From patchwork Mon Sep 2 11:27:10 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ilya Maximets X-Patchwork-Id: 1156594 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=openvswitch.org (client-ip=140.211.169.12; helo=mail.linuxfoundation.org; envelope-from=ovs-dev-bounces@openvswitch.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=samsung.com Authentication-Results: ozlabs.org; dkim=fail reason="signature verification failed" (1024-bit key; unprotected) header.d=samsung.com header.i=@samsung.com header.b="WUBLGBLM"; dkim-atps=neutral Received: from mail.linuxfoundation.org (mail.linuxfoundation.org [140.211.169.12]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 46MSX3487Lz9s7T for ; Mon, 2 Sep 2019 21:28:51 +1000 (AEST) Received: from mail.linux-foundation.org (localhost [127.0.0.1]) by mail.linuxfoundation.org (Postfix) with ESMTP id 92C23DBC; Mon, 2 Sep 2019 11:27:28 +0000 (UTC) X-Original-To: ovs-dev@openvswitch.org Delivered-To: ovs-dev@mail.linuxfoundation.org Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id B5D74DB5 for ; Mon, 2 Sep 2019 11:27:27 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mailout1.w1.samsung.com (mailout1.w1.samsung.com [210.118.77.11]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id EA96E5D3 for ; Mon, 2 Sep 2019 11:27:26 +0000 (UTC) Received: from eucas1p1.samsung.com (unknown [182.198.249.206]) by mailout1.w1.samsung.com (KnoxPortal) with ESMTP id 20190902112725euoutp0180844f4a3aa5efb41e16622ce58a1337~Am4BJJvEh2557125571euoutp01x for ; Mon, 2 Sep 2019 11:27:25 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 mailout1.w1.samsung.com 20190902112725euoutp0180844f4a3aa5efb41e16622ce58a1337~Am4BJJvEh2557125571euoutp01x DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=samsung.com; s=mail20170921; t=1567423645; bh=At0oXA0yh178485tSIbzmQZcH39iuwmOp2euf5wcTkY=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=WUBLGBLMATuQnS2fW+cblGdzTSeFABB8SR2GDf9s48rJSDY/H3DVE76egA5BrAW4G 4z/qRkSo1akMe4fT2cJ4kUX/j5VP54ZlmDtA0MVHco/MtRE4clCCeXzr64FlLf527n swaTpx0QYwGD3zRurRsTGODZeQc5WQGVlbpjB8kU= Received: from eusmges1new.samsung.com (unknown [203.254.199.242]) by eucas1p1.samsung.com (KnoxPortal) with ESMTP id 20190902112725eucas1p11eff21fd0db6e6a0c058ced42f417fe4~Am4AofXfm3085430854eucas1p1x; Mon, 2 Sep 2019 11:27:25 +0000 (GMT) Received: from eucas1p2.samsung.com ( [182.198.249.207]) by eusmges1new.samsung.com (EUCPMTA) with SMTP id 20.A5.04469.C9CFC6D5; Mon, 2 Sep 2019 12:27:24 +0100 (BST) Received: from eusmtrp2.samsung.com (unknown [182.198.249.139]) by eucas1p1.samsung.com (KnoxPortal) with ESMTPA id 20190902112724eucas1p10be06a9fede425def829b9fe69094872~Am3-14xnj1961219612eucas1p1A; Mon, 2 Sep 2019 11:27:24 +0000 (GMT) Received: from eusmgms1.samsung.com (unknown [182.198.249.179]) by eusmtrp2.samsung.com (KnoxPortal) with ESMTP id 20190902112723eusmtrp2366d7cfdccc355fac6c1d7452902b957~Am3-n6A7N2554125541eusmtrp2c; Mon, 2 Sep 2019 11:27:23 +0000 (GMT) X-AuditID: cbfec7f2-569ff70000001175-93-5d6cfc9c34b9 Received: from eusmtip1.samsung.com ( [203.254.199.221]) by eusmgms1.samsung.com (EUCPMTA) with SMTP id DD.F9.04166.B9CFC6D5; Mon, 2 Sep 2019 12:27:23 +0100 (BST) Received: from imaximets.rnd.samsung.ru (unknown [106.109.129.180]) by