[PATCH net-next v2 0/2] be2net: add vxlan offload support

[PATCH net-next v2 0/2] be2net: add vxlan offload support

[Sathya Perla]

The first patch adds the FW cmds needed to configure the Skyhawk-R
chip for supporting VxLAN offloads. The second patch implements the
ndo_add/del_vxlan_port() methods and the plumbing for supporting
RX/TX csum, TSO and RSS steering offloads for VxLAN traffic.

v2 changes: NETIF_F_SG need not be set for hw_enc_features by the driver
as it is done by the stack.

Pls apply. Thanks.

Sathya Perla (2):
  be2net: add FW cmds needed for VxLAN offloads
  be2net: csum, tso and rss steering offload support for VxLAN

 drivers/net/ethernet/emulex/benet/be.h      |    7 +-
 drivers/net/ethernet/emulex/benet/be_cmds.c |  135 ++++++++++++++++++++-------
 drivers/net/ethernet/emulex/benet/be_cmds.h |   69 ++++++++++----
 drivers/net/ethernet/emulex/benet/be_hw.h   |    3 +-
 drivers/net/ethernet/emulex/benet/be_main.c |  130 +++++++++++++++++++++++---
 5 files changed, 277 insertions(+), 67 deletions(-)

[PATCH net-next v2 1/2] be2net: add FW cmds needed for VxLAN offloads

[Sathya Perla]

This patch adds support for the FW cmds needed for VxLAN offloads
on Skyhawk-R:
1) The VxLAN UDP port needs to be configured via the port-desc of
   SET_PROFILE_CONFIG_v1 cmd.
   This patch re-factors the be_set_profile_config() code (used so far
   only for setting VF QoS) to be used to set any type of descriptor.
2) The MANAGE_IFACE_FILTERS cmds is needed to convert a normal interface
   into a tunnel interface. This allows for RSS to work even on the inner
   TCP/UDP headers of VxLAN traffic.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
---
 drivers/net/ethernet/emulex/benet/be_cmds.c |  135 ++++++++++++++++++++-------
 drivers/net/ethernet/emulex/benet/be_cmds.h |   69 ++++++++++----
 drivers/net/ethernet/emulex/benet/be_main.c |    8 +-
 3 files changed, 154 insertions(+), 58 deletions(-)

diff --git a/drivers/net/ethernet/emulex/benet/be_cmds.c b/drivers/net/ethernet/emulex/benet/be_cmds.c
index cf5afe7..d1ec15a 100644
--- a/drivers/net/ethernet/emulex/benet/be_cmds.c
+++ b/drivers/net/ethernet/emulex/benet/be_cmds.c
@@ -3503,14 +3503,11 @@ err:
 	return status;
 }
 
-/* Currently only Lancer uses this command and it supports version 0 only
- * Uses sync mcc
- */
-int be_cmd_set_profile_config(struct be_adapter *adapter, u32 bps,
-			      u8 domain)
+int be_cmd_set_profile_config(struct be_adapter *adapter, void *desc,
+			      int size, u8 version, u8 domain)
 {
-	struct be_mcc_wrb *wrb;
 	struct be_cmd_req_set_profile_config *req;
+	struct be_mcc_wrb *wrb;
 	int status;
 
 	spin_lock_bh(&adapter->mcc_lock);
@@ -3522,44 +3519,116 @@ int be_cmd_set_profile_config(struct be_adapter *adapter, u32 bps,
 	}
 
