* [PATCH 1/3] lib: dpdk spec to skip red for ingress policer
@ 2022-12-22 1:39 Rakesh Kudurumalla
2022-12-22 1:39 ` [PATCH 2/3] app/testpmd: add skip red for testpmd Rakesh Kudurumalla
` (3 more replies)
0 siblings, 4 replies; 70+ messages in thread
From: Rakesh Kudurumalla @ 2022-12-22 1:39 UTC (permalink / raw)
To: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko
Cc: dev, adrien.mazarguil, Rakesh Kudurumalla
Dropping of packets based on RED can be skipped
with meter action, when RED is configured using
rte_eth_cman_config_set()
Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com>
---
lib/ethdev/rte_flow.h | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h
index b60987db4b..d72374a0b8 100644
--- a/lib/ethdev/rte_flow.h
+++ b/lib/ethdev/rte_flow.h
@@ -2203,6 +2203,16 @@ enum rte_flow_action_type {
*/
RTE_FLOW_ACTION_TYPE_DROP,
+ /**
+ * Skip packet drops base on RED
+ *
+ * Dropping of packets is based on RED when RED is configured
+ * uing rte_eth_cman_config_set(). RED drop can be skipped
+ * with this action
+ *
+ */
+ RTE_FLOW_ACTION_TYPE_SKIP_RED,
+
/**
* Enables counters for this flow rule.
*
--
2.25.1
^ permalink raw reply related [flat|nested] 70+ messages in thread* [PATCH 2/3] app/testpmd: add skip red for testpmd 2022-12-22 1:39 [PATCH 1/3] lib: dpdk spec to skip red for ingress policer Rakesh Kudurumalla @ 2022-12-22 1:39 ` Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2022-12-22 1:39 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: dev, adrien.mazarguil, Rakesh Kudurumalla added support for testpmd application to accept skip_red action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..a1094bf00f 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_RED, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_RED, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_RED] = { + .name = "skip_red", + .help = "disable dropping of packets based on RED", + .priv = PRIV_ACTION(SKIP_RED, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH 3/3] net/cnxk: skip red drop for ingress policer 2022-12-22 1:39 [PATCH 1/3] lib: dpdk spec to skip red for ingress policer Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 2/3] app/testpmd: add skip red for testpmd Rakesh Kudurumalla @ 2022-12-22 1:39 ` Rakesh Kudurumalla 2022-12-22 3:01 ` [PATCH 1/3] lib: dpdk spec to skip red " Stephen Hemminger 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2022-12-22 1:39 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: dev, adrien.mazarguil, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ 2 files changed, 51 insertions(+) diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index 651ef45ea8..4fe4aa320b 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -222,6 +222,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index 0fa18f01c7..ff8fa35428 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -359,6 +359,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_RED) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -398,6 +401,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_RED) + fmp->actions[i].skip_red = true; } } } @@ -1307,6 +1314,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1484,6 +1530,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2022-12-22 1:39 [PATCH 1/3] lib: dpdk spec to skip red for ingress policer Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 2/3] app/testpmd: add skip red for testpmd Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2022-12-22 3:01 ` Stephen Hemminger 2022-12-22 5:27 ` Jerin Jacob 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla 3 siblings, 1 reply; 70+ messages in thread From: Stephen Hemminger @ 2022-12-22 3:01 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko, dev, adrien.mazarguil On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > Dropping of packets based on RED can be skipped > with meter action, when RED is configured using > rte_eth_cman_config_set() > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Should this be more general and apply to all congestion management options. Assuming the hardware can do something better than RED. ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2022-12-22 3:01 ` [PATCH 1/3] lib: dpdk spec to skip red " Stephen Hemminger @ 2022-12-22 5:27 ` Jerin Jacob 2022-12-26 17:00 ` Ori Kam 0 siblings, 1 reply; 70+ messages in thread From: Jerin Jacob @ 2022-12-22 5:27 UTC (permalink / raw) To: Stephen Hemminger Cc: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko, dev, adrien.mazarguil On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger <stephen@networkplumber.org> wrote: > > On Thu, 22 Dec 2022 07:09:02 +0530 > Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > > Dropping of packets based on RED can be skipped > > with meter action, when RED is configured using > > rte_eth_cman_config_set() > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > Should this be more general and apply to all congestion > management options. Assuming the hardware can do something > better than RED. Yes. We can use "enum rte_cman_mode mode" in the descriptor to future-proof. ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2022-12-22 5:27 ` Jerin Jacob @ 2022-12-26 17:00 ` Ori Kam 2023-01-10 6:42 ` Rakesh Kudurumalla 0 siblings, 1 reply; 70+ messages in thread From: Ori Kam @ 2022-12-26 17:00 UTC (permalink / raw) To: Jerin Jacob, Stephen Hemminger Cc: Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) Hi All, > -----Original Message----- > From: Jerin Jacob <jerinjacobk@gmail.com> > Sent: Thursday, 22 December 2022 7:27 > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > <stephen@networkplumber.org> wrote: > > > > On Thu, 22 Dec 2022 07:09:02 +0530 > > Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > > > > Dropping of packets based on RED can be skipped > > > with meter action, when RED is configured using > > > rte_eth_cman_config_set() > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > Should this be more general and apply to all congestion > > management options. Assuming the hardware can do something > > better than RED. > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to future- > proof. I'm missing the idea of this new action, I understand that is related to Jerin congestion patches. But I fail to see why we need it? Is it to mark some metadata that will have some effect on the congestion result? (I assume the system is implemented in the HW) ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2022-12-26 17:00 ` Ori Kam @ 2023-01-10 6:42 ` Rakesh Kudurumalla 2023-01-18 8:09 ` Rakesh Kudurumalla 0 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-01-10 6:42 UTC (permalink / raw) To: Ori Kam, Jerin Jacob, Stephen Hemminger Cc: NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) > -----Original Message----- > From: Ori Kam <orika@nvidia.com> > Sent: Monday, December 26, 2022 10:30 PM > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > <stephen@networkplumber.org> > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > <ferruh.yigit@amd.com>; Andrew Rybchenko > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > External Email > > ---------------------------------------------------------------------- > Hi All, > > > -----Original Message----- > > From: Jerin Jacob <jerinjacobk@gmail.com> > > Sent: Thursday, 22 December 2022 7:27 > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > <stephen@networkplumber.org> wrote: > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 > > > Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > > > > > > Dropping of packets based on RED can be skipped with meter action, > > > > when RED is configured using > > > > rte_eth_cman_config_set() > > > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > Should this be more general and apply to all congestion management > > > options. Assuming the hardware can do something better than RED. > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to future- > > proof. > > I'm missing the idea of this new action, I understand that is related to Jerin > congestion patches. > But I fail to see why we need it? Is it to mark some metadata that will have > some effect on the congestion result? (I assume the system is implemented > in the HW) Yes. It is implemented in HW. Congestion management is applied on ethdev Rx queue using rte_eth_cman_config() API. Once it is configured, it applies to all the packets that steer towards that particular ethdev Rx queue. This feature help to skip the congestion management processing based on the packet color identified by the rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants to bypass the RED congestion processing for all GREEN color packet can be expressed though this API proposal. ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-01-10 6:42 ` Rakesh Kudurumalla @ 2023-01-18 8:09 ` Rakesh Kudurumalla 2023-01-23 13:13 ` Jerin Jacob 2023-01-26 15:13 ` Ori Kam 0 siblings, 2 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-01-18 8:09 UTC (permalink / raw) To: Ori Kam, Jerin Jacob, Stephen Hemminger Cc: NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) > -----Original Message----- > From: Rakesh Kudurumalla > Sent: Tuesday, January 10, 2023 12:12 PM > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > Stephen Hemminger <stephen@networkplumber.org> > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > -----Original Message----- > > From: Ori Kam <orika@nvidia.com> > > Sent: Monday, December 26, 2022 10:30 PM > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > <stephen@networkplumber.org> > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > Thomas > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > policer > > > > External Email > > > > ---------------------------------------------------------------------- > > Hi All, > > > > > -----Original Message----- > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > <stephen@networkplumber.org> wrote: > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > action, when RED is configured using > > > > > rte_eth_cman_config_set() > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > > > Should this be more general and apply to all congestion management > > > > options. Assuming the hardware can do something better than RED. > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > future- proof. > > > > I'm missing the idea of this new action, I understand that is related > > to Jerin congestion patches. > > But I fail to see why we need it? Is it to mark some metadata that > > will have some effect on the congestion result? (I assume the system > > is implemented in the HW) > > Yes. It is implemented in HW. Congestion management is applied on ethdev > Rx queue using rte_eth_cman_config() API. Once it is configured, it applies to > all the packets that steer towards that particular ethdev Rx queue. This > feature help to skip the congestion management processing based on the > packet color identified by the rte_flow meter object. For example, If one Rx > queue configured as RED congestion and application wants to bypass the > RED congestion processing for all GREEN color packet can be expressed > though this API proposal. Hi Ori Kam, Let me know if above information would give clear idea on skip RED action ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-01-18 8:09 ` Rakesh Kudurumalla @ 2023-01-23 13:13 ` Jerin Jacob 2023-01-26 15:13 ` Ori Kam 1 sibling, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-01-23 13:13 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Ori Kam, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) On Wed, Jan 18, 2023 at 1:39 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > > > > -----Original Message----- > > From: Rakesh Kudurumalla > > Sent: Tuesday, January 10, 2023 12:12 PM > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > > Stephen Hemminger <stephen@networkplumber.org> > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > -----Original Message----- > > > From: Ori Kam <orika@nvidia.com> > > > Sent: Monday, December 26, 2022 10:30 PM > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > <stephen@networkplumber.org> > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > > Thomas > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > policer > > > > > > External Email > > > > > > ---------------------------------------------------------------------- > > > Hi All, > > > > > > > -----Original Message----- > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > > action, when RED is configured using > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > > > > > Should this be more general and apply to all congestion management > > > > > options. Assuming the hardware can do something better than RED. > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > > future- proof. > > > > > > I'm missing the idea of this new action, I understand that is related > > > to Jerin congestion patches. > > > But I fail to see why we need it? Is it to mark some metadata that > > > will have some effect on the congestion result? (I assume the system > > > is implemented in the HW) > > > > Yes. It is implemented in HW. Congestion management is applied on ethdev > > Rx queue using rte_eth_cman_config() API. Once it is configured, it applies to > > all the packets that steer towards that particular ethdev Rx queue. This > > feature help to skip the congestion management processing based on the > > packet color identified by the rte_flow meter object. For example, If one Rx > > queue configured as RED congestion and application wants to bypass the > > RED congestion processing for all GREEN color packet can be expressed > > though this API proposal. > > Hi Ori Kam, > > Let me know if above information would give clear idea on skip RED action Rakesh, Adding extra rewording to make it clear. git comment: ethdev: introduce flow action to skip congestion management configuration Using rte_eth_cman_config_set() API the application can configure ethdev Rx queue's congestion mechanism. Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the congestion configuration applied to the given ethdev Rx queue. This feature helps to skip the congestion management processing based on per flow or the packet color identified by the rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants to bypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action. code: /** * Skip congestion management configuration * * Skip the congestion management configuration applied using rte_eth_cman_config_set() * API on the given ethdev Rx queue. */ RTE_FLOW_ACTION_TYPE_SKIP_CMAN ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-01-18 8:09 ` Rakesh Kudurumalla 2023-01-23 13:13 ` Jerin Jacob @ 2023-01-26 15:13 ` Ori Kam 2023-01-27 6:23 ` Jerin Jacob 1 sibling, 1 reply; 70+ messages in thread From: Ori Kam @ 2023-01-26 15:13 UTC (permalink / raw) To: Rakesh Kudurumalla, Jerin Jacob, Stephen Hemminger Cc: NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) > -----Original Message----- > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Sent: Wednesday, 18 January 2023 10:10 > > > > -----Original Message----- > > From: Rakesh Kudurumalla > > Sent: Tuesday, January 10, 2023 12:12 PM > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > > Stephen Hemminger <stephen@networkplumber.org> > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > -----Original Message----- > > > From: Ori Kam <orika@nvidia.com> > > > Sent: Monday, December 26, 2022 10:30 PM > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > <stephen@networkplumber.org> > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > > Thomas > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > Adrien > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > policer > > > > > > External Email > > > > > > ---------------------------------------------------------------------- > > > Hi All, > > > > > > > -----Original Message----- > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > > action, when RED is configured using > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > > > > > Should this be more general and apply to all congestion management > > > > > options. Assuming the hardware can do something better than RED. > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > > future- proof. > > > > > > I'm missing the idea of this new action, I understand that is related > > > to Jerin congestion patches. > > > But I fail to see why we need it? Is it to mark some metadata that > > > will have some effect on the congestion result? (I assume the system > > > is implemented in the HW) > > > > Yes. It is implemented in HW. Congestion management is applied on > ethdev > > Rx queue using rte_eth_cman_config() API. Once it is configured, it applies > to > > all the packets that steer towards that particular ethdev Rx queue. This > > feature help to skip the congestion management processing based on the > > packet color identified by the rte_flow meter object. For example, If one > Rx > > queue configured as RED congestion and application wants to bypass the > > RED congestion processing for all GREEN color packet can be expressed > > though this API proposal. > > Hi Ori Kam, > > Let me know if above information would give clear idea on skip RED action I think so, to put it in my own words, when setting this the selected packet is treated as green packet? If so, can we use the meter_color field? If you want the packet to be green just set the field to green? Best, Ori ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-01-26 15:13 ` Ori Kam @ 2023-01-27 6:23 ` Jerin Jacob 2023-02-01 17:49 ` Ori Kam 0 siblings, 1 reply; 70+ messages in thread From: Jerin Jacob @ 2023-01-27 6:23 UTC (permalink / raw) To: Ori Kam Cc: Rakesh Kudurumalla, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) On Thu, Jan 26, 2023 at 8:43 PM Ori Kam <orika@nvidia.com> wrote: > > > > > -----Original Message----- > > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > Sent: Wednesday, 18 January 2023 10:10 > > > > > > > -----Original Message----- > > > From: Rakesh Kudurumalla > > > Sent: Tuesday, January 10, 2023 12:12 PM > > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > > > Stephen Hemminger <stephen@networkplumber.org> > > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) <thomas@monjalon.net>; > > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact-Adrien > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > > > > > -----Original Message----- > > > > From: Ori Kam <orika@nvidia.com> > > > > Sent: Monday, December 26, 2022 10:30 PM > > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > > <stephen@networkplumber.org> > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > > > Thomas > > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > Adrien > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > > policer > > > > > > > > External Email > > > > > > > > ---------------------------------------------------------------------- > > > > Hi All, > > > > > > > > > -----Original Message----- > > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > > > action, when RED is configured using > > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > > > > > > > Should this be more general and apply to all congestion management > > > > > > options. Assuming the hardware can do something better than RED. > > > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > > > future- proof. > > > > > > > > I'm missing the idea of this new action, I understand that is related > > > > to Jerin congestion patches. > > > > But I fail to see why we need it? Is it to mark some metadata that > > > > will have some effect on the congestion result? (I assume the system > > > > is implemented in the HW) > > > > > > Yes. It is implemented in HW. Congestion management is applied on > > ethdev > > > Rx queue using rte_eth_cman_config() API. Once it is configured, it applies > > to > > > all the packets that steer towards that particular ethdev Rx queue. This > > > feature help to skip the congestion management processing based on the > > > packet color identified by the rte_flow meter object. For example, If one > > Rx > > > queue configured as RED congestion and application wants to bypass the > > > RED congestion processing for all GREEN color packet can be expressed > > > though this API proposal. > > > > Hi Ori Kam, > > > > Let me know if above information would give clear idea on skip RED action > > I think so, to put it in my own words, when setting this the selected packet is treated as > green packet? > > If so, can we use the meter_color field? If you want the packet to be green just set the > field to green? It is already there in one form. See following in existing header file. /** * Meter policy */ struct rte_mtr_meter_policy_params { /** * Policy action list per color. * actions[i] potentially represents a chain of rte_flow actions * terminated by the END action, exactly as specified by the rte_flow * API for the flow definition, and not just a single action. */ const struct rte_flow_action *actions[RTE_COLORS]; }; > > Best, > Ori > ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-01-27 6:23 ` Jerin Jacob @ 2023-02-01 17:49 ` Ori Kam 2023-02-01 18:37 ` Jerin Jacob 0 siblings, 1 reply; 70+ messages in thread From: Ori Kam @ 2023-02-01 17:49 UTC (permalink / raw) To: Jerin Jacob Cc: Rakesh Kudurumalla, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) > -----Original Message----- > From: Jerin Jacob <jerinjacobk@gmail.com> > Sent: Friday, 27 January 2023 8:23 > To: Ori Kam <orika@nvidia.com> > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; Stephen Hemminger > <stephen@networkplumber.org>; NBU-Contact-Thomas Monjalon > (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit <ferruh.yigit@amd.com>; > Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; > NBU-Contact-Adrien Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > Subject: Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > On Thu, Jan 26, 2023 at 8:43 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > -----Original Message----- > > > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > Sent: Wednesday, 18 January 2023 10:10 > > > > > > > > > > -----Original Message----- > > > > From: Rakesh Kudurumalla > > > > Sent: Tuesday, January 10, 2023 12:12 PM > > > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > > > > Stephen Hemminger <stephen@networkplumber.org> > > > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) > <thomas@monjalon.net>; > > > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > Adrien > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > From: Ori Kam <orika@nvidia.com> > > > > > Sent: Monday, December 26, 2022 10:30 PM > > > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > > > <stephen@networkplumber.org> > > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > > > > Thomas > > > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > > Adrien > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > > > policer > > > > > > > > > > External Email > > > > > > > > > > ---------------------------------------------------------------------- > > > > > Hi All, > > > > > > > > > > > -----Original Message----- > > > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > > > > action, when RED is configured using > > > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla > <rkudurumalla@marvell.com> > > > > > > > > > > > > > > Should this be more general and apply to all congestion > management > > > > > > > options. Assuming the hardware can do something better than > RED. > > > > > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > > > > future- proof. > > > > > > > > > > I'm missing the idea of this new action, I understand that is related > > > > > to Jerin congestion patches. > > > > > But I fail to see why we need it? Is it to mark some metadata that > > > > > will have some effect on the congestion result? (I assume the system > > > > > is implemented in the HW) > > > > > > > > Yes. It is implemented in HW. Congestion management is applied on > > > ethdev > > > > Rx queue using rte_eth_cman_config() API. Once it is configured, it > applies > > > to > > > > all the packets that steer towards that particular ethdev Rx queue. This > > > > feature help to skip the congestion management processing based on > the > > > > packet color identified by the rte_flow meter object. For example, If > one > > > Rx > > > > queue configured as RED congestion and application wants to bypass > the > > > > RED congestion processing for all GREEN color packet can be expressed > > > > though this API proposal. > > > > > > Hi Ori Kam, > > > > > > Let me know if above information would give clear idea on skip RED > action > > > > I think so, to put it in my own words, when setting this the selected packet > is treated as > > green packet? > > > > If so, can we use the meter_color field? If you want the packet to be green > just set the > > field to green? > > It is already there in one form. See following in existing header file. > > /** > * Meter policy > */ > struct rte_mtr_meter_policy_params { > /** > * Policy action list per color. > * actions[i] potentially represents a chain of rte_flow actions > * terminated by the END action, exactly as specified by the rte_flow > * API for the flow definition, and not just a single action. > */ > const struct rte_flow_action *actions[RTE_COLORS]; > }; > > Sorry I'm not sure I understand, I know we can have colors in the meter, but this feature is about somehow telling the rxq to skip the red packet and treat it as green right? > > > > Best, > > Ori > > ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-02-01 17:49 ` Ori Kam @ 2023-02-01 18:37 ` Jerin Jacob 2023-02-02 7:31 ` Ori Kam 0 siblings, 1 reply; 70+ messages in thread From: Jerin Jacob @ 2023-02-01 18:37 UTC (permalink / raw) To: Ori Kam Cc: Rakesh Kudurumalla, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) On Wed, Feb 1, 2023 at 11:19 PM Ori Kam <orika@nvidia.com> wrote: > > > > > -----Original Message----- > > From: Jerin Jacob <jerinjacobk@gmail.com> > > Sent: Friday, 27 January 2023 8:23 > > To: Ori Kam <orika@nvidia.com> > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; Stephen Hemminger > > <stephen@networkplumber.org>; NBU-Contact-Thomas Monjalon > > (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit <ferruh.yigit@amd.com>; > > Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; > > NBU-Contact-Adrien Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > Subject: Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > On Thu, Jan 26, 2023 at 8:43 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > > > > > -----Original Message----- > > > > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > Sent: Wednesday, 18 January 2023 10:10 > > > > > > > > > > > > > -----Original Message----- > > > > > From: Rakesh Kudurumalla > > > > > Sent: Tuesday, January 10, 2023 12:12 PM > > > > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob <jerinjacobk@gmail.com>; > > > > > Stephen Hemminger <stephen@networkplumber.org> > > > > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) > > <thomas@monjalon.net>; > > > > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > Adrien > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > From: Ori Kam <orika@nvidia.com> > > > > > > Sent: Monday, December 26, 2022 10:30 PM > > > > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > > > > <stephen@networkplumber.org> > > > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU-Contact- > > > > > Thomas > > > > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > > > Adrien > > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > > > > policer > > > > > > > > > > > > External Email > > > > > > > > > > > > ---------------------------------------------------------------------- > > > > > > Hi All, > > > > > > > > > > > > > -----Original Message----- > > > > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > > > > > > > Dropping of packets based on RED can be skipped with meter > > > > > > > > > action, when RED is configured using > > > > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla > > <rkudurumalla@marvell.com> > > > > > > > > > > > > > > > > Should this be more general and apply to all congestion > > management > > > > > > > > options. Assuming the hardware can do something better than > > RED. > > > > > > > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the descriptor to > > > > > > > future- proof. > > > > > > > > > > > > I'm missing the idea of this new action, I understand that is related > > > > > > to Jerin congestion patches. > > > > > > But I fail to see why we need it? Is it to mark some metadata that > > > > > > will have some effect on the congestion result? (I assume the system > > > > > > is implemented in the HW) > > > > > > > > > > Yes. It is implemented in HW. Congestion management is applied on > > > > ethdev > > > > > Rx queue using rte_eth_cman_config() API. Once it is configured, it > > applies > > > > to > > > > > all the packets that steer towards that particular ethdev Rx queue. This > > > > > feature help to skip the congestion management processing based on > > the > > > > > packet color identified by the rte_flow meter object. For example, If > > one > > > > Rx > > > > > queue configured as RED congestion and application wants to bypass > > the > > > > > RED congestion processing for all GREEN color packet can be expressed > > > > > though this API proposal. > > > > > > > > Hi Ori Kam, > > > > > > > > Let me know if above information would give clear idea on skip RED > > action > > > > > > I think so, to put it in my own words, when setting this the selected packet > > is treated as > > > green packet? > > > > > > If so, can we use the meter_color field? If you want the packet to be green > > just set the > > > field to green? > > > > It is already there in one form. See following in existing header file. > > > > /** > > * Meter policy > > */ > > struct rte_mtr_meter_policy_params { > > /** > > * Policy action list per color. > > * actions[i] potentially represents a chain of rte_flow actions > > * terminated by the END action, exactly as specified by the rte_flow > > * API for the flow definition, and not just a single action. > > */ > > const struct rte_flow_action *actions[RTE_COLORS]; > > }; > > > > > Sorry I'm not sure I understand, > I know we can have colors in the meter, but this feature is about > somehow telling the rxq to skip the red packet and treat it as green right? Yes. When rte_mtr_meter_policy_params::actions[RTE_COLOR_GREEN] set as RTE_FLOW_ACTION_TYPE_SKIP_CMAN for the given meter object, it is indicating to SKIP the CMAN configuration applied to the rxq on the downstream path if meter assigns a GREEN color. (RQ section is as usual as existing path, either via ethdev RSS or rte_flow RSS action or rte_fow Queue action). > > > > > > > Best, > > > Ori > > > ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-02-01 18:37 ` Jerin Jacob @ 2023-02-02 7:31 ` Ori Kam 2023-02-06 3:31 ` Jerin Jacob 0 siblings, 1 reply; 70+ messages in thread From: Ori Kam @ 2023-02-02 7:31 UTC (permalink / raw) To: Jerin Jacob Cc: Rakesh Kudurumalla, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) > -----Original Message----- > From: Jerin Jacob <jerinjacobk@gmail.com> > Sent: Wednesday, 1 February 2023 20:37 > > On Wed, Feb 1, 2023 at 11:19 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > -----Original Message----- > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > Sent: Friday, 27 January 2023 8:23 > > > To: Ori Kam <orika@nvidia.com> > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; Stephen > Hemminger > > > <stephen@networkplumber.org>; NBU-Contact-Thomas Monjalon > > > (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > <ferruh.yigit@amd.com>; > > > Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; > > > NBU-Contact-Adrien Mazarguil (EXTERNAL) > <adrien.mazarguil@6wind.com> > > > Subject: Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > On Thu, Jan 26, 2023 at 8:43 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > Sent: Wednesday, 18 January 2023 10:10 > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > From: Rakesh Kudurumalla > > > > > > Sent: Tuesday, January 10, 2023 12:12 PM > > > > > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob > <jerinjacobk@gmail.com>; > > > > > > Stephen Hemminger <stephen@networkplumber.org> > > > > > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) > > > <thomas@monjalon.net>; > > > > > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > > Adrien > > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > > From: Ori Kam <orika@nvidia.com> > > > > > > > Sent: Monday, December 26, 2022 10:30 PM > > > > > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > > > > > <stephen@networkplumber.org> > > > > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU- > Contact- > > > > > > Thomas > > > > > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > > > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU- > Contact- > > > > > Adrien > > > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > > > > > policer > > > > > > > > > > > > > > External Email > > > > > > > > > > > > > > ---------------------------------------------------------------------- > > > > > > > Hi All, > > > > > > > > > > > > > > > -----Original Message----- > > > > > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > > > > > > > > > Dropping of packets based on RED can be skipped with > meter > > > > > > > > > > action, when RED is configured using > > > > > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla > > > <rkudurumalla@marvell.com> > > > > > > > > > > > > > > > > > > Should this be more general and apply to all congestion > > > management > > > > > > > > > options. Assuming the hardware can do something better than > > > RED. > > > > > > > > > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the > descriptor to > > > > > > > > future- proof. > > > > > > > > > > > > > > I'm missing the idea of this new action, I understand that is related > > > > > > > to Jerin congestion patches. > > > > > > > But I fail to see why we need it? Is it to mark some metadata that > > > > > > > will have some effect on the congestion result? (I assume the > system > > > > > > > is implemented in the HW) > > > > > > > > > > > > Yes. It is implemented in HW. Congestion management is applied on > > > > > ethdev > > > > > > Rx queue using rte_eth_cman_config() API. Once it is configured, it > > > applies > > > > > to > > > > > > all the packets that steer towards that particular ethdev Rx queue. > This > > > > > > feature help to skip the congestion management processing based > on > > > the > > > > > > packet color identified by the rte_flow meter object. For example, > If > > > one > > > > > Rx > > > > > > queue configured as RED congestion and application wants to > bypass > > > the > > > > > > RED congestion processing for all GREEN color packet can be > expressed > > > > > > though this API proposal. > > > > > > > > > > Hi Ori Kam, > > > > > > > > > > Let me know if above information would give clear idea on skip RED > > > action > > > > > > > > I think so, to put it in my own words, when setting this the selected > packet > > > is treated as > > > > green packet? > > > > > > > > If so, can we use the meter_color field? If you want the packet to be > green > > > just set the > > > > field to green? > > > > > > It is already there in one form. See following in existing header file. > > > > > > /** > > > * Meter policy > > > */ > > > struct rte_mtr_meter_policy_params { > > > /** > > > * Policy action list per color. > > > * actions[i] potentially represents a chain of rte_flow actions > > > * terminated by the END action, exactly as specified by the rte_flow > > > * API for the flow definition, and not just a single action. > > > */ > > > const struct rte_flow_action *actions[RTE_COLORS]; > > > }; > > > > > > > > Sorry I'm not sure I understand, > > I know we can have colors in the meter, but this feature is about > > somehow telling the rxq to skip the red packet and treat it as green right? > > Yes. When rte_mtr_meter_policy_params::actions[RTE_COLOR_GREEN] set > as > RTE_FLOW_ACTION_TYPE_SKIP_CMAN for the given meter object, > it is indicating to SKIP the CMAN configuration applied to the rxq on > the downstream path if meter assigns a GREEN color. > (RQ section is as usual as existing path, either via ethdev RSS or > rte_flow RSS action or rte_fow Queue action). > Sorry for all my questions just trying to find the best solution, Some follow up questions, 1. except the skip_cman what other actions is going to be in the action list? Queue? Jump? 2. why can't we use the meter color as input if the cman should be skipped or not? Thanks, Ori > > > > > > > > > > Best, > > > > Ori > > > > ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer 2023-02-02 7:31 ` Ori Kam @ 2023-02-06 3:31 ` Jerin Jacob 0 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-06 3:31 UTC (permalink / raw) To: Ori Kam Cc: Rakesh Kudurumalla, Stephen Hemminger, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko, dev, NBU-Contact-Adrien Mazarguil (EXTERNAL) On Thu, Feb 2, 2023 at 1:01 PM Ori Kam <orika@nvidia.com> wrote: > > > > > -----Original Message----- > > From: Jerin Jacob <jerinjacobk@gmail.com> > > Sent: Wednesday, 1 February 2023 20:37 > > > > On Wed, Feb 1, 2023 at 11:19 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > > > > > -----Original Message----- > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > Sent: Friday, 27 January 2023 8:23 > > > > To: Ori Kam <orika@nvidia.com> > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; Stephen > > Hemminger > > > > <stephen@networkplumber.org>; NBU-Contact-Thomas Monjalon > > > > (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > <ferruh.yigit@amd.com>; > > > > Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; > > > > NBU-Contact-Adrien Mazarguil (EXTERNAL) > > <adrien.mazarguil@6wind.com> > > > > Subject: Re: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > On Thu, Jan 26, 2023 at 8:43 PM Ori Kam <orika@nvidia.com> wrote: > > > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > > > > > Sent: Wednesday, 18 January 2023 10:10 > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > > From: Rakesh Kudurumalla > > > > > > > Sent: Tuesday, January 10, 2023 12:12 PM > > > > > > > To: Ori Kam <orika@nvidia.com>; Jerin Jacob > > <jerinjacobk@gmail.com>; > > > > > > > Stephen Hemminger <stephen@networkplumber.org> > > > > > > > Cc: NBU-Contact-Thomas Monjalon (EXTERNAL) > > > > <thomas@monjalon.net>; > > > > > > > Ferruh Yigit <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU-Contact- > > > > Adrien > > > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > > > Subject: RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress policer > > > > > > > > > > > > > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > > > From: Ori Kam <orika@nvidia.com> > > > > > > > > Sent: Monday, December 26, 2022 10:30 PM > > > > > > > > To: Jerin Jacob <jerinjacobk@gmail.com>; Stephen Hemminger > > > > > > > > <stephen@networkplumber.org> > > > > > > > > Cc: Rakesh Kudurumalla <rkudurumalla@marvell.com>; NBU- > > Contact- > > > > > > > Thomas > > > > > > > > Monjalon (EXTERNAL) <thomas@monjalon.net>; Ferruh Yigit > > > > > > > > <ferruh.yigit@amd.com>; Andrew Rybchenko > > > > > > > > <andrew.rybchenko@oktetlabs.ru>; dev@dpdk.org; NBU- > > Contact- > > > > > > Adrien > > > > > > > > Mazarguil (EXTERNAL) <adrien.mazarguil@6wind.com> > > > > > > > > Subject: [EXT] RE: [PATCH 1/3] lib: dpdk spec to skip red for ingress > > > > > > > > policer > > > > > > > > > > > > > > > > External Email > > > > > > > > > > > > > > > > ---------------------------------------------------------------------- > > > > > > > > Hi All, > > > > > > > > > > > > > > > > > -----Original Message----- > > > > > > > > > From: Jerin Jacob <jerinjacobk@gmail.com> > > > > > > > > > Sent: Thursday, 22 December 2022 7:27 > > > > > > > > > > > > > > > > > > On Thu, Dec 22, 2022 at 8:32 AM Stephen Hemminger > > > > > > > > > <stephen@networkplumber.org> wrote: > > > > > > > > > > > > > > > > > > > > On Thu, 22 Dec 2022 07:09:02 +0530 Rakesh Kudurumalla > > > > > > > > > > <rkudurumalla@marvell.com> wrote: > > > > > > > > > > > > > > > > > > > > > Dropping of packets based on RED can be skipped with > > meter > > > > > > > > > > > action, when RED is configured using > > > > > > > > > > > rte_eth_cman_config_set() > > > > > > > > > > > > > > > > > > > > > > Signed-off-by: Rakesh Kudurumalla > > > > <rkudurumalla@marvell.com> > > > > > > > > > > > > > > > > > > > > Should this be more general and apply to all congestion > > > > management > > > > > > > > > > options. Assuming the hardware can do something better than > > > > RED. > > > > > > > > > > > > > > > > > > Yes. We can use "enum rte_cman_mode mode" in the > > descriptor to > > > > > > > > > future- proof. > > > > > > > > > > > > > > > > I'm missing the idea of this new action, I understand that is related > > > > > > > > to Jerin congestion patches. > > > > > > > > But I fail to see why we need it? Is it to mark some metadata that > > > > > > > > will have some effect on the congestion result? (I assume the > > system > > > > > > > > is implemented in the HW) > > > > > > > > > > > > > > Yes. It is implemented in HW. Congestion management is applied on > > > > > > ethdev > > > > > > > Rx queue using rte_eth_cman_config() API. Once it is configured, it > > > > applies > > > > > > to > > > > > > > all the packets that steer towards that particular ethdev Rx queue. > > This > > > > > > > feature help to skip the congestion management processing based > > on > > > > the > > > > > > > packet color identified by the rte_flow meter object. For example, > > If > > > > one > > > > > > Rx > > > > > > > queue configured as RED congestion and application wants to > > bypass > > > > the > > > > > > > RED congestion processing for all GREEN color packet can be > > expressed > > > > > > > though this API proposal. > > > > > > > > > > > > Hi Ori Kam, > > > > > > > > > > > > Let me know if above information would give clear idea on skip RED > > > > action > > > > > > > > > > I think so, to put it in my own words, when setting this the selected > > packet > > > > is treated as > > > > > green packet? > > > > > > > > > > If so, can we use the meter_color field? If you want the packet to be > > green > > > > just set the > > > > > field to green? > > > > > > > > It is already there in one form. See following in existing header file. > > > > > > > > /** > > > > * Meter policy > > > > */ > > > > struct rte_mtr_meter_policy_params { > > > > /** > > > > * Policy action list per color. > > > > * actions[i] potentially represents a chain of rte_flow actions > > > > * terminated by the END action, exactly as specified by the rte_flow > > > > * API for the flow definition, and not just a single action. > > > > */ > > > > const struct rte_flow_action *actions[RTE_COLORS]; > > > > }; > > > > > > > > > > > Sorry I'm not sure I understand, > > > I know we can have colors in the meter, but this feature is about > > > somehow telling the rxq to skip the red packet and treat it as green right? > > > > Yes. When rte_mtr_meter_policy_params::actions[RTE_COLOR_GREEN] set > > as > > RTE_FLOW_ACTION_TYPE_SKIP_CMAN for the given meter object, > > it is indicating to SKIP the CMAN configuration applied to the rxq on > > the downstream path if meter assigns a GREEN color. > > (RQ section is as usual as existing path, either via ethdev RSS or > > rte_flow RSS action or rte_fow Queue action). > > > > Sorry for all my questions just trying to find the best solution, > Some follow up questions, > 1. except the skip_cman what other actions is going to be in the action list? Queue? Jump? No specific restriction from speciation POV. Queue, Jump, RSS most useful candidates. > 2. why can't we use the meter color as input if the cman should be skipped or not? Because, cman skip is happening in ethdev Rx and in our case meter object action is the one trigger for this. > > Thanks, > Ori > > > > > > > > > > > > > > Best, > > > > > Ori > > > > > ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v2 1/3] lib: skip congestion management configuration 2022-12-22 1:39 [PATCH 1/3] lib: dpdk spec to skip red for ingress policer Rakesh Kudurumalla ` (2 preceding siblings ...) 2022-12-22 3:01 ` [PATCH 1/3] lib: dpdk spec to skip red " Stephen Hemminger @ 2023-01-25 7:32 ` Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (3 more replies) 3 siblings, 4 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-01-25 7:32 UTC (permalink / raw) To: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Skip the congestion management configuration applied using rte_eth_cman_config_set() API on the given ethdev Rx queue. Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- V2: Rename action name to skip congestion management lib/ethdev/rte_flow.h | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..e62b57acfd 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2203,6 +2203,15 @@ enum rte_flow_action_type { */ RTE_FLOW_ACTION_TYPE_DROP, + /** + * Skip congestion management configuration + * + * Skip the congestion management configuration applied using + * rte_eth_cman_config_set() API on the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, + /** * Enables counters for this flow rule. * -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v2 2/3] app/testpmd: add skip cman support for testpmd 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla @ 2023-01-25 7:32 ` Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-01-25 7:32 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v2 3/3] net/cnxk: skip red drop for ingress policer 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-01-25 7:32 ` Rakesh Kudurumalla 2023-01-25 9:26 ` [PATCH v2 1/3] lib: skip congestion management configuration Jerin Jacob 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-01-25 7:32 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ 2 files changed, 51 insertions(+) diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index ea8c70b8b7..7c53122b9f 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -238,6 +238,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v2 1/3] lib: skip congestion management configuration 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-01-25 9:26 ` Jerin Jacob 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-01-25 9:26 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko, jerinj, ndabilpuram, dev On Wed, Jan 25, 2023 at 1:03 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > Skip the congestion management configuration applied using > rte_eth_cman_config_set() API on the given ethdev Rx queue. 1) Add more comments. Also starts with ethdev: example, ethdev: introduce flow action to skip congestion management configuration Using rte_eth_cman_config_set() API the application can configure ethdev Rx queue's congestion mechanism. Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the congestion configuration applied to the given ethdev Rx queue. This feature helps to skip the congestion management processing based on per flow or the packet color identified by the rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action 2) Update doc/guides/prog_guide/rte_flow.rst for new action and 3/3 patch for doc/guides/nics/features/ pmd specific doc ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v3 1/3] lib: skip congestion management configuration 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-01-25 9:26 ` [PATCH v2 1/3] lib: skip congestion management configuration Jerin Jacob @ 2023-02-09 7:24 ` Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (3 more replies) 3 siblings, 4 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 7:24 UTC (permalink / raw) To: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Skip the congestion management configuration applied using rte_eth_cman_config_set() API on the given ethdev Rx queue. Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- v3: Updated comments and programmers guide regrading new action doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 20 ++++++++++++++++++++ 2 files changed, 42 insertions(+) diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..0737b877da 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..5ae00e2245 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2203,6 +2203,26 @@ enum rte_flow_action_type { */ RTE_FLOW_ACTION_TYPE_DROP, + /** + * Skip congestion management configuration + * + * ethdev: introduce flow action to skip congestion + * management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + * This feature helps to skip the congestion management processing + * based on per flow or the packet color identified by + * rte_flow meter object. For example, If one Rx queue configured as + * RED congestion and application wants tobypass the RED congestion + * processing for all GREEN color packet can be expressed though + * RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, + /** * Enables counters for this flow rule. * -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v3 2/3] app/testpmd: add skip cman support for testpmd 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla @ 2023-02-09 7:24 ` Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 7:24 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v3 3/3] net/cnxk: skip red drop for ingress policer 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-09 7:24 ` Rakesh Kudurumalla 2023-02-09 7:31 ` [PATCH v3 1/3] lib: skip congestion management configuration Jerin Jacob 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 7:24 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ 3 files changed, 52 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..753da67c5a 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -80,6 +80,7 @@ vxlan_gpe = Y [rte_flow actions] count = Y +skip_cman = Y drop = Y flag = Y mark = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v3 1/3] lib: skip congestion management configuration 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-09 7:31 ` Jerin Jacob 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-09 7:31 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko, jerinj, ndabilpuram, dev On Thu, Feb 9, 2023 at 12:54 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > Skip the congestion management configuration applied using > rte_eth_cman_config_set() API on the given ethdev Rx queue. > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > --- > v3: Updated comments and programmers guide regrading > new action > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > lib/ethdev/rte_flow.h | 20 ++++++++++++++++++++ > 2 files changed, 42 insertions(+) > > diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst > index 3e6242803d..0737b877da 100644 > --- a/doc/guides/prog_guide/rte_flow.rst > +++ b/doc/guides/prog_guide/rte_flow.rst > @@ -1840,6 +1840,28 @@ Drop packets. > | no properties | > +---------------+ > > + > +Action: ``SKIP_CMAN`` > +^^^^^^^^^^^^^^^^^^^ > + > +Skip congestion management on received packets > + > +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx > + queue's congestion mechanism.Once applied packets congestion configuration > + is bypassed on that particular ethdev Rx queue for all packets directed > + to that receive queue > + > +.. _table_rte_flow_action_skip_cman: > + > +.. table:: SKIP_CMAN > + > + +---------------+ > + | Field | > + +===============+ > + | no properties | > + +---------------+ > + > + > Action: ``COUNT`` > ^^^^^^^^^^^^^^^^^ > > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > index b60987db4b..5ae00e2245 100644 > --- a/lib/ethdev/rte_flow.h > +++ b/lib/ethdev/rte_flow.h > @@ -2203,6 +2203,26 @@ enum rte_flow_action_type { > */ > RTE_FLOW_ACTION_TYPE_DROP, > > + /** > + * Skip congestion management configuration > + * > + * ethdev: introduce flow action to skip congestion > + * management configuration This supposes to be git commit heading. > + * > + * Using rte_eth_cman_config_set() API the application > + * can configure ethdev Rx queue's congestion mechanism. > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > + * congestion configuration applied to the given ethdev Rx queue. This paragraph is OK here. > + * > + * This feature helps to skip the congestion management processing > + * based on per flow or the packet color identified by > + * rte_flow meter object. For example, If one Rx queue configured as > + * RED congestion and application wants tobypass the RED congestion > + * processing for all GREEN color packet can be expressed though > + * RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Move this to git commit log > + */ > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > + > /** > * Enables counters for this flow rule. > * > -- > 2.25.1 > ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v4 1/3] ethdev: skip congestion management configuration 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-09 7:31 ` [PATCH v3 1/3] lib: skip congestion management configuration Jerin Jacob @ 2023-02-09 8:35 ` Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (3 more replies) 3 siblings, 4 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 8:35 UTC (permalink / raw) To: Ori Kam, Thomas Monjalon, Ferruh Yigit, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- v4: updated commit message doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 2 files changed, 33 insertions(+) diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..0737b877da 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..f4eb4232d4 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { */ RTE_FLOW_ACTION_TYPE_DROP, + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, + /** * Enables counters for this flow rule. * -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v4 2/3] app/testpmd: add skip cman support for testpmd 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla @ 2023-02-09 8:35 ` Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 8:35 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-09 8:35 ` Rakesh Kudurumalla 2023-02-09 17:14 ` Jerin Jacob 2023-02-09 14:51 ` [PATCH v4 1/3] ethdev: skip congestion management configuration Ori Kam 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla 3 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-09 8:35 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ 3 files changed, 52 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..753da67c5a 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -80,6 +80,7 @@ vxlan_gpe = Y [rte_flow actions] count = Y +skip_cman = Y drop = Y flag = Y mark = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer 2023-02-09 8:35 ` [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-09 17:14 ` Jerin Jacob 0 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-09 17:14 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao, jerinj, dev On Thu, Feb 9, 2023 at 2:06 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > Dropping of packets is based on action configured > to meter.If both skip_red and drop actions are configured > then tail dropping in invoked else if only drop action is > configured then RED drop is invoked.This action is supported > only when RED is configured using rte_eth_cman_config_set() > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Fix CI reported issue at http://mails.dpdk.org/archives/test-report/2023-February/350229.html https://patches.dpdk.org/project/dpdk/patch/20230209083558.958566-3-rkudurumalla@marvell.com/ > --- > doc/guides/nics/features/cnxk.ini | 1 + > drivers/net/cnxk/cnxk_ethdev.h | 1 + > drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ > 3 files changed, 52 insertions(+) > > diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini > index f81628da77..753da67c5a 100644 > --- a/doc/guides/nics/features/cnxk.ini > +++ b/doc/guides/nics/features/cnxk.ini > @@ -80,6 +80,7 @@ vxlan_gpe = Y > > [rte_flow actions] > count = Y > +skip_cman = Y > drop = Y > flag = Y > mark = Y > diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h > index f0eab4244c..ba35873124 100644 > --- a/drivers/net/cnxk/cnxk_ethdev.h > +++ b/drivers/net/cnxk/cnxk_ethdev.h > @@ -168,6 +168,7 @@ struct policy_actions { > uint16_t queue; > uint32_t mtr_id; > struct action_rss *rss_desc; > + bool skip_red; > }; > }; > > diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c > index dcfa4223d5..27a6e4ef3d 100644 > --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c > +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c > @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, > if (action->type == RTE_FLOW_ACTION_TYPE_VOID) > supported[i] = true; > > + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + supported[i] = true; > + > if (!supported[i]) > return update_mtr_err(i, error, true); > } > @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, > fmp->actions[i].action_fate = > action->type; > } > + > + if (action->type == > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + fmp->actions[i].skip_red = true; > } > } > } > @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) > cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; > } > > +static void > +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, > + struct roc_nix_bpf_cfg *cfg) > +{ > + struct cnxk_mtr_policy_node *policy = mtr->policy; > + > + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || > + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || > + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { > + if (policy->actions[RTE_COLOR_GREEN].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_GREEN].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_YELLOW].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_YELLOW].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_RED].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_RED].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_RED; > + } > + } > +} > + > static void > nix_precolor_table_map(struct cnxk_meter_node *mtr, > struct roc_nix_bpf_precolor *tbl, > @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) > if (!mtr[i]->is_used) { > memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); > nix_mtr_config_map(mtr[i], &cfg); > + for (j = 0; j < mtr[i]->rq_num; j++) { > + rq = &dev->rqs[mtr[i]->rq_id[j]]; > + nix_mtr_config_red(mtr[i], rq, &cfg); > + } > rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, > lvl_map[mtr[i]->level], > &cfg); > -- > 2.25.1 > ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH v4 1/3] ethdev: skip congestion management configuration 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-09 14:51 ` Ori Kam 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Ori Kam @ 2023-02-09 14:51 UTC (permalink / raw) To: Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Ferruh Yigit, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev Hi Rakesh, > -----Original Message----- > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Sent: Thursday, 9 February 2023 10:36 > Subject: [PATCH v4 1/3] ethdev: skip congestion management configuration > > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > --- Acked-by: Ori Kam <orika@nvidia.com> Best, Ori ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v5 1/3] ethdev: skip congestion management configuration 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-09 14:51 ` [PATCH v4 1/3] ethdev: skip congestion management configuration Ori Kam @ 2023-02-10 8:14 ` Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (3 more replies) 3 siblings, 4 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:14 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao, Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> --- v5: Updated doc doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 5 files changed, 36 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 510cc6679d..b93f053a2b 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -198,6 +198,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..4efadcc87b 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..f4eb4232d4 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { */ RTE_FLOW_ACTION_TYPE_DROP, + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, + /** * Enables counters for this flow rule. * -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v5 2/3] app/testpmd: add skip cman support for testpmd 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla @ 2023-02-10 8:14 ` Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:14 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v5 3/3] net/cnxk: skip red drop for ingress policer 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-10 8:14 ` Rakesh Kudurumalla 2023-02-10 8:16 ` [PATCH v5 1/3] ethdev: skip congestion management configuration Jerin Jacob 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:14 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++++ 3 files changed, 52 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index 7947c044bb..afbb6dc8c9 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -80,6 +80,7 @@ vxlan_gpe = Y [rte_flow actions] count = Y +skip_cman = Y drop = Y flag = Y mark = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v5 1/3] ethdev: skip congestion management configuration 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-10 8:16 ` Jerin Jacob 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-10 8:16 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao, Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko, jerinj, dev On Fri, Feb 10, 2023 at 1:44 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> > --- > v5: Updated doc > doc/guides/nics/features/cnxk.ini | 1 + > doc/guides/nics/features/cnxk_vf.ini | 1 + > doc/guides/nics/features/default.ini | 1 + > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > lib/ethdev/rte_flow.h | 11 +++++++++++ > 5 files changed, 36 insertions(+) > > diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini > index f81628da77..7947c044bb 100644 > --- a/doc/guides/nics/features/cnxk.ini > +++ b/doc/guides/nics/features/cnxk.ini > @@ -94,4 +94,5 @@ queue = Y > represented_port = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y > diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini > index c4ee32a9ad..873e1dcc0a 100644 > --- a/doc/guides/nics/features/cnxk_vf.ini > +++ b/doc/guides/nics/features/cnxk_vf.ini > @@ -83,4 +83,5 @@ pf = Y > queue = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y cnxk specific . Move to 3/3 patch. > diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini > index 510cc6679d..b93f053a2b 100644 > --- a/doc/guides/nics/features/default.ini > +++ b/doc/guides/nics/features/default.ini > @@ -198,6 +198,7 @@ set_tag = > set_tp_dst = > set_tp_src = > set_ttl = > +skip_cman = > vf = > vxlan_decap = > vxlan_encap = > diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst > index 3e6242803d..4efadcc87b 100644 > --- a/doc/guides/prog_guide/rte_flow.rst > +++ b/doc/guides/prog_guide/rte_flow.rst > @@ -1840,6 +1840,28 @@ Drop packets. > | no properties | > +---------------+ > > + > +Action: ``SKIP_CMAN`` > +^^^^^^^^^^^^^^^^^^^^^ > + > +Skip congestion management on received packets > + > +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx > + queue's congestion mechanism.Once applied packets congestion configuration > + is bypassed on that particular ethdev Rx queue for all packets directed > + to that receive queue > + > +.. _table_rte_flow_action_skip_cman: > + > +.. table:: SKIP_CMAN > + > + +---------------+ > + | Field | > + +===============+ > + | no properties | > + +---------------+ > + > + > Action: ``COUNT`` > ^^^^^^^^^^^^^^^^^ > > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > index b60987db4b..f4eb4232d4 100644 > --- a/lib/ethdev/rte_flow.h > +++ b/lib/ethdev/rte_flow.h > @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { > */ > RTE_FLOW_ACTION_TYPE_DROP, > > + /** > + * Skip congestion management configuration > + * > + * Using rte_eth_cman_config_set() API the application > + * can configure ethdev Rx queue's congestion mechanism. > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > + * congestion configuration applied to the given ethdev Rx queue. > + * > + */ > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > + > /** > * Enables counters for this flow rule. > * > -- > 2.25.1 > ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-10 8:16 ` [PATCH v5 1/3] ethdev: skip congestion management configuration Jerin Jacob @ 2023-02-10 8:26 ` Rakesh Kudurumalla 2023-02-10 8:26 ` [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (5 more replies) 3 siblings, 6 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:26 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> --- v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 3 files changed, 34 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 510cc6679d..b93f053a2b 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -198,6 +198,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..4efadcc87b 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..f4eb4232d4 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { */ RTE_FLOW_ACTION_TYPE_DROP, + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, + /** * Enables counters for this flow rule. * -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla @ 2023-02-10 8:26 ` Rakesh Kudurumalla 2023-02-10 23:08 ` Ferruh Yigit 2023-02-10 8:26 ` [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (4 subsequent siblings) 5 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:26 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd 2023-02-10 8:26 ` [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-10 23:08 ` Ferruh Yigit 2023-02-14 13:12 ` Ferruh Yigit 0 siblings, 1 reply; 70+ messages in thread From: Ferruh Yigit @ 2023-02-10 23:08 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > added support for testpmd application to accept > skip_cman action while configuring policy action > Can you please give some sample testpmd commands with new action in commit log, it helps. > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > --- > app/test-pmd/cmdline_flow.