From nobody Tue Apr 28 10:33:55 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 3C785C433EF for ; Wed, 1 Jun 2022 08:42:34 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1350912AbiFAImd (ORCPT ); Wed, 1 Jun 2022 04:42:33 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39844 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S238208AbiFAImK (ORCPT ); Wed, 1 Jun 2022 04:42:10 -0400 Received: from mail-pg1-x530.google.com (mail-pg1-x530.google.com [IPv6:2607:f8b0:4864:20::530]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id B21225D1AD for ; Wed, 1 Jun 2022 01:42:08 -0700 (PDT) Received: by mail-pg1-x530.google.com with SMTP id v15so1255364pgk.11 for ; Wed, 01 Jun 2022 01:42:08 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=bytedance-com.20210112.gappssmtp.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=WmU4UwvLkRpE2mZ/DCl6f6FWKY7wBCHG/pflgzby8H4=; b=tnrUJg8tieFOQvEjlXtGKXYXWuXqLJAo+VhEWExqX4USma159XIuswtwbhywuiLJnP 4plLXBt15ItqBPQYC/ECisE5UxukMCTPpT6w0TU8I0I6xxcIoHnxQ0FCVAv+EhP9sY7E YEYyM55DVA2SnI1YaE4Ck7YxIMjxeq5xsmskuhd0yHwlyfVc9TqEndA97oOvgrlxLCVy aXlGCNuEvMx/oh0ItPYVGIFUAdbK/fWIZYmPQMxxbyOZmMHbko1odbiAK9XPBn3tDtxj 2KNTC8Ab2oFGBcL9cbAN3uPxwm4fn4Y7vDFkiAfmeyhiV37bauR6awZ07WHQKDfyKFEX D3DA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=WmU4UwvLkRpE2mZ/DCl6f6FWKY7wBCHG/pflgzby8H4=; b=PoMAKWMJB5j+rdxJ2emKHLJ/qDNAcxUSb3+OL80Xi+vhYzybj5mp33X0GQIMXy4LD8 WMf3aPaCCKJKmxoWapd7vSIvOdMVGvRRcijFroiJ1o3DPSo7AlucOUHv8WZbRevlRZI9 3ygRcPu/WV5f15VaGXhMu7gHtBnzqtDYuOMDOziyJG5rmB+eqwd6kSs8RfOiXm7wAhL2 b5KPOPkYTrQfb0d8rrLUBCcf96Xwwk17EkqxxXWrhl2FRK9ttRsIJTvyEEd8rW9Bp9GQ 1fHTMeTLcsIl/5YTBARLw6ABxABSVTbDPdHdXHhOXBy5KAeMAQs/o7SiSBsZK9j1fIWC sC5w== X-Gm-Message-State: AOAM530F78BYbZadYKNmU4v9kXTCiJY8plvJG+X6x8KSFQJMZVlNyaDc I3wXquXeyzJXBaQn8p+c5STtRA== X-Google-Smtp-Source: ABdhPJy7VwL1Z+I53VMH/Thn03fwQznsYUkLVOLEpU9IWf1180K2Yo7WZWQIARLfIhjfCiCvXk63lw== X-Received: by 2002:a65:63d9:0:b0:374:6b38:c6b3 with SMTP id n25-20020a6563d9000000b003746b38c6b3mr56543153pgv.195.1654072928241; Wed, 01 Jun 2022 01:42:08 -0700 (PDT) Received: from C02F52LSML85.bytedance.net ([139.177.225.241]) by smtp.gmail.com with ESMTPSA id u11-20020a63d34b000000b003c14af505f6sm827952pgi.14.2022.06.01.01.42.02 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Wed, 01 Jun 2022 01:42:07 -0700 (PDT) From: Feng zhou To: ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org, kafai@fb.com, songliubraving@fb.com, yhs@fb.com, john.fastabend@gmail.com, kpsingh@kernel.org Cc: netdev@vger.kernel.org, bpf@vger.kernel.org, linux-kernel@vger.kernel.org, duanxiongchun@bytedance.com, songmuchun@bytedance.com, wangdongdong.6@bytedance.com, cong.wang@bytedance.com, zhouchengming@bytedance.com, zhoufeng.zf@bytedance.com Subject: [PATCH v4 1/2] bpf: avoid grabbing spin_locks of all cpus when no free elems Date: Wed, 1 Jun 2022 16:41:48 +0800 Message-Id: <20220601084149.13097-2-zhoufeng.zf@bytedance.com> X-Mailer: git-send-email 2.30.1 (Apple Git-130) In-Reply-To: <20220601084149.13097-1-zhoufeng.zf@bytedance.com> References: <20220601084149.13097-1-zhoufeng.zf@bytedance.