Re: [PATCH 1/7] async: Asynchronous function calls to speed up kernel boot

[Cornelia Huck <cornelia.huck@de.ibm.com>]

On Tue, 13 Jan 2009 13:48:59 -0700,
Jonathan Corbet <corbet@lwn.net> wrote:

> [A somewhat belated question...]
> 
> As I read the patch, I find the async_entry structure:
> 
> > +struct async_entry {
> > +	struct list_head list;
> > +	async_cookie_t   cookie;
> > +	async_func_ptr	 *func;
> > +	void             *data;
> > +	struct list_head *running;
> > +};
> 
> The "running" field is, presumably, meant to hold a pointer to the
> "running" queue to be used when this function is actually run.  But, then,
> I see:
> 
> > +async_cookie_t async_schedule(async_func_ptr *ptr, void *data)
> > +{
> > +	return __async_schedule(ptr, data, &async_pending);
> > +}
> > +EXPORT_SYMBOL_GPL(async_schedule);
> 
> It seems to me that you wanted &async_running there, no?
> 
> However, it doesn't matter in the current form of the patch:
> 
> > +/*
> > + * pick the first pending entry and run it
> > + */
> > +static void run_one_entry(void)
> > +{
> > +	unsigned long flags;
> > +	struct async_entry *entry;
> > +	ktime_t calltime, delta, rettime;
> > +
> > +	/* 1) pick one task from the pending queue */
> > +
> > +	spin_lock_irqsave(&async_lock, flags);
> > +	if (list_empty(&async_pending))
> > +		goto out;
> > +	entry = list_first_entry(&async_pending, struct async_entry, list); 
> > +
> > +	/* 2) move it to the running queue */
> > +	list_del(&entry->list);
> > +	list_add_tail(&entry->list, &async_running);
> > +	spin_unlock_irqrestore(&async_lock, flags);
> 
> Given the way things are designed, don't you want to add the entry to
> entry->running, rather than unconditionally to async_running?  If not, I
> don't see how calls to async_synchronize_cookie_special() can work right.
> 
> Of course, I'm probably just confused...enlighten me?

No, you're not confused, the code is :)

async_schedule() should pass in async_running as the running
list, and run_one_entry() should put the entry to be run on
the provided running list instead of always on the generic one.

Reported-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

---
 kernel/async.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

--- linux-2.6.orig/kernel/async.c
+++ linux-2.6/kernel/async.c
@@ -133,7 +133,7 @@ static void run_one_entry(void)
 
 	/* 2) move it to the running queue */
 	list_del(&entry->list);
-	list_add_tail(&entry->list, &async_running);
+	list_add_tail(&entry->list, entry->running);
 	spin_unlock_irqrestore(&async_lock, flags);
 
 	/* 3) run it (and print duration)*/
@@ -207,7 +207,7 @@ static async_cookie_t __async_schedule(a
 
 async_cookie_t async_schedule(async_func_ptr *ptr, void *data)
 {
-	return __async_schedule(ptr, data, &async_pending);
+	return __async_schedule(ptr, data, &async_running);
 }
 EXPORT_SYMBOL_GPL(async_schedule);