From patchwork Fri Nov 3 12:25:44 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Prarit Bhargava X-Patchwork-Id: 10039951 X-Patchwork-Delegate: lenb@kernel.org Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 751C760384 for ; Fri, 3 Nov 2017 12:26:06 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 5B9BF295B3 for ; Fri, 3 Nov 2017 12:26:06 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 50514295BF; Fri, 3 Nov 2017 12:26:06 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=2.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id AD64A295BE for ; Fri, 3 Nov 2017 12:26:05 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756179AbdKCM0E (ORCPT ); Fri, 3 Nov 2017 08:26:04 -0400 Received: from mx1.redhat.com ([209.132.183.28]:35750 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755927AbdKCM0D (ORCPT ); Fri, 3 Nov 2017 08:26:03 -0400 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.phx2.redhat.com [10.5.11.16]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id F1D39C0587F4; Fri, 3 Nov 2017 12:26:02 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com F1D39C0587F4 Authentication-Results: ext-mx08.extmail.prod.ext.phx2.redhat.com; dmarc=none (p=none dis=none) header.from=redhat.com Authentication-Results: ext-mx08.extmail.prod.ext.phx2.redhat.com; spf=fail smtp.mailfrom=prarit@redhat.com Received: from praritdesktop.bos.redhat.com (prarit-guest.khw.lab.eng.bos.redhat.com [10.16.186.145]) by smtp.corp.redhat.com (Postfix) with ESMTP id 34C1A5C548; Fri, 3 Nov 2017 12:26:02 +0000 (UTC) From: Prarit Bhargava To: linux-pm@vger.kernel.org Cc: lenb@kernel.org, len.brown@intel.com, charles.rose@dell.com, rafael@kernel.org, suravee.suthikulpanit@amd.com, Prarit Bhargava Subject: [RESEND PATCH 2/8] turbostat: Fix node and siblings lookup data Date: Fri, 3 Nov 2017 08:25:44 -0400 Message-Id: <20171103122550.13341-3-prarit@redhat.com> In-Reply-To: <20171103122550.13341-1-prarit@redhat.com> References: <20171103122550.13341-1-prarit@redhat.com> X-Scanned-By: MIMEDefang 2.79 on 10.5.11.16 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.32]); Fri, 03 Nov 2017 12:26:03 +0000 (UTC) Sender: linux-pm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pm@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP The turbostat code only looks at thread_siblings_list to determine if processing units/threads are on the same the core. This works well on Intel systems which have a shared L1 instruction and data cache. This does not work on AMD systems which have shared L1 instruction cache but separate L1 data caches. Other utilities also check sibling's core ID to determine if the processing unit shares the same core. Additionally, the cpu_topology *cpus list used in topology_probe() can be used elsewhere in the code to simplify things. Export *cpus to the entire turbostat code, and add Processing Unit/Thread IDs information to each cpu_topology struct. Confirm that the thread is on the same core as indicated by thread_siblings_list. Signed-off-by: Prarit Bhargava Cc: Len Brown --- tools/power/x86/turbostat/turbostat.c | 114 +++++++++++++++++++++++----------- 1 file changed, 79 insertions(+), 35 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 6135aa99f8ad..33d35e19164d 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -149,6 +149,7 @@ char *progname; cpu_set_t *cpu_present_set, *cpu_affinity_set, *cpu_subset; size_t cpu_present_setsize, cpu_affinity_setsize, cpu_subset_size; #define MAX_ADDED_COUNTERS 16 +#define BITMASK_SIZE 32 struct thread_data { struct timeval tv_begin; @@ -245,6 +246,13 @@ struct system_summary { struct pkg_data packages; } average; +struct cpu_topology { + int physical_package_id; + int logical_cpu_id; + int node_id; + int physical_core_id; + cpu_set_t *put_ids; /* Processing Unit/Thread IDs */ +} *cpus; struct topo_params { int num_packages; @@ -2188,6 +2196,8 @@ void free_fd_percpu(void) void free_all_buffers(void) { + int i; + CPU_FREE(cpu_present_set); cpu_present_set = NULL; cpu_present_setsize = 0; @@ -2220,6 +2230,12 @@ void free_all_buffers(void) free(irq_column_2_cpu); free(irqs_per_cpu); + + for (i = 0; i <= topo.max_cpu_num; ++i) { + if (cpus[i].put_ids) + CPU_FREE(cpus[i].put_ids); + } + free(cpus); } @@ -2300,35 +2316,57 @@ int get_core_id(int cpu) return parse_int_file("/sys/devices/system/cpu/cpu%d/topology/core_id", cpu); } -int get_num_ht_siblings(int cpu) +int get_node_id(struct cpu_topology *thiscpu) { char path[80]; FILE *filep; - int sib1; - int matches = 0; - char character; - char str[100]; - char *ch; - - sprintf(path, "/sys/devices/system/cpu/cpu%d/topology/thread_siblings_list", cpu); - filep = fopen_or_die(path, "r"); + int i; + int cpu = thiscpu->logical_cpu_id; - /* - * file format: - * A ',' separated or '-' separated set of numbers - * (eg 1-2 or 1,3,4,5) - */ - fscanf(filep, "%d%c\n", &sib1, &character); - fseek(filep, 0, SEEK_SET); - fgets(str, 100, filep); - ch = strchr(str, character); - while (ch != NULL) { - matches++; - ch = strchr(ch+1, character); + for (i = 0; i <= topo.max_cpu_num; i++) { + sprintf(path, "/sys/devices/system/cpu/cpu%d/node%i/cpulist", + cpu, i); + filep = fopen(path, "r"); + if (!filep) + continue; + fclose(filep); + return i; } + return -1; +} +int get_thread_siblings(struct cpu_topology *thiscpu) +{ + char path[80], character; + FILE *filep; + unsigned long map; + int shift, sib_core; + int cpu = thiscpu->logical_cpu_id; + int offset = topo.max_cpu_num + 1; + + thiscpu->put_ids = CPU_ALLOC((topo.max_cpu_num + 1)); + if (!thiscpu->put_ids) + return -1; + CPU_ZERO(thiscpu->put_ids); + + sprintf(path, + "/sys/devices/system/cpu/cpu%d/topology/thread_siblings", cpu); + filep = fopen_or_die(path, "r"); + do { + offset -= BITMASK_SIZE; + fscanf(filep, "%lx%c", &map, &character); + for (shift = 0; shift < BITMASK_SIZE; shift++) { + if ((map >> shift) & 0x1) { + sib_core = get_core_id(shift + offset); + if (sib_core == thiscpu->physical_core_id) + CPU_SET(shift + offset, + thiscpu->put_ids); + } + } + } while (!strncmp(&character, ",", 1)); fclose(filep); - return matches+1; + + return CPU_COUNT(thiscpu->put_ids); } /* @@ -2423,7 +2461,7 @@ void set_max_cpu_num(void) "/sys/devices/system/cpu/cpu0/topology/thread_siblings", "r"); while (fscanf(filep, "%lx,", &dummy) == 1) - topo.max_cpu_num += 32; + topo.max_cpu_num += BITMASK_SIZE; fclose(filep); topo.max_cpu_num--; /* 0 based */ } @@ -4325,10 +4363,6 @@ void topology_probe() int max_core_id = 0; int max_package_id = 0; int max_siblings = 0; - struct cpu_topology { - int core_id; - int physical_package_id; - } *cpus; /* Initialize num_cpus, max_cpu_num */ set_max_cpu_num(); @@ -4385,20 +4419,32 @@ void topology_probe() fprintf(outf, "cpu%d NOT PRESENT\n", i); continue; } - cpus[i].core_id = get_core_id(i); - if (cpus[i].core_id > max_core_id) - max_core_id = cpus[i].core_id; + cpus[i].logical_cpu_id = i; + + /* get package information */ cpus[i].physical_package_id = get_physical_package_id(i); if (cpus[i].physical_package_id > max_package_id) max_package_id = cpus[i].physical_package_id; - siblings = get_num_ht_siblings(i); + /* get numa node information */ + cpus[i].node_id = get_node_id(&cpus[i]); + + /* get core information */ + cpus[i].physical_core_id = get_core_id(i); + if (cpus[i].physical_core_id > max_core_id) + max_core_id = cpus[i].physical_core_id; + + /* get thread information */ + siblings = get_thread_siblings(&cpus[i]); if (siblings > max_siblings) max_siblings = siblings; + if (debug > 1) - fprintf(outf, "cpu %d pkg %d core %d\n", - i, cpus[i].physical_package_id, cpus[i].core_id); + fprintf(outf, "cpu %d pkg %d node %d core %d\n", + i, cpus[i].physical_package_id, + cpus[i].node_id, + cpus[i].physical_core_id); } topo.num_cores_per_pkg = max_core_id + 1; if (debug > 1) @@ -4417,8 +4463,6 @@ void topology_probe() topo.num_threads_per_core = max_siblings; if (debug > 1) fprintf(outf, "max_siblings %d\n", max_siblings); - - free(cpus); } void