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Type: text/plain; charset="utf-8" From: Feng Zhou This patch add is_empty in pcpu_freelist_head to check freelist having free or not. If having, grab spin_lock, or check next cpu's freelist. Before patch: hash_map performance ./map_perf_test 1 0:hash_map_perf pre-alloc 975345 events per sec 4:hash_map_perf pre-alloc 855367 events per sec 12:hash_map_perf pre-alloc 860862 events per sec 8:hash_map_perf pre-alloc 849561 events per sec 3:hash_map_perf pre-alloc 849074 events per sec 6:hash_map_perf pre-alloc 847120 events per sec 10:hash_map_perf pre-alloc 845047 events per sec 5:hash_map_perf pre-alloc 841266 events per sec 14:hash_map_perf pre-alloc 849740 events per sec 2:hash_map_perf pre-alloc 839598 events per sec 9:hash_map_perf pre-alloc 838695 events per sec 11:hash_map_perf pre-alloc 845390 events per sec 7:hash_map_perf pre-alloc 834865 events per sec 13:hash_map_perf pre-alloc 842619 events per sec 1:hash_map_perf pre-alloc 804231 events per sec 15:hash_map_perf pre-alloc 795314 events per sec hash_map the worst: no free ./map_perf_test 2048 6:worse hash_map_perf pre-alloc 28628 events per sec 5:worse hash_map_perf pre-alloc 28553 events per sec 11:worse hash_map_perf pre-alloc 28543 events per sec 3:worse hash_map_perf pre-alloc 28444 events per sec 1:worse hash_map_perf pre-alloc 28418 events per sec 7:worse hash_map_perf pre-alloc 28427 events per sec 13:worse hash_map_perf pre-alloc 28330 events per sec 14:worse hash_map_perf pre-alloc 28263 events per sec 9:worse hash_map_perf pre-alloc 28211 events per sec 15:worse hash_map_perf pre-alloc 28193 events per sec 12:worse hash_map_perf pre-alloc 28190 events per sec 10:worse hash_map_perf pre-alloc 28129 events per sec 8:worse hash_map_perf pre-alloc 28116 events per sec 4:worse hash_map_perf pre-alloc 27906 events per sec 2:worse hash_map_perf pre-alloc 27801 events per sec 0:worse hash_map_perf pre-alloc 27416 events per sec 3:worse hash_map_perf pre-alloc 28188 events per sec ftrace trace 0) | htab_map_update_elem() { 0) 0.198 us | migrate_disable(); 0) | _raw_spin_lock_irqsave() { 0) 0.157 us | preempt_count_add(); 0) 0.538 us | } 0) 0.260 us | lookup_elem_raw(); 0) | alloc_htab_elem() { 0) | __pcpu_freelist_pop() { 0) | _raw_spin_lock() { 0) 0.152 us | preempt_count_add(); 0) 0.352 us | native_queued_spin_lock_slowpath(); 0) 1.065 us | } | ... 