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 B89B7C61DA4 for ; Mon, 6 Mar 2023 12:40:02 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 4E62D6B0072; Mon, 6 Mar 2023 07:40:02 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 46FF6280001; Mon, 6 Mar 2023 07:40:02 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 310816B0074; Mon, 6 Mar 2023 07:40:02 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0012.hostedemail.com [216.40.44.12]) by kanga.kvack.org (Postfix) with ESMTP id 1B5AD6B0072 for ; Mon, 6 Mar 2023 07:40:02 -0500 (EST) Received: from smtpin03.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay04.hostedemail.com (Postfix) with ESMTP id E86411A0C26 for ; Mon, 6 Mar 2023 12:40:01 +0000 (UTC) X-FDA: 80538430602.03.5856774 Received: from out30-101.freemail.mail.aliyun.com (out30-101.freemail.mail.aliyun.com [115.124.30.101]) by imf25.hostedemail.com (Postfix) with ESMTP id BC923A0007 for ; Mon, 6 Mar 2023 12:39:58 +0000 (UTC) Authentication-Results: imf25.hostedemail.com; dkim=none; dmarc=pass (policy=none) header.from=alibaba.com; spf=pass (imf25.hostedemail.com: domain of xhao@linux.alibaba.com designates 115.124.30.101 as permitted sender) smtp.mailfrom=xhao@linux.alibaba.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1678106400; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=iIusJw/5F2DlEURCv+4uTnylFVKVafxSg2w+AcrRB6U=; b=WUkAz3kZ+e/ozoFSpmoQxnlgUZ2NM3llnt7huA3/46WJAKMthhg4IkSRJoUoJlGCxrHKGn d1apo72B/ftWKLFOlvGPQSuqx1xEt+tX6kIlqOLKwYIfGCgdYKw+s5kgdfyHQM3B/3/Unu FJ/w4SgEWtrTI2gMNLS3W0HwJfbTTiM= ARC-Authentication-Results: i=1; imf25.hostedemail.com; dkim=none; dmarc=pass (policy=none) header.from=alibaba.com; spf=pass (imf25.hostedemail.com: domain of xhao@linux.alibaba.com designates 115.124.30.101 as permitted sender) smtp.mailfrom=xhao@linux.alibaba.com ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1678106400; a=rsa-sha256; cv=none; b=77q59u6HHH3DPj//yizMdSSf3C7GUKzyDAqqhzFkrr3p7gLtOXcDj5HOzdNbHKGcKXiUM1 kOUJuXBMrRqgUExO8hMIxz4HMhPfZwgyf/ElhbHhvmxi0cvGOnCY3Uzb8OXYhUJN+IAtQ/ TIqylQYv8WjXs0Q1d/RVSpW2sjv1U7Q= X-Alimail-AntiSpam:AC=PASS;BC=-1|-1;BR=01201311R191e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=ay29a033018046051;MF=xhao@linux.alibaba.com;NM=1;PH=DS;RN=9;SR=0;TI=SMTPD_---0VdGQIlo_1678106391; Received: from 30.240.100.122(mailfrom:xhao@linux.alibaba.com fp:SMTPD_---0VdGQIlo_1678106391) by smtp.aliyun-inc.com; Mon, 06 Mar 2023 20:39:53 +0800 Message-ID: <047de2dd-e4e5-287f-2270-3ce356854aa7@linux.alibaba.com> Date: Mon, 6 Mar 2023 20:39:52 +0800 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.8.0 Subject: Re: [PATCH v3 5/5] try_to_unmap_one: batched remove rmap, update folio refcount To: Yin Fengwei , linux-mm@kvack.org, akpm@linux-foundation.org, willy@infradead.org, mike.kravetz@oracle.com, sidhartha.kumar@oracle.com, naoya.horiguchi@nec.com, jane.chu@oracle.com, david@redhat.com References: <20230306092259.3507807-1-fengwei.yin@intel.com> <20230306092259.3507807-6-fengwei.yin@intel.com> From: haoxin In-Reply-To: <20230306092259.3507807-6-fengwei.yin@intel.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-Rspamd-Queue-Id: BC923A0007 X-Stat-Signature: syc7wm57hrpiw7rp6r4y6dup8gihpedy X-Rspam-User: X-Rspamd-Server: rspam08 X-HE-Tag: 1678106398-329725 X-HE-Meta: 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 mvjrmFfx udXXWhNb1RPL0EOBJC6lJxKchxFnhhVjZtrDAA8y/M+LQpJu5K1vpX7h2fzWpG4C6iZNTMxYOV3M19sdV0Jc93bWPiSVO7GDjZEs2VZLBHiQ+AYQAEip4piDF4kR0yRoj+Cr4Goq2vAOfyY8pT9xLPJ/m+9kQszovC6J13puwHg1QNBs= 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: 在 2023/3/6 下午5:22, Yin Fengwei 写道: > If unmap one page fails, or the vma walk will skip next pte, > or the vma walk will end on next pte, batched remove map, > update folio refcount. > > Signed-off-by: Yin Fengwei > --- > include/linux/rmap.h | 1 + > mm/page_vma_mapped.c | 30 +++++++++++++++++++++++++++ > mm/rmap.c | 48 ++++++++++++++++++++++++++++++++++---------- > 3 files changed, 68 insertions(+), 11 deletions(-) > > diff --git a/include/linux/rmap.h b/include/linux/rmap.h > index d2569b42e21a..18193d1d5a8e 100644 > --- a/include/linux/rmap.h > +++ b/include/linux/rmap.h > @@ -424,6 +424,7 @@ static inline void