From 0b40774ef06cee7c3334dcee068f06d331ff2749 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:31 +0200 Subject: cgroup: use ns_common_init() Don't cargo-cult the same thing over and over. Acked-by: Tejun Heo Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/namespace.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index 144a464e45c6..0391b6ab0bf1 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -21,20 +21,16 @@ static void dec_cgroup_namespaces(struct ucounts *ucounts) static struct cgroup_namespace *alloc_cgroup_ns(void) { - struct cgroup_namespace *new_ns; + struct cgroup_namespace *new_ns __free(kfree) = NULL; int ret; new_ns = kzalloc(sizeof(struct cgroup_namespace), GFP_KERNEL_ACCOUNT); if (!new_ns) return ERR_PTR(-ENOMEM); - ret = ns_alloc_inum(&new_ns->ns); - if (ret) { - kfree(new_ns); + ret = ns_common_init(&new_ns->ns, &cgroupns_operations, true); + if (ret) return ERR_PTR(ret); - } - refcount_set(&new_ns->ns.count, 1); - new_ns->ns.ops = &cgroupns_operations; - return new_ns; + return no_free_ptr(new_ns); } void free_cgroup_ns(struct cgroup_namespace *ns) -- cgit v1.2.3 From 8e199cd6e3303d8f9352030ab9e4a221f277bd51 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:35 +0200 Subject: pid: use ns_common_init() Don't cargo-cult the same thing over and over. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/pid_namespace.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 7098ed44e717..20ce4052d1c5 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -102,17 +102,15 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns if (ns->pid_cachep == NULL) goto out_free_idr; - err = ns_alloc_inum(&ns->ns); + err = ns_common_init(&ns->ns, &pidns_operations, true); if (err) goto out_free_idr; - ns->ns.ops = &pidns_operations; ns->pid_max = PID_MAX_LIMIT; err = register_pidns_sysctls(ns); if (err) goto out_free_inum; - refcount_set(&ns->ns.count, 1); ns->level = level; ns->parent = get_pid_ns(parent_pid_ns); ns->user_ns = get_user_ns(user_ns); -- cgit v1.2.3 From 7b0e2c83624b02a8a11f04f0581721d95ea6e4a6 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:36 +0200 Subject: time: use ns_common_init() Don't cargo-cult the same thing over and over. Reviewed-by: Jan Kara Reviewed-by: Thomas Gleixner Signed-off-by: Christian Brauner --- kernel/time/namespace.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 667452768ed3..0be93d8f2896 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -88,22 +88,19 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns, goto fail; err = -ENOMEM; - ns = kmalloc(sizeof(*ns), GFP_KERNEL_ACCOUNT); + ns = kzalloc(sizeof(*ns), GFP_KERNEL_ACCOUNT); if (!ns) goto fail_dec; - refcount_set(&ns->ns.count, 1); - ns->vvar_page = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_ZERO); if (!ns->vvar_page) goto fail_free; - err = ns_alloc_inum(&ns->ns); + err = ns_common_init(&ns->ns, &timens_operations, true); if (err) goto fail_free_page; ns->ucounts = ucounts; - ns->ns.ops = &timens_operations; ns->user_ns = get_user_ns(user_ns); ns->offsets = old_ns->offsets; ns->frozen_offsets = false; -- cgit v1.2.3 From 00ed42285c46e2c84223e3ec5e1d07638948b4d1 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:37 +0200 Subject: user: use ns_common_init() Don't cargo-cult the same thing over and over. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/user_namespace.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index 682f40d5632d..98f4fe84d039 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -124,12 +124,11 @@ int create_user_ns(struct cred *new) goto fail_dec; ns->parent_could_setfcap = cap_raised(new->cap_effective, CAP_SETFCAP); - ret = ns_alloc_inum(&ns->ns); + + ret = ns_common_init(&ns->ns, &userns_operations, true); if (ret) goto fail_free; - ns->ns.ops = &userns_operations; - refcount_set(&ns->ns.count, 1); /* Leave the new->user_ns reference with the new user namespace. */ ns->parent = parent_ns; ns->level = parent_ns->level + 1; -- cgit v1.2.3 From 09337e064cbb692d5e0e34d7e1e2f2c53166f91a Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:38 +0200 Subject: uts: use ns_common_init() Don't cargo-cult the same thing over and over. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/utsname.c | 16 ++-------------- 1 file changed, 2 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/utsname.c b/kernel/utsname.c index b1ac3ca870f2..02037010b378 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -27,16 +27,6 @@ static void dec_uts_namespaces(struct ucounts *ucounts) dec_ucount(ucounts, UCOUNT_UTS_NAMESPACES); } -static struct uts_namespace *create_uts_ns(void) -{ - struct uts_namespace *uts_ns; - - uts_ns = kmem_cache_alloc(uts_ns_cache, GFP_KERNEL); - if (uts_ns) - refcount_set(&uts_ns->ns.count, 1); - return uts_ns; -} - /* * Clone a new ns copying an original utsname, setting refcount to 1 * @old_ns: namespace to clone @@ -55,17 +45,15 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns, goto fail; err = -ENOMEM; - ns = create_uts_ns(); + ns = kmem_cache_zalloc(uts_ns_cache, GFP_KERNEL); if (!ns) goto fail_dec; - err = ns_alloc_inum(&ns->ns); + err = ns_common_init(&ns->ns, &utsns_operations, true); if (err) goto fail_free; ns->ucounts = ucounts; - ns->ns.ops = &utsns_operations; - down_read(&uts_sem); memcpy(&ns->name, &old_ns->name, sizeof(ns->name)); ns->user_ns = get_user_ns(user_ns); -- cgit v1.2.3 From 885fc8ac0a4dc70f5d87b80b0977292870e35c60 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:40 +0200 Subject: nstree: make iterator generic Move the namespace iteration infrastructure originally introduced for mount namespaces into a generic library usable by all namespace types. Signed-off-by: Christian Brauner --- kernel/Makefile | 2 +- kernel/nstree.c | 233 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 234 insertions(+), 1 deletion(-) create mode 100644 kernel/nstree.