0) | _raw_spin_unlock() { 0) 0.254 us | preempt_count_sub(); 0) 0.555 us | } 0) + 25.188 us | } 0) + 25.486 us | } 0) | _raw_spin_unlock_irqrestore() { 0) 0.155 us | preempt_count_sub(); 0) 0.454 us | } 0) 0.148 us | migrate_enable(); 0) + 28.439 us | } The test machine is 16C, trying to get spin_lock 17 times, in addition to 16c, there is an extralist. after patch: hash_map performance ./map_perf_test 1 0:hash_map_perf pre-alloc 969348 events per sec 10:hash_map_perf pre-alloc 906526 events per sec 11:hash_map_perf pre-alloc 904557 events per sec 9:hash_map_perf pre-alloc 902384 events per sec 15:hash_map_perf pre-alloc 912287 events per sec 14:hash_map_perf pre-alloc 905689 events per sec 12:hash_map_perf pre-alloc 903680 events per sec 13:hash_map_perf pre-alloc 902631 events per sec 8:hash_map_perf pre-alloc 875369 events per sec 4:hash_map_perf pre-alloc 862808 events per sec 1:hash_map_perf pre-alloc 857218 events per sec 2:hash_map_perf pre-alloc 852875 events per sec 5:hash_map_perf pre-alloc 846497 events per sec 6:hash_map_perf pre-alloc 828467 events per sec 3:hash_map_perf pre-alloc 812542 events per sec 7:hash_map_perf pre-alloc 805336 events per sec hash_map worst: no free ./map_perf_test 2048 7:worse hash_map_perf pre-alloc 391104 events per sec 4:worse hash_map_perf pre-alloc 388073 events per sec 5:worse hash_map_perf pre-alloc 387038 events per sec 1:worse hash_map_perf pre-alloc 386546 events per sec 0:worse hash_map_perf pre-alloc 384590 events per sec 11:worse hash_map_perf pre-alloc 379378 events per sec 10:worse hash_map_perf pre-alloc 375480 events per sec 12:worse hash_map_perf pre-alloc 372394 events per sec 6:worse hash_map_perf pre-alloc 367692 events per sec 3:worse hash_map_perf pre-alloc 363970 events per sec 9:worse hash_map_perf pre-alloc 364008 events per sec 8:worse hash_map_perf pre-alloc 363759 events per sec 2:worse hash_map_perf pre-alloc 360743 events per sec 14:worse hash_map_perf pre-alloc 361195 events per sec 13:worse hash_map_perf pre-alloc 360276 events per sec 15:worse hash_map_perf pre-alloc 360057 events per sec 0:worse hash_map_perf pre-alloc 378177 events per sec ftrace trace 0) | htab_map_update_elem() { 0) 0.317 us | migrate_disable(); 0) | _raw_spin_lock_irqsave() { 0) 0.260 us | preempt_count_add(); 0) 1.803 us | } 0) 0.276 us | lookup_elem_raw(); 0) | alloc_htab_elem() { 0) 0.586 us | __pcpu_freelist_pop(); 0) 0.945 us | } 0) | _raw_spin_unlock_irqrestore() { 0) 0.160 us | preempt_count_sub(); 0) 0.972 us | } 0) 0.657 us | migrate_enable(); 0) 8.669 us | } It can be seen that after adding this patch, the map performance is almost not degraded, and when free=3D0, first check is_empty instead of directly acquiring spin_lock. As for why to add is_empty instead of directly judging head->first, my understanding is this, head->first is frequently modified during updating map, which will lead to invalid other cpus's cache, and is_empty is after freelist having no free elems will be changed, the performance will be bett= er. Co-developed-by: Chengming Zhou Signed-off-by: Chengming Zhou Signed-off-by: Feng Zhou --- kernel/bpf/percpu_freelist.c | 28 +++++++++++++++++++++++++--- kernel/bpf/percpu_freelist.h | 1 + 2 files changed, 26 insertions(+), 3 deletions(-) diff --git