[PATCH] mm/balloon: fix NR_BALLOON_PAGES accounting on migration failure

[PATCH] mm/balloon: fix NR_BALLOON_PAGES accounting on migration failure

[shao.mingyin]

From: Shao Mingyin <shao.mingyin@zte.com.cn>

mm/balloon: fix NR_BALLOON_PAGES accounting on migration failure

A balloon driver's -ENOENT migration result means that the old page was
deflated but the replacement was not inflated. vmw_balloon returns this
when its deflate operation succeeds and the subsequent inflate fails.

balloon_page_migrate() accounts the deflation and frees the old page,
but does not decrement NR_BALLOON_PAGES. Each such failure therefore
leaves one extra page in the per-node count, which also contributes to
the Balloon value in /proc/meminfo.

Decrement the counter on the old page's node, matching normal dequeue
accounting.

Fixes: 4d689474e1b2 ("balloon_compaction: update the NR_BALLOON_PAGES state")
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
---
Build-tested mm/balloon.o with W=1 and CONFIG_BALLOON_MIGRATION=y on
x86_64. The vmw_balloon -ENOENT callback path was reviewed statically.
No VMware hardware or hypervisor runtime test was performed.


 mm/balloon.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/mm/balloon.c b/mm/balloon.c
index 96a8f1e20bc66ae46a571a4f51243789bb362925..44c211ba8e6b0b86236ec0e05ca1eb4b9c444a8e 100644
--- a/mm/balloon.c
+++ b/mm/balloon.c
@@ -313,6 +313,7 @@ static int balloon_page_migrate(struct page *newpage, struct page *page,
 	} else {
 		/* Old page was deflated but new page not inflated. */
 		__count_vm_event(BALLOON_DEFLATE);
+		dec_node_page_state(page, NR_BALLOON_PAGES);
 
 		if (b_dev_info->adjust_managed_page_count)
 			adjust_managed_page_count(page, 1);

base-commit: 84abb955b1015477aa11a96e2f3ac1b938b77f48
-- 
2.43.0

Re: [PATCH] mm/balloon: fix NR_BALLOON_PAGES accounting on migration failure

[Lorenzo Stoakes (ARM)]

On Sun, Nov 15, 2026 at 12:47:08AM +0800, shao.mingyin@zte.com.cn wrote:
> From: Shao Mingyin <shao.mingyin@zte.com.cn>
>
> mm/balloon: fix NR_BALLOON_PAGES accounting on migration failure
>
> A balloon driver's -ENOENT migration result means that the old page was
> deflated but the replacement was not inflated. vmw_balloon returns this
> when its deflate operation succeeds and the subsequent inflate fails.
>
> balloon_page_migrate() accounts the deflation and frees the old page,
> but does not decrement NR_BALLOON_PAGES. Each such failure therefore
> leaves one extra page in the per-node count, which also contributes to
> the Balloon value in /proc/meminfo.
>
> Decrement the counter on the old page's node, matching normal dequeue
> accounting.
>
> Fixes: 4d689474e1b2 ("balloon_compaction: update the NR_BALLOON_PAGES state")
> Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>

NAK.

this is generated by some AI slopper's script, spamming the kernel. The
patches I looked at were wrong/broken, so no to anything from this person.

They've sent 300 mails in under four weeks spanniing: ata, scsi, target,
ufs, nvme/nvmet, ubifs, ublk, zloop, dm, md, mmc, virtio-blk, mfd, gpio,
pwm, spi, PCI, ASoC, ACPI, net, vfs mount, block partitions, and now mm.

and 17 series across 7 subsystems over the past 23 hours.

We need a way to ban emails from people like this...

--
Cheers, Lorenzo