eusmtip1.samsung.com (KnoxPortal) with ESMTPA id 20190902112723eusmtip10aa4557b41aacd146cd5e6a3a2034e56~Am3-FP_b92569025690eusmtip1h; Mon, 2 Sep 2019 11:27:23 +0000 (GMT) From: Ilya Maximets To: ovs-dev@openvswitch.org Date: Mon, 2 Sep 2019 14:27:10 +0300 Message-Id: <20190902112711.2919-3-i.maximets@samsung.com> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20190902112711.2919-1-i.maximets@samsung.com> X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFvrCIsWRmVeSWpSXmKPExsWy7djP87pz/uTEGqyfpmOxfUUXm8WV9p/s FhsfnmW1WHvoA7vF3E/PGS2uX+JxYPPYOesuu8fiPS+ZPJ7d/M/o8X7fVTaPvi2rGANYo7hs UlJzMstSi/TtErgy3kzfwVww2bbizPmzLA2M9/W6GDk4JARMJK6s9u9i5OIQEljBKLFvzitG COcLo8SsqQuZIJzPjBK/Fx9i7mLkBOtoX/aBHSKxnFFiWstHKOcHo8STtz/YQarYBHQkTq0+ wghiiwhIS7zufcMKUsQscJhRYlfvQzaQhLBAgMSbORtZQWwWAVWJL58vMYMcxStgJXHyFyPE NnmJ1RsOgG3mFLCW6Ni4kA1kjoTAezaJ189uMkM84SJxoScfol5Y4tXxLewQtozE/53zmSDs eon7LS8ZIXo7GCWmH/oHlbCX2PL6HDvIHGYBTYn1u/QhRjpKTN8QBGHySdx4KwhSzAxkTto2 HWopr0RHmxDEDBWJ3weXQ4NHSuLmu89QB3hIvOjeDg2dPkaJtv07WCYwys9C2LWAkXEVo3hq aXFuemqxYV5quV5xYm5xaV66XnJ+7iZGYFo4/e/4px2MXy8lHWIU4GBU4uHl+JQdK8SaWFZc mXuIUYKDWUmEN3RPTqwQb0piZVVqUX58UWlOavEhRmkOFiVx3mqGB9FCAumJJanZqakFqUUw WSYOTqkGRoPiJpmcllMKPGZ/0m7w7PObu2r7Gv3MS0IrbltLbt6p+OPImtp7V9dtkVhYx7Xn bsLpjQETEpdOKEm+bTsz9e89GZ0XW395vNH61+4aZq4kUNT0YZa/X47vXY7JPyYpO+48Wpgs OTPauvLrxBtvq5m22hcumfj4kH2K887J/luWeEvO7V63LkaJpTgj0VCLuag4EQA5RI6iBwMA AA== X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFrrBLMWRmVeSWpSXmKPExsVy+t/xu7qz/+TEGlyfZmmxfUUXm8WV9p/s FhsfnmW1WHvoA7vF3E/PGS2uX+JxYPPYOesuu8fiPS+ZPJ7d/M/o8X7fVTaPvi2rGANYo/Rs ivJLS1IVMvKLS2yVog0tjPQMLS30jEws9QyNzWOtjEyV9O1sUlJzMstSi/TtEvQy3kzfwVww 2bbizPmzLA2M9/W6GDk5JARMJNqXfWDvYuTiEBJYyijx9PMpZoiElMSPXxdYIWxhiT/Xutgg ir4xSjT/OMkCkmAT0JE4tfoII4gtIiAt8br3DStIEbPAcUaJVZ07mEASwgJ+EntWLgFrYBFQ lfjy+RLQBg4OXgEriZO/GCEWyEus3nAAbDGngLVEx8aFbCC2EFDJ9MVfGCcw8i1gZFjFKJJa WpybnltsqFecmFtcmpeul5yfu4kRGKrbjv3cvIPx0sbgQ4wCHIxKPLwcn7JjhVgTy4orcw8x SnAwK4nwhu7JiRXiTUmsrEotyo8vKs1JLT7EaAp000RmKdHkfGAc5ZXEG5oamltYGpobmxub WSiJ83YIHIwREkhPLEnNTk0tSC2C6WPi4JRqYIyWaLz0qXXzxkOLH3OulD6bvSPs8uF1t/xl swIYZD0KkoQOyPirWJnvPm23+pHcvhXsk7wWsL7yW8DasXTBluYNlt8+i6dNqHgmv9OaRTcv M55XN+y41eatGe7TV3+yEt1o3hSRyB14Tur4TI/3T15cOrjiD8+ZCXFbntVPNHntlDf7KNOU je5KLMUZiYZazEXFiQBTSwzuawIAAA== X-CMS-MailID: 20190902112724eucas1p10be06a9fede425def829b9fe69094872 X-Msg-Generator: CA X-RootMTR: 20190902112724eucas1p10be06a9fede425def829b9fe69094872 X-EPHeader: CA CMS-TYPE: 201P X-CMS-RootMailID: 20190902112724eucas1p10be06a9fede425def829b9fe69094872 References: <20190902112711.2919-1-i.maximets@samsung.com> X-Spam-Status: No, score=-7.0 required=5.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on smtp1.linux-foundation.org Cc: Ilya Maximets , David Marchand Subject: [ovs-dev] [PATCH v2 2/3] dpif-netdev-perf: Fix TSC frequency for non-DPDK case. X-BeenThere: ovs-dev@openvswitch.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Sender: ovs-dev-bounces@openvswitch.org Errors-To: ovs-dev-bounces@openvswitch.org Unlike 'rte_get_tsc_cycles()' which doesn't need any specific initialization, 'rte_get_tsc_hz()' could be used only after successfull call to 'rte_eal_init()'. 