From patchwork Thu Sep 2 08:26:40 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Anton Ivanov X-Patchwork-Id: 1523519 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=140.211.166.138; helo=smtp1.osuosl.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Received: from smtp1.osuosl.org (smtp1.osuosl.org [140.211.166.138]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4H0Yv85CSnz9sVw for ; Thu, 2 Sep 2021 18:27:16 +1000 (AEST) Received: from localhost (localhost [127.0.0.1]) by smtp1.osuosl.org (Postfix) with ESMTP id B4AAA8276B; Thu, 2 Sep 2021 08:27:11 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp1.osuosl.org ([127.0.0.1]) by localhost (smtp1.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id fim5kJ0Qz-hG; Thu, 2 Sep 2021 08:27:07 +0000 (UTC) Received: from lists.linuxfoundation.org (lf-lists.osuosl.org [IPv6:2605:bc80:3010:104::8cd3:938]) by smtp1.osuosl.org (Postfix) with ESMTPS id 52F0781B01; Thu, 2 Sep 2021 08:27:06 +0000 (UTC) Received: from lf-lists.osuosl.org (localhost [127.0.0.1]) by lists.linuxfoundation.org (Postfix) with ESMTP id AD9C3C0021; Thu, 2 Sep 2021 08:27:04 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@lists.linuxfoundation.org Received: from smtp2.osuosl.org (smtp2.osuosl.org [140.211.166.133]) by lists.linuxfoundation.org (Postfix) with ESMTP id 96670C001A for ; Thu, 2 Sep 2021 08:27:02 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by smtp2.osuosl.org (Postfix) with ESMTP id 86F7440102 for ; Thu, 2 Sep 2021 08:27:02 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp2.osuosl.org ([127.0.0.1]) by localhost (smtp2.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0TlY95YYNZHE for ; Thu, 2 Sep 2021 08:27:01 +0000 (UTC) X-Greylist: from auto-whitelisted by SQLgrey-1.8.0 Received: from www.kot-begemot.co.uk (ivanoab7.miniserver.com [37.128.132.42]) by smtp2.osuosl.org (Postfix) with ESMTPS id A8B954000B for ; Thu, 2 Sep 2021 08:27:01 +0000 (UTC) Received: from tun252.jain.kot-begemot.co.uk ([192.168.18.6] helo=jain.kot-begemot.co.uk) by www.kot-begemot.co.uk with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mLi3i-0003Ra-16 for ovs-dev@openvswitch.org; Thu, 02 Sep 2021 08:26:58 +0000 Received: from jain.kot-begemot.co.uk ([192.168.3.3]) by jain.kot-begemot.co.uk with esmtp (Exim 4.92) (envelope-from ) id 1mLi3Y-000128-QB; Thu, 02 Sep 2021 09:26:49 +0100 From: anton.ivanov@cambridgegreys.com To: ovs-dev@openvswitch.org Date: Thu, 2 Sep 2021 09:26:40 +0100 Message-Id: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> X-Mailer: git-send-email 2.20.1 MIME-Version: 1.0 X-Clacks-Overhead: GNU Terry Pratchett Cc: Anton Ivanov Subject: [ovs-dev] [OVN Patch v5 1/4] northd: Disable parallel processing for logical_dp_groups 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" From: Anton Ivanov Work on improving processing with dp_groups enabled has discovered that the locking mechanism presently in use is not reliable. Disabling parallel processing if dp_groups are enabled until the root cause is determined and fixed. Signed-off-by: Anton Ivanov --- northd/ovn-northd.c | 2 +- ovs | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c index af413aba4..8c653cf52 100644 --- a/northd/ovn-northd.c +++ b/northd/ovn-northd.c @@ -12861,7 +12861,7 @@ build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, } } - if (use_parallel_build) { + if (use_parallel_build && (!use_logical_dp_groups)) { struct hmap *lflow_segs; struct lswitch_flow_build_info *lsiv; int index; From patchwork Thu Sep 2 08:26:41 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Anton Ivanov X-Patchwork-Id: 1523518 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=140.211.166.138; helo=smtp1.osuosl.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Received: from smtp1.osuosl.org (smtp1.osuosl.org [140.211.166.138]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4H0Yv55fxcz9sVw for ; Thu, 2 Sep 2021 18:27:12 +1000 (AEST) Received: from localhost (localhost [127.0.0.1]) by smtp1.osuosl.org (Postfix) with ESMTP id CB48C817FF; Thu, 2 Sep 2021 08:27:08 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp1.osuosl.org ([127.0.0.1]) by localhost (smtp1.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id tJiVN0F57vYR; Thu, 2 Sep 2021 08:27:05 +0000 (UTC) Received: from lists.linuxfoundation.org (lf-lists.osuosl.org [IPv6:2605:bc80:3010:104::8cd3:938]) by smtp1.osuosl.org (Postfix) with ESMTPS id 1B2A0818A2; Thu, 2 Sep 2021 08:27:04 +0000 (UTC) Received: from lf-lists.osuosl.org (localhost [127.0.0.1]) by lists.linuxfoundation.org (Postfix) with ESMTP id E8597C001A; Thu, 2 Sep 2021 08:27:03 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@lists.linuxfoundation.org Received: from smtp1.osuosl.org (smtp1.osuosl.org [IPv6:2605:bc80:3010::138]) by lists.linuxfoundation.org (Postfix) with ESMTP id 940B4C000E for ; Thu, 2 Sep 2021 08:27:02 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by smtp1.osuosl.org (Postfix) with ESMTP id 7E434818A2 for ; Thu, 2 Sep 2021 08:27:02 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp1.osuosl.org ([127.0.0.1]) by localhost (smtp1.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id xEbtVshO_q-i for ; Thu, 2 Sep 2021 08:27:01 +0000 (UTC) X-Greylist: from auto-whitelisted by SQLgrey-1.8.0 Received: from www.kot-begemot.co.uk (ivanoab7.miniserver.com [37.128.132.42]) by smtp1.osuosl.org (Postfix) with ESMTPS id 9AFB1817FF for ; Thu, 2 Sep 2021 08:27:01 +0000 (UTC) Received: from tun252.jain.kot-begemot.co.uk ([192.168.18.6] helo=jain.kot-begemot.co.uk) by www.kot-begemot.co.uk with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mLi3j-0003Rb-29 for ovs-dev@openvswitch.org; Thu, 02 Sep 2021 08:26:59 +0000 Received: from jain.kot-begemot.co.uk ([192.168.3.3]) by jain.kot-begemot.co.uk with esmtp (Exim 4.92) (envelope-from ) id 1mLi3d-000128-B7; Thu, 02 Sep 2021 09:26:53 +0100 From: anton.ivanov@cambridgegreys.com To: ovs-dev@openvswitch.org Date: Thu, 2 Sep 2021 09:26:41 +0100 Message-Id: <20210902082643.3916-2-anton.ivanov@cambridgegreys.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> References: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> MIME-Version: 1.0 X-Clacks-Overhead: GNU Terry Pratchett Cc: Anton Ivanov Subject: [ovs-dev] [OVN Patch v5 2/4] northd: Resize the hash to correct parameters after build 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" From: Anton Ivanov Parallel builds may result in suboptimal hash bucket sizing. In the absense of dp-groups this does not matter as the hash is purely storage and not used for lookups during the build. Such a hash needs to be resized to a correct size at the end of the build to ensure that any lookups during the lflow reconcilliation phase are done as fast as possible. Signed-off-by: Anton Ivanov --- northd/ovn-northd.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c index 8c653cf52..71458ff4e 100644 --- a/northd/ovn-northd.c +++ b/northd/ovn-northd.c @@ -13057,6 +13057,11 @@ build_lflows(struct northd_context *ctx, struct hmap *datapaths, igmp_groups, meter_groups, lbs, bfd_connections); + /* Parallel build may result in a suboptimal hash. Resize the + * hash to a correct size before doing lookups */ + + hmap_expand(&lflows); + if (hmap_count(&lflows) > max_seen_lflow_size) { max_seen_lflow_size = hmap_count(&lflows); } From patchwork Thu Sep 2 08:26:42 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Anton Ivanov X-Patchwork-Id: 1523520 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=140.211.166.138; helo=smtp1.osuosl.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Received: from smtp1.osuosl.org (smtp1.osuosl.org [140.211.166.138]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4H0YvC5MrVz9sPf for ; Thu, 2 Sep 2021 18:27:19 +1000 (AEST) Received: from localhost (localhost [127.0.0.1]) by smtp1.osuosl.org (Postfix) with ESMTP id 9BCF5827A8; Thu, 2 Sep 2021 08:27:12 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp1.osuosl.org ([127.0.0.1]) by localhost (smtp1.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id cZfTNSV5i-iv; Thu, 2 Sep 2021 08:27:10 +0000 (UTC) Received: from lists.linuxfoundation.org (lf-lists.osuosl.org [140.211.9.56]) by smtp1.osuosl.org (Postfix) with ESMTPS id 5767682572; Thu, 2 Sep 2021 08:27:09 +0000 (UTC) Received: from lf-lists.osuosl.org (localhost [127.0.0.1]) by lists.linuxfoundation.org (Postfix) with ESMTP id 3EDEAC001A; Thu, 2 Sep 2021 08:27:09 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@lists.linuxfoundation.org Received: from smtp1.osuosl.org (smtp1.osuosl.org [140.211.166.138]) by lists.linuxfoundation.org (Postfix) with ESMTP id D9B58C000E for ; Thu, 2 Sep 2021 08:27:07 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by smtp1.osuosl.org (Postfix) with ESMTP id D163582572 for ; Thu, 2 Sep 2021 08:27:07 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp1.osuosl.org ([127.0.0.1]) by localhost (smtp1.