From patchwork Tue Mar 16 02:30:02 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Miaohe Lin X-Patchwork-Id: 12141073 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=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham 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 BA83FC433DB for ; Tue, 16 Mar 2021 02:30:49 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id D7CAE6502A for ; Tue, 16 Mar 2021 02:30:48 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org D7CAE6502A Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=huawei.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id 4C9CA6B006C; Mon, 15 Mar 2021 22:30:48 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 47A696B006E; Mon, 15 Mar 2021 22:30:48 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 342B36B0070; Mon, 15 Mar 2021 22:30:48 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0158.hostedemail.com [216.40.44.158]) by kanga.kvack.org (Postfix) with ESMTP id 1A69B6B006C for ; Mon, 15 Mar 2021 22:30:48 -0400 (EDT) Received: from smtpin07.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay05.hostedemail.com (Postfix) with ESMTP id BBEE51803DDA6 for ; Tue, 16 Mar 2021 02:30:47 +0000 (UTC) X-FDA: 77924159334.07.2CD56D9 Received: from szxga06-in.huawei.com (szxga06-in.huawei.com [45.249.212.32]) by imf29.hostedemail.com (Postfix) with ESMTP id 915FFDA for ; Tue, 16 Mar 2021 02:30:46 +0000 (UTC) Received: from DGGEMS409-HUB.china.huawei.com (unknown [172.30.72.60]) by szxga06-in.huawei.com (SkyGuard) with ESMTP id 4Dzy0Q2PkNzkYtQ; Tue, 16 Mar 2021 10:29:10 +0800 (CST) Received: from huawei.com (10.175.104.175) by DGGEMS409-HUB.china.huawei.com (10.3.19.209) with Microsoft SMTP Server id 14.3.498.0; Tue, 16 Mar 2021 10:30:33 +0800 From: Miaohe Lin To: , CC: , , , , , Subject: [PATCH v3] hugetlb_cgroup: fix imbalanced css_get and css_put pair for shared mappings Date: Mon, 15 Mar 2021 22:30:02 -0400 Message-ID: <20210316023002.53921-1-linmiaohe@huawei.com> X-Mailer: git-send-email 2.19.1 MIME-Version: 1.0 X-Originating-IP: [10.175.104.175] X-CFilter-Loop: Reflected X-Stat-Signature: zjxghqhb6detzmhw77eybrtiisjyc8sw X-Rspamd-Server: rspam02 X-Rspamd-Queue-Id: 915FFDA Received-SPF: none (huawei.com>: No applicable sender policy available) receiver=imf29; identity=mailfrom; envelope-from=""; helo=szxga06-in.huawei.com; client-ip=45.249.212.32 X-HE-DKIM-Result: none/none X-HE-Tag: 1615861846-111358 X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: The current implementation of hugetlb_cgroup for shared mappings could have different behavior. Consider the following two scenarios: 1.Assume initial css reference count of hugetlb_cgroup is 1: 1.1 Call hugetlb_reserve_pages with from = 1, to = 2. So css reference count is 2 associated with 1 file_region. 1.2 Call hugetlb_reserve_pages with from = 2, to = 3. So css reference count is 3 associated with 2 file_region. 1.3 coalesce_file_region will coalesce these two file_regions into one. So css reference count is 3 associated with 1 file_region now. 2.Assume initial css reference count of hugetlb_cgroup is 1 again: 2.1 Call hugetlb_reserve_pages with from = 1, to = 3. So css reference count is 2 associated with 1 file_region. Therefore, we might have one file_region while holding one or more css reference counts. This inconsistency could lead to imbalanced css_get() and css_put() pair. If we do css_put one by one (i.g. hole punch case), scenario 2 would put one more css reference. If we do css_put all together (i.g. truncate case), scenario 1 will leak one css reference. The imbalanced css_get() and css_put() pair would result in