Skip to contentSkip to conversation

enable UDP ecpri configure in dcf

PATCH Superseded Jeff Guo 9 participants 91 messages Dec 16, 2020 – Jan 20, 2021 Network drivers

lore
Share & tools
lore

Jeff Guo

author
#0 lore raw

cover enable UDP ecpri configure in dcf

Enabling ecpri UDP tunnel port configure in dcf.

Jeff Guo (5):
ethdev: add new tunnel type for ecpri net/ice/base: add new UDP tunnel type for ecpri net/ice: add ecpri package type
net/ice: enable ecpri tunnel port configure in dcf app/testpmd: add new UDP tunnel port for ecpri

 app/test-pmd/cmdline.c               |   7 +-
 drivers/net/ice/base/ice_flex_pipe.c |   1 +
 drivers/net/ice/base/ice_flex_type.h |   1 +
 drivers/net/ice/ice_dcf_ethdev.c     |  67 +++++++++++++++
 drivers/net/ice/ice_ethdev.c         |   7 +-
 drivers/net/ice/ice_ethdev.h         |   1 +
 drivers/net/ice/ice_fdir_filter.c    |  66 ++++++++++++++
 drivers/net/ice/ice_hash.c           | 111 ++++++++++++++++++++++++
 drivers/net/ice/ice_rxtx.c           |  32 +++++++
 drivers/net/ice/ice_switch_filter.c  | 124 +++++++++++++++++++++++++++
 lib/librte_ethdev/rte_ethdev.h       |   1 +
 11 files changed, 415 insertions(+), 3 deletions(-)
Signature · 1 line
2.20.1
author
#1 lore raw

PATCH 1/5 ethdev: add new tunnel type for ecpri

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
 lib/librte_ethdev/rte_ethdev.h | 1 +
 1 file changed, 1 insertion(+)
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
Signature · 1 line
2.20.1
Reviewed-by on v1 1/5
#2 lore raw

Re: [dpdk-dev] [dpdk-dev 21.02 1/5] ethdev: add new tunnel type for ecpri

Quoted 11 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Wednesday, December 16, 2020 4:59 PM
To: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]; Guo, Jia <[email protected]>
Subject: [dpdk-dev 21.02 1/5] ethdev: add new tunnel type for ecpri

Quoted 3 lines

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

Suggest to separate into 2 patcheset,
one for API/testpmd change (1, 5), one for PMD implementation (2-4) for easy review then merge.

Quoted 18 lines
---
 lib/librte_ethdev/rte_ethdev.h | 1 +
 1 file changed, 1 insertion(+)
lib/librte_ethdev/rte_ethdev.h +1 -1
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
--

2.20.1

author
#3 lore raw

PATCH 2/5 net/ice/base: add new UDP tunnel type for ecpri

In reply to Jeff Guo

Add new UDP tunnel type for ecpri.

---
 drivers/net/ice/base/ice_flex_pipe.c | 1 +
 drivers/net/ice/base/ice_flex_type.h | 1 +
 2 files changed, 2 insertions(+)
drivers/net/ice/base/ice_flex_pipe.c +1 -0
index 7594df1696..fa8494e28d 100644
@@ -13,6 +13,7 @@
 static const struct ice_tunnel_type_scan tnls[] = {
 	{ TNL_VXLAN,		"TNL_VXLAN_PF" },
 	{ TNL_GENEVE,		"TNL_GENEVE_PF" },
+	{ TNL_ECPRI,		"TNL_UDP_ECPRI_PF" },
 	{ TNL_LAST,		"" }
 };
 
drivers/net/ice/base/ice_flex_type.h +1 -0
index 1dd57baccd..bf720753c1 100644
@@ -516,6 +516,7 @@ struct ice_pkg_enum {
 enum ice_tunnel_type {
 	TNL_VXLAN = 0,
 	TNL_GENEVE,
+	TNL_ECPRI,
 	TNL_LAST = 0xFF,
 	TNL_ALL = 0xFF,
 };
Signature · 1 line
2.20.1
author
#4 lore raw

PATCH 3/5 net/ice: add ecpri package type

In reply to Jeff Guo

Add new ecpri package type and process the new package type when handle each flow engines.

---
 drivers/net/ice/ice_ethdev.c        |   7 +-
 drivers/net/ice/ice_ethdev.h        |   1 +
 drivers/net/ice/ice_fdir_filter.c   |  66 +++++++++++++++
 drivers/net/ice/ice_hash.c          | 111 +++++++++++++++++++++++++
 drivers/net/ice/ice_rxtx.c          |  32 +++++++
 drivers/net/ice/ice_switch_filter.c | 124 ++++++++++++++++++++++++++++
 6 files changed, 340 insertions(+), 1 deletion(-)
drivers/net/ice/ice_ethdev.c +6 -1
index 9a5d6a559f..ae1a22572f 100644
@@ -71,7 +71,8 @@ static struct proto_xtr_ol_flag ice_proto_xtr_ol_flag_params[] = {
 #define ICE_DFLT_OUTER_TAG_TYPE ICE_AQ_VSI_OUTER_TAG_VLAN_9100
 
 #define ICE_OS_DEFAULT_PKG_NAME		"ICE OS Default Package"
-#define ICE_COMMS_PKG_NAME			"ICE COMMS Package"
+#define ICE_COMMS_PKG_NAME		"ICE COMMS Package"
+#define ICE_WIRELESS_EDGE_PKG_NAME	"ICE Wireless Edge Package"
 #define ICE_MAX_RES_DESC_NUM        1024
 
 static int ice_dev_configure(struct rte_eth_dev *dev);
@@ -1807,6 +1808,10 @@ ice_load_pkg_type(struct ice_hw *hw)
 	else if (!strncmp((char *)hw->active_pkg_name, ICE_COMMS_PKG_NAME,
 		ICE_PKG_NAME_SIZE))
 		package_type = ICE_PKG_TYPE_COMMS;
+	else if (!strncmp((char *)hw->active_pkg_name,
+			  ICE_WIRELESS_EDGE_PKG_NAME,
+			  ICE_PKG_NAME_SIZE))
+		package_type = ICE_PKG_TYPE_WIRELESS_EDGE;
 	else
 		package_type = ICE_PKG_TYPE_UNKNOWN;
 
drivers/net/ice/ice_ethdev.h +1 -0
index 899f446cde..7f8d0e9087 100644
@@ -147,6 +147,7 @@ enum ice_pkg_type {
 	ICE_PKG_TYPE_UNKNOWN,
 	ICE_PKG_TYPE_OS_DEFAULT,
 	ICE_PKG_TYPE_COMMS,
+	ICE_PKG_TYPE_WIRELESS_EDGE,
 };
 
 struct ice_adapter;
drivers/net/ice/ice_fdir_filter.c +66 -0

92 lines not on this page · raw message

drivers/net/ice/ice_hash.c +111 -0

135 lines not on this page · raw message

drivers/net/ice/ice_rxtx.c +32 -0
index d052bd0f1b..0cf7d6abb8 100644
@@ -1948,8 +1948,40 @@ ice_dev_supported_ptypes_get(struct rte_eth_dev *dev)
 		RTE_PTYPE_UNKNOWN
 	};
 
+	static const uint32_t ptypes_wireless_edge[] = {
+		/* refers to ice_get_default_pkt_type() */
+		RTE_PTYPE_L2_ETHER,
+		RTE_PTYPE_L2_ETHER_TIMESYNC,
+		RTE_PTYPE_L2_ETHER_LLDP,
+		RTE_PTYPE_L2_ETHER_ARP,
+		RTE_PTYPE_L3_IPV4_EXT_UNKNOWN,
+		RTE_PTYPE_L3_IPV6_EXT_UNKNOWN,
+		RTE_PTYPE_L4_FRAG,
+		RTE_PTYPE_L4_ICMP,
+		RTE_PTYPE_L4_NONFRAG,
+		RTE_PTYPE_L4_SCTP,
+		RTE_PTYPE_L4_TCP,
+		RTE_PTYPE_L4_UDP,
+		RTE_PTYPE_TUNNEL_GRENAT,
+		RTE_PTYPE_TUNNEL_IP,
+		RTE_PTYPE_INNER_L2_ETHER,
+		RTE_PTYPE_INNER_L3_IPV4_EXT_UNKNOWN,
+		RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN,
+		RTE_PTYPE_INNER_L4_FRAG,
+		RTE_PTYPE_INNER_L4_ICMP,
+		RTE_PTYPE_INNER_L4_NONFRAG,
+		RTE_PTYPE_INNER_L4_SCTP,
+		RTE_PTYPE_INNER_L4_TCP,
+		RTE_PTYPE_INNER_L4_UDP,
+		RTE_PTYPE_TUNNEL_GTPC,
+		RTE_PTYPE_TUNNEL_GTPU,
+		RTE_PTYPE_UNKNOWN
+	};
+
 	if (ad->active_pkg_type == ICE_PKG_TYPE_COMMS)
 		ptypes = ptypes_comms;
+	else if (ad->active_pkg_type == ICE_PKG_TYPE_WIRELESS_EDGE)
+		ptypes = ptypes_wireless_edge;
 	else
 		ptypes = ptypes_os;
 
drivers/net/ice/ice_switch_filter.c +124 -0

180 lines not on this page · raw message

Signature · 1 line
2.20.1
author
#5 lore raw

PATCH 4/5 net/ice: enable ecpri tunnel port configure in dcf

In reply to Jeff Guo

Add ecpri tunnel port add and rm ops to configure ecpri udp tunnel port in dcf.

---
 drivers/net/ice/ice_dcf_ethdev.c | 67 ++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
drivers/net/ice/ice_dcf_ethdev.c +67 -0
index b0b2ecb0d6..08265db79f 100644
@@ -26,6 +26,13 @@
 #include "ice_dcf_ethdev.h"
 #include "ice_rxtx.h"
 
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+
 static uint16_t
 ice_dcf_recv_pkts(__rte_unused void *rx_queue,
 		  __rte_unused struct rte_mbuf **bufs,
@@ -870,6 +877,64 @@ ice_dcf_link_update(__rte_unused struct rte_eth_dev *dev,
 	return 0;
 }
 
+/* Add UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+		ret = ice_create_tunnel(parent_hw, TNL_VXLAN,
+					udp_tunnel->udp_port);
+		break;
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_create_tunnel(parent_hw, TNL_ECPRI,
+					udp_tunnel->udp_port);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+/* Delete UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_destroy_tunnel(parent_hw, udp_tunnel->udp_port, 0);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
 static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.dev_start               = ice_dcf_dev_start,
 	.dev_stop                = ice_dcf_dev_stop,
@@ -892,6 +957,8 @@ static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.allmulticast_enable     = ice_dcf_dev_allmulticast_enable,
 	.allmulticast_disable    = ice_dcf_dev_allmulticast_disable,
 	.filter_ctrl             = ice_dcf_dev_filter_ctrl,
+	.udp_tunnel_port_add	 = ice_dcf_dev_udp_tunnel_port_add,
+	.udp_tunnel_port_del	 = ice_dcf_dev_udp_tunnel_port_del,
 };
 
 static int
Signature · 1 line
2.20.1
author
#6 lore raw

PATCH 5/5 app/testpmd: add new UDP tunnel port for ecpri

In reply to Jeff Guo

Add new UDP tunnel port params for ecpri configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
 app/test-pmd/cmdline.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
app/test-pmd/cmdline.c +5 -2
index 2ccbaa039e..af08e48e2e 100644
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
 	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
 	} else {
 		printf("Invalid tunnel type\n");
 		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
 				 "add#rm");
 cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
 	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
 cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
 			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
 	.f = cmd_cfg_tunnel_udp_port_parsed,
 	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
 	.tokens = {
 		(void *)&cmd_config_tunnel_udp_port_port,
 		(void *)&cmd_config_tunnel_udp_port_config,
Signature · 1 line
2.20.1
author
#7 lore raw

v2 cover add new UDP tunnel port for ecpri

In reply to Jeff Guo

Add new UDP tunnel type and port params for ecpri configuration.

v2:
separate patch set

Jeff Guo (2):
ethdev: add new tunnel type for ecpri
app/testpmd: add new UDP tunnel port for ecpri

 app/test-pmd/cmdline.c         | 7 +++++--
 lib/librte_ethdev/rte_ethdev.h | 1 +
 2 files changed, 6 insertions(+), 2 deletions(-)
Signature · 1 line
2.20.1
author
#8 lore raw

PATCH v2 1/2 ethdev: add new tunnel type for ecpri

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
 lib/librte_ethdev/rte_ethdev.h | 1 +
 1 file changed, 1 insertion(+)
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
Signature · 1 line
2.20.1
#9 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

24/12/2020 07:59, Jeff Guo:

Quoted 4 lines

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

[...]

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

author
#10 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 19 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 AM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 16 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 3 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Oh, the ABI break should be a problem.

Quoted 3 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#11 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 36 lines · Jia
Quoted 19 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 AM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 16 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
  	RTE_TUNNEL_TYPE_IP_IN_GRE,
  	RTE_L2_TUNNEL_TYPE_E_TAG,
  	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
  	RTE_TUNNEL_TYPE_MAX,
  };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Quoted 1 line

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Which legacy API is kept? (I've tried to cleanup drivers, but it is not always easy to do and I relied on driver maintainers)

If you are talking about an API to set UDP port for UDP-based tunnels, yes it is kept right now since there is no replacement yet.

It looks like eCRPI could be encapsulated in UDP, but I don't know if UDP port is fixed for the tunnel type or requires configuring.

#12 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

In reply to Guo, Jia

07/01/2021 10:32, Guo, Jia:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 7 lines

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 15 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Quoted 3 lines

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

I think rte_flow is a complete replacement with more features. You can match, encap, decap.
There is even a new API to get tunnel infos after decap. What is missing?

Quoted 9 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Oh, the ABI break should be a problem.

Quoted 2 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#13 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 38 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 PM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 10:32, Guo, Jia:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 22 lines

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Quoted 5 lines
Quoted 3 lines

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add
A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe, ecpri...)
In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way. so naturally it should be a device (port) level configuration but not a rte_flow rule for match, encap, decap...
So I think it's not a good idea to replace the rte_eth_dev_udp_tunnel_port_add with rte_flow config
and also there is no existing rte_flow_action can cover this requirement unless we introduce some new one.

Quoted 7 lines

You can match, encap, decap.
There is even a new API to get tunnel infos after decap. What is missing?

Quoted 2 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

Thanks
Qi

Quoted 9 lines
Quoted 7 lines

Oh, the ABI break should be a problem.

Quoted 2 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#14 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 13:47, Zhang, Qi Z:

Quoted 41 lines
Quoted 41 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 PM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 10:32, Guo, Jia:

Quoted 27 lines

From: Thomas Monjalon <[email protected]>

Quoted 7 lines

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 15 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add
A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe, ecpri...)
In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way. so naturally it should be a device (port) level configuration but not a rte_flow rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Quoted 8 lines

So I think it's not a good idea to replace
the rte_eth_dev_udp_tunnel_port_add with rte_flow config and also there is no existing rte_flow_action
can cover this requirement unless we introduce some new one.

Quoted 3 lines

You can match, encap, decap.
There is even a new API to get tunnel infos after decap. What is missing?

I still don't see which use case is missing.

Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed. Why it is not flagged by libabigail?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

Oh, the ABI break should be a problem.

Quoted 2 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#15 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 5 lines · Thomas Monjalon
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

I suspect this is because the enum is not referenced in any object... all I see is an integer with a comment that it should be filled with values from the enum.

Signature · 1 line
David Marchand
#16 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 10 lines · David Marchand
Quoted 5 lines · Thomas Monjalon
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

I suspect this is because the enum is not referenced in any object... all I see is an integer with a comment that it should be filled with values from the enum.

I am not sure about the full context but David is right in theory.

If the enum is not reachable from a publically exported interface (function or global variable) then it won't we considered as being part of the ABI and changes to that enum won't be reported.

I am not sure if that is what is happening in this particular case, though.

Cheers,

Signature · 1 line
		Dodji
#17 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

In reply to Thomas Monjalon
Quoted 67 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 9:34 PM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 13:47, Zhang, Qi Z:

Quoted 43 lines
Quoted 43 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 PM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 10:32, Guo, Jia:

Quoted 28 lines

From: Thomas Monjalon <[email protected]>

Quoted 8 lines

24/12/2020 07:59, Jeff Guo:

Quoted 2 lines

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 15 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 7 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 6 lines
Quoted 1 line

so naturally it should be a device (port) level configuration but not a rte_flow

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is "now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port SHOULD be configurable."

Thanks
Qi

Quoted 33 lines
Quoted 9 lines

So I think it's not a good idea to replace the rte_eth_dev_udp_tunnel_port_add with rte_flow config and also there is no existing rte_flow_action can cover this requirement unless we introduce some new one.

Quoted 3 lines

You can match, encap, decap.
There is even a new API to get tunnel infos after decap. What is missing?

I still don't see which use case is missing.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-compatibility tool

shows negative :) , thanks Ferruh' s guide.

Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

Oh, the ABI break should be a problem.

Quoted 2 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#18 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 16:24, Zhang, Qi Z:

Quoted 43 lines

From: Thomas Monjalon <[email protected]>

Quoted 35 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 7 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 6 lines
Quoted 1 line

so naturally it should be a device (port) level configuration but not a rte_flow

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Quoted 8 lines

Yes, in general It is a rule, it matches a udp packet's dst port and the action is "now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port SHOULD be configurable."

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

#19 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 72 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday, January 8, 2021 12:59 AM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 7 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port SHOULD be configurable."

Quoted 3 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end
then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?
if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

Quoted 3 lines
#20 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 72 lines · Qi Z
Quoted 72 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday, January 8, 2021 12:59 AM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
  	RTE_TUNNEL_TYPE_IP_IN_GRE,
  	RTE_L2_TUNNEL_TYPE_E_TAG,
  	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
  	RTE_TUNNEL_TYPE_MAX,
  };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 7 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port SHOULD be configurable."

Quoted 8 lines
Quoted 3 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end
then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

Quoted 8 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?
if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

Quoted 3 lines
#21 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 77 lines · Ferruh Yigit
Quoted 76 lines · Qi Z
Quoted 75 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday, January 8, 2021 12:59 AM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
       RTE_TUNNEL_TYPE_IP_IN_GRE,
       RTE_L2_TUNNEL_TYPE_E_TAG,
       RTE_TUNNEL_TYPE_VXLAN_GPE,
+    RTE_TUNNEL_TYPE_ECPRI,
       RTE_TUNNEL_TYPE_MAX,
   };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 7 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect.  but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port.  Some early implementations of VXLAN have used other values for the destination port.  To enable interoperability with these implementations, the destination port SHOULD be configurable."

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

Quoted 6 lines
Quoted 4 lines

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end
then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?
if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

#22 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 10:29, Andrew Rybchenko:

Quoted 72 lines · Ferruh Yigit
Quoted 71 lines · Qi Z
Quoted 70 lines

From: Thomas Monjalon <[email protected]>

Quoted 68 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
      RTE_TUNNEL_TYPE_IP_IN_GRE,
      RTE_L2_TUNNEL_TYPE_E_TAG,
      RTE_TUNNEL_TYPE_VXLAN_GPE,
+    RTE_TUNNEL_TYPE_ECPRI,
      RTE_TUNNEL_TYPE_MAX,
  };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 8 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that
match for a vlxan as pattern will take effect. but somehow, I think they are
not rules in the same domain, just like we have dedicate API for mac/vlan filter,
we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we
need this. https://tools.ietf.org/html/rfc7348).

Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port SHOULD be configurable."

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow
rule?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below
#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end
then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Quoted 4 lines
Quoted 2 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction.
All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#23 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 89 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday, January 8, 2021 6:36 PM
To: Yigit, Ferruh <[email protected]>; Zhang, Qi Z <[email protected]>; Guo, Jia <[email protected]>; Andrew Rybchenko <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 10:29, Andrew Rybchenko:

Quoted 58 lines · Ferruh Yigit
Quoted 57 lines · Qi Z
Quoted 56 lines

From: Thomas Monjalon <[email protected]>

Quoted 55 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
      RTE_TUNNEL_TYPE_IP_IN_GRE,
      RTE_L2_TUNNEL_TYPE_E_TAG,
      RTE_TUNNEL_TYPE_VXLAN_GPE,
+    RTE_TUNNEL_TYPE_ECPRI,
      RTE_TUNNEL_TYPE_MAX,
  };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 8 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this. https://tools.ietf.org/html/rfc7348).

