Skip to contentSkip to conversation

bpf: Fix UAF in task local storage

PATCH Applied KP Singh 3 participants 4 messages Jun 2, 2023 – Jun 5, 2023

Subsystems (4)
lore
Share & tools
lore

KP Singh

author
#0 lore raw

PATCH v2 1/1 bpf: Fix UAF in task local storage

When task local storage was generalized for tracing programs, the bpf_task_local_storage callback was moved from a BPF LSM hook callback for security_task_free LSM hook to it's own callback. But a failure case in bad_fork_cleanup_security was missed which, when triggered, led to a dangling task owner pointer and a subsequent use-after-free. Move the bpf_task_storage_free to the very end of free_task to handle all failure cases.

This issue was noticed when a BPF LSM program was attached to the task_alloc hook on a kernel with KASAN enabled. The program used bpf_task_storage_get to copy the task local storage from the current task to the new task being created.

Fixes a10787e6d58c ("bpf: Enable task local storage for tracing programs")

---

* v1 -> v2

Move the bpf_task_storage_free to free_task as suggested by Martin

 kernel/fork.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/fork.c +1 -1
index ed4e01daccaa..cb20f9f596d3 100644
@@ -627,6 +627,7 @@ void free_task(struct task_struct *tsk)
 	arch_release_task_struct(tsk);
 	if (tsk->flags & PF_KTHREAD)
 		free_kthread_struct(tsk);
+	bpf_task_storage_free(tsk);
 	free_task_struct(tsk);
 }
 EXPORT_SYMBOL(free_task);
@@ -979,7 +980,6 @@ void __put_task_struct(struct task_struct *tsk)
 	cgroup_free(tsk);
 	task_numa_free(tsk, true);
 	security_task_free(tsk);
-	bpf_task_storage_free(tsk);
 	exit_creds(tsk);
 	delayacct_tsk_free(tsk);
 	put_signal_struct(tsk->signal);
Signature · 1 line
2.41.0.rc0.172.g3f132b7071-goog
reviewer Acked-by on v2 1/1
#1 lore raw
Quoted 9 lines · KP Singh

When task local storage was generalized for tracing programs, the bpf_task_local_storage callback was moved from a BPF LSM hook callback for security_task_free LSM hook to it's own callback. But a failure case in bad_fork_cleanup_security was missed which, when triggered, led to a dangling task owner pointer and a subsequent use-after-free. Move the bpf_task_storage_free to the very end of free_task to handle all failure cases.

Quoted 8 lines

This issue was noticed when a BPF LSM program was attached to the task_alloc hook on a kernel with KASAN enabled. The program used bpf_task_storage_get to copy the task local storage from the current task to the new task being created.

Fixes a10787e6d58c ("bpf: Enable task local storage for tracing programs")

Thanks,
Song

Quoted 33 lines

---

* v1 -> v2

Move the bpf_task_storage_free to free_task as suggested by Martin

 kernel/fork.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/fork.c +1 -2
index ed4e01daccaa..cb20f9f596d3 100644
@@ -627,6 +627,7 @@ void free_task(struct task_struct *tsk)
        arch_release_task_struct(tsk);
        if (tsk->flags & PF_KTHREAD)
                free_kthread_struct(tsk);
+       bpf_task_storage_free(tsk);
        free_task_struct(tsk);
 }
 EXPORT_SYMBOL(free_task);
@@ -979,7 +980,6 @@ void __put_task_struct(struct task_struct *tsk)
        cgroup_free(tsk);
        task_numa_free(tsk, true);
        security_task_free(tsk);
-       bpf_task_storage_free(tsk);
        exit_creds(tsk);
        delayacct_tsk_free(tsk);
        put_signal_struct(tsk->signal);
--

2.41.0.rc0.172.g3f132b7071-goog

reviewer Applied
#2 lore raw
In reply to KP Singh
Quoted 30 lines · KP Singh

When task local storage was generalized for tracing programs, the bpf_task_local_storage callback was moved from a BPF LSM hook callback for security_task_free LSM hook to it's own callback. But a failure case in bad_fork_cleanup_security was missed which, when triggered, led to a dangling task owner pointer and a subsequent use-after-free. Move the bpf_task_storage_free to the very end of free_task to handle all failure cases.

This issue was noticed when a BPF LSM program was attached to the task_alloc hook on a kernel with KASAN enabled. The program used bpf_task_storage_get to copy the task local storage from the current task to the new task being created.

Fixes a10787e6d58c ("bpf: Enable task local storage for tracing programs")

---

* v1 -> v2

Move the bpf_task_storage_free to free_task as suggested by Martin

  kernel/fork.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
kernel/fork.c +0 -0
index ed4e01daccaa..cb20f9f596d3 100644

kernel/fork.c void free_task(struct task_struct *tsk) · line 630

Quoted 4 lines
kernel/fork.c +1 -0
@@ -627,6 +627,7 @@ void free_task(struct task_struct *tsk)
  	arch_release_task_struct(tsk);
  	if (tsk->flags & PF_KTHREAD)
  		free_kthread_struct(tsk);
+	bpf_task_storage_free(tsk);

Applied. Thanks for the fix.

A followup question, does it need a "notrace" to be added to bpf_task_storage_free to ensure no task storage can be recreated?

Quoted 11 lines
kernel/fork.c +0 -1
  	free_task_struct(tsk);
  }
  EXPORT_SYMBOL(free_task);
@@ -979,7 +980,6 @@ void __put_task_struct(struct task_struct *tsk)
  	cgroup_free(tsk);
  	task_numa_free(tsk, true);
  	security_task_free(tsk);
-	bpf_task_storage_free(tsk);
  	exit_creds(tsk);
  	delayacct_tsk_free(tsk);
  	put_signal_struct(tsk->signal);
author
#3 lore raw
Quoted 39 lines · Martin KaFai Lau
Quoted 30 lines · KP Singh

When task local storage was generalized for tracing programs, the bpf_task_local_storage callback was moved from a BPF LSM hook callback for security_task_free LSM hook to it's own callback. But a failure case in bad_fork_cleanup_security was missed which, when triggered, led to a dangling task owner pointer and a subsequent use-after-free. Move the bpf_task_storage_free to the very end of free_task to handle all failure cases.

This issue was noticed when a BPF LSM program was attached to the task_alloc hook on a kernel with KASAN enabled. The program used bpf_task_storage_get to copy the task local storage from the current task to the new task being created.

Fixes a10787e6d58c ("bpf: Enable task local storage for tracing programs")

---

* v1 -> v2

Move the bpf_task_storage_free to free_task as suggested by Martin

  kernel/fork.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
kernel/fork.c +0 -0
index ed4e01daccaa..cb20f9f596d3 100644

kernel/fork.c void free_task(struct task_struct *tsk) · line 630

Quoted 4 lines
kernel/fork.c +1 -0
@@ -627,6 +627,7 @@ void free_task(struct task_struct *tsk)
      arch_release_task_struct(tsk);
      if (tsk->flags & PF_KTHREAD)
              free_kthread_struct(tsk);
+     bpf_task_storage_free(tsk);

Applied. Thanks for the fix.

Quoted 2 lines

A followup question, does it need a "notrace" to be added to bpf_task_storage_free to ensure no task storage can be recreated?

Should we do a notrace or a BTF set based deny list?