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 5DAD6C3DA59 for ; Tue, 16 Jul 2024 19:02:01 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id BF5BC6B0085; Tue, 16 Jul 2024 15:02:00 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id B7F4B6B0088; Tue, 16 Jul 2024 15:02:00 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 9F88A6B0093; Tue, 16 Jul 2024 15:02:00 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0015.hostedemail.com [216.40.44.15]) by kanga.kvack.org (Postfix) with ESMTP id 7CB606B0085 for ; Tue, 16 Jul 2024 15:02:00 -0400 (EDT) Received: from smtpin13.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay04.hostedemail.com (Postfix) with ESMTP id 280661A0776 for ; Tue, 16 Jul 2024 19:02:00 +0000 (UTC) X-FDA: 82346535600.13.9582170 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by imf08.hostedemail.com (Postfix) with ESMTP id C7EEA160029 for ; Tue, 16 Jul 2024 19:01:57 +0000 (UTC) Authentication-Results: imf08.hostedemail.com; dkim=pass header.d=redhat.com header.s=mimecast20190719 header.b=JVVPDz6A; spf=pass (imf08.hostedemail.com: domain of peterx@redhat.com designates 170.10.133.124 as permitted sender) smtp.mailfrom=peterx@redhat.com; dmarc=pass (policy=none) header.from=redhat.com ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1721156498; 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: in-reply-to:in-reply-to:references:references:dkim-signature; bh=CcYHHlg5fx3Cb0ytsohRRyGcvg4CjT5/itwmV+qR6D0=; b=ItuiZjSmiEvbl8nd+vXrugM8dd6Rk1WJcElSHPdnrDnO61dl5IyCyPjzIkFRuBTt2j9JjW incsnaA5KzhdHZegj6V4glJJmojoZ6UABe7LVR5mDxeOMWbWOvVGLJjQcp6osu7j5U/aVC BCwy2eb1q38YDR0GV53mRDF3N71KFEo= ARC-Authentication-Results: i=1; imf08.hostedemail.com; dkim=pass header.d=redhat.com header.s=mimecast20190719 header.b=JVVPDz6A; spf=pass (imf08.hostedemail.com: domain of peterx@redhat.com designates 170.10.133.124 as permitted sender) smtp.mailfrom=peterx@redhat.com; dmarc=pass (policy=none) header.from=redhat.com ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1721156498; a=rsa-sha256; cv=none; b=mkpU2dJkxnkPQau28sfYBHwu7CRwpzxpO6i+h2AizZo+eJ/mgkfBy0hNeaDxR5yhvBnzLt wTLHDBpH5E8qRvdLg2L1EqzZA647im7pumOYqVsri9sKFPxrYOYVwCeG129LnbPAd24tBA HHjiH0oArZ3dHwWWGOQ1TqNIfKo8M/4= DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1721156517; 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: in-reply-to:in-reply-to:references:references; bh=CcYHHlg5fx3Cb0ytsohRRyGcvg4CjT5/itwmV+qR6D0=; b=JVVPDz6A1ar/kMgfO/vqOKWRSI97nk2axAq61sj1tlxkRPF5Ka9CDgT2rvFoWkKsb8ozl4 MPjnA7twgTRAiB6DqRCHYEBwhe90ZqlR2HfCp0bT9QEUyZjhLaFwMjj6Rw28Pipa/0ekEV yJBa/wFinJ/oawutXYTdbdiRAi791Rg= Received: from mail-qt1-f198.google.com (mail-qt1-f198.google.com [209.85.160.198]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-634-qShaQIXsOoCrRkBHGPKi1Q-1; Tue, 16 Jul 2024 15:01:55 -0400 X-MC-Unique: qShaQIXsOoCrRkBHGPKi1Q-1 Received: by mail-qt1-f198.google.com with SMTP id d75a77b69052e-447f84c361eso10174881cf.2 for ; Tue, 16 Jul 2024 12:01:55 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1721156515; x=1721761315; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:x-gm-message-state:from:to:cc:subject:date :message-id:reply-to; bh=CcYHHlg5fx3Cb0ytsohRRyGcvg4CjT5/itwmV+qR6D0=; b=fke//YjvUU2QItor3vQ+xLpQW6y/dRXHmfNn+dBqItXrngZhqyHf/F3V9IXaagnrsP 4JWP8GYPfTUJneEBgzWJysc0sDeeakauSyb45nBtP34zdE4TJ6ILEja7mTccuq5uOaFi z2qX/vM294RhNXLvUTmSdbbXd4FBB1/B5Klutoz6makuQdj3XdNlbL6E69m8D+AfbjSU gLNxi4D25H5voZRHx26q8xzyx9u5NCvmD8YuJCUnxeL7M8mdMsdO3a5z2xF7aphyaS3Z DcVkDKCfimhpwQgs0Myp3xbSPgzOuht8ypgSto456W199LYFKdrikAQznsOYahYcfSnc ArKg== X-Forwarded-Encrypted: i=1; AJvYcCXWO8jB/UHS6I2hdqkoGz48o6emR9HWyN/2YVYC/wPfgOsS7Izqv6KuADS2V3xNuSv8wuYpIP2IJymRTXaqm/BRCxU= X-Gm-Message-State: AOJu0Yx7po6BQ8iz6OHaqNlV6aZYWtWAyu/Q6Y+dblLrpxSjVsUWm8D0 NmI7yiBlLVDdel0d/JwwCxvK1gAK6RcxZednf0NgrtVYwherl9oHIqasrgivW+B2bsMUFNrZqN8 5saGrC3Blv8CKPh2VaOk/kkf0XLj+xFudyXuKavb6JTppFlst X-Received: by 2002:a05:620a:46a7:b0:79d:595b:d4ea with SMTP id af79cd13be357-7a179f15aa8mr307764385a.2.1721156514739; Tue, 16 Jul 2024 12:01:54 -0700 (PDT) X-Google-Smtp-Source: AGHT+IHOzvMR5tbj+oFC3kdKi/Z+6CGo9jm2FxkGsW2GFU+DmGvAarOMuuX7Ftf95fm/up+VVG8a3A== X-Received: by 2002:a05:620a:46a7:b0:79d:595b:d4ea with SMTP id af79cd13be357-7a179f15aa8mr307761785a.2.1721156514279; Tue, 16 Jul 2024 12:01:54 -0700 (PDT) Received: from x1n (pool-99-254-121-117.cpe.net.cable.rogers.com. [99.254.121.117]) by smtp.gmail.com with ESMTPSA id af79cd13be357-7a160b92a4asm320843085a.12.2024.07.16.12.01.52 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 16 Jul 2024 12:01:53 -0700 (PDT) Date: Tue, 16 Jul 2024 15:01:50 -0400 From: Peter Xu To: Yan Zhao Cc: "linux-kernel@vger.kernel.org" , "linux-mm@kvack.org" , Andrew Morton , Alex Williamson , Jason Gunthorpe , Al Viro , Dave Hansen , Andy Lutomirski , Peter Zijlstra , Thomas Gleixner , Ingo Molnar , Borislav Petkov , "Kirill A . Shutemov" , "x86@kernel.org" , "Tian, Kevin" , Pei Li , David Hildenbrand , David Wang <00107082@163.com>, Bert Karwatzki , Sergey Senozhatsky Subject: Re: [PATCH] mm/x86/pat: Only untrack the pfn range if unmap region Message-ID: References: <20240712144244.3090089-1-peterx@redhat.com> MIME-Version: 1.0 In-Reply-To: X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=utf-8 Content-Disposition: inline X-Rspamd-Server: rspam11 X-Rspamd-Queue-Id: C7EEA160029 X-Stat-Signature: xye8yahnnt3re97czmk4bb6i7cyp99ii X-Rspam-User: X-HE-Tag: 1721156517-243317 X-HE-Meta: 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 dwrPgjtW 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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 Tue, Jul 16, 2024 at 05:13:29PM +0800, Yan Zhao wrote: > On Mon, Jul 15, 2024 at 10:29:59PM +0800, Peter Xu wrote: > > On Mon, Jul 15, 2024 at 03:08:58PM +0800, Yan Zhao wrote: > > > On Fri, Jul 12, 2024 at 10:42:44AM -0400, Peter Xu wrote: > > > > This patch is one patch of an old series [1] that got reposted standalone > > > > here, with the hope to fix some reported untrack_pfn() issues reported > > > > recently [2,3], where there used to be other fix [4] but unfortunately > > > > which looks like to cause other issues. The hope is this patch can fix it > > > > the right way. > > > > > > > > X86 uses pfn tracking to do pfnmaps. AFAICT, the tracking should normally > > > > start at mmap() of device drivers, then untracked when munmap(). However > > > > in the current code the untrack is done in unmap_single_vma(). This might > > > > be problematic. > > > > > > > > For example, unmap_single_vma() can be used nowadays even for zapping a > > > > single page rather than the whole vmas. It's very confusing to do whole > > > > vma untracking in this function even if a caller would like to zap one > > > > page. It could simply be wrong. > > > > > > > > Such issue won't be exposed by things like MADV_DONTNEED won't ever work > > > > for pfnmaps and it'll fail the madvise() already before reaching here. > > > > However looks like it can be triggered like what was reported where invoked > > > > from an unmap request from a file vma. > > > > > > > > There's also work [5] on VFIO (merged now) to allow tearing down MMIO > > > > pgtables before an munmap(), in which case we may not want to untrack the > > > > pfns if we're only tearing down