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Bi9oQssb1nWU for ; Thu, 2 Sep 2021 08:27:05 +0000 (UTC) X-Greylist: from auto-whitelisted by SQLgrey-1.8.0 Received: from www.kot-begemot.co.uk (ivanoab7.miniserver.com [37.128.132.42]) by smtp1.osuosl.org (Postfix) with ESMTPS id 774878194C for ; Thu, 2 Sep 2021 08:27:05 +0000 (UTC) Received: from tun252.jain.kot-begemot.co.uk ([192.168.18.6] helo=jain.kot-begemot.co.uk) by www.kot-begemot.co.uk with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mLi3p-0003Ri-Bp for ovs-dev@openvswitch.org; Thu, 02 Sep 2021 08:27:04 +0000 Received: from jain.kot-begemot.co.uk ([192.168.3.3]) by jain.kot-begemot.co.uk with esmtp (Exim 4.92) (envelope-from ) id 1mLi3h-000128-SU; Thu, 02 Sep 2021 09:26:56 +0100 From: anton.ivanov@cambridgegreys.com To: ovs-dev@openvswitch.org Date: Thu, 2 Sep 2021 09:26:42 +0100 Message-Id: <20210902082643.3916-3-anton.ivanov@cambridgegreys.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> References: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> MIME-Version: 1.0 X-Clacks-Overhead: GNU Terry Pratchett Cc: Anton Ivanov Subject: [ovs-dev] [OVN Patch v5 3/4] northd: Optimize dp groups operations 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" From: Anton Ivanov Remove full hash walks to form lflow dp_groups and add them to the overall parallelizeable lflow build. Make processing of "with dp groups" and "without" in build_lflows independent to allow these to run in parallel after the updates to the parallel API have been merged. Signed-off-by: Anton Ivanov --- northd/ovn-northd.c | 537 +++++++++++++++++++++++++++----------------- 1 file changed, 332 insertions(+), 205 deletions(-) diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c index 71458ff4e..28af790bc 100644 --- a/northd/ovn-northd.c +++ b/northd/ovn-northd.c @@ -86,6 +86,11 @@ struct northd_state { bool paused; }; +struct lflow_state { + struct hmap single_od; + struct hmap multiple_od; +}; + static const char *ovnnb_db; static const char *ovnsb_db; static const char *unixctl_path; @@ -4304,8 +4309,15 @@ struct ovn_lflow { const char *where; }; -static void ovn_lflow_destroy(struct hmap *lflows, struct ovn_lflow *lflow); -static struct ovn_lflow *ovn_lflow_find(const struct hmap *lflows, +static void ovn_lflow_destroy(struct lflow_state *lflows, + struct ovn_lflow *lflow); +static struct ovn_lflow *do_ovn_lflow_find(const struct hmap *lflows, + const struct ovn_datapath *od, + enum ovn_stage stage, + uint16_t priority, const char *match, + const char *actions, + const char *ctrl_meter, uint32_t hash); +static struct ovn_lflow *ovn_lflow_find(const struct lflow_state *lflows, const struct ovn_datapath *od, enum ovn_stage stage, uint16_t priority, const char *match, @@ -4363,7 +4375,7 @@ static struct hashrow_locks lflow_locks; * Version to use when locking is required. */ static void -do_ovn_lflow_add(struct hmap *lflow_map, struct ovn_datapath *od, +do_ovn_lflow_add(struct lflow_state *lflow_map, struct ovn_datapath *od, uint32_t hash, enum ovn_stage stage, uint16_t priority, const char *match, const char *actions, const char *io_port, const struct ovsdb_idl_row *stage_hint, @@ -4374,10 +4386,32 @@ do_ovn_lflow_add(struct hmap *lflow_map, struct ovn_datapath *od, struct ovn_lflow *lflow; if (use_logical_dp_groups) { - old_lflow = ovn_lflow_find(lflow_map, NULL, stage, priority, match, - actions, ctrl_meter, hash); + old_lflow = do_ovn_lflow_find(&lflow_map->single_od, NULL, stage, + priority, match, + actions, ctrl_meter, hash); if (old_lflow) { hmapx_add(&old_lflow->od_group, od); + /* Found, different, od count went up. Move to multiple od. */ + if (hmapx_count(&old_lflow->od_group) > 1) { + hmap_remove(&lflow_map->single_od, &old_lflow->hmap_node); + if (use_parallel_build) { + hmap_insert_fast(&lflow_map->multiple_od, + &old_lflow->hmap_node, hash); + } else { + hmap_insert(&lflow_map->multiple_od, + &old_lflow->hmap_node, hash); + } + } + } else { + /* Not found, lookup in multiple od. */ + old_lflow = do_ovn_lflow_find(&lflow_map->multiple_od, NULL, + stage, priority, match, + actions, ctrl_meter, hash); + if (old_lflow) { + hmapx_add(&old_lflow->od_group, od); + } + } + if (old_lflow) { return; } } @@ -4392,16 +4426,19 @@ do_ovn_lflow_add(struct hmap *lflow_map, struct ovn_datapath *od, nullable_xstrdup(ctrl_meter), ovn_lflow_hint(stage_hint), where); hmapx_add(&lflow->od_group, od); + + /* Insert "fresh" lflows into single_od. */ + if (!use_parallel_build) { - hmap_insert(lflow_map, &lflow->hmap_node, hash); + hmap_insert(&lflow_map->single_od, &lflow->hmap_node, hash); } else { - hmap_insert_fast(lflow_map, &lflow->hmap_node, hash); + hmap_insert_fast(&lflow_map->single_od, &lflow->hmap_node, hash); } } /* Adds a row with the specified contents to the Logical_Flow table. */ static void -ovn_lflow_add_at(struct hmap *lflow_map, struct ovn_datapath *od, +ovn_lflow_add_at(struct lflow_state *lflow_map, struct ovn_datapath *od, enum ovn_stage stage, uint16_t priority, const char *match, const char *actions, const char *io_port, const char *ctrl_meter, @@ -4465,27 +4502,76 @@ ovn_lflow_add_at(struct hmap *lflow_map, struct ovn_datapath *od, ACTIONS, NULL, CTRL_METER, NULL) static struct ovn_lflow * -ovn_lflow_find(const struct hmap *lflows, const struct ovn_datapath *od, +do_ovn_lflow_find(const struct hmap *lflows, const struct ovn_datapath *od, enum ovn_stage stage, uint16_t priority, const char *match, const char *actions, const char *ctrl_meter, uint32_t hash) { struct ovn_lflow *lflow; - HMAP_FOR_EACH_WITH_HASH (lflow, hmap_node, hash, lflows) { - if (ovn_lflow_equal(lflow, od, stage, priority, match, actions, - ctrl_meter)) { - return lflow; + if (lflows) { + HMAP_FOR_EACH_WITH_HASH (lflow, hmap_node, hash, lflows) { + if (ovn_lflow_equal(lflow, od, stage, priority, match, actions, + ctrl_meter)) { + return lflow; + } } } return NULL; } +static struct ovn_lflow * +ovn_lflow_find(const struct lflow_state *lflows, const struct ovn_datapath *od, + enum ovn_stage stage, uint16_t priority, + const char *match, const char *actions, const char *ctrl_meter, + uint32_t hash) +{ + struct ovn_lflow *lflow = + do_ovn_lflow_find(&lflows->single_od, od, stage, + priority, match, actions, + ctrl_meter, hash); + if (!lflow) { + lflow = do_ovn_lflow_find(&lflows->multiple_od, od, stage, + priority, match, + actions, ctrl_meter, hash); + } + + return lflow; +} + +static inline bool +hmap_safe_remove(struct hmap *hmap, struct hmap_node *node) +{ + struct hmap_node **bucket = &hmap->buckets[node->hash & hmap->mask]; + while (*bucket != node && *bucket != NULL) { + bucket = &(*bucket)->next; + } + if (*bucket == node) { + *bucket = node->next; + hmap->n--; + return true; + } + return false; +} + +static void +remove_lflow_from_lflows(struct lflow_state *lflows, struct ovn_lflow *lflow) +{ + if (use_logical_dp_groups && use_parallel_build) { + lock_hash_row(&lflow_locks, lflow->hmap_node.hash); + } + if (!hmap_safe_remove(&lflows->multiple_od, &lflow->hmap_node)) { + hmap_remove(&lflows->single_od, &lflow->hmap_node); + } + if (use_logical_dp_groups && use_parallel_build) { + unlock_hash_row(&lflow_locks, lflow->hmap_node.hash); + } +} static void -ovn_lflow_destroy(struct hmap *lflows, struct ovn_lflow *lflow) +ovn_lflow_destroy(struct lflow_state *lflows, struct ovn_lflow *lflow) { if (lflow) { if (lflows) { - hmap_remove(lflows, &lflow->hmap_node); + remove_lflow_from_lflows(lflows, lflow); } hmapx_destroy(&lflow->od_group); free(lflow->match); @@ -4497,6 +4583,7 @@ ovn_lflow_destroy(struct hmap *lflows, struct ovn_lflow *lflow) } } + /* Appends port security constraints on L2 address field 'eth_addr_field' * (e.g. "eth.src" or "eth.dst") to 'match'. 