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 B64A8C83F1B for ; Mon, 14 Jul 2025 13:16:30 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 52F9A6B007B; Mon, 14 Jul 2025 09:16:30 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 4DFCC6B0089; Mon, 14 Jul 2025 09:16:30 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 3CE546B008A; Mon, 14 Jul 2025 09:16:30 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0017.hostedemail.com [216.40.44.17]) by kanga.kvack.org (Postfix) with ESMTP id 276496B007B for ; Mon, 14 Jul 2025 09:16:30 -0400 (EDT) Received: from smtpin30.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay01.hostedemail.com (Postfix) with ESMTP id ABD191D4A7F for ; Mon, 14 Jul 2025 13:16:29 +0000 (UTC) X-FDA: 83662919298.30.AF62341 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by imf24.hostedemail.com (Postfix) with ESMTP id 5791C18000F for ; Mon, 14 Jul 2025 13:16:27 +0000 (UTC) Authentication-Results: imf24.hostedemail.com; dkim=pass header.d=redhat.com header.s=mimecast20190719 header.b=KYCJGS4l; spf=pass (imf24.hostedemail.com: domain of luizcap@redhat.com designates 170.10.133.124 as permitted sender) smtp.mailfrom=luizcap@redhat.com; dmarc=pass (policy=quarantine) header.from=redhat.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1752498987; h=from:from:sender:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references:dkim-signature; bh=Ib77k7eAfBr6oqef0Ex2EcxB2hnN68oB0e1cVj8uYzo=; b=6DB48RxM0pRGkTlwktL7q3YSD2jw7AEsC7DBId6jz4CvUt3w0FIUw0mE6+lj/g4RDZVHo0 J2PbrFpvscbgZQc9ZLUEjX4R4haccGFWNlUhdwhETIbeU3+A8SoKymCHFka2U1/U5YStHt +0g+rl4aJzNnnya0H88XK2NdoNWlXuU= ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1752498987; a=rsa-sha256; cv=none; b=mGg1H+soAf4O/h8A1sf5tl2Q0II25reVXuOnvseg2YKMY2z6t1ubgq5sXYJpFnrsjH2ovN nZkuWeVwtqP3LG/SdmMxjfMq0Nv2qVhTSqMkfLv5qskEK8gskesbwGKYWQ8QPs3Gh4Js75 dejKHukUwoz42Yi6ve40eLd9qaV3gpI= ARC-Authentication-Results: i=1; imf24.hostedemail.com; dkim=pass header.d=redhat.com header.s=mimecast20190719 header.b=KYCJGS4l; spf=pass (imf24.hostedemail.com: domain of luizcap@redhat.com designates 170.10.133.124 as permitted sender) smtp.mailfrom=luizcap@redhat.com; dmarc=pass (policy=quarantine) header.from=redhat.com DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1752498986; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=Ib77k7eAfBr6oqef0Ex2EcxB2hnN68oB0e1cVj8uYzo=; b=KYCJGS4lcWaqhZtfV9It5xgN2ILYvgO41FaQH3Pnj1eQI17i7i8W0ybiE6okqNJn+oRLUd A7pO3E812D4aPjsTGK9wNzSdqJGF2ioSM9xFsNe2W1bAHPCiBHaOQburjWRji0SWwAxWZx 8emLPbCK4Z32l1NYSY81cIZMHDF6mVg= Received: from mail-qk1-f200.google.com (mail-qk1-f200.google.com [209.85.222.200]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-604-MuE3gHJsMq6s6yLr76Yl_g-1; Mon, 14 Jul 2025 09:16:25 -0400 X-MC-Unique: MuE3gHJsMq6s6yLr76Yl_g-1 X-Mimecast-MFC-AGG-ID: MuE3gHJsMq6s6yLr76Yl_g_1752498985 Received: by mail-qk1-f200.google.com with SMTP id af79cd13be357-7e278d83457so166985585a.2 for ; Mon, 14 Jul 2025 06:16:25 