From: David Woodhouse <dwmw2@infradead.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>,
"shenkai (D)" <shenkai8@huawei.com>,
"Schander, Johanna 'Mimoja' Amelie" <mimoja@amazon.com>,
LKML <linux-kernel@vger.kernel.org>,
Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
X86 ML <x86@kernel.org>, "H. Peter Anvin" <hpa@zytor.com>,
hewenliang4@huawei.com, hushiyuan@huawei.com,
luolongjun@huawei.com, hejingxian <hejingxian@huawei.com>
Subject: Re: [PATCH] use x86 cpu park to speedup smp_init in kexec situation
Date: Tue, 16 Feb 2021 21:18:24 +0000 [thread overview]
Message-ID: <ff1f4bcd728c3500f80259d1bd9319f6d4cabef2.camel@infradead.org> (raw)
In-Reply-To: <5039f6178715dc4725a8c7f071dfd9ef5d70ae43.camel@infradead.org>
[-- Attachment #1: Type: text/plain, Size: 2563 bytes --]
On Tue, 2021-02-16 at 15:10 +0000, David Woodhouse wrote:
> Actually it breaks before that, in rcu_cpu_starting(). A spinlock
> around that, an atomic_t to let the APs do their TSC sync one at a time
> (both in the above tree now), and I have a 75% saving on CPU bringup
> time for my 28-thread Haswell:
Here's a dual-socket Skylake box (EC2 c5.metal)... before:
[ 1.449015] smp: Bringing up secondary CPUs ...
[ 1.449358] x86: Booting SMP configuration:
[ 1.449578] .... node #0, CPUs: #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16 #17 #18 #19 #20 #21 #22 #23
[ 1.514986] .... node #1, CPUs: #24 #25 #26 #27 #28 #29 #30 #31 #32 #33 #34 #35 #36 #37 #38 #39 #40 #41 #42 #43 #44 #45 #46 #47
[ 1.644978] .... node #0, CPUs: #48 #49 #50 #51 #52 #53 #54 #55 #56 #57 #58 #59 #60 #61 #62 #63 #64 #65 #66 #67 #68 #69 #70 #71
[ 1.711970] .... node #1, CPUs: #72 #73 #74 #75 #76 #77 #78 #79 #80 #81 #82 #83 #84 #85 #86 #87 #88 #89 #90 #91 #92 #93 #94 #95
[ 1.781438] Brought CPUs online in 1063391744 cycles
[ 1.782289] smp: Brought up 2 nodes, 96 CPUs
[ 1.782515] smpboot: Max logical packages: 2
[ 1.782738] smpboot: Total of 96 processors activated (576364.89 BogoMIPS)
... and after:
[ 1.445661] smp: Bringing up secondary CPUs ...
[ 1.446004] x86: Booting SMP configuration:
[ 1.446047] .... node #0, CPUs: #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16 #17 #18 #19 #20 #21 #22 #23
[ 1.451048] .... node #1, CPUs: #24 #25 #26 #27 #28 #29 #30 #31 #32 #33 #34 #35 #36 #37 #38 #39 #40 #41 #42 #43 #44 #45 #46 #47
[ 1.455046] .... node #0, CPUs: #48 #49 #50 #51 #52 #53 #54 #55 #56 #57 #58 #59 #60 #61 #62 #63 #64 #65 #66 #67 #68 #69 #70 #71
[ 1.462050] .... node #1, CPUs: #72 #73 #74 #75 #76 #77 #78 #79 #80 #81 #82 #83 #84 #85 #86 #87 #88 #89 #90 #91 #92 #93 #94 #95
[ 1.465983] Brought 96 CPUs to x86/cpu:kick in 68802688 cycles
[ 0.094170] smpboot: CPU 26 Converting physical 1 to logical package 2
[ 1.468078] Brought 96 CPUs to x86/cpu:wait-init in 5178300 cycles
[ 1.476112] Brought CPUs online in 23479546 cycles
[ 1.476298] smp: Brought up 2 nodes, 96 CPUs
[ 1.476520] smpboot: Max logical packages: 2
[ 1.476743] smpboot: Total of 96 processors activated (576000.00 BogoMIPS)
So the CPU bringup went from 334ms to 31ms on that one.
I might try that without spewing over 1KiB to the serial port at 115200
baud during the proceedings, and see if it makes a bigger difference :)
[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5174 bytes --]
next prev parent reply other threads:[~2021-02-16 21:19 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <87ft22dxop.fsf@nanos.tec.linutronix.de>
[not found] ` <27357c74bdc3b52bdf59e6f48cd8690495116d64.camel@infradead.org>
[not found] ` <877dnedt7l.fsf@nanos.tec.linutronix.de>
[not found] ` <87zh09tcqz.fsf@nanos.tec.linutronix.de>
2021-02-16 13:53 ` David Woodhouse
2021-02-16 15:10 ` David Woodhouse
2021-02-16 21:18 ` David Woodhouse [this message]
2021-12-08 14:14 ` David Woodhouse
2021-12-08 14:50 ` Paul E. McKenney
2021-12-08 15:10 ` David Woodhouse
2021-12-08 16:57 ` David Woodhouse
2021-12-08 17:35 ` Paul E. McKenney
2021-12-08 18:32 ` David Woodhouse
2021-12-08 19:03 ` Paul E. McKenney
2021-12-08 20:35 ` David Woodhouse
2021-12-08 21:09 ` Paul E. McKenney
2020-12-15 14:46 shenkai (D)
2020-12-15 16:31 ` Andy Lutomirski
2020-12-15 21:20 ` Thomas Gleixner
2020-12-16 8:45 ` shenkai (D)
2020-12-16 10:12 ` Thomas Gleixner
2020-12-16 14:18 ` shenkai (D)
2020-12-16 15:31 ` Thomas Gleixner
2020-12-17 14:53 ` shenkai (D)
2021-01-07 15:18 ` David Woodhouse
2021-01-19 12:12 ` David Woodhouse
2021-01-21 14:55 ` Thomas Gleixner
2021-01-21 15:42 ` David Woodhouse
2021-01-21 17:34 ` David Woodhouse
2021-02-01 10:36 ` David Woodhouse
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=ff1f4bcd728c3500f80259d1bd9319f6d4cabef2.camel@infradead.org \
--to=dwmw2@infradead.org \
--cc=bp@alien8.de \
--cc=hejingxian@huawei.com \
--cc=hewenliang4@huawei.com \
--cc=hpa@zytor.com \
--cc=hushiyuan@huawei.com \
--cc=linux-kernel@vger.kernel.org \
--cc=luolongjun@huawei.com \
--cc=luto@kernel.org \
--cc=mimoja@amazon.com \
--cc=mingo@redhat.com \
--cc=shenkai8@huawei.com \
--cc=tglx@linutronix.de \
--cc=x86@kernel.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
all inboxes | Powered by JetHome®