From: Steffen Klassert Subject: [PATCH 3/4] padata: Add some code comments Date: Fri, 14 May 2010 13:46:06 +0200 Message-ID: <20100514114606.GD2184@secunet.com> References: <20100514114336.GB2184@secunet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-kernel@vger.kernel.org, linux-crypto@vger.kernel.org To: Herbert Xu , Andrew Morton Return-path: Received: from a.mx.secunet.com ([195.81.216.161]:56863 "EHLO a.mx.secunet.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752671Ab0ENLsU (ORCPT ); Fri, 14 May 2010 07:48:20 -0400 Content-Disposition: inline In-Reply-To: <20100514114336.GB2184@secunet.com> Sender: linux-crypto-owner@vger.kernel.org List-ID: Signed-off-by: Steffen Klassert --- include/linux/padata.h | 53 ++++++++++++++++++++++++++++++++++++++++++++++++ kernel/padata.c | 50 +++++++++++++++++++++++++++++++++++++++++++- 2 files changed, 101 insertions(+), 2 deletions(-) diff --git a/include/linux/padata.h b/include/linux/padata.h index 64836a6..e8aac0f 100644 --- a/include/linux/padata.h +++ b/include/linux/padata.h @@ -26,6 +26,17 @@ #include #include +/* + * struct padata_priv - Embedded to the users data structure. + * + * @list: List entry, to attach to the padata lists. + * @pd: Pointer to the internal control structure. + * @cb_cpu: Callback cpu for serializatioon. + * @seq_nr: Sequence number of the parallelized data object. + * @info: Used to pass information from the parallel to the serial function. + * @parallel: Parallel execution function. + * @serial: Serial complete function. + */ struct padata_priv { struct list_head list; struct parallel_data *pd; @@ -36,11 +47,29 @@ struct padata_priv { void (*serial)(struct padata_priv *padata); }; +/* + * struct padata_list + * + * @list: List head. + * @lock: List lock. + */ struct padata_list { struct list_head list; spinlock_t lock; }; +/* + * struct padata_queue - The percpu padata queues. + * + * @parallel: List to wait for parallelization. + * @reorder: List to wait for reordering after parallel processing. + * @serial: List to wait for serialization after reordering. + * @pwork: work struct for parallelization. + * @swork: work struct for serialization. + * @pd: Backpointer to the internal control structure. + * @num_obj: Number of objects that are processed by this cpu. + * @cpu_index: Index of the cpu. + */ struct padata_queue { struct padata_list parallel; struct padata_list reorder; @@ -52,6 +81,20 @@ struct padata_queue { int cpu_index; }; +/* + * struct parallel_data - Internal control structure, covers everything + * that depends on the cpumask in use. + * + * @pinst: padata instance. + * @queue: percpu padata queues. + * @seq_nr: The sequence number that will be attached to the next object. + * @reorder_objects: Number of objects waiting in the reorder queues. + * @refcnt: Number of objects holding a reference on this parallel_data. + * @max_seq_nr: Maximal used sequence number. + * @cpumask: cpumask in use. + * @lock: Reorder lock. + * @timer: Reorder timer. + */ struct parallel_data { struct padata_instance *pinst; struct padata_queue *queue; @@ -64,6 +107,16 @@ struct parallel_data { struct timer_list timer; }; +/* + * struct padata_instance - The overall control structure. + * + * @cpu_notifier: cpu hotplug notifier. + * @wq: The workqueue in use. + * @pd: The internal control structure. + * @cpumask: User supplied cpumask. + * @lock: padata instance lock. + * @flags: padata flags. + */ struct padata_instance { struct notifier_block cpu_notifier; struct workqueue_struct *wq; diff --git a/kernel/padata.c b/kernel/padata.c index ec6b8b7..629bef3 100644 --- a/kernel/padata.c +++ b/kernel/padata.c @@ -152,6 +152,23 @@ out: } EXPORT_SYMBOL(padata_do_parallel); +/* + * padata_get_next - Get the next object that needs serialization. + * + * Return values are: + * + * A pointer to the control struct of the next object that needs + * serialization, if present in one of the percpu reorder queues. + * + * NULL, if all percpu reorder queues are empty. + * + * -EINPROGRESS, if the next object that needs serialization will + * be parallel processed by another cpu and is not yet present in + * the cpu's reorder queue. + * + * -ENODATA, if this cpu has to do the parallel processing for + * the next object. + */ static struct padata_priv *padata_get_next(struct parallel_data *pd) { int cpu, num_cpus, empty, calc_seq_nr; @@ -173,7 +190,7 @@ static struct padata_priv *padata_get_next(struct parallel_data *pd) /* * Calculate the seq_nr of the object that should be - * next in this queue. + * next in this reorder queue. */ overrun = 0; calc_seq_nr = (atomic_read(&queue->num_obj) * num_cpus) @@ -248,15 +265,36 @@ static void padata_reorder(struct parallel_data *pd) struct padata_queue *queue; struct padata_instance *pinst = pd->pinst; + /* + * We need to ensure that only one cpu can work on dequeueing of + * the reorder queue the time. Calculating in which percpu reorder + * queue the next object will arrive takes some time. A spinlock + * would be highly contended. Also it is not clear in which order + * the objects arrive to the reorder queues. So a cpu could wait to + * get the lock just to notice that there is nothing to do at the + * moment. Therefore we use a trylock and let the holder of the lock + * care for all the objects enqueued during the holdtime of the lock. + */ if (!spin_trylock_bh(&pd->lock)) return; while (1) { padata = padata_get_next(pd); + /* + * All reorder queues are empty, or the next object that needs + * serialization is parallel processed by another cpu and is + * still on it's way to the cpu's reorder queue, nothing to + * do for now. + */ if (!padata || PTR_ERR(padata) == -EINPROGRESS) break; + /* + * This cpu has to do the parallel processing of the next + * object. It's waiting in the cpu's parallelization queue, + * so exit imediately. + */ if (PTR_ERR(padata) == -ENODATA) { del_timer(&pd->timer); spin_unlock_bh(&pd->lock); @@ -274,6 +312,11 @@ static void padata_reorder(struct parallel_data *pd) spin_unlock_bh(&pd->lock); + /* + * The next object that needs serialization might have arrived to + * the reorder queues in the meantime, we will be called again + * from the timer function if noone else cares for it. + */ if (atomic_read(&pd->reorder_objects) && !(pinst->flags & PADATA_RESET)) mod_timer(&pd->timer, jiffies + HZ); @@ -348,6 +391,7 @@ void padata_do_serial(struct padata_priv *padata) } EXPORT_SYMBOL(padata_do_serial); +/* Allocate and initialize the internal cpumask dependend resources. */ static struct parallel_data *padata_alloc_pd(struct padata_instance *pinst, const struct cpumask *cpumask) { @@ -417,6 +461,7 @@ static void padata_free_pd(struct parallel_data *pd) kfree(pd); } +/* Flush all objects out of the padata queues. */ static void padata_flush_queues(struct parallel_data *pd) { int cpu; @@ -440,6 +485,7 @@ static void padata_flush_queues(struct parallel_data *pd) BUG_ON(atomic_read(&pd->refcnt) != 0); } +/* Replace the internal control stucture with a new one. */ static void padata_replace(struct padata_instance *pinst, struct parallel_data *pd_new) { @@ -706,7 +752,7 @@ EXPORT_SYMBOL(padata_alloc); /* * padata_free - free a padata instance * - * @ padata_inst: padata instance to free + * @padata_inst: padata instance to free */ void padata_free(struct padata_instance *pinst) { -- 1.5.6.5