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index c60623448235..b807516a1b43 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -8,7 +8,7 @@ obj-y = fork.o exec_domain.o panic.o \ sysctl.o capability.o ptrace.o user.o \ signal.o sys.o umh.o workqueue.o pid.o task_work.o \ extable.o params.o \ - kthread.o sys_ni.o nsproxy.o \ + kthread.o sys_ni.o nsproxy.o nstree.o \ notifier.o ksysfs.o cred.o reboot.o \ async.o range.o smpboot.o ucount.o regset.o ksyms_common.o diff --git a/kernel/nstree.c b/kernel/nstree.c new file mode 100644 index 000000000000..bbe8bedc924c --- /dev/null +++ b/kernel/nstree.c @@ -0,0 +1,233 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include +#include +#include + +struct ns_tree mnt_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(mnt_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(mnt_ns_tree.ns_tree_lock), + .type = CLONE_NEWNS, +}; + +struct ns_tree net_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(net_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(net_ns_tree.ns_tree_lock), + .type = CLONE_NEWNET, +}; +EXPORT_SYMBOL_GPL(net_ns_tree); + +struct ns_tree uts_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(uts_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(uts_ns_tree.ns_tree_lock), + .type = CLONE_NEWUTS, +}; + +struct ns_tree user_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(user_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(user_ns_tree.ns_tree_lock), + .type = CLONE_NEWUSER, +}; + +struct ns_tree ipc_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(ipc_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(ipc_ns_tree.ns_tree_lock), + .type = CLONE_NEWIPC, +}; + +struct ns_tree pid_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(pid_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(pid_ns_tree.ns_tree_lock), + .type = CLONE_NEWPID, +}; + +struct ns_tree cgroup_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(cgroup_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(cgroup_ns_tree.ns_tree_lock), + .type = CLONE_NEWCGROUP, +}; + +struct ns_tree time_ns_tree = { + .ns_tree = RB_ROOT, + .ns_list = LIST_HEAD_INIT(time_ns_tree.ns_list), + .ns_tree_lock = __SEQLOCK_UNLOCKED(time_ns_tree.ns_tree_lock), + .type = CLONE_NEWTIME, +}; + +DEFINE_COOKIE(namespace_cookie); + +static inline struct ns_common *node_to_ns(const struct rb_node *node) +{ + if (!node) + return NULL; + return rb_entry(node, struct ns_common, ns_tree_node); +} + +static inline int ns_cmp(struct rb_node *a, const struct rb_node *b) +{ + struct ns_common *ns_a = node_to_ns(a); + struct ns_common *ns_b = node_to_ns(b); + u64 ns_id_a = ns_a->ns_id; + u64 ns_id_b = ns_b->ns_id; + + if (ns_id_a < ns_id_b) + return -1; + if (ns_id_a > ns_id_b) + return 1; + return 0; +} + +void __ns_tree_add_raw(struct ns_common *ns, struct ns_tree *ns_tree) +{ + struct rb_node *node, *prev; + + VFS_WARN_ON_ONCE(!ns->ns_id); + + write_seqlock(&ns_tree->ns_tree_lock); + + VFS_WARN_ON_ONCE(ns->ops->type != ns_tree->type); + + node = rb_find_add_rcu(&ns->ns_tree_node, &ns_tree->ns_tree, ns_cmp); + /* + * If there's no previous entry simply add it after the + * head and if there is add it after the previous entry. + */ + prev = rb_prev(&ns->ns_tree_node); + if (!prev) + list_add_rcu(&ns->ns_list_node, &ns_tree->ns_list); + else + list_add_rcu(&ns->ns_list_node, &node_to_ns(prev)->ns_list_node); + + write_sequnlock(&ns_tree->ns_tree_lock); + + VFS_WARN_ON_ONCE(node); +} + +void __ns_tree_remove(struct ns_common *ns, struct ns_tree *ns_tree) +{ + VFS_WARN_ON_ONCE(RB_EMPTY_NODE(&ns->ns_tree_node)); + VFS_WARN_ON_ONCE(list_empty(&ns->ns_list_node)); + VFS_WARN_ON_ONCE(ns->ops->type != ns_tree->type); + + write_seqlock(&ns_tree->ns_tree_lock); + rb_erase(&ns->ns_tree_node, &ns_tree->ns_tree); + list_bidir_del_rcu(&ns->ns_list_node); + RB_CLEAR_NODE(&ns->ns_tree_node); + write_sequnlock(&ns_tree->ns_tree_lock); +} +EXPORT_SYMBOL_GPL(__ns_tree_remove); + +static int ns_find(const void *key, const struct rb_node *node) +{ + const u64 ns_id = *(u64 *)key; + const struct ns_common *ns = node_to_ns(node); + + if (ns_id < ns->ns_id) + return -1; + if (ns_id > ns->ns_id) + return 1; + return 0; +} + + +static struct ns_tree *ns_tree_from_type(int ns_type) +{ + switch (ns_type) { + case CLONE_NEWCGROUP: + return &cgroup_ns_tree; + case CLONE_NEWIPC: + return &ipc_ns_tree; + case CLONE_NEWNS: + return &mnt_ns_tree; + case CLONE_NEWNET: + return &net_ns_tree; + case CLONE_NEWPID: + return &pid_ns_tree; + case CLONE_NEWUSER: + return &user_ns_tree; + case CLONE_NEWUTS: + return &uts_ns_tree; + case CLONE_NEWTIME: + return &time_ns_tree; + } + + return NULL; +} + +struct ns_common *ns_tree_lookup_rcu(u64 ns_id, int ns_type) +{ + struct ns_tree *ns_tree; + struct rb_node *node; + unsigned int seq; + + RCU_LOCKDEP_WARN(!rcu_read_lock_held(), "suspicious ns_tree_lookup_rcu() usage"); + + ns_tree = ns_tree_from_type(ns_type); + if (!ns_tree) + return NULL; + + do { + seq = read_seqbegin(&ns_tree->ns_tree_lock); + node = rb_find_rcu(&ns_id, &ns_tree->ns_tree, ns_find); + if (node) + break; + } while (read_seqretry(&ns_tree->ns_tree_lock, seq)); + + if (!node) + return NULL; + + VFS_WARN_ON_ONCE(node_to_ns(node)->ops->type != ns_type); + + return node_to_ns(node); +} + +/** + * ns_tree_adjoined_rcu - find the next/previous namespace in the same + * tree + * @ns: namespace to start from + * @previous: if true find the previous namespace, otherwise the next + * + * Find the next or previous namespace in the same tree as @ns. If + * there is no next/previous namespace, -ENOENT is returned. + */ +struct ns_common *__ns_tree_adjoined_rcu(struct ns_common *ns, + struct ns_tree *ns_tree, bool previous) +{ + struct list_head *list; + + RCU_LOCKDEP_WARN(!rcu_read_lock_held(), "suspicious ns_tree_adjoined_rcu() usage"); + + if (previous) + list = rcu_dereference(list_bidir_prev_rcu(&ns->ns_list_node)); + else + list = rcu_dereference(list_next_rcu(&ns->ns_list_node)); + if (list_is_head(list, &ns_tree->ns_list)) + return ERR_PTR(-ENOENT); + + VFS_WARN_ON_ONCE(list_entry_rcu(list, struct ns_common, ns_list_node)->ops->type != ns_tree->type); + + return list_entry_rcu(list, struct ns_common, ns_list_node); +} + +/** + * ns_tree_gen_id - generate a new namespace id + * @ns: namespace to generate id for + * + * Generates a new namespace id and assigns it to the namespace. All + * namespaces types share the same id space and thus can be compared + * directly. IOW, when two ids of two namespace are equal, they are + * identical. + */ +u64 ns_tree_gen_id(struct ns_common *ns) +{ + guard(preempt)(); + ns->ns_id = gen_cookie_next(&namespace_cookie); + return ns->ns_id; +} -- cgit v1.2.3 From 7c60593985331e7839ec3fea6328a3253a325e82 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:42 +0200 Subject: cgroup: support ns lookup Support the generic ns lookup infrastructure to support file handles for