From: "Paul E. McKenney" <paulmck@kernel.org>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc: bpf <bpf@vger.kernel.org>, Daniel Borkmann <daniel@iogearbox.net>,
Kernel Team <Kernel-team@fb.com>
Subject: Re: slow sync rcu_tasks_trace
Date: Wed, 9 Sep 2020 04:38:58 -0700 [thread overview]
Message-ID: <20200909113858.GF29330@paulmck-ThinkPad-P72> (raw)
In-Reply-To: <CAADnVQK_AiX+S_L_A4CQWT11XyveppBbQSQgH_qWGyzu_E8Yeg@mail.gmail.com>
On Tue, Sep 08, 2020 at 07:34:20PM -0700, Alexei Starovoitov wrote:
> Hi Paul,
>
> Looks like sync rcu_tasks_trace got slower or we simply didn't notice
> it earlier.
>
> In selftests/bpf try:
> time ./test_progs -t trampoline_count
> #101 trampoline_count:OK
> Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
>
> real 1m17.082s
> user 0m0.145s
> sys 0m1.369s
>
> But with the following hack:
> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> index 7dd523a7e32d..c417b817ec5d 100644
> --- a/kernel/bpf/trampoline.c
> +++ b/kernel/bpf/trampoline.c
> @@ -217,7 +217,7 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr)
> * programs finish executing.
> * Wait for these two grace periods together.
> */
> - synchronize_rcu_mult(call_rcu_tasks, call_rcu_tasks_trace);
> +// synchronize_rcu_mult(call_rcu_tasks, call_rcu_tasks_trace);
>
> I see:
> time ./test_progs -t trampoline_count
> #101 trampoline_count:OK
> Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
>
> real 0m1.588s
> user 0m0.131s
> sys 0m1.342s
>
> It takes an extra minute to do 40 sync rcu_tasks_trace calls.
> It means that every sync takes more than a second.
> That feels excessive.
>
> Doing:
> - synchronize_rcu_mult(call_rcu_tasks, call_rcu_tasks_trace);
> + synchronize_rcu();
> is also fast:
> time ./test_progs -t trampoline_count
> #101 trampoline_count:OK
> Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
>
> real 0m2.089s
> user 0m0.139s
> sys 0m1.282s
>
> sync rcu_tasks() is fast too:
> - synchronize_rcu_mult(call_rcu_tasks, call_rcu_tasks_trace);
> + synchronize_rcu_tasks();
> time ./test_progs -t trampoline_count
> #101 trampoline_count:OK
> Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
>
> real 0m2.209s
> user 0m0.117s
> sys 0m1.344s
>
> so it's really something going on with sync rcu_tasks_trace.
> Could you please take a look?
I am guessing that your .config has CONFIG_TASKS_TRACE_RCU_READ_MB=n.
If I am wrong, please try CONFIG_TASKS_TRACE_RCU_READ_MB=y.
Otherwise (or alternatively), could you please try booting with
rcupdate.rcu_task_ipi_delay=50? The default value is 500, or half a
second on a HZ=1000 system, which on a busy system could easily result
in the grace-period delays that you are seeing. The value of this
kernel boot parameter does interact with the tasklist-scan backoffs,
so its effect will not likely be linear.
Do either of those approaches help?
Thanx, Paul
next prev parent reply other threads:[~2020-09-09 11:45 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-09 2:34 slow sync rcu_tasks_trace Alexei Starovoitov
2020-09-09 11:38 ` Paul E. McKenney [this message]
2020-09-09 15:10 ` Jiri Olsa
2020-09-09 17:02 ` Paul E. McKenney
2020-09-09 17:12 ` Alexei Starovoitov
2020-09-09 17:35 ` Paul E. McKenney
2020-09-09 18:04 ` Alexei Starovoitov
2020-09-09 19:39 ` Paul E. McKenney
2020-09-09 19:48 ` Alexei Starovoitov
2020-09-09 21:04 ` Paul E. McKenney
2020-09-09 21:22 ` Paul E. McKenney
2020-09-10 5:27 ` Paul E. McKenney
2020-09-10 18:33 ` Alexei Starovoitov
2020-09-10 18:51 ` Paul E. McKenney
2020-09-10 19:04 ` Alexei Starovoitov
2020-09-10 20:24 ` Paul E. McKenney
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=20200909113858.GF29330@paulmck-ThinkPad-P72 \
--to=paulmck@kernel.org \
--cc=Kernel-team@fb.com \
--cc=alexei.starovoitov@gmail.com \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
/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).