the pgtables. IOW, we may want to keep the > > > > pfn tracking information as those pfn mappings can be restored later with > > > > the same vma object. Currently it's not an immediate problem for VFIO, as > > > > VFIO uses UC- by default, but it looks like there's plan to extend that in > > > > the near future. > > > > > > > > IIUC, this was overlooked when zap_page_range_single() was introduced, > > > > while in the past it was only used in the munmap() path which wants to > > > > always unmap the region completely. E.g., commit f5cc4eef9987 ("VM: make > > > > zap_page_range() callers that act on a single VMA use separate helper") is > > > > the initial commit that introduced unmap_single_vma(), in which the chunk > > > > of untrack_pfn() was moved over from unmap_vmas(). > > > > > > > > Recover that behavior to untrack pfnmap only when unmap regions. > > > > > > > > [1] https://lore.kernel.org/r/20240523223745.395337-1-peterx@redhat.com > > > > [2] https://groups.google.com/g/syzkaller-bugs/c/FeQZvSbqWbQ/m/tHFmoZthAAAJ > > > > [3] https://lore.kernel.org/r/20240712131931.20207-1-00107082@163.com > > > > [4] https://lore.kernel.org/all/20240710-bug12-v1-1-0e5440f9b8d3@gmail.com/ > > > > [5] https://lore.kernel.org/r/20240523195629.218043-1-alex.williamson@redhat.com > > > > > > > > Cc: Alex Williamson > > > > Cc: Jason Gunthorpe > > > > Cc: Al Viro > > > > Cc: Dave Hansen > > > > Cc: Andy Lutomirski > > > > Cc: Peter Zijlstra > > > > Cc: Thomas Gleixner > > > > Cc: Ingo Molnar > > > > Cc: Borislav Petkov > > > > Cc: Kirill A. Shutemov > > > > Cc: x86@kernel.org > > > > Cc: Yan Zhao > > > > Cc: Kevin Tian > > > > Cc: Pei Li > > > > Cc: David Hildenbrand > > > > Cc: David Wang <00107082@163.com> > > > > Cc: Bert Karwatzki > > > > Cc: Sergey Senozhatsky > > > > Signed-off-by: Peter Xu > > > > --- > > > > > > > > NOTE: I massaged the commit message comparing to the rfc post [1], the > > > > patch itself is untouched. Also removed rfc tag, and added more people > > > > into the loop. Please kindly help test this patch if you have a reproducer, > > > > as I can't reproduce it myself even with the syzbot reproducer on top of > > > > mm-unstable. Instead of further check on the reproducer, I decided to send > > > > this out first as we have a bunch of reproducers on the list now.. > > > > --- > > > > mm/memory.c | 5 ++--- > > > > 1 file changed, 2 insertions(+), 3 deletions(-) > > > > > > > > diff --git a/mm/memory.c b/mm/memory.c > > > > index 4bcd79619574..f57cc304b318 100644 > > > > --- a/mm/memory.c > > > > +++ b/mm/memory.c > > > > @@ -1827,9 +1827,6 @@ static void unmap_single_vma(struct mmu_gather *tlb, > > > > if (vma->vm_file) > > > > uprobe_munmap(vma, start, end); > > > > > > > > - if (unlikely(vma->vm_flags & VM_PFNMAP)) > > > > - untrack_pfn(vma, 0, 0, mm_wr_locked); > > > > - > > > Specifically to VFIO's case, looks it doesn't matter if untrack_pfn() is > > > called here, since remap_pfn_range() is not called in mmap() and fault > > > handler, and therefore (vma->vm_flags & VM_PAT) is always 0. > > > > Right when with current repo, but I'm thinking maybe we should have VM_PAT > > there.. > Yes, I agree. > > But, currently for VFIO, it cannot call io_remap_pfn_range() in the fault > handler since vm_flags_set() requires mmap lock held for write while > the fault handler can only hold mmap lock for read. > So, it relies on ioremap()/iounmap() to reserve/de-reserve memtypes, > without VM_PAT being set in vma. Right, neither vm_flags_set() nor io_remap_pfn_range() should be called in a fault handler. They should only be called in mmap() phase. When you mentioned ioremap(), it's only for the VGA device, right? I don't see it's used in the vfio-pci's fault handler. > > > > > See what reserve_pfn_range() does right now: it'll make sure only one owner > > reserve this area, e.g. memtype_reserve() will fail if it has already been > > reserved. Then when succeeded as the first one to reserve the region, > > it'll make sure this mem type to also be synchronized to the kernel map > > (memtype_kernel_map_sync). > > > > So I feel like when MMIO disabled for a VFIO card, we may need to call > > reserve_pfn_range(), telling the kernel the mem type VFIO uses, even if the > > pgtable will be empty, and even if currently it's always UC- for now: > > > > vfio_pci_core_mmap(): > > vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); > > > > Then the memtype will be reserved even if it cannot be used. Otherwise I > > don't know whether there's side effects of kernel identity mapping where it > > mismatch later with what's mapped in the userspace via the vma, when MMIO > > got enabled again: pgtable started to be injected with a different memtype > > that specified only in the vma's pgprot. > Current VFIO relies on pci_iomap() to reserve MMIO pfns as UC-, thus > no VM_PAT in vmas. > > Calling remap_pfn_range() in the mmap() will cause problem to support > dynamic faulting. Looks there's still a window even with an immediate > unmap after mmap(). It can be conditionally injected. See Alex's commit (not yet posted anywhere, only used in our internal testing so far): https://github.com/xzpeter/linux/commit/f432e0e46c34e493943034be4cb9d6eb638f57d1 > > Also, calling remap_pfn_range() in mmap() may not meet the requirement of > drivers to dynamic switch backend resources, e.g. as what's in > cxl_mmap_fault(), since one remap_pfn_range() cannot reserve memtypes for > all backend resources at once. > > So, is there any way for the driver to reserve memtypes and set VM_PAT in > fault handler? I must confess I'm not familiar with memtype stuff, and just started looking recently. Per my reading so far, we have these multiple ways of doing memtype reservations, and no matter which API we use (ioremap / track_pfn_remap / pci_iomap), the same memtype needs to be used otherwise the reservation will fail. Here ioremap / pci_iomap will involve one more vmap so that the virtual mapping will present already after the API returns. Then here IMHO track_pfn_remap() is the one that we should still use in page-level mmap() controls, because so far we don't want to map any MMIO range yet, however we want to say "hey this VMA maps something special", by reserving these memtype and set VM_PAT. We want the virtual mapping only later during a fault(). In short, it looks to me the right thing we should do is to manually invoke track_pfn_remap() in vfio-pci's mmap() hook. if (!vdev->pm_runtime_engaged && __vfio_pci_memory_enabled(vdev)) ret = remap_pfn_range(vma, vma->vm_start, pfn, req_len, vma->vm_page_prot); else // TODO: manually invoke track_pfn_remap() here vm_flags_set(vma, VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP); Then the vma is registered, and untrack_pfn() should be automatically done when vma unmaps (and that relies on this patch to not do that too early). >From that POV, I still think this patch does the right thing and should be merged, even if we probably have one more issue to fix as David Wang reported.. I'll need to think about how to do that right, as I think that'll be needed as long as pfnmaps will support fault()s: it means when munmap() the pgtable may not present, and PAT cannot rely on walking the pgtable to know the base PFN anymore. Thanks, -- Peter Xu