Re: [PATCH v4 1/3] time: add "NOW() good" indicator

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Tue, Jun 30, 2026 at 04:06:00PM +0200, Jan Beulich wrote:
> printk_start_of_line() checks for a value of 0 right now. In order to be
> able to have NOW() return at least monotonically increasing values, that
> needs replacing by an explicit indicator.
> 
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> ---
> Arm and RISC-V may want to consider whether their initial get_cycles()
> can't be moved yet earlier, such that the indicator also can be set
> yet earlier.
> ---
> v4: Add barriers.
> v3: New.
> 
> --- a/xen/arch/arm/time.c
> +++ b/xen/arch/arm/time.c
> @@ -145,6 +145,8 @@ void __init preinit_xen_time(void)
>          panic("Timer: Cannot initialize platform timer\n");
>  
>      boot_count = get_cycles();
> +    smp_wmb();
> +    NOW_good = true;
>  }
>  
>  static void __init init_dt_xen_time(void)
> --- a/xen/arch/riscv/time.c
> +++ b/xen/arch/riscv/time.c
> @@ -87,6 +87,8 @@ void __init preinit_xen_time(void)
>          panic("%s: ACPI isn't supported\n", __func__);
>  
>      boot_clock_cycles = get_cycles();
> +    smp_wmb();
> +    NOW_good = true;
>  
>      /* set_xen_timer must have been set by sbi_init() already */
>      ASSERT(set_xen_timer);
> --- a/xen/arch/x86/time.c
> +++ b/xen/arch/x86/time.c
> @@ -2660,6 +2660,7 @@ void __init early_time_init(void)
>  
>      set_time_scale(&t->tsc_scale, tmp);
>      t->stamp.local_tsc = boot_tsc_stamp;
> +    NOW_good = true;

Would you need a barrier here to ensure compiler doesn't re-order the
writes?  Maybe using ACCESS_ONCE(), or a smp_wmb() like it's used in
other arches?

Thanks, Roger.