From 28aaa9c39945b7925a1cc1d513c8f21ed38f5e4f Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 26 Feb 2026 10:43:55 +0100 Subject: kthread: consolidate kthread exit paths to prevent use-after-free Guillaume reported crashes via corrupted RCU callback function pointers during KUnit testing. The crash was traced back to the pidfs rhashtable conversion which replaced the 24-byte rb_node with an 8-byte rhash_head in struct pid, shrinking it from 160 to 144 bytes. struct kthread (without CONFIG_BLK_CGROUP) is also 144 bytes. With CONFIG_SLAB_MERGE_DEFAULT and SLAB_HWCACHE_ALIGN both round up to 192 bytes and share the same slab cache. struct pid.rcu.func and struct kthread.affinity_node both sit at offset 0x78. When a kthread exits via make_task_dead() it bypasses kthread_exit() and misses the affinity_node cleanup. free_kthread_struct() frees the memory while the node is still linked into the global kthread_affinity_list. A subsequent list_del() by another kthread writes through dangling list pointers into the freed and reused memory, corrupting the pid's rcu.func pointer. Instead of patching free_kthread_struct() to handle the missed cleanup, consolidate all kthread exit paths. Turn kthread_exit() into a macro that calls do_exit() and add kthread_do_exit() which is called from do_exit() for any task with PF_KTHREAD set. This guarantees that kthread-specific cleanup always happens regardless of the exit path - make_task_dead(), direct do_exit(), or kthread_exit(). Replace __to_kthread() with a new tsk_is_kthread() accessor in the public header. Export do_exit() since module code using the kthread_exit() macro now needs it directly. Reported-by: Guillaume Tucker Tested-by: Guillaume Tucker Tested-by: Mark Brown Tested-by: David Gow Cc: Link: https://lore.kernel.org/all/20260224-mittlerweile-besessen-2738831ae7f6@brauner Co-developed-by: Linus Torvalds Fixes: 4d13f4304fa4 ("kthread: Implement preferred affinity") Signed-off-by: Linus Torvalds Signed-off-by: Christian Brauner --- kernel/exit.c | 6 ++++++ kernel/kthread.c | 41 +++++------------------------------------ 2 files changed, 11 insertions(+), 36 deletions(-) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 8a87021211ae..ede3117fa7d4 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -896,11 +896,16 @@ static void synchronize_group_exit(struct task_struct *tsk, long code) void __noreturn do_exit(long code) { struct task_struct *tsk = current; + struct kthread *kthread; int group_dead; WARN_ON(irqs_disabled()); WARN_ON(tsk->plug); + kthread = tsk_is_kthread(tsk); + if (unlikely(kthread)) + kthread_do_exit(kthread, code); + kcov_task_exit(tsk); kmsan_task_exit(tsk); @@ -1013,6 +1018,7 @@ void __noreturn do_exit(long code) lockdep_free_task(tsk); do_task_dead(); } +EXPORT_SYMBOL(do_exit); void __noreturn make_task_dead(int signr) { diff --git a/kernel/kthread.c b/kernel/kthread.c index 20451b624b67..791210daf8b4 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -85,24 +85,6 @@ static inline struct kthread *to_kthread(struct task_struct *k) return k->worker_private; } -/* - * Variant of to_kthread() that doesn't assume @p is a kthread. - * - * When "(p->flags & PF_KTHREAD)" is set the task is a kthread and will - * always remain a kthread. For kthreads p->worker_private always - * points to a struct kthread. For tasks that are not kthreads - * p->worker_private is used to point to other things. - * - * Return NULL for any task that is not a kthread. - */ -static inline struct kthread *__to_kthread(struct task_struct *p) -{ - void *kthread = p->worker_private; - if (kthread && !(p->flags & PF_KTHREAD)) - kthread = NULL; - return kthread; -} - void get_kthread_comm(char *buf, size_t buf_size, struct task_struct *tsk) { struct kthread *kthread = to_kthread(tsk); @@ -193,7 +175,7 @@ EXPORT_SYMBOL_GPL(kthread_should_park); bool kthread_should_stop_or_park(void) { - struct kthread *kthread = __to_kthread(current); + struct kthread *kthread = tsk_is_kthread(current); if (!kthread) return false; @@ -234,7 +216,7 @@ EXPORT_SYMBOL_GPL(kthread_freezable_should_stop); */ void *kthread_func(struct task_struct *task) { - struct kthread *kthread = __to_kthread(task); + struct kthread *kthread = tsk_is_kthread(task); if (kthread) return kthread->threadfn; return NULL; @@ -266,7 +248,7 @@ EXPORT_SYMBOL_GPL(kthread_data); */ void *kthread_probe_data(struct task_struct *task) { - struct kthread *kthread = __to_kthread(task); + struct kthread *kthread = tsk_is_kthread(task); void *data = NULL; if (kthread) @@ -309,19 +291,8 @@ void kthread_parkme(void) } EXPORT_SYMBOL_GPL(kthread_parkme); -/** - * kthread_exit - Cause the current kthread return @result to kthread_stop(). - * @result: The integer value to return to kthread_stop(). - * - * While kthread_exit can be called directly, it exists so that - * functions which do some additional work in non-modular code such as - * module_put_and_kthread_exit can be implemented. - * - * Does not return. - */ -void __noreturn kthread_exit(long result) +void kthread_do_exit(struct kthread *kthread, long result) { - struct kthread *kthread = to_kthread(current); kthread->result = result; if (!list_empty(&kthread->affinity_node)) { mutex_lock(&kthread_affinity_lock); @@ -333,9 +304,7 @@ void __noreturn kthread_exit(long result) kthread->preferred_affinity = NULL; } } - do_exit(0); } -EXPORT_SYMBOL(kthread_exit); /** * kthread_complete_and_exit - Exit the current kthread. @@ -683,7 +652,7 @@ void kthread_set_per_cpu(struct task_struct *k, int cpu) bool kthread_is_per_cpu(struct task_struct *p) { - struct kthread *kthread = __to_kthread(p); + struct kthread *kthread = tsk_is_kthread(p); if (!kthread) return false; -- cgit v1.2.3 From e6b899f08066e744f89df16ceb782e06868bd148 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 26 Feb 2026 14:50:09 +0100 Subject: nsfs: tighten permission checks for ns iteration ioctls Even privileged services should not necessarily be able to see other privileged service's namespaces so they can't leak information to each other. Use may_see_all_namespaces() helper that centralizes this policy until the nstree adapts. Link: https://patch.msgid.link/20260226-work-visibility-fixes-v1-1-d2c2853313bd@kernel.org Fixes: a1d220d9dafa ("nsfs: iterate through mount namespaces") Reviewed-by: Jeff Layton Cc: stable@kernel.org # v6.12+ Signed-off-by: Christian Brauner --- kernel/nscommon.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/nscommon.c b/kernel/nscommon.c index bdc3c86231d3..3166c1fd844a 100644 --- a/kernel/nscommon.c +++ b/kernel/nscommon.c @@ -309,3 +309,9 @@ void __ns_ref_active_get(struct ns_common *ns) return; } } + +bool may_see_all_namespaces(void) +{ + return (task_active_pid_ns(current) == &init_pid_ns) && + ns_capable_noaudit(init_pid_ns.user_ns, CAP_SYS_ADMIN); +} -- cgit v1.2.3 From 8d76afe84fa2babf604b3c173730d4d2b067e361 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Thu, 26 Feb 2026 14:50:11 +0100 Subject: nstree: tighten permission checks for listing Even privileged services should not necessarily be able to see other privileged service's namespaces so they can't leak information to each other. Use may_see_all_namespaces() helper that centralizes this policy until the nstree adapts. Link: https://patch.msgid.link/20260226-work-visibility-fixes-v1-3-d2c2853313bd@kernel.org Fixes: 76b6f5dfb3fd ("nstree: add listns()") Reviewed-by: Jeff Layton Cc: stable@kernel.org # v6.19+ Signed-off-by: Christian Brauner --- kernel/nstree.c | 29 ++++------------------------- 1 file changed, 4 insertions(+), 25 deletions(-) (limited to 'kernel') diff --git a/kernel/nstree.c b/kernel/nstree.c index f36c59e6951d..6d12e5900ac0 100644 --- a/kernel/nstree.c +++ b/kernel/nstree.c @@ -515,32 +515,11 @@ static inline bool __must_check ns_requested(const struct klistns *kls, static inline bool __must_check may_list_ns(const struct klistns *kls, struct ns_common *ns) { - if (kls->user_ns) { - if (kls->userns_capable) - return true; - } else { - struct ns_common *owner; - struct user_namespace *user_ns; - - owner = ns_owner(ns); - if (owner) - user_ns = to_user_ns(owner); - else - user_ns = &init_user_ns; - if (ns_capable_noaudit(user_ns, CAP_SYS_ADMIN)) - return true; - } - - if (is_current_namespace(ns)) + if (kls->user_ns && kls->userns_capable) return true; - - if (ns->ns_type != CLONE_NEWUSER) - return false; - - if (ns_capable_noaudit(to_user_ns(ns), CAP_SYS_ADMIN)) + if (is_current_namespace(ns)) return true; - - return false; + return may_see_all_namespaces(); } static inline void ns_put(struct ns_common *ns) @@ -600,7 +579,7 @@ static ssize_t do_listns_userns(struct klistns *kls) ret = 0; head = &to_ns_common(kls->user_ns)->ns_owner_root.ns_list_head; - kls->userns_capable = ns_capable_noaudit(kls->user_ns, CAP_SYS_ADMIN); + kls->userns_capable = may_see_all_namespaces(); rcu_read_lock(); -- cgit v1.2.3