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 Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by smtp.lore.kernel.org (Postfix) with ESMTP id 21586C433EF for ; Tue, 15 Mar 2022 14:00:35 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 834C58D0002; Tue, 15 Mar 2022 10:00:34 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 7E59B8D0001; Tue, 15 Mar 2022 10:00:34 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 6D3B68D0002; Tue, 15 Mar 2022 10:00:34 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0241.hostedemail.com [216.40.44.241]) by kanga.kvack.org (Postfix) with ESMTP id 5D6BD8D0001 for ; Tue, 15 Mar 2022 10:00:34 -0400 (EDT) Received: from smtpin24.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay03.hostedemail.com (Postfix) with ESMTP id 0E3DE8249980 for ; Tue, 15 Mar 2022 14:00:34 +0000 (UTC) X-FDA: 79246780788.24.4693FDA Received: from szxga01-in.huawei.com (szxga01-in.huawei.com [45.249.212.187]) by imf28.hostedemail.com (Postfix) with ESMTP id 164DEC000E for ; Tue, 15 Mar 2022 14:00:32 +0000 (UTC) Received: from canpemm500002.china.huawei.com (unknown [172.30.72.54]) by szxga01-in.huawei.com (SkyGuard) with ESMTP id 4KHw4R3Bw9zfYq8; Tue, 15 Mar 2022 21:59:03 +0800 (CST) Received: from [10.174.177.76] (10.174.177.76) by canpemm500002.china.huawei.com (7.192.104.244) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256) id 15.1.2308.21; Tue, 15 Mar 2022 22:00:30 +0800 Subject: Re: [PATCH v2] mm/hwpoison: set PageHWPoison after taking page lock in memory_failure_hugetlb() To: =?UTF-8?B?SE9SSUdVQ0hJIE5BT1lBKOWggOWPoyDnm7TkuZ8p?= CC: Naoya Horiguchi , Andrew Morton , Mike Kravetz , Yang Shi , "linux-kernel@vger.kernel.org" , Linux-MM References: <20220314021337.333781-1-naoya.horiguchi@linux.dev> <58732610-36a4-1f05-c09d-a5536013772d@huawei.com> <20220315054935.GA3454497@hori.linux.bs1.fc.nec.co.jp> From: Miaohe Lin Message-ID: <1770029b-fd59-4eb1-c891-5a2ba4beef9c@huawei.com> Date: Tue, 15 Mar 2022 22:00:29 +0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101 Thunderbird/78.6.0 MIME-Version: 1.0 In-Reply-To: <20220315054935.GA3454497@hori.linux.bs1.fc.nec.co.jp> Content-Type: text/plain; charset="utf-8" Content-Language: en-US X-Originating-IP: [10.174.177.76] X-ClientProxiedBy: dggems702-chm.china.huawei.com (10.3.19.179) To canpemm500002.china.huawei.com (7.192.104.244) X-CFilter-Loop: Reflected X-Rspamd-Server: rspam11 X-Rspamd-Queue-Id: 164DEC000E Authentication-Results: imf28.hostedemail.com; dkim=none; spf=pass (imf28.hostedemail.com: domain of linmiaohe@huawei.com designates 45.249.212.187 as permitted sender) smtp.mailfrom=linmiaohe@huawei.com; dmarc=pass (policy=quarantine) header.from=huawei.com X-Rspam-User: X-Stat-Signature: 7jn5jb58wbg7ew7u9pdnmdinzfkjpxzq X-HE-Tag: 1647352832-936292 Content-Transfer-Encoding: quoted-printable 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: On 2022/3/15 13:49, HORIGUCHI NAOYA(=E5=A0=80=E5=8F=A3 =E7=9B=B4=E4=B9=9F= ) wrote: > On Mon, Mar 14, 2022 at 03:10:25PM +0800, Miaohe Lin wrote: >> On 2022/3/14 10:13, Naoya Horiguchi wrote: >>> From: Naoya Horiguchi >>> >>> There is a race condition between memory_failure_hugetlb() and hugetl= b >>> free/demotion, which causes setting PageHWPoison flag on the wrong pa= ge >>> (which was a hugetlb when memory_failure() was called, but was remove= d >>> or demoted when memory_failure_hugetlb() is called). This results in >>> killing wrong processes. So set PageHWPoison flag with holding page = lock, >> >> It seems hold page lock could not help solve this race condition as hu= getlb >> page demotion is not required