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 3052EC54E65 for ; Thu, 22 May 2025 00:24:10 +0000 (UTC) Received: by kanga.kvack.org (Postfix) id 9629F6B0085; Wed, 21 May 2025 20:24:09 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 914266B0088; Wed, 21 May 2025 20:24:09 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 804FA6B0089; Wed, 21 May 2025 20:24:09 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from relay.hostedemail.com (smtprelay0011.hostedemail.com [216.40.44.11]) by kanga.kvack.org (Postfix) with ESMTP id 60E276B0085 for ; Wed, 21 May 2025 20:24:09 -0400 (EDT) Received: from smtpin23.hostedemail.com (a10.router.float.18 [10.200.18.1]) by unirelay03.hostedemail.com (Postfix) with ESMTP id CA67FBF527 for ; Thu, 22 May 2025 00:24:08 +0000 (UTC) X-FDA: 83468646576.23.05E06A9 Received: from out-186.mta1.migadu.com (out-186.mta1.migadu.com [95.215.58.186]) by imf17.hostedemail.com (Postfix) with ESMTP id CF05D40003 for ; Thu, 22 May 2025 00:24:06 +0000 (UTC) Authentication-Results: imf17.hostedemail.com; dkim=pass header.d=linux.dev header.s=key1 header.b=nsDeIZWy; spf=pass (imf17.hostedemail.com: domain of roman.gushchin@linux.dev designates 95.215.58.186 as permitted sender) smtp.mailfrom=roman.gushchin@linux.dev; dmarc=pass (policy=none) header.from=linux.dev ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=hostedemail.com; s=arc-20220608; t=1747873447; 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=wrH9jBCKBEOrByQtkD+ofBzbo0nje3TFiKainJjRQrc=; b=T2WF7ElSHp0G8m9hKr7PAetuyLp6VyvgCUVAIE3chxRCY3UfHDXvQlMIl2fyOqRkRmpF60 yerRtG7w9htb9KBasm68JsednrCdmp9sZ9qoyPCHXAE+f6VKdU8Iw+uwdSC8T6BizWABH5 gU9VM00gojYOyyAoGUE8CdJTxjhBNPM= ARC-Authentication-Results: i=1; imf17.hostedemail.com; dkim=pass header.d=linux.dev header.s=key1 header.b=nsDeIZWy; spf=pass (imf17.hostedemail.com: domain of roman.gushchin@linux.dev designates 95.215.58.186 as permitted sender) smtp.mailfrom=roman.gushchin@linux.dev; dmarc=pass (policy=none) header.from=linux.dev ARC-Seal: i=1; s=arc-20220608; d=hostedemail.com; t=1747873447; a=rsa-sha256; cv=none; b=q4RkNnaHVwv0OWl+d4wm47T6eYl9/vBlUblCefVoEce8/Npr3oScRSb6p41XRdJaXmBr9q StIzFb8W5uqH77WCnRYONfpJn/4149B+UzzvKqVIOchNjrArHaIaYd84XtHdK+9MXk2Tc2 LBXVlfp9o3YeGcLxho2kM2wuCGYPPtU= X-Report-Abuse: Please report any abuse attempt to abuse@migadu.com and include these headers. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.dev; s=key1; t=1747873444; 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=wrH9jBCKBEOrByQtkD+ofBzbo0nje3TFiKainJjRQrc=; b=nsDeIZWy0XQL+FstjkzXh/izAGEdF4IebwLXaH5pOGKGp2q24y3Q0lptRQ/59ZhA2jC6iF vUQqf64nQljBBhW0zsOXjhIF5haPqkB14hEUVM69YztfqBmNIrqvHq6r4geqcZ4lVQZtU2 OMyWkZaqLeeL/68drUJeGbGFQavzRnU= From: Roman Gushchin To: Jann Horn Cc: Hugh Dickins , Andrew Morton , Peter Zijlstra , linux-kernel@vger.kernel.org, linux-mm@kvack.org, Will Deacon , "Aneesh Kumar K.V" , Nick Piggin , linux-arch@vger.kernel.org Subject: Re: [PATCH v5] mmu_gather: move tlb flush for VM_PFNMAP/VM_MIXEDMAP vmas into free_pgtables() In-Reply-To: (Jann Horn's message of "Wed, 21 May 2025 02:38:20 +0200") References: <20250520214813.3946964-1-roman.gushchin@linux.dev> Date: Wed, 21 May 2025 17:23:57 -0700 Message-ID: <878qmpwl0i.fsf@linux.dev> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-Migadu-Flow: FLOW_OUT X-Rspamd-Server: rspam12 X-Rspamd-Queue-Id: CF05D40003 X-Stat-Signature: i1rqhodksx7xy6x618kk5og389dotyms X-Rspam-User: X-HE-Tag: 1747873446-823696 X-HE-Meta: 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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: Jann Horn writes: > On Tue, May 20, 2025 at 11:50=E2=80=AFPM Roman Gushchin > wrote: >> Commit b67fbebd4cf9 ("mmu_gather: Force tlb-flush VM_PFNMAP vmas") >> added a forced tlbflush to tlb_vma_end(), which is required to avoid a >> race between munmap() and unmap_mapping_range(). However it added some >> overhead to other paths where tlb_vma_end() is used, but vmas are not >> removed, e.g. madvise(MADV_DONTNEED). >> >> Fix this by moving the tlb flush out of tlb_end_vma() into new >> tlb_flush_vmas() called from free_pgtables(), somewhat similar to the >> stable version of the original