linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: wolfgang huang <wolfgang9277@126.com>
To: akpm@linux-foundation.org
Cc: linux-mm@kvack.org, wolfgang huang <huangjinhui@kylinos.cn>
Subject: [PATCH] mm/khugepaged: make reserved memory adaptively
Date: Sat,  2 Sep 2023 13:26:11 +0800	[thread overview]
Message-ID: <20230902052611.144972-1-wolfgang9277@126.com> (raw)

From: wolfgang huang <huangjinhui@kylinos.cn>

In the 64k page configuration of ARM64, the size of THP is 512MB,
which usually reserves almost 5% of memory. However, the probability
of THP usage is not high, especially in the madvise configure. and
THP is not usually used, but a large amount of memory is reserved
for THP use, resulting in a lot of memory waste.

So a dynamic method of adjusting reserved memory is proposed. when
scan and trying to merge into THP, the reserved memory is pulled up,
and when the system does not need to synthesize THP, the reserved
memory is kept low. this method can save a lot of memory in scenarios
without synthesizing THP, and can effectively pull up the reserved
memory to support when THP synthesis is needed. especially in the
case of madvise configure, the running effect will be better.

Signed-off-by: wolfgang huang <huangjinhui@kylinos.cn>
---
 mm/khugepaged.c | 36 +++++++++++++++++++++++++++++++++++-
 1 file changed, 35 insertions(+), 1 deletion(-)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 88433cc25d8a..41a837e618a5 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -129,6 +129,10 @@ static struct khugepaged_scan khugepaged_scan = {
 	.mm_head = LIST_HEAD_INIT(khugepaged_scan.mm_head),
 };
 
+/* khugepaged should scan or not, trying to be thp, default as false */
+static unsigned int khugepaged_thp_scan_state;
+static void set_recommended_min_free_kbytes(void);
+
 #ifdef CONFIG_SYSFS
 static ssize_t scan_sleep_millisecs_show(struct kobject *kobj,
 					 struct kobj_attribute *attr,
@@ -2549,6 +2553,33 @@ static void khugepaged_wait_work(void)
 		wait_event_freezable(khugepaged_wait, khugepaged_wait_event());
 }
 
+static int khugepaged_threshold(void)
+{
+	/* thp size threshold check */
+	if ((PAGE_SIZE << HPAGE_PMD_ORDER) >= SZ_512M)
+		return true;
+	return false;
+}
+
+static void khugepaged_update_wmarks(void)
+{
+	if (!khugepaged_threshold())
+		return;
+
+	/* __khugepaged_enter push khugepaged to work, raise watermark */
+	if (khugepaged_has_work()) {
+		/* Once set, do not repeat distrub watermark */
+		if (!khugepaged_thp_scan_state) {
+			khugepaged_thp_scan_state = true;
+			mutex_lock(&khugepaged_mutex);
+			set_recommended_min_free_kbytes();
+			mutex_unlock(&khugepaged_mutex);
+		}
+	} else {
+		khugepaged_thp_scan_state = false;
+	}
+}
+
 static int khugepaged(void *none)
 {
 	struct khugepaged_mm_slot *mm_slot;
@@ -2559,6 +2590,7 @@ static int khugepaged(void *none)
 	while (!kthread_should_stop()) {
 		khugepaged_do_scan(&khugepaged_collapse_control);
 		khugepaged_wait_work();
+		khugepaged_update_wmarks();
 	}
 
 	spin_lock(&khugepaged_mm_lock);
@@ -2576,7 +2608,9 @@ static void set_recommended_min_free_kbytes(void)
 	int nr_zones = 0;
 	unsigned long recommended_min;
 
-	if (!hugepage_flags_enabled()) {
+	if (!hugepage_flags_enabled() ||
+		(khugepaged_threshold() &&
+		 !khugepaged_thp_scan_state)) {
 		calculate_min_free_kbytes();
 		goto update_wmarks;
 	}
-- 
2.34.1



             reply	other threads:[~2023-09-02  5:26 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-02  5:26 wolfgang huang [this message]
2023-09-05 22:10 ` [PATCH] mm/khugepaged: make reserved memory adaptively Yang Shi

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=20230902052611.144972-1-wolfgang9277@126.com \
    --to=wolfgang9277@126.com \
    --cc=akpm@linux-foundation.org \
    --cc=huangjinhui@kylinos.cn \
    --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 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).