Re: [PATCH 1/4] namespace: take lock_mount_hash() directly when changing flags

[Jann Horn <jannh@google.com>]

On Tue, Jul 14, 2020 at 6:16 PM Christian Brauner
<christian.brauner@ubuntu.com> wrote:
> Changing mount options always ends up taking lock_mount_hash() but when
> MNT_READONLY is requested and neither the mount nor the superblock are
> not already MNT_READONLY we end up taking the lock, dropping it, and
> retaking it to change the other mount attributes. Instead of this,
> acquire the lock once when changing mount properties. This simplifies
> the locking in these codepath, makes them easier to reason about and
> avoids having to reacquire the lock right after dropping it.
[...]
> diff --git a/fs/namespace.c b/fs/namespace.c
[...]
> @@ -463,7 +463,6 @@ static int mnt_make_readonly(struct mount *mnt)
>  {
>         int ret = 0;
>
> -       lock_mount_hash();
>         mnt->mnt.mnt_flags |= MNT_WRITE_HOLD;
>         /*
>          * After storing MNT_WRITE_HOLD, we'll read the counters. This store
> @@ -497,15 +496,12 @@ static int mnt_make_readonly(struct mount *mnt)
>          */
>         smp_wmb();
>         mnt->mnt.mnt_flags &= ~MNT_WRITE_HOLD;
> -       unlock_mount_hash();
>         return ret;
>  }

It might be a good idea, instead of completely removing the locking
calls here, to replace them with lockdep_assert_held(...).
(Currently that doesn't appear much in core VFS code though.)