Re: [PATCH v1 2/3] fs/mnt_idmapping: introduce an invalid_mnt_idmap

[Christian Brauner]

On Fri, Sep 06, 2024 at 04:34:52PM GMT, Alexander Mikhalitsyn wrote:
> Link: https://lore.kernel.org/linux-fsdevel/20240904-baugrube-erhoben-b3c1c49a2645@brauner/
> Suggested-by: Christian Brauner <brauner@xxxxxxxxxx>
> Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@xxxxxxxxxxxxx>
> ---
>  fs/mnt_idmapping.c            | 22 ++++++++++++++++++++--
>  include/linux/mnt_idmapping.h |  1 +
>  2 files changed, 21 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/mnt_idmapping.c b/fs/mnt_idmapping.c
> index 3c60f1eaca61..cbca6500848e 100644
> --- a/fs/mnt_idmapping.c
> +++ b/fs/mnt_idmapping.c
> @@ -32,6 +32,15 @@ struct mnt_idmap nop_mnt_idmap = {
>  };
>  EXPORT_SYMBOL_GPL(nop_mnt_idmap);
>  
> +/*
> + * Carries the invalid idmapping of a full 0-4294967295 {g,u}id range.
> + * This means that all {g,u}ids are mapped to INVALID_VFS{G,U}ID.
> + */
> +struct mnt_idmap invalid_mnt_idmap = {
> +	.count	= REFCOUNT_INIT(1),
> +};
> +EXPORT_SYMBOL_GPL(invalid_mnt_idmap);
> +
>  /**
>   * initial_idmapping - check whether this is the initial mapping
>   * @ns: idmapping to check
> @@ -75,6 +84,8 @@ vfsuid_t make_vfsuid(struct mnt_idmap *idmap,
>  
>  	if (idmap == &nop_mnt_idmap)
>  		return VFSUIDT_INIT(kuid);
> +	if (idmap == &invalid_mnt_idmap)
> +		return INVALID_VFSUID;

Could possibly deserve an:

if (unlikely(idmap == &invalid_mnt_idmap))
	return INVALID_VFSUID;

and technically I guess we could also do:

if (likely(idmap == &nop_mnt_idmap))
	return VFSUIDT_INIT(kuid);

but not that relevant for this patch.