a/kernel/bpf/percpu_freelist.c b/kernel/bpf/percpu_freelist.c index 3d897de89061..4d55a81ba896 100644 --- a/kernel/bpf/percpu_freelist.c +++ b/kernel/bpf/percpu_freelist.c @@ -16,9 +16,11 @@ int pcpu_freelist_init(struct pcpu_freelist *s) =20 raw_spin_lock_init(&head->lock); head->first =3D NULL; + head->is_empty =3D true; } raw_spin_lock_init(&s->extralist.lock); s->extralist.first =3D NULL; + s->extralist.is_empty =3D true; return 0; } =20 @@ -32,6 +34,8 @@ static inline void pcpu_freelist_push_node(struct pcpu_fr= eelist_head *head, { node->next =3D head->first; head->first =3D node; + if (head->is_empty) + WRITE_ONCE(head->is_empty, false); } =20 static inline void ___pcpu_freelist_push(struct pcpu_freelist_head *head, @@ -130,14 +134,19 @@ static struct pcpu_freelist_node *___pcpu_freelist_po= p(struct pcpu_freelist *s) orig_cpu =3D cpu =3D raw_smp_processor_id(); while (1) { head =3D per_cpu_ptr(s->freelist, cpu); + if (READ_ONCE(head->is_empty)) + goto next_cpu; raw_spin_lock(&head->lock); node =3D head->first; if (node) { head->first =3D node->next; + if (!head->first) + WRITE_ONCE(head->is_empty, true); raw_spin_unlock(&head->lock); return node; } raw_spin_unlock(&head->lock); +next_cpu: cpu =3D cpumask_next(cpu, cpu_possible_mask); if (cpu >=3D nr_cpu_ids) cpu =3D 0; @@ -146,10 +155,15 @@ static struct pcpu_freelist_node *___pcpu_freelist_po= p(struct pcpu_freelist *s) } =20 /* per cpu lists are all empty, try extralist */ + if (READ_ONCE(s->extralist.is_empty)) + return NULL; raw_spin_lock(&s->extralist.lock); node =3D s->extralist.first; - if (node) + if (node) { s->extralist.first =3D node->next; + if (!s->extralist.first) + WRITE_ONCE(s->extralist.is_empty, true); + } raw_spin_unlock(&s->extralist.lock); return node; } @@ -164,15 +178,20 @@ ___pcpu_freelist_pop_nmi(struct pcpu_freelist *s) orig_cpu =3D cpu =3D raw_smp_processor_id(); while (1) { head =3D per_cpu_ptr(s->freelist, cpu); + if (READ_ONCE(head->is_empty)) + goto next_cpu; if (raw_spin_trylock(&head->lock)) { node =3D head->first; if (node) { head->first =3D node->next; + if (!head->first) + WRITE_ONCE(head->is_empty, true); raw_spin_unlock(&head->lock); return node; } raw_spin_unlock(&head->lock); } +next_cpu: cpu =3D cpumask_next(cpu, cpu_possible_mask); if (cpu >=3D nr_cpu_ids) cpu =3D 0; @@ -181,11 +200,14 @@ ___pcpu_freelist_pop_nmi(struct pcpu_freelist *s) } =20 /* cannot pop from per cpu lists, try extralist */ - if (!raw_spin_trylock(&s->extralist.lock)) + if (READ_ONCE(s->extralist.is_empty) || !raw_spin_trylock(&s->extralist.l= ock)) return NULL; node =3D s->extralist.first; - if (node) + if (node) { s->extralist.first =3D node->next; + if (!s->extralist.first) + WRITE_ONCE(s->extralist.is_empty, true); + } raw_spin_unlock(&s->extralist.lock); return node; } diff --git a/kernel/bpf/percpu_freelist.h b/kernel/bpf/percpu_freelist.h index 3c76553cfe57..9e4545631ed5 100644 --- a/kernel/bpf/percpu_freelist.h +++ b/kernel/bpf/percpu_freelist.h @@ -9,6 +9,7 @@ struct pcpu_freelist_head { struct pcpu_freelist_node *first; raw_spinlock_t lock; + bool is_empty; }; =20 struct pcpu_freelist { --=20 2.20.1 From nobody Tue Apr 28 10:33:56 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 71B95C433EF for ; Wed, 1 Jun 2022 08:42:26 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1343669AbiFAImX (ORCPT ); Wed, 1 Jun 2022 04:42:23 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:40038 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1348845AbiFAImQ (ORCPT ); Wed, 1 Jun 2022 04:42:16 -0400 Received: from mail-pj1-x102b.google.com (mail-pj1-x102b.google.com [IPv6:2607:f8b0:4864:20::102b]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 12989694AF for ; Wed, 1 Jun 2022 01:42:14 -0700 (PDT) Received: by mail-pj1-x102b.google.com with SMTP id n13-20020a17090a394d00b001e30a60f82dso5515699pjf.5 for ; Wed, 01 Jun 2022 01:42:14 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=bytedance-com.20210112.gappssmtp.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=B9vwKfbKHzDOEY1WrU024PdJUi8+ydTG61v0lkvD1q4=; b=nDKbEO4tZCSRRrSzh63hDu/iN425iuSK9EBeOyAKxNfyRH/f2IOm+NoRQDRvBCREVO PaRk0fuPmkOM2XoCmt3ulyij1N/Wrq7ee5hpTaYj1D3wnxOi2+qZdY92EbCrrcPRkbzI rxYow6zzLPDBClNz5/llPR7yXHPukzhuv6T4ANUlJVUhmXe+KU+Hk7gIMH5ozSGyYk05 g08oIUWKvs+S71KmJDx4CgEg/vOMbXnPuiviyZIPStoxVPAyHRV8woDEUiVBZ1WcI2II lsLH3BEK++XLOpH7R958TXbBY9M53ykeS6W7vLCSe0WWesao4ZyfcktoCALqEpJStcAL pygg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=B9vwKfbKHzDOEY1WrU024PdJUi8+ydTG61v0lkvD1q4=; b=K+vXR8XD5bjYT3NP8eFdykLyvXGbIoKGaAURC6YTqFZ/ZQsfrBddtjsweiWIh5a70o XxJzKm7cthYgT0rJDab2I9kOUZSliUSsSocLnUkaL6NfXFYoalZI8z5X4HS+EhN4jBW/ ismok7PBF2apIvCmsdMPsV3sz01wlsMTPbq9/v5Se8+2Y2VJzsg8T1Pbky9TEoc1rE5o q6bgA3pyjaaQtHl/fLY4J2vxFG6IL4dojW3hQ3fQ9FxE8JVy6RFr6FmHU0orcxzfq2qc /9y8sMkksQh4LFtUTXkrJkHyU9btH6vi0tsKNFlhMZ1nODBAeHYAVDSL5GCZkQHadD9k l0Ww== X-Gm-Message-State: AOAM531OhgIqw2YpJWZ/t3+JI52JttHW5SoYAVYPhYufg2w7Pi2V4f66 v7msP9WWs/a1UL1ffVQsesL9Og== X-Google-Smtp-Source: ABdhPJzck7sBi084+4Z7YhNWh2YIjxwOSXzmOy4Ww6SR0trDAp44Jq1YQ+Xa9zQq40HcPjQJSdHY5Q== X-Received: by 2002:a17:902:7048:b0:15f:34c1:bae0 with SMTP id h8-20020a170902704800b0015f34c1bae0mr64511498plt.71.1654072934453; Wed, 01 Jun 2022 01:42:14 -0700 (PDT) Received: from C02F52LSML85.bytedance.net ([139.177.225.241]) by smtp.gmail.com with ESMTPSA id u11-20020a63d34b000000b003c14af505f6sm827952pgi.14.2022.06.01.01.42.08 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Wed, 01 Jun 2022 01:42:14 -0700 (PDT) From: Feng zhou To: ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org, kafai@fb.com, songliubraving@fb.com, yhs@fb.com, john.fastabend@gmail.com, kpsingh@kernel.org Cc: netdev@vger.kernel.org, bpf@vger.kernel.org, linux-kernel@vger.kernel.org, duanxiongchun@bytedance.com, songmuchun@bytedance.com, wangdongdong.6@bytedance.com, cong.wang@bytedance.com, zhouchengming@bytedance.com, zhoufeng.zf@bytedance.com Subject: [PATCH v4 2/2] selftest/bpf/benchs: Add bpf_map benchmark Date: Wed, 1 Jun 2022 16:41:49 +0800 Message-Id: <20220601084149.13097-3-zhoufeng.zf@bytedance.com> X-Mailer: git-send-email 2.30.1 (Apple Git-130) In-Reply-To: <20220601084149.13097-1-zhoufeng.zf@bytedance.com> References: <20220601084149.13097-1-zhoufeng.zf@bytedance.