a non-zero reference when we try to destroy the hugetlb cgroup. The hugetlb cgroup directory is removed __but__ associated resource is not freed. This might result in OOM or can not create a new hugetlb cgroup in a busy workload ultimately. In order to fix this, we have to make sure that one file_region must hold exactly one css reference. So in coalesce_file_region case, we should release one css reference before coalescence. Also only put css reference when the entire file_region is removed. The last thing to note is that the caller of region_add() will only hold one reference to h_cg->css for the whole contiguous reservation region. But this area might be scattered when there are already some file_regions reside in it. As a result, many file_regions may share only one h_cg->css reference. In order to ensure that one file_region must hold exactly one css reference, we should do css_get() for each file_region and release the reference held by caller when they are done. Fixes: 075a61d07a8e ("hugetlb_cgroup: add accounting for shared mappings") Reported-by: kernel test robot (auto build test ERROR) Reviewed-by: Mike Kravetz Signed-off-by: Miaohe Lin Cc: Aneesh Kumar K.V Cc: Wanpeng Li Cc: Mina Almasry Cc: Mike Kravetz Cc: --- v2->v3: reshape some comments suggested by Mike Kravetz. collect Reviewed-by tag from Mike Kravetz. v1->v2: fix auto build test ERROR --- include/linux/hugetlb_cgroup.h | 15 +++++++++++-- mm/hugetlb.c | 41 ++++++++++++++++++++++++++++++---- mm/hugetlb_cgroup.c | 10 +++++++-- 3 files changed, 58 insertions(+), 8 deletions(-) diff --git a/include/linux/hugetlb_cgroup.h b/include/linux/hugetlb_cgroup.h index 2ad6e92f124a..0bff345c4bc6 100644 --- a/include/linux/hugetlb_cgroup.h +++ b/include/linux/hugetlb_cgroup.h @@ -113,6 +113,11 @@ static inline bool hugetlb_cgroup_disabled(void) return !cgroup_subsys_enabled(hugetlb_cgrp_subsys); } +static inline void hugetlb_cgroup_put_rsvd_cgroup(struct hugetlb_cgroup *h_cg) +{ + css_put(&h_cg->css); +} + extern int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, struct hugetlb_cgroup **ptr); extern int hugetlb_cgroup_charge_cgroup_rsvd(int idx, unsigned long nr_pages, @@ -138,7 +143,8 @@ extern void hugetlb_cgroup_uncharge_counter(struct resv_map *resv, extern void hugetlb_cgroup_uncharge_file_region(struct resv_map *resv, struct file_region *rg, - unsigned long nr_pages); + unsigned long nr_pages, + bool region_del); extern void hugetlb_cgroup_file_init(void) __init; extern void hugetlb_cgroup_migrate(struct page *oldhpage, @@ -147,7 +153,8 @@ extern void hugetlb_cgroup_migrate(struct page *oldhpage, #else static inline void hugetlb_cgroup_uncharge_file_region(struct resv_map *resv, struct file_region *rg, - unsigned long nr_pages) + unsigned long nr_pages, + bool region_del) { } @@ -185,6 +192,10 @@ static inline bool hugetlb_cgroup_disabled(void) return true; } +static inline void hugetlb_cgroup_put_rsvd_cgroup(struct hugetlb_cgroup *h_cg) +{ +} + static inline int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, struct hugetlb_cgroup **ptr) { diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 8fb42c6dd74b..09814be60412 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -280,6 +280,17 @@ static void record_hugetlb_cgroup_uncharge_info(struct hugetlb_cgroup *h_cg, nrg->reservation_counter = &h_cg->rsvd_hugepage[hstate_index(h)]; nrg->css = &h_cg->css; + /* + * The caller will hold exactly one h_cg->css reference for the + * whole contiguous reservation region. But this area might be + * scattered when there are already some file_regions reside in + * it. As a result, many file_regions may share