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 X-Spam-Level: X-Spam-Status: No, score=-19.0 required=3.0 tests=BAYES_00,DKIMWL_WL_HIGH, DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,MENTIONS_GIT_HOSTING,SPF_HELO_NONE,SPF_PASS, 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 E6D72C433DB for ; Tue, 23 Mar 2021 00:49:22 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 5CCF36199F for ; Tue, 23 Mar 2021 00:49:22 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 5CCF36199F Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id 752AB6B00FF; Mon, 22 Mar 2021 20:49:21 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 6EB7F6B0101; Mon, 22 Mar 2021 20:49:21 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 4E0E96B0100; Mon, 22 Mar 2021 20:49:21 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0168.hostedemail.com [216.40.44.168]) by kanga.kvack.org (Postfix) with ESMTP id 2B1876B00FD for ; Mon, 22 Mar 2021 20:49:21 -0400 (EDT) Received: from smtpin21.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay02.hostedemail.com (Postfix) with ESMTP id DB7B06D99 for ; Tue, 23 Mar 2021 00:49:20 +0000 (UTC) X-FDA: 77949305280.21.70CB7F2 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [63.128.21.124]) by imf01.hostedemail.com (Postfix) with ESMTP id C71EB5001522 for ; Tue, 23 Mar 2021 00:49:19 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1616460559; 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; bh=hD2W80XSCvgjJyAJdeh/8rKOZd4SgNsGsbdJSKqKW+4=; b=VbJfe1TA2GCsWlw2vYyemLVN/HfPaJWwTYFK23M02D2XqYNeVM55EWCoF3JE2Tveu6g67G lP1zIxdrwkFelPIZSAIrbPis0eZWVK6UaXPf4kXWpEsquKrJIWdflkMdJg4sarOBZQaKhP vEEHxFZMyXiOj/418Hu6/xUjiZ4VF4I= Received: from mail-qk1-f199.google.com (mail-qk1-f199.google.com [209.85.222.199]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-496-1B-aw2QZN42ZQc7lUdU3jw-1; Mon, 22 Mar 2021 20:49:16 -0400 X-MC-Unique: 1B-aw2QZN42ZQc7lUdU3jw-1 Received: by mail-qk1-f199.google.com with SMTP id b127so789715qkf.19 for ; Mon, 22 Mar 2021 17:49:16 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=rvoI1qdKEfspGJCPdlDFMoYFs7ZhJOMq3uHtD+HwICI=; b=F2T1LsmWwtkTamV2ir6t6RZBCsoT45haGMjLwF3WuhkKvSuiNRtEvZd7iGMb+BtWcX 0Z77WC7m07V+fVX8rHYaq+TWjjOtt2wKETE8KcUYwnbjJaab3/fLDqURYZXVMNThXl8a gLYyS1rOt/8lZKgq7Q80LVWkKidLmKwFtg9fElz0zqmSiH+/uAppyEimJSX1apMPduZ6 MMdyRCf26cOoksQhhi06zOQgAmXuxea1jFLK1L82MMqAis0RzilvdfNUW4WMxAac6rSJ mZbI4H3S5I2bKzUJjXeEQmSN7Ks4M/MKqu5+ptW26f9BUdKf0JibwzKp20CdosFgPrWZ yF+w== X-Gm-Message-State: AOAM533rUQ2nhAhCBJjzf8/h0cMmZTdKgaAgJiv2pl0/1lcqUgEMU7Xy cIAE5zhdSmiAhuLVBmO1yTKbpS5nry+jZNA5eqaE3lWw9LC0aP114yEr9AOGdk0FbFwIYUvK1fv GwkvEdw2twM0= X-Received: by 2002:ac8:6f2e:: with SMTP id i14mr2339152qtv.277.1616460555972; Mon, 22 Mar 2021 17:49:15 -0700 (PDT) X-Google-Smtp-Source: ABdhPJxa8nFhBKigj+cy7jQp1pDPBmbwrf4XYyXLlKuRA437qcGajMfEx+HTBlRrPU7LH8eL0xJKww== X-Received: by 2002:ac8:6f2e:: with SMTP id i14mr2339127qtv.277.1616460555441; Mon, 22 Mar 2021 17:49:15 -0700 (PDT) Received: from localhost.localdomain (bras-base-toroon474qw-grc-82-174-91-135-175.dsl.bell.ca. [174.91.135.175]) by smtp.gmail.com with ESMTPSA id n6sm5031793qtx.22.2021.03.22.17.49.13 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 22 Mar 2021 17:49:14 -0700 (PDT) From: Peter Xu To: linux-kernel@vger.kernel.org, linux-mm@kvack.org Cc: "Kirill A . Shutemov" , Jerome Glisse , Mike Kravetz , Matthew Wilcox , Andrew Morton , Axel Rasmussen , Hugh Dickins , peterx@redhat.com, Nadav Amit , Andrea Arcangeli , Mike Rapoport Subject: [PATCH 00/23] userfaultfd-wp: Support shmem and hugetlbfs Date: Mon, 22 Mar 2021 20:48:49 -0400 Message-Id: <20210323004912.35132-1-peterx@redhat.com> X-Mailer: git-send-email 2.26.2 MIME-Version: 1.0 Authentication-Results: relay.mimecast.com; auth=pass smtp.auth=CUSA124A263 smtp.mailfrom=peterx@redhat.com X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable X-Stat-Signature: czcfw9phnjygomnmz7tmsgq7jnwxccdi X-Rspamd-Server: rspam02 X-Rspamd-Queue-Id: C71EB5001522 Received-SPF: none (redhat.com>: No applicable sender policy available) receiver=imf01; identity=mailfrom; envelope-from=""; helo=us-smtp-delivery-124.mimecast.com; client-ip=63.128.21.124 X-HE-DKIM-Result: pass/pass X-HE-Tag: 1616460559-326507 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: This patchset is based on tag v5.12-rc3-mmots-2021-03-17-22-26. To run the= =0D selftest, need to apply the two patches to fix minor mode page leak:=0D =0D https://lore.kernel.org/lkml/20210322175132.36659-1-peterx@redhat.com/=0D https://lore.kernel.org/lkml/20210322204836.1650221-1-axelrasmussen@google.= com/=0D =0D Since I didn't get any NACK in the previous RFC series for months, I decide= d to=0D remove the RFC tag starting from this version, so this is v1 of uffd-wp sup= port=0D on shmem & hugetlb.=0D =0D The whole series can also be found online [1].=0D =0D The major comment I'd like to get is on the new idea of swap special pte. = That=0D comes from suggestions from both Hugh and Andrea and I appreciated a lot fo= r=0D those discussions.=0D =0D In short, the so-called "swap special pte" in this patchset is a new type o= f=0D pte that doesn't exist in the past, but it got used initially in this serie= s in=0D file-backed memories. It is used to persist information even if the ptes g= ot=0D dropped meanwhile when the page cache still existed. For example, when=0D splitting a file-backed huge pmd, we could be simply dropping the pmd entry= =0D then wait until another fault coming. It's okay in the past since all=0D information in the pte can be retained from the page cache when the next pa= ge=0D fault triggers. However in this case, uffd-wp is per-pte information which= =0D cannot be kept in page cache, so that information needs to be maintained=0D somehow still in the pgtable entry, even if the pgtable entry is going to b= e=0D dropped. Here instead of replacing with a none entry, we used the "swap=0D special pte". Then when the next page fault triggers, we can observe orig_= pte=0D to retain this information.=0D =0D I'm copy-pasting some commit message from the patch "mm/swap: Introduce the= =0D idea of special swap ptes", where it tried to explain this pte in another a= ngle:=0D =0D We used to have special swap entries, like migration entries, hw-poison= =0D entries, device private entries, etc.=0D =0D Those "special swap entries" reside in the range that they need to be a= t least=0D swap entries first, and their types are decided by swp_type(entry).=0D =0D This patch introduces another idea called "special swap ptes".=0D =0D It's very easy to get confused against "special swap entries", but a sp= eical=0D swap pte should never contain a swap entry at all. It means, it's ille= gal to=0D call pte_to_swp_entry() upon a special swap pte.