linux-kernel - Re: [PATCH 0/5] Make Cluster Scheduling Configurable

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: <20211206220550.GA18698@ranerica-svr.sc.intel.com>
Date:   Mon, 6 Dec 2021 14:05:50 -0800
From:   Ricardo Neri <ricardo.neri-calderon@...ux.intel.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Tim Chen <tim.c.chen@...ux.intel.com>,
        Arjan Van De Ven <arjan.van.de.ven@...el.com>,
        Ricardo Neri <ricardo.neri@...el.com>,
        Len Brown <len.brown@...el.com>,
        Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>,
        Artem Bityutskiy <artem.bityutskiy@...ux.intel.com>,
        Chen Yu <yu.c.chen@...el.com>,
        Song Bao Hua <song.bao.hua@...ilicon.com>,
        yangyicong <yangyicong@...wei.com>,
        Michael Larabel <Michael@...haelLarabel.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 0/5] Make Cluster Scheduling Configurable

On Sat, Dec 04, 2021 at 10:14:02AM +0100, Peter Zijlstra wrote:
> On Fri, Dec 03, 2021 at 12:32:37PM -0800, Tim Chen wrote:
> > Tim Chen (5):
> >   scheduler: Create SDTL_SKIP flag to skip topology level
> >   scheduler: Add SD_CLUSTER topology flag to cluster sched domain
> >   scheduler: Add runtime knob sysctl_sched_cluster
> >   scheduler: Add boot time enabling/disabling of cluster scheduling
> >   scheduler: Default cluster scheduling to off on x86 hybrid CPU
> 
> s/scheduler:/sched:/, surely?
> 
> >  .../admin-guide/kernel-parameters.txt         |  4 +
> >  arch/x86/kernel/smpboot.c                     | 26 +++++++
> >  drivers/base/arch_topology.c                  | 23 +++++-
> >  include/linux/sched/sd_flags.h                |  7 ++
> >  include/linux/sched/sysctl.h                  |  6 ++
> >  include/linux/sched/topology.h                |  3 +-
> >  include/linux/topology.h                      |  7 ++
> >  kernel/sched/core.c                           |  1 +
> >  kernel/sched/sched.h                          |  6 ++
> >  kernel/sched/topology.c                       | 75 ++++++++++++++++++-
> >  kernel/sysctl.c                               | 11 +++
> >  11 files changed, 163 insertions(+), 6 deletions(-)
> 
> *groan*,... I was more thinking of something like so.
> 
> ---
>  arch/x86/kernel/smpboot.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
> index ac2909f0cab3..617012f4619f 100644
> --- a/arch/x86/kernel/smpboot.c
> +++ b/arch/x86/kernel/smpboot.c
> @@ -579,6 +579,17 @@ static struct sched_domain_topology_level x86_numa_in_package_topology[] = {
>  	{ NULL, },
>  };
>  
> +static struct sched_domain_topology_level x86_hybrid_topology[] = {
> +#ifdef CONFIG_SCHED_SMT
> +	{ cpu_smt_mask, x86_smt_flags, SD_INIT_NAME(SMT) },
> +#endif
> +#ifdef CONFIG_SCHED_MC
> +	{ cpu_coregroup_mask, x86_core_flags, SD_INIT_NAME(MC) },
> +#endif
> +	{ cpu_cpu_mask, SD_INIT_NAME(DIE) },
> +	{ NULL, },
> +};
> +
>  static struct sched_domain_topology_level x86_topology[] = {
>  #ifdef CONFIG_SCHED_SMT
>  	{ cpu_smt_mask, x86_smt_flags, SD_INIT_NAME(SMT) },
> @@ -1469,8 +1480,11 @@ void __init native_smp_cpus_done(unsigned int max_cpus)
>  
>  	calculate_max_logical_packages();
>  
> +	/* XXX for now assume numa-in-package and hybrid don't overlap */
>  	if (x86_has_numa_in_package)
>  		set_sched_topology(x86_numa_in_package_topology);
> +	if (cpu_feature_enabled(X86_FEATURE_HYBRID_CPU))
> +		set_sched_topology(x86_hybrid_topology);
>  
>  	nmi_selftest();
>  	impress_friends();

FWIW, I tested this and now I don't see cluster scheduling interfering
with asymmetric packing when running on a x86 hybrid topology.

Tested-by: Ricardo Neri <ricardo.neri-calderon@...ux.intel.com>