 	req = embedded_payload(wrb);
-
 	be_wrb_cmd_hdr_prepare(&req->hdr, CMD_SUBSYSTEM_COMMON,
 			       OPCODE_COMMON_SET_PROFILE_CONFIG, sizeof(*req),
 			       wrb, NULL);
+	req->hdr.version = version;
 	req->hdr.domain = domain;
 	req->desc_count = cpu_to_le32(1);
-	req->nic_desc.hdr.desc_type = NIC_RESOURCE_DESC_TYPE_V0;
-	req->nic_desc.hdr.desc_len = RESOURCE_DESC_SIZE_V0;
-	req->nic_desc.flags = (1 << QUN) | (1 << IMM) | (1 << NOSV);
-	req->nic_desc.pf_num = adapter->pf_number;
-	req->nic_desc.vf_num = domain;
-
-	/* Mark fields invalid */
-	req->nic_desc.unicast_mac_count = 0xFFFF;
-	req->nic_desc.mcc_count = 0xFFFF;
-	req->nic_desc.vlan_count = 0xFFFF;
-	req->nic_desc.mcast_mac_count = 0xFFFF;
-	req->nic_desc.txq_count = 0xFFFF;
-	req->nic_desc.rq_count = 0xFFFF;
-	req->nic_desc.rssq_count = 0xFFFF;
-	req->nic_desc.lro_count = 0xFFFF;
-	req->nic_desc.cq_count = 0xFFFF;
-	req->nic_desc.toe_conn_count = 0xFFFF;
-	req->nic_desc.eq_count = 0xFFFF;
-	req->nic_desc.link_param = 0xFF;
-	req->nic_desc.bw_min = 0xFFFFFFFF;
-	req->nic_desc.acpi_params = 0xFF;
-	req->nic_desc.wol_param = 0x0F;
-
-	/* Change BW */
-	req->nic_desc.bw_min = cpu_to_le32(bps);
-	req->nic_desc.bw_max = cpu_to_le32(bps);
+	memcpy(req->desc, desc, size);
+
 	status = be_mcc_notify_wait(adapter);
 err:
 	spin_unlock_bh(&adapter->mcc_lock);
 	return status;
 }
 
+/* Mark all fields invalid */
+void be_reset_nic_desc(struct be_nic_res_desc *nic)
+{
+	memset(nic, 0, sizeof(*nic));
+	nic->unicast_mac_count = 0xFFFF;
+	nic->mcc_count = 0xFFFF;
+	nic->vlan_count = 0xFFFF;
+	nic->mcast_mac_count = 0xFFFF;
+	nic->txq_count = 0xFFFF;
+	nic->rq_count = 0xFFFF;
+	nic->rssq_count = 0xFFFF;
+	nic->lro_count = 0xFFFF;
+	nic->cq_count = 0xFFFF;
+	nic->toe_conn_count = 0xFFFF;
+	nic->eq_count = 0xFFFF;
+	nic->link_param = 0xFF;
+	nic->acpi_params = 0xFF;
+	nic->wol_param = 0x0F;
+	nic->bw_min = 0xFFFFFFFF;
+	nic->bw_max = 0xFFFFFFFF;
+}
+
+int be_cmd_config_qos(struct be_adapter *adapter, u32 bps, u8 domain)
+{
+	if (lancer_chip(adapter)) {
+		struct be_nic_res_desc nic_desc;
+
+		be_reset_nic_desc(&nic_desc);
+		nic_desc.hdr.desc_type = NIC_RESOURCE_DESC_TYPE_V0;
+		nic_desc.hdr.desc_len = RESOURCE_DESC_SIZE_V0;
+		nic_desc.flags = (1 << QUN_SHIFT) | (1 << IMM_SHIFT) |
+					(1 << NOSV_SHIFT);
+		nic_desc.pf_num = adapter->pf_number;
+		nic_desc.vf_num = domain;
+		nic_desc.bw_max = cpu_to_le32(bps);
+
+		return be_cmd_set_profile_config(adapter, &nic_desc,
+						 RESOURCE_DESC_SIZE_V0,
+						 0, domain);
+	} else {
+		return be_cmd_set_qos(adapter, bps, domain);
+	}
+}
+
+int be_cmd_manage_iface(struct be_adapter *adapter, u32 iface, u8 op)
+{
+	struct be_mcc_wrb *wrb;
+	struct be_cmd_req_manage_iface_filters *req;
+	int status;
+
+	if (iface == 0xFFFFFFFF)
+		return -1;
+
+	spin_lock_bh(&adapter->mcc_lock);
+
+	wrb = wrb_from_mccq(adapter);
+	if (!wrb) {
+		status = -EBUSY;
+		goto err;
+	}
+	req = embedded_payload(wrb);
+
+	be_wrb_cmd_hdr_prepare(&req->hdr, CMD_SUBSYSTEM_COMMON,
+			       OPCODE_COMMON_MANAGE_IFACE_FILTERS, sizeof(*req),
+			       wrb, NULL);
+	req->op = op;
+	req->target_iface_id = cpu_to_le32(iface);
+
+	status = be_mcc_notify_wait(adapter);
+err:
+	spin_unlock_bh(&adapter->mcc_lock);
+	return status;
+}
+
+int be_cmd_set_vxlan_port(struct be_adapter *adapter, __be16 port)
+{
+	struct be_port_res_desc port_desc;
+
+	memset(&port_desc, 0, sizeof(port_desc));
+	port_desc.hdr.desc_type = PORT_RESOURCE_DESC_TYPE_V1;
+	port_desc.hdr.desc_len = RESOURCE_DESC_SIZE_V1;
+	port_desc.flags = (1 << IMM_SHIFT) | (1 << NOSV_SHIFT);
+	port_desc.link_num = adapter->hba_port_num;
+	if (port) {
+		port_desc.nv_flags = NV_TYPE_VXLAN | (1 << SOCVID_SHIFT) |
+					(1 << RCVID_SHIFT);
+		port_desc.nv_port = swab16(port);
+	} else {
+		port_desc.nv_flags = NV_TYPE_DISABLED;
+		port_desc.nv_port = 0;
+	}
+
+	return be_cmd_set_profile_config(adapter, &port_desc,
+					 RESOURCE_DESC_SIZE_V1, 1, 0);
+}
+
 int be_cmd_get_if_id(struct be_adapter *adapter, struct be_vf_cfg *vf_cfg,
 		     int vf_num)
 {
diff --git a/drivers/net/ethernet/emulex/benet/be_cmds.h b/drivers/net/ethernet/emulex/benet/be_cmds.h
index fda3e88..b60e4d5 100644
--- a/drivers/net/ethernet/emulex/benet/be_cmds.h
+++ b/drivers/net/ethernet/emulex/benet/be_cmds.h
@@ -222,6 +222,7 @@ struct be_mcc_mailbox {
 #define OPCODE_COMMON_GET_FN_PRIVILEGES			170
 #define OPCODE_COMMON_READ_OBJECT			171
 #define OPCODE_COMMON_WRITE_OBJECT			172
+#define OPCODE_COMMON_MANAGE_IFACE_FILTERS		193
 #define OPCODE_COMMON_GET_IFACE_LIST			194
 #define OPCODE_COMMON_ENABLE_DISABLE_VF			196
 
