From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-9.1 required=3.0 tests=DKIM_SIGNED,DKIM_VALID, DKIM_VALID_AU,INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE, SPF_PASS,USER_AGENT_GIT autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 2075DC04AAF for ; Mon, 20 May 2019 05:42:45 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id D4E55206B6 for ; Mon, 20 May 2019 05:42:44 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1558330965; bh=lWt5d1InJFN1ghf1lvShlp5HkkwKBbGCM4Ej4W5ghvM=; h=From:To:Cc:Subject:Date:In-Reply-To:References:List-ID:From; b=gCfZxUwIUUgDKpAjK5vqr0MF0lf0LWKv3etKQa9tJhQEZWs8dYslqBqsDSseDaDpy fC5kyJAne4SlL5eNowUxiTlR4A68omBOv5iuwTCkC46RK7KEnYJqQlPIWjP82uD9wT R5gqZVWFwAhnZwss5D3sq86zN8KZ7g4Gt4OqqRdM= Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1730306AbfETFmj (ORCPT ); Mon, 20 May 2019 01:42:39 -0400 Received: from new4-smtp.messagingengine.com ([66.111.4.230]:39715 "EHLO new4-smtp.messagingengine.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727057AbfETFmj (ORCPT ); Mon, 20 May 2019 01:42:39 -0400 Received: from compute3.internal (compute3.nyi.internal [10.202.2.43]) by mailnew.nyi.internal (Postfix) with ESMTP id 7AE6FDC02; Mon, 20 May 2019 01:42:37 -0400 (EDT) Received: from mailfrontend1 ([10.202.2.162]) by compute3.internal (MEProxy); Mon, 20 May 2019 01:42:37 -0400 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d= messagingengine.com; h=cc:content-transfer-encoding:date:from :in-reply-to:message-id:mime-version:references:subject:to :x-me-proxy:x-me-proxy:x-me-sender:x-me-sender:x-sasl-enc; s= fm2; bh=9XDuSxwKrHTa5OuRmvXZIbQcLUkhgPbDvSR6tXiHu4g=; b=ZcsCbH2n kx1kNHGoZK7ikUL9wnmYUEIsxrCWj1knqAFlyLdJaK9uiLS4iwz5G6z8XnewR98P De3WXWOu9CSDdOa9H4CPAzqGZhflIdG8VTY1+mkbcNeVpIoMLEqoaiCKuEDKQOHq HitUxfue0bnJaGOnMPDRx7f5SXaX4p/RBYpc1kJKQNcn9BqhxRNp5wyMNC1K8Fp8 M4o6TSgZ+aaec/tmYMbuJ2fdiNGwLkapRM7eW3288aSZR+rc2lr0UlsRxsHRWE4u G+tcdDg2vlwDKOCGt2/jhDtYvydr8vwlB8kSArZcUjXvmFFU4spRPENzaXHDQsAT nq5LQTuebmFuXA== X-ME-Sender: X-ME-Proxy-Cause: gggruggvucftvghtrhhoucdtuddrgeduuddruddtjedguddtudcutefuodetggdotefrod ftvfcurfhrohhfihhlvgemucfhrghsthforghilhdpqfgfvfdpuffrtefokffrpgfnqfgh necuuegrihhlohhuthemuceftddtnecusecvtfgvtghiphhivghnthhsucdlqddutddtmd enucfjughrpefhvffufffkofgjfhgggfestdekredtredttdenucfhrhhomhepfdfvohgs ihhnucevrdcujfgrrhguihhnghdfuceothhosghinheskhgvrhhnvghlrdhorhhgqeenuc fkphepuddvgedrudeiledrudehiedrvddtfeenucfrrghrrghmpehmrghilhhfrhhomhep thhosghinheskhgvrhhnvghlrdhorhhgnecuvehluhhsthgvrhfuihiivgepuddu X-ME-Proxy: Received: from eros.localdomain (124-169-156-203.dyn.iinet.net.au [124.169.156.203]) by mail.messagingengine.com (Postfix) with ESMTPA id 5638E8005B; Mon, 20 May 2019 01:42:30 -0400 (EDT) From: "Tobin C. Harding" To: Andrew Morton , Matthew Wilcox Cc: "Tobin C. Harding" , Roman Gushchin , Alexander Viro , Christoph Hellwig , Pekka Enberg , David Rientjes , Joonsoo Kim , Christopher Lameter , Miklos Szeredi , Andreas Dilger , Waiman Long , Tycho Andersen , Theodore Ts'o , Andi Kleen , David Chinner , Nick Piggin , Rik van Riel , Hugh Dickins , Jonathan Corbet , linux-mm@kvack.org, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org Subject: [RFC PATCH v5 12/16] slub: Enable moving objects to/from specific nodes Date: Mon, 20 May 2019 15:40:13 +1000 Message-Id: <20190520054017.32299-13-tobin@kernel.org> X-Mailer: git-send-email 2.21.0 In-Reply-To: <20190520054017.32299-1-tobin@kernel.org> References: <20190520054017.32299-1-tobin@kernel.org> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org We have just implemented Slab Movable Objects (object migration). Currently object migration is used to defrag a cache. On NUMA systems it would be nice to be able to control the source and destination nodes when moving objects. Add CONFIG_SMO_NODE to guard this feature. CONFIG_SMO_NODE depends on CONFIG_SLUB_DEBUG because