Re: [RFC/RFT] CFI: Add support for gcc CFI in aarch64

From: Peter Zijlstra <peterz@infradead.org>
To: Dan Li <ashimida.1990@gmail.com>
Cc: Masahiro Yamada <masahiroy@kernel.org>,
	Michal Marek <michal.lkml@markovi.net>,
	Nick Desaulniers <ndesaulniers@google.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will@kernel.org>,
	Sami Tolvanen <samitolvanen@google.com>,
	Kees Cook <keescook@chromium.org>,
	Nathan Chancellor <nathan@kernel.org>, Tom Rix <trix@redhat.com>,
	"Paul E. McKenney" <paulmck@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	Josh Poimboeuf <jpoimboe@kernel.org>,
	Frederic Weisbecker <frederic@kernel.org>,
	"Eric W. Biederman" <ebiederm@xmission.com>,
	Marco Elver <elver@google.com>,
	Christophe Leroy <christophe.leroy@csgroup.eu>,
	Song Liu <song@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Uros Bizjak <ubizjak@gmail.com>,
	Kumar Kartikeya Dwivedi <memxor@gmail.com>,
	Juergen Gross <jgross@suse.com>,
	Luis Chamberlain <mcgrof@kernel.org>,
	Borislav Petkov <bp@suse.de>,
	Masami Hiramatsu <mhiramat@kernel.org>,
	Dmitry Torokhov <dmitry.torokhov@gmail.com>,
	Aaron Tomlin <atomlin@redhat.com>,
	Kalesh Singh <kaleshsingh@google.com>,
	Yuntao Wang <ytcoode@gmail.com>,
	Changbin Du <changbin.du@intel.com>,
	linux-kbuild@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, llvm@lists.linux.dev
Subject: Re: [RFC/RFT] CFI: Add support for gcc CFI in aarch64
Date: Mon, 19 Dec 2022 11:40:19 +0100	[thread overview]
Message-ID: <Y6A/k7/KrCCDuux6@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20221219061758.23321-1-ashimida.1990@gmail.com>

On Sun, Dec 18, 2022 at 10:17:58PM -0800, Dan Li wrote:

> In the compiler part[4], there are some differences from Sami's
> implementation[3], mainly including:
> 
> 1. When a typeid mismatch is detected, the cfi_check_failed function
>    will be called instead of the brk instruction. This function needs
>    to be implemented by the compiler user.
>    If there are user mode programs or other systems that want to use
>    this feature, it may be more convenient to use a callback (so this
>    compilation option is set to -fsanitize=cfi instead of kcfi).

This is not going to be acceptible for x86_64.

> 2. A reserved typeid (such as 0x0U on the aarch64 platform) is always
>    inserted in front of functions that should not be called indirectly.
>    Functions that can be called indirectly will not use this hash value,
>    which prevents instructions/data before the function from being used
>    as a typeid by an attacker.
> 
> 3. Some bits are ignored in the typeid to avoid conflicts between the
>    typeid and the instruction set of a specific platform, thereby
>    preventing an attacker from bypassing the CFI check by using the
>    instruction as a typeid, such as on the aarch64 platform:
>    * If the following instruction sequence exists:
> 	  400620:       a9be7bfd        stp     x29, x30, [sp, #-32]!
> 	  400624:       910003fd        mov     x29, sp
> 	  400628:       f9000bf3        str     x19, [sp, #16]
>    * If the expected typeid of the indirect call is exactly 0x910003fd,
>      the attacker can jump to the next instruction position of any
>      "mov x29,sp" instruction (such as 0x400628 here).
> 
> 4. Insert a symbol __cfi_<function> before each function's typeid,
>    which may be helpful for fine-grained KASLR implementations (or not?).
> 
> 5. The current implementation of gcc only supports the aarch64 platform.

What, if any, are the plans for x86_64 support?