page_vma_mapped_walk_done(struct page_vma_mapped_walk *pvmw) > } > > bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw); > +bool pvmw_walk_skip_or_end_on_next(struct page_vma_mapped_walk *pvmw); > > /* > * Used by swapoff to help locate where page is expected in vma. > diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c > index 4e448cfbc6ef..19e997dfb5c6 100644 > --- a/mm/page_vma_mapped.c > +++ b/mm/page_vma_mapped.c > @@ -291,6 +291,36 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw) > return false; > } > > +/** > + * pvmw_walk_skip_or_end_on_next - check if next pte will be skipped or > + * end the walk > + * @pvmw: pointer to struct page_vma_mapped_walk. > + * > + * This function can only be called with correct pte lock hold > + */ > +bool pvmw_walk_skip_or_end_on_next(struct page_vma_mapped_walk *pvmw) > +{ > + unsigned long address = pvmw->address + PAGE_SIZE; > + > + if (address >= vma_address_end(pvmw)) > + return true; If vma_address_end is exactly equal to next address(pvmw->address + PAGE_SIZE) , does this mean that we are ignored to unmap the last page here ? so there can just use ' > '      ' if (address > vma_address_end(pvmw))' . I may have misunderstood, please correct me. > + > + if ((address & (PMD_SIZE - PAGE_SIZE)) == 0) > + return true; > + > + if (pte_none(*pvmw->pte)) > + return true; > + > + pvmw->pte++; > + if (!check_pte(pvmw)) { > + pvmw->pte--; > + return true; > + } > + pvmw->pte--; > + > + return false; > +} > + > /** > * page_mapped_in_vma - check whether a page is really mapped in a VMA > * @page: the page to test > diff --git a/mm/rmap.c b/mm/rmap.c > index bb3fcb8df579..a64e9cbb52dd 100644 > --- a/mm/rmap.c > +++ b/mm/rmap.c > @@ -1741,6 +1741,26 @@ static bool try_to_unmap_one_page(struct folio *folio, > return false; > } > > +static void folio_remove_rmap_and_update_count(struct folio *folio, > + struct page *start, struct vm_area_struct *vma, int count) > +{ > + if (count == 0) > + return; > + > + /* > + * No need to call mmu_notifier_invalidate_range() it has be > + * done above for all cases requiring it to happen under page > + * table lock before mmu_notifier_invalidate_range_end() > + * > + * See Documentation/mm/mmu_notifier.rst > + */ > + folio_remove_rmap_range(folio, start, count, vma, > + folio_test_hugetlb(folio)); > + if (vma->vm_flags & VM_LOCKED) > + mlock_drain_local(); > + folio_ref_sub(folio, count); > +} > + > /* > * @arg: enum ttu_flags will be passed to this argument > */ > @@ -1748,10 +1768,11 @@ static bool try_to_unmap_one(struct folio *folio, struct vm_area_struct *vma, > unsigned long address, void *arg) > { > DEFINE_FOLIO_VMA_WALK(pvmw, folio, vma, address, 0); > - struct page *subpage; > + struct page *start = NULL; > bool ret = true; > struct mmu_notifier_range range; > enum ttu_flags flags = (enum ttu_flags)(long)arg; > + int count = 0; > > /* > * When racing against e.g. zap_pte_range() on another cpu, > @@ -1812,26 +1833,31 @@ static bool try_to_unmap_one(struct folio *folio, struct vm_area_struct *vma, > break; > } > > - subpage = folio_page(folio, > + if (!start) > + start = folio_page(folio, > pte_pfn(*pvmw.pte) - folio_pfn(folio)); > ret = try_to_unmap_one_page(folio, vma, > range, pvmw, address, flags); > if (!ret) { > + folio_remove_rmap_and_update_count(folio, > + start, vma, count); > page_vma_mapped_walk_done(&pvmw); > break; > } > + count++; > > /* > - * No need to call mmu_notifier_invalidate_range() it has be > - * done above for all cases requiring it to happen under page > - * table lock before mmu_notifier_invalidate_range_end() > - * > - * See Documentation/mm/mmu_notifier.rst > + * If next pte will be skipped in page_vma_mapped_walk() or > + * the walk will end at it, batched remove rmap and update > + * page refcount. We can't do it after page_vma_mapped_walk() > + * return false because the pte lock will not be hold. > */ > - page_remove_rmap(subpage, vma, false); > - if (vma->vm_flags & VM_LOCKED) > - mlock_drain_local(); > - folio_put(folio); > + if (pvmw_walk_skip_or_end_on_next(&pvmw)) { > + folio_remove_rmap_and_update_count(folio, > + start, vma, count); > + count = 0; > + start = NULL; > + } > } > > mmu_notifier_invalidate_range_end(&range);