'rte_eal_init()' estimates the TSC frequency for later use by 'rte_get_tsc_hz()'. Fairly said, we're not allowed to use 'rte_get_tsc_cycles()' before initializing DPDK too, but it works this way for now and provides correct results. This patch provides TSC frequency estimation code that will be used in two cases: * DPDK is not compiled in, i.e. DPDK_NETDEV not defined. * DPDK compiled in but not initialized, i.e. other_config:dpdk-init=false This change is mostly useful for AF_XDP netdev support, i.e. allows to use dpif-netdev/pmd-perf-show command and various PMD perf metrics. Signed-off-by: Ilya Maximets Acked-by: William Tu Reviewed-by: David Marchand --- lib/dpdk-stub.c | 6 ++++ lib/dpdk.c | 6 ++++ lib/dpdk.h | 1 + lib/dpif-netdev-perf.c | 75 ++++++++++++++++++++++++++++++++---------- lib/dpif-netdev-perf.h | 2 ++ lib/dpif-netdev.c | 9 +++++ 6 files changed, 81 insertions(+), 18 deletions(-) diff --git a/lib/dpdk-stub.c b/lib/dpdk-stub.c index e55be5750..c332c217c 100644 --- a/lib/dpdk-stub.c +++ b/lib/dpdk-stub.c @@ -68,6 +68,12 @@ dpdk_per_port_memory(void) return false; } +bool +dpdk_available(void) +{ + return false; +} + void print_dpdk_version(void) { diff --git a/lib/dpdk.c b/lib/dpdk.c index f31e1580c..fc58de55a 100644 --- a/lib/dpdk.c +++ b/lib/dpdk.c @@ -518,6 +518,12 @@ dpdk_per_port_memory(void) return per_port_memory; } +bool +dpdk_available(void) +{ + return dpdk_initialized; +} + void dpdk_set_lcore_id(unsigned cpu) { diff --git a/lib/dpdk.h b/lib/dpdk.h index 7dab83775..736a64279 100644 --- a/lib/dpdk.h +++ b/lib/dpdk.h @@ -41,6 +41,7 @@ const char *dpdk_get_vhost_sock_dir(void); bool dpdk_vhost_iommu_enabled(void); bool dpdk_vhost_postcopy_enabled(void); bool dpdk_per_port_memory(void); +bool dpdk_available(void); void print_dpdk_version(void); void dpdk_status(const struct ovsrec_open_vswitch *); #endif /* dpdk.h */ diff --git a/lib/dpif-netdev-perf.c b/lib/dpif-netdev-perf.c index e7ed49e7e..baf90b0f4 100644 --- a/lib/dpif-netdev-perf.c +++ b/lib/dpif-netdev-perf.c @@ -17,9 +17,11 @@ #include #include +#include "dpdk.h" #include "dpif-netdev-perf.h" #include "openvswitch/dynamic-string.h" #include "openvswitch/vlog.h" +#include "ovs-numa.h" #include "ovs-thread.h" #include "timeval.h" @@ -43,21 +45,59 @@ uint64_t iter_cycle_threshold; static struct vlog_rate_limit latency_rl = VLOG_RATE_LIMIT_INIT(600, 600); -#ifdef DPDK_NETDEV -static uint64_t -get_tsc_hz(void) -{ - return rte_get_tsc_hz(); -} -#else -/* This function is only invoked from PMD threads which depend on DPDK. - * A dummy function is sufficient when building without DPDK_NETDEV. */ -static uint64_t -get_tsc_hz(void) +static uint64_t tsc_hz = 1; + +void +pmd_perf_estimate_tsc_frequency(void) { - return 1; -} +#ifdef DPDK_NETDEV + if (dpdk_available()) { + tsc_hz = rte_get_tsc_hz(); + } + if (tsc_hz > 1) { + VLOG_INFO("DPDK provided TSC frequency: %"PRIu64" KHz", tsc_hz / 1000); + return; + } #endif + struct ovs_numa_dump *affinity; + struct pmd_perf_stats s; + uint64_t start, stop; + + /* DPDK is not available or returned unreliable value. + * Trying to estimate. */ + affinity = ovs_numa_thread_getaffinity_dump(); + if (affinity) { + const struct ovs_numa_info_core *core; + + FOR_EACH_CORE_ON_DUMP (core, affinity) { + /* Setting affinity to a single core from the affinity mask to + * avoid re-scheduling to