only one css + * reference. In order to ensure that one file_region must hold + * exactly one h_cg->css reference, we should do css_get for + * each file_region and leave the reference held by caller + * untouched. + */ + css_get(&h_cg->css); if (!resv->pages_per_hpage) resv->pages_per_hpage = pages_per_huge_page(h); /* pages_per_hpage should be the same for all entries in @@ -293,6 +304,14 @@ static void record_hugetlb_cgroup_uncharge_info(struct hugetlb_cgroup *h_cg, #endif } +static void put_uncharge_info(struct file_region *rg) +{ +#ifdef CONFIG_CGROUP_HUGETLB + if (rg->css) + css_put(rg->css); +#endif +} + static bool has_same_uncharge_info(struct file_region *rg, struct file_region *org) { @@ -316,6 +335,7 @@ static void coalesce_file_region(struct resv_map *resv, struct file_region *rg) prg->to = rg->to; list_del(&rg->link); + put_uncharge_info(rg); kfree(rg); rg = prg; @@ -327,6 +347,7 @@ static void coalesce_file_region(struct resv_map *resv, struct file_region *rg) nrg->from = rg->from; list_del(&rg->link); + put_uncharge_info(rg); kfree(rg); } } @@ -659,7 +680,7 @@ static long region_del(struct resv_map *resv, long f, long t) del += t - f; hugetlb_cgroup_uncharge_file_region( - resv, rg, t - f); + resv, rg, t - f, false); /* New entry for end of split region */ nrg->from = t; @@ -680,7 +701,7 @@ static long region_del(struct resv_map *resv, long f, long t) if (f <= rg->from && t >= rg->to) { /* Remove entire region */ del += rg->to - rg->from; hugetlb_cgroup_uncharge_file_region(resv, rg, - rg->to - rg->from); + rg->to - rg->from, true); list_del(&rg->link); kfree(rg); continue; @@ -688,13 +709,13 @@ static long region_del(struct resv_map *resv, long f, long t) if (f <= rg->from) { /* Trim beginning of region */ hugetlb_cgroup_uncharge_file_region(resv, rg, - t - rg->from); + t - rg->from, false); del += t - rg->from; rg->from = t; } else { /* Trim end of region */ hugetlb_cgroup_uncharge_file_region(resv, rg, - rg->to - f); + rg->to - f, false); del += rg->to - f; rg->to = f; @@ -5128,6 +5149,10 @@ bool hugetlb_reserve_pages(struct inode *inode, */ long rsv_adjust; + /* + * hugetlb_cgroup_uncharge_cgroup_rsvd() will put the + * reference to h_cg->css. See comment below for detail. + */ hugetlb_cgroup_uncharge_cgroup_rsvd( hstate_index(h), (chg - add) * pages_per_huge_page(h), h_cg); @@ -5135,6 +5160,14 @@ bool hugetlb_reserve_pages(struct inode *inode, rsv_adjust = hugepage_subpool_put_pages(spool, chg - add); hugetlb_acct_memory(h, -rsv_adjust); + } else if (h_cg) { + /* + * The file_regions will hold their own reference to + * h_cg->css. So we should release the reference held + * via hugetlb_cgroup_charge_cgroup_rsvd() when we are + * done. + */ + hugetlb_cgroup_put_rsvd_cgroup(h_cg); } } return true; diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c index f68b51fcda3d..603a131e262d 100644 --- a/mm/hugetlb_cgroup.c +++ b/mm/hugetlb_cgroup.c @@ -391,7 +391,8 @@ void hugetlb_cgroup_uncharge_counter(struct resv_map *resv, unsigned long start, void hugetlb_cgroup_uncharge_file_region(struct resv_map *resv, struct file_region *rg, - unsigned long nr_pages) + unsigned long nr_pages, + bool region_del) { if (hugetlb_cgroup_disabled() || !resv || !rg || !nr_pages) return; @@ -400,7 +401,12 @@ void hugetlb_cgroup_uncharge_file_region(struct resv_map *resv, !resv->reservation_counter) { page_counter_uncharge(rg->reservation_counter, nr_pages * resv->pages_per_hpage); - css_put(rg->css); + /* + * Only do css_put(rg->css) when we delete the entire region + * because one file_region must hold exactly one css reference. + */ + if (region_del) + css_put(rg->css); } }