.dst_ip, &cf->ip.v4.dst_ip,
+				   ARRAY_SIZE(tcf.dst_ip)))
+				continue;
+		/* for ipv6, mask is set for all sixteen bytes (4 words) */
+		if (cfilter.n_proto == ETH_P_IPV6 && mask.dst_ip[3])
+			if (memcmp(&cfilter.ip.v6.dst_ip6, &cf->ip.v6.dst_ip6,
+				   sizeof(cfilter.ip.v6.src_ip6)))
+				continue;
+		if (mask.vlan_id)
+			if (cfilter.vlan_id != cf->vlan_id)
+				continue;
+
+		hlist_del(&cf->cloud_node);
+		kfree(cf);
+		vf->num_cloud_filters--;
+	}
+
+err:
+	return i40e_vc_send_resp_to_vf(vf, VIRTCHNL_OP_DEL_CLOUD_FILTER,
+				       aq_ret);
+}
+
+/**
+ * i40e_vc_add_cloud_filter
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer
+ *
+ * This function adds a cloud filter programmed as TC filter for ADq
+ **/
+static int i40e_vc_add_cloud_filter(struct i40e_vf *vf, u8 *msg)
+{
+	struct virtchnl_filter *vcf = (struct virtchnl_filter *)msg;
+	struct virtchnl_l4_spec mask = vcf->mask.tcp_spec;
+	struct virtchnl_l4_spec tcf = vcf->data.tcp_spec;
+	struct i40e_cloud_filter *cfilter = NULL;
+	struct i40e_pf *pf = vf->pf;
+	struct i40e_vsi *vsi = NULL;
+	i40e_status aq_ret = 0;
+	int i, ret;
+
+	if (!test_bit(I40E_VF_STATE_ACTIVE, &vf->vf_states)) {
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (!vf->adq_enabled) {
+		dev_info(&pf->pdev->dev,
+			 "VF %d: ADq is not enabled, can't apply cloud filter\n",
+			 vf->vf_id);
+		aq_ret = I40E_ERR_PARAM;
+		goto err;
+	}
+
+	if (i40e_validate_cloud_filter(vf, vcf)) {
+		dev_info(&pf->pdev->dev,
+			 "VF %d: Invalid input/s, can't apply cloud filter\n",
+			 vf->vf_id);
+			aq_ret = I40E_ERR_PARAM;
+			goto err;
+	}
+
+	cfilter = kzalloc(sizeof(*cfilter), GFP_KERNEL);
+	if (!cfilter)
+		return -ENOMEM;
+
+	/* parse destination mac address */
+	for (i = 0; i < ETH_ALEN; i++)
+		cfilter->dst_mac[i] = mask.dst_mac[i] & tcf.dst_mac[i];
+
+	/* parse source mac address */
+	for (i = 0; i < ETH_ALEN; i++)
+		cfilter->src_mac[i] = mask.src_mac[i] & tcf.src_mac[i];
+
+	cfilter->vlan_id = mask.vlan_id & tcf.vlan_id;
+	cfilter->dst_port = mask.dst_port & tcf.dst_port;
+	cfilter->src_port = mask.src_port & tcf.src_port;
+
+	switch (vcf->flow_type) {
+	case VIRTCHNL_TCP_V4_FLOW:
+		cfilter->n_proto = ETH_P_IP;
+		if (mask.dst_ip[0] & tcf.dst_ip[0])
+			memcpy(&cfilter->ip.v4.dst_ip, tcf.dst_ip,
+			       ARRAY_SIZE(tcf.dst_ip));
+		else if (mask.src_ip[0] & tcf.dst_ip[0])
+			memcpy(&cfilter->ip.v4.src_ip, tcf.src_ip,
+			       ARRAY_SIZE(tcf.dst_ip));
+		break;
+	case VIRTCHNL_TCP_V6_FLOW:
+		cfilter->n_proto = ETH_P_IPV6;
+		if (mask.dst_ip[3] & tcf.dst_ip[3])
+			memcpy(&cfilter->ip.v6.dst_ip6, tcf.dst_ip,
+			       sizeof(cfilter->ip.v6.dst_ip6));
+		if (mask.src_ip[3] & tcf.src_ip[3])
+			memcpy(&cfilter->ip.v6.src_ip6, tcf.src_ip,
+			       sizeof(cfilter->ip.v6.src_ip6));
+		break;
+	default:
+		/* TC filter can be configured based on different combinations
+		 * and in this case IP is not a part of filter config
+		 */
+		dev_info(&pf->pdev->dev, "VF %d: Flow type not configured\n",
+			 vf->vf_id);
+	}
+
+	/* get the VSI to which the TC belongs to */
+	vsi = pf->vsi[vf->ch[vcf->action_meta].vsi_idx];
+	cfilter->seid = vsi->seid;
+	cfilter->flags = vcf->field_flags;
+
+	/* Adding cloud filter programmed as TC filter */
+	if (tcf.dst_port)
+		ret = i40e_add_del_cloud_filter_big_buf(vsi, cfilter, true);
+	else
+		ret = i40e_add_del_cloud_filter(vsi, cfilter, true);
+	if (ret) {
+		dev_err(&pf->pdev->dev,
+			"VF %d: Failed to add cloud filter, err %s aq_err %s\n",
+			vf->vf_id, i40e_stat_str(&pf->hw, ret),
+			i40e_aq_str(&pf->hw, pf->hw.aq.asq_last_status));
+		goto err;
+	}
+
+	INIT_HLIST_NODE(&cfilter->cloud_node);
+	hlist_add_head(&cfilter->cloud_node, &vf->cloud_filter_list);
+	vf->num_cloud_filters++;
+err:
+	return i40e_vc_send_resp_to_vf(vf, VIRTCHNL_OP_ADD_CLOUD_FILTER,
+				       aq_ret);
+}
+
 /**
  * i40e_vc_add_qch_msg: Add queue channel and enable ADq
  * @vf: pointer to the VF info
@@ -3036,6 +3470,11 @@ static int i40e_vc_add_qch_msg(struct i40e_vf *vf, u8 *msg)
 
