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=-2.2 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=no 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 9C588C432C1 for ; Tue, 24 Sep 2019 16:35:52 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 3D47220872 for ; Tue, 24 Sep 2019 16:35:52 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 3D47220872 Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=arm.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id 9CE396B0006; Tue, 24 Sep 2019 12:35:51 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id 9A5166B0008; Tue, 24 Sep 2019 12:35:51 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 8BB4B6B000A; Tue, 24 Sep 2019 12:35:51 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0062.hostedemail.com [216.40.44.62]) by kanga.kvack.org (Postfix) with ESMTP id 6B5D96B0006 for ; Tue, 24 Sep 2019 12:35:51 -0400 (EDT) Received: from smtpin27.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay04.hostedemail.com (Postfix) with SMTP id 1DC59840E for ; Tue, 24 Sep 2019 16:35:51 +0000 (UTC) X-FDA: 75970365702.27.brush87_8a0a1efae8f52 X-HE-Tag: brush87_8a0a1efae8f52 X-Filterd-Recvd-Size: 7739 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by imf29.hostedemail.com (Postfix) with ESMTP for ; Tue, 24 Sep 2019 16:35:48 +0000 (UTC) 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 5589A142F; Tue, 24 Sep 2019 09:35:47 -0700 (PDT) Received: from arrakis.emea.arm.com (arrakis.cambridge.arm.com [10.1.196.78]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id A49143F694; Tue, 24 Sep 2019 09:35:44 -0700 (PDT) Date: Tue, 24 Sep 2019 17:35:42 +0100 From: Catalin Marinas To: Jia He Cc: "Justin He (Arm Technology China)" , Will Deacon , Mark Rutland , James Morse , Marc Zyngier , Matthew Wilcox , "Kirill A. Shutemov" , "linux-arm-kernel@lists.infradead.org" , "linux-kernel@vger.kernel.org" , "linux-mm@kvack.org" , Suzuki Poulose , Punit Agrawal , Anshuman Khandual , Alex Van Brunt , Robin Murphy , Thomas Gleixner , Andrew Morton , =?iso-8859-1?B?Suly9G1l?= Glisse , Ralph Campbell , "Kaly Xin (Arm Technology China)" , nd Subject: Re: [PATCH v8 3/3] mm: fix double page fault on arm64 if PTE_AF is cleared Message-ID: <20190924163542.GI41214@arrakis.emea.arm.com> References: <20190921135054.142360-1-justin.he@arm.com> <20190921135054.142360-4-justin.he@arm.com> <20190923170433.GE10192@arrakis.emea.arm.com> <20190924103324.GB41214@arrakis.emea.arm.com> <6267b685-5162-85ac-087f-112303bb7035@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <6267b685-5162-85ac-087f-112303bb7035@gmail.com> User-Agent: Mutt/1.10.1 (2018-07-13) Content-Transfer-Encoding: quoted-printable 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: On Tue, Sep 24, 2019 at 11:29:07PM +0800, Jia He wrote: > On 2019/9/24 18:33, Catalin Marinas wrote: > > On Tue, Sep 24, 2019 at 06:43:06AM +0000, Justin He (Arm Technology C= hina) wrote: > > > Catalin Marinas wrote: > > > > On Sat, Sep 21, 2019 at 09:50:54PM +0800, Jia He wrote: > > > > > /* > > > > > * This really shouldn't fail, because the page is there > > > > > * in the page tables. But it might just be unreadable, > > > > > * in which case we just give up and fill the result with > > > > > * zeroes. > > > > > */ > > > > > - if (__copy_from_user_inatomic(kaddr, uaddr, PAGE_SIZE)) > > > > > + if (__copy_from_user_inatomic(kaddr, uaddr, PAGE_SIZE)) { > > > > > + /* Give a warn in case there can be some obscure > > > > > + * use-case > > > > > + */ > > > > > + WARN_ON_ONCE(1); > > > > That's more of a question for the mm guys: at this point we do th= e > > > > copying with the ptl released; is there anything else that could = have > > > > made the pte old in the meantime? I think unuse_pte() is only cal= led on > > > > anonymous vmas, so it shouldn't be the case here. > > > > If we need to hold the ptl here, you could as well have an enclosing > > kmap/kunmap_atomic (option 2) with some goto instead of "return false= ". >=20 > I am not 100% sure that I understand your suggestion well, so I > drafted the patch Well, however you think the code is cleaner really. The copy/paste didn't work well, tabs disappeared (or rather the Exchange server corrupting outgoing emails) but I'll try to comment below: > -static inline void cow_user_page(struct page *dst, struct page *src, > unsigned long va, struct vm_area_struct *vma) > +static inline bool cow_user_page(struct page *dst, struct page *src, > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0struct