Re: [PATCH -v2] x86/split_lock: Restructure the unwieldy switch-case in sld_state_show()

[Luck, Tony]

On Thu, Feb 26, 2026 at 03:50:33PM +0100, Borislav Petkov wrote:

Only need the cpuhp_setup_state() for the sld_state == sld_warn case.

> +	if (boot_cpu_has(X86_FEATURE_SPLIT_LOCK_DETECT)) {
> +		pr_info("#AC: crashing the kernel on kernel split_locks and %s on user-space split_locks\n", action);
> +
	    if (sld_state == sld_warn) {
> +		/*
> +		 * This is handling the case where a CPU goes offline at the
> +		 * moment where split lock detection is disabled in the warn
> +		 * setting, see split_lock_warn(). It doesn't have any effect
> +		 * in the fatal case.
> +                 */
> +		if (cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "x86/splitlock", NULL, splitlock_cpu_offline) < 0)
> +			pr_warn("No splitlock CPU offline handler\n");
            }
> +
> +	} else if (boot_cpu_has(X86_FEATURE_BUS_LOCK_DETECT)) {
> +		pr_info("#DB: %s on user-space bus_locks\n", action);
>  	}
>  }

-Tony