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=-0.8 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_PASS,URIBL_BLOCKED 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 D4DA3C433F4 for ; Thu, 30 Aug 2018 16:06:21 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 8767B20658 for ; Thu, 30 Aug 2018 16:06:21 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 8767B20658 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727466AbeH3UJK (ORCPT ); Thu, 30 Aug 2018 16:09:10 -0400 Received: from mga05.intel.com ([192.55.52.43]:51126 "EHLO mga05.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726592AbeH3UJK (ORCPT ); Thu, 30 Aug 2018 16:09:10 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by fmsmga105.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 30 Aug 2018 09:06:19 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.53,307,1531810800"; d="scan'208";a="84719735" Received: from 2b52.sc.intel.com ([143.183.136.52]) by fmsmga004.fm.intel.com with ESMTP; 30 Aug 2018 09:06:18 -0700 Message-ID: <1535644924.26689.7.camel@intel.com> Subject: Re: [RFC PATCH v3 12/24] x86/mm: Modify ptep_set_wrprotect and pmdp_set_wrprotect for _PAGE_DIRTY_SW From: Yu-cheng Yu To: Jann Horn Cc: the arch/x86 maintainers , "H . Peter Anvin" , Thomas Gleixner , Ingo Molnar , kernel list , linux-doc@vger.kernel.org, Linux-MM , linux-arch , Linux API , Arnd Bergmann , Andy Lutomirski , Balbir Singh , Cyrill Gorcunov , Dave Hansen , Florian Weimer , hjl.tools@gmail.com, Jonathan Corbet , keescook@chromiun.org, Mike Kravetz , Nadav Amit , Oleg Nesterov , Pavel Machek , Peter Zijlstra , ravi.v.shankar@intel.com, vedvyas.shanbhogue@intel.com Date: Thu, 30 Aug 2018 09:02:04 -0700 In-Reply-To: References: <20180830143904.3168-1-yu-cheng.yu@intel.com> <20180830143904.3168-13-yu-cheng.yu@intel.com> Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.18.5.2-0ubuntu3.2 Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, 2018-08-30 at 17:49 +0200, Jann Horn wrote: > On Thu, Aug 30, 2018 at 4:43 PM Yu-cheng Yu > wrote: > > > > > > When Shadow Stack is enabled, the read-only and PAGE_DIRTY_HW PTE > > setting is reserved only for the Shadow Stack.  To track dirty of > > non-Shadow Stack read-only PTEs, we use PAGE_DIRTY_SW. > > > > Update ptep_set_wrprotect() and pmdp_set_wrprotect(). > > > > Signed-off-by: Yu-cheng Yu > > --- > >  arch/x86/include/asm/pgtable.h | 42 > > ++++++++++++++++++++++++++++++++++ > >  1 file changed, 42 insertions(+) > > > > diff --git a/arch/x86/include/asm/pgtable.h > > b/arch/x86/include/asm/pgtable.h > > index 4d50de77ea96..556ef258eeff 100644 > > --- a/arch/x86/include/asm/pgtable.h > > +++ b/arch/x86/include/asm/pgtable.h > > @@ -1203,7 +1203,28 @@ static inline pte_t > > ptep_get_and_clear_full(struct mm_struct *mm, > >  static inline void ptep_set_wrprotect(struct mm_struct *mm, > >                                       unsigned long addr, pte_t > > *ptep) > >  { > > +       pte_t pte; > > + > >         clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte); > > +       pte = *ptep; > > + > > +       /* > > +        * Some processors can start a write, but ending up seeing > > +        * a read-only PTE by the time they get to the Dirty bit. > > +        * In this case, they will set the Dirty bit, leaving a > > +        * read-only, Dirty PTE which looks like a Shadow Stack > > PTE. > > +        * > > +        * However, this behavior has been improved and will not > > occur > > +        * on processors supporting Shadow Stacks.  