Re: [PATCHv2 08/11] dmapool: consolidate page initialization

[Christoph Hellwig]

> +	page->in_use = 0;
> +	page->offset = 0;

This move makes total sense to me.

>  		return NULL;
>  
>  	spin_lock_irqsave(&pool->lock, flags);
> -
> -	list_add(&page->page_list, &pool->page_list);
> +	pool_initialise_page(pool, page);

.. but this now moves all of pool_initialise_page inside the
pool->lock critical section.  Is that really a good idea?