From: Marcelo Tosatti <mtosatti@redhat.com>
To: Christoph Lameter <cl@linux.com>
Cc: Aaron Tomlin <atomlin@atomlin.com>,
Frederic Weisbecker <frederic@kernel.org>,
Andrew Morton <akpm@linux-foundation.org>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
Marcelo Tosatti <mtosatti@redhat.com>
Subject: [PATCH v2 10/11] mm/vmstat: switch vmstat shepherd to flush per-CPU counters remotely
Date: Thu, 09 Feb 2023 12:02:00 -0300 [thread overview]
Message-ID: <20230209153204.901518530@redhat.com> (raw)
In-Reply-To: 20230209150150.380060673@redhat.com
Now that the counters are modified via cmpxchg both CPU locally
(via the account functions), and remotely (via cpu_vm_stats_fold),
its possible to switch vmstat_shepherd to perform the per-CPU
vmstats folding remotely.
This fixes the following two problems:
1. A customer provided some evidence which indicates that
the idle tick was stopped; albeit, CPU-specific vmstat
counters still remained populated.
Thus one can only assume quiet_vmstat() was not
invoked on return to the idle loop. If I understand
correctly, I suspect this divergence might erroneously
prevent a reclaim attempt by kswapd. If the number of
zone specific free pages are below their per-cpu drift
value then zone_page_state_snapshot() is used to
compute a more accurate view of the aforementioned
statistic. Thus any task blocked on the NUMA node
specific pfmemalloc_wait queue will be unable to make
significant progress via direct reclaim unless it is
killed after being woken up by kswapd
(see throttle_direct_reclaim())
2. With a SCHED_FIFO task that busy loops on a given CPU,
and kworker for that CPU at SCHED_OTHER priority,
queuing work to sync per-vmstats will either cause that
work to never execute, or stalld (i.e. stall daemon)
boosts kworker priority which causes a latency
violation
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Index: linux-2.6/mm/vmstat.c
===================================================================
--- linux-2.6.orig/mm/vmstat.c
+++ linux-2.6/mm/vmstat.c
@@ -2007,6 +2007,23 @@ static void vmstat_shepherd(struct work_
static DECLARE_DEFERRABLE_WORK(shepherd, vmstat_shepherd);
+#ifdef CONFIG_HAVE_CMPXCHG_LOCAL
+/* Flush counters remotely if CPU uses cmpxchg to update its per-CPU counters */
+static void vmstat_shepherd(struct work_struct *w)
+{
+ int cpu;
+
+ cpus_read_lock();
+ for_each_online_cpu(cpu) {
+ cpu_vm_stats_fold(cpu);
+ cond_resched();
+ }
+ cpus_read_unlock();
+
+ schedule_delayed_work(&shepherd,
+ round_jiffies_relative(sysctl_stat_interval));
+}
+#else
static void vmstat_shepherd(struct work_struct *w)
{
int cpu;
@@ -2026,6 +2043,7 @@ static void vmstat_shepherd(struct work_
schedule_delayed_work(&shepherd,
round_jiffies_relative(sysctl_stat_interval));
}
+#endif
static void __init start_shepherd_timer(void)
{
next prev parent reply other threads:[~2023-02-09 15:34 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-09 15:01 [PATCH v2 00/11] fold per-CPU vmstats remotely Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 01/11] mm/vmstat: remove remote node draining Marcelo Tosatti
2023-02-28 15:53 ` David Hildenbrand
2023-02-28 19:36 ` Marcelo Tosatti
2023-03-02 10:10 ` David Hildenbrand
2023-03-21 15:20 ` Mel Gorman
2023-03-21 17:31 ` Marcelo Tosatti
2023-03-02 17:21 ` Peter Xu
2023-03-02 17:27 ` Peter Xu
2023-03-02 19:17 ` Marcelo Tosatti
2023-03-02 18:56 ` Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 02/11] this_cpu_cmpxchg: ARM64: switch this_cpu_cmpxchg to locked, add _local function Marcelo Tosatti
2023-03-02 10:42 ` David Hildenbrand
2023-03-02 10:51 ` David Hildenbrand
2023-03-02 14:32 ` Marcelo Tosatti
2023-03-02 20:53 ` Peter Xu
2023-03-02 21:04 ` Marcelo Tosatti
2023-03-02 21:25 ` Peter Xu
2023-03-03 15:39 ` Marcelo Tosatti
2023-03-03 15:47 ` Marcelo Tosatti
2023-03-15 23:56 ` Christoph Lameter
2023-03-16 10:54 ` Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 03/11] this_cpu_cmpxchg: loongarch: " Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 04/11] this_cpu_cmpxchg: S390: " Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 05/11] this_cpu_cmpxchg: x86: " Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 06/11] this_cpu_cmpxchg: asm-generic: " Marcelo Tosatti
2023-02-09 15:01 ` [PATCH v2 07/11] convert this_cpu_cmpxchg users to this_cpu_cmpxchg_local Marcelo Tosatti
2023-03-02 20:54 ` Peter Xu
2023-02-09 15:01 ` [PATCH v2 08/11] mm/vmstat: switch counter modification to cmpxchg Marcelo Tosatti
2023-03-02 10:47 ` David Hildenbrand
2023-03-02 14:47 ` Marcelo Tosatti
2023-03-02 16:20 ` Peter Xu
2023-03-02 19:11 ` Marcelo Tosatti
2023-03-02 20:06 ` Peter Xu
2023-02-09 15:01 ` [PATCH v2 09/11] mm/vmstat: use cmpxchg loop in cpu_vm_stats_fold Marcelo Tosatti
2023-03-01 22:57 ` Peter Xu
2023-03-02 13:55 ` Marcelo Tosatti
2023-03-02 21:19 ` Peter Xu
2023-03-03 15:17 ` Marcelo Tosatti
2023-02-09 15:02 ` Marcelo Tosatti [this message]
2023-03-02 21:01 ` [PATCH v2 10/11] mm/vmstat: switch vmstat shepherd to flush per-CPU counters remotely Peter Xu
2023-03-02 21:16 ` Marcelo Tosatti
2023-03-02 21:30 ` Peter Xu
2023-02-09 15:02 ` [PATCH v2 11/11] mm/vmstat: refresh stats remotely instead of via work item Marcelo Tosatti
2023-02-23 14:54 ` [PATCH v2 00/11] fold per-CPU vmstats remotely Marcelo Tosatti
2023-02-24 2:34 ` Hillf Danton
2023-02-27 19:41 ` Marcelo Tosatti
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=20230209153204.901518530@redhat.com \
--to=mtosatti@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=atomlin@atomlin.com \
--cc=cl@linux.com \
--cc=frederic@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.