Quoted 1 line
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't have.

#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement solution.

Quoted 26 lines
Quoted 6 lines
Quoted 4 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction.
All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#24 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Hi,

Quoted 99 lines

-----Original Message-----
From: Zhang, Qi Z <[email protected]>
Sent: Saturday, January 9, 2021 3:01 AM
Subject: RE: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 3 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for

ecpri

Quoted 59 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 57 lines · Ferruh Yigit
Quoted 56 lines · Qi Z
Quoted 55 lines

From: Thomas Monjalon <[email protected]>

Quoted 54 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:

Quoted 9 lines

--- a/lib/librte_ethdev/rte_ethdev.h
+++ b/lib/librte_ethdev/rte_ethdev.h
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
RTE_TUNNEL_TYPE_IP_IN_GRE,
RTE_L2_TUNNEL_TYPE_E_TAG,
RTE_TUNNEL_TYPE_VXLAN_GPE,
+ RTE_TUNNEL_TYPE_ECPRI,
RTE_TUNNEL_TYPE_MAX,
};

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 8 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this.

https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftools.ietf
.org%2Fhtml%2Frfc7348&amp;data=04%7C01%7Corika%40nvidia.com%7C46b2 d8f48944422f0d9008d8b43a2293%7C43083d15727340c1b7db39efd9ccc17a%7 C0%7C0%7C637457509081543237%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC 4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&a mp;sdata=RYWFMjuxkcUZ982kK2s44tBAjf%2FTkDyaa7VEybCtxOo%3D&amp;res erved=0).

Quoted 18 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Quoted 8 lines

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't have.

#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement solution.

Let me see if I understand it correctly.
In your case, the issue is that you need to configure the HW to parse the packet correctly right? It is not about the matching it is about the configuration of the HW, you wish to tell the HW that the packet should be parsed by different means correct?

If this is the case it sounds to me that you should use rte_flow and if the user adds the following rule:
#flow create pattern eth / ivp4 / udp port is 4789 / .. action ..... You simply need to configure your HW to support the ecpri configuration.

Quoted 26 lines
Quoted 26 lines
Quoted 6 lines
Quoted 4 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction. All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#25 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

10/01/2021 11:46, Ori Kam:

Quoted 103 lines

Hi,

Quoted 101 lines

-----Original Message-----
From: Zhang, Qi Z <[email protected]>
Sent: Saturday, January 9, 2021 3:01 AM
Subject: RE: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 3 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for

ecpri

Quoted 59 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 57 lines · Ferruh Yigit
Quoted 56 lines · Qi Z
Quoted 55 lines

From: Thomas Monjalon <[email protected]>

Quoted 54 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 24 lines

From: Thomas Monjalon <[email protected]>

Quoted 23 lines

07/01/2021 10:32, Guo, Jia:

Quoted 21 lines

From: Thomas Monjalon <[email protected]>

Quoted 16 lines

24/12/2020 07:59, Jeff Guo:
--- a/lib/librte_ethdev/rte_ethdev.h
+++ b/lib/librte_ethdev/rte_ethdev.h
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
RTE_TUNNEL_TYPE_IP_IN_GRE,
RTE_L2_TUNNEL_TYPE_E_TAG,
RTE_TUNNEL_TYPE_VXLAN_GPE,
+ RTE_TUNNEL_TYPE_ECPRI,
RTE_TUNNEL_TYPE_MAX,
};

We tried to remove all these legacy API in DPDK 20.11.
Andrew decided to not remove this one because it is not yet
completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way.

Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

Quoted 8 lines
Quoted 5 lines

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take effect. but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this.

https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftools.ietf
.org%2Fhtml%2Frfc7348&amp;data=04%7C01%7Corika%40nvidia.com%7C46b2 d8f48944422f0d9008d8b43a2293%7C43083d15727340c1b7db39efd9ccc17a%7 C0%7C0%7C637457509081543237%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC 4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&a mp;sdata=RYWFMjuxkcUZ982kK2s44tBAjf%2FTkDyaa7VEybCtxOo%3D&amp;res erved=0).

Quoted 18 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't have.

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

Quoted 4 lines
Quoted 4 lines

#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement solution.

The documentation does not say why it is useful.
With rte_flow you don't need it because a flow is specified with its action.

Quoted 39 lines

Let me see if I understand it correctly.
In your case, the issue is that you need to configure the HW to parse the packet correctly right? It is not about the matching it is about the configuration of the HW, you wish to tell the HW that the packet should be parsed by different means correct?

If this is the case it sounds to me that you should use rte_flow and if the user adds the following rule:
#flow create pattern eth / ivp4 / udp port is 4789 / .. action ..... You simply need to configure your HW to support the ecpri configuration.

Quoted 26 lines
Quoted 26 lines
Quoted 6 lines
Quoted 4 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction. All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#26 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 130 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Monday, January 11, 2021 5:24 PM
To: Zhang, Qi Z <[email protected]>; Yigit, Ferruh <[email protected]>; Guo, Jia <[email protected]>; Andrew Rybchenko <[email protected]>; Ori Kam <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Gregory Etelson <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

10/01/2021 11:46, Ori Kam:

Quoted 54 lines

Hi,

Quoted 52 lines

-----Original Message-----
From: Zhang, Qi Z <[email protected]>
Sent: Saturday, January 9, 2021 3:01 AM
Subject: RE: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 4 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for

ecpri

Quoted 39 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 37 lines · Ferruh Yigit
Quoted 36 lines · Qi Z
Quoted 35 lines

From: Thomas Monjalon <[email protected]>

Quoted 34 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 33 lines

From: Thomas Monjalon <[email protected]>

Quoted 32 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 31 lines

From: Thomas Monjalon <[email protected]>

Quoted 19 lines

07/01/2021 10:32, Guo, Jia:

Quoted 18 lines

From: Thomas Monjalon <[email protected]>
24/12/2020 07:59, Jeff Guo:
--- a/lib/librte_ethdev/rte_ethdev.h
+++ b/lib/librte_ethdev/rte_ethdev.h
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
RTE_TUNNEL_TYPE_IP_IN_GRE,
RTE_L2_TUNNEL_TYPE_E_TAG,
RTE_TUNNEL_TYPE_VXLAN_GPE,
+ RTE_TUNNEL_TYPE_ECPRI,
RTE_TUNNEL_TYPE_MAX,
};

We tried to remove all these legacy API in DPDK 20.11.
Andrew decided to not remove this one because it is
not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Quoted 5 lines
Quoted 3 lines

Agree but seems that the legacy api and driver legacy
implementation still keep in this release, and there is
no a general way to replace the legacy by rte_flow right now.

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type

(e.g.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line

vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar

way.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

rule for match, encap, decap...

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take

effect.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this.

https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fto
ols.ietf .org%2Fhtml%2Frfc7348&amp;data=04%7C01%7Corika%40nvidia.com%7C

46b2

Quoted 1 line
Quoted 1 line

d8f48944422f0d9008d8b43a2293%7C43083d15727340c1b7db39efd9ccc17a %7

Quoted 1 line
Quoted 1 line

C0%7C0%7C637457509081543237%7CUnknown%7CTWFpbGZsb3d8eyJWIjoi MC

Quoted 1 line
Quoted 1 line

4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&a

Quoted 1 line
Quoted 1 line

mp;sdata=RYWFMjuxkcUZ982kK2s44tBAjf%2FTkDyaa7VEybCtxOo%3D&amp;r es

Quoted 24 lines
Quoted 24 lines

erved=0).

Quoted 22 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Quoted 5 lines
Quoted 2 lines
Quoted 2 lines

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't have.

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can only be configured by a PF driver or a privileged VF base on our security model. A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use rte_eth_dev_udp_tunnel_port_add to create tunnel port for eCPRI, them this will impact on all VFs in the same PF.
2. A normal VF driver can create rte_flow rule that match specific patch for queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

So it does not help to have a rte_flow that match dst udp port as tunnel while have an action in the same rule.

Quoted 57 lines
Quoted 7 lines
Quoted 7 lines

#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement solution.

The documentation does not say why it is useful.
With rte_flow you don't need it because a flow is specified with its action.

Quoted 2 lines

Let me see if I understand it correctly.
In your case, the issue is that you need to configure the HW to parse the

packet correctly right?

Quoted 2 lines

It is not about the matching it is about the configuration of the HW, you wish to tell the HW that the packet should be parsed by different means

correct?

Quoted 35 lines

If this is the case it sounds to me that you should use rte_flow and if the user adds the following rule:
#flow create pattern eth / ivp4 / udp port is 4789 / .. action ..... You simply need to configure your HW to support the ecpri configuration.

Quoted 26 lines
Quoted 26 lines
Quoted 6 lines
Quoted 4 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction. All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#27 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

11/01/2021 12:26, Zhang, Qi Z:

Quoted 107 lines

From: Thomas Monjalon <[email protected]>

Quoted 106 lines

10/01/2021 11:46, Ori Kam:

Quoted 40 lines

From: Zhang, Qi Z <[email protected]>

Quoted 39 lines

From: Thomas Monjalon <[email protected]>

Quoted 38 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 37 lines · Ferruh Yigit
Quoted 36 lines · Qi Z
Quoted 35 lines

From: Thomas Monjalon <[email protected]>

Quoted 34 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 33 lines

From: Thomas Monjalon <[email protected]>

Quoted 32 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 31 lines

From: Thomas Monjalon <[email protected]>

Quoted 24 lines

07/01/2021 10:32, Guo, Jia:
From: Thomas Monjalon <[email protected]>
24/12/2020 07:59, Jeff Guo:
--- a/lib/librte_ethdev/rte_ethdev.h
+++ b/lib/librte_ethdev/rte_ethdev.h
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
RTE_TUNNEL_TYPE_IP_IN_GRE,
RTE_L2_TUNNEL_TYPE_E_TAG,
RTE_TUNNEL_TYPE_VXLAN_GPE,
+ RTE_TUNNEL_TYPE_ECPRI,
RTE_TUNNEL_TYPE_MAX,
};

We tried to remove all these legacy API in DPDK 20.11.
Andrew decided to not remove this one because it is
not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy
implementation still keep in this release, and there is
no a general way to replace the legacy by rte_flow right now.

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add A packet with specific dst udp port will be recognized as a specific tunnel packet type

(e.g.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line

vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar

way.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 2 lines

so naturally it should be a device (port) level configuration but not a rte_flow

rule for match, encap, decap...

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take

effect.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this.

https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fto
ols.ietf .org%2Fhtml%2Frfc7348&amp;data=04%7C01%7Corika%40nvidia.com%7C

46b2

Quoted 1 line
Quoted 1 line

d8f48944422f0d9008d8b43a2293%7C43083d15727340c1b7db39efd9ccc17a %7

Quoted 1 line
Quoted 1 line

C0%7C0%7C637457509081543237%7CUnknown%7CTWFpbGZsb3d8eyJWIjoi MC

Quoted 1 line
Quoted 1 line

4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&a

Quoted 1 line
Quoted 1 line

mp;sdata=RYWFMjuxkcUZ982kK2s44tBAjf%2FTkDyaa7VEybCtxOo%3D&amp;r es

Quoted 26 lines
Quoted 26 lines

erved=0).

Quoted 19 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't have.

Quoted 8 lines
Quoted 2 lines

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can only be configured by a PF driver or a privileged VF base on our security model. A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use rte_eth_dev_udp_tunnel_port_add to create tunnel port for eCPRI, them this will impact on all VFs in the same PF.
2. A normal VF driver can create rte_flow rule that match specific patch for queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

Whaooh! A normal Intel VF is not allowed to match the tunnel it wants if not enabled by a priviledged VF?
I would say it is a HW design flaw, but that's not the question.

Quoted 2 lines

So it does not help to have a rte_flow that match dst udp port as tunnel while have an action in the same rule.

Now I understand you are truly looking for a device configuration. But as it looks more as a HW limitation,
I don't think it should be part of ethdev.
The fact is that it is already part of ethdev...

I don't know what to say. I need to digest how Intel limitation is "still" trying to drive some parts of the "generic" API.

Quoted 48 lines
Quoted 48 lines
Quoted 5 lines
Quoted 5 lines

#flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement solution.

The documentation does not say why it is useful.
With rte_flow you don't need it because a flow is specified with its action.

Quoted 2 lines

Let me see if I understand it correctly.
In your case, the issue is that you need to configure the HW to parse the

packet correctly right?

Quoted 2 lines

It is not about the matching it is about the configuration of the HW, you wish to tell the HW that the packet should be parsed by different means

correct?

Quoted 32 lines

If this is the case it sounds to me that you should use rte_flow and if the user adds the following rule:
#flow create pattern eth / ivp4 / udp port is 4789 / .. action ..... You simply need to configure your HW to support the ecpri configuration.

Quoted 25 lines
Quoted 25 lines
Quoted 6 lines
Quoted 4 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction.
All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#28 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 145 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Monday, January 11, 2021 7:38 PM
To: Yigit, Ferruh <[email protected]>; Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Andrew Rybchenko <[email protected]>; Ori Kam <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Gregory Etelson <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

11/01/2021 12:26, Zhang, Qi Z:

Quoted 34 lines

From: Thomas Monjalon <[email protected]>

Quoted 33 lines

10/01/2021 11:46, Ori Kam:

Quoted 32 lines

From: Zhang, Qi Z <[email protected]>

Quoted 31 lines

From: Thomas Monjalon <[email protected]>

Quoted 30 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 29 lines · Ferruh Yigit
Quoted 28 lines · Qi Z
Quoted 27 lines

From: Thomas Monjalon <[email protected]>

Quoted 26 lines

07/01/2021 16:24, Zhang, Qi Z:

Quoted 25 lines

From: Thomas Monjalon <[email protected]>

Quoted 24 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 23 lines

From: Thomas Monjalon <[email protected]>
07/01/2021 10:32, Guo, Jia:
From: Thomas Monjalon <[email protected]>
24/12/2020 07:59, Jeff Guo:
--- a/lib/librte_ethdev/rte_ethdev.h
+++ b/lib/librte_ethdev/rte_ethdev.h
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
RTE_TUNNEL_TYPE_IP_IN_GRE,
RTE_L2_TUNNEL_TYPE_E_TAG,
RTE_TUNNEL_TYPE_VXLAN_GPE,
+ RTE_TUNNEL_TYPE_ECPRI,
RTE_TUNNEL_TYPE_MAX,
};

We tried to remove all these legacy API in DPDK 20.11.
Andrew decided to not remove this one because it
is not yet completely replaced by rte_flow in all drivers.
However, I am against continuing to update this API.
The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver
legacy implementation still keep in this release,
and there is no a general way to replace the legacy by

rte_flow right now.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

I think rte_flow is a complete replacement with more

features.

Quoted 28 lines
Quoted 28 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by
rte_eth_dev_udp_tunnel_port_add A packet with
specific dst udp port will be recognized as a
specific tunnel packet type

(e.g.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 1 line

vxlan, vxlan-gpe,

ecpri...) In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar

way.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

so naturally it should be a device (port) level
configuration but not a rte_flow

rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel. What is the usage?

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 2 lines

Yes, in general It is a rule, it matches a udp packet's dst port and the action is

"now the packet is identified as vxlan packet" then all other rte_flow rules that match for a vlxan as pattern will take

effect.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

but somehow, I think they are not rules in the same domain, just like we have dedicate API for mac/vlan filter, we'd better have a dedicate API for this also. ( RFC for Vxlan explains why we need this.

https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%
2Fto
ols.ietf .org%2Fhtml%2Frfc7348&amp;data=04%7C01%7Corika%40nvidia.com

%7C

Quoted 3 lines
Quoted 3 lines

46b2

Quoted 1 line
Quoted 1 line

d8f48944422f0d9008d8b43a2293%7C43083d15727340c1b7db39efd9ccc17a

Quoted 3 lines
Quoted 3 lines

%7

Quoted 1 line
Quoted 1 line

C0%7C0%7C637457509081543237%7CUnknown%7CTWFpbGZsb3d8eyJWIjoi

Quoted 3 lines
Quoted 3 lines

MC

Quoted 1 line
Quoted 1 line

4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&a

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

mp;sdata=RYWFMjuxkcUZ982kK2s44tBAjf%2FTkDyaa7VEybCtxOo%3D&amp;r

Quoted 15 lines
Quoted 15 lines

es

Quoted 14 lines
Quoted 14 lines

erved=0).

Quoted 13 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

"Destination Port: IANA has assigned the value 4789 for the VXLAN UDP port, and this value SHOULD be used by default as the destination UDP port. Some early implementations of VXLAN have used other values for the destination port. To enable interoperability with these implementations, the destination port

SHOULD be configurable."

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

I think if we have a rte_flow action type that can be
used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action
set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have

a better idea.

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't

have.

Quoted 5 lines
Quoted 3 lines

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can

only be configured by a PF driver or a privileged VF base on our security model.

Quoted 3 lines

A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use rte_eth_dev_udp_tunnel_port_add

to create tunnel port for eCPRI, them this will impact on all VFs in the same PF.

Quoted 8 lines
Quoted 1 line

2. A normal VF driver can create rte_flow rule that match specific patch for

queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

Whaooh! A normal Intel VF is not allowed to match the tunnel it wants if not enabled by a priviledged VF?

I would say it is a HW design flaw, but that's not the question.

Why you think this is a design flaw? in real case, is it a typical requirement that different VF need different tunnel port for eCPRI (or VxLan) on the same PF? I believe it's not necessary to make it as a per VF resource in most cases, and I will be surprise if a driver that allow any VF to change the share resource without any privilege control.

Btw I guess mlx NIC has more flexible way to handle ecpri tunnel, just curious how it works, what's the expected result of below rules?

1. create flow eth / ipv4 / udp dst is 1234 / ecpri msgtype is 0 / ... to queue 0
2. create flow eth / ipv4 / udp dst is 5678 / ecrpi msgtype is 1 / ... to queue 1.

So both 1234 and 5678 will be regarded as an ECPRI packet? Or only the first one will work? does dst udp port is always needed if an ecpri pattern is involved?

Quoted 66 lines
Quoted 3 lines

So it does not help to have a rte_flow that match dst udp port as tunnel while have an action in the same rule.

Now I understand you are truly looking for a device configuration.
But as it looks more as a HW limitation, I don't think it should be part of ethdev. The fact is that it is already part of ethdev...

I don't know what to say. I need to digest how Intel limitation is "still" trying to drive some parts of the "generic" API.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

#flow create eth/ipv4/udp port is 4789/... action
set_tunnel_type VxLAN

I think rte_eth_dev_udp_tunnel_port_add does this job well already, if we plan to move it to rte_flow, at least we need a replacement

solution.

Quoted 22 lines
Quoted 22 lines

The documentation does not say why it is useful.
With rte_flow you don't need it because a flow is specified with its action.

Quoted 3 lines

Let me see if I understand it correctly.
In your case, the issue is that you need to configure the HW to parse the

packet correctly right?

Quoted 3 lines

It is not about the matching it is about the configuration of the HW, you wish to tell the HW that the packet should be parsed by different means

correct?

Quoted 9 lines

If this is the case it sounds to me that you should use rte_flow and if the user adds the following rule:
#flow create pattern eth / ivp4 / udp port is 4789 / .. action ..... You simply need to configure your HW to support the ecpri configuration.

Quoted 2 lines
Quoted 2 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

Isn't this more a device configuration than filtering, not sure about using rte_flow for this.

+1

A device configuration? No, setting an UDP port is a stack

configuration.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

BTW, are we going to move all other filter like mac , VLAN filter/strip/insert into rte_flow finally?

Yes I think it should be the direction.
All of this can be achieved with rte_flow.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

if that's the plan, though I don't have much inputs for this right now, but I think we may not need to prevent new features be added based on current API if it does not introduce more complexity and not break anything.

If we continue updating old API, we are just forking ourself:
having 2 APIs for the same feature is a non-sense. We must remove APIs which are superseded by rte_flow.

#29 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

11/01/2021 15:02, Zhang, Qi Z:

Quoted 42 lines

From: Thomas Monjalon <[email protected]>

Quoted 41 lines

11/01/2021 12:26, Zhang, Qi Z:

Quoted 25 lines

From: Thomas Monjalon <[email protected]>

Quoted 24 lines

10/01/2021 11:46, Ori Kam:

Quoted 23 lines

From: Zhang, Qi Z <[email protected]>

Quoted 22 lines

From: Thomas Monjalon <[email protected]>

Quoted 15 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 7 lines · Ferruh Yigit
Quoted 6 lines · Qi Z
Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

Yes the port number is free.
But isn't it more natural to specify this port number as part of the rte_flow rule?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't

have.

