Hi!

On Wed, Jul 20, 2022 at 11:21:32PM +1000, Nicholas Piggin wrote:
> We want to move away from using SMT prioroty updates for cpu_relax, and

(typo, "priority")

> +#define spin_begin()                                                 \
> +do {                                                                 \
> +     asm volatile(ASM_FTR_IFCLR(                                     \
> +             "or 1,1,1", /* HMT_LOW */                               \
> +             "nop",/* POWER10 onward uses pause_short (wait 2,0) */  \
> +             %0) :: "i" (CPU_FTR_ARCH_31) : "memory");               \
> +} while (0)

Is that nop patched later?  Or should you change the comment, maybe?

> +#define spin_cpu_relax()                                             \
> +do {                                                                 \
> +     asm volatile(ASM_FTR_IFCLR(                                     \
> +             /* Pre-POWER10 uses low / medium priority nops */       \
> +             "nop",                                                  \

"nop" aka "or 0,0,0" does not change program priority?  Medium low would
be "or 6,6,6", not sure if that is the ppr you wanted here?

> +             /* POWER10 onward uses pause_short (wait 2,0) */        \
> +             PPC_WAIT(2, 0),                                         \
> +             %0) :: "i" (CPU_FTR_ARCH_31) : "memory");               \
> +} while (0)
> +
> +#define spin_end()                                                   \
> +do {                                                                 \
> +     asm volatile(ASM_FTR_IFCLR(                                     \
> +             "or 2,2,2", /* HMT_MEDIUM */                            \
> +             "nop",/* POWER10 onward uses pause_short (wait 2,0) */  \
> +             %0) :: "i" (CPU_FTR_ARCH_31) : "memory");               \
> +} while (0)

Same comment as for spin_begin.

Reviewed-by: Segher Boessenkool <seg...@kernel.crashing.org>


Segher

Reply via email to