mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Jianlin Shi <shijianlin11@foxmail.com>
To: linux-mm@kvack.org
Cc: vbabka@kernel.org, akpm@linux-foundation.org,
	linux-kernel@vger.kernel.org
Subject: [PATCH v2] mm/page_alloc: only update lowmem_reserve_ratio on sysctl write
Date: Fri, 31 Jul 2026 11:42:26 +0800	[thread overview]
Message-ID: <tencent_DB426A4AAD75DA751446F443770939B43205@qq.com> (raw)
In-Reply-To: <tencent_FFD4F4D728AAE8A8AE0AF277A59854A29A06@qq.com>

lowmem_reserve_ratio_sysctl_handler() ignores the return value of
proc_dointvec_minmax() and always calls setup_per_zone_lowmem_reserve(),
even for read operations.

Reading /proc/sys/vm/lowmem_reserve_ratio should not recompute per-zone
lowmem_reserve[] and totalreserve_pages.  Only do so when the sysctl is
written, matching min_free_kbytes and watermark_scale_factor handlers.

Also propagate errors from proc_dointvec_minmax() instead of ignoring
them.

Drop the manual "< 1 -> 0" sanitization loop in the handler and set
.extra1 = SYSCTL_ZERO on the ctl_table entry so proc_dointvec_minmax()
enforces the minimum on write (suggested by Vlastimil Babka).

Compatibility note:
Previously a read also sanitized sysctl_lowmem_reserve_ratio[] and
called setup_per_zone_lowmem_reserve(), which rewrites each zone's
lowmem_reserve[] and recalculates pgdat->totalreserve_pages /
totalreserve_pages (visible via /proc/zoneinfo "protection" and used
by page allocation fallback and dirty-limit accounting).  After this
change only a write does that.  Documentation describes the meaning of
the ratio and the derived protection pages, but does not document any
read side-effect.

Worst case for odd userspace that treated a read as a refresh of those
derived values: lowmem_reserve[] and totalreserve_pages remain at their
last written/setup values until the next write of this sysctl, or until
another existing updater runs (e.g. adjust_managed_page_count() on
managed-page changes, or init/watermark setup paths).  Until then,
allocation fallback into lower zones and per-node dirtyable memory
(node_dirtyable_memory() subtracts pgdat->totalreserve_pages) may not
reflect a refresh that such userspace expected from the read alone.
Normal readers that only consume the ratio array are unaffected.

Changes in v2:
- Add .extra1 = SYSCTL_ZERO to the ctl_table entry
- Remove the manual sanitization loop (negative writes now return
  -EINVAL instead of being silently coerced to 0)

Link: https://lore.kernel.org/linux-mm/tencent_FFD4F4D728AAE8A8AE0AF277A59854A29A06@qq.com/

Signed-off-by: Jianlin Shi <shijianlin11@foxmail.com>
---
 mm/page_alloc.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 0387d2afd..a7381327d 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -6683,16 +6683,15 @@ static int sysctl_min_slab_ratio_sysctl_handler(const struct ctl_table *table, i
 static int lowmem_reserve_ratio_sysctl_handler(const struct ctl_table *table,
 		int write, void *buffer, size_t *length, loff_t *ppos)
 {
-	int i;
+	int rc;
 
-	proc_dointvec_minmax(table, write, buffer, length, ppos);
+	rc = proc_dointvec_minmax(table, write, buffer, length, ppos);
+	if (rc)
+		return rc;
 
-	for (i = 0; i < MAX_NR_ZONES; i++) {
-		if (sysctl_lowmem_reserve_ratio[i] < 1)
-			sysctl_lowmem_reserve_ratio[i] = 0;
-	}
+	if (write)
+		setup_per_zone_lowmem_reserve();
 
-	setup_per_zone_lowmem_reserve();
 	return 0;
 }
 
@@ -6791,6 +6790,7 @@ static const struct ctl_table page_alloc_sysctl_table[] = {
 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
 		.mode		= 0644,
 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
+		.extra1		= SYSCTL_ZERO,
 	},
 #ifdef CONFIG_NUMA
 	{
-- 
2.43.0


  parent reply	other threads:[~2026-07-31  3:42 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-28 10:32 [PATCH] " Jianlin Shi
2026-07-30 15:28 ` Vlastimil Babka (SUSE)
2026-07-31  3:36   ` Jianlin Shi
2026-07-31  3:42 ` Jianlin Shi [this message]
2026-07-31  8:06   ` [PATCH v2] " Vlastimil Babka (SUSE)
2026-08-01  7:36   ` Johannes Weiner
2026-08-01 15:11 ` [PATCH v3] " Jianlin Shi
2026-08-01 18:44   ` Andrew Morton
2026-08-02 14:31     ` Jianlin Shi
2026-08-03  8:27     ` Vlastimil Babka (SUSE)
2026-08-04 12:20       ` Joel Granados
2026-08-04 14:33         ` Vlastimil Babka (SUSE)
2026-08-06  8:27         ` Joel Granados
2026-08-02 14:57 ` [PATCH v4] " Jianlin Shi
2026-08-04  0:48   ` Andrew Morton
2026-08-06  7:32     ` Jianlin Shi
2026-08-06  8:27 ` [PATCH v5] " Jianlin Shi
2026-08-06 13:50   ` Johannes Weiner
2026-08-06 21:20     ` Andrew Morton
2026-08-07  9:45       ` Joel Granados
2026-08-07 10:05   ` Vlastimil Babka (SUSE)

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=tencent_DB426A4AAD75DA751446F443770939B43205@qq.com \
    --to=shijianlin11@foxmail.com \
    --cc=akpm@linux-foundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=vbabka@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®