Quoted 5 lines
Quoted 3 lines

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can

only be configured by a PF driver or a privileged VF base on our security model.

Quoted 3 lines

A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use rte_eth_dev_udp_tunnel_port_add

to create tunnel port for eCPRI, them this will impact on all VFs in the same PF.

Quoted 1 line

2. A normal VF driver can create rte_flow rule that match specific patch for

queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

Quoted 8 lines
Quoted 4 lines

Whaooh! A normal Intel VF is not allowed to match the tunnel it wants if not enabled by a priviledged VF?

I would say it is a HW design flaw, but that's not the question.

Why you think this is a design flaw? in real case,
is it a typical requirement that different VF
need different tunnel port for eCPRI (or VxLan) on the same PF?

They are different VFs, so why should they use the same UDP port? Anyway it doesn't need to be typical to be allowed.

Quoted 3 lines

I believe it's not necessary to make it as a per VF resource in most cases, and I will be surprise if a driver that
allow any VF to change the share resource without any privilege control.

The thing is that a flow rule should not be a shared resource. In Intel devices, it seems the UDP port of a protocol is supposed to be shared with all VFs, but it looks a very specific assumption, or limitation.
I wonder how we can document this and ask the user to call rte_eth_dev_udp_tunnel_port_add(), because of some devices. Anyway, this is currently poorly documented.

Quoted 5 lines

Btw I guess mlx NIC has more flexible way to handle ecpri tunnel, just curious how it works, what's the expected result of below rules?

1. create flow eth / ipv4 / udp dst is 1234 / ecpri msgtype is 0 / ... to queue 0
2. create flow eth / ipv4 / udp dst is 5678 / ecrpi msgtype is 1 / ... to queue 1.

It should move the eCPRI packets to the right queue, taking into consideration the UDP port and the message type. Of course there may be some bugs :)

Quoted 1 line

So both 1234 and 5678 will be regarded as an ECPRI packet?

Yes, both should be considered eCPRI.

Quoted 1 line

Or only the first one will work?

I am not aware of such limitation.

Quoted 1 line

does dst udp port is always needed if an ecpri pattern is involved?

No, the UDP part is optional.

#30 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 67 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Monday, January 11, 2021 10:54 PM
To: Yigit, Ferruh <[email protected]>; Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Andrew Rybchenko <[email protected]>; Ori Kam <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Gregory Etelson <[email protected]>; [email protected]; [email protected]; [email protected]; Bing Zhao <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

11/01/2021 15:02, Zhang, Qi Z:

Quoted 36 lines

From: Thomas Monjalon <[email protected]>

Quoted 35 lines

11/01/2021 12:26, Zhang, Qi Z:

Quoted 26 lines

From: Thomas Monjalon <[email protected]>

Quoted 25 lines

10/01/2021 11:46, Ori Kam:

Quoted 24 lines

From: Zhang, Qi Z <[email protected]>

Quoted 23 lines

From: Thomas Monjalon <[email protected]>

Quoted 15 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 7 lines · Ferruh Yigit
Quoted 6 lines · Qi Z
Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

Yes the port number is free.
But isn't it more natural to specify this port
number as part of the rte_flow rule?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't

have.

Quoted 6 lines
Quoted 3 lines

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can

only be configured by a PF driver or a privileged VF base on our security

model.

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use
rte_eth_dev_udp_tunnel_port_add

to create tunnel port for eCPRI, them this will impact on all VFs in the same

PF.

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

2. A normal VF driver can create rte_flow rule that match specific patch for

queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

Whaooh! A normal Intel VF is not allowed to match the tunnel it wants if not enabled by a priviledged VF?

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

I would say it is a HW design flaw, but that's not the question.

Why you think this is a design flaw? in real case, is it a typical requirement that different VF need different tunnel port for eCPRI (or VxLan) on the same PF?

They are different VFs, so why should they use the same UDP port? Anyway it doesn't need to be typical to be allowed.

Yes, of cause, your can support different UDP tunnel port for different VF, but there are lots of alternative ways to isolate VFs, its just not a big deal for most real use case. The typical requirement is some customer want eCPRI with UDP port A, while another one want UDP port B, and our NIC is good enough to support both cases separately. There are seldom cases that different eCPRI tunnel port need to be deployed on the same NIC or same port.
so from my view, it's a reasonable design compromise that lose minor software flexibility but get a more simplified firmware and save more hardware resource from unnecessary usage.

Quoted 11 lines
Quoted 4 lines

I believe it's not necessary to make it as a per VF resource in most cases, and I will be surprise if a driver that allow any VF to change the share resource without any privilege control.

The thing is that a flow rule should not be a shared resource.
In Intel devices, it seems the UDP port of a protocol is supposed to be shared with all VFs, but it looks a very specific assumption, or limitation. I wonder how we can document this and ask the user to call rte_eth_dev_udp_tunnel_port_add(), because of some devices. Anyway, this is currently poorly documented.

OK, let me check the document to see if anything we can improve.

Quoted 11 lines
Quoted 6 lines

Btw I guess mlx NIC has more flexible way to handle ecpri tunnel, just curious how it works, what's the expected result of below rules?

1. create flow eth / ipv4 / udp dst is 1234 / ecpri msgtype is 0 / ... to queue 0 2. create flow eth / ipv4 / udp dst is 5678 / ecrpi msgtype is 1 / ...

to queue 1.

It should move the eCPRI packets to the right queue, taking into consideration the UDP port and the message type.
Of course there may be some bugs :)

I guess it is not just some bugs, I saw below note in Mellanox latest MLX5 driver. "eCPRI over UDP layer is not yet supported right now",
but this is not the question, I believe your answers are all fit for the VxLan case :)

For VxLAN offload I note below statement from your user manual

*If you configure multiple UDP ports for offload and exceed the total number of ports supported by hardware, then those additional ports will still function properly, but will not benefit from any of the stateless offloads.

Looks like you have a port limitation, additional port that above this number will not work with offload like RSS/steering ...,that's fine. So my understanding the port resource is not just a regular rule in your general flow table.
The questions is how many is the limitation ? does each VF has its own resource pool?
If they are shared, how do you manage these ports?
What if one malicious VF used up all the tunnel ports, does another VF still get chance to create its own?

Quoted 13 lines
Quoted 2 lines

So both 1234 and 5678 will be regarded as an ECPRI packet?

Yes, both should be considered eCPRI.

Quoted 1 line

Or only the first one will work?

I am not aware of such limitation.

Quoted 1 line

does dst udp port is always needed if an ecpri pattern is involved?

No, the UDP part is optional.

#31 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

12/01/2021 03:14, Zhang, Qi Z:

Quoted 72 lines

From: Thomas Monjalon <[email protected]>

Quoted 52 lines

11/01/2021 15:02, Zhang, Qi Z:

Quoted 36 lines

From: Thomas Monjalon <[email protected]>

Quoted 35 lines

11/01/2021 12:26, Zhang, Qi Z:

Quoted 26 lines

From: Thomas Monjalon <[email protected]>

Quoted 25 lines

10/01/2021 11:46, Ori Kam:

Quoted 24 lines

From: Zhang, Qi Z <[email protected]>

Quoted 23 lines

From: Thomas Monjalon <[email protected]>

Quoted 15 lines

08/01/2021 10:29, Andrew Rybchenko:

Quoted 7 lines · Ferruh Yigit
Quoted 6 lines · Qi Z
Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

Yes the port number is free.
But isn't it more natural to specify this port
number as part of the rte_flow rule?

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

I think if we have a rte_flow action type that can be used to set a packet's tunnel type xxx, like below #flow create eth/ipv4/udp port is 4789/... action set_tunnel_type VxLAN / end then we may replace it with rte_flow, but I'm not sure if it's necessary, please share if you have a better idea.

Quoted 3 lines

Of course we can specify the UDP port in rte_flow rule. Please check rte_flow_item_udp.
That's a basic of rte_flow.

Its not about the pattern match, it's about the action, what we need is a rte_flow action to "define a packet's tunnel type", but we don't

have.

Quoted 6 lines
Quoted 3 lines

A packet type alone is meaningless.
It is always associated to an action, this is what rte_flow does.

As I mentioned in previous, this is a device (port) level configuration, so it can

only be configured by a PF driver or a privileged VF base on our security

model.

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

A typical usage in a NFV environment could be:

1. A privileged VF (e.g. ice_dcf PMD) use
rte_eth_dev_udp_tunnel_port_add

to create tunnel port for eCPRI, them this will impact on all VFs in the same

PF.

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

2. A normal VF driver can create rte_flow rule that match specific patch for

queue steering or apply RSS for eCPRI packets, but it DON'T have the permission to define the tunnel port.

Whaooh! A normal Intel VF is not allowed to match the tunnel it wants if not enabled by a priviledged VF?

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

I would say it is a HW design flaw, but that's not the question.

Why you think this is a design flaw? in real case, is it a typical requirement that different VF need different tunnel port for eCPRI (or VxLan) on the same PF?

They are different VFs, so why should they use the same UDP port? Anyway it doesn't need to be typical to be allowed.

Yes, of cause, your can support different UDP tunnel port for different VF, but there are lots of alternative ways to isolate VFs, its just not a big deal for most real use case. The typical requirement is some customer want eCPRI with UDP port A, while another one want UDP port B, and our NIC is good enough to support both cases separately. There are seldom cases that different eCPRI tunnel port need to be deployed on the same NIC or same port.
so from my view, it's a reasonable design compromise that lose minor software flexibility but get a more simplified firmware and save more hardware resource from unnecessary usage.

Quoted 5 lines
Quoted 4 lines

I believe it's not necessary to make it as a per VF resource in most cases, and I will be surprise if a driver that allow any VF to change the share resource without any privilege control.

Quoted 8 lines
Quoted 6 lines

The thing is that a flow rule should not be a shared resource.
In Intel devices, it seems the UDP port of a protocol is supposed to be shared with all VFs, but it looks a very specific assumption, or limitation. I wonder how we can document this and ask the user to call rte_eth_dev_udp_tunnel_port_add(), because of some devices. Anyway, this is currently poorly documented.

OK, let me check the document to see if anything we can improve.

Thank you for trying to improve the doc.

Quoted 17 lines
Quoted 10 lines
Quoted 5 lines

Btw I guess mlx NIC has more flexible way to handle ecpri tunnel, just curious how it works, what's the expected result of below rules?

1. create flow eth / ipv4 / udp dst is 1234 / ecpri msgtype is 0 / ... to queue 0 2. create flow eth / ipv4 / udp dst is 5678 / ecrpi msgtype is 1 / ...

to queue 1.

It should move the eCPRI packets to the right queue, taking into consideration the UDP port and the message type.
Of course there may be some bugs :)

I guess it is not just some bugs, I saw below note in Mellanox latest MLX5 driver. "eCPRI over UDP layer is not yet supported right now",
but this is not the question, I believe your answers are all fit for the VxLan case :)

For VxLAN offload I note below statement from your user manual

Quoted 8 lines

*If you configure multiple UDP ports for offload and exceed the total number of ports supported by hardware, then those additional ports will still function properly, but will not benefit from any of the stateless offloads.

Looks like you have a port limitation, additional port that above this number will not work with offload like RSS/steering ...,that's fine. So my understanding the port resource is not just a regular rule in your general flow table.
The questions is how many is the limitation ? does each VF has its own resource pool?
If they are shared, how do you manage these ports?
What if one malicious VF used up all the tunnel ports, does another VF still get chance to create its own?

Sorry I don't know exactly what are the limitations. From DPDK point of view, when a flow rule cannot be created, it returns an error and the app must handle.
Yes the app must handle limitations because there is no magic with hardware offloads: hardware are all more or less limited, that's a sad truth of our finite world ;)

#32 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

In reply to Thomas Monjalon
Quoted 67 lines · Thomas Monjalon

07/01/2021 13:47, Zhang, Qi Z:

Quoted 41 lines
Quoted 41 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Thursday, January 7, 2021 6:12 PM
To: Guo, Jia <[email protected]>
Cc: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

07/01/2021 10:32, Guo, Jia:

Quoted 27 lines

From: Thomas Monjalon <[email protected]>

Quoted 7 lines

24/12/2020 07:59, Jeff Guo:

Quoted 1 line

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 3 lines

[...]

Quoted 15 lines

lib/librte_ethdev/rte_ethdev.h enum rte_eth_tunnel_type { · line 1224

Quoted 9 lines
lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
  	RTE_TUNNEL_TYPE_IP_IN_GRE,
  	RTE_L2_TUNNEL_TYPE_E_TAG,
  	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
  	RTE_TUNNEL_TYPE_MAX,
  };

We tried to remove all these legacy API in DPDK 20.11. Andrew decided to not remove this one because it is not yet completely replaced by rte_flow in all drivers. However, I am against continuing to update this API. The opposite work should be done: migrate to rte_flow.

Agree but seems that the legacy api and driver legacy implementation still keep in this release, and there is no a general way to replace the legacy by rte_flow right now.

Quoted 8 lines
Quoted 1 line

I think rte_flow is a complete replacement with more features.

Thomas, I may not agree with this.

Actually the "enum rte_eth_tunnel_type" is used by rte_eth_dev_udp_tunnel_port_add
A packet with specific dst udp port will be recognized as a specific tunnel packet type (e.g. vxlan, vxlan-gpe, ecpri...)
In Intel NIC, the API actually changes the configuration of the packet parser in HW but not add a filter rule and I guess all other devices may enable it in a similar way. so naturally it should be a device (port) level configuration but not a rte_flow rule for match, encap, decap...

I don't understand how it helps to identify an UDP port if there is no rule for this tunnel.
What is the usage?

Quoted 8 lines

So I think it's not a good idea to replace
the rte_eth_dev_udp_tunnel_port_add with rte_flow config and also there is no existing rte_flow_action
can cover this requirement unless we introduce some new one.

Quoted 3 lines

You can match, encap, decap.
There is even a new API to get tunnel infos after decap. What is missing?

I still don't see which use case is missing.

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed. Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

Quoted 10 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

Oh, the ABI break should be a problem.

Quoted 2 lines

PS: please Cc ethdev maintainers for such patch, thanks.
tip: use --cc-cmd devtools/get-maintainer.sh

Thanks for your helpful tip.

#33 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 10:22, Ferruh Yigit:

Quoted 8 lines · Thomas Monjalon
Quoted 7 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 6 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

#34 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 15 lines · Thomas Monjalon

08/01/2021 10:22, Ferruh Yigit:

Quoted 14 lines · Thomas Monjalon
Quoted 12 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 8 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

Quoted 7 lines
Quoted 2 lines

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Yes it is exposed to the application. But in runtime does it exchanged between library and application is the issue I think. For this case it seems it is not, so not an ABI break.

#35 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 11:43, Ferruh Yigit:

Quoted 19 lines · Thomas Monjalon
Quoted 18 lines

08/01/2021 10:22, Ferruh Yigit:

Quoted 16 lines · Thomas Monjalon
Quoted 4 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

07/01/2021 10:32, Guo, Jia:

Quoted 1 line

From: Thomas Monjalon <[email protected]>

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of RTE_TUNNEL_TYPE_MAX.

Yes that may break the ABI but fortunately the checking-abi-compatibility tool shows negative :) , thanks Ferruh' s guide.
https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

Quoted 7 lines
Quoted 3 lines

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Yes it is exposed to the application. But in runtime does it exchanged between library and application is the issue I think.
For this case it seems it is not, so not an ABI break.

If I create a table of size RTE_TUNNEL_TYPE_MAX with DPDK 20.11, I will get an overflow when writing to the new ECPRI index. The question is: can I receive the ECPRI value dynamically from ethdev? If yes, it is an ABI breakage. But I cannot think of such case now.

#36 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 42 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday 8 January 2021 14:06
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; Yigit, Ferruh <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>; Kinsella, Ray <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 11:43, Ferruh Yigit:

Quoted 9 lines · Thomas Monjalon
Quoted 8 lines

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h

so

Quoted 3 lines
Quoted 2 lines

it is exposed to the application.
I think it is a case of ABI breakage.

Quoted 7 lines
Quoted 1 line

Yes it is exposed to the application. But in runtime does it

exchanged

Quoted 2 lines

between library and application is the issue I think. For this case it seems it is not, so not an ABI break.

If I create a table of size RTE_TUNNEL_TYPE_MAX with DPDK 20.11, I will get an overflow when writing to the new ECPRI index.

I guess the question is - are you likely to do this?

Quoted 2 lines

The question is: can I receive the ECPRI value dynamically from ethdev? If yes, it is an ABI breakage. But I cannot think of such case now.

Ray K

#37 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 15:07, Kinsella, Ray:

Quoted 29 lines

From: Thomas Monjalon <[email protected]>

Quoted 28 lines

08/01/2021 11:43, Ferruh Yigit:

Quoted 9 lines · Thomas Monjalon
Quoted 8 lines

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept within the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h

so

Quoted 4 lines
Quoted 2 lines

it is exposed to the application.
I think it is a case of ABI breakage.

Yes it is exposed to the application. But in runtime does it

Quoted 8 lines
Quoted 6 lines

exchanged

Quoted 2 lines

between library and application is the issue I think. For this case it seems it is not, so not an ABI break.

If I create a table of size RTE_TUNNEL_TYPE_MAX with DPDK 20.11, I will get an overflow when writing to the new ECPRI index.

I guess the question is - are you likely to do this?

As said below, no I cannot think about such a case myself.

Quoted 2 lines
Quoted 2 lines

The question is: can I receive the ECPRI value dynamically from ethdev? If yes, it is an ABI breakage. But I cannot think of such case now.

#38 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

In reply to Thomas Monjalon
Quoted 32 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday 8 January 2021 10:24
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; Yigit, Ferruh <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>; Kinsella, Ray <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 5 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

Quoted 8 lines
Quoted 1 line

As long as the enum values not sent to the application and kept

within

Quoted 1 line

the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Really a lot depends on context, Thomas is right it is hard to predict how these _MAX values are used.

We have seen cases in the past where _MAX enumeration values have been used to size arrays the like - I don't immediately see that issue here. My understanding is that the only consumer of this enumeration is rte_eth_dev_udp_tunnel_port_add and rte_eth_dev_udp_tunnel_port_delete, right? On face value, impact looks negligible.

I will take a look at why libabigail doesn't complain.

#39 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 41 lines · Ray
Quoted 32 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday 8 January 2021 10:24
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; Yigit, Ferruh <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>; Kinsella, Ray <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 5 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

Quoted 8 lines
Quoted 1 line

As long as the enum values not sent to the application and kept

within

Quoted 1 line

the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Quoted 6 lines

Really a lot depends on context, Thomas is right it is hard to predict how these _MAX values are used.

We have seen cases in the past where _MAX enumeration values have been used to size arrays the like - I don't immediately see that issue here. My understanding is that the only consumer of this enumeration is rte_eth_dev_udp_tunnel_port_add and rte_eth_dev_udp_tunnel_port_delete, right? On face value, impact looks negligible.

I will take a look at why libabigail doesn't complain.

Application can use the enum, including MAX as they desire, we can't really assume anything there.

In previous case, library was providing an enum value back to application. And the problem was application can use those values blindly and new unexpected values may cause trouble.

For this case, even the application create a table with RTE_TUNNEL_TYPE_MAX size, library is not sending any type of this enum to application to cause any problem, at least abigail seems not able to finding any instance of it.

#40 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 75 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Friday 8 January 2021 14:27
To: Kinsella, Ray <[email protected]>; Thomas Monjalon <[email protected]>; Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 6 lines · Ray
Quoted 6 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday 8 January 2021 10:24
To: Guo, Jia <[email protected]>; Zhang, Qi Z

<[email protected]>;

Quoted 5 lines
Quoted 5 lines

Yigit, Ferruh <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>; Kinsella,

Ray

Quoted 2 lines
Quoted 2 lines

<[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel

type

Quoted 21 lines
Quoted 21 lines

for ecpri

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept

within

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Really a lot depends on context, Thomas is right it is hard to

predict how these _MAX values are used.

Quoted 2 lines

We have seen cases in the past where _MAX enumeration values have

been used to size arrays the like - I don't immediately see that issue here. My understanding is that the only consumer of this enumeration is rte_eth_dev_udp_tunnel_port_add and rte_eth_dev_udp_tunnel_port_delete, right? On face value, impact looks negligible.

