From: Rik van Riel <riel@redhat.com>
To: linux-kernel@vger.kernel.org
Cc: aquini@redhat.com, walken@google.com, eric.dumazet@gmail.com,
lwoodman@redhat.com, knoel@redhat.com, chegu_vinod@hp.com,
raghavendra.kt@linux.vnet.ibm.com, mingo@redhat.com
Subject: [PATCH -v4 4/5] x86,smp: keep spinlock delay values per hashed spinlock address
Date: Fri, 25 Jan 2013 14:18:42 -0500 [thread overview]
Message-ID: <20130125141842.58010c25@annuminas.surriel.com> (raw)
In-Reply-To: <20130125140553.060b8ced@annuminas.surriel.com>
From: Eric Dumazet <eric.dumazet@gmail.com>
Eric Dumazet found a regression with the first version of the spinlock
backoff code, in a workload where multiple spinlocks were contended,
each having a different wait time.
This patch has multiple delay values per cpu, indexed on a hash
of the lock address, to avoid that problem.
Eric Dumazet wrote:
I did some tests with your patches with following configuration :
tc qdisc add dev eth0 root htb r2q 1000 default 3
(to force a contention on qdisc lock, even with a multi queue net
device)
and 24 concurrent "netperf -t UDP_STREAM -H other_machine -- -m 128"
Machine : 2 Intel(R) Xeon(R) CPU X5660 @ 2.80GHz
(24 threads), and a fast NIC (10Gbps)
Resulting in a 13 % regression (676 Mbits -> 595 Mbits)
In this workload we have at least two contended spinlocks, with
different delays. (spinlocks are not held for the same duration)
It clearly defeats your assumption of a single per cpu delay being OK :
Some cpus are spinning too long while the lock was released.
We might try to use a hash on lock address, and an array of 16 different
delays so that different spinlocks have a chance of not sharing the same
delay.
With following patch, I get 982 Mbits/s with same bench, so an increase
of 45 % instead of a 13 % regression.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Acked-by: Rafael Aquini <aquini@redhat.com>
Reviewed-by: Michel Lespinasse <walken@google.com>
diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c
index 05f828b..3d30f6e 100644
--- a/arch/x86/kernel/smp.c
+++ b/arch/x86/kernel/smp.c
@@ -23,6 +23,7 @@
#include <linux/interrupt.h>
#include <linux/cpu.h>
#include <linux/gfp.h>
+#include <linux/hash.h>
#include <asm/mtrr.h>
#include <asm/tlbflush.h>
@@ -134,12 +135,26 @@ static bool smp_no_nmi_ipi = false;
#define DELAY_FIXED_1 (1<<DELAY_SHIFT)
#define MIN_SPINLOCK_DELAY (1 * DELAY_FIXED_1)
#define MAX_SPINLOCK_DELAY (16000 * DELAY_FIXED_1)
-DEFINE_PER_CPU(unsigned, spinlock_delay) = { MIN_SPINLOCK_DELAY };
+#define DELAY_HASH_SHIFT 6
+struct delay_entry {
+ u32 hash;
+ u32 delay;
+};
+static DEFINE_PER_CPU(struct delay_entry [1 << DELAY_HASH_SHIFT], spinlock_delay) = {
+ [0 ... (1 << DELAY_HASH_SHIFT) - 1] = {
+ .hash = 0,
+ .delay = MIN_SPINLOCK_DELAY,
+ },
+};
+
void ticket_spin_lock_wait(arch_spinlock_t *lock, struct __raw_tickets inc)
{
__ticket_t head = inc.head, ticket = inc.tail;
__ticket_t waiters_ahead;
- unsigned delay = __this_cpu_read(spinlock_delay);
+ u32 hash = hash32_ptr(lock);
+ u32 slot = hash_32(hash, DELAY_HASH_SHIFT);
+ struct delay_entry *ent = &__get_cpu_var(spinlock_delay[slot]);
+ u32 delay = (ent->hash == hash) ? ent->delay : MIN_SPINLOCK_DELAY;
unsigned loops = 1;
for (;;) {
@@ -171,11 +186,12 @@ void ticket_spin_lock_wait(arch_spinlock_t *lock, struct __raw_tickets inc)
* to get it back to a good value quickly.
*/
if (delay >= 2 * DELAY_FIXED_1)
- delay -= max(delay/32, DELAY_FIXED_1);
+ delay -= max_t(u32, delay/32, DELAY_FIXED_1);
break;
}
}
- __this_cpu_write(spinlock_delay, delay);
+ ent->hash = hash;
+ ent->delay = delay;
}
/*
next prev parent reply other threads:[~2013-01-25 19:20 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-25 19:05 [PATCH -v4 0/5] x86,smp: make ticket spinlock proportional backoff w/ auto tuning Rik van Riel
2013-01-25 19:16 ` [PATCH -v4 1/5] x86,smp: move waiting on contended ticket lock out of line Rik van Riel
2013-01-25 19:17 ` [PATCH -v4 2/5] x86,smp: proportional backoff for ticket spinlocks Rik van Riel
2013-01-25 19:17 ` [PATCH -v4 3/5] x86,smp: auto tune spinlock backoff delay factor Rik van Riel
2013-01-26 12:03 ` Ingo Molnar
2013-01-25 19:18 ` Rik van Riel [this message]
2013-01-27 13:04 ` [PATCH -v4 4/5] x86,smp: keep spinlock delay values per hashed spinlock address Michel Lespinasse
2013-02-06 20:10 ` Rik van Riel
2013-02-09 23:50 ` Michel Lespinasse
2013-01-25 19:19 ` [PATCH -v4 5/5] x86,smp: limit spinlock delay on virtual machines Rik van Riel
2013-01-26 12:00 ` Ingo Molnar
2013-01-26 12:47 ` Borislav Petkov
2013-02-04 13:50 ` Rik van Riel
2013-02-04 14:02 ` Borislav Petkov
2013-02-06 17:05 ` Rik van Riel
2013-01-28 18:18 ` Stefano Stabellini
2013-01-25 19:20 ` [DEBUG PATCH -v4 6/5] x86,smp: add debugging code to track spinlock delay value Rik van Riel
2013-01-26 7:21 ` [PATCH -v4 0/5] x86,smp: make ticket spinlock proportional backoff w/ auto tuning Mike Galbraith
2013-01-26 12:05 ` Ingo Molnar
2013-01-26 13:10 ` Mike Galbraith
2013-01-28 6:49 ` Raghavendra K T
2013-02-03 15:28 ` Chegu Vinod
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=20130125141842.58010c25@annuminas.surriel.com \
--to=riel@redhat.com \
--cc=aquini@redhat.com \
--cc=chegu_vinod@hp.com \
--cc=eric.dumazet@gmail.com \
--cc=knoel@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lwoodman@redhat.com \
--cc=mingo@redhat.com \
--cc=raghavendra.kt@linux.vnet.ibm.com \
--cc=walken@google.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
all inboxes | Powered by JetHome®