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1752498985; x=1753103785; h=content-transfer-encoding:in-reply-to:content-language:references :cc:to:from:subject:user-agent:mime-version:date:message-id :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=Ib77k7eAfBr6oqef0Ex2EcxB2hnN68oB0e1cVj8uYzo=; b=EDshLh1RPV1oRT+Av6yahyPZU6BkBx6w2fIEd7jnU/sN95ZmO8gOhNQCcMOhtYt0px 7jf32zXQSsAGEJdYZ7OV+vYp0xNaEEYywWYBMVx2bVnq+NlU8LAsoYuMqNDOKqrH/MDP LCziI8LzjAkvLAsc1yenVEl7bT7kex7tZkEf/UzQJYlm8ckUyCzRs6BOh4+nwDLBKqFY agv50RHmW3EzPtMwEgZXCzFVkLMpkVz9CsNVj0QQ0z+ZDH6hUm8OUlkE28uWN+rEHSSm V0TzXO93exJhcCHs4X+sODDJBQZNejlQ6voVB944+FiOWtB+WOPC+le/VjqFc9us8D6G K1yg== X-Forwarded-Encrypted: i=1; AJvYcCWGVsjWnTHAEhF4EDhdCFZNe3Du/ngXXgJ2PQvoNYnxeLaXWeZyIrmP5MWfs/FbQjJpsVCDxMSNig==@kvack.org X-Gm-Message-State: AOJu0Yw8Cx1TLKNE6/GrmZvhP8bgq8U8Usjs67GrAXy7Dh+w23LCRKZU kPo20rov9cHlp3AzXfUBORBhWFORrn8Q00hjfPiuQZui0c4124lvkRaJGEs2syE/Nito61PzMiQ 6kB7d6rMsvPbuagFT4NT/wOnvpIYs8o9S6KmNDqSCV2BL7gLFMcqF X-Gm-Gg: ASbGncsKH8ew6JbfsscNZfoREo9/y8aMprtMiHsrk7pONU9lUX5DlpvA32Z+92v+4jD 9YseFvFpT1WM7VZS7rkC0VMRqd4nfxWIKxptdvm0yoxIIHyrVyicg39oJsGeFkagIxmnt9JOUT8 ldwcHt9b2UzT4iE9qi/7fhyCQBpC9zr44PA+Ph9Ft1kT+JnCi8tW4vkHIcG52W/PCiDkGCaLFsF s8vN55wPUPaau0b8aUMWJIPUrwjYlLBFcrB4svV3rv6l9lFfda1arl1WFnLkXpL437tV9lz8eLM W4Kdnsoi26C0/si9luVyjkmWzuitC2yFkMhYSS1Z X-Received: by 2002:a05:620a:290f:b0:7e3:3604:5047 with SMTP id af79cd13be357-7e336045076mr2677385a.38.1752498984777; Mon, 14 Jul 2025 06:16:24 -0700 (PDT) X-Google-Smtp-Source: AGHT+IFPUrFGR4YDRevL4SLsVtbgKHacggj9VWxy5bSHPv6taO97ydd55xZFzgAfWdcEMNR+dSDZcw== X-Received: by 2002:a05:620a:290f:b0:7e3:3604:5047 with SMTP id af79cd13be357-7e336045076mr2674085a.38.1752498984323; Mon, 14 Jul 2025 06:16:24 -0700 (PDT) Received: from [192.168.2.110] ([70.29.229.84]) by smtp.gmail.com with ESMTPSA id af79cd13be357-7dcdbb0ceaesm506633285a.13.2025.07.14.06.16.23 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Mon, 14 Jul 2025 06:16:24 -0700 (PDT) Message-ID: <017e17e9-c9d7-4dc2-90bc-0ef06289007c@redhat.com> Date: Mon, 14 Jul 2025 09:16:13 -0400 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH v2 2/4] mm/util: introduce snapshot_page() From: Luiz Capitulino To: Shivank Garg , david@redhat.com, willy@infradead.org, akpm@linux-foundation.org Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, sj@kernel.org References: <88d956248f7528b7998ef00ca8742061839d1036.1751914235.git.luizcap@redhat.com> <3942336d-20e6-457d-8b53-e198eed5d9d9@amd.com> <60a1a526-52dd-478a-9ed4-79e6428743ac@redhat.com> In-Reply-To: <60a1a526-52dd-478a-9ed4-79e6428743ac@redhat.com> X-Mimecast-Spam-Score: 0 X-Mimecast-MFC-PROC-ID: Hxb0qhcd4U2PI_4Sp4Lpxh3b8wc-83YWP6uhaLiUWes_1752498985 X-Mimecast-Originator: redhat.com Content-Language: en-US, en-CA Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit X-Rspamd-Queue-Id: 5791C18000F X-Stat-Signature: ntgqsb7qa9c3tg47a4zif4e1cyordx6k X-Rspam-User: X-Rspamd-Server: rspam07 X-HE-Tag: 1752498987-190774 X-HE-Meta: U2FsdGVkX187f9fxYhaSbvRuN5gNXv3SdClanNmq/FsSiHVa39BSJEV3TJRIEkaaSM3/G/fE7qlozqzEQ66haBJfM1okUh/Xl3Tv+pwYZ3CLv7dtPEUdfirBapd6UbUGotUD2ouv/fAeptChY1nrHho+FiZn2woCXRSf03Ka43rQsIiTv6McYUOJrZcrlKIgeR9UjX8SC8kcKE4+oqjfHVT+w8ftv0kjviu6sFo+kyvPPA/h0N1JuRpEftSaO+54fwN/+U9+s5IiY7p3wrg6VB76FxyH03ExLyru1SmWU0OjNA6rn/p3zYda32SU7GAifz6m4+XLcR9kzff9GgFA1vERXjEpUjNCYEbKn6HOkNvYtb679BIlr6e5pC8j1sORP2Ll5jc1E9E3yO89fka4QohdIROAebHUYMbMyxK+M7fxWR87rcD6o3pdJIEK7pxitnNC3TN6iwu3yEEqjLtzjl+ziU1yXou9pICDQPxie7T7PYj2wXBspgW/yCToNJzKLEj1xHDGXlrPEeooPTrLKPeyl+ChBOcevcf2w28xSIaVarIdWgorY+d6uyOaGGjgt5F4LkJd4ezhTNe3OGa3S9V4/CWAzXWqgf+oxiZElANl3yCSBnF41BS15W+fipZ90FZFkqIMSA71Y3nujm52UFGeXpdwtWYGJaPi+uf8sxfYLdqv6TiHGexuYwjwotF02dIVPN+lvsV4Q2BjtyJKnJQ+OelKbfENWko2+Z+SMXBLmagZJ5b/xVSQnAJE08i5fT5BimQ+3alZCVyb4QVjT5XKnuWZhbHgnwNyH7gkT1mLYgKIgDisO1KKYnopJrfvlpku1/QEPpNNiMSxJ/v9tErnu0QNIaXUz5rv9a6YqgM8UJrubnshWkGL0SgMNvO4u55gSz/W0tr+/lR/UidZR/mb62qr5bRVT0IQfdFkn0+D2Jm5SpBZXreyRwFesHh000+UoCA3oObg4zGHjSU UDzypf1U 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 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: List-Subscribe: List-Unsubscribe: On 2025-07-08 12:59, Luiz Capitulino wrote: > On 2025-07-08 01:49, Shivank Garg wrote: >> >> >> On 7/8/2025 12:20 AM, Luiz Capitulino wrote: >>> This commit refactors __dump_page() into snapshot_page(). >>> >>> snapshot_page() tries to take a faithful snapshot of a page and its >>> folio representation. The snapshot is returned in the struct >>> page_snapshot parameter along with additional flags that are best >>> retrieved at snapshot creation time to reduce race windows. >>> >>> This function is intended to be used by callers that need a stable >>> representation of a struct page and struct folio so that pointers >>> or page information doesn't change while working on a page. >>> >>> The idea and original implementation of snapshot_page() comes from >>> Matthew Wilcox with suggestions for improvements from David Hildenbrand. >>> All bugs and misconceptions are mine. >>> >>> Signed-off-by: Luiz Capitulino >>> --- >>>   include/linux/mm.h | 19 ++++++++++++ >>>   mm/debug.c         | 42 +++---------------------- >>>   mm/util.c          | 77 ++++++++++++++++++++++++++++++++++++++++++++++ >>>   3 files changed, 100 insertions(+), 38 deletions(-) >>> >>> diff --git a/include/linux/mm.h b/include/linux/mm.h >>> index 0ef2ba0c667a..090968c6eebb 100644 >>> --- a/include/linux/mm.h >>> +++ b/include/linux/mm.h >>> @@ -4184,4 +4184,23 @@ static inline bool page_pool_page_is_pp(struct page *page) >>>   } >>>   #endif >>> +#define PAGE_SNAPSHOT_FAITHFUL     (1 << 0) >>> +#define PAGE_SNAPSHOT_PG_FREE      (1 << 1) >>> +#define PAGE_SNAPSHOT_PG_IDLE      (1 << 2) >>> + >>> +struct page_snapshot { >>> +    struct folio folio_snapshot; >>> +    struct page page_snapshot; >>> +    unsigned long pfn; >>> +    unsigned long idx; >>> +    unsigned long flags; >>> +}; >>> + >>> +static inline bool snapshot_page_is_faithful(const struct page_snapshot *ps) >>> +{ >>> +    return ps->flags & PAGE_SNAPSHOT_FAITHFUL; >>> +} >>> + >>> +void snapshot_page(struct page_snapshot *ps, const struct page *page); >>> + >>>   #endif /* _LINUX_MM_H */ >>> diff --git a/mm/debug.c b/mm/debug.c >>> index 907382257062..7349330ea506 100644 >>> --- a/mm/debug.c >>> +++ b/mm/debug.c >>> @@ -129,47 +129,13 @@ static void __dump_folio(struct folio *folio, struct page *page, >>>   static void __dump_page(const struct page *page) >>>   { >>> -    struct folio *foliop, folio; >>> -    struct page precise; >>> -    unsigned long head; >>> -    unsigned long pfn = page_to_pfn(page); >>> -    unsigned long idx, nr_pages = 1; >>> -    int loops = 5; >>> - >>> -again: >>> -    memcpy(&precise, page, sizeof(*page)); >>> -    head = precise.compound_head; >>> -    if ((head & 1) == 0) { >>> -        foliop = (struct folio *)&precise; >>> -        idx = 0; >>> -        if (!folio_test_large(foliop)) >>> -            goto dump; >>> -        foliop = (struct folio *)page; >>> -    } else { >>> -        foliop = (struct folio *)(head - 1); >>> -        idx = folio_page_idx(foliop, page); >>> -    } >>> +    struct page_snapshot ps; >>> -    if (idx < MAX_FOLIO_NR_PAGES) { >>> -        memcpy(&folio, foliop, 2 * sizeof(struct page)); >>> -        nr_pages = folio_nr_pages(&folio); >>> -        if (nr_pages > 1) >>> -            memcpy(&folio.__page_2, &foliop->__page_2, >>> -                   sizeof(struct page)); >>> -        foliop = &folio; >>> -    } >>> - >>> -    if (idx > nr_pages) { >>> -        if (loops-- > 0) >>> -            goto again; >>> +    snapshot_page(&ps, page); >>> +    if (!snapshot_page_is_faithful(&ps)) >>>           pr_warn("page does not match folio\n"); >>> -        precise.compound_head &= ~1UL; >>> -        foliop = (struct folio *)&precise; >>> -        idx = 0; >>> -    } >>> -dump: >>> -    __dump_folio(foliop, &precise, pfn, idx); >>> +    __dump_folio(&ps.folio_snapshot, &ps.page_snapshot, ps.pfn, ps.idx); >>>   } >>>   void dump_page(const struct page *page, const char *reason) >>> diff --git a/mm/util.c b/mm/util.c >>> index 0b270c43d7d1..c38d213be83f 100644 >>> --- a/mm/util.c >>> +++ b/mm/util.c >>> @@ -25,6 +25,7 @@ >>>   #include >>>   #include >>>   #include >>> +#include >>>   #include >>> @@ -1171,3 +1172,79 @@ int compat_vma_mmap_prepare(struct file *file, struct vm_area_struct *vma) >>>       return 0; >>>   } >>>   EXPORT_SYMBOL(compat_vma_mmap_prepare); >>> + >>> +static void set_flags(struct page_snapshot *ps, const struct folio *folio, >>> +              const struct page *page) >>> +{ >>> +    /* >>> +     * Caveats on high order pages: PG_buddy and PG_slab will only be set >>> +     * on the head page. >>> +     */ >>> +    if (PageBuddy(page)) >>> +        ps->flags |= PAGE_SNAPSHOT_PG_FREE; >>> +    else if (page_count(page) == 0 && is_free_buddy_page(page)) >>> +        ps->flags |= PAGE_SNAPSHOT_PG_FREE; >>> + >>> +    if (folio_test_idle(folio)) >>> +        ps->flags |= PAGE_SNAPSHOT_PG_IDLE; >>> +} >>> + >>> +/* >>> + * Create a snapshot of a page and store its struct page and struct folio >>> + * representations in a struct