[PATCH] proc: only bump parent nlink when registering directories

[Krzysztof Wilczyński]

proc_register() increments the parent directory's link count for every
entry it registers, while remove_proc_entry() and remove_proc_subtree()
decrement it only when the removed entry is a directory.  Regular files
thus inflate the parent's count while they exist, and leak one link
permanently on every create and remove cycle.

For example, /proc/bus/pci/00 with twenty-two device files and no
subdirectories reports nlink 24 instead of 2, and SR-IOV VF enable
and disable cycles, each creating and removing the VF config space
entries under /proc/bus/pci/<bus>, inflate the link count of that
directory without bound.

Before commit e06689bf5701 ("proc: change ->nlink under
proc_subdir_lock"), the increment lived in proc_mkdir_data() and
proc_create_mount_point(), and was therefore applied only to
directories.  Moving it into proc_register() to bring it under
proc_subdir_lock dropped the S_ISDIR check.

Thus, increment the parent link count only for directories, matching
the decrements in remove_proc_entry() and remove_proc_subtree().

Fixes: e06689bf5701 ("proc: change ->nlink under proc_subdir_lock")
Cc: stable@xxxxxxxxxxxxxxx # v5.5+
Signed-off-by: Krzysztof Wilczyński <kwilczynski@xxxxxxxxxx>
---
 fs/proc/generic.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/proc/generic.c b/fs/proc/generic.c
index 8bb81e58c9d8..3ac825c79313 100644
--- a/fs/proc/generic.c
+++ b/fs/proc/generic.c
@@ -404,7 +404,8 @@ struct proc_dir_entry *proc_register(struct proc_dir_entry *dir,
 		write_unlock(&proc_subdir_lock);
 		goto out_free_inum;
 	}
-	dir->nlink++;
+	if (S_ISDIR(dp->mode))
+		dir->nlink++;
 	write_unlock(&proc_subdir_lock);
 
 	return dp;
-- 
2.54.0