Re: [PATCH] riscv: Implement smp_cond_load8/16() with Zawrs

[Andrew Jones <ajones@ventanamicro.com>]

On Sun, Dec 15, 2024 at 10:22:53PM -0500, guoren@kernel.org wrote:
> From: Guo Ren <guoren@linux.alibaba.com>
> 
> RISC-V code uses the queued spinlock implementation, which calls
> the macros smp_cond_load_acquire for one byte. So, complement the
> implementation of byte and halfword versions.
> 
> Signed-off-by: Guo Ren <guoren@linux.alibaba.com>
> Signed-off-by: Guo Ren <guoren@kernel.org>
> ---
>  arch/riscv/include/asm/cmpxchg.h | 38 +++++++++++++++++++++++++++++---
>  1 file changed, 35 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/riscv/include/asm/cmpxchg.h b/arch/riscv/include/asm/cmpxchg.h
> index 4cadc56220fe..2bd42a11ff8f 100644
> --- a/arch/riscv/include/asm/cmpxchg.h
> +++ b/arch/riscv/include/asm/cmpxchg.h
> @@ -365,16 +365,48 @@ static __always_inline void __cmpwait(volatile void *ptr,
>  {
>  	unsigned long tmp;
>  
> +	u32 *__ptr32b;
> +	ulong __s, __val, __mask;
> +
>  	asm goto(ALTERNATIVE("j %l[no_zawrs]", "nop",
>  			     0, RISCV_ISA_EXT_ZAWRS, 1)
>  		 : : : : no_zawrs);
>  
>  	switch (size) {
>  	case 1:
> -		fallthrough;
> +		__ptr32b = (u32 *)((ulong)(ptr) & ~0x3);
> +		__s = ((ulong)(ptr) & 0x3) * BITS_PER_BYTE;
> +		__val = val << __s;
> +		__mask = 0xf << __s;

This mask should be 0xff and the mask below should be 0xffff.

> +
> +		asm volatile(
> +		"	lr.w	%0, %1\n"
> +		"	and	%0, %0, %3\n"
> +		"	xor	%0, %0, %2\n"
> +		"	bnez	%0, 1f\n"
> +			ZAWRS_WRS_NTO "\n"
> +		"1:"
> +		: "=&r" (tmp), "+A" (*(__ptr32b))
> +		: "r" (__val), "r" (__mask)
> +		: "memory");
> +		break;
>  	case 2:
> -		/* RISC-V doesn't have lr instructions on byte and half-word. */
> -		goto no_zawrs;
> +		__ptr32b = (u32 *)((ulong)(ptr) & ~0x3);
> +		__s = ((ulong)(ptr) & 0x2) * BITS_PER_BYTE;
> +		__val = val << __s;
> +		__mask = 0xff << __s;
> +
> +		asm volatile(
> +		"	lr.w	%0, %1\n"
> +		"	and	%0, %0, %3\n"
> +		"	xor	%0, %0, %2\n"
> +		"	bnez	%0, 1f\n"
> +			ZAWRS_WRS_NTO "\n"
> +		"1:"
> +		: "=&r" (tmp), "+A" (*(__ptr32b))
> +		: "r" (__val), "r" (__mask)
> +		: "memory");
> +		break;
>  	case 4:
>  		asm volatile(
>  		"	lr.w	%0, %1\n"
> -- 
> 2.40.1
>

Thanks,
drew