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > > diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c > index 88108498e0..fc003e0096 100644 > --- a/app/test-pmd/cmdline_flow.c > +++ b/app/test-pmd/cmdline_flow.c > @@ -472,6 +472,7 @@ enum index { > ACTION_END, > ACTION_VOID, > ACTION_PASSTHRU, > + ACTION_SKIP_CMAN, > ACTION_JUMP, > ACTION_JUMP_GROUP, > ACTION_MARK, > @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { > ACTION_END, > ACTION_VOID, > ACTION_PASSTHRU, > + ACTION_SKIP_CMAN, > ACTION_JUMP, > ACTION_MARK, > ACTION_FLAG, > @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { > .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), > .call = parse_vc, > }, > + [ACTION_SKIP_CMAN] = { > + .name = "skip_cman", > + .help = "bypass cman on received packets", > + .priv = PRIV_ACTION(SKIP_CMAN, 0), > + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), > + .call = parse_vc, > + }, > [ACTION_JUMP] = { > .name = "jump", > .help = "redirect traffic to a given group", ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd 2023-02-10 23:08 ` Ferruh Yigit @ 2023-02-14 13:12 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-14 13:12 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Aman Singh, Yuying Zhang, jerinj Cc: ndabilpuram, dev On 2/10/2023 11:08 PM, Ferruh Yigit wrote: > On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: >> added support for testpmd application to accept >> skip_cman action while configuring policy action >> > > Can you please give some sample testpmd commands with new action in > commit log, it helps. > reminder of this comment, v8 doesn't have this. please check following as sample: "app/testpmd: add shared indirect action support" https://git.dpdk.org/next/dpdk-next-net/commit/?id=c13f2654cd3c67e526be0 >> Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> >> --- >> app/test-pmd/cmdline_flow.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c >> index 88108498e0..fc003e0096 100644 >> --- a/app/test-pmd/cmdline_flow.c >> +++ b/app/test-pmd/cmdline_flow.c >> @@ -472,6 +472,7 @@ enum index { >> ACTION_END, >> ACTION_VOID, >> ACTION_PASSTHRU, >> + ACTION_SKIP_CMAN, >> ACTION_JUMP, >> ACTION_JUMP_GROUP, >> ACTION_MARK, >> @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { >> ACTION_END, >> ACTION_VOID, >> ACTION_PASSTHRU, >> + ACTION_SKIP_CMAN, >> ACTION_JUMP, >> ACTION_MARK, >> ACTION_FLAG, >> @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { >> .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), >> .call = parse_vc, >> }, >> + [ACTION_SKIP_CMAN] = { >> + .name = "skip_cman", >> + .help = "bypass cman on received packets", >> + .priv = PRIV_ACTION(SKIP_CMAN, 0), >> + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), >> + .call = parse_vc, >> + }, >> [ACTION_JUMP] = { >> .name = "jump", >> .help = "redirect traffic to a given group", > ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla 2023-02-10 8:26 ` [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-10 8:26 ` Rakesh Kudurumalla 2023-02-10 23:18 ` Ferruh Yigit 2023-02-10 9:27 ` [PATCH v6 1/3] ethdev: skip congestion management configuration Jerin Jacob ` (3 subsequent siblings) 5 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-10 8:26 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 2 ++ doc/guides/nics/features/cnxk_vf.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++ 4 files changed, 54 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..afbb6dc8c9 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -80,6 +80,7 @@ vxlan_gpe = Y [rte_flow actions] count = Y +skip_cman = Y drop = Y flag = Y mark = Y @@ -94,4 +95,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer 2023-02-10 8:26 ` [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-10 23:18 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-10 23:18 UTC (permalink / raw) To: Rakesh Kudurumalla, Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini > index f81628da77..afbb6dc8c9 100644 > --- a/doc/guides/nics/features/cnxk.ini > +++ b/doc/guides/nics/features/cnxk.ini > @@ -80,6 +80,7 @@ vxlan_gpe = Y > > [rte_flow actions] > count = Y > +skip_cman = Y > drop = Y > flag = Y > mark = Y > @@ -94,4 +95,5 @@ queue = Y > represented_port = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y 'skip_cman' added twice. ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla 2023-02-10 8:26 ` [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-10 8:26 ` [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-10 9:27 ` Jerin Jacob 2023-02-10 23:10 ` Ferruh Yigit ` (2 subsequent siblings) 5 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-10 9:27 UTC (permalink / raw) To: Rakesh Kudurumalla Cc: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko, jerinj, ndabilpuram, dev On Fri, Feb 10, 2023 at 1:56 PM Rakesh Kudurumalla <rkudurumalla@marvell.com> wrote: > > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> Acked-by: Jerin Jacob <jerinj@marvell.com> > --- > v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch > doc/guides/nics/features/default.ini | 1 + > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > lib/ethdev/rte_flow.h | 11 +++++++++++ > 3 files changed, 34 insertions(+) > > diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini > index 510cc6679d..b93f053a2b 100644 > --- a/doc/guides/nics/features/default.ini > +++ b/doc/guides/nics/features/default.ini > @@ -198,6 +198,7 @@ set_tag = > set_tp_dst = > set_tp_src = > set_ttl = > +skip_cman = > vf = > vxlan_decap = > vxlan_encap = > diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst > index 3e6242803d..4efadcc87b 100644 > --- a/doc/guides/prog_guide/rte_flow.rst > +++ b/doc/guides/prog_guide/rte_flow.rst > @@ -1840,6 +1840,28 @@ Drop packets. > | no properties | > +---------------+ > > + > +Action: ``SKIP_CMAN`` > +^^^^^^^^^^^^^^^^^^^^^ > + > +Skip congestion management on received packets > + > +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx > + queue's congestion mechanism.Once applied packets congestion configuration > + is bypassed on that particular ethdev Rx queue for all packets directed > + to that receive queue > + > +.. _table_rte_flow_action_skip_cman: > + > +.. table:: SKIP_CMAN > + > + +---------------+ > + | Field | > + +===============+ > + | no properties | > + +---------------+ > + > + > Action: ``COUNT`` > ^^^^^^^^^^^^^^^^^ > > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > index b60987db4b..f4eb4232d4 100644 > --- a/lib/ethdev/rte_flow.h > +++ b/lib/ethdev/rte_flow.h > @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { > */ > RTE_FLOW_ACTION_TYPE_DROP, > > + /** > + * Skip congestion management configuration > + * > + * Using rte_eth_cman_config_set() API the application > + * can configure ethdev Rx queue's congestion mechanism. > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > + * congestion configuration applied to the given ethdev Rx queue. > + * > + */ > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > + > /** > * Enables counters for this flow rule. > * > -- > 2.25.1 > ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-10 9:27 ` [PATCH v6 1/3] ethdev: skip congestion management configuration Jerin Jacob @ 2023-02-10 23:10 ` Ferruh Yigit 2023-02-13 12:34 ` Ori Kam 2023-02-11 0:35 ` Ferruh Yigit 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla 5 siblings, 1 reply; 70+ messages in thread From: Ferruh Yigit @ 2023-02-10 23:10 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> > --- > v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch > doc/guides/nics/features/default.ini | 1 + > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > lib/ethdev/rte_flow.h | 11 +++++++++++ Should the release notes needs to be updated for this? ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-10 23:10 ` Ferruh Yigit @ 2023-02-13 12:34 ` Ori Kam 2023-02-13 13:54 ` Ferruh Yigit 0 siblings, 1 reply; 70+ messages in thread From: Ori Kam @ 2023-02-13 12:34 UTC (permalink / raw) To: Ferruh Yigit, Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Andrew Rybchenko Cc: jerinj, ndabilpuram, dev > -----Original Message----- > From: Ferruh Yigit <ferruh.yigit@amd.com> > Sent: Saturday, 11 February 2023 1:10 > > On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > > Introduce new flow action to skip congestion management configuration > > This feature helps to skip the congestion management processing > > based on per flow or the packet color identified by > > rte_flow meter object. For example, If one Rx queue configured as > > RED congestion and application wants tobypass the RED congestion > > processing for all GREEN color packet can be expressed though > > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > > Acked-by: Ori Kam <orika@nvidia.com> > > --- > > v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch > > doc/guides/nics/features/default.ini | 1 + > > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > > lib/ethdev/rte_flow.h | 11 +++++++++++ > > Should the release notes needs to be updated for this? Yes, I think so it is a new rte_flow action ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-13 12:34 ` Ori Kam @ 2023-02-13 13:54 ` Ferruh Yigit 2023-02-13 15:44 ` Jerin Jacob 0 siblings, 1 reply; 70+ messages in thread From: Ferruh Yigit @ 2023-02-13 13:54 UTC (permalink / raw) To: Ori Kam, Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Andrew Rybchenko, jerinj Cc: ndabilpuram, dev On 2/13/2023 12:34 PM, Ori Kam wrote: > > >> -----Original Message----- >> From: Ferruh Yigit <ferruh.yigit@amd.com> >> Sent: Saturday, 11 February 2023 1:10 >> >> On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: >>> Introduce new flow action to skip congestion management configuration >>> This feature helps to skip the congestion management processing >>> based on per flow or the packet color identified by >>> rte_flow meter object. For example, If one Rx queue configured as >>> RED congestion and application wants tobypass the RED congestion >>> processing for all GREEN color packet can be expressed though >>> RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action >>> >>> Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> >>> Acked-by: Ori Kam <orika@nvidia.com> >>> --- >>> v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch >>> doc/guides/nics/features/default.ini | 1 + >>> doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ >>> lib/ethdev/rte_flow.h | 11 +++++++++++ >> >> Should the release notes needs to be updated for this? > > Yes, I think so it is a new rte_flow action Hi Jerin, what do you think about this? ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-13 13:54 ` Ferruh Yigit @ 2023-02-13 15:44 ` Jerin Jacob 2023-02-13 15:53 ` Ferruh Yigit 0 siblings, 1 reply; 70+ messages in thread From: Jerin Jacob @ 2023-02-13 15:44 UTC (permalink / raw) To: Ferruh Yigit Cc: Ori Kam, Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Andrew Rybchenko, jerinj, ndabilpuram, dev On Mon, Feb 13, 2023 at 7:25 PM Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > On 2/13/2023 12:34 PM, Ori Kam wrote: > > > > > >> -----Original Message----- > >> From: Ferruh Yigit <ferruh.yigit@amd.com> > >> Sent: Saturday, 11 February 2023 1:10 > >> > >> On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > >>> Introduce new flow action to skip congestion management configuration > >>> This feature helps to skip the congestion management processing > >>> based on per flow or the packet color identified by > >>> rte_flow meter object. For example, If one Rx queue configured as > >>> RED congestion and application wants tobypass the RED congestion > >>> processing for all GREEN color packet can be expressed though > >>> RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > >>> > >>> Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > >>> Acked-by: Ori Kam <orika@nvidia.com> > >>> --- > >>> v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch > >>> doc/guides/nics/features/default.ini | 1 + > >>> doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > >>> lib/ethdev/rte_flow.h | 11 +++++++++++ > >> > >> Should the release notes needs to be updated for this? > > > > Yes, I think so it is a new rte_flow action > > Hi Jerin, what do you think about this? Yes. Rakesh already send a new version with this change. Thanks. ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-13 15:44 ` Jerin Jacob @ 2023-02-13 15:53 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-13 15:53 UTC (permalink / raw) To: Jerin Jacob Cc: Ori Kam, Rakesh Kudurumalla, NBU-Contact-Thomas Monjalon (EXTERNAL), Andrew Rybchenko, jerinj, ndabilpuram, dev On 2/13/2023 3:44 PM, Jerin Jacob wrote: > On Mon, Feb 13, 2023 at 7:25 PM Ferruh Yigit <ferruh.yigit@amd.com> wrote: >> >> On 2/13/2023 12:34 PM, Ori Kam wrote: >>> >>> >>>> -----Original Message----- >>>> From: Ferruh Yigit <ferruh.yigit@amd.com> >>>> Sent: Saturday, 11 February 2023 1:10 >>>> >>>> On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: >>>>> Introduce new flow action to skip congestion management configuration >>>>> This feature helps to skip the congestion management processing >>>>> based on per flow or the packet color identified by >>>>> rte_flow meter object. For example, If one Rx queue configured as >>>>> RED congestion and application wants tobypass the RED congestion >>>>> processing for all GREEN color packet can be expressed though >>>>> RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action >>>>> >>>>> Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> >>>>> Acked-by: Ori Kam <orika@nvidia.com> >>>>> --- >>>>> v6: Moved cnxk.ini and cnxk_vf.ini to cnxk specific patch >>>>> doc/guides/nics/features/default.ini | 1 + >>>>> doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ >>>>> lib/ethdev/rte_flow.h | 11 +++++++++++ >>>> >>>> Should the release notes needs to be updated for this? >>> >>> Yes, I think so it is a new rte_flow action >> >> Hi Jerin, what do you think about this? > > Yes. Rakesh already send a new version with this change. Thanks. v7 doesn't have release notes update, that is why I am pinging again. ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla ` (3 preceding siblings ...) 2023-02-10 23:10 ` Ferruh Yigit @ 2023-02-11 0:35 ` Ferruh Yigit 2023-02-11 5:16 ` Jerin Jacob 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla 5 siblings, 1 reply; 70+ messages in thread From: Ferruh Yigit @ 2023-02-11 0:35 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, David Marchand On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > index b60987db4b..f4eb4232d4 100644 > --- a/lib/ethdev/rte_flow.h > +++ b/lib/ethdev/rte_flow.h > @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { > */ > RTE_FLOW_ACTION_TYPE_DROP, > > + /** > + * Skip congestion management configuration > + * > + * Using rte_eth_cman_config_set() API the application > + * can configure ethdev Rx queue's congestion mechanism. > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > + * congestion configuration applied to the given ethdev Rx queue. > + * > + */ > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > + Inserting new enum item in to the middle of the enum upsets the ABI checks [1], can it go to the end? [1] 1 function with some indirect sub-type change: [C] 'function size_t rte_flow_copy(rte_flow_desc*, size_t, const rte_flow_attr*, const rte_flow_item*, const rte_flow_action*)' at rte_flow.c:1092:1 has some indirect sub-type changes: parameter 1 of type 'rte_flow_desc*' has sub-type changes: in pointed to type 'struct rte_flow_desc' at rte_flow.h:4326:1: type size hasn't changed 1 data member changes (1 filtered): type of 'rte_flow_action* actions' changed: in pointed to type 'struct rte_flow_action' at rte_flow.h:3775:1: type size hasn't changed 1 data member change: type of 'rte_flow_action_type type' changed: type size hasn't changed 1 enumerator insertion: 'rte_flow_action_type::RTE_FLOW_ACTION_TYPE_SKIP_CMAN' value '8' 50 enumerator changes: 'rte_flow_action_type::RTE_FLOW_ACTION_TYPE_COUNT' from value '8' to '9' at rte_flow.h:2216:1 ... ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v6 1/3] ethdev: skip congestion management configuration 2023-02-11 0:35 ` Ferruh Yigit @ 2023-02-11 5:16 ` Jerin Jacob 0 siblings, 0 replies; 70+ messages in thread From: Jerin Jacob @ 2023-02-11 5:16 UTC (permalink / raw) To: Ferruh Yigit Cc: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko, jerinj, ndabilpuram, dev, David Marchand On Sat, Feb 11, 2023 at 6:05 AM Ferruh Yigit <ferruh.yigit@amd.com> wrote: > > On 2/10/2023 8:26 AM, Rakesh Kudurumalla wrote: > > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > > index b60987db4b..f4eb4232d4 100644 > > --- a/lib/ethdev/rte_flow.h > > +++ b/lib/ethdev/rte_flow.h > > @@ -2203,6 +2203,17 @@ enum rte_flow_action_type { > > */ > > RTE_FLOW_ACTION_TYPE_DROP, > > > > + /** > > + * Skip congestion management configuration > > + * > > + * Using rte_eth_cman_config_set() API the application > > + * can configure ethdev Rx queue's congestion mechanism. > > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > > + * congestion configuration applied to the given ethdev Rx queue. > > + * > > + */ > > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > > + > > Inserting new enum item in to the middle of the enum upsets the ABI > checks [1], can it go to the end? Yes. > > > > > [1] > 1 function with some indirect sub-type change: > > [C] 'function size_t rte_flow_copy(rte_flow_desc*, size_t, const > rte_flow_attr*, const rte_flow_item*, const rte_flow_action*)' at > rte_flow.c:1092:1 has some indirect sub-type changes: > parameter 1 of type 'rte_flow_desc*' has sub-type changes: > in pointed to type 'struct rte_flow_desc' at rte_flow.h:4326:1: > type size hasn't changed > 1 data member changes (1 filtered): > type of 'rte_flow_action* actions' changed: > in pointed to type 'struct rte_flow_action' at > rte_flow.h:3775:1: > type size hasn't changed > 1 data member change: > type of 'rte_flow_action_type type' changed: > type size hasn't changed > 1 enumerator insertion: > > 'rte_flow_action_type::RTE_FLOW_ACTION_TYPE_SKIP_CMAN' value '8' > 50 enumerator changes: > 'rte_flow_action_type::RTE_FLOW_ACTION_TYPE_COUNT' > from value '8' to '9' at rte_flow.h:2216:1 > ... ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v7 1/3] ethdev: skip congestion management configuration 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla ` (4 preceding siblings ...) 2023-02-11 0:35 ` Ferruh Yigit @ 2023-02-12 9:00 ` Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (2 more replies) 5 siblings, 3 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-12 9:00 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> --- v7: Moved skip_cman macro to the end doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 3 files changed, 34 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 510cc6679d..b93f053a2b 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -198,6 +198,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..4efadcc87b 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..6ab632df9d 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2736,6 +2736,17 @@ enum rte_flow_action_type { * No associated configuration structure. */ RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, + + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, }; /** -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v7 2/3] app/testpmd: add skip cman support for testpmd 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla @ 2023-02-12 9:00 ` Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla 2 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-12 9:00 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v7 3/3] net/cnxk: skip red drop for ingress policer 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-12 9:00 ` Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla 2 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-12 9:00 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++ 4 files changed, 53 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v8 1/4] ethdev: skip congestion management configuration 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-14 13:04 ` Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (4 more replies) 2 siblings, 5 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-14 13:04 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> --- v8: updated release notes doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 3 files changed, 34 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 510cc6679d..b93f053a2b 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -198,6 +198,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..4efadcc87b 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..6ab632df9d 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2736,6 +2736,17 @@ enum rte_flow_action_type { * No associated configuration structure. */ RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, + + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, }; /** -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla @ 2023-02-14 13:04 ` Rakesh Kudurumalla 2023-02-14 13:21 ` Ferruh Yigit 2023-02-14 13:04 ` [PATCH v8 3/4] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (3 subsequent siblings) 4 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-14 13:04 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd 2023-02-14 13:04 ` [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-14 13:21 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-14 13:21 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev On 2/14/2023 1:04 PM, Rakesh Kudurumalla wrote: > added support for testpmd application to accept > skip_cman action while configuring policy action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > --- > app/test-pmd/cmdline_flow.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > > diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c > index 88108498e0..fc003e0096 100644 > --- a/app/test-pmd/cmdline_flow.c > +++ b/app/test-pmd/cmdline_flow.c > @@ -472,6 +472,7 @@ enum index { > ACTION_END, > ACTION_VOID, > ACTION_PASSTHRU, > + ACTION_SKIP_CMAN, > ACTION_JUMP, > ACTION_JUMP_GROUP, > ACTION_MARK, > @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { > ACTION_END, > ACTION_VOID, > ACTION_PASSTHRU, > + ACTION_SKIP_CMAN, > ACTION_JUMP, > ACTION_MARK, > ACTION_FLAG, > @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { > .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), > .call = parse_vc, > }, > + [ACTION_SKIP_CMAN] = { > + .name = "skip_cman", > + .help = "bypass cman on received packets", > + .priv = PRIV_ACTION(SKIP_CMAN, 0), > + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), > + .call = parse_vc, > + }, > [ACTION_JUMP] = { > .name = "jump", > .help = "redirect traffic to a given group", Hi Ori, any objection to this patch? ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v8 3/4] net/cnxk: skip red drop for ingress policer 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-14 13:04 ` Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 4/4] doc: update release notes for 23_03.rst Rakesh Kudurumalla ` (2 subsequent siblings) 4 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-14 13:04 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++++ 4 files changed, 53 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v8 4/4] doc: update release notes for 23_03.rst 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 3/4] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-14 13:04 ` Rakesh Kudurumalla 2023-02-14 13:19 ` Ferruh Yigit 2023-02-14 13:13 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Ferruh Yigit 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla 4 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-14 13:04 UTC (permalink / raw) Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Updated release notes on new action to skip congestion management configuration Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/rel_notes/release_23_03.rst | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index 1fa101c420..a58e92071c 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -55,6 +55,12 @@ New Features Also, make sure to start the actual text at the margin. ======================================================= +* **Added support to skip congestion management configuration.** + + Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip + congestion management processing based on per flow or packet color + identified by rte_flow meter object + * **Updated AMD axgbe driver.** * Added multi-process support. @@ -72,6 +78,11 @@ New Features Added support for SHAKE hash algorithm in cn9k and cn10k. +* **Updated Marvell cnxk ethdev PMD.** + + Added support to skip RED using new action + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` + * **Updated Intel QuickAssist Technology (QAT) crypto driver.** * Added support for SHA3 224/256/384/512 plain hash in QAT GEN 3. -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v8 4/4] doc: update release notes for 23_03.rst 2023-02-14 13:04 ` [PATCH v8 4/4] doc: update release notes for 23_03.rst Rakesh Kudurumalla @ 2023-02-14 13:19 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-14 13:19 UTC (permalink / raw) To: Rakesh Kudurumalla; +Cc: jerinj, ndabilpuram, dev On 2/14/2023 1:04 PM, Rakesh Kudurumalla wrote: > Updated release notes on new action to skip > congestion management configuration > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> please distribute this patch to relevant patches, like flow api update can go into first patch (ethdev one), driver update to the one that updates driver. No need to have a separate patch for release updates. > --- > doc/guides/rel_notes/release_23_03.rst | 11 +++++++++++ > 1 file changed, 11 insertions(+) > > diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst > index 1fa101c420..a58e92071c 100644 > --- a/doc/guides/rel_notes/release_23_03.rst > +++ b/doc/guides/rel_notes/release_23_03.rst > @@ -55,6 +55,12 @@ New Features > Also, make sure to start the actual text at the margin. > ======================================================= > > +* **Added support to skip congestion management configuration.** > + > + Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip > + congestion management processing based on per flow or packet color > + identified by rte_flow meter object > + Can you please rebase on latest next-net, this part is different in latest head? Also there are a few other flow api related release notes updates, can you please keep the note similar to existing ones? > * **Updated AMD axgbe driver.** > > * Added multi-process support. > @@ -72,6 +78,11 @@ New Features > > Added support for SHAKE hash algorithm in cn9k and cn10k. > > +* **Updated Marvell cnxk ethdev PMD.** > + > + Added support to skip RED using new action > + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` > + > * **Updated Intel QuickAssist Technology (QAT) crypto driver.** > > * Added support for SHA3 224/256/384/512 plain hash in QAT GEN 3. ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v8 1/4] ethdev: skip congestion management configuration 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-14 13:04 ` [PATCH v8 4/4] doc: update release notes for 23_03.rst Rakesh Kudurumalla @ 2023-02-14 13:13 ` Ferruh Yigit 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla 4 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-14 13:13 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev On 2/14/2023 1:04 PM, Rakesh Kudurumalla wrote: > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> Jerin also acked the patch, please include that too. ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v9 1/3] ethdev: skip congestion management configuration 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla ` (3 preceding siblings ...) 2023-02-14 13:13 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Ferruh Yigit @ 2023-02-15 6:25 ` Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (3 more replies) 4 siblings, 4 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 6:25 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> Acked-by: Jerin Jacob <jerinj@marvell.com> --- v9: Distributed doc updates to relevant patches doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ doc/guides/rel_notes/release_23_03.rst | 6 ++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 4 files changed, 40 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 510cc6679d..b93f053a2b 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -198,6 +198,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 3e6242803d..4efadcc87b 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1840,6 +1840,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index bf90feba68..1dd10e26db 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -60,6 +60,12 @@ New Features Added lock annotations attributes so that clang can statically analyze lock correctness. +* **Added support to skip congestion management configuration.** + + Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip + congestion management processing based on per flow or packet color + identified by rte_flow meter object + * **Updated AMD axgbe driver.** * Added multi-process support. diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index b60987db4b..6ab632df9d 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2736,6 +2736,17 @@ enum rte_flow_action_type { * No associated configuration structure. */ RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, + + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, }; /** -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v9 2/3] app/testpmd: add skip cman support for testpmd 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla @ 2023-02-15 6:25 ` Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla ` (2 subsequent siblings) 3 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 6:25 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 88108498e0..fc003e0096 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -472,6 +472,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1825,6 +1826,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5159,6 +5161,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-15 6:25 ` Rakesh Kudurumalla 2023-02-15 9:45 ` Ferruh Yigit 2023-02-15 10:00 ` Nithin Kumar Dabilpuram 2023-02-15 9:43 ` [PATCH v9 1/3] ethdev: skip congestion management configuration Ferruh Yigit 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla 3 siblings, 2 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 6:25 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + doc/guides/rel_notes/release_23_03.rst | 4 +++ drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++ 5 files changed, 57 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index 1dd10e26db..9f53ded5fb 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -98,6 +98,10 @@ New Features ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter required for eth_rx, eth_tx, crypto and timer eventdev adapters. +* **Updated Marvell cnxk ethdev PMD.