Without this > > +        * guarantee, a transition to a non-present PTE and flush > > the > > +        * TLB would be needed. > > +        * > > +        * When change a writable PTE to read-only and if the PTE > > has > > +        * _PAGE_DIRTY_HW set, we move that bit to _PAGE_DIRTY_SW > > so > > +        * that the PTE is not a valid Shadow Stack PTE. > > +        */ > > +       pte = pte_move_flags(pte, _PAGE_DIRTY_HW, _PAGE_DIRTY_SW); > > +       set_pte_at(mm, addr, ptep, pte); > >  } > I don't understand why it's okay that you first atomically clear the > RW bit, then atomically switch from DIRTY_HW to DIRTY_SW. Doesn't > that > mean that between the two atomic writes, another core can > incorrectly > see a shadow stack? Yes, we had that concern earlier and checked. On processors supporting Shadow Stacks, that will not happen. Yu-cheng From mboxrd@z Thu Jan 1 00:00:00 1970 From: Yu-cheng Yu Subject: Re: [RFC PATCH v3 12/24] x86/mm: Modify ptep_set_wrprotect and pmdp_set_wrprotect for _PAGE_DIRTY_SW Date: Thu, 30 Aug 2018 09:02:04 -0700 Message-ID: <1535644924.26689.7.camel@intel.com> References: <20180830143904.3168-1-yu-cheng.yu@intel.com> <20180830143904.3168-13-yu-cheng.yu@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Jann Horn Cc: the arch/x86 maintainers , "H . Peter Anvin" , Thomas Gleixner , Ingo Molnar , kernel list , linux-doc@vger.kernel.org, Linux-MM , linux-arch , Linux API , Arnd Bergmann , Andy Lutomirski , Balbir Singh , Cyrill Gorcunov , Dave Hansen , Florian Weimer , hjl.tools@gmail.com, Jonathan Corbet , keescook@chromiun.org, Mike Kravetz , Nadav Amit , Oleg Nesterov , Pavel Machek List-Id: linux-api@vger.kernel.org On Thu, 2018-08-30 at 17:49 +0200, Jann Horn wrote: > On Thu, Aug 30, 2018 at 4:43 PM Yu-cheng Yu > wrote: > > > > > > When Shadow Stack is enabled, the read-only and PAGE_DIRTY_HW PTE > > setting is reserved only for the Shadow Stack.  To track dirty of > > non-Shadow Stack read-only PTEs, we use PAGE_DIRTY_SW. > > > > Update ptep_set_wrprotect() and pmdp_set_wrprotect(). > > > > Signed-off-by: Yu-cheng Yu > > --- > >  arch/x86/include/asm/pgtable.h | 42 > > ++++++++++++++++++++++++++++++++++ > >  1 file changed, 42 insertions(+) > > > > diff --git a/arch/x86/include/asm/pgtable.h > > b/arch/x86/include/asm/pgtable.h > > index 4d50de77ea96..556ef258eeff 100644 > > --- a/arch/x86/include/asm/pgtable.h > > +++ b/arch/x86/include/asm/pgtable.h > > @@ -1203,7 +1203,28 @@ static inline pte_t > > ptep_get_and_clear_full(struct mm_struct *mm, > >  static inline void ptep_set_wrprotect(struct mm_struct *mm, > >                                       unsigned long addr, pte_t > > *ptep) > >  { > > +       pte_t pte; > > + > >         clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte); > > +       pte = *ptep; > > + > > +       /* > > +        * Some processors can start a write, but ending up seeing > > +        * a read-only PTE by the time they get to the Dirty bit. > > +        * In this case, they will set the Dirty bit, leaving a > > +        * read-only, Dirty PTE which looks like a Shadow Stack > > PTE. > > +        * > > +        * However, this behavior has been improved and will not > > occur > > +        * on processors supporting Shadow Stacks.  Without this > > +        * guarantee, a transition to a non-present PTE and flush > > the > > +        * TLB would be needed. > > +        * > > +        * When change a writable PTE to read-only and if the PTE > > has > > +        * _PAGE_DIRTY_HW set, we move that bit to _PAGE_DIRTY_SW > > so > > +        * that the PTE is not a valid Shadow Stack PTE. > > +        */ > > +       pte = pte_move_flags(pte, _PAGE_DIRTY_HW, _PAGE_DIRTY_SW); > > +       set_pte_at(mm, addr, ptep, pte); > >  } > I don't understand why it's okay that you first atomically clear the > RW bit, then atomically switch from DIRTY_HW to DIRTY_SW. Doesn't > that > mean that between the two atomic writes, another core can > incorrectly > see a shadow stack? Yes, we had that concern earlier and checked. On processors supporting Shadow Stacks, that will not happen. Yu-cheng From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pl1-f198.google.com (mail-pl1-f198.google.com [209.85.214.198]) by kanga.kvack.org (Postfix) with ESMTP id