From: Beata Michalska <beata.michalska@arm.com>
To: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: linux-kernel@vger.kernel.org, peterz@infradead.org,
mingo@redhat.com, juri.lelli@redhat.com,
vincent.guittot@linaro.org, valentin.schneider@arm.com,
corbet@lwn.net, rdunlap@infradead.org, linux-doc@vger.kernel.org
Subject: Re: [PATCH v6 2/3] sched/topology: Rework CPU capacity asymmetry detection
Date: Wed, 2 Jun 2021 20:54:35 +0100 [thread overview]
Message-ID: <20210602195435.GB18136@e120325.cambridge.arm.com> (raw)
In-Reply-To: <8ea4cfc2-514b-6b5c-7269-7720a54dbb39@arm.com>
On Wed, Jun 02, 2021 at 09:09:54PM +0200, Dietmar Eggemann wrote:
> On 27/05/2021 17:38, Beata Michalska wrote:
>
> [...]
>
> > +/*
> > + * Verify whether there is any CPU capacity asymmetry in a given sched domain.
> > + * Provides sd_flags reflecting the asymmetry scope.
> > + */
> > +static inline int
> > +asym_cpu_capacity_classify(struct sched_domain *sd,
> > + const struct cpumask *cpu_map)
> > +{
> > + struct asym_cap_data *entry;
> > + int sd_asym_flags = 0;
> > + int asym_cap_count = 0;
> > + int asym_cap_miss = 0;
> > +
> > + /*
> > + * Count how many unique CPU capacities this domain spans across
> > + * (compare sched_domain CPUs mask with ones representing available
> > + * CPUs capacities). Take into account CPUs that might be offline:
> > + * skip those.
> > + */
> > + list_for_each_entry(entry, &asym_cap_list, link) {
> > + if (cpumask_intersects(sched_domain_span(sd),
> > + cpu_capacity_span(entry)))
> > + ++asym_cap_count;
> > + else if (cpumask_intersects(cpu_capacity_span(entry), cpu_map))
>
> nit: `sd span, entry span` but `entry span, cpu_map`. Why not `cpu_map, entry span`?
>
Cannot recall any reason for that.
> > + ++asym_cap_miss;
> > + }
> > + /* No asymmetry detected */
> > + if (WARN_ON_ONCE(!asym_cap_count) || asym_cap_count == 1)
> > + goto leave;
> > +
> > + sd_asym_flags |= SD_ASYM_CPUCAPACITY;
> > +
> > + /*
> > + * All the available capacities have been found within given sched
> > + * domain: no misses reported.
> > + */
> > + if (!asym_cap_miss)
> > + sd_asym_flags |= SD_ASYM_CPUCAPACITY_FULL;
> > +
> > +leave:
> > + return sd_asym_flags;
> > +}
>
> Everything looks good except that I like this more compact version better, proposed in:
>
> https://lkml.kernel.org/r/YK9ESqNEo+uacyMD@hirez.programming.kicks-ass.net
>
> And passing `const struct cpumask *sd_span` instead of `struct
> sched_domain *sd` into the function.
>
I do understand the parameter argument, but honestly don't see much difference
in naming and switching single return for asymmetric topologies vs two return
statement, but if that is more preferred/readable version I do not mind changing
that as well.
Thanks for the review.
---
BR
B.
>
> diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
> index 77b73abbb9a4..0de8eebded9f 100644
> --- a/kernel/sched/topology.c
> +++ b/kernel/sched/topology.c
> @@ -1290,13 +1290,11 @@ static LIST_HEAD(asym_cap_list);
> * Provides sd_flags reflecting the asymmetry scope.
> */
> static inline int
> -asym_cpu_capacity_classify(struct sched_domain *sd,
> +asym_cpu_capacity_classify(const struct cpumask *sd_span,
> const struct cpumask *cpu_map)
> {
> struct asym_cap_data *entry;
> - int sd_asym_flags = 0;
> - int asym_cap_count = 0;
> - int asym_cap_miss = 0;
> + int count = 0, miss = 0;
>
> /*
> * Count how many unique CPU capacities this domain spans across
> @@ -1305,27 +1303,20 @@ asym_cpu_capacity_classify(struct sched_domain *sd,
> * skip those.
