Re: [RFC PATCH] x86 / hibernate: Eliminate the redundant smp_ops.play_dead assignment

[Rafael J. Wysocki]

On Mon, Feb 24, 2025 at 8:45 AM WangYuli <wangyuli@xxxxxxxxxxxxx> wrote:
>
> It's unnecessary to re-initialize smp_ops.play_dead to play_dead as
> it naturally goes back to play_dead in the freshly booted kernel upon
> device resume.

Is it really?

And what if freeze_secondary_cpus() returns an error, for one example?

> Suggested-by: Huacai Chen <chenhuacai@xxxxxxxxxxx>
> Co-developed-by: Wentao Guan <guanwentao@xxxxxxxxxxxxx>
> Signed-off-by: Wentao Guan <guanwentao@xxxxxxxxxxxxx>
> Signed-off-by: WangYuli <wangyuli@xxxxxxxxxxxxx>
> ---
>  arch/x86/power/cpu.c | 9 ++-------
>  1 file changed, 2 insertions(+), 7 deletions(-)
>
> diff --git a/arch/x86/power/cpu.c b/arch/x86/power/cpu.c
> index 63230ff8cf4f..023cf9421467 100644
> --- a/arch/x86/power/cpu.c
> +++ b/arch/x86/power/cpu.c
> @@ -297,9 +297,6 @@ static void __noreturn resume_play_dead(void)
>
>  int hibernate_resume_nonboot_cpu_disable(void)
>  {
> -       void (*play_dead)(void) = smp_ops.play_dead;
> -       int ret;
> -
>         /*
>          * Ensure that MONITOR/MWAIT will not be used in the "play dead" loop
>          * during hibernate image restoration, because it is likely that the
> @@ -316,13 +313,11 @@ int hibernate_resume_nonboot_cpu_disable(void)
>          * resume) sleep afterwards, and the resumed kernel will decide itself
>          * what to do with them.
>          */
> -       ret = cpuhp_smt_enable();
> +       int ret = cpuhp_smt_enable();
>         if (ret)
>                 return ret;
>         smp_ops.play_dead = resume_play_dead;
> -       ret = freeze_secondary_cpus(0);
> -       smp_ops.play_dead = play_dead;
> -       return ret;
> +       return freeze_secondary_cpus(0);
>  }
>  #endif
>
> --
> 2.47.2
>