@@ -1824,18 +1825,33 @@ struct be_cmd_req_set_ext_fat_caps {
 #define PORT_RESOURCE_DESC_TYPE_V1		0x55
 #define MAX_RESOURCE_DESC			264
 
-/* QOS unit number */
-#define QUN					4
-/* Immediate */
-#define IMM					6
-/* No save */
-#define NOSV					7
+#define IMM_SHIFT				6	/* Immediate */
+#define NOSV_SHIFT				7	/* No save */
 
 struct be_res_desc_hdr {
 	u8 desc_type;
 	u8 desc_len;
 } __packed;
 
+struct be_port_res_desc {
+	struct be_res_desc_hdr hdr;
+	u8 rsvd0;
+	u8 flags;
+	u8 link_num;
+	u8 mc_type;
+	u16 rsvd1;
+
+#define NV_TYPE_MASK				0x3	/* bits 0-1 */
+#define NV_TYPE_DISABLED			1
+#define NV_TYPE_VXLAN				3
+#define SOCVID_SHIFT				2	/* Strip outer vlan */
+#define RCVID_SHIFT				4	/* Report vlan */
+	u8 nv_flags;
+	u8 rsvd2;
+	__le16 nv_port;					/* vxlan/gre port */
+	u32 rsvd3[19];
+} __packed;
+
 struct be_pcie_res_desc {
 	struct be_res_desc_hdr hdr;
 	u8 rsvd0;
@@ -1856,6 +1872,8 @@ struct be_pcie_res_desc {
 struct be_nic_res_desc {
 	struct be_res_desc_hdr hdr;
 	u8 rsvd1;
+
+#define QUN_SHIFT				4 /* QoS is in absolute units */
 	u8 flags;
 	u8 vf_num;
 	u8 rsvd2;
@@ -1896,16 +1914,6 @@ enum mc_type {
 	vNIC2 = 0x07
 };
 
-struct be_port_res_desc {
-	struct be_res_desc_hdr hdr;
-	u8 rsvd0;
-	u8 flags;
-	u8 rsvd1;
-	u8 mc_type;
-	u16 rsvd2;
-	u32 rsvd3[20];
-} __packed;
-
 /* Is BE in a multi-channel mode */
 static inline bool be_is_mc(struct be_adapter *adapter)
 {
@@ -1940,7 +1948,7 @@ struct be_cmd_req_set_profile_config {
 	struct be_cmd_req_hdr hdr;
 	u32 rsvd;
 	u32 desc_count;
-	struct be_nic_res_desc nic_desc;
+	u8 desc[RESOURCE_DESC_SIZE_V1];
 };
 
 struct be_cmd_resp_set_profile_config {
@@ -1999,6 +2007,26 @@ struct be_cmd_req_set_ll_link {
 	u32 link_config; /* Bit 0: UP_DOWN, Bit 9: PLINK */
 };
 