** + + Added support to skip RED using new action + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` Removed Items ------------- diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer 2023-02-15 6:25 ` [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-15 9:45 ` Ferruh Yigit 2023-02-15 10:00 ` Nithin Kumar Dabilpuram 1 sibling, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-15 9:45 UTC (permalink / raw) To: Rakesh Kudurumalla, Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev On 2/15/2023 6:25 AM, Rakesh Kudurumalla wrote: > Dropping of packets is based on action configured > to meter.If both skip_red and drop actions are configured > then tail dropping in invoked else if only drop action is > configured then RED drop is invoked.This action is supported > only when RED is configured using rte_eth_cman_config_set() > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Ack from driver maintainers required to proceed. > --- > doc/guides/nics/features/cnxk.ini | 1 + > doc/guides/nics/features/cnxk_vf.ini | 1 + > doc/guides/rel_notes/release_23_03.rst | 4 +++ > drivers/net/cnxk/cnxk_ethdev.h | 1 + > drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++ > 5 files changed, 57 insertions(+) > > diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini > index f81628da77..7947c044bb 100644 > --- a/doc/guides/nics/features/cnxk.ini > +++ b/doc/guides/nics/features/cnxk.ini > @@ -94,4 +94,5 @@ queue = Y > represented_port = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y > diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini > index c4ee32a9ad..873e1dcc0a 100644 > --- a/doc/guides/nics/features/cnxk_vf.ini > +++ b/doc/guides/nics/features/cnxk_vf.ini > @@ -83,4 +83,5 @@ pf = Y > queue = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y > diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst > index 1dd10e26db..9f53ded5fb 100644 > --- a/doc/guides/rel_notes/release_23_03.rst > +++ b/doc/guides/rel_notes/release_23_03.rst > @@ -98,6 +98,10 @@ New Features > ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter > required for eth_rx, eth_tx, crypto and timer eventdev adapters. > > +* **Updated Marvell cnxk ethdev PMD.** > + > + Added support to skip RED using new action > + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` > Can you please move this entry alphabetically ordered by vendor to the block where has ethdev drivers, a little above in this section. > Removed Items > ------------- > diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h > index f0eab4244c..ba35873124 100644 > --- a/drivers/net/cnxk/cnxk_ethdev.h > +++ b/drivers/net/cnxk/cnxk_ethdev.h > @@ -168,6 +168,7 @@ struct policy_actions { > uint16_t queue; > uint32_t mtr_id; > struct action_rss *rss_desc; > + bool skip_red; > }; > }; > > diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c > index dcfa4223d5..27a6e4ef3d 100644 > --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c > +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c > @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, > if (action->type == RTE_FLOW_ACTION_TYPE_VOID) > supported[i] = true; > > + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + supported[i] = true; > + > if (!supported[i]) > return update_mtr_err(i, error, true); > } > @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, > fmp->actions[i].action_fate = > action->type; > } > + > + if (action->type == > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + fmp->actions[i].skip_red = true; > } > } > } > @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) > cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; > } > > +static void > +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, > + struct roc_nix_bpf_cfg *cfg) > +{ > + struct cnxk_mtr_policy_node *policy = mtr->policy; > + > + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || > + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || > + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { > + if (policy->actions[RTE_COLOR_GREEN].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_GREEN].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_YELLOW].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_YELLOW].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_RED].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_RED].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_RED; > + } > + } > +} > + > static void > nix_precolor_table_map(struct cnxk_meter_node *mtr, > struct roc_nix_bpf_precolor *tbl, > @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) > if (!mtr[i]->is_used) { > memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); > nix_mtr_config_map(mtr[i], &cfg); > + for (j = 0; j < mtr[i]->rq_num; j++) { > + rq = &dev->rqs[mtr[i]->rq_id[j]]; > + nix_mtr_config_red(mtr[i], rq, &cfg); > + } > rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, > lvl_map[mtr[i]->level], > &cfg); ^ permalink raw reply [flat|nested] 70+ messages in thread
* RE: [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer 2023-02-15 6:25 ` [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-15 9:45 ` Ferruh Yigit @ 2023-02-15 10:00 ` Nithin Kumar Dabilpuram 1 sibling, 0 replies; 70+ messages in thread From: Nithin Kumar Dabilpuram @ 2023-02-15 10:00 UTC (permalink / raw) To: Rakesh Kudurumalla, Kiran Kumar Kokkilagadda, Sunil Kumar Kori, Satha Koteswara Rao Kottidi Cc: Jerin Jacob Kollanukkaran, dev, Rakesh Kudurumalla Acked-by: Nithin Dabilpuram<ndabilpuram@marvell.com> > -----Original Message----- > From: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Sent: Wednesday, February 15, 2023 11:56 AM > To: Nithin Kumar Dabilpuram <ndabilpuram@marvell.com>; Kiran Kumar Kokkilagadda > <kirankumark@marvell.com>; Sunil Kumar Kori <skori@marvell.com>; Satha Koteswara Rao > Kottidi <skoteshwar@marvell.com> > Cc: Jerin Jacob Kollanukkaran <jerinj@marvell.com>; dev@dpdk.org; Rakesh Kudurumalla > <rkudurumalla@marvell.com> > Subject: [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer > > Dropping of packets is based on action configured > to meter.If both skip_red and drop actions are configured > then tail dropping in invoked else if only drop action is > configured then RED drop is invoked.This action is supported > only when RED is configured using rte_eth_cman_config_set() > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > --- > doc/guides/nics/features/cnxk.ini | 1 + > doc/guides/nics/features/cnxk_vf.ini | 1 + > doc/guides/rel_notes/release_23_03.rst | 4 +++ > drivers/net/cnxk/cnxk_ethdev.h | 1 + > drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++ > 5 files changed, 57 insertions(+) > > diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini > index f81628da77..7947c044bb 100644 > --- a/doc/guides/nics/features/cnxk.ini > +++ b/doc/guides/nics/features/cnxk.ini > @@ -94,4 +94,5 @@ queue = Y > represented_port = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y > diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini > index c4ee32a9ad..873e1dcc0a 100644 > --- a/doc/guides/nics/features/cnxk_vf.ini > +++ b/doc/guides/nics/features/cnxk_vf.ini > @@ -83,4 +83,5 @@ pf = Y > queue = Y > rss = Y > security = Y > +skip_cman = Y > vf = Y > diff --git a/doc/guides/rel_notes/release_23_03.rst > b/doc/guides/rel_notes/release_23_03.rst > index 1dd10e26db..9f53ded5fb 100644 > --- a/doc/guides/rel_notes/release_23_03.rst > +++ b/doc/guides/rel_notes/release_23_03.rst > @@ -98,6 +98,10 @@ New Features > ``rte_event_dev_config::nb_single_link_event_port_queues`` parameter > required for eth_rx, eth_tx, crypto and timer eventdev adapters. > > +* **Updated Marvell cnxk ethdev PMD.** > + > + Added support to skip RED using new action > + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` > > Removed Items > ------------- > diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h > index f0eab4244c..ba35873124 100644 > --- a/drivers/net/cnxk/cnxk_ethdev.h > +++ b/drivers/net/cnxk/cnxk_ethdev.h > @@ -168,6 +168,7 @@ struct policy_actions { > uint16_t queue; > uint32_t mtr_id; > struct action_rss *rss_desc; > + bool skip_red; > }; > }; > > diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c > index dcfa4223d5..27a6e4ef3d 100644 > --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c > +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c > @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, > if (action->type == RTE_FLOW_ACTION_TYPE_VOID) > supported[i] = true; > > + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + supported[i] = true; > + > if (!supported[i]) > return update_mtr_err(i, error, true); > } > @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, > fmp->actions[i].action_fate = > action->type; > } > + > + if (action->type == > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) > + fmp->actions[i].skip_red = true; > } > } > } > @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct > roc_nix_bpf_cfg *cfg) > cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; > } > > +static void > +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, > + struct roc_nix_bpf_cfg *cfg) > +{ > + struct cnxk_mtr_policy_node *policy = mtr->policy; > + > + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || > + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || > + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { > + if (policy->actions[RTE_COLOR_GREEN].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_GREEN].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_YELLOW].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_YELLOW].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = > + ROC_NIX_BPF_ACTION_RED; > + } > + if (policy->actions[RTE_COLOR_RED].action_fate == > + RTE_FLOW_ACTION_TYPE_DROP) { > + if (policy->actions[RTE_COLOR_RED].skip_red) > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_DROP; > + else > + cfg->action[ROC_NIX_BPF_COLOR_RED] = > + ROC_NIX_BPF_ACTION_RED; > + } > + } > +} > + > static void > nix_precolor_table_map(struct cnxk_meter_node *mtr, > struct roc_nix_bpf_precolor *tbl, > @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) > if (!mtr[i]->is_used) { > memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); > nix_mtr_config_map(mtr[i], &cfg); > + for (j = 0; j < mtr[i]->rq_num; j++) { > + rq = &dev->rqs[mtr[i]->rq_id[j]]; > + nix_mtr_config_red(mtr[i], rq, &cfg); > + } > rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, > lvl_map[mtr[i]->level], > &cfg); > -- > 2.25.1 ^ permalink raw reply [flat|nested] 70+ messages in thread
* Re: [PATCH v9 1/3] ethdev: skip congestion management configuration 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-15 9:43 ` Ferruh Yigit 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla 3 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-15 9:43 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev On 2/15/2023 6:25 AM, Rakesh Kudurumalla wrote: > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> > Acked-by: Jerin Jacob <jerinj@marvell.com> > --- > v9: Distributed doc updates to relevant patches > > doc/guides/nics/features/default.ini | 1 + > doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ > doc/guides/rel_notes/release_23_03.rst | 6 ++++++ > lib/ethdev/rte_flow.h | 11 +++++++++++ > 4 files changed, 40 insertions(+) > > diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini > index 510cc6679d..b93f053a2b 100644 > --- a/doc/guides/nics/features/default.ini > +++ b/doc/guides/nics/features/default.ini > @@ -198,6 +198,7 @@ set_tag = > set_tp_dst = > set_tp_src = > set_ttl = > +skip_cman = > vf = > vxlan_decap = > vxlan_encap = > diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst > index 3e6242803d..4efadcc87b 100644 > --- a/doc/guides/prog_guide/rte_flow.rst > +++ b/doc/guides/prog_guide/rte_flow.rst > @@ -1840,6 +1840,28 @@ Drop packets. > | no properties | > +---------------+ > > + > +Action: ``SKIP_CMAN`` > +^^^^^^^^^^^^^^^^^^^^^ > + > +Skip congestion management on received packets > + > +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx > + queue's congestion mechanism.Once applied packets congestion configuration > + is bypassed on that particular ethdev Rx queue for all packets directed > + to that receive queue > + > +.. _table_rte_flow_action_skip_cman: > + > +.. table:: SKIP_CMAN > + > + +---------------+ > + | Field | > + +===============+ > + | no properties | > + +---------------+ > + > + > Action: ``COUNT`` > ^^^^^^^^^^^^^^^^^ > > diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst > index bf90feba68..1dd10e26db 100644 > --- a/doc/guides/rel_notes/release_23_03.rst > +++ b/doc/guides/rel_notes/release_23_03.rst > @@ -60,6 +60,12 @@ New Features > Added lock annotations attributes so that clang can statically analyze lock > correctness. > > +* **Added support to skip congestion management configuration.** > + > + Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip > + congestion management processing based on per flow or packet color > + identified by rte_flow meter object > + Can you please rebase on next-net [1], there you will see other flow API entries? [1] https://dpdk.org/git/next/dpdk-next-net > * **Updated AMD axgbe driver.** > > * Added multi-process support. > diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h > index b60987db4b..6ab632df9d 100644 > --- a/lib/ethdev/rte_flow.h > +++ b/lib/ethdev/rte_flow.h > @@ -2736,6 +2736,17 @@ enum rte_flow_action_type { > * No associated configuration structure. > */ > RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, > + > + /** > + * Skip congestion management configuration > + * > + * Using rte_eth_cman_config_set() API the application > + * can configure ethdev Rx queue's congestion mechanism. > + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the > + * congestion configuration applied to the given ethdev Rx queue. > + * > + */ > + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, > }; > > /** ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v10 1/3] ethdev: skip congestion management configuration 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla ` (2 preceding siblings ...) 2023-02-15 9:43 ` [PATCH v9 1/3] ethdev: skip congestion management configuration Ferruh Yigit @ 2023-02-15 17:42 ` Rakesh Kudurumalla 2023-02-15 17:42 ` [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (2 more replies) 3 siblings, 3 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 17:42 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> Acked-by: Jerin Jacob <jerinj@marvell.com> --- v10: Rebased on top of dpdk-next-net doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ doc/guides/rel_notes/release_23_03.rst | 6 ++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 4 files changed, 40 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 976a020985..49e41b8002 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -200,6 +200,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 161c5dce53..76045386ba 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1832,6 +1832,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index ad23355e54..286cffab95 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -97,6 +97,12 @@ New Features ``rte_flow_configure``. * Added ``host_port_id`` to ``rte_flow_port_attr`` structure. +* **Added support to skip congestion management configuration.** + + * Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip + congestion management processing based on per flow or packet color + identified by rte_flow meter object + * **Updated AMD axgbe driver.** * Added multi-process support. diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index 6a51442b8c..069b2210f1 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2820,6 +2820,17 @@ enum rte_flow_action_type { * No associated configuration structure. */ RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, + + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, }; /** -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla @ 2023-02-15 17:42 ` Rakesh Kudurumalla 2023-02-16 12:19 ` Ferruh Yigit 2023-02-15 17:42 ` [PATCH v10 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla 2 siblings, 1 reply; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 17:42 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 63a0b36622..e4b17aa22d 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -488,6 +488,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1896,6 +1897,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5385,6 +5387,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd 2023-02-15 17:42 ` [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-16 12:19 ` Ferruh Yigit 0 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-16 12:19 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev On 2/15/2023 5:42 PM, Rakesh Kudurumalla wrote: > added support for testpmd application to accept > skip_cman action while configuring policy action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> A comment seems missed, please check v6 & v8 [1], can you please send a new version with it? [1] https://inbox.dpdk.org/dev/67c958be-09e7-5c70-9a59-d471daeab4e8@amd.com/ ^ permalink raw reply [flat|nested] 70+ messages in thread
