Re: [RFC PATCH v1 15/28] riscv/mm: Implement map_shadow_stack() syscall

[Deepak Gupta]

On Thu, Jan 25, 2024 at 01:24:16PM -0800, Charlie Jenkins wrote:
> On Wed, Jan 24, 2024 at 10:21:40PM -0800, debug@xxxxxxxxxxxx wrote:
> > From: Deepak Gupta <debug@xxxxxxxxxxxx>
> >
> > As discussed extensively in the changelog for the addition of this
> > syscall on x86 ("x86/shstk: Introduce map_shadow_stack syscall") the
> > existing mmap() and madvise() syscalls do not map entirely well onto the
> > security requirements for guarded control stacks since they lead to
> > windows where memory is allocated but not yet protected or stacks which
> > are not properly and safely initialised. Instead a new syscall
> > map_shadow_stack() has been defined which allocates and initialises a
> > shadow stack page.
> >
> > This patch implements this syscall for riscv. riscv doesn't require token
> > to be setup by kernel because user mode can do that by itself. However to
> > provide compatiblity and portability with other architectues, user mode can
> > specify token set flag.
> >
> > Signed-off-by: Deepak Gupta <debug@xxxxxxxxxxxx>
> > ---
> >  arch/riscv/kernel/Makefile      |   2 +
> >  arch/riscv/kernel/usercfi.c     | 150 ++++++++++++++++++++++++++++++++
> >  include/uapi/asm-generic/mman.h |   1 +
> >  3 files changed, 153 insertions(+)
> >  create mode 100644 arch/riscv/kernel/usercfi.c
> >
> > diff --git a/arch/riscv/kernel/Makefile b/arch/riscv/kernel/Makefile
> > index fee22a3d1b53..8c668269e886 100644
> > --- a/arch/riscv/kernel/Makefile
> > +++ b/arch/riscv/kernel/Makefile
> > @@ -102,3 +102,5 @@ obj-$(CONFIG_COMPAT)		+= compat_vdso/
> >
> >  obj-$(CONFIG_64BIT)		+= pi/
> >  obj-$(CONFIG_ACPI)		+= acpi.o
> > +
> > +obj-$(CONFIG_RISCV_USER_CFI) += usercfi.o
> > diff --git a/arch/riscv/kernel/usercfi.c b/arch/riscv/kernel/usercfi.c
> > new file mode 100644
> > index 000000000000..35ede2cbc05b
> > --- /dev/null
> > +++ b/arch/riscv/kernel/usercfi.c
> > @@ -0,0 +1,150 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright (C) 2023 Rivos, Inc.
> Nit: Should be updated to 2024

noted

> > + * Deepak Gupta <debug@xxxxxxxxxxxx>
> > + */
> > +
> > +#include <linux/sched.h>
> > +#include <linux/bitops.h>
> > +#include <linux/types.h>
> > +#include <linux/mm.h>
> > +#include <linux/mman.h>
> > +#include <linux/uaccess.h>
> > +#include <linux/sizes.h>
> > +#include <linux/user.h>
> > +#include <linux/syscalls.h>
> > +#include <linux/prctl.h>
> > +#include <asm/csr.h>
> > +#include <asm/usercfi.h>
> > +
> > +#define SHSTK_ENTRY_SIZE sizeof(void *)
> > +
> > +/*
> > + * Writes on shadow stack can either be `sspush` or `ssamoswap`. `sspush` can happen
> > + * implicitly on current shadow stack pointed to by CSR_SSP. `ssamoswap` takes pointer to
> > + * shadow stack. To keep it simple, we plan to use `ssamoswap` to perform writes on shadow
> > + * stack.
> > + */
> > +static noinline unsigned long amo_user_shstk(unsigned long *addr, unsigned long val)
> > +{
> > +	/*
> > +	 * In case ssamoswap faults, return -1.
> > +	 * Never expect -1 on shadow stack. Expect return addresses and zero
> > +	 */
> > +	unsigned long swap = -1;
> > +
> > +	__enable_user_access();
> > +	asm_volatile_goto(
> > +				".option push\n"
> > +				".option arch, +zicfiss\n"
> > +#ifdef CONFIG_64BIT
> > +				"1: ssamoswap.d %0, %2, %1\n"
> > +#else
> > +				"1: ssamoswap.w %0, %2, %1\n"
>
> A SSAMOSWAP macro that conditionally defines this would be cleaner

Yes I need to do that. Infact I need to gate CONFIG_RISCV_USER_CFI behind
some riscv-gnu toolchain version as well. Becuase not all toolchain versions will
recognize this.

> > +#endif
> > +				_ASM_EXTABLE(1b, %l[fault])
> > +				RISCV_ACQUIRE_BARRIER
> > +				".option pop\n"
> > +				: "=r" (swap), "+A" (*addr)
>
> I just ran into this on one of my patches that not every compiler
> supports output args in asm goto blocks. You need to guard this with the
> kconfig option CC_HAS_ASM_GOTO_TIED_OUTPUT. Unfortunately, that means
> that this code needs two versions, or you can choose to gate CFI behind
> this option, it's supported by recent versions of GCC/CLANG.

Thanks.
I'll gate behind CC_HAS_ASM_GOTO_TIED_OUTPUT. Earlier versions of GCC/CLANG
won't have CFI support in them anyways.

> For readability it is also nice to use labels for the asm variables such
> as `"=r" (swap)` can be `[swap] "=r" (swap)` and then replace %0 with
> %[swap].

noted, will do that. I copied it from gcc asm snippet `amoswap` somewhere in
kernel. Goes without saying, I am terrible with gcc asm syntax.

> - Charlie
>
> > +				: "r" (val)
> > +				: "memory"
> > +				: fault