+/************** Manage IFACE Filters *******************/
+#define OP_CONVERT_NORMAL_TO_TUNNEL		0
+#define OP_CONVERT_TUNNEL_TO_NORMAL		1
+
+struct be_cmd_req_manage_iface_filters {
+	struct be_cmd_req_hdr hdr;
+	u8  op;
+	u8  rsvd0;
+	u8  flags;
+	u8  rsvd1;
+	u32 tunnel_iface_id;
+	u32 target_iface_id;
+	u8  mac[6];
+	u16 vlan_tag;
+	u32 tenant_id;
+	u32 filter_id;
+	u32 cap_flags;
+	u32 cap_control_flags;
+} __packed;
+
 int be_pci_fnum_get(struct be_adapter *adapter);
 int be_fw_wait_ready(struct be_adapter *adapter);
 int be_cmd_mac_addr_query(struct be_adapter *adapter, u8 *mac_addr,
@@ -2073,7 +2101,7 @@ int be_cmd_get_seeprom_data(struct be_adapter *adapter,
 int be_cmd_set_loopback(struct be_adapter *adapter, u8 port_num,
 			u8 loopback_type, u8 enable);
 int be_cmd_get_phy_info(struct be_adapter *adapter);
-int be_cmd_set_qos(struct be_adapter *adapter, u32 bps, u32 domain);
+int be_cmd_config_qos(struct be_adapter *adapter, u32 bps, u8 domain);
 void be_detect_error(struct be_adapter *adapter);
 int be_cmd_get_die_temperature(struct be_adapter *adapter);
 int be_cmd_get_cntl_attributes(struct be_adapter *adapter);
@@ -2114,7 +2142,8 @@ int be_cmd_get_func_config(struct be_adapter *adapter,
 			   struct be_resources *res);
 int be_cmd_get_profile_config(struct be_adapter *adapter,
 			      struct be_resources *res, u8 domain);
-int be_cmd_set_profile_config(struct be_adapter *adapter, u32 bps, u8 domain);
+int be_cmd_set_profile_config(struct be_adapter *adapter, void *desc,
+			      int size, u8 version, u8 domain);
 int be_cmd_get_active_profile(struct be_adapter *adapter, u16 *profile);
 int be_cmd_get_if_id(struct be_adapter *adapter, struct be_vf_cfg *vf_cfg,
 		     int vf_num);
@@ -2122,3 +2151,5 @@ int be_cmd_enable_vf(struct be_adapter *adapter, u8 domain);
 int be_cmd_intr_set(struct be_adapter *adapter, bool intr_enable);
 int be_cmd_set_logical_link_config(struct be_adapter *adapter,
 					  int link_state, u8 domain);
+int be_cmd_set_vxlan_port(struct be_adapter *adapter, __be16 port);
+int be_cmd_manage_iface(struct be_adapter *adapter, u32 iface, u8 op);
diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c
index a61f967..8790e54 100644
--- a/drivers/net/ethernet/emulex/benet/be_main.c
+++ b/drivers/net/ethernet/emulex/benet/be_main.c
@@ -1346,11 +1346,7 @@ static int be_set_vf_tx_rate(struct net_device *netdev,
 		return -EINVAL;
 	}
 
-	if (lancer_chip(adapter))
-		status = be_cmd_set_profile_config(adapter, rate / 10, vf + 1);
-	else
-		status = be_cmd_set_qos(adapter, rate / 10, vf + 1);
-
+	status = be_cmd_config_qos(adapter, rate / 10, vf + 1);
 	if (status)
 		dev_err(&adapter->pdev->dev,
 				"tx rate %d on VF %d failed\n", rate, vf);
@@ -3124,7 +3120,7 @@ static int be_vf_setup(struct be_adapter *adapter)
 		 * Allow full available bandwidth
 		 */
 		if (BE3_chip(adapter) && !old_vfs)
-			be_cmd_set_qos(adapter, 1000, vf+1);
+			be_cmd_config_qos(adapter, 1000, vf + 1);
 
 		status = be_cmd_link_status_query(adapter, &lnk_speed,
 						  NULL, vf + 1);
-- 
1.7.1

RE: [PATCH net-next v2 1/2] be2net: add FW cmds needed for VxLAN offloads

[Sathya Perla]

