Received-SPF: pass (google.com: domain of linux-kernel+bounces-188572-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Precedence: bulk
MIME-Version: 1.0
References: <20240403234054.2020347-1-debug@rivosinc.com> <20240403234054.2020347-23-debug@rivosinc.com>
In-Reply-To: <20240403234054.2020347-23-debug@rivosinc.com>
From: Andy Chiu <andy.chiu@sifive.com>
Date: Fri, 24 May 2024 17:46:16 +0800
Message-ID: <CABgGipW4ZTFLh1dkiRuWD0WP4RRkfhyFCc+RsUjCD2EkA5GhSQ@mail.gmail.com>
Subject: Re: [PATCH v3 22/29] riscv sigcontext: adding cfi state field in sigcontext
To: Deepak Gupta <debug@rivosinc.com>
Cc: paul.walmsley@sifive.com, rick.p.edgecombe@intel.com, broonie@kernel.org, 
	Szabolcs.Nagy@arm.com, kito.cheng@sifive.com, keescook@chromium.org, 
	ajones@ventanamicro.com, conor.dooley@microchip.com, cleger@rivosinc.com, 
	atishp@atishpatra.org, alex@ghiti.fr, bjorn@rivosinc.com, 
	alexghiti@rivosinc.com, samuel.holland@sifive.com, conor@kernel.org, 
	linux-doc@vger.kernel.org, linux-riscv@lists.infradead.org, 
	linux-kernel@vger.kernel.org, devicetree@vger.kernel.org, linux-mm@kvack.org, 
	linux-arch@vger.kernel.org, linux-kselftest@vger.kernel.org, corbet@lwn.net, 
	palmer@dabbelt.com, aou@eecs.berkeley.edu, robh+dt@kernel.org, 
	krzysztof.kozlowski+dt@linaro.org, oleg@redhat.com, akpm@linux-foundation.org, 
	arnd@arndb.de, ebiederm@xmission.com, Liam.Howlett@oracle.com, vbabka@suse.cz, 
	lstoakes@gmail.com, shuah@kernel.org, brauner@kernel.org, 
	jerry.shih@sifive.com, hankuan.chen@sifive.com, greentime.hu@sifive.com, 
	evan@rivosinc.com, xiao.w.wang@intel.com, charlie@rivosinc.com, 
	apatel@ventanamicro.com, mchitale@ventanamicro.com, dbarboza@ventanamicro.com, 
	sameo@rivosinc.com, shikemeng@huaweicloud.com, willy@infradead.org, 
	vincent.chen@sifive.com, guoren@kernel.org, samitolvanen@google.com, 
	songshuaishuai@tinylab.org, gerg@kernel.org, heiko@sntech.de, bhe@redhat.com, 
	jeeheng.sia@starfivetech.com, cyy@cyyself.name, maskray@google.com, 
	ancientmodern4@gmail.com, mathis.salmen@matsal.de, cuiyunhui@bytedance.com, 
	bgray@linux.ibm.com, mpe@ellerman.id.au, baruch@tkos.co.il, alx@kernel.org, 
	david@redhat.com, catalin.marinas@arm.com, revest@chromium.org, 
	josh@joshtriplett.org, shr@devkernel.io, deller@gmx.de, omosnace@redhat.com, 
	ojeda@kernel.org, jhubbard@nvidia.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Deepak,

On Thu, Apr 4, 2024 at 7:42=E2=80=AFAM Deepak Gupta <debug@rivosinc.com> wr=
ote:
>
> Shadow stack needs to be saved and restored on signal delivery and signal
> return.
>
> sigcontext embedded in ucontext is extendible. Adding cfi state in there
> which can be used to save cfi state before signal delivery and restore
> cfi state on sigreturn
>
> Signed-off-by: Deepak Gupta <debug@rivosinc.com>
> ---
>  arch/riscv/include/uapi/asm/sigcontext.h | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/arch/riscv/include/uapi/asm/sigcontext.h b/arch/riscv/includ=
e/uapi/asm/sigcontext.h
> index cd4f175dc837..5ccdd94a0855 100644
> --- a/arch/riscv/include/uapi/asm/sigcontext.h
> +++ b/arch/riscv/include/uapi/asm/sigcontext.h
> @@ -21,6 +21,10 @@ struct __sc_riscv_v_state {
>         struct __riscv_v_ext_state v_state;
>  } __attribute__((aligned(16)));
>
> +struct __sc_riscv_cfi_state {
> +       unsigned long ss_ptr;   /* shadow stack pointer */
> +       unsigned long rsvd;             /* keeping another word reserved =
in case we need it */
> +};
>  /*
>   * Signal context structure
>   *
> @@ -29,6 +33,7 @@ struct __sc_riscv_v_state {
>   */
>  struct sigcontext {
>         struct user_regs_struct sc_regs;
> +       struct __sc_riscv_cfi_state sc_cfi_state;

I am concerned about this change as this could potentially break uabi.
Let's say there is a pre-CFI program running on this kernel. It
receives a signal so the kernel lays out the sig-stack as presented in
this structure. If the program accesses sc_fpregs, it would now get
sc_cfi_state. As the offset has changed, and the pre-CFI program has
not been re-compiled.

>         union {
>                 union __riscv_fp_state sc_fpregs;
>                 struct __riscv_extra_ext_header sc_extdesc;
> --
> 2.43.2
>

There may be two ways to deal with this. One is to use a different
signal ABI for CFI-enabled programs. This may complicate the user
space because new programs will have to determine whether it should
use the CFI-ABI at run time. Another way is to follow what Vector does
for signal stack. It adds a way to introduce new extensions on signal
stack without impacting ABI.

Please let me know if I misunderstand anything, thanks.

Cheers,
Andy