page_snapshot. >>> + * >>> + * @ps: struct page_snapshot to store the page snapshot >>> + * @page: the page we want to snapshot >>> + * >>> + * Note that creating a faithful snapshot of a page may fail if the page >>> + * compound keeps changing (eg. due to folio split). In this case we set >>> + * ps->faithful to false and the snapshot will assume that @page refers >>> + * to a single page. >>> + */ >>> +void snapshot_page(struct page_snapshot *ps, const struct page *page) >>> +{ >>> +    unsigned long head, nr_pages = 1; >>> +    struct folio *foliop, folio; >>> +    int loops = 5; >>> + >>> +    ps->pfn = page_to_pfn(page); >>> +    ps->flags = PAGE_SNAPSHOT_FAITHFUL; >>> + >>> +again: >>> +    memcpy(&ps->page_snapshot, page, sizeof(*page)); >>> +    head = ps->page_snapshot.compound_head; >>> +    if ((head & 1) == 0) { >>> +        foliop = (struct folio *)&ps->page_snapshot; >>> +        ps->idx = 0; >>> +        if (!folio_test_large(foliop)) { >>> +            set_flags(ps, page_folio(page), page); >>> +            goto out; >>> +        } >>> +        foliop = (struct folio *)page; >>> +    } else { >>> +        foliop = (struct folio *)(page->compound_head - 1); >>                       ^^^^ >> should we use ps->page_snapshot here? >> IIUC, page may change due to race. > > You're right the race exists, but we can't make foliop point to the > snapshot because we do pointer arithmetic and other operations that need > the real page memory address to work properly. While working on v3 I realized it should be fine (and correct) to use the head pointer from ps->page_snapshot like the original __dump_page() does (assuming this is what you meant for me to change). So, I changed this for v3. > >>> +        ps->idx = folio_page_idx(foliop, page); >>> +    } >>> + >>> +    if (ps->idx < MAX_FOLIO_NR_PAGES) { >>> +        memcpy(&folio, foliop, 2 * sizeof(struct page)); >>> +        nr_pages = folio_nr_pages(&folio); >>> +        if (nr_pages > 1) >>> +            memcpy(&folio.__page_2, &foliop->__page_2, >>> +                   sizeof(struct page)); >>> +        set_flags(ps, foliop, page); >>> +        foliop = &folio; >>> +    } >>> + >>> +    if (ps->idx > nr_pages) { >>> +        if (loops-- > 0) >>> +            goto again; >>> +        ps->page_snapshot.compound_head &= ~1UL; >> >> Should we use clear_compound_head() for clarity? > > Yes, we can do that. > > Andrew, since you applied this series already, can you squash the change > below into this patch? If not, I can send v3. > > diff --git a/mm/util.c b/mm/util.c > index c38d213be83f..ce4200529772 100644 > --- a/mm/util.c > +++ b/mm/util.c > @@ -1239,7 +1239,7 @@ void snapshot_page(struct page_snapshot *ps, const struct page *page) >      if (ps->idx > nr_pages) { >          if (loops-- > 0) >              goto again; > -        ps->page_snapshot.compound_head &= ~1UL; > +        clear_compound_head(&ps->page_snapshot); >          foliop = (struct folio *)&ps->page_snapshot; >          ps->flags = 0; >          ps->idx = 0;