Re: [PATCH] Per-task watchers: Enable inheritance

[Peter Williams]

Matt Helsley wrote:
> This allows per-task watchers to implement inheritance of the same function
> and/or data in response to the initialization of new tasks. A watcher might
> implement inheritance using the following notifier_call snippet:
>
> int my_notify_func(struct notifier_block *nb, unsigned long val, void *t)
> {
> 	struct task_struct *tsk = t;
> 	struct notifier_block *child_nb;
> 	
> 	switch(get_watch_event(val)){
> 	case WATCH_TASK_INIT: /* use container_of() to associate extra data */
> 		child_nb = kzalloc(sizeof(struct notifier_block), GFP_KERNEL);
> 		if (!child_nb)
> 			return NOTIFY_DONE;
> 		child_nb->notifier_call = my_notify_func;
> 		register_per_task_watcher(tsk, child_nb);
> 		return NOTIFY_OK;
> 	case WATCH_TASK_FREE:
> 		unregister_per_task_watcher(tsk, nb);
> 		kfree(nb);
> 		return NOTIFY_OK;
>
> Compile tested only. Peter, is this useful to you?

I think that it's what I want (i.e. the implementation is what I would 
have done) but I'm confused by you reference to inheritance.  My concern 
is to NOT inherit the data (via the notifier_block) but to have separate 
data for each task which is why I was concerned about not finding where 
"notify" was being initialized on boot.

What I'm doing is using an ordinary watcher to catch new tasks being 
created via WATCH_TASK_INIT and attaching a per task watcher to them at 
that time.  As per your suggestion the notifier_block for the per task 
watcher is contained in a struct which contains the data that I need to 
maintain for each task.  So two layers of watchers :-)

It will be a good test of your mechanism if I can get it to work.

It'll probably take me another couple of days to complete this code as 
I'm having to figure out how it all hangs together as I go.  I'll let 
you know when I've finished.

Peter
--
Peter Williams                                   pwil3058@xxxxxxxxxxxxxx

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/