> */
> list_for_each_entry(entry, &asym_cap_list, link) {
> - if (cpumask_intersects(sched_domain_span(sd),
> - cpu_capacity_span(entry)))
> - ++asym_cap_count;
> - else if (cpumask_intersects(cpu_capacity_span(entry), cpu_map))
> - ++asym_cap_miss;
> + if (cpumask_intersects(sd_span, cpu_capacity_span(entry)))
> + ++count;
> + else if (cpumask_intersects(cpu_map, cpu_capacity_span(entry)))
> + ++miss;
> }
> - /* No asymmetry detected */
> - if (WARN_ON_ONCE(!asym_cap_count) || asym_cap_count == 1)
> - goto leave;
>
> - sd_asym_flags |= SD_ASYM_CPUCAPACITY;
> + if (WARN_ON_ONCE(!count) || count == 1) /* No asymmetry */
> + return 0;
>
> - /*
> - * All the available capacities have been found within given sched
> - * domain: no misses reported.
> - */
> - if (!asym_cap_miss)
> - sd_asym_flags |= SD_ASYM_CPUCAPACITY_FULL;
> + if (miss) /* Partial asymmetry */
> + return SD_ASYM_CPUCAPACITY;
>
> -leave:
> - return sd_asym_flags;
> + /* Full asymmetry */
> + return SD_ASYM_CPUCAPACITY | SD_ASYM_CPUCAPACITY_FULL;
> }
>
> static inline void asym_cpu_capacity_update_data(int cpu)
> @@ -1510,6 +1501,7 @@ sd_init(struct sched_domain_topology_level *tl,
> struct sd_data *sdd = &tl->data;
> struct sched_domain *sd = *per_cpu_ptr(sdd->sd, cpu);
> int sd_id, sd_weight, sd_flags = 0;
> + struct cpumask *sd_span;
>
> #ifdef CONFIG_NUMA
> /*
> @@ -1557,10 +1549,11 @@ sd_init(struct sched_domain_topology_level *tl,
> #endif
> };
>
> - cpumask_and(sched_domain_span(sd), cpu_map, tl->mask(cpu));
> - sd_id = cpumask_first(sched_domain_span(sd));
> + sd_span = sched_domain_span(sd);
> + cpumask_and(sd_span, cpu_map, tl->mask(cpu));
> + sd_id = cpumask_first(sd_span);
>
> - sd->flags |= asym_cpu_capacity_classify(sd, cpu_map);
> + sd->flags |= asym_cpu_capacity_classify(sd_span, cpu_map);
>
> WARN_ONCE((sd->flags & (SD_SHARE_CPUCAPACITY | SD_ASYM_CPUCAPACITY)) ==
> (SD_SHARE_CPUCAPACITY | SD_ASYM_CPUCAPACITY),
> --
> 2.25.1
next prev parent reply other threads:[~2021-06-02 19:54 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-27 15:38 [PATCH v6 0/3] Rework CPU capacity asymmetry detection Beata Michalska
2021-05-27 15:38 ` [PATCH v6 1/3] sched/core: Introduce SD_ASYM_CPUCAPACITY_FULL sched_domain flag Beata Michalska
2021-05-27 15:38 ` [PATCH v6 2/3] sched/topology: Rework CPU capacity asymmetry detection Beata Michalska
2021-06-02 12:50 ` Valentin Schneider
2021-06-02 13:03 ` Beata Michalska
2021-06-02 19:09 ` Dietmar Eggemann
2021-06-02 19:54 ` Beata Michalska [this message]
2021-05-27 15:38 ` [PATCH v6 3/3] sched/doc: Update the CPU capacity asymmetry bits Beata Michalska
2021-06-02 19:10 ` [PATCH v6 0/3] Rework CPU capacity asymmetry detection Dietmar Eggemann
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210602195435.GB18136@e120325.cambridge.arm.com \
--to=beata.michalska@arm.com \
--cc=corbet@lwn.net \
--cc=dietmar.eggemann@arm.com \
--cc=juri.lelli@redhat.com \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=rdunlap@infradead.org \
--cc=valentin.schneider@arm.com \
--cc=vincent.guittot@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).