linux-kernel - Re: [PATCH 3/8] sched_ext: idle: Extend topology optimizations to all tasks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Z9SL9chJFYswO5MU@gpd3>
Date: Fri, 14 Mar 2025 21:05:09 +0100
From: Andrea Righi <arighi@...dia.com>
To: Tejun Heo <tj@...nel.org>
Cc: David Vernet <void@...ifault.com>, Changwoo Min <changwoo@...lia.com>,
	bpf@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/8] sched_ext: idle: Extend topology optimizations to
 all tasks

On Fri, Mar 14, 2025 at 08:21:49AM -1000, Tejun Heo wrote:
> Hello,
> 
> On Fri, Mar 14, 2025 at 10:45:35AM +0100, Andrea Righi wrote:
> ...
> > -	if (p->nr_cpus_allowed >= num_possible_cpus()) {
> > -		if (static_branch_maybe(CONFIG_NUMA, &scx_selcpu_topo_numa))
> > -			numa_cpus = numa_span(prev_cpu);
> > -
> > -		if (static_branch_maybe(CONFIG_SCHED_MC, &scx_selcpu_topo_llc))
> > -			llc_cpus = llc_span(prev_cpu);
> > +	if (static_branch_maybe(CONFIG_NUMA, &scx_selcpu_topo_numa)) {
> > +		struct cpumask *cpus = numa_span(prev_cpu);
> > +
> > +		if (cpus && !cpumask_equal(cpus, p->cpus_ptr)) {
> > +			if (cpumask_subset(cpus, p->cpus_ptr)) {
> > +				numa_cpus = cpus;
> > +			} else {
> > +				numa_cpus = this_cpu_cpumask_var_ptr(local_numa_idle_cpumask);
> > +				if (!cpumask_and(numa_cpus, cpus, p->cpus_ptr))
> > +					numa_cpus = NULL;
> > +			}
> > +		}
> > +	}
> > +	if (static_branch_maybe(CONFIG_SCHED_MC, &scx_selcpu_topo_llc)) {
> > +		struct cpumask *cpus = llc_span(prev_cpu);
> > +
> > +		if (cpus && !cpumask_equal(cpus, p->cpus_ptr)) {
> > +			if (cpumask_subset(cpus, p->cpus_ptr)) {
> > +				llc_cpus = cpus;
> > +			} else {
> > +				llc_cpus = this_cpu_cpumask_var_ptr(local_llc_idle_cpumask);
> > +				if (!cpumask_and(llc_cpus, cpus, p->cpus_ptr))
> > +					llc_cpus = NULL;
> > +			}
> > +		}
> > 
> 
> Wouldn't it still make sense to special case p->nr_cpus_allowed >=
> num_possible_cpus()? That'd be vast majority of cases and we can skip all
> the cpumask comparisons for them.

Yeah, that's probably a good idea, I'll re-add that in the next version.

Thanks,
-Andrea