Re: [PATCH,RFC] livepatch: fix race between fork and klp_reverse_transition

[Petr Mladek <pmladek@suse.com>]

On Wed 2022-07-20 12:10:23, Rik van Riel wrote:
> When a KLP fails to apply, klp_reverse_transition will clear the
> TIF_PATCH_PENDING flag on all tasks, except for newly created tasks
> which are not on the task list yet.
> 
> Meanwhile, fork will copy over the TIF_PATCH_PENDING flag from the
> parent to the child early on, in dup_task_struct -> setup_thread_stack.
> 
> Much later, klp_copy_process will set child->patch_state to match
> that of the parent.
> 
> However, the parent's patch_state may have been changed by KLP loading
> or unloading since it was initially copied over into the child.
> 
> This results in the KLP code occasionally hitting this warning in
> klp_complete_transition:
> 
>         for_each_process_thread(g, task) {
>                 WARN_ON_ONCE(test_tsk_thread_flag(task, TIF_PATCH_PENDING));
>                 task->patch_state = KLP_UNDEFINED;
>         }

I see.

> This patch will set, or clear, the TIF_PATCH_PENDING flag in the child
> process depending on whether or not it is needed at the time
> klp_copy_process is called, at a point in copy_process where the
> tasklist_lock is held exclusively, preventing races with the KLP
> code.
> 
> This should prevent this warning from triggering again in the
> future.
> 
> I have not yet figured out whether this would also help with races in
> the other direction, where the child process fails to have TIF_PATCH_PENDING
> set and somehow misses a transition, or whether the retries in
> klp_try_complete_transition would catch that task and help it transition
> later.
> 
> Signed-off-by: Rik van Riel <riel@surriel.com>
> Reported-by: Breno Leitao <leitao@debian.org>
> ---
>  kernel/livepatch/transition.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/livepatch/transition.c b/kernel/livepatch/transition.c
> index 5d03a2ad1066..7a90ad5e9224 100644
> --- a/kernel/livepatch/transition.c
> +++ b/kernel/livepatch/transition.c
> @@ -612,7 +612,15 @@ void klp_copy_process(struct task_struct *child)
>  {
>  	child->patch_state = current->patch_state;
>  
> -	/* TIF_PATCH_PENDING gets copied in setup_thread_stack() */
> +	/*
> +	 * The parent process may have gone through a KLP transition since
> +	 * the thread flag was copied in setup_thread_stack earlier. Set
> +	 * the flag according to whether this task needs a KLP transition.
> +	 */
> +	if (child->patch_state != klp_target_state)
> +		set_tsk_thread_flag(child, TIF_PATCH_PENDING);
> +	else
> +		clear_tsk_thread_flag(child, TIF_PATCH_PENDING);
>  }

I am afraid that it is more complicated.

If the parent process might have gone through a KLP transition
then also the transition might have finished and klp_target_state might be
KLP_UNDEFINED. We must not set TIF_PATCH_PENDING in this case.

Now, we might race with klp_complete_transition() at any stage. It
might be before or after setting task->patch_state = KLP_UNDEFINED.
And it might be before or after setting klp_target_state =
KLP_UNDEFINED.

The great thing is that we could not race with
klp_update_patch_state() that would be migrating current because
we are current.

So, the easiest solution would be to copy the flag from current
once again here:


/* Called from copy_process() during fork */
void klp_copy_process(struct task_struct *child)
{
	/*
	 * The parent process may have gone through a KLP transition since
	 * the thread flag was copied in setup_thread_stack earlier.
	 * Copy also the flag once again here.
	 *
	 * The operation is serialized against all klp_*_transition()
	 * operations by tasklist_lock. The only exception is
	 * klp_update_patch_state(current). But it could not race
	 * because we are current.
	 */
	if (test_tsk_thread_flag(current, TIF_PATCH_PENDING))
		set_tsk_thread_flag(child, TIF_PATCH_PENDING);
	else
		clear_tsk_thread_flag(child, TIF_PATCH_PENDING);

	child->patch_state = current->patch_state;
}


I hope that I did not miss anything. It is Friday.

Best Regards,
Petr