Re: [PATCH v3] do_wait: make PIDTYPE_PID case O(1) instead of O(n)

[ebiederm@xmission.com (Eric W. Biederman)]

Jim Newsome <jnewsome@torproject.org> writes:

> On 3/10/21 16:40, Eric W. Biederman wrote:
>>> +// Optimization for waiting on PIDTYPE_PID. No need to iterate
> through child
>>> +// and tracee lists to find the target task.
>>
>> Minor nit:  C++ style comments look very out of place in this file
>>             which uses old school C /* */ comment delimiters for
>>             all of it's block comments.
>
> Will do
>
>>> +static int do_wait_pid(struct wait_opts *wo)
>>> +{
>>> +	struct task_struct *target = pid_task(wo->wo_pid, PIDTYPE_PID);
>>                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> This is subtle change in behavior.
>> 
>> Today on the task->children list we only place thread group leaders.
>
> Shouldn't we allow waiting on clone children if __WALL or __WCLONE is set?
>
> This is already checked later in `eligible_child`, called from
> `wait_consider_task`, so I *think* the current form should already do
> the right thing. Now I'm confused though how the general path (through
> `do_wait_thread`) works if clone children aren't on the task->children
> list...?
>
> (In any case it seems this will need another version with at least an
> explanatory comment here)

What I am worried about are not clone children.  AKA ordinary children
that have a different exit signal but CLONE_THREAD children that are
never put on the children list so are naturally excluded from today's
do_wait (except in the case of ptrace). These are also known as threads.

Maybe I am missing it but I don't see anything in wait_consider_task
or in the way that you are calling it that would exclude CLONE_THREAD
children for the non-ptrace case.

Eric