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Type: text/plain; charset="utf-8" From: Feng Zhou Add benchmark for hash_map to reproduce the worst case that non-stop update when map's free is zero. before patch: Setting up benchmark 'bpf-hashmap-ful-update'... Benchmark 'bpf-hashmap-ful-update' started. 1:hash_map_full_perf 107796 events per sec 2:hash_map_full_perf 108072 events per sec 3:hash_map_full_perf 112169 events per sec 4:hash_map_full_perf 111423 events per sec 5:hash_map_full_perf 110778 events per sec 6:hash_map_full_perf 121336 events per sec 7:hash_map_full_perf 98676 events per sec 8:hash_map_full_perf 105860 events per sec 9:hash_map_full_perf 109930 events per sec 10:hash_map_full_perf 123434 events per sec 11:hash_map_full_perf 125374 events per sec 12:hash_map_full_perf 121979 events per sec 13:hash_map_full_perf 123014 events per sec 14:hash_map_full_perf 126219 events per sec 15:hash_map_full_perf 104793 events per sec after patch: Setting up benchmark 'bpf-hashmap-ful-update'... Benchmark 'bpf-hashmap-ful-update' started. 0:hash_map_full_perf 1219230 events per sec 1:hash_map_full_perf 1320256 events per sec 2:hash_map_full_perf 1196550 events per sec 3:hash_map_full_perf 1375684 events per sec 4:hash_map_full_perf 1365551 events per sec 5:hash_map_full_perf 1318432 events per sec 6:hash_map_full_perf 1222007 events per sec 7:hash_map_full_perf 1240786 events per sec 8:hash_map_full_perf 1190005 events per sec 9:hash_map_full_perf 1562336 events per sec 10:hash_map_full_perf 1385241 events per sec 11:hash_map_full_perf 1387909 events per sec 12:hash_map_full_perf 1371877 events per sec 13:hash_map_full_perf 1561836 events per sec 14:hash_map_full_perf 1388895 events per sec 15:hash_map_full_perf 1579054 events per sec Signed-off-by: Feng Zhou --- tools/testing/selftests/bpf/Makefile | 4 +- tools/testing/selftests/bpf/bench.c | 2 + .../benchs/bench_bpf_hashmap_full_update.c | 96 +++++++++++++++++++ .../run_bench_bpf_hashmap_full_update.sh | 11 +++ .../bpf/progs/bpf_hashmap_full_update_bench.c | 40 ++++++++ 5 files changed, 152 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/bpf/benchs/bench_bpf_hashmap_fu= ll_update.c create mode 100755 tools/testing/selftests/bpf/benchs/run_bench_bpf_hashma= p_full_update.sh create mode 100644 tools/testing/selftests/bpf/progs/bpf_hashmap_full_upda= te_bench.c diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests= /bpf/Makefile index 3820608faf57..b968649c7aa1 100644 --- a/tools/testing/selftests/bpf/Makefile +++ b/tools/testing/selftests/bpf/Makefile @@ -549,6 +549,7 @@ $(OUTPUT)/bench_ringbufs.o: $(OUTPUT)/ringbuf_bench.ske= l.h \ $(OUTPUT)/bench_bloom_filter_map.o: $(OUTPUT)/bloom_filter_bench.skel.h $(OUTPUT)/bench_bpf_loop.o: $(OUTPUT)/bpf_loop_bench.skel.h $(OUTPUT)/bench_strncmp.o: $(OUTPUT)/strncmp_bench.skel.h +$(OUTPUT)/bench_bpf_hashmap_full_update.o: $(OUTPUT)/bpf_hashmap_full_upda= te_bench.skel.h $(OUTPUT)/bench.o: bench.h testing_helpers.h $(BPFOBJ) $(OUTPUT)/bench: LDLIBS +=3D -lm $(OUTPUT)/bench: $(OUTPUT)/bench.o \ @@ -560,7 +561,8 @@ $(OUTPUT)/bench: $(OUTPUT)/bench.o \ $(OUTPUT)/bench_ringbufs.o \ $(OUTPUT)/bench_bloom_filter_map.o \ $(OUTPUT)/bench_bpf_loop.o \ - $(OUTPUT)/bench_strncmp.o + $(OUTPUT)/bench_strncmp.o \ + $(OUTPUT)/bench_bpf_hashmap_full_update.o $(call msg,BINARY,,$@) $(Q)$(CC) $(CFLAGS) $(LDFLAGS) $(filter %.a %.o,$^) $(LDLIBS) -o $@ =20 diff --git a/tools/testing/selftests/bpf/bench.c b/tools/testing/selftests/= bpf/bench.c index f973320e6dbf..35de886d9a05 100644 --- a/tools/testing/selftests/bpf/bench.c +++ b/tools/testing/selftests/bpf/bench.c @@ -397,6 +397,7 @@ extern const struct bench bench_hashmap_with_bloom; extern const struct bench bench_bpf_loop; extern const struct bench bench_strncmp_no_helper; extern const struct bench bench_strncmp_helper; +extern const struct bench bench_bpf_hashmap_full_update; =20 static const struct bench *benchs[] =3D { &bench_count_global, @@ -431,6 +432,7 @@ static const struct bench *benchs[] =3D { &bench_bpf_loop, &bench_strncmp_no_helper, &bench_strncmp_helper, + &bench_bpf_hashmap_full_update, }; =20 static void setup_benchmark() diff --git a/tools/testing/selftests/bpf/benchs/bench_bpf_hashmap_full_upda= te.c b/tools/testing/selftests/bpf/benchs/bench_bpf_hashmap_full_update.c new file mode 100644 index 000000000000..cec51e0ff4b8 --- /dev/null +++ b/tools/testing/selftests/bpf/benchs/bench_bpf_hashmap_full_update.c @@ -0,0 +1,96 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2022 Bytedance */ + +#include +#include "bench.h" +#include "bpf_hashmap_full_update_bench.skel.h" +#include "bpf_util.h" + +/* BPF triggering benchmarks */ +static struct ctx { + struct bpf_hashmap_full_update_bench *skel; +} ctx; + +#define MAX_LOOP_NUM 10000 + +static void validate(void) +{ + if (env.consumer_cnt !=3D 1) { + fprintf(stderr, "benchmark doesn't support multi-consumer!\n"); + exit(1); + } +} + +static void *producer(void *input) +{ + while (true) { + /* trigger the bpf program */ + syscall(__NR_getpgid); + } + + return NULL; +} + +static void *consumer(void *input) +{ + return NULL; +} + +static void measure(struct bench_res *res) +{ +} + +static void setup(void) +{ + struct bpf_link *link; + int map_fd, i, max_entries; + + setup_libbpf(); + + ctx.skel =3D bpf_hashmap_full_update_bench__open_and_load(); + if (!ctx.skel) { + fprintf(stderr, "failed to open skeleton\n"); + exit(1); + } + + ctx.skel->bss->nr_loops =3D MAX_LOOP_NUM; + + link =3D bpf_program__attach(ctx.skel->progs.benchmark); + if (!link) { + fprintf(stderr, "failed to attach program!