Re: [dpdk-dev] [PATCH v2 2/5] ethdev: introduce new tunnel VXLAN-GPE

[Adrien Mazarguil <adrien.mazarguil@6wind.com>]

On Tue, Apr 10, 2018 at 09:00:33PM +0800, Xueming Li wrote:
> VXLAN-GPE enables VXLAN for all protocols. Protocol link:
> https://datatracker.ietf.org/doc/draft-ietf-nvo3-vxlan-gpe/
> 
> Signed-off-by: Xueming Li <xuemingl@mellanox.com>

Adding a new rte_flow pattern item in the middle of enum rte_flow_item_type
breaks ABI compatibility. It's fine for 18.05 because prior series already
destroyed it, however for this patch you need to choose between:

- Adding the new entry at the end of the enum and modifying the rest of the
  code to follow the same order (preferred approach when not doing a full
  API overhaul).

*or*

- Stating in the commit log what functions are impacted by ABI changes as in
  "ethdev: remove DUP action from flow API" [1].

Also you must add a new "Item: ``VXLAN_GPE``" section to
doc/guides/prog_guide/rte_flow.rst (look for "VXLAN" for clues).

Otherwise patch is mostly fine, just a few comments below.

[1] http://dpdk.org/ml/archives/dev/2018-April/096526.html

> ---
>  lib/librte_ether/rte_eth_ctrl.h  |  3 ++-
>  lib/librte_ether/rte_flow.c      |  1 +
>  lib/librte_ether/rte_flow.h      | 27 +++++++++++++++++++++++++++
>  lib/librte_mbuf/rte_mbuf.c       |  3 +++
>  lib/librte_mbuf/rte_mbuf.h       |  1 +
>  lib/librte_mbuf/rte_mbuf_ptype.c |  1 +
>  lib/librte_mbuf/rte_mbuf_ptype.h | 13 +++++++++++++
>  lib/librte_net/rte_ether.h       | 25 +++++++++++++++++++++++++
>  8 files changed, 73 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/librte_ether/rte_eth_ctrl.h b/lib/librte_ether/rte_eth_ctrl.h
> index 668f59acb..5ea8ae24c 100644
> --- a/lib/librte_ether/rte_eth_ctrl.h
> +++ b/lib/librte_ether/rte_eth_ctrl.h
> @@ -54,7 +54,8 @@ extern "C" {
>  #define RTE_ETH_FLOW_VXLAN              19 /**< VXLAN protocol based flow */
>  #define RTE_ETH_FLOW_GENEVE             20 /**< GENEVE protocol based flow */
>  #define RTE_ETH_FLOW_NVGRE              21 /**< NVGRE protocol based flow */
> -#define RTE_ETH_FLOW_MAX                22
> +#define RTE_ETH_FLOW_VXLAN_GPE          22 /**< VXLAN-GPE protocol based flow */
> +#define RTE_ETH_FLOW_MAX                23
>  
>  /**
>   * Feature filter types
> diff --git a/lib/librte_ether/rte_flow.c b/lib/librte_ether/rte_flow.c
> index 3d8116ebd..fb710fff7 100644
> --- a/lib/librte_ether/rte_flow.c
> +++ b/lib/librte_ether/rte_flow.c
> @@ -50,6 +50,7 @@ static const struct rte_flow_desc_data rte_flow_desc_item[] = {
>  	MK_FLOW_ITEM(TCP, sizeof(struct rte_flow_item_tcp)),
>  	MK_FLOW_ITEM(SCTP, sizeof(struct rte_flow_item_sctp)),
>  	MK_FLOW_ITEM(VXLAN, sizeof(struct rte_flow_item_vxlan)),
> +	MK_FLOW_ITEM(VXLAN_GPE, sizeof(struct rte_flow_item_vxlan_gpe)),

Should be at the end of this array if you choose to not impact ABI.

>  	MK_FLOW_ITEM(MPLS, sizeof(struct rte_flow_item_mpls)),
>  	MK_FLOW_ITEM(GRE, sizeof(struct rte_flow_item_gre)),
>  	MK_FLOW_ITEM(E_TAG, sizeof(struct rte_flow_item_e_tag)),
> diff --git a/lib/librte_ether/rte_flow.h b/lib/librte_ether/rte_flow.h
> index bed727df8..c7cfc201a 100644
> --- a/lib/librte_ether/rte_flow.h
> +++ b/lib/librte_ether/rte_flow.h
> @@ -256,6 +256,13 @@ enum rte_flow_item_type {
>  	RTE_FLOW_ITEM_TYPE_VXLAN,
>  
>  	/**
> +	 * Matches a VXLAN-GPE header.
> +	 *
> +	 * See struct rte_flow_item_vxlan_gpe.
> +	 */
> +	RTE_FLOW_ITEM_TYPE_VXLAN_GPE,
> +
> +	/**

Ditto for the enum definition.

