[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YByxkDi0Ruhb0AA8@kernel.org>
Date: Fri, 5 Feb 2021 04:46:40 +0200
From: Jarkko Sakkinen <jarkko@...nel.org>
To: David Howells <dhowells@...hat.com>
Cc: sprabhu@...hat.com, christian@...uner.io, selinux@...r.kernel.org,
keyrings@...r.kernel.org, linux-api@...r.kernel.org,
linux-fsdevel@...r.kernel.org,
linux-security-module@...r.kernel.org,
linux-kernel@...r.kernel.org, containers@...ts.linux-foundation.org
Subject: Re: [PATCH 1/2] Add namespace tags that can be used for matching
without pinning a ns
On Thu, Feb 04, 2021 at 05:47:39PM +0000, David Howells wrote:
> Add a ns tag struct that consists of just a refcount. It's address can be
> used to compare namespaces without the need to pin a namespace. Just the
> tag needs pinning.
>
> Signed-off-by: David Howells <dhowells@...hat.com>
> ---
>
> fs/namespace.c | 18 ++++++++----------
> include/linux/ns_common.h | 23 +++++++++++++++++++++++
> include/linux/proc_ns.h | 38 +++++++++++++++++++++++++++++++++++---
> init/version.c | 9 ++++++++-
> ipc/msgutil.c | 7 ++++++-
> ipc/namespace.c | 8 +++-----
> kernel/cgroup/cgroup.c | 5 +++++
> kernel/cgroup/namespace.c | 6 +++---
> kernel/pid.c | 5 +++++
> kernel/pid_namespace.c | 18 +++++++++---------
> kernel/time/namespace.c | 13 +++++--------
> kernel/user.c | 5 +++++
> kernel/user_namespace.c | 7 +++----
> kernel/utsname.c | 24 +++++++++++++-----------
> net/core/net_namespace.c | 38 +++++++++++++++-----------------------
> 15 files changed, 146 insertions(+), 78 deletions(-)
>
> diff --git a/fs/namespace.c b/fs/namespace.c
> index 9d33909d0f9e..f8da9be8c6f7 100644
> --- a/fs/namespace.c
> +++ b/fs/namespace.c
> @@ -3238,10 +3238,9 @@ static void dec_mnt_namespaces(struct ucounts *ucounts)
>
> static void free_mnt_ns(struct mnt_namespace *ns)
> {
> - if (!is_anon_ns(ns))
> - ns_free_inum(&ns->ns);
> dec_mnt_namespaces(ns->ucounts);
> put_user_ns(ns->user_ns);
> + destroy_ns_common(&ns->ns);
> kfree(ns);
> }
>
> @@ -3269,18 +3268,17 @@ static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *user_ns, bool a
> dec_mnt_namespaces(ucounts);
> return ERR_PTR(-ENOMEM);
> }
> - if (!anon) {
> - ret = ns_alloc_inum(&new_ns->ns);
> - if (ret) {
> - kfree(new_ns);
> - dec_mnt_namespaces(ucounts);
> - return ERR_PTR(ret);
> - }
> +
> + ret = init_ns_common(&new_ns->ns, anon);
> + if (ret) {
> + destroy_ns_common(&new_ns->ns);
> + kfree(new_ns);
> + dec_mnt_namespaces(ucounts);
> + return ERR_PTR(ret);
> }
> new_ns->ns.ops = &mntns_operations;
> if (!anon)
> new_ns->seq = atomic64_add_return(1, &mnt_ns_seq);
> - refcount_set(&new_ns->ns.count, 1);
> INIT_LIST_HEAD(&new_ns->list);
> init_waitqueue_head(&new_ns->poll);
> spin_lock_init(&new_ns->ns_lock);
> diff --git a/include/linux/ns_common.h b/include/linux/ns_common.h
> index 0f1d024bd958..45174ad8a435 100644
> --- a/include/linux/ns_common.h
> +++ b/include/linux/ns_common.h
> @@ -3,14 +3,37 @@
> #define _LINUX_NS_COMMON_H
>
> #include <linux/refcount.h>
> +#include <linux/slab.h>
>
> struct proc_ns_operations;
>
> +/*
> + * Comparable tag for namespaces so that namespaces don't have to be pinned by
> + * something that wishes to detect if a namespace matches a criterion.
> + */
> +struct ns_tag {
> + refcount_t usage;
Is that indentation necessary? I'd put just a space.
> +};
> +
> struct ns_common {
> atomic_long_t stashed;
> const struct proc_ns_operations *ops;
> + struct ns_tag *tag;
> unsigned int inum;
> refcount_t count;
> };
>
> +static inline struct ns_tag *get_ns_tag(struct ns_tag *tag)
> +{
> + if (tag)
> + refcount_inc(&tag->usage);
> + return tag;
> +}
> +
> +static inline void put_ns_tag(struct ns_tag *tag)
> +{
> + if (tag && refcount_dec_and_test(&tag->usage))
> + kfree(tag);
> +}
> +
> #endif
> diff --git a/include/linux/proc_ns.h b/include/linux/proc_ns.h
> index 75807ecef880..9fb7eb403923 100644
> --- a/include/linux/proc_ns.h
> +++ b/include/linux/proc_ns.h
> @@ -64,13 +64,45 @@ static inline void proc_free_inum(unsigned int inum) {}
>
> #endif /* CONFIG_PROC_FS */
>
> -static inline int ns_alloc_inum(struct ns_common *ns)
> +/**
> + * init_ns_common - Initialise the common part of a namespace
Nit: init_ns_common()
> + * @ns: The namespace to initialise
> + * @anon: The namespace will be anonymous
> + *
> + * Set up the common part of a namespace, assigning an inode number and
> + * creating a tag. Returns 0 on success and a negative error code on failure.
> + * On failure, the caller must call destroy_ns_common().
I've used lately (e.g. arch/x86/kernel/cpu/sgx/ioctl.c) along the lines:
* Return:
* - 0: Initialization was successful.
* - -ENOMEM: Out of memory.
Looking at the implementation, I guess this is a complete representation of
what it can return?
The driving point here is that this nicely lines up when rendered with
"make htmldocs".
> + */
> +static inline int init_ns_common(struct ns_common *ns, bool anon)
> {
> + struct ns_tag *tag;
> +
> + tag = kzalloc(sizeof(*tag), GFP_KERNEL);
> + if (!tag)
> + return -ENOMEM;
> +
> + refcount_set(&tag->usage, 1);
> + ns->tag = tag;
> + ns->inum = 0;
> atomic_long_set(&ns->stashed, 0);
> - return proc_alloc_inum(&ns->inum);
> + refcount_set(&ns->count, 1);
> +
> + return anon ? 0 : proc_alloc_inum(&ns->inum);
> }
>
> -#define ns_free_inum(ns) proc_free_inum((ns)->inum)
> +/**
> + * destroy_ns_common - Clean up the common part of a namespace
Nit: destroy_ns_common()
/Jarkko
Powered by blists - more mailing lists