namespaces. Acked-by: Tejun Heo Signed-off-by: Christian Brauner --- kernel/cgroup/cgroup.c | 2 ++ kernel/cgroup/namespace.c | 7 +++++-- 2 files changed, 7 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index 312c6a8b55bb..092e6bf081ed 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -59,6 +59,7 @@ #include #include #include +#include #include #define CREATE_TRACE_POINTS @@ -6312,6 +6313,7 @@ int __init cgroup_init(void) WARN_ON(register_filesystem(&cpuset_fs_type)); #endif + ns_tree_add(&init_cgroup_ns); return 0; } diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index 0391b6ab0bf1..fc12c416dfeb 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -5,7 +5,7 @@ #include #include #include - +#include /* cgroup namespaces */ @@ -30,16 +30,19 @@ static struct cgroup_namespace *alloc_cgroup_ns(void) ret = ns_common_init(&new_ns->ns, &cgroupns_operations, true); if (ret) return ERR_PTR(ret); + ns_tree_add(new_ns); return no_free_ptr(new_ns); } void free_cgroup_ns(struct cgroup_namespace *ns) { + ns_tree_remove(ns); put_css_set(ns->root_cset); dec_cgroup_namespaces(ns->ucounts); put_user_ns(ns->user_ns); ns_free_inum(&ns->ns); - kfree(ns); + /* Concurrent nstree traversal depends on a grace period. */ + kfree_rcu(ns, ns.ns_rcu); } EXPORT_SYMBOL(free_cgroup_ns); -- cgit v1.2.3 From 488acdcec8e24377506934a95e0ba21619073e8f Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:45 +0200 Subject: pid: support ns lookup Support the generic ns lookup infrastructure to support file handles for namespaces. Signed-off-by: Christian Brauner --- kernel/pid_namespace.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 20ce4052d1c5..228ae20299f9 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -23,6 +23,7 @@ #include #include #include +#include #include #include "pid_sysctl.h" @@ -122,6 +123,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns ns->memfd_noexec_scope = pidns_memfd_noexec_scope(parent_pid_ns); #endif + ns_tree_add(ns); return ns; out_free_inum: @@ -147,6 +149,7 @@ static void delayed_free_pidns(struct rcu_head *p) static void destroy_pid_namespace(struct pid_namespace *ns) { + ns_tree_remove(ns); unregister_pidns_sysctls(ns); ns_free_inum(&ns->ns); @@ -473,6 +476,7 @@ static __init int pid_namespaces_init(void) #endif register_pid_ns_sysctl_table_vm(); + ns_tree_add(&init_pid_ns); return 0; } -- cgit v1.2.3 From b36c823b9a4be5b0c8e38c3fd60cade7d41c216c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:46 +0200 Subject: time: support ns lookup Support the generic ns lookup infrastructure to support file handles for namespaces. Reviewed-by: Thomas Gleixner Signed-off-by: Christian Brauner --- kernel/time/namespace.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 0be93d8f2896..408f60d0a3b6 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include #include @@ -104,6 +105,7 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns, ns->user_ns = get_user_ns(user_ns); ns->offsets = old_ns->offsets; ns->frozen_offsets = false; + ns_tree_add(ns); return ns; fail_free_page: @@ -250,11 +252,13 @@ out: void free_time_ns(struct time_namespace *ns) { + ns_tree_remove(ns); dec_time_namespaces(ns->ucounts); put_user_ns(ns->user_ns); ns_free_inum(&ns->ns); __free_page(ns->vvar_page); - kfree(ns); + /* Concurrent nstree traversal depends on a grace period. */ + kfree_rcu(ns, ns.ns_rcu); } static struct time_namespace *to_time_ns(struct ns_common *ns) @@ -487,3 +491,8 @@ struct time_namespace init_time_ns = { .ns.ops = &timens_operations, .frozen_offsets = true, }; + +void __init time_ns_init(void) +{ + ns_tree_add(&init_time_ns); +} -- cgit v1.2.3 From 2f5243cbba6cff66cc8b43bdc14853282f5b1e67 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:47 +0200 Subject: user: support ns lookup Support the generic ns lookup infrastructure to support file handles for namespaces. Signed-off-by: Christian Brauner --- kernel/user_namespace.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index 98f4fe84d039..ade5b6806c5c 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -21,6 +21,7 @@ #include #include #include +#include static struct kmem_cache *user_ns_cachep __ro_after_init; static DEFINE_MUTEX(userns_state_mutex); @@ -158,6 +159,7 @@ int create_user_ns(struct cred *new) goto fail_keyring; set_cred_user_ns(new, ns); + ns_tree_add(ns); return 0; fail_keyring: #ifdef CONFIG_PERSISTENT_KEYRINGS @@ -200,6 +202,7 @@ static void free_user_ns(struct work_struct *work) do { struct ucounts *ucounts = ns->ucounts; parent = ns->parent; + ns_tree_remove(ns); if (ns->gid_map.nr_extents > UID_GID_MAP_MAX_BASE_EXTENTS) { kfree(ns->gid_map.forward); kfree(ns->gid_map.reverse); @@ -218,7 +221,8 @@ static void free_user_ns(struct work_struct *work) retire_userns_sysctls(ns); key_free_user_ns(ns); ns_free_inum(&ns->ns); - kmem_cache_free(user_ns_cachep, ns); + /* Concurrent nstree traversal depends on a grace period. */ + kfree_rcu(ns, ns.ns_rcu); dec_user_namespaces(ucounts); ns = parent; } while (refcount_dec_and_test(&parent->ns.count)); @@ -1412,6 +1416,7 @@ const struct proc_ns_operations userns_operations = { static __init int user_namespaces_init(void) { user_ns_cachep = KMEM_CACHE(user_namespace, SLAB_PANIC | SLAB_ACCOUNT); + ns_tree_add(&init_user_ns); return 0; } subsys_initcall(user_namespaces_init); -- cgit v1.2.3 From 58f976d41fd915acd2403a2b8e0eef8e9c478357 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:48 +0200 Subject: uts: support ns lookup Support the generic ns lookup infrastructure to support file handles for namespaces. Signed-off-by: Christian Brauner --- kernel/utsname.