linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Yajun Deng <yajun.deng@linux.dev>
To: akpm@linux-foundation.org, mike.kravetz@oracle.com,
	muchun.song@linux.dev, glider@google.com, elver@google.com,
	dvyukov@google.com, rppt@kernel.org, david@redhat.com,
	osalvador@suse.de
Cc: linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	kasan-dev@googlegroups.com, Yajun Deng <yajun.deng@linux.dev>
Subject: [PATCH 3/4] mm: Set page count and mark page reserved in reserve_bootmem_region
Date: Fri, 22 Sep 2023 15:09:22 +0800	[thread overview]
Message-ID: <20230922070923.355656-4-yajun.deng@linux.dev> (raw)
In-Reply-To: <20230922070923.355656-1-yajun.deng@linux.dev>

memmap_init_range() would set page count of all pages, but the free
pages count would be reset in __free_pages_core(). These two are
opposite operations. It's unnecessary and time-consuming when it's
in MEMINIT_EARLY context.

Set page count and mark page reserved in reserve_bootmem_region when
in MEMINIT_EARLY context, and change the context from MEMINIT_LATE
to MEMINIT_EARLY in __free_pages_memory.

At the same time, the init list head in reserve_bootmem_region isn't
need. As it already done in __init_single_page.

The following data was tested on an x86 machine with 190GB of RAM.

before:
free_low_memory_core_early()	342ms

after:
free_low_memory_core_early()	286ms

Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
---
 mm/memblock.c   |  2 +-
 mm/mm_init.c    | 20 ++++++++++++++------
 mm/page_alloc.c |  8 +++++---
 3 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/mm/memblock.c b/mm/memblock.c
index a32364366bb2..9276f1819982 100644
--- a/mm/memblock.c
+++ b/mm/memblock.c
@@ -2089,7 +2089,7 @@ static void __init __free_pages_memory(unsigned long start, unsigned long end)
 		while (start + (1UL << order) > end)
 			order--;
 
-		memblock_free_pages(start, order, MEMINIT_LATE);
+		memblock_free_pages(start, order, MEMINIT_EARLY);
 
 		start += (1UL << order);
 	}
diff --git a/mm/mm_init.c b/mm/mm_init.c
index 0a4437aae30d..1cc310f706a9 100644
--- a/mm/mm_init.c
+++ b/mm/mm_init.c
@@ -718,7 +718,7 @@ static void __meminit init_reserved_page(unsigned long pfn, int nid)
 		if (zone_spans_pfn(zone, pfn))
 			break;
 	}
-	__init_single_page(pfn_to_page(pfn), pfn, zid, nid, true, false);
+	__init_single_page(pfn_to_page(pfn), pfn, zid, nid, false, false);
 }
 #else
 static inline void pgdat_set_deferred_range(pg_data_t *pgdat) {}
@@ -756,8 +756,8 @@ void __meminit reserve_bootmem_region(phys_addr_t start,
 
 			init_reserved_page(start_pfn, nid);
 
-			/* Avoid false-positive PageTail() */
-			INIT_LIST_HEAD(&page->lru);
+			/* Set page count for the reserve region */
+			init_page_count(page);
 
 			/*
 			 * no need for atomic set_bit because the struct
@@ -888,9 +888,17 @@ void __meminit memmap_init_range(unsigned long size, int nid, unsigned long zone
 		}
 
 		page = pfn_to_page(pfn);
-		__init_single_page(page, pfn, zone, nid, true, false);
-		if (context == MEMINIT_HOTPLUG)
-			__SetPageReserved(page);
+
+		/* If the context is MEMINIT_EARLY, we will set page count and
+		 * mark page reserved in reserve_bootmem_region, the free region
+		 * wouldn't have page count and reserved flag and we don't
+		 * need to reset pages count and clear reserved flag in
+		 * __free_pages_core.
+		 */
+		if (context == MEMINIT_EARLY)
+			__init_single_page(page, pfn, zone, nid, false, false);
+		else
+			__init_single_page(page, pfn, zone, nid, true, true);
 
 		/*
 		 * Usually, we want to mark the pageblock MIGRATE_MOVABLE,
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 6c4f4531bee0..6ac58c5f3b00 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1285,9 +1285,11 @@ void __free_pages_core(struct page *page, unsigned int order, enum meminit_conte
 	unsigned int loop;
 
 	/*
-	 * When initializing the memmap, __init_single_page() sets the refcount
-	 * of all pages to 1 ("allocated"/"not free"). We have to set the
-	 * refcount of all involved pages to 0.
+	 * When initializing the memmap, memmap_init_range sets the refcount
+	 * of all pages to 1 ("allocated"/"not free") in hotplug context. We
+	 * have to set the refcount of all involved pages to 0. Otherwise,
+	 * we don't do it, as reserve_bootmem_region only set the refcount on
+	 * reserve region ("allocated") in early context.
 	 */
 	if (context != MEMINIT_EARLY) {
 		prefetchw(p);
-- 
2.25.1



  parent reply	other threads:[~2023-09-22  7:10 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-22  7:09 [PATCH 0/4] mm: Don't set and reset page count in MEMINIT_EARLY Yajun Deng
2023-09-22  7:09 ` [PATCH 1/4] mm: pass set_count and set_reserved to __init_single_page Yajun Deng
2023-09-22  7:47   ` Matthew Wilcox
2023-09-22  7:48     ` David Hildenbrand
2023-09-22  8:08       ` Mike Rapoport
2023-09-22  7:09 ` [PATCH 2/4] mm: Introduce MEMINIT_LATE context Yajun Deng
2023-09-22  7:09 ` Yajun Deng [this message]
2023-09-22  7:09 ` [PATCH 4/4] mm: don't set page count in deferred_init_pages Yajun Deng

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=20230922070923.355656-4-yajun.deng@linux.dev \
    --to=yajun.deng@linux.dev \
    --cc=akpm@linux-foundation.org \
    --cc=david@redhat.com \
    --cc=dvyukov@google.com \
    --cc=elver@google.com \
    --cc=glider@google.com \
    --cc=kasan-dev@googlegroups.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mike.kravetz@oracle.com \
    --cc=muchun.song@linux.dev \
    --cc=osalvador@suse.de \
    --cc=rppt@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