B90B76B5143 for ; Thu, 30 Aug 2018 12:06:20 -0400 (EDT) Received: by mail-pl1-f198.google.com with SMTP id 90-v6so4138779pla.18 for ; Thu, 30 Aug 2018 09:06:20 -0700 (PDT) Received: from mga09.intel.com (mga09.intel.com. [134.134.136.24]) by mx.google.com with ESMTPS id b2-v6si7264778plm.202.2018.08.30.09.06.19 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Thu, 30 Aug 2018 09:06:19 -0700 (PDT) Message-ID: <1535644924.26689.7.camel@intel.com> Subject: Re: [RFC PATCH v3 12/24] x86/mm: Modify ptep_set_wrprotect and pmdp_set_wrprotect for _PAGE_DIRTY_SW From: Yu-cheng Yu Date: Thu, 30 Aug 2018 09:02:04 -0700 In-Reply-To: References: <20180830143904.3168-1-yu-cheng.yu@intel.com> <20180830143904.3168-13-yu-cheng.yu@intel.com> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: owner-linux-mm@kvack.org List-ID: To: Jann Horn Cc: the arch/x86 maintainers , "H . Peter Anvin" , Thomas Gleixner , Ingo Molnar , kernel list , linux-doc@vger.kernel.org, Linux-MM , linux-arch , Linux API , Arnd Bergmann , Andy Lutomirski , Balbir Singh , Cyrill Gorcunov , Dave Hansen , Florian Weimer , hjl.tools@gmail.com, Jonathan Corbet , keescook@chromiun.org, Mike Kravetz , Nadav Amit , Oleg Nesterov , Pavel Machek , Peter Zijlstra , ravi.v.shankar@intel.com, vedvyas.shanbhogue@intel.com On Thu, 2018-08-30 at 17:49 +0200, Jann Horn wrote: > On Thu, Aug 30, 2018 at 4:43 PM Yu-cheng Yu > wrote: > > > > > > When Shadow Stack is enabled, the read-only and PAGE_DIRTY_HW PTE > > setting is reserved only for the Shadow Stack.A A To track dirty of > > non-Shadow Stack read-only PTEs, we use PAGE_DIRTY_SW. > > > > Update ptep_set_wrprotect() and pmdp_set_wrprotect(). > > > > Signed-off-by: Yu-cheng Yu > > --- > > A arch/x86/include/asm/pgtable.h | 42 > > ++++++++++++++++++++++++++++++++++ > > A 1 file changed, 42 insertions(+) > > > > diff --git a/arch/x86/include/asm/pgtable.h > > b/arch/x86/include/asm/pgtable.h > > index 4d50de77ea96..556ef258eeff 100644 > > --- a/arch/x86/include/asm/pgtable.h > > +++ b/arch/x86/include/asm/pgtable.h > > @@ -1203,7 +1203,28 @@ static inline pte_t > > ptep_get_and_clear_full(struct mm_struct *mm, > > A static inline void ptep_set_wrprotect(struct mm_struct *mm, > > A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A unsigned long addr, pte_t > > *ptep) > > A { > > +A A A A A A A pte_t pte; > > + > > A A A A A A A A clear_bit(_PAGE_BIT_RW, (unsigned long *)&ptep->pte); > > +A A A A A A A pte = *ptep; > > + > > +A A A A A A A /* > > +A A A A A A A A * Some processors can start a write, but ending up seeing > > +A A A A A A A A * a read-only PTE by the time they get to the Dirty bit. > > +A A A A A A A A * In this case, they will set the Dirty bit, leaving a > > +A A A A A A A A * read-only, Dirty PTE which looks like a Shadow Stack > > PTE. > > +A A A A A A A A * > > +A A A A A A A A * However, this behavior has been improved and will not > > occur > > +A A A A A A A A * on processors supporting Shadow Stacks.A A Without this > > +A A A A A A A A * guarantee, a transition to a non-present PTE and flush > > the > > +A A A A A A A A * TLB would be needed. > > +A A A A A A A A * > > +A A A A A A A A * When change a writable PTE to read-only and if the PTE > > has > > +A A A A A A A A * _PAGE_DIRTY_HW set, we move that bit to _PAGE_DIRTY_SW > > so > > +A A A A A A A A * that the PTE is not a valid Shadow Stack PTE. > > +A A A A A A A A */ > > +A A A A A A A pte = pte_move_flags(pte, _PAGE_DIRTY_HW, _PAGE_DIRTY_SW); > > +A A A A A A A set_pte_at(mm, addr, ptep, pte); > > A } > I don't understand why it's okay that you first atomically clear the > RW bit, then atomically switch from DIRTY_HW to DIRTY_SW. Doesn't > that > mean that between the two atomic writes, another core can > incorrectly > see a shadow stack? Yes, we had that concern earlier and checked. On processors supporting Shadow Stacks, that will not happen. Yu-cheng