'ps_addrs', with 'n_ps_addrs' * elements, is the collection of port_security constraints from an @@ -4624,7 +4711,7 @@ build_port_security_ipv6_flow( * - Priority 80 flow to drop ARP and IPv6 ND packets. */ static void -build_port_security_nd(struct ovn_port *op, struct hmap *lflows, +build_port_security_nd(struct ovn_port *op, struct lflow_state *lflows, const struct ovsdb_idl_row *stage_hint) { struct ds match = DS_EMPTY_INITIALIZER; @@ -4705,7 +4792,7 @@ build_port_security_nd(struct ovn_port *op, struct hmap *lflows, */ static void build_port_security_ip(enum ovn_pipeline pipeline, struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, const struct ovsdb_idl_row *stage_hint) { char *port_direction; @@ -5109,7 +5196,7 @@ ls_get_acl_flags(struct ovn_datapath *od) */ static void build_lswitch_input_port_sec_op( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *actions, struct ds *match) { @@ -5153,7 +5240,7 @@ build_lswitch_input_port_sec_op( */ static void build_lswitch_input_port_sec_od( - struct ovn_datapath *od, struct hmap *lflows) + struct ovn_datapath *od, struct lflow_state *lflows) { if (od->nbs) { @@ -5164,7 +5251,7 @@ build_lswitch_input_port_sec_od( static void build_lswitch_learn_fdb_op( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *actions, struct ds *match) { if (op->nbsp && !op->n_ps_addrs && !strcmp(op->nbsp->type, "") && @@ -5191,7 +5278,7 @@ build_lswitch_learn_fdb_op( static void build_lswitch_learn_fdb_od( - struct ovn_datapath *od, struct hmap *lflows) + struct ovn_datapath *od, struct lflow_state *lflows) { if (od->nbs) { @@ -5212,7 +5299,7 @@ build_lswitch_learn_fdb_od( */ static void build_lswitch_output_port_sec_op(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match, struct ds *actions) { @@ -5257,7 +5344,7 @@ build_lswitch_output_port_sec_op(struct ovn_port *op, * (priority 100). */ static void build_lswitch_output_port_sec_od(struct ovn_datapath *od, - struct hmap *lflows) + struct lflow_state *lflows) { if (od->nbs) { ovn_lflow_add(lflows, od, S_SWITCH_OUT_PORT_SEC_IP, 0, "1", "next;"); @@ -5269,7 +5356,7 @@ build_lswitch_output_port_sec_od(struct ovn_datapath *od, static void skip_port_from_conntrack(struct ovn_datapath *od, struct ovn_port *op, enum ovn_stage in_stage, enum ovn_stage out_stage, - uint16_t priority, struct hmap *lflows) + uint16_t priority, struct lflow_state *lflows) { /* Can't use ct() for router ports. Consider the following configuration: * lp1(10.0.0.2) on hostA--ls1--lr0--ls2--lp2(10.0.1.2) on hostB, For a @@ -5299,7 +5386,7 @@ skip_port_from_conntrack(struct ovn_datapath *od, struct ovn_port *op, static void build_stateless_filter(struct ovn_datapath *od, const struct nbrec_acl *acl, - struct hmap *lflows) + struct lflow_state *lflows) { if (!strcmp(acl->direction, "from-lport")) { ovn_lflow_add_with_hint(lflows, od, S_SWITCH_IN_PRE_ACL, @@ -5318,7 +5405,7 @@ build_stateless_filter(struct ovn_datapath *od, static void build_stateless_filters(struct ovn_datapath *od, struct hmap *port_groups, - struct hmap *lflows) + struct lflow_state *lflows) { for (size_t i = 0; i < od->nbs->n_acls; i++) { const struct nbrec_acl *acl = od->nbs->acls[i]; @@ -5342,7 +5429,7 @@ build_stateless_filters(struct ovn_datapath *od, struct hmap *port_groups, static void build_pre_acls(struct ovn_datapath *od, struct hmap *port_groups, - struct hmap *lflows) + struct lflow_state *lflows) { /* Ingress and Egress Pre-ACL Table (Priority 0): Packets are * allowed by default. */ @@ -5471,7 +5558,7 @@ ls_has_lb_vip(struct ovn_datapath *od) } static void -build_pre_lb(struct ovn_datapath *od, struct hmap *lflows, +build_pre_lb(struct ovn_datapath *od, struct lflow_state *lflows, struct hmap *lbs) { /* Do not send ND packets to conntrack */ @@ -5550,7 +5637,7 @@ build_pre_lb(struct ovn_datapath *od, struct hmap *lflows, } static void -build_pre_stateful(struct ovn_datapath *od, struct hmap *lflows) +build_pre_stateful(struct ovn_datapath *od, struct lflow_state *lflows) { /* Ingress and Egress pre-stateful Table (Priority 0): Packets are * allowed by default. */ @@ -5602,7 +5689,7 @@ build_pre_stateful(struct ovn_datapath *od, struct hmap *lflows) } static void -build_acl_hints(struct ovn_datapath *od, struct hmap *lflows) +build_acl_hints(struct ovn_datapath *od, struct lflow_state *lflows) { /* This stage builds hints for the IN/OUT_ACL stage. Based on various * combinations of ct flags packets may hit only a subset of the logical @@ -5775,7 +5862,7 @@ build_acl_log(struct ds *actions, const struct nbrec_acl *acl, } static void -build_reject_acl_rules(struct ovn_datapath *od, struct hmap *lflows, +build_reject_acl_rules(struct ovn_datapath *od, struct lflow_state *lflows, enum ovn_stage stage, struct nbrec_acl *acl, struct ds *extra_match, struct ds *extra_actions, const struct ovsdb_idl_row *stage_hint, @@ -5818,7 +5905,7 @@ build_reject_acl_rules(struct ovn_datapath *od, struct hmap *lflows, } static void -consider_acl(struct hmap *lflows, struct ovn_datapath *od, +consider_acl(struct lflow_state *lflows, struct ovn_datapath *od, struct nbrec_acl *acl, bool has_stateful, const struct shash *meter_groups, struct ds *match, struct ds *actions) @@ -6042,7 +6129,7 @@ build_port_group_lswitches(struct northd_context *ctx, struct hmap *pgs, } static void -build_acls(struct ovn_datapath *od, struct hmap *lflows, +build_acls(struct ovn_datapath *od, struct lflow_state *lflows, struct hmap *port_groups, const struct shash *meter_groups) { bool has_stateful = od->has_stateful_acl || od->has_lb_vip; @@ -6256,7 +6343,7 @@ build_acls(struct ovn_datapath *od, struct hmap *lflows, } static void -build_qos(struct ovn_datapath *od, struct hmap *lflows) { +build_qos(struct ovn_datapath *od, struct lflow_state *lflows) { struct ds action = DS_EMPTY_INITIALIZER; ovn_lflow_add(lflows, od, S_SWITCH_IN_QOS_MARK, 0, "1", "next;"); @@ -6317,7 +6404,7 @@ build_qos(struct ovn_datapath *od, struct hmap *lflows) { } static void -build_lb_rules(struct hmap *lflows, struct ovn_northd_lb *lb, +build_lb_rules(struct lflow_state *lflows, struct ovn_northd_lb *lb, struct ds *match, struct ds *action, struct shash *meter_groups) { @@ -6388,7 +6475,7 @@ build_lb_rules(struct hmap *lflows, struct ovn_northd_lb *lb, } static void -build_stateful(struct ovn_datapath *od, struct hmap *lflows) +build_stateful(struct ovn_datapath *od, struct lflow_state *lflows) { /* Ingress and Egress stateful Table (Priority 0): Packets are * allowed by default. */ @@ -6427,7 +6514,7 @@ build_stateful(struct ovn_datapath *od, struct hmap *lflows) } static void -build_lb_hairpin(struct ovn_datapath *od, struct hmap *lflows) +build_lb_hairpin(struct ovn_datapath *od, struct lflow_state *lflows) { /* Ingress Pre-Hairpin/Nat-Hairpin/Hairpin tabled (Priority 0). * Packets that don't need hairpinning should continue processing. @@ -6485,7 +6572,7 @@ build_lb_hairpin(struct ovn_datapath *od, struct hmap *lflows) /* Build logical flows for the forwarding groups */ static void -build_fwd_group_lflows(struct ovn_datapath *od, struct hmap *lflows) +build_fwd_group_lflows(struct ovn_datapath *od, struct lflow_state *lflows) { if (!(!od->nbs || !od->nbs->n_forwarding_groups)) { @@ -6698,7 +6785,7 @@ static void build_lswitch_rport_arp_req_self_orig_flow(struct ovn_port *op, uint32_t priority, struct ovn_datapath *od, - struct hmap *lflows) + struct lflow_state *lflows) { struct sset all_eth_addrs = SSET_INITIALIZER(&all_eth_addrs); struct ds eth_src = DS_EMPTY_INITIALIZER; @@ -6792,7 +6879,7 @@ arp_nd_ns_match(const char *ips, int addr_family, struct ds *match) static void build_lswitch_rport_arp_req_flow_for_reachable_ip(const char *ips, int addr_family, struct ovn_port *patch_op, struct ovn_datapath *od, - uint32_t priority, struct hmap *lflows, + uint32_t priority, struct lflow_state *lflows, const struct ovsdb_idl_row *stage_hint) { struct ds match = DS_EMPTY_INITIALIZER; @@ -6831,7 +6918,7 @@ build_lswitch_rport_arp_req_flow_for_reachable_ip(const char *ips, static void build_lswitch_rport_arp_req_flow_for_unreachable_ip(const char *ips, int addr_family, struct ovn_datapath *od, uint32_t priority, - struct hmap *lflows, const struct ovsdb_idl_row *stage_hint) + struct lflow_state *lflows, const struct ovsdb_idl_row *stage_hint) { struct ds match = DS_EMPTY_INITIALIZER; @@ -6856,7 +6943,7 @@ static void build_lswitch_rport_arp_req_flows(struct ovn_port *op, struct ovn_datapath *sw_od, struct ovn_port *sw_op, - struct hmap *lflows, + struct lflow_state *lflows, const struct ovsdb_idl_row *stage_hint) { if (!op || !op->nbrp) { @@ -6983,7 +7070,7 @@ build_dhcpv4_options_flows(struct ovn_port *op, struct lport_addresses *lsp_addrs, struct ovn_port *inport, bool is_external, struct shash *meter_groups, - struct hmap *lflows) + struct lflow_state *lflows) { struct ds match = DS_EMPTY_INITIALIZER; @@ -7076,7 +7163,7 @@ build_dhcpv6_options_flows(struct ovn_port *op, struct lport_addresses *lsp_addrs, struct ovn_port *inport, bool is_external, struct shash *meter_groups, - struct hmap *lflows) + struct lflow_state *lflows) { struct ds match = DS_EMPTY_INITIALIZER; @@ -7126,7 +7213,7 @@ build_dhcpv6_options_flows(struct ovn_port *op, static void build_drop_arp_nd_flows_for_unbound_router_ports(struct ovn_port *op, const struct ovn_port *port, - struct hmap *lflows) + struct lflow_state *lflows) { struct ds match = DS_EMPTY_INITIALIZER; @@ -7191,7 +7278,7 @@ is_vlan_transparent(const struct ovn_datapath *od) } static void -build_lswitch_flows(struct hmap *datapaths, struct hmap *lflows) +build_lswitch_flows(struct hmap *datapaths, struct lflow_state *lflows) { /* This flow table structure is documented in ovn-northd(8), so please * update ovn-northd.8.xml if you change anything. */ @@ -7226,7 +7313,7 @@ build_lswitch_flows(struct hmap *datapaths, struct hmap *lflows) static void build_lswitch_lflows_pre_acl_and_acl(struct ovn_datapath *od, struct hmap *port_groups, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups, struct hmap *lbs) { @@ -7249,7 +7336,7 @@ build_lswitch_lflows_pre_acl_and_acl(struct ovn_datapath *od, * 100). */ static void build_lswitch_lflows_admission_control(struct ovn_datapath *od, - struct hmap *lflows) + struct lflow_state *lflows) { if (od->nbs) { /* Logical VLANs not supported. */ @@ -7275,7 +7362,7 @@ build_lswitch_lflows_admission_control(struct ovn_datapath *od, static void build_lswitch_arp_nd_responder_skip_local(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match) { if (op->nbsp) { @@ -7295,7 +7382,7 @@ build_lswitch_arp_nd_responder_skip_local(struct ovn_port *op, * (priority 50). */ static void build_lswitch_arp_nd_responder_known_ips(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct hmap *ports, struct shash *meter_groups, struct ds *actions, @@ -7515,7 +7602,7 @@ build_lswitch_arp_nd_responder_known_ips(struct ovn_port *op, * (priority 0)*/ static void build_lswitch_arp_nd_responder_default(struct ovn_datapath *od, - struct hmap *lflows) + struct lflow_state *lflows) { if (od->nbs) { ovn_lflow_add(lflows, od, S_SWITCH_IN_ARP_ND_RSP, 0, "1", "next;"); @@ -7526,7 +7613,7 @@ build_lswitch_arp_nd_responder_default(struct ovn_datapath *od, * (priority 110)*/ static void build_lswitch_arp_nd_service_monitor(struct ovn_northd_lb *lb, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *actions, struct ds *match) { @@ -7574,7 +7661,7 @@ build_lswitch_arp_nd_service_monitor(struct ovn_northd_lb *lb, * priority 100 flows. */ static void build_lswitch_dhcp_options_and_response(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups) { if (op->nbsp) { @@ -7630,7 +7717,7 @@ build_lswitch_dhcp_options_and_response(struct ovn_port *op, * (priority 0). */ static void build_lswitch_dhcp_and_dns_defaults(struct ovn_datapath *od, - struct hmap *lflows) + struct lflow_state *lflows) { if (od->nbs) { ovn_lflow_add(lflows, od, S_SWITCH_IN_DHCP_OPTIONS, 0, "1", "next;"); @@ -7646,7 +7733,7 @@ build_lswitch_dhcp_and_dns_defaults(struct ovn_datapath *od, */ static void build_lswitch_dns_lookup_and_response(struct ovn_datapath *od, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups) { if (od->nbs && ls_has_dns_records(od->nbs)) { @@ -7675,7 +7762,7 @@ build_lswitch_dns_lookup_and_response(struct ovn_datapath *od, * binding the external ports. */ static void build_lswitch_external_port(struct ovn_port *op, - struct hmap *lflows) + struct lflow_state *lflows) { if (op->nbsp && lsp_is_external(op->nbsp)) { @@ -7690,7 +7777,7 @@ build_lswitch_external_port(struct ovn_port *op, * (priority 70 - 100). */ static void build_lswitch_destination_lookup_bmcast(struct ovn_datapath *od, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *actions, struct shash *meter_groups) { @@ -7782,7 +7869,7 @@ build_lswitch_destination_lookup_bmcast(struct ovn_datapath *od, * (priority 90). */ static void build_lswitch_ip_mcast_igmp_mld(struct ovn_igmp_group *igmp_group, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *actions, struct ds *match) { @@ -7859,7 +7946,7 @@ static struct ovs_mutex mcgroup_mutex = OVS_MUTEX_INITIALIZER; /* Ingress table 22: Destination lookup, unicast handling (priority 50), */ static void build_lswitch_ip_unicast_lookup(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct hmap *mcgroups, struct ds *actions, struct ds *match) @@ -8279,7 +8366,7 @@ get_outport_for_routing_policy_nexthop(struct ovn_datapath *od, } static void -build_routing_policy_flow(struct hmap *lflows, struct ovn_datapath *od, +build_routing_policy_flow(struct lflow_state *lflows, struct ovn_datapath *od, struct hmap *ports, const struct nbrec_logical_router_policy *rule, const struct ovsdb_idl_row *stage_hint) @@ -8344,8 +8431,8 @@ build_routing_policy_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -build_ecmp_routing_policy_flows(struct hmap *lflows, struct ovn_datapath *od, - struct hmap *ports, +build_ecmp_routing_policy_flows(struct lflow_state *lflows, + struct ovn_datapath *od, struct hmap *ports, const struct nbrec_logical_router_policy *rule, uint16_t ecmp_group_id) { @@ -8815,7 +8902,7 @@ find_static_route_outport(struct ovn_datapath *od, struct hmap *ports, } static void -add_ecmp_symmetric_reply_flows(struct hmap *lflows, +add_ecmp_symmetric_reply_flows(struct lflow_state *lflows, struct ovn_datapath *od, const char *port_ip, struct ovn_port *out_port, @@ -8895,7 +8982,7 @@ add_ecmp_symmetric_reply_flows(struct hmap *lflows, } static void -build_ecmp_route_flow(struct hmap *lflows, struct ovn_datapath *od, +build_ecmp_route_flow(struct lflow_state *lflows, struct ovn_datapath *od, struct hmap *ports, struct ecmp_groups_node *eg) { @@ -8978,7 +9065,7 @@ build_ecmp_route_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -add_route(struct hmap *lflows, struct ovn_datapath *od, +add_route(struct lflow_state *lflows, struct ovn_datapath *od, const struct ovn_port *op, const char *lrp_addr_s, const char *network_s, int plen, const char *gateway, bool is_src_route, const struct ovsdb_idl_row *stage_hint, @@ -9040,7 +9127,7 @@ add_route(struct hmap *lflows, struct ovn_datapath *od, } static void -build_static_route_flow(struct hmap *lflows, struct ovn_datapath *od, +build_static_route_flow(struct lflow_state *lflows, struct ovn_datapath *od, struct hmap *ports, const struct parsed_route *route_) { @@ -9135,7 +9222,7 @@ static void build_lrouter_nat_flows_for_lb(struct ovn_lb_vip *lb_vip, struct ovn_northd_lb *lb, struct ovn_northd_lb_vip *vips_nb, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match, struct ds *action, struct shash *meter_groups) { @@ -9345,7 +9432,8 @@ next: } static void -build_lswitch_flows_for_lb(struct ovn_northd_lb *lb, struct hmap *lflows, +build_lswitch_flows_for_lb(struct ovn_northd_lb *lb, + struct lflow_state *lflows, struct shash *meter_groups, struct ds *match, struct ds *action) { @@ -9396,7 +9484,7 @@ build_lswitch_flows_for_lb(struct ovn_northd_lb *lb, struct hmap *lflows, */ static void build_lrouter_defrag_flows_for_lb(struct ovn_northd_lb *lb, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match) { if (!lb->n_nb_lr) { @@ -9442,7 +9530,8 @@ build_lrouter_defrag_flows_for_lb(struct ovn_northd_lb *lb, } static void -build_lrouter_flows_for_lb(struct ovn_northd_lb *lb, struct hmap *lflows, +build_lrouter_flows_for_lb(struct ovn_northd_lb *lb, + struct lflow_state *lflows, struct shash *meter_groups, struct ds *match, struct ds *action) { @@ -9593,7 +9682,7 @@ lrouter_nat_is_stateless(const struct nbrec_nat *nat) */ static inline void lrouter_nat_add_ext_ip_match(struct ovn_datapath *od, - struct hmap *lflows, struct ds *match, + struct lflow_state *lflows, struct ds *match, const struct nbrec_nat *nat, bool is_v6, bool is_src, ovs_be32 mask) { @@ -9660,7 +9749,7 @@ build_lrouter_arp_flow(struct ovn_datapath *od, struct ovn_port *op, const char *ip_address, const char *eth_addr, struct ds *extra_match, bool drop, uint16_t priority, const struct ovsdb_idl_row *hint, - struct hmap *lflows) + struct lflow_state *lflows) { struct ds match = DS_EMPTY_INITIALIZER; struct ds actions = DS_EMPTY_INITIALIZER; @@ -9710,7 +9799,7 @@ build_lrouter_nd_flow(struct ovn_datapath *od, struct ovn_port *op, const char *sn_ip_address, const char *eth_addr, struct ds *extra_match, bool drop, uint16_t priority, const