From mboxrd@z Thu Jan 1 00:00:00 1970 From: Balbir Singh Subject: Re: [PATCH v3 00/11] mm: Hardened usercopy Date: Mon, 18 Jul 2016 18:26:43 +1000 Message-ID: <1468830403.2800.0.camel@gmail.com> References: <1468619065-3222-1-git-send-email-keescook@chromium.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <1468619065-3222-1-git-send-email-keescook@chromium.org> Sender: linux-ia64-owner@vger.kernel.org To: Kees Cook , linux-kernel@vger.kernel.org Cc: Daniel Micay , Josh Poimboeuf , Rik van Riel , Casey Schaufler , PaX Team , Brad Spengler , Russell King , Catalin Marinas , Will Deacon , Ard Biesheuvel , Benjamin Herrenschmidt , Michael Ellerman , Tony Luck , Fenghua Yu , "David S. Miller" , x86@kernel.org, Christoph Lameter , Pekka Enberg , David Rientjes , Joonsoo Kim , Andrew Morton , Andy Lutomirski , Borislav Petkov , Mathias Krause List-Id: linux-arch.vger.kernel.org On Fri, 2016-07-15 at 14:44 -0700, Kees Cook wrote: > Hi, >=C2=A0 > [I'm going to carry this series in my kspp -next tree now, though I'd > really love to have some explicit Acked-bys or Reviewed-bys. If you'v= e > looked through it or tested it, please consider it. :) (I added Valdi= s > and mpe's Tested-bys where they seemed correct, thank you!)] >=C2=A0 > This is a start of the mainline port of PAX_USERCOPY[1]. After I star= ted > writing tests (now in lkdtm in -next) for Casey's earlier port[2], I = kept > tweaking things further and further until I ended up with a whole new > patch series. To that end, I took Rik and other people's feedback alo= ng > with other changes and clean-ups. >=C2=A0 > Based on my understanding, PAX_USERCOPY was designed to catch a > few classes of flaws (mainly bad bounds checking) around the use of > copy_to_user()/copy_from_user(). These changes don't touch get_user()= and > put_user(), since these operate on constant sized lengths, and tend t= o be > much less vulnerable. There are effectively three distinct protection= s in > the whole series, each of which I've given a separate CONFIG, though = this > patch set is only the first of the three intended protections. (Gener= ally > speaking, PAX_USERCOPY covers what I'm calling CONFIG_HARDENED_USERCO= PY > (this) and CONFIG_HARDENED_USERCOPY_WHITELIST (future), and > PAX_USERCOPY_SLABS covers CONFIG_HARDENED_USERCOPY_SPLIT_KMALLOC > (future).) >=C2=A0 > This series, which adds CONFIG_HARDENED_USERCOPY, checks that objects > being copied to/from userspace meet certain criteria: > - if address is a heap object, the size must not exceed the object's > =C2=A0 allocated size. (This will catch all kinds of heap overflow fl= aws.) > - if address range is in the current process stack, it must be within= the > =C2=A0 current stack frame (if such checking is possible) or at least= entirely > =C2=A0 within the current process's stack. (This could catch large le= ngths that > =C2=A0 would have extended beyond the current process stack, or overf= lows if > =C2=A0 their length extends back into the original stack.) > - if the address range is part of kernel data, rodata, or bss, allow = it. > - if address range is page-allocated, that it doesn't span multiple > =C2=A0 allocations. > - if address is within the kernel text, reject it. > - everything else is accepted >=C2=A0 > The patches in the series are: > - Support for arch-specific stack frame checking (which will likely b= e > =C2=A0 replaced in the future by Josh's more comprehensive unwinder): > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A01- mm: Implement stac= k frame object validation > - The core copy_to/from_user() checks, without the slab object checks= : > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A02- mm: Hardened userc= opy > - Per-arch enablement of the protection: > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A03- x86/uaccess: Enabl= e hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A04- ARM: uaccess: Enab= le hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A05- arm64/uaccess: Ena= ble hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A06- ia64/uaccess: Enab= le hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A07- powerpc/uaccess: E= nable hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A08- sparc/uaccess: Ena= ble hardened usercopy > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A09- s390/uaccess: Enab= le hardened usercopy > - The heap allocator implementation of object size checking: > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A010- mm: SLAB hardened userc= opy support > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A011- mm: SLUB hardened userc= opy support >=C2=A0 > Some notes: >=C2=A0 > - This is expected to apply on top of -next which contains fixes for = the > =C2=A0 position of _etext on both arm and arm64, though it has minor = conflicts > =C2=A0 with KASAN that are trivial to fix up. Living in -next are als= o tests > =C2=A0 for this protection in lkdtm, prefixed with USERCOPY_. >=C2=A0 > - I couldn't detect a measurable performance change with these featur= es > =C2=A0 enabled. Kernel build times were unchanged, hackbench was unch= anged, > =C2=A0 etc. I think we could flip this to "on by default" at some poi= nt, but > =C2=A0 for now, I'm leaving it off until I can get some more definiti= ve > =C2=A0 measurements. I would love if someone with greater familiarity= with > =C2=A0 perf could give this a spin and report results. >=C2=A0 > - The SLOB support extracted from grsecurity seems entirely broken. I > =C2=A0 have no idea what's going on there, I spent my time testing SL= AB and > =C2=A0 SLUB. Having someone else look at SLOB would be nice, but this= series > =C2=A0 doesn't depend on it. >=C2=A0 > Additional features that would be nice, but aren't blocking this seri= es: >=C2=A0 > - Needs more architecture support for stack frame checking (only x86 = now, > =C2=A0 but it seems Josh will have a good solution for this soon). >=C2=A0 >=C2=A0 > Thanks! >=C2=A0 > -Kees >=C2=A0 > [1] https://grsecurity.net/download.php "grsecurity - test kernel pat= ch" > [2] http://www.openwall.com/lists/kernel-hardening/2016/05/19/5 >=C2=A0 > v3: > - switch to using BUG for better Oops integration > - when checking page allocations, check each for Reserved > - use enums for the stack check return for readability > Thanks looks good so far! I'll try and test it and report back Balbir=C2=A0 -- To unsubscribe from this list: send the line "unsubscribe linux-ia64" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pa0-f68.google.com ([209.85.220.68]:36062 "EHLO mail-pa0-f68.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751426AbcGRI0X (ORCPT ); Mon, 18 Jul 2016 04:26:23 -0400 Message-ID: <1468830403.2800.0.camel@gmail.com> Subject: Re: [PATCH v3 00/11] mm: Hardened usercopy From: Balbir Singh Date: Mon, 18 Jul 2016 18:26:43 +1000 In-Reply-To: <1468619065-3222-1-git-send-email-keescook@chromium.org> References: <1468619065-3222-1-git-send-email-keescook@chromium.org> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-arch-owner@vger.kernel.org List-ID: To: Kees Cook , linux-kernel@vger.kernel.org Cc: Daniel Micay , Josh Poimboeuf , Rik van Riel , Casey Schaufler , PaX Team , Brad Spengler , Russell King , Catalin Marinas , Will Deacon , Ard Biesheuvel , Benjamin Herrenschmidt , Michael Ellerman , Tony Luck , Fenghua Yu , "David S. Miller" , x86@kernel.org, Christoph Lameter , Pekka Enberg , David Rientjes , Joonsoo Kim , Andrew Morton , Andy Lutomirski , Borislav Petkov , Mathias Krause , Jan Kara , Vitaly Wool , Andrea Arcangeli , Dmitry Vyukov , Laura Abbott , linux-arm-kernel@lists.infradead.org, linux-ia64@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, sparclinux@vger.kernel.org, linux-arch@vger.kernel.org, linux-mm@kvack.org, kernel-hardening@lists.openwall.com Message-ID: <20160718082643.23j_WGvVUjXekmREZZTRBtyQDzhjogJUZWE1GaDM-fM@z> On Fri, 2016-07-15 at 14:44 -0700, Kees Cook wrote: > Hi, >  > [I'm going to carry this series in my kspp -next tree now, though I'd > really love to have some explicit Acked-bys or Reviewed-bys. If you've > looked through it or tested it, please consider it. :) (I added Valdis > and mpe's Tested-bys where they seemed correct, thank you!)] >  > This is a start of the mainline port of PAX_USERCOPY[1]. After I started > writing tests (now in lkdtm in -next) for Casey's earlier port[2], I kept > tweaking things further and further until I ended up with a whole new > patch series. To