> -----Original Message-----
> From: Or Gerlitz [mailto:or.gerlitz@gmail.com]
> 
> On Mon, Mar 24, 2014 at 8:00 AM, Sathya Perla <sathya.perla@emulex.com> >
> 
> +void be_reset_nic_desc(struct be_nic_res_desc *nic)
> > +{
> > +       memset(nic, 0, sizeof(*nic));
> > +       nic->unicast_mac_count = 0xFFFF;
> > +       nic->mcc_count = 0xFFFF;
> > +       nic->vlan_count = 0xFFFF;
> > +       nic->mcast_mac_count = 0xFFFF;
> > +       nic->txq_count = 0xFFFF;
> > +       nic->rq_count = 0xFFFF;
> > +       nic->rssq_count = 0xFFFF;
> > +       nic->lro_count = 0xFFFF;
> > +       nic->cq_count = 0xFFFF;
> > +       nic->toe_conn_count = 0xFFFF;
> > +       nic->eq_count = 0xFFFF;
> > +       nic->link_param = 0xFF;
> > +       nic->acpi_params = 0xFF;
> > +       nic->wol_param = 0x0F;
> > +       nic->bw_min = 0xFFFFFFFF;
> > +       nic->bw_max = 0xFFFFFFFF;
> 
> maybe some sort of memset can do as well?

This is old code that I just wrapped-up in a routine as a part of this patch, to help
with re-factoring of be_cmd_set_profile_config() needed for vxlan setting.
I'd like to take up any cleanup of this code as a part of a separate patch and not this patch-set.

Also, if you observe the be_nic_res_desc{} struct, the fields that require
re-setting with 0xFFFF (count fields and params that are to remain unchanged)
are all not contiguous. So, just memset-ing a region of memory doesn't necessarily work here.

thanks,
-Sathya

[PATCH net-next v2 2/2] be2net: csum, tso and rss steering offload support for VxLAN

[Sathya Perla]

This patch mainly implements the add/del_vxlan_port() methods by invoking
the needed FW cmds for supporting VxLAN offloads for Skyhawk-R.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
---
 drivers/net/ethernet/emulex/benet/be.h      |    7 +-
 drivers/net/ethernet/emulex/benet/be_hw.h   |    3 +-
 drivers/net/ethernet/emulex/benet/be_main.c |  122 +++++++++++++++++++++++++--
 3 files changed, 123 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/emulex/benet/be.h b/drivers/net/ethernet/emulex/benet/be.h
index a587c8a..8ccaa25 100644
--- a/drivers/net/ethernet/emulex/benet/be.h
+++ b/drivers/net/ethernet/emulex/benet/be.h
@@ -296,6 +296,7 @@ struct be_rx_compl_info {
 	u8 qnq;
 	u8 pkt_type;
 	u8 ip_frag;
+	u8 tunneled;
 };
 
 struct be_rx_obj {
@@ -371,10 +372,11 @@ enum vf_state {
 #define BE_FLAGS_WORKER_SCHEDULED		(1 << 3)
 #define BE_FLAGS_VLAN_PROMISC			(1 << 4)
 #define BE_FLAGS_NAPI_ENABLED			(1 << 9)
-#define BE_UC_PMAC_COUNT		30
-#define BE_VF_UC_PMAC_COUNT		2
 #define BE_FLAGS_QNQ_ASYNC_EVT_RCVD		(1 << 11)
+#define BE_FLAGS_VXLAN_OFFLOADS			(1 << 12)
 
+#define BE_UC_PMAC_COUNT			30
+#define BE_VF_UC_PMAC_COUNT			2
 /* Ethtool set_dump flags */
 #define LANCER_INITIATE_FW_DUMP			0x1
 
@@ -494,6 +496,7 @@ struct be_adapter {
 	u32 sli_family;
 	u8 hba_port_num;
 	u16 pvid;
+	__be16 vxlan_port;
 	struct phy_info phy;
 	u8 wol_cap;
 	bool wol_en;
diff --git a/drivers/net/ethernet/emulex/benet/be_hw.h b/drivers/net/ethernet/emulex/benet/be_hw.h
index 28ac8dd..3bd1985 100644
--- a/drivers/net/ethernet/emulex/benet/be_hw.h
+++ b/drivers/net/ethernet/emulex/benet/be_hw.h
@@ -407,7 +407,8 @@ struct amap_eth_rx_compl_v1 {
 	u8 vntagp;		/* dword 2 */
 	u8 header_len[8];	/* dword 2 */
 	u8 header_split[2];	/* dword 2 */
-	u8 rsvd1[13];		/* dword 2 */
+	u8 rsvd1[12];		/* dword 2 */
+	u8 tunneled;
 	u8 valid;		/* dword 2 */
 	u8 rsshash[32];		/* dword 3 */
 } __packed;
diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c
index 8790e54..66f8bfe 100644
--- a/drivers/net/ethernet/emulex/benet/be_main.c
+++ b/drivers/net/ethernet/emulex/benet/be_main.c
@@ -23,6 +23,7 @@
 #include <linux/aer.h>
 #include <linux/if_bridge.h>
 #include <net/busy_poll.h>
+#include <net/vxlan.h>
 