commit: >> commit 895428ee124a ("mm: Force TLB flush for PFNMAP mappings before >> unlink_file_vma()"). >> >> Note, that if tlb->fullmm is set, no flush is required, as the whole >> mm is about to be destroyed. >> >> Signed-off-by: Roman Gushchin >> Cc: Jann Horn >> Cc: Peter Zijlstra >> Cc: Will Deacon >> Cc: "Aneesh Kumar K.V" >> Cc: Andrew Morton >> Cc: Nick Piggin >> Cc: Hugh Dickins >> Cc: linux-arch@vger.kernel.org >> Cc: linux-mm@kvack.org >> >> --- >> v5: >> - tlb_free_vma() -> tlb_free_vmas() to avoid extra checks >> >> v4: >> - naming/comments update (by Peter Z.) >> - check vma->vma->vm_flags in tlb_free_vma() (by Peter Z.) >> >> v3: >> - added initialization of vma_pfn in __tlb_reset_range() (by Hugh D.) >> >> v2: >> - moved vma_pfn flag handling into tlb.h (by Peter Z.) >> - added comments (by Peter Z.) >> - fixed the vma_pfn flag setting (by Hugh D.) >> --- >> include/asm-generic/tlb.h | 49 +++++++++++++++++++++++++++++++-------- >> mm/memory.c | 2 ++ >> 2 files changed, 41 insertions(+), 10 deletions(-) >> >> diff --git a/include/asm-generic/tlb.h b/include/asm-generic/tlb.h >> index 88a42973fa47..8a8b9535a930 100644 >> --- a/include/asm-generic/tlb.h >> +++ b/include/asm-generic/tlb.h >> @@ -58,6 +58,11 @@ >> * Defaults to flushing at tlb_end_vma() to reset the range; helps w= hen >> * there's large holes between the VMAs. >> * >> + * - tlb_free_vmas() >> + * >> + * tlb_free_vmas() marks the start of unlinking of one or more vmas >> + * and freeing page-tables. >> + * >> * - tlb_remove_table() >> * >> * tlb_remove_table() is the basic primitive to free page-table dire= ctories >> @@ -399,7 +404,10 @@ static inline void __tlb_reset_range(struct mmu_gat= her *tlb) >> * Do not reset mmu_gather::vma_* fields here, we do not >> * call into tlb_start_vma() again to set them if there is an >> * intermediate flush. >> + * >> + * Except for vma_pfn, that only cares if there's pending TLBI. >> */ >> + tlb->vma_pfn =3D 0; > > This looks dodgy to me. Can you explain here in more detail why this > is okay? Looking at current mainline, tlb->vma_pfn is only set to 1 > when tlb_start_vma() calls into tlb_update_vma_flags(); it is never > set again after tlb_start_vma(), so I don't think it's legal to just > clear it in the middle of a VMA. > > If we had something like this callgraph on a VM_MIXEDMAP mapping, with > an intermediate TLB flush in the middle of the VM_MIXEDMAP mapping: > > tlb_start_vma() > [sets tlb->vma_pfn] > zap_pte_range > tlb_flush_mmu_tlbonly > __tlb_reset_range > [clears tlb->vma_pfn] > zap_pte_range > [zaps more PTEs and queues a pending TLB flush] > tlb_end_vma() > free_pgtables > tlb_free_vmas > [checks for tlb->vma_pfn] > > then tlb_free_vmas() will erroneously not do a flush when it should've > done one, right? Good catch! > > Why does it even matter to you whether tlb->vma_pfn ever gets reset? I > think more or less at worst you do one extra TLB flush in some case > involving a munmap() across multiple VMAs including a mix of > VM_PFNMAP/VM_MIXEDMAP and non-VM_PFNMAP/VM_MIXEDMAP VMAs (which is > already a fairly weird scenario on its own), with the region being > operated on large enough or complicated enough that you already did at > least one TLB flush, and the unmap sufficiently large or with > sufficient address space around it that page tables are getting freed, > or something like that? That seems like a scenario in which one more > flush isn't going to be a big deal. Agree. > > If you wanted to do this properly, I think you could do something like: > > - add another flag tlb->current_vma_pfn that tracks whether the > current vma is pfnmap/mixedmap > - reset tlb->vma_pfn on TLB flush > - set tlb->vma_pfn again if a TLB flush is enqueued while > tlb->current_vma_pfn is true > > But that seems way too complicated, so I would just delete these three > lines from the patch. Agree, except tlb->vma_pfn needs to be initialized somewhere. This is primarily why these lines were added based on the feedback to the original version. However we missed the race you pointed at. I guess instead tlb->vma_pfn can be initialized in __tlb_gather_mmu(). I'll send out the fixed version shortly. Thank you!