Quoted 3 lines

I will take a look at why libabigail doesn't complain.

Application can use the enum, including MAX as they desire, we can't really assume anything there.

Quoted 8 lines

In previous case, library was providing an enum value back to application. And the problem was application can use those values blindly and new unexpected values may cause trouble.

For this case, even the application create a table with RTE_TUNNEL_TYPE_MAX size, library is not sending any type of this enum to application to cause any problem, at least abigail seems not able to finding any instance of it.

Yes - it makes any problem associated with unlikely then.

Ray K

#41 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

In reply to Ferruh Yigit
Quoted 60 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Friday 8 January 2021 14:27
To: Kinsella, Ray <[email protected]>; Thomas Monjalon <[email protected]>; Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 6 lines · Ray
Quoted 6 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Friday 8 January 2021 10:24
To: Guo, Jia <[email protected]>; Zhang, Qi Z

<[email protected]>;

Quoted 5 lines
Quoted 5 lines

Yigit, Ferruh <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; Dodji Seketeli <[email protected]>; Kinsella,

Ray

Quoted 2 lines
Quoted 2 lines

<[email protected]>
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel

type

Quoted 28 lines
Quoted 27 lines

for ecpri

08/01/2021 10:22, Ferruh Yigit:

Quoted 7 lines · Thomas Monjalon
Quoted 6 lines

07/01/2021 13:47, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 4 lines

07/01/2021 10:32, Guo, Jia:

Quoted 3 lines

From: Thomas Monjalon <[email protected]>

Quoted 2 lines

Sorry, it is a nack.
BTW, it is probably breaking the ABI because of

RTE_TUNNEL_TYPE_MAX.

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Yes that may break the ABI but fortunately the checking-abi-

compatibility tool shows negative :) , thanks Ferruh' s guide.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

https://github.com/ferruhy/dpdk/actions/runs/468859673

That's very strange. An enum value is changed.
Why it is not flagged by libabigail?

As long as the enum values not sent to the application and kept

within

Quoted 1 line

the library, changing their values shouldn't be problem.

But RTE_TUNNEL_TYPE_MAX is part of lib/librte_ethdev/rte_ethdev.h so it is exposed to the application.
I think it is a case of ABI breakage.

Quoted 10 lines
Quoted 1 line

Really a lot depends on context, Thomas is right it is hard to

predict how these _MAX values are used.

Quoted 2 lines

We have seen cases in the past where _MAX enumeration values have

been used to size arrays the like - I don't immediately see that issue here. My understanding is that the only consumer of this enumeration is rte_eth_dev_udp_tunnel_port_add and rte_eth_dev_udp_tunnel_port_delete, right? On face value, impact looks negligible.

Quoted 2 lines

I will take a look at why libabigail doesn't complain.

So I spent some time looking a bit closer at why libabigail didn't complain,
I am summarizing it is because there is no symbol that obviously uses enum rte_eth_tunnel_type. The prot_type field in rte_eth_udp_tunnel is declared as a uint8_t, not as enum rte_eth_tunnel_type.

Is there a particular reason an enumerated field would be declared as unsigned int instead?

/**
* UDP tunneling configuration.
* Used to config the UDP port for a type of tunnel.
* NICs need the UDP port to identify the tunnel type.
* Normally a type of tunnel has a default UDP port, this structure can be used
* in case if the users want to change or support more UDP port. */
struct rte_eth_udp_tunnel {
uint16_t udp_port; /**< UDP port used for the tunnel. */ uint8_t prot_type; /**< Tunnel type. Defined in rte_eth_tunnel_type. */
};

Quoted 4 lines

Application can use the enum, including MAX as they desire, we can't really assume anything there.

Quoted 8 lines

In previous case, library was providing an enum value back to application. And the problem was application can use those values blindly and new unexpected values may cause trouble.

For this case, even the application create a table with RTE_TUNNEL_TYPE_MAX size, library is not sending any type of this enum to application to cause any problem, at least abigail seems not able to finding any instance of it.

I agree - I think this has marginal risk.

author
#43 lore raw

Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 15 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Thursday, January 14, 2021 10:35 PM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 1/2] ethdev: add new tunnel type for ecpri

Quoted 1 line · Jeff Guo

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 1 line
Quoted 7 lines
Quoted 2 lines

But since you will be sending a new version of the set, can you please include a short release notes update?

Sure, please check v3 again, thanks.

author
#44 lore raw

PATCH v2 2/2 app/testpmd: add new UDP tunnel port for ecpri

In reply to Jeff Guo

Add new UDP tunnel port params for ecpri configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
 app/test-pmd/cmdline.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
app/test-pmd/cmdline.c +5 -2
index 2ccbaa039e..af08e48e2e 100644
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
 	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
 	} else {
 		printf("Invalid tunnel type\n");
 		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
 				 "add#rm");
 cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
 	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
 cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
 			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
 	.f = cmd_cfg_tunnel_udp_port_parsed,
 	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
 	.tokens = {
 		(void *)&cmd_config_tunnel_udp_port_port,
 		(void *)&cmd_config_tunnel_udp_port_config,
Signature · 1 line
2.20.1
#45 lore raw

Re: [dpdk-dev] [dpdk-dev v2 2/2] app/testpmd: add new UDP tunnel port for ecpri

app/test-pmd/cmdline.c

Quoted 12 lines · Jeff Guo

Add new UDP tunnel port params for ecpri configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
  app/test-pmd/cmdline.c | 7 +++++--
  1 file changed, 5 insertions(+), 2 deletions(-)
app/test-pmd/cmdline.c +0 -0

Can you please update the 'cmd_help_long_parsed()' for command update, and update the testpmd documentation, 'testpmd_funcs.rst'?

Quoted 32 lines

index 2ccbaa039e..af08e48e2e 100644

app/test-pmd/cmdline.c +5 -2
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
  		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
  	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
  		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
  	} else {
  		printf("Invalid tunnel type\n");
  		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
  				 "add#rm");
  cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
  	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
  cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
  	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
  			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
  cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
  	.f = cmd_cfg_tunnel_udp_port_parsed,
  	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
  	.tokens = {
  		(void *)&cmd_config_tunnel_udp_port_port,
  		(void *)&cmd_config_tunnel_udp_port_config,
author
#46 lore raw

Re: [dpdk-dev] [dpdk-dev v2 2/2] app/testpmd: add new UDP tunnel port for ecpri

Quoted 23 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Thursday, January 14, 2021 10:34 PM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]
Subject: Re: [dpdk-dev] [dpdk-dev v2 2/2] app/testpmd: add new UDP tunnel port for ecpri

Quoted 11 lines · Jeff Guo

Add new UDP tunnel port params for ecpri configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 port config 0 testpmd> udp_tunnel_port rm ecpri 6789

---
  app/test-pmd/cmdline.c | 7 +++++--
  1 file changed, 5 insertions(+), 2 deletions(-)
Quoted 5 lines

app/test-pmd/cmdline.c

Quoted 1 line
app/test-pmd/cmdline.c +0 -0

Can you please update the 'cmd_help_long_parsed()' for command update, and update the testpmd documentation, 'testpmd_funcs.rst'?

Oh, sure, that is what I should do for this patch set. Please expect next version.

Quoted 39 lines

app/test-pmd/cmdline.c cmd_cfg_tunnel_udp_port_parsed(void

Quoted 4 lines

index 2ccbaa039e..af08e48e2e 100644

app/test-pmd/cmdline.c +0 -0
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void

*parsed_result,

app/test-pmd/cmdline.c cmdline_parse_token_string_t

Quoted 9 lines
app/test-pmd/cmdline.c +2 -0
  		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
  	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
  		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
  	} else {
  		printf("Invalid tunnel type\n");
  		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t

cmd_config_tunnel_udp_port_action =

app/test-pmd/cmdline.c cmdline_parse_token_string_t · line 9212

Quoted 2 lines
app/test-pmd/cmdline.c +0 -0
  				 "add#rm");
  cmdline_parse_token_string_t

cmd_config_tunnel_udp_port_tunnel_type =

app/test-pmd/cmdline.c cmdline_parse_token_string_t · line 9213

Quoted 1 line
app/test-pmd/cmdline.c +0 -0
  	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port,

tunnel_type,

app/test-pmd/cmdline.c cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =

Quoted 4 lines
app/test-pmd/cmdline.c +1 -1
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
  cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
  	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port,

udp_port,

app/test-pmd/cmdline.c cmdline_parse_token_num_t

Quoted 2 lines
app/test-pmd/cmdline.c +0 -0
  			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t

cmd_config_tunnel_udp_port_value =

app/test-pmd/cmdline.c cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =

Quoted 4 lines
app/test-pmd/cmdline.c +0 -1
  cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
  	.f = cmd_cfg_tunnel_udp_port_parsed,
  	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm

vxlan|geneve|vxlan-gpe <udp_port>",

Quoted 6 lines
app/test-pmd/cmdline.c +2 -0
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
  	.tokens = {
  		(void *)&cmd_config_tunnel_udp_port_port,
  		(void *)&cmd_config_tunnel_udp_port_config,
Reviewed-by on the v2 series
#47 lore raw

Re: [dpdk-dev] [dpdk-dev v2 0/2] add new UDP tunnel port for ecpri

In reply to Jeff Guo
Quoted 20 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Thursday, December 24, 2020 3:00 PM
To: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]; Guo, Jia <[email protected]>
Subject: [dpdk-dev v2 0/2] add new UDP tunnel port for ecpri

Add new UDP tunnel type and port params for ecpri configuration.

v2:
separate patch set

Jeff Guo (2):
ethdev: add new tunnel type for ecpri
app/testpmd: add new UDP tunnel port for ecpri

Quoted 6 lines
 app/test-pmd/cmdline.c         | 7 +++++--
 lib/librte_ethdev/rte_ethdev.h | 1 +
 2 files changed, 6 insertions(+), 2 deletions(-)

--
2.20.1

author
#48 lore raw

v2 cover enable UDP ecpri configure in dcf

In reply to Jeff Guo

Enabling ecpri UDP tunnel port configure in dcf.

v2:
refactor PTYPE parsing and add related sharecode patch separate patch set

Jeff Guo (6):
net/ice/base: add package PTYPE enable information net/ice: refactor package type parsing net/ice/base: add new UDP tunnel type for ecpri net/ice: add PTYPE mapping for ecpri net/iavf: add PTYPE mapping for ecpri net/ice: enable ecpri tunnel port configure in dcf

 drivers/net/iavf/iavf_rxtx.c         |  44 +++++++++
 drivers/net/ice/base/ice_flex_pipe.c |  80 +++++++++++++++
 drivers/net/ice/base/ice_flex_pipe.h |   3 +
 drivers/net/ice/base/ice_flex_type.h |  20 ++++
 drivers/net/ice/base/ice_type.h      |   1 +
 drivers/net/ice/ice_acl_filter.c     |   3 +-
 drivers/net/ice/ice_dcf_ethdev.c     |  67 +++++++++++++
 drivers/net/ice/ice_fdir_filter.c    |  63 ++----------
 drivers/net/ice/ice_generic_flow.c   |  27 +++++-
 drivers/net/ice/ice_generic_flow.h   |   9 +-
 drivers/net/ice/ice_hash.c           |  47 ++-------
 drivers/net/ice/ice_rxtx.c           |  44 +++++++++
 drivers/net/ice/ice_switch_filter.c  | 139 +++------------------------
 13 files changed, 323 insertions(+), 224 deletions(-)
Signature · 1 line
2.20.1
author
#49 lore raw

PATCH v2 1/6 net/ice/base: add package PTYPE enable information

Scan the 'Marker PType TCAM' session to retrieve the Rx parser PTYPE enable information from the current package.

---
 drivers/net/ice/base/ice_flex_pipe.c | 79 ++++++++++++++++++++++++++++
 drivers/net/ice/base/ice_flex_pipe.h |  3 ++
 drivers/net/ice/base/ice_flex_type.h | 19 +++++++
 drivers/net/ice/base/ice_type.h      |  1 +
 4 files changed, 102 insertions(+)
drivers/net/ice/base/ice_flex_pipe.c +79 -0
index 7594df1696..987bea5623 100644
@@ -315,6 +315,84 @@ ice_pkg_enum_entry(struct ice_seg *ice_seg, struct ice_pkg_enum *state,
 	return entry;
 }
 
+/**
+ * ice_hw_ptype_ena - check if the PTYPE is enabled or not
+ * @hw: pointer to the HW structure
+ * @ptype: the hardware PTYPE
+ */
+bool ice_hw_ptype_ena(struct ice_hw *hw, u16 ptype)
+{
+	return ptype < ICE_FLOW_PTYPE_MAX &&
+	       ice_is_bit_set(hw->hw_ptype, ptype);
+}
+
+/**
+ * ice_marker_ptype_tcam_handler
+ * @sect_type: section type
+ * @section: pointer to section
+ * @index: index of the Marker PType TCAM entry to be returned
+ * @offset: pointer to receive absolute offset, always 0 for ptype TCAM sections
+ *
+ * This is a callback function that can be passed to ice_pkg_enum_entry.
+ * Handles enumeration of individual Marker PType TCAM entries.
+ */
+static void *
+ice_marker_ptype_tcam_handler(u32 sect_type, void *section, u32 index,
+			      u32 *offset)
+{
+	struct ice_marker_ptype_tcam_section *marker_ptype;
+
+	if (!section)
+		return NULL;
+
+	if (sect_type != ICE_SID_RXPARSER_MARKER_PTYPE)
+		return NULL;
+
+	if (index > ICE_MAX_MARKER_PTYPE_TCAMS_IN_BUF)
+		return NULL;
+
+	if (offset)
+		*offset = 0;
+
+	marker_ptype = (struct ice_marker_ptype_tcam_section *)section;
+
+	if (index >= LE16_TO_CPU(marker_ptype->count))
+		return NULL;
+
+	return marker_ptype->tcam + index;
+}
+
+/**
+ * ice_fill_hw_ptype - fill the enabled PTYPE bit information
+ * @hw: pointer to the HW structure
+ */
+static void
+ice_fill_hw_ptype(struct ice_hw *hw)
+{
+	struct ice_marker_ptype_tcam_entry *tcam;
+	struct ice_seg *seg = hw->seg;
+	struct ice_pkg_enum state;
+
+	ice_zero_bitmap(hw->hw_ptype, ICE_FLOW_PTYPE_MAX);
+	if (!seg)
+		return;
+
+	ice_memset(&state, 0, sizeof(state), ICE_NONDMA_MEM);
+
+	do {
+		tcam = (struct ice_marker_ptype_tcam_entry *)
+			ice_pkg_enum_entry(seg, &state,
+					   ICE_SID_RXPARSER_MARKER_PTYPE, NULL,
+					   ice_marker_ptype_tcam_handler);
+		if (tcam &&
+		    LE16_TO_CPU(tcam->addr) < ICE_MARKER_PTYPE_TCAM_ADDR_MAX &&
+		    LE16_TO_CPU(tcam->ptype) < ICE_FLOW_PTYPE_MAX)
+			ice_set_bit(LE16_TO_CPU(tcam->ptype), hw->hw_ptype);
+
+		seg = NULL;
+	} while (tcam);
+}
+
 /**
  * ice_boost_tcam_handler
  * @sect_type: section type
@@ -1511,6 +1589,7 @@ enum ice_status ice_init_pkg(struct ice_hw *hw, u8 *buf, u32 len)
 		 */
 		ice_init_pkg_regs(hw);
 		ice_fill_blk_tbls(hw);
+		ice_fill_hw_ptype(hw);
 		ice_get_prof_index_max(hw);
 	} else {
 		ice_debug(hw, ICE_DBG_INIT, "package load failed, %d\n",
drivers/net/ice/base/ice_flex_pipe.h +3 -0
index 214c7a2837..66965d0975 100644
@@ -48,6 +48,9 @@ bool ice_tunnel_port_in_use(struct ice_hw *hw, u16 port, u16 *index);
 bool
 ice_tunnel_get_type(struct ice_hw *hw, u16 port, enum ice_tunnel_type *type);
 
+/* RX parser PType functions */
+bool ice_hw_ptype_ena(struct ice_hw *hw, u16 ptype);
+
 /* XLT2/VSI group functions */
 enum ice_status
 ice_vsig_find_vsi(struct ice_hw *hw, enum ice_block blk, u16 vsi, u16 *vsig);
drivers/net/ice/base/ice_flex_type.h +19 -0
index 1dd57baccd..9213856ac1 100644
@@ -472,6 +472,25 @@ struct ice_boost_tcam_section {
 	sizeof(struct ice_boost_tcam_entry), \
 	sizeof(struct ice_boost_tcam_entry))
 
+/* package Marker PType TCAM entry */
+struct ice_marker_ptype_tcam_entry {
+#define ICE_MARKER_PTYPE_TCAM_ADDR_MAX	1024
+	__le16 addr;
+	__le16 ptype;
+	u8 keys[20];
+};
+
+struct ice_marker_ptype_tcam_section {
+	__le16 count;
+	__le16 reserved;
+	struct ice_marker_ptype_tcam_entry tcam[STRUCT_HACK_VAR_LEN];
+};
+
+#define ICE_MAX_MARKER_PTYPE_TCAMS_IN_BUF ICE_MAX_ENTRIES_IN_BUF( \
+	ice_struct_size((struct ice_marker_ptype_tcam_section *)0, tcam, 1) - \
+	sizeof(struct ice_marker_ptype_tcam_entry), \
+	sizeof(struct ice_marker_ptype_tcam_entry))
+
 struct ice_xlt1_section {
 	__le16 count;
 	__le16 offset;
drivers/net/ice/base/ice_type.h +1 -0
index 6b8d44f0b4..8ba7ded9a2 100644
@@ -985,6 +985,7 @@ struct ice_hw {
 	ice_declare_bitmap(fdir_perfect_fltr, ICE_FLTR_PTYPE_MAX);
 	struct ice_lock rss_locks;	/* protect RSS configuration */
 	struct LIST_HEAD_TYPE rss_list_head;
+	ice_declare_bitmap(hw_ptype, ICE_FLOW_PTYPE_MAX);
 };
 
 /* Statistics collected by each port, VSI, VEB, and S-channel */
Signature · 1 line
2.20.1
author
#50 lore raw

PATCH v2 2/6 net/ice: refactor package type parsing

In reply to Jeff Guo

If the PTYPE support of the package could be checked, no need to parse different PTYPE list for each type of the package. So, refactor the package type parsing mechanism in each flow engins.

