Received: by 2002:ad5:474a:0:0:0:0:0 with SMTP id i10csp105860imu; Mon, 10 Dec 2018 17:11:06 -0800 (PST) X-Google-Smtp-Source: AFSGD/UsgpEj+6UUnUBBWI81FTLMPwQHqp+6jtG0gpkQSvCnOHkU1hOdwlOet7PCHh46J0swblnh X-Received: by 2002:a63:2263:: with SMTP id t35mr12806369pgm.69.1544490666410; Mon, 10 Dec 2018 17:11:06 -0800 (PST) ARC-Seal: i=1; a=rsa-sha256; t=1544490666; cv=none; d=google.com; s=arc-20160816; b=m72wT0sRgzTzXoopjSbiM3iNGc5+YaMt1mg6QUaJGy348SHSNWwx+Ds0CBfd5/n03A fovES8kGpFiN66PCh412P2roclL0q91gbS7UzXXXY3txGV8UHFJmHJS0H3Mh/DnW+52L A9UrdNrfz0A8JBujruZEv6MAnK1kDNFEa4byBjAYjov06KIRb5HmDbBGt04PwKKv3Fe3 TXU55bOJtcpLIty6XmOp1uVrJO0K5B7UhEoPkXUPimcFa1ksuFOQIhR7ZKNqOUHL+47H Il4gG8uvl6CCLNT78W7W+SRqtkSL7AvjHHvWd4Sl8KpNJmDftmmgIsR08WFoma4J/bq/ IeqQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=list-id:precedence:sender:references:in-reply-to:message-id:date :subject:cc:to:from; bh=44qP4UsydjGC77SKhN4MAkgaKSiEKUy1LUuzw8HoBkI=; b=pEAWO59NosdVmRzZGgZZq3H9KNSwIw+Ew3xkByI1Q0msap+qucWKx68j2NdMUHsPwl dvtXo71Vzwcgh23kdJfQLLSL9SLtwYdDGTFxSFzAyFb6MIDvJJYiHUPYRt2M0QfH+ahc IvdRI2KgT9YrGyRvjG91lGNFUmQa22RJ5eTytMAvA8YUMjEjWBY9AVxzvlC4+xJJ6fGn PlRulaaopb7eArsmuRa8zHEp4R1O6W9fQFtf6pPkGpCW9+xkYVxe9P7aD4SV+VKiwpVp LyP5T5sd5+Oz5O6wmt2AkuhGpeasPjvfxvxVQVVGbT2/T95YqAZmOXLgfjxJVthYEa9F i/AA== ARC-Authentication-Results: i=1; mx.google.com; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=intel.com Return-Path: Received: from vger.kernel.org (vger.kernel.org. [209.132.180.67]) by mx.google.com with ESMTP id t186si11956850pfd.68.2018.12.10.17.10.51; Mon, 10 Dec 2018 17:11:06 -0800 (PST) Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67; Authentication-Results: mx.google.com; spf=pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) smtp.mailfrom=linux-kernel-owner@vger.kernel.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=intel.com Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729520AbeLKBFw (ORCPT + 99 others); Mon, 10 Dec 2018 20:05:52 -0500 Received: from mga02.intel.com ([134.134.136.20]:24625 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729289AbeLKBFv (ORCPT ); Mon, 10 Dec 2018 20:05:51 -0500 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga001.jf.intel.com ([10.7.209.18]) by orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 10 Dec 2018 17:05:49 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.56,340,1539673200"; d="scan'208";a="117705189" Received: from unknown (HELO localhost.lm.intel.com) ([10.232.112.69]) by orsmga001.jf.intel.com with ESMTP; 10 Dec 2018 17:05:48 -0800 From: Keith Busch To: linux-kernel@vger.kernel.org, linux-acpi@vger.kernel.org, linux-mm@kvack.org Cc: Greg Kroah-Hartman , Rafael Wysocki , Dave Hansen , Dan Williams , Keith Busch Subject: [PATCHv2 06/12] node: Add heterogenous memory performance Date: Mon, 10 Dec 2018 18:03:04 -0700 Message-Id: <20181211010310.8551-7-keith.busch@intel.com> X-Mailer: git-send-email 2.13.6 In-Reply-To: <20181211010310.8551-1-keith.busch@intel.com> References: <20181211010310.8551-1-keith.busch@intel.com> Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Heterogeneous memory systems provide memory nodes with different latency and bandwidth performance attributes. Create an interface for the kernel to register the attributes for the primary memory initiators under the memory node. If the system provides this information, applications can then query the node attributes when deciding which node to request memory. The following example shows the new sysfs hierarchy for a node exporting performance attributes: # tree /sys/devices/system/node/nodeY/primary_initiator_access /sys/devices/system/node/nodeY/primary_initiator_access |-- read_bandwidth |-- read_latency |-- write_bandwidth `-- write_latency The bandwidth is exported as MB/s and latency is reported in nanoseconds. Memory accesses from an initiator node that is not one of the memory's primary compute nodes may encounter a performance penalty that does not match the performance reported for primary memory initiators. As an example of what you may be able to do with this, let's say we have a PCIe storage device, /dev/nvme0n1, attached to a particular node, and we want to run IO to it using the fastest memory with primary access from the same node as that PCIe device. The following shell script is such an example to achieve that goal: #!