 MODULE_VERSION(DRV_VER);
 MODULE_DEVICE_TABLE(pci, be_dev_ids);
@@ -718,10 +719,23 @@ static inline u16 be_get_tx_vlan_tag(struct be_adapter *adapter,
 	return vlan_tag;
 }
 
+/* Used only for IP tunnel packets */
+static u16 skb_inner_ip_proto(struct sk_buff *skb)
+{
+	return (inner_ip_hdr(skb)->version == 4) ?
+		inner_ip_hdr(skb)->protocol : inner_ipv6_hdr(skb)->nexthdr;
+}
+
+static u16 skb_ip_proto(struct sk_buff *skb)
+{
+	return (ip_hdr(skb)->version == 4) ?
+		ip_hdr(skb)->protocol : ipv6_hdr(skb)->nexthdr;
+}
+
 static void wrb_fill_hdr(struct be_adapter *adapter, struct be_eth_hdr_wrb *hdr,
 		struct sk_buff *skb, u32 wrb_cnt, u32 len, bool skip_hw_vlan)
 {
-	u16 vlan_tag;
+	u16 vlan_tag, proto;
 
 	memset(hdr, 0, sizeof(*hdr));
 
@@ -734,9 +748,15 @@ static void wrb_fill_hdr(struct be_adapter *adapter, struct be_eth_hdr_wrb *hdr,
 		if (skb_is_gso_v6(skb) && !lancer_chip(adapter))
 			AMAP_SET_BITS(struct amap_eth_hdr_wrb, lso6, hdr, 1);
 	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
-		if (is_tcp_pkt(skb))
+		if (skb->encapsulation) {
+			AMAP_SET_BITS(struct amap_eth_hdr_wrb, ipcs, hdr, 1);
+			proto = skb_inner_ip_proto(skb);
+		} else {
+			proto = skb_ip_proto(skb);
+		}
+		if (proto == IPPROTO_TCP)
 			AMAP_SET_BITS(struct amap_eth_hdr_wrb, tcpcs, hdr, 1);
-		else if (is_udp_pkt(skb))
+		else if (proto == IPPROTO_UDP)
 			AMAP_SET_BITS(struct amap_eth_hdr_wrb, udpcs, hdr, 1);
 	}
 
@@ -1467,9 +1487,10 @@ static void be_rx_stats_update(struct be_rx_obj *rxo,
 static inline bool csum_passed(struct be_rx_compl_info *rxcp)
 {
 	/* L4 checksum is not reliable for non TCP/UDP packets.
-	 * Also ignore ipcksm for ipv6 pkts */
+	 * Also ignore ipcksm for ipv6 pkts
+	 */
 	return (rxcp->tcpf || rxcp->udpf) && rxcp->l4_csum &&
-				(rxcp->ip_csum || rxcp->ipv6);
+		(rxcp->ip_csum || rxcp->ipv6) && !rxcp->err;
 }
 
 static struct be_rx_page_info *get_rx_page_info(struct be_rx_obj *rxo)
@@ -1612,6 +1633,8 @@ static void be_rx_compl_process(struct be_rx_obj *rxo, struct napi_struct *napi,
 	skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
 	if (netdev->features & NETIF_F_RXHASH)
 		skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
+
+	skb->encapsulation = rxcp->tunneled;
 	skb_mark_napi_id(skb, napi);
 
 	if (rxcp->vlanf)
@@ -1668,6 +1691,8 @@ static void be_rx_compl_process_gro(struct be_rx_obj *rxo,
 	skb_record_rx_queue(skb, rxo - &adapter->rx_obj[0]);
 	if (adapter->netdev->features & NETIF_F_RXHASH)
 		skb_set_hash(skb, rxcp->rss_hash, PKT_HASH_TYPE_L3);
+
+	skb->encapsulation = rxcp->tunneled;
 	skb_mark_napi_id(skb, napi);
 