---
 drivers/net/ice/ice_acl_filter.c    |   3 +-
 drivers/net/ice/ice_fdir_filter.c   |  63 ++-----------
 drivers/net/ice/ice_generic_flow.c  |  27 +++++-
 drivers/net/ice/ice_generic_flow.h  |   9 +-
 drivers/net/ice/ice_hash.c          |  47 ++--------
 drivers/net/ice/ice_switch_filter.c | 139 ++++------------------------
 6 files changed, 64 insertions(+), 224 deletions(-)
drivers/net/ice/ice_acl_filter.c +2 -1
index f7dbe53574..363ce68318 100644
@@ -914,7 +914,8 @@ ice_acl_parse(struct ice_adapter *ad,
 	int ret;
 
 	memset(filter, 0, sizeof(*filter));
-	item = ice_search_pattern_match_item(pattern, array, array_len, error);
+	item = ice_search_pattern_match_item(ad, pattern, array, array_len,
+					     error);
 	if (!item)
 		return -rte_errno;
 
drivers/net/ice/ice_fdir_filter.c +10 -53

105 lines not on this page · raw message

drivers/net/ice/ice_generic_flow.c +23 -4
index 1429cbc3b6..34a209af9d 100644
@@ -2029,6 +2029,21 @@ ice_pattern_skip_void_item(struct rte_flow_item *items,
 	rte_memcpy(items, pe, sizeof(struct rte_flow_item));
 }
 
+static bool
+ice_pattern_is_support(__rte_unused struct ice_adapter *ad,
+		       const struct rte_flow_item *pattern)
+{
+	const struct rte_flow_item *pb = pattern;
+
+	while (pb != RTE_FLOW_ITEM_TYPE_END) {
+		if (!ice_hw_ptype_ena(&ad->hw, pb->type))
+			return false;
+		pb++;
+	}
+
+	return true;
+}
+
 /* Check if the pattern matches a supported item type array */
 static bool
 ice_match_pattern(enum rte_flow_item_type *item_array,
@@ -2047,10 +2062,11 @@ ice_match_pattern(enum rte_flow_item_type *item_array,
 }
 
 struct ice_pattern_match_item *
-ice_search_pattern_match_item(const struct rte_flow_item pattern[],
-		struct ice_pattern_match_item *array,
-		uint32_t array_len,
-		struct rte_flow_error *error)
+ice_search_pattern_match_item(struct ice_adapter *ad,
+			      const struct rte_flow_item pattern[],
+			      struct ice_pattern_match_item *array,
+			      uint32_t array_len,
+			      struct rte_flow_error *error)
 {
 	uint16_t i = 0;
 	struct ice_pattern_match_item *pattern_match_item;
@@ -2083,6 +2099,9 @@ ice_search_pattern_match_item(const struct rte_flow_item pattern[],
 
 	ice_pattern_skip_void_item(items, pattern);
 
+	if (!ice_pattern_is_support(ad, pattern))
+		return NULL;
+
 	for (i = 0; i < array_len; i++)
 		if (ice_match_pattern(array[i].pattern_list,
 					items)) {
drivers/net/ice/ice_generic_flow.h +5 -4
index 434d2f425d..0dcb620809 100644
@@ -593,10 +593,11 @@ int ice_register_parser(struct ice_flow_parser *parser,
 void ice_unregister_parser(struct ice_flow_parser *parser,
 		struct ice_adapter *ad);
 struct ice_pattern_match_item *
-ice_search_pattern_match_item(const struct rte_flow_item pattern[],
-		struct ice_pattern_match_item *array,
-		uint32_t array_len,
-		struct rte_flow_error *error);
+ice_search_pattern_match_item(struct ice_adapter *ad,
+			      const struct rte_flow_item pattern[],
+			      struct ice_pattern_match_item *array,
+			      uint32_t array_len,
+			      struct rte_flow_error *error);
 int
 ice_flow_redirect(struct ice_adapter *ad,
 		  struct ice_flow_redirect *rd);
drivers/net/ice/ice_hash.c +8 -39

78 lines not on this page · raw message

drivers/net/ice/ice_switch_filter.c +16 -123

190 lines not on this page · raw message

Signature · 1 line
2.20.1
author
#51 lore raw

PATCH v2 3/6 net/ice/base: add new UDP tunnel type for ecpri

In reply to Jeff Guo

Add new UDP tunnel type for ecpri.

---
 drivers/net/ice/base/ice_flex_pipe.c | 1 +
 drivers/net/ice/base/ice_flex_type.h | 1 +
 2 files changed, 2 insertions(+)
drivers/net/ice/base/ice_flex_pipe.c +1 -0
index 987bea5623..31a4695343 100644
@@ -13,6 +13,7 @@
 static const struct ice_tunnel_type_scan tnls[] = {
 	{ TNL_VXLAN,		"TNL_VXLAN_PF" },
 	{ TNL_GENEVE,		"TNL_GENEVE_PF" },
+	{ TNL_ECPRI,		"TNL_UDP_ECPRI_PF" },
 	{ TNL_LAST,		"" }
 };
 
drivers/net/ice/base/ice_flex_type.h +1 -0
index 9213856ac1..f75a1005f3 100644
@@ -535,6 +535,7 @@ struct ice_pkg_enum {
 enum ice_tunnel_type {
 	TNL_VXLAN = 0,
 	TNL_GENEVE,
+	TNL_ECPRI,
 	TNL_LAST = 0xFF,
 	TNL_ALL = 0xFF,
 };
Signature · 1 line
2.20.1
author
#52 lore raw

PATCH v2 4/6 net/ice: add PTYPE mapping for ecpri

In reply to Jeff Guo

Until the new ecpri PTYPE be added into the RTE lib, just mapping ecpri to the PTYPE of RTE_PTYPE_L4_UDP in ice pmd.

---
 drivers/net/ice/ice_rxtx.c | 44 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
drivers/net/ice/ice_rxtx.c +44 -0
index d052bd0f1b..95be6f6791 100644
@@ -3880,6 +3880,50 @@ ice_get_default_pkt_type(uint16_t ptype)
 			RTE_PTYPE_TUNNEL_GTPU |
 			RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
 			RTE_PTYPE_INNER_L4_ICMP,
+
+		/* IPv4 --> UDP ECPRI */
+		[372] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[373] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[374] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[375] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[376] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[377] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[378] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[379] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[380] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[381] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+
+		/* IPV6 --> UDP ECPRI */
+		[382] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[383] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[384] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[385] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[386] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[387] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[388] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[389] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[390] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[391] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
 		/* All others reserved */
 	};
 
Signature · 1 line
2.20.1
author
#53 lore raw

PATCH v2 5/6 net/iavf: add PTYPE mapping for ecpri

In reply to Jeff Guo

Until the new ecpri PTYPE be added into the RTE lib, just mapping ecpri to the PTYPE of RTE_PTYPE_L4_UDP in iavf pmd.

---
 drivers/net/iavf/iavf_rxtx.c | 44 ++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
drivers/net/iavf/iavf_rxtx.c +44 -0
index 21d508b3f4..2800dd6250 100644
@@ -3193,6 +3193,50 @@ iavf_get_default_ptype_table(void)
 			RTE_PTYPE_TUNNEL_GTPU |
 			RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
 			RTE_PTYPE_INNER_L4_ICMP,
+
+		/* IPv4 --> UDP ECPRI */
+		[372] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[373] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[374] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[375] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[376] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[377] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[378] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[379] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[380] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[381] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+
+		/* IPV6 --> UDP ECPRI */
+		[382] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[383] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[384] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[385] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[386] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[387] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[388] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[389] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[390] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[391] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
 		/* All others reserved */
 	};
 
Signature · 1 line
2.20.1
author
#54 lore raw

PATCH v2 6/6 net/ice: enable ecpri tunnel port configure in dcf

In reply to Jeff Guo

Add ecpri tunnel port add and rm ops to configure ecpri udp tunnel port in dcf.

---
 drivers/net/ice/ice_dcf_ethdev.c | 67 ++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
drivers/net/ice/ice_dcf_ethdev.c +67 -0
index b0b2ecb0d6..08265db79f 100644
@@ -26,6 +26,13 @@
 #include "ice_dcf_ethdev.h"
 #include "ice_rxtx.h"
 
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+
 static uint16_t
 ice_dcf_recv_pkts(__rte_unused void *rx_queue,
 		  __rte_unused struct rte_mbuf **bufs,
@@ -870,6 +877,64 @@ ice_dcf_link_update(__rte_unused struct rte_eth_dev *dev,
 	return 0;
 }
 
+/* Add UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+		ret = ice_create_tunnel(parent_hw, TNL_VXLAN,
+					udp_tunnel->udp_port);
+		break;
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_create_tunnel(parent_hw, TNL_ECPRI,
+					udp_tunnel->udp_port);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+/* Delete UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_destroy_tunnel(parent_hw, udp_tunnel->udp_port, 0);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
 static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.dev_start               = ice_dcf_dev_start,
 	.dev_stop                = ice_dcf_dev_stop,
@@ -892,6 +957,8 @@ static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.allmulticast_enable     = ice_dcf_dev_allmulticast_enable,
 	.allmulticast_disable    = ice_dcf_dev_allmulticast_disable,
 	.filter_ctrl             = ice_dcf_dev_filter_ctrl,
+	.udp_tunnel_port_add	 = ice_dcf_dev_udp_tunnel_port_add,
+	.udp_tunnel_port_del	 = ice_dcf_dev_udp_tunnel_port_del,
 };
 
 static int
Signature · 1 line
2.20.1
author
#55 lore raw

v3 cover net/ice: refactor PTYPE parsing

In reply to Jeff Guo

If the capability of a PTYPE within a specific package could be negotiated, no need to maintain a different PTYPE list for each type of the package when parsing PTYPE. So refactor the PTYPE parsing mechanism for each flow engines.

v3:
separate the patch set from the ecpri configure patch set

Jeff Guo (3):
net/ice/base: add package PTYPE enable information net/ice/base: add PTYPE value
net/ice: refactor PTYPE parsing

 drivers/net/ice/base/ice_flex_pipe.c |  79 ++++++++++++++
 drivers/net/ice/base/ice_flex_pipe.h |   3 +
 drivers/net/ice/base/ice_flex_type.h | 157 ++++++++++++++++++++-------
 drivers/net/ice/base/ice_type.h      |   1 +
 drivers/net/ice/ice_acl_filter.c     |   3 +-
 drivers/net/ice/ice_fdir_filter.c    |  63 ++---------
 drivers/net/ice/ice_generic_flow.c   | 132 ++++++++++++++++++++--
 drivers/net/ice/ice_generic_flow.h   |   9 +-
 drivers/net/ice/ice_hash.c           |  47 ++------
 drivers/net/ice/ice_switch_filter.c  | 139 +++---------------------
 10 files changed, 366 insertions(+), 267 deletions(-)
Signature · 1 line
2.20.1
author
#56 lore raw

PATCH v3 1/3 net/ice/base: add package PTYPE enable information

Scan the 'Marker PType TCAM' session to retrieve the Rx parser PTYPE enable information from the current package.

---
 drivers/net/ice/base/ice_flex_pipe.c | 79 ++++++++++++++++++++++++++++
 drivers/net/ice/base/ice_flex_pipe.h |  3 ++
 drivers/net/ice/base/ice_flex_type.h | 19 +++++++
 drivers/net/ice/base/ice_type.h      |  1 +
 4 files changed, 102 insertions(+)
drivers/net/ice/base/ice_flex_pipe.c +79 -0
index 96aed3b795..96628ae5dc 100644
@@ -316,6 +316,84 @@ ice_pkg_enum_entry(struct ice_seg *ice_seg, struct ice_pkg_enum *state,
 	return entry;
 }
 
+/**
+ * ice_hw_ptype_ena - check if the PTYPE is enabled or not
+ * @hw: pointer to the HW structure
+ * @ptype: the hardware PTYPE
+ */
+bool ice_hw_ptype_ena(struct ice_hw *hw, u16 ptype)
+{
+	return ptype < ICE_FLOW_PTYPE_MAX &&
+	       ice_is_bit_set(hw->hw_ptype, ptype);
+}
+
+/**
+ * ice_marker_ptype_tcam_handler
+ * @sect_type: section type
+ * @section: pointer to section
+ * @index: index of the Marker PType TCAM entry to be returned
+ * @offset: pointer to receive absolute offset, always 0 for ptype TCAM sections
+ *
+ * This is a callback function that can be passed to ice_pkg_enum_entry.
+ * Handles enumeration of individual Marker PType TCAM entries.
+ */
+static void *
+ice_marker_ptype_tcam_handler(u32 sect_type, void *section, u32 index,
+			      u32 *offset)
+{
+	struct ice_marker_ptype_tcam_section *marker_ptype;
+
+	if (!section)
+		return NULL;
+
+	if (sect_type != ICE_SID_RXPARSER_MARKER_PTYPE)
+		return NULL;
+
+	if (index > ICE_MAX_MARKER_PTYPE_TCAMS_IN_BUF)
+		return NULL;
+
+	if (offset)
+		*offset = 0;
+
+	marker_ptype = (struct ice_marker_ptype_tcam_section *)section;
+
+	if (index >= LE16_TO_CPU(marker_ptype->count))
+		return NULL;
+
+	return marker_ptype->tcam + index;
+}
+
+/**
+ * ice_fill_hw_ptype - fill the enabled PTYPE bit information
+ * @hw: pointer to the HW structure
+ */
+static void
+ice_fill_hw_ptype(struct ice_hw *hw)
+{
+	struct ice_marker_ptype_tcam_entry *tcam;
+	struct ice_seg *seg = hw->seg;
+	struct ice_pkg_enum state;
+
+	ice_zero_bitmap(hw->hw_ptype, ICE_FLOW_PTYPE_MAX);
+	if (!seg)
+		return;
+
+	ice_memset(&state, 0, sizeof(state), ICE_NONDMA_MEM);
+
+	do {
+		tcam = (struct ice_marker_ptype_tcam_entry *)
+			ice_pkg_enum_entry(seg, &state,
+					   ICE_SID_RXPARSER_MARKER_PTYPE, NULL,
+					   ice_marker_ptype_tcam_handler);
+		if (tcam &&
+		    LE16_TO_CPU(tcam->addr) < ICE_MARKER_PTYPE_TCAM_ADDR_MAX &&
+		    LE16_TO_CPU(tcam->ptype) < ICE_FLOW_PTYPE_MAX)
+			ice_set_bit(LE16_TO_CPU(tcam->ptype), hw->hw_ptype);
+
+		seg = NULL;
+	} while (tcam);
+}
+
 /**
  * ice_boost_tcam_handler
  * @sect_type: section type
@@ -1541,6 +1619,7 @@ enum ice_status ice_init_pkg(struct ice_hw *hw, u8 *buf, u32 len)
 		 */
 		ice_init_pkg_regs(hw);
 		ice_fill_blk_tbls(hw);
+		ice_fill_hw_ptype(hw);
 		ice_get_prof_index_max(hw);
 	} else {
 		ice_debug(hw, ICE_DBG_INIT, "package load failed, %d\n",
drivers/net/ice/base/ice_flex_pipe.h +3 -0
index d4679cc940..066ff95e13 100644
@@ -54,6 +54,9 @@ bool ice_tunnel_port_in_use(struct ice_hw *hw, u16 port, u16 *index);
 bool
 ice_tunnel_get_type(struct ice_hw *hw, u16 port, enum ice_tunnel_type *type);
 
+/* RX parser PType functions */
+bool ice_hw_ptype_ena(struct ice_hw *hw, u16 ptype);
+
 /* XLT2/VSI group functions */
 enum ice_status
 ice_vsig_find_vsi(struct ice_hw *hw, enum ice_block blk, u16 vsi, u16 *vsig);
drivers/net/ice/base/ice_flex_type.h +19 -0
index 62cc81b49c..9b9503b3ba 100644
@@ -472,6 +472,25 @@ struct ice_boost_tcam_section {
 	sizeof(struct ice_boost_tcam_entry), \
 	sizeof(struct ice_boost_tcam_entry))
 
+/* package Marker PType TCAM entry */
+struct ice_marker_ptype_tcam_entry {
+#define ICE_MARKER_PTYPE_TCAM_ADDR_MAX	1024
+	__le16 addr;
+	__le16 ptype;
+	u8 keys[20];
+};
+
+struct ice_marker_ptype_tcam_section {
+	__le16 count;
+	__le16 reserved;
+	struct ice_marker_ptype_tcam_entry tcam[STRUCT_HACK_VAR_LEN];
+};
+
+#define ICE_MAX_MARKER_PTYPE_TCAMS_IN_BUF ICE_MAX_ENTRIES_IN_BUF( \
+	ice_struct_size((struct ice_marker_ptype_tcam_section *)0, tcam, 1) - \
+	sizeof(struct ice_marker_ptype_tcam_entry), \
+	sizeof(struct ice_marker_ptype_tcam_entry))
+
 struct ice_xlt1_section {
 	__le16 count;
 	__le16 offset;
drivers/net/ice/base/ice_type.h +1 -0
index bb2cfd07fd..86e93c34a1 100644
@@ -995,6 +995,7 @@ struct ice_hw {
 	struct ice_lock rss_locks;	/* protect RSS configuration */
 	struct LIST_HEAD_TYPE rss_list_head;
 	struct ice_vlan_mode_ops vlan_mode_ops;
+	ice_declare_bitmap(hw_ptype, ICE_FLOW_PTYPE_MAX);
 };
 
 /* Statistics collected by each port, VSI, VEB, and S-channel */
Signature · 1 line
2.20.1
author
#57 lore raw

PATCH v3 2/3 net/ice/base: add PTYPE value

In reply to Jeff Guo

Add some macros for some PType value.

---
 drivers/net/ice/base/ice_flex_type.h | 138 +++++++++++++++++++--------
 1 file changed, 99 insertions(+), 39 deletions(-)
drivers/net/ice/base/ice_flex_type.h +99 -39
index 9b9503b3ba..da408ec49c 100644
@@ -267,45 +267,105 @@ enum ice_sect {
 };
 