we use the full list. Implement moving all objects (including those in full slabs) to a specific node. Expose this functionality to userspace via a sysfs entry. Add sysfs entry: /sysfs/kernel/slab//move With this users get access to the following functionality: - Move all objects to specified node. echo "N1" > move - Move all objects from specified node to other specified node (from N1 -> to N2): echo "N1 N2" > move This also enables shrinking slabs on a specific node: echo "N1 N1" > move Signed-off-by: Tobin C. Harding --- mm/Kconfig | 7 ++ mm/slub.c | 249 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 256 insertions(+) diff --git a/mm/Kconfig b/mm/Kconfig index ee8d1f311858..aa8d60e69a01 100644 --- a/mm/Kconfig +++ b/mm/Kconfig @@ -258,6 +258,13 @@ config ARCH_ENABLE_THP_MIGRATION config CONTIG_ALLOC def_bool (MEMORY_ISOLATION && COMPACTION) || CMA +config SMO_NODE + bool "Enable per node control of Slab Movable Objects" + depends on SLUB && SYSFS + select SLUB_DEBUG + help + On NUMA systems enable moving objects to and from a specified node. + config PHYS_ADDR_T_64BIT def_bool 64BIT diff --git a/mm/slub.c b/mm/slub.c index 2157205df7ba..9582f2fc97d2 100644 --- a/mm/slub.c +++ b/mm/slub.c @@ -4336,6 +4336,106 @@ static void move_slab_page(struct page *page, void *scratch, int node) s->migrate(s, vector, count, node, private); } +#ifdef CONFIG_SMO_NODE +/* + * kmem_cache_move() - Attempt to move all slab objects. + * @s: The cache we are working on. + * @node: The node to move objects away from. + * @target_node: The node to move objects on to. + * + * Attempts to move all objects (partial slabs and full slabs) to target + * node. + * + * Context: Takes the list_lock. + * Return: The number of slabs remaining on node. + */ +static unsigned long kmem_cache_move(struct kmem_cache *s, + int node, int target_node) +{ + struct kmem_cache_node *n = get_node(s, node); + LIST_HEAD(move_list); + struct page *page, *page2; + unsigned long flags; + void **scratch; + + if (!s->migrate) { + pr_warn("%s SMO not enabled, cannot move objects\n", s->name); + goto out; + } + + scratch = alloc_scratch(s); + if (!scratch) + goto out; + + spin_lock_irqsave(&n->list_lock, flags); + + list_for_each_entry_safe(page, page2, &n->partial, lru) { + if (!slab_trylock(page)) + /* Busy slab. Get out of the way */ + continue; + + if (page->inuse) { + list_move(&page->lru, &move_list); + /* Stop page being considered for allocations */ + n->nr_partial--; + page->frozen = 1; + + slab_unlock(page); + } else { /* Empty slab page */ + list_del(&page->lru); + n->nr_partial--; + slab_unlock(page); + discard_slab(s, page); + } + } + list_for_each_entry_safe(page, page2, &n->full, lru) { + if (!slab_trylock(page)) + continue; + + list_move(&page->lru, &move_list); + page->frozen = 1; + slab_unlock(page); + } + + spin_unlock_irqrestore(&n->list_lock, flags); + + list_for_each_entry(page, &move_list, lru) { + if (page->inuse) + move_slab_page(page, scratch, target_node); + } + kfree(scratch); + + /* Bail here to save taking the list_lock */ + if (list_empty(&move_list)) + goto out; + + /* Inspect results and dispose of pages */ + spin_lock_irqsave(&n->list_lock, flags); + list_for_each_entry_safe(page, page2, &move_list, lru) { + list_del(&page->lru); + slab_lock(page); + page->frozen = 0; + + if (page->inuse) { + if (page->inuse == page->objects) { + list_add(&page->lru, &n->full); + slab_unlock(page); + } else { + n->nr_partial++; + list_add_tail(&page->lru, &n->partial); + slab_unlock(page); + } + } else { + slab_unlock(page); + discard_slab(s, page); + } + } + spin_unlock_irqrestore(&n->list_lock, flags); +out: + return atomic_long_read(&n->nr_slabs); +} +#endif /* CONFIG_SMO_NODE */ + /* * kmem_cache_defrag() - Defragment node. * @s: cache we are working on. @@ -4450,6 +4550,32 @@ static unsigned long kmem_cache_defrag(struct