 	if (rxcp->vlanf)
@@ -1704,6 +1729,8 @@ static void be_parse_rx_compl_v1(struct be_eth_rx_compl *compl,
 					       compl);
 	}
 	rxcp->port = AMAP_GET_BITS(struct amap_eth_rx_compl_v1, port, compl);
+	rxcp->tunneled =
+		AMAP_GET_BITS(struct amap_eth_rx_compl_v1, tunneled, compl);
 }
 
 static void be_parse_rx_compl_v0(struct be_eth_rx_compl *compl,
@@ -2834,6 +2861,9 @@ static int be_open(struct net_device *netdev)
 
 	netif_tx_start_all_queues(netdev);
 	be_roce_dev_open(adapter);
+
+	if (skyhawk_chip(adapter))
+		vxlan_get_rx_port(netdev);
 	return 0;
 err:
 	be_close(adapter->netdev);
@@ -4120,6 +4150,78 @@ static int be_ndo_bridge_getlink(struct sk_buff *skb, u32 pid, u32 seq,
 				       BRIDGE_MODE_VEPA : BRIDGE_MODE_VEB);
 }
 
+static void be_disable_vxlan_offloads(struct be_adapter *adapter)
+{
+	if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS)
+		be_cmd_manage_iface(adapter, adapter->if_handle,
+				    OP_CONVERT_TUNNEL_TO_NORMAL);
+
+	if (adapter->vxlan_port)
+		be_cmd_set_vxlan_port(adapter, 0);
+
+	adapter->flags &= ~BE_FLAGS_VXLAN_OFFLOADS;
+	adapter->vxlan_port = 0;
+}
+
+static void be_add_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
+			      __be16 port)
+{
+	struct be_adapter *adapter = netdev_priv(netdev);
+	struct device *dev = &adapter->pdev->dev;
+	int status;
+
+	if (lancer_chip(adapter) || BEx_chip(adapter))
+		return;
+
+	if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS) {
+		dev_warn(dev, "Cannot add UDP port %d for VxLAN offloads\n",
+			 be16_to_cpu(port));
+		dev_info(dev,
+			 "Only one UDP port supported for VxLAN offloads\n");
+		return;
+	}
+
+	status = be_cmd_manage_iface(adapter, adapter->if_handle,
+				     OP_CONVERT_NORMAL_TO_TUNNEL);
+	if (status) {
+		dev_warn(dev, "Failed to convert normal interface to tunnel\n");
+		goto err;
+	}
+
+	status = be_cmd_set_vxlan_port(adapter, port);
+	if (status) {
+		dev_warn(dev, "Failed to add VxLAN port\n");
+		goto err;
+	}
+	adapter->flags |= BE_FLAGS_VXLAN_OFFLOADS;
+	adapter->vxlan_port = port;
+
+	dev_info(dev, "Enabled VxLAN offloads for UDP port %d\n",
+		 be16_to_cpu(port));
+	return;
+err:
+	be_disable_vxlan_offloads(adapter);
+	return;
+}
+
+static void be_del_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
+			      __be16 port)
+{
+	struct be_adapter *adapter = netdev_priv(netdev);
+
+	if (lancer_chip(adapter) || BEx_chip(adapter))
+		return;
+
+	if (adapter->vxlan_port != port)
+		return;
+
+	be_disable_vxlan_offloads(adapter);
+
+	dev_info(&adapter->pdev->dev,
+		 "Disabled VxLAN offloads for UDP port %d\n",
+		 be16_to_cpu(port));
+}
+
 static const struct net_device_ops be_netdev_ops = {
 	.ndo_open		= be_open,
 	.ndo_stop		= be_close,
@@ -4142,14 +4244,22 @@ static const struct net_device_ops be_netdev_ops = {
 	.ndo_bridge_setlink	= be_ndo_bridge_setlink,
 	.ndo_bridge_getlink	= be_ndo_bridge_getlink,
 #ifdef CONFIG_NET_RX_BUSY_POLL
-	.ndo_busy_poll		= be_busy_poll
+	.ndo_busy_poll		= be_busy_poll,
 #endif
+	.ndo_add_vxlan_port	= be_add_vxlan_port,
+	.ndo_del_vxlan_port	= be_del_vxlan_port,
 };
 
 static void be_netdev_init(struct net_device *netdev)
 {
 	struct be_adapter *adapter = netdev_priv(netdev);
 