* [PATCH v10 3/3] net/cnxk: skip red drop for ingress policer 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla 2023-02-15 17:42 ` [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-15 17:42 ` Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla 2 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-15 17:42 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Nithin Dabilpuram <ndabilpuram@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + doc/guides/rel_notes/release_23_03.rst | 5 +++ drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++ 5 files changed, 58 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index 286cffab95..51bffc13e5 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -116,6 +116,11 @@ New Features * Added support for meter options. * Added support for rte_flow meter action. +* **Updated Marvell cnxk ethdev PMD.** + + * Added support to skip RED using new action + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` + * **Updated Mellanox mlx5 driver.** * Added support for matching on ICMPv6 ID and sequence fields. diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v11 1/3] ethdev: skip congestion management configuration 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla 2023-02-15 17:42 ` [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-15 17:42 ` [PATCH v10 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-16 14:51 ` Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla ` (2 more replies) 2 siblings, 3 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-16 14:51 UTC (permalink / raw) To: Ferruh Yigit, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla Introduce new flow action to skip congestion management configuration This feature helps to skip the congestion management processing based on per flow or the packet color identified by rte_flow meter object. For example, If one Rx queue configured as RED congestion and application wants tobypass the RED congestion processing for all GREEN color packet can be expressed though RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Ori Kam <orika@nvidia.com> Acked-by: Jerin Jacob <jerinj@marvell.com> --- v11: Updated sample command to apply congestion configuration with dpdk-testpmd application doc/guides/nics/features/default.ini | 1 + doc/guides/prog_guide/rte_flow.rst | 22 ++++++++++++++++++++++ doc/guides/rel_notes/release_23_03.rst | 6 ++++++ lib/ethdev/rte_flow.h | 11 +++++++++++ 4 files changed, 40 insertions(+) diff --git a/doc/guides/nics/features/default.ini b/doc/guides/nics/features/default.ini index 976a020985..49e41b8002 100644 --- a/doc/guides/nics/features/default.ini +++ b/doc/guides/nics/features/default.ini @@ -200,6 +200,7 @@ set_tag = set_tp_dst = set_tp_src = set_ttl = +skip_cman = vf = vxlan_decap = vxlan_encap = diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst index 8ad01ef05c..6b412e8585 100644 --- a/doc/guides/prog_guide/rte_flow.rst +++ b/doc/guides/prog_guide/rte_flow.rst @@ -1832,6 +1832,28 @@ Drop packets. | no properties | +---------------+ + +Action: ``SKIP_CMAN`` +^^^^^^^^^^^^^^^^^^^^^ + +Skip congestion management on received packets + +- Using ``rte_eth_cman_config_set()``, application can configure ethdev Rx + queue's congestion mechanism.Once applied packets congestion configuration + is bypassed on that particular ethdev Rx queue for all packets directed + to that receive queue + +.. _table_rte_flow_action_skip_cman: + +.. table:: SKIP_CMAN + + +---------------+ + | Field | + +===============+ + | no properties | + +---------------+ + + Action: ``COUNT`` ^^^^^^^^^^^^^^^^^ diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index 2ca30b3b49..39085a19a1 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -98,6 +98,12 @@ New Features * Added ``host_port_id`` in ``rte_flow_port_attr`` structure to reference the port hosting the shared objects. +* **Added support to skip congestion management configuration.** + + * Added a new action ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` to skip + congestion management processing based on per flow or packet color + identified by rte_flow meter object + * **Updated AMD axgbe driver.** * Added multi-process support. diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h index fb984a1097..8b9c0f02c3 100644 --- a/lib/ethdev/rte_flow.h +++ b/lib/ethdev/rte_flow.h @@ -2820,6 +2820,17 @@ enum rte_flow_action_type { * No associated configuration structure. */ RTE_FLOW_ACTION_TYPE_SEND_TO_KERNEL, + + /** + * Skip congestion management configuration + * + * Using rte_eth_cman_config_set() API the application + * can configure ethdev Rx queue's congestion mechanism. + * Introducing RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action to skip the + * congestion configuration applied to the given ethdev Rx queue. + * + */ + RTE_FLOW_ACTION_TYPE_SKIP_CMAN, }; /** -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v11 2/3] app/testpmd: add skip cman support for testpmd 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla @ 2023-02-16 14:51 ` Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-16 16:18 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Ferruh Yigit 2 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-16 14:51 UTC (permalink / raw) To: Ori Kam, Aman Singh, Yuying Zhang Cc: jerinj, ndabilpuram, dev, Rakesh Kudurumalla added support for testpmd application to accept skip_cman action while configuring policy action Below is sample command given after configuring meter profile for which congestion configuration is applied for red and yellow color packets and bypassed for green color packets in dpdk-testpmd application add port meter policy 0 200 g_actions drop / skip_cman / end y_actions drop / end r_actions drop / end Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> --- app/test-pmd/cmdline_flow.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c index 63a0b36622..e4b17aa22d 100644 --- a/app/test-pmd/cmdline_flow.c +++ b/app/test-pmd/cmdline_flow.c @@ -488,6 +488,7 @@ enum index { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_JUMP_GROUP, ACTION_MARK, @@ -1896,6 +1897,7 @@ static const enum index next_action[] = { ACTION_END, ACTION_VOID, ACTION_PASSTHRU, + ACTION_SKIP_CMAN, ACTION_JUMP, ACTION_MARK, ACTION_FLAG, @@ -5385,6 +5387,13 @@ static const struct token token_list[] = { .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), .call = parse_vc, }, + [ACTION_SKIP_CMAN] = { + .name = "skip_cman", + .help = "bypass cman on received packets", + .priv = PRIV_ACTION(SKIP_CMAN, 0), + .next = NEXT(NEXT_ENTRY(ACTION_NEXT)), + .call = parse_vc, + }, [ACTION_JUMP] = { .name = "jump", .help = "redirect traffic to a given group", -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* [PATCH v11 3/3] net/cnxk: skip red drop for ingress policer 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla @ 2023-02-16 14:51 ` Rakesh Kudurumalla 2023-02-16 16:18 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Ferruh Yigit 2 siblings, 0 replies; 70+ messages in thread From: Rakesh Kudurumalla @ 2023-02-16 14:51 UTC (permalink / raw) To: Nithin Dabilpuram, Kiran Kumar K, Sunil Kumar Kori, Satha Rao Cc: jerinj, dev, Rakesh Kudurumalla Dropping of packets is based on action configured to meter.If both skip_red and drop actions are configured then tail dropping in invoked else if only drop action is configured then RED drop is invoked.This action is supported only when RED is configured using rte_eth_cman_config_set() Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> Acked-by: Nithin Dabilpuram <ndabilpuram@marvell.com> --- doc/guides/nics/features/cnxk.ini | 1 + doc/guides/nics/features/cnxk_vf.ini | 1 + doc/guides/rel_notes/release_23_03.rst | 5 +++ drivers/net/cnxk/cnxk_ethdev.h | 1 + drivers/net/cnxk/cnxk_ethdev_mtr.c | 50 ++++++++++++++++++++++++++ 5 files changed, 58 insertions(+) diff --git a/doc/guides/nics/features/cnxk.ini b/doc/guides/nics/features/cnxk.ini index f81628da77..7947c044bb 100644 --- a/doc/guides/nics/features/cnxk.ini +++ b/doc/guides/nics/features/cnxk.ini @@ -94,4 +94,5 @@ queue = Y represented_port = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/nics/features/cnxk_vf.ini b/doc/guides/nics/features/cnxk_vf.ini index c4ee32a9ad..873e1dcc0a 100644 --- a/doc/guides/nics/features/cnxk_vf.ini +++ b/doc/guides/nics/features/cnxk_vf.ini @@ -83,4 +83,5 @@ pf = Y queue = Y rss = Y security = Y +skip_cman = Y vf = Y diff --git a/doc/guides/rel_notes/release_23_03.rst b/doc/guides/rel_notes/release_23_03.rst index 39085a19a1..53d5b1416c 100644 --- a/doc/guides/rel_notes/release_23_03.rst +++ b/doc/guides/rel_notes/release_23_03.rst @@ -117,6 +117,11 @@ New Features * Added support for meter options. * Added support for rte_flow meter action. +* **Updated Marvell cnxk ethdev PMD.** + + * Added support to skip RED using new action + ``RTE_FLOW_ACTION_TYPE_SKIP_CMAN`` + * **Updated NVIDIA mlx5 driver.** * Added support for matching on ICMPv6 ID and sequence fields. diff --git a/drivers/net/cnxk/cnxk_ethdev.h b/drivers/net/cnxk/cnxk_ethdev.h index f0eab4244c..ba35873124 100644 --- a/drivers/net/cnxk/cnxk_ethdev.h +++ b/drivers/net/cnxk/cnxk_ethdev.h @@ -168,6 +168,7 @@ struct policy_actions { uint16_t queue; uint32_t mtr_id; struct action_rss *rss_desc; + bool skip_red; }; }; diff --git a/drivers/net/cnxk/cnxk_ethdev_mtr.c b/drivers/net/cnxk/cnxk_ethdev_mtr.c index dcfa4223d5..27a6e4ef3d 100644 --- a/drivers/net/cnxk/cnxk_ethdev_mtr.c +++ b/drivers/net/cnxk/cnxk_ethdev_mtr.c @@ -358,6 +358,9 @@ cnxk_nix_mtr_policy_validate(struct rte_eth_dev *dev, if (action->type == RTE_FLOW_ACTION_TYPE_VOID) supported[i] = true; + if (action->type == RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + supported[i] = true; + if (!supported[i]) return update_mtr_err(i, error, true); } @@ -397,6 +400,10 @@ cnxk_fill_policy_actions(struct cnxk_mtr_policy_node *fmp, fmp->actions[i].action_fate = action->type; } + + if (action->type == + RTE_FLOW_ACTION_TYPE_SKIP_CMAN) + fmp->actions[i].skip_red = true; } } } @@ -1306,6 +1313,45 @@ nix_mtr_config_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_cfg *cfg) cfg->action[ROC_NIX_BPF_COLOR_RED] = ROC_NIX_BPF_ACTION_DROP; } +static void +nix_mtr_config_red(struct cnxk_meter_node *mtr, struct roc_nix_rq *rq, + struct roc_nix_bpf_cfg *cfg) +{ + struct cnxk_mtr_policy_node *policy = mtr->policy; + + if ((rq->red_pass && rq->red_pass >= rq->red_drop) || + (rq->spb_red_pass && rq->spb_red_pass >= rq->spb_red_drop) || + (rq->xqe_red_pass && rq->xqe_red_pass >= rq->xqe_red_drop)) { + if (policy->actions[RTE_COLOR_GREEN].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_GREEN].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_GREEN] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_YELLOW].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_YELLOW].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_YELLOW] = + ROC_NIX_BPF_ACTION_RED; + } + if (policy->actions[RTE_COLOR_RED].action_fate == + RTE_FLOW_ACTION_TYPE_DROP) { + if (policy->actions[RTE_COLOR_RED].skip_red) + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_DROP; + else + cfg->action[ROC_NIX_BPF_COLOR_RED] = + ROC_NIX_BPF_ACTION_RED; + } + } +} + static void nix_precolor_table_map(struct cnxk_meter_node *mtr, struct roc_nix_bpf_precolor *tbl, @@ -1483,6 +1529,10 @@ nix_mtr_configure(struct rte_eth_dev *eth_dev, uint32_t id) if (!mtr[i]->is_used) { memset(&cfg, 0, sizeof(struct roc_nix_bpf_cfg)); nix_mtr_config_map(mtr[i], &cfg); + for (j = 0; j < mtr[i]->rq_num; j++) { + rq = &dev->rqs[mtr[i]->rq_id[j]]; + nix_mtr_config_red(mtr[i], rq, &cfg); + } rc = roc_nix_bpf_config(nix, mtr[i]->bpf_id, lvl_map[mtr[i]->level], &cfg); -- 2.25.1 ^ permalink raw reply related [flat|nested] 70+ messages in thread
* Re: [PATCH v11 1/3] ethdev: skip congestion management configuration 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla @ 2023-02-16 16:18 ` Ferruh Yigit 2 siblings, 0 replies; 70+ messages in thread From: Ferruh Yigit @ 2023-02-16 16:18 UTC (permalink / raw) To: Rakesh Kudurumalla, Ori Kam, Thomas Monjalon, Andrew Rybchenko Cc: jerinj, ndabilpuram, dev On 2/16/2023 2:51 PM, Rakesh Kudurumalla wrote: > Introduce new flow action to skip congestion management configuration > This feature helps to skip the congestion management processing > based on per flow or the packet color identified by > rte_flow meter object. For example, If one Rx queue configured as > RED congestion and application wants tobypass the RED congestion > processing for all GREEN color packet can be expressed though > RTE_FLOW_ACTION_TYPE_SKIP_CMAN flow action > > Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com> > Acked-by: Ori Kam <orika@nvidia.com> > Acked-by: Jerin Jacob <jerinj@marvell.com> Series applied to dpdk-next-net/main, thanks. ^ permalink raw reply [flat|nested] 70+ messages in thread
end of thread, other threads:[~2023-02-16 16:18 UTC | newest] Thread overview: 70+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2022-12-22 1:39 [PATCH 1/3] lib: dpdk spec to skip red for ingress policer Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 2/3] app/testpmd: add skip red for testpmd Rakesh Kudurumalla 2022-12-22 1:39 ` [PATCH 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2022-12-22 3:01 ` [PATCH 1/3] lib: dpdk spec to skip red " Stephen Hemminger 2022-12-22 5:27 ` Jerin Jacob 2022-12-26 17:00 ` Ori Kam 2023-01-10 6:42 ` Rakesh Kudurumalla 2023-01-18 8:09 ` Rakesh Kudurumalla 2023-01-23 13:13 ` Jerin Jacob 2023-01-26 15:13 ` Ori Kam 2023-01-27 6:23 ` Jerin Jacob 2023-02-01 17:49 ` Ori Kam 2023-02-01 18:37 ` Jerin Jacob 2023-02-02 7:31 ` Ori Kam 2023-02-06 3:31 ` Jerin Jacob 2023-01-25 7:32 ` [PATCH v2 1/3] lib: skip congestion management configuration Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-01-25 7:32 ` [PATCH v2 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-01-25 9:26 ` [PATCH v2 1/3] lib: skip congestion management configuration Jerin Jacob 2023-02-09 7:24 ` [PATCH v3 " Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-09 7:24 ` [PATCH v3 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-09 7:31 ` [PATCH v3 1/3] lib: skip congestion management configuration Jerin Jacob 2023-02-09 8:35 ` [PATCH v4 1/3] ethdev: " Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-09 8:35 ` [PATCH v4 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-09 17:14 ` Jerin Jacob 2023-02-09 14:51 ` [PATCH v4 1/3] ethdev: skip congestion management configuration Ori Kam 2023-02-10 8:14 ` [PATCH v5 " Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-10 8:14 ` [PATCH v5 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-10 8:16 ` [PATCH v5 1/3] ethdev: skip congestion management configuration Jerin Jacob 2023-02-10 8:26 ` [PATCH v6 " Rakesh Kudurumalla 2023-02-10 8:26 ` [PATCH v6 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-10 23:08 ` Ferruh Yigit 2023-02-14 13:12 ` Ferruh Yigit 2023-02-10 8:26 ` [PATCH v6 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-10 23:18 ` Ferruh Yigit 2023-02-10 9:27 ` [PATCH v6 1/3] ethdev: skip congestion management configuration Jerin Jacob 2023-02-10 23:10 ` Ferruh Yigit 2023-02-13 12:34 ` Ori Kam 2023-02-13 13:54 ` Ferruh Yigit 2023-02-13 15:44 ` Jerin Jacob 2023-02-13 15:53 ` Ferruh Yigit 2023-02-11 0:35 ` Ferruh Yigit 2023-02-11 5:16 ` Jerin Jacob 2023-02-12 9:00 ` [PATCH v7 " Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-12 9:00 ` [PATCH v7 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 2/4] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-14 13:21 ` Ferruh Yigit 2023-02-14 13:04 ` [PATCH v8 3/4] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-14 13:04 ` [PATCH v8 4/4] doc: update release notes for 23_03.rst Rakesh Kudurumalla 2023-02-14 13:19 ` Ferruh Yigit 2023-02-14 13:13 ` [PATCH v8 1/4] ethdev: skip congestion management configuration Ferruh Yigit 2023-02-15 6:25 ` [PATCH v9 1/3] " Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-15 6:25 ` [PATCH v9 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-15 9:45 ` Ferruh Yigit 2023-02-15 10:00 ` Nithin Kumar Dabilpuram 2023-02-15 9:43 ` [PATCH v9 1/3] ethdev: skip congestion management configuration Ferruh Yigit 2023-02-15 17:42 ` [PATCH v10 " Rakesh Kudurumalla 2023-02-15 17:42 ` [PATCH v10 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-16 12:19 ` Ferruh Yigit 2023-02-15 17:42 ` [PATCH v10 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 2/3] app/testpmd: add skip cman support for testpmd Rakesh Kudurumalla 2023-02-16 14:51 ` [PATCH v11 3/3] net/cnxk: skip red drop for ingress policer Rakesh Kudurumalla 2023-02-16 16:18 ` [PATCH v11 1/3] ethdev: skip congestion management configuration Ferruh Yigit
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.