vm_fault *vmf) > =A0{ > +=A0=A0=A0 struct vm_area_struct *vma =3D vmf->vma; > +=A0=A0=A0 struct mm_struct *mm =3D vma->vm_mm; > +=A0=A0=A0 unsigned long addr =3D vmf->address; > +=A0=A0=A0 bool ret; > +=A0=A0=A0 pte_t entry; > +=A0=A0=A0 void *kaddr; > +=A0=A0=A0 void __user *uaddr; > + > =A0=A0=A0=A0 debug_dma_assert_idle(src); >=20 > +=A0=A0=A0 if (likely(src)) { > +=A0=A0=A0 =A0=A0=A0 copy_user_highpage(dst, src, addr, vma); > +=A0=A0=A0 =A0=A0=A0 return true; > +=A0=A0=A0 } > + > =A0=A0=A0=A0 /* > =A0=A0=A0=A0 =A0* If the source page was a PFN mapping, we don't have > =A0=A0=A0=A0 =A0* a "struct page" for it. We do a best-effort copy by > =A0=A0=A0=A0 =A0* just copying from the original user address. If that > =A0=A0=A0=A0 =A0* fails, we just zero-fill it. Live with it. > =A0=A0=A0=A0 =A0*/ > -=A0=A0=A0 if (unlikely(!src)) { > -=A0=A0=A0 =A0=A0=A0 void *kaddr =3D kmap_atomic(dst); > -=A0=A0=A0 =A0=A0=A0 void __user *uaddr =3D (void __user *)(va & PAGE_M= ASK); > +=A0=A0=A0 kaddr =3D kmap_atomic(dst); > +=A0=A0=A0 uaddr =3D (void __user *)(addr & PAGE_MASK); > + > +=A0=A0=A0 /* > +=A0=A0=A0 =A0* On architectures with software "accessed" bits, we woul= d > +=A0=A0=A0 =A0* take a double page fault, so mark it accessed here. > +=A0=A0=A0 =A0*/ > +=A0=A0=A0 vmf->pte =3D pte_offset_map_lock(mm, vmf->pmd, addr, &vmf->p= tl); > +=A0=A0=A0 if (arch_faults_on_old_pte() && !pte_young(vmf->orig_pte)) { I'd move the pte_offset_map_lock() inside the 'if' block as we don't want to affect architectures that handle old ptes automatically. > +=A0=A0=A0 =A0=A0=A0 if (!likely(pte_same(*vmf->pte, vmf->orig_pte))) { > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 /* > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0* Other thread has already handled th= e fault > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0* and we don't need to do anything. I= f it's > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0* not the case, the fault will be tri= ggered > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0* again on the same address. > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 =A0*/ > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 ret =3D false; > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 goto pte_unlock; > +=A0=A0=A0 =A0=A0=A0 } > + > +=A0=A0=A0 =A0=A0=A0 entry =3D pte_mkyoung(vmf->orig_pte); > +=A0=A0=A0 =A0=A0=A0 if (ptep_set_access_flags(vma, addr, vmf->pte, ent= ry, 0)) > +=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 update_mmu_cache(vma, addr, vmf->pte); > +=A0=A0=A0 } >=20 > +=A0=A0=A0 /* > +=A0=A0=A0 =A0* This really shouldn't fail, because the page is there > +=A0=A0=A0 =A0* in the page tables. But it might just be unreadable, > +=A0=A0=A0 =A0* in which case we just give up and fill the result with > +=A0=A0=A0 =A0* zeroes. > +=A0=A0=A0 =A0*/ > +=A0=A0=A0 if (__copy_from_user_inatomic(kaddr, uaddr, PAGE_SIZE)) { > =A0=A0=A0=A0 =A0=A0=A0 /* > -=A0=A0=A0 =A0=A0=A0 =A0* This really shouldn't fail, because the page = is there > -=A0=A0=A0 =A0=A0=A0 =A0* in the page tables. But it might just be unre= adable, > -=A0=A0=A0 =A0=A0=A0 =A0* in which case we just give up and fill the re= sult with > -=A0=A0=A0 =A0=A0=A0 =A0* zeroes. > +=A0=A0=A0 =A0=A0=A0 =A0* Give a warn in case there can be some obscure > +=A0=A0=A0 =A0=A0=A0 =A0* use-case > =A0=A0=A0=A0 =A0=A0=A0 =A0*/ > -=A0=A0=A0 =A0=A0=A0 if (__copy_from_user_inatomic(kaddr, uaddr, PAGE_S= IZE)) > -=A0=A0=A0 =A0=A0=A0 =A0=A0=A0 clear_page(kaddr); > -=A0=A0=A0 =A0=A0=A0 kunmap_atomic(kaddr); > -=A0=A0=A0 =A0=A0=A0 flush_dcache_page(dst); > -=A0=A0=A0 } else > -=A0=A0=A0 =A0=A0=A0 copy_user_highpage(dst, src, va, vma); > +=A0=A0=A0 =A0=A0=A0 WARN_ON_ONCE(1); > +=A0=A0=A0 =A0=A0=A0 clear_page(kaddr); > +=A0=A0=A0 } > + > +=A0=A0=A0 ret =3D true; > + > +pte_unlock: > +=A0=A0=A0 pte_unmap_unlock(vmf->pte, vmf->ptl); Since the locking would be moved in the 'if' block above, we need another check here before unlocking: if (arch_faults_on_old_pte() && !pte_young(vmf->orig_pte)) pte_unmap_unlock(vmf->pte, vmf->ptl); You could probably replace the two calls to arch_faults_on_old_pte() with a single bool variable initialisation, something like: force_mkyoung =3D arch_faults_on_old_pte() && !pte_young(vmf->orig_pte) and only check for "if (force_mkyoung)" in both cases. > +=A0=A0=A0 kunmap_atomic(kaddr); > +=A0=A0=A0 flush_dcache_page(dst); > + > +=A0=A0=A0 return ret; > =A0} --=20 Catalin