struct ovsdb_idl_row *hint, - struct hmap *lflows, struct shash *meter_groups) + struct lflow_state *lflows, struct shash *meter_groups) { struct ds match = DS_EMPTY_INITIALIZER; struct ds actions = DS_EMPTY_INITIALIZER; @@ -9764,7 +9853,7 @@ build_lrouter_nd_flow(struct ovn_datapath *od, struct ovn_port *op, static void build_lrouter_nat_arp_nd_flow(struct ovn_datapath *od, struct ovn_nat *nat_entry, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups) { struct lport_addresses *ext_addrs = &nat_entry->ext_addrs; @@ -9787,7 +9876,7 @@ build_lrouter_nat_arp_nd_flow(struct ovn_datapath *od, static void build_lrouter_port_nat_arp_nd_flow(struct ovn_port *op, struct ovn_nat *nat_entry, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups) { struct lport_addresses *ext_addrs = &nat_entry->ext_addrs; @@ -9854,7 +9943,7 @@ build_lrouter_port_nat_arp_nd_flow(struct ovn_port *op, static void build_lrouter_drop_own_dest(struct ovn_port *op, enum ovn_stage stage, uint16_t priority, bool drop_snat_ip, - struct hmap *lflows) + struct lflow_state *lflows) { struct ds match_ips = DS_EMPTY_INITIALIZER; @@ -9911,7 +10000,8 @@ build_lrouter_drop_own_dest(struct ovn_port *op, enum ovn_stage stage, } static void -build_lrouter_force_snat_flows(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_force_snat_flows(struct lflow_state *lflows, + struct ovn_datapath *od, const char *ip_version, const char *ip_addr, const char *context) { @@ -9938,7 +10028,7 @@ build_lrouter_force_snat_flows(struct hmap *lflows, struct ovn_datapath *od, static void build_lrouter_force_snat_flows_op(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (!op->nbrp || !op->peer || !op->od->lb_force_snat_router_ip) { @@ -10009,7 +10099,7 @@ build_lrouter_force_snat_flows_op(struct ovn_port *op, } static void -build_lrouter_bfd_flows(struct hmap *lflows, struct ovn_port *op, +build_lrouter_bfd_flows(struct lflow_state *lflows, struct ovn_port *op, struct shash *meter_groups) { if (!op->has_bfd) { @@ -10064,7 +10154,7 @@ build_lrouter_bfd_flows(struct hmap *lflows, struct ovn_port *op, */ static void build_adm_ctrl_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows) + struct ovn_datapath *od, struct lflow_state *lflows) { if (od->nbr) { /* Logical VLANs not supported. @@ -10083,7 +10173,7 @@ build_check_pkt_len_action_string(struct ovn_port *op, struct ds *actions); */ static void build_adm_ctrl_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (op->nbrp) { @@ -10135,7 +10225,7 @@ build_adm_ctrl_flows_for_lrouter_port( * lflows for logical routers. */ static void build_neigh_learning_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct ds *match, struct ds *actions, struct shash *meter_groups) { @@ -10239,7 +10329,7 @@ build_neigh_learning_flows_for_lrouter( * for logical router ports. */ static void build_neigh_learning_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (op->nbrp) { @@ -10302,7 +10392,7 @@ build_neigh_learning_flows_for_lrouter_port( * Adv (RA) options and response. */ static void build_ND_RA_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match, struct ds *actions, struct shash *meter_groups) { @@ -10433,7 +10523,8 @@ build_ND_RA_flows_for_lrouter_port( /* Logical router ingress table ND_RA_OPTIONS & ND_RA_RESPONSE: RS * responder, by default goto next. (priority 0). */ static void -build_ND_RA_flows_for_lrouter(struct ovn_datapath *od, struct hmap *lflows) +build_ND_RA_flows_for_lrouter(struct ovn_datapath *od, + struct lflow_state *lflows) { if (od->nbr) { ovn_lflow_add(lflows, od, S_ROUTER_IN_ND_RA_OPTIONS, 0, "1", "next;"); @@ -10459,7 +10550,7 @@ build_ND_RA_flows_for_lrouter(struct ovn_datapath *od, struct hmap *lflows) */ static void build_ip_routing_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *ports,struct hmap *lflows) + struct ovn_port *op, struct hmap *ports,struct lflow_state *lflows) { if (op->nbrp) { @@ -10506,7 +10597,7 @@ build_ip_routing_flows_for_lrouter_port( static void build_static_route_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct hmap *ports, struct hmap *bfd_connections) { if (od->nbr) { @@ -10560,7 +10651,7 @@ build_static_route_flows_for_lrouter( */ static void build_mcast_lookup_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (od->nbr) { @@ -10629,7 +10720,7 @@ build_mcast_lookup_flows_for_lrouter( * advances to the next table for ARP/ND resolution. */ static void build_ingress_policy_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct hmap *ports) { if (od->nbr) { @@ -10663,7 +10754,7 @@ build_ingress_policy_flows_for_lrouter( /* Local router ingress table ARP_RESOLVE: ARP Resolution. */ static void build_arp_resolve_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows) + struct ovn_datapath *od, struct lflow_state *lflows) { if (od->nbr) { /* Multicast packets already have the outport set so just advance to @@ -10680,7 +10771,8 @@ build_arp_resolve_flows_for_lrouter( } static void -routable_addresses_to_lflows(struct hmap *lflows, struct ovn_port *router_port, +routable_addresses_to_lflows(struct lflow_state *lflows, + struct ovn_port *router_port, struct ovn_port *peer, struct ds *match, struct ds *actions) { @@ -10721,7 +10813,7 @@ routable_addresses_to_lflows(struct hmap *lflows, struct ovn_port *router_port, */ static void build_arp_resolve_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct hmap *ports, struct ds *match, struct ds *actions) { @@ -11044,7 +11136,8 @@ build_arp_resolve_flows_for_lrouter_port( } static void -build_icmperr_pkt_big_flows(struct ovn_port *op, int mtu, struct hmap *lflows, +build_icmperr_pkt_big_flows(struct ovn_port *op, int mtu, + struct lflow_state *lflows, struct shash *meter_groups, struct ds *match, struct ds *actions, enum ovn_stage stage, struct ovn_port *outport) @@ -11137,7 +11230,8 @@ build_check_pkt_len_action_string(struct ovn_port *op, struct ds *actions) static void build_check_pkt_len_flows_for_lrp(struct ovn_port *op, - struct hmap *lflows, struct hmap *ports, + struct lflow_state *lflows, + struct hmap *ports, struct shash *meter_groups, struct ds *match, struct ds *actions) { @@ -11190,7 +11284,7 @@ build_check_pkt_len_flows_for_lrp(struct ovn_port *op, * */ static void build_check_pkt_len_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct hmap *ports, struct ds *match, struct ds *actions, struct shash *meter_groups) @@ -11225,7 +11319,7 @@ build_check_pkt_len_flows_for_lrouter( */ static void build_gateway_redirect_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (!od->nbr) { @@ -11263,7 +11357,7 @@ build_gateway_redirect_flows_for_lrouter( * and sends an ARP/IPv6 NA request (priority 100). */ static void build_arp_request_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows, + struct ovn_datapath *od, struct lflow_state *lflows, struct ds *match, struct ds *actions, struct shash *meter_groups) { @@ -11342,7 +11436,7 @@ build_arp_request_flows_for_lrouter( */ static void build_egress_delivery_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match, struct ds *actions) { if (op->nbrp) { @@ -11384,7 +11478,7 @@ build_egress_delivery_flows_for_lrouter_port( static void build_misc_local_traffic_drop_flows_for_lrouter( - struct ovn_datapath *od, struct hmap *lflows) + struct ovn_datapath *od, struct lflow_state *lflows) { if (od->nbr) { /* L3 admission control: drop multicast and broadcast source, localhost @@ -11439,7 +11533,7 @@ build_misc_local_traffic_drop_flows_for_lrouter( static void build_dhcpv6_reply_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match) { if (op->nbrp && (!op->l3dgw_port)) { @@ -11458,7 +11552,7 @@ build_dhcpv6_reply_flows_for_lrouter_port( static void build_ipv6_input_flows_for_lrouter_port( - struct ovn_port *op, struct hmap *lflows, + struct ovn_port *op, struct lflow_state *lflows, struct ds *match, struct ds *actions, struct shash *meter_groups) { @@ -11620,7 +11714,7 @@ build_ipv6_input_flows_for_lrouter_port( static void build_lrouter_arp_nd_for_datapath(struct ovn_datapath *od, - struct hmap *lflows, + struct lflow_state *lflows, struct shash *meter_groups) { if (od->nbr) { @@ -11670,7 +11764,7 @@ build_lrouter_arp_nd_for_datapath(struct ovn_datapath *od, /* Logical router ingress table 3: IP Input for IPv4. */ static void build_lrouter_ipv4_ip_input(struct ovn_port *op, - struct hmap *lflows, + struct lflow_state *lflows, struct ds *match, struct ds *actions, struct shash *meter_groups) { @@ -11958,7 +12052,8 @@ build_lrouter_ipv4_ip_input(struct ovn_port *op, } static void -build_lrouter_in_unsnat_flow(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_in_unsnat_flow(struct lflow_state *lflows, + struct ovn_datapath *od, const struct nbrec_nat *nat, struct ds *match, struct ds *actions, bool distributed, bool is_v6) { @@ -12021,7 +12116,7 @@ build_lrouter_in_unsnat_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -build_lrouter_in_dnat_flow(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_in_dnat_flow(struct lflow_state *lflows, struct ovn_datapath *od, const struct nbrec_nat *nat, struct ds *match, struct ds *actions, bool distributed, ovs_be32 mask, bool is_v6) @@ -12108,7 +12203,8 @@ build_lrouter_in_dnat_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -build_lrouter_out_undnat_flow(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_out_undnat_flow(struct lflow_state *lflows, + struct ovn_datapath *od, const struct nbrec_nat *nat, struct ds *match, struct ds *actions, bool distributed, struct eth_addr mac, bool is_v6) @@ -12155,7 +12251,8 @@ build_lrouter_out_undnat_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -build_lrouter_out_snat_flow(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_out_snat_flow(struct lflow_state *lflows, + struct ovn_datapath *od, const struct nbrec_nat *nat, struct ds *match, struct ds *actions, bool distributed, struct eth_addr mac, ovs_be32 mask, @@ -12248,7 +12345,7 @@ build_lrouter_out_snat_flow(struct hmap *lflows, struct ovn_datapath *od, } static void -build_lrouter_ingress_flow(struct hmap *lflows, struct ovn_datapath *od, +build_lrouter_ingress_flow(struct lflow_state *lflows, struct ovn_datapath *od, const struct nbrec_nat *nat, struct ds *match, struct ds *actions, struct eth_addr mac, bool distributed, bool is_v6) @@ -12384,7 +12481,8 @@ lrouter_check_nat_entry(struct ovn_datapath *od, const struct nbrec_nat *nat, /* NAT, Defrag and load balancing. */ static void -build_lrouter_nat_defrag_and_lb(struct ovn_datapath *od, struct hmap *lflows, +build_lrouter_nat_defrag_and_lb(struct ovn_datapath *od, + struct lflow_state *lflows, struct hmap *ports, struct ds *match, struct ds *actions) { @@ -12606,7 +12704,7 @@ struct lswitch_flow_build_info { struct hmap *datapaths; struct hmap *ports; struct hmap *port_groups; - struct hmap *lflows; + struct lflow_state *lflows; struct hmap *mcgroups; struct hmap *igmp_groups; struct shash *meter_groups; @@ -12827,12 +12925,6 @@ init_lflows_thread_pool(void) } } -/* TODO: replace hard cutoffs by configurable via commands. These are - * temporary defines to determine single-thread to multi-thread processing - * cutoff. - * Setting to 1 forces "all parallel" lflow build. - */ - static void noop_callback(struct worker_pool *pool OVS_UNUSED, void *fin_result OVS_UNUSED, @@ -12842,10 +12934,24 @@ noop_callback(struct worker_pool *pool OVS_UNUSED, /* Do nothing */ } +static void +lflow_merge_callback(struct worker_pool *pool OVS_UNUSED, + void *fin_result, + void *result_frags, + int index) +{ + struct lflow_state *result = (struct lflow_state *)fin_result; + struct lflow_state *res_frags = (struct lflow_state *)result_frags; + + fast_hmap_merge(&result->single_od, &res_frags[index].single_od); + hmap_destroy(&res_frags[index].single_od); +} + static void build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, - struct hmap *port_groups, struct hmap *lflows, + struct hmap *port_groups, + struct lflow_state *lflows, struct hmap *mcgroups, struct hmap *igmp_groups, struct shash *meter_groups, struct hmap *lbs, @@ -12862,7 +12968,7 @@ build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, } if (use_parallel_build && (!use_logical_dp_groups)) { - struct hmap *lflow_segs; + struct lflow_state *lflow_segs; struct lswitch_flow_build_info *lsiv; int index; @@ -12881,7 +12987,9 @@ build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, * on a per-bucket level instead of merging hash frags */ lsiv[index].lflows = lflows; } else { - fast_hmap_init(&lflow_segs[index], lflows->mask); + fast_hmap_init(&lflow_segs[index].single_od, + lflows->single_od.mask); + hmap_init(&lflow_segs[index].multiple_od); lsiv[index].lflows = &lflow_segs[index]; } @@ -12904,7 +13012,9 @@ build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, if (use_logical_dp_groups) { run_pool_callback(build_lflows_pool->pool, NULL, NULL, noop_callback); } else { - run_pool_hash(build_lflows_pool->pool, lflows, lflow_segs); + run_pool_callback(build_lflows_pool->pool, + &lflows->single_od, lflow_segs, + lflow_merge_callback); } for (index = 0; index < build_lflows_pool->pool->size; index++) { @@ -13036,6 +13146,57 @@ ovn_sb_set_lflow_logical_dp_group( static ssize_t max_seen_lflow_size = 128; +static void +reconcile_lflow(struct ovn_lflow *lflow, struct northd_context *ctx, + struct hmap *dp_groups, struct lflow_state *lflows) +{ + const struct sbrec_logical_flow *sbflow; + const char *pipeline = ovn_stage_get_pipeline_name(lflow->stage); + uint8_t table = ovn_stage_get_table(lflow->stage); + + sbflow = sbrec_logical_flow_insert(ctx->ovnsb_txn); + if (lflow->od) { + sbrec_logical_flow_set_logical_datapath(sbflow, lflow->od->sb); + } + ovn_sb_set_lflow_logical_dp_group(ctx, dp_groups, + sbflow, &lflow->od_group); + sbrec_logical_flow_set_pipeline(sbflow, pipeline); + sbrec_logical_flow_set_table_id(sbflow, table); + sbrec_logical_flow_set_priority(sbflow, lflow->priority); + sbrec_logical_flow_set_match(sbflow, lflow->match); + sbrec_logical_flow_set_actions(sbflow, lflow->actions); + if (lflow->io_port) { + struct smap tags = SMAP_INITIALIZER(&tags); + smap_add(&tags, "in_out_port", lflow->io_port); + sbrec_logical_flow_set_tags(sbflow, &tags); + smap_destroy(&tags); + } + sbrec_logical_flow_set_controller_meter(sbflow, lflow->ctrl_meter); + + /* Trim the source locator lflow->where, which looks something like + * "ovn/northd/ovn-northd.c:1234", down to just the part following the + * last slash, e.g. "ovn-northd.c:1234". */ + const char *slash = strrchr(lflow->where, '/'); +#if _WIN32 + const char *backslash = strrchr(lflow->where, '\\'); + if (!slash || backslash > slash) { + slash = backslash; + } +#endif + const char *where = slash ? slash + 1 : lflow->where; + + struct smap ids = SMAP_INITIALIZER(&ids); + smap_add(&ids, "stage-name", ovn_stage_to_str(lflow->stage)); + smap_add(&ids, "source", where); + if (lflow->stage_hint) { + smap_add(&ids, "stage-hint", lflow->stage_hint); + } + sbrec_logical_flow_set_external_ids(sbflow, &ids); + smap_destroy(&ids); + + ovn_lflow_destroy(lflows, lflow); +} + /* Updates the Logical_Flow and Multicast_Group tables in the OVN_SB database, * constructing their contents based on the OVN_NB database. */ static void @@ -13046,50 +13207,40 @@ build_lflows(struct northd_context *ctx, struct hmap *datapaths, struct hmap *lbs, struct hmap *bfd_connections, bool ovn_internal_version_changed) { - struct hmap lflows; + struct lflow_state lflows; - fast_hmap_size_for(&lflows, max_seen_lflow_size); - if (use_parallel_build) { - update_hashrow_locks(&lflows, &lflow_locks); + fast_hmap_size_for(&lflows.single_od, max_seen_lflow_size); + fast_hmap_size_for(&lflows.multiple_od, max_seen_lflow_size); + if (use_parallel_build && use_logical_dp_groups) { + update_hashrow_locks(&lflows.single_od, &lflow_locks); } + + build_lswitch_and_lrouter_flows(datapaths, ports, port_groups, &lflows, mcgroups, igmp_groups, meter_groups, lbs, bfd_connections); /* Parallel build may result in a suboptimal hash. Resize the - * hash to a correct size before doing lookups */ - - hmap_expand(&lflows); + * hash to a correct size before doing lookups. + * We need to do that only to the multiple_od hash. Single_od + * is recreated during processing, it will be the correct size + * at the point where reconciliation starts lookups. + */ - if (hmap_count(&lflows) > max_seen_lflow_size) { - max_seen_lflow_size = hmap_count(&lflows); - } + hmap_expand(&lflows.multiple_od); - /* Collecting all unique datapath groups. */ + /* Collecting all unique datapath groups. Works only with + * multiple_od hash and is not dependent on single_od. + * Can run in a separate thread after OVN updates to the + * newer version of the parallel API. + */ struct hmap dp_groups = HMAP_INITIALIZER(&dp_groups); - struct hmapx single_dp_lflows = HMAPX_INITIALIZER(&single_dp_lflows); - struct ovn_lflow *lflow; - HMAP_FOR_EACH (lflow, hmap_node, &lflows) { + struct ovn_lflow *lflow, *next_lflow; + + HMAP_FOR_EACH (lflow, hmap_node, &lflows.multiple_od) { uint32_t