* [PATCH] hugetlbfs: report inode limits without a memory subpool
@ 2026-11-01 12:26 Hungyu Lin
2026-11-01 12:58 ` Aveline Noir
0 siblings, 1 reply; 2+ messages in thread
From: Hungyu Lin @ 2026-11-01 12:26 UTC (permalink / raw)
To: Muchun Song
Cc: Oscar Salvador, David Hildenbrand, Andrew Morton, linux-mm,
linux-fsdevel, linux-kernel, Hungyu Lin
A hugetlbfs mount can limit inodes with nr_inodes independently of size
or min_size. hugetlbfs_statfs() only fills f_files and f_ffree when a
memory subpool exists, so an inode-only limit is enforced but its inode
counts are reported as zero.
For example, a fresh mount with nr_inodes=4 reports zero total and free
inodes instead of 4 and 3. This hides the inode limit from tools such
as df -i, even when creating another file fails with ENOSPC.
Report the inode counters when an inode limit is configured,
independently of the memory subpool. Preserve the existing zero and -1
results for unlimited inode counts.
Fixes: 74a8a65c5261 ("[PATCH] Fix hugetlbfs_statfs() reporting of block limits")
Signed-off-by: Hungyu Lin <dennylin0707@gmail.com>
---
fs/hugetlbfs/inode.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index 01c4d6f44ebe..e94ee8e17004 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -1113,6 +1113,8 @@ static int hugetlbfs_statfs(struct dentry *dentry, struct kstatfs *buf)
- sbinfo->spool->used_hpages;
buf->f_bavail = buf->f_bfree = free_pages;
spin_unlock_irq(&sbinfo->spool->lock);
+ }
+ if (sbinfo->spool || sbinfo->max_inodes >= 0) {
buf->f_files = sbinfo->max_inodes;
buf->f_ffree = sbinfo->free_inodes;
}
--
2.43.0
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] hugetlbfs: report inode limits without a memory subpool
2026-11-01 12:26 [PATCH] hugetlbfs: report inode limits without a memory subpool Hungyu Lin
@ 2026-11-01 12:58 ` Aveline Noir
0 siblings, 0 replies; 2+ messages in thread
From: Aveline Noir @ 2026-11-01 12:58 UTC (permalink / raw)
To: Hungyu Lin
Cc: Andrew Morton, David Hildenbrand, linux-fsdevel, linux-kernel,
linux-mm, Muchun Song, Oscar Salvador
Hi Hungyu,
On Sun, 01 Nov 2026 12:26:53 +0000 Hungyu Lin wrote:
> Report the inode counters when an inode limit is configured,
> independently of the memory subpool. Preserve the existing zero and -1
> results for unlimited inode counts.
Tested in an x86-64 QEMU/KVM guest and all checks passed.
Inode-only counts are fixed, and unlimited counts are unchanged.
Makes sense to me.
Reviewed-by: Aveline Noir <jm5905938@gmail.com>
Tested-by: Aveline Noir <jm5905938@gmail.com>
Thanks,
Aveline Noir
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-11-01 13:08 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-01 12:26 [PATCH] hugetlbfs: report inode limits without a memory subpool Hungyu Lin
2026-11-01 12:58 ` Aveline Noir
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®