Re: [PATCH 08/10] sched_ext: idle: introduce SCX_PICK_IDLE_NODE

[Yury Norov <yury.norov@gmail.com>]

On Fri, Dec 20, 2024 at 04:11:40PM +0100, Andrea Righi wrote:
> Introduce a flag to restrict the selection of an idle CPU to a specific
> NUMA node.
> 
> Signed-off-by: Andrea Righi <arighi@nvidia.com>
> ---
>  kernel/sched/ext.c      |  1 +
>  kernel/sched/ext_idle.c | 11 +++++++++--
>  2 files changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/sched/ext.c b/kernel/sched/ext.c
> index 143938e935f1..da5c15bd3c56 100644
> --- a/kernel/sched/ext.c
> +++ b/kernel/sched/ext.c
> @@ -773,6 +773,7 @@ enum scx_deq_flags {
>  
>  enum scx_pick_idle_cpu_flags {
>  	SCX_PICK_IDLE_CORE	= 1LLU << 0,	/* pick a CPU whose SMT siblings are also idle */
> +	SCX_PICK_IDLE_NODE	= 1LLU << 1,	/* pick a CPU in the same target NUMA node */

SCX_FORCE_NODE or SCX_FIX_NODE?

>  };
>  
>  enum scx_kick_flags {
> diff --git a/kernel/sched/ext_idle.c b/kernel/sched/ext_idle.c
> index 444f2a15f1d4..013deaa08f12 100644
> --- a/kernel/sched/ext_idle.c
> +++ b/kernel/sched/ext_idle.c
> @@ -199,6 +199,12 @@ static s32 scx_pick_idle_cpu(const struct cpumask *cpus_allowed, int node, u64 f
>  		cpu = pick_idle_cpu_from_node(cpus_allowed, n, flags);
>  		if (cpu >= 0)
>  			break;
> +		/*
> +		 * Check if the search is restricted to the same core or
> +		 * the same node.
> +		 */
> +		if (flags & SCX_PICK_IDLE_NODE)
> +			break;

Yeah, if you will give a better name for the flag, you'll not have to
comment the code.

>  	}
>  
>  	return cpu;
> @@ -495,7 +501,8 @@ static s32 scx_select_cpu_dfl(struct task_struct *p, s32 prev_cpu,
>  		 * Search for any fully idle core in the same LLC domain.
>  		 */
>  		if (llc_cpus) {
> -			cpu = pick_idle_cpu_from_node(llc_cpus, node, SCX_PICK_IDLE_CORE);
> +			cpu = scx_pick_idle_cpu(llc_cpus, node,
> +						SCX_PICK_IDLE_CORE | SCX_PICK_IDLE_NODE);

You change it from scx_pick_idle_cpu() to pick_idle_cpu_from_node()
in patch 7 just to revert it back in patch 8...

You can use scx_pick_idle_cpu() in patch 7 already because
scx_builtin_idle_per_node is always disabled, and you always
follow the NUMA_FLAT_NODE path.  Here you will just add the
SCX_PICK_IDLE_NODE flag. 

That's the point of separating functionality and control patches. In
patch 7 you may need to mention explicitly that your new per-node
idle masks are unconditionally disabled, and will be enabled in the
last patch of the series, so some following patches will detail the
implementation.

>  			if (cpu >= 0)
>  				goto cpu_found;
>  		}
> @@ -533,7 +540,7 @@ static s32 scx_select_cpu_dfl(struct task_struct *p, s32 prev_cpu,
>  	 * Search for any idle CPU in the same LLC domain.
>  	 */
>  	if (llc_cpus) {
> -		cpu = pick_idle_cpu_from_node(llc_cpus, node, 0);
> +		cpu = scx_pick_idle_cpu(llc_cpus, node, SCX_PICK_IDLE_NODE);
>  		if (cpu >= 0)
>  			goto cpu_found;
>  	}
> -- 
> 2.47.1