=0D =0D Make the uffd-wp special pte to be the first special swap pte.=0D =0D Before this patch, is_swap_pte()=3D=3Dtrue means one of the below:=0D =0D (a.1) The pte has a normal swap entry (non_swap_entry()=3D=3Dfalse).= For=0D example, when an anonymous page got swapped out.=0D =0D (a.2) The pte has a special swap entry (non_swap_entry()=3D=3Dtrue).= For=0D example, a migration entry, a hw-poison entry, etc.=0D =0D After this patch, is_swap_pte()=3D=3Dtrue means one of the below, where= case (b) is=0D added:=0D =0D (a) The pte contains a swap entry.=0D =0D (a.1) The pte has a normal swap entry (non_swap_entry()=3D=3Dfalse).= For=0D example, when an anonymous page got swapped out.=0D =0D (a.2) The pte has a special swap entry (non_swap_entry()=3D=3Dtrue).= For=0D example, a migration entry, a hw-poison entry, etc.=0D =0D (b) The pte does not contain a swap entry at all (so it cannot be pass= ed=0D into pte_to_swp_entry()). For example, uffd-wp special swap pte.= =0D =0D Hugetlbfs needs similar thing because it's also file-backed. I directly re= used=0D the same special pte there, though the shmem/hugetlb change on supporting t= his=0D new pte is different since they don't share code path a lot.=0D =0D Patch layout=0D =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=0D =0D Part (1): Shmem support, this is where the special swap pte is introduced.= =0D Some zap rework is needed within the process:=0D =0D shmem/userfaultfd: Take care of UFFDIO_COPY_MODE_WP=0D mm: Clear vmf->pte after pte_unmap_same() returns=0D mm/userfaultfd: Introduce special pte for unmapped file-backed mem=0D mm/swap: Introduce the idea of special swap ptes=0D shmem/userfaultfd: Handle uffd-wp special pte in page fault handler=0D mm: Drop first_index/last_index in zap_details=0D mm: Introduce zap_details.zap_flags=0D mm: Introduce ZAP_FLAG_SKIP_SWAP=0D mm: Pass zap_flags into unmap_mapping_pages()=0D shmem/userfaultfd: Persist uffd-wp bit across zapping for file-backed=0D shmem/userfaultfd: Allow wr-protect none pte for file-backed mem=0D shmem/userfaultfd: Allows file-back mem to be uffd wr-protected on thps= =0D shmem/userfaultfd: Handle the left-overed special swap ptes=0D shmem/userfaultfd: Pass over uffd-wp special swap pte when fork()=0D =0D Part (2): Hugetlb support, we need to disable huge pmd sharing for uffd-wp= =0D because not compatible just like uffd minor mode. The rest is the changes= =0D required to teach hugetlbfs understand the special swap pte too that introd= uced=0D with the uffd-wp change:=0D =0D hugetlb/userfaultfd: Hook page faults for uffd write protection=0D hugetlb/userfaultfd: Take care of UFFDIO_COPY_MODE_WP=0D hugetlb/userfaultfd: Handle UFFDIO_WRITEPROTECT=0D hugetlb: Pass vma into huge_pte_alloc()=0D hugetlb/userfaultfd: Forbid huge pmd sharing when uffd enabled=0D mm/hugetlb: Introduce huge version of special swap pte helpers=0D mm/hugetlb: Move flush_hugetlb_tlb_range() into hugetlb.h=0D hugetlb/userfaultfd: Unshare all pmds for hugetlbfs when register wp=0D hugetlb/userfaultfd: Handle uffd-wp special pte in hugetlb pf handler=0D hugetlb/userfaultfd: Allow wr-protect none ptes=0D hugetlb/userfaultfd: Only drop uffd-wp special pte if required=0D =0D Part (3): Enable both features in code and test=0D =0D userfaultfd: Enable write protection for shmem & hugetlbfs=0D userfaultfd/selftests: Enable uffd-wp for shmem/hugetlbfs=0D =0D Tests=0D =3D=3D=3D=3D=3D=3D=3D=3D=3D=0D =0D I've tested it using either userfaultfd kselftest program, but also with=0D umapsort [2] which should be even stricter. Tested page swapping in/out du= ring=0D umapsort.