\n"); + exit(1); + } + + /* fill hash_map */ + map_fd =3D bpf_map__fd(ctx.skel->maps.hash_map_bench); + max_entries =3D bpf_map__max_entries(ctx.skel->maps.hash_map_bench); + for (i =3D 0; i < max_entries; i++) + bpf_map_update_elem(map_fd, &i, &i, BPF_ANY); +} + +void hashmap_report_final(struct bench_res res[], int res_cnt) +{ + unsigned int nr_cpus =3D bpf_num_possible_cpus(); + int i; + + for (i =3D 0; i < nr_cpus; i++) { + u64 time =3D ctx.skel->bss->percpu_time[i]; + + if (!time) + continue; + + printf("%d:hash_map_full_perf %lld events per sec\n", + i, ctx.skel->bss->nr_loops * 1000000000ll / time); + } +} + +const struct bench bench_bpf_hashmap_full_update =3D { + .name =3D "bpf-hashmap-ful-update", + .validate =3D validate, + .setup =3D setup, + .producer_thread =3D producer, + .consumer_thread =3D consumer, + .measure =3D measure, + .report_progress =3D NULL, + .report_final =3D hashmap_report_final, +}; diff --git a/tools/testing/selftests/bpf/benchs/run_bench_bpf_hashmap_full_= update.sh b/tools/testing/selftests/bpf/benchs/run_bench_bpf_hashmap_full_u= pdate.sh new file mode 100755 index 000000000000..1e2de838f9fa --- /dev/null +++ b/tools/testing/selftests/bpf/benchs/run_bench_bpf_hashmap_full_update.= sh @@ -0,0 +1,11 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0 + +source ./benchs/run_common.sh + +set -eufo pipefail + +nr_threads=3D`expr $(cat /proc/cpuinfo | grep "processor"| wc -l) - 1` +summary=3D$($RUN_BENCH -p $nr_threads bpf-hashmap-ful-update) +printf "$summary" +printf "\n" diff --git a/tools/testing/selftests/bpf/progs/bpf_hashmap_full_update_benc= h.c b/tools/testing/selftests/bpf/progs/bpf_hashmap_full_update_bench.c new file mode 100644 index 000000000000..aa93a03f961d --- /dev/null +++ b/tools/testing/selftests/bpf/progs/bpf_hashmap_full_update_bench.c @@ -0,0 +1,40 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2022 Bytedance */ + +#include "vmlinux.h" +#include +#include "bpf_misc.h" + +char _license[] SEC("license") =3D "GPL"; + +#define MAX_ENTRIES 1000 + +struct { + __uint(type, BPF_MAP_TYPE_HASH); + __type(key, u32); + __type(value, u64); + __uint(max_entries, MAX_ENTRIES); +} hash_map_bench SEC(".maps"); + +u64 __attribute__((__aligned__(256))) percpu_time[256]; +u64 nr_loops; + +static int loop_update_callback(__u32 index, u32 *key) +{ + u64 init_val =3D 1; + + bpf_map_update_elem(&hash_map_bench, key, &init_val, BPF_ANY); + return 0; +} + +SEC("fentry/" SYS_PREFIX "sys_getpgid") +int benchmark(void *ctx) +{ + u32 key =3D bpf_get_prandom_u32() % MAX_ENTRIES + MAX_ENTRIES; + u32 cpu =3D bpf_get_smp_processor_id(); + u64 start_time =3D bpf_ktime_get_ns(); + + bpf_loop(nr_loops, loop_update_callback, &key, 0); + percpu_time[cpu & 255] =3D bpf_ktime_get_ns() - start_time; + return 0; +} --=20 2.20.1