From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: "Rybalchenko, Kirill" <kirill.rybalchenko@intel.com>,
"dev@dpdk.org" <dev@dpdk.org>
Cc: "Rybalchenko, Kirill" <kirill.rybalchenko@intel.com>,
"Chilikin, Andrey" <andrey.chilikin@intel.com>,
"Xing, Beilei" <beilei.xing@intel.com>,
"Wu, Jingjing" <jingjing.wu@intel.com>
Subject: Re: [dpdk-dev] [PATCH v2 1/4] net/i40e: implement dynamic mapping of sw flow types to hw pctypes
Date: Mon, 4 Sep 2017 16:49:27 +0000 [thread overview]
Message-ID: <2601191342CEEE43887BDE71AB9772584F2463D4@irsmsx105.ger.corp.intel.com> (raw)
In-Reply-To: <1504278166-32769-2-git-send-email-kirill.rybalchenko@intel.com>
Hi Kirill,
> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Kirill Rybalchenko
> Sent: Friday, September 1, 2017 4:03 PM
> To: dev@dpdk.org
> Cc: Rybalchenko, Kirill <kirill.rybalchenko@intel.com>; Chilikin, Andrey <andrey.chilikin@intel.com>; Xing, Beilei <beilei.xing@intel.com>;
> Wu, Jingjing <jingjing.wu@intel.com>
> Subject: [dpdk-dev] [PATCH v2 1/4] net/i40e: implement dynamic mapping of sw flow types to hw pctypes
>
> Implement dynamic mapping of software flow types to hardware pctypes.
> This allows to add new flow types and pctypes for DDP without changing
> API of the driver. The mapping table is located in private
> data area for particular network adapter and can be individually
> modified with set of appropriate functions.
>
> Signed-off-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
> ---
> v2
> Re-arrange patchset to avoid compillation errors.
> Remove usage of statically defined flow types and pctypes.
> ---
> drivers/net/i40e/i40e_ethdev.c | 347 ++++++++++----------------------------
> drivers/net/i40e/i40e_ethdev.h | 16 +-
> drivers/net/i40e/i40e_ethdev_vf.c | 36 ++--
> drivers/net/i40e/i40e_fdir.c | 51 +++---
> drivers/net/i40e/i40e_flow.c | 2 +-
> drivers/net/i40e/i40e_rxtx.c | 57 +++++++
> drivers/net/i40e/i40e_rxtx.h | 1 +
> 7 files changed, 190 insertions(+), 320 deletions(-)
>
> diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
> index 8e0580c..56a96f5 100644
> --- a/drivers/net/i40e/i40e_ethdev.c
> +++ b/drivers/net/i40e/i40e_ethdev.c
> @@ -1062,6 +1062,7 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
> return 0;
> }
> i40e_set_default_ptype_table(dev);
> + i40e_set_default_pctype_table(dev);
> pci_dev = RTE_ETH_DEV_TO_PCI(dev);
> intr_handle = &pci_dev->intr_handle;
>
> @@ -2971,7 +2972,7 @@ i40e_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
> dev_info->hash_key_size = (I40E_PFQF_HKEY_MAX_INDEX + 1) *
> sizeof(uint32_t);
> dev_info->reta_size = pf->hash_lut_size;
> - dev_info->flow_type_rss_offloads = I40E_RSS_OFFLOAD_ALL;
> + dev_info->flow_type_rss_offloads = pf->adapter->flow_types_msk;
>
> dev_info->default_rxconf = (struct rte_eth_rxconf) {
> .rx_thresh = {
> @@ -6562,104 +6563,36 @@ i40e_vsi_delete_mac(struct i40e_vsi *vsi, struct ether_addr *addr)
>
> /* Configure hash enable flags for RSS */
> uint64_t
> -i40e_config_hena(uint64_t flags, enum i40e_mac_type type)
> +i40e_config_hena(uint64_t flags, struct i40e_adapter *adapter)
> {
As a nit here and in few other functions below, to keep more conventional order of parameters:
i40e_config_hena(struct i40e_adapter *adapter, ....)
probably even better: 'const struct i40e_adapter *adapter' whenever possible.