 /* Packet Type (PTYPE) values */
-#define ICE_PTYPE_MAC_PAY		1
-#define ICE_PTYPE_IPV4FRAG_PAY		22
-#define ICE_PTYPE_IPV4_PAY		23
-#define ICE_PTYPE_IPV4_UDP_PAY		24
-#define ICE_PTYPE_IPV4_TCP_PAY		26
-#define ICE_PTYPE_IPV4_SCTP_PAY		27
-#define ICE_PTYPE_IPV4_ICMP_PAY		28
-#define ICE_PTYPE_IPV6FRAG_PAY		88
-#define ICE_PTYPE_IPV6_PAY		89
-#define ICE_PTYPE_IPV6_UDP_PAY		90
-#define ICE_PTYPE_IPV6_TCP_PAY		92
-#define ICE_PTYPE_IPV6_SCTP_PAY		93
-#define ICE_PTYPE_IPV6_ICMP_PAY		94
-#define ICE_MAC_IPV4_GTPC_TEID		325
-#define ICE_MAC_IPV6_GTPC_TEID		326
-#define ICE_MAC_IPV4_GTPC		327
-#define ICE_MAC_IPV6_GTPC		328
-#define ICE_MAC_IPV4_GTPU		329
-#define ICE_MAC_IPV6_GTPU		330
-#define ICE_MAC_IPV4_GTPU_IPV4_FRAG	331
-#define ICE_MAC_IPV4_GTPU_IPV4_PAY	332
-#define ICE_MAC_IPV4_GTPU_IPV4_UDP_PAY	333
-#define ICE_MAC_IPV4_GTPU_IPV4_TCP	334
-#define ICE_MAC_IPV4_GTPU_IPV4_ICMP	335
-#define ICE_MAC_IPV6_GTPU_IPV4_FRAG	336
-#define ICE_MAC_IPV6_GTPU_IPV4_PAY	337
-#define ICE_MAC_IPV6_GTPU_IPV4_UDP_PAY	338
-#define ICE_MAC_IPV6_GTPU_IPV4_TCP	339
-#define ICE_MAC_IPV6_GTPU_IPV4_ICMP	340
-#define ICE_MAC_IPV4_GTPU_IPV6_FRAG	341
-#define ICE_MAC_IPV4_GTPU_IPV6_PAY	342
-#define ICE_MAC_IPV4_GTPU_IPV6_UDP_PAY	343
-#define ICE_MAC_IPV4_GTPU_IPV6_TCP	344
-#define ICE_MAC_IPV4_GTPU_IPV6_ICMPV6	345
-#define ICE_MAC_IPV6_GTPU_IPV6_FRAG	346
-#define ICE_MAC_IPV6_GTPU_IPV6_PAY	347
-#define ICE_MAC_IPV6_GTPU_IPV6_UDP_PAY	348
-#define ICE_MAC_IPV6_GTPU_IPV6_TCP	349
-#define ICE_MAC_IPV6_GTPU_IPV6_ICMPV6	350
+#define ICE_PTYPE_MAC_PAY			1
+#define ICE_MAC_PTP				2
+#define ICE_MAC_LLDP				6
+#define ICE_MAC_ARP				11
+#define ICE_PTYPE_IPV4FRAG_PAY			22
+#define ICE_PTYPE_IPV4_PAY			23
+#define ICE_PTYPE_IPV4_UDP_PAY			24
+#define ICE_PTYPE_IPV4_TCP_PAY			26
+#define ICE_PTYPE_IPV4_SCTP_PAY			27
+#define ICE_PTYPE_IPV4_ICMP_PAY			28
+#define ICE_MAC_IPV4_TUN_PAY			43
+#define ICE_MAC_IPV4_TUN_IPV4_FRAG		44
+#define ICE_MAC_IPV4_TUN_IPV4_PAY		45
+#define ICE_MAC_IPV4_TUN_IPV4_UDP_PAY		46
+#define ICE_MAC_IPV4_TUN_IPV4_TCP		48
+#define ICE_MAC_IPV4_TUN_IPV4_SCTP		49
+#define ICE_MAC_IPV4_TUN_IPV4_ICMP		50
+#define ICE_MAC_IPV4_TUN_IPV6_FRAG		51
+#define ICE_MAC_IPV4_TUN_IPV6_PAY		52
+#define ICE_MAC_IPV4_TUN_IPV6_UDP_PAY		53
+#define ICE_MAC_IPV4_TUN_IPV6_TCP		55
+#define ICE_MAC_IPV4_TUN_IPV6_SCTP		56
+#define ICE_MAC_IPV4_TUN_IPV6_ICMPV6		57
+#define ICE_MAC_IPV4_TUN_ICE_MAC_PAY		58
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_FRAG	59
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_PAY	60
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_UDP_PAY	61
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_TCP	63
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_SCTP	64
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_ICMP	65
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_FRAG	66
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_PAY	67
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_UDP_PAY	68
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_TCP	70
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_SCTP	71
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_ICMPV6	72
+#define ICE_PTYPE_IPV6FRAG_PAY			88
+#define ICE_PTYPE_IPV6_PAY			89
+#define ICE_PTYPE_IPV6_UDP_PAY			90
+#define ICE_PTYPE_IPV6_TCP_PAY			92
+#define ICE_PTYPE_IPV6_SCTP_PAY			93
+#define ICE_PTYPE_IPV6_ICMP_PAY			94
+#define ICE_MAC_IPV4_ESP			160
+#define ICE_MAC_IPV6_ESP			161
+#define ICE_MAC_IPV4_AH				162
+#define ICE_MAC_IPV6_AH				163
+#define ICE_MAC_IPV4_NAT_T_ESP			164
+#define ICE_MAC_IPV6_NAT_T_ESP			165
+#define ICE_MAC_IPV4_NAT_T_IKE			166
+#define ICE_MAC_IPV6_NAT_T_IKE			167
+#define ICE_MAC_IPV4_NAT_T_KEEP			168
+#define ICE_MAC_IPV6_NAT_T_KEEP			169
+#define ICE_MAC_PPPOD_PAY			300
+#define ICE_MAC_PPPOE_PAY			301
+#define ICE_MAC_PPPOE_IPV4_FRAG			302
+#define ICE_MAC_PPPOE_IPV4_PAY			303
+#define ICE_MAC_PPPOE_IPV4_UDP_PAY		304
+#define ICE_MAC_PPPOE_IPV4_TCP			305
+#define ICE_MAC_PPPOE_IPV4_SCTP			306
+#define ICE_MAC_PPPOE_IPV4_ICMP			307
+#define ICE_MAC_PPPOE_IPV6_FRAG			308
+#define ICE_MAC_PPPOE_IPV6_PAY			309
+#define ICE_MAC_PPPOE_IPV6_UDP_PAY		310
+#define ICE_MAC_PPPOE_IPV6_TCP			311
+#define ICE_MAC_PPPOE_IPV6_SCTP			312
+#define ICE_MAC_PPPOE_IPV6_ICMPV6		313
+#define ICE_MAC_IPV4_GTPC_TEID			325
+#define ICE_MAC_IPV6_GTPC_TEID			326
+#define ICE_MAC_IPV4_GTPC			327
+#define ICE_MAC_IPV6_GTPC			328
+#define ICE_MAC_IPV4_GTPU			329
+#define ICE_MAC_IPV6_GTPU			330
+#define ICE_MAC_IPV4_GTPU_IPV4_FRAG		331
+#define ICE_MAC_IPV4_GTPU_IPV4_PAY		332
+#define ICE_MAC_IPV4_GTPU_IPV4_UDP_PAY		333
+#define ICE_MAC_IPV4_GTPU_IPV4_TCP		334
+#define ICE_MAC_IPV4_GTPU_IPV4_ICMP		335
+#define ICE_MAC_IPV6_GTPU_IPV4_FRAG		336
+#define ICE_MAC_IPV6_GTPU_IPV4_PAY		337
+#define ICE_MAC_IPV6_GTPU_IPV4_UDP_PAY		338
+#define ICE_MAC_IPV6_GTPU_IPV4_TCP		339
+#define ICE_MAC_IPV6_GTPU_IPV4_ICMP		340
+#define ICE_MAC_IPV4_GTPU_IPV6_FRAG		341
+#define ICE_MAC_IPV4_GTPU_IPV6_PAY		342
+#define ICE_MAC_IPV4_GTPU_IPV6_UDP_PAY		343
+#define ICE_MAC_IPV4_GTPU_IPV6_TCP		344
+#define ICE_MAC_IPV4_GTPU_IPV6_ICMPV6		345
+#define ICE_MAC_IPV6_GTPU_IPV6_FRAG		346
+#define ICE_MAC_IPV6_GTPU_IPV6_PAY		347
+#define ICE_MAC_IPV6_GTPU_IPV6_UDP_PAY		348
+#define ICE_MAC_IPV6_GTPU_IPV6_TCP		349
+#define ICE_MAC_IPV6_GTPU_IPV6_ICMPV6		350
+#define ICE_MAC_IPV4_PFCP_NODE			351
+#define ICE_MAC_IPV4_PFCP_SESSION		352
+#define ICE_MAC_IPV6_PFCP_NODE			353
+#define ICE_MAC_IPV6_PFCP_SESSION		354
+#define ICE_MAC_IPV4_L2TPV3			360
+#define ICE_MAC_IPV6_L2TPV3			361
+
 
 /* Attributes that can modify PTYPE definitions.
  *
Signature · 1 line
2.20.1
author
#58 lore raw

PATCH v3 3/3 net/ice: refactor PTYPE parsing

In reply to Jeff Guo

If the capability of a PTYPE within a specific package could be negotiated, no need to maintain a different PTYPE list for each type of the package when parsing PTYPE. So refactor the PTYPE parsing mechanism for each flow engines.

---
 drivers/net/ice/ice_acl_filter.c    |   3 +-
 drivers/net/ice/ice_fdir_filter.c   |  63 ++-----------
 drivers/net/ice/ice_generic_flow.c  | 132 ++++++++++++++++++++++++--
 drivers/net/ice/ice_generic_flow.h  |   9 +-
 drivers/net/ice/ice_hash.c          |  47 ++--------
 drivers/net/ice/ice_switch_filter.c | 139 ++++------------------------
 6 files changed, 165 insertions(+), 228 deletions(-)
drivers/net/ice/ice_acl_filter.c +2 -1
index f7dbe53574..363ce68318 100644
@@ -914,7 +914,8 @@ ice_acl_parse(struct ice_adapter *ad,
 	int ret;
 
 	memset(filter, 0, sizeof(*filter));
-	item = ice_search_pattern_match_item(pattern, array, array_len, error);
+	item = ice_search_pattern_match_item(ad, pattern, array, array_len,
+					     error);
 	if (!item)
 		return -rte_errno;
 
drivers/net/ice/ice_fdir_filter.c +10 -53

105 lines not on this page · raw message

drivers/net/ice/ice_generic_flow.c +124 -8

164 lines not on this page · raw message

drivers/net/ice/ice_generic_flow.h +5 -4
index 434d2f425d..0dcb620809 100644
@@ -593,10 +593,11 @@ int ice_register_parser(struct ice_flow_parser *parser,
 void ice_unregister_parser(struct ice_flow_parser *parser,
 		struct ice_adapter *ad);
 struct ice_pattern_match_item *
-ice_search_pattern_match_item(const struct rte_flow_item pattern[],
-		struct ice_pattern_match_item *array,
-		uint32_t array_len,
-		struct rte_flow_error *error);
+ice_search_pattern_match_item(struct ice_adapter *ad,
+			      const struct rte_flow_item pattern[],
+			      struct ice_pattern_match_item *array,
+			      uint32_t array_len,
+			      struct rte_flow_error *error);
 int
 ice_flow_redirect(struct ice_adapter *ad,
 		  struct ice_flow_redirect *rd);
drivers/net/ice/ice_hash.c +8 -39

78 lines not on this page · raw message

drivers/net/ice/ice_switch_filter.c +16 -123

190 lines not on this page · raw message

Signature · 1 line
2.20.1
author
#59 lore raw

v4 cover net/ice: refactor PTYPE parsing

In reply to Jeff Guo

If the capability of a PTYPE within a specific package could be negotiated, no need to maintain a different PTYPE list for each type of the package when parsing PTYPE. So refactor the PTYPE parsing mechanism for each flow engines.

v4:
rebase patch and add some more PTYPE macros

v3:
separate the patch set from the ecpri configure patch set

Jeff Guo (2):
net/ice/base: add PTYPE value
net/ice: refactor PTYPE parsing

 drivers/net/ice/base/ice_flex_type.h | 189 +++++++++++++++++++++------
 drivers/net/ice/ice_acl_filter.c     |   3 +-
 drivers/net/ice/ice_fdir_filter.c    |  63 ++-------
 drivers/net/ice/ice_generic_flow.c   | 132 +++++++++++++++++--
 drivers/net/ice/ice_generic_flow.h   |   9 +-
 drivers/net/ice/ice_hash.c           |  51 ++------
 drivers/net/ice/ice_switch_filter.c  | 139 +++-----------------
 7 files changed, 315 insertions(+), 271 deletions(-)
Signature · 1 line
2.20.1
author
#60 lore raw

PATCH v4 1/2 net/ice/base: add PTYPE value

Add some macros for some PType value.

---
 drivers/net/ice/base/ice_flex_type.h | 189 +++++++++++++++++++++------
 1 file changed, 150 insertions(+), 39 deletions(-)
drivers/net/ice/base/ice_flex_type.h +150 -39
index ad620f7892..c7f92b9150 100644
@@ -267,45 +267,156 @@ enum ice_sect {
 };
 
 /* Packet Type (PTYPE) values */
-#define ICE_PTYPE_MAC_PAY		1
-#define ICE_PTYPE_IPV4FRAG_PAY		22
-#define ICE_PTYPE_IPV4_PAY		23
-#define ICE_PTYPE_IPV4_UDP_PAY		24
-#define ICE_PTYPE_IPV4_TCP_PAY		26
-#define ICE_PTYPE_IPV4_SCTP_PAY		27
-#define ICE_PTYPE_IPV4_ICMP_PAY		28
-#define ICE_PTYPE_IPV6FRAG_PAY		88
-#define ICE_PTYPE_IPV6_PAY		89
-#define ICE_PTYPE_IPV6_UDP_PAY		90
-#define ICE_PTYPE_IPV6_TCP_PAY		92
-#define ICE_PTYPE_IPV6_SCTP_PAY		93
-#define ICE_PTYPE_IPV6_ICMP_PAY		94
-#define ICE_MAC_IPV4_GTPC_TEID		325
-#define ICE_MAC_IPV6_GTPC_TEID		326
-#define ICE_MAC_IPV4_GTPC		327
-#define ICE_MAC_IPV6_GTPC		328
-#define ICE_MAC_IPV4_GTPU		329
-#define ICE_MAC_IPV6_GTPU		330
-#define ICE_MAC_IPV4_GTPU_IPV4_FRAG	331
-#define ICE_MAC_IPV4_GTPU_IPV4_PAY	332
-#define ICE_MAC_IPV4_GTPU_IPV4_UDP_PAY	333
-#define ICE_MAC_IPV4_GTPU_IPV4_TCP	334
-#define ICE_MAC_IPV4_GTPU_IPV4_ICMP	335
-#define ICE_MAC_IPV6_GTPU_IPV4_FRAG	336
-#define ICE_MAC_IPV6_GTPU_IPV4_PAY	337
-#define ICE_MAC_IPV6_GTPU_IPV4_UDP_PAY	338
-#define ICE_MAC_IPV6_GTPU_IPV4_TCP	339
-#define ICE_MAC_IPV6_GTPU_IPV4_ICMP	340
-#define ICE_MAC_IPV4_GTPU_IPV6_FRAG	341
-#define ICE_MAC_IPV4_GTPU_IPV6_PAY	342
-#define ICE_MAC_IPV4_GTPU_IPV6_UDP_PAY	343
-#define ICE_MAC_IPV4_GTPU_IPV6_TCP	344
-#define ICE_MAC_IPV4_GTPU_IPV6_ICMPV6	345
-#define ICE_MAC_IPV6_GTPU_IPV6_FRAG	346
-#define ICE_MAC_IPV6_GTPU_IPV6_PAY	347
-#define ICE_MAC_IPV6_GTPU_IPV6_UDP_PAY	348
-#define ICE_MAC_IPV6_GTPU_IPV6_TCP	349
-#define ICE_MAC_IPV6_GTPU_IPV6_ICMPV6	350
+#define ICE_PTYPE_MAC_PAY			1
+#define ICE_MAC_PTP				2
+#define ICE_MAC_LLDP				6
+#define ICE_MAC_ARP				11
+#define ICE_PTYPE_IPV4FRAG_PAY			22
+#define ICE_PTYPE_IPV4_PAY			23
+#define ICE_PTYPE_IPV4_UDP_PAY			24
+#define ICE_PTYPE_IPV4_TCP_PAY			26
+#define ICE_PTYPE_IPV4_SCTP_PAY			27
+#define ICE_PTYPE_IPV4_ICMP_PAY			28
+#define ICE_MAC_IPV4_IPV4_FRAG			29
+#define ICE_MAC_IPV4_IPV4_PAY			30
+#define ICE_MAC_IPV4_IPV4_UDP_PAY		31
+#define ICE_MAC_IPV4_IPV4_TCP			33
+#define ICE_MAC_IPV4_IPV4_SCTP			34
+#define ICE_MAC_IPV4_IPV4_ICMP			35
+#define ICE_MAC_IPV4_IPV6_FRAG			36
+#define ICE_MAC_IPV4_IPV6_PAY			37
+#define ICE_MAC_IPV4_IPV6_UDP_PAY		38
+#define ICE_MAC_IPV4_IPV6_TCP			40
+#define ICE_MAC_IPV4_IPV6_SCTP			41
+#define ICE_MAC_IPV4_IPV6_ICMPV6		42
+#define ICE_MAC_IPV4_TUN_PAY			43
+#define ICE_MAC_IPV4_TUN_IPV4_FRAG		44
+#define ICE_MAC_IPV4_TUN_IPV4_PAY		45
+#define ICE_MAC_IPV4_TUN_IPV4_UDP_PAY		46
+#define ICE_MAC_IPV4_TUN_IPV4_TCP		48
+#define ICE_MAC_IPV4_TUN_IPV4_SCTP		49
+#define ICE_MAC_IPV4_TUN_IPV4_ICMP		50
+#define ICE_MAC_IPV4_TUN_IPV6_FRAG		51
+#define ICE_MAC_IPV4_TUN_IPV6_PAY		52
+#define ICE_MAC_IPV4_TUN_IPV6_UDP_PAY		53
+#define ICE_MAC_IPV4_TUN_IPV6_TCP		55
+#define ICE_MAC_IPV4_TUN_IPV6_SCTP		56
+#define ICE_MAC_IPV4_TUN_IPV6_ICMPV6		57
+#define ICE_MAC_IPV4_TUN_ICE_MAC_PAY		58
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_FRAG	59
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_PAY	60
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_UDP_PAY	61
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_TCP	63
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_SCTP	64
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV4_ICMP	65
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_FRAG	66
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_PAY	67
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_UDP_PAY	68
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_TCP	70
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_SCTP	71
+#define ICE_MAC_IPV4_TUN_ICE_MAC_IPV6_ICMPV6	72
+#define ICE_PTYPE_IPV6FRAG_PAY			88
+#define ICE_PTYPE_IPV6_PAY			89
+#define ICE_PTYPE_IPV6_UDP_PAY			90
+#define ICE_PTYPE_IPV6_TCP_PAY			92
+#define ICE_PTYPE_IPV6_SCTP_PAY			93
+#define ICE_PTYPE_IPV6_ICMP_PAY			94
+#define ICE_MAC_IPV6_IPV4_FRAG			95
+#define ICE_MAC_IPV6_IPV4_PAY			96
+#define ICE_MAC_IPV6_IPV4_UDP_PAY		97
+#define ICE_MAC_IPV6_IPV4_TCP			99
+#define ICE_MAC_IPV6_IPV4_SCTP			100
+#define ICE_MAC_IPV6_IPV4_ICMP			101
+#define ICE_MAC_IPV6_IPV6_FRAG			102
+#define ICE_MAC_IPV6_IPV6_PAY			103
+#define ICE_MAC_IPV6_IPV6_UDP_PAY		104
+#define ICE_MAC_IPV6_IPV6_TCP			106
+#define ICE_MAC_IPV6_IPV6_SCTP			107
+#define ICE_MAC_IPV6_IPV6_ICMPV6		108
+#define ICE_MAC_IPV6_TUN_PAY			109
+#define ICE_MAC_IPV6_TUN_IPV4_FRAG		110
+#define ICE_MAC_IPV6_TUN_IPV4_PAY		111
+#define ICE_MAC_IPV6_TUN_IPV4_UDP_PAY		112
+#define ICE_MAC_IPV6_TUN_IPV4_TCP		114
+#define ICE_MAC_IPV6_TUN_IPV4_SCTP		115
+#define ICE_MAC_IPV6_TUN_IPV4_ICMP		116
+#define ICE_MAC_IPV6_TUN_IPV6_FRAG		117
+#define ICE_MAC_IPV6_TUN_IPV6_PAY		118
+#define ICE_MAC_IPV6_TUN_IPV6_UDP_PAY		119
+#define ICE_MAC_IPV6_TUN_IPV6_TCP		121
+#define ICE_MAC_IPV6_TUN_IPV6_SCTP		122
+#define ICE_MAC_IPV6_TUN_IPV6_ICMPV6		123
+#define ICE_MAC_IPV6_TUN_MAC_PAY		124
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_FRAG		125
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_PAY		126
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_UDP_PAY	127
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_TCP		129
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_SCTP		130
+#define ICE_MAC_IPV6_TUN_MAC_IPV4_ICMP		131
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_FRAG		132
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_PAY		133
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_UDP_PAY	134
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_TCP		136
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_SCTP		137
+#define ICE_MAC_IPV6_TUN_MAC_IPV6_ICMPV6	138
+#define ICE_MAC_IPV4_ESP			160
+#define ICE_MAC_IPV6_ESP			161
+#define ICE_MAC_IPV4_AH				162
+#define ICE_MAC_IPV6_AH				163
+#define ICE_MAC_IPV4_NAT_T_ESP			164
+#define ICE_MAC_IPV6_NAT_T_ESP			165
+#define ICE_MAC_IPV4_NAT_T_IKE			166
+#define ICE_MAC_IPV6_NAT_T_IKE			167
+#define ICE_MAC_IPV4_NAT_T_KEEP			168
+#define ICE_MAC_IPV6_NAT_T_KEEP			169
+#define ICE_MAC_CONTROL				278
+#define ICE_MAC_PPPOD_PAY			300
+#define ICE_MAC_PPPOE_PAY			301
+#define ICE_MAC_PPPOE_IPV4_FRAG			302
+#define ICE_MAC_PPPOE_IPV4_PAY			303
+#define ICE_MAC_PPPOE_IPV4_UDP_PAY		304
+#define ICE_MAC_PPPOE_IPV4_TCP			305
+#define ICE_MAC_PPPOE_IPV4_SCTP			306
+#define ICE_MAC_PPPOE_IPV4_ICMP			307
+#define ICE_MAC_PPPOE_IPV6_FRAG			308
+#define ICE_MAC_PPPOE_IPV6_PAY			309
+#define ICE_MAC_PPPOE_IPV6_UDP_PAY		310
+#define ICE_MAC_PPPOE_IPV6_TCP			311
+#define ICE_MAC_PPPOE_IPV6_SCTP			312
+#define ICE_MAC_PPPOE_IPV6_ICMPV6		313
+#define ICE_MAC_IPV4_GTPC_TEID			325
+#define ICE_MAC_IPV6_GTPC_TEID			326
+#define ICE_MAC_IPV4_GTPC			327
+#define ICE_MAC_IPV6_GTPC			328
+#define ICE_MAC_IPV4_GTPU			329
+#define ICE_MAC_IPV6_GTPU			330
+#define ICE_MAC_IPV4_GTPU_IPV4_FRAG		331
+#define ICE_MAC_IPV4_GTPU_IPV4_PAY		332
+#define ICE_MAC_IPV4_GTPU_IPV4_UDP_PAY		333
+#define ICE_MAC_IPV4_GTPU_IPV4_TCP		334
+#define ICE_MAC_IPV4_GTPU_IPV4_ICMP		335
+#define ICE_MAC_IPV6_GTPU_IPV4_FRAG		336
+#define ICE_MAC_IPV6_GTPU_IPV4_PAY		337
+#define ICE_MAC_IPV6_GTPU_IPV4_UDP_PAY		338
+#define ICE_MAC_IPV6_GTPU_IPV4_TCP		339
+#define ICE_MAC_IPV6_GTPU_IPV4_ICMP		340
+#define ICE_MAC_IPV4_GTPU_IPV6_FRAG		341
+#define ICE_MAC_IPV4_GTPU_IPV6_PAY		342
+#define ICE_MAC_IPV4_GTPU_IPV6_UDP_PAY		343
+#define ICE_MAC_IPV4_GTPU_IPV6_TCP		344
+#define ICE_MAC_IPV4_GTPU_IPV6_ICMPV6		345
+#define ICE_MAC_IPV6_GTPU_IPV6_FRAG		346
+#define ICE_MAC_IPV6_GTPU_IPV6_PAY		347
+#define ICE_MAC_IPV6_GTPU_IPV6_UDP_PAY		348
+#define ICE_MAC_IPV6_GTPU_IPV6_TCP		349
+#define ICE_MAC_IPV6_GTPU_IPV6_ICMPV6		350
+#define ICE_MAC_IPV4_PFCP_NODE			351
+#define ICE_MAC_IPV4_PFCP_SESSION		352
+#define ICE_MAC_IPV6_PFCP_NODE			353
+#define ICE_MAC_IPV6_PFCP_SESSION		354
+#define ICE_MAC_IPV4_L2TPV3			360
+#define ICE_MAC_IPV6_L2TPV3			361
+
 
 /* Attributes that can modify PTYPE definitions.
  *
Signature · 1 line
2.20.1
author
#61 lore raw

PATCH v4 2/2 net/ice: refactor PTYPE parsing

In reply to Jeff Guo

If the capability of a PTYPE within a specific package could be negotiated, no need to maintain a different PTYPE list for each type of the package when parsing PTYPE. So refactor the PTYPE parsing mechanism for each flow engines.