 	/* set this flag only after making sure all inputs are sane */
 	vf->adq_enabled = true;
+	/* num_req_queues is set when user changes number of queues via ethtool
+	 * and this causes issue for default VSI(which depends on this variable)
+	 * when ADq is enabled, hence reset it.
+	 */
+	vf->num_req_queues = 0;
 
 	/* reset the VF in order to allocate resources */
 	i40e_vc_notify_vf_reset(vf);
@@ -3065,11 +3504,12 @@ static int i40e_vc_del_qch_msg(struct i40e_vf *vf, u8 *msg)
 	}
 
 	if (vf->adq_enabled) {
+		i40e_del_all_cloud_filters(vf);
 		i40e_del_qch(vf);
 		vf->adq_enabled = false;
 		vf->num_tc = 0;
 		dev_info(&pf->pdev->dev,
-			 "Deleting Queue Channels for ADq on VF %d\n",
+			 "Deleting Queue Channels and cloud filters for ADq on VF %d\n",
 			 vf->vf_id);
 	} else {
 		dev_info(&pf->pdev->dev, "VF %d trying to delete queue channels but ADq isn't enabled\n",
@@ -3223,6 +3663,12 @@ int i40e_vc_process_vf_msg(struct i40e_pf *pf, s16 vf_id, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_CHANNELS:
 		ret = i40e_vc_del_qch_msg(vf, msg);
 		break;
+	case VIRTCHNL_OP_ADD_CLOUD_FILTER:
+		ret = i40e_vc_add_cloud_filter(vf, msg);
+		break;
+	case VIRTCHNL_OP_DEL_CLOUD_FILTER:
+		ret = i40e_vc_del_cloud_filter(vf, msg);
+		break;
 	case VIRTCHNL_OP_UNKNOWN:
 	default:
 		dev_err(&pf->pdev->dev, "Unsupported opcode %d from VF %d\n",
@@ -3788,6 +4234,16 @@ int i40e_ndo_set_vf_trust(struct net_device *netdev, int vf_id, bool setting)
 	i40e_vc_disable_vf(vf);
 	dev_info(&pf->pdev->dev, "VF %u is now %strusted\n",
 		 vf_id, setting ? "" : "un");
+
+	if (vf->adq_enabled) {
+		if (!vf->trusted) {
+			dev_info(&pf->pdev->dev,
+				 "VF %u no longer Trusted, deleting all cloud filters\n",
+				 vf_id);
+			i40e_del_all_cloud_filters(vf);
+		}
+	}
+
 out:
 	return ret;
 }
diff --git a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
index 6e794c8a1ef3..6852599b2379 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
+++ b/drivers/net/ethernet/intel/i40e/i40e_virtchnl_pf.h
@@ -128,6 +128,8 @@ struct i40e_vf {
 	bool adq_enabled; /* flag to enable adq */
 	u8 num_tc;
 	struct i40evf_channel ch[I40E_MAX_VF_VSI];
+	struct hlist_head cloud_filter_list;
+	u16 num_cloud_filters;
 
 	/* RDMA Client */
 	struct virtchnl_iwarp_qvlist_info *qvlist_info;
-- 
2.14.3

Re: [net-next 00/12][pull request] 40GbE Intel Wired LAN Driver Updates 2018-02-14

[David Miller]

From: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Date: Wed, 14 Feb 2018 09:45:27 -0800

> This patch series enables the new mqprio hardware offload mechanism
> creating traffic classes on VFs for XL710 devices. The parameters
> needed to configure these traffic classes/queue channels are provides
> by the user via the tc tool. A maximum of four traffic classes can be
> created on each VF. This patch series also enables application of cloud
> filters to each of these traffic classes. The cloud filters are applied
> using the tc-flower classifier.
> 
> Example:
>     1. tc qdisc add dev vf0 root mqprio num_tc 4 map 0 0 0 0 1 2 2 3\
>         queues 2@0 2@2 1@4 1@5 hw 1 mode channel
>     2. tc qdisc add dev vf0 ingress
>     3. ethtool -K vf0 hw-tc-offload on
>     4. ip link set eth0 vf 0 spoofchk off
>     5. tc filter add dev vf0 protocol ip parent ffff: prio 1 flower dst_ip\
>         192.168.3.5/32 ip_proto udp dst_port 25 skip_sw hw_tc 2
> 
> The following are changes since commit 1d631583ae5e40681dc9a41d7cfa845222c3c7ff:
>   Merge branch 'net-dev-Make-protocol-ptr-dependent-on-CONFIG'
> and are available in the git repository at:
>   git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queue 40GbE

Looks great, pulled, thanks Jeff.