+	if (skyhawk_chip(adapter)) {
+		netdev->hw_enc_features |= NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
+					   NETIF_F_TSO | NETIF_F_TSO6 |
+					   NETIF_F_GSO_UDP_TUNNEL;
+		netdev->hw_features |= NETIF_F_GSO_UDP_TUNNEL;
+	}
 	netdev->hw_features |= NETIF_F_SG | NETIF_F_TSO | NETIF_F_TSO6 |
 		NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_RXCSUM |
 		NETIF_F_HW_VLAN_CTAG_TX;
-- 
1.7.1

Re: [PATCH net-next v2 2/2] be2net: csum, tso and rss steering offload support for VxLAN

[Or Gerlitz]

On Mon, Mar 24, 2014 at 8:00 AM, Sathya Perla <sathya.perla@emulex.com> wrote:
> This patch mainly implements the add/del_vxlan_port() methods by invoking
> the needed FW cmds for supporting VxLAN offloads for Skyhawk-R.

[...]
> +static void be_add_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
> +                             __be16 port)
> +{
> +       struct be_adapter *adapter = netdev_priv(netdev);
> +       struct device *dev = &adapter->pdev->dev;
> +       int status;
> +
> +       if (lancer_chip(adapter) || BEx_chip(adapter))
> +               return;
> +
> +       if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS) {
> +               dev_warn(dev, "Cannot add UDP port %d for VxLAN offloads\n",
> +                        be16_to_cpu(port));
> +               dev_info(dev,
> +                        "Only one UDP port supported for VxLAN offloads\n");
> +               return;
> +       }
> +
> +       status = be_cmd_manage_iface(adapter, adapter->if_handle,
> +                                    OP_CONVERT_NORMAL_TO_TUNNEL);
> +       if (status) {
> +               dev_warn(dev, "Failed to convert normal interface to tunnel\n");
> +               goto err;
> +       }
> +
> +       status = be_cmd_set_vxlan_port(adapter, port);

so the flow be_cmd_set_vxlan_port --> be_cmd_set_profile_config goes
to sleep, right? well, it's wrong doing it from  here since the ndo is
called from the vxlan driver under spinlock, you should easily get a
nice stack trace if you test your code with spinlock/atomic debugging

RE: [PATCH net-next v2 2/2] be2net: csum, tso and rss steering offload support for VxLAN

[Sathya Perla]

> -----Original Message-----
> From: Or Gerlitz [mailto:or.gerlitz@gmail.com]
> 
> On Mon, Mar 24, 2014 at 8:00 AM, Sathya Perla <sathya.perla@emulex.com> wrote:
> > This patch mainly implements the add/del_vxlan_port() methods by invoking
> > the needed FW cmds for supporting VxLAN offloads for Skyhawk-R.
> 
> [...]
> > +static void be_add_vxlan_port(struct net_device *netdev, sa_family_t sa_family,
> > +                             __be16 port)
> > +{
> > +       struct be_adapter *adapter = netdev_priv(netdev);
> > +       struct device *dev = &adapter->pdev->dev;
> > +       int status;
> > +
> > +       if (lancer_chip(adapter) || BEx_chip(adapter))
> > +               return;
> > +
> > +       if (adapter->flags & BE_FLAGS_VXLAN_OFFLOADS) {
> > +               dev_warn(dev, "Cannot add UDP port %d for VxLAN offloads\n",
> > +                        be16_to_cpu(port));
> > +               dev_info(dev,
> > +                        "Only one UDP port supported for VxLAN offloads\n");
> > +               return;
> > +       }
> > +
> > +       status = be_cmd_manage_iface(adapter, adapter->if_handle,
> > +                                    OP_CONVERT_NORMAL_TO_TUNNEL);
> > +       if (status) {
> > +               dev_warn(dev, "Failed to convert normal interface to tunnel\n");
> > +               goto err;
> > +       }
> > +
> > +       status = be_cmd_set_vxlan_port(adapter, port);
> 
> so the flow be_cmd_set_vxlan_port --> be_cmd_set_profile_config goes
> to sleep, right? well, it's wrong doing it from  here since the ndo is
> called from the vxlan driver under spinlock, you should easily get a
> nice stack trace if you test your code with spinlock/atomic debugging

be_cmd_set_profile_config() does not sleep. 
But, I guess it could be made to sleep() in the future due to a design change...
Just to future-proof this code, I could fix this up with some workqueue processing..

thanks,
-Sathya