[v11,2/2] arm64: Enable BTI for main executable as well as the interpreter

Currently for dynamically linked ELF executables we only enable BTI for
the interpreter, expecting the interpreter to do this for the main
executable. This is a bit inconsistent since we do map main executable and
is causing issues with systemd's MemoryDenyWriteExecute feature which is
implemented using a seccomp filter which prevents setting PROT_EXEC on
already mapped memory and lacks the context to be able to detect that
memory is already mapped with PROT_EXEC.

Resolve this by adding a sysctl abi.bti_main which causes the kernel to
checking the BTI property for the main executable and enable BTI if it
is present when doing the initial mapping. This sysctl is disabled by
default.

Signed-off-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
---
 arch/arm64/include/asm/elf.h | 15 ++++++++---
 arch/arm64/kernel/process.c  | 52 +++++++++++++++++++++++++++---------
 2 files changed, 51 insertions(+), 16 deletions(-)

Hi,

On 3/8/22 07:22, Mark Brown wrote:
> Currently for dynamically linked ELF executables we only enable BTI for
> the interpreter, expecting the interpreter to do this for the main
> executable. This is a bit inconsistent since we do map main executable and
> is causing issues with systemd's MemoryDenyWriteExecute feature which is
> implemented using a seccomp filter which prevents setting PROT_EXEC on
> already mapped memory and lacks the context to be able to detect that
> memory is already mapped with PROT_EXEC.
> 
> Resolve this by adding a sysctl abi.bti_main which causes the kernel to
> checking the BTI property for the main executable and enable BTI if it
> is present when doing the initial mapping. This sysctl is disabled by
> default.

This seems less than ideal, maybe the default can be flipped with a 
CONFIG option?

> 
> Signed-off-by: Mark Brown <broonie@kernel.org>
> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
> ---
>   arch/arm64/include/asm/elf.h | 15 ++++++++---
>   arch/arm64/kernel/process.c  | 52 +++++++++++++++++++++++++++---------
>   2 files changed, 51 insertions(+), 16 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/elf.h b/arch/arm64/include/asm/elf.h
> index 5cc002376abe..c4aa60db76a4 100644
> --- a/arch/arm64/include/asm/elf.h
> +++ b/arch/arm64/include/asm/elf.h
> @@ -251,12 +251,21 @@ struct arch_elf_state {
>   	int flags;
>   };
>   
> -#define ARM64_ELF_BTI		(1 << 0)
> +#define ARM64_ELF_INTERP_BTI		(1 << 0)
> +#define ARM64_ELF_EXEC_BTI		(1 << 1)
>   
>   #define INIT_ARCH_ELF_STATE {			\
>   	.flags = 0,				\
>   }
>   
> +static inline int arm64_elf_bti_flag(bool is_interp)
> +{
> +	if (is_interp)
> +		return ARM64_ELF_INTERP_BTI;
> +	else
> +		return ARM64_ELF_EXEC_BTI;
> +}
> +
>   static inline int arch_parse_elf_property(u32 type, const void *data,
>   					  size_t datasz, bool compat,
>   					  bool has_interp, bool is_interp,
> @@ -272,9 +281,9 @@ static inline int arch_parse_elf_property(u32 type, const void *data,
>   		if (datasz != sizeof(*p))
>   			return -ENOEXEC;
>   
> -		if (system_supports_bti() && has_interp == is_interp &&
> +		if (system_supports_bti() &&
>   		    (*p & GNU_PROPERTY_AARCH64_FEATURE_1_BTI))
> -			arch->flags |= ARM64_ELF_BTI;
> +			arch->flags |= arm64_elf_bti_flag(is_interp);
>   	}
>   
>   	return 0;
> diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
> index 5369e649fa79..82aaf361fa17 100644
> --- a/arch/arm64/kernel/process.c
> +++ b/arch/arm64/kernel/process.c
> @@ -701,23 +701,49 @@ core_initcall(tagged_addr_init);
>   #endif	/* CONFIG_ARM64_TAGGED_ADDR_ABI */
>   
>   #ifdef CONFIG_BINFMT_ELF
> +static unsigned int bti_main;
> +
>   int arch_elf_adjust_prot(int prot, const struct arch_elf_state *state,
>   			 bool has_interp, bool is_interp)
>   {
> -	/*
> -	 * For dynamically linked executables the interpreter is
> -	 * responsible for setting PROT_BTI on everything except
> -	 * itself.
> -	 */
> -	if (is_interp != has_interp)
> -		return prot;
> -
> -	if (!(state->flags & ARM64_ELF_BTI))
> -		return prot;
> -
> -	if (prot & PROT_EXEC)
> +	if ((prot & PROT_EXEC) &&
> +	    (is_interp || !has_interp || bti_main) &&
> +	    (state->flags & arm64_elf_bti_flag(is_interp)))
>   		prot |= PROT_BTI;
>   
>   	return prot;
>   }
> -#endif
> +
> +#ifdef CONFIG_ARM64_BTI
> +/*
> + * If this sysctl is enabled then we will apply PROT_BTI to the main
> + * executable as well as the dynamic linker if it has the appropriate
> + * ELF note.  It is disabled by default, in which case we will only
> + * apply PROT_BTI to the dynamic linker or static binaries.
> + */
> +static struct ctl_table bti_main_sysctl_table[] = {
> +	{
> +		.procname	= "bti_main",
> +		.mode		= 0644,
> +		.data		= &bti_main,
> +		.maxlen		= sizeof(int),
> +		.proc_handler	= proc_dointvec_minmax,
> +		.extra1		= SYSCTL_ZERO,
> +		.extra2		= SYSCTL_ONE,
> +	},
> +	{ }
> +};
> +
> +static int __init bti_main_init(void)
> +{
> +	if (!system_supports_bti())
> +		return 0;
> +
> +	if (!register_sysctl("abi", bti_main_sysctl_table))
> +		return -EINVAL;
> +	return 0;
> +}
> +core_initcall(bti_main_init);
> +#endif /* CONFIG_ARM64_BTI */
> +
> +#endif /* CONFIG_BINFMT_ELF */