> uint64_t hena = 0;
> + int i;
>
> if (!flags)
> return hena;
>
> - if (flags & ETH_RSS_FRAG_IPV4)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_FRAG_IPV4;
> - if (flags & ETH_RSS_NONFRAG_IPV4_TCP) {
> - if (type == I40E_MAC_X722) {
> - hena |= (1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_TCP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_TCP_SYN_NO_ACK);
> - } else
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_TCP;
> - }
> - if (flags & ETH_RSS_NONFRAG_IPV4_UDP) {
> - if (type == I40E_MAC_X722) {
> - hena |= (1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_UDP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_UNICAST_IPV4_UDP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV4_UDP);
> - } else
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_UDP;
> - }
> - if (flags & ETH_RSS_NONFRAG_IPV4_SCTP)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_SCTP;
> - if (flags & ETH_RSS_NONFRAG_IPV4_OTHER)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV4_OTHER;
> - if (flags & ETH_RSS_FRAG_IPV6)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_FRAG_IPV6;
> - if (flags & ETH_RSS_NONFRAG_IPV6_TCP) {
> - if (type == I40E_MAC_X722) {
> - hena |= (1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_TCP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_TCP_SYN_NO_ACK);
> - } else
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_TCP;
> + for (i = 0; i < I40E_FLOW_TYPE_MAX; i++) {
> + if (flags & (1ULL << i))
> + hena |= adapter->pcypes_tbl[i];
> }
> - if (flags & ETH_RSS_NONFRAG_IPV6_UDP) {
> - if (type == I40E_MAC_X722) {
> - hena |= (1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_UDP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_UNICAST_IPV6_UDP) |
> - (1ULL << I40E_FILTER_PCTYPE_NONF_MULTICAST_IPV6_UDP);
> - } else
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_UDP;
> - }
> - if (flags & ETH_RSS_NONFRAG_IPV6_SCTP)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_SCTP;
> - if (flags & ETH_RSS_NONFRAG_IPV6_OTHER)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_NONF_IPV6_OTHER;
> - if (flags & ETH_RSS_L2_PAYLOAD)
> - hena |= 1ULL << I40E_FILTER_PCTYPE_L2_PAYLOAD;
>
> return hena;
> }
>
...
>
> -enum i40e_filter_pctype
> -i40e_flowtype_to_pctype(uint16_t flow_type)
> -{
> - static const enum i40e_filter_pctype pctype_table[] = {
> - [RTE_ETH_FLOW_FRAG_IPV4] = I40E_FILTER_PCTYPE_FRAG_IPV4,
> - [RTE_ETH_FLOW_NONFRAG_IPV4_UDP] =
> - I40E_FILTER_PCTYPE_NONF_IPV4_UDP,
> - [RTE_ETH_FLOW_NONFRAG_IPV4_TCP] =
> - I40E_FILTER_PCTYPE_NONF_IPV4_TCP,
> - [RTE_ETH_FLOW_NONFRAG_IPV4_SCTP] =
> - I40E_FILTER_PCTYPE_NONF_IPV4_SCTP,
> - [RTE_ETH_FLOW_NONFRAG_IPV4_OTHER] =
> - I40E_FILTER_PCTYPE_NONF_IPV4_OTHER,
> - [RTE_ETH_FLOW_FRAG_IPV6] = I40E_FILTER_PCTYPE_FRAG_IPV6,
> - [RTE_ETH_FLOW_NONFRAG_IPV6_UDP] =
> - I40E_FILTER_PCTYPE_NONF_IPV6_UDP,
> - [RTE_ETH_FLOW_NONFRAG_IPV6_TCP] =
> - I40E_FILTER_PCTYPE_NONF_IPV6_TCP,
> - [RTE_ETH_FLOW_NONFRAG_IPV6_SCTP] =
> - I40E_FILTER_PCTYPE_NONF_IPV6_SCTP,
> - [RTE_ETH_FLOW_NONFRAG_IPV6_OTHER] =
> - I40E_FILTER_PCTYPE_NONF_IPV6_OTHER,
> - [RTE_ETH_FLOW_L2_PAYLOAD] = I40E_FILTER_PCTYPE_L2_PAYLOAD,
> - };
> +uint16_t
> +i40e_flowtype_to_pctype(uint16_t flow_type, struct i40e_adapter *adapter)
> +{
> + int i;
> + uint64_t pctype_mask;
>
> - return pctype_table[flow_type];
> + if (flow_type < I40E_FLOW_TYPE_MAX) {
> + pctype_mask = adapter->pcypes_tbl[flow_type];
> + for (i = I40E_PCTYPE_MAX - 1; i >= 0; i--) {
> + if (pctype_mask & (1ULL << i))
> + return (uint16_t)i;
So, each pcypes_tbl[] would always have only one bit set?
If so, wouldn't it be more convenient to store only bit index?
> + }
> + }
> + return 0;
As I can see, that function would return 0 for both
adapter->pcypes_tbl[flow_type] == 0 and adapter->pcypes_tbl[flow_type] == 1
Is that intended?