>  	 * Matches a E_TAG header.
>  	 *
>  	 * See struct rte_flow_item_e_tag.
> @@ -676,6 +683,26 @@ static const struct rte_flow_item_vxlan rte_flow_item_vxlan_mask = {
>  #endif
>  
>  /**
> + * RTE_FLOW_ITEM_TYPE_VXLAN_GPE.
> + *
> + * Matches a VXLAN-GPE header.

You should name the current IETF draft pending a proper RFC:

 Matches a VXLAN-GPE header (draft-ietf-nvo3-vxlan-gpe-05).

> + */
> +struct rte_flow_item_vxlan_gpe {
> +	uint8_t flags; /**< Normally 0x0c (I and P flag). */
> +	uint8_t rsvd0[2]; /**< Reserved, normally 0x0000. */
> +	uint8_t protocol; /**< Protocol type. */
> +	uint8_t vni[3]; /**< VXLAN identifier. */
> +	uint8_t rsvd1; /**< Reserved, normally 0x00. */
> +};
> +
> +/** Default mask for RTE_FLOW_ITEM_TYPE_VXLAN_GPE. */
> +#ifndef __cplusplus
> +static const struct rte_flow_item_vxlan_gpe rte_flow_item_vxlan_gpe_mask = {
> +	.vni = "\xff\xff\xff",
> +};
> +#endif

Again if you choose to not impact ABI, this should be moved further down,
after the last item definition for consistency.