that end, I took Rik and other people's feedback along > with other changes and clean-ups. >  > Based on my understanding, PAX_USERCOPY was designed to catch a > few classes of flaws (mainly bad bounds checking) around the use of > copy_to_user()/copy_from_user(). These changes don't touch get_user() and > put_user(), since these operate on constant sized lengths, and tend to be > much less vulnerable. There are effectively three distinct protections in > the whole series, each of which I've given a separate CONFIG, though this > patch set is only the first of the three intended protections. (Generally > speaking, PAX_USERCOPY covers what I'm calling CONFIG_HARDENED_USERCOPY > (this) and CONFIG_HARDENED_USERCOPY_WHITELIST (future), and > PAX_USERCOPY_SLABS covers CONFIG_HARDENED_USERCOPY_SPLIT_KMALLOC > (future).) >  > This series, which adds CONFIG_HARDENED_USERCOPY, checks that objects > being copied to/from userspace meet certain criteria: > - if address is a heap object, the size must not exceed the object's >   allocated size. (This will catch all kinds of heap overflow flaws.) > - if address range is in the current process stack, it must be within the >   current stack frame (if such checking is possible) or at least entirely >   within the current process's stack. (This could catch large lengths that >   would have extended beyond the current process stack, or overflows if >   their length extends back into the original stack.) > - if the address range is part of kernel data, rodata, or bss, allow it. > - if address range is page-allocated, that it doesn't span multiple >   allocations. > - if address is within the kernel text, reject it. > - everything else is accepted >  > The patches in the series are: > - Support for arch-specific stack frame checking (which will likely be >   replaced in the future by Josh's more comprehensive unwinder): >         1- mm: Implement stack frame object validation > - The core copy_to/from_user() checks, without the slab object checks: >         2- mm: Hardened usercopy > - Per-arch enablement of the protection: >         3- x86/uaccess: Enable hardened usercopy >         4- ARM: uaccess: Enable hardened usercopy >         5- arm64/uaccess: Enable hardened usercopy >         6- ia64/uaccess: Enable hardened usercopy >         7- powerpc/uaccess: Enable hardened usercopy >         8- sparc/uaccess: Enable hardened usercopy >         9- s390/uaccess: Enable hardened usercopy > - The heap allocator implementation of object size checking: >        10- mm: SLAB hardened usercopy support >        11- mm: SLUB hardened usercopy support >  > Some notes: >  > - This is expected to apply on top of -next which contains fixes for the >   position of _etext on both arm and arm64, though it has minor conflicts >   with KASAN that are trivial to fix up. Living in -next are also tests >   for this protection in lkdtm, prefixed with USERCOPY_. >  > - I couldn't detect a measurable performance change with these features >   enabled. Kernel build times were unchanged, hackbench was unchanged, >   etc. I think we could flip this to "on by default" at some point, but >   for now, I'm leaving it off until I can get some more definitive >   measurements. I would love if someone with greater familiarity with >   perf could give this a spin and report results. >  > - The SLOB support extracted from grsecurity seems entirely broken. I >   have no idea what's going on there, I spent my time testing SLAB and >   SLUB. Having someone else look at SLOB would be nice, but this series >   doesn't depend on it. >  > Additional features that would be nice, but aren't blocking this series: >  > - Needs more architecture support for stack frame checking (only x86 now, >   but it seems Josh will have a good solution for this soon). >  >  > Thanks! >  > -Kees >  > [1] https://grsecurity.net/download.php "grsecurity - test kernel patch" > [2] http://www.openwall.com/lists/kernel-hardening/2016/05/19/5 >  > v3: > - switch to using BUG for better Oops integration > - when checking page allocations, check each for Reserved > - use enums for the stack check return for readability > Thanks looks good so far! I'll try and test it and report back Balbir