From: Peter Zijlstra <peterz@infradead.org>
To: Frederic Weisbecker <frederic@kernel.org>
Cc: LKML <linux-kernel@vger.kernel.org>,
Jacek Anaszewski <jacek.anaszewski@gmail.com>,
Wanpeng Li <wanpengli@tencent.com>,
"Rafael J . Wysocki" <rjw@rjwysocki.net>,
Rik van Riel <riel@surriel.com>,
Thomas Gleixner <tglx@linutronix.de>,
Yauheni Kaliuta <yauheni.kaliuta@redhat.com>,
Ingo Molnar <mingo@kernel.org>,
Viresh Kumar <viresh.kumar@linaro.org>,
Pavel Machek <pavel@ucw.cz>
Subject: Re: [PATCH 11/14] sched/kcpustat: Introduce vtime-aware kcpustat accessor for CPUTIME_SYSTEM
Date: Thu, 24 Oct 2019 13:50:34 +0200 [thread overview]
Message-ID: <20191024115034.GA4114@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20191016025700.31277-12-frederic@kernel.org>
On Wed, Oct 16, 2019 at 04:56:57AM +0200, Frederic Weisbecker wrote:
> +static int kcpustat_field_vtime(u64 *cpustat,
> + struct vtime *vtime,
> + enum cpu_usage_stat usage,
> + int cpu, u64 *val)
> +{
> + unsigned int seq;
> + int err;
> +
> + do {
> + seq = read_seqcount_begin(&vtime->seqcount);
> +
> + /*
> + * We raced against context switch, fetch the
> + * kcpustat task again.
> + */
> + if (vtime->cpu != cpu && vtime->cpu != -1) {
> + err = -EAGAIN;
> + continue;
Did that want to be break?
> + }
> +
> + /*
> + * Two possible things here:
> + * 1) We are seeing the scheduling out task (prev) or any past one.
> + * 2) We are seeing the scheduling in task (next) but it hasn't
> + * passed though vtime_task_switch() yet so the pending
> + * cputime of the prev task may not be flushed yet.
> + *
> + * Case 1) is ok but 2) is not. So wait for a safe VTIME state.
> + */
> + if (vtime->state == VTIME_INACTIVE) {
> + err = -EAGAIN;
> + continue;
Idem.
If so, you can do return -EAGAIN here, and return 0 at the end and get
rid of err.
Also, if you're spin-waiting here, there should probably be a
cpu_relax() before the return -EAGAIN.
And in case that is so, you probably want the rcu_read_lock() section
below _inside_ the do{}while loop, such that the RCU section doesn't
cover the entire spin-wait.
> + }
> +
> + err = 0;
> +
> + *val = cpustat[usage];
> +
> + if (vtime->state == VTIME_SYS)
> + *val += vtime->stime + vtime_delta(vtime);
> +
> + } while (read_seqcount_retry(&vtime->seqcount, seq));
> +
> + return err;
> +}
> +
> +u64 kcpustat_field(struct kernel_cpustat *kcpustat,
> + enum cpu_usage_stat usage, int cpu)
> +{
> + u64 val;
> + int err;
> + u64 *cpustat = kcpustat->cpustat;
> +
> + if (!vtime_accounting_enabled_cpu(cpu))
> + return cpustat[usage];
> +
> + /* Only support sys vtime for now */
> + if (usage != CPUTIME_SYSTEM)
> + return cpustat[usage];
> +
> + rcu_read_lock();
> +
> + do {
> + struct rq *rq = cpu_rq(cpu);
> + struct task_struct *curr;
> + struct vtime *vtime;
> +
> + curr = rcu_dereference(rq->curr);
This is indeed safe now (relies on commit
5311a98fef7d ("tasks, sched/core: RCUify the assignment of rq->curr")
)
> + if (WARN_ON_ONCE(!curr)) {
> + val = cpustat[usage];
> + break;
> + }
> +
> + vtime = &curr->vtime;
> + err = kcpustat_field_vtime(cpustat, vtime, usage, cpu, &val);
> + } while (err == -EAGAIN);
> +
> + rcu_read_unlock();
> +
> + return val;
> +}
> +
> #endif /* CONFIG_VIRT_CPU_ACCOUNTING_GEN */
> --
> 2.23.0
>
next prev parent reply other threads:[~2019-10-24 11:51 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-16 2:56 [PATCH 00/14] sched/nohz: Make kcpustat's CPUTIME_SYSTEM vtime aware v2 (Partially fix kcpustat on nohz_full) Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 01/14] sched/vtime: Record CPU under seqcount for kcpustat needs Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 02/14] sched/cputime: Add vtime idle task state Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 03/14] sched/cputime: Add vtime guest " Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 04/14] context_tracking: Remove context_tracking_active() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 05/14] context_tracking: s/context_tracking_is_enabled/context_tracking_enabled() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] context_tracking: Rename context_tracking_is_enabled() => context_tracking_enabled() tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 06/14] context_tracking: Rename context_tracking_is_cpu_enabled() to context_tracking_enabled_this_cpu() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 07/14] context_tracking: Introduce context_tracking_enabled_cpu() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 08/14] sched/vtime: Rename vtime_accounting_cpu_enabled() to vtime_accounting_enabled_this_cpu() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 09/14] sched/vtime: Introduce vtime_accounting_enabled_cpu() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 10/14] context_tracking: Check static key on context_tracking_enabled_*cpu() Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 11/14] sched/kcpustat: Introduce vtime-aware kcpustat accessor for CPUTIME_SYSTEM Frederic Weisbecker
2019-10-24 11:50 ` Peter Zijlstra [this message]
2019-10-25 1:25 ` Frederic Weisbecker
2019-10-25 2:03 ` [PATCH 11/14 v2] " Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 12/14] procfs: Use vtime aware kcpustat accessor to fetch CPUTIME_SYSTEM Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:56 ` [PATCH 13/14] cpufreq: " Frederic Weisbecker
2019-10-16 3:40 ` Viresh Kumar
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-16 2:57 ` [PATCH 14/14] leds: " Frederic Weisbecker
2019-10-29 9:52 ` [tip: sched/core] " tip-bot2 for Frederic Weisbecker
2019-10-24 0:45 ` [GIT PULL] sched/nohz: Make kcpustat's CPUTIME_SYSTEM vtime aware Frederic Weisbecker
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=20191024115034.GA4114@hirez.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=frederic@kernel.org \
--cc=jacek.anaszewski@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=pavel@ucw.cz \
--cc=riel@surriel.com \
--cc=rjw@rjwysocki.net \
--cc=tglx@linutronix.de \
--cc=viresh.kumar@linaro.org \
--cc=wanpengli@tencent.com \
--cc=yauheni.kaliuta@redhat.com \
/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).