On Tue, Mar 08, 2022 at 12:01:17PM -0600, Jeremy Linton wrote:
> On 3/8/22 07:22, Mark Brown wrote:
> > Currently for dynamically linked ELF executables we only enable BTI for
> > the interpreter, expecting the interpreter to do this for the main
> > executable. This is a bit inconsistent since we do map main executable and
> > is causing issues with systemd's MemoryDenyWriteExecute feature which is
> > implemented using a seccomp filter which prevents setting PROT_EXEC on
> > already mapped memory and lacks the context to be able to detect that
> > memory is already mapped with PROT_EXEC.
> > 
> > Resolve this by adding a sysctl abi.bti_main which causes the kernel to
> > checking the BTI property for the main executable and enable BTI if it
> > is present when doing the initial mapping. This sysctl is disabled by
> > default.
> 
> This seems less than ideal, maybe the default can be flipped with a CONFIG
> option?

I'm not keen on config options changing the ABI. If there's a good
chance that this feature won't be turned on (via sysfs) in distros with
MDWE, I'd rather drop the whole series than maintain unused code in the
kernel.

(we can go back fixing this properly at the mprotect() level, as long as
systemd folk are willing to move away from eBPF for MDWE; happy to
provide kernel patches to start the discussion)

On Mon, Mar 14, 2022 at 08:57:49AM +0000, Catalin Marinas wrote:
> On Tue, Mar 08, 2022 at 12:01:17PM -0600, Jeremy Linton wrote:

> > > Resolve this by adding a sysctl abi.bti_main which causes the kernel to
> > > checking the BTI property for the main executable and enable BTI if it
> > > is present when doing the initial mapping. This sysctl is disabled by
> > > default.

> > This seems less than ideal, maybe the default can be flipped with a CONFIG
> > option?

> I'm not keen on config options changing the ABI. If there's a good
> chance that this feature won't be turned on (via sysfs) in distros with
> MDWE, I'd rather drop the whole series than maintain unused code in the
> kernel.

I think it's more just that it's a pain to have to also update userspace
for something fairly low risk than that it's an insurmountable obstacle.