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 vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 6D966C001B0 for ; Mon, 17 Jul 2023 13:14:00 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229820AbjGQNN7 (ORCPT ); Mon, 17 Jul 2023 09:13:59 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:52570 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229496AbjGQNN5 (ORCPT ); Mon, 17 Jul 2023 09:13:57 -0400 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by lindbergh.monkeyblade.net (Postfix) with ESMTP id 7242594 for ; Mon, 17 Jul 2023 06:13:56 -0700 (PDT) Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 481F9C15; Mon, 17 Jul 2023 06:14:39 -0700 (PDT) Received: from [10.57.76.30] (unknown [10.57.76.30]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id A63E23F67D; Mon, 17 Jul 2023 06:13:53 -0700 (PDT) Message-ID: Date: Mon, 17 Jul 2023 14:13:52 +0100 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.13.0 Subject: Re: [PATCH v3 1/4] mm: Non-pmd-mappable, large folios for folio_add_new_anon_rmap() To: David Hildenbrand , Andrew Morton , Matthew Wilcox , "Kirill A. Shutemov" , Yin Fengwei , Yu Zhao , Catalin Marinas , Will Deacon , Anshuman Khandual , Yang Shi , "Huang, Ying" , Zi Yan , Luis Chamberlain Cc: linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org References: <20230714160407.4142030-1-ryan.roberts@arm.com> <20230714161733.4144503-1-ryan.roberts@arm.com> From: Ryan Roberts In-Reply-To: Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 17/07/2023 14:00, David Hildenbrand wrote: > On 14.07.23 18:17, Ryan Roberts wrote: >> In preparation for FLEXIBLE_THP support, improve >> folio_add_new_anon_rmap() to allow a non-pmd-mappable, large folio to be >> passed to it. In this case, all contained pages are accounted using the >> order-0 folio (or base page) scheme. >> >> Signed-off-by: Ryan Roberts >> Reviewed-by: Yu Zhao >> Reviewed-by: Yin Fengwei >> --- >>   mm/rmap.c | 28 +++++++++++++++++++++------- >>   1 file changed, 21 insertions(+), 7 deletions(-) >> >> diff --git a/mm/rmap.c b/mm/rmap.c >> index 0c0d8857dfce..f293d072368a 100644 >> --- a/mm/rmap.c >> +++ b/mm/rmap.c >> @@ -1278,31 +1278,45 @@ void page_add_anon_rmap(struct page *page, struct >> vm_area_struct *vma, >>    * This means the inc-and-test can be bypassed. >>    * The folio does not have to be locked. >>    * >> - * If the folio is large, it is accounted as a THP.  As the folio >> + * If the folio is pmd-mappable, it is accounted as a THP.  As the folio >>    * is new, it's assumed to be mapped exclusively by a single process. >>    */ >>   void folio_add_new_anon_rmap(struct folio *folio, struct vm_area_struct *vma, >>           unsigned long address) >>   { >> -    int nr; >> +    int nr = folio_nr_pages(folio); >> >> -    VM_BUG_ON_VMA(address < vma->vm_start || address >= vma->vm_end, vma); >> +    VM_BUG_ON_VMA(address < vma->vm_start || >> +            address + (nr << PAGE_SHIFT) > vma->vm_end, vma); >>       __folio_set_swapbacked(folio); >> >> -    if (likely(!folio_test_pmd_mappable(folio))) { >> +    if (!folio_test_large(folio)) { > > Why remove the "likely" here? The patch itself does not change anything about > that condition. Good question; I'm not sure why. Will have to put it down to bad copy/paste fixup. Will put it back in the next version. > >>           /* increment count (starts at -1) */ >>           atomic_set(&folio->_mapcount, 0); >> -        nr = 1; >> +        __page_set_anon_rmap(folio, &folio->page, vma, address, 1); >> +    } else if (!folio_test_pmd_mappable(folio)) { >> +        int i; >> + >> +        for (i = 0; i < nr; i++) { >> +            struct page *page = folio_page(folio, i); >> + >> +            /* increment count (starts at -1) */ >> +            atomic_set(&page->_mapcount, 0); >> +            __page_set_anon_rmap(folio, page, vma, >> +                    address + (i << PAGE_SHIFT), 1); >> +        } >> + >> +        /* increment count (starts at 0) */ > > That comment is a bit misleading. We're not talking about a mapcount as in the > other cases here. Correct, I'm talking about _nr_pages_mapped, which starts 0, not -1 like _mapcount. The comment was intended to be in the style used in other similar places in rmap.c. I could change it to: "_nr_pages_mapped is 0-based, so set it to the number of pages in the folio" or remove it entirely? What do you prefer? > >> +        atomic_set(&folio->_nr_pages_mapped, nr); >>       } else { >>           /* increment count (starts at -1) */ >>           atomic_set(&folio->_entire_mapcount, 0); >>           atomic_set(&folio->_nr_pages_mapped, COMPOUND_MAPPED); >> -        nr = folio_nr_pages(folio); >> +        __page_set_anon_rmap(folio, &folio->page, vma, address, 1); >>           __lruvec_stat_mod_folio(folio, NR_ANON_THPS, nr); >>       } >> > > Apart from that, LGTM. >