=0D =0D If anyone would like to try umapsort, need to use an extremely hacked versi= on=0D of umap library [3], because by default umap only supports anonymous. So t= o=0D test it we need to build [3] then [2].=0D =0D Any comment would be greatly welcomed. Thanks,=0D =0D [1] https://github.com/xzpeter/linux/tree/uffd-wp-shmem-hugetlbfs=0D [2] https://github.com/LLNL/umap-apps=0D [3] https://github.com/xzpeter/umap/tree/peter-shmem-hugetlbfs=0D =0D Peter Xu (23):=0D shmem/userfaultfd: Take care of UFFDIO_COPY_MODE_WP=0D mm: Clear vmf->pte after pte_unmap_same() returns=0D mm/userfaultfd: Introduce special pte for unmapped file-backed mem=0D mm/swap: Introduce the idea of special swap ptes=0D shmem/userfaultfd: Handle uffd-wp special pte in page fault handler=0D mm: Drop first_index/last_index in zap_details=0D mm: Introduce zap_details.zap_flags=0D mm: Introduce ZAP_FLAG_SKIP_SWAP=0D mm: Pass zap_flags into unmap_mapping_pages()=0D shmem/userfaultfd: Persist uffd-wp bit across zapping for file-backed=0D shmem/userfaultfd: Allow wr-protect none pte for file-backed mem=0D shmem/userfaultfd: Allows file-back mem to be uffd wr-protected on=0D thps=0D shmem/userfaultfd: Handle the left-overed special swap ptes=0D shmem/userfaultfd: Pass over uffd-wp special swap pte when fork()=0D hugetlb/userfaultfd: Hook page faults for uffd write protection=0D hugetlb/userfaultfd: Take care of UFFDIO_COPY_MODE_WP=0D hugetlb/userfaultfd: Handle UFFDIO_WRITEPROTECT=0D mm/hugetlb: Introduce huge version of special swap pte helpers=0D hugetlb/userfaultfd: Handle uffd-wp special pte in hugetlb pf handler=0D hugetlb/userfaultfd: Allow wr-protect none ptes=0D hugetlb/userfaultfd: Only drop uffd-wp special pte if required=0D mm/userfaultfd: Enable write protection for shmem & hugetlbfs=0D userfaultfd/selftests: Enable uffd-wp for shmem/hugetlbfs=0D =0D arch/arm64/kernel/mte.c | 2 +-=0D arch/x86/include/asm/pgtable.h | 28 +++=0D fs/dax.c | 10 +-=0D fs/hugetlbfs/inode.c | 15 +-=0D fs/proc/task_mmu.c | 14 +-=0D fs/userfaultfd.c | 38 ++--=0D include/asm-generic/hugetlb.h | 10 +=0D include/asm-generic/pgtable_uffd.h | 3 +=0D include/linux/hugetlb.h | 25 ++-=0D include/linux/mm.h | 50 ++++-=0D include/linux/mm_inline.h | 43 ++++=0D include/linux/shmem_fs.h | 5 +-=0D include/linux/swapops.h | 39 +++-=0D include/linux/userfaultfd_k.h | 46 +++++=0D include/uapi/linux/userfaultfd.h | 7 +-=0D mm/gup.c | 2 +-=0D mm/hmm.c | 2 +-=0D mm/hugetlb.c | 167 ++++++++++++---=0D mm/khugepaged.c | 14 +-=0D mm/madvise.c | 4 +-=0D mm/memcontrol.c | 2 +-=0D mm/memory.c | 249 ++++++++++++++++++-----=0D mm/migrate.c | 4 +-=0D mm/mincore.c | 2 +-=0D mm/mprotect.c | 63 +++++-=0D mm/mremap.c | 2 +-=0D mm/page_vma_mapped.c | 6 +-=0D mm/rmap.c | 8 +=0D mm/shmem.c | 31 ++-=0D mm/swapfile.c | 2 +-=0D mm/truncate.c | 17 +-=0D mm/userfaultfd.c | 37 ++--=0D tools/testing/selftests/vm/userfaultfd.c | 9 +-=0D 33 files changed, 776 insertions(+), 180 deletions(-)=0D =0D --=20=0D 2.26.2=0D =0D