From patchwork Thu Sep 17 14:32:34 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chris Mi X-Patchwork-Id: 1366179 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=openvswitch.org (client-ip=140.211.166.133; helo=hemlock.osuosl.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=nvidia.com Received: from hemlock.osuosl.org (smtp2.osuosl.org [140.211.166.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4BsfbG2FhXz9sRR for ; Fri, 18 Sep 2020 00:33:30 +1000 (AEST) Received: from localhost (localhost [127.0.0.1]) by hemlock.osuosl.org (Postfix) with ESMTP id D28A38764C; Thu, 17 Sep 2020 14:33:28 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from hemlock.osuosl.org ([127.0.0.1]) by localhost (.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id gXx+PC4liSy4; Thu, 17 Sep 2020 14:33:25 +0000 (UTC) Received: from lists.linuxfoundation.org (lf-lists.osuosl.org [140.211.9.56]) by hemlock.osuosl.org (Postfix) with ESMTP id 3597487520; Thu, 17 Sep 2020 14:33:25 +0000 (UTC) Received: from lf-lists.osuosl.org (localhost [127.0.0.1]) by lists.linuxfoundation.org (Postfix) with ESMTP id C7D07C089F; Thu, 17 Sep 2020 14:33:24 +0000 (UTC) X-Original-To: dev@openvswitch.org Delivered-To: ovs-dev@lists.linuxfoundation.org Received: from silver.osuosl.org (smtp3.osuosl.org [140.211.166.136]) by lists.linuxfoundation.org (Postfix) with ESMTP id ACCCEC0864 for ; Thu, 17 Sep 2020 14:33:23 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by silver.osuosl.org (Postfix) with ESMTP id 9A8612E205 for ; Thu, 17 Sep 2020 14:33:23 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from silver.osuosl.org ([127.0.0.1]) by localhost (.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1fSrffJOtjBF for ; Thu, 17 Sep 2020 14:33:00 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by silver.osuosl.org (Postfix) with ESMTP id 952B02E1A6 for ; Thu, 17 Sep 2020 14:32:53 +0000 (UTC) Received: from Internal Mail-Server by MTLPINE1 (envelope-from cmi@nvidia.com) with SMTP; 17 Sep 2020 17:32:48 +0300 Received: from dev-r630-04.mtbc.labs.mlnx (dev-r630-04.mtbc.labs.mlnx [10.75.205.14]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 08HEWZj8030984; Thu, 17 Sep 2020 17:32:48 +0300 From: Chris Mi To: dev@openvswitch.org Date: Thu, 17 Sep 2020 22:32:34 +0800 Message-Id: <20200917143234.16804-10-cmi@nvidia.com> X-Mailer: git-send-email 2.21.1 In-Reply-To: <20200917143234.16804-1-cmi@nvidia.com> References: <20200917143234.16804-1-cmi@nvidia.com> MIME-Version: 1.0 Cc: elibr@nvidia.com, Chris Mi Subject: [ovs-dev] [PATCH v3 9/9] netdev-offload-tc: Add offload support for sFlow X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: ovs-dev-bounces@openvswitch.org Sender: "dev" Create a unique group ID to map the sFlow info when offloading sFlow action to TC. When showing the offloaded datapath flows, translate the group ID from TC sample action to sFlow info using the mapping. Signed-off-by: Chris Mi Reviewed-by: Eli Britstein --- lib/netdev-offload-tc.c | 131 +++++++++++++++++++++++++++++++++++++--- lib/tc.c | 59 ++++++++++++++++++ lib/tc.h | 9 ++- 3 files changed, 188 insertions(+), 11 deletions(-) diff --git a/lib/netdev-offload-tc.c b/lib/netdev-offload-tc.c index 8b62b745a..55b53b235 100644 --- a/lib/netdev-offload-tc.c +++ b/lib/netdev-offload-tc.c @@ -981,6 +981,18 @@ parse_tc_flower_to_match(struct tc_flower *flower, action = flower->actions; for (i = 0; i < flower->action_count; i++, action++) { switch (action->type) { + case TC_ACT_SAMPLE: { + const struct gid_node *node; + + node = gid_find(action->sample.action_group_id); + if (!node) { + VLOG_ERR_RL(&error_rl, "gid node is NULL, gid: %d", + action->sample.action_group_id); + return ENOENT; + } + nl_msg_put(buf, node->sflow.sflow, node->sflow.sflow_len); + } + break; case TC_ACT_VLAN_POP: { nl_msg_put_flag(buf, OVS_ACTION_ATTR_POP_VLAN); } @@ -1660,6 +1672,74 @@ flower_match_to_tun_opt(struct tc_flower *flower, const struct flow_tnl *tnl, flower->mask.tunnel.metadata.present.len = tnl->metadata.present.len; } +static int +parse_userspace_userdata(const struct nlattr *actions, + struct dpif_sflow_attr *sflow_attr) +{ + const struct nlattr *nla; + unsigned int left; + + NL_NESTED_FOR_EACH_UNSAFE (nla, left, actions) { + if (nl_attr_type(nla) == OVS_USERSPACE_ATTR_USERDATA) { + sflow_attr->userdata = CONST_CAST(void *, nl_attr_get(nla)); + sflow_attr->userdata_len = nl_attr_get_size(nla); + return 0; + } + } + + VLOG_ERR_RL(&error_rl, "%s: no OVS_USERSPACE_ATTR_USERDATA attribute", + __func__); + return EINVAL; +} + +static int +parse_action_userspace(const struct nlattr *actions, + struct dpif_sflow_attr *sflow_attr) +{ + const struct nlattr *nla; + unsigned int left; + + NL_NESTED_FOR_EACH_UNSAFE (nla, left, actions) { + if (nl_attr_type(nla) == OVS_ACTION_ATTR_USERSPACE) { + return parse_userspace_userdata(nla, sflow_attr); + } + } + + VLOG_ERR_RL(&error_rl, "%s: no OVS_ACTION_ATTR_USERSPACE attribute", + __func__); + return EINVAL; +} + +static int +parse_sample_action(const struct nlattr *actions, + struct dpif_sflow_attr *sflow_attr, + struct tc_action *tc_action) +{ + const struct nlattr *nla; + unsigned int left; + int ret = EINVAL; + + sflow_attr->sflow = actions; + sflow_attr->sflow_len = actions->nla_len; + + NL_NESTED_FOR_EACH_UNSAFE (nla, left, actions) { + if (nl_attr_type(nla) == OVS_SAMPLE_ATTR_ACTIONS) { + ret = parse_action_userspace(nla, sflow_attr); + } else if (nl_attr_type(nla) == OVS_SAMPLE_ATTR_PROBABILITY) { + tc_action->type = TC_ACT_SAMPLE; + tc_action->sample.action_rate = UINT32_MAX / nl_attr_get_u32(nla); + } else { + return EINVAL; + } + } + + if (tc_action->sample.action_rate) { + return ret; + } + + return EINVAL; +} + static int netdev_tc_flow_put(struct netdev *netdev, struct match *match, struct nlattr *actions, size_t actions_len, @@ -1676,6 +1756,7 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, struct tc_action *action; bool recirc_act = false; uint32_t block_id = 0; + uint32_t group_id = 0; struct nlattr *nla; struct tcf_id id; uint32_t chain; @@ -1965,7 +2046,8 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, NL_ATTR_FOR_EACH(nla, left, actions, actions_len) { if (flower.action_count >= TCA_ACT_MAX_NUM) { VLOG_DBG_RL(&rl, "Can only support %d actions", TCA_ACT_MAX_NUM); - return EOPNOTSUPP; + err = EOPNOTSUPP; + goto out; } action = &flower.actions[flower.action_count]; if (nl_attr_type(nla) == OVS_ACTION_ATTR_OUTPUT) { @@ -1975,7 +2057,8 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, if (!outdev) { VLOG_DBG_RL(&rl, "Can't find netdev for output port %d", port); - return ENODEV; + err = ENODEV; + goto out; } action->out.ifindex_out = netdev_get_ifindex(outdev); action->out.ingress = is_internal_port(netdev_get_type(outdev)); @@ -2013,7 +2096,7 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, err = parse_put_flow_set_action(&flower, action, set, set_len); if (err) { - return err; + goto out; } if (action->type == TC_ACT_ENCAP) { action->encap.tp_dst = info->tp_dst_port; @@ -2026,7 +2109,7 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, err = parse_put_flow_set_masked_action(&flower, action, set, set_len, true); if (err) { - return err; + goto out; } } else if (nl_attr_type(nla) == OVS_ACTION_ATTR_CT) { const struct nlattr *ct = nl_attr_get(nla); @@ -2034,7 +2117,7 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, err = parse_put_flow_ct_action(&flower, action, ct, ct_len); if (err) { - return err; + goto out; } } else if (nl_attr_type(nla) == OVS_ACTION_ATTR_CT_CLEAR) { action->type = TC_ACT_CT; @@ -2053,17 +2136,25 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, struct dpif_sflow_attr sflow_attr; memset(&sflow_attr, 0, sizeof sflow_attr); - gid_alloc_ctx(&sflow_attr); + if (flower.tunnel) { + sflow_attr.tunnel = CONST_CAST(struct flow_tnl *, tnl); + } + parse_sample_action(nla, &sflow_attr, action); + group_id = gid_alloc_ctx(&sflow_attr); + action->sample.action_group_id = group_id; + flower.action_count++; } else { VLOG_DBG_RL(&rl, "unsupported put action type: %d", nl_attr_type(nla)); - return EOPNOTSUPP; + err = EOPNOTSUPP; + goto out; } } if ((chain || recirc_act) && !info->recirc_id_shared_with_tc) { VLOG_ERR_RL(&error_rl, "flow_put: recirc_id sharing not supported"); - return EOPNOTSUPP; + err = EOPNOTSUPP; + goto out; } if (get_ufid_tc_mapping(ufid, &id) == 0) { @@ -2075,20 +2166,30 @@ netdev_tc_flow_put(struct netdev *netdev, struct match *match, prio = get_prio_for_tc_flower(&flower); if (prio == 0) { VLOG_ERR_RL(&rl, "couldn't get tc prio: %s", ovs_strerror(ENOSPC)); - return ENOSPC; + err = ENOSPC; + goto out; } flower.act_cookie.data = ufid; flower.act_cookie.len = sizeof *ufid; block_id = get_block_id_from_netdev(netdev); - id = tc_make_tcf_id_chain(ifindex, block_id, chain, prio, hook); + id = tc_make_tcf_id_chain(ifindex, block_id, chain, prio, hook, group_id); err = tc_replace_flower(&id, &flower); if (!err) { if (stats) { memset(stats, 0, sizeof *stats); } add_ufid_tc_mapping(netdev, ufid, &id); + } else { + goto out; + } + + return 0; + +out: + if (group_id) { + gid_free(group_id); } return err; @@ -2298,6 +2399,15 @@ netdev_tc_init_flow_api(struct netdev *netdev) return 0; } +static const struct dpif_sflow_attr * +netdev_tc_sflow_attr_get(uint32_t gid) +{ + const struct gid_node *node; + + node = gid_find(gid); + return node ? &node->sflow : NULL; +} + const struct netdev_flow_api netdev_offload_tc = { .type = "linux_tc", .flow_flush = netdev_tc_flow_flush, @@ -2308,4 +2418,5 @@ const struct netdev_flow_api netdev_offload_tc = { .flow_get = netdev_tc_flow_get, .flow_del = netdev_tc_flow_del, .init_flow_api = netdev_tc_init_flow_api, + .sflow_attr_get = netdev_tc_sflow_attr_get, }; diff --git a/lib/tc.c b/lib/tc.c index 8761304c9..128d7f637 100644 --- a/lib/tc.c +++ b/lib/tc.c @@ -31,6 +31,7 @@ #include #include #include +#include #include #include #include @@ -1289,6 +1290,38 @@ nl_parse_act_gact(struct nlattr *options, struct tc_flower *flower) return 0; } +static const struct nl_policy sample_policy[] = { + [TCA_SAMPLE_PARMS] = { .type = NL_A_UNSPEC, + .min_len = sizeof(struct tc_sample), + .optional = false, }, + [TCA_SAMPLE_PSAMPLE_GROUP] = { .type = NL_A_U32, + .optional = false, }, + [TCA_SAMPLE_RATE] = { .type = NL_A_U32, + .optional = false, }, +}; + +static int +nl_parse_act_sample(struct nlattr *options, struct tc_flower *flower) +{ + struct nlattr *sample_attrs[ARRAY_SIZE(sample_policy)]; + struct tc_action *action; + + if (!nl_parse_nested(options, sample_policy, sample_attrs, + ARRAY_SIZE(sample_policy))) { + VLOG_ERR_RL(&error_rl, "failed to parse sample action options"); + return EPROTO; + } + + action = &flower->actions[flower->action_count++]; + action->type = TC_ACT_SAMPLE; + action->sample.action_group_id = + nl_attr_get_u32(sample_attrs[TCA_SAMPLE_PSAMPLE_GROUP]); + action->sample.action_rate = + nl_attr_get_u32(sample_attrs[TCA_SAMPLE_RATE]); + + return 0; +} + static const struct nl_policy mirred_policy[] = { [TCA_MIRRED_PARMS] = { .type = NL_A_UNSPEC, .min_len = sizeof(struct tc_mirred), @@ -1697,6 +1730,8 @@ nl_parse_single_action(struct nlattr *action, struct tc_flower *flower, /* Added for TC rule only (not in OvS rule) so ignore. */ } else if (!strcmp(act_kind, "ct")) { nl_parse_act_ct(act_options, flower); + } else if (!strcmp(act_kind, "sample")) { + nl_parse_act_sample(act_options, flower); } else { VLOG_ERR_RL(&error_rl, "unknown tc action kind: %s", act_kind); err = EINVAL; @@ -2292,6 +2327,23 @@ nl_msg_put_act_mirred(struct ofpbuf *request, int ifindex, int action, nl_msg_end_nested(request, offset); } +static void +nl_msg_put_act_sample(struct ofpbuf *request, uint32_t rate, uint32_t group_id) +{ + size_t offset; + + nl_msg_put_string(request, TCA_ACT_KIND, "sample"); + offset = nl_msg_start_nested(request, TCA_ACT_OPTIONS | NLA_F_NESTED); + { + struct tc_sample parm = { .action = TC_ACT_PIPE }; + + nl_msg_put_unspec(request, TCA_SAMPLE_PARMS, &parm, sizeof parm); + nl_msg_put_u32(request, TCA_SAMPLE_RATE, rate); + nl_msg_put_u32(request, TCA_SAMPLE_PSAMPLE_GROUP, group_id); + } + nl_msg_end_nested(request, offset); +} + static inline void nl_msg_put_act_cookie(struct ofpbuf *request, struct tc_cookie *ck) { if (ck->len) { @@ -2551,6 +2603,13 @@ nl_msg_put_flower_acts(struct ofpbuf *request, struct tc_flower *flower) nl_msg_end_nested(request, act_offset); } break; + case TC_ACT_SAMPLE: { + act_offset = nl_msg_start_nested(request, act_index++); + nl_msg_put_act_sample(request, action->sample.action_rate, + action->sample.action_group_id); + nl_msg_end_nested(request, act_offset); + } + break; case TC_ACT_OUTPUT: { if (!released && flower->tunnel) { act_offset = nl_msg_start_nested(request, act_index++); diff --git a/lib/tc.h b/lib/tc.h index cc2ad025d..143e225d1 100644 --- a/lib/tc.h +++ b/lib/tc.h @@ -171,6 +171,7 @@ enum tc_action_type { TC_ACT_MPLS_SET, TC_ACT_GOTO, TC_ACT_CT, + TC_ACT_SAMPLE, }; enum nat_type { @@ -253,6 +254,11 @@ struct tc_action { bool force; bool commit; } ct; + + struct { + uint32_t action_rate; + uint32_t action_group_id; + } sample; }; enum tc_action_type type; @@ -292,11 +298,12 @@ tc_make_tcf_id(int ifindex, uint32_t block_id, uint16_t prio, static inline struct tcf_id tc_make_tcf_id_chain(int ifindex, uint32_t block_id, uint32_t chain, - uint16_t prio, enum tc_qdisc_hook hook) + uint16_t prio, enum tc_qdisc_hook hook, uint32_t group_id) { struct tcf_id id = tc_make_tcf_id(ifindex, block_id, prio, hook); id.chain = chain; + id.sflow_group_id = group_id; return id; }