Konstantin
> }
>
next prev parent reply other threads:[~2017-09-04 16:49 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <1503569908-104074-1-git-send-email-kirill.rybalchenko@intel.com>
2017-09-01 15:02 ` [dpdk-dev] [PATCH v2 0/4] net/i40e: implement dynamic mapping of flow types to pctypes Kirill Rybalchenko
2017-09-01 15:02 ` [dpdk-dev] [PATCH v2 1/4] net/i40e: implement dynamic mapping of sw flow types to hw pctypes Kirill Rybalchenko
2017-09-04 16:49 ` Ananyev, Konstantin [this message]
2017-09-08 16:58 ` Ferruh Yigit
2017-09-01 15:02 ` [dpdk-dev] [PATCH v2 2/4] net/i40e: add new functions to manipulate with pctype mapping table Kirill Rybalchenko
2017-09-04 17:24 ` Iremonger, Bernard
2017-09-08 17:01 ` Ferruh Yigit
2017-09-01 15:02 ` [dpdk-dev] [PATCH v2 3/4] app/testpmd: add new commands to manipulate with pctype mapping Kirill Rybalchenko
2017-09-04 17:29 ` Iremonger, Bernard
2017-09-08 17:02 ` Ferruh Yigit
2017-09-01 15:02 ` [dpdk-dev] [PATCH v2 4/4] ethdev: remove unnecessary check for new flow type Kirill Rybalchenko
2017-09-08 17:01 ` Ferruh Yigit
2017-09-04 17:16 ` [dpdk-dev] [PATCH v2 0/4] net/i40e: implement dynamic mapping of flow types to pctypes Iremonger, Bernard
2017-09-20 14:32 ` [dpdk-dev] [PATCH v3 0/6] " Kirill Rybalchenko
2017-09-20 14:32 ` [dpdk-dev] [PATCH v3 1/6] net/i40e: remove unnecessary bit operations Kirill Rybalchenko
2017-09-20 14:32 ` [dpdk-dev] [PATCH v3 2/6] net/i40e: add definition for invalid pctype Kirill Rybalchenko
2017-09-22 7:29 ` Xing, Beilei
2017-09-20 14:33 ` [dpdk-dev] [PATCH v3 3/6] net/i40e: implement dynamic mapping of sw flow types to hw pctypes Kirill Rybalchenko
2017-09-25 9:44 ` Xing, Beilei
2017-09-20 14:33 ` [dpdk-dev] [PATCH v3 4/6] net/i40e: add new functions to manipulate with pctype mapping table Kirill Rybalchenko
2017-09-22 7:27 ` Xing, Beilei
2017-09-20 14:33 ` [dpdk-dev] [PATCH v3 5/6] app/testpmd: add new commands to manipulate with pctype mapping Kirill Rybalchenko
2017-09-25 11:54 ` Xing, Beilei
2017-09-20 14:33 ` [dpdk-dev] [PATCH v3 6/6] ethdev: remove unnecessary check for new flow type Kirill Rybalchenko
2017-10-02 15:08 ` [dpdk-dev] [PATCH v4 0/5] net/i40e: implement dynamic mapping of flow types to pctypes Kirill Rybalchenko
2017-10-02 15:08 ` [dpdk-dev] [PATCH v4 1/5] net/i40e: remove unnecessary bit operations Kirill Rybalchenko
2017-10-02 15:08 ` [dpdk-dev] [PATCH v4 2/5] net/i40e: implement dynamic mapping of sw flow types to hw pctypes Kirill Rybalchenko
2017-10-02 15:09 ` [dpdk-dev] [PATCH v4 3/5] net/i40e: add new functions to manipulate with pctype mapping table Kirill Rybalchenko
2017-10-02 15:09 ` [dpdk-dev] [PATCH v4 4/5] app/testpmd: add new commands to manipulate with pctype mapping Kirill Rybalchenko
2017-10-03 21:42 ` Ferruh Yigit
2017-10-02 15:09 ` [dpdk-dev] [PATCH v4 5/5] ethdev: remove unnecessary check for new flow type Kirill Rybalchenko
2017-10-03 21:44 ` [dpdk-dev] [PATCH v4 0/5] net/i40e: implement dynamic mapping of flow types to pctypes Ferruh Yigit
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 " Kirill Rybalchenko
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 1/5] net/i40e: remove unnecessary bit operations Kirill Rybalchenko
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 2/5] net/i40e: implement dynamic mapping of sw flow types to hw pctypes Kirill Rybalchenko
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 3/5] net/i40e: add new functions to manipulate with pctype mapping table Kirill Rybalchenko
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 4/5] app/testpmd: add new commands to manipulate with pctype mapping Kirill Rybalchenko
2017-10-04 12:52 ` [dpdk-dev] [PATCH v5 5/5] ethdev: remove unnecessary check for new flow type Kirill Rybalchenko
2017-10-04 21:48 ` [dpdk-dev] [PATCH v5 0/5] net/i40e: implement dynamic mapping of flow types to pctypes Ferruh Yigit
2017-10-05 1:28 ` Ferruh Yigit
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2601191342CEEE43887BDE71AB9772584F2463D4@irsmsx105.ger.corp.intel.com \
--to=konstantin.ananyev@intel.com \
--cc=andrey.chilikin@intel.com \
--cc=beilei.xing@intel.com \
--cc=dev@dpdk.org \
--cc=jingjing.wu@intel.com \
--cc=kirill.rybalchenko@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).