---
 drivers/net/ice/ice_acl_filter.c    |   3 +-
 drivers/net/ice/ice_fdir_filter.c   |  63 ++-----------
 drivers/net/ice/ice_generic_flow.c  | 132 ++++++++++++++++++++++++--
 drivers/net/ice/ice_generic_flow.h  |   9 +-
 drivers/net/ice/ice_hash.c          |  51 ++--------
 drivers/net/ice/ice_switch_filter.c | 139 ++++------------------------
 6 files changed, 165 insertions(+), 232 deletions(-)
drivers/net/ice/ice_acl_filter.c +2 -1
index f7dbe53574..363ce68318 100644
@@ -914,7 +914,8 @@ ice_acl_parse(struct ice_adapter *ad,
 	int ret;
 
 	memset(filter, 0, sizeof(*filter));
-	item = ice_search_pattern_match_item(pattern, array, array_len, error);
+	item = ice_search_pattern_match_item(ad, pattern, array, array_len,
+					     error);
 	if (!item)
 		return -rte_errno;
 
drivers/net/ice/ice_fdir_filter.c +10 -53

105 lines not on this page · raw message

drivers/net/ice/ice_generic_flow.c +124 -8

164 lines not on this page · raw message

drivers/net/ice/ice_generic_flow.h +5 -4
index 434d2f425d..0dcb620809 100644
@@ -593,10 +593,11 @@ int ice_register_parser(struct ice_flow_parser *parser,
 void ice_unregister_parser(struct ice_flow_parser *parser,
 		struct ice_adapter *ad);
 struct ice_pattern_match_item *
-ice_search_pattern_match_item(const struct rte_flow_item pattern[],
-		struct ice_pattern_match_item *array,
-		uint32_t array_len,
-		struct rte_flow_error *error);
+ice_search_pattern_match_item(struct ice_adapter *ad,
+			      const struct rte_flow_item pattern[],
+			      struct ice_pattern_match_item *array,
+			      uint32_t array_len,
+			      struct rte_flow_error *error);
 int
 ice_flow_redirect(struct ice_adapter *ad,
 		  struct ice_flow_redirect *rd);
drivers/net/ice/ice_hash.c +8 -43

82 lines not on this page · raw message

drivers/net/ice/ice_switch_filter.c +16 -123

190 lines not on this page · raw message

Signature · 1 line
2.20.1
Acked-by on the v4 series Applied
#62 lore raw

Re: [dpdk-dev] [dpdk-dev v4 0/2] net/ice: refactor PTYPE parsing

In reply to Jeff Guo
Quoted 35 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Wednesday, January 13, 2021 1:31 PM
To: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]; Guo, Jia <[email protected]>; Su, Simei <[email protected]>
Subject: [dpdk-dev v4 0/2] net/ice: refactor PTYPE parsing

If the capability of a PTYPE within a specific package could be negotiated, no need to maintain a different PTYPE list for each type of the package when parsing PTYPE. So refactor the PTYPE parsing mechanism for each flow engines.

v4:
rebase patch and add some more PTYPE macros

v3:
separate the patch set from the ecpri configure patch set

Jeff Guo (2):
net/ice/base: add PTYPE value
net/ice: refactor PTYPE parsing

 drivers/net/ice/base/ice_flex_type.h | 189 +++++++++++++++++++++------
 drivers/net/ice/ice_acl_filter.c     |   3 +-
 drivers/net/ice/ice_fdir_filter.c    |  63 ++-------
 drivers/net/ice/ice_generic_flow.c   | 132 +++++++++++++++++--
 drivers/net/ice/ice_generic_flow.h   |   9 +-
 drivers/net/ice/ice_hash.c           |  51 ++------
 drivers/net/ice/ice_switch_filter.c  | 139 +++-----------------
 7 files changed, 315 insertions(+), 271 deletions(-)
Quoted 2 lines

--
2.20.1

Applied to dpdk-next-net-intel.

Thanks
Qi

author
#63 lore raw

v3 cover enable UDP ecpri configure in dcf

In reply to Jeff Guo

Enabling ecpri UDP tunnel port configure in dcf.

v3:
seperate patch set.

v2:
refactor PTYPE parsing and add related sharecode patch separate patch set

Jeff Guo (3):
net/ice: add PTYPE mapping for ecpri net/iavf: add PTYPE mapping for ecpri net/ice: enable ecpri tunnel port configure in dcf

 drivers/net/iavf/iavf_rxtx.c     | 44 +++++++++++++++++++++
 drivers/net/ice/ice_dcf_ethdev.c | 67 ++++++++++++++++++++++++++++++++
 drivers/net/ice/ice_rxtx.c       | 44 +++++++++++++++++++++
 3 files changed, 155 insertions(+)
Signature · 1 line
2.20.1
author
#64 lore raw

PATCH v3 1/3 net/ice: add PTYPE mapping for ecpri

Until the new ecpri PTYPE be added into the RTE lib, just mapping ecpri to the PTYPE of RTE_PTYPE_L4_UDP in ice pmd.

---
 drivers/net/ice/ice_rxtx.c | 44 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
drivers/net/ice/ice_rxtx.c +44 -0
index d052bd0f1b..95be6f6791 100644
@@ -3880,6 +3880,50 @@ ice_get_default_pkt_type(uint16_t ptype)
 			RTE_PTYPE_TUNNEL_GTPU |
 			RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
 			RTE_PTYPE_INNER_L4_ICMP,
+
+		/* IPv4 --> UDP ECPRI */
+		[372] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[373] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[374] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[375] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[376] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[377] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[378] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[379] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[380] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[381] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+
+		/* IPV6 --> UDP ECPRI */
+		[382] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[383] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[384] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[385] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[386] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[387] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[388] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[389] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[390] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[391] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
 		/* All others reserved */
 	};
 
Signature · 1 line
2.20.1
author
#65 lore raw

PATCH v3 2/3 net/iavf: add PTYPE mapping for ecpri

In reply to Jeff Guo

Until the new ecpri PTYPE be added into the RTE lib, just mapping ecpri to the PTYPE of RTE_PTYPE_L4_UDP in iavf pmd.

---
 drivers/net/iavf/iavf_rxtx.c | 44 ++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)
drivers/net/iavf/iavf_rxtx.c +44 -0
index d53d7b9840..1ddbad0f1f 100644
@@ -3197,6 +3197,50 @@ iavf_get_default_ptype_table(void)
 			RTE_PTYPE_TUNNEL_GTPU |
 			RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN |
 			RTE_PTYPE_INNER_L4_ICMP,
+
+		/* IPv4 --> UDP ECPRI */
+		[372] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[373] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[374] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[375] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[376] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[377] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[378] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[379] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[380] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[381] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV4_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+
+		/* IPV6 --> UDP ECPRI */
+		[382] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[383] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[384] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[385] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[386] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[387] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[388] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[389] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[390] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
+		[391] = RTE_PTYPE_L2_ETHER | RTE_PTYPE_L3_IPV6_EXT_UNKNOWN |
+			RTE_PTYPE_L4_UDP,
 		/* All others reserved */
 	};
 
Signature · 1 line
2.20.1
author
#66 lore raw

PATCH v3 3/3 net/ice: enable ecpri tunnel port configure in dcf

In reply to Jeff Guo

Add ecpri tunnel port add and rm ops to configure ecpri udp tunnel port in dcf.

---
 drivers/net/ice/ice_dcf_ethdev.c | 67 ++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
drivers/net/ice/ice_dcf_ethdev.c +67 -0
index a9e78064d4..c8357f5054 100644
@@ -26,6 +26,13 @@
 #include "ice_dcf_ethdev.h"
 #include "ice_rxtx.h"
 
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel);
+
 static uint16_t
 ice_dcf_recv_pkts(__rte_unused void *rx_queue,
 		  __rte_unused struct rte_mbuf **bufs,
@@ -870,6 +877,64 @@ ice_dcf_link_update(__rte_unused struct rte_eth_dev *dev,
 	return 0;
 }
 
+/* Add UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+		ret = ice_create_tunnel(parent_hw, TNL_VXLAN,
+					udp_tunnel->udp_port);
+		break;
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_create_tunnel(parent_hw, TNL_ECPRI,
+					udp_tunnel->udp_port);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
+/* Delete UDP tunneling port */
+static int
+ice_dcf_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+				struct rte_eth_udp_tunnel *udp_tunnel)
+{
+	struct ice_dcf_adapter *adapter = dev->data->dev_private;
+	struct ice_adapter *parent_adapter = &adapter->parent;
+	struct ice_hw *parent_hw = &parent_adapter->hw;
+	int ret = 0;
+
+	if (!udp_tunnel)
+		return -EINVAL;
+
+	switch (udp_tunnel->prot_type) {
+	case RTE_TUNNEL_TYPE_VXLAN:
+	case RTE_TUNNEL_TYPE_ECPRI:
+		ret = ice_destroy_tunnel(parent_hw, udp_tunnel->udp_port, 0);
+		break;
+	default:
+		PMD_DRV_LOG(ERR, "Invalid tunnel type");
+		ret = -EINVAL;
+		break;
+	}
+
+	return ret;
+}
+
 static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.dev_start               = ice_dcf_dev_start,
 	.dev_stop                = ice_dcf_dev_stop,
@@ -892,6 +957,8 @@ static const struct eth_dev_ops ice_dcf_eth_dev_ops = {
 	.allmulticast_enable     = ice_dcf_dev_allmulticast_enable,
 	.allmulticast_disable    = ice_dcf_dev_allmulticast_disable,
 	.filter_ctrl             = ice_dcf_dev_filter_ctrl,
+	.udp_tunnel_port_add	 = ice_dcf_dev_udp_tunnel_port_add,
+	.udp_tunnel_port_del	 = ice_dcf_dev_udp_tunnel_port_del,
 };
 
 static int
Signature · 1 line
2.20.1
Acked-by on the v3 series Applied
#67 lore raw
In reply to Jeff Guo
Quoted 17 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Wednesday, January 13, 2021 10:05 PM
To: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]; Guo, Jia <[email protected]>; Su, Simei <[email protected]>
Subject: [dpdk-dev v3 0/3] enable UDP ecpri configure in dcf

Enabling ecpri UDP tunnel port configure in dcf.

v3:
seperate patch set.

Quoted 6 lines

v2:
refactor PTYPE parsing and add related sharecode patch separate patch set

Jeff Guo (3):
net/ice: add PTYPE mapping for ecpri
net/iavf: add PTYPE mapping for ecpri

For patch, 1,2 for new PTYPE mapping.

Applied to dpdk-next-net-intel.

Quoted 1 line

net/ice: enable ecpri tunnel port configure in dcf

For patch 3, it is depends on below patches
1. enable ecpri tunnel type .
http://patchwork.dpdk.org/patch/85702/
2. refine the description for rte_eth_dev_udp_tunnel_port_add
http://patchwork.dpdk.org/patch/86406/

Thomas, Ferruh:
Please let us know if any issue for above patch merge.
Thanks
Qi

#68 lore raw
Quoted 32 lines · Qi Z
Quoted 17 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Wednesday, January 13, 2021 10:05 PM
To: Zhang, Qi Z <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>
Cc: [email protected]; Guo, Jia <[email protected]>; Su, Simei <[email protected]>
Subject: [dpdk-dev v3 0/3] enable UDP ecpri configure in dcf

Enabling ecpri UDP tunnel port configure in dcf.

v3:
seperate patch set.

Quoted 6 lines

v2:
refactor PTYPE parsing and add related sharecode patch separate patch set

Jeff Guo (3):
net/ice: add PTYPE mapping for ecpri
net/iavf: add PTYPE mapping for ecpri

For patch, 1,2 for new PTYPE mapping.

Applied to dpdk-next-net-intel.

Quoted 1 line

net/ice: enable ecpri tunnel port configure in dcf

Quoted 8 lines

For patch 3, it is depends on below patches
1. enable ecpri tunnel type .
http://patchwork.dpdk.org/patch/85702/
2. refine the description for rte_eth_dev_udp_tunnel_port_add
http://patchwork.dpdk.org/patch/86406/

Thomas, Ferruh:
Please let us know if any issue for above patch merge.

First dependency already merged to the next-net and other is a documentation patch but still going with the merged patches and will get third patch with this set as separate patch.

author
#69 lore raw

v3 cover add new UDP tunnel port for ecpri

In reply to Jeff Guo

v3:
seperate the patch set and add some document

Jeff Guo (2):
ethdev: add new tunnel type for ecpri
app/testpmd: add new UDP tunnel port for ecpri

 app/test-pmd/cmdline.c                      | 9 ++++++---
 doc/guides/rel_notes/release_21_02.rst      | 3 ++-
 doc/guides/testpmd_app_ug/testpmd_funcs.rst | 2 +-
 lib/librte_ethdev/rte_ethdev.h              | 1 +
 4 files changed, 10 insertions(+), 5 deletions(-)
Signature · 1 line
2.20.1
author
#70 lore raw

PATCH v3 1/2 ethdev: add new tunnel type for ecpri

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
 doc/guides/rel_notes/release_21_02.rst | 3 ++-
 lib/librte_ethdev/rte_ethdev.h         | 1 +
 2 files changed, 3 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_02.rst +2 -1
index b1bb2d8679..e5168d1312 100644
@@ -110,7 +110,8 @@ ABI Changes
    Also, make sure to start the actual text at the margin.
    =======================================================
 
-* No ABI change that would break compatibility with 20.11.
+* ethdev: the structure ``rte_eth_tunnel_type`` has added one parameter
+  ``RTE_TUNNEL_TYPE_ECPRI`` for ecpri UDP port configuration.
 
 
 Known Issues
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
Signature · 1 line
2.20.1
author
#71 lore raw

PATCH v3 2/2 app/testpmd: add new UDP tunnel port for ecpri

In reply to Jeff Guo

Add new UDP tunnel port params for ecpri configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
 app/test-pmd/cmdline.c                      | 9 ++++++---
 doc/guides/testpmd_app_ug/testpmd_funcs.rst | 2 +-
 2 files changed, 7 insertions(+), 4 deletions(-)
app/test-pmd/cmdline.c +6 -3
index 2ccbaa039e..596dd239fe 100644
@@ -845,7 +845,7 @@ static void cmd_help_long_parsed(void *parsed_result,
 			"fdir_inset|fdir_flx_inset clear all"
 			"    Clear RSS|FDIR|FDIR_FLX input set completely for some pctype\n\n"
 
-			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve (udp_port)\n\n"
+			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve|ecpri (udp_port)\n\n"
 			"    Add/remove UDP tunnel port for tunneling offload\n\n"
 
 			"port config <port_id> rx_offload vlan_strip|"
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
 	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
 	} else {
 		printf("Invalid tunnel type\n");
 		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
 				 "add#rm");
 cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
 	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
 cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
 			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
 	.f = cmd_cfg_tunnel_udp_port_parsed,
 	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
 	.tokens = {
 		(void *)&cmd_config_tunnel_udp_port_port,
 		(void *)&cmd_config_tunnel_udp_port_config,
doc/guides/testpmd_app_ug/testpmd_funcs.rst +1 -1
index 9be450066e..842ac42ce6 100644
@@ -2375,7 +2375,7 @@ port config udp_tunnel_port
 
 Add/remove UDP tunnel port for VXLAN/GENEVE tunneling protocols::
 
-    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe (udp_port)
+    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe|ecpri (udp_port)
 
 port config tx_metadata
 ~~~~~~~~~~~~~~~~~~~~~~~
Signature · 1 line
2.20.1
author
#72 lore raw

v4 cover add new UDP tunnel port configure for eCPRI

In reply to Jeff Guo

Add new UDP tunnel port configure for eCPRI protocol features

v4:
add doc in release note for all related eCPRI features.

v3:
seperate the patch set and add some document

Jeff Guo (2):
ethdev: add new tunnel type for eCPRI
app/testpmd: add new UDP tunnel port for eCPRI

 app/test-pmd/cmdline.c                      |  9 ++++++---
 doc/guides/rel_notes/release_21_02.rst      | 15 ++++++++++++++-
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  2 +-
 lib/librte_ethdev/rte_ethdev.h              |  1 +
 4 files changed, 22 insertions(+), 5 deletions(-)
Signature · 1 line
2.20.1
author
#73 lore raw

PATCH v4 1/2 ethdev: add new tunnel type for eCPRI

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
 doc/guides/rel_notes/release_21_02.rst | 15 ++++++++++++++-
 lib/librte_ethdev/rte_ethdev.h         |  1 +
 2 files changed, 15 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_02.rst +14 -1
index b1bb2d8679..2de6afdb85 100644
@@ -61,6 +61,18 @@ New Features
 
   * Added support for Stingray2 device.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol configure feature. 
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+
 
 Removed Items
 -------------
@@ -110,7 +122,8 @@ ABI Changes
    Also, make sure to start the actual text at the margin.
    =======================================================
 
-* No ABI change that would break compatibility with 20.11.
+* ethdev: the structure ``rte_eth_tunnel_type`` has added one parameter
+  ``RTE_TUNNEL_TYPE_ECPRI`` for eCPRI UDP port configuration.
 
 
 Known Issues
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
Signature · 1 line
2.20.1
author
#74 lore raw

PATCH v4 2/2 app/testpmd: add new UDP tunnel port for eCPRI

In reply to Jeff Guo

Add new UDP tunnel port params for eCPRI configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
 app/test-pmd/cmdline.c                      | 9 ++++++---
 doc/guides/testpmd_app_ug/testpmd_funcs.rst | 2 +-
 2 files changed, 7 insertions(+), 4 deletions(-)
app/test-pmd/cmdline.c +6 -3
index 2ccbaa039e..596dd239fe 100644
@@ -845,7 +845,7 @@ static void cmd_help_long_parsed(void *parsed_result,
 			"fdir_inset|fdir_flx_inset clear all"
 			"    Clear RSS|FDIR|FDIR_FLX input set completely for some pctype\n\n"
 
-			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve (udp_port)\n\n"
+			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve|ecpri (udp_port)\n\n"
 			"    Add/remove UDP tunnel port for tunneling offload\n\n"
 
 			"port config <port_id> rx_offload vlan_strip|"
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
 	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
 	} else {
 		printf("Invalid tunnel type\n");
 		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
 				 "add#rm");
 cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
 	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
 cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
 			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
 	.f = cmd_cfg_tunnel_udp_port_parsed,
 	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
 	.tokens = {
 		(void *)&cmd_config_tunnel_udp_port_port,
 		(void *)&cmd_config_tunnel_udp_port_config,
doc/guides/testpmd_app_ug/testpmd_funcs.rst +1 -1
index 9be450066e..842ac42ce6 100644
@@ -2375,7 +2375,7 @@ port config udp_tunnel_port
 
 Add/remove UDP tunnel port for VXLAN/GENEVE tunneling protocols::
 
-    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe (udp_port)
+    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe|ecpri (udp_port)
 
 port config tx_metadata
 ~~~~~~~~~~~~~~~~~~~~~~~
Signature · 1 line
2.20.1
author
#75 lore raw

v5 cover add new UDP tunnel port configure for eCPRI

In reply to Jeff Guo

Add new UDP tunnel port configure for eCPRI protocol features

v5:
fix a coding style issue.

v4:
add doc in release note for all related eCPRI features.

v3:
seperate the patch set and add some document

Jeff Guo (2):
ethdev: add new tunnel type for eCPRI
app/testpmd: add new UDP tunnel port for eCPRI

 app/test-pmd/cmdline.c                      |  9 ++++++---
 doc/guides/rel_notes/release_21_02.rst      | 15 ++++++++++++++-
 doc/guides/testpmd_app_ug/testpmd_funcs.rst |  2 +-
 lib/librte_ethdev/rte_ethdev.h              |  1 +
 4 files changed, 22 insertions(+), 5 deletions(-)
Signature · 1 line
2.20.1
author
#76 lore raw

PATCH v5 1/2 ethdev: add new tunnel type for eCPRI

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
 doc/guides/rel_notes/release_21_02.rst | 15 ++++++++++++++-
 lib/librte_ethdev/rte_ethdev.h         |  1 +
 2 files changed, 15 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_02.rst +14 -1
index b1bb2d8679..80f71be8e6 100644
@@ -61,6 +61,18 @@ New Features
 
   * Added support for Stingray2 device.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol configure feature.
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+
 
 Removed Items
 -------------
@@ -110,7 +122,8 @@ ABI Changes
    Also, make sure to start the actual text at the margin.
    =======================================================
 
-* No ABI change that would break compatibility with 20.11.
+* ethdev: the structure ``rte_eth_tunnel_type`` has added one parameter
+  ``RTE_TUNNEL_TYPE_ECPRI`` for eCPRI UDP port configuration.
 
