From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753271Ab1AXPc3 (ORCPT ); Mon, 24 Jan 2011 10:32:29 -0500 Received: from rcsinet10.oracle.com ([148.87.113.121]:64456 "EHLO rcsinet10.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751855Ab1AXPc1 convert rfc822-to-8bit (ORCPT >); Mon, 24 Jan 2011 10:32:27 -0500 Date: Mon, 24 Jan 2011 10:31:09 -0500 From: Konrad Rzeszutek Wilk To: matthieu castet Cc: Ian Campbell , Kees Cook , Jeremy Fitzhardinge , "keir.fraser@eu.citrix.com" , "mingo@redhat.com" , "hpa@zytor.com" , "sliakh.lkml@gmail.com" , "jmorris@namei.org" , "linux-kernel@vger.kernel.org" , "rusty@rustcorp.com.au" , "torvalds@linux-foundation.org" , "ak@muc.de" , "davej@redhat.com" , "jiang@cs.ncsu.edu" , "arjan@infradead.org" , "tglx@linutronix.de" , "sfr@canb.auug.org.au" , "mingo@elte.hu" , Stefan Bader , konrad@kernel.org Subject: Re: [tip:x86/security] x86: Add NX protection for kernel data Message-ID: <20110124153109.GA970@dumpdata.com> References: <20110120150618.GC5092@dumpdata.com> <1295537856.14780.54.camel@zakaz.uk.xensource.com> <20110120190531.GA9687@dumpdata.com> <4D3899AB.60207@free.fr> <20110120210436.GA1810@dumpdata.com> <20110120211939.GA32262@dumpdata.com> <20110120215556.GA29700@dumpdata.com> <4D39FDA2.2000305@free.fr> <20110122051134.GA6905@dumpdata.com> <4D3C3ADB.1020609@free.fr> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <4D3C3ADB.1020609@free.fr> User-Agent: Mutt/1.5.20 (2009-06-14) Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sun, Jan 23, 2011 at 03:27:39PM +0100, matthieu castet wrote: > Konrad Rzeszutek Wilk a écrit : > >On Fri, Jan 21, 2011 at 10:41:54PM +0100, matthieu castet wrote: > >>Konrad Rzeszutek Wilk a écrit : > >>>>- * .data and .bss should always be writable. > >>>>+ * .data and .bss should always be writable, but xen won't like > >>>>+ * if we make page table rw (that live in .data or .bss) > >>>> */ > >>>>+#ifdef CONFIG_X86_32 > >>>> if (within(address, (unsigned long)_sdata, (unsigned long)_edata) || > >>>>- within(address, (unsigned long)__bss_start, (unsigned long)__bss_stop)) > >>>>- pgprot_val(required) |= _PAGE_RW; > >>>>+ within(address, (unsigned long)__bss_start, (unsigned long)__bss_stop)) { > >>>>+ unsigned int level; > >>>>+ if (lookup_address(address, &level) && (level != PG_LEVEL_4K)) > >>>>+ pgprot_val(forbidden) |= _PAGE_RW; > >>>>+ } > >>>>+#endif > >>>> #if defined(CONFIG_X86_64) && defined(CONFIG_DEBUG_RODATA) > >>>> > >>>>fyi, it does make it boot. > >>>Hold it.. ccache is a wonderful tool but I think I've just "rebuilt" the > >>>binaries with the .bss HPAGE_ALIGN aligment by mistake, so this path got never > >>>taken. > >>> > >>> > >>Ok, > >> > >>ATM I saw the following solution to solve the problem : > >>1) remove the data/bss check in static_protections, it was introduced by NX patches (64edc8ed). But I am not sure it > >>is really needed anymore. > >>2) add ". = ALIGN(HPAGE_SIZE)" somewhere after init section. But if we want not to be allocated in image we > >>should put it before bss. And if we want to be freed after init, we should put before .init.end. > >>This mean moving .smp_locks (and .data_nosave when x86 will be added) before init section. I have no idea of the impact. > >>3) add some logic in arch/x86/xen/mmu.c, that will ignore RW page setting for the page table marked RO. > >>4) make static_protections take and old_prot argument, and only apply RW .data/.bss requirement if page is already RW. > >> > >>If possible I will go for 1). > > > >Sounds good. Just send me the patch and I will test it. > > Ok, what give you the attached patch. > > I don't know if I should give the printk or not. I would say get rid of the printk. It does not really help the users. Here is an excerpt of 2.6.38-rc2 with this patch: 7.247448] NX-protecting the kernel data: 2412k [ 7.252489] RO page for 0xc15a0000 in bss/data. [ 7.253052] RO page for 0xc15a1000 in bss/data. [ 7.253052] RO page for 0xc15a3000 in bss/data. [ 7.365104] mv used greatest stack depth: 6616 bytes left So Tested-by: Konrad Rzeszutek Wilk (I tested on baremetal x86,x86_64 and Xen x86 and x86_64) > > > Matthieu > diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c > index 8b830ca..eec93c5 100644 > --- a/arch/x86/mm/pageattr.c > +++ b/arch/x86/mm/pageattr.c > @@ -256,7 +256,6 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address, > unsigned long pfn) > { > pgprot_t forbidden = __pgprot(0); > - pgprot_t required = __pgprot(0); > > /* > * The BIOS area between 640k and 1Mb needs to be executable for > @@ -283,11 +282,13 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address, > __pa((unsigned long)__end_rodata) >> PAGE_SHIFT)) > pgprot_val(forbidden) |= _PAGE_RW; > /* > - * .data and .bss should always be writable. > + * .data and .bss should always be writable, but xen won't like > + * if we make page table rw (that live in .data or .bss) > */ > if (within(address, (unsigned long)_sdata, (unsigned long)_edata) || > within(address, (unsigned long)__bss_start, (unsigned long)__bss_stop)) > - pgprot_val(required) |= _PAGE_RW; > + if ((pgprot_val(prot) & _PAGE_RW) == 0) > + printk(KERN_INFO "RO page for 0x%lx in bss/data.\n", address); > > #if defined(CONFIG_X86_64) && defined(CONFIG_DEBUG_RODATA) > /* > @@ -327,7 +328,6 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address, > #endif > > prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden)); > - prot = __pgprot(pgprot_val(prot) | pgprot_val(required)); > > return prot; > }