hash = hash_int(hmapx_count(&lflow->od_group), 0); struct ovn_dp_group *dpg; - - ovs_assert(hmapx_count(&lflow->od_group)); - - if (hmapx_count(&lflow->od_group) == 1) { - /* There is only one datapath, so it should be moved out of the - * group to a single 'od'. */ - const struct hmapx_node *node; - HMAPX_FOR_EACH (node, &lflow->od_group) { - lflow->od = node->data; - break; - } - hmapx_clear(&lflow->od_group); - /* Logical flow should be re-hashed later to allow lookups. */ - hmapx_add(&single_dp_lflows, lflow); - continue; - } - dpg = ovn_dp_group_find(&dp_groups, &lflow->od_group, hash); if (!dpg) { dpg = xzalloc(sizeof *dpg); @@ -13099,18 +13250,34 @@ build_lflows(struct northd_context *ctx, struct hmap *datapaths, } /* Adding datapath to the flow hash for logical flows that have only one, - * so they could be found by the southbound db record. */ - const struct hmapx_node *node; + * so they could be found by the southbound db record. + * Not dependent on multiple_od - same comment as above applies here. + */ + + struct hmap processed_single; + hmap_init(&processed_single); + uint32_t hash; - HMAPX_FOR_EACH (node, &single_dp_lflows) { - lflow = node->data; + struct hmapx_node *node; + HMAP_FOR_EACH_POP (lflow, hmap_node, &lflows.single_od) { + HMAPX_FOR_EACH (node, &lflow->od_group) { + lflow->od = node->data; + break; + } + hmapx_clear(&lflow->od_group); hash = hmap_node_hash(&lflow->hmap_node); - hmap_remove(&lflows, &lflow->hmap_node); hash = ovn_logical_flow_hash_datapath(&lflow->od->sb->header_.uuid, hash); - hmap_insert(&lflows, &lflow->hmap_node, hash); + hmap_insert(&processed_single, &lflow->hmap_node, hash); } - hmapx_destroy(&single_dp_lflows); + + hmap_destroy(&lflows.single_od); + + lflows.single_od = processed_single; + hmap_moved(&lflows.single_od); + + max_seen_lflow_size = MAX(hmap_count(&lflows.single_od), + hmap_count(&lflows.multiple_od)); /* Push changes to the Logical_Flow table to database. */ const struct sbrec_logical_flow *sbflow, *next_sbflow; @@ -13215,54 +13382,14 @@ build_lflows(struct northd_context *ctx, struct hmap *datapaths, free(od); } - struct ovn_lflow *next_lflow; - HMAP_FOR_EACH_SAFE (lflow, next_lflow, hmap_node, &lflows) { - const char *pipeline = ovn_stage_get_pipeline_name(lflow->stage); - uint8_t table = ovn_stage_get_table(lflow->stage); - - sbflow = sbrec_logical_flow_insert(ctx->ovnsb_txn); - if (lflow->od) { - sbrec_logical_flow_set_logical_datapath(sbflow, lflow->od->sb); - } - ovn_sb_set_lflow_logical_dp_group(ctx, &dp_groups, - sbflow, &lflow->od_group); - sbrec_logical_flow_set_pipeline(sbflow, pipeline); - sbrec_logical_flow_set_table_id(sbflow, table); - sbrec_logical_flow_set_priority(sbflow, lflow->priority); - sbrec_logical_flow_set_match(sbflow, lflow->match); - sbrec_logical_flow_set_actions(sbflow, lflow->actions); - if (lflow->io_port) { - struct smap tags = SMAP_INITIALIZER(&tags); - smap_add(&tags, "in_out_port", lflow->io_port); - sbrec_logical_flow_set_tags(sbflow, &tags); - smap_destroy(&tags); - } - sbrec_logical_flow_set_controller_meter(sbflow, lflow->ctrl_meter); - - /* Trim the source locator lflow->where, which looks something like - * "ovn/northd/ovn-northd.c:1234", down to just the part following the - * last slash, e.g. "ovn-northd.c:1234". */ - const char *slash = strrchr(lflow->where, '/'); -#if _WIN32 - const char *backslash = strrchr(lflow->where, '\\'); - if (!slash || backslash > slash) { - slash = backslash; - } -#endif - const char *where = slash ? slash + 1 : lflow->where; - - struct smap ids = SMAP_INITIALIZER(&ids); - smap_add(&ids, "stage-name", ovn_stage_to_str(lflow->stage)); - smap_add(&ids, "source", where); - if (lflow->stage_hint) { - smap_add(&ids, "stage-hint", lflow->stage_hint); - } - sbrec_logical_flow_set_external_ids(sbflow, &ids); - smap_destroy(&ids); - - ovn_lflow_destroy(&lflows, lflow); + HMAP_FOR_EACH_SAFE (lflow, next_lflow, hmap_node, &lflows.single_od) { + reconcile_lflow(lflow, ctx, &dp_groups, &lflows); + } + HMAP_FOR_EACH_SAFE (lflow, next_lflow, hmap_node, &lflows.multiple_od) { + reconcile_lflow(lflow, ctx, &dp_groups, &lflows); } - hmap_destroy(&lflows); + hmap_destroy(&lflows.single_od); + hmap_destroy(&lflows.multiple_od); struct ovn_dp_group *dpg; HMAP_FOR_EACH_POP (dpg, node, &dp_groups) { From patchwork Thu Sep 2 08:26:43 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Anton Ivanov X-Patchwork-Id: 1523521 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=2605:bc80:3010::137; helo=smtp4.osuosl.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Received: from smtp4.osuosl.org (smtp4.osuosl.org [IPv6:2605:bc80:3010::137]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 4H0YvJ6KJmz9sPf for ; Thu, 2 Sep 2021 18:27:24 +1000 (AEST) Received: from localhost (localhost [127.0.0.1]) by smtp4.osuosl.org (Postfix) with ESMTP id 3942B42510; Thu, 2 Sep 2021 08:27:22 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp4.osuosl.org ([127.0.0.1]) by localhost (smtp4.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Wu3sphMiMBkE; Thu, 2 Sep 2021 08:27:18 +0000 (UTC) Received: from lists.linuxfoundation.org (lf-lists.osuosl.org [140.211.9.56]) by smtp4.osuosl.org (Postfix) with ESMTPS id 5A1EE4252E; Thu, 2 Sep 2021 08:27:17 +0000 (UTC) Received: from lf-lists.osuosl.org (localhost [127.0.0.1]) by lists.linuxfoundation.org (Postfix) with ESMTP id 38963C000E; Thu, 2 Sep 2021 08:27:17 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@lists.linuxfoundation.org Received: from smtp2.osuosl.org (smtp2.osuosl.org [140.211.166.133]) by lists.linuxfoundation.org (Postfix) with ESMTP id 2C2FAC001A for ; Thu, 2 Sep 2021 08:27:16 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by smtp2.osuosl.org (Postfix) with ESMTP id E85E0406F3 for ; Thu, 2 Sep 2021 08:27:15 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from smtp2.osuosl.org ([127.0.0.1]) by localhost (smtp2.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0N_-WRe9i2Z7 for ; Thu, 2 Sep 2021 08:27:14 +0000 (UTC) X-Greylist: from auto-whitelisted by SQLgrey-1.8.0 Received: from www.kot-begemot.co.uk (ivanoab7.miniserver.com [37.128.132.42]) by smtp2.osuosl.org (Postfix) with ESMTPS id C2756406E0 for ; Thu, 2 Sep 2021 08:27:11 +0000 (UTC) Received: from tun252.jain.kot-begemot.co.uk ([192.168.18.6] helo=jain.kot-begemot.co.uk) by www.kot-begemot.co.uk with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mLi3t-0003Rs-Ql for ovs-dev@openvswitch.org; Thu, 02 Sep 2021 08:27:10 +0000 Received: from jain.kot-begemot.co.uk ([192.168.3.3]) by jain.kot-begemot.co.uk with esmtp (Exim 4.92) (envelope-from ) id 1mLi3k-000128-I9; Thu, 02 Sep 2021 09:27:01 +0100 From: anton.ivanov@cambridgegreys.com To: ovs-dev@openvswitch.org Date: Thu, 2 Sep 2021 09:26:43 +0100 Message-Id: <20210902082643.3916-4-anton.ivanov@cambridgegreys.com> X-Mailer: git-send-email 2.20.1 In-Reply-To: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> References: <20210902082643.3916-1-anton.ivanov@cambridgegreys.com> MIME-Version: 1.0 X-Clacks-Overhead: GNU Terry Pratchett Cc: Anton Ivanov Subject: [ovs-dev] [OVN Patch v5 4/4] northd: Restore parallel build with dp_groups 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" From: Anton Ivanov Restore parallel build with dp groups using rwlock instead of per row locking as an underlying mechanism. This provides improvement ~ 10% end-to-end on ovn-heater under virutalization despite awakening some qemu gremlin which makes qemu climb to silly CPU usage. The gain on bare metal is likely to be higher. Signed-off-by: Anton Ivanov --- northd/ovn-northd.c | 183 +++++++++++++++++++++++++++++++++----------- 1 file changed, 137 insertions(+), 46 deletions(-) diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c index 28af790bc..f5d49143d 100644 --- a/northd/ovn-northd.c +++ b/northd/ovn-northd.c @@ -59,6 +59,7 @@ #include "unixctl.h" #include "util.h" #include "uuid.h" +#include "ovs-thread.h" #include "openvswitch/vlog.h" VLOG_DEFINE_THIS_MODULE(ovn_northd); @@ -4369,7 +4370,26 @@ ovn_lflow_init(struct ovn_lflow *lflow, struct ovn_datapath *od, static bool use_logical_dp_groups = false; static bool use_parallel_build = true; -static struct hashrow_locks lflow_locks; +static struct ovs_rwlock flowtable_lock; + +static void ovn_make_multi_lflow(struct ovn_lflow *old_lflow, + struct ovn_datapath *od, + struct lflow_state *lflow_map, + uint32_t hash) +{ + hmapx_add(&old_lflow->od_group, od); + hmap_remove(&lflow_map->single_od, &old_lflow->hmap_node); + if (use_parallel_build) { + hmap_insert_fast(&lflow_map->multiple_od, &old_lflow->hmap_node, hash); + } else { + hmap_insert(&lflow_map->multiple_od, &old_lflow->hmap_node, hash); + } +} + +#ifdef __clang__ +#pragma clang diagnostic push +#pragma clang diagnostic ignored "-Wthread-safety" +#endif /* Adds a row with the specified contents to the Logical_Flow table. * Version to use when locking is required. @@ -4385,57 +4405,133 @@ do_ovn_lflow_add(struct lflow_state *lflow_map, struct ovn_datapath *od, struct ovn_lflow *old_lflow; struct ovn_lflow *lflow; + /* Fast Path. + * See if we can get away without writing - grab a rdlock and check + * if we can get away with as little work as possible. + */ + if (use_logical_dp_groups) { - old_lflow = do_ovn_lflow_find(&lflow_map->single_od, NULL, stage, - priority, match, + if (use_parallel_build) { + ovs_rwlock_rdlock(&flowtable_lock); + } + old_lflow = do_ovn_lflow_find(&lflow_map->single_od, + NULL, stage, priority, match, actions, ctrl_meter, hash); if (old_lflow) { - hmapx_add(&old_lflow->od_group, od); - /* Found, different, od count went up. Move to multiple od. */ - if (hmapx_count(&old_lflow->od_group) > 1) { - hmap_remove(&lflow_map->single_od, &old_lflow->hmap_node); + if (!hmapx_contains(&old_lflow->od_group, od)) { + /* od not in od_group, we need to add it and move to + * multiple. */ if (use_parallel_build) { - hmap_insert_fast(&lflow_map->multiple_od, - &old_lflow->hmap_node, hash); - } else { - hmap_insert(&lflow_map->multiple_od, - &old_lflow->hmap_node, hash); + /* Upgrade the lock to write, we are likely to + * modify data. */ + ovs_rwlock_unlock(&flowtable_lock); + ovs_rwlock_wrlock(&flowtable_lock); + + /* Check if someone got ahead of us and the flow is already + * in multiple. */ + if (!hmap_contains(&lflow_map->single_od, + &old_lflow->hmap_node)) { + /* Someone did get ahead of us, add the od to the + * group. */ + hmapx_add(&old_lflow->od_group, od); + goto done_update_unlock; + } } + ovn_make_multi_lflow(old_lflow, od, lflow_map, hash); + goto done_update_unlock; } - } else { - /* Not found, lookup in multiple od. */ + } + if (!old_lflow) { + /* Not found in single, lookup in multiple od. */ old_lflow = do_ovn_lflow_find(&lflow_map->multiple_od, NULL, stage, priority, match, actions, ctrl_meter, hash); if (old_lflow) { - hmapx_add(&old_lflow->od_group, od); + if (!hmapx_contains(&old_lflow->od_group, od)) { + if (use_parallel_build) { + /* Upgrade lock to write.*/ + ovs_rwlock_unlock(&flowtable_lock); + ovs_rwlock_wrlock(&flowtable_lock); + } + hmapx_add(&old_lflow->od_group, od); + } } } +done_update_unlock: + if (use_parallel_build) { + ovs_rwlock_unlock(&flowtable_lock); + } if (old_lflow) { return; } } - lflow = xmalloc(sizeof *lflow); - /* While adding new logical flows we're not setting single datapath, but - * collecting a group. 'od' will be updated later for all flows with only - * one datapath in a group, so it could be hashed correctly. */ - ovn_lflow_init(lflow, NULL, stage, priority, - xstrdup(match), xstrdup(actions), - io_port ? xstrdup(io_port) : NULL, - nullable_xstrdup(ctrl_meter), - ovn_lflow_hint(stage_hint), where); - hmapx_add(&lflow->od_group, od); - - /* Insert "fresh" lflows into single_od. */ + /* Slow Path. + * We could not get away with minimal mostly ro amount of work, + * lock with rw and try to do an insert (may end up repeating + * some of what we do for ro). */ + if (use_logical_dp_groups && use_parallel_build) { + ovs_rwlock_wrlock(&flowtable_lock); + } if (!use_parallel_build) { + lflow = xmalloc(sizeof *lflow); + /* While adding new logical flows we are not setting single datapath, + * but collecting a group. 'od' will be updated later for all flows + * with only one datapath in a group, so it could be hashed correctly. + */ + ovn_lflow_init(lflow, NULL, stage, priority, + xstrdup(match), xstrdup(actions), + io_port ? xstrdup(io_port) : NULL, + nullable_xstrdup(ctrl_meter), + ovn_lflow_hint(stage_hint), where); + hmapx_add(&lflow->od_group, od); hmap_insert(&lflow_map->single_od, &lflow->hmap_node, hash); } else { + if (use_logical_dp_groups) { + /* Search again in case someone else got here first. */ + old_lflow = do_ovn_lflow_find(&lflow_map->single_od, + NULL, stage, priority, match, + actions, ctrl_meter, hash); + if (old_lflow) { + if (!hmapx_contains(&old_lflow->od_group, od)) { + ovn_make_multi_lflow(old_lflow, od, lflow_map, hash); + } + goto done_add_unlock; + } + /* Unlikely, but possible, more than one thread got here + * ahead of us while we were wating to acquire a write lock. */ + old_lflow = do_ovn_lflow_find(&lflow_map->multiple_od, NULL, + stage, priority, match, + actions, ctrl_meter, hash); + if (old_lflow) { + hmapx_add(&old_lflow->od_group, od); + goto done_add_unlock; + } + } + lflow = xmalloc(sizeof *lflow); + /* While adding new logical flows we're not setting single datapath, + * but collecting a group. 'od' will be updated later for all + * flows with only * one datapath in a group, so it could be hashed + * correctly. */ + ovn_lflow_init(lflow, NULL, stage, priority, + xstrdup(match), xstrdup(actions), + io_port ? xstrdup(io_port) : NULL, + nullable_xstrdup(ctrl_meter), + ovn_lflow_hint(stage_hint), where); + hmapx_add(&lflow->od_group, od); hmap_insert_fast(&lflow_map->single_od, &lflow->hmap_node, hash); } +done_add_unlock: + if (use_logical_dp_groups && use_parallel_build) { + ovs_rwlock_unlock(&flowtable_lock); + } } +#ifdef __clang__ +#pragma clang diagnostic pop +#endif + /* Adds a row with the specified contents to the Logical_Flow table. */ static void ovn_lflow_add_at(struct lflow_state *lflow_map, struct ovn_datapath *od, @@ -4453,15 +4549,8 @@ ovn_lflow_add_at(struct lflow_state *lflow_map, struct ovn_datapath *od, priority, match, actions); - if (use_logical_dp_groups && use_parallel_build) { - lock_hash_row(&lflow_locks, hash); - do_ovn_lflow_add(lflow_map, od, hash, stage, priority, match, - actions, io_port, stage_hint, where, ctrl_meter); - unlock_hash_row(&lflow_locks, hash); - } else { - do_ovn_lflow_add(lflow_map, od, hash, stage, priority, match, - actions, io_port, stage_hint, where, ctrl_meter); - } + do_ovn_lflow_add(lflow_map, od, hash, stage, priority, match, + actions, io_port, stage_hint, where, ctrl_meter); } /* Adds a row with the specified contents to the Logical_Flow table. */ @@ -4555,15 +4644,9 @@ hmap_safe_remove(struct hmap *hmap, struct hmap_node *node) static void remove_lflow_from_lflows(struct lflow_state *lflows, struct ovn_lflow *lflow) { - if (use_logical_dp_groups && use_parallel_build) { - lock_hash_row(&lflow_locks, lflow->hmap_node.hash); - } if (!hmap_safe_remove(&lflows->multiple_od, &lflow->hmap_node)) { hmap_remove(&lflows->single_od, &lflow->hmap_node); } - if (use_logical_dp_groups && use_parallel_build) { - unlock_hash_row(&lflow_locks, lflow->hmap_node.hash); - } } static void @@ -12967,7 +13050,7 @@ build_lswitch_and_lrouter_flows(struct hmap *datapaths, struct hmap *ports, } } - if (use_parallel_build && (!use_logical_dp_groups)) { + if (use_parallel_build) { struct lflow_state *lflow_segs; struct lswitch_flow_build_info *lsiv; int index; @@ -13197,6 +13280,8 @@ reconcile_lflow(struct ovn_lflow *lflow, struct northd_context *ctx, ovn_lflow_destroy(lflows, lflow); } +static bool needs_init = true; + /* Updates the Logical_Flow and Multicast_Group tables in the OVN_SB database, * constructing their contents based on the OVN_NB database. */ static void @@ -13211,8 +13296,15 @@ build_lflows(struct northd_context *ctx, struct hmap *datapaths, fast_hmap_size_for(&lflows.single_od, max_seen_lflow_size); fast_hmap_size_for(&lflows.multiple_od, max_seen_lflow_size); - if (use_parallel_build && use_logical_dp_groups) { - update_hashrow_locks(&lflows.single_od, &lflow_locks); + if (use_parallel_build && use_logical_dp_groups && needs_init) { + ovs_rwlock_init(&flowtable_lock); + /* This happens on first run with parallel+dp_groups. + * db_run will re-read use_parallel_build from config and + * reset it. This way we get correct sizing for + * parallel + dp_groups by doing one single threaded run + * on the first iteration. */ + use_parallel_build = false; + needs_init = false; } @@ -15139,7 +15231,6 @@ main(int argc, char *argv[]) daemonize_complete(); - init_hash_row_locks(&lflow_locks); use_parallel_build = can_parallelize_hashes(false); /* We want to detect (almost) all changes to the ovn-nb db. */