Re: [kernel-hardening] Re: [PATCH] arm64: usercopy: Implement stack frame object validation

From: Will Deacon <will.deacon@arm.com>
To: Kees Cook <keescook@chromium.org>
Cc: Yann Droneaud <ydroneaud@opteya.com>,
	Keun-O Park <kpark3469@gmail.com>,
	"kernel-hardening@lists.openwall.com"
	<kernel-hardening@lists.openwall.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Mark Rutland <mark.rutland@arm.com>,
	James Morse <james.morse@arm.com>,
	Pratyush Anand <panand@redhat.com>,
	keun-o.park@darkmatter.ae,
	AKASHI Takahiro <takahiro.akashi@linaro.org>
Subject: Re: [kernel-hardening] Re: [PATCH] arm64: usercopy: Implement stack frame object validation
Date: Thu, 26 Jan 2017 17:47:02 +0000	[thread overview]
Message-ID: <20170126174702.GN14167@arm.com> (raw)
In-Reply-To: <CAGXu5jKgEketcC0jcxejw+Bmg0qUNUR7xtB8=CAstKpO-bdwfA@mail.gmail.com>

On Thu, Jan 26, 2017 at 09:36:44AM -0800, Kees Cook wrote:
> On Thu, Jan 26, 2017 at 8:40 AM, Yann Droneaud <ydroneaud@opteya.com> wrote:
> > Le mercredi 25 janvier 2017 à 13:54 +0000, Will Deacon a écrit :
> >> diff --git a/arch/arm64/include/asm/thread_info.h
> >> > b/arch/arm64/include/asm/thread_info.h
> >> > index 46c3b93..f610c44 100644
> >> > --- a/arch/arm64/include/asm/thread_info.h
> >> > +++ b/arch/arm64/include/asm/thread_info.h
> >> > @@ -68,7 +68,62 @@ struct thread_info {
> >> > +   const void *oldframe;
> >> > +   const void *callee_fp = NULL;
> >> > +   const void *caller_fp = NULL;
> >> > +
> >> > +   oldframe = __builtin_frame_address(1);
> >> > +   if (oldframe) {
> >> > +           callee_fp = __builtin_frame_address(2);
> >> > +           if (callee_fp)
> >> > +                   caller_fp = __builtin_frame_address(3);
> >> > +   }
> >> >
> >> Which compilers have you tested this with? The GCC folks don't
> >> guarantee a frame layout, and they have changed it in the past, so I
> >> suspect this is pretty fragile. In particularly, if
> >> __builtin_frame_address just points at the frame record, then I don't
> >> think you can make assumptions about the placement of local variables
> >> and arguments with respect to that.
> >>
> >
> > https://gcc.gnu.org/onlinedocs/gcc-6.3.0/gcc/Return-Address.html#index-
> > g_t_005f_005fbuiltin_005fframe_005faddress-3701
> >
> > "Calling this function with a nonzero argument can have unpredictable
> >  effects, including crashing the calling program. As a result, calls
> >  that are considered unsafe are diagnosed when the -Wframe-address
> >  option is in effect. Such calls should only be made in debugging
> >  situations."
> 
> It does work, though, and given the CONFIG_FRAME_POINTER check, I
> think this is fine. The kernel explicitly disables -Wframe-address
> since it gets used in a number of places.

I would prefer to use the existing unwind_frame, as suggested by James,
if possible. I really don't like relying on unpredictable compiler behaviour
if we don't have to!

Will