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/utsname.c b/kernel/utsname.c index 02037010b378..64155417ae0c 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -13,6 +13,7 @@ #include #include #include +#include #include static struct kmem_cache *uts_ns_cache __ro_after_init; @@ -58,6 +59,7 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns, memcpy(&ns->name, &old_ns->name, sizeof(ns->name)); ns->user_ns = get_user_ns(user_ns); up_read(&uts_sem); + ns_tree_add(ns); return ns; fail_free: @@ -93,10 +95,12 @@ struct uts_namespace *copy_utsname(unsigned long flags, void free_uts_ns(struct uts_namespace *ns) { + ns_tree_remove(ns); dec_uts_namespaces(ns->ucounts); put_user_ns(ns->user_ns); ns_free_inum(&ns->ns); - kmem_cache_free(uts_ns_cache, ns); + /* Concurrent nstree traversal depends on a grace period. */ + kfree_rcu(ns, ns.ns_rcu); } static inline struct uts_namespace *to_uts_ns(struct ns_common *ns) @@ -162,4 +166,5 @@ void __init uts_ns_init(void) offsetof(struct uts_namespace, name), sizeof_field(struct uts_namespace, name), NULL); + ns_tree_add(&init_uts_ns); } -- cgit v1.2.3 From d7afdf889561058068ab46fd8f306c70ef29216a Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 12 Sep 2025 13:52:49 +0200 Subject: ns: add to__ns() to respective headers Every namespace type has a container_of(ns, , ns) static inline function that is currently not exposed in the header. So we have a bunch of places that open-code it via container_of(). Move it to the headers so we can use it directly. Reviewed-by: Aleksa Sarai Signed-off-by: Christian Brauner --- kernel/cgroup/namespace.c | 5 ----- kernel/pid_namespace.c | 5 ----- kernel/time/namespace.c | 5 ----- kernel/user_namespace.c | 5 ----- kernel/utsname.c | 5 ----- 5 files changed, 25 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index fc12c416dfeb..5a327914b565 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -89,11 +89,6 @@ struct cgroup_namespace *copy_cgroup_ns(unsigned long flags, return new_ns; } -static inline struct cgroup_namespace *to_cg_ns(struct ns_common *ns) -{ - return container_of(ns, struct cgroup_namespace, ns); -} - static int cgroupns_install(struct nsset *nsset, struct ns_common *ns) { struct nsproxy *nsproxy = nsset->nsproxy; diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 228ae20299f9..9b327420309e 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -345,11 +345,6 @@ int reboot_pid_ns(struct pid_namespace *pid_ns, int cmd) return 0; } -static inline struct pid_namespace *to_pid_ns(struct ns_common *ns) -{ - return container_of(ns, struct pid_namespace, ns); -} - static struct ns_common *pidns_get(struct task_struct *task) { struct pid_namespace *ns; diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 408f60d0a3b6..20b65f90549e 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -261,11 +261,6 @@ void free_time_ns(struct time_namespace *ns) kfree_rcu(ns, ns.ns_rcu); } -static struct time_namespace *to_time_ns(struct ns_common *ns) -{ - return container_of(ns, struct time_namespace, ns); -} - static struct ns_common *timens_get(struct task_struct *task) { struct time_namespace *ns = NULL; diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index ade5b6806c5c..cfb0e28f2779 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -1325,11 +1325,6 @@ bool current_in_userns(const struct user_namespace *target_ns) } EXPORT_SYMBOL(current_in_userns); -static inline struct user_namespace *to_user_ns(struct ns_common *ns) -{ - return container_of(ns, struct user_namespace, ns); -} - static struct ns_common *userns_get(struct task_struct *task) { struct user_namespace *user_ns; diff --git a/kernel/utsname.c b/kernel/utsname.c index 64155417ae0c..a682830742d3 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -103,11 +103,6 @@ void free_uts_ns(struct uts_namespace *ns) kfree_rcu(ns, ns.ns_rcu); } -static inline struct uts_namespace *to_uts_ns(struct ns_common *ns) -{ - return container_of(ns, struct uts_namespace, ns); -} - static struct ns_common *utsns_get(struct task_struct *task) { struct uts_namespace *ns = NULL; -- cgit v1.2.3 From f74ca6da113d5d4b21c00bb4da3f3c137162b4fe Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 17 Sep 2025 12:28:02 +0200 Subject: nscommon: move to separate file It's really awkward spilling the ns common infrastructure into multiple headers. Move it to a separate file. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/Makefile | 2 +- kernel/nscommon.c | 21 +++++++++++++++++++++ 2 files changed, 22 insertions(+), 1 deletion(-) create mode 100644 kernel/nscommon.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index b807516a1b43..1f48f7cd2d7b 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -8,7 +8,7 @@ obj-y = fork.o exec_domain.o panic.o \ sysctl.o capability.o ptrace.o user.o \ signal.o sys.o umh.o workqueue.o pid.o task_work.o \ extable.o params.o \ - kthread.o sys_ni.o nsproxy.o nstree.o \ + kthread.o sys_ni.o nsproxy.o nstree.o nscommon.o \ notifier.o ksysfs.o cred.o reboot.o \ async.o range.o smpboot.o ucount.o regset.o ksyms_common.o diff --git a/kernel/nscommon.c b/kernel/nscommon.c new file mode 100644 index 000000000000..ebf4783d0505 --- /dev/null +++ b/kernel/nscommon.c @@ -0,0 +1,21 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include + +int ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, + bool alloc_inum) +{ + if (alloc_inum) { + int ret; + ret = proc_alloc_inum(&ns->inum); + if (ret) + return ret; + } + refcount_set(&ns->count, 1); + ns->stashed = NULL; + ns->ops = ops; + ns->ns_id = 0; + RB_CLEAR_NODE(&ns->ns_tree_node); + INIT_LIST_HEAD(&ns->ns_list_node); + return 0; +} -- cgit v1.2.3 From 86cdbae5c61c6b8c0a2adc78dbbb0314b3254a9c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 17 Sep 2025 12:28:05 +0200 Subject: mnt: simplify ns_common_init() handling Assign the reserved MNT_NS_ANON_INO sentinel to anonymous mount namespaces and cleanup the initial mount ns allocation. This is just a preparatory patch and the ns->inum check in ns_common_init() will be dropped in the next patch. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/nscommon.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/nscommon.c b/kernel/nscommon.c index ebf4783d0505..e10fad8afe61 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -5,7 +5,7 @@ int ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, bool alloc_inum) { - if (alloc_inum) { + if (alloc_inum && !ns->inum) { int ret; ret = proc_alloc_inum(&ns->inum); if (ret) -- cgit v1.2.3 From 5612ff3ec588be09f11a9424db6d1186bcdeb3fa Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 17 Sep 2025 12:28:07 +0200 Subject: nscommon: simplify initialization There's a lot of information that namespace implementers don't need to know about at all. Encapsulate this all in the initialization helper. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/namespace.c | 2 +- kernel/nscommon.c | 17 ++++++++--------- kernel/pid_namespace.c | 2 +- kernel/time/namespace.c | 2 +- kernel/user_namespace.c | 2 +- kernel/utsname.c | 2 +- 6 files changed, 13 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index 5a327914b565..d928c557e28b 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -27,7 +27,7 @@ static struct cgroup_namespace *alloc_cgroup_ns(void) new_ns = kzalloc(sizeof(struct cgroup_namespace), GFP_KERNEL_ACCOUNT); if (!new_ns) return ERR_PTR(-ENOMEM); - ret = ns_common_init(&new_ns->ns, &cgroupns_operations, true); + ret = ns_common_init(new_ns, &cgroupns_operations); if (ret) return ERR_PTR(ret); ns_tree_add(new_ns); diff --git a/kernel/nscommon.c b/kernel/nscommon.c index e10fad8afe61..c3a90bb665ad 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -1,21 +1,20 @@ // SPDX-License-Identifier: GPL-2.0-only #include +#include -int ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, - bool alloc_inum) +int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, int inum) { - if (alloc_inum && !ns->inum) { - int ret; - ret = proc_alloc_inum(&ns->inum); - if (ret) - return ret; - } refcount_set(&ns->count, 1); ns->stashed = NULL; ns->ops = ops; ns->ns_id = 0; RB_CLEAR_NODE(&ns->ns_tree_node); INIT_LIST_HEAD(&ns->ns_list_node); - return 0; + + if (inum) { + ns->inum = inum; + return 0; + } + return proc_alloc_inum(&ns->inum); } diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 9b327420309e..170757c265c2 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -103,7 +103,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns if (ns->pid_cachep == NULL) goto out_free_idr; - err = ns_common_init(&ns->ns, &pidns_operations, true); + err = ns_common_init(ns, &pidns_operations); if (err) goto out_free_idr; diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 20b65f90549e..ce8e952104a7 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -97,7 +97,7 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns, if (!ns->vvar_page) goto fail_free; - err = ns_common_init(&ns->ns, &timens_operations, true); + err = ns_common_init(ns, &timens_operations); if (err) goto fail_free_page; diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index cfb0e28f2779..db9f0463219c 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -126,7 +126,7 @@ int create_user_ns(struct cred *new) ns->parent_could_setfcap = cap_raised(new->cap_effective, CAP_SETFCAP); - ret = ns_common_init(&ns->ns, &userns_operations, true); + ret = ns_common_init(ns, &userns_operations); if (ret) goto fail_free; diff --git a/kernel/utsname.c b/kernel/utsname.c index a682830742d3..399888be66bd 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -50,7 +50,7 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns, if (!ns) goto fail_dec; - err = ns_common_init(&ns->ns, &utsns_operations, true); + err = ns_common_init(ns, &utsns_operations); if (err) goto fail_free; -- cgit v1.2.3 From be5f21d3985f00827e09b798f7a07ebd6dd7f54a Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 17 Sep 2025 12:28:08 +0200 Subject: ns: add ns_common_free() And drop ns_free_inum(). Anything common that can be wasted centrally should be wasted in the new common helper. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/namespace.c | 2 +- kernel/nscommon.c | 5 +++++ kernel/pid_namespace.c | 4 ++-- kernel/time/namespace.c | 2 +- kernel/user_namespace.c | 4 ++-- kernel/utsname.c | 2 +- 6 files changed, 12 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index d928c557e28b..16ead7508371 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -40,7 +40,7 @@ void free_cgroup_ns(struct cgroup_namespace *ns) put_css_set(ns->root_cset); dec_cgroup_namespaces(ns->ucounts); put_user_ns(ns->user_ns); - ns_free_inum(&ns->ns); + ns_common_free(ns); /* Concurrent nstree traversal depends on a grace period. */ kfree_rcu(ns, ns.ns_rcu); } diff --git a/kernel/nscommon.c b/kernel/nscommon.c index c3a90bb665ad..7c1b07e2a6c9 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -18,3 +18,8 @@ int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, } return proc_alloc_inum(&ns->inum); } + +void __ns_common_free(struct ns_common *ns) +{ + proc_free_inum(ns->inum); +} diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 170757c265c2..27e2dd9ee051 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -127,7 +127,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns return ns; out_free_inum: - ns_free_inum(&ns->ns); + ns_common_free(ns); out_free_idr: idr_destroy(&ns->idr); kmem_cache_free(pid_ns_cachep, ns); @@ -152,7 +152,7 @@ static void destroy_pid_namespace(struct pid_namespace *ns) ns_tree_remove(ns); unregister_pidns_sysctls(ns); - ns_free_inum(&ns->ns); + ns_common_free(ns); idr_destroy(&ns->idr); call_rcu(&ns->rcu, delayed_free_pidns); diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index ce8e952104a7..d49c73015d6e 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -255,7 +255,7 @@ void free_time_ns(struct time_namespace *ns) ns_tree_remove(ns); dec_time_namespaces(ns->ucounts); put_user_ns(ns->user_ns); - ns_free_inum(&ns->ns); + ns_common_free(ns); __free_page(ns->vvar_page); /* Concurrent nstree traversal depends on a grace period. */ kfree_rcu(ns, ns.ns_rcu); diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index db9f0463219c..32406bcab526 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -165,7 +165,7 @@ fail_keyring: #ifdef CONFIG_PERSISTENT_KEYRINGS key_put(ns->persistent_keyring_register); #endif - ns_free_inum(&ns->ns); + ns_common_free(ns); fail_free: kmem_cache_free(user_ns_cachep, ns); fail_dec: @@ -220,7 +220,7 @@ static void free_user_ns(struct work_struct *work) #endif retire_userns_sysctls(ns); key_free_user_ns(ns); - ns_free_inum(&ns->ns); + ns_common_free(ns); /* Concurrent nstree traversal depends on a grace period. */ kfree_rcu(ns, ns.ns_rcu); dec_user_namespaces(ucounts); diff --git a/kernel/utsname.c b/kernel/utsname.c index 399888be66bd..95d733eb2c98 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -98,7 +98,7 @@ void free_uts_ns(struct uts_namespace *ns) ns_tree_remove(ns); dec_uts_namespaces(ns->ucounts); put_user_ns(ns->user_ns); - ns_free_inum(&ns->ns); + ns_common_free(ns); /* Concurrent nstree traversal depends on a grace period. */ kfree_rcu(ns, ns.ns_rcu); } -- cgit v1.2.3 From 07897b38eadf5a370a6001790239f23036d5b970 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 18 Sep 2025 12:11:50 +0200 Subject: pid: port to ns_ref_*() helpers Stop accessing ns.count directly. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/pid_namespace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 27e2dd9ee051..162f5fb63d75 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -169,7 +169,7 @@ static void destroy_pid_namespace_work(struct work_struct *work) parent = ns->parent; destroy_pid_namespace(ns); ns = parent; - } while (ns != &init_pid_ns && refcount_dec_and_test(&ns->ns.count)); + } while (ns != &init_pid_ns && ns_ref_put(ns)); } struct pid_namespace *copy_pid_ns(unsigned long flags, @@ -184,7 +184,7 @@ struct pid_namespace *copy_pid_ns(unsigned long flags, void put_pid_ns(struct pid_namespace *ns) { - if (ns && ns != &init_pid_ns && refcount_dec_and_test(&ns->ns.count)) + if (ns && ns != &init_pid_ns && ns_ref_put(ns)) schedule_work(&ns->work); } EXPORT_SYMBOL_GPL(put_pid_ns); -- cgit v1.2.3 From 96d997ea5ad1911cc393ffdb5c928b532f2f921a Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 18 Sep 2025 12:11:52 +0200 Subject: user: port to ns_ref_*() helpers Stop accessing ns.count directly. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/user_namespace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index 32406bcab526..f9df45c46235 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -225,7 +225,7 @@ static void free_user_ns(struct work_struct *work) kfree_rcu(ns, ns.ns_rcu); dec_user_namespaces(ucounts); ns = parent; - } while (refcount_dec_and_test(&parent->ns.count)); + } while (ns_ref_put(parent)); } void __put_user_ns(struct user_namespace *ns) -- cgit v1.2.3 From 024596a4e2802e457a9f92af79f246fa9631f8de Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 18 Sep 2025 12:11:59 +0200 Subject: ns: rename to __ns_ref Make it easier to grep and rename to ns_count. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/cgroup.c | 2 +- kernel/nscommon.c | 2 +- kernel/pid.c | 2 +- kernel/time/namespace.c | 2 +- kernel/user.c | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index 092e6bf081ed..a0e24adceef0 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -219,7 +219,7 @@ static bool have_favordynmods __ro_after_init = IS_ENABLED(CONFIG_CGROUP_FAVOR_D /* cgroup namespace for init task */ struct cgroup_namespace init_cgroup_ns = { - .ns.count = REFCOUNT_INIT(2), + .ns.__ns_ref = REFCOUNT_INIT(2), .user_ns = &init_user_ns, .ns.ops = &cgroupns_operations, .ns.inum = PROC_CGROUP_INIT_INO, diff --git a/kernel/nscommon.c b/kernel/nscommon.c index 7c1b07e2a6c9..7aa2be6a0c32 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -5,7 +5,7 @@ int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, int inum) { - refcount_set(&ns->count, 1); + refcount_set(&ns->__ns_ref, 1); ns->stashed = NULL; ns->ops = ops; ns->ns_id = 0; diff --git a/kernel/pid.c b/kernel/pid.c index c45a28c16cd2..e222426f745d 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -71,7 +71,7 @@ static int pid_max_max = PID_MAX_LIMIT; * the scheme scales to up to 4 million PIDs, runtime. */ struct pid_namespace init_pid_ns = { - .ns.count = REFCOUNT_INIT(2), + .ns.__ns_ref = REFCOUNT_INIT(2), .idr = IDR_INIT(init_pid_ns.idr), .pid_allocated = PIDNS_ADDING, .level = 0, diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index d49c73015d6e..d70bdfb7b001 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -480,7 +480,7 @@ const struct proc_ns_operations timens_for_children_operations = { }; struct time_namespace init_time_ns = { - .ns.count = REFCOUNT_INIT(3), + .ns.__ns_ref = REFCOUNT_INIT(3), .user_ns = &init_user_ns, .ns.inum = PROC_TIME_INIT_INO, .ns.ops = &timens_operations, diff --git a/kernel/user.c b/kernel/user.c index f46b1d41163b..17a742fb4e10 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -65,7 +65,7 @@ struct user_namespace init_user_ns = { .nr_extents = 1, }, }, - .ns.count = REFCOUNT_INIT(3), + .ns.__ns_ref = REFCOUNT_INIT(3), .owner = GLOBAL_ROOT_UID, .group = GLOBAL_ROOT_GID, .ns.inum = PROC_USER_INIT_INO, -- cgit v1.2.3 From 7cf730321132e726ff949c6f3c0d5c598788f7a2 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Fri, 19 Sep 2025 11:29:49 +0200 Subject: ns: use inode initializer for initial namespaces Just use the common helper we have. Signed-off-by: Christian Brauner --- kernel/cgroup/cgroup.c | 2 +- kernel/pid.c | 2 +- kernel/time/namespace.c | 2 +- kernel/user.c | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index a0e24adceef0..245b43ff2fa4 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -222,7 +222,7 @@ struct cgroup_namespace init_cgroup_ns = { .ns.__ns_ref = REFCOUNT_INIT(2), .user_ns = &init_user_ns, .ns.ops = &cgroupns_operations, - .ns.inum = PROC_CGROUP_INIT_INO, + .ns.inum = ns_init_inum(&init_cgroup_ns), .root_cset = &init_css_set, }; diff --git a/kernel/pid.c b/kernel/pid.c index e222426f745d..7e8c66e0bf67 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -77,7 +77,7 @@ struct pid_namespace init_pid_ns = { .level = 0, .child_reaper = &init_task, .user_ns = &init_user_ns, - .ns.inum = PROC_PID_INIT_INO, + .ns.inum = ns_init_inum(&init_pid_ns), #ifdef CONFIG_PID_NS .ns.ops = &pidns_operations, #endif diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index d70bdfb7b001..7aa4d6fedd49 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -482,7 +482,7 @@ const struct proc_ns_operations timens_for_children_operations = { struct time_namespace init_time_ns = { .ns.__ns_ref = REFCOUNT_INIT(3), .user_ns = &init_user_ns, - .ns.inum = PROC_TIME_INIT_INO, + .ns.inum = ns_init_inum(&init_time_ns), .ns.ops = &timens_operations, .frozen_offsets = true, }; diff --git a/kernel/user.c b/kernel/user.c index 17a742fb4e10..b2a53674d506 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -68,7 +68,7 @@ struct user_namespace init_user_ns = { .ns.