 
 Known Issues
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
 	RTE_TUNNEL_TYPE_IP_IN_GRE,
 	RTE_L2_TUNNEL_TYPE_E_TAG,
 	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
 	RTE_TUNNEL_TYPE_MAX,
 };
 
Signature · 1 line
2.20.1
#77 lore raw

Re: [dpdk-dev] [dpdk-dev v5 1/2] ethdev: add new tunnel type for eCPRI

Quoted 20 lines · Jeff Guo

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel type.

---
  doc/guides/rel_notes/release_21_02.rst | 15 ++++++++++++++-
  lib/librte_ethdev/rte_ethdev.h         |  1 +
  2 files changed, 15 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_02.rst +2 -0
index b1bb2d8679..80f71be8e6 100644
@@ -61,6 +61,18 @@ New Features
 
    * Added support for Stingray2 device.
 
+* **Updated the Intel ice driver.**
+

doc/guides/rel_notes/release_21_02.rst New Features · line 75

Quoted 10 lines
doc/guides/rel_notes/release_21_02.rst +10 -0
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol configure feature.
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+

These are not related to the patch, so dropping from the patch.

doc/guides/rel_notes/release_21_02.rst ABI Changes · line 127

Quoted 11 lines
doc/guides/rel_notes/release_21_02.rst +2 -1
  Removed Items
  -------------
@@ -110,7 +122,8 @@ ABI Changes
     Also, make sure to start the actual text at the margin.
     =======================================================
 
-* No ABI change that would break compatibility with 20.11.
+* ethdev: the structure ``rte_eth_tunnel_type`` has added one parameter
+  ``RTE_TUNNEL_TYPE_ECPRI`` for eCPRI UDP port configuration.
 

This is not an ABI break, so should not be in this section, also not an API cange, we can't put there too.
And this change is not big enough to add the new features, perhaps better to remove this and add the PMD feature updates as you did above for the relevant sets, so I am dropping this as well.

Quoted 15 lines
doc/guides/rel_notes/release_21_02.rst +0 -0
  Known Issues
lib/librte_ethdev/rte_ethdev.h +1 -0
index f5f8919186..2cbce958cf 100644
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
  	RTE_TUNNEL_TYPE_IP_IN_GRE,
  	RTE_L2_TUNNEL_TYPE_E_TAG,
  	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
  	RTE_TUNNEL_TYPE_MAX,
  };
 
author
#78 lore raw

Re: [dpdk-dev] [dpdk-dev v5 1/2] ethdev: add new tunnel type for eCPRI

Quoted 51 lines

-----Original Message-----
From: Ferruh Yigit <[email protected]>
Sent: Friday, January 15, 2021 8:23 PM
To: Guo, Jia <[email protected]>; Zhang, Qi Z <[email protected]>; [email protected]; [email protected]; Iremonger, Bernard <[email protected]>; Lu, Wenzhuo <[email protected]>; Xing, Beilei <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Su, Simei <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kinsella, Ray <[email protected]>; [email protected]; [email protected]
Subject: Re: [dpdk-dev v5 1/2] ethdev: add new tunnel type for eCPRI

Quoted 1 line · Jeff Guo

Add type of RTE_TUNNEL_TYPE_ECPRI into the enum of ethdev tunnel

type.

Quoted 15 lines
---
  doc/guides/rel_notes/release_21_02.rst | 15 ++++++++++++++-
  lib/librte_ethdev/rte_ethdev.h         |  1 +
  2 files changed, 15 insertions(+), 1 deletion(-)
doc/guides/rel_notes/release_21_02.rst +0 -0
index b1bb2d8679..80f71be8e6 100644

doc/guides/rel_notes/release_21_02.rst New Features · line 66

Quoted 6 lines
doc/guides/rel_notes/release_21_02.rst +3 -0
@@ -61,6 +61,18 @@ New Features
 
    * Added support for Stingray2 device.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements,

including:

doc/guides/rel_notes/release_21_02.rst New Features · line 68

Quoted 2 lines
doc/guides/rel_notes/release_21_02.rst +2 -0
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol

configure feature.

doc/guides/rel_notes/release_21_02.rst New Features

Quoted 6 lines
doc/guides/rel_notes/release_21_02.rst +6 -0
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI

protocol features.

Quoted 4 lines

doc/guides/rel_notes/release_21_02.rst New Features · line 75

Quoted 1 line
doc/guides/rel_notes/release_21_02.rst +1 -0
+

These are not related to the patch, so dropping from the patch.

Ok.

Quoted 7 lines
Quoted 7 lines
doc/guides/rel_notes/release_21_02.rst +0 -0
  Removed Items
  -------------
@@ -110,7 +122,8 @@ ABI Changes
     Also, make sure to start the actual text at the margin.
     =======================================================
 
Quoted 12 lines

doc/guides/rel_notes/release_21_02.rst ABI Changes · line 127

Quoted 5 lines
doc/guides/rel_notes/release_21_02.rst +3 -1
-* No ABI change that would break compatibility with 20.11.
+* ethdev: the structure ``rte_eth_tunnel_type`` has added one
+parameter
+  ``RTE_TUNNEL_TYPE_ECPRI`` for eCPRI UDP port configuration.
 

This is not an ABI break, so should not be in this section, also not an API cange, we can't put there too.
And this change is not big enough to add the new features, perhaps better to remove this and add the PMD feature updates as you did above for the relevant sets, so I am dropping this as well.

Ok, I will use another coming patch to update the doc. Thanks.

Quoted 15 lines
Quoted 15 lines
doc/guides/rel_notes/release_21_02.rst +0 -0
  Known Issues
(quoted diff) +0 -0

b/lib/librte_ethdev/rte_ethdev.h index f5f8919186..2cbce958cf 100644

lib/librte_ethdev/rte_ethdev.h +1 -0
@@ -1219,6 +1219,7 @@ enum rte_eth_tunnel_type {
  	RTE_TUNNEL_TYPE_IP_IN_GRE,
  	RTE_L2_TUNNEL_TYPE_E_TAG,
  	RTE_TUNNEL_TYPE_VXLAN_GPE,
+	RTE_TUNNEL_TYPE_ECPRI,
  	RTE_TUNNEL_TYPE_MAX,
  };
 
author
#79 lore raw

PATCH v5 2/2 app/testpmd: add new UDP tunnel port for eCPRI

In reply to Jeff Guo

Add new UDP tunnel port params for eCPRI configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

---
 app/test-pmd/cmdline.c                      | 9 ++++++---
 doc/guides/testpmd_app_ug/testpmd_funcs.rst | 2 +-
 2 files changed, 7 insertions(+), 4 deletions(-)
app/test-pmd/cmdline.c +6 -3
index 2ccbaa039e..596dd239fe 100644
@@ -845,7 +845,7 @@ static void cmd_help_long_parsed(void *parsed_result,
 			"fdir_inset|fdir_flx_inset clear all"
 			"    Clear RSS|FDIR|FDIR_FLX input set completely for some pctype\n\n"
 
-			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve (udp_port)\n\n"
+			"port config (port_id) udp_tunnel_port add|rm vxlan|geneve|ecpri (udp_port)\n\n"
 			"    Add/remove UDP tunnel port for tunneling offload\n\n"
 
 			"port config <port_id> rx_offload vlan_strip|"
@@ -9175,6 +9175,8 @@ cmd_cfg_tunnel_udp_port_parsed(void *parsed_result,
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_GENEVE;
 	} else if (!strcmp(res->tunnel_type, "vxlan-gpe")) {
 		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_VXLAN_GPE;
+	} else if (!strcmp(res->tunnel_type, "ecpri")) {
+		tunnel_udp.prot_type = RTE_TUNNEL_TYPE_ECPRI;
 	} else {
 		printf("Invalid tunnel type\n");
 		return;
@@ -9209,7 +9211,7 @@ cmdline_parse_token_string_t cmd_config_tunnel_udp_port_action =
 				 "add#rm");
 cmdline_parse_token_string_t cmd_config_tunnel_udp_port_tunnel_type =
 	TOKEN_STRING_INITIALIZER(struct cmd_config_tunnel_udp_port, tunnel_type,
-				 "vxlan#geneve#vxlan-gpe");
+				 "vxlan#geneve#vxlan-gpe#ecpri");
 cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 	TOKEN_NUM_INITIALIZER(struct cmd_config_tunnel_udp_port, udp_port,
 			      RTE_UINT16);
@@ -9217,7 +9219,8 @@ cmdline_parse_token_num_t cmd_config_tunnel_udp_port_value =
 cmdline_parse_inst_t cmd_cfg_tunnel_udp_port = {
 	.f = cmd_cfg_tunnel_udp_port_parsed,
 	.data = NULL,
-	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe <udp_port>",
+	.help_str = "port config <port_id> udp_tunnel_port add|rm vxlan|"
+		"geneve|vxlan-gpe|ecpri <udp_port>",
 	.tokens = {
 		(void *)&cmd_config_tunnel_udp_port_port,
 		(void *)&cmd_config_tunnel_udp_port_config,
doc/guides/testpmd_app_ug/testpmd_funcs.rst +1 -1
index 9be450066e..842ac42ce6 100644
@@ -2375,7 +2375,7 @@ port config udp_tunnel_port
 
 Add/remove UDP tunnel port for VXLAN/GENEVE tunneling protocols::
 
-    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe (udp_port)
+    testpmd> port config (port_id) udp_tunnel_port add|rm vxlan|geneve|vxlan-gpe|ecpri (udp_port)
 
 port config tx_metadata
 ~~~~~~~~~~~~~~~~~~~~~~~
Signature · 1 line
2.20.1
Reviewed-by on v5 2/2
#80 lore raw

Re: [dpdk-dev] [dpdk-dev v5 2/2] app/testpmd: add new UDP tunnel port for eCPRI

Quoted 7 lines · Jeff Guo

Add new UDP tunnel port params for eCPRI configuration, the command as below:

testpmd> port config 0 udp_tunnel_port add ecpri 6789 testpmd> port config 0 udp_tunnel_port rm ecpri 6789

#81 lore raw

Re: [dpdk-dev] [dpdk-dev v5 0/2] add new UDP tunnel port configure for eCPRI

In reply to Jeff Guo
Quoted 8 lines · Jeff Guo

Add new UDP tunnel port configure for eCPRI protocol features

v5:
fix a coding style issue.

v4:
add doc in release note for all related eCPRI features.

Quoted 7 lines

v3:
seperate the patch set and add some document

Jeff Guo (2):
ethdev: add new tunnel type for eCPRI
app/testpmd: add new UDP tunnel port for eCPRI

Series applied to dpdk-next-net/main, thanks.

author
#82 lore raw

PATCH v4 1/1 net/ice: enable eCPRI tunnel port configure in dcf

In reply to Jeff Guo

Add eCPRI tunnel port add and rm ops to configure eCPRI UDP tunnel port in dcf.

---
 doc/guides/rel_notes/release_21_02.rst | 12 +++++
 drivers/net/ice/ice_dcf_ethdev.c       | 67 ++++++++++++++++++++++++++
 2 files changed, 79 insertions(+)
doc/guides/rel_notes/release_21_02.rst +12 -0
index 3bb0b9a9ed..c84f0bfad9 100644
@@ -72,6 +72,18 @@ New Features
 
   * Added inner UDP/IPv4 support for VXLAN IPv4 GSO.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol feature.
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+
 
 Removed Items
 -------------
drivers/net/ice/ice_dcf_ethdev.c +67 -0

85 lines not on this page · raw message

Signature · 1 line
2.20.1
author
#83 lore raw

PATCH v4 1/1 net/ice: enable eCPRI tunnel port configure in dcf

In reply to Jeff Guo

Add eCPRI tunnel port add and rm ops to configure eCPRI UDP tunnel port in dcf.

---
 doc/guides/rel_notes/release_21_02.rst | 12 +++++
 drivers/net/ice/ice_dcf_ethdev.c       | 67 ++++++++++++++++++++++++++
 2 files changed, 79 insertions(+)
doc/guides/rel_notes/release_21_02.rst +12 -0
index 3bb0b9a9ed..c84f0bfad9 100644
@@ -72,6 +72,18 @@ New Features
 
   * Added inner UDP/IPv4 support for VXLAN IPv4 GSO.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol feature.
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+
 
 Removed Items
 -------------
drivers/net/ice/ice_dcf_ethdev.c +67 -0

85 lines not on this page · raw message

Signature · 1 line
2.20.1
author
#84 lore raw

PATCH v5 1/1 net/ice: enable eCPRI tunnel port configure in dcf

In reply to Jeff Guo

Add eCPRI tunnel port add and rm ops to configure eCPRI UDP tunnel port in dcf.

---
v5:
rebase patch

v4:
add doc

---
 doc/guides/rel_notes/release_21_02.rst | 12 +++++
 drivers/net/ice/ice_dcf_ethdev.c       | 67 ++++++++++++++++++++++++++
 2 files changed, 79 insertions(+)
doc/guides/rel_notes/release_21_02.rst +12 -0
index 78ec2758e2..6b7870779f 100644
@@ -106,6 +106,18 @@ New Features
 
   * Added Double VLAN support for DCF switch QinQ filtering.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol feature.
+
+* **Updated Intel iavf driver.**
+
+  Updated iavf PMD with new features and improvements, including:
+
+  * Added support for FDIR/RSS packet steering for flow type eCPRI protocol features.
+
 
 Removed Items
 -------------
drivers/net/ice/ice_dcf_ethdev.c +67 -0

85 lines not on this page · raw message

Signature · 1 line
2.20.1
Acked-by on v5 1/1 Applied
#85 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

Quoted 45 lines

-----Original Message-----
From: Guo, Jia <[email protected]>
Sent: Tuesday, January 19, 2021 12:19 PM
To: Zhang, Qi Z <[email protected]>; [email protected]; Yigit, Ferruh <[email protected]>; [email protected]; Iremonger, Bernard <[email protected]>; Lu, Wenzhuo <[email protected]>; Xing, Beilei <[email protected]>
Cc: Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Guo, Jia <[email protected]>; Su, Simei <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kinsella, Ray <[email protected]>; [email protected]; [email protected]
Subject: [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

Add eCPRI tunnel port add and rm ops to configure eCPRI UDP tunnel port in dcf.

---
v5:
rebase patch

v4:
add doc

---
 doc/guides/rel_notes/release_21_02.rst | 12 +++++
 drivers/net/ice/ice_dcf_ethdev.c       | 67 ++++++++++++++++++++++++++
 2 files changed, 79 insertions(+)
doc/guides/rel_notes/release_21_02.rst +5 -0
index 78ec2758e2..6b7870779f 100644
@@ -106,6 +106,18 @@ New Features
 
   * Added Double VLAN support for DCF switch QinQ filtering.
 
+* **Updated the Intel ice driver.**
+
+  Updated the Intel ice driver with new features and improvements, including:
+
+  * Added support for UDP dynamic port assignment for eCPRI protocol
Quoted 1 line

feature.

Need to claim this is a feature for ice DCF, reword to

Added support for UDP dynamic port assignment for eCPRI tunnel in DCF.

Applied to dpdk-next-net-intel.

Thanks
Qi

#86 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

20/01/2021 11:14, Zhang, Qi Z:

Quoted 5 lines

From: Guo, Jia <[email protected]>

Quoted 2 lines

+ * Added support for UDP dynamic port assignment for eCPRI protocol feature.

Need to claim this is a feature for ice DCF

What is DCF?

#87 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

Quoted 17 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Wednesday, January 20, 2021 6:20 PM
To: Guo, Jia <[email protected]>; Yigit, Ferruh <[email protected]>; Lu, Wenzhuo <[email protected]>; Xing, Beilei <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: [email protected]; Iremonger, Bernard <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Su, Simei <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kinsella, Ray <[email protected]>; [email protected]; [email protected]
Subject: Re: [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

Quoted 10 lines

20/01/2021 11:14, Zhang, Qi Z:

Quoted 5 lines

From: Guo, Jia <[email protected]>

Quoted 2 lines

+ * Added support for UDP dynamic port assignment for eCPRI protocol feature.

Need to claim this is a feature for ice DCF

What is DCF?

Its "Device Config Function" (DCF), please check ice.rst for detail.

Regards
Qi

#88 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

20/01/2021 11:23, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

What is DCF?

Its "Device Config Function" (DCF), please check ice.rst for detail.

Thank you.
As you know, I am interested in the privileged rights. This is what I found in ice.rst:

A DCF PMD bounds to the device's trusted VF with ID 0 [...]
#. Enable the VF0 trust on::
ip link set dev enp24s0f0 vf 0 trust on

Does it mean only VF 0 can be trusted?

#89 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

Quoted 28 lines

-----Original Message-----
From: Thomas Monjalon <[email protected]>
Sent: Wednesday, January 20, 2021 6:31 PM
To: Guo, Jia <[email protected]>; Yigit, Ferruh <[email protected]>; Lu, Wenzhuo <[email protected]>; Xing, Beilei <[email protected]>; Zhang, Qi Z <[email protected]>
Cc: [email protected]; Iremonger, Bernard <[email protected]>; Wu, Jingjing <[email protected]>; Yang, Qiming <[email protected]>; Wang, Haiyue <[email protected]>; [email protected]; Su, Simei <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Kinsella, Ray <[email protected]>; [email protected]; [email protected]
Subject: Re: [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

20/01/2021 11:23, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

What is DCF?

Its "Device Config Function" (DCF), please check ice.rst for detail.

Thank you.
As you know, I am interested in the privileged rights. This is what I found in ice.rst:

Quoted 6 lines

A DCF PMD bounds to the device's trusted VF with ID 0 [...]
#. Enable the VF0 trust on::
ip link set dev enp24s0f0 vf 0 trust on

Does it mean only VF 0 can be trusted?

Yes, currently we only allow VF0 have DCF capability, but that limitation might be removed at some time.

Quoted 1 line
#90 lore raw

Re: [dpdk-dev] [dpdk-dev v5] net/ice: enable eCPRI tunnel port configure in dcf

20/01/2021 11:36, Zhang, Qi Z:

Quoted 12 lines

From: Thomas Monjalon <[email protected]>

Quoted 11 lines

20/01/2021 11:23, Zhang, Qi Z:

Quoted 5 lines

From: Thomas Monjalon <[email protected]>

Quoted 3 lines

What is DCF?

Its "Device Config Function" (DCF), please check ice.rst for detail.

Thank you.
As you know, I am interested in the privileged rights. This is what I found in ice.rst:

Quoted 8 lines
Quoted 6 lines

A DCF PMD bounds to the device's trusted VF with ID 0 [...]
#. Enable the VF0 trust on::
ip link set dev enp24s0f0 vf 0 trust on

Does it mean only VF 0 can be trusted?

Yes, currently we only allow VF0 have DCF capability, but that limitation might be removed at some time.

OK thanks for the explanations.