> +
> +/**
>   * RTE_FLOW_ITEM_TYPE_E_TAG.
>   *
>   * Matches a E-tag header.
> diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.c
> index 091d388d3..dc90379e5 100644
> --- a/lib/librte_mbuf/rte_mbuf.c
> +++ b/lib/librte_mbuf/rte_mbuf.c
> @@ -405,6 +405,7 @@ const char *rte_get_tx_ol_flag_name(uint64_t mask)
>  	case PKT_TX_TUNNEL_IPIP: return "PKT_TX_TUNNEL_IPIP";
>  	case PKT_TX_TUNNEL_GENEVE: return "PKT_TX_TUNNEL_GENEVE";
>  	case PKT_TX_TUNNEL_MPLSINUDP: return "PKT_TX_TUNNEL_MPLSINUDP";
> +	case PKT_TX_TUNNEL_VXLAN_GPE: return "PKT_TX_TUNNEL_VXLAN_GPE";
>  	case PKT_TX_MACSEC: return "PKT_TX_MACSEC";
>  	case PKT_TX_SEC_OFFLOAD: return "PKT_TX_SEC_OFFLOAD";
>  	default: return NULL;
> @@ -439,6 +440,8 @@ rte_get_tx_ol_flag_list(uint64_t mask, char *buf, size_t buflen)
>  		  "PKT_TX_TUNNEL_NONE" },
>  		{ PKT_TX_TUNNEL_MPLSINUDP, PKT_TX_TUNNEL_MASK,
>  		  "PKT_TX_TUNNEL_NONE" },
> +		{ PKT_TX_TUNNEL_VXLAN_GPE, PKT_TX_TUNNEL_MASK,
> +		  "PKT_TX_TUNNEL_NONE" },
>  		{ PKT_TX_MACSEC, PKT_TX_MACSEC, NULL },
>  		{ PKT_TX_SEC_OFFLOAD, PKT_TX_SEC_OFFLOAD, NULL },
>  	};
> diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> index 62740254d..1839cf2ed 100644
> --- a/lib/librte_mbuf/rte_mbuf.h
> +++ b/lib/librte_mbuf/rte_mbuf.h
> @@ -210,6 +210,7 @@ extern "C" {
>  #define PKT_TX_TUNNEL_GENEVE  (0x4ULL << 45)
>  /**< TX packet with MPLS-in-UDP RFC 7510 header. */
>  #define PKT_TX_TUNNEL_MPLSINUDP (0x5ULL << 45)
> +#define PKT_TX_TUNNEL_VXLAN_GPE (0x6ULL << 45)
>  /* add new TX TUNNEL type here */
>  #define PKT_TX_TUNNEL_MASK    (0xFULL << 45)
>  
> diff --git a/lib/librte_mbuf/rte_mbuf_ptype.c b/lib/librte_mbuf/rte_mbuf_ptype.c
> index 1feefacc6..49106c7df 100644
> --- a/lib/librte_mbuf/rte_mbuf_ptype.c
> +++ b/lib/librte_mbuf/rte_mbuf_ptype.c
> @@ -65,6 +65,7 @@ const char *rte_get_ptype_tunnel_name(uint32_t ptype)
>  	case RTE_PTYPE_TUNNEL_GTPU: return "TUNNEL_GTPU";
>  	case RTE_PTYPE_TUNNEL_ESP: return "TUNNEL_ESP";
>  	case RTE_PTYPE_TUNNEL_L2TP: return "TUNNEL_L2TP";
> +	case RTE_PTYPE_TUNNEL_VXLAN_GPE: return "TUNNEL_VXLAN_GPE";
>  	default: return "TUNNEL_UNKNOWN";
>  	}
>  }
> diff --git a/lib/librte_mbuf/rte_mbuf_ptype.h b/lib/librte_mbuf/rte_mbuf_ptype.h
> index b9a338110..7caf83312 100644
> --- a/lib/librte_mbuf/rte_mbuf_ptype.h
> +++ b/lib/librte_mbuf/rte_mbuf_ptype.h
> @@ -423,6 +423,19 @@ extern "C" {
>   */
>  #define RTE_PTYPE_TUNNEL_L2TP               0x0000a000
>  /**
> + * VXLAN-GPE (VXLAN Generic Protocol Extension) tunneling packet type.
> + *
> + * Packet format:
> + * <'ether type'=0x0800
> + * | 'version'=4, 'protocol'=17
> + * | 'destination port'=4790>
> + * or,
> + * <'ether type'=0x86DD
> + * | 'version'=6, 'next header'=17
> + * | 'destination port'=4790>
> + */
> +#define RTE_PTYPE_TUNNEL_VXLAN_GPE          0x0000b000
> +/**
>   * Mask of tunneling packet types.
>   */
>  #define RTE_PTYPE_TUNNEL_MASK               0x0000f000
> diff --git a/lib/librte_net/rte_ether.h b/lib/librte_net/rte_ether.h
> index a271d1c86..a64814179 100644
> --- a/lib/librte_net/rte_ether.h
> +++ b/lib/librte_net/rte_ether.h
> @@ -311,6 +311,31 @@ struct vxlan_hdr {
>  /**< VXLAN tunnel header length. */
>  
>  /**
> + * VXLAN-GPE protocol header.
> + * Contains the 8-bit flag, 8-bit next-protocol, 24-bit VXLAN Network
> + * Identifier and Reserved fields (16 bits and 8 bits).

Another reference to the current IETF draft here shouldn't hurt.

> + */
> +struct vxlan_gpe_hdr {
> +	uint8_t vx_flags; /**< flag (8). */
> +	uint8_t reserved[2]; /**< Reserved (16). */
> +	uint8_t proto; /**< next-protocol (8). */
> +	uint32_t vx_vni;   /**< VNI (24) + Reserved (8). */
> +} __attribute__((__packed__));
> +
> +/* VXLAN-GPE next protocol types */
> +#define VXLAN_GPE_TYPE_IPv4 1 /**< IPv4 Protocol. */
> +#define VXLAN_GPE_TYPE_IPv6 2 /**< IPv6 Protocol. */
> +#define VXLAN_GPE_TYPE_ETH  3 /**< Ethernet Protocol. */
> +#define VXLAN_GPE_TYPE_NSH  4 /**< NSH Protocol. */
> +#define VXLAN_GPE_TYPE_MPLS 5 /**< MPLS Protocol. */
> +#define VXLAN_GPE_TYPE_GBP  6 /**< GBP Protocol. */
> +#define VXLAN_GPE_TYPE_VBNG 7 /**< vBNG Protocol. */
> +
> +#define ETHER_VXLAN_GPE_HLEN (sizeof(struct udp_hdr) + \
> +			      sizeof(struct vxlan_gpe_hdr))
> +/**< VXLAN-GPE tunnel header length. */
> +
> +/**
>   * Extract VLAN tag information into mbuf
>   *
>   * Software version of VLAN stripping
> -- 
> 2.13.3

-- 
Adrien Mazarguil
6WIND