Re: [PATCH] kthread: Don't depend on work queues

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

Oleg Nesterov <oleg@tv-sign.ru> writes:

> On 04/10, Eric W. Biederman wrote:
>> 
>>  static int kthread(void *_create)
>>  {
>>  	struct kthread_create_info *create = _create;
>>  	int (*threadfn)(void *data);
>>  	void *data;
>> -	sigset_t blocked;
>>  	int ret = -EINTR;
>>  
>> -	kthread_exit_files();
>> -
>> -	/* Copy data: it's on keventd's stack */
>> +	/* Copy data: it's on kthread's stack */
>>  	threadfn = create->threadfn;
>>  	data = create->data;
>>  
>> -	/* Block and flush all signals (in case we're not from keventd). */
>> -	sigfillset(&blocked);
>> -	sigprocmask(SIG_BLOCK, &blocked, NULL);
>> -	flush_signals(current);
>> -
>> -	/* By default we can run anywhere, unlike keventd. */
>> -	set_cpus_allowed(current, CPU_MASK_ALL);
>
> The above is OK, but I believe you should add set_cpus_allowed() to
> kthreadd_setup(). Note that kthreadd() is forked after init_idle().

init_idle()?  Yep in sched_init.  I wondered where that happened
for the initial idle thread.  Ugh. That is the first thing that
we fix after we fork init as well.  So yes it would be a good
idea to fix that...

I was expecting init_idle() to happen after we forked kthreadd. I almost
wonder if I should move init_idle....

Anyway that is one bug caught.

Eric