Kmemleak is reporting the following leak in ramfs_init_fs_context:

  unreferenced object 0xffff000080302480 (size 8):
    comm "swapper/0", pid 0, jiffies 4294667306
    hex dump (first 8 bytes):
      ed 01 00 00 00 00 00 00                          ........
    backtrace (crc bea23b04):
      __kmalloc_cache_noprof+0x28c/0x5a0
      ramfs_init_fs_context+0x5c/0xf8
      rootfs_init_fs_context+0x30/0x40
      alloc_fs_context+0x4c0/0x6a8
      fs_context_for_mount+0x28/0x40
      vfs_kern_mount+0x40/0x220
      init_mount_tree+0x80/0x440
      mnt_init+0x138/0x170
      vfs_caches_init+0x2c/0x48
      start_kernel+0x350/0x3f8

When init pivot_root()s away from a ramfs-backed boot rootfs and
unmounts it, the rootfs superblock is killed with kill_anon_super() and
the struct ramfs_fs_info allocated by ramfs_init_fs_context() is never
freed.

Kill a ramfs-backed rootfs with ramfs_kill_sb(). A tmpfs-backed one keeps
kill_anon_super(), as shmem frees its info in ->put_super.

Fixes: 576ee5dfd459 ("fs: add immutable rootfs")
Signed-off-by: Breno Leitao <[email protected]>
---
 init/do_mounts.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/init/do_mounts.c b/init/do_mounts.c
index 95e0b3a0f711bd..b088021b1546d8 100644
--- a/init/do_mounts.c
+++ b/init/do_mounts.c
@@ -503,10 +503,18 @@ static int rootfs_init_fs_context(struct fs_context *fc)
        return ramfs_init_fs_context(fc);
 }
 
+static void rootfs_kill_sb(struct super_block *sb)
+{
+       if (IS_ENABLED(CONFIG_TMPFS) && is_tmpfs)
+               kill_anon_super(sb);
+       else
+               ramfs_kill_sb(sb);
+}
+
 struct file_system_type rootfs_fs_type = {
        .name           = "rootfs",
        .init_fs_context = rootfs_init_fs_context,
-       .kill_sb        = kill_anon_super,
+       .kill_sb        = rootfs_kill_sb,
 };
 
 void __init init_rootfs(void)

---
base-commit: d419d682257d813f6cb7a28e42b35d0edf4ad038
change-id: 20261016-rootfs-kill-sb-4f1b597dab24

Best regards,
--  
Breno Leitao <[email protected]>


Reply via email to