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=-9.0 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_PASS,UNPARSEABLE_RELAY, 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 76AEFC43381 for ; Sat, 23 Mar 2019 04:45:23 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 426C6218E2 for ; Sat, 23 Mar 2019 04:45:23 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727393AbfCWEpV (ORCPT ); Sat, 23 Mar 2019 00:45:21 -0400 Received: from out30-54.freemail.mail.aliyun.com ([115.124.30.54]:58092 "EHLO out30-54.freemail.mail.aliyun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727183AbfCWEpS (ORCPT ); Sat, 23 Mar 2019 00:45:18 -0400 X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R641e4;CH=green;DM=||false|;FP=0|-1|-1|-1|0|-1|-1|-1;HT=e01f04397;MF=yang.shi@linux.alibaba.com;NM=1;PH=DS;RN=14;SR=0;TI=SMTPD_---0TNPuxAM_1553316293; Received: from e19h19392.et15sqa.tbsite.net(mailfrom:yang.shi@linux.alibaba.com fp:SMTPD_---0TNPuxAM_1553316293) by smtp.aliyun-inc.com(127.0.0.1); Sat, 23 Mar 2019 12:45:01 +0800 From: Yang Shi To: mhocko@suse.com, mgorman@techsingularity.net, riel@surriel.com, hannes@cmpxchg.org, akpm@linux-foundation.org, dave.hansen@intel.com, keith.busch@intel.com, dan.j.williams@intel.com, fengguang.wu@intel.com, fan.du@intel.com, ying.huang@intel.com Cc: yang.shi@linux.alibaba.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: [PATCH 02/10] mm: mempolicy: introduce MPOL_HYBRID policy Date: Sat, 23 Mar 2019 12:44:27 +0800 Message-Id: <1553316275-21985-3-git-send-email-yang.shi@linux.alibaba.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1553316275-21985-1-git-send-email-yang.shi@linux.alibaba.com> References: <1553316275-21985-1-git-send-email-yang.shi@linux.alibaba.com> Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Introduce a new NUMA policy, MPOL_HYBRID. It behaves like MPOL_BIND, but since we need migrate pages from non-DRAM node (i.e. PMEM node) to DRAM node on demand, MPOL_HYBRID would do page migration on numa fault, so it would have MPOL_F_MOF set by default. The NUMA balancing stuff will be enabled in the following patch. Signed-off-by: Yang Shi --- include/uapi/linux/mempolicy.h | 1 + mm/mempolicy.c | 56 +++++++++++++++++++++++++++++++++++++----- 2 files changed, 51 insertions(+), 6 deletions(-) diff --git a/include/uapi/linux/mempolicy.h b/include/uapi/linux/mempolicy.h index 3354774..0fdc73d 100644 --- a/include/uapi/linux/mempolicy.h +++ b/include/uapi/linux/mempolicy.h @@ -22,6 +22,7 @@ enum { MPOL_BIND, MPOL_INTERLEAVE, MPOL_LOCAL, + MPOL_HYBRID, MPOL_MAX, /* always last member of enum */ }; diff --git a/mm/mempolicy.c b/mm/mempolicy.c index af171cc..7d0a432 100644 --- a/mm/mempolicy.c +++ b/mm/mempolicy.c @@ -31,6 +31,10 @@ * but useful to set in a VMA when you have a non default * process policy. * + * hybrid Only allocate memory on specific set of nodes. If the set of + * nodes include non-DRAM nodes, NUMA balancing would promote + * the page to DRAM node. + * * default Allocate on the local node first, or when on a VMA * use the process policy. This is what Linux always did * in a NUMA aware kernel and still does by, ahem, default. @@ -191,6 +195,17 @@ static int mpol_new_bind(struct mempolicy *pol, const nodemask_t *nodes) return 0; } +static int mpol_new_hybrid(struct mempolicy *pol, const nodemask_t *nodes) +{ + if (nodes_empty(*nodes)) + return -EINVAL; + + /* Hybrid policy would promote pages in page fault */ + pol->flags |= MPOL_F_MOF; + pol->v.nodes = *nodes; + return 0; +} + /* * mpol_set_nodemask is called after mpol_new() to set up the nodemask, if * any, for the new policy. mpol_new() has already validated the nodes @@ -401,6 +416,10 @@ void mpol_rebind_mm(struct mm_struct *mm, nodemask_t *new) .create = mpol_new_bind, .rebind = mpol_rebind_nodemask, }, + [MPOL_HYBRID] = { + .create = mpol_new_hybrid, + .rebind = mpol_rebind_nodemask, + }, }; static void migrate_page_add(struct page *page, struct list_head *pagelist, @@ -782,6 +801,8 @@ static void get_policy_nodemask(struct mempolicy *p, nodemask_t *nodes) return; switch (p->mode) { + case MPOL_HYBRID: + /* Fall through */ case MPOL_BIND: /* Fall through */ case MPOL_INTERLEAVE: @@ -1721,8 +1742,12 @@ static int apply_policy_zone(struct mempolicy *policy, enum zone_type zone) */ static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy) { - /* Lower zones don't get a nodemask applied for MPOL_BIND */ - if (unlikely(policy->mode == MPOL_BIND) && + /* + * Lower zones don't get a nodemask applied for MPOL_BIND + * or MPOL_HYBRID. + */ + if (unlikely((policy->mode == MPOL_BIND) || + (policy->mode == MPOL_HYBRID)) && apply_policy_zone(policy, gfp_zone(gfp)) && cpuset_nodemask_valid_mems_allowed(&policy->v.nodes)) return &policy->v.nodes; @@ -1742,7 +1767,9 @@ static int policy_node(gfp_t gfp, struct mempolicy *policy, * because we might easily break the expectation to stay on the * requested node and not break the policy. */ - WARN_ON_ONCE(policy->mode == MPOL_BIND && (gfp & __GFP_THISNODE)); + WARN_ON_ONCE((policy->mode == MPOL_BIND || + policy->mode == MPOL_HYBRID) && + (gfp & __GFP_THISNODE)); } return nd; @@ -1786,6 +1813,8 @@ unsigned int mempolicy_slab_node(void) case MPOL_INTERLEAVE: return interleave_nodes(policy); + case MPOL_HYBRID: + /* Fall through */ case MPOL_BIND: { struct zoneref *z; @@ -1856,7 +1885,8 @@ static inline unsigned interleave_nid(struct mempolicy *pol, * @addr: address in @vma for shared policy lookup and interleave policy * @gfp_flags: for requested zone * @mpol: pointer to mempolicy pointer for reference counted mempolicy - * @nodemask: pointer to nodemask pointer for MPOL_BIND nodemask + * @nodemask: pointer to nodemask pointer for MPOL_BIND or MPOL_HYBRID + * nodemask * * Returns a nid suitable for a huge page allocation and a pointer * to the struct mempolicy for conditional unref after allocation. @@ -1871,14 +1901,16 @@ int huge_node(struct vm_area_struct *vma, unsigned long addr, gfp_t gfp_flags, int nid; *mpol = get_vma_policy(vma, addr); - *nodemask = NULL; /* assume !MPOL_BIND */ + /* assume !MPOL_BIND || !MPOL_HYBRID */ + *nodemask = NULL; if (unlikely((*mpol)->mode == MPOL_INTERLEAVE)) { nid = interleave_nid(*mpol, vma, addr, huge_page_shift(hstate_vma(vma))); } else { nid = policy_node(gfp_flags, *mpol, numa_node_id()); - if ((*mpol)->mode == MPOL_BIND) + if ((*mpol)->mode == MPOL_BIND || + (*mpol)->mode == MPOL_HYBRID) *nodemask = &(*mpol)->v.nodes; } return nid; @@ -1919,6 +1951,8 @@ bool init_nodemask_of_mempolicy(nodemask_t *mask) init_nodemask_of_node(mask, nid); break; + case MPOL_HYBRID: + /* Fall through */ case MPOL_BIND: /* Fall through */ case MPOL_INTERLEAVE: @@ -1966,6 +2000,7 @@ bool mempolicy_nodemask_intersects(struct task_struct *tsk, * nodes in mask. */ break; + case MPOL_HYBRID: case MPOL_BIND: case MPOL_INTERLEAVE: ret = nodes_intersects(mempolicy->v.nodes, *mask); @@ -2170,6 +2205,8 @@ bool __mpol_equal(struct mempolicy *a, struct mempolicy *b) return false; switch (a->mode) { + case MPOL_HYBRID: + /* Fall through */ case MPOL_BIND: /* Fall through */ case MPOL_INTERLEAVE: @@ -2325,6 +2362,9 @@ int mpol_misplaced(struct page *page, struct vm_area_struct *vma, unsigned long polnid = pol->v.preferred_node; break; + case MPOL_HYBRID: + /* Fall through */ + case MPOL_BIND: /* @@ -2693,6 +2733,7 @@ void numa_default_policy(void) [MPOL_BIND] = "bind", [MPOL_INTERLEAVE] = "interleave", [MPOL_LOCAL] = "local", + [MPOL_HYBRID] = "hybrid", }; @@ -2768,6 +2809,8 @@ int mpol_parse_str(char *str, struct mempolicy **mpol) if (!nodelist) err = 0; goto out; + case MPOL_HYBRID: + /* Fall through */ case MPOL_BIND: /* * Insist on a nodelist @@ -2856,6 +2899,7 @@ void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol) else node_set(pol->v.preferred_node, nodes); break; + case MPOL_HYBRID: case MPOL_BIND: case MPOL_INTERLEAVE: nodes = pol->v.nodes; -- 1.8.3.1