__ns_ref = REFCOUNT_INIT(3), .owner = GLOBAL_ROOT_UID, .group = GLOBAL_ROOT_GID, - .ns.inum = PROC_USER_INIT_INO, + .ns.inum = ns_init_inum(&init_user_ns), #ifdef CONFIG_USER_NS .ns.ops = &userns_operations, #endif -- cgit v1.2.3 From d7610cb7454bbd8bf6d58f71b0ed57155d3c545f Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Mon, 22 Sep 2025 14:42:36 +0200 Subject: ns: simplify ns_common_init() further Simply derive the ns operations from the namespace type. Acked-by: Thomas Gleixner Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/namespace.c | 2 +- kernel/pid_namespace.c | 2 +- kernel/time/namespace.c | 2 +- kernel/user_namespace.c | 2 +- kernel/utsname.c | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index 16ead7508371..04c98338ac08 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -27,7 +27,7 @@ static struct cgroup_namespace *alloc_cgroup_ns(void) new_ns = kzalloc(sizeof(struct cgroup_namespace), GFP_KERNEL_ACCOUNT); if (!new_ns) return ERR_PTR(-ENOMEM); - ret = ns_common_init(new_ns, &cgroupns_operations); + ret = ns_common_init(new_ns); if (ret) return ERR_PTR(ret); ns_tree_add(new_ns); diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index 162f5fb63d75..a262a3f19443 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -103,7 +103,7 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns if (ns->pid_cachep == NULL) goto out_free_idr; - err = ns_common_init(ns, &pidns_operations); + err = ns_common_init(ns); if (err) goto out_free_idr; diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 7aa4d6fedd49..9f26e61be044 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -97,7 +97,7 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns, if (!ns->vvar_page) goto fail_free; - err = ns_common_init(ns, &timens_operations); + err = ns_common_init(ns); if (err) goto fail_free_page; diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index f9df45c46235..e1559e8a8a02 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -126,7 +126,7 @@ int create_user_ns(struct cred *new) ns->parent_could_setfcap = cap_raised(new->cap_effective, CAP_SETFCAP); - ret = ns_common_init(ns, &userns_operations); + ret = ns_common_init(ns); if (ret) goto fail_free; diff --git a/kernel/utsname.c b/kernel/utsname.c index 95d733eb2c98..00001592ad13 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -50,7 +50,7 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns, if (!ns) goto fail_dec; - err = ns_common_init(ns, &utsns_operations); + err = ns_common_init(ns); if (err) goto fail_free; -- cgit v1.2.3 From 5890f504ef543190beae2a4e244bbfa7c3e0b57c Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Mon, 22 Sep 2025 14:42:37 +0200 Subject: ns: add ns_debug() Add ns_debug() that asserts that the correct operations are used for the namespace type. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/nscommon.c | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 53 insertions(+) (limited to 'kernel') diff --git a/kernel/nscommon.c b/kernel/nscommon.c index 7aa2be6a0c32..3cef89ddef41 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -2,6 +2,55 @@ #include #include +#include + +#ifdef CONFIG_DEBUG_VFS +static void ns_debug(struct ns_common *ns, const struct proc_ns_operations *ops) +{ + switch (ns->ops->type) { +#ifdef CONFIG_CGROUPS + case CLONE_NEWCGROUP: + VFS_WARN_ON_ONCE(ops != &cgroupns_operations); + break; +#endif +#ifdef CONFIG_IPC_NS + case CLONE_NEWIPC: + VFS_WARN_ON_ONCE(ops != &ipcns_operations); + break; +#endif + case CLONE_NEWNS: + VFS_WARN_ON_ONCE(ops != &mntns_operations); + break; +#ifdef CONFIG_NET_NS + case CLONE_NEWNET: + VFS_WARN_ON_ONCE(ops != &netns_operations); + break; +#endif +#ifdef CONFIG_PID_NS + case CLONE_NEWPID: + VFS_WARN_ON_ONCE(ops != &pidns_operations); + break; +#endif +#ifdef CONFIG_TIME_NS + case CLONE_NEWTIME: + VFS_WARN_ON_ONCE(ops != &timens_operations); + break; +#endif +#ifdef CONFIG_USER_NS + case CLONE_NEWUSER: + VFS_WARN_ON_ONCE(ops != &userns_operations); + break; +#endif +#ifdef CONFIG_UTS_NS + case CLONE_NEWUTS: + VFS_WARN_ON_ONCE(ops != &utsns_operations); + break; +#endif + default: + VFS_WARN_ON_ONCE(true); + } +} +#endif int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, int inum) { @@ -12,6 +61,10 @@ int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, RB_CLEAR_NODE(&ns->ns_tree_node); INIT_LIST_HEAD(&ns->ns_list_node); +#ifdef CONFIG_DEBUG_VFS + ns_debug(ns, ops); +#endif + if (inum) { ns->inum = inum; return 0; -- cgit v1.2.3 From 10cdfcd37ade7ce736bc4a1927680f390a6b1f7b Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 24 Sep 2025 13:33:58 +0200 Subject: nstree: make struct ns_tree private Don't expose it directly. There's no need to do that. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/nstree.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'kernel') diff --git a/kernel/nstree.c b/kernel/nstree.c index bbe8bedc924c..ecc88b013eff 100644 --- a/kernel/nstree.c +++ b/kernel/nstree.c @@ -4,6 +4,20 @@ #include #include +/** + * struct ns_tree - Namespace tree + * @ns_tree: Rbtree of namespaces of a particular type + * @ns_list: Sequentially walkable list of all namespaces of this type + * @ns_tree_lock: Seqlock to protect the tree and list + * @type: type of namespaces in this tree + */ +struct ns_tree { + struct rb_root ns_tree; + struct list_head ns_list; + seqlock_t ns_tree_lock; + int type; +}; + struct ns_tree mnt_ns_tree = { .ns_tree = RB_ROOT, .ns_list = LIST_HEAD_INIT(mnt_ns_tree.ns_list), -- cgit v1.2.3 From 4055526d35746ce8b04bfa5e14e14f28bb163186 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 24 Sep 2025 13:33:59 +0200 Subject: ns: move ns type into struct ns_common It's misplaced in struct proc_ns_operations and ns->ops might be NULL if the namespace is compiled out but we still want to know the type of the namespace for the initial namespace struct. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/cgroup/cgroup.c | 1 + kernel/cgroup/namespace.c | 1 - kernel/nscommon.c | 5 +++-- kernel/nsproxy.c | 4 ++-- kernel/nstree.c | 8 ++++---- kernel/pid.c | 1 + kernel/pid_namespace.c | 2 -- kernel/time/namespace.c | 3 +-- kernel/user.c | 1 + kernel/user_namespace.c | 1 - kernel/utsname.c | 1 - 11 files changed, 13 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index 245b43ff2fa4..9b75102e81cb 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -224,6 +224,7 @@ struct cgroup_namespace init_cgroup_ns = { .ns.ops = &cgroupns_operations, .ns.inum = ns_init_inum(&init_cgroup_ns), .root_cset = &init_css_set, + .ns.ns_type = ns_common_type(&init_cgroup_ns), }; static struct file_system_type cgroup2_fs_type; diff --git a/kernel/cgroup/namespace.c b/kernel/cgroup/namespace.c index 04c98338ac08..241ca05f07c8 100644 --- a/kernel/cgroup/namespace.c +++ b/kernel/cgroup/namespace.c @@ -137,7 +137,6 @@ static struct user_namespace *cgroupns_owner(struct ns_common *ns) const struct proc_ns_operations cgroupns_operations = { .name = "cgroup", - .type = CLONE_NEWCGROUP, .get = cgroupns_get, .put = cgroupns_put, .install = cgroupns_install, diff --git a/kernel/nscommon.c b/kernel/nscommon.c index 3cef89ddef41..92c9df1e8774 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -7,7 +7,7 @@ #ifdef CONFIG_DEBUG_VFS static void ns_debug(struct ns_common *ns, const struct proc_ns_operations *ops) { - switch (ns->ops->type) { + switch (ns->ns_type) { #ifdef CONFIG_CGROUPS case CLONE_NEWCGROUP: VFS_WARN_ON_ONCE(ops != &cgroupns_operations); @@ -52,12 +52,13 @@ static void ns_debug(struct ns_common *ns, const struct proc_ns_operations *ops) } #endif -int __ns_common_init(struct ns_common *ns, const struct proc_ns_operations *ops, int inum) +int __ns_common_init(struct ns_common *ns, u32 ns_type, const struct proc_ns_operations *ops, int inum) { refcount_set(&ns->__ns_ref, 1); ns->stashed = NULL; ns->ops = ops; ns->ns_id = 0; + ns->ns_type = ns_type; RB_CLEAR_NODE(&ns->ns_tree_node); INIT_LIST_HEAD(&ns->ns_list_node); diff --git a/kernel/nsproxy.c b/kernel/nsproxy.c index 5f31fdff8a38..8d62449237b6 100644 --- a/kernel/nsproxy.c +++ b/kernel/nsproxy.c @@ -545,9 +545,9 @@ SYSCALL_DEFINE2(setns, int, fd, int, flags) if (proc_ns_file(fd_file(f))) { ns = get_proc_ns(file_inode(fd_file(f))); - if (flags && (ns->ops->type != flags)) + if (flags && (ns->ns_type != flags)) err = -EINVAL; - flags = ns->ops->type; + flags = ns->ns_type; } else if (!IS_ERR(pidfd_pid(fd_file(f)))) { err = check_setns_flags(flags); } else { diff --git a/kernel/nstree.c b/kernel/nstree.c index ecc88b013eff..b24a320a11a6 100644 --- a/kernel/nstree.c +++ b/kernel/nstree.c @@ -106,7 +106,7 @@ void __ns_tree_add_raw(struct ns_common *ns, struct ns_tree *ns_tree) write_seqlock(&ns_tree->ns_tree_lock); - VFS_WARN_ON_ONCE(ns->ops->type != ns_tree->type); + VFS_WARN_ON_ONCE(ns->ns_type != ns_tree->type); node = rb_find_add_rcu(&ns->ns_tree_node, &ns_tree->ns_tree, ns_cmp); /* @@ -128,7 +128,7 @@ void __ns_tree_remove(struct ns_common *ns, struct ns_tree *ns_tree) { VFS_WARN_ON_ONCE(RB_EMPTY_NODE(&ns->ns_tree_node)); VFS_WARN_ON_ONCE(list_empty(&ns->ns_list_node)); - VFS_WARN_ON_ONCE(ns->ops->type != ns_tree->type); + VFS_WARN_ON_ONCE(ns->ns_type != ns_tree->type); write_seqlock(&ns_tree->ns_tree_lock); rb_erase(&ns->ns_tree_node, &ns_tree->ns_tree); @@ -197,7 +197,7 @@ struct ns_common *ns_tree_lookup_rcu(u64 ns_id, int ns_type) if (!node) return NULL; - VFS_WARN_ON_ONCE(node_to_ns(node)->ops->type != ns_type); + VFS_WARN_ON_ONCE(node_to_ns(node)->ns_type != ns_type); return node_to_ns(node); } @@ -225,7 +225,7 @@ struct ns_common *__ns_tree_adjoined_rcu(struct ns_common *ns, if (list_is_head(list, &ns_tree->ns_list)) return ERR_PTR(-ENOENT); - VFS_WARN_ON_ONCE(list_entry_rcu(list, struct ns_common, ns_list_node)->ops->type != ns_tree->type); + VFS_WARN_ON_ONCE(list_entry_rcu(list, struct ns_common, ns_list_node)->ns_type != ns_tree->type); return list_entry_rcu(list, struct ns_common, ns_list_node); } diff --git a/kernel/pid.c b/kernel/pid.c index 7e8c66e0bf67..0c2dcddb317a 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -85,6 +85,7 @@ struct pid_namespace init_pid_ns = { #if defined(CONFIG_SYSCTL) && defined(CONFIG_MEMFD_CREATE) .memfd_noexec_scope = MEMFD_NOEXEC_SCOPE_EXEC, #endif + .ns.ns_type = ns_common_type(&init_pid_ns), }; EXPORT_SYMBOL_GPL(init_pid_ns); diff --git a/kernel/pid_namespace.c b/kernel/pid_namespace.c index a262a3f19443..f5b222c8ac39 100644 --- a/kernel/pid_namespace.c +++ b/kernel/pid_namespace.c @@ -443,7 +443,6 @@ static struct user_namespace *pidns_owner(struct ns_common *ns) const struct proc_ns_operations pidns_operations = { .name = "pid", - .type = CLONE_NEWPID, .get = pidns_get, .put = pidns_put, .install = pidns_install, @@ -454,7 +453,6 @@ const struct proc_ns_operations pidns_operations = { const struct proc_ns_operations pidns_for_children_operations = { .name = "pid_for_children", .real_ns_name = "pid", - .type = CLONE_NEWPID, .get = pidns_for_children_get, .put = pidns_put, .install = pidns_install, diff --git a/kernel/time/namespace.c b/kernel/time/namespace.c index 9f26e61be044..530cf99c2212 100644 --- a/kernel/time/namespace.c +++ b/kernel/time/namespace.c @@ -462,7 +462,6 @@ out: const struct proc_ns_operations timens_operations = { .name = "time", - .type = CLONE_NEWTIME, .get = timens_get, .put = timens_put, .install = timens_install, @@ -472,7 +471,6 @@ const struct proc_ns_operations timens_operations = { const struct proc_ns_operations timens_for_children_operations = { .name = "time_for_children", .real_ns_name = "time", - .type = CLONE_NEWTIME, .get = timens_for_children_get, .put = timens_put, .install = timens_install, @@ -480,6 +478,7 @@ const struct proc_ns_operations timens_for_children_operations = { }; struct time_namespace init_time_ns = { + .ns.ns_type = ns_common_type(&init_time_ns), .ns.__ns_ref = REFCOUNT_INIT(3), .user_ns = &init_user_ns, .ns.inum = ns_init_inum(&init_time_ns), diff --git a/kernel/user.c b/kernel/user.c index b2a53674d506..0163665914c9 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -65,6 +65,7 @@ struct user_namespace init_user_ns = { .nr_extents = 1, }, }, + .ns.ns_type = ns_common_type(&init_user_ns), .ns.__ns_ref = REFCOUNT_INIT(3), .owner = GLOBAL_ROOT_UID, .group = GLOBAL_ROOT_GID, diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c index e1559e8a8a02..03cb63883d04 100644 --- a/kernel/user_namespace.c +++ b/kernel/user_namespace.c @@ -1400,7 +1400,6 @@ static struct user_namespace *userns_owner(struct ns_common *ns) const struct proc_ns_operations userns_operations = { .name = "user", - .type = CLONE_NEWUSER, .get = userns_get, .put = userns_put, .install = userns_install, diff --git a/kernel/utsname.c b/kernel/utsname.c index 00001592ad13..a8cdc84648ee 100644 --- a/kernel/utsname.c +++ b/kernel/utsname.c @@ -146,7 +146,6 @@ static struct user_namespace *utsns_owner(struct ns_common *ns) const struct proc_ns_operations utsns_operations = { .name = "uts", - .type = CLONE_NEWUTS, .get = utsns_get, .put = utsns_put, .install = utsns_install, -- cgit v1.2.3 From af075603f27b0f6e05f1bdf64bad42fa7cfb033b Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 24 Sep 2025 13:34:00 +0200 Subject: ns: drop assert Otherwise we warn when e.g., no namespaces are configured but the initial namespace for is still around. Reviewed-by: Jan Kara Signed-off-by: Christian Brauner --- kernel/nscommon.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/nscommon.c b/kernel/nscommon.c index 92c9df1e8774..c1fb2bad6d72 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -46,8 +46,6 @@ static void ns_debug(struct ns_common *ns, const struct proc_ns_operations *ops) VFS_WARN_ON_ONCE(ops != &utsns_operations); break; #endif - default: - VFS_WARN_ON_ONCE(true); } } #endif -- cgit v1.2.3