Re: [PATCH 0/4] allow multiple kthreadd's

[J. Bruce Fields]

On Tue, May 05, 2020 at 05:09:56PM -0400, Tejun Heo wrote:
> Hello,
> 
> On Tue, May 05, 2020 at 05:01:18PM -0400, J. Bruce Fields wrote:
> > On Mon, May 04, 2020 at 10:15:14PM -0400, J. Bruce Fields wrote:
> > > Though now I'm feeling greedy: it would be nice to have both some kind
> > > of global flag, *and* keep kthread->data pointing to svc_rqst (as that
> > > would give me a simpler and quicker way to figure out which client is
> > > conflicting).  Could I take a flag bit in kthread->flags, maybe?
> > 
> > Would something like this be too hacky?:
> 
> It's not the end of the world but a bit hacky. I wonder whether something
> like the following would work better for identifying worker type so that you
> can do sth like
> 
>  if (kthread_fn(current) == nfsd)
>         return kthread_data(current);
>  else
>         return NULL;     

Yes, definitely more generic, looks good to me.

--b.

> 
> Thanks.
> 
> diff --git a/kernel/kthread.c b/kernel/kthread.c
> index bfbfa481be3a..4f3ab9f2c994 100644
> --- a/kernel/kthread.c
> +++ b/kernel/kthread.c
> @@ -46,6 +46,7 @@ struct kthread_create_info
>  struct kthread {
>  	unsigned long flags;
>  	unsigned int cpu;
> +	int (*threadfn)(void *);
>  	void *data;
>  	struct completion parked;
>  	struct completion exited;
> @@ -152,6 +153,13 @@ bool kthread_freezable_should_stop(bool *was_frozen)
>  }
>  EXPORT_SYMBOL_GPL(kthread_freezable_should_stop);
>  
> +void *kthread_fn(struct task_struct *task)
> +{
> +	if (task->flags & PF_KTHREAD)
> +		return to_kthread(task)->threadfn;
> +	return NULL;
> +}
> +
>  /**
>   * kthread_data - return data value specified on kthread creation
>   * @task: kthread task in question
> @@ -244,6 +252,7 @@ static int kthread(void *_create)
>  		do_exit(-ENOMEM);
>  	}
>  
> +	self->threadfn = threadfn;
>  	self->data = data;
>  	init_completion(&self->exited);
>  	init_completion(&self->parked);
> 
> -- 
> tejun