/bin/bash DEV_NODE=/sys/devices/system/node/node$(cat /sys/block/nvme0n1/device/device/numa_node) BEST_WRITE_BW=0 BEST_MEM_NODE=0 for i in $(ls -d ${DEV_NODE}/primary_target*); do tmp=$(cat ${i}/primary_initiator_access/write_bandwidth); if ((${tmp} > ${BEST_WRITE_BW})); then BEST_WRITE_BW=${tmp} BEST_MEM_NODE=$(echo ${i} | sed s/^.*primary_target//g) fi done numactl --membind=${BEST_MEM_NODE} \ --cpunodebind=$(cat ${DEV_NODE}/primary_cpu_nodelist) \ -- fio --filename=/dev/nvme0n1 --bs=4k --name=access-test Signed-off-by: Keith Busch --- drivers/base/Kconfig | 8 ++++++++ drivers/base/node.c | 44 ++++++++++++++++++++++++++++++++++++++++++++ include/linux/node.h | 22 ++++++++++++++++++++++ 3 files changed, 74 insertions(+) diff --git a/drivers/base/Kconfig b/drivers/base/Kconfig index 3e63a900b330..6014980238e8 100644 --- a/drivers/base/Kconfig +++ b/drivers/base/Kconfig @@ -149,6 +149,14 @@ config DEBUG_TEST_DRIVER_REMOVE unusable. You should say N here unless you are explicitly looking to test this functionality. +config HMEM_REPORTING + bool + default y + depends on NUMA + help + Enable reporting for heterogenous memory access attributes under + their non-uniform memory nodes. + source "drivers/base/test/Kconfig" config SYS_HYPERVISOR diff --git a/drivers/base/node.c b/drivers/base/node.c index 50412ce3fd7d..768612c06c56 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -99,6 +99,50 @@ static DEVICE_ATTR_RO(primary_cpu_nodelist); static DEVICE_ATTR(cpumap, S_IRUGO, node_read_cpumask, NULL); static DEVICE_ATTR(cpulist, S_IRUGO, node_read_cpulist, NULL); +#ifdef CONFIG_HMEM_REPORTING +const struct attribute_group node_access_attrs_group; + +#define ACCESS_ATTR(name) \ +static ssize_t name##_show(struct device *dev, \ + struct device_attribute *attr, \ + char *buf) \ +{ \ + return sprintf(buf, "%d\n", to_node(dev)->hmem_attrs.name); \ +} \ +static DEVICE_ATTR_RO(name); + +ACCESS_ATTR(read_bandwidth) +ACCESS_ATTR(read_latency) +ACCESS_ATTR(write_bandwidth) +ACCESS_ATTR(write_latency) + +static struct attribute *access_attrs[] = { + &dev_attr_read_bandwidth.attr, + &dev_attr_read_latency.attr, + &dev_attr_write_bandwidth.attr, + &dev_attr_write_latency.attr, + NULL, +}; + +const struct attribute_group node_access_attrs_group = { + .name = "primary_initiator_access", + .attrs = access_attrs, +}; + +void node_set_perf_attrs(unsigned int nid, struct node_hmem_attrs *hmem_attrs) +{ + struct node *node; + + if (WARN_ON_ONCE(!node_online(nid))) + return; + node = node_devices[nid]; + node->hmem_attrs = *hmem_attrs; + if (sysfs_create_group(&node->dev.kobj, &node_access_attrs_group)) + pr_info("failed to add performance attribute group to node %d\n", + nid); +} +#endif + #define K(x) ((x) << (PAGE_SHIFT - 10)) static ssize_t node_read_meminfo(struct device *dev, struct device_attribute *attr, char *buf) diff --git a/include/linux/node.h b/include/linux/node.h index 3d06de045cbf..71abaf0d4f4b 100644 --- a/include/linux/node.h +++ b/include/linux/node.h @@ -17,8 +17,27 @@ #include #include +#include #include +#ifdef CONFIG_HMEM_REPORTING +/** + * struct node_hmem_attrs - heterogeneous memory performance attributes + * + * read_bandwidth: Read bandwidth in MB/s + * write_bandwidth: Write bandwidth in MB/s + * read_latency: Read latency in nanoseconds + * write_latency: Write latency in nanoseconds + */ +struct node_hmem_attrs { + unsigned int read_bandwidth; + unsigned int write_bandwidth; + unsigned int read_latency; + unsigned int write_latency; +}; +void node_set_perf_attrs(unsigned int nid, struct node_hmem_attrs *hmem_attrs); +#endif + struct node { struct device dev; nodemask_t primary_mem_nodes; @@ -27,6 +46,9 @@ struct node { #if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_HUGETLBFS) struct work_struct node_work; #endif +#ifdef CONFIG_HMEM_REPORTING + struct node_hmem_attrs hmem_attrs; +#endif }; struct memory_block; -- 2.14.4