another core while sleeping. */ + ovs_numa_thread_setaffinity_core(core->core_id); + break; + } + } + + start = cycles_counter_update(&s); + /* Using xnanosleep as it's interrupt resistant. + * Sleeping only 100 ms to avoid holding the main thread for too long. */ + xnanosleep(1E8); + stop = cycles_counter_update(&s); + + if (affinity) { + /* Restoring previous affinity. */ + ovs_numa_thread_setaffinity_dump(affinity); + ovs_numa_dump_destroy(affinity); + } + + if (stop <= start) { + VLOG_WARN("TSC source is unreliable."); + tsc_hz = 1; + } else { + tsc_hz = (stop - start) * 10; + } + + VLOG_INFO("Estimated TSC frequency: %"PRIu64" KHz", tsc_hz / 1000); +} /* Histogram functions. */ @@ -170,7 +210,6 @@ pmd_perf_format_overall_stats(struct ds *str, struct pmd_perf_stats *s, double duration) { uint64_t stats[PMD_N_STATS]; - uint64_t tsc_hz = get_tsc_hz(); double us_per_cycle = 1000000.0 / tsc_hz; if (duration == 0) { @@ -555,7 +594,7 @@ pmd_perf_end_iteration(struct pmd_perf_stats *s, int rx_packets, cum_ms->timestamp = now; } /* Do the next check after 4 us (10K cycles at 2.5 GHz TSC clock). */ - s->next_check_tsc = cycles_counter_update(s) + get_tsc_hz() / 250000; + s->next_check_tsc = cycles_counter_update(s) + tsc_hz / 250000; } } @@ -585,7 +624,7 @@ pmd_perf_set_log_susp_iteration(struct pmd_perf_stats *s, " duration=%"PRIu64" us\n", s->log_reason, susp->timestamp, - (1000000L * susp->cycles) / get_tsc_hz()); + (1000000L * susp->cycles) / tsc_hz); new_end_it = history_add(s->iterations.idx, log_it_after + 1); new_range = history_sub(new_end_it, s->log_begin_it); @@ -615,7 +654,7 @@ pmd_perf_log_susp_iteration_neighborhood(struct pmd_perf_stats *s) " duration=%"PRIu64" us\n", s->log_reason, susp->timestamp, - (1000000L * susp->cycles) / get_tsc_hz()); + (1000000L * susp->cycles) / tsc_hz); pmd_perf_format_iteration_history(&log, s, range); VLOG_WARN_RL(&latency_rl, @@ -729,7 +768,7 @@ pmd_perf_log_set_cmd(struct unixctl_conn *conn, log_it_after = it_after; log_q_thr = q_thr; log_us_thr = us_thr; - iter_cycle_threshold = (log_us_thr * get_tsc_hz()) / 1000000L; + iter_cycle_threshold = (log_us_thr * tsc_hz) / 1000000L; unixctl_command_reply(conn, ""); } diff --git a/lib/dpif-netdev-perf.h b/lib/dpif-netdev-perf.h index 244813ffe..ce369375b 100644 --- a/lib/dpif-netdev-perf.h +++ b/lib/dpif-netdev-perf.h @@ -233,6 +233,8 @@ cycles_counter_get(struct pmd_perf_stats *s) return s->last_tsc; } +void pmd_perf_estimate_tsc_frequency(void); + /* A nestable timer for measuring execution time in TSC cycles. * * Usage: diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 75d85b2fd..17323696f 100644 --- a/lib/dpif-netdev.c +++ b/lib/dpif-netdev.c @@ -1517,9 +1517,18 @@ create_dp_netdev(const char *name, const struct dpif_class *class, struct dp_netdev **dpp) OVS_REQUIRES(dp_netdev_mutex) { + static struct ovsthread_once tsc_freq_check = OVSTHREAD_ONCE_INITIALIZER; struct dp_netdev *dp; int error; + /* Avoid estimating TSC frequency for dummy datapath to not slow down + * unit tests. */ + if (!dpif_netdev_class_is_dummy(class) + && ovsthread_once_start(&tsc_freq_check)) { + pmd_perf_estimate_tsc_frequency(); + ovsthread_once_done(&tsc_freq_check); + } + dp = xzalloc(sizeof *dp); shash_add(&dp_netdevs, name, dp);