Re: [PATCH v2] livepatch: fix race between fork and klp_reverse_transition

[Petr Mladek]

On Fri 2022-07-22 15:01:06, Rik van Riel wrote:
> v2: a better approach, suggested by Petr (thank you)
> ---8<---
> 
> 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;
>         }
> 
> 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.

It should fix these races as well. Both task->patch_state and
TIF_PATCH_PENDING flag are almost always modified under tasklist_lock.

One exception is klp_update_patch_state(current) but it could not
race because klp_copy_process() is called under spinlock.
So that "current" can't sleep and can't get migrated in the middle of
klp_copy_process().

Another exception is klp_check_and_switch_task() that is called
under p->pi_lock. It prevents rescheduling and the task will be
migrated only when sleeping. As a result "current" again
can't get migrated inside klp_copy_process().

Finally, the state of "idle" tasks (idle_task(cpu)) is updated
without tasklist_lock. But they are not forked so that we are
on safe side.


> Signed-off-by: Rik van Riel <riel@xxxxxxxxxxx>
> Reported-by: Breno Leitao <leitao@xxxxxxxxxx>

We should update the commit message and mention also the other
two locations where the state is manipulated without tasklist_lock.
I am sorry that I did not mention it on Friday.

Also we should remove "I have not figured yet whether". The patch
should fix these races as well.

With the above changes:

Reviewed-by: Petr Mladek <pmladek@xxxxxxxx>


Best Regards,
Petr