to hold the page lock. Could you please = explain >> this a bit more? >=20 > Sorry, the last line in the paragraph need change. What prevents the cu= rrent > race is hugetlb_lock, not page lock. The page lock is here to prevent = the > race with hugepage allocation (not directly related to the current issu= e, > but it's still necessary). Many thanks for clarifying this. >=20 >> >> BTW:Is there some words missing or here should be 'page lock.' instead= of 'page lock,' ? >=20 > I should use a period here, I'll fix it. >=20 > [...] >=20 >>> @@ -1503,24 +1502,11 @@ static int memory_failure_hugetlb(unsigned lo= ng pfn, int flags) >>> int res; >>> unsigned long page_flags; >>> =20 >>> - if (TestSetPageHWPoison(head)) { >>> - pr_err("Memory failure: %#lx: already hardware poisoned\n", >>> - pfn); >>> - res =3D -EHWPOISON; >>> - if (flags & MF_ACTION_REQUIRED) >>> - res =3D kill_accessing_process(current, page_to_pfn(head), flags)= ; >>> - return res; >>> - } >>> - >>> - num_poisoned_pages_inc(); >>> - >>> if (!(flags & MF_COUNT_INCREASED)) { >>> res =3D get_hwpoison_page(p, flags); >>> if (!res) { >> >> In this (res =3D=3D 0) case, hugetlb page could be dissolved via __pag= e_handle_poison. >> But since PageHWPoison is not set yet, we can't set the PageHWPoison t= o the correct >> page. Think about the below code in dissolve_free_huge_page: >> /* >> * Move PageHWPoison flag from head page to the raw >> * error page, which makes any subpages rather than >> * the error page reusable. >> */ >> if (PageHWPoison(head) && page !=3D head) { >> SetPageHWPoison(page); >> ClearPageHWPoison(head); >> } >> >> SetPageHWPoison won't be called for the error page. Or am I miss somet= hing? >=20 > No, you're right. We need call page_handle_poison() instead of > __page_handle_poison(). >=20 > @@ -1512,7 +1512,7 @@ static int memory_failure_hugetlb(unsigned long p= fn, int flags) > } > unlock_page(head); > res =3D MF_FAILED; > - if (__page_handle_poison(p)) { > + if (page_handle_poison(p, true, false)) { > page_ref_inc(p); > res =3D MF_RECOVERED; > } >=20 This one looks good to me. >=20 >=20 >> >>> lock_page(head); >>> if (hwpoison_filter(p)) { >>> - if (TestClearPageHWPoison(head)) >>> - num_poisoned_pages_dec(); >>> unlock_page(head); >>> return -EOPNOTSUPP; >>> } >>> @@ -1553,13 +1539,16 @@ static int memory_failure_hugetlb(unsigned lo= ng pfn, int flags) >>> page_flags =3D head->flags; >>> =20 >>> if (hwpoison_filter(p)) { >>> - if (TestClearPageHWPoison(head)) >>> - num_poisoned_pages_dec(); >>> put_page(p); >>> res =3D -EOPNOTSUPP; >>> goto out; >>> } >>> =20 >>> + if (TestSetPageHWPoison(head)) >>> + goto already_hwpoisoned; >>> + >>> + num_poisoned_pages_inc(); >>> + >>> /* >>> * TODO: hwpoison for pud-sized hugetlb doesn't work right now, so >>> * simply disable it. In order to make it work properly, we need >>> @@ -1585,6 +1574,14 @@ static int memory_failure_hugetlb(unsigned lon= g pfn, int flags) >>> out: >>> unlock_page(head); >>> return res; >>> +already_hwpoisoned: >>> + put_page(p); >>> + unlock_page(head); >> >> Generally speaking, we should do unlock_page before put_page or page m= ight be disappeared >> before we unlock the page. This should be ok when memory_failure succe= eds to handle the >> page previously as it holds one extra page refcnt. But it might be pro= blematic when >> memory_failure failed to handle the page last time. We might be the la= st user here. >=20 > OK, so another code path in "if (hwpoison_filter)@ block seems to need > the same change in the order. You're right. >=20 > Thanks, > Naoya Horiguchi >=20 Many thanks for your patch.