Linux Security Modules development
 help / color / mirror / Atom feed
From: Alexander Mikhalitsyn <alex@amutable.com>
To: Christian Brauner <brauner@kernel.org>, linux-fsdevel@vger.kernel.org
Cc: linux-security-module@vger.kernel.org,
	Alexander Viro	 <viro@zeniv.linux.org.uk>,
	Jan Kara <jack@suse.cz>,
	bpf@vger.kernel.org, Zhiqi Zhang <zhiqizhangg@gmail.com>
Subject: Re: [PATCH 2/2] ns: don't call the namespace LSM hooks for kernel-internal namespaces
Date: Fri, 20 Nov 2026 15:03:07 +0100	[thread overview]
Message-ID: <ecb63d95f150bd91bbd8a3c83211055461bc9594.camel@amutable.com> (raw)
In-Reply-To: <20261120-work-ns-mount_subtree-v1-2-06337377f94a@kernel.org>

On Fri, 2026-11-20 at 13:29 +0100, Christian Brauner wrote:
> mount_subtree() allocates a private anonymous mount namespace so that
> automounts triggered while it looks up the path have somewhere to go.
> That namespace never becomes anyone's namespace and it's gone again
> before mount_subtree() returns. btrfs goes through it on every mount:
> 
>   btrfs_get_tree()
>   -> btrfs_get_tree_subvol()
>      -> mount_subvol()
>         -> mount_subtree()
> 
> and NFS on every NFSv4 mount and whenever it follows a referral:
> 
>   nfs4_try_get_tree() / nfs4_get_referral_tree()
>   -> do_nfs4_mount()
>      -> mount_subtree()
> 
> But __ns_common_init() calls security_namespace_init() for it like
> for
> any other namespace. Since commit 15dd86021b57a ("landlock: Enforce
> namespace use restrictions") Landlock enforces
> LANDLOCK_PERMISSION_NAMESPACE_USE in its namespace_init hook. So a
> task
> in a domain that handles that permission without allowing CLONE_NEWNS
> can't mount btrfs or NFSv4 anymore. mount(2) fails with EPERM even
> though the caller never asked for a mount namespace.
> 
> Mark such namespaces as kernel-internal and don't call the namespace
> LSM hooks for them at all. That includes namespace_free. Currently
> security_namespace_free() only calls the hook if an LSM requested a
> namespace blob and none does. But we shouldn't rely on that.
> Otherwise
> a bpf program attached to namespace_free would see namespaces go away
> that it never saw being created.
> 
> The anonymous mount namespaces created by open_tree(OPEN_TREE_CLONE)
> and fsmount() are handed to userspace and keep going through the
> hooks.
> Landlock covers them on purpose.
> 
> Fixes: f675d2e95569 ("lsm: add LSM blob and hooks for namespaces")
> Reported-by: Zhiqi Zhang <zhiqizhangg@gmail.com>
> Closes:
> https://lore.kernel.org/20261114141323.19029-1-zhiqizhangg@gmail.com
> Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>

Looks good to me.

Reviewed-by: Alexander Mikhalitsyn <alex@amutable.com>

> ---
>  fs/namespace.c                     |  9 +++++++--
>  include/linux/ns/ns_common_types.h |  2 ++
>  include/linux/ns_common.h          | 20 +++++++++++++-------
>  kernel/nscommon.c                  |  9 ++++++---
>  4 files changed, 28 insertions(+), 12 deletions(-)
> 
> diff --git a/fs/namespace.c b/fs/namespace.c
> index 756f9011cbf1d..82738b818c1d0 100644
> --- a/fs/namespace.c
> +++ b/fs/namespace.c
> @@ -2355,6 +2355,7 @@ void drop_collected_paths(const struct path
> *paths, const struct path *prealloc)
>  
>  enum mnt_ns_flags_t {
>  	MNT_NS_ANON	= (1 << 0),
> +	MNT_NS_KERNEL	= (1 << 1),
>  };
>  
>  static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *,
> @@ -4377,10 +4378,14 @@ static void free_mnt_ns(struct mnt_namespace
> *ns)
>  static struct mnt_namespace *alloc_mnt_ns(struct user_namespace
> *user_ns,
>  					  enum mnt_ns_flags_t flags)
>  {
> +	enum ns_common_flags_t ns_flags = 0;
>  	struct mnt_namespace *new_ns;
>  	struct ucounts *ucounts;
>  	int ret;
>  
> +	if (flags & MNT_NS_KERNEL)
> +		ns_flags |= NS_COMMON_KERNEL;
> +
>  	ucounts = inc_mnt_namespaces(user_ns);
>  	if (!ucounts)
>  		return ERR_PTR(-ENOSPC);
> @@ -4392,7 +4397,7 @@ static struct mnt_namespace
> *alloc_mnt_ns(struct user_namespace *user_ns,
>  	}
>  
>  	if (flags & MNT_NS_ANON)
> -		ret = ns_common_init_inum(new_ns, MNT_NS_ANON_INO);
> +		ret = ns_common_init_inum(new_ns, MNT_NS_ANON_INO,
> ns_flags);
>  	else
>  		ret = ns_common_init(new_ns);
>  	if (ret) {
> @@ -4515,7 +4520,7 @@ struct dentry *mount_subtree(struct vfsmount
> *m, const char *name)
>  	struct path path;
>  	int err;
>  
> -	ns = alloc_mnt_ns(&init_user_ns, MNT_NS_ANON);
> +	ns = alloc_mnt_ns(&init_user_ns, MNT_NS_ANON |
> MNT_NS_KERNEL);
>  	if (IS_ERR(ns)) {
>  		mntput(m);
>  		return ERR_CAST(ns);
> diff --git a/include/linux/ns/ns_common_types.h
> b/include/linux/ns/ns_common_types.h
> index 47a5cfe2326a3..5f8b0930ae065 100644
> --- a/include/linux/ns/ns_common_types.h
> +++ b/include/linux/ns/ns_common_types.h
> @@ -102,6 +102,7 @@ extern const struct proc_ns_operations
> utsns_operations;
>   *   __ns_ref_active = 1 and remain active forever.
>   *
>   * @ns_type: type of namespace (e.g., CLONE_NEWNET)
> + * @ns_flags: NS_COMMON_* flags
>   * @stashed: cached dentry to be used by the vfs
>   * @ops: namespace operations
>   * @inum: namespace inode number (quickly recycled for non-initial
> namespaces)
> @@ -113,6 +114,7 @@ struct ns_common {
>  		refcount_t __ns_ref; /* do not use directly */
>  	} ____cacheline_aligned_in_smp;
>  	u32 ns_type;
> +	u32 ns_flags;
>  	struct dentry *stashed;
>  	const struct proc_ns_operations *ops;
>  	unsigned int inum;
> diff --git a/include/linux/ns_common.h b/include/linux/ns_common.h
> index e8b2bf0da5ede..524b77904d3b4 100644
> --- a/include/linux/ns_common.h
> +++ b/include/linux/ns_common.h
> @@ -8,9 +8,14 @@
>  #include <uapi/linux/sched.h>
>  #include <uapi/linux/nsfs.h>
>  
> +enum ns_common_flags_t {
> +	/* Kernel-internal namespace. The LSM namespace hooks never
> see it. */
> +	NS_COMMON_KERNEL	= (1 << 0),
> +};
> +
>  bool is_current_namespace(struct ns_common *ns);
>  int __ns_common_init(struct ns_common *ns, u32 ns_type, const struct
> proc_ns_operations *ops,
> -		     int inum, u64 ns_id);
> +		     int inum, u64 ns_id, enum ns_common_flags_t
> flags);
>  void __ns_common_free(struct ns_common *ns);
>  struct ns_common *__must_check ns_owner(struct ns_common *ns);
>  
> @@ -47,13 +52,14 @@ static __always_inline bool is_ns_init_id(const
> struct ns_common *ns)
>  			
> ns_common_type(__ns),				\
>  			
> to_ns_operations(__ns),			\
>  			 (((__ns) == ns_init_ns(__ns)) ?
> ns_init_inum(__ns) : 0), \
> -			 (((__ns) == ns_init_ns(__ns)) ?
> ns_init_id(__ns) : 0))
> +			 (((__ns) == ns_init_ns(__ns)) ?
> ns_init_id(__ns) : 0), \
> +			 0)
>  
> -#define ns_common_init_inum(__ns, __inum)        \
> -	__ns_common_init(to_ns_common(__ns),     \
> -			 ns_common_type(__ns),   \
> -			 to_ns_operations(__ns), \
> -			 __inum, 0)
> +#define ns_common_init_inum(__ns, __inum, __flags) \
> +	__ns_common_init(to_ns_common(__ns),       \
> +			 ns_common_type(__ns),     \
> +			 to_ns_operations(__ns),   \
> +			 __inum, 0, __flags)
>  
>  #define ns_common_free(__ns) __ns_common_free(to_ns_common((__ns)))
>  
> diff --git a/kernel/nscommon.c b/kernel/nscommon.c
> index 4438c6b663f91..20fc0d0e98212 100644
> --- a/kernel/nscommon.c
> +++ b/kernel/nscommon.c
> @@ -55,7 +55,7 @@ static void ns_debug(struct ns_common *ns, const
> struct proc_ns_operations *ops)
>  #endif
>  
>  int __ns_common_init(struct ns_common *ns, u32 ns_type, const struct
> proc_ns_operations *ops,
> -		     int inum, u64 ns_id)
> +		     int inum, u64 ns_id, enum ns_common_flags_t
> flags)
>  {
>  	int ret = 0;
>  
> @@ -66,6 +66,7 @@ int __ns_common_init(struct ns_common *ns, u32
> ns_type, const struct proc_ns_ope
>  #endif
>  	ns->ops = ops;
>  	ns->ns_type = ns_type;
> +	ns->ns_flags = flags;
>  	ns_tree_node_init(&ns->ns_tree_node);
>  	ns_tree_node_init(&ns->ns_unified_node);
>  	ns_tree_node_init(&ns->ns_owner_node);
> @@ -85,7 +86,8 @@ int __ns_common_init(struct ns_common *ns, u32
> ns_type, const struct proc_ns_ope
>  	/* Allocate the id before the LSM hook so it can track the
> namespace by it. */
>  	__ns_tree_gen_id(ns, ns_id);
>  
> -	ret = security_namespace_init(ns);
> +	if (!(flags & NS_COMMON_KERNEL))
> +		ret = security_namespace_init(ns);
>  	if (ret) {
>  		if (!inum)
>  			proc_free_inum(ns->inum);
> @@ -106,7 +108,8 @@ int __ns_common_init(struct ns_common *ns, u32
> ns_type, const struct proc_ns_ope
>  
>  void __ns_common_free(struct ns_common *ns)
>  {
> -	security_namespace_free(ns);
> +	if (!(ns->ns_flags & NS_COMMON_KERNEL))
> +		security_namespace_free(ns);
>  
>  	if (ns->inum > MNT_NS_INO_SPECIAL_MAX)
>  		proc_free_inum(ns->inum);

      parent reply	other threads:[~2026-11-20 14:03 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-20 12:29 [PATCH 0/2] ns: don't call the namespace LSM hooks for kernel-internal namespaces Christian Brauner
2026-11-20 12:29 ` [PATCH 1/2] mount: pass flags to alloc_mnt_ns() Christian Brauner
2026-11-20 12:32   ` sashiko-bot
2026-11-20 14:02   ` Alexander Mikhalitsyn
2026-11-20 12:29 ` [PATCH 2/2] ns: don't call the namespace LSM hooks for kernel-internal namespaces Christian Brauner
2026-11-20 12:33   ` sashiko-bot
2026-11-20 14:03   ` Alexander Mikhalitsyn [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ecb63d95f150bd91bbd8a3c83211055461bc9594.camel@amutable.com \
    --to=alex@amutable.com \
    --cc=bpf@vger.kernel.org \
    --cc=brauner@kernel.org \
    --cc=jack@suse.cz \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-security-module@vger.kernel.org \
    --cc=viro@zeniv.linux.org.uk \
    --cc=zhiqizhangg@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox