Re: [PATCH RESEND v7 6/6] arm64: futex: support futex with FEAT_LSUI

[Will Deacon <will@kernel.org>]

On Sat, Aug 16, 2025 at 04:19:29PM +0100, Yeoreum Yun wrote:
> Current futex atomic operations are implemented with ll/sc instructions
> and clearing PSTATE.PAN.
> 
> Since Armv9.6, FEAT_LSUI supplies not only load/store instructions but
> also atomic operation for user memory access in kernel it doesn't need
> to clear PSTATE.PAN bit anymore.
> 
> With theses instructions some of futex atomic operations don't need to
> be implmented with ldxr/stlxr pair instead can be implmented with
> one atomic operation supplied by FEAT_LSUI.
> 
> However, some of futex atomic operations still need to use ll/sc way
> via ldtxr/stltxr supplied by FEAT_LSUI since there is no correspondant
> atomic instruction or doesn't support word size operation.
> (i.e) eor, cas{mb}t
> 
> But It's good to work without clearing PSTATE.PAN bit.
> 
> Signed-off-by: Yeoreum Yun <yeoreum.yun@arm.com>
> ---
>  arch/arm64/include/asm/futex.h | 130 ++++++++++++++++++++++++++++++++-
>  1 file changed, 129 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/include/asm/futex.h b/arch/arm64/include/asm/futex.h
> index 22a6301a9f3d..ece35ca9b5d9 100644
> --- a/arch/arm64/include/asm/futex.h
> +++ b/arch/arm64/include/asm/futex.h
> @@ -9,6 +9,8 @@
>  #include <linux/uaccess.h>
>  #include <linux/stringify.h>
> 
> +#include <asm/alternative.h>
> +#include <asm/alternative-macros.h>
>  #include <asm/errno.h>
> 
>  #define LLSC_MAX_LOOPS	128 /* What's the largest number you can think of? */
> @@ -115,11 +117,137 @@ __llsc_futex_cmpxchg(u32 __user *uaddr, u32 oldval, u32 newval, u32 *oval)
>  	return ret;
>  }
> 
> +#ifdef CONFIG_AS_HAS_LSUI
> +
> +#define __LSUI_PREAMBLE	".arch_extension lsui\n"
> +
> +#define LSUI_FUTEX_ATOMIC_OP(op, asm_op, mb)				\
> +static __always_inline int						\
> +__lsui_futex_atomic_##op(int oparg, u32 __user *uaddr, int *oval)	\
> +{									\
> +	int ret = 0;							\
> +	int oldval;							\
> +									\
> +	uaccess_ttbr0_enable();						\
> +	asm volatile("// __lsui_futex_atomic_" #op "\n"			\
> +	__LSUI_PREAMBLE							\
> +"1:	" #asm_op #mb "	%w3, %w2, %1\n"					\
> +"2:\n"									\
> +	_ASM_EXTABLE_UACCESS_ERR(1b, 2b, %w0)				\
> +	: "+r" (ret), "+Q" (*uaddr), "=r" (oldval)			\
> +	: "r" (oparg)							\
> +	: "memory");							\
> +	uaccess_ttbr0_disable();					\
> +									\
> +	if (!ret)							\
> +		*oval = oldval;						\
> +									\
> +	return ret;							\
> +}
> +
> +LSUI_FUTEX_ATOMIC_OP(add, ldtadd, al)
> +LSUI_FUTEX_ATOMIC_OP(or, ldtset, al)
> +LSUI_FUTEX_ATOMIC_OP(andnot, ldtclr, al)
> +LSUI_FUTEX_ATOMIC_OP(set, swpt, al)
> +
> +static __always_inline int
> +__lsui_futex_atomic_and(int oparg, u32 __user *uaddr, int *oval)
> +{
> +	return __lsui_futex_atomic_andnot(~oparg, uaddr, oval);
> +}
> +
> +static __always_inline int
> +__lsui_futex_atomic_eor(int oparg, u32 __user *uaddr, int *oval)
> +{
> +	unsigned int loops = LLSC_MAX_LOOPS;
> +	int ret, oldval, tmp;
> +
> +	uaccess_ttbr0_enable();
> +	/*
> +	 * there are no ldteor/stteor instructions...
> +	 */

*sigh*

Were these new instructions not added with futex in mind?

I wonder whether CAS would be better than exclusives for xor...

> +static __always_inline int
> +__lsui_futex_cmpxchg(u32 __user *uaddr, u32 oldval, u32 newval, u32 *oval)
> +{
> +	int ret = 0;
> +	unsigned int loops = LLSC_MAX_LOOPS;
> +	u32 val, tmp;
> +
> +	uaccess_ttbr0_enable();
> +	/*
> +	 * cas{al}t doesn't support word size...
> +	 */

What about just aligning down and doing a 64-bit cas in that case?

Will