From nobody Wed Dec 17 14:04:43 2025 Received: from m16.mail.163.com (m16.mail.163.com [117.135.210.4]) by smtp.subspace.kernel.org (Postfix) with ESMTP id 2947F4C6B; Sat, 22 Jun 2024 07:17:51 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=117.135.210.4 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1719040676; cv=none; b=JT/oEJu1jFC+690dj8Ea6MxWe7Gorfx1D9NZNu/AlVXr+HnLj90qamtV/fJ3TFTCNDI6dOgwQ3Jn73QsQODBjG2qcgE/D/i80qpGgoNERrNU8foju9zwK2/8SjlMUb6Mcl2kiW3WJQAlL+IGfQ/4s4YjK7zR6xdehaGakD1wrnU= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1719040676; c=relaxed/simple; bh=T9O4nMfNL264snMcAy6vophUdVlTs3RvdJVMqt8zS+w=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=mV6G9zBS2L+nTOdBqghGZyMQeeT7LrxBJaiAJm5XAcpZczAUWof4tF8ULuqR2EE89yECTZmc3l1P17zWX/dyxUHABN5/Ack8wRbBiwEMVS8CSVcgCrhYueFHefdGJWVXLdtVirixOpluUw+oW5Oz9FQ9oKm2wQde4FyeNUr7QNc= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=163.com; spf=pass smtp.mailfrom=163.com; dkim=pass (1024-bit key) header.d=163.com header.i=@163.com header.b=lGeJCGuR; arc=none smtp.client-ip=117.135.210.4 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=163.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=163.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=163.com header.i=@163.com header.b="lGeJCGuR" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=163.com; s=s110527; h=From:Subject:Date:Message-Id:MIME-Version; bh=2bzIC 7usDj/I+d/64DMmEa2VWZHcXlRlo9WqOfdcwd4=; b=lGeJCGuRyzummNUru9dOS GQc0Ao2aVBmxDe1dickbop4B32JMPDakCovfpFThG6DVR9MLzlr8QVAVKpnihaa1 4vW0Wv23qP+Lg/lCJcTrMmHmUp17I3zYCp1RKNKEx9hsYwwmd5CXZMTzwfjfGFgR bVOyqgwy/HExb/lyetyseY= Received: from localhost (unknown [101.132.132.191]) by gzga-smtp-mta-g2-2 (Coremail) with SMTP id _____wD3HybTeXZmUqwUAA--.6297S2; Sat, 22 Jun 2024 15:14:28 +0800 (CST) From: Xavier To: tj@kernel.org, longman@redhat.com Cc: mkoutny@suse.com, lizefan.x@bytedance.com, hannes@cmpxchg.org, cgroups@vger.kernel.org, linux-kernel@vger.kernel.org, torvalds@linux-foundation.org, akpm@linux-foundation.org, Xavier Subject: [PATCH-cpuset v6 2/2] cpuset: use Union-Find to optimize the merging of cpumasks Date: Sat, 22 Jun 2024 15:14:24 +0800 Message-Id: <20240622071424.215778-3-xavier_qy@163.com> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20240622071424.215778-1-xavier_qy@163.com> References: <20240622071424.215778-1-xavier_qy@163.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-CM-TRANSID: _____wD3HybTeXZmUqwUAA--.6297S2 X-Coremail-Antispam: 1Uf129KBjvJXoWxAFWfur4kKFyUZFWUJrW3Awb_yoWrCw47pF s3Gay29rWrJryUGwsYkay8Z34SkaykJa1Utw13Gw1rArnrA3Z2va40qFs5KFWUuryDCF1U uFnxKr47WryUKFDanT9S1TB71UUUUU7qnTZGkaVYY2UrUUUUjbIjqfuFe4nvWSU5nxnvy2 9KBjDUYxBIdaVFxhVjvjDU0xZFpf9x07U-J5wUUUUU= X-CM-SenderInfo: 50dyxvpubt5qqrwthudrp/1tbiYx0FEGV4I3AumQAFsX Content-Type: text/plain; charset="utf-8" The process of constructing scheduling domains involves multiple loops and repeated evaluations, leading to numerous redundant and ineffective assessments that impact code efficiency. Here, we use Union-Find to optimize the merging of cpumasks. By employing path compression and union by rank, we effectively reduce the number of lookups and merge comparisons. Signed-off-by: Xavier --- kernel/cgroup/cpuset.c | 97 +++++++++++++++++------------------------- 1 file changed, 38 insertions(+), 59 deletions(-) diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index fe76045aa5..98e9df57c4 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -45,6 +45,7 @@ #include #include #include +#include =20 DEFINE_STATIC_KEY_FALSE(cpusets_pre_enable_key); DEFINE_STATIC_KEY_FALSE(cpusets_enabled_key); @@ -172,9 +173,6 @@ struct cpuset { */ int attach_in_progress; =20 - /* partition number for rebuild_sched_domains() */ - int pn; - /* for custom sched domain */ int relax_domain_level; =20 @@ -1007,7 +1005,7 @@ static int generate_sched_domains(cpumask_var_t **dom= ains, struct cpuset *cp; /* top-down scan of cpusets */ struct cpuset **csa; /* array of all cpuset ptrs */ int csn; /* how many cpuset ptrs in csa so far */ - int i, j, k; /* indices for partition finding loops */ + int i, j; /* indices for partition finding loops */ cpumask_var_t *doms; /* resulting partition; i.e. sched domains */ struct sched_domain_attr *dattr; /* attributes for custom domains */ int ndoms =3D 0; /* number of sched domains in result */ @@ -1015,6 +1013,8 @@ static int generate_sched_domains(cpumask_var_t **dom= ains, struct cgroup_subsys_state *pos_css; bool root_load_balance =3D is_sched_load_balance(&top_cpuset); bool cgrpv2 =3D cgroup_subsys_on_dfl(cpuset_cgrp_subsys); + struct uf_node *nodes =3D NULL; + int nslot_update; =20 doms =3D NULL; dattr =3D NULL; @@ -1102,31 +1102,26 @@ static int generate_sched_domains(cpumask_var_t **d= omains, if (root_load_balance && (csn =3D=3D 1)) goto single_root_domain; =20 - for (i =3D 0; i < csn; i++) - csa[i]->pn =3D i; - ndoms =3D csn; - -restart: - /* Find the best partition (set of sched domains) */ - for (i =3D 0; i < csn; i++) { - struct cpuset *a =3D csa[i]; - int apn =3D a->pn; - - for (j =3D 0; j < csn; j++) { - struct cpuset *b =3D csa[j]; - int bpn =3D b->pn; - - if (apn !=3D bpn && cpusets_overlap(a, b)) { - for (k =3D 0; k < csn; k++) { - struct cpuset *c =3D csa[k]; + if (!cgrpv2) { + nodes =3D uf_nodes_alloc(csn); + if (!nodes) + goto done; =20 - if (c->pn =3D=3D bpn) - c->pn =3D apn; - } - ndoms--; /* one less element */ - goto restart; + /* Merge overlapping cpusets */ + for (i =3D 0; i < csn; i++) { + for (j =3D i + 1; j < csn; j++) { + if (cpusets_overlap(csa[i], csa[j])) + uf_union(&nodes[i], &nodes[j]); } } + + /* Count the total number of domains */ + for (i =3D 0; i < csn; i++) { + if ((nodes[i].parent =3D=3D &nodes[i]) || !nodes[i].parent) + ndoms++; + } + } else { + ndoms =3D csn; } =20 /* @@ -1159,48 +1154,32 @@ static int generate_sched_domains(cpumask_var_t **d= omains, } =20 for (nslot =3D 0, i =3D 0; i < csn; i++) { - struct cpuset *a =3D csa[i]; - struct cpumask *dp; - int apn =3D a->pn; - - if (apn < 0) { - /* Skip completed partitions */ - continue; - } - - dp =3D doms[nslot]; - - if (nslot =3D=3D ndoms) { - static int warnings =3D 10; - if (warnings) { - pr_warn("rebuild_sched_domains confused: nslot %d, ndoms %d, csn %d, i= %d, apn %d\n", - nslot, ndoms, csn, i, apn); - warnings--; - } - continue; - } - - cpumask_clear(dp); - if (dattr) - *(dattr + nslot) =3D SD_ATTR_INIT; + nslot_update =3D 0; for (j =3D i; j < csn; j++) { - struct cpuset *b =3D csa[j]; - - if (apn =3D=3D b->pn) { - cpumask_or(dp, dp, b->effective_cpus); + if (uf_find(&nodes[j]) =3D=3D &nodes[i]) { + struct cpumask *dp =3D doms[nslot]; + + if (i =3D=3D j) { + nslot_update =3D 1; + cpumask_clear(dp); + if (dattr) + *(dattr + nslot) =3D SD_ATTR_INIT; + } + cpumask_or(dp, dp, csa[j]->effective_cpus); cpumask_and(dp, dp, housekeeping_cpumask(HK_TYPE_DOMAIN)); if (dattr) - update_domain_attr_tree(dattr + nslot, b); - - /* Done with this partition */ - b->pn =3D -1; + update_domain_attr_tree(dattr + nslot, csa[j]); } } - nslot++; + if (nslot_update) + nslot++; } BUG_ON(nslot !=3D ndoms); =20 done: + if (nodes) + uf_nodes_free(nodes); + kfree(csa); =20 /* --=20 2.45.2