kmem_cache *s, return n->nr_partial; } +#ifdef CONFIG_SMO_NODE +/* + * kmem_cache_move_to_node() - Move all slab objects to node. + * @s: The cache we are working on. + * @node: The target node to move objects to. + * + * Attempt to move all slab objects from all nodes to @node. + * + * Return: The total number of slabs left on emptied nodes. + */ +static unsigned long kmem_cache_move_to_node(struct kmem_cache *s, int node) +{ + unsigned long left = 0; + int nid; + + for_each_node_state(nid, N_NORMAL_MEMORY) { + if (nid == node) + continue; + + left += kmem_cache_move(s, nid, node); + } + + return left; +} +#endif + /** * kmem_defrag_slabs() - Defrag slab caches. * @node: The node to defrag or -1 for all nodes. @@ -5594,6 +5720,126 @@ static ssize_t shrink_store(struct kmem_cache *s, } SLAB_ATTR(shrink); +#ifdef CONFIG_SMO_NODE +static ssize_t move_show(struct kmem_cache *s, char *buf) +{ + return 0; +} + +/* + * parse_move_store_input() - Parse buf getting integer arguments. + * @buf: Buffer to parse. + * @length: Length of @buf. + * @arg0: Return parameter, first argument. + * @arg1: Return parameter, second argument. + * + * Parses the input from user write to sysfs file 'move'. Input string + * should contain either one or two node specifiers of form Nx where x + * is an integer specifying the NUMA node ID. 'N' or 'n' may be used. + * n/N may be omitted. + * + * e.g. + * echo 'N1' > /sysfs/kernel/slab/cache/move + * or + * echo 'N0 N2' > /sysfs/kernel/slab/cache/move + * + * Regex matching accepted forms: '[nN]?[0-9]( [nN]?[0-9])?' + * + * FIXME: This is really fragile. Input must be exactly correct, + * spurious whitespace causes parse errors. + * + * Return: 0 if an argument was successfully converted, or an error code. + */ +static ssize_t parse_move_store_input(const char *buf, size_t length, + long *arg0, long *arg1) +{ + char *s, *save, *ptr; + int ret = 0; + + if (!buf) + return -EINVAL; + + s = kstrdup(buf, GFP_KERNEL); + if (!s) + return -ENOMEM; + save = s; + + if (s[length - 1] == '\n') { + s[length - 1] = '\0'; + length--; + } + + ptr = strsep(&s, " "); + if (!ptr || strcmp(ptr, "") == 0) { + ret = 0; + goto out; + } + + if (*ptr == 'N' || *ptr == 'n') + ptr++; + ret = kstrtol(ptr, 10, arg0); + if (ret < 0) + goto out; + + if (s) { + if (*s == 'N' || *s == 'n') + s++; + ret = kstrtol(s, 10, arg1); + if (ret < 0) + goto out; + } + + ret = 0; +out: + kfree(save); + return ret; +} + +static bool is_valid_node(int node) +{ + int nid; + + for_each_node_state(nid, N_NORMAL_MEMORY) { + if (nid == node) + return true; + } + return false; +} + +/* + * move_store() - Move objects between nodes. + * @s: The cache we are working on. + * @buf: String received. + * @length: Length of @buf. + * + * Writes to /sys/kernel/slab//move are interpreted as follows: + * + * echo "N1" > move : Move all objects (from all nodes) to node 1. + * echo "N0 N1" > move : Move all objects from node 0 to node 1. + * + * 'N' may be omitted: + */ +static ssize_t move_store(struct kmem_cache *s, const char *buf, size_t length) +{ + long arg0 = -1; + long arg1 = -1; + int ret; + + ret = parse_move_store_input(buf, length, &arg0, &arg1); + if (ret < 0) + return -EINVAL; + + if (is_valid_node(arg0) && is_valid_node(arg1)) + (void)kmem_cache_move(s, arg0, arg1); + else if (is_valid_node(arg0)) + (void)kmem_cache_move_to_node(s, arg0); + + /* FIXME: What should we be returning here? */ + return length; +} +SLAB_ATTR(move); +#endif /* CONFIG_SMO_NODE */ + #ifdef CONFIG_NUMA static ssize_t remote_node_defrag_ratio_show(struct kmem_cache *s, char *buf) { @@ -5718,6 +5964,9 @@ static struct attribute *slab_attrs[] = { &reclaim_account_attr.attr, &destroy_by_rcu_attr.attr, &shrink_attr.attr, +#ifdef CONFIG_SMO_NODE + &move_attr.attr, +#endif &slabs_cpu_partial_